# Polygenic risk score and phenome-wide association study of the Epstein-Barr virus antibody response

## TL;DR

This study identifies genetic factors influencing Epstein-Barr virus antibody responses and develops a predictive genetic score linked to autoimmune diseases.

## Contribution

A multi-ancestry polygenic risk score for EBV antibody response was developed and validated across diverse populations.

## Key findings

- A significant genetic association was found at chromosome 6 within the MHC region.
- A multi-ancestry PRS model showed predictive performance with AUCs between 65% and 72%.
- PheWAS analyses revealed a strong link between EBV antibody response and multiple sclerosis.

## Abstract

The Epstein-Barr virus (EBV) infection is nearly ubiquitous and has established links to malignancy and autoimmune disease. Here we evaluate the genetic factors influencing the humoral immune response to EBV and establish a polygenic risk score (PRS) for anti-EBNA1 responses.

We conducted a multi-biobank genetic study for the serologic humoral IgG antibody response to EBV-EBNA1, including data from the UK Biobank (UKB, N = 9695 individuals) and the Milieu Intérieur (MI) cohort from Institute Pasteur (IP) (N = 1000), as well as GWAS summary statistics for individuals of African ancestry (N = 4365). We divided the cohort into discovery and validation, performed GWAS analyses, and developed a PRS using a Bayesian multi-ancestry approach application (PRS-CSx). After successfully validating PRS predictive performance, we then applied PheWAS analyses to all UKB datasets.

Consistent with a previous report, our GWAS analyses identified an association at chromosome 6 within the MHC region as the most significant SNP (rs6927022; p = 8.21 × 10−113). Meta-analyses of all cohorts identified several regions outside of the MHC region with the strongest effects at 10p12 within the armadillo repeat containing three gene (ARMC3) (rs10763372; p = 9.93 × 10−08) and at 10q22 upstream of Neuregulin-3 (NRG3) (rs61851627; p = 8.62 × 10−9). A multi-ancestry PRS model was then trained and developed, obtaining successful predictive performance in validation cohorts (AUC ranges from 65% to 72%). Finally, PheWAS analyses using the developed PRS confirmed a significant positive association with multiple sclerosis (p = 1.14 × 10−10), and also identified novel negative associations with other spectra of autoimmune disease, including Celiac disease (2.03 × 10−127).

We developed and validated a multi-ancestry PRS for EBNA1 IgG response that enables genetic profiling of EBV antibody responsiveness in genotyped cohorts lacking serologic measurements. PheWAS results support shared and divergent genetic relationships between EBV antibody response and autoimmune disease, motivating mechanistic and longitudinal follow-up studies. Further research is needed to evaluate the future implementation of this PRS in the clinic to improve long-term health outcomes.