Nelson Guerreiro, Sebastien Jeay, Stephane Ferretti, Claire Fabre, Astrid Jullion and Christophe Meille
Novartis, Basel, Switzerland
Objectives: The purpose of this work was to investigate the predictability from preclinical to clinical of a model-based analysis capturing drug action on tumor size kinetics, and its application in the clinical development of HDM201 for exposure-efficacy relationships, optimization of dose regimen selection, and predicting clinical outcomes.
Methods: Plasma concentrations, and tumor size data were collected from preclinical SJSA-1 (a wild-type p53 and MDM2-amplified osteosarcoma cell line known to be sensitive to MDM2 inhibition and p53 reactivation) tumor bearing rats and from the ongoing Phase I study of HDM201 in patients with TP53 wild-type solid tumors [1]. Population PKPD modeling was used to investigate the pharmacokinetics of HDM201 linked to the time-course of the anti-tumor effect in both the SJSA-1 xenograft rat, as well as in clinic as a function of dose and treatment schedule. In patients, tumor size assessments were conducted, with longitudinal sum of largest diameter (SLD) change post-dose relative to baseline. This data was obtained from the on-going first-in-human CHDM201X2101 trial. This dataset included 85 patients randomized to receive NVP-HDM201 every 3 weeks (Q3W; n = 26; 12.5 to 350 mg), QW (D1, D8 of a 4-wk cycle; n = 20; 120 to 200 mg), QD for 14 consecutive days (4-wk cycle; n = 20; 1 to 20 mg), and QD for 7 consecutive days (4-wk cycle; n = 19; 15 to 25 mg). Firstly, preclinical tumor growth inhibition (TGI) modeling was applied to understand scheduling requirements for antitumor activity in clinic, and secondly, TGI modeling was applied on clinical data to validate the translational PKPD modeling approach [2].
Results: The PKPD relationship among HDM201 exposure and TGI in the preclinical SJSA-1 xenograft model was well characterized. A signal distribution model with a saturating growth best described the time-course of longitudinal tumor growth as a function of dose and treatment schedule [3]. Incorporation of a resistance component was necessary to characterize tumor regrowth in the presence of HDM201 across multiple treatment cycles. The results from preclinical PKPD modeling did not identify a schedule dependency with tumor growth inhibition, allowing for the derivation of an average drug concentration of 44 ng/ml as a requirement for tumor stasis in SJSA-1 per cycle. This value was further adjusted to account for the tumor resistance component in which the tumor cells were less responsive to treatment, as well for rat to human difference in free drug fraction. A structurally similar TGI model but with exponential growth was then utilized for PKPD modeling of the clinical time course of longitudinal tumor SLD data from patients treated with escalating doses of HDM201 and at different treatment schedules. PKPD modeling of clinical data confirmed human tumor size modeling is in line with preclinical modeling projection with a derived average drug concentration of 68 ng/mL required for tumor stasis. These projected concentrations appeared consistent with clinical responses.
Conclusions: For both preclinical and clinical data, tumor size measurements were adequately described by a single consolidated model structure that captured continuous tumor size with a combination of growth regression and resistance terms. Both preclinical and clinical TGI models indicated that average concentration per cycle was a predictor of tumor size response, demonstrating efficacy to be independent on schedule, and allowing for schedule adjustment to identify optimal treatment strategies. This work demonstrates that PKPD modeling can be used for predictive translational pharmacology from nonclinical to clinical development, helping guide decisions on dose escalation and dosing regimen selection for HDM201 in clinic.
References:
[1] Hyman, D. et al. Dose- and regimen-finding phase I study of NVP-HDM201 in patients (pts) with TP53 wild-type (wt) advanced tumors; European Journal of Cancer. 2016; 69(1): S128–S129.
[2] Mager D, Jusko W. Development of Translational Pharmacokinetic–Pharmacodynamic Models. Clinical pharmacology and therapeutics. 2008; 83(6):909-912.
[3] Yang J1, Mager DE, Straubinger RM. Comparison of two pharmacodynamic transduction models for the analysis of tumor therapeutic responses in model systems. American Association of Pharmaceutical Scientists J. 2010; 12(1):1-10.
Reference: PAGE 27 (2018) Abstr 8633 [www.page-meeting.org/?abstract=8633]
Poster: Drug/Disease Modelling - Oncology