Kinetics of Mycobacterium tuberculosis-specific IFN-γ responses and sputum bacillary clearance in HIV-infected adults during treatment of pulmonary tuberculosis

Summary In HIV-uninfected adults with pulmonary tuberculosis (TB), anti-TB treatment is associated with changes in Mycobacterium tuberculosis (Mtb)-specific immune responses, which correlate with sputum bacillary load. It is unclear if this occurs in HIV-infected TB patients. We investigated changes in Mtb-specific immune responses and sputum bacillary clearance during anti-TB treatment in HIV-infected and HIV-uninfected adults with pulmonary TB. Sputum bacillary load was assessed by smear microscopy and culture. Mtb-specific IFN-γ secreting peripheral blood mononuclear cells were enumerated using an ELISPOT assay following stimulation with PPD, ESAT-6 and CFP-10. The baseline frequency of Mtb-specific IFN-γ secreting cells was lower in HIV-infected than HIV-uninfected patients (median PPD 32 vs. 104 Spot Forming Units (SFU), p = 0.05; CFP-10 19 vs. 74 SFU, p = 0.01). ESAT-6-specific IFN-γ secreting cells and sputum bacillary load declined progressively during treatment in both HIV-infected and HIV-uninfected patients. HIV infection did not influence the 2-month sputum culture conversion rate (Odds Ratio 0.89, p = 0.95). These findings suggest that changes in ESAT-6-specific immune responses during anti-TB treatment correspond with changes in sputum bacillary load irrespective of host HIV infection status. The utility of Mtb-specific IFN-γ responses as a proxy measure of treatment response in HIV-infected TB patients warrants further evaluation in other settings.


Introduction
Tuberculosis (TB) is a leading cause of morbidity and mortality worldwide. Cell-mediated immune responses are pivotal in the host's response to Mycobacterium tuberculosis (Mtb) infection and IFN-g produced predominantly by CD4 þ T lymphocytes is a crucial component of this response [1e4]. Production of IFN-g in response to Mtb-specific antigens is commonly used as a marker of potentially protective immunity against Mtb [5e7]. The 20-to 40-fold increased risk of developing active TB among HIV-infected individuals, particularly those with low blood CD4 þ T lymphocyte counts [8] underscores the importance of T cell-mediated adaptive immunity in protection against disease caused by Mtb.
Successful treatment of pulmonary TB is associated with reduction in sputum bacillary load and assessment of sputum by smear microscopy and culture after two months of anti-TB treatment to monitor early microbiological response is recommended by the International Union Against Tuberculosis and Lung Disease (IUATLD) [9]. Both microbiological assessments, however, have limitations. First, the sensitivity of sputum smear microscopy for acid-fast bacilli (AFB) is low, particularly when the bacillary load is reduced by anti-TB treatment [10]. Second, Mtb grows slowly in culture and may take up to 8 weeks or longer before a positive culture becomes detectable [10]. Other laboratory markers such as C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) have been used to monitor response to anti-TB treatment [11,12], but they are not specific to TB. Therefore, there is need to identify other laboratory parameters specific to TB which could be useful rapid surrogate markers of response during treatment of pulmonary TB irrespective of the patient's HIV infection status.
In HIV-uninfected adults with pulmonary TB, a good response to anti-TB treatment is associated with changes in peripheral blood immune parameters, including T-cell subpopulations, cellular proliferative and cytokine responses [13e16]. Data from previous studies suggest a correlation between disease activity, bacterial load and IFN-g production by sensitized lymphocytes in response to Mtb-specific antigens Early-Secreted Antigenic Target 6 (ESAT-6) and Culture Filtrate Protein 10 (CFP-10) during treatment of pulmonary TB [17e22]. How Mtb-specific immune responses relate to microbiological responses during treatment of pulmonary TB in HIV-infected adults is, however, incompletely understood.
We addressed this knowledge gap by conducting a prospective observational cohort study of adults receiving treatment for confirmed pulmonary TB in Malawi where over 50% of adults with sputum smear-positive pulmonary TB are HIV-infected [23]. The aim of the study was to assess the relationship between host Mtb-specific immune responses and sputum bacillary load during the first 2 months of anti-TB treatment in HIV-infected pulmonary TB patients.

Study population
HIV-infected and HIV-uninfected adults aged !17 years were recruited at the Queen Elizabeth Central Hospital (QECH) in Blantyre, Malawi. Participants were patients with microbiologically confirmed pulmonary TB whose ZiehleNeelsen (ZN)-stained sputum smears were graded !1þ positive for AFB at direct microscopy [10]. Asymptomatic volunteers with no clinical evidence of active disease or previous history of TB treatment were also recruited from communities surrounding QECH as controls. All participants were BCG-vaccinated at birth. Patients were recruited before commencing anti-TB treatment and were followed up to 56 days of treatment, while controls were seen once at recruitment. Anti-TB treatment was given as short course chemotherapy consisting of rifampicin, isoniazid, pyrazinamide and ethambutol for 2 months (intensive phase), followed by rifampicin and isoniazid for 4 months (continuation phase) according to national guidelines. Written informed consent was obtained from all study participants and the research ethics committees of the Malawi College of Medicine (COMREC) and the Liverpool School of Tropical Medicine approved the study.

Sample collection
Peripheral blood and sputum samples were collected from patients before and after 14, 28 and 56 days of anti-TB treatment. Peripheral blood was collected from controls at recruitment only.

Processing of sputum for mycobacterial culture and detection of Mtb
Sputum samples digested and decontaminated by the N-acetyl L-cysteine (NALC) (SigmaeAldrich, Germany) and 3% sodium hydroxide (NaOH) (VWR, Belgium) method were processed for mycobacterial culture using the Bactec™ MGIT™ 960 system (Becton Dickinson, USA) as previously described [24]. Days to positivity (DTP), defined as the time it took for MGIT cultures to become positive were used as an inverse measure of the bacillary load in the sputum sample. Smears were prepared from positive MGIT cultures, stained with ZN stain and examined for AFB by light microscopy. The presence of Mtb in all cultures that were positive for AFBs was confirmed using the MPT64 antigen test (Becton Dickinson, USA) according to the manufacturer's instructions. MGIT cultures were positive for Mtb if both smear microscopy for AFB and MPT64 antigen test results were positive. They were reported negative if there was no growth after 42 days incubation.

Antigens and enumeration of Mtb-specific IFN-g secreting cells
Peripheral blood mononuclear cells (PBMCs) were isolated from heparinised blood using the gradient centrifugation technique as previously described [25]. Freshly isolated PBMCs were analysed for IFN-g production using an 18-hour enzyme-linked immunospot (ELISPOT) assay as described elsewhere [26]. Antigens/peptides were added individually to duplicate wells containing 0.25 Â 10 6 PBMCs at 10 mg/ml purified protein derivative (PPD) (Statens Serum Institute, Copenhagen, Denmark), 5 mg/ml CFP-10, 5 mg/ml ESAT-6 (Peptide and Protein Research, UK) and 5 mg/ml phytohaemagglutinin (PHA) (SigmaeAldrich, Germany) as the positive control. Additional duplicate wells left unstimulated were the negative control. Spot Forming Units (SFU) were quantified using an automated ELISPOT reader (AID Autoimmune Diagnostic GmbH, Strassberg, Germany), and data were expressed as SFU per million PBMCs. Control values (from unstimulated wells) were subtracted from antigenestimulated conditions, and responses were scored as positive if antigen-stimulated wells contained ! 5 SFU more than unstimulated wells as described previously [27].

Statistical analysis
Data analysis and graphical presentations were performed using GraphPad Prism 5 (GraphPad Software, USA). Non-paired comparisons were done either by ManneWhitney U or Kruskall Wallis and Dunn's multiple comparison tests. Logistic regression was used to determine the kinetics of bacillary clearance, while One-way ANOVA was used to determine the kinetics of Mtb-specific immune responses over time. Linear regression was used to determine the association between CD4 count and Mtb-specific immune responses. Results are given as medians with inter-quartile ranges (IQR). Differences were considered statistically significant when p < 0.05.

Participant characteristics
We recruited 63 sputum smear-and culture-positive pulmonary TB patients and 27 asymptomatic controls (Table 1). Among HIV-infected patients 44% (12/27) had a CD4 þ T-cell count of 200 cells/ml at recruitment. All participants on antiretroviral therapy (ART) received the first-line regimen for Malawi at the time of the study consisting of stavudine, lamivudine and nevirapine. Nine ART-naïve patients commenced ART within 2 weeks of starting anti-TB treatment according to national guidelines. All patients had good clinical response to anti-TB treatment during the study period. Chest radiographs were not repeated to monitor response to treatment during follow-up unless clinically indicated by worsening respiratory symptoms and signs. ESR, CRP and plasma HIV viral load were not monitored during the study.

HIV infection is not associated with failure to clear Mtb from sputum during the intensive phase of anti-TB treatment
To assess the effect of HIV on microbiological responses, we determined the sputum bacillary load before, and at days 14, 28 and 56 of anti-TB treatment.  Figure 1A) but was significant for smear conversion (30% vs. 12%, p ¼ 0.0029, Figure 1B). However, the proportion of patients who successfully cleared Mtb after 2 months of treatment was similar between the two groups (Day 56 smear conversion, HIVÀ 85% vs. HIVþ 86%, Figure 1B; culture conversion, HIVÀ 68% vs. HIVþ 50%, Figure 1C; OR 0.89 [IQR 0.02e32.1], p ¼ 0.95). While these findings suggest that clearance of Mtb from sputum very early during anti-TB treatment may be slower in HIV-infected than HIV-uninfected individuals, HIV infection is not associated with failure to clear bacilli as duration of anti-TB treatment increases.   3.3. HIV-infected TB patients have low frequency of PPD-and CFP-10 specific IFN-g secreting cells before commencing anti-TB treatment Next, we compared the Mtb antigen-specific IFN-g responses between HIV-infected and HIV-uninfected TB patients before the start of anti-TB treatment. The frequency of IFN-g secreting cells following ex vivo stimulation of PBMCs with PPD, ESAT-6 and CFP-10 was measured using an ELISPOT assay. We found that the frequency of PPD-and CFP-10-specific IFN-g secreting cells was lower in HIV-infected than HIV-uninfected patients ( Figure 2A and 2B). There was no significant difference in the frequency of ESAT-6-specific IFN-g secreting cells between the two groups (median ESAT-6 72 [IQR 14e330] vs. 108 [IQR 18e256], p ¼ 0.8923). The number of antigen-specific IFN-g secreting cells did not correlate with the blood CD4 þ T-cell count at recruitment (PPD, r 2 ¼ 0.004, p ¼ 0.77; CFP-10, r 2 ¼ 0.04, p ¼ 0.32) suggesting that the blunted IFN-g responses may reflect a preferential loss or dysfunction of PPD-and CFP-10-specific CD4 þ T cells in HIV-infected individuals as reported previously [28,29].
To determine if the antigen-specific responses observed in TB patients were due to active TB, we compared ELISPOT responses between TB patients and asymptomatic controls. The frequency of ESAT-6-and CFP-10-specific IFN-g secreting cells was higher in HIV-uninfected TB patients than HIV-uninfected controls ( Figure 2B and 2C). The frequency of PPD-specific IFN-g secreting cells was not significantly different between HIV-uninfected patients and controls (median PPD 104 [IQR 38e190] vs. 40 [IQR 15e193], p ¼ 0.4255). ESAT-6-specific responses were also higher in HIVinfected TB patients compared to HIV-infected controls (median ESAT-6 72 [IQR 14e330] vs. 10 [IQR 2e26] SFU, p ¼ 0.001; Figure 2C). The frequencies of PPD-and CFP-10-specific responses were not significantly different between HIV-infected patients and controls (median PPD 32 [ ¼ 0.1155). These findings suggest that high ELISPOT responses to Mtb-specific antigens ESAT-6 and CFP-10 in TB patients reflect high bacillary loads during active disease. They also suggest that HIV infection may differentially impact host Mtb-specific responses, with CFP-10-specific responses affected more than ESAT-6-specific responses. The absence of a significant difference in PPD-specific responses between patients and controls may be due to previous BCG vaccination and comparable exposure to environmental mycobacteria.

Kinetics of PPD-, ESAT-6-and CFP-10-specific IFN-g secreting cells during the intensive phase of anti-TB treatment
We examined changes in Mtb antigen-specific IFN-g responses during the first 2 months of anti-TB treatment by comparing the frequencies of IFN-g secreting cells before (day 0) and after 14, 28 and 56 days of treatment in HIV-infected and HIV-uninfected TB patients. Although there were no statistically significant differences between day 0 and day 56 in responses to PPD (HIVÀ, p ¼ 0.6888; HIVþ, p ¼ 0.0846), ESAT-6 (HIVÀ, p ¼ 0.1220; HIVþ, p ¼ 0.9923) or CFP-10 (HIVÀ, p ¼ 0.4508; HIVþ, p ¼ 0.2417), the frequency of PPD-specific IFN-g secreting cells showed an increasing trend from day 0 to day 56 in both HIV-infected and HIV-uninfected patients (Table 2, Figure 3A and 3B). In contrast, the frequency of ESAT-6-specific IFN-g secreting cells showed a decreasing trend from day 0 to day 56 in both HIV-infected and HIV-uninfected patients (Table 2, Figure 3C and 3D). There was no clear trend in the frequency of CFP-10-specific IFN-g secreting cells in both HIV-infected and HIV-uninfected patients (Table 2, Figure 3E and 3F). These findings suggest similar kinetics of PPD-, ESAT-6-and CFP-10-specific IFN-g secreting cells during anti-TB treatment between HIV-infected and HIV-uninfected TB patients although the trends in responses differ between antigens. Differences in antigen loads during treatment may underlie the different trends in antigenspecific responses.

Discussion
The present study investigated the kinetics of Mtb-specific immunity and sputum microbiological responses during the intensive phase of anti-TB treatment in adults with pulmonary TB. We have shown that while HIV infection may be associated with slow sputum bacillary clearance very early in the course of anti-TB treatment, it did not impact the sputum conversion rate after 2 months of treatment. These findings are consistent with, and advance what was reported previously that HIV infection does not influence the time to sputum conversion in sputum smear-positive pulmonary TB patients [30]. Furthermore, HIV-infected patients had lower frequencies of PPD-and CFP-10-specific IFN-g secreting cells before the start of anti-TB treatment than HIV-uninfected patients, perhaps due to loss or dysfunction of Mtb-specific CD4 þ T-cells [28,29]. Together, the findings imply that HIV-associated impaired host Mtb-specific immune responses before commencing anti-TB treatment do not predict subsequent microbiological failure.
The RD1-coded antigens ESAT-6 and CFP-10 are specific to Mtb and were reported to be more specific than PPD for detecting Mtb in HIV-uninfected patients with active TB [31]. Consistent with this report, we found a higher frequency of ESAT-6-and CFP-10-specific IFN-g secreting cells in HIV-uninfected TB patients than HIV-uninfected controls. In contrast, only ESAT-6-specific IFN-g secreting cell responses were significantly different between HIV-infected TB patients and HIV-infected controls, suggesting a differential impact of HIV infection on Mtb antigen-specific immune responses. CFP-10-specific responses appear to be influenced more by HIV infection than ESAT-6-specific responses. This observation underscores the previously reported limited accuracy of RD1-coded Mtb antigen-specific immune responses for diagnosing active TB in HIV-infected individuals [32].
During the first 2 months of anti-TB treatment the frequency of ESAT-6-specific IFN-g secreting cells showed a decreasing trend with increasing duration of treatment in both HIV-infected and HIV-uninfected patients while the opposite was true for PPD-specific responses. Other studies have reported similar trends in the kinetics of cytokine responses to PPD, ESAT-6 and CFP-10 in HIV-infected and HIV-uninfected children [33,34] and HIV-uninfected adults during treatment of pulmonary TB [16]. While the reasons for this dichotomy are not clear, we speculate that the decline in Mtb antigen-specific immune responses may be explained by a decline in antigen load as a consequence of successful anti-TB treatment since these proteins are secreted by live and actively metabolizing bacilli [35]. The differences may also reflect inadequate expansion of a small reservoir of ESAT-6-and CFP-10-specific CD4 þ T cells during anti-TB treatment [36,37]. In contrast, the increase in the frequency of PPD-specific immune responses with increasing duration of anti-TB treatment may reflect an appropriate response to an increasing load of mycobacterial proteins released by dying bacilli early during the intensive phase of treatment. The increase in PPD-specific responses may also be due to rapid expansion of a large reservoir of PPD-specific CD4 þ T cells established following previous BCG vaccination and exposure to environmental mycobacteria [36,37]. Although our study did not look beyond the first two months of treatment, others have shown that the PPD-specific IFN-g response eventually wanes after three to six months of anti-TB treatment [16,34].
This study had limitations. First, the number of patients recruited was small, raising the possibility that small differences in immune responses between HIV-infected and HIV-uninfected patients may have been missed. Second, patients were followed up for the first 2 months of anti-TB treatment only, so it was not possible to link changes in immune and microbiological parameters during the follow up period with outcomes at the end of 6 months of treatment.

Conclusion
This study has shown that HIV infection does not influence the 2-month sputum smear and culture conversion rate in HIV-