Adaptive immune responses to Mycobacterium tuberculosis (M.tb) infection and disease vary widely and are heterogeneous between infected individuals [1]. During the early stages of infection with tuberculosis there is a high T helper (Th) 1, and to some extent Th17, immune response to M.tb antigens [2, 3]. As mycobacteria multiply and M.tb infection progresses to disease, Th1 and Th17 responses tend to be suppressed and Th2 and regulatory responses emerge. Of note, interleukin (IL)-4, is associated with marked immunopathology and increased severity of the disease [4]. Certain M.tb strains, such as some subtypes of the W-Beijing strain, are associated with prominent Th2 type responses and inhibition of Th1 immune responses, and these strains are characterised by greater severity of tuberculosis infection and disease [5, 6]. The effects of IL-4 are accentuated in the presence of the Th1 cytokine tumour necrosis factor alpha (TNFα) [7, 8].

Persons with active tuberculosis also tend to have increased production of IL-10 compared to those with latent infection [9]. M.tb uses the lipoarabinomannan (LAM) molecule in the cell wall to induce immune regulation pathways through transforming growth factor beta (TGFβ) and regulatory T cells that suppress protective immune responses [10–12]. This same molecule, when mannosylated, interacts with mannose receptors to drive dendritic cells to release the regulatory cytokine IL-10 [13]. The IL-10 inhibits antigen processing and presentation during tuberculosis infection [14] and suppresses Th1, Th2 and Th17 function [15–18]. Thus, there is a delicate balance between pro-inflammatory immune responses that recognise M.tb infection, and conceal it by promoting granuloma formation, and regulatory immune responses that, while supressing Th1, Th2 and Th17 function to prevent immunopathology, may support mycobacterial replication and disease progression [18–22].

Extensive work has been carried out on treating persons with latent tuberculosis using Isoniazid Preventive Therapy (IPT). Early studies suggested a lifetime protection against active tuberculosis [23–25], but later studies have shown waning protection following short term isoniazid based regimens [26, 27]. Despite this, benefits of IPT have been demonstrated in high risk groups such as those living with HIV infection and who are latently infected with tuberculosis, where IPT reduces the risk of re-activation and progression to active tuberculosis by over 70 %, and that of mortality by 25 % [28]. Currently, the World Health Organisation (WHO) recommends the use of IPT in HIV positive individuals and children under the age of 5 years who are exposed to active tuberculosis [29].

We derived a hypothesis that household contacts of active tuberculosis patients with latent tuberculosis infection would present with mixed Th1/Th2 cytokine profiles, dominated by high concentrations of Th1 cytokines (due to recent exposure), but accompanied by Th2 and regulatory cytokine production as a result of the on-going immune modulating effects of M.tb infection. Regulatory profiles might be further enhanced in populations highly exposed to environmental mycobacteria, as is the case in Uganda [30].

We further hypothesised that treatment of these latently infected people with isoniazid would halt or reverse the induction of the Th2 and regulatory immune responses, causing a shift towards Th1 type immune responses. Previous studies in active tuberculosis have shown that treatment reverses the immune equilibrium from Th2 responses back to Th1 immune dominance [25, 31]. A similar effect might contribute to the long term protective effects of IPT.

We therefore initiated an exploratory trial nested in a parent cohort to investigate the effect of IPT among household contacts of active tuberculosis patients found to have latent tuberculosis infection, by comparing the immune responses among those who received six months of IPT with those who did not receive IPT.

In this study, we found that the household contacts of active tuberculosis patients who had evidence of latent M.tb infection had strong M.tb-specific Th1 cytokine responses at baseline. Contrary to our hypothesis, we found low production of Th2 cytokines which supports evidence that marked Th2 cytokine responses are markers of TB disease severity rather than of latent infection [4, 31]. In addition, we found very low production of the regulatory cytokine IL-10 and therefore a remote possibility of on-going immune modulation by M.tb in these contacts. In this study we used supernatants from the QFN assay to measure the different cytokines. This assay is designed to measure IFNγ release and may therefore not be optimal for measuring other cytokines including IL-10.

Some studies have also demonstrated that healthy contacts with stable latent tuberculosis infection present with high Th1 and low Th2 cytokine profiles caused by up-regulation of the IL-4 antagonist IL-4δ2 [2, 36]. Immune responses in people with latent tuberculosis vary with the different stages of infection [1, 37] and may also be influenced by prior exposure to environmental bacteria [30], the time since infection with M.tb [38], the infective dose of M.tb [19], and the dynamic status between controlling infection and progressing to disease [39]. Prior exposure with BCG immunisation at birth can also be an attributable factor to the sustained high Th1 responses in this population [40]. Our baseline findings are consistent with recent low-dose exposure to M.tb infection among contacts with an effective, immune controlling response to the M.tb infection [41]. This is supported by the high rate of reversions of the TST even in the no IPT arm, and lack of progression to active TB.

At the end of six months, we found that IPT led to a decrease in the levels of Th1 cytokines and not of the Th2 or regulatory cytokines as anticipated. Reduction in IFNγ cytokine responses following IPT has been demonstrated before [38, 42–45]. One possible explanation is effective clearance of dormant M.tb by isoniazid, leading to a decline in circulating effector cells and leaving only long-lived memory T cells which secrete low Th1 cytokine concentrations on re-exposure to M.tb antigens [46]. Added to this, it is possible that the presence of BCG bacilli from previous immunisation at birth sustains expression of CD4 effector memory cells, and that clearance of BCG bacilli by isoniazid therapy reduces the number of these cells [40]. A third possibility is a reduction in M.tb specific IFNγ and IL-2 producing T-cells [42, 47] due to isoniazid induced apoptosis [48] however, this data has only been generated in one animal study. There have been some concerns about the reduction of M.tb related cytokine responses that may be protective after treatment with isoniazid: this might increase the risk of M.tb re-infection or re-activation, in settings with high exposure to M.tb [26, 27, 49]. Some studies have addressed this by prolonging the duration of IPT [50, 51]. The incidence of acquiring active TB following IPT or no IPT in these trials was similar in the post IPT period, however these trials were in HIV patients who are highly susceptible to acquiring active TB disease. Effects might be different in HIV negative individuals.

We found that IPT caused a relative decline in the antibody concentration to CFP-10 between baseline and follow up, which was similar with the anti-ESAT concentrations though the effect on the latter was statistically weak. There was no effect of IPT on antibody concentrations to PPD. The effect of treatment of active tuberculosis on M.tb-specific antibody responses is variable with some studies showing a decrease [52, 53], while others a sustained increase up to 2 years following chemotherapy [54, 55]. In M.tb infection, CFP-10 and ESAT-6 secretory antigens form a complex which is involved in mycobacterial virulence [56]. The fact that the decline in M.tb-specific antibody responses was only seen with CFP-10 reflects an effect of IPT on M.tb specific responses. This could be explained by antibody involvement in clearing mycobacterial antigens by formation of antigen-antibody complexes [57], though this may be more pronounced for antigens such as LAM [58].

The PPD responses on the other hand may not only be M.tb specific but could be due to exposure to environmental mycobacteria or non-tuberculosis mycobacteria (NTM). Also, even if the IPT led to a reduction in PPD responses that are “M.tb specific”, this effect could be “drowned” by the PPD responses induced as a result of continuous exposure by NTM encountered in the environment. We have previously shown high antibody titres to PPD that were maintained for up to 40 years after BCG immunisation [59]. The majority of our household contacts were immunised with BCG and this previous exposure (as well as prior exposure to environmental mycobacteria) may have contributed to the sustained antibody concentrations to PPD. It is also possible that IPT led to mycobacterial killing and release of antigens resulting in sustained antibody production to PPD while, on the contrary, reducing exposure to the antigens secreted by living organisms. Similar to T cell based responses, antibody responses vary with the type of antigen, stage and bacterial load of M.tb infection [60]. A study with a longer follow up duration would be beneficial to determine any later change in the PPD antibody concentrations.

To assess the effect of IPT on reduction of tuberculosis infection, we analysed the change in QFN and TST tests at the end of follow up. There was only one QFN reversion, and this occurred in the IPT arm, similar to another study conducted by Johnson, et al., in South Africa [43]. In our study, TST reversion was more common in the IPT arm than in the no IPT arm, but this was a weak difference. Johnson, et al., found that it was common to obtain TST reversions following IPT [41], and the average TST diameter for those that reverted changed from 16.3 mm to 1.4 mm after IPT, again similar to our study. There is still contradictory evidence regarding the effect of IPT on TST reactions. The earlier Public Health Service trials reported in North America showed that reversions of the TST reactions were as frequent in the placebo groups as in the isoniazid groups, in keeping with our results [25].

This exploratory study included small numbers of subjects in each IPT treatment arm giving us limited power to draw solid conclusions. Apparent effects may have occurred by chance, and real effects may have been missed. For example, the TST reversion rate appears to be higher in the IPT arm, but the p-valve was not statistically significant. The small sample size resulted mainly from our strict eligibility criteria which required a household contact to be positive on both the QFN and TST tests, yet there were a large number of household contacts that had discordant QFN and TST test results. The impact of IPT in the subgroups positive only on QFN would be of interest. A larger study would be warranted to investigate effects of IPT further. Due to the nature of the study design, we were unable to know the duration of latent tuberculosis infection for each household contact. As discussed above, this is important for the interpretation of our findings [1]. Placing a TST can influence subsequent immune responses to mycobacterial antigen [61] but this is unlikely to have biased our results since samples for QFN were taken before the TST was placed at both time points, and all participants had the same TST exposure. We were unable to satisfactorily monitor for compliance to the IPT. However, the observed differences between the two arms are consistent with an effect of treatment among the participants who were provided with the drug.

These findings provide further evidence that there is an effect of IPT on the Th1 cytokine responses to M.tb-specific antigens, and the M.tb-specific antibody concentration, particularly to CFP-10, but gives no evidence of an effect on the Th2, regulatory and Th17 cytokine profiles, or on M.tb-specific antibody concentrations to PPD or ESAT-6. This could reflect a reduction in the mycobacterial burden, and hence be taken as a positive response to treatment, promoting the desired outcome of disease prevention. Alternatively, these results might perhaps be a direct effect of the drug. The implications of the decline in T cell and antibody responses are not clear, given that correlates of protection against tuberculosis are not well defined [62, 63]. Longer follow up studies have been carried out, mainly in the USA, which showed that the risk of subsequent progress to active TB disease reduced with IPT [24, 25]. We lack information from similar studies carried out in highly TB-endemic settings, where re-exposure is very common. Could a reduction in potentially protective immune responses increase vulnerability upon re-exposure, possibly out-weighing the benefit of preventing reactivation of latent infection? In such settings, there might be value in a larger study investigating the risks and benefits of short term IPT for low risk household contacts of patients with active TB.