Objective: To evaluate the accuracy of Gleason grading of prostate cancer in multiple core biopsies, compared with the final Gleason score of total prostatectomy specimens, and to investigate whether the prediction of the correct Gleason score is improved by increasing the number of biopsies.
Patients and methods: Before total prostatectomy, 121 men had a mean (range) of 10.0 (8-14) transrectal ultrasonography (TRUS)-guided core biopsies taken from the apex, mid-medial, mid-lateral and basal regions, from the transition zone and from lesions detected on TRUS. The biopsies and prostatectomy specimens were reviewed and the Gleason scores assessed.
Results: The preoperative biopsies predicted the prostatectomy Gleason score exactly in 45.5% of the patients and within one Gleason score in 93.4%. The biopsies under-graded the prostate cancer in 38.8% and overgraded it in 15.7%. The weighted kappa value for exact agreement was 0.502. If one biopsy was positive for cancer, the prostatectomy Gleason score was predicted correctly in 43.8% and within one score in 93.8%, compared with 53.8% and 92.3%, respectively, if cancer was found in at least seven biopsies. If the mid-lateral and transition zone biopsies had been excluded from the biopsy protocol, 5% of the cancers would have been undetected. Among the remaining 115 cancers, grading accuracy only improved from 43.5% to 45.2% by adding biopsies to the sextant protocol.
Conclusion: Despite a statistically significant agreement between biopsy and prostatectomy Gleason score, under-grading remains a major problem. The prediction of the prostatectomy Gleason score is only marginally improved by increasing the number of biopsies.