Abstract Body


Most people with HIV exhibit viral rebound within weeks of analytical treatment interruption (ATI), but some can control replication for prolonged periods of time. The mechanisms mediating control upon ATI remain poorly understood, and may involve Natural Killer (NK) cells, potent antiviral immune effectors. The objective of this study was to use CyTOF to identify NK cell features that can predict time-to-rebound upon ATI.


We analyzed samples from two ATI cohorts: the REDUC cohort, which consisted of 17 participants that received therapeutic vaccination (Vacc4x) with the HDAC inhibitor Romidepsin prior to ATI, and the ACTG A5345 cohort, which consisted of 33 chronic- and 12 acute-treated participants that underwent ATI in the absence of any intervention. We phenotyped pre-ATI PBMCs using an NK cell CyTOF panel that includes markers of differentiation and activation states, homing receptors, inhibitory and activating receptors, and intracellular factors. We performed unsupervised clustering analyses to define NK cell clusters, optimized the number of clusters with a leave-one-out cross-validation model, and looked for associations with time-to-rebound, with HIV reservoir size as determined by IPDA, and with time of treatment initiation (chronic vs. acute).


Clustering analyses for the REDUC cohort revealed a cluster that was significantly (p< 0.05) associated with increased time-to-rebound: an individual with 2.1 times the average number of cells belonging to this cluster rebounds on average 3 days later. The NK cells in this cluster were atypical in that they were CD56-CD16+ and exhibited a memory signature. Clustering analyses for the ACTG A5345 cohort revealed that individuals with a lower frequency of intact provirus were significantly (p< 0.01) associated with a different cluster of CD56dimCD16+ NK cells expressing high levels of the inhibitory receptor Siglec-7 associated with functional NK responses. Acute-treated participants were also more likely to have NK cells that expressed high levels of the memory NK cell antigens CXCR6 (p< 0.01) and CD49a (p=0.06).


As CD56-CD16+ NK cells have been associated with bnAb production, these results suggest a potential role of NK cells and HIV-specific antibodies in control of viral replication. Siglec-7+ NK cells may temporarily control viral rebound during ATI. The findings with CXCR6 and CD49a implicate a role for memory NK cells in the preserved immune responses found in acute-treated individuals.