Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Benchmarking Alzheimer’s disease prediction : personalised risk assessment using polygenic risk scores across various methodologies and genome-wide studies

Bellou, Eftychia ; Kim, Woori ; Leonenko, Ganna ; Tao, Feifei ; Simmonds, Emily ; Wu, Ying ; Mattsson-Carlgren, Niklas LU orcid ; Hansson, Oskar LU orcid ; Nagle, Michael W. and Escott-Price, Valentina (2025) In Alzheimer's Research and Therapy 17(1).
Abstract

Background: The success of selecting high risk or early-stage Alzheimer’s disease individuals for the delivery of clinical trials depends on the design and the appropriate recruitment of participants. Polygenic risk scores (PRS) show potential for identifying individuals at risk for Alzheimer’s disease (AD). Our study comprehensively examines AD PRS utility using various methods and models. Methods: We compared the PRS prediction accuracy in ADNI (N = 568) and BioFINDER (N = 766) cohorts using five disease risk modelling approaches, three PRS derivation methods, two AD genome-wide association study (GWAS) statistics and two sets of SNPs: the whole genome and microglia-selective regions only. Results: The best prediction accuracy was... (More)

Background: The success of selecting high risk or early-stage Alzheimer’s disease individuals for the delivery of clinical trials depends on the design and the appropriate recruitment of participants. Polygenic risk scores (PRS) show potential for identifying individuals at risk for Alzheimer’s disease (AD). Our study comprehensively examines AD PRS utility using various methods and models. Methods: We compared the PRS prediction accuracy in ADNI (N = 568) and BioFINDER (N = 766) cohorts using five disease risk modelling approaches, three PRS derivation methods, two AD genome-wide association study (GWAS) statistics and two sets of SNPs: the whole genome and microglia-selective regions only. Results: The best prediction accuracy was achieved when modelling genetic risk by using two predictors: APOE and remaining PRS (AUC = 0.72–0.76). Microglial PRS showed comparable accuracy to the whole genome (AUC = 0.71–0.74). The individuals’ risk scores differed substantially, with the largest discrepancies (up to 70%) attributable to the GWAS statistics used. Conclusions: Our work benchmarks the best PRS derivation and modelling strategies for AD genetic prediction.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; ; ; ; and
author collaboration
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Alzheimer’s disease, Polygenic risk score, Risk prediction
in
Alzheimer's Research and Therapy
volume
17
issue
1
article number
6
publisher
BioMed Central (BMC)
external identifiers
  • scopus:85214383857
  • pmid:39762974
ISSN
1758-9193
DOI
10.1186/s13195-024-01664-9
language
English
LU publication?
yes
additional info
Publisher Copyright: © The Author(s) 2025.
id
810b9692-f3b4-47fe-b979-63a4e441d499
date added to LUP
2025-03-24 13:38:18
date last changed
2025-07-14 20:19:09
@article{810b9692-f3b4-47fe-b979-63a4e441d499,
  abstract     = {{<p>Background: The success of selecting high risk or early-stage Alzheimer’s disease individuals for the delivery of clinical trials depends on the design and the appropriate recruitment of participants. Polygenic risk scores (PRS) show potential for identifying individuals at risk for Alzheimer’s disease (AD). Our study comprehensively examines AD PRS utility using various methods and models. Methods: We compared the PRS prediction accuracy in ADNI (N = 568) and BioFINDER (N = 766) cohorts using five disease risk modelling approaches, three PRS derivation methods, two AD genome-wide association study (GWAS) statistics and two sets of SNPs: the whole genome and microglia-selective regions only. Results: The best prediction accuracy was achieved when modelling genetic risk by using two predictors: APOE and remaining PRS (AUC = 0.72–0.76). Microglial PRS showed comparable accuracy to the whole genome (AUC = 0.71–0.74). The individuals’ risk scores differed substantially, with the largest discrepancies (up to 70%) attributable to the GWAS statistics used. Conclusions: Our work benchmarks the best PRS derivation and modelling strategies for AD genetic prediction.</p>}},
  author       = {{Bellou, Eftychia and Kim, Woori and Leonenko, Ganna and Tao, Feifei and Simmonds, Emily and Wu, Ying and Mattsson-Carlgren, Niklas and Hansson, Oskar and Nagle, Michael W. and Escott-Price, Valentina}},
  issn         = {{1758-9193}},
  keywords     = {{Alzheimer’s disease; Polygenic risk score; Risk prediction}},
  language     = {{eng}},
  number       = {{1}},
  publisher    = {{BioMed Central (BMC)}},
  series       = {{Alzheimer's Research and Therapy}},
  title        = {{Benchmarking Alzheimer’s disease prediction : personalised risk assessment using polygenic risk scores across various methodologies and genome-wide studies}},
  url          = {{http://dx.doi.org/10.1186/s13195-024-01664-9}},
  doi          = {{10.1186/s13195-024-01664-9}},
  volume       = {{17}},
  year         = {{2025}},
}