A nomogram and random forest model for predicting liver metastasis in patients with early-onset colorectal cancer.

The incidence of colorectal cancer (CRC) in individuals under the age of 50 has increased. Liver metastasis (LM) is the most common metastasis in CRC patients and is associated with a poor prognosis. This study aimed to use public databases to identify the risk factors for LM in early-onset colorectal cancer (EOCRC) patients and develop a nomogram to quantify the risk of LM. We retrospectively collected data of EOCRC patients diagnosed from 2010 to 2015 in the Surveillance, Epidemiology, and End Results (SEER) database. Univariate and multivariate logistic analysis were used to screen and validate the risk factors for LM in EOCRC patients, and a nomogram was established based on these factors. Calibration curve, area under the receiver operating curve (AUC), and decision curve analysis (DCA) were developed to evaluate the accuracy of the model. A total of 2567 EOCRC patients were included and randomly divided into a training set (n = 1797) and a validation set (n = 770) at a ratio of 7:3. Univariate and multivariate analyses showed that N stage, pretreatment CEA, bone metastasis, and lung metastasis were independent risk factors. The AUCs of the training set and validation set were 0.7958 and 0.7653, respectively, and the calibration curve also demonstrated good accuracy and predictive ability. DCA indicated that it was more clinically relevant than the traditional TN staging. We constructed a Random Forest model, and calculated the SHapley Additive exPlanations (SHAP) values to determine variables importance and visualize the results. We developed a nomogram to predict the risk of LM in EOCRC patients, and the model was internally validated with good accuracy and reliability. It can assist doctors in risk assessment and clinical decision-making.