Nomogram for individualised prediction of liver failure risk after hepatectomy in patients with resectable hepatocellular carcinoma: the evidence from ultrasound data.

OBJECTIVES

This study sought to develop a clinical nomogram for predicting post-hepatectomy liver failure (PHLF) among patients with resectable hepatocellular carcinoma (HCC).

METHODS

The nomogram was established based on data obtained from a prospective study on 136 consecutive patients with resectable HCC undergoing hepatectomy from January 2015 to December 2015 in our centre. Another 80 patients in our centre served as an independent internal validation set. The predictive accuracy and discriminative ability of the nomogram were determined by concordance index (C-index), calibration curve and compared with commonly predictive systems.

RESULTS

PHLF occurred in 30.9% of patients in the derivation set, including 36 and six patients with Grades A and B, respectively. The statistical nomogram built on the basis of platelet count, serum bilirubin, serum GGT, clinical signs of portal hypertension and shear wave elastography had good calibration and discriminatory abilities, with C-indices of 0.85. These models showed satisfactory goodness-of-fit and discrimination abilities in the independent validation set with C-indices of 0.824 for PHLF. The areas under the receiver-operator characteristic (ROC) curve using our methods were greater than those of conventional predictive systems in the validation patients (corresponding C-indices, 0.572-0.701).

CONCLUSIONS

This novel nomogram provides good preoperative prediction of PHLF in patients with resectable HCC.

KEY POINTS

• The nomogram was built by platelet count, bilirubin, GGT, CSPH and SWE. • The nomogram showed good calibration and discriminatory abilities in the different sets. • Compared with other models, the nomogram indicated better discriminatory capability.