Interobserver Variability in Assessing Pathologic Response to Preoperative Treatment in Rectal Cancer: Standardization of an Evaluation Method and Comparisons Between Published Scales

Background Evaluating tumor response of rectal cancer to preoperative chemoradiotherapy (NCRT) has a prognostic value on overall survival; however, grading tumor response is a controversial issue due to lack of reproducibility and the lack of information about the standardization of the evaluation. Methods We performed this study to examine the variability between observers' assessment of the pathological responses to NCRT using a systematic quantitative grading system based on a percentage of tumor response against the proportion of residual tumor burden. As a secondary aim, we classified the tumor response according to six published systems to determine the correlation between the observers into each grading system. Results From 70 cases, the mean age was 60.6 +/- 11.78 years, 36 (51.47%) patients were female, the pathological T stage was pT3 in 48.6% of cases, pT2 in 32.9%, pT1 in 11.4% and 7.1% in pT4, whereas 40% had lymph node metastasis. The median lymph node count was ten lymph nodes (range 6-43). Our method of tumor regression evaluation has a good intraclass correlation (ICC) value. From the scales compared regarding interobserver agreement, the Ryan's and Royal College of Pathologists showed fair agreement (but good ICC); the scales from Dworak, Becker, and Rizk showed substantial agreement (and good to excellent ICC values); and the scale from Rodel showed almost-perfect agreement. Results All the evaluated systems showed good interobserver agreement, but the best interobserver agreement was reached with the Rodel's scale.