Chronobiology International, 2014; 31(10): 1115–1122 ! Informa Healthcare USA, Inc. ISSN: 0742-0528 print / 1525-6073 online DOI: 10.3109/07420528.2014.957301
ORIGINAL ARTICLE
Polymorphisms in circadian genes, night work and breast cancer: Results from the GENICA study
Chronobiol Int Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 12/03/14 For personal use only.
Sylvia Rabstein1, Volker Harth2, Christina Justenhoven3, Beate Pesch1, Sabine Plo¨ttner1, Evelyn Heinze1, Anne Lotz1, Christian Baisch4, Markus Schiffermann4, Hiltrud Brauch5, Ute Hamann6 Yon Ko4, and Thomas Bru¨ning1 on behalf of the GENICA Consortium 1
Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of the RuhrUniversita¨t Bochum (IPA), Bochum, Germany, 2Institute for Occupational Medicine and Maritime Medicine (ZfAM), University Medical Centre Hamburg-Eppendorf, Hamburg, Germany, 3Bioscientia Center for Human Genetics, Ingelheim, Germany, 4Department of Internal Medicine, Evangelische Kliniken Bonn gGmbH, Johanniter Krankenhaus, Bonn, Germany, 5Dr. Margarete Fischer-Bosch – Institute of Clinical Pharmacology, Stuttgart, and University of Tu¨bingen, Germany, and 6Molecular Genetics of Breast Cancer, German Cancer Research Center (DKFZ), Heidelberg, Germany
Objectives: The role of genetic variants and environmental factors in breast cancer etiology has been intensively studied in the last decades. Gene-environment interactions are now increasingly being investigated to gain more insights into the development of breast cancer, specific subtypes, and therapeutics. Recently, night shift work that involves circadian disruption has gained rising interest as a potential non-genetic breast cancer risk factor. Here, we analyzed genetic polymorphisms in genes of cellular clocks, melatonin biosynthesis and signaling and their association with breast cancer as well as gene–gene and gene–night work interactions in a German case-control study on breast cancer. Methods: GENICA is a population-based case-control study on breast cancer conducted in the Greater Region of Bonn. Associations between seven polymorphisms in circadian genes (CLOCK, NPAS2, ARTNL, PER2 and CRY2), genes of melatonin biosynthesis and signaling (AANAT and MTNR1B) and breast cancer were analyzed with conditional logistic regression models, adjusted for potential confounders for 1022 cases and 1014 controls. Detailed shift-work information was documented for 857 breast cancer cases and 892 controls. Gene–gene and gene–shiftwork interactions were analyzed using model-based multifactor dimensionality reduction (mbMDR). Results: For combined heterozygotes and rare homozygotes a slightly elevated breast cancer risk was found for rs8150 in gene AANAT (OR 1.17; 95% CI 1.01–1.36), and a reduced risk for rs3816358 in gene ARNTL (OR 0.82; 95% CI 0.69–0.97) in the complete study population. In the subgroup of shift workers, rare homozygotes for rs10462028 in the CLOCK gene had an elevated risk of breast cancer (OR for AA vs. GG: 3.53; 95% CI 1.09–11.42). Shift work and CLOCK gene interactions were observed in the two-way interaction analysis. In addition, gene–shiftwork interactions were detected for MTNR1B with NPAS2 and ARNTL. Conclusions: In conclusion, the results of our population-based case-control study support a putative role of the CLOCK gene in the development of breast cancer in shift workers. In addition, higher order interaction analyses suggest a potential relevance of MTNR1B with the key transcriptional factor NPAS2 with ARNTL. Hence, in the context of circadian disruption, multivariable models should be preferred that consider a wide range of polymorphisms, e.g. that may influence chronotype or light sensitivity. The investigation of these interactions in larger studies is needed. Keywords: Case-control study, genetic variants, shift work
INTRODUCTION The role of genetic variants and environmental factors in breast cancer etiology has been intensively studied in the last decades. Genetic variation is estimated to explain about 30% of familial breast cancers (Mavaddat et al., 2010; Maxwell & Nathanson, 2013). Among the
most important known risk factors are menstrual and reproductive history and menopausal hormone therapy. As low penetrance variants might interact with environmental factors and contribute to breast cancer development, gene–environment interactions are now increasingly being investigated to gain more insights
Submitted February 10, 2014, Returned for revision August 11, 2014, Accepted August 18, 2014
Correspondence: Sylvia Rabstein, PhD, Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of the Ruhr-Universita¨t Bochum (IPA), Bu¨rkle-de-la-Camp-Platz 1, 44789 Bochum, Germany. Tel: +49 234 3024595. Fax: +49 234 302-4505. E-mail:
[email protected] 1115
Chronobiol Int Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 12/03/14 For personal use only.
1116
S. Rabstein et al.
into the development of breast cancer, specific subtypes and therapeutics (Garcia-Closas et al., 2013; Michailidou et al., 2013; Nickels et al., 2013; Schoeps et al., 2014). Recently, night shift work that involves circadian disruption has gained rising interest as a potential nongenetic risk factor (Straif et al., 2007). Since the International Agency for Research on Cancer classified ‘‘night work that includes circadian disruption’’ as probably carcinogenic (group 2A) in 2007, several studies were published on the association between shift work and breast cancer risk (Fritschi et al., 2013; Grundy et al., 2013a, b; Hansen & Lassen, 2012; Hansen & Stevens, 2012; Knutsson et al., 2013; Lie et al., 2011, 2013; Menegaux et al., 2013; Pesch et al., 2010; Pronk et al., 2010; Rabstein et al., 2013). Different exposure definitions of night-shift exposure have been used including lifetime number of night shifts, years in shift systems with permanent and/or rotational night shift, number of subsequent night shifts, different time frames for start and finish of night work, and night shift work before first birth. Furthermore, the influence of sleep quality and chronotype-specific subgroup analysis of night shift risks were investigated (Fritschi et al., 2013; Girschik et al., 2013; Hansen & Lassen, 2012). Most of these studies show elevated risk estimates with higher levels of investigated exposure metrics. However, the assessment of shift work, chronodisruption and breast cancer in epidemiology still remains one important issue (Erren & Morfeld, 2014; Stevens et al., 2011). Three meta-analyses were recently published on shift work and breast cancer and have been intensively discussed (Ijaz et al., 2013a, b; Jia et al., 2013; Kamdar et al., 2013; Stevens & Hansen, 2013; Stevens et al., 2013; Wang et al., 2013). The most prominent hypothesized mechanisms linking shift work with breast cancer is the lightat-night (LAN) hypothesis (Davis et al., 2001; Fritschi et al., 2011; Stevens, 2005, 2009). It implies changes in hormonal secretion and a wide range of cellular processes due to altered N-acetyl-5-methoxytryptamine (melatonin) secretion when exposed to light during the night. The hormone melatonin is secreted by the pineal gland with daily and seasonal rhythm, specifically in response to darkness. One of the key enzymes for the biosynthesis of melatonin is arylalkylamine N-acetyltransferase (AANAT) whose activity is higher at night than during the day. The interplay between melatonin and cellular clocks in peripheral tissue may be mediated via melatonin receptors (Blask et al., 2011; Reppert et al., 1995). Among the various physiological functions, melatonin has been suggested to entail oncostatic properties (Simonneaux & Ribelayga, 2003). Especially for hormone-dependent tumors increased melatonin levels have been found to be associated with decreased breast cancer risk (Cos & Sa´nchez-Barcelo´, 2000; Mediavilla et al., 2010; Schernhammer & Hankinson, 2009).
From transcriptome-profiling studies it has been concluded that up to 10% of genes have a circadian pattern of expression (Brown et al., 2008; Gachon et al., 2004). On the molecular level the cellular clocks are organized in several transcriptional–translational feedback loops that regulate the expression of circadian genes. The key transcriptional activator of the cellular clocks is a heterodimer that consists of either circadian locomotor output cycles kaput (CLOCK) or neuronal PAS domain containing protein 2 (NPAS2) and aryl hydrocarbon receptor nuclear translocator-like (ARNTL), also known as BMAL1. It binds to E-box elements in the promoters of period homologues (PERs) and cryptochrome (CRYs) genes and activates their transcription. PER and CRY proteins heterodimerize in the cytosol, translocate to the nucleus and inhibit their own expression upon interaction with the CLOCK/ ARNTL complex. Several other circadian genes like CSNK1E, REV-ERBA, TIMELESS are also involved in the cellular clocks (for review, see Kelleher et al., 2014; Takahashi et al., 2008). They may act as tumor suppressors at the systemic and cellular levels due to their involvement in cell proliferation, apoptosis, cell cycle control and DNA damage response (Fu & Lee, 2003). Genetic polymorphisms in circadian genes and in genes relevant for the melatonin biosynthesis and signaling may influence the rhythmic interplay between external ‘‘zeitgebers’’, primary and secondary biological rhythms. Hence, they may contribute to the variations in internal desynchronization caused by night shift work. Several studies have investigated polymorphisms in circadian genes in studies of breast cancer (Dai et al., 2011; Deming et al., 2012; Grundy et al., 2013b; Monsees et al., 2012; Truong et al., 2014; Zienolddiny et al., 2013). So far, no clear pattern can be seen from the results from these studies that observed different selections and number of polymorphisms as well as different populations. Here, we investigated whether polymorphisms in circadian genes and genes of melatonin biosynthesis and signaling have an influence on the chronodisruptive effects of night shift work and hence modify the breast cancer risk of night shift work. We analyzed seven variants in a set of circadian candidate genes and additionally investigated gene–gene and gene–night work interactions the large population-based German breast cancer case-control study GENICA with a detailed assessment of shift work history.
MATERIAL AND METHODS Study population and breast cancer risk factors GENICA is a population-based case-control study that was conducted in the Greater Region of Bonn, Germany (Pesch et al., 2005). Overall, 1143 breast cancer cases and 1155 frequency-matched population controls were enrolled between 2000 and 2004 with response rates of 88% for cases and 67% for controls. Inclusion criteria Chronobiology International
Chronobiol Int Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 12/03/14 For personal use only.
Circadian genes, night work and breast cancer comprised age 80 years and Caucasian ethnicity. Incident cases were women with cancer diagnosed within six months before enrolment. Data on known and suspected risk factors, including a detailed occupational history, were obtained by in-person interviews. Information on shift work – including ever or never having worked in shift work or night shift work with time frames – was assessed between 2004 and 2007 in subsequent telephone interviews for 857 GENICA cases and 892 controls (Kropp et al., 2007; Pesch et al., 2010). Among 247 GENICA women reporting shift work, 223 women (104 cases, 119 controls) could be reached for another telephone interview to collect detailed shiftwork information for each occupational period. Night work was defined as working the full time period between midnight and 5 a.m. Shift work and other information was truncated at the date of the core interview. All participants gave written informed consent. The GENICA study is in line with the ethical standards for the conduct of human research and was approved by the ethics committee of the University of Bonn (Portaluppi et al., 2010).
Selection of polymorphisms and genotype analysis Seven polymorphisms in genes of the circadian clock and genes of melatonin biosynthesis and melatonin signaling were selected for genotyping. These included rs10462028 in CLOCK, rs11123857 in NPAS2, rs3816358 in ARNTL, rs10838524 in CRY2, and rs10462023 in PER2 for the circadian clock, rs8150 in AANAT for the melatonin biosynthesis, and rs10765576 in MTNR1B for melatonin signaling. Polymorphisms were selected considering a minor allele frequency above 0.15 and known or putative functional consequences. Genotyping was performed by blood-derived DNA using SequenomÕ matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) (Sequenom, San Diego, CA). Call rates were above 98% for all polymorphisms. A sample of 5% was randomly selected for repeated assays and concordance was 100% for these duplicates. Fisher’s exact tests for Hardy– Weinberg-Equilibrium were applied for genotypes in cases and controls for quality checks resulting in p values above 0.10. Statistical analyses Analyses of associations between potential confounders, polymorphisms and breast cancer Potential confounders were checked for the inclusion in adjusted models (Pesch et al., 2010). The set of potential confounders comprised menopausal status (pre- or post-menopausal), education, breast cancer in mother or sister, parity (nulliparous, 1–2 children, 3 children), age at first birth (nulliparous, 525 years, 25 to 530 years, 30 years), duration of oral contraceptive (OC) use and duration of HT use (never, 40 to 510 years, 10 years), body-mass index (BMI) (522.5 kg/m2, 22.5 to 525 kg/m2, 25 to 530 kg/m2, 30 kg/m2), !
Informa Healthcare USA, Inc.
1117
smoking as packyears (0 packyears, 40 to 15 packyears, 15 packyears), number of mammograms until two years before interview, and lifetime breast-feeding in months. Women were considered pre-menopausal if they reported bleedings in the year of interview and no bilateral oophorectomy. The associations between polymorphisms and breast cancer risk were investigated for the complete study population with available DNA samples with logistic regression models conditional on age (20–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70 years). Adjusted models included the confounders’ family history of breast cancer, HT use, and number of mammograms. Analyses were repeated in the group of women who were ever employed in shift work for at least one year. Risk estimates were calculated as odds ratios (ORs) with 95% confidence intervals (CIs) by using PHREG procedure in SAS software, version 9.2 (SAS Institute Inc., Cary, NC). Permutation tests for each SNP were provided with 10 000 permutations. Bayesian False Discovery Probability was calculated with a prior probability of 0.15 for significant associations, roughly assuming that one of the selected seven polymorphisms might be of relevance in breast cancer development and chosing the threshold for noteworthiness as 0.8 (Wakefield, 2007).
Gene–gene and gene–shiftwork interactions For the analysis of gene–gene and gene–shiftwork interactions we applied model-based Multifactor Dimensionality Reduction (mbMDR) described in detail elsewhere (Calle et al., 2008). This is a modelbased approach to identify combinations of multilocus genotypes that are associated with high or low risks of disease while adjusting for confounders. For this method, we created a shift work variable with three categories based on shift work exposure with the reference category of no shift work exposure and with the highest category of a long duration of night shift work (never employment in shift work or never employment, ever shift or less than 20 years of night work, night work for 20 years or longer). We used 1000fold permutation testing and a co-dominant model. For the search of interactions two adjustment strategies were applied, (1) adjustment for potential main effects of the included variables, and adjusted for confounder’s age, family history of breast cancer, hormone therapy use (HT use), and number of mammograms until two years before the interview; (2) adjustment for confounders only. Results are presented for p values below 0.05. RESULTS A description of the study population with ORs for selected confounders is shown in Table 1. Higher breast cancer risks were observed for long duration of HT use,
1118
S. Rabstein et al.
TABLE 1. Characteristics of the study population for breast cancer cases and controls.
Chronobiol Int Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 12/03/14 For personal use only.
Cases (N ¼ 1022) Age (years) –550 226 50–560 299 60–570 337 70–80 160 p Fisher’s exact test Menopausal status Pre 249 Post 759 Hormone therapy (years) Never 507 40–510 245 10 266 Smoking (in packyears) 0 589 40–515 166 15 204 Body-mass index (kg/m2) 522.5 305 22.5–525 257 25–530 305 30 155 Breast cancer in mother or No 890 Yes 132 No. of mammogramsb None 149 1–510 667 10 204
(22.1%) (29.3%) (33.0%) (15.7%)
Controls (N ¼ 1014) 226 289 335 164
ORa (95% CI)
(22.3%) (28.5%) (33.0%) (16.2%) 0.979
(24.7%) 235 (23.6%) 1 (75.3%) 763 (76.5%) 0.90 (0.65–1.24) (49.8%) 508 (50.2%) 1 (24.1%) 290 (28.7%) 0.84 (0.67–1.05) (26.1%) 214 (21.2%) 1.30 (1.01–1.67) (61.4%) 554 (58.6%) 1 (17.3%) 185 (19.6%) 0.84 (0.65–1.07) (21.3%) 206 (21.8%) 0.92 (0.73–1.16) (29.8%) (25.2%) (29.8%) (15.2%) sisters (87.1%) (12.9%)
300 243 324 146
(29.6%) (24.0%) (32.0%) (14.4%)
1 0.96 (0.76–1.22) 0.89 (0.70–1.13) 1.01 (0.76–1.34)
939 (92.6%) 1 75 (7.4%) 1.85 (1.38–2.50)
(14.6%) 175 (17.3%) 1 (65.4%) 691 (68.3%) 1.16 (0.90–1.50) (20.0%) 146 (14.4%) 1.71 (1.24–2.37)
Statistical significance (p50.05) is indicated in bold. a Logistic regression conditional on age in 5-year groups. b Number of mammograms for cases calculated as number of mammograms until two years before diagnosis.
a family history of breast cancer, and 10 or more mammograms. Genotype frequencies and risk estimates of seven polymorphisms in circadian genes and genes of melatonin biosynthesis and signaling in the complete study population and in the subgroup of women that were ever shift workers for at least one year are given in Table 2. For combined heterozygotes and rare homozygotes a slightly elevated breast cancer risk was found for rs8150 in gene AANAT (OR 1.17; 95% CI 1.01–1.36), and a reduced risk for rs3816358 in gene ARNTL (OR 0.82; 95% CI 0.69–0.97) in the complete study population. For these variants, no noticeable risk effects were detected in the subgroup of ever shift workers. Conversely, rare homozygotes for rs10462028 in CLOCK had an elevated risk of 3.53 (95% CI 1.09–11.42) compared to common homozygotes in the subgroup of shift workers, while no associations were observed in the complete study population (GA vs. GG: OR 0.94 (95% CI 0.79–1.12) and AA versus GG: OR 0.88 (95% CI 0.68–1.16), respectively. Bayesian False Discovery Probabilities for these associations were below 0.8, indicating a noteworthiness of associations. Gene–shiftwork work interactions with
permutation p values below 0.05 are presented in Table 3. Shift work and CLOCK gene interactions were observed in the two-way interaction analysis, but not in the analysis of higher order interactions. For gene–gene interactions, 2 two-way interactions (ARNTL/MTNR1B and MTNR1B/NPAS2) with a p value below 0.05 were observed that resulted in lowest interaction p values of three-way gene–gene interactions (MTNR1B/ARNTL/ NPAS2, p ¼ 0.001 in model adjusted for confounders, and p ¼ 0.002 in model adjusted for main effects and confounders, data not shown). Hence, genotype combinations of MTNR1B, NPAS2 and ARNTL might be relevant in the context of higher order interactions. High effect estimates were found for gene shift work combinations with variant rs10462023 in PER2.
DISCUSSION In this study, we investigated the associations between polymorphisms in candidate genes of the cellular clocks, melatonin biosynthesis, and melatonin signaling and breast cancer. We analyzed the interactions between these polymorphisms and shift work. We observed a higher risk of breast cancer in carriers of the AA genotype of CLOCK gene variant rs10462028 in women that were ever shift workers. Previous studies reported significant associations between CLOCK gene variants and breast cancer with different selections of polymorphisms. So far, the function of CLOCK polymorphism rs10462028 that is situated in 30 UTR region is not known. Varying frequencies of genotype AA were found in different populations with higher proportions in Europeans (see Database SNP URL: http://www.ncbi.nlm.nih.gov). Breast cancer risk estimates of polymorphisms were found to be both increased and decreased. Reduced risk estimates were seen in two studies that investigated a high number of polymorphisms together with shift work (Grundy et al., 2013b; Monsees et al., 2012). Three studies observed higher risk estimates for CLOCK gene variants for the complete study population or in the subgroup of postmenopausal women (Dai et al., 2011; Hoffman et al., 2010; Truong et al., 2014). Hoffmann et al. found that hypermethylation of CLOCK promoter reduced the risk of breast cancer. In another study by Zhu et al. (2011), long-term shift work resulted in a promoter hypomethylation of the gene. Therefore, a potential role of CLOCK in breast cancer etiology seems plausible. The role of melatonin in the link between shift work and breast cancer is still unclear. A recent paper by Archer and colleagues showed that the timing of expression of the majority of genes with a circadian rhythm was influenced by changes in the sleep–wake rhythm, while the melatonin rhythm was not shifted due to low light influence (Archer et al., 2014). This may point to a minor role of melatonin in temporal organization of molecular circadian processes. However, night-shift work is often connected to LAN and a Chronobiology International
!
Informa Healthcare USA, Inc.
CC CA AA CA + AA
GG GA AA GA + AA
GG GA AA GA + AA GG GA AA GA + AA AA AG GG AG + GG GG GA AA GA + AA
ARNTL rs3816358
CLOCK rs10462028
CRY2 rs10838524
331 478 199 677 329 473 199 672 500 405 85 490 398 461 142 603
454 430 108 538
778 192 13 205
493 401 95 496
301 471 224 695 304 491 201 692 451 438 104 542 394 448 154 602
433 440 122 562
734 239 22 261
554 366 75 441
Controls (N ¼ 1014) N
1 0.92 (0.75–1.13) 0.81 (0.63–1.03) 0.89 (0.73–1.07) 1 0.89 (0.73–1.09) 0.91 (0.71–1.17) 0.90 (0.74–1.08) 1 0.83 (0.69–1.00) 0.74 (0.54–1.01) 0.81 (0.68–0.97) 1 1.02 (0.84–1.23) 0.91 (0.70–1.19) 0.99 (0.83–1.19)
1 0.93 (0.77–1.12) 0.84 (0.63–1.13) 0.91 (0.77–1.09)
1 0.76 (0.61–0.94) 0.56 (0.28–1.12) 0.74 (0.60–0.91)
1 1.23 (1.02–1.48) 1.42 (1.03–1.97) 1.26 (1.06–1.51)
OR (95% CI)b
1 0.94 0.85 0.91 1 0.91 0.92 0.91 1 0.90 0.80 0.88 1 1.01 0.94 1.00 (0.85–1.21) (0.74–1.21) (0.84–1.18)
(0.76–1.06) (0.60–1.06) (0.75–1.03)
(0.76–1.09) (0.73–1.16) (0.77–1.09)
(0.78–1.13) (0.68–1.05) (0.76–1.08)
1 0.94 (0.79–1.12) 0.88 (0.68–1.16) 0.93 (0.79–1.09)
1 0.83 (0.70–1.00) 0.70 (0.39–1.25) 0.82 (0.69–0.97)
1 1.14 (0.97–1.35) 1.28 (0.96–1.70) 1.17 (1.01–1.36)
OR (95% CI)c
Overall
0.712
0.057
0.517
0.237
0.481
0.013 (0.772)
0.024 (0.787)
pperm (BFDP)d
29 33 22 55 30 34 19 53 39 36 8 44 37 35 11 46
35 36 12 48
67 15 1 16
51 27 7 34
Cases (N ¼ 85) N
28 50 22 72 36 45 19 64 47 44 9 53 32 56 12 68
44 50 5 55
76 20 3 23
57 35 8 43
Controls (N ¼ 101) N
1 0.63 0.85 0.70 1 0.92 1.20 1.01 1 0.98 1.23 1.02 1 0.53 0.74 0.57
(0.28–1.01) (0.28–1.96) (0.31–1.05)
(0.52–1.84) (0.43–3.57) (0.56–1.86)
(0.46–1.83) (0.53–2.72) (0.54–1.89)
(0.32–1.26) (0.38–1.91) (0.37–1.33)
1 0.96 (0.51–1.80) 3.30 (1.03–10.6) 1.16 (0.63–2.11)
1 0.88 (0.41–1.87) 0.37 (0.04–3.75) 0.81 (0.39–1.68)
1 0.87 (0.45–1.67) 0.99 (0.33–2.98) 0.89 (0.48–1.64)
OR (95% CI)b
1 0.57 0.81 0.65 1 0.76 0.87 0.79 1 1.08 0.92 1.05 1 0.66 0.84 0.70
(0.33–1.31) (0.31–2.30) (0.36–1.33)
(0.56–2.07) (0.29–2.90) (0.56–1.96)
(0.37–1.54) (0.36–2.07) (0.41–1.52)
(0.28–1.18) (0.35–1.89) (0.33–1.26)
1 0.94 (0.48–1.85) 3.53 (1.09–11.42) 1.19 (0.63–2.25)
1 0.78 (0.35–1.74) 0.24 (0.02–2.70) 0.70 (0.33–1.51)
1 0.88 (0.45–1.72) 0.87 (0.27–2.82) 0.88 (0.47–1.64)
OR (95% CI)c
Shift workersa
Statistical significance (p50.05) is indicated in bold. a Shift work was defined as ever employment in shift work for at least one year. b Logistic regression conditional on age in 5-year groups. c Logistic regression conditional on age in 5-year groups, adjusted for family history of breast cancer, hormone replacement use, and number of mammograms. d perm p : p value of permutation testing, BFDP: Bayesian False Discovery Probability.
PER2 rs10462023
NPAS2 rs11123857
MTNR1B rs10765576
GG GC CC GC + CC
AANAT rs8150
Polymorphism
Cases (N ¼ 1022) N
0.156
0.988
0.778
0.344
0.087 (0.845)
0.631
0.897
pperm (BFDP)d
TABLE 2. Associations between candidate polymorphisms in circadian genes, melatonin biosynthesis, melatonin signaling and breast cancer in the whole study population and in shift workers.
Chronobiol Int Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 12/03/14 For personal use only.
Circadian genes, night work and breast cancer 1119
1120
S. Rabstein et al.
TABLE 3. Gene–shiftwork interactions assessed by model-based multifactor dimensionality reduction. Factors
NHa
eb
Two-wayd,e Shift work CLOCK 1 3.12 Three-waye Adjusted for confounders Shift work MTNR1B NPAS2 1 1.26 Shift work MTNR1B ARTNL 1 1.26 Shift work AANAT PER2 1 4.28 Shift work CRY2 PER2 1 5.82 Adjusted for main effects and confounders Shift work NPAS2 PER2 3 1.43 Shift work CRY2 PER2 1 5.82
NLb
eb
0
ppermc 0.021
1 1 0 1
0.75 0.75 0.54
0.042 0.032 0.045 0.026
1 1
0.78 0.54
0.034 0.048
Chronobiol Int Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 12/03/14 For personal use only.
a
Number of high-risk combinations of variables. Number of low-risk combinations of variables. c perm p : p value of permutation testing. d Adjusted for main effects and confounders. e Shift work in categories ‘‘never employment in shift work or never employment’’, ‘‘ever shift or less than 20 years of night work’’, and ‘‘night work for 20 years or longer’’. b
change in melatonin rhythm. Archer et al. also reported that the time course of expression for a small fraction of genes (about 1%) remained in synchrony with the melatonin rhythm when the sleep–wake rhythm was shifted. In our study, we analyzed a candidate polymorphism in gene AANAT, which is a key enzyme in melatonin biosynthesis and has been found to be associated with major depression and bipolar disorder (Soria et al., 2010). In a study by Zienolddiny et al. (2013), two different genetic variants of AANAT showed elevated breast cancer risk estimates. This effect was not seen in the study by Deming et al. (2012) that included the same polymorphism in a Chinese population. We found a higher risk of breast cancer for AANAT polymorphisms rs8150 for the combined genotype GC and CC for the complete study population and not in shift workers. Model-based MDR revealed a high risk combination with shift work and PER2. However, the highest effect levels were observed for different combinations including PER2. Hence, our data would rather point to a minor role of rs8150 in breast cancer development and in interaction with shift work. We observed a significant three-way interaction between MTNR1B/NPAS2/ARNTL that suggests a role of the melatonin receptors in higher order interactions in the circadian regulation. Melatonin receptors are known to be the main mediators of the anti-proliferative effects of melatonin (Hill et al., 2011). Polymorphisms in this gene were found to be related to the regulation of insulin secretion (Zhang et al., 2013). A set of 14 polymorphisms in MTNR1A, MTNR1B and AANAT was investigated in the Shanghai Breast Cancer Study, where polymorphism MTNR1B rs10765576 showed a lower breast cancer risk (Deming et al., 2012). We observed no significant main effect for this polymorphism but noteworthy interactions with shift work/NPAS2, and shift
work/ARNTL that included not only high- but also lowrisk combinations. One of the observed gene–gene three-way interactions with p values below 0.01 was MTNR1B/NPAS2/ARNTL. Monsees and colleagues also found an interaction for a polymorphism in NPAS2 with shift work. We observed high effect levels for combinations with the period gene PER2. Expression levels of PER2 were found to be decreased in breast tumors (Winter et al., 2007). In addition, PER2 expression was found to be associated with an up-regulation of p53 in breast cancer cells (Xiang et al., 2008). Dai et al. (2011) also found higher risks for combinations of PER2 with CLOCK. Generally, gene–gene interactions have to be interpreted with caution as combinations that are repeatedly found in lower-order as well as higherorder interactions may simply reflect one specific gene–gene combination. Here, we concluded that small subgroups of interactions between several genes might be of importance. Different origins of study populations as well as differing SNP selections make comparisons and interpretation difficult. In addition, shift work exposures might be different regarding shift systems, industries and other factors. The fraction of shift workers, especially night workers, in female populations varies. In the study by Grundy and colleagues, 31% of women were night shift workers for at least 2 years. This fraction seems quite large in comparison to the German GENICA population with about 10% of women working ever night shift. Legislation and lifestyle differences may explain these disagreements to some extent. We reported elevated but non-significant risks for long duration of night-shift work in the GENICA study population, and a significant association for estrogen receptor negative tumors (Pesch et al., 2010; Rabstein et al., 2013). Hence, stratified analyses might be important in the investigation of the association between shift work and breast cancer. Several other studies also investigated the effects of polymorphisms for specific subgroups of women. However, we decided not to present further subgroups, e.g. by menopausal status or by body-mass index, as they lead to similar but non-significant risk estimates due to a smaller number of subjects. For example, women who were employed in night work for at least two years also showed an increased but non-significant risk estimate for CLOCK rs10462028 of OR 3.51 (95% CI 0.58–21.3, data not shown). Noteworthiness of associations, e.g. using Bayesian False Discovery Probability, strongly depends on the a priori assumptions. This study was not initially designed to investigate questions regarding night work and, hence, has a rather explorative character. The strength of our study is that we chose a statistical method for interaction analysis that allows adjustment for important confounders like family history of breast cancer. We did not assess chronotype which is one of the shortcomings of this study. However, it is not yet Chronobiology International
Chronobiol Int Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 12/03/14 For personal use only.
Circadian genes, night work and breast cancer clear if current chronotype at the time of interview in a retrospective study reflects the chronotype at the time of exposure, e.g. exposure to shift work. Recently, the role of chronotype in shift work research has been increasingly discussed. Hansen & Lassen (2012) found a nearly four-fold breast cancer risk for high night work exposure in women with morning preference. Allebrandt & Roenneberg (2008) illustrated that various circadian polymorphisms may interfere with respect to chronotype. Therefore, analysis of multiple variants and their gene–gene interactions in sufficiently large study populations is important. Although the analysis of SNPs may not necessarily have a preventive character, it may provide helpful insights for the identification of relevant processes in cancer etiology. The link between cancer development and shift work is still unclear, especially with respect to tissue specificity. Night work may act as a promoting factor in the presence of occupational exposures. Therefore, the analysis of genetic polymorphisms in shift-workers is one possible step in disentangling the link between night work, chronodisruption and cancer etiology. In conclusion, the results of our population-based case-control study support a possible role of the CLOCK gene in the development of breast cancer under shift work. Higher order interaction analyses reveal a potential relevance of MTNR1B and the key transcriptional factor NPAS2 with ARNTL. The investigation of interactions of specific polymorphism and shift work in larger studies is needed.
DECLARATION OF INTEREST The authors report no conflicts of interest. SR, BP, SP, EH, AL and TB, as staff of the Institute for Prevention and Occupational Medicine (IPA), are employed at the ‘‘Berufsgenossenschaft Rohstoffe und chemische Industrie’’ (BG RCI), a public body, which is a member of the study’s main sponsor, the German Social Accident Insurance. IPA is an independent research institute of the Ruhr-Universita¨t Bochum. The authors are independent from the German Social Accident Insurance in study design, access to the collected data, responsibility for data analysis and interpretation, and the right to publish. The views expressed in this paper are those of the authors and not necessarily those of the sponsor. The authors alone are responsible for the content and writing of the paper. This work was supported by the German Federal Ministry of Education and Research (BMBF) grants 01KW9975/5, 01KW9976/8, 01KW9977/0, 01KW0114 and 01KH0411, Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Ruhr-University Bochum (IPA), Robert Bosch Foundation, Stuttgart, Evangelische Kliniken Bonn GmbH, and Deutsches Krebsforschungszentrum, Heidelberg. !
Informa Healthcare USA, Inc.
1121
REFERENCES Allebrandt KV, Roenneberg T. (2008). The search for circadian clock components in humans: New perspectives for association studies. Braz J Med Biol Res. 41:716–21. Archer SN, Laing EE, Mo¨ller-Levet CS, et al. (2014). Mistimed sleep disrupts circadian regulation of the human transcriptome. Proc Natl Acad Sci U.S.A. 111:E682–91. Blask DE, Hill SM, Dauchy RT, et al. (2011). Circadian regulation of molecular, dietary, and metabolic signaling mechanisms of human breast cancer growth by the nocturnal melatonin signal and the consequences of its disruption by light at night. J Pineal Res. 51:259–69. Brown SA, Kunz D, Dumas A, et al. (2008). Molecular insights into human daily behavior. Proc Natl Acad Sci U.S.A. 105: 1602–7. Calle ML, Urrea V, Vellalta G, et al. (2008). Improving strategies for detecting genetic patterns of disease susceptibility in association studies. Stat Med. 27:6532–46. Cos S, Sa´nchez-Barcelo´ EJ. (2000). Melatonin and mammary pathological growth. Front Neuroendocrinol. 21:133–70. Dai H, Zhang L, Cao M, et al. (2011). The role of polymorphisms in circadian pathway genes in breast tumorigenesis. Breast Cancer Res Treat. 127:531–40. Davis S, Mirick DK, Stevens RG. (2001). Night shift work, light at night, and risk of breast cancer. J Natl Cancer Inst. 93: 1557–62. Deming SL, Lu W, Beeghly-Fadiel A, et al. (2012). Melatonin pathway genes and breast cancer risk among Chinese women. Breast Cancer Res Treat. 132:693–9. Erren TC, Morfeld P. (2014). Computing chronodisruption: How to avoid potential chronobiological errors in epidemiological studies of shift work and cancer. Chronobiol Int. 31:589–99. Fritschi L, Erren TC, Glass DC, et al. (2013). The association between different night shiftwork factors and breast cancer: A case-control study. Br J Cancer. 109:2472–80. Fritschi L, Glass DC, Heyworth JS, et al. (2011). Hypotheses for mechanisms linking shiftwork and cancer. Med Hypotheses. 77: 430–6. Fu L, Lee CC. (2003). The circadian clock: Pacemaker and tumour suppressor. Nat Rev Cancer. 3:350–61. Gachon F, Nagoshi E, Brown SA, et al. (2004). The mammalian circadian timing system: From gene expression to physiology. Chromosoma. 113:103–12. Garcia-Closas M, Couch FJ, Lindstrom S, et al. (2013). Genomewide association studies identify four ER negative-specific breast cancer risk loci. Nat Genet. 45:392–8, 398e1–2. Girschik J, Heyworth J, Fritschi L. (2013). Self-reported sleep duration, sleep quality, and breast cancer risk in a populationbased case-control study. Am J Epidemiol. 177:316–27. Grundy A, Richardson H, Burstyn I, et al. (2013a). Increased risk of breast cancer associated with long-term shift work in Canada. Occup Environ Med. 70:831–8. Grundy A, Schuetz JM, Lai AS, et al. (2013b). Shift work, circadian gene variants and risk of breast cancer. Cancer Epidemiol. 37: 606–12. Hansen J, Lassen CF. (2012). Nested case-control study of night shift work and breast cancer risk among women in the Danish military. Occup Environ Med. 69:551–6. Hansen J, Stevens RG. (2012). Case-control study of shift-work and breast cancer risk in Danish nurses: Impact of shift systems. Eur J Cancer. 48:1722–9. Hill SM, Blask DE, Xiang S, et al. (2011). Melatonin and associated signaling pathways that control normal breast epithelium and breast cancer. J Mammary Gland Biol Neoplasia. 16:235–45. Hoffman AE, Zheng T, Yi C, et al. (2010). The core circadian gene Cryptochrome 2 influences breast cancer risk, possibly by mediating hormone signaling. Cancer Prev Res (Phila). 3: 539–48.
Chronobiol Int Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 12/03/14 For personal use only.
1122
S. Rabstein et al.
Ijaz S, Verbeek J, Seidler A, et al. (2013a). Night-shift work and breast cancer – A systematic review and meta-analysis. Scand J Work Environ Health. 39:431–47. Ijaz S, Verbeek J, Seidler A, et al. (2013b). Response to letter to the editor, re: Ijaz S, et al. ‘‘Night-shift work and breast cancer – a systematic review and meta-analysis". Scand J Work Environ Health. 39:633–4. Jia Y, Lu Y, Wu K, et al. (2013). Does night work increase the risk of breast cancer? A systematic review and meta-analysis of epidemiological studies. Cancer Epidemiol. 37:197–206. Kamdar BB, Tergas AI, Mateen FJ, et al. (2013). Night-shift work and risk of breast cancer: A systematic review and metaanalysis. Breast Cancer Res Treat. 138:291–301. Kelleher FC, Rao A, Maguire A. (2014). Circadian molecular clocks and cancer. Cancer Lett. 342:9–18. Knutsson A, Alfredsson L, Karlsson B, et al. (2013). Breast cancer among shift workers: Results of the WOLF longitudinal cohort study. Scand J Work Environ Health. 39:170–7. Kropp S, Terboven T, Hedicke J, et al. (2007). Good agreement between physician and self-reported hormone therapy data in a case-control study. J Clin Epidemiol. 60:1280–7. Lie JS, Kjuus H, Zienolddiny S, et al. (2011). Night work and breast cancer risk among Norwegian nurses: Assessment by different exposure metrics. Am J Epidemiol. 173:1272–9. Lie JS, Kjuus H, Zienolddiny S, et al. (2013). Breast cancer among nurses: Is the intensity of night work related to hormone receptor status? Am J Epidemiol. 178:110–17. Mavaddat N, Antoniou AC, Easton DF, Garcia-Closas M. (2010). Genetic susceptibility to breast cancer. Mol Oncol. 4:174–91. Maxwell KN, Nathanson KL. (2013). Common breast cancer risk variants in the post-COGS era: A comprehensive review. Breast Cancer Res. 15:212. Mediavilla MD, Sanchez-Barcelo EJ, Tan DX, et al. (2010). Basic mechanisms involved in the anti-cancer effects of melatonin. Curr Med Chem. 17:4462–81. Menegaux F, Truong T, Anger A, et al. (2013). Night work and breast cancer: A population-based case-control study in France (the CECILE study). Int J Cancer. 132:924–31. Michailidou K, Hall P, Gonzalez-Neira A, et al. (2013). Large-scale genotyping identifies 41 new loci associated with breast cancer risk. Nat Genet. 45:353–61, 361e1–2. Monsees GM, Kraft P, Hankinson SE, et al. (2012). Circadian genes and breast cancer susceptibility in rotating shift workers. Int J Cancer. 131:2547–52. Nickels S, Truong T, Hein R, et al. (2013). Evidence of geneenvironment interactions between common breast cancer susceptibility loci and established environmental risk factors. PLoS Genet. 9:e1003284. Pesch B, Harth V, Rabstein S, et al. (2010). Night work and breast cancer – Results from the German GENICA study. Scand J Work Environ Health. 36:134–41. Pesch B, Ko Y, Brauch H, et al. (2005). Factors modifying the association between hormone-replacement therapy and breast cancer risk. Eur J Epidemiol. 20:699–711. Portaluppi F, Smolensky MH, Touitou Y. (2010). Ethics and methods for biological rhythm research on animals and human beings. Chronobiol Int. 27:1911–29. Pronk A, Ji B, Shu X, et al. (2010). Night-shift work and breast cancer risk in a cohort of Chinese women. Am J Epidemiol. 171: 953–9. Rabstein S, Harth V, Pesch B, et al. (2013). Night work and breast cancer estrogen receptor status – results from the German GENICA study. Scand J Work Environ Health. 39:448–55. Reppert SM, Godson C, Mahle CD, et al. (1995). Molecular characterization of a second melatonin receptor expressed in
human retina and brain: The Mel1b melatonin receptor. Proc Natl Acad Sci U.S.A. 92:8734–8. Schernhammer ES, Hankinson SE. (2009). Urinary melatonin levels and postmenopausal breast cancer risk in the Nurses’ Health Study cohort. Cancer Epidemiol Biomarkers Prev. 18:74–9. Schoeps A, Rudolph A, Seibold P, et al. (2014). Identification of new genetic susceptibility loci for breast cancer through consideration of gene-environment interactions. Genet Epidemiol. 38: 84–93. Simonneaux V, Ribelayga C. (2003). Generation of the melatonin endocrine message in mammals: A review of the complex regulation of melatonin synthesis by norepinephrine, peptides, and other pineal transmitters. Pharmacol Rev. 55:325–95. Soria V, Martı´nez-Amoro´s E, Escaramı´s G, et al. (2010). Resequencing and association analysis of arylalkylamine Nacetyltransferase (AANAT) gene and its contribution to major depression susceptibility. J Pineal Res. 49:35–44. Stevens RG. (2005). Circadian disruption and breast cancer: From melatonin to clock genes. Epidemiology. 16:254–8. Stevens RG. (2009). Light-at-night, circadian disruption and breast cancer: Assessment of existing evidence. Int J Epidemiol. 38: 963–70. Stevens RG, Hansen J, Costa G, et al. (2011). Considerations of circadian impact for defining ‘shift work’ in cancer studies: IARC Working Group Report. Occup Environ Med. 68:154–62. Stevens RG, Hansen J. (2013). Letter to the editor, re: Night-shift work and risk of breast cancer (Kamdar et al., doi:10.1007/ s10549-013-2433-1). Breast Cancer Res Treat. 139:291–4. Stevens RG, Hansen J, Schernhammer ES, Davis S. (2013). Response to Ijaz S, et al. ‘‘Night-shift work and breast cancer – A systematic review and meta-analysis’’. Scand J Work Environ Health. 39:631–2. Straif K, Baan R, Grosse Y, et al. (2007). Carcinogenicity of shiftwork, painting, and fire-fighting. Lancet Oncol. 8:1065–6. Takahashi JS, Hong H, Ko CH, McDearmon EL. (2008). The genetics of mammalian circadian order and disorder: Implications for physiology and disease. Nat Rev Genet. 9: 764–75. Truong T, Liquet B, Menegaux F, et al. (2014). Breast cancer risk, nightwork, and circadian clock gene polymorphisms. Endocr Relat Cancer. 21:629–38. Wakefield J. (2007). A Bayesian measure of the probability of false discovery in genetic epidemiology studies. Am J Hum Genet. 81: 208–27. Wang F, Yeung KL, Chan WC, et al. (2013). A meta-analysis on dose-response relationship between night shift work and the risk of breast cancer. Ann Oncol. 24:2724–32. Winter SL, Bosnoyan-Collins L, Pinnaduwage D, Andrulis IL. (2007). Expression of the circadian clock genes Per1 and Per2 in sporadic and familial breast tumors. Neoplasia. 9: 797–800. Xiang S, Coffelt SB, Mao L, et al. (2008). Period-2: A tumor suppressor gene in breast cancer. J Circadian Rhythms. 6:4. Zhang C, Bao W, Rong Y, et al. (2013). Genetic variants and the risk of gestational diabetes mellitus: A systematic review. Hum Reprod Update. 19:376–90. Zhu Y, Stevens RG, Hoffman AE, Tjonneland A, et al. (2011). Epigenetic impact of long-term shiftwork: Pilot evidence from circadian genes and whole-genome methylation analysis. Chronobiol Int. 28:852–61. Zienolddiny S, Haugen A, Lie JS, et al. (2013). Analysis of polymorphisms in the circadian-related genes and breast cancer risk in Norwegian nurses working night shifts. Breast Cancer Res. 15:R53.
Chronobiology International