Sintawat Wangsiricharoen M.D., Sunida Rewsuwan M.D., Nopporn Satabongkoch M.D., Samreung Rangdaeng M.D.
Department of Pathology, Faculty of Medicine, Chiang Mai University
Objective: To evaluate diagnostic agreement and compare the performance (accuracy) between rapid onsite evaluation (ROSE) and the ﬁ nal cytologic diagnosis of salivary gland specimens.
Methods: Patients with salivary gland lesions who underwent ﬁ ne-needle aspiration (FNA) with ROSE during 2009 to 2013 were evaluated. Patient clinical characteristics, ROSE, ﬁ nal cytologic diagnosis and histopathologic diagnosis were obtained. The cytologic diagnosis was categorized into six groups: benign, atypical, suspicious for malignancy, malignant, indeterminate and inadequate. Agreement and performance were assessed by Kappa statistic and receiver operating characteristic (ROC) curve analysis, respectively.
Results: A total of 347 patients (median age 52 years) underwent FNA with ROSE for 386 lesions including parotid glands (64.2%), submandibular glands (29.5%), and minor salivary glands (1.8%). On follow-up, 171 (44.3%) had the histopathologic diagnosis, 134 (34.7%) had clinical follow-up and 81 (21%) were lost to follow-up. Agreement between on-site and the ﬁ nal cytologic diagnosis was good to excellent (simple kappa = 76% [95% CI, 0.68-0.84]; weighted kappa = 81% [95% CI, 0.73-0.89]). The on-site interpretation was changed in the ﬁ nal cytologic diagnosis in 26 lesions (7.1%). The ﬁ nal cytologic interpretation yielded higher sensitivity (75% vs 69.23%) and similar speciﬁ city (91.59% vs 93.46%); however, no signiﬁ cant difference was found in performance between ROSE and the ﬁ nal cytologic diagnosis (area under the ROC curve = 82.39% vs 84.54%, p = 0.826).
Conclusions: There was excellent agreement and comparably good performance between rapid on-site and the ﬁ nal cytologic evaluation in detecting malignant lesions of salivary glands.