S359 Personalized Prognostic Modeling in Mucinous Adenocarcinoma (MAC) of the Colon With Liver Involvement

Introduction: Mucinous adenocarcinoma (MA) is a distinct colorectal cancer subtype with poor prognosis and limited response to therapy, especially when liver metastases are present. We applied machine learning (ML) to identify prognostic factors and improve outcome prediction in this population. Methods: Data were obtained from the SEER database (2004-2021). Patients who met any of the following criteria were excluded: diagnosis not confirmed by histology; previous history of cancer or other concurrent malignancies; or unknown data. To identify prognostic variables, we conducted Cox regression analysis and constructed prognostic models using ML algorithms to predict the 5-year survival. Patient records were randomly divided into training (70%) and validation (30%) sets. Model performance was evaluated using the area under the receiver operating characteristic curve. We also investigated the role of multiple therapeutic options using Kaplan-Meier survival analysis. Results: A total of 1,950 patients were included. Most patients (64.5%) were aged <71 years and 50.5% were women. The majority were White (77.1%), followed by Black (16.9%). The most common primary tumor sites were the cecum (36.1%) and ascending colon (20.7%). Most patients presented with an advanced T stage (T4, 47.4%) and nodal involvement (N2, 45.2%). Surgery was performed in 91.6% of the cases, chemotherapy in 67.6%, and radiation therapy in only 2.8%. Patients aged <71 years had higher 5-year survival rates (OS, 13.6%; CSS, 14.6%) than those aged ≥71 years (overall survival [OS], 3.4%; cancer specific survival [CSS], 4.1%). Radiation was associated with better outcomes (OS, 24.2%; CSS, 24.2%) than no radiation (OS: 9.4%, CSS: 10.4%). Among racial groups, Asian/Pacific Islanders had the highest survival (OS: 15.9%, CSS: 22.6%), followed by White (OS: 9.8%, CSS: 10.5%) and Black patients (OS: 8.4%, CSS: 8.6%). Chemotherapy was associated with improved survival (OS: 12.4%, CSS: 13.5%) compared to no chemotherapy (OS: 4.9%, CSS: 5.5%). The highest 5-year CSS was observed in the ascending (15.7%) and descending colon (13.6%), followed by the hepatic flexure (13.4%), splenic flexure (10.8%), transverse colon (11.9%), sigmoid colon (8.7%), and cecum (7.8%). Random forest was the most accurate models. The ML models identified the T stage as the most significant prognostic factor, followed by age. Conclusion: ML enhances personalized medicine by identifying key prognostic factors in MAC with liver metastasis, which can guide treatment decisions for better patient management.