PRISM-AD quantifies clinical and genetic contributions to cognitive risk prediction

Abstract

Background: Older adults with high inherited Alzheimer's disease (AD) susceptibility can still be cognitively normal (CN). An overall risk estimate alone does not show the cognitive and genetic contributions within a forecast. Objective: To quantify clinical and genetic contributions to cognitive prognosis. Methods: PRISM-AD combined nonlinear clinical functions and a polygenic risk score (PRS) term in 4,435 European-ancestry participants. Participant-level internal validation evaluated first recorded mild cognitive impairment (MCI) after CN in 908 participants and AD dementia after MCI in 714, within three or five years. Separately fitted comparators used identical inputs. Post hoc clinical-course comparisons used regression with robust standard errors and Benjamini-Hochberg (BH) correction. Results: PRISM-AD using current assessments achieved average precision of 0.2996 and 0.4094 for three- and five-year CN-to-MCI, with estimates close to logistic regression. No PRISM-AD PRS increment met BH q < 0.05. Three-year first-MCI forecasts characterized risk contributions in 893 initially CN participants. A retrospective comparison over available follow-up included 95 participants with later AD and 701 with CN at their last diagnostic assessment. Modeled genetic factors were higher in later AD in nine of ten shared cognitive-risk deciles. After adjustment for modeled cognitive risk and fold, the genetic-factor ratio was 1.135 (nominal 95% CI, 1.073 to 1.201, BH q < 0.001). Conclusions: Higher genetic contributions were associated with later AD at comparable modeled cognitive risk. These exploratory profiles require independent validation before use in cognitive follow-up or prevention research.

Prediction of cognitive impairment in Parkinson’s disease using TabPFN: an integrative analysis of clinical data, longitudinal trajectories, and lipidomics
Zikai Zhou, Jie Liu, Lijuan Wang +5
Frontiers in aging neuroscience16 Oct 2026Peer-reviewedCC BYFull textGeneral medicineEurope PMC
Machine learning-derived Alzheimer's disease dimensions: neuroanatomical, cognitive, clinical, functional, and genetic risk heterogeneity signatures
H. N. Le, S. Ishrat, B. Yaghooti +7
medRxiv2 Sep 2026PreprintCC BY-NCFull textMedical imaging & radiologymedRxiv
Personalized Prediction of Venetoclax-Associated Tumor Lysis Syndrome in Acute Myeloid Leukemia Using Drug Exposure, Genetic Polymorphisms, and Clinical-Laboratory Characteristics.
Chanchan Wang, Xiaoyan Lin, Yonggong Yang +6
Drug design, development and therapy29 Oct 2026Peer-reviewedCC BY-NCFull textGeneral medicineEurope PMCPMC
Dissociable contributions of cortical thickness and surface area to cognitive aging: evidence from multiple longitudinal cohorts.
Ina Demetriou, Marta Correia, Didac Vidal-Piñeiro +8
Cerebral cortex (New York, N.Y. : 1991)1 Oct 2026Peer-reviewedCC BYFull textPsychiatry & mental healthEurope PMC
Genetic and non-genetic contributions to early spontaneous miscarriage in women with PCOS undergoing PGT-A and single euploid blastocyst transfer: a retrospective cohort study
Zhiqin Bu, Xiaofei Ge, Jing Sun +1
BMC Pregnancy and Childbirth31 Jul 2026Peer-reviewedCC BY-NC-ND 4.0Full textObstetrics, gynaecology & paediatricsPMC

Cite

BibTeX RIS

BibTeX

@misc{dai2026prismad,
  title = {{PRISM-AD quantifies clinical and genetic contributions to cognitive risk prediction}},
  author = {Y. Dai and C. S. Wood and X. Jiang and Y.-C. Hsu},
  howpublished = {medRxiv},
  year = {2026},
  month = nov,
  doi = {10.64898/2026.11.03.26367206},
  url = {https://www.medrxiv.org/content/10.64898/2026.11.03.26367206v1},
  note = {CC BY},
}

RIS

TY  - UNPB
TI  - PRISM-AD quantifies clinical and genetic contributions to cognitive risk prediction
AU  - Y. Dai
AU  - C. S. Wood
AU  - X. Jiang
AU  - Y.-C. Hsu
PY  - 2026
DA  - 2026/11/03
PB  - medRxiv
DO  - 10.64898/2026.11.03.26367206
UR  - https://www.medrxiv.org/content/10.64898/2026.11.03.26367206v1
AB  - Background: Older adults with high inherited Alzheimer's disease (AD) susceptibility can still be cognitively normal (CN). An overall risk estimate alone does not show the cognitive and genetic contributions within a forecast. Objective: To quantify clinical and genetic contributions to cognitive prognosis. Methods: PRISM-AD combined nonlinear clinical functions and a polygenic risk score (PRS) term in 4,435 European-ancestry participants. Participant-level internal validation evaluated first recorded mild cognitive impairment (MCI) after CN in 908 participants and AD dementia after MCI in 714, within three or five years. Separately fitted comparators used identical inputs. Post hoc clinical-course comparisons used regression with robust standard errors and Benjamini-Hochberg (BH) correction. Results: PRISM-AD using current assessments achieved average precision of 0.2996 and 0.4094 for three- and five-year CN-to-MCI, with estimates close to logistic regression. No PRISM-AD PRS increment met BH q < 0.05. Three-year first-MCI forecasts characterized risk contributions in 893 initially CN participants. A retrospective comparison over available follow-up included 95 participants with later AD and 701 with CN at their last diagnostic assessment. Modeled genetic factors were higher in later AD in nine of ten shared cognitive-risk deciles. After adjustment for modeled cognitive risk and fold, the genetic-factor ratio was 1.135 (nominal 95% CI, 1.073 to 1.201, BH q < 0.001). Conclusions: Higher genetic contributions were associated with later AD at comparable modeled cognitive risk. These exploratory profiles require independent validation before use in cognitive follow-up or prevention research.
ER  -

Abstract

Related

Cite