(M1230-05-32) Using a Semi-Mechanistic Model to Predict Efficacious Dosing Regimens and Optimal Drug Design Properties for a LNP-Delivered mRNA UGT1A1 Replacement Therapy in Crigler-Najjar Syndrome Type 1 Patients

Purpose: Crigler-Najjar syndrome type 1 (CN1) is an autosomal recessive disease where bilirubin accumulates to toxic levels in the blood as a result of the reduced enzymatic activity of glucuronosyltransferase family 1 member A1 (UGT1A1). With no approved drugs for CN1, patients are often treated with blue light, but as they age, this is insufficient to control bilirubin levels. While liver transplants can be used to restore UGT1A1 activity, an alternative approach considers using gene therapy to restore UGT1A1 activity by delivery of mRNA encoding UGT1A1 protein via lipid nanoparticle (LNP). Utilizing a semi-mechanistic model, we explored the feasibility of dose amounts and frequencies that could sustain reductions in bilirubin levels over time and investigated the therapeutic design variables that most strongly influence dose projections to help prioritize optimization efforts.

Methods: A semi-mechanistic model was developed to describe an mRNA therapy delivered to the liver via a LNP. The model includes LNP delivery to the liver, endosomal escape of the mRNA, translation of the target protein UGT1A1, and UGT1A1 catalysis of bilirubin to bilirubin monoglucuronide. The model parameters were set to typical values for lipid-delivered mRNA therapeutics taken from the literature. The UGT1A1 half-life was taken to be a standard value for an intracellular enzyme. To illustrate the usage of the model for a feasibility assessment, no parameters were derived from clinical or preclinical studies.

Results: Model simulations predict that prohibitively large doses are required to maintain the bilirubin reduction target of 25% of baseline over time. The model predicts that doses larger than 0.07 mg/kg achieve sustained reduction in bilirubin concentration at Q1W dosing frequency; however, increasing the dosing interval requires increasing doses, and at Q3W doses need to be prohibitively high to maintain desired bilirubin levels throughout the dosing period. This is because the mRNA is cleared at an exponential rate inside the cells, creating a nonlinear relationship between dosing frequency and sustainability of response. To achieve longer dose intervals with lower dose amounts, it may be possible to engineer the properties of the lipid or the mRNA for improved efficacy. Three drug design parameters were scanned over a realistic range of values to assess the possibility of attaining the feasibility criteria for bilirubin reduction at 0.5 mg/kg dose Q3W. The efficiency of mRNA escape from endosomes is known to be a major bottleneck in intracellular drug delivery and can be potentially modified by optimizing the lipid composition of the LNP. The model predicts that increasing endosomal escape efficiency may provide some benefit but not enough to attain the feasibility criteria at this dose amount and interval. Similarly, more efficient translation, which can be achieved through codon-optimization of the mRNA, does not result in enough benefit to reach feasibility. The stability of the mRNA, on the other hand, which is represented in the model by the intracellular mRNA half life, is a key design variable that impacts the feasibility for UGT1A1 replacement and its pharmacology. We performed a 2D scan on dose and mRNA half life to quantify the effect of mRNA stability on bilirubin reduction at other dose levels. The model predicts that as the mRNA half life increases, the dose required for a given amount of bilirubin reduction is reduced substantially. With an mRNA half life of 30 hours, the drug achieves 25% reduction at all times at 0.5 mg/kg. Increasing the half life to 40 hours reduces the required dose five-fold to 0.1 mg/kg.

Conclusion: In summary, a semi-mechanistic model was developed and parameterized to predict efficacious dose and dosing intervals and to help prioritize optimization of liver-targeted mRNA enzyme design to treat CN1. Using this model, we determined that it is feasible for liver UGT1A1 to restrict bilirubin to non-toxic levels at frequent (Q1W-Q2W) dosing intervals, but dose amounts would be prohibitively high at less frequent intervals (Q3W). With further model analysis, we identified UGT1A1 mRNA stability as the primary design variable that may be modified to enable a longer dosing interval at reasonable dose amounts. This case study highlights how mechanistic modeling can be used for making early feasibility assessments for novel gene therapy treatments and how modeling can help inform the drug design targets for optimal efficacy of these therapeutics.

References: Apgar JF, Tang JP, Singh P, Balasubramanian N, Burke J, Hodges MR, et al. Quantitative Systems Pharmacology Model of hUGT1A1-modRNA Encoding for the UGT1A1 Enzyme to Treat Crigler-Najjar Syndrome Type 1. CPT Pharmacometrics Syst Pharmacol. 2018 Jun;7(6):404–12.

Model diagram

A 2D scan on dose and mRNA half life to quantify the effect of mRNA stability on bilirubin reduction over a range of dose levels. The color bar indicates the percent of bilirubin reduction from baseline at the end of the dosing interval.