Pancreatic fine-needle aspiration cytopathology: an analysis of the CAP NGC program for pancreatic FNA 2003-2011.

INTRODUCTION

The College of American Pathologists monitors quality in cytologic analysis in its nongynecologic cytology sample program. We report the performance of participating laboratories in pancreatic fine-needle aspiration sample analysis.

MATERIALS AND METHODS

We evaluated 23,079 responses to 392 pancreatic fine-needle aspiration slide challenges that were collected between January 6, 2003 and December 31, 2011. The analysis examined concordance to the reference diagnosis as well as performance of conventional Papanicolaou smears, Romanowsky smears, CytoSpin and ThinPrep preparations. A nonlinear mixed model was fit with 3 factors: reference diagnosis, reader type, and preparation type.

RESULTS

Overall concordance rate was 93.2%, 94.8% for ductal adenocarcinoma, and 96.2% for interpretation of malignancy in cases of neuroendocrine tumors. There was no difference in performance between pathologists and cytotechnologists. In negative/benign preparations, there was a 76.3% concordance to the reference diagnosis. There was 89.2% sensitivity for diagnosis of malignancy when adenocarcinoma was present and 72.8% specificity for a benign non-neoplastic diagnosis with a tendency to overcall and demonstrate insecurity by providing a number of incorrect diagnoses for benign entities. Sensitivity of an exact diagnosis of neuroendocrine lesion when a neuroendocrine tumor is present was 79%. Concordance for diagnosis of mucinous cystic neoplasm without cytologic atypia was problematic at 46.4% with participants yielding an erroneous interpretation of adenocarcinoma one-third of the time.

CONCLUSIONS

Participants performed well in recognizing adenocarcinoma, but they overcalled negative samples. Findings can provide focus for education and suggest that efforts be directed at benign pancreatic samples, neuroendocrine cytomorphology, and mucinous neoplasms.