Persistent metabolic changes in HIV-infected patients during the first year of combination antiretroviral therapy

Abstract

The HIV-human metabolic relationship is a complex interaction convoluted even more by antiretroviral therapy (cART) and comorbidities. The ability of cART to undo the HIV induced metabolic dysregulation is unclear and under-investigated. Using targeted metabolomics and multiplex immune biomarker analysis, we characterized plasma samples obtained from 18 untreated HIV-1-infected adult patients and compared these to a non-HIV infected (n = 23) control population. The biogenic amine perturbations during an untreated HIV infection implicated altered tryptophan- nitrogen- and muscle metabolism. Furthermore, the lipid profiles of untreated patients were also significantly altered compared to controls. In untreated HIV infection, the sphingomyelins and phospholipids correlated negatively to markers of infection IP-10 and sIL-2R whereas a strong association was found between triglycerides and MCP-1. In a second cohort, we characterized plasma samples obtained from 28 HIV-1-infected adult patients before and 12 months after the start of cART, to investigate the immune-metabolic changes associated with cART. The identified altered immune-metabolic pathways of an untreated HIV infection showed minimal change after 12 months of cART. In conclusion, 12 months of cART impacts only mildly on the metabolic dysregulation underlying an untreated HIV infection and provide insights into the comorbidities present in virally suppressed HIV patients.

Figure 1

Plasma metabolic characterization of untreated HIV vs controls. Volcano plot of (a) the biogenic amine profile. (b) The lipid profile. (c) The signalling lipid profile. Volcano plots are representative of the −log10(Mann-Whitney p-value) on the y-axis with the x-axis showing the log2(Fold change) of the metabolite, between the controls and the untreated HIV patients (population A) respectively. Dashed lines represent the respective significance thresholds, with the significant metabolites identifiable by coloured symbols with either a name or corresponding class colour identifier. CE – Cholesterol esters; CER – Ceramides; PC – Phosphatidylcholines; PC-O - Plasmalogen PCs; PE – Phosphatidylethanolamines; PE-O – Plasmalogens PEs; SM - Sphingomyelins; TG - Triglycerides; LPA – Lysophosphatidic acid; COX - Cyclooxygenase; CYP450 – Cytochrome P450; LOX - Lipoxygenase; RNS – Reactive nitrogen species; ROS – Reactive oxygen species.

Figure 2

Plasma metabolic characterization of response to 12 months’ cART. Modified volcano plots showing (a) the biogenic amine profile, and (b) the lipid profile. The modified volcano plots are representative of the −log10(FDR adjusted paired t-test p-value) on the y-axis with the x-axis showing the scaled mean of the paired differences per metabolite, between the baseline and 12 months cART follow-up sample (patient population B). The horizontal dashed line represents a false discovery rate adjusted q-value < 0.05. Significant metabolites are identifiable by coloured symbols with either a name or corresponding class colour identifier. CE – Cholesterol esters; LPC – Lysophosphatidylcholines; PC – Phosphatidylcholines; PC-O - Plasmalogen PCs; PE – Phosphatidylethanolamines; PE-O – Plasmalogens PEs; SM – Sphingomyelins.

Figure 3

Immune-metabolic networks of controls, untreated HIV and the influence of cART. (a) Control immune-metabolic correlation network based on correlating the absolute levels of plasma metabolites and plasma cytokines in the control group. (b) Untreated HIV immune-metabolic correlation network based on correlating the absolute levels of plasma metabolites and plasma cytokines in the untreated HIV group (Population B). (c) The immune-metabolic correlation network following 12 months cART. Using the paired samples of population B, the relative change in levels of 12 months’ cART compared to their baseline levels (untreated HIV sample) were determined for plasma metabolites and cytokines and subsequently correlated. For all three networks, spearman correlations were performed and significant correlations were defined as having a strict p-value (two-tailed) < 0.01.