Oral epithelial dysplasia grading and risk stratification: a systematic review of binary vs. three-tier systems and prognostic biomarkers

Oral epithelial dysplasia (OED) is the histologic hallmark of oral potentially malignant disorders (OPMDs). The conventional three-tier World Health Organization (WHO) grading system is limited by significant inter-observer variability, particularly in the moderate category, prompting proposals for an alternative binary grading system. To systematically review the diagnostic reliability, prognostic accuracy for malignant transformation (MT), and clinical utility of WHO versus binary grading systems, alongside key immunohistochemical biomarkers in OED. A systematic search of PubMed, Scopus, Web of Science, and Cochrane Library was conducted (January 2017–December 2025). Studies were selected according to PRISMA 2020 guidelines. Case reports, animal studies, systematic reviews, correspondence articles, and AI-focused papers were excluded. Sixteen primary studies met the inclusion criteria. Evidence on the comparative reliability and prognostic value of binary versus WHO grading systems was heterogeneous and appeared to correlate with study scale. Among six studies directly comparing both systems, four larger and more recent cohorts ( n = 137–878) found the WHO system to have comparable or superior inter-observer agreement and/or prediction of MT, while two smaller studies ( n = 48–148) favored the binary system. The largest study identified ( n = 878) reported significantly higher inter-observer reproducibility for the WHO system (κ = 0.47, P < 0.05) compared with the binary system (κ = 0.139, P = 0.39), while the best prognostic performance was achieved when both systems were combined (OR = 14.28). Podoplanin (PDPN) showed a promising association with malignant transformation across two independent cohorts (HR = 6.48–10.15, P < 0.05). In a single longitudinal cohort, reduced FANCD2 expression, particularly when combined with histologic grading, was associated with subsequent malignant transformation. The comparative performance of binary versus WHO grading systems for OED appears to depend on study scale, and larger recent cohorts do not confirm the reproducibility or prognostic advantage of the binary system reported in earlier, smaller studies. Rather than supporting replacement of the WHO classification, these findings favor a combined-systems approach, supplemented by emerging biomarkers such as PDPN, pending confirmation in large-scale prospective and meta-analytic studies. Integrating emerging prognostic biomarkers, particularly PDPN and FANCD2, may provide additional biological information for risk stratification; however, these biomarkers require independent prospective validation before clinical implementation.”