A head-to-head comparison of four grading systems for oral epithelial dysplasia.

Abstract

Aims: Oral epithelial dysplasia (OED) carries a risk of malignant transformation to oral squamous cell carcinoma. Clinical risk stratification for these patients is challenging, and reliant upon histological grading. The World Health Organisation (WHO) grading system is the current gold standard, although the binary system, two- and six-point prognostic models have also been proposed. This study assesses the interobserver agreement and malignant transformation outcomes for these four grading systems.

Methods and results: Up to 5 years of outcome data were collected for this retrospective cohort of 137 patients. Archived slides were reviewed by three pathologists, and grades for the WHO, binary, two- and six-point systems were assigned. Interobserver agreement was assessed with Light's kappa coefficient. Kaplan-Meier and Cox regression survival analyses were used to assess the correlation of each grading system with malignant transformation. The WHO, binary, two- and six-point systems had kappa coefficients of 0.42, 0.31, 0.17 and 0.41, respectively. All grading systems stratified lesions by malignant transformation risk, except the two-point model. Moderate OED (WHO) did not show an increased risk of malignant transformation, while severe OED had a hazard ratio (HR) of 13.7 (P = 0.02). The high-risk category for the binary and six-point systems had HRs of 4.67 (P = 0.03) and 5.28 (P = 0.03), respectively.

Conclusions: The interobserver agreement of the WHO, binary and six-point systems is comparable. The six-point and binary systems provided the most useful risk stratification. This study highlights the potential value of the six-point prognostic model for OED grading, which has comparable performance with the current gold standard.