Geographical heterogeneity in the American Joint committee on Cancer oral cancer staging and prognostic implications

•AJCC 8 increased complexity of staging with no improvement in predictive value.•Geographical variations in DOI & ENE resulted in different rates of stage migration.•Heterogeneity increased when AJCC 8 was applied to a diverse population.•This may result in inaccurate and unreliable prognostication. The AJCC 8th edition (AJCC 8) has introduced depth of invasion (DOI) and extranodal extension (ENE) into staging for oral squamous cell carcinoma (OSCC). Although validations have been performed on institutional datasets have shown a good performance, particularly in early OSCC, there have been no studies on diverse patient populations that determine the impact on prognostic heterogeneity. Retrospective analysis of 4710 patients with oral squamous cell carcinoma (OSCC) treated with surgery +/- adjuvant therapy in 8 institutions in Australia, North America and Asia. With overall survival (OS) as endpoint, the prognostic performance of AJCC 7th and 8th editions were compared using Akaike Information Criterion (AIC), Bayesian Information Criteria (BIC), Harrell's concordance index (C-index). When comparing AJCC 8 to AJCC 7, the heterogeneity in prediction of OS increased for T-category and N-category while remaining unchanged for TNM staging, suggesting AJCC 8 increased complexity with no improvement in predictive value. There were significant differences in median DOI and incidence of ENE between geographical regions, resulting in dissimilar rates of stage-migration when adopting AJCC 8. In an attempt to improve prognostic performance, AJCC 8 introduced more variables; however heterogeneity in these results in significant geographical differences in model discrimination and performance. Caution should be applied as this may result in inaccurate and unreliable prognostic predictions that may impact treatment recommendations.