Antiretroviral therapy suppressed participants withlow CD4RT-cell counts segregate according toopposite immunological phenotypes

Abstract

Background:The failure to increase CD4þT-cell counts in some antiretroviral therapysuppressed participants (immunodiscordance) has been related to perturbed CD4þT-cell homeostasis and impacts clinical evolution.Methods:We evaluated different definitions of immunodiscordance based on CD4þT-cell counts (cutoff) or CD4þT-cell increases from nadir value (DCD4) using super-vised random forest classification of 74 immunological and clinical variables from 196antiretroviral therapy suppressed individuals. Unsupervised clustering was performedusing relevant variables identified in the supervised approach from 191 individuals.Results:Cutoff definition of CD4þcell count 400 cells/ml performed better than anyother definition in segregating immunoconcordant and immunodiscordant individuals(85% accuracy), using markers of activation, nadir and death of CD4þT cells.Unsupervised clustering of relevant variables using this definition revealed largeheterogeneity between immunodiscordant individuals and segregated participants intothree distinct subgroups with distinct production, programmed cell-death protein-1(PD-1) expression, activation and death of T cells. Surprisingly, a nonnegligible numberof immunodiscordant participants (22%) showed high frequency of recent thymicemigrants and low CD4þT-cell activation and death, very similar to immunoconcor-dant participants. Notably, human leukocyte antigen - antigen D related (HLA-DR)PD-1 and CD45RA expression in CD4þT cells allowed reproducing subgroup segre-gation (81.4% accuracy). Despite sharp immunological differences, similar andpersistently low CD4þvalues were maintained in these participants over time.Conclusion:A cutoff value of CD4þT-cell count 400 cells/ml classified betterimmunodiscordant and immunoconcordant individuals than anyDCD4 classification.Immunodiscordance may present several, even opposite, immunological patterns thatare identified by a simple immunological follow-up. Subgroup classification may helpclinicians to delineate diverse approaches that may be needed to boost CD4þT-cellrecovery.