Table 5: Predictive value of the prognostic stratification schemes in Table 4 using TROG 96.01 trial data with 5-year prostate cancer-specific mortality (± one standard error) as an endpoint.

AuthorVariableCutpoints, mortality, number of patientsCutpoints, Mortality, Number of patientsc-index*

PSADT≥6 months<6 months0.675
Okotie et al. [8] %
(58%) (42%)

D’Amico et al. [6]PSADT≥12 months<12 months0.598
Tollefson et al. [14] %
(28%) (72%)

GS≤7>7
TTBF>3 years≤3 years >3 years≤3 years 0.694
Freedland et al. [10]PSADT≥9 months<9 months ≥9 months <9 months ≥9 months<9 months ≥9 months <9 months
0%0% % %
(29%) (12%) (8%) (31%) (3%) (1%)  (1%) (14%)

GS≤7>70.654
Zhou et al. [13]PSADT≥3 months<3 months ≥3 months<3 months
(70%) (10%) (16%) (4%)

Stephenson et al. [9]GS≤7>70.647
PSADT>10 months≤10 months >10 months≤10 months
(32%) (48%) (4%) (16%)

Buyyounouski et al. [15]GS≤6>60.682
TTBF≥1.5 years <1.5 years ≥1.5 years <1.5 years
(29%) (7%) (44%) (20%)

: number of TROG 96.01 subjects in study stratum; ( %): percentage of total number of TROG 96.01 subjects with biochemical failure ( ) in the study; GS: Gleason score; PSADT: PSA doubling time; TTBF: time to biochemical failure.
*Harrell’s concordance index (higher c-index indicates that the model has a better predictive power).
Due to subject numbers, subsets with PSADT <3 months and >15 months are not presented in the Freedland study.