Risk factors for increased immune reconstitution in response to Mycobacterium tuberculosis antigens in tuberculosis HIV-infected, antiretroviral-naïve patients

Patients enrolled in this study were TB-HIV-infected patients who participated in a randomized clinical trial (RCT) conducted from April 2006 to August 2012 at the Tuberculosis Clinic of the National Institute of Infectious Diseases Evandro Chagas (INI), Fundação Oswaldo Cruz, Rio de Janeiro, Brazil. The aim of this RCT was to evaluate the efficacy and safety of the concomitant use of anti-TB regimens, including rifampicin and efavirenz-based cART, in two different doses of efavirenz (600 and 800 mg) registered at clinicaltrials.​gov [11].

The cART and anti-TB therapies were prescribed according to the Brazilian Ministry of Health Recommendations at the time of the trial [12, 13], and efavirenz combined with two nucleoside analogs or a nucleoside analog and tenofovir was offered in two different doses: 600 and 800 mg according to the randomization. Patients with low adherence to cART or anti-TB therapy and more than three missing visits were excluded from this analysis.

The cART was initiated during the baseline visit (D0), which occurred 30 ± 10 days after the introduction of anti-TB therapy. Follow-up visits were performed at 30, 60, 90 and 180 days after cART initiation. Demographic and clinical data as well as blood samples were collected at the baseline and follow-up visits.

During the follow-up, patients were assessed for IRIS. IRIS is defined as a documented worsening of signs or symptoms of TB while on an appropriate anti-TB therapy and cART that cannot be explained by any other diseases, resistance to TB drugs, low adherence or by an adverse reaction [14].

All the patients who agreed to participation in the immune response study signed an informed consent form. This protocol was approved by the INI Ethical Board (CAE: 0052.0.009.000–10), which is affiliated with the Brazilian National Ethics Council (CONEP).

CD4⁺ T cell counts and quantification of HIV VL were performed at each visit. BD Multitest monoclonal antibodies specific for CD45⁺, CD3⁺, CD4⁺ and CD8⁺ and conjugated to PerCP, FITC, APC and PE, respectively, were used to determine absolute counts of CD4⁺ and CD8⁺ T cell subsets according to the manufacturer’s instructions (BD Biosciences, Franklin Lakes, NJ, USA). The samples were evaluated using a FACSCalibur (BD, USA) and Multiset software (BD, USA). Plasma samples were obtained by centrifugation within 4 h of blood collection and aliquots were stored at −86 °C until ready for VL determinations that were performed according to the manufacturer’s guidelines (NASBA, Organon Teknika, Boxtel, The Netherlands; branched DNA assay, Versant HIV-1 RNA 3.0, Siemens, Tarrytown, USA). The lowest established VL detection limit (LDL) was 80 copies/ml.

Peripheral blood mononuclear cells (PBMCs) were obtained from heparinized whole blood and were processed immediately after blood collection. PBMCs were isolated by density gradient centrifugation using Histopaque 1077 (Sigma–Aldrich, USA), cryopreserved in 90% fetal bovine serum (FBS-Gibco, Invitrogen, USA) and 10% dimethylsulfoxide (Sigma–Aldrich, USA) and stored in liquid nitrogen to be further subjected to the ELISpot assay.

Briefly, 96-well plates (Millipore, USA) were coated with capture anti-IFN-γ mAb (Diaclone, France) diluted 1/100 in PBS and incubated overnight at 4 °C. After several washes with PBS, the wells were blocked for 2 h at room temperature with complete culture medium. PBMCs were added at an input cell number of 1 × 10⁵ cells/well and were unstimulated (negative control) or stimulated with phytohemagglutinin (positive control), and with different Mtb-related antigens in duplicates: PPD (Statens Serum Institut-Denmark), ESAT-6 (Statens Serum Institut-Denmark) or 38 kDa/CFP-10 (kindly donated by LIONEX Immunodiagnostic GmbH, Braunschweig, Germany). All of the stimuli were used at a concentration of 5 μg/well in complete culture medium [RPMI 1640 (Sigma, USA) supplemented with 10% of FBS, Penicillin–Streptomycin (10,000 U–10,000 μg/mL), L-glutamine 200 mM, non-essential amino acids 10 mM, and sodium pyruvate 100 mM (all from Invitrogen, USA)]. After 40 h of incubation at 37 °C and a 5% CO₂ humid atmosphere, the plates were washed 3 times with PBS, 3 times with PBS supplemented with 0.1% Tween 20 (Sigma-Aldrich) and 3 times with PBS followed by 2 h of incubation at 37 °C with 100 μl per well biotinylated detection mAb and diluted 1/100 in PBS supplemented with 1% BSA (Diaclone) (PBS 1% BSA). The wells were then washed with PBS, and 100 μl of substrate (5-bromo-4-chloro-3-indolyl phosphate/NBT; Diaclone) was added per well. The colorimetric reaction was terminated after 10 min at room temperature by washing the wells several times with water. The spots were counted using an automated ELISpot reader (CTL Analyzers LLC, Cellular Technology, USA). The results are expressed as spot-forming cells (SFC)/million PBMCs. The response was considered positive if ≥50 SFC/10⁶ PBMCs were obtained after subtraction of the mean background obtained with non-stimulated cells [15].

Demographic characteristics and clinical, virological and immunological parameters were assessed for TB-HIV patients at baseline and during follow-up visits. Chi-squared distribution for categorical variables and the Mann-Whitney test for continuous variables were used to compare groups (included vs. excluded patients). Wilcoxon sign-rank, and McNemar tests were used to compare the immune responses, which were expressed as numerical (SFC/10⁶ PBMCs) and categorical (positive or negative) variables within patients at different study time points, respectively. We fitted a log-linear model where each follow-up visit represents one observation, and the patients were considered at the cluster-level with random effects and modeled by a working autoregressive (AR1) correlation structure. Then, the adjustment of the covariance was made, taking into account observations of the same individual and their proximity in time, although the fixed effects had been obtained through the mean value of the visits. The numerical outcome variables were the production of IFN-γ in response to 38 kDa/CFP-10 or PPD, and both were transformed into log scales for analysis. The univariable statistical analysis was made by modeling the variables one by one. The multivariable statistical analysis was made using a backward stepwise method where all of the covariates were included in the initial model but only the covariates that presented statistical significance with alpha <10% were retained in the final model. The analysis was performed using R v3.1.0 software [16].

Eighty-eight TB-HIV patients agreed to participate in the study. Twenty-seven patients were excluded due to low adherence to HIV treatment (n = 9) or for attending less than three follow-up visits (n = 18). There were no statistically significant differences in demographic, clinical and laboratory data at baseline among the included and excluded patients (Table 1). Among the included patients, we observed a predominance of men and a similar proportion of pulmonary and disseminated TB. Moreover, median CD4⁺ T cells count were 141 cells/mm³ and 77% had a VL ≥ 30,000 copies/ml. IRIS incidence was 11% (7 patients).

The ELISpot results for the presence of IFN-γ producing T cells in response to PPD, ESAT-6 and 38 kDa/CFP-10 antigens are shown in Fig. 1(a, b and c, respectively). The proportion of responders to PPD, ESAT-6 and 38 kDa/CFP-10 at baseline was approximately 49%, 14% and 20%, respectively, and increased significantly after 30 days of cART (except for ESAT-6) and then stabilized (38 kDa/CFP-10) or slightly decreased (PPD and ESAT-6) between 90 and 180 days (Fig. 1d, e and f). The proportion of responders to PPD remained higher than that for all of the other antigens during the follow-up.

Multivariate analysis was conducted to identify factors associated with IFN-γ production in response to PPD (Table 2) and to 38 kDa/CFP-10 antigens (Table 3). In patients with CD4⁺ T cell counts <200 cells/mm³ at baseline, the mean of IFN-γ production in response to the PPD antigen was 5.547 (95% CI = [1.348; 22.821], p = 0.018) times that of the patients with CD4⁺ T cell counts ≥200 cells/mm³. Similarly, the IFN-γ production in patients aged 32 years or less was 2.808 (95% CI = [1.13; 6.98]; p = 0.026) times that of patients aged 38–44 years. The other variables independently associated with IFN-γ production in response to PPD were the site of TB, efavirenz dose, follow-up CD4 ≥ 200 cells/mm³ and follow-up CD4 < 200 cells/mm³ (Table 2). Regarding follow-up CD4⁺ T cell counts, it was estimated that for each increase of one unit in CD4⁺ T cell counts, the IFN-γ production in response to PPD increased 1.011 units (95% CI = [1.002; 1.02], p = 0.016) since CD4 counts were <200; however, when follow-up CD4⁺ T cell counts were ≥200 cells/mm³, the IFN-γ production in response to PPD increased 1.006 units (95% CI = [1.003; 1.009], p = 0.016), p < 0.001) for each increase of one unit in CD4⁺ T cell counts (Table 2). Further adjusting for gender did not change our results.

Factors associated with the production of IFN-γ in response to 38 kDa/CFP-10 were detectable HIV VL and CD4⁺ T cell counts at follow-up visits in the range ≥ 200 cells/mm³ with an increase of 1.002 (95% CI = [1; 1.003]; p = 0.013) units of IFN-γ production for each increase of one unit of CD4⁺ T cell counts. The other variables that were independently associated were efavirenz dose and follow-up % CD8⁺/38⁺ T cells (Table 3). Additionally, a trend (significance between 5% and 10%) for the association with IFN-γ production in response to 38 kDa/CFP-10 was observed for extrapulmonary TB cases compared with pulmonary TB.

The IRIS diagnosis occurred at a median of 51 days after the introduction of cART. Patients with IRIS showed an increased immune response compared with patients without IRIS, and had a marked increase in IFN-γ production by T cells in response both to PPD and, notably, to 38 kDa/CFP-10 antigens after 60 days of cART followed by a reduction, which was in contrast to patients without IRIS (Fig. 2a and b).