Longitudinal patterns of inflammatory mediators after acute HIV infection correlate to intact and total reservoir

Abstract

Background: Despite the beneficial effects of antiretroviral therapy (ART) initiation during acute HIV infection (AHI), residual immune activation remains a hallmark of treated HIV infection.

Methods: Plasma concentrations of 40 mediators were measured longitudinally in 39 early treated participants of a Belgian AHI cohort (HIV+) and in 21 HIV-negative controls (HIV-). We investigated the association of the inflammatory profile with clinical presentation, plasma viral load, immunological parameters, and in-depth characterization of the HIV reservoir.

Results: While levels of most soluble mediators normalized with suppressive ART, we demonstrated the persistence of a pro-inflammatory signature in early treated HIV+ participants in comparison to HIV- controls. Examination of these mediators demonstrated a correlation with their levels during AHI, which seemed to be viremia-driven, and suggested involvement of an activated myeloid compartment, IFN-γ-signaling, and inflammasome-related pathways. Interestingly, some of these pro-inflammatory mediators correlated with a larger reservoir size and slower reservoir decay. In contrast, we also identified soluble mediators which were associated with favorable effects on immunovirological outcomes and reservoir, both during and after AHI.

Conclusion: These data highlight how the persistent pro-inflammatory profile observed in early ART treated individuals is shaped during AHI and is intertwined with viral dynamics.

Keywords: HIV reservoir; acute HIV infection; chemokines; cytokines; early ART; inflammation; plasma soluble mediators.

Figure 1

Frequency of acute retroviral syndrome (ARS) and ARS signs and symptoms at the time of HIV diagnosis.

Figure 2

The intact HIV reservoir declines faster than the total HIV reservoir after early treated acute HIV infection. (A, B) Box plots displaying total (A) and intact (B) HIV DNA measurements (log10 transformed) at T0, UD, and UD+1 for HIV+ participants diagnosed and treated in Fiebig stages II-III (purple) and Fiebig stages IV-VI (green). Dots are colored according to Fiebig stages. P-values were calculated using Mann Whitney U tests. (C) Box plots displaying total (empty circles) and intact (filled circles) HIV DNA measurements (log10 transformed) at T0, UD, and UD+1. P-values comparing measurements across timepoints were calculated with Friedman's test and post-hoc Dunn's test. P-values comparing total and intact HIV DNA were calculated using paired Wilcoxon signed-rank tests. (D) Box plot displaying the ratio of intact to total HIV DNA (%) at UD (orange) and UD+1 (yellow). P-value was calculated using a paired Wilcoxon signed-rank test. (E), Dot plot displaying the decline in HIV reservoir between UD (orange) and UD+1 (yellow) in longitudinally paired samples for total (empty circles) and intact (filled circles) HIV DNA. (F) Violin plot displaying the difference in fold decrease of the HIV reservoir between UD and UD+1 for total (empty circles) and intact (filled circles) HIV DNA. P-value was calculated using a paired Wilcoxon signed-rank test.

Figure 3

Acute HIV-1 infection is characterized by a pleiotropic release of systemic soluble mediators with distinct temporal patterns. Heatmap displaying individual batch-corrected log2 transformed, row-scaled levels of soluble mediators for early treated HIV-positive (HIV+) and HIV-negative (HIV-) participants per timepoint. Columns represent plasma samples at the different timepoints. Rows represent soluble mediators and are clustered with Spearman's rank correlation coefficient as distance metric.

Figure 4

The persistent pro-inflammatory signature after early ART correlates to pretreatment inflammation, which is likely driven by viremia. (A) Box plots displaying plasma concentrations of CXCL11, CXCL10, and IL-4 at the UD+1 timepoint in HIV+ participants presenting with (orange) and without (blue) ARS at diagnosis and HIV- participants (grey). P-values were calculated using a linear model with batch as a fixed factor. (B) Box plots displaying plasma concentrations of IFN-a2a and CXCL10 at the UD+1 timepoint in HIV+ participants diagnosed and treated in Fiebig stages II-III (purple) and Fiebig stages IV-VI (green) and HIV- participants (grey). Dots are colored according to Fiebig stages. P-values were calculated using a linear model with batch as a fixed factor. (C, D) Correlation matrixes between plasma concentrations of soluble mediators during suppressed viremia (UD+1) and laboratory parameters at baseline (T0) (C) and plasma concentrations of soluble mediators at baseline (T0) (D). Correlation coefficients and p-values were calculated with Spearman's rank test for pairwise complete observations. Positive and negative correlations are depicted in red and blue, respectively. Significant correlations are indicated by *, p<0.05; **, p<0·01; ***, p<0·001. (E) Correlation matrix between plasma concentrations of soluble mediators and laboratory parameters of HIV+ participants during acute HIV infection (T0). Correlation coefficients and p-values were calculated with Spearman's rank test for pairwise complete observations. Positive and negative correlations are depicted in red and blue, respectively. Significant correlations are indicated by *, p<0·05; **, p<0.01; ***, p<0·001. All soluble mediator concentrations were batch-corrected and log2 transformed. ARS, acute retroviral syndrome; AST, aspartate aminotransferase; VL, viral load; ns, not significant.

Figure 5

The CD4/CD8 ratio and time to viral suppression are correlated with distinct soluble mediators following acute HIV infection. (A), Correlations between CD4/CD8 ratio during suppressed viremia (UD+1) and plasma concentrations of CXCL10 at T0 (burgundy) and UD+1 (yellow). (B) Correlations between time to undetectable HIV viral load and plasma concentrations of CXCL1 at T0 (burgundy) and CX3CL1 at UD+1 (yellow). All soluble mediator concentrations were batch-corrected and log2 transformed. Correlation coefficients and p-values were calculated with Spearman's rank test. VL, viral load.

Figure 6

Inflammatory profiles in the course of acute HIV infection show bidirectional correlations to the viral reservoir. (A, B) Correlations of HIV reservoir at UD+1 in relation with soluble mediators at UD+1. Inverse correlation of total HIV DNA with plasma concentrations of CX3CL1 (A) and correlation of intact HIV DNA (B) with plasma concentrations of CX3CL1, CXCL11, and IL-27. (C, D) Correlations of HIV reservoir at UD+1 with soluble mediators at T0. Inverse correlation of total HIV DNA in relation with plasma concentrations of GM-CSF (C) and correlation of intact HIV DNA (D) in relation with plasma concentrations of IL-2 and TNF-β (inverse) and IL-27. (E) Correlations of fold decrease in total HIV DNA between UD and UD+1 in relation with plasma concentrations of CX3CL1 and CXCL8 (inverse) at UD+1. (F) Correlations of fold decrease in intact HIV DNA between UD and UD+1 in relation with plasma concentrations of CXCL8 (inverse) at UD+1. All concentrations were batch-corrected and log2 transformed. Correlation coefficients and p-values were calculated with Spearman's rank test.