Machine Learning Attempts for Predicting Human Subcutaneous Bioavailability of Monoclonal Antibodies

Another string to your bow: machine learning prediction of the pharmacokinetic properties of small molecules

INTRODUCTION

Prediction of pharmacokinetic (PK) properties is crucial for drug discovery and development. Machine-learning (ML) models, which use statistical pattern recognition to learn correlations between input features (such as chemical structures) and target variables (such as PK parameters), are being increasingly used for this purpose. To embed ML models for PK prediction into workflows and to guide future development, a solid understanding of their applicability, advantages, limitations, and synergies with other approaches is necessary.

AREAS COVERED

This narrative review discusses the design and application of ML models to predict PK parameters of small molecules, especially in light of established approaches including in vitro-in vivo extrapolation (IVIVE) and physiologically based pharmacokinetic (PBPK) models. The authors illustrate scenarios in which the three approaches are used and emphasize how they enhance and complement each other. In particular, they highlight achievements, the state of the art and potentials of applying machine learning for PK prediction through a comphrehensive literature review.

EXPERT OPINION

ML models, when carefully crafted, regularly updated, and appropriately used, empower users to prioritize molecules with favorable PK properties. Informed practitioners can leverage these models to improve the efficiency of drug discovery and development process.

Predicting the clinical subcutaneous absorption rate constant of monoclonal antibodies using only the primary sequence: a machine learning approach

Subcutaneous injections are an increasingly prevalent route of administration for delivering biological therapies including monoclonal antibodies (mAbs). Compared with intravenous delivery, subcutaneous injections reduce administration costs, shorten the administration time, and are strongly preferred from a patient experience point of view. An understanding of the absorption process of a mAb from the injection site to the systemic circulation is critical to the process of subcutaneous mAb formulation development. In this study, we built a model to predict the absorption rate constant (ka), which denotes how fast a mAb is absorbed from the site of administration. Once trained, our model (enabled by the XGBoost algorithm in machine learning) can predict the ka of a mAb following a subcutaneous injection using in silico molecular properties alone (generated from the primary sequence). Our model does not need clinically observed plasma concentration-time data; this is a novel capability not previously achieved in predictive pharmacokinetic models. The model also showed improved performance when benchmarked against a recently reported mechanistic model that relied on clinical data to predict subcutaneous absorption of mAbs. We further interpreted the model to understand which molecular properties affect the absorption rate and showed that our findings are consistent with previous studies evaluating subcutaneous absorption through direct experimentation. Taken altogether, this study reports the development, validation, benchmarking, and interpretation of a model that can predict the clinical ka of a mAb using its primary sequence as the only input.

Advancing algorithmic drug product development: Recommendations for machine learning approaches in drug formulation

Artificial intelligence is a rapidly expanding area of research, with the disruptive potential to transform traditional approaches in the pharmaceutical industry, from drug discovery and development to clinical practice. Machine learning, a subfield of artificial intelligence, has fundamentally transformed in silico modelling and has the capacity to streamline clinical translation. This paper reviews data-driven modelling methodologies with a focus on drug formulation development. Despite recent advances, there is limited modelling guidance specific to drug product development and a trend towards suboptimal modelling practices, resulting in models that may not give reliable predictions in practice. There is an overwhelming focus on benchtop experimental outcomes obtained for a specific modelling aim, leaving the capabilities of data scraping or the use of combined modelling approaches yet to be fully explored. Moreover, the preference for high accuracy can lead to a reliance on black box methods over interpretable models. This further limits the widespread adoption of machine learning as black boxes yield models that cannot be easily understood for the purposes of enhancing product performance. In this review, recommendations for conducting machine learning research for drug product development to ensure trustworthiness, transparency, and reliability of the models produced are presented. Finally, possible future directions on how research in this area might develop are discussed to aim for models that provide useful and robust guidance to formulators.

Global parameter optimisation and sensitivity analysis of antivenom pharmacokinetics and pharmacodynamics

In recent years it has become possible to design snakebite antivenoms with diverse pharmacokinetic properties. Owing to the pharmacokinetic variability of venoms, the choice of antivenom scaffold may influence a treatment's neutralisation coverage. Computation offers a useful medium through which to assess the pharmacokinetics and pharmacodynamics of envenomation-treatment systems, as antivenoms with identical neutralising capacities can be simulated. In this study, we simulate envenomation and treatment with a variety of antivenoms, to define the properties of effective antivenoms. Systemic envenomation and treatment were described using a two-compartment pharmacokinetic model. Treatment of Naja sumatrana and Cryptelytrops purpureomaculatus envenomation was simulated with a set of 200,000 theoretical antivenoms across 10 treatment time delays. These two venoms are well-characterised and have differing pharmacokinetic properties. The theoretical antivenom set varied across molecular weight, dose, kon, koff, and valency. The best and worst treatments were identified using an area under the curve metric, and a global sensitivity analysis was performed to quantify the influence of the input parameters on treatment outcome. The simulations show that scaffolds of diverse molecular formats can be effective. Molecular weight and valency have a negligible direct impact on treatment outcome, however low molecular weight scaffolds offer more flexibility across the other design parameters, particularly when treatment is delayed. The simulations show kon to primarily mediate treatment efficacy, with rates above 105 M-1s-1 required for the most effective treatments. koff has the greatest impact on the performance of less effective scaffolds. While the same scaffold preferences for improved treatment are seen for both model snakes, the parameter bounds for C. purpureomaculatus envenomation are more constrained. This paper establishes a computational framework for the optimisation of antivenom design.

Predicting Human Bioavailability of Subcutaneously Administered Monoclonal Antibodies Using Non-human Primate Linear Clearance and Antibody Isoelectric Point

The prediction of bioavailability is one of the major barriers in the clinical translation of subcutaneously (SC) administered therapeutic monoclonal antibodies (mAbs) due to the lack of reliable in vitro and preclinical in vivo predictive models. Recently, multiple linear regression (MLR) models were developed to predict human SC bioavailability of mAbs using human linear clearance (CL) and isoelectric point (pI) of the whole antibody or Fv regions as independent variables. Unfortunately, these models cannot be applied to mAbs at the preclinical development stage because human CLs of these mAbs are unknown. In this study, we predicted human SC bioavailability of mAbs using preclinical data only by two approaches. In the first approach, allometric scaling was used to predict human linear CL from non-human primate (NHP) linear CL. The predicted human CL and the pI of the whole antibody or Fv regions were then incorporated into two previously published MLR models to predict the human bioavailability of 61 mAbs. In the second approach, two MLR models were developed using NHP linear CL and the pI of whole antibody or Fv regions of 41 mAbs in a training set. The two models were validated using an independent test dataset containing 20 mAbs. The four MLR models generated 77-85% of predictions within 0.8- to 1.2-fold deviations from observed human bioavailability. Overall, this study demonstrated that human SC bioavailability of mAbs at the preclinical stage could be predicted using NHP CL and pI of mAbs.

Artificial intelligence for compound pharmacokinetics prediction

Optimisation of compound pharmacokinetics (PK) is an integral part of drug discovery and development. Animal in vivo PK data as well as human and animal in vitro systems are routinely utilised to evaluate PK in humans. In recent years machine learning and artificial intelligence (AI) emerged as a major tool for modelling of in vivo animal and human PK, enabling prediction from chemical structure early in drug discovery, and therefore offering opportunities to guide the design and prioritisation of molecules based on relevant in vivo properties and, ultimately, predicting human PK at the point of design. This review presents recent advances in machine learning and AI models for in vivo animal and human PK for small-molecule compounds as well as some examples for antibody therapeutics.

Predicting Human Bioavailability of Subcutaneously Administered Fusion Proteins and Monoclonal Antibodies Using Human Intravenous Clearance or Antibody Isoelectric Point

There has been an increasing trend towards subcutaneous (SC) delivery of fusion proteins and monoclonal antibodies (mAbs) in recent years versus intravenous (IV) administration. The prediction of bioavailability is one of the major barriers in clinical translation of SC-administered therapeutic proteins due to a lack of reliable in vitro and preclinical in vivo predictive models. In this study, we explored the relationships between human SC bioavailability and physicochemical or pharmacokinetic properties of 19 Fc- or albumin-fusion proteins and 98 monoclonal antibodies. An inverse linear correlation was observed between human SC bioavailability and intravenous clearance (CL) or isoelectric point (pI). Multivariate regression models were developed using intravenous CL and pI of a training set (N = 59) as independent variables. The predictive models of mAbs were validated with an independent test set (N = 33). Two linear regression models resulted in 24 (73%) and 27 (82%) among 33 predictions within 0.8- to 1.2-fold deviations. Due to the small sample size of dataset, regression model validation was not conducted for fusion proteins. Overall, this study demonstrated that CL- and pI-based multivariate regression models could be used to predict human SC bioavailability of mAbs.

Monoclonal antibody and protein therapeutic formulations for subcutaneous delivery: high-concentration, low-volume vs. low-concentration, high-volume

Biologic drugs are used to treat a variety of cancers and chronic diseases. While most of these treatments are administered intravenously by trained healthcare professionals, a noticeable trend has emerged favoring subcutaneous (SC) administration. SC administration of biologics poses several challenges. Biologic drugs often require higher doses for optimal efficacy, surpassing the low volume capacity of traditional SC delivery methods like autoinjectors. Consequently, high concentrations of active ingredients are needed, creating time-consuming formulation obstacles. Alternatives to traditional SC delivery systems are therefore needed to support higher-volume biologic formulations and to reduce development time and other risks associated with high-concentration biologic formulations. Here, we outline key considerations for SC biologic drug formulations and delivery and explore a paradigm shift: the flexibility afforded by low-to-moderate-concentration drugs in high-volume formulations as an alternative to the traditionally difficult approach of high-concentration, low-volume SC formulation delivery.

High concentration formulation developability approaches and considerations

The growing need for biologics to be administered subcutaneously and ocularly, coupled with certain indications requiring high doses, has resulted in an increase in drug substance (DS) and drug product (DP) protein concentrations. With this increase, more emphasis must be placed on identifying critical physico-chemical liabilities during drug development, including protein aggregation, precipitation, opalescence, particle formation, and high viscosity. Depending on the molecule, liabilities, and administration route, different formulation strategies can be used to overcome these challenges. However, due to the high material requirements, identifying optimal conditions can be slow, costly, and often prevent therapeutics from moving rapidly into the clinic/market. In order to accelerate and derisk development, new experimental and in-silico methods have emerged that can predict high concentration liabilities. Here, we review the challenges in developing high concentration formulations, the advances that have been made in establishing low mass and high-throughput predictive analytics, and advances in in-silico tools and algorithms aimed at identifying risks and understanding high concentration protein behavior.

Non-specificity as the sticky problem in therapeutic antibody development

Antibodies are highly potent therapeutic scaffolds with more than a hundred different products approved on the market. Successful development of antibody-based drugs requires a trade-off between high target specificity and target binding affinity. In order to better understand this problem, we here review non-specific interactions and explore their fundamental physicochemical origins. We discuss the role of surface patches - clusters of surface-exposed amino acid residues with similar physicochemical properties - as inducers of non-specific interactions. These patches collectively drive interactions including dipole-dipole, π-stacking and hydrophobic interactions to complementary moieties. We elucidate links between these supramolecular assembly processes and macroscopic development issues, such as decreased physical stability and poor in vivo half-life. Finally, we highlight challenges and opportunities for optimizing protein binding specificity and minimizing non-specificity for future generations of therapeutics.

Advanced Formulations/Drug Delivery Systems for Subcutaneous Delivery of Protein-Based Biotherapeutics

Multiple advanced formulations and drug delivery systems (DDSs) have been developed to deliver protein-based biotherapeutics via the subcutaneous (SC) route. These formulations/DDSs include high-concentration solution, co-formulation of two or more proteins, large volume injection, protein cluster/complex, suspension, nanoparticle, microparticle, and hydrogel. These advanced systems provide clinical benefits related to efficacy and safety, but meanwhile, have more complicated formulations and manufacturing processes compared to conventional solution formulations. To develop a fit-for-purpose formulation/DDS for SC delivery, scientists need to consider multiple factors, such as the primary indication, targeted site, immunogenicity, compatibility, biopharmaceutics, patient compliance, etc. Next, they need to develop appropriate formulation (s) and manufacturing processes using the QbD principle and have a control strategy. This paper aims to provide a comprehensive review of advanced formulations/DDSs recently developed for SC delivery of proteins, as well as some knowledge gaps and potential strategies to narrow them through future research.

Designing antibodies as therapeutics

Antibody therapeutics are a large and rapidly expanding drug class providing major health benefits. We provide a snapshot of current antibody therapeutics including their formats, common targets, therapeutic areas, and routes of administration. Our focus is on selected emerging directions in antibody design where progress may provide a broad benefit. These topics include enhancing antibodies for cancer, antibody delivery to organs such as the brain, gastrointestinal tract, and lungs, plus antibody developability challenges including immunogenicity risk assessment and mitigation and subcutaneous delivery. Machine learning has the potential, albeit as yet largely unrealized, for a transformative future impact on antibody discovery and engineering.

Machine-learning-guided early drug discovery of small molecules

Machine learning (ML) approaches have been widely adopted within the early stages of the drug discovery process, particularly within the context of small-molecule drug candidates. Despite this, the use of ML is still limited in the pharmacokinetic/pharmacodynamic (PK/PD) application space. Here, we describe recent progress and the role of ML used in preclinical drug discovery. We summarize the advances and current strategies used to predict ADME properties of small molecules based on their structures, and predict structures based on the desired properties for molecular screening and optimization. Finally, we discuss the use of ML to predict PK to rank the ability of drug candidates to achieve appropriate exposures and hence provide important insights into safety and efficacy. Teaser: In this short review we highlight key advances in the application of machine learning to early drug discovery for small molecules in the preclinical setting.