A multiscale systems pharmacology framework to predict drug-class specific prophylactic efficacy of antivirals against HIV
Sulav Duwal (1), Laura Dickinson (2), Saye Khoo (2), Max von Kleist (1)
(1) Department of Mathematics & Computer Science, Free University Berlin, Germany (2) Institute of Translational Medicine, University of Liverpool, United Kingdom
Objectives: Despite intense research, a complete cure and an effective vaccine against HIV have remained elusive till now. This emphasizes the importance of prevention strategies such as pre-exposure prophylaxis(PrEP) to curb the spread of HIV[1]. Currently, there are more than 30 antiretrovirals belonging to 4 major drug class for HIV-1 treatment and only two of them belonging to nucleoside reverse transcriptase inhibitor (NRTI) are approved for PrEP. The challenge is to rationally prioritize other antiretrovirals for PrEP repurposing and optimize the administration schemes. To this end, we extended the previously developed system pharmacology framework for NRTI for all antivirals and tailor-designed the hybrid deterministic-stochastic algorithm for efficient computation.
Methods: Previously, we built a multiscale modular systems pharmacology pipeline to assess the prophylactic efficacy of antivirals belonging to nucleotide reverse transcriptase inhibitors [2]. The pipeline meaningfully integrates processes of various scales. This includes modelling and simulation of the molecular mechanism of action at microscale level to meso-, macro- and population scale processes, such as the drug pharmacokinetics, viral replication dynamics, vertical viral transmission, up to the long-term infection probabilities after repeated virus exposure, akin to a clinical trial. In our recent works, we extended the pipeline for all antiviral classes considering their respective mode-of-action [3]. We utilized the branching process theory to derive drug-class specific concentration-prophylactic efficacy curve (dose-response curve).
Secondly, we employed a recently developed hybrid deterministic-stochastic algorithm based on Monte Carlo technique known as EXTRANDE [4]. EXTRANDE utilizes the thinning technique which allows exact simulation of a stochastic process (viral dynamics) embedded in a dynamically changing environment (antiviral pharmacokinetics). We extended EXTRANDE by introducing stopping criteria based on dynamically adapting extinction simplex derived from the branching process. This guarantees that the probability of falsely classifying a trajectory as an infection event is below a user-defined threshold, while the computational run-time is optimal for the user-defined threshold [3].
Results: In vitro measured drug potency (IC50, IC90) usually guides the design of PrEP trials [5]. We showed that such direct translation of in vitro drug potency to prophylactic efficacy is systematically misleading [3]. Except for reverse transcriptase inhibitor, the in vitro potency overestimates the prophylactic efficacy. Furthermore, we derived drug-class specific concentration-prophylactic efficacy curves which properly translates the in vitro measured drug potency to prophylactic efficacy. We observed that the shape of the concentration-prophylactic efficacy for co-receptor antagonists, reverse transcriptase inhibitors and integrase inhibitors is a classical Emax equation, whereas for protease inhibitors it is a power function.
Using the framework, we benchmarked all the treatment-approved antivirals and predicted that oral darunavir, efavirenz, nevirapine, etravirine and rilpivirine may provide complete protection at clinically relevant concentrations against wildtype virus [3]. Utilizing the population pharmacokinetics of dolutegravir, we assessed various prevention strategies based on dolutegravir. We predicted that the plasma concentrations of 145.18 and 722.23nM prevent 50- and 90% sexual transmissions respectively.
Conclusions: Herein presented systems pharmacology framework and algorithm can be used to select or rule-out PrEP candidates based on their prophylactic efficacies. Moreover, the algorithm allows for exact and efficient simulation of various administration strategies. This can be used to assess, design and optimize the administration strategies.
References:
[1] http://www.unaids.org/sites/default/files/media_asset/2016-prevention-gap-report_en.pdf
[2] Duwal S et al. (2016) CPT Pharmacometrics Syst Pharmacol, 5(7):377–387.
[3] Duwal S et al. (2017) Submitted.
[4] Voliotis M et al. (2016) PLoS Comput Biol., 12(6):e1004923.
[5] McGowan I et al. (2016) Lancet HIV, 3:e569–e578