A prospective multicenter trial evaluating diagnostic validity of multivariate analysis and individual serum marker in differential diagnosis of pancreatic cancer from benign pancreatic diseases.

Abstract

Conclusion: A multivariate analysis of CAMPAS-PX2 can increase its diagnostic accuracy in differential diagnosis of pancreatic cancer from benign pancreatic or extrapancreatic disease, when compared with CA19-9 alone. However, the improvement in diagnostic accuracy is still not satisfactory in spite of an elaborate combination of serum markers in diagnosis for pancreatic cancer. Optimal combination of a sensitive serum marker and another diagnostic modality, such as ultrasonography, can be a practical way to improve important diagnostic and cost-effectiveness in diagnosis for pancreatic cancer.

Background: No specific biological test has yet been developed for diagnosis of pancreatic cancer, although increasing numbers of tumor markers become available. For improvement in the diagnostic and cost effectiveness, it is important to select optimal combination of several serum markers relatively independent of each other.

Methods: A new model of discriminant function, computer-aided multivariate and pattern analysis system for pancreatic cancer examination 2 (CAMPAS-PX2), was developed based on the data of the 23 serum tumor markers from the first prospective trial (1) to differentiate between pancreatic cancer and benign pancreatobiliary disease by logistic regression analysis using a stepwise selection method. In 243 patients suspected of having pancreatic pancreatic cancer by a multicenter prospective study, the diagnostic value of the multivariate analysis, CAMPAS-PX2, was compared with the 23 markers.

Results: Pancreatic cancer was subsequently identified in 27 patients. Positive in disease, negative in health, and area under receiver operating characteristic curve were significantly higher by CAMPAS-PX2 (89, 87, 91%) than by CA 19-9 (78, 82, 84%), the most sensitive marker among the 23 markers.