Conditional survival of patients with hepatocellular carcinoma undergoing curative intent resection: Analysis of the national cancer database

Abstract

Background: Hepatocellular carcinoma (HCC) is a lethal malignancy rising in both incidence and mortality worldwide. Conditional survival (CS) estimates represent the probability that a patient will survive additional number of years, because the patient has already survived a given amount of time, and better define the survival potential over time. We analyzed the CS estimates of resected patients from the time of surgical resection. Methods: A retrospective analysis of patients who had curative intent surgical resection for HCC was performed using the National Cancer Data Base (NCDB) from 2004-2014. Crude survival curves were generated using the Kaplan-Meier method. For the purposes of the CS analysis, we assessed the probability of surviving an additional 5 years (CS5), given that the patient had already survived 2 years. As such, the 5-year CS for patients who had survived (S) “X” years was computed as CS5 S(x+5)/S(x). CS estimates were also stratified by clinicopathologic variables of interest. Results: The NCDB captured 11,357 patients who underwent resection for HCC. The characteristics of the patients and tumors include: male gender (74.1%), white race (78.7%), Charlson Deyo score of 0-1 (72.5%), intermediate tumor grade (53.1%), margin-negative resection (88.5%), 30-day mortality (3.2%) and 90-day mortality (5.4%). Table 1 shows that, for example, for the OS, the unconditional 5-year survival probability is 65.9%, but given that those have survived 2 years, the additional 5-year survival probability is 84.7% (in red). Also, the unconditional 4-year survival probability is 71.0%, given that the person survived 2 or 3 years, the 4-year survival probabilities are 85.1 and 78.6% respectively (in blue). Similarly, the 5-year CS for patients who have survived 2 years is as follows for: age < 65 (92.9%), intermediate grade (85.2%), margin-negative resection (84.8%), no radiation therapy (84.6%), and no systemic therapy (81.2%). The overall survival was highly statistically significant between different groups within each of these variables (p < 0.0004). Conclusion: CS estimates following resection in HCC patients reflect an improved survival for a given length of time survived compared to the respective unconditional survival. However, the CS declines for each successive year survived reflecting other variables that (perhaps not oncologic) that impact survival in HCC. CS estimates provides more accurate prognostic information for the individual resected HCC patients, and can be used by clinical clinicians to guide further therapy and follow up years after resection.