MYC amplification at diagnosis drives therapy-induced hypermutation of recurrent glioma

Jiguang Wang,Quanhua Mu,Ruichao Chai,Hanjie Liu,Yingxi Yang,Zheng Zhao,Ming Hong Lui,Zhaoshi Bao,Dong Song,Biaobin Jiang,Jason Sa,Hee Jin Cho,Yuzhou Chang,Kaitlin Hao Yi Chan,Danson Shek Chun Loi,Sindy Sing Ting Tam,Aden Ka Yin Chan,Angela Wu,Wai San Poon,H.K. Ng,Danny Chan,Antonio Iavarone,Do-Hyun Nam,Tao Jiang

DOI: https://doi.org/10.21203/rs.3.rs-138020/v1

2021-01-05

Abstract:Abstract Clonal evolution drives cancer progression and therapeutic resistance 1-2 . Recent longitudinal analyses revealed divergent clonal dynamics in adult diffuse gliomas 3–11 . However, the early genomic and epigenomic factors that steer post-treatment molecular trajectories remain unknown. To track evolutionary predictors, we analyzed sequencing and clinical data of matched initial-recurrent tumor pairs from 511 adult diffuse glioma patients. Using machine learning we developed methods capable of predicting grade progression and hypermutation from tumor characteristics at diagnosis. Strikingly, MYC copy number gain in initial tumors emerged as a key factor predicting development of hypermutation under temozolomide (TMZ) treatment. The driving role of MYC in TMZ-associated hypermutagenesis has been experimentally validated in a model of TMZ-induced hypermutator using both patient-derived gliomaspheres and established glioma cell lines. Subsequent studies showed that c-Myc binding to open chromatin and transcriptionally active regions increases the vulnerability of genomic regions to TMZ-induced mutagenesis. Consequently, MYC target genes, including the key mismatch repair genes, develop loss-of-function mutations, thus triggering the hypermutation process. This study reveals MYC as an early predictor of cancer evolution and provides a machine learning platform for predicting cancer dynamics to improve patient management.