Purpose: The higher number of high grade cancers noted in the Prostate Cancer Prevention Trial may have arisen due to a previously unknown association between prostate volume and sextant biopsy derived grade, rendering it more difficult to detect high grade cancer in men with a larger prostate (volume-grade bias). A basic tenet of measurement theory is that repeat measurement improves validity. We determined whether an extended pattern prostate biopsy technique could overcome this bias.
Materials and methods: We reviewed the record of 679 patients who underwent systematic extended (10 cores or greater) biopsy with isoechoic prostate transrectal ultrasound and prostate specific antigen less than 10 ng/ml. Since specimens were separately labeled, we were able to compare the grade of the first 6 cores vs that of the extended pattern. We determined the highest grade achieved using traditional sextant biopsy and the change induced by marginal samples across volume tertiles, hypothesizing that if upgrading did not occur by volume, additional biopsies would be ineffective for minimizing this bias.
Results: Prostate cancer detected using a 6-core technique revealed 179 of 679 cancers (26.4%) vs 240 of 679 (35.4%) using the extended core technique (p <0.001). The marginal cancer detection rate increased significantly as prostate volume increased. Cancer detection rates for the 1st, 2nd and 3rd tertiles of prostate volume were increased by 16 of 227, 17 of 226 and 28 of 226 cases, respectively (p = 0.05). With respect to Gleason score, upgrading from Gleason 6 to 7 was observed in 14 patients (7.9%) due to the additional procured cores (p <0.001). However there was no association among the various prostate volumes (p = 0.87).
Conclusions: Although more high grade cancers are detected with extended pattern biopsy, there is no differential upgrading with respect to prostate volume. Based on these observations extended prostate sampling in trials of agents that decrease prostate volume would have minimal impact on volume-grade associations.