Summary
A medRxiv preprint reports five DNA-methylation clusters in 138 Wilms tumours and a classifier that identified WT1-driven tumours with about 90% accuracy. The researchers say using these patterns to guide genetic reanalysis uncovered previously unrecognised WT1 variants, including one that led to surveillance enrolment.
A medRxiv preprint reports that DNA-methylation patterns can separate Wilms tumours into subgroups associated with tumour biology and genetic predisposition. In profiling 138 tumours, the researchers identified five clusters, including one enriched for Beckwith-Wiedemann Spectrum tumours and a distinct WT1-driven subgroup. They also report that a classifier identified WT1-driven tumours with about 90% accuracy across independent validation cohorts.
The team used the subgroup patterns to guide reanalysis of genetic data, uncovering previously unrecognised pathogenic WT1 variants. One newly identified germline variant led to enrolment in surveillance, according to the abstract.
How tumour methylation can inform genetic analysis
DNA methylation consists of chemical marks on DNA that can influence how genes are regulated. Patterns of these marks can reflect the biological state of a tumour, providing a different kind of information from the DNA sequence variants sought in conventional genetic testing.
The researchers’ approach uses the tumour’s methylation profile to identify patterns associated with particular subgroups, then directs attention to relevant genetic findings. This may be useful because, as the authors note, predisposition can be difficult to recognise when clinical features vary or genetic variants are mosaic or cryptic. The reported result is a way to complement genetic analysis, not a replacement for it.
What the study reports—and what it means
The study identifies five molecular clusters among the 138 profiled Wilms tumours. A random-forest classifier—a machine-learning method that combines decisions from multiple decision trees—predicted the WT1-driven tumour subgroup with approximately 90% accuracy in independent validation cohorts. The abstract does not give the validation cohort sizes or sensitivity and specificity, so the accuracy figure is best read as a research result rather than a complete measure of performance in clinical testing.
The report is a preprint. Its findings suggest that tumour methylation profiling could help clinicians and researchers recognise cases warranting focused genetic review, including cases that standard diagnostic workflows may have missed. Further assessment in defined clinical settings would be needed to establish how the approach performs as part of routine care.