Concordance between pulmonologist performed rapid on-site evaluation (ROSE) and pathological diagnosis in EBUS-TBNA samples

Introduction: Sometimes it’s not easy have a pathologist to perform ROSE during an EBUS-TBNA procedure. Our objective was to analyze the concordance between ROSE performed by pulmonologists and the definitive pathological diagnosis. Methods: Retrospective analysis of EBUS-TBNA performed in our center, from february 2016 to september 2019, in which ROSE was performed by a pulmonologist through a fresh staining with panoptic rapid immersion technique. Results were reported as valid/suggestive, valid/non-suggestive and non-valid/non-suggestive, depending on lymphocytes and atypical cells presence. A comparison with the definitive pathological diagnosis is done, in all cases performed and those in which a malignancy diagnosis was obtained. Results: 453 EBUS-TBNA procedures analyzed in the study period, 198 met study criteria. Average age 64.08, 155 male sex. 437 punctures of hiliomediastinal lymph nodes (LN) were performed, average 3.66 punctures per LN. Punctured LN stations were 7 (n=140), 4R (n=104), 4L (n=58), 11L (n=34), 2R (n=31), 11R (n=29), 10R (n=19), 10L (n=8), 2L (n=8), 8 (n=4) and 3p (n=2). Valid/suggestive result was obtained in 104 patients (52.3%), valid/non-suggestive in 81 (40.9%) and non-valid /non-suggestive in 13 (6.6%). The overall concordance was considered as a moderate degree (kappa index 0.494). Concordance related to malignancy ROSE results (valid/suggestive) is considered as a high degree relation (kappa index 0.952), with sensitivity 98.9%, specificity 92.3% and NPV 92.3%. Conclusions: ROSE of EBUS-TBNA mediastinal LN cytological samples performed by pulmonologists show a high concordance in case of malignancy results.