Background: Assessing genetic risk before a cancer diagnosis

Among children referred for testing because of suspected genetic conditions, subsequent malignancy incidence was higher in those carrying pathogenic or likely pathogenic germline variants than in comparison groups. Published in Nature Medicine, this study moved beyond analyses centered on patients already diagnosed with cancer and tracked tumor occurrence in children who had undergone genetic testing for other symptoms.

Germline variants contribute to some genetic disorders and tumor predisposition. Prior studies cited in the paper reported that about 10% of children with cancer carry germline variants. However, explaining subsequent risk when such variants are found before a cancer diagnosis is a separate question. Informing families about secondary findings unrelated to the original reason for testing and discussing follow-up care require evidence on clinical outcomes by variant classification.

Key findings: Malignancy incidence differed by variant classification

The researchers analyzed 75,602 infants and young children referred for testing because of suspected genetic conditions between January 2016 and January 2025. At testing, 53.5% were neonates aged 28 days or younger. Neurological, metabolic, and immune-related symptoms were among the main reasons for testing; this was not a screening study that randomly recruited children from the general population.

The analysis used data from clinical exome sequencing (CES), accounting for 70%, and whole-exome sequencing (WES), accounting for 30%. The team reassessed variants in 139 tumor-susceptibility genes and combined retrospective medical records with prospective follow-up data.

Of the 75,399 participants without a tumor at testing, follow-up data were available for 64,187, representing 85.1%. Observation from the date of testing until tumor occurrence or May 31, 2025, identified 208 newly diagnosed tumors: 66 malignant and 142 benign. The full recruitment period does not represent the follow-up duration for every participant.

Among patients with follow-up data, 1,187 carried pathogenic (P) or likely pathogenic (LP) variants. Their malignancy incidence was 3.23 cases per 1,000 person-years, compared with 0.236 cases per 1,000 person-years among those carrying only variants of uncertain significance leaning toward likely pathogenic (VUS-LP) and 0.272 cases per 1,000 person-years in the remaining group, whose variants fell into neither category. Person-years measure the observation time accumulated across participants. These rates cannot be directly converted into an individual child's annual probability of developing cancer or lifetime penetrance. VUS-LP must also be distinguished from P/LP.

Implications and outlook: Observational evidence to support counseling

The findings add evidence for discussing tumor risk and considering surveillance in children whose evaluation for a genetic condition identifies a P/LP variant. However, this observational study identified an association with risk; it did not establish the variants' molecular mechanisms of carcinogenesis or demonstrate the effectiveness of surveillance interventions. Nor did it evaluate the performance of a tool for predicting when an individual will develop a tumor.

The characteristics of the study population matter for interpretation. Results from patients referred for severe noncancer symptoms are difficult to apply directly to healthy children in the general population. The analysis was also limited to 139 genes, and follow-up data were unavailable for some patients. Relatively short follow-up and underrepresentation of brain tumors leave open the possibility that later-onset tumor risks were not fully captured. Broader gene analysis and long-term follow-up are needed to assess individual risk estimates and the clinical utility of surveillance strategies.

The study population, incidence comparisons, analytical methods, and limitations can be found in the original paper.