Association of common variations of 8q24 with the risk of prostate cancer in Koreans and a review of the Asian population
J.Y.J. and S.P. contributed equally to this study.
Kang Hyun Lee, Center for Prostate Cancer, National Cancer Center, Goyang 410-769, Korea. e-mail: firstname.lastname@example.org
What's known on the subject? and What does the study add?
The association between subjects with the genetic variation of 8q24 and the risk of development of prostate cancer in Korean men was found. As a result of haplotype analysis, [AGC] and [CTA] carriers showed a significant association with prostate cancer risk. This is clinically meaningful as an initial study on genetic susceptibility to prostate cancer in Korean men and the first report of 8q24 haplotypes in an Asian population.
- • To determine the association between genetic variation of 8q24 with prostate cancer risk in Korean men.
PATIENTS AND METHODS
- • With a hospital-based case-control study design, we enrolled 194 patients with prostate cancer and 169 healthy controls from visitors for cancer screening.
- • DNA samples were obtained from peripheral blood for the analysis of single nucleotide polymorphisms (SNPs). Three SNPs of 8q24, including rs16901979, rs6983267, and rs1447295, were genotyped on cases and controls.
- • The subjects with the rs1447295 CA or AA genotype had a higher risk of prostate cancer than the CC genotype.
- • The A allele at SNP rs1447295 was associated with the incidence of prostate cancer.
- • The rs16901979 CA genotype carriers had a higher risk of prostate cancer than the CC genotype.
- • Individuals with the [AGC] and [CTA] haplotypes had a significantly increased risk of prostate cancer compared with the [CTC] haplotype ([AGC] with adjusted odds ratio [OR] 1.79; 95% confidence interval [CI] 1.09–2.96; P= 0.022; [CTA] with adjusted OR 5.17; 95% CI 2.40–11.15; P < 0.001).
- • The genetic variation of 8q24 is associated with the risk of prostate cancer in Korean men.
- • Individuals with the [AGC] and [CTA] haplotypes had a significant association with prostate cancer risk.
genome-wide association study
single nucleotide polymorphism
generalised multifactor dimensionality reduction
Prostate cancer is the most common cancer among men in most Western populations, with the exception of skin cancer, with 192 280 new cases expected in 2009 in the USA . Generally, it is thought that the incidence of prostate cancer is lower in Asian countries ; however, the occurrence of prostate cancer in Korea is steadily increasing and the incidental detection of prostate cancer is not uncommon. The incidence of prostate cancer had one of the highest rates of increase (12.6% annual increase) among all cancers in Korea [3–5]. Prostate cancer is a very heterogenous disease. The aetiology of prostate cancer remains largely unknown, but genetic and environmental factors are likely to be involved. Increasing age, high dietary fat intake, and cigarette smoking are considered the common risk factors [6,7]; however, only a small percentage of individuals exposed to these risk factors develop prostate cancer. Predisposing genetic factors probably contribute to such discrepancies, together with a combination of environmental risk factors. Based on the accumulating epidemiological and genetic evidence, genetic variation is a significant factor in determining the risk of prostate cancer. A genome-wide association study (GWAS) reported that genetic polymorphisms of several genes at 8q24 have a significant association with prostate cancer in different races [8–17]. Using genetic linkage analysis to identify the prostate cancer susceptibility locus, three single nucleotide polymorphisms (SNPs), including rs1447295, rs6983267, and rs16901979, have been suggested as potential SNPs within 8q24 that may contribute to disease risk [9,11,18,19]. However, polymorphisms of 8q24 have not been studied in Korean men. Furthermore, there are no reports of an association between the 8q24 haplotype and prostate cancer risk. We conducted the presents study to determine the association between genetic variations of 8q24 with prostate cancer risk in Korean men.
PATIENTS AND METHODS
This study was designed as a hospital-based case-control study. Between January 2005 and February 2009, we enrolled 194 consecutive patients who were newly diagnosed with histologically confirmed prostate cancer at the National Cancer Center in Korea. The healthy controls registered at the National Cancer Center for cancer screening. All 169 controls had no evidence of a malignancy and no history of cancer. We excluded the possibility of prostate cancer in all controls based on the PSA level and DRE. If the PSA level was persistently elevated >4 ng/mL, a prostate needle biopsy was performed. All patients underwent radical prostatectomy for the treatment of prostate cancer after prostate cancer was diagnosed via needle biopsy. Gleason grading was determined by one experienced pathologist. All pathological analyses were based on the findings in the prostatectomy specimens.
The purpose of and requirements for this study were clearly explained to all study subjects who signed informed consent forms when they agreed to participate. The study subjects voluntarily donated blood, and the blood samples were registered and stored at −70 °C on-site for the following studies. Information on demographic characteristics was obtained from personal questionnaires. Both the patients and controls were of the same ethnic origin (Korean). This study was approved by the Institutional Review Board of the National Cancer Center in Korea (IRB No. NCCNCS05-049).
DNA samples from 5 mL of peripheral blood were prepared using a QIAamp Blood Kit (QIAGEN, Germany) according to the manufacturer's instructions. We selected three SNPs (rs16901979, rs6983267, and rs1447295) and genotyping analyses were performed using the SEQUENOM Mass Array system. The basic information of the three SNPs at 8q24 is shown in Supplemental Table S1, including primers, primer sequences, and restriction enzymes. Genotyping was performed using the iPLEX Gold assay on the Mass ARRAY® platform (Sequenom, USA) based on MALDI-TOF spectrometry by the manufacturer's instructions, and the resulting genotype data were collected by Typer v4.0 .
Demographic and clinical characteristics in cases and controls were compared using a chi-square test for categorical variables and an independent t-test for continuous variables. The Hardy-Weinberg equilibrium (HWE) was separately tested for random mating in cases and controls for each SNP. The association between 8q24 polymorphisms and prostate cancer risk was analysed for genotypic frequencies of each SNP using an unconditional logistic regression model. Additionally, to assess for the presence of an association for the linear trend in proportion by the number of variant alleles in each SNP, we used the Cochran-Armitage test for trend, which is a commonly used genotype-based test for case-control genetic association studies. The cumulative effect of combined genotypes on prostate cancer risk was estimated by counting the number of genotypes associated with prostate cancer on the basis of the best-fitting genetic inheritance from single SNP analysis.
The haplotype estimation was carried out using the PHASE version 2.1 program (http://www.stat.washington.edu/stephens/software.html), which implements a Bayesian statistical method for reconstructing haplotypes. Because of the uncertainty of estimated haplotype pairs, we used the pair with the highest probability for each individual. The association between 8q24 haplotypes and prostate cancer risk was analysed using an unconditional logistic regression in two ways: considering the haplotypic frequencies of each haplotype; and considering the haplotype-pair of each person. Additionally, the associations between genotypes/haplotypes and clinical parameters, e.g. PSA level, Gleason score, and pathological stage, were investigated using an unconditional logistic regression. The odds ratios (ORs) and 95% CIs were estimated. We used a multivariable logistic regression model to adjust for age as a potential confounding factor for prostate cancer risk. We used the generalised multifactor dimensionality reduction (GMDR) method, which permits the adjustment for discrete and quantitative covariates with unbalanced case-control studies and overcomes the limitation arising from a relatively small sample size . GMDR software and GMDR permutation perl script (http://www.healthsystem.virginia.edu/internet/Addiction-Genomics) were used for this purpose. Empirical P-values were calculated via 5000 permuted datasets. All other statistical analyses were tested using SAS (version 9.1; SAS Institute, Inc., Cary, NC, USA). The reported P-values are two-sided and a P < 0.05 was considered to indicate statistical significance.
CHARACTERISTICS OF THE STUDY POPULATION
The features of 194 patients with prostate cancer and 169 healthy controls are given in Table 1. For age, PSA level, and alcohol consumption status, there were significant differences between the cases and controls. Patients with prostate cancer were older, had higher PSA levels, and were more likely to be current consumers of alcohol. In contrast, there were no significant differences for smoking status, body mass index, family history of prostate cancer, and a medical history of hypertension (Table 1).
Table 1. Demographic and clinical characteristics of study subjects
|Mean (sd; median):|| || || |
| Age, years||66.52 (7.16; 68.0)||60.13 (3.30; 61.0)||<0.001|
| Serum PSA level, ng/mL||21.28 (28.15; 9.8)||1.47 (1.63; 1.0)||<0.001|
|N (%):|| || || |
| Serum PSA level, ng/mL|| || ||<0.001|
| <4||11 (5.7)||160 (94.7)|| |
| ≥4 to <10||87 (44.9)||8 (4.7)|| |
| ≥10 to <20||43 (22.2)||1 (0.6)|| |
| ≥20||53 (27.3)||0 (0)|| |
|Mean (sd; median):|| || || |
| BMI, kg/m2||24.35 (2.60; 24.1)||24.42 (2.34; 24.2)||0.800|
|N (%):|| || || |
| BMI, kg/m2:|| || ||0.388|
| <25||127 (65.5)||102 (61.1)|| |
| ≥25||67 (34.5)||65 (38.9)|| |
|Mean (sd; median):|| || || |
| Smoking, pack-years||21.93 (21.75; 18.3)||19.24 (20.60; 14.9)||0.264|
|N (%):|| || || |
| Smoking status:|| || ||0.672|
| 1. Never||52 (26.8)||42 (24.9)|| |
| 2. Yes||142 (73.2)||127 (75.2)|| |
| Drinking status:|| || ||0.002|
| 1. Never||68 (35.1)||34 (20.1)|| |
| 2. Yes||126 (65.0)||135 (79.9)|| |
| Hypertension:|| || ||0.502|
| 1. No||120 (61.9)||109 (65.3)|| |
| 2. Yes||74 (38.1)||58 (34.7)|| |
| Family history of prostate cancer:|| || ||0.132|
| 1. No||186 (95.9)||146 (98.7)|| |
| 2. Yes||8 (4.1)||2 (1.4)|| |
| Gleason score:|| || || |
| 2–6||99 (51.0)||–|| |
| 7||62 (32.0)||–|| |
| 8–10||33 (17.0)||–|| |
| pT stage:|| || || |
| ≤pT2b||40 (20.7)||–|| |
| pT2c||88 (45.4)||–|| |
| pT3||50 (25.8)||–|| |
| N+||10 (5.2)||–|| |
| pT0||6 (3.1)||–|| |
DISTRIBUTION OF ALLELES AND GENOTYPES OF 8Q24 POLYMORPHISMS
The distribution of 8q24 alleles and genotypes, and the association with the development of prostate cancer are shown in Table 2. The allele frequency analysis showed that the A allele of rs1447295 was more likely to be present in patients with prostate cancer than in controls (adjusted OR 2.02; 95% CI 1.28–3.21; P= 0.003). The genotype frequency analysis showed that two SNPs (rs1447295 and rs16901979) had a statistically significant association with the risk of prostate cancer. The subjects with the rs1447295 CA genotype had a higher risk of prostate cancer than the CC genotype (adjusted OR 2.14; 95% CI 1.23–3.72; P= 0.007). The individuals with the rs1447295 CA or AA genotype also had a higher prostate cancer risk (adjusted OR 2.22; 95% CI 1.31–3.79; P= 0.003). These data indicate that the A allele at SNP rs1447295 is strongly associated with the incidence of prostate cancer (Cochran-Armitage trend test P < 0.001). The subjects with the rs16901979CA genotype had a higher risk of prostate cancer than the CC genotype (adjusted OR 1.69; 95% CI 1.01–2.81; P= 0.045). Patients with the rs16901979 CA or AA genotype had a difference in risk, but not a statistically significant difference (adjusted OR 1.56; 95% CI 0.96–2.52; P= 0.073). For the rs6983267 SNP, the association with prostate cancer risk did not reach statistical significance.
Table 2. Frequency of 8q24 polymorphisms and association with prostate cancer risk
|rs16901979||C/C||99 (51.0)||100 (59.2)||1.00|| ||1.00|| ||0.211|
|C/A||81 (41.8)||57 (33.7)||1.44 (0.93–2.23)||0.106||1.69 (1.01–2.81)||0.045|| |
|A/A||14 (7.2)||12 (7.1)||1.18 (0.52–2.68)||0.695||1.04 (0.40–2.66)||0.939|| |
|C/A + A/A||95 (49.0)||69 (40.8)||1.39 (0.92–2.11)||0.121||1.56 (0.96–2.52)||0.073|| |
|C||279 (71.9)||257 (76.0)||1.00|| ||1.00|| || |
|A||109 (28.1)||81 (24.0)||1.24 (0.89–1.73)||0.207||1.29(0.88–1.90)||0.197|| |
|rs6983267||T/T||56 (28.9)||51 (30.3)||1.00|| ||1.00|| ||0.366|
|G/T||92 (47.4)||86 (51.2)||0.97 (0.60–1.57)||0.915||0.93 (0.53–1.62)||0.799|| |
|G/G||46 (23.7)||31 (18.5)||1.35 (0.75–2.45)||0.319||1.55 (0.78–3.07)||0.210|| |
|G/T + G/G||138 (71.1)||117 (69.6)||1.07 (0.68–1.69)||0.756||1.08 (0.64–1.83)||0.761|| |
|T||204 (52.6)||188 (56.0)||1.00|| ||1.00|| || |
|G||184 (47.4)||148 (44.0)||1.15 (0.85–1.54)||0.363||1.22 (0.87–1.70)||0.260|| |
|rs1447295||C/C||114 (59.1)||127 (75.6)||1.00|| ||1.00|| ||<0.001|
|C/A||67 (34.7)||38 (22.6)||1.96 (1.23–3.15)||0.005||2.14 (1.23–3.72)||0.007|| |
|A/A||12 (6.2)||3 (1.8)||4.46 (1.23–16.19)||0.023||3.09 (0.72–13.28)||0.130|| |
|C/A + A/A||79 (40.9)||41 (24.4)||2.15 (1.36–3.38)||0.001||2.22 (1.31–3.79)||0.003|| |
|C||295 (76.4)||292 (86.9)||1.00|| ||1.00|| || |
|A||91 (23.6)||44 (13.1)||2.05 (1.38–3.04)||<0.001||2.02 (1.28–3.21)||0.003|| |
HAPLOTYPE ANALYSIS FOR 8Q24 GENE POLYMORPHISMS
Table 3 shows the frequency distribution and association analysis of haplotypes for SNPs in 8q24. In comparison with the [CTC] haplotype consisting of three SNPs (rs16901979, rs6983267 and rs1447295), an individual with the [AGC] haplotype had a significant risk of prostate cancer (adjusted OR 1.79; 95% CI 1.09–2.96; P= 0.022). In addition, subjects with the [CTA] haplotype had a higher risk of prostate cancer than subjects with the [CTC] haplotype (adjusted OR 5.17; 95% CI 2.40–11.15; P < 0.001). In addition, subjects with a copy of the [CTC] haplotype had a decreased risk of prostate cancer (adjusted OR 0.52; 95% CI 0.36–0.77; trend test, P < 0.001), and subjects carrying a copy of the [CTA] haplotype were shown to have a four-fold increased risk of prostate cancer compared with those who did not carry a copy of the [CTA] haplotype (adjusted OR 4.16; 95% CI 1.93–8.96; trend test, P < 0.001). When haplotypes consisted of two SNPs, including rs16901979 and rs1447295, the analysis results were similar to those involving three SNPs; the subjects with the [CA] haplotype had a higher risk of prostate cancer than the those with the [CC] haplotype (adjusted OR 2.17; 95% CI 1.33–3.55; P= 0.002; Table 3).
Table 3. Frequency of 8q24 haplotype and association with prostate cancer risk
|rs16901979 rs6983267 rs1447295|| || || || || || |
| 3–1 [CTC]||124 (32.0)||149 (44.1)||1.00|| ||1.00|| |
| 3–2 [CGC]||75 (19.3)||66 (19.5)||1.37 (0.91–2.05)||0.134||1.43 (0.90–2.28)||0.130|
| 3–3 [ATC]||31 (8.0)||25 (7.4)||1.49 (0.84–2.66)||0.177||1.36 (0.70–2.63)||0.363|
| 3–4 [AGC]||67 (17.3)||54 (16.0)||1.49 (0.97–2.29)||0.069||1.79 (1.09–2.96)||0.022|
| 3–5 [CTA ]||44 (11.3)||14 (4.1)||3.78 (1.98–7.21)||<0.001||5.17 (2.40–11.15)||<0.001|
| 3–6 [CGA]||36 (9.3)||28 (8.3)||1.55 (0.89–2.67)||0.120||1.42 (0.74–2.69)||0.289|
| 3–7 [ATA]||5 (1.3)||1 (0.3)||6.01 (0.69–52.09)||0.104||3.60 (0.33–38.98)||0.293|
| 3–8 [AGA]||6 (1.6)||1 (0.3)||7.21 (0.86–60.64)||0.069||5.63 (0.47–67.67)||0.173|
|rs16901979 rs1447295|| || || || || || |
| 2–1 [CC]||199 (51.29)||215 (63.61)||1.00|| ||1.00|| |
| 2–2 [AC]||98 (25.26)||79 (23.37)||1.34 (0.94–1.91)||0.105||1.45 (0.96–2.17)||0.076|
| 2–3 [CA]||80 (20.62)||42 (12.43)||2.06 (1.35–3.13)||<0.001||2.17 (1.33–3.55)||0.002|
| 2–4 [AA]||11 (2.84)||2 (0.59)||5.94 (1.30–27.13)||0.022||3.97 (0.71–22.32)||0.117|
|Haplotype pair|| || || || || || |
| 3–1 [C T C]|| || || || || || |
| 0||82 (42.3)||50 (29.6)||1.00|| ||1.00|| |
| 1 copy||100 (51.5)||89 (52.7)||0.69 (0.44–1.08)||0.102||0.60 (0.35–1.02)||0.057|
| 2 copies||12 (6.2)||30 (17.8)||0.24 (0.11–0.52)||<0.001||0.24 (0.10–0.56)||0.001|
| 1 or 2 copies||112 (57.7)||119 (70.4)||0.57 (0.37–0.89)||0.013||0.51 (0.31–0.84)||0.009|
| Trend†|| || ||0.55 (0.40–0.77)||<0.001||0.52 (0.36–0.77)||<0.001|
| 3–5 [C T A]|| || || || || || |
| 0||152 (78.4)||155 (91.7)||1.00|| ||1.00|| |
| 1 copy||40 (20.6)||14 (8.3)||2.91 (1.52–5.57)||0.001||4.03 (1.84–8.82)||<0.001|
| 2 copies||2 (1.0)||0||–|| ||–|| |
| 1 or 2 copies||42 (21.6)||14 (8.3)||3.06 (1.61–5.83)||<0.001||4.23 (1.94–9.21)||<0.001|
| Trend†|| || ||3.04 (1.61–5.71)||<0.001||4.16 (1.93–8.96)||<0.001|
| 2–1[C C]|| || || || || || |
| 0||46 (23.7)||25 (14.8)||1.00|| ||1.00|| |
| 1 copy||97 (50.0)||73 (43.2)||0.72 (0.41–1.28)||0.266||0.68 (0.35–1.33)||0.263|
| 2 copies||51 (26.3)||71 (42.0)||0.39 (0.21–0.72)||0.002||0.37 (0.18–0.74)||0.005|
| 1 or 2 copies||148 (76.3)||144 (45.2)||0.56 (0.33–0.96)||0.034||0.52 (0.28–0.97)||0.040|
| Trend†|| || ||0.61 (0.45–0.82)||0.001||0.59 (0.42–0.84)||0.003|
| 2–3 [C A]|| || || || || || |
| 0||120 (61.9)||129 (76.3)||1.00|| ||1.00|| |
| 1 copy||68 (35.0)||38 (22.5)||1.92 (1.20–3.07)||0.006||2.16 (1.24–3.77)||0.007|
| 2 copies||6 (3.1)||2 (1.2)||3.23 (0.64–16.29)||0.157||2.19 (0.39–12.43)||0.376|
| 1 or 2 copies||74 (38.1)||40 (23.7)||1.99 (1.26–3.14)||0.003||2.16 (1.26–3.71)||0.005|
| Trend†|| || ||1.89 (1.24–2.89)||0.003||1.97 (1.20–3.23)||0.007|
CUMULATIVE EFFECT OF ‘AT-RISK’ GENOTYPE OF SNPS
To evaluate the possible cumulative effects of the risk polymorphisms defined in the single SNP analysis, we used an age-adjusted multivariable logistic regression model on combinations of ‘at-risk’ genotypes. Two genotypes (CA of rs16901979 and CA or AA of rs14477295) that showed a significant association with prostate cancer risk were included in the analysis of cumulative effect compared with the ‘no-risk’ genotype reference. The combination of these two risk genotypes not in linkage disequilibrium had a significant association with prostate cancer, and a two-fold increase in risk was noted in subjects with more ‘at-risk’ genotypes (adjusted OR 2.11; 95% CI 1.42–3.12, trend test, P < 0.001; Table 4). In particular, a man with risk genotypes in both rs16901979 and rs14477295 appeared to have almost five-times the risk of prostate cancer than a man with no risk genotype (adjusted OR 4.92; 95% CI 1.93–12.58, trend test, P < 0.001; Table 4).
Table 4. Combined effect of polymorphisms in 8q24 (rs16901979<C/A> and rs1447295<C/A, A/A>) on prostate cancer risk
|0||59 (30.6)||82 (48.8)||1.00|| ||1.00|| |
|1||108 (56.0)||75 (44.6)||2.00 (1.28–3.13)||0.002||1.97 (1.18–3.29)||0.009|
|2||26 (13.5)||11 (6.6)||3.28 (1.51–7.17)||0.003||4.92 (1.93–12.58)||<0.001|
|Trend†|| || ||1.89 (1.34–2.65)||<0.001||2.11 (1.42–3.12)||<0.001|
CROSS-VALIDATION OF MAIN EFFECTS
When the GMDR modelling was performed on rs1447295 and haplotype [CTA] as the best single factor, the permutation testing P-values were 0.004 and <0.001, respectively (Supplemental Table S2).
ASSOCIATION OF THE GENOTYPES/HAPLOTYPES WITH PSA LEVEL, GLEASON SCORE, AND PATHOLOGICAL STAGE OF PROSTATE CANCER
We also determined the association of SNPs with clinical factors, including PSA level, Gleason score, and pathological stage (pT stage). We used the Gleason score from needle biopsies because 47 patients had received temporary androgen ablation therapy before surgery. For association analysis, the patients were divided into two subgroups based on their tumour Gleason score from needle biopsy (Gleason score < 7 and ≥7). The PSA level was analysed as a categorical variable in two PSA categories (<10 and >10 ng/mL). Patients were divided into two subgroups depending on the pathological tumour stage (≤pT2b and ≥pT2c). For the PSA level, subjects with rs1447295 CA or AA genotypes had significantly higher PSA levels (>10 ng/mL) at the time of diagnosis than those with the CC genotype (adjusted OR 2.00; 95% CI 1.12–3.59; P= 0.020). None of the three SNPs were significantly associated with pT stage and Gleason score (Supplemental Table S3). However, when the haplotypes of these three SNPs were considered, patients with a copy of the [AGC] haplotype had higher pathological tumour stages (with [AGC] adjusted OR 2.49; 95% CI 1.11–5.56; P= 0.026) and patients with the [CTA] haplotype had higher PSA levels (with [CTA] adjusted OR 2.55; 95% CI 1.24–5.25; P= 0.011: Supplemental Table S4).
ASSOCIATION OF COMMON 8Q24 VARIATIONS WITH PROSTATE CANCER IN ASIAN POPULATION
Table 5 [14,16,22–26] shows the summary of the previously published studies that examined 8q24 variations with prostate cancer in Asian populations. While the prostate cancer incidence is low in Asian countries, several studies have shown significant genetic association between 8q24 and prostate cancer risk. Detailed results are further described in the Discussion.
Table 5. Association of common 8q24 variations with prostate cancer in Asian populations
|Terada et al. ||Native Japanese||rs1447295||AA||NA||1.23||0.086||A||21.9/18.6||1.28 (1.01–1.61)||0.041|
|Liu et al. ||Native Japanese||rs1447295||CA + AA||44.5/32.53||1.25 (1.07–2.11)||0.02||A||25.4/18.7||1.23* (1.01–1.79)||0.04|
|Liu et al. ||Native Japanese||rs6983561||CA + CC||NA||1.615 (1.185–2.202)||0.003||C||24.5/18.1||1.550 (1.198–2.014)||0.001|
|rs4430796||GA + AA||NA||1.762 (1.105–2.819)||0.017||A||69.3/63.0||1.350 (1.081–1.685)||0.008|
|Chen et al. ||Taiwanese||rs1447295||CA + AA||36.8/24.9||1.75 (1.26–2.44)||0.001||A||19.3/13.8||1.49 (1.12–1.99)||0.07|
|Chen et al. ||Taiwanese||rs16901979||CA + AA||55.3/48.4||1.32 (0.97–1.79)||0.046||A||32.9/27.8||1.28 (1.01–1.62)||0.046|
|rs6983561||CA + CC||58.3/47.9||1.45 (1.06–1.97)||0.022||C||34.9/27.7||1.40 (1.11–1.77)||0.006|
|Zheng et al. ||Chinese||rs1016343||NA||NA||NA||NA||T||45.3/33.0||2.07 (1.35–3.20)||<0.001|
|Tan et al. ||Asian Indian||rs1447295||CA + AA||32.0/22.6||1.60 (1.01–2.52)||NA||A||17.3/12.6||NA||0.18|
|Present study||Korean||rs1447295||CA + AA||40.9/24.4||2.22 (1.31–3.79)||0.003||A||23.6/13.1||2.02* (1.28–3.21)||0.003|
|rs16901979||CA||41.8/33.7||1.69 (1.01–2.81)||0.045||A||28.1 / 24.0||1.29* (0.88–1.90)||0.197|
In the present study, we identified an association between genetic variations of 8q24 and the risk of prostate cancer in Korean men. A cumulative effect of significant SNPs on susceptibility of prostate cancer was found and haplotype analyses revealed that the [AGC] and [CTA] haplotypes had a potential relationship with increased prostate cancer risk. Based on a recent meta-analysis that include multi-ethnic cohorts, Liu et al.  reported moderate effects of 31 SNPs on prostate cancer and 14 independent prostate risk loci were identified. The present study is of interest due to the identification of an association between common variations of 8q24 and susceptibility to prostate cancer in Koreans, despite the limited genetic data involving the Asian population. Because GWAS was performed in many diseases, SNPs in genetic studies have been actively used to find the cause of prostate cancer at the level of the gene . Based on the GWAS related to prostate cancer, researchers have detected several genes at different loci. In particular, 8q24 has been the focus of many targeted SNP studies in diverse ethnic groups [8–10,14,15,22,23]. Based on the genetic studies involving prostate cancer families in Iceland , rs1447295 (‘at-risk’ allele A) at 8q24.21 was the first region to be strongly associated with prostate cancer risk in case-control studies of Caucasians [9,11,18,19]. In contrast, this SNP only showed a weak association among African-American men . For the second region, rs16901979 (‘at-risk’ allele A) showed strong effects in both Caucasians and African-Americans [8,11]. Finally, rs6983267 (‘at-risk’ allele G), located between two regions, was shown to be strongly associated with prostate cancer [9,11]. In the present study, the CA or AA genotype of rs1447295 had the strongest risk of prostate cancer among the three SNPs. We also showed that the CA genotype of rs16901979 was significantly associated with susceptibility to prostate cancer. In contrast, there was no association in any genotype of rs6983267. The subjects with the rs1447295 CA or AA genotype showed an association with development of prostate cancer with an increased risk (OR 2.22), which is a similar risk in comparison with the Caucasian population [8,9,11,12]. Despite a lower incidence of prostate cancer in Asian countries, there are some data that have shown the genetic association between 8q24 and the risk of prostate cancer in this population (Table 5) [14,16,22–26]. For rs1447295 in most Asian populations, the A allele or a genotype with any A was associated with an increased susceptibility to prostate cancer. The rs6983267 G allele did not have a significant association with prostate cancer in Japanese and Asian Indians, which is a similar finding with Korean men in the present study [24,26]. In Taiwanese and Asian Indians, the Rs16901979 A allele was significantly related with susceptibility to prostate cancer, which was similar to Korean subjects in whom the CA genotype of rs16901976 had a borderline significant relationship [14,26].
About 25% of the control group in the present study had the A allele (CA or AA genotype) of rs1447295, which means that healthy Korean men in the present study probably carry a potential risk for prostate cancer susceptibility (Table 2). Perhaps of greater interest, two SNPs, such as the CA or AA genotype of rs1447295 and the CA genotype of rs16901979, showed a strong cumulative effect for an increased risk of prostate cancer. These SNPs could be applied as useful variables for personal susceptibility to prostate cancer.
All of the patients included in the present study underwent radical prostatectomy and for patients from whom the entire prostate specimen was available for accurate and reliable grading and staging, there was an opportunity to determine the associations between clinical factors and genetic variations. The associations between pT stage and [AGC] haplotype, and PSA level and [CTA] haplotype were found at three SNPs of 8q24, as well as the relationship between the PSA level and rs1447295. However, there is limited data to interpret the association between genetic variations and aggressiveness of prostate cancer because there was no association between SNPs and Gleason score. In Caucasian populations, Wang et al.  reported a significant association between the SNP at rs1447295 with aggressive prostate cancer (Gleason score ≥ 8). In a Japanese population, the rs1447295 A allele was more significantly associated with disease in aggressive prostate cancer (PSA level ≥ 20 ng/mL, Gleason score ≥ 8, or clinical stage ≥ T3; ). In contrast, Freedman et al.  reported no such difference in association with tumour grade, although different groups were compared (Gleason score ≥ 8 vs <8). These inconsistent results may be due to different races and the characteristics of participants, as well as limitations in sample size according to each study.
The HapMap project has established the common pattern of DNA sequence variation in the human genome for several populations . When we included three SNPs for haplotype analysis, carriers of the [AGC] or [CTA] haplotype exhibited a higher risk for prostate cancer. When haplotypes consisted of two SNPs with significance in each, we also found consistent significance of haplotype in terms of risk for prostate cancer. Beuten et al.  reported that major haplotypes, containing the non-risk alleles, conferred protection against prostate cancer in Caucasian and Hispanic men. The present study is the first study involving an Asian population and haplotypes of common variations of 8q24 that has shown significance for prostate cancer.
Despite the new findings that the haplotypes of 8q24 SNPs are related to prostate cancer risk in Korean men, there are some limitations of the present study. First, for demographic characteristics, the patients were older than the controls. It was not feasible to match the patients with prostate cancer with healthy controls of the same age because the patients who registered for cancer screening at our institution were generally younger than the patients with cancer. To minimise the effect of the different age distribution in cases and controls, we performed an age-adjusted analysis to identify the risk SNPs and haplotypes. Furthermore, our primary interest lies on the genotype and its association with prostate cancer risk and it is unlikely that the genotype is affected by age. Thus, we think that the bias possibly caused by different age distributions among cases and controls would be minimal for the purposes of the present study. Second, the sample size was relatively small. Our main findings on the association between polymorphisms/haplotype and prostate cancer risk were validated using GMDR, which remained significant despite relatively small sample sizes. Further studies with a larger sample may be needed to confirm the associations between the haplotypes of SNPs in the 8q24 gene and prostate cancer. Last, only three SNPs with a strong association of prostate cancer in Western populations were examined in the present study, but this is clinically meaningful as an initial study on genetic susceptibility of prostate cancer in Korean men and the first report of 8q24 haplotypes in an Asian population.
In conclusion, we found an association between subjects with the genetic variation of 8q24 and the risk of development of prostate cancer in Korean men. As a result of haplotype analysis, [AGC] and [CTA] carriers showed a significant association with prostate cancer risk.
This study was supported by Korean National Cancer Center grants (nos. 0810220 and 0910221). All experiments were performed at the Genomics Core Facility in the National Cancer Center.
CONFLICT OF INTEREST