Genetic variation in RNASEL associated with prostate cancer risk and progression

Introduction

Exposure to infectious agents, hormones or dietary carcinogens may contribute to inflammation in the prostate (1). Intraprostatic inflammation may alter the tissue microenvironment, lead to cellular and genetic damage, and promote proliferation and angiogenesis, which may drive cancer progression (2). Genetic variation in immune and inflammatory response pathways may modulate the relationship between inflammation and prostate cancer pathogenesis and progression. RNASEL, a candidate hereditary prostate cancer gene (HPC1) located at chromosome 1q25, encodes ribonuclease L (RNASEL), an enzyme in the interferon (IFN)-induced antiviral 2-5A system. Upon exposure to IFNs, cells express numerous antiviral genes, including IFN-stimulated genes, which encode 2-5A synthetases (2',5'-oligoadenylate synthetase, OAS) (3). When activated by the presence of dsRNA, OAS proteins produce 2′,5′-linked oligoadenylates (2-5A) from adenosine triphosphate; the short-lived signaling molecules serve to activate latent RNASEL protein (3). RNASEL employs several mechanisms to fight viral infections, including cleavage of viral RNA and induction of apoptosis (4,5). Mutations in the gene may reduce the ability to fight viral infections, as demonstrated by RNASEL knockout mice that have impaired IFN-α activity, suppression of apoptosis and an increased susceptibility to viral infections (6). RNASEL has been linked to hereditary prostate cancer since the mid-1990s when it was included in the first linkage peak identified for prostate cancer in a genome-wide scan of 91 high-risk families (7). However, recent genome-wide association studies of sporadic prostate cancer have not identified RNASEL as a susceptibility locus (8–10). A common missense mutation in RNASEL, R462Q (rs486907, G→A), has been studied extensively in genetic epidemiology studies of prostate cancer risk. The homozygous genotype (AA) has 15% prevalence in Caucasian populations and causes a functionally significant amino acid substitution resulting in a 3-fold decrease in enzymatic activity (4,11,12). Epidemiological studies have been mixed, with some finding an increased risk of prostate cancer with at least one copy of the variant allele (12–14), whereas others report no association (15–19). A recent meta-analysis of seven studies reported no significant association of R462Q with prostate cancer across all ethnic groups or by family history status (20). However, R462Q may still affect prostate cancer progression and outcomes as most prior studies did not report on these potential associations.

A second missense mutation in RNASEL, D541E (rs627928, T→G), has also been examined in studies of prostate cancer risk (21). Results for this single-nucleotide polymorphism (SNP) have been inconsistent across populations and ethnic groups; however, a meta-analysis of six studies reported an increased risk for prostate cancer among Caucasians with at least one copy of the variant (G) allele [odds ratio (OR): 1.27, 95% confidence interval (CI): 1.13–1.44) (20). In addition, a recent study found a complementary 36% decreased risk for men homozygous for the T allele (CI: 0.44–0.92) (14). Both R462Q and D541E variants are located in exonic regions toward the 5′ end of the gene. Several familial studies have identified other less common mutations in RNASEL associated with prostate cancer, including truncating mutation E265X, frameshift mutation 471delAAAG and M1I in the initiation codon (4,22).

In this study, we comprehensively explore RNASEL by capturing genetic variation across a 25.6 kb region, including the gene and 5 kb upstream and downstream. Our analysis builds upon past studies by selecting 11 SNPs and investigating a number of prostate cancer end points, including associations with advanced and high-grade cancers and prostate cancer survival. To our knowledge, no study has yet examined the relationship between RNASEL and prostate cancer survival, although given the inflammatory mechanism, it is a plausible hypothesis. Because RNASEL is involved in innate immunity, and certain mutations may affect the antiviral response, we also explored the possibility that mutations in RNASEL may alter levels of circulating inflammatory biomarkers interleukin-6 (IL-6), C-reactive protein (CRP) and tumor necrosis factor-alpha receptor 2 (TNFR2). We therefore explored the effects of RNASEL genetic variants on prostate cancer risk and progression, as well as inflammatory markers, in the Physicians’ Health Study.

Materials and methods

Results

There were 1286 cases and 1264 controls with data on RNASEL included in the analyses. Characteristics of study participants can be found in Table I. Eleven tagging SNPs were chosen to account for genetic variation in RNASEL (rs12034888, rs12757998, rs635261, rs12729828, rs10911099, rs627839, rs11807829, rs533259, rs627928, rs486907 and rs682585). One SNP, rs12729828, violated Hardy–Weinberg equilibrium, with P = 0.0002. Although we included the SNP in our analyses, results should be interpreted with caution. All other SNPs were in Hardy–Weinberg equilibrium and were in line with observed frequencies reported in the HapMap CEPH sample set.

Among the cases, 70% were diagnosed during the PSA era. We observed a significant increased risk of prostate cancer with variant rs12757998, located downstream of the gene (AA versus GG; OR: 1.63, 95% CI: 1.18–2.25, Table II). Unlike previous reports in the literature, we found no increased risk of prostate cancer with R462Q (AA versus GG; OR: 0.96, 95% CI: 0.75–1.24) or D541E (GG versus TT; OR: 1.03, 95% CI: 0.82–1.28). Dominant models were employed for four SNPs with <5% rare homozygous genotypes but produced the same results as the codominant models (Table II).

We further assessed subgroups of cases defined by Gleason grade or stage at diagnosis (Table III). Variant rs12757998 (AA) was associated with a significantly increased risk of having a high-grade tumor (OR: 1.90, 95% CI: 1.25–2.89), although not with advanced stage disease (OR: 1.44, 95% CI: 0.79–2.63). R462Q was not significantly associated with advanced stage disease overall or in the PSA era; however, there was a significantly increased risk in the pre-PSA era with the rare homozygous genotype (OR: 1.93, 95% CI: 1.05–3.56). Similar results were observed with D541E, which was null overall and in the PSA era, yet significantly increased risk of advanced stage disease in individuals with one (OR: 2.18, 95% CI: 1.20–4.00) or two (OR: 2.04, 95% CI: 1.08–3.85) copies of the variant allele among cases diagnosed before 1992 (supplementary Table I is available at Carcinogenesis Online). The remaining SNPs do not appear to be significantly associated with high-grade or advanced stage disease.

We confirmed 176 lethal prostate cancers (prostate cancer-specific deaths or bony metastases) among the cases in our analysis by the end of follow-up. On average, those who developed bony metastases or died from prostate cancer had 6.5 years of follow-up since diagnosis compared with 10.6 years for non-fatal cases of prostate cancer. Survival analyses did not yield significant associations between any SNP and survival following prostate cancer diagnosis in this population, including R462Q (hazard ratio: 1.05, 95% CI: 0.66–1.67) and D541E (hazard ratio: 1.02, 95% CI: 0.66–1.57). The results of the survival analysis can be found in supplementary Table II (available at Carcinogenesis Online).

For cases in the analyses of inflammatory biomarkers, the median time between blood collection and prostate cancer diagnosis was 9.9 years (range 0.3–17.9 years). We excluded 24 men who returned blood samples >5 days after blood draw, as well as 69 cases and 50 matched controls with <5 years between blood draw and diagnosis from biomarker analyses, leaving 538 cases and 420 controls with genetic, plasma biomarker and BMI data.

RNASEL variants appear to affect circulating levels of plasma inflammatory biomarkers CRP (Figure 1) and IL-6 (Figure 2). Variant rs12757998, which increased risk of both overall prostate cancer and high-grade tumors in this analysis, was also associated with elevated serum levels of both CRP (P = 0.02) and IL-6 (P = 0.05) in linear regression models. Participants with two copies of the variant allele had on average 14% higher plasma CRP levels (1.12 versus 0.96 mg/l) and 18% higher plasma IL-6 levels (1.78 versus 1.46 pg/ml) than individuals with the wild-type form (GG). We also observed a significant association between variant rs12034888, located downstream of the gene, and reduced TNFR2 levels (P = 0.03), although there are very few individuals who are homozygous for the variant allele (T) in this study (n = 8). We did not detect any significant interactions between median levels of biomarkers and genotype in relation to prostate cancer risk or progression.

We did not detect significant associations between any of the genotypes and PSA levels among cases diagnosed >5 years after blood draw and controls together, or among controls alone, with P-values from linear regression models ranging from 0.32 (rs533259) to 0.92 (rs12034888). Despite reports of diminished capacity to induce apoptosis with mutated RNASEL protein, there were no apparent associations between RNASEL genotype and percent of tumor cells undergoing apoptosis (mean 2.3%; range 0–30%). The results of linear regression analysis of the TUNEL data resulted in non-significant P-values ranging from 0.09 (rs12034888) to 0.84 (rs682585). The R462Q missense mutation was not associated with apoptosis (P = 0.84).

Discussion

In our study expanding coverage of genetic variation in RNASEL, we identified new associations with the SNP rs12757998, which appeared marginally associated with overall prostate cancer risk and high-grade disease. Located downstream of the gene, rs12757998 is not strongly correlated with either missense mutation R462Q (r² = 0.08) or D541E (r² = 0.02), suggesting an independent association with prostate cancer. In addition to increasing the risk of overall and high-grade disease by >50% when compared with the wild-type genotype, variant rs12757998 was also linked to higher levels of circulating inflammatory biomarkers CRP and IL-6. Taken together, these results suggest a potential inflammatory mechanism linking the mutation to prostate cancer, although an exact biological mechanism is currently unknown. Significant SNPs downstream of the gene could be involved in gene regulation or could be in linkage disequilibrium with a functional SNP within the gene that we did not directly genotype. The SNP rs12757998 tags nearby SNP rs12723593 with r² = 1.0, suggesting the putative SNP could be either of these with equal probability, as well as any other known or unknown SNP in strong linkage disequilibrium.

The associations between variant 12757998 and circulating inflammatory biomarkers are novel findings. IL-6, a cytokine with both pro- and anti-inflammatory properties, plays a key role in mediating inflammation, including the acute-phase response, which promotes CRP production. One recent study found a significant positive association between IL-6 levels and prostate cancer incidence among men with BMI <25 kg/m² (24). Elevated prediagnostic CRP, a non-specific marker of systemic inflammation, has been associated with cancer risk and survival in some (27,28) but not all (29,30) studies. CRP levels have also been associated with poor survival among men with androgen-independent prostate cancer (31). Levels of both inflammatory biomarkers have been observed to be higher in men with prostate cancer when compared with healthy men or men with benign conditions (32,33). Relationships between variation in RNASEL and these biomarkers have not been reported in the literature.

We did not replicate previously reported associations for missense mutations R462Q and D541E with overall prostate cancer risk in this study population. However, the potential modification by calendar year of diagnosis is a novel finding. Cases diagnosed after 1992 during the ‘PSA era’ are more probably to be diagnosed by regular PSA screening and thus at an earlier stage. In contrast, men diagnosed before PSA screening was commonplace are more probably to have presented with symptoms of their disease and thus at a more advanced stage. Furthermore, pre-PSA era cases may better reflect the natural history of prostate cancer and the subsequent shift toward PSA-detected early stage cancers could have diluted any association between the well-studied missense mutations and advanced stage disease. However, because we are looking at a subset of the data, these findings could be due to chance.

The SNP rs12757998 also appeared recently in an analysis of the San Antonio Center for Biomarkers of Risk of Prostate Cancer (SABOR) cohort, although with somewhat conflicting results (34). In the SABOR cohort, having at least one copy of the variant (A) allele of rs12757998 was associated with reduced prostate cancer risk only among Hispanic men with a family history of the disease (OR: 0.41, 95% CI: 0.19–0.88); separate results for men carrying two copies of the variant allele were not shown. They did not report significant findings for rs12757998 among Caucasian men. The study also investigated 10 other SNPs in RNASEL, including rs635261, rs10911099 and rs533259, which also appear in our analysis. Similar to our results, they did not find any significant associations between prostate cancer and these three SNPs (34). A recent study of genetic variants among familial prostate cancer cases also reported null findings for rs533259, as well as for rs11807829, which is consistent with the results of our study (14). Although several genome-wide association studies failed to identify SNPs in RNASEL as significantly associated with sporadic prostate cancer, genetic variation in RNASEL may still impact disease risk. Genome-wide association studies have lower power to detect rare alleles and associations of smaller magnitude, and thus, our study design may be more appropriate in these cases.

There are several strengths to consider when assessing the findings of this study. The prospective nature of our analysis allowed us to ascertain exposure information before disease onset. In particular, this is important for the biomarker analyses, where disease and treatment could affect circulating levels of inflammatory markers. We have long and complete follow-up, with reasonably accurate reporting by physician participants. Residual confounding from variations in socioeconomic status and access to healthcare is unlikely in this population and is not probably to influence results of a genetic study. We have also restricted our study sample to self-reported Caucasians to reduce the influence of population stratification.

Our large study population gives us sufficient power (>80%) to detect an association of OR ≥1.4 for each of the 11 SNPs. The large sample size also allows us to conduct subgroup analyses and investigate relationships with disease stage and grade. However, although we have considerable power in this study, we remain underpowered to detect very small effects of RNASEL on prostate cancer or interactions with plasma biomarkers. The relatively small number of prostate cancer deaths may have restricted the power of the survival analysis to detect associations in the more rare genotypes; no clear relation between RNASEL and prostate cancer-specific survival emerged from our data.

As with other genetic studies, multiple testing is of concern. A conservative approach to correct for multiple testing is the Bonferroni correction, where the alpha value (0.05) is divided by the number of tests. With the 11 SNPs in this study, the corrected alpha value would be 0.005. After applying this conservative measure, the result for variant rs12757998 and total prostate cancer would maintain statistical significance, with a P-value of 0.004 from the likelihood ratio test, as would the association between variant rs12757998 and high Gleason grade tumors, with a likelihood ratio P-value of 0.003. However, the associations between rs12757998 and inflammatory biomarkers CRP and IL-6 would not remain statistically significant after correcting for multiple testing, with original P-values of 0.02 and 0.05, respectively. Because the associations between variant rs12757998 and prostate cancer remain significant after conservative correction for multiple testing, the SNP deserves replication in future studies.

The generalizability of our study results across racial or ethnic groups may be limited as our population was restricted to Caucasians. There have been discrepant findings across ethnic groups for carriage of the variant allele of R462Q. Several studies report an increased risk (13,35) or no association (16) in African-Americans, inconsistent observations in Caucasian men (12,13,35), an increased risk among Hispanic men (34,35) and an inverse association in a Japanese population (36). The recent analysis of the SABOR cohort also found an increased risk of prostate cancer among Hispanic men with two copies of the variant (G) allele of D541E. The same study also saw a 42% reduced risk of prostate cancer limited to Hispanic men with two copies of the variant (A) allele of rs682585, a SNP which was also selected in our analysis, although results were non-significant (34). The results from the SABOR cohort were not replicated among Caucasians, suggesting that the relationship between prostate cancer and RNASEL may vary by ethnic group.

Furthermore, our biomarker data are based upon a single baseline blood sample and may not be representative of long-term levels of inflammatory biomarkers. Multiple samples over the course of time could be more informative, possibly reducing within-person variation. In addition, CRP and IL-6 are markers of systemic inflammation and not exclusive to the prostate. The findings of increased CRP and IL-6 levels with variant rs12757998 suggest that an association between rs12757998 and prostate cancer may be mediated through inflammation; however, these results could also indicate that men with two copies of the variant allele have higher levels of systemic inflammation not necessarily intraprostatic inflammation. From our analysis of PSA levels, we found that the SNPs in RNASEL do not raise PSA, suggesting that the SNPs are probably associated with prostate cancer for a reason other than increasing PSA. We did not find any significant interactions between biomarkers and genotype in relation to prostate cancer risk or progression. It is possible that biomarker levels drawn closer to date of diagnosis may be more informative in this regard.

Recently, the novel xenotropic murine leukemia virus-related virus (XMRV) was discovered in prostate cancer tissue differentially by R462Q genotype (37). Men with the QQ genotype were more probably to have evidence of an infection than men who were heterozygous or wild-type, indicating the mutation might impede viral clearance (37). Although this result was not replicated in a subsequent study of 105 non-familial prostate cancer cases (38), future studies on RNASEL may expand the analysis to explore the relation of xenotropic murine leukemia virus-related virus infection with other SNPs.

The results from this study further support the proposed link between RNASEL and prostate cancer. The additional association between variant rs12757998 and levels of inflammatory biomarkers CRP and IL-6 lends credence to the findings and suggests inflammation as a possible mechanism linking rs12757998 to prostate cancer risk and progression. These results should be validated in other populations.

Supplementary material

Supplementary Tables I and II can be found at http://carcin.oxfordjournals.org/

Funding

National Cancer Institute (CA-058684 to M.J.S.); the National Research Service Award Training Program in Cancer Epidemiology from the National Institutes of Health (T32 CA-09001 to J.R.S. and K.L.P.); the Harvard University Franklin Reynolds Fund (to M.S.M.); L.A.M. is a Milken Scholar, Prostate Cancer Foundation. The Physicians’ Health Study is supported by the National Cancer Institute (CA-34944, CA-40360, CA-097193); the National Heart, Lung, and Blood Institute (HL-26490, HL-34595), Bethesda, MD.

Supplementary Material