Journal of Cancer Epidemiology and Prevention Open Access

Keywords

Breast cancer; Epidemiology; Fat/omega-3; Omega-6/fish oil; Single nucleotide polymorphisms

Introduction

Experimental evidence suggests that higher ω-3 relative to ω-6 could reduce mammary carcinogenesis through mechanisms related to inflammation, oxidative stress, and estrogen metabolism [1]. A recently published meta-analysis of populationbased prospective cohort studies reported an approximately null association for dietary intake of polyunsaturated fatty acids (PUFAs) among studies that were conducted using populations from the United States (U.S.) [2]. This is inconsistent with findings from studies conducted in Asia [3] that show risk reductions for ω-3 intake. PUFAs may affect carcinogenesis via multiple biologic pathways [1]. Thus, we hypothesized that interactions with biologically relevant genetic polymorphisms with breast cancer may help to clarify the biologically plausible association with PUFAs. We further hypothesized that consideration of the ratio of ω-3 to ω-6 intake (ω-3:ω-6) may enhance our examination of interactions with breast cancer, given the two PUFA subgroups compete for the same metabolic enzymes [1]. Potential effect measure modification by genetic polymorphisms involved in PUFA metabolism has been previously considered by two studies [4,5] which examined only ω-3, but not intake relative to ω-6. Further, each study considered either glutathione S-transferase (GST) [4] or lipoxygenase enzymes [5], but not other genes involved in other biologically plausible pathways.

Materials and Methods

To address our hypothesis that the interaction between PUFA intake and breast cancer incidence would be evident when we considered the interaction between the ratio of ω-3:ω-6 intake and multiple genetic polymorphisms in the inflammation, oxidative stress, and estrogen metabolism pathways, we used case-control resources from the Long Island Breast Cancer Study Project (LIBCSP), a population-based study. Details of the parent study methods have been published [6].

Study population

Cases were women diagnosed with first primary in situ or invasive breast cancer between August 1, 1996 and July 31, 1997, who were residents of Nassau and Suffolk counties on Long Island, NY. Cases were identified by contacting hospital pathology departments daily or 2-3 times per week. Controls were identified from these same two counties using random digit dialing (women <65 years), and Health Care Finance Administration rosters (women ≥ 65 years). Controls were frequency-matched to the expected age distribution of the cases by 5-year age group. The LIBCSP includes 1508 case and 1556 control participants, ranging in age from 20 to 98 years. Approximately 67% were postmenopausal, and 93% self-reported as white, which is consistent with the underlying distribution of Nassau and Suffolk counties at the time of data collection.

Dietary and covariate assessment

LIBCSP participants completed an interviewer-administered risk factor questionnaire within approximately three months of diagnosis (cases), and within six months of identification (controls). Nearly all cases (98%) and controls (98%) completed a self-administered 101-item modified Block food frequency questionnaire (FFQ) to assess dietary intake in the year prior to the interview. PUFA intake was estimated by linking FFQ responses with nutrient values available from the U.S. Department of Agriculture databases for total ω-3 and total ω-6 PUFAs [7]. Alphalinolenic acid, docosapentaenoic acid, eicosapentaenoic acid, and docosahexanoic acid were included in the estimate for total ω-3 intake; whereas, linoleic acid and arachidonic acid subtypes comprised the estimate for total ω-6 intake. Excluding individuals with extreme total energy intake (>3 standard deviations from the mean) yielded 1463 cases and 1500 controls. Given ω-3 and ω-6 compete for the same enzymes [1], we primarily considered the ratio of ω-3:ω-6 intake [which ranged from 0.10-0.17 (25-75^th percentiles, respectively) among controls [3]. However, because this ratio may not accurately represent absolute intakes for each PUFA individually, we also considered separately total ω-3 and ω-6 (with corresponding ranges from 0.49-1.30 and 3.68-10.10 grams/day, respectively [3]).

Genotyping

Blood samples were collected from cases (74%) and controls (74%) at the time of the case-control interview and were used as the DNA source for genotyping [8-14]. Briefly, DNA was isolated from mononuclear cells in whole blood, which was separated by Ficoll (Sigma Chemical Co., St. Louis, MO) using standard phenol and chloroform-isoamyl alcohol extraction and RNase treatment [10]. Genotyping for inflammation genes [PTGS-2 (rs201417, rs5275), FAS (rs2234767, rs1800682), FASL rs763110, PPAR-α rs1800206, TNF-α rs1800629] used the following assays: Taqman 5’-Nuclease Assay (Applied Biosystems, Foster City, CA) and AcycloPrime^™-FP SNP Detection Kit obtained from Perkin Elmer Life Sciences (Boston, MA) [9,11,12]. The same assay was used for aromatase gene [CYP17 rs743572 with a 10 μM probe [8,14]]. For oxidative stress genes (CAT rs1001179, MPO rs2333227, MnSOD rs4880, GPX1 rs1050450, GSTA1 rs3957356, GSTP1 rs1695, COMT (rs4680, rs737865)], BioServe Biotechnologies (Laurel, MD) performed the genotyping using high-throughput, matrix assisted, laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry of Sequenom, Inc. (San Diego, CA). Gene deletions for GSTM1 and GSTT1 were determined by a multiplex polymerase chain reaction method, with the constitutively present gene β-globulin as an internal positive control [13].

Eighteen polymorphisms in 15 genes were selected spanning three biologically plausible pathways for PUFA metabolism, including inflammation, oxidative stress, and estrogen metabolism pathways. Variants affecting polyphen prediction (GPX1), transcription factor binding prediction (PTGS-2 rs20417, FAS, FASL, TNF-α, MPO, CAT, GSTA1, COMT rs737865, CYP17), miRNA binding (PTGS-2 rs5275, GPX1), 3D conformation (PPAR-α, COMT rs4680), or splicing regulation (PPAR-α, FAS rs2234767, GPX1, GSTP1, COMT rs4680) were considered as putatively functional variants as defined in the NIEHS SNPInfo WebServer [15].

Data were missing for some genetic polymorphisms, primarily due to laboratory failures. The final analytic sample size included a maximum of 1035 cases and 1075 controls, but varied by genetic variant (Supplemental Table S1).

Tests for Hardy-Weinberg equilibrium (HWE) were conducted among controls. Only PTGS-2 rs20417 and MPO deviated significantly from HWE (p=0.003 and 0.04, respectively). However, the observer agreement for PTGS-2 and MPO in 8% of a randomly selected sample was high (κ=0.99 and 0.91, respectively), and the failure rate was low for both polymorphisms (<1%). Also, the allele frequencies for both polymorphisms were similar to other studies [16].

Statistical analysis

Unconditional logistic regression was used to estimate interactions on the additive (using single-referent) and multiplicative (using a multiplicative interaction term) scales. To maximize cell sample size, PUFA intake measures were dichotomized at the median value (ratio of ω-3:ω-6=0.14, ω-3=0.83, and ω-6=6.31), and genotypes were dichotomized according to a dominant model and categorized into “high” and “low” risk groups based upon the putative function of the variant allele, which was determined using the existing literature (Supplemental Table S1). Additive interactions were evaluated using the Relative Excess Risk Due to Interaction (RERI) and the corresponding 95% CI [17]. Multiplicative interactions were evaluated using the likelihood ratio test (LRT, α=0.05). From among known risk factors for breast cancer, we identified no potential confounders, using a changein- estimate criterion of 10%. Thus, all interaction models were adjusted for the frequency matching factor five-year age group and total energy intake (kcal/day). We considered multiple comparisons using the false discovery rate (FDR, q=0.05) [18]. Analyses were conducted using SAS, version 9.2 (Cary, NC).

Results

When we explored interactions between the ratio of ω-3:ω-6 intake and genetic polymorphisms in the inflammatory, oxidative stress, and estrogen metabolism pathways, we found little evidence to support an additive (Table 1) or multiplicative (Supplemental Table S2) interaction for the majority of polymorphisms considered. However, an interaction was observed on the additive (RERI=-0.47, 95% CI=-0.92, -0.02) and multiplicative scales (LRT χ²=4.63, p value=0.03) for FASL rs763110. Specifically, we observed reduced OR for breast cancer among those with a low ratio of ω-3:ω-6 intake with the high risk FASL genotype on the multiplicative scale (OR=0.79, 95% CI=0.61, 1.02). The estimate was null on the additive scale (OR=1.06, 95% CI=0.81, 1.38), which was nearly 50% less than what would have been observed under an additive model [(expected ORs=1.52 (additive), and 1.58 (multiplicative)] as indicated by the statistically significant RERI. For the majority of other interactions examined, breast cancer risk remained high for low ratio of ω-3:ω-6 intake regardless of genotype, with the exception of the FASL rs763110 and COMT rs4680. When we considered PUFA types separately, no additive interaction was observed between total ω-3 intake and FASL (RERI=0.02; 95% CI=-0.33, 0.68); but for total ω-6 intake the interaction (RERI=-0.17, 95% CI=-0.59, 0.24) was in a similar direction to what was observed for the ω-3:ω-6 ratio. However, after considering multiple comparisons, we could not rule out chance as an explanation for our findings.

Table 1: Multivariable^a-adjusted odds ratios (ORs) and 95% confidence intervals (95% CIs) for the risk of breast cancer for the interaction between ω-3/ω-6 ratio and putatively functional genetic polymorphisms^b evaluated on the additive scale in the LIBCSP, 1996-1997.

Discussion

In our study we examined the interaction between PUFA intake and genetic polymorphisms from three biologically plausible pathways with breast cancer, and observed multiplicative and additive interaction between FASL rs763110 and the ratio of ω-3:ω-6 intake among a population-based sample of women from Long Island, NY. However, we could not rule out chance after adjusting for multiple comparisons. No other notable interactions were observed for polymorphisms from the inflammatory, oxidative stress, and estrogen metabolism pathways with breast cancer incidence.

For the interaction observed with FASL rs763110 high-risk genotype, we found a reduced OR for those women with a low ratio of ω-3:ω-6 PUFAs, which was contrary to our a priori biologic hypothesis. Amplification of FASL has been reported among breast cancer cases [19], and the arachidonic acidderived eicosanoid, prostaglandin E2 (PGE2), has been reported to increase FASL production [20]. Thus, a priori we hypothesized that the allele suggested to increase basal expression of FASL (i.e., C allele) along with low ratio of ω-3:ω-6 intake would increase breast cancer risk. A meta-analysis examining FASL rs763110 also reported increased risk for breast cancer for the same high risk allele [21] (although the authors did not consider interactions with PUFA intake). It is possible that the effect of FASL expression may be more important when considered as part of the FAS-FASL ligand-receptor system (part of the tumor necrosis factor superfamily) which can lead to apoptosis [22]. Thus, apoptosis resulting from FAS-FASL ligand-receptor system could potentially explain the reduced risk observed for the hypothesized highest risk subgroup in our study. Previously reported main effects for FASL and breast cancer incidence were nearly null in the LIBCSP [9]. Given these LIBCSP findings, the interaction between PUFAs and FASL requires further investigation.

Our study expands upon the limited findings from two previous breast cancer studies [4,5], which considered only ω-3 interactions with GSTs or lipoxygenases. Our approach has several advantages, including consideration of interactions between the ω-3:ω-6 ratio and genes involved in multiple biologically plausible pathways related to PUFA metabolism, which we evaluated on additive and multiplicative scales. We also utilized a population for whom dietary intake of total ω-3 PUFAs is likely to be higher compared to the other regions in the U.S.; women in New York City have been reported to consume more fish, a major dietary source of ω-3 PUFAs, compared to estimates from the National Health and Nutrition Examination Survey (NHANES) [23]. We did not observe an association between ω-3:ω-6 ratio and breast cancer incidence in a previous report [3]. However, it is still statistically and biologically possible that examining PUFA-gene interactions in multiple relevant biologic pathways, among a population with relatively high fish intake (a major ω-3 source) [3], could have revealed noteworthy effect modification.

There are several limitations to our study. First, our results are generalizable to only European-American women, for whom the incidence of breast cancer compared to other racial/ethnic groups remains high [24]. Importantly, it is possible that the frequency of the at-risk allele for relevant PUFA metabolism genes differ by racial subgroup [25], and thus it may be important to identify whether the PUFA-gene interactions and breast cancer risk vary among different populations. Furthermore, although we examined interactions with multiple polymorphisms spanning several biologic pathways, our selected genes are not exhaustive. For studies with larger sample sizes, it may be beneficial to examine additional genetic polymorphisms, such as genes involved in the in vivo metabolism of ω-3 and ω-6. Finally, after adjusting for multiple comparisons, we could not rule out chance for the interaction with FASL rs763110. However, we selected variants based on their putative function, thus, correcting for multiple comparisons may be overly conservative.

In this first study to consider the interaction between ω-3:ω-6 ratio and genetic polymorphisms from three biologically plausible pathways with breast cancer incidence, we noted an interaction on additive and multiplicative scales for FASL rs763110, but not for the other 17 polymorphisms considered. Additional research is needed to help clarify our findings, which were not consistent with our a priori hypothesis nor could we rule out chance once we adjusted for multiple comparisons.

References

1. Chenais B, Blanckaert V (2012) The janus face of lipids in human breast cancer: how polyunsaturated Fatty acids affect tumor cell hallmarks. Int J Breast Can 2012: 712536.
2. Cao Y, Hou L, Wang W (2016) Dietary total fat and fatty acids intake, serum fatty acids and risk of breast cancer: A meta-analysis of prospective cohort studies: Fat, fatty acids, and breast cancer. Int J Cancer 138:1894-1904.
3. Khankari NK, Bradshaw PT, Steck SE, He K, Olshan AF, et al. (2015) Polyunsaturated fatty acid interactions and breast cancer incidence: a population-based case-control study on Long Island, New York. Ann Epidemiol 25: 929-935.
4. Gago-Dominguez M, Castelao JE, Sun CL, Berg DVD, Koh WP, et al. (2004) Marine n-3 fatty acid intake, glutathione S-transferase polymorphisms and breast cancer risk in post-menopausal Chinese women in Singapore. Carcinogenesis 25: 2143-2147.
5. Wang J, John EM, Ingles SA (2008) 5-Lipoxygenase and 5-Lipoxygenase-Activating Protein Gene Polymorphisms, Dietary Linoleic Acid, and Risk for Breast Cancer. Cancer Epidemiol Biomark Prev Publ Am Assoc Cancer Res Cosponsored Am Soc Prev Oncol 17: 2748-2754.
6. Gammon MD, Neugut AI, Santella RM, Teitelbaum SL, Britton JA, et al. (2002) The Long Island Breast Cancer Study Project: description of a multi-institutional collaboration to identify environmental risk factors for breast cancer. Breast Cancer Res Treat 74: 235-254.
7. US Department of Agriculture (2012) Agricultural Research Service, USDA National Nutrient Database for Standard Reference, Release 25. Nutrient Data Laboratory Home Page.
8. Chen Y, Gammon MD, Teitelbaum SL, Britton JA, Terry MB, et al. (2008) Estrogen-biosynthesis gene CYP17 and its interactions with reproductive, hormonal and lifestyle factors in breast cancer risk: results from the Long Island Breast Cancer Study Project. Carcinogenesis29: 766-771.
9. Crew KD, Gammon MD, Terry MB, Zhang FF, Agrawal M, et al. (2007) Genetic polymorphisms in the apoptosis-associated genes FAS and FASL and breast cancer risk. Carcinogenesis 28:2548-2551.
10. Gammon MD, Santella RM, Neugut AI, Eng SM, Teitelbaum SL, et al. (2002) Environmental toxins, and breast cancer on Long Island. I. Polycyclic aromatic hydrocarbon DNA adducts. Cancer Epidemiol Biomark Prev Publ Am Assoc Cancer Res Cosponsored Am Soc Prev Oncol11:677-685.
11. Golembesky AK, Gammon MD, North KE, Bensen JT, Schroeder JC, et al. (2008) Peroxisome proliferator-activated receptor-alpha (PPARA) genetic polymorphisms and breast cancer risk: A Long Island ancillary study. Carcinogenesis 29:1944-1949.
12. Shen J, Gammon MD, Terry MB, Teitelbaum SL, Neugut AI, et al. (2006) Genetic polymorphisms in the cyclooxygenase-2 gene, use of nonsteroidal anti-inflammatory drugs, and breast cancer risk. Breast Cancer Res 8:R71.
13. Steck SE, Gaudet MM, Britton JA, Teitelbaum SL, Terry MB, et al. (2007) Interactions among GSTM1, GSTT1 and GSTP1 polymorphisms, cruciferous vegetable intake and breast cancer risk. Carcinogenesis 28:1954-1959.
14. Talbott KE, Gammon MD, Kibriya MG, Chen Y, Teitelbaum SL, et al. (2008) A CYP19 (aromatase) polymorphism is associated with increased premenopausal breast cancer risk. Breast Cancer Res Treat111:481-487.
15. Xu Z, Taylor JA (2009) SNPinfo: integrating GWAS and candidate gene information into functional SNP selection for genetic association studies. Nucleic Acids Res37:W600-W605.
16. Lee YS, Kim H, Wu TX, Wang XM, Dionne RA (2006) Genetically mediated interindividual variation in analgesic responses to cyclooxygenase inhibitory drugs. Clin Pharmacol Ther79:407-418.
17. Hosmer DW, Lemeshow S (1992) Confidence interval estimation of interaction. Epidemiol Camb Mass 3:452-456.
18. Benjamini Y, Hochberg Y (1995) Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. R Stat Soc 57:289.
19. Mullauer L, Mosberger I, Grusch M, Rudas M, Chott A (2000) Fas ligand is expressed in normal breast epithelial cells and is frequently up-regulated in breast cancer. J Pathol190:20-30.
20. O’Callaghan G, Kelly J, Shanahan F, Houston A (2008) Prostaglandin E2 stimulates Fas ligand expression via the EP1 receptor in colon cancer cells. Br J Cancer 99:502-512.
21. Xu L, Zhou X, Jiang F, Qiu MT, Zhang Z, et al. (2013) FASL rs763110 Polymorphism Contributes to Cancer Risk: An Updated Meta-Analysis Involving 43,295 Subjects. PLoS ONE 8:e74543.
22. Nagata S, Golstein P (1995) The Fas death factor. Science267:1449-1456.
23. Mahaffey KR, Clickner RP, Jeffries RA (2009) Adult women’s blood mercury concentrations vary regionally in the United States: association with patterns of fish consumption (NHANES 1999-2004). Environ Health Perspect 117:47-53.
24. DeSantis C, Ma J, Bryan L, Jemal A (2014) Breast cancer statistics, 2013: Breast Cancer Statistics, 2013. CA Cancer J Clin 64:52-62.
25. Mathias RA, Sergeant S, Ruczinski I, Torgerson DG, Hugenschmidt CE, et al. (2011) The impact of FADS genetic variants on ω6 polyunsaturated fatty acid metabolism in African Americans. BMC Genet 12:50.