Platelet-Albumin Score as a Sensitive Measure for Surgical Risk Prediction and Survival Outcomes of Patients with Hepatocellular Carcinoma.

BACKGROUND

Given that patients with hepatocellular carcinoma (HCC) usually suffer from dual diseases (i.e., HCC and underlying liver disease), a complete survival estimation model is difficult to establish because both the oncological stage and the underlying liver function affect the survival outcome.

METHODS

A new surgical risk model for estimating the survival of patients undergoing resection for HCC was created using a cohort treated between 1995 and 2013 (training set, n = 889), and its efficacy was then validated using a cohort treated between 2014 and 2017 (validation set, n = 310).

RESULTS

The following statistical model was developed based on the results of multivariate analysis: albumin-platelet (PAL) score = - 0.777 × albumin (g/dL) - 0.575 × log10 (platelet count [104/μL]) (cut-off value, - 3.77 and - 3.04 for grading). A time-dependent receiver-operating curve analysis revealed that the area under the curve for 3-year survival was 0.644 in the training set and 0.666 in the validation set. The incidences of postoperative morbidity were 14.0% for PAL grade 1, 18.7% for PAL grade 2, and 26.1% for PAL grade 3 (P = 0.039), while the incidences of refractory ascites were 2.2, 7.1, and 12.5% (P = 0.005), respectively, in the training set. The reproducibility of these results was confirmed in the validation set with morbidity rates of 13.5, 23.3, and 40.7% (P = 0.003), respectively, and the incidences of refractory ascites were 0.7, 10.7, and 22.2% (P < 0.0001), respectively.

CONCLUSIONS

The PAL score can be used as a grading system for the stratification of survival outcomes and surgical risks of patients undergoing HCC resection.