Variability in the Management of Adults With Pulmonary Nontuberculous Mycobacterial Disease

Abstract

(See the Editorial Commentary by Salfinger and Somoskovi on pages 1138–40.)

There are more than 190 nontuberculous mycobacteria (NTM) species [1, 2], and although most are nonpathogenic, some cause serious human disease [3]. Nontuberculous mycobacteria are primarily environmental saprophytes that are found in soil, water, and dust and may be acquired via inhalation, ingestion, or direct inoculation. Pulmonary NTM disease is most common in adults who are negative for human immunodeficiency virus (HIV) [4].

In many countries, particularly where the incidence of tuberculosis (TB) is low, both the prevalence of pulmonary NTM and the mortality rate associated with NTM infections are increasing [4–8]. A US study of Medicare Part B beneficiaries showed that the prevalence of NTM increased from 20 to 47 per 100 000 persons between 1997 and 2007, an increase of 8.2% per year [5]. A more recent report estimated that the number of pulmonary NTM cases in the United States increased by at least 2-fold between 2010 and 2014 [9]. Limited data suggest that pulmonary NTM is associated with increased risk of hospitalization and reduced quality of life [10, 11]. Current American Thoracic Society (ATS) and Infectious Diseases Society of America (IDSA) guidelines recommend treatment with multiple drugs for many months [12], although data demonstrating that such treatment reduces morbidity or mortality are lacking.

Most clinical studies of pulmonary NTM in North America have focused on the experiences of a single institution or region, potentially limiting generalizability of results [13, 14]. Meta-analyses and case series have analyzed composite data from smaller cohorts with some success [15–20], which still warrants multicenter epidemiological studies. Therefore, this multicenter epidemiological study was conducted to determine mycobacterial species and risk factors associated with pulmonary NTM, describe variability in patient management, determine treatment outcomes, and identify factors that affect treatment outcomes.

METHODS

Study Design and Setting

We performed a multicenter, 10-year, retrospective cohort study of patients with NTM evaluated at 7 Vaccine and Treatment Evaluation Units (VTEUs) in the United States. Participating institutions included Saint Louis University, Emory University, University of Iowa, Baylor College of Medicine, Duke University, Vanderbilt University Medical Center, and Kaiser Permanente Washington Health Research Institute. The VTEU network is funded under contract by the National Institute of Allergy and Infectious Diseases and has been a ready resource for conducting clinical trials and research of vaccine and treatments for infectious diseases since 1962. The study was approved by the institution review boards of all participating VTEUs.

Inclusion and Exclusion Criteria

Medical records of patients who were 18 years of age or older with NTM isolated from a respiratory specimen and in whom the diagnosis of pulmonary NTM was made between 1 January 2005 and 1 July 2015 were eligible for the study. Exclusion criteria were history of HIV infection and concurrent diagnosis of TB as determined by culture or molecular tests, such as polymerase chain reaction at the time of NTM diagnosis or during follow-up visits.

Data Collection

Patient data were abstracted through medical record review by licensed nurses, clinical research coordinators, subspecialty training fellows, or physicians who were trained in protocol-specific data collection procedures. Patients with NTM were identified for review of eligibility using International Classification of Diseases (ICD), 9th and 10th revision, codes. All patients had NTM culture-positive respiratory specimens. Clinical, laboratory, radiology, and treatment data were evaluated through 30 months post-NTM diagnosis using electronic data collection forms standardized across sites by the Statistical and Data Coordinating Center (SDCC; The Emmes Corporation, LLC). Abstracted data were reviewed and queried by the SDCC for data quality and adherence to protocol data collection procedures across sites. All participating sites used laboratories that adhere to Clinical and Laboratory Standards Institute guidelines for drug susceptibility testing. More than 3986 medical records of patients with possible pulmonary NTM were reviewed, and data were collected from 322 patients who were diagnosed with pulmonary NTM. Two sites that had fewer than 100 total eligible patients included all patients; 4 sites randomly selected patients’ records from each of the study years; and 1 site randomly selected records from 3 time periods (2005–2008, 2009–2012, 2013–2015). Upon further review of abstracted data, 18 patients were unlikely to have pulmonary NTM, and 2 did not have positive NTM culture results recorded. Therefore, the data from the remaining 302 patients were analyzed (Figure 1).

Figure 1.

Schematic of total medical records reviewed and the number of patients with pulmonary NTM who were negative for HIV analyzed for different outcome measures. Abbreviations: HIV, human immunodeficiency virus; NTM, nontuberculous mycobacteria; PNTM, pulmonary nontuberculous mycobacteria.

Open in new tabDownload slide

Definitions

Microbiologic cure was defined as 1 or more negative sputum or bronchoalveolar lavage (BAL) culture without reversion to culture positivity after 1 year of anti-NTM therapy [12, 21]. Clinical cure was defined as patient-reported or objective improvement in symptoms sustained until the end (minimum) of NTM treatment [21, 22]. The presence of observed or reported symptoms was documented by review of clinical case records, and changes in symptoms (improved, worsened, or no change) from the previous visit were adjudicated by reviewers. Comorbidities were defined as the recorded diagnoses or ICD diagnosis codes at the time of NTM diagnosis.

Statistical Analysis

Demographic and clinical characteristics, radiologic and microbiologic results, and treatment types and duration were reported using descriptive statistics in the full cohort and stratified by NTM species. Comorbidities, clinical symptoms, treatment characteristics, and NTM species were compared between NTM cases who did or did not meet criteria for clinical and microbiologic cure using Pearson’s chi-square or Fisher’s exact tests for categorical variables and Wilcoxon 2-sample tests for continuous variables. Unconditional logistic regression was used to estimate unadjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the associations of characteristics with cure outcomes. Analyses were repeated among the subgroup of patients with Mycobacterium avium-intracellulare complex (MAC) to describe characteristics associated with cure among this most frequent NTM type. Demographic, clinical, diagnostic, and treatment characteristics were also compared by vital status at the end of follow-up among all patients with NTM and among patients with MAC. Characteristics associated with cure and mortality in bivariate analyses at a significance level of less than 0.1 were included as independent variables in separate multivariate logistic regression models of cure and mortality as the dependent variables. Kaplan-Meier curves of time to mortality over follow-up were plotted and compared by log-rank test for select predictors that were identified as independently associated with vital status at the end of follow-up in the multivariate analyses.

RESULTS

Data from 297 patients negative for HIV diagnosed with pulmonary NTM were used for analysis (Figure 1). Of these 297 cases, 245 (82.5%) were begun on treatment; 45 (15.2%) were followed without treatment; and the treatment status of 7 (2.4%) patients was unknown. Table 1 shows demographic characteristics at the time of NTM diagnosis. Sixty-four percent of patients with NTM were female and 61% were white. The median age was 64 years (interquartile range [IQR], 55–73 years), and median body mass index (BMI) was 21.9 kg/m² (IQR, 19.5–25.0). Half of patients (49.5%) had public insurance, while 27 had private insurance. A total of 213 of 297 (72%) met ATS/IDSA diagnostic criteria for pulmonary NTM. The most frequent NTM species was MAC (247/297, 83.2%), followed by Mycobacterium kansasii (23/297, 7.7%), and Mycobacterium abscessus (12/297, 4%). Eight (2.7%) patients had more than 1 NTM species isolated.

Table 2 shows the clinical presentation features and comorbidities at the time of NTM diagnosis. Eighty-five percent of patients had 1 or more chronic diseases, and 20.5% had an acute comorbidity at the time of pulmonary NTM diagnosis. Thirty-seven (14.7%) were previously treated for pulmonary NTM. Nearly half (44.4%) used selected drugs such as vitamin D (56/297, 18.9%), statins (54/297, 18.2%), and glucocorticoids (32/297, 10.8%) at presentation. Cough was the most common presenting symptom (86.4%); most patients with cough (64.8%) reported it to be productive and 17.0% of patients reported hemoptysis. Half (50.6%) of the patients reported shortness of breath, 29.2% had unintentional weight loss, 26.9% had fatigue, and 14.4% had fever.

Radiologic and microbiologic results at diagnosis are shown in Table 3. At the time of diagnosis, pulmonary nodules, bronchiectasis, infiltrate/opacity, and/or cavitary lesion(s) were seen in 155 (52.2%), 128 (43.1%), 92 (31%), and 80 (26.9%) patients, respectively. A single culture-positive sputum specimen was used to satisfy the microbiologic criteria for NTM diagnosis in 116 (40.6%) patients. More than 1 culture-positive sputum specimen was used in 36 (12.6%) patients, and a culture-positive bronchial specimen (BAL and/or biopsy) was used for diagnosis in 134 (46.9%) patients. Eighty-six patients (28.5%) had drug susceptibility results at the time of diagnosis. Among the 86 tested NTM isolates, in vitro drug resistance to ethambutol, rifampin, and amikacin was 48.5%, 40%, and 43.1%, respectively. Baseline clarithromycin resistance among MAC isolates was 4.8%.

Treatment characteristics and clinical course are shown in Table 4. The majority (83%) of patients with pulmonary NTM were begun on anti-NTM treatment. Among treated patients, 210 (85.4%) received at least 3 drugs, 26 (10.6%) received 2 drugs, and 10 (4%) received 1 drug. Two hundred twenty-nine (93.1%) patients received an exclusively oral regimen, while the remaining patients received a combination of oral and parenteral therapy. On average, treatment was continued for 13.7 ± 9.6 (mean ± SD) months. One or more drugs in the treatment regimen were discontinued or replaced in 97 (32.7%) of the 245 patients who received antimicrobial treatment. Among these 97 patients, 56.6% experienced an adverse drug effect, making it the most common reason for a change in the treatment regimen. Gastrointestinal symptoms, such as nausea and diarrhea, were the most common adverse drug effects.

The total person-years of follow-up in this study was 506, with a median follow-up of 24.2 months (IQR, 14.5–29.7). Eighty-three (29.2%) patients with NTM required hospitalization after NTM diagnosis, and among those with known vital status, 44 (15.7%) died during follow-up. Approximately 45% of patients had symptom improvement at 3 months, and of 142 patients who had follow-up cultures performed at 12–24 months, 62.4% were negative. Among 138 treated patients with data available for evaluation of treatment outcome, 78 (56.5%) met criteria for clinical or microbiologic cure. Table 5 shows that adherence to ATS/IDSA diagnostic guidelines did not significantly change the treatment outcome, but adherence to the ATS/IDSA treatment guidelines was significantly more common in patients who were cured (OR, 4.6; 95% CI, 2.0–10.4; P < .001).

Factors associated with clinical cure alone are shown in Supplementary Table 1. Clinical symptoms (particularly fatigue), meeting ATS/IDSA diagnostic criteria, and adherence to ATS/IDSA treatment guidelines were more common in patients who experienced clinical cure (Supplementary Table 1). However, multivariable logistic regression analysis showed that daily treatment course, meeting ATS/IDSA diagnostic criteria, adherence to ATS/IDSA treatment guidelines, or number of antibiotics administered did not predict clinical cure (Supplementary Table 2). Because MAC was the most common NTM isolated from our patients, we performed a subgroup analysis (Supplementary Table 3). Patients with MAC with a radiologic finding of tree-in-bud opacities and those who received ethambutol or rifampin/rifabutin in the initial treatment regimen were more likely to be cured (OR [95% CI], 2.5 [1.1–5.6] and 3.2 [1.1–9.4], respectively). The number and frequency of follow-up cultures are shown in Supplementary Table 4. Factors associated with microbiologic cure are shown in Supplementary Table 5. As expected, patients who had microbiologic cure had significantly shorter duration of treatment (P < .05) and the number of patients with pulmonary MAC was significantly higher in the group who failed treatment (P < .05).

Among 44 patients who died, 34 (77.3%) had pulmonary disease due to MAC; therefore, we analyzed predictors of mortality among patients with MAC (Table 6). Of 221 patients with MAC whose vital status over the follow-up period was known, survivors were more likely to be female (OR, .34; 95% CI, .16–.72; P = .005) and have tree-in-bud opacities (OR, .21; 95% CI, .07–.64; P = .002) on chest imaging. In addition, survivors were less likely to have weight loss (OR, 3.43; 95% CI, 1.59–7.42; P = .003), chronic kidney disease (OR, 5.96; 95% CI, 1.61–22.04; P = .012), and infiltrates/opacities (OR, 2.26; 95% CI, 1.05–4.86; P = .04) or cavitation (OR, 3.43; 95% CI, 1.58–7.44; P = .002) on chest imaging. Multivariate logistic regression analysis (Table 7) showed that female sex (adjusted OR [aOR], .37; 95% CI, .16–.85) and tree-in-bud opacities (aOR, .28; 95% CI, .09–.87) on imaging studies were associated with increased odds of survival, and pulmonary cavitation (aOR, 3.18; 95% CI, 1.37–7.37) was associated with decreased odds of survival in patients with pulmonary NTM. Kaplan-Meier analysis estimated that overall survival was higher for females and patients without cavitation with a log-rank test P of .02 and .002, respectively (Figure 2).