Abstract Background Interstitial lung disease (ILD) is frequently associated with delayed evaluation despite advanced imaging findings such as honeycombing. Timely rheumatologic assessment is critical because ILD may represent the initial manifestation of a connective tissue disease (CTD). Identifying CTD enables appropriate therapy—such as antifibrotics or immunosuppressants—which can slow disease progression and improve outcomes. Methods Natural language processing (NLP) was applied to radiology narrative reports of chest CT scans over a 12-month period to identify the term “honeycombing.” A retrospective chart review was conducted using EPIC to collect demographics, CTD diagnoses, serologic testing, and treatment history. Patients aged 85 years or older were excluded. Results Among 153 charts containing the term “honeycombing,” 100 were confirmed as true positives, with 22 patients aged 85 or older excluded from analysis. Of the remaining 78 patients, 16 had a documented CTD diagnosis, with an average age of 68 years compared to 74 years in those without CTD; females represented 62.5% of the CTD group versus 25.8% of the non-CTD group. CTD diagnoses included 13 cases of rheumatoid arthritis (RA), 3 Sjögren’s syndrome, 1 lupus, and 1 inflammatory myopathy, with two patients having dual RA and Sjögren’s diagnoses. Serologic testing revealed that 31% of CTD patients were antinuclear antibody (ANA)-positive, 25% rheumatoid factor (RF)-positive, and 19% cyclic citrullinated peptide (CCP)-positive, while positivity rates in non-CTD patients were 15%, 8%, and 2%, respectively. Screening was incomplete: half of CTD patients lacked ANA testing, and only 6 of 16 received guideline-based ANA, RF, and CCP testing. Treatment patterns showed limited antifibrotic use (3 patients) and more frequent immunosuppressant therapy (10 patients), including mycophenolate, rituximab, and azathioprine, with three patients receiving combination therapy. Pulmonary evaluation occurred in 13 CTD patients, and 11 underwent pulmonary function tests, averaging two tests per patient. Conclusions Accurate ILD treatment depends on identifying underlying etiology. Rheumatologic evaluation can reclassify patients as CTD-ILD, guiding therapy that often combines immunosuppressants and antifibrotics. The American Thoracic Society recommends ANA, RF, and CCP testing in all idiopathic pulmonary fibrosis patients to exclude CTD-ILD; however, 62% of CTD patients in this cohort lacked complete biomarker assessment. NLP-based identification of honeycombing offers an opportunity to prompt rheumatologic evaluation, improve diagnostic accuracy, and optimize treatment strategies to slow ILD progression. This abstract is funded by: None
Mahally et al. (Fri,) studied this question.