(029) Risk factors for malignant transformation of oral epithelial dysplasia: a systematic review and meta-regression analysis

Background: Oral epithelial dysplasia is commonly known as a precancerous condition in oral cavity. While there were a large number of studies, including systematic reviews and meta-analyses, evaluating the malignant transformation rate (MTR) of dysplasia with a wide range of reported estimates, the heterogeneity across studies has never been adequately investigated.

Objectives: To determine factors associated with the heterogeneity in the reported MTRs of oral epithelial dysplasia in terms of study or population characteristics.

Methods: We searched PubMed, Embase, and Cochrane Library to identify studies reporting the MTR from patients with histologically confirmed oral epithelial dysplasia. Fifty-four studies with 9048 patients were included in the analysis. A meta-regression approach with the random-effects model on the MTR after logit transformation was used. Study design, region, start date of enrollment, follow-up duration, gender, age, grade of dysplasia, site of dysplasia, smoking status, and alcohol intake were examined as potential factors. Univariable as well as multiple meta-regression modelling was conducted and the significance level was not strictly applied in the exploratory nature.

Results: The higher MTR was found in the studies with the longer follow-up although it was not statistically significant (p=.326). The severity of dysplasia and the site of dysplasia in the tongue or floor of the mouth were the prominent factors for large MTR (p <.001, p=.021, respectively) and they became more significant after adjustment for follow-up duration. There was a tendency for the lower MTR to be observed in studies with more number of smokers (p=.146). However, the studies with the more smokers were the studies with the shorter follow-up (p=.345), the less proportion of dysplasia in the tongue or floor of the mouth (p=.004), in the younger (p=.063), and with more male (p <.001) patients. The p-value for smokers, therefore, became enlarged after adjustment for those covariates. A significantly higher MTR was observed in the studies in Europe compared to those in Asia (p=.036). However, the proportion of patients with severe dysplasia was significantly higher in those European studies, and the p-value for the difference became insignificant after adjustment for it (p=.375).

Conclusions: Our finding identified the site and severity of dysplasia as significant risk factors and follow-up duration as a major confounder. The final model including follow-up, site of dysplasia, and the severity accounted for 91.86% of the heterogeneity of MTRs reported in the existing literature.