Purpose: Recent data show that genomic changes are frequently subclonal in neuroblastoma and may occur in an undetectable fraction in a primary tumor. Alternative analytic strategies need to be evaluated to obtain a more complete picture of the genomic landscape and to circumvent an often insufficient amount of tumor tissue. Isolating cell-free tumor DNA from more easily accessible patient biomaterial could surmount these challenges. We investigated the feasibility of this approach to use cell-free peripheral blood (PB) and bone marrow (BM) plasma for neuroblastoma tumor genome analyses.
Experimental Design: Cell-free DNA (cfDNA) isolated either from BM or PB or both from 14 stage M neuroblastoma patients were analyzed by ultra-high-density SNP arrays (23) and by low-coverage whole genome sequencing (WGS) on an Illumina HiSeq2500 (6/23) and compared with genomic data from the corresponding tumor or GD2-enriched disseminated tumor cells from BM (BM-DTCs).
Results: Genomic aberrations were unambiguously detectable in cfDNA samples using SNP array analysis (17/23) and WGS (6/6) and included typical segmental chromosomal aberrations (SCA) and MYCN amplifications as well as atypical SCAs, amplicons and micro-deletions. A high concordance between genomic aberrations found in patient-matched cfDNAs from BM plasma and BM-DTCs (3/3) was found. However, discordances were detected between BM and PB cfDNA samples obtained at the same time point (3/6).
Conclusions: Cell-free tumor DNA can serve as an additional source for tumor genome analysis allowing the identification of genetically distinct tumor clones which are otherwise difficult to detect in the primary tumor. Our findings contradict the widely accepted view that all bodily fluids present an identical picture of tumor genetics. Comparing datasets from different intra-patient locations will produce a more complete representation of tumor heterogeneity, thus improving our understanding of tumor dynamics and progression in individual patients.