Alzheimer's Polygenic Risk Scores Are Not Interchangeable: Evidence from 1,752 Models

Abstract

IntroductionPolygenic risk scores (PRS) may improve Alzheimers disease (AD) risk prediction before symptom onset, yet choosing an appropriate model can be challenging.

MethodsUsing the standardized GenoPred pipeline, 1,752 PRS models (9 algorithms; 584 configurations; 3 genome-wide association studies) were evaluated and stratified by genetic ancestry and APOE diplotype. PRS models were evaluated using 11,200 clinical or autopsy-confirmed AD cases and 19,321 controls age [≥]65 from the Alzheimers Disease Sequencing Project Release 5.

ResultsPRS results were not consistent across methodologies (Spearmans {rho}: -0.49 to 1), with >95% of individuals having PRS in both the top and bottom risk deciles. Top-performing PRS were effective at stratifying AD risk across ancestries (AFR: P=2.04x10- 26; AMR: P=4.57x10-21; EAS: P=6.21x10-40; EUR: P=7.90x10-187).

DiscussionPRS parameters should be optimized for each ancestry. Contradictory signals across methodologies underscore the need for carefully choosing suitable PRS methods and fine-tuning algorithmic parameters to ensure accuracy and consistency.

Data AvailabilityAccess to the ADSP is controlled by The National Institute on Aging Genetics of Alzheimers Disease (NIAGADS). All scripts used to analyze the data are freely available at https://github.com/jmillerlab/prs_comparisons.

Rare extreme polygenic risk scores strongly indicate Alzheimer's disease risk
E. L. Ward, P. T. Nelson, Y. Katsumata +4
medRxiv8 Sep 2026PreprintCC BY-NCFull textHuman geneticsmedRxiv
Pathway‐specific polygenic risk scores modify the association between amyloid and tau PET accumulation in patients with Alzheimer's disease
Carolina Valentim, Jannis Denecke, Simon Frerich +7
Alzheimer's & Dementia27 Oct 2026Peer-reviewedCC BY 4.0Full textPhysiologyPMC
Hybrid risk scores integrating polygenic and clinical variables for endometriosis prediction
O. Goroshchuk, D. Koller
medRxiv31 Aug 2026PreprintAll rights reservedFull textEpidemiology & public healthmedRxiv
Adult Patient Perspectives on Psychiatric Polygenic Risk Scores
C. W. McFarland, A. Merner, P. M. Trotter +8
medRxiv24 Sep 2026PreprintCC BYFull textHuman geneticsmedRxiv
Polygenic risk scores for LATE-neuropathologic change: Implications for clinical utility.
Jung-Min Pyun, Young Ho Park, Lincoln M P Shade +4
Alzheimer's & dementia : the journal of the Alzheimer's Association1 Oct 2026Peer-reviewedCC BY-NC-NDFull textPsychiatry & mental healthEurope PMCPMC

Cite

BibTeX RIS

BibTeX

@misc{ward2026alzheimers,
  title = {{Alzheimer's Polygenic Risk Scores Are Not Interchangeable: Evidence from 1,752 Models}},
  author = {E. L. Ward and P. T. Nelson and Y. Katsumata and D. W. Fardo and G. A. Jicha and Alzheimer's Disease Neuroimaging Initiative and J. B. Miller},
  howpublished = {medRxiv},
  year = {2026},
  month = sep,
  doi = {10.64898/2026.09.02.26362060},
  url = {https://www.medrxiv.org/content/10.64898/2026.09.02.26362060v1},
  note = {CC BY-NC},
}

RIS

TY  - UNPB
TI  - Alzheimer's Polygenic Risk Scores Are Not Interchangeable: Evidence from 1,752 Models
AU  - E. L. Ward
AU  - P. T. Nelson
AU  - Y. Katsumata
AU  - D. W. Fardo
AU  - G. A. Jicha
AU  - Alzheimer's Disease Neuroimaging Initiative
AU  - J. B. Miller
PY  - 2026
DA  - 2026/09/02
PB  - medRxiv
DO  - 10.64898/2026.09.02.26362060
UR  - https://www.medrxiv.org/content/10.64898/2026.09.02.26362060v1
AB  - IntroductionPolygenic risk scores (PRS) may improve Alzheimers disease (AD) risk prediction before symptom onset, yet choosing an appropriate model can be challenging.  MethodsUsing the standardized GenoPred pipeline, 1,752 PRS models (9 algorithms; 584 configurations; 3 genome-wide association studies) were evaluated and stratified by genetic ancestry and APOE diplotype. PRS models were evaluated using 11,200 clinical or autopsy-confirmed AD cases and 19,321 controls age [≥]65 from the Alzheimers Disease Sequencing Project Release 5.  ResultsPRS results were not consistent across methodologies (Spearmans {rho}: -0.49 to 1), with >95% of individuals having PRS in both the top and bottom risk deciles. Top-performing PRS were effective at stratifying AD risk across ancestries (AFR: P=2.04x10- 26; AMR: P=4.57x10-21; EAS: P=6.21x10-40; EUR: P=7.90x10-187).  DiscussionPRS parameters should be optimized for each ancestry. Contradictory signals across methodologies underscore the need for carefully choosing suitable PRS methods and fine-tuning algorithmic parameters to ensure accuracy and consistency.  Data AvailabilityAccess to the ADSP is controlled by The National Institute on Aging Genetics of Alzheimers Disease (NIAGADS). All scripts used to analyze the data are freely available at https://github.com/jmillerlab/prs_comparisons.
ER  -

Abstract

Related

Cite