Methods Patients with NTM isolated from respiratory specimens between 2010–2013 at our centre were reviewed. Chest X-rays (CXR) nearest the date of positive NTM culture were read independently by two consultant Radiologists for 5 categories of abnormality (nodules, cavities, bronchiectasis, bronchial wall thickness [BWT] and consolidation) in each of 6 zones. A consensus result was agreed where discrepant. CXR results were recorded as “normal” or “abnormal” overall and for each category per zone. The total number of zones affected in all categories was summed to provide a measure of radiological extent of disease (with a maximum score of 30), e.g. a patient with cavitation in 2 zones and bronchiectasis in 3 would score 5/30. Results were compared to clinical and microbiological data (including time-to-positivity in liquid culture, TTP).

Results Of 79 patients, 44/79 (56%) were male, median age 63 years [IQR_53;75]. CXR was performed median 5-days [IQR_2;28] from sample collection. Inter-rater CXR agreement was 92%, kappa 0.57 [95% CI: 0.52–0.63]. 58/79 (75%) of subjects had an abnormal CXR [Table 1]: half had nodules and BWT, with 41%, 23% and 20% bronchiectasis, consolidation and cavities respectively. Using symptoms present in >1/3 patients, only sputum and difficulty breathing were significantly associated with CXR score (p = 0.04; p = 0.01). M.avium and M.intracellulare were the most common NTM isolated [Table 1]. The highest abnormal median score for CXR was for M.xenopi (6/30) followed by M.perigrinum (5/30), and M.abscessus (4.5/30). For M.kansasii there was significant correlation between CXR score and TTP (r = −0.82 ; p = 0.01) with cavitation being associated with a significantly lower TTP (4.5 vs 17.5 days; p = 0.03). CXR score did not predict whether or not a patient started treatment (OR 1.07; p = 0.40).

Conclusion CXR abnormalities were present in 75% of patients in whom NTM was isolated. Unlike M.tuberculosis, the extent of radiographic changes correlated poorly with clinical symptoms or mycobacterial burden. Better simple, repeatable measures of NTM disease severity are required.