CA19-9 as an independent risk factor for survival in hepatocellular carcinoma.
253 Background: High alpha-fetoprotein (AFP) is associated with worse survival in hepatocellular carcinoma (HCC) patients. CA19-9 is a glycoprotein biomarker of biliary and pancreatic cancer. To our knowledge, no prospective study has examined CA19-9 as a prognostic biomarker in HCC. We hypothesized that it may add to our ability to risk stratify patients. Methods: We prospectively enrolled 122 patients with newly diagnosed HCC between 10/2008 and 7/2012 and followed until July 23, 2012. HCC was defined by AASLD criteria. CA 19-9 and AFP were measured on enrollment: 50% patients had CA19-9 >37 U/mL and 22% had CA 19-9>100. Pathology specimens were available for 50% of patients and none were consistent with mixed HCC-cholangiocarcinoma (HCC-CC) or pure CC. We evaluated the relationship between CA19-9 and overall survival using Kaplan Meier curves, univariate, and multivariate Cox Proportional Hazards Models. Results: The mean age of the cohort was 62.1 and 79% were male. 59% had underlying HCV and 16% had HBV. In univariate analysis, CA19-9>100 predicted worse survival (HR=3.44, 95% CI: 1.79-6.61, p=0.0002). Other cut-points for CA19-9 yielded similar results, although not all were significant. In a multivariate model including CP Score (B vs. A), Milan (outside vs. within), and AFP (>100 vs. ≤100), CA19-9 >100 remained an independent risk factor for increased mortality (HR 3.95, 95% CI: 1.95-7.97, p=0.0001). CP score and Milan status had HRs of 2.57 (95% CI: 1.31-5.02, p=0.006) and 5.27 (95% CI: 2.05-13.57 p=0.0006). AFP was not an independent risk factor for worse survival with a HR of 1.69 (95% CI: 0.85-3.37, p=0.14). We also looked at models using different cut-points for AFP; CA19-9 added additional risk stratification in all models. Among patients with AFP>100, median survival was 15.5 months vs. 3.8 months when comparing those with CA19-9 ≤100 with CA19-9>100 (p=0.01, log-rank). Conclusions: CA 19-9 is a readily available biomarker that may be a novel predictor of survival in HCC patients. In our multivariate model, CA 19-9>100 almost quadrupled mortality risk. It might be particularly useful to risk stratify patients with normal AFP and those with very high AFP being considered for transplant.