BACKGROUND

CA19-9 is a carbohydrate tumor-associated antigen which is frequently upregulated in pancreatobiliary neoplasia. However, it may also be elevated in patients with jaundice in the absence of a tumor due to biliary obstruction, and in other non-hepato-pancreatico-biliary conditions. This study aimed to evaluate whether CA19-9 levels could accurately differentiate between benign and malignant pancreatobiliary disease.

METHODS

All patients referred to a single surgeon for investigation of pancreaticobiliary disease in 2003 in whom a firm diagnosis had been established were included. For malignant disease, a histological diagnosis was required but for benign disease a firm radiological diagnosis was deemed adequate. The patients were divided into 4 categories: pancreatic adenocarcinoma (PCa); cholangiocarcinoma (CCa); chronic pancreatitis (CP) and biliary calculous disease (Calc). Bilirubin and alkaline phosphatase levels corresponding to the point of assessment of CA19-9 were also noted.

RESULTS

Final diagnoses were made of pancreatic adenocarcinoma (PCa, n=73), cholangiocarcinoma (CCa, n=19), ampullary carcinoma (Amp, n=7), neuroendocrine carcinoma (Neu, n=4), duodenal carcinoma (Duo, n=3), chronic pancreatitis (CP, n=115), and biliary calculous disease (Calc, n=27). Median CA19-9 levels (U/ml) were: PCa, 653; CCa, 408; Duo, 403; Calc, 27; CP, 19; Neu, 10.5; Amp, 8 (reference range: 0-37). The CA19-9 levels were significantly greater for malignant than for benign disease, could differentiate PCa from CCa/Duo, and were significantly higher in unresectable than in resectable PCa. The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for CA19-9 were 84.9%, 69.7%, 67.7% and 86.1%, respectively. A ROC analysis provided an area under the curve for CA19-9 of 0.871 (0.820-0.922), giving an optimal CA19-9 of 70.5 U/ml for differentiating benign from malignant pathology. Using this cut-off, the sensitivity was 82.1%, while specificity, PPV and NPV improved to 85.9%, 81.3% and 86.5%, respectively. When standard radiology was included (US/CT/MRCP) in the decision process, the results improved to 97.2%, 88.7%, 86.6%, and 97.7%. For benign disease, the CA19-9 correlated directly with the serum bilirubin, but for malignant disease, CA19-9 levels were elevated independent of the bilirubin level.

CONCLUSIONS

CA19-9 is useful in the differentiation of pancreatobiliary disease and when using an optimized cut-off and combining with routine radiology, the diagnostic yield is improved significantly, thus stressing the importance of a multi-disciplinary approach to pancreatobiliary disease.