Abstract A007: Mosaicism in cancer-susceptibility genes in early-onset childhood cancer

Abstract Background: The prevalence and contribution of postzygotic mosaic variants to early childhood cancers remain poorly understood compared to constitutional germline variants. Mosaic variants are challenging to identify in peripheral blood due to low allelic fractions, tissue specificity, and confounding effects of adult clonal hematopoiesis. Traditional approaches often struggle to distinguish these from later-life somatic events or tumor-derived sequences, underestimating their clinical importance. Leveraging newborn dried blood spots (DBS) provides a powerful, unbiased platform to systematically assess mosaic variant prevalence in cancer predisposition. Methods: Children born in Michigan (1987-2020) who developed solid/CNS tumors before age 8 were identified via the Michigan Cancer Registry and linked to DBS biobank records using birth registry data. DNA was extracted from 3.2 mm DBS punches and underwent targeted next-generation sequencing for a panel of previously reported 11 genes (RB1, TP53, WT1, RET, SMARCB1, PTCH1, SUFU, APC, DICER1, ALK, PHOX2B) and NF1 with a median 500X depth. Deleterious single-nucleotide variants and indels were called based on ClinVar and predicted high-impact consequences. To identify mosaic variants, variant allele fraction (VAF) distributions for all pathogenic variants (PVs) were analyzed using a Gaussian mixture model (GMM) to separate presumed germline and mosaic variants. The threshold for putative mosaicism was set at a VAF of 0.379 based on the GMM, which also corresponded to 3 standard deviations below the mean VAF for presumed heterozygous germline variants. Results: A total of 1,943 DBS samples from infants with solid/CNS tumors were identified, including 76 diagnosed at birth. The most frequent histologies were neuroblastoma (n=451), Wilms tumor (n=281), and retinoblastoma (n=169). Among 174 PVs detected, 26 (14.9%) were mosaic – mainly in RB1 (n=15), as well as NF1 (n=5), TP53 (n=4), SMARCB1 (n=1), and APC (n=1). Mosaicism was evaluated across all genes, but clinical correlations were restricted to RB1 due to sample size. For retinoblastoma, 62 cases had a PV, including 16 deemed putatively mosaic (VAF range 11.0–37.7%, median 26.4%). Mosaic variants were seen in 8% of bilateral and 8.8% of unilateral cases. Lower VAF was associated with later disease onset (adjusted R2=0.37, p<0.001). No mosaic variants were observed among patients with advanced disease (n=15) or early mortality (n=16). Conclusion: In a large cohort of early-onset pediatric cancer patients with DBS evaluated for PVs in 12 established cancer predisposition genes, 14.5% of PV-carriers had a single putative mosaic variant detected. Mosaic RB1 PVs were associated with an earlier age of disease onset and more favorable clinical features in retinoblastoma, although definitive conclusions regarding disease severity are limited by the cohort size. Future work will apply advanced postzygotic mosaic variant identification methods and expand the panel of cancer susceptibility genes to improve the detection and interpretation of mosaic PVs. Citation Format: Sneh Patel, Emily Iannetta, Nicole Taikeff, Abigail Cinelli, Andy Bhattacharjee, Richard Parad, Lisa Diller, Riaz Gillani. Mosaicism in cancer-susceptibility genes in early-onset childhood cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Discovery and Innovation in Pediatric Cancer— From Biology to Breakthrough Therapies; 2025 Sep 25-28; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2025;85(18_Suppl_2):Abstract nr A007.