Model-Based First-in-Human Dose Prediction Strategies

Abstract

Model-based first-in-human (FIH) dose prediction strategies are revolutionizing early-phase clinical drug development by introducing quantitative pharmacological principles to optimize safety and efficacy. These approaches integrate preclinical pharmacokinetic (PK), pharmacodynamic (PD), and toxicological data using mechanistic and statistical modeling to predict a rational initial human dose. This review provides a comprehensive analysis of model-based FIH dose prediction, examining key methodologies, current evidence, clinical relevance, and implications for future drug development. The strengths, limitations, and evolving guideline recommendations are discussed to inform clinicians, researchers, and regulatory professionals on best practices and emerging trends in model-informed clinical trial design.

Introduction

The transition from preclinical research to first-in-human studies represents a pivotal stage in drug development, carrying inherent risks for both participants and sponsors. Traditional FIH dose selection often relied on empirical allometric scaling or safety factors applied to no-observed-adverse-effect levels (NOAEL). However, these approaches may inadequately capture human pharmacology and interspecies differences, leading to suboptimal safety margins or inefficient escalation schemes. Model-based dose prediction strategies employ quantitative frameworks such as physiologically-based pharmacokinetic (PBPK) models, population PK/PD modeling, and exposure-response analysis to integrate diverse data sources and simulate human drug disposition and effect. These methodologies are endorsed by regulatory agencies and are increasingly incorporated into investigational new drug (IND) submissions to enhance translational accuracy and patient safety.

Epidemiology / Disease Burden

While dose prediction itself is not tied to a specific disease, the imperative for accurate FIH dose selection is underscored by the growing complexity and volume of novel therapeutics entering clinical pipelines, especially in oncology, immunology, and rare diseases. Failures in early-phase trials due to inadequate dose selection contribute to drug attrition rates estimated at 30–40% in phase I, representing a significant burden on healthcare resources, patient safety, and development timelines. The need for robust model-based prediction aligns with the increasing prevalence of personalized and targeted therapies, where interpatient variability and narrow therapeutic indices are common.

Pathophysiology

Model-based FIH dose prediction is grounded in an understanding of drug absorption, distribution, metabolism, and excretion (ADME) as well as mechanism of action (MoA) at molecular, cellular, and systemic levels. Mechanistic modeling incorporates species-specific differences in enzyme expression, receptor affinity, transporter activity, and target biology. For biologics, models may account for target-mediated drug disposition and immunogenicity, while for small molecules, hepatic and renal clearance pathways are parameterized. These models simulate human pharmacology by extrapolating from animal and in vitro systems, integrating physiological variability and disease-modified states, thereby providing a pathophysiological rationale for dose selection.

Risk Factors

Risk factors influencing FIH dose prediction include interspecies differences in drug metabolism, incomplete understanding of human target biology, and limitations of preclinical models in recapitulating human disease physiology. Additional risks arise from novel modalities such as gene therapies, antibody-drug conjugates, or cell-based therapies, where human-specific responses may be unpredictable. Patient-specific factors including genetic polymorphisms, comorbidities, and concomitant medications may further modulate drug exposure and response, underscoring the need for individualized model-based approaches. Regulatory guidance emphasizes the identification and quantification of such risks through robust modeling and sensitivity analyses prior to FIH dosing.

Clinical Features

In the context of FIH studies, clinical features refer to the anticipated pharmacological effects and safety signals predicted by model-based simulations. These models estimate the concentration-time profiles, potential for adverse events, therapeutic window, and biomarkers of drug activity. Early identification of dose-limiting toxicities, maximal tolerated dose, and pharmacodynamic endpoints is facilitated by virtual patient populations and scenario testing. The clinical translation of these features informs trial design, cohort selection, and adaptive dosing strategies, promoting patient safety while enabling efficient dose escalation.

Diagnosis

Diagnosis in the dose prediction paradigm involves the systematic evaluation of preclinical data quality, translational relevance, and model performance. This includes validation of input parameters against human in vitro systems, retrospective analysis of similar compounds, and cross-validation with historical clinical data. Diagnostic criteria for model adequacy include predictive accuracy, precision, and robustness under uncertainty. Regulatory agencies may require justification of model assumptions, sensitivity analyses, and transparent reporting of predictive intervals as part of IND submissions.

Treatment & Management

Model-based FIH dose prediction directly informs the treatment and management of study participants in early-phase trials. Dosing regimens are tailored based on predicted PK/PD profiles, with real-time model updating as clinical data accrue. This enables dynamic risk mitigation, rapid identification of optimal biological dose, and minimization of exposure to subtherapeutic or toxic levels. Management strategies may incorporate Bayesian adaptive designs, exposure-response modeling, and incorporation of translational biomarkers to refine dosing in subsequent cohorts or patient populations.

Recent Advances / Emerging Therapies

Recent advances in model-based dose prediction include the integration of artificial intelligence (AI) and machine learning algorithms to automate data extraction, parameter estimation, and predictive analytics. The use of multi-scale modeling linking molecular, cellular, and organismal data enhances the mechanistic understanding of drug action. Emerging therapies such as gene editing, RNA therapeutics, and bispecific antibodies pose novel challenges for dose prediction, necessitating the development of new modeling paradigms and translational tools. Regulatory initiatives, including the FDA's Model-Informed Drug Development (MIDD) Pilot Program, are fostering collaboration between industry and regulators to standardize best practices and accelerate innovation.

Guideline Recommendations

Current guidelines from regulatory bodies such as the FDA, EMA, and ICH recommend the use of model-informed approaches for FIH dose selection, advocating for transparent documentation, rigorous validation, and iterative model refinement. Guidance documents emphasize the importance of integrating all available data preclinical, in vitro, in silico, and clinical to justify dose selection and escalation schemes. Model-based approaches are particularly encouraged for high-risk compounds, novel modalities, and special populations (e.g., pediatrics, geriatrics, patients with organ impairment). Collaboration with regulatory agencies early in development is advised to align expectations and facilitate regulatory review.

Conclusion

Model-based first-in-human dose prediction strategies represent a transformative advancement in clinical drug development, enabling more precise, safe, and efficient translation of novel therapeutics from bench to bedside. By integrating mechanistic and statistical modeling with empirical data, these approaches address key challenges in dose selection and risk mitigation. Ongoing advances in modeling science, regulatory guidance, and computational tools promise to further enhance the reliability and impact of model-based strategies. Adoption of these methods is essential for optimizing early-phase trials, safeguarding patient welfare, and accelerating the development of innovative therapies.