Grading prostate cancer.

Abstract

Histologic tumor grade is a strong predictor of outcome for men with prostate cancer. All existing grading systems successfully identify well-differentiated cancer, which progresses slowly, and poorly differentiated cancer, which progresses rapidly, but they are less successful in subdividing most moderately differentiated cancers, which have an intermediate malignant potential. The Gleason system, the de facto standard for grading, identifies histologic patterns by the degree of glandular differentiation without relying on morphogenetic or histogenetic models; it reflects tumor heterogeneity by combining primary and secondary patterns into a cancer score. Modifications that have been proposed for Gleason grading include morphometric nuclear grading, grouping of grades, estimating the amount of high-grade cancer (Gleason patterns 4 and 5), and including the cribriform pattern as Gleason pattern 4 rather than 3. Most variants of prostate cancer are high grade (Gleason patterns 4 and 5), including small cell undifferentiated carcinoma, signet ring cell carcinoma, sarcomatoid carcinoma, and carcinosarcoma. The Gleason system can be reproduced by most investigators, although there is a small but significant level of interobserver and intraobserver variability that is unavoidable. When compared with matched prostatectomy specimens, contemporary 18-gauge needle core biopsy underestimates tumor grade in 33% to 45% of cases and overestimates grade in 4% to 32% of cases, similar to results with traditional 14-gauge biopsies. Grading errors are common in biopsy specimens with small amounts of tumor and low-grade tumor, and are probably due to tissue sampling error and tumor heterogeneity. Upgrading of prostate cancer may occur after radiation therapy but is common after androgen-deprivation therapy. Univariate and multivariate analyses of prognosis in prostate cancer almost always identify cancer grade as one of the most significant predictors of patient outcome. The combination of cancer grade with other prognostic variables to create a multiple prognostic index should allow greater precision in predicting outcome for individual patients.