This paper is in Nature Communications is here: http://go.nature.com/2BgeQKk
In recent years, my lab focused on understanding whether epigenetic processes mediate the impact of substance abuse on HIV and other medical outcomes. The intravenous use of illicit drugs (IDU) is a major risk factor for infection with HIV, hepatitis, and other blood-born pathogens. However, it is often forgotten that ongoing IDU is a risk factors for poorer medical outcomes for infected individuals. We hypothesized that IDU might induce changes in the genome-wide pattern of methylation in critical elements of the immune system, particularly various populations of white blood cells (WBC), that produce lasting changes that undermine immune function and result in poorer medical outcomes.
The current study tested this hypothesis. Using epigenome-wide association analysis, we previously identified DNA methylation sites in the epigenome of human WBCs that differed significantly between HIV-infected and uninfected individuals. This study reports the impact of IDU and HIV on the methylation status of these CpG sites and on a particularly important health outcome measure derived from several facets of clinical status, called HIV clinical frailty.
In other words, we profiled genome-wide DNA methylation of CpG sites in DNA samples extracted from HIV-infected blood from individuals with and without IDU. After adjusting for a number of confounding factors, we identified 6 epigenome-wide significant signals located in the genes involving immune function and HIV pathogenesis. Enrichment analysis also revealed biological pathways involving processes implicated in antigen presentation. We further bioinformatically classified IDU/HCV versus non-IDU/non-HCV phenotypes based on methylation signatures.
In a replication sample, 6 CpG sites identified in the discovery sample showed non-significant trends for similar association with IDU in HIV-individuals. We believe that the failure to achieve statistical significance in the second sample was due to reduced statistical power in that analysis. Combining the two sample sets, we replicated 6 CpG sites and identified additional 4 CpG sites.
Our ultimate goal is to identify a methylation signature to predict HIV outcomes. We then applied machine-learning methods to predict HIV frailty. We used the discovery sample as a training set and an independent sample served as a testing set. We used 789 probes that clustered two phenotypes and robustly differentiated patients with an area under the curve of 79.3% for high frailty and 82.3% for low frailty.
A limitation of the current study was that IDU was highly comorbid with HCV, mirroring the characteristics of IDU populations. The high rate of IDU and HCV comorbidity prevented us from being able to fully dissociate IDU and HCV effects in the current study.
The current study suggests that epigenetic analyses may yield important clinical biomarkers related to HIV outcomes, particularly the interaction of IDU and HIV. Further validation of these epigenetic marks is justified, particularly with regards to understanding the pathophysiology of the convergent effects of IDU and HIV.
Ke Xu, MD, PhD, Assistant Professor of Psychiatry, Yale school of Medicine