Unsupervised proteomic stratification of people living with HIV reveals inflammatory biotypes beyond conventional clinical classifications.

BACKGROUND: Conventional HIV-1 management relies on clinical phenotypes, assuming a uniform inflammatory status and overlooking biological heterogeneity. We aimed to employ a high-throughput proteomic approach to stratify people living with HIV (PLWH) based on plasma inflammatory profiles and evaluate their association with comorbidity burden. METHODS: This cross-sectional study included 30 PLWH (10 treatment-naïve, 10 ART-suppressed, and 10 elite controllers (EC)) and 10 seronegative controls. Unsupervised hierarchical clustering was applied to 87 inflammatory proteins (analyzed by proximity extension assay and ELISA). A Random Forest model stabilized with 1,000 trees identified the variables defining the biotypes. Comorbidity burden was assessed within the virally suppressed groups (ART-suppressed and EC). RESULTS: Clustering revealed two distinct biological profiles independent of clinical classifications: a high-inflammation biotype (n=11) and a low-inflammation biotype (n=19). Within the treatment-naïve, elite controller, and ART-suppressed groups, 60%, 30%, and 20% of participants, respectively, were classified as belonging to the high-inflammation biotype. PDL1, CD40, TNF, LAP-TGFbeta1, and SLAMF1 were the top biological drivers of this separation. A trend for a higher comorbidity burden was observed in the high-inflammation biotype compared to the low-inflammation group (p=0.06). CONCLUSIONS: Plasma inflammatory profiling stratifies PLWH into distinct biological biotypes that bypass conventional classifications. The observed differences in comorbidity burden between inflammatory biotypes warrant further investigation in larger prospective studies to clarify their clinical relevance and potential implications for long-term health outcomes.