False‐negative rate of endoscopic ultrasound‐guided fine‐needle aspiration for pancreatic solid and cystic lesions with matched surgical resections as the gold standard

BACKGROUND The diagnosis of pancreatic tumors is often complicated because of sampling and interpretive challenges. The current study was performed to determine the rates, types, and causes of diagnostic discrepancies. METHODS The authors retrospectively reviewed cytology cases from 2004 to 2010 using matched surgical resection cases as the gold standard. RESULTS A total of 733 cases were divided into 3 categories: 1) positive or suspicious (290 cases); 2) negative or atypical (403 cases); and 3) unsatisfactory (40 cases). Of these cases, 101 fine‐needle aspiration (FNA) cases had matched surgical resections including 58 positive diagnoses, 39 negative diagnoses, and 4 unsatisfactory diagnoses. All 19 discrepant cases represented false‐negative diagnoses without any false‐positive cases noted, which included 2 cases with interpretive errors (10%) and 17 cases with sampling errors (90%). All matched cytology cases were divided into 5 subgroups based on the type of lesion or type of error and were analyzed for sensitivity and specificity. The sampling error rate in cystic lesions (8 of 24; 33%) was significantly higher than that in solid lesions (9 of 73; 12%). The false‐negative rate in the interpretive error group (3%) was significantly lower than that in the sampling error group (23%). CONCLUSIONS The results of the current study confirm that pancreatic endoscopic ultrasound‐guided FNA diagnosis has a very low false‐positive rate but a relatively high false‐negative rate using matched surgical resections as the gold standard. The major cause of a false‐negative cytology diagnosis is sampling error and the rate of sampling error in cystic lesions is significantly higher than that in solid lesions. Cancer (Cancer Cytopathol) 2013;121:449‐58. © 2013 American Cancer Society .