Patient information list

Published: 18-10-2019| Version 1 | DOI: 10.17632/gm6mc2f4v7.1
xiaoli Cai


Pertinent clinical data, including patient demographics and laboratory test results,were collected from archived medical records. The following parameters were extracted from pathologic reports: histologic grade; tumor location; tumor size; residual tumor (R) classification; perineural invasion; and T and N stages of PDAC, as specified by American Joint Committee on Cancer (AJCC, 8th edition). In each patient, follow-up CE-CT or magnetic resonance imaging studies and laboratory profiles were obtained every 3-6 months postoperatively, defining DFS as the interval from date of surgery to tumor recurrence or most recent follow-up visit and OS as the period from date of surgery to patient death or most recent follow-up visit. Specimens removed during curative pancreatic resections were fixed in 10% formalin, sampling tumors thereafter for routine processing and standard hematoxylin & eosin (H&E) staining. To quantify fibrous tissue content, Masson’s trichrome and Sirius Red stains were also performed. Histologic evaluations of all specimens were reviewed by two experienced pathologists ,both credited with >10 years of experience in pancreatic pathology and blinded to clinical/radiologic patient data. Image Pro software, v6.0.0.260 (Media Cybernetics Inc, Rockville, MD, USA) was invoked to evaluate fibrotic stromal fractions. A total of 12 Semantic features were incorporated into the PDAC radiology reporting template , allocated as follows: location (head or uncinate process), shape, tumor-pancreas interface (conspicuity, scored 1-5), clear fat plane (yes/no), arterial and venous contact (scored 0-4), and positive secondary signs (pancreatic duct abruption, main pancreatic duct or common bile duct dilatation, parenchymal atrophy, contour abnormality, and presence of adenopathy). With respect to quantitative imaging features, preoperative CT images were first imported into the open-source 3D Slicer v4.10 application ( for precise image registration in each patient. Both rigid and deformable registration (elastix registration toolbox) was performed. Next, volumes of interest(VOI) in tumor and adjacent pancreas were segmented manually by one radiologist, avoiding pancreatic and bile ducts, vessels, duodenum, and artifacts. Segmentation was performed in transverse sections, slice by slice, at pancreatic phase. Registered volumes and VOIs were then saved and imported into Matlab R2018a-v9.4.0 (Math Works, Natick, MA, USA) to acquire CT attenuations at each pixel and to calculate volumetric mean attenuations of tumor and parenchyma at all registered phases. Delta values signified differences between two mean attenuations and ratios denoted proportions of attenuations between two volumes.