AIM: To determine the predictive value of biochemistry and imaging in the diagnosis of malignant biliary strictures. METHODS: 31 patients with suspected non-calculous biliary obstruction were consecutively enrolled. A biochemical profile, ultrasound (U/S), HIDA scan and Cholangiography (ERCP or PTC) were obtained at study entry. Stricture etiology was determined based on clinical outcome by one year and/or cytology and histology. Statistical analysis utilized t tests and Mann-Whitney for non-distributive data. RESULTS: 29/31 patients had biliary strictures, 15/29 had malignant strictures. The mean age of the malignant cohort was 73.9 and the benign cohort 53.9 (P=0.0005). Statistically significant differences between the malignant vs benign groups included: ALT 235.2 vs 66.9 IU/L (P=0.002), AST 189.8 vs 84.6 IU/L (P=0.009), ALK PHOS 840.2 vs 361.1 IU/L (P=0.001), BILIRUBIN 317.8 vs 22.1 mmol/L (P<0.001) and BILE ACIDS 242.5 vs 73.2 mmol/L (P=0.001). Threshold analysis using ROC curves demonstrated that a bilirubin level of 75 mmol/L was most predictive of malignant strictures (log odds ratio 16.24). Likelihood ratios were not significantly greater than chance for the other biochemical parameters. Intrahepatic duct dilatation or a pancreatic/intraductal mass on U/S was more prevalent in patients with malignant strictures (14/15 vs 5/14 for dilatation; P=.002 Fisher's exact test). Of 12 patients with malignancy undergoing HIDA scan, 9 showed functional obstruction while 3 showed hepatocellular dysfunction. Among patients with benign strictures, 5 had functional obstruction, 4 had hepatocellular dysfunction and 5 HIDA scans were normal. Cholangiographic characterization of strictures revealed an equal distribution of smooth (8/15) and irregular (5/15) strictures in the malignant group; 10/14 benign strictures were smooth. There was a significant difference in stricture length between malignant vs benign cohorts, 30.30 mm vs 9.20 mm (P=0.0039). Threshold analysis using ROC curves revealed a stricture length of 25 mm was most consistent with malignancy (sensitivity 79%, specificity 75%, log odds ratio 11.23). A multivariate model could not be constructed because in our data set, serum bilirubin perfectly discriminated between benign and malignant strictures. CONCLUSIONS: Serum bilirubin and cholangiographic stricture length are most predictive of malignancy. Ultrasound was useful in predicting malignant strictures in the presence of intrahepatic duct dilatation and pancreatic/intraductal mass. Strictures with a "benign" cholangiographic appearance (smooth or regular) are frequently malignant.
|Original language||English (US)|
|State||Published - 1997|
ASJC Scopus subject areas