Abstract
Circadian genes are responsible for maintaining the ancient adaptation of a 24-hour circadian rhythm and influence a variety of cancer-related biological pathways, including the regulation of sex hormone levels. However, few studies have been undertaken to investigate the role of circadian genes in the development of prostate cancer, the most common cancer type among men (excluding non-melanoma skin cancer). The current genetic association study tested the circadian gene hypothesis in relation to prostate cancer by genotyping a total of 41 tagging and amino acid altering SNPs in ten circadian-related genes in a population-based case-control study of Caucasian men (N=1,308 cases and 1,266 controls). Our results showed that at least one SNP in nine core circadian genes (rs885747 and rs2289591 in PER1, rs7602358 in PER2, rs1012477 in PER3, rs1534891 in CSNK1E, rs12315175 in CRY1, rs2292912 in CRY2, rs7950226 in ARNTL, rs11133373 in CLOCK, and rs1369481, rs895521, and rs17024926 in NPAS2) was significantly associated with susceptibility to prostate cancer (either overall risk or risk of aggressive disease), and the risk estimate for four SNPs in three genes (rs885747 and rs2289591 in PER1, rs1012477 in PER3 and rs11133373 in CLOCK) varied by disease aggressiveness. Further analyses of haplotypes were consistent with these genotyping results. Findings from this candidate gene association study support the hypothesis of a link between genetic variants in circadian genes and prostate cancer risk, warranting further confirmation and mechanistic investigation of circadian biomarkers in prostate tumorigenesis.
Keywords: Circadian Genes, Prostate Cancer, Genetic Polymorphisms
Introduction
The universal 24-hour oscillation, which arises from an ancient adaptation to the rotation of the earth, is one of the fundamental components of our biological world and has a profound impact on biochemical, physiological, and behavioral processes in almost all living organisms. Disrupted circadian rhythm caused by 'light-at-night' was hypothesized over a decade ago to explain, in part, the high risk of human breast and prostate cancers in industrialized societies (1, 2). An expert panel assembled by the International Agency for Research on Cancer (IARC) recently concluded that “shift-work that involves circadian disruption is probably carcinogenic to humans” based on “limited evidence in humans” and “sufficient evidence in experimental animals for the carcinogenicity of light during the daily dark period” (3).
The role of circadian disruption in prostate tumorigenesis has been supported by observations of pilots, flight attendants, and shift workers, who are more likely to have disrupted circadian cycles due to abnormal work hours. Studies from pilot cohorts have shown an excess risk of prostate cancer (4–6) that increases with increasing number of flight hours (6). Similarly, male rotating-shift workers have a higher risk of prostate cancer as compared with day- or night-only workers (7, 8). Furthermore, studies of the blind who have a reduced or abolished sensitivity to environmental light exposures have observed a reduced risk of prostate cancer (9, 10), which provides additional evidence suggesting a role for circadian disruption in the etiology of prostate cancer.
Circadian rhythms are controlled and maintained by several core circadian genes via transcription-translation feedback loops that include positive activators, e.g., CLOCK, neuronal PAS domain protein 2 (NPAS2), and aryl hydrocarbon receptor nuclear translocator-like (ARNTL), negative effectors, e.g., two cryptochrome (CRY1 and CRY2) and three period (PER1, PER2, and PER3) genes, and modulators (casein kinase 1, ε (CSNK1E); see Kondratov et al. (11) for an extensive discussion on the molecular organization of the circadian system). Emerging data have demonstrated that circadian genes are also involved in regulating cancer-related biological pathways such as cell proliferation and apoptosis by controlling the expression of tumor suppressor genes, cell cycle genes, and genes encoding caspases and transcription factors (reviewed in (12)).
Given the potential role of circadian genes in tumorigenesis, it has been hypothesized that genetic variations in these genes could be associated with an individual’s susceptibility to cancer, especially hormone-related cancers such as prostate cancer (13). In the current study, we tested the circadian gene hypothesis by systematically genotyping genetic variants in 10 circadian genes in subjects from two population-based case-control studies of prostate cancer.
Subjects, Materials and Methods
Study Population
The study population consists of participants from two population-based case-control studies in residents of King County, Washington (Study I and Study II), which have been described previously (14, 15). Briefly, subjects diagnosed with histologically confirmed prostate cancer were ascertained from the Seattle-Puget Sound SEER cancer registry. In Study I, cases were diagnosed between January 1, 1993, and December 31, 1996 and were 40–64 years of age at diagnosis. In Study II, cases were diagnosed between January 1, 2002, and December 31, 2005 and were 35–74 years of age at diagnosis. Overall, 2,244 eligible prostate cancer patients were identified, 1,754 (78.2%) were interviewed, and blood samples yielding sufficient DNA for genotyping were drawn from 1,457 (83.1%) cases who completed a study interview. Demographic and clinical characteristics did not differ between cases that did and did not provide a DNA sample. A comparison group of controls without a self-reported physician’s diagnosis of prostate cancer was identified using one-step random digit telephone dialing. Controls were frequency matched to cases by five-year age groups and recruited evenly throughout each ascertainment period for cases. A total of 2,448 men were identified who met the eligibility criteria, 1,645 (67.2%) were interviewed and blood samples were drawn and DNA prepared from 1,352 men (82.2%) using standard protocols. Demographic characteristics did not differ between controls that did and did not provide a DNA sample. For the current analyses, only Caucasian participants with DNA were included (1,308 cases and 1,266 controls).
Data Collection
Subjects in both studies completed in-person interviews conducted by trained male interviewers using similar standardized questionnaires. Questions pertained to the time prior to reference date, which was the date of prostate cancer diagnosis for cases and for controls, was a pre-assigned random date that approximated the distribution of cases’ diagnosis dates. The questionnaire collected information on social and demographic factors, family structure and cancer history, and medical history including prostate cancer screening history. Clinical information on cases, including Gleason score, stage of disease and serum PSA level at diagnosis was obtained from the cancer registry. All study procedures were reviewed and approved by the Fred Hutchinson Cancer Research Center institutional review board and written informed consent was obtained from all study subjects before participation.
SNP Selection and Genotyping
SNPs were identified using the Tagger algorithm (16), which is implemented in the Haploview interface (17) of HapMap’s genome browser, Release 22. SNP sets were identified to capture all of the variation found within each circadian gene with r2 ≥ 0.8. In addition to the tagging SNPs, seven SNPs that cause amino acid changes were included in the genotyping panel. We also examined data from the National Cancer Institute (NCI) Cancer Genetic Markers of Susceptibility (CGEMS) project (http://cgems.cancer.gov/) (18) and of the eight SNPs found to be significantly associated with prostate cancer risk in CGEMS (global P<0.05), six were included in the current study. Due to SNP compatibility limitations associated with the genotyping assay, two were unable to be included in the current study. Genotyping for all SNPs was performed at the National Human Genome Research Institute (Dr. Ostrander’s laboratory) using the SNPlex Genotyping System (Applied Biosystems, Inc., Foster City, CA) according to the manufacturer’s protocol. The details of this assay have been described previously (19, 20). Briefly, the genotyping system employs multiplex PCR technology and capillary electrophoresis in order to determine the genotypes of up to 48 different SNPs per SNPlex array. The GeneMapper software package (Applied Biosystems) was used to assign genotypes for each SNP. Replicate samples (n=141) were interspersed throughout all genotyping batches, and the concordance levels for blind duplicate samples were > 99% for all SNPs assayed. All genotyping scores, including quality control data, were re-checked by different laboratory personnel and the accuracy of each assay was confirmed.
Statistical Analysis
All statistical analyses were performed using SAS Version 9.1 (SAS Institute, Cary, NC), unless otherwise noted. A chi-square test was used to test for departures from Hardy-Weinberg equilibrium (HWE) for each SNP in the control population. SNPs not in HWE (p<0.01) were excluded from subsequent analyses. A goodness-of-fit chi-square test was used to determine whether the data were best described using a recessive, dominant or co-dominant model. Since no SNPs were significant under a recessive model only, and using the homozygous common plus heterozygous genotypes as the referent category did not dramatically alter the risk estimates for any SNP, all genotyping results are presented under both a co-dominant and dominant model. Adjusted odds ratios (ORs) with 95% confidence intervals (CI) and two-sided P-values were calculated by unconditional logistic regression to estimate the relative risk associated with each genotype. Adjustments were made to control for age and family history of prostate cancer in first-degree relatives. In addition, an indicator variable for study (I or II) was tested for confounding effects in the overall model, and no evidence of confounding was detected. Further, an interaction term for study and the combined variant genotypes for all significant SNPs in the overall population was tested, and no strong effect modification was identified (P>0.01). As such, the study covariate was not included in the final models. In addition to the overall associations, separate risk estimates were obtained for less aggressive and more aggressive tumors using multinomial logistic regression. Tests for interaction were also conducted using a Wald chi-square comparing the parameter estimates obtained for each genotype in less aggressive versus more aggressive cases. A binomial test was conducted to determine the probability of obtaining as many or more significant SNPs by chance alone, as were actually observed in our analysis.
Haplotype estimates were calculated by the PHASE program (21) using the genotyping information from all markers in each gene. Odds ratios and 95% confidence intervals for each haplotype were determined by unconditional multivariate logistic regression, using all other haplotypes as the referent category. The same covariates that were included in the main effects model were also used in the haplotype model. Separate risk estimates were not obtained for haplotypes with a frequency less than 1%, but these individuals remained in the analysis as members of the reference group.
Results
Compared to controls, prostate cancer cases reported a higher proportion of first-degree relatives with a history of prostate cancer, a greater frequency of prostate-specific antigen (PSA) screening and a higher serum PSA level (p<0.001; Table 1). Genotyping was completed for 44 SNPs in nine core circadian genes and one circadian related gene, TIMELESS. A list of these SNPs, including the minor allele frequency (MAF) detected in this study population and p-values for HWE, is presented in Table 2. The MAF of each SNP was very similar to that found in the CEPH HapMap population (www.hapmap.org/), which like this study, is population-based. Three SNPs (rs11133376 in CLOCK, rs1534891 in CSNK1E and rs2291739 in TIMELESS) did not fit HWE (P<0.01) and were thus excluded from further analysis.
Table 1.
Cases (N=1308) | Controls (N=1266) | p-value* | |||
---|---|---|---|---|---|
N | % | N | % | ||
Age at Reference Date | |||||
35–49 | 102 | 7.8 | 107 | 8.4 | |
50–54 | 188 | 14.4 | 178 | 14.1 | |
55–59 | 325 | 24.8 | 343 | 27.1 | |
60–64 | 395 | 30.2 | 334 | 26.4 | |
65–69 | 153 | 11.7 | 160 | 12.6 | |
70–74 | 145 | 11.1 | 144 | 11.4 | |
BMI | |||||
<25.0 | 429 | 32.8 | 389 | 30.7 | |
25.0–29.9 | 637 | 48.7 | 618 | 48.8 | |
≥30 | 242 | 18.5 | 259 | 20.5 | 0.34 |
Smoking History | |||||
Never | 522 | 39.9 | 541 | 42.7 | |
Former | 631 | 48.2 | 561 | 44.3 | |
Current | 155 | 11.9 | 164 | 13.0 | 0.13 |
Family History of PC‡ | |||||
No | 1025 | 78.4 | 1124 | 88.8 | |
Yes | 283 | 21.6 | 142 | 11.2 | <0.0001 |
PSA Screening History | |||||
Never | 135 | 10.3 | 167 | 13.2 | |
DRE only | 223 | 17.1 | 483 | 38.1 | |
PSA | 950 | 72.6 | 616 | 48.7 | <0.0001 |
PSA Level ± (ng/mL) | |||||
0–3.9 | 178 | 13.6 | 1175 | 92.8 | |
4–9.9 | 722 | 55.2 | 74 | 5.8 | |
10–19.9 | 190 | 14.5 | 15 | 1.2 | |
≥20 | 118 | 9.0 | 2 | 0.2 | |
Missing | 100 | 7.7 | 0 | 0.0 | <0.0001 |
Gleason Score | |||||
2–4 | 67 | 5.1 | |||
5–6 | 680 | 52.0 | |||
7=3+4 | 355 | 27.1 | |||
7=4+3 | 76 | 5.8 | |||
8–10 | 126 | 9.6 | |||
Missing | 4 | 0.3 | |||
Stage of Disease | |||||
Local | 1022 | 78.1 | |||
Regional | 254 | 19.4 | |||
Distant | 32 | 2.5 | |||
Aggressiveness† | |||||
Less | 873 | 66.7 | |||
More | 435 | 33.3 |
Chi-square test
Family history of cancer in a first-degree relative
Serum PSA at diagnosis for cases and at interview for controls
Less aggressive: localized stage, Gleason score ≤7 (3+4), and a diagnostic PSA level <20 ng/mL; More aggressive: regional or distant stage, or Gleason score ≥7 (4+3) or a diagnostic PSA level ≥20 ng/mL
Table 2.
dbSNP ID | Gene | SNP Typea | Allele (Major/Minor) |
MAF | P for HWE |
Location |
---|---|---|---|---|---|---|
rs2278749 | ARNTL | Tagging | G/A | 0.184 | 0.320 | 11p15 |
rs2290035 | ARNTL | Tagging | A/T | 0.457 | 0.227 | 11p15 |
rs3816360 | ARNTL | Tagging | G/A | 0.353 | 0.706 | 11p15 |
rs4757144 | ARNTL | Tagging | A/G | 0.409 | 0.901 | 11p15 |
rs6486122 | ARNTL | Tagging | T/C | 0.335 | 0.810 | 11p15 |
rs7950226 | ARNTL | Tagging | G/A | 0.457 | 0.356 | 11p15 |
rs11133373 | CLOCK | Tagging | C/G | 0.341 | 0.162 | 4q12 |
rs11133376 | CLOCK | Tagging | T/C | 0.462 | 0.000 | 4q12 |
rs1399953 | CLOCK | Tagging | G/A | 0.266 | 0.850 | 4q12 |
rs17776421 | CLOCK | Tagging | G/A | 0.383 | 0.776 | 4q12 |
rs10778528 | CRY1 | Tagging | T/G | 0.430 | 0.650 | 12q23-q24.1 |
rs10778534 | CRY1 | Tagging | T/C | 0.362 | 0.063 | 12q23-q24.1 |
rs12315175 | CRY1 | Tagging | T/C | 0.194 | 0.954 | 12q23-q24.1 |
rs7297614 | CRY1 | CGEMS | T/C | 0.392 | 0.759 | 12q23-q24.1 |
rs11605924 | CRY2 | Tagging | A/C | 0.472 | 0.035 | 11p11.2 |
rs12281674 | CRY2 | Tagging | A/G | 0.030 | 0.893 | 11p11.2 |
rs2292910 | CRY2 | Tagging | C/A | 0.316 | 0.245 | 11p11.2 |
rs2292912 | CRY2 | Tagging | C/G | 0.217 | 0.619 | 11p11.2 |
rs135757 | CSNK1E | CGEMS | G/A | 0.245 | 0.160 | 22q13.1 |
rs1534891 | CSNK1E | CGEMS | C/T | 0.124 | 0.001 | 22q13.1 |
rs1997644 | CSNK1E | Tagging | A/G | 0.467 | 0.039 | 22q13.1 |
rs5757037 | CSNK1E | Tagging | G/A | 0.369 | 0.039 | 22q13.1 |
rs6001093 | CSNK1E | CGEMS | T/C | 0.184 | 0.172 | 22q13.1 |
rs1369481 | NPAS2 | Tagging | G/A | 0.300 | 0.315 | 2q11.2 |
rs1562313 | NPAS2 | CGEMS | G/A | 0.208 | 0.782 | 2q11.2 |
rs17024926 | NPAS2 | Tagging | T/C | 0.329 | 0.504 | 2q11.2 |
rs2305160 | NPAS2 | AA change | C/T | 0.340 | 0.418 | 2q11.2 |
rs4074920 | NPAS2 | Tagging | A/G | 0.245 | 0.041 | 2q11.2 |
rs7565018 | NPAS2 | Tagging | A/G | 0.236 | 0.725 | 2q11.2 |
rs895521 | NPAS2 | CGEMS | G/A | 0.240 | 0.258 | 2q11.2 |
rs965519 | NPAS2 | Tagging | A/G | 0.153 | 0.657 | 2q11.2 |
rs2253820 | PER1 | Tagging | G/A | 0.154 | 0.881 | 17p13.1-p12 |
rs2289591 | PER1 | Tagging | G/T | 0.243 | 0.495 | 17p13.1-p12 |
rs885747 | PER1 | Tagging | C/G | 0.456 | 0.983 | 17p13.1-p12 |
rs11695472 | PER2 | Tagging | A/C | 0.269 | 0.072 | 2q37.3 |
rs7602358 | PER2 | Tagging | T/G | 0.238 | 0.221 | 2q37.3 |
rs1012477 | PER3 | Tagging | C/G | 0.143 | 0.179 | 1p36.23 |
rs10462020 | PER3 | AA change | T/G | 0.192 | 0.120 | 1p36.23 |
rs11121023 | PER3 | Tagging | G/A | 0.219 | 0.584 | 1p36.23 |
rs228697 | PER3 | AA change | C/G | 0.099 | 0.781 | 1p36.23 |
rs2640909 | PER3 | AA change | T/C | 0.295 | 0.327 | 1p36.23 |
rs2291739 | TIMELESS | AA change | T/C | 0.411 | 0.002 | 12q12-q13 |
rs774027 | TIMELESS | AA change | T/A | 0.453 | 0.456 | 12q12-q13 |
rs774047 | TIMELESS | AA change | C/T | 0.451 | 0.511 | 12q12-q13 |
AA = amino acid, CGEMS = Cancer Genetic Markers of Susceptibility
Genotypes of circadian genes and prostate cancer risk
Of the remaining 41 SNPs, 17 were significantly associated with risk of prostate cancer (either overall risk or risk of more aggressive disease), whereas ∼ 6 (41 × 0.05 × 3) would be expected by chance alone (P=0.0001; Table 3). Specifically, assuming a dominant model three SNPs located in NPAS2 (rs895521: OR=0.83; 95% CI, 0.70–0.97; rs1369481: OR=0.81; 95% CI, 0.69–0.95; and rs17024926: OR=1.25; 95% CI, 1.07–1.47) and rs7950226 in ARNTL (OR=1.22: 95% CI, 1.02–1.46) were significantly associated with prostate cancer risk, as was the homozygous rare genotype of rs1534891 in CSNK1E (OR=2.65: 95% CI, 1.16–5.95). We also analyzed these SNPs for interaction with family history of prostate cancer in a first-degree relative. Only one SNP, rs17024926 in NPAS2, showed significant interaction (P for interaction=0.013 for the group with any variant allele), with a significant association observed only among men with no family history (OR=1.36: 95% CI, 1.15–1.63). In addition, since heterodimerization between the protein products of ARNTL and NPAS2 is required for proper functioning, and each of these genes harbored at least one significant SNP, we tested for interactions among each of the significant variants in these genes (rs7950226 in ARNTL combined with rs1369481, rs895521, or rs17024926 in NPAS2). In each case, elevated relative risks were observed for individuals carrying the risk allele at both loci, but the interactions were not statistically significant.
Table 3.
All | Less Aggressive | More Aggressive | ||||||
---|---|---|---|---|---|---|---|---|
Genotype | Controls | Cases | OR* | Cases | OR* | Cases. | OR* | |
N | N | (95% CI) | N | (95% CI) | N | (95% CI) | P-Value† | |
PER1, rs885747, tag | ||||||||
C/C | 367 | 375 | Ref. | 239 | Ref. | 136 | Ref. | |
C/G | 615 | 639 | 1.00 (0.83–1.20) | 432 | 1.05 (0.86–1.30) | 207 | 0.90 (0.70–1.16) | 0.24 |
G/G | 257 | 238 | 0.90 (0.72–1.14) | 170 | 1.01 (0.78–1.31) | 68 | 0.71 (0.51–0.99) | 0.05 |
C/G+G/G | 872 | 877 | 0.97 (0.82–1.15) | 602 | 1.04 (0.86–1.27) | 275 | 0.84 (0.66–1.07) | 0.10 |
PER1, rs2289591, tag | ||||||||
G/G | 702 | 696 | Ref. | 487 | Ref. | 209 | Ref. | |
G/T | 461 | 459 | 1.00 (0.84–1.18) | 296 | 0.91 (0.75–1.10) | 163 | 1.19 (0.94–1.51) | 0.04 |
T/T | 68 | 79 | 1.24 (0.88–1.75) | 47 | 1.05 (0.71–1.55) | 32 | 1.70 (1.08–2.66) | 0.05 |
G/T+T/T | 529 | 538 | 1.03 (0.87–1.20) | 343 | 0.93 (0.78–1.11) | 195 | 1.25 (1.00–1.57) | 0.02 |
PER2, rs7602358, tag | ||||||||
T/T | 706 | 669 | Ref. | 436 | Ref. | 233 | Ref. | |
T/G | 421 | 464 | 1.15 (0.97–1.36) | 327 | 1.24 (1.03–1.50) | 137 | 0.98 (0.77–1.25) | 0.07 |
G/G | 76 | 74 | 1.01 (0.72–1.42) | 45 | 0.94 (0.64–1.39) | 29 | 1.13 (0.72–1.79) | 0.46 |
T/G+G/G | 497 | 538 | 1.13 (0.96–1.33) | 372 | 1.19 (1.00–1.43) | 166 | 1.00 (0.80–1.26) | 0.16 |
PER3, rs1012477, tag | ||||||||
C/C | 914 | 899 | Ref. | 584 | Ref. | 315 | Ref. | |
C/G | 290 | 331 | 1.15 (0.96–1.39) | 238 | 1.28 (1.04–1.56) | 93 | 0.92 (0.71–1.21) | 0.02 |
G/G | 31 | 27 | 0.85 (0.50–1.45) | 20 | 0.97 (0.54–1.73) | 7 | 0.63 (0.27–1.45) | 0.33 |
C/G+G/G | 321 | 358 | 1.12 (0.94–1.34) | 258 | 1.25 (1.03–1.52) | 100 | 0.89 (0.69–1.16) | 0.02 |
CSNK1E, rs1534891, cgems | ||||||||
C/C | 941 | 966 | Ref. | 641 | Ref. | 325 | Ref. | |
C/T | 292 | 266 | 0.88 (0.73–1.06) | 183 | 0.91 (0.73–1.12) | 83 | 0.82 (0.62–1.08) | 0.50 |
T/T | 8 | 23 | 2.65 (1.16–5.95) | 18 | 3.09 (1.32–7.21) | 5 | 1.70 (0.55–5.29) | 0.25 |
C/T+T/T | 300 | 289 | 0.93 (0.77–1.12) | 201 | 0.97 (0.79–1.19) | 88 | 0.85 (0.64–1.11) | 0.35 |
CRY1, rs12315175, tag | ||||||||
T/T | 805 | 778 | Ref. | 515 | Ref. | 263 | Ref. | |
T/C | 387 | 406 | 1.09 (0.92–1.30) | 279 | 1.14 (0.94–1.38) | 127 | 1.01 (0.79–1.29) | 0.38 |
C/C | 47 | 63 | 1.47 (0.99–2.18) | 44 | 1.55 (1.01–2.39) | 19 | 1.30 (0.75–2.26) | 0.53 |
T/C+C/C | 434 | 469 | 1.13 (0.96–1.34) | 323 | 1.18 (0.98–1.42) | 146 | 1.04 (0.82–1.32) | 0.33 |
CRY2, rs2292912, tag | ||||||||
C/C | 763 | 812 | Ref. | 555 | Ref. | 257 | Ref. | |
C/G | 414 | 392 | 0.88 (0.74–1.05) | 252 | 0.83 (0.68–1.00) | 140 | 1.00 (0.79–1.27) | 0.14 |
G/G | 61 | 49 | 0.79 (0.54–1.17) | 34 | 0.81 (0.52–1.25) | 15 | 0.76 (0.43–1.37) | 0.86 |
C/G+G/G | 475 | 441 | 0.87 (0.74–1.03) | 286 | 0.82 (0.69–0.99) | 155 | 0.97 (0.77–1.22) | 0.19 |
ARNTL, rs7950226, tag | ||||||||
G/G | 367 | 317 | Ref. | 213 | Ref. | 104 | Ref. | |
G/A | 588 | 631 | 1.22 (1.01–1.48) | 430 | 1.24 (1.01–1.54) | 201 | 1.19 (0.90–1.56) | 0.76 |
A/A | 262 | 276 | 1.22 (0.97–1.53) | 177 | 1.16 (0.90–1.51) | 99 | 1.33 (0.97–1.83) | 0.44 |
G/A+A/A | 850 | 907 | 1.22 (1.02–1.46) | 607 | 1.22 (1.00–1.49) | 300 | 1.23 (0.95–1.59) | 0.94 |
CLOCK, rs11133373, tag | ||||||||
C/C | 520 | 575 | Ref. | 377 | Ref. | 198 | Ref. | |
C/G | 571 | 533 | 0.85 (0.72–1.01) | 353 | 0.86 (0.71–1.04) | 180 | 0.82 (0.65–1.04) | 0.71 |
G/G | 131 | 134 | 0.91 (0.69–1.19) | 102 | 1.06 (0.79–1.42) | 32 | 0.62 (0.41–0.95) | 0.02 |
C/G+G/G | 702 | 667 | 0.86 (0.73–1.01) | 455 | 0.90 (0.75–1.08) | 212 | 0.79 (0.63–0.99) | 0.26 |
NPAS2, rs1369481, tag | ||||||||
G/G | 600 | 677 | Ref. | 456 | Ref. | 221 | Ref. | |
G/A | 535 | 484 | 0.81 (0.68–0.96) | 326 | 0.81 (0.67–0.98) | 158 | 0.80 (0.63–1.02) | 0.95 |
A/A | 104 | 93 | 0.80 (0.59–1.09) | 60 | 0.77 (0.55–1.09) | 33 | 0.86 (0.57–1.32) | 0.62 |
G/A+A/A | 639 | 577 | 0.81 (0.69–0.95) | 386 | 0.80 (0.67–0.96) | 191 | 0.81 (0.65–1.02) | 0.92 |
NPAS2, rs895521, cgems | ||||||||
G/G | 709 | 778 | Ref. | 519 | Ref. | 259 | Ref. | |
G/A | 466 | 423 | 0.85 (0.71–1.0) | 291 | 0.87 (0.73–1.05) | 132 | 0.79 (0.62–1.00) | 0.42 |
A/A | 64 | 50 | 0.71 (0.48–1.05) | 32 | 0.68 (0.44–1.06) | 18 | 0.77 (0.44–1.32) | 0.70 |
G/A+A/A | 530 | 473 | 0.83 (0.70–0.97) | 323 | 0.85 (0.71–1.02) | 150 | 0.79 (0.62–0.99) | 0.52 |
NPAS2, rs17024926, tag | ||||||||
T/T | 563 | 499 | Ref. | 330 | Ref. | 169 | Ref. | |
T/C | 536 | 594 | 1.26 (1.06–1.49) | 394 | 1.26 (1.04–1.53) | 200 | 1.25 (0.99–1.59) | 0.95 |
C/C | 139 | 153 | 1.23 (0.94–1.60) | 111 | 1.34 (1.00–1.78) | 42 | 1.01 (0.68–1.49) | 0.17 |
T/C+C/C | 675 | 747 | 1.25 (1.07–1.47) | 505 | 1.28 (1.07–1.53) | 242 | 1.20 (0.96–1.51) | 0.62 |
Adjusted for age and family history of prostate cancer in a first-degree relative
Test for homogeneity of ORs for less versus more aggressive disease groups
In addition to the main effects model, a multinomial regression analysis was conducted to determine whether any of the circadian gene variants were associated with a composite variable indicating aggressive disease. For these analyses, controls were compared simultaneously to prostate cancer cases stratified into two groups: 1) less aggressive (localized stage, Gleason score ≤7 (3+4), and a diagnostic PSA level <20 ng/mL); and 2) more aggressive (regional or distant stage, or Gleason score ≥7 (4+3) or a diagnostic PSA level ≥20 ng/mL). These results showed that eight SNPs were significantly associated with risk of less aggressive prostate cancer, including five assuming a dominant model: rs1012477 in PER3 (OR=1.25; 95% CI, 1.03–1.52), rs2292912 in CRY2 (OR=0.82; 95% CI, 0.69–0.99), rs7950226 in ARNTL (OR=1.22; 95% CI, 1.00–1.49), and SNPs rs17024926 and rs1369481 in NPAS2 (OR=1.28; 95% CI, 1.07–1.53; OR=0.80; 95% CI, 0.67–0.96, respectively); two assuming a recessive model: rs1534891 in CSNK1E (OR=3.09; 95% CI, 1.32–7.21) and rs12315175 in CRY1 (OR=1.55; 95% CI, 1.01–2.39); and, finally, the heterozygous genotype of rs7602358 in PER2 (OR=1.24; 95% CI, 1.03–1.50). Four SNPs were significantly associated with risk of more aggressive prostate cancer, including two assuming a dominant model: rs11133373 in CLOCK (OR=0.79; 95% CI, 0.63–0.99), and rs895521 in NPAS2 (OR=0.79; 95% CI, 0.62–0.99); and the homozygous rare genotypes of two SNPs in PER1 [rs885747 (OR=0.71; 95% CI, 0.51–0.99), and rs2289591 (OR=1.70; 95% CI, 1.08–2.66)]. In total, the effect of one or more genotypes in four SNPs varied significantly according to disease aggressiveness (P for interaction<0.05), including two SNPs in PER1 (rs2289591 and rs885747), one SNP in PER3 (rs1012477), and one SNP in CLOCK (rs11133373). Of note, the effect of all five SNPs that were significantly associated with prostate cancer risk in the full population did not vary significantly by aggressiveness. Genotyping results, including prostate cancer risk estimates for each of the non-significant SNPs, are provided in Supplementary Table 1.
Haplotypes of circadian genes and prostate cancer risk
In addition to the single-SNP analyses, haplotypes were estimated using the genotyping data from all SNPs in each gene. The haplotype analysis identified seven haplotypes in two different genes that were significantly associated with prostate cancer risk in the full population (Table 4). These included four different haplotypes in ARNTL, with a combined frequency of 13.94% in the population, as well as three haplotypes in NPAS2 with a combined frequency of 16.11%. The effect of one common haplotype in PER1 (24.82%), varied significantly according to aggressiveness, as it was significantly associated with prostate cancer risk in the more aggressive disease group (OR=1.23; 95% CI, 1.03–1.48, P for interaction<0.05) but not in the less aggressive disease group (OR=0.96; 95% CI, 0.83–1.11).
Table 4.
All | Less Aggressive | More Aggressive | |||||||
---|---|---|---|---|---|---|---|---|---|
Haplotype | Frequency | Controls | Cases | OR* | Cases | OR* | Cases | OR* | |
% | N | N | 95% CI | N | 95% CI | N | 95% CI | P-Value† | |
ARNTL | |||||||||
GAGGTA | 6.00% | 124 | 176 | 1.42 (1.12–1.80) | 117 | 1.40 (1.08–1.82) | 59 | 1.46 (1.06–1.82) | 0.81 |
GTGGTA | 5.36% | 115 | 153 | 1.32 (1.02–1.69) | 105 | 1.36 (1.03–1.79) | 48 | 1.23 (0.87–1.75) | 0.59 |
ATAATA | 1.38% | 26 | 43 | 1.65 (1.01–2.71) | 29 | 1.65 (0.96–2.83) | 14 | 1.65 (0.85–3.18) | 0.99 |
GTAGCA | 1.20% | 38 | 22 | 0.56 (0.33–0.96) | 17 | 0.65 (0.36–1.17) | 5 | 0.39 (0.15–0.99) | 0.30 |
NPAS2 | |||||||||
GAGGTAAC | 11.63% | 259 | 322 | 1.25 (1.05–1.49) | 216 | 1.25 (1.03–1.52) | 106 | 1.26 (0.99–1.61) | 0.93 |
AAAGTAAT | 2.34% | 71 | 46 | 0.64 (0.44–0.94) | 27 | 0.56 (0.36–0.88) | 19 | 0.80 (0.48–1.33) | 0.25 |
GAGGCGGT | 2.14% | 67 | 40 | 0.58 (0.39–0.87) | 26 | 0.57 (0.36–0.90) | 14 | 0.62 (0.35–1.11) | 0.78 |
PER1 | |||||||||
CGT | 24.82% | 599 | 623 | 1.05 (0.92–1.19) | 394 | 0.96 (0.83–1.11) | 229 | 1.23 (1.03–1.48) | 0.01 |
Adjusted for age and family history of prostate cancer in a first-degree relative
Test for homogeneity of ORs for less versus more aggressive disease groups
Discussion
In this hypothesis driven association study of ten candidate circadian genes, three, ARNTL, CSNK1E, and NPAS2, were significantly associated with susceptibility to prostate cancer in the full population, and SNPs in an additional six genes, CLOCK, CRY1, CRY2, PER1, PER2, and PER3, were significantly associated with a subgroup comprised of less or more aggressive cancers. These results support the hypothesis of a link between genetic variations in circadian genes and prostate cancer, and are consistent with previous genetic association findings from breast cancer (22, 23), prostate cancer (24), and non-Hodgkin’s lymphoma (NHL) (21).
It is not unexpected that alterations in these genes might have a broad impact on cancer susceptibility since as many as 10% of all genes in the mammalian genome are under some form of circadian control, including a number of transcripts with relevance to tumorigenesis (25). In addition, the molecular clockwork of the circadian system operates via interacting feedback loops among each of the core genes. As such, disturbances in one gene could result in phenotypic consequences across the entire system. As a result of this interconnectedness, if alterations to one or a few of the circadian genes were in fact causally associated with cancer risk, this phenotype could potentially be elicited by perturbing any one of the core circadian genes, with the consequence that a number of circadian gene variants could be significantly associated with disease risk without being directly causal.
Our results are also congruent with findings from a recently released NCI genome-wide association study on prostate cancer that showed some significant associations between variants in circadian genes and prostate cancer risk. The CGEMS project genotyped 550,000 SNPs in 1,182 prostate cancer cases and 1,174 controls from the Prostate, Lung, Cervical, and Ovarian Cancer Screening Trial (18). A total of 104 SNPs in circadian genes were included in the CGEMS project. Eight of these, located in four genes, NPAS2 (2 SNPs), CSNK1E (3 SNPs), CRY1 (2 SNPs), and CRY2 (1 SNP), were significantly associated with prostate cancer risk (p < 0.05 for all SNPs), including two (NPAS2 rs895521 and CRY2 rs12281674) with p ≤ 0.01. Six of these eight variants were genotyped in the current study and of these, two, CSNK1E rs1534891 and NPAS2 rs895521, were also significantly associated with prostate cancer risk. There are several possibilities as to why the results of the two studies differed. First, our study is population-based and geographically homogenous, while CGEMS is neither. Since prostate cancer rates vary widely by geography and prostate cancer is known to be a genetically heterogeneous disease, it is not surprising that different genetic variants are highlighted in these two studies. Nevertheless, both studies support a role of circadian genes in prostate cancer, which warrants further confirmation and validation in future studies.
The clock-cancer connection in prostate cancer is further supported by discoveries demonstrating that the circadian clock may function as a tumor suppressor at the systemic, cellular, and molecular levels through its involvement in cell proliferation, apoptosis, cell cycle control, and DNA damage response (26–28). For example, mice with mutant mPer2 had impaired DNA damage responses to gamma irradiation, which made them more cancer-prone (26). CSNK1E also functions in promoting cell proliferation by stabilizing β-Catenin (29). β-Catenin can interact with transcription factors of the T-cell-specific transcription factor/lymphoid enhancer factor-1 (TCF/LEF) family to regulate transcription (30) and promote tumorigenesis (31). PER1 also plays an important role in regulating growth and DNA damage control, and it interacts with proteins in the cell-cycle pathway (32). A recent study also demonstrated that the circadian gene NPAS2 affects several gene pathways associated with the DNA damage response, including cell cycle, apoptosis, and DNA repair (33). As such, polymorphisms in circadian genes may change their protein expression and activity, thereby affecting downstream cancer-related biological pathways.
Apart from their tumor suppressor role, circadian genes might affect prostate cancer risk by mediating levels of serum androgen (13). This in turn may act as a carcinogen by increasing cellular proliferation, thereby increasing the chance of random DNA copy errors. The prostate is an androgen-dependent organ, since androgens are essential for its normal growth and maintenance. Animal studies have shown that large amounts of androgens given to rodents can induce prostate cancer (34). A human study has also observed that long-term exposure to high levels of androgens can promote the development of prostate cancer (35), although a large pooled analysis of 18 prospective studies did not support this finding (36). Further, men taking the drug finasteride, which reduces androgen exposure within the prostate, compared to men on placebo were shown to have about a 25% lower prevalence of prostate cancer over a seven year period (37). Levels of testosterone, the principle circulating androgen, have also been shown to play a pivotal role in the differentiation and maintenance of prostate cancer cells (38). The impact of circadian rhythms on the expression of androgens has been previously documented in animals (39). In men, a circadian pattern in the levels of serum total testosterone has also been observed, and this circadian rhythm becomes blunted with normal aging (40). Circadian rhythms in plasma levels of cortisol, dehydroepiandrosterone, delta 4-androstenedione, testosterone and dihydrotestosterone have been detected in healthy young men (41). In addition, circadian gene variants have been correlated to alterations in serum sex hormone levels, including testosterone and sex hormone binding globulin (42). These observations demonstrate the potential impact of the circadian rhythm on androgen expression, and suggest that circadian gene variants may alter the risk of hormonally-mediated cancers by influencing the regulation of hormone pathways.
Among all ten circadian genes examined, NPAS2 gives the most robust association with prostate cancer risk. Importantly, NPAS2 has been linked to the risk of two other cancer types: hormone-related breast cancer (23) and immune-related NHL (43). The possible involvement of NPAS2 in tumorigenesis has been supported by an animal study demonstrating that ARNTL/NPAS2 heterodimers directly suppress transcription of the oncogene c-myc (26). A recent in vitro study further showed that the circadian gene NPAS2 affects several pathways associated with the DNA damage response, including the cell cycle, apoptosis, and DNA repair (33). Although the explicit mechanisms underlying these associations remain unclear, there is strong evidence indicating that NPAS2 may serve as a biomarker for an individual’s risk of cancers. As such, further study is needed in order to elucidate the functional significance of NPAS2 variants, and to further understand their role in carcinogenesis.
In conclusion, the genotype and haplotype approaches, using both tagging and amino acid altering SNPs in this population-based case-control study, provide comprehensive evidence supporting a role for the molecular clockwork pathway in prostate tumorigenesis. The strengths of this study include the relatively large sample size available for genotyping and the novel findings associated with the emerging area of circadian genetics in cancer. In addition, while previous epidemiologic studies have noted a link between circadian disruption and prostate cancer risk, no previous study has systematically examined the effects of variants in circadian genes on prostate cancer susceptibility. A major limitation is one that is shared with most genetic epidemiology studies, i.e., the lack of functional data relating the variants under study to a well-defined phenotypic effect. Nevertheless, the genetic polymorphisms identified in this study will build a molecular basis for future mechanistic studies and phenotypic assessments of circadian genes in relation to prostate cancer. Given the newly identified role of the circadian clock in tumorigenesis, unraveling the association between circadian genes and prostate cancer will add to our understanding of a fundamental aspect of cellular processes in tumorigenesis, and may facilitate the development of novel risk and prognostic biomarkers for prostate cancer.
Supplementary Material
Acknowledgments
This work was supported by the National Institutes of Health (grants R01 CA56678, R01 CA082664, R01 CA092579, R01 CA122676, and R03 CA110937), with additional support from the Fred Hutchinson Cancer Research Center, Yale University and the National Human Genome Research Institute.
References
- 1.Stevens RG. Electric power use and breast cancer: a hypothesis. Am J Epidemiol. 1987;125:556–561. doi: 10.1093/oxfordjournals.aje.a114569. [DOI] [PubMed] [Google Scholar]
- 2.Stevens RG, Davis S, Thomas DB, Anderson LE, Wilson BW. Electric power, pineal function, and the risk of breast cancer. Faseb J. 1992;6:853–860. doi: 10.1096/fasebj.6.3.1740235. [DOI] [PubMed] [Google Scholar]
- 3.Straif K, Baan R, Grosse Y, et al. Carcinogenicity of shift-work, painting, and fire-fighting. The Lancet Oncology. 2007;8:1065–1066. doi: 10.1016/S1470-2045(07)70373-X. [DOI] [PubMed] [Google Scholar]
- 4.Band PR, Le ND, Fang R, et al. Cohort study of Air Canada pilots: mortality, cancer incidence, and leukemia risk. Am J Epidemiol. 1996;143:137–143. doi: 10.1093/oxfordjournals.aje.a008722. [DOI] [PubMed] [Google Scholar]
- 5.Irvine D, Davies DM. British Airways flightdeck mortality study, 1950–1992. Aviat Space Environ Med. 1999;70:548–555. [PubMed] [Google Scholar]
- 6.Pukkala E, Aspholm R, Auvinen A, et al. Cancer incidence among 10,211 airline pilots: a Nordic study. Aviat Space Environ Med. 2003;74:699–706. [PubMed] [Google Scholar]
- 7.Kubo T, Ozasa K, Mikami K, et al. Prospective cohort study of the risk of prostate cancer among rotating-shift workers: findings from the Japan collaborative cohort study. Am J Epidemiol. 2006;164:549–555. doi: 10.1093/aje/kwj232. [DOI] [PubMed] [Google Scholar]
- 8.Conlon M, Lightfoot N, Kreiger N. Rotating shift work and risk of prostate cancer. Epidemiology. 2007;18:182–183. doi: 10.1097/01.ede.0000249519.33978.31. [DOI] [PubMed] [Google Scholar]
- 9.Pukkala E, Ojamo M, Rudanko SL, Stevens RG, Verkasalo PK. Does incidence of breast cancer and prostate cancer decrease with increasing degree of visual impairment. Cancer Causes Control. 2006;17:573–576. doi: 10.1007/s10552-005-9005-6. [DOI] [PubMed] [Google Scholar]
- 10.Feychting M, Osterlund B, Ahlbom A. Reduced cancer incidence among the blind. Epidemiology. 1998;9:490–494. [PubMed] [Google Scholar]
- 11.Kondratov RV, Gorbacheva VY, Antoch MP. The role of mammalian circadian proteins in normal physiology and genotoxic stress responses. Curr Top Dev Biol. 2007;78:173–216. doi: 10.1016/S0070-2153(06)78005-X. [DOI] [PubMed] [Google Scholar]
- 12.Fu L, Lee CC. The circadian clock: pacemaker and tumour suppressor. Nat Rev Cancer. 2003;3:350–361. doi: 10.1038/nrc1072. [DOI] [PubMed] [Google Scholar]
- 13.Zhu Y, Zheng T, Stevens RG, Zhang Y, Boyle P. Does "Clock" Matter in Prostate Cancer? Cancer Epidemiology, Biomarkers and Prevention. 2006;15:3–5. doi: 10.1158/1055-9965.EPI-05-0631. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Stanford JL, Wicklund KG, McKnight B, Daling JR, Brawer MK. Vasectomy and risk of prostate cancer. Cancer Epidemiol Biomarkers Prev. 1999;8:881–886. [PubMed] [Google Scholar]
- 15.Agalliu I, Salinas CA, Hansten PD, Ostrander EA, Stanford JL. Statin use and risk of prostate cancer: results from a population-based epidemiologic study. Am J Epidemiol. 2008;168:250–260. doi: 10.1093/aje/kwn141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.de Bakker PI, Yelensky R, Pe'er I, Gabriel SB, Daly MJ, Altshuler D. Efficiency and power in genetic association studies.[see comment] Nature Genetics. 2005;37:1217–1223. doi: 10.1038/ng1669. [DOI] [PubMed] [Google Scholar]
- 17.Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21:263–265. doi: 10.1093/bioinformatics/bth457. [DOI] [PubMed] [Google Scholar]
- 18.Core Genotyping Facility, Office of Cancer Genomics, Division of Cancer Epidemiology and Genetics NCI. Bethesda (MD): National Institutes of Health The Cancer Genetic Markers of Susceptibility Project. Vol. 2006
- 19.Tobler AR, Short S, Andersen MR, et al. The SNPlex genotyping system: a flexible and scalable platform for SNP genotyping. J Biomol Tech. 2005;16:398–406. [PMC free article] [PubMed] [Google Scholar]
- 20.De la Vega FM, Lazaruk KD, Rhodes MD, Wenz MH. Assessment of two flexible and compatible SNP genotyping platforms: TaqMan SNP Genotyping Assays and the SNPlex Genotyping System. Mutat Res. 2005;573:111–135. doi: 10.1016/j.mrfmmm.2005.01.008. [DOI] [PubMed] [Google Scholar]
- 21.Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet. 2001;68:978–989. doi: 10.1086/319501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zhu Y, Brown HN, Zhang Y, Stevens RG, Zheng T. Period3 structural variation: a circadian biomarker associated with breast cancer in young women. Cancer Epidemiol Biomarkers Prev. 2005;14:268–270. [PubMed] [Google Scholar]
- 23.Zhu Y, Stevens RG, Leaderer D, et al. Non-synonymous polymorphisms in the circadian gene NPAS2 and breast cancer risk. Breast Cancer Res Treat. 2008;107:421–425. doi: 10.1007/s10549-007-9565-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Chu LW, Zhu Y, Yu K, et al. Variants in circadian genes and prostate cancer risk: a population-based study in China. Prostate Cancer Prostatic Dis. 2008;11:342–348. doi: 10.1038/sj.pcan.4501024. [DOI] [PubMed] [Google Scholar]
- 25.Storch KF, Lipan O, Leykin I, et al. Extensive and divergent circadian gene expression in liver and heart. Nature. 2002;417:78–83. doi: 10.1038/nature744. [DOI] [PubMed] [Google Scholar]
- 26.Fu L, Pelicano H, Liu J, Huang P, Lee C. The circadian gene Period2 plays an important role in tumor suppression and DNA damage response in vivo. Cell. 2002;111:41–50. doi: 10.1016/s0092-8674(02)00961-3. [DOI] [PubMed] [Google Scholar]
- 27.Matsuo T, Yamaguchi S, Mitsui S, Emi A, Shimoda F, Okamura H. Control Mechanism of the Circadian Clock for Timing of Cell Division in Vivo. Science. 2003;302:255–259. doi: 10.1126/science.1086271. [DOI] [PubMed] [Google Scholar]
- 28.Granda TG, Liu XH, Smaaland R, et al. Circadian regulation of cell cycle and apoptosis proteins in mouse bone marrow and tumor. Faseb J. 2005;19:304–306. doi: 10.1096/fj.04-2665fje. [DOI] [PubMed] [Google Scholar]
- 29.Lee E, Salic A, Kirschner MW. Physiological regulation of [beta]-catenin stability by Tcf3 and CK1epsilon. J Cell Biol. 2001;154:983–993. doi: 10.1083/jcb.200102074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.van de Wetering M, de Lau W, Clevers H. WNT signaling and lymphocyte development. Cell. 2002;109 Suppl:S13–S19. doi: 10.1016/s0092-8674(02)00709-2. [DOI] [PubMed] [Google Scholar]
- 31.Morin PJ. beta-catenin signaling and cancer. Bioessays. 1999;21:1021–1030. doi: 10.1002/(SICI)1521-1878(199912)22:1<1021::AID-BIES6>3.0.CO;2-P. [DOI] [PubMed] [Google Scholar]
- 32.Gery S, Komatsu N, Baldjyan L, Yu A, Koo D, Koeffler HP. The circadian gene per1 plays an important role in cell growth and DNA damage control in human cancer cells. Mol Cell. 2006;22:375–382. doi: 10.1016/j.molcel.2006.03.038. [DOI] [PubMed] [Google Scholar]
- 33.Hoffman AE, Zheng T, Ba Y, Zhu Y. The circadian gene NPAS2, a putative tumor suppressor, is involved in DNA damage response. Mol Cancer Res. 2008;6:1461–1468. doi: 10.1158/1541-7786.MCR-07-2094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Brown CE, Warren S, Chute RN, Ryan KJ, Todd RB. Hormonally induced tumors of the reproductive system of parabiosed male rats. Cancer Res. 1979;39:3971–3976. [PubMed] [Google Scholar]
- 35.Gann PH, Hennekens CH, Ma J, Longcope C, Stampfer MJ. Prospective study of sex hormone levels and risk of prostate cancer. J Natl Cancer Inst. 1996;88:1118–1126. doi: 10.1093/jnci/88.16.1118. [DOI] [PubMed] [Google Scholar]
- 36.Roddam AW, Allen NE, Appleby P, Key TJ. Endogenous sex hormones and prostate cancer: a collaborative analysis of 18 prospective studies. J Natl Cancer Inst. 2008;100:170–183. doi: 10.1093/jnci/djm323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Thompson IM, Goodman PJ, Tangen CM, et al. The influence of finasteride on the development of prostate cancer. N Engl J Med. 2003;349:215–224. doi: 10.1056/NEJMoa030660. [DOI] [PubMed] [Google Scholar]
- 38.Soronen P, Laiti M, Torn S, et al. Sex steroid hormone metabolism and prostate cancer. J Steroid Biochem Mol Biol. 2004;92:281–286. doi: 10.1016/j.jsbmb.2004.10.004. [DOI] [PubMed] [Google Scholar]
- 39.Francavilla A, Eagon PK, DiLeo A, et al. Circadian rhythm of hepatic cytosolic and nuclear estrogen and androgen receptors. Gastroenterology. 1986;91:182–188. doi: 10.1016/0016-5085(86)90456-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Plymate SR, Tenover JS, Bremner WJ. Circadian variation in testosterone, sex hormone-binding globulin, and calculated non-sex hormone-binding globulin bound testosterone in healthy young and elderly men. J Androl. 1989;10:366–371. doi: 10.1002/j.1939-4640.1989.tb00120.x. [DOI] [PubMed] [Google Scholar]
- 41.Guignard MM, Pesquies PC, Serrurier BD, Merino DB, Reinberg AE. Circadian rhythms in plasma levels of cortisol, dehydroepiandrosterone, delta 4-androstenedione, testosterone and dihydrotestosterone of healthy young men. Acta Endocrinol (Copenh) 1980;94:536–545. doi: 10.1530/acta.0.0940536. [DOI] [PubMed] [Google Scholar]
- 42.Chu LW, Zhu Y, Yu K, et al. Correlation between circadian gene variants and serum levels of sex steroids and insulin-like growth factor-I. Cancer Epidemiol Biomarkers Prev. 2008;17:3268–3273. doi: 10.1158/1055-9965.EPI-08-0073. [DOI] [PubMed] [Google Scholar]
- 43.Zhu Y, Leaderer D, Guss C, et al. Ala394Thr polymorphism in the clock gene NPAS2: a circadian modifier for the risk of non-Hodgkin's lymphoma. Int J Cancer. 2007;120:432–435. doi: 10.1002/ijc.22321. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Johansson C, Willeit M, Smedh C, et al. Circadian clock-related polymorphisms in seasonal affective disorder and their relevance to diurnal preference. Neuropsychopharmacology. 2003;28:734–739. doi: 10.1038/sj.npp.1300121. [DOI] [PubMed] [Google Scholar]
- 45.Lam RW, Levitan RD. Pathophysiology of seasonal affective disorder: a review. J Psychiatry Neurosci. 2000;25:469–480. [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.