A multiancestry polygenic risk score for Alzheimer's disease is associated with cognitive decline and neuropathological hallmarks in diverse populations.

APOE-independent polygenic risk score built from genome-wide association study summary statistics applied to cohorts of European, African American, Caribbean Hispanic and East Asian ancestry.

Polygenic risk scores developed so far for Alzheimer's disease perform inconsistently across ancestries. The authors built an APOE-independent multiancestry score from genome-wide association study summary statistics applied to European ancestry, African American, Caribbean Hispanic and East Asian cohorts, then evaluated its performance in a large independent multiancestry dataset and validated it in several additional cohorts. The score was significantly associated with disease in European ancestry, African American, Caribbean Hispanic and Native American Hispanic groups, with adjusted odds ratios between 1.14 and 1.52 per score standard deviation, and between 1.21 and 1.65 in replication cohorts. It was also associated with poorer memory, executive function and language performance, greater neuropathological burden, reduced hippocampal volume, lower cerebrospinal fluid amyloid-beta 42 and elevated total tau and phosphorylated tau, with stronger phosphorylated tau associations in women.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

The strength of this work is not the effect size, which is modest, but its anchoring on biological measures independent of clinical diagnosis — neuropathology, hippocampal volume, cerebrospinal fluid markers — which makes the association hard to attribute to misclassification. An odds ratio of 1.14 to 1.52 per standard deviation remains far from individual use, and the authors say as much: the score is meaningful only as one component of multimodal stratification. That a score built to be APOE-independent holds across four ancestry groups is the real advance, in a field where almost every tool has been calibrated on European populations.

Analysis by Dr Thibaut Benquey