Related Experiment Video
Updated: Oct 29, 2026
Establishment and Evaluation of a Risk Prediction Model for Pathological Escalation of Gastric Low-Grade Intraepithelial Neoplasia
Published on: February 16, 2024
WHO three-tier versus binary grading for predicting malignant transformation in oral epithelial dysplasia: a
Pawat Sripodok1, Nichapat Panutyotin1, Saharat Liampeng2
1Department of Oral and Maxillofacial Pathology, Faculty of Dentistry, Mahidol University, Bangkok, Thailand.
Background:
Grading of oral epithelial dysplasia (OED) is essential for guiding clinical management and predicting risk of malignant transformation (MT) to oral squamous cell carcinoma. While the World Health Organization (WHO) three-tier system (mild/moderate/severe) and binary system (low-grade [LG]/high-grade [HG]) are both used clinically, no study has compared their prognostic performance across all grading categories using network meta-analysis (NMA). This study aimed to compare their prognostic performances in predicting MT risk within and between both systems.
Methods:
Systematic search of six databases was conducted to identify cohort studies reporting MT rates of OED graded by three-tier and/or binary systems. The performance of each system was assessed by estimating the pooled area under curve (AUC). Multilevel mixed-effects logistic regression-based NMA estimated odds ratios (OR) for MT risk.
Results:
Forty-one studies encompassing 12,964 lesions were included. The binary system showed marginally higher discriminative power (AUC: 0.68) than modified WHO models (AUC: 0.64). From NMA, within-system analyses confirmed statistical significance including HG versus LG dysplasia (OR :4.63), severe versus mild dysplasia (OR: 4.42), moderate versus mild (OR: 2.66) and moderate versus severe dysplasia (OR: 1.66). Cross-system comparisons confirmed prognostic concordance at upper and lower categories (severe ≈ HG; mild ≈ LG) while moderate dysplasia showed significant differences from both LG and HG categories (ORs: 2.58 and 1.79, respectively).
Conclusions:
This NMA identifies moderate dysplasia as a distinct prognostic entity with MT risk trending closer to severe dysplasia/HG. While the binary system offers practical simplicity, it obscures moderate dysplasia, a distinct prognostic entity. The distinct MT risk of moderate dysplasia suggests that proactive intervention or shortened follow-up intervals may be considered. Pathologists should be aware of the limitations of applied grading system as this influences clinicians in interpreting MT risk and determining appropriate management for patients.
