To compare Gleason scores (GS) originally assigned in the mid 1990s with the current pathologic evaluation of the same prostatectomy slides, and to assess the GS migration effect on outcome in patients undergoing surgical treatment of prostate cancer.

We reviewed medical charts of consecutive patients who underwent a radical prostatectomy for T2-T3 prostate cancer at our Medical Center between 1995 and 1997. Prostate specimen slides of 204 patients were reviewed and GS was reassigned in a blinded fashion by a single uropathologist in 2008. GS distributions were compared, and original and re-evaluated GS were assessed for predictive ability in survival regression models.

GS distribution differed significantly between the mid 1990s and the current evaluation (P < .001), with the average reevaluated GS higher than the initial one (6.14 vs 6.39, P < .001). The GS was upgraded in 63 cases (30.9%) and downgraded in 25 (12.3%) at reevaluation. The initial GS was predictive (P = .002) of prostate-specific antigen recurrence (PSAR), whereas the newly assigned GS was not (P = .393). However, grouping reassigned GS into risk groups (low < 7, moderate = 7 and high > 7) yielded a better PSAR definition. Survival curves of initial GS could not distinguish between moderate- and high-risk groups, although reassigned GS curves showed statistically significant differences between all risk groups.

These results suggest that interpretation of pathologists played a significant role in the GS shift and propose that the contemporary GS remains a useful prognostic factor of PSAR when stratified in risk categories, although the single GS value may not be as important.