We searched for deletions in the germ-line genome among 498 aggressive prostate cancer cases and 494 controls from a population-based study in Sweden [CAncer of the Prostate in Sweden (CAPS)] using Affymetrix SNP arrays. By comparing allele intensities of ∼500,000 SNP probes across the genome, a germ-line deletion at 2p24.3 was observed to be significantly more common in cases (12.63%) than in controls (8.28%); P = 0.028. To confirm the association, we genotyped this germ-line copy number variation (CNV) in additional subjects from CAPS and from Johns Hopkins Hospital (JHH). Overall, among 4,314 cases and 2,176 controls examined, the CNV was significantly associated with prostate cancer risk [odds ratio (OR), 1.25; 95% confidence interval (95% CI), 1.06–1.48; P = 0.009]. More importantly, the association was stronger for aggressive prostate cancer (OR, 1.31; 95% CI, 1.08–1.58; P = 0.006) than for nonaggressive prostate cancer (OR, 1.19; 95% CI, 0.98–1.45; P = 0.08). The biological effect of this germ-line CNV is unknown because no known gene resides in the deletion. Results from this study represent the first novel germ-line CNV that was identified from a genome-wide search and was significantly, but moderately, associated with prostate cancer risk. Additional confirmation of this association and functional studies are warranted. [Cancer Res 2009;69(6):2176–9]
- prostate cancer
A consistent finding of genetic susceptibility to prostate cancer suggests that there are altered germ-line DNA sequences that affect the function and/or expression of specific genes. These germ-line alterations include single nucleotide substitutions (SNP) and deletions/gains of a string of nucleotides such as copy number variations (CNV). Whereas the importance of SNPs and their associations with disease risk are well established, there is an increasing appreciation for a potential role of CNVs in disease risk ( 1– 6). CNVs are found to be common in the genome; however, their associations with risk of common diseases are understudied. This gap is largely due to a lack of high-throughput and cost-effective methods for detecting germ-line CNVs in the genome and for genotyping target CNVs in large numbers of cases and controls.
In this study, we report association findings of prostate cancer risk with a common germ-line CNV at 2p24.3, detected from a genome-wide analysis and confirmed using a simple PCR method among a large number of prostate cancer patients and control subjects from Sweden and the United States.
Materials and Methods
Subjects in the discovery stage of a genome-wide association study. Subjects were a subset of men from CAPS (CAncer of the Prostate in Sweden), a population-based case-control study in Sweden ( 7, 8). Prostate cancer patients were identified and recruited from four of the six regional cancer registries in Sweden. These case subjects were classified as having aggressive disease if they met any of the following criteria: T3/4, N+, M+, Gleason score sum ≥8, or PSA >50 ng/mL; otherwise, they were classified as nonaggressive disease. Control subjects were recruited concurrently with case subjects. We selected 498 aggressive prostate cancer cases and 494 unaffected controls matching the age distribution of cases for the discovery stage of genome-wide association study (GWAS).
Subjects in the confirmation stage. Two study populations were included for the confirmation stage. The first were the remaining subjects from CAPS, including 733 aggressive cases, 1,619 nonaggressive cases, and 1,287 controls from Sweden. The second study population was a hospital-based prostate cancer case-control study of European Americans from Johns Hopkins Hospital (JHH; ref. 9). It included 1,527 cases and 482 controls of European descent (by self-report). We defined more aggressive and less aggressive disease based on tumor stage and Gleason score. Tumors with a Gleason score of 7 or higher or stage pT3 or higher or N+ or M1 (i.e., either high-grade or non–organ-confined disease) were defined as more aggressive. Tumors with a Gleason score of 6 or lower and stage pT2/N0 (i.e., cancer confined to the prostate) were defined as less aggressive. The study received institutional approval and complied with the Health Insurance Portability and Accountability Act regulations. The study received institutional approval.
Genotyping of Germ-Line CNVs
Detection of germ-line deletions in the genome. Germ-line CNVs were inferred from SNP intensity data of the Affymetrix 500K SNP arrays among 498 aggressive prostate cancer cases and 494 controls of CAPS ( 8). The dChipSNP computer software was used to estimate DNA copy number from allele intensity for each of the ∼500,000 SNPs ( 10). We used the following criteria to define a germ-line deletion: (a) DNA copy number ≤1.5 and ≤0.5 for a hemizygous or homozygous deletion, respectively; (b) a homozygous genotype (expected in a hemizygous deletion) or no genotype call (expected in a homozygous deletion); and (c) must involve at least three consecutive SNPs with a minimum physical length of 1 kb. We focused on only germ-line deletions in this study because SNP information can be used to improve the quality of defining deletions.
Confirmation of germ-line deletions using quantitative PCR. Three subjects with the deletion and one subject without the deletion at 2p24.3 were selected for independent confirmation by quantitative real-time PCR using the ABI Prism 7000 Sequence Detection System ( 11).
Mapping breakpoints of the germ-line deletion at 2p24.3. Subjects who were inferred to have the germ-line deletion at 2p24.3 (n = 100) were selected for the breakpoint analysis (Supplementary Fig. S1). Based on the putative breakpoints deduced from the allele intensity of the SNP probes of Affymetrix 500K SNP arrays, PCR primers were designed based on the flanking genomic sequence of the target CNV. We then sequenced the PCR products using dye-terminator chemistry (BigDye, ABI) and determined the exact breakpoints.
Measurement of germ-line deletion at 2p24.3. We used a regular PCR method to measure the germ-line 2p24.3 deletion among all cases and controls from CAPS and JHH. Deleted and wild-type genotypes at the CNV can be detected simultaneously based on the different lengths of PCR products. The primer set is available on request. The same PCR conditions were used as described above.
Tests for Hardy-Weinberg equilibrium were done separately among case patients and control subjects using Fisher's exact test. Haplotype blocks were estimated using a computer program Haploview ( 12), and a default Gabriel method ( 13) was used to define a haplotype block. Differences in the carrier frequencies of the deletion between case patients and control subjects were tested using a χ2 test with 1 degree of freedom (two-sided). Odds ratio (OR) and 95% confidence interval (95% CI) were estimated using an unconditional logistic regression with adjustment for age and geographic region (for CAPS only).
Results and Discussion
We identified 1,740 deletions in the genome that have a frequency of >1% in 498 prostate cancer cases and 494 controls. We then compared the frequencies of these deletions in cases and controls (Supplementary Fig. S2) and identified 76 unique putative deleted regions that were associated with prostate cancer risk (P < 0.05). To reduce the measurement errors in defining DNA copy numbers and to identify a subset of putative deletions with high confidence to be true CNVs, we examined the distributions of DNA copy numbers among the subjects. This second step identified seven independent regions where three groups of subjects carrying 2, 1, and 0 copies were distinguishable with high confidence. To rule out artifactual copy number changes due to technical issues (e.g., mutations in restriction enzyme sites of Sty or Nsp used in the Affymetrix SNP array protocol can lead to reduced DNA copy number), we developed a quantitative real-time PCR assay as an independent molecular method to further confirm the deletions. One region at 2p24.3 was selected for a quantitative real-time PCR analysis because it was the only deletion that had frequency >5% in the controls. The concordance in detecting the germ-line deletion using the SNP array and the quantitative real-time PCR analysis was 100% in three subjects with the deletion and one subject without the deletion. Among the subjects in the GWAS whose DNA copy number could be reliably determined, the deletion was found in 60 of 475 (12.63%) cases and 40 of 483 (8.28%) controls. The difference was statistically significantly, with a nominal P value of 0.028.
This germ-line deletion at 2p24.3 involves six consecutive SNPs in the Affymetrix 500K SNP arrays (one on Nsp and five on Sty arrays) and spans 3,670 bp between rs16861868 (14,656,949 bp) and rs2380618 (14,660,690; Fig. 1A ). It overlaps with a known CNV region [locus 166 of the Database of Genomic Variants ( 2) and CNP119 of the UCSC Genome Browser on Human May 2004 Assembly ( 14– 16)]. However, the reported frequencies from these studies, ranging from 0.1% to 6%, are considerably lower than what we observed in our study. This discrepancy could be due to different genotyping platforms and the criteria used to define CNV in different studies, as well as the geographic differences of study subjects ( 17). It is also interesting to note that none of the six SNPs was associated with prostate cancer risk (P > 0.05) in the subjects of our CAPS GWAS study.
To map the deletion boundaries, we designed a set of PCR primers flanking the putative breakpoints deduced from the allele intensity of the SNP probes of Affymetrix 500K SNP arrays. Genomic DNA samples from 100 subjects who carried the deletion CNV were amplified using these PCR primers and the resulting fragments were directly sequenced. We found that all 100 subjects carried the same 5,947-bp nucleotide deletion between 14,662,627 and 14,668,573 bp ( Fig. 1B). Because of the uniform boundaries of the CNV, we were able to design a single set of primers to amplify two possible different lengths of PCR fragments, one representing the deleted allele and the other representing the wild-type (nondeleted) allele, in a single reaction to identify DNA samples with homozygous deletion, hemizygous deletion, or no deletion (marked by red, purple, and blue arrows, respectively, in Fig. 1C).
We then used this PCR method to genotype this germ-line deletion among the remaining subjects from CAPS (2,386 cases and 1,213 controls) and all study subjects from JHH (1,453 cases and 480 controls). Within the CAPS and JHH study populations, the germ-line deletion was in Hardy-Weinberg equilibrium among controls and cases (P ≥ 0.05). Whereas the germ-line deletion was higher in both sets of cases than in controls, the difference was only statistically significant in the JHH population (P = 0.03, two-sided; Table 1 ). When all 4,314 cases and 2,176 controls were examined, the deletion was significantly associated with prostate cancer risk (OR, 1.25; 95% CI, 1.06–1.48; P = 0.009). Importantly, a stronger association of the germ-line deletion with prostate cancer risk was found among aggressive cases (OR, 1.31; 95% CI, 1.08–1.58; P = 0.006) than among nonaggressive cases (OR, 1.19; 95% CI, 0.98–1.45; P = 0.08) in the combined data from CAPS and JHH. However, the difference in ORs was not statistically significant (P = 0.33).
The functional effect of this CNV is not clear because no known gene resides in the ∼6-kb deletion. However, there are only two genes, AY928977 and FAM84A, located within 100 kb of this CNV. AY928977 is a spliced fusion transcript between the first 14 exons of the anthrax toxin receptor I gene (ANTX1) and novel neuroblastoma gene 1 (NNG1; ref. 18). However, the ANTX1-NNG1 fusion transcript is less likely to play an essential role in carcinogenesis due to its limited expression in only one neuroblastoma cell line and its sublines but not in 70 primary neuroblastoma tumors ( 18). FAM84A is a component of extracellular matrices and has been found to be up-regulated in several tumors ( 19) including prostate cancer. Notably, exogenous FAM84A expression was found to increase cell motility in NIH3T3 cells and may contribute to metastasis ( 19). The effect of this CNV on the expression of FAM84A is unclear because it is not within the proximal regulatory promoter of FAM84A. However, it is likely that uncharacterized distal regulatory elements that either activate or suppress FAM84A expression exist within this CNV. Additionally, the existence of yet unidentified small regulatory RNAs in this CNV is another testable hypothesis. Therefore, despite the lack of direct functional evidence of the involvement of this CNV, there remains the potential for this CNV to affect prostate carcinogenesis.
Although we report one of the first germ-line CNVs associated with prostate cancer risk in general populations, we acknowledge that this study has several important limitations. First, the overall statistical evidence for the prostate cancer association is moderate. Therefore, it remains possible that the association finding is due to chance alone. Second, the observed association results could be affected by measurement error because the PCR method used to measure the germ-line deletion has not been extensively evaluated. However, when measuring the germ-line deletion, we observed an excellent concordance rate (98%) between the PCR method and the Affymetrix SNP arrays among ∼1,000 subjects, providing some assurance. Finally, our study on global germ-line CNVs is incomplete because only a small fraction of germ-line deletions in the genome were systematically evaluated among a small number of case and control subjects. Additional genome-wide studies on germ-line CNVs using higher-resolution SNP arrays among a large number of cases and controls are needed.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: National Cancer Institute grants CA129684, CA105055, CA106523, and CA95052 (J. Xu) and grants CA112517 and CA58236 (W.B. Isaacs); Department of Defense grant PC051264 (J. Xu); Swedish Cancer Society (Cancerfonden; to H. Grönberg); and Swedish Academy of Sciences (Vetenskapsrådet; to H. Grönberg).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank all the study subjects who participated in the CAPS study and the urologists who included their patients in the CAPS study. We acknowledge the contribution of multiple physicians and researchers in designing and recruiting study subjects, including Dr. Hans-Olov Adami (for CAPS) and Drs. Bruce J. Trock, Alan W. Partin, and Patrick C. Walsh (for JHH).
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
W. Liu and J. Sun contributed equally to the study.
- Received August 14, 2008.
- Revision received November 13, 2008.
- Accepted January 30, 2009.
- ©2009 American Association for Cancer Research.