Diagnostic Accuracy of Endoscopic Ultrasound-Guided Fine-Needle Aspiration in Patients with Pancreatic Adenocarcinoma Using Histology as the Gold Standard
Shantel Hebert-Magee, Amanda Treece, Faisal Mukhtar, Isam Eltoum. University of Alabama at Birmingham, Birmingham, AL
Background: Endoscopic ultrasound (EUS)–guided fine-needle aspiration (FNA) of the pancreas has been well established, by numerous single center studies, as an effective minimally invasive modality in confirming pancreatic malignancy. We recently performed a meta-analysis of EUS-FNA of pancreatic adenocarcinoma (PA) in solid pancreatic lesions to determine the true diagnostic accuracy since variance amongst the published reports exists. We found that less than half (17/40) of the eligible studies consistently had rapid on-site examination (ROSE) provided by a cytopathologist, only 7/40 used histology alone, and none of the studies solely focused on PA. Due the lack of data on the diagnostic accuracy of PA by EUS-FNA, we decided to evaluate this modality using ROSE and histology alone as the reference standard since the results of our meta-analysis suggested that a composite reference standard tends to inflate the diagnostic accuracy.
Design: We reviewed our ten year experience with EUS-FNA in patients with solid pancreatic lesions; data were assessed from all patients who presented with CT imaging of a pancreatic mass and underwent EUS-FNA at our institution between January 1, 2000 and December 1, 2010.
Results: EUS-FNA was performed in 2397 patients; 500 patients had subsequent histological follow-up. A final diagnosis of PA was established in 274/500 patients (54.8%); 242 true-positives, 224 true-negatives, and 2 false-positives. The sensitivity, specificity, and accuracy of EUS-FNA for diagnosis of PA were 88.68%, 99.1%, and 93.4%, respectively. The positive predictive value (PPV) and negative predictive value (NPV) of EUS-FNA for PA were 99.18% [95% confidence interval (CI) 97.07 to 99.78%] and 87.84% (CI 83.26 to 91.30%), respectively.
Conclusions: We conclude that the diagnostic accuracy of EUS-FNA of PA is slightly lower when histology alone is used as the reference standard, suggesting clinical follow-up as a reference component portends bias.
Tuesday, March 20, 2012 1:00 PM
Poster Session IV # 286, Tuesday Afternoon