Latent Tuberculosis Infection Treatment and T-Cell Responses to Mycobacterium tuberculosis-specific Antigens

The identification and treatment of persons with latent tuberculosis infection (LTBI) who are at high risk of progression to active disease is an important tuberculosis (TB) control and elimination strategy in many low-TB burden countries (1-3). However, apart from observing treated persons for disease progression, there is currently no means to ascertain the success (or otherwise) of LTBI treatment. A simple test that can be used as a marker of cure or to predict an increased risk of progression to active disease after LTBI treatment would be a vital addition to the existing tools for TB control.

The recent availability of the Mycobacterium tuberculosis antigen-specific interferon (IFN)-? release assays (IGRAs) represents a significant advance in the field of TB diagnostics (4-6). These new blood tests are expected to eventually replace the century-old tuberculin skin test (TST) for LTBI testing (7, 8). These assays measure the T-cell IFN-? responses to M. tuberculosis-specific peptides derived from early secretory antigenic target 6-kD protein (ESAT-6) and culture filtrate protein 10 (CFP-10). These antigens are encoded by the region-of-difference-1 genomic region present in all strains of M. tuberculosis but absent in most nontuberculous mycobacteria and all Mycobacterium bovis bacillus Calmette-Guèrin (BCG) vaccine strains. The commercially available IGRAs are the T-SPOT.TB assay (Oxford Immunotec, Abingdon, UK), which enumerates individual, activated, M. tuberculosis-specific T cells using enzyme-linked immunospot (ELISPOT) technology, and the whole blood QuantiFERON-TB Gold assay (Cellistis, Victoria, Australia). The overnight (16-24 h) incubation period of these assays permits the stimulation of effector T cells that have recently encountered the TB antigens in vivo, without stimulation of the longlived memory T cells. It is hypothesized that the frequency of effector T cells falls as the mycobacterial antigen load (reflecting bacterial load) declines with treatment, and that measuring T-cell responses to M. tuberculosis-specific antigens may thus be useful to determine the effectiveness of treatment (9). If this is correct, then these responses could potentially be important clinical and research tools to monitor the effect of treatment for LTBI or TB, and as surrogate markers of cure or predictors of relapse.

We used the T-SPOT.TB assay to test this hypothesis in a cohort of contacts who received LTBI treatment under program conditions at the Singapore TB Control Unit (TBCU).

METHODS

Study Participants

The subjects comprised adult and adolescent close contacts of sputum acid-fast bacilli smear or culture-positive pulmonary or laryngeal TB index cases recruited into a prospective study evaluating the T-SPOT. TB for contact screening under routine program conditions at the Singapore TBCU from January to September 2005. These contacts were exposed to the infectious index patients for at least 8 hours in the household, family, workplace, or school setting. Ethics approval for this study was granted by the National Healthcare Group's domain-specific institutional review board.

Contact Screening and LTBI Treatment Protocol

The contacts were screened with the T-SPOT.TB assay, the TST (two tuberculin units of purified protein derivative of the RT 23 strains [2 TU RT 23 PPD; Statens Serum Institute, Copenhagen, Denmark]), and for symptoms of active disease. They were also asked about comorbidities such as diabetes mellitus, infection with HIV, renal failure, malignancy, and use of immunosuppressive or steroid therapy. Exposure to TB in the remote past and the presence and number of BCG scars were noted. HIV testing was not offered as part of the TBCU's contact screening protocol. Those with history of previous TB or LTBI were screened only for symptoms and with chest radiograph in the presence of such symptoms.

Contacts with positive T-SPOT.TB results were offered LTBI treatment after exclusion of active disease with chest radiograph, symptom screen, and general physical examination. The treatment regimen used was daily isoniazid for 6 months. Contacts exposed to isoniazid-resistant index cases were prescribed the 4-month rifampicin regimen. Those with fibrotic scarring on chest radiograph were prescribed 4 months of rifampicin and isoniazid. Treatment was self-administered. The contacts were reviewed every 4 to 6 weeks at the Contact clinic, with their treatment adherence verbally assessed at each visit by the TBCU health care worker.

T-SPOT.TB Assay

The T-SPOT.TB assay was performed according to the manufacturer's instructions. The test result was considered positive if either or both of panel A (containing antigens derived from ESAT-6) or panel B (containing antigens derived from CFP-10) had six or more spots than the negative control and this number was at least twice the number of spots in the negative control. The spots were read using the ELISPOT plate reader (AID-GmbH, Strassberg, Germany). The results were visually checked and, if deemed necessary, corrected by manual counting. The laboratory technicians were blinded to the subject identifiers.

Statistical Analysis

Data were entered into SPSS (version 13; SPSS, Inc., Chicago, IL) for analysis. The Pearson's ?^sup 2^ test was used to compare the proportions in two groups and the McNemar test was used for comparing paired proportions. The Mann-Whitney U test was used to compare unpaired results and the Wilcoxon signed rank test compared paired observations. The level of statistical significance was p < 0.05.

RESULTS

Nine hundred and nine contacts completed the screening protocol, of whom 404 (44.4%) tested T-SPOT.TB positive. Seventy-two (17.8%) contacts were positive only to ESAT-6, 191 (47.2%) contacts were positive only to CFP-10, and 141 (34.9%) contacts were positive to both. Three hundred and twenty-five contacts with positive T-SPOT.TB results commenced LTBI treatment; 250 (76.9%) of these contacts completed treatment. Repeat T-SPOT.TB tests at treatment completion were performed for 226 contacts; these contacts form the subject pool for this study analysis. The vast majority (216/226, or 96%) received daily isoniazid for 6 months. Seven contacts received daily rifampicin for 4 months, and three received daily rifampicin and isoniazid for 4 months.

The characteristics of the 325 contacts who began LTBI treatment are shown in Table 1. There was no significant difference in the characteristics of those who underwent repeat T-SPOT.TB testing upon treatment completion (n = 226) and those who did not (n = 99). The latter group comprised 75 contacts who did not complete treatment and 24 who completed treatment but did not undergo T-SPOT.TB testing at treatment completion. None of the contacts self-reported to be HIV positive. Diabetes mellitus was the predominant comorbidity and was present in 5% of contacts.

The distributions of the TST readings for the 226 subjects were as follows: 108 (47.8%) had TST readings of 15 mm or more, 109 (48.2%) had TST readings of 10 to 14 mm, six (2.6%) had TST readings of 5 to 9 mm, and three (1.3%) had TST readings of less than 5 mm. One hundred and fifty-one (67%) had at least one BCG scar.

Pre-LTBI Treatment T-SPOT.TB Results

Of the 226 pretreatment tests, 120 (53.1%) were positive to ESAT-6 and 181 (80.1%) positive to CFP-10. Forty-five (19.9%) were positive to ESAT-6 only, 106 (46.9%) to CFP-10 only, and 75 (33.1%) to both (Figure 1). Contacts positive to CFP-10 only were significantly younger than those who were positive to ESAT-6, whether alone or in combination with CFP-10 (median age, 36.0 vs. 47.1 or 47.0 yr, respectively; p < 0.001, Kruskal-Wallis test). The mean number of spot-forming cells (SFCs)/ 2.5 × 10^sup 5^ peripheral blood mononuclear cells (PBMCs) above negative control in the ESAT-6 test panel was 15.35 (range, -2 to 147), whereas that for CFP-10 was 23.03 (range, -2 to 193). The median number of SFCs/2.5 × 10^sup 5^ PBMCs above negative control was 6 in the ESAT-6 panel, and 11 in the CFP-10 panel.

Post-LTBI Treatment T-SPOT.TB Results

Using the manufacturer's criteria, the T-SPOT.TB reverted to negative in 85 (37.6%) contacts at LTBI treatment completion. Of the 141 positive post-treatment tests, 43 (30.5%) were positive to ESAT-6 only, 33 (23.4%) to CFP-10 only, and 65 (46.1%) to both peptides (Figure 1). There was a statistically significant change (as reflected by test reversion) in the T-cell response to CFP-10 with LTBI treatment (p < 0.001, McNemar test) (Table 2), whereas the response to ESAT-6 was unchanged with treatment (p = 0.081, McNemar test) (Table 3). Of the 106 contacts who were positive pretreatment only to CFP-10, 69 (65.1%) reverted this response post-treatment. Of the 45 contacts who were positive pretreatment only to ESAT-6, 13 (28.9%) reverted this response post-treatment. Of the 75 contacts who tested positive pretreatment to both antigens, 9 (12.0%) reverted both responses, 10 (13.3%) reverted only to CFP-10, and 4 (5.3%) reverted only to ESAT-6 post-treatment. CFP-10 reverters were significantly younger than nonreverters (median age, 36.9 vs. 43.9 yr; p = 0.013, Mann-Whitney U test). There was no significant age difference between ESAT-6 reverters and nonreverters (median age, 44.0 vs. 47.8 yr; p = 0.386, Mann-Whitney U test).

Post-treatment, the mean number of SFCs/2.5 × 10^sup 5^ PBMCs above negative control in the ESAT-6 panel was 13.75 (range, -15 to 211), whereas that for the CFP-10 panel was 13.61 (range, -13 to 157). The median number of SFCs/2.5 × 10^sup 5^ PBMCs above negative control was 4.5 for the ESAT-6 and 4.0 for the CFP-10 panels. The difference in the median SFCs pre- and post-treatment was significant for CFP-10 (11 vs. 4; p < 0.001, Wilcoxon signed rank test), but not for ESAT-6 (6 vs. 4.5; p = 0.116, Wilcoxon signed rank test). The median fall in SFCs/2.5 × 10^sup 5^ PBMCs in response to ESAT-6 was zero, and the median fall in response to CFP-10 was 6 SFCs. The difference in the pre- and post-treatment responses to these two antigens was statistically significant (p < 0.001, Mann-Whitney U test). The numbers of SFCs pre- and post-treatment in response to ESAT-6 and CFP-10 are shown in Figure 2.

There was no correlation of TST size with the change in CFP-10 responses with treatment (Figure 3).

Direction of Change of ESAT-6 and CFP-10 Responses with LTBI Treatment

The various permutations of change are shown in Table 4. The proportions of responses that fell, rose, or remained unchanged with treatment were significantly different between ESAT-6 and CFP-10 (p = 0.004, ?^sup 2^ test).

One hundred and ten (48.7%) contacts demonstrated a fall in SFCs in their response to ESAT-6 (median fall, 5 SFCs), compared with 160 (70.8%) contacts in the response to CFP-10 (median fall, 10 SFCs). Eighty-eight (38.9%) contacts demonstrated a rise in frequency of SFCs (median rise, 4 SFCs) in their response to ESAT-6, as compared with 52 (23%) contacts in their response to CFP-10 (median rise, 6 SFCs). Among the 88 contacts with a rise in response to ESAT-6 with treatment, the highest proportion (59.1%, n = 52) had a concomitant fall in response to CFP-10. Taking into account the likely low reproducibility of the exact spot counts, we also recategorized the direction of change of the responses such that an interval change of ±5 SFCs was classified as "unchanged." Using this classification, we found that the majority (52%) of the responses to CFP-10 fell with treatment, whereas the majority (59%) of the responses to ESAT-6 remained unchanged. A statistically significant difference in the directions of change in the responses to the two antigens was still found (p < 0.001).

DISCUSSION

To our knowledge, this is the largest cohort in whom the effect of LTBI treatment on T-cell responses to M. tuberculosis-specific antigens has been reported, and the first using the commercial T-SPOT.TB assay. The assay's test format enabled the effect of LTBI treatment on the individual antigens ESAT-6 and CFP-10 to be evaluated separately. LTBI treatment resulted in a significant fall in the quantitative response to CFP-10 and a significant reversion rate for this repsonse, without a similar effect on the response to ESAT-6. Although an isolated positive response to CFP-10 had accounted for the highest numer (46.9%) of positive T-SPOT.TB test pretreatment, an isolated response in this test panel accounted for the least number (23.4%) of positive T-SPOT.TB test post-treatment. Those who were positive pretreatment to ESAT-6, whether alone or in combination with CFP-10, were significantly older than those who were positive to CFP-10 alone. Contacts who reverted their response to CFP-10 post-treatment were significantly younger than those with nonreversion. The majority (62.3%) of our contacts remained T-SPOT.TB positive by the manufacturer's criteria at the end of LTBI treatment.

A limitation of this study was that the medications were self-administered; hence, treatment adherence could not be assured. We would, however, assume that our contacts who completed treatment were fairly motivated to take their medications because they had returned regularly for follow-up and for their end-of-treatment T-SPOT.TB test. We also did not have a group of untreated contacts in whom the T-cell responses were monitored over time. Reports from other investigators to date, however, support the assumption that positive ELISPOT responses generally remained unchanged in the absence of any treatment intervention (10, 11).

Published longitudinal studies on the influence of TB and LTBI treatment on T-cell response to M. tuberculosis-specific antigens have mostly involved small numbers of subjects. Three early reports using the ELISPOT technology showed declines in T-cell response to ESAT-6 with treatment of active TB disease in 12, 5, and 18 patients (12-14). Nicol and coworkers subsequently reported in 10 children an early increase in ELISPOT responses to ESAT-6 and CFP-10 followed by a decline at 3 and 6 months of TB treatment (15). A study by Aiken and coworkers on 89 patients with TB in the Gambia showed declines in ESAT-6 and CFP-10 ELISPOT counts after successful TB treatment at 12 months after diagnosis (16). Wilkinson and coworkers showed that LTBI treatment with 12 weeks of rifampin and isoniazid was associated with a statistically significant increase in the numbers of IFN-?-producing T cells (expressed as the mean of the summed numbers of SFCs/10^sup 6^ PBMCs in response to ESAT-6, CFP-10, 38-kD, and a-crystallin 1) within 26 ± 4 days of treatment, followed by a decrease by the end of the treatment period (10). This pilot study was, however, insufficiently powered to show a significant difference in the responses before and after treatment in the 16 subjects who were monitored to the end of treatment. Ewer and coworkers reported an increase in the ELISPOT responses to both ESAT-6 and CFP-10 at the 6-month time point in 38 schoolchildren in the United Kingdom who received 3 months of rifampicin and isoniazid LTBI treatment (11). Although there was a subsequent overall decline in the responses to both antigens, the tests remained persistently positive in 84% of the schoolchildren at 18 months. Pai and coworkers also found persistently elevated IFN-? responses, at 4 and 10 months after 6 months of isoniazid LTBI treatment, in 10 health care workers in rural India (17). They used the Quanti-FERON-TB Gold In-tube (QFT) assay in which whole blood is incubated in blood collection tubes precoated with ESAT-6, CFP-10, and TB 7.7 peptides. The authors attributed their finding to ongoing, intensive exposure in a high TB transmission setting.

An unresolved issue in the use and interpretation of the IGRAs in serial testing is the definition of a test conversion or reversion. The optimal threshold to distinguish these events from nonspecific variations is not known. Pai and coworkers compared serial testing using the QFT with the TST (1 TU RT 23 PPD) in Indian health care workers (18), and found a stronger agreement using the QFT conversion definition of IFN-? = 0.70 IU/ml, rather than the manufacturer's recommendation of IFN-? = 0.35 IU/ml, with TST conversion (defined as a 10-mm increment). Pai and colleagues' study also reported QFT reversion in 9 of 38 health care workers. That QFT nonreverters had significantly higher baseline IFN-? levels than reverters was not surprising because the IFN-? response in the former would have had to drop much more for the result to become negative. Interestingly, the QFT reverters were more likely to have baseline TST readings of less than 10 mm compared with the nonreverters, and the authors suggest that these baseline positive QFT results may have been falsely positive. It was not stated whether any of the QFT reverters had received LTBI treatment. In this present study, which focuses on IGRA reversion post-LTBI treatment, repeat TST post-treatment would not be useful for comparison with the IGRA result because the significance of TST reversion is not known. Although it may be argued that the reversion seen in responses to CFP-10 with LTBI treatment may be due to nonspecific variations, we nevertheless found a significant difference between the responses to this antigen and those to ESAT-6 with treatment.

Using the T-SPOT.TB assay, we found a differential effect of LTBI treatment on the separate T-cell response to ESAT-6 and CFP-10 in our cohort of 226 contacts. The significant fall in the response to CFP-10 with LTBI treatment would be consistent with the hypothesis that there is clearing of this antigen as the mycobacterial load declines with treatment. Although the lack of a corresponding decline in the response to ESAT-6 could reflect true persistence of the antigen (or the host immune response to it) despite treatment, factors that could account for this nondecline should be considered. A possibility could be that 6 months of isoniazid may be too short a treatment duration for LTBI. In the American Thoracic Society 2000 statement (1), the recommended length of isoniazid treatment from 6 to 9 months was revised based on evidence from the Alaskan isoniazid preventive treatment trials (19). The 6-month treatment duration is still, however, widely prescribed, perhaps due to difficulties in patient acceptance and ensuring treatment completion with the 9-month regimen. There will no doubt be reports forthcoming on the effect of this duration of LTBI treatment on T-cell responses, which should shed further light on this issue. Interestingly, we also found the same pattern of response (i.e., no change in response to ESAT-6 vs. a fall in response to CFP-10) in our 10 contacts treated with rifampin-containing regimens (data not shown). Whether sufficient time had elapsed for the clearance of the antigen and the immune response to wane after treatment completion is also uncertain. Our contacts who were still positive at treatment completion will have their T-SPOT.TB repeated 6 months later to determine any further fall in T-cell response. Last, we believe that exogeneous reinfection is highly unlikely in Singapore's setting, which has an intermediate TB incidence rate of 39 new cases per 100,000 individuals (20). Nonetheless, if any of the above factors were indeed responsible for the nondecline in response to ESAT-6, one would reasonably expect a similar nondecline in the response to CFP-10, which was not the case in our contacts.

The role of ESAT-6 and CFP-10 proteins in TB pathogenesis has yet to be defined. The genes encoding these antigens are transcribed together (21); these two proteins form a tight 1:1 complex, and evidence suggests that they are as active as the complex (22). The differential effect of LTBI treatment on the T-cell responses to these individual antigens in our contacts might, however, suggest that CFP-10 or the host immune response to it may have a different function from that of ESAT-6. Most of the early research activity had focused on the ESAT-6 antigen; publications pertaining to CFP-10 per se have been relatively scanty. Ferrand and colleagues showed strong ESAT-6 responses in a cross-sectional study on 10 patients who recently completed TB treatment, suggesting the persistence of this antigen after successful treatment (23). Similar findings have been reported by other investigators (24, 25), and it has been postulated that this antigen may have a role in protective immunity. Evidence for the protective role of ESAT-6 against M. tuberculosis infection has also been demonstrated in the mouse TB model (26). We postulate that the T-cell response to ESAT-6 may persist as a "scar" of previously treated or "quiescent" infection, whereas that to CFP-10 may be more indicative of "active" (i.e., recently acquired), albeit "latent" infection because this response appears to be influenced by LTBI treatment. That the majority of our contacts with a rise or no change in SFCs to ESAT-6 post-treatment exhibited a concomitant fall in SFCs to CFP-10 may lend support to this. Our finding that contacts with positive pretreatment responses to ESAT-6 were significantly older than those with isolated CFP-10 responses would also further favor this speculation.

We have assumed that the overlapping peptides used in the commercial assay have equivalent potency to that of whole ESAT-6 and CFP-10. The antigenic equivalence of human T-cell responses to the ESAT-6 and CFP-10 proteins and to mixtures of synthetic peptides has been previously demonstrated by Arend and coworkers (27). Interestingly, these researchers also showed that the responses to these two antigens varied between individuals with different HLA-DR types, with responses to CFP-10 being significantly higher in the presence of HLA-DR15. A study in West African twins has also reported that memory T-cell responses to "short-term culture filtrate" and peptides from the ESAT-6 protein are subject to genetic regulation (28). These findings would obviously have important implications in the performance and use of the IGRAs in different populations.

In conclusion, we show, for the first time, that LTBI treatment resulted in a significant fall in the T-cell response to CFP-10, while having no effect on the response to ESAT-6 as measured using the T-SPOT.TB assay in 226 TB contacts treated under general program conditions. Our findings suggest that the quantitative response to CFP-10 may be a potentially useful treatment-monitoring tool. In contrast, the response to ESAT-6 appeared to persist, at least in the immediate post-treatment period. Longer term follow-up of treated and untreated persons with LTBI with serial IGRA testing should provide more information on the utility of these responses for treatment monitoring, and as surrogate markers of cure or risk of progression to disease.

Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript.

Acknowledgment: The authors thank all the contacts who participated in the study. They thank the staff of the TBCU Contact clinic, and laboratory technicians Chai Lim Ng and Agampodi Pereira. They thank Siew Pang Chan and Shen Liang for rendering statistical expertise and Professor Soo Chuan Poh for critical comments on the manuscript. Oxford Immunotec provided the T-SPOT.TB kits at a reduced price.