Introduction It is now anticipated that HIV-infected adults who have access to modern antiretroviral therapy (ART) should be able to suppress HIV replication indefinitely. Although treatment-mediated increases in the peripheral CD4 count are associated with reduced morbidity and mortality, compared to age-matched individuals without HIV infection, those on ART have a higher risk of morbidity and mortality. This risk is predicted in part by the on therapy CD4 count, although achieving an apparent normal CD4 count may not fully restore health [1]–[5]. Indeed, it has been shown that even those treated patients with CD4+ T cell counts above 500 cells/mm3, a further CD4+ T cell count increase is still associated with a slight benefit in terms of mortality [6]. The decreased life expectancy during ART-mediated viral suppression is largely explained by a higher than expected risk of non-AIDS-morbidity, a term that entails a group of conditions generally associated with aging, including cardiovascular, renal, liver, neurologic, and bone disease, as well as cancer [4], [7], [8]. While the mechanisms driving the increased burden of aging-associated disease in HIV-infected individuals are not fully understood, an emerging body of evidence suggests that persistent innate and adaptive immune dysfunction and/or activation are major risk factors [9]–[12]. Many of the immunologic abnormalities that persist during therapy are similar to those observed in the elderly, raising the hypothesis that age-associated decline in immune function (“immunosenescence”) contributes to disease progression and adverse outcomes [13]–[16]. Markers of innate immune activation [e.g. interleukin (IL)-6, high-sensitivity C reactive protein (hs-CRP) and soluble CD14 (sCD14)], coagulation (fibrinogen, D-dimers), bacterial translocation (lipopolysaccharide), and T cell activation (HLADR and CD38 co-expression) are elevated despite effective ART and associated with subsequent morbidity and mortality, even after adjustment for CD4+ T cell count [17]–[21]. Induction of indoleamine 2,3-dioxygenase-1 (IDO) in monocytes and dendritic cells occurs during HIV infection and has been associated with impairment of the mucosal immunity and the maintenance of a chronic inflammatory state [22]. Collectively, these observations strongly suggest that an underlying mechanism not captured by CD4+ T cell count and HIV replication might be contributing to disease progression. The importance of CD4 counts as a strong predictor of opportunistic infections and non-AIDS events has been widely investigated, but little attention has been paid to the prognostic significance of CD8 counts. During untreated HIV infection, CD8 counts increase as CD4 counts decline [23]. During ART-mediated viral suppression, some individuals achieving CD4 counts above 500 cells/mm3 experience a simultaneous decline in CD8 counts, leading to normalization of the CD4/CD8 ratio. Others, however, maintain high levels of circulating CD8+ T cells, and hence a persistently low CD4/CD8 ratio [24]. Among elderly HIV-uninfected adults, inversion of the CD4/CD8 ratio ( 0.1% pp65/IE-specific IFN-γ+ CD8+ T cell responses by cytokine flow cytometry (ten-fold increase over limit of detection) as previously described [44]. Statistical methods Cross-sectional pairwise comparisons between groups were performed using Wilcoxon rank sum tests. Since a “normal” CD4/CD8 ratio remains poorly defined, for the between-group comparisons of T cell subsets and percentages of activated/senescent CD8+ T cells, we classified individuals according to the lowest quartile (≤0.4) and highest quartile (≥1.0) of SCOPE participants with ≥500 CD4+ T cells/mm3. A CD4/CD8 ratio ≤0.4 has been defined previously as the best cutoff that may predict serious non-AIDS events in well-treated HIV-infected patients [36], and 1.0 has been suggested in the general population as the cutoff for the “immune risk profile” associated with immunosenescence and mortality [26], [45]. To assess the intra-individual variability of the CD4/CD8 ratio, we used data from the control arms of ART intensification trials with raltegravir [38] and maraviroc [39], [40] to calculate the coefficient of variation (standard deviation/mean) for the CD4+ and CD8+ T cell counts and for the CD4/CD8 ratio. To analyze the association between the CD4/CD8 ratio and the KT ratio in SOCA, we fitted a linear regression model using CD4/CD8 ratio as the dependent variable, and KT ratio as the explanatory variable, adjusting the model by age, gender, time under viral suppression and CD4 nadir. To evaluate the relative contribution of the CD4+ and CD8+ T cells to this association, we also fitted another model with both CD4+ and CD8+ T cells in which the CD4/CD8 ratio was not considered because of colinearity, adjusting for the same covariates. We analyzed the correlations between the CD4/CD8 ratio in blood, with the ratio in lymph nodes and in GALT. For the GALT CD4/CD8 ratio measured in the MVC and RAL studies, we used only baseline measurements (before ART intensification). Since a different panel of antibodies was used for each study for flow-cytometry analysis, we fitted a linear regression analysis adjusting by the source study. We analyzed the impact of early ART initiation on the CD4/CD8 ratio in the OPTIONS cohort among recently HIV-infected participants, focusing on those who either started ART within six months of infection (early ART) or who deferred therapy for at least two years (later ART) [37]. Longitudinal changes in CD4 and CD8 counts and in the CD4/CD8 ratio were assessed using linear mixed models with random intercepts. Age, gender, and pre-ART CD4 counts were included in multivariate analyses as fixed-effects. Interaction terms were created to assess whether these changes over time differed significantly between the early and later ART initiators. Changes in slopes before and after ART time points were assessed using linear splines. We used data from the Madrid and SOCA cohorts to evaluate whether the CD4/CD8 ratio might be a marker of non-AIDS-related morbidity and mortality, respectively. In the nested case-control analysis in the Madrid cohort, cases who developed serious non-AIDS events and had ≥500 CD4+ T cells/mm3, were each matched to one controls by age, sex, nadir CD4, and proximal CD4 counts (N = 66). In the nested case-control study of immunological predictors of mortality in SOCA, cases with non-accidental death who had PBMC and plasma samples available within 18 months of death with confirmed plasma HIV RNA levels 500 cells/mm3 stratified by a normal (4th quartile, ≥1) or low (1st quartile, ≤0.4) CD4/CD8 ratio. HIV-infected individuals with low CD4/CD8 ratio had lower percentages of TN, TCM, and TTR CD8+ cells, higher TEM and TEMRA (A), and higher absolute counts (B) of all subsets compared to those with higher CD4/CD8 ratio and with healthy controls. 10.1371/journal.ppat.1004078.g002 Figure 2 Percentages and absolute counts of CD8+ activation phenotypes among HIV-/CMV+ individuals and ART-suppressed HIV-infected patients with CD4 counts >500 cells/mm3 stratified by a normal (4th quartile, ≥1, in green) or low (1st quartile, ≤0.4, in red) CD4/CD8 ratio. Subjects with low CD4/CD8 ratio showed higher percentages (A) and absolute counts (B) of HLADR+, CD28− and CD28−CD57+, and higher absolute counts of PD1+ cells (B). There were no differences in HIV-infected individuals in the proportion of CD28−CD8+ T cells expressing CD57, being significantly lower in both groups compared to HIV-/CMV+ controls. We sought to validate these findings among effectively treated subjects (undetectable viral load, ≥500 CD4+ T cells/mm3) within the SOCA cohort (general characteristics summarized in Table S3 ), and found comparable correlations between the CD4/CD8 ratio and different phenotypes of activated/senescent CD8+ T cells among ART-suppressed subjects with CD4>500 T cells/mm3. The most consistent correlates of the CD4/CD8 ratio were the %HLADR+CD38+ CD8+ T cells (Rho = −0.507, P 500 cells/mm3 stratified by a normal (4th quartile, ≥1) or low (1st quartile, ≤0.4) CD4/CD8 ratio. Individuals with low CD4/CD8 ratio had decreased frequencies of CD4+ TTR and decreased absolute counts of TN, TCM, and TTM CD4+ T cells compared to those HIV-infected patients with normal CD4/CD8 ratio and with healthy controls. (TIF) Click here for additional data file. Figure S3 Intra-individual variability of the CD4/CD8 ratio compared to CD4+ and CD8+ T cell counts. Using data from 38 HIV-infected patients on ART-mediated HIV-RNA suppression in whom a median of 11 determinations of CD4+ and CD8+ T cells measurements were performed during a median of 81 weeks, we calculated the coefficient of variation –within subject standard deviation (blue lines) and the within subject mean (red plus symbols)– for the CD4+ T cell counts, CD8+ T cell counts and the CD4/CD8 ratio. The mean coefficient of variation was significantly lower for the CD4/CD8 ratio (12%) compared to CD4+ T cell counts (16%, P = 0.017) and for CD8+ T cell counts (18%, P = 0.001). (TIF) Click here for additional data file. Table S1 Antibodies used for T-cell immunophenotyping. (DOCX) Click here for additional data file. Table S2 Characteristics of chronically HIV-infected participants and HIV negative controls in SCOPE. (DOCX) Click here for additional data file. Table S3 Characteristics of HIV-infected participants in SOCA cohort. (DOCX) Click here for additional data file. Table S4 General characteristics of participants in the lymph node and GALT analysis. (DOCX) Click here for additional data file. Table S5 General characteristics of OPTIONS participants. (DOCX) Click here for additional data file. Table S6 General characteristics of participants in the Madrid cohort nested study. (DOCX) Click here for additional data file. Table S7 Description of non-AIDS events in the Madrid cohort and causes of death in the SOCA cohort. (DOCX) Click here for additional data file. Text S1 Additional information on the cohorts and the clinical trials analyzed in this work. (DOCX) Click here for additional data file.