Table 3.

Multivariable analysis for SNPs and prostate cancer outcomes

SNP	No. of patients	Biochemical recurrence			Clinical metastases			Prostate cancer–specific death
		n = 351			n = 146			n = 91
		HR	95% CI^*	P	HR	95% CI	P	HR	95% CI	P
rs10486567	635	1.29	1.00-1.65	0.046	1.42	0.96-2.10	0.083	1.1	0.66-1.82	0.7
rs4962416	628	1.08	0.84-1.39	0.5	0.87	0.58-1.29	0.5	0.85	0.50-1.45	0.6
rs7920517	631	1.53	1.17-2.01	0.002	1.11	0.73-1.68	0.6	1.06	0.61-1.84	0.8
rs7931342	632	1.03	0.80-1.32	0.8	2.01	1.35-3.00	0.0006	1.81	1.07-3.05	0.027
rs10896449	649	1.01	0.79-1.28	1	1.88	1.28-2.77	0.001	1.83	1.10-3.04	0.021
rs2735839	643	0.88	0.67-1.17	0.4	1.26	0.84-1.87	0.3	2.22	1.35-3.66	0.002
rs7008482	617	0.85	0.66-1.09	0.19	0.87	0.59-1.28	0.5	1	0.60-1.65	1

NOTE: Each SNP is individually assessed in separate multivariable models, controlling for age at prostate cancer diagnosis, PSA at diagnosis, clinical stage, and biopsy Gleason grade. Shown are SNPs with univariate associations at P < 0.05 by the 2 df test.

↵*The first P value shown for each SNP was determined by a 2 df test for a global difference across genotypes; the second P value shown was determined by a log-additive (linear trend) model.