Background. There has been extensive use of serum tumor markers in diagnosing pancreatic adenocarcinoma. There is no tumor marker, however, that alone has sufficient diagnostic accuracy. It is necessary to know which combination of tumor markers should be used to detect pancreatic cancer, with respect to clinical usefulness and cost effectiveness.
Methods. Serum levels of 17 kinds of tumor markers were determined in 145 patients and 40 healthy volunteers. Thirty‐five patients with proven pancreatic adenocarcinoma, and 32 with benign pancreatobiliary disease (14 chronic pancreatitis and 18 biliary stones) were selected. For analysis of the usefulness of each tumor marker to differentiate these two groups, scatterplot and relative operating characteristic (ROC) analyses were used. A multivariate discriminant system to differentiate these two groups was developed using stepwise discriminant analysis by backward elimination selection.
Results. The significance of each tumor marker varied according to the tumor volume. By ROC analysis, the markers were divided into four subgroups according to their usefulness in discriminating pancreatic adenocarcinoma from benign pancreatobiliary disease. A discriminant system consisting of two different discriminant functions using nine tumor markers (CA 19–9, DUPAN‐2, TPA, elastase‐1, lipase, amylase, γ‐glutamyl transpeptidase, alkaline phosphatase, and lactate dehydrogenase) was developed and designated CAMPAS‐P; it could differentiate between all 35 cases of pancreatic adenocarcinoma and 32 cases of benign pancreatobiliary disease. On the other hand, CAMPAS‐P showed a low positive rate in pancreatic tumors of unusual histologic type, and poor organ‐specific diagnostic ability in various digestive organ malignancies.
Conclusions. CAMPAS‐P may be very useful for differential diagnosis between pancreatic adenocarcinoma and benign pancreatobiliary disease.