Biomarkers of progression after HIV acute/early infection: Nothing compares to CD4+ T-cell count?

GABRIELA TURK; NATALIA LAUFER; CÉSAR TRIFONE; MARÍA PÍA HOLGADO; HORACIO SALOMÓN; MARÍA MAGDALENA GHERARDI; YANINA GHIGLIONE; ROMINA COLOCCINI; OMAR SUED; MARÍA JULIA RUIZ; MARÍA PAULA CARUSO; LUIS D. GIAVEDONNI; DANIEL RABINOVICH; MACARENA HORMANSTORFER; JIMENA SALIDO; JULIANA FALIVENE; MARÍA INÉS FIGUEROA; MARÍA DE LOS ÁNGELES PANDO; PEDRO A. PURY

Paris

Conferencia; International AIDS Conference (AIDS 2017); 2017

Background: Progression of HIV infection is variable amongindividuals. Despite implementation of effective ART, definition of diseaseprogression biomarkers is still fundamental. Apart from CD4+ T-cellcount (CD4TC) and viral load (VL), several parameters have been individuallyproposed as biomarkers by our group and others. Here, we aimed to categorizetheir predictive potential using decision trees and analyze their possibleimplementation in the clinical setting. Methods: A total of seventy-five subjects were enrolledduring acute/early HIV infection (<6 months postinfection). CD4TC and VLdeterminations were performed at enrollment (baseline sample) and during 1year. This study only included samples and data from subjects whileoff-treatment. Immune activation (HLA-DR and CD38 expression), HIV-specificimmune response (ELISPOT) and HLA haplotype were determined in a subset of 41individuals at baseline sample. Within this group, plasma levels of 39cytokines were determined by Luminex in 27 individuals. Progression was definedas CD4TC decreasing below 350 cells/ml or experiencing AIDS-related B/C events within 12 monthspost-infection. Data was analyzed by machine learning and non-parametricmethods and adjusted for multiple comparisons. Variable hierarchization wasperformed by Weka correlation based feature selection and J48 decision tree. Results: Plasma IL-10, IP-10, sIL-2Ra and TNF-a directly correlated with baseline VL whileIL-2, TNF-a, FGF-2 andMip-1b inversely correlated with CD4+T-cell activation (p<0.05). However, none of these cytokines had good predictivevalue to distinguish progressors from non-progressors. Similarly, immune activation,HIV-specific immune responses and HLA haplotypes had lower discrimination powerwhen compared to clinical parameters (CD4TC and VL). Baseline CD4TC was themost potent variable to distinguish progressors from non-progressors with acut-off of 436 cells/ml (accuracy=0.93, k-Cohen=0.85). Discussion: In our cohort, baseline CD4TC was thestrongest predictor of disease progression early after infection. Limiteddiscerning power of the other factors might be related to frequency,variability and/or sampling time. Surprisingly, high baseline CD4TCs wereobserved even in subjects that progressed rapidly, reinforcing the importanceof early ART initiation. Also, efforts should be made to develop and makeavailable CD4TC determination techniques to all possible settings.  Future studies based on decision trees toidentify biomarkers of posttreatment control are warrantied.