Somatic Copy Number Alteration in Circulating Tumor DNA for Monitoring of Pediatric Patients with Cancer

Abstract

Pediatric tumors share few recurrent mutations and are instead characterized by copy number alterations (CNAs). The cell-free DNA (cfDNA) is a prominent source for the detection of cancer-specific biomarkers in plasma. We profiled CNAs in the tumor tissues for further evaluation of alterations in 1q, MYCN and 17p in the circulating tumor DNA (ctDNA) in the peripheral blood at diagnosis and follow-up using digital PCR. We report that among the different kinds of tumors (neuroblastoma, Wilms tumor, Ewing sarcoma, rhabdomyosarcoma, leiomyosarcoma, osteosarcoma and benign teratoma), neuroblastoma presented the greatest amount of cfDNA, in correlation with tumor volume. Considering all tumors, cfDNA levels correlated with tumor stage, metastasis at diagnosis and metastasis developed during therapy. In the tumor tissue, at least one CNA (at CRABP2, TP53, surrogate markers for 1q and 17p, respectively, and MYCN) was observed in 89% of patients. At diagnosis, CNAs levels were concordant between tumor and ctDNA in 56% of the cases, and for the remaining 44%, 91.4% of the CNAs were present only in cfDNA and 8.6% only in the tumor. Within the cfDNA, we observed that 46% and 23% of the patients had MYCN and 1q gain, respectively. The use of specific CNAs as targets for liquid biopsy in pediatric patients with cancer can improve diagnosis and should be considered for monitoring of the disease response.

Keywords: 1q; MYCN; cell-free DNA; circulating tumor DNA; copy number alterations; digital PCR.

Figure 1

Overview of the study. Tumor tissue and whole blood were collected from pediatric patients diagnosed with cancer. Tumor tissue DNA was extracted and underwent either bisulfite conversion followed by hybridization in the EPIC BeadChip Methylation Arrays or SALSA MLPA assay. CNAs were obtained from either EPIC or MLPA data. Tumors with 1q gain, MYCN gain or TP53 gain/loss were selected for dPCR. Plasma was used for cfDNA extraction and for CNAs’ evaluation. WT, Wilms tumor; NB, neuroblastoma; ES, Ewing sarcoma; RMS, rhabdomyosarcoma; LMS, leiomyosarcoma; OS, osteosarcoma; and BT, benign teratoma.

Figure 2

cfDNA levels at diagnosis were evaluated regarding the tumor characteristics. Statistically significant comparison where indicated by “*”. (A): cfDNA ng per mL plasma across tumor histologies (p = 0.0201, Kruskal–Wallis test). (B): cfDNA ng per mL plasma correlated with tumor stage (I/II–IV: p = 0.0227, III–IV: p = 0.0303). (C): Metastatic patients (n = 17) had higher levels of cfDNA at diagnosis compared with nonmetastatic patients (n = 18, p = 0.0153, Mann–Whitney U test). (D): Patients who developed metastasis during treatment (n = 11) also had higher levels of cfDNA compared to those who did not (n = 19, p = 0.0264, Mann–Whitney U test).

Figure 3

Correlation between cfDNA ng per mL plasma and tumor volume (Spearman’s correlation and linear regression). (A): cfDNA ng per mL plasma did not correlate with volume when all tumors were used. Considering each tumor type, (B): there was a correlation for neuroblastoma but not for (C): Wilms tumor or (D): sarcomas (Ewing sarcoma, embryonal rhabdomyosarcoma, osteosarcoma and leiomyosarcoma).

Figure 4

MYCN, CRABP2 (1q) and TP53 (17p) CN for tumor and peripheral blood. NB, neuroblastoma; WT, Wilms tumor; SARC, sarcomas; and BT, benign teratoma. Numbers refer to copy number of the target regions relative to RNAseP. The parameters for gain and loss are also displayed.

Figure 5

CtDNA CNA in follow-up samples from seven cases. CNAs in the tumor samples at diagnosis were characterized by methylome or MLPA. Genes evaluated in the ctDNA are labeled in blue. The ctDNA levels at diagnosis and in follow-up samples (graphs on the right) were determined by dPCR. The numbers contained within the arrows represent the time (in months) between the analyses performed at diagnosis and at follow-up. Chemo, chemotherapy.