The factors associated with mortality and progressive disease of nontuberculous mycobacterial lung disease: a systematic review and meta-analysis

Study selection

A total of 1648 records were identified using the Medline, Embase, Cochrane Central Register of Controlled Trials, and Web of Science databases. No additional records from other sources were included. After the removal of duplicate studies, 1029 studies were screened and 73 potentially eligible studies were retrieved for the full-text review. Finally, 41 studies satisfied the eligibility criteria (Fig. 1). The reasons for excluding the 32 studies are described in Appendix S1. Of the 41 studies, 40 were observational studies and one was an RCT. Of the observational studies, 34 were retrospective, four were prospective, and two had data from both the prospective and retrospective cohorts.

Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) flow diagram for the systematic review and meta-analysis.

Baseline characteristics of the included studies and patients

A total of 41 studies were included in this review, which evaluated factors associated with progressive disease and death in 10,452 patients (Table 1). Thirty-four studies were conducted in Asia, four studies were conducted in Europe, and three were conducted in North America. Most studies were conducted in Japan (n = 21), followed by South Korea (n = 9). The mean age was 64.9 years and 36.4% were men. M. avium complex (MAC) infection was the most common (85.1%), followed by M. abscessus (7.8%). In the radiologic pattern, 69.6% of the patients had nodules, bronchiectasis, or nodular bronchiectatic pattern, 17.5% had a cavity or fibrocavitary pattern, and 0.7% had consolidation.

Risk of bias assessment

The risk of bias (ROB) assessment for all included studies is described in Appendix S2. Most studies had high ROBs for study attrition due to lack of information on the sample attrition. Moreover, there were moderate-to-high ROBs in the study participation and adjustment for confounders, which was due to the low participation rate, selective enrollment in the study, and insufficient consideration for other factors.

All-cause mortality

A total of 7,103 patients from 23 studies were included in the review of factors associated with all-cause mortality^{9,11,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33}. The pooled adjusted RRs and unadjusted RRs, which estimated the association between factors and all-cause mortality, were visualized by forest plots (Fig. 2 and Figure S1). The pooled risk ratios for each factor are presented as a forest plot in Appendix S3. The factors significantly associated with all-cause mortality and progressive disease in univariable analysis were summarized in Appendix S4. Categorical variables and continuous variables were separately analyzed for the association with all-cause mortality and progressive disease. The definition of some categorical variables defined differently in each study is outlined in Appendix S5.

Forest plots displaying the pooled hazard ratios estimating the association between factors and all-cause mortality. (a) Analysis of adjusted hazard ratios. (b) Analysis of adjusted odds ratios. The numbers within the parentheses mean the numbers of studies analyzed. All adjusted risk ratios were estimated in multivariable analysis.

The overall rate of all-cause mortality was 20% (95% CI 17–24%). The heterogeneity of the effect estimate (I²) was 94.8%. The factors significantly associated with all-cause mortality, clinical or radiographic progressive disease were summarized in Table 2. Multivariable analysis showed that increasing age (adjusted HR = 1.052; 95% CI 1.031–1.074; I² = 19.6%; seven studies), the elderly (adjusted HR = 3.005; 95% CI 2.329–3.876; I² = 22.6%; five studies), male (adjusted HR = 2.406; 95% CI 1.900–3.047; I² = 46.3%; ten studies), and low body mass index (BMI) (adjusted HR = 1.934; 95% CI 1.581–2.366; I² = 0; four studies) had a significant association with increased all-cause mortality, while increasing BMI (adjusted HR = 0.832; 95% CI 0.788–0.878; I² = 0; six studies) was significantly associated with decreased all-cause mortality.

Among the underlying diseases, history of tuberculosis (TB) (adjusted HR = 2.749; 95% CI 1.341–5.637; I² = 0; two studies), diabetes (adjusted HR = 2.062; 95% CI 1.194–3.562; I² = 30.7%; three studies), chronic heart disease (adjusted HR = 1.959; 95% CI 1.093–3.509; I² = 61.3%; two studies), malignancy (adjusted HR = 2.213; 95% CI 1.680–2.914; I² = 11.7%; five studies), and systemic immunosuppression (adjusted HR = 2.126; 95% CI 1.311–3.450; I² = 0; two studies) were significantly associated with increased all-cause mortality in multivariable analysis. In one study, chronic liver disease was significantly associated with increased all-cause mortality (adjusted HR = 1.860; 95% CI 1.242–2.785) after adjusting for covariates⁹.

Among the patients’ symptoms, hemoptysis was significantly associated with decreased all-cause mortality after adjusting covariates in one study (adjusted HR = 0.542; 95% CI 0.414–0.709)²⁰.

Among radiologic findings, presence of cavity (adjusted HR = 2.380; 95% CI 1.866–3.037; I² = 0; 10 studies; adjusted OR = 3.176; 95% CI 1.369–7.369; one study) and consolidative pattern (adjusted HR = 4.895; 95% CI 2.997–7.996; I² = 0; two studies) were significantly associated with increased all-cause mortality in multivariable analysis.

Among the laboratory findings, acid-fast bacillus (AFB) smear positivity (adjusted HR = 2.456; 95% CI 1.460–4.130; I² = 0; two studies), increasing erythrocyte sedimentation rate (ESR) (adjusted HR = 1.020; 95% CI 1.016–1.024; I² = 0; two studies), and hypoalbuminemia (adjusted HR = 3.770; 95% CI 2.697–5.270; I² = 0; three studies) were significantly associated with increased all-cause mortality in multivariable analysis. In single studies, anemia (adjusted HR = 5.547; 95% CI 1.235–24.916)³¹, increasing platelet count (adjusted OR = 1.090; 95% CI 1.008–1.178)¹¹, increasing C-reactive protein (CRP) (adjusted HR = 1.220; 95% CI 1.058–1.407)¹⁹, high CRP (adjusted HR = 8.960; 95% CI 1.657–48.462)²⁸, and high ESR (adjusted HR = 1.849; 95% CI 1.140–2.999)²⁰ were significantly associated with increased all-cause mortality after adjusting for covariates.

After adjusting for covariates, treatment with rifamycin regimen in M. xenopi lung disease was significantly associated with decreased all-cause mortality (adjusted HR = 0.330; 95% CI 0.151–0.723) in one study¹⁵.

Clinical progressive disease with treatment

A total of 2,782 patients from 10 studies were included in the review of factors associated with clinical progressive disease with treatment^{12,17,34,35,36,37,38,39,40,41}. Figures 3 and S2 present the pooled adjusted RRs and unadjusted RRs, which estimated the association between factors and clinical progressive disease using forest plots. The pooled risk ratios for each factor were presented in Appendix S6.

Forest plots displaying the pooled hazard ratios estimating the association between factors and clinical progressive disease with treatment. (a) Analysis of adjusted hazard ratios. (b) Analysis of adjusted odds ratios. The numbers within the parentheses mean the numbers of studies analyzed. All adjusted risk ratios were estimated in multivariable analysis.

The overall rate of clinical progressive disease was 46% (95% CI 39%–53%; I² = 91.0%). In baseline demographics, low BMI was significantly associated with decreased clinical progression after adjusting for covariates in one study (adjusted OR = 0.515; 95% CI 0.310–0.855)³⁶. Increasing age was significantly associated with decreased clinical progression in multivariable analysis (adjusted HR = 0.976; 95% CI 0.967–0.985; I² = 0; three studies; adjusted OR = 0.950; 95% CI 0.920–0.980; one study).

Among underlying diseases, history of TB (adjusted HR = 1.230; 95% CI 1.009–1.499)³⁸ and Aspergillus co-infection (adjusted OR = 5.330; 95% CI 1.107–25.662)¹² were significantly associated with increased clinical progression after adjusting for covariates in single studies.

Regarding the patient’s symptoms, loss of weight (adjusted OR = 2.822; 95% CI 1.271–6.268; I² = 0; two studies) was significantly associated with increased clinical progression in multivariable analysis. In one study, cough (adjusted HR = 1.360; 95% CI 1.053–1.756) and increased sputum (adjusted HR = 1.470; 95% CI 1.131–1.911) were significantly associated with increased clinical progression after adjusting for covariates³⁸.

Among the radiologic findings, presence of cavity (adjusted HR = 3.460; 95% CI 2.273–5.267; one study; adjusted OR = 5.324; 95% CI 2.323–12.199; I² = 0; two studies) was significantly associated with increased clinical progression in multivariable analysis.

Among the laboratory findings, AFB smear positivity (adjusted OR = 2.132; 95% CI 1.393–3.263; I² = 0; two studies) was significantly associated with increased clinical progression in multivariable analysis.

Radiographic progressive disease

Total 1,439 patients from 11 studies were included in the review of factors associated with radiographic progressive disease^{20,30,42,43,44,45,46,47,48,49,50}. Figures 4 and S3 present the pooled adjusted RRs and unadjusted RRs, which estimated the association between factors and progressive disease using forest plots. The pooled risk ratios for each factor were presented in Appendix S7.

Forest plots displaying the pooled hazard ratios estimating the association between factors and radiographic progressive disease. (a) Analysis of adjusted hazard ratios. (b) Analysis of adjusted odds ratios. The numbers within the parentheses mean the numbers of studies analyzed. All adjusted risk ratios were estimated in multivariable analysis.

The overall rate of radiographic progressive disease was 43% (95% CI 31%–55%, I² = 95.1%). In single studies, increasing age (adjusted OR = 1.120; 95% CI 1.038–1.209)⁴⁹ and being elderly (adjusted OR = 2.980; 95% CI 1.041–8.534)⁴⁸ was significantly associated with increased radiographic progression after adjusting for covariates.

Interstitial lung disease (adjusted HR = 2.191; 95% CI 1.325–3.623) was significantly associated with increased radiographic progression after adjusting for covariates in one study²⁰.

After adjusting for covariates, presence of cavity (adjusted HR = 1.651; 95% CI 1.181–2.308; one study; adjusted OR = 3.283; 95% CI 1.405–7.673; I² = 0; two studies) was significantly associated with increased radiographic progression. Additionally, the consolidative pattern (adjusted OR = 16.150; 95% CI 4.048–64.429) was significantly associated with increased radiographic progression after adjusting for covariates in one study⁴⁹.

Among the laboratory findings, anemia (adjusted HR = 1.852; 95% CI 1.265–2.712)²⁰, high CRP (adjusted HR = 1.520; 95% CI 1.081–2.137)²⁰, and leukocytosis (adjusted OR = 3.440; 95% CI 1.139–10.390)⁴⁸ were significantly associated with increased radiographic progression after adjusting for covariates in single studies.

Heterogeneity between studies

Seventeen pooled effect estimates were identified as having substantial heterogeneity (Appendix S8), 10 of which were analyzed for all-cause mortality, four were analyzed for clinical progressive disease with treatment, and three were analyzed for radiographic progressive disease as an outcome.

Publication bias and certainty of evidence

Publication bias was observed in two estimates (adjusted HR of age and presence of cavity for all-cause mortality) when assessed by Egger’s and Begg’s tests (Appendix S9). After the trim and fill method, the overall effect size remains unchanged. The levels of certainty of evidence regarding associated factors are described in Appendix S10. All associated factors had very low or low levels of evidence certainty.