A Critical Assessment of Postneoadjuvant Therapy Pancreatic Cancer Regression Grading Schemes With a Proposal for a...

Currently, there is no consensus on the optimal tumor response score (TRS) system to assess regression in pancreatic cancers resected after neoadjuvant therapy. We developed a novel TRS (Royal North Shore [RNS] system) based on estimating the percentage of tumor bed occupied by viable cancer and categorized into 3 tiers: grade 1 (≤10%), grade 2 (11% to 75%), and grade 3 (>75%). We assessed 147 resected carcinomas with this and other TRS systems (College of American Pathologists [CAP], MD Anderson Cancer Center [MDACC], and Evans). The 3-tiered RNS system predicted median survival after surgery for grades 1, 2, and 3 of 54, 23, and 9 months, respectively (P<0.05). The CAP, MDACC, and Evans systems also predicted survival (P<0.05) but less consistently. The median survival for MDACC and CAP grade 0 (complete regression) was less than MDACC grade 1 and CAP grades 1 and 2. There was no difference in survival between CAP grades 2 and 3 (P=0.960), Evans grades 1 and 2a (P=0.395), and Evans grades 2a and 2b (P=0.587). Interobserver concordance was weak for CAP (κ=0.431), moderate for MDACC (κ=0.691), minimal for Evans (κ=0.307), and moderate to strong for RNS (κ=0.632 to 0.84). Of age, sex, size, stage, grade, perineural and vascular invasion, extrapancreatic extension, margin status, and RNS score, only RNS score, vascular invasion, and extrapancreatic extension predicted survival in univariate analysis. Only extrapancreatic extension (P=0.034) and RNS score (P<0.0001) remained significant in multivariate analysis. We conclude that the RNS system is a reproducible and powerful predictor of survival after resection for pancreatic cancers treated with neoadjuvant therapy and should be investigated in larger cohorts.