Fig. 5: Combinatorial cyclin D1, cyclin E1, p16, and Rb biomarker signature predicts response to BLU-222+ribociclib in breast PDX models.

a Kendall’s tau correlation between BLU-222 + ribociclib-induced TGI in PDX models and gene expression for a single biomarker of interest. b Pearson’s correlation between observed BLU-222+ribociclib-induced TGI and predicted BLU-222 + ribociclib TGI estimated from a weighted, multivariate biomarker signature developed from a generalized linear model. c Differential observed TGI between predicted BLU-222 + ribociclib responder (predicted TGI ≥ 70%) and predicted BLU-222+ribociclib nonresponder (predicted TGI < 70%). d Receiver operator curves for the prediction models for each single biomarker and combinations of CCND1, CCNE1, RB1, and CDKN2A. e CDK2i, ribociclib, or the combination in a predicted CDK2i + CDK4/6i responder PDX model. PDX.003.014 is an ER+ breast cancer model derived from a patient who had progressed on CDK4/6i. f Antitumor activity in a predicted CDK2i + CDK4/6i combination nonresponder PDX model. PDX.003.234 is an ER+ breast cancer model derived from a CDK4/6i-naïve patient previously treated with chemotherapy. BLU-222 studies dosed as follows: BLU-222 60 mg/kg BID, ribociclib 50 mg/kg QD, BLU-222+ribociclib 60 mg/kg BID/50 mg/kg QD. BLU-2256 studies dosed as follows: BLU-2256 10 mg/kg QD, ribociclib 75 mg/kg QD, BLU-2256+ribociclib 10 mg/kg QD/75 mg/kg QD. Mean tumor volume and SEM are plotted, n = 6-8 per group. TGI from vehicle is noted: 2-way ANOVA, **P < 0.01, ***P < 0.001, ns = not significant.