S2511 Identifying Prognostic Factors in Appendiceal (AC) and Cecal Adenocarcinoma (CA) With Liver Metastasis Through Machine Learning Analysis

Introduction: AC and CA with liver metastases are rare gastrointestinal cancers that remain poorly studied. This study aimed to apply machine learning (ML) to identify prognostic factors and improve survival prediction in patients with liver metastases from AC or CA. 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. A validation method incorporating the area under the curve (AUC) of the receiver operating characteristic curve was used to validate the accuracy and reliability of the ML models. We also investigated the role of therapeutic options using Kaplan-Meier survival analysis. Results: A total of 6,596 patients were included, of whom 191 (2.9%) had AC and 6,405 (97.1%) had CA. Most patients were younger than 69 years (60.1%), 51.2% were woman, and 76.6% were White. Compared with CA, AC cases were more likely to present with advanced T4 disease (71.2% vs 43.7%) and had higher rates of N2 nodal involvement (26.7% vs 51.6%). The most common treatment modality was surgery combined with chemotherapy (59.9%) followed by surgery alone (32.6%). At 5 years, AC showed better survival than CA. The 5-year OS and CSS rates for AC were both 15.5%, whereas CA had a slightly lower survival rate (OS: 12.3%, CSS: 13.9%). In AC, surgery alone was associated with excellent survival (OS, 89.4%; CSS, 93.3%), whereas surgery with chemotherapy yielded lower outcomes (OS, 49.2%; CSS, 51.1%). For CA, chemotherapy alone had the lowest 5-year survival rate (OS, 2.4%; CSS, 2.9%), followed by surgery alone (OS, 9.2%; CSS, 11.4%). The highest survival rate was observed with surgery and chemotherapy combined (OS: 15.0%, CSS: 16.5%), whereas the addition of chemoradiotherapy showed limited benefit (OS and CSS: 5.7%). Gradient Boosting was the most accurate ML models. ML algorithms identified T stage as the most significant prognostic factor, followed by age. Conclusion: ML plays a crucial role in personalized medicine by identifying key prognostic factors in AC and CA with liver metastasis. Further research is needed to assess the benefits of adjuvant therapy for managing these rare malignancies.