Background/aims: Staging laparoscopy for suspected pancreatic neoplasia is not widely accepted due to its low yield. The aim of this study was to determine if serum carbohydrate antigen (CA19-9) levels could be used to improve the selection of patients for staging laparoscopy.
Methods: The data from a prospectively collected database (1997-2004) with 159 patients who had computed tomography-predicted resectable disease and who had undergone laparoscopic staging were analysed to determine if a low preoperative CA19-9 level (< or =150 kU/l, or < or =300 kU/l with a bilirubin >35 micromol/l) identified patients in whom laparoscopy was not useful.
Results: The CA19-9 level was >150 kU/l in 96 patients of whom 75 (78%) were considered resectable following laparoscopic assessment. There were 63 patients with a CA19-9 < or =150 kU/l of whom 60 (95%) were considered resectable following laparoscopic assessment. The sensitivity, specificity, positive predictive value and negative predictive value for CA19-9 < or =150 kU/l in predicting that laparoscopic assessment would judge patients as resectable were 44, 88, 95 and 22%, respectively. A cut-off level of < or =300 kU/l in patients with a bilirubin >35 micromol/l produced values of 30, 94, 94 and 28%, respectively. By using CA19-9 < or =150 kU/l, laparoscopy could have been avoided in 40% of patients, increased to 55% of patients with adjustment for the presence of jaundice; concomitantly, the yield from laparoscopy would have been increased from 15 to 22 and 25%, respectively.
Conclusion: Use of serum CA19-9 levels would increase the efficiency of laparoscopic staging in patients with suspected pancreatic malignancy.