Anson K. Abraham (1), Doug C. Wilson (2), Glareh Azadi (3), Gulesi Ayanoglu (4), Charo Garrido (5), Rachel Altura (6)
(1) Quantitative Pharmacology & Pharmacometrics, PPDM, (Upper Gwynedd, PA); (2) Profiling and Expression, (Palo Alto, CA); (3) ADME, PPDM, (Palo Alto, CA); (4) Bioanalytical, PPDM, (Palo Alto, CA); (5) Research Science, (Kenilworth, NJ); (6) Clinical Research, (Rahway, NJ), MRL, Merck & Co., Inc., Kenilworth, NJ, USA.
Introduction: During development of biologic drugs such as monoclonal antibodies (mAb), a major challenge can be posed by the lack of homology between human and non-human primate or murine target protein/receptor, thus limiting the utility of preclinical studies for FIH dose selection in some cases. Furthermore, in-vivo abundance and turnover of human target/protein is typically unknown during the early stages of drug discovery and development. In oncology, such challenges often lead to a very low FIH starting dose, resulting in sub-therapeutic drug exposures for many patients with life-threatening diseases. For such targets in oncology, a FIH clinical study design informed by PKPD analysis, is outlined.
Objectives:
For a mAb drug with lack of pre-clinical cross-reactivity:
- Predict a safe starting FIH dose
- Outline a dose-escalation clinical study design
- Predict the highest clinical dose to ensure proof of pharmacology
Methods:
The proposed pharmacology-based model consists of membrane-bound and soluble forms of the therapeutic target. The model included zero-order synthesis of membrane target, first-order formation of soluble target from the membrane target and first-order degradation of the soluble target. The PK of the mAb drug was characterized using a two-compartment linear PK model. For simulations, population level fixed and random effect parameters of pembrolizumab were taken as representative of a typical mAb [1]. The model allowed binding of mAb to both, membrane and soluble forms of the target. The only pharmacological information available for the target was binding affinity (KD) of drug to the human target and soluble target concentrations (10-24 ng/mL) measured in human serum. A wide range of turnover rates for membrane (5-100 /day) and soluble targets (0.01-0.2 /day) were considered for simulations [2].
A multiple dosing regimen with a 21-day dosing interval was adopted for all simulations. For informing the starting FIH dose and highest clinical dose, simulations for membrane target occupancy profiles were conducted for a range of doses (3 mg to 2400 mg) using distinct combinations of membrane target concentrations and target turnover rates considering variabilities wherein inter-individual variabilities (log-normal distribution) were assigned to PK and system parameters.
For dose escalation, accelerated titration design (ATD) with 1 patient per dose level until adequate target engagement (i.e., required for pharmacological effects), followed by a transition to a more conventional modified toxicity probability interval (mTPI) design was planned. The ATD to mTPI transition dose was informed using the model by predicting the dose required for >75% target engagement.
All simulations were conducted using “ubiquity Ver 1.0.0” package [3] in R (version 3.4.3).
Results: Using the model, a starting FIH dose of 3 mg was predicted to achieve <50% target occupancy at PK Cmax. By using measured soluble target concentrations and exploring a range of turnover rates [2], the simulation space was constrained to a distinct set of nine potential scenarios with respect to target occupancy. These nine scenarios included specific combinations of membrane target concentrations (0.06, 1, 10 nM) and target turnover half-life (0.1, 1, 16 h). If target turnover half-life was 0.1h and baseline target concentration was either 1 nM or 10 nM, then membrane target occupancy at all clinically feasible doses could not be sustained over a 21-day dosing interval. For a target with rapid turnover half-life of 0.1h and 21-day dosing interval, the membrane target expression had to be low i.e.<= 0.06 nM to achieve >75% target occupancy at ≥700 mg mAb dose. For the remaining scenarios, >75% membrane target occupancy could always be achieved by >=1600 mg mAb dose over a 21-day dosing interval, thus informing the highest clinical dose. Simulations predicted approximately 80% membrane target occupancy at the 100 mg dose Ctrough suggesting that the ATD to mTPI transition could occur at this dose.
Conclusions: In conclusion, a systems pharmacology model was used to address the issue of lack of pre-clinical pharmacology data. Given the unknowns related to the target pharmacology, these model-based results guided the initial clinical design, which would be updated with emerging clinical data. This work shows how FIH clinical study design can be informed in cases of limited pre-clinical pharmacology and PK-PD data.
References:
[1] Ahamadi, M et al. “Model-Based Characterization of the Pharmacokinetics of Pembrolizumab: A Humanized Anti-PD-1 Monoclonal Antibody in Advanced Solid Tumors” CPT: pharmacometrics & systems pharmacology vol. 6,1 (2016): 49-57.
[2] http://www.quantpharm.com/pdf_files/TMDD_Talk_PAGE_2011.pdf
[3] https://github.com/john-harrold/ubiquity
Reference: PAGE 28 (2019) Abstr 8854 [www.page-meeting.org/?abstract=8854]
Poster: Drug/Disease Modelling - Oncology