Tumor analysis of BRCA carriers reveals genomic similarities although separated by time Authors: Tal Falick Michaeli, Avital Granit Mizrahi, Batia Azria, Ofra Maymon, Shai Rosenberg, Jonathan Monin, Esther Braitbart Cohen, Myriam Maoz, Luna Kadiuri, Hovav Nechushtan, Amichay Meyrowitz, Tamar Peretz Published: 2024 ABSTRACT AbstractAmong the dominant pathogenic genes (PG) in breast cancer are BRCA1/2. Knowing whether a patient carry one of these alterations is meaningful as it affects management. A substantial question is to what extent are the genomic profile of a tumor and its characteristics affected by the germline profile of BRCA1/2 and what is the possible contribution of other environmental factors. Here, we compared the molecular characteristics of two subsequent primary breast cancers in three women (6 primary breast cancers) BRCA PG carriers, and in two of them also a primary lung cancer. Comparing two different tumors in the same patient neutralizes the contribution of other germline changes, and may demonstrate possible effects of other external insults occurring between the first and second tumor. Nonetheless, epigenetic changes resulting from early life events will be present in all tumors that the patient has developed. Our analysis suggests that tumors arising from the same tissue in the same patient share similar molecular characteristics, albeit occurring in different times, as the patient is exposed to a variety of external stimuli.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12672-024-01577-x.Keywords: BRCA1/2 carriers, Exome analysis, Breast cancer, Molecular signature, Second primary Introduction Breast cancer (BC) is the most prevalent female malignancy worldwide. The pathogenesis of BC is mostly sporadic, with familial genetic cases accounting for 5–10% [1]. Among the molecular changes occurring in BC, BRCA1/2 pathogenic genes (PG) are probably the most studied. These PGs are highly penetrant and associated with the Hereditary Breast/Ovarian Cancer Syndrome, an autosomal-dominant inherited trait predisposing women to both breast and ovarian cancer [2, 3]. Women with BRCA PG have a lifetime risk of developing breast cancer (38%–87%) and ovarian cancer (16.5%–73%). In addition to female breast and ovarian cancers, BRCA PG are associated with, male breast cancer, pancreatic, stomach colorectal and prostate cancer [1]. The wide variation in penetrance, can be attributed to a variety of modifiers, either germline changes, or external factors, either environmental or lifestyle [4–6]. To date it is unclear what is the crucial age of external exposure that affects penetrance. Better understanding of the contribution of both germline modifiers and external insults, may lead to better personalized screening programs for BRCA carriers, and possible advanced recommendation for prevention, thus reducing morbidity and mortality in BRCA PG carriers. We recently demonstrated that among BRCA carriers, daughters of BRCA1/2 mothers with a similar genomic profile develop BC at a younger age compared to their mothers [7], pointing to the existence of external factors affecting penetrance, presumably early life events. Germline testing is important for therapy management, since carriers of BRCA1/2 PG derive enhanced benefit from platinum-based chemotherapy or poly (ADP-ribose) polymerase inhibitors (PARPi) and in addition, according to a recent report, might predict response to immunotherapy [8, 9] Moreover, genetic testing for BRCA1/2 PG helps clinicians target unaffected high-risk women for close surveillance that has a significant survival benefit [10] as well as for preventive measures. Up to 60% of BRCA1 PG BCs are triple-negative BC [11, 12] yet the clinical spectrum is wider and some BRCA1/2 carriers suffer from multiple tumors [13]. BC in BRCA1/2 PG carriers are characterized by a specific mutational pattern, Signature 3 [14]. The aim of the study was to investigate the relative contribution of genetic modifiers, as well as external modifiers, in particular early life events, on the development of breast cancer in BRCA carriers. We evaluated the genomic profile of a tumor, and how it is affected aby the germline profile of the patient. We compared the molecular characteristics of two subsequent primary breast cancers in three women (6 primaryBCs) BRCA PG carriers, and in two of them also a primary lung cancer. Comparing two different tumors in the same patient neutralizes the contribution of other germline changes, and may demonstrate possible effects of other external insults occurring between the first and second tumor. However, epigenetic changes resulting from early life events will be expressed in all tumors that the patient has developed. Our analysis suggests that tumors arising from the same tissue in the same patient share similar molecular characteristics, albeit occurring at different times, as the patient is exposed to a variety of external stimuli. Materials and methods We carried out WES on 6 primary breast cancer, 2 from each individual, as compared to germline WES in each patient. Moreover, we sequenced 2 lung cancer samples from 2 different patients. DNA was extracted from FFPE slides using the QIAamp DNA Mini Kit according to the manufacturer’s protocol. The quality and quantity of the DNA were assessed using a Nanodrop spectrophotometer (Thermo Scientific) and Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific), respectively. DNA integrity was further evaluated using an Agilent 2100 Bioanalyzer. Exome libraries were prepared using the KAPA HyperPrep Kit (Roche) following the manufacturer’s guidelines. Exome capture was performed using the Roche MedExome Nimblegen capture kit according to the manufacturer's protocol.The final captured libraries were pooled in equimolar amounts and loaded onto the Illumina NextSeq 500 using the Mid Output Kit v2 (300 cycles) for paired-end 150 bp sequencing, sequencing was conducted according to the Illumina protocol.FASTQ files were filtered for adapter contamination and low quality bases were trimmed using trim_galore, wrapper tool around Cutadapt and FastQC, version 0.6.1 [15]. Filtered FASTQ files were mapped to human assembly GRCh38 with bwa-mem2, a software package for mapping DNA sequences against a large reference genome, version 2.2.1 [16]. Duplicates were removed with Picard tool, a set of Java command line tools for manipulating high-throughput sequencing (HTS) data and formats [17]. Specifically, Picard tool MarkDuplicates, version 2.23.8 was applied with REMOVE_DUPLICATES flag turned on. Our pre-QA depth coverage figures across all regions of interest (exons) are presented in Supplementary Fig. 1a with 3 samples being well covered by more than 550 reads per region, two samples exhibited moderate coverage above 200 and only one sample featuring passable depth coverage of 140 reads per region. Post QA depth coverage per base was also analyzed for a threshold minimal depth coverage of 10 reads per base. This was calculated utilizing random sampling of 1,000 exons and reveals that more than 50% of the bases fulfill the minimum threshold (Supplementary Fig. 1b). We conclude that a substantial number of individual bases within our targeted regions are adequately represented and sequenced. Copy number variations were explored with CNVkit: Genome-wide copy number from high-throughput sequencing, version 0.9.9 [18–20]. CNVkit batch command was applied for examining somatic CNV by comparing primary, secondary and lung tumors each against its germline PBMC (normal) sample. Targets are set as GRCh38 exons downloaded from Ensemble Biomart [21]. Low coverage sites were dropped. Genes exhibiting CNV were then extracted with CNVkit tool genemetrics by setting CNV threshold of 0.4 and at least five reporting probes per genes. Mutual CNV gene sets were determined, and VENN plots were created using eulerr [22] package. P-value for obtaining the number of genes in the overlapping set, given the two original groups of genes, was calculated by direct probability calculation utilizing hypergeometric distribution (Figs. 1, 5 and 6). Genes in CNV segments. VENN diagram showing the overlapped genes in CNV segments between all of the primary breast tumors (A), and the second primary breast tumors (B). Assuming that about 20,000 genes are candidates for being mutual, the p-value for obtaining 56 mutuals out of two groups sized 941 and 448 is 2.5E− 11 Characterization of lung samples. a Venn diagram for shared SNVs for lung samples. Assuming that ~ 20,000 genes are candidates for being mutual, the p-value for obtaining 1123 mutual out of two groups sized 1748 and 8919 is 4.8E− 67. b Pathway analysis based on SNVs. All of the mutual genes associated with SNVs of the Lung tumors (L) Venn plot for SNV based genes for patients’ primary and second-primary breast cancer. Patient numbers 1, 2, 3 correspond to Venn's a, b, c. Statistics are as follows: assuming that ~ 20,000 genes are candidates for being mutual, the p-value for obtaining 2,610 shared genes out of two groups sized 5591 and 6212 is 1.9E− 187 (a). The p-value for obtaining 2616 shared genes out of two groups sized 4417 and 8258 is 1.2E− 163 (b). The p-value for obtaining 1606 shared genes out of two groups sized 2639 and 7820 is 4.5E− 130 (c) Somatic single nucleotide variations (SNVs) were called with GATK Mutec2 tool version 4.2.0.0 [23]. Each tumor sample is compared with its normal matched germline peripheral blood mononuclear cells (PBMC) sample. The resultant Variant Calling Format (VCF) file was then filtered with GATK tool FilterMutectCalls at its default settings. Nucleotide variations were aggregated per sample and tested against COSMIC mutational signatures (Fig. 2). Signature refitting is performed with R package MutationalPatterns version 3.2.0 [24] by solving a least-squares non-negative constrained problem for the mutation matrix. Figure 2 shows linear combination of the top 5 contributing signatures for each sample, some of which are recurring. Two sets of COSMIC signatures were tested: the newer version 3.2 [25] and the previous version 2.0 [26] which allows for cancer type mapping. Mutational signature. Plots show the top mutational signatures that most closely reconstruct the mutation profile of each sample. Signature contributions are quantified by the size of each colored bar and normalized to a total of 100. Two mutation sets have been tested, version 2 which consists of 30 signatures (a) and the newer version 3.2 which consists of 60 signatures (b). Primary breast tumors (A), second primary breast tumors (B), lung tumor samples (L) Mutual genes exhibiting SNP were selected for groups A, B and L. These mutual genes do not necessarily show the same SNP between samples of the group. Pathways were searched for these sets of genes using enrichR R package version 3.0 [27–29] limited to Biological, Cellular and Molecular GO terms and to KEGG and Wiki pathways. Results are ranked by enrichR combined score which weighs p-values and z-score results. Figures 3, 4 and 5 present the top scored pathways after applying a threshold limit 20, adjusted p-value of 0.05 and at least 5 genes per pathway. Bar lengths are log-transformed for better visualization. Combined score multiplies p-value and z-score as described in [27]. Pathway analysis based on SNVs. a Venn diagram for shared SNVs for primary breast cancer (“A”) samples. p-value (upper limit) < 3.7E− 271. b All of the mutual genes associated with SNVs of the A samples Pathway analysis based on SNVs. a Venn diagram for shared SNVs for second primary breast cancer (“B”) samples. p-value (upper limit) < 5.1E− 146. b All of the mutual genes associated with SNVs of the B samples GSEA [30] Hallmark pathway analysis (Supplementary Fig. 2) was used applying the top 500 genes on CNV segments in each of the shared CNVs of all samples of A and B. Results Three female patients were included in our study. All three women carry BRCA PG, as described in Table 1. All three developed BC in both breasts with an interval between the occurrences. Patients 1 and 2 are Ashkenazi Jewish women, patient 3 is an Iraqi Jewish woman. Table 1Patient characteristicsPatient 1Patient 2Patient 3ClinVar mutationBRCA1 c.5266dupCBRCA2 c.5946delBRCA1 c.224_227delDemographicsAshkenazi born in IsraelAshkenazi born in IsraelNon-Ashkenazi born in IraqAge*503060Sample siteLeft breastLungRight breastLeft breastLeft breastRight breastLeft breastRight breastLungSample labelA1L1B1A2B2A3B3L3Years*0341151Grade33333333Tumor (mm)1525152015204015.625Node’s status1010NA0100Estrogen receptor staining grading (0–3)3NA033130NAHER2 staining grading (0–3)1NA1NANANANA0NAMode of SurgeryLump-ectomyLob-ectomyMast-ectomyMast-ectomyLocal excisionMast-ectomyMast-ectomyMast-ectomyLob-ectomyAdjuvant Hormonal therapy (yes/no)YesNoNoNoNoNoYesNoNaChemotherapy treatment (yes/no)YesNoYesNoNoYesNoNoNoAdjuvant radiotherapyYesNoNoNoYesNoNoNoNoAge* = Age of the patient at first diagnosis, Year* = Years separating the initial and following diagnosis Patient characteristics Age* = Age of the patient at first diagnosis, Year* = Years separating the initial and following diagnosis Patient 2 had left chest wall recurrence after left mastectomy, and subsequently developed primary BC. Exome analysis was performed on the chest wall recurrence and the subsequent contralateral tumor. Two patients had PVs in BRCA1, One patient had a PV in BRCA2. All initial tumors were ER positive. All subsequent tumors were triple negative. Patients 1 and 3 received hormonal therapy after the first diagnosis, patient 3 refused hormonal therapy after the initial surgery and after the chest wall recurrence. Lung cancer was concomitant with the first BC in patient 1, and was not in the radiation field in patient 3. Two women also developed primary lung cancer, as well (Table 1). Exome sequencing was generated for germline DNA (peripheral blood mononuclear cells = PBMC), tumor DNA—primary and second-primary of BC, as well as exome sequencing from lung tumors for patient number 1 and 3. First, we wanted to explore the similarities in all A samples, which are the primary BC samples for each patient. Secondly, we wanted to compare the similarities in B samples, representing the second-primary BC samples. We carried out copy number variation (CNV) analysis for all mutual A and B samples (Fig. 1). We discovered 941 genes corresponding to these CNVs in A samples, and 448 genes in B samples. We then took the top 500 genes (as calculated by the highest absolute value of CNV) and carried out pathway analysis of Hallmark pathways in oncology. We discovered pathways of DNA repair and P53 pathways. The Myogenesis pathway was enriched in both A and B groups, and others as shown in Supplementary Fig. 2. By looking at the mutual A and B genes we discovered 56 genes which are statistically significantly enriched. However, pathway analysis did not identify enriched pathways. Next, we examined whether primary BC and second-primary BC occurring years after the first diagnosis, especially following hormonal treatment or chemotherapy, exhibit different phenotypes influenced by this environmental exposure. In order to answer this question, we employed two different molecular signatures algorithms. For each sample, the top 5 most contributing mutational signatures are found using the function fit_to_signatures of R package MutationalPatterns. A combined list is then created, consisting of the top 5 signatures from all samples. Most of the top signatures are repeated across two or more samples, thus lowering the total number of signatures in the combined list. Results are presented in Fig. 2. For visual clarity only, each sample is restricted to be presented by up to 10 mutational signatures. In general, most samples pairs harbor similar proportions of the contributing signatures with some small differences. In the first version, Signature 1, which usually exists in all cancer types, is not ranked in the top 5 of sample 1B (Fig. 2a). Lung samples differ from breast cancer samples by their loss of Signature 8 (BC marker), loss of Signature 29 and a possible increase (intragroup) in Signature 11. Lung samples also differ in Signature 30 (BC marker) which is maintained and elevated in 1L but lost in 3L. Also, 1L is the only sample harboring Signature 23 in its top 5 signatures (Fig. 2a). In the second updated version of mutational signature (Fig. 2b), observable features of lung samples are the addition of SBS23, and a decrease in SBS3 (a BC marker) in 1L and lost in 3L. Taken together, it appears that in terms of mutational signatures we did not detect any significant changes between tumors arising from the same tissue. Following this, we focused on single nucleotide variations (SNVs). We correlated the genes associated with the SNV in all mutual A samples and discovered enriched pathways (Fig. 3), mostly for extracellular matrix (ECM) organization and epigenetic regulation within the cells. Moreover, we could detect neurological-associated pathways presumably indicating motility of the cells. Results are ranked by enrichR combined score which weighs p-values and z-score results (see Methods). We conducted the same analysis for B samples (Fig. 4). We identified the PI3K signaling pathway among the top enriched pathways, as well as the Notch signaling pathway. Again, we found ECM-associated pathways in several characterized pathways. 1748 SNVs were associated with sample L1 from patient number 1, occurring at the second site of tumor formation (Table 1). 8919 SNVs were related for L3, which is the third site of tumor in that patient. Overlapping SNV were 1123 (Fig. 5a), generating the pathways in Fig. 5b. Lung tumor samples showed diverse pathways such as metabolic pathways associated with hormonal secretion, ECM-related pathways, regulation of transcription pathways and carcinogenesis pathways (Fig. 5b). When looking at SNVs for each patient for primary and secondary BC tumors we detected significant overlap between the two cancers (Fig. 6), reinforcing our findings of mutational signature analysis above and demonstrating similarities in the molecular alteration from a tumor of the same origin, but not occurring in a synchronous manner. Discussion BRCA PG carriers are known to have an increased risk for BC as well as other malignancies. BC can occur throughout life several times in the same patient in the ipsilateral as well the contralateral breast. WES detects a broad repertoire of somatic and/or germline alterations in an unbiased manner. The frequency and distribution of these mutations serve as an imprint of mutational processes or exposures that contribute to tumor development and progression [31]. We aimed to dissect the possible role of germline versus external modifiers, that affect the development of breast cancer in BRCA PG carriers. We have demonstrated that the second breast tumor shares significant genomic similarities with the first tumor, thus pointing to the fact that life events occurring after the development of the initial breast tumor did not impact significantly tumor characteristics. These similarities can be explained either by the strong effect of other germline modifiers, or the impact of early life events, or probably both. One important question we address in our unique data set is whether the second primary BC is influenced by former systemic treatments given to the patient or other external factors. Our study carries small-sample limitations. However, using several analyses including mutation signature, SNV, CNV and pathway analysis, we demonstrate that the molecular characteristics of the second primary BC are quite similar to the first primary. The mutation signature analysis could suggests little environmental contribution to tumor development in these BRCA carriers or accumulation of other changes in the epigenome, given the HRD state of the carriers' genome. Of note, that the two lung tumors did not develop on a background of radiation therapy, as demonstrated in the result section, by the tumor time of occurrence and molecular signature. Interestingly, the initial BC was ER positive, while subsequent BC was triple negative. This can be explained only partially by possible exposure to anti-hormonal therapy for the first tumor, since patient 2 was not exposed to such therapy. Additionally, our results suggest that tumor samples arising from the same tissue, breast or lung, more resemble one another across all patients than to tumors arising in the same patient, which occurs from different organ origin. This analysis is in line with our knowledge of tumor features [14, 32, 33]. Pathways analysis reinforce these findings, showing BC-related pathways in BC WES data as compared to lung tumor related pathways, which show different pathways. Furthermore, our pathway analysis captured mutual pathways as was described recently in a paper investigating driver mutations in primary and recurrent BC and ovarian cancer in BRCA carries [34]. We recently published a paper showing Signature 3, also known as the BRCA1/2 associated pattern of genome-wide mutations, to be universal in all of our analyzed BC BRCA carriers [12], however, Signature 3 was not consistently elevated in the patients in the current study. These changes and the switch in receptors status of the patients, between primary and second primary tumors can be attributed, at least in part, to heterogeneity of the tumors. We evaluated 8 exomes in three female patients. Within the limitation of small sample size, we can carefully stipulate that primary BC in the same women carry strong similarity to each other although developed subsequently, with the second tumors being exposed to more external insults than the first BC. In different BC patients, the diversity of tumor characteristics can be attributed to various genetic modifiers from patient to patient. Our unique data set, which is comprised of samples of 2 primary tumors in the same woman, lowers down the effect of genetic modifiers and enables us to focus on external modifiers. These findings do not preclude the impact of external modifiers on the development of BC. It may suggest that early life exposures do affect later on BC diagnosis, and that these early external modifiers carry dominant effects on the character and development of subsequent cancers. Such early life events, which differs along the recent decades, may explain the genetic anticipation of BC among BRCA PG carriers. Puzzling is the different pattern of tumors by organ origin and possible interaction with specific organ-associated microenvironmental factors that give rise to different molecular profiles. Further research is warranted to elucidate genotype–phenotype interactions in the evolutions of breast cancer, as a multifactorial disease, and the role of the organ associated microenvironment in cancer development in BRCA1/2 PG carriers. Better evaluation of germ line modifiers, and understanding external early life events, can provide better, personalized screening programs, and preventing manners. Limitations: The conclusions we draw from our data is limited due to small cohort composed of six women. Supplementary Information Additional file1 (PDF 311 KB) Additional file2 (PDF 273 KB) Acknowledgements Not applicable. Author contributions T.F.M., A.G.M, B.A., O.M., S.R., J.M., E.B.C., M.M., L.K., H.N., A.M., and T.P. collected and assembled data and interpretation; T.F.M., J.M. and T.P. analyzed data; T.F.M. and T.P design and wrote the manuscript. Funding No funding was received. Data availability Data is provided within the manuscript or supplementary information files. Declarations Competing interests The authors declare no competing interests. Ethical approval and consent to participate The study was approved by the Institutional Review Board (IRB) of Hadassah Medical Organization. Patient consent for publication Not applicable. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Amichay Meyrowitz and Tamar Peretz share equal senior authorship contribution. References Associated Data Supplementary Materials Additional file1 (PDF 311 KB) Additional file2 (PDF 273 KB) Data Availability Statement Data is provided within the manuscript or supplementary information files.