Reliability and sensitivity of frozen-section pancreatic biopsy.

A collaborative Veterans Administration and University of California, Davis Medical Center group of 586 patients with histologically proved pancreatic carcinoma was reviewed. During laparotomy, 159 patients underwent 251 frozen-section pancreatic biopsies with subsequent permanent section examination of the same tissue block. All 112 positive frozen-section diagnoses were corroborated on permanent sectioning. The 47 patients with false-negative biopsy specimens were equally divided between sampling and interpretation error. We conclude that in this group of 159 pancreatic cancer patients, 30% failed to be correctly diagnosed by intraoperative frozen-section biopsy. This failure was due to patient sampling and interpretation error in equal proportion. Interpretation error rates were not influenced by the type or number of biopsies. Patient sampling error is apparently reduced by repeated biopsy, and specimen sampling error occurred less frequently with wedge biopsy.