Genetic Analysis of the RNASEL Gene in Hereditary, Familial, and Sporadic Prostate Cancer

7150 Vol. 10, 7150–7156, November 1, 2004 Clinical Cancer Research

Featured Article Genetic Analysis of the RNASEL Gene in Hereditary, Familial, and Sporadic Prostate Cancer

Fredrik Wiklund,1 Bjo¨rn-Anders Jonsson,2 0.77; 95% confidence interval, 0.59–1.00) and reduced risk Anthony J. Brookes,3 Linda Stro¨mqvist,3 of prostate cancer in carriers of two different haplotypes being completely discordant. Jan Adolfsson,4 Monica Emanuelsson,1 5 5 Conclusions: Considering the high quality in genotyp- Hans-Olov Adami, Katarina Augustsson-Ba¨lter, ing and the size of this study, these results provide solid 1 and Henrik Gro¨nberg evidence against a major role of RNASEL in prostate cancer Department of 1Radiation Sciences, Oncology, and 2Medical etiology in Sweden. Biosciences, Pathology, University of Umeå, Umeå, Sweden; and 3Center for Genomics and Bioinformatics, 4Oncologic Center, Department of Surgical Sciences, and 5Department of Medical INTRODUCTION Epidemiology and Biostatistics, Karolinska Institutet, Sweden Accumulating evidence from epidemiologic and genetic studies indicates that hereditary predisposition has a consider- ABSTRACT able impact on the development of prostate cancer. Linkage analyses suggest that a number of chromosomal regions harbor Purpose: The RNASEL gene has been proposed as a prostate cancer susceptibility genes. However, lack of consis- candidate gene for the HPC1 locus through a positional tency between studies additionally indicates that prostate cancer cloning and candidate gene approach. Cosegregation be- is a genetically heterogeneous disorder, with multiple genetic tween the truncating mutation E265X and disease in a he- and environmental factors involved in its etiology (1). In 1996, reditary prostate cancer (HPC) family and association be- the first prostate cancer susceptibility locus, HPC1, was mapped tween prostate cancer risk and the common missense to chromosome 1q24-25 (2). Subsequent studies of linkage to variant R462Q has been reported. To additionally evaluate HPC1 have proven inconclusive; some confirmed linkage (3– the possible role of RNASEL in susceptibility to prostate 6), although others did not (7–11). However, in a pooled anal- cancer risk, we performed a comprehensive genetic analysis ysis of 772 families from nine international groups confirmatory of sequence variants in RNASEL in the Swedish population. linkage evidence was found with a maximum heterogeneity Experimental Design: Using 1624 prostate cancer cases logarithm of odds score of 1.4 (12). and 801 unaffected controls, the truncating mutation E265X RNASEL maps to HPC1 and is involved in the IFN-regu- and five common sequence variants, including the two missense lated 2Ј-5Ј–linked oligoadenylates system that mediates cell mutations R462Q and D541E, were evaluated for association proliferation and apoptosis (13, 14) and has been suggested as a between genotypes/haplotypes and prostate cancer risk. candidate tumor suppressor gene (15). RNASEL was recently Results: The prevalence of E265X carriers among un- identified as a candidate gene for HPC1 through a positional affected controls and prostate cancer patients was almost cloning and candidate gene approach. A truncating mutation identical (1.9 and 1.8% in controls and cases, respectively), (E265X) and an initiation codon mutation (M1I) was reported to and evidence for segregation of E265X with disease was not cosegregate within two hereditary prostate cancer (HPC; ref. 2) observed within any HPC family. Overall, the analyses of families linked to HPC1 (16). Furthermore, loss of the wild-type common sequence variants provided limited evidence for allele and reduced RNASEL activity was observed in microdis- association with prostate cancer risk. We found a marginally sected tumors carrying a germ-line mutation. In Finnish HPC significant inverse association between the missense muta- families, RNASEL mutations did not explain disease segregation D541E and sporadic prostate cancer risk (odds ratio, tion; however, the truncating mutation E265X was associated with prostate cancer risk (17). In addition to rare mutations, several polymorphisms in RNASEL have been reported. A common missense variant of Received 5/19/04; accepted 7/26/04. RNASEL, R462Q, has been associated both with increased (17, Grant support: F. Wiklund received support from Swedish Cancer 18) and decreased (19, 20) risk of HPC. Recently, Xiang et al. Society (Cancerfonden), Lion’s Cancer Foundation, and Stiftelsen fo¨r (21) reported that the R462Q variant reduces the ability of Strategisk Forskning (Genome Program). RNase L to cause apoptosis in response to activation by 2-5A. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked Hence, this polymorphism might have a functional role in pros- advertisement in accordance with 18 U.S.C. Section 1734 solely to tate cancer development. Another missense variant, D541E, was indicate this fact. associated with an increased risk for prostate cancer in a study Requests for reprints: Fredrik Wiklund, Department of Radiation from Japan (19), although several other studies found no asso- Sciences, Oncology, University of Umeå, 901 87 Umeå, Sweden. Phone: 46-90-7852270; Fax: 46-90-127464; E-mail: fredrik.wiklund@ ciation (17, 18, 20). oc.umu.se. Additional evaluation of the impact that the RNASEL gene ©2004 American Association for Cancer Research. have on prostate cancer risk are required. Our aim was to assess

the possible role of RNASEL sequence variants on prostate (same as for cases). Subsequently, eight controls were excluded cancer risk in a Swedish population. We used a large popula- because linkage to the Swedish Cancer Registry revealed that tion-based case-control study, composed of 1624 prostate cancer they were diagnosed with a cancer of the prostate before inclu- cases and 801 controls. The truncating mutation E265X and sion. At the time of this study, DNA samples were available for several common polymorphisms in RNASEL were evaluated, 801 controls and 1427 cases (1247 sporadic prostate cancer both for individual association and through haplotype analyses. cases, 131 familial prostate cancer cases, and 49 HPC cases). Patients with Familial Prostate Cancer and HPC. Re- MATERIALS AND METHODS cruitment of Swedish families with prostate cancer is an ongo- Patients with Prostate Cancer and Controls. This ing activity that started at the Department of Oncology, Univer- study used men enrolled into the population based case-control sity of Umeå, in 1995. Identification of families has been based study Cancer Prostate in Sweden. A detailed description of the mainly on referrals by collaborating urologists and oncologists Cancer Prostate in Sweden study has been given elsewhere (22). throughout Sweden but also on self-reported family history of The Cancer Prostate in Sweden study is composed of all prostate prostate cancer from cases included into the Cancer Prostate in cancer patients, ages Յ 79 years, diagnosed between July 1, Sweden study, described above. Blood has been collected for as 2001, and September 30, 2002, from the central and northern many family members as possible. For the present study, all part of Sweden and all patients, ages Յ 65 years, diagnosed in affected family members with DNA information available (ex- the same calendar period from the southeastern part of Sweden cluding cases from the Cancer Prostate in Sweden cohort al- and the area of Stockholm. In total, 1961 prostate cancer pa- ready selected for this study) were included, resulting in 184 tients were invited for participation and 1444 (74%) agreed to HPC cases and 25 familial prostate cancer cases. All diagnoses participate. All these cases donated a blood sample and an- of prostate cancer in the families were confirmed both by swered a questionnaire, including questions regarding family reference to the Cancer Registry and by direct examination of history for prostate cancer. For all cases with at least one medical records. reported family member with prostate cancer, based on the Total Study Population. Our total study population is initial questionnaire, a more detailed family history was ob- composed of 1636 prostate cancer patients and 801 unaffected tained through a second questionnaire followed by a telephone controls. Of the 1636 prostate cancer patients, 156 (9.5%) and interview. 233 (14.2%) were classified as familial prostate cancer and HPC Of the 1444 cases included in the study, 143 were classified cases, respectively. Pertinent characteristics of the study popu- as familial prostate cancer cases (two relatives with verified lation are given in Table 1. Written informed consent was prostate cancer in a nuclear family) and 51 as HPC cases (three obtained from each subject. The ethical committee at the Karo- or more members with prostate cancer in a nuclear family). linska Institutet and University of Umeå approved the study. Detailed clinical data, including tumor-node-metastasis stage, Selection of RNASEL Single Nucleotide Polymorphisms Gleason score, prostate-specific antigen level at time of diag- (SNPs). Located on chromosome 1q22, the RNASEL gene nosis, means of diagnosis and primary treatment were obtained with eight exons is ϳ13 kb. To comprehensively evaluate as- through linkage to the National Prostate Cancer Registry. Cases sociation between prostate cancer risk and common RNASEL were classified as either localized (tumor stage I-II and Gleason sequence variants, we used the following approach. First, we sum Ͻ 8/grade I-II and prostate-specific antigen Ͻ 100) or defined the target region for selection of SNPs as 3 kb of the advanced (tumor stage III-IV or Gleason sum Ն 8/grade III or promoter, all exons, introns, and 3Ј-untranslated region. We prostate-specific antigen Ն 100). then searched the public database National Center for Biotech- The continuously updated Swedish Population Registry nology Information (www.ncbi.nlm.nih.gov/SNP) within this encompasses the individually unique national registration num- target region, resulting in the identification of 34 SNPs. A subset ber assigned to all residents. From this register, control subjects of nine SNPs were then selected, with the criteria of a reported were randomly selected and frequency matched according to minor allele frequency of at least 5%, a density of one SNP per geographical origin (northern part of Sweden versus southeast- three kb, and as low as possible proportion of repetitive se- ern part of Sweden and the area of Stockholm) and to the quences. Preferentially validated SNPs were selected, and addi- expected age distribution of the cases (within 5-year age tional attention was paid to SNPs in the promoter region. Be- groups). The controls were recruited during the same calendar sides the two reported missense variants, R462Q and D541E, period as the cases. Of 1697 controls invited, 866 (51%) agreed which were decided to be included in the analyses in advance, to participate and to complete a blood draw and questionnaire this subset consisted of four SNPs in the promoter region, one

Table 1 Pertinent characteristics of the study population Patient group Number Mean age* (range) Proportion of advanced tumors† Controls 801 67.8 (45.5–80.0) Sporadic prostate cancer 1247 66.7 (47.3–80.4) 512/1182 (43.3%) Familial prostate cancer 156 65.8 (47.0–80.0) 51/124 (41.1%) Hereditary prostate cancer 233 65.3 (43.0–81.0) 15/51 (29.4%) * Age defined as age at diagnosis for cases and age at blood sampling for controls. † Information of tumor characteristics only available for cases ascertained through the Cancer Prostate in Sweden study.

Table 2 RNASEL sequence variants selected for association analysis dbSNP Location Position* Nucleotide change Amino acid change Cause of exclusion† rs2274509 5Ј-UTR Ϫ2,575 AϾG rs618685 5Ј-UTR Ϫ659 GϾA Failed assay rs3845458 5Ј-UTR Ϫ511 GϾA Monomorphic rs620070 5Ј-UTR Ϫ389 AϾG PCR failure rs486907 Exon 2 793 GϾT E265X rs486907 Exon 2 1,385 GϾA R462Q rs627928 Exon 4 4,605 GϾT D541E rs579006 Intron 4 6,606 TϾC PCR failure rs672527 Intron 5 10,721 GϾA rs627839 3Ј-UTR 13,861 AϾC * Positions are relative to the A in the initiation codon (ATG) from RNASEL genomic DNA (NT_004487). † Four SNPs were excluded from additional analysis based on genotyping results in 96 individuals. Abbreviations: UTR, untranslated region; dbSNP, public database National Center for Biotechnology Information.

SNP in intron 4, one SNP in intron 5, and one SNP in 3Ј- were designed by custom DFold software (27) provided by untranslated region (Table 2). These nine SNPs were genotyped DynaMetrix Ltd. (Hertfordshire, United Kingdom). These oli- in a randomly selected subset of 96 control subjects from the gonucleotides were provided and HPLC purified, by Biomers Cancer Prostate in Sweden study. This resulted in the elimina- GmbH (Ulm, Germany). The dynamic allele-specific hybridization of four SNPs due to failed assay (one SNP), PCR failure tion PCRs entailed amplifying short genomic fragments span- (two SNPs), and monomorphic result (one SNP; see Table 2). ning the variant of interest, with one of the primers carrying a The remaining five SNPs were all in Hardy-Weinberg equilib- 5Ј-biotin label. Amplifications were performed in 5 ␮Lof rium. volume, containing 1 to 2 ng of genomic DNA, 0.38 ␮mol/L Haplotypes were estimated with a method proposed by biotinylated primer, 0.75 ␮mol/L nonbiotinylated primer, 0.03 Stephens et al. (23), as implemented in the software PHASE. units of AmpliTaq Gold (Applied Biosystems, Inc.), 10% dim- Using a Markov Chain Monte Carlo approach, this method ethylsulphoxide, 1ϫ AmpliTaq Gold Buffer, including 1.5 incorporates a statistical model for the distribution of unresolved mmol/L MgCl2 (Applied Biosystems, Inc.), and 0.2 mmol/L haplotypes based on coalescent theory. Haplotype-tagging SNPs each deoxynucleoside triphosphate. Thermal cycling was con- were determined by using the haplotype-tagging SNP2 software. ducted on a MBS 384 device (Thermo-Hybaid, Thermo Electron Because removing any one of the five SNPs resulted in a subset Corporation, Milford, MA) as follows: 1ϫ (10 minutes at 94°C) of haplotype-tagging SNPs that retained Ͻ95% of the diversity and 35ϫ (15 seconds at 94°C, 30 seconds at annealing temper- observed among the 96 control individuals, all five SNPs were ature). To verify successful amplification, 0.5 ␮L of several genotyped in the total study population. In addition to the five randomly chosen samples were examined on a 3.0% low-melt SNPs, the truncating stop mutation E265X was also genotyped agarose gel. in all individuals. Dynamic allele-specific hybridization analysis of the PCR Genotyping. DNA samples were extracted from leuko- product was conducted on membrane macroarrays, with the cytes with standard methods. Each DNA plate contained two dynamic allele-specific hybridization-2 protocol (25). Briefly, Centre d’Etude du Polymorphisme Humain controls, a water this entailed transferring samples to the membrane array by blank and blinded internal replicates. Two methods were used for genotyping. The stop mutation (E265X) and the two mis- centrifugation (28) or robotic gridding. Resulting individual sense mutations (R462Q and D541E) were genotyped using a arrays with up to 9600 distinct features were rinsed in 0.1 mol/L 5Ј-nuclease assay with TaqMan MGB probes. Primers and NaOH to denature the PCR products. They were then exposed to probes were designed with the Assay-by-Design service (Ap- 2 mL of HE buffer [0.1 mol/L HEPES and 10 mmol/L EDTA plied Biosystems, Inc., Foster City, CA). All reactions were (pH 7.9)] containing 4 nmol of the appropriate ROX end-labeled performed in 10 ␮L of volume consisting of 10 ng of genomic probe. After heating to 85°C and cooling to room temperature, DNA, 900 nmol/L of each primer, 200 nmol/L of each probe, the membrane was briefly rinsed in HE buffer. The array was and 5 ␮L of TaqMan universal master-mix. PCR cycling con- then soaked in 40 mL of HE buffer containing SYBR GreenI ditions were as follows: 50°C for 2 minutes, 95°C for 10 dye at 1:20,000 dilution for up to 3 hours. Using a dynamic minutes, followed by 40 cycles of 92°C for 15 seconds and 60°C allele-specific hybridization-2 device (DynaMetrix Ltd.), the for 1 minute. The samples were analyzed on an ABI PRISM membrane was taken through a dynamic allele-specific hybrid- 7900 HT sequence detection system. ization heating ramp (heating at 3°C/minute from room temper- A dynamic allele-specific hybridization technique was ature to 85°C), whereas fluorescence from the ROX acceptor adapted to genotype the remaining three SNPs. To thoroughly dye on the probe was monitored. Data were collected at intervals evaluate the genotype results, the missense variant R462Q was of 0.5°C. We used fluorescence changes with temperature also genotyped with the dynamic allele-specific hybridization (DNA melting profiles) to distinguish different alleles. This was technique. Dynamic allele-specific hybridization was performed done by the dynamic allele-specific hybridization-2 device soft- as described previously (24–26). For this, two PCR primers and ware that uses negative derivatives of fluorescence against tem- one dynamic allele-specific hybridization probe per target SNP perature to reveal peaks of denaturation rate (target-probe melt-

ing temperatures; Tm) and thereby automatically assign DNA Centre d’Etude du Polymorphisme Humain samples gave con- samples into genotype groups. sistent genotypes. No significant deviation of the genotype Statistical Methods. We tested Hardy-Weinberg equi- frequencies from those expected under Hardy-Weinberg equilibrium for each sequence variant and pairwise linkage disequi- librium was observed for any of the six sequence variants, librium between all sequence variants using a replication neither among the 801 control individuals nor among the 1504 method (29). For each test, 10,000 permutations were per- unrelated prostate cancer cases (all P values Ͼ 0.19; range, 0.19 formed, and the Fisher probability test statistic of each replicate to 1.00). We found strong evidence for pairwise linkage dis- was calculated from the new corresponding multilocus table. equilibrium between these variants (all P values Ͻ 0.0001) and Empirical P values for each test were estimated as the propor- pairwise DЈ estimates between adjacent SNPs were high (mean, tion of replicate data sets found to be as probable as or less 0.94; range, 0.76 to 1.00). probable than the observed data, as implemented in the software The truncating mutation, E265X, was observed in 15 package Genetic Data Analysis. (1.9%) controls, in 23 (1.9%) sporadic prostate cancer cases, We used unconditional logistic regression to evaluate the and in five (1.4%) familial prostate cancer or HPC cases (Table association of each of the sequence variants with prostate cancer 3). There was no significant difference in proportion of mutation risk with the statistical software STATA. To adjust for the carriers between controls and sporadic prostate cancer cases or matching, we included indicator variables, representing each between controls and familial prostate cancer or HPC cases. combination of age category (5-year age groups) and geograph- Among cases with HPC, four individuals (2.1%) were observed ical region (northern part of Sweden versus southeastern part of with the stop mutation. This proportion was not significantly Sweden and the area of Stockholm), in the regression model. A different from the proportion observed among controls. The four likelihood ratio test of a covariate equal to number of rare alleles HPC cases carrying the stop mutations are members of three (0, 1, or 2), corresponding to multiplicative allelic effects on the different families. In one family, including five affected brothers genotype relative risks, was used to test for association between with available DNA samples, two were mutation carriers, genotypes and disease. To account for genotype correlations whereas the other three did not carry the mutation. In the other among individuals from the same family (in comparing familial two HPC families, including one affected mutation carrier, no prostate cancer or HPC cases with controls), a modified sand- DNA samples were available from any other family member. wich variance estimator in clustered logistic regression analysis The median age at sampling for the 15 control individuals was used (30). Because the likelihood used for estimation is not carrying the mutation was 4 years younger than for the control a true likelihood, a Wald test was applied to test for association between genotypes and prostate cancer risk. Association between haplotypes and prostate cancer risk was evaluated with a score test proposed by Schaid et al. (31), Table 3 Genotype frequencies of RNASEL sequence variants in as implemented in the software HAPLO.SCORE for the R controls, patients with SPC and patients with FPC or HPC programming language. On the basis of a generalized linear Familial model, this method allows adjustment for possible confounding Sporadic prostate cancer Controls prostate cancer and HPC variables and provides both global and haplotype-specific tests. Variant and In these analyses, age and geographical region was adjusted for genotype No. % No. % P* No. % P† (as described above) and haplotypes with frequencies Ͻ 0.005 Ϫ2575 AϾG were pooled into a common group. Empirical P values, based on AA 268 35.8 400 34.6 0.474 144 40.1 0.355 10,000 simulations, were computed for the global score test and AG 371 49.6 584 50.5 161 44.8 each of the haplotype-specific score tests. GG 109 14.6 173 15.0 54 15.0 E265X No correction was made for multiple comparisons. All GG 780 98.1 1204 98.1 0.988 345 98.6 0.655 reported P values are two sided. GT 15 1.9 23 1.9 5 1.4 R462Q GG 297 37.3 451 36.4 0.417 146 38.0 0.805 RESULTS GA 384 48.2 598 48.3 180 46.9 AA 115 14.4 189 15.3 58 15.1 In total, 14,466 genotypes were assayed, of which, 13,988 D541E (96.7%) provided a successful result. For the two missense GG 257 32.5 407 33.5 0.090 115 33.1 0.622 mutations (R462Q and D541E) and the stop mutation (E265X), GT 372 47.0 601 49.4 167 48.1 the success rate was 99.5, 96.7, and 97.4%, respectively. The TT 162 20.5 208 17.1 65 18.7 ϩ Ͼ lowest success rate was observed for the two SNPs located in 10721 G A GG 381 50.8 599 51.7 0.408 192 53.0 0.966 the promoter and in intron 5, 93.3% in both cases. Quality GA 313 41.7 462 39.9 141 39.0 control of genotyping results based on included Centre d’Etude AA 56 7.5 98 8.5 29 8.0 du Polymorphisme Humain controls (59 samples), repeated ϩ13861 AϾC study samples (from 115 individuals), and independent geno- AA 211 27.4 316 26.3 0.917 105 27.5 0.871 AC 386 50.1 622 51.7 188 49.2 typing of the R462Q variant in two different laboratories (with CC 173 22.5 264 22.0 89 23.3 two different genotyping methods) provided an estimated error * Likelihood ratio test based on a logistic regression model ad- rate of 0.3%. Repeated samples revealed 19 discordant geno- justed for age and geographical region. typing results of 2,901 samples verified. Repeated samples with † Wald test based on clustered logistic regression model with a discordant genotyping result were blanked in the analysis. All robust variance estimate adjusted for age and geographical region.

individuals who were not carriers (P ϭ 0.08, Mann-Whitney with controls (2.9 versus 1.5%, P ϭ 0.010). This haplotype also test). Stratified analyses based on age (Ͻ65 years or Ͼ65 years) occurred at lower frequency among familial prostate cancer and and tumor aggressiveness (localized or locally advanced) did HPC cases (1.6%) compared with controls, yet there was no not reveal any significant differences in proportion of mutation significant difference (P ϭ 0.181). The other haplotype signif- carriers between controls and prostate cancer cases. icantly associated with prostate cancer risk was the GGTGC No significant difference in genotype frequencies between haplotype (containing the opposite allele of each SNP to the sporadic prostate cancer cases, familial prostate cancer cases, or above associated haplotype) that also occurred at higher fre- HPC cases and controls were observed for any of the five quencies among controls (3.1%) compared with sporadic pros- common sequence variants based on a multiplicative genetic tate cancer cases (2.5%, P ϭ 0.415) and familial prostate cancer/ model (Table 3). Assuming a dominant or recessive allelic effect HPC cases (0.8%, P ϭ 0.020). The stop mutation E265X was in on prostate cancer risk did not reveal any significant differences all cases observed on the most frequent haplotype AGTGC. between cases and controls for any of the studied variants, Haplotype analyses, stratified by age and tumor aggressiveness, except for the missense variant D541E. With a recessive model, did not reveal any additional support for association between D541E showed a significant inverse association of the less RNASEL sequence variants and prostate cancer risk (data not common allele between individuals with sporadic prostate can- shown). cer and controls (␹2 ϭ 4.86, P ϭ 0.03). However, between familial prostate cancer cases or HPC cases and controls, no significant differences were observed under a recessive model. DISCUSSION In subgroup analyses based on age and tumor aggressiveness, no Linkage analyses of high-risk prostate cancer families pro- significant association between these variants and prostate can- vide convincing evidence that the HPC1 locus is likely to harbor cer risk was observed. a prostate cancer susceptibility gene. In 2002, Carpten et al. (16) Haplotype frequencies of the five common sequence vari- proposed RNASEL as a plausible candidate gene for this region ants were estimated in familial prostate cancer and HPC cases through a positional cloning and candidate gene approach. To (selecting only the proband among related cases) and sporadic test the hypothesis that RNASEL sequence variants are associ- prostate cancer cases and controls (Table 4). A global score test ated with prostate cancer risk, we genotyped five common provided no significant difference in haplotype frequency be- SNPs, including the missense variants R462Q and D541E, and tween sporadic prostate cancer cases and controls (P ϭ 0.46), the rare stop mutation E265X in 801 unaffected men and 1636 nor did a test of the maximum among all score-specific statistics prostate cancer patients. Overall, our study provided limited (P ϭ 0.10). This latter test should have greater statistical power support for the hypothesis that RNASEL is a prostate cancer compared with the global test if only a few haplotypes were susceptibility gene. associated with prostate cancer risk. Comparing familial pros- In the initial report of RNASEL as a candidate susceptibility tate cancer and HPC cases with controls did not provide any gene for prostate cancer, an index case from each of 26 families support for association between the haplotypes and prostate at high risk for prostate cancer were screened for mutations in cancer risk (global score test, P ϭ 0.33; max-score test, P ϭ the RNASEL gene. Among these families, eight were linked to 0.25). Individual haplotype analyses revealed two haplotypes the HPC1 region with at least four affected individuals sharing significantly associated with prostate cancer risk. The frequency an HPC1 haplotype. In this selected set of families, the stop of the AAGAA haplotype of SNPs Ϫ2575 AϾG, R462Q GϾA, mutation E265X, detected in one (3.8%) of the index cases, D541E GϾT, ϩ10721 GϾA, and ϩ13861 AϾC was signifi- cosegregated with disease status within that family. Ro¨kman et cantly decreased in sporadic prostate cancer cases compared al. (17) reported the detection of E265X in five (4.3%) index

Table 4 Estimated RNASEL haplotype frequencies in controls, patients with SPC and patients with FPC or HPC Sporadic prostate Familial prostate cancer cancer and HPC Controls Haplotype % % Score* P† % Score* P† A G T G C 24.2 23.8 Ϫ0.69 0.492 21.4 Ϫ0.77 0.461 G A G A A 19.4 21.2 1.33 0.183 18.7 Ϫ0.46 0.620 A G T G A 13.9 13.5 Ϫ0.49 0.625 16.9 0.99 0.319 G A G G C 11.7 12.2 0.24 0.814 13.4 0.74 0.467 A G G G C 8.7 9.0 0.62 0.534 11.3 1.45 0.142 A G G G A 6.1 6.6 0.76 0.445 5.6 0.15 0.885 G A G G A 3.5 2.9 Ϫ0.68 0.495 3.6 0.36 0.720 G G T G C 3.1 2.5 Ϫ0.81 0.415 0.8 Ϫ2.29 0.020 A A G A A 2.9 1.5 Ϫ2.59 0.010 1.6 Ϫ1.29 0.181 A G G A A 2.6 2.7 0.30 0.763 0.7 Ϫ0.94 0.363 A G T A A 2.3 2.6 0.07 0.945 4.2 1.27 0.192 G G G A A 0.6 0 0 Ϫ2,575 AϾG R462Q GϾA D541E GϾT ϩ1,0721 GϾA ϩ13,861 AϾC * Score test statistic for association between haplotypes and prostate cancer risk. † Empirical P values based on 10,000 replications.

patients from 116 Finnish families with HPC. In one of these ences were observed between patients with familial prostate five families, suggestive evidence of segregation between the cancer or HPC and controls for this variant, irrespective of mutation and prostate cancer was observed. In that study, they choice of genetic model (recessive, multiplicative, or dominant). also compared the frequency of E265X mutations in unselected This missense variant was significantly associated with an in- prostate cancer cases (n ϭ 492) and control individuals (n ϭ creased prostate cancer risk in a study from Japan based on 101 566), without finding any significant association. In a study of familial prostate cancer cases and 105 controls (19), yet other 95 men with HPC, originating from 75 unrelated families, Chen studies have failed to find an association between this variant et al. (32) found the E265X mutation in one family, with two of and prostate cancer risk (17, 18, 20). three affected brothers carrying the mutation. From these re- Haplotype analysis of the five SNPs provided weak support sults, it is clear that only a small fraction of prostate cancer for RNASEL as a risk factor for prostate cancer. Global tests for families can be explained by segregation of the E265X variant. association between haplotypes and prostate cancer risk were all We observed virtually identical frequencies of E265X mu- nonsignificant; however, two specific haplotypes were signifi- tation carriers among sporadic prostate cancer cases and control cantly associated with prostate cancer risk. For both haplotypes, individuals (1.9% in both groups). Among familial prostate higher frequency was observed among controls compared with cancer and HPC patients, expected to be enriched for genetic prostate cancer cases, indicating protective effects. It is there- variants associated with prostate cancer risk, only 5 (1.4%) of fore puzzling that these haplotypes were completely discordant 350 patients carried the E265X mutation; 4 of these had HPC for each of the five SNPs, and, as discussed above, due to the and were members of three unrelated families. None of these many statistical tests performed, the chance of false positive families provided evidence for cosegregation between the mu- findings has to be considered. tation and disease status; in the only family with DNA samples Our study design has a certain limitation. There is the available from other family members, two of five affected possibility of the existence of other common variants in brothers were mutation carriers. Therefore, our results provide RNASEL that are not well represented by the five SNPs geno- no support for the E265X mutation being a high-penetrance risk typed in the present study. We used the public database National variant for prostate cancer. Furthermore, the complete lack of Center for Biotechnology Information for selection of SNPs, association between E265X and prostate cancer risk in our study which is suboptimal compared with sequencing an adequate provided no evidence for E265X being a modifier variant for number of individuals from the study population to uncover all prostate cancer risk. existing variants. However, considering the strong linkage dis- The common missense mutation R462Q has been implicated in prostate cancer risk both from epidemiologic and func- equilibrium between the studied SNPs, we believe that it is most tional studies. Using 423 prostate cancer cases and 454 sibling likely that we were able to captured the major part of the genetic controls, Casey et al. (18) estimated that heterozygous carriers variation in the RNASEL gene in our study. Furthermore, be- of this variant have 50% greater risk of prostate cancer than cause of the large number of individuals in this study, the noncarriers, and homozygous carriers have more than double the possibility that our results represent false negative findings is of risk. They also reported a reduction in enzymatic activity of the minimal concern. For example, comparing sporadic prostate variant to one-third of normal activity. Additional functional cancer patients with controls, using the genotype frequencies evidence for this variants role in prostate cancer development observed for the R462Q variant among the controls and assum- comes from a recent study observing that the R462Q variant ing a multiplicative genetic model, we have Ͼ87% power to reduced the ability of RNase L to cause apoptosis in response to detect an odds ratio of 1.5 for homozygotes at a significance activation by 2-5A (21). Ro¨kman et al. (17) reported moderate level of 5% (two-sided test). evidence for an increased risk of prostate cancer for homozy- In summary, we failed to support a role of the rare trun- gous carriers of the R462Q variant in 66 index patients from cating mutation E265X in Swedish prostate cancer. In addition, HPC families. In addition, Wang et al. (20) reported a signifi- analyses of common sequence variants, both individually and cant association between this variant and prostate cancer risk through haplotypes, provided limited evidence for association among 473 affected men from 181 families; however, they with prostate cancer risk. Considering the large and well-char- found an opposite trend with odds ratios of 0.8 for heterozygotes acterized study population and the high quality in genotyping, and 0.5 for homozygotes. these results provide strong evidence against a role of RNASEL Despite the functional and epidemiologic findings support- in prostate cancer etiology in Sweden. ing a role for R462Q in prostate cancer susceptibility, no significant associations between this variant and prostate cancer risk was observed in our study. Nor did we find any significant ACKNOWLEDGMENTS association with prostate cancer risk for any of the other four We thank all study participants in the Cancer Prostate in Sweden common sequence variants studied, except for the missense study and all family members who participated in this study, Ulrika mutation D541E, which under a recessive genetic model, pro- Lund for skillfully coordinating the study center at Karolinska Institute, all urologists, including their patients in the Cancer Prostate in Sweden vided marginally significant evidence for association with spo- ϭ study, and all urologists providing clinical data to the National Registry radic prostate cancer (P 0.03), with an estimated odds ratio of of Prostate Cancer. We also thank Karin Andersson, Susan Lindh, 0.77 (95% confidence interval, 0.59–1.00) for homozygous Gabriella Thore´n Berglund, and Margareta Åswa¨rd at the Regional carriers compared with homozygous noncarriers. However, be- Cancer Registries. In addition, we thank So¨ren Holmgren and the cause of the large number of tests conducted in our study, this personnel at the Medical Biobank in Umeå for skilfully handling the could represent a false positive finding. Moreover, no differ- blood samples.

REFERENCES 16. Carpten J, Nupponen N, Isaacs S, et al. Germline mutations in the ribonuclease L gene in families showing linkage with HPC1. Nat Genet 1. Easton DF, Schaid DJ, Whittemore AS, Isaacs WJ. Where are the 2002;30(2):181–4. prostate cancer genes? A summary of eight genome wide searches. Prostate 2003;57(4):261–9. 17. Ro¨kman A, Ikonen T, Seppa¨la¨ EH, et al. Germline alterations of the RNASEL gene, a candidate HPC1 gene at 1q25, in patients and families 2. Smith JR, Freije D, Carpten JD, et al. Major susceptibility locus for with prostate cancer. Am J Hum Genet 2002;70(5):1299–304. prostate cancer on chromosome 1 suggested by a genome-wide search. 18. Casey G, Neville PJ, Plummer SJ, et al. RNASEL Arg462Gln Science (Wash. DC) 1996;274(5291):1371–4. variant is implicated in up to 13% of prostate cancer cases. Nat Genet 3. Cooney KA, McCarthy JD, Lange E, et al. Prostate cancer suscep- 2002;32(4):581–3. tibility locus on chromosome 1q: a confirmatory study. J Natl Cancer 19. Nakazato H, Suzuki K, Matsui H, Ohtake N, Nakata S, Yamanaka Inst (Bethesda) 1997;89(13):955–9. H. Role of genetic polymorphisms of the RNASEL gene on familial 4. Goddard KA, Witte JS, Suarez BK, Catalona WJ, Olson JM. Model- prostate cancer risk in a Japanese population. Br J Cancer 2003;89(4): free linkage analysis with covariates confirms linkage of prostate cancer 691–6. to chromosomes 1 and 4. Am J Hum Genet 2001;68(5):1197–206. 20. Wang L, McDonnell SK, Elkins DA, et al. Analysis of the RNA- 5. Gro¨nberg H, Smith J, Emanuelsson M, et al. In Swedish families SEL gene in familial and sporadic prostate cancer. Am J Hum Genet with hereditary prostate cancer, linkage to the HPC1 locus on chromo- 2002;71(1):116–23. some 1q24-25 is restricted to families with early-onset prostate cancer. 21. Xiang Y, Wang Z, Murakami J, et al. Effects of RNase L mutations Am J Hum Genet 1999;65(1):134–40. associated with prostate cancer on apoptosis induced by 2Ј,5Ј-oligoad- 6. Neuhausen SL, Farnham JM, Kort E, Tavtigian SV, Skolnick MH, enylates. Cancer Res 2003;63(20):6795–801. Cannon-Albright LA. Prostate cancer susceptibility locus HPC1 in Utah 22. Zheng SL, Augustsson-Ba¨lter K, Chang B, et al. Sequence variants of high-risk pedigrees. Hum Mol Genet 1999;8(13):2437–42. toll-like receptor 4 are associated with prostate cancer risk: results from the Cancer Prostate in Sweden Study. Cancer Res 2004;64(8):2918–22. 7. Berry R, Schroeder JJ, French AJ, et al. Evidence for a prostate cancer-susceptibility locus on chromosome 20. Am J Hum Genet 2000; 23. Stephens M, Smith NJ, Donnelly P. A new statistical method for 67(1):82–91. haplotype reconstruction from population data. Am J Hum Genet 2001; 68(4):978–89. 8. Berthon P, Valeri A, Cohen-Akenine A, et al. Predisposing gene for 24. Howell WM, Jobs M, Gyllensten U, Brookes AJ. Dynamic allele- early-onset prostate cancer, localized on chromosome 1q42.2-43. Am J specific hybridization. A new method for scoring single nucleotide Hum Genet 1998;62(6):1416–24. polymorphisms. Nat Biotechnol 1999;17(1):87–8. 9. Eeles RA, Durocher F, Edwards S, et al. Linkage analysis of chro- 25. Jobs M, Howell WM, Stromqvist L, Mayr T, Brookes AJ. DASH-2: mosome 1q markers in 136 prostate cancer families. The Cancer Re- flexible, low-cost, and high-throughput SNP genotyping by dynamic search Campaign/British Prostate Group U. K. Familial Prostate Cancer allele-specific hybridization on membrane arrays. Genome Res 2003; Study Collaborators. Am J Hum Genet 1998;62(3):653–8. 13(5):916–24. 10. Goode EL, Stanford JL, Chakrabarti L, et al. Linkage analysis of 26. Prince JA, Feuk L, Howell WM, et al. Robust and accurate single 150 high-risk prostate cancer families at 1q24-25. Genet Epidemiol nucleotide polymorphism genotyping by dynamic allele-specific hybrid- 2000;18(3):251–75. ization (DASH): design criteria and assay validation. Genome Res 11. McIndoe RA, Stanford JL, Gibbs M, et al. Linkage analysis of 49 2001;11(1):152–62. high-risk families does not support a common familial prostate cancer- 27. Fredman D, Jobs M, Brookes AJ. DFold: PCR design that mini- susceptibility gene at 1q24-25. Am J Hum Genet 1997;61(2):347–53. mizes secondary structure and optimizes downstream genotyping appli- 12. Xu J. Combined analysis of hereditary prostate cancer linkage to cations. Hum Mutat 2004;24(1):1–8. 1q24-25: results from 772 hereditary prostate cancer families from the 28. Jobs M, Howell WM, Brookes AJ. Creating arrays by centrifuga- International Consortium for Prostate Cancer Genetics. Am J Hum tion. Biotechniques 2002;32(6):1322–9. Genet 2000;66(3):945–57. 29. Weir BS. Genetic data analysis II. Sunderland, United Kingdom: 13. Castelli JC, Hassel BA, Maran A, et al. The role of 2Ј-5Ј oligoad- Sinauer; 1996. enylate-activated ribonuclease L in apoptosis. Cell Death Differ 1998; 30. Wooldridge JM. Econometric analysis of cross section and panel 5(4):313–20. data. Cambridge, MA: Massachusetts Institute of Technology; 2002. 14. Hassel BA, Zhou A, Sotomayor C, Maran A, Silverman RH. A 31. Schaid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA. dominant negative mutant of 2–5A-dependent RNase suppresses antipro- Score tests for association between traits and haplotypes when linkage liferative and antiviral effects of interferon. EMBO J 1993;12(8):3297–304. phase is ambiguous. Am J Hum Genet 2002;70(2):425–34. 15. Lengyel P. Tumor-suppressor genes: news about the interferon 32. Chen H, Griffin AR, Wu YQ, et al. RNASEL mutations in hered- connection. Proc Natl Acad Sci USA 1993;90(13):5893–5. itary prostate cancer. J Med Genet 2003;40(3):e21.

Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2004 American Association for Cancer Research. Genetic Analysis of the RNASEL Gene in Hereditary, Familial, and Sporadic Prostate Cancer

Fredrik Wiklund, Björn-Anders Jonsson, Anthony J. Brookes, et al.

Clin Cancer Res 2004;10:7150-7156.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/10/21/7150

Cited articles This article cites 30 articles, 7 of which you can access for free at: http://clincancerres.aacrjournals.org/content/10/21/7150.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/10/21/7150.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 Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/10/21/7150. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.