Reproducibility of Gleason grading of prostate cancer can be improved by the use of reference images

Objectives. The Gleason system has become the international standard for grading prostate cancer. Its interobserver reproducibility is fairly good but needs improvement. A national effort to standardize the Gleason grading in Sweden was evaluated. Methods. A teaching set of 40 images illustrating the Gleason grades was distributed to 85 Swedish pathologists, including 73 specialists and 12 residents. Their ability to assign correct grades (1 to 5) to a series of microphotographs was tested before and after the distribution of the teaching set. Each test included 20 images, graded by an expert panel. Results. The proportion of correct grades improved from 70.5% in the first test to 86.6% in the second test (P < 0.001). Undergrading decreased from 22.5% to 9.4%. Grading errors of more than one step were uncommon in test 1 (1.9%) and almost disappeared in test 2 (0.1%). The average kappa value of tests 1 and 2 was 0.58 and 0.78, respectively. Conclusions. The results of this study reveal that a set of reference images may significantly improve the reproducibility of grading. This method is inexpensive and based on modern imaging techniques. The reference images can be easily distributed to a large number of pathologists, either as paper copies or through the Internet (http://www.pathology.ks.se/egevad/gleason.html), and can, thereby, be readily available in everyday practice. UROLOGY 57: 291-295, 2001. (C) 2001, Elsevier Science Inc.