Genome-wide association meta-analysis of Alzheimer's disease in Colombian, Argentinian, and Chilean populations

Latin American populations are underrepresented in Alzheimer’s disease (AD) genome-wide association studies (GWAS), limiting ancestry-specific risk prediction. To address this gap, a GWAS was conducted with 254 AD cases and 241 controls from Colombia, followed by a trans-ancestry meta-analysis that incorporated cohorts from Colombia, Chile, Argentina, and the European EADB consortium. AD polygenic risk scores (AD-PRS) derived from the Latin American GWAS were compared with those from European GWAS. APOE -ε4 showed the strongest association in Colombians (OR=3.33, P =1.81×10 ^-9 ). The meta-analysis identified five loci, including a population-specific signal at FBXO33/MIA2 (OR=1.09, P =8.39×10 ^-9 ). Additionally, a missense coding variant at GCAT , first identified in a Caribbean Hispanic cohort with enriched Native American ancestry (NAM), was validated in our admixed population with strong NAM ancestry (OR=2.55, P =1.00×10 ^-4 ). The accuracy of European-derived AD-PRSs was largely driven by APOE and declined with increasing NAM, whereas Latin American-derived AD-PRS captured ancestry-specific polygenic risk beyond APOE adjustment. These findings indicate that AD risk in Colombians reflects both shared and ancestry-specific genetic effects. In addition, our results suggest that genetic datasets from admixed Latin American populations are beginning to reach the scale and resolution necessary not only to replicate established AD loci but also to identify population-specific variants, including novel signals and refined associations at previously implicated loci. These results underscore the importance of ancestry-matched GWAS and PRS development to achieve equitable and biologically accurate AD risk prediction.