Abstract
BACKGROUND: Genetic predisposition is particularly common in children with the kidney cancer, Wilms tumor. In 10% of these children, this manifests as a family history of Wilms tumor or bilateral disease. The frequency and spectrum of underlying changes have not been systematically investigated. METHODS: We analyzed 129 children with suspected Wilms tumor predisposition, 20 familial cases, and 109 children with bilateral disease, enrolled over 30 years in the German SIOP93-01/GPOH and SIOP2001 studies. We used whole exome, whole genome, and targeted DNA sequencing, together with MLPA and targeted methylation assays on tumor, blood, and normal kidney to determine predisposing changes. RESULTS: Predisposing variants were identified in 117/129 children, comprising DNA variants (57%) and epigenetic changes (34%). Most children had predisposition variants in genes previously implicated in Wilms tumor: most prominently WT1 (n = 35) and less frequently TRIM28, REST, DIS3L2, CTR9, DICER1, CDC73, and NONO. Nine children carried germline mutations in cancer predisposition genes not considered Wilms tumor predisposition genes, such as CHEK2, CDKN2A, BLM, BRCA2, STK11, and FMN2. Predisposition via epigenetic BWS-IC1 alterations occurred as early somatic events, reflected by partial (mosaic) loss of imprinting or loss of heterozygosity at the IGF2/H19 locus in normal kidney or blood. These patients rarely had a clinical diagnosis of Beckwith-Wiedemann syndrome (BWS). Especially WT1-driven tumors follow a stereotypical pathway of germline WT1 mutations becoming homozygous in renal precursor lesions through 11p LOH, which concomitantly activates imprinted IGF2 expression, with subsequent WNT pathway activation leading to tumor growth. There is a high rate of multicentric tumors, which may have previously been missed in unilateral tumors. While Wilms tumor predisposition genes relied on somatic inactivation of the second allele, this was different for general cancer predisposition genes. The latter cases were often associated with additional oncogenic alterations, similar to tumors with epigenetic predisposition. CONCLUSIONS: We identified two main mechanisms of Wilms tumor predisposition: either germline genetic alterations of Wilms tumor and, less frequently, general cancer genes; or postzygotic mosaic imprinting defects activating IGF2. These findings inform future genetic screening and risk assessment of affected children and lend support to liquid biopsy screening for enhanced therapeutic stratification.