Integrative Analyses of Primary and Relapse Tumor Samples Identifies Genomic Changes During Cancer Progression

Abstract
The rise of next-generation sequencing approaches to comprehensively assess and interpret the molecular genetic tumor landscape has fueled efforts to tailor medical care for individual cancer patients. While progress in therapies has been significant for many cancer types, death of cancer results, in the vast majority of cases, not from the initial disease manifestation, but from metastasis or recurrence. This helds also true for neuroblastoma, the most common extracranial tumor of childhood. In this particular tumor, therapies are quite successful in reducing the initial tumor burden, but relapses cannot be treated with curative intent. In order to identify specific genetic aberrations at recurrence of disease, we performed whole-exome sequencing of neuroblastoma at relapse and compared their mutational profiles to the corresponding primary tumors and constitutional DNAs. These data were integrated with arrayCGH, methylation and mRNA expression profiles and validated using preclinical NB models. Global allele frequency analyses suggested an enrichment of pre-existing mutations during disease progression. None of the patients presented with acquired mutations in previously identified NB driver genes at relapse. Exome sequencing and phylogenetic reconstruction using multiple relapses from a single patient in our cohort provided unique insight into molecular genetic evolution in neuroblastoma. Clonal branches evolved from the primary tumor gave rise to independent relapses colonizing different sites. To assess not only spatial but temporal tumor evolution, we also analyzed allelic frequencies of mutated genes deduced from both exome sequencing and amplicon resequencing in primary-relapse tumor pairs in our patient cohort. Results from both techniques suggest that most relapse-specific mutations evolve either de novo or from a rare subpopulation present in the primary tumor. From a global view, clonal selection during disease progression and relapse can occur independently at various time points and metastatic localizations. These findings underpin the need for multiregional sequencing when more than one relapse tumor is detected in a single patient in order to define promising individualized therapy approaches.