Agreement between rapid on-site evaluation and the final cytological diagnosis of salivary gland specimens

Objective: Rapid on-site evaluation (ROSE) has been shown to be beneficial in determining the adequacy and a preliminary diagnosis in multiple organ systems. However, little is known regarding the diagnostic values and agreement of ROSE compared with a final cytological diagnosis in salivary gland fine needle aspiration (FNA). The aim of the present study was to evaluate agreement and compare accuracy between ROSE and a final cytological diagnosis of salivary gland FNA, using a final histological diagnosis as a gold standard. Methods: All patients with salivary gland lesions who underwent FNA with ROSE during 2009 to 2013 were evaluated. The patient demographic data, clinical characteristics, ROSE, final cytological diagnosis and final histological diagnosis were obtained. Agreement and accuracy were assessed by Kappa statistic and receiver-operating characteristic (ROC) curve analysis, respectively. Results: Of 386 FNAs with ROSE, 248 (64%) lesions were from parotid glands, and 171 (44%) had histological follow-up results. Agreement between ROSE and the final cytological diagnosis was good to excellent (simple kappa = 0.76; linear weighted kappa = 0.81). There were discrepancies in 25 lesions (6.8%). Of these, 10 (2.7%) were major discrepancies, resulting in a change in clinical management. Only one of 36 malignant lesions by ROSE was downgraded to suspicious. No significant difference was found in accuracy between two interpretations (area under the curve 0.82 vs 0.84, P=.17). Conclusion: These findings suggest that, in addition to an excellent agreement and comparably good accuracy between ROSE and final cytological evaluation, malignant results of ROSE may be useful to facilitate an early clinical decision.