Agreement validation between axial imaging modalities and endoscopic ultrasonography in staging resectability of pancreatic cancer

Autor: Faaiza Mahmoud, William Small, Timothy Ozga, Brendan Martin, C M Dudiak, Nghia Pham, Ebenezer Enchia, Sam G. Pappas, Thomas Aquisto, Aakash Ahuja, Tejvir Nanda, Tarita O. Thomas, Nihal Gooneratne, Neil Gupta, Adam Gliniewicz
Rok vydání: 2017
Předmět:
Zdroj: Journal of Clinical Oncology. 35:273-273
ISSN: 1527-7755
0732-183X
Popis: 273 Background: Vascular involvement is a key determinant of resectability and treatment sequencing in pancreatic cancer (PC). Current staging requires multidetector CT (MDCT) and endoscopic ultrasonography (EUS), but there are still occasions when vascular involvement cannot be fully assessed. Current NCCN guidelines recommend using magnetic resonance imaging (MRI) for clarifying these ambiguities. The primary aim of our study is to access the accuracy of CT, MRI, and EUS in staging resectabilty of PC. The secondary aim is to assess internal consistency in the evaluation of vascular involvement among different staging modalities. Methods: We retrospectively reviewed 40 medical records of patients who were treated at LUMC for PC from 2006 to present. We evaluated 8 MDCT scans and 8 MRI scans in addition to their EUS reports. Seven board certified radiologists analyzed the 16 imaging studies using a standardized template. Unweighted Kappa statistic and observed agreement analysis was calculated to assess inter-observer agreement on vessel involvement between the imaging studies compared to the EUS reports. We sought to examine degree of concordance for involvement of the superior mesenteric artery (SMA), superior mesenteric vein (SMV), and main portal vein (MPV) with the pancreatic tumor. Results: Low Kappa and observed agreement in respect to vascular involvement between MDCT and EUS was found for SMV (K = 0.33, 62.5%), SMA (K = -0.2, 62.5%), and MPV (K = 0.38, 75%), indicating low concordance. Low Kappa and observed agreement in respect to vascular involvement between MRI and EUS was found for SMV (K = 0.09, 37.5%), SMA (K = 0.14, 62.5%), and MPV (K = 0.47, 75%). Low Kappa and observed agreement in respect to vascular involvement between MRI and MDCT in comparison to EUS was found for SMV (K = 0.11, 50%), SMA (K = 0.20, 75%), and MPV (K = 0.14, 62.5%). Conclusions: Concordance was shown within each imaging modality when analyzed by radiologists. Due to high discordance across different staging modalities, however, MRI and CT scans complement each other in determining degree of vascular involvement. We believe both imaging modalities should be used in staging borderline resectable PC.
Databáze: OpenAIRE