Key Points

Abstract

Introduction

Germline genetic factors play a substantial role in the development of childhood cancer.¹ Identifying cancer predisposition syndromes (CPSs) in children is important, as it enables surveillance for early detection of second cancers and genetic counseling and testing of relatives. In some CPSs, a modification in treatment is needed.^1,2 Traditionally, the diagnosis of CPSs in children with cancer is based on clinical suspicion prompting referral to a clinical geneticist. However, the diagnostic approach is increasingly shifting toward a genotype-first approach.^3,4,5

To improve CPS diagnostics among children with cancer, it is essential to compare the effect of germline sequencing of all childhood cancer predisposition genes vs (targeted) genetic testing based on clinical selection. However, to our knowledge, recent reports on a phenotype-first approach in large, unselected childhood cancer cohorts are lacking. For this, we retrospectively reviewed medical records of children with a new diagnosis of a neoplasm in the Netherlands during a 19-month period (June 2018 to December 2019) before introducing whole-exome sequencing (WES) in the diagnostic work-up. This nationwide, unselected cohort provides insight on the diagnostic yield of a phenotype-first approach and can potentially be used as a reference cohort for genotype-first studies.

Methods

Results

Patients Identified With a CPS

In 71 of 824 patients (8.6%) a CPS was identified, including 26 different syndromes (Figure 1). In 68 patients (96%), the CPS was identified by a phenotype-first approach. In 3 patients, the CPS was identified by performing WES in a genotype-first approach as part of a precision medicine program or tumor diagnostic workup. The prevalence of CPSs ranged from 5.4% for patients with hematologic neoplasms to 7.8% for central nervous system tumors, and 12.6% for solid tumors. Down syndrome (n=10) and neurofibromatosis type 1 (n=10) were the most common CPSs diagnosed. Most CPSs were molecularly confirmed (69 of 71 cases). In 66 patients (93%), the identified CPS explained neoplasm development. In 2 of these patients, who both had a diagnosis of a high-grade glioma and germline TP53 pathogenic variant, the CPS was not identified by the phenotype-first approach, but rather detected by performing WES as part of a precision medicine program or tumor diagnostic work-up. In 5 patients, germline pathogenic variants were identified in genes for which there are currently no evidence for a causal role in the development of neoplasms in these children (ie, CHEK2 in acute myeloid leukemia, heterozygous PMS2 in Wilms tumor, SDHA in osteosarcoma, MSH6::FBXO11 in diffuse large B-cell lymphoma,⁸ and Turner syndrome in neuroblastoma).

Open in a separate window

Figure 1.

Identification of 71 Cancer Predisposition Syndromes in a National, Unselected Cohort of 824 Children With a Neoplasm

AA indicates aplastic anemia; BWS, Beckwith-Wiedemann spectrum; CMMRD, constitutional mismatch repair deficiency; FA, Fanconi anemia; feo, feochromocytoma; GCT, germ cell tumor; HGG, high-grade glioma; IMF, infantile myofibromatosis; leu, leukemia; LGG, low-grade glioma; lymph, lymphoma; MB, medulloblastoma; MDS, myelodysplastic syndrome; NBL, neuroblastoma; NET, neuroendocrine tumor; NF1, neurofibromatosis type 1; OS, osteosarcoma; PNST, peripheral nerve sheath tumor; PPB, pleuropulmonary blastoma; RB, retinoblastoma; RMS, rhabdomyosarcoma; SCCOHT, small-cell carcinoma of the ovary hypercalcemic type; SDS Shwachman-Diamond syndrome; TSC, tuberous sclerosis complex; WT, Wilms tumor.

^aIncluding molecular diagnosis for BWS based on methylation-specific MLPA nontumor kidney tissue.⁶

^bPatients were described in case reports.^8,9

Diagnostic Process of Identified CPSs

Twenty-nine of the 71 patients with a CPS (41%) had received a diagnosis of this syndrome before they developed a neoplasm. Most of these children had specific phenotypic features that allowed for a clinical diagnosis (eg, signs of Down syndrome or neurofibromatosis type 1). In patients in whom genetic predisposition was identified after they had developed a neoplasm, the specific type of neoplasm was the most frequent indicator (35 of 42 patients [83.3%]) for referral to a clinical geneticist (Figure 2A), although often in combination with other features. In most patients, the genetic predisposition was revealed by performing targeted genetic testing, being either sequencing of 1 or 2 genes or targeted methylation analysis (Figure 2, B and C).

Open in a separate window

Figure 2.

Diagnostic Process of Identified Cancer Predisposition Syndromes After Neoplasm Development

A, Specific indicators that initiated referral for clinical genetic assessment in 42 patients for whom genetic predisposition was identified after they had developed a neoplasm. The total number of patients per indicator for referral are noted in the column labels. Each column corresponds to a specific combination, and bar charts on top show the number of patients per combination. The filled-in dots show which indicator for referral is part of a combination. The feature “child with 2 or more neoplasms” is defined as the presence of bilateral, multifocal, or metachronous primary neoplasms. B, Number of genetic tests that were performed per patient. C, Type of genetic tests that were performed. SNV, single-nucleotide variation; WES, whole-exome sequencing.

^aPart of the tumor diagnostic workup or precision medicine study.

^bIncluding a kidney tumor predisposition gene panel that was offered to all children with Wilms tumors as part of the WES-KidTs study.⁶

Discussion

In this national unselected cohort study of 824 children with a neoplasm, 8.6% received a diagnosis of a CPS, of which most (96%) were identified by a phenotype-driven approach. The tumor entity was the most frequent reason for genetic testing. In contrast to many adult-onset CPSs, family history only played a limited role in recognition of the underlying predisposition. A possible explanation for this is the high contribution of de novo variants. In this cohort, 37 of 58 patients (64%) with sufficient information about inheritance mode carried a de novo pathogenic variant. In addition, 3 patients received a diagnosis of an autosomal-recessive CPS.

The prevalence of CPSs identified in the unselected cohort was similar to the 7% to 12% that was reported in earlier genotype-based studies.^3,10,11 In a recent prospective sequencing study, a higher prevalence of germline cancer-predisposing variants was reported (18%), but heterozygous variants in genes associated with autosomal recessive inherited CPSs were included.⁵ Whereas the mostly phenotype-driven diagnosis of CPSs in the current unselected cohort revealed a CPS prevalence that was similar to that of earlier genotype-based studies, the spectrum of CPS diagnoses is different between the 2 approaches. For example, Down syndrome, often excluded from these large sequence studies, was found to be the most common CPS in the current cohort, whereas relatively few children were identified with a germline TP53 pathogenic variant. The higher prevalence of certain CPSs, like Li-Fraumeni syndrome, in genotype-based studies is partly due to the biased selection of cancer types typically found in Li-Fraumeni syndrome (such as adrenocortical tumors and hypodiploid acute lymphocytic leukemia).^3,10 However, a phenotype-driven approach is also associated with an underestimation of CPSs with less prominent clinical features.

We hypothesize that identification of CPSs in children with cancer can be optimized by combining comprehensive phenotyping (clinical history and examination) with systematic genetic tests, even in patients without specific clinical features. This is illustrated in a recent study in patients with Wilms tumors in which a comprehensive and stepwise approach of diagnostic genetic testing and research-based WES analysis was performed.⁶ Whether such a combined approach is also useful in children with other types of cancer has to be proven. To our knowledge, to date, only 2 studies have combined comprehensive clinical data and WES/whole-genome sequencing in an unselected cohort of pediatric patients.^4,12 Pathogenic germline variants were found in 6.9% and 14.6% of patients, but relatively few children were included.

Conclusions

In this cohort study of children with a neoplasm, the prevalence of CPSs identified by a phenotype-driven approach was 8.6%. Whereas the CPS prevalence in this cohort was similar to that in earlier genotype-based studies, the spectrum of CPS was different. To determine whether CPS genes sequencing among all children with cancer can serve as a replacement or add-on test, the diagnostic yield, as well as other relevant aspects, such as costs and the number of unsolicited findings, should be compared with that of the classical phenotype-driven diagnostic approach.

Notes

References