BetaEntity Annotation Prototype
← Back to diseases

Annotated abstract

Serum KL-6 and lung ultrasound B-lines: a combined non-invasive model for screening and predicting interstitial lung disease in idiopathic inflammatory myopathy

rmdopen · 2026-05-21 · canonical JSON source

3 visible annotations · policy: published · automated confidence ≥ 75.00%

Document resource

Background Interstitial lung disease (ILD) is a major pulmonary complication of idiopathic inflammatory myopathy (IIM), where early diagnosis improves outcomes. While high-resolution CT (HRCT) remains the gold standard, its radiation exposure poses concerns. Serum Krebs von den Lungen-6 (KL-6) and lung ultrasound (LUS) B-lines offer non-invasive alternatives, though their optimal diagnostic cut-offs and combined utility for IIM-ILD remain undefined. This study aimed to establish these cut-offs, evaluate diagnostic performance and develop a clinical prediction model.Methods In this single-centre observational study, 162 patients diagnosed with IIM between 2020 and 2024 were enrolled. All underwent serum KL-6 testing, and 120 received systematic 50-point LUS examinations. Using HRCT as the reference standard for ILD diagnosis, receiver operating characteristics (ROC) curve analysis was used to determine optimal cut-offs and construct a combined nomogram.Results KL-6 levels were significantly elevated in the ILD group (n=113) compared with non-ILD (n=49). The optimal KL-6 cut-off was 553 U/mL (area under the curve (AUC)=0.895, sensitivity 69%, specificity 95.9%). For LUS B-lines number, 25 was recommended as the clinical threshold (sensitivity 98.8%). Their combination enhanced diagnostic performance (AUC=0.984). An online prediction model demonstrated strong clinical applicability. In an exploratory analysis of the rapid-progressive ILD subgroups, KL-6 and B-lines showed only modest predictive value (AUC ≈ 0.65).Conclusion KL-6 ≥553 U/mL and B-lines ≥25 are effective screening thresholds, with combined use significantly improving diagnostic accuracy. The prediction model provides a practical tool for early identification in similar clinical settings. To optimise and generalise its use, external validation in multicentre cohorts is warranted.