Knöchel J (1), Nilsson C (1), Carlsson B (2), Wernevik L (2), Schumi J (3), Hofherr A (2), Johanson P (4), Rydén-Bergsten T (2), Rekic D (1)
[1] Clinical Pharmacology and Quantitative Pharmacology, Clinical Pharmacology & Safety Sciences, R&D, AstraZeneca, Gothenburg, Sweden [2] Research and Early Development, Cardiovascular, Renal and Metabolism, BioPharmaceuticals R&D, AstraZeneca, Gothenburg, Sweden [3] Early Biometrics and Statistical Innovation, Data Science & Artificial Intelligence, BioPharmaceuticals R&D, AstraZeneca, Gaithersburg MD, USA [4] Research and Late Development, Cardiovascular, Renal and Metabolism, BioPharmaceuticals R&D, AstraZeneca, Gothenburg, Sweden
Objectives:
To this day, statin treatment is the standard of care for the management of atherosclerotic cardiovascular disease (ASCVD), however, target low density lipoprotein cholesterol (LDL-C) levels are not reached in all patients. The discovery of the proprotein convertase subtilisin/kexin type 9 (PCSK9) and its ability to regulate the LDL receptor (LDL-C) has resulted in clinical development of PCSK9 monoclonal antibody inhibitors [1-2] and a silencing RNA that inhibit the production of PCSK9 [3]. This provided a unique opportunity for model-informed drug development (MIDD) of a novel PCSK9 antisense oligonucleotide inhibitor. Integrating the abundance of data from older PCSK9 inhibitors as well as semi-mechanistic pharmacodynamic modelling, this case study spans both the technical modelling decisions as well as drug development impact of MIDD.
Methods:
Data delivery in a Phase 1 single ascending dose study was streamlined so that a PCSK9 kinetic pharmacodynamic (K-PD) model could be developed [4]. At each delivery the model parameters were updated and a new dose prediction made based on the current model. The model was then used to design Phase 2 studies. The K-PD model focused on PCSK9 and used the quantitative relationship to LDL-C based on available literature data. In this part a total of 1112 samples from 84 subjects were used.
The PCSK9 K-PD model was used to conduct clinical trial simulations to evaluate the effect of AZD8233 Q4W dosing on PCSK9 steady-state levels and to calculate the probability of technical success (PTS) to help design Phase 2.
In the population model, which was used for therapeutic dose prediction, 152 subjects from Phase 1 and Phase 2 studies were included. In contrast to the modelling work done up to this point, this model included a semi-mechanistic approach for the link between PCSK9 and LDL-C to allow for estimation of both markers at the same time.
The model was used to conduct a stepwise covariate modelling (SCM) search with forward inclusion (p<0.05) and backward deletion (p<0.01) and a subgroup analysis with each individual’s empirical bayes parameter estimates assuming steady treatment with 60 mg Q4W dosing. The following covariates were assessed in the SCM analysis: dose, age, bodyweight, renal function (moderate and mild impairment), race (White and Black), sex, statin treatment, baseline PCSK9 and baseline LDL-C levels.
Results:
Based on clinical trial simulations (CTS) using the PCSK9 K-PD model, the predicted therapeutic dose was 49 mg (90% CI: 42, 59). However, a dose needed to reach >90% of PTS of observing a mean PCSK9 reduction of 90% in the proposed phase 2 study was estimated to be 53 mg. In order to achieve a reduction of 90% in PCSK9 for 90% of the patients with 90% PTS a dose of 97 mg would be needed. Using these results from the CTS as well as PTS, the 15, 50 and 90 mg dose were selected to be investigated in the Phase 2 study.
The time course of PCSK9 and LDL-C response to AZD8233 in the Phase 2 study was well described by the developed population K-PD model. Simulations with the model predicted LDL-C changes from baseline of 69.3%, 71%, 72.5%, 73.4%, and 74% at week 24, after repeated every 4 weeks doses of 50, 60, 70, 80, and 90 mg , respectively. The predicted LDL-C change from baseline at week 24 for the placebo arm was 0.03%.
Stepwise covariate modelling identified body weight, sex, age, statin treatment therapy, baseline LDL-C and race as statistically significant covariates for several structural parameters of the K-PD model. However, their overall impact on PCSK9 and LDL-C reduction was negligible. No covariate resulted in more than ±2% difference in LDL-C reduction to the reference subject. The subgroup analysis also showed that the difference in LDL-C response was negligible in subjects on high vs low statin intensity treatment, and in subjects with LDL-C baseline above or below 100 mg/dL. The largest difference in LDL-C effect (measured as percent change from baseline, assuming treatment with 60 mg AZD8233) was found between men and women (71 vs 74%) and in subjects below 75 kg versus subjects above 100 kg (75% vs 71%). These differences were not considered clinical relevant [5,6].
Conclusions:
This case study demonstrates the power of MIDD to accelerate clinical development by using data from a single dose Phase 1 study and safety from an ongoing multiple ascending dose (MAD) study to design and start Phase 2 in parallel with Phase 1 MAD and inform decision making in clinical development of a novel PCSK9 inhibitor.
References:
[1] Gibbs et al 2017, Impact of Target-Mediated Elimination on the Dose and Regimen of Evolocumab, a Human Monoclonal Antibody Against Proprotein Convertase Subtilisin / Kexin Type 9 (PCSK9). J Clin Pharmacol 2017 57:616-626
[2] Nicolas, Xavier, et al. “Population Pharmacokinetic/Pharmacodynamic Analysis of Alirocumab in Healthy Volunteers or Hypercholesterolemic Subjects Using an Indirect Response Model to Predict Low-Density Lipoprotein Cholesterol Lowering: Support for a Biologics License Application .” Clinical Pharmacokinetics, vol. 58, no. 1, 2019, pp. 115–30, doi:10.1007/s40262-018-0670-5.
[3] Kathman et al 2018, Population dose-response modelling of inclisiran, a novel siRNA inhibitor to PCSK9, in patients with high cardiovascular risk with elevated LDL cholesterol. 119th Annual Meeting of the American-Society-for-Clinical-Pharmacology-and-Therapeutics (ASCPT) – Breaking Down Barriers to Effective Patient Care (Wiley, 2018)
[4] Jacqmin, P., et al. “Modelling Response Time Profiles in the Absence of Drug Concentrations?: Definition and Performance Evaluation of the K – PD Model.” Journal of Pharmacokinetics and Pharmacodynamics, vol. 34, no. 1, 2007, pp. 57–85, doi:10.1007/s10928-006-9035-z.
[5] Vaughan et al 2003, Update on Statins: 2003. Circulation 2003 110(7):886-892
doi: 10.1161/01.CIR.0000139312.10076.BA
[6] Roberts 1997, The rule of 5 and the rule of 7 in lipid-lowering by statin drugs. The American Journal of Cardiology 1997 80(1):106-107
doi: https://doi.org/10.1016/S0002-9149(97)00298-1
Reference: PAGE 30 (2022) Abstr 9974 [www.page-meeting.org/?abstract=9974]
Poster: Drug/Disease Modelling - Other Topics