Objective This study reports further refinement of a prediction scoring system, which was established in 1980 as a guide to determine a safe limit for hepatectomy, based on 10 years of use. Summary Background Data In the past, whether major resection was safe was judged empiricially from the net resection volume or the residual hepatic volume combined with the patient's liver function. However, such judgment was not based on objectively defined criteria. Methods Patients with hepatocellular carcinoma (HCC; n=376) and metastatic cancer (n=58) who had hepatectomy at some time from 1981 through 1990 were entered into this study. A prediction score (PS) was computed using a multiple regression equation that consists of computed tomographic scan-estimated resection rate, indocyanine green retention rate, and the patient's age. A PS greater than 55 was classified as a risky zone, a PS of 45 to 55 was considered borderline. and a PS less than 45 was a safe zone. Results With HCC and chronic liver disease, all patients in the risky zone died, whereas 33% in the borderline zone died and 7.3% died who were in the safe zone. With metastatic cancer with normal liver, all patients in the risky zone died, whereas no patient in either the borderline or safe zones died. The major cause of death in the risky zone was liver failure due to excessive resection. In the borderline and safe zones, liver failure developed primarily after abdominal sepsis or pulmonary infection, particularly for those with adverse prognostic factors such as disturbed glucose tolerance, lower platelet count, and higher indocyanine green retention rate. Conclusion Prediction scores can eliminate deaths related to excessive resection for patients with normal or injured livers. When patients have adverse prognostic factors, careful surgery and postoperative management is mandatory to avoid liver failure triggered by intra- or extra-abdominal sepsis, even if the score remains in a borderline or safe zone.
