Evaluation of the Interobserver Agreement on the Resectability Status of Patients With a Pancreatic Cancer

Pancreatic adenocarcinoma frequently invades peripancreatic major vessels (common hepatic artery, celiac axis, superior mesenteric artery and mesenteric-portal confluence), requiring vascular resection or precluding the possibility of a radical resection. To Asses local resectability in pancreatic cancer patients is a key step to the whole process of care and it can change dramatically the outcomes. It is also extremely important in the reproducibility of different scientific studies that stratified or included patients on this feature. Resectability status classification of pancreatic adenocarcinoma, according to the definition of the National Comprehensive Cancer Network (NCCN) is widely used in current surgical and oncological practice to define the degree of vascular invasion: this classification describes three different classes of tumor depending on the local extension and vascular involvement. Resectable cancers are defined when no contact with the main vessels can be detected, whereas borderline resectable and non resectable tumors have a different extent of vascular infiltration. The recommended staging exam to define vascular involvement of pancreatic adenocarcinoma is contrast-enhanced CT scan; NCCN recommends that all patients with suspicion of pancreatic adenocarcinoma have a dedicated pancreas protocol CT scan as part of the initial assessment. From the attribution of a patient to a resectability class (resectable, borderline resectable or non resectable) derives the choice of a specific therapeutic pathway, as suggested by NCCN guidelines. International validated criteria to define local respectability are available but they can be prone to subjective interpretation. An evaluation of interobserver agreement on local resectability for pancreatic cancer in a multicenter setting is needed to understand the reliability and reproducibility of this system.

The primary aim of the study will be the assessment of interobserver variability to define vascular invasion and resectability status on CT scans (according to the last version of the NCCN Classification).One senior radiologist and one senior surgeon from different Italian high-volume Institutions will separately assess the degree of vascular invasion by pancreatic cancer and the definition of resectability status according to NCCN. 69 CT scans, randomly selected from a pool of 138 studies (provided by each center and collected by the study coordinator) of non-metastatic consecutive patients with pancreatic cancers that signed the informed consent, will be evaluated. CT scans will be provided by participating centres, that will send them to the coordinating centre until reaching the planned enrolment. The images and related DVDs will be made anonymous and identified just by serial number and initials of the patient. The CT scan reports will be collected by each centre and made available by the coordinating centre for subsequent analysis. The criteria used to define the resectability status are those developed by the NCCN and included in the latest guidelines. Observers will be blinded to the CT-scan reports, as well as to the clinical history of the patient, and to the votes of the other raters. The degree of each vessel invasion and assignment of each tumor to a category of resectability status (resectable - borderline resectable - non resectable) will be defined separately by all raters, by fulfilling a template adopted and reported in NCCN guidelines. An interrater reliability analysis using the Kappa statistic will be performed to determine consistency among observers on the respectability status. The consistency among observer will be also evaluated for each point of the criteria and section of the template to asses vascular involvement in detail.

In case of significant fair or poor agreement (K < 0.50), the participating raters (consensus group), including the study coordinator, will meet to discuss the cases that will have less than 100% agreement among all the raters. Scans will be reviewed, an open discussion ensued. Based on the discussion, the consensus group will develop proposals to address the specific areas of disagreement. The CT scans not select for the first evaluation will be than evaluated, using the proposal of the consensus group by a new set of rater (one senior radiologist and one senior surgeon from each institution involved) and the new Kappa statistic will be calculated.