Persistent Mycobacterium tuberculosis bioaerosol release in a tuberculosis-endemic setting

Pioneering studies linking symptomatic disease and cough-mediated release of Mycobacterium tuberculosis (Mtb) established the infectious origin of tuberculosis (TB), simultaneously informing the pervasive notion that pathology is a prerequisite for Mtb transmission. Our prior work has challenged this assumption: by sampling TB clinic attendees, we detected equivalent release of Mtb-containing bioaerosols by confirmed TB patients and individuals not receiving a TB diagnosis, and we demonstrated a time-dependent reduction in Mtb bioaerosol positivity during six-months’ follow-up, irrespective of anti-TB chemotherapy. Now, by extending bioaerosol sampling to a randomly selected community cohort, we show that Mtb release is common in a TB-endemic setting: of 89 participants, 79.8% (71/89) produced Mtb bioaerosols independently of QuantiFERON-TB Gold status, a standard test for Mtb infection; moreover, during two-months’ longitudinal sampling, only 2% (1/50) were serially Mtb bioaerosol negative. These results necessitate a reframing of the prevailing paradigm of Mtb transmission and infection, and may explain the current inability to elucidate Mtb transmission networks in TB-endemic regions.


Introduction
Infection with Mycobacterium tuberculosis (Mtb) encompasses a spectrum of outcomes (Drain et al., 2018, Barry et al., 2009, Lin and Flynn, 2018, Coussens et al., 2024).At its mildest, the immune system clears or contains infecting Mtb bacilli with minimal discernible impact on the host (Richards et al., 2023).At its most severe, pulmonary Mtb infection results in advanced disease that is associated with a 40-70% fatality rate in the absence of effective treatment (Hermans et al., 2015).Despite the rarity of progression to active disease at an individual level and the availability of effective chemotherapy (Richards et al., 2023, Horton et al., 2023), tuberculosis (TB) remains a leading global cause of mortality (WHO, 2023), with notiSication rates surpassing those reported in the early 1900s in certain settings (Hermans et al., 2015).
In endemic regions, most incident TB is thought to arise from the recent (<2 years) transmission of Mtb (Shah et al., 2017, Dowdy andBehr, 2022).However, with only 1-30% of new Mtb infections traceable to known TB cases (Crampin et al., 2006, Glynn et al., 2015, Middelkoop et al., 2015, Verver et al., 2004), it is likely that numerous undetected sources of Mtb transmission exist within these communities.Consistent with this possibility, there is growing interest in the notion that early TB states -encompassing Mtb infection, subclinical, and clinical TB (Coussens et al., 2024) -might provide a previously overlooked reservoir of transmission (Kendall et al., 2021, Nguyen et al., 2023).Establishing the propensity for Mtb release -often equated with "infectiousness" -during early TB is therefore essential, but very difSicult to accomplish (Dinkele et al., 2024).
Sputum diagnosis is generally considered the gold standard for assessing infectiousness among TB patients (Pai et al., 2023).Through this lens, data from extensive community TB screening efforts utilizing sputum-GeneXpert -which have identiSied large numbers of asymptomatic cases (Moyo et al., 2022) -support the notion of unrecognized transmitters in high-burden communities.And, while these results accommodate the potential for Mtb transmission to occur prior to the development of recognizable TB symptoms (Nguyen et al., 2023), they nevertheless rely on sputum production (itself a symptom of disease) for the detection of Mtb.This dependency on sputum production imposes a signiSicant constraint on attempts to detect individuals with early TB, and is exacerbated by the difSiculties encountered in obtaining useable samples from many individuals (Pai et al., 2023).
Bioaerosol sampling offers a non-invasive method to collect peripheral lung Sluid and/or particulate matter independent of the speciSic disease state (Johnson and Morawska, 2009).
Moreover, the detection of Mtb in bioaerosols signiSies the release of bacilli by an individual, tentatively offering a concurrent metric of transmission risk (Williams et al., 2023, Jones-López et . al., 2013).Despite these advantages, the utilization of bioaerosol sampling has predominantly been restricted to individuals with bacteriologically conSirmed TB.This owes mainly to the challenges inherent in handling extremely paucibacillary samples as well as the assumed importance of pathology for Mtb release -in turn reinforcing the primacy of sputum bacterial burden as key predictor of infectiousness.Insights gained from the COVID-19 pandemic, however, established the plausibility of asymptomatic transmission (Shaikh et al., 2023), with analogous recent results suggesting the cough-independent aerosolization of Mtb (Dinkele et al., 2022, Williams et al., 2020).It seems possible, therefore, that individuals might progress from Mtb infected (often inferred via interferon gamma release assay; IGRA (Cohen et al., 2019)) to infectious before producing sputum.
Since 2013, we have studied the spontaneous generation of Mtb-containing bioaerosols using the Respiratory Aerosol Sampling Chamber (RASC), a highly sensitive, adaptable personal clean room designed to capture all particulate matter (including Mtb bacilli) released by individuals (Dinkele et al., 2021).By combining liquid capture of bioaerosols in the RASC with DMN-trehalose-enabled microscopic detection and visualization of Mtb bacilli, we recently reported that ~90% of clinic attendees with presumptive TB produced bioaerosols containing viable Mtb bacilli (Patterson et al., 2024).Surprisingly, the high proportion of Mtb bioaerosol-positivity occurred irrespective of Sinal TB diagnosis and included both notiSied TB patients -who commenced anti-TB chemotherapy -and those not diagnosed with TB (who did not initiate treatment).Moreover, during six-months' follow-up, the rates of decline in symptom severity and Mtb bioaerosol release were equivalent in both groups.And, at the six-month study endpoint, approximately 20% of all participants -including treated TB patients and those who did not receive a TB diagnosisremained Mtb bioaerosol-positive despite clinical resolution, consistent with prior observations from radiological assays (Malherbe et al., 2016).
Given the high-prevalence of Mtb bioaerosol release among clinic attendees not diagnosed with TB, as well as the frequency of Mtb bioaerosol release in conSirmed TB patients on completion of standard anti-TB treatment, we hypothesized that the production of Mtb bioaerosols by individuals living in TB-endemic communities might be more common than previously thought.
To investigate this possibility, we utilized our bioaerosol sampling platform (Dinkele et al., 2022) to investigate Mtb release in 89 randomly selected participants recruited into two consecutive cohorts: the Sirst was a cross-sectional community survey comprising 39 participants, and the second was a longitudinal observational study in which 50 individuals consented to be sampled at three separate timepoints over two months.As detailed below, our results suggest that earlystage Mtb infection is pervasive and may be an important contributor to the transmission of Mtb in high TB-burden communities. .

The high prevalence of Mtb bioaerosol release occurs independently of respiratory maneuver
To investigate the prevalence of Mtb bioaerosol release in a TB-endemic setting, we initiated a cross-sectional community survey in Masiphumelele, Cape Town, with recruitment randomized geospatially by parcel of land, or "erf" (Fig. 1).From 10 erfs, we sequentially enrolled 39 individuals (1-12/erf [median=3]) who were assessed in Mtb bioaerosol release, sputum-GeneXpert MTB/RIF (GXP), and QuantiFERON-TB Gold (QFT) assays.The cohort had high rates of Human ImmunodeSiciency Virus (HIV) infection (13%) and previous TB (18%).Two participants who produced Mtb bioaerosols at the screening visit (5%) were GXP-positive at follow-up (one of these participants had completed treatment for TB in 2020, two years prior to the study), providing immediate evidence of undiagnosed TB in the community (Table 1).participants were recruited from randomly selected erfs in Masiphumelele.At a ;irst screening visit, participants produced bioaerosol samples from three respiratory maneuvers; forced vital capacity (FVC), tidal breathing (TiBr), and induced cough.Samples were processed and visualized independently by microscopists blinded to all sample information.Owing to the high prevalence of Mtb bioaerosol positivity, all participants were brought back for a follow-up visit during which blood and sputum were collected for QFT and GXP analysis, respectively.(B) 50 participants were recruited into a longitudinal observational study of Mtb release.Blood and sputum samples were collected at baseline for QFT and GXP analyses, respectively.Two equivalent bioaerosol samples were collected during ten minutes of tidal breathing with deep breaths taken at 30-second intervals.These samples were processed and imaged independently on nanowell-arrayed microscope slides by microscopists blinded to all sample information.This process was repeated at two weeks and two months post initial recruitment.
Advanced lung pathology and chronic cough are often viewed as essential for Mtb release (Turner and Bothamley, 2015).This assumption is, however, incompatible with the notion that Mtb may transmit during early-stage infection.
Here, we examined the potential for Mtb release from a randomly selected community cohort, considering both unrelated cough (Esmail et al., 2018) and tidal breathing (Dinkele et al., 2022) as potential mechanisms.This was done by comparing Mtb aerosolization from three independently sampled respiratory maneuvers; namely, forced vital capacity (FVC -a deep breathing respiratory maneuver), tidal breathing (TiBr), and cough, according to our previously described methodology (Dinkele et al., 2022).When comparing the respiratory maneuvers, the percentage of Mtb-positive samples ranged from 51.1-67.6%, with no signiSicant differences observed in the likelihood of producing Mtb between the three respiratory maneuvers (Fig. 2A &   B).Moreover, no differences were observed in the average number of bacteria identiSied per sample (Fig. 2C & D).Across FVC, TiBr, and cough, the average counts were 2.9 (95% CI: 1.9; 4.0), 2.2 (95% CI: 1.4; 3.2), and 2.0 (95% CI: 1.1; 3.1) Mtb bacilli per sample, respectively.Overall, we found that 79.5% (31/39) of the participants produced at least one positive bioaerosol sample when looking across all three maneuvers.Together, these data suggest that the prevalence of Mtb release is high in a TB-endemic setting.

Altering the bioaerosol sampling algorithm did not reduce Mtb detection ef:iciency
The observation that ~80% of a randomly selected community cohort produced Mtb bioaerosols was reminiscent of our previous work, which demonstrated the high (~90%) prevalence of Mtb release among clinic attendees with presumed TB, regardless of Sinal diagnosis or respiratory maneuver (Patterson et al., 2024).However, the cross-sectional community survey only provided a snapshot in time of Mtb bioaerosol release.To investigate the variability of Mtb release through time, we repeated our random community sampling strategy in recruiting a further 50 individuals from 20 erfs (1-7/erf [median=2]) into a longitudinal observational study of Mtb release (Fig. 1).Bioaerosols were collected at baseline, and follow-up visits were scheduled at two weeks and two months post baseline, in accordance with previously described time intervals (Patterson et al., 2024).The characteristics of this cohort were not dissimilar to the Sirst 39 participants (Table 1), except for HIV status -where the difference was driven largely by the proportion of individuals reporting their status as "unknown".
Given the observation that FVC and TiBr were sufSicient to aerosolize Mtb, we reasoned our bioaerosol sampling algorithm could be simpliSied by removing induced cough.Therefore, rather than collecting three Sive-minute samples (~15 minutes), each from a different respiratory maneuver, we implemented a single sampling algorithm which was repeated twice: ten minutes of tidal breathing with deep breaths every 30 seconds (~20 minutes sampling in total).This enabled a head-to-head comparison of duplicate bioaerosol samples at a single visit and allowed for an assessment of variation through time.We found comparable results between Sirst and second cohorts in both the proportion of participants producing at least one positive sample and the total number of aerosolized Mtb (Fig. 3A-C).Average Mtb counts at baseline varied from 0-16, with 20% (10/50) of the participants producing two negative bioaerosol samples (Fig. 4A).
Discrepant samples (i.e., where one sample was negative and the other positive), were relatively rare [22% (11/50)] with a low maximum sample difference (3 bacilli).There was a strong correlation between the Sirst and second bioaerosol sample (Fig. 4B) and a high degree of   3).Surprisingly, neither QFT status -the conventional marker of Mtb infection -nor HIV status were associated with the associated with Mtb release.Overall, these data suggest that individuals who aerosolize Mtb do so consistently within short time frames and independently of previously accepted markers of infection.

Persistent Mtb release among a randomly selected community cohort is common
Based on the observation from baseline sampling that 94% of matched samples differed by less than Sive, we reasoned that differences in the average count between two consecutive samples through time of Sive or greater could be considered signiSicant.According to this deSinition, most [74% (32/43)] participants were constant low-level producers of Mtb over two months (Fig. 5A).
The rate of bioaerosol positivity was relatively consistent through time, ranging from 76.7-90.7%(Fig. 5B & C).Most surprisingly, only one individual was negative across all three visits (representing six negative bioaerosol samples).Of 43 individuals with complete data for all three timepoints (baseline, two weeks, two months), 19% (8/43) and 9% (4/43) were negative at one and two visits, respectively.Although individual participants varied in their bioaerosol count from visit to visit, there was no overall trend in the number of Mtb detected per visit (Fig. 5D & E).This lack of secular trend in Mtb bioaerosol release among a randomly selected community cohort supports our previous hypotheses that Mtb bioaerosol clearance is immune driven (Patterson et al., 2024).

Discussion.
The use of new tools in TB research has enabled unexpected insights that question established models of pathogenicity.For Mtb transmission, the increasing awareness that TB disease is not required for Mtb bioaerosol release interrogates previous assumptions linking pathology to infectiousness (Dinkele et al., 2024), in turn raising important questions about the implications for new diagnostics, drugs, and vaccines (Fortune, 2024).In work immediately preceding this report, we detected high-prevalence Mtb bioaerosol release among TB clinic attendees diagnosed as not having TB, as well as the persistent release of Mtb bioaerosols post-treatment among bacteriologically conSirmed TB cases (Patterson et al., 2024)a Sinding consistent with separate work which used positron emission tomography-computed tomography (PET-CT) imaging and Mtb mRNA detection (Malherbe et al 2016) to conclude that apparently curative treatment for TB might not eradicate all Mtb bacteria.One implication of our previous results was that release of Mtb bioaerosols within TB-endemic communities might be considerably more prevalent than previously understood.To address that possibility, we initiated .the work reported here, in which we broadened our sampled population to screen a random selection of individuals from the larger of the two communities served in the earlier study.In doing so, we modiSied the sampling protocol slightly to exclude unnecessary forced coughing and, importantly, demonstrated high reproducibility in Mtb capture numbers when repeated samples were obtained from the same individual at the same visit.
The results were again very unexpected: we identiSied aerosolized Mtb in 80% of participants recruited at random from a high TB-burden community.Moreover, Mtb release occurred independently of QFT results, the test conventionally used to detect current or previous exposure to Mtb (Cohen et al., 2019).Together, these observations suggest that current methods for detecting Mtb infection and estimating patient infectiousness are insufSicient, and that Mtb release is far more common in TB-endemic settings than previous appreciated.They also imply a large reservoir of potential transmitters that remains invisible to national TB programs, although deSinitive demonstration of Mtb transmission from this population of relatively well individuals is necessary.Further studies are planned to establish the attributable fraction of TB transmission from this population; the immediate implications, though, are that a focus on treatment as the sole intervention to prevent transmission will require reassessing if the attributable fraction is signiSicant in high TB-burdened settings.In addition, the potential utility of immune strategies including vaccination might be negatively impacted if these organisms are able to avoid immune surveillance or are in a relatively immune-privileged site.
Given the assumed importance of lung pathology and chronic cough in TB transmission (Turner and Bothamley, 2015), there are currently few mechanistic explanations for how bacilli are aerosolized during early-stage Mtb infection.Notably, from longitudinal sampling of 50 participants over 2 months, it was apparent that most individuals were transiently or persistently releasing Mtb, with only one individual returning negative Mtb-bioaerosol results across all three visits (representing six negative bioaerosol samples).Despite variation in Mtb release per participant, no overall trend was observed in Mtb release through time for the cohort.This observation supports our previous inference that, among a predominantly symptomatic cohort of clinic attendees with presumptive TB, the reduction in Mtb release through time might be largely immune driven (Patterson et al., 2024).
These results might offer new insights into the Mtb host-pathogen interaction.It is tempting to suggest, for example, that the persistent bioaerosol release of Mtb within this community could indicate a homeostatic interaction between the bacillus and its human host.Although half of the participants were QFT positive, there was no correlation with aerosol Mtb exhalation.This could be explained if the aerosolized Mtb phenotype remains immunologically unrecognized or is present only in a relatively immune-privileged site.Notably, HIV status was not associated with aerosolized Mtb, further implying a lack of immune interaction between host and pathogen.In this context, recent work suggesting intercellular replication of Mtb within bioSilm-like cords might be apposite (Mishra et al., 2023).Determining the anatomical location of bacilli released in bioaerosols in asymptomatic individuals is a key research priority.
Our results must be interpreted in the light of the study's limitations, including sample size and composition, as well as the potential for the false-positive identiSication of Mtb.Participants were recruited from randomly selected erfs during the workday between Monday-Friday.Our sample was therefore biased to individuals both willing and able to participate in the study.However, the average age, QFT status, and HIV prevalence were as expected for the community (Middelkoop et al., 2014), suggesting that our observations are internally reliable within Masiphumelele.It seems unlikely, therefore, that the manner of participant recruitment would signiSicantly impact the results of this study.
Our assay for the detection of Mtb release is based on the microscopic detection of bacilli matching speciSic morphological characteristics, which also take up with DMN-tre (Dinkele et al., 2021).
This technique is potentially subject to the false-positive identiSication of Mtb, given that other actinomycetales produce the enzyme required for the probe incorporation (Kamariza et al., 2018).However, we have previously used the RASC for Mtb detection by colony formation and droplet-digital PCR (ddPCR) (Wood et al., 2016, Patterson et al., 2018, Patterson et al., 2024).More recently, we successfully obtained whole-genome sequence data from three bioaerosol samples after Sifty-day culture, importantly demonstrating the presence of Mtb, with no non-tubercular mycobacteria or other actinomycetales found (Patterson et al., 2024).
With these limitations in mind, we interpret our results as indicating that Mtb bioaerosol release is common in this community and is not detectable by standard measures of Mtb infection.If correct, these observations motivate for an urgent reframing of the prevailing paradigm of Mtb transmission and infection.The notion that infection with Mtb constitutes the primary determinant of TB risk in high-TB burden settings has dominated thinking about new TB interventions.Our work using advanced breath aerosol collection technology challenges this assumption: in successive studies, we have detected release of Mtb-containing bioaerosols in conSirmed TB patients and a majority of randomly selected community members, and we have demonstrated the time-dependent reduction (but not elimination) of Mtb bioaerosol positivity in TB clinic attendees irrespective of TB chemotherapy.These observations reinforce the axiom that, while Mtb infection is necessary for TB, it is not sufSicient: instead, a poorly understood combination of host, environmental, and mycobacterial factors determine disease risk.
In turn, these Sindings have important implications on the development and testing of new vaccines and prognostic models, as well as the design of future transmission studies and .intervention strategies.Although the signiSicance for TB control and the attributable transmission risk from this population remains to be established, these data provide a plausible explanation for the difSiculty in linking transmission chains in high-burden regions.Moreover, since the bacilli produced by randomly selected community members appear quantitatively and qualitatively indistinguishable from those released by conSirmed TB patients (at least using existing tools to interrogate phenotypic and/or genetic adaptations), it seems plausible that there is some contribution to ongoing Mtb transmission.By extrapolating to the broader community, the implication appears unavoidable that there are far greater numbers of disease-and symptom-free individuals with prolonged shedding of Mtb in communal settings than those with TB disease.

Study population and participant recruitment
This study was conducted in Masiphumelele, a peri-urban township located south of Cape Town, South Africa.This residential area is divided into demarcated parcels of land, called "erfs" under South African legislation, each with its own unique numeric identiSier.Erfs were randomly selected and all consenting participants over 14 years of age were eligible for recruitment into this study.Recruitment was conducted between February-December 2022, with ethical approval from the Human Research Ethics Committee of the University of Cape Town (HREC no.

The Respiratory Aerosol Sampling Chamber
The RASC is a purpose-built personal clean room equipped with a high efSiciency bioaerosol collection system, which captures all exhaled particulate matter (bioaerosols) at 100-300 L/min in 15 mL of collection medium (sterilized phosphate-buffered saline supplemented with Polymyxin B, Amphotericin B, Nalidixic acid, Trimethoprim and Azilocillin (PANTA) [Becton Dickinson]).Bioaerosol samples were concentrated and stained with DMN-trehalose (Olilux Biosciences Inc.) overnight, as previously reported, before transfer to nanowell devices for imaging (Dinkele et al., 2021).

Bioaerosol generation and sample processing
This study was conducted in two phases (Fig. 1).In the Sirst phase, participants 1-39 underwent bioaerosol collection during three respiratory maneuvers: FVC, TiBr, and induced cough, as previously described (Dinkele et al., 2022).The sequence of these maneuvers remained consistent throughout the study.During FVC and cough sampling, participants performed the designated maneuver 15 times within a Sive-minute period, as directed by the study nurse.For tidal breath .sampling, participants were instructed to breathe normally into the bioaerosol collection system for Sive minutes.Each bioaerosol sample underwent independent collection, processing, and enumeration.Enumeration of Mtb bacilli was conducted by depositing concentrated and DMNtrehalose-stained bioaerosol samples onto nanowell devices.Bioaerosol positivity was determined by aggregating counts from all three maneuvers.Given the high proportion of bioaerosol positivity within this cohort, all participants were scheduled for a follow-up visit to obtain blood and sputum samples for subsequent QFT and GXP analysis.
The second phase involved the recruitment of participants 40-89 for a longitudinal examination of Mtb bioaerosols.This phase comprised 50 individuals sampled on three occasions: baseline, two weeks, and two months.At each visit, two equivalent bioaerosol samples were collected, with each involving ten minutes of tidal breathing interspersed with deep breaths every 30 seconds.
Subsequently, these samples were concentrated and arrayed on nanowell devices.Blood and sputum specimens were obtained at baseline for QFT and GXP analysis, respectively.In all cases, microscopists were blinded to the origin of the sample (which included an empty booth negative control) and enumerated putative Mtb bacilli per sample.

QuantiFERON-TB Gold assay
In this study, BARC SA were contracted to perform and interpret QFT assays according to the manufacturer's instructions (Qiagen).

Sputum collection and the GeneXpert assay
For all participants, sputum sampling was attempted.In those capable of producing sputum, a GXP assay was performed and interpreted according to the manufacturer's instructions (Cepheid).Individuals unable to produce sputum were considered sputum-GXP negative.

Statistical methods
For the descriptive statistics, a Fisher's Exact Test was used for the categorical variables given the relatively small sample size.To assess age, the normality and variance of the data were assessed, and a Student's T Test was used for the comparison.Generalized linear regression was used to assess the bioaerosol results, either as a dichotomous outcome (logistic regression) or a count outcome (negative binomial regression).These outcomes were regressed against age, biological sex, previous TB, QFT results, and HIV status.For the respiratory maneuver comparison, we used the equivalent mixed effects regression models to account for the fact that each participant produced three samples. .

Figure 1 :
Figure 1: Participant recruitment and bioaerosol sampling algorithms for the two cohorts.(A) 39participants were recruited from randomly selected erfs in Masiphumelele.At a ;irst screening visit, participants produced bioaerosol samples from three respiratory maneuvers; forced vital capacity (FVC), tidal breathing (TiBr), and induced cough.Samples were processed and visualized independently by microscopists blinded to all sample information.Owing to the high prevalence of Mtb bioaerosol positivity, all participants were brought back for a follow-up visit during which blood and sputum were collected for QFT and GXP analysis, respectively.(B) 50 participants were recruited into a longitudinal observational study of Mtb release.Blood and sputum samples were collected at baseline for QFT and GXP analyses, respectively.Two equivalent bioaerosol samples were collected during ten minutes of tidal breathing with deep breaths taken at 30-second intervals.These samples were processed and imaged independently on nanowell-arrayed microscope slides by microscopists blinded to all sample information.This process was repeated at two weeks and two months post initial recruitment.

Figure 2 :
Figure 2: The production of aerosolized Mtb during all three respiratory maneuvers in the first cohort.(A) The percentage of samples in which putative Mtb were detected (turquoise) or absent (purple) from forced vital capacity [FVC (67.6%)], tidal breathing [TiBr (51.4%)], and cough (51.1%).(B) Results of a logistic regression comparing the odds of a positive bioaerosol result compared to TiBr.(C) Box and density plots comparing the total number of Mtb detected between the three respiratory maneuvers.(D) Results of a negative binomial regression comparing the number of Mtb detected between the three respiratory maneuvers.OR = odds ratio, IRR = incident rate ratio, CI = confidence interval, BA = bioaerosol.

Figure 3 :
Figure 3: The production of aerosolized Mtb between the two cohorts.(A) The percentage of samples in which putative Mtb were detected (green) or absent (purple).The odds of a positive bioaerosol sample were equivalent between the two groups (OR = 1.03, 95% CI = 0.36;2.92,p = 0.952) (B) Box and whisker and (C) equivalent density plots comparing the total number of Mtb detected between the two cohorts.The rates at which Mtb were produced during the two samplings were equivalent (IRR = 0.797, 95% CI = 0.47;1.33,p = 0.386).OR = odds ratio, IRR = incident rate ratio, CI = confidence interval, BA = bioaerosol.

Figure 4 :
Figure 4: The production of aerosolized Mtb during two equivalent respiratory maneuvers in the second cohort.(A) Plot of the mean Mtb (DMN-tre positive) count from two samples, with error bars representing the range.No lines indicate equal counts, green lines indicate one count = 0, blue lines indicate two counts > 0. (B) Plot of the Mtb (DMN-tre positive) counts of the first and second samples at baseline (r = 0.810, p < 0.0001) with a fitted line representing a 1:1 correlation.(C) A Bland-Altman plot indicating the level of agreement between the first and second samples, with a (D) histogram showing the frequency of each count difference.Most samples differed by either 0 or ±1 (60%) and 94% of the samples differed by four or less.

Figure 5 :
Figure 5: The production of aerosolized Mtb through time in the second cohort.(A) Total Mtb (DMN-tre positive) counts (sum of the two samples) through time, stratified by time trend.(B) The percentage of samples in which putative Mtb were detected (turquoise) or absent (purple) at each of the visits.(C) Results of a logistic regression comparing the odds of a negative BA result compared to T0. (D) Box and density plots comparing the total number of Mtb bacilli (DMN-tre positive) detected at each visit.(E) Results of a negative binomial regression comparing the number of Mtb bacilli (DMN-tre positive) detected at each visit.OR = odds ratio, IRR = incident rate ratio, CI = confidence interval, BA = bioaerosol.

Table 1 : Summary of participant demographic and clinical information from the two cohorts.
NA = Not available (Incomplete or failed assay), * Excluded from the analysis .

Table 2 : Results of a logistic regression assessing the odds of a positive bioaerosol sample
. OR = odds ratio, CI = confidence interval.*Indeterminate results removed, # Unknown status removed

Table 3 : Results of a negative binomial regression assessing the number of Mtb per participant. IRR
* Indeterminate results removed, # Unknown status removed