Ma Fei; Wen Tingyu; Wang Jun; Xie Guotong; Yi Zongbi
Competition Sponsor: Chinese Academy of Medical Sciences
Although the successful development of antibody drugs and small-molecule tyrosinase inhibitors (TKI) targeting HER2 has significantly improved the survival time of HER2-positive breast cancer patients, most patients will eventually develop primary or acquired resistance during long-term treatment. Preliminary studies of our group suggested that HER2-positive breast cancer is often accompanied by CDK12 co-amplification, which can promote tumor growth by activating the PI3K/AKT signaling pathway. HER2/CDK12 co-amplified patients are more resistant to HER2 TKI lapatinib than CDK12 non-amplified patients, and they are associated with poor median progression-free survival. Meanwhile, dual inhibition of HER2/CDK12 will prominently benefit the outcomes of patients with HER2-positive breast cancer by sensitizing or re-sensitizing the tumors to anti-HER2 TKIs treatment. Since there are no approved CDK12 inhibitors, this study intends to construct a graph neural network model for drug molecule representation through the self-supervised pre-training method of graph neural network, assist in the screening of new CDK12 small molecule inhibitors, and test its anti-tumor activity in vivo and in vitro.