General Surgery Residency University of Pittsburgh Pittsburgh, Pennsylvania, United States
Disclosure: Disclosure information not submitted.
Participants should be aware of the following financial/non-financial relationships:
Tanner Wilson: Disclosure information not submitted.
Introduction: Elevated levels of carcinoembryonic antigen (CEA) are a negative prognostic factor in colon cancer. Current recommendations do not consider elevated CEA levels when determining the need for adjuvant chemotherapy in stage II patients. Instead, adjuvant chemotherapy in stage II patients is reserved for those with high-risk features such as positive margins, lymphovascular invasion (LVI), and high tumor grade. We hypothesized that CEA is also a high-risk feature for which adjuvant chemotherapy is beneficial.
Methods: This retrospective study examined patients from the National Cancer Database (NCDB) diagnosed with colon cancer between 2004-2018. We included patients with stage II disease with high-risk features. Preoperative CEA was defined as either normal or elevated as recorded in the NCDB. Overall survival (OS) was estimated using the Cox proportional hazards model, with LASSO regularization used for feature selection. Standardized model coefficients were used to compare importance of predictors and are reported as hazard ratios (HR). Computations were performed in Python using scikit-learn and scikit-survival with a 20% held out test set.
Results: A total of 8,107 patients (mean age 70.6, 42.3% male, 87.6% white) met inclusion criteria. Of these, 25.8% received adjuvant chemotherapy. CEA was elevated in 38.6% of patients. Among patients who did not receive adjuvant chemotherapy, OS was worse when the CEA level was elevated in addition to other high-risk factors (Figure 1). Adjuvant chemotherapy improved OS (p < 0.00001). CEA’s prognostic value for OS (HR=1.13) was nearly as high as that of positive margins (HR=1.15) and much higher than that of LVI (HR=1.04). Notably, LASSO did not select high tumor grade as a predictive feature (HR=1) when CEA was included; our overall model yielded a test set concordance index of C=0.733.
Conclusions: Our findings demonstrate that in stage II colon cancer patients, an elevated CEA level is an independent high-risk feature that has a survival predictive capacity rivaling that of accepted high-risk features and should be considered in recommendations regarding adjuvant chemotherapy in stage II patients.
Learning Objectives:
Describe indications for adjuvant chemotherapy in stage II colon cancer
Identify high-risk features of stage II colon cancer associated with worse prognosis
