Objective To evaluate the prognostic value of tumor burden score (TBS) combined with serum tumor markers in patients with resectable colorectal liver metastasis (CRLM) after curative hepatectomy.

Methods We retrospectively retrieved and analyzed clinicalpathological data of patients with CRLM who underwent curative hepatectomy between Jan 2012 to Dec 2023 in The First Affiliated Hospital of Chongqing Medical University. Serological tumor markers and TBS were utilized to generate prediction model, where the serological index was derived from the sum of carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) levels, and TBS was calculated from preoperative radiological images. The cutoff values were determined as the sum of CEA and CA19-9 levels and preoperative TBS. The patients were then categorized based on prediction parameters into low-risk (lacking any high-level index), medium-risk (with one high-level index) and high-risk (with two high-level indices) groups. Overall survival (OS) and disease-free survival (DFS) were compared using Kaplan-Meier curves, and multivariate Cox regression analysis was performed to determine possible prognostic risk factors. The receiver operating characteristic curve (ROC) and C-index were employed to assess the discriminatory power of the prediction model.

Results Median follow-up time was 43(4–112) months. Kaplan–Meier survival analysis suggested that the median OS time and 5-year OS of low-, median-, and high-risk groups were 64, 54, and 30 months and 55%, 38%, and 23% respectively (P=0.0019). The median DFS time and 3-year DFS were 36, 22, and 11 months, and 47%, 35%, and 16% respectively (P<0.001). Multivariate Cox regression analysis revealed that independent prognostic risk factors for postoperative OS were the sum of CEA and CA19-9 level (hazard rstio HR=1.81, 95% confidence interval CI: 1.09-3.01, P=0.021), TBS (HR=1.42, 95%CI:1.02-2.84, P=0.032), bilobar metastasis (HR=1.99, 95%CI:1.25–3.16, P=0.004), and primary tumor nodal invasion (HR=1.51, 95%CI:1.37–3.34, P=0.028). Multivariate Logistic regression analysis showed that the sum of CEA and CA19-9 levels (odds ratio OR=2.44, 95%CI 1.26-4.71, P=0.008) and TBS (OR=2.95, 95%CI:1.50-5.78, P=0.002) associated significantly with postoperative recurrence. The ROC curve and C-index revealed that serum tumor markers combined with TBS (area under the curve=0.630, 95%CI: 0.552–0.707, P<0.05; C-index=0.627) were more effective in predicting the prognosis of patients with CRLM than either CEA or TBS alone.

Conclusions The prediction model based on serum tumor markers combined with TBS demonstrated better efficacy than any single serological or radiological index and may provide new insights aiding individualized and accurate clinical decision-making for patients with CRLM.