Risk prediction models for oral squamous cell carcinoma and oral potentially malignant disorders in Asian populations: a systematic review and meta-analysis.

BACKGROUND

Oral squamous cell carcinoma (OSCC) and oral potentially malignant disorders (OPMDs) represent a substantial public health burden across Asia, where tobacco use, betel quid chewing, and alcohol consumption are prevalent risk factors. Numerous multivariable risk prediction models have been developed to support early detection and targeted screening; however, their accuracy and applicability in Asian populations remain unclear. This study evaluated the diagnostic performance, methodological quality, and clinical readiness of OSCC/OPMD risk prediction models developed or validated in Asian settings.

METHODS

Following PRISMA and PRISMA-DTA guidelines, a comprehensive search of PubMed, Embase, Cochrane Library, Scopus, Google Scholar and grey literature was conducted for studies published from January 2000 to December 2025. Eligible studies included multivariable prediction models developed or validated among asymptomatic adults in Asian countries. Data extraction and PROBAST risk-of-bias assessments were completed independently by two reviewers. Hierarchical summary receiver operating characteristic (HSROC) modelling was used to pool sensitivity and specificity, with heterogeneity explored using meta-regression. Publication bias was assessed using Deeks' test. All analyses were performed in RStudio.

RESULTS

Eleven studies met the inclusion criteria, and seven contributed to the meta-analysis. Pooled sensitivity was 0.84 (95% CI: 0.78-0.89), and pooled specificity was 0.80 (95% CI: 0.67-0.89), demonstrating strong case-detection performance with acceptable false-positive rates. Substantial heterogeneity was observed (sensitivity I2 = 91.4%; specificity I2 = 99.2%). Meta-regression showed significantly higher sensitivity among externally validated models and those predicting combined OSCC/OPMD outcomes. No significant publication bias was detected.

CONCLUSION

Risk prediction models for OSCC and OPMDs in Asian populations demonstrate favourable diagnostic performance; however, substantial heterogeneity and methodological limitations warrant cautious interpretation before clinical implementation. Future research should prioritise large-scale external validations, improved calibration reporting, and harmonised risk thresholds to enhance regional applicability for risk-based oral cancer screening.

SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/search.