Paired tumor sequencing and germline testing in breast cancer management: An experience of a single academic center

Abstract Background Genetic testing for cancer predisposition is recommended to women with breast cancer who meet the criteria for such testing. After the FDA approvals of the poly ADP ribose polymerase (PARP) inhibitors, olaparib and talazoparib, for treatment of metastatic breast cancer, carrying germline mutations in BRCA1 and BRCA2 genes, the genetic testing result has become critical in their care. With the recent FDA approval of alpelisib for the treatment of PIK3CA‐mutated hormone‐receptor positive metastatic breast cancer, tumor molecular profiling to identify somatic mutations and potential molecularly targeted agents is increasingly utilized in the treatment of advanced breast cancer. Aim Combining germline and somatic sequencing (paired testing) offers an advantage over a single technique approach. Our study evaluates the role of paired testing on the management of breast cancer patients. Methods and Results Forty‐three breast cancer patients treated at Rush University Medical Center underwent paired germline and somatic variant testing in 2015 to 2017. A retrospective chart review was conducted with the analysis of demographic, clinical, and genomic data. Three actionable germline variants were found in the CHEK2 (2) and ATM (1) genes. 95% of tumors had somatic mutations. Seventy‐seven percent of tumors had genomic alterations targetable with agents approved for breast cancer and 88% had molecular targets for agents approved for other cancers. Clinical examples of such use are described and potential future directions of tumor and paired testing are discussed. Conclusions Germline variants were present in a relatively small patient group not routinely tested for inherited alterations. Potentially targetable somatic alterations were identified in the majority of breast cancers. Paired testing is a feasible and efficient approach that delivers valuable information for the care of breast cancer patients and eliminates serial testing.

Background: Genetic testing for cancer predisposition is recommended to women with breast cancer who meet the criteria for such testing. After the FDA approvals of the Poly ADP Ribose Polymerase (PARP) inhibitors olaparib and talazoparib for treatment of metastatic breast cancer carrying germline mutations in BRCA1 and BRCA2 genes, the genetic testing result has become critical in their care. With the recent FDA approval of alpelisib for the treatment of PIK3CA-mutated hormonereceptor positive metastatic breast cancer, tumor molecular profiling to identify somatic mutations and potential molecularly targeted agents is increasingly utilized in the treatment of advanced breast cancer. Aims: Combining germline and somatic sequencing (paired testing) offers an advantage over a single technique approach. Our study evaluates the role of paired testing on the management of breast cancer patients. Methods and Results: 43 breast cancer patients treated at Rush University Medical Center underwent paired germline and somatic variant testing in 2015 -2017. A retrospective chart review was conducted with the analysis of demographic, clinical, and genomic data. Three actionable germline variants were found in the CHEK2 (2) and ATM (1)  Clinical examples of such use are described and potential future directions of tumor and paired testing are discussed. Conclusions: Germline variants were present in a relatively small patient group not routinely tested for inherited alterations. Potentially targetable somatic alterations were identified in the majority of breast cancers. Paired testing is a feasible and efficient approach that delivers valuable information for the care of breast cancer patients and eliminates serial testing. Keywords: Breast cancer, genomic profiling, next-generation sequencing, germline testing, cancer predisposition, clinical utility Recent advances in precision oncology allow for the administration of more effective and less toxic treatments, leading to improved patient outcomes. Molecularly targeted agents (MTAs) such as imatinib in chronic myelogenous leukemia or trastuzumab in Her-2/neu-positive breast cancer have shown significant prolongation of overall survival in the corresponding malignancies and sometimes, cures. [1] A meta-analysis of Phase I trials with more than 13,000 patients showed that biomarker-driven MTAs as compared to non-personalized treatments resulted in a higher response rates (30.6% vs 4.9%) and better progression-free survival (PFS; 5.7 mo vs 2.9 mo). [2] In addition, MTAs usually have fewer side effects (on-target or off-target) than cytotoxic agents. There is an everincreasing number of clinical trials underway testing new targets and matching therapies. [3] In January of 2018, a PARP-inhibitor olaparib became the first genomically-based treatment approved by the United States FDA (Food and Drug Administration) for patients with metastatic breast cancer (MBC), specifically those who carry germline BRCA1 or BRCA2 mutations. [4,5] Subsequently, two more MTAs (talazoparib and alpelisib) have been FDAapproved for metastatic breast cancer with a germline BRCA mutation (talazoparib) and somatic PIK3CA alteration (alpelisib). [6,7] Advances such as these confirm the promise of molecularly based approaches in breast cancer (BC) management. Currently, therapy in BC is guided by the presence of estrogen and progesterone receptors as well as human epidermal growth factor receptor 2 (HER2) protein overexpression or amplification on pathologic examination. [8][9][10][11] The introduction of next-generation DNA sequencing (NGS) and recently, mapping , whole cancer genomes to find cancer "drivers", allowed to identify promising MTAs. [12,13] At present, however, alpelisib is the only MTA which requires identification of PIK3CA mutation in the tumor to predict its efficacy.
Tumor molecular profiling (TMP) is usually employed in advanced BC when all standard treatments have been exhausted. [14] Since the technology of TMP with NGS became available, many experts considered it investigational and appropriate only in a clinical trial setting. Yet in March of 2018 Centers for Medicare & Medicaid Services made the decision that NGS is "reasonable and necessary" in advanced cancer for Medicare beneficiaries, opening the doors for wider use of this laboratory testing.
[15] NGS is also employed in germline hereditary predisposition testing, which typically involves the use of multi-gene panels on constitutional DNA isolated from blood cells.
Germline testing provides valuable information on cancer surveillance and risk reduction for patients and their biological relatives. It may also provide information for MTA selection, such as identifying a germline BRCA1/2 mutation in a patient with metastatic BC who may benefit from treatment with Poly ADP Ribose Polymerase (PARP) inhibitors. Combining germline testing with TMP (paired testing or PT) may provide a time and resource efficient option to reveal molecular targets for therapy and genetic predisposition to cancer. [16] We performed a pilot study of PT in a series of 43 BC patients to assess the feasibility and clinical impact of this testing approach at a single academic medical center. We discuss the implications of PT results for treatment selection, subsequent cancer surveillance, and screening of family members. We also discuss efficiency of the PT approach.

Methods:
We performed a retrospective chart review of women with the diagnosis of BC who underwent PT when deemed appropriate by their oncologists in the context of routine clinical care at Rush University Medical Center from November 2015 to February 2017. The following data were collected: patient age, ethnic background, clinical stage at diagnosis, histology, ER/PR/HER2 receptor status, BC treatment history, and PT results. All patients received pre-test risk assessment and genetic counseling to inform them about PT. Post-test genetic counseling was provided for patients and/or their families who tested positive for a pathogenic or likely pathogenic germline variant.
The study was approved by the Rush University Institutional Review Board (IRB# 17041703-IRB01). Participants were deidentified before group statistical analyses.

Laboratory analysis
DNA was extracted from a primary tumor for the majority of subjects while a metastatic lesion was analyzed when the primary tumor was not available. Blood or saliva was also submitted for each patient for germline DNA extraction. Samples were processed as previously described. [16] PT consisted of a custom probebased NGS tumor panel (Illumina HiSeq) for the detection of single nucleotide variants, small insertions and deletions in 142 genes that frequently harbor somatic and/or germline mutations in cancer (Online Resource Supplemental Table 1). The NGS panel used detects gene fusions and structural variants, such as tandem duplications and inversions, in 15 frequently disrupted oncogenes and tumor suppressors. Tumor tissue and a matched blood specimen were analyzed using a custom bioinformatics pipeline to differentiate between somatic and germline mutations, allowing for precise variant classification. Briefly, paired normal samples were analyzed using Novoalign V3.02.07 to align FASTQ reads to a reference sequence (hg19) and GATK (V3.2.2) to generate variants and no/low coverage reports. Germline variants were filtered using a Q score of 30, coverage of 10X, het ratio of 10% and filtered out if determined to be a sequencing artifact or common polymorphism utilizing population frequency data from NCBI dbSNP, NHLBI Exome Sequencing Project (ESP), 1000 Genomes, and internal Ambry data. In tumor-normal analysis mode, Varscan2 (v2.3.6), was used to detect somatic variants as low as 3% minor allele frequency. Structural variants were annotated using Oncofuse v1.0.7 and DELLY v.0.6.1 respectively.
Germline genetic variants were assessed using Ambry's five-tier classification framework based on guidelines published by the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (pathogenic; likely pathogenic; variant of uncertain significance; likely benign; benign). [17,18] Tumor specimens were also analyzed using the Affymetrix OncoScan platform, an array technology for high resolution copy number variant detection that can detect single copy amplifications, hemizygous deletions and copy neutral loss of heterozygosity.
Alterations identified in TMP and associations with MTAs were reported based on peer reviewed studies and other publicly available resources. Germline variants when identified in blood or tumor were reported for the following genes: APC, BRCA1, BRAC2, MLH1, MSH2, MSH6, MUTYH, PMS2, PTEN, RB1, RET, SDHAF2, SDHB, SDHC, SDHD, STK11, TP53, TSC1, TSC2, and VHL. In some cases, germline testing was performed on additional cancer susceptibility genes as clinically indicated based on genetic risk assessment. Actionable germline alterations were defined as those alterations associated with a currently available option for cancer surveillance, prevention or treatment in either the patient or their close family members.

Results:
Age at diagnosis, ethnicity, stage at the time of testing, histologic type, and tumor markers of breast cancer for 43 patients are reported in Table 1 Somatic genomic alterations in the tumors tested were highly prevalent with 95% (42/43) of tumors having at least one reported mutation with an average of 6.5 (range 0-17) mutations per tumor. Overall, 77% of tumors contained at least one target for BC therapy, and 88% contained an FDA-approved targeted therapy for another type of cancer, potentially available off-label for BC. ( Table 2) A total of 278 alterations in 81 genes were clinically reported in Figure 1, and are broken down by ER receptor status in Supplemental Graph 1. Hemizygous loss of a gene was the most common alteration (n=96) while missense alterations (n=79) and gene amplifications (n=66) were also frequent (Online Resource Supplemental Table 2). Genes known to be frequently mutated in BC (BRCA1, BRCA2, CDH1, PIK3CA, PTEN, and TP53) accounted for 40% of alterations. 22 genes altered in this cohort of patients had an associated MTA for breast cancer while alterations in additional genes added to the number of patients who were potentially eligible for an MTA ( Figure 2). Patient characteristics in relation to the alterations can be found in Online Resource Supplemental Table   3.
Three patients were found to have actionable germline variants that were not previously suspected including one pathogenic mutation in CHEK2 (c.1100delC), a likely pathogenic variant in the CHEK2 gene (c.1427C>T), and a likely pathogenic splice site variant in the ATM gene (2466+1G>C). None of these three patients met the then current National Comprehensive Cancer Network (NCCN) germline genetic testing criteria.
[22] We have performed TMP on breast cancers of the three germline carriers described above ( Table 3). Two of these tumors had multiple molecular alterations.

Clinical Application of PT Results:
Because 47% of our sample (20 patients) consisted of patients with early-stage breast cancer, real time clinical care was usually not impacted by PT but rather, adjuvant or neoadjuvant treatment was based on stage, pathologic characteristics of the tumor including tumor markers, and in some cases on the results of molecular prognostic assays such as Oncotype DX ® and Mammaprint ® . In one patient (Patient C), an MTA was considered in adjuvant setting and in two women (Patient A and Patient B), MTAs were administered when BC recurred.  Table 3) was diagnosed with inflammatory triple-negative left-sided IDC, clinical stage III, grade 3, at age 66. She received four cycles of neoadjuvant dose-dense doxorubicin and cyclophosphamide, followed by four cycles of paclitaxel, and underwent left modified radical mastectomy which demonstrated significant residual BC: ypTis N3a. After radiation to the left chest wall, adjuvant chemotherapy with carboplatin was attempted but not tolerated due to rash and weakness. She was diagnosed with histologically confirmed recurrence of triple-negative BC and metastases to intra-thoracic lymph nodes a year after her surgery. PT was performed on the primary tumor and revealed four somatic genomic alterations including KRAS amplification, BARD1 p.A724V, TP53 p. K139_P142del, and TP53 copy neutral LOH. Based on these findings, she was prescribed a suggested MTA off-label, Trametinib, a mitogen-activated protein kinase kinase (MEK) inhibitor which targets KRAS amplification. Although it does not directly inhibit MEK, Trametinib has been shown to limit tumor progression via CD8 T-cell mediated factors by altering signaling along the RAS-ERK pathway in cancers with KRAS amplification.
[23] Unfortunately, trametinib was discontinued one month later due to progression. Table 3) was diagnosed with clinical stage III, grade 3, triple-negative rightsided IDC at age 62. She received four cycles of neoadjuvant dose-dense doxorubicin and cyclophosphamide, followed by 12 cycles of weekly paclitaxel. She underwent subsequent right modified radical mastectomy with residual tumor noted, ypT2N0. She then completed radiation and received six cycles of adjuvant chemotherapy with capecitabine. She also was diagnosed with BC recurrence and histologically-confirmed pulmonary metastases one year after surgery. She underwent PT using a tissue sample from her mastectomy, notable for four somatic genomic alterations: PIK3R1 p.R461_E462delinsQ, FGF3 amplification, FGF4 amplification, and HER2-Neu (ERBB2) mutation, p.S310F. Given the finding of HER2-Neu mutation, she was offered participation in phase II study with neratinib.

Patient 435327 (Supplemental
[24]. Because somatic HER2-Neu mutations activate the HER2-Neu oncogene without its amplification, Trastuzumab and other commonly used HER2-Neu directed agents are usually ineffective in HER2-Neu-mutated BC. Neratinib is a small molecule which irreversibly inhibits both HER2-Neu amplifications and mutations.
[25] The patient enrolled in this trial but it was discontinued 3 months later due to cancer progression.
Patient 487820 (Supplemental Table 3  and a hemizygous loss of NF2, yielding four possible therapies including an mTOR inhibitor everolimus, approved in combination with anti-hormonal therapy for metastatic BC. Given the significant amount of residual disease and this molecular alteration, she was evaluated for the adjuvant trial with everolimus.
[26] However, she was found to be ineligible because positive lymph node biopsy was not followed by axillary lymph node dissection based on patient and surgeon preference. She underwent adjuvant radiation therapy and started anti-hormonal therapy with tamoxifen. At the time of this writing, she remains without evidence of disease 3.5 years after surgery.

Discussion:
In our study, 95% of all tumors analyzed demonstrated at least one somatic mutation presumed to be involved in tumorigenesis. The most commonly detected mutation was in TP53 gene, followed by CDH1 and PIK3CA. This correlates with other studies, which have described TP53 as the most commonly mutated gene in human cancers.
[27] As seen in Figure 1, the rate of gene alteration in TP53 was less than 12% and that in PIK3CA was less than 6%, which are lower than reported results for breast cancer. Those studies, however, include noncoding regions with high mutation frequencies not all of which can be classified as driver mutations. Mutational signatures extended to genome rearrangements, characterized by tandem duplications or deletions, appear to be associated with defective homologous recombination-based DNA repair.
[28] Analysis of all classes of somatic mutation across exons, introns and intergenic regions might generate a higher overall mutation rate, not all of which will be driver mutations. In general, mutations in tumor suppressor genes (loss of function mutations) like TP53 are more difficult to target than mutations in oncogenes (gain of function mutations).
[29] PARP-inhibitors is the first class of MTA that targets mutations in tumor suppressor genes (BRCA1 and BRCA2). Preliminary data suggest that these agents can also elicit synthetic lethality in the presence of mutations in other tumor suppressor such as TP53 and PTEN which are mutated in 37% and 3% of all BCs respectively and represented 11.8% and 5.  When analyzing the results of MTA clinical trials, it is important to remember that they depend on target identification in tumor tissue and availability of matching MTAs. In the initial phase of the NCI sponsored MATCH (Molecular Analysis for Therapy Choice) trial, only 9% of patients had a matched targeted therapy, when there were only 10 MTAs and 10 corresponding arms in the study. However, the percentage of patients increased to 23% when the number of MTAs and the respective arms in the study went up to 24% . [36] Previous studies have revealed the prevalence of germline variants discovered in paired testing. A retrospective study of Stanford patients with BRCA1/2 somatic mutations found 55.7% were positive for pathogenic BRCA1/2 germline mutations, confirming "a second hit" hypothesis. [37] In a study of patients with advanced cancer diagnoses, presumed pathogenic germline variants were found in 17 of 269 (6.3%) breast cancer patients (6.3%). [38] In another study of 1040 breast, prostate, renal, pancreatic, and colon cancer patients, 182 (17.5%) had clinically actionable mutations conferring cancer susceptibility. Of these, 101 patients would not have had these mutations detected using clinical guidelines. Germline findings led to predictive testing in the families of 13 individuals (1.3%), including 6 for whom genetic evaluation would not have been initiated by guideline-based testing. [39] In our small sample, we identified three patients (7.0% of those tested) with previously unrecognized actionable germline mutations in CHEK2 and ATM genes which is consistent with these data.
Variants in CHEK2 and ATM can impact cancer surveillance and cancer screening and prevention for blood relatives. [40,41] For our patients, the discovery of these actionable germline mutations led to recommendations for increased breast cancer surveillance with the addition of screening breast MRIs (for both ATM and CHEK2 mutation carriers) and more frequent colonoscopies for colon cancer surveillance (for CHEK2 mutation carriers only).
The same recommendations would apply to their blood relatives if they carry these mutations. Currently, germline genetic testing for hereditary predisposition to breast and other cancers is only offered to individuals who meet specific testing criteria set forth by different societal guidelines, most commonly, the National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines. The NCCN guidelines recommend genetic testing for BRCA1/2 mutations based on age at breast cancer diagnosis and burden of breast, ovarian and other related cancers in the family. While these guidelines are useful to identify individuals who should be considered for genetic testing, germline mutations in the BRCA1/2 and other BC susceptibility genes may be missed in individuals who do not meet these criteria as their sensitivity is limited. [43,44] PT may allow for detecting pathogenic germline mutations in such BC patients. Importantly, somatic BRCA 1/2 mutations in the tumors of BRCA 1/2 mutation carriers can cause functional reversal of the germline mutation and restoration of the wild-type BRCA which results in resistance to Carboplatin and PARP-inhibitors in these patients. As opposed to "a second-hit" mutation which affects a normal allele, a "reversing" mutation usually affects the allele with a germline mutation. [45] Ongoing studies evaluate a potential benefit of PARP-inhibitors in the presence of somatic The same genomic alterations can be a target for a chemotherapy agent carboplatin which is known to double the objective tumor response rate in women with metastatic BC and germline BRCA1/2 mutations.
[47] HR deficiency in early-stage BC as measured by a HRD (Homologous Recombination Deficiency) molecular assay was associated with a higher chance to benefit from Carboplatin in neoadjuvant setting. [48,49] In a study of triple negative breast cancer, 59% of cases with homologous-recombination-repair deficiency had better outcome on In our study, all three germline mutation carriers had mutations in HR pathway (CHEK2 and ATM genes).
In addition, the somatic alterations in two of these three tumors also included many genes in HR pathway (BRCA1 and BRCA2, NF1 and NF2, FANCA) ( Table 3). It is logical to suggest that these somatic alterations (frequently, hemizygous losses) in the same pathway played a key role in the tumor development initiated by the germline mutation. Given this constellation of germline and somatic variants, it is likely that these cancers would respond to PARP-inhibitors and carboplatin if they recur.
As discussed earlier, TMP and PT data are usually not incorporated in treatment decisions for early BC. (PCAWG) Consortium found cancer genomes contained 4-5 driver mutations, however, in around 5% of cases no drivers were identified, suggesting that cancer driver discovery is not yet complete. [12] Our study showed that the use of PT is feasible in an actual clinical setting and a single academic

Ethical Statement
The study was approved by the Rush University Institutional Review Board (IRB# 17041703-IRB01).