Interobserver reproducibility of percent Gleason grade 4/5 in total prostatectomy specimens.

Abstract

Recently the percent Gleason grade 4/5 was proposed as a predictor of the outcome of prostate cancer and it has been shown that it adds prognostic information to that given by Gleason score. To our knowledge the interobserver variability of percent Gleason grade 4/5 has not yet been investigated. We assessed the percent Gleason grade 4/5, including the identification of high grade patterns and estimation of the percent tumor area, which is potentially more difficult than conventional Gleason grading. A consecutive series of 69 total prostatectomy specimens was reviewed. A single slide per specimen was circulated among 4 genitourinary pathologists, who assessed Gleason score and the percent Gleason grade 4/5. Results were compared pairwise and a weighted kappa was calculated for Gleason score and the percent Gleason grade 4/5. The 4 observers had a mean weighted kappa for Gleason score and the percent Gleason grade 4/5 of 0.52 to 0.66 (overall mean 0.56) and 0.58 to 0.72 (overall mean 0.66), respectively. The best agreement for percent Gleason grade 4/5 was in 2 pathologists at the same department (weighted kappa 0.86). Transition zone tumors had a lower weighted kappa for Gleason score but a higher weighted kappa for percent Gleason grade 4/5 than peripheral zone tumors. In cases of the greatest disagreement in the percent Gleason grade 4/5 crush artifact, cribriform cancer and high grade PIN within the tumor were significantly more common. An intraobserver reproducibility of weighted kappa 0.91 was achieved for Gleason score and the percent Gleason grade 4/5. Interobserver reproducibility of the percent Gleason grade 4/5 is substantial and at least as good as that of the Gleason score. Hence, concern about interobserver variability should not deter pathologists from using the percent Gleason grade 4/5 as a prognostic marker for prostate cancer.