Analysis of 436,390 genetic variants in 9,765 elderly individuals implicates TOMM40, MARK4, CLPTM1 and VDAC1/FSTL4 in the inverse relationship between Alzheimer's and cancer




Zhou, Zhengyang
Pathak, Gita A.
Phillips, Nicole


0000-0003-3943-0895 (Pathak, Gita A.)

Journal Title

Journal ISSN

Volume Title



Purpose A rapidly aging demographic, aged 65 and older, is expected to double by 2060 reaching 98 million and creating demands for better healthcare. While the number of co-occurring diseases increase in the aging population, Alzheimer’s and cancer have been reported to be inversely related – a lower than expected probability of the secondary disease after the primary disease diagnosis. This fueled our interest in exploring genetic variation that is responsible for the inverse relationship between AD and cancer. Rationale & Hypothesis Age is a risk factor for both AD and cancer, and our goal was to compare late-onset AD with two most prevalent age-related cancers – breast and prostate cancer. We hypothesize that harmonizing the age to study the cross-phenotypic effects of genetic variants between AD and cancer against a common control group will identify genetic variants that contribute to their inverse relationship. Methods Genomic SNP data from ADNI (Alzheimer’s Disease Neuroimaging Initiative), ADGC (Alzheimer’s Disease Genetics Consortium) and BPC3 (Breast and Prostate Cancer Care Consortium) which contained 757, 6065, and 11893 individuals respectively included genotypes for up to 700,000 SNP markers. Standard quality control measures were implemented. Individuals with age of disease onset between 60- 80 years were included, and Bayesian multinomial regression was used to compare cases (AD and cancer) against controls in a two-stage replication study. Results A total of 4 SNPs that replicated in the two study stages. In males, two risk loci were significant with opposite odds ratios – rs2075650 mapped to TOMM40 , and an intergenic SNP- rs4298154 on chr 4. Since TOMM40 is near APOE region, we conditioned on APOE SNPs – rs429358 and rs769449, to identify secondary hits. 8 SNPs in the MARK4 region were significant with the inverse hit. In females, rs2075650 was also significant, and conditional analysis resulted in variants in CLPTM1. A non-coding SNP in the VDAC1/ FSTL4 region was also replicated in females. Conclusion Our novel approach has identified four genic regions that have cross-phenotypic effects when comparing AD and cancer: TOMM40/APOE, MARK4, CLPTM1, and VDAC/FSTL4. These genes have been previously implicated independently in AD and cancer and are known to be involved in mitochondrial pathways; however, this is the first study to directly demonstrate that genetic variability in these genes underlies the inverse comorbidity of AD and cancer.