Germline Cancer Predisposition Results From the National Cancer Institute—Children's Oncology Group Pediatric MATCH Trial

Abstract

PURPOSE

Precision oncology trials have generally focused on tumor testing to identify actionable alterations. The National Cancer Institute—Children's Oncology Group Pediatric MATCH trial incorporated return of germline results to assess feasibility of reporting in a cooperative group setting and characterize germline cancer predisposition in patients with refractory cancers.

PATIENTS AND METHODS

Tumor and blood DNA from patients 1-21 years of age with treatment-refractory solid tumors, non-Hodgkin lymphomas, or histiocytic disorders underwent cancer gene panel sequencing. Clinical germline reports returned to 151 study sites included pathogenic/likely pathogenic (P/LP) germline variants found in 38 cancer predisposition genes (CPGs). European Society of Medical Oncology (ESMO) recommendations for germline follow-up of tumor variants in CPGs were assessed.

RESULTS

Both tumor and germline reports were completed for 1,167 patients (87.5% of enrolled). A total of 295 tumor reports (25%) included 361 CPG variants of which 70 variants (19.4%) were found in the germline sample. Three additional germline-only CPG variants resulted in 73 (6.3%) of 1,167 germline reports containing variants across 21 CPGs previously associated with pediatric and/or adult cancers. Among frequently mutated CPGs in tumors, concurrent germline findings ranged from 8/32 NF1 (25.0%) and 25/163 TP53 (15.3%) to zero of 27 ALK and 18 PTEN tumor variants. ESMO guidelines recommended clinical follow-up for 110 (30.5%) of 361 tumor CPG variants which included 40 (57.1%) of 70 germline variants.

CONCLUSION

Coordinated germline and tumor panel testing was feasible and revealed P/LP CPG variants in 6.3% of the Pediatric MATCH cohort. Tumor variant fraction, germline association of CPG with tumor type, and adult-oriented guidelines were not predictive of germline status, emphasizing the need for systematic germline follow-up after tumor genomic testing for pediatric patients.

INTRODUCTION

Genomic research has provided new insight into the diversity of genetic changes occurring in pediatric cancers.^1,2 Much of this work has focused on characterizing somatic (tumor-specific) alterations, primarily in newly diagnosed tumors, to understand tumor biology and identify molecular targets for therapy. In parallel, studies of the germline basis of pediatric cancers have identified an underlying major cancer predisposition disorder (CPD) in 8%-13% of mixed disease cohorts,^3-6 with variable frequency in specific tumor types and subtypes.^7-9 These studies have demonstrated that family history alone is not a sufficient predictor of inherited cancer susceptibility and that current referral criteria for genetic evaluation may miss CPD diagnoses,^3,5 prompting consideration of more widespread germline screening in pediatric cancer care.

CONTEXT

• Key Objective

• To determine the feasibility of performing and reporting matched tumor and germline panel testing in a nationwide precision oncology trial for children with relapsed or refractory solid tumors.

• Knowledge Generated

• Both tumor and germline reports were completed for nearly 90% of enrolled patients. Overall, 25% of tumor reports had a potential cancer predisposition finding of which 20% were germline, leading to a 6.3% germline rate. Current frameworks to predict germline status missed many germline results.

• Relevance

• National treatment–directed precision oncology trial efficiently returned germline results. In the pediatric setting, systematic germline follow-up after tumor genomic testing is recommended.

Clinical tumor molecular profiling most frequently uses cancer gene–targeted next-generation sequencing (NGS) panels. Although combined tumor and germline NGS is performed in some laboratories,¹⁰ tumor-only testing is more common, with a primary focus on alterations that will guide selection of cancer therapies. However, tumor NGS also has the potential to identify variants in cancer predisposition genes (CPGs) that are suggestive of an underlying CPD.¹¹ Recommendations from professional societies^11-13 and incorporation of tumor findings into germline testing criteria, often for analysis of tumors from adult patients, by the National Comprehensive Cancer Network guidelines¹⁴ exemplify growing awareness of this issue. The European Society of Medical Oncology (ESMO) developed specific recommendations for germline follow-up of tumor variants detected in CPGs based on analysis of more than 17,000 tumor-germline pairs in predominantly adult patients with cancer. This guidance incorporates tumor variant allele fraction (VAF), phenotype match of gene with the tumor, actionability, and—for each CPG—the germline fraction of previously observed tumor variants.¹⁵ The applicability of these recommendations for pediatric patients with cancer has not been assessed.

Prospective clinical trials that use NGS to identify biomarker-selected patients for evaluation of cancer therapies have rarely incorporated analysis or return of germline testing results.^16,17 Follow-up for such tumor findings requires recognition of the variant of concern, referral to a genetic specialist, and consideration of subsequent germline testing including insurance coverage. In contrast, the National Cancer Institute—Children's Oncology Group (NCI-COG) Pediatric MATCH screening trial for children, adolescents, and young adults with refractory cancers included both tumor and germline testing, in parallel, using the same tumor-focused panel with reporting for all patients enrolled between July 2017 and December 2021. Details of the methods used in Pediatric MATCH for tumor analysis and data regarding treatment arm matching and tumor NGS results have been reported.¹⁸ We describe here the germline results including the proportion of tumor variants in CPGs found in the germline and assess the ESMO recommendations for germline referral in this nationwide precision oncology trial.

PATIENTS AND METHODS

Patient Selection and Clinical Data Collection

Patients age 1-21 years with recurrent or refractory solid tumors, non-Hodgkin lymphomas, or histiocytic disorders treated at US-based COG sites were eligible for the Pediatric MATCH screening trial (ClinicalTrials.gov identifier: NCT03155620; see Protocol). Demographic and clinical data were reported by study sites at enrollment, including whether the patient was known to have a clinical diagnosis of a genetic disease. Molecular status was not requested. The trial was approved by the NCI Central Institutional Review Board. All patients or their parents gave signed informed consent and assent was obtained as appropriate.

Sample Submission and Processing

Formalin-fixed, paraffin-embedded tumor and blood samples were submitted to the COG Biospecimen Bank at Nationwide Children's Hospital. Nucleic acids (tumor DNA and RNA; blood DNA) were isolated and sent to one of three Pediatric MATCH laboratories for NGS.¹⁸ Blood DNA was only sent for testing in study patients with an adequate tumor sample to attempt NGS and was not completed for patients who went off study before tumor testing had been completed.

Germline and Tumor NGS

Germline and tumor DNA were sequenced using an Oncomine AmpliSeq cancer gene panel (ThermoFisher Scientific, Waltham, MA) designed for the detection of tumor alterations with the potential to guide selection of molecularly targeted therapies, including single nucleotide variants (SNVs), insertions and deletions (indels), and amplifications in 161 cancer genes.^18,19 A VAF of 0.05 was required for reporting SNVs and 0.10 for indels. The median read coverage was 1,788 (range, 299-4,316) for sites of reported tumor variants and 1,990 (range, 642-5,538) for sites of reported germline variants across the three testing laboratories. A software application (Pediatric MATCHBox) facilitated the processing, analysis, and review of tumor and germline sequence data. The panel was not validated for reporting of copy number losses or consensus splice site mutations. In a small number of cases, a germline splice site variant detected during manual review was reported.

Tumor and Germline Variant Reporting

Before study initiation, the Pediatric MATCH Germline Reporting Committee (Appendix Table A1) selected 38 study-defined CPGs on the DNA panel that convey adult or pediatric cancer susceptibility for germline reporting (Appendix Table A2). For each sample, pathogenicity of germline variants was defined following American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP) guidelines.²⁰ Before reporting, results were reviewed in a weekly multidisciplinary meeting that included members of the Pediatric MATCH germline committee and at least one clinical genetics laboratory director. Tumor and germline NGS reports were separately returned to study sites. Germline results were not used for any treatment decisions in the trial. Language in germline reports highlighted limitations of the panel for germline testing, and patients where a CPD is suspected should have a formal genetics evaluation.

Follow-Up of Positive Germline Reports

Positive germline NGS reports were defined as those containing a pathogenic or likely pathogenic (P/LP) variant in a CPG. For each positive report, a follow-up e-mail was sent from the study genetic counselor to the patient's primary oncologist, COG principal investigator, and clinical research associate at the study site. The e-mail provided contact information for the MATCH germline team (genetic counselor, medical geneticist, and oncologist with expertise in cancer surveillance) for questions. The e-mail also asked whether the patient had a previous clinical diagnosis of the CPD identified on the report and whether that diagnosis had been previously confirmed by genetic testing.

Variant Analysis

Reported CPG variants were annotated by variant type (eg, SNV or indel), tumor and germline VAF, whether the variant is predicted to be truncating, ClinVar database classification level as of August 2022,²¹ and whether germline analysis would have been recommended per ESMO guidelines.¹⁵ GeneReviews were used to determine whether P/LP variants in the CPG are associated with predisposition to the patient's tumor type (ie, on-tumor).²² A more recent association between IDH1 germline variants (Ollier disease) and glioma²³ was included. Tumor loss of heterozygosity (LOH) was defined as an increase in tumor VAF of 0.20 compared with the germline report.²⁴

Statistical Analysis

Chi-square tests (or Fisher's exact tests as appropriate) were used to compare germline variant rates between groups. The Wilcoxon rank-sum test or two-sample t test was used to compare continuous measures between groups. Statistical significance was set at P value <.05. Data were analyzed using SAS version 9.4 and R version 4.0.2.

RESULTS

Description of Tumor/Germline Cohort

A total of 1,334 patients from 151 study sites across the United States were enrolled in the Pediatric MATCH screening protocol between July 2017 and December 2021 (see study flow diagram in Fig 1). A blood sample was received for 1,285 patients (96.3% of those enrolled) including samples from three patients with a history of allogeneic bone marrow transplantation that were not analyzed further. Germline NGS was initiated for 1,202 patients (90.1% of enrolled) with adequate tumor for sequencing, but 19 patients came off study before testing was completed. Germline results were reported for 1,183 patients (88.7% of enrolled) with 16 patients only having germline results as tumor testing was not successful. The focus of this analysis is the 1,167 patients (87.5% of enrolled) with both tumor and germline NGS results being reported, defined as the Pediatric MATCH Tumor/Germline (T/G) Cohort.

FIG 1.

Flow diagram for germline reporting conducted in the Pediatric MATCH Trial. The 1,167 patients with both tumor and germline results reported comprise the Tumor/Germline Cohort. NGS, next-generation sequencing.

Patients in the T/G Cohort ranged in age from 1 to 21 years (median 13 years) with a slight predominance of males (55.8%; Table 1). Most patients were described as White (67.8%) and of non-Hispanic or Latino ethnicity (72.8%). No significant demographic differences were reported in the T/G Cohort compared with patients not in the T/G Cohort (Table 1).^18,25

TABLE 1.

Pediatric MATCH Screening Trial Patient Characteristics

Characteristic	Patients	T/G Cohort	Pts with CPG Tumor Variant	Pts With CPG Germline Variant	T/G Cohort With Germline Variant, %	P Value for T/G Cohort (n = 1,167) v Non-T/G Cohort (n = 167)	P Value for T/G Cohort With Germline Variant (n = 73) v Without (n = 1,094)
Total number of patients	1,334	1,167	295	73	6.3
Sex, No. (%)						.679	.044
Female	587 (44)	516 (44)	131 (44)	24 (33)	4.7
Male	747 (56)	651 (56)	164 (56)	49 (67)	7.5
Age, years, median (range)	13 (1-21)	13 (1-21)	13 (1-21)	14 (2-21)		.179	.400
Age categories, years, No. (%)
<5	160 (12)	127 (11)	27 (9)	8 (11)	6.3
≥5 and <15	633 (47)	556 (48)	142 (48)	34 (46)	6.1
≥ 15	541 (41)	484 (42)	126 (43)	31 (42)	6.4
Race, No. (%)						.688	.312
White	895 (67)	791 (68)	186 (63)	51 (70)	6.4
Black or African American	198 (15)	168 (14)	50 (17)	9 (12)	5.3
Asian	52 (4)	44 (4)	13 (4)	2 (3)	4.5
Native Hawaiian or Pacific Islander	11 (1)	10 (1)	3 (1)	2 (3)	20.0
American Indian or Alaska Native	5 (<1)	4 (<1)	2 (1)	1 (1)	25.0
Multiple races	21 (2)	18 (2)	5 (2)	1 (1)	5.6
Not reported	152 (11)	132 (11)	36 (12)	7 (10)	5.3
Ethnicity, No. (%)						.168	.946
Hispanic or Latino	304 (23)	270 (23)	68 (23)	17 (23)	6.3
Not Hispanic or Latino	970 (73)	849 (73)	216 (73)	54 (74)	6.4
Not reported	60 (4)	48 (4)	11 (4)	2 (3)	4.2
Tumor category, No. (%)						.046	.442
Non-CNS solid tumor	969 (73)	861 (74)	211 (72)	50 (68)	5.8
CNS solid tumor	319 (24)	267 (23)	79 (27)	21 (29)	7.9
Lymphoma/Histiocytosis	46 (3)	39 (3)	5 (2)	2 (3)	5.1
Known cancer predisposition disorder reported, No. (%)						.347	<.001
Yes	54 (4)	45 (4)	25 (8)	21 (29)	46.7
No	1,280 (96)	1,122 (96)	270 (92)	52 (71)	4.6

NOTE. The table reflects data collected at screening enrollment including age at study entry.

Abbreviations: CPG, cancer predisposition genes; Pts, patients; T/G, tumor/germline.

Solid tumors occurring outside the CNS comprised 73.8% of diagnoses, followed by CNS tumors (22.9%) and lymphomas and histiocytoses (3.3%; Table 2). More than half of the patients had sarcomas (53.3%). In the T/G Cohort, a history of known CPD was reported at study entry for 45 patients (3.9%) comprising 14 different CPDs (Appendix Table A3) with only neurofibromatosis type 1 (n = 17, 1.5%), Li-Fraumeni syndrome (n = 14, 1.2%), and Lynch syndrome (n = 3, 0.3%) reported in more than one patient.

TABLE 2.

Tumor Diagnoses of NCI-COG Pediatric MATCH Screening Trial Patients

Diagnosis	T/G Cohort, No.	Pts With CPG Tumor Variant, No. (%)	Pts With CPG Germline Variant, No. (%)
Non-CNS solid tumors (total)	861	211 (25)	50 (6)
Osteosarcoma	224	57 (25)	19 (8)
Rhabdomyosarcoma	155	40 (26)	8 (5)
Ewing sarcoma	132	27 (20)	6 (5)
Neuroblastoma	68	25 (37)	2 (3)
Wilms tumor	62	18 (29)	2 (3)
Sarcomas - other	53	12 (23)	3 (6)
Undifferentiated sarcoma/NOS	25	4 (16)	1 (4)
Carcinomas	30	10 (33)	3 (10)
Synovial sarcoma	20	1 (5)	1 (5)
Hepatoblastoma	13	0 (0)	0 (0)
Germ cell tumors	18	3 (17)	0 (0)
Malignant rhabdoid tumor	12	4 (33)	0 (0)
Desmoplastic small round cell tumor	13	2 (15)	1 (8)
Hepatocellular carcinoma	12	1 (8)	1 (8)
Other non-CNS tumor	24	7 (29)	3 (13)
CNS solid tumors (total)	267	79 (30)	21 (8)
High-grade glioma	78	38 (49)	10 (13)
Low-grade glioma	58	11 (19)	6 (10)
Ependymoma	40	3 (8)	0 (0)
Medulloblastoma	27	9 (33)	2 (7)
Atypical teratoid/Rhabdoid tumor	20	10 (50)	2 (10)
Mixed glial or glioneuronal tumor	13	2 (15)	0 (0)
Primitive neuroectodermal tumor	12	2 (17)	0 (0)
Other CNS tumor	19	4 (21)	1 (5)
Lymphomas/Histiocytoses (total)	39	5 (13)	2 (5)
Lymphomas	24	5 (21)	2 (8)
Histiocytic disorders	15	0 (0)	0 (0)
All diagnoses	1,167	295 (25.3)	73 (6.3)

Abbreviations: CPG, cancer predisposition genes; NCI-COG, National Cancer Institute—Children's Oncology Group; NOS, not otherwise specified; Pts, patients; T/G, Tumor/Germline.

Frequency of Tumor and Germline CPG Variants

Variants in genes designated as CPGs were reported in 295 (25.3%) of 1,167 tumor NGS reports, with 53 reports (4.5%) having more than one CPG variant, for a total of 361 CPG variants detected in 31 genes (Fig 2A, Table 3, Data Supplement, Table S1). Of these 361 CPG tumor variants, 70 (19.4%) were also found in the blood sample (referred to as germline) and classified by the reporting lab as P/LP per ACMG/AMP criteria in 70 different patients. In three additional cases, a germline variant (one each in BRCA1, CHEK2, and NF1) was not reported in the matched tumor sample. Altogether, 73 (6.3%) of the 1,167 patients in the T/G Cohort had a positive germline report (Fig 2A and Table 3, Data Supplement, Table S1). Analysis of the G/T cohort revealed no significant difference in the rate of germline variants detected by patient age, race, or ethnicity (Table 1). Of tumor types with >50 samples, high- and low-grade glioma were the histologies with the highest germline rate, 12.8% and 10.3%, respectively (Table 2).

FIG 2.

Comparison of cancer predisposition gene variants in tumor and germline analysis. (A) CPGs with variants reported. Graph demonstrates the tumor and germline sequencing results. Tumor sequencing results (n = 361 variant). Germline sequencing results (n = 73 variants). For further annotated data on the CPG variants reported, refer to the Data Supplement (Table S1). (B) Tumor VAFs for tumor CPG variants. The VAF for tumor CPG variants that were also detected in the germline sample (GL+ patients) is shown on the left. The VAF for tumor CPG variants that were not detected in the germline sample (GL– patients) is shown on the right. The mean of each population is noted by a horizontal line and significant difference noted by asterisks. CPG, cancer predisposition genes; GL, germline; VAF, variant allele fraction.

TABLE 3.

Characteristics of Tumor and Germline Cancer Predisposition Gene Variants Detected in Pediatric MATCH Patients

Gene	Tumor Variants Detected, No.	Tumor Variants Met ESMO Criteria for Germline Analysis, No. (%)	Germline Variants Detected, No. (%)	Germline Variants with Tumor LOH, No. (%)	Germline Variants Recommended by ESMO Criteria, No. (%)
TP53	163	51 (31)	25 (15)	20 (80)	15 (60)
NF1	32	18 (56)	8 (25)	4 (50)	7 (88)
ALK	27	0 (0)a	0 (0)	NA	NA
PTEN	18	0 (0)a,b	0 (0)	NA	NA
RB1	16	15 (94)	3 (19)	3 (100)	3 (100)
SMARCB1	14	0(0)a,b	1 (7)	1 (100)	0 (0)
ATM	9	0 (0)a,b	5 (56)	1 (20)	0 (0)
BRCA2	9	7 (78)	5 (56)	0 (0)	5 (100)
PTPN11	8	0 (0)a,b	0 (0)	NA	NA
TSC2	7	4 (57)	1 (14)	0 (0)	1 (100)
PTCH1	5	0 (0)a,b	1 (20)	1 (100)	0 (0)
SMARCA4	5	0 (0)a,b	1 (20)	0 (0)	0 (0)
POLE	5	0 (0)	0 (0)	NA	NA
CHEK2	4	0 (0)a,b	4 (100)	0 (0)	0 (0)
IDH1	4	0 (0)a	1 (25)	0 (0)	0 (0)
MSH2	4	4 (100)	1 (25)	0 (0)	1 (100)
NF2	4	0 (0)a	0 (0)	NA	NA
TSC1	4	0 (0)a	0 (0)	NA	NA
NBN	3	0 (0)a	2 (67)	0 (0)	0 (0)
MSH6	3	3 (100)	1 (33)	1 (100)	1 (100)
BRCA1	2	2 (100)	2 (100)	0 (0)	2 (100)
FANCI	2	0 (0)a	2 (100)	0 (0)	0 (0)
MLH1	2	2 (100)	2 (100)	2 (100)	2 (100)
PALB2	2	2 (100)	2 (100)	0 (0)	2 (100)
SLX4	2	0 (0)a	1 (50)	0 (0)	0 (0)
PMS2	2	1 (50)	0 (0)	NA	NA
FANCD2	1	0 (0)a	1 (100)	0 (0)	0 (0)
RET	1	1 (100)	1 (100)	0 (0)	1 (100)
EGFR	1	0 (0)a	0 (0)	NA	NA
RAD51C	1	0 (0)	0 (0)	NA	NA
SMAD4	1	0 (0)a	0 (0)	NA	NA
Total	361	110 (30)	70 (19)c	33 (47)	40 (57)

Abbreviations: ESMO, European Society for Medical Oncology; LOH, loss of heterozygosity; NA, not applicable.

^a Gene not included on original ESMO guideline list of genes for potential germline follow-up.

^b Gene added to updated ESMO guideline list of genes for potential germline follow-up.

^c Variants in three genes identified in germline only (BRCA1, CHEK2, and NF1) are excluded from table.

Description of Tumor and Germline CPG Variants

The mean tumor VAF was higher for CPG variants that were also detected in the germline compared with those that were tumor only (0.64 v 0.51, P < .001) but with broad overlap of the distribution (Fig 2B). No tumor variant with a VAF of <0.3 (SNV) or <0.2 (indel) was detected in the germline (0/60; Data Supplement, Table S1).

The 73 germline variants were reported in 21 different CPGs (Table 3). Seventy-five percent of germline variants (n = 53) were from seven genes (TP53, NF1, BRCA2, ATM, CHEK2, BRCA1, and RB1) with variants in TP53 (n = 25, 34.2% of germline results) most common (Fig 2A). The fraction of reported tumor CPG variants that were germline in origin varied markedly by gene (Table 3). For example, no tumor variants in ALK (n = 27) or PTEN (n = 18) were detected in the germline, compared with results for TP53 25/163 (15.3%), NF1 8/32 (25.0%), and RB1 3/16 (18.7%). Tumor LOH was observed for 33 (47.1%) of 70 CPG variants detected in both tumor and germline (Table 3, Data Supplement, Table S1) including 20 (80.0%) of 25 in TP53. Only two tumors demonstrated second somatic variants in NF1 and TP53 each in low-grade astrocytomas.

Forty-two (57.5%) of the 73 germline CPG variants were in genes reported to be associated with predisposition to the patient's tumor type (including 30 tumors with LOH), 21 (28.8%) were in genes associated primarily with adult-onset cancers, six (8.2%) were single variants associated with recessive disease, and four (5.5%) were in genes associated with pediatric tumor types other than the patient's diagnosis (eg, TP53 in atypical teratoid rhabdoid tumor; Data Supplement, Table S1). Of note, the 17 tumor variants detected in breast cancer susceptibility genes (BRCA1, BRCA2, CHEK2 and PALB2) had a unique pattern with a high rate of germline findings (n = 13, 76.5%) but no evidence of tumor LOH as reported recently.²⁶

Assessing ESMO Recommendations for Germline Follow-Up of Tumor CPG Variants

ESMO guidelines use tumor and variant characteristics to predict need for germline follow-up.¹⁵ Applying these criteria, 110 (30.4%) of the 361 tumor CPG variants would merit germline follow-up, of which 40 (36.3%) were confirmed to be germline variants (Table 3, Data Supplement, Table S2). However, the guidelines did not recommend follow-up for 30 (42.9%) of 70 tumor CPG variants found to be germline, for example, nine variants in genes not considered sufficiently penetrant (ATM and CHEK2), nine TP53 variants in pediatric patients with brain tumor, and six single variants in genes for recessive disorders (Data Supplement, Table S2).

Previous Knowledge of Germline CPG Variants and Patient CPD

Of the germline positive cases, previous knowledge of genetic conditions was confirmed for 48 (65.8%) of 73 patients derived from site-submitted clinical data on study survey, pathology report, molecular pathology report, and/or oncologist responses to positive germline result notification e-mails. Documentation of the germline variant was reported in 29 (39.7%) and a CPD was clinically suspected but not molecularly proven in seven additional patients (9.5%). The 12 CPG variants (25.0%), and the related CPD confirmed to be a new finding via oncologist e-mail included four variants in TP53 (osteosarcoma [OS] × 2, rhabdomyosarcoma, glioblastoma), two variants in BRCA2 (OS and high-grade B-cell lymphoma), three variants in CHEK2 (Wilms tumor, Ewing sarcoma, glioblastoma), one variant in PALB2 (Ewing sarcoma), and heterozygous variants in the recessive cancer genes NBN and FANCI (both in OS).

Conversely, of the 45 patients with a clinical diagnosis reported at study entry, only 21 (46.6%; Appendix Table A3) had positive MATCH germline testing; in six patients, the causative CPG was not on the panel (eg, APC) and 18 had a negative MATCH result for the gene in question (eg, 10 patients with neurofibromatosis type 1 and five with Li-Fraumeni Syndrome).

DISCUSSION

Pediatric MATCH has demonstrated the feasibility of incorporating clinical germline testing into a collaborative nationwide precision oncology trial, with results reported for nearly 90% of patients enrolled on study. Germline reporting required additional resources and infrastructure including costs of germline sample processing and sequencing, a separate team of laboratory diagnosticians with experience in hereditary cancer testing to classify and report germline CPG variants, a second pipeline for review and delivery of germline reports, and clinical genomics support team including a genetic counselor and medical geneticists to address clinical questions or identify local genetic resources. However, only five oncologists who responded to the study e-mail about positive germline results requested additional assistance, suggesting that extensive genetics support may not be perceived as necessary in the context of such trials.

The 6.3% detection rate for germline CPG variants does not provide evidence for a greater frequency of cancer predisposition diagnoses in the Pediatric MATCH cohort of children, adolescent, and young adults with advanced and refractory tumors compared with pediatric patients with newly diagnosed cancers, for whom rates of 8%-13% have been reported.^3-6

However, the germline variant detection rate from Pediatric MATCH is clearly an underestimate given that the cancer gene panel and analytic pipeline were designed to detect tumor variants that could guide selection of targeted therapies, and not all causative CPGs were included on the panel and key CPG alteration types (including copy number loss and splice site mutations) were not evaluated. In the context of negative tumor findings, patients with substantial concern for a CPD, such as those identified by the McGill Interactive Pediatric OncoGenetic Guidelines, should still be referred for genetics evaluation given the limitations in treatment-focused testing platforms.^27,28

One approach to reduce the number of tumor variants resulting in subsequent germline testing is to apply additional clinical and molecular parameters, such as those developed by ESMO. These criteria would have reduced germline follow-up for 69.6% of the tumor variants in CPGs in our study cohort. However, 42.9% of germline-positive cases did not meet the original ESMO recommendations including multiple missed germline TP53 variants. Pediatric MATCH data have some similarities to results of adult studies,^29,30 for example, most MATCH tumor CPG variants in breast cancer susceptibility genes were of germline origin. However, as anticipated in the original ESMO guidelines and reflected in updated guidelines following this analysis,³¹ special considerations are needed for pediatric patients. They now include pediatric-focused genes (noted in Table 3) and change from excluding TP53 variants identified in brain tumors (of which 33% were germline) to say these may be excluded for germline testing, a term which may lead to inconsistent referral. Overall, in Pediatric MATCH, 15% of all tumor TP53 variants were germline in origin, consistent with other recent studies³² suggesting that the germline fraction of TP53 variants in pediatric cancers is markedly higher than the 2%-3% reported in adult populations.^29,30 More than 40% of MATCH germline CPG variants were in genes not currently defined as associated with the patient's tumor type, including single variants in rare recessive disorders and genes with established roles for adult-onset cancer predisposition without evidence of tumor LOH and whose relevance to pediatric cancer is under investigation.^33-35 This is consistent with previous studies where about half of the variants in CPGs in unselected pediatric patients with cancer occur in patients who do not meet germline testing criteria.^3,5

A primary goal of the parallel tumor/germline testing approach used in Pediatric MATCH was to decrease the burden of follow-up germline testing for the 25% of patients with a tumor CPG variant detected. This may be particularly important for patients with advanced disease enrolled in early-phase clinical trials for whom referral for a separate genetic evaluation may not be prioritized or possible before death by disease. Although universal follow-up of tumor CPG variants will increase predisposition diagnoses, more than 75% of tumor CPG variants reported (20% of trial participants) were not present in the germline sample. Thus, germline testing as part of the trial reduces the need for subsequent referral for those participants without additional concern for a CPD diagnosis.

In conclusion, coordinated germline and tumor panel testing was feasible and revealed P/LP CPG variants in 6.3% of the Pediatric MATCH cohort. Incorporation into future precision oncology study designs of rapid stepwise germline testing for all subjects with detection of tumor CPG variants could substantially reduce germline testing costs compared with MATCH while still efficiently providing germline results to patients and parents.

APPENDIX

TABLE A1.

Committee Membership for the NCI-COG Pediatric Match Germline Reporting Committee

Committee Member	Institution
Steven Joffe, MD, MPH (co-chair)	Children's Hospital of Philadelphia, University of Pennsylvania
Sharon E. Plon, MD, PhD (co-chair)	Texas Children's Hospital, Baylor College of Medicine
Jacquelyn Biegel, PhD	Children's Hospital of Los Angeles
Sarah Scollon, MS, CGC	Texas Children's Hospital, Baylor College of Medicine
Lisa Diller, MD	Boston Children's Hospital, Dana Farber Cancer Institute
Conrad Fernandez, MD	IWK Health Center, Dalhousie University
Shasikant Kulkarni, PhD, MBA	Baylor College of Medicine
David Malkin, MD	Toronto Hospital for Sick Children, University of Toronto
George Miles, MD, PhD	Baylor College of Medicine
Jennifer Oberg, EdD	Columbia University
D. Will Parsons	Texas Children's Hospital, Baylor College of Medicine
Mary Relling, PharmD	St Jude Children's Research Hospital
Josh Schiffman, MD	Huntsman Cancer Institute, University of Utah
Lisa Schwartz, PhD	Children's Hospital of Philadelphia, University of Pennsylvania
Douglas Stewart, MD, PhD	National Cancer Institute
James V. Tricoli, PhD	National Cancer Institute

Abbreviation: NCI-COG, National Cancer Institute—Children's Oncology Group.

TABLE A2.

Pediatric MATCH Cancer Panel Genes Analyzed for Germline Cancer Susceptibility

Hotspot Genes (n = 12)	Tumor Suppressor Genes (n = 26)
ALK	ATM
CBL	BRCA1
CDK4	BRCA2
CHEK2	FANCA
EGFR	FANCD2
GATA2	FANCI
IDH1	MLH1
IDH2	MSH2
PTPN11	MSH6
RET	NBN
SMAD4	NF1
TERT	NF2
	PALB2
	PMS2
	POLE1
	PTCH1
	PTEN
	RAD51C
	RB1
	SLX4
	SMARCA4
	SMARCB1
	STK11
	TP53
	TSC1
	TSC2

TABLE A3.

Reported Genetic Diagnoses of Pediatric MATCH Screening Trial Patients

CPD	All Enrolled Patients (n = 1,334)	Tumor/Germline Cohort Patients (n = 1,167)
Neurofibromatosis type 1	18	17
Li-Fraumeni syndrome	18	14
Lynch syndrome	3	3
Beckwith-Weidemann syndrome	1	0
Dubowitz syndrome	1	0
Familial adenomatous polyposis	1	1
FGFR1 germline variant—glioneuronal tumor	1	0
Gorlin syndrome	1	1
Multiple endocrine neoplasia type 2B	1	1
Noonan syndrome	1	1
Noonan-like syndrome	1	1
Ollier disease	1	1
Familial retinoblastoma	1	1
Rhabdoid predisposition syndrome	1	0
Rothmund-Thomson syndrome	1	1
SDHB germline variant	1	1
Simpson-Golabi-Behmel syndrome	1	1
Trisomy 21	1	1
Total, No. (%)	54 (4.0)	45 (3.9)

Abbreviation: CPD, Cancer Predisposition Disorder.

AUTHOR CONTRIBUTIONS

Conception and design: Sarah Scollon, Sharon E. Plon, Steven Joffe, Jaclyn A. Biegel, David R. Patton, Brent Coffey, Gregory J. Tsongalis, Mark J. Routbort, Nilsa Ramirez, Todd A. Alonzo, Stacey L. Berg, Elizabeth Fox, Brenda Weigel, Jeffrey S. Abrams, Margaret Mooney, Naoko Takebe, James V. Tricoli, Katherine A. Janeway, Nita L. Seibel, D. Williams Parsons

Financial support: David R. Patton, Gregory J. Tsongalis

Administrative support: David R. Patton, Gregory J. Tsongalis

Provision of study materials or patients: David R. Patton, Brenda Weigel, Katherine A. Janeway

Collection and assembly of data: Sarah Scollon, Sharon E. Plon, Jaclyn A. Biegel, Shashikant Kulkarni, David R. Patton, Brent Coffey, Cynthia L. Winter, Mark J. Routbort, Nilsa Ramirez, Lauren Saguilig, Brenda Weigel, Naoko Takebe, James V. Tricoli, Katherine A. Janeway, D. Williams Parsons

Data analysis and interpretation: Sarah Scollon, Sharon E. Plon, Steven Joffe, Jaclyn A. Biegel, Shashikant Kulkarni, George Miles, David R. Patton, Brent Coffey, Lauren Saguilig, Jin Piao, Todd A. Alonzo, Stacey L. Berg, Douglas S. Hawkins, Naoko Takebe, James V. Tricoli, Katherine A. Janeway, D. Williams Parsons

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).