Application of the Milan System for Reporting Salivary Gland Cytopathology: Experience of an academic institution in a tertiary academic medical center.

Background: The Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) represents a standardized reporting system for salivary gland lesions. The recent literature has demonstrated a wide range of data regarding range of malignancy (ROM) and interobserver variability. The objective of the current study was to evaluate the reproducibility and interobserver agreement of MSRSGC, and establish the ROM in a unique patient population residing within a designated Health Professional Shortage Area.

Methods: A total of 380 salivary gland fine-needle aspiration cases were obtained over a 3-year period. Corresponding cytology reports and slides were reviewed in a blinded fashion by a panel of cytopathologists and recategorized using MSRSGC. ROM was calculated by cytohistologic correlation in 176 cases. Agreement between review of reports and slides and interobserver reliability were determined using kappa statistics.

Results: The ROMs per MSRSGC category based on review of reports and slides were as follows: 4% and 0%, respectively, for nonneoplastic; 22% and 0%, respectively, for nondiagnostic; 42.9% and 48%, respectively, for atypia of undetermined significance; 1.6% and 1.9%, respectively, for benign-neoplastic; 17.9% and 15.6%, respectively, for salivary gland neoplasm of uncertain malignant potential; 81.8% and 71.4%, respectively, for suspicious for malignancy; and 100% and 90.5%, respectively, for malignant. There was a 59.2% overall agreement between review of reports and slides with regard to recategorizing salivary gland lesions (kappa, 0.51). The interobserver reliability demonstrated a 64.6% agreement (weighted kappa, 0.59).

Conclusions: The ROMs at the study institution appeared comparable to those in the published literature. There was moderate overall agreement among cytopathologists and low interobserver agreement with regard to the indeterminate categories. Image-guided fine-needle aspiration specimens; rapid onsite adequacy; and integration of clinical, imaging, and ancillary studies can improve diagnostic accuracy among indeterminate lesions.