Validation of a Nomogram to Predict the Risk of Perioperative Blood Transfusion for Liver Resection.

BACKGROUND

Nomograms may be important clinical tools to estimate the preoperative risk of transfusion and allow for preemptive arrangements for alternatives to allogeneic blood transfusions.

METHODS

A multicentric international cohort of 1345 patients who underwent hepatectomy for benign or malign liver diseases was used to validate a nomogram developed by the Memorial Sloan-Kettering Cancer Center.

RESULTS

A total of 449 (33.3 %) patients received a blood transfusion after hepatectomy. Several variables were associated with the need of transfusion on univariate analysis: age, BMI, hemoglobin, PT-INR, bilirubin, AST, ALT, GGT, albumin, primary liver cancer, and number of segments resected. The MSKCC nomogram, including the number of segments resected, diagnosis (primary vs. non-primary), extrahepatic organ resection, as well as platelet and hemoglobin levels, had a good predictive ability (AUC = 0.69). The frequency of patients transfused ranged from 19 % for patients who were at "low risk" (<20 % risk to be transfused) up to 68 % for patients at "high risk" (>70 % risk to be transfused). The nomogram was tested in a multivariable model including other factors associated with risk of transfusion. The final model included age (OR 1.02, 95 % CI 1.01-1.03, p < 0.001), PT-INR (OR 1.54, 95 % CI 1.01-2.36, p = 0.048), and bilirubin (OR 1.86, 95 % CI 1.09-3.18, p = 0.021). The prediction ability for the integrated prediction model was AUC = 0.73.

CONCLUSION

The MSKCC nomogram was an effective clinical tool able to predict the perioperative risk of transfusion in our independent external validation. The inclusion of patient age, as well as factors associated with liver functional status (bilirubin and PT-INR), improved the predictive ability of the MSKCC nomogram.