A multi-trait epigenome-wide association study identified DNA methylation signature of inflammation among people with HIV

Abstract

Inflammation underlies many conditions causing excess morbidity and mortality among people with HIV (PWH). A handful of single-trait epigenome-wide association studies (EWAS) have suggested that inflammation is associated with DNA methylation (DNAm) among PWH. Multi-trait EWAS may further improve statistical power and reveal pathways in common between different inflammatory markers. We conducted single-trait EWAS of three inflammatory markers (soluble CD14, D-dimers, and interleukin 6) in the Veteran Aging Cohort Study (n = 920). The study population was all male PWH with an average age of 51 years, and 82.3% self-reported as Black. We then applied two multi-trait EWAS methods-CPASSOC and OmniTest-to combine single-trait EWAS results. CPASSOC and OmniTest identified 189 and 157 inflammation-associated DNAm sites respectively, of which 112 overlapped. Among the identified sites, 56% were not significant in any single-trait EWAS. Top sites were mapped to inflammation-related genes including IFITM1, PARP9 and STAT1. These genes were significantly enriched in pathways such as "type I interferon signaling" and "immune response to virus". We demonstrate that multi-trait EWAS can improve the discovery of inflammation-associated DNAm sites, genes, and pathways. These DNAm sites suggest molecular mechanisms in response to inflammation associated with HIV and might hold the key to addressing persistent inflammation in PWH.

Keywords: D-dimer; EWAS; HIV; IL-6; sCD14.

Figure 1

Outline of multi-trait epigenome-wide associations studies of inflammatory markers in the Veteran Aging Cohort Study (VACS).

Figure 2. Quantile-quantile (QQ) and Manhattan plots from meta-analyses of multi-trait epigenome-wide associations studies in VACSn450K and VACS-850K.

(A) and (B) show the results from the cross-phenotype association test (CPASSOC) while (C) and (D) show the results from the omnibus test (OMNITEST). QQ plots (A and C) compare the observed P-values to the expected P-values. X-axis represents −log10(expected P-values) and Y-axis represents −log10(observed P-values). The red lines indicate the distribution of expected P-values (solid diagonal line) and their 95% confidence interval (dotted lines). The inflation factors estimate the amount of inflation by comparing observed P-values to expected P-values under the hypothesis of no effect. Dashed lines in (B and D) indicate genome-wide significance after adjusting for multiple testing.

Figure 3

Overlapping CpG sites identified by the cross-phenotype association test (CPASSOC) and the omnibus test (OMNITEST) in (A) VACS-450K, (B) VACS-850K and (C) meta-analyses.

Figure 4

Comparison between the beta-coefficients of the CpG sites associated with (A) IL-6, (B) sCD14, (C) D-Dimer in VACS and the ones associated with C-reactive protein in the general population.

Figure 5. Gene set enrichment analysis by missMethyl using CpG sites identified from meta-analyses of multi-trait epigenome-wide associations studies in VACS-450K and VACS-850K.

(A) shows the top 20 GO terms from the cross-phenotype association test (CPASSOC) while (B) shows the top 20 GO terms from the omnibus test (OmniTest). Blue bars indicate consistently significant top 20 GO terms between CPASSOC and OmniTest and grey bars indicate unique top 20 GO terms in each method.