Background
Tuberculosis (TB) is the leading infectious cause of death world-over. TB is caused by infection with the tubercle bacillus
Mycobacterium tuberculosis (M.
tb) [
1,
2]. An estimated 1/3 of the global human population is thought to be infected with M.
tb. Majority (90–95%) of these infections remain asymptomatic (a.k.a latent M.
tb infection, LTBI) [
3‐
5]. Only 10-5% progress to active TB (ATB) disease. In 2016, there were an estimated 1.3 million TB deaths among HIV-negative people (down from 1.7 million in 2000) and an additional 374 000 deaths among HIV-positive people. About 10.4 million people fell ill with TB in 2016: 90% were adults, 65% were male, 10% were people living with HIV (74% in Africa) and 56% were in five countries: India, Indonesia, China, the Philippines and Pakistan [
3‐
5]. Hopes of totally controlling TB have been dampened because of the (i) difficulty of developing an effective vaccine, (ii) expensive and time-consuming diagnostic process, (iii) necessity of many months of treatment, (iv) increase in HIV-TB co-infections, and (iv) emergence of drug-resistant cases in the 1980s [
3‐
5].
Because LTBI forms the vast reservoir from which ATB accrues, it has been proposed that identifying those incipient forms of LTBI that are at high-risk of progressing to ATB and treating or offering them chemo-prophylaxis, can drastically reduce global TB incidence [
3,
5]. Accurate designation of high-risk or incipient LTBI is,
however, currently impossible. Specifically, the two available methods (the tuberculin skin test-TST and interferon gamma release assays- IGRA) for testing for LTBI cannot designate high-risk LTBI. TST if positive provides evidence of M
.tb infection. That said, many HIV infected patients will have a negative skin test despite M
.tb infection or disease, due to anergy. “Two stage or booster test” is not a substitute to anergy testing. However, the same might have some utility in detecting M
.tb infection in anergic HIV-TB co-infected patients [
6]. TST underestimates the prevalence of LTBI in endemic countries, requires trained health care staff to correctly perform and accurately read the results, and also demands a second patient visit [
7‐
13]. The test is neither useful to rule in disease nor in high TB prevalence settings to identify eligible individuals for prophylaxis. Interferon-γ release assay (IGRA) is used to diagnose LTBI and is particularly useful in profoundly ill patients and those with severe malnutrition. Two
in vitro IGRA tests are available: QuantiFERON- TB Gold (Cellestis, USA) and the T SPOT-TB test (Oxford Immunotec, USA). Both use an enzyme- linked immunospot assay to quantify the number of peripheral blood mononuclear cells (PBMCs) producing IFN- γ in response to TB-specific antigen stimulation (ESAT-6 and CFP10). Both assays have sensitivity (as measured in patients with ATB) comparable to that of the TST, but are significantly more expensive [
14]. IFN-γ assays do not differentiate between LTBI and ATB or between immune reconstitution inflammatory syndrome (IRIS) and failure. Studies suggest that IGRAs are ideal for serial testing because these can be repeated without boosting [
15‐
17]. These are also unaffected by previous BCG vaccination and require fewer patient visits. However, WHO recommended against the use of IGRAs for diagnosis of active or latent TB, in resource-limited settings [
18,
19].
There is epidemiologically evidenced high risk for acquiring ATB (over 20% per year) among HIV-1 co-infected persons [
20]. This has been used to argue the incipient nature of LTBI among HIV+ve people and is the globally acceptable rationale underlying recommendations for isoniazid-INH chemo-prophylaxis. Overall, TB is the most common opportunistic infection (OI) among HIV-infected individuals, and co-infected individuals are at high risk of death [
21‐
23]. TB may occur at any stage of HIV disease and is frequently the first recognized presentation of underlying HIV infection [
24,
25]. As compared to people without HIV, people living with HIV (PLWH) have a 20-fold higher risk of developing TB [
26,
27] and the risk continues to increase as CD4 cell counts progressively decline [
24,
26,
27]. Although antiretroviral therapy (ART) can reduce the incidence of TB both at the individual and population level, PLWH on ART still have higher TB incidence rates and a higher risk of dying from TB [
28,
29]. However, co-administration of ART along with anti-TB therapy presents several management challenges, including drug-drug interactions, overlapping drug toxicities and immune reconstitution syndrome. This emphasizes the importance of routine TB screening among PLWHA to not only identify those without TB, but if possible, prevent TB by chemoprophylaxis as well as to diagnose and promptly treat TB. In absence of accurate tests for high-risk LTBI, WHO recommends INH chemo-prophylaxis for all PLWHA [
30‐
32].
Our group recently demonstrated the capability of TMKmt Ag and Ab assays to differentiate between ATBI and LTBI or NTB [
33]. In the current study, we—riding on the evidence that LTBI among HIV-1 co-infected persons is high-risk, set out to examine LTBI diagnostic potential of TMKmt assays. Specifically, in light of the above described existing epidemiological data pointing to the high-risk of LTBI among HIV-1 co-infected persons to progress to ATB relative to their HIV-1 negative counterparts, it was argued that a point cross-sectional study of TMKmt Ag levels among HIV-ve LTBI relative to HIV-ve LTBI participants, can inform the diagnostic potential of TMKmt Ag assays for high risk LTBI. As a secondary outcome, we also examined the ability of TMKmt host specific IgM and IgG to differentiate between NTB and LTBI relative to the QuantiFERON-TB GOLD® assay.
Discussion
We present data to support the view that TMKmt Ag level are a potentially more precise and accurate biomarker for incipient LTBI relative to existing assays. Brust B, et al. (2011) recently intimated that those pathogen antigens whose secretory levels depend on the physiology of M.
tb represent the best candidates for research and development of TB immune-diagnostics [
34]. Our group has previously shown that TMKmt Ag levels represent a predictive (foretelling) surrogate biomarker for both
in-vitro and
in-vivo growth and proliferation of M.
tb [
33,
35,
36]. Growth and proliferation is a physiologic change that assails M.
tb exit from dormancy [
37‐
39]. Arguably, there should be more active growth and proliferation of M.
tb among high-risk LTBI relative to low risk LTBI. Here, using a conceivably novel design of cross-sectional study, we present validation that TMKmt Ag levels are objectively a surrogate biomarker for high-risk LTBI.
First, we show similarly low levels of TMKmt host specific IgM (ODs <0.03) captured by EIAs premised on our custom peptide epitopes (see Figs.
1 and
2 respectively) among both NTB and TB exposed household contacts (see Fig.
3 and Tables
1 and
2). For details, see Additional file
1. Exposure to M.
tb within high TB endemic areas likely occurs immediately following birth [
1‐
5]. As a consequence, IgM levels are bound to only be high within 1 to 3 months following the initial exposure to M.
tb after which they wane and are replaced by IgG levels. Assays of levels of TMKmt host specific IgM are therefore not usable for differentiating between NTB and LTBI, but might be relevant towards evaluating TB exposure among natives of low TB endemic areas (objectively NTB) who travel to TB high endemic areas and return home within 1-3 months.
Second, we show that levels of TMKmt host specific IgG can clearly differentiate NTB from LTBI. Specifically, IgG levels were 0.5155+0.07675 (95% CI: 0.3528 to 0.6782) and 0.3277+ 0.04226 (95% CI: 0.2381 to 0.4173) among NTB controls (see Table
3) compared to 1.020+0.1183 (95% CI: 0.7854 to 1.256) and 1.209 +0.1209 (95% CI: 0.9689 to 1.449) (see Table
4) among LTBI (all captured by the TMKmt epitope peptides UG-peptide 1 and 2 based EIAs, respectively (see Fig.
4). This data is consistent with our prior work that found that levels of IgG among NTB are =OD<0.88 (95% CI: 0.1527 to 0.8751) compared to LTBI=0.255>OD<1.00 (95% CI: 0.2690 to 0.6396) [
33]. Indeed, a subsequent attempt to re-categorize the 40 TB exposed household contacts as NTB and LTBI by the QuantiFERON TB Gold® assay alongside the TMKmt-IgG assay revealed high IgG levels among both clusters (see Fig.
5); suggesting that QuantiFERON TB Gold® might be inaccurate towards differentiating LTBI from NTB [
3‐
19]. For details, see Additional file
1. Interestingly, our early work also showed higher TMKmt host specific IgG among HIV+ve persons relative to the HIV-ve persons, a finding we attributed to the myco-septicemia that assails the disseminated nature of TB disease among HIV-1 co-infected persons [
33]. These data support the high-risk nature of LTBI among HIV+ve co-infected persons. It is on basis of these and existing epidemiological data that we moved to examine the ability of TMKmt Ag detection as a surrogate for incipient LTBI (assuming that all LTBI among PLWHA is high-risk)[
20‐
30,
33].
In line with the primary hypothesis of this study (that LTBI among HIV+ve persons is high risk)—we show that TMKmt Ag levels among HIV+ve LTBI are above those of HIV-ve LTBI. Specifically, we show that HIV+ve LTBI presents with higher TMKmt Ag levels (>0.14: 0.2676 ± 0.0197 [95% CI: 0.2279 to 0.3073]) relative to HIV-ve LTBI (<0.14: 0.1069 ± 0.01628[95% CI: 0.07385 to 0.14]) (see Table
5). These data support the prevailing WHO & STOP TB Partnership recommendations for isoniazid (INH) prophylaxis among PLWHA within high TB endemic areas [
25‐
30]. As was the case for TMKmt host specific IgG levels, we similarly noted that some QuantiFERON-TB Gold® assay pre-qualified NTB controls had TMKmt Ag levels that lay within ranges of HIV-ve LTBI (0.1013 ± 0.02505 [95% CI: 0.0421 to 0.1606]) (see Figs.
6,
7 and
8). For details, see Additional file
1. Two incidental HIV-ve LTBI had high TMKmt Ag levels, possibly due to false negative HIV-1 test, or another form of physiological or biological immune-deficiency.
In order to offer a hint on the performance of TMKmt Ag and host specific IgG levels for the designation of TB status, we a re-evaluated in-house data from our prior study data for receiver operator characteristics [
33].
On one hand: (a) for n=128 and prevalence of 80.0 (95% CI: 73.0,86.9), the sensitivity, specificity, PPV and NPV of pre-set TMKmt Ag capture-EIA-OD cut-off for differentiating ATB from NTB alone at 95% CI were respectively
: 99.0 (94.7,100.0),
68.0 (46.5,85.1),
92.7 (86.2,96.8), and
94.4 (72.7,99.9) [yielding a ROC-area of
83.5 (95% CI: 74.1, 92.9)] compared to
96.6 (91.4, 99.1),
21.1 (9.5, 37.3),
78.9 (71.2,85.3) and
66.7 (34.9,90.1) [ROC-area
58.8 (52.0,65.6)] obtained using pre-set TMKmt host-specific IgG-EIA-OD cut-offs on n=154 participants and prevalence of 74.0 (68.0, 81.9).
On the other hand, (b) for n=220 and prevalence of
63.0 (95% CI: 56.0, 69.1), the sensitivity, specificity, PPV and NPV of pre-set TMKmt Ag capture-EIA-OD cut-off for differentiating ATB from both LTBI & NTB at 95% CI were respectively:
73.9 (65.8,81.0),
90.2 (81.7,95.7),
92.7 (86.2,96.8) and
67.3 (57.7,75.9) [ ROC-area of
82.1 (77.2,87.0)] compared to
92.6 (86.3,96.5),
34.8 (21.4,50.2),
78.9 (71.2,85.3) and
64.0 (42.5,82.0) [ROC-area
63.7 (56.3,71.0)] obtained using pre-set TMKmt host-specific IgG-EIA-OD cut-offs for
n=167 participants and prevalence of 72.0 (65.0,79.1) (for a summary, see Tables
6,
7 and
8). These sensitivity results are above the documented values for smear microscopy for AFBs (~45%) and close or above those of GenXpert® (72.5 %) [
19]. The observed high prevalence rates are as a result of using already pre-qualified specimen rather than undertaking a prospective recruitment and testing.
Table 6
Cut-offs values associated with TB status: UG-Peptide 1 based TMKmt host specific Antibody (IgG) capture EIAs (HIV+ve)
Active TB (ATB) | ATB=OD>1.00 | 1.170 to 1.528 |
Latent M.tb Infection (LTBI) | LTBI=0.255>OD<1.00 | 0.2690 to 0.6396 |
No TB (NTB) | NTB=OD<0.88 | 0.1527 to 0.8751 |
Table 7
Cut-offs values for TB status: PAb-655 based TMKmt Antigen capture EIAs (HIV+ve)
Active TB (ATB) | ATB=OD>0.490 | 0.7446 to 0.8715 |
Latent M.tb Infection (LTBI) | LTBI=0.26>OD<0.490 | 0.4325 to 0.4829 |
No TB (NTB) | NTB=OD<0.26 | 0.1675 to 0.2567 |
Table 8
Receiver operator characterization of TMKmt Ag and host specific IgG EIA-ODs for detecting TB exposure and or disease status among HIV positive individuals
TMKmt Ag Capture |
n | 128 | 220 |
Prevalence | 80.0 (73.0,86.9) | 63.0 (56.0,69.1) |
Sensitivity | 99.0 (94.7,100.0) | 73.9 (65.8,81.0) |
Specificity | 68.0 (46.5,85.1) | 90.2 (81.7,95.7) |
PPV | 92.7 (86.2,96.8) | 92.7 (86.2,96.8) |
NPV | 94.4 (72.7,99.9) | 67.3 (57.7,75.9) |
ROC area | 83.5 (74.1,92.9) | 82.1 (77.2,87.0) |
TMKmt Ab Capture |
n | 154 | 167 |
Prevalence | 74.0 (68.0, 81.9) | 72.0 (65.0,79.1) |
Sensitivity | 96.6 (91.4, 99.1) | 92.6 (86.3,96.5) |
Specificity | 21.1 (9.5, 37.3) | 34.8 (21.4,50.2) |
PPV | 78.9 (71.2,85.3) | 78.9 (71.2,85.3) |
NPV | 66.7 (34.9,90.1) | 64.0 (42.5,82.0) |
ROC | 58.8 (52.0,65.6) | 63.7 (56.3,71.0) |
A key limitation of our work—as is the case for all projects that aim to advance novel biomarkers for LTBI, is the absence of a gold standard for precisely designating LTBI and NTB [
3‐
19]. As noted, we maneuvered around this challenge using epidemiological data around LTBI in context of HIV-1 co-infection; and the same design might be relevant to global TB diagnostic biomarker studies [
4,
6‐
14,
40‐
44,
45‐
49]. However, the best approach would be a prospective follow-up cohort to recruit high-risk LTBI among HIV+ve persons that eventually develop TB.
Second, our sample size is small and a larger validation study is needed. Third, much as host immunological and transcriptomic profiles have been studied in an attempt to expose markers of TB disease progression, our group focused on a marker for growth and proliferation of the pathogen (M.
tb) as a surrogate for disease progression. Although several M.
tb targets inclusive of lip-arabinomannose (LAM), IP-10, early secretory antigen 6(ESAT-6) and colony filtrate protein 10 (CFP-10) have previously been developed for the purpose of detecting TB, none meets the criteria for designating delineating high-risk LTBI [
3‐
7]. On part of the host, however, recent data report that sequential inflammatory processes and a whole blood RNA signature might possess TB disease progression detection capability that are better than the afore listed M.
tb targets [
50,
51]. It might therefore be worthwhile to adopt assays of TMKmt Ag levels as a new pathogen target for expanded clinical testing towards accurately designating incipient TB. Last but more important to note is that INF-Ƴ responses—which are a correlate of memory, might not present the best biomarker for differentiating between high and low-risk LTBI especially within high TB endemic areas [
3‐
19].
Acknowledgements
We thank Mr. Ronald Ssenyonga who did the secondary analyses for sensitivity, specificity, PPV, NPV and ROCs. Ms. Geraldine Nalwadde and Ms. Joanitta Basemera (Dept of Medical Microbiology, MakCHS) alongside Ms. Harriet Nambooze (MEPI-MESAU Office, MakCHS) for their administrative assistance. Staff of the Immunology and the Mycobacteriology Level III at MaKCHS offered us invaluable technical support. Also, Mr. Nestar B Mugaba and Ms. Christine Zaake at the Sida/SAREC Office (Directorate of Research and Graduate Training, Makerere University) offered us administrative assistance. Dr. Irene Andia provided samples for the TB exposed house-hold contacts. Investigators for the Makerere University-Case Western Reserve University TB Research Unit (TBRU) provided broadly consented participant serum samples for LTBI. Dr. Benson Ouma provided the gift of 9 serum samples of TB naive American donors used as negative controls.