Multi-ancestry meta-analysis of genome-wide association studies discovers 67 new loci associated with chronic back pain

Abstract

This multi-ancestry meta-analysis of genome-wide association studies (GWAS) investigated the genetic factors underlying chronic back pain (CBP) in a sample from the Million Veteran Program comprised of 553,601 Veterans of African (19.2%), European (72.6%), and Hispanic (8.2%) ancestry. The results revealed novel (N = 67) and known (N = 20) genome-wide significant loci associated with CBP, with 43 independent variants replicating in a non-overlapping contemporary meta-GWAS of the spinal pain dorsalgia phenotype. The most significant novel variant was rs12533005 (chr7:114416000, p = 1.61 × 10^-20, OR = 0.96 (95% CI: 0.95-0.97), EA = C, EAF = 0.39), in an intron of the FOXP2 gene. In silico functional characterization revealed enrichment in brain and pituitary tissues. Mendelian randomization analysis of 62 variants for CBP-MVP revealed 48 with causal links to dorsalgia. Notably, four genes (INPP5B, DRD2, HTT, SLC30A6) associated with these variants are targets of existing drugs. Our findings more than double the number of previously reported genetic predictors across all spinal pain phenotypes.

Fig. 1. Multi-ancestry Manhattan plot of meta-analysis GWAS of EHR defined CBP-MVP.

The y axis represents -log₁₀ p-values from two-sided z-test for meta-analyses effect estimates. The dotted line represents genome-wide significance at p < 5 × 10⁻⁸.

Fig. 2

Post-Hoc MiXeR power estimates in MVP EUR, HIS and AFR strata.

Fig. 3. Genetic overlap between EUR ancestry male and female participants, using MiXeR.

Venn diagram reports numbers in thousands of variant risk loci (standard error) related to the first trait (gray circle), second trait (orange), and the mutually shared variants (gray overlap). Genetic correlation is depicted as a number and red progress bar. The two center plots show conditional QQ plots of trait 1 on subset 2 and trait 2 on subset 1. The model log-likelihoods based on the number of causal variants are shown in the rightmost plot. The solid blue line reflects the average likelihood across the 20 MiXeR runs. The dotted blue line reflect the likelihood of individual runs.

Fig. 4. Summary of 491 significant LDSC genetic correlations (r_gs) between EUR MVP-CBP and various traits presented by domain.

R_g is presented next to domain name, number of significant correlations out of total number per domain presented parenthetically. Triangles represent the average r_g per domain, and squares represent the minimum and maximum r_g within the domain.