Interobserver Variability in Grading of Response to Neoadjuvant Chemoradiotherapy in Rectal Adenocarcinomas Using Five Grading Systems
P Roy, S Serra, R Chetty. University Health Network, Toronto, ON, Canada
Background: Neo-adjuvant chemoradiation (NACRT) is a standard treatment option for locally advanced rectal adenocarcinoma. Various scoring systems are currently in use to assess the regression of the tumour following CT/RT. The robustness of any scoring system depends on interobserver agreement based on defined criteria. There is a paucity of studies that test the interobserver agreement of the grading systems commonly used in clinical practice for rectal adenocarcinoma.
Design: Cases of rectal adenocarcinomas treated with NACRT followed by surgery were identified from the archives of the Department of Pathology at the Toronto General Hospital. Three observers independently examined the slides to assess the degree of regression using five different scoring systems: the Dvorak's regression scoring system (DRS), Royal College of Pathologist Cancer Dataset method (RCPath), the Rectal Cancer Regression Grade(RCRG), the modified Rectal Cancer Regression Grade(m-RCRG), and the Tumour Regression Grade (TRG) . The crude percentage of overall agreement was calculated and the inter-observer variability for each system was calculated using the Fleiss' modification of Cohen's kappa κ.
Results: 41 cases of rectal adenocarcinoma were graded for degree of regression. The calculated percentage of overall agreement and the corresponding κ values for interobserver agreement were 71.5% and 0.57 for DRS; 70.7% and 0.45 for RCPath; 69.1% and 0.50 for RCRG; 62.6% and 0.42 for m-RCRG; 55.2% and 0.38 for TRG. Using Landis and Koch's method for interpreting κ values, a moderate agreement (0.41-0.60) was noted with all the systems except the TRG for which the agreement was fair (0.21-0.40). We found the Dvorak's system the easiest to apply and with the highest degree of agreement among the 5 systems.
Conclusions: The interobserver variability of the systems for assessing degree of regression in rectal adenocarcinomas post NARCT must be factored into their interpretation. The Dvorak's system is the most reproducible of the systems currently used in clinical practice.
Monday, March 22, 2010 9:30 AM
Poster Session I Stowell-Orbison/Surgical Pathology/Autopsy Awards Poster Session # 108, Monday Morning