E264: Association of Blood Transfusion During Surgery for Colorectal Liver Metastasis with Outcomes: Propensity Score Analysis of Multicenter Database

Introduction: Hepatectomy is the treatment of choice for colorectal liver metastases (CLM). This procedure can be associated with significant blood loss requiring blood product transfusion. Previous studies suggested no effect of transfusion on overall survival (OS). We examined outcomes of CLM resection with and without transfusion using propensity score (PS) matched analysis.

Methods: We analyzed the multi-institutional Colorectal Liver Operative Metastasis International Collaborative (COLOMIC) database of patients with resectable CLM from 2000 to 2018. Unadjusted survival curves were built comparing patients who received at least one unit of blood product intraoperatively (Transfused) versus those who did not. Multivariate Cox proportional hazards models were built. Using PS matching for Transfused, matched cohorts were built and analyzed.

Results: 929 patients were available for analysis. 354 (38.1%) were transfused versus 575 (61.9%) not. The median OS in the transfused group was 39.8 months vs 52.8 (p = < 0.01). Transfused patients had more estimated blood loss (EBL), (459 vs 790 ml, p < 0.01), and had greater 90-day mortality, 5 (0.9%) vs 10 (2.8%), p = 0.02. The unmatched hazard ratio (HR) from a multivariate Cox model for Transfused was 1.36 (95% CI 1.02 - 1.80, p = 0.03) and 1.08 for each unit transfused (95% CI 1.00 – 1.17, p = 0.06). PS matching was performed for likelihood of transfusion in a multivariate logistic regression model. Significant variables included CEA (p = 0.02), BMI (p < 0.01), Functional dependence (p = 0.01), Diabetes (p = 0.045), Major hepatectomy (p = 0.032), Open approach (p < 0.01) and largest liver lesion size (p < 0.01). Matched data on Transfused were used to build a multivariate Cox model. Transfused remained significant with a HR of 1.46 (95% CI 1.08 - 1.98, p = 0.013). The HR for each unit transfused did not differ from the unmatched model. Other significant HRs from the matched model included BMI, 1.03 (95% CI 1.02 - 1.05, p < 0.01), Mutant Kras, 1.63 (95% CI 1.30 - 2.03, p < 0.01), liver lesion number, 1.11 (95% CI 1.05 - 1.17, p < 0.01) and liver lesion size, 1.00 (95% CI 1.00 to 1.01, p < 0.01). These models accounted for preoperative laboratory data, comorbidities, approaches and EBL.

Conclusions: Receiving at least one blood product operatively for CLM was significantly associated with decreased overall survival both before and after PS matching, on multivariate Cox models. Preoperative optimization of anemia and intraoperative blood conservation protocols should be initiated to decrease blood product transfusion.