Antiretroviral therapy (ART) leads to a dramatic reduction in HIV-related morbidity and mortality [1] and currently suppresses viral replication in the vast majority of compliant patients [2]. Most treatment guidelines recommend early initiation of ART, based mainly on nucleoside reverse transcriptase inhibitors (NRTI) combined with non-nucleoside reverse transcriptase inhibitors (NNRTI) or protease inhibitors (PI) [3, 4]. However, despite viral suppression and quantitative immune restoration in most patients, signs of immune activation and inflammation persist [5, 6]. This may be due in part to ongoing low-level viral replication [7], the coinfection with other chronic viruses such as cytomegalovirus and Epstein-Barr virus [8, 9] and to the consequences of mucosal immune dysfunction that is characterized by a profound depletion of CD4 + T-cells during the early acute infection, and a progressive loss of the ability to maintain the intestinal barrier function, allowing translocation of the intestinal microbial flora into the systemic circulation which induces immune activation and inflammation cascades [10]. An additional factor that may contribute to this persistence is the perturbation of immune regulatory mechanisms, such as regulatory T-cells or the immune regulatory cytokines such as interleukin-10 [11].

By comparison with HIV-uninfected individuals, several studies have shown an excess of comorbidities in HIV-infected patients on virologically effective treatment, including metabolic disorders [12], decreased bone mineral density with an increased risk of fractures [13], an elevated cardiovascular risk [14], and cancers [15]. These comorbidities have been attributed to a higher prevalence of standard risk factors such as smoking in HIV-infected patients, as well as to past or present HIV-induced immune depression reflected by the CD4 cell nadir or the CD4/CD8 cell ratio [16, 17], and to the effect of some antiretroviral drugs [12, 18, 19]. In addition, elevated levels of markers of inflammation, such as interleukin-6 (IL-6) and C-reactive protein (CRP), have been linked to increased morbidity and mortality in both the general and HIV-infected populations [20–22]. The SMART study was the first to demonstrate that mortality in this setting is linked to HIV replication, inflammatory markers such as IL-6 and soluble CD14 (sCD14), and coagulation markers such as D-dimers [23]. Since then, several studies have shown an association between plasma markers of inflammation and the risk of non AIDS-defining events [24, 25]. In the ALLRT cohort, elevated levels of IL-6, sCD14, D-dimers and soluble tumor necrosis factor receptors (sTNFR1 and sTNFR2), both prior to and during ART, were associated with the occurrence of non AIDS-related morbidities and death [26].

Now that ART is recommended for all HIV-infected patients, and life expectancy is greatly prolonged by suppressive ART despite the lack of viral eradication, it is important to compare the potential impact of different antiretroviral strategies on residual immune activation and inflammation. There are some arguments to suggest that different drugs from different classes may have different impact on immune activation and inflammation. In the Spiral study, a switch from PI-based therapy to a raltegravir-containing regimen in patients with suppressed viremia led to a decrease in biomarkers associated with inflammation, insulin resistance and hypercoagulability [27].

Few studies have examined the impact of first-line ART on biomarkers of inflammation and immune activation, and they have given discordant results [28–31]. These discordant results might reflect differences in baseline HIV disease status, and/or differences in the virological response, and/or treatment switches.

We therefore explored the impact of different first-line ARV regimens on soluble markers of inflammation and immune activation during the first two years of effective therapy, while controlling for potential confounders such as HIV replication, and changes in the initial regimen.

This observational study of changes in markers of inflammation and immune activation during the first two years of unchanged virologically effective first-line cART regimens (tenofovir- or abacavir-based NRTI backbone plus efavirenz, atazanavir/r or lopinavir/r) shows that the choice of the third antiretroviral agent influenced the degree of decline in markers of T-lymphocyte and macrophage activation. Plasma levels of IL-6, IP-10 and MIG fell by at least 40% in all the treatment groups, while the decline in hs-CRP levels failed to reach statistical significance and sCD14 levels were unchanged. The only observed difference between the treatment groups was that the T-lymphocyte and macrophage activation markers IP-10 and MIG fell less markedly in patients receiving ATV/r than in patients receiving EFV; no difference was found between LPV/r and EFV. Our patient population represents the naïve patients who initiated cART between January 2006 and December 2009 with a moderate immune deficit and a median viral load slightly below 100000 copies/ml and who experienced a rapid virological response in usual care in France.

Several aspects of study design are important in studying the influence of different antiretroviral drugs on biomarker variations in patients starting first-line therapy. First, it is important to control for viral replication, which is a major cause of persistent immunological activation and inflammation. As a result, the approach used in randomized trials, which includes all patients independently of their virological response, may not be appropriate. Furthermore, removal of patients with uncontrolled viral load from a randomized trial will result in a simple observational study. Despite the lack of randomization in our study, baseline characteristics were well balanced across the treatment groups, and factors that might have influenced the choice of cART regimen or the time course of the markers of interest were systematically included in the analyses. In addition, all the patients under care in the Infectious Diseases department of Pitié-Salpêtrière Hospital were screened for eligibility criteria, thus avoiding any major selection bias. However, one cannot exclude that unrecognized confounding regarding treatment selection could be present. In our study, the strict inclusion criteria regarding the viral replication resulted in a limited sample size. However, even with this sample size, significant differences were detected for some of the comparisons. In addition, most p-values of non-significant tests were above 0.15 indicating that for most comparisons, power was not an issue. It is possible that the freezing of plasma sample could have modified the marker levels and diminished our ability to detect differences. However, all the groups of antiretroviral regimens were studied similarly and this should not have biased the comparison between groups. Moreover, as in most published studies, studying thawed plasma allowed to study all marker dosages simultaneously in the same experiment, limiting the inter-individual variability. The high variability of inflammatory markers between individuals could have precluded seeing differences between treatment groups.

Elevated IL-6 and CRP levels before antiretroviral treatment initiation or after treatment interruption have been linked to a higher risk of AIDS-defining events and death [23, 34]. In our study, IL-6 levels fell during the first two years of effective antiretroviral therapy, and the decline did not differ significantly between the NRTI backbones or across the different third agents. These results are consistent with those of the HEAT and ACTG A5224 trials [28, 29]. With regard to hs-CRP, we found that its levels fell slightly although not significantly, with marked inter-individual differences and no differential effect of the studied antiretroviral drugs. In the ACTG A5224, HEAT and NICE trials, CRP levels either rose or fell after cART initiation, while a differential effect of antiretroviral drugs was found in ACTG A5224 trial but not in the HEAT or the NICE trials [28–30]. The high inter-individual variability, differences in study designs, such as discussed above, and in baseline characteristics may explain the discordant results.

Elevated plasma sCD14 levels are an independent predictor of death among HIV-infected patients [35]. SCD14, a marker of monocyte activation that correlates with HIV viremia, and is also considered to be a marker of microbial translocation across the intestinal mucosa, correlating positively with plasma lipopolysaccharide levels as being its soluble receptor [36, 37]. Despite 2 years of effective antiretroviral therapy, sCD14 levels did not change significantly, in keeping with previous reports, suggesting that regimens used in our study may not restore the intestinal barrier function, resulting in persistent microbial translocation and immune activation [36, 38]. Previous study has reported that among treated HIV-infected patients, persistent HIV-DNA in the gut correlates with levels of microbial translocation and immune activation [39]. Interestingly, Taiwo B et al. have recently reported that an NRTI-sparing regimen consisting of boosted darunavir plus raltegravir led to a decline in sCD14, IL-6 and IP-10 levels [40], suggesting that integrase inhibitors may be more effective in this respect than other antiretroviral drugs.

Persistent immune activation despite virologically effective therapy has been linked to immunological failure [41, 42]. While IL-6, hs-CRP, sCD14 and IP-10 have been studied in other studies, MIG has not been evaluated in the context of naïve patients initiating cART. IP-10 (CXCL-10) and MIG (CXCL-9), two chemokines induced by interferon gamma, specifically target lymphocytes, particularly activated T cells, as well as macrophages, and are critical mediators of T cell migration during T-cell-dependent immune responses [43]. High levels of these chemokines reflect immunological activation in HIV-infected patients [44]. A recent study has shown that elevated plasma IP-10 levels in primary HIV-1 infection are strongly predictive of rapid HIV disease progression [45]. In our study, IP-10 and MIG levels showed the largest decline among the studied biomarkers during the first two years of effective cART, running parallel to the fall in plasma viremia. IP-10 and MIG were also the only markers for which a differential effect of the studied antiretroviral drugs was observed. The fall in both markers was larger with EFV than with ATV/r. To investigate whether these differences could be explained by the change in CD4 cell count, analyses were made adjusting for the change of CD4 cell count and differences were still significant and could not be explained by the change in CD4 cell count. We hypothesize that other ART-mediated mechanisms such as the more rapid decay in HIV-RNA in the first 14 days of treatment initiation associated with EFV than with ATV/r observed in a randomized study [46] could explain, at least partly, the observed difference between EFV and ATV/r. However, it is unclear whether this has any clinical implications. To our knowledge, the differential impact of antiretroviral drugs on IP-10 and MIG has not previously been studied in naïve HIV-infected patients.

The differential effect of the studied antiretroviral drugs on changes of two immune activation markers such as IP-10 and MIG levels suggests that these markers could be worthwhile when evaluating new antiretroviral drugs. The presence of sCD14 despite two years of viral suppression reflects the persistence of immune activation, and may contribute to the maintenance of residual viremia and HIV DNA reservoirs [47]. With the growing interest for finding a cure for HIV infection and with the preoccupation of long-term management of HIV-infected individuals, it is key to assess the capacity of new drugs and new combinations of drugs to drive immune activation and inflammation down.