Pancreatic Cancer Primary Tumor Size on Multidetector Row CT vs. Pathologic Specimen: Implications for Radiation Treatment Planning

Abstract

There is often a discrepancy between pancreatic cancer primary tumor size as imaged on CT, and that found on pathologic specimen after surgical resection. Prior reports have focused on sensitivity of multidetector row CT (MDR-CT) for determining surgical resectability, but not on metrics such as tumor size. Dimensions of the primary tumor are increasingly relevant in the era of highly conformal treatment such as stereotactic body radiotherapy (SBRT), in which the gross tumor volume (GTV) is typically expanded by only 2–3 mm to generate a planning target volume (PTV). Ninety-seven patients with pancreatic cancer were retrospectively evaluated at two hospitals. All patients underwent MDR-CT imaging prior to surgical resection. One patient received neoadjuvant chemotherapy. Primary endpoints were maximum dimension (in mm) of the primary tumor as reported by the radiologist on CT and by the pathologist on resected gross specimen. Maximum dimension measured on endoscopic ultrasound (EUS) was analyzed if available. Median age at surgery was 63. Pathology was pancreatic adenocarcinoma in 86 patients (89%). Eighty patients (82%) had pT3 disease, 35 patients (36%) had a positive margin, and 77 patients (79%) had positive lymph nodes. Eighty-seven patients (90%) had CT scans available for review. The CT scans were performed a median of 19 days before surgery. Median primary tumor size (maximum dimension) on pathology was 32 mm. Seventy-three patients (84%) had a maximum tumor dimension that was larger on pathologic specimen as compared to CT scan. Median and mean differences were 9 mm and 12 mm larger on pathologic specimen compared to CT (interquartile range, 4–21 mm, signed rank p < 0.0001). In patients for whom both CT scan and EUS were performed (n = 46), largest tumor dimension was better estimated by EUS, with median difference compared to pathologic specimen being 5 mm for EUS and 9 mm for CT (median improvement 3 mm, interquartile range, 0–11 mm, signed rank p < 0.002). Findings were similar between hospitals. The CT scans significantly under-represent the size of pancreatic cancer primary tumors as compared to findings on pathologic specimen in resectable cases, even in the MDR-CT era. While this investigation only examined the largest tumor dimension, our findings imply a marked under-representation of the GTV for this disease, raising the possibility that a clinical tumor volume (CTV) utilizing a larger margin should be incorporated into CT-based planning for highly conformal treatment such as SBRT. Interestingly, EUS appeared more accurate than CT for defining primary tumor size, suggesting that it could play a new role in determining CTV.