Purpose: There is some evidence that Gleason scores (GS) have shifted over time, although documentation of the extent to which this has occurred and its clinical significance is sparse.
Methods and materials: Three types of analyses were performed to characterize the GS shift. First, the time-related changes in GS, T-stage category, and PSA for all 983 patients treated with conformal radiotherapy at our institution were examined between 1992 and 1997. Then a matched-pair analysis of patients treated without androgen deprivation was implemented. Patients were matched for grouped GS, grouped PSA, T category, and grouped radiation dose; 242 patients (Cohort 1) treated between 1992 and 1994 were compared to 242 patients treated between 1995 and 1997 (Cohort 2). Follow-up for each patient in Cohort 1 was truncated to match the length of follow-up for his counterpart in Cohort 2 (median follow-up 42 vs. 47 months). The ASTRO consensus definition of a rising PSA was used. Finally, the pathology slides of 106 patients treated with radiotherapy at another institution between 1987 and 1993 underwent a blind second review by one of the study pathologists (P.T.) in 1998.
Results: The percentages of prostate cancer patients treated with radiotherapy at our institution from 1992 to 1994 vs. 1995 to 1997 were GS 2-5: 49% vs. 16%, GS 6: 27% vs. 59%, GS 7: 20% vs. 19%, and GS > or =8: 8.3% vs. 6%. There was no clear difference in T category over time with 92-96% having T1-T2 disease. The percentage of patients with a PSA < or = 10 ng/mL rose rapidly from 35% in 1992 to 63% in 1994. In corroboration of these findings, the pathologic review of specimens from 106 patients at another institution yielded changes between the initial reading to the recent reading of 50% to 6% for GS 2-5, 22% to 31% for GS 6, 16% to 25% for GS 7, and 12% to 25% for GS 8-10 (p < 0.0001). From the matched-pair analysis, 5-year Kaplan-Meier bNED rates were 68% and 82% (p = 0.03) favoring Cohort 2.
Conclusions: During the study period, pretreatment PSA values declined and T category remained stable; yet, GS increased. Upgrading of GS would be expected to lead to an improvement in outcome for all GS groups because of the Will Rogers effect, which explains in part the results of the matched-pair analysis. The GS shift confounds retrospective series spanning the 1990s.