Preoperative CT Classification of the Resectability of Pancreatic Cancer: Interobserver Agreement

Abstract

P ancreatic cancer has a dismal prognosis with a 5-year survival rate of only 8% (1). However, surgical resection with a negative margin may be a potentially curative treatment option for this disease. Recently, the concept of borderline resectable pancreatic cancer has emerged to indicate a special category of potential resectability that may help refine the conventional resectable or unresectable binary stratification system (2). Although there is no widespread consensus on the ideal treatment for borderline resectable pancreatic cancer (3-5), neoadjuvant treatment has been suggested to enhance the margin-negative (ie, R0) resection rate for this disease and to enable better patient selection for surgical procedures (6). In this regard, accurate categorization of patients according to their resectability status at initial workup would have a major impact on the management decision (2,7). Furthermore, proper selection of surgical candidates would offer the best chance to cure patients with resectable disease while preventing futile laparotomies in those with unresectable diseases.The National Comprehensive Cancer Network guidelines (version 2.2017; 8) recommend pancreatic protocol multi-detector row CT, including pancreatic and portal venous phases, preferably CT angiography with thin-section axial images, multiplanar reformats, and maximal intensity projection or three-dimensional volumetric images, for the initial evaluation of patients who are clinically suspected of having pancreatic cancers. Indeed, for the prediction of the resectability of pancreatic cancer, CT shows good diagnostic performance (9-11). Currently, in an attempt to standardize the imaging interpretation of the local resectability of pancreatic cancers, several guidelines define the vessels that should be assessed and the imaging criteria to define the tumor-vascular relationship: abutment (,180°),

Radiomics-Assisted Presurgical Prediction for Surgical Portal Vein-Superior Mesenteric Vein Invasion in Pancreatic Ductal Adenocarcinoma

“…Interestingly, though the interobserver agreement between the radiologists was moderate, their sensitivities were both unsatisfactory. The diagnostic performance of the radiologists was concordant with that of similar studies on CT (16)(17)(18)(19); about 20% of cases were false negative according to the qualitative imaging features. Though a recent study reported that endoscopic ultrasound elastography could improve diagnostic performance of vascular invasion in PDAC, the elastography technique was not commonly used for all patients (28).…”

“…invasion. Concordance to the previous study, pancreatitis-related changes usually blur the contact region between the solid tumor and adjacent vessels; this led to discrepancy in the interpretation of the degree of tumor-vascular contact (abutment vs encasement, <180°vs >180°, respectively) and may have caused the interobserver variability (19).…”

“…Vascular morphologic changes on visual assessment, such as vascular deformation, narrowing, or teardrop sign, sometimes were difficult to judge the presence of invasion, this could result in low sensitivity of radiologists in detecting PV-SMV invasion. Concordance to the previous study, pancreatitis-related changes usually blur the contact region between the solid tumor and adjacent vessels; this led to discrepancy in the interpretation of the degree of tumor-vascular contact (abutment vs encasement, <180° vs >180°, respectively) and may have caused the interobserver variability ( 19 ).…”

“…Unfortunately, such above qualitative imaging findings do not accurately classify vascular involvement, especially in peripancreatic venous involvement ( 16 – 18 ). Also, a recent study showed that agreements in the interpretation of tumor-vascular relationships were low among different observers ( 19 ).…”

Systematic Analysis of Accuracy in Predicting Complete Oncological Resection in Pancreatic Cancer Patients—Proposal of a New Simplified Borderline Resectability Definition

“…Furthermore, the false positive assessment of arterial encasement occurs frequently [11]. Joo et al recently performed a study assessing interrater reliability among radiologists evaluating resectability in PDAC [12]. The authors demonstrated an interrater reliability as low as 30%, and in cases determined as borderline resectable, resection rates varied from 0% to 74%.…”