Background
Tuberculosis (TB) continues to be one of the deadliest infectious diseases. In 2015 WHO reported an increase in the world prevalence and mortality rate of active tuberculosis (TB) disease cases [
1], particularly in the developing world. Latent TB infection (LTBI), defined as the absence of clinical symptomatology in the presence of infection, affects an estimated one-third to one-fourth of the human population [
2], and as a result of reactivation disease represents the primary source of active TB.
LTBI is traditionally identified as a positive reaction (>5–10 mm induration) in response to the intradermal injection of a protein purified extract of
Mycobacterium tuberculosis (
Mtb), named tuberculin | [
3]. Although sensitive, the specificity of the intradermal reaction is compromised by previous vaccination with
M. bovis BCG and/or infection with some non-tuberculous mycobacteria (NTM) [
4]. More recently, the Interferon Gamma Release Assays (IGRAs) have contributed to a higher specificity in detection of Mtb infection since these assays utilize proteins encoded in the
Mtb, but not in the BCG genome, such as ESAT-6 and CFP10 [
5‐
8]. Nevertheless, both assays have demonstrated a low predictive value for progression to the active forms of TB [
9,
10].
During the last few years, in vivo and in vitro evidence has indicated that
Mtb adapts its transcriptional signature to the microenvironmental conditions posed by the host cells, such as macrophages and dendritic cells, and in granulomas of the infected host [
11‐
14]. In addition, these transcriptional changes seem to be necessary for infection establishment [
11,
12]. In conditions such as hypoxia, acidic pH, nutrient starvation, and high concentrations of oxygen (ROIs) and nitrogen (RNIs) reactive intermediates and CO
2 that may all be present inside granulomas [
15,
16],
Mtb requires a group of approximately 50 genes known as the
dosR regulon [
11,
13,
17] to survive and enter dormancy. Indeed, the expression of
dosR regulon encoded genes has been associated with the non-replicative persistence of
Mtb [
11,
12], showing that some members of this regulon are playing an important role in the maintenance of the latency condition. On the other hand, a family of five genes able to induce resuscitation of dormant bacilli, termed resuscitation-promoting factors (
rpf), has also been identified in the
Mtb genome [
18]. It has been described that the proteins encoded by the
rpf genes (
rpfA-E) are capable of stimulating the mycobacterial growth of non-replicating cells obtained in vitro [
18,
19] and play a significant role in the in vivo persistence and reactivation of chronic infection in mice [
20,
21]. Additionally, ex vivo studies, have demonstrated that Rpf proteins increased the recovery of
Mtb from sputum of the patient with active TB [
22] and improved the sensitivity of culture-based Mtb test in samples that require long culture times [
23].
Given the interest in proteins encoded by the
dosR regulon and the
rpf genes, the cellular immune response to some of these members has been studied in different human populations of Africa, Asia, Europe, and America, demonstrating significantly increased responses of LTBI individuals compared to active TB [
24‐
36]. Our studies in a TB endemic community in the city of Medellín, Colombia, have also provided evidence that the
dosR encoded Rv1737c (NarK, nitrate reductase), Rv2029c (PFKB, phosphofructokinase B), the hypothetical protein Rv2628, and the resuscitation-promoting factors (Rpf), Rv0867c (RpfA) and Rv2389c (RpfD), induced higher production of IFNγ and a higher frequency of T-cells with a CD45RO
+CD27
+ (Tcm) phenotype in 7-day stimulation assays of peripheral blood mononuclear cells (PBMCs) of LTBI compared to active TB [
29,
30,
33]. Interestingly, this higher response of PBMC from LTBI compared to active TB to Rv1737c, Rv2029c and Rv2628 has also been observed in Africa, Asia, Europe, India and Brazil [
24‐
27,
37], suggesting an immune response independent of the human genetic and environmental background, and possibly of the circulating
Mtb strains, and thus suggesting the presence of a prevalent immune response to DosR antigens in LTBI.
Given the problems of sensitivity and specificity associated with the immune response to ESAT-6 and CFP10 Mtb antigens, the search for immune response biomarkers that more efficiently classify LTBI from active TB is a top priority for the prevalence and incidence of active TB to be reduced. In this study, using variables of the immune response, Receiver Operating Characteristics (ROC curves), CHi-squared Automatic Interaction Detection (CHAID) and logistic regression (LR), we found that the Rv2029c antigen of Mtb is a novel classifier of LTBI vs. active TB with high specificity and sensitivity.
Discussion
The immune response to PPD and IGRA antigens has a low power to discriminate between LTBI and PTB [
44,
45]. Indeed, the predictive value of IGRAs for progression to TB disease is low and slightly but not significantly higher than that of the TST [
9,
45]. Therefore, new biomarkers are urgently needed to facilitate the diagnosis of LTBI [
3].
Herein we have evaluated the potential of the
Mtb antigens E6-C10, the DosR Rv1737c, Rv2029c, and Rv2628, and the Rpf RpfA and RpfD to discriminate latent TB infection from active TB. The DosR antigens evaluated in the present study induced a higher production of IFNγ in stimulated PBMCs from LTBI compared to PTB. Also, the production of IFNγ to DosR antigens showed a high probability to discriminate disease status (AUC >0.70). Different studies, including ours, indicate that LTBI preferentially recognizes DosR antigens compared to PTB in different human populations [
24‐
26,
29,
30,
32,
33,
35,
37]. In this study, Rv2029c (pfkB) was included in the model that better-predicted disease status, with a correct classification of 78.3% of LTBI and 88.9% of PTB, according to the analysis based on decision trees (CHAID). Additionally, the logistic regression analysis using the forward method also showed Rv2029c as the antigen that better discriminates between those two conditions (data not shown).
Recent studies have highlighted the immunological importance to Rv2029c. Vaccination of mice with DosR antigens, including Rv2029c induced strong humoral and/or cellular Th1-type (interleukin-2 and gamma interferon) immune responses [
47]. In humans, a stronger response to Rv2029c in LTBI compared to PTB has been reported in the Netherlands [
24], Africa [
25,
35], Japan [
26], China [
36], Brazil [
27], and Colombia [
29,
30,
33]. Remarkably, in a one-year longitudinal study in Brazilian subjects recently exposed to TB, classified by IGRA and TST positivity, PBMCs stimulation with latency antigens, including Rv2029c, it was found that combining the IFNγ responders to Rv2029c, Rv2031c plus Rv2034 detected 90.3% of IGRA-RD1(+) and 66.7% of TST(+) contacts, while 95% were identified by classifying them as TST(+) IGRA-RD(+) and 11% as TST(−) IGRA(−). Moreover, in the follow-up, the TST converters also demonstrated an IFNγ conversion to Rv2029c and Rv2031c, whereas the only TB incident case was detected via IGRA-Rv2029c and TST previous to developing TB [
25]. In another study, the LTBI diagnostic performance of Rv2029c was higher than Rv2628 and Rv1813c by ROC evaluation [
36]. Furthermore, in a study in of the in vitro immune response to the DosR antigens Rv1733c, Rv2029c, Rv2628 before and after 2-week anti-tuberculosis treatment in Ghanaians PTB, it was found that the second week of effective chemotherapy was characterized by a general increase in cytokine response to
Mtb-specific antigens suggesting improvement in cellular response to therapy [
48]. Thus, our studies strengthen the observation that Rv2029c may constitute a relevant biomarker of LTBI. Besides, our findings that Rv2029 was included in the model of logistic regression analysis that best predicted latent and active TB, along with E6-C10 and Rpf antigens, may suggest that Rv2029c is a useful diagnostic candidate who might increase the capacity to discriminate between LTBI and PTB in combination with antigens currently used as such as ESAT-6 and CFP-10, even though much larger studies need to be performed to validate the present results.
Similar to the DosR antigens, the Rpf antigens (RpfA and RpfD) induced a higher production of IFNγ in stimulated PBMCs from LTBI compared to PTB and showed a high probability to discriminate disease status (AUC >0.70). Additionally, these antigens were included in the model that predicted the disease status. The results obtained in this study are consistent with those reported by Chegou and colleagues [
49], who found that Rv0867c (RpfA) and Rv2389c (RpfD) were included in antigen combinations discriminating between HHC and PTB. It has been reported that
rpf genes are differentially expressed at different growth stages, and stress conditions, with
rpfA and
rpfD, mainly expressed during early resuscitation [
50]. Studies conducted in different human population, including ours, indicate that an immune response to RpfA and RpfD antigens has been preferentially found in LTBI suggesting that the immune response to Rpf antigens may play a protective role against bacilli reactivation [
29,
34,
35]. It has been suggested that the bacilli may indeed still be replicating but are controlled by the host immune response during LTBI infection [
51,
52]. So, the host may be exposed to antigens from the different metabolic states of
Mtb in vivo, and the response to these antigens may be detected in in-vitro stimulation assays. Thus, our results suggest that RpfA and RpD antigens also could be added to current diagnostic tests to improve the capacity to discriminate between LTBI and PTB.
On the other hand, and somewhat unexpectedly, our results show that the fusion protein E6-C10 induced a higher production of IFNγ in stimulated PBMCs from LTBI compared to PTB and that the production of IFNγ to E6-C10 antigens showed a high probability to differentiate between LTBI and PTB (AUC >0.70). Although E6-C10 was not selected as the best predictor of the disease status, according to the analysis based on decision trees (CHAID) and the logistic regression analysis using the forward method, E6-C10 was included in the model that better discriminates between LTBI and PTB. This result is consistent with those reported by Chegou and colleagues [
49], who found that E6-C10 was included in the most of the antigen combinations discriminating between presence and absence of TB disease. ESAT-6 and CFP-10 antigens are encoded by genes located within the region of difference 1 (RD1) of the
Mtb genome, a chromosomal segment absent in the BCG vaccine strains and most of the NTM [
7,
8]. So, the utilization of these antigens in the commercially available IGRAs has contributed to a higher specificity for infection detection in comparison to the tuberculin skin test [
44,
45]. The RD-1 antigens are described to be secreted during
Mtb active replication [
6,
53]. Additionally, it has been described that the expression of genes encoding early stage proteins such as ESAT-6 is repressed during the stationary phase of
Mtb growth in the lungs of chronically infected mice [
54], suggesting that they might not be expressed optimally during later stages of
Mtb infection and likely play a much less dominant role during LTBI than proteins from the DosR regulón or Rpf antigens. Collectively these results suggest that the use the DosR antigen Rv2029c and RpfA and Rpf in T cell assays (IGRAs), in addition to E6-C10, could enhance the ability to differentiate LTBI from TB disease especially in a high-burden setting where a mixture of recent and old infections is commonly found [
38].
Our study presents some limitations such as the small sample size which therefore should be extended and validated in a more significant number of participants and different human populations; the use of a 7-day in vitro culture assay rather than a more user friendly assay; the absence of a healthy control group; and the lack of longitudinal evaluation of progression to disease. Although a recent infection is more detectable in a short-term stimulation, it has been argued that long-term stimulation is more sensitive to the detection of LTBI than those with short-term stimulation times particularly in regions of high endemicity in which a mixture of recent and old infections are frequently found [
32,
55,
56]. Thus, long-term stimulation may be better to measure central memory T-cell responses [
55‐
57]. By using long-term stimulation conditions, our group has previously reported the enhanced ability to detect Tcm cells (CD45RO
+CD27
+) in response to mycobacterial antigens [
29,
30,
58,
59]. For that reason, we used long-term cultures to define the LTBI status and compared the immune response between the study groups. As previously pointed out, future works should include a prospective evaluation the antigens and host markers identified in this study, using larger sample size, different geographic settings and using preferably short-term assays to detect effector cell responses. Future studies should also evaluate the antigens and host markers in different study populations as such as children, individuals with immune suppression (e.g., due to HIV coinfection, therapy with TNF-α inhibitors, or due to type 2 diabetes), in TB patients after anti-TB therapy, in patients with extrapulmonary TB, and also in individuals with other lung diseases [
49].