DNA from multiple viral species is associated with Alzheimer's disease risk

Abstract

Introduction: Multiple infectious agents, including viruses, bacteria, fungi, and protozoa, have been linked to Alzheimer's disease (AD) risk by independent lines of evidence. We explored this association by comparing the frequencies of viral species identified in a large sample of AD cases and controls.

Methods: DNA sequence reads that did not align to the human genome in sequences were mapped to viral reference sequences, quantified, and then were tested for association with AD in whole exome sequences (WES) and whole genome sequences (WGS) datasets.

Results: Several viruses were significant predictors of AD according to the machine learning classifiers. Subsequent regression analyses showed that herpes simplex type 1 (HSV-1) (odds ratio [OR] = 3.71, p = 8.03 × 10-4) and human papillomavirus 71 (HPV-71; OR = 3.56, p = 0.02), were significantly associated with AD after Bonferroni correction. The phylogenetic-related cluster of Herpesviridae was significantly associated with AD in several strata of the data (p < 0.01).

Discussion: Our results support the hypothesis that viral infection, especially HSV-1, is associated with AD risk.

Keywords: Alzheimer's disease; Alzheimer's disease sequencing project; antiviral agents; herpes simplex; human papillomavirus; torque teno viruses; whole exome sequencing; whole genome sequencing.

FIGURE 1

Frequency of viral reads by tissue source and type of sequencing. Proportion of total viral reads mapping to individual species in (A) whole exome sequence (WES) data from blood, (B) whole genome (WGS) sequence data from blood, (C) WES data from brain in WES, and (D) WGS data from brain. The innermost circle shows the proportion of all viral reads between Alzheimer's disease (AD) cases and controls within each of these subsets. The middle ring shows the proportion of viral reads mapping to a viral family within AD cases and controls, and the outer ring is the breakdown between viral species within a viral family

FIGURE 2

Top virus predictors of Alzheimer's disease (AD) by tissue source and type of sequencing. Bar charts of the ML weighted algorithm for (A) whole exome sequence (WES) data from brain, (B) whole genome (WGS) sequence data from brain, (C) WES data from blood in WES, and (D) WGS data from blood. Each feature within each subset is assigned a score created by summing the accuracy of the ML prediction model in which it improved the prediction of AD. The top 15 features are shown in each bar chart though several other viruses improved the prediction models