RESUMEN
OBJECTIVES: Our main purpose and research question were to analyze and quantify whether there were significant differences in the time to develop cancer among patients with oral leukoplakia (OL), comparing the more susceptible cases to those with the least susceptibility to malignancy. MATERIALS AND METHODS: We followed 224 cases of OL after surgical or CO2 laser treatment for a mean time of 6.4 years. A Bayesian mixture cure model based on the Weibull distribution was used to model the relationship between our variables and cancer risk. In this model type, the population is considered a mixture of individuals who are susceptible or non-susceptible to developing cancer. The statistical model estimates the probability of cure (incidence model) and then infers the time to malignancy. The model was adjusted using the R-package INLA using default priors. RESULTS: Histology type (moderate or severe dysplasia) and tongue location showed hazard ratios (HR) of 3.19 (95% CI [1.05-8.59]) and 4.78 (95% CI [1.6-16.61]), respectively. Both variables increased the risk of malignant transformation, thus identifying a susceptible subpopulation with reduced time required to develop cancer, as with non-homogeneous leukoplakias. The median time for cancer development was 4 years and 5 months, with a minimum of 9 months after the diagnosis of OL and a maximum of 15 years and 2 months. CONCLUSIONS: Susceptible patients with non-homogeneous leukoplakia, dysplasia, or leukoplakia in the tongue develop cancer earlier than those with homogeneous OL and those without dysplasia. CLINICAL RELEVANCE: The novel contribution of this research is that, until now, the time it took for oral leukoplakias to develop cancer based on whether they were homogeneous or non-homogeneous, and if they have or not epithelial dysplasia, had not been comparatively described and quantified. As a final result, the time to malignant transformation in non-homogeneous and dysplastic leukoplakias is significantly shorter.