Pancreatic FNA in 1000 cases: a comparison of imaging modalities.

BACKGROUND: Image-guided FNA is a popular method for evaluating pancreatic lesions, but few large studies on pancreatic FNA exist.

METHODS: Cytologic material, imaging reports, and clinical follow-up information were reviewed from pancreatic FNA cases performed over a 5-year period.

RESULTS: A total of 1050 pancreatic FNAs were obtained by EUS (n = 843), US (n = 140), and CT (n = 67). On-site assessment was performed in 89.2% (n = 937) of cases. Findings were as follows: positive for neoplasm 48.9% (n = 503), negative 29.1% (n = 306), descriptive 10% (n = 105), suspicious 5.9% (n = 62), atypical/inconclusive 4.6% (n = 48), and nondiagnostic/inadequate 1.5% (n = 26). Follow-up in the form of histology or at least 6 months of clinical observation was available for 61.2% (n = 643). There was an overall false-positive rate of 0.3% and a false-negative rate of 14.3%. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were as follows: 79.4, 99.0, 99.4, 67.9, 84.5 for the total series, respectively; 79.9, 98.8, 99.2, 72.5, 86.5 for EUS, respectively; 77.9, 100, 100, 48.6, 81.7 for US, respectively; and 78.6, 100, 100, 47.1, 82.0 for CT, respectively. In general, accuracy was not influenced by lesion size or site, number of FNA passes, or number of procedures per patient. After controlling for lesion size, EUS resulted in greater accuracy than US or CT when evaluating lesions <3 cm ( p = 0.015).

CONCLUSIONS: All imaging modalities showed moderate to high sensitivity, specificity, and accuracy. Logistic regression analysis showed that for lesions <3 cm, the EUS method had higher accuracy than US or CT. No statistically significant difference was seen for larger lesions or for the number of FNA passes.