Pediatric oncology patients with germline pathogenic variants in adult-onset cancer predisposition genes

Abstract

Background:

Cancer predisposition syndromes caused by germline pathogenic variants (GPV) in adult-onset cancer predisposition genes (aoCPG) are those for which there is low risk for cancer in children, with genetic testing and screening for these conditions typically deferred until adulthood. GPV in aoCPG have been identified in pediatric oncology patients, but in these cases the potential contribution of the aoCPG to cancer development is often unknown. We investigated the role GPV in aoCPG may play in childhood cancer development.

Procedure:

Results of paired tumor-germline sequencing from pediatric oncology patients enrolled 5/2012–10/2023 were analyzed for frequency of GPV in aoCPG. Germline testing included analysis of up to 182 cancer predisposition genes. Tumor loss-of-heterozygosity, presence of second somatic pathogenic variant, immunohistochemical stain for protein expression, and/or tumor mutation burden were used to determine possible causation.

Results:

Of 954 participants, 42 (4.4%) had GPV in aoCPG. Six of these 42 participants (14.3%) had tumor findings indicating their GPV in an aoCPG likely contributed to cancer development: three patients with Lynch syndrome (two anaplastic astrocytomas, one giant cell glioblastoma) and one each with GPV in ATM (craniopharyngioma and diffuse high-grade glioma), BRIP1 (atypical teratoid rhabdoid tumor), and CHEK2 (mixed germ cell tumor of pineal gland).

Conclusions:

These findings contribute to the literature suggesting that, rarely, GPV in aoCPG may contribute to cancer diagnoses in children, raising the question of how tumors in these cases may present differently in children than adults. Increased knowledge about potential childhood cancer risks related to what have historically been considered aoCPG could modify predictive genetic testing recommendations for children and enhance existing cancer screening protocols.

Introduction

Approximately 10–25% of pediatric oncology patients have an underlying hereditary risk for cancer^1–3. There are several well-described hereditary cancer predisposition syndromes that cause childhood-onset cancers. However, increased use of paired tumor-germline sequencing in pediatric oncology settings has resulted in identification of germline pathogenic/likely pathogenic variants (GPV) in adult-onset cancer predisposition genes (aoCPG) in children with cancer^4–6. Historically, there has been thought to be no expected increased risk for cancer in children with aoCPG, although emerging data suggests there is low penetrance of tumors in childhood. While there are reports of pediatric malignancies in families with GPV in aoCPG, little is known about the potential contribution of the aoCPG to these diagnoses⁶. Professional organizations typically recommend deferring genetic testing for these aoCPG until adulthood and medical management for these conditions begins at or after the age of 18^7–10. The rationale for deferring testing includes preserving the child’s right to make an autonomous decision, their right to an open future (given concerns for discrimination and/or stigma related to a genetic diagnosis), fears about psychological harms, concerns about communication of results to children, and lack of actionability until adulthood^7,8. While increasing use of comprehensive genetic testing such as whole exome/genome sequencing may incidentally identify aoCPG in pediatric patients, guidelines continue to suggest limiting genetic testing of minors for aoCPG outside of these scenarios^6,11,12. However, this topic is more fluid in clinical practice, with families expressing interest in genetic testing for their minor children and some providers reporting ordering genetic testing for aoCPG for individuals who are under the age of 18^13–16.

While many large studies have reported findings of GPV in aoCPG in pediatric oncology populations, most have been unable to assess causality³. Prior work in large pediatric cancer cohorts has identified enrichment of GPV in several genes linked to aoCPG including BRCA1, BRCA2, PALB2, and Lynch syndrome-associated genes (MLH1, MSH2/EPCAM, MSH6, PMS2), but lacked molecular tumor data such as immunohistochemical (IHC) staining, tumor loss-of-heterozygosity (LOH), tumor mutation burden, and/or presence of a second somatic pathogenic variant, to aid in the interpretation of causality^3,17–19. A few case series have reported tumor findings associated with aoCPG, mostly pediatric-onset malignancies including colorectal cancer, brain tumors, and osteosarcoma with associated IHC/tumor findings suggestive of Lynch syndrome^2,6,20–23. A large study reported the majority (81%) of pediatric patients with somatic mutations in an aoCPG were confirmed to have this same finding present in the germline, with 10% of these having tumor findings consistent with an association with the GPV²⁴. Here, we present findings of GPV in aoCPG from a pediatric oncology cohort with both tumor and germline data, providing the opportunity to assess the possible role of GPV in the development of pediatric cancer.

Methods

Study participants

All participants were part of the Personalized Medicine based on Molecular Profiling of Pediatric and Young Adult Patients with Cancer (PEDS-ONCOSEQ) study, which was approved by the Michigan Medicine Institutional Review Board (HUM00056496). Patients up to age 25 seen in our Pediatric Hematology/Oncology Clinic with a suspected or diagnosed cancer or rare tumor were eligible for participation, with a focus on those with relapsed/refractory cancers. Individuals with rare, non-cancerous conditions, such as vascular malformations or inflammatory myofibroblastic tumors were also eligible for enrollment. Enrollment began in May 2012, and we screened all patients enrolled through October 2023 for this study.

All subjects provided informed consent for participation. A patient/guardian signed informed consent on behalf of participants under age 18, with children aged 10 and over asked to provide assent for their participation. Participants aged 18 and older signed their own consent forms. Tumor and germline samples were collected for all patients. Consent included the option for patients to opt-out of receiving germline findings (not disclosed on report or to clinical team), with the exception of mandatory disclosure of germline findings with associated treatment implications for the participant’s current diagnosis. Only GPV were disclosed on sequencing reports; germline variants of uncertain significance were not reported. Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Michigan^25,26. REDCap (Research Electronic Data Capture) is a secure, web-based software platform designed to support data capture for research studies, providing 1) an intuitive interface for validated data capture; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for data integration and interoperability with external sources. A subset of the sequenced cohort has been previously described, with solid tumors making up the majority of the diagnoses, and brain tumors, neuroblastoma, and sarcomas representing most of these malignancies¹.

Molecular analysis

The use of integrative clinical exome (tumor and germline DNA) and transcriptome (tumor RNA) sequencing in our population has been previously described¹. Briefly, paired-end whole exome libraries from tumor samples were matched with normal DNA and transcriptome libraries. Aligned exome and transcriptome sequences were analyzed to identify putative somatic mutations, insertions and deletions (indels), copy-number alterations, gene fusions, and gene expression as described previously¹. Germline analysis included 161–182 genes associated with both pediatric- and adult-onset hereditary cancer risks (Supplementary Table 1). The germline multigene panel was updated throughout the study as new information about gene-disease relationships emerged. Pathogenicity of germline variants were determined through a review of the published literature, public databases including but not limited to ClinVar, the Human Genome Mutation Database, the Leiden Open Variation Database, variant specific databases, and the American College of Medical Genetics and Genomics guidelines for interpretation of sequence variants. Variants are not re-reviewed for updates to classification. Genes were defined as aoCPG if standard of care medical management (i.e. guidelines through National Comprehensive Cancer Network and similar organizations) for the associated increased cancer risks suggests that screening and/or prevention begins at age 18 or older.

Cohort identification

The sub-population for this study consisted of participants with a GPV identified in an aoCPG. A list of the 182 genes interrogated for germline pathogenic variants and designation of which were deemed adult-onset is available in Supplementary Table 1. Genes in which no GPV were identified in our cohort were not examined to determine if cancer risk was adult- versus childhood-onset. Tumor data were analyzed for participants meeting this criterion to determine if there were tumor features (IHC staining, tumor LOH, tumor mutation burden, and/or presence of a second somatic pathogenic variant) suggestive of an association between the GPV and cancer development.

Rates of pathogenic variants

The prevalence of GPV identified in aoCPG was compared to the general population incidence to assess enrichment for any conditions in our cohort. When general population prevalence estimates varied in the literature or were reported as a range, the commonly accepted upper estimate was used so that any resulting bias would be towards the null hypothesis. Significance was calculated using Fisher’s exact test.

Results

Germline findings

Nine hundred and fifty-four participants were enrolled between 5/2012–10/2023. GPV in aoCPG were identified in 42/954 (4.4%) individuals (Supplementary Table 2). Some patients had multiple GPV for a total of 45 GPV. GPV were identified in the CHEK2 (17, 37.8% of pathogenic variants), BRCA2 (5, 11.1%), HOXB13 (4, 8.9%), MITF (4, 8.9%), ATM (3, 6.7%), RAD50 (2, 4.4%), MSH6 (2, 4.4%), BRIP1 (2, 4.4%), BRCA1 (2, 4.4%), MLH1 (1, 2.2%), BARD1 (1, 2.2%), PALB2 (1, 2.2%), and CTNNA1 (1, 2.2%) genes. Six cases (14.3% of participants with GPV in aoCPG) had tumor findings supportive of a potential causative role of the identified GPV in cancer development (Table 1). A summary of the identification process of the study cohort is displayed in Figure 1.

TABLE 1.

Summary of patients’ characteristics with tumor findings suggesting a causative relationship between the identified germline pathogenic variant and cancer development.

Patient	Age at diagnosis (years)	Sex assigned at birth	Diagnosis	Germline pathogenic variant	Transcript number	Tumor finding(s) supporting association	Treatment implications
1	17	Male	Diffuse high-grade glioma	ATM c.1236–2delA (splicing)	NM_000051.4	LOH for ATM	PARP inhibitor
2	9	Male	Atypical teratoid rhabdoid tumor	BRIP1 c.508–1G>C (splicing) Additional finding: TP53 c.818G>A; p.Arg273His	NM_032043.2	LOH for BRIP1 LOH for TP53	No
3	20	Male	Mixed germ cell tumor of pineal gland	CHEK2 c.1427C>T; p.T476M	NM_007194.3	LOH for CHEK2	No
4	15	Male	Anaplastic astrocytoma	MLH1 c.1415_1416delGA; p.R472fs	NM_000249.3	IHC (absent MLH1/PMS2), hypermutation	Immune checkpoint inhibitor
5	17	Male	Giant cell glioblastoma	MSH6 c.3158_3159delGT; p.C1053fs	NM_000179.2	IHC (absent MSH6), LOH for MSH6, hypermutation	Immune checkpoint inhibitor
6	22	Female	Anaplastic astrocytoma	MSH6 c.3439–2A>G (splicing)	NM_000179.2	LOH for MSH6, hypermutation	Immune checkpoint inhibitor

LOH: loss-of-heterozygosity

IHC: immunohistochemistry

PARP: poly adenosine diphosphate-ribose polymerase

Figure 1:

Identification of patients in this case series from total group of paired tumor-germline participants.

Case series

Patient 1

17-year-old diagnosed with adamantinomatous craniopharyngioma at age 13, which was treated with gross total resection and focal proton radiation. He presented to the emergency department at age 17 with right hemiparalysis and speech and cognitive difficulties. A brain MRI identified masses in the basal ganglia and thalamus, which were biopsied and were consistent with diffuse high-grade glioma, WHO Grade 4, clinically noted to be radiotherapy-induced secondary to his craniopharyngioma treatment. Sequencing completed after his glioma diagnosis identified a GPV in ATM (c.1236–2delA) with LOH in the tumor in addition to somatic aberrations including MET amplification and CDKN2A/2B homozygous deletion, H3-wildtype, and IDH-wildtype. The craniopharyngioma was not available for sequencing. He initially received standard of care therapy with chemoradiation (hypo-fractionated 40.5 Gy radiotherapy with concurrent temozolomide followed by Avastin infusions), but adjuvant temozolomide was not started due to severe side effects including transfusion-dependent thrombocytopenia, a severe vasculitis rash, and pseudomonas bacteremia. After a very good radiographic and clinical response by one month after chemoradiation, targeted therapy with a poly adenosine diphosphate-ribose polymerase (PARP) inhibitor (olaparib) was initiated due to the known GPV in ATM. However, olaparib was discontinued after about one week due to significant side effects, including nausea, vomiting, hematemesis, and hematochezia. The multiple areas of residual tumor have been stable on surveillance MR imaging, and additional therapy has been deferred in the absence of clinical or radiographic progression. The patient has no known family history of cancer.

Patient 2

9-year-old presented with new-onset seizures, headaches, and vomiting. Brain MRI identified a left parieto-occipital lesion that was biopsied and returned as atypical teratoid rhabdoid tumor. This tumor was treated with left parietal craniotomy, focal proton radiotherapy, and chemotherapy via modified Intergroup Rhabdomyosarcoma Study-III protocol (multi-agent). Sequencing of the resected specimen identified somatic point mutations in PTEN, BAP1, a SMARCB1 fusion, and multiple regions of uniparental disomy. GPV were identified in the BRIP1 and TP53 genes (both located on chromosome 17), with LOH for the entire chromosome 17, including both of these genes, found in the tumor. While a clinical trial that enrolls individuals with cancer and a GPV in BRIP1 on PARP inhibitors was identified, the patient was not eligible based on age. The patient had symptomatic metastatic progressive disease despite treatment, and re-resection with additional radiation were completed approximately a year after initial resection. Repeat sequencing was completed on the re-excision specimen; SMARCB1 fusion was absent, and the tumor was positive for expression of SMARCB1 and SMARCA4. Attempts to initiate PARP inhibitor were denied by insurance and the patient passed away at age 11. Their family history at the time of testing was notable only for melanoma in the patient’s father diagnosed as a teenager, although the family health history for other paternal relatives is unknown.

Patient 3

20-year-old presented with headaches and hydrocephalus. Biopsy of an identified pineal mass showed mixed germ cell tumor with yolk sac (60%), germinoma (10%), and 30% adenocarcinoma (possible teratoma). He received chemotherapy (per ACNS1123 regimen: alternating carboplatin/etoposide and ifosfamide/etoposide) followed by proton radiation to the brain and spine due to residual disease. The patient had continued disease progression which was treated with salvage chemotherapy (gemcitabine/paclitaxel/oxaliplatin) and resection. Sequencing was completed following resection, with somatic mutations with potential treatment implications including AKT1, KRAS, and MAP2K1, although none of these somatic aberrations impacted treatment. A GPV was identified in the CHEK2 gene with LOH in the tumor, but no related targeted therapies were available. The patient received additional chemotherapy (etoposide/ifosfamide/cisplatin) followed by tandem autologous stem cell transplants. Locally recurrent disease led to a plan for additional radiation, however, the patient passed away before this was initiated. There was no family history of cancer noted.

Patient 4

15-year-old presented with new onset seizures. A brain MRI identified a left temporal parietal lesion which was diagnosed as anaplastic astrocytoma on biopsy. Sequencing of this specimen identified a hypermutated tumor (H3-wildtype, IDH-wildtype) with absent MLH1 and PMS2 staining and paired germline testing confirmed the presence of a MLH1 GPV consistent with a diagnosis of Lynch syndrome, with no additional GPV identified. The patient completed proton bean radiation then received therapy with an immune checkpoint inhibitor (PD1 inhibitor, pembrolizumab). Although his disease was initially stable, treatment was ultimately discontinued due to widespread progression and the patient passed away about 18 months after his diagnosis. This patient had a family history of colorectal cancer diagnosed in his father at age 38, with his father having a known diagnosis of Lynch syndrome prior to the patient’s cancer diagnosis.

Patient 5

17-year-old presented with headaches, vomiting, cold sensation in his hand, and bright spots in his vision. A brain MRI identified two left-sided temporal enhancing lesions, which were confirmed as giant cell glioblastoma, WHO Grade 4. The patient was treated with a gross total resection and focal radiation therapy. The tumor was hypermutated (IDH-wildtype) and a GPV in MSH6 was identified, with accompanying tumor LOH and loss of MSH6 on IHC. He received one cycle of lomustine but was switched to immune checkpoint inhibitors (PD1 inhibitor, nivolumab and CTLA-4 inhibitor, ipilumimab) after the sequencing results identified a diagnosis of Lynch syndrome. He had no evidence of tumor return for almost a year on this therapy but ultimately progressed and passed away. Family history was notable only for pancreatic cancer in his paternal grandfather.

Patient 6

22-year-old presented with worsening headaches, with a head CT identifying a mass primarily within the left frontal lobe that was diagnosed as anaplastic astrocytoma. She had been tested for Lynch syndrome via multigene panel testing about two years prior to her cancer diagnosis based on a known maternal family history of this condition (MSH6 GPV) and was aware of her Lynch syndrome diagnosis prior to enrolling in this sequencing study. She had initiated high-risk cancer screening, which included colonoscopy and upper endoscopy based on her age (both normal). The brain mass was excised, with sequencing completed on the sample. Results were consistent with a hypermutated tumor, notable for a somatic mutation in IDH1 and LOH for MSH6. She completed radiation and initiated treatment with an immune checkpoint inhibitor (PD1 inhibitor, pembrolizumab) but unfortunately experienced disease progression. She was planning to pursue an alternative treatment, however, she passed away before another therapy could be initiated. Family history was notable for ovarian cancer in the patient’s mother, breast, ovarian, uterine, stomach, and kidney cancer in a maternal great-aunt, early onset colorectal cancer in another maternal great-aunt, and ovarian cancer in a third maternal great-aunt, all of whom had a confirmed diagnosis of Lynch syndrome.

Rates of pathogenic variants

The rate of GPV in different adult-onset cancer predisposition syndromes were compared to the rate of these conditions in the general population to determine if there was enrichment for any conditions in our cohort. The rate of GPV in our cohort did not differ significantly from those in the general population.

Discussion

Here, we present results from a cohort of pediatric oncology patients enrolled in a paired tumor-germline sequencing study, focusing on those with GPV identified in aoCPG. Tumor features were analyzed alongside germline results with the goal of identifying potential associations between the identified GPV and cancer development. Most participants with GPV in aoCPG (85.7%) did not have tumor characteristics suggesting a causative role of the GPV in cancer development. Our findings support that, generally, aoCPG are not responsible for pediatric cancer development. Additionally, findings suggest that deferral of germline testing for familial GPV in aoCPG for unaffected children until adulthood is usually appropriate. However, we present a limited number of pediatric oncology cases where tumor data support a causative role of a GPV in an aoCPG.

The potential connection between the ATM, BRIP1, and CHEK2 GPV and the glioma, atypical teratoid rhabdoid tumor, and mixed germ cell tumor, respectively, were all unexpected based on the known tumor spectrum associated with these conditions. Heterozygous GPV in ATM are associated with an increased risk for breast, pancreatic, and ovarian cancers, with preliminary evidence suggesting potential increased risk for prostate cancer²⁷. GPV in BRIP1 predispose to ovarian cancer. In this case, the potential contribution of the BRIP1 GPV to tumor development is difficult to discern given the presence of a TP53 GPV in the same patient, with prior cases of atypical teratoid rhabdoid tumor reported in other individuals with GPV in TP53^28,29. GPV in CHEK2 have been associated with an increased risk for a number of malignancies, however, the particular GPV identified here (p.T476M) has an association with breast cancer risk only^27,30. Cancer screenings for individuals with ATM and CHEK2 GPV generally begin by age 30, and risk-reducing surgery for individuals with BRIP1 GPV are recommended around age 45²⁷. Prior knowledge of these GPV in these pediatric patients would not have impacted cancer screening recommendations given their young age.

In contrast, all three individuals with Lynch syndrome-associated GPV had brain tumors, which fall within the known cancer risk spectrum associated with this condition. Brain cancer is a rare manifestation of Lynch syndrome, with the cumulative risk through age 80 estimated at 0.7%−1.7% (MLH1), 2.5–7.7% (MSH2/EPCAM), 0.8–1.8% (MSH6), and 0.6-≤1% (PMS2) compared to the general population risk of 0.5%³¹. This is generally considered an adult-onset malignancy; the estimated average age of presentation is 43–54 years for those with GPV in MSH6 and 40 years for those with GPV in PMS2³¹. Estimates regarding the average age of onset for patients with GPV in MLH1 and MSH2/EPCAM are unavailable³¹. Despite the increased cancer risk, no specific high-risk screening of the brain is recommended for patients with Lynch syndrome. Guidelines recommend educating patients regarding signs and symptoms of neurologic cancer and the importance of prompt reporting of abnormal symptoms to their physicians³¹. Therefore, even in patients known to have Lynch syndrome (such as Patient 6 in this case series) brain malignancies are not detected by current screening guidelines. However, knowledge of a genetic diagnosis is useful, as management of certain tumors, such as IDH1-mutant gliomas, may be different in individuals with Lynch syndrome or another condition where targeted therapies are available^32,33. While childhood onset brain tumors have been reported in the autosomal recessive condition constitutional mismatch repair deficiency, the patients reported here had molecular analysis ruling out other GPV in Lynch syndrome-associated genes and none had clinical findings characteristic of constitutional mismatch repair deficiency, such as additional malignancies or café au lait macules.

Limitations of our work include that tumors may have other mechanisms of inactivation not captured here. In particular, mutational signatures in patients with GPV in the homologous recombination repair support a role as a driver of pediatric cancer development, even in the absence of a second somatic hit, but were not available in our analysis³⁴. The measures used to determine potential causality, including LOH, IHC, tumor mutation burden, and a second somatic hit, may be imperfect measures of association. In particular, the presence of multiple regions of uniparental disomy in Patient 2’s tumor suggest the possibility that broader genomic disruption, rather than a specific second-hit event, could account for LOH for BRIP1 and TP53. Therefore, this work may over or underestimate the contribution of the identified GPV to cancer development. There may be other potential contributing genetic factors that explain the patient’s phenotype. For example, Patient 2 had GPV identified in both BRIP1 and TP53, with GPV in TP53 previously reported in patients with atypical teratoid rhabdoid tumors^28,29. The small number of patients with GPV in each gene and with each cancer type limited our ability to assess gene-disease relationships.

Conclusion

Overall, our findings suggest that a small proportion of pediatric cancer diagnoses are due to GPV that have historically been considered to only cause adult-onset cancer predisposition. While our cohort is too small to draw definitive conclusions, these findings suggest that this topic is nuanced and discussion with families should avoid using absolutes when reviewing potential risks for childhood cancers in the case of aoCPG. Ultimately, the increasing use of paired tumor-germline sequencing, particularly in the pediatric population, may provide increased knowledge to inform predictive genetic testing recommendations for children and screening protocols.

Context summary.

Key objective:

What can paired tumor/germline sequencing tell us about the contribution of germline pathogenic variants (GPV) in adult-onset cancer predisposition genes (aoCPG) to pediatric cancer development?

Knowledge generated:

Six pediatric patients had tumor findings suggesting a potential role of an aoCPG (ATM, BRIP1, CHEK2, MLH1, MSH6) in their cancer development. Identification of GPV in aoCPG in the pediatric population can impact therapy selection, future cancer screening, and familial testing as well as contribute to our knowledge regarding the tumor spectrum associated with these cancer predisposition syndromes.

Relevance:

In rare instances, GPV in aoCPG may contribute to pediatric cancers. Reports of these cases can inform predictive genetic testing and screening recommendations.

Abbreviations:

GPV

Germline pathogenic/likely pathogenic variant(s)

aoCPG

Adult-onset cancer predisposition gene(s)

PARP

Poly adenosine diphosphate-ribose polymerase

IHC

Immunohistochemistry

LOH

Loss-of-heterozygosity

PEDS-ONCOSEQ

Personalized Medicine based on Molecular Profiling of Pediatric and Young Adult Patients with Cancer