[ Chest Infections Original Research ]

Risk Factors for Nontuberculous

Mycobacterial Pulmonary Disease
A Systematic Literature Review and Meta-Analysis
Michael R. Loebinger, PhD; Jennifer K. Quint, PhD; Roald van der Laan, PhD; Marko Obradovic;
Rajinder Chawla, PhD; Amit Kishore, PhD; and Jakko van Ingen, MD, PhD

BACKGROUND: Nontuberculous mycobacterial pulmonary disease (NTM-PD) is widely

underdiagnosed, and certain patient groups, such as those with underlying respiratory dis-
eases, are at increased risk of developing the disease. Understanding patients at risk is
essential to allow for prompt testing and diagnosis and appropriate management to prevent
disease progression.
RESEARCH QUESTION: What are the risk factors for NTM-PD that should prompt a physician
to consider NTM testing and diagnosis?
STUDY DESIGN AND METHODS: Electronic searches of PubMed and EMBASE were conducted in
July 2021 for the period 2011-2021. Inclusion criteria were studies of patients with NTM-PD
with associated risk factors. Data were extracted and assessed using the Newcastle-Ottawa
Scale. Data analysis was conducted using the R-based “meta” package. Only studies that re-
ported association outcomes for cases with NTM-PD compared with control participants
(healthy populations or participants without NTM-PD) were considered for the meta-analysis.
RESULTS: Of the 9,530 searched publications, 99 met the criteria for the study. Of these, 24
formally reported an association between possible risk factors and the presence of NTM-PD
against a control population and were included in the meta-analysis. Comorbid respiratory
disease was associated with a significant increase in the OR for NTM-PD (bronchiectasis [OR,
21.43; 95% CI, 5.90-77.82], history of TB [OR, 12.69; 95% CI, 2.39-67.26], interstitial lung
disease [OR, 6.39; 95% CI, 2.65-15.37], COPD [OR, 6.63; 95% CI, 4.57-9.63], and asthma [OR,
4.15; 95% CI, 2.81-6.14]). Other factors noted to be associated with an increased risk of NTM-
PD were the use of inhaled corticosteroids (OR 4.46; 95% CI, 2.13-9.35), solid tumors (OR, 4.66;
95% CI, 1.04-20.94) and the presence of pneumonia (OR, 5.54; 95% CI, 2.72-11.26).
INTERPRETATION: The greatest risk for NTM-PD is conferred by comorbid respiratory dis-
eases such as bronchiectasis. These findings could help with identification of patient pop-
ulations at risk for NTM-PD to drive prompt testing and appropriate initiation of therapy.
CHEST 2023; 164(5):1115-1124

KEY WORDS: clinical practice; nontuberculous mycobacteria; NTM-PD; risk factors

ABBREVIATIONS: CF = cystic fibrosis; HR = hazard ratio; NTM = Data in this manuscript have been presented previously at the World
nontuberculous mycobacteria; NTM-PD = nontuberculous mycobac- Bronchiectasis and NTM Conference, Prague, June 30- July 2, 2022.
terial pulmonary disease; RR = relative risk CORRESPONDENCE TO: Michael R. Loebinger, PhD; email: m.loebinger@
AFFILIATIONS: From the Royal Brompton Hospital and NHLI, Im- rbht.nhs.uk
perial College London, London, England (M. R. L. and J. K. Q.); Copyright Ó 2023 The Author(s). Published by Elsevier Inc under li-
Insmed B.V. (R. v. d. L.), Utrecht, The Netherlands; Insmed Germany cense from the American College of Chest Physicians. This is an open
GmbH (M. O.), Frankfurt am Main, Germany; Accuscript Consultancy access article under the CC BY license (http://creativecommons.org/
(R. C. and A. K.), Ludhiana, Punjab, India; the Department of Medical licenses/by/4.0/).
Microbiology (J. v. I.), Radboud University Medical Center, Nijmegen, DOI: https://doi.org/10.1016/j.chest.2023.06.014
The Netherlands.

chestjournal.org 1115
 associated with significant morbidity and mortality.4-7
Take-home Points Incidence rates of NTM-PD are rising, but the disease
Study Question: What are the risk factors for remains challenging to diagnose and treat.8 Clinical
developing nontuberculous mycobacterial pulmo- symptoms of NTM-PD are nonspecific and often
nary disease (NTM-PD)? overlap with those of other underlying respiratory
Results: Comorbid respiratory disease, including conditions, leading to a high proportion of patients with
bronchiectasis, COPD, and history of tuberculosis, NTM-PD remaining undiagnosed.9-11 Testing for NTM-
was associated with the highest risk for NTM-PD; the PD is important to allow for prompt diagnosis and, after
use of inhaled corticosteroids, solid tumors, and the this, appropriate management—which should be
presence of pneumonia were also important factors. individualized and may include guideline-based therapy
Interpretation: This is the first study to provide an or nonpharmacologic treatment such as airway
overview of risk for NTM-PD for a comprehensive clearance—can be initiated to prevent disease
set of potential factors. progression.12-14

Previous data have highlighted that particular groups of

Nontuberculous mycobacteria (NTM) are ubiquitous in patients are at risk of NTM-PD.15 Testing for NTM-PD
the environment1,2 and encompass more than 200 is recommended by some society guidelines when risk
species, which differ in their propensity to cause factors, clinical symptoms, or suspicious radiologic
pulmonary disease.2 Nontuberculous mycobacteria findings are found14,16; however, no study has provided
pulmonary disease (NTM-PD) is a serious, chronic, rare a comprehensive insight into risk factors for NTM-PD,
infectious disease that often presents in people with including underlying comorbidities and patient
underlying lung conditions such as bronchiectasis, cystic characteristics. The aim of this current study is to
fibrosis (CF), and COPD.3 NTM-PD can be a substantial identify risk factors for NTM-PD that can be used to
burden for patients, contributing to lung function determine which patients are at risk of infection so that
decline and reduced health-related quality of life, and is prompt and appropriate diagnoses can be made.

Study Design and Methods Data Extraction

Protocol and Registration From the studies identified, information on study characteristics
(study name, year, author, digital object identifier, study design,
The systematic literature review has been registered in the
aim, study duration, country, inclusion and exclusion criteria, study
International Prospective Register of Systematic Reviews (www.crd.
groups, and sample size); patient characteristics (age, sex, ethnicity,
york.ac.uk/prospero; record ID: 347379) and has been conducted in
case definition, disease severity, NTM species data, and patient
line with the Preferred Reporting Items for Systematic Review and
symptoms); NTM-PD association with comorbidities (including
Meta-analyses guidelines.17 Institutional review board approval was
respiratory disorders, cancers, immune disorders, transplants,
not required.
immunosuppression, cardiovascular disease, renal disease, and
diabetes); and NTM-PD association with risk factors (age, sex,
Search Strategy ethnicity, BMI, FEV1, macrolide therapy, and concomitant bacterial
All authors determined a search strategy that included both free-text and fungal infection) were collated into a data extraction table
words and medical subject headings. The search string can be found (e-Table 3).
in e-Table 1 and was used to perform searches in MEDLINE via
PubMed and EMBASE. Searches were undertaken in July 2021. Quality and Risk of Bias Assessment
Internal validity was determined through multiple reviews of the data
Study Selection (M. O., M. R. L., J. v. I., J. K. Q., R. C., A. K.). Three reviewers (J. K. Q.,
Inclusion and exclusion criteria for publications are shown in R. C., and A. K.) independently assessed the quality of each study using
e-Table 2. Search parameters were limited to material published the Newcastle-Ottawa Scale, a bias evaluation tool recommended by
in English between 2011 and 2021. A pilot phase was used to the Cochrane Collection (e-Table 3).18,19 The review satisfied the
assess the concordance of decisions between reviewers and the criteria of the ROBIS tool.20
relevance and utility of the screening criteria before continuing.
Two authors (R. C. and A. K.) undertook the first-pass evaluation Data Synthesis and Analysis
of studies, reviewing titles and abstracts, and then there was a Data analysis was conducted by two reviewers (R. C. and A. K.) using
second-pass evaluation of published full texts to determine which the R-based “meta” package. An Excel-based feasibility sheet was
met eligibility criteria for subsequent review and discussion by all developed that included details of study characteristics, population
the authors. All authors independently reviewed potentially characteristics, risk factors, subgroups, crude or adjusted association
relevant studies to determine their eligibility and the strength of data, 95% CI, level of significance, and natural log-transformed
the evidence. Any disagreements were resolved through author association data. Only studies that reported association outcomes for
discussion. cases with NTM-PD compared with control participants (healthy

1116 Original Research [ 164#5 CHEST NOVEMBER 2023 ]

populations or participants without NTM-PD) were considered for the can be roughly interpreted as follows: 0% to 40%, might not be
meta-analysis. Studies that did not include a reference group without important; 30% to 60%, may represent moderate heterogeneity;
NTM-PD or did not report 95% CI for the risk estimates were not 50% to 90%, may represent substantial heterogeneity; 75% to 100%,
selected for the meta-analysis. considerable heterogeneity. All the studies had an observational
The adjusted data (adjusted OR, adjusted hazard ratio [HR], or study design, and a higher heterogeneity (I2 > 50%) was expected;
adjusted relative risk [RR]) were used wherever available, but crude thus, a random effects model was used in the meta-analysis. The
data were used when adjusted data were not available. Heterogeneity meta-analysis was performed with the subgroups analysis as per
across returned data was evaluated using I2 statistics. Heterogeneity population type and data format type (OR, HR, or RR).

Results Sixty studies reported a formal analysis of association

(OR, HR, or RR) between risk factors and occurrence
Systematic Literature Review
of NTM-PD, most of the studies reporting an
Searches returned an initial total of 9,530 publications, association for one or two different risk factors,
which after screening led to the final selection of 99 although two studies evaluated data for 1021 and 1222
studies (Fig 1; e-Table 1). Of the 99 publications risk factors, respectively. e-Table 4 summarizes the
identified, 90 were full-text articles and nine were identified published association between
congress abstracts; they included 96 primary studies and comorbidities, use of particular medications, clinical
three secondary publications. An overview of included history, patient demographics, and concomitant
studies is shown in e-Table 3. infection.
Identification

Records identified, N = 9,530

A. Embase, n = 5,461
B. PubMed, n = 4,069

Duplicates, n = 2,284
Screening

Records screened Title/abstract based Exclude, n = 6,339

N = 7,246 • Duplicate: 491
• Disease not of interest: 2,856
• Animal/in vitro: 492
• Study design not of interest: 373
• Outcome not of interest: 779
• Publication type not of interest: 748
• Review: 427
• Language (non-English): 173
Eligibility

Citations assessed for eligibility

N = 907 Exclude, n = 808
• Duplicate: 32
• Disease not of interest: 49
• Animal/in vitro: 15
• Study design not of interest: 7
• Outcome not of interest: 623
• Publication type not of interest: 35
Included

Included studies, N = 99 • Review: 27

(Full text, n = 90, congress abstracts, n = 9) • Language (non-English): 20
(Primary, n = 96, secondary, n = 3)
Exclude, n = 36
• Non-epidemiological risk factors: 32
• Epidemiological risk factors without
Meta-analysis

sufficient detail: 3
• Genetic risk factor: 1
Included for meta-analysis
N = 24

Figure 1 – Preferred Reporting Items for Systematic Review and Meta-Analyses flowchart depicting the selection of studies and data in the analysis.

chestjournal.org 1117
The comorbidities most cited with a formal association term use of macrolides for underlying disease, although
with NTM-PD were COPD (n ¼ 8), history of TB not statistically significant, was suggested as being
(n ¼ 7), and immunosuppression (n ¼ 10); diabetes and possibly protective against NTM-PD (OR, 0.80; 95% CI,
cancer were also prominent, reporting an association in 0.47-1.39) (Fig 2H; e-Table 7, e-Fig 1G, H). Female sex
more than five studies. Other clinical and demographic was a nonsignificant risk factor for NTM, with an OR of
factors citing an association with NTM-PD included 1.27 (95% CI, 0.95-1.69) (e-Table 7, e-Fig 1I).
bronchiectasis, BMI (healthy weight, overweight, and
Other marginal or statistically nonsignificant
underweight), sex, radiologic findings, concomitant
associations with NTM-PD were observed for lung
bacterial infection with Pseudomonas aeruginosa or
function (FEV1%) (OR, 1.01; 95% CI, 0.97-1.05),
Staphylococcus aureus, FEV1 % predicted, ethnicity, age,
diabetes (OR, 1.04; 95% CI, 0.63-1.72), renal disease
and macrolide use. Association analysis suggested an
(OR, 1.62; 95% CI, 0.87-3.03), cancer (OR, 1.39; 95% CI,
increased risk of NTM-PD with most of the parameters
0.70-2.68), healthy weight (OR, 0.96; 95% CI, 0.85-1.09),
in at least one study, except FEV1 % predicted.
and infection with Pseudomonas aeruginosa (OR, 1.12;
95% CI, 0.87-1.44) or Staphylococcus aureus (OR, 2.42;
Meta-Analysis
95% CI, 0.55-10.63) (e-Table 7).
Of the identified publications, 24 formally reported an
association between possible risk factors and the
presence of NTM-PD against a control population (cases Discussion
without NTM or healthy control participants without The systematic literature review and meta-analysis aimed
NTM-PD) and were included in the meta-analysis to identify potential risk factors associated with NTM-PD
(e-Table 5).15,21-43 Of these 24 studies, three reported reported in the literature and to pool results across relevant
HR, one reported RR, and the remainder (n ¼ 20) studies. The study brings these data together into a single
reported (crude or adjusted) OR. A total of 36 studies publication, allowing extensive comprehension and
reporting possible risk factors associated with NTM-PD comparison of the associations of various risk factors with
were excluded from the analysis (e-Table 6). NTM-PD. The most frequently reported risk factors across
Results for the meta-analysis indicate that comorbid selected publications were use of immunosuppressants (10
respiratory disease was associated with a significant studies), sex (nine studies), COPD comorbidity (eight
increase in the OR for NTM-PD, with ORs ranging from studies), and history of or suspected TB (seven studies).
4 to 21 (e-Table 7). Bronchiectasis showed the strongest Bronchiectasis, which is considered the most relevant risk
association (OR, 21.43; 95% CI, 5.90-77.82), followed by factor for NTM, was assessed in four studies.
a history of TB (OR, 12.69; 95% CI, 2.39-67.26), Results for the meta-analysis indicated that comorbid
interstitial lung disease (OR, 6.39; 95% CI, 2.65-15.37), respiratory disease is associated with a significant increase
COPD (OR, 6.63; 95% CI, 4.57-9.63), and asthma (OR, in the risk of NTM-PD, with the strongest association
4.15; 95% CI, 2.81-6.14) (Fig 2A-E). Other factors noted seen for bronchiectasis. Other comorbidities noted to be
to be associated with an increased risk of NTM-PD were associated with an increased risk of NTM-PD were solid
immunosuppression, including use of inhaled tumors, cardiovascular disease, and use of
corticosteroids (OR 4.46; 95% CI, 2.13-9.35), oral immunosuppressants such as inhaled corticosteroids or
corticosteroids (OR 3.37; CI, 0.82-13.75), and other anti-tumor necrosis factor-alpha treatment for
immunosuppressants such as tacrolimus or rheumatoid arthritis. The association between solid
mycophenolate mofetil (OR 2.60; 95% CI, 0.69-9.79), as tumors and risk of NTM-PD may be a result of cancer
well as anti-tumor necrosis factor-alpha treatment for therapy rather than cancer itself; similarly, the risk for
rheumatoid arthritis (OR, 2.13; 95% CI, 1.24-3.65), solid NTM infection in patients with interstitial lung disease
tumors (OR, 4.66; 95% CI, 1.04-20.94), the presence of may be a result of disease rather than a risk factor per se.
pneumonia (OR, 5.54; 95% CI, 2.72-11.26), It is also worth highlighting that the changing landscape
cardiovascular disease (OR, 1.73; 95% CI, 1.01-2.97), of immune suppression and the increased use of agents,
and low BMI (being underweight) (OR, 3.04; 95% CI, such as biologics, that may have promoted mycobacterial
1.95-4.73) (Fig 2F, G; e-Table 7, e-Fig 1A-F). In contrast, disease in recent years.44 Introduction of these agents
increasing BMI (OR, 0.82; 95% CI, 0.71-0.95) or high may have significant implications for the development of
BMI (being overweight) (OR, 0.73; 95% CI, 0.58-0.97) NTM-PD and may not be adequately represented in
were associated with being protective factors, and long- older studies. Future evaluation of NTM-PD in people

1118 Original Research [ 164#5 CHEST NOVEMBER 2023 ]

 A
Bronchiectasis
Study or OR OR
Subgroup logOR SE Weight IV, Random, 95% CI IV, Random, 95% CI

Pop = General pop

Andrejak 2013 5.2338 1.0300 17.1% 187.50 [25.00-1417.00]
Axson 2019* 3.3707 0.2393 28.3% 29.10 [18.20-46.50]
Total (95% CI) 45.4% 56.43 [9.83-323.98]
Heterogeneity Tau 2 = 1.1764; Chi 2 = 3.1, df = 1 (P = .08); I 2 = 68%

Pop = TB symptoms
Tan 2021 1.6014 0.1326 29.1% 4.96 [3.83-6.44]
Xu 2019* 2.9386 0.4716 25.5% 18.89 [7.54-47.88]
Total (95% CI) 54.6% 8.97 [2.44-32.96]
Heterogeneity Tau = 1.7741; Chi = 7.45, df = 1 (P < .01); I 2 = 87%
2 2

Total (95% CI) 100.0% 21.43 [5.90-77.82]

Heterogeneity Tau 2 = 0.4727; Chi 2 = 54.76, df = 3 (P < .01); I 2 = 95%
Test for overall effect: Z = 4.66 (P < .01) 0.001 0.1 1 10 1000
Test for subgroup differences: Chi 2 = 2.74, df = 1 (P = .10)

B
Tuberculosis
Study or OR OR
Subgroup logOR SE Weight IV, Random, 95% CI IV, Random, 95% CI

Pop = General pop

Andrejak 2013 5.1835 0.5966 13.7% 178.30 [55.40-574.30]
Axson 2019* 4.2017 0.1558 14.6% 66.80 [49.20-90.60]
Total (95% CI) 28.3% 92.17 [37.35-227.48]
Heterogeneity Tau 2 = 0.2918; Chi 2 = 2.54, df = 1 (P = .11); I 2 = 61%

Pop = COPD
Wang 2021* 3.8238 0.3031 14.4% 45.78 [25.27-82.91]

Pop = RA
Liao 2016 1.7192 0.4161 14.2% 5.58 [2.47-12.62]

Pop = TB symptoms
Tan 2021 –0.0305 0.1120 14.7% 0.97 [0.78-1.21]
Wu 2014 0.4947 0.1691 14.6% 1.64 [1.18-2.29]
Xu 2019* 2.5588 0.5828 13.8% 12.92 [3.24-31.82]
Total (95% CI) 43.0% 2.11 [0.95-4.69]
Heterogeneity Tau 2 = 0.4074; Chi 2 = 23.64, df = 6 (P < .01); I 2 = 92%

Total (95% CI) 100% 12.69 [2.39-67.26]

2
Heterogeneity Tau 2 = 4.9243; Chi 2 = 623.17, df = 6 (P < .01); I = 99%
Test for overall effect: Z = 2.99 (P < .01) 0.01 0.1 1 10 100
2
Test for subgroup differences: Chi = 57.49, df = 3 (P < .01)

C
Interstitial lung disease
Study or OR OR
Subgroup logOR SE Weight IV, Random, 95% CI IV, Random, 95% CI

Pop = General pop

Miller 2017 2.4069 0.1190 27.1% 11.10 [8.78-14.00]
Axson 2019* 1.9155 0.2729 25.1% 6.79 [3.98-11.60]
Total (95% CI) 52.2% 9.23 [5.79-14.72]
Heterogeneity Tau 2 = 0.0765; Chi 2 = 2.73, df = 1 (P = .10); I 2 = 63%

Pop = COPD
Wang 2021* 1.0647 0.0747 27.4% 2.90 [2.50-3.35]

Pop = RA
Liao 2016 2.1066 0.5060 20.4% 8.22 [3.05-22.17]

Total (95% CI) 100.0% 6.39 [2.65-15.37]

Heterogeneity Tau 2 = 0.7265; Chi 2 = 95.66, df = 3 (P < .01); I 2 = 97%
Test for overall effect: Z = 4.146 (P < .01) 0.1 0.5 1 2 10
Test for subgroup differences: Chi 2 = 24.92, df = 2 (P < .10)

Figure 2 – A-H, Identified risk factors for NTM-PD. All data are quoted as adjusted OR (aOR) unless stated; *crude risk estimate OR; #adjusted risk
ratio (aRR); †adjusted hazard ratio; CF ¼ cystic fibrosis; df ¼ degrees of freedom; ILD ¼ interstitial lung disease; IV ¼ inverse variance; pop ¼
population; RA ¼ rheumatoid arthritis; TE ¼ treatment effect.

receiving newer immunosuppressive therapies would be reflux disease was highlighted as a risk factor in five
interesting and welcome. Some comorbid conditions, publications through the systematic literature review
such as gastroesophageal reflux disease, were not included process,45-49 but only one publication among the
in this study because of lack of data. Gastroesophageal identified studies had appropriate data for meta-analysis

chestjournal.org 1119
 D
COPD
Study or OR OR
Subgroup logOR SE Weight IV, Random, 95% CI IV, Random, 95% CI
Pop = General pop
Andrejak 2013 2.7537 2.7537 14.2% 15.70 [11.40-21.50]
Axson 2019* 1.9769 1.9769 15.1% 7.22 [5.86-8.89]
Marras 2016* 2.3351 0.0215 15.8% 10.33 [9.90-10.77]
Miller 2017 1.6827 0.0601 15.6% 5.38 [4.78-6.05]
Uno 2020* 1.8017 0.3873 9.6% 6.06 [2.78-12.69]

Total (95% CI) 70.3% 8.40 [5.72-12.31]

Heterogeneity Tau2 = 0.1639; Chi2 = 122.23, df = 4 (P = .10); I 2 = 97%

Pop = NCFB
Mirsaeidi 2013 2.2601 0.7742 4.4% 9.58 [0.13-2.66]

Pop = RA
Liao 2016 2.1506 0.5775 6.4% 8.59 [2.77-26.64]

Pop = TB symptoms
Tan 2021 0.7178 0.1961 13.6% 2.05 [1.40-3.02]
Xu 2019* 1.1848 0.6701 5.3% 3.27 [0.88-12.17]
Total (95% CI) 18.9% 2.13 [1.47-3.08]
2 2 2
Heterogeneity Tau = 0; Chi = 0.45, df = 1 (P = .50); I = 0%

Total (95% CI) 100.0% 6.63 [4.57-9.63]

Heterogeneity Tau2 = 0.2286; Chi2 = 185.75, df = 8 (P < .01); I2 = 96%
Test for overall effect: Z = 9.946 (P < .01)
0.1 0.5 1 2 10
Test for subgroup differences: Chi2 = 28.21, df = 3 (P = .10)

E
Asthma
Study or OR OR
Subgroup logOR SE Weight IV, Random, 95% CI IV, Random, 95% CI
Pop = General pop
Andrejak 2013 2.0541 0.2047 19.2% 7.80 [5.20-11.60]
Axson 2019* 1.0682 0.1002 22.7% 2.91 [2.39-3.54]
Marras 2016* 1.7299 0.0275 23.9% 5.64 [5.35-5.96]
Uno 2020* 1.5326 0.1514 21.1% 4.63 [3.43-6.21]
Total (95% CI) 86.9% 4.85 [3.30-7.12]
Heterogeneity Tau = 0.1362; Chi = 44.81, df = 3 (P < .01); I2 = 93%
2 2

Pop = TB symptoms
Tan 2021 0.3853 0.3710 13.1% 1.47 [0.71-3.04]

Total (95% CI) 100.0% 4.15 [2.81-6.14]

Heterogeneity Tau2 = 0.1657; Chi2 = 57.02, df = 4 (P < .01); I2 = 93%
Test for overall effect: Z = 7.14 (P < .01)
Test for subgroup differences: Chi 2 = 8.09, df = 1 (P < .01) 0.1 0.5 1 2 10

F
Inhaled corticosteroids
Study or OR OR
Subgroup logOR SE Weight IV, Random, 95% CI IV, Random, 95% CI
Pop = General pop
Andrejak 2013 3.1905 0.3647 22.0% 24.30 [11.90-49.70]
Axson 2019* 1.5041 0.0891 27.2% 4.50 [3.78-5.36]
Total (95% CI) 49.1% 10.08 [1.93-52.52]
Heterogeneity Tau = 1.3515; Chi = 20.18, df = 1 (P < .01); I2 = 95%
2 2

Pop = PD
Brode 2017 0.6206 0.0754 27.3% 1.86 [1.60-2.15]
Liu 2018 0.9203 0.2973 23.6% 2.51 [1.40-4.49]
Total (95% CI) 50.9% 1.89 [1.64-2.19]
Heterogeneity Tau = 0; Chi = 0.95, df = 1 (P = .33 .01); I 2 = 0%
2 2

Total (95% CI) 100.0% 4.46 [2.13-9.35]

Heterogeneity Tau2 = 0.5168; Chi2 = 93.03, df = 3 (P < .01); I 2 = 97%
Test for overall effect: Z = 3.96 (P < .01) 0.1 0.5 1 2 10
Test for subgroup differences: Chi 2 = 3.91, df = 3 (P = .05)

Figure 2 – Continued

producing an equivocal OR.48 This is despite general Despite being considered the most relevant risk factor
clinical consensus that gastroesophageal reflux disease is a for NTM, bronchiectasis was less frequently evaluated
likely risk factor for NTM-PD because of aspiration into than other risk factors in the studies in this research
the lungs and has been associated with higher bacterial because it was frequently considered an exclusion
burden and more severe radiologic findings.50 criterion. However, many studies report high numbers

1120 Original Research [ 164#5 CHEST NOVEMBER 2023 ]

 G
Pneumonia
Study or OR OR
Subgroup logOR SE Weight IV, Random, 95% CI IV, Random, 95% CI

Pop = General pop

Andrejak 2013 2.2824 0.8368 10.8% 9.80 [1.90-50.50]
Miller 2017 2.0580 0.1086 25.1% 7.83 [6.33-9.69]
Uno 2020* 2.9386 0.3686 20.3% 18.89 [9.36-39.70]
Total (95% CI) 56.3% 10.86 [5.59-21.10]
Heterogeneity Tau = 0.2034; Chi = 5.29, df = 2 (P = .07); I 2 = 62%
2 2

Pop = CF
Viviani 2016 0.3492 0.4613 18.1% 1.42 [0.57-3.50]

Pop = COPD
Wang 2021* 1.1217 0.0547 25.6% 3.07 [2.76-3.42]
Total (95% CI) 100.0% 5.54 [2.72-11.26]
Heterogeneity Tau 2 = 0.5101; Chi 2 = 84.28, df = 4 (P < .01); I 2 = 95%
Test for overall effect: Z = 4.73 (P < .01) 0.1 0.5 1 2 10
Test for subgroup differences: Chi 2 = 16.53, df = 2 (P < .01)

H
Macrolide therapy
Study or OR OR
Subgroup logOR SE Weight IV, Random, 95% CI IV, Random, 95% CI

Pop = CF
Adjemian 2018# –0.2231 0.0191 35.4% 0.80 [0.77-0.83]
Binder 2013* –0.9163 0.1768 31.0% 0.40 [0.30-0.60]
Viviani 2016 0.4331 0.1080 33.6% 1.54 [1.25-1.91]
Total (95% CI) 100.0% 0.80 [0.47-1.39]
Heterogeneity Tau 2 = 0.2191; Chi 2 = 51.86, df = 2 (P < .01); I 2 = 96%

Total (95% CI) 100.0% 0.80 [0.47-1.39]

Heterogeneity Tau 2 = 0.2191; Chi 2 = 51.86, df = 2 (P < .01); I 2 = 96%
Test for overall effect: Z = –0.78 (P = .44) 0.5 1 2
Test for subgroup differences: Chi 2 = 0.00, df = 0 (P = NA)

Figure 2 – Continued

of patients with nodular bronchiectatic NTM-PD and is pneumonia analysis also did not include data on
suggested to be almost universal in patients with pneumonia events, and in those in which data were
noncavitary NTM-PD. These patients can be considered obtained from patients with active pneumonia, data
to have bronchiectasis, although whether bronchiectasis relating to disease duration were lacking.
is a risk factor or consequence of NTM-PD is unclear.
Regardless, clearly patients with bronchiectasis will likely Limited studies in the public domain explored the
have the highest percentage of new NTM cases, the impact of receiving a solid organ transplant on the risk
exploration, diagnosis, and management of which of NTM-PD.28,32 Longworth et al32 reported that for
warrants further study. patients receiving a solid organ transplant, lung vs non-
lung transplantation significantly increased the risk for
Weight appeared to be associated with NTM-PD risk; NTM infection (OR, 11.49; 95% CI, 3.86-34.18) and
being underweight was suggested as a risk factor and bilateral lung transplant conferred the highest risk
being overweight appeared to suggest a protective effect (OR, 16.22; 95% CI, 4.24-62.08).32 Kabbani et al28
against the risk of disease. However, for patients in these specifically evaluated the risk for NTM infection in post-
studies who were underweight, it is not possible to gauge lung transplantation patients and reported that having
whether being underweight predisposed patients to an granuloma or positive mycobacterial cultures at the time
increased risk of NTM-PD or whether being infected of transplantation was associated with an increased risk
with NTM or being treated for NTM-PD reduced of post-transplantation mycobacterial infection (HR, 1.8;
appetite and impacted metabolism, leading to weight 95% CI, 1.024-3.154; HR, 2.08; 95% CI, 1.01-4.29,
loss. Similarly, the result obtained for pneumonia is respectively).28 Although solid organ transplant data are
interesting in that it is unclear whether risk due to reported in e-Table 7, there is insufficient evidence to
pneumonia is a consequence per se of the infecting comment further.
organism, underlying respiratory disease, or being
treated with medications such as oral Sex differences that appeared in this study suggestive of
corticosteroids.15,33,37,39,40 The studies used for the a higher risk for NTM-PD for female participants over

chestjournal.org 1121
male participants were seen, but this was primarily studies analyzed for macrolide use differed substantially
driven by a significant association in populations with in terms of data collection. Viviani et al39 explored risk
bronchiectasis (OR, 3.85; 95% CI, 2.10-7.08), a disease in a population with CF only in patients using
that predominantly occurs in female patients, which is macrolides for any duration within the previous
reflected by the low numbers of male patients included 12 months, whereas Binder et al24 explored associative
in the studies.30,34 In contrast, populations with CF, risk with duration of macrolide use, suggesting that
which are more evenly distributed regarding sex, show a duration is a key factor in protection against NTM
trend toward a decreased risk for NTM-PD in female infection. Adjemian et al23 similarly stratified use of
patients (OR, 0.90; 95% CI, 0.79-1.03).23,24,39 It could be macrolides into less than 1 year, 1 to 2 years, and 3 to 5
that the underlying disease rather than female sex itself years.
is responsible for the apparent increased risk for NTM-
The limitations of these data include a high degree of
PD in female patients, although this was not a significant
heterogeneity across studies, including different baseline
association. The concept of sex in relation to NTM-PD
study populations, different methodologies, and
risk may be a useful avenue of further research to
different population sizes, as seen by the I2 values, which
elucidate a causal or incidental role.
ranged from 50% to 99% across all suggested associative
Immunosuppression can arise from a range of factors. Given the high degree of heterogeneity and
medications, and in this meta-analysis, the role of accompanying large CIs, the outcomes of this meta-
inhaled and oral corticosteroids in raising the risk for analysis need to be viewed cautiously, particularly where
NTM-PD has been discussed.15,22,25,31,43 However, outcomes are reflective either of general populations or
understanding the risk of specific immunosuppressive from other populations with underlying diseases or
medication is hampered by the paucity of data, and data conditions. Such caution means the results presented
that can be combined for analysis. Developing an OR for here are useful for identifying risk factors and
a broad heading of immunosuppression provided a understanding which patients may be at an increased
value of 2.26 (95% CI, 1.53-3.33), suggesting that risk of NTM-PD but cannot provide insight into which
any immunosuppressive medication confers an factor or factors should be considered over others.
increased risk of NTM infection. However, when Variation in the studies available for analysis mean each
immunosuppressant drugs were grouped according to is subject to strengths and weaknesses, which ideally
type (oral corticosteroids, inhaled corticosteroids, and would be evaluated on a study-by-study basis, with
others), it can be seen that ORs of 4.46 (95% CI, 2.13- aggregate data then presented. It would be interesting to
9.35), 3.37 (95% CI, 0.82-13.75), and 2.60 (95% CI, 0.69- see, in further studies, whether this approach might
9.79) for oral corticosteroids, inhaled corticosteroids, change the results observed here or strengthen the
and other immunosuppressants such as tacrolimus, observations seen.
respectively, were found (e-Table 7). These data suggest
There is also a paucity of data regarding the attributable
that whilst all immunosuppressants may present a risk
risk for NTM-PD, and meta-analyses can only evaluate
for NTM infection, inhaled corticosteroids present the
risk factors previously identified in other studies. There
greatest risk. What cannot be known from these data,
was an under-representation of risk factors in published
however, is if the greatest risk of NTM-PD from
data that run counter to clinical expertise, such as
immunosuppression due to inhaled corticosteroids is a
gastroesophageal reflux disease, and studies did not
consequence of immunosuppressant therapy or the
specifically report on symptomatology and how it may
underlying rationale for use.
be associated with NTM-PD, despite this being an
A range of other factors suggestive of an association did important element in defining who should be tested for
not reach statistical significance, and whether this is NTM. Data derivation in defined patient groups with
because no association truly exists or because lack of underlying lung disease such as CF may mean the data
significance is driven by a lack of available data, and data obtained are not representative of a more general
of high enough quality for interrogation and analysis, is population, and some identified risk factors cannot be
unclear. For data to be appropriate for such analysis, considered to be causal for NTM-PD. Given that many
likely they would need to be obtained prospectively, patients with NTM-PD are known to be older with
using predefined validated outcome measures. Similarly, comorbid conditions such as a respiratory disease, how
studies exploring macrolide use demonstrated a the interplay of multiple factors may increase the risk of
marginal protective effect for NTM-PD. The three disease in an individual remains unknown. Finally,

1122 Original Research [ 164#5 CHEST NOVEMBER 2023 ]

NTM-PD diagnosis is complex, and although all individualize the decision to treat pharmacologically
included studies were published after the initial or provide regular monitoring.
development of the NTM diagnostic guidelines in 2007,
there is a lack of uniformity in the definitions of NTM- Funding/Support
PD among the publications included. The study was funded by Insmed B.V.

Interpretation Financial/Nonfinancial Disclosures

These data suggest that, in line with current clinical The authors have reported to CHEST the following: M.
thinking, comorbid respiratory disease is associated R. L. reports receiving honorarium from Insmed,
with a high risk for NTM-PD, but other factors such AstraZeneca, Chiesi, Grifols, Savara, Armata; J. K. Q. has
as cancer and immunosuppression also predispose received grants from The Health Foundation, Medical
patients to an increased risk of disease. Further Research Council, GlaxoSmithKline, Bayer, Boehringer
studies to explore the impact of multiple possible risk Ingelheim, Asthma UK-British Lung Foundation, HDR
factors that predispose individual patients to NTM- UK, Chiesi, and AstraZeneca and personal fees for
PD would be of interest and warranted. advisory board participation or speaking fees from
Understanding the risk factors for NTM-PD, as GlaxoSmithKline, Boehringer Ingelheim, AstraZeneca,
explored in this meta-analysis, may help to identify Chiesi, Insmed, and Bayer; J. v. I. reports honorarium for
patients who do not yet have notable symptoms and speaking or advisory boards from Janssen
to initiate testing to diagnose NTM-PD early. In line Pharmaceuticals, Insmed, Spero Therapeutics, and
with this, additional resources are needed to improve Paratek; R. v. d. L. is an employee of Insmed B.V.; M. O.
clinician knowledge surrounding appropriate is an employee of Insmed Germany GmbH; R. C. and A.
management after diagnosis, emphasizing the need to K. are employees of Accuscript Consultancy.

Acknowledgments review. Semin Respir Crit Care Med. primary care data in the UK. Eur Respir J.
2018;39(3):336-342. 2020;56(4):2000045.
Author contributions: R. C. and A. K. were
responsible authors for data collation and data 3. Cowman S, van Ingen J, Griffith DE, 10. Ringshausen FC, Ewen R, Multmeier J,
preparation. All authors were responsible for Loebinger MR. Non-tuberculous et al. Predictive modeling of
mycobacterial pulmonary disease. Eur nontuberculous mycobacterial pulmonary
data analysis and interpretation. All authors
Respir J. 2019;54(1):1900250. disease epidemiology using German
equally contributed to the conception and health claims data. Int J Infect Dis.
outline of the manuscript. All authors critically 4. Asakura T, Ishii M, Ishii K, et al. Health- 2021;104:398-406.
reviewed each draft and equally contributed to related QOL of elderly patients with
the revisions. All authors reviewed and pulmonary M. avium complex disease in a 11. Griffith DE, Aksamit T, Brown-Elliott BA,
approved the final manuscript for submission. university hospital. Int J Tuberc Lung Dis. et al. An official ATS/IDSA statement:
2018;22(6):695-703. diagnosis, treatment, and prevention of
Role of sponsors: R. v. d. L. and M. O., nontuberculous mycobacterial diseases.
5. Mehta M, Marras TK. Impaired health- Am J Respir Crit Care Med. 2007;175(4):
employees of Insmed B. V., were involved in
related quality of life in pulmonary 367-416.
the design of the study, the collection and nontuberculous mycobacterial disease.
analysis of the data, and the preparation of Respir Med. 2011;105(11):1718-1725. 12. Wagner D, van Ingen J, Adjemian J, et al.
the manuscript. Annual prevalence and treatment
6. Park HY, Jeong BH, Chon HR, et al. Lung estimates of nontuberculous
Other contributions: Medical writing function decline according to clinical mycobacterial pulmonary disease in
assistance and editorial support was provided course in nontuberculous mycobacterial Europe: a NTM-NET collaborative study.
by Highfield, Oxford, England. This lung disease. Chest. 2016;150(6): Eur Respir J. 2014;44(Suppl 58):P1067.
assistance was sponsored by Insmed. 1222-1232.
13. Park TY, Chong S, Jung JW, et al. Natural
Additional information: The e-Figure and 7. Diel R, Lipman M, Hoefsloot W. High course of the nodular bronchiectatic form
e-Tables are available online under mortality in patients with Mycobacterium of Mycobacterium avium complex lung
"Supplementary Data." avium complex lung disease: a systematic disease: long-term radiologic change
review. BMC Infect Dis. 2018;18(1):206. without treatment. PLoS One.
8. van Ingen J, Obradovic M, Hassan M, 2017;12(10):e0185774.
References et al. Nontuberculous mycobacterial lung 14. Daley CL, Iaccarino JM, Lange C, et al.
1. Hoefsloot W, van Ingen J, Andrejak C, disease caused by Mycobacterium avium Treatment of nontuberculous
et al. The geographic diversity of complex: disease burden, unmet needs, mycobacterial pulmonary disease: an
nontuberculous mycobacteria isolated and advances in treatment developments. official ATS/ERS/ESCMID/IDSA clinical
from pulmonary samples: an NTM-NET Expert Rev Respir Med. 2021;15(11): practice guideline. Eur Respir J.
collaborative study. Eur Respir J. 1387-1401. 2020;56(1):2000535.
2013;42(6):1604-1613. 9. Doyle OM, van der Laan R, Obradovic M, 15. Andréjak C, Nielsen R, Thomsen V, et al.
2. Zweijpfenning SMH, Ingen JV, et al. Identification of potentially Chronic respiratory disease, inhaled
Hoefsloot W. Geographic distribution of undiagnosed patients with corticosteroids and risk of non-
nontuberculous mycobacteria isolated nontuberculous mycobacterial lung tuberculous mycobacteriosis. Thorax.
from clinical specimens: a systematic disease using machine learning applied to 2013;68(3):256-362.

chestjournal.org 1123
16. Haworth CS, Banks J, Capstick T, et al. 28. Kabbani D, Kozlowski HN, Cervera C, nontuberculous mycobacterial lung
British Thoracic Society guidelines for et al. Granuloma in the explanted lungs: disease among patients with chronic
the management of non-tuberculous Infectious causes and impact on post-lung obstructive pulmonary disease:
mycobacterial pulmonary disease transplant mycobacterial infection. Development of a predictive algorithm
(NTM-PD). Thorax. 2017;72(Suppl 2): Transpl Infect Dis. 2020;22(3):e13262. using claims data. Am J Respir Crit Care
ii1-ii64. 29. Liao TL, Lin CF, Chen YM, Liu HJ, Med. 2021;203(9):A3940.
17. Moher D, Shamseer L, Clarke M, et al. Chen DY. Risk factors and outcomes of 41. Wu J, Zhang Y, Li J, et al. Increase in
Preferred reporting items for systematic nontuberculous mycobacterial disease nontuberculous mycobacteria isolated in
review and meta-analysis protocols among rheumatoid arthritis patients: a Shanghai, China: Results from a
(PRISMA-P) 2015 statement. Syst Rev. case-control study in a TB-endemic area. population-based study. PLoS One.
2015;4(1):1. Sci Rep. 2016;6:29443. 2014;9(10):e109736.
18. G.A. Wells, B. Shea, D. O’Connell, et al. 30. Lim SY, Lee YJ, Park JS, et al. Association 42. Xu J, Li P, Zheng S, Shu W, Pang Y.
The Newcastle-Ottawa Scale (NOS) for of low fat mass with nontuberculous Prevalence and risk factors of pulmonary
assessing the quality of nonrandomised mycobacterial infection in patients with nontuberculous mycobacterial infections
studies in meta-analyses, 2014. Accessed bronchiectasis. Medicine (Baltimore). in the Zhejiang Province of China.
September 27, 2023. https://www.ohri.ca/ 2021;100(14):e25193. Epidemiol Infect. 2019;147:e269.
programs/clinical_epidemiology/oxford.asp 31. Liu VX, Winthrop KL, Lu Y, et al.
43. Brode SK, Campitelli MA, Kwong JC,
19. Higgins JPT, Thomas J, Chandler J, et al, Association between inhaled
et al. The risk of mycobacterial infections
eds. Cochrane Handbook for Systematic corticosteroid use and pulmonary
associated with inhaled corticosteroid use.
Reviews of Interventions version 6.3 nontuberculous mycobacterial infection.
Eur Respir J. 2017;50(3):1700037.
(updated February 2022). Cochrane; 2022. Ann Am Thorac Soc. 2018;15(10):
www.training.cochrane.org/handbook 1169-1176. 44. Evangelatos G, Bamias G, Kitas GD,
20. Whiting P, Savovic J, Higgins JP, et al. 32. Longworth SA, Vinnard C, Lee I, et al. Kollias G, Sfikakis P. The second decade of
ROBIS: A new tool to assess risk of bias in Risk factors for nontuberculous anti-TNF-a therapy in clinical practice:
systematic reviews was developed. J Clin mycobacterial infections in solid organ new lessions and future directions in the
Epidemiol. 2016;69:225-234. transplant recipients: a case-control study. COVID-19 era. Rheumatol Int.
Transpl Infect Dis. 2014;16(1):76-83. 2022;42(9):1493-1511.
21. Tan Y, Deng Y, Yan X, et al.
Nontuberculous mycobacterial pulmonary 33. Miller D, Powell W, Zhang Q, Garshick E. 45. Aliberti S, Spotti M, Sotgiu G, et al. A first
disease and associated risk factors in Predictors of nontuberculous look at the italian registry on pulmonary
China: A prospective surveillance study. mycobacterial lung disease in US veterans. non-tuberculous mycobacteria (NTM)-
J Infect. 2021;83(1):46-53. Chest. 2017;152(4):A153. IRENE. Eur Respir J. 2020;56:4344.
22. Axson EL, Bual N, Bloom CI, Quint JK, 34. Mirsaeidi M, Hadid W, Ericsoussi B, 46. Nagano H, Kinjo T, Nei Y, et al.
Bual N. Risk factors and secondary care Rodgers D, Sadikot RT. Non-tuberculous Causative species of nontuberculous
utilisation in a primary care population mycobacterial disease is common in mycobacterial lung disease and
with non-tuberculous mycobacterial patients with non-cystic fibrosis comparative investigation on clinical
disease in the UK. Eur J Clin Microbiol bronchiectasis. Int J Infect Dis. features of Mycobacterium abscessus
Infect Dis. 2019;38(1):117-124. 2013;17(11):e1000-e1004. complex disease: a retrospective analysis
35. Shah SK, McAnally KJ, Seoane L, et al. for two major hospitals in a subtropical
23. Adjemian J, Olivier KN, Prevots DR. region of Japan. PLoS One. 2017;12(10):
Epidemiology of pulmonary Analysis of pulmonary non-tuberculous
mycobacterial infections after lung e0186826.
nontuberculous mycobacterial sputum
positivity in patients with cystic fibrosis in transplantation. Transpl Infect Dis. 47. Pedrero S, Tabernero E, Arana-Arri E,
the United States, 2010-2014. Ann Am 2016;18(4):585-591. et al. Changing epidemiology of
Thorac Soc. 2018;15(7):817-826. 36. Song JH, Kim BS, Kwak N, Han K, Yim JJ. nontuberculous mycobacterial lung
Impact of body mass index on disease over the last two decades in a
24. Binder AM, Adjemian J, Olivier KN,
development of nontuberculous region of the Basque Country. ERJ Open
Prevots DR. Epidemiology of
mycobacterial pulmonary disease. Eur Res. 2019;5(4):00110-2018.
nontuberculous mycobacterial infections
and associated chronic macrolide use Respir J. 2021;57(2):2000454. 48. Pyarali FF, Schweitzer M, Bagley V, et al.
among persons with cystic fibrosis. Am J 37. Uno S, Asakura T, Morimoto K, et al. Increasing non-tuberculous mycobacteria
Respir Crit Care Med. 2013;188(7): Comorbidities associated with infections in veterans with COPD and
807-812. nontuberculous mycobacterial disease in association with increased risk of
25. Brode SK, Jamieson FB, Ng R, et al. Japanese adults: A claims-data analysis. mortality. Front Med (Lausanne). 2018;5:
Increased risk of mycobacterial infections BMC Pulm Med. 2020;20(1):262. 311.
associated with anti-rheumatic 38. Uwamino Y, Nishimura T, Sato Y, et al. 49. Ringshausen FC, Wagner D, de Roux A,
medications. Thorax. 2015;70(7):677-682. Low serum estradiol levels are related to et al. Prevalence of nontuberculous
26. Cavalli Z, Reynaud Q, Bricca R, et al. High Mycobacterium avium complex lung mycobacterial pulmonary disease,
incidence of non-tuberculous disease: A cross-sectional study. BMC Germany, 2009-2014. Emerg Infect Dis.
mycobacteria-positive cultures among Infect Dis. 2019;19(1):1055. 2016;22(6):1102-1105.
adolescent with cystic fibrosis. J Cyst 39. Viviani L, Harrison MJ, Zolin A, 50. Kim T, Yoon JH, Yang B, et al. Healthcare
Fibros. 2017;16(5):579-584. Haworth CS, Floto RA. Epidemiology of utilization and medical cost of
27. Chu H, Li B, Zhao L, et al. Chest imaging nontuberculous mycobacteria (NTM) gastrointestinal reflux disease in non-
comparison between non-tuberculous and amongst individuals with cystic fibrosis tuberculous mycobacterial pulmonary
tuberculosis mycobacteria in sputum acid (CF). J Cyst Fibros. 2016;15(5):619-623. disease: a population-based study, South
fast bacilli smear-positive patients. Eur 40. Wang P, Marras TK, Alemao E, et al. Korea 2009-2017. Front Med. 2022;9:
Respir J. 2015;46:2429-2439. Identifying potentially undiagnosed 793453.

1124 Original Research [ 164#5 CHEST NOVEMBER 2023 ]