Cough Frequency During Treatment Associated With Baseline Cavitary Volume and Proximity to the Airway in Pulmonary TB

Background Cough frequency, and its duration, is a biomarker that can be used in low-resource settings without the need of laboratory culture and has been associated with transmission and treatment response. Radiologic characteristics associated with increased cough frequency may be important in understanding transmission. The relationship between cough frequency and cavitary lung disease has not been studied. Methods We analyzed data in 41 adults who were HIV negative and had culture-confirmed, drug-susceptible pulmonary TB throughout treatment. Cough recordings were based on the Cayetano Cough Monitor, and sputum samples were evaluated using microscopic observation drug susceptibility broth culture; among culture-positive samples, bacillary burden was assessed by means of time to positivity. CT scans were analyzed by a US-board-certified radiologist and a computer-automated algorithm. The algorithm evaluated cavity volume and cavitary proximity to the airway. CT scans were obtained within 1 month of treatment initiation. We compared small cavities (≤ 7 mL) and large cavities (> 7 mL) and cavities located closer to (≤ 10 mm) and farther from (> 10 mm) the airway to cough frequency and cough cessation until treatment day 60. Results Cough frequency during treatment was twofold higher in participants with large cavity volumes (rate ratio [RR], 1.98; P = .01) and cavities located closer to the airway (RR, 2.44; P = .001). Comparably, cough ceased three times faster in participants with smaller cavities (adjusted hazard ratio [HR], 2.89; P = .06) and those farther from the airway (adjusted HR, 3.61;, P = .02). Similar results were found for bacillary burden and culture conversion during treatment. Conclusions Cough frequency during treatment is greater and lasts longer in patients with larger cavities, especially those closer to the airway.

During 2016, there were estimated to be 10.4 million new TB cases worldwide, causing 1.7 million deaths. 1 TB is transmitted mostly through coughing, [2][3][4] which has been associated with increased bacillary burden. 5 Cough can be assessed easily throughout treatment, but its relationship with cavitary lung disease, to our knowledge, has not been studied. 6 Identifying radiologic characteristics associated with increased cough frequency is important in understanding transmission and evaluating treatment response. 2,5,7,8 Infectivity of TB is different for each individual, with some infecting more than others, so transmission in TB is considered heterogeneous. [9][10][11][12][13][14][15][16][17][18][19] In this longitudinal study, we sought to investigate whether there is an association between cough frequency, and its duration, with radiologic characteristics, such as cavitary volume and cavitary proximity to the airway. We also evaluated whether bacillary burden and culture conversion were associated with these radiologic characteristics.

Study Design
This was a prospective cohort study conducted in two tertiary hospitals in Lima, Peru. The detailed study protocol has been published previously. 20 Study participants were at least 18 years old, and their pulmonary TB diagnosis and drug sensitivity were assessed by means of microscopic observation drug susceptibility (MODS) broth culture assay. [20][21][22][23][24][25][26] In this report, we restricted analyses to participants who had a strain that was susceptible to isoniazid and rifampicin and who did not have HIV (Fig 1) because immune status and drugresistant strains affect radiologic manifestation. 27,28 The Cayetano Cough Monitor (CayeCoM) was used to record participants' data daily during the first 14 days of treatment and at days 21, 30, and 60. Recordings started at 9:00 AM. 20 A cough episode included all independent cough events that occurred without a 2-second pause, no cough was a cough frequency # 0.7 cough events per hour, and cough cessation was two consecutive recordings with no cough. 5 Sputum was obtained on days 0, 3, 7, 14, 21, and 60 of treatment. Bacillary burden was assessed through time to positivity (TTP) of cultures 5,29,30 in all MODS culture-positive sputum samples, and culture conversion was defined as the first negative culture with no subsequent positive cultures. 5 The study data for cough frequency and bacillary burden has been published. 5,31 A baseline chest CT scan was obtained within 31 days of treatment initiation in all participants enrolled in the study who consented, similar to methods used in a previous TB study in participants who were drug susceptible and HIV negative. 28

Radiologic Imaging
Scans were obtained (Aquilion 64, Toshiba) and analyzed by using a free Digital Imaging and Communications in Medicine viewer. Our computer-automated algorithm detected and measured the volume of the cavitary lesion and its proximity to the airway.
A previous algorithm used in small animals 32 has been improved in performance for human CT scans by using a more accurate lung segmentation algorithm. 33,34 The validation methods of this higherresolution algorithm are described in the supplementary methods section of e-Appendix 1. In the case of multiple cavities, we used the cumulative volume of all cavities for analyses.
Fuzzy connectedness methods 35 were used to segment the airway in high-resolution CT scans (< 4-mm section thickness). The proximity of the cavitary lesion to the airway was determined using Euclidean distance transform. [36][37][38] If multiple cavities were present, then the cavity closest to the bronchi was used to determine proximity to the airway. To evaluate other radiologic features, a US board-certified radiologist (P. C.) evaluated each scan to indicate presence or absence of consolidation, cavitation, pneumatocele, atelectasis, fibrosis, bronchiectasis, pericardial effusion, pleural effusion, lymphadenopathy, miliary spread, and pneumothorax.

Statistical Analysis
Data analysis was performed using software (Stata/SE 14.0, Stata Corp). P values # .05 were considered statistically significant, and data are shown following recommended numeric presentation. Percentages presented as integers, mean difference (MD) is shown to one decimal place, rate ratio (RR) and hazard ratio (HR) are shown based on the rule of four. 39,40 Sciences, National Institutes of Health, Bethesda, MD; the Section of Infectious Diseases & Immunity and Wellcome Trust Imperial College Centre for Global Health Research (Drs Friedland and Evans), Imperial College London, and the TB Centre Cavitary disease was evaluated based on its volume and proximity to the airway according to data from the computer-automated algorithm. We chose 7 mL as the cutoff between a small and a large cavity and 10 mm as the cutoff between a cavity positioned closer to and farther from the airway to the closest edge of the cavity (inner wall). Cutoff analyses showed significance at these values (P < .001 for both) (e- Fig 1). In addition, the presence of bronchiectasis, atelectasis, pleural effusion, and lymphadenopathy were assessed by the radiologist. Other features were too skewed to be compared.
We evaluated baseline cavitary lung disease (cavitary volume and proximity from the cavitary lesion to the airway) with pretreatment cough frequency (negative binomial model) and pretreatment TTP (linear regression), adjusting for age and sex, respectively. We also assessed the association between baseline cavitary lung disease and longitudinal cough frequency results during treatment by using a negative binomial model adjusting for age, culture positivity or negativity, sex, treatment day, and treatment day squared, with a random intercept for study participant; covariates were chosen based on previous analyses. 5 Baseline cavitary lung disease and longitudinal TTP during treatment were assessed using a linear regression model adjusting for age, cough rate, sex, treatment day, and treatment day squared. A Cox proportional hazards model, unadjusted and adjusted to age and sex, was used to evaluate baseline cavitary lung disease and its effect on cough cessation and culture conversion.
In addition, we used the same analyses to evaluate the presence of baseline atelectasis, bronchiectasis, pleural effusion, and lymphadenopathy on cough frequency before and during treatment, TTP, cough cessation, and culture conversion. For all analyses described, given the small sample size and the exploratory nature of these analyses, no correction for multiple comparisons was made.

Ethics
This study was conducted in accordance with the Declaration of Helsinki. 41 This study also was conducted with institutional review board approval by each participating hospital; Universidad Peruana Cayetano Heredia (SIDISI:57183); Asociación Benefica PRISMA in Lima, Peru; and Johns Hopkins University in Baltimore, Maryland (IRB00001676).

Results
There were 64 participants with available CT scans, but three scans were of poor image quality (recorded as JPEG format instead of Digital Imaging and Communications in Medicine), and two were incomplete (not enough cross-sectional images) and therefore could not be read. After excluding participants with drug-resistant strains, HIV-positive status, or CT scanning performed after 1 month of treatment, 41 participants were available for analysis. The 41 participants in the study group had a total of 695 recordings, but 37% had to be excluded for technical reasons (e- Table 1). After exclusion, there were 18 participants with pretreatment cough recordings. The median length of recordings was 21 hours. Sixty-eight percent of participants were male, with a median age at enrollment of 30 years (interquartile range, 23-50 years). CT scans were obtained a median of 13 days after treatment initiation (interquartile range, 7-21 days). Demographic and radiologic characteristics of participants are shown in Table 1. According to the radiologist, the most common findings were cavitary lesions (98%), consolidations (93%), bronchiectasis (68%), atelectasis (29%), lymphadenopathy (20%), and pleural effusion (17%). Only three participants had pneumatocele or pericardial effusion reported, only one participant had fibrosis reported, and miliary spread and pneumothorax were not reported.
CT scans with adequate quality were used (n ¼ 41) (Fig 1). In a sensitivity-specificity analysis of cavitation detection, the computer-automated algorithm had a sensitivity of 95% and a specificity of 100% (e- Table 2). The validation of the higher-resolution computerautomated algorithm is shown in the supplementary results (e-Appendix 1, e- Fig 2).

Cough Frequency Associations
Baseline cavitary volume and proximity to the airway were not associated with pretreatment cough frequency (e -Tables 3-5). However, results of our multivariable analyses showed that cough frequency during treatment in participants with larger cavities was nearly double that of participants with smaller cavities RR, 1.98; 95% CI, 1.17-3.35; P ¼ .01) ( Table 2). Similarly, participants with cavity lesions located farther from the airway had significantly less cough frequency during treatment than did patients with closer proximities (RR, 0.41; 95% CI, 0.248-0.68; P ¼ .001) ( Table 3). When we analyzed both cavity volume and distance to the airway, combined, we found that only distance to the airway was significant during treatment (RR, 0.376; 95% CI, 0.196-0.72; P ¼ .003) ( Table 4). Older age had a strong trend for more cough frequency during treatment in our models (Tables 2-4).

Cough Cessation Assessment
Cough cessation tended to be three times faster among participants with smaller cavities than among those with larger cavities, but this finding was not statistically significant (adjusted HR, 2.89; 95% CI, 0.95-8.8; P ¼ .06). The probabilities of cough cessation by day 60, were 69% for small cavities and 31% for large cavities (Fig 2). Furthermore, the hazard for cough cessation was significantly three times higher among participants with cavities located > 10 mm from the airway than among those with cavities located # 10 mm from the airway (adjusted HR, 3.61; 95% CI, 1.26-10.4; P ¼ .02). Cough frequency was used as an outcome in a negative binomial regression to test for risk factors that would increase cough frequency during treatment.
In the partially adjusted model, we adjusted for treatment day and treatment day squared. In the fully adjusted model, we adjusted for treatment day, treatment day squared, MODS culture positivity, age, and sex. The volume of the cavity in milliliters was calculated through a computer-automated algorithm that analyzed CT scans on the basis of the voxel size of the cavitary lesion. Participants with no cavities were included in this analysis as having 0-mL volume. MODS ¼ microscopic observation drug susceptibility; Obs ¼ observations; RR ¼ rate ratio. Cough frequency was used as an outcome in a negative binomial regression to test for risk factors that would increase cough frequency during treatment.

chestjournal.org
By day 60, the probabilities for cough cessation were 37% for closer distances and 75% for farther distances (Fig 3). The presence of other radiologic features was not associated with cough cessation.

Culture Conversion Assessment
Culture conversion hazard tended to be two times higher among patients with smaller cavities than among those with larger cavities, but this finding was not statistically significant (adjusted HR, 2.07; 95% CI, 0.90-4.7; P ¼ .09). By day 60, the probabilities of culture conversion are 100% for small cavities and 73% for large cavities (Fig 2). Similarly, those with lesions located farther from the airway tended to have a higher culture conversion hazard but this was not statistically significant (adjusted HR, 2.00; 95% CI, 0.95-4.2; P ¼ .07). Culture conversion probabilities, by day 60, were 83% for closer distances and 100% for farther distances (Fig 3). The presence of other radiologic features was not associated with culture conversion.

Discussion
Despite the importance of cough in TB transmission, there is a lack of research on this topic, 2,7 and a recent clinical guideline demonstrated that cough duration and cavitary lung disease have not been studied. 6 An increase in cough frequency, as well as delayed cough cessation, heightens the theoretical chances for that patient to expel TB aerosols into the air, 42,43 increasing the risk of transmission. 44,45 Our study demonstrated that higher cough frequency during treatment, as well as delayed time to cough cessation, are associated with larger cavitary volume, especially cavities closer to the airway.
environments. 48 A diagnostic algorithm could determine quickly the most likely contagious patients, as well as identify potential patients who might not respond well to treatment given their increased disease burden. 46 However, the most important factor to diminish transmission is effective treatment, 5,49,50 and other factors (cough strength, sputum viscosity, cough hygiene, social interaction) also would need to be evaluated for this algorithm to build on current scores. 51 We observed that a larger cavity volume and a closer proximity to the airway was associated with higher cough frequency during treatment, higher bacillary burden before and during treatment, delayed cough cessation, and culture conversion. Previous studies support the association between larger cavitary volume and higher bacillary burden before treatment, 52 as well as a relationship between closer proximity to the airway and higher bacillary burden before treatment. 53 When evaluating both volume and proximity, combined, we found that of these two, proximity to the airway seems to play a larger role for both cough frequency during treatment and bacillary burden before and during treatment. The closer the cavity is to the airway, the more inflammation causes increased cough frequency during treatment, and the better oxygen access is provide an optimum microenvironment for Mycobacterium tuberculosis growth. 54,55 Previous studies show that M tuberculosis grows better within the macrophages of the luminal surface of the cavitary lesion because of better oxygen access, coupled with a lack of T lymphocytes, which diminishes the interactions between T cells and macrophages that clear mycobacteria. 54,55 Patients with more severe infection might have bronchial obstruction that can lead to a lung collapse (atelectasis), which in turn can act as a one-way valve that ultimately increases cough frequency. 56,57 However, pleural effusion is a hypersensitivity reaction that could cause a systemic response resulting in cough, independent of bacillary burden. [58][59][60] Our study also supports the suggested relationship between radiologic extent of the disease, based on CXR, 61 and cough frequency. 62 A limitation is that 18 participants had pretreatment recordings, and nearly one-third of recordings had to be excluded due to technical limitations. We did not identify bias when comparing participants with at least 10 excluded recordings with those with fewer than 10 excluded recordings. Chest radiography (CXR) is usually the imaging modality of choice in TB control programs, but CT scans are more sensitive for detecting pleural and parenchymal lesions. [63][64][65][66][67][68] Nearly all participants had cavities, so we could not evaluate or compare the cough frequency between patients with a cavity and patients without a cavity. The presence of a cavity has not been associated with cough-generated aerosols. 49 Given that CXR are obtained in a two-dimensional fashion, it would not have enabled us to evaluate threedimensional volume and proximity to the airway. A strength of our investigation is that our cough measurements with CayeCoM were validated previously, [69][70][71] as was the algorithm used to evaluate cavity volume and proximity to the airway. 32,38 Our sample size was similar to those in other CT scanning studies in TB, and the small delay in CT scanning after starting treatment is unlikely to affect results because major changes in cavity structure are uncommon in the first month of treatment. 28,52,53,64 Conclusions To our knowledge, this is the first report regarding an association between cough frequency during treatment and cavitary lung disease. Our study demonstrates an association between cough frequency during treatment, and its duration, with cavitary volume and cavitary proximity to the airway. Younger patients, with small cavitary lesions, especially lesions farther from the airway, may present with minimal cough and sputum samples with low bacillary burden (ie, be smear negative). These patients likely would have cough symptoms later than those with cavities close to the bronchi and, if they are not cultured, may be missed by smear alone. Similarly, if a patient is has a large cavity diagnosed, especially close to the airway, this patient has an increased risk for coughing more during treatment and should be monitored closely for the possibility of expelling more M tuberculosis to the environment. chestjournal.org