Comprehensive molecular comparison of BRCA1 hypermethylated and BRCA1 mutated triple negative breast cancers

Abstract

Homologous recombination deficiency (HRD) is a defining characteristic in BRCA-deficient breast tumors caused by genetic or epigenetic alterations in key pathway genes. We investigated the frequency of BRCA1 promoter hypermethylation in 237 triple-negative breast cancers (TNBCs) from a population-based study using reported whole genome and RNA sequencing data, complemented with analyses of genetic, epigenetic, transcriptomic and immune infiltration phenotypes. We demonstrate that BRCA1 promoter hypermethylation is twice as frequent as BRCA1 pathogenic variants in early-stage TNBC and that hypermethylated and mutated cases have similarly improved prognosis after adjuvant chemotherapy. BRCA1 hypermethylation confers an HRD, immune cell type, genome-wide DNA methylation, and transcriptional phenotype similar to TNBC tumors with BRCA1-inactivating variants, and it can be observed in matched peripheral blood of patients with tumor hypermethylation. Hypermethylation may be an early event in tumor development that progress along a common pathway with BRCA1-mutated disease, representing a promising DNA-based biomarker for early-stage TNBC.

Conflict of interest statement

J.H. has received speakers honoraria and travel support from Roche, advisory board fees from MSD, Novartis and Roche, and institutional research grants from Cepheid and Novartis. Anna Ehinger has received speakers honoraria from Novartis, Amgen, Roche, and advisory board fees from Roche. Ana Bosch has participated in advisory boards for Novartis and Pfizer, and has received travel support from Roche. The remaining authors declare no competing interests.

Figures

Fig. 1. Study scheme, performed analyses, and cohorts used.

Gray boxes indicate a cohort of samples.

Fig. 2. BRCA1 hypermethylation, gene expression, and HRD association.

a Hierarchical clustering (ward.D2 linkage, Euclidean distance) of DNA methylation data (beta-values shown as a heatmap) for 30 CpGs associated with the BRCA1 gene (transcription start site (TSS): −1500b to +500 bp) in Illumina MethylationEPIC data for 235 SCAN-B TNBCs, including 57 tumors classified as hypermethylated by pyrosequencing (black column sample annotation bars). Gray CpG annotation bars (rows) indicate a promoter associated CpGs according to Illumina EPIC annotations. bBRCA1 mRNA expression (FPKM) across the 237 SCAN-B cases stratified by gene abrogation status. p Values calculated using t-test. Top axis shows number of cases per group. c Left: proportions of rearrangement signature 3 (RS3) versus patient stratifications based on BRCA1/2 mutation status, BRCA1 methylation status, and BRCA1 LOH in the total SCAN-B cohort, excluding the small HRDetect-intermediate group. For non-BRCA1/2 cases, tumors are stratified by HRDetect-high or low classification. Right: proportion of deletions with microhomology across the same patient subgroups. Top axes show number of cases per group. All cases do not have assigned RS3 rearrangements. dBRCA1 CpG allele methylation versus estimated tumor % by the WGS specific Battenberg algorithm (https://github.com/cancerit/cgpBattenberg) for all 237 SCAN-B cases. Black dotted line corresponds to a linear regression fit for the 57 hypermethylated cases specifically. e Circos plot and depiction of mutational substitution (S3, S8, and S13) and rearrangement signatures (RS2 and RS3) as defined in ref. of PD35990a. This case harbors both a BRCA2 variant and BRCA1 hypermethylation but has a genetic phenotype of BRCA1-deficient cancer. Circos plot depicting from outermost rings heading inwards: karyotypic ideogram outermost. Base substitutions next, plotted as rainfall plots (log10 intermutation distance on radial axis, Ring with short green lines, insertions; ring with short red lines, deletions. Major copy number allele ring (green, gain), minor copy number allele ring (red, loss), Central lines represent rearrangements (green, tandem duplications; red, deletions; blue, inversions; gray, interchromosomal events). FPKM fragments per kilobase of transcript per Million mapped reads. All p values reported from statistical tests are two-sided. Source data are provided as a Source Data file.

Fig. 3. BRCA1 methylation in peripheral blood, gene expression subtypes, and prognosis after therapy.

a Distribution of age at diagnosis for 43 germline screened SCAN-B patients harboring germline BRCA1 loss of function variants (n = 9) or no germline variants (n = 34) further stratified by tumor BRCA1 hypermethylation status. bBRCA1 CpG allele methylation frequency in matched peripheral blood DNA from non-germline SCAN-B patients stratified by their tumor methylation status (methylated = 1) from panel (a). cBRCA1 CpG allele methylation frequency in peripheral blood DNA from a combined analysis of 105 SCAN-B cases analyzed using the same instrument settings and reagent lots, including 55 of 57 hypermethylated cases, and stratified by patient age and tumor methylation status. Hard brackets ([]) imply ≥ or ≤, respectively. d Age at diagnosis for 237 SCAN-B patients stratified by BRCA1 and BRCA2 status. The number of methylated cases is less than 57 as two cases have concurrent BRCA2 mutations (one germline, one somatic nonpathogenic). e Molecular subtype proportions in BRCA1 hypermethylated SCAN-B cases for PAM50, CIT, IntClust 10, and TNBCtype. CIT subtypes; mApo, molecular apocrine. IntClust 10 subtypes; cluster 10 corresponding to basal-like tumors by other subtyping schemes. TNBCtype subtypes; BL1, basal-like 1: BL 2, basal-like 2: IM, immunomodulatory: M, mesenchymal: MSL, mesenchymal stem-like: LAR, luminal androgen receptor: UNC, uncertain. f Univariate Cox regression using invasive disease-free survival (IDFS) as clinical endpoint for different variables in 149 SCAN-B patients eligible for outcome analysis after standard of care adjuvant chemotherapy. HR: hazard ratio. NHG: grade, G2 equals grade 2, G3 equals grade 3. N0: node-negative. A Zph p value < 0.05 corresponds to that the proportional hazard assumption is not fulfilled. An (n =) indication in the right axis indicates that not all 149 cases were used due to missing values. g Kaplan–Meier analysis using IDFS as clinical endpoint for SCAN-B patients eligible for outcome analysis after adjuvant chemotherapy. Top panel shows the 149 patients stratified by BRCA1 hypermethylation status alone, center panel shows stratification including also BRCA1-null cases, and bottom panel a comparison between only hypermethylated and BRCA1-null patients. All p values reported from statistical tests are two-sided. Source data are provided as a Source Data file.

Fig. 4. Genetic phenotypes of BRCA1 hypermethylated and BRCA1-null TNBC.

The 57 hypermethylated and 25 BRCA1-null SCAN-B cases were combined with 27 BRCA1-null cases from Nik-Zainal et al.. a Frequency of copy number alterations across the genome for BRCA1 hypermethylated and BRCA1-null cases. b Frequency of copy number amplification for driver genes defined in Nik-Zainal et al.. c Frequency of mutations (insertions, deletions, substitutions) for driver genes defined in Nik-Zainal et al.. Only genes with >1% alteration in the BRCA1-null cohort are shown. Displayed mutations in BRCA1 are somatic. d Total number of substitutions, indels, and rearrangements per sample for BRCA1 hypermethylated versus BRCA1-null groups. Only cases sequenced to at least 30-fold depth included. p Values calculated using Wilcoxon’s test. Top axes show number of cases per group. e Left panel shows distribution of mutational signature (defined in Nik-Zainal et al.) proportions per sample between hypermethylated versus BRCA1-null cases. Proportions are calculated as the number of substitutions for a signature divided by the total number of substitutions from all signatures. Right panel shows proportion of the APOBEC Substitution Signature 2 in non-basal-like tumors from Nik-Zainal et al.. Top axis indicates number of samples per group. All outliers are not shown due to y-axis scale. f Distribution of rearrangement signature proportions per sample between hypermethylated and BRCA1-null cases. g Hierarchical clustering of combined substitution and rearrangement signature proportions using Pearson correlation and Ward.D linkage in the 109 combined cases. h Principal component analysis of proportions of substitution and rearrangement signatures in the 109 combined cases, illustrated by the first two principal components representing most variation. i Principal component analysis of the proportions of the contributions of HRDetect components (as defined in ref. ) per sample (obtained from), illustrated by the first two principal components representing most variation. The analysis only included the 25 BRCA1-null and 57 hypermethylated SCAN-B cases. All p values reported from statistical tests are two-sided. Source data are provided as a Source Data file.

Fig. 5. DNA methylation and transcriptional characteristics in BRCA1 hypermethylated and BRCA1-null TNBC.

a Hierarchical clustering (ward.D2 linkage, Euclidean distance) of DNA methylation data (beta-values, heatmap) from 235 SCAN-B cases for 32 CpGs associated with the BRCA1 gene. The 32 CpGs were differentially methylated between 25 BRCA1-null and 57 BRCA1 hypermethylated SCAN-B cases through supervised analysis of Illumina EPIC data. Black sample annotations correspond to hypermethylated cases by pyrosequencing, gray to unmethylated. Gray CpG annotation bars (rows) indicate belonging to the canonical BRCA1 promoter (+500 to −1500 base pairs, chr17:43124984–43126983). BRCA1 mRNA expression (FPKM) is shown for each case as bar plot expression above the CpG heatmap. b Unsupervised hierarchical gene expression-based clustering of 52 BRCA1-null and 57 BRCA1 hypermethylated cases combined from the SCAN-B and Jönsson et al. cohorts using Pearson correlation as distance and Ward.D linkage based on 7224 genes with standard deviation >0.6 in mean-centered expression across all samples. Colored boxes indicate the seven top subclusters defined from the hierarchical tree. TNBCtype subtypes; BL1, basal-like 1: BL 2, basal-like 2: IM, immunomodulatory: M, mesenchymal: MSL, mesenchymal stem-like: UNC, uncertain. PAM50 and TNBCtype subtypes were not available for the Jönsson et al. cases. c Summarized results from consensus clustering, using a range between 3 and 7 cluster solutions (k, columns) and the same gene set and samples as in (b). For each cluster solution, the percentage of BRCA1 hypermethylated (left matrix) and BRCA1-null (right matrix) cases in the different defined clusters are shown (rows). d Principal component analysis based on the expression of the 7224 genes to relate variance in gene expression with BRCA1 status. A more intense red color for a principal component (columns) with a variable (rows) implies a stronger association with variance in expression for that component. Components represent variation in decreasing strength (first component largest variation with contributed percentage variation listed). A variable strongly associated with variation in expression is thus represented by an intense red principal component capturing a high percentage of the variation. All p values reported from statistical tests are two-sided. Source data are provided as a Source Data file.

Fig. 6. Immune cell infiltration phenotypes in BRCA1-null and hypermethylated cases.

a PD-L1 scoring of 53 BRCA1 hypermethylated and 25 BRCA1-null SCAN-B cases using the Roche SP-142 antibody that is evaluated in immune cells. To the left, proportion of positive cases (≥1%), to the right distribution of actual PD-L1 scores for the two groups. b Scoring of tumor infiltrating lymphocytes (TILs) in 52 BRCA1 hypermethylated and 24 BRCA1-null SCAN-B cases based on available whole section H&E slides. c Total number of expressed neoantigens per sample as calculated from substitutions by NeoPredPipe for BRCA1-null and BRCA1 hypermethylated SCAN-B cases. d Neoantigens as shown in (c), but stratified for sample groups also by PD-L1 IHC status. Four hypermethylated cases did not have available PD-L1 immunohistochemistry data. All p values reported from statistical tests are two-sided. Source data are provided as a Source Data file.