Endoscopic ultrasound (EUS) enhances pre-resection stage determination compared to CT or PET for ampullary cancer

Joyce Wong, MD, Zachary J Thompson, PhD, Cynthia Harris, MD, Jason Klapman, MD, Shivakumar Vignesh, MD, Barbara Centeno, MD, Pamela J Hodul, MD. Moffitt Cancer Center.

Introduction: Ampullary cancers are relatively uncommon neoplasms; conventional imaging has limited sensitivity in determining pre-resection stage. This study set out to evaluate the utility of endoscopic ultrasound (EUS) as an adjunct to computed tomography (CT) and positron electron tomography (PET) in the setting of ampullary neoplasms.

Methods: A database of patients undergoing resection for ampullary cancer from 1996-2013 was retrospectively reviewed. Patients undergoing pre-operative imaging with CT, PET and EUS were included in this study. Frequency of detection and agreement between imaging modalities were performed with Kappa statistics and exact binomial confidence intervals, where applicable. Goodman-Kruskal gamma statistics were used to quantify the correlation between imaging and pathologic stage.

Results: 91 patients underwent resection for ampullary cancer; in this cohort, 68 (75%) had adenocarcinoma, an additional 17 (19%) had adenocarcinoma, intestinal type. Six (6%) had uncommon pathology: papillary carcinoma and mucinous carcinoma. 79 (87%) had a pre-operative CT scan; 25 (33%) had a mass identified in the ampullary region on CT. 43 (47%) had a PET scan, of which 33 (77%) demonstrated avidity in the ampullary region. 56 (62%) underwent EUS; 49 (88%) of which demonstrated a mass. When correlating patients who underwent both CT and EUS, the Kappa correlation (95% CI) for presence of mass was 0.12 (-0.01-0.24), with EUS demonstrating a mass more frequently than CT, p<0.001. When comparing PET detection of an ampullary mass compared to EUS, Kappa correlation was -0.1 (-0.37-0.17).

90 (99%) patients underwent pancreaticoduodenectomy; one (1%) patient underwent ampullectomy. Final pathology demonstrated 26 (29%) with stage 3 disease, 36 (40%) with stage 2 disease, 28 (31%) with stage 1 disease, and 1 (1%) with no residual tumor identified. Pre-operative EUS stage was categorized into stage 1 or stage 2/3 and compared with final pathology. Gamma correlation was only 0.3 (-0.31-0.91) between EUS tumor (T) stage and final T stage; 16 (73%) with T stage 3 or 4 were identified by EUS, versus 7 (41%) of T stage 1 or 2 corresponding to EUS T stage 1 or 2. Gamma correlation was 0.74 (0.34-1.00) between EUS stage and final pathologic stage. Correlation was improved in stage 2 and 3 disease, with 25 (83%) with EUS stage 2/3 correlating to pathologic stage 2/3 disease; only 4 (50%) with stage 1 disease had a corresponding stage 1 by EUS. Presence of mass on CT, dilated biliary duct on CT, or PET avidity of the ampullary region were not correlated with final pathologic stage.

Conclusions: Ampullary cancers present a diagnostic challenge to clinicians, often with conventional imaging providing limited staging information. This study supports the use of EUS as an adjunct that may provide improved localized staging information, particularly for more advanced cancers.

View Poster

« Return to SAGES 2014 abstract archive