Exposure-Response Analysis of Tovorafenib (DAY101) Relating Objective Response to Exposure, Patient- and Disease Characteristics
John Maringwa1, Anna Largajolli1, YuWei-Lin1, Katrina Hui1, Yin Edwards1, James Smith1, Eleni Venetsanakos2, Elly Barry2, Samuel C Blackman2, S. Y. Amy Cheung1
(1) Certara, Integrated Drug Development, Princeton, NJ, USA, (2) Day One Biopharmaceuticals, Inc, USA
Introduction: Tovorafenib (DAY101) is an oral, highly selective, CNS-penetrant, Type II pan RAF kinase inhibitor being developed for the treatment of pediatric low-grade glioma and solid tumors harboring RAF alterations. In contrast to approved Type I RAF inhibitors, tovorafenib inhibits both wild-type BRAF and CRAF/RAF kinases and, hyperactivated signaling resulting from B/CRAF fusions, including the KIAA1549: BRAF fusion[1]. This analysis focused on exploring the relation between the probability (prob) of being a treatment responder and tovorafenib exposure and other potential factors including demographics and patient (pt)/disease characteristics using data from C28001/NCT01425008 in adult pts with advanced solid tumors[2].
Objectives: The analysis aimed to assess the impact of tovorafenib’s exposure, pt, and disease characteristics on response as assessed by the Response Evaluation Criteria in Solid Tumors (RECIST) definition of responder vs non-responder, in order to confirm recommended Ph2 dose (RP2D) and support future study design.
Methods: The focus was on the tumor objective response (OR) as assessed via the RECIST criteria (complete and partial responder (responder) vs stable and progressive disease (non-responder)). A developed PopPK model for tovorafenib was used to predict individual-specific tovorafenib exposure metrics including the steady-state area the curve (AUC), min (Cmin) or max (Cmax) drug concentration, and time-averaged exposure, which were subsequently related to OR in an exposure-response (ER) analysis. Graphical exploration of response in relation to exposure and other potential factors eg, mutation status and previous treatment, informed on trends in the data. The prob of being a responder was related to exposure and patient/disease characteristics formally in a logistic regression model. The exposure metric with the strongest impact on the response at univariate assessment (smallest p-value) formed the basis for covariate evaluation. No covariate effects were assessed in the absence of a univariate statistically significant exposure effect (p-value<0.05).
Results:
Data from 135 pts in the dose-escalation+dose-expansion phases of study C28001 was available. The RECIST response was not assessed in 14 pts, hence data from 121 pts was explored. In ~43% (52/121) of the pts, mutation status was unknown. PK information was not available in 22 pts, hence the models were developed based on data from 99 pts.
The response was somewhat larger in pts with a mutation and in previously untreated pts compared to other pts. The observed response in pts with a BRAF mutation, either naïve or previously treated (9/22 pts) was 0.41 (95% CI: 0.25; 0.59) compared to 0.030 (95% CI: 0.012, 0.076) in other pts (3/99). Based on the univariate assessment using data available in C28001, Cmin was identified as the best predictor of response (p=0.008) to support the RP2D. No statistically significant effect of demographics on response was identified in the presence of Cmin. Pts with a BRAF mutation and no previous treatment (naïve) were found to have a statistically significantly larger response compared to all other pts (p=0.02). Considering patients with a BRAF mutation (naïve or previously treated), the difference from all other pts was more pronounced (p=0.007). Simulations confirmed the increase in response with ↑ Cmin notably with more frequent dosing of Q2D to Q4D. Across all pts, the simulated response at the RP2D of 600mg QW of 0.040 (95% CI: 0.010-0.080) was somewhat lower than the response for 200mg Q2D at 0.13 (95%CI: 0.06; 0.20), noting the relatively wide confidence intervals
Conclusions: The prob of being a treatment responder as assessed by the RECIST criteria was identified to increase with ↑ minimum tovorafenib concentration and the simulated response for the RP2D did not seem to differ much from that based on more frequent dosing. Pts with a BRAF mutation, regardless of previous treatment status, appeared to have a larger prob of being a responder compared to other pts. Demographics did not appear to impact response once exposure was accounted for. The number of pts with a BRAF mutation was relatively small and hence these results need cautious interpretation. The marginal increase in efficacy/OR for more frequent doses was outweighed by an increase in AEs of G3 rash[3] along with findings from the PopPK-longitudinal safety analysis[4], supporting 600mg QW as the RP2D.
References:
[1] Sun Y, Alberta JA, Pilarz C, Calligaris D, Chadwick EJ, Ramkissoon SH, et al. A brain penetrant RAF dimer antagonist for the noncanonical BRAF oncoprotein of pediatric low-grade astrocytomas. Neuro Oncol. 2017 Jun 1;19(6):774-785.
[2] Clinical Study Report C28001: An Open-Label, Phase 1, Dose Escalation Study of MLN2480 in Patients with Relapsed or Refractory Solid Tumors Followed by a Dose Expansion Phase in Patients With Metastatic Melanoma. Millennium Pharmaceuticals, Inc.31 May 2018
[3] Edwards AY, Largajolli A, Lin, YW et.al Exposure-safety analysis of Tovorafenib (DAY101), an investigational oral pan-RAF kinase inhibitor, in oncology patients with solid tumors. 2022 PAGE 2022 Abstract.
[4] Hui K, Lin, YW, Largajolli A et. al Utilizing model informed drug development to support study design for Tovorafenib (DAY101), a novel, type II, pan RAF kinase inhibitor: a focus on longitudinal safety markers. PAGE 2022 Abstract.