Graph models for phenotype and genotype association between oral mucosa and submandibular gland tumorigenesis in rat

Abstract

In recent years, success of statistics in field of genetics has been the identification of genes that affect the process of disease. Experimental models using animals enable early stages of tumor development to be studied. The aim of this study was to apply graph models to assess the association between the observed phenotypic changes in rat oral mucosa and induced tumorigenesis in the submandibular gland (SMG). We studied changes in oncogenes TP53 and bcl-2, histopathological and immunomarker variables in samples of oral mucosa and SMG of Wistar male rats, 60 days old and 180 g in weight, in which tumorigenesis was induced in their SMG by a 0.5% solution of 9,10-dimethyl-1,2-benzanthracene in acetone. A set of linear structural equations were defined, with each formula indicating the response variables and the direct influences. In graph models, saliva was considered as a latent variable. The association was analyzed using Graphical Gaussian Markov models and odd ratios. About 40% of animals treated with 9, 10-dimethyl-1, 2-benzanthracene showed histological alterations in the epithelial basal strata of their oral mucosa only at 150 days. Statistical models indicated a relationship between gene alteration in gene bcl-2 in the SMG and histological changes observed in the oral mucosa (P = 0.04). Graph statistical model with one latent variable allows to conclude that these results associated with other clinical parameters may be useful in detecting early changes in SMG tumorigenesis. Furthermore, the design of randomized sampling of oral mucosa allows to validate these results and establish a reliable methodology for presumptive diagnosis or screening in the future.