GE-21 * DRASTIC GENOMIC DIVERGENCE OF RECURRENT MEDULLOBLASTOMA INVALIDATES TARGETED THERAPIES DISCOVERED AT DIAGNOSIS

Recurrence of the childhood brain cancer medulloblastoma is almost universally fatal. The current strategy for clinical trials is to test novel therapies at the time of recurrence, using targets discovered at diagnosis. To define the relationship between naïve and post-treatment medulloblastoma, we developed a functional genomic transposon-driven mouse-model. Deep sequencing of the commonly inserted genes in diagnostic and recurrent tumors revealed dramatic divergence post-therapy, with recurrences demonstrating a convergence on the TP53 pathway. Loss of function insertions in Tp53 are clonal in the recurrence, result in loss of expression of the Tp53 target gene P21, and arise through selection of minor subclones present before therapy. The majority of insertions identified at diagnosis are no longer present at recurrence, and therefore represent futile targets for therapy. To validate these findings, we studied 36 pairs of human primary/recurrent tumors by whole genome sequencing. When comparing SNVs, indels or structural aberrations, we found minimal genetic overlap between the dominant clones at diagnosis versus recurrence. Recurrent tumors have more genomic aberrations than their paired diagnostic sample, and commonly display alterations in the TP53 (SHH), or the TP53 and TGF-β pathways (Group 3/4 tumors). Clonal dynamics between primary and recurrent tumors was measured using ultra-deep sequencing (1,500X). Overall, recurrent tumors share little genetic identity with their diagnostic tumor, arise through clonal selection, target specific signaling pathways to avoid cell death, and assume clinical dominance after therapy. Our findings undermine the key assumption that diagnostic medulloblastoma samples provide rational targets for therapy in recurrent disease, and provide a simple reason for the failure of prior trials of targeted therapy. Repeated genetic convergence on specific signaling pathways at the time of recurrence suggests that it may be possible to develop ‘anticipatory therapy’ delivered up front to prevent the later emergence of minor resistant clones driving recurrence.