jpad journal

AND option

OR option



M.W. Weiner1,2,3, P.S. Aisen4, L.A. Beckett5, R.C. Green6, W. Jagust7, J.C. Morris8, O. Okonkwo9, R.J. Perrin8, R.C. Petersen10, M. Rivera Mindt11, A.J. Saykin12,13, L.M. Shaw14, A.W. Toga15, J.Q. Trojanowski16 and the Alzheimer’s Disease Neuroimaging Initiative (ADNI)


1. Department of Veterans Affairs Medical Center, Center for Imaging of Neurodegenerative Diseases, San Francisco, CA, USA; 2. Departments of Radiology, Medicine, Psychiatry, and Neurology, University of California, San Francisco, CA, USA; 3. Principal Investigator of the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the Brain Health Registry; 4. Alzheimer’s Therapeutic Research Institute, University of Southern California, San Diego, CA, USA; 5. Division of Biostatistics, Department of Public Health Sciences, University of California, Davis, CA, USA; 6. Department of Neurology and Center for Neuroscience, University of California, Davis, Davis, CA, USA; 7. Helen Wills Neuroscience Institute, University of California Berkeley, Berkeley, CA, USA; 8. Departments of Pathology and Immunology and of Neurology, Washington University School of Medicine, Saint Louis, MO, USA; 9. Wisconsin Alzheimer’s Disease Research Center and Department of Medicine, University of Wisconsin School of Medicine and Public Health, Madison, WI, USA; 10. Department of Neurology, Mayo Clinic, Rochester, MN, USA; 11. Department of Psychology, Latin American Latino Studies Institute, and African and African American Studies, Fordham University, New York, NY, USA, and Departments of Neurology and Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY, USA; 12. Department of Radiology and Imaging Sciences and Indiana Alzheimer’s Disease Research Center, Indiana University School of Medicine, Indianapolis, IN, USA; 13. Department of Medical and Molecular Genetics, Indiana University School of Medicine, Indianapolis, IN, USA; 14. Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA; 15. Laboratory of Neuroimaging, USC Stevens Institute of Neuroimaging and Informatics, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA; 16. Department of Geriatric Medicine and Gerontology, University of Pennsylvania Health System. Philadelphia, PA, USA.

Corresponding Author: Michael W. Weiner, NCIRE, 4150 Clement St, San Francisco, CA 94121, USA. E-mail address:

J Prev Alz Dis 2021;
Published online July 22, 2021,


The accelerated approval of aducanumab (AduhelmTM) by the US FDA is a momentous event. For the first time, a therapeutic agent that targets the neurobiology of Alzheimer’s disease (AD) is available for clinical use (1, 2). In addition to the FDA approval of aducanumab, the FDA has also provided “Breakthrough therapy designation” for Lilly’s Donanemab and Eisai’s Lecnemab which also are monoclonal antibodies that remove brain amyloid plaques and may slow cognitive decline. Aducanumab approval will impact clinical practice. The effects on AD clinical research will be profound in both positive and negative ways. This Editorial reflects the opinion of the leadership of the Alzheimer’s Disease Neuroimaging Initiative (ADNI), a large multisite longitudinal observational study with the goal of validating biomarkers for clinical trials. ADNI data have been used to help design and statistically power many AD clinical trials, including the aducanumab studies.


Positive Impacts

Setting aside interpretation of trial data statistical issues, we hope that some patients who receive aducanumab will benefit from aducanumab-induced brain amyloid reduction with slowing of their cognitive and functional decline.
We believe that FDA approval of aducanumab will help change the public perception of AD. Many in the public consider cognitive decline and dementia to be a normal consequence of aging with no effective treatment. This nihilistic view prevents patients and their families from seeking medical evaluation and is one reason why recruitment of participants into AD clinical research, including therapeutic trials, is difficult. Unfortunately, this perception is commonly held by many healthcare providers. The recent approval should help all healthcare professionals to understand that cognitive decline and dementia in older people are caused by diseases, most commonly AD, and that a medical workup for such patients is needed now that an approved treatment which may slow decline is available. We hope that the widespread media attention surrounding the approval will serve to educate the public, promote more patient visits to their healthcare providers, and lead to greater participation in clinical research.
Another potential outcome of aducanumab approval may be more investment by government and the private sector in neurodegenerative disease research. For some time, there has been negative press coverage about failed trials and speculation that amyloid is a failed target. In the past, we have heard from many in the investment community that there is no interest in treatments aimed at amyloid, because they are ineffective. The mood should now shift as the scientific community (admittedly with some exceptions) recognizes that treatments aimed at removing amyloid can be approved. We likewise hope that investment will now extend to all plausible therapeutics aimed at targets such as tau, α-synuclein, and TDP-43, as well as cerebrovascular disease, inflammation, metabolic dysregulation, and other disease mechanisms. The presence of such co-pathologies may contribute to continued cognitive decline after amyloid removal. In the future, more effective treatment for AD may be through combination therapy aimed at multiple pathologies. Therefore, one positive outcome of the aducanumab experience may be increased awareness of the need for more research to develop diagnostics and therapeutics for these co-pathologies.
The approval of aducanumab will likely also increase investment in diagnostics. The use of amyloid PET scans and cerebrospinal fluid amyloid and tau peptide assays will increase because of aducanumab approval. There is huge excitement in the AD research community about plasma tests for amyloid, phosphorylated tau, neurofilament light and others; blood tests are certainly less expensive and less invasive than PET scans or lumbar punctures, and more widely available.
Most AD clinical studies, including clinical trials such as the aducanumab trials, draw participant groups that profoundly under-represent people of ethnoculturally diverse backgrounds and people with lower education, and those from low resource backgrounds. It is unclear to what extent insurers will pay for aducanumab and its associated diagnostic costs. Therefore, the approval of aducanumab once again emphasizes the same disparities in healthcare we have seen repeatedly in other disease fields and with COVID-19. All of us involved in this work must strive to be more inclusive and to help reduce dementia disparities.


Negative Impacts

The approval of aducanumab will have some negative effects on AD research. One concern is that this approval might lead to false hope, ultimately causing disappointment and disillusionment. The reported clinical benefits are modest and may not be apparent to individuals. There are also potentially serious side effects, specifically amyloid-related imaging abnormalities (ARIA), which requires MRI monitoring for detection. It is possible that if the treatment is used by inexperienced clinicians, adverse consequences may become more common. Taken together with the costs of treatment and associated diagnostics, this may end up leading to a public sense of disappointment and reduced participation in clinical research.
We believe that in the short term, aducanumab approval is likely to make all types of AD clinical research including observational studies like ADNI, and randomized clinical trials, more difficult. Many AD experts report being swamped with calls from their patients and others asking about aducanumab and requesting visits. As aducanumab becomes available to more and more doctors, especially those involved in research, the resources which are currently devoted to research (clinician time, staff, examination rooms, PET and MRI scanners) will be diverted to provide treatment. COVID-19 has severely slowed clinical AD research and we can expect aducanumab approval to further delay the recovery of research activities. Of course, in time, the growing need for AD specialists, including physicians, neuropsychologists, and nurses, clinical space, and scanners will result in more investment and growth in this area. But research institutions, especially universities where most clinical AD research is performed, are very slow to respond to growing needs. For example, it may take years to find or build space for new MRI or PET scanners in the university environment.
Another way that aducanumab will make clinical studies more difficult is that many patients and their families will choose to be treated with the approved aducanumab instead of joining an observational study such as ADNI or participating in a randomized trial of an unapproved (however promising) treatment. We’ve seen this story play out in other disease fields – the first approved treatment becomes the preferred treatment for many, and this greatly complicates observational studies and other treatment trials. On the other hand, many studies and trials will adjust their protocols and analysis plans to allow use of aducanumab by participants.
In summary, the FDA approval of aducanumab will have both positive and negative effects. We hope that at least some appropriately treated patients may benefit, and the public and clinicians will become much better educated about AD. Clinical research will be complicated by inclusion of people taking disease-altering therapy. Public resources, notably Medicare, will be strained by the need to provide aducanumab to the large number of appropriate patients. We anticipate a rapid surge in investment in AD treatment centers, diagnostic technologies, as well in developing improved therapeutics which slow, stop, and prevent the progressive symptoms of AD and other neurodegenerative diseases.


Disclosures: Dr. Aisen’s work is supported by research agreements with Lilly and Eisai and research grants from NIA and the Alzheimer’s Association; he receives consulting fees from Biogen, Roche, Merck, Abbvie, Immunobrain Checkpoint Rainbow Medical and Shionogi. Dr. Beckett is supported by NIH grants U01AG024904, R01AG062517, and B639943. Dr. Green is supported by NIH grants AG024904, AG047866, HG009922, HL143295, OD026553, TR003201, HG008685 and the Franca Sozzani Fund. Dr. Green has no financial interests in any pharmaceutical companies. : Dr. Jagust receives grant funding from the NIH/NIA, the Rainwater Foundation, and Roche/Genentech. He serves as a consultant to Bioclinica, Biogen, and CuraSen. Dr. Morris is funded by NIH grants P30 AG066444; P01AG003991; P01AG026276; U19 AG032438; and U19 AG024904. Neither Dr. Morris nor his family owns stock or has equity interest (outside of mutual funds or other externally directed accounts) in any pharmaceutical or biotechnology company. Dr. Perrin’s laboratory provides autopsy/tissue donationservices for local participants enrolled in Biogen’s clinical trials of ALS treatments. Dr. Petersen receives funding from NIH grants P30 AG0620677, U01 AG006786, U01 AG024904 and U24 AG057437, is a consultant for Roche, Merck, Biogen , and Eisai, and serves on the Data Safety Monitoring Board of Genentech. Dr. Rivera Mindt’s work is supported by NIH/NIA grants (R01AG065110, R13AG071313, SC3GM141996, R01AG066471-01A1, and U19AG024904), the Alzheimer’s Association (AARGD-16-446038), and the Genentech Health Equity Fund (G-89294). Dr. Saykin is supported by multiple NIH grants (P30 AG010133, P30 AG072976, R01 AG019771, R01 AG057739, U01 AG024904, R01 LM013463, R01 AG068193, T32 AG071444, and U01 AG068057). He has also received support from Avid Radiopharmaceuticals, a subsidiary of Eli Lilly (in kind contribution of PET tracer precursor); Bayer Oncology (Scientific Advisory Board); and Springer-Nature Publishing (Editorial Office Support as Editor-in-Chief, Brain Imaging and Behavior). Dr. Shaw is supported by NIH/NIA grants, ADI biomarker Core, Penn ADRC Biomarker Core, the M.J. Fox Foundations, and Roche IIS AD biomarker studies. Dr. Shaw receives honoraria from the Biogen teaching program on fluid biomarkers, and the Fujirebio teaching program on biomarkers. Drs. Beckett, Okonkwo, Rivera-Mindt, Toga Trojanowski have no conflicts to report.



1. US Food & Drug Administration FDA grants accelerated approval for Alzheimer’s drug. Date: 2021 Date accessed: June 17, 2021
2. Cummings J., Aisen P., Apostolova L.G. et al. Aducanumab: Appropriate Use Recommendations. J Prev Alz Dis; in press; DOI:



B. Kallmyer, M. Daven, L. Thornhill, K. Clifford, R. Conant, M. Carrillo


Alzheimer’s Association, 225 N. Michigan Ave. Floor 17, Chicago, IL, USA.

Corresponding Author: Maria C. Carrillo, PhD, Alzheimer’s Association, 225 N. Michigan Ave. Floor 17, Chicago, IL 60601, E-mail:, (312)335-5722


The more than 6 million Americans living with Alzheimer’s disease face a future filled with progressive loss of their cognitive abilities ending with certain death (1). They will eventually require help in all aspects of daily living, and that help is provided by over 11 million unpaid caregivers (2). At this time, Alzheimer’s remains a clinical diagnosis and unfortunately, many individuals who would meet the diagnostic criteria are not diagnosed (3). The Food and Drug Administration’s (FDA) accelerated approval of aducanumab (Aduhelm™) as a treatment for Alzheimer’s makes early detection, accurate diagnosis and quality care even more critical, to ensure individuals receive the most benefit at the earliest point possible. Furthermore, the approval of this treatment opens up a new landscape in Alzheimer’s care that comes with many implications for effective public policy to enhance access to quality care.
This drug is a complement to comprehensive care, not a substitute for care. All individuals living with Alzheimer’s, at every stage of the disease, should have access to high quality comprehensive care, including care planning, management, and coordination. However, access to quality dementia care is limited for many older adults, due to a shortage of specialty physicians (4). Primary care providers have limited capacity and expertise to assist in all the areas of need associated with dementia (5), and especially traditionally non-medical activities such as counseling, education, and referrals to community-based organizations (6). Meeting the growing demand for dementia care requires increasing the specialty workforce and improving capacity in primary care, including the expansion of successful collaborative and coordinated care models/programs that use primary care providers. Pilot programs for individuals with dementia have reduced hospital and emergency room visits and nursing home placement (7). A change in payment structure to value-based payment is also necessary to enable effective dementia care management. The Comprehensive Care for Alzheimer’s Act (S. 1125/H.R. 2517) would ask the Center for Medicare & Medicaid Innovation (CMMI) to implement a dementia care management model to test the effectiveness of comprehensive care management services. This dementia care management model would provide comprehensive care services including caregiver education and support, ensure patients have access to providers with dementia care expertise, and reimburse providers through payment based on performance.
Black and Hispanic Americans are at increased risk for Alzheimer’s and related dementia yet have been underrepresented in the Aduhelm clinical trials and most other clinical trials for dementia treatments in the United States. Stigma, cultural differences, awareness and understanding, the ability to obtain a diagnosis, manage the disease, and access care and support services for dementia vary widely depending on race, ethnicity, geography and socioeconomic status (8). The Alzheimer’s Association supports the bipartisan Equity in Neuroscience and Alzheimer’s Clinical Trials (ENACT) Act (S. 1548/H.R. 3085), which would increase the participation of underrepresented populations in Alzheimer’s and other dementia clinical trials by expanding education and outreach to these populations, encouraging the diversity of clinical trial staff, and reducing participation burden, among other priorities.
The accelerated approval of Aduhelm is an important step in addressing the underlying biology of Alzheimer’s disease, as the first FDA-approved treatment to reduce one of the defining brain changes of Alzheimer’s, known as amyloid plaques. However, it is not a cure, and more research is needed to determine if removing amyloid plaques will be effective to slow clinical decline. It is also very clear that more research is necessary to advance other therapeutic approaches that include the acceleration of anti tau approaches and so many others, for the ultimate goal of treating this complex disease with combination therapies and to treat populations that respond differently. That is why it is imperative that Congress continue its commitment to increased investments in Alzheimer’s research at the National Institutes of Health (NIH). Recent NIH funding increases have laid the foundation for breakthroughs in diagnosis, treatment, and prevention, and enabled significant advances in understanding the complexities of Alzheimer’s, but there is still much left to be done. Investment in Alzheimer’s research is only a fraction of what’s been applied over time, with great success, to address other major diseases. An increase of $289 million in Alzheimer’s research at the NIH in Fiscal Year 2022 would enable scientists to maximize every opportunity for success.
Until disease modifying treatments are developed and widely available and accessible, a public health approach to Alzheimer’s disease plays a critical role in promoting prevention, early detection, accurate diagnosis, comprehensive care management, and support for caregivers. Through their mandated responsibility to protect the public’s health, governmental public health agencies can advance proven strategies to support and maintain the health, well-being, and productivity of caregivers. In 2018, Congress acted decisively to address Alzheimer’s as an urgent and growing public health threat through the passage of the bipartisan BOLD Infrastructure for Alzheimer’s Act. This law authorizes $100 million over five years for the Centers for Disease Control and Prevention (CDC) to build a robust Alzheimer’s public health infrastructure across the country focused on public health actions. Congressional appropriations have allowed CDC to award funding to three Public Health Centers of Excellence focused on risk reduction, caregiving, and early detection, and 16 public health departments across the country. While these BOLD implementation efforts are important steps forward, CDC must receive the full $20 million authorized in the law for FY2022 to ensure the meaningful impact that Congress intended.
Early detection, an accurate diagnosis and access to dementia care are now more important than ever. The possibility for some individuals living with dementia to benefit from a new type of treatment presents new opportunities and new challenges for public and private leadership, to increase clinical capabilities and institute policies for access to quality care. The Alzheimer’s Association will do everything in its power to ensure access to any treatment, diagnostic tests needed during the treatment process, and other associated costs for all who could benefit. Ensuring access is part of our commitment to ensuring quality, person-centered care for all who are affected by this devastating disease. The Alzheimer’s Association likewise calls on leaders in health care and health policy to improve access to dementia detection, diagnosis and care, and the Association is here to work with others to this end.


Disclosures: All Authors are full time employees of the Alzheimer’s Association. Organizational disclosures can be found at:



1. Rajan KB, Weuve J, Barnes LL, McAninch EA, Wilson RS, Evans DA. Population estimate of people with clinical AD and mild cognitive impairment in the United States (2020-2060). Alzheimers Dement 2021;17. In press.
2. Alzheimer’s Association. 2021 Alzheimer’s Disease Facts and Figures. Alzheimers Dement 2021;17(3).
3. Alzheimer’s Association. 2021 Alzheimer’s Disease Facts and Figures. Alzheimers Dement 2021;17(3).
4. Alzheimer’s Association. 2020 Alzheimer’s disease facts and figures. Alzheimer’s Dement., 16: 433.
5. Bernstein et al. 2019
6. Reuben et al., 2003; Reuben et al., 2010
7. Alzheimer’s Association. 2020 Alzheimer’s disease facts and figures. Alzheimer’s Dement., 16: 439.
8. Alzheimer’s Association. 2021 Alzheimer’s Disease Facts and Figures. Alzheimers Dement 2021;17(3).



P. Scheltens1, E.G.B. Vijverberg1,2


1. Alzheimer Center Amsterdam, Department of Neurology, Amsterdam Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam, The Netherlands; 2. Brain Research Center, Amsterdam, The Netherlands

Corresponding Author: Philip Scheltens, MD,PhD, Alzheimercentrum Amsterdam, Amsterdam UMC, Locatie VUmc, De Boelelaan 1117/1118, 1081 HZ Amsterdam, T: +31 (0)20 4440816,


June 7, 2021 will not likely be forgotten soon by many Alzheimer Disease (AD) researchers. To paraphrase a famous quote: ”a small step for man, but a giant leap forwards for the field”. That day, AduhelmTM (aducanumab) was approved by the US Food and Drug administration (FDA), because of its profound effect on amyloid plaques as shown by amyloid PET as a surrogate marker. Although this decision was not met with great enthusiasm uniformly, the general feeling during a 4 hour webinar hosted by the Alzheimer Association (1) was that the benefits seem to outweigh the risks, but proper guidance was needed on who would be eligible for treatment, because all attendees felt the label was far too broad and unspecific. The latter was amended on july 8, jointly by FDA and Biogen to more accurately reflect the eligible patient population that was studied in the Phase III program.
In this issue of JPAD, Cummings and colleagues report Appropriate Use guidelines constructed by an Expert Panel (2). Not surprisingly, The Expert Panel recommends that the appropriate use of aducanumab in real-world clinical practice should pragmatically mimic the use of aducanumab in the EMERGE and ENGAGE clinical trials that led the FDA to approve aducanumab. Importantly, not mentioned explicitly in the label, they stress that a patient who is considered for treatment with Aduhelm, should be amyloid positive, using any of the approved amyloid ligands, visually read by an experienced nuclear medicine physician, or by having abnormal CSF levels. We would like to specify this by stating that both CSF abeta 1-42 and p-tau 181 should be abnormal (or an abnormal ratio ptau/abeta), in agreement with the biological definition of AD (3). In order to lower the number of patients undergoing expensive PET scanning or invasive lumbar punction procedures, the availability of plasma tests to prescreen patients will be highly valuable (4). This is becoming a reality quite soon and will reduce costs for the healthcare system as well.
The Expert Panel stresses the patiënt-centered discussion around the decision to start treatment and assessing APOE status. The latter is important given the higher risk of ARIA in E4 carriers but even more because of a possible even higher treatment effect (5). This shared decision making around these important issues requires more time and effort from an experienced AD physician, most often a neurologist, than previously needed and thought (6).
The Expert panel makes important recommendations to monitor patients for ARIA using a similar scheme as was employed in the trials. It may be emphasized that ARIA is an MRI phenomenon that remained asymptomatic in 75% of the participants in the trials. It is conceivable that this is actually lower in real world practice, since in the trials patients were specifically asked for side effects. That said, the treating physician should be alert for side effects and should taper the treatment upon the appearance of ARIA, as per the criteria laid out by Cummings et al. In case of severe symptomatology, treatment with steroids is highly effective, but rarely needed.
One category of patients that were not included in the trials and not mentioned in the citeria, but who may appeal for treatment to their phsyicians is familial AD: patients carrying a PSEN1, 2 or APP mutation. When symptomatic and amyloid positive, they may be eligible for treatment and the same recommendations will apply to them.
The Expert Panel has given timely advice and it is hoped that all sites around the US will apply these recommendations in order to minimize regional differences, to optimize treatment and monitoring with an expensive drug in a vulnerable population of patients and it forms a solid basis for future treatment with other monoclonal antibodies for AD. When EMA approves aducanumab in the EU, these Appropriate Use guidelines will help formulating similar advice to EU clinicians, to secure the right patient is treated in the most optimal way.



2. Cummings J., Aisen P., Apostolova L.G. et al. Aducanumab: Appropriate Use Recommendations. J Prev Alz Dis; in press; DOI:
3. Jack CR Jr, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, Holtzman DM, Jagust W, Jessen F, Karlawish J, Liu E, Molinuevo JL, Montine T, Phelps C, Rankin KP, Rowe CC, Scheltens P, Siemers E, Snyder HM, Sperling R; Contributors. NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement. 2018 Apr;14(4):535-562.
4. Thijssen EH, Verberk IMW, Vanbrabant J, Koelewijn A, Heijst H, Scheltens P, van der Flier W, Vanderstichele H, Stoops E, Teunissen CE. Highly specific and ultrasensitive plasma test detects Abeta(1-42) and Abeta(1-40) in Alzheimer’s disease. Sci Rep. 2021 May 6;11(1):9736.
5. Briefing document Biogen and FDA, November 6, 2020
6. Fruijtier AD, Visser LNC, Bouwman FH, Lutz R, Schoonenboom N, Kalisvaart K, Hempenius L, Roks G, Boelaarts L, Claus JJ, Kleijer M, de Beer M, van der Flier WM, Smets EMA. What patients want to know, and what we actually tell them: The ABIDE project. Alzheimers Dement (N Y). 2020 Dec 16;6(1):e12113.



R. Doody*


Dr. Doody is a Vice President at Genentech, part of the Roche Group, and Franchise Head for Alzheimer’s Disease and Neurodegeneration in late-stage Neuroscience at F. Hoffman LaRoche. She is also Chair of the Alzheimer’s Association Research Roundtable. This editorial represents her personal view and not the position of any group with which she is associated.

Corresponding Author: Rachelle Doody, F. Hoffmann-La Roche Ltd, Basel, Switzerland,


The recent accelerated approval of Aducanumab has been the most mixed-consequence and historical decision that has taken place in the public health arena during my professional career. When I search for earlier comparisons of similar moment, the National Cancer Act of 1971 comes to mind, based upon what I have learned about it from Siddhartha Muckherjee’s “biography of cancer (1).” That, too, evidently involved passionate scientists with widely differing opinions, politics, and the loud voices of influential advocacy groups. There was a large outcry protesting the interference of politics and social concerns into the integrity of the scientific process. The approval of the National Cancer Act enabled many diverse efforts to treat a host of cancers, often treatments that would have been considered too high risk before the Act, resulting in some treatments that devastated patients, and some treatments and innovative approaches (e.g. combinations) that eventually worked.
More recently, in 1987, the approval of Azidothymidine (AZT) as a treatment for HIV/AIDS also shares similarities with the approval of Aducanumab (2). There was a global crisis due to increasing numbers of patients affected by the disease, an unclear scientific understanding of the cause of the disease, and no treatments other than drugs to treat secondary infections. The Food and Drug Administration had to decide about approval quickly, reviewing an application that was based upon a small trial with methodological flaws; a trial that was stopped early (in that case because of efficacy); and a reported treatment effect that many questioned as being not clinically meaningful. Further, the drug was associated with substantial side effects. Many scientists and clinicians denounced the regulatory process, accusing the agency of responding to social forces instead of upholding scientific standards. There were calls to investigate the Agency. Once on the market, the cost of the drug was prohibitive for many infected patients. Many observers have credited the approval of AZT with spurring additional research into the causes and treatments for HIV/AIDS, research which eventually led to dozens of approved drugs, largely based upon combination approaches that would not have been possible without so many mechanisms of action to choose from. Yet the value of AZT itself for the prevention and treatment of HIV/AIDS remains controversial even today.
In our own historic moment, what can we who are involved say about the consequences of the FDA approving Aducanumab? We cannot know the ultimate outcome, or the impact that this decision will have for patients, research and society as a whole in the years to come, but can only share our personal experiences, our worries, and our hopes. In that vein, I speak for myself, and not as a representative of any company or organization with which I am affiliated.
The approval of Aducanumab will likely help some patients to decline more slowly from their Alzheimer’s condition than they otherwise would have. This is a huge accomplishment for patients, clinicians, researchers and the army of dedicated company employees who worked for years to achieve this outcome. As far as we know today, the patients most likely to be helped are those similar to the patients included in the EMERGE trial. Yet the inconsistent outcomes in EMERGE compared to those in ENGAGE will temper the enthusiasm of some clinicians for even trying the drug. Often the persistence of a treatment is as much a function of the clinician as it is of the patient. If clinicians have difficulty discerning benefits to patients, which is often the case when an illness is expected to progress despite treatment, it will re-enforce their lack of enthusiasm for treatment.
The Aducanumab prescribing information does not limit the drug’s use to this population of individuals who are most likely to benefit. Biogen has added to the label, after the initial approval, that treatment should be initiated in patients with MCI or mild AD, but there is no guidance regarding when to stop treatment and the wide indication statement “for the treatment of AD,” remains. Some clinician researchers are attempting to help by making their own recommendations to guide clinical practice (3). Many commentators have reacted against the wide indication, thinking that it goes well beyond the evidence, and exposes patients who may be unable to benefit to unnecessary risks, and the lack of limitations increases the cost to payers. But the wide label also facilitates our thinking about AD as a series of neuropathological changes in the brain, reflected by biomarkers, as opposed to waiting for the clinical manifestations to identify the disease. This wider conception of AD could help to reverse some historical and artificial constraints on AD drug development. In most diseases, we do not have to develop drugs for patients one severity segment at a time. We do not, for example, have drugs that are developed and approved specifically for metabolic syndrome, or just for early diabetes, or moderate diabetes or severe diabetes. The time and resources lost by doing drug development in this step-by-step way are enormous. I have often thought that we were painting ourselves into a corner in AD research by studying AD drugs in one population segment at a time, beginning with the cholinesterase inhibitors (4-7). We did it, in part, because of the nihilism associated with the hypothesis that one could actually treat AD. Drug developers feared that wide labels would work against approval and reimbursement. The field also pursued this development strategy because of the lack of outcome measures that we were sure could capture a benefit across all of the disease stages. Maybe one consequence of the FDA approval of Aducanumab with a wide label is that we can start to think of AD in a more natural way, on a continuum, and change our expectations for therapies. Perhaps treatments should be expected to benefit more patients, even if the effect varies by stage or severity. Instead of limiting drugs by stage, our efforts to identify predictors of response can accelerate as a way to personalize the treatments and limit wasted resources.
The approval of Aducanumab will likely also cause hardship for some patients. As a practicing clinician in an academic neurology clinic for over 20 years, I accompanied many patients and their families on their journey with AD. Our practice and research setting emphasized follow up of the patient for the duration of their disease, sometimes from the time of detection in our Healthy Aging Program, through years of Mild Cognitive Impairment, to mild, moderate and severe stages of AD (8). Although we offered clinical trials to every eligible patient, the percentage who participated steadily went down over the years, once symptomatic treatments came on the market. Yet some patients participated in multiple clinical studies (with no treatments offered) and clinical trials because they thought that better disease understanding and treatments were needed and wanted to help make that happen. Other patients and their families looked for any and all potential treatments outside of clinical trials, regardless of the cost, and regardless of where they had to go to get the treatment. People in this latter group were not necessarily financially privileged. I saw families who sold their vehicles, or took out a second mortgage on their home, or crowd-sourced from family and friends just to try drugs and treatments that they could get outside of a clinical trial. Sometimes these were marketed drugs for which information in the media suggesting their use for AD, and so they sought off-label use. Or sometimes the drug was already in a clinical trial for AD, but available by prescription for other indications. These off-label uses of drugs for AD were not covered by insurance. Often, families could not raise enough money to continue the treatment for the hypothesized duration that was necessary for a treatment effect. Even knowing this, they would try it anyway for a few doses. So, it is likely that some families will risk serious economic insecurity by trying to keep up with co-pays for Aducanumab, or to pay for it out of pocket. Other patients will be denied access by healthcare systems who have to make difficult decisions about what drugs to cover or not to cover, and these patients and families who cannot get the treatment will suffer emotionally from thinking that there is a potentially effective treatment that they cannot obtain for themselves or their loved one. While the FDA approval of Aducanumab may help some patients, it may cause tremendous hardship for others.
What are the consequences of Aducanumab’s approval for research? Proper informed consent requires that patients who are currently in a clinical trial of an AD therapy, or patients considering entering one must be told about all available alternatives, including all approved therapies. Since trials usually involve double-blind periods where there is a chance that the patient will get placebo, some patients will opt for the newly approved drug by prescription instead and drop out of ongoing trials. This could render several years of their own and/ or of a company’s investment into an ongoing trial as worthless, if dropout rates are too high to draw conclusions about the treatment under study. If enough patients exit ongoing studies or fail to enroll in new ones that are placebo-controlled, which are still the gold standard for drug development, the development of additional new treatments will stall.
Many trials may be pressured to allow the new drug as a background treatment, despite the questions about efficacy, and to the detriment of the scientific questions being asked in the study. Since the clinical effects of Aducanumab have been variable across studies, it would be difficult when designing trials to predict the rate of change in placebo-treated patients on this background, affecting study power calculations and increasing the cost of trials. A requirement to allow background Aducanumab would circumvent the use of digital twins or other real world control groups, slowing down innovative approaches to drug development. Safety monitoring for trials of new drugs in which background treatment with Aducanumab is allowed would have to factor in drug interactions as well as adverse events attributable to Aducanumab. If sponsors want to reduce variability caused by having some patients on treatment and some not, they can require all participants to be on Aducanumab (an add-on study design as opposed to allowing the drug as background therapy), but the cost to the sponsor of supplying Aducanumab to participants could be cost-prohibitive. And because of the wide drug label, these considerations regarding background treatment could theoretically apply to all AD trials, from early AD to advanced AD. Finally, the wide label for Aducanumab, even after the qualifiying statement added by Biogen after the approval, still leaves the door open for the drug to be used by prescription for primary and secondary prevention. Trials in these populations are large and long in duration, so that commitments to provide Aducanumab to some or all of the participants in such prevention trials would be impractical. If there is widespread use of Aducanumab in preclinical or very early stages of AD outside of clinical trials, new treatments that might be ideally suited to these stages of the disease may never be developed. Already, the approval of Aducanumab has raised such challenges for ongoing platform studies, such as the Dominantly Inherited AD Network, as there are no data on the dose or safety of Aducanumab for this genetic population, but patients who progress in these trials may expect to receive the treatment.
In conclusion, the FDA approval of Aducanumab is a hand-wringing historical development. People will be helped and hurt, research will be advanced, slowed, and in some cases potentially made impossible. Public confidence in the regulatory process has once again been called into question, and the ability of most people to even understand the risks and benefits of the new drug has been compromised by the complex study results and contested regulatory decision. Change is inevitable, and there can be no progress in science without change. So each person involved in this present-day history-making event is free to decide what he/she would have done differently, or thinks of the drug approval, and most importantly, how they plan to conduct their AD clinical and research efforts, given the news. The hope is that, collectively, we will go on to do better at preventing and treating Alzheimer’s disease.



1. The Emperor of All Maladies, Siddhartha Muckerjee, Simon &Schuster, 2010.
2. The Story Behind the First AIDs Drug, Alice Park, Time Magazine, March 19, 2017.
3. Cummings J., Aisen P., Apostolova L.G. et al. Aducanumab: Appropriate Use Recommendations. J Prev Alz Dis; in press; DOI:
4. Rogers, S., Doody, R.S., Mohs, R., Friedhoff, L., and the Donepezil Study Group. Donepezil Improves Cognition and Global Function in Alzheimer’s Disease: A 15-week Double-Blind Placebo-Controlled Study. Archives of Internal Medicine, 1998; 158, 1021-1031
5. Rogers, S.L., Farlow, M.R., Doody, R.S., Mohs, R., Friedhoff, L.T. A 24-week, Double-Blind, Placebo-Controlled Trial of Donepezil in Patients with Alzheimer’s Disease. Neurology, 1998; 50, 136-145
6. Black S, Doody R, Li H, McRae T, Xu Y, Sun Y, Perdomo CA, Richardson S. Donepezil preserves cognition and global function in severe Alzheimer’s disease patients. Neurology 2007; 69:459-469.
7. Doody R, Ferris S, Salloway S, Sun Y, Goldman R, Xu Y, Murthy A. Donepezil Treatment of Patients with Mild Cognitive Impairment: a 48-week, Randomized, Placebo-Controlled trial. Neurology, 2009; 72(18):1555-1561.
8. Doody, R.S., Pavlik, V., Massman, P., Kenan, M., Yeh, S., Powell, S., Cooke, N., Dyer, C., Demirovic, J., Waring, S., Chan, W. Changing patient characteristics and survival experience in an Alzheimer’s disease center patient cohort. Dementia and Geriatric Cognitive Disorders, 2005;20:198-208



S. Gauthier1, P. Rosa-Neto2


1. Department of Neurology and Neurosurgery, McGill University, Montreal, Canada; 2. Departments of Neurology and Neurosurgery, McGill University, Canada.

Corresponding Author: Serge Gauthier, Emeritus Professor, Department of Neurology and Neurosurgery, McGill University, Montreal, Canada,


The authors of these recommendations must be congratulated on rapidly putting together a workable set of guidelines on the best use in clinical practice of the first drug approved (1) in a new class of medications acting on the excessive deposition of amyloid plaques in the brain of persons with early symptoms of Alzheimer’s disease (AD). These guidelines have been written despite the current lack of peer-reviewed publications about the Phase III pivotal studies so changes may be required once all the data is available in the public domain (2, 3, 4).
There has been a good effort at suggesting clinical instruments that are familiar to clinicians and less time consuming than rating scales used in randomized clinical trials (RCT). Nevertheless, the Clinical Dementia Rating (CDR) scale is a comprehensive and reliable instrument that should be considered for annual follow-up visits and help determine if there is sufficient clinical stability to justify treatment continuation for another year.
`Start rules` are clearly defined, based on what is known of the participant populations in the Phase III studies. There are few comments about persons younger than 50 in the early symptomatic stages of familial autosomal dominant AD or with Down’s syndrome, and these special populations require their own independent RCT to determine effective doses and safety profile, even with aducanumab. The exclusion of persons with `evidence of stroke` need more details since many patients with early symptomatic AD have asymptomatic lacunar infarcts in non-strategic brain regions.
There has been an effort at defining `stop rules`, including severe symptoms in the presence of ARIA, inability to reach therapeutic dose, loss of access to clinical and brain imaging monitoring, but the authors stopped short of stating that this treatment should be stopped when reaching a moderate stage of dementia. This will likely be a requirement from payers and the next set of use guidelines may operationally define moderate dementia such as CDR global score of 2, MMSE lower than 19 at least twice, having lost autonomy on key instrumental activities of daily living.
Optimal use of amyloid PET and CSF analysis is very well documented in these guidelines, and the authors hint at potential time and cost savings using ApoE genotype and plasma p-tau isoforms as a case-finding tool to identify patients that are amyloid positive. There is indeed realistic hope that a significant number of candidates for anti-amyloid drugs will not need amyloid PET or CSF analysis if they are ApoE4 homozygotes or have elevated plasma p-tau 181.
Future recommendations could address the need of confirmatory data on clinical efficacy, taking advantage of the FDA requirement for another placebo-controlled study: it is an opportunity to establish if anti-amyloid therapy can be stopped once the amyloid load has been rectified, through a repeat PET amyloid scan after 12 or 18 months, followed by randomization to continuation of aducanumab, to placebo, to an anti-tau drug or a combination of the two drug class. This randomized delayed start factorial design may go a long way in influencing future therapy of AD, as well as adaptive designs as described by Bateman et al, 2016 (5) .
Although written to answer the specific needs of clinical use of aducanumab right now in the USA, these guidelines will likely serve as blueprint for other drugs of that class. This is one big step for our field.



1. US Food & Drug Administration FDA grants accelerated approval for Alzheimer’s drug. Date: 2021 Date accessed: June 17, 2021
2. Cummings J, Aisen P, Lemere C, Atri A, Sabbagh M, Salloway S. Aducanumab produced a clinically meaningful benefit in association with amyloid lowering. Alzheimers Res Ther. 2021 May 10;13(1):98. doi: 10.1186/s13195-021-00838-z. PMID: 33971962; PMCID: PMC8111757.
3. Walsh S, Merrick R, Milne R, Brayne C. Aducanumab for Alzheimer’s disease? BMJ. 2021 Jul 5;374:n1682. doi: 10.1136/bmj.n1682. PMID: 34226181.
4. Alexander GC, Karlawish J. The Problem of Aducanumab for the Treatment of Alzheimer Disease. Ann Intern Med. 2021 Jun 17. doi: 10.7326/M21-2603. Epub ahead of print. PMID: 34138642.
5. Bateman RJ, Benzinger TL, Berry S, Clifford DB, Duggan C, Fagan AM et al. The DIAN-TU Next Generation Alzheimer prevention trial: adaptive design and disease progression model. Alzheimer`s & Dementia, 2017, 13(1) 8-19. DOI:10.1016/j.alz.2016.07.005



K. Sato1,2, T. Mano2, R. Ihara3, K. Suzuki4, Y. Niimi5, T. Toda2, T. Iwatsubo1, A. Iwata3, for Alzheimer’s disease Neuroimaging Initiative, Japanese Alzheimer’s disease Neuroimaging Initiative, and The A4 Study Team


1. Department of Neuropathology, Graduate School of Medicine, The University of Tokyo, Japan; 2. Department of Neurology, The University of Tokyo Hospital, japan; 3. Department of Neurology, Tokyo Metropolitan Geriatric Medical Center Hospital, Japan; 4. Division of Neurology, Internal Medicine, National Defense Medical College, Japan; 5. Unit for Early and Exploratory Clinical Development, The University of Tokyo Hospital, Japan

Corresponding Author: Dr. Atsushi Iwata, Department of Neurology, Tokyo Metropolitan Geriatric Medical Center Hospital, 35-2 Sakaecho Itabashi-ku, Tokyo 173-0015, Japan, Phone: 81-3-3964-1141, FAX: 81-3-3964-2963, , E-mails:

J Prev Alz Dis 2021;
Published online July 5, 2021,



Background: Models that can predict brain amyloid beta (Aβ) status more accurately have been desired to identify participants for clinical trials of preclinical Alzheimer’s disease (AD). However, potential heterogeneity between different cohorts and the limited cohort size have been the reasons preventing the development of reliable models applicable to the Asian population, including Japan.
Objectives: We aim to propose a novel approach to predict preclinical AD while overcoming these constraints, by building models specifically optimized for ADNI or for J-ADNI, based on the larger samples from A4 study data.
Design & Participants: This is a retrospective study including cognitive normal participants (CDR-global = 0) from A4 study, Alzheimer Disease Neuroimaging Initiative (ADNI), and Japanese-ADNI (J-ADNI) cohorts.
Measurements: The model is made up of age, sex, education years, history of AD, Clinical Dementia Rating-Sum of Boxes, Preclinical Alzheimer Cognitive Composite score, and APOE genotype, to predict the degree of amyloid accumulation in amyloid PET as Standardized Uptake Value ratio (SUVr). The model was at first built based on A4 data, and we can choose at which SUVr threshold configuration the A4-based model may achieve the best performance area under the curve (AUC) when applied to the random-split half ADNI or J-ADNI subset. We then evaluated whether the selected model may also achieve better performance in the remaining ADNI or J-ADNI subsets.
Result: When compared to the results without optimization, this procedure showed efficacy of AUC improvement of up to approximately 0.10 when applied to the models “without APOE;” the degree of AUC improvement was larger in the ADNI cohort than in the J-ADNI cohort.
Conclusions: The obtained AUC had improved mildly when compared to the AUC in case of literature-based predetermined SUVr threshold configuration. This means our procedure allowed us to predict preclinical AD among ADNI or J-ADNI second-half samples with slightly better predictive performance. Our optimizing method may be practically useful in the middle of the ongoing clinical study of preclinical AD, as a screening to further increase the prior probability of preclinical AD before amyloid testing.

Key words: Amyloid beta, preclinical Alzheimer’s disease, machine learning, predictive model.




Preclinical Alzheimer’s disease (AD), which corresponds to positive brain amyloid beta (Aβ) accumulation in healthy individuals without an evidence of cognitive decline (1-3), is getting focused as the target of clinical trials aiming to develop disease-modifying therapies for AD (4). Positive amyloid accumulation on amyloid positron emission tomography (PET) or lowered levels of Aβ42 in the cerebrospinal fluid (CSF) are used as the gold standard to include participants into clinical trials for preclinical AD (1).
It is estimated that approximately one-third of cognitive normal elderly individuals have positive Aβ (5), which means that if randomly selected, it is necessary to screen 3 times more clinically eligible participants by PET amyloid imaging or CSF lumbar puncture to determine if they are actually amyloid positive or not. Indeed, in the A4 study in which 1,000 participants were included to conduct a double-blinded randomized clinical trial of solanezumab versus a placebo (6), more than 10,000 clinically normal individuals were initially screened, and then the eligible 3,300 participants were further screened by PET amyloid imaging.
If we have some predictive index that can increase the prior probability for the positive Aβ accumulation, the above cost/labor-consuming screening processes could become more efficient with a smaller number of participants requiring PET screening (7, 8). For example, an earlier study reported predicting Aβ of cognitive normal participants from an Alzheimer’s disease Neuroimaging Initiative (ADNI) cohort (9) used demographic features of age, sex, education, APOE ε4 status, and cognitive scores, increasing the positive predictive value to 0.65 compared to the reference prevalence of 0.41 (7).
Meanwhile, in case of a Japanese cohort such as the Japanese Alzheimer’s disease Neuroimaging Initiative (J-ADNI) (10-12) cohort, there is a concern in deriving similar predictive models from this cohort due to the limited number of eligible cognitive normal participants. There are fewer than 100 participants included without lack of the necessary data in the J-ADNI (10, 12), so it is considered difficult to construct statistically robust models trained and validated within the Japanese cohort alone to date.
On the other hand, it might be also unsatisfying to apply the models derived from the external population out of the Japanese cohort directly, due to the potential heterogeneity of study participants among different cohorts. In other words, models derived from Anti-Amyloid treatment in Asymptomatic Alzheimer’s disease (A4), ADNI, or Australian Imaging Biomarkers and Lifestyle Study of Ageing (AIBL) cohort data (13) might not always be applicable to the J-ADNI cohort as they are, since the variable importance of each feature in the model can differ depending on the cohort, due to the difference in the distribution of participants’ basic demographics. For example, baseline age and education or even the proportion of those with positive Aβ are shown to be significantly different between ADNI and J-ADNI cohorts (10). These problems might have prevented the development of clinical models that effectively predict preclinical AD in a Japanese cohort.
As one of the solutions to overcome these constraints, here we propose to utilize models trained based on the A4 cohort data, which is a large dataset with more than 3,000 participants as of late 2019. Since the data characteristics of A4 participants and Japanese cohort (i.e. J-ADNI here) participants could somehow differ as we mentioned above, we optimized the A4-based models, thereby making the models more suitable to the J-ADNI cohort. Our proposing procedure is composed of two stages: the first is to generate numerous patterns of prediction models based on the A4 data with the varying standard uptake value ratio (SUVr) thresholds, and the second is to find the most appropriate SUVr threshold configuration among them so that the model based on the SUVr configuration would perform best in the randome-half of J-ADNI (or ADNI) dataset. The SUVr threshold is the critical cut-off to determine if there is amyloid accumulation in the PET or not (14) but is not always strictly established in the A4 study cohort, so adjusting the SUVr threshold leads to the varied allocation of amyloid positive/negative binary status in each case of the original A4 data. This is the operational procedure made solely for the purpose of identifying the best-performing models for other cohort data, and then we evaluate whether the obtained model based on the determined SUVr threshold can also take the better performance in the remaining J-ADNI (or ADNI) subset. Such ‘optimization’ procedure might allow us to build more flexible models, thereby enhancing the applicability of the obtained models to any external cohorts such as J-ADNI or ADNI. Practically, our proposed method might be useful as a predictive index available in the actual clinical study settings for preclinical AD, e.g., as a screening to increase the prior probability of preclinical AD just in the middle of ongoing preclinical AD studies.



Data acquisition and preprocessing

This study was approved by the University of Tokyo Graduate School of Medicine institutional ethics committee (ID: 11628-(3)). Informed consent is not required because this was observational study using publicly available data. We used the datasets of the A4 study and ADNI obtained from the Laboratory of Neuro Imaging (LONI) ( in October 2019 and the J-ADNI dataset obtained from National Bioscience Database Center (NBDC) ( in June 2018 with the approval of the data access committee. 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), PET, other biological markers, and clinical and neuropsychological assessments can be combined to measure the progression of mild cognitive impairment (MCI) and early AD. For up-to-date information, see
In this study, we used the data of cognitive normal participants. General inclusion criteria for the cognitive normal participants were determined in reference to an earlier study on the preclinical Alzheimer cognitive composite (PACC) (3), defined as follows: participants ages 65 to 85 years old (* 60 to 84 years old for cases from the J-ADNI cohort) at the time of screening with a global Clinical Dementia Rating (CDR-global) score of 0, with MMSE score (27-30) and Delayed Recall score on the Logical Memory IIa subtest (8-15) for participants with 13 or more years of education, or with MMSE (25-30) and Delayed Recall score (6-13) for participants with 12 years or less of education.
To determine Aβ accumulation status in A4 study cohort, with/without (binary) positive Aβ-PET (florbetapir) at a varying threshold level of Standardized Uptake Value ratios (SUVr) (value corresponding to the ‘Composite_Summary’ in the ‘A4_PETSUVR.csv’ file) was used (Supplemental Table 1). Meanwhile, in the ADNI data, due to the limited number of eligible participants with missing data, we used CSF Aβ42 < 192 pg/mL (values of median batch in the ‘UPENNBIOMK_MASTER.csv’ file) as the criterion for positive Aβ accumulation (15). In the J-ADNI cohort, cases with CSF Aβ42 < 333 pg/mL (values in the ‘pub_csf_apoe.tsv’ file) (10) or with positive findings on the visual assessment of PiB-PET results (as listed in the ‘pub_petqc.tsv’ file) (16) were determined as positive Aβ.
We used the following clinical and laboratory features, which are available commonly in A4, ADNI, and J-ADNI datasets, as exploratory variables to include into the models: age at baseline, sex (male or female: binary), education years, with/without parental history of AD (binary), with/without elevated Clinical Dementia Rating sum of boxes (CDR-SB) at baseline (≥0.5 or not: binary), with/without APOE ε4 allele(s) (binary), and the baseline PACC score. Other features such as brain MRI or blood test results as used in our previous studies (11, 17, 18) were not included because they are not always available from A4, ADNI, and J-ADNI cohorts in a unified manner. Since the A4 study dataset up to 2019 contains baseline data alone and the participants’ sequential changes have not been available, we also used the baseline data alone from the ADNI and J-ADNI datasets. The parental history of AD was regarded as positive if there was a statement that the participant’s father or mother had been diagnosed with AD, and it was regarded as negative if there was no such statement or the data were missing.
The PACC score (3) is the composite score, which is calculated from the sum of Z scores from 4 items: (1) the Total Recall score from the Free and Cued Selective Reminding Test (FCSRT), (2) Delayed Recall score on the Logical Memory IIa subtest from the Wechsler Memory Scale, (3) Digit Symbol Substitution Test score from the Wechsler Adult Intelligence Scale-Revised, and (4) MMSE total score [3]. Since the PACC score was not calculated in the ADNI and J-ADNI studies, we calculated the virtual PACC score by using the score of “LDELTOTAL: for Logical Delayed, the score of ”DIGITSCOR” for the Digit Symbol Substitution Test, and the total MMSE score. Furthermore, instead of using the FCSRT test score, which was not conducted in ADNI and J-ADNI studies, we used the delayed recall scores of ADAS-cog13 (Q4) in ADNI and J-ADNI datasets as in the earlier study (3). The Z scores of each of the 4 items were calculated within each ADNI or J-ADNI cohort in reference to the data of the cognitive normal cohort as allocated at baseline (“DX_bl” of “CN” (cognitive normal) or “SMC” (subjective memory complaints) in ADNI and the “COHORT” of “NL”: (normal) in J-ADNI.
Missing data were handled by using the list-wise method: samples with missing data in the above modeling features were excluded from the analysis. Eventually, we included n = 3233 unique eligible cases of the A4 study cohort, n = 86 eligible cases of the ADNI cohort, and n = 50 eligible cases of the J-ADNI cohort.

Concepts of our proposed method

Here we explain how to demonstrate the practical effectiveness of our proposed ideas. First, we built a large number of models based on the varying SUVr configurations (Figures 1A, 1B) to predict positive Aβ within the A4 cohort data. Then, we evaluated the performance of these models, as calculated by area under the curve (AUC) as a performance metric of binary prediction models available regardless of the threshold value, in each of the half-split subgroups of the external cohort from A4 comprised of cognitive normal participants (Figure 1C, 1D). Suppose we know the Aβ status of each case in subgroup1 (Figure 1C), while we do not know the status of each case in subgroup2 (Figure 1D). When we compare the distribution of predictive performance results across all SUVr configurations (from 1 to k here) between the subgroups (Figure 1E), the true correlation should fall into the significantly negative (Figure 1F), non-significant (Figure 1G), or significantly positive (Figure 1H) categories. If we can observe that the actual correlation is consistently positive using the various datasets for evaluation (e.g. ADNI and J-ADNI here), to find which model’s SUVr configuration achieves the highest performance in one subgroup, this would also result in the near-highest performance in the rest another subgroup with unknown Aβ status (Figure 2A & 2B). We call the procedure to find the SUVr configuration with the highest AUC in one subgroup the ‘optimization of models’.
The significant-positive correlation (Figure 1F) is the prerequisite for this optimization. Although the half-split subgroups derived from the same cohort might tend to have a positive correlation due to the similar variance in their participants’ demographical data, such a tendency is not always validated, especially in cohorts which are far smaller (e.g. ADNI or J-ADNI cognitive normal cases) than the A4 cohort. If the correlation between subgroups occasionally becomes negative (Figure 1F) or non-significant (Figure 1G), the optimization will not work. Therefore, our goal in this study was to confirm that the correlation between the half-split validation subgroups (Figures 1C versus 1D) is reproducibly significantly-positive (Figure 1F-H) and then to assess the degree of AUC improvement by employing this optimization procedure (Figures 2A & 2B), using the ADNI and J-ADNI datasets as validation.

Figure 1. Conceptual outline of our proposed method

We at first built a large number of models based on varying SUVr configurations (A, B), then we evaluated the performance of these models in each of the half-split external cohort (= ADNI or J-ADNI here) subgroups of cognitive normal participants (C, D). We supposed that we know the Aβ status of each case in subgroup1, while we do not know the statuses in subgroup2. When we compare the predictive performance results’ distribution across different SUVr configurations (from 1 to k) between the external cohort half-split subgroups (e.g. ADNI or J-ADNI), the actual correlation should fall into the either significantly negative (F), non-significant (G), or significantly positive (H).


Processing workflow: model training and performance evaluation

A detailed data processing workflow is outlined in Supplemental Figure 1. The target of A4-cohort predictive models is whether they are with/without positive Aβ-PET (florbetapir) (binary) which are determined at varying SUVr threshold levels. In the model training, the SUVr threshold continuously varied by 0.01 from 0.99 to 1.47, corresponding to the [mean – 0.5 SD] and the [mean + 2 SD] of SUVr distribution in all the A4 data. Furthermore, we excluded the Aβ-negative cases with an SUVr barely lower than the threshold, between which the margin range is varied, in order to exclude possible false-negative cases. This exclusion procedure substantially also acts to exclude possible false-positive cases, clarifying the difference between cases with and without positive Aβ. Simultaneously adjusting with the above SUVr threshold, the “exclusion range” (Supplemental Figure 1A) is also adjusted continuously by 0.01 from 0 to 0.09, where 0.09 corresponds to [0.5 SD] of A4-SUVr. Taken together, cases whose SUVr is higher than the [threshold value] are defined as Aβ-positive, and the cases whose SUVr is lower than the [threshold value – exclusion range value] are defined as Aβ-negative (Supplemental Figure 1A). We here define this way of varying Aβ allocation and the eligible case inclusion as “SUVr configuration,” which is used to generate a large number of models (Figure 1B). This SUVr configuration can be changed into 48 SUVr threshold patterns *10 exclusion range patterns = 480 combination patterns in total.
Since the small proportion of cases within the exclusion range is eliminated, the eligible A4 dataset A_k, which is from the A4 cohort cases (n = 3233), is slightly different depending on each SUVr configuration k (k=1,2,…480) (Supplemental Figure 1B). Then a randomly selected 70% of A_k were further picked up as the A4 training subgroup A’k; using this A’k subgroup, we trained a model M_k predictive for positive Aβ (Supplemental Figure 1C). For the model Mk, we separately constructed 2 types of models, one of them including APOE ε4 status into its features (denoted as “model with APOE”), and another not including APOE ε4 status into the model (“model without APOE”) (Supplemental Figure 1C). This is because APOE ε4 is one of the strongest determinants of the CSF Aβ42 level (19), while a model without APOE ε4 status would be more convenient to use as a screening index. The training was conducted with 10-fold cross validation and by a penalized generalized linear regression (GLM) algorithm using R package “caret” (20). Automated optimization of penalized GLM hyperparameters was conducted with grid-search by the caret function.
Then the predictive performance of the model M_k was validated in the ADNI and J-ADNI cohort data, out of the original A4. We split the ADNI and J-ADNI cohorts into half subgroups (“subgroup1” & “subgroup2”) randomly (Supplemental Figure 1G, in Figures 1C & 1D) while retaining equal proportions of Aβ positive between the half subgroups using the “caret” package function (“createDataPartition”), then we aimed to compare the performance between ADNI subgroups or between J-ADNI subgroups. The predictive performance was measured with the metric of area under the curve (AUC), which is calculated by the predicted probability for the positive Aβ of each case in the applied dataset (Supplemental Figure 1D).
Since the randomly sampled A’k subgroup yields a slightly different model (Supplemental Figure 1C) every sampling time, we repeated the above processing steps (B-D: circled with gray color) 5 times in each k (shown with dagger mark [†]). We named the median from 5 times of AUC results as the vXi,k (Supplemental Figure 1E), which means it is derived from the k-th configuration-based model Mk applied to the subgroup Xi.
As the configuration can vary for 480 types as described above, the full validation results (k=1, 2, …480) for one subgroup are represented by a vector with a length of 480. For example, when one cohort X (= ADNI or J-ADNI) data are split into subgroup X1 and subgroup X2, vectors representing the results for these subgroups, which correspond to the result list of Figures 1C and 1D, are described as follows (Supplemental Figure 1F):


Then we measured the correlation between V_ADNI1 and VADNI2, and between VJADNI1 and VJADNI2.
The above process (steps A-F) was repeatedly performed for each ADNI and J-ADNI half-split subgroup (Supplemental Figure 1G), which are randomly separated 30 times in total (shown with the asterisk [*]), eventually yielding 30 sets of [VADNI1,VADNI2,VJADNI1,VJADNI2].
Next, we again explain how the ‘optimization’ is conducted using Figure 1 & 2, the example scatter plot of VX1 (plotted on X-axis) versus VX2 (plotted on Y-axis) across all 480 patterns of SUVr configurations in one randomization time (*). On this plot, the Pearson correlation between the VX1 and VX2 was R = 0.967 (p < 0.001). When we choose the ka of which vX1,ka takes maximum among the VX1, the performance AUC with the same ka-th SUVr configuration (= vX2,ka) would also be approximately the highest among VX2. In other words, based on the assumption that the correlation between the vectors VX1 versus VX2 is significantly-positive (as in Figure 1H), we can optimize the predictive model in reference to the half subgroup X1 alone so that the model takes the most of the best performance both in X1 and the rest from the half subgroup X2 of which performance distribution is unknown to us, by choosing the k of which vX1,k is the highest among VX1.
And when we choose the kb of which vX1,kb takes minimum among the VX1, the performance AUC with the same kb-th SUVr configuration (= vX2,kb) would also be approximately the lowest among VADNI2: the difference between the vX2,ka and vX2,kb just corresponds to the theoretically-maximum AUC improvement expected to be achievable by the present “optimization” procedure (Figure 2A).
Furthermore, we compared the optimized result and the non- optimized result based on the conventionally-used SUVr configuration (e.g., threshold of 1.15 (21)). Supposing an i-th SUVr configuration with a threshold of 1.15 and exclusion range of 0, we measured the difference between the above vX2,ka and the resulting AUC vX2,i of the i-th configuration in subgroup2. This difference just corresponds to the AUC improvement expected to be achieved by using this optimization procedure [Figure 2B], compared to the conventional settings when not using “optimization” as in earlier studies.

Figure 2. Evaluation of performance improvement

If we could observe that the actual correlation is consistently positive using the several datasets for evaluation (e.g. ADNI and J-ADNI here) as in the Figure 1H, the “optimization” to take the SUVr configuration of the model achieving the highest performance in one subgroup would also result in the near-highest performance in the rest of another subgroup with unknown Aβ status (A, B).


Statistical analysis

All data handling and statistical analysis were performed using the software R 3.5.1 (R Foundation for Statistical Computing, Vienna, Austria). For numerical data, we used median and interquartile ranges (IQR) for summarization and the Wilcoxon rank sum test or analysis of variance (ANOVA) test for comparisons between groups. For categorical data, we used frequency and percentage for summarizing and Fisher’s exact test for the group comparison. For calculating correlations between two numerical vectors, we used Pearson’s correlation. A P-value less than 0.05 was regarded as statistically significant if not mentioned otherwise.



Overview of the demographical distribution of the included cohorts

Basic demographics are shown in Supplemental Table 1, revealing slight differences among the data of the 3 included cohorts (A4, ADNI, and J-ADNI). The J-ADNI cohort participants had a significantly younger median age, were more predominantly male, and had fewer years of education than the other 2 cohorts. There was no significant difference among the 3 cohorts in the distribution of CDR-SB, parental history of AD, APOE ε4 status, and the baseline PACC.
In addition, we also evaluated the performance of each single feature for predicting positive Aβ in each of the included cohorts. A heatmap of AUC result values as the predictive performance of the corresponding features (in columns) in each corresponding cohort (in rows) is shown in Supplemental Figure 2A. For the A4 cohort on this heatmap, a SUVr threshold of 1.15 was used (21). Each of the features except for APOE has a different level of association with the positive Aβ status, depending on the cohort.

Cohort-specific “optimization» of models

Next, we obtained the predictive performance of models based on the varying SUVr configuration evaluated with AUC in the ADNI and J-ADNI subgroups. We visualized the examples of the result vectors VADNI1, VADNI2, VJADNI1, and VJADNI2, summarizing the AUC from 480 different SUVr configurations (48 types of SUVr thresholds × 10 types of exclusion ranges: Supplemental Figure 1A) by converting them into heatmap matrices for clarify (Supplemental Figure 2B). Each cell in the heatmaps represent the performance AUC value of the model based on the corresponding SUVr configuration, where the row denotes the SUVr threshold and the column denotes the exclusion range. For the results from ADNI (Supplemental Figure 2B, left) or J-ADNI (Supplemental Figure 2B, right) data, we can see that the AUC performance results distribute differently depending on the SUVr configuration, and that the AUC performance results distribute differently largely depending on the cohort.
By choosing the darkest cell in the heatmap from the ADNI-subgroup1 (Supplemental Figure 2B), we can select the SUVr for each model’s performance as that which is the highest in the ADNI subgroup1. As there was a positive correlation of R = 0.767 (p < 0.001) between the AUC heatmap of the ADNI subgroup1 and ADNI subgroup2 (Supplemental Figure 2B, left), the selected SUVr configuration would also take the near-best performance when applied to the rest of ADNI subgroup2. The same is true to the pair of J-ADNI subgroups (Supplemental Figure 2, right), between which there was a positive correlation of R = 0.493 (p < 0.001). This “optimization” procedure is generally cohort-specific since each cohort has specific spatial distribution of the resulting AUC heatmaps. Conversely, by choosing the lightest cell in the heatmap from ADNI subgroup1, the selected SUVr configuration would also show the near-lowest performance when applied to the rest of the ADNI subgroup2. The difference between the near-highest and the near-lowest AUC within subgroup2 corresponds to the “expected maximum AUC improvement achievable by optimization,” the difference between the worst AUC when the “optimization” was not used, and the best AUC when the “optimization” was used.
Now the set of [VADNI1,VADNI2,VJADNI1,VJADNI2] (as in Supplemental Figure 2B) is repeatedly obtained for 30 times of ADNI and J-ADNI randomization (Supplemental Figure 1G, [*]), at first based on the model “with APOE”: Figure 3A shows the distribution of the obtained correlation coefficients (as in Figure 1F-H) between VADNI1 and VADNI2 (summarized as Figure 3A, [a] & [b]) or between VJADNI1 and VJADNI2 (summarized as Figure 3A, [c] & [d]), repeated 30 times in total. In the ADNI cohort of models “with APOE” (Figure 3A [a]), Pearson’s correlation coefficient between the ADNI subgroup1 and subgroup2 was a mean of 0.897 (the mean’s 95% CI: 0.877 – 0.917), and the correlation coefficient > 0 and p-value < 0.05 were simultaneously observed in 30/30 of randomization (*) trials, fully meeting the prerequisite of our “optimization” method. The expected maximum AUC improvement width was a mean of 0.077 (the mean’s 95% CI: 0.069 – 0.085) (Figure 3B [a]), and the expected AUC improvement when compared to the AUC in a model of SUVr threshold 1.15 was a mean of 0.033 (95% CI: 0.022 – 0.043) (Figure 3C [a]), e.g. AUC value improved from 0.724 to 0.774 in a representative case. Similarly, in the ADNI cohort by models “without APOE” (Figure 3A [b]), the correlation coefficient was a mean of 0.517 (the mean’s 95% CI: 0.444 – 0.582), and the correlation coefficient > 0 and p-value < 0.05 were simultaneously observed in 30/30 of randomization trials (*). The expected maximum AUC improvement width was a mean of 0.107 (the mean’s 95% CI: 0.086 – 0.129) (Figure 3B [b]), and the expected AUC improvement when compared to the AUC in a model of SUVr with a threshold of 1.15 was a mean of 0.075 (95% CI: 0.057 – 0.093) (Figure 3C [b]), e.g. AUC value improved from 0.61 to 0.69 in a representative case. In comparison, the expected maximum AUC improvement achievable by the “optimization” was greater in models “without APOE” than in models “with APOE” (Figure 3B) in ADNI (Figure 3B [a] versus [b], and Figure 3C [a] versus [b]).

Figure 3. Correlation coefficients between the resultant AUC vectors from half-split subgroups and the degree of AUC improvement

Box plots show the distribution of the obtained correlation coefficients (as in Figure 1F-H) between V_ADNI1 versus V_ADNI2 (A, [a] & [b]), or between V_JADNI1 and V_JADNI2 (A, [c] & [d]), repeated 30 times in total. Each box corresponds to the range between the lower and upper quartiles (Q1 and Q3, respectively), and the range between whiskers corresponds to the data distribution within the range of [Q1 – 1.5*IQR, Q3 + 1.5*IQR]. In the ADNI cohort (A, [a] & [b]), 30/30 of results both with models “with APOE” (A, [a]) or “without APOE” (A, [b]) showed a significantly positive correlation between V_ADNI1 versus V_ADNI2. In the J-ADNI cohort, 22/30 results of models “with APOE” (A, [c]) were significantly positive, and 26/30 results of models “without APOE” (A, [d]) were significantly positive. The expected maximum AUC improvement achievable by the “optimization” (B), and the expected AUC improvement achievable by “optimization” when compared to the model based on the SUVr threshold of 1.15 without optimization (C) are plotted. In all models ([a]-[d]), the mean of “expected AUC improvement” was significantly higher than 0 (i.e. its lower 95% CI > 0), and a model “without APOE” in the ADNI cohort had approximately 0.10 of AUC improvement.


In the J-ADNI cohort models “with APOE” (Figure 3A [c]), the correlation coefficient between J-ADNI subgroup1 and subgroup2 was a mean of 0.301 (the mean’s 95% CI: 0.107 – 0.495), and a significant and positive correlation was observed in 22/30 randomization (*) trials, showing occasionally unsuccessful “optimization.” The expected maximum AUC improvement width was a mean of 0.011 (the mean’s 95% CI: 0.001 – 0.020) (Figure 3B [c]), and the expected AUC improvement when compared to the AUC in a model of SUVr with a threshold 1.15 was a mean of 0.009 (95% CI: 0.003 – 0.016) (Figure 3C [c]), e.g. AUC value showed few improvement from 0.65 to 0.65 in a representative case. Furthermore, in the J-ADNI cohort models “without APOE” (Figure 3A [d]), the correlation coefficient was a mean of 0.353 (95% CI: 0.258 – 0.448), and a significant and positive correlation was observed in 26/30 randomization trials (*), mostly meeting the “optimization” prerequisite. The expected maximum AUC improvement width was a mean of 0.086 (95% CI: 0.060 – 0.113) (Figure 3B [d]), and the expected AUC improvement when compared to the AUC in a model of the SUVr threshold of 1.15 was a mean of 0.019 (95% CI: 0.007 – 0.030) (Figure 3C [d]), e.g. AUC value improved from 0.61 to 0.64 in a representative case. The models “without APOE” showed a higher expected maximum AUC improvement achievable with the “optimization” than the models “with APOE” (Figure 3B [c] versus [d], and Figure 3C [c] versus [d]).



In this retrospective study, we demonstrated our attempts to optimize the A4 study-derived predictive models to be applicable to external cohort datasets, including ADNI and J-ADNI. The proposed method has novelty in that we operationally manipulated the positive Aβ allocation in the original training data of A4, thereby enabling the achievement of the best-performing model when applied to the external cohorts, including ADNI and J-ADNI. The obtained AUC had improved mildly when compared to the AUC in case of literature-based predetermined SUVr threshold configuration. This means our ‘optimization’ procedure allowed us to obtain preclinical AD models for ADNI or J-ADNI with slightly better predictive performance. Our method may be practically useful in the middle of ongoing clinical study of preclinical AD, as a screening to further increase the prior probability of preclinical AD among the remaining samples before their amyloid testing.
The motivation of this study was mainly based on the concern as to the direct application of the A4 study-derived models to J-ADNI cohort, due to the differences in the distribution of participants’ baseline demographics such as age, sex, education years, ethnicity, the proportion of positive Aβ (Supplemental Table 1), or any unexamined clinical, laboratory, or genetic factors. It is known that such differences in the probability distributions of each feature between the training and validation datasets lead to failures in accurate prediction. “Transfer learning” is used in the field of deep learning as one of its solutions, enabling us to apply the trained model to the dataset origin of other domains. Thus, if utilized in our settings, it would enable us to apply the dataset from a different regional population with the smaller sample size (22, 23). However, our approach is based on conventional machine learning and is different from ‘transfer learning’, which we have not used since even the Aβ status in the original training data (= A4 study cohort) has not been definitely determined yet. If the biologically-corroborated criteria for the Aβ status are established within the original A4 cohort, transfer learning would be employable for building models effectively applicable to ADNI or J-ADNI datasets.
As expected, the efficacy of “optimization,” which is measured by the degree of AUC improvement compared to the resulting AUC of not using the “optimization,” was higher than 0 in average. The degree of maximum improvement in AUC (Figure 3B) and the degree of AUC improvement compared to the SUVr threshold of 1.15 (Figure 3C) are both approximately 0.10 in models “without APOE” applied to the ADNI cohort (Figure 3B[b], 3C[b]), which means this optimization procedure is expected when applied to the models “without APOE.” Although showing a smaller improvement, the models “without APOE” applied to the J-ADNI cohort also had a higher AUC than in the case of the models with any SUVr configuration (Figure 3B[d]) or with a conventional SUVr threshold 1.15 (Figure 3C[d]). This difference between ADNI and J-ADNI in their degree of AUC improvement may be due to the difference in their size of samples or in the degree of inter-cohort variation as represented by the different amyloid positivity rate.
Generally, the degree of AUC improvement (Figure 3B, 3C) tended to be higher in models “without APOE’”([b], [d]) than in models “with APOE” ([a], [c]), which means the performance is expected to improve by optimization much larger models “without APOE” than models “with APOE,” probably reflecting the high importance of APOE ε4 status as a variable for predicting positive Aβ. In addition, when the model “with APOE” was used, only 22/30 of a randomized half-split of the J-ADNI dataset led to a significantly positive correlation between VJADNI1 and VJADNI2, while it was more frequent (26/30) when the model “without APOE” was used. These results suggest that the current optimization methods are more reliably and effectively used in models not including APOE ε4 status as features than those including it.
The current approach to adjust SUVr configuration consisted of the SUVr threshold and the exclusion of cases whose SUVr is barely lower than the threshold, is no more than an operational procedure here and is not biologically-validated in a strict sense. In this point, we need to be careful in the interpretation on the obtained final model or its variables’ importance that it is the “transferred” model and does not have certain biological basis on its own. For example, when we identified one feature (e.g. higher PACC) with high variable importance in the final model, the potential biological association between that feature and the Aβ positivity may be smaller than in the case of conventional non-transferred models.
Our study has some limitations. First, while the degree/frequency of positive correlation between the result vectors (Figure 3A) might be influenced by the size of the validating cohort datasets or their intra-cohort data variability, as suggested by our results where the efficacy of “optimization” showed smaller improvement and lower reliability when applied to the J-ADNI cohort than to the ADNI cohort, we have not examined the detailed conditions (e.g. sample size) required for the validation of datasets to be eligible for the “optimization” procedure. Further validation may be needed in other external cohorts with various kinds of sample sizes. Second, in the case of the single multi-center clinical trial to which we attempt to apply our method practically, there may be uncertainty whether the two subgroups collected from different facilities truly have a similar distribution in their demographical features, which is the pre-requisite for the external application of the current methods. Also, the extent to which the difference in inter-subgroup feature distribution can be allowed may be uncertain, and the sample size required to alleviate the potentially underlying variance between subgroups may also remain uncertain. Third, the proposed method manipulates the original training data distribution so as to be specifically best-performing in the validation cohort of interest, so the final model is not reversely applicable to the original A4 cohort data or to other cohorts with different demographical distributions. The fourth limitation is related to the PACC calculation in ADNI and J-ADNI: the validity of using ADAS-cog 13 (Q4) as a substitution of FCSRT, and the validity of setting ‘”NL” cohort data as a reference of PACC calculation. And the fifth is that the proposed method takes a certain amount of computational times, since model training and validation are repeatedly needed: 30 times of ADNI or J-ADNI splits for each [5 times of A4 training subgroup splits and model validations for each k (480 patterns in total)], eventually requiring us to calculate 30*5*480 = 72,000 times of model training and validation. This is actually one of the reasons why we used penalized GLM as the prediction algorithm here, which takes shorter computational time than other types of algorithms such as random forest or support vector machine, and it is designed to have a smaller risk of over-fitting to the training data. If possible, other algorithms should also be tried (24). And lastly, used 3 cohorts referred to different modality of amyloid tests (i.e., florbetapir-PET in A4, CSF in ADNI, and CSF and PiB-PET in J-ADNI), possibly lowering the applicability of our method.
To conclude, we proposed a novel method to obtain preclinical Aβ predictive models specifically optimized to the cohort of interest in order to achieve extrapolative application out of the original training data. This optimization procedure showed efficacy of up to 0.10 of AUC improvement when used in combination with the models “without APOE.” Our method may be practically useful in the mid of the actual clinical study of preclinical AD, as a screening to further increase the prior probability of preclinical AD before amyloid testing.


Funding: This study was supported by Japan Agency for Medical Research and Development grants JP21dk0207057, JP21dk0207048, and JP20dk0207028.

Description about the ADNI: 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.

Description about the A4 study: The A4 Study is a secondary prevention trial in preclinical Alzheimer’s disease, aiming to slow cognitive decline associated with brain amyloid accumulation in clinically normal older individuals. The A4 Study is funded by a public-private-philanthropic partnership, including funding from the National Institutes of Health-National Institute on Aging, Eli Lilly and Company, Alzheimer’s Association, Accelerating Medicines Partnership, GHR Foundation, an anonymous foundation and additional private donors, with in-kind support from Avid and Cogstate. The companion observational Longitudinal Evaluation of Amyloid Risk and Neurodegeneration (LEARN) Study is funded by the Alzheimer’s Association and GHR Foundation. The A4 and LEARN Studies are led by Dr. Reisa Sperling at Brigham and Women’s Hospital, Harvard Medical School and Dr. Paul Aisen at the Alzheimer’s Therapeutic Research Institute (ATRI), University of Southern California. The A4 and LEARN Studies are coordinated by ATRI at the University of Southern California, and the data are made available through the Laboratory for Neuro Imaging at the University of Southern California. The participants screening for the A4 Study provided permission to share their de-identified data in order to advance the quest to find a successful treatment for Alzheimer’s disease. We would like to acknowledge the dedication of all the participants, the site personnel, and all of the partnership team members who continue to make the A4 and LEARN Studies possible. The complete A4 Study Team list is available on:

Conflicts of interest: The authors have no conflict of interest to disclose.

Ethical standards: This study was approved by the University of Tokyo Graduate School of Medicine institutional ethics committee (ID: 11628-(3)).





1. Sperling RA, Aisen PS, Beckett LA, Bennett DA, Craft S, Fagan AM, Iwatsubo T, Jack CR Jr, Kaye J, Montine TJ, Park DC, Reiman EM, Rowe CC, Siemers E, Stern Y, Yaffe K, Carrillo MC, Thies B, Morrison-Bogorad M, Wagster MV, Phelps CH. Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 2011 May;7(3):280-92.
2. Jack CR Jr, Knopman DS, Jagust WJ, Petersen RC, Weiner MW, Aisen PS, Shaw LM, Vemuri P, Wiste HJ, Weigand SD, Lesnick TG, Pankratz VS, Donohue MC, Trojanowski JQ. Tracking pathophysiological processes in Alzheimer’s disease: an updated hypothetical model of dynamic biomarkers. Lancet Neurol. 2013 Feb;12(2):207-16.
3. Donohue MC, Sperling RA, Salmon DP, Rentz DM, Raman R, Thomas RG, Weiner M, Aisen PS; Australian Imaging, Biomarkers, and Lifestyle Flagship Study of Ageing; Alzheimer’s Disease Neuroimaging Initiative; Alzheimer’s Disease Cooperative Study. The preclinical Alzheimer cognitive composite: measuring amyloid-related decline. JAMA Neurol. 2014 Aug;71(8):961-70.
4. Cummings J. The National Institute on Aging-Alzheimer’s Association Framework on Alzheimer’s disease: Application to clinical trials. Alzheimers Dement. 2019 Jan;15(1):172-178.
5. Jansen WJ, Ossenkoppele R, Knol DL, Tijms BM, Scheltens P, Verhey FR, Visser PJ; Amyloid Biomarker Study Group, Aalten P, Aarsland D, Alcolea D, Alexander M, Almdahl IS, Arnold SE, Baldeiras I, Barthel H, van Berckel BN, Bibeau K, Blennow K, Brooks DJ, van Buchem MA, Camus V, Cavedo E, Chen K, Chetelat G, Cohen AD, Drzezga A, Engelborghs S, Fagan AM, Fladby T, Fleisher AS, van der Flier WM, Ford L, Förster S, Fortea J, Foskett N, Frederiksen KS, Freund-Levi Y, Frisoni GB, Froelich L, Gabryelewicz T, Gill KD, Gkatzima O, Gómez-Tortosa E, Gordon MF, Grimmer T, Hampel H, Hausner L, Hellwig S, Herukka SK, Hildebrandt H, Ishihara L, Ivanoiu A, Jagust WJ, Johannsen P, Kandimalla R, Kapaki E, Klimkowicz-Mrowiec A, Klunk WE, Köhler S, Koglin N, Kornhuber J, Kramberger MG, Van Laere K, Landau SM, Lee DY, de Leon M, Lisetti V, Lleó A, Madsen K, Maier W, Marcusson J, Mattsson N, de Mendonça A, Meulenbroek O, Meyer PT, Mintun MA, Mok V, Molinuevo JL, Møllergård HM, Morris JC, Mroczko B, Van der Mussele S, Na DL, Newberg A, Nordberg A, Nordlund A, Novak GP, Paraskevas GP, Parnetti L, Perera G, Peters O, Popp J, Prabhakar S, Rabinovici GD, Ramakers IH, Rami L, Resende de Oliveira C, Rinne JO, Rodrigue KM, Rodríguez-Rodríguez E, Roe CM, Rot U, Rowe CC, Rüther E, Sabri O, Sanchez-Juan P, Santana I, Sarazin M, Schröder J, Schütte C, Seo SW, Soetewey F, Soininen H, Spiru L, Struyfs H, Teunissen CE, Tsolaki M, Vandenberghe R, Verbeek MM, Villemagne VL, Vos SJ, van Waalwijk van Doorn LJ, Waldemar G, Wallin A, Wallin ÅK, Wiltfang J, Wolk DA, Zboch M, Zetterberg H. Prevalence of cerebral amyloid pathology in persons without dementia: a meta-analysis. JAMA. 2015 May 19;313(19):1924-38.
6. Sperling RA, Rentz DM, Johnson KA, Karlawish J, Donohue M, Salmon DP, Aisen P. The A4 study: stopping AD before symptoms begin? Sci Transl Med. 2014 Mar 19;6(228):228fs13.
7. Insel PS, Palmqvist S, Mackin RS, Nosheny RL, Hansson O, Weiner MW, Mattsson N. Assessing risk for preclinical β-amyloid pathology with APOE, cognitive, and demographic information. Alzheimers Dement (Amst). 2016 Aug 3;4:76-84.
8. Ansart M, Epelbaum S, Gagliardi G, Colliot O, Dormont D, Dubois B, Hampel H, Durrleman S; Alzheimer’s Disease Neuroimaging Initiative* and the INSIGHT-preAD study. Reduction of recruitment costs in preclinical AD trials: validation of automatic pre-screening algorithm for brain amyloidosis. Stat Methods Med Res. 2020 Jan;29(1):151-164.
9. Petersen RC, Aisen PS, Beckett LA, Donohue MC, Gamst AC, Harvey DJ, Jack CR Jr, Jagust WJ, Shaw LM, Toga AW, Trojanowski JQ, Weiner MW. Alzheimer’s Disease Neuroimaging Initiative (ADNI): clinical characterization. Neurology. 2010 Jan 19;74(3):201-9.
10. Iwatsubo T, Iwata A, Suzuki K, Ihara R, Arai H, Ishii K, Senda M, Ito K, Ikeuchi T, Kuwano R, Matsuda H; Japanese Alzheimer’s Disease Neuroimaging Initiative, Sun CK, Beckett LA, Petersen RC, Weiner MW, Aisen PS, Donohue MC; Alzheimer’s Disease Neuroimaging Initiative. Japanese and North American Alzheimer’s Disease Neuroimaging Initiative studies: Harmonization for international trials. Alzheimers Dement. 2018 Aug;14(8):1077-1087.
11. Iwata A, Iwatsubo T, Ihara R, Suzuki K, Matsuyama Y, Tomita N, Arai H, Ishii K, Senda M, Ito K, Ikeuchi T, Kuwano R, Matsuda H; Alzheimer’s Disease Neuroimaging Initiative; Japanese Alzheimer’s Disease Neuroimaging Initiative.Effects of sex, educational background, and chronic kidney disease grading on longitudinal cognitive and functional decline in patients in the Japanese Alzheimer’s Disease Neuroimaging Initiative study. Alzheimers Dement (N Y). 2018 Jul 12;4:765-774.
12. Ihara R, Iwata A, Suzuki K, Ikeuchi T, Kuwano R, Iwatsubo T; Japanese Alzheimer’s Disease Neuroimaging Initiative. Clinical and cognitive characteristics of preclinical Alzheimer’s disease in the Japanese Alzheimer’s Disease Neuroimaging Initiative cohort. Alzheimers Dement (N Y). 2018 Nov 26;4:645-651.
13. Ellis KA, Bush AI, Darby D, De Fazio D, Foster J, Hudson P, Lautenschlager NT, Lenzo N, Martins RN, Maruff P, Masters C, Milner A, Pike K, Rowe C, Savage G, Szoeke C, Taddei K, Villemagne V, Woodward M, Ames D; AIBL Research Group. The Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging: methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer’s disease. Int Psychogeriatr. 2009 Aug;21(4):672-87.
14. Clark CM, Schneider JA, Bedell BJ, Beach TG, Bilker WB, Mintun MA, Pontecorvo MJ, Hefti F, Carpenter AP, Flitter ML, Krautkramer MJ, Kung HF, Coleman RE, Doraiswamy PM, Fleisher AS, Sabbagh MN, Sadowsky CH, Reiman EP, Zehntner SP, Skovronsky DM; AV45-A07 Study Group. Use of florbetapir-PET for imaging beta-amyloid pathology. JAMA. 2011 Jan 19;305(3):275-83.
15. Shaw LM, Vanderstichele H, Knapik-Czajka M, Clark CM, Aisen PS, Petersen RC, Blennow K, Soares H, Simon A, Lewczuk P, Dean R, Siemers E, Potter W, Lee VM, Trojanowski JQ; Alzheimer’s Disease Neuroimaging Initiative. Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects. Ann Neurol. 2009 Apr;65(4):403-13.
16. Yamane T, Ishii K, Sakata M, Ikari Y, Nishio T, Ishii K, Kato T, Ito K, Senda M; J-ADNI Study Group. Inter-rater variability of visual interpretation and comparison with quantitative evaluation of 11C-PiB PET amyloid images of the Japanese Alzheimer’s Disease Neuroimaging Initiative (J-ADNI) multicenter study. Eur J Nucl Med Mol Imaging. 2017 May;44(5):850-857.
17. Sato K, Mano T, Ihara R, Suzuki K, Tomita N, Arai H, Ishii K, Senda M, Ito K, Ikeuchi T, Kuwano R, Matsuda H, Iwatsubo T, Toda T, Iwata A; Alzheimer’s Disease Neuroimaging Initiative, and Japanese Alzheimer’s Disease Neuroimaging Initiative. Lower Serum Calcium as a Potentially Associated Factor for Conversion of Mild Cognitive Impairment to Early Alzheimer’s Disease in the Japanese Alzheimer’s Disease Neuroimaging Initiative. J Alzheimers Dis. 2019;68(2):777-788.
18. Sato K, Mano T, Matsuda H, Senda M, Ihara R, Suzuki K, Arai H, Ishii K, Ito K, Ikeuchi T, Kuwano R, Toda T, Iwatsubo T, Iwata A; Japanese Alzheimer’s Disease Neuroimaging Initiative. Visualizing modules of coordinated structural brain atrophy during the course of conversion to Alzheimer’s disease by applying methodology from gene co-expression analysis. Neuroimage Clin. 2019 Jul 25;24:101957.
19. Lautner R, Palmqvist S, Mattsson N, Andreasson U, Wallin A, Pålsson E, Jakobsson J, Herukka SK, Owenius R, Olsson B, Hampel H, Rujescu D, Ewers M, Landén M, Minthon L, Blennow K, Zetterberg H, Hansson O; Alzheimer’s Disease Neuroimaging Initiative. Apolipoprotein E genotype and the diagnostic accuracy of cerebrospinal fluid biomarkers for Alzheimer disease. JAMA Psychiatry. 2014 Oct;71(10):1183-91.
20. Max Kuhn. Contributions from Jed Wing, Steve Weston, Andre Williams, Chris Keefer, Allan Engelhardt, Tony Cooper, Zachary Mayer, Brenton Kenkel, the R Core Team, Michael Benesty, Reynald Lescarbeau, Andrew Ziem, Luca Scrucca, Yuan Tang, Can Candan and Tyler Hunt. (2018). caret: Classification and Regression Training. R package version 6.0-81. (
21. Pascoal TA, Mathotaarachchi S, Shin M, Park AY, Mohades S, Benedet AL, Kang MS, Massarweh G, Soucy JP, Gauthier S, Rosa-Neto P; Alzheimer’s Disease Neuroimaging Initiative.Amyloid and tau signatures of brain metabolic decline in preclinical Alzheimer’s disease. Eur J Nucl Med Mol Imaging. 2018 Jun;45(6):1021-1030.
22. Yosinski J., Clune J., Bengio Y., Lipson H. NIPS; 2014. How Transferable are Features in Deep Neural Networks? pp. 3320–3328.
23. Wee CY, Liu C, Lee A, Poh JS, Ji H, Qiu A; Alzheimers Disease Neuroimage Initiative. Cortical graph neural network for AD and MCI diagnosis and transfer learning across populations. Neuroimage Clin. 2019;23:101929.
24. Sato K, Ihara R, Suzuki K, Niimi Y, Toda T, Jimenez-Maggiora G, Langford O, Donohue MC, Raman R, Aisen PS, Sperling RA, Iwata A, Iwatsubo T. Predicting amyloid risk by machine learning algorithms based on the A4 screen data: Application to the Japanese Trial-Ready Cohort study. Alzheimers Dement (N Y). 2021 Mar 24;7(1):e12135.



J. Kaye1, P. Aisen2, R. Amariglio3, R. Au4, C. Ballard5, M. Carrillo6, H. Fillit7, T. Iwatsubo8, G. Jimenez-Maggiora2, S. Lovestone9, F. Natanegara10, K. Papp3, M.E. Soto11, M. Weiner12, B. Vellas13, and the EU/US CTAD Task Force


*EU/US/CTAD TASK FORCE: Sandrine Andrieu (Toulouse); Matthew Barton (Raleigh); Randall Bateman (Saint Louis); Monika Baudler (Basel); Joanne Bell (Willmington); Kevin Biglan (Indianapolis); Adam Boxer (San Francisco); Sasha Bozeat (Basel); Claudine Brisard (Issy les Moulineaux); Miroslaw Brys (Indianapolis); Marc Cantillon (Gilbert); Bill Chan (Beijing); Ivan Cheung (Woodcliff Lake); Min Cho (Woodcliff Lake); Julia Coelho (San Francisco); Shobha Dhadda (Woodcliff Lake); Daniel Di Giusto (Basel); Michael Donohue (San Diego); Rachelle Doody (Basel); John Dwyer (Washington); Michael Egan (North Wales); Rianne Esquivel (Malvern); Wendy Galpern (New Jersey); Harald Hampel (Woodcliff Lake); Jason Hassenstab (St Louis); David Henley (New Jersey); Joseph Herring (North Wales); Carole Ho (South San Francisco); Michael Irizarry (Woodcliff Lake); Keith Johnson (Boston); Geoffrey Kerchner (South San Francisco); Gene Kinney (South San Francisco); Shailaja Korukonda (Woodcliff Lake); Lynn Kramer (Woodcliff Lake); Jaren Landen (Cambridge); Ishani Landri (Woodcliff Lake); Lars Lannfelt (Uppsala); Valérie Legrand (Nanterre); Manoj Malhotra (Woodcliff Lake); Eric McDade (St Louis); Francisco Nogueira (South San Francisco); Gerald Novak (New Jersey); Gunilla Osswald (Stockholm); Susanne Ostrowitzki (South San Francisco); Amanda Paley (New York); Martin Rabe (Woodcliff Lake); Rema Raman (San Diego); Elena Ratti (Cambridge); Laurie Ryan (Bethesda); Stephen Salloway (Providence); Peter Schüler (Langen); Hiroshi Sekiya (Malvern); Jiong Shi (Las Vegas); Melanie Shulman (Cambridge); Eric Siemers (Zionsville); John Sims (Indianapolis); Kaycee Sink (South San Francisco); Reisa Sperling (Boston); Joyce Suhy (Newark); Jina Swartz (London); Pierre Tariot (Phoenix); Edmond Teng (South San Francisco); Jacques Touchon (Montpellier); Martin Traber (Basel); Andrea Vergallo (Woodcliff Lake); Judy Walker (Singapore); Jon Walsh (San Francisco); Alette Wessels (Indianapolis); Haichen Yang (North Wales); Wagner Zago (San Francisco); Kenton Zavitz (Cambridge).

1. Layton Aging and Alzheimer’s Disease Center, School of Medicine, Oregon Health and Science University, Portland, OR, USA; 3. Alzheimer’s Therapeutic Research Institute, University of Southern California, San Diego, CA, USA; 3. Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; 4. Boston University, Boston, MA, USA; 5. University of Exeter, Exeter, UK; 6. Alzheimer’s Association, Chicago IL, USA; 7. Alzheimer Drug Discovery Foundation, New York, NY, USA; 8. University of Tokyo, Tokyo, japan; 9. Janssen Pharmaceutical Company, High Wycombe, UK; 10. Eli Lilly and Company, Indianapolis IIN, USA; 11. Toulouse University Hospital, Toulouse, France; 12. University of California, San Francisco CA, USA; 13. Gerontopole, INSERM U1027, Alzheimer’s Disease Research and Clinical Center, Toulouse University Hospital, Toulouse, France

Corresponding Author: Jeffrey Kaye, Layton Aging and Alzheimer’s Disease Center, School of Medicine, Oregon Health and Science University, Portland, OR, USA,

Published online June 28, 2021,



The 2020 COVID-19 pandemic has disrupted Alzheimer’s disease (AD) clinical studies worldwide. Digital technologies may help minimize disruptions by enabling remote assessment of subtle cognitive and functional changes over the course of the disease. The EU/US Clinical Trials in Alzheimer’s Disease (CTAD) Task Force met virtually in November 2020 to explore the opportunities and challenges associated with the use of digital technologies in AD clinical research. While recognizing the potential of digital tools to accelerate clinical trials, improve the engagement of diverse populations, capture clinically meaningful data, and lower costs, questions remain regarding the stability, validity, generalizability, and reproducibility of digital data. Substantial concerns also exist regarding regulatory acceptance and privacy. Nonetheless, the Task Force supported further exploration of digital technologies through collaboration and data sharing, noting the need for standardization of digital readouts. They also concluded that while it may be premature to employ remote assessments for trials of novel experimental medications, remote studies of non-invasive, multi-domain approaches may be feasible at this time.

Key words: Alzheimer’s disease, clinical outcomes, digital tools, remote assessments.



The 2020 COVID-19 pandemic disrupted clinical trials worldwide as health care facilities became overwhelmed and lock-down conditions shuttered clinical trial sites, forcing both study staff and trial participants to stay home (1–3). Nearly every ongoing clinical study has felt the impact, although the degree of impact varies. Alzheimer’s disease (AD) studies have been acutely affected for multiple reasons, including: 1) the increased vulnerability to COVID-19 of people with or at risk of AD due to their advanced age and high prevalence of comorbid conditions; 2) the effects of even mild cognitive decline on the ability of participants to comply with the operational changes in how trials are conducted (e.g., social distancing from study staff and study partners, changes in study protocols); and 3) effects on clinical endpoints due to isolation, confusion, and behavioral and psychological symptoms of dementia (BPSD) such as agitation, apathy, and depression (1, 4). Other disruptions common to all clinical trials including those for AD interventions include: 1) reduced clinical capacity; 2) increased costs due to the need for increased personal protective equipment (PPE) and infrastructure for altered intervention delivery; 3) delays in ethics approvals as institutional review boards (IRBs) are swamped with COVID-19 related protocols; 4) reluctance of patients and participants to be in contact with clinics and health providers; and 5) suspension of patient screening and recruitment.
Study sponsors have responded to these challenges in different ways; for example, by mailing drugs to trial participants to ensure participants have access to interventions; limiting clinic visits through the establishment of alternative trial sites; allowing a brief hiatus from dosing; and implementing home infusions. Funders have also responded with innovations. In an analysis of their funding portfolio during the latter half of 2020, the Alzheimer’s Drug Discovery Foundation (ADDF) found that about 70% of grantees expected their studies to be delayed, and some requested no-cost extensions. Delays may harm not only budgets but data analyses as well, since data interpretation can be muddled due to long-term disruptions, missing data, and protocol deviations; with clinical studies more affected than preclinical studies. Observational studies have also been affected. For example, Swedish BioFINDER2 halted enrollment, and the Rush University aging cohort studies stopped performing autopsies due to lack of PPE. In response to these challenges, the Alzheimer’s Association introduced a Rapid Program in Dementia (RAPID) funding program to provide bridge funding for Association-funded research during the pandemic.
In March 2020, the U.S. Food and Drug Administration (FDA) issued guidelines on the conduct of clinical trials during the pandemic and updated this guidance in December 2020 (5). The guidance aims to protect the safety of trial participants while ensuring good clinical practice and trial integrity. It also recognized the likely use of remote assessments and new technologies for data acquisition.
In recent years, digital technologies that enable remote assessment and gather real-world granular data on a variety of cognitive and functional outcomes have received increased attention and interest in the AD drug development community (6). Recognizing the urgency of developing these technologies due to the pandemic, the EU/US CTAD Task Force met in November 2020 to address the topic, bringing together an international group of clinical investigators from academia and industry along with funders. They explored emerging digital technologies that are being used in clinical research as well as the challenges that have yet to be addressed to realize the potential of these technologies. First and foremost, they recognized that global challenges such as the COVID-19 pandemic require global solutions.


Opportunities and challenges to make trials more robust using digital technologies

Even if there were no pandemic, clinical trials as currently conducted are too slow and costly to adequately respond to a disease such as Alzheimer’s, which currently affects around 50 million people worldwide and for which there is no approved disease-modifying treatment (7). The difficulty of conducting clinical trials for AD treatments is exacerbated by the use of conventional outcome measures that are insensitive to change early in disease and sensitive to environmentally-induced variation. By capturing alternative outcome measures in real-world settings, digital devices have the potential to mitigate these problems, thereby expediting trials, lowering costs, and enabling more effective clinical trials. Digital tools can also promote inclusiveness and social justice, increase the pool of participants, decrease sample size, and increase trial effectiveness by improving the generalizability and representation of vulnerable populations. Smartphones hold particular promise in the realm of digital tools since they are among the most penetrating technologies available, with over three billion users worldwide (8) .
However, converting trial outcomes from in-person to remote measures is also associated with many challenges. Among the most critical aspects to be considered are usability and the reliability and fidelity of technology-related measures (9). Consumer-oriented devices and apps must be scaled for use in clinical trial settings and confounds introduced by remote assessment must be addressed. User training and support will be critical to reduce user errors. Technologies themselves should not drive their use in clinical trials; rather the use case should drive the technology.
In the AD space, digital tools have shown particular promise in capturing subtle cognitive and functional changes in preclinical AD. Yet there remain questions regarding their ability to serve as outcome measures due to questions around feasibility, reliability, and validity.


Implementing digital devices in AD clinical studies

As the COVID-19 pandemic has necessitated moving clinical trials from specialized centers to home-based virtual assessments using both passive and active measures, investigators are learning lessons that have the potential to accelerate all clinical trials for dementia. Indirect measures of behavior, function, and physiology are being explored as cognitive and functional endpoints for clinical trials by individual research groups as well as technology companies. For example, Elektra Labs has created the Atlas platform, which takes a data science approach to clinical trials through the application of a wide selection of connected sensors and physiological and behavioral markers.
Many research groups that have developed and/or implemented digital technologies in clinical studies presented data and plans to the Task Force, providing lessons learned as well as strategies for moving forward:


The Collaborative Aging Research Using Technology (CART) Initiative funded by the US National Institutes of Health and the Department of Veterans Affairs, has developed a multi-functional digital assessment platform that is technology agnostic, use case flexible, sharable, and scalable (10). The platform, an extension of the Oregon Center for Aging & Technology (ORCATECH) assessment system at Oregon Health and Science University has been deployed to 301 older adults for up to three years in four US regions. It enables unobtrusive, long-term assessment across multiple domains (e.g., mobility, cognitive function, sleep) through the use of many different sensors and connected devices (e.g., electronic pillboxes, bed mats, actigraphs, online reporting on laptops, tablets or smartphones). The platform is currently deployed by multiple groups in the United States and Canada, with plans to extend the platform to France (in the community-wide INSPIRE study based at Toulouse University Hospital) and Australia. Testing in two cohorts — a low-income housing cohort in Portland, Oregon, and an African-American Minority Aging Research Study (MARS) cohort in Chicago – gathered data spanning the time period of the pandemic demonstrating increased loneliness and decreased physical activity (Figure 1). In a proof-of-concept trial, the platform has also demonstrated feasibility for monitoring and predicting agitation among residents of memory care facilities (11), suggesting that it may be possible to demonstrate the efficacy of treatments for agitation with sensors that monitor life-space activity.

Figure 1. Pervasive Computing and Continuous Monitoring Demonstrate Effects of COVID-19 Pandemic on Loneliness and Physical Activity

In two separate cohorts, digital data streams provided by the CART ORCATECH platform demonstrated changes in mood (through online self-reported questionnaires) and physical activity (through actigraphy) that coincided with the imposition of stay-at-home orders because of the pandemic.



The public private partnership Remote Assessment of Disease and Relapse – AD (RADAR-AD), funded by the Innovative Medicines Initiative (IMI) has embraced an “internet of things” (IOT) approach similar to the CART/ORCATECH system to measure early decline in cognitive function using low friction digital devices across a range of challenges. A study is planned that will enroll 200 participants over age 50 including those with preclinical AD, MCI due to AD, mild-to-moderate AD, and age and gender matched controls. RADAR earlier launched a multi-site prospective cohort study to determine the feasibility, usability, and acceptability of remote measurement technology (RMT) for major depressive disorder (MDD) (12).
Remote digital measurements that may be used in RADAR-AD include adaptations of multiple instruments to be administered using a table or smartphone interface such as the Amsterdam-instrumental activities of daily living questionnaire (A-IADL-Q) (13) to assess orientation, planning skills, and memory; and the Patient-Reported Outcomes Measurement Information Systems (PROMIS) general health measures to assess quality of life, disabilities, pain, and fatigue (14).
The RADAR base app, using a smartphone and passive sensors, will assess mobility, social communication, time at home, general activity, and sleep duration and quality. A residential movement detector will also assess gait and mobility at home and time spent in different areas of the home. The Mezurio smartphone app will capture high-frequency data related to cognition, including sleep, planning, memory, mood and anxiety, voice and speech, and typing and swiping (15).

Harvard Aging Brain Study (HABS)

In its search for sensitive measures of subtle cognitive decline in preclinical AD, The Harvard Aging Brain Study (HABS) developed an iPad-based Computerized Cognitive Composite (C3) and subsequently a smartphone-based tool for remote assessment of memory. The C3 on the iPAD comprises 3 components: a Behavioral Pattern Separation task, a Face-Name Associative Memory Exam, and the Cogstate Brief Battery. Development of the C3 included its use in the A4 study, a secondary prevention trial in clinically normal older adults with abnormal levels of amyloid who were randomized to receive placebo or solanezumab. This initial study showed that when the drug was administered in screening period, the amyloid-positive group performed worse on C3 compared to the Preclinical Alzheimer Cognitive Composite (PACC) (16, 17). The C3 has also been shown to provide reliable data when participants complete the assessments at home on an iPad (18). The remote use of C3 allows for more frequent assessments, which provide new information about subtle cognitive changes (19).
HABS has also developed the Boston Remote Assessment for Neurocognitive Health (BRANCH), which combines two paired associative learning tasks, an associated memory test, and a continuous visual recognition task accessible on a smartphone. In a recent study, BRANCH was shown to be correlated with scores on the paper-based preclinical Alzheimer’s cognitive composite (PACC5) (20).
HABS investigators have also been using BRANCH to assess learning curves using data from clinically normal older adults who completed 5 consecutive days of BRANCH. Early evidence suggests that a steeper learning curve is associated with better performance on the PACC, although the results were not statistically significant. Nonetheless, capturing learning curves using a smartphone over several days appears to be feasible and valid and may be useful to prescreen potential trial participants for biomarker status and identify those at greater risk for imminent cognitive decline and most likely to benefit from treatment. Cognitive learning curves may also serve as a paradigm for exploring psychopharmacokinetic profiles of medications to assess for efficacy over relatively short time periods.

The Trial-Ready Cohort for Preclinical and Prodromal Alzheimer’s Disease (TRC-PAD)

The Trial-Ready Cohort for Preclinical and Prodromal Alzheimer’s Disease (TRC-PAD) was established to accelerate drug development for AD by ensuring timely recruitment of targeted individuals into optimally designed preclinical and prodromal trials. TRC-PAD gathers multiple types of data including digital data from wearables and sensors through its informatics platform (TRC-PAD IP). In concert with TRC-PAD, the Alzheimer’s APT Webstudy registry was established along with an analytics platform, a referral management system, and a data management system (21).
The APT Webstudy has been working to optimize digital smartphone-based tools to assess cognition, such as the Cogstate Brief Battery. The study plans to implement this tool in the AHEAD 3-45 secondary prevention trials in early preclinical and preclinical AD and in the CT1812 trial in early AD (22). APT is also exploring alternative remote testing sensitive to amyloid elevation.

WHO – Integrated Care for Older People (ICOPE) Monitor

In its World Report on Healthy Ageing, the World Health Organization (WHO) redefined healthy aging in terms of functional ability, which is a product of an individual’s intrinsic capacity (IC) interacting with the environment (23). In order to implement integrated care for older people (ICOPE), an international partnership of researchers have developed a tool for assessing IC (24) as well as a mobile app available on smartphones that enable self-assessment by individuals to provide health care professionals with data on intrinsic capacity (25). Initial results from more than 5300 older individuals in the ICOPE Digital Cohort report high levels of IC impairments in multiple domains, including in vision, hearing, psychological health, and cognition. Recognizing that additional cognitive assessments are needed, ICOPE data will be combined with the Alzheimer’s Prevention Trial (APT) Webstudy data beginning in 2021. The ICOPE integrative approach from WHO provides the opportunity to detect memory complaints or decline, taking into consideration declines in other ICs, such as vision, hearing, mood, nutrition.

Brain Health Registry

With a goal of improving clinical trial recruitment, the Brain Health Registry (BHR) was established in 2014 as a website and online registry that collects digital data from potential clinical trial participants and study partners through self-report questionnaires and neuropsychological tests. Among the over 70,000 individuals enrolled in BHR, there have been over 106,000 referrals and 5,000 enrolled in other clinical studies. Through co-enrollment, BHR and clinical trial data can be linked.
BHR is also involved in methods development and validation of new online methods, including an electronic Clinical Dementia Rating (CDR) scale, a digital financial capacity instrument, and other digital assessment tools. In support of blood biomarkers studies, BHR has also collaborated with Quest Diagnostics to scale up blood collection using remote identification through the registry.

Other Studies Incorporating Digital Assessment Tools

Many other clinical research programs are exploring using digital measures of cognition and function (e.g., web- or mobile- device based cognitive testing, automated speech and language assessment, or use of sensor-embedded home-based device responsives). For example, the Framingham Heart Study (FHS) is developing a smartphone platform with multiple applications including digital voice. Voice (spoken utterances and language) is measured through device agnostic technologies and FHS research shows that linguistic and acoustic measures can be used as biomarkers of cognitive function (26–28). However, while speech and language assessments are likely to be very sensitive, additional work is needed to determine the best way to capture speech samples.
The Deep and Frequent Phenotyping (DFP) study in the United Kingdom has also incorporated digital measures from wearables and smartphones to capture highly granular data on gait, memory, navigational ability, and other measures that may serve as proxies for cognition and function (29). The study aims to identify an optimal set of markers for patient stratification.


Potential benefits and challenges associated with incorporating digital tools in trials

Digital devices offer the potential to expedite clinical trials, reduce the sample sizes or duration of studies, lower the cost of acquiring and analyzing data, enable the recruitment of more diverse and representative populations, reduce the need for specialized raters, reduce rater burden, and increase the clinical meaningfulness of outcome data collected to determine efficacy of an intervention. During a pandemic, digital tools also may provide the means for remote assessments, which could mitigate problems associated with closure of clinical trial sites and the safety concerns associated with in-person assessments and interventions.
Combining cognitive and functional endpoints in clinical trials is well established through the use of questionnaires in symptomatic stages of disease; and in preclinical disease, reports from both participants and study partners regarding high-level functional activities appear to provide valuable information on disease trajectory. Pairing digital cognitive measures with digital functional assessments may offer increased sensitivity in preclinical disease. The HABS team has shown that functional inventories can be administered remotely through the use of REDCap software, saving time and reducing data errors without sacrificing data quality.
The technologies employed in various digital devices provide additional potential advantages:
• They allow data to be time stamped, which enables assessment of everyday function.
• They can improve participant compliance by providing study participants and partners with reminders to take medications and engage in study-related activities.
• Audio and video recordings can provide a level of biometric validation to ensure that study participants themselves (rather than study partners) are providing data.
• Passive sensing of activities such as driving and computer use have demonstrated unique personalized patterns that can be captured using relatively simple algorithms.
• Passive sensors reduce participant burden, thus potentially reducing drop-outs and producing outcomes that are easily understandable and relevant in a real-world context.
• Meta-data such as time taken to complete a questionnaire or remember to take medication on time may also reflect changes in episodic memory, a key domain affected in AD.
• Passive assessment of sleep may prove useful since disrupted sleep has been shown to be a mediator of amyloid deposition in many studies.

To achieve the promise of digital devices will require addressing many challenges including those related to reliability, validation, user acceptability, regulatory approval, privacy and security. While regulators have indicated a willingness to consider digital data, much work remains to be completed before digital measures are accepted as primary or secondary outcomes. Indeed, the FDA issued guidance on the conduct of trials during a pandemic which includes increased flexibility around trial procedures intended to maintain participant safety, but less flexibility around efficacy measures. Task Force members also questioned whether regulators would accept remotely acquired data where there is no assurance that study participants and not study partners completed questionnaires or cognitive outcome assessments.
While digital registries may enable outreach to otherwise under-represented populations, engagement and retention of trial participants through digital means alone may be insufficient. Indeed, studies have reported high compliance rates that they attribute to personalized rather than anonymous digital engagement of study participants (30).
Transitioning cognitive assessments to smartphones may enable more frequent and sensitive measurement among large numbers of participants in a short period of time. However, smartphones may introduce additional challenges with regard to data fidelity and security. Other technological challenges related to digital data include the difficulty of analyzing continuous data and deconvoluting diverse data captured by devices. Addressing these challenges will be advanced by wider collaboration with the computer science and data science community.
Digital readouts are diverse and non-standardized, which results in impediments to data aggregation and sharing of data, both of which are essential to advance the field. Collaboration among technology developers is needed to ensure that raw data from which readouts on different devices are derived will be made available to the research community. Privacy concerns represent an additional roadblock to data sharing. Federated data management platforms have been developed that provide researchers with access to multiple types of data from multiple sources (31). The Alzheimer’s Disease Data Initiative has also established the AD Workbench (ADWB) platform to promote data, information, and tool sharing across the AD research community (32), although no digital biomarker data are yet incorporated in this relatively new effort.


Moving Forward

Despite the challenges of conducting clinical studies during the pandemic, the AD clinical research community remains committed to advancing current studies and launching new studies. Indeed, Task Force participants suggested that the COVID-19 experience may fuel preparations for better clinical trials that mitigate both the risks associated with the pandemic and more common complications of clinical trials for example by broadening the reach of recruitment efforts.
Among the possibilities discussed by the Task Force was the potential to design completely digital or hybrid trials combining virtual/telemedicine and physical visits. For example, telemedicine consultations in the pre-screening phase of a trial could help verify inclusion and exclusion criteria, provide information to potential trial participants, and obtain electronic consent. Virtual visits could also be used for trial components that do not require technically invasive procedures. The INSPIRE study team at Toulouse University Hospital tested this strategy following the first COVID lockdown in France and found a high rate of acceptance among participants (33).
Digital devices may provide new analytical methods as well, for example, by enabling assessment of intraindividual change in a high frequency way. Digital tools implemented during the pandemic may also provide increased understanding to the effects of daily stress, anxiety, and poor sleep on cognitive outcomes. They may also enable continued assessment of cognitive function in participants post-infection with COVID-19.
To ensure that clinical studies continue through the pandemic and its aftermath, the Task Force recommended that regulators and institutional review boards provide guidance yet maintain flexibility with regard to changes in applications/devices and statistical plans and allow for virtual and home-based visits and remote monitoring. Developers must consider alternate or additional trial sites, and funders should expect unanticipated modifications to protocols and funding needs.
The Task Force concluded its session with a discussion of the field’s readiness to conduct remote clinical trials. Task Force members urged caution in employing remote assessments for trials of experimental medications; the digital measures at this time may be most productively embedded in parallel with conventional measures in early-stage Phase 1 and 2 trials, demonstrating their fidelity or even superiority to current commonly used clinical outcome measures. Enthusiasm was expressed for using digital technologies in remote studies of non-invasive interventions such as studies of vitamins, exercise, and mind-body or multidomain approaches. Trials of repurposed drugs for which safety data exists may also be feasible as entirely remote trials. For example, because of the COVID-19 pandemic, the Alzheimer’s Disease Cooperative Study (ADCS) is attempting to conduct the PEACE-AD (prazosin for disruptive agitation in Alzheimer’s disease) trial (34) as a remote-only trial.


Acknowledgements: The authors thank Lisa J. Bain for assistance in the preparation of this manuscript.

Conflicts of interest: The Task Force was partially funded by registration fees from industrial participants. These corporations placed no restrictions on this work. Dr. Kaye reports grants from Merck, Eisai, Genentech, and Abbvie; and serves on a Data Safety Monitoring Committee at Eli Lilly. Dr. Aisen reports grants from Janssen, Eli Lilly, Eisai, NIA, the Alzheimer’s Association, and FNIH; and consulting fees from Biogen, Roche, Merck, Abbvie, Lundbeck, Proclara, and Immunobrain Checkpoint. Dr. Amariglio declares there are no conflicts. Dr. Au is on the scientific advisory board of Signant Health and a consultant to Biogen. Dr. Ballard reports grants from Arcadia pharmaceutical company and Lundbeck; personal fees from Arcadia, Lundbeck, Roche, Otsuka, Novartis, Eli Lilly, Suven, Sunovion, ADDEX, and Exciva; personal fees and other from Synexus, Novo Nordisk; and consulting fees from Biogen. Dr. Carillo declares there are no conflicts. Dr. Fillit declares there are no conflicts. Dr. Iwatsubo declares there are no conflicts. Dr. Jimenez-Maggiora declares there are no conflicts. Dr. Lovestone is an employee of Janssen Medical Ltd and a co-founder of Akrivia Health Ltd . Dr. Natanegara is an employee of Eli Lilly and Company. Dr. Papp declares there are no conflicts. Dr. Soto declares payment as consultant or advisor from Avanir, Acadia. Dr. Weiner reports grants from Siemens, Biogen, and Johnson & Johnson, NIH, California Department of Health, University of Michigan, Hillblom Foundation, Alzheimer’s Association, State of California, Kevin and Connie Shanahan, GE, VUMc, American Catholic University, The Stroke Foundation, and the Veteran’s Administration; has served on advisory boards for Eli Lilly, Cerecin/Accera, Roche, Alzheon Inc., Merck Sharp & Dohme Corp, Nestle/Nestec; and has consulted with or acted as a speaker to Cerecin/Accera Inc., BioClinica, Nestle/Nestec, Genentech, FUJIFILM-Toyama Chemical, and T3D Therapeutics; and holds stock options with Alzheon Inc, Alzeca, and Anven. Dr. Vellas reports grants from Lilly, Merck, Roche, Lundbeck, Biogen, grants from Alzheimer’s Association, European Commission, personal fees from Lilly, Merck, Roche, Biogen, outside the submitted work.

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. Morris MC, Tangney CC, Wang Y, et al. MIND diet slows cognitive decline 1. van Dorn A. COVID-19 and readjusting clinical trials. The Lancet. 2020 Aug 22;396(10250):523–4.
2. Fleming TR, Labriola D, Wittes J. Conducting Clinical Research During the COVID-19 Pandemic: Protecting Scientific Integrity. JAMA. 2020 Jul 7;324(1):33–4.
3. Sathian B, Asim M, Banerjee I, Pizarro AB, Roy B, van Teijlingen ER, et al. Impact of COVID-19 on clinical trials and clinical research: A systematic review. Nepal J Epidemiol. 2020 Sep;10(3):878–87.
4. Brown EE, Kumar S, Rajji TK, Pollock BG, Mulsant BH. Anticipating and Mitigating the Impact of the COVID-19 Pandemic on Alzheimer’s Disease and Related Dementias. Am J Geriatr Psychiatry. 2020 Jul;28(7):712–21.
5. Food and Drug Administration. Conduct of Clinical Trials of Medical Products During the COVID-19 Public Health Emergency Guidance for Industry, Investigators, and Institutional Review Boards. U.S. Department of Health and Human Services,; 2020.
6. Gold M, Amatniek J, Carrillo MC, Cedarbaum JM, Hendrix JA, Miller BB, et al. Digital technologies as biomarkers, clinical outcomes assessment, and recruitment tools in Alzheimer’s disease clinical trials. Alzheimers Dement (N Y). 2018;4:234–42.
7. World Health Organization. Dementia [Internet]. 2020 [cited 2021 Jan 6]. Available from:
8. Statista. Number of smarphone users worldwide from 2016 to 2021 [Internet]. Statista. 2020 [cited 2021 Jan 26]. Available from:
9. Lindauer A, Seelye A, Lyons B, Dodge HH, Mattek N, Mincks K, et al. Dementia Care Comes Home: Patient and Caregiver Assessment via Telemedicine. Gerontologist. 2017 Oct 1;57(5):e85–93.
10. Beattie Z, Miller LM, Almirola C, Au-Yeung W-TM, Bernard H, Cosgrove KE, et al. The Collaborative Aging Research Using Technology Initiative: An Open, Sharable, Technology-Agnostic Platform for the Research Community. Digit Biomark. 2020;4(Suppl 1):100–18.
11. Au-Yeung W-TM, Miller L, Beattie Z, Dodge HH, Reynolds C, Vahia I, et al. Sensing a problem: Proof of concept for characterizing and predicting agitation. Alzheimers Dement (N Y). 2020;6(1):e12079.
12. Matcham F, Barattieri di San Pietro C, Bulgari V, de Girolamo G, Dobson R, Eriksson H, et al. Remote assessment of disease and relapse in major depressive disorder (RADAR-MDD): a multi-centre prospective cohort study protocol. BMC Psychiatry. 2019 Feb 18;19(1):72.
13. Stringer G, Leroi I, Sikkes SAM, Montaldi D, Brown LJE. Enhancing ‘meaningfulness’ of functional assessments: UK adaptation of the Amsterdam IADL questionnaire. International Psychogeriatrics. undefined/ed;1–12.
14. Tatsuoka C, DeMarco L, Smyth KA, Wilkes S, Howland M, Lerner AJ, et al. Evaluating PROMIS Physical Function Measures in Older Adults at Risk for Alzheimer’s Disease. Gerontol Geriatr Med [Internet]. 2016 Sep 5 [cited 2021 Jan 8];2. Available from:
15. Lancaster C, Koychev I, Blane J, Chinner A, Wolters L, Hinds C. Evaluating the Feasibility of Frequent Cognitive Assessment Using the Mezurio Smartphone App: Observational and Interview Study in Adults With Elevated Dementia Risk. JMIR Mhealth Uhealth [Internet]. 2020 Apr 2 [cited 2021 Jan 8];8(4). Available from:
16. Papp KV, Rentz DM, Maruff P, Sun C-K, Raman R, Donohue MC, et al. The Computerized Cognitive Composite (C3) in A4, an Alzheimer’s Disease Secondary Prevention Trial. J Prev Alzheimers Dis. 2021;8(1):59–67.
17. Papp KV, Rentz DM, Orlovsky I, Sperling RA, Mormino EC. Optimizing the preclinical Alzheimer’s cognitive composite with semantic processing: The PACC5. Alzheimers Dement (N Y). 2017 Nov;3(4):668–77.
18. Rentz D, Dekhtyar D, Sherman J, Burnham S, Blacker D, Aghjayan S, et al. The Feasibility of At-Home iPad Cognitive Testing For Use in Clinical Trials. 2015 [cited 2021 Jan 8]; Available from:
19. Samaroo A, Amariglio RE, Burnham S, Sparks P, Properzi M, Schultz AP, et al. Diminished Learning Over Repeated Exposures (LORE) in preclinical Alzheimer’s disease. Alzheimers Dement (Amst). 2020;12(1):e12132.
20. Papp KV, Samaroo AH, Chou H-CL, Buckley RF, Rentz D, Sperling RA, et al. Repeated memory-based assessments: Implications for clinical trials and practice. Alzheimer’s & Dementia. 2020;16(S6):e038143.
21. Jimenez-Maggiora GA, Bruschi S, Raman R, Langford O, Donohue M, Rafii MS, et al. TRC-PAD: Accelerating Recruitment of AD Clinical Trials through Innovative Information Technology. J Prev Alzheimers Dis. 2020;7(4):226–33.
22. Grundman M, Morgan R, Lickliter JD, Schneider LS, DeKosky S, Izzo NJ, et al. A phase 1 clinical trial of the sigma-2 receptor complex allosteric antagonist CT1812, a novel therapeutic candidate for Alzheimer’s disease. Alzheimers Dement (N Y). 2019 Jan 23;5:20–6.
23. Beard JR, Officer A, de Carvalho IA, Sadana R, Pot AM, Michel J-P, et al. The World report on ageing and health: a policy framework for healthy ageing. Lancet. 2016 May 21;387(10033):2145–54.
24. Takeda C, Guyonnet S, Sumi Y, Vellas B, Araujo de Carvalho I. Integrated Care for Older People and the Implementation in the INSPIRE Care Cohort. J Prev Alzheimers Dis. 2020;7(2):70–4.
25. Sanchez-Rodriguez D, Annweiler C, Gillain S, Vellas B. Implementation of the Integrated Care of Older People (ICOPE) App in Primary Care: New Technologies in Geriatric Care During Quarantine of COVID-19 and Beyond. J Frailty Aging. 2020 May 6;1–2.
26. Thomas JA, Burkhardt HA, Chaudhry S, Ngo AD, Sharma S, Zhang L, et al. Assessing the Utility of Language and Voice Biomarkers to Predict Cognitive Impairment in the Framingham Heart Study Cognitive Aging Cohort Data. J Alzheimers Dis. 2020;76(3):905–22.
27. Thomas NWD, Beattie Z, Marcoe J, Wright K, Sharma N, Mattek N, et al. An Ecologically Valid, Longitudinal, and Unbiased Assessment of Treatment Efficacy in Alzheimer Disease (the EVALUATE-AD Trial): Proof-of-Concept Study. JMIR Res Protoc. 2020 May 27;9(5):e17603.
28. Lin H, Karjadi C, Ang TFA, Prajakta J, McManus C, Alhanai TW, et al. Identification of digital voice biomarkers for cognitive health. Exploration of Medicine. 2020 Dec 31;1(6):406–17.
29. Koychev I, Lawson J, Chessell T, Mackay C, Gunn R, Sahakian B, et al. Deep and Frequent Phenotyping study protocol: an observational study in prodromal Alzheimer’s disease. BMJ Open. 2019 Mar 23;9(3):e024498.
30. Sano M, Egelko S, Zhu CW, Li C, Donohue MC, Ferris S, et al. Participant satisfaction with dementia prevention research: Results from Home-Based Assessment trial. Alzheimers Dement. 2018;14(11):1397–405.
31. Toga AW, Bhatt P, Ashish N. Global data sharing in Alzheimer’s disease research. Alzheimer Dis Assoc Disord. 2016;30(2):160–8.
32. Alzheimer’s Disease Data Initiative. AD Workbench [Internet]. ADDI. [cited 2021 Jan 11]. Available from:
33. Takeda C, Guyonnet S, Ousset PJ, Soto M, Vellas B. Toulouse Alzheimer’s Clinical Research Center Recovery after the COVID-19 Crisis: Telemedicine an Innovative Solution for Clinical Research during the Coronavirus Pandemic. J Prev Alzheimers Dis. 2020;7(4):301–4.
34. Alzheimer Disease Cooperative Study. PEACE-AD [Internet]. ADCS PEACE-AD. [cited 2021 Jan 11]. Available from:




T. Iwatsubo1,2, Y. Niimi2, H. Akiyama3


Department of Neuropathology, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan; 2. Unit for Early and Exploratory Clinical Development, The University of Tokyo Hospital, Tokyo, Japan; 3. Yokohama Brain and Spine Center, Yokohama, Japan

Corresponding Author: Takeshi Iwatsubo, Department of Neuropathology, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo Bunkyo-ku, Tokyo 113-0033, Japan. E-mail:, Phone +81-3-5841-3541, FAX +81-3-5841-3613.

J Prev Alz Dis 2021;
Published online June 30, 2021,


Alzheimer’s disease and Dementia is endemic in Japan

A recent report from the Hisayama study, a representative regional cohort in Japan, showed that the lifetime risk of dementia in the Japanese elderly population has exceeded 50%, in which >50% of all dementia cause is comprised by Alzheimer’s disease (AD) (1). The total number of individuals with dementia in Japan is more than >5 million, and the global costs of dementia is estimated to exceed 14.5 trillion yen (equivalent to ~133 billion USD). Thus, there is an urgent and compelling need for the prevention of dementia, especially that of AD.
As an introduction to the excellent pieces of papers on AD prevention featured in this special issue, here we try to briefly overview the history, current status and future perspectives of AD research in Japan.


Dawn of the molecular neuropathology of AD in Japan

A milestone discovery that brought about a breakthrough in AD research dates back to mid 1980s: Yasuo Ihara and Nobuyuki Nukina identified hyperphosphorylated species of tau as an integral component of paired helical filaments in AD brains (2). This finding boosted the progress in molecular neuropathology of tau protein in Japan, ranging from the clinical and neuropathological studies of amyotrophic lateral sclerosis and parkinsonism-dementia complex uniquely observed in the Kii peninsula, which shares common features to that of Guam (3), to the development of a series of tau-oriented fluid and PET biomarkers (4-6).
Next came the molecular characterization of amyloid β peptides (Aβ), especially that of Aβ that initially get deposited in amyloid plaques of AD brains (7). The invention of sandwich ELISAs making use of antibodies that selectively discriminate the C-terminal extent of Aβ opened up venues for the development of plasma-based Aβ biomarkers (8,9), as well as characterization of unique APP mutations affecting the Aβ aggregation (10). Finally, the identification of presenilin polypeptides as the determinant of Aβ42 overproduction (11) led to the structure-function analyses of the presenilin complex that represents the catalytic subunit of γ-secretase (12).


Delineating the AD trajectories toward disease-modification: J-ADNI and A4 studies

As the pathomechanisms of AD are being elucidated, establishing methods to detect the progression of AD in its early stages (e.g., mild cognitive impairment; MCI) in a multicenter trial setting using neuroimaging and fluid biomarkers, and building up a database delineating the natural history of the early stage of AD, have become paramount toward the ultimate goal of precise evaluation of the efficacy of disease-modifying therapies (DMTs). For this purpose, AD Neuroimaging Initiative (ADNI) has been conducted in North America since 2004, which has set a firm basis for the current global clinical trials of AD. In 2007, the Japanese (J-) ADNI study was launched as a multicenter, longitudinal observational study using an almost identical protocol to ADNI’s. J-ADNI was successfully concluded in 2014, making the J-ADNI database obtained from 537 individuals with AD, MCI, and normal cognition available for worldwide data sharing (13). Notably, profiles of decline in cognitive or functional measures in the prodromal AD population in J-ADNI and North American ADNI were remarkably similar, supporting the feasibility of bridging of clinical trials in prodromal AD between Asia and western countries (13). Also, J-ADNI sample repository has provided researchers with precious resources including genome; in this issue, Mano and colleagues have clearly shown that peripheral blood BRCA1 methylation positively correlated with the major Alzheimer’s disease risk factors in the J-ADNI participants (14).
The second stage of J-ADNI (J-ADNI2/AMED preclinical AD study) was planned to focus on the preclinical (asymptomatic) and prodromal AD stages with biomarker verification of AD pathology. Senda and colleagues show the PET and biomarker profiles of the preclinical and prodromal cohort; importantly, this study represents the first academic multicencer tau-PET study in Japan utilizing flortaucipir as a tracer (15).
Currently, the anti-amyloid treatment in asymptomatic AD (A4) study is being conducted in 1169 preclinical AD participants from North America, Australia and Japan as a double-blind, randomized trial for 4.5 years using solanezumab as a prevention drug candidate, and the screening, pre-randomization data have been shared for research (16). Making use of the ADNI, J-ADNI and A4 screen data, Sato and colleagues propose a novel approach to predict preclinical AD (17). In view of the forthcoming series of preclinical AD trials using new AD DMTs (e.g., the AHEAD study), a method for efficiently recruiting individuals with preclinical AD is mandatory. To this end, we have started a collaboration with the Trial Ready Cohort for the Prevention of AD (TRC-PAD) in the US, and launched a trial ready cohort for preclinical AD in Japan, by combining a web-based, screening registry (J-TRC webstudy) and an in-person longitudinal study implementing cognitive assessments by Preclinical AD Cognitive Composite (PACC), amyloid PET (either by florbetapir or flutemetamol) and blood testing (J-TRC onsite study). Sato and colleagues have employed machine learning models fitted to the A4 screen data, to predict the amyloid PET results from the J-TRC webstudy data (18). Such an attempt will facilitate and ensure the effective recruitment of preclinical AD participants toward the upcoming prevention trials.


Non-pharmacological intervention into the lifestyle and modifiable risks

In addition to the pharmacological intervention approaches using DMTs in the early stages of AD, multi-domain interventions targeting modifiable risk factors for dementia in the lifestyle of older adults as evidenced by epidemiological studies is attracting enormous attention, since the Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER) has demonstrated that a multidomain lifestyle intervention (i.e., dietary counseling, physical exercise, cognitive training, and vascular risk monitoring) as a large randomized controlled trial can ameliorate cognitive decline in older adults at increased risk of developing dementia (19). In 2019, Arai, Sakurai and colleagues at the National Center for Geriatrics and Gerontology in Aichi have initiated the Japan-Multimodal Intervention Trial for Prevention of Dementia (J-MINT) as a member of the WW-FINGERS Network, sponsored by the Japan Agency for Medical Research and Development and the Ministry of Economy, Trade and Industry, Japan, aiming to verify whether multi-domain intervention consisting of management of vascular risk factors, group-based physical exercise and self-monitoring of physical activity, nutritional counseling, and cognitive training, could prevent the progression of cognitive decline in the Japanese elderly population (20). Another unique feature of the J-MINT trial is that the implementation of relevant interventional measures is led by private sectors, headed by SOMPO care Inc.
Successful interventions into the life-style to prevent cognitive decline and dementia in humans readily elicit reverse-translational approaches to elucidate the mechanism whereby metabolic stress aggravates AD pathophysiology in model animals. Wakabayashi and colleagues have characterized the effects of diet-induced insulin resistance on amyloid pathology (21), and shown that diet-induced and age-related endoplasmic reticulum stress can be attenuated by the administration of tauroursodeoxycholic acid, a chemical chaperone compound applicable to humans (22). Such an interplay between the translational and reverse-translational studies will facilitate the development of effective combinatorial interventions that retards the symptomatic and pathophysiological progression of cognitive decline in the elderly.


Future perspectives on the social implementation of research outcomes on dementia

The original contributions published in this special issue represent the favorable combinations of basic and clinical, as well as pharmacological and non-pharmacological approaches aiming at preventing AD and dementia, that are currently underway in Japan. Lastly, we would like to emphasize that the optimal coordination of the academic activities with different disciplines, often with those of private sectors, has been supported by the unique scholarly activities of the Japan Society for Dementia Research, the principal academic society on dementia comprised of a wide variety of members, e.g., from neurology, psychiatry, geriatrics, basic and social sciences. Development of prevention therapies on AD and dementia is a huge task on humans of the 21st century, which should involve multiple sectors, i.e., academia, industry, government, non-profit organizations, society, and individuals threatened to cognitive decline, under the spirit of public-private partnership, toward the ultimate goal of reducing the burden of cognitive decline on human being.


Conflict of interests: The author declares there are no conflicts.



1. Yoshida D, Ohara T, Hata J, Shibata M, Hirakawa Y, Honda T, Furuta Y, Oishi E, Sakata S, Kanba S, Kitazono T, Ninomiya T. Lifetime cumulative incidence of dementia in a community-dwelling elderly population in Japan. Neurology 2020; 95, e508-e518. doi: 10.1212/WNL.0000000000009917.
2. Ihara Y, Nukina N, Miura R, Ogawara M. Phosphorylated tau protein is integrated into paired helical filaments in Alzheimer’s disease. J Biochem 1986; 99, 1807-1810. doi: 10.1093/oxfordjournals.jbchem.a135662.
3. Kuzuhara S, Kokubo Y, Sasaki R, Narita Y, Yabana T, Hasegawa M, Iwatsubo T: Familial amyotrophic lateral sclerosis and parkinsonism-dementia complex of the Kii peninsula of Japan: clinical and neuropathological study and tau analysis. Ann Neurol 2001; 49, 501-511.
4. Iwatsubo T, Odaka A, Suzuki N, Mizusawa H, Nukina N, Ihara Y: Visualization of Aβ42(43) and Aβ40 in senile plaques with end-specific Aβ-monoclonals: Evidence that an initially deposited species is Aβ42(43). Neuron 1994; 13, 45-53. doi: 10.1016/0896-6273(94)90458-8
5. Kosaka T, Imagawa M, Seki K, Arai H, Sasaki H, Tsuji S, Asami-Odaka A, Fukushima T, Imai K, Iwatsubo T: The βAPP717 Alzheimer mutation increases the percentage of plasma amyloid β protein ending at Aβ42(43). Neurology 1997; 48, 741-745. doi: 10.1212/wnl.48.3.741.
6. Wakutani Y, Watanabe K, Adachi Y, Wada-Isoe K, Urakami K, Ninomiya H, Saido TC, Hashimoto T, Iwatsubo T, Nakashima K: Novel amyloid precursor protein missense mutation (D678N) in familial Alzheimer’s disease. J Neurol Neurosurg Psychiatr 2004; 75, 1039-1042. doi: 10.1136/jnnp.2003.010611.
7. Tomita T, Takikawa R, Koyama A, Morohashi Y, Takasugi N, Saido TC, Maruyama K, Iwatsubo T: C terminus of presenilin is required for overproduction of amyloidogenic Aβ42 through stabilization and endoproteolysis of presenilin. J Neurosci 1999; 19, 10627-10634. doi: 10.1523/JNEUROSCI.19-24-10627.1999.
8. Takasugi N, Tomita T, Hayashi I, Tsuruoka M, Niimura M, Takahashi Y, Thinakaran G, Iwatsubo T: The role of presenilin cofactors in the γ-secretase complex. Nature 2003; 422, 438-441. doi: 10.1038/nature01506.
9. Iwatsubo T, Iwata A, Suzuki K, Ihara R, Arai H, Ishii K, Senda M, Ito K, Ikeuchi T, Kuwano R, Matsuda H, Sun CK, Beckett LA, Petersen RC, Weiner MW, Aisen PS, Donohue MC: Japanese and North American Alzheimer’s Disease Neuroimaging Initiative studies: harmonization for international trials. Alzheimers Dementia 2018; 14, 1077-1087. doi: 10.1016/j.jalz.2018.03.009.
10. Mano T, Sato K, Ikeuchi T, Toda T, Iwatsubo T, Iwata A: Peripheral blood BRCA1 methylation positively correlates with major Alzheimer’s disease risk factors. J Prev Alzheimers Dis 2021; doi: 10.14283/jpad.2021.31
11. Senda M, Ishii K, Ito K, Ikeuchi T, Matsuda H, Iwatsubo T, Iwata A, Ihara R, Suzuki K, Kasuga K, Ikari Y, Niimi Y, Arai H, Tamaoka A, Arahata Y, Itoh Y, Tachibana H, Ichimiya Y, Washizuka S, Odawara T, Ishii K, Ono K, Yokota T, Nakanishi A, Matsubara E, Mori H, Shimada H: A Japanese multicenter study on PET and other biomarkers for subjects with potential preclinical and prodromal Alzheimer’s disease. J Prev Alzheimers Dis 2021; doi: 10.14283/jpad.2021.37
12. Sperling RA, Donohue MC, Raman R, Sun CK, Yaari R, Holdridge K, Siemers E, Johnson KA, Aisen PS. Association of factors with elevated amyloid burden in clinically normal older individuals. JAMA Neurol 2020; 77, 735-745. doi: 10.1001/jamaneurol.2020.0387.
13. Sato K, Mano T, Ihara R, Suzuki K, Toda T, Iwatsubo T, Iwata A: Cohort-specific optimization of models predicting preclinical Alzheimer’s disease, to enhance screening performance in the middle of preclinical Alzheimer’s disease clinical studies. J Prev Alzheimers Dis 2021; doi: 10.14283/jpad.2021.37
14. Sato K, Ihara R, Suzuki K, Niimi Y, Toda T, Jimenez-Maggiora G, Langford O, Donohue MC, Raman R, Aisen PS, Sperling RA, Iwata A, Iwatsubo T: Predicting amyloid risk by machine learning algorithms based on the A4 screen data: application to the Japanese Trial-Ready Cohort study. Alzheimers Dementia: Translational Research & Clinical Interventions 2021; 7, e12135. doi: 10.1002/trc2.12135.
15. Ngandu T, Lehtisalo J, Solomon A, Levalahti E, Ahtiluoto S, Antikainen R, Backman L, Hanninen T, Jula A, Laatikainen T, Lindstrom J, Mangialasche F, Paajanen T, Pajala S, Peltonen M, Rauramaa R, Stigsdotter-Neely A, Strandberg T, Tuomilehto J, Soininen H, Kivipelto M. A 2 year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at-risk elderly people (FINGER): a randomised controlled trial. Lancet 2015;385, 2255-2263. doi: 10.1016/S0140-6736(15)60461-5.
16. Sugimoto T, Sakurai T, Akatsu H, Doi T, Fujiwara Y, Hirakawa A, Kinoshita F, Kuzuya M, Lee S, Matsuo K, Michikawa M, Ogawa S, Otsuka R, Sato K, Shimada H, Suzuki H, Suzuki H, Takechi H, Takeda S, Umegaki H, Wakayama S, Arai H: The Japan-Multimodal Intervention Trial for Prevention of Dementia (J-MINT): The Study Protocol for an 18-Month, Multicenter, Randomized, Controlled Trial. J Prev Alzheimers Dis 2021; 10.14283/jpad.2021.29
17. Ochiai T, Nagayama T, Matsui K, Amano K, Sano T, Wakabayashi T, Iwatsubo T: Tauroursodeoxycholic acid attenuates diet-induced and age-related peripheral endoplasmic reticulum stress and cerebral amyloid pathology in a mouse model of Alzheimer’s disease. J Prev Alzheimers Dis 2021; doi. 10.14283/jpad.2021.33



M. Senda1, K. Ishii2, K. Ito3, T. Ikeuchi4, H. Matsuda5, T. Iwatsubo6, A. Iwata7, R. Ihara7, K. Suzuki8, K. Kasuga4, Y. Ikari1,9, Y. Niimi6, H. Arai10, A. Tamaoka11, Y. Arahata3, Y. Itoh12, H. Tachibana13, Y. Ichimiya14, S. Washizuka15, T. Odawara16, K. Ishii17, K. Ono18, T. Yokota19, A. Nakanishi20, E. Matsubara21, H. Mori12, H. Shimada12


1. Kobe City Medical Center General Hospital, Japan; 2. Tokyo Metropolitan Institute of Gerontology, Japan; 3. National Center for Geriatrics and Gerontology, Japan; 4. Niigata University, Japan; 5. National Center of Neurology and Psychiatry, Japan; (currently, Southern Tohoku Drug Development and Cyclotron Research Center, Japan); 6. The University of Tokyo, Japan; 7. The University of Tokyo, Japan; (currently, Tokyo Metropolitan Geriatric Hospital, Japan); 8. The University of Tokyo, Japan; (currently, National Defense Medical College, Japan); 9. Osaka University, Japan; 10. Tohoku University, Japan; 11. University of Tsukuba, Japan; 12. Osaka City University, Japan; 13. Kobe University, Japan; 14. Juntendo Tokyo Koto Geriatric Medical Center, Japan; 15. Shinshu University, Japan; 16. Yokohama City University, Japan; 17. Kindai University, Japan; 18. Showa University, Japan; 19. Tokyo Medical and Dental University, Japan; 20. Osaka City Kosaiin Hospital, Japan; 21. Oita University, Japan

Corresponding Author: Michio Senda, Division of Molecular Imaging Research Kobe City Medical Center General Hospital (KCGH), 2-1-1 Minatojima-Minamimachi, Chuo-ku, Kobe 650-0047 Japan, E-mail:, Phone: 81-78-304-5212, Fax: 81-78-304-5201.



BACKGROUND: PET (positron emission tomography) and CSF (cerebrospinal fluid) provide the “ATN” (Amyloid, Tau, Neurodegeneration) classification and play an essential role in early and differential diagnosis of Alzheimer’s disease (AD).
OBJECTIVE: Biomarkers were evaluated in a Japanese multicenter study on cognitively unimpaired subjects (CU) and early (E) and late (L) mild cognitive impairment (MCI) patients.
MEASUREMENTS: A total of 38 (26 CU, 7 EMCI, 5 LMCI) subjects with the age of 65-84 were enrolled. Amyloid-PET and FDG-PET as well as structural MRI were acquired on all of them, with an additional tau-PET with 18F-flortaucipir on 15 and CSF measurement of Aβ1-42, P-tau, and T-tau on 18 subjects. Positivity of amyloid and tau was determined based on the positive result of either PET or CSF.
RESULTS: The amyloid positivity was 13/38, with discordance between PET and CSF in 6/18. Cortical tau deposition quantified with PET was significantly correlated with CSF P-tau, in spite of discordance in the binary positivity between visual PET interpretation and CSF P-tau in 5/8 (PET-/CSF+). Tau was positive in 7/9 amyloid positive and 8/16 amyloid negative subjects who underwent tau measurement, respectively. Overall, a large number of subjects presented quantitative measures and/or visual read that are close to the borderline of binary positivity, which caused, at least partly, the discordance between PET and CSF in amyloid and/or tau. Nine subjects presented either tau or FDG-PET positive while amyloid was negative, suggesting the possibility of non-AD disorders.
CONCLUSION: Positivity rate of amyloid and tau, together with their relationship, was consistent with previous reports. Multicenter study on subjects with very mild or no cognitive impairment may need refining the positivity criteria and cutoff level as well as strict quality control of the measurements.

Key words: Alzheimer’s disease, PET, CSF biomarker, amyloid, tau.




Early and differential diagnosis of Alzheimer’s disease (AD) has been drawing more and more attention these days as the target population of the therapeutic trials has shifted toward the early phases of the AD continuum. Biomarkers including PET, MRI and cerebrospinal fluid(CSF)/plasma play an essential role in such early phases, where clinical manifestation and behavioral findings are limited. Jack et al (1) extracted three markers, i.e., amyloid (A), tau (T) and neurodegeneration (N), and proposed the “ATN” classification for differential diagnosis of AD continuum. PET provides imaging and quantification of amyloid and tau deposition as well as neurodegeneration evaluable with 18F-fluorodeoxyglucose(FDG)-PET. Amyloid and tau can also be evaluated with CSF sampling, and recently with plasma as well, and MRI volumetry has also been used as a marker of neurodegeneration.
In Japan, a large-scale prospective observational study called J-ADNI (Japanese Alzheimer’s Disease Neuroimaging Initiative) was completed (2), in which a total of 537 subjects were enrolled, comprising 154 cognitively unimpaired subjects (CU), 234 MCI and 149 AD patients.
Then, a new version of J-ADNI was designed by the same group, named “AMED Preclinical AD Study”, which focused on CU and MCI and acquired amyloid-PET and FDG-PET on all subjects. Part of the subjects also underwent a tau-PET scan and/or a CSF sampling. The objective of the study was to evaluate PET and MRI images and CSF biomarkers in CU and MCI subjects in Japan, compare those biomarkers between modalities to explore their reliability and usefulness in such early-phase subjects, and obtain a rough idea of the fractions of ATN-based classifications. This report summarizes the results of the study that was recently completed.




The study was a non-randomized prospective observational study, and was designed and conducted in accordance with the ethical principles as proclaimed in the Declaration of Helsinki. The study protocol was first approved by Ethical Committee of Osaka City University Graduate School of Medicine (site of leading PI) and registered as UMIN000019926, and was later re-approved by Osaka City University Hospital Certified Review Board when the Japanese Law on Clinical Research was enacted, and was registered as jRCTs051180239. The protocol was also approved by each participating site according to the Japanese regulations and ethics guidance. The tau-PET portion of the study was designed as a nominally separate add-on study when tau-PET became available, though limited, later in the course of the research project, and was approved and registered as jRCTs051190065.
The subjects were enrolled at a total of 14 clinical sites and consisted of 26 CU subjects and 12 MCI patients (7 early (E) MCI and 5 late (L) MCI as classified below) based on the neuropsychological tests.
Written informed consent was obtained from each subject and the study partner such as a family member of the subject.
The CU subjects were 65-82 years old without any memory problem and CDR-J=0. The MMSE-J score was 25-30 (higher than or equal to 24), and the delayed recall score of WMS-R logical memory (WMS-R LM II) ranged from 4 to 19 except for one subject (See footnote of Table 1).
The MCI subjects were 65-84 years old with objective persistent memory impairment reported by the study partner and CDR-J=0.5 with memory box score being 0.5 or higher. The MMSE-J score was 27-30 (higher than or equal to 24). The WMS-R LM II was used to classify the MCI subjects into EMCI (WMS-R LM II = 3-6, 5-9, 9-11) and LMCI (≤2, ≤4, ≤8) depending on the educational years (0-7, 8-15, ≥16 years, respectively), and was 6-18 for EMCI and 0-8 for LMCI.

CSF measurements and genotyping

CSF was collected from 18 subjects by lumbar puncture and stored in polypropylene tubes at -80℃ until biochemical analysis. CSF concentration of Aβ1-42 was analyzed using V-PLEX Aβ Peptide Panel 1 kit with MESO QuickPlex SQ120 (MesoScale Discovery, Rockville, MD). CSF phosphorylated tau (P-tau) and total tau (T-tau) were measured using commercially available ELISA kits, INNOTEST hTAU and PHOSPHO-TAU (181P) (Fujirebio Europe, Belgium), respectively, according to the manufacturer’s instructions. Stability of the results was monitored in the Alzheimer’s Association QC program. Cutoff values (Aβ42<378.7 pg/mL, P-tau>29.1 pg/mL, and T-tau>88.8 pg/mL) that best discriminated PiB-PET positive AD patients from PiB-PET negative CU subjects were determined using independent J-ADNI cohort (2). Because the CSF assays used in this study were different from those used in J-ADNI study, calibration between two assays were performed.
APOE genotyping (rs429358 and rs7412) was performed by Taq-Man based assay using blood samples.

MRI imaging

The brain MRI was acquired for each subject using a 3-Tesla or 1.5-Tesla scanner. The structural 3D-T1 images (MP-RAGE or IR-SPGR) were analyzed with FreeSurfer (Ver. 6.0) to measure the regional cerebral gray-matter volumes. Because the absolute volumetry depends on the version of the software and other conditions, the regional atrophy of the subject was derived as z-score using mean and SD of the baseline scan for the 26 CU subjects of this study. The volume of 8 regions in the temporal lobe (right and left entorhinal cortex, parahippocampal gyrus, hippocampus, and amygdala) were summed up and the z score was derived as a measure of the temporal lobe atrophy for each subject.

PET image acquisition

All subjects underwent an amyloid-PET and an FDG-PET. Each PET imaging site, together with the PET camera, was qualified, in which the reconstruction parameters were determined for each PET camera so that all the PET cameras satisfied the image quality criteria with the Hoffman 3D brain phantom and the uniform cylindrical phantom (3).
For amyloid PET, either 11C-PiB (PiB), 18F-florbetapir (FBP) or 18F-flutemetamol (FMM) was used for 23, 13, and 2 subjects, respectively. The injection activity was 555MBq, 370MBq, 185MBq, the uptake time (start of emission scan post injection) was 50min, 50min, 90min, and the scan duration was 20min, 20min, 30min, for PiB, FBP, and FMM, respectively.
For the FDG scan, after at least 4 hours of fasting, the subject was administered with 185 MBq of 18F-FDG in a quiet, dimly lit room while resting in a reclining chair or bed, and the subject remained in the condition until several minutes before the start of the scanning session. The PET emission data was acquired for 30 minutes starting at 30 minutes post injection.
Tau-PET was performed with 18F-flortaucipir (FTP) on 15 subjects. Because tau-PET was not ready until late in the course of the research project, the time span from amyloid to tau-PET ranged from 1.0 to 2.0 (mean 1.56) years. The subject was administered with 240.5 MBq of FTP and a 30 min emission scan started 75 min post injection.
No adverse effects were observed at the PET scans of this study.

PET image analysis

The amyloid PET images were binary interpreted visually in a blind manner by the readers who were qualified for this study, and the adjudicator (K.I.) confirmed them. The PiB images were interpreted visually using the criteria adopted in J-ADNI (4), and the FMM and FBP images were interpreted with each vendor’s criteria.
As a quantitative analysis of the amyloid PET, mean cortical standardized uptake value ratio (mcSUVR) of PiB images was computed using the cerebellar cortex as a reference based on the method of J-ADNI, and the cutoff value of 1.5 was used to determine the quantitative positivity (4). The FMM images were analyzed with CORTEX ID (GE Healthcare) to derive mcSUVR using the pons as a reference, for which the cutoff value of 0.58 was used for the quantitative positivity (5). The FBP images were analyzed with MIMneuro (MIM Software) to derive mcSUVR using the whole cerebellum as a reference, for which the cutoff value of 1.10 was used for the quantitative positivity (6).
The FDG images together with the semiquantitative 3D-SSP results were visually interpreted by three independent readers followed by a consensus read in the same way as J-ADNI (7), and the images were classified into N1 (normal), N2 (reflecting atrophy), N3, P1 (AD pattern), P2 (FTD pattern), P3, and P1+ (DLB pattern) (8). No one presented N3 or P3 in this study. The DLB pattern criteria was interpreted in a broader sense to include cases with occipital hypometabolism extending to neighboring areas even if typical temporoparietal hypometabolism was not observed. The FDG images were also quantified with AD t-sum (9) using the module PALZ in the PMOD software package (Ver. 3.2; PMOD Technologies, Zurich, Switzerland), which were then converted into PET score [10] that reflects the severity of temporoparietal hypometabolism (AD pattern).
The FTP images were interpreted and classified into AD negative, AD+ and AD++, according to the vendor’s criteria that regards cortical uptake except anterior temporal as AD-related ( The FTP-PET was also analyzed with MUBADA-PERSI method to derive SUVR over the area affected by AD process (posterior temporal, occipital, parietal and part of frontal cortex) with white matter as a reference (11, 12).

Follow up

Whenever possible, each subject was followed up every year with a general clinical interview with neuropsychological tests, an MRI scan, and an FDG-PET scan. As a result, 1-year follow-up data were acquired on 32 subjects, and 2-year follow-up on 5 subjects.

Statistical methods

Because the number of subjects was small, descriptive results were presented in general. Proportion of positivity was compared between groups using chi-square tests, in which EMCI and LMCI were combined to increase the number of observations. Statistical tests were also performed on the Pearson correlation coefficient between two variables.



Findings of each subject

Table 1 describes findings of each subject as classified according to the ATN concept. In this study, amyloid (A) was interpreted as positive (A+) when either PET or CSF Aβ was positive. Tau (T) was interpreted as positive (T+) when either PET or CSF P-tau was positive; negative (T-) when either of them was obtained and neither of them were positive; and was “na” (not available) (Tna) when neither of them were obtained. Neurodegeneration (N) was interpreted as positive (N+) when the consensus visual read of FDG-PET showed a progressive pattern (P1, P2, or P1+), and negative (N-) when it was a non-progressive pattern (N1 or N2).

Table 1. Findings for each subject and ATN classification


The amyloid positivity rate was 13/38 overall (6/26 CU, 4/7 EMCI, 3/5 LMCI, p>0.05 between CU and MCI), while it was 8/38 based on the PET alone (3/26 CU, 2/7 EMCI, 3/5 LMCI, p>0.05 between CU and MCI).
Tau was positive for 7, negative for 2 and not available for 4 out of the 13 A+ subjects, being 2, 2, 2 and 5, 0, 2 out of the 6 A+ CU and 7 A+ MCI subjects, respectively.
FDG-PET showed a progressive pattern in 6/13 A+ subjects (3/6 CU, 1/4 EMCI, 2/3 LMCI) as compared to 3/25 A- subjects (0/20 CU, 1/3 EMCI, 2/2 LMCI). Significant difference was observed in the FDG-PET positivity (N+) proportion between A+ and A- (p<0.05) as well as between CU (3/26) and MCI (6/12) (p<0.05).
Of interest, tau was positive for as many as 8 (negative for 8, not available for 9) out of the 25 amyloid negative subjects, indicating tau deposition without AD pathological process. It should be noted that all the 8 A-T+ subjects was tau positive due to CSF test, in spite of negative tau PET for two of them.
Association of APOE genotypes with amyloid PET (p>0.6) or CSF Aβ (p>0.5), or with any other biomarkers, was not observed for the presence of E4, probably due to the small number of subjects.

Representative cases

Figures 1 (#24, LMCI) and 2 (#22, CU) depict a case with prodromal AD (A+T+N+) and preclinical AD (A+T+N-), respectively. PET and CSF were discordant for “A” and/or “T” in both cases, which may be related to visually equivocal images and near-cutoff level quantified values. In the case of Figure 2, CSF P-tau was positive while tau PET was negative, consistent with the report of earlier and more sensitive positivity of CSF P-tau than tau-PET in the AD continuum (13).
Four cases (1 CU, 1 EMCI, 2 LMCIs) showed a mild/partial DLB pattern in FDG-PET marked with “P1+” in Table 1, featuring hypometabolism in the occipital cortex extending into surrounding areas but not showing a typical AD pattern of temporo-parietal hypometabolism. Amyloid was positive for 3/4 and tau was positive for 4/4. Figure 3 (#26, CU) depicts one of them.

Figure 1. PiB, FTP and FDG-PET of a female LMCI patient in her 70s (#24) interpreted as prodromal AD

Amyloid PET with PiB was visually negative, as the left parietal mild accumulation did not reach the cortical surface (arrow). However, the subject was classified as “A+” because quantitative analysis revealed SUVR (1.57) above cutoff. The CSF Aβ was negative (399.8 pg/mL). The FTP-PET showed abnormal tau accumulation in the left posterior temporal lobe (arrow), typical of AD process. Note off-target uptake of FTP in choroid plexus (arrowheads), substantia nigra, and striatum. The FDG-PET was read as temporo-parietal hypometabolism indicating AD pattern in the baseline that progressed in two years (arrows). PET score and MRI z-score also increased in two years: from 0.76 to 1.08 and from 2.7 to 3.2, respectively.

Figure 2. PiB, FTP and FDG-PET of a female CU subject of her 70s (#22) interpreted as preclinical AD

PiB-PET revealed positive amyloid accumulation in the left temporal and parietal areas (arrows). Tau PET with FTP acquired 1 year later was negative, because mild activity along the cortical rim was interpreted as off-target uptake by the meninges (short arrows) and that the left anterior temporal uptake was considered non-pathological within the AD continuum (long arrow). CSF P-tau was positive. FDG-PET showed a normal pattern.

Figure 3. FBP, FTP and FDG-PET of a female CU subject of her 70s (#26)

FBP-PET presented negative amyloid, and tau was negative in FTP-PET, although CSF showed positive Aβ (317.9pg/mL) and P-tau (38.2pg/mL). FDG-PET revealed a DLB pattern, presenting occipital hypometabolism (long arrows) extending to the right temporal and parietal cortex (short arrows), which progressed 1 year later. Note cingulate island sign denoting preserved metabolism in the posterior cingulate cortex (arrowheads).


Association between PET and CSF

For the 18 subjects, in which CSF data were obtained, amyloid positivity by CSF agreed with that by PET in 12 cases while 6 showed a discordance (Table 1). The rate of discordance was consistent with previous reports and may be caused by various factors (13).
Quantified tau uptake (SUVR) measured with FTP-PET using MUBADA-PERSI method was significantly correlated with CSF P-tau (r=0.92, p<0.001, n=8) (Figure 4). Although the cutoff for SUVR with MUBADA-PERSI SUVR is not established yet, the visual read of the FTP-PET was positive only for two of them. In CSF P-tau, however, 7 out of the 8 subjects showed P-tau above the cutoff level, indicating a discordance in the tau positivity between PET and CSF. This is consistent with recent investigations that reported earlier or more sensitive positivity of CSF P-tau than tau-PET in the AD continuum (i.e. in amyloid positive subjects), because secretion of soluble p-tau to CSF is increased by Aβ pathology before tau begins to accumulate in the brain (14).

Figure 4. Scatter plot of tau uptake (SUVR) quantified with FTP-PET and CSF P-tau

Red marks indicate PET-positive cases by visual read. Arrow indicates cutoff value for CSF P-tau.



PET/CSF discordance for amyloid and tau

This study suffers limitations such as the small number of subjects, poor follow-up records, and lack of tau-PET and CSF measurement for a large fraction of the subjects. However, some findings are notable.
The rate of amyloid positivity based on the combination of PET and CSF (6/26=23% for CU, 7/12=58% for MCI) was consistent with previous reports including J-ADNI. Discordance of positivity between amyloid-PET and CSF Aβ was observed in 6 subjects (5 PET-/CSF+, 1 PET+/CSF-), suggesting higher sensitivity of CSF, which was also consistent with the ADNI data on CU and MCI (13).
The rate of tau positivity was 4/6 for amyloid PET-/CSF Aβ+ or amyloid PET+/CSF Aβ-, and 2/2 for amyloid PET+/CSF Aβ+ in this study (Table 1). This was agreeable with the above ADNI data, in which the former two groups presented significantly lower CSF p-tau and PET-measured tau deposition than the latter and suggested earlier manifestations of AD process (13).
It is known that CSF p-tau is quantitatively associated with PET-measured tau deposition, especially in the AD continuum, and that CSF p-tau rises in the earlier stage than the pathological uptake of FTP-PET accrues (14). The result of the present study indicated a similar association in spite of the small number of subjects (Figure 4), in which FTP-PET was quantified with MUBADA, and that CSF p-tau was more sensitive than visual read of FTP-PET. In the early phase of AD continuum, CSF p-tau and FTP-PET could be considered to reflect different pathological changes, as the former may indicate excess secretion of p-tau to CSF and the latter represents tissue tau deposition. The present study adopted the criteria of visual FTP-PET interpretation to determine the positivity, by which the anterior temporal FTP uptake was considered insignificant. Because MUBADA VOI covers wide cortical areas, it may not be a sensitive measure of early tau deposition in the AD continuum that begins in the temporal cortex. Although a recent study suggests the earliest tau deposition at the rhinal cortex located in the anteromedial temporal lobe (15), it was hard to quantify the pathological FTP uptake in that region due to off-target binding to the choroid plexus in the present study (Figure 2). In that sense, use of FTP was another limitation of this study. Tau PET drugs with little off-target binding such as 18F-MK-6240 or 18F-PI-2620 may be more suitable for evaluation of the earliest stage of AD.

Neurodegeneration marker

In this study only visual assessment of FDG-PET was used to evaluate the “N” (neurodegeneration) marker, and quantitative PET score was not used so that one “N-” subject presented a high PET score (#4). Although CSF T-tau and MRI-volumetry are also regarded as an N-marker, their association with FDG-PET remains to be investigated as they represent different pathophysiology.
Hypometabolism depicted by FDG-PET reflects reduced neuronal activity in general, regardless of pathophysiology. No subjects showed A+T-N+ in this study, which agrees with the concept of tau deposition leading to neurodegeneration in AD continuum, although such manifestation, if occurred, might have suggested combined AD and non-AD processes. Outside the AD continuum (A-), however, FDG-PET neurodegeneration was positive in 3/5 MCI (2/2 LMCI) subjects and was not observed in CU subjects (0/20), which is consistent with the above notion and agrees with previous reports (16).

Binary criteria

A large number of subjects presented quantitative measures and/or visual read that are close to the borderline of binary positivity in the present study, which caused, at least partly, discordance between PET and CSF in amyloid and/or tau, such as the cases in Figures 1 and 2. This is understandable because most of the subjects in the present study were in the early phase of AD continuum or of a non-AD disease if any, and that the current criteria and cutoff level have been derived from differential diagnosis of AD patients from CU subjects. To deal with early-phase subjects having no or very mild cognitive impairment, the criteria and cutoff level of the biomarkers may need refining, and the data acquisition may need strict quality control.

Non-AD disorders

The present study revealed 8 A-T+ subjects. Because two of them (#37, #38) were LMCI patients and showed AD or DLB pattern in FDG-PET, they are considered SNAP and to have cognitive impairment due to non-AD disorders (17). The other six are CU subjects and had non-progressive FDG pattern, and may suggest a very early stage of various non-AD tauopathy such as PART or normal aging process (17). There is also a possibility of false-positive CSF P-tau as the value was close to the cutoff, ranging 31.8-38.2 pg/mL for 5 of the 6 subjects. Since biomarkers in non-AD tauopathies are not well understood, further investigations are needed.
Another EMCI subject (#36) presented “A-T-N+” with FTD pattern in FDG-PET, and was suspected of early stage of non-amyloid non-tau FTLD.
The present study also revealed 4 subjects presenting DLB pattern in FDG-PET, with occipital hypometabolism extending to surrounding areas (Figure 3). Three of them were amyloid positive and all were tau positive. It is not clear whether they were preclinical or prodromal stage of atypical AD, or DLB with or without amyloid deposition.
These findings suggest that a significant fraction of the subjects in this study might be related to non-AD disorders such as DLB, SNAP, PART, argyrophilic grain disease (AGD), and TDP-43 proteinopathy (like LATE) (17). Even if they are amyloid positive, there is a possibility of incidental amyloid deposition. Therefore, possibility of non-AD disorders should always be considered when clinical trials targeting preclinical or prodromal AD are designed and subjects are selected based on the biomarkers.



In conclusion, this study confirmed the known changes of PET and CSF biomarkers in preclinical and prodromal AD, and at the same time, suggested difficulties of determining the criteria and cutoff level of those biomarkers to evaluate such subjects as well as the possibility of unsolicited inclusion of early-phase non-AD disorders.


Acknowledgements: We are grateful for the materials and technical supports for the PET imaging by Fujifilm Toyama Chemical, Avid Radiopharmaceuticals/Eli Lilly Japan, and GE Healthcare. PET centers that imaged the subjects but did not belong to the clinical site that enrolled the subjects are also acknowledged, including Tohoku University Cyclotron and Radioisotope Center (CYRIC), Tsukuba Advanced Imaging Center (AIC), Tokyo Metropolitan Institute of Gerontology (TMIG), Aizawa Hospital, MI Clinic, and Kobe City Medical Center General Hospital (KCGH). We thank all the people who participated in this study in the clinical and imaging sites as well as in the Core sites.

Conflict of interest: The following conflicts of interest are disclosed by the authors. Senda reports provision of devices, cassettes, and precursors from Avid/Eli Lilly Japan and GE, funding as PI of clinical trials sponsored by Eli Lilly, Eisai, Biogen, Cerveau and Merck, as well as leadership role in the Japanese Society of Nuclear Medicine as board member, congress chair and committee chair. Ikeuchi reports grants from AMED (JP19dk0207020, JP20dk0202028, JP20dm0207073). Matsuda reports a grant from AMED (19dk0207020h0005), intramural grants from National Center of Neurology and Psychiatry, and an entrusted research fund from Nihon Medi-Physics Co. Ltd. Iwatsubo reports a grant from an anonymous Foundation. Iwata reports grants from AMED (19dk0207020h0005, 16dk0207028h0001). Ikari is a full time employee of CMIC Inc. as well as graduate student of Osaka University. Washizuka reports research funding from AMED and pharmaceutical companies including Otsuka, Eisai, Pfizer, Daiichi-Sankyo, Tsumura, Mochida, Astellas, Shionogi, Takeda, Sumitomo-Dainippon, as well as honoraria from such pharmaceutical companies. Kazunari Ishii reports honoraria from Nihon Medi-Physics. Yokota reports licensing and collaboration research with Takeda Pharmaceutical Company. Nakanishi reports research funding from Eisai and Elli Lilly Japan as well as leadership role as a director in the Japan Society for Dementia Research. Shimada reports grants from AMED (19dk0207020h0005, 20dk0207028h0005). The other authors have nothing to disclose.

Funding: This study was financially supported by grants from Japan Agency for Medical Research and Development (AMED) 19dk0207020h0005, 20dk0202028h0005 and 20dm0207073h003, as well as by an anonymous Foundation.



1. Jack CR, Jr., Bennett DA, Blennow K, et al. A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers. Neurology 2016;87:539-547.
2. Iwatsubo T, Iwata A, Suzuki K, et al. Japanese and North American Alzheimer’s Disease Neuroimaging Initiative studies: Harmonization for international trials. Alzheimers Dement 2018;14:1077-1087.
3. Ikari Y, Akamatsu G, Nishio T, et al. Phantom criteria for qualification of brain FDG and amyloid PET across different cameras. EJNMMI physics 2016;3:23.
4. Yamane T, Ishii K, Sakata M, et al. Inter-rater variability of visual interpretation and comparison with quantitative evaluation of (11)C-PiB PET amyloid images of the Japanese Alzheimer’s Disease Neuroimaging Initiative (J-ADNI) multicenter study. Eur J Nucl Med Mol Imaging 2017;44:850-857.
5. Thurfjell L, Lilja J, Lundqvist R, et al. Automated quantification of 18F-flutemetamol PET activity for categorizing scans as negative or positive for brain amyloid: concordance with visual image reads. J Nucl Med 2014;55:1623-1628.
6. Clark CM, Pontecorvo MJ, Beach TG, et al. Cerebral PET with florbetapir compared with neuropathology at autopsy for detection of neuritic amyloid-β plaques: a prospective cohort study. The Lancet Neurology 2012;11:669-678.
7. Yamane T, Ikari Y, Nishio T, et al. Visual-statistical interpretation of (18)F-FDG-PET images for characteristic Alzheimer patterns in a multicenter study: inter-rater concordance and relationship to automated quantitative evaluation. AJNR Am J Neuroradiol 2014;35:244-249.
8. Silverman DHS, Small GW, Chang CY, et al. Positron Emission Tomography in Evaluation of DementiaRegional Brain Metabolism and Long-term Outcome. JAMA 2001;286:2120-2127.
9. Herholz K, Salmon E, Perani D, et al. Discrimination between Alzheimer dementia and controls by automated analysis of multicenter FDG PET. NeuroImage 2002;17:302-316.
10. Herholz K, Westwood S, Haense C, Dunn G. Evaluation of a calibrated (18)F-FDG PET score as a biomarker for progression in Alzheimer disease and mild cognitive impairment. J Nucl Med 2011;52:1218-1226.
11. Southekal S, Devous MD, Sr., Kennedy I, et al. Flortaucipir F 18 Quantitation Using Parametric Estimation of Reference Signal Intensity. J Nucl Med 2018;59:944-951.
12. Devous MD, Sr., Joshi AD, Navitsky M, et al. Test-Retest Reproducibility for the Tau PET Imaging Agent Flortaucipir F 18. J Nucl Med 2018;59:937-943.
13. Reimand J, Collij L, Scheltens P, Bouwman F, Ossenkoppele R, Alzheimer’s Disease Neuroimaging I. Association of amyloid-beta CSF/PET discordance and tau load 5 years later. Neurology 2020;95:e2648-e2657.
14. Mattsson-Carlgren N, Andersson E, Janelidze S, et al. Aβ deposition is associated with increases in soluble and phosphorylated tau that precede a positive Tau PET in Alzheimer’s disease. Science advances 2020;6:eaaz2387.
15. Sanchez JS, Becker JA, Jacobs HIL, et al. The cortical origin and initial spread of medial temporal tauopathy in Alzheimer’s disease assessed with positron emission tomography. Science translational medicine 2021;13:eabc0655.
16. Coutinho AM, Busatto GF, de Gobbi Porto FH, et al. Brain PET amyloid and neurodegeneration biomarkers in the context of the 2018 NIA-AA research framework: an individual approach exploring clinical-biomarker mismatches and sociodemographic parameters. Eur J Nucl Med Mol Imaging 2020;47:2666-2680.
17. Jack CR, Jr., Knopman DS, Chételat G, et al. Suspected non-Alzheimer disease pathophysiology–concept and controversy. Nature reviews Neurology 2016;12:117-124.



I. Choi1, H. La Monica1, S.L. Naismith2, A. Rahmanovic2, L. Mowszowski2, N. Glozier1


1. Central Clinical School, Brain and Mind Centre, Faculty of Medicine and Health, University of Sydney, Australia; 2. Charles Perkins Centre, Faculty of Science, School of Psychology and the Brain and Mind Centre, University of Sydney, Australia

Corresponding Author: Dr Isabella Choi, 94 Mallett Street, Camperdown, NSW 2050, Australia,, +612 8627 7240.

J Prev Alz Dis 2021;
Published online June 23, 2021,



Communicating personal Alzheimer’s disease risk profiles based on validated risk algorithms may improve public knowledge about risk reduction, and initiate action. This proof of concept pilot trial aimed to test whether this is feasible and potentially effective and/or harmful. Older at-risk adults (N=24) were provided with their personal Alzheimer’s disease risk profile online, which contained information on their personal risk level, scores and tailored recommendations to manage modifiable risk factors. After receiving the risk profile, participants were significantly more accurate in identifying risk and protective factors, and revised their perceived risk to be lower than their initial estimate. There was no apparent harm seen in psychological distress or dementia-related worry. This shows preliminary support for the feasibility of delivering personal dementia risk profiles to low risk, help-seeking older adults in an online format. A definitive trial examining behavioural outcomes and testing in groups with higher risk profiles is now warranted.

Key words: Risk communication, health literacy, psychological distress, prevention, Alzheimer’s disease.



Up to a third of Alzheimer’s disease cases can be prevented through improved education and reduction of modifiable risk factors (1). Growing evidence from multi-domain interventions shows that targeting modifiable risk factors can reduce risk of Alzheimer’s disease (AD) and improve cognition (2, 3). However, lack of knowledge about dementia and its risk factors among the public is a major barrier to individuals implementing behavioural and lifestyle change and, in turn, to dementia risk reduction (4).
Having an accurate understanding of one’s personal risk of future disease is considered essential for engaging in behaviours for risk reduction. Most health behaviour change models, including the Health Belief Model, identify four major constructs that surround health behaviour: health literacy, perceived susceptibility, motivation to change, and perceived barriers to change (5). However, there is poor health literacy of dementia risk, risk factors, and prevention strategies among the public. A systematic review found that almost half of the general public agreed that dementia was a normal and non-preventable part of ageing (6). Mental activity, healthy diet, physical exercise and social engagement were the most commonly nominated ways to reduce risk, but other well established risk factors such as vascular risk (including smoking, high blood pressure, high cholesterol, obesity), low education, poor mental health, brain trauma, and environmental toxins were rarely mentioned (6-8). Only around 25% of Australians were confident they could reduce risk (8). However, there is support that improving dementia health literacy has a positive impact on risk reduction. People who had a strong belief that dementia risk could be reduced, had moderate to high knowledge of risk-reduction behaviours, or had high confidence that risk reduction can be achieved were almost twice as likely to take action to reduce risk (9).
Communicating personal dementia risk level and risk factors to at-risk adults, based on validated AD risk algorithms, may be one way to improve dementia risk knowledge and engage people in risk reduction behaviours. A number of dementia risk algorithms have been validated for the general population and can identify those with high risk with acceptable predictive ability (10). For instance, the Australian National University Alzheimer’s Disease Risk Index (ANU-ADRI) includes various modifiable risk factors that have been validated for middle-age and older adults that are easily assessed via self-report (11, 12). These evidence-based algorithms, along with personalised risk factor feedback and recommendations to reduce risk, can be delivered online for wide access, allowing people to screen for their risk without having had to first consult with a physician. Older adults are able to use technology proficiently and over 90% of help-seeking older adults with some degree of cognitive impairment use the Internet at home, most commonly for emails (13). Such online dementia risk algorithms already exist and are made freely available to the public, for instance, the CAIDE Risk Score App allows users to detect their dementia risk, obtain information on modifiable risk factors, and receive suggestions on how to modify their risk (14).
Surveys have found that there is high interest among older adults in knowing their risk of AD (15, 16) While there may be potential benefits to disclosing risk, there are also concerns that this could cause negative psychological outcomes. For instance, about 30% of older adults who were interested in knowing their risk also actively worried about developing AD (17). Evidence from a systematic review suggests that disclosure of increased AD risk was not associated with anxiety or depression, but did lead to heightened test-related distress, long-term care insurance uptake and health-related behaviour changes (18).
This pilot trial explores the feasibility and acceptability of communicating online personal dementia risk profiles to at-risk older adults and the impact on dementia health literacy, motivation to engage in behaviour change, and potential harmful psychological effects.




This is an uncontrolled proof of concept pilot trial. The study was approved by the University of Sydney Human Research Ethics Committee (protocol number: 2019/669) and registered with the Australian New Zealand Clinical Trials Registry (ANZCTRN12619001242112p).


People attending the Healthy Brain Ageing Clinic, a specialised memory research clinic for people aged 50 years or older in Sydney, Australia, who were clinically diagnosed with mild cognitive impairment (MCI) or subjective cognitive decline (SCD) between October 2019 to May 2020 were recruited. Those who had dementia or pre-existing severe cognitive impairment due to neurological conditions, psychosis, intellectual disability, substance misuse, stroke, or acquired brain injury were excluded.
As part of standard clinic procedure, all clinic attendees completed self-report measures and were assessed by a geriatrician or neurologist, a psychologist, and a neuropsychologist for a review of medical and psychiatric history, mood, and cognitive functioning. Diagnoses were rated according to consensus, including at least two neuropsychologists and one specialist, and were used to exclude those with a dementia diagnosis and other exclusion criteria. Within two weeks of the clinic assessment, attendees received neuropsychological test result feedback over the phone from a neuropsychologist. The clinic does not provide treatment but refers attendees to suitable clinical investigations, e.g. sleep studies, if warranted, as well as clinical trials for which they are deemed eligible.
After receiving neuropsychological test feedback from a neuropsychologist, eligible attendees were invited via a telephone call to participate in the current study. Interested people had to have an email account and were emailed a survey link via REDCap (Research Electronic Data Capture), a web-based research management platform, with the Participant Information Statement and Consent Form. Participants gave consent to extract relevant data collected from their recent clinic assessment to populate their personal dementia risk profile.


Participants completed self-report baseline measures online via REDCap. Participants’ demographic and risk information were extracted from their standard clinic assessment to compile their personal dementia risk profile using the ANU-ADRI (11). Risk factors in the model included: age, gender, highest level and total number of years of education completed, body mass index, diabetes, depression, high cholesterol, traumatic brain injury, smoking status, alcohol intake, social engagement, physical activity, cognitive activity, and diet.
Within two weeks of completing the baseline self-report measures, participants were emailed a pdf document with their personal dementia risk profile. The risk profile contained standard information about dementia, an explanation of their personal dementia risk profile, and information about the ANU-ADRI risk model. Participants were presented with a visual representation of their risk level in the form of a thermometer showing their risk from 0 to 100, along with an explainer “Your risk of developing dementia is low/ moderate/ high. It is estimated that XX out of 100 people with your risk factors will develop dementia in their lifetime” (Figure 1). They were reminded that this is an estimate based on their risk factors rather than a definitive guarantee, and that there are some risk factors they cannot change but some they could potentially work on to reduce their risk. Participants also received a summary of the dementia risk factors included in the risk model and their scores on each risk factor. They were told why the risk factor was important for brain functioning and were given tailored recommendations to manage each risk factor based on their risk factor score, as well as links for more information.
One week after receiving their risk profile, participants received an email asking them to complete the online post-intervention measures in REDCap. After completing all study measures, participants were reimbursed with a $20 gift card in return for their time.

Figure 1. Example of the risk level and risk factor feedback provided in the personal risk profile




Primary outcome: Dementia health literacy

Participants were asked “How likely do you think that you will get Alzheimer’s disease in your lifetime?” to assess perceived risk on a scale, where 0%=certain not to happen and 100%=certain to happen. To examine accuracy of perceived risk, the participant’s perceived risk was subtracted from their ANU-ADRI risk estimate. We recoded the difference (D) into a categorical variable, with <−10% indicating overestimation, >10% indicating underestimation, and accurate if −10% ≤ D ≤ 10%, in accordance with previous studies (19). Similarly, to examine change in perceived risk, participants’ perceived risk at post-intervention was subtracted from their perceived risk at baseline (d), and <-10% indicates increased perceived risk, >10% indicates reduced perceived risk, and −10% ≤ d ≤ 10% indicates no change.
We adapted the dementia risk and protective factors questionnaire in the MijnBreincoach survey (20) to assess for knowledge of dementia risk factors. We included five additional modifiable risk and protective factors that were identified in the ANU-ADRI (11) and the Lancet Commission for dementia prevention (21) (i.e. traumatic brain injury, social activity, sleep, education, and age), totalling 19 risk factors. Additional questions asked about barriers to improving brain health, confidence in risk reduction (8), and worry about getting dementia (7).

Secondary outcome: Motivation to Change Lifestyle and Health Behaviours for Dementia

The Motivation to Change Lifestyle and Health Behaviours for Dementia Risk Reduction (MCLHB-DRR) Scale is designed to assess beliefs and attitudes about lifestyle and health behavioural changes for dementia risk reduction among middle-aged and older adults (22). The scale includes (27) items matched onto seven subscales that reflect the seven concepts of the Health Behaviour Model. All items are rated on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). The scale has moderate to high internal consistencies for the seven subscales, and moderate test-retest reliability (18). Cronbach’s alpha for each of the subscales in this study were: perceived susceptibility (.916), perceived severity (.331), perceived benefits (.715), perceived barriers (.878), cues to action (.656), general health motivation (.638), and self-efficacy (.615).

Secondary outcome: Psychological distress

The K10 is a commonly used screening scale for non-specific psychological distress validated for use in Australia (23). The K10 has also been demonstrated as having moderate sensitivity to symptom change in an Australian sample (24). Scores on the K10 range from 10 to 50, and a score of 30 or more indicates a severe level of distress. Cronbach’s alpha in this study was 0.841.

Secondary outcome: Dementia-related worry

The Dementia Worry Scale was used to assess dementia-related worry (25). It has strong internal consistency and test-retest reliability. It consists of 12 items with scores ranging from 15 to 60. Cronbach’s alpha in this study was 0.908.

User evaluation

We adapted a five-point scale (from 1= not at all to 5= completely) previously used to assess user experience of a dementia information website (26). Participants were asked whether the information provided was engaging and easy to understand as well as how helpful they found the risk profile and how much they felt they had learned (from 1=nothing at all to 5=a great deal). Additionally, participants were asked if they required more information about their dementia risk profile and were given the option to discuss their experience of using the risk profile with a researcher in a telephone interview.

Data analysis

Data was analysed using SPSS version 23.0. Descriptive statistics regarding participant and baseline characteristics were analysed. Fischer’s exact tests and paired samples t-tests were used to test for differences between outcome measures pre- and post- receiving the dementia risk profile. All p-values were two-sided with an alpha of 0.05 to test for significance.



Demographics and baseline characteristics

Overall, 24 eligible participants participated in the trial (Figure 2). Participants’ ages ranged from 53 to 87, with a mean age of 69.54 years (SD 7.69). Over half of the participants (54%) were female and majority spoke English as a first language (83%). Majority were tertiary educated (75%), 21% had completed a trade certificate, and 4% had completed high school. Majority were retired (58%), 29% were employed, and 13% were unemployed. Over half were married or in a de facto relationship (58%), 25% were widowed or divorced, and 17% had never married. The majority (75%) of participants had MCI and 25% had SCD.
Almost all participants (96%; 23/24) were considered to have Low Risk of developing AD (ANU-ADRI score of less than 17), and one participant (4%) was considered as having High Risk (ANU-ADRI score of greater than 27). Participants’ perceived risk of developing AD ranged from 10-99 (M=51.63, SD=23.85), with the majority of participants overestimating their personal risk (87.5%; 21/24).

Figure 2. Participant flow

Pre-post change on dementia health literacy

All participants completed the post-intervention questionnaires. After receiving their personal dementia risk profile, participants’ perceived risk of developing AD ranged from 3-81 (M=38.83, SD=25.02). There was a significant decrease in perceived risk among the group (p = .010). A total of 18 participants (75%) still overestimated their level of risk whereas the remaining 6 correctly identified their level of risk (25%). Eleven participants (45.8%) reported a reduction in their perceived risk, eleven (45.8%) reported no change, and two participants’ (8.3%) perceived risk had increased.
The average number of correctly identified risk and protective factors increased from a mean of 11.42 items (SD= 4.50) at baseline to 13.96 items (SD= 3.98) (t1,23= -3.839, p=.001) at follow-up.

Pre-post change on motivation to engage in behaviour change and psychological effects

There was a significant reduction on the perceived susceptibility subscale of the MCLHB (t1,23=4.416, p<.001) from baseline (M=12.86; SD=3.30) to 1-week follow up (M=10.29; SD=4.21), but no change on the other subscales. There was no change on the K10 or Dementia Worry Scale.
Participants’ self-reported worry about getting dementia was significantly reduced at follow-up, from 2.75 (SD=.85) to 2.29 (SD=.91) (t1,23= 3.412, p= .002). The most common barriers to reducing risk factors at baseline were lack of knowledge (45.8%; 11/24), followed by health problems (25%; 6/24). At 1-week follow up, the most common barriers were lack of motivation (29.2%; 7/24), health problems (29.2%; 7/24), and difficulty with organisation (25%; 6/24). There was no change in confidence to take action to change risk.

User evaluation

Overall, 70.9% (17/24) of participants agreed that the information in the personal risk profile was engaging, 79.2% (19/24) agreed the information was easy to understand, 79.2% (19/24) agreed it was helpful, and 79.2% (19/24) reported they learned a good deal from their personal risk profile.
Two participants participated in the optional telephone interview with a researcher. Both expressed surprise at their lower than expected AD risk feedback, and identified difficulty addressing some of their risk factors (e.g. getting motivated to exercise). One participant agreed that seeking guidance from a health professional may support them to work on their risk factors.



This pilot study aimed to explore the feasibility, acceptability and potential impact of providing an online personal dementia risk profile to help-seeking older adults at risk of developing AD on risk knowledge, motivation to change health-related behaviours, and psychological effects. To our knowledge, this is the first study focusing on the effects of communicating personal risk profile using risk algorithms based on epidemiological risk factors. Communication of the personal dementia risk profile led to more accurate knowledge of AD risk factors and improved understanding of perceived susceptibility among patients with MCI and SCD. Importantly, there was no negative effect of communicating the personal risk profile online on psychological distress or dementia-related worry among our participants. Participants mostly had a low risk of developing AD, but still reported reduced worry about getting dementia after receiving their risk profile. These findings support the feasibility and acceptability of using dementia risk algorithms to deliver personal risk profiles to low risk older adults in an online format, and indicate that providing this information can improve AD health literacy without a negative impact on psychological wellbeing.
However, there was little evidence in this study that providing personal risk profiles as a standalone intervention was sufficient to motivate change in behaviours to address AD risk factors. Although providing the dementia risk profile addressed one main barrier for risk reduction at baseline (i.e. lack of knowledge of dementia risk factors), participants reported that lack of motivation, health problems, and difficulty with organisation became the main barriers after receiving their risk profile. This suggests that older adults need extra support to effect behavioural change. The personal dementia risk profile could potentially be used as part of a collaborative, shared decision-making approach to address these barriers by guiding and engaging users, carers and clinicians to choose several high impact or easy-to-change risk factors to focus on, and by providing feedback on the change in risk level if risk factors are modified. Trials are underway to test the impact of a tailor-made online lifestyle programme targeting modifiable risk factors on risk score and health behaviours and compliance to health advice (27). There may also be a role for clinicians to follow up with specific guidance on addressing health problems and to assist the older adult to develop a personalised risk reduction plan. A recent rapid review on approaches to healthy ageing interventions for older adults demonstrated that optimal interventions are those that incorporate collaborative approaches with shared decision-making and behavioural change techniques (28). In this regard, the personal dementia risk profile represents a useful tool that clinicians can draw on to present evidence-based, tailored, health and risk information, which in turn can stimulate a collaborative decision-making process around which health/lifestyle factors to target and how best to achieve long-lasting behaviour change.
An interesting finding was that most participants overestimated their risk of developing AD, even after receiving their personal risk profile. This is possibly reflective of our cohort which was composed of help-seeking older adults seeking an evaluation at a memory clinic and were concerned with developing dementia. The continuing high levels of perceived risk at follow up is unsurprising given that previous research has found that even among individuals who accurately recalled their communicated AD risk, over 50% did not fully adjust their perceived risk to match the communicated risk, and that high baseline AD risk perception was the strongest predictor of overestimation of risk (29). It is also possible that participants may not have readily accepted the communicated risk after receiving lower than expected risk feedback, as seen by interview participants expressing surprise at their communicated risk.
Nonetheless, this has implications for supporting clinicians to communicate AD risk information to people with MCI. A survey found that 90% of neurologists said they counselled MCI patients on their risk of AD in general terms but only 60% communicated AD risk in numeric terms (30). Our findings provide preliminary support that patients with MCI or SCD understood numeric AD risk information and risk factors even when it is communicated online without support, and that the multifaceted approach with a clear visual representation and accompanying explanatory text may have facilitated understanding. Clinicians are encouraged to discuss numeric risk estimates with patients with visual aids, explain how these are estimated from risk models, and explore reasons for discord to improve risk appraisal.
There are several important considerations in interpreting the findings. The study sample was a well-educated inner-city cohort who have expressed concern about their memory and were highly motivated to seek help. Participants already knew that they did not have dementia. Their relatively positive reactions to their personal risk profile may reflect their personal interest in managing their brain health or because they were reassured of having low risk. It is unclear whether a general population or primary care sample, who had subjective memory concerns and not assessed for AD, would have similar reactions. It should be noted that a number of older adults approached to take part in the trial declined due to not wanting to know their risk or because they felt overwhelmed. Further research is needed to explore these concerns about knowing one’s personal risk.
The majority of our sample had low risk of developing AD, and it is unknown how moderate or high risk adults would respond to their personal dementia risk profile. There is some indication that high risk individuals, such as those who screen positive to genetic biomarkers, have heightened test-related distress (18). In order for dementia risk profiles to be widely and safely distributed to older adults in public health programs, particularly if they are to be delivered online in the absence of immediate clinical support, it is important to understand how moderate or high risk older adults react to their personal dementia risk profile and to monitor any potential adverse reactions. Finally, this was a pilot trial with a small sample size and short-term follow up. Longer-term follow up and randomised controlled trials to examine effects of communication of personal risk of developing AD are required.
The application of dementia risk algorithms to identify those at risk and to promote and encourage risk reduction behaviour is still in its early stages. This study provides preliminary support for the utility of using risk models that incorporate accessible and potentially modifiable risk factors to communicate personal dementia risk profiles to at-risk older adults.


Acknowledgements: The authors would like to acknowledge Professor Kaarin J. Anstey and Dr Sarang Kim for permission to use the ANU Alzheimer’s Disease Risk Index and their advice on adapting it to the Healthy Brain Ageing clinic measures. We thank the participants who have helped make this research possible.

Funding: This study was supported by a Dementia Australia Research Foundation project grant award. 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 standards: The study was approved by the University of Sydney Human Research Ethics Committee (protocol number: 2019/669) and registered with the Australian New Zealand Clinical Trials Registry (ANZCTRN12619001242112p).

Conflict of interest: The authors confirm that there are no known conflicts of interest associated with this publication.



1. Norton S, Matthews FE, Barnes DE, Yaffe K, Brayne C. Potential for primary prevention of Alzheimer’s disease: an analysis of population-based data. The Lancet Neurology. 2014;13(8):788-794
2. Isaacson RS, Hristov H, Saif N, et al. Individualized clinical management of patients at risk for Alzheimer’s dementia. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2019;15(12):1588-1602
3. Ngandu T, Lehtisalo J, Solomon A, et al. A 2 year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at-risk elderly people (FINGER): a randomised controlled trial. The Lancet. 2015;385(9984):2255-2263 S0140-6736(15)60461-5.
4. Kim S, Sargent-Cox KA, Anstey KJ. A qualitative study of older and middle-aged adults’ perception and attitudes towards dementia and dementia risk reduction. J Adv Nurs. 2015;71(7):1694-1703
5. Janz NK, Becker MH. The Health Belief Model: A Decade Later. Health Educ Q. 1984;11:1-47
6. Cations M, Radisic G, Crotty M, Laver KE. What does the general public understand about prevention and treatment of dementia? A systematic review of population-based surveys. PLoS One. 2018;13(4):e0196085
7. Low L-F, Anstey KJ. Dementia literacy: Recognition and beliefs on dementia of the Australian public. Alzheimer’s & Dementia. 2009;5(1):43-49
8. Smith BJ, Ali S, Quach H. Public knowledge and beliefs about dementia risk reduction: a national survey of Australians. BMC Public Health. 2014;14(1):661
9. Smith BJ, Ali S, Quach H. The motivation and actions of Australians concerning brain health and dementia risk reduction. Health Promot J Austr. 2015;26(2):115-121
10. Hou X-H, Feng L, Zhang C, Cao X-P, Tan L, Yu J-T. Models for predicting risk of dementia: a systematic review. J Neurol Neurosurg Psychiatry. 2018:jnnp-2018-318212
11. Anstey KJ, Cherbuin N, Herath PM. Development of a New Method for Assessing Global Risk of Alzheimer’s Disease for Use in Population Health Approaches to Prevention. Prevention Science. 2013;14(4):411-421
12. Anstey KJ, Cherbuin N, Herath PM, et al. A Self-Report Risk Index to Predict Occurrence of Dementia in Three Independent Cohorts of Older Adults: The ANU-ADRI. PLoS One. 2014;9(1):e86141
13. LaMonica HM, English A, Hickie IB, et al. Examining Internet and eHealth Practices and Preferences: Survey Study of Australian Older Adults With Subjective Memory Complaints, Mild Cognitive Impairment, or Dementia. J Med Internet Res. 2017;19(10):e358
14. Sindi S, Calov E, Fokkens J, et al. The CAIDE Dementia Risk Score App: The development of an evidence-based mobile application to predict the risk of dementia. Alzheimer’s & Dementia : Diagnosis, Assessment & Disease Monitoring. 2015;1(3):328-333
15. Sheffrin M, Stijacic Cenzer I, Steinman MA. Desire for predictive testing for Alzheimer’s disease and impact on advance care planning: a cross-sectional study. Alzheimers Res Ther. 2016;8(1):55.
16. Wikler EM, Blendon RJ, Benson JM. Would you want to know? Public attitudes on early diagnostic testing for Alzheimer’s disease. Alzheimers Res Ther. 2013;5(5):43
17. Roberts JS, McLaughlin SJ, Connell CM. Public beliefs and knowledge about risk and protective factors for Alzheimer’s disease. Alzheimer’s & Dementia. 2014;10(5):S381-S389
18. Bemelmans SASA, Tromp K, Bunnik EM, et al. Psychological, behavioral and social effects of disclosing Alzheimer’s disease biomarkers to research participants: a systematic review. Alzheimers Res Ther. 2016;8(1):46
19. Rimer BK, Halabi S, Sugg Skinner C, et al. Effects of a mammography decision-making intervention at 12 and 24 months. Am J Prev Med. 2002;22:247-257
20. Heger I, Deckers K, van Boxtel M, et al. Dementia awareness and risk perception in middle-aged and older individuals: baseline results of the MijnBreincoach survey on the association between lifestyle and brain health. BMC Public Health. 2019;19(1):678
21. Livingston G, Sommerlad A, Orgeta V, et al. Dementia prevention, intervention, and care. The Lancet. 2017;390(10113):2673-2734
22. Kim S, Sargent-Cox K, Cherbuin N, Anstey KJ. Development of the motivation to change lifestyle and health Behaviours for Dementia Risk Reduction Scale. Dement Geriatr Cogn Dis Extra. 2014;4(2):172-183
23. Kessler R, Andrews G, Colpe L, et al. Short screening scales to monitor population prevalences and trends in non-specific psychological distress. Psychol Med. 2002;32(06):959-976
24. Perini SJ, Slade T, Andrews G. Generic effectiveness measures: Sensitivity to symptom change in anxiety disorders. J Affect Disord. 2006;90:123-130
25. Kinzer A, Suhr JA. Dementia worry and its relationship to dementia exposure, psychological factors, and subjective memory concerns. Applied Neuropsychology: Adult. 2016;23(3):196-204
26. Farrow M. User perceptions of a dementia risk reduction website and its promotion of behavior change. JMIR research protocols. 2013;2(1):e15-e15
27. Vrijsen J, Abu-Hanna A, Maeckelberghe ELM, et al. Uptake and effectiveness of a tailor-made online lifestyle programme targeting modifiable risk factors for dementia among middle-aged descendants of people with recently diagnosed dementia: study protocol of a cluster randomised controlled trial (Demin study). BMJ Open. 2020;10(10):e039439
28. Owusu-Addo E, Ofori-Asenso R, Batchelor F, Mahtani K, Brijnath B. Effective implementation approaches for healthy ageing interventions for older people: A rapid review. Arch Gerontol Geriatr. 2021;92:104263
29. Linnenbringer E, Roberts JS, Hiraki S, Cupples LA, Green RC. “I know what you told me, but this is what I think:” Perceived risk of Alzheimer disease among individuals who accurately recall their genetics-based risk estimate. Genet Med. 2010;12(4):219-227
30. Roberts JS, Karlawish JH, Uhlmann WR, Petersen RC, Green RC. Mild cognitive impairment in clinical care: A survey of American Academy of Neurology members. Neurology. 2010;75(5):425-431