Prediction of percent Gleason grade 4/5 by multiple core biopsies

To evaluate whether percent Gleason grade 4/5 (i.e. the proportion of a tumor occupied by high-grade cancer) can be predicted by multiple needle biopsies.

In 115 men, 8-14 (mean 10) biopsies were taken, including eight from standardized positions (apex, mid-medial, mid-lateral and base). Biopsies were reviewed and cancer lengths measured. All men underwent radical prostatectomy. The prostatectomy specimens were totally embedded and tumor volume measured planimetrically. Gleason scores and percent Gleason grade 4/5 were assessed for both biopsy and prostatectomy specimens.

Percent Gleason grade 4/5 in prostatectomy specimens was predicted correctly in 34% of cases and within 10%, 20% and 30% in 55%, 64% and 73% of cases, respectively. Biopsies had a sensitivity, specificity and accuracy for Gleason grade 4/5 of 62%, 87% and 69%, respectively. Positive and negative predictive values were 93% and 45%, respectively. The weighted kappa value for agreement was slightly higher for Gleason score (0.685) than for percent Gleason grade 4/5 (0.573). The univariate correlation for percent Gleason grade 4/5 in biopsies and the main tumor was r=0.62, r(2)=0.39 (p<0.001). In univariate logistic regression, percent Gleason grade 4/5 on biopsies predicted the presence of any Gleason grade 4/5 cancer in the main tumor (p=0.009).

Gleason grade 4/5 in prostatectomy specimens correlates with findings in preoperative biopsies. Whether this measure will be used in routine practice remains to be seen.