jpad journal

AND option

OR option


R.E. Curiel Cid1, E.A. Crocco1, M. Kitaigorodsky1, L. Beaufils2, P.A. Peña2, G. Grau1, U. Visser2, D.A. Loewenstein1

1. Center for Cognitive Neuroscience and Aging, Department of Psychiatry and Behavioral Sciences, University of Miami Miller School of Medicine, 1695 NW 9th Avenue, Miami, Florida, 33136. U.S.A; 2. Department of Computer Science, University of Miami, 1365 Memorial Drive, Coral Gables, Florida 33146, U.S.A.

Corresponding Author: Rosie E. Curiel, Psy.D., Associate Professor of Psychiatry & Behavioral Sciences, University of Miami Miller School of Medicine, 1695 NW 9th Avenue, Suite 3202, Miami, FL 33136.

J Prev Alz Dis 2021;
Published online January 19, 2021,



BACKGROUND: The Loewenstein Acevedo Scales of Semantic Interference and Learning (LASSI-L) is a novel and increasingly employed instrument that has outperformed widely used cognitive measures as an early correlate of elevated brain amyloid and neurodegeneration in prodromal Alzheimer’s Disease (AD). The LASSI-L has distinguished those with amnestic mild cognitive impairment (aMCI) and high amyloid load from aMCI attributable to other non-AD conditions. The authors designed and implemented a web-based brief computerized version of the instrument, the LASSI-BC, to improve standardized administration, facilitate scoring accuracy, real-time data entry, and increase the accessibility of the measure.
Objective: The psychometric properties and clinical utility of the brief computerized version of the LASSI-L was evaluated, together with its ability to differentiate older adults who are cognitively normal (CN) from those with amnestic Mild Cognitive Impairment (aMCI).
Methods: After undergoing a comprehensive uniform clinical and neuropsychological evaluation using traditional measures, older adults were classified as cognitively normal or diagnosed with aMCI. All participants were administered the LASSI-BC, a computerized version of the LASSI-L. Test-retest and discriminant validity was assessed for each LASSI-BC subscale.
Results: LASSI-BC subscales demonstrated high test-retest reliability, and discriminant validity was attained.
Conclusions: The LASSI-BC, a brief computerized version of the LASSI-L is a valid and useful cognitive tool for the detection of aMCI among older adults.

Key words: Computerized test, mild cognitive impairment, Alzheimer’s disease, semantic intrusion errors, semantic interference, clinical trials.



Alzheimer’s disease (AD) is a devastating condition that is expected to significantly impact the rapidly aging population. Important advancements have been made to identify novel candidate biomarkers of AD, and a research framework to stage the disease from its preclinical stage onward has been proposed, with the aim of establishing a biological definition of the disease (1). Despite these formidable advances, neuropsychological assessment remains an essential component of the evaluative process because cognitive impairment is a fundamental defining symptom of AD that emerges early, at a certain point in the transition from the preclinical to clinically symptomatic stages of the disease. Further, cognitive changes are used to detect and track disease progression over time and a measurable change in cognitive ability represents a potentially meaningful clinical outcome (2). Thus, the identification of cognitive markers that are sensitive to detecting early disease states and converge with biological markers of AD pathology, have become increasingly necessary in terms of identifying individuals at risk, monitoring disease progression, and ascertaining treatment efficacy (3).
Traditional paper-and-pencil cognitive measures employed for the detection of AD-related Mild Cognitive Impairment (MCI) are often insensitive to detecting subtle cognitive changes that occur during preclinical or prodromal disease states (5, 6). There is a developing body of literature, however, that cognitive stress paradigms can measure subtle deficiencies that are highly implicated in early AD disease states among older adults. One such paradigm that measures semantic interference in memory, the Loewenstein-Acevedo Scales for Semantic Interference and Learning (LASSI-L), was sensitive enough to differentiate older adults who are cognitively unimpaired from those with subjective memory complaints, and early amnestic MCI (7, 8). On this memory measure, proactive semantic interference (PSI) deficits and particularly, the inability to recover from PSI (frPSI) was also highly associated with brain amyloid load in older adults with otherwise normal performance on a traditional battery of cognitive tests (9). The LASSI-L has outperformed other widely used memory measures in detecting prodromal AD in both English and Spanish (10, 11), and has been found to be useful in different cultural/language groups (7, 11, 12). In addition to measuring the total number of correct targets recalled on individual LASSI-L subscales, there is evidence that semantic intrusion errors may have specific utility in the assessment of prodromal AD. Loewenstein and colleagues (4) found that semantic intrusion errors sensitive to PSI and frPSI on the LASSI-L could differentiate amyloid positive aMCI groups from amyloid negative aMCI groups with non-AD diagnoses.
While it is recognized that intrusion errors represent early manifestations of neurodegenerative brain disease, a potential limitation of previous approaches is that the number of intrusion errors are often highly dependent on an individual’s total responses on a particular trial. Thus, even a seemingly modest number of intrusion errors may actually represent an at-risk cognitive profile, depending on the total number of responses that are correct. For example, an individual may make a minimal number of intrusion errors on a given trial, which may appear to be clinically insignificant. However, if the number of total responses is low, even a modest number of intrusion errors may indicate impaired inhibitory processes and underlying brain pathology. As a result, we recently developed a novel method to evaluate semantic intrusion errors utilizing the percentage of intrusion errors (PIE) in relation to total correct responses (13). This method takes into account the observation that the number of intrusion errors a person makes is often highly dependent on their total responses on a particular trial. Thus, even a seemingly modest number of intrusion errors may represent an at-risk cognitive profile. PIE demonstrated high levels of sensitivity and specificity in differentiating CN from amyloid positive persons with preclinical AD and preliminary work suggests that it is a novel and sensitive index of early memory dysfunction (11, 13).
Traditional paper and pencil neuropsychological assessments are lengthy, require a skilled examiner, are vulnerable to human error in administration and scoring, and associated with practice effects. Moreover, some of these measures have been found to be biased among diverse ethnic/cultural and language groups. To address some of these concerns, computerized testing batteries have been developed to explore a more suitable option to mitigate some of the abovementioned limitations (14-17). However, these too have limitations in early detection of AD-associated cognitive impairment. For example, many of these computerized batteries are relatively successful at distinguishing between older adults with normal cognition and those with dementia or late stage MCI, but lack the predictive power needed to move the field forward, which is to correctly classify individuals with MCI and/or earlier on the disease continuum, and do so in a manner that is validated for use among different ethnic/cultural and language groups. This highlights a major problem with many traditional computerized batteries; they are automated versions of traditional paper-and-pencil cognitive assessment paradigms that lack sensitivity to detect AD-associated cognitive decline, and employ the same paradigms originally developed for the assessment of dementia or traumatic brain injury (17).
Recent work by Curiel and associates (5-12) led to the development of a brief computerized version of the LASSI-L, the LASSI-BC, which incorporates all the elements of this well-established cognitive stress test. The LASSI-BC is currently being studied extensively in a longitudinal study of at-risk aging adults. This novel computerized version of the instrument does not require a skilled examiner, is web-based and can remotely run on most browser-capable devices. Moreover, it is intuitive and appropriate for use among older adults that are either predominantly English or Spanish-speaking and who have varying ethnic/cultural backgrounds including Hispanics and African Americans.
In this first validation study, we examine the psychometric properties of the LASSI-BC. We also evaluate the clinical utility several LASSI-BC subscales as it relates to their ability to differentiate older adults with normal cognition from those with aMCI on measures of: i) proactive semantic interference, ii) the failure to recover from proactive semantic interference, iii) retroactive semantic interference and iv) the percentage of intrusion errors in relation to total cued recall responses by the participant. Performance on these specific subscales were selected a priori because, as noted above, our previous work using the paper-and-pencil LASSI-L has robustly demonstrated that these particular subscales are the most sensitive to cognitive breakdowns associated with MCI due to preclinical and prodromal AD.



This study included 64 older adults that were evaluated as part of an IRB-approved longitudinal investigation funded by the National Institute on Aging. An experienced clinician administered a standard clinical assessment protocol, which included the Clinical Dementia Rating Scale (CDR) (18), and the Mini-Mental State Examination (MMSE) (19). Subsequently, a neuropsychological battery was independently administered in either Spanish or English dependent on the participant’s dominant and preferred language. Spanish language evaluations were completed with equivalent standardized neuropsychological tests and appropriate age, education, and cultural/language normative data (20-23). Proficient bilingual (Spanish/English) psychometricians performed all the testing.
Diagnostic groups were classified using the following criteria:

Amnestic MCI group (aMCI) (n=25)

Participants met Petersen’s criteria (24)) for MCI and evidenced all of the following: a) subjective cognitive complaints by the participant and/or collateral informant; b) evidence by clinical evaluation or history of memory or other cognitive decline; c) Global Clinical Dementia Rating scale of 0.5 (18); d) below expected performance on delayed recall of the HVLT-R (23) or delayed paragraph recall from the National Alzheimer’s Coordinating Center -Unified Data Set (NACC-UDS) (25) as measured by a score that is 1.5 SD or more below the mean using age, education, and language-related norms.

Cognitively Normal Group (n=39)

Participants were classified as cognitively normal if all of the following criteria were met: a) no subjective cognitive complaints made by the participant and a collateral informant; b) no evidence by clinical evaluation or history of memory or other cognitive decline after an extensive interview with the participant and an informant; c) Global CDR score of 0; d) performance on all traditional neuropsychological tests (e.g.: Category Fluency (26), Trails A and B (27), WAIS-IV Block Design subtest (28) was not more than 1.0 SD below normal limits for age, education, and language group.

Loewenstein-Acevedo Scales for Semantic Interference and Learning, Brief Computerized Version (LASSI-BC)

The LASSI-BC was not used for diagnostic determination in this study. This computerized cognitive stress test is a novel computerized measure that is briefer than the paper-and-pencil LASSI-L, taking approximately 10 to 12 minutes to complete. The LASSI-BC contains the elements of the original LASSI-L which demonstrated the greatest differentiation between aMCI, PreMCI and CN older adults in previous studies. For example, free recall preceding the cued recall trials of the LASSI-L added time to the administration but was never as effective as cued recall in distinguishing among diagnostic groups. Developed in collaboration with the University of Miami Department of Computer Science, the LASSI-BC is a remotely accessible test available in both English and Spanish. As a web application, it can be run on devices that can run Google Chrome, including desktop computers, laptops, tablets, or even smartphones. While the LASSI-BC is a fully self-administered test with all verbal responses recorded and scored by the computer, for the purposes of this validation study, a trained study team member was present for each administration to systematically record responses, which provided a double check on the accuracy of data. The LASSI-BC utilizes Google Cloud Speech API , which has been successfully implemented for use with older adults. The test leverages Google Cloud’s Speech to Text software in conjunction with a backup lexicon for understanding the participants’ spoken words. The lexicon is designed to account for variations in participant’s pronunciation by allowing for words that the computer “mishears” to serve as alternatives to the actual word being spoken. Lexicons were chosen based on observations from participants during the test.
Upon initiating the examination, the participant is instructed in both audio and visual formats. They will see 15 words belonging to one of three semantic categories: fruits, musical instruments, or articles of clothing (five words per category). The words are then individually presented on the screen and audio for a 6-second interval. This presentation facilitates optimal encoding and storage of the to-be-remembered information. Further, this instruction style has been easily understood and accepted by older adults during pilot studies in the course of developing the LASSI-BC. After the computer presents all 15 words, participants are presented with each category cue (e.g., fruits) and asked to recall the words that belonged to that category. Participants are then presented with the same target stimuli for a second learning trial with subsequent cued recall to strengthen the acquisition and recall of the List A targets. The exposure to the semantically related list (i.e., List B) is then conducted in the same manner as exposure to List A. List B consists of 15 words different from List A, all of which belong to each of the three categories used in List A (i.e., fruits, musical instruments, and articles of clothing). Following the presentation of the List B words, the person is asked to recall each of the List B words that belonged to each of the categories. List B words are presented again, followed by a second category-cued recall trial. Finally, to assess retroactive semantic interference, participants are asked to free recall the original List A words. Primary measures used in this study are the second cued recall score for List A (maximum learning), first cued recall score for List B (susceptibility to proactive semantic interference), second cued recall of List B (failure to recover from proactive semantic interference), and the third cued recall of List A (retroactive semantic interference). In addition, we evaluated the novel ratio used with the LASSI-L, that takes into account the percentage of intrusion errors (PIE) as a function of total responses on subscales that measure proactive semantic interference and the failure to recover from proactive semantic interference. Specifically, the ratio is denoted as follows: Total Intrusion Errors/ (Total Intrusion Errors + Total Correct Responses) for LASSI-BC Cued B1 (a measure of susceptibility to proactive semantic interference) and LASSI-BC Cued B2 recall (a measure of recovery from proactive semantic interference).



The computerized version of the LASSI-BC had psychometric properties that compared favorably to the test-retest reliabilities obtained on the original paper-and-pencil LASSI-L (7). As depicted in Table 1, CN (n=39) and aMCI (n=25) groups did not differ in terms of age, sex, or language of evaluation. Individuals diagnosed as aMCI, although well educated (Mean =14.26; SD=3.5), had less educational attainment relative to their cognitively normal counterparts (Mean =16.32; SD=2.3). As expected, aMCI participants also had lower mean MMSE scores (Mean =26.04; SD=2.3).

Table 1. Demographic Characteristics and Computerized LASSI-BC Scores among Participants who are Cognitively Normal and with Amnestic Mild Cognitive Impairment


Test-retest reliability

Test-retest reliability data was obtained on a subset of 15 older adults diagnosed with aMCI using Petersen’s criteria (24) for each of the LASSI-BC subscales. The mean age was 73.4 (SD=6.3); education 15.4 (SD=3.6); and the mean MMSE score for this group was 26.6 (SD=2.2). These individuals (60% primary English-speakers and 60% female) were administered the LASSI-BC on two occasions, within a 4 to 39-week interval (Mean =13.9.; SD=10.6 weeks). In our pilot work, we found robust test-retest correlations ranging from 0.55 to 0.721 on the subscales that have shown to be the most sensitive measures of cognitive decline in the original paper-and-pencil version. In this study, test-retest comparisons were conducted for Cued Recall A2 (measures maximum learning), Cued Recall B1 (measures proactive semantic interference), and Cued Recall B2 (measures the failure to recover from proactive semantic interference). One-tailed Pearson Product Moment Correlation Coefficients were obtained given the directional hypotheses concerning test-retest relationships. High, statistically significant test-retest reliabilities were obtained for Cued A2 Recall (r=.726; p<.001); Cued Recall B1 (r=.529; p=0.021); Cued Recall B2 (r=.555; p=0.016).

Discriminant validity

As depicted in Table 1, LASSI-BC scales sensitive to maximum learning (Cued A2), vulnerability to proactive semantic interference (Cued B1) and the failure to recover from proactive semantic interference (Cued B2) were statistically significant in discriminating between older adults with amnestic MCI and cognitively normal counterparts. These results were identical when demographic variables such as education were entered in the model as covariates
We then calculated areas under the Receiver Operating Characteristic (ROC) curve for LASSI-BC correct responses as well as the PIE indices for Cued B1 and Cued B2 subscales. We selected these measures a priori given that performance on these specific subscales have traditionally been the most discriminant measures on the paper-and-pencil form of the LASSI-L.
As shown in Table 2, an optimal cut-point of 5 by Youden’s criteria on correct responses for Cued Recall B1, yielded a sensitivity of 84.6% and a specificity of 86.8%. An optimal cut-point of 9 by Youden’s criteria on correct responses provided on Cued Recall B2, yielded statistically significant areas under the ROC curve of .868 (SE=0.88) and .824 (SE=.051), respectively.

Table 2. Classification of aMCI versus Cognitively Normal Participants on the LASSI-BC


We subsequently examined an optimal cut-point for PIE on the Cued Recall B1 and the Cued Recall B2 subscales. For PIE on Cued Recall B1, the area under the ROC was .879 (SE=.06) with a sensitivity of 92.9% and specificity of 80%, respectively using an optimal cut-point of .2540. For PIE on Cued Recall B2, the area under the ROC was .801 (SE=.07), using an optimal cut-point of .2159, which yielded a sensitivity of 78.6% and specificity of 68.0%. We selected these specific subscales because they have shown to be the strongest predictors of aMCI in the paper-and-pencil form of the LASSI-L.
We subsequently entered the statistically significant LASSI-BC subscales (Cued Recall B1 and Cued Recall B2) into a stepwise logistic regression. As seen in Table 3, the first variable to enter the logistic regression model was PIE on Cued B1 [B=6.86 (SE=1.67) Wald=17.07, p<0.001)]. On the second step of the logistic regression model, correct responses on Cued Recall B2 entered the model [B=-.34 (SE=.128), Wald= 17.1 (p=.008)]. Combining PIE Cued Recall B1 and correct responses on Cued Recall B2, yielded an overall sensitivity of 80% and specificity of 89.7%. It should be noted logistic regression weighs overall classification in a manner that favors the largest diagnostic group (in this case CN participants). Nonetheless, ROC and stepwise regression models yielded similar results indicating excellent discriminative ability.
In sum, our findings support that the LASSI-BC has equal or better psychometric properties than the original paper-and-pencil LASSI-L and demonstrates that computerized administration is both feasible, well accepted, and has excellent discriminant properties.

Table 3. Step-wise Logistic Regression Using Proactive Semantic Interference Measures on the Computerized LASSI-BC



The present study was designed to examine the psychometric properties of the LASSI-BC, the brief computerized version of the LASSI-L, a cognitive stress test that utilizes a novel cognitive assessment paradigm based on semantic interference in memory. In studies conducted in the United States and abroad, the LASSI-L has shown great utility in detecting cognitive changes among individuals during the preclinical and prodromal stages of AD (4, 29) and has been found to be appropriate for use among diverse ethnic/cultural and language groups (11, 30, 12). The paradigm that this measure employs is unique in that it explicitly and from the outset organizes the examinee’s learning around specific semantic categories, which promotes active encoding, reduces the use of individualized learning strategies that can help or hinder performance, increases depth of initial learning, and is designed to tap an individual’s vulnerability to semantic interference.
The current investigation examined all salient subscales of the LASSI-BC, which were selected based on previous work with the paper-and-pencil versions. The computerized version evidenced good test-retest reliability for participants diagnosed with aMCI. Scores on all LASSI-BC subscales were higher for cognitively normal older adults, relative to aMCI participants. In addition, high levels of discriminant validity were obtained in differentiating aMCI from cognitively normal groups based on ROC analyses and logistic regression.
A potential limitation of this first validation study is that we employed modest numbers of participants who were tested in either English or Spanish on the LASSI-BC. Although, our overall findings were highly significant and the paper-and-pencil LASSI-L has been validated in different languages (i.e.- Spanish speakers in Argentina, Spanish speakers in Spain, Spanish speakers from Mexico, etc.) and with different ethnic/cultural groups (European Americans, Hispanics and African Americans), such future comparisons should be made with the LASSI-BC. Further, additional studies with the LASSI-BC will include evaluating the diagnostic utility of this computerized cognitive stress test to differentiate older adults earlier on the preclinical continuum of AD, and relate performance to biomarkers of AD pathology, as well as compare it to other traditional and widely used cognitive measures in the field.
There has been an increase in the number of computerized tests developed including the CogState (31) and the Cognition Battery from the NIH Toolbox (16), but limitations exist. For example, one of the most widely-used computerized cognitive batteries for the assessment of MCI is the CogState. As part of the Mayo Clinic Study on Aging, Mielke and associates (32) administered the CogState to eighty-six participants diagnosed with MCI who were found to have worse performance than cognitively healthy individuals; however, it is likely that individuals classified as MCI ranged from early states of MCI to late MCI, the latter of which is more cognitively similar to early dementia in terms of neuropsychological test performance, limiting evidence that this measure in sensitive to preclinical cognitive change. Further, the authors noted that their results are not generalizable to other ethnicities due to the demographic makeup of the region (Minnesota, USA). Another study conducted by Mielke and colleagues (33) aimed to examine performance on the CogState with neuroimaging biomarkers (MRI, FDG PET, and amyloid PET) among cognitively normal participants aged 51-71; however, only weak associations were found between CogState subtests and biomarkers of neurodegeneration.
With the rapidly aging population, early detection of cognitive decline in individuals at risk for AD and related disorders has become a global priority. Accurately identifying at risk individuals through the detection and monitoring of subtle, albeit sensitive cognitive changes that transpire early in the disease course is an important initiative and computerized cognitive outcome measures have the potential to greatly reduce burden for participants, clinical researchers and clinicians.
The development of computerized cognitive tests for older adults has significantly increased during the past decade. In fact, available systematic reviews have identified more than a dozen computerized measures designed to detect dementia or MCI (34, 35, 36). Moreover, the use of computerized assessments with older adults has been found to be feasible and reliable (37, 38). A recent meta-analysis has shown relatively good diagnostic accuracy, and authors further concluded that their performance distinguishing individuals with MCI and dementia is comparable with traditional paper-pencil neuropsychological measures (35). It is anticipated that as technology advances, clinical trials will include validated computerized testing to sensitively capture cognitive performance, particularly in large-scale secondary prevention efforts (39). The impact of this technological advancement in computerized, web-based cognitive testing has the potential to facilitate remote deliverability, allow for real-time data entry, improves standardization, and reduces administration and scoring errors. Moreover, computerized assessment can more readily monitor longitudinal cognitive changes for each individual, facilitating a precision-based approach. It is critical; however, that emerging cognitive tests move beyond simply computerizing outdated, insensitive cognitive paradigms and instead invest in the development and validation of cognitive paradigms that are sensitive and specific to early cognitive breakdowns that occur during the preclinical stages of AD. These too should exhibit sensitivity to biomarkers of AD (e.g., amyloid load, tau deposition, and neurodegeneration in AD-prone regions). Doing so may address some of the most critical challenges facing clinical trials including proper selection of at-risk participants, and monitoring meaningful cognitive change over time.

Funding: This research was funded by the National Institute of Aging Grant 1 R01 AG047649-01A1 (David Loewenstein, PI), 1 R01 AG047649-01A1 (Rosie Curiel Cid, PI) 5 P50 AG047726602 1Florida Alzheimer’s Disease Research Center (Todd Golde, PI), 8AZ. The sponsors had no role in the design and conduct of the study; in the collection analysis, and interpretation of data; in the preparation of the manuscript; or in the review or approval of the manuscript.

Ethical standard: This research study was conducted in alignment with the Declaration of Helsinki and through the approval of the University of Miami Institutional Review Board.
Conflict of interest: Drs. Curiel and Loewenstein have intellectual property used in this study.


1. Jack Jr CR, Bennett DA, Blennow K, et al. NIA-AA research framework: toward a biological definition of Alzheimer’s disease. Alzheimer’s & Dementia. 2018 Apr;14(4):535-62.
2. Harvey PD, Cosentino S, Curiel R, et al. Performance-based and observational assessments in clinical trials across the Alzheimer’s disease spectrum. Innovations in clinical neuroscience. 2017 Jan; 14(1-2):30.
3. Edmonds EC, Delano-Wood L, Galasko DR, et al. Subtle cognitive decline and biomarker staging in preclinical Alzheimer’s disease. Journal of Alzheimer’s disease. 2015 Jan 1;47(1):231-42.
4. Loewenstein DA, Curiel RE, Duara R, et al. Novel cognitive paradigms for the detection of memory impairment in preclinical Alzheimer’s disease. Assessment. 2018 Apr25;(3):348-59.
5. Brooks L, Loewenstein D. Assessing the progression of mild cognitive impairment to Alzheimer’s disease: current trends and future directions. Alzheimer’s Research & Therapy. 2010;2(28):28-28.
6. Crocco E, Curiel RE, Acevedo A, et al. An evaluation of deficits in semantic cueing and proactive and retroactive interference as early features of Alzheimer’s disease. The American Journal of Geriatric Psychiatry. 2014 Sep 1;22(9):889-97.
7. Curiel RE, Crocco E, Acevedo A, et al. A new scale for the evaluation of proactive and retroactive interference in mild cognitive impairment and early Alzheimer’s disease. Aging. 2013;1(1):1000102.
8. Loewenstein DA, Curiel RE, Greig MT, et al. A novel cognitive stress test for the detection of preclinical Alzheimer disease: discriminative properties and relation to amyloid load. The American Journal of Geriatric Psychiatry. 2016 Oct 1;24(10):804-13.
9. Matías-Guiu JA, Curiel RE, Rognoni T, Valles-Salgado M, Fernández-Matarrubia M, Hariramani R, Fernández-Castro A, Moreno-Ramos T, Loewenstein DA, Matías-Guiu J. Validation of the Spanish version of the LASSI-L for diagnosing mild cognitive impairment and Alzheimer’s disease. Journal of Alzheimer’s Disease. 2017 Jan 1;56(2):733-42.
10. Rosselli M, Loewenstein DA, Curiel RE, et al. Effects of bilingualism on verbal and nonverbal memory measures in mild cognitive impairment. Journal of the International Neuropsychological Society. 2019 Jan;25(1):15-28.
11. Capp KE, Curiel Cid, RE, Crocco EA, et al. Semantic Intrusion Error Ratio Distinguishes Between Cognitively Impaired and Cognitively Intact African American Older Adults. Journal of Alzheimer’s Disease. 2019 Dec 23(Preprint):1-6.
12. Matias-Guiu JA, Cabrera-Martín MN, Curiel RE, et al. Comparison between FCSRT and LASSI-L to detect early stage Alzheimer’s disease. Journal of Alzheimer’s Disease. 2018 Jan 1;61(1):103-11.
13. Crocco, E, Curiel RE, Grau, G. Percentage of intrusion errors predicts patterns of cognitive change in older adults. (Under Review). Journal of Alzheimer’s Disease.
14. Beaumont JL, Havlik R, Cook KF, et al. Norming plans for the NIH Toolbox. Neurology. 2013;80(11 Suppl 3):S87–S92.
15. Saxton J, Morrow L, Eschman A, et al. Computer assessment of mild cognitive impairment. Postgraduate medicine. 2009 Mar 1;121(2):177-85.
16. Weintraub S, Dikmen SS, Heaton RK, et al. Cognition assessment using the NIH Toolbox. Neurology. 2013 Mar 12;80(11 Supplement 3):S54-64.
17. Parsons TD, Courtney CG, Arizmendi BJ, et al. Virtual reality stroop task for neurocognitive assessment. InMMVR 2011 Feb 16 (pp. 433-439).
17. Morris JC. Clinical dementia rating: a reliable and valid diagnostic and staging measure for dementia of the Alzheimer type. International psychogeriatrics. 1997 Dec;9(S1):173-6.
18. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”: a practical method for grading the cognitive state of patients for the clinician. Journal of psychiatric research. 1975 Nov 1;12(3):189-98.
19. Arango-Lasprilla JC, Rivera D, Aguayo A, et al. Trail making test: Normative data for the Latin American Spanish speaking adult population. NeuroRehabilitation. 2015 Jan 1;37(4):639-61.
20. Arango-Lasprilla JC, Rivera D, Garza MT, et al. Hopkins verbal learning test–revised: Normative data for the Latin American Spanish speaking adult population. NeuroRehabilitation. 2015 Jan 1;37(4):699-718.
21. Benson G, de Felipe J, Sano M. Performance of Spanish-speaking community-dwelling elders in the United States on the Uniform Data Set. Alzheimer’s & Dementia. 2014 Oct;10:S338-43.
22. Peña-Casanova J, Quinones-Ubeda S, Gramunt-Fombuena N,et al. Spanish Multicenter Normative Studies (NEURONORMA Project): norms for verbal fluency tests. Archives of Clinical Neuropsychology. 2009 Jun 1;24(4):395-411.
23. Brandt J. The Hopkins Verbal Learning Test: Development of a new memory test with six equivalent forms. The Clinical Neuropsychologist. 1991 Apr 1;5(2):125-42.
24. Petersen RC. Mild cognitive impairment as a diagnostic entity. Journal of internal medicine. 2004 Sep;256(3):183-94.
25. Beekly DL, Ramos EM, Lee WW, Deitrich WD, Jacka ME, Wu J, Hubbard JL, Koepsell TD, Morris JC, Kukull WA. The National Alzheimer’s Coordinating Center (NACC) database: the uniform data set. Alzheimer Disease & Associated Disorders. 2007 Jul 1;21(3):249-58.
26. Binetti G, Magni E, Cappa SF, et al. Semantic memory in Alzheimer’s disease: an analysis of category fluency. Journal of Clinical and Experimental Neuropsychology. 1995 Feb 1;17(1):82-9.
27. Reitan RM. Validity of the Trail Making Test as an indicator of organic brain damage. Perceptual and motor skills. 1958 Dec;8(3):271-6.
28. Wechsler D. Wechsler adult intelligence scale–Fourth Edition (WAIS–IV). San Antonio, TX: NCS Pearson. 2008;22(498):816-27.
29. Crocco EA, Loewenstein DA, Curiel RE, et al. A novel cognitive assessment paradigm to detect Pre-mild cognitive impairment (PreMCI) and the relationship to biological markers of Alzheimer’s disease. Journal of psychiatric research. 2018 Jan 1;96:33-8.
30. Curiel Cid RE, Loewenstein DA, Rosselli M, et al. A cognitive stress test for prodromal Alzheimer’s disease: Multiethnic generalizability. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring. 2019 Dec;11(C):550-9.
31. Darby D, Collie A, McStephen M, Maruff P. Reliable detection of asymptomatic longitudinal cognitive decline in healthy community dwelling volunteers. Journal of the American Geriatrics Society. 2004 Apr;52.
32. Mielke MM, Machulda MM, Hagen CE, Edwards KK, Roberts RO, Pankratz VS, Knopman DS, Jack Jr CR, Petersen RC. Performance of the CogState computerized battery in the Mayo Clinic Study on Aging. Alzheimer’s & Dementia. 2015 Nov 1;11(11):1367-76.
33. Mielke MM, Weigand SD, Wiste HJ, Vemuri P, Machulda MM, Knopman DS, Lowe V, Roberts RO, Kantarci K, Rocca WA, Jack Jr CR. Independent comparison of CogState computerized testing and a standard cognitive battery with neuroimaging. Alzheimer’s & Dementia. 2014 Nov 1;10(6):779-89.
34. Zygouris S, Tsolaki M. Computerized cognitive testing for older adults: a review. American Journal of Alzheimer’s Disease & Other Dementias®. 2015 Feb;30(1):13-28.
35. Chan JY, Kwong JS, Wong A, Kwok TC, Tsoi KK. Comparison of computerized and paper-and-pencil memory tests in detection of mild cognitive impairment and dementia: A systematic review and meta-analysis of diagnostic studies. Journal of the American Medical Directors Association. 2018 Sep 1;19(9):748-56.
36. De Roeck EE, De Deyn PP, Dierckx E, Engelborghs S. Brief cognitive screening instruments for early detection of Alzheimer’s disease: a systematic review. Alzheimer’s research & therapy. 2019 Dec 1;11(1):21.
37. Wild K, Howieson D, Webbe F, Seelye A, Kaye J. Status of computerized cognitive testing in aging: a systematic review. Alzheimer’s & Dementia. 2008 Nov 1;4(6):428-37.
38. Pankratz VS, Roberts RO, Mielke MM, Knopman DS, Jack CR, Geda YE, Rocca WA, Petersen RC. Predicting the risk of mild cognitive impairment in the Mayo Clinic Study of Aging. Neurology. 2015 Apr 7;84(14):1433-42.
39. Buckley RF, Sparks KP, Papp KV, Dekhtyar M, Martin C, Burnham S, Sperling RA, Rentz DM. Computerized cognitive testing for use in clinical trials: a comparison of the NIH Toolbox and Cogstate C3 batteries. The journal of prevention of Alzheimer’s disease. 2017;4(1):3.


X. Fu1,*, W. Yu2,*, M. Ke2, X. Wang1, J. Zhang1, T. Luo1, P.J. Massman3,4, R.S. Doody3, Y. Lü1,*

1. Department of Geriatrics, The First Affiliated Hospital of Chongqing Medical University, Chongqing 400016, China; 2. Institute of Neuroscience, Chongqing Medical University, Chongqing 400016, China; 3. Department of Neurology, Baylor College of Medicine, Houston, TX USA at the time this work was done. Now Genentech/Roche, Basel, Switzerland; 4. Department of Psychology, University of Houston, Houston, TX USA; *Authors contributed equally and are co-first authors of the study.

Corresponding Authors: Prof. Yang Lü, 1 Youyi Road, Yuzhong District, Chongqing 400016, China, Tel: +86-23-89011622, Fax: +86-23-68811487, E-mail:

J Prev Alz Dis 2020;
Published online December 21, 2020,



BACKGROUND: A specialized instrument for assessing the cognition of patients with severe Alzheimer’s disease (AD) is needed in China.
Objectives: To validate the Chinese version of the Baylor Profound Mental Status Examination (BPMSE-Ch).
Design: The BPMSE is a simplified scale which has proved to be a reliable and valid tool for evaluating patients with moderate to severe AD, it is worthwhile to extend the use of it to Chinese patients with AD.
Setting: Patients were assessed from the Memory Clinic Outpatient.
Participants: All participants were diagnosed as having probable AD by assessment.
Measurements: The BPMSE was translated into Chinese and back translated. The BPMSE-Ch was administered to 102 AD patients with a Mini-Mental State Examination (MMSE) score below 17. We assessed the internal consistency, reliability, and construct validity between the BPMSE-Ch and MMSE, Severe Impairment Battery (SIB), Global Deterioration Scale (GDS-1), Geriatric Depression Scale(GDS-2), Instrumental Activities of Daily Living (IADL), Physical Self-Maintenance Scale (PSMS), Neuropsychiatric Inventory (NPI) and Clinical Dementia Rating (CDR).
Results: The BPMSE-Ch showed good internal consistency (α = 0.87); inter-rater and test-retest reliability were both excellent, ranging from 0.91 to 0.99. The construct validity of the measure was also supported by significant correlations with MMSE, SIB. Moreover, as expected, the BMPSE-Ch had a lower floor effect than the MMSE, but a ceiling effect existed for patients with MMSE scores above 11.
Conclusions: The BPMSE-Ch is a reliable and valid tool for evaluating cognitive function in Chinese patients with severe AD.

Key words: Alzheimer’s disease, Baylor Profound Mental Status Examination, Chinese version, severe dementia, validation.

Abbreviations: AD: Alzheimer’s disease; ADAS-cog: Alzheimer’s Disease Assessment Scale-Cognitive section; ANOVA: A one-way analysis of variance; BPMSE: Baylor Profound Mental Status Examination; BPMSE-Ch: Chinese version of the Baylor Profound Mental Status Examination; BPMSE-Ch-cog: Cognition subscale of Chinese version of the Baylor Profound Mental Status Examination; BPMSE-Ch-behav: Behavior subscale of Chinese version of the Baylor Profound Mental Status Examination; CDR: Clinical Dementia Rating; FAST: Functional Assessment Staging; GDS-1: Global Deterioration Scale; GDS-2: Geriatric Depression Scale: IADL, Instrumental Activities of Daily Living; MMSE: Mini-Mental State Examination; NPI: Neuropsychiatric Inventory; PSMS: physical self-maintenance scale; SIB: Severe Impairment Battery.



Alzheimer’s disease (AD) is a common neurodegenerative disorder among mainly elderly persons worldwide. The manifestations of AD include deterioration in cognition, memory and activities of daily living. It is usually accompanied by behavioral and psychological symptoms (1).
Currently, China is facing serious issues related to having an aging population. Persons aged 60 or older account for 17.3% of the total population (2). The prevalence of all-cause dementia over age 65 is about 6% in China, and AD makes up about 65% of all cases (3, 4). The rough prevalence of AD in China has reported to ranges from 7 per 1000 people to 66 per 1000 individuals (5). In a population-based cross-sectional survey, 10276 residents aged 65 year or older were drawn from Beijing (northern-eastern), Zhengzhou (northern-central), Guiyang (southern-western) and Guangzhou (southern-eastern). This survey showed that the prevalence of AD was 3.21% in a total of 10276 residents (6). Despite the fact that China has the relatively high AD prevalence, few studies of AD were conducted to research excellent methods for AD diagnosing and evaluating.
It seems unquestionable that AD is gradually evolving into a crucial social problem and presents a major challenge for health-care in China. However, awareness of AD and dementia in general is inadequate in China, leading to delayed diagnosis and initiation of treatment (7, 8).Therefore, many patients do not get evaluated until moderate to severe stages of the disease (9, 10). Moreover, once these patients present for an evaluation, tools to assess them are limited (11). Hence, better instruments are needed for the accurate assessment of patients with advanced AD.
A variety of neuropsychological and functional measures have been utilized to assess mental status and dementia severity both cross-sectionally and longitudinally. Frequently-used instruments include the Mini-Mental Status Examination (MMSE) (12), Severe Impairment Battery (SIB) (13), Alzheimer’s Disease Assessment Scale-Cognitive section (ADAS-Cog) (14), Geriatric Deterioration Scale (GDS-1) (15), Functional Assessment Staging Tool(FAST) (16) and Clinical Dementia Rating (CDR) (17). However, these scales show some limitations in patients with moderate to severe AD. The MMSE and ADAS-cog are not optimal for evaluating patients with severe AD because both contain a lot of verbal information and; therefore, the results may be confounded by language disorders and/or low level of education. SIB is a suitable tool to evaluate patients with severe dementia. However, this test takes more than 30 minutes to administer, which often exceeds the attention capacity of most patients with severe AD (18). The NPI is usually used to evaluate neuropsychiatric symptoms, but it is largely dependent on the description from caregivers (19). Overall, it is clear that a convenient and effective assessment instrument for measuring cognitive function in patients with severe AD is highly needed.
The Baylor Profound Mental State Examination (BPMSE) developed by Doody RS et al, is a simplified scale which has proved to be a reliable and valid tool for evaluating patients with moderate to severe AD (20). And in Doody’s study, European American accounted for about 82% of the original population. Thus, it is worthwhile to extend the use of the BPMSE to Severely demented patients from different cultural backgrounds. To date, there have been three translated versionsof the BPMSE, including Korean, Danish and Spanish (21-23). A study of the Korean version has shown that the BPMSE is a rapid, easy and valid scale for measuring cognitive function in patients with moderate to severe AD, particularly in patients with MMSE below 12. Similarly, a study utilizing the Danish version indicated that the BPMSE is a stable and strong instrument, and was recommended as an appropriate measure of dementia severity in patients with more sever impairment. Adaptation of the Spanish version revealed that BPMSE that the BPMSE is a useful tool for assessing cognitive function, even in daily medical practice focusing on patients with severe AD.
In China, there is no applicable scale for assessing patients with severe AD. Therefore, the aim of our study was to develop a Chinese version of the BPMSE (BPMSE-Ch) and to evaluate the psychometric properties of this version in Chinese patients with AD.




The original version of BPMSE consists of three parts, including the cognition subscale which includes 25 questions, the behavior subscale which includes 10 items to rate the presence or absence of behavioral problems, and 2 qualitative observations of language and social interaction. The cognition subscale assesses four areas: language, orientation, attention, and motor skills. The BPMSE total cognition subscale has a score between 0and 25: maximum 5 scores for orientation, 11 scores for language, 4scores for attention and 5 scores for motor skills. BPMSE behavior subscale score has a score between 0 (no behavioral disturbances) and 10 (all behavioral disturbances). In present study, we did not study the 2 qualitative observations about communication and social interactions.
Firstly, the original version of BPMSE was translated into Chinese with Mandarin by two bilingual translators whose mother tongue was Chinese. Then, the two Chinese versions were discussed by our team with gerontologists, a neurologist, a psychologist and an English expert, and the final Chinese version was formulated based on this input. Finally, two other translators of English philology back translated the final Chinese version into English to confirm consistency with the original version.


Patients were recruited from the Memory Clinic, Department of Geriatrics, The First Affiliated Hospital of Chongqing Medical University.

Enrollment criteria

(a) All participants were diagnosed as having probable AD according to National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer’s Disease and Related Disorders Association criteria (NINCDS-ADRDA) (24); (b) Patients with MMSE <17 were included; (c) This study was approved by the Ethical Committee of The First Affiliated Hospital of Chongqing Medical University on human research; (d) Informed consent was obtained from all participants or their family members.

Exclusion criteria

(a) Patients were excluded if they had other neurological or psychiatric disorders or clinically significant medical conditions (e.g., acute infections, cancer, organ failure etc.,); (b) Patients had severely impaired communication abilities (e.g., global aphasia, deafness, blindness, muteness etc.,); (c) Patients had a history of head trauma, sedative drugs use or substance abuse.


The following measures were administered to all enrolled patients: BPMSE-Ch, MMSE, SIB, GDS-1, GDS-2, IADL, PSMS, NPI, and CDR. All tests were given on the same day. Two trained physicians in our clinic administered the BPMSE-Ch to evaluate a subset of enrolled patients consecutively and independently in order to examine inter-rater reliability. Finally, to investigate test-retest reliability, some patients were randomly chosen to be given the BPMSE-Cha second time within 30days of the first administration. It took 5 minutes on average to administer the BPMSE-Ch.

Statistical analyses

Internal consistency was assessed by computing coefficient α. Inter-rater reliability was assessed by correlation and paired t-test of the two scores obtained by different professionals on the same day. And the test-retest reliability was also calculated with correlational and paired t-test analyses using scores obtained on the same patient within 30 days. The correlations between the BPMSE-Ch and other measures including the SIB, MMSE, GDS-1, GDS-2, IADL, PSMS, NPI and CDR were calculated with Pearson correlations in order to evaluate construct validity. In addition, patients were divided into dementia severity groups using the MMSE and CDR, and differences between those groups were analyzed by conducting a one-way analysis of variance (ANOVA) and Scheffé’s test. Statistical analyses were performed with SPSS 20.0 for Windows.



Demographic characteristics and test performances

102 patients (male: 35, female: 67) were included in our study, the mean age of the patients was 77.76, ranging between 64 and 93. The mean years of education was 7.95, ranging from 0 to 16 years. The specific variations were showed in Table 1.

Table 1. Demographic characteristics and Scores on Instruments

Abbreviations: MMSE, Mini-Mental State Examination; BPMSE-Ch-cog, Cognition subscale of Chinse version of the Baylor Profound Mental Status Examination; BPMSE-Ch-behav, Behavior subscale of Chinse version of the Baylor Profound Mental Status Examination; SIB, Severe Impairment Battery; NPI, Neuropsychiatric Inventory; SD, standard deviation.



In our study, the coefficient α which could reflect the inter-correlations for items on the BPMSE-Ch cognition (BPMSE-Ch-cog) subscale, was 0.87. Furthermore, significant correlations were found among all the BPMSE-Ch-cog components, as seen in Table 2. Inter-rater and test-retest reliability were showed in Table 3.

Table 2. Correlations among BPMSE-Ch-cogsubscales

Correlation coefficients by Pearson correlation. * p< 0.001.


Table 3. Inter-rater and test-retest reliability

Abbreviations: BPMSE-Ch-cog, Cognition subscale of Chinse version of the Baylor Profound Mental Status Examination; BPMSE-Ch-behav, Behavior subscale of Chinse version of the Baylor Profound Mental Status Examination.
Correlation coefficients by Pearson correlation. n = Number of patients. All p values <0.001.


52 patients were tested twice by two trained doctors simultaneously and independently to determine the inter-rater reliability. The correlation between two total cognition subscale scores was 0.99 (p < 0.001) and there was no significant difference (paired t (51) = +1.84, p > 0.05) between the two scores (Mean = 0.17, SD = 0.68). The correlation between two behavior subscale scores was 0.92 (p < 0.001).
42 patients were tested twice by a same doctor within 30 day-interval for the test-retest reliability. The test-retest correlation between two total cognition scores was 0.99 (p < 0.001). Similarly, there was no significant difference (paired t (41) = +1.18, p > 0.05) between the two scores obtained at two time points (Mean = 0.14, SD = 0.78). The test-retest correlation between two behavior scores was 0.94 (p < 0.001).


Construct validity of the BPMSE-Ch was showed in Table 4. The correlations between the BPMSE-Ch-cog and MMSE (0.76), SIB (0.78), GDS-1 (-0.26), GDS-2 (0.16), PSMS (-0.26), IADL (-0.36), NPI (-0.41), CDR (-0.54) were calculated by Pearson correlation. The results showed that the construct validity of BPMSE-cog was very good (r=0.78) for SIB and good for MMSE (0.76). In addition, the relationship between BPMSE-Ch behavior subscale (BPMSE-Ch-behav) and NPI was analyzed (0.54, p < 0.001, Table 4).

Table 4. Concurrent validity of BPMSE-Ch

Abbreviations: BPMSE-Ch, Chinse version of the Baylor Profound Mental Status Examination; BPMSE-Ch-cog, Cognition subscale of Chinse version of the Baylor Profound Mental Status Examination; BPMSE-Ch-behav, Behavior subscale of Chinse version of the Baylor Profound Mental Status Examination; MMSE, Mini-Mental State Examination; SIB, Severe Impairment Battery; GDS1, Global Deterioration Scale; GDS2, Geriatric Depression Scale; IADL, Instrumental Activities of Daily Living; PSMS, physical self-maintenance scale; CDR, Clinical Dementia Rating; NPI, Neuropsychiatric Inventory.


Ceiling and floor effects

The relationship between BPMSE-Ch-cog and MMSE was revealed on a scatterplot (Supplementary Figure 1A). The range of 0 to 5 scores on the MMSE corresponded to a substantial range of 2 to 24 scores on BPMSE-Ch-cog, indicating that the BPMSE-Ch had no floor effect. In addition, it was found that patients scoring 12 to 16 on MMSE had the BPMSE-Ch-cog scores ranging from 20 to 25 (Mean: 23.08, SD: 1.08, Table 5). This demonstrated that BPMSE-Ch showed a ceiling effect among patients who were at a relative moderate level of dementia.


The relationship between BPMSE-Ch-cog and SIB scores is displayed (Supplementary Figure 1B). The relatively highR2=0.61 indicated that BPMSE-Ch-cog showed a strong association with the SIB, which demonstrated that the BPMSE-Ch was a sensitive tool for assessing patients with severe AD.

BPMSE-Ch-cog score stratified by MMSE levels

Table 5 presented that BPMSE-Ch-cog differentiated all the enrolled patients belonging to different severity groups according to the MMSE scores (F = 56.7, p <0.001). Patients in the MMSE Group 1 (range 16-12) had a BPMSE-Ch score of 23.08 ± 1.08, patients in the MMSE Group 2 (range 7-11) had a BPMSE-Ch score of 21.25 ± 3.53, and patients in the MMSE Group 3 (range 0-6) had a further reduced BPMSE-Ch score of 12.50 ± 6.69. From the results of Table 5, it was found that the differences in total BPMSE-Ch-cog score as well as in its four subcomponents scores between the Group 2 and Group 3 was significant (p < 0.001).

Table 5. Three severity groups according to the MMSE

Abbreviations: MMSE, Mini-Mental State Examination; BPMSE-Ch-cog, Cognition subscale of Chinse version of the Baylor Profound Mental Status Examination; SD, standard deviation; n = Number of patients. One-way ANOVA test. NS = Nonsignificant; 1. By Scheffé’s analysis


BPMSE-Ch-cog score stratified by CDR levels

BPMSE-Ch-cog differentiates the patients into different groups according to the CDR stage (F = 16.0, p < 0.001) (Supplementary Table 1). It was observed that the mean BPMSE-Ch-cog and subcomponents scores declined as the CDR stage increased (Supplementary Table 1). Furthermore, at Group 1 (CDR = 0.5), the total score of BPMSE-Ch-cog ranged from 23 to 25(Mean = 24.33, SD = 1.15); at Group 2 (CDR = 1), the total score of BPMSE-Ch-cog ranged from 19 to 25 (Mean = 22.86, SD = 1.42); at Group 3 (CDR = 2), the total BPMSE-Ch-cog score ranged from 2 to 25 (Mean = 19.82, SD = 5.59); at Group 4 (CDR = 3), the total BPMSE-Ch-cog score ranged from 2 to 24 (Mean = 12.50, SD = 7.29). It was observed that as the CDR stage increased, the corresponding range of BPMSE-Ch-cog became wide. Moreover, it was also shown that significant differences of total BPMSE-Ch-cog score and subcomponents scores existed between Group 3 and Group 4(Supplementary Table 1). All above suggested that BPMSE-Ch measured in a way different from CDR, and could differentiate levels of cognition at high CDR stages. Discussion The present study shows that BPMSE-Ch is a reliable, stable and valid instrument for assessing cognition in patients with severe AD. Internal consistency is robust, inter-rater reliability is near-perfect for both the BPMSE-Ch-cog and BPMSE-Ch-behav subscales, and test-retest reliability is also excellent. Furthermore, excellent construct validity was found referring to significant correlations with SIB (r=0.78), MMSE (r=0.76). These findings are consistent with the results of previous adoptions of Korean, Spanish, and Danish versions of the BPMSE. BPMSE-Ch-cog scores were strongly associated with MMSE, SIB ratings, indicating that the BPMSE-Ch-cog can differentiate well among patients with AD with differing degrees of cognitive impairment, particularly in the more severe end of the dementia spectrum, which of course is its primary intended use. In this regard, BPMSE-Ch-cog do not display floor effects in severely demented patients, as measured by the MMSE. Also, BPMSE-Ch-cog scores are strongly associated with SIB scores (while displaying a lower floor than the SIB), and its administration time is much shorter (only 5 minutes on average versus 30 minutes for the SIB). It further suggests that BPMSE-Ch is an efficient tool. Relative low correlations are also shown between BPMSE-Ch-cog scores and PSMS and IADL functional scores, demonstrating that the BPMSE-Ch can only partly measure cognitive abilities relevant to the abilities needed to function in daily life. We thought the possible reason is that the most enrolled patients would have reached maximum impairment of activities of daily living. It supposed that a certain degree of ceiling effects existed in IADL and PSMS tests. AlthoughGDS-1 is an available tool used to evaluate not only cognition but also the abilities to maintain daily life, participation in adverse activities and it is useful for the severe AD cases (25-27), it is a synthetic grade evaluation tool. The forced-choice format would place most enrolled patients into high stages. This might be the reason that the correlation between BPMSE-Ch and GDS-1 is low. Behavioral and psychological symptoms of dementia (BPSD) in patients with Alzheimer’s disease have a strong correlation with cognitive impairment and impairment in activities of daily living. NPI is a common tool for BPMSD evaluating. The BPMSE-Ch-behav selectively focused on disruptive behaviors. In this study, it has been found that there is a moderate correlation between BPMSE-Ch-behavand NPI. While the NPI is obtained by questions to the primary caregiver and is a complex and time-consuming process. Therefore, it indicated that BPMSE-Ch is also a relative practicable instrument to evaluate the behavioral and psychological symptoms in patients with severe dementia. The correlation between BPMSE-Ch-cog and GDS-2 is not significant (r = 0.16, p > 0.001). There are two possible reasons. Firstly, BPMSE-Ch-cog does not involve questions directed against depressive symptoms and is not intended to evaluate for depression. Secondly, it has been reported that patients with moderate-severe AD have relatively low GDS-2 scores (28), which is similar to our study. It suggests that patients with moderate-severe AD have no obvious depression symptoms. In our study, the highest GDS-2score seen was 24; therefore, GDS-2 sometimes shows a good complementary assessment for depression. Because the BPMSE measures clinical features distinct from the GDS-2 the absence of correlation is not surprising.
Regarding its suitability for use with severely impaired patients, it has been observed that the BPMSE-Ch-cog differentiates well between patients with MMSE scores 0-6 and those with MMSE scores 7-11, but not as well between patients with scores of 12-16 and those with scores 7-11. This indicates that the BPMSE-Ch, like its versions in other languages, is most appropriate to use with patients who are more severely impaired (with MMSE score of 11 or below). Similarly, analyses of patients in different CDR stages reveals that the total BPMSE-Ch score and subcomponent scores differ significantly between patients in CDR stage 2 versus those in CDR stage 3, and patients in both of these more severely impaired CDR stages exhibited a wide range of scores, with substantial variability. These results lend further support to the use of the BPMSE-Ch with severely impaired patients.
In conclusion, the BPMSE-Ch is a convenient, stable, reliable and valid scale to assess cognition in patients with moderate-severe AD, and is most appropriately used with patients who have MMSE scores 11 or below. And in future work, we should popularize the BPMSE-Ch in other areas of China including rural areas to research the properties about BPMSE. We believe that it would be beneficial for this instrument to be widely used for evaluating cognitive functioning of patients with severe AD in China.


Acknowledgments: Funding Information: This study was supported by grants from National Key R&D Program of China (2018YFC2001700), General Project of Technological Innovation and Application Development of Chongqing Science & Technology Bureau (cstc2019jscx-msxmX0239), Key project of Social undertakings and people’s livelihood security of Chongqing Science & Technology Commission (cstc2017shms-zdyfX0009) and Postgraduate Research Innovation Project of Chongqing(CYS16122), Particularly, we greatly thank Dr. Sergio Salmerón (Department of Geriatrics, Hospital General de Villarrobledo, Albacete, Spain) for the assistance in making a translation of BPMSE.

Conflict of Interest: The authors declare that they have no potential competing interests

Ethics approval and consent to participate: The study was approved by the Ethics Committee of The First Affiliated Hospital of Chongqing Medical University and has been performed in accordance with the ethical standards laid down in the Declaration of Helsinki and its later amendments.

Authors’ Contributions: Rachelle S. Doody and Yang Lü designed the study. Xue Fu, Weihua Yu and Yang Lü collected the data and wrote the paper. Yang Lü, Paul J. Massman and Rachelle S. Doody revised the manuscript: Xia Wang, Jia Zhang and Tao Luo analyzed data and assisted with writing the article.




1. J.C. Morris, K. Blennow, L. Froelich, et al. Harmonized diagnostic criteria for Alzheimer’s disease: recommendations, Journal of internal medicine2014; 275(3): 204-13.
2. F. Li, S. Chen, C. Wei, J. Jia. Monetary costs of Alzheimer’s disease in China: protocol for a cluster-randomised observational study, BMC neurology 2017; 17(1):15.
3. J. Jia, A. Zhou, C. Wei, et al. The prevalence of mild cognitive impairment and its etiological subtypes in elderly Chinese, Alzheimer’s & dementia : the journal of the Alzheimer’s Association 2014; 10(4): 439-47.
4. Y. Zhang, Y. Xu, H. Nie, et al. Prevalence of dementia and major dementia subtypes in the Chinese populations: a meta-analysis of dementia prevalence surveys, 1980-2010, Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia 2012;19(10): 1333-7.
5. K.Y. Chan, W. Wang, J.J. Wu, et al. Epidemiology of Alzheimer’s disease and other forms of dementia in China, 1990-2010: a systematic review and analysis, Lancet (London, England) 2013; 381(9882): 2016-23.
6. J. Jia, F. Wang, C. Wei, et al. The prevalence of dementia in urban and rural areas of China, Alzheimer’s & dementia : the journal of the Alzheimer’s Association 2014, 10(1): 1-9.
7. D. Liu, G. Cheng, L. An, et al. Public Knowledge about Dementia in China: A National WeChat-Based Survey, International journal of environmental research and public health2019; 16(21).
8. X. Li, W. Fang, N. Su, Y. Liu, S. Xiao, Z. Xiao, Survey in Shanghai communities: the public awareness of and attitude towards dementia, Psychogeriatrics : the official journal of the Japanese Psychogeriatric Society2011; 11(2): 83-9.
9. M. Zhao, X. Lv, M. Tuerxun, et al. Delayed help seeking behavior in dementia care: preliminary findings from the Clinical Pathway for Alzheimer’s Disease in China (CPAD) study, International psychogeriatrics 2016; 28(2): 211-9.
10. D. Peng, Z. Shi, J. Xu, et al. Demographic and clinical characteristics related to cognitive decline in Alzheimer disease in China: A multicenter survey from 2011 to 2014, Medicine2016; 95(26): e3727.
11. F.A. Schmitt, W. Ashford, C. Ernesto, et al. The severe impairment battery: concurrent validity and the assessment of longitudinal change in Alzheimer’s disease. The Alzheimer’s Disease Cooperative Study, Alzheimer disease and associated disorders1997; 11 Suppl 2: S51-6.
12. M.F. Folstein, S.E. Folstein, P.R. McHugh, «Mini-mental state». A practical method for grading the cognitive state of patients for the clinician, Journal of psychiatric research 1975; 12(3): 189-98.
13. J. Saxton, A.A. Swihart, Neuropsychological assessment of the severely impaired elderly patient, Clinics in geriatric medicine 1989; 5(3): 531-43.
14. S.J. Cano, H.B. Posner, M.L. Moline, et al. The ADAS-cog in Alzheimer’s disease clinical trials: psychometric evaluation of the sum and its parts, Journal of neurology, neurosurgery, and psychiatry2010; 81(12): 1363-8.
15. B. Reisberg, S.H. Ferris, M.J. de Leon, T. Crook. The Global Deterioration Scale for assessment of primary degenerative dementia, The American journal of psychiatry1982; 139(9): 1136-9.
16. B. Reisberg. Functional assessment staging (FAST), Psychopharmacology bulletin 1988; 24(4): 653-9.
17. C.P. Hughes, L. Berg, W.L. Danziger, L.A. Coben, R.L. Martin, A new clinical scale for the staging of dementia, The British journal of psychiatry : the journal of mental science 1982; 140: 566-72.
18. G.M. Peavy, D.P. Salmon, V.A. Rice, et al. Neuropsychological assessment of severely demeted elderly: the severe cognitive impairment profile, Archives of neurology 1996; 53(4): 367-72.
19. K.L. Lanctot, J. Amatniek, S. Ancoli-Israel, et al. Neuropsychiatric signs and symptoms of Alzheimer’s disease: New treatment paradigms, Alzheimer’s & dementia (New York, N. Y.)2017; 3(3): 440-449.
20. R.S. Doody, S.L. Strehlow, P.J. Massman, E.P. Feher, C. Clark, J.R. Roy, Baylor profound mental status examination: a brief staging measure for profoundly demented Alzheimer disease patients, Alzheimer disease and associated disorders 1999; 13(1): 53-9.
21. A. Korner, A. Brogaard, I. Wissum, U. Petersen, The Danish version of the Baylor Profound Mental State Examination, Nordic journal of psychiatry2012; 66(3): 198-202.
22. H.R. Na, S.H. Lee, J.S. Lee, R.S. Doody, S.Y. Kim, Korean version of the Baylor Profound Mental Status Examination: a brief staging measure for patients with severe Alzheimer’s disease, Dementia and geriatric cognitive disorders 2009; 27(1): 69-75.
23. S. Salmeron, I. Huedo, M. Lopez-Utiel, et al. Validation of the Spanish version of the Baylor Profound Mental Status Examination, Journal of Alzheimer’s disease 2016; 49(1): 73-8.
24. G. McKhann, D. Drachman, M. Folstein, R. Katzman, D. Price, E.M. Stadlan, Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease, Neurology1984; 34(7): 939-44.
25. R.H. Paul, R.A. Cohen, D.J. Moser, et al. The global deterioration scale: relationships to neuropsychological performance and activities of daily living in patients with vascular dementia, Journal of geriatric psychiatry and neurology 2002; 15(1): 50-4.
26. J.S. Kim, C.W. Won, B.S. Kim, H.R. Choi, Predictability of various serial subtractions on global deterioration scale according to education level, Korean journal of family medicine2013; 34(5): 327-33.
27. S.H. Choi, B.H. Lee, S. Kim, et al. Interchanging scores between clinical dementia rating scale and global deterioration scale, Alzheimer disease and associated disorders2003; 17(2): 98-105.
28. A.J. Midden, B.T. Mast, Differential item functioning analysis of items on the Geriatric Depression Scale-15 based on the presence or absence of cognitive impairment, Aging & mental health 2017; 1-7.



M. Moline1, S. Thein2, M. Bsharat1, N. Rabbee1, M. Kemethofer-Waliczky3, G. Filippov1, N. Kubota4, S. Dhadda1


1. Eisai Inc., Woodcliff Lake, NJ, USA; 2. Pacific Research Network – an ERG Portfolio Company, San Diego, CA, USA; 3. The Siesta Group, Vienna, Austria; 4. Eisai Co. Ltd., Tokyo, Japan.

Corresponding Authors: Margaret Moline, PhD, Clinical Research, Eisai, Inc., 100 Tice Boulevard, Woodcliff Lake, NJ 07677, USA, Phone: +1 (201) 949-4226, Fax: +1 (201) 949-4595, E-mail:

J Prev Alz Dis 2021;1(8):7-18
Published online December 3, 2020,



BACKGROUND: Irregular sleep-wake rhythm disorder (ISWRD) is a common sleep disorder in individuals with Alzheimer’s disease dementia (AD-D).
OBJECTIVES: This exploratory phase 2 proof-of-concept and dose-finding clinical trial evaluated the effects of lemborexant compared with placebo on circadian rhythm parameters, nighttime sleep, daytime wakefulness and other clinical measures of ISWRD in individuals with ISWRD and mild to moderate AD-D.
DESIGN: Multicenter, randomized, double-blind, placebo-controlled, parallel-group study.
SETTING: Sites in the United States, Japan and the United Kingdom.
PARTICIPANTS: Men and women 60 to 90 years of age with documentation of diagnosis with AD-D and Mini-Mental State Exam (MMSE) score 10 to 26.
INTERVENTION: Subjects were randomized to placebo or one of four lemborexant treatment arms (2.5 mg, 5 mg, 10 mg or
15 mg) once nightly at bedtime for 4 weeks.
MEASUREMENTS: An actigraph was used to collect subject rest-activity data, which were used to calculate sleep-related, wake-related and circadian rhythm–related parameters. These parameters included least active 5 hours (L5), relative amplitude of the rest-activity rhythm (RA) and mean duration of sleep bouts (MDSB) during the daytime. The MMSE and the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) were used to assess for changes in cognitive function.
RESULTS: Sixty-two subjects were randomized and provided data for circadian, daytime and nighttime parameters (placebo, n = 12; lemborexant 2.5 mg [LEM2.5], n = 12; lemborexant
5 mg [LEM5], n = 13, lemborexant 10 mg [LEM10], n = 13 and lemborexant 15 mg [LEM15], n = 12). Mean L5 showed a decrease from baseline to week 4 for LEM2.5, LEM5 and LEM15 that was significantly greater than with placebo (all p < 0.05), suggesting a reduction in restlessness. For RA, LS mean change from baseline to week 4 versus placebo indicated greater distinction between night and day with all dose levels of lemborexant, with significant improvements seen with LEM5 and LEM15 compared with placebo (both p < 0.05). The median percentage change from baseline to week 4 in MDSB during the daytime indicated a numerical decrease in duration for LEM5, LEM10 and LEM15, which was significantly different from placebo for LEM5 and LEM15 (p < 0.01 and p = 0.002, respectively).
There were no serious treatment-emergent adverse events or worsening of cognitive function, as assessed by the MMSE and ADAS-Cog. Lemborexant was well tolerated. No subjects discontinued treatment.
CONCLUSIONS: This study provides preliminary evidence of the potential utility of lemborexant as a treatment to address both nighttime and daytime symptoms in patients with ISWRD and AD-D.

Key words: Irregular sleep-wake rhythm disorder, Alzheimer’s disease, lemborexant.



Individuals with Alzheimer’s disease dementia (AD-D) commonly exhibit sleep disorders, particularly irregular sleep-wake rhythm disorder (ISWRD) (1, 2). ISWRD is a circadian rhythm sleep disorder, distinct from insomnia, which is characterized by the irregular distribution of sleep bouts across the 24-hour period rather than consolidated sleep at night (3). The most common symptoms of ISWRD are chronic sleep maintenance problems during the nighttime and a high level of daytime sleepiness (3). The pathology of ISWRD includes neuronal activity loss in the suprachiasmatic nucleus, a structure within the hypothalamus that controls circadian rhythms, and the pineal gland (3, 4).
The lack of a well-defined circadian pattern of sleep-wake behavior in patients with AD-D can present a challenge for caregivers (5). There are no pharmacologic treatments currently approved for ISWRD. The American Academy of Sleep Medicine (AASM) strongly recommends against the use of sedative-hypnotics in these patients owing to safety concerns, including increased risk of falls (6). Melatonin has not demonstrated efficacy in improving sleep in individuals with Alzheimer’s disease in clinical studies (7, 8), and the AASM does not recommend its use in elderly patients with dementia (6). Light therapy has been investigated as a potential nonpharmacologic treatment to improve sleep quality in patients with Alzheimer’s disease and related dementias (9). The AASM recommends its use versus no treatment in elderly patients with dementia (6), as some improvements in behavioral disorders have been reported (10). However, this recommendation was given a “strength value” of “Weak For,” as the quality of evidence was considered very low, as evaluated by the GRADE approach (6).
Consolidation of nighttime sleep and daytime wakefulness are the main goals of treatment for
ISWRD (3). Recent evidence suggests that a dysfunctional orexin system may play a role in the neuropathology of ISWRD (11, 12). Elevated orexin levels have been associated with both disturbed sleep and impaired cognition in patients with Alzheimer’s disease (11). Therapies targeting the orexin system, such as a dual orexin receptor antagonist (DORA), may improve sleep in individuals with Alzheimer’s disease (2, 13).
Lemborexant is a DORA that has been approved recently in the United States (14), Canada, and Japan for the treatment of insomnia in adults. In the pivotal phase 3 studies E2006-G000-304 (Study 304; SUNRISE-1; identifier NCT02783729) and E2006-G000-303 (Study 303; SUNRISE-2; identifier NCT02952820), lemborexant treatment provided significant benefit compared with placebo on polysomnogram-based and self-reported sleep onset and sleep maintenance outcomes over 1 month (Study 304), and patient-reported sleep onset and sleep maintenance outcomes over 6 months (Study 303), in subjects with insomnia disorder (15, 16). In both phase 3 clinical studies, lemborexant was well tolerated.
Here we describe results from an exploratory
phase 2 proof-of-concept and dose-finding clinical trial (E2006-G000-202 [Study 202]; identifier NCT03001557) that evaluated the effects of lemborexant compared with placebo on circadian rhythm parameters, nighttime sleep, daytime wakefulness and other clinical measures of ISWRD, in individuals with ISWRD and mild to moderate AD-D.



Study participants

This study enrolled men and women 60 to 90 years of age with documentation of diagnosis with AD-D on the basis of the National Institute on Aging/Alzheimer’s Association Diagnostic Guidelines and Mini-Mental State Exam (MMSE) (17) score 10 to 26. Subjects met criteria for circadian rhythm sleep disorder, irregular sleep-wake type (Diagnostic and Statistical Manual of Mental Disorders [5th edition]), and the International Statistical Classification of Diseases, Tenth Revision, as follows: complaint by the subject or caregiver of difficulty sleeping during the night and/or excessive daytime sleepiness associated with multiple irregular sleep bouts during a 24-hour period. Subjects also had frequency of complaint of sleep and wake fragmentation ≥ 3 days per week; duration of complaint of sleep and wake fragmentation ≥ 3 months; and mean sleep efficiency (SE) < 87.5% in the nocturnal sleep period and mean wake efficiency (WE) < 87.5% during the wake period, as measured by actigraphy during the screening period; and, as confirmed by actigraphy, a combination of sleep bouts of > 10 minutes during the wake period plus wake bouts of > 10 minutes during the sleep period, totaling at least 4 bouts per 24-hour period, ≥ 3 days per week. Subjects could also have no more than mild sleep apnea and be able to tolerate wearing an actigraphy device. Individuals with dementia other than AD-D and sleep disorders other than ISWRD were excluded. Additional details of major exclusion criteria are provided in the supplementary material.

Ethical Standards

This study received approval from the relevant Institutional Review Boards and Independent Ethics Committees and was conducted in adherence to Good Clinical Practice guidelines as required by the principles of the Declaration of Helsinki and the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. All protocol amendments were reviewed and approved by the institutional review board or independent ethics committee before study implementation. Details of protocol amendments are available on
Subjects or their legal representative signed the informed consent form. Caregivers signed a separate consent form. For the subject to enroll, there had to be one or more persons responsible to provide the required information for assessments, complete the sleep log for actigraphy and ensure that the subject was dosed at the appropriate time.

Study design

This multicenter, randomized, double-blind, placebo-controlled, parallel-group study was conducted at 57 sites: 47 in the United States, 9 in Japan and 1 in the United Kingdom, and started December 20, 2016. This study had three phases: the prerandomization phase, the randomization (core) phase and the extension phase (figure 1a). Here we present results from the prerandomization and randomization phases only, which completed July 26, 2018 (primary completion date); the extension phase completed on April 17, 2020.

Figure 1. (a) Study design; (b) Subject disposition. Visit 1 = Screening. Visit 2 = Caregiver visit; download actigraphy data. Visit 3 = Confirm eligibility and dispense study drug. Visit 4 = Subject and caregiver visit; download actigraphy data and perform safety assessments. Visit 5 = End-of-treatment assessments; download actigraphy data. Visit 6 = End-of-study assessments; download actigraphy data. *Sleep study: before randomization, the investigator was required to review a report detailing the potential subject’s Apnea-Hypopnea Index or equivalent. †Includes 14 subjects who were rescreened once and one subject who was rescreened twice. ‡Subjects were allowed to rescreen. Seven subjects rescreened and failed the second screening. Therefore, there were 151 individuals who were screen failures and 158 primary reasons for screen failure. BL, baseline, R, randomization, V, visit

*Including 14 subjects who rescreened once and 1 subject who rescreened twice; †Subjects were allowed to rescreen. Seven subjects rescreened and failed the second screening. Therefore, there are 151 individuals who were screen failures and 158 primary reasons for screen failure.


The prerandomization phase comprised a screening period and a baseline period. Eligible subjects were provided with an actigraph to wear continuously for at least the first 14 days of screening. During the screening period, subjects underwent a polysomnogram either at home or in the clinic to rule out moderate to severe sleep apnea (≥ 15 events per hour of sleep). Subjects who met eligibility criteria after at least 2 weeks of actigraphy could enter the randomization phase, in which they were randomized (1:1:1:1:1) to placebo or one of four lemborexant treatment arms (2.5 mg [LEM2.5], 5 mg [LEM5], 10 mg [LEM10] or 15 mg [LEM15]), stratified by country, for 4 weeks. Randomization was based on a computer-generated randomization scheme that was reviewed and approved by an independent statistician. Subjects and all personnel involved with the conduct and interpretation of the study, including investigators, site personnel and sponsor staff, were blinded to the treatment codes. Study drug was dispensed to the caregiver and was administered within 5 minutes of bedtime during the treatment period. Following the 4-week treatment period, there was a 2-week follow-up period without study medication to assess for possible rebound ISWRD symptoms and for safety. Eligible participants could enter an open-label extension phase for up to 30 months, or until program discontinuation, after the 2-week follow-up period.

Subjects were asked to wear an actigraphy device (MotionWatch8, CamNtech, Boerne, TX) continuously on their nondominant wrist for at least 14 days to qualify and for 28 days during placebo or lemborexant treatment. Subjects also wore the actigraph during the follow-up period. Actigraphy data were collected in 30-second epochs and scored centrally using a customized algorithm. The in-bed intervals and times when the actigraphs were removed from the wrists were provided to the central reader based on the sleep logs completed by the caregivers. At a minimum, participants were required to wear the actigraph for 5 complete days out of 7 days’ data. A day was considered complete as long as data from 90% of the 24-hour period was able to be scored.


This study evaluated the efficacy of lemborexant compared with placebo on changes from baseline in circadian, nighttime and daytime endpoints. Mean changes from baseline were evaluated over each week of treatment with lemborexant versus placebo for the following endpoints. All actigraphy-derived parameters were calculated based on the logged time in bed (nighttime) or logged time out of bed (daytime) as reported in the sleep log.

Circadian rhythm–related endpoints

Circadian rhythm–related endpoints included the least active 5 hours (L5), L5 start time (L5ST), most active 10 hours (M10), relative amplitude of the rest-activity rhythm (RA), interdaily stability (IS) and intradaily variability (IV). L5 was defined as the average activity across the least active 5-hour period of 24-hour sleep-wake rhythm (higher values indicate restlessness). For L5ST, the numbers represent clock times, with the two digits after the decimal point representing percentage of 60 minutes. M10 was defined as the average activity during the most active 10-hour period per 24-hour period (low levels indicating inactivity). RA was calculated as the difference between M10 and L5 divided by M10 plus L5. RA standardizes for activity-level differences across subjects and reflects strength of circadian signal; values closer to 1 represent rhythms with higher relative amplitudes. IS was derived by the ratio between the variance of the average 24-hour pattern around the mean and the overall variance, and gives an indication of the stability of the sleep-wake rhythm across days, and varies from zero (low stability) to 1 (high stability). IV was derived by the ratio of the mean squares of the difference between all successive hours (first derivative) and the mean squares around the grand mean (overall variance). IV gives an indication of ISWRD by quantifying the number and strength of transitions between rest and activity bouts, with a higher number indicating more fragmentation.

Daytime wake endpoints

Endpoints related to daytime wake included WE, wake fragmentation index (WFI) and mean number and mean duration of sleep bouts during the daytime. These endpoints were derived by actigraphy. WE was defined as wake time per daytime hours and calculated as 100% × the total duration of wake epochs during the wake period (ie, the time outside of the sleep period) divided by the duration of the daytime hours. WFI, which characterizes transitions between wake and sleep throughout the day, was calculated as the sum of an immobility index and a fragmentation index, with immobility index equal to epochs of immobility outside of the defined sleep period × 100, and fragmentation index equal to the number of ≤ 1-minute periods of mobility/total number of periods of mobility outside of the sleep period × 100. The mean number and mean duration of sleep bouts that occurred during the hours outside of the nocturnal sleep period were assessed, where a sleep bout was defined as continuous sleep of 10 minutes or longer. Lastly, total sleep time (TST) during the daytime, defined as minutes of sleep during the day, was also assessed.

Nighttime sleep endpoints

Endpoints related to nighttime sleep included actigraphy-derived SE, actigraphy-derived sleep fragmentation index (SFI) and TST during the nighttime. SE was calculated as 100% times the total duration of sleep epochs during the nocturnal sleep period. SFI was calculated as the sum of a movement index and a fragmentation index, with movement index = (epochs of wake per time in bed) × 100 and fragmentation index = (number of ≤ 1-minute periods of immobility/total number of periods of immobility of all durations during the nocturnal sleep period) × 100. This outcome measures the transitions between sleep and wake throughout the night; higher values indicate fragmented sleep. TST during the night was defined as minutes of sleep during the nighttime. The mean number and duration of wake bouts that occurred during the nocturnal sleep period, where a wake bout was defined as continuous wake of
10 minutes or longer, were also assessed.

Additional assessments

The MMSE (17) and the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) (18) were administered prior to and at the end of treatment to assess for changes in cognitive function. The Clinician’s Global Impression of Change–ISWRD version, the Neuropsychiatric Inventory (19) and the Sleep Disorders Inventory (18) were also assessed in this study, but these data will be reported separately.

Statistical analyses

The study objectives reflect the exploratory nature of this phase 2 study and were not categorized as primary or secondary, following a protocol amendment (Protocol Amendment 6; June 20, 2018).

Sample size

The sample size of this proof-of-concept study was approximately 60 subjects, reduced from approximately 125 subjects following a protocol amendment (Protocol Amendment 6; June 20, 2018). Sample size was reduced following the amending of the objectives and endpoints to reflect the exploratory nature of the proof-of-concept study. All statistical tests were based on the 5% level of significance (two-sided), unless otherwise stated. No multiplicity adjustments were made. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC). Responder analyses, network analyses and corresponding visualizations were created using the R statistical software package (20).

Populations analyzed

Efficacy analyses were performed on the Full Analysis Set (FAS) unless otherwise specified. The FAS was defined as the group of randomized subjects who received at least one dose of randomized study drug and had at least one post-dose efficacy measurement. The Safety Analysis Set (SAS) was defined as the group of randomized subjects who received at least one dose of randomized study drug and had at least one post-dose safety assessment.
Demographic and other baseline characteristics for the SAS were summarized for each treatment group using descriptive statistics. For all actigraphy parameters, baseline was defined as the average value during the designated days of screening. For L5, M10, RA, IS and IV parameters, the weekly averages were calculated by the actigraphy vendor. For these variables, the last record of the screening period was considered as the baseline (generally the average of the last 7 days) of the screening period. Efficacy evaluations in this study mainly focused on numerical changes for summary statistics and their clinical significance based on the limited number of subjects.
The change from baseline to week 4 of the following endpoints was analyzed using mixed models for repeated measures (MMRM) analysis on the FAS for lemborexant versus placebo: L5, M10, RA, IS, IV, mean WE, mean WFI, TST during the day, mean number and mean duration of sleep bouts during the daytime, mean SE, mean SFI, TST during the night, mean number and mean duration of wake bouts during the nighttime. The MMRM model included all data and was adjusted for the corresponding baseline value, country, treatment, visit (week 1, week 2, week 3 and week 4) and treatment-by-visit interaction. The MMRM model accounted for any missing data, and assumed that missing data were missing at random. An unstructured covariance matrix was used and, if the model failed to converge, an autoregressive matrix was used. Where data were normally distributed, least squares (LS) means, difference in LS means of each lemborexant dose compared with placebo,
95% confidence intervals and p values at the appropriate time point were presented.
To identify relevant efficacy variables, a Gaussian graphical model was developed post hoc using the R statistical software package. Regularization method was applied to infer a sparse network topology of interconnectedness among the efficacy variables.
Mean change from baseline in L5, average L5ST, mean duration of sleep bouts and average number of wake bouts were analyzed post hoc for LEM5 versus placebo at week 1, week 2, week 3 and week 4 using an MMRM model, adjusted for region and baseline value of the variable. Mean and standard error were plotted from the model at each time point to represent any longitudinal trends graphically.
The mean duration of sleep bouts during the daytime, one of the network analysis–identified variables, was analyzed separately post hoc to determine the treatment effect. Boxplots were produced, and percentage change from baseline at week 4 was compared for each dose group versus placebo. The Wilcoxon test was performed to compare pairwise means of each treatment dose with placebo.
Changes from baseline in the MMSE and ADAS-Cog were analyzed using analysis of covariance, adjusted for baseline value and country.

Responder analyses

Responder analyses were also conducted, in which responders were defined separately as:
• Subjects whose mean activity level dropped from baseline at week 4 during L5 (sleep) and whose mean duration of sleep bouts during the wake period decreased from baseline at week 4. A nominal threshold of 5% (rather than 0) was applied for the definition.
• Subjects whose mean duration of sleep bouts during the wake period decreased from baseline at week 4, whose mean RA of sleep-wake cycle improved from baseline at week 4 and whose mean IS of sleep-wake cycle improved.

In responder analyses, the percentage change from baseline at week 4 was used as the metric for change for each variable.


All subjects underwent routine safety assessments at specified visits, including questioning regarding treatment-emergent adverse events (TEAEs) and serious adverse events (SAEs); suicidality (assessed using an electronic version of the Columbia–Suicide Severity Rating Scale) (21); electrocardiograms; vital signs, weight; hematology and blood chemistry analysis; and urinalysis.



In total, 214 subjects were screened, 63 were randomized and 62 completed the randomization phase of this study and comprised the FAS and SAS (figure 1b). Fifty subjects randomized to lemborexant (12, 13, 13 and 12 subjects in the LEM2.5, LEM5, LEM10 and LEM15 groups, respectively) and 12 subjects randomized to placebo received at least one dose of study drug. All 62 subjects received study drug for the entire treatment period. Treatment groups were generally balanced with respect to most demographic variables across the five groups; however, the number of males versus females was not fully balanced across all groups (table 1). Baseline actigraphy characteristics were consistent with the presence of ISWRD (table 2). Mean baseline MMSE score was comparable across the five treatment groups and indicated mild to moderate Alzheimer’s disease (supplementary table S1).

Table 1. Baseline demographics

BMI, body mass index; SD, standard deviation.

Table 2. Summary of change from baseline to week 4 for circadian rhythm–related, daytime and nighttime outcomes

*Based on a mixed model for repeated measure analysis adjusted for baseline value, country, visit and treatment-by-visit interaction. †Numbers represent clock times, with the two digits after the decimal point representing percentage of 60 minutes. ‡Sleep fragmentation index was calculated based on the logged time in bed. CI, confidence interval; L5, least active 5-hour period per 24-hour period; LS, least squares; PBO, placebo; SD, standard deviation.


Efficacy outcomes

Network analysis of efficacy variables

As efficacy variables are interrelated, an advanced network analysis was performed to elucidate the relational structure of circadian rhythm variables and treatment (supplementary figure S1). The main efficacy variables identified from the network analysis were the mean duration of sleep bouts during the daytime, activity level during L5, start time of the L5 period and number of wake bouts at night.

Circadian rhythm–related outcomes

At week 4, mean L5 showed a significantly greater decrease from baseline versus placebo for LEM2.5, LEM5 and LEM15 (table 2), indicating a quieter and more restful nighttime sleep. When examined longitudinally over 4 weeks for the LEM5 dose, consistent improvements (decreases) from baseline in L5 were observed after each week of treatment (figure 2a).
Mean baseline L5ST ranged from 24.08 to 25.24 hours (corresponding to ~12:05 am to ~1:15 am) across all groups, meaning that L5 was occurring during the nighttime (table 2). Numerical LS mean decreases from baseline in L5ST were observed at week 4 with LEM5 and LEM15, which were not significantly different from placebo (table 2). Over the 4 weeks of treatment with LEM5, there was no consistent change in L5ST, suggesting no phase shift in the timing of the L5 of the circadian sleep-wake rhythm (figure 2b).
Only LEM5 demonstrated a numerical improvement versus placebo in the LS mean change from baseline in M10, but this treatment difference was not statistically significant (table 2). LS mean treatment difference in change from baseline indicated higher RAs with all dose levels of lemborexant compared with placebo, with significant increases seen with LEM5 and LEM15 (table 2). LEM5 demonstrated improvements in IS and IV versus placebo at week 4, but these improvements did not reach statistical significance compared with placebo (table 2).

Figure 2. Longitudinal plots of mean change from baseline in circadian, daytime and nighttime efficacy variables over 4 weeks of treatment for LEM5 versus placebo analyzed by mixed effects repeated measures analysis. (a) L5; (b) L5ST; (c) MDSB during the daytime; (d) WB during the night. Error bars represent SE. Mean and SEs were plotted from mixed models for repeated measures analyses. L5, mean least active 5-hour period per 24-hour period; L5ST, mean start hour of L5 (HH); LEM5, lemborexant 5 mg; MDSB, mean duration of sleep bouts (minutes); SE, standard error; WB, mean number of wake bouts


Daytime endpoints

Of the LEM doses, only LEM5 demonstrated a numerical increase from baseline in LS mean WE during the daytime, a numerical reduction from baseline in LS mean WFI (lower values indicate more consolidated wake during the daytime) and a numerical reduction from baseline in LS mean TST during the daytime at week 4; though these changes were not significantly different from placebo (table 2).
In the longitudinal analysis, greater numerical decreases from baseline in mean duration of sleep bouts during the daytime were observed in the LEM5 group compared with placebo across each study week (figure 2c), but the week 4 analysis showed no statistically significant treatment difference versus placebo (table 2).
Median percentage change from baseline to week 4 in mean duration of sleep bouts during the daytime indicated a decrease in duration with LEM5, LEM10 and LEM15 (supplementary figure S2). The greatest decreases occurred in the LEM5 and LEM15 treatment groups, and these changes were statistically significantly different versus placebo (p < 0.01 and p = 0.002, respectively).

Nighttime endpoints

LS mean changes from baseline to week 4 in nighttime endpoints indicated numerical increases in SE for LEM2.5 and LEM5, numerical improvements in SFI with LEM2.5, LEM5 and LEM15, indicating more consolidated (ie, less fragmented) sleep, and numerical improvements in mean TST during the night with LEM5 and LEM15; none of these changes were statistically significantly different versus placebo (table 2). Decreases from baseline to
week 4 in LS mean number of wake bouts during the night were observed in the LEM2.5 and LEM5 groups which were significantly greater than with placebo (table 2). The LS mean duration of wake bouts during the night increased for the LEM2.5, LEM5 and LEM15 groups, but these differences were not statistically significant compared with placebo.
When analyzed by treatment week, consistent decreases (improvements) from baseline in the mean number of wake bouts were observed at each time point for the LEM5 group, whereas increases from baseline were observed in the placebo group at Weeks 1, 2 and 4 (figure 2d).

Responder analyses

After 4 weeks, a greater percentage of subjects in each lemborexant treatment group met post hoc responder criteria, defined as > 5% decreases from baseline in both L5 and mean duration of sleep bouts during the daytime, compared with placebo (supplementary figure S3a). Additionally, after 4 weeks, a greater percentage of subjects in each lemborexant treatment group, versus placebo, met the more restrictive post hoc responder criteria, defined as changes from baseline at 4 weeks of > 0% for mean RA and IS, and < 0% for mean duration of sleep bouts during wake (supplementary figure S3b).

Cognitive assessments and safety outcomes

In this study, no significant worsening of cognition, as assessed by MMSE and ADAS-Cog, was observed by the end of the treatment period (supplementary table S1). The incidence of TEAEs was slightly higher for the highest dose of LEM15 (50.0%) compared with placebo (33.3%), and similar to placebo in the other lemborexant groups (23.1-30.8%) (table 3). Across the treatment groups, four subjects reported TEAEs of moderate severity; one subject in the LEM15 group reported somnolence of moderate severity. One severe TEAE, arthralgia, was reported by one subject in the LEM15 group. There were no deaths, no treatment-emergent SAEs and no TEAEs leading to study drug discontinuation reported (table 3). The most common TEAEs (reported in two or more subjects in any lemborexant group) were constipation, somnolence, arthralgia, headache and nightmare, and those events were not reported for placebo, LEM2.5 or LEM5. No falls or confusion were observed and no suicidality was reported in any lemborexant-treated subjects.

Table 3. Summary of treatment-emergent adverse events*†

*A TEAE is defined as an AE with onset date on or after the first dose of study drug up to 14 days after the last dose of study drug. †For each row category, a subject with two or more TEAEs in that category is counted only once. ‡If a subject had a single incident of an AE (Preferred Term) with a missing severity, the subject was counted in the ‘Missing’ category for that Preferred Term. If a subject had two or more AEs in the same system organ class (or with the same Preferred Term) with different severities, then the event with the maximum severity was used for that subject. Subjects with missing AE severity are counted under the ‘Missing’ category unless the subject already has another AE with severe intensity, in which case the subject is counted in the ‘Severe’ category. AE, adverse event; PBO, placebo; TEAE, treatment-emergent adverse event.



This exploratory randomized clinical study is the first to investigate the use of a drug affecting orexin neurotransmission in a patient population with ISWRD. Treatment with lemborexant improved 24-hour circadian rhythm variables, as demonstrated by increased RA, and helped to consolidate nighttime sleep by decreasing L5. Subjects were able to have longer, more restful and less fragmented sleep, a key goal in the treatment of ISWRD (3). Lemborexant exhibited treatment benefit, as detected by the interconnected efficacy variables in ISWRD patients on their circadian rhythm. Results of this study provide preliminary evidence that treatment with lemborexant may improve both 24-hour circadian rhythm variables and nocturnal sleep variables and impact the duration of daytime unplanned naps in subjects with ISWRD and AD-D. Additionally, these results suggest that proof-of-concept was established, as objective endpoints were identified that both characterized ISWRD in this patient population and were clinically relevant.
LEM5 appeared to be the most consistently effective dose in improving circadian rhythm–related, wake-related and sleep-rated actigraphy variables in this study. LEM5 demonstrated significant treatment differences versus placebo at week 4 in improving L5, RA and mean number of wake bouts during the night. LEM5 also resulted in less daytime sleep, as demonstrated by the greater numerical decreases from baseline in mean duration of sleep bouts during the day compared with placebo during each study week. Importantly, numerically higher RAs in circadian sleep-wake rhythms (ie, more distinction between night and day) were seen with all lemborexant dose levels.
Lemborexant was generally well tolerated in this population of individuals with Alzheimer’s disease and ISWRD. The rate of TEAEs was low, no treatment-emergent SAEs were reported and no new safety concerns were identified in this study. The safety profile in this study population was consistent with that observed in adult subjects with insomnia (15, 16). Additionally, treatment with lemborexant did not worsen the cognitive functions of this population of subjects with ISWRD and AD-D.
Dysregulation of the sleep-wake cycle is a common problem in patients with Alzheimer’s disease (22). One potential consequence for patients with Alzheimer’s disease who suffer from sleep disorders is an increased likelihood of institutionalization (23). However, at this time, the lack of approved pharmacologic treatments for patients with ISWRD and AD-D represents an unmet medical need. Some evidence is available to support the use of nonpharmacologic interventions, such as light therapy, behavioral techniques and increased social and physical activity during the daytime, to improve sleep in patients with Alzheimer’s disease (9, 10, 24). Both the American Geriatric Society and the AASM discourage the use of benzodiazepines in older adults (6, 25), as this drug class has been shown to be significantly associated with falls in the elderly population (26).
DORAs, which block the orexin system, may have the potential to improve sleep in patients with AD-D. Data regarding the treatment of insomnia (not ISWRD) in patients with mild to moderate Alzheimer’s disease have recently been added to the prescribing information for the DORA suvorexant (27).
Strengths of this study include the use of actigraphy, which can capture the full 24-hour sleep-wake pattern in treatment trials and has been a common method for assessing sleep in individuals with Alzheimer’s disease (28). Study limitations include the small sample size, which was, in part, due to slow recruitment. Additionally, the study duration was only 1 month.
These results provide important new information regarding the potential utility of lemborexant to address both nighttime and daytime symptoms that affect sleep-related quality of life of patients with ISWRD and AD-D, as well as reduce the burden of patients’ sleep disturbances on their caregivers and families. Further evaluation in future clinical trials is warranted to confirm the value of lemborexant in this patient population.


Funding: This study was sponsored by Eisai Inc. The sponsor participated in the design and conduct of the study; the collection, analysis and interpretation of data; and the preparation, review and approval of the manuscript.

Acknowledgement: Medical writing assistance was provided by Rebecca Jarvis, PhD, of ProScribe – part of the Envision Pharma Group and was funded by Eisai Inc. Envision Pharma Group’s services complied with international guidelines for Good Publication Practice (GPP3).

Declaration of conflicting interests: Drs Moline, Rabbee, Filippov and Dhadda are employees of Eisai Inc. Dr Bsharat is formerly an employee of Eisai Inc. Mr Kubota is an employee of Eisai Co. Ltd. Dr Thein is the director and founder of Pacific Research Network, which received funding from the study sponsor, Eisai Inc., for the conduct of this study. Mr Kemethofer is an employee of The Siesta Group, the central actigraphy scoring vendor.

Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.



1. Brzecka A, Leszek J, Ashraf GM, et al. Sleep disorders associated with Alzheimer’s disease: a perspective. Front Neurosci 2018;12:330.
2. Musiek ES, Xiong DD, Holtzman DM. Sleep, circadian rhythms, and the pathogenesis of Alzheimer disease. Exp Mol Med 2015;47:e148.
3. Zee PC, Vitiello MV. Circadian rhythm sleep disorder: irregular sleep wake rhythm type. Sleep Med Clin 2009;4:213-218.
4. Zhu L, Zee PC. Circadian rhythm sleep disorders. Neurol Clin 2012;30:1167-1191.
5. Gehrman P, Gooneratne NS, Brewster GS, Richards KC, Karlawish J. Impact of Alzheimer disease patients’ sleep disturbances on their caregivers. Geriatr Nurs 2018;39:60-65.
6. Auger RR, Burgess HJ, Emens JS, Deriy LV, Thomas SM, Sharkey KM. Clinical practice guideline for the treatment of intrinsic circadian rhythm sleep-wake disorders: advanced sleep-wake phase disorder (ASWPD), delayed sleep-wake phase disorder (DSWPD), non-24-hour sleep-wake rhythm disorder (N24SWD), and irregular sleep-wake rhythm disorder (ISWRD). An update for 2015: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med 2015;11:1199-1236.
7. Serfaty M, Kennell-Webb S, Warner J, Blizard R, Raven P. Double blind randomised placebo controlled trial of low dose melatonin for sleep disorders in dementia. Int J Geriatr Psychiatry 2002;17:1120-1127.
8. Singer C, Tractenberg RE, Kaye J, et al. A multicenter, placebo-controlled trial of melatonin for sleep disturbance in Alzheimer’s disease. Sleep 2003;26:893-901.
9. Figueiro MG, Plitnick BA, Lok A, et al. Tailored lighting intervention improves measures of sleep, depression, and agitation in persons with Alzheimer’s disease and related dementia living in long-term care facilities. Clin Interv Aging 2014;9:1527-1537.
10. Mishima K, Okawa M, Hishikawa Y, Hozumi S, Hori H, Takahashi K. Morning bright light therapy for sleep and behavior disorders in elderly patients with dementia. Acta Psychiatr Scand 1994;89:1-7.
11. Liguori C, Nuccetelli M, Izzi F, et al. Rapid eye movement sleep disruption and sleep fragmentation are associated with increased orexin-A cerebrospinal-fluid levels in mild cognitive impairment due to Alzheimer’s disease. Neurobiol Aging 2016;40:120-126.
12. Liguori C, Romigi A, Nuccetelli M, et al. Orexinergic system dysregulation, sleep impairment, and cognitive decline in Alzheimer disease. JAMA Neurol 2014;71:1498-1505.
13. Duncan MJ, Farlow H, Tirumalaraju C, et al. Effects of the dual orexin receptor antagonist DORA-22 on sleep in 5XFAD mice. Alzheimers Dement 2019;5:70-80.
14. Dayvigo (lemborexant) tablets [package insert]. Woodcliff Lake, NJ: Eisai Inc.; 2019.
15. Rosenberg R, Murphy P, Zammit G, et al. Comparison of lemborexant with placebo and zolpidem tartrate extended release for the treatment of older adults with insomnia disorder: a phase 3 randomized clinical trial. JAMA Netw Open 2019;2:e1918254.
16. Karppa M, Yardley J, Pinner K, et al. Long-term efficacy and tolerability of lemborexant compared with placebo in adults with insomnia disorder: results from the phase 3 randomized clinical trial SUNRISE 2. Sleep 2020.
17. Folstein MF, Folstein SE, McHugh PR. « Mini-mental state ». A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198.
18. Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimer’s disease. Am J Psychiatry 1984;141:1356-1364.
19. Trzepacz PT, Saykin A, Yu P, et al. Subscale validation of the neuropsychiatric inventory questionnaire: comparison of Alzheimer’s disease neuroimaging initiative and national Alzheimer’s coordinating center cohorts. Am J Geriatr Psychiatry 2013;21:607-622.
20. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing. 2013;
21. Posner K, Brown GK, Stanley B, et al. The Columbia-Suicide Severity Rating Scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults. Am J Psychiatry 2011;168:1266-1277.
22. Liguori C, Spanetta M, Izzi F, et al. Sleep-wake cycle in Alzheimer’s disease is associated with tau pathology and orexin dysregulation. J Alzheimers Dis 2020;74:501-508.
23. Pollak CP, Perlick D. Sleep problems and institutionalization of the elderly. J Geriatr Psychiatry Neurol 1991;4:204-210.
24. McCurry SM, Gibbons LE, Logsdon RG, Vitiello MV, Teri L. Nighttime insomnia treatment and education for Alzheimer’s disease: a randomized, controlled trial. J Am Geriatr Soc 2005;53:793-802.
25. American Geriatrics Society 2019 Beers Criteria Update Expert Panel. American Geriatrics Society 2019 Updated AGS Beers Criteria(R) for Potentially Inappropriate Medication Use in Older Adults. J Am Geriatr Soc 2019;67:674-694.
26. Woolcott JC, Richardson KJ, Wiens MO, et al. Meta-analysis of the impact of 9 medication classes on falls in elderly persons. Arch Intern Med 2009;169:1952-1960.
27. Belsomra (suvorexant) tablets [package insert]. Whitehouse Station, NJ: Merck & Co., Inc.; 2020.
28. Camargos EF, Louzada FM, Nóbrega OT. Wrist actigraphy for measuring sleep in intervention studies with Alzheimer’s disease patients: application, usefulness, and challenges. Sleep Med Rev 2013;17:475-488.



G. Klein1, P. Delmar1, G.A. Kerchner1, C. Hofmann1, D. Abi-Saab1, A. Davis2, N. Voyle2, M. Baudler1,3, P. Fontoura1, R. Doody1,3


1. F. Hoffmann-La Roche Ltd, Basel, Switzerland; 2. Roche Products Ltd, Welwyn Garden City, UK; 3. Genentech Inc., South San Francisco, CA, USA.

Corresponding Authors: Gregory Klein, Biomarkers and Translational Technology, Neuroscience and Rare Diseases, Basel, Switzerland. Email:, Phone: (+41) 616820759

J Prev Alz Dis 2021;1(8):3-6
Published online December 3, 2020,



Previous findings from the positron emission tomography (PET) substudy of the SCarlet RoAD and Marguerite RoAD open-label extension (OLE) showed gantenerumab doses up to 1200 mg every 4 weeks administered subcutaneously resulted in robust beta-amyloid (Aβ) plaque removal over 24 months in people with prodromal-to-moderate Alzheimer’s disease (AD). In this 36-month update, we demonstrate continued reduction, with mean (standard error) centiloid values at 36 months of -4.3 (7.5), 0.8 (6.7), and 4.7 (8.0) in the SCarlet RoAD (double-blind pooled placebo and active groups), Marguerite RoAD double-blind placebo, and Marguerite RoAD double-blind active groups respectively, representing a change of -57.0 (10.3), -90.3 (9.0), and -74.9 (10.5) centiloids respectively. These results demonstrate that prolonged gantenerumab treatment, at doses up to 1200 mg, reduces amyloid plaque levels below the amyloid positivity threshold. The ongoing GRADUATE Phase III trials will evaluate potential clinical benefits associated with gantenerumab-induced amyloid-lowering in people with early (prodromal-to-mild) AD.

Key words: Gantenerumab, Alzheimer’s disease, positron emission tomography, amyloid.




Alzheimer’s disease (AD) accounts for 60–80% of all cases of dementia globally (1). Currently, the available treatments for AD offer only limited benefits and there is an urgent need for disease-modifying therapies that reverse neuropathologic changes, or slow or stop neurodegeneration (1, 2).
AD pathogenesis is driven by the gradual accumulation of beta-amyloid (Aβ) plaques and neurofibrillary tangles (NFTs) in the brain (1, 3). In vitro and in vivo evidence suggests that soluble Aβ oligomers and insoluble Aβ plaques contribute to cognitive failure by causing neuronal loss, synaptic dysfunction and disconnection syndromes (4, 5). The recognition of Aβ accumulation as the earliest identifiable marker of AD has led to the development of amyloid positron emission tomography (PET), a neuroimaging technique that can be utilized to visualize Aβ accumulation that helps improve diagnostic accuracy and may also facilitate appropriate participant selection in clinical trials (6).
Gantenerumab is a fully humanized, anti-Aβ immunoglobulin (Ig) G1 that binds to Aβ species with high affinity for aggregated forms, including oligomers and plaques, and is thought to remove Aβ via microglia-mediated phagocytosis (7, 8). The long-term, pharmacodynamic effect of gantenerumab-induced Aβ plaque removal in participants with prodromal-to-mild and mild-to-moderate AD is currently being investigated in the PET substudies of the Phase III SCarlet RoAD (SR; NCT01224106) and Marguerite RoAD (MR; NCT02051608) open-label extension (OLE) studies, respectively (7). Interim results showed robust Aβ plaque removal with gantenerumab doses up to
1200 mg administered subcutaneously, with mean amyloid reductions of 59 centiloids and 51% of participants below the Aβ positivity threshold after 24 months (7). Here, we tested whether amyloid signal plateaus or continues to decline with continued therapy in the 36-month results of the ongoing OLE PET substudy.



Participants and study design

Complete details of the study designs and methodologies of SR and MR, the associated OLE studies, and the OLE PET substudies have been previously reported (7-9). Briefly, participants in the SR trial who received double-blind treatment and had ≥1 follow-up visit and those who were currently enrolled in the MR trial were eligible for participation in the OLE. Various titration schemes were used to allow OLE participants to gradually reach the target dose of gantenerumab 1200 mg per month while decreasing the risk of amyloid-related imaging abnormality (ARIA)-related adverse events. The target gantenerumab dose was reached within 6 to 10 months for SR OLE participants, and 2 to 6 months in MR OLE participants.
Participants of the OLE substudy were divided into three cohorts based on their prior exposure to gantenerumab and their stage of AD. The SR cohort included SR participants with all SR treatment arms pooled together (received gantenerumab 105 mg or 225 mg or placebo every 4 weeks during the double-blind phase), all participants in the SR cohort were off treatment for 16 to 19 months prior to OLE higher dosing. The MR double-blind placebo cohort (MR-DBP) included participants in the MR trial who received placebo during the double-blind phase and the MR double-blind active cohort (MR-DBA) included participants of the MR trial who received either 105 or 225 mg gantenerumab during the double-blind phase.

Amyloid-β plaque PET imaging and quantification

Amyloid PET scans were obtained at baseline and at 12, 24, and 36 months after baseline using intravenous
370 MBq 18F-florbetapir, with each 15-minute scan obtained 50 minutes after 18F-florbetapir injection. Participants who received a PET scan during the double-blind phase, within 9 to 12 months of OLE dosing, were not scanned at OLE baseline to minimize participant burden.
Volume-weighted, gray matter-masked standard uptake value ratios (SUVR) were calculated for six bilateral cortical regions using the Automated Anatomical Labeling (AAL) template, normalized by a cerebellar cortex reference region (10, 11). SUVR values were then converted to centiloid values as previously described, using the following linear transformation: Centiloid = SUVR*184.12 – 233.72 (7, 12). The threshold for amyloid positivity has been previously established as 24 centiloids, which corresponds to 1.40 SUVR units. The amyloid positivity threshold represents the quantitative threshold that best discriminates pathologically verified absence of plaques or sparse plaques from moderate-to-frequent plaques (13).

Statistical analysis

This analysis included all study participants who had a PET scan at OLE baseline (or 9–12 months prior to OLE dosing) and received ≥1 follow-up scan. PET centiloid values were analyzed using a mixed model for repeated measures (MMRM), with treatment visit, treatment group, and the interaction for treatment group by visit as independent variables. An unstructured covariance matrix was used to capture within-participant correlation.



Participant characteristics

A total of 67 participants with at least 1 post-baseline scan were enrolled in the OLE PET substudy (SR, n = 19; MR-DBP, n = 27; MR-DBA, n = 21). A total of 30 participants completed the 36-month scan (SR, n = 10; MR-DBP, n = 12; MR-DBA, n = 8). The baseline characteristics for both the overall population and the 36-month completers are shown in Table 1. More than half of the participants in each cohort were Apolipoprotein E (APOE) ε4 carriers (SR, 89%; MR-DBP, 67%; MR-DBA, 52%). Across all three cohorts, the mean [SE] baseline amyloid burden in centiloids was above the positivity threshold (SR, 49.6 [12.1]; MR-DBP, 91.1 [9.6]; MR-DBA, 79.6 [10.9]).

Table 1. Baseline characteristics of participants enrolled in the SR, MR-DBP, and MR-DBA cohorts, including 36-month completers

APOEε4, Apolipoprotein E; IQR, Interquartile range; MMSE, Mini-Mental State Examination; MR-DBA, Marguerite RoAD double-blind active; MR-DBP, Marguerite RoAD double-blind placebo; OLE, open-label extension; SE, standard error; SD, standard deviation; SR, SCarlet RoAD.


Amyloid PET results

Consistent with our previous report, reductions in mean amyloid burden were observed across cohorts after 12 and 24 months of open-label therapy, with 37% and 52% of participants, respectively, reaching levels below the amyloid positivity threshold (Figure 1) (7). Continued reductions beyond 24 months were observed after 36 months, with mean amyloid levels approaching zero centiloids across all cohorts. The absolute mean (SE) amyloid burden after 36 months were -4.3 (7.5), 0.8 (6.7), and 4.7 (8.0) centiloids for the SR, MR-DBP and MR-DBA cohorts, respectively, representing a change of -57.0 (10.3), -90.3 (9.0), -74.9 (10.5) centiloids respectively. Furthermore, the proportion of participants below the amyloid positivity threshold was 24 of 30 participants (80%) at 36 months (Figure 1).

Figure 1. Reduction of amyloid burden towards zero centiloids after 36 months of open-label therapy

*LS mean (SE); †Analyzed using an MMRM; LS, least-squares; MMRM, mixed model for repeated measures; MR-DBA, Marguerite RoAD double-blind active; MR-DBP, Marguerite RoAD double-blind placebo; SE, standard error; SR, SCarlet RoAD; SUVR, standard uptake value ratio.



This 36-month OLE PET substudy investigated the effect of gantenerumab on Aβ plaque removal on participants with prodromal-to-moderate AD. Prior results have shown that while the three cohorts began with considerably different mean baseline centiloid values, all three cohorts demonstrated a mean centiloid value just below the amyloid positivity threshold after
24 months of treatment with gantenerumab 1200 mg every 4 weeks. The latest results showed continued Aβ reduction with gantenerumab treatment below the amyloid positivity threshold, without plateau, with 80% of completers below the amyloid positivity threshold after 36 months of open-label therapy. Mean centiloid values of all three cohorts at this time are near a value of zero, which represents the mean amyloid burden expected in a healthy control group (12). Given that the SR and MR-DBA groups may have experienced some amyloid reduction due to low-dose gantenerumab treatment during the double-blind period of the SR and MR studies, the 90-centiloid reduction seen in the MR-DBP group represents the amyloid reduction that could be expected in a treatment-naïve population. The consistent reduction in Aβ suggests that gantenerumab is able to remove Aβ species successfully.
These findings may translate to clinical benefit in people with prodromal-to-mild AD as other studies with aducanumab and lecanemab (BAN2401) have observed amyloid PET reduction as well as clinical efficacy (7, 14, 15). Specifically, in a Phase Ib placebo-controlled study, aducanumab demonstrated reduced brain amyloid plaque levels after 24 months with a reduction in clinical decline as measured by the Clinical Dementia Rating–Sum of Boxes (CDR-SB) and Mini-Mental State Examination (MMSE) (14). In a Phase II placebo-controlled study, lecanemab produced a dose-dependent reduction in amyloid plaque levels after
18 months and a reduction in clinical decline as measured by AD Composite Score (15). In light of these studies, the current PET results suggest that the process of Aβ reduction at the gantenerumab dose of 1200 mg every
4 weeks has the potential to produce clinical benefits. The precise relation between amyloid reduction and clinical benefit is still an open question, including the question of whether a reduction to below amyloid positivity or to centiloid zero makes a difference in the clinical outcome and management of patients with early AD. The ongoing GRADUATE Phase III program evaluates the safety and efficacy of gantenerumab, subcutaneously administered, in participants with early AD. This program includes two global, double-blind, placebo-controlled trials in people with early AD, designed to maximize exposure to gantenerumab and to prospectively examine the correlation between amyloid-lowering and clinical outcomes.


Funding: This study was sponsored by F. Hoffmann-La Roche Ltd, Basel, Switzerland.

Acknowledgments: We would like to thank all the participants and their families, the investigators and site staff, and the entire study team for their time and commitment to the SCarlet RoAD and Marguerite RoAD OLE studies. Medical writing support was provided by Joshua Quartey, BSc, of Health Interactions and was funded by F. Hoffmann-La Roche Ltd.

Conflict of interest disclosures: GK, PD, GAK, CH, DA-S and PF were full-time employees of F. Hoffmann-La Roche Ltd during the conduct of the study. GK, PD, GAK, CH, DA-S, NV and PF are shareholders in F. Hoffmann-La Roche Ltd. AD and NV were full-time employees of Roche Products Ltd during the conduct of the study. AD is currently employed at the MRC Clinical Trials Unit at UCL. MB and RD are full-time employees and shareholders in F. Hoffmann-La Roche Ltd and Genentech Inc. CH has an Alzheimer’s disease-related patent planned which is relevant to this study.

Ethical standards: Institutional Review Boards (IRBs) approved the SCarlet RoAD and Marguertie RoAD studies, and all participants gave informed consent before participating.

Data sharing statement: Qualified researchers may request access to individual patient-level data through the clinical study data request platform: https://vivli. org. Further details on Roche’s criteria for eligible studies are available here: For further details on Roche’s Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, see here: development/who_we_are_how_we_work/clinical_trials/our_commitment_to_ data_sharing.htm

Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.


1. Alzheimer’s Association. 2019 Alzheimer’s disease facts and figures. Alzheimers Dement 2019; 15:321-387.
2. Wang S, Mims PN, Roman RJ, et al. Is beta-amyloid accumulation a cause or consequence of Alzheimer’s disease? J Alzheimers Parkinsonism Dement 2016; 1:007.
3. Lee JC, Kim SJ, Hong S, et al. Diagnosis of Alzheimer’s disease utilizing amyloid and tau as fluid biomarkers. Exp Mol Med 2019; 51:1-10.
4. Mucke L & Selkoe D. Neurotoxicity of amyloid β-protein: Synaptic and network dysfunction. Cold Spring Harb Perspect Med 2012; 2:a006338.
5. Delbeuck X, Van der Linden M, Collette F. Alzheimer’s disease as a disconnection syndrome? Neuropsychol Rev 2003; 13:79-92.
6. Ishii K. Amyloid positron emission tomography in the therapeutic strategies for Alzheimer’s disease. Brain Nerve 2017; 69:809-818.
7. Klein G, Delmar P, Voyle N, et al. Gantenerumab reduces amyloid-β plaques in patients with prodromal to moderate Alzheimer’s disease: A PET substudy interim analysis. Alzheimers Res Ther 2019; 11:101.
8. Ostrowitzki S, Lasser RA, Dorflinger E, et al. A phase III randomized trial of gantenerumab in prodromal Alzheimer’s disease. Alzheimers Res Ther 2017; 9:95.
9. Abi-Saab D, Andjelkovic M, Pross N, et al. MRI findings in the open label extension of the Marguerite RoAD study in patients with mild Alzheimer’s disease. J Prev Alz Dis 2017; 4:339 (P336).
10. Fleisher AS, Chen K, Liu X, et al. Using positron emission tomography and florbetapir F18 to image cortical amyloid in patients with mild cognitive impairment or dementia due to Alzheimer disease. Arch Neurol 2011; 68:1404-1411.
11. Barthel H, Gertz HJ, Dresel S, et al. Cerebral amyloid-β PET with florbetaben (18F) in patients with Alzheimer’s disease and healthy controls: A multicentre phase 2 diagnostic study. Lancet Neurol 2011; 10:424-435.
12. Klunk WE, Koeppe RA, Price JC, et al. The Centiloid Project: Standardizing quantitative amyloid plaque estimation by PET. Alzheimers Dement 2015; 11:1-15.e11-14.
13. Navitsky M, Joshi AD, Kennedy I, et al. Standardization of amyloid quantitation with florbetapir standardized uptake value ratios to the Centiloid scale. Alzheimers Dement 2018; 14:1565-1571.
14. von Rosenstiel P, Gheuens S, Chen T, et al. Aducanumab titration dosing regimen: 24-month analysis from PRIME, a randomized, double-blind, placebo-controlled Phase 1b study in patients with prodromal or mild Alzheimer’s disease. Neurology 2018; 90:Abstract S2.003.
15. Swanson C, Zhang Y, Dhadda S, et al. Treatment of early AD subjects with BAN2401, an anti-Aβ protofibril monoclonal antibody, significantly clears amyloid plaque and reduces clinical decline. Alzheimers Dement 2018; 14 (Suppl):P1668. DT-1601-1607.



P. Daunt1, C.G. Ballard2, B. Creese2, G. Davidson3, J. Hardy4, O. Oshota1, R.J. Pither1, A.M. Gibson1 for the Alzheimer’s Disease Neuroimaging Initiative*


1. Cytox Limited, Manchester, UK; 2. University of Exeter Medical School, Exeter, UK; 3. Ledcourt Associates Limited, Cambridge, UK.; 4. UK Dementia Research Institute, University College London, London, UK.

Corresponding Authors: Alex Gibson, Cytox Ltd., John Eccles House, Robert Robinson Avenue, Oxford Science Park, Oxford, OX4 4GP, United Kingdom. Email: Tel:+44 (0)1865 338018

J Prev Alz Dis 2021;1(8):78-83
Published online November 11, 2020,



BACKGROUND: There is a clear need for simple and effective tests to identify individuals who are most likely to develop Alzheimer’s Disease (AD) both for the purposes of clinical trial recruitment but also for improved management of patients who may be experiencing early pre-clinical symptoms or who have clinical concerns.
OBJECTIVES: To predict individuals at greatest risk of progression of cognitive impairment due to Alzheimer’s Disease in individuals from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) using a polygenic risk scoring algorithm. To compare the performance of a PRS algorithm in predicting cognitive decline against that of using the pTau/Aß1-42 ratio CSF biomarker profile.
DESIGN: A longitudinal analysis of data from the Alzheimer’s Disease Neuroimaging Initiative study conducted across over 50 sites in the US and Canada.
SETTING: Multi-center genetics study.
PARTICPANTS: 515 subjects who upon entry to the study were diagnosed as cognitively normal or with mild cognitive impairment.
MEASUREMENTS: Use of genotyping and/or whole genome sequencing data to calculate polygenic risk scores and assess ability to predict subsequent cognitive decline as measured by CDR-SB and ADAS-Cog13 over 4 years
RESULTS: The overall performance for predicting those individuals who would decline by at least 15 ADAS-Cog13 points from a baseline mild cognitive impairment in 4 years was 72.8% (CI:67.9-77.7) AUC increasing to 79.1% (CI: 75.6-82.6) when also including cognitively normal participants. Assessing mild cognitive impaired subjects only and using a threshold of greater than 0.6, the high genetic risk participant group declined, on average, by 1.4 points (CDR-SB) more than the low risk group over 4 years. The performance of the PRS algorithm tested was similar to that of the pTau/Aß1-42 ratio CSF biomarker profile in predicting cognitive decline.
CONCLUSION: Calculating polygenic risk scores offers a simple and effective way, using DNA extracted from a simple mouth swab, to select mild cognitively impaired patients who are most likely to decline cognitively over the next four years.

Key words: Polygenic risk, cognitive decline, Alzheimer’s disease.



Alzheimer’s disease (AD) is the most common form of dementia with nearly 50 million people affected globally and an estimated economic impact of $818 billion (1).
As well as having a clear heritable component (2), AD is genetically complex. Neuropathologically, the disease is characterized by extracellular senile plaques containing β -amyloid (Aβ) and intracellular neurofibrillary tangles containing hyperphosphorylated tau protein. A relatively small number of dominant mutations in the amyloid precursor and presenilin genes are known to cause early onset Alzheimer’s disease. Over the past two decades, genome wide association studies (GWAS) have identified multiple loci and single nucleotide polymorphisms (SNPs) associated with the much more common, late-onset or sporadic form of the disease (LOAD) (3-5). Apolipoprotein E (ApoE) is a major cholesterol carrier that supports lipid transport and injury repair in the brain. The ε4 allele of ApoE (ApoE4) has been found to be a primary genetic risk factor for AD, associated with increased risk for both early-onset AD and LOAD (6, 7). Although only 20-30% of humans are ApoE4 carriers, these individuals account for up to 60% of all Alzheimer’s disease cases. In addition, ApoE4 is associated with an increased risk of lower age of onset (8, 9), making this an important subset of the population at high risk of developing AD.
Development of polygenic risk scoring (PRS) algorithms that can capture all the genetic contribution towards the risk of developing AD (10) is an attractive strategy to allow better clinical trials for AD prevention. PRS approaches have demonstrated accuracies of between 75 and 84% for predicting onset of AD when including APOE, sex and age in addition to PRS (11), In particular the PRS approach as developed by Escott-Price et al (12), is built as a sum of the weighted contributed of 10,000s of SNPs where the weights are the β-coefficients of each SNP association with the disease. In contrast to other PRS algorithms, where fewer SNPs have been used (for example just 31 SNPs (13)) this approach includes SNPs that are not considered as having genome wide significance in GWAS studies. However, inclusion of this vastly increased number of variants which alone carry sub-threshold significance provides an additive contribution to the overall performance that may be substantive and also reduce risk that performance is not lost when being applied across different cohorts.
Until now the analyses performed using this particular approach have been carried out to predict those individuals diagnosed with AD or MCI (14) versus those who are cognitively normal, though PRS algorithms have been used to look at a variety of AD pathology and risk by Altmann et al (15). Here we look to see how the PRS performs in predicting those individuals most likely to decline cognitively independent of whether they have cognitive impairment on entry or not.
Currently, the most frequently used approach to enrich clinical trial recruitment with participants who have increased likelihood of progressive cognitive and functional decline has been to focus on identifying individuals who are positive for amyloid biomarkers. In addition, measurement of tau in CSF often with amyloid levels, is increasing in use. We therefore also compare the ability to predict decline using PRS against that of using CSF tau and amyloid measurements.



Data used in the preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database ( The ADNI was launched in 2003 as a public-private partnership, led by Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test whether serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early Alzheimer’s disease (AD).

Sample Description

ADNI is an on-going longitudinal study that has been established to develop methods for early detection of AD and subsequent monitoring of disease trajectory using clinical, imaging and genetic data (16). Data for this analysis was collected from 515 participants, who entered the study with a diagnosis of Mild Cognitive Impairment or considered cognitively normal. In addition, 47 individuals diagnosed with AD were used to check the algorithm was performing as expected to differentiate AD cases from cognitively normal controls. All participants in addition to having suitable genetic data available had at least 4 years’ worth of follow up cognitive testing and imaging scans. Upon entry into the study 199 individuals were cognitively normal and 316 diagnosed as MCI. The average age of the total group was 73.2 years, with the CN group being on average approximately 3 years older than those diagnosed with MCI (75.1y and 72.0 y respectively). ADAS-Cog13 scores for cognitively normal and MCI groups upon entry were 9.0 and 14.9 respectively and at the 4 year assessment the average scores obtained were 9.6 and 19.8, clearly showing that on average the MCI group continued to decline compared with little change in the average score of the CN group. CSF biomarker data were not available for all participants, so analyses performed to compare PRS with biomarker (tau and amyloid) as a predictor for subsequent cognitive decline were carried out on 290 MCI subjects. Table 1a shows the classification of the ADNI dataset at baseline and changes to cognitive performance as measured by ADAS-Cog13 after 4 years. Similarly, Table 1b describes the sub-group that also had CSF biomarker data available.

Table 1. Characteristics of participants used in the analyses


Genotyping Procedures and Quality Control

The ADNI samples were genotyped using with Whole Genome Sequencing and/or the Illumina Omni 2.5M BeadChip array. Quality control checks were performed using PLINK software ( (17). Checks included exclusion of SNPs with missingness greater than 0.02, minor allele frequency of less than 0.01 and SNPs with Hardy-Weinberg equilibrium p-value less than 1 x 10-6 were also excluded. After such checks 8,990,292 SNPs were left for analysis of which approximately 114,000 were used as part of the polygenic risk scoring algorithm.

Calculation of Polygenic Risk Scores

A specifically built, proprietary software called SNPfitR was used for all subsequent PRS calculations. The PRS calculations are based on a pre-determined logistic regression model based on the modelling of the association between the incidences of variants within a large panel of SNPs with a known links to AD to the presence of the disease in a substantial cohort of subjects (Escott-Price et al12). Subject age, gender and presence of both APOE4 and APOE2 proteins are included as covariates. The software calculates the normalised sum of the individual scores weighted by their effect sizes for each SNP, adds the values for the covariates and derives the predicted risk from the model equation.
Effect sizes were determined from the IGAP study. The score contribution from SNPs with missing values were imputed based on the population frequency of the effect allele for that SNP.

Statistical Analysis

The polygenic risk scores generated were exported for the analysis presented.
The statistical analysis software package JMP 14.1.0 was used to carry out all data manipulation and analysis. The ROC analysis and AUC calculations were performed using the add in ‘Model Classification Explorer’. Values were cross checked with the AUC calculations carried out in the software.



Association of AD PRS with onset of Alzheimer’s Disease

As a check that the algorithm was performing as expected polygenic risks scores were also generated for 47 Alzheimer Disease cases and compared with those generated for the 199 cognitively normal individuals. The accuracy of prediction of clinical AD cases (n = 47) versus cognitively normal control (n = 199) was 80% AUC. Furthermore, as shown in Figure 1, PRS score is clearly associated with tau levels and, as expected, diagnostic classification. In this analysis, data were plotted in this heat density format to illustrate a clear relationship between the pTau/Aß1-42 ratio (Hansson et al18) and the stage of disease. It can be seen that those subjects classified at AD, late-MCI (LMCI) or early-MCI (EMCI), predominantly sit in the group with a PRS of 0.6 or above, whereas cognitively normal (CN) individuals tend to be in the 0.6 or lower range of the PRS scores. Importantly, as shown in Figure 2, there is a substantial overlap between different APOE genotype sub-groups. As expected, ApoE4 carriers fall within the higher end (0.6 and above) end of the PRS range and ApoE2 carriers at the lower end. However, ApoE3/3 homozygotes, representing some 60% of the Caucasian population, span the high and low ends of the PRS spectrum, thereby demonstrating the additional genetic risk information provided by the risk algorithm over APOE alone.

Figure 1. Density plots showing relationship between full PRS, pTau/Aβ(1-42) ratio and diagnostic classification (CN, EMCI, LMCI, AD)

Figure 2. Distribution of risks scores across the MCI population (n = 316) coloured by APOE genotype


Association of AD PRS with cognitive decline from an MCI baseline

Rather than using specific clinical diagnosis to categorise cases as previously used, predicting the cognitive decline likely due to AD from either an MCI or cognitively normal baseline was performed. Table 2 shows the predictive accuracy of identifying those individuals who are most likely to decline as measured by ADAS-Cog13 testing from an MCI or CN baseline, irrespective of cognitive status at baseline. The analyses were performed defining significant progression as 5-point, 10-point or 15-point decline at their 4 year follow up assessment. Though we report the accuracy for predicting decline from a cognitively normal state, the number of individuals that decline significantly within the time period is relatively low and thus results cannot be considered statistically significant. However, of the 316 individuals who entered the study with an MCI diagnosis, significant numbers had declined by at least 5 points (107), 10 points (61) and 15 points (39) on the ADAS-Cog13 scale to allow meaningful prediction accuracies to be measured. In addition to the full PRS algorithm (APOE + PRS + Age + Sex) being used to generate risk scores, prediction accuracies based on APOE status alone and total genetic risk (APOE + PRS) were calculated. The best prediction accuracy is seen for testing cases that have declined by at least 10 points at 4 years versus those that have remained cognitively stable (< 5-point decline) with an AUC of 74.8%, compared with 67.4% for APOE alone and 73.5% for APOE + PRS. A similar performance is seen when predicting those individuals with 15-point decline. In both analyses all those individuals had polygenic risk scores in the upper half of the distribution. When looking at smaller changes in cognitive performance over 4 years, addition of the polygenic risk score term to APOE did not impart greater performance. In all cases addition of age and gender as co-variates did not add any additional predictive performance in this particular group presumably due to the particular age and sex distribution between the CN and MCI groups in this particular cohort. Given that the mean age of those that declined and those that remained relatively stable were similar the contribution provided by age to the overall risk score for both groups would, in turn, be broadly equivalent.

Table 2. Performance of polygenic risk scoring algorithm to predict cognitive decline up to 4 years after entry to study

* PRS = all risk associated with the genetics other than that contribution from APOE

To evaluate whether the full algorithm could predict cognitive decline as defined by predetermined thresholds and be compared with that predicted by CSF biomarker status (figure 3), the MCI population where both genetics, CSF and CDR-SB assessment data were available was studied (n=290). There was a significant difference in progression (as defined by CDR-SB) between patients whose risk score was greater than 0.6 (n=196) versus the group whose score was less 0.6 (n=94) as early as 6 months after baseline assessment. 0.6 was chosen as a threshold based on an optimal balance between sensitivity and specificity (data not shown here) High risk patients progressed, on average, by approximately 1 point over 24 months and 2 points over 48 months compared with low risk patients who on average decline 0.2 and 0.4 points over the same timepoints. A similar evaluation was carried out to compare the predictive performance using CSF biomarker positivity as determined by a pTau/Aβ(1-42) ratio using the cut off of 0.02818 and CSF Aβ(1-42) with a threshold of 880pg/mL18. Again, there was a significant difference in progression between biomarker positive and negative patients. pTau/Aβ(1-42) ratio positive patients progressed, on average, by 1.1 and 2.9 points over 24 and 48 months respectively, whereas there was an average decline of 0.1 and 0.2 points for the negative group. Similarly using Aβ(1-42) CSF levels only, the amyloid positive group progress by 1 and 2.6 points at 24 and 48 months respectively whilst the negative group only progressed by 0,3 points on average over 48 months. The performance of the PRS was broadly similar to that of either CSF biomarker measurement in identifying those subjects at highest and lowest risk of cognitive decline on the CDR-SB scale. Furthermore, a similar analysis was performed on APOE3 homozygote individuals (n=125) only (figure 3). Again, using a threshold of 0.6 to determine the high risk group (n =49), a difference a measured by a change of CDR-SB between the two groups was shown 12 months with a clear difference at 36 months. The high risk group declined, on average by 1.5 points at 36 months compared with the low risk group who only declined, on average, by 0.5 points.

Figure 3. Time course of clinical progression in patients with MCI over 48 months. Average with standard errors by PRS group (orange >0.6; blue <0.6 at baseline) for all APOE genotypes and for APOE homozygotes only, pTau/Aβ(1-42) group (orange > 0.028; blue <0.028) and Aβ(1-42) (orange < 880pg/mL; blue >880pg/mL)



This study was designed to demonstrate the potential utility of a specific PRS algorithm in identifying individuals at highest risk of clinically significant cognitively decline within a specific time period. Previously most studies reporting the use of PRS approaches have been used to differentiate two populations with clearly different clinical phenotypes (AD versus CN) and thus not necessarily demonstrating how this approach could be used prospectively. The results of these analyses show that using polygenic scoring algorithms which have been designed to understand the genetic risk of future onset of Alzheimer’s Disease, can be applied to enrich trial populations with individuals who are more likely to decline cognitively within a certain time period.
Though APOE genotype remains an important genetic risk factor within this cohort, it is clear there is an additional genetic component that should be considered in assessing genetic risk. This will subsequently allow further risk stratification within APOE genotypes such as identifying APOE3 homozygotes who are at relatively higher risk even compared with some APOE4 carriers. This has implications in the design of clinical trials where in many trial designs possession of at least one APOE4 allele is used as an enrichment strategy in prevention trials.
It is broadly accepted that CSF-tau/amyloid ratios are a reasonable predictor of future cognitive decline (18-20) though definitive studies have yet to be performed, and testing for amyloid alone, via PET imaging or CSF remains the standard method to enrich trials with patient most likely to decline cognitively. This study shows that PRS predictions, are able to perform to a similar level in predicting further progression, as measured by CDR-SB, in patients who have an MCI diagnosis. Importantly this genetic risk assessment can be more easily accessed (cost and patient burden) through whole blood or mouth swab testing, rather than by performing an invasive lumbar puncture procedure and subsequent CSF testing; such invasive procedures are particularly challenging in elderly subjects who may be relatively cognitively robust (early MCI or prodromal). The PRS algorithm therefore represents a promising method to facilitate broad screening of potential trial participants in order to identify those at highest risk for cognitive decline. Further confirmatory testing, via the use of more invasive and expensive CSF and/or PET imaging, could then be focussed on a significantly reduced number of individuals for final patient recruitment decisions. Furthermore, a combination of PRS and tau levels (underlying genetic risk coupled with manifestation of that risk through pathology) may provide a more optimal model for likelihood of subsequent onset of AD in early symptomatic or pre-symptomatic individuals. Whilst there may be specific reasons why amyloid or tau biomarkers may be required for clinical trials focussing on treatments specifically targeting amyloid or tau, PRS may have advantages for therapies with different treatment targets independent of potential mechanisms.
Further studies will be important to determine the added value of combing amyloid/tau and PRS markers and to fully determine the utility of PRS in predicting cognitive decline in cognitively normal individuals.
It is recognised that this work has considered genetic risk together with age and gender in developing a model for predicting further development of cognitive symptoms and so does not consider other risk factors that are known to influence onset and development of disease. Combining both genetic and lifestyle risk factors for the purposes of identifying those individuals most at risk of Alzheimer’s Disease is likely to add further to the predictive accuracy.

Study Limitations

This study is not without limitations, with sample size being the primary shortcoming. This was particularly relevant in evaluating the APOE E3 homozygote only sub-group. Furthermore, studies with larger sample sizes across all diagnostic categories, including those declining from a cognitively normal baseline, is important to understand broader utility. As with most studies of this nature, observing similar performance in alternative cohorts is important and is critical towards the understanding and confirmation of polygenic risk score assessment for use in clinical trial recruitment and in clinical practice.


*Data used in preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database ( As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at:

Acknowledgements: Data collection and sharing for this project was funded by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (National Institutes of Health Grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH-12-2-0012). ADNI is funded by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: AbbVie, Alzheimer’s Association; Alzheimer’s Drug Discovery Foundation; Araclon Biotech; BioClinica, Inc.; Biogen; Bristol-Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann-La Roche Ltd and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & Development, LLC.; Johnson & Johnson Pharmaceutical Research & Development LLC.; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics. The Canadian Institutes of Health Research is providing funds to support ADNI clinical sites in Canada. Private sector contributions are facilitated by the Foundation for the National Institutes of Health ( The grantee organization is the Northern California Institute for Research and Education, and the study is coordinated by the Alzheimer’s Therapeutic Research Institute at the University of Southern California. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of Southern California. We also acknowledge Prof. Julie Williams, Prof. Valentina Escott-Price, Dr Rebecca Sims and Dr Eftychia Bellou from the University of Cardiff for their advice on adaptation and implementation of of the polygenic risk algorithm.

Funding: Funding for this study was provided under an Innovate UK grant (Project No 5195).

Conflict of Interest: P. Daunt, A.Gibson, O.Oshota and R. Pither are all employees of Cytox Ltd. G. Davidson received payment from Cytox Ltd. for work done both within and outside the scope of this article.

Ethical Standards: The ADNI protocols were approved by all the Institutional Review Boards of the participating institutions. Only data from volunteers who had provided written informed consent were used to complete these analyses.

Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.



1. Prince M, Wimo A, Guerchet M et al. World Alzheimer Report 2015: the Global Impact of Dementia – An analysis of prevalence, incidence, cost and trends. Alzheimer’s Dis Int 2015; 84:425
2. Gatz M, Reynolds CA, Fratiglioni L, et al. Role of genes and environments for explaining Alzheimer’s disease. Arch Gen Psychiatry 2006; 63: 168-174
3. Lambert JC, Ibrahim-Verbaas CA, Harold D, et al. Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nat Genet 2013; 45: 1452-1458
4. Naj, AC, Jun G, Beecham, GW et al. Common variants at MS4A4/MS4A6E, CD2AP, CD33 and EPHA1 are associated with late-onset Alzheimer’s disease. Nat Genet 2011; 43: 436-441
5. Seshadri, S, Fitzpatrick AL, Arfan Ikram M et al. Genome-wide analysis of genetic loci associated with Alzheimer disease. JAMA 2010; 303: 1832-1840
6. Corder EH, Saunders AM, Strittmatter WJ, et al. Gene Dose of Apolipoprotein-E Type-4 Allele and the Risk of Alzheimers-Disease in Late-Onset Families. Science1993;261:921–923.
7. Farrer LA, Cupples LA, Haines JL, et al. Effects of age, sex, and ethnicity on the association between apolipoprotein E Genotype and Alzheimer Disease: a Meta-analysis. JAMA 1997;278:1349–1356.
8. Blacker D, Haines JL, Rodes L, et al. ApoE-4 and age at onset of Alzheimer’s disease: the NIMH genetics initiative. Neurology 1997;48;139-147
9. Bonham LW, Geier EG, Fan CC, et al. Age-dependent effects of APOE ε4 in preclinical Alzheimer’s disease. Ann Clin Transl Neurol 2016;3;668-677
10. Stocker H, Möllers T, Perna L, et al, The genetic risk of Alzheimer’s disease beyond APOE ε4: systematic review of Alzheimer’s genetic risk scores. Transl Psych 2018;8;166-174
11. Escott-Price V, Myers A, Huentelman AJ, et al, Polygenic risk score analysis of pathologically confirmed Alzheimer’s disease. Ann Neurol 2017;82;311-314
12. Escott-Price V, Sims R, Williams J et al, Common polygenic variation enhances risk prediction for Alzheimer’s disease. Brain 2015;138;3673-3684
13. Desikan RS, Fan CC, Wang Y et al, Genetic assessment of age-associated Alzheimer disease risk: development and validation of polygenic hazard score. PLoS Med 2017 e1002258
14. Escott-Price V, Leonenko G, Sims R et al, Polygenic risk and hazard scores for Alzheimer’s disease prediction. Ann Clin Transl Neurol 2019;6;456-465
15. Altmann A, Schott JM, Scelsi et al, A comprehensive analysis of methods for assessing polygenic burden on Alzheimer’s disease pathology and risk beyond APOE. Brain Commun 2020;2;fcz047
16. Peterson R, Aisen PS, Beckett LA et al, Alzheimer’s Disease Neuroimaging Initiative (ADNI): clinical characterization. Neurology 2010;74;201-209
17. Chang CC, Chow CC, Tellier LCAM et al, Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience 2015;4;s13742
18. Hansson O, Seibyl J, Stomrud et al, CSF biomarkers of Alzheimer’s disease concord with amyloid-β PET and predict clinical progression: A study of fully automated immunoassays in BioFINDER and ADNI cohorts. Alzhemers Dement 2018;14;1470-1481.
19. Ritchie C, Smailagic N, Noel-Storr et al, CSF tau and CSF tau/ABeta ratio for the diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2017;3;CD10803.
20. Wolfsgruber S, Polcher A, Koppara A et al. Cerebrospinal fluid biomarkers and clinical progression in patients with subjective cognitive decline and mild cognitive impairment. Alzheimers Dis 2017;58;939-950.


J. Park1, Y. Kim2, M. Woo1

1. School of Sport and Exercise Science, University of Ulsan, Nam-gu Ulsan, Korea; 2. Department of Physical Education, Sejong University, Gunja-dong, Gwangjin-gu, Seoul, Korea

Corresponding Author: School of Sport and Exercise Science, University of Ulsan, 93 Daehak-ro, Nam-gu, Ulsan 44610, Korea, Tel: +82-52-259-2380, Fax: +82-52-259-1696, E-mail:

J Prev Alz Dis 2021;1(8):92-99
Published online October 19, 2020,


Aims: This study examined the interactive effect of physical fitness and Apolipoprotein e4 on intelligence and cortical networking in adolescents.
Methods: Participants were middle school students consisting of 10 and 8 high- and low-fit e4 carriers (e4+), respectively, and 14 and 10 high- and low-fit non-carriers (e4−), respectively. Inter- and intra-hemispheric coherences were calculated to examine cortico-cortical communication during intelligence test.
Results: Coherence in low-fit e4+ was lower than in high-fit e4+, while coherence in low-fit e4- was similar to or higher than in high-fit e4-.
Conclusion: the presence of the e4 allele can decrease neural networking 50-60 years before Alzheimer’s disease onset: however, physical fitness may compensate for the negative impact of genotype. Moreover, the beneficial effects of physical fitness may differ depending on functional states of the adolescent brain according to the presence of the e4 allele.

Key words: Physical fitness, apolipoprotein, Alzheimer’s disease, adolescent, EEG coherence.



Alzheimer’s disease (AD) is a progressive neurodegenerative brain disease leading to cortical dysfunction and anatomical atrophy (1). Genetic factors account for approximately 40–50% of AD cases (2). Genes encoding apolipoprotein E (ApoE) are associated with AD. There are three main ApoE genotypes: e2, e3, and e4. Of these, e4 is related to the accumulation of beta amyloids and formation of neurofibrillary tangles (3). Approximately 70% of patients with AD have least one e4 allele (4). Furthermore, as the number of e4 alleles present in individuals increases from 0 to 2, the incidence rate of AD increases from 20% to 90%, with the average age of onset lowering from 84 to 64 years (3). These findings indicate that e4 is a major risk factor for AD.
A previous neuroimaging study has investigated the relationship between e4 and brain function in healthy middle-aged subjects. They found that e4 carriers show hypoactivation in the hippocampus, prefrontal, temporal, and parietal regions when compared with e4 non-carriers (5). In addition, college-aged e4 carriers have lower cerebral blood flow than e4 non-carriers (6). In addition, children and adolescent e4 carriers have thinner cerebral cortices (7) and lower brain networking (8) than e4 non-carriers. These findings suggest that individuals with a genetic risk factor for dementia begin to exhibit anatomical and functional abnormalities in the brain several decades before the onset of AD. To date, there are no effective treatments that cure AD; therefore, it is crucial to establish preventive measures for those with genetic risk factors.
Exercise has positive effects on physical fitness (ES = 0.69) and cognitive function (ES = 0.57) in patients with AD (9). Six months of aerobic exercise can reduce the symptoms of AD (10). Interestingly, the beneficial effect of physical activity on cognitive function has been observed in young e4 carriers decades before AD onset. Studies that have examined the benefits of physical activity on e4 carriers have reported that cognitive function and cortical activation in both middle-aged (11, 12) and college-aged (13) groups with high activity levels were higher than in e4 carriers with low activity levels and not significantly different from that of e4 non-carriers. Therefore, physical activity could offset the cognitive decline associated with a genetic predisposition for AD by helping to maintain normal brain functioning.
To date, most studies that have examined the relationship between AD-related genes and physical activity or fitness have been conducted on adults (11-13). Few studies have assessed the role of these genetic risk factors in adolescents who are undergoing rapid neurobiological development. In contrast to the simple cognitive tasks that have been widely adopted in previous studies, this study employed a non-verbal test of intelligence to investigate the relationship between physical fitness and AD-related genes. This test excludes the influence of acquired experiences and requires sufficient cognitive effort. In addition, we conducted coherence analysis, which is known to have the highest predictive validity of intelligence among EEG markers, to measure neural efficiency during cognitive tasks (14).
This study investigated the benefits of physical fitness in adolescent e4 carriers and non-carriers by examining their neural efficiency during a non-verbal intelligence test. We hypothesized that high-fit adolescent e4 carriers would show higher non-verbal intelligence and more active cortical networking than low-fit e4 carriers.


Materials and Methods


Among a total of 1500 students, 100 high-fit (level 1) and 100 low-fit (level 4 or 5) were selected predicated on the Physical Activity Promotion System (PAPS). Students who were willing to participate and received consent from their parents received genetic testing. According to the genetic test results, the final participant lists included in the analyses–excluding those eliminated due to EEG data artifacts or withdrawal from the study–contained 10 high-fit e4+ (13.9 ± 0.74 years), 8 low-fit e4+ (14.13 ± 0.64 years), 14 high-fit e4− (13.92 ± 0.86 years), and 10 low-fit e4− (13.80 ± 0.79 years) carriers. This study was conducted conforming to the provisions of the Declaration of Helsinki. All participants and their legal representatives signed a written consent form and were informed that they were free to withdraw from the study at any time.

Experimental Task and Apparatus

Physical fitness measurement

The Physical Activity Promotion System (PAPS) was developed by the Ministry of Education of the Republic of Korea to evaluate health and fitness of children and adolescents. The PAPS measures four physical fitness factors: cardiovascular endurance, muscle power, flexibility, and agility, which are rated from levels 1 (high-fit) to 5 (low-fit) according to performance. The fitness level for each factor is determined according to the fitness evaluation guideline of the PAPS (15). Cardiovascular endurance was measured as the time (minutes) taken to complete 1.6 km as quickly as possible by running and walking. Muscle power was determined using a hand grip strength test, recording the highest score. Flexibility was assessed with a sit and reach test, where participants bent the upper body forward without bending the knees. The record maintained for 2 seconds was measured twice and the highest score was evaluated. Agility was assessed as the time (seconds) taken to complete the 50 m course. The fitness level of participants was calculated by averaging the levels of the four factors (Table 1). In this study, level 1 was defined as the high-fit group, and level 4 and 5 were categorized as the low-fit group. Measurements were recorded by certified, professional health and fitness instructors, and the reliability (r) of all test items was 0.69–0.83 (16).

Table 1. Participant’s body composition and physical fitness


Comprehensive Test of Nonverbal Intelligence-Second Edition

The Comprehensive Test of Nonverbal Intelligence-Second Edition (CTONI-2) is a non-verbal intelligence test designed to measure high-order cognitive abilities from ages 6–89 with minimal influence of educational experience or linguistic ability. The CTONI-2 measures three cognitive abilities: (1) analogical reasoning, (2) categorical classification, and (3) sequential reasoning. Each of these abilities is assessed using 6 subtests: (1) pictorial analogies, (2) geometric analogies, (3) pictorial categories, (4) geometric categories, (5) pictorial sequences, and (6) geometric sequences. In compliance with the CTONI-2 manual, a computerized version was created using the LabVIEW program. The questions were automatically presented on the computer screen and participants were instructed to respond by pressing the button of the number they thought was correct as quickly as possible. The stimuli (questions) were presented with one-second inter-trial intervals. The task was automatically terminated after three consecutive incorrect responses and the next question appeared. The order of the six subtests was randomized to exclude any order effect. The accuracy of the CTONI-2 was calculated as the number of correct responses divided by the total number of questions multiplied by 100. The Response Time (RT) was determined as the time elapsed from stimulus presentation to response completion for correct responses.


DNA from buccal cells was collected by swabbing the mouths of each participant and amplified using polymerase change reaction (PCR). PCR was performed using the primers [F-5’ GTC TCC TTC TCT GGG CCT CT 3’; R-5’ CAC CTC GTC CAG GCG GTC G 3’] and run in a reaction volume of 25 μl using 1 μl, or 10 picomole, of DNA. The thermal cycling profile consisted of incubation at 95 °C for 15 minutes followed by 30 cycles of 95 °C for 20 s, 60 °C for 40 s, and 72 °C for 40 s, with a final 5 min incubation at 72 °C. Purified PCR products underwent a sequencing reaction using BigDye® Terminator v3.1 Cycle Sequencing Kits and were sequenced in an ABI 3730XL DNA Analyzer. Genotypes were determined from nucleotide sequences.

Experimental Design and Procedure

Genetic testing was performed for participants who submitted a consent form, including parental consent. Using a cotton swab to collect buccal cells, the collected samples were sealed and sent to a genetic testing company. e4 carriers were defined as having one or more e4 alleles and no e2 alleles (e3/e4, e4/e4), while e4 non-carriers were randomly selected from individuals without an e4 allele. Students who met the selection criteria of the genetic and physical fitness requirements were informed of their results, the cognitive test procedures (e.g., the intelligence test and EEG measurement), and compensation. A consent form was obtained again from parents and students who wished to participate in further testing.
After arriving at the laboratory, participants were fitted with electrode caps. Gel was injected into the EEG electrodes and impedances of all electrodes were kept below 5 KΩ. Participants were provided with examples of the CTONI tasks to help them understand the test and the task order was randomized. Participants were instructed to respond as quickly and accurately as possible to the test items presented on the screen. It took approximately 30 min to complete the 6 test items.

EEG Data Collection and Coherence Analysis

During the CTONI-2 test, scalp EEG was recorded from F3, F4, C3, C4, P3, P4, T3, T4, O1, and O2 of a stretchable Lycra cap (Electro-Cap International Inc., Eaton, OH, USA) according to the international 10–20 system (17). The sampling rate was 256 Hz using the WEEG-32 system (LXE 3232-RF; Laxtha, Daejeon, South Korea). The left (A1) and right (A2) earlobes were connected to serve as reference electrodes, and the frontal midline (Fpz) was used as the ground site. The impedance of all electrodes was kept below 5 kΩ. A 1–100 Hz band-pass filter and 60-Hz notch-filter were used for the 24 dB/octave roll-off to remove artifacts. The EEG data measured during CTONI were epoched from each stimulus onset to response (from the point of presentation of the question to the participant’s response). All epoched EEG data, signals exceeding ±100 μV were excluded from the analysis. Ocular correction was performed to exclude any noise including eye blinking (EOG). Coherence analyses were conducted to investigate inter-hemispheric coherence from F3-F4, C3-C4, T3-T4, P3-P4, and O1-O2 electrode pairs, and intra-hemispheric coherence from F3-C3, F3-P3, F3-O1, F3-T3, C3-P3, C3-O1, C3-T3, P3-O1, P3-T3, O1-T3, F4-C4, F4-P4, F4-O2, F4-T4, C4-P4, C4-O2, C4-T4, P4-O2, P4-T4, and O2-T4 electrode pairs.
Coherence is defined as ︱Cxy(f) ︱2 , where:

with Xi(f) and Yi(f) representing the Fourier transforms that take the time series for electrode sites X and Y, respectively. EEG coherence values were calculated at 1-Hz frequency bins and averaged across frequency bands to obtain power values for the bandwidths of interest (8–10 Hz, 10–13 Hz, and 13–22 Hz). The Fisher z-transformation was performed for all coherence values prior to analysis to ensure normal distribution (18).

Statistical Analysis

Two way 2 (Physical fitness: high, low) × 2 (genotype: e4+, e4−) ANOVAs conducted. Dependent variables were CTONI scores, accuracy, reaction time, and inter- and intra- hemispheric EEG coherence values. All variables were analyzed by SPSS 18, The α value was set to 0.05.



CTONI IQ scores, subtest converted scores, accuracy, and RT

Analyses of the K-CTONI-2 IQ scores, subtest converted scores, accuracy, and RT revealed no significant main effect or interaction effect.

Table 2. Statistical results of CTONI-2 IQ, t-scores, response accuracy, and Response Time (RT) of subtest in low-fit and high-fit groups

Interaction effect of physical fitness and genotype in intra-hemispheric coherence

In the analysis of intra-hemispheric coherence during CTONI task performance, significant fitness and genotype interactions were observed for geometric analogies at high alpha C3-P3 (F(1, 37) = 6.40, p = 0.016) and beta C4-O2 (F(1, 37) = 4.27, p = 0.046); for pictorial categories at high alpha C4-O2 (F(1, 37) = 4.10, p = 0.05), P4-O2 (F(1, 37) = 4.34, p = 0.044), beta P3-O1 (F(1, 37) = 6.05, p = 0.019), P3-T3 (F(1, 37) = 7.44, p = 0.01), F4-O2 (F(1, 37) = 4.19, p = 0.048), C4-P4 (F(1, 37) = 6.75, p = 0.013), C4-O2 (F(1, 37) = 5.97, p = 0.019), P4-O2 (F(1, 37) = 5.37, p = 0.026), gamma F4-P4 (F(1, 37) = 4.50, p = 0.041), F4-O2 (F(1, 37) = 4.52, p = 0.04), C4-P4 (F(1, 37) = 5.47, p = 0.025), C4-O2 (F(1, 37) = 5.72, p = 0.022), C4-T4 (F(1, 37) = 5.85, p = 0.021), P4-O2 (F(1, 37) = 5.46, p = 0.025), and P4-T4 (F(1, 37) = 5.25, p = 0.028); for geometric categories at beta C4-P4 (F(1, 37) = 6.95, p = 0.012) and C4-O2 (F(1, 37) = 4.26, p = 0.046), and gamma P4-O2 (F(1, 37) = 4.32, p = 0.045); for pictorial sequence at theta P4-O2 (F(1, 38) = 5.08, p = 0.03); and for geometric sequence at high alpha C3-P3 (F(1, 37) = 5.44, p = 0.025) and beta C4-O2 (F(1, 37) = 4.65, p = 0.038). The interaction patterns between fitness and genotype are shown in (Figs. 2–3).

Figure 2. Inter- and Intra-hemisphere coherence electrode pairs showing significant interactions between physical fitness and genotype

Figure 3. Inter- and Intra-hemisphere coherence electrode pairs showing significant interactions between physical fitness and genotype


Interaction effect of physical fitness and genotype in inter-hemispheric coherence

In the analysis of inter-hemispheric coherence during CTONI task performance, significant fitness and genotype interaction effects emerged for geometric analogies at beta P3-P4 (F(1, 37) = 5.40, p = 0.026), for pictorial categories at high alpha O1-O2 (F(1, 37) = 4.26, p = 0.046), beta P3-P4 (F(1, 37) = 7.85, p = 0.008) and O1-O2 (F(1, 37) = 5.81, p = 0.021), gamma P3-P4 (F(1, 37) = 5.38, p = 0.026) and T3-T4 (F(1, 37) = 4.93, p = 0.033); for geometric categories at high alpha O1-O2 (F(1, 37) = 5.62, p = 0.023), beta P3-P4 (F(1, 37) = 8.44, p = 0.006), and gamma P3-P4 (F(1, 37) = 4.27, p = 0.046); and for pictorial sequence at high alpha O1-O2 (F(1, 38) = 4.56, p = 0.039), and beta P3-P4 (F(1, 38) = 6.33, p = 0.016), and for geometric sequence at beta P3-P4 (F(1, 37) = 8.55, p = 0.006). The interaction patterns between fitness and genotype are shown in (Figs. 2–3).



This study investigated cognitive functioning and the synchronization between brain regions in adolescent e4 carriers and non-carriers depending on physical fitness. The present study only found effects of the genotype and fitness in the analyses of EEG coherence during the CTONI-2 test, without any significant effects in performance outcomes.
The absence of behavioral differences in our study is consistent with the previous studies where only neuropsychological effects emerged without behavioral differences on cognitive tasks between adult e4 carriers and non-carriers in their early twenties (13) and between adolescent e4 carriers and non-carriers (8). Such findings are not surprising since a meta-analysis which investigated effects of ApoE e4 on cognitive ability of young carriers (age 2-40 years) found no significant differences to have the impact of the genotype on cognitive domains including intelligence, attention, memory, and executive functioning (19). Considering that the participants in our study were young adolescents in their teens, it is possible that the effects of ApoE e4 is not detrimental enough to be manifested by behavior. When there is structural and functional degradation, a healthy brain activates compensatory mechanisms to buffer against cognitive and behavioral decline (20), which makes it difficult to detect the changes until the impairment reaches a tipping point or threshold. In a study which examined functional connectivity during memory encoding in young adult e4 carriers and non-carriers, the e4+ group exhibited greater functional activation and connectivity of the medial temporal lobes, without a performance difference between groups, as evidence of the e4 influence on brain function long before the onset of the disease (21). In another study where event-related potentials were explored, the e4 carriers demonstrated delayed information processing and less effective attention allocation relative to the non-carriers without behavioral differences (13). Therefore, despite the absence of a performance difference, the EEG coherence itself can be interpreted as a marker for cognitive decline, with higher coherence reflecting lower neural efficiency or greater cognitive decline (22).
The coherence analysis on the interaction effect between genotype and fitness revealed a different coherence pattern between e4+ and e4− that was dependent on physical fitness level (Fig. 3). The high-fit e4+ individuals exhibited higher coherence than the low-fit e4+ individuals. In contrast, the opposite pattern was shown in e4− individuals, where the high-fit e4− group showed lower coherence than the low-fit e4− group. These results suggested that low-fit adolescents carrying the e4 allele have reduced cortical networking, which is similar to middle-aged (11) and college-aged e4 (13) carriers. Furthermore, the coherence of low-fit e4+ individuals was significantly lower than that of the low-fit e4− individuals. This suggests that e4 carriers who are physically less fit may be more susceptible to the negative effects associated with e4 retention, such as degradation of brain function and synapse transmission, even during adolescence (23). In contrast, the high-fit e4+ adolescents exhibited higher coherence than the low-fit e4+ adolescents and higher or not significantly different coherence when compared with the e4− adolescents. Higher coherence in the high-fit e4+ group may be a function of reduced neural efficiency; therefore, greater cortical communication is required to perform the same task. This increase in activation may reflect a compensatory mechanism in the brain to offset the cognitive declines due to the presence of e4 alleles. This interpretation is supported by Woo (13), which showed that high-fit young adult e4 carriers had higher attentional allocation during task performance than low-fit e4 carriers.
The coherence of the high-fit e4− group was similar to or lower than that of low-fit e4− group. Furthermore, the low-fit e4− group exhibited higher coherence than the low-fit e4+ group, which suggests that the cognitive function of low-fit e4− individuals may be less negatively affected compared with the low-fit e4+ individuals due to the absence of the genetic risk factor. In this case, fitness may improve the efficiency of brain functioning rather than compensating for functional degradation (as shown in e4+ carriers). Colcombe et al. (24) have reported that groups with high fitness or participating in aerobic exercise for 6 months exhibit enhanced neural efficiency with reduced activation in task-irrelevant brain regions. Similarly, one study examined coherence differences associated with physical fitness during non-verbal intelligence task. They found that the high-fit group demonstrate higher neural efficiency than the low-fit group; therefore, they engage less cortical networking to perform the same task (25). These data suggest that the way fitness affects cortical networking differs depending on the presence of the AD-related genotype. Nonetheless, fitness plays a positive role in performing cognitive tasks.
The interaction effects between genotype and fitness on both inter- and intra-hemispheric coherence were found in high frequency bands (high-alpha, beta, and gamma) at the parietal, occipital, and central regions (Fig. 2). These brain regions and frequency bands are commonly associated with functional degradation caused by the e4 genotype. Studies investigating the neural efficiency of adult e4 carriers in their twenties, adolescence, and patients with AD (8, 26) have found functional decline primarily in high frequency bands at the parietal, occipital, and central regions. These findings suggest that the brain regions affected by the e4 genotype might be the parietal and occipital lobes, which are responsible for spatio-temporal thinking and three-dimensional representation, and encoding and storing visual and spatial information, respectively (27). The results of the present study highlight that functional deterioration of these brain regions may also occur in adolescence with a genetic risk of AD.

One notable result of our inter- and intra- hemispheric coherence analyses was that the largest number of electrode pairs that showed interactions between fitness and genotype did so during the pictorial categories of the test. These categories are designed to evaluate relational reasoning by selecting pictures that are closest to the two associated pictures presented in the example. The pictorial categories of the intelligence test require the retrieval of information from long-term memory; therefore, patients with AD, or those at risk of developing AD, may demonstrate functional deterioration during these types of tests. This may explain why approximately 70% of the interaction effects between fitness and genotype in the intra-hemispheric coherence analysis emerged in pictorial categories in the present study.
In this study, the effects of the e4 allele retention and physical fitness only emerged in coherence results, not in the performance results. Since different cognitive tasks involve distinct brain networks according to task requirements, the difference in the pattern of cortical activation may induce different electrophysiological and behavioral outcomes (28). Therefore, the future study needs to be replicated using different types of cognitive tasks to support the interpretation of the current findings. This study, unfortunately, did not confirm if body weight or BMI caused meaningful confounding on the results. Considering that body weight or BMI is closely related to physical fitness (29), the next studies need to investigate how these variables have influences on the cognitive decline caused by the e4 allele.
In summary, the findings of the present study suggest that physical fitness may contribute to compensating for the possible e4 allele-related cognitive decline by increasing cortical communication between the brain regions and to benefiting adolescents not carrying the e4 allele in a way that enables task performance with reduced cortical networking owing to increased neural efficiency. Taken together, this study showed that the mechanisms by which physical fitness influences cognitive function may differ according to the presence or absence of a genetic risk factor for AD.

Conflict of interest: No conflict of interest

Funding: This study was supported by the National Research Foundation of Korea (NRF-2013R1A1A3010059).

Ethical Standards: This work was conducted in accordance with principles set forth by the Declaration of Helsinki. All volunteers gave written informed consent before participating.


1. Serrano-Pozo A, Frosch MP, Masliah E, Hyman BT. Neuropathological alterations in Alzheimer disease. Cold Spring Harb Perspect Med 2011;1(1):a006189.
2. Prince M, Bryce R, Albanese E, Wimo A, Ribeiro W, Ferri CP. The global prevalence of dementia: a systematic review and metaanalysis. Alzheimers Dement 2013;9(1):63-75.
3. Corder EH, Saunders AM, Strittmatter WJ, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 1993;261(5123):921-3.
4. Roses AD. The Alzheimer diseases. Curr Opin in Neurobiol 1996;6(5):644-50.
5. Reiman EM, Uecker A, Caselli RJ, et al. Hippocampal volumes in cognitively normal persons at genetic risk for Alzheimer’s disease. Ann Neurol 1998;44(2):288-91.
6. Scarmeas N, Habeck CG, Hilton J, et al. APOE related alterations in cerebral activation even at college age. J Neurol Neurosurg Psychiatry 2005;76(10):1440-4.
7. Shaw P, Lerch JP, Pruessner JC, et al. Cortical morphology in children and adolescents with different apolipoprotein E gene polymorphisms: an observational study. Lancet Neurol 2007;6(6):494-500.
8. Woo M, Kim Y. Cortical functional connections and fluid intelligence in adolescent APOE ε4 carriers. Dement Geriatr Cogn Disord 2017;44(3-4):153-9.
9. Heyn P, Abreu BC, Ottenbacher KJ. The effects of exercise training on elderly persons with cognitive impairment and dementia: A meta-analysis1. Arch Phys Med Rehabil 2004;85(10):1694-704.
10. Yu F, Thomas W, Nelson NW, Bronas UG, Dysken M, Wyman JF. Impact of 6-month aerobic exercise on Alzheimer’s symptoms. J Appl Gerontol 2015;34(4):484-500.
11. Deeny SP, Poeppel D, Zimmerman JB, et al. Exercise, APOE, and working memory: MEG and behavioral evidence for benefit of exercise in epsilon4 carriers. Biol Psychol 2008;78(2):179-87.
12. Schuit AJ, Feskens EJ, Launer LJ, Kromhout DA. Physical activity and cognitive decline, the role of the apolipoprotein e4 allele. Med Sci Sports Exerc 2001;33(5):772-7.
13. Woo M. Cardiovascular fitness modifies the relationship between genotype and neurocognitive function during executive challenge in late adolescence. 2008. Dissertation. University of Maryland.
14. Thatcher RW, North D, Biver C. EEG and intelligence: Relations between EEG coherence, EEG phase delay and power. Clin Neurophysiol 2005;116(9):2129–41.
15. Seoul National University Institute of Sport Science. Physical Activity Promotion System. 2009.
16. Ko BG, Kim YR, Sung BJ, et al. Development of criteria for Korea youth fitness award. Korean J Sport Sci 2005;16:44-63.
17. Jasper HH. The ten-twenty electrode system of the International Federation. Electroencephalogr Clin Neurophysiol 1958;10:370-5.
18. Deeny SP, Hillman CH, Janelle CM, Hatfield BD. Cortico-cortical communication and superior performance in skilled marksmen: An EEG coherence analysis. J Sport Exerc Psychol 2003;25(2):188–204.
19. Weissberger GH, Nation DA, Nguyen CP, Bondi MW, Han SD. Meta-analysis of cognitive ability differences by apolipoprotein e genotype in young humans. Neurosci Biobehav Rev. 2018;94:49–58.
20. Bookheimer SY, Strojwas MH, Cohen MS, et al. Patterns of brain activation in people at risk for Alzheimer’s disease. N Engl J Med 2000;343(7):450–6.
21. Dennis NA, Browndyke JN, Stokes J, Need A, Burke JR, Welsh-Bohmer KA, et al. Temporal lobe functional activity and connectivity in young adult APOE ε4 carriers. Alzheimer’s Dement. 2010;6(4):303–11.
22. Deeny SP, Haufler AJ, Saffer M, Hatfield BD. Electroencephalographic coherence during visuomotor performance: A comparison of cortico-cortical communication in experts and novices. J Mot Behav 2009;41(2):106–16.
23. Klein RC, Mace BE, Moore SD, Sullivan PM. Progressive loss of synaptic integrity in human apolipoprotein E4 targeted replacement mice and attenuation by apolipoprotein E2. Neuroscience 2010;171(4):1265–72.
24. Colcombe SJ, Kramer AF, Erickson KI, et al. Cardiovascular fitness, cortical plasticity, and aging. Proc Natl Acad Sci U S A 2004;101(9):3316–21.
25. Jung H, Woo M, Woo J. The Difference of non-verbal intelligence and cortical networking according to physical fitness: analysis of EEG coherence. Korean J of Sport Psychology 2015;26(4):97-110.
26. Koelewijn L, Lancaster TM, Linden D, et al. Oscillatory hyperactivity and hyperconnectivity in young APOE-ε4 carriers and hypoconnectivity in alzheimer’s disease. Elife 2019;8:1–25.
27. Kolb B, Whishaw IQ. Introduction to Brain and Behavior. 2013. Worth Publishers. N.Y.
28. Fitzgerald PB, Srithiran A, Benitez J, Daskalakis ZZ, Oxley TJ, Kulkarni J, et al. An fMRI study of prefrontal brain activation during multiple tasks in patients with major depressive disorder. Hum Brain Mapp. 2008;29(4):490–501.
29. Dumith SC, Ramires VV, Souza MA, Moraes DS, Petry FG, Oliveira ES, et al. Overweight/obesity and physical fitness among children and adolescents. J Phys Act Heal. 2010;7(5):641–8.


A.G. Juby1, D.R. Brocks2, D.A. Jay1, C.M.J. Davis1, D.R. Mager3

1. Division of Geriatrics, Department of Medicine, Edmonton, Alberta, Canada; 2. Faculty of Pharmacy and Pharmaceutical Sciences, Edmonton, Alberta, Canada;
3. Department of Agriculture Food and Nutrition Science; University of Alberta, Edmonton, Alberta, Canada

Corresponding Author: Angela G Juby, Professor of Medicine, Division of Geriatrics, Department of Medicine, 1-198 Clinical Sciences Building, 11350 83 Avenue, Edmonton, T6G 2P4, Canada, Tel: 1 780 492 6233, Fax: 1 780 492 2874, Email:

J Prev Alz Dis 2021;1(8):19-28
Published online October 9, 2020,



Objectives, Design, Setting: The ketogenic effect of medium chain triglyceride (MCT) oil offers potential for Alzheimer’s disease prevention and treatment. Limited literature suggests a linear B-hyroxybutyrate (BHB) response to increasing MCT doses. This pharmacokinetic study evaluates factors affecting BHB response in three subject groups.
Participants: Healthy subjects without cognitive deficits <65years, similarly healthy subjects >=65years, and those with Alzheimer’s Disease were assessed.
Intervention: Different doses (0g,14g, 28g, 42g) of MCT oil (99.3% C8:0) were administered, followed by fasting during the study period.
Measurements: BHB measured by finger prick sampling hourly for 5 hours after ingestion. Each subject attended four different days for each ascending dose. Data was also collected on body composition, BMI, waist/hip ratio, grip strength, gait speed, nutrient content of pre-study breakfast and side effects.
Results: Twenty-five participants: eight healthy; average age of 44yr (25-61), nine healthy; 79yr (65-90) and eight with AD; 78.6yr (57-86) respectively. Compiled data showed the expected linear dose response relationship. No group differences, with baseline corrected area under the blood vs. time curve (r2=0.98) and maximum concentrations (r2=0.97). However, there was notable individual variability in maximum BHB response (42g dose: 0.4 -2.1mM), and time to reach maximum BHB response both, within and between individuals. Variability was unrelated to age, sex, sarcopenic or AD status. Visceral fat, BMI, waist/hip ratio and pretest meal CHO and protein content all affected the BHB response (p<0.001).
Conclusion: There was a large inter-individual variability, with phenotype effects identified. This highlights challenges in interpreting clinical responses to MCT intake.

Key words: Alzheimer’s disease, B-hydroxybutyrate, medium chain triglyceride (MCT), pharmacokinetic, body composition, coconut oil.

Abbreviations: AD: Alzheimer’s Disease; AUC: area under the blood concentration time curve; BHB: Beta hydroxybutyrate; BMI: Body Mass Index; Cmax: maximum blood concentration; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders, 4th edition; MCT: Medium Chain Triglyceride; NINCDS-ADRDA: National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association; tmax: time of Cmax



Nutritionally induced ketosis (either by diet and/or MCT oil or esters) is being increasingly studied for the prevention and treatment of AD. The rationale is based on research that clearly identifies cerebral glucose hypometabolism in pre-symptomatic and symptomatic AD. This impaired metabolism and blood flow has been shown to be correctable when supplying the brain with ketones-through direct infusion (1) or via a ketogenic agent (2). This knowledge creates the opportunity for avenues of investigation for new approaches in the prevention and management of AD (3, 4). Extensive discussion of this topic is beyond the scope of this paper, but is covered in a recent review (5).
Ketones (acetone, acetoacetate and betahydroxybutyrate (BHB)) are produced endogenously under conditions of reduced glucose availability, primarily from Beta oxidation of fatty acids (6). Therapeutic levels of BHB (for treatment of medical conditions such as epilepsy and obesity) require a very low calorie, or very low carbohydrate (CHO), ketogenic diet to be maintained long term. A very low calorie ketogenic diet, requires strict medical supervision and is unsustainable in the long term because of compliance and biochemical side effects (7). A very low CHO, ketogenic, iso caloric diet is more sustainable [8]. Other methods of nutritional ketone generation such as MCT oil supplementation may serve as possible solutions, either alone or as an adjunct to ketogenic diets.
MCTs (commonly produced from coconut or palm oil), with a chain length of 6-10 carbons, are metabolized differently from other triglycerides (short and long chain), and result in ketone production, even in the face of adequate blood glucose. Gastric and pancreatic lipases hydrolyse MCT into medium chain fatty acids (MCFA) enabling rapid absorption from the gut, where the majority is transported in the portal vein to the liver, with rapid diffusion into hepatocytes (9, 10). MCFAs have a high propensity for oxidation, behaving more like glucose than fat in oxidative pathways (10). The resultant increased acetyl-Co A in the liver leads to ketogenesis and ketone release into the circulation and therefore immediate energy production, without adipose tissue deposition (11). Because ketones generated in the liver cannot be utilized by the liver for energy, all therefore flow from the liver to extra-hepatic tissues as fuel (11).
MCT in the form of oil or tablet esters, is increasingly being used to promote ketogenesis by members of the public. Its purported benefits are for weight loss, increased energy, and cognitive enhancement (11-15). MCT intake however can be associated with gastrointestinal side effects which can decrease tolerability and the sustainability of their ketogenic effect (16).
In previous studies evaluating ketogenesis for various clinical effects (13, 15), pooled BHB response has been measured with little comment on potential confounding factors such as age, BMI, or body composition.
This study seeks to further our understanding of factors affecting BHB response to MCT oil by evaluating the impact of: Age – by using three groups of subjects of differing age and health status; Dose – by giving each subject three incremental doses of MCT oil and measuring their individual serum BHB ketone response; Phenotype – by assessing the impact of variables such as muscle mass and function, and body composition on BHB response; Tolerance – by detailed monitoring for side effects at different doses for each individual; Disease state – by including healthy elderly and young, and a group with AD to provide comparative data between these subgroups.
This information will enable more meaningful design and interpretation of future clinical trials in all areas of possible therapeutic ketone use (eg. seizures, weight management), but especially for the prevention and management of increasing numbers of the population with AD for which there are currently only limited treatment options.




This is a single centre, open-label, dose response study in three groups of subjects: healthy subjects under 65 years, healthy subjects 65 years and older, and subjects over 50 years of age with confirmed Alzheimer’s disease. The study was carried out in Edmonton, Alberta, Canada.
Alzheimer’s subjects were included to assess their tolerance and BHB response to the MCT oil, to see if it differed from healthy subjects of a similar age. They had a diagnosis of probable dementia of the Alzheimer’s type (mild to moderate severity) based on the revised National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) (16) and Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) (17) criteria, of at least 12 months duration.
Exclusion criteria included consumption of a ketogenic diet, prior use of MCT oil or esters (within 6 months), or a diagnosis of diabetes mellitus.
Power calculations showed a minimum requirement of eight subjects in each group (alpha=0.05, beta=0.8) receiving all doses would be adequate to assess for BHB response and tolerance.

Study design and intervention

Each subject’s individual response to varying doses of MCT (Bulletproof Brain Octane®) oil was measured. Subjects attended for four different study days. Each day they arrived having eaten breakfast. Details of the participant’s pre-study meal were recorded, and the nutritional components (CHO, protein, fat, fiber) and calories calculated at the study completion, using Cronometer® software. In the case of AD subjects, breakfast details were confirmed by their caregivers. Subjects were encouraged to eat a similar pre-study breakfast on each study day. This was consumed 60-90 minutes prior to the study start.
They were all given a 150ml standardized fruit drink. On study Day 1 it contained 0ml (0g) MCT oil, Day 2 it was mechanically stirred with 15ml (=14g) MCT oil, Day 3 with 30ml (=28g) and Day 4 with 45ml (=42g). The oil and fruit juice was mixed with a battery operated milk frother immediately prior to consumption, to improve misability of the MCT in the juice. The consumption of the MCT oil drink was directly supervised by the investigators and was consumed in a single drink of 150ml fruit juice and oil in 1-3 swallows (0.5-1 minute). After their study drink the subjects were only allowed to consume water.
The study doses of 14g, 28g and 42g were chosen based on published studies that have found a minimum daily dose of 30g MCT is required to generate adequate brain ketone uptake to theoretically replace, in part, the deficit in brain glucose uptake and/or use in AD (12).
Neither subjects nor investigators were blinded to their dose as this was not practical with the incremental MCT dose design. All participants were occupied with sedentary activities (puzzles, board games, movies, reading, socializing) between their hourly testing for the study duration, on each study day.
Each subject was also evaluated for weight, body composition (Omron Full Body Sensor Body Composition and Monitor Scale, Omron Healthcare Co, Ltd, China), BMI, waist/hip circumference, grip strength (Almedic Dynamometer®, Japan) and walking speed using the 10m walking test (18). Although AD subjects were included, cognitive function was not a study outcome given the short study design, but the information in these subjects is applicable to longer AD studies where cognitive outcomes are more feasible (19).


At baseline and hourly for 5 hours (h) thereafter, subjects had measurements of serum BHB by finger-prick testing for blood glucose and BHB done using the standardized protocol for the FreeStyle Precision Neo Blood Glucose and Ketone Monitoring System (Manufactured by Abbott Diabetes Care, Canada) (20). The tests strips used were Precision Blood B-ketone and Precision Blood Glucose strips, and used in accordance with the manufacturer’s instructions. The lower limit of quantification for BHB using this test is 0.1mM. Values were reported numerically. Values below 0.1mM were registered as 0.0mM on the device. These BHB ketostrips have been validated in previous studies for specificity, sensitivity, accuracy and precision (21, 22), with good accuracy (r=0.97) compared to whole blood BHB levels, and precision (%CV) no more than 3.1-3.8%. Random participants had a repeat BHB analysis from different finger prick sites at the same time for test re-test validity. The glucometers were all checked for accuracy daily, using the manufacturer’s standardized solutions, and met the standardization criteria.
As the test oil (Bulletproof Brain Octane®) is an over-the-counter product and not a pharmaceutical product, it was felt to be important for the validity of this study for the reported triglyceride content of the MCT test oil to be independently verified in an unrelated research laboratory. All the study oil came from the same lot number (1901279039, expiration 01/22) as that tested in the independent laboratory. The manufacturer stated 15ml contains 14g of MCT, but does not specify the oil specific gravity.
Side effects were recorded as per standard Health Canada protocols and reporting. They were classified as mild, moderate or severe based on pre-specified criteria (23). Side effects were actively solicited at each hour of testing, as well as being self reported. Any symptoms were classified and recorded.

Ethics and Registration

This study was approved by the Health Products and Food Branch, of Health Canada, (HC6-24-c186660) and the local University of Alberta Health Research Ethics Board (Pro 000087958). Procedures followed were also in accordance with the Helsinki Declaration of 1975 as revised in 1983. All subjects (or their caregivers) had read an Information sheet (mailed prior to the first study visit). At the start of the study all signed a Consent Form if they were cognitively able to understand the study process. For those with cognitive impairment, an Assent Form was signed by the participant and their legal representative. All AD subjects in this study were accompanied by their caregivers throughout the study duration. These documents were reviewed and approved by the local Research Ethics Board as well as Health Canada. Clinical trial registration: Identifier NCT04389983.

Statistical Analysis

Pharmacokinetic analysis

The BHB blood concentration versus time data after administration of MCT oil were analyzed using noncompartmental pharmacokinetic methods. The maximum blood concentrations (Cmax) and the time to attainment of Cmax (tmax) from each subject were obtained from inspection of the raw data. The area under the blood concentration vs. time curve for each subject, from the time of dosing to the last measured concentration, were calculated using the linear trapezoidal rule. The baseline-corrected AUC (BC AUC) after MCT oil were determined by subtracting the AUC0-5h after the 0 g MCT dose (representing the endogenous BHB concentrations) from the AUC0-5h after the MCT oil. The baseline-corrected Cmax (BC Cmax) was also determined by subtracting from each sampling time measurement after MCT oil, the corresponding measure of BHB in the same patient after no MCT oil was administered. The time at which the BC Cmax occurred (BC tmax) was also denoted.


Power calculations showed a requirement for 8-12 subjects per group to allow adequate statistical precision to notice a difference of 0.1mM in BHB for a p value of 0.05 for pharmacokinetic data.
Continuous variables were described with mean values and standard deviations. For descriptive statistics, paired t tests were used to analyze differences in BMI, grip strength, body composition, waist/hip ratio, gait speed and dietary macronutients between subjects, and between groups. Relationships between MCT dose and measures of systemic exposure (Cmax and AUC) of the BHB were assessed by linear regression. To assess intra-subject increases in exposure between each dose, mean differences in Cmax and AUC for each subject were determined and the 95% confidence intervals calculated and assessed. Statistical tests were two-tailed with significance set at p<0.05.



Twenty-five subjects were enrolled, and all completed all 4 days and all 4 doses of the study drink. Eight (young group), nine (healthy senior group) and eight (Alzheimer’s dementia) subjects were in the three groups, with an average age of 44 years (25-61), 79 years (65-90) and 78.6 years (57-86) respectively. By chance, all the AD subjects whose caregivers wanted them to participate were male. See Table 1 for details. Subjects attended consecutive days for each increased dose. Research days were Monday, Thursday and Friday, so based on the subject’s availability the maximum time between study visits was 7 days, with an average of 2 days, and a minimum of 1 day.

Table 1. Study subject demographics

Table 1. Study subject demographics

Data is expressed as median (range). Variables with different superscripts are statistically significant at p<0.05.


Detailed nutritional analysis of reported pre-study breakfast each day confirmed that subjects were consistent with their breakfast nutrient intake as requested. Daily breakfast calorie intake ranged from 166-1545 kcal (average 485.4kcal), 4.5-138 g CHO (average 49.4g), 14-56.7g protein (average 24.4g) and 2.0-46.1g fat (average 19.1g) There were no significant differences in energy and fat intake noted in each subject for their daily breakfast over the four days of the study. See Table 2 for details. There was a statistically significant difference between the under and over 65 year healthy groups, and the healthy groups compared to the AD group, for breakfast fat and fiber intake. CHO intake only differed between the two healthy groups. The standardized study fruit drink contained 17.4g of CHO.

Table 2. Nutrient content of pre-study breakfast meal. ≥65; healthy groups versus AD

*denotes statistically significant difference between groups: healthy <65 versus healthy ≥65; or all healthy versus AD, at p<0.05


The MCT test oil (Bulletproof Brain Octane®) was verified as 99.3% C8, 0.6% C10 and 0.1% C12:0 by an independent lipid research laboratory (see acknowledgements).
Baseline BHB levels were in the range expected for non-fasting subjects on a non-ketogenic diet (0.0-0.3 mM). Individual responses in BHB level varied at each dose, and the pattern of individual response also varied between and within individuals for each dose. See Figure 1. Only three subjects were on statins (low dose) : one healthy elderly; and two AD subjects. They did not appear to have any blunting of response with all three reaching peak BHB level of 0.8-1.1mM with the highest dose, which was within the average for the entire group.

Figure 1. Mean+CI (confidence interval) in absolute BHB (mM) with each MCT dose


Endogenously produced BHB was detected in all patients and a full AUC was calculable in the subjects. The median tmax of BHB without MCT oil was 3h after the sampling was started in the patients. As the dose of MCT oil was introduced and increased, there were associated increases in the mean measures of AUC and Cmax. (Table 3). Considerable inter-subject variability was apparent in the measures of overall exposure in the absence of MCT oil; this variability generally dropped as the dose of MCT oil increased.


Table 3. Summary of pharmacokinetic exposure data (mean+SD). Observed and baseline-corrected (BC) values are shown

Variables are mean +/- standard deviation; (AUC: area under the blood concentration time curve; h: hour; Cmax: maximum blood concentration; tmax: time of Cmax; BC: baseline corrected)

To ascertain how much the dose of MCT oil contributed to the BHB levels, baseline correction of the AUC and Cmax were undertaken by subtracting the baseline values from concentrations (to obtain Cmax) or baseline AUC from AUC after MCT oil. When this was done increases in exposure with increases in MCT oil doses were still apparent, again with notable inter-subject variability. Despite the inter-individual variability, compiled pharmocokinetic evaluation of pooled data produced overall a highly linear response between the mean measures of Cmax and AUC0-5h (raw and baseline-corrected) and the administered dose (See Figure 2).

Figure 2. Mean+SD exposure data plotted versus the dose of MCT oil. (AUC Area Under Curve, Cmax: maximum B-Hydroxybutyrate response)


To better understand the nature of the increase exposure with increases in the dose of MCT oil, the mean intra-subject increase in concentrations between discreet doses was calculated along with the 95% confidence intervals of those differences (Table 4). For every dose increment it was apparent that there were significant increases in exposure in the subjects. It was also apparent that between the first to third doses the increases in exposure were nearly the same, thus indicating overall dose proportionality within individuals.

Table 4. Mean intra-individual differences between doses in exposure error between doses with 95% confidence intervals shown in parentheses

All differences between pairs of the rising doses were significantly different at p<0.05; (AUC: area under the blood concentration time curve; h hour; Cmax: maximum blood


When assessing factors that may be responsible for this individual variation, the BHB level was found to be unrelated to: AD status; age; sex; and body weight. Percentage muscle, and measures of muscle function (grip strength and gait speed), ie. parameters reflecting sarcopenic status, did not appear to affect BHB responses. However, for BMI and body composition, increasing BMI, increasing percentage visceral fat, and increasing waist/hip circumference were all associated with decreased BHB serum levels 2-3h post MCT ingestion (p<0.001), and this was also seen in the AUC responses for all the doses. Apo E4 status was not evaluated as this data was only available for eight (32%) of participants, and only two of these had at least one ApoE4 allele. Therefore there was not enough power in this study to assess for any effect. Table 5. Adverse events (AE) (Some subjects reported several side effects). (g: gram, n: number of subjects) MCT Dose (g) 0 14 28 42 Total AE (n) 0 3 13 13 Abdo cramp/indigestion 0 1 5 7 Nausea 0 0 6 3 Burping/bloating 0 0 1 1 Diarrhoea 0 0 1 1 Fatigue 0 1 0 2 Dizziness 0 0 1 2 Headache 0 1 3 3 Brain fog 0 0 2 1 Sweating/clamminess 0 0 2 1 Perhaps not unexpectedly, increasing CHO and protein in the pre-study breakfast meal was associated with lower levels of serum BHB at 2-5h (p=0.002) at all dosing levels. These effects were also apparent in the AUC response, at different MCT doses, at all time points (p=0.008). Although the study drink contained CHO, this was identical for all participants and did not affect the baseline BHB as that was measured immediately before the study drink was consumed. It may have influenced the maximum BHB level obtained but this impact would have been consistent across the study days and the study participants. There was no apparent effect of total kcal, or fiber. With respect to gait speed, those generating a higher BHB response had a faster walking speed in the “fast walk” component of the gait evaluation, consistent across all doses (p=0.05). Increasing grip strength was related to lower BHB tmax (p=0.03). There were no serious adverse events. Four participants (16%) experienced “influenza-like” symptoms (headache, diaphoresis, fatigue) with either 28g or 42g MCT oil, or both, lasting approximately 2h. Blood glucose measurements done at the time of these symptoms were in the normal range. None of these symptoms were reported by the subjects to be severe enough for study discontinuation, and those experiencing adverse events at 28g had no concerns about returning for the 42g test dose study day. Eighteen (72%) experienced some minor gastrointestinal discomfort, not necessarily related to the measured blood BHB level. 28% of subjects experienced no side effects. In all cases, all the symptoms resolved completely within minutes of eating the post study meal. See Table 5. There was no change in percentage body fat, percentage muscle and percentage visceral fat from Day 1 to Day 4.

Table 5. Adverse events (AE) (Some subjects reported several side effects). (g: gram, n: number of subjects)



MCT oil led to therapeutic BHB levels in many of our subjects. In healthy individuals BHB is generally less than 0.5mM. Under physiological low glucose conditions, ketones rarely reached over 3mM, but can reach 5-7mM after prolonged (one week) fasting (24).Levels of 4-6mM stimulate increases in insulin secretion resulting in a reduced production and increased urinary excretion of ketones. Only in dysregulatory conditions such as diabetic ketoacidosis can levels reach 20mM (25). Nutritional ketosis with a ketogenic diet alone can be difficult to maintain, especially in adults with cognitive impairment. Even in a group of highly motivated parents of children with intractable seizures, 33% discontinued their ketogenic diet (4:1 fat:CHO and protein) within 12 months because of side effects or lack of seizure benefit (26). However, they did reach BHB levels up to 5mM in some subjects. In addition, ketogenic diets vary in their nutritional content. In one study in adults, of a ketogenic diet (1.8:1 fat: CHO and protein) alone (without additional MCT oil) their highest level of BHB was 0.7 +/- 0.62mM (27). To address this, researchers have combined ketogenic diets with very low calories (7) or low carbohydrate intake, or have added MCT oil to a ketogenic diet (12, 27-30). With these interventions, BHB levels ranged from >0.7mM to >2.0mM. In this current study, with the highest MCT dose (42g), the highest individual level was 2.1mM.
Assuming non-saturable absorption mechanisms, a linear relationship would be expected between dose and blood measures. Although there was the expected linear dose response in our compiled data (shown in Tables 3,4 and Figure 2), the marked individual variability in pattern of response, degree of response and side effects was somewhat unexpected. Not only did individuals have a unique pattern of BHB response to the MCT dose, this pattern was not necessarily the same with subsequent doses. Some patients showed evidence of multiple peaks in their BHB blood concentration versus time profiles. This may have been due to factors such as variability in absorption secondary to gastrointestinal mixing, or presence of food components, or the delay attributed to the lymphatic pathway of absorption (31).

Covariate analysis

A covariate analysis was performed to possibly predict a “good” responder from an “average” or “non” responder. Looking at the impact of the pre-study breakfast to account, there seemed to be an effect of the baseline CHO and protein content of the meal, with higher levels reducing the BHB response. The significant effect of visceral fat, BMI and waist/hip circumference in reducing BHB response, suggests that insulin sensitivity may play a role in the response of an individual to MCT-induced nutritional ketosis. Fasting insulin was not measured, but visceral adiposity is associated with an increased risk of metabolic syndrome, one component of which is insulin resistance (32).
There was no impact of age or cognitive health observed here. AD subjects were included because of our ongoing interest in the role of MCT oil as a source of nutritional ketones for influencing cognitive function. AD is associated with cerebral insulin resistance and perhaps also peripheral insulin resistance (33). Ongoing research is looking at MCT induced nutritional ketosis as a possible cognitive enhancer in AD subjects (6, 14, 15), so it is an important group in whom to assess their tolerance and response to the ketogenic effects of MCT. There was no statistical signal suggesting the AD group responded any differently to the healthy groups with respect to BHB response or tolerance. Cognitive change after the MCT administration was not assessed, as that was not an objective in this study. Other authors have suggested cognitive benefits with MCT oil or with ketone monoesters (9, 10). Discussion of this data is beyond the scope of this pharmacokinetic paper, but covered in several recent reviews (3, 34).


We are aware of only two other MCT specific pharmacokinetic studies in publication, and both evaluate ketone esters rather than MCT oil as a source of nutritional ketosis. One study involving a synthetic ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate was performed (35) in 54 healthy subjects between 18-45 years. In 36 subjects the kinetics of the monoester were evaluated after 0.42, 1.07 and 2.14g/kg body weight, administered for 5 consecutive days. Additionally, 18 subjects were evaluated in a single dose (0.14, 0.357, or 0.714g/kg) pharmacokinetic analysis, where each subject only had one test dose. Their subjects, unlike in this study, were fasting at baseline. For their three doses, the Cmax of BHB was 0.28, 1.00 and 3.30mM respectively. The tmax ranged from 1.5 to 2.5h. They did not comment on individual BHB responses with any of the doses, only the pooled response, nor was there mention of baseline-correction to account for endogenous BHB production. The current study found BHB to vary considerably between individuals. They assessed BMI (but not body composition or waist/hip ratio) but did not comment on its relationship to BHB. The use of a synthetic ketone may have produced different pharmacokinetics than MCT oil.
In the second study, Shivva and colleagues [36] used the same ketone monoester in their subsequent single dose study with five different doses in 37 young healthy volunteers, followed for 5-7h post dose. Only total percentage fat mass was reported, and the same method of BHB measurement was used as in the current study. Their objective was to develop a pharmacokinetic prediction model, including accounting for endogenous ketone production and the other nutrients in their study drink. They found that lean body weight and sex affected the BHB response, contrary to the present study. In their conclusion they stated the “pharmacokinetics of BHB is complicated” and encouraged more research.
Although these agents differ from the ketogenic agent used in this study, it does provide some comparable data on Cmax and some information on factors affecting the dose response. In addition, investigators in this field are evaluating ketone esters (15) as possible therapies for AD given that their tablet/powdered form is more convenient than liquid MCT.

Ketogenic response

In a feasibility study for a ketogenic diet in AD, Taylor et al (14) administered their participant’s ketones as 1.5-3 tablespoons of MCT oil (17.5-42g) daily with food, as well as consuming a ketogenic diet. Only 60-80% of subjects were able to consume their targeted MCT dose because of gastrointestinal side effects, and there was an increase in average monthly BHB from baseline. Although AUC was unreported, there was significant individual variation in BHB levels (0-1.6 mM) reached at the maximum time point, which occurred in the first month of their study. With the significant individual variation in MCT dose taken, this variable response is hard to interpret. Their study was daily MCT over three months making compliance more challenging. This differs from the present study design where compliance was 100% because it was only a single dose on each day, meaning subjects were able to consume exactly the same dose as one another, making between group and individual comparisons possible.
Freemantle et al (37) examined three age groups of healthy subjects, provided a low CHO (3g) ketogenic (110g fat) diet with MCT oil. They measured serum BHB hourly, and breath acetone and plasma insulin over 6h. BHB levels rose from 0.1mM to 1.3mM with peak BHB response at 2-4h post dose. Plasma insulin peaked at 1-2h with no reported difference between the groups. The insulin level was not correlated with individual BHB response, but there were higher plasma glucose (but not BHB or insulin) and lipids in the elderly group compared to younger subjects.
Vandenberghe and colleagues (38) compared different combinations of MCT oils (coconut, tricaprylin (C8), and tricaprin (C10)) to assess the ketogenic potential. Their tricaprylin product (95%C8) most closely resembles the MCT oil used in the current study. They looked at total ketones (BHB plus acetacetate) making the actual levels achieved difficult to compare. However, they did find their C8 oil significantly increased plasma ketones by 2.88 +- 1.9 mM above baseline when given with a meal, and increased it further when an additional dose was given on an empty stomach. It was the most ketogenic of the oils in their study. Unfortunately, they did not discuss side effects in their test subjects.
Different MCT formulations, including an emulsified formulation have been investigated in young healthy (mean 31 years) to examine their influence on BHB response (39). Their MCT was a mix of 60% C8:0 and 40% C10:0, rather than the 99.3% C8:0 oil in this study. Their subjects were fasting at baseline, and then consumed a standardized breakfast with their test drink. They were followed for 4h post dose, as opposed to the 5h follow-up in this study. Their data showed emulsification of the MCT drink increased BHB response, with less diarrhea and a stronger correlation with C8:0, and not C10:0 serum levels, and ketogenic response. They did not correct for any body composition differences between subjects.
Adding exercise to MCT is another strategy to increase the ketogenic response [40]. In a five day study the authors evaluated 15g twice daily of their MCT oil (55%C8 35%C10) and 30mins daily of aerobic exercise. This combination resulted in 69% higher plasma ketone levels (and AUC nearly doubled) in their normoglycemic subjects than either intervention alone. The absolute level achieved cannot be compared to this study as theirs was for total ketones. Unfortunately there was no report on adverse events.

Side effects

Side effects (nausea, vomiting, bloating, abdominal cramps, diarrhea) with MCT oil ingestion are well documented (10), and limit the amount of MCT oil that can be ingested at any one time, usually to 25-30g (10). Somewhat surprising in this study there was no clear relationship between gastrointestinal side effects and dose. For subjects reporting abdominal pain, headache, and diaphoresis at the 28g dose, they did not necessarily experience them (or any side effects) with the higher dose, even in the face of a higher blood BHB. Furthermore, some subjects with higher blood BHB than those experiencing symptoms, were asymptomatic. There appeared to be a lack of correlation between the level of BHB and report of side effects, both within and between individuals. In this study, the overall tolerance with the 42g dose was greater than anticipated from the literature (10). The gastrointestinal side effects were expected. The brain fog, dizziness, headaches and diaphoresis reported by some was not expected. Blood glucose assessment done at the time of the symptoms showed that these were not related to low or high blood glucose. Ketones have been reported to increase cerebral blood flow equally in healthy and AD subjects (1, 2) but unfortunately neither studied reported data on adverse events in their subjects. Apo E4 negative subjects had higher cerebral blood flow. In this study there was not Apo E4 data on all subjects, but the cognitive side effects were reported by the non-AD subjects, except for one AD subject who interestingly, was know to be Apo E4 negative. Keto-adapatation (12, 30) is unlikely to be the cause given the single dose and short duration design. Tolerance has been shown to improve with gradual dose titration, so most authors (us included) design longer studies to include a titration period (19, 39).
Of most interest for investigators is that clinical ketosis symptoms/side effects did not seem to be a good guide of the subject’s current BHB level.

New data

This study differs from these previously published studies in that it collected an AUC of exposure that included an AUC after a 0g dose, which allowed accurate correction for baseline AUC, as opposed to just relying on a zero time point, resulting in more robust AUC estimates. In addition, other studies have not corrected for BMI, or visceral adiposity which this study suggests are important factors affecting the BHB response to MCT oil supplementation. There may therefore need to be consideration of higher doses of MCT in studies involving subjects with elevated BMI, visceral adiposity, or increased waist/hip ratio, in order to elicit the desired BHB response. The individual response variability identified highlights the difficulty in interpreting pooled data, and subsequently extrapolating it to individuals with the assumption of achieving the same response. The apparent lack of association between absolute BHB level and side effects, and varied individual tolerance shown differs from other published studies.


Although the sample size appears small, a priori testing illustrated that a minimum of eight subjects per group was sufficient to detect differences in the primary outcomes.
All subjects were Caucasian Canadians, possibly limiting generalizable to other demographics. All the AD subjects were male, but as no overall sex differences in response were identified, that may not negate the applicability of the data to female AD subjects. Subjects were un-blinded to their MCT dose, although this is unlikely to have influenced blood concentration measures. The only disease state included was AD (without DM), preventing safety assessment in DM. Although study duration was only 5h the majority of subjects had a peak in BHB response around 3h, suggesting that a longer duration was not critical. Extending the period could have been associated with production of fasting-related endogenous ketones (a response which we did start to see even at hour 5 in some subjects). Research suggesting 8 hour follow-up was based on a two 10g dose protocol (4). Our study design prevented comment on longer duration MCT intake with respect to tolerance, and modified BHB response. We did not measure acetoacetate levels. We acknowledge that the ratio of acetoacetate and BHB can vary with different MCT products and is affected by additional oils such as coconut (38). But, in the case of >95% pure C8 oil (as in this study) these authors showed that the acetoacetate/BHB ratio was similar to their control oil. Urbain and colleagues showed the course of blood and urine ketones to be very similar (27) so we feel confident that the serum BHB level alone is a reflection of total body ketone production, and pattern of ketone response, but acknowledge that the actual BHB level may under-reflect the total body ketone level. We did not measure fasting insulin or HOMA IR in this study, but in another study with 20 AD subjects on daily MCT followed for 15months, we saw no effect of these parameters (19). The goal of the study was not to assess adverse events specifically based on dose, but adverse events in general. We do acknowledge that the serial dosing may have masked some adverse events, but the serial dosing was necessary as previous studies have suggested that single doses over 15g may be difficult to tolerate. Ethically, we did not want to expose subjects to the 42g dose if they had significant side effects at the 28g dose. In addition, we know from our previous experience, that symptoms resolve immediately with CHO consumption. This has also been shown in studies evaluating timing of MCT around meals (4).


All subjects completed all doses of the study. There was a large age range (from 25-90 years) and AD subjects were included. Parameters of body composition were evaluated and these were ultimately found to be of paramount importance in the BHB response. As discussed, other kinetics studies have not included or analyzed these variables. In addition, there was correction for baseline BHB production ensuring that the measured BHB response was due to the consumed MCT oil.



This dose response study with supplementary MCT oil looks at individual and group responses. It shows the expected linear response in ketone production (as measured by BHB levels) in the pooled data, which is not impacted by age, sex, muscle or total fat mass, grip strength, or AD status. However, it shows there is a marked individual response in the BHB level achieved with varying MCT doses, which is influenced by BMI and visceral adiposity. Individual tolerance to elevated BHB levels is variable and, unexpectedly, was not dependent on the absolute BHB level. Future studies, assessing response in clinical outcomes to defined MCT doses (for example, AD and cognition) will need to take these important variables into consideration in order to accurately design the studies and interpret the results.


Acknowledgements: The participants who gave up their time to be in the study. Vickie Baker (Registered Nurse) who assisted with data collection. Dr Vera Mazurak‘s lipid research laboratory for independently verifying the triglyceride content of the test oil.

Contributions of Authors: AGJ: designed the research, conducted the research, analyzed data, wrote the paper and had primary responsibility for the final content. DRB was involved in study design, and performed the pharmacokinetic statistical analysis. DAJ was involved in conducting the research, and data analysis. CMJD was involved in conducting the research, and data analysis. DRM performed the non-pharmacokinetic statistical analysis. All authors read and approved the final manuscript.

Conflict of Interest: Angela G Juby, Dion R Brocks, David A Jay, Christopher MJ Davis, Diana R Mager, all have no conflicts of interest with respect to this study.

Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.



1. Hasselbalch SG, Madsen PL, Hageman LP, Olsen KS, Justesen N, Holm S, Paulson OB. Changes in cerebral blood flow and carbohydrate metabolism during acute hyeprketonemia. Am Physiol Soc 1996;270:E746-51
2. Torosyan N, Sethanandha C, Grill JD, Dilley ML, Lee J, Cummings JL, Ossinalde C, Silverman DH. Chnges in regional cerebral blood flow associated with a 45 day course of the ketogenic agent, caprylidene, in patients with mild to moderate Alzheimer’s disease: results of a randomized, double blind, pilot study. Exp Gerontol 2018;111:118-21
3. Cunnane SC, Courchesne-Loyer A, Vandenberge C, St-Pierre V, Fortier M, Hennebelle M, et al. Can ketones help rescue brain fuel supply in later life? Implications for cognitive health during aging and the treatment of Alzheimer’s disease. Front Mol Neurosci 2016;9:53
4. Vandenberghe C, St-Pierre V, Fortier M, Castellano CA, Cuenoud B, Cunnane SC. Medium chain triglycerides modulate the ketogenic effect of a metabolic switch. Front Nutr 2020;7:3
5. Chatterjee P, Fernando M, Fernando B, Dias CB, Shah T, Silva R, et al. Potential of coconut oil and medium chain triglycerides in the prevention and treatment of Alzheimer’s disease. Mech Ageing Dev 2020; 186:111209
6. McPherson PAC, McEneny J. The biochemistry of ketogenesis and its role in weight management, neurological disease and oxidative stress. J Physiol Biochem 2012 68: 141-151
7. Caprio M, Infante M, Moriconi E, Armani A, Fabbri A, Mantovani G, Mariani S, Lubrano C, Poggiogalle E, Migliaccio S, et al. Very-low-calorie ketogenic diet (VLCKD) in the management of metabolic diseases: systematic review and consensus statement from the Italian Society of Endocrinology (SIE). J Endocrinol Investig 2019 42:1365-1386
8. Manninen AH. Metabolic effects of the very-low-carbohydrate diets: misunderstood “villains” of human metabolism. J Int Soc Sports Nutr 2004;1:7-11
9. Ramirez M, Amate L, Gil A. Absorption and distribution of dietary fatty acids from different sources. Early Hum Develop 2001;Suppl 65:S95-S101
10. Marten B, Pfeuffer M, Schrezenmeir J. Review: Medium-chain triglycerides. Int Dairy J 2006; 16:1374-82
11. Bach AC. Ingenbleek Y, Frey A. The usefulness of dietary medium chain triglycerides in body weight control: fact or fancy? J Lip Res 1996 37(4):708-726
12. Corchesne-Loyer A, Fortier M, Tremblay-Mercier J, Chouinard-Watkins R, Roy M, Nugent S, et al. Stimulation of mild, sustained ketonemia by medium chain triacylglycerols in healthy humans: estimated potential contribution to brain energy metabolism. Nutrition 2013;29:635-40
13. Hall KD, Chen KY, Guo J, Lam YY, Leibel RL, Mayer LES, Reitman ML, et al. Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men. Am J Clin Nutr 2016;104:324-33
14. Taylor MK, Sullivan DK, Mahnken JD, Burns JM, Swerdlow RH. Feasability and efficacy data form a ketogenic diet intervention in Alzheimer’s disease. Alzh Dement 4:28-36, 2018
15. Henderson ST, Vogel JL, Barr LJ, Garvin F, Jones JJ, Constantini LC. Study of the ketogenic agent AC1202 in mild to moderate Alzheimer’s Disease: A randomized, double-blind, Placebo controlled trial. Nutr Metab 2009:6:31
16. Dubois B, Feldman HH, Jacova C, DeKosky ST, Barberger-Gateau P, Cummings J, Delacourte A, et al. Research criteria for the diagnosis of Alzheimer’s disease: revising the NINCDS-ADRDA criteria. Lancet Neurol 2007;6:734-46
17. American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision. Washington, DC, American Psychiatric Press, 2000
18. Bohannon RW. Comfortable and maximum walking speed of adults aged 20-79 years: reference values and determinants. Age Ageing 1997;26:15-9
19. Juby A, Mager D, Davis C, Jay D, Blackburn T. Use of medium chain triglyceride (MCT) oil in subjects with Alzheimer’s disease: a randomized double-blind placebo cross over study, with an open label extension. Can Geriatr J 2020;23:1;3 (CCD Oral Presentations)
20. Abbott Laboratories Co 2020
21. Guerci B, Benichou M, Floriot M, Bohme P, Fougnot S, Franck P, Drouin P. Accuracy of an electrochemical sensor for measuring capillary blood ketones by fingerstick samples during metabolic deterioration after continuous subcutaneous insulin infusion interruption in type 1 diabetic patients. Diabetic Care 2003;26(4):1137-41
22. Forrow NJ, Sanghera GS, Walters SJ, Watkin JL. Development of a commercial amperometric biosensor electrode for the ketone D-3-hydroxybutyrate. Biosens Bioelectron 2005; 20(8):1617-25
23. Government of Canada, Clinical trials for Natural health products, Appendix 8:Adverse Reaction report form for Clinical trials.
24. Owen OE, Morgan AP, Kemp HG, Sullivan JM, Herrera MG, Cahill Jr GF. Brain metabolism during fasting. J Clin Invest 1967;46:1589-95
25. Cahill Jr GF, Veech RL. Ketoacids? Good medicine? Trans Am Clin Climatol Assoc 2003;114:149-61
26. Buchhalter JR, D’Alfonso S, Connolly M, Fung E, Michoulas A, et al. The relationship between D-beta-hydroxybutyrate blood concentrations and seizure control in children treated with the ketogenic diet for medically intractable epilepsy. Epilepsia Open 2017; 2:317-321
27. Urbain P, Bertz H. Monitoring for compliance with a ketogenic diet: what is the best time of day to test for urinary ketosis? Nutr Metab 2016 13:77-8
28. Volek JS, LaFountain RA, Dituro P. Extended ketogenic diet and physical training in military personnel. Military Med 2019; 184:199-200
29. Hallberg SJ, Gershuni VM, Hazbun TL, Athinarayanan SJ. Reversing Type 2 diabetes: a narrative review of the evidence. Nutrients 2019;11:766
30. Phinney SD, Bistrian BR, Evans JW, Gervino E, Blackburn GL. The human metabolic response to chronic ketosis without caloric restriction: preservation of submaximal exercise capability with reduced carbohydrate oxidation. Metabol 1983;32:769-776
31. Brocks DR, Davies NM. Lymphatic drug absorption via the enterocytes: pharmacokinetic simulation, modeling, and considerations for optimal drug development. J Pharm Pharm Sci 2018;21:254s-270s
32. Griffith ML, Younk LM, Davis SN. Visceral adiposity, insulin resistance, and Type 2 Diabetes. Am J Lifestyle Med 2010; 4:230-243
33. Willette AA, Bendlin BB, Starks BS, Birdsill AC, Johnson SC, Christian BT, Okonkwo OC, La Rue A, Hermann BP, Koscik RL, et al. Association of insulin resistance with cerebral glucose uptake in late-middle aged adults at risk for Alzheimer’s disease. JAMA Neurol 2015;72(9):1013-1020
34. McDonald TJW, Cervenka MC. The expanding role of ketogenic diets in adult neurological disorders. Brain Sci 2018;8:148
35. Clarke K, Tchabanenko K, Pawlosky R, Carter E, Todd King M, Musa-Veloso K, Ho M, Roberts A, Robertson J, VanItallie TB, Veech RL. Kinetics , safety and tolerability of (R)-3-hydroxybutyl (R)-3-hydroxybutyrate in healthy adult subjects. Regul Toxicol Pharmacol 2012; 63: 401-408
36. Shivva V, Cox PJ, Clarle K, Veech RL, Tucker IG, Duffull SB. The population pharmacokinetics of D-B-hydroxybutyrate following administration of (R)-3-3hydroxybutyrate (R)-3-hydroxybutyrate. The AAPS J 2016 18(3):678-688
37. Freemantle E, Vandal M, Tremblay-Mercier J, Plourde M, Poirier J, Cunnane SC. Metabolic response to a ketogenic breakfast in the healthy elderly. J Nutr, Health, Aging. 2009 13:293-298
38. Vandenberghe C, St-Pierre V. Perotti T, Fortier M, Castellano CA, Cunnane SC. Tricaprylin alone increases plasma ketone response more than coconut oil or other medium-chain triglycerides: an acute crossover study in healthy adults. Curr Dev Nutr 2017;1:1-5
39. Courchesne-Loyer A, Lowry CM, St-Pierre V, Vandenberghe C, Fortier M, Castellano CA, et al. Emulsification increases the acute ketogenic effect and bioavailability of medium-chain triglycerides in humans. Curr Dev Nutr 2017;1:e000851
40. Vandenberghe C, Castellano CA, Maltais M, Fortier M, St-Pierre V, Dionne J, Cunnane SC. A short term intervention combining aerobic exercise with medium chain triglycerides (MCT) is more ketogenic than either MCT or aerobic exercise alone: a comparison of nomoglycemic and prediabetic older women. Appl Physiol Nutr Metab 2019;44:66-73



S. Walter1, O.G. Langford1, T.B. Clanton1, G.A. Jimenez-Maggiora1, R. Raman1, M.S. Rafii1, E.J. Shaffer1, R.A. Sperling2, J.L. Cummings3, P.S. Aisen1 and the TRC-PAD Investigators*


1. Alzheimer’s Therapeutic Research Institute, University of Southern California, San Diego, CA, USA; 2. Center for Alzheimer Research and Treatment, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; 3. Department of Brain Health, School of Integrated Health Sciences, University of Las Vegas, Nevada; Cleveland Clinic Lou Ruvo Center for Brain Health, USA; * TRC-PAD investigators are listed at

Corresponding Author: S. Walter, Alzheimer’s Therapeutic Research Institute, University of Southern California, San Diego, CA, USA,

J Prev Alz Dis 2020;4(7):234-241
Published online August 13, 2020,


BACKGROUND: The Trial-Ready Cohort for Preclinical and Prodromal Alzheimer’s disease (TRC-PAD) aims to accelerate enrollment for Alzheimer’s disease (AD) clinical trials by remotely identifying and tracking individuals who are at high risk for developing symptoms of AD, and referring these individuals to in-person cognitive and biomarker evaluation with the purpose of engaging them in clinical trials. A risk algorithm using statistical modeling to predict brain amyloidosis will be refined as TRC-PAD advances with a maturing data set.
Objectives: To provide a summary of the steps taken to build this Trial-Ready cohort (TRC) and share results of the first 3 years of enrollment into the program.
Design: Participants are remotely enrolled in the Alzheimer Prevention Trials (APT) Webstudy with quarterly assessments, and through an algorithm identified as potentially at high risk, referred to clinical sites for biomarker confirmation, and enrolled into the TRC.
Setting: Both an online study and in-clinic non-interventional cohort study.
Participants: APT Webstudy participants are aged 50 or older, with an interest in participation in AD therapeutic trials. TRC participants must have a study partner, stable medical condition, and elevated brain amyloid, as measured by amyloid positron emission tomography or cerebrospinal fluid analysis. Additional risk assessments include apolipoprotein E genotyping.
Measurements: In the APT Webstudy, participants complete the Cognitive Function Index and Cogstate Brief Battery. The TRC includes the Preclinical Alzheimer’s Cognitive Composite, comprised of the Free and Cued Selective Reminding Test, the Delayed Paragraph Recall score on the Logical Memory IIa test from the Wechsler Memory Scale, the Digit-Symbol Substitution test from the Wechsler Adult Intelligence Scale-Revised, and the Mini Mental State Examination total score (1).
Results: During the first 3 years of this program, the APT Webstudy has 30,650 consented participants, with 23 sites approved for in person screening, 112 participants have been referred for in-clinic screening visits with eighteen enrolled to the TRC. The majority of participants consented to APT Webstudy have a family history of AD (62%), identify as Caucasian (92.5%), have over twelve years of formal education (85%), and are women (73%). Follow up rates for the first quarterly assessment were 38.2% with 29.5% completing the follow up Cogstate Battery.
Conclusions: After successfully designing and implementing this program, the study team’s priority is to improve diversity of participants both in the APT Webstudy and TRC, to continue enrollment into the TRC to our target of 2,000, and to improve longitudinal retention, while beginning the process of referring TRC participants into clinical trials.

Key words: Alzheimer’s disease, prevention, webstudy, remote study.



The Trial-Ready Cohort for Preclinical and Prodromal Alzheimer’s disease (TRC-PAD) program aims to accelerate enrollment into clinical trials for AD by building a cohort of biomarker-confirmed eligible participants. The first stage of the program is remote recruitment of participants to the Alzheimer Prevention Trials (APT) Webstudy (2). Participants are followed with quarterly assessments, and through an algorithm identified as potentially at high risk, and referred to clinical sites. Participants are screened for Trial-ready cohort (TRC) eligibility, involving cognitive testing, genotyping and amyloid biomarker measures, and then if eligible, enrolled and followed longitudinally until an appropriate clinical trial becomes available. Separate papers summarize the program design and implementation considerations (3), the complex informatics infrastructure (4), the algorithm to predict brain amyloidosis and risk for AD (5), and recruitment strategies (2). Here we expand on the experience of the TRC-PAD program during its initial three years.

Study network and infrastructure

The TRC-PAD program is the result of extensive collaboration between multiple principal investigators (PIs), online National Registries, the Coordinating Center, and the network of clinical trial sites. Registries that referred participants to the APT Webstudy were the Alzheimer’s Prevention Registry (APR), The Alzheimer’s Association TrialMatch, Brain Health Registry (BHR), and Healthy Brains, as well as registries managed by clinical trial sites. The study was coordinated by the Alzheimer’s Therapeutic Research Institute (ATRI) at the University of Southern California (USC). Periodic updates were provided to the clinical trial sites involved in the program during the development phase. A small group of “vanguard sites” were selected first, with their study teams providing feedback on the referral process, before expanding to the total sites. Each clinical site participating in the TRC receives modest financial support for local recruitment and referral efforts, separate from their reimbursement for TRC participant visits. In 2019, the TRC-PAD program became affiliated with the Alzheimer’s Clinical Trials Consortium (ACTC) with scientific guidance provided by the ACTC Steering Committee.

Regulatory oversight

The APT Webstudy is overseen by the University of Southern California (USC) Institutional Review Board (IRB), which reviews and approves all participant-facing content, including the informed consent documents, web pages, emails, newsletters, and quarterly testing reminders. The IRB provided initial approval for the APT Webstudy in November 2017 and the Webstudy launched four weeks later (Figure 1). IRBs overseeing Registries also reviewed recruitment materials. The protocol describing in-person visits, screening, and enrollment in the TRC is overseen by Advarra IRB, the central IRB. In some cases, the local IRBs that oversee the clinical trial sites also required review of materials.

APT Webstudy Participant support

Support is provided in-house by the APT Webstudy team at the USC Alzheimer’s Therapeutic Research Institute (ATRI). Participants may telephone or email the study team with their questions. Using a ticketing and tagging system, each issue is tracked centrally, which allows the support team to identify patterns and trends. Questions are triaged to subject matter experts when needed; for example, to a clinician or technical team member. Issues are reviewed centrally at regular intervals and used to improve the website and study communications.

Retention tools

Retention of study participants and capturing longitudinal assessments, particularly cognitive testing, are critical to the program aims. The APT Webstudy team developed a participant engagement platform to optimize the Webstudy experience. Each participant is provided results of their clinical and cognitive testing over the course of the study. Reminder emails alert participants when the next quarterly assessment is due. In addition, a quarterly newsletter called “Alzheimer’s Research Today” is emailed to all participants, including updates from the field of AD research, describing upcoming studies, and providing information on new features of the Webstudy.

APT Webstudy experience

In order to register for the APT Webstudy, participants are asked to log in using either their existing social login credentials, or to create an account by providing a username, email address and password. Once logged on, participants are considered ‘registered.’ The Webstudy is designed as a ‘walk through’ experience, with each new section opening after completion of the former section.

Step 1

Personal profile. Participants provide basic information including age, race and ethnicity, education, zip code, whether they have been diagnosed with Alzheimer’s disease, and whether they are interested in participating in future AD clinical trials and are willing to share information with clinical sites near them.

Step 2

Consent. Each participant is asked to indicate whether they agree to participate or do not agree to participate. The consent form is displayed online and may be downloaded. Consent is required to move forward and may be revoked at any time.

Step 3

Lifestyle. Participants are asked brief questions about diet and lifestyle. Questions about prior genetic and amyloid testing were added in January 2019, 12 months after the APT Webstudy launched. Participants enrolled prior to this question being included are prompted to respond to these questions the next time they sign on.

Step 4

Remote Cognitive and Functional Assessments. The Cognitive Function Instrument (CFI) is a 15-item participant-reported questionnaire (6, 7). This assessment captures the participant’s perceived ability to perform high level functional tasks in daily life, as well as their sense of overall cognitive functional ability. The participant self-reported CFI has been validated in prior studies to provide early indication of future cognitive decline (7). The Cogstate Brief Battery (CBB), comprised of four simple playing card tasks measuring psychomotor speed and recent memory (8), is used to assess cognition and memory function. The One-Card Learning Test has shown particular sensitivity to amyloid-related decline in preclinical and prodromal AD (9).

Step 5: Review Scores

After completing the remote assessments, the participant can review their CFI scores in a ‘Dashboard’ view. There is a description below the score of the test, explaining what the scores might mean, (e.g. “An increasing score over time might mean cognitive decline”). CBB scores are processed within 2-5 days, and participants are notified by email when scores are available. The website description of the CBB emphasizes that the tool is used for research, and that a change in score between -10 to +10 is considered normal. After completion, the cognitive test questions are no longer available to the participant, and the next available testing date is displayed (3 months from previous test date).

Clinical Site Referrals

Data from the APT Webstudy are evaluated monthly using an adaptive algorithm. This algorithm uses statistical models to assess each participant’s risk of AD amyloidosis (5). In order to be referred, participants must have consented to the APT Webstudy, agreed to share information with researchers, and provided a valid zip code. Participants are ranked by their predicted risk, and those with the highest risk are referred to the nearest TRC-PAD site based on their zip code. Site referrals are provided via a secure web-based tool, the Site Referral System (SRS), with the flow of participants customized to meet the site’s capacity. Site staff reach out to participants using their preferred method of contact, conduct prescreening, and if the participants are interested and appears to be eligible, invites them for an in-person screening visit to confirm eligibility for the Trial-Ready Cohort (TRC).

Trial Ready Cohort (TRC)

The eligibility criteria for TRC-PAD broadly encompass both current and upcoming clinical trials in prodromal and preclinical AD, with the aim of enrolling 2,000 participants; approximately 1,000 preclinical and 1,000 prodromal. Screening is conducted in multiple phases, first confirming clinical and cognitive eligibility and performing apolipoprotein E (APOE) genetic testing. Using this additional information, the participant’s risk assessment is updated and reviewed centrally before screening proceeds to amyloid testing, either by positron emission tomography (PET) imaging or cerebrospinal fluid (CSF) collection by lumbar puncture. Following procedures that were designed and refined for the Anti-Amyloid Treatment in Asymptomatic Alzheimer’s Disease (A4) study (10, 11), participants are told whether they are eligible for the TRC. A 21 CFR Part 11 compliant electronic data capture system was developed by the TRC-PAD study team to manage participant data (4). Broad data-sharing in the TRC consent allow the data to be potentially used as run-in data for downstream clinical trials, minimizing participant burden. Once enrolled in the TRC, participants are followed with clinical and cognitive assessments every 6 months until a clinical trial becomes available at their site. The decision to screen for a clinical trial is entirely that of the participants, with appropriate guidance from their clinician. The protocol is designed to allow participants to re-enter the cohort after participation in another study or a break for any other reason.



APT Webstudy Enrollment

The first major increase in APT Webstudy enrollment followed an article in a San Diego newspaper in February 2018, which resulted in over 2,000 consented participants. Gradually other recruitment initiatives were rolled out, resulting in 10,000 participants in January 2019, doubling to 20,000 participants in August 2019. As of the data cut for this manuscript (April 20, 2020) there are 30,650 participants consented to the APT Webstudy. More details on APT Webstudy recruitment methods and metrics are described in another paper in this series (2).

Table 1. APT and TRC Demographics (April 20, 2020)


APT Webstudy Demographics

Participants consenting to the APT Webstudy range in age from 17 to 94 with the mean age of 64.5, and 98.8% of participants are over the age of 50. A majority (73.0%) of Webstudy participants are female. 62.2% have a parent or sibling diagnosed with AD, and 4.6% report a diagnosis of AD. 85% have post-secondary education, with 14% reporting high school or equivalent education. Participants are 92.5% Caucasian, with 2.3% Hispanic/Latino, 1.5% African American, 1.4% Asian, and 0.2% American Indian and 0.1% Pacific Islander. 53.2% of participants are retired or not working, 30.6% are working full time, and 14.7% part-time. Within the US, geographic distribution of participants is broad, with participants in 59.1% of US counties (2). About half of APT Webstudy participants report no medical concerns, with the other half most commonly reporting high blood pressure (30%), diabetes (8%), or vascular disease (4%). In terms of lifestyle, 74% exercise 1 or more hours per week, and 81% do not drink alcohol regularly (defined as 2 drinks per day or more). Most participants prefer being contacted by email (78.7%) over a phone call (3.1%).

TRC Enrollment

The first referral from SRS to a TRC-PAD Site was in July 2019 with the first screening visit conducted one month later (Figure 1). As additional sites were approved to enroll, screening activity increased to 20 screens per month in late 2019 and into early 2020. As of April 20, 2020, 1,178 participants have been referred to SRS, and 171 (14%) (Figure 2) have been subsequently referred to be screened for TRC. 112 TRC screening visits have been conducted at 9 Sites, with 54 TRC participants completing the amyloid testing, resulting in 25 participants eligible for enrollment into the TRC. 18 participants have completed a baseline visit (Figure 2).

Figure 1. TRC-PAD Program Timeline

igure 2. TRC-PAD Program Funnel


TRC Demographics

Of the 112 participants with an in-person screening visit, participants are aged 60-79, (mean 71.1 SD 10.6), 49.5% are women, and 93% identify as Caucasian.

APT Webstudy Retention and Drop-outs

Participants were most likely to drop from the Webstudy at the point of consent, with 3,307 (9.7%) registering for the Webstudy but not completing the consent and 8,850 (28.9% of consented participants) not completing the initial CBB. Based on feedback from participants through the user support desk, the missed cognitive assessments are due to technical challenges and lack of compatibility of the CBB with smart phones. Retention is a challenge in the Webstudy with only 10,393 (38.2%) returning for their 2nd visit and 7,220 (28.8%) returning for visit 3. 538 participants have completed up to 8 follow up visits. The CBB retention has been lower, with 8,025 (29.5%) completing testing for the 2nd visit, and 5,777 (23%) for the 3rd visit. 461 participants have completed the CBB for up to the 8th follow up visit.

User Support

Since launching the APT Webstudy, over 1,900 inquiries have been received from users, with a majority (78%) received by email. The most frequent reason for support (38%) is regarding the Cogstate testing. 19% of support requests are related to logging into the Webstudy, 7% are questions related to the scores for CFI or Cogstate, 7% are for non Cogstate-related technical support, and the remainder are miscellaneous support needs. Most inquiries require more than one response and took more than 2 days to resolve. Phone inquiries require an average of 20 minutes of staff time to resolve.

Self-report of prior testing

13.03% of the APT Webstudy participants report undergoing prior APOE testing. Of these 3,989 participants with prior testing, 28% report not carrying the APOE-4 risk gene, 33% report one copy of the APOE-4 allele, and 9% reported having 2 copies. 5% reported that they carry the risk gene but do not know the details, and 23% didn’t know the results. In contrast, only 4.03% of participants had prior amyloid testing, with 2.86% having a prior PET Scan, and 1.17% a prior lumbar puncture.

APT clinical and cognitive assessment

Nearly every participant that signed consent completed the CFI (97%), with a majority scoring in normal ranges (Figure 3). 65% of Webstudy participants completed the initial Cogstate testing (Figure 4).

Figure 3. APT Webstudy Cognitive Function Instrument (CFI)

Figure 4. APT Webstudy Cogstate One card learning



We have demonstrated that it is feasible to build a cohort of remotely-consented and enrolled participants with normal cognition, with broad geographic distribution using an unsupervised cognitive assessment battery to evaluate for increased risk for future cognitive decline. This first stage of the TRC-PAD program represents the best in what collaborative science can achieve. The partnership between the PI’s, an experienced Coordinating Center, the network of sites, academic partners, and the valuable experience and advice of investigators overseeing the APR, TrialMatch, HealthyBrains, and BHR have been critical to this success.
In general, the group of individuals enrolled in the APT Webstudy are similar to those enrolled in clinical trials, with most being highly educated and Caucasian, and a majority reporting a family history of AD. We were intentional in designing the APT assessments to be as brief as possible, and believe that low drop-out rates during initial visit is due to this. The most commonly reported problem leading to missing information on the CBB was incompatibility with smart phones; we expect that compatibility will be improved in the future.
Retention to the APT Webstudy is comparable to what has been reported by online Registry studies (12) and remains a significant challenge. Capturing longitudinal information is an important goal of TRC-PAD. More work is needed to understand why participants are not returning, in order to improve content, language, and presentation.
The APT Webstudy and TRC have both recruited a mostly white and highly educated group, which limits the representativeness of clinical trial participants using this program to the general population. We hope to improve accessibility of the APT Webstudy with the recently released Spanish translation and Spanish-language user support.
Providing consistent and knowledgeable user support for a remote Webstudy has been critical to success. We have found great value in using a centralized ticketing system, which consolidates multiple communication channels (e.g. email, telephone) and allows the study team to identify trends and prioritize development and refinement of procedures.
Ultimately, the success of TRC-PAD will be measured by efficient referral of representative participants from TRC-PAD into clinical trials. Can we predict brain amyloid elevation using Webstudy data augmented by in-person assessment, APOE genotyping and eventually plasma amyloid peptide testing (3) to reduce screening amyloid PET expenses? Can we reduce the long recruitment and screening timelines seen in studies like A4 and early symptomatic-stage AD trials? Can we minimize participant and site burden through efficient design and data-sharing between TRC-PAD and clinical trials? How do we enroll an inclusive group of individuals who are representative of the population at greatest risk for cognitive decline due to AD? TRC-PAD remains a work in progress. Continuing adjustments to its design are essential to optimizing its value.


Funding: The study was supported by a grant from NIA/NIH (R01AG053798). The sponsors had no role in the design and conduct of the study, in the collection, analysis, and interpretation of data, in the preparation of the manuscript, or in the review or approval of the manuscript.

Acknowledgments: We would like to acknowledge and thank our participants, the teams at each of the clinical sites, and the USC Alzheimer’s Therapeutic Research Institute (ATRI) Coordinating Center team members whose work made this study possible. In particular, Devon Gessert, Yuliana Cabrera, Emily Voeller, Stefani Bruschi, Jia-Shing So, Marian Wong, Rosio Gonzalez-Beristain, and Godfrey Coker. A full list of TRC-PAD investigators is at:

Ethical standard: Institutional Review Boards (IRBs) approved these studies, and all participants gave informed consent before participating.

Conflict of interest: The authors report grants from National Institute on Aging, during the conduct of the study. None of the authors have additional financial interests, relationships or affiliations relevant to the subject of this manuscript.

Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.



1. Mormino EC, Papp KV, Rentz DM, et al. Early and late change on the preclinical Alzheimer’s cognitive composite in clinically normal older individuals with elevated amyloid β. Alzheimers Dement 2017;13(9):1004-1012
2. Walter S, Clanton TB, Langford OG, Recruitment into the Alzheimer Prevention Trials (APT) Webstudy for a Trial-Ready Cohort for Preclinical and Prodromal Alzheimer’s Disease (TRC-PAD). J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.46
3. Aisen PS, Sperling R., Cummings J, et al. The Trial-Ready Cohort for Preclinical/Prodromal Alzheimer’s Disease (TRC-PAD) Project: An Overview. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.45
4. Jimenez-Maggiora GA , Bruschi S., Raman R, et al. TRC-PAD: Accelerating Recruitment of AD Clinical Trials through Innovative Information Technology. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.48
5. Langford O, Raman R, Sperling RA, et al. Predicting Amyloid Burden to Accelerate Recruitment of Secondary Prevention Clinical Trials. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.44
6. Walsh SP, Raman R, Jones KB, Aisen PS. ADCS Prevention Instrument Project: The Mail-In Cognitive Function Screening Instrument (MCFSI). Alzheimer Dis Assoc Disord 2006;20(4 Suppl 3):S170-8
7. Amariglio RE, Donohue MC, Marshall GA, et al. Tracking Early Decline in Cognitive Function in Older Individuals at Risk for Alzheimer Disease Dementia. JAMA Neurol 2015;72(4):446-454
8. Crook TH, Kay GG, Larrabee GJ, et al. Computer-based cognitive testing. Neuropsychol assess of Neuropsychiatr. and Neuromedical Disord 2009:84-100
9. Darby DG, Brodtmann A, Pietrzak RH, et al. Episodic Memory Decline Predicts Cortical Amyloid Status in Community-Dwelling Older Adults. J Alzheimers Dis 2011;27(3):627-637
10. Sperling RA, Rentz, DM, Johnson KA, et al. The A4 Study: Stopping AD before Symptoms Begin? Sci Transl Med 2014 Mar 19; 6(228): 228fs13
11. Sperling, RA, Donohue, MC, Raman, R, Sun, et al. Association of Factors with Elevated Amyloid burden in Clinically Normal Older Individuals. JAMA Neurol 2020 Apr 6;e200387. Doi: 10.1001/jamaneurol.2020.0387
12. Weiner MW, Nosheny R, Camacho M, et al. The Brain Health Registry: An internet-based platform for recruitment, assessment, and longitudinal monitoring of participants for neuroscience studies. Alzheimers Dement 2018;14(8):1063-1076

What do you want to do ?

New mail

What do you want to do ?

New mail

What do you want to do ?

New mail

What do you want to do ?

New mail

What do you want to do ?

New mail


S. Walter1, T.B. Clanton1, O.G. Langford1, M.S. Rafii1, E.J. Shaffer1, J.D. Grill3, G.A. Jimenez-Maggiora1, R.A. Sperling2, J.L. Cummings4, P.S. Aisen1 and the TRC-PAD Investigators*

1. Alzheimer’s Therapeutic Research Institute, University of Southern California, San Diego, CA, USA; 2. Center for Alzheimer Research and Treatment, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; 3. Institute for Memory Impairments and Neurological Disorders, University of California, Irvine;
4. Department of Brain Health, School of Integrated Health Sciences, University of Las Vegas, Nevada; Cleveland Clinic Lou Ruvo Center for Brain Health, USA;* TRC-PAD Investigators are listed at

Corresponding Author: S. Walter, Alzheimer’s Therapeutic Research Institute, University of Southern California, San Diego, CA, USA,

J Prev Alz Dis 2020;4(7):219-225
Published online August 11, 2020,



BACKGROUND: The Alzheimer Prevention Trials (APT) Webstudy is the first stage in establishing a Trial-ready Cohort for Preclinical and Prodromal Alzheimer’s disease (TRC-PAD). This paper describes recruitment approaches for the APT Webstudy.
Objectives: To remotely enroll a cohort of individuals into a web-based longitudinal observational study. Participants are followed quarterly with brief cognitive and functional assessments, and referred to Sites for in-clinic testing and biomarker confirmation prior to enrolling in the Trial-ready Cohort (TRC).
Design: Participants are referred to the APT Webstudy from existing registries of individuals interested in brain health and Alzheimer’s disease research, as well as through central and site recruitment efforts. The study team utilizes Urchin Tracking Modules (UTM) codes to better understand the impact of electronic recruitment methods.
Setting: A remotely enrolled online study.
Participants: Volunteers who are at least 50 years old and interested in Alzheimer’s research.
Measurements: Demographics and recruitment source of participant where measured by UTM.
Results: 30,650 participants consented to the APT Webstudy as of April 2020, with 69.7% resulting from referrals from online registries. Emails sent by the registry to participants were the most effective means of recruitment. Participants are distributed across the US, and the demographics of the APT Webstudy reflect the referral registries, with 73.1% female, 85.0% highly educated, and 92.5% Caucasian.
Conclusions: We have demonstrated the feasibility of enrolling a remote web-based study utilizing existing registries as a primary referral source. The next priority of the study team is to engage in recruitment initiatives that will improve the diversity of the cohort, towards the goal of clinical trials that better represent the US population.

Key words: Trial-ready cohort, online registry, remote recruitment, web-based, preclinical, Alzheimer’s disease, prevention.



Identifying eligible participants for early intervention Alzheimer’s disease (AD) clinical trials continues to be a significant challenge in the field (1, 2). The overarching aim of the Trial-Ready Cohort in Preclinical and Prodromal Alzheimer’s Disease (TRC-PAD) program is to accelerate enrollment for early stage AD clinical trials (3). This will be accomplished by identifying and screening participants to confirm eligibility for these trials, including amyloid biomarker confirmation, and then monitoring and maintaining engagement with these participants through longitudinal visits until an appropriate trial is available. The considerations behind the design of TRC-PAD are described by Aisen et al. (4). The first step in establishing the Trial-ready Cohort (TRC) was to recruit participants into the Alzheimer Prevention Trials (APT) Webstudy, an online assessment tool designed to serve as a feeder to the in-person TRC-PAD cohort. We projected the APT Webstudy would require between 25,000 and 50,000 participants, with at least 20% participants from under-represented communities, in order to identify enough eligible participants for a planned TRC of n=2,000. The APT Webstudy program requires secure and scalable informatics infrastructure (5), as well as an algorithm to identify participants and rank them by risk of brain amyloidosis and development of AD dementia (6). These elements of the program are described in separate papers in this series.
The APT Webstudy was launched as clinical trials have increasingly utilized web-based tools, including registries, to improve efficiency in screening (7-9). Although leveraging registries to recruit for clinical trials is not a new concept, the establishment of online registries has broadened access to participants who are interested and eligible for studies (10-13). Going further than remote recruitment, Orri et al (14) conducted the first entirely web-based clinical trial run under an Investigational New Drug (IND) application. Digital tools allow researchers to optimize the use of mobile technologies in clinical trials, respond to the preferences of participants (15), and measure and fine-tune communication methods (16). To our knowledge, TRC-PAD is the first program inviting participants from various existing registries to a join a longitudinal Webstudy with identification and referral of high-risk individuals to an in-person TRC. In this article, we describe the preliminary experience of efforts to recruit to APT Webstudy, including from national and local registries, as a unifying path to enrollment in TRC-PAD.



APT Webstudy Experience

Participants log in using either their existing social login credentials or by creating an account and providing a username, email address and password. Once logged on, participants are considered ‘registered.’ The Webstudy is designed as a ‘walk through’ experience, with each new section opening after completion of the former. The sections are: Step 1: Personal Profile; Step 2: Consent; Step 3 Lifestyle; Step 4: Remote cognitive and functional assessments; Step 5: Review scores. Sections are described in more detail in a separate paper in this series (17). The questionnaires and assessments were designed to be brief with a target duration of 15 minutes.


APT Webstudy participants are recruited from multiple sources. For the purposes of this paper, the term registry refers to a online registry, study, or service matching individuals interested in participating in studies or clinical trials to prevent or delay AD dementia. Early in its development, the TRC-PAD study team established partnerships with each of the largest “Feeder” registries, and in collaboration with the managing team or investigators, developed a referral strategy based on the registry’s unique population and existing communication pathways. Each strategy began small and was expanded when we were able to ensure the stability of the Webstudy infrastructure, as well as our capacity to provide user support. Outreach took the form of direct email campaigns highlighting the APT Webstudy on the registry website, e-newsletters, and social media posts. In addition to referrals from registries, both central and site-based strategies were employed.

UTM Codes

Urchin Tracking Modules (UTM) were generated to track participants that registered for the APT Webstudy in response to digital outreach, and were embedded in emails, webpages, and social media advertisements. For some registries, although various outreach activities were utilized, all responses linked back through the registry website, requiring use of a single UTM, and limiting our ability to understand the response rates to different digital communications. Recruitment strategies that did not utilize a UTM included printed materials (i.e., brochures, newsletters and magazines) and earned media (i.e., online and print newspaper articles).

The Alzheimer’s Prevention Registry (APR) (

APR was launched in October 2012 by the Banner Alzheimer’s Institute with the aim of providing a shared resource to the AD scientific community to facilitate enrollment in studies to prevent AD. In 2015, APR began offering an optional APOE genotyping program (GeneMatch) to members ages 55-75 to help match individuals to research studies. As of August 2018, APR enrolled a total 320,000 participants with 75,351 agreeing to the GeneMatch program, and approximately 75,000 agreeing to be contacted by researchers (18). APR participants are primarily women (65.6%) and Caucasian (45.5%); 1.8% are Hispanic/Latino and less than 1% are from other underrepresented groups. It should be noted that these percentages are a reflection of only the 60.8% of APR participants who provided their Race or Ethnicity (Table 1) (19). 14% of APR participants are age 50-59, 35% age 60-69, and 23% age 70-79 (Table 1). The APT Webstudy recruitment strategy began with a pilot phase in April 2018, with batches of emails sent from APR to 7,293 individuals (Figure 1). This was followed by an article in the APR quarterly newsletter introducing the APT Webstudy and posts on APR’s social media accounts. In January 2019, emails were sent in batches to 75,000 registrants inviting them to join the APT Webstudy. In March and April 2020, follow up emails were sent to participants who had not opened the email or clicked the link for the APT Webstudy, with additional reminders scheduled for May 2020.

Alzheimer’s Association TrialMatch (

Alzheimer’s Association TrialMatch ( is a free online matching service that utilizes user’s information to generate a custom report of clinical trials for which they may be a good fit. TrialMatch has a large pool of 322,997 users, with 134,148 providing contact and personal information. Individuals enrolled in TrialMatch indicate whether they are a healthy volunteer (52.8%), a caregiver looking for clinical trials for someone else such as a family member with AD (31.7%), or a person living with the disease looking for trials (13.3%). A small percent (2.2%) of users are entered into TrialMatch by a physician or researcher. Individuals under 50 comprise 35% of the Healthy Volunteers and 20% of all TrialMatch participants. 69% of TrialMatch are over the age of 50. Participants are 73.4% Caucasian, 4.5% Hispanic/Latino, and 65% are women. Women comprise 78% of the healthy controls and 54% of caregivers looking for trials for someone else. 22% of TrialMatch users either care for someone with a diagnosis of AD or have a diagnosis of AD. The first APT Webstudy recruitment campaign began in March 2019, with direct emails targeting 48,000 TrialMatch users living within 200 miles of potential TRC-PAD clinical sites. An additional 33,000 users were invited to join APT Webstudy beginning in December 2019. Emails were sent in batches of 5,000 twice a week, and is ongoing at the time of this manuscript.

The Brain Health Registry (BHR) (

The Brain Health Registry (BHR) ( collects longitudinal health, cognitive, and lifestyle data through detailed self-report questionnaires and online cognitive tests (Cogstate, Lumosity, and MemTrax) (16). BHR was launched in 2014 and currently has baseline data on 56,982 participants. BHR participants are 80.9% Caucasian, 5.3% Hispanic/Latino, 73.9% women, with 73% of participants over the age of 50 (20) (Table 1). The BHR team sent emails to 18,240 participants inviting them to register for the APT Webstudy beginning in March 2019 (Figure 1). Emails were sent in batches of 500 every week. If participants do not respond, two follow-up emails are sent, with a second set of reminder emails 231 and 238 days from their initial email contact. The BHR team also featured the APT Webstudy in their e-newsletter.

Table 1. Feeder Registries and APT Demographics

The Cleveland Clinic Healthy Brains Registry (


The Cleveland Clinic Healthy Brains Registry ( is a longitudinal, web-based symptomatic and lifestyle assessment (21), with over 13,000 registrants, and over half expressing interest in enrolling into clinical trials. HealthyBrains has registrants and newsletter subscribers from across the nation. The highest number of registrants in the US states of Ohio, Nevada, California and Florida. Registrants were invited to join the APT Webstudy through an article on the HealthyBrains website in May 2018, followed by features in two newsletters, sent by email (Figure 1).

Figure 1. Alzheimer Prevention Trials (APT) Webstudy: Feeder Registry Recruitment Campaign Timeline


UCI Consent-to-Contact (C2C) Registry (

UCI Consent-to-Contact (C2C) Registry ( is a confidential online tool to help match local volunteers in Orange County, CA, with research studies at the University of California, Irvine (22). Registrants enroll by providing an email address or by phoning the research site, remotely completing a series of questions regarding medical history and research interests. Beginning in July 2019, 7,300 C2C participants were invited by email to join the APT Webstudy (Figure 1).

Other sources

Anticipating that the registry-based approach would have limitations, especially in identifying eligible participants from under-represented groups, the APT Webstudy team developed recruitment strategies utilizing the TRC-PAD site network as well as other central activities. Sites participating in the TRC-PAD cohort study were identified early in the development of the program, with some agreeing to work locally to recruit participants to the APT Webstudy. Each of the TRC sites were invited to utilize their own databases of individuals interested in clinical research and email information about the APT Webstudy. The TRC-PAD study team provided flyers, postcards, newsletter and email template language, social media content and leaflets describing the APT Webstudy. Language for these materials was approved by the Institutional Review Board (IRB) and UTM codes were generated where appropriate. Sites also held community outreach events, partnered with other local community organizations to share information about the study, advertised on social media, and posted information about the Webstudy on their own webpages. Central recruitment efforts included generating earned media including newspaper and online and print edition magazine articles, local TV interviews, and posting the study on websites for clinical trials and AD. The earned media stories included an article in the San Diego Union Tribune in January 2018, two letters to the editor in May 2019, in local papers that have circulations of 80,000 (Charleston, SC) and 150,000 (Lexington, KY) respectively. Grand Magazine published an online piece about the APT Webstudy on August 12, 2019, generating 54,000 impressions. The Saturday Evening Post, with a circulation of 302,000 and majority of readers over the age of 45, included APT in its January/February 2020 print edition. So far, the only paid advertising was in the form of Facebook advertisements. Facebook ads ran in eight markets for two weeks in November 2018 for a cost of $12,000, and six markets for 5 weeks in August-September 2019 for a cost of $3,000. The ads were targeted geographically and to the largest minority population in each location, based around the location of TRC sites.



APT Webstudy Enrollment: At the time of preparing this mansuscript, there are 30,650 participants consented to the APT Webstudy. Recruitment strategies for the first year were a mix of central and local efforts (Figure 1). The first notable increase was in January 2018 following local newspaper coverage. In March 2018, email referrals were piloted for APR Registry. In April 2018, APR and HealthyBrains introduced the Webstudy in their newsletters. In the first year, 388 participants per month consented to the APT Webstudy on an average. The APR email referrals began in earnest in January 2019, leading to a dramatic increase in consented participants, with 5,196 consenting in January 2019 (Figure 1). This was followed by email referrals from TrialMatch and BHR. In the second year, participants consented to the APT Webstudy on an average of 1,514 per month.


Participants in the APT Webstudy have a mean age of 64.56 with a majority of participants ages 50-59 (28.9%) and 60-69 (44.1%) (Table 1). Most participants identify as women (73.0%), white (92.5%) and more than high school level education (85.0%). 2.3% of APT Webstudy participants self describe as Hispanic/Latino. Although most participants are retired or not working (53.2%), a significant percentage are employed either full (30.6%) or part-time (14.7%) (Table 2). A majority of participants have a family history of AD (62.6%) and do not have a personal diagnosis of AD (94.6%). Further details on lifestyle and medical history are provided on Tables 2 and 3.

Table 2. APT Webstudy Health and Lifestyle

Table 3. APT Webstudy Recruitment by Referral Sourc


Enrollment by Referral sources

At this point in the recruitment to the APT Webstudy, registries were the primary source of participants, with referrals resulting in 69.69% of consented individuals, according to UTM codes. APR was by far the biggest contributer with 38.98% of all APT Webstudy consented participants, followed by 25.40% referred by TrialMatch. Those referred by APR were also slightly more likely to both register and consent to APT (Table 3). All together 15.9% of the APR participants that were contacted consented to APT, compared to 9.8% or less for other registries. Email (32.92%) and websites (40.78%) were the most common mode of referral, however website visits were largely driven by email campaigns. Central media efforts that could be tracked with UTM resulted in 234 participants. The central Facebook ads accounted for 7,800 and 3,000 clicks which translated to 0.15% of consenting participants.

Geographic Distribution

APT Webstudy participants reside in each of the 50 United States (US), the District of Columbia, and Puerto Rico. States with the highest number of consented participants include California (16.63%), Florida (5.65%), New York (4.67%), Texas (4.66%), and Virginia (4.38%). International location is not currently collected. Participants consented to the APT Webstudy reside in 1931 (or 60%) of US counties. The top ten counties with participants consented to APT are San Diego County, CA (n=1621); Orange County, CA (n=861) Maricopa County, AZ (n=764), Los Angeles County, CA (n=612), Cook County, IL (n=443) Charleston County, SC (n=384), Fayette County, KY (n=279), King County, WA (n=270) Pima County, AZ (n=239) and Middlesex County, MA (n=238) (Figure 2).

Figure 2. APT Webstudy Enrollment: Heatmap of US Counties



We have demonstrated that online registries are not only feasible but they are an excellent method to identify and recruit participants for a Webstudy. Participants in a registry have already demonstrated an interest in research and willingness to provide information about themselves. In addition, registries have communication infrastructure and digital platforms designed to engage individuals through educational materials, newsletters and other outreach, which may lead to higher rates of referral. UTM codes were shown to be an effective method to track the referral source in this study. The strategy that yielded highest rates of responses was to first feature the APT Webstudy in the registry’s newsletter, followed by direct email communication to registrants. Although not tracked with separate UTM codes, the consistent increase of participants demonstrates that sending second and third emails to non-responders produces additional participants. Although central media efforts and social media advertising were piloted in this first stage of recruitment, this strategy has not been fully explored as a potential source for remotely enrolled participants.
The registries used in this study had a contact-to-consent rate ranging from 1.8%-15.9%, despite having very similar composition of registrants. This brings up several questions as to best practices. Was the higher rate of consent from APR compared to BHR due to the fact that APR directly targets individuals interested in clinical trials? Could the observed rate of consent to contacted participant be influenced by the level of engagement utilized by the respective registries?
It is not surprising that the demographics of participants in the APT Webstudy are similar in demographics to the registries that referred the majority of participants. However, understanding why such a large majority of participants are women is important. Further research may reveal both barriers to in-person research and preferences for online studies. The low rate of Hispanic/Latino involvement in APT Webstudy can likely be attributed to 2 factors, (1) the low rates of Hispanic/Latino participants in the referral registries and (2) the APT Webstudy and recruitment materials had not been translated into Spanish.
We acknowledge that the APT Webstudy has an inherent selection bias, in that participants must have access to the internet in order to participate. This disproportionately excludes many people from under-represented communities, where according to recent Pew reports, only 57% of Hispanic and African American adults own a laptop or a tablet (23), compared to 82% of Caucasians. Although those over 65 years of age are more likely to use a desktop or tablet to access the internet, lower income Americans, those with less than college education, and black and Hispanic populations, are all more likely to use a cell phone to access the internet (24). Although the APT Webstudy is mobile-friendly, the cognitive testing at present requires use of a tablet or computer. The study team is considering changes to cognitive testing that will allow for the use of smart phones and expand accessibility to all communities. Other researchers (25) have demonstrated that text messages can be an effective communication channel with research participants. Would people be more responsive to a text message inviting them to return for a study visit?
The Spanish language version of the APT Webstudy was launched early in 2020, with efforts underway to optimize the cultural sensitivity of the Webstudy and all participant-facing content. A key aim of the study is to engage in recruitment initiatives that will improve the diversity of the cohort, towards the goal of clinical trials that better represent the US population. For the African-American community in particular, recruitment campaigns will highlight disparities in Alzheimer’s disease risk and care, and the role research and clinical trials can play in effecting change.
This study has several limitations. The feeder registries differ in numerous ways, including sample sizes, aims or purpose, geographic distribution, length of time from when participants were first engaged with, and frequency of participant engagement. The current analyses did not account for these differences. Similarly, varying levels of data were available for participants in feeder registries, preventing combination of data streams for more sophisticated analyses of recruitment efficiency. Recruitment from feeder registries was peformed over multiple years, introducing potential confounding by time. Quantification of site level efforts toward recruitment was minimal, limiting our ability to understand the efficacy of site level efforts relative to using central efforts or these feeder registries.
In conclusion, this study demonstrates the feasibility of recruiting from feeder registries into a common platform for identifying potentially eligible participants for a Trial-ready cohort. A robust sample was assembled in a relatively short period of time that is anticipated to play a key role in the national AD clinical trial agenda.


Acknowledgements: From the Alzheimer’s Assocation, our thanks to Keith Fargo, Stephen Hall, and Martha Tierney. From APR: Jessica Langbaum, Cassandra Kettenhoven, and Nellie High. From Brain Health Registry: Rachel Nosheny, and Joseph Eichenbaum. From University California Irvine Registry we’d like to thank Meagan Witbracht. Coordinating Center staff providing support to APT Webstudy participants are Godfrey Coker and Rocio Gonzalez-Beristain. The informatics development team is Stefania Burschi, Jia-Shing So, and Marian Wong.

Funding: The study was supported by R01AG053798 from NIA/NIH. The sponsors had no role in the design and conduct of the study; in the collection, analysis, and interpretation of data; in the preparation of the manuscript; or in the review or approval of the manuscript.

Ethical standard: Institutional Review Boards (IRBs) approved these studies, and all participants gave informed consent before participating.

Conflict of interest: The authors report grants from National Institute on Aging, during the conduct of the study. None of the authors have additional financial interests, relationships or affiliations relevant to the subject of this manuscript.

Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.



1. Fargo KN, Carrillo MC, Weiner MW, Potter WZ, Khachaturian Z. The crisis in recruitment for clinical trials in Alzheimers and dementia: An action plan for solutions. Alzheimers Dement 2016;12(11):1113-1115.
2. Sperling RA, Jack CR Jr, Aisen PS. Testing the right target and the right drug at the right stage. Sci Transl Med. 2011 Nov 30; 3(111):111cm33.
3. Sperling R, Cummings J, Donohue M, Aisen P. Global Alzheimer’s Platform Trial Ready Cohorts for the Prevention of Alzheimer’s Dementia. J Prev Alzheimers Dis 2016;3:185-7.
4. Aisen PS, Sperling R., Cummings J, et al. The Trial-Ready Cohort for Preclinical/Prodromal Alzheimer’s Disease (TRC-PAD) Project: An Overview. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.45.
5. Jimenez-Maggiora GA , Bruschi S., Raman R, et al. TRC-PAD: Accelerating Recruitment of AD Clinical Trials through Innovative Information Technology. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.48.
6. Langford O, Raman R, Sperling RA, et al. Predicting Amyloid Burden to Accelerate Recruitment of Secondary Prevention Clinical Trials. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.44.
7. Mohebati A, Knutson A, Zhou XK, et al. A web-based screening and accrual strategy for a cancer prevention clinical trial in healthy smokers. Contemp Clin Trials 2012;33(5):942-948.
8. Parker G, Fletcher K, Blanch B, Greenfield L. The ‘real world’ utility of a web-based bipolar disorder screening measure. Acta Psychiatr Scand 2013;127(5):373-380.
9. Thibault-Halman G, Rivers CS, Bailey CS, et al. Predicting Recruitment Feasibility for Acute Spinal Cord Injury Clinical Trials in Canada Using National Registry Data. JNeurotrauma 2017;34(3):599-606.
10. Wessel J, Gupta J, Groot MD. Factors Motivating Individuals to Consider Genetic Testing for Type 2 Diabetes Risk Prediction. Plos One 2016;11(1).
11. Russo R, Coultas D, Ashmore J, et al. Chronic obstructive pulmonary disease self-management activation research trial (COPD–SMART): Results of recruitment and baseline patient characteristics. Contemp Clin Trials 2015;41:192-201.
12. Andersen MR, Schroeder T, Gaul M, Moinpour C, Urban N. Using a Population-Based Cancer Registry for Recruitment of Newly Diagnosed Patients With Ovarian Cancer. Am J Clin Oncol 2005;28(1):17-20.
13. Hein A, Gass P, Walter CB, et al. Computerized patient identification for the EMBRACA clinical trial using real-time data from the PRAEGNANT network for metastatic breast cancer patients. Breast Cancer Res Treat 2016;158(1):59-65.
14. Orri M, Lipset CH, Jacobs BP, Costello AJ, Cummings SR. Web-based trial to evaluate the efficacy and safety of tolterodine ER 4mg in participants with overactive bladder: REMOTE trial. Contemp Clin Trials 2014;38(2):190-197.
15. Perry B, Geoghegan C, Lin L, et al. Patient preferences for using mobile technologies in clinical trials. Contemp Clin Trials Commun 2019;15:100399.
16. Baca-Motes K, Edwards AM, Waalen J, et al. Digital recruitment and enrollment in a remote nationwide trial of screening for undiagnosed atrial fibrillation: Lessons from the randomized, controlled mSToPS trial. Contemp Clin Trials Commun 2019;14:100318.
17. Walter S, Langford OG, Clanton TB, et al. The Trial-Ready Cohort for Preclinical and Prodromal Alzheimer’s Disease (TRC-PAD): Experience from the First 3 Years. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.47.
18. Langbaum JB, Karlawish J, Roberts JS, et al. GeneMatch: A novel recruitment registry using at-home APOE genotyping to enhance referrals to Alzheimer’s prevention studies. Alzheimers Dement 2019;15:515-24.
19. Langbaum JB, High N, Nichols J, Kettenhoven C, Reiman EM, Tariot PN. The Alzheimer’s Prevention Registry: a large internet-based participant recruitment registry to accelerate referrals to Alzheimer’s-focused studies. J Prev Alz Dis 2020; DOI:10.14283/jpad.2020.31.
20. Weiner MW, Nosheny R, Camacho M, et al. The Brain Health Registry: An internet-based platform for recruitment, assessment, and longitudinal monitoring of participants for neuroscience studies. Alzheimers Dement 2018;14(8):1063-1076.
21. Zhong K, Cummings J. From Registry to Randomization. J Prev Alzheimers Dis 2016;3(3):123-126. doi:10.14283/jpad.2016.100
22. Grill, JD, Hoang, D, Gillen, DL et al. Constructing a Local Potential Participant Registry to Improve Alzheimer’s Disease Clinical Research Recruitment. J Alzheimers Dis 2018;63(3):1055-1063.
23. Perrin A, Turner E. Smartphones help blacks, Hispanics bridge some – but not all – digital gaps with whites. Pew Research Study. 2019 Accessed April 5, 2020.
24. Anderson, M. Mobile Technology and home Broadband 2019. Pew Research Study 2019. Accessed May 14, 2020
25. Lincoln KD, Chow TW, Gaines BF BrainWorks: A Comparative Effectiveness Trial to Examine Alzheimer’s Disease Education for Community-Dwelling African Americans. Am J Geriatr Psychiatr. 2019 Jan;27(1):53-61.

What do you want to do ?

New mail



P.S. Aisen1, R.A. Sperling2, J. Cummings3, M.C. Donohue3, O. Langford3, G.A. Jimenez-Maggiora3, R.A. Rissman4, M.S. Rafii3, S. Walter3, T. Clanton3, R. Raman3


1. Alzheimer’s Therapeutic Research Institute, University of Southern California, San Diego, CA, USA; 2. Center for Alzheimer Research and Treatment, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; 3. Department of Brain Health, School of Integrated Health Sciences, University of Las Vegas, Nevada; Cleveland Clinic Lou Ruvo Center for Brain Health, USA; 4. Department of Neurosciences, University of California San Diego, San Diego, CA, USA

Corresponding Author: PS Aisen, Alzheimer’s Therapeutic Research Institute, University of Southern California, San Diego, CA, USA,

J Prev Alz Dis 2020;4(7):208-212
Published online August 11, 2020,



The Trial-Ready Cohort for Preclinical/prodromal Alzheimer’s Disease (TRC-PAD) project is a collaborative effort to establish an efficient mechanism for recruiting participants into very early stage Alzheimer’s disease trials. Clinically normal and mildly symptomatic individuals are followed longitudinally in a web-based component called the Alzheimer’s Prevention Trial Webstudy (APT Webstudy), with quarterly assessment of cognition and subjective concerns. The Webstudy data is used to predict the likelihood of brain amyloid elevation; individuals at relatively high risk are invited for in-person assessment in the TRC screeing phase, during which a cognitive battery is administered and Apolipoprotein E genotype is obtained followed by reassessment of risk of amyloid elevation. After an initial validation study, plasma amyloid peptide ratios will be included in this risk assessment. Based on this second risk calculation, individuals may have amyloid testing by PET scan or lumbar puncture, with those potentially eligible for trials followed in the TRC, while the rest are invited to remain in the APT Webstudy. To date, over 30,000 individuals have participated in the Webstudy; enrollment in the TRC is in its early stage..

Key words: Trial-Ready Cohort, Alzheimer’s disease.



The critical need for effective disease-slowing therapy for Alzheimer’s disease (AD) is among the most important health care challenges. Advances in understanding the biology of AD reveal that the disease has a 15-20 year preclinical period during which individuals are cognitively normal but have fibrillar brain amyloid, a prodromal phase during which mild cognitive impairment is present, and a dementia phase with more severe cognitive and functional compromise (1). Disease-modifying therapies may best be evaluated in the early stages of AD when it seems most feasible to preserve cognition and forestall decline. Amyloid changes are the earliest identifiable biological changes of AD, and recent trials of anti-amyloid agents in patients with prodromal AD and very mild AD dementia suggest that anti-amyloid approaches may be viable therapeutics in early stages of the illness. Aducanumab, BAN2401 and gantenerumab have all had outcomes in recent trials suggesting clinical or biological benefit in symptomatic participants []; suggesting that that earlier treatment, before the extensive accumulation of amyloid plaques and irreversible synaptic damage, may provide clinically meaningful gains. Indeed, it may be the case that all disease-modifying strategies may need to target individuals that are at early, preclinical points on the Alzheimer’s continuum (2).

Very early, large, intervention trials are feasible. The A4 (Anti-Amyloid treatment in Asymptomatic Alzheimer’s) trial, for example, is a multicenter trial in sporadic pre-symptomatic AD that demonstrated that clinically normal individuals 65 years of age and older can be screened for amyloid elevation using positron emission tomography (PET) and enrolled in a long-term, placebo-controlled treatment study (3). A total of 1169 individuals were randomized into A4, though the recruitment process took over three years.
Recruitment challenges are especially severe for trials in preclinical and prodromal AD populations, in which the minimal nature or absence of cognitive symptoms means that individuals do not to seek medical care for memory decline. While AD dementia trials typically recruit from medical practices and clinics specializing in caring for patients with cognitive disorders, the preclinical AD population requires a different approach. Clinically normal A4 participants were identified by screening on the basis of age alone. As expected, about 30% of asymptomatic individuals 65 years or older were amyloid positive by PET.
Therefore a large number of cognitively normal individuals needed to be screened with a lengthy and expensive process (including education, behavioral assessment prior to scanning, then scanning and disclosure) in order to fully enroll this prevention trial. Thus, early stage trials require a method to efficiently connect with individuals who are concerned about their risk for AD and pre-screening to identify those individuals who are at high-risk in order to reduce the high costs and delay associated with a high screen-fail rate (4).
The Trial-Ready Cohort in Preclinical/prodromal Alzheimer’s disease (TRC-PAD) program grew out of a series of meetings of academic and industry investigators, organized by the Global Alzheimer Platform (GAP) to address the challenges of early stage trial recruitment (5). An academic team filed a successful application to the National Institute on Aging, and the program was launched in early 2018. This overview summarizes the considerations behind the design and implementation of TRC-PAD.


Figure 1. The TRC-PAD program. APOE – apolipoprotein E genotype; CFI – Cognitive Function Instrument (6, 7); LP – lumbar puncture; Cogstate – Cogstate Brief Battery (8); PACC – Preclinical Alzheimer Cognitive Composite (9)


Overall design elements and the APT Webstudy

The TRC-PAD project aims to establish a recruitment infrastructure for early stage AD trials that will shorten the enrollment period from years to months. Participants are drawn from existing registries (“feeders”) plus media and outreach efforts to join the Alzheimer Prevention Trials Webstudy (APT Webstudy). The Webstudy is an online tool designed to collect brief information on demographics, family history, medical history, and subjective cognitive concerns. Unsupervised cognitive assessment collects data on intellectual and memory function relevant to possible early AD. Participants are asked to return to the site quarterly to provide longitudinal cognitive and subjective data. Each participant’s demographic and cognitive data inform his/her individualized risk assessment. The APT Webstudy data is analyzed in an adaptive algorithm using statistical models to determine likelihood of elevation in brain amyloid; initial algorithms are based primarily on analysis of the pre-randomization data from the A4 trial (10). Based on the risk determination, as well as proximity to active TRC-PAD clinical sites and the entry criteria for available trials, individuals may be invited for in-person assessment (including Preclinical Alzheimer Cognitive Composite (PACC) testing and apolipoprotein E (APOE) genotyping) and, based on the updated risk assessment, amyloid testing by amyloid PET or lumbar puncture for measurement of cerebrospinal fluid (CSF) amyloid peptides. Those with amyloid results consistent with AD are invited to be cohort participants, followed in-person longitudinally and ready for enrollment into trials. Those without amyloid abnormalities continue to be followed remotely in the APT Webstudy to continue to provide data for updated risk assessments.
The demographic characteristics of individuals currently enrolled in the APT Webstudy are provided in the companion paper in this issue (11).


Building on existing registries

In addition to common strategies such as earned media coverage and social media advertising, we sought to build on prior efforts to connect with the concerned, aging population through registries. Examples of such registries are the Brain Health Registry (BHR) (2), the Alzheimer’s Prevention Registry (APR) (13) and the Alzheimer’s Association TrialMatch program ( We partnered with investigators from these efforts to inform and invite registrants to the APT Webstudy. The APR, with 75,000 registrants agreeing to be contacted by researchers, was particularly successful in generating Webstudy participants.
While we have exceeded our anticipated rate of accrual with 30,000 consented participants to the APT Webstudy, and a rate of 1,000 participants consenting every month in the past year, we have not been successful in attracting an inclusive group of participants representative of the U.S. population (Walter et al, 2020). The priority of this next phase of the program will be to address this deficiency through recruitment in Spanish language, and other community-based approaches.


Designing a low-burden, informative longitudinal study to assess risk

A challenge noted by registries in the field is that participant retention can be low. In the APT Webstudy, we can estimate risk of amyloid elevation using cross-sectional data from the pre-randomization phase of A4, but longitudinal change in subjective concerns and cognitive performance are expected to significantly improve accuracy. We have tried to improve retention by minimizing participant burden, keeping follow-up visits to 20 minutes or less, and by optimizing engagement, through sharing of graphical representations of longitudinal performance as well as up-to-date information on available and expected therapeutic trials using the Webstudy itself as well as quarterly newsletters. We provide timely responses, by email or phone, to all queries from participants. These efforts are ongoing; more work toward this goal is required.


SRS: a data system to connect high-risk Webstudy participants to TRC sites for in-person testing

Webstudy participants determined to have relatively high risk for amyloid elevation in brain and are located near a TRC-PAD clinical site are invited to have in-person assessments to screen for enrollment into the TRC. In addition to predicted amyloid PET SUVr levels, the selection process considers demographics to achieve diversity, particularly important since Webstudy participants tend to be homogeneous. At this time, the final selections are manually reviewed; after gaining more experience with the system, we will increase automation. Selected participants are presented to site teams through the Site Referral System (SRS) described in a later paper in this series (15). In instances where participants in the APT Webstudy are do not reside close to a TRC-PAD site, they are provided with the opportunity to download and print a report that displays their performance on the various assessments as well as an explanation of the assessments, that they can review with their healthcare provider.
TRC sites are provided a list of potential participants on a monthly basis; the size of the geographic referral area and the number of participants to be referred customized based on site capacity and recruitment needs.


Recalculation of risk and assessment of potential trial eligibility to select for amyloid imaging

Amyloid testing, by PET or CSF analysis, is an expensive and somewhat invasive component of the assessment of early stage trial eligibility. TRC-PAD aims to dramatically reduce the number of amyloid tests required to recruit trial participants. The first in-person visit of participants referred via the SRS to TRC sites includes confirmation of demographic information, medical and neurological assessment, cognitive testing with the PACC and APOE genotyping; these data allow a more precise prediction of brain amyloid level. APOE genotype in particular substantially improves prediction of brain amyloid; if APOE genotype is included in the risk assessment, almost all selected would be APOE ε4 carriers. Our target trial sample will have a distribution of APOE genotypes representative of the AD population, meaning 30-40% APOE ε4 non-carriers. We therefore assess risk separately for carriers and non-carriers to allow control over final genetic distribution. Again, the selection process for amyloid testing permits adjustment to support diversity goals.


Enrollment in TRC based on SUVr or CSF amyloid peptide ratio

Eligibility criteria for the TRC is based on criteria, including amyloid levels, for preclinical and prodromal clinical trials anticipated to be available at each site. The AHEAD 3-45 platform, a public private partnership collaboration of the NIA Alzheimer’s Clinical Trials Consortium and Eisai Pharmaceuticals including most TRC sites, is currently in its start-up phase; this program will enroll clinically normal individuals with elevated and intermediate levels of amyloid. Current TRC amyloid requirements are based on the this platform. Amyloid-eligible individuals are invited to join the TRC for semiannual in-person reassessment including PACC testing.


Connection to early-stage clinical trials

The informatics architecture for TRC-PAD envisions use of longitudinal TRC data as run-in data for therapeutic trials. The system is seamlessly integrated with the Alzheimer’s Treatment Research Institute/Alzheimer’s Clinical Trial Consortium (ATRI/ACTC) Electronic Data System (EDC), is 21 CFR Part 11 compliant, and supports the inclusion of TRC data in trial datsets.
Selection of TRC participants to specific trials available at a site is based on the preferences of participants in discussions with their site investigators. While TRC-PAD procedures are designed with ongoing or coming ATRI/ACTC trials in mind, participants may choose to be screened for any available trials. Additionally, the TRC is designed to allow participants to return after beling either screened or participating in a clinical trial, meeting the important need to the field of retaining and following screen fails.


Incorporation of plasma abeta ratios into the TRC-PAD amyloid risk assessment

In a newly funded revision of the TRC-PAD program, we are now in the process of integrating plasma amyloid peptide ratio assays into the final risk assessment in-person screening prior to brain amyloid testing. The promise of plasma amyloid ratio testing has been confirmed by two independent labs using different immunoprecipitation/mass spectrometry approaches (16, 17); each finds a strong association between plasma ratios and brain amyloid load. Encouraging results have also been reported using an automated immunoassay (18). We will assess these methods by obtaining plasma prior to brain amyloid testing for the initial few hundred APT Webstudy participants to undergo brain amyloid PET. The optimal pre-processing approach and assay methodology will then be incorporated into the risk algorithm for the remaining participants. We expect a substantial improvement in accuracy of our algorithm, as well as a significant reduction the number of negative amyloid PET scans and CSF draws, reducing burden to participants and high cost of screening.


TRC-PAD and Primary Prevention of AD

Our ultimate goal is the primary prevention of AD. This will require monitoring individuals prior to amyloid elevation in brain to identify characteristics (demographic, genetic, biochemical, clinical) that predict later amyloid elevation, enabling the selection of high-risk people for primary prevention trials involving reducing production or promoting clearance of amyloid peptides. We believe that the APT Webstudy, with the addition of remote acquisition of DNA, and longitudinal collection of blood to assess Aβ42/Aβ40 ratios over time (5), will provide the necessary infrastructure for this effort. Plasma assays of Aβ42/Aβ40 followed longitudinally will be key; encouraging data suggest that plasma amyloid ratios predict later amyloid PET positivity (16).
The TRC-PAD program is a work in progress. While we have passed our initial target of 25,000 registrants in the APT Webstudy, TRC screening and amyloid testing are still in a very early stage, and validation of a plasma abeta ratio assay is still in the future. Many investigators across the U.S. and around the world are contributing to the continued optimization and implementation of TRC-PAD. We hope that this program will accelerate recruitment into early intervention AD trials and facilitate work toward the primary prevention of AD.


Acknowledgements: The authors are grateful for the enormous contributions of the entire TRC-PAD team, listed at:

Funding: The study was supported primarily by R01 AG053798 from NIA/NIH. The sponsors had no role in the design and conduct of the study; in the collection, analysis, and interpretation of data; in the preparation of the manuscript; or in the review or approval of the manuscript.

Conflicts of interest: This work was supported by grants from National Institute on Aging. None of the authors have additional financial interests, relationships or affiliations relevant to the subject of this manuscript.

Ethical Standards: Institutional Review Boards approved these studies, and all participants gave informed consent before participating.

Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.



1. Jack CR, Jr., Albert MS, Knopman DS, et al. Introduction to the recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 2011;7(3):257-262.
2. Aisen PS, Cummings J, Jack CR, Jr., et al. On the path to 2025: understanding the Alzheimer’s disease continuum. Alzheimers Res Ther. 2017;9(1):60.
3. Sperling RA, Donohue MC, Raman R, et al. Association of Factors With Elevated Amyloid Burden in Clinically Normal Older Individuals. JAMA Neurol. 2020.
4. Rafii MS, Aisen PS. Alzheimer’s Disease Clinical Trials: Moving Toward Successful Prevention. CNS Drugs. 2019;33(2):99-106.
5. Sperling R, Cummings J, Donohue M, Aisen P. Global Alzheimer’s Platform Trial Ready Cohorts for the Prevention of Alzheimer’s Dementia. J Prev Alzheimers Dis. 2016;3(4):185-187.
6. Walsh SP, Raman R, Jones KB, Aisen PS, Alzheimer’s Disease Cooperative Study G. ADCS Prevention Instrument Project: the Mail-In Cognitive Function Screening Instrument (MCFSI). Alzheimer Dis Assoc Disord. 2006;20(4 Suppl 3):S170-178.
7. Amariglio RE, Donohue MC, Marshall GA, et al. Tracking early decline in cognitive function in older individuals at risk for Alzheimer disease dementia: the Alzheimer’s Disease Cooperative Study Cognitive Function Instrument. JAMA Neurol. 2015;72(4):446-454.
8. Maruff P, Thomas E, Cysique L, et al. Validity of the CogState brief battery: relationship to standardized tests and sensitivity to cognitive impairment in mild traumatic brain injury, schizophrenia, and AIDS dementia complex. Arch Clin Neuropsychol. 2009;24(2):165-178.
9. Donohue MC, Sperling RA, Salmon DP, et al. The preclinical Alzheimer cognitive composite: measuring amyloid-related decline. JAMA Neurol. 2014;71(8):961-970.
10. Langford O, Raman R, Sperling RA, et al. Predicting Amyloid Burden to Accelerate Recruitment of Secondary Prevention Clinical Trials. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.44
11. Walter S, Langford OG, Clanton TB, et al. The Trial-Ready Cohort for Preclinical and Prodromal Alzheimer’s Disease (TRC-PAD): Experience from the First 3 Years. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.47
12. Weiner MW, Nosheny R, Camacho M, et al. The Brain Health Registry: An internet-based platform for recruitment, assessment, and longitudinal monitoring of participants for neuroscience studies. Alzheimers Dement. 2018;14(8):1063-1076.
13. Langbaum JB, Karlawish J, Roberts JS, et al. GeneMatch: A novel recruitment registry using at-home APOE genotyping to enhance referrals to Alzheimer’s prevention studies. Alzheimers Dement. 2019;15(4):515-524.
14. Walter S, Clanton TB, Langford OG, Recruitment into the Alzheimer Prevention Trials (APT) Webstudy for a Trial-Ready Cohort for Preclinical and Prodromal Alzheimer’s Disease (TRC-PAD). J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.46
15. Jimenez-Maggiora GA , Bruschi S., Raman R, et al. TRC-PAD: Accelerating Recruitment of AD Clinical Trials through Innovative Information Technology. J Prev Alz Dis 2020; DOI: 10.14283/jpad.2020.48.
16. Schindler SE, Bollinger JG, Ovod V, et al. High-precision plasma beta-amyloid 42/40 predicts current and future brain amyloidosis. Neurology. 2019;93(17):e1647-e1659.
17. Nakamura A, Kaneko N, Villemagne VL, et al. High performance plasma amyloid-beta biomarkers for Alzheimer’s disease. Nature. 2018;554(7691):249-254.
18. Palmqvist S, Janelidze S, Stomrud E, et al. Performance of Fully Automated Plasma Assays as Screening Tests for Alzheimer Disease-Related beta-Amyloid Status. JAMA Neurol. 2019.

What do you want to do ?

New mail

What do you want to do ?

New mail