Pediatric solid tumors encompass a heterogeneous group of rare malignancies that constitute approximately 40% of all childhood cancers1. Among them, soft-tissue tumors including embryonal tumors, germ cell tumors, and renal tumors account for approximately 10% of cases, whereas bone and soft-tissue sarcomas comprise ~20% of all cases2,3,4. The cytogenetic and molecular genetic characterization of pediatric solid tumors is used clinically to aid in diagnosis and determine prognosis, and in certain cases, guide treatment5. Identification of germline copy number alterations (CNAs) and sequence variants in cancer predisposition genes are also important for patient management, specifically to address the risk for second malignancies in the patient or cancer in a family member.

Liquid biopsy (LB) assays for pediatric patients with solid tumors have the potential to transform patient care by providing a less invasive alternative to diagnostic biopsies for identifying genomic aberrations that can inform diagnosis, risk stratification, and therapeutic options as well as enable earlier detection of disease progression compared with conventional radiographic imaging6,7,8,9,10,11. The clinical effectiveness of plasma-based LB approaches, however, depends on optimizing the detection of circulating tumor DNA (ctDNA) fragments and distinguishing them from cfDNA derived predominantly from hematopoietic cells12,13.

Recent advances in LB test development have focused primarily on adult-type malignant epithelial tumors defined by a spectrum of recurrent dominant activating mutations14,15,16. Pediatric tumors, however, are more often characterized by CNAs, loss of tumor suppressor genes, epigenetic modifications, and large-scale structural rearrangements with breakpoints that are variable and located in the intronic regions of the genome5,17,18. Furthermore, the incidence of pediatric solid tumors is low, and the number of histologic, genomic, and clinical subtypes is large, suggesting that the combined use of a pan-cancer assay and a target-specific approach may be better suited for clinical applications. Most importantly, the clinical development of LB assays for pediatric solid tumor patients has been limited by the requirement for relatively large volumes of blood, urine, or cerebrospinal fluid (CSF) to isolate sufficient amounts of cancer-derived nucleic acids for analysis.

The feasibility of employing NGS and droplet digital PCR-based assays utilizing cfDNA derived from CSF, plasma, or the aqueous humor of the eye for pediatric central nervous system (CNS) tumors, solid tumors, or retinoblastoma, respectively, has recently been described10,19,20,21,22,23,24,25,26,27. In the diagnostic setting, ultra-low-pass WGS (ULP-WGS) analysis of plasma ctDNA was effectively used to distinguish malignant peripheral nerve sheath tumors (MPNST) from the benign lesion, plexiform neurofibroma (PN), in patients with the neurofibromatosis type 1 (NF1) cancer predisposition syndrome28. By using in silico enrichment of short cfDNA fragments and copy number analysis, MPNST samples were found to be enriched with ctDNA when compared to PN, and treatment response was correlated with the ctDNA-derived estimate of tumor burden28. To assess response to treatment, Liu et al. assessed measurable residual disease (MRD) in 123 children with medulloblastoma using LP-WGS of cfDNA derived from CSF10. The presence of MRD in the CSF was found to be associated with a higher risk of relapse. The presence of ctDNA in plasma has been proposed as a prognostic biomarker in pediatric solid tumors (reviewed in ref. 7), and targeted detection of EWSR1 and FOXO1 fusions in Ewing sarcoma and alveolar rhabdomyosarcoma (ARMS) has been demonstrated in small cohorts of patients29,30.

The Precision Medicine Program in Pediatric and Adolescent Patients with Recurrent Malignancies (MAPPYACTS) employed whole exome sequencing (WES) of tumor tissue and cfDNA from plasma to identify targeted therapies in patients with relapsed/recurrent non-CNS solid tumors31. Notably, 57% of the somatic SNVs were detected in both the tumor and the cfDNA, whereas 31% were specific to the tumor and 11% were specific to the cfDNA, reflecting tumor heterogeneity, or possibly technical limitations of the assays31. Significantly more mutations were detected in patients with metastatic (66%) vs localized (47%) disease.

The studies published to date have demonstrated the feasibility of using cfDNA from plasma to detect CNAs, sequence variants, and/or gene fusions, but have focused on disease-specific cohorts, or patients with recurrent/refractory disease. The aim of the present study was to evaluate the potential clinical applicability of a plasma-based LB assay for newly diagnosed and relapsed pediatric patients with a variety of solid tumors. In this approach, LP-WGS was performed using cfDNA to detect CNAs and mutations. Targeted panel-based sequencing of cfDNA was also performed to identify specific gene fusions. Clinical validation of these assays will allow for their implementation in a prospective pan-cancer setting as an aid in diagnosis and to monitor response to therapy.


Patient clinical characteristics

Patients were deemed eligible for this study if they had a newly diagnosed or recurrent malignant bone or soft-tissue sarcoma or germ cell, hepatic, thyroid, or renal tumor. Written informed consent (and assent when appropriate) was obtained from all patients, their parents, or a legal guardian to participate in this study under a Children’s Hospital Los Angeles Institutional Review Board-approved protocol (CHLA-19-00146). Written informed consent was also obtained for non-oncologic controls, their parents, or a legal guardian to participate in this study under a Children’s Hospital Los Angeles Institutional Review Board-approved protocol (CHLA-19-00230).

One hundred and forty-three samples from 73 eligible patients and 19 non-oncologic controls were analyzed. The patients were diagnosed as having sarcoma (n = 44), renal tumor (n = 12), hepatic tumor (n = 5), malignant germ cell tumor (n = 10), or thyroid carcinoma (n = 2). The age of the patients ranged from six months to 28 years (median 12 years) and there were 37 males and 36 females (Table 1). The median age of the controls was 11 years (Supplemental Table 1). A median of two samples were analyzed per patient (range, 1–6). Forty-eight of the patients were enrolled at diagnosis, although 11 had received chemotherapy and/or undergone surgery prior to obtaining a blood sample for the study. Thirty-five of the patients presented with localized disease. Twenty-five patients were enrolled at the time of a recurrence (two with local and 23 with distant disease). Six patients had received chemotherapy and/or had undergone surgery for relapse prior to enrollment. None of the patients received prior radiation therapy. Two patients had surgery and chemotherapy prior to study enrollment (Table 1 and Supplemental Table 1). Samples were collected at the time of initial diagnosis or recurrence, during therapy, at the time of recurrence/progression, where applicable, and during follow-up (i.e., after treatment).