Persistent chronic respiratory symptoms despite TB cure is poorly correlated with lung function

SUMMARY BACKGROUND: Persistent respiratory symptoms and lung function deficits are common after patients with TB. We aimed to define the burden of post-TB lung disease (PTLD) and assess associations between symptoms and impairment in two high TB incidence communities. METHODS: This was a cross-sectional survey of adults in Cape Town, South Africa who completed TB treatment 1–5 years previously. Questionnaires, spirometry and 6-minute walking distance (6MWD) were used to assess relationships between outcome measures and associated factors. RESULTS: Of the 145 participants recruited (mean age: 42 years, range: 18–75; 55 [38%] women), 55 (38%) had airflow obstruction and 84 (58%) had low forced vital capacity (FVC); the mean 6MWD was 463 m (range: 240–723). Respiratory symptoms were common: chronic cough (n = 27, 19%), wheeze (n = 61, 42%) and dyspnoea (modified MRC dyspnoea score 3 or 4: n = 36, 25%). There was poor correlation between FVC or obstruction and 6MWD. Only low body mass index showed consistent association with outcomes on multivariable analyses. Only 19 (13%) participants had a diagnosis of respiratory disease, and 16 (11%) currently received inhalers. CONCLUSION: There was substantial burden of symptoms and physiological impairment in this “cured” population, but poor correlation between objective outcome measures, highlighting deficits in our understanding of PTLD.


S U M M A R Y B A C K G R O U N D :
Persistent respiratory symptoms and lung function deficits are common after patients with TB. We aimed to define the burden of post-TB lung disease (PTLD) and assess associations between symptoms and impairment in two high TB incidence communities. M E T H O D S : This was a cross-sectional survey of adults in Cape Town, South Africa who completed TB treatment 1-5 years previously. Questionnaires, spirometry and 6-minute walking distance (6MWD) were used to assess relationships between outcome measures and associated factors. R E S U LT S : Of the 145 participants recruited (mean age: 42 years, range: 18-75; 55 [38%] women), 55 (38%) had airflow obstruction and 84 (58%) had low forced vital capacity (FVC); the mean 6MWD was 463 m (range: 240-723). Respiratory symptoms were common: chronic cough (n ¼ 27, 19%), wheeze (n ¼ 61, 42%) and dyspnoea (modified MRC dyspnoea score 3 or 4: n ¼ 36, 25%). There was poor correlation between FVC or obstruction and 6MWD. Only low body mass index showed consistent association with outcomes on multivariable analyses. Only 19 (13%) participants had a diagnosis of respiratory disease, and 16 (11%) currently received inhalers. C O N C L U S I O N : There was substantial burden of symptoms and physiological impairment in this ''cured'' population, but poor correlation between objective outcome measures, highlighting deficits in our understanding of PTLD. K E Y W O R D S : post-TB; spirometry; restriction; physiology; obstruction TB is one of the most common respiratory infections, and South Africa has one of the highest TB incidence rates in the world-around 615 per 100,000 population. 1,2 There is a growing understanding that the effects of TB do not end with the completion of anti-TB chemotherapy. 3,4 Residual, post-TB lung disease (PTLD) includes fibrosis, lung volume loss, cavitation, pleural disease, gas trapping and bronchiectasis, and can manifest with varying severity. 5 In the multinational burden of obstructive lung disease study, people who completed TB treatment had a two-fold higher risk of developing both obstructive and restrictive lung disease compared to the general population. 6 There is currently little evidence about how PTLD phenotypes progress or relate to symptoms. A prospective study of 405 patients in Malawi found 44% with bronchiectasis and 9% with one destroyed lobe on high-resolution computed tomography scanning. 7 After 1 year, almost a third reported residual respiratory symptoms, one in five had a decline in forced expiratory volume in 1 sec (FEV 1 ) of 100 ml, and 16% reported at least one respiratory exacerbation. In a cross-sectional study in suburban Cape Town, South Africa, approximately half of participants who had completed TB treatment in the previous 5 years reported ongoing respiratory symptoms. 8 Better understanding of the interplay between symptoms, physiology and physical performance is needed, as well as predictive models of who may develop symptoms or require treatment for those at highest risk of PTLD. This is most relevant in lowresource, high-burden settings such as sub-Saharan Africa, where healthcare resources and diagnostics are limited.
In the present study, we evaluated the burden of symptoms, spirometric and physiological impairment in a community-based population who had successfully completed TB treatment, and investigated the relationship between symptoms and lung function.

Setting, participants and sampling
We conducted a cross-sectional study of adults who had successfully completed TB treatment more than 1 year but less than 5 years prior to study enrolment. Potential participants who had completed an episode of TB treatment between 2013 and 2017 were identified through the Ikhwezi and Ravensmead Clinic registers in suburban Cape Town, as described elsewhere, 8 and invited to participate. We used electronic TB registers from both primary healthcare facilities to obtain a random sample of individuals who were recorded for treatment during this time frame, and with a documented standard treatment outcome of either ''cure'' or ''treatment completed''. Study recruitment and follow-up occurred between October and December 2018 (Ikhwezi) and March and November 2019 (Ravensmead). Individuals were contacted during study visits at the home address recorded in the treatment register. Individuals were eligible for this study if they were at least 18 years old, had at least one episode of treated TB (as per list), and provided informed written consent. Exclusion criteria were pregnancy, currently receiving TB treatment, contra-indications to performing spirometry and acute illness (likely to introduce measurement error).

Outcomes
Primary outcome was the prevalence of symptoms, physiological impairment and spirometric abnormalities. Secondary outcomes included correlation between the primary outcomes and a model for spirometric and 6-min walking distance (6MWD) results.

Data collection
Participants completed the IMPALA (International Multidisciplinary Programme to Address Lung Health and TB in Africa) symptoms, smoking, environmental exposures, life exposures, previous episodes of TB and nutrition questionnaires. 9 Data were collected electronically using KoBo Toolbox (Harvard Humanitarian Initiative, Cambridge, MA, USA). 10 Breathlessness was rated using the modified Medical Research Council (mMRC) scale. 11 Demographic data were captured and managed using REDCap (Vanderbilt University, Nashville, TN, USA) electronic data capture tools hosted at Stellenbosch University, Tygerberg, South Africa. 12 Spirometry and the 6-min walking test were conducted at the study locations. Spirometry was performed before and after inhaled salbutamol 400 mcg via a metereddose inhaler using EasyOne w (ndd Medical Technologies, Zurich, Switzerland) device according to the American Thoracic Society/European Respiratory Society guidelines. Post-bronchodilator spirometry data are presented here.Normal spirometric ranges were defined by GLI (Global Lung Function Initiative) 2012, and abnormal results were defined those less than the lower limit of normal (LLN) (Zscore ,-1.64). 13 Statistical methods Data were presented as means, standard deviations, medians, ranges or percentages based on the type and distribution of data. Pearson's and Spearman's correlation were used for the correlation analysis. Student's t-test, analysis of variance (ANOVA), v 2 and Kruskal-Wallis tests were used for univariable analysis. Linear or logistic regression analysis was used for the multivariable analysis. Outcome variables were forced vital capacity (FVC) % predicted, 6-min walking distance (6MWD) and forced expiratory volume in 1 s (FEV 1 )/FVC , LLN. Variables were included in the model based on exposures of interest and possible confounders. Variables were retained to create the smallest Akaike Information Criterion (AIC) in the final backwards multivariable regression analysis. Significance was defined as P , 0.05, 95% confidence intervals (CIs) have been provided where appropriate. Data were analysed using R (R Computing, Vienna, Austria). 14

Ethical considerations
Ethical approval was obtained from Stellenbosch University, Tygerberg, South Africa; (N18/05/056) and the Liverpool School of Tropical Medicine, Liverpool, UK . Permission to conduct the study was obtained from the Western Cape Department of Health, Cape Town, South Africa; all participants provided written informed consent prior to participating in the study.

Descriptive data
A total of 432 participants from the registers were contacted, 39 had died (9.0%) and 277 were invited to take part ( Figure 1). We recruited 145 study participants (Figure 1), of which 55 (38%) were women. The mean age was 42 years (range 18-75, Table 1). Over half of the participants left school at or before age 15 (66%). Of the 145 participants, 24 were living with HIV, 26 had hypertension and 19 diabetes; 19 had a respiratory disease diagnosis and 16 were prescribed inhalers. Cough, phlegm production, and wheeze were present in respectively 19%, 17% and 42% of participants; 55 (38%) had been treated more than once for TB, with a range of up to 6 times. Most participants reported having their first episode of TB as adults. Seven (5%) did not complete treatment on at least one occasion and 10 (7%) had been treated for multi-or extensively drug-resistant TB (MDR-and XDR-TB) at least once ( Figure 1).
There was a high prevalence but low intensity of smoking: 100 (69%) reported current or previous smoking, but participants only accumulated a median of 6.7 pack-years; 51 (35%) participants reported smoking substances other than cigarettes, the most common being cannabis and waterpipe.

Uni-and multivariable analysis
The main outcome variables were FVC (% predicted) to measure restriction, FEV 1 /FVC ratio as per the LLN (to measure obstruction) and 6MWD (as a physiological measure). For the univariable analysis, all participants with data for that variable were included. All those with fully completed data sets across all variables (n ¼ 107) were included in the multivariable analysis.
For the 84 participants with low FVC (Z-score ,1.64 of normal), the mean 6MWD was 465 m compared to 458 m of patients with normal FVC (P ¼ 0.69, 95% CI for difference of the mean: -43 to 28). 12 The mean 6MWD in patients with airflow obstruction was 450 m, compared to 469 m in patients without airflow obstruction (P ¼ 0.31, 95% CI for difference of mean -19 to 56).
FVC (% predicted), BMI, alcohol consumption and chronic bronchitis symptoms were statistically significantly associated in univariable analysis (Table 2). BMI, alcohol consumption and symptoms of chronic bronchitis were present in the final multivariable, adjusted regression model (R 2 0.14; AIC 594).
6MWD, age, BMI, highest level of education, HIV  status, age at first TB, FEV 1 (in L) and FVC (in L) were statistically significant in the univariable analysis ( Table 3). The multivariable linear regression model included age, BMI, alcohol consumption, highest level of education, HIV status, chronic cough, chronic bronchitis, smoking status and smoking substances other than cigarettes (R 2 0.37; AIC 920). Spirometric obstruction was statistically significantly influenced only by BMI in the uni-and multivariable logistic regression analysis (

DISCUSSION
This cohort of 145 participants who had successfully completed TB treatment between 1 and 5 years previously had high burden of symptoms, short 6MWD and spirometric obstruction. We were unable to find strong associations between risk factors and symptoms, spirometric or physiological impairment, and there was poor correlation between outcome measures themselves (spirometric and functional).
Despite the high burden of respiratory abnormalities, only 19 participants had a formal diagnosis of respiratory disease and only 16 received inhalers at enrolment. Almost a third reported significant breathlessness (mMRC 2), more than 40% attacks of wheezing, and about 1 in 5 chronic cough and phlegm. It is surprising to note that 38% had chronic airflow obstruction and the mean 6MWD was 463 m, as this was a young community population. A hundred participants were current or ex-smokers; however, smoking intensity was low (median consumption: 6.7 pack years), implying cigarette smoke a less likely cause for symptoms. Smoking other substances and exposures to potentially damaging inhalants were common; however, neither these, nor cigarette smoking were associated with objective outcome measures. Seventeen participants reported at least three previous episodes of TB and most first contracted TB as middle-aged adults, but the number of TB episodes, drug resistance and age of first episode were not associated with outcomes, as has been suggested elsewhere. 15 A very high prevalence of spirometric obstruction has been described in previous multi-centre, multinational studies. 16 A similar study in rural South Africa found comparatively more preserved lung function and fewer symptoms but much reduced 6MWD (mean: 294 m) compared to our population. 17 A Malawian cohort demonstrated lower symptom burden than our population, with 12% having symptom-related impairment in their ability to work 1 year after treatment completion, with a mean 6MWD of 611 m. 7 Whether there is an effect of urbanisation on PTLD needs to be established.
The associations between various commonly reported objective measures of respiratory function were poor. This adds complexity to the syndrome of PTLD, which is known for its heterogeneity, and amplifies the difficulty in the epidemiological measurement and modelling of disability in PTLD. The persistent symptomatology observed and the welldescribed excess mortality (9% of our sample population had died) cannot be ignored; 18,19 however, our knowledge of the interplay between spirometry, symptoms and physiology in PTLD is incomplete. Studies of PTLD in Mexico and the United States described a negative correlation between spirometry abnormalities and symptoms. 20,21 These correlations have also been investigated in other respiratory conditions, such as asthma, with inconsistent results. [22][23][24][25][26][27] In chronic obstructive pul-monary disease (COPD), dyspnoea has been established as a marker for mortality, with mMRC scores of 3 or 4 associated with a relative risk of death of 8.3 and 61 respectively. 28 These conflicting data indicate that for now we should continue investigating PTLD patients multi-dimensionally until better outcome data are available.
The models derived suggest some factors may be associated with the spirometry, physiological and symptomatic impairments described. Chronic cough or chronic bronchitis symptoms appeared in final models for 6MWD and FVC but not spirometric obstruction, and could be explored as one potential screening tool for impairment. Low BMI and alcohol use were frequently associated, but the reason for the association is unclear. Both could be confounders, associated with a common unknown independent variable, for example, poor nutrition or poor social circumstances, or both could be effect modifiers in the pathway to PTLD, or potentially even a consequence (rather than a cause) of chronic lung impairment. The usual cautions need to be employed when interpreting cross-sectional associations and attributing either effect modification or prediction to the outcome. Although participants were assessed in detail regarding their symptoms, spirometry and functional capacity, we did not consider their radiological findings. Chest X-ray is likely to be a less sensitive measure of post-TB impairment than spirometry, as it cannot identifypredominantly small-airway PTLD. 29 The baseline data reflect the population sampledlimited education, almost 20% HIV prevalence, with an experience of food shortages and a low-normal BMI, so the high burden of observed PTLD could have particularly significant economic, morbidity and mortality implications in this vulnerable population.
Clearly there is a need for a better understanding of the pathophysiology of PTLD and the relationship between objective physiological measures, symptoms and meaningful outcomes such as mortality, exacerbations and morbidity (including psychosocial impacts). It is likely that a multimodal approach will be needed, analogous to that used for other respiratory diseases, for example, the BODE (Body-mass index, airflow Obstruction, Dyspnea, and Exercise) index in COPD. 30,31 Developing such a tool for PTLD would be helpful to risk stratify patients in limited healthcare settings such as sub-Saharan Africa. Although self-selection through participation and the sample size may have impacted our study findings, unlike previous studies of PTLD, patients were not actively healthcare-seeking and were selected at random from a TB register between 1 and 5 years after treatment completion. These high rates of residual abnormalities after successful treatment, combined with low rates of chronic respiratory diagnosis or treatment, again highlights the need to look beyond usual outcome measures in TB trials and control programmes. Given the significant symptoms and functional impairment in this young population, urgent prognostic assessments for PTLD are needed, and pulmonary rehabilitation programmes may play a significant role in its treatment. [32][33][34][35] More work is required on the longterm evolution of PTLD to identify individuals at risk of chronic PTLD and to investigate effective prevention and treatment strategies.

CONCLUSION
This population exhibited a high burden of selfreported symptoms, airflow obstruction and functional impairment 1-5 years after TB treatment completion, yet only a minority had received a diagnosis of chronic disease or were receiving treatment. There was poor correlation between physiology, functional capacity and symptoms, with only low BMI most consistently associated with outcomes. This highlights the need for a better understanding of disease pathophysiology and multimodal assessments to determine meaningful objective measurements in PTLD. More work is needed on the prognosis of and interventions for PTLD, considering that more than 58 million people have survived TB in the last two decades alone. Meme