Proposal for a tumor budding predictive score derived from endoscopic biopsy samples in colorectal cancer

Background In colorectal cancer, tumor budding is highlighted as both a prognostic indicator and a predictor of chemosensitivity. However, tumor budding has a serious drawback because of unattainable preoperative assessment, thereby, making it not applicable to decision-making on treatment strategies. Recently, high expressions of seven genes (i.e., MSLN , SLC4A11 , WNT11 , SCEL , RUNX2 , MGAT3 , and FOXC1 ) were shown to be associated with high-budding colorectal cancers. This study aimed to propose a budding prediction system using selected RNAs extracted from biopsy specimens. Methods The RNA expression levels in 86 surgically resected samples and in 104 samples obtained by colonoscopy before surgery were investigated. Results The tumor surface expressions of four exclusive genes (i.e., MSLN , SLC4A11 , SCEL , and MGAT3 ) were correlated with the tumor budding grade. Subsequently, the logit P value calculated by multiple logistic regression analysis using the four surface gene expressions was set as the following budding predictive score: Logit ( P ) = − 0.55 + 0.27* MSLN + 0.16* SLC4A11 + 0.06* MGAT3 + 0.21* SCEL. The effectiveness of the model using colorectal cancer biopsy samples was well corroborated prospectively. Conclusion The budding predictive score that we developed using endoscopic biopsy specimens was clarified to have a high potential for preoperative use.