
==== Front
Nat Commun
Nat Commun
Nature Communications
2041-1723
Nature Publishing Group UK London

52122
10.1038/s41467-024-52122-x
Article
Mycobacterium tuberculosis cough aerosol culture status associates with host characteristics and inflammatory profiles
Nduba Videlis 1
http://orcid.org/0000-0002-5067-0788
Njagi Lilian N. 1
Murithi Wilfred 1
Mwongera Zipporah 1
http://orcid.org/0009-0003-8740-2351
Byers Jodi 2
Logioia Gisella 3
Peterson Glenna 3
http://orcid.org/0000-0001-9806-259X
Segnitz R. Max 3
http://orcid.org/0000-0001-9416-0686
Fennelly Kevin 4
Hawn Thomas R. 3
http://orcid.org/0000-0002-4721-9099
Horne David J. dhorne@uw.edu

23
1 https://ror.org/04r1cxt79 grid.33058.3d 0000 0001 0155 5938 Centre for Respiratory Diseases Research, Kenya Medical Research Institute, Nairobi, Kenya
2 https://ror.org/00cvxb145 grid.34477.33 0000 0001 2298 6657 Department of Global Health, University of Washington, Seattle, WA USA
3 https://ror.org/00cvxb145 grid.34477.33 0000 0001 2298 6657 Department of Medicine, University of Washington, Seattle, WA USA
4 grid.94365.3d 0000 0001 2297 5165 Division of Intramural Research, National Heart, Lung and Blood Institute (NHLBI), National Institutes of Health, Bethesda, MD USA
1 9 2024
1 9 2024
2024
15 760415 12 2023
27 8 2024
© The Author(s) 2024
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Interrupting transmission events is critical to tuberculosis control. Cough-generated aerosol cultures predict tuberculosis transmission better than microbiological or clinical markers. We hypothesize that highly infectious individuals with pulmonary tuberculosis (positive for cough aerosol cultures) have elevated inflammatory markers and unique transcriptional profiles compared to less infectious individuals. We performed a prospective, longitudinal study using cough aerosol sampling system. We enrolled 142 participants with treatment-naïve pulmonary tuberculosis in Kenya and assessed the association of clinical, microbiologic, and immunologic characteristics with Mycobacterium tuberculosis aerosolization and transmission in 129 household members. Contacts of the forty-three aerosol culture-positive participants (30%) are more likely to have a positive interferon-gamma release assay (85% vs 53%, P = 0.006) and higher median IFNγ level (P < 0.001, 4.28 IU/ml (1.77-5.91) vs. 0.71 (0.01-3.56)) compared to aerosol culture-negative individuals. We find that higher bacillary burden, younger age, larger mean upper arm circumference, and host inflammatory profiles, including elevated serum C-reactive protein and lower plasma TNF levels, associate with positive cough aerosol cultures. Notably, we find pre-treatment whole blood transcriptional profiles associate with aerosol culture status, independent of bacillary load. These findings suggest that tuberculosis infectiousness is associated with epidemiologic characteristics and inflammatory signatures and that these features may identify highly infectious persons.

Using cough-generated aerosol cultures, authors probe the inflammatory markers and epidemiological characteristics of individuals with pulmonary tuberculosis, in association with infection state.

Subject terms

Bacterial host response
Tuberculosis
Infection
Tuberculosis
https://doi.org/10.13039/100000060 U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID) 5R01AI150815 UH2AI152621 TW011817-01 5R01AI150815 TW011817-01 TW011817-01 TW011817-01 5R01AI150815 TW011817-01 Nduba Videlis Njagi Lilian N. Murithi Wilfred Hawn Thomas R. Horne David J. U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID)U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID)https://doi.org/10.13039/100006108 U.S. Department of Health & Human Services | NIH | National Center for Advancing Translational Sciences (NCATS) UL1 TR002319 Horne David J. U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID)U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID)U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID)U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID)U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID)U.S. Department of Health & Human Services | NIH | National Institute of Allergy and Infectious Diseases (NIAID)University of Washington/Fred Hutch Center for AIDS Research (AI027757)issue-copyright-statement© Springer Nature Limited 2024
==== Body
pmcIntroduction

Tuberculosis (TB), a leading infectious disease-related cause of death1, is spread person-to-person through the inhalation of aerosolized bacilli. A key step for TB elimination is interrupting transmission events to prevent new acquisition of infection and disease. However, knowledge gaps in understanding the biology and determinants of TB transmission undermine efforts to develop interventions2,3. These gaps include uncertainties around the determinants of infectiousness, poor estimations of individual infectiousness, and the lack of accurate and convenient biomarkers of infectiousness. Similar to other infectious diseases4–6, “superspreaders”, individuals with TB who are responsible for the majority of Mycobacterium tuberculosis (Mtb) transmission7–11 may have an outsized role in TB endemic settings. Modeling studies suggest greatly amplified returns by focusing control efforts on the minority of persons with TB who are most infectious12. Although the existence of “superspreaders” in Mtb transmission remains unproven, investigations into this phenotype may provide insights into TB pathophysiology and inform the development, implementation, and evaluation of targeted public health interventions to eliminate TB3.

Epidemiologic studies of Mtb infectiousness suggest that several host characteristics, including younger age, higher bacillary load, cough features, and greater contact time, are associated with increased transmission13–18. Overall, these associations are weak and suggest that additional factors regulate infectivity. Studies by Wells and Riley from the pre-chemotherapeutic era demonstrated that a subset of patients with TB generated small droplets capable of infecting guinea pigs15,19–21. More recent studies focused on people living with HIV (PLWH) similarly found that a subset of patients transmitted Mtb22. Compared to sputum smear and culture assessment, direct measurement of Mtb aerosolization with a cough aerosol sampling system (CASS)23–28 is a more accurate predictor of transmission among household contacts (HHCs)24,26. A recent study suggested that sputum bacillary load and clinical characteristics, including fewer symptoms and a stronger cough, predict Mtb aerosolization29. Studies using a Respiratory Aerosol Sampling Chamber (RASC), which directly measures viable Mtb in bioaerosols30–32, as well as those that used face mask sampling33 suggested that tidal breathing and non-cough mechanisms may be important drivers of transmission34. Although these studies demonstrate features of airborne Mtb transmission, the biological mechanisms remain poorly understood, including whether immune pathways modulate aerosolization. In addition, while CASS and RASC are important research tools, a simple diagnostic test or algorithm that identifies superspreaders has not been developed and would have a significant impact on efficient resource allocation for TB control35,36.

To address these gaps in knowledge, we designed the TB Aerobiology, Immunology and Transmission (TBAIT) study. We performed a prospective, longitudinal study using CASS, enrolled 142 participants with pulmonary TB, and assessed the association of clinical, microbiologic, and immunologic characteristics with Mtb aerosolization. In addition, we measured interferon-gamma release assay (IGRA) responses in HHCs to assess Mtb transmission risk by cough aerosol culture (CAC) status of the index participant.

Results

Cough aerosol culture results among participants with pulmonary TB

To examine the biology of Mtb aerosolization and transmission, we enrolled 142 individuals with microbiologically confirmed pulmonary TB in Nairobi, Kenya. The median age of participants was 35 years (interquartile range (IQR), 27–44) and ranged from 18 to 100 years; 27% were women (Table 1). All participants identified as Black. Most participants were symptomatic including cough in 95% and weight loss in 85%. Most participants had medium (32%) or high (39%) GeneXpert semi-quantitative grades and findings of cavitations on chest x-rays (70%). Only 11% of participants were living with HIV, of whom one-half were taking antiretroviral therapy at enrollment. As compared to participants identified through passive case finding (n = 116), those identified through active case finding (n = 26) were older (mean 43 vs. 35 years, p-value 0.005), less likely to report a cough (81% vs. 98%, p-value < 0.001), less likely to have cavitary disease on chest x-ray (38% vs 77%, p-value < 0.001) and had higher mean GeneXpert Ct values (24.0 vs. 19.2, p-value < 0.001); there was no difference in the frequency of women or PLWH.Table 1 Characteristics of Participants with TB (Index participants) overall and by cough aerosol culture (CAC) status

Characteristics	Total n = 142 N(%) or (med, IQR)	CAC +  n = 43	CAC- n = 99	Odds Ratio	p-value	
Age, years	35 (27, 44)	32 (24, 39)	36 (28, 48)	0.95 (0.91 − 0.98)	0.006	
Female	39 (27%)	8 (19%)	31 (31%)	0.50 (0.21 − 1.21)	0.12	
BMI (n = 141)	19 (17, 21)	19 (18, 21)	19 (17, 21)	1.02 (0.92 − 1.14)	0.69	
MUACa, cm (n = 141)	23 (21, 25)	24 (21, 26)	22 (21, 25)	1.14 (1.00 − 1.30)	0.04	
PLWHb	16 (11%)	4 (9%)	12 (12%)	0.74 (0.23 – 2.45)	0.88	
Taking ART?	8 (50%)	4 (100%)	4 (33%)		0.02	
CD4 T-cell count, cells/µL (n = 15)	193 (67, 231)	218 (105, 397)	178 (67, 216)	1.00 (1.00 − 1.01)	0.63	
Prior history of TB (n = 140)	27 (19%)	8 (19%)	19 (20%)	0.94 (0.37 − 2.35)	0.89	
Enrolled through active case finding	26 (18%)	0	26 (26%)	–	–	
Household residents	2 (1, 4)	2 (1, 4)	2 (1, 4)	–	0.94	
Chest X-ray with cavitary findings	99 (70%)	38 (88%)	61 (62%)	4.73 (1.71 − 13.08)	0.003	
Number of chest X-ray quadrants with TB involvement (n = 140)	–	–	–	1.50 (1.02 − 2.20)	0.04	
0	1 (1%)	0	1 (1%)	–	–	
1	47 (34%)	8 (19%)	39 (40%)	–	–	
2	53 (38%)	18 (42%)	35 (36%)	–	–	
3	28 (20%)	15 (35%)	13 (13%)	–	–	
4	11 (8%)	2 (5%)	9 (9%)	–	–	
Cough (n = 141)	133 (95%)	43 (100%)	91 (93%)	–	0.07	
Fever (n = 141)	92 (65%)	35 (81%)	57 (58%)	3.15 (1.32 − 7.49)	0.10	
Weight loss (n = 141)	120 (85%)	41 (95%)	79 (81%)	1.00 (0.97 − 1.03)	0.76	
Night sweats	108 (77%)	36 (84%)	72 (73%)	1.86 (0.74 − 4.69)	0.19	
Cough duration (days)(n = 134)	8 (4,12)	8 (4, 12)	6 (4, 12)	1.03 (1.0 − 1.07)	0.10	
LCQ Scorec (n = 140)	14.5 (12.2, 16.7)	13.1 (10.4, 14.4)	15.5 (12.8, 17.2)	0.81 (0.71 − 0.92)	0.001	
CPFd, mL	260 (170, 370)	225 (170, 350)	270 (170, 370)	1.00 (1.0 − 1.0)	0.23	
Current tobacco use	32 (23%)	9 (21%)	23 (23%)	0.86 (0.36-2.06)	0.74	
Xpert Ct	18.8 (16.6, 23.3)	16.4 (14.6, 18.1)	19.7 (17.8, 25.1)	0.73 (0.63 − 0.84)	0.000006	
Xpert Semiquantitative Grade	–	–	–	2.12 (1.35 − 3.22)	0.001	
Trace	4 (3%)	0	4 (4%)	–	–	
Very Low	9 (6%)	1 (2%)	8 (8%)	–	–	
Low	28 (20%)	3 (7%)	25 (26%)	–	–	
Medium	45 (32%)	15 (35%)	30 (30%)	–	–	
High	56 (39%)	24 (56%)	32 (32%)	–	–	
Rifampin resistance detected	1 (1%)	0	1 (1%)	–	–	
Sputum Smear Grade	–	–	–	1.62 (1.16 − 2.26)	0.004	
Negative	15 (11%)	1 (2%)	14 (14%)	–	–	
Scanty	9 (6%)	2 (5%)	7 (7%)	–	–	
1 +	33 (23%)	5 (12%)	28 (28%)	–	–	
2 +	35 (25%)	16 (37%)	19 (19%)	–	–	
3 +	50 (35%)	19 (44%)	31 (31%)	–	–	
TTDe (days) (n = 136)	5 (4, 9)	4 (3, 5)	6 (4, 10)	0.75 (0.63 − 0.88)	0.0005	
Sputum Appearance	–	–	–	–	–	
Thin	46 (32%)	10 (23%)	36 (36%)	REF		
Thick	91 (64%)	32 (74%)	59 (60%)	1.95 (0.86 − 4.44)	0.11	
Bloody	5 (4%)	1 (2%)	4 (4%)	0.90 (0.09 − 8.98)	0.93	
CRP, mg/dL (n = 141)	67 (26, 96)	88 (57, 131)	51 (15, 86)	1.02 (1.01 − 1.03)	0.00005	
WBC	7.1 (5.7, 9.1)	7.7 (6.4, 10.1)	6.8 (5.6, 8.6)	1.19 (1.03 − 1.36)	0.02	
Granulocytes	5.1 (3.8, 7.0)	5.7 (4.7, 7.8)	4.8 (3.6, 6.6)	1.20 (1.03 − 1.39)	0.02	
Lymphocytes	1.4 (1.1, 1.8)	1.5 (1.1, 1.9)	1.4 (1.0, 1.8)	1.20 (0.72 – 1.99)	0.49	
Monocytes	0.4 (0.2, 0.5)	0.3 (0.3, 0.5)	0.4 (0.2, 0.5)	0.77 (0.29 – 2.07)	0.61	
HbA1C, % (n = 141)	5.7 (5.3, 6.0)	5.8 (5.5, 6.1)	5.6 (5.1, 6.0)	1.1 (0.79 − 1.45)	0.67	
Hgb	13 (11.9, 14.2)	13.0 (12.0, 13.9)	13.1 (11.8, 14.3)	1.00 (0.84 − 1.19)	1	
aMUAC, Mid-upper arm circumference; bPLWH, a person living with HIV; c LCQ, Leicester Cough Questionnaire; dCPF, Cough peak flow; eTTD, Time to detection of Mtb growth in liquid media

Characteristics were compared by cough aerosol culture status using bivariate logistic regression, except for cough, which was tested using a chi-square statistic. No adjustments were made for multiple comparisons. Source data are provided as a Source Data file.

The median CRP value among all participants was high at 67 mg/L (IQR 26–96). Forty-three participants (30%) were cough aerosol culture-positive and had higher measures of bacillary burden, including higher AFB-smear grades and shorter time to detection of Mtb growth in liquid culture, compared to cough aerosol culture-negative participants (Table 1). Cough aerosol culture-positive participants were more likely to have cavitary lung disease and at least two quadrants affected by TB on chest x-ray than cough aerosol culture-negative persons. GeneXpert cycle threshold was significantly lower and semi-quantitative grade significantly higher among cough aerosol culture-positive compared to cough aerosol culture-negative participants. More cough aerosol culture-positive participants (91%) had GeneXpert semi-quantitative grades of medium or high compared to 62% of cough aerosol culture-negative individuals. The correlation between GeneXpert cycle threshold and cough aerosol culture status was − 0.44, a moderate correlation which suggests that additional factors determine aerosolization.

Index cough aerosol culture positivity associated with IGRA positive result in household contacts

We next evaluated IGRA results in household contacts to determine whether cough aerosol culture status is associated with infectiousness. Of the 142 participants with TB, 30 lived alone and were not eligible for the household contact study, and 64 had household members who could not be contacted, or the household members declined enrollment. Index participants (N = 48) with enrolled household members compared to index participants (N = 94) without enrolled household members did not differ by age (p-value 0.12), HIV status (p-value 0.74), MUAC (p-value 0.60), or CASS status (0.17). (Supplementary Table 5) However, there were differences in the proportion of women (42% vs. 20%, p-value 0.007), Xpert Ct values (20.5 vs. 18.1, p-value 0.01), and the frequency of cavitations on chest X-rays (58% vs. 76%, p-value 0.04) between those with and without enrolled household members, respectively. We enrolled 129 household contacts and obtained QFT tests in 116 (13 declined phlebotomy or had unsuccessful phlebotomy) (Table 2). In bivariate analyses, household contacts did not differ in gender or age by cough aerosol culture status of the index participant. The overall frequency of QFT-positive results among household contacts was 60% and differed by cough aerosol culture status of the index participant: excluding indeterminate results, 85%, and 53%, for contacts of cough aerosol culture-positive and -negative participants, respectively (p-value 0.006). Among household contacts less than 10 years of age, QFT-positive results were also more common (p-value 0.01) among contacts of CAC-positive persons (9 of 10 participants, 90%) than contacts of CAC-negative persons (14 of 31 participants, 45%). Cough aerosol culture-positive contacts had a higher mean IGRA IFNγ level compared to cough aerosol culture-negative individuals (p-value 0.001, 4.25 IU/ml vs. 1.35 IU/ml). Among participants who were QFT-positive, the mean IFN-γ levels were significantly higher among contacts of cough aerosol culture-positive persons (n = 22, 4.76, SD 2.58) compared to contacts of cough aerosol culture-negative persons (n = 48, 3.36, SD 1.97, P = 0.02).Table 2 Household Contact Characteristics overall and by index cough aerosol culture (CAC) status

Characteristics	Total n = 129 N(%) or (med, IQR)	CAC + n = 27	CAC- n = 102	p-value	
Household Contact Characteristics (n = 129)	
Age, years (med, IQR)	12 (5, 29)	12 (5, 28)	13 (5, 32)	0.54	
Age groups	
< 5 years	31 (24%)	6 (22%)	25 (25%)	0.58	
5–10 years	20 (15%)	6 (22%)	14 (14%)	–	
10–15 years	19 (15%)	5 (19%)	14 (14%)	–	
> 15 years	59 (46%)	10 (37%)	49 (48%)	–	
Women	66 (51%)	15 (56%)	51 (50%)	0.61	
BMIb (n = 122)	19 (16, 24)	19 (15, 25)	20 (16, 24)	0.39	
MUACc, cm (n = 123)	21.5 (16, 25.5)	19.2 (15.5, 29.5)	22 (16.2, 25)	0.87	
PLWHa (n = 99)	4 (4%)	1 (5%)	3 (4%)	0.76	
Characteristics of Index Participants (n = 48) with whom Household Contacts Resided	
Age	34 (27, 47)	37 (29, 41)	33 (27,47)	0.49	
Women	48 (37%)	10 (37%)	38 (37%)	0.98	
PLWHa	17 (13%)	2 (7%)	15 (15%)	0.32	
BMIb (n = 126)	19 (18, 23)	18 (18, 21)	19 (18, 23)	0.21	
MUACb (n = 126)	23.2 (20.8, 25.6)	22.9 (22.4, 25.5)	23.3 (20.8, 25.8)	0.38	
Xpert Ct (n = 129)	20.5 (17.6, 26.7)	19.0 (16.6, 20.4)	22.7 (17.8, 27.3)	0.0002	
Xpert Grade (n = 129)	–	–	–	0.003	
Trace	3 (2%)	0	3 (3%)	–	
Very Low	17 (13%)	0	17 (17%)	–	
Low	30 (23%)	2 (7%)	28 (27%)	–	
Medium	53 (41%)	19 (70%)	34 (33%)	–	
High	26 (20%)	6 (22%)	20 (20%)	–	
Smear Grade (n = 129)	–	–	–	0.25	
Negative	23 (18%)	4 (15%)	19 (19%)	–	
Scanty	5 (4%)	1 (4%)	4 (4%)	–	
1 +	42 (33%)	7 (26%)	35 (34%)	–	
2 +	27 (21%)	10 (37%)	17 (17%)	–	
3 +	32 (25%)	5 (19%)	26 (26%)	–	
TTDd (days) (n = 122)	5.5 (4, 12)	5 (4, 6)	6 (4, 12)	0.04	
Cavitary CXR (n = 126)	74 (59%)	24 (89%)	50 (51%)	0.001	
CRP (n = 124)	46 (19, 75)	51 (26, 96)	46 (7, 75)	0.11	
CXR quadrants (n = 126)	–	–	–	0.001	
0	0	0	0	–	
1	49 (39%)	6 (22%)	43 (43%)	–	
2	54 (43%)	11 (41%)	43 (43%)	–	
3	18 (14%)	10 (37%)	8 (8%)	–	
4	5 (4%)	0	5 (5%)	–	
Number of contacts for each index participant (% of households)	–	–	–	0.02	
1 HHC	16 (33%)	4 (36%)	12 (32%)	–	
2 HHCs	9 (19%)	3 (27%)	6 (16%)	–	
3 HHCs	8 (17%)	1 (9%)	7 (19%)	–	
4 HHCs	7 (15%)	1 (9%)	6 (16%)	–	
5 HHCs	5 (10%)	2 (18%)	3 (8%)	–	
6 HHCs	3 (6%)	0	3 (8%)	–	
QuantiFERON (QFT) Result in HHC (n = 116)	
QFT-neg	44 (38%)	3 (12%)	41 (46%)	0.006	
QFT-pos	70 (60%)	22 (85%)	48 (53%)	–	
Indeterminate	2 (2%)	1 (4%)	1 (1%)	–	
Mean IFN-γ (IU/mL) (SD) (indeterminate excluded)	2.34 (2.55)	4.18 (2.89)	1.83 (2.20)	0.0002	
Median IFN-γ (IU/mL) (IQR) (indeterminate excluded)	1.02 (0.02 − 4.45)	4.28 (1.77 − 5.91)	0.71 (0 − 3.56)	–	
QFT-positive HHCs, Mean IFN-γ (IU/mL) (SD) (n = 70)	3.80 (2.25)	4.76 (2.58)	3.36 (1.97)	0.02	
QFT-positive HHCs, Median IFN-γ (IU/mL) (IQR) (n = 70)	3.71 (1.85-5.06)	4.59 (3.35–6.97)	3.46 (1.82–4.79)	–	
aPLWH, person living with HIV; bBMI, Body mass index; cMUAC, Mid-Upper Arm Circumference; dTTD = Time to detection of Mtb growth in liquid media

Using bivariate logistic regression, characteristics were compared by cough aerosol culture status. No adjustments were made for multiple comparisons. The source data are provided as a Source Data file.

To establish the background prevalence of QFT-positive results in Nairobi residents, we evaluated QFT status in adult outpatients who presented for health care due to cough (n = 45, median age 39 years (IQR, 34-45)) and to HIV prevention and care centers (n = 121 PLWH; n = 122 without HIV; overall median age 36 years (IQR, 28-44). After excluding indeterminate results (n = 2 and n = 34, respectively), we found that the prevalence of QFT-positive results was, 51% (n = 22) and 51% (n = 121), respectively. Among the 45 persons who presented to health care with a cough, QFT status did not differ by HIV status (PLWH = 17, p-value 0.66). In participants recruited at HIV care centers, QFT results differed by HIV status: 43% of PLWH (48 of 111) and 60% of persons without HIV (59 of 98) (p-value 0.01) were QFT positive. The estimated Nairobi background prevalence of QFT test positivity did not differ from contacts of cough aerosol culture-negative participants (p-value 0.33) but was significantly lower than contacts of cough aerosol culture-positive participants (p-value < 0.001).

Using mixed effects logistic regression models, we next evaluated bivariate associations between measures of index case TB infectiousness and QFT results in contacts. (Table 3) Using multivariable models with random intercepts to account for clustering by index participant, we evaluated predictors of interest for associations with QFT positivity in household contacts and found that the best-performing model included index cough aerosol culture status and HIV status, and household contact’s age. (Table 3).Table 3 Mixed methods bivariate & multivariable analyses of index and household contact characteristics associated with QFT result in household contacts (excluding indeterminate results)

Predictorsa	OR	95% CI	p-value	aOR	95% CI	p-value	
Household Contact Characteristics	
Age, years	1.04	1.00–1.07	0.049	1.03	0.99–1.06	0.10	
Women	1.03	0.41–2.58	0.95	–	–	–	
PLWHa	2.16	0.16–28.45	0.56	–	–	–	
Index Characteristics	
Cough aerosol culture-positive	8.51	1.62–44.78	0.01	8.81	1.46–53.20	0.02	
Age, years	1.02	0.98–1.06	0.28	–	–		
Women	1.56	0.47–5.15	0.46	–	–		
PLWHa	0.09	0.01–0.65	0.02	0.12	0.02–0.88	0.04	
Xpert Ct	0.89	0.78–1.00	0.06	–	–	–	
Xpert semi-quantitative grade	1.50	0.85–2.63	0.16	–	–	–	
TTDb (days)	0.88	0.73–1.06	0.18	–	–	–	
AFB-smear grade	1.40	0.93–2.12	0.11	–	–	–	
CRP level, mg/dL	1.01	1.00–1.03	0.10	–	–	–	
Cavitary chest X-ray	2.85	0.93–8.69	0.07	–	–	–	
CXR Quadrants	1. 38	0.69–2.75	0.36	–	–	–	
Body Mass Index	0.98	0.85–1.13	0.79	–	–	–	
MUAC†	0.94	0.79–1.12	0.51	–	–	–	
aPLWH, person living with HIV; bTTD, Time to detection of Mtb growth in liquid media; †MUAC, Mid-upper arm circumference

aPredictors were selected based on our conceptual model (see Supplementary Fig. 1), tested in forward selection stepwise multivariable models with random intercepts to account for clustering by index participant, and compared with likelihood ratio test. The best-performing model with adjusted odds ratios is shown in the final three columns. No adjustments were made for multiple comparisons. Source data are provided as a Source Data file.

Models of index characteristics associated with cough aerosol culture status

We evaluated host characteristics associated with cough aerosol culture status based on a conceptual framework that included potential confounders (Table 1 and Supplementary Fig. 2). On bivariate analyses, younger age, higher MUAC, cavitary disease on chest X-ray, TB-related abnormalities in more chest X-ray quadrants, and higher cough-related impairment effects on quality of life (lower Leicester Cough Questionnaire score) were associated with cough aerosol culture-positive status. Factors associated with higher bacillary burden, including lower GeneXpert Ct value, higher GeneXpert grade, higher sputum smear grade, and shorter time-to-detection of Mtb culture growth, were all associated with cough aerosol culture-positive status. Higher WBC counts (total and granulocyte count) and CRP levels were also associated with cough aerosol culture-positive status (Fig. 1a). We compared multivariable models for predictors associated with cough aerosol culture-positive status and found that the best-performing model included lower GeneXpert Ct value, lower age, higher CRP level, higher MUAC, and shorter TTD of culture growth (Table 4). The coefficient of determination (R2) of this model was 0.37, suggesting that other factors, in addition to the evaluated independent variables, determine TB infectiousness.Fig. 1 Association of blood and sputum inflammatory and microbiologic markers with cough aerosol culture positivity.

Inflammatory and microbiologic markers were measured in blood and sputum at baseline visits and correlated with cough aerosol culture positivity. a Serum CRP, Blood WBC, sputum bacillary load determined by GeneXpert Ct value, and sputum culture time to detection (n = 99 CAC-; n = 43 CAC + ). b–d Sputum and plasma cytokines measured by ELISA; CXCL8, IL1B, IL6, and TNF measurements at diagnosis were correlated with cough aerosol culture positivity (b), bacillary (c), and chest x-ray lung cavitation load (panels b-d: sputum CXCL8: n = 81 CAC-, n = 37 CAC +; sputum IL1B: n = 80 CAC-, n = 37 CAC +; sputum IL6: n = 82 CAC-, n = 38 CAC +; plasma IL6 & TNF: n = 91 CAC-, n = 41 CAC +) (d). For c, d cough aerosol culture-positive and cough aerosol culture-negative individuals are depicted with red or black circles, respectively. Figures (a), (b), and (d) are Tukey-style box plot statistics with the median line, hinges indicating the 1st and 3rd quartiles, and whiskers extending to maximum and minimum values within 1.5x the interquartile range of the upper and lower hinges. The graphs in (c) display the linear model fit to the data with shaded error regions indicating the 95% confidence interval of the fit. Analyses shown in 1(a) panels reflect bivariate logistic regression described in Table 1, those shown in 1(b) and 1(d) panels used two-sided Wilcoxon rank sum test with continuity correction, and 1(c) panels reflect linear models fit to log10 transformed cytokine concentration. ELISA was performed with technical duplicates on the same sample with highly discordant samples removed from further analysis; results were averaged for each participant. Gray circle, unknown cough aerosol culture status. Source data are provided as a Source Data file. *p < 0.05, **p < 0.01, ***p < 0.001.

Table 4 Multivariable model of index characteristics associated with cough aerosol culture positivity

Characteristics	OR	95% CI	p-value	aOR	95% CI	p-value	
MUAC,a cm	1.14	1.00−1.30	0.04	1.42	1.17–1.73	0.0004	
CRP level, mg/dL	1.02	1.01−1.03	< 0.001	1.02	1.01–1.03	0.001	
Xpert Ct	0.73	0.63−0.84	< 0.001	0.76	0.64–0.90	0.002	
Age, years	0.95	0.92−0.99	0.006	0.94	0.90–0.99	0.02	
TTDb, days	0.75	0.63−0.88	< 0.001	0.80	0.64–1.00	0.05	
Female	0.50	0.21−1.21	0.12	–	–	–	
BMI	1.02	0.92–1.14	0.69	–	–	–	
Sputum smear grade	1.62	1.16–2.26	0.005	–	–	–	
Xpert semi-quantitative grade	2.12	1.35–3.22	0.001	–	–	–	
PLWHc	0.74	0.23–2.45	0.88	–	–	–	
Fever	3.15	1.32−7.49	0.10	–	–	–	
Cough Duration	1.03	1.00−1.07	0.10	–	–	–	
LCQd Score	0.81	0.71−0.92	0.001	–	–	–	
CPFe	1.00	1.00–1.00	0.23	–	–	–	
Cavitary Chest X-ray	4.73	1.71−13.08	0.003	–	–	–	
Chest x-ray quadrants	1.50	1.02−2.20	0.04	–	–	–	
WBC	1.19	1.03−1.36	0.02	–	–	–	
aMUAC, Mid-Upper Arm Circumference; bTTD, Time to detection of Mtb growth in liquid media; cPLWH, Person living with HIV; dLCQ, Leicester Cough Questionnaire; eCPF, Cough peak flow

Predictors were selected based on our conceptual model (see Supplementary Fig. 1), tested in forward selection stepwise multivariable models, and compared with the likelihood ratio test. The best-performing model with adjusted odds ratios is shown in the final three columns. No adjustments were made for multiple comparisons. Source data are provided as a Source Data file.

Performance of clinical prediction rule to identify cough aerosol culture-positive persons

We developed and evaluated the performance of a clinical prediction rule to predict cough aerosol culture-positive persons as a means to identify those who are likely to be highly infectious. We included predictors from our multivariable logistic regression model that are readily available to a clinician during the initial visit. For this reason, we removed the time-to-detection of culture growth but kept the CRP level as there are point-of-care versions of this test37. We also used Xpert Ultra semi-quantitative grade, which is the result generally reported to clinicians and is derived from the Ct value. Using the coefficients obtained in the multivariable logistic regression analysis, we derived the following prediction equation for cough aerosol culture-positive status where outcomes were coded according to weighted scores based on the beta coefficient: CRP > 42.9 + Age < 43.8 years + MUAC > 22.8 cm + Xpert semi-quantitative grade (high, medium, or low/very low/trace). A total risk score is calculated by adding the risk points and has an optimum cut point of 15 points for predicting cough aerosol culture-positive status with an estimated sensitivity of 0.86 (95% CI, 0.74−0.95) and specificity of 0.65 (95% CI, 0.56−0.74). (Supplementary Tables 2 and 3) The prediction rule based on the calculated risk score had an area under the receiver operating curve (AUROC) of 0.84 (95% CI: 0.77–0.91) (Supplementary Fig. 3). The model’s Somers’ Dxy index was 0.71; the equivalent in bootstrap validation was 0.66, with an optimism estimate of 0.051, indicating good stability of the model in internal validation.

Sputum and plasma cytokine analysis of associations with cough aerosol culture positivity, bacterial load, and cavitary lung disease

To further examine the association of inflammatory markers with cough aerosol culture-positive status, we analyzed sputum and plasma levels of CXCL8 (IL-8), IL-1β, TNF, and IL-6 (Fig. 1b and Supplementary Table 4). TNF in sputum and IL-1β and CXCL8 in plasma were nearly undetectable in preliminary subgroup testing and were not examined further. The sputum concentration of CXCL8 was significantly higher (p-value 0.03) among the cough aerosol culture-positive compared to cough aerosol culture-negative participants. Sputum IL-1β (p-value 0.052) and IL-6 (p-value 0.36) were not different by CAC status (Fig. 1b). Higher IL-6, CXCL8, and IL-1β levels were associated with greater bacillary burden (measured by GeneXpert cycle threshold, p-value 0.0009, 0.007 and 0.00007, respectively, Fig. 1c). IL-1β, but not IL-6 or CXCL8, was positively associated with cavitary lung disease (p-value 0.0007) across cough aerosol culture status.

Plasma IL-6 was associated with bacillary burden (p = 1.69 × 10−6) and cavitary disease (p = 0.005) and was higher in CAC-positive compared to CAC-negative participants (p = 0.03). Plasma TNF was associated with cavitary disease (p = 0.013), but not bacillary burden, and was lower in CAC-positive compared to CAC-negative participants (p-value 0.0007). For cytokines that had a significant association with cough aerosol status in bivariate models (sputum CXCL8, plasma TNF, plasma IL-6), we evaluated associations in multivariate logistic regression models in which we adjusted for age, sex, cavitary disease on chest x-ray and GeneXpert Ct value. We found that plasma TNF was independently associated with CAC status (adjusted p-value 0.007, Supplementary Table 4).

Whole blood transcriptomic signatures and analysis of associations with cough aerosol culture positivity and sputum bacterial load

We selected 58 subjects for whole blood transcriptomic analysis which were a subgroup of the full cohort of 142. We initiated this experiment after 29 CAC + and 29 CAC- subjects were available for analysis from the initial phase of enrollment. We found no differences in clinical and biologic variables (age, sex, HIV status, Xpert Ct value, cavitary CXR, and degree of lung involvement (CXR quadrants)) when comparing participants included in the transcriptomic subgroups (CAC + and CAC-) with the remaining cohort (Supplementary Tables 6 and 7). We used a genome-wide approach to determine whether a specific host inflammatory signature was associated with cough aerosol culture-positive status independent of other diseases. We measured whole blood RNA-seq profiles from a pre-treatment sample of persons with pulmonary TB and found differentially expressed genes associated with cough aerosol culture-positive status (100 genes with FDR < 0.2). While a strong majority of the genes (78%) were best fit by models including cough aerosol culture status alone, 16% of genes were best fit with the inclusion of bacillary load with a smaller percentage best explained by more complex models with age and cavitary disease (Supplementary Fig. 4a). After a stepwise evaluation of covariates and assessment of model fit (based on improvement of > 10% of genes), bacillary load (measured by Ct value) was the only covariate included in the final model. We examined covariate-adjusted expression values and found that no differentially expressed genes (DEGs) were independently associated with cough aerosol culture positivity (Fig. 2a, b and Supplementary Fig. 2c). In contrast, many DEGs remained associated with bacterial burden (1129 genes with FDR < 0.2; 40 genes with FDR < 0.05; min FDR = 0.0008, Supplementary Table 7). Based on these findings, DEGs are associated with several features of TB presentation but are not independently associated with cough aerosol culture-positive status after adjusting for bacillary burden.Fig. 2 Whole blood transcriptional signatures and association with cough aerosol culture positivity and bacillary burden.

a Differential gene expression associated with cough aerosol culture positivity and bacillary load. The volcano plot shows the gene-wise covariate-adjusted log2 fold change associated with each effect as well as each gene’s Benjamini-Hochberg adjusted false discovery rate (FDR). b Distributions of unadjusted p-values (x-axis) and FDR (y-axis) for each effect. c Gene set enrichment of Hallmark pathways in differential gene expression of cough aerosol culture positivity (red) and bacillary load (yellow). Displayed at left are all pathways for which any effect showed enrichment at an FDR threshold of 0.2. For each effect, pathways with an enrichment FDR < 0.2 are indicated with solid points while open points reflect FDR > = 0.2; diamond-shaped points locate two pathways uniquely enriched in the expression signal associated with cough aerosol culture positivity. Pathways are organized by their maximum enrichment score for any of the two effects. The Venn diagram at right displays the number of pathways enriched in each effect at FDR < 0.2 with stratification by concordance of directionality of effect. d Leading edge genes driving enrichment of Hallmark pathways uniquely associated with cough aerosol culture-positive status. Points indicate the adjusted fold change associated with cough aerosol culture positivity (y position is arbitrary jitter), are colored according to the sign of FC, and filled/labeled points indicate leading edge genes in the GSEA analysis. Gray violins indicate the distribution of FC values within each pathway, and a rug plot along the x-axis indicates the full distribution of FCs estimates for the cough aerosol culture-positive effect. Black curves indicate the running enrichment score deviation from zero along the y-axis. Source data are provided as a Source Data file.

We next used Gene Set Enrichment Analysis (GSEA) to examine whether transcriptional signatures were associated with cough aerosol culture-positive status. Using an adjusted model with a bacillary burden as a covariate and assessing for concordant directionality of effect, several gene sets were associated with bacillary load or cough aerosol culture-positive status and represented a diverse set of cellular processes (Fig. 2c). Although there was overlap among cough aerosol culture-positive and bacillary burden associations, the majority of gene sets were associated with concordant directionality with either, but not both, traits (Fig. 2c). We found eight Hallmark gene sets that were associated either exclusively with cough aerosol culture-positive status, or with opposite directionality as bacillary burden. Hallmark_Angiogenesis was associated with cough aerosol culture-positive status, while seven gene sets were associated with cough aerosol culture-negative status (Fig. 2c and Supplementary Fig. 5). There were 10 leading-edge genes in the Angiogenesis gene set with functional associations with cellular proliferation, ligand:receptor interactions, and the extracellular matrix. (Fig. 2d and Supplementary Table 8).

Discussion

The primary findings of TBAIT suggest that host inflammatory signatures are associated with Mtb aerosolization independent of bacillary load and cavitary lung disease. We extend previous studies that determined several non-immunologic host factors that are associated with cough aerosol culture status25,29. We found that higher serum CRP levels, sputum and plasma cytokines, and whole blood transcriptional signatures were associated with Mtb aerosolization. These data highlight potential insights into the biology of Mtb transmission events as well as biomarkers to identify highly infectious individuals. Based on our findings and prior studies, cough aerosol cultures are superior to sputum smear analysis in predicting Mtb transmission events and are likely the best estimators of TB infectiousness currently24,26–28. Although cough aerosol culture-negative patients may transmit Mtb, cough aerosol cultures allow for a more accurate assessment of relative infectiousness than traditional measures of sputum bacillary load. Prior studies found that bacillary burden, mucoid sputum, stronger cough, and higher Karnofsky performance score were associated with cough aerosol culture positivity25,29.

Theron et al. in the largest cough aerosol culture study of TB patients to date29, found that higher peak cough flow rate, higher bacillary load, lack of HIV infection, and lower “TB symptom score” were independent risk factors for cough aerosol culture positivity. In the Theron study, a lower TB symptom score indicated a lower burden of findings attributable to TB disease as it is a summation of points for TB-related symptoms (cough, hemoptysis, dyspnea, chest pain, fever, night sweats), TB-related signs (anemia, tachycardia), lung auscultation findings, and malnutrition (low BMI, low MUAC). Our study adds the association between cough aerosol culture-positive status and higher serum CRP levels, an acute phase reactant that is produced in the liver and is a non-specific marker of inflammation. CRP has been extensively evaluated as a triage test for TB, particularly among PLWH38, is endorsed by WHO as a TB screen, and is available as a point-of-care test37. We also identified an independent association between lower plasma TNF levels and culture-positive cough aerosols. Our finding of an association between cough aerosol status and MUAC, but not BMI, may seem counterintuitive and may represent a statistical artifact. Alternatively, we hypothesize that MUAC and BMI measure related but different aspects of nutritional status: BMI reflects fat mass while MUAC measures fat and muscle mass.39. Low MUAC was a better predictor than BMI of all-cause mortality in older adults39–41, and in persons with TB, MUAC (unlike BMI) was independently associated with cavitary lung disease42, Cachexia, severe weight loss that results from muscle atrophy and the loss of adipose tissue, may correlate with differential immune cell responses43. For example, using parasitic infections in the murine model, mice deficient for CD4 + T cells experienced muscle and fat wasting as opposed to CD8 + T cell-deficient mice, which experienced only fat wasting44.

Taken together, these findings suggest common features of a cough aerosol culture-positive phenotype: younger persons with few symptoms (aside from cough), preserved muscle mass, a higher bacillary burden, and higher systemic inflammation. If accurate, the proposed phenotype of highly infectious persons with TB would likely describe individuals who are younger16–18,29, and active rather than moribund25,29, increasing the opportunities for TB transmission events. We developed a clinical prediction tool to identify highly infectious persons with TB (cough aerosol culture positive) that could have utility in TB control interventions, for example, to inform contact investigations and TB preventive therapy administration, and to improve precision in clinical trials of TB preventive therapy and vaccines. Like other clinical prediction rules, the model performance is predicated on having data for all of the characteristics that contribute to the risk score. While promising, the risk score that we present requires external validation and should not be used at this time for clinical (non-research) purposes.

Potential mechanisms underlying increased Mtb aerosolization and transmission include high sputum bacillary loads, Mtb strain features, cough characteristics (e.g., propulsive strength or frequency), and host inflammation. A balance in the host inflammatory response is needed to prevent the dissemination of Mtb (resulting in part from insufficient inflammation) while limiting tissue destruction and other complications from excessively robust inflammatory responses45,46. The stimulation of inflammatory pathways that regulate Mtb aerosolization and transmission may exert additional evolutionary pressure on the immune response to Mtb. To explore potential mechanisms of Mtb aerosolization, we discovered a whole blood transcriptomic signature associated with cough aerosol culture -positivity after adjustments for bacillary load. Given the timing of sample collection at the time of diagnosis, we cannot distinguish whether index case inflammatory pathways cause Mtb aerosolization or vice versa. If the inflammatory pathways associated with Mtb aerosolization are regulated by immunogenetically encoded mechanisms, then the Angiogenesis signaling pathway, which was enriched in cough aerosol culture-positive participants, may offer causal insights.

Of the 10 leading edge genes differentiating cough aerosol culture-positive versus -negative status, several are involved in the extracellular matrix (COL5A2, FGFR1, ITGAV) and cell development and proliferation (JAG1, PDGFA, FSTL1) pathways which offer potential insights into aerosolization mechanisms. For example, extracellular molecules secreted by these pathways could provide an alternative milieu surrounding the bacterium, which modulates the surface structure or metabolic state of Mtb and its subsequent capacity to transmit. Previous work in the M. marinum zebrafish model uncovered an important role for vascularization and angiogenesis in granuloma formation and control of bacterial dissemination47,48. Enzymatically modified trehalose dimycolate (TDM), an immunogenic cell wall lipid, induced angiogenesis through a VEGF pathway. These studies highlight potential mechanisms that could influence Mtb aerosolization via modulation of angiogenesis-dependent inflammatory pathways. Similarly, cytokine pathways (e.g., Interferon Alpha Response and Interferon Gamma Response gene sets enriched in cough aerosol culture-negative individuals) from activated immune cells could modulate Mtb’s state and survival before, during, and after propulsion from the host airway. Regardless of the causal pathway, the inflammatory profiles provide a potential biomarker of infectiousness that could rapidly identify Mtb superspreaders.

Our study supports prior findings of significant individual host variation in infectiousness among patients with TB and suggests that factors beyond bacillary burden determine infectiousness19,21,22,25,26,49. While CASS, modeled on cough aerosol production, is the best-studied method for assessing infectiousness50, recent studies have drawn attention to detectable Mtb from non-cough respiratory maneuvers32–34,51. Williams and colleagues demonstrated that when persons with confirmed and suspected TB wore face masks with collection strips (face mask sampling or FMS) while breathing normally, Mtb DNA was detected in up to 90% of persons33. The authors subsequently found a modest association between FMS detection of Mtb and increased TB infection in close contacts52. RASC is a device to capture bioaerosols that are then evaluated for viable Mtb using a fluorescent trehalose analog32,34. In a study of 38 GeneXpert-positive participants who performed respiratory maneuvers (tidal breathing, voluntary cough, forced vital capacity) while seated in the RASC, 88% of participants produced at least one sample positive for Mtb, and all three maneuvers were equally likely to produce viable Mtb34. Further investigations are needed to determine differences in findings from CASS, FMS, and RASC, especially as they relate to actual transmission events.

Our study has several limitations. First, we did not evaluate whether Mtb microbiologic factors are associated with cough aerosol culture status. A previous investigation of Mtb genetic variants and cough aerosol culture positivity did not demonstrate associations29. Second, 95% of participants reported cough, and our findings may not apply to persons with TB without cough. Recent studies that have detected TB in exhaled breath call into question whether cough is essential and/or a primary driver of TB transmission.33,34,52,53. Third, our transcriptomic data derives from whole blood which precludes cell-specific insights. Despite this limitation, pathway analysis suggests possible cellular sources of mechanisms that can be tested in future studies. Fourth, we were unable to assess the causality of immunologic pathways associated with cough aerosol culture positivity which may precede Mtb aerosolization or be a consequence of it. However, we did evaluate for possible confounding and adjusted for bacillary load to identify gene sets that are independently associated with cough aerosol culture status. Fifth, sputum collection methods are not standardized and have more technical heterogeneity in comparison to other biological samples. Despite this challenge, sputum provides a direct assessment of the primary site of TB disease, where pulmonary immune responses are compartmentalized and differ from blood. In addition, sputum cytokines have been evaluated as biomarkers in the diagnosis of TB and treatment monitoring54–56. However, to our knowledge, no prior studies have evaluated their association with infectiousness measured by culturable aerosols. In addition, the index participants who contributed to our evaluations of risk factors for HHC IGRA results differed from index participants without enrolled HHCs, which introduces selection bias and limits generalizability. Although our random effects models should address concerns around selection bias57, the generalizability of the HHC IGRA status remains a limitation. Finally, we enrolled participants during the COVID-19 pandemic, and the impact of SARC-CoV-2 on CRP level, transcriptional profiles, or other study variables is not known.

There were also several strengths to our study. First, all participants underwent CASS procedures prior to the initiation of anti-tuberculosis therapy, which is known to rapidly impact cough aerosol culture results29,58. Second, we adhered to a rigorous definition of cough aerosol culture status in which we excluded participants who had more than two CASS plates contaminated with overgrowth. Third, we demonstrated that cough aerosol culture-positive status was strongly associated with evidence of TB transmission in household contacts based on QFT results. The annual risk of TB infection differs by age group in high-burden settings59, and notably, there was no difference in the ages of HHCs by CAC status. Among household contacts less than 10 years of age, who have a lower annual risk of infection compared to adolescents and adults, QFT-positive results were more common (p-value 0.01) among contacts of CAC-positive persons (9 of 10 participants, 90%) than contacts of CAC-negative persons (14 of 33 participants, 42%).

Our findings, along with those from prior studies, suggest that host characteristics and biomarkers may identify the most infectious patients with TB. CRP, already recommended as a screening test for TB by the World Health Organization, may have a role in identifying highly infectious persons with TB60,61. While persons with pulmonary TB who are cough aerosol culture-negative patients may transmit Mtb, identifying the most infectious persons would allow targeted interventions to support TB control efforts, such as isolation, true direct observation of treatment, and drug-susceptibility testing to confirm that treatment is effective. TB control could also be supported through enhanced investigations to identify contacts of the most infectious persons to evaluate for active TB and provide TB preventive therapy given the higher likelihood of recent transmission and progression to disease. Further elucidation of inflammatory transcriptional signatures associated with TB infectiousness is important not only for understanding mechanisms and developing new therapies but also for the possibility of developing a diagnostic tool to identify individuals who are the most infectious.

Methods

Ethical approvals

This study complies with all relevant ethical regulations and was approved by the Kenya Medical Research Institute Scientific and Ethics Review Unit (048/3988), the Kenyatta National Hospital/University of Nairobi Institutional Review Board (KNH-ERC/A/375), and the University of Washington Institutional Review Board (STUDY00009209). All participants 18 years and older provided informed consent. For participants less than 18 years, informed consent was obtained from the legal guardian and participants ages 13 to 18 years also provided informed assent. Participants were compensated the equivalent of $4 per study visit for their time.

Study design

Study setting & participants

Between March 1, 2021, and March 30, 2023, we enrolled adults with newly diagnosed pulmonary TB in

Nairobi, Kenya. The TB incidence rate in Kenya was estimated to be 558 per 100,000 adults in 201562. Participants were enrolled either through outpatient TB and respiratory clinics where they had presented for healthcare (passive case finding) as a convenience sample or were diagnosed with pulmonary TB through a Nairobi-based household TB prevalence survey (active case finding). Participants identified through active case finding underwent chest x-ray (regardless of symptoms) and cough assessment; those with an abnormal chest x-ray and/or who reported a current cough were asked to provide two sputum samples for AFB-smear, -culture, and GeneXpert testing. All participants were diagnosed with TB based on a positive GeneXpert test result, either GeneXpert MTB/RIF (Xpert MTB/RIF) or GeneXpert Ultra (Xpert Ultra). AFB culture was subsequently performed on sputum samples. A final diagnosis of pulmonary TB was based on a positive GeneXpert test unless the result was “trace positive” from the Xpert Ultra assay, in which case culture confirmation of Mtb was required. We did not include persons in this study identified through active case finding who were GeneXpert negative/culture-positive due to delays in their TB diagnosis, who were trace positive/culture-negative as we concluded that they did not have active pulmonary based on four negative sputum cultures, and those who were diagnosed on weekends or declined enrollment. Participants had not initiated anti-TB treatment at the time of enrollment and study interventions. Potential participants who were unable to consent in the study languages (Ki-Swahili, English), did not provide a home location, planned to move from the area within six months, declined study procedures, or were currently imprisoned were not eligible for the study. We also enrolled the household contacts of participants with pulmonary TB to assess for evidence of TB transmission events. Household contacts were eligible for enrollment if they resided and slept in the household for at least 60 days prior to enrollment. There were no age restrictions on the eligibility of household contacts. Participants’ sex and race were recorded based on self-identification. Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Washington63.

We estimated the frequency of positive IGRAs in Nairobi residents who were not known to have active TB or had recent close contact with a person with active pulmonary TB. Participants were enrolled in two studies. As part of a study of cough analysis for TB detection64, we recruited adult outpatients (n = 45) from the same TB and respiratory clinics as TBAIT participants with TB (described above) who were identified through passive case finding and presented with cough. All participants underwent sputum GeneXpert testing and chest X-rays to exclude pulmonary TB. In addition, we report results from a study to evaluate treatment responses during isoniazid preventive therapy65. From December 2019 to December 2020, we recruited adults with (n = 121) and without (n = 122) HIV from three HIV prevention and care centers in Nairobi. We assessed for active TB using the WHO four-symptom screen, followed by sputum GeneXpert and chest X-rays among those with one or more symptoms. To calculate the prevalence of QFT positive tests, we excluded indeterminate results.

Study procedures

Participants with TB

We collected sputa from participants at enrollment (“spot”) and the following morning (“morning”). AFB-smear and culture were performed on both samples, and GeneXpert testing was performed on the spot sample. We collected separate sputum for cytokine testing in addition to whole blood for laboratory testing and mRNA analysis.

CASS consists of a six-stage Andersen Cascade Impactor (Thermo Fischer Scientific, Rockford, IL) within a larger stainless-steel chamber attached to the tubing and with a vacuum pump creating an airflow through the system23,25. We calibrated the vacuum pump flow rate (28.3 L/min) using a primary flow calibrator (Model 4046, TSI, Inc., Shoreview, MN) and then maintained and monitored that using a marked field rotameter (SKC, Inc., Eighty Four, PA). Each of the six stages of the Andersen cascade impactor holds a Middlebrook 7H10 or 7H11 solid agar plate, on which aerosolized particles impact based on particle size. The agar plates were loaded into the Andersen impactors in a sterile fashion; the impactors were loaded into the cylinder before each study. Prior to cough peak flow measurements and sputum collection, participants coughed into a mouthpiece on tubing connected to the chamber for 5 min. Plates were incubated at 37 °C for 8 weeks and observed weekly for growth. When growth was identified as Mtb, colony-forming units (CFU) were counted. Agar plates with non-acid fast bacilli growth were considered contaminated and discarded. The cylinder and components were autoclaved after each use. Mtb growth was first detected at a median of 4 weeks (interquartile range (IQR), 4–6) and was most frequently detected on plates four (22 with growth), five29, and six19, corresponding to particle sizes of 2.1–3.3 µm, 1.1–2.0 µm, and 0.65–1.1 µm, respectively23. (Supplementary Fig. 1) Among cough aerosol culture participants, the maximal plate CFUs, defined for each participant as the highest CFU count for any plate with Mtb growth, ranged from 1 to 76 with a median of 12 CFUs (IQR, 4–27).

For cough peak flow (CPF) measurements, we instructed participants to cough as forcefully as possible through a disposable mouthpiece attached to a Vitalograph peak flow meter (Ennis, Ireland). The procedure was performed three times, and results were recorded in L/min. The highest recorded value was used in analyses. Enrollment posteroanterior (PA) chest x-rays were obtained and interpreted by a member of the study team (DJH, a pulmonologist) for the presence of cavitations and number of quadrants with changes attributed to TB disease; DJH was blinded to CASS results in the review of chest x-rays. Study personnel interviewed participants to collect demographic information, HIV history, and TB history. To assess cough-related effects on quality of life, we administered the Leicester Cough Questionnaire (LCQ), a validated health status measure for adults with chronic cough ranging from 3–21 with a lower score indicating greater impairment66.

Household contacts

Consenting household contacts were interviewed to assess for symptoms consistent with TB and to collect demographic information and exposure history. Household contacts underwent chest x-ray (PA unless age < 10 years, in which case PA and lateral images were obtained) and phlebotomy for interferon-gamma release assay and HIV testing. Household contacts who were suspected of having TB based on an abnormal chest x-ray and/or the presence of symptoms (cough, fever, weight loss) had sputum collected for GeneXpert testing.

Laboratory assays

Sputum volume and quality were visually assessed by laboratory personnel. We performed concentrated sputum smears on all samples using Auramine O staining with fluorescence smear microscopy, read according to the WHO AFB scale. Sputum was processed using the NALC-NaOH-NaCitrate method, and 0.5 ml of the pellet inoculated in mycobacterial culture (MGIT) Mycobacterium Growth Indicator 4 mL tubes (Becton-Dickinson, Franklin Lakes, NJ) supplemented with BD BBLTM MGITTM PANTA and incubated in an automated BACTEC MGIT 960 machine for growth determination. Samples that had zero(0) growth units at 42 days were confirmed negative. Broth cultures that were flagged as positive by the MGIT 960 had the time to detection recorded, and the presence of acid-fast bacilli was verified using Ziehl-Neelsen smear microscopy with isolates identified as Mtb using the MGIT TBc Identification Test (Becton-Dickinson Diagnostic Instrument Systems, Sparks, MD). Further speciation was not performed. Xpert MTB/RIF or Xpert Ultra (Cepheid, Sunnyvale, CA) were performed on raw sputum samples. GeneXpert results were recorded as a cycle threshold (Ct) value and as a semi-quantitative grade: Negative, Very Low, Low, Medium, and High, with an additional category of “Trace” for Xpert Ultra. Results from GeneXpert testing for rifampin resistance were recorded as negative, positive, or indeterminate. For the CASS culture plates, after the preparation and autoclaving of 7H10 or 7H11 solid media, appropriate volumes of OADC (Oleic acid Albumin Dextrose Catalase) and antibiotics (amphotericin B, carbenicillin, polymixin B, and trimethoprim lactate) were added to inhibit contaminant growth.

C-reactive protein (CRP) was measured from serum samples using a Cobas C 111 chemistry analyzer (Roche Diagnostics Ltd, Liechtenstein, Switzerland) according to the manufacturer’s instructions. The detection range was 0.6–350 mg/L. We offered HIV testing to participants unless a participant was known to be living with HIV or if testing had been performed within the six months prior to enrollment. Whole blood for QuantiFERON-TB Plus testing (QFT-Plus, Qiagen Diagnostics; Hamburg, Germany) was drawn into lithium heparin blood collection tubes, and 1 mL amounts were transferred into Nil, Mitogen, TB1, and TB2 tubes. All QFT-Plus tubes were incubated at 37 °C within 6 h of collection. QFT-Plus processing and interpretation were performed according to the manufacturer’s instructions67. If TB1-nil and/or TB2-nil were > 0.35 IU/ml and > 25% of nil value (with Nil < 8.0 IU/ml), then the QFT-Plus test was considered positive. If Nil > 8.0 IU/mL or Mitogen–Nil < 0.5 IU/ml with Nil < 8.0 IU/ml and negative antigen-Nil results, then QFT-Plus was defined as indeterminate. QFT-Plus results were negative if Nil < 8.0 IU/ml with either antigen-Nil values < 0.35 IU/ml IFN-γ or < 25% of Nil). Indeterminate results were repeated, with the second result reported.

Sputum and plasma cytokines

Four cytokines were examined in sputum and blood. TNF, IL6, IL1B, and CXCL8 were chosen based on their central roles in regulating inflammatory pathways and TB pathogenesis68. Spot sputum was digested by adding an equal volume of 10% Sputolysin (Millipore, Merck KGaA, Darmstadt, Germany). The mixture was vortexed and incubated at 37 °C for 15 min. The digested sample was centrifuged at 500 × g for 10 min. The supernatant was preserved by the addition of 40 µl of 25X cOmpleteTM protease inhibitor cocktail solution (Roche, Merck KGaA, Darmstadt, Germany) per 1 ml of sample and stored at − 80 °C. Plasma was tested directly. Sputum supernatants and plasma were tested for CXCL8 (IL-8), Interleukin 1 beta (IL-1β), Interleukin 6 (1L-6), and TNF using sandwich ELISA according to the manufacturer’s instructions. (R&D Systems Inc. Minneapolis, USA.). In pilot sputum cytokine studies, TNF was not detectable at high levels and was not examined further. In pilot plasma cytokine studies, CXCL8 and IL1B were not detectable at high levels and were not examined further.

Whole blood transcriptomics: RNA isolation, RNASeq, data processing and analysis

PAXgene tubes were thawed at room temperature, and RNA was isolated using PAXgene miRNA spin columns (Qiagen), followed by globin reduction using GlobinClear Human (ThermoFisher). To generate sequencing libraries, total RNA (0.5 ng) was added to the reaction buffer from the SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing (Takara), and reverse transcription was performed, followed by PCR amplification to generate full-length amplified cDNA. Sequencing libraries were constructed using the NexteraXT DNA sample preparation kit (Illumina) to generate Illumina-compatible barcoded libraries. Libraries were pooled and quantified using a Qubit® Fluorometer (Life Technologies). Sequencing of pooled libraries was carried out on a NextSeq 2000 sequencer (Illumina) with paired-end 59-base reads, using a NextSeq P3 sequencing kit (Illumina) with a target depth of 5 million reads per sample. Base calls were processed to FASTQs on BaseSpace (Illumina), and a base call quality-trimming step was applied to remove low-confidence base calls from the ends of reads. The FASTQs were aligned to the GRCh38 human reference genome using STAR v.2.4.2a, and gene counts were generated using htseq-count. QC and metrics analysis was performed using the Picard family of tools (v1.134). Counts were assigned to gene exons using RSEM 1.3.0. Further RNA sequencing data filtering and analysis were performed in R 4.2.369. We removed libraries with fewer than 1,000,000 total reads or a median coefficient of variation of coverage (median CV) greater than 0.8; this resulted in the removal of 2 sequencing libraries, both of which were clear outliers by these metrics. Libraries were reduced to protein-coding genes, and principal component analysis (PCA) did not detect any strong outliers based on RNA composition. Protein-coding counts were normalized using the trimmed mean of M-values normalization method, filtered to protein-coding genes with at least 5% of libraries containing at least 1 count per million (CPM), and finally, converted to log2 CPM using the voom-normalization procedure within the R package “limma”70.

Data analysis

Host characteristics and household contact analyses

Our primary outcome of interest was cough aerosol culture status. Participants were considered cough aerosol culture-positive if one or more of the six CASS plates were positive for Mtb growth. Participants were considered cough aerosol culture-negative if no CASS plates were positive for Mtb growth and no more than two of the six plates were discarded as contaminated with fungal or bacterial overgrowth. Participant characteristics were compared using the chi-square test, Student’s t test, or Wilcoxon rank-sum test as appropriate. We assessed associations between predictors and outcomes using bivariate and multivariable logistic regression. Multivariable models were developed using forward selection stepwise regression, evaluating variables with p-values ≤ 0.20 in bivariate analyses and retaining covariates which significantly improved model fit based on likelihood-ratio tests. We assessed multicollinearity between our independent variables using variance inflation factors and condition indices. All statistical tests were two-sided with α = 0.05.

We assessed associations between host TB characteristics and cough aerosol culture-positive status using bivariate and multivariable logistic regression models in which we evaluated predictor variables based on our conceptual model presented as a directed acyclic graph (DAG, Supplementary Fig. 2)71. Based on published studies of TB transmission and cough aerosol status18,25,29, we categorized as predictors characteristics that we hypothesized as determining the cough aerosol culture-positive phenotype including younger age, functional status (body mass index (BMI), mid-upper arm circumference (MUAC), TB symptoms, cough duration, Leicester Cough Questionnaire (LCQ) score, prior history of TB, and cough peak flow)29 and systemic inflammation (CRP, hemoglobin, white blood cell count and differential, hemoglobin A1C percent, and sputum appearance). In the DAG, we categorized HIV status, bacillary burden (GeneXpert semi-quantitative grade, GeneXpert Ct value, AFB-smear grade, time to detection (TTD) of Mtb growth in liquid culture, presence of chest X-ray cavitation, and number of radiographic quadrants with TB-related changes), and sex as potential confounders due to their shared associations with the exposure and outcome72. Healthcare engagement is considered a mediator, a node that is on the causal pathway from exposure to outcome, and should not be controlled in models. Pathogen-related factors, which are not included in this study, would be considered effect modifiers as they may contribute to the outcome and modify the effect of other causes of the outcome. GeneXpert Ct values were assigned using the smallest Ct value from any of the probes targeting the rpoB gene; participants with Xpert Ultra trace positive results (for whom rpoB probe Ct values were 0) were assigned a Ct value of 35, near the highest detectable Ct value for Xpert Ultra.

To evaluate associations between cough aerosol culture status and evidence of TB transmission, we investigated QFT-Plus responses in household contacts by cough aerosol culture status of the index participant. QFT-Plus responses were dichotomized as positive (≥ 0.35 IU/mL) or negative (< 0.35 IU/mL) after excluding indeterminate responses. We also evaluated the absolute interferon-gamma response based on the larger value of either TB antigen 1 – nil or TB antigen 2 – nil. To evaluate associations with QuantiFERON positive results in household contacts, we used bivariate and multivariable logistic regression models with random intercepts (melogit command in Stata) to account for clustering by index participant and compared multivariable models with the likelihood-ratio test. and R: A Language and Environment for Statistical Computing.

Risk score modeling

Analyses were performed to develop a risk score for clinical decision-making to identify highly infectious (cough aerosol culture-positive) persons with pulmonary TB. We evaluated covariates from our multivariable logistic regression model that could be applied at the time of a patient’s diagnosis with pulmonary TB. Included variables were categorized, and the optimal cut-points were determined using the Youden index (J) method, the point maximizing the Youden function, which is the difference between true positive rate and false positive rate over all possible cut-point values73. These values were then used in a multivariable logistic regression model. The risk score was generated by dividing the beta coefficients from the logistic regression model by the smallest beta in the model, multiplying by 5, and rounding to the nearest integer. Internal validation was conducted using the bootstrap resampling method with 5000 replications. We evaluated Somers’ Dxy index, a rank correlation between predicted probabilities and observed responses, to evaluate model performance74. Somers’ Dxy index takes values between − 1 and 1, with the latter demonstrating agreement between predicted and observed responses.

Cytokine analyses

Association of cytokine concentrations (log10 pg/ml) with cough aerosol culture status and cavitary disease was done by Wilcoxon rank sum test with continuity correction. Bivariate logistic regression models were used to compare cytokine concentration with cough aerosol culture status, GeneXpert cycle thresholds, and number of chest x-ray quadrants with changes attributed to TB disease. In multivariate analyses, we evaluated associations between the outcome of cough aerosol culture status and cytokine concentrations adjusting for age, sex, cavitations on chest x-ray, and GeneXpert cycle threshold.

RNASeq Analysis: estimation of differential expression

Fold changes in gene expression were estimated using linear models implemented in the R package “kimma”.75. In addition to the primary fixed effect of cough aerosol culture, model selection considered for inclusion of clinical covariates that were significantly associated with cough aerosol culture status in simple bivariate modeling, and which were complete for our RNASeq subset. We compared single covariate additions of bacillary load (GeneXpert Ct), age, and cavitary disease, assessing trends in model fit using the Akaike Information Criterion (AIC) across the genome (Supplementary Fig. 4a). We then performed stepwise model addition, comparing the impact on AIC at each step, and only retained covariates which significantly improved model fit (delta AIC < − 2) for at least 10% of genes (Supplementary Fig. 4B). This resulted in an expression model accounting for bacillary load in addition to the cough aerosol culture phenotype (Supplementary Fig. 4c, d).

RNASeq Analysis: gene set enrichment

We used a competitive gene set test to evaluate the enrichment of gene expression differences between cough aerosol culture status (positive vs. negative). We used gene set enrichment analysis (GSEA) using the pre-ranked approach employed in the fast GSEA (FGSEA) method and implemented in the R package “fgsea”76, and tested for enrichment of the Hallmark annotated biological pathways accessed via the mSigDB database77. Log2 Fold change estimates (log2 FC) for cough aerosol culture status, adjusted for the effect of bacillary load, were used to pre-rank genes for GSEA. Estimates for the effects of bacillary load were separately tested for enrichment in GSEA to identify enriched pathways unique to cough aerosol culture status. For each clinical predictor (cough aerosol culture status, bacillary load), we used kimma estimated log2FC to pre-rank genes and used gene-set permutation with 1000 permutations to randomize gene ordering75.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information

Peer Review File

Reporting Summary

Source data

Source data

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-52122-x.

Acknowledgements

This research was funded by the National Institutes of Health/NIAID (NIH grant 5R01AI150815 - DJH/TRH/VN and UH2AI152621 - DJH), NIH D43 TW011817-01 (VN/LNN/WB/DJH/TRH), and the Firland Foundation (TRH). The University of Washington/Fred Hutch Center for AIDS Research (AI027757) provided support to the J.B. REDCap database for data collection was supported by the National Center for Advancing Translational Sciences of the NIH (UL1 TR002319). K.F. was funded entirely by the NHLBI Division of Intramural Research. We would like to thank the participation of the individual study participants and their families. We thank Dr. Lucy Kijaro, Dr. Joy Githua, Dr. Jacqueline Mirera, Robi Chacha, Lenis Njagi, Geoffrey Onchiri, Ruth Munyasya, Caroline Epiche, Isaac Kibet, Stella Nthambi, Kevin Munge, Japherson Mecha, Patrick Isinidu, Joash Omolo, Hastings Koech, Inviolata Sakwa and all other KEMRI CRDR Nairobi staff for their support in data collection. We thank Madison Jones for laboratory assistance.

Author contributions

V.N., K.F., T.R.H., and D.J.H. designed the study. L.N.N., V.N., G.L., T.R.H., and D.J.H. led participant enrollment and study interventions. K.F. supervised CASS studies. W.M., Z.M., and G.P. led laboratory investigations under the supervision of T.R.H. and V.N. Statistical analyses were performed by D.J.H., R.M.S, and J.B. D.J.H. and T.R.H. drafted the manuscript, and all authors edited and reviewed the final manuscript. All collaborators of this study have fulfilled the criteria for authorship required by Nature Portfolio journals and have been included as authors, as their participation was essential for the design and implementation of the study. Roles and responsibilities were agreed among collaborators ahead of the research. This work includes findings that are locally relevant; this was determined in collaboration with local partners. We obtained local ethics reviews of all research in this paper. This research was not severely restricted or prohibited in the setting of the researchers and did not result in stigmatization, discrimination or personal risk to participants. When available, we included local and regional research relevant to our study in citations.

Peer review

Peer review information

Nature Communications thanks Mariana Araújo-Pereira, Delia Goletti, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The RNA-seq data generated in this study have been deposited in the dbGaP database (Home-dbGaP-NCBI(nih.gov)) under accession code phs003727.v1.p1. Source data are provided in this paper.

Competing interests

The authors declare no competing interests.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. World Health Organization. Global Tuberculosis Report 2017. Geneva, Switzerland: http://apps.who.int/iris/bitstream/10665/259366/1/9789241565516-eng.pdf?ua=1 (2017).
2. Yuen CM Turning off the tap: stopping tuberculosis transmission through active case-finding and prompt effective treatment Lancet 2015 386 2334 2343 10.1016/S0140-6736(15)00322-0 26515675
Yuen, C. M. et al. Turning off the tap: stopping tuberculosis transmission through active case-finding and prompt effective treatment. Lancet 386, 2334–2343 (2015).26515675 10.1016/S0140-6736(15)00322-0
3. Auld SC Research roadmap for tuberculosis transmission science: Where do we go from here and how will we know when we’re there? J. Infect. Dis. 2017 216 S662 S668 10.1093/infdis/jix353 29112744
Auld, S. C. et al. Research roadmap for tuberculosis transmission science: Where do we go from here and how will we know when we’re there? J. Infect. Dis. 216, S662–S668 (2017).29112744 10.1093/infdis/jix353
4. Woolhouse ME Heterogeneities in the transmission of infectious agents: implications for the design of control programs Proc. Natl. Acad. Sci. USA 1997 94 338 342 10.1073/pnas.94.1.338 8990210
Woolhouse, M. E. et al. Heterogeneities in the transmission of infectious agents: implications for the design of control programs. Proc. Natl. Acad. Sci. USA 94, 338–342 (1997).8990210 10.1073/pnas.94.1.338
5. Smith DL Dushoff J Snow RW Hay SI The entomological inoculation rate and Plasmodium falciparum infection in African children Nature 2005 438 492 495 10.1038/nature04024 16306991
Smith, D. L., Dushoff, J., Snow, R. W. & Hay, S. I. The entomological inoculation rate and Plasmodium falciparum infection in African children. Nature 438, 492–495 (2005).16306991 10.1038/nature04024
6. Teicher A Super-spreaders: a historical review Lancet Infect. Dis. 2023 23 e409 e417 10.1016/S1473-3099(23)00183-4 37352877
Teicher, A. Super-spreaders: a historical review. Lancet Infect. Dis. 23, e409–e417 (2023).37352877 10.1016/S1473-3099(23)00183-4
7. Ypma RJ Altes HK van Soolingen D Wallinga J van Ballegooijen WM A sign of superspreading in tuberculosis: highly skewed distribution of genotypic cluster sizes Epidemiology 2013 24 395 400 10.1097/EDE.0b013e3182878e19 23446314
Ypma, R. J., Altes, H. K., van Soolingen, D., Wallinga, J. & van Ballegooijen, W. M. A sign of superspreading in tuberculosis: highly skewed distribution of genotypic cluster sizes. Epidemiology 24, 395–400 (2013).23446314 10.1097/EDE.0b013e3182878e19
8. McCreesh N White RG An explanation for the low proportion of tuberculosis that results from transmission between household and known social contacts Sci. Rep. 2018 8 5382 10.1038/s41598-018-23797-2 29599463
McCreesh, N. & White, R. G. An explanation for the low proportion of tuberculosis that results from transmission between household and known social contacts. Sci. Rep. 8, 5382 (2018).29599463 10.1038/s41598-018-23797-2
9. Snider DE Jr. Kelly GD Cauthen GM Thompson NJ Kilburn JO Infection and disease among contacts of tuberculosis cases with drug-resistant and drug-susceptible bacilli Am. Rev. Respir. Dis. 1985 132 125 132 3925826
Snider, D. E. Jr., Kelly, G. D., Cauthen, G. M., Thompson, N. J. & Kilburn, J. O. Infection and disease among contacts of tuberculosis cases with drug-resistant and drug-susceptible bacilli. Am. Rev. Respir. Dis. 132, 125–132 (1985).3925826
10. Kline SE Hedemark LL Davies SF Outbreak of tuberculosis among regular patrons of a neighborhood bar N. Engl. J. Med. 1995 333 222 227 10.1056/NEJM199507273330404 7791838
Kline, S. E., Hedemark, L. L. & Davies, S. F. Outbreak of tuberculosis among regular patrons of a neighborhood bar. N. Engl. J. Med. 333, 222–227 (1995).7791838 10.1056/NEJM199507273330404
11. Lee RS Reemergence and amplification of tuberculosis in the Canadian arctic J. Infect. Dis. 2015 211 1905 1914 10.1093/infdis/jiv011 25576599
Lee, R. S. et al. Reemergence and amplification of tuberculosis in the Canadian arctic. J. Infect. Dis. 211, 1905–1914 (2015).25576599 10.1093/infdis/jiv011
12. Lloyd-Smith JO Schreiber SJ Kopp PE Getz WM Superspreading and the effect of individual variation on disease emergence Nature 2005 438 355 359 10.1038/nature04153 16292310
Lloyd-Smith, J. O., Schreiber, S. J., Kopp, P. E. & Getz, W. M. Superspreading and the effect of individual variation on disease emergence. Nature 438, 355–359 (2005).16292310 10.1038/nature04153
13. Melsew YA Risk factors for infectiousness of patients with tuberculosis: a systematic review and meta-analysis Epidemiol. Infect. 2018 146 345 353 10.1017/S0950268817003041 29338805
Melsew, Y. A. et al. Risk factors for infectiousness of patients with tuberculosis: a systematic review and meta-analysis. Epidemiol. Infect. 146, 345–353 (2018).29338805 10.1017/S0950268817003041
14. Turner RD Bothamley GH Cough and the transmission of tuberculosis J. Infect. Dis. 2015 211 1367 1372 10.1093/infdis/jiu625 25387581
Turner, R. D. & Bothamley, G. H. Cough and the transmission of tuberculosis. J. Infect. Dis. 211, 1367–1372 (2015).25387581 10.1093/infdis/jiu625
15. Donald PR Droplets, dust and guinea pigs: an historical review of tuberculosis transmission research, 1878-1940 Int J. Tuberc. Lung Dis. 2018 22 972 982 10.5588/ijtld.18.0173 30092861
Donald, P. R. et al. Droplets, dust and guinea pigs: an historical review of tuberculosis transmission research, 1878-1940. Int J. Tuberc. Lung Dis. 22, 972–982 (2018).30092861 10.5588/ijtld.18.0173
16. van Geuns HA Meijer J Styblo K Results of contact examination in Rotterdam, 1967-1969 Bull. Int Union Tuberc. 1975 50 107 121 1218286
van Geuns, H. A., Meijer, J. & Styblo, K. Results of contact examination in Rotterdam, 1967-1969. Bull. Int Union Tuberc. 50, 107–121 (1975).1218286
17. Zurcher K Estimating tuberculosis transmission risks in a primary care clinic in South Africa: modeling of environmental and clinical data J. Infect. Dis. 2022 225 1642 1652 10.1093/infdis/jiab534 35039860
Zurcher, K. et al. Estimating tuberculosis transmission risks in a primary care clinic in South Africa: modeling of environmental and clinical data. J. Infect. Dis. 225, 1642–1652 (2022).35039860 10.1093/infdis/jiab534
18. Borgdorff MW Nagelkerke NJ de Haas PE van Soolingen D Transmission of Mycobacterium tuberculosis depending on the age and sex of source cases Am. J. Epidemiol. 2001 154 934 943 10.1093/aje/154.10.934 11700248
Borgdorff, M. W., Nagelkerke, N. J., de Haas, P. E. & van Soolingen, D. Transmission of Mycobacterium tuberculosis depending on the age and sex of source cases. Am. J. Epidemiol. 154, 934–943 (2001).11700248 10.1093/aje/154.10.934
19. Riley RL Wells WF Mills CC Nyka W McLean RL Air hygiene in tuberculosis: quantitative studies of infectivity and control in a pilot ward Am. Rev. Tuberc. 1957 75 420 431 13403171
Riley, R. L., Wells, W. F., Mills, C. C., Nyka, W. & McLean, R. L. Air hygiene in tuberculosis: quantitative studies of infectivity and control in a pilot ward. Am. Rev. Tuberc. 75, 420–431 (1957).13403171
20. Sultan L Tuberculosis disseminators. A study of the variability of aerial infectivity of tuberculous patients Am. Rev. Respir. Dis. 1960 82 358 369 13835667
Sultan, L. et al. Tuberculosis disseminators. A study of the variability of aerial infectivity of tuberculous patients. Am. Rev. Respir. Dis. 82, 358–369 (1960).13835667
21. Riley RL Aerial dissemination of pulmonary tuberculosis. A two-year study of contagion in a tuberculosis ward. 1959 Am. J. Epidemiol. 1995 142 3 14 10.1093/oxfordjournals.aje.a117542 7785671
Riley, R. L. et al. Aerial dissemination of pulmonary tuberculosis. A two-year study of contagion in a tuberculosis ward. 1959. Am. J. Epidemiol. 142, 3–14 (1995).7785671 10.1093/oxfordjournals.aje.a117542
22. Escombe AR The infectiousness of tuberculosis patients coinfected with HIV PLoS Med. 2008 5 e188 10.1371/journal.pmed.0050188 18798687
Escombe, A. R. et al. The infectiousness of tuberculosis patients coinfected with HIV. PLoS Med. 5, e188 (2008).18798687 10.1371/journal.pmed.0050188
23. Fennelly KP Cough-generated aerosols of Mycobacterium tuberculosis: a new method to study infectiousness Am. J. Respir. Crit. Care Med. 2004 169 604 609 10.1164/rccm.200308-1101OC 14656754
Fennelly, K. P. et al. Cough-generated aerosols of Mycobacterium tuberculosis: a new method to study infectiousness. Am. J. Respir. Crit. Care Med. 169, 604–609 (2004).14656754 10.1164/rccm.200308-1101OC
24. Acuna-Villaorduna C Cough-aerosol cultures of Mycobacterium tuberculosis in the prediction of outcomes after exposure. A household contact study in Brazil PLoS ONE 2018 13 e0206384 10.1371/journal.pone.0206384 30372480
Acuna-Villaorduna, C. et al. Cough-aerosol cultures of Mycobacterium tuberculosis in the prediction of outcomes after exposure. A household contact study in Brazil. PLoS ONE 13, e0206384 (2018).30372480 10.1371/journal.pone.0206384
25. Fennelly KP Variability of infectious aerosols produced during coughing by patients with pulmonary tuberculosis Am. J. Respir. Crit. Care Med. 2012 186 450 457 10.1164/rccm.201203-0444OC 22798319
Fennelly, K. P. et al. Variability of infectious aerosols produced during coughing by patients with pulmonary tuberculosis. Am. J. Respir. Crit. Care Med. 186, 450–457 (2012).22798319 10.1164/rccm.201203-0444OC
26. Jones-Lopez EC Cough aerosols of mycobacterium tuberculosis in the prediction of incident tuberculosis disease in household contacts Clin. Infect. Dis. 2016 63 10 20 10.1093/cid/ciw199 27025837
Jones-Lopez, E. C. et al. Cough aerosols of mycobacterium tuberculosis in the prediction of incident tuberculosis disease in household contacts. Clin. Infect. Dis. 63, 10–20 (2016).27025837 10.1093/cid/ciw199
27. Jones-Lopez EC Cough aerosols of Mycobacterium tuberculosis predict new infection: a household contact study Am. J. Respir. Crit. Care Med. 2013 187 1007 1015 10.1164/rccm.201208-1422OC 23306539
Jones-Lopez, E. C. et al. Cough aerosols of Mycobacterium tuberculosis predict new infection: a household contact study. Am. J. Respir. Crit. Care Med. 187, 1007–1015 (2013).23306539 10.1164/rccm.201208-1422OC
28. Jones-Lopez EC Cough aerosol cultures of mycobacterium tuberculosis: Insights on TST / IGRA discordance and transmission dynamics PLoS ONE 2015 10 e0138358 10.1371/journal.pone.0138358 26394149
Jones-Lopez, E. C. et al. Cough aerosol cultures of mycobacterium tuberculosis: Insights on TST / IGRA discordance and transmission dynamics. PLoS ONE 10, e0138358 (2015).26394149 10.1371/journal.pone.0138358
29. Theron G Bacterial and host determinants of cough aerosol culture positivity in patients with drug-resistant versus drug-susceptible tuberculosis Nat. Med. 2020 26 1435 1443 10.1038/s41591-020-0940-2 32601338
Theron, G. et al. Bacterial and host determinants of cough aerosol culture positivity in patients with drug-resistant versus drug-susceptible tuberculosis. Nat. Med. 26, 1435–1443 (2020).32601338 10.1038/s41591-020-0940-2
30. Wood R Real-time investigation of tuberculosis transmission: Developing the respiratory aerosol sampling chamber (RASC) PLoS ONE 2016 11 e0146658 10.1371/journal.pone.0146658 26807816
Wood, R. et al. Real-time investigation of tuberculosis transmission: Developing the respiratory aerosol sampling chamber (RASC). PLoS ONE 11, e0146658 (2016).26807816 10.1371/journal.pone.0146658
31. Patterson B Bioaerosol sampling of patients with suspected pulmonary tuberculosis: a study protocol BMC Infect. Dis. 2020 20 587 10.1186/s12879-020-05278-y 32770954
Patterson, B. et al. Bioaerosol sampling of patients with suspected pulmonary tuberculosis: a study protocol. BMC Infect. Dis. 20, 587 (2020).32770954 10.1186/s12879-020-05278-y
32. Dinkele R Capture and visualization of live Mycobacterium tuberculosis bacilli from tuberculosis patient bioaerosols PLoS Pathog. 2021 17 e1009262 10.1371/journal.ppat.1009262 33524021
Dinkele, R. et al. Capture and visualization of live Mycobacterium tuberculosis bacilli from tuberculosis patient bioaerosols. PLoS Pathog. 17, e1009262 (2021).33524021 10.1371/journal.ppat.1009262
33. Williams CM Exhaled Mycobacterium tuberculosis output and detection of subclinical disease by face-mask sampling: prospective observational studies Lancet Infect. Dis. 2020 20 607 617 10.1016/S1473-3099(19)30707-8 32085847
Williams, C. M. et al. Exhaled Mycobacterium tuberculosis output and detection of subclinical disease by face-mask sampling: prospective observational studies. Lancet Infect. Dis. 20, 607–617 (2020).32085847 10.1016/S1473-3099(19)30707-8
34. Dinkele R Aerosolization of Mycobacterium tuberculosis by Tidal Breathing Am. J. Respir. Crit. Care Med. 2022 206 206 216 10.1164/rccm.202110-2378OC 35584342
Dinkele, R. et al. Aerosolization of Mycobacterium tuberculosis by Tidal Breathing. Am. J. Respir. Crit. Care Med. 206, 206–216 (2022).35584342 10.1164/rccm.202110-2378OC
35. Behr MA Edelstein PH Ramakrishnan L Revisiting the timetable of tuberculosis BMJ 2018 362 k2738 10.1136/bmj.k2738 30139910
Behr, M. A., Edelstein, P. H. & Ramakrishnan, L. Revisiting the timetable of tuberculosis. BMJ 362, k2738 (2018).30139910 10.1136/bmj.k2738
36. Dowdy DW Azman AS Kendall EA Mathema B Transforming the fight against tuberculosis: targeting catalysts of transmission Clin. Infect. Dis. 2014 59 1123 1129 10.1093/cid/ciu506 24982034
Dowdy, D. W., Azman, A. S., Kendall, E. A. & Mathema, B. Transforming the fight against tuberculosis: targeting catalysts of transmission. Clin. Infect. Dis. 59, 1123–1129 (2014).24982034 10.1093/cid/ciu506
37. Gentile I., et al. The role of CRP POC testing in the fight against antibiotic overuse in European primary care: Recommendations from a European expert panel. Diagnostics 13, 10.3390/diagnostics13020320 (2023).
38. Dhana A Tuberculosis screening among ambulatory people living with HIV: a systematic review and individual participant data meta-analysis Lancet Infect. Dis. 2022 22 507 518 10.1016/S1473-3099(21)00387-X 34800394
Dhana, A. et al. Tuberculosis screening among ambulatory people living with HIV: a systematic review and individual participant data meta-analysis. Lancet Infect. Dis. 22, 507–518 (2022).34800394 10.1016/S1473-3099(21)00387-X
39. de Hollander EL Bemelmans WJ de Groot LC Associations between changes in anthropometric measures and mortality in old age: a role for mid-upper arm circumference? J. Am. Med. Dir. Assoc. 2013 14 187 193 10.1016/j.jamda.2012.09.023 23168109
de Hollander, E. L., Bemelmans, W. J. & de Groot, L. C. Associations between changes in anthropometric measures and mortality in old age: a role for mid-upper arm circumference? J. Am. Med. Dir. Assoc. 14, 187–193 (2013).23168109 10.1016/j.jamda.2012.09.023
40. Schaap LA Quirke T Wijnhoven HAH Visser M Changes in body mass index and mid-upper arm circumference in relation to all-cause mortality in older adults Clin. Nutr. 2018 37 2252 2259 10.1016/j.clnu.2017.11.004 29195733
Schaap, L. A., Quirke, T., Wijnhoven, H. A. H. & Visser, M. Changes in body mass index and mid-upper arm circumference in relation to all-cause mortality in older adults. Clin. Nutr. 37, 2252–2259 (2018).29195733 10.1016/j.clnu.2017.11.004
41. Wijnhoven HA Low mid-upper arm circumference, calf circumference, and body mass index and mortality in older persons J. Gerontol. A Biol. Sci. Med. Sci. 2010 65 1107 1114 10.1093/gerona/glq100 20547497
Wijnhoven, H. A. et al. Low mid-upper arm circumference, calf circumference, and body mass index and mortality in older persons. J. Gerontol. A Biol. Sci. Med. Sci. 65, 1107–1114 (2010).20547497 10.1093/gerona/glq100
42. Getnet F Delay in diagnosis of pulmonary tuberculosis increases the risk of pulmonary cavitation in pastoralist setting of Ethiopia BMC Pulm. Med. 2019 19 201 10.1186/s12890-019-0971-y 31694601
Getnet, F. et al. Delay in diagnosis of pulmonary tuberculosis increases the risk of pulmonary cavitation in pastoralist setting of Ethiopia. BMC Pulm. Med. 19, 201 (2019).31694601 10.1186/s12890-019-0971-y
43. Baazim H Antonio-Herrera L Bergthaler A The interplay of immunology and cachexia in infection and cancer Nat. Rev. Immunol. 2022 22 309 321 10.1038/s41577-021-00624-w 34608281
Baazim, H., Antonio-Herrera, L. & Bergthaler, A. The interplay of immunology and cachexia in infection and cancer. Nat. Rev. Immunol. 22, 309–321 (2022).34608281 10.1038/s41577-021-00624-w
44. Redford SE Varanasi SK Sanchez KK Thorup NR Ayres JS CD4+ T cells regulate sickness-induced anorexia and fat wasting during a chronic parasitic infection Cell Rep. 2023 42 112814 10.1016/j.celrep.2023.112814 37490905
Redford, S. E., Varanasi, S. K., Sanchez, K. K., Thorup, N. R. & Ayres, J. S. CD4+ T cells regulate sickness-induced anorexia and fat wasting during a chronic parasitic infection. Cell Rep. 42, 112814 (2023).37490905 10.1016/j.celrep.2023.112814
45. Dorhoi A Kaufmann SH Perspectives on host adaptation in response to Mycobacterium tuberculosis: modulation of inflammation Semin Immunol. 2014 26 533 542 10.1016/j.smim.2014.10.002 25453228
Dorhoi, A. & Kaufmann, S. H. Perspectives on host adaptation in response to Mycobacterium tuberculosis: modulation of inflammation. Semin Immunol. 26, 533–542 (2014).25453228 10.1016/j.smim.2014.10.002
46. Matty MA Roca FJ Cronan MR Tobin DM Adventures within the speckled band: heterogeneity, angiogenesis, and balanced inflammation in the tuberculous granuloma Immunol. Rev. 2015 264 276 287 10.1111/imr.12273 25703566
Matty, M. A., Roca, F. J., Cronan, M. R. & Tobin, D. M. Adventures within the speckled band: heterogeneity, angiogenesis, and balanced inflammation in the tuberculous granuloma. Immunol. Rev. 264, 276–287 (2015).25703566 10.1111/imr.12273
47. Oehlers SH Interception of host angiogenic signalling limits mycobacterial growth Nature 2015 517 612 615 10.1038/nature13967 25470057
Oehlers, S. H. et al. Interception of host angiogenic signalling limits mycobacterial growth. Nature 517, 612–615 (2015).25470057 10.1038/nature13967
48. Walton EM Cyclopropane modification of trehalose dimycolate drives granuloma angiogenesis and mycobacterial growth through VEGF signaling Cell Host Microbe 2018 24 514 25 e6 10.1016/j.chom.2018.09.004 30308157
Walton, E. M. et al. Cyclopropane modification of trehalose dimycolate drives granuloma angiogenesis and mycobacterial growth through VEGF signaling. Cell Host Microbe 24, 514–25 e6 (2018).30308157 10.1016/j.chom.2018.09.004
49. Riley RL Infectiousness of air from a tuberculosis ward. Ultraviolet irradiation of infected air: comparative infectiousness of different patients Am. Rev. Respir. Dis. 1962 85 511 525 14492300
Riley, R. L. et al. Infectiousness of air from a tuberculosis ward. Ultraviolet irradiation of infected air: comparative infectiousness of different patients. Am. Rev. Respir. Dis. 85, 511–525 (1962).14492300
50. Coussens A. K., et al. Classification of early tuberculosis states to guide research for improved care and prevention: an international Delphi consensus exercise. Lancet Respir Med. 6, 484–498 (2024).
51. Williams CM Face mask sampling for the detection of Mycobacterium tuberculosis in expelled aerosols PLoS ONE 2014 9 e104921 10.1371/journal.pone.0104921 25122163
Williams, C. M. et al. Face mask sampling for the detection of Mycobacterium tuberculosis in expelled aerosols. PLoS ONE 9, e104921 (2014).25122163 10.1371/journal.pone.0104921
52. Williams CM Exhaled Mycobacterium tuberculosis predicts incident infection in household contacts Clin. Infect. Dis. 2023 76 e957 e964 10.1093/cid/ciac455 36350995
Williams, C. M. et al. Exhaled Mycobacterium tuberculosis predicts incident infection in household contacts. Clin. Infect. Dis. 76, e957–e964 (2023).36350995 10.1093/cid/ciac455
53. Patterson B Aerosolization of viable Mycobacterium tuberculosis bacilli by tuberculosis clinic attendees independent of sputum-Xpert Ultra status Proc. Natl. Acad. Sci. USA 2024 121 e2314813121 10.1073/pnas.2314813121 38470917
Patterson, B. et al. Aerosolization of viable Mycobacterium tuberculosis bacilli by tuberculosis clinic attendees independent of sputum-Xpert Ultra status. Proc. Natl. Acad. Sci. USA 121, e2314813121 (2024).38470917 10.1073/pnas.2314813121
54. Ribeiro-Rodrigues R Sputum cytokine levels in patients with pulmonary tuberculosis as early markers of mycobacterial clearance Clin. Diagn. Lab Immunol. 2002 9 818 823 12093679
Ribeiro-Rodrigues, R. et al. Sputum cytokine levels in patients with pulmonary tuberculosis as early markers of mycobacterial clearance. Clin. Diagn. Lab Immunol. 9, 818–823 (2002).12093679
55. Ota MO Rapid diagnosis of tuberculosis using ex vivo host biomarkers in sputum Eur. Respir. J. 2014 44 254 257 10.1183/09031936.00209913 24627540
Ota, M. O. et al. Rapid diagnosis of tuberculosis using ex vivo host biomarkers in sputum. Eur. Respir. J. 44, 254–257 (2014).24627540 10.1183/09031936.00209913
56. Heslop R Changes in host cytokine patterns of TB patients with different bacterial loads detected using 16S rRNA analysis PLoS ONE 2016 11 e0168272 10.1371/journal.pone.0168272 27992487
Heslop, R. et al. Changes in host cytokine patterns of TB patients with different bacterial loads detected using 16S rRNA analysis. PLoS ONE 11, e0168272 (2016).27992487 10.1371/journal.pone.0168272
57. Bell A Fairbrother M Jones K Fixed and random effects models: making an informed choice Qual. Quant. 2019 53 1051 1074 10.1007/s11135-018-0802-x
Bell, A., Fairbrother, M. & Jones, K. Fixed and random effects models: making an informed choice. Qual. Quant. 53, 1051–1074 (2019).10.1007/s11135-018-0802-x
58. Acuna-Villaorduna C Host determinants of infectiousness in smear-positive patients With pulmonary tuberculosis Open Forum Infect. Dis. 2019 6 ofz184 10.1093/ofid/ofz184 31205972
Acuna-Villaorduna, C. et al. Host determinants of infectiousness in smear-positive patients With pulmonary tuberculosis. Open Forum Infect. Dis. 6, ofz184 (2019).31205972 10.1093/ofid/ofz184
59. Dowdy DW Behr MA Are we underestimating the annual risk of infection with Mycobacterium tuberculosis in high-burden settings? Lancet Infect. Dis. 2022 22 e271 e278 10.1016/S1473-3099(22)00153-0 35526558
Dowdy, D. W. & Behr, M. A. Are we underestimating the annual risk of infection with Mycobacterium tuberculosis in high-burden settings? Lancet Infect. Dis. 22, e271–e278 (2022).35526558 10.1016/S1473-3099(22)00153-0
60. Yew WW Leung CC Are some people not safer after successful treatment of tuberculosis? Am. J. Respir. Crit. Care Med. 2005 171 1324 1325 10.1164/rccm.2502005 15941841
Yew, W. W. & Leung, C. C. Are some people not safer after successful treatment of tuberculosis? Am. J. Respir. Crit. Care Med. 171, 1324–1325 (2005).15941841 10.1164/rccm.2502005
61. Martinez-Gonzalez N. A., et al. Point-of-care C-reactive protein testing to reduce antibiotic prescribing for respiratory tract infections in primary care: Systematic review and meta-analysis of randomised controlled trials. Antibiotics 9, 10.3390/antibiotics9090610 (2020).
62. Enos M Kenya tuberculosis prevalence survey 2016: Challenges and opportunities of ending TB in Kenya PLoS ONE 2018 13 e0209098 10.1371/journal.pone.0209098 30586448
Enos, M. et al. Kenya tuberculosis prevalence survey 2016: Challenges and opportunities of ending TB in Kenya. PLoS ONE 13, e0209098 (2018).30586448 10.1371/journal.pone.0209098
63. Harris PA Research electronic data capture (REDCap)-a metadata-driven methodology and workflow process for providing translational research informatics support J. Biomed. Inf. 2009 42 377 381 10.1016/j.jbi.2008.08.010
Harris, P. A. et al. Research electronic data capture (REDCap)-a metadata-driven methodology and workflow process for providing translational research informatics support. J. Biomed. Inf. 42, 377–381 (2009).10.1016/j.jbi.2008.08.010
64. Sharma M TBscreen: A passive cough classifier for tuberculosis screening with a controlled dataset Sci. Adv. 2024 10 eadi0282 10.1126/sciadv.adi0282 38170773
Sharma, M. et al. TBscreen: A passive cough classifier for tuberculosis screening with a controlled dataset. Sci. Adv. 10, eadi0282 (2024).38170773 10.1126/sciadv.adi0282
65. Njagi L. N., Nduba V., Mureithi M. W., Mecha J. O. Prevalence and predictors of tuberculosis infection among people living with HIV in a high tuberculosis burden context. BMJ Open Respir Res.10.1136/bmjresp-2022-001581 (2023).
66. Birring SS Development of a symptom specific health status measure for patients with chronic cough: Leicester Cough Questionnaire (LCQ) Thorax 2003 58 339 343 10.1136/thorax.58.4.339 12668799
Birring, S. S. et al. Development of a symptom specific health status measure for patients with chronic cough: Leicester Cough Questionnaire (LCQ). Thorax 58, 339–343 (2003).12668799 10.1136/thorax.58.4.339
67. Hopewell PC Pai M Maher D Uplekar M Raviglione MC International standards for tuberculosis care Lancet Infect. Dis. 2006 6 710 725 10.1016/S1473-3099(06)70628-4 17067920
Hopewell, P. C., Pai, M., Maher, D., Uplekar, M. & Raviglione, M. C. International standards for tuberculosis care. Lancet Infect. Dis. 6, 710–725 (2006).17067920 10.1016/S1473-3099(06)70628-4
68. Mayer-Barber K. D., Barber D. L. Innate and Adaptive Cellular Immune Responses to Mycobacterium tuberculosis Infection. Cold Spring Harb Perspect Med. 5, a018424 (2015).
69. R. Core Team. R: A language and environment for statistical computing. 4.2.3 ed: (Foundation for Statistical Computing, Vienna, Austria; 2023).
70. Law CW Chen Y Shi W Smyth GK voom: Precision weights unlock linear model analysis tools for RNA-seq read counts Genome Biol. 2014 15 R29 10.1186/gb-2014-15-2-r29 24485249
Law, C. W., Chen, Y., Shi, W. & Smyth, G. K. voom: Precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biol. 15, R29 (2014).24485249 10.1186/gb-2014-15-2-r29
71. Greenland S Pearl J Robins JM Causal diagrams for epidemiologic research Epidemiology 1999 10 37 48 10.1097/00001648-199901000-00008 9888278
Greenland, S., Pearl, J. & Robins, J. M. Causal diagrams for epidemiologic research. Epidemiology 10, 37–48 (1999).9888278 10.1097/00001648-199901000-00008
72. Digitale JC Martin JN Glymour MM Tutorial on directed acyclic graphs J. Clin. Epidemiol. 2022 142 264 267 10.1016/j.jclinepi.2021.08.001 34371103
Digitale, J. C., Martin, J. N. & Glymour, M. M. Tutorial on directed acyclic graphs. J. Clin. Epidemiol. 142, 264–267 (2022).34371103 10.1016/j.jclinepi.2021.08.001
73. Youden WJ Index for rating diagnostic tests Cancer 1950 3 32 35 10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>3.0.CO;2-3 15405679
Youden, W. J. Index for rating diagnostic tests. Cancer 3, 32–35 (1950).15405679 10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>3.0.CO;2-3
74. Somers RH A new asymmetric measure of association for ordinal variables Am. Sociol. Rev. 1962 27 799 811 10.2307/2090408
Somers, R. H. A new asymmetric measure of association for ordinal variables. Am. Sociol. Rev. 27, 799–811 (1962).10.2307/2090408
75. Dill-McFarland K. A., et al. Kimma: flexible linear mixed effects modeling with kinship covariance for RNA-seq data. Bioinformatics. 39, btad279 (2023).
76. Korotkevich G., et al. Fast gene set enrichment analysis. Preprint at bioRxiv 10.1101/060012 (2021).
77. Subramanian A Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles Proc. Natl. Acad. Sci. USA 2005 102 15545 15550 10.1073/pnas.0506580102 16199517
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545–15550 (2005).16199517 10.1073/pnas.0506580102
