Using CDKN2A loss in the context of wildtype TP53 to predict sensitivity for the MDM2 inhibitor milademetan.

Abstract

3136 Background: MDM2 is an E3 ubiquitin ligase that plays a critical role in the degradation of the tumor suppressor p53. Milademetan (RAIN-32) is an orally available, small molecule inhibitor of MDM2 that disrupts the MDM2-p53 complex thereby restoring p53 activity. Approximately 50% of tumors harbor wildtype (WT) TP53 and thus may be susceptible to strategies that reactivate p53. The CDKN2A gene is altered in more than 15% of all tumors (TCGA PanCancer Atlas) and encodes two proteins, p14ARF and p16, which are inhibitors of p53 and cyclin dependent kinases, respectively. Given the role of p14ARF in regulating the MDM2-p53 pathway, we investigated the use of CDKN2A loss in the context of WT TP53 as a strategy for selection of patients who might benefit from milademetan. Methods: N/A. Results: We evaluated the sensitivity of 215 cancer cell lines to milademetan treatment (Ishizawa et al., 2018) by CDKN2A and TP53 status. The median IC50 of CDKN2A homozygous (HZ) loss vs. non-HZ loss was 8,620 vs. 10,000 nM. However, when we assessed CDKN2A HZ loss with WT TP53 versus mutant TP53 the median IC50 was 79.5 vs. 10,000 nM demonstrating that the use of both CDKN2A and TP53 was better able to discriminate sensitive vs. resistant cell lines. To validate these in vitro findings, we tested milademetan in 5 xenograft models with CDKN2A HZ loss and WT TP53, all of which demonstrated tumor growth inhibition with milademetan. As suppression of p53 activity by MDM2 amplification (Kato et al. 2017) or CDKN2A loss (Adib et al. 2021) has been associated with resistance to immune checkpoint inhibitors (ICI), we also tested the combination of anti-PD1 with milademetan in the colon-26 syngeneic model ( CDKN2A HZ loss) and observed a significant enhancement in tumor growth inhibition compared to milademetan or anti-PD1 alone. Based on the differential sensitivity to milademetan using both CDKN2A loss and WT TP53 status we evaluated TCGA Pan-Cancer Atlas data to estimate the frequency of these genetic co-alterations. Among solid tumors types the most frequent percentage of these co-alterations included glioblastoma, mesothelioma, melanoma, bladder, sarcoma, pancreatic and NSCLC. Overall, the percentage of all tumors with co-alteration of CDKN2A HZ loss and WT TP53 was 6.2%. Patients with CDKN2A HZ loss had a significantly worse overall survival than those without CDKN2A HZ loss (median OS of 29.7 vs. 97.4 months, p < 0.0001), and this was maintained when accounting for tumor type in multivariate analysis (p < 0.0001). Conclusions: Milademetan showed evidence of preclinical anti-tumor activity across multiple tumor types with CDKN2A loss and WT TP53. In vivo data supported potential synergy of milademetan with an ICI in this genetic subset. A clinical trial evaluating the safety and efficacy of milademetan plus atezolizumab in advanced solid tumors with CDKN2A HZ loss and WT TP53 (MANTRA-4) is planned.