PURPOSE: To assess the impact of perioperative blood transfusion on overall and disease-free survival in patients undergoing curative resection for cholangiocarcinoma. METHODS: In a single-center study, 128 patients undergoing curative resection for cholangiocarcinoma between 2001 and 2010 were assessed. The median follow-up period was 19 months. Transfused and nontransfused patients were compared by Cox regression and propensity score analyses. RESULTS: Overall, 38 patients (29.7 %) received blood transfusions. The patient characteristics were highly biased with respect to receiving transfusions (propensity score 0.69 ± 0.22 vs. 0.11 ± 0.16, p < 0.001). In the unadjusted analysis, blood transfusion was associated with a 105 % increased risk of mortality [hazard ratio (HR) 2.05, 95 % CI 1.19-3.51, p = 0.010]. In the multivariate (HR 1.14, 95 % CI 0.52-2.48, p = 0.745) and the propensity score-adjusted Cox regression (HR 1.02, 95 % CI 0.39-2.62, p = 0.974), blood transfusion had no influence on overall survival. Similarly, in the propensity score-adjusted Cox regression (HR 0.62, 95 % CI 0.24-1.58, p = 0.295), no relevant effect of blood transfusion on disease-free survival was observed. CONCLUSIONS: To our knowledge, this is the first propensity score-based analysis providing compelling evidence that the worse oncological outcome after curative resection for advanced cholangiocarcinoma in patients receiving perioperative blood transfusions is caused by the clinical circumstances requiring the transfusions, not by the blood transfusions themselves.
PURPOSE: To assess the impact of perioperative blood transfusion on overall and disease-free survival in patients undergoing curative resection for cholangiocarcinoma. METHODS: In a single-center study, 128 patients undergoing curative resection for cholangiocarcinoma between 2001 and 2010 were assessed. The median follow-up period was 19 months. Transfused and nontransfused patients were compared by Cox regression and propensity score analyses. RESULTS: Overall, 38 patients (29.7 %) received blood transfusions. The patient characteristics were highly biased with respect to receiving transfusions (propensity score 0.69 ± 0.22 vs. 0.11 ± 0.16, p < 0.001). In the unadjusted analysis, blood transfusion was associated with a 105 % increased risk of mortality [hazard ratio (HR) 2.05, 95 % CI 1.19-3.51, p = 0.010]. In the multivariate (HR 1.14, 95 % CI 0.52-2.48, p = 0.745) and the propensity score-adjusted Cox regression (HR 1.02, 95 % CI 0.39-2.62, p = 0.974), blood transfusion had no influence on overall survival. Similarly, in the propensity score-adjusted Cox regression (HR 0.62, 95 % CI 0.24-1.58, p = 0.295), no relevant effect of blood transfusion on disease-free survival was observed. CONCLUSIONS: To our knowledge, this is the first propensity score-based analysis providing compelling evidence that the worse oncological outcome after curative resection for advanced cholangiocarcinoma in patients receiving perioperative blood transfusions is caused by the clinical circumstances requiring the transfusions, not by the blood transfusions themselves.
Authors: Annemiek M Dekker; Jimme K Wiggers; Robert J Coelen; Rowan F van Golen; Marc G H Besselink; Olivier R C Busch; Joanne Verheij; Markus W Hollmann; Thomas M van Gulik Journal: HPB (Oxford) Date: 2016-01-06 Impact factor: 3.647