Looking beyond year 1 in the molecular era of pediatric brain tumor diagnosis: confirmatory testing of germline variants found on tumor sequencing

Purpose Somatic molecular profiling of pediatric brain tumors aids with the diagnosis and treatment of patients with a variety of high- and low-grade central nervous system neoplasms. Here, we report follow-up targeted germline evaluation for patients with possible germline variants following tumor only testing in the initial year in which somatic molecular testing was implemented at a single institution. Patients and Methods Somatic testing was completed for all tumors of the central nervous system (CNS) undergoing diagnostic workup at Seattle Children’s Hospital during the study period of November 2015 to November 2016. Sequencing was performed in a College of American Pathologists-accredited, Clinical Laboratory Improvements Amendments-certified laboratory using UW-OncoPlex™ assay (version 5), a DNA-based targeted next generation sequencing panel validated to detect genetic alterations in 262 cancer-related genes. We tracked subsequent clinical evaluation and testing on a subgroup of this cohort found to have potential germline variants of interest. Results Molecular sequencing of 88 patients’ tumors identified 31 patients with variants that warranted consideration of germline testing. To date, 19 (61%) patients have been tested. Testing confirmed germline variants for ten patients (31% of those identified for testing), one with two germline variants (NF1 and mosaic TP53). Eight (26%) patients died before germline testing was sent. One patient (13%) has not yet had testing. Conclusion Clinically validated molecular profiling of pediatric brain tumors identifies patients who warrant further germline evaluation. Despite this, only a subset of these patients underwent the indicated confirmatory sequencing. Further work is needed to identify barriers and facilitators to this testing, including the role of genetic counseling and consideration of upfront paired somatic-germline testing.


Introduction
Over the prior two decades, large pediatric oncology sequencing studies have demonstrated the feasibility of upfront genomic testing at the time of high-risk diagnosis and relapse of childhood cancer (1,2).While access to sequencing via research protocols has increased for children with cancer over the prior decade, there is currently no consensus regarding best-practice diagnostics for routine clinical practice.
NGS assays designed to detect single-nucleotide variants, insertions and deletions, copy number changes, and rearrangements in genes selected for their clinical significance in cancer, can identify alterations that clarify diagnosis or suggest molecularly targeted therapeutics.In some cases, somatic sequencing can also result in secondary findings, identifying variants in the tumor sample that appear likely to be present in the patient's germline, which may or may not be related to the development of the patient's cancer (3)(4)(5).
Our group previously reported the findings from a prospective cohort study where clinically validated molecular profiling of pediatric brain tumors aided in the diagnosis and treatment of patients with a variety of high-and low-grade primary, newlydiagnosed and relapsed brain tumors (6).A subset of the patients undergoing tumor only testing during the first year in which somatic clinical targeted sequencing was performed were found to have possible germline variants (6).We sought to evaluate how these results may have impacted further clinical evaluation and testing in this subgroup of the cohort.

Materials and methods
All patients with tumors of the central nervous system (CNS) undergoing diagnostic evaluation at Seattle Children's Hospital during the study period from November 2015 to November 2016 had their tumors sequenced using the UW-OncoPlex ™ assay (version 5; https://testguide.labmed.uw.edu/view/OPX)?(7).This multiplexed targeted next-generation sequencing panel, designed for the detection of genetic alterations in 262 cancer-related genes, is performed in a College of American Pathologists (CAP)accredited and Clinical Laboratory Improvements Amendments (CLIA)-certified laboratory.Patients with tumor diagnosis by imaging alone were excluded from this study.The patients in this study consented to tumor banking, biology, and return of results (Seattle Children's Hospital Institutional Review Board No. 00000506).Data were extracted from the medical record.
The molecular tumor results were discussed at a monthly molecular brain tumor conference, including pediatric neurooncology, pediatric pathology, molecular pathology, genetic counseling, and neurosurgery along with a research coordinator (6).Each month, all new patients were systematically reviewed in a presentation format consisting of clinical features and magnetic resonance imaging presented by a neuro-oncologist (S.E.S.L.), histologic findings and photomicrographs by a pediatric pathologist (B.L.C.), molecular results by a molecular pathologist (C.M.L.) followed by group discussion including laboratory genetic counseling (S.M.S.).Variants were identified for potential follow-up constitutional testing if they met any of the following criteria: pathogenic or likely pathogenic variants in genes known to be associated with cancer predisposition; or large indels, structural rearrangements or exonic deletions/duplications in genes known to be associated with cancer predisposition; or variants in a patient with clinical features, family history or tumor type suggestive of a constitutional predisposition related to the involved variant (8).The variant allele frequency (VAF) was not used to exclude variants warranting additional follow-up constitutional testing based on other features.Details of the laboratory workflow and methods have previously been published in a Data Supplement (6).
We tracked subsequent clinical evaluation and testing on a subgroup of this cohort who were found to have variants considered to be potentially germline.Individuals in this cohort (or their parents/guardians) were informed of the results and the recommendation for germline testing by their clinical team.Germline confirmatory testing involved the collection a peripheral blood sample or otherwise available non-tumor tissue.Results available in the electronic medical record were tracked on this subgroup of the cohort.

Results
Molecular sequencing was performed on CNS tumors of 88 patients.At the time of biopsy or surgical resection of their tumor, study participants' ages ranged from 1 month to 21 years (median 7 years).Sequencing identified 31 patients (35%) whose tumors harbored variants warranting consideration of germline testing (Figure 1).The details of these results were previously published (6).To date, 19 patients (61% of those with recommended germline follow-up) have been tested (Table 1) (6).Thirteen (68%) of these patients who underwent testing met with an oncology genetic counselor who coordinated testing, while the other patients had the test coordinated by an oncologist at the institution (2, 11%), a provider at another institution (3, 16%), or an established genetic counselor given known germline genetic diagnosis (1, 5%).Germline testing was performed on peripheral blood for 18 patients and on non-affected brain tissue for one patient (patient 7) who had died prior to confirmatory testing.Confirmatory germline testing was performed a median of 13 months following initial tumor testing and ranged from 3 to 57 months.The patients' ages at the time of confirmatory testing ranged from 9 months to 21 years (median 13 years).Of these 19 patients, testing confirmed germline variants for 10 patients (52% of those tested; 11.4% of the study population) as of the last follow-up.The confirmed germline findings were identified in ATM, BAP1, BRIP1, FANCA, MSH2, NF1, NF2, RAD51D, and TP53.One patient was found to have two germline variants (NF1 and mosaic TP53) (Table 1).Two other patients each had two somatic variants warranting testing, with one of the two variants in each patient confirmed to be germline.Of these 11 confirmed germline variants (in 10 patients), three (27%) had a corresponding tumor VAF less than 40%.Of the 10 patients with a confirmed germline variant, 8 patients (80%) received a family referral for genetic counseling and 3 (30%) had at least one relative undergo testing (Table 1).
For some patients, confirmation of a germline variant clarified their diagnosis or was relevant to their treatment plan.Sequencing of one patient's high grade pleomorphic xanthoastrocytoma (PXA) demonstrated a somatic BRAF variant and a TP53 variant, which was confirmed to be present in the germline (patient 39, Table 1).These findings directly impacted their treatment plan, with the use of a BRAF inhibitor, rather than radiation therapy.Another patient's posterior fossa tumor was found to have an ATM variant, confirmed on germline testing (patient 61, Table 1).If this variant had been identified prior to initiating therapy, the concern for radiation sensitivity with a germline ATM variant would have likely impacted management (9).Unfortunately, this result was not available prior to treatment, and the patient ultimately experienced radiation necrosis (10).Another patient with medulloblastoma was found to have a germline variant of uncertain significance (VUS) in SUFU (patient 83, Table 1).This prompted cascade testing in her mother who was found to have the same variant.The patient's maternal grandmother had a basal cell carcinoma diagnosed in her 40s and reported several other clinical features that could be consistent with Gorlin syndrome, including an extra rib, extra teeth, and history of oophorectomy for ruptured     ovarian cyst.For this reason, sequencing of her maternal grandmother's basal cell carcinoma was performed, but the variant was not identified.Twelve (39%) of the patients with somatic variants indicating a possible germline variant have not had confirmatory testing.Eight (26%) of these patients died before genetic counseling and germline testing coordination.Three (10%) of the patients had variants that were reclassified and ultimately did not have germline testing recommended.The remaining patient (3%) had insurance authorization for testing, which has not yet been performed.No patients who were offered confirmatory testing actively declined.

Discussion
Molecular profiling of pediatric brain tumors is typically pursued with the priority of identifying potentially clinically significant variants.However, somatic sequencing may uncover potential germline variants, as demonstrated by this cohort.These variants can be important for informing a patient's own cancer risk, can translate into life-saving surveillance and risk reduction interventions for self and family members, andif identified in timecan sometimes impact therapeutic decisions (3,4).The cohort on which we report includes several patients whose secondary germline findings could have influenced their treatment plan or surveillance care.Additionally, the majority of the patients with confirmed germline variants had at least one family member referred for genetic counseling and consideration of cascade testing.In one case, resultant cascade testing in family members assisted in diagnostic clarification (patient 83, Table 1).This patient, diagnosed with a medulloblastoma, was found to have a germline VUS in SUFU.Given the patient's family history of basal cell carcinoma, one might suspect this variant was pathogenic.not receiving the sequencing results and recommendation until after their child's completion of upfront tumor-directed treatment.While for some families a delay in this discussion may have allowed more "mental space" to consider germline testing, others may have preferred to hear about this information sooner, particularly if the findings of the germline test had the potential to impact their child's treatment plan.For those patients with confirmed germline variants, uptake of cascade testing in family members was variable.This is not unique to our cohort; in general, cascade testing remains suboptimal due to a variety of barriers at the patient-, provider-, and health systems-level (22)(23)(24).
The implications of this study should be considered in the context of a few limitations.First, this study reports on a small cohort of patients with brain tumors at a single institution.A limitation is that we did not consider germline testing of all variants found on somatic sequencing and therefore cannot validate our selection method prior to pursuing testing.Finally, the follow-up details of patient testing were determined by chart review, which may have simplified or misrepresented the extent to which the sequencing impacted the patients' care or discussions about further testing.
Experience from this cohort highlights important considerations as upfront molecular sequencing becomes ever more frequent in childhood cancer care, both on clinical research protocols, such as the Molecular Characterization Initiative funded by the National Cancer Institute, and as part of the standard clinical care of patients (25).Additional research is needed to better delineate the impact of molecular sequencing both on patients' clinical care and outcomes, as well as the psychosocial and financial impact on patients and their families.

FIGURE 1 Flow
FIGURE 1Flow chart of eligible patients during the study period from November 2015 to November 2016.

TABLE 1
Description of variants found in tumors prompting recommendation for germline testing.

TABLE 1 Continued
CNS, central nervous system; VAF, variant allele frequency; dx, diagnosis; n/a, not applicable.*Confirmatory germline testing was performed on peripheral blood samples, unless otherwise noted.