2971.Full-Text.Pdf

Published OnlineFirst March 28, 2014; DOI: 10.1158/1078-0432.CCR-13-2567

Clinical Cancer Imaging, Diagnosis, Prognosis Research

Steroidogenic Germline Polymorphism Predictors of Prostate Cancer Progression in the Estradiol Pathway

Eric Levesque 1,2, Isabelle Laverdiere 1, Etienne Audet-Walsh1, Patrick Caron1,Melanie Rouleau1, Yves Fradet2, Louis Lacombe2, and Chantal Guillemette1

Abstract Purpose: Reliable biomarkers that predict prostate cancer outcomes are urgently needed to improve and personalize treatment approaches. With this goal in mind, we individually and collectively appraised common genetic polymorphisms related to estradiol metabolic pathways to find prostate cancer prognostic markers. Methods: The genetic profiles of 526 men with organ-confined prostate cancer were examined to find common genetic polymorphisms related to estradiol metabolic pathways and these findings were replicated in a cohort of 213 men with more advanced disease (follow-up time for both cohorts, >7.4 years). Specifically, we examined 71 single-nucleotide polymorphisms (SNP) in SULT2A1, SULT2B1, CYP1B1, COMT, CYP3A4, CYP3A5, CYP3A43, NQO1, and NQO2 and assessed the impact of the SNPs alone and in combination on prostate cancer progression and on circulating hormone levels. Results: According to a multivariate analysis, CYP1B1 (rs1800440), COMT (rs16982844), and SULT2B1 (rs12460535, rs2665582, rs10426628) were significantly associated with prostate cancer progression and hormone levels. Remarkably, by combining the SNP information with previously identified HSD17B2 markers, the patients could be stratified into four distinct prognostic subgroups. The most prominent association was observed for the eight-marker combination [CYP1B1 (rs1800440), SULT2B1 (rs12460535, rs2665582, and rs10426628), and HSD17B2 (rs4243229, rs1364287, rs2955162, and rs1119933)]. Conclusion: This study identified specific germline variations in estradiol metabolism–related pathways, namely CYP1B1, SULT2B1,andHSD17B2, as novel prognostic markers that are cumulatively associated with increased risk of prostate cancer progression. This panel of markers warrants additional investigation and validation to help stratify patients according to their risk of progression. Clin Cancer Res; 20(11); 2971–83. 2014 AACR.

Introduction Simple, noninvasive, and reliable molecular markers are Prostate cancer is the most common cancer in men and necessary to identify men at high risk for prostate cancer the second leading cause of cancer death in North America progression. Such prognostic markers for use in localized, (1). A clinically relevant, prognostic molecular signature locally advanced, and metastatic settings would consider- defining aggressive and indolent prostate cancer is eagerly ably improve prostate cancer management. awaited to personalize treatment approaches. Indeed, pros- Recently, circulating tumor cells and whole-blood gene- tate cancer is a complex genetic disease, and its clinical expression signatures were elegantly shown to be potential heterogeneity underscores the need to identify biomarkers biomarkers of cancer progression, and use of these markers for prognostication purposes at all stages of prostate cancer. enabled identification of several differentially regulated genes in immune or androgen signaling that may be useful for predicting outcomes for patients with castration-resis- Authors' Affiliations: 1Pharmacogenomics Laboratory, Centre Hospitalier tant prostate cancer (2, 3). Another useful and much sim- Universitaire de Quebec (CHU de Quebec) Research Center and Faculty of pler approach to predict prostate cancer outcomes relies on Pharmacy; and 2CHU de Quebec Research Center and Faculty of Medicine, Laval University, Quebec, Canada identifying key inherited genetic variations involved in disease progression (4, 5). In fact, potential molecular Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). determinants of clinical outcomes have been identified as components encoded by immune system genes (6, 7), sex Corresponding Author: Chantal Guillemette, CHU de Quebec Research Center, Laval University, 2705 Boul Laurier, R4720 Quebec G1V 4G2, steroid–related genes (4, 8), and, with conflicting results, Canada. Phone: 418-654-2296; Fax: 418-654-2761; E-mail: prostate cancer risk alleles (9–12). The importance of [email protected] studying host steroidogenic pathways comes from observa- doi: 10.1158/1078-0432.CCR-13-2567 tions that prostate cells possess the enzymatic machinery for 2014 American Association for Cancer Research. intracrine conversion of precursors into more potent

www.aacrjournals.org 2971

Levesque et al.

to 90% to 100% (20). Moreover, knockout of CYP19A1 Translational Relevance (encoding aromatase) in mice, which prevents estrogen Although there has been a decline in mortality rate due production, elevated the circulating T levels of the mice and to early detection and better therapies, our ability to caused enlargement of their prostates, but this knockout did predict prostate cancer progression and metastatic not cause the mice to develop prostate cancer (21, 22). These behavior of a patient’s cancer is still very limited. This observations are consistent with the conclusion that both variability in clinical outcome underscores the need to androgen and estrogen are required for the development of identify physiopathologic changes, in tumor cells and its prostate cancer. Furthermore, Cussenot and colleagues dem- microenvironment, associated with disease progression. onstrated the role of key estrogen metabolism–related genes, Despite the fact that estrogens have been overshadowed e.g., CYP1B1 and catechol-O-methyltransferase (COMT),ina by androgens in the last decades, several pieces of evi- large and homogeneous French cohort with prostate cancer dence suggested that estrogens cooperate with andro- carcinogenesis (23). Given these important findings sup- gens to perpetuate carcinogenesis. We describe host porting the contribution of estrogen to prostate cancer, we genetic variations in the estradiol pathway associated sought to comprehensively determine if single-nucleotide with hormone levels and lethal form of prostate polymorphisms (SNP) in genes associated with estradiol- cancer. Single-nucleotide polymorphisms distributed related metabolic pathways, either individually or in com- across CYP1B1, HSD17B2, catechol-O-methyltransfer- bination, would be predictors of prostate cancer progression. ase (COMT), and SULT2B1 were associated with pro- We established germline genetic signature in these pathways gression and survival. Indeed, based on our findings, associated with cancer progression in localized prostate patients could be stratified into four distinct prognostic cancers and then verified these findings in locally advanced subgroups. Thus, the cumulative impact of such markers prostate cancers to evaluate the influence of the signature on provides clinically relevant information from a single overall survival. blood sample, is invariant with time, overcomes tumor heterogeneity, and identifies potential targets to opti- Patients and Methods mize disease management. Our data offer a high trans- Clinical data lational potential that could lead to a personalize The study included 739 Caucasian patients with prostate approach based on the presence or absence of the cancer divided into two independent cohorts. The first molecular signature and identify novel promising drug- cohort was composed of 526 patients with localized pros- gable targets. tate cancer, and the second was composed of 213 patients with locally advanced disease. Patients had undergone surgical radical prostatectomies at l’Hotel-Dieu^ de Quebec Hospital (Quebec City, Canada) before being enrolled in hormones (13, 14) and that these pathways are dysregu- the study. The localized and advanced prostate cancer lated in prostate cancer (15). Moreover, the consequence of cohorts were recruited between 1999 and 2002, and sustained and active steroidogenesis in cancer cells is 1982 and 2002, respectively (5). The median follow-up highlighted by the observation that castration-resistant times were 7.4 and 7.8 years for the localized and locally prostate cancer can still be driven by sex steroid hormones advanced populations, respectively. Before surgery, each despite a low circulating testosterone (T) level (14). Several participant provided written consent for genetic analysis. germline polymorphisms in sex steroid metabolic pathways The local research ethics committee approved the research have been shown to be associated with localized and protocol. advanced prostate cancer outcomes related to a meaningful influence of inherited genetic variations in the androgenic Genetic analysis pathway on progression (4, 8). Therefore, in addition to DNA was extracted from peripheral blood mononuclear tumor markers, circulating tumor cells, and whole-blood cells obtained at time of diagnosis. For genotyping, PCR gene-expression profiles (2, 3, 16), host polymorphisms amplification was performed on germline DNA, and the might assist in predicting tumor behavior and serve as a products were sequenced by Sequenom iPLEX matrix- simple and reliable assay easily analyzed from germline assisted laser desorption/ionization–time-of-flight mass DNA of any human. spectrometry. The SNPs were chosen so that most of the Because we know that steroidogenic pathways are impor- common haplotype diversity in Caucasians would be tant to prostate cancer progression, we focused on identify- included. A region covering the exons, introns, and 5 kb ing genetic polymorphisms related to sex steroid metabolic of the 50 and 30 untranslated regions immediately adjacent genes associated with endogenous hormone levels and pros- to each gene was screened using a haplotype-tagging SNP tate cancer progression with an emphasis on the biotrans- (htSNP) strategy to maximize coverage (5). formation pathways involving estradiol and its metabolites. Seventy-one SNPs were genotyped in localized prostate Previous important studies have suggested that estrogen cancer and sixteen SNPs were tested in the locally affects prostate carcinogenesis (17–19). In rodents, when advanced cases. For both populations, the average geno- even a short course of estradiol is combined with T treatment, type call rate for all SNPs was 98%. SNPs with a missing the incidence of prostate cancer increases from 35% to 40% call rate 5% were excluded. Negative controls were

2972 Clin Cancer Res; 20(11) June 1, 2014 Clinical Cancer Research

Estrogen-Related Genes and Prostate Cancer Prognostic Stratiﬁcation

included in each analysis, and quality controls included Kaplan–Meier analyses, log-rank tests, and univariate Cox 5% blind duplicates. regression analyses were also carried out for each variant. P values were considered significant if <0.05. False discovery Measurement of endogenous steroid levels rates (q values) were calculated to determine the degree to Plasma samples were collected from the localized cohort which the tests were prone to false positives, using the R on the morning of the surgical procedure (4). We excluded QVALUE package (http://genomics.princeton.edu/storey- patients who had received neoadjuvant hormone treatment lab/qvalue/). and those with missing genotype information for the stud- To adjust for differences in the absolute levels of sex ied SNPs. Steroids were measured by validated gas chro- steroids, we calculated the residuals of the natural logarithm matography–mass spectrometry or liquid chromatogra- of each hormone level regressed on age at blood donation phy–tandem mass spectrometry (4). Deuterated steroids and smoking status. The associations with SNPs were per- were added into each sample, and quality controls were formed by regression of the hormone residuals for each SNP included in each run. The lower quantification limits were using two categories: recessive and dominant with one T (0.03 ng/mL), dihydrotestosterone (DHT; 0.005 ng/mL), degree of freedom. We considered the association of an androsterone (ADT; 0.025 ng/mL), ADT-glucuronide SNP with the variation in hormone levels to be significant if (ADT-G; 1 ng/mL), 3a-diol-3G and 3a-diol-17G (0.25 the P value was <0.05. To facilitate the comparison between ng/mL), dehydroepiandrosterone (DHEA; 0.1 ng/mL), groups, we displayed the hormone level as untransformed DHEA-sulfate (DHEA-S; 0.075 mg/mL), E1-S (0.075 ng/ data using the geometric mean and the SEM. Statistical mL), estrone (E1; 0.005 ng/mL), estradiol (E2; 0.001 ng/ analyses were performed using SAS Statistical Software mL), androstenedione (4-dione; 0.05 ng/mL), androstene- version 9.2 (SAS Institute) and PASW statistics version 17 diol (5-diol; 0.05 ng/mL). Coefficients of variation for the (SPSS Inc.). intra- and interassays were 10.0%, and accuracy values (given in parentheses) were T (100.5%), DHT (103.5%), Results ADT (96.5%), ADT-G (99.5%), 3a-diol-3G (89.2%), 3a- Experimental approach diol-17G (100.2%), DHEA (96.13%), DHEA-S (99.9%), Seventy-one htSNPs distributed in the selected genes E1-S (100.45%), E1 (106.9%), E2 (103.15%), 4-dione encoding proteins involved in estrogen metabolism, name- (95.5%), and 5-diol (95.33%; ref. 4). ly genes encoding cytochrome P450s (CYP1B1, CYP3A4, CYP3A5, and CYP3A43), COMT, quinone reductases Statistical analyses (NQO1 and NQO2), and sulfotransferases (SULT2A1 and To assess the association of the SNPs with time to bio- SULT2B1), were studied in 526 Caucasians with localized chemical recurrence (BCR), disease progression, and death, prostate cancer in relation to time to BCR. We subsequently each htSNP was first categorized with the common genomic studied 16 markers in the advanced cohort (213 cases) to model because the function of most of these htSNPs was further evaluate their potential relationships to prostate unknown. Minor-allele homozygotes with a frequency of cancer progression and survival. There was no evidence for <2% were grouped with the heterozygotes. A sample size of deviation from Hardy–Weinberg equilibrium for the stud- 526 patients provided a power of at least 80% to detect a HR ied SNPs (n ¼ 71). of 1.75 for genetic variant with a frequency of 25%. Cox regression was performed for the frequency of occurrence of Genetic analyses of localized prostate cancer each SNP, with adjustments made for clinicopathologic The relative frequencies of the SNPs in patients with variables. After performing a Cox regression using the cancer and their corresponding HRs (95% confidence inter- genomic model, the associations of SNPs with clinical vals, CI) are shown in Table 2 and Supplementary Tables outcomes were also evaluated for genomic modes of trans- S1–S5. After making adjustments for known clinicopatho- mission (dominant and recessive). In dominant model, for logic variables, ten SNPs, three in COMT, two in SULT2B1, a SNP with a major allele "A" and a minor allele "b," the one in CYP1B1 and NQO1, and three in NQO2 were found collective genotypes ("Ab" þ "bb") are compared with to be significantly associated with time to BCR. Of note, a reference genotype "AA." For recessive model, "bb" is com- significant association was observed for the CYP1B1 variant pared with collective ("AA" þ "Ab") reference group. HR of rs1800440 involved in the initial carcinogenic step related reference genotype group is arbitrary fixed at 1.00. All to 4-hydroxy catechol estrogen (CE) biosynthesis (HR, 1.64; covariables were categorized as shown in Table 1, and 95% CI, 1.10–2.46; P ¼ 0.016; q ¼ 0.12). The three SNPs in 4% of the covariables were missing. For localized prostate the COMT variants (rs11705619; rs165849; rs9332377), cancer cases, the censoring variable was time to BCR, based encoding COMT involved in CE inactivation in the prostate, on a prostate-specific antigen (PSA) cut-off value 0.3 mg/L were associated with time to BCR. One additional COMT (4). For locally advanced cases, disease progression was variant (rs16982844) was associated with a trend toward a defined as resistance to androgen-deprivation therapy, greater risk of progression (HR, 1.73; 95% CI, 0.98–3.06; P metastasis, and/or death (5). Multivariate models included ¼ 0.06). The SULT2B1 variants rs1246053 and rs2665582 PSA at diagnosis, Gleason score, pathologic tumor stage, age were associated with BCR (HR, 1.99; 95% CI, 1.13–3.49; P at diagnosis, neoadjuvant therapy, smoking status, adjuvant ¼ 0.017; q ¼ 0.58 and HR, 0.50; 95% CI, 0.26–0.96; P ¼ therapy, surgical margins, and nodal invasion status. 0.037; q ¼ 0.17, respectively). NQO1 rs2917670 was

www.aacrjournals.org Clin Cancer Res; 20(11) June 1, 2014 2973

Levesque et al.

Table 1. Clinical and pathologic characteristics of the patients with prostate cancer

Localized prostate Locally advanced prostate Characteristics cancer (n ¼ 526) cancer (n ¼ 213) Age at diagnosis (y) Mean 63.3 63.8 SD 6.8 6.2 Range 43.5–80.7 47.0–81.7 Median follow-up (months)a 88.8 93.4 Biochemical recurrence (%)a 130 (24.7) 132 (63)b PSA at diagnosis (ng/mL)a 10 362 (69) 81 (38) >10–20 103 (20) 66 (31) >20 56 (11) 65 (31) Pathologic Gleason scorea 6 158 (31) 32 (15) 7 244 (48) 75 (36) 8 107 (21) 102 (49) Pathologic tumor stagea pT ¼ T2 313 (60) 28 (13) pT ¼ T3a 131 (25) 59 (28) pT T3b 77 (15) 124 (59) Nodal invasiona N0 481 (92) 0 (0) Nþ 44 (8) 213 (100) Neoadjuvant hormone therapya Yes 31 (6) 54 (25) No 495 (94) 159 (75) Adjuvant hormone therapya Yes 30 (6) 94 (44) No 496 (94) 118 (56) Margin statusa Negative 368 (70) 69 (32) Positive 154 (30) 144 (68)

Abbreviations: N, node; PSA, prostate-speciﬁc antigen. aP < 0.001 for the two cohorts. bIn the locally advanced prostate cancer cohort, BCR data were missing for 4 patients.

associated with a lower risk of progression (HR, 0.53; 95% respectively). The COMT rs16982844 was signiﬁcantly asso- CI, 0.29–0.97; P ¼ 0.039; q ¼ 0.58), whereas NQO2 variants ciated with an increased risk of progressive disease (HR, were associated with an increased risk of BCR (Table 2 and 3.40; 95% CI, 1.63–7.10; P ¼ 0.001; q ¼ 0.004) and death Supplementary Table S5). Indeed, NQO2 rs10223369, (HR, 3.86; 95% CI, 1.61–9.29; P ¼ 0.003; q ¼ 0.027). rs1143684, and rs6920900 were associated with a higher Moreover, SULT2B1 rs12460535, rs2665582, and risk of progression (HR, 1.80; 95% CI, 1.24–2.60; P ¼ rs10426628 were signiﬁcantly associated with certain 0.002; q ¼ 0.06; HR, 1.51; 95% CI, 1.04–2.19; P ¼ 0.03; patient outcomes as follows: SULT2B1 rs10426628 and q ¼ 0.17; and HR, 1.89; 95% CI, 1.19–3.00; P ¼ 0.007; q ¼ rs2665582 were associated with a lower risk of progression 0.10, respectively). (HR, 0.49; 95% CI, 0.26–0.91; P ¼ 0.025; q ¼ 0.056 and HR, 0.38; 95% CI, 0.13–1.09; P ¼ 0.07; q ¼ 0.14, Genetic analyses in locally advanced cases respectively; Table 3), whereas SULT2B1 rs12460535 was Four of the 16 SNPs were also found in the genes of associated with progression (HR, 2.44; 95% CI, 1.10–5.41; patients with advanced prostate cancer (Tables 3 and 4 and P ¼ 0.028; q ¼ 0.43). Supplementary Tables S6 and S7). CYP1B1 rs1800440 was associated with progression-free survival and mortality Cumulative association of adverse genotypes (HR, 2.58; 95% CI, 1.47–4.52; P ¼ 0.0009; q ¼ 0.004 and We postulated that the association of the markers with HR, 3.25; 95% CI, 1.58–6.70; P ¼ 0.001; q ¼ 0.018, prostate cancer progression might be stronger if the SNPs

2974 Clin Cancer Res; 20(11) June 1, 2014 Clinical Cancer Research

Table 2. SNPs associated with time to BCR in 526 patients with localized prostate cancer

Cox regression analysesc d

Heterozygous Homozygous Secondary model Published OnlineFirstMarch28,2014;DOI:10.1158/1078-0432.CCR-13-2567 clincancerres.aacrjournals.org

SNP Polymorphisma BCRb No BCRb HR (95% CI) P HR (95% CI) P Model HR (95% CI) P COMT rs11705619 T > C 8/34/88 22/144/225 0.60 (0.39–0.92) 0.020 1.07 (0.51–2.23) 0.86 Dom 0.66 (0.45–0.98) 0.041 rs165849 A > G 15/60/55 34/165/195 1.31 (0.88–1.94) 0.19 2.01 (1.11–3.62) 0.021 Rec 1.74 (1.01–3.01) 0.047 rs16982844 C > A 0/16/114 0/36/358 Dom 1.73 (0.98–3.06) 0.06 rs4646316 C > T 6/44/79 21/148/223 0.74 (0.50–1.11) 0.15 0.51 (0.20–1.29) 0.16 Dom 0.71 (0.48–1.05) 0.08 rs9332377 C > T 6/40/84 10/103/280 Dom 1.60 (1.09–2.36) 0.017 CYP1B1 rs1800440 A > G 4/38/88 13/94/287 Dom 1.64 (1.10–2.46) 0.016 NQO1 on September 27, 2021. © 2014American Association for Cancer rs2917670 G > A 13/65/52 75/179/139 1.21 (0.81–1.79) 0.35 0.58 (0.30–1.12) 0.10 Rec 0.53 (0.29–0.97) 0.039 Research. > –

rs689453 G A 2/8/118 2/63/328 Dom 0.57 (0.29 1.10) 0.09 Strati Prognostic Cancer Prostate and Genes Estrogen-Related NQO2 rs10223369 C > T 15/59/56 21/159/214 1.78 (1.20–2.64) 0.004 1.87 (1.02–3.45) 0.044 Dom 1.80 (1.24–2.60) 0.002 rs1143684 T > C 9/46/73 16/115/262 1.49 (1.00–2.20) 0.049 1.64 (0.78–3.46) 0.19 Dom 1.51 (1.04–2.19) 0.030 rs2070999 G > A 19/54/55 61/196/136 0.70 (0.47–1.05) 0.08 0.67 (0.39–1.17) 0.16 Dom 0.69 (0.47–1.01) 0.06 rs6920900 G > C 34/72/24 85/201/108 1.92 (1.19–3.11) 0.008 1.82 (1.04–3.18) 0.037 Dom 1.89 (1.19–3.00) 0.007 rs9405188 A > C 9/40/79 18/148/227 0.77 (0.51–1.15) 0.20 1.88 (0.89–3.96) 0.10 Rec 2.06 (0.99–4.29) 0.053 SULT2B1 rs10426628 G > A 4/35/91 13/115/265 0.72 (0.48–1.10) 0.13 0.82 (0.26–2.60) 0.73 Dom 0.73 (0.49–1.10) 0.13 rs12460535 G > A 18/48/63 27/162/203 1.00 (0.68–1.49) 0.99 1.99 (1.10–3.59) 0.022 Rec 1.99 (1.13–3.49) 0.017 rs12611137 C > T 4/56/70 21/128/244 1.40 (0.97–2.04) 0.08 0.97 (0.30–3.14) 0.96 Dom 1.37 (0.95–1.98) 0.09

lnCne e;2(1 ue1 2014 1, June 20(11) Res; Cancer Clin rs2665582 G > A 0/14/114 3/46/344 Dom 0.50 (0.26–0.96) 0.037 rs3848542 C > T 4/51/70 42/162/181 0.76 (0.51–1.12) 0.17 0.33 (0.12–0.92) 0.034 Rec 0.37 (0.14–1.03) 0.06

NOTE: P value 0.05 are in bold characters. Abbreviations: Dom, dominant; Rec, recessive. aThe major allele is to the left of the > sign and the minor allele is to the right. bThe numbers of minor-allele homozygotes, heterozygotes, and major-allele homozygotes are presented left to right. cCox regression models included PSA at diagnosis, Gleason score, pathologic T stage, age at diagnosis, neoadjuvant hormone therapy, smoking status, adjuvant therapy, surgical margin, and nodal invasion status. The major-allele homozygotes served as the reference group with a ﬁxed HR of 1.00. When minor-allele homozygotes were rarely found (frequency 2%), they were combined with the heterozygotes; therefore, only the dominant/recessive model is shown (genomic model is empty). d

Minor-allele homozygotes. ﬁ cation 2975 2976 lnCne e;2(1 ue1 2014 1, June 20(11) Res; Cancer Clin L Downloaded from vsu tal. et evesque

Table 3. Association of SNPs with progression-free survival in 213 prostate cancer patients with locally advanced disease

c Published OnlineFirstMarch28,2014;DOI:10.1158/1078-0432.CCR-13-2567

clincancerres.aacrjournals.org Cox regression analyses Heterozygous Homozygousd Secondary model

SNP Polymorphisma Progb No Progb HR (95% CI) P HR (95% CI) P Model HR (95% CI) P COMT rs11705619 T > C 1/20/34 9/45/94 Dom 0.74 (0.39–1.40) 0.36 rs165849 A > G 3/28/25 23/56/73 1.17 (0.64–2.12) 0.61 0.57 (0.16–1.97) 0.37 Rec 0.51 (0.16–1.70) 0.28 rs16982844 C > A 0/11/45 1/15/137 Dom 3.40 (1.63–7.10) 0.001 rs9332377 C > T 2/17/37 5/41/107 Dom 1.31 (0.70–2.45) 0.40 CYP1B1 rs1800440 A > G 3/26/28 4/37/112 Dom 2.58 (1.47–4.52) 0.0009 NQO1 on September 27, 2021. © 2014American Association for Cancer > – – –

Research. rs2917670 G A 14/19/23 24/70/59 0.73 (0.37 1.43) 0.36 1.25 (0.61 2.58) 0.54 Rec 1.45 (0.75 2.81) 0.26 rs689453 G > A 0/5/46 0/13/135 Dom 1.18 (0.45–3.09) 0.74 NQO2 rs10223369 C > T 6/22/26 11/65/75 1.26 (0.68–2.34) 0.47 1.30 (0.47–3.59) 0.61 Dom 1.27 (0.70–2.28) 0.43 rs1143684 T > C 4/20/32 6/50/96 1.39 (0.76–2.55) 0.29 1.68 (0.48–5.85) 0.42 Dom 1.43 (0.80–2.55) 0.23 rs2070999 G > A 6/26/24 26/75/51 0.86 (0.46–1.59) 0.63 0.69 (0.27–1.77) 0.44 Dom 0.82 (0.45–1.48) 0.51 rs6920900 G > C 17/27/11 31/82/36 0.97 (0.46–2.08) 0.94 1.73 (0.73–4.10) 0.21 Rec 1.77 (0.92–3.41) 0.09 SULT2B1 rs10426628 G > A 0/15/41 6/62/85 Dom 0.49 (0.26–0.91) 0.025 rs12460535 G > A 9/23/25 14/75/64 0.71 (0.38–1.30) 0.26 2.05 (0.88–4.76) 0.10 Rec 2.44 (1.10–5.41) 0.028 rs12611137 C > T 1/17/37 1/46/105 Dom 1.35 (0.73–2.52) 0.34 rs2665582 G > A 0/4/52 0/26/123 Dom 0.38 (0.13–1.09) 0.07 rs3848542 C > T 5/22/24 10/53/76 1.25 (0.67–2.32) 0.49 1.65 (0.60–4.51) 0.33 Rec 1.48 (0.57–3.85) 0.42

NOTE: P value 0.05 are in bold characters. Abbreviations: Prog, progression; Dom, dominant; Rec, recessive.

lnclCne Research Cancer Clinical aThe major allele is to the left of the > sign and the minor allele is to the right. bThe numbers of minor-allele homozygotes, heterozygotes, and major-allele homozygotes are presented left to right. cCox regression models included PSA at diagnosis, Gleason score, pathologic T stage, age at diagnosis, neoadjuvant hormone therapy, smoking status, adjuvant therapy, surgical margin, and nodal invasion status. The major-allele homozygotes served as the reference group with a ﬁxed HR of 1.00. When minor-allele homozygotes were rarely found (frequency 2%), they were combined with the heterozygotes; therefore, only the dominant/recessive model is shown (genomic model is empty). dMinor-allele homozygotes. www.aacrjournals.org Downloaded from

Table 4. Association of SNPs with overall survival in 213 prostate cancer patients with locally advanced disease

Cox regression analysesc Heterozygous Homozygousd Secondary model Published OnlineFirstMarch28,2014;DOI:10.1158/1078-0432.CCR-13-2567 clincancerres.aacrjournals.org SNP Polymorphisma Deadb Aliveb HR (95% CI) P HR (95% CI) P Model HR (95% CI) P COMT rs11705619 T > C 1/15/21 9/50/108 0.74 (0.32–1.71) 0.48 0.75 (0.09–5.99) 0.79 Dom 0.74 (0.33–1.68) 0.47 rs165849 A > G 1/18/18 25/67/80 1.40 (0.65–3.01) 0.39 0.32 (0.04–2.63) 0.29 Rec 0.26 (0.03–2.00) 0.20 rs16982844 C > A 0/8/30 1/18/153 Dom 3.86 (1.61–9.29) 0.003 rs9332377 C > T 1/9/27 6/50/117 1.05 (0.43–2.60) 0.91 2.15 (0.27–17.17) 0.47 Dom 1.12 (0.47–2.67) 0.79 CYP1B1 rs1800440 A > G 2/20/16 5/44/124 Dom 3.25 (1.58–6.70) 0.001 NQO1 rs2917670 G > A 7/14/17 31/76/65 0.66 (0.29–1.52) 0.33 0.80 (0.30–2.17) 0.66 Rec 1.00 (0.40–2.45) 0.99

Research. NQO2 rs10223369 C > T 3/16/16 14/71/84 1.60 (0.74–3.43) 0.23 0.61 (0.13–2.90) 0.54 Rec 0.50 (0.11–2.26) 0.37 Strati Prognostic Cancer Prostate and Genes Estrogen-Related rs1143684 T > C 2/13/20 8/56/108 1.44 (0.68–3.04) 0.34 0.63 (0.08–5.05) 0.66 Dom 1.33 (0.64–2.76) 0.45 rs2070999 G > A 2/20/13 29/82/61 1.29 (0.59–2.84) 0.52 0.50 (0.11–2.32) 0.38 Rec 0.43 (0.10–1.82) 0.25 rs6920900 G > C 11/16/8 37/93/38 0.62 (0.24–1.59) 0.32 1.15 (0.40–3.26) 0.80 Rec 1.62 (0.71–3.68) 0.25 SULT2B1 rs10426628 G > A 0/7/29 6/69/98 Dom 0.40 (0.17–0.91) 0.029 rs12460535 G > A 6/19/13 17/79/77 1.28 (0.58–2.79) 0.54 2.70 (0.90–8.14) 0.08 Rec 2.36 (0.87–6.41) 0.09 rs12611137 C > T 1/12/24 1/51/119 1.82 (0.82–4.01) 0.14 0.72 (0.08–6.62) 0.77 Dom 1.64 (0.76–3.56) 0.21 rs2665582 G > A 0/2/36 0/28/140 Dom 0.36 (0.08–1.56) 0.17 rs3848542 C > T 5/14/16 10/61/85 1.04 (0.47–2.34) 0.92 2.58 (0.86–7.74) 0.09 Rec 2.53 (0.91–7.01) 0.07

lnCne e;2(1 ue1 2014 1, June 20(11) Res; Cancer Clin NOTE: P value 0.05 are in bold characters. Abbreviations: Dom, dominant; Rec, recessive. aThe major allele is to the left of the > sign and the minor allele is to the right. bThe numbers of minor-allele homozygotes, heterozygotes, and major-allele homozygotes are presented left to right. cCox regression models included PSA at diagnosis, Gleason score, pathologic T stage, age at diagnosis, neoadjuvant hormone therapy, smoking status, adjuvant therapy, surgical margin, and nodal invasion status. The major-allele homozygotes served as the reference group with a ﬁxed HR of 1.00. When minor-allele homozygotes were rarely found (frequency 2%), they were combined with the heterozygotes; therefore, only the dominant/recessive model is shown (genomic model is empty). dMinor-allele homozygotes. ﬁ cation 2977 Published OnlineFirst March 28, 2014; DOI: 10.1158/1078-0432.CCR-13-2567

Levesque et al.

rs12460535, rs2665582, rs10426628) were assessed for signiﬁcant association with prostate cancer progression and overall patient survival (Fig. 1). Using these eight SNPs (of which SULT2B1 rs2665582 and rs10426628 are protective), patients were stratiﬁed into four prognostic subgroups (Fig. 1). Together, the combination of these markers had a cumulative association with PCa progression ranging from 87% (group 1) to 55% (group 4) in BCR-free survival in localized PCa (Fig. 1, panel A) and from 95% (group 1) to 9% (group 4) in PFS in locally advanced disease (Fig. 1, panel B). Indeed, the addition of each risk allele increased the risk of progression (localized PCa: HR, 1.73; 95% CI, 1.402.14; p ¼ 5107; locally advanced PCa: HR, 2.56; 95% CI, 1.783.69; p ¼ 4 107). Survival curves of the locally advanced cohort are showninFig.1,panelC. For localized disease, addition of the genetic markers to the baseline clinicopathologic model, used to predict BCR, modestly increased the area under the curve (AUC) values and the concordance index of each model. Indeed, a maximal increase of approximately 2% was found for the AUC (baseline model: 0.753; baseline model plus germline data: 0.778) and the concordance index (baseline model: 0.747; baseline model plus germline data: 0.767). For advanced disease, the maximal increase was approximately 3% for AUC (baseline model: 0.641; baseline model plus germline data: 0.671) and 4% for the concordance index (baseline model: 0.651; baseline model plus germline data: 0.694).

Relationship between prostate cancer outcomes and endogenous sex steroid hormones We found positive associations between germline markers linked to prostate cancer outcomes and circulating endogenous sex steroid levels, although such associations were not observed for estradiol, estrone, and estrone-sulfate, the only three estrogens measured in our study. COMT rs16982844 and CYP1B1 rs1800440 were associated, respectively, with higher levels of adrenal precursors (DHEA, DHEA-S, and 5-diol) and lower levels of inactive DHT metabolites (3a-DIOL-3 and -17 glucuronides; Table 5). Also, the protective SULT2B1 rs10426628 allele, asso-

Figure 1. Association of inherited variations in CYP1B1, SULT2B1,and ciated with a reduced risk of progression, was persistently HSD17B2 with PCa progression. Prognostic SNPs located in CYP1B1, associated with reduced circulating levels of many of the SULT2B1, and HSD17B2 were combined in relation to biochemical measured steroids and increased levels of inactive glucuro- recurrence (BCR) in (A) the Caucasian cohort with localized PCa (n ¼ 526), nide conjugates. A similar effect was observed for the (B) progression-free survival, and (C) overall survival in the locally protective SNPs in SULT2B1 rs12460535 and rs2665582 advanced cohort (n ¼ 213). Log-rank (LR) p values are shown in each frame. (Table 5).

associated with these interconnected estrogen biotransfor- Discussion mation pathways were assessed in combination. Because We report herein novel inherited molecular prognostic we had demonstrated a prominent role in prostate cancer markers for prostate cancer found in CYP1B1, SULT2B1, progression for common genetic variants in the HSD17B2- COMT, NQO1,andNQO2 associated with multiple estro- related conversion of estradiol or estrone, we also included gen-related pathways (Fig. 2). It is well established that variants of this gene in the combinatorial analyses. Thus, steroid hormones play a central role in prostate cancer the cumulative effects of the most prognostically pro- progression. Indeed, the improved outcomes for patients mising SNPs (CYP1B1 rs1800440, HSD17B2 rs1364287, on androgen deprivation, abiraterone acetate (24, 25) rs4243229, rs2955162, rs1119933, and SULT2B1 and MDV-3100 (26) therapies clearly demonstrate the

2978 Clin Cancer Res; 20(11) June 1, 2014 Clinical Cancer Research

Table 5. Association between positive markers in CYP1B1, COMT and SULT2B1, and endogenous steroid hormones levels. Published OnlineFirstMarch28,2014;DOI:10.1158/1078-0432.CCR-13-2567 clincancerres.aacrjournals.org Secondary model: dominantb Secondary model: recessiveb Pc Homozygous major allele 1 minor allele 1 minor allele Homozygous minor allele Hormone SNP Polymorphismsa Mean SEM Mean SEM Mean SEM Mean SEM Dom Rec COMT DHEA (ng/mL) rs16982844 C > A 1.57 0.06 2.03 0.20 0.009 DHEA-S (mg/mL) rs16982844 C > A 0.78 0.03 1.02 0.14 0.007 5-diol (pg/mL) rs16982844 C > A 508.08 14.45 615.87 46.65 0.033 CYP1B1 3a-diol3G (ng/mL) rs1800440 A > G 1.69 0.08 1.51 0.10 1.66 0.07 1.20 0.15 0.053 0.018 3a-diol17G (ng/mL) rs1800440 A > G 3.21 0.13 2.86 0.18 3.15 0.11 2.18 0.35 0.10 0.018

Research. 5-diol (pg/mL) rs10426628 G > A 536.85 17.23 478.02 23.14 518.90 14.19 470.23 62.73 0.028 0.29 srgnRltdGnsadPott acrPonsi Strati Prognostic Cancer Prostate and Genes Estrogen-Related Testo (ng/mL) rs10426628 G > A 3.80 0.08 3.47 0.12 3.70 0.07 3.19 0.34 0.010 0.09 DHT (pg/mL) rs10426628 G > A 314.58 8.77 294.28 11.56 310.43 7.14 234.94 35.79 0.07 0.011 ADT (pg/mL) rs12460535 G > A 176.62 5.95 178.44 10.25 179.85 6.17 152.98 14.13 0.93 0.029 rs10426628 G > A 180.30 5.38 168.28 14.08 177.68 5.89 136.76 27.47 0.15 0.022 rs2665582 G > A 179.08 6.27 153.71 14.20 176.35 5.82 133.56 92.63 0.028 0.17 3b-diol (pg/mL) rs10426628 G > A 22.51 0.74 20.69 1.28 22.16 0.66 15.19 3.47 0.13 0.011 rs2665582 G > A 21.81 0.65 22.78 2.71 22.01 0.65 11.86 6.36 0.72 0.028 3a-diol17G (ng/mL) rs10426628 G > A 2.98 0.12 3.40 0.20 3.10 0.11 3.60 0.59 0.049 0.50

NOTE: Only positive associations are shown. P value 0.05 are in bold characters. Abbreviations: Dom, dominant; Rec, recessive. lnCne e;2(1 ue1 2014 1, June 20(11) Res; Cancer Clin aThe major allele is to the left of the > sign and the minor allele is to the right. bTotal steroid levels (geometric mean SEM). cP values from linear regression of the natural logarithm of each hormone level regressed on age at blood donation and smoking status on the SNP under each model. ﬁ cation 2979 Published OnlineFirst March 28, 2014; DOI: 10.1158/1078-0432.CCR-13-2567

Levesque et al.

Figure 2. The estrogen and catechol estrogen biosynthetic pathways. 17bHSD, 17b-hydroxysteroid dehydrogenase; 17bHSD2, 17b-hydroxysteroid dehydrogenase type 2; CYP1B1, cytochrome P450 1B1; SULT2B1, sulfotransferase 2B1; 4-OHE1, 4-hydroxy-estrone; 4-OHE2, 4-hydroxy-estradiol; 2-OHE1, 2-hydroxy-estrone; 2-OHE2, 2-hydroxy-estradiol; 2-MeOHE1, 2-methoxy-estrone; 2-MeOHE2, 2-methoxy-estradiol; 4-MeOHE1, 4-methoxy-estrone; 4-MeOHE2, 4-methoxy-estradiol; E1, estrone; E2, estradiol; Q, quinone; CYP1A1, cytochrome P450 1A1; NQO1, NAD (P) H dehydrogenase, quinone 1; NQO2, NAD (P) H dehydrogenase, quinone 2.

importance of androgens in prostate cancer progression of protective or less reactive metabolites; therefore, these at all disease stages. Previous studies have suggested that, genes may be involved in host defence (Fig. 2). in addition to androgens, estrogens may contribute to CYP1B1 rs1800440 has the amino acid replacement Asn prostate cancer development and progression (18, ! Ser at position 453, which subjects it to proteasomal 19, 23, 27). Studies performed with humans and rodents degradation in a recombinant system (17, 29). This variant suggested that estrogens and their metabolites could be is associated with lower levels of 2-hydroxy and 16a- drivers of carcinogenesis (18, 28). Furthermore, CYP1B1 hydroxy estrogens in women with a family history of breast and COMT are associated with prostate cancer risk (23), cancer, thus revealing its role in estrogen biotransformation highlighting the significance of these genes in prostate in humans (30). Moreover, CYP1B1 is overexpressed in cancer carcinogenesis. To our knowledge, however, a prostate cancer and may, in addition to its role in estrogen comprehensive assessment of the genes related to estro- biotransformation, also catalyze 6-hydroxylation of T, and gen pathways has not been made in relation to a prog- therefore this gene would have a dual role in the sex steroid nostic point of view. We thus evaluated the impact of hormonal pathway. This function of CYP1B1 reflects single and cumulative adverse and protective genetic changes observed in androgen glucuronide levels in asso- markers associated with prostate cancer progression. This ciation with rs1800440 (Table 5). The complex role of the study demonstrates that germline polymorphisms in CYP1B1 enzyme in androgen and estrogen metabolism, genes associated with estrogen and CE metabolic path- combined with its role in prostate cancer risk and progres- ways are associated with prostate cancer progression in sion, definitely warrants additional investigation. Interest- Caucasians with localized and locally advanced disease. ingly, NQO1 and NQO2 SNPs are associated with progres- Remarkably, positive findings were observed for both sion only in localized prostate cancer, suggesting a potential cohorts for variants of CYP1B1, SULT2B1,andCOMT, role for this pathway early in the course of disease. Possibly, which suggests a role for these genes in prostate cancer the combination of polymorphisms in these estrogen-relat- progression. The cumulative association of the poly- ed genes modifies sex steroid exposure, e.g., estrogen and CE morphisms found for these genes also suggests that E2 exposure, or their relative abundance (compared with , its metabolites, and/or their repercussion on the andro- androgens) and thereby favors cancer progression. This gen axis are potentially involved in prostate cancer postulate remains to be demonstrated. Moreover, the progression. In addition, their roles in regulating the known association of HSD17B2 with cancer progression concentration of E2 and other biologically active meta- (5, 31) may also be related to its ability to catalyze the bolites, such as CYP1B1 involved in the production of 4- conversion of CE metabolites in addition to its role in hydroxy-metabolites of E2 and E1, may result in the converting E2 to E1 (Fig. 2). generation of large amounts of reactive oxygen species Given the interrelationship among these pathways, we (17), which may cause DNA damage and thereby enhance assessed the cumulative effect of variations in the related genes prostate carcinogenesis and progression. Indeed, several on prostate cancer progression to establish if a combination of studies have demonstrated the role of certain estrogen markers would be more predictive than individual SNPs. As metabolites, e.g., CE, in DNA-adduct formation and car- hypothesized, although the presence of each SNP is modestly cinogenicity (28). Such mutagenic compounds include informative, their cumulative association is more predictive of mostly 4-hydroxy-metabolites of estrogens, which can be outcome. Remarkably, the cumulative association of these further oxidized to CE quinones. In contrast, COMT, SNPs with the HSD17B2 markers stratifies these patients into SULT2B1,andNQO1/2 reduce the accumulation of reac- four prognostic subgroups (Fig. 1). The most promising tive oxygen species (28) and are involved in the formation eight-marker signature [CYP1B1, (rs1800440), SULT2B1

2980 Clin Cancer Res; 20(11) June 1, 2014 Clinical Cancer Research

Estrogen-Related Genes and Prostate Cancer Prognostic Stratiﬁcation

(rs12460535, rs2665582, rs10426628), and HSD17B2 most promising individual germline variants associated with (rs4243229, rs1364287, rs2955162, rs1119933)] has a cumu- outcomes will have OR/HR values around 2.0. In our study, lative association with PCa progression in localized and in two markers (rs16982844 and rs1800440) are above this locally advanced disease. Indeed, the BCR-free and progres- threshold. Interestingly, these two SNPs are also associated sion-free survival stratifications are respectively 87%, 80%, with significant changes in circulating steroid hormone 72%, and 55% in localized and 95%, 87%, 73%, and 9% in levels, which clearly reinforce their role in disease progres- locally advanced diseases (Fig. 1). sion. However, these stronger associations observed in the In support of their role(s) in cancer progression and the advanced cohort might overestimate the magnitude of the plausibility of our hypothesis, it is notable that these germ- effect compared with results obtained in localized disease, line variations are associated with alterations in circulating especially due to smaller size of the cohort and to the fact that hormone levels, which may create a specific environment our endpoints in this population included not only death but needed for cell population expansion. Of importance is the also androgen-deprivation therapy and metastasis. Thus, relationship observed for the SULT2B1 htSNPs and several before clinical translation, our findings require further val- circulating steroids. The protective SNP rs10426628, locat- idation in larger, independent, and interethnic prostate can- ed in intron 3 of SULT2B1, is associated with lower levels of cer cohorts. systemic T, DHT, and androsterone and higher levels of In conclusion, the cumulative association of inherited glucuronides, illustrating the broad effect on hormonal molecular markers in the estrogenic pathways is certain exposure associated with this SNP. Therefore, in addition to become attractive prognostic markers for prostate to its potential role on CE inactivation, this SULT2B1 cancer progression. Indeed, the cumulative impact of variant seems to affect events upstream of the steroidogen- such markers provides clinically relevant information esis of cholesterol, DHEA, and sulfatation of pregnenolone, from a single blood sample, which requires only a thereby controlling steroid bioavailability of sex steroid simple methodology, is invariant with time, overcomes precursors. tumor heterogeneity, and may identify potential targets To improve our understanding of the steroidogenic for a more personalized approach. The impact of such pathways in patients with prostate cancer, it will be host markers will help us understand genetic factors crucial to integrate the levels and ratios of circulating associated with hormone levels in the microenviron- androgens, estrogens, and their metabolites in indivi- ment of prostate cancer cells, which may promote para- duals with a normal prostate and patients with prostate crine-dependent tumor propagation. It will be critical to cancer to completely evaluate the hormonal milieu asso- determine the influence of these markers on intracrine ciated with disease aggressiveness and progression. Over- steroid conversion, which undeniably also supports cell all, the presence of such variations in the steroidogenic proliferation. In addition to previous findings about pathway may have a major consequence(s) on fine para- prostate cancer risk, variations in estrogen-related crine and intracrine regulation of key steroid carcinogenic genes such as ESR1, HSD17B2, CYP1B1, SULT2B1,and drivers. The role of estrogens and their derivatives defi- COMT point toward a polygenic contribution of estro- nitely deserves more attention because they can have gen-related genes with prostate cancer progression. The diverse effects on progression such as (i) several 4- cumulative association of CYP1B1 (rs1800440), hydroxy-CE metabolites can still bind the estrogen recep- SULT2B1 (rs12460535, rs2665582, and rs10426628) tor with very high affinity, the latter also being positively with HSD17B2 (rs4243229, rs1364287, rs2955162, and associated with prostate cancer progression (4), (ii) 4- rs1119933) may help stratify patients into four distinct hydroxy-CE metabolites can increase DNA-adduct forma- subgroups for the purpose of establishing an appropri- tion as reported for several other cancers (28), creating a ate prognosis. Overall, it is expected that combinations favorable environment for additional genomic mutagen- of markers in steroidogenic pathways, rather than SNP esis and proliferation, and (iii) MeOHEs can contribute to considered individually, will be more valuable at pre- the prevention of cancer progression given their antipro- dicting outcomes. Our findings may ultimately lead to liferative, proapoptotic, and antiangiogenic functions. an improved understanding of host–tumor/hormonal The strengths of our study include the large number of interactions underlying cancer progression and certainly patients, a candidate gene approach related to a specific constitute an additional step toward a more personal- steroidogenic pathway, the substantial plausibility of an ized approach to prostate cancer management. association(s) based on the biologic functions of the selected genes, expression of their corresponding enzymes in the fl human prostate, repeated associations in the two indepen- Disclosure of Potential Con icts of Interest E. Levesque, C. Guillemette, L. Lacombe, and Y. Fradet have been named dent and different types of cohorts, a significant follow-time inventors on a patent application owned by Laval University on a previous (7 years), associations between circulating steroid levels work related to this study. No potential conflicts of interest were disclosed by and progression-free, and overall survival endpoints. the other authors. Limitations of the study are related to the absence of data about circulating estrogen derivatives and the smaller cohort Authors' Contributions of patients with advanced disease. On the basis of our results Conception and design: E. Levesque, C. Guillemette (4, 5, 32) and those of others (8, 33, 34), it is expected that the Development of methodology: E. Levesque, C. Guillemette

www.aacrjournals.org Clin Cancer Res; 20(11) June 1, 2014 2981

Levesque et al.

Acquisition of data (provided animals, acquired and managed patients, Grant Support provided facilities, etc.): E. Levesque, I. Laverdiere, E. Audet-Walsh, This work was supported by Canadian research grants from Prostate P. Caron, M. Rouleau, Y. Fradet, L. Lacombe, C. Guillemette Cancer Canada (to E. Levesque), Cancer Research Society (to C. Guillemette), Analysis and interpretation of data (e.g., statistical analysis, biosta- the Canada Research Chair Program (to C. Guillemette), and the FRQ-S tistics, computational analysis): E. Levesque, I. Laverdiere, E. Audet- Innovation fund to the CHU Research Centre (to C. Guillemette, E. Levesque, Walsh, M. Rouleau, C. Guillemette L. Lacombe, and Y. Fradet). E. Levesque is recipient of a Prostate Cancer Writing, review, and/or revision of the manuscript: E. Levesque, I. Laver- Canada rising star award (RS2013-55). I. Laverdiere, E. Audet-Walsh, and M. diere, E. Audet-Walsh, M. Rouleau, Y. Fradet, L. Lacombe, C. Guillemette Rouleau are recipients of a Frederick Banting and Charles Best Canada Administrative, technical, or material support (i.e., reporting or orga- Graduate Scholarship awards. C. Guillemette holds the Canada Research nizing data, constructing databases): E. Levesque, Y. Fradet, L. Lacombe, Chair in pharmacogenomics. I. Laverdiere is a recipient of a clinician- C. Guillemette scientist scholarship from the FRQ-S. Study supervision: E. Levesque, C. Guillemette The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked Acknowledgments advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate The authors thank the personnel of the genetic platform, particularly this fact. Sylvie Desjardins, for help with genotyping, and the personnel of the clinical research platform, particularly Sun Makosso-Kallyth, for support with the Received September 18, 2013; revised March 3, 2014; accepted March 23, statistical analysis. 2014; published OnlineFirst March 28, 2014.

References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer prostate cancer: a mechanism for castration-resistant tumor growth. statistics. CA Cancer J Clin 2011;61:69–90. Cancer Res 2008;68:4447–54. 2. Olmos D, Brewer D, Clark J, Danila DC, Parker C, Attard G, et al. 15. Mitsiades N, Sung CC, Schultz N, Danila DC, He B, Eedunuri VK, et al. Prognostic value of blood mRNA expression signatures in castration- Distinct patterns of dysregulated expression of enzymes involved in resistant prostate cancer: a prospective, two-stage study. Lancet androgen synthesis and metabolism in metastatic prostate cancer Oncol 2012;13:1114–24. tumors. Cancer Res 2012;72:6142–52. 3. Miyamoto DT, Lee RJ, Stott SL, Ting DT, Wittner BS, Ulman M, et al. 16. Resel Folkersma L, San Jose Manso L, Galante Romo I, Moreno Sierra Androgen receptor signaling in circulating tumor cells as a marker of J, Olivier Gomez C. Prognostic significance of circulating tumor cell hormonally responsive prostate cancer. Cancer Discov 2012;2: count in patients with metastatic hormone-sensitive prostate cancer. 995–1003. Urology 2012;80:1328–32. 4. Levesque E, Huang SP, Audet-Walsh E, Lacombe L, Bao BY, Fradet Y, 17. Sissung TM, Price DK, Sparreboom A, Figg WD. Pharmacogenetics et al. Molecular markers in key steroidogenic pathways, circulating and regulation of human cytochrome P450 1B1: implications in hor- steroid levels, and prostate cancer progression. Clin Cancer Res mone-mediated tumor metabolism and a novel target for therapeutic 2013;19:699–709. intervention. Mol Cancer Res 2006;4:135–50. 5. Audet-Walsh E, Bellemare J, Lacombe L, Fradet Y, Fradet V, Douville 18. Bosland MC, Mahmoud AM. Hormones and prostate carcinogenesis: P, et al. The impact of germline genetic variations in hydroxysteroid androgens and estrogens. J Carcinog 2011;10:33. (17-Beta) dehydrogenases on prostate cancer outcomes after pros- 19. Ellem SJ, Risbridger GP. The dual, opposing roles of estrogen in the tatectomy. Eur Urol 2012;62:88–96. prostate. Ann NY Acad Sci. 2009;1155:174–86. 6. Ahn J, Kibel AS, Park JY, Rebbeck TR, Rennert H, Stanford JL, et al. 20. Bosland MC, Ford H, Horton L. Induction at high incidence of Prostate cancer predisposition loci and risk of metastatic disease and ductal prostate adenocarcinomas in NBL/Cr and Sprague-Dawley prostate cancer recurrence. Clin Cancer Res 2011;17:1075–81. Hsd:SD rats treated with a combination of testosterone and 7. Lin DW, FitzGerald LM, Fu R, Kwon EM, Zheng SL, Kolb S, et al. estradiol-17 beta or diethylstilbestrol. Carcinogenesis 1995;16: Genetic variants in the LEPR, CRY1, RNASEL, IL4, and ARVCF genes 1311–7. are prognostic markers of prostate cancer-specific mortality. Cancer 21. McPherson SJ, Wang H, Jones ME, Pedersen J, Iismaa TP, Wreford N, Epidemiol Biomarkers Prev 2011;20:1928–36. et al. Elevated androgens and prolactin in aromatase-deficient mice 8. Ross RW, Oh WK, Xie W, Pomerantz M, Nakabayashi M, Sartor O, et al. cause enlargement, but not malignancy, of the prostate gland. Endo- Inherited variation in the androgen pathway is associated with the crinology 2001;142:2458–67. efficacy of androgen-deprivation therapy in men with prostate cancer. 22. Jarred RA, McPherson SJ, Bianco JJ, Couse JF, Korach KS, Risbridger J Clin Oncol 2008;26:842–7. GP. Prostate phenotypes in estrogen-modulated transgenic mice. 9. Cheng I, Plummer SJ, Neslund-Dudas C, Klein EA, Casey G, Rybicki Trends Endocrinol Metab 2002;13:163–8. BA, et al. Prostate cancer susceptibility variants confer increased risk 23. Cussenot O, Azzouzi AR, Nicolaiew N, Fromont G, Mangin P, Cormier of disease progression. Cancer Epidemiol Biomarkers Prev 2010;19: L, et al. Combination of polymorphisms from genes related to estrogen 2124–32. metabolism and risk of prostate cancers: the hidden face of estrogens. 10. Gallagher DJ, Vijai J, Cronin AM, Bhatia J, Vickers AJ, Gaudet MM, J Clin Oncol 2007;25:3596–602. et al. Susceptibility loci associated with prostate cancer progression 24. Fizazi K, Scher HI, Molina A, Logothetis CJ, Chi KN, Jones RJ, et al. and mortality. Clin Cancer Res 2010;16:2819–32. Abiraterone acetate for treatment of metastatic castration-resistant 11. Wiklund FE, Adami HO, Zheng SL, Stattin P, Isaacs WB, Gronberg H, prostate cancer: final overall survival analysis of the COU-AA-301 et al. Established prostate cancer susceptibility variants are not asso- randomised, double-blind, placebo-controlled phase 3 study. Lancet ciated with disease outcome. Cancer Epidemiol Biomarkers Prev Oncol 2012;13:983–92. 2009;18:1659–62. 25. Ryan CJ, Smith MR, de Bono JS, Molina A, Logothetis CJ, de Souza P, 12. Shui IM, Lindstrom S, Kibel AS, Berndt SI, Campa D, Gerke T, et al. et al. Abiraterone in metastatic prostate cancer without previous Prostate cancer (PCa) risk variants and risk of fatal PCa in the National chemotherapy. N Engl J Med 2013;368:138–48. Cancer Institute Breast and Prostate Cancer Cohort Consortium. Eur 26. Scher HI, Fizazi K, Saad F, Taplin ME, Sternberg CN, Miller K, et al. Urol 2014 Jan 4. [Epub ahead of print]. Increased survival with enzalutamide in prostate cancer after chemo- 13. Labrie F. Intracrinology. Mol Cell Endocrinol 1991;78:C113–8. therapy. N Engl J Med 2012;367:1187–97. 14. Montgomery RB, Mostaghel EA, Vessella R, Hess DL, Kalhorn TF, 27. Ellem SJ, Risbridger GP. Treating prostate cancer: a rationale for Higano CS, et al. Maintenance of intratumoral androgens in metastatic targeting local oestrogens. Nat Rev Cancer 2007;7:621–7.

2982 Clin Cancer Res; 20(11) June 1, 2014 Clinical Cancer Research

Estrogen-Related Genes and Prostate Cancer Prognostic Stratiﬁcation

28. Cavalieri EL, Rogan EG. Unbalanced metabolism of endogenous genome of castration-resistant prostate cancer. Cancer Res 2012; estrogens in the etiology and prevention of human cancer. J Steroid 72:616–25. Biochem Mol Biol 2011;125:169–80. 32. Levesque E, Laverdiere I, Lacombe L, Caron P, Rouleau M, Turcotte V, 29. Bandiera S, Weidlich S, Harth V, Broede P, Ko Y, Friedberg T. Pro- et al. Importance of 5alpha-reductase gene polymorphisms on circu- teasomal degradation of human CYP1B1: effect of the Asn453Ser lating and intraprostatic androgens in prostate cancer. Clin Cancer Res polymorphism on the post-translational regulation of CYP1B1 expres- 2014;20:576–84. sion. Mol Pharmacol 2005;67:435–43. 33. Penney KL, Pyne S, Schumacher FR, Sinnott JA, Mucci LA, Kraft PL, 30. Greenlee H, Chen Y, Kabat GC, Wang Q, Kibriya MG, Gurvich I, et al. et al. Genome-wide association study of prostate cancer mortality. Variants in estrogen metabolism and biosynthesis genes and urinary Cancer Epidemiol Biomarkers Prev 2010;19:2869–76. estrogen metabolites in women with a family history of breast cancer. 34. Yang M, Xie W, Mostaghel E, Nakabayashi M, Werner L, Sun T, et al. Breast Cancer Res Treat 2007;102:111–7. SLCO2B1 and SLCO1B3 may determine time to progression for 31. FriedlanderTW,RoyR,TomlinsSA,NgoVT,KobayashiY,Aza- patients receiving androgen deprivation therapy for prostate cancer. meera A, et al. Common structural and epigenetic changes in the J Clin Oncol 2011;29:2565–73.

www.aacrjournals.org Clin Cancer Res; 20(11) June 1, 2014 2983

Steroidogenic Germline Polymorphism Predictors of Prostate Cancer Progression in the Estradiol Pathway

Éric Lévesque, Isabelle Laverdière, Étienne Audet-Walsh, et al.

Clin Cancer Res 2014;20:2971-2983. Published OnlineFirst March 28, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-13-2567

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2014/03/28/1078-0432.CCR-13-2567.DC1

Cited articles This article cites 33 articles, 17 of which you can access for free at: http://clincancerres.aacrjournals.org/content/20/11/2971.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/20/11/2971.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/20/11/2971. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.