Differences in DNA methylation by extent of breast cancer family history in unaffected women

Abstract

Breast cancer clusters within families but genetic factors identified to date explain only a portion of this clustering. Lower global DNA methylation in white blood cells (WBC) has been associated with increased breast cancer risk. We examined whether WBC DNA methylation varies by extent of breast cancer family history in unaffected women from high-risk breast cancer families. We evaluated DNA methylation levels in LINE-1, Alu and Sat2 in 333 cancer-free female family members of the New York site of the Breast Cancer Family Registry, the minority of which were known BRCA1 or BRCA2 mutation carriers. We used generalized estimated equation models to test for differences in DNA methylation levels by extent of their breast cancer family history after adjusting for age. All unaffected women had at least one sister affected with breast cancer. LINE-1 and Sat2 DNA methylation levels were lower in individuals with 3 or more (3+) first-degree relatives with breast cancer relative to women with only one first-degree relative. For LINE-1, Alu, and Sat2, having 3+ affected first-degree relatives was associated with a decrease of 23.4% (95%CI = −46.8%, 0.1%), 17.9% (95%CI = −39.5%, 3.7%) and 11.4% (95% CI = −20.3%, −2.5%), respectively, relative to individuals with only one affected first-degree relative, but the results were only statistically significant for Sat2. Individuals having an affected mother had 17.9% lower LINE-1 DNA methylation levels (95% CI = −28.8%, −7.1%) when compared with those not having an affected mother. No associations were observed for Alu or Sat2 by maternal breast cancer status. If replicated, these results indicate that lower global WBC DNA methylation levels in families with extensive cancer histories may be one explanation for the clustering of cancers in these families. Family clustering of disease may reflect epigenetic as well as genetic and shared environmental factors.

Keywords: :

• breast cancer
• family history
• LINE-1
• Sat2
• DNA methylation
• white blood cells

Introduction

Approximately 20–30% of breast cancer is attributed to familial aggregation.¹ Family history of breast cancer is a well-established risk factor; with individuals with at least one affected first degree relative such as a daughter, a sister, or mother facing a increased risk of at least 80%, and individuals with three or more affected relatives facing an increase in risk closer to 4-fold compared to women without a family history of breast cancer.² This increase in disease risk associated with family history has been shown on both an absolute and a relative scale, being higher when relatives are diagnosed earlier .³

Genetic factors such as high-penetrance BRCA1 and BRCA2 variants have been identified and are well characterized; however, the majority of women in families with breast cancer do not have mutations in these genes.⁴ The analysis of genome wide association studies have led to the identification of new genetic susceptibility loci, which might help explain sporadic as well as familial breast cancer in families without BRCA1/2 mutations; however, these findings reveal a large amount of family clustering that remains unexplained.⁵ More recent interest has focused on additional mechanisms such as epigenetic changes to help explain the occurrence of breast cancer within families. DNA methylation is one type of epigenetic change that has been associated with cancer, and recent studies have shown that global changes in DNA methylation in peripheral tissue such as blood are correlated with increased risk of different cancers, including breast.^6-10 Our group also found that global DNA methylation levels, measured by the methyl acceptance assay, and levels in the repeat Sat2 were lower in women with breast cancer when compared with their unaffected sisters.^7,8 Therefore, it has been hypothesized that lower global DNA methylation in peripheral tissue may be an indicator of an individual’s susceptibility to breast cancer.

A previously published study investigating changes in levels of DNA methylation in a 16-y period in 21 families with at least 2 generations, found that the changes in global DNA methylation levels cluster in members of the same family, indicating the possibility of a genetic contribution to the maintenance of these levels in the cell.¹¹ In addition, we have previously showed that different global DNA methylation markers are correlated between sisters discordant for breast cancer suggesting that these markers might represent shared genetic and/or environmental determinants.^7,8 In a study investigating colorectal cancer risk, DNA methylation levels of LINE-1 in tumor DNA were associated with a family history of colorectal cancer.¹² We have also reported differences in global DNA methylation biomarkers in blood of girls with a family history of breast cancer when compared with those without.¹³ These results raise the question of whether global DNA methylation might represent, at the molecular level, a marker of inherited breast cancer susceptibility. To expand on these findings, we conducted an analysis in the New York site of the Breast Cancer Family Registry (BCFR) of women unaffected with cancer but who had extensive pedigree data available to examine whether in healthy women DNA methylation levels measured in white blood cells (WBC) were different by extent of family history.

Results

In Table 1 we report overall demographic and genetic characteristics for the study subjects. Fifty six percent of the unaffected women were non-Hispanic white, while 44% were Hispanic or of other ethnicity. A small fraction of women were carriers of BRCA1 (3.9%) or BRCA2 (2.4%) mutations. Of all the participants, 64% had 1 first-degree relative affected by breast cancer, 33% had 2 relatives affected, and 3% had 3 or more first degree relatives affected.

Table 1. Demographic and genetic characteristics of the study participants, unaffected women of the New York site of the BCFR with at least one sister with breast cancer

	N	Mean (SD) or %
Age
Continuous	333	47.3 (11.1)
Range		18–84
Race
Non-Hispanic White	187	56.2
Hispanic/ Other	146	43.8
BRCA1 Mutation
Positive	13	3.9
Negative	320	96.1
BRCA2 Mutation
Positive	8	2.4
Negative	325	97.6
Number of breast cancer cases in 1st and 2nd degree relatives
1	157	47.1
2	107	32.1
3+	69	20.7
Number of breast cancer cases in 1st degree relatives
1	213	64.0
2	109	32.7
3+	11	3.3
Number of breast cancer cases in siblings
1	293	88.0
2, 3	40	12.0
Breast cancer in mothers
Yes	76	22.8
No	257	77.2

We investigated the difference in DNA methylation of LINE-1, Alu and Sat2 by number of family members with breast cancer relative to those with only one first-degree and second-degree or first-degree relative affected. In Table 2 we show the percent of DNA methylation of the study participants by extent of breast cancer family history. Table 3 has the age-adjusted generalized estimating equation (GEE) estimates and confidence intervals for the overall number of affected first-degree relatives. For LINE-1, Alu, and Sat2, having 3+ affected first-degree relatives was associated with a decrease of 23.4% (95%CI = −46.8%, 0.1%), 17.9% (95%CI = −39.5%, 3.7%) and 11.4% (95% CI = −20.3%, −2.5%), respectively, relative to individuals with only 1 affected first-degree relative, but the results were only statistically significant for Sat2. We considered separately the effect of having an affected mother and the number of sisters affected on the levels of these DNA methylation biomarkers. Unaffected women who had an affected mother had a statistically significant decrease of 17.9% in LINE-1 DNA methylation (95% CI = −28.8%, −7.1%), while there were no overall differences for the other studied markers (Alu and Sat2 levels were 3.2% (95% CI = −8.5%, 14.8%) and -0.4% (95% CI = −5.6%, 4.8%) between women with and without an affected mother). There were no differences in DNA methylation levels between women with 2 or more affected sisters relative to 1 affected sibling: for LINE-1 (3.4% 95% CI = −12.8%, 19.6%), Alu (−2.0% 95% CI = −16.7%, 12.6%) and Sat2 (−2.1% 95%CI = −8.6%, 4.4%) (Table 3).

Table 2. DNA methylation levels of the repetitive elements LINE-1, Alu, and Sat2 for unaffected women of the New York site of the BCFR by the number of relatives with a history of breast cancer.

	# Affected relatives	Repetitive element DNA methylation (%)
		LINE-1		Alu		Sat2
		N	Mean (SD)	N	Mean (SD)	N	Mean (SD)
First and Second Degree Relatives
	1	156	115.9 (59.9)	157	100.3 (60.5)	156	45.9 (39.2)
	2	107	97.4 (51.2)	107	92.1 (43.7)	107	40.9 (25.1)
	3+	69	108.6 (66.9)	69	105.4 (39.0)	69	42.1 (27.9)
P value			0.05		0.22		0.45
First Degree Relatives
	1	212	111.9 (55.3)	213	98.3 (54.7)	212	43.8 (34.9)
	2	109	104.0 (67.1)	109	101.0 (45.4)	109	44.2 (29.9)
	3+	11	84.9 (43.9)	11	84.6 (49.2)	11	30.6 (19.6)
P value			0.22		0.59		0.42
Siblings
	1	292	107.0 (52.3)	293	98.3 (49.5)	292	43.2 (31.6)
	2, 3	40	118.8 (96.2)	40	101.9 (65.9)	40	45.5 (41.7)
P value			0.45		0.74		0.74
Mothers
	No	256	113.4 (63.0)	257	99.6 (56.3)	256	44.1 (35.7)
	Yes	76	91.7 (40.3)	76	95.8 (31.1)	76	41.4 (21.2)
P value			0.0005		0.44		0.41

Table 3. Changes in WBC DNA methylation levels in LINE-1, Alu, and Sat2 with overall first-degree relative, maternal and sibling breast cancer history for unaffected women of the New York site of the BCFR. Estimates and 95% confidence intervals from age-adjusted GEE models

	# Affected relatives	Repetitive element DNA methylation (%)
		LINE-1		Alu		Sat2
		β	95% CI	β	95% CI	β	95% CI
First Degree Relatives
	1	Reference		Reference		Reference
	2	−9.9	−20.6, 0.8	5.9	−4.7, 16.6	0.4	−4.3, 5.2
	3+	−23.4	−46.8, 0.1	−17.9	−39.5, 3.7	−11.4	−20.3, −2.5
Siblings
	1	Reference		Reference		Reference
	2, 3	3.4	−12.8, 19.6	−2.0	−16.7, 12.6	−2.1	−8.6, 4.4
Mother
	No	Reference		Reference		Reference
	Yes	−17.9	−28.8, −7.1	3.2	−8.5, 14.8	−0.4	−5.6, 4.8

We also performed additional analyses exploring the effect of maternal breast cancer family history on DNA methylation levels of LINE-1 when stratified by race/ethnicity (Table 4). White non-Hispanic women with an affected mother had a decrease in methylation of 12.4% (95% CI = −23.3%, −1.5%) for LINE-1 DNA. Hispanic women with a mother affected by breast cancer had a decrease in LINE-1 DNA methylation levels of 18.8% (95% CI = −42.7%, 5.1%) (Table 4). We did not observe differences by race/ethnicity on the other markers studied (data not shown).

Table 4. Changes in WBC DNA methylation levels of LINE-1, Alu, and Sat2 by maternal breast cancer history stratified by race/ethnicity in unaffected women of the New York site of the BCFR. Estimates and 95% confidence intervals from age-adjusted GEE models

Race/Ethnicity	Maternal breast cancer history	Repetitive element DNA methylation (%)
		LINE-1		Alu		Sat2
		β	95% CI	β	95% CI	β	95% CI
White	No	Reference		Reference		Reference
	Yes	−12.4	−23.3, −1.5	0.7	−9.9, 11.4	−0.1	−5.1, 4.9
Hispanic/Other	No	Reference		Reference		Reference
	Yes	−18.8	−42.7, 5.1	5.9	−19.8, 31.7	2.4	−10.0, 14.7

Discussion

We found that DNA methylation at the LINE-1 repetitive element was lower for individuals with more extensive family histories particularly if these individuals had a mother affected by the disease in addition to a sister. We also found that levels of Alu and Sat2 were lower in those study participants with an extensive family history in first-degree relatives, but the differences were statistically significant only for Sat2. We only examined DNA methylation levels in unaffected women so their WBC levels were not affected by cancer.

The LINE-1 repeat is the most common retrotransposable element in the genome.¹⁴ Human genomes have approximately half a million copies of this repeat, a small fraction of which are able to undergo active retrotransposition and of which a small portion are commonly analyzed for DNA methylation levels.¹⁵ One of the mechanisms used by the cell to keep these retrotransposons from being transcribed is hypermethylation. Conversely, lower levels of DNA methylation at these sequences can lead to their activation and initiate disease processes such as carcinogenesis.¹⁶ We have previously reported that girls with a family history had lower DNA methylation at LINE-1 and Alu repeats.¹³ In addition, a recent study found that LINE-1 levels in tumor DNA were associated with a family history of colorectal cancer, suggesting this DNA methylation mark might be important in understanding epigenetic inherited cancer risk. Results of the current study are in agreement with these findings and suggest that lower methylation of repeated sequences might be an indicator of extensive breast cancer family history.

Limited published evidence suggests that DNA methylation levels in the LINE-1 repeat might be inherited. A study found levels of DNA methylation at LINE-1 were correlated between parents and their children and that a gender specific pattern of inheritance might exist.¹⁷ Similarly, a separate study found levels of LINE-1 and Alu DNA methylation were correlated in maternal and infant pair bloods.¹⁸ These results and our current findings suggest levels of DNA methylation in the LINE-1 repetitive element might be inherited and might provide information on disease susceptibility.

While associations between breast cancer risk in women and global hypomethylation measured by other markers have been reported, little evidence to date suggests an association between LINE-1 and breast cancer risk.^6-8,19 However, lower levels of LINE-1 have been associated with increased risk of gastric, head and neck, bladder, liver, and testicular cancer.^17,20-23 Increased breast cancer risk has been found to be related to lower DNA methylation of the repetitive element Sat2.⁸ In this study, we found a strong indication that LINE-1 DNA methylation might be associated with a breast cancer family history. In addition we observed a decrease in Sat2 DNA methylation in individuals with extensive breast cancer family history in first-degree relatives. We should highlight that in our analysis there are only 11 participants with three or more breast cancer affected first degree relatives. While this is a limitation of the study, we found that the inverse association (e.g., more relatives, lower levels of DNA methylation markers) was consistent for 3 of the markers studied, albeit it was only statistically significant for Sat2. We also observed differences by whether both the mother and sister were affected relative to just having a sister affected, further suggesting that the relation between DNA methylation and family history of breast cancer warrants replication and evaluation in larger studies. Our study also did not take into account family size, a fact that could influence the lack of differences by number of relatives observed in some of the associations.

Previously published studies have identified race and ethnicity as demographic variables related to differential levels of DNA methylation of repetitive sequences.^22,24,25 Similarly, it has been shown that family history might differ by race and/or ethnicity, although effect modification by race and/or ethnicity has not been fully consistent.^26-29 Some of these studies suggest non-Hispanic White women with a breast cancer family history are at higher risk of disease when diagnosed before the age of 45,^26,27 while others seem to find no differences by race/ethnicity,²⁸ or higher risk for Hispanic women with a positive family history.²⁹ In spite of the fact that our results when stratifying by ethnicity supported that an inverse association between LINE-1 DNA methylation and maternal breast cancer was only statistically significant in non-Hispanic White women, Hispanic women also had decreases in LINE-1 methylation, although not statistically significant, if their mother and sister were both affected relative to Hispanic women with just an affected sister. We had a limited sample size, which might have affected our ability to detect an association. In addition, Hispanic populations have a diverse genetic composition, and studies published to date have only investigated family history in Mexican-American women or women of Mexican descent. Future studies should investigate further the possible effect of race/ethnicity on the association between WBC DNA methylation biomarkers and breast cancer family history.

It is important to consider that all individuals included in this analysis have a sister who has had breast cancer. Therefore, we could not assess the separate contribution of sibling family history on DNA methylation of repetitive elements. An analysis of extent of family history and DNA methylation levels has not been reported previously, however studies have looked at breast cancer risk and extent of family history. In the Nurses’ Health Study a history of breast cancer in the mother and in a female sibling conferred a similar amount of risk.³⁰ The same group also reported that the amount of disease risk resulting from having one affected sibling, did not change with increasing number of affected sisters.³ While we are not able to compare individuals without affected siblings, these data support our finding that our molecular marker might parallel disease susceptibility.

In this study, presence of disease in the mother predicted lower levels of LINE-1 DNA methylation but an increasing number of affected siblings did not, which might indicate a particular contribution of the mother’s underlying disease susceptibility to the levels of epigenetic markers on her children. These results help support the idea that epigenetic marks and family history both might represent shared environmental, behavioral and genetic factors which affect disease risk. These results also suggest epigenetic biomarkers measured in WBC, such as DNA methylation of the repetitive element LINE-1, may differ by the extent of breast cancer family history and that global levels of DNA methylation might be a molecular marker of inherited breast cancer susceptibility.

Methods

Study population and assessment of family history

The New York site of the BCFR recruited high risk breast and/or ovarian cancer families from clinical and community settings within the metropolitan New York area (for details see refs. 7,8,31-34) In the present study, we included 333 unaffected women who had a sister who was affected with breast cancer participating in the New York BCFR for whom we had available DNA methylation data. In addition to biospecimens used for the methylation analyses, participants provided family pedigree data and individual epidemiological data. We obtained written informed consent from all research participants and the IRB at Columbia approved the study protocol

Global DNA methylation measures

DNA extraction, bisulfite modification and LINE-1, Alu and Sat2 methylation levels

DNA extraction from total WBC was performed by standard salting-out procedures. The DNA was quantified and aliquots of 500ng were bisulfite-treated using the EZ DNA methylation kit (Zymo Research, catalog number D5005) following the manufacturer’s protocol. We used sequences of probes and forward and reverse primers of LINE-1-M1, Alu-M2, and Sat2-M1, and methods as described in Weisenberger et al. to determine the levels of fully methylated LINE-1, Alu, and Sat2 repeats.³⁵ Conditions for real-time PCR followed previously reported protocols.⁸ Universal methylated DNA from ZYMO Research served as a fully methylated reference sample, and an Alu-based control reaction (Alu-C4)³⁵ was used to measure the levels of input DNA to normalize the signal for each methylation reaction. MethyLight data was expressed as percent of methylated reference (PMR) values. The number of repeats of the reference repetitive sequence (Alu-C4) vary among individuals, which results in some PMR values being above 100%. We calculated PMR using the equation below:

PMR = 100% * 2^−ΔΔCt

Where the ΔΔCt = [(target repeat in sample − Alu-C4 in sample) − (target repeat in 100% fully methylated reference sample − Alu-C4 in 100% fully methylated reference sample)].

Each MethyLight reaction was performed in duplicate wells and PMR values were calculated using the average Ct for both samples. The inter assay coefficients of variation range from 5.6% to 15.6% for the studied markers. We conducted all laboratory assays blinded to epidemiological data.

Statistical analysis

We used ANOVA to test for differences in each DNA methylation marker by family history of breast cancer. We compiled family history information on all female and male family members to categorize the extent of family history for each individual. We used age-adjusted generalized estimating equation (GEE) linear regression models to examine associations between each methylation marker and family history of first-degree relatives. We did not adjust for other covariates, as they were not associated with the family history (main exposure) and the DNA methylation marker. We log transformed PMR values representing DNA methylation percentages because these were not normally distributed before running the GEE linear regression models. No adjustment for multiple testing was considered necessary because the measures are highly correlated with each other. All analyses were performed with SAS software 9.3 (SAS Institute).

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We would like to gratefully acknowledge all of the families that participate in the New York site of the BCFR. We also would like to gratefully acknowledge Ann Johnston Cloud, MPH; Ashley Tsai, MPH, Julie Flom, MPH; Jennifer Ferris, MPH; and Irina Gurvich for their dedication to the BCFR and for their valuable input and discussion in the presentation of these results. This work was supported by an award from the Breast Cancer Research Foundation and NIH grants, R01CA159868, R01094069, and by the National Cancer Institute, National Institutes of Health under RFA-CA-06–503 and through cooperative agreements with members of the BCFR and Principal Investigators, including Columbia University (U01 CA69398). The content of this manuscript does not necessarily reflect the views or policies of the National Cancer Institute or any of the collaborating centers in the BCFR, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government or the BCFR.

10.4161/epi.26880