Liu H, Marsh TW, Shi X, Renton AE, Bowling KM, Ziegemeier E, Wang G, Cao Y, Aristel A, Li J, Dickson A, Perrin RJ, Goate AM, Fernández V, Day GS, Doering M, Daniels A, Gordon BA, Benzinger TL, Hassenstab J, Ibanez L, Supnet-Bell C, Xiong C, Allegri R, Berman SB, Fox NC, Ryan N, Huey ED, Vöglein J, Noble JM, Roh JH, Jucker M, Laske C, Ikeuchi T, Sanchez-Valle R, Schofield PR, Chrem Mendez P, Chhatwal JP, Farlow M, Lee JH, Levey AI, Levin J, Lopera F, Martins R, Niimi Y, Rosa-Neto P, Morris JC, Bateman RJ, Karch CM, Cruchaga C, McDade E, Llibre-Guerra JJ. The landscape of autosomal-dominant Alzheimer's disease: global distribution and age of onset. Brain. 2025 Feb 4; Epub 2025 Feb 4 PubMed.

Comments

- John Hardy
  Institute of Neurology, UCL
- Posted: 18 Apr 2025
- News: Trial-Eligible Genetic Variants Now Marked in Alzforum Mutations Database
This is a useful review which will serve as a reference for the field. As I assume the authors would agree, it will be great to expand this into African populations. In the 2010 Guerreiro paper (cited) we found that several benign variants were actually African alleles.

View all comments by John Hardy

- Gael Nicolas
  Rouen University Hospital
- Posted: 18 Apr 2025
- News: Trial-Eligible Genetic Variants Now Marked in Alzforum Mutations Database
This article is a very useful resource for Alzheimer's genetics. The amount of work done for variant curation is massive. I would like to thank the DIAN team for the data they are bringing to the community with this paper along with their previous work. I am glad to learn that the Alzforum database is adding information based on this paper.

Regarding the algorithm itself, I think it is very useful to put a threshold of frequency at the level of the most recurrent PSEN1 pathogenic variant, A79V. This kind of gene/disease-specific criteria is helpful to refine the ACMG-AMP recommendations for a specific disease, as has been done in many other Mendelian disorders. This might help non-experts interpret variants in APP, PSEN1, or PSEN2. However, I think it should be revised to be placed at the level of the A713T APP variant for diagnostic purposes, while keeping it at the level of A79V PSEN1 for the DIAN-TU eligibility list.

I would like to bring to the attention of Alzforum readers that the algorithm is not a pathogenicity algorithm but rather eligibility criteria for a trial, as the authors clearly state. Some variants might be pathogenic, or likely pathogenic, and still not eligible for DIAN-TU. For example, the APP A713T variant is a well known likely pathogenic variant, known by experts, but gets misclassified as variant of uncertain significance in Clinvar (and in this paper and in Alzforum!), and the variant is not eligible for DIAN-TU. This is most likely because this is a variant of incomplete penetrance, and its ages of onset are highly variable and unpredictable. Thus, it is logical not to consider this variant for a preventive clinical trial based on the estimated age at onset although it is likely pathogenic.

Because the algorithm is not a variant pathogenicity classifier, we should for that purpose still rely on the ACMG-AMP recommendations. Of note, in the ACMG-AMP recommendations "putative pathogenic" or "putative risk factor" categories do not exist. Here, adding a putative risk factor category may be a bit misleading. Considering a variant as a risk factor requires accurate statistics (significant enrichment in cases) and, while some variants show reduced penetrance in APP or PSEN2, for example, their status as a risk factor remains unclear. From my perspective, a variant that does not fulfill all the pathogenicity criteria but lies somewhere between pathogenic and benign should be a variant of uncertain significance, in a Mendelian point of view, and its status as a putative risk factor is another question.

I do not know an example of a variant that would show Aβ levels altered similarly to known mutations but does not meet any of the supportive criteria. More importantly, if there is no segregation data available and the variant still shows Aβ levels altered similarly to known mutations and supportive criteria, I think it is most likely to be pathogenic, not "putative risk factor." Maybe I misunderstood the segregation criterion at this point, but it is often difficult to have sufficient segregation information in a given family. The segregation argument is unclear to me. How many informative meioses are required? What is an informative meiosis based on age of onset in affected relatives (risk of phenocopies) and age at last visit for asymptomatic relatives?

Regarding AAO predictability, I think this article confirms previous work by this team and others, and is very useful for the community. Obviously, the EYO is interesting as it provides an idea. Even so, it remains difficult to use in clinical practice for individual prediction, given the dispersion of ages of onset which, although limited compared to many other autosomal-dominant neurocognitive disorders, remains not negligible. We all know families with so-called outliers, hence research on modifiers remains an interesting topic.

Adding the DIAN-TU eligibility data in the Alzforum APP, PSEN1 and PSEN2 datasets is helpful. This adds a criterion of confidence for professionals involved in Alzheimer's disease genetics, because it means that the variant they are looking at has been reviewed. It thus adds to the information present in the Clinvar database or other pathogenic variants databases. Also, it is most useful for patients, families, and medical doctors who might assess eligibility for the trial, so that they can contact the DIAN-TU team, if the trial is open in their country.

However, I would like to remind patients, families, and professionals who are not experts in AD genetics to first rely on the individual genetics lab report to assess pathogenicity, and, if there is any doubt, request secondary advice from an AD genetics expert lab if the analysis was done in a generalist genetics lab.

View all comments by Gael Nicolas

- Jorge Jesus Llibre Guerra
  Washington University School of Medicine in St. Louis
- Posted: 18 Apr 2025
- News: Trial-Eligible Genetic Variants Now Marked in Alzforum Mutations Database
The DIAN-TU eligibility algorithm provides a framework to ensure that only variants with the highest level of evidence for pathogenicity in ADAD are included in clinical trials. It’s not intended to replace the ACMG-AMP classification but rather to complement it, with a specific focus on ADAD. It also sets a framework that incorporates evidence from biomarkers, neuropathology, functional studies, and family data, which can evolve as new evidence emerges, ensuring the field continues to move toward precision prevention in AD. This helps ensure scientific rigor, protects trial participants, and provides clarity for researchers, clinicians, and families navigating genetic testing and trial eligibility.

The expanded analysis of variant-specific age at onset provides an important tool for the prediction of when symptoms may begin in ADAD mutation carriers. In our study, we validated that variant AAO has a strong predictive value. This information is already used in DIAN-TU trials to calculate Estimated Years to Onset, which guides participant selection and trial design. Importantly, refining AAO estimates also helps define the optimal window for intervention, particularly in the context of primary and secondary prevention trials. This allows us to more precisely target individuals at different disease stages and design trials that are better aligned with the biology of ADAD progression.

We recommend that families, researchers, or clinicians who are interested in determining trial eligibility for a particular ADAD variant that is NOT included in the published list reach out to DIAN directly (via https://dian.wustl.edu). Importantly, variant eligibility is dynamic. If a variant is not currently eligible, it could become eligible as new data emerge, particularly if additional families are identified, biomarker data are collected, or functional studies support pathogenicity. Families can contribute to this process by participating in observational studies like DIAN-OBS, sharing clinical and family history, and enabling further genetic and biomarker research.

Looking ahead, expanding access to genetic testing globally, improving representation of understudied populations, and increasing data-sharing will be essential to refine variant classification and age-at-onset prediction. Ultimately, this work brings us closer to precision prevention in AD—allowing us to design interventions tailored not only to individuals at risk but also to the specific genetic and biological characteristics of their disease.

View all comments by Jorge Jesus Llibre Guerra

Make a Comment

To make a comment you must login or register.

Paper

Tools

Comments

Make a Comment

This paper appears in the following:

Mutations

News