Needle-based confocal laser endomicroscopy for real-time granuloma detection

Background and Objective: Needle-based confocal laser endomicroscopy (nCLE) allows real-time microscopic imaging at the needle tip. nCLE malignancy criteria are used for tool-in-lesion confirmation during bronchoscopic lung nodule analysis. However, to date, nCLE criteria for granulomas are lacking. The aim was to identify and validate nCLE granuloma criteria and assess if blinded raters can distinguish malignant from granulomatous nCLE videos. Methods: In patients with suspected sarcoidosis, nCLE-imaging of mediastinal lymph nodes was performed during endoscopic ultrasound procedures, followed by needle aspiration. nCLE granuloma criteria were identified by comparison with pathology and final diagnoses. Additionally, nCLE-videos of granulomatous lung nodules part of prospective trials and clinical care were compared to the proposed nCLE granuloma criteria. Blinded raters validated nCLE videos of sarcoid and reactive mediastinal lymph nodes and malignant and granulomatous lung nodules twice. Results: Granuloma criteria were identified (brighter-toned, homogeneous and well-demarcated lesions) based on nCLE-imaging in 14 sarcoidosis patients. Raters evaluated 26 nCLE-videos obtained in lymph nodes (n = 15 sarcoidosis; n = 11 reactive and total of 260 ratings). Granuloma criteria were recognized with 88% accuracy. The inter-observer (kappa = 0.63, 95% CI 0.54-0.72) and intra-observer reliability (kappa = 0.70 +/- 0.06) were substantial. Based on 12 nCLE-videos obtained in lung nodules (n = 4 granulomas, n = 6 malignancy, n = 2 malignancy + granulomas and total of 120 ratings) granuloma and malignancy criteria were recognized with 92% and 75% accuracy. Conclusion: nCLE imaging facilitates real-time granuloma visualization. Blinded raters accurately and consistently recognized granulomas on nCLE-imaging and distinguished nCLE granuloma criteria from malignancy. Our data show the potential of nCLE as a real-time bronchoscopic guidance tool for lung nodule analysis.