Transforming growth factor-β and matrix metalloproteinases as potential biomarkers of fibrotic lesions induced by tuberculosis: a systematic review and meta-analysis

Abstract

Objectives

Very few studies and limited information are available regarding the mechanism of fibrosis in tuberculosis (TB). This study aimed to identify, describe and synthesise potential biomarkers of the development of tissue fibrosis induced by TB through a systematic method and meta-analysis.

Methods

A literature search was performed using keywords according to the topic from electronic databases (ScienceDirect and PubMed) and other methods (websites, organisations and citations). Studies that matched predetermined eligibility criteria were included. The quality assessment tool used was the Quality Assessment of Diagnostic Accuracy Score 2, and the data obtained were processed using Review Manager V.5.3.

Results

Of the 305 studies, 7 met the eligibility criteria with a total sample of 365. The results of the meta-analysis showed that the post-TB group of patients with pulmonary parenchymal fibrosis had a higher transforming growth factor (TGF)-β level (6.09) than the control group (1.82), with a 4.27 (95% CI: 0.92 to 7.61) mean difference. Moreover, patients with residual pleural thickening post-TB had a higher mean of TGF-β (0.61) than the control group (0.56), with a 0.05 (95% CI: 0.04 to 0.06) mean difference. Besides TGF-β, our qualitative synthesis also found that matrix metalloproteinase-1 might have a role in forming and developing pulmonary tissue fibrosis, thus, could be used as a predictor marker in the formation of fibrotic lesions in patients with TB. In addition, several other biomarkers were assessed in the included studies, such as tumour necrosis factor-α, interleukin (IL)-4, IL-8, IL-10, plasminogen activator inhibitor-1 and platelet-derived growth factor. However, this study is not intended to examine these biomarkers.

Conclusions

There were differences in the results of TGF-β levels in patients with fibrotic lesions compared with controls. TGF-β might be a biomarker of fibrotic tissue formation or increased pulmonary tissue fibrosis in post-TB patients. However, further studies are needed on a larger scale.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• There is no limitation to the time frame used in the literature search of this study.

• Subgroup analysis was done for different outcomes: parenchymal lung fibrosis and residual pleural thickening.

• Other biomarkers found in included studies are discussed for their potentiality.

• Small sample size.

• Language selection of the literature is limited to English and Bahasa.

Introduction

Globally, there were more than 1.3 million deaths due to tuberculosis (TB), and it ranked as the second leading cause of death from a single infectious agent after COVID-19 in 2020.¹ With its large number, TB is one of the tremendous health threats to human. Minor abnormalities to massive respiratory failure can increase the risk of TB mortality.^{2 3} Despite being a curable disease and can be successfully treated with antibiotics, more than 50% of TB survivors have persistent pulmonary/extrapulmonary impairment, one of which is tissue fibrosis.⁴

Tissue fibrosis can result from a long-term tissue injury that includes activation of fibroblast and extracellular matrix deposition, leading to distortion of normal tissue architecture, such as thickening and stiffening.⁵ This can happen in the lung parenchyma as a result of pulmonary TB and extra lung, usually due to tuberculous pleuritis (TBP).^{6 7}

Excessive collagen deposition and scarring can occur throughout TB disease, during and after treatment.⁸ It can happen through some immune pathways or as subsequent granulomas and cavitations.⁶ But the factors that promote fibrotic processes precisely in TB are still poorly understood. Transforming growth factor (TGF)-β is believed to be the principal mediator of fibrogenesis by stimulating fibroblast differentiation into myofibroblast.⁹ Matrix metalloproteinases (MMPs) are considered to have a major role in tissue remodelling and inflammatory conditions.¹⁰

However, to date, there has not been any systematic study that explores fibrosis in TB. Thus, we conducted a systematic review to identify, describe and synthesise potential biomarkers of the development of TB-induced tissue fibrosis.

Method

Eligibility criteria

The following study criteria were used in this systematic review: (1) population: patients >18 years old who had pulmonary TB or currently undergoing treatment for TB; (2) exposures are TGF-β or MMP; (3) outcome is the incidence of pulmonary fibrosis or the severity of pulmonary fibrosis; (4) observational study design (cohort/case–control/cross-sectional); (5) articles in Indonesian or English.

Literature searching

Literature study was conducted on 15 April 2022 from various databases: ScienceDirect and PubMed using the keywords (((pulmonary tuberculosis) OR (pulmonary tuberculoses)) AND ((Transforming growth factor beta) OR (TGF-beta) OR (Matrix Metalloproteinase) OR (MMP) OR (risk factor)) AND ((Pulmonary fibrosis) OR (Lung fibrosis))). In addition, several valid studies outside of the database were included if they match the criteria. We also screened reference lists of published reviews to identify additional relevant studies. More details can be seen in the online supplemental file 1.

Selection process

After the search was carried out, duplicate studies from multiple sources were identified and then excluded. The remaining literature was screened by reading the titles and abstracts of all studies obtained from the search. At least two independent reviewers carried out the screening process. Literature studies that match the eligibility criteria were included, while those that do not meet the requirements were excluded for reasons. If there is any conflict in categorising the studies, it was discussed until an agreement is reached. The results of the literature screening are reported using the Preferred Reporting Items for Systematic Review and Meta-Analyses¹¹ (online supplemental file 2).

Data collection

Two reviewers performed data extraction from eligible studies. The extracted data are compared, and differences are resolved through discussion. The first reviewer collected the data in Microsoft Excel and double-checked them for accuracy. When the information regarding the data was unclear, we contacted the authors to provide further details. If the author does not reply, the study is withdrawn with the consent of other reviewers.

The following data were collected for all included studies: (1) main author; (2) year of publication; (3) place where the research was conducted; (4) characteristics of the sample (race, age, sex); (5) number of samples; (6) type of exposure; (7) type of outcome; (8) value or number of incidence of exposure; (9) incidence rate data from the outcome.

Quality assessment

We evaluated the quality and risk of bias of all included studies for the four domains of the Quality Assessment of Diagnostic Accuracy Score 2 (QUADAS-2), including patient selection, index test, reference standard, and flow and timing. Each domain was evaluated using a set of QUADAS-2 guiding questions. Items were scored as ‘high risk’, ‘low risk’, ‘some concerns’ or ‘unclear/no information’. The overall risk of bias was evaluated as ‘high risk’ for studies with more than one area of high risk, ‘some concerns’ for all studies that included one area of high risk, ‘low risk’ for all studies with two or more areas of low risk and no high risk, and ‘no information/unclear’ for all studies with three or more areas of unclear/no information concerns and no high risk (online supplemental file 3).¹² The results of the risk of bias assessment for all included studies will be presented in the form of traffic light and weighted bar plots.

Statistical analysis

We planned to analyse continuous outcomes by calculating mean difference (MD) and their 95% CIs. Differences in outcome units reported in studies are then converted following the international system. Then, the data are tabulated visually using Microsoft Excel before meta-analysis is performed.

Given the complexity of the results of included studies, we conclude two types of outcomes that are judged to be associated with TB-induced fibrotic lesions.

1. Formation of new fibrotic lesions in patients who previously did not have any.

2. Increased fibrotic lesions in patients with pre-existing fibrous lesions.

In conducting a meta-analysis, we combined these two outcomes into one type of outcome: fibrosis/increased. We do this because this study focused on the influence of a biomarker on the formation or worsening of fibrotic lesions in patients infected with TB.

The data obtained were processed using Review Manager V.5.3. Meta-analysis was performed by entering the mean value of TGF-β in the fibrosis/increased fibrosis lesion and non-fibrosis/stable fibrosis lesion groups to measure the MD (95% CI) of studies that met the criteria for inclusion in the quantitative analysis. The heterogeneity of the statistical analysis was seen from the I² value. The fixed-effects model is used if I²<40%, while the random-effects model is used if I²≥40%.

In conditions where high heterogeneity is found, it can be considered to do a subgroup meta-analysis by dividing based on the location of the lesion, the first group of lesions in the lung parenchyma and the second group of lesions on the pleura (pleural thickening). The results of the analysis are said to be significant if the CI value of the overall MD does not touch the zero point.

Publication bias

The funnel plot was used to evaluate publication bias subjectively. Publication bias is high if the distribution of studies in the funnel plot is not symmetrical. Conversely, if the study distribution is evenly distributed and symmetrical in the funnel plot, it can be said that publication bias is low.

Patient and public involvement

Patients and/or the public were not involved in this review and meta-analysis.

Results

Literature search and result of screening

After searching literature studies from various databases (PubMed and ScienceDirect) using the keywords, 299 studies were obtained; outside of the database, we found six studies that are also related to the topic, so we found a total of 305 studies. The study results found are studies published from 1968 to 15 April 2022. Then, the studies were filtered according to predetermined criteria. Four duplicate studies were excluded. The remaining 301 study titles and abstracts were screened independently by two reviewers. A total of 294 studies were excluded because they did not comply with the eligibility criteria. As the final result, 7 studies met the criteria with a total sample size of 365 and were included for qualitative and quantitative analyses. Full details of the search and filter results are presented in figure 1.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart.

Characteristics of eligible studies

The seven included studies were primary studies conducted in five countries with a total sample size of 365. Of the seven studies, one used a case–control design,¹³ two used a cross-sectional design^{14 15} and four used a prospective cohort study design.^{10 16–18} Five studies assessed the outcome of parenchymal lung fibrosis, while the other two assessed residual pleural thickening (RPT). All studies assessed the potential biomarker in the formation of TB-induced fibrotic lesions. Five studies evaluated the correlation between fibrotic lesions and TGF-β level, and three studies assessed the correlation between fibrotic lesions with MMP-1 level. Full details of the characteristics of the included studies are presented in table 1 and online supplemental file 4.

Table 1.

Characteristics of included studies

First author (year)	Location	Study design	Sample size	Mean age (SD)	Treatment status	Specimen	Biomarker	Outcome
Christine (2019)13	Indonesia	Case–control	TB=51	40.75 (14.3)	Completed anti-TB treatment	Serum	TGF-β	Parenchymal lung fibrosis
Chen (2015)16	Taiwan	Prospective cohort	TB=18 C=18	TB=62 (24) C=72 (20)	Completed anti-TB treatment	Pleural fluid	TNF-α, MMP-1, MMP-7, MMP-9,	Residual pleural fibrosis
Wang (2010)10	Taiwan	Prospective cohort	TB=98 C=49	TB=50 (17.9) C=46.1 (14.9)	Completed anti-TB treatment	Blood	MMP-1, MMP-9, MMP-12	Parenchymal lung fibrosis
Seiscento (2007)17	Brazil	Prospective cohort	TB=50 C=8	TB=39 (18) C=64 (14)	Completed anti-TB treatment	Serum and pleural fluid	TGF-β	Pleural thickening
Astuti (2018)14	Indonesia	Cross-sectional	TB=20 C=10	NA	Did not receive anti-TB drugs yet	Plasma	TGF-β, TNF-α, IGF-1	Parenchymal lung fibrosis
Ameglio (2005)18	Rome	Prospective cohort	TB=13	34	Completed anti-TB treatment	Serum and bronchoalveolar lavage fluid	TGF-β, IFN-γ, TNF-α, IL-4, IL-10, PDGF-BB	Stable or increasing fibrosis score
Shanmugasundaram (2022)15	India	Cross-sectional	PTBS=10 PTBWS=10 C=10	PTBS=46.2 (10.05) PTBWS=37.1 (9.52) C=37.5 (9.95)	Completed anti-TB treatment	Serum	TGF-β, MMP-1, IFN-γ	Residual chest radiographic abnormality

C, control group; IFN-γ, interferon-γ; IGF-1, insulin-like growth factor-1; IL, interleukin; MMP, matrix metalloproteinase; NA, not available; PDGF-BB, platelet-derived growth factor-BB; PTBS, post-tuberculosis sequelae; PTBWS, post-tuberculosis without sequelae; TB, tuberculosis group; TGF-β, transforming growth factor-β; TNF-α, tumour necrosis factor-α.

Quality assessment result

The quality assessment results using QUADAS-2 show that four studies fall into the low-risk category, three fall into the some concerns category and none falls into the high-risk category (figure 2). In particular, the risk of bias for patient selection tends to have a low risk of bias because none of the included studies used a case–control design, but three studies were included in the some concerns category because the sampling method was not explained. For the domain index test, all studies are given a low-risk score because the calculation of the index test is considered objective and quantitative by a calibrated measuring instrument, so the risk of bias in assessing the index test is certainly low even though there is no mention of a blind method for measuring the index test. Regarding reference standards, some studies are considered high risk because researchers determine whether or not fibrosis exists without using a blind method, so the risk of bias is high, because reading the results of a radiology photo is subjective and highly dependent on the reader. Lastly, for flow and timing, two studies explained that their studies used prospective studies, but the timing of index test and reference standard measurements was not explained in detail. Therefore, the results were given no information.

Figure 2.

Summary of the Quality Assessment of Diagnostic Accuracy Score 2 risk of bias assessment.

TGF-β levels in fibrotic lesion induced by TB

Five studies assessed the TGF-β level as a potential biomarker in forming fibrotic lesions in patients with TB with a total of 197 participants.13–15 17 18 Four studies reported that the TGF-β level had a significantly higher level in the presence of tissue fibrosis compared with non-fibrosis.13 14 17 18 Meanwhile, one study reported that TGF-β levels were higher in fibrotic lesions than in non-fibrotic lesions, but the difference was statistically insignificant.¹⁵ Furthermore, two studies assessed the difference in TGF-β levels in minimal lung fibrosis and extensive lung fibrotic lesion compared with the healthy group.^{14 17} Of the two studies, one study reported that the level of TGF-β was significantly higher in extensive lung fibrotic lesions compared with minimal fibrotic lesions.¹⁴ Meanwhile, the other study reported an opposite result might be due to the small size of samples used in the study.¹⁷ All of these studies showed that the TGF-β level has promising results as a predictor marker in the presence of lung fibrosis in patients with TB.

MMP-1 levels in fibrotic lesion induced by TB

Three studies assessed the MMP-1 level as a potential biomarker in the formation of fibrotic lesions in patients with TB, with a total of 202 participants.^{10 15 16} Of three studies, one reported that MMP-1 polymorphisms had a significant correlation with the formation of parenchymal lung fibrosis in patients with TB. Furthermore, this study showed that the frequency of MMP-1 was significantly higher in patients with TB with moderate to advanced pulmonary fibrosis compared with minimal to mild fibrosis.¹⁰ Meanwhile, two other studies reported the association of MMP-1 level with residual pleural fibrosis.^{15 16} Of two studies, one study measured pleural fluid levels of MMP-1 and showed that MMP-1 was positively associated with residual pleural fibrosis of TBP.¹⁶ Other studies showed serum MMP-1 level was significantly higher in post-TB patients with RPT sequelae compared with patients without any sequelae.¹⁵ All studies showed that the MMP-1 level could be used as a predictor marker in the formation of fibrotic lesions in patients with TB.

Other potential biomarkers in fibrotic lesion induced by TB

Apart from TGF-β and MMP which were the main outcomes assessed in this systematic review, we found several other biomarkers that were also reported in the included studies and had a correlation with the formation or development of fibrosis in patients with TB.

Two studies described changes in tumour necrosis factor (TNF)-α levels. One study reported that TNF-α was correlated positively with effusion shadowing and residual pleural fibrosis. Another study showed that plasma level of TNF-α in minimal lesion of TB was higher than in extensive lesion, but the result was not significant. Dissimilar trend of TNF-α results from the studies might be due to different sampling methods used by Chen et al (pleural fluid) and Astuti et al (plasma). We assumed that TNF-α is also an essential mediator in the pathogenesis of TB and its residual fibrotic lesions.^{14 16}

Besides MMP-1, other MMPs were also assessed by two studies, but the results show no association between MMP-9 and MMP-12 with pulmonary fibrosis.^{10 16} One study showed insulin-like growth factor (IGF)-1 levels in the pulmonary TB groups with minimal and extensive lesions were significantly higher than that of the healthy control group that represents the process of pulmonary fibrosis.¹⁴

Ameglio et al observed a significant increase in cytokines platelet-derived growth factor-BB (PDGF-BB) levels in ELF (epithelial lining fluid) in the second group (the group with increased high resolution computed tomography (HRCT) scores for fibrosis) after 6-month treatment. We assumed that fibrosis might also be related to PDGF-BB values. Furthermore, this study found a significant increase in interleukin (IL)-4 and IL-10 levels in ELF. We suggest that there is a local synthesis of anti-inflammatory cytokines, a possible indication of an incompletely resolved process, possibly leading to the induction of fibrotic mechanisms.¹⁸

Statistical test result (meta-analysis)

Continuous data in the form of the mean TGF-β serum levels and SD from each group (fibrosis lesion and non-fibrosis lesion) were collected from five included studies. Of the five studies included, two studies presented data in the form of the median of TGF-β serum levels, so the data were converted into the form of the mean of TGF-β serum levels using a formula described by Hozo et al.¹⁹ Meanwhile, to estimate the SD of these two studies, the reviewers proceeded the data using also the formula described by Hozo et al.¹⁹

Moreover, one included study by Seiscento et al^{14 17} presented the TGF-β levels data in three groups based on the severity of fibrotic lesion formation, which is a minimal fibrotic lesion and extensive fibrotic lesion compared with the healthy group, respectively. So, the reviewers divided each data of the study into two subgroup studies based on the severity of the fibrotic lesion formation. In total, data in the form of the mean TGF-β serum levels and SD from each group (fibrosis/increased fibrotic lesion and non-fibrosis/stable fibrotic lesion) were collected from five included studies.

Furthermore, the data are applied in a statistical test using the Review Manager V.5.3 to evaluate the MD using the random-effects model. During meta-analysis, we did a subgroup analysis by dividing the data into two study subgroups based on the study’s outcome: parenchymal lung fibrosis and RPT. The results of the parenchymal lung fibrosis subgroup showed that the post-TB group of patients had a higher TGF-β level (6.09) than the control group (1.82), with a 4.27 (95% CI: 0.92 to 7.61) MD and I²=89%. Moreover, the RPT subgroup showed that the post-TB group of patients had a higher mean of TGF-β (0.61) than the control group (0.56), with a 0.05 (95% CI: 0.04 to 0.06) MD and I²=0%. In total, the results showed that the post-TB group of patients had a higher TGF-β level (1.82) than the control group (0.91), with a significant MD of 0.91 (95% CI: 0.27 to 1.54). Overall, the results were as follows: effect Z=2.81 (p=0.005) and heterogeneity of τ²=0.40; Χ²=65.51; df=5 (p<0.00001); I²=92%. These results are presented in the form of a forest plot in figure 3A.

Figure 3.

(A) Forest plot subgroup meta-analysis (fibrotic lesion/increased fibrosis and non-fibrotic lesion/stable fibrosis) and (B) funnel plot subgroup meta-analysis (fibrotic lesion/increased fibrosis and non-fibrotic lesion/stable fibrosis).

Among the five studies included in the meta-analysis, there were differences in the topical location of the outcomes. Three studies assessed fibrotic lesions in the lung parenchyma, while the other two studies assessed fibrotic lesions in the pleural lining. This could be one of the reasons for the high heterogeneity in the results of the meta-analysis.

After conducting a subgroup analysis, the subgroup results for the RPT outcome obtained an I² value of 0%, but for the parenchymal lung fibrosis outcome, the heterogeneity was still high at 89%; this could be influenced by differences in the characteristics used in the three studies. Studies by Ameglio et al and Christine et al conducted an assessment of patients who completed TB treatment, while Astuti et al conducted an assessment of patients who did not receive anti-TB drug yet; the outcome assessed by Ameglio et al was also an increase in fibrotic lesions, in contrast to the other two studies which assessed for the presence or absence of fibrotic lesions.

In this review, we only proceeded with a meta-analysis of the correlation between TGF-β serum levels and the formation/increase of fibrotic lesions induced by TB due to the availability of data that contain at least two studies. For MMP-1, quantitative analysis could not be proceeded due to unavailable data.

Publication bias

The funnel plot (figure 3B) shows an asymmetrical shape of the study distribution, suggesting that there is a risk of publication bias in this meta-analysis, especially in studies belonging to the subgroup with parenchymal lung fibrosis outcomes.

Discussion

Repair or damaged tissue is a natural mechanism after injury.²⁰ However, if this process becomes aberrant, it may lead to the growth of a permanent fibrotic scar at the site of tissue injury. Basically, wound healing has four stages that include a coagulation phase, an inflammatory cell migration phase, a fibroblast proliferation/migration/activation phase, and a tissue remodelling phase. Fibrosis is likely to develop if any stage in the wound healing programme is dysregulated or when the damaging stimulus persists.^{6 21}

Specific host and pathogen factors causing persistent pulmonary/extrapulmonary impairment after TB remain unclear. Therefore, in this systematic review and meta-analysis, we sought to explore factors contributing to the development of tissue fibrosis induced by TB infection. TGF-β is considered the main cytokine associated with fibrogenesis.^{21 22} TGF-β is produced in a latent form and can be activated through reactive oxygen and nitrogen species, plasmin protease pathway, thrombospondin and CD36, low pH, hypoxia and MMPs.²³ Active TGF-β stimulates fibroblast differentiation into myofibroblast and then produces a greater amount of extracellular matrix protein.⁹ The environment of TB granuloma may hold the conditions that prolong the activation of TGF-β, including hypoxia, metalloproteases and nitrogen radicals, meaning that the disease process itself activates TGF-β locally at the site of infection.^24–26 The result of our meta-analysis was parallel to the immunological process, which showed a significant MD of 4.27 (95% CI: 0.92 to 7.61) in the parenchymal lung fibrosis subgroup and a significant MD of 0.05 (95% CI: 0.04 to 0.06) in the RPT subgroup.

MMPs are a family of 25 proteases that can degrade extracellular matrix, leading to tissue remodelling and fibrosis.^{27 28} Dysregulated activities of these MMPs may play a critical part in the altered collagen homeostasis during TB infection.^{10 29} Our systematic study found that MMP-1 is the type that is prone to develop tissue fibrosis after TB infection. Excessive secretion of MMP-1 may lead the fibroblast or other structural cells of the lung to produce TGF-β, degrade collagen type I and III in the lung parenchyma, and is also associated with the degree of residual pleural fibrosis in TBP.^30–32 These are in line with three studies that observed MMP changes in patients with TB infection, which show that MMP-1 is positively correlated with parenchymal lung and pleural fibrosis.^{10 15 16}

Other factors are also presented in this systematic study, such as TNF-α, IGF, IL-4, IL-10, etc.^{14 16 18} But most of the studies that we found through systematic searching led to the association of TGF-β and tissue fibrosis after TB infection. The result of this meta-analysis further strengthens its level of evidence. This proposes that TGF-β might be the critical pathway in modulating the fibrotic process and could be useful as a biomarker of parenchymal or pleural fibrosis. In addition, it also raises the possibility of using targeted therapy such as TGF-β inhibitor as adjunctive therapy to TB antibiotics. Further research is warranted to explore clinically applicable biomarkers as a diagnostic and prognostic value of pulmonary/pleural fibrosis induced by TB infection.

The diversity of outcomes assessed to see the effect of biomarkers on fibrotic lesions limits the conduction of a better meta-analysis. This is also why in this study, we combined the outcome of the formation of new fibrotic lesions and the increase of fibrotic lesions into the same outcome. Furthermore, most studies poorly assessed exposure, and some did not report exposure details, preventing meta-analysis for several outcomes.

During the review process, we strive to reduce bias that can occur at each stage. Screening studies and assessing the risk of bias for each study were also carried out independently by at least two reviewers. There are some limitations to this study: we did not register the protocol for this review, language selection is limited to English and Bahasa, the studies included have a small sample size and the treatment status of patients in all studies is not homogeneous.

Conclusion

Based on this systematic review, it can be concluded that increased TGF-β expression is associated with the formation of fibrotic tissue in post-TB patients. MMP-1, TNF-α, IL-4, IL-8, IL-10, plasminogen activator inhibitor-1 and PDGF levels were also correlated with the presence of fibrotic tissue in post-TB patients.

The results of this study intend to be the basis for application of targeted therapy in TB or post-TB patients to prevent the formation of fibrotic tissue and further impairment.

Data availability statement

Data are available upon reasonable request. The original data presented in the study are included in the article. Further inquiries can be directed to the corresponding author.

Ethics statements

Patient consent for publication

Not required.

Ethics approval

This study does not involve human participants so ethical approval was not required.