Background
Globally, about seven million new and relapse Tuberculosis (TB) cases were diagnosed in 2018 with a treatment success rate of about 85% [
1]. However, in a great proportion of up to 70% of TB patients TB disease does not end with microbiological cure [
2‐
6]. As a consequence of active lung infection with
Mycobacterium tuberculosis (Mtb), many patients undergo parenchymal structure damage, including bronchovascular distortion, excessive fibrosis, bronchiectasis and pleural thickening [
2]. These anatomical changes can result in chronic chest symptoms and a long-term reduction of overall lung function, including ventilatory and respiratory failure. In the clinical assessment of a relevant proportion of post TB patients, abnormalities in spirometry could be found [
7‐
9]. However, not many studies were performed so far in which the lung function of former TB patients was systematically assessed by spirometry. In the majority of these studies a onetime measurement, usually during or directly after TB treatment, was performed. This approach, however, does only provide limited information on the evolution and chronification of lung impairment (LI) during and after pulmonary TB disease. Further, the scientific evidence on the type and severity of respiratory impairment pattern is scarce.
Ravimohan et al. [
2], described several, probably genetically defined, immunological and inflammatory pathways which may result in the destruction of lung tissue and consecutive function impairment. However, the clinical, behavioural or microbiological risk factors for the development of pulmonary sequelae has not been studied in depth yet. In the few existing and relatively small studies partly contradicting evidence was found for an association of post TB lung disease with smoking habits, Human Immunodeficiency Virus (HIV)-status, sex and chest x-ray findings [
3,
5,
10‐
18].
The objective of this study was to prospectively assess the evolution of long-term pulmonary function impairment in spirometry and to get first insights on the associated risk factors in pulmonary TB patients from Maputo, Mozambique. In the absence of a common case definition for chronic lung diseases after TB [
19], in this study we used the term “lung impairment (LI)” for spirometric measurements which were below the lower limit of normality (LLN) threshold.
Discussion
In this prospective study, the majority of pulmonary TB patients had abnormal lung function in spirometry during and after TB treatment. This finding is in line with several other publications and our own meta-analysis (unpublished data) showing that up to 70% of previous TB patients suffer from chronic and clinically relevant lung function impairment [
3,
5,
11,
32]. Currently, there is not much data available on the long term medical but also socio-economic consequences of chronic lung damage after TB [
19]. It is very likely, that many patients with LI presenting with continuous symptoms might be incorrectly classified as recurrent TB cases and referred to unnecessary TB treatment. In the majority, however, LI remains undiagnosed and untreated, possibly resulting in increased mortality compared to the standard population [
33,
34].
Due to the prospective study design, lung function could be measured during and after TB treatment. Like in few other prospective studies [
12,
35], an increase in FVC and FEV1 over time was observed in most subjects that in young adults might be related to muscular and cardiovascular improvement after TB cure. However, this improvement was usually not significant and led to a normalization of lung function in only 15% of subjects with LI at week 8. In general, almost all participants with an abnormal lung function 1 year after TB treatment, already had lung function impairment early during treatment. The data indicate that potential medical interventions aiming to preserve lung function should probably be applied rather early during TB treatment. Currently, there are no guidelines on the diagnosis, follow up and treatment of patients with post TB lung disease [
8]. In order to develop concrete therapeutic strategies, we need a better understanding of the heterogeneity of pulmonary disease after TB and the underlying patho-mechanisms [
19,
36]. Most likely, future preventive treatment options will depend on the causal pathway and/or the prevailing type of LI [
36].
It is worth noting that opposed to the findings in other studies from Africa [
5,
11,
18,
32], almost all our participants had pulmonary restriction (low FVC) after TB and not pulmonary obstruction (low FEV1/FVC-ratio). Ravimohan et al. described different, genetically defined, immunological pathways that are defining lung outcome, including the type of ventilation impairment [
37]. In addition to the genetic constitution of our study population, also the composition of other background risks such as (indoor) air pollution, prevalence of childhood pulmonary infections or risk behaviour could be different to other study populations and thus, could explain the high prevalence of pulmonary restriction. Further, age per se is a risk factor for chronic lung diseases and especially Chronic Obstructive Respiratory Disease (COPD). With a median age of about 30 years, our study cohort was relatively young. Therefore, the prevalence of obstruction could also be lower than in other studies.
Despite the small sample size, we performed univariable regression analysis in order to get an insight into possible risk factors for LI after TB. The results indicate that factors that might be related to a stronger inflammatory response in study participants (BMI < 18.5, CD4 ≥ 200/μl, low haemoglobin) increase the risk for LI. Further, cigarette smoking and alcohol drinking beyond a certain threshold (e.g., ≥10PY, 150 g/80 g alcohol per week) were also increasing the risk for long term LI. Possible causal pathway could either be via direct damage of lung tissue or immune cells or due to poorer health seeking behaviour. Also, a certain level of pre-existing lung damage cannot be excluded in heavy smokers and alcohol drinkers (see below). The finding that light smokers have a lower risk for LI could be explained by the fact that those who can afford to buy cigarettes but do not smoke much may have better access to health care or better physical conditions (related to wealth) that are protective of LI after TB. Reasons for the association of female sex with LI could be differences in immune response, nutrition status, delayed access to medical care, higher exposure to indoor air pollution, etc., compared to men. We did not find any association between HIV-status and LI in our study. Previous research found contradicting evidence for sex, smoking behaviour and also HIV-status as risk factors for TB lung outcome [
3,
5,
10‐
18]. With regards to the role of HIV-coinfection for the development of LI, one reason for the heterogeneity of findings across the few available studies could be the differences in the main type of LI found. While it could be shown that HIV may be associated with obstructive lung disease [
14,
38,
39], its role in the development of restrictive lung impairment, which was the dominant LI type of our study, is not known. Interestingly, an association of increasing CD4 counts with reduced FEV1 was also observed in another study [
37]. However, like this manuscript, the majority of existing publications was based on rather small study cohorts which usually prevented sub-group analyses or assessment of bias such as confounding and effect modification.
In the context of data analysis, we faced several challenges: in the absence of a Mozambican spirometric reference standard, we used South African prediction equations as basis for the analysis of spirometry data (except for FEV1/FVC ratio) as there was an even poorer fit with the reference values for Africans published by the Global Lung Initiative (unpublished data). In order to get an indication to what extent the South African standard is applicable to our TB study population in Maputo we collected lung function data in healthy volunteers (no TB in the past, no chronic or acute lung disease) from the same neighbourhood as the TB patients. The slight shift to lower lung volumes in the healthy volunteers compared to the South African standard (Fig.
4) could either mean that there is a certain level of pre-existing LI in the study population or that the South African standard is not ideal for the population in Maputo. In any case, the data suggest that the observed LI in previous TB patients is largely related to TB, especially in those patients with relatively low z-scores, representing moderate or severe abnormalities in spirometry. In this article, we present the results for FVC in addition to FEV1 in order to avoid the impression that our study participants suffer from airway obstruction. In the majority of them, however, FEV1 was reduced as a consequence of low FVC. In the absence of clinical guidelines on post TB LI, it is unclear whether spirometry (alone) is suitable to identify those patients who suffer most from post TB lung disease or would benefit most from pulmonary follow up and treatment. It might be that additional investigations, e.g., chest x-ray, cardiac investigation, exercise tests or symptoms screens, are needed to identify patients with manifest comorbidities of LI in order to prioritize supportive treatment, given that the underlying pulmonary damage is currently not curable.
A recent publication showed that reduced, but still normal, values for FEV1 of above -2SD of predicted were associated with significantly increased mortality [
9]. This supports the strategy to use spirometry to identify patients with LI after TB who are possibly at highest risks for increased morbidity and mortality. However, there is a need of studies in larger cohorts to firstly get more robust data on the prevalence, type and severity of LI in previous TB patients. Secondly, the role of risk factors and the underlying causal pathways need to be addressed in those studies in order to develop strategies for treatment and prevention of LI. Finally, as the basis for future research, guidelines on measurement and classification of LI as well as agreed strategies on data analysis are urgently needed.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.