Significant Individual Variation Between Pathologists in the Evaluation of Colon Cancer Specimens After Complete Mesocolic Excision

BACKGROUND:After the introduction of complete mesocolic excision, a new pathological evaluation of the resected colon cancer specimen was introduced. This concept has quickly gained acceptance and is often used to compare surgical quality. The grading of colon cancer specimens is likely to depend on both surgical quality and the training of the pathologist. OBJECTIVE:The purpose of this study was to validate the principles of the pathological evaluation of colon cancer specimens. DESIGN:This was an exploratory study. SETTINGS:The study was conducted in Aarhus, Denmark, and Leeds, United Kingdom. PATIENTS:Colon cancers specimens were used. MAIN OUTCOME MEASURES:The agreement of gradings between participants was of interest. Four specialist GI pathologists and 2 abdominal surgeons evaluated 2 rounds of colon cancer specimens, each at 2 separate time points. Each round contained 50 specimens. After the first round, a protocol of detailed principles for the grading procedure was agreed on. Results from an experienced pathologist were considered as the reference results. RESULTS:In the first round, the distribution of gradings between participants showed substantial variation. In the second round, the variation was reduced. Intraobserver agreement was mostly fair to good, whereas interobserver agreement was frequently poor. This did not significantly change from round 1 to round 2. LIMITATIONS:The small sample size of 100 specimens provided a very small number of specimens resected in the muscularis propria plane, which renders the evaluation of this group potentially unreliable. The evaluations were made on photos and not on fresh specimens. CONCLUSIONS:This study demonstrates significant variation in the pathological evaluation of colon cancer specimens. It demonstrates that it cannot be used in clinical studies, and care should be taken when comparing results between different hospitals.