Ultrasonically guided histological and cytological fine needle biopsies of the pancreas. Reliability and reproducibility of diagnoses.

In 100 consecutive patients ultrasonically guided histological and cytological fine needle biopsy specimens were obtained from pancreatic lesions using two different needles with an outer diameter of 0.6 mm. Specimens taken by both cytological and histological fine needle biopsy were examined blindly by two pathologists. When related to the final and reliable diagnosis obtained in 57 patients, the predictive value of a malignant diagnosis was 1.00 for both types of biopsy. The predictive value for a benign diagnosis was 0.25 for histological specimens for both examiners and 0.33 and 0.45 for the two evaluations of the cytological specimens. False benign diagnoses seemed to be related to both sampling error and difficulties in interpreting the biopsy specimens. The intraobserver and interobserver kappa values concerning reproducibility of diagnoses were higher for histological specimens (0.80 and 0.74) than for cytological specimens (0.70 and 0.61). Consistent malignant diagnoses, however, occurred more often with cytological specimens (51 cases) than with histological specimens (39 cases) (p less than 0.05) and consistent diagnoses of insufficient material were more common with histological specimens (18 cases v six cases). Cytological fine needle biopsy seems to be the method of choice if only one method is used and a 0.6 mm needle is used.