Predictors of HIV rebound differ by timing of antiretroviral therapy initiation

Abstract

BACKGROUNDIdentifying factors that predict the timing of HIV rebound after treatment interruption will be crucial for designing and evaluating interventions for HIV remission.METHODSWe performed a broad evaluation of viral and immune factors that predict viral rebound (AIDS Clinical Trials Group A5345). Participants initiated antiretroviral therapy (ART) during chronic (N = 33) or early (N = 12) HIV infection with ≥ 2 years of suppressive ART and restarted ART if they had 2 viral loads ≥ 1,000 copies/mL after treatment interruption.RESULTSCompared with chronic-treated participants, early-treated individuals had smaller and fewer transcriptionally active HIV reservoirs. A higher percentage of HIV Gag-specific CD8+ T cell cytotoxic response was associated with lower intact proviral DNA. Predictors of HIV rebound timing differed between early- versus chronic-treated participants, as the strongest reservoir predictor of time to HIV rebound was level of residual viremia in early-treated participants and intact DNA level in chronic-treated individuals. We also identified distinct sets of pre-treatment interruption viral, immune, and inflammatory markers that differentiated participants who had rapid versus slow rebound.CONCLUSIONThe results provide an in-depth overview of the complex interplay of viral, immunologic, and inflammatory predictors of viral rebound and demonstrate that the timing of ART initiation modifies the features of rapid and slow viral rebound.TRIAL REGISTRATIONClinicalTrials.gov NCT03001128FUNDINGNIH National Institute of Allergy and Infectious Diseases, Merck.

Keywords: AIDS vaccine; AIDS/HIV; Adaptive immunity.

Figure 1. Relationship between HIV reservoir measures and antibody levels before treatment interruption.

Correlation plots with Spearman rho values. ***P < 0.001, **P < 0.01, *P < 0.05. CA-RNA, unspliced cell-associated RNA; CA-DNA, total HIV proviral DNA; SCA, integrase single-copy assay; Total HIV-1, total HIV proviral DNA by the intact proviral DNA assay (IPDA); IUPM, infectious units per million resting CD4⁺ cells by the differentiation quantitative viral outgrowth assay; LAg-Avidity, HIV-1 limiting antigen avidity enzyme immunoassay.

Figure 2. HIV reservoir comparison between early- and chronic-treated participants.

Unspliced cell-associated HIV RNA, total HIV DNA by the intact proviral DNA assay (IPDA), intact proviral DNA (IPD) by the IPDA, and infectious units per million resting CD4⁺ cells (IUPM) by the differentiation quantitative viral outgrowth assay (dQVOA). Open circles represent values that are below the limit of quantification. Median lines and fold-change between early- and chronic-treated participants are shown. P values by the Wilcoxon rank sum test.

Figure 3. Higher percentage of HIV Gag-specific CD107a⁺ CD8⁺ T cells is associated with lower intact HIV-1 DNA levels.

Data points are color-coded by timing of ART initiation and rapid versus slow viral rebound. Rapid rebound is defined as meeting ART restart criteria < 4 weeks, and slow rebound is defined as meeting ART restart criteria on or after week 4.

Figure 4. Correlation of reservoir, immune, and inflammatory markers with timing of HIV rebound.

(A and B) Correlation map with lines connecting factors between categories (antibody, cellular, soluble inflammatory markers, and viral) meeting either (A) Spearman r > 0.35 and unadjusted P < 0.05 or (B) r < –0.35 and unadjusted P < 0.05. (C and D) Volcano plot of Spearman correlations with timing of HIV rebound in (C) acute/early-treated and (D) chronic-treated participants. Factors associated with earlier rebound are on the left and factors associated with delayed rebound on the right. Padj values were adjusted P values using the Benjamini-Hochberg method, and those with Padj < 0.25 are included given the exploratory nature of this analysis.

Figure 5. Viral and immune predictors of HIV rebound timing.

Sparse variant partial least squares discriminant analysis (sPLS-DA) of individuals with rapid (< 4 weeks) or slow (≥ 4 weeks) viral rebound. PLS-DA score derived from features after sPLS-DA feature down-selection, demonstrating distinction between rapid and slow viral rebound groups in (A) acute/early-treated and (C) chronic-treated participants. Variable importance in projection (VIP) scores for selected features in (B) acute/early-treated and (D) chronic-treated participants. expl. var, explained variance.