The Benefits and Limitations of On-Site Smear Review During EUS-FNA for Pancreatic Adenocarcinoma

Abstract

Purpose: EUS-guided FNA (EUS-FNA) of pancreatic masses often relies on rapid, on-site cytopathologic interpretation following each “pass” to maximize diagnostic yield. Anecdotally, many endosonographers and cytopathologists believe the first pass is the most diagnostic. We aimed to determine the test characteristics of on-site cytopathologic interpretation for EUS-FNA of pancreatic adenocarcinoma. Methods: We retrospectively identified patients sequentially referred to our medical center for EUS-FNA of pancreatic adenocarcinoma. For each procedure, a cytopathologist was present and performed both an on-site rapid histologic “smear” interpretation following each pass and later histologic confirmation of adenocarcinoma using the “cell block.” The main outcome of interest was the pass at which on-site cytopathologic interpretation was diagnostic for adenocarcinoma. Results: From March 2007 until April 2009, 120 patients underwent EUS-FNA that was diagnostic for pancreatic adenocarcinoma at our medical center. Although each cell block confirmed adenocarcinoma, the on-site cytopathologist interpretation of the smears diagnosed 90 masses as adenocarcinoma, 23 as “suspicious” and 7 as “atypical.” Of those diagnosed with adenocarcinoma based on the smear, 42% were diagnosed on the first needle pass, 14% on the second pass and 13% on the third pass (p=0.01). No significant confounding was demonstrated via regression analysis when controlling for the size of the lesion, location, needle gauge, needle type, endosonographer or interpreting pathologist. Patients diagnosed on the first pass underwent less total passes than those diagnosed on subsequent passes (median 4 vs. 5, p<0.01). In patients who underwent more than 6 passes, only 3% were diagnostic for adenocarcinoma vs. 20% being labeled suspicious or atypical (p<0.05). The procedural complication rate in patients undergoing more than six passes was 38%, although there was no difference in the rate between those with diagnostic smears and those with atypical or suspicious smears (p=1.00). Conclusion: In patients with confirmed pancreatic adenocarcinoma on the cell block, rapid, on-site cytopathologic interpretation is most diagnostic on the first pass with very low yield on the preliminary smears following the sixth pass. If the clinical suspicion for adenocarcinoma is high even after six non-diagnostic passes on smear review, additional passes may be useful for the cell block, but at a cost of higher complication rates.