A working group classification of focal prostate atrophy lesions
Focal atrophy is extremely common in prostate specimens. Although there are distinct histologic variants, the terminology is currently nonstandardized and no formal classification has been tested for interobserver reliability. This lack of standardization hampers the ability to study the biologic and clinical significance of these lesions. After informal and formal meetings by a number of the authors, focal atrophy lesions were categorized into 4 distinct subtypes as follows: (i) simple atrophy, (ii) simple atrophy with cyst formation, (iii) postatrophic hyperplasia, and (iv) partial atrophy. In phase 1 of the study, pathologists with varying levels of experience in prostate pathology were invited to view via the Internet a set of "training" images with associated descriptions of lesions considered typical of each subtype. In phase 2 of the study, each participant provided diagnoses on a series of 140 distinct "test" images that were viewed over the Internet. These test images consisted of the 4 subtypes of atrophy and images of normal epithelium, high grade prostatic intraepithelial neoplasia, and carcinoma. The diagnoses for each image from each pathologist were compared with a set of "standard" diagnoses and the kappa statistic was computed. Thirty-four pathologists completed both phases of the study. The interobserver reliability (median kappa) for classification of lesions as normal, cancer, prostatic intraepithelial neoplasia, or focal atrophy was 0.97. The median kappa for the classification of atrophy lesions into the 4 subtypes was 0.80. The median percent agreement with the standard diagnosis for the atrophy subtypes were: simple 60.6%, simple with cyst formation 100%; postatrophic hyperplasia 87.5%; partial atrophy 93.9%. The lower percentage for simple atrophy reflected a propensity to diagnose some of these as simple atrophy with cyst formation. Seven pathologists completed the phase 2 analysis a second time, and their intraobserver reproducibility was excellent. Three of 4 pathologists with low agreement with the standard diagnosis for simple atrophy improved their scores after repeating the analysis after re-examination of the "training set" of images. In conclusion, these criteria for variants of focal prostate atrophy may facilitate studies to examine the relation between various patterns of prostate atrophy and prostate cancer.