Abstract

Abstract Worldwide, approximately 405 000 cases of oral cancer (OSCC) are diagnosed each year, with a rising incidence in many countries. Despite advances in surgery and radiotherapy, which remain the standard treatment options, the mortality rate has remained largely unchanged for decades, with a 5‐year survival rate of around 50%. OSCC is a heterogeneous disease, staged currently using the TNM classification, supplemented with pathological information from the primary tumour and loco‐regional lymph nodes. Although patients with advanced disease show reduced survival, there is no single pathological or molecular feature that identifies aggressive, early‐stage tumours. We retrospectively analysed 282 OSCC patients for disease mortality, related to clinical, pathological, and molecular features based on our previous functional studies [EGFR, αvβ6 integrin, smooth muscle actin (SMA), p53, p16, EP4]. We found that the strongest independent risk factor of early OSCC death was a feature of stroma rather than tumour cells. After adjusting for all factors, high stromal SMA expression, indicating myofibroblast transdifferentiation, produced the highest hazard ratio (3.06, 95% CI 1.65–5.66) and likelihood ratio (3.6; detection rate: false positive rate) of any feature examined, and was strongly associated with mortality, regardless of disease stage. Functional assays showed that OSCC cells can modulate myofibroblast transdifferentiation through αvβ6‐dependent TGF‐β1 activation and that myofibroblasts promote OSCC invasion. Finally, we developed a prognostic model using Cox regression with backward elimination; only SMA expression, metastasis, cohesion, and age were significant. This model was independently validated on a patient subset (detection rate 70%; false positive rate 20%; ROC analysis 77%, p < 0.001). Our study highlights the limited prognostic value of TNM staging and suggests that an SMA‐positive, myofibroblastic stroma is the strongest predictor of OSCC mortality. Whether used independently or as part of a prognostic model, SMA identifies a significant group of patients with aggressive tumours, regardless of disease stage. Copyright © 2011 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.