Poor utility of current nomograms assessing the risk of intraoperative blood transfusion in patients undergoing liver resection for hepatocellular carcinoma and proposal of a new model

Background: The Memorial Sloan Kettering Cancer Center nomogram, the predictive scoring system of Yamamoto et al, and the 3-point transfusion risk score of Lemke et al are models used to determine the probability of receiving intraoperative blood transfusion in patients undergoing liver resection. However, the external validity of these models remains unknown. The objective of this study was to evaluate their predictive performance in an external cohort of patients with hepatocellular carcinoma. We also aimed to identify predictors of blood transfusion and develop a new predictive model for blood transfusion.

Methods: Post hoc analysis of our prospective database of 1,081 patients undergoing liver resection for hepatocellular carcinoma from 2001 to 2018. The predictive performance of current prediction models was evaluated using C statistics. Demographic and clinical variables as predictors of blood transfusion were assessed. Using logistic regression, an alternative model was created.

Results: The Lemke transfusion risk score performed better than the Memorial Sloan Kettering Cancer Center nomogram (0.69, 95% confidence interval 0.66-0.73 vs 0.66, 95% liver resection 0.62-0.69) (P < .001). The model from Yamamoto et al performed comparably with no statistically significant differences found through pairwise comparison. In our alternative model, hemoglobin level, albumin level, liver resection type, and tumor size were independent predictors of blood transfusion. The new HATS model obtained a C statistic of 0.74 (95% confidence interval 0.71-0.78), performing significantly better than the previous 3 models (P ≤ 0.001 for all).

Conclusion: The existing Memorial Sloan Kettering Cancer Center, Yamamoto et al, and Lemke et al had nomograms with the suboptimal accuracy of predicting risk of intraoperative blood transfusion in patients undergoing liver resection for hepatocellular carcinoma. The proposed HATS model was more accurate at predicting patients at risk of blood transfusion.