Background
Antiretroviral therapy (ART) is an effective treatment for people living with HIV (PLWH) that can suppress the virus and prevent disease progression. However, the virus persists latently and rebounds in a majority of PLWH during analytical treatment interruption (ATI) with variation in time to rebound. Currently, there is a lack of unbiased identification of ‘omics’ biomarkers that predict viral rebound. The RV254 Thai cohort of PLWH who initiated ART during acute HIV-1 infection (AHI) then underwent ATI provided a unique opportunity for us to use multiomic profiling and machine learning (ML) with explainable AI to identify signatures that could predict viral rebound following ATI.
Methods
Peripheral blood cells and plasma were obtained from 27 PLWH from ATI studies with no other intervention in Thailand. Participants were diagnosed during AHI and were on ART for a median of 3 years. Covariates that affect viral load (VL) rebound were first evaluated using ML. Longitudinal multiomics data including proteomics, simultaneous single-cell transcriptomics, chromatin accessibility and surface protein expression at AHI (week 0), 60 weeks after ART initiation, and immediately prior to ATI were analyzed. Cell state identification, differential expression, and pathway analyses were performed to identify markers of rebound.
Results
ML approaches identified faster time to VL suppression upon ART initiation as the strongest predictor of time to delayed rebound which was then used as a covariate in our multiomics analyses. Single-cell CITE-seq differential expression analyses showed that CD8+ T cells had the highest number of genes associating with time to rebound at all three timepoints. Surface protein markers measured at the ART timepoints of delayed rebound were markers of T cell activation (CD2, CD48) and proliferation (CD45), while early rebound markers were related to inhibition of immune responses (CTLA-4, PD1, PD-L1).
Conclusions
Single-cell omics methods identified the immune population with the strongest effect on HIV-1 rebound. Specific markers of CD8+ T cell activation or inhibition measured during ART associated with time to rebound. Understanding the CD8+ T cell phenotype associated with time to rebound using pathway analysis and integration of other clinical and immunologic data could provide further insight into how natural variation can impact time to viral rebound and inform new intervention strategies focused on CD8+ T cell activation to advance HIV-1 remission.
Table or Figure
