Investigating a Wilms tumor DNA methylation signature in liquid biopsies

For adult cancers, plasma ctDNA interrogation of genomic alterations has the potential to detect cancers before radiographic evidence, supporting prognostication and monitoring of therapeutic response in real-time. Despite its remarkable potential, ctDNA detection is often limited to advanced disease, with very few low-grade tumors shedding sufficient ctDNA with genomic lesions characteristic of the cancer. In contrast to adult tumors, the genomes of early childhood and embryonal tumors are typically stable with low somatic and copy number alterations that are often benign and heterogeneous, making detection even more difficult. Thus, a diagnostic tool that detects DNA methylation (DNAm) changes in ctDNA in addition to genomic alterations could enable earlier cancer detection and improve outcomes. Wilms tumor (WT) is the most prevalent renal cancer in children. Our group has shown that DNAm aberrations exist in all WTs regardless of the presence of genomic aberrations, and that DNAm alterations are often the earliest pathogenic change in early lesions. Here, we present preliminary findings of our methyl-classifiers' ability to detect WT from plasma cfDNA by enzymatic methyl-sequencing (EM-seq). Using EM-seq, we are currently generating DNAm data from plasma drawn at diagnosis in 35 patients with WT and 24 controls. In the initial pilot analysis of 17 patients with Wilms tumors and 4 controls, our WT-methyl-classifier was not only able to reliably identify patients with WT, but also stratified patient cfDNA samples into two subtypes with methylation profiles similar to those reported in primary WT tissues. Additionally, we showed corroborating evidence of WT ctDNA in plasma cfDNA samples: First, copy number alterations associated with WT were present such as chromosome 1p, 11p, and 16q loss, and chromosome 1q gain. Secondly, methylation at commonly differentially methylated regions in WT plasma samples were concordant with tumor tissue - for example increased and decreased methylation at H19 and KCNQ1OT1 imprinting control regions respectively. Thirdly, cfDNA from patients with WT had smaller average DNA fragment size compared to controls - a pattern that has been described in other cancer types. Lastly, the majority of matched tumor tissues shared methylation subgroup profiles and CNAs with their cfDNA counterparts, while discordant cases likely resulted from normal tissue contamination. In summary, we demonstrated that WT ctDNA is robustly detected in cfDNA through measurement of DNAm profiles. As prognostic information such as DNAm subgroup can be derived from these data, this assay is well positioned for translation into clinical molecular diagnostics. We are currently optimizing this assay for early detection of WT in patients with predisposition and for monitoring of treatment response.