APOE and Polygenic Risk Shift Alzheimer Onset by 10 Years
Key Takeaways
- Among people with an extra copy of APP across APP duplication and Down syndrome, Alzheimer disease onset was earlier on average in APP duplication while overall variability in onset was similar between groups.
- APOE ε2 was associated with later Alzheimer disease onset across APP-driven populations.
- APOE ε4 and higher non-APOE Alzheimer disease genetic risk were each associated with earlier onset, with similar effect directions in APP duplication and Down syndrome.
- The lowest- versus highest-risk combined genetic profiles corresponded to about a 10-year difference in predicted median onset, and pathway analyses pointed to amyloid beta, angiogenesis, and Down syndrome endocytosis signals.
In a multicenter retrospective cohort reported by Groeneveld and colleagues in Alzheimer’s & Dementia, investigators analyzed 100 APP duplication carriers and 957 individuals with Down syndrome (DS) who carried an extra copy of APP. Alzheimer disease (AD) age at onset (AAO) was defined differently across cohorts, unaffected participants were censored at last follow-up, and Cox models accounted for cohort. Apolipoprotein E (APOE) ε2 and ε4 were compared with ε3/ε3, and the Alzheimer disease genetic risk score (AD-GRS) excluded APOE and chromosome 21 variants and was built from 77 variants after filtering. Pathway-specific scores covered amyloid beta metabolism, immune response, endocytosis, cholesterol or lipid metabolism, and angiogenesis.
Mean AAO was 50.7 ± 6.6 years in APP duplication versus 53.3 ± 5.5 years in DS (P = 0.0005), while overall variability was similar and prodromal DS onset was comparable to APP duplication when examined separately. In the combined cohort, APOE ε2 was associated with later onset (HR 0.47 [0.32-0.67]), whereas APOE ε4 (HR 1.5 [1.2-1.9]) and the AD-GRS (HR 1.3 [1.1-1.4] per SD) were associated with earlier onset. No interaction was detected between APOE and AD-GRS, and subgroup estimates were directionally similar in APP duplication and DS, with wider uncertainty in the smaller APP duplication group.
Modeled genetic-risk differences in age at onset corresponded to predicted median AAO of 49.4 versus 58.8 years in APP duplication and 50.1 versus 60.1 years in DS, a 9.4- and 10-year gap respectively. Within each APOE genotype, higher AD-GRS values tracked with earlier predicted onset, pathway-specific analyses implicated amyloid beta and angiogenesis overall with an endocytosis association in DS, and single-variant meta-analysis suggested the AD-GRS signal was not driven by one variant. Duplication length and duplicated chromosome 21 gene content were not associated with AAO.
Direct comparison of absolute onset ages between APP duplication and DS is limited because AAO was defined differently across cohorts, and recognition of Alzheimer disease in DS may lag because of baseline intellectual disability, comorbidities, and atypical presentations. The authors interpreted the modest mean difference between groups as more definitional than biological, and the much smaller APP duplication subgroup widened confidence intervals and left some subgroup estimates non-significant despite similar effect directions. In that context, APOE and polygenic risk functioned as modifiers of onset timing rather than deterministic individual forecasts, and the authors framed the findings as evidence of overlap between APP-driven and sporadic Alzheimer disease biology.
APOE genotype and a non-APOE AD-GRS were associated with AAO in people carrying an extra copy of APP, and the modeled contrast between the lowest- and highest-risk genetic profiles approached a decade in both APP duplication and DS. The authors further suggested that pathways tied to amyloid beta, angiogenesis, and endocytosis may explain part of the heterogeneity in APP-driven Alzheimer disease.
Clinician Questions
How was age at onset defined differently in APP duplication and Down syndrome in this analysis?
In APP duplication, AAO was the age at first symptom onset reported by the patient or family and could include cognitive decline, hemorrhages, or seizures. In DABNI, AAO was the age at prodromal AD diagnosis or the age at first visit if dementia was already established, whereas in ACE/HUP it was the age at clinical AD dementia diagnosis. Those cohort-specific definitions limit direct comparison of absolute onset ages across APP duplication and Down syndrome populations.
What did the non-APOE Alzheimer disease genetic risk score include in people with an extra copy of APP?
The non-APOE AD-GRS used genome-wide significant AD variants from a recent genome-wide association study, excluded APOE and chromosome 21 variants, removed low-quality single nucleotide polymorphisms, and retained 77 variants after filtering. The score was standardized within each population subgroup and modeled as a cumulative measure of sporadic AD susceptibility rather than a single-gene effect.
Were the APOE and polygenic risk signals similar in APP duplication and Down syndrome?
The direction of effect was similar in both populations: APOE ε2 tracked with later onset, whereas APOE ε4 and higher AD-GRS tracked with earlier onset. APP duplication estimates were less statistically certain because that subgroup was much smaller, and the combined analyses did not detect an interaction between APOE and AD-GRS.
Did APP duplication length or other duplicated chromosome 21 genes explain variation in Alzheimer disease onset?
Within the APP duplication subset, duplication length was not associated with AAO, and none of the tested duplicated chromosome 21 genes present in at least 5 carriers showed a significant association. In this analysis, variation in Alzheimer disease onset within APP duplication was not explained by the size of the duplication or by the presence of individual duplicated chromosome 21 genes that could be assessed.