Putting morphology to the test: An established classification scheme reliably stratifies salivary gland cytology by risk of malignancy with substantial interobserver agreement

Interobserver variability in cytopathology: How much do we agree?

Interobserver variability remains a major challenge for cytopathologists despite the development of standardized reporting and classification systems. Indeed, whereas moderate-to-good interobserver agreement is generally achievable when the differential diagnosis between benign and malignant entities is straightforward, high levels of variability make the diagnostic interpretation of atypical and suspicious samples not consistent. This review explores the landscape of interobserver agreement in cytopathology across different anatomical sites.

Multi-institutional validation of a modified scheme for subcategorizing salivary gland neoplasm of uncertain malignant potential (SUMP)

BACKGROUND

The salivary gland neoplasm of uncertain malignant potential (SUMP) category in the Milan System is diagnostically challenging. This study aims to validate a modified scheme for subcategorizing SUMP in a large multi-institutional cohort.

METHODS

Retrospective review of salivary gland fine-needle aspirations (FNAs) from 10 institutions were classified based on the Milan System. Cases diagnosed as SUMP with available cytology slides and surgical follow-up were retrieved for review and subcategorized based on a modified scheme as follows: basaloid SUMP (B1: absent/scant nonfibrillary matrix; B2: presence of nonfibrillary/mixed-type matrix), oncocytic/oncocytoid SUMP (O1: with mucinous background; O2: without mucinous background), and SUMP not otherwise specified (NOS).

RESULTS

A total of 742 (7.5%) cases from 9938 consecutive salivary gland FNAs were classified as SUMP. Among them, 525 (70.8%) had surgical follow-up and 329 (62.7%) were available for review. The overall risk of malignancy (ROM) of SUMP was 40.4%. There were 156 cases (47.4%) subcategorized as basaloid SUMP with a ROM of 36.5%, 101 (30.7%) as oncocytic/oncocytoid SUMP with a ROM of 52.5%, and 72 (21.9%) as SUMP NOS with a ROM of 31.9%. The ROM of oncocytic/oncocytoid SUMP was significantly higher than basaloid SUMP (P = .0142) and SUMP NOS (P = .0084). No significant differences in ROM were noted between B1 and B2 (36.7% vs 36.4%, P = 1.0000) and O1 and O2 (65.2% vs 48.7%, P = .2349).

CONCLUSIONS

The ROM of oncocytic/oncocytoid SUMP was 52.5% and significantly higher than that of basaloid SUMP (36.5%, P = .0142) and SUMP NOS (31.9%, P = .0084), whereas no significant differences in ROM were noted for cases with different types of extracellular matrix or background material.

Salivary gland neoplasms with basaloid features in the era of the Milan system for reporting salivary gland cytology: Classification and interobserver agreement

BACKGROUND

The Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) has been shown to have moderate to good reproducibility for categorization of salivary gland fine-needle aspiration (FNA) specimens. Less is known of its accuracy and interobserver reproducibility for categorization of the diagnostically difficult group of basaloid neoplasms.

METHODS

Forty-five salivary gland specimens with a basaloid morphology (pleomorphic and monomorphic adenomas and adenoid cystic carcinomas) were independently assigned by seven cytopathologists to one of the MSRSGC categories. Interobserver agreement was assessed for average agreement, chance expected agreement and by Cohen's κ and diagnostic accuracy. Correlation of the salivary gland neoplasm of unknown malignant potential (SUMP) category with histologic diagnosis and benign or malignant designation along with interobserver reproducibility were calculated.

RESULTS

Average observed agreement for assignment to the MSRSGC was 46% and Cohen's κ = 0.2%. The SUMP category did not correlate with tumor type or with the benign or malignant nature of the neoplasm. Diagnostic specificity and sensitivity were 92% and 100% for consensus diagnosis, but were 76% and 77% for individual diagnoses.

CONCLUSION

The interobserver agreement in categorizing basaloid neoplasms by the MSRSGC is poorer than for salivary gland lesions overall. This reflects the difficulty in diagnosing basaloid neoplasms. Nonetheless, diagnostic accuracy appears similar to that of salivary gland neoplasms as a whole.

Analysis of the risk of malignancy associated with the basaloid and oncocytic subtypes of the salivary gland neoplasm of unknown malignant potential (SUMP) category in the Milan system

BACKGROUND

The salivary gland neoplasm of unknown malignant potential (SUMP) category reflects the cytomorphologic overlap and complexity of reporting salivary gland cytology in the Milan system. It includes neoplasms for which a diagnosis of a specific entity cannot be made and, more importantly, for which a carcinoma cannot be entirely excluded. For risk stratification, the subcategorization of SUMP based on the predominant cell type is recommended. This study was aimed at evaluating the risk of neoplasm (RON) and the risk of malignancy (ROM) of the basaloid and oncocytic subtypes of the SUMP category.

METHODS

A retrospective analysis of 482 salivary gland fine-needle aspirations from 2012 to 2019 resulted in 48 SUMP cases. The cytology of these cases was reviewed and reclassified as the basaloid or oncocytic subtype. Surgical follow-up was available for 36 cases. The RON and ROM for each subtype were calculated.

RESULTS

The RON and ROM were 100% and 23%, respectively, for monomorphic basaloid tumors and 88% and 58.8%, respectively, for monomorphic oncocytic tumors. The ROM for basaloid tumors was 8.3% without matrix/with minimal matrix and 60% with an nonfibrillary matrix. The ROM for oncocytic tumors was 50% without a cystic background and 60% with a cystic or mucinous background. The difference was not statistically significant for either of the subgroups.

CONCLUSIONS

Even though statistically not significant in our study, the differential ROMs within the oncocytic and basaloid subgroups help in the risk stratification of SUMP cases. Further subcategorization based on the stroma and background helps in limiting the differential diagnosis but does not necessarily add to the value of the risk stratification.