Background
Pulmonary tuberculosis is the most common clinical manifestation of
Mycobacterium tuberculosis (MTB) infection. Among extrapulmonary presentations pleural tuberculosis (PLTB) is the most frequent in many countries, representing 15% of the extrapulmonary cases globally reported in 2016 [
1‐
3]. Moreover, PLTB is the major cause of pleural effusions, responsible for approximately 50% of all related diagnoses in Brazil [
4,
5]. The disease generally affects immunocompetent young adults, but, although patients may spontaneously heal, there is a risk of developing active tuberculosis (TB) in the absence of specific drug administration. Thus, early diagnosis and adequate treatment are required to avoid PLTB evolution to tuberculous empyema or pleural fibrosis [
6,
7].
Microbiological methods depend on the presence of bacilli in clinical specimens, therefore displaying low sensitivity in paucibacilary cases, like PLTB patients. The use of molecular techniques is costly and restricted to centers with adequate structure and specialized professionals. Histopathological examination of pleural biopsy samples comprises the standard PLTB reference test, but exhibits variable sensitivity, is expensive, time-consuming, requires skilled personnel and an invasive procedure [
8,
9]. Serological tests based on antibodies (Ab) response may be an attractive alternative as a diagnostic method, as they are simple, fast and easy to operate, allowing their application in the public health system, where the diagnostic demand is high. In addition, they are not dependent on the presence of the bacilli in clinical specimens. In a previous study, Araujo et al. [
10] demonstrated that the MT10.3:MPT64 fused antigen is recognized by immunoglobulin A (IgA) in the pleural fluid (PF) of PLTB patients, at high sensitivity (81.4%) and specificity (95.5%). Furthermore, the cloned fusion protein based on MPT64 and MT10.3 epitopes, termed F2 (MT10.3
(1M-40S):MPT64m
(91L-205A):MT10.3
(41S-96)), was described to be predominantly recognized by serum IgA in pulmonary TB [
11]. The PE (Pro-Glu) and PPE (Pro-Pro-Glu) molecules are coding by two families of genes responsible for about 10% of the MTB genome [
12]. Proteins belonging to PPE family are involved in many aspects of TB pathogenesis, including antibody recognition and bacilli persistence ability in granulomas [
13‐
16]. Moreover, their antigenic variation has been associated with the ability of determined microbial pathogens to evade the immune system [
17]. The PPE59 protein coding gene
rv3429c belonged the RD11, region absent in
M. bovis. The actual role of PPE59 in metabolism or in the clinical evolution of MTB infection is not well known, but it is capable of inducing a cell-mediated response by IFN-γ and interleukin-10 [
13,
15]. However, no clues on the reactivity to these antigens (Ags) by Ab of PF and/or serum of PLTB cases are available.
On the other hand, the use of adenosine deaminase (ADA) biochemical test may contribute to a more efficient and differential diagnosis in PLTB cases [
1,
18‐
20]. Although it is a fast, cheap, reproducible and an easy to perform assay [
8,
9], it presents several limitations: the predictive value depends on the local TB prevalence, low sensitivity in immunocompromised patients, false-positive results due to cross-reactivity with other diseases presenting pleural effusion, such as lymphoma, vascular collagen diseases and bacterial empyema. Furthermore, due a wide range of results from region to region should be used with caution in countries with low TB incidence [
21‐
23]. Therefore, identification of new biomarkers and, consequently, the development of diagnostic test that differentiate PLTB from other pathologies presenting exudative pleural effusion may have a significant impact on primary care. In this context, here we investigated potential PLTB biomarkers, validating and/or evaluating the accuracy of dosing IgA and IgG response to mycobacterial single Ags MT10.3, MPT64 and PPE59, fusion proteins MT10.3:MPT64 and the novel F2, in PF and sera from patients with pleural effusion using an in-house ELISA method.
Discussion
Simple, fast and accurate diagnostic methods in PLTB are still required. For instance, the reference histopathological examination for pleural biopsy was virtually not performed in the present sampling as compared to Araújo et al [
10]. On the contrary, the diagnosis herein relied on clinical suspicion and the simple ADA test for all patients, reaching 79% sensitivity. However, PF IgA MT10.3:MPT64 (86.2%) and it combined results with /−MPT64 or /−F2 identified all, except one, PLTB negative ADA, and, despite missing three or two of the positive ADA PLTB patients, the ELISA improved pleurisy tuberculosis detection (89.6%). The combinatory results for PF (1:50) IgA MT10.3:MPT64/−F2 and ADA displays a double advantage, as, besides increasing sensitivity (96.6%), the PF can be used at the same dilution for both fused Ags when compared to the combinatory results of different PF dilutions used to obtain the same high sensitivity, thus making them suitable as a diagnosis test to aid in PLTB diagnosis. Moreover, the combinations of tests results can obviate the need for a pleural biopsy during the initial diagnostic and, as the ELISA and ADA are fast techniques and of low costs, easy to operate and to apply in public health systems, they favor a rapid auxiliary PLTB diagnosis. Recently, authors have suggested the use of ADA in combination with other diagnostic tests, such as interferon-gamma (IFN-γ) release assays, and in the search for new biomarkers [
18‐
20,
29‐
32]. The present study provides a simpler alternative ELISA for ADA combination.
Notably, the better fused antigen MT10.3:MPT64 performance for PF IgA detection corroborates previous studies [
10], although the lower reactivity on single antigens MT10.3 (20.7%) and MPT64 (24.1%) was not expected and is in disagreement with the previous report (72%) assessing populations carrying similar pleurisy morbidities to the present study [
33]. One explanation could be antigen degradation, as the batches used herein were old and underwent several freezing and thawing cycles. On the other hand, for serum, the present results corroborate those reported by Silva et al. [
34] where, even evaluating pulmonary TB, both single antigens failed to show high sensitivity (< 34.5%). Nevertheless, reduced recognition of these antigens by serum IgA and IgG (including to PPE59) was also detected.
It is well known that IgA is a predominant mucosal/serous immunoglobulin, also secreted in the pleural space [
35,
36]. However, IgA against mycobacterial Ags has been found in pulmonary TB patient sera by several authors [
11,
37‐
40]. Zhao et al. [
41] suggested Ab-based tests in plasma to identify Beijing MTB infection. In our study, IgG and IgA resulted in lower serum positivity compared to PF results, but the frequency of positive results is higher to that in microbiological tests, therefore useful in their absence and other, less sensitive, conventional PLTB diagnostic tests. The use of recombinant fused antigens in diagnostic tests constitutes an interesting strategy to increase accuracy and simplicity. However, fusion of different antigens may change reactivity according to gene position in their construction [
10,
11,
42]. The higher sensitivity elicited by MT10.3:MPT64 may be related to this fact, since the fusion protein expressed by opposite genes according to that construction seems to display low stability and quality (personal communication).
As expected, histopathological examinations demonstrated higher sensitivity than the conventional microbiological methods for PLTB diagnosis, but lower compared to the developed ELISA. Nonetheless, among the patients with negative results or no information, the ELISA PF IgA-MT10.3:MPT64 was able to identify most PLTB patients, which could be supportive of the hypothesis that, in some cases, pleural effusion may also be caused by the possible entry of mycobacterial antigens into the pleural space, thereby stimulating the presence of IgA at this site [
43]. Conversely, few OPL patients diagnosed by histopathology presented false-positive ELISA IgA-MT10.3:MPT64 results (1/16, 6.2%).
The F2 (MT10.3(1M-40S):MPT64m(91L-205A):MT10.3(41S-96)) containing peptides of both antigens led to decreased mean levels of reactivity to almost half among PLTB cases (0.553 ± 0.230), but no change for OPL cases (0.351 ± 0.212) compared to the full fused genes (0.821 ± 0.213 or 0.387 ± 0.188, respectively). However, F2 recognition by a PLTB group added sensitivity to the combinatory of results using both fusion antigens for PF IgA detection. This may be explained by exposure of other epitopes not available in the full gene construction or single antigens.
Until now, no publications associating the F2 protein chimera or PPE59 in PF IgA detection in PLTB patient are available. Despite their lower sensitivity compared to MT10.3:MPT64, sensitivity was higher to that AFB smear exams. Among microbiological tests used for TB diagnosis, AFB is the flagship, as it is easy to apply, rapid and cheap, although exhibiting low sensitivity, nevertheless considering that the ELISA has the potential to be point of care (POC) test, similar results to those found in bacterioscopy assessments may be obtained quickly. Serum IgA-F2 elicited the best reactivity compared to the other antigens, detecting half of the PLTB cases (51.7%); although in pulmonary TB this positivity was evidenced for IgG–F2 [
11], a discrepancy which may be related to the different clinical TB forms. It is possible that the compliance of the fused protein, where the repositioning of the gene fragments led to the formation of new epitopes and, consequently, obfuscation of others, could positively affect IgA or IgG recognition in PL or pulmonary TB patient sera. The third single antigen used in this study, PPE59, performed similarly, at low reactivity for serum IgG in PLTB, as previously observed for pulmonary and different extrapulmonary TB cases (25 and 0%, respectively) (unpublished data). However, a higher PF (1:100) IgA-PPE59 (62.1%) sensitivity compared to that unpublished data (54 and 28%, respectively) is described herein for the first time. To date, PPE59 it is known to induce cell-mediated response by IFN-γ and interleukin-10 in pulmonary TB [
12,
15] and, as a result of our study, immunodominance of PF IgA but non-IgG in PLTB patient. Therefore, PPE59 protein displays potential as specific TB diagnosis markers and may assist in the diagnostic investigations performed in PLTB.
Some Bacille Calmette Guérin (BCG) vaccine strains, including the one used in Brazil, contain the gene coding for the MPT64 antigen and it has been hypothesized that tumor cells and the BCG strain may share antigens [
10,
44,
45]. In the present study, cross-reactivity was observed, as in previous similar studies [
10,
26,
46,
47]. In our and in the other similar Brazilian [
10] studies the majority of participants were BCG vaccinated and, more important, they live in a city with a high TB incidence rate [
48,
49]. Thus, it is possible the OPL patients may have presented TB as undiagnosed co-morbidity or were LTBI, since some of them presented QFT-GIT® positive results. Lupus autoimmune disorder based diagnostics rely solely on clinical examination, negative PF culture or AFB and positive for the ADA assay. Another possible explanation is that, 64% of OPL cases presented cancer and higher mean IgA levels in both PF and serum, as well as PF IgG, compared to other OPL cases. False-positive ELISA results in patients with lung cancer have been reported by other authors [
33,
35].
This study has some limitations, such as: 1) in the absence of a precise diagnostic tests for PLTB and the limitations of the ADA, it is possible that patients were in some stage of TB status, including LTBI, but were mistakenly classified as OPL because not all the resources available for the TB diagnosis were applied, 2) patients recruitment was performed in a single healthcare center, which contributes to the limitation in the number of the cases included, 3) failure to follow the clinical evolution of the patients, which would make it possible to ascertain TB disease in false-positive cases in the tests proposed here.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.