Model for End-stage Liver Disease score fails to predict perioperative outcome after hepatic resection for hepatocellular carcinoma in patients without cirrhosis

Background: The Model for End-stage Liver Disease (MELD) score was developed to reflect the hepatocellular reserve in patients with cirrhosis. We hypothesized that the MELD score would not be predictive of perioperative outcome after hepatic resection in patients without cirrhosis. Methods: We performed a case-control study of all consecutive patients from 1995 through 2005 undergoing hepatic resection for HCC. Results: Group A (21 patients without cirrhosis) had a mean age of 57 years, which was similar to control group B (25 patients with cirrhosis), with a mean age of 60 years. The mean tumor size in group A was 9.8 cm compared with that of group B, which was 4.8 cm (P = .03). The American Joint Committee on Cancer stage in group A was I in 14%, II in 5%, and III in 81% versus I in 48%, II in 16%, and 111 in 36% in group B (P = .002). Eighty-six percent of group A patients had a major hepatic resection (>2 segments) compared with 40% in group B (P = .001). The perioperative morbidity and mortality were 24% and 4.8%, respectively, in group A compared with 64% (P = .006) and 20% (P = .12) in group B. The mean preoperative, postoperative, and delta MELD scores were 7.0, 13.0, and 5.0, respectively, in group A compared with 9.6, 16.8, and 7.2 in group B (P = NS). In group A, none of the MELD score parameters accurately predicted perioperative outcomes despite a higher number of patients who had major hepatic resection. In group B, a preoperative MELD score of 9 or greater was associated with a higher overall perioperative morbidity (84% vs 41%, P = .03). Perioperative mortality (n = 6; 13%) was significantly higher in patients with a postoperative MELD score of 15 or higher (P = .02) and a delta MELD score of 10 or higher (P = .03). Conclusions: Perioperative MELD score fails to predict perioperative outcomes after hepatic resection for hepatocellular carcinoma in patients without cirrhosis. Other predictive parameters need to be developed for this group of patients.