Predictive modeling for the germline pathogenic variant of the APC gene in patients with adenomatous polyposis: proposing a new APC score.

Abstract

The precise diagnosis and medical management of patients with suspected familial adenomatous polyposis should be based on genetic testing, which may not always be available. Therefore, establishing a new model for predicting the likelihood of a germline pathogenic variant (GPV) of APC based on its clinical manifestations could prove to be useful in clinical practice. The presence of GPVs of APC gene was investigated in 162 patients with adenomatous polyposis (≥ 10 polyps) using a multigene panel or single-gene testing. To generate a predictive model for GPV of the APC gene, a logistic regression analysis was performed using the clinicopathological variables available at the time of the diagnosis of adenomatous polyposis. Ninety (55.6%) patients had GPV of the APC gene. According to a multivariate logistic regression analysis, age < 40years, polyps ≥ 100, fundic gland polyposis, and a family history of colorectal polyposis were found to be independent predictors of the GPV of APC and were used to establish a formula for predicting the GPV of APC using the four predictors. The prediction model had an area under the curve of 0.91 (0.86-0.96) according to a receiver operating characteristic analysis. The model for predicting the GPV of APC will help patients with adenomatous polyposis and physicians make decisions about genetic testing.