Table 3.

Ki67 dichotomized at 5% cutoff: multivariable Cox proportional hazard models stratified by center for RFS.

Model	Factor	Comparison	Hazard Ratio	95% LCL	95% UCL	Overall p-value
RFS Model 1 (N = 706, #Events = 319)	Ki67% weighted average	>=5% vs. <5%	1.471	1.110	1.950	0.0007
	Margin	Pos vs. Neg	1.598	1.248	2.047	0.0002
	Seminal vesicle invasion	Yes vs. No	2.072	1.430	3.002	0.0001
	Gleason score	3+4 vs. 6	1.297	0.972	1.732	0.004
		4+3 vs. 6	1.808	1.273	2.568
		8–10 vs. 6	1.626	1.126	2.348
	Log(PSA)	1 unit increase	1.536	1.288	1.831	<.0001
RFS Model 2 (N = 706, #Events = 319)	Ki67% maximum	>=5% vs. <5%	1.314	1.033	1.672	0.03
	Margin	Pos vs. Neg	1.596	1.246	2.043	0.0002
	Seminal vesicle invasion	Yes vs. No	2.119	1.462	3.070	<.0001
	Gleason score	3+4 vs. 6	1.317	0.986	1.757	0.0002
		4+3 vs. 6	1.876	1.324	2.658
		8–10 vs. 6	1.670	1.159	2.406
	Log(PSA)	1 unit increase	1.505	1.264	1.792	<.0001

Recurrence free survival (RFS) event is defined as any recurrence, metastasis, or prostate cancer death. Models 1 and 2 are separate models using Ki67 weighted average % positive (PI=2.19) and maximum % positive (PI=3.11), respectively. Only significant factors were included in the final models.