M. szulgai,M. simiae,M. peregrinum,

M. szulgai,M. simiae,M. peregrinum,M. scrofulaceum andM. terrae; and 16S ribosomal RNA
sequencing forM. shimoidei [13,14]. Patients with at least two separate culture-positive sputum
samples or one culture-positive bronchial washing or lung tissue sample during the study period
were included. Patients with a single positive sputum culture or PCR positive-culture negative
samples were excluded.
Diagnosis of NTM-LD
Diagnosis was performed by specialists in infectious disease and/or respiratory medicine according to 2007 American Thoracic Society (ATS)/Infectious Diseases Society of America
(IDSA) criteria [1]. In accordance with the criteria, patients with suspected infection with contaminated environmental Mycobacterium species were excluded from study participation. Patients who had been diagnosed with NTM-LD prior to the current study were also excluded.
Epidemiological Study
Among all the patients who had presented at the 11 hospitals during the study period, 975 had
been diagnosed with NTM-LD according to the 2007 ATS/IDSA criteria. After dividing the
prefecture into five regions on the basis of administrative boundaries and medical care zones,
the patients were classified into one region based on the residence listed in their medical records.
To estimate the annual incidence of NTM-LD in each region, data regarding the annual
population of each region were collected from the Nagasaki prefectural government, and data
regarding the incidence of all forms of tuberculosis collected from the healthcare center of each region. In the analysis of the radiological features, 228 patients were excluded due to lack of accurate information about chest computed tomography (CT) scan (Fig 1). Based on the review of the results of chest CT scan, the remaining 747 patients were classified into four patterns according to their radiological characteristics: the nodular-bronchiectatic (NB), cavitary (CAV), hypersensitivity pneumonitis-like, or unclassifiable pattern. The NB pattern and CAV pattern were defined by chest CT scan showing multifocal bronchiectasis with multiple small nodules or cavitary opacities, respectively. If a patient had the characteristics of both the NB and CAV patterns, the patient was classified in accordance with the dominant pattern. Using these data, geographic variations in NTM-LD incidence, isolates, and radiological features were assessed.
Patient Characteristics and Prognostic Factors
Of the 975 patients, 374 were excluded from analysis due to lack of sufficient clinical information in their medical records (Fig 1). The clinical course of the disease was assessed after 1 year of diagnosis. Patient outcome was evaluated by review of (1) clinical, (2) radiological and (3) microbiological responses. Clinical response was defined as resolution of symptoms such as fever, weight loss, cough, sputum, hemoptysis and general fatigue. Radiological response was defined as the clearing or improvement of opacity on chest radiographs/CT. Microbiological response was determined by culturing of sputum. Patients who had shown improvement or no change in the 3 factors and had experienced no deterioration were defined as having achieved “stability”. Patients who had not met the criteria for “stability” definition and had subsequently died from all causes were classified as having experienced “deterioration”.
Statistical Analysis
The data are presented as mean ± SD values. Statistical significance was evaluated using the
chi-square test or the two-tailed Mann-Whitney test, with the level of significance set at