Genetic testing for cancer predisposition syndromes in pediatric cancer patients: current practices and recommendations - a systematic review

Overall, most studies included cohorts with broadly similar structures; however, differences in cohort composition were observed. Cohort sizes varied considerably, ranging from 31 to 3,975 participants. In some studies, cohorts consisted exclusively of patients with a specific tumour type, such as medulloblastoma²⁷, rhabdomyosarcoma²⁹, or paediatric non-medullary thyroid cancer³⁴, whereas in most studies inclusion was based primarily on a diagnosis of cancer rather than tumour subtype. This heterogeneity likely reflects differences in study objectives, patient selection strategies, and the rapid evolution of genetic testing technologies in paediatric oncology.

Some studies applied more complex cohort definitions, for example including childhood cancer patients with clinical features suggestive of a genetic cancer predisposition⁴¹, French patients with clinical suspicion of Li-Fraumeni syndrome⁴⁰, or individuals undergoing either SGT or MGPT.³⁶

Some studies focused on survivors of childhood cancer, such as adult 10-year survivors of childhood cancer diagnosed before 15 years of age between 1960 and 2006 in the United States³², or survivors who were more than five years from initial cancer diagnosis.³⁰ This highlights that germline predisposition is clinically relevant not only at primary diagnosis but also during long-term survivorship.

A notable exception was the study Differences in TP53 Mutation Carrier Phenotypes Emerge From Panel-Based Testing. This study differed from most other included publications, as it primarily evaluated the proportion of TP53 variant carriers identified through SGT compared with MGPT. The cohort composition was also distinct, including both individuals with a history of cancer and individuals without cancer who underwent testing for other indications, provided that the testing panel included TP53. Furthermore, this study did not impose age restrictions, meaning that children were included but were not exclusively represented.³⁶ The remaining studies were more similar in design, with the exception of the ALTE03N1 study, which included both childhood cancer survivors with subsequent neoplasms (cases) and survivors without subsequent neoplasms (controls), all of whom underwent germline genetic testing.³⁸ Given that this literature review did not focus on the occurrence of new primary tumours, both the control and case groups provide relevant information for our analysis. In addition, the case group offers an interesting additional perspective on genetic predisposition in patients with multiple primary tumours.

With regard to patient age at diagnosis, our analysis primarily focused on paediatric cancer populations. Most studies included children and adolescents or children and young adults. However, exceptions were observed, as noted above, where children represented only a subset of the cohort.³⁶ In addition, in more complex study designs that combined multiple subcohorts, age distributions were not uniform. For example, in the medulloblastoma study, the retrospective cohort had no age restriction, whereas four prospective cohorts applied different age limits (≤5 years, 3–21 years, 3–39 years, and all ages).²⁷ More detailed cohort descriptions and age restrictions are provided in Supplementary Table 1.

The studies included in our systematic review also differed substantially in the number of analysed genes. Overall, the number of genes analyzed ranged from 1 to 1,048. For example, the included meta-analysis focused on 10 genes that were common across all included datasets.³⁵ In general, it could be observed that studies with more tumor-type–specific cohorts tended to focus on a smaller number of genes. For instance, in the study Revisiting Li-Fraumeni Syndrome from TP53 Mutation Carriers, which included patients with clinical suspicion of Li-Fraumeni syndrome, only a single gene (TP53) was analyzed⁴⁰, which is clinically justified given the underlying syndrome. Similarly, in the medulloblastoma cohort, only six genes were analyzed.²⁷ However, this was not a consistent rule, as in the rhabdomyosarcoma cohort, despite tumor specificity, 70 genes were analyzed.²⁹ This variability likely reflects differences in clinical indication for testing, availability of sequencing technologies, and evolving understanding of cancer predisposition in pediatric oncology.

Testing approaches ranged from SGT to MGPT. Overall, the most commonly used sequencing approaches were WGS and WES, performed for example using blood or tumour tissue samples²⁷, and in some studies using trio analysis (patient and both parents).²⁸ In the meta-analysis, additional methods included next-generation sequencing based (NGS-based) gene panels and targeted sequencing approaches.³⁵ The increasing use of broad sequencing approaches reflects the shift toward comprehensive germline risk assessment but also increases the likelihood of detecting variants with uncertain clinical relevance.

The TP53-focused study directly compared SGT and MGPT, illustrating methodological differences in variant detection. Within this study, several types of multigene panels were used, including comprehensive cancer panels, breast cancer panels, gynecological cancer panels, colon/gastrointestinal panels, pancreatic cancer panels, and kidney cancer panels.³⁶

The majority of the remaining studies employed MGPT as the primary testing approach, while the ALTE03N1 study systematically tested both cases and controls for pathogenic or likely pathogenic germline variants.³⁸

The proportion of genetic predisposition varied substantially across the included studies. This variability is expected, given the differences in cohort structure, the spectrum of analysed genes, and the applied testing methodologies. The prevalence of germline pathogenic or likely pathogenic variants ranged from 0.2%to 47.5%. Intuitively, studies focusing on a smaller number of genes are likely to report a lower proportion of genetic predisposition, resulting in a lower diagnostic yield. Conversely, one could argue that cohorts with a narrower tumour-type focus and fewer analysed genes may include patients with a higher pre-test probability for a specific genetic predisposition, thereby increasing the yield.

In line with this, the meta-analysis reported a prevalence of only 2.99%across the 10 evaluated genes.³⁵ Similarly study Differences in TP53 Mutation Carrier Phenotypes Emerge From PanelBased Testing, focusing on a single gene, showed that 4.1%of individuals tested with SGT were TP53-positive compared with 0.2%of those tested with MGPT.³⁶

In the ALTE03N1 cohort, the prevalence was 14.43%among survivors with subsequent neoplasms and 6.08%among controls.³⁸ This pattern is not unexpected, as it is well established that genetic predisposition confers an increased risk for specific cancer types and therefore also for the development of additional primary tumours in genetically predisposed individuals.⁴⁹

Studies in children and adolescents (or childhood cancer survivors) that analysed a larger number of genes and included tumour-nonspecific cohorts reported prevalences ranging from 5.8%to 47.5%. The lowest proportion was observed in the study Genetic Risk for Subsequent Neoplasms Among Long-Term Survivors of Childhood Cancer, with a prevalence of only 5.8%.³⁰ However, it should be noted that this cohort consisted of childhood cancer survivors (≥ 5 years post-diagnosis). One must consider that not all children with cancer survive long term, and those who do not survive are excluded from such cohorts, which may artificially lower the observed prevalence of genetic predisposition.

Interestingly, a study with a similar design which also included childhood cancer survivors (≥ 10 years post-diagnosis), reported a prevalence of 11.8%.³¹ Although this study included approximately 1.2-fold fewer participants, both cohorts comprised at least 2,450 individuals. Thus, both cohorts can be considered substantially large in the context of germline predisposition studies. This raises the question of what might explain the roughly two-fold difference in prevalence, given that the number of analysed genes was identical and both studies used WGS (30-fold coverage in the first study and mean coverage of 36.8x in the second).

Another outlier was the 47.5%prevalence reported in study from Denmark.³³ Several factors could contribute to this exceptionally high proportion. The cohort may have included individuals with a more apparent suspicion of genetic predisposition, or the very young age of the patients (more than half were under 5 years old) may have played a role, as environmental influences are less likely to contribute to cancer development in this age group.

The remaining prevalence estimates aligned well with theoretical expectations, as the proportion of genetic predisposition in childhood cancer is generally estimated at around 10%.⁵⁰ The included literature review also reported comparable values, ranging from 8.5%to 35.5%.³⁹

Taken together, these findings highlight that reported prevalence of germline predisposition is highly context-dependent and must always be interpreted in light of cohort selection, testing breadth, and clinical indication.

As it is documented, the leading cause of premature mortality among childhood cancer survivors are secondary neoplasms (SNs).^51–53 Most of them are associated with prior radiotherapy⁵⁴, as well as exposure to alkylating agents, platinum compounds, and anthracyclines⁵⁵, yet growing evidence suggests that underlying pathogenic or likely pathogenic variants (P/LPVs) in cancer predisposition genes (CPGs) may further influence risk.^56,57 These findings underscore the potential value of integrating CPG status into clinical practice, not only to identify survivor subgroups at highest risk for SNs, but also to guide decisions on which children should undergo genetic screening to refine risk assessment and implement targeted surveillance or risk-reduction strategies. In one of the studies by Zhou et al., carriers of P/LPV had 4.26-fold increased odds of developing SNs (95%CI = 2.36–7.69), with stronger associations observed in survivors exposed to platinum-based chemotherapy and in those not treated with radiotherapy. Survivors were then stratified using a clinical risk classifier into low-risk (22%) and moderate-to-high-risk groups (78%). Most carriers of at least one P/LPV, including all TP53 and RB1 pathogenic variant carriers, were classified as moderate-to-high risk. TP53 carriers most commonly developed secondary central nervous system tumours, osteosarcomas, soft tissue sarcomas, or breast cancer, consistent with the known tumour spectrum associated with this gene. These observations support referral of survivors at moderate-to-high risk for genetic counselling and highlight the utility of mutation status in guiding individualized surveillance strategies.³⁸ Follow-up provided by a dedicated physician adhering to current guidelines probably ensures optimal long-term care for survivors.⁵⁸

In the second part of this systematic review, we summarized recent guidelines and recommendations for genetic counselling and testing in children diagnosed with cancer, based on a review of the available literature identified through an online search. Seven relevant publications were included, and their key concepts are synthesized in Table 2 (Results). This section we considered particularly important, as growing evidence suggests that an increasing proportion of childhood cancers is associated with CPGs, and advances in knowledge and technology continue to expand the spectrum of identifiable genes and hereditary syndromes, enabling more personalized treatment and targeted surveillance for patients and their families.⁵⁹ Pediatric oncologists play a central role in identifying children at risk for CPSs; however, only a limited number have sufficient specialized training to manage this process independently.^60,61 Therefore, we aimed to compile and compare the most up-to-date guidelines within a single framework, focusing on differences in patient selection for referral to genetic counselling and/or testing, the timing of these interventions, and the clinical, familial, and tumour-related criteria applied.

Genetic testing is becoming increasingly rapid, affordable, and widely available, and many pediatric oncology centres have incorporated paired tumour–germline sequencing into routine diagnostics. Nonetheless, the key question remains which patients should undergo comprehensive germline testing.⁴² Should testing be offered universally to all children with cancer, or should it follow a more conservative, phenotype-driven approach?

Currently, three main strategies are described for identifying patients with CPS: (i) a phenotype-driven approach, in which genetic testing is offered to selected high-risk patients based on clinical evaluation; (ii) a phenotype-agnostic approach, in which all patients undergo comprehensive testing of genes associated with CPSs using large gene panels; and (iii) a combined approach, in which all patients are tested for a limited gene set, followed by additional phenotype-driven testing when clinically indicated.⁴²

Most guidelines agree that a defined clinical suspicion of a CPS should precede referral for genetic counselling and testing. The SIOPE Host Genome Working Group, following its meeting in Paris in November 2022, concluded in a 2024 publication that all children with cancer should undergo clinical screening for CPS, with targeted genetic testing guided by clinical indications. The use of comprehensive, phenotype-agnostic CPS gene panels in all pediatric patients was recommended primarily for research purposes, rather than routine clinical practice, to evaluate their impact and expand understanding of CPS phenotypes.⁴²

Most guideline publications support a phenotype-driven approach, and a variety of clinical screening tools have been developed to identify children at increased risk for CPS. Across most approaches, there is a broad consensus in the criteria used, which consistently address: (i) family history, assessed through evaluation of unaffected children or children with cancer⁴³, three-generation pedigrees⁶², or focused assessment of the affected child⁴⁸; (ii) tumour characteristics, such as rare tumours, malignancies typically associated with adulthood but occurring in childhood, tumours with distinct histological or molecular subtypes, tumours linked to known CPS⁴³, or the use of predefined tumour lists prompting consideration of an underlying CPS^43,48,62; (iii) congenital or other anomalies, including dysmorphic features, congenital malformations, growth abnormalities, characteristic cutaneous findings (e.g. café-au-lait macules, abnormal pigmentation)^43,48,62; (iv) the occurrence of multiple primary malignancies, including multiple or bilateral primary tumours of the same type⁴³, secondary, bilateral, multifocal, or metachronous malignancies⁶², or the development of multiple cancers, one of them under 18 years where subsequent malignancies are not attributable to therapy⁴⁸; and (v) excessive treatment-related toxicity.^43,48,62

A decision-support tool is also described in The McGill Interactive Pediatric OncoGenetic Guidelines (MIPOGG): An approach to identifying pediatric oncology patients most likely to benefit from a genetic evaluation. MIPOGG is based on tumour-specific algorithms with sequential “yes/no” questions, allowing clinicians to assess the likelihood that a tumour is caused by a germline mutation. The algorithms combine tumour-specific and universal criteria (detailed in Table 2) and include a direct referral category for tumours with a high probability of germline origin. The tool is currently available in a non-digital format, with a point-of-care mobile application planned for future implementation.⁴⁵

Two articles describe an alternative, phenotypeagnostic approach, currently most extensively implemented in Sweden, where genetic testing is offered to all children with cancer regardless of clinical phenotype. Since 2021, Sweden has provided WGS of tumour and normal tissue as part of precision diagnostics for all pediatric cancer patients⁴⁶, and since May 2024, has additionally incorporated WGS and whole-transcriptome sequencing (WTS) into routine diagnostic care at the time of cancer diagnosis.⁴⁷

Globally, however, the phenotype-driven approach remains predominant in clinical practice. Bakhuizen et al. showed that systematic clinical assessment followed by targeted referral for genetic evaluation identifies over 90%of patients with a (likely) causal CPS.^7,45,63 While broad genetic testing of all paediatric cancer patients may enhance understanding of cancer aetiology and is valuable in research settings, it cannot replace comprehensive clinical evaluation, as some CPSs may otherwise remain undetected.^64,65

The phenotype-driven approach generally increases clinical relevance by focusing testing on patients with a clear suspicion of CPS, thereby increasing the likelihood of identifying clinically actionable pathogenic variants. Use of smaller gene panels also reduce the detection of variants of uncertain significance or incidental findings, which can create uncertainty for both healthcare professionals and patients and increase the psychological burden on families.^66,67 However, this approach may be limited by incomplete or unreliable family history, variable penetrance, and de novo mutations.^68,69 In addition, the presence of congenital or other anomalies does not necessarily indicate an underlying CPS^70,71, and their absence does not exclude it, as some clinical features may only become apparent later in life.⁷² In such cases, careful clinical judgment is essential, as inaccurate or incomplete assessment may lead to suboptimal patient selection for genetic evaluation.

Advocates of the phenotype-agnostic approach argue that universal testing supports precision medicine by enabling molecularly informed treatment decisions.⁴⁶ National pilot data indicate improved tumour classification or identification of prognostic markers in approximately 50%of cases, as well as earlier diagnosis and more individualized long-term follow-up.^46,47 Nevertheless, widespread implementation remains limited by substantial financial, organizational, and workforce demands.

Although current guidelines predominantly favour a phenotype-driven strategy, each approach has inherent strengths and limitations. Ultimately, the choice of diagnostic strategy depends on local resources, technical capacity, availability of clinical genetic expertise, and regulatory frameworks.⁴² So far, no consensus has been reached on the optimal approach for children with cancer; however, in most settings, the phenotype-driven strategy currently provides the most balanced and sustainable option).

A potential limitation of this review is that only articles published in English were included, which may have led to the exclusion of potentially relevant studies published in other languages. Finally, it should be noted that in the Current Studies section of this review, the literature search was limited to articles identified through keyword searches in the PubMed database. This methodological approach may have resulted in the omission of relevant studies indexed in other biomedical databases. Consequently, a more comprehensive search across multiple databases might have identified additional publications, which could further strengthen the evidence base and broaden the population-level generalizability of our conclusions.