Immunological Age Prediction in HIV-Infected, ART-Treated Individuals

Abstract

Abstract Anti-retroviral therapy (ART) improves life expectancy in people living with HIV (PWH), but it remains unclear how chronic HIV infection affects normal aging of the immune system. Plasma cell-free protein expression and immune phenotypes were assessed in blood from ART treated PWH (19–77yrs, n=106) and age-matched, HIV-negative controls (HC, n=103). Age-associated parameters were identified in HC and PWH independently by univariate Spearman correlation (p<0.05) to obtain a reduced dataset for downstream statistical modeling. Machine learning algorithms were applied to generate predictive models for age as a continuous variable. Plasma biomarker analysis resulted in 8-parameter inflammatory index that was higher in PWH compared to HC and correlated with age. Combining flow and plasma data generated 277 and 491 age-associated parameters (out of total 1,357) in HC and PWH, respectively. PWH exhibited different immunological profiles associated with age compared to HC. However, CD38 expression on T cells declined with age in both HC and PWH. Predictive modeling using age-associated parameters in HC generated a 14-parameter signature with 76.9% accuracy to predict chronological age in HC, however this model failed to predict age in PWH (18.6% accuracy) highlighting the effect of HIV infection on immunological aging. A 25-parameter model (IMmunological Age Prediction, IMAP-25) however showed 70% and 53% accuracy in HC and PWH and was used to show an accelerated aging rate in PWH by 5.6 yrs compared to HC. We applied a statistical approach to generate immune signatures, easily monitored in human blood samples, that could be used as an indicator of one’s ‘immunological age’ during ART-treated HIV infection.