Early Postoperative Biomarkers Associated with Subsequent Operative Source Control in Anastomotic Leakage After Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy: A Multicenter Retrospective Cohort Study

Background/Objectives: Anastomotic leakage is a serious complication after cytoreductive surgery with hyperthermic intraperitoneal chemotherapy (CRS-HIPEC). Among patients who develop leakage, early identification of those requiring operative source control remains difficult. This study evaluated the discriminatory performance of early postoperative inflammatory and metabolic biomarkers for subsequent operative source control. Methods: This multicenter retrospective cohort study screened 540 patients who underwent CRS-HIPEC between January 2015 and January 2025. Anastomotic leakage was identified in 51 patients. Four patients underwent reoperation before completion of the postoperative day (POD) 3 laboratory assessment: two for anastomotic leakage and two for postoperative bleeding. Three additional patients were excluded because the clinical or laboratory data required for the primary analyses were unavailable. The final cohort comprised 44 patients, of whom 24 (54.5%) subsequently underwent operative source control for anastomotic leakage and 20 (45.5%) were managed without reoperation through one month. C-reactive protein (CRP), albumin, neutrophil-to-lymphocyte ratio (NLR), white blood cell count, glucose, CRP-to-albumin ratio (CAR), and leukoglycemic index (LGI) were assessed on POD1 and POD3. All POD1 and POD3 laboratory measurements were obtained before operative source control in patients who subsequently underwent reoperation. Discriminatory performance was evaluated using receiver operating characteristic analysis. Post hoc, data-informed Firth penalized logistic regression models were used to explore the associations of POD1 CRP and POD3 NLR with operative source control after adjustment for BMI and HIPEC agent. Results: NLR was significantly higher in patients who subsequently required operative source control on both POD1 (median, 14.9 vs. 9.8; p < 0.001) and POD3 (21.0 vs. 9.8; p < 0.001). The increase in NLR from POD1 to POD3 was also greater in the operative group (median change, 4.8 vs. 1.0; p = 0.009). POD3 NLR showed the highest observed discriminatory performance (area under the curve [AUC], 0.901; 95% confidence interval [CI], 0.810–0.992), while POD1 NLR (AUC, 0.834; 95% CI, 0.717–0.951) and POD1 CRP (AUC, 0.798; 95% CI, 0.649–0.947) also showed good observed discrimination. POD3 CRP-to-albumin ratio showed moderate discrimination, whereas POD3 LGI showed more limited performance. In post hoc exploratory Firth models adjusted for BMI and HIPEC agent, POD1 CRP per 10 mg/L increase (adjusted odds ratio [aOR], 1.52; 95% CI, 1.10–2.56; p = 0.006) and POD3 NLR per 5-unit increase (aOR, 3.91; 95% CI, 1.57–20.23; p = 0.001) remained associated with operative source control. All ROC-derived cutoff values and their operating characteristics were data-derived, intended solely for hypothesis generation, and should not be used for clinical decision-making. Conclusions: Within this selected retrospective cohort, POD3 NLR showed the highest observed discriminatory performance and remained associated with subsequent operative source control in a post hoc exploratory adjusted model. POD1 NLR and POD1 CRP also provided exploratory discriminatory information, while the greater increase in NLR from POD1 to POD3 suggests that serial biomarker trajectories warrant further investigation. These findings, including the data-derived cutoff estimates, are intended solely for hypothesis generation and should not be used for clinical decision-making. Prospective external validation is required.