Abstract
Background Normal cell BRCA1 epimutations have been associated with increased risk of triple-negative breast cancer (TNBC). However, the fraction of TNBCs that may have BRCA1 epimutations as their underlying cause is unknown.
Methods To address this question, we analyzed BRCA1 methylation status in breast cancer tissue and matched white blood cells (WBC) from 411 patients with primary breast cancer, including 66 TNBCs, applying a highly sensitive sequencing assay, allowing allele-resolved methylation assessment. Further, to assess the time of origin and the characteristics of normal cell BRCA1 methylation, we analyzed umbilical cord blood of 1260 newborn girls.
Results We found concordant tumor and mosaic WBC BRCA1 epimutations in 10 out of 66 patients with TNBC and in four out of six patients with estrogen receptor (ER)-low expression (<10%) tumors (combined: 14 out of 72; 19.4%; 95% CI 11.1–30.5). In contrast, we found concordance in only three out of 221 patients with ER≥10% tumors and zero out of 116 patients with HER2-positive tumors. Intraindividually, BRCA1 epimutations affected the same allele in normal and tumor cells. Assessing BRCA1 methylation in umbilical WBCs from girls, we found mosaic, predominantly monoallelic BRCA1 epimutations, with qualitative features similar to those in adults, in 113/1260 (9.0%) of individuals.
Conclusions Our findings reveal prenatal BRCA1 epimutations to be the underlying cause of around 20% of TNBC and low-ER expression breast cancers.
Background
Aberrant gene promoter methylation, or epimutations, are observed in many cancer types. While such epimutations may be passenger events of limited biological importance, it is well established that promoter methylation of tumor suppressor genes (TSGs) may contribute as driving forces in tumor progression [1, 2].
BRCA1 germline mutations are the most frequent cause of hereditary breast and ovarian cancers [3]. Most cancers arising in BRCA1 mutation carriers belong to the triple-negative subclass of breast cancers (TNBC) and the high-grade serous subclass of ovarian cancers (HGSOC). Contrasting BRCA2 [4], BRCA1 is frequently methylated in sporadic TNBC and HGSOC tumors [5–7], and it is well established that such promoter methylation is associated with repressed BRCA1 transcription [8, 9]. TNBCs with BRCA1 methylation have a gene expression profile closely resembling TNBCs arising in BRCA1 mutation carriers [5]. While BRCA1 promoter methylation and BRCA1 mutations seem to a large extent to be mutually exclusive in both TNBCs and HGSOCs [5, 10], conflicting evidence indicates similarities and differences between BRCA1 promoter methylation and mutations regarding therapy sensitivity in breast cancer [6, 11, 12].
Constitutional epimutations are defined as aberrant normal tissue methylation occurring in early life, generally affecting all three germ layers [13]. There are two types; secondary epimutations, caused by specific genetic aberrations, and primary epimutations, for which no underlying genetic factor is found [13]. Contrasting secondary epimutations, primary epimutations often present in a low-level, mosaic pattern, affecting only a small fraction of cells [10]. While secondary constitutional methylation of BRCA1 has been observed in a few families with an elevated risk of breast and ovarian cancer [14–16], the question of primary constitutional methylation as a cancer risk factor has remained controversial [4, 10, 17-23]. However, in a recent study we found white blood cell (WBC) BRCA1 promoter methylation to predict an elevated risk of incident TNBC as well as HGSOC >5 years after blood sampling in healthy women [24]. While these findings indicate that BRCA1 methylation may arise in normal cells subsequently developing into cancer precursors, several key questions need to be addressed. First, we do not know whether WBC BRCA1 mosaic methylation arise prenatally (constitutional) or may be acquired postnatally (somatic normal tissue methylation). In case of constitutional methylation, we need to address whether such methylation may be fully developed across the promoter prenatally or exist as a precursor for subsequent development at a later stage. Second, in case BRCA1 methylated cells arising by a prenatal clonal expansion as constitutional methylation, one would expect a qualitatively similar, allele-specific WBC methylation in newborns as that recorded in adults. Third, if BRCA1-methylated WBCs represent constitutional methylation, and methylated WBCs and breast cancer precursor cells share a common embryonic clonal origin, one would expect a similar allele-specific BRCA1 methylation profile [24] in WBCs and matched BRCA1-methylated tumors from the same individual. Fourth, we need to assess the fraction of TNBCs arising from constitutionally BRCA1-methylated cells, i.e. the fraction of TNBCs, previously considered as “sporadic”, that could be explained by underlying BRCA1 methylation. While we recently reported the hazard ratio for incident TNBC with respect to WBC BRCA1 methylation to be 2.35 [24], such a hazard ratio provides an indirect estimate for the fraction of tumors actually derived by this mechanism [25]. Moreover, the fact that the median age of women enrolled in our previous study was 62 years, indicates that a substantial fraction of TNBCs that may have been overlooked due to diagnosis at a younger age.
To address these questions, we evaluated the incidence, magnitude, intraindividual tissue concordance and allele specificity of BRCA1 methylation in tumor and matched WBC from 411 patients with primary breast cancer, including 66 TNBCs. In addition, we analyzed WBC BRCA1 methylation in 1260 umbilical cord blood samples from newborn girls.
METHODS
Patients and tissue sampling
In the present study, we included all patients enrolled in three neoadjuvant breast cancer studies (EPITAX, DDP and PETREMAC) [6, 26, 27] from which pretreatment tumor tissue and WBC DNA samples were available for analysis (Fig. 1). The studies were approved by the Regional Ethics Committee (273/96-82.96, 06/3077 and 2015/1493), and all patients provided written informed consent at enrolment. The DDP and PETREMAC trials were registered under ClinicalTrials.gov (NCT00496795 and NCT02624973) while the EPITAX was conducted prior to ClinicalTrials implementation.
All patients underwent an incisional or Tru-Cut tumor biopsy prior to treatment. Tumor biopsies were snap-frozen in liquid nitrogen at removal and stored in liquid nitrogen, while WBC specimens were stored at –80°C after centrifugation of EDTA whole blood and plasma removal.
The Norwegian Mother, Father and Child Cohort study (MoBa) is an ongoing cohort study enrolling more than 110 000 newborns and their parents [28]. For the present study, we randomly selected 1260 newborn girls, including 420 born prematurely (before 36 weeks of gestation) and 840 girls born at normal term (39–41 weeks of gestation). Procedures for sample collection, DNA extraction and storage have been described previously [29]. No difference in BRCA1 methylation was observed between premature and normal term newborns, and the samples were therefore treated as a unified cohort in the present study.
Sample preparation
Procedures for DNA and RNA extraction from tumor and WBC samples are outlined in Additional file 1. In brief, genomic DNA for methylation analyses was extracted from tumor and WBC samples using QIAamp DNA Mini kit (Qiagen, Valencia, CA) and total RNA for gene expression analysis was extracted from tumor tissue using the RNeasy Mini kit (Qiagen, Valencia, CA).
Methylation sequencing
For BRCA1 methylation analysis, DNA bisulfite conversion, amplification and sequencing was performed as described previously [24] Briefly, 500 ng of genomic DNA were bisulfite converted and subjected to BRCA1 gene promoter fragment amplification using four pairs of primers, that do not overlap with any of the CpG dinucleotides (GRCh38 genomic coordinates: CpG00–13 chr17:43125624–43126026, CpG14–31 chr17:43125270–43125640, CpG17–34 chr17:43125171–43125550, CpG33–49 chr17:43124861–43125249; Supplementary Fig. S1). All four amplicons were combined, indexed, and sequenced by 2×226 bp reads using Illumina MiSeq System (Illumina, San Diego, CA), resulting in an ultradeep coverage of about 30000x for each amplicon. As previously described for case-control analyses [24], the region covering CpGs 17–34 is considered biologically relevant and was used as the main measure for methylation calling. The overlapping amplicon covering CpGs 33–49, also covered SNP rs799905, and was used for allele-specific methylation assessment.
Molecular subtyping
Estrogen and progesterone receptor (ER, PgR) as well as HER2 status was determined upon inclusion in each of the clinical trials. For the present analyses, we defined the cutoff for ER positivity as 1%. Since the EPITAX trial, conducted in 1997–2003, used a cutoff of 10%, all cases where ER status was recorded as <10% were re-examined according to standard criteria and classified as either <1% or 1–9%.
All tumors were assigned to intrinsic subtypes based on mRNA expression profiling according to the classification by Perou et al. [30] using either RNA sequencing (DDP and PETREMAC trials) or mRNA microarrays (EPITAX trial) (for details, see Additional file 1). While the percentage of ER-positive tumors is lower and the percentage of HER2+ and TNBC/ER-low tumors is higher in the oldest (EPITAX trial) subset, as compared to the DDP and PETREMAC trials, the same difference is present in the full study population, [6, 26, 27] reflecting a higher enrolment of patients with HER2+/TNBC when the EPITAX trial was actively recruiting in the 1990s.
BRCA1 mutation status assessment
Data on BRCA1 mutation status for patients were collected from our previous genetic analyses [6, 27]. For cases lacking previous genetic data, we performed targeted sequencing of a cancer gene panel, as previously described, [31] and drew BRCA1 mutation status from the generated data. For consistency, all detected variants were re-audited for pathogenicity according to the ClinVar database [32].
Data analyses
For BRCA1 methylation analysis, sequencing reads were mapped/aligned to the GRCh38 reference genome using the Illumina DRAGEN Bio-IT Platform (v3.6.3). Cytosine methylation and its allele specificity was evaluated using the epialleleR R package (v1.3.5) [33]. A single quantitative metric of methylation (hypermethylated variant epiallele frequency, VEF) was obtained by averaging frequencies of hypermethylated epialleles for two amplicons covering CpGs 14–31 and 17–34 as previously reported [24]. The cutoffs for methylation positivity were determined computationally, following the same predefined approach as previously reported [24] (for details, see Additional file 1). The cutoffs were defined as 6.96×10−4 for BRCA1 methylation in WBC and 4.71×10−2 for tumors. The differences between the cutoffs in WBC and tumors also reflect a biological rationale: WBC BRCA1 methylation is expected to present a low-level, mosaic pattern [10, 34], while tumors arising from BRCA1-methylated cells are expected to harbor a large fraction of methylated cells due to clonal expansion. Notably, for sensitivity analyses all WBC data from newborns in the present study were also re-assessed applying the cutoff defined by WBC-analyses of cancer patients and vice versa. These sensitivity analyses had no impact on the biological conclusion (see Additional file 1 for details).
Statistical analysis
Correlation in methylation status between tumor and corresponding WBC or normal breast tissue samples were evaluated using Fisher’s exact test. Methylation frequencies were presented with confidence intervals. R software environment for statistical computing (v4.1.2) was used for all statistical analyses.
RESULTS
Study population characteristics
To assess BRCA1 methylation in breast cancer patients, we included all patients enrolled in three neoadjuvant breast cancer studies (the EPITAX, DDP and PETREMAC trials [6, 26, 27] from whom pretreatment tumor tissue and matched WBC DNA samples were available. Among a total of 600 patients screened, 411 had both blood and tumor pretreatment samples successfully analyzed and were included in the final results. The reasons for exclusion are presented in Fig. 1 and details of the patients included are given in Table 1.
To assess allele-specific mosaic BRCA1 methylation in newborns, 1260 umbilical cord blood samples were drawn from the Norwegian Mother, Father and Child Cohort study (MoBa) [35], as listed in Table 2.
Concordant WBC and tumor BRCA1 methylation
Based on the previously detected association between constitutional BRCA1 methylation and risk of TNBC, we analyzed concordance of BRCA1 methylation in tumors and matched WBC samples to assess the fraction of TNBCs potentially caused by underlying constitutional BRCA1 methylation. In total, 17 out of 66 (25.8%; 95% CI 15.8–38.0) patients with TNBC harbored tumor BRCA1 methylation (Table 3). Notably, among these 17 patients, 10 (58.9%; CI 32.9–81.6%) also carried WBC BRCA1 methylation (WBC and tumor tissue methylation concordance: P < 0.001; Fig. 2). Thus, 15.2% (95% CI 7.5–26.1%) of all TNBCs revealed concordant tumor and WBC BRCA1 methylation.
Regarding tumors with ER expression within 1–9%, genomic profiling has revealed these tumors to mirror gene expression profiles recorded in TNBC [36]. In this subgroup, four out of six patients harbored BRCA1 methylation in their tumor tissue, all revealing concordant WBC methylation (Fig. 2). Grouping the TNBC and ER-low (1–9%) tumors together, concordant tumor and WBC BRCA1 methylation was observed in 14 out of 72 patients (19.1%; 95% CI 11.1–30.5%). Notably, these 14 constituted the majority of the 21 patients with BRCA1 methylated TNBC or ER low (1–9%) tumors (66.7%; 95% CI 43.0–85.4%).
For HER2-negative tumors expressing ER≥10%, six out of 221 revealed BRCA1 tumor methylation with only three of these patients revealing concordant WBC BRCA1 methylation. Further, the lowest methylation frequency was observed among HER2-positive tumors (independent of ER expression). Here, one out of 116 tumors revealed BRCA1 methylation in the tumor tissue, and this patient was negative for WBC BRCA1 methylation (Fig. 2).
As for patients harboring BRCA1 methylation in both tumor and WBC, BRCA1 methylation levels in the tumors were 36–103 fold higher than in blood, consistent with clonal expansion of cells with methylated BRCA1 alleles (Fig. 3A). Intratumoral levels of BRCA1 methylation were similar for TNBCs (7.6–78.8%), ER-low (10.7–36.4%), and the remaining non-TNBC methylated tumors (13.8–82.4%).
Taken together, these findings indicate that around 20% of all TNBC/ER-low breast cancers and around 65% of all BRCA1-methylated tumors occur in individuals with underlying constitutional BRCA1 methylation.
Allele-specific concordance of BRCA1 methylation in WBC and tumor tissue
While concordant BRCA1 methylation in tumor tissue and matched blood samples may suggest a common clonal origin, we sought to provide further evidence for this hypothesis by assessing the allele specificity of the BRCA1 methylation in tumors and blood.
Allele-specific methylation may be detected in cases heterozygous for the SNP rs799905, since this polymorphism is located in the area that is covered by the BRCA1 methylation assay (see Methods and Additional file 1; Fig. S1). Among 17 patients carrying concordant BRCA1 methylation in WBC and tumor tissue, SNP rs799905 genotype information was lacking and/or could not be linked to methylation in two patients. For the 15 informative individuals, seven were homozygous for the reference allele, two were homozygous for the alternative allele, while six were heterozygous. The allelic distribution of BRCA1 methylation for WBCs and tumor samples among these six informative heterozygous cases is depicted in Fig. 3B (green dots). BRCA1 methylation was enriched on the same allele in the tumor tissue and WBC in five of these individuals, indicating a shared clonal origin of the methylated normal and tumor cells. The sixth patient revealed comparable levels of BRCA1 methylation of both rs799905 alleles in blood (with a slight preference for methylation on the alternative allele) while the tumor carried methylation of the reference allele. Most likely, this patient harbored two independent subclones of BRCA1-methylated normal cells, with one clone giving rise to the tumor cells.
As the tumor samples were not subject to microdissection, they contain a number of different types of benign cells including normal breast epithelium, fibroblasts, circulating WBCs and macrophages [37]. Among individuals harboring WBC but not tumor BRCA1 methylation (n = 27), 17 revealed small traces of BRCA1-methylated cells in the tumor biopsies, below the defined threshold for classification of tumors as methylation-positive but above the methylation threshold applied to WBC samples. This is consistent with low-level mosaic BRCA1 methylation in normal breast cells and/or other normal cells present in the tumor biopsies. Among these 17 patients, four were heterozygous for rs799905 and thus informative for allele-specific methylation status. These four all revealed the low-level BRCA1 methylation in their tumor biopsies to share the same magnitude and allele specificity as the methylation in the matched WBCs (Fig. 3B, red dots).
In addition, one patient with a BRCA1 methylated TNBC and WBC BRCA1 methylation just below the formal cutoff for positivity could be assessed for allelic methylation concordance. This patient also revealed a similar allele specific BRCA1 methylation in tumor and WBC samples (Fig. 3B, blue dot).
Taken together, these findings reveal an allelic concordance between BRCA1 methylation in WBC and matched cancer or benign tissue in the breast cancer samples, indicating that the methylated tumors have arisen from methylated normal cells in the affected mosaic individuals.
Intrinsic breast cancer subtypes and BRCA1 methylation
The distribution of BRCA1 methylation was determined among the intrinsic subtypes of breast cancers, based on their mRNA signatures [30]. The subtype distribution of BRCA1-methylated tumors did not differ between TNBCs and non-TNBCs, neither was there a difference between those harboring concordant WBC BRCA1 methylation and those that did not (Fig. 2A). Regarding TNBCs and ER<10% tumors with concordant tumor and WBC BRCA1 methylation, 11 out of 14 were basal-like, two were normal-like, while one tumor expressed a HER2-enriched profile, despite absence of HER2 gene amplification or positive protein staining. Interestingly, among the three ER≥10% tumors revealing concordant tumor and WBC BRCA1 methylation, two revealed a basal-like profile, while the remaining one was classified as luminal A.
BRCA1 mutations and methylation
Among the 411 tumors analyzed, nine harbored BRCA1 pathogenic variants (four somatic and five germline; Fig. 3C; Additional file 1; Table S1). None of the patients with germline BRCA1 pathogenic variants revealed either WBC or tumor BRCA1 methylation. While one patient with a somatic BRCA1 mutation harbored tumor tissue BRCA1 methylation in concert, no BRCA1 methylation was detected in this patient’s WBCs, indicating the tumor BRCA1 methylation, similar to the mutation, to be a somatic event.
Frequency and allele specificity of BRCA1 methylation in WBC from newborns
To assess the hypothesis of early origin (Fig. 4A) and potential dynamics of BRCA1 epimutations and their allele specificity, we analyzed umbilical cord blood samples from 1260 newborn girls. We found 113 out of 1260 newborns to be BRCA1 methylation-positive (9.0%; 95% CI 7.5–10.7%), with average beta value distribution (Fig. 4B) as well as hypermethylated variant epiallele frequencies (VEF) in a very similar range as seen in adult breast cancer patients (7.0×10-4 to 5.9×10- 2 and 7.2×10-4 to 7.6×10-2, respectively; p>0.1; Fig. 4D). Also, similar to the intramolecular methylation pattern seen in adults, there was a relatively sharp distinction between non-methylated and hypermethylated epialleles, with the majority of methylation-positive epialleles being close to fully methylated (i.e. methylated on all CpGs), and very few epialleles having intermediate methylation levels (i.e. methylated at 20-80% of epialleles; Fig. 4C). No particular single CpG or stretch of consecutive CpGs in the methylated region stood out as systematically more or less methylated than others.
Among the 113 newborns revealing WBC BRCA1 methylation, sufficient amounts of DNA for SNP rs799905 assessment were available from 89. Among these, we found 40/89 (44.9%) of the girls to be heterozygous for SNP rs799905, while 12/89 (13.5%) and 37/89 (41.6%) were homozygous for the alternative and reference alleles, respectively. This distribution is similar to that recently established in adult women in the US Women’s Health Initiative study [24]. Intraindividually, among the 40 girls heterozygous for rs799905, methylation was located predominantly on one specific allele, similar to the pattern seen in adults (Fig. 4E). Further, across these heterozygous individuals, there was a similar overall distribution of methylated alleles between the SNP rs799905 alternative and reference allele (p > 0.10).
Taken together, our findings confirm BRCA1 epimutations in newborns to be predominantly monoallelic intraindividually but to occur independently of promoter haplotype across individuals, and with low VEF values, resembling the findings in adults.
DISCUSSION
Epigenetic regulation plays a key role in normal cell function during life and is influenced by genetic as well as environmental factors. Epimutations may occur either as somatic events during life or as so-called constitutional methylation, arising in utero, in which case it may affect normal tissues derived from all three germ layers [13]. While epimutations causing gene silencing of tumor suppressors like MGMT, MLH1 and BRCA1 are well known across many malignancies [6, 8, 38-41], in general such epimutations are considered somatic. However, the seminal discovery by Hitchins and colleagues of inherited MLH1 methylation in a family prone to colon cancer [42] sparked interest in constitutional methylation of TSG promoters as a potential underlying cause of cancer [13, 43].
While WBC BRCA1 methylation has been associated with an elevated hazard ratio of triple-negative breast cancer, to this end the quantitative contribution of normal tissue BRCA1 methylation to TNBC and, potentially, non-TNBC, has remained unknown due to lack of studies evaluating concordant BRCA1 methylation in tumor tissue and matched WBC. Analyzing BRCA1 methylation in matched blood and tumor tissue of patients with both TNBC and non-TNBC, we found a strong correlation between tumor tissue and WBC BRCA1 methylation in TNBC and tumors revealing a low ER expression (1–9%). In this group, 29.2% of tumors were BRCA1-methylated, and 19.4% harbored concordant tumor and WBC methylation. While no previous data exist regarding the incidence of concordant BRCA1 methylation in tumor and WBCs, our findings for tumor methylation in total (somatic plus constitutional origin) aligns with the finding by Glodzik et al [5] of 30% of TNBCs carrying tissue BRCA1 promoter methylation. Importantly, if a percentage as high as 20% of TNBCs/ER-low breast cancers having concordant WBC methylation is reproduced across other cohorts, this would mean that a substantially larger fraction of TNBCs may be caused by constitutional methylation than by pathogenic germline variants.
In contrast, among breast cancers with HER2 overexpression or ER levels of ≥10%, constitutional BRCA1 tumor methylation was a rare event. However, interestingly, two out of three tumors with constitutional methylation revealed a basal-like gene expression signature.
Recently, we reported WBC BRCA1 methylation to be predominantly monoallelic, enriched on the same allele across the vast majority of normal blood cells in affected adult individuals [24]. Here, we significantly extended this observation by showing concordant allele-specific methylation in normal and malignant cells in our cancer patients. This is consistent with a common clonal origin of methylated normal and cancer cells, which supports the hypothesis that these BRCA1-methylated tumors have arisen from methylated normal cells.
In addition, in four patients informative for methylation allele specificity in WBC, we found the tumor samples to reveal low-level BRCA1 methylation, likely reflecting a fraction of benign cells in the biopsy. Here, we found similar allelic concordance for BRCA1 methylation in WBC and the presumed normal breast tissue. While such low-level methylation in theory could reflect small subclones of somatically methylated tumor cells, the chance of allelic concordance across four paired datasets is <7%. In contrast, the finding of similar allele-specific methylation in WBC and normal breast tissue is what one would expect in cases of constitutional (prenatal) methylation. While the term “non-malignant cells” in a breast cancer biopsy covers different cell types [37] of ectodermal as well as mesenchymal origin, constitutional methylation generally affects cells derived from all three embryonic germ layers [13].
While our patient data are consistent with constitutional BRCA1 methylation, a pivotal question is whether normal tissue BRCA1 methylation develops as a complete promoter methylation (effectively repressing BRCA1 expression) in utero or arise as a partially functional epimutation, subsequently developing into complete promoter methylation in normal cells postnatally. While we and others previously detected WBC BRCA1 methylation among newborn girls [10, 44], these studies assessed methylation status by conventional analyses (qualitative methylation-specific PCR) preventing detailed assessment of allele specificity and quantitative characteristics of methylation. Notably, there is evidence showing that gene promoter methylation may develop in a step-wise manner [45], thus, methylation previously recorded by MSP in newborns could present a pre-methylation step or a qualitatively completely different methylation process from the one detected in normal cells of adults. In the present study, we found BRCA1 methylation in newborn girls to qualitatively and quantitatively mirror the one seen in adult cancer patients and previously recorded in healthy adults [24], indicating that the methylation observed in newborns and in adults is the same molecular feature.
Taken together, these findings, in concert with our findings of similar allelic methylation status in WBC and tumor tissue in adults, are consistent with a common clonal origin of all BRCA1-methylated cells within each patient, indicating that methylation may have arisen as a single-cell event during early embryonic development with subsequent clonal expansion across all germ layers.
Analyzing WBCs collected from patients diagnosed with their breast cancer, it is important to exclude the possibility that WBC methylation is due to contamination from the tumor, either as circulating tumor DNA or circulating tumor cells. Yet, tumor contamination is unlikely to be the cause of WBC BRCA1 methylation for several reasons. First; considering the number of circulating tumor cells, even among patients with a substantial cancer burden, such cells account for less than one in a million cells [46]. Second, as for circulating tumor DNA [47, 48], the plasma volume currently required for detection of tumor-derived genomic aberrations in blood samples is far above any possible plasma remnants in our WBC assay. Finally, in our recent study [24] we confirmed WBC BRCA1 methylation in healthy women to predict subsequent incident TNBC as well as HGSOC >5 years after sampling, providing proof for normal cell BRCA1 methylation to be a precursor for TNBC and HGSOC.
Cancer patients may have a different WBC subfraction composition as compared to healthy individuals. Thus, a potential uncertainty in the present study relates to cancer-related changes in the WBC subfraction composition. While global methylation patterns vary between leukocyte subfractions [49, 50], examining BRCA1 methylation status across previously reported datasets from adults [51], newborns and corresponding 5-year old children [29, 52] we detected no difference in BRCA1 methylation status between the different WBC subfractions [10]. Thus, the observed differences in BRCA1 methylation may not be a consequence of differences in WBC subfraction composition.
Taken together, we consider these findings to validate and justify the use of WBC BRCA1 methylation as a marker of constitutional methylation in most individuals, including patients diagnosed with their primary breast cancer.
In our recent case-control WHI study [24], WBC BRCA1 methylation was associated with an increased risk of incident TNBC (hazard ratio, HR 2.5) and HGSOC (HR 1.8). Notably, in that study the median age at inclusion was 62 years, yet, TNBCs are known in general to be detected at an earlier age compared to other breast cancer subtypes [5]. Regarding HGSOC, the risk estimates in the WHI study is lower than that previously observed (HR 2.2–2.9) in a hospital-based cohort study in Norwegian women in which methylation was assessed by MSP [10]. Thus, the possibility exists that the lifetime risk for TNBC is higher than what we recorded in the WHI study. While the number of cases in the present study is limited, our finding that constitutional BRCA1 methylation may account for 19.4% of all triple-negative and ER-low breast cancers is high given the observed BRCA1 methylation frequency in the population (5.6% among non-cancer females in the US [24] and 9.0% among healthy newborns in the current study). Since constitutional BRCA1 methylation affects a small fraction of normal cells in the individual, one may assume that the background incidence of BRCA1 unmethylated breast cancers (including TNBC) is similar among carriers and non-carriers of BRCA1 constitutional methylation. Based on this, our present data indicates that constitutional BRCA1 methylation could be associated with a hazard ratio perhaps as high as 4–5 for TNBC/ER-low BC development [25].
In summary, we find constitutional BRCA1 methylation, as defined by WBC methylation, to be linked to TNBC/ER-low BC. As for patients diagnosed with TNBC or ER-low tumors, our findings indicate that BRCA1 promoter methylation should be explored as a potential risk factor for subsequent cancer development. Moreover, comparing cancers carrying the methylation on a constitutional versus somatic background should be performed to fully elucidate potential pathogenic consequences.The presence of BRCA1 methylation on the same BRCA1 allele in WBC and breast cancer DNA in the same patients adds strong support to the hypothesis that BRCA1-methylated tumors may arise from constitutionally BRCA1-methylated normal cells, likely initiated as an early, prenatal event [13]. Thus, our findings conceptually differ from normal tissue global methylation signatures designed for early cancer detection [53–55]. Our findings urge for further research exploring potential causes of BRCA1 methylation, as well as studies exploring potential constitutional methylation of other tumor suppressor genes, with respect to cancer risk. To this end, cancers have generally been classified in two main groups; those arising on a background of pathogenic germline variants and those regarded as spontaneous tumors, with a grey zone of low-risk variants and genes, in-between. Our findings point toward the dawn of a new era, suggesting a substantial number of cancers may develop from early epimutated cell clones.
Supplementary Information
The online version contains supplementary material available at …………..
Additional file 1. Contains Supplementary Methods, Supplementary Fig.s S1–S6 and Supplementary Table S1. Fig. S1: Genomic structure of the BRCA1 promoter area, positions of CpGs, single-nucleotide variations, and amplified regions Fig. S2: Distributions of VEF values (CpG14–34 average methylation metric), corresponding density functions and cutoff bounds. Fig. S3: Scatter plots and density histograms showing the relation of VEF and beta values for blood and tumor samples. Fig. S4: Scatter plots and density histograms showing the relation of VEF and beta values for newborn blood samples. Fig. S5: PCR bias assessed for VEF and methylation beta values. Fig. S6. Intrinsic breast cancer subtypes based on gene expression analysis of tumors. Table S1. BRCA1 methylation status in TNBC cases with pathogenic (germline or somatic) BRCA1 mutations
Data Availability
Haukeland University Hospital and the University of Bergen support the dissemination of research data that has been generated, and increased cooperation between investigators. Trial data is collected, stored, and disseminated according to institutional guidelines and in accordance with national laws and regulations to ensure the quality, integrity, and use of clinical data. Study protocol, including plan for statistical analyses, is available online via previous publications. Signed informed consent forms are stored at each participating hospital and are available for monitoring by regulatory authorities. After publication and upon formal request, raw data, including de-identified individual participant data and a data dictionary defining each field in the data set, will be shared according to institutional procedures. Requests are via a standard pro forma describing the nature of the proposed research and extent of data requirements. Data recipients are required to enter a formal data sharing agreement which describes the conditions for release and requirements for data transfer, storage, archiving, publication, and intellectual property. Requests are reviewed by the EPITAX, DDP and PETREMAC study teams in terms of scientific merit and ethical considerations, including patient consent. Data sharing is permitted if proposed projects have a sound scientific or patient benefit rationale, as agreed by the study team and with approval from the trials' co-investigators as required. Samples from the MoBa study were analyzed blinded to the identity of the participants. Since BRCA1 WBC methylation has been found associated with an elevated risk of HGSOC and TNBC, under Norwegian law, this means that WBC BRCA1 methylation status may be considered predictive testing. Thus, our Regional Ethics Committee approved this part of the study provided no identification key or link to clinical information was stored. Thus, all molecular data from newborns in the present study were handled anonymously after analysis. Raw molecular data are available from the authors upon reasonable individual request, but any link to parameters in the MoBa databank is not possible.
Authors contribution
PEL and SK conceived the study. ON and SK developed the laboratory methods, and ON was in primary charge of the statistical analysis and performed all the epigenome analysis. PEL (EPITAX, DDP and PETREMAC) and HPE (DDP and PETREMAC) were in charge of the clinical studies. BG, SL, ESB, HE; JG and SG provided clinical material for the analysis. EAMJ was in charge of the immunochemistry, while SY, LM, BL and RL performed the epigenetic and subtype analysis. PEL, ON and SK drafted the manuscript, while all authors participated in editing of the final version.
Funding
This work was supported by unrestricted grants from The K.G. Jebsen Foundation, Helse Vest, The Norwegian Research Council and The Norwegian Cancer Society. The funders had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. All authors had access to the data and vouch for its accuracy and completeness. All authors were involved in the decision to submit the manuscript for publication.
Data availability and materials
Haukeland University Hospital and the University of Bergen support the dissemination of research data that has been generated, and increased cooperation between investigators. Trial data is collected, stored, and disseminated according to institutional guidelines and in accordance with national laws and regulations to ensure the quality, integrity, and use of clinical data. Study protocol, including plan for statistical analyses, is available online via previous publications. Signed informed consent forms are stored at each participating hospital and are available for monitoring by regulatory authorities. After publication and upon formal request, raw data, including de-identified individual participant data and a data dictionary defining each field in the data set, will be shared according to institutional procedures. Requests are via a standard pro forma describing the nature of the proposed research and extent of data requirements. Data recipients are required to enter a formal data sharing agreement which describes the conditions for release and requirements for data transfer, storage, archiving, publication, and intellectual property. Requests are reviewed by the EPITAX, DDP and PETREMAC study teams in terms of scientific merit and ethical considerations, including patient consent. Data sharing is permitted if proposed projects have a sound scientific or patient benefit rationale, as agreed by the study team and with approval from the trials’ co-investigators as required.
Samples from the MoBa study were analyzed blinded to the identity of the participants. Since BRCA1 WBC methylation has been found associated with an elevated risk of HGSOC and TNBC [24], under Norwegian law, this means that WBC BRCA1 methylation status may be considered predictive testing. Thus, our Regional Ethics Committee approved this part of the study provided no identification key or link to clinical information was stored. Thus, all molecular data from newborns in the present study were handled anonymously after analysis. Raw molecular data are available from the authors upon reasonable individual request, but any link to parameters in the MoBa databank is not possible.
Declarations
Ethics approval and consent to participate
All participants in the clinical studies provided written informed consent, and the current study is covered by each consent. All studies were approved by the Regional Ethics Committee (reference numbers: 273/96-82.96, 06/3077 and 2015/1493).
Competing interests
Research Funding (to Institution): Astellas Oncology (BG), AstraZeneca (HPE, BG, SK, PEL, ESB), Celgene (BG), Novartis (HPE, PEL), Pfizer (HPE, SK, BG, PEL).
Honoraria: Amgen (HPE), AstraZeneca (HPE, BG, TA, EAMJ, SK, PEL), Abbvie (PEL), Bristol-Myers-Squibb (HPE, JG), Dagens Medisin (HPE, PEL), Eli Lilly (JG), HAI Interaktiv AS (HPE), MSD (JG), Novartis (HPE, JG, SK), Pfizer (HPE, EAJ, SK), Pierre Fabre (HPE, JG, SK, PEL), Roche (HPE, BG, TA, PEL).
Consulting or Advisory Role: Aptitude Health (HPE), Astellas Oncology (BG), AstraZeneca (ESB, JG, PEL), Daiichi Sankyo (ESB, HPE), Eli Lilly (ESB, HPE, JG), Gilead (ESB), Laboratorios Farmaceuticos Rovi (PEL), MSD (ESB, HPE, JG), Novartis (ESB, HPE, JG), Pfizer (HPE), Pierre Fabre (HPE), Roche (HPE, BG)
Expert Testimony: Pfizer (HPE).
Travel, Accomodations, Expenses: AstraZeneca (HPE), Pierre Fabre (HPE, BG, PEL), Roche (BG).
Speakers’ Bureau: Akademikonferens (PEL), Aptitude Health (PEL), AstraZeneca (JG), Bristol-Myers-Squibb (JG), MSD (JG), Novartis (JG), Pfizer (JG), Pierre Fabre (JG).
Patents, Royalties, Other Intellectual Property: Patent EP2389450 A1 (SK), Patent WO 2012/ 010661 (SK), Cytovation (PEL).
All remaining authors have declared no conflicts of interest.
Author details
1K.G. Jebsen Center for Genome-Directed Cancer Therapy, Department of Clinical Science, University of Bergen; Bergen, Norway.
2Department of Oncology, Haukeland University Hospital; Bergen, Norway.
3Department of Hematology and Oncology, Stavanger University Hospital; Stavanger, Norway.
4Cancer Clinic, St. Olavs Hospital, Trondheim University Hospital; Trondheim, Norway.
5Department of Clinical and Molecular Medicine, Norwegian University of Science and Technology; Trondheim, Norway.
6Department of Oncology, University Hospital of North Norway; Tromsø, Norway.
7Department of Surgery, Haugesund Hospital; Haugesund, Norway.
8Department of Oncology, Akershus University Hospital; Lørenskog, Norway.
9Institute of Clinical Medicine, University of Oslo, Norway.
10Department of Pathology, Stavanger University Hospital; Stavanger, Norway.
11Department of Chemistry, Bioscience and Environmental Engineering; Stavanger University, Stavanger, Norway.
Acknowledgements
We gratefully acknowledge the technical assistance of N.K. Duong, C. Eriksen, E. De Faveri, and G. Iversen, University of Bergen and Haukeland University Hospital.