Improved sensitivity of fuzzy logic based tumor marker profiles for diagnosis of pancreatic carcinoma versus benign pancreatic disease.

BACKGROUND: The definite diagnosis of pancreatic tumors of unknown origin remains a clinical challenge. Imaging techniques may fail to differentiate malignant tumors from inflammation, especially in chronic pancreatitis. In a considerable number of cases, the definitive diagnosis needs laparotomy. Single tumor markers as CA 19-9 or CEA are of limited value in these cases because of their limited sensitivity and specificity. This study was performed to find out, whether a classification method based on fuzzy logic analysis of tumor marker profiles is feasible in patients with pancreatic carcinoma and benign pancreatic disease. PATIENTS AND METHODS: Tumor markers and other clinical and laboratory parameters of 74 consecutive patients, either with histologically proved pancreatic carcinoma (n = 43) or presumed benign pancreatic disease (n = 31) assessed by ultrasound, endoscopic retrograde pancreatography, or computed tomography were analysed in order to detect their diagnostic value in the multi-dimensional approach of fuzzy logic analysis.
RESULTS: Given a specificity of 95%, sensitivity was considerably increased by the fuzzy logic method (83%) compared with the best single tumor marker CA 19-9 (65%). CA 19-9, CYFRA 21-1 and CA 15-3 gave the most considerable contributions to the classification system, whereas CEA, CA 125, CA 72-4, AFP, liver enzymes, lipase, amylase, bilirubin, and ferritin were of no or little value.
CONCLUSIONS: Tumor marker profiles analysed with the fuzzy logic method may have an improved sensitivity compared with single standard tumor markers. The diagnostic efficacy should be proved in patients with pancreatic tumors of unknown origin.