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

TURK, GABRIELA; LAUFER, N; TRIFONE, C; HOLGADO, MP; SALOMON, H; GHERARDI, MM; SUED, O; HOSMANSTORFER, M; SALIDO, J; FALIVENE, J; FIGUEROA, MI; PANDO, MA; PURY, P; GHIGLIONE, Y; COLOCCINI, R; RUIZ, MJ; CARUSO, MP; GIAVEDONI, L; RABINOVICH, RD

Paris

Conferencia; 9th IAS Conference on HIV Science (IAS 2017); 2017

Background: Progression of HIV infection is variable among individuals. Despite implementationof effective ART, definition of disease progression biomarkers is stillfundamental. Apart from CD4+ T-cell count (CD4TC) and viral load(VL), several parameters have been individually proposed as biomarkers by ourgroup and others. Here, we aimed to categorize their predictive potential usingdecision trees and analyze their possible implementation in the clinicalsetting.Methods: A total of seventy-five subjects were enrolled during acute/early HIVinfection (< 6 months postinfection). CD4TC and VL determinations wereperformed at enrollment (baseline sample) and during 1 year. This study only includedsamples and data from subjects while off-treatment. Immune activation (HLA-DRand CD38 expression), HIV-specific immune response (ELISPOT) and HLA haplotypewere determined in a subset of 41 individuals at baseline sample. Within thisgroup, plasma levels of 39 cytokines were determined by Luminex in 27individuals. Progression was defined as CD4TC decreasing below 350 cells/μl or experiencing AIDS-related B/C events within12 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 treeResults: Plasma IL-10, IP-10, sIL-2R and TNF- directly correlated withbaseline VL while IL-2, TNF-, FGF-2 and Mip-1 inversely correlatedwith CD4+ T-cell activation (p< 0.05). However, none of thesecytokines had good predictive value to distinguish progressors fromnon-progressors. Similarly, immune activation, HIV-specific immune responsesand HLA haplotypes had lower discrimination power when compared to clinicalparameters (CD4TC and VL). Baseline CD4TC was the most potent variable todistinguish progressors from non-progressors with a cut-off of 436 cells/μl (accuracy=0.93, k-Cohen=0.85)Conclusions: In our cohort, baseline CD4TC was the strongest predictor of diseaseprogression early after infection. Limited discerning power of the otherfactors might be related to frequency, variability and/or sampling time. Surprisingly,high baseline CD4TCs were observed even in subjects that progressed rapidly,reinforcing the importance of early ART initiation. Also, efforts should bemade to develop and make available CD4TC determination techniques to allpossible settings. Future studies basedon decision trees to identify biomarkers of posttreatment control arewarrantied.