Quantitative Medicine to Support Drug Development and Regulatory Decisions: A UK Regulatory Approach Informed by ICH M15

ABSTRACT

Quantitative medicine supports modern drug development and regulatory decision‐making by integrating non‐clinical and clinical data to inform dose selection, trial optimization, and extrapolation across populations and benefit–risk assessment. The Medicines and Healthcare products Regulatory Agency (MHRA) has adopted quantitative approaches across the product lifecycle within a risk‐proportionate regulatory framework. Internationally, the ICH M15 guideline provides a harmonized framework for planning, evaluation and documentation of model‐informed drug development (MIDD) evidence which is aligned with MHRA practice. This perspective describes the application of quantitative medicine to UK regulatory decision‐making using two case studies: sugemalimab (Eqjubi) in oncology and lecanemab (Leqembi) in Alzheimer's disease. The case studies are assessed using ICH M15 concepts, including context of use, model influence, and consequence of a wrong decision. For sugemalimab, quantitative pharmacokinetic modeling supported a high‐impact dosing decision with clinically meaningful consequences of a wrong decision requiring proportionate scrutiny and mitigation of uncertainty. For lecanemab, multiple quantitative models with lower individual influence supported interpretation of pharmacokinetic variability, biomarkers, and clinical outcomes within the overall benefit–risk assessment. These case studies demonstrate how quantitative models can support decisions with differential impact ranging from critical dosing decisions to interpretation of complex data. Regulatory policy that is aligned with ICH M15 can be used to support the use of quantitative medicine to link complex modeling approaches with regulatory decision‐making and enable patient access to medicines.

1. Introduction

Quantitative medicine plays an increasingly important role in modern drug development and regulatory decision‐making. Quantitative medicine methods include a wide range of approaches such as population pharmacokinetics (population PK), exposure–response (ER) analysis, pharmacokinetic/pharmacodynamic (PK/PD) modelling and systems pharmacology approaches. These methods allow integration of diverse data generated during the non‐clinical and clinical stages of drug development to support dose selection, clinical trial optimization, extrapolation between different populations and to determine benefit–risk.

In response to regulatory challenges, the Medicines and Healthcare products Regulatory Agency (MHRA) employs the use of quantitative medicine to allow innovation‐led, risk‐proportionate regulation and to improve patient access to all therapies. This includes the application of model‐informed approaches across the product lifecycle and the use of real‐world data (RWD) and real‐world evidence (RWE) to complement clinical trial evidence. Richards and Hudson indicated that modern regulators must remain at the forefront of emerging scientific methodologies such as quantitative modeling to optimize development pathways, strengthen benefit–risk assessment and accelerate access to medicines in areas of unmet medical need [1]. In the past, the MHRA worked with the European Medicines Agency (EMA) to publish guidelines on the reporting of population pharmacokinetic and physiologically based pharmacokinetic modeling. In the latter guidance the concept of model impact was used to support the level of model evaluation required [2, 3]. More recently, the MHRA has published the “Real‐World Evidence Scientific Dialogue Programme,” which promotes early engagement with developers on RWE methodologies to clarify regulatory expectations and support informed decision‐making throughout the medicinal product lifecycle [4]. This program reflects the wider data strategy to promote data‐driven innovation while maintaining strict standards for safety, quality and efficacy.

At the international level, the International Council for Harmonization (ICH) has recently published a harmonized framework for the evaluation of quantitative evidence in the ICH Harmonized Guideline: General Principles for Model Informed Drug Development (M15) [5]. The guideline was developed by the ICH M15 Expert Working Group (EWG) including participation from the MHRA, among other regulatory authorities from ICH regions. The guideline provides general principles for planning, evaluation, and documentation of model‐informed drug development (MIDD) evidence. The guideline employs concepts such as context of use (COU), proportionality between model evaluation and model risk, and transparent communication of uncertainty and limitations of the model‐based evidence. These principles align with the MHRA practice which is based on the view that quantitative models are valuable tools to support regulatory decisions but that appropriate model evaluation and validation are performed. Although ICH M15 does not recommend specific modeling techniques, it provides a common language and assessment framework to support multidisciplinary understanding and consistent regulatory evaluation.

In this perspective manuscript, we illustrate how quantitative medicine is applied to support regulatory decision‐making from a UK perspective. We use two case studies for the approval of Sugemalimab (Eqjubi) in oncology and Lecanemab (Leqembi) in Alzheimer's disease. These case studies were assessed before the formal Step 5 adoption of ICH M15. However, the principles that were used reflect established MHRA practice. Therefore, ICH M15 is expected to formalize and harmonize existing regulatory practices rather than introducing significant changes to regulatory evaluation of MIDD. These examples demonstrate how models of differing complexity and influence can be incorporated into regulatory decisions when their context of use, decision impact, and limitations are clearly considered and communicated in line with ICH M15 principles.

2. Case Studies

Application of Key assessment elements as described in ICH M15 [5]. The population PK and PK/PD models described in these case studies were developed by the applicant and were subsequently refined in response to regulatory questions raised during the assessment.

2.1. Sugemalimab (Eqjubi)

2.1.1. Context of Use (ICH M15)

The context of use for one aspect of quantitative pharmacokinetics (PK) modeling in the assessment of sugemalimab (Eqjubi) was to inform whether the fixed dosing regimen evaluated in the pivotal clinical program would provide adequate exposure in the intended entire UK population, particularly in patients with higher body weight relative to those represented in the pivotal clinical study. Population PK modeling and simulation were provided to explore exposure across an extended weight range and with consideration of exposure–response models, to assess whether an alternative dosing strategy could achieve exposure comparable to the reference population.

2.1.2. Model Risk (ICH M15)

Under the ICH M15 framework, assessment of model risk takes into consideration both the influence of the model on the regulatory decision and the consequence if the decision proves to be incorrect. In the case of sugemalimab, the quantitative model directly informed the recommended dosing strategy with limited PK data available in the target population. This represents a moderate to high model influence. Sugemalimab is administered in an oncology setting where adequate systemic exposure is required to ensure therapeutic efficacy, whereas excessive exposure may lead to increased risk of immune‐related toxicity. Although sugemalimab does not represent a classic narrow therapeutic index drug, inappropriate dose selection could plausibly result in loss of efficacy at lower exposures. The primary concern was related to potential underexposure in patients with higher body weight as this population was under‐represented in the clinical development program. In contrast, no specific safety signal that requires dose reduction in lower body weight patients was identified. Therefore, the consequence of a wrong decision with respect to dose selection was considered clinically significant. Taking into consideration this consequence and the central role of the model in informing dose justification, the overall model risk was considered high which warrants a proportionate level of scrutiny. In line with ICH M15 principles, the rigor of model evaluation and transparency of assumptions were commensurate with model influence and the potential consequence of a wrong decision. This was reflected in conservative use of model outputs, precise communication of uncertainty and regulatory reliance on the model only within the totality of evidence.

2.1.3. Key Uncertainties and Risk Mitigation

Key uncertainties included limitations of exposure–response analyses derived mainly from a single dose level and potential effects of η‐shrinkage on diagnostic interpretation. These uncertainties were mitigated by selecting PK exposure to match a clinically validated reference population rather than relying on causal ER inference. This resulted in the Population PK model being the influential model. Complementary diagnostics were used to assess the robustness of the population model in terms of error on model parameters and comparison with post hoc estimation of parameters and consideration of how well the profiles and variability were captured in the model using visual predictive checks (VPCs) for each population. The MHRA's scientific evaluation and acceptance of this quantitative evidence are documented in the Public Assessment Report (PAR) for Eqjubi [6].

2.2. Lecanemab (Leqembi)

2.2.1. Context of Use (ICH M15)

For lecanemab (Leqembi), quantitative medicine was applied during different stages of regulatory assessment with a number of related contexts of use. These included the characterization of pharmacokinetic variability, to evaluate the potential need for dose adjustment based on intrinsic and extrinsic factors and to support the interpretation of biomarker and clinical outcome data within the overall benefit–risk assessment.

2.2.2. Model Risk (ICH M15), Key Uncertainties, and Risk Mitigation

The influence of lecanemab quantitative models varied by model type. The Population PK model was used to inform exposure characterization and covariate evaluation. In addition, exposure–biomarker and integrated analyses provided supportive influence in the interpretation of biomarker trajectories and clinical outcomes. These models were not used as standalone evidence of clinical benefit but as tools to support interpretation of a comprehensive clinical data package. Therefore, the model influence was considered moderate to low as regulatory conclusions were driven mainly by clinical evidence rather than model outputs alone. Lecanemab has a relatively wide therapeutic index and the dosing strategy is supported by extensive clinical experience and predefined monitoring recommendations. As a result, modest deviations in exposure predictions derived from quantitative models were unlikely to lead to immediate or irreversible patient harm. Therefore, the consequence of a wrong decision informed by these models was considered low as dosing and risk management decisions were not solely dependent on the modeling outcome.

In this situation, overall model risk was mitigated by the limited influence of the models on regulatory decisions and by the availability of complementary clinical and biomarker data. In line with ICH M15, model evaluation and outputs were assessed in a conservative way. Uncertainty was evaluated and conclusions considered in line with the supportive interpretation instead of confirmatory decision‐making. This proportional approach confirmed that the level of model evaluation and documentation was appropriate to the anticipated consequence of a wrong decision.

Key uncertainties for lecanemab related to whether residual inter‐individual variability in exposure, remaining after inclusion of identified covariates, could result in clinically meaningful differences across patient subgroups. Covariates such as body weight, sex, albumin, ethnicity and anti‐drug antibody status influenced pharmacokinetic parameters. However, none of these effects were considered clinically meaningful. In addition, the consequence of a wrong decision was considered low because of the relatively wide therapeutic index and absence of clinically meaningful exposure differences. These uncertainties were mitigated by indicating that the magnitude of covariate effects on exposure was not clinically relevant and did not warrant dose adjustment. In line with ICH M15, the modeling strategy, level of documentation and regulatory reliance were proportionate to the lower model influence and consequence of a wrong decision. The MHRA's assessment of these quantitative elements is described in the PAR for Leqembi [7].

3. Discussion

Figure 1 illustrates model risk that is determined by the combination of model influence and the consequence of a wrong decision. Sugemalimab occupies a higher‐risk region due to greater model influence and clinically meaningful consequences of incorrect dose selection. In contrast, Lecanemab occupies a lower‐risk region which reflects limited model influence and lower consequence. Table 1 summarizes how the ICH M15 framework was applied to case studies of Sugemalimab and Lecanemab. The table highlights common principles and important differences in context of use, model influence and decision impact. These examples show that quantitative models can be used in different ways ranging from guiding dosing decisions to assist the interpretation of data over time while being assessed using the same framework. Although model influence was moderate or low at the assessed stage, similar modeling approaches may play a more influential role at later stages when regulatory questions evolve which is consistent with the iterative nature of MIDD. Additionally, whilst both case studies focused on population PK modeling for monoclonal antibodies, ICH M15 principles are also applicable to other therapeutic modalities and a range of quantitative methods, as illustrated in Table 2. These include small molecules, cell therapies and gene therapies which may have different model influence and consequence of a wrong decision. However, these can be assessed using the same principles. From a clinical and regulatory science perspective, the cases demonstrate the importance of ensuring that the level of model evaluation is commensurate with model risk (considering model influence and the consequence of a wrong decision), as described in ICH M15. In Sugemalimab case, a clearly defined context of use and high decision impact justified the regulatory reliance on robust PK exposure matching with key uncertainties acknowledged and mitigated. In contrast, the Lecanemab case demonstrates how multiple models with varying influence could be integrated to support understanding of disease biology, treatment effects and monitoring considerations without making more claims than the model could support. For both products, the key regulatory conclusions were generally consistent with those reached by other major agencies. For example, the sugemalimab dose of 1500 mg for individuals weighing more than 115 kg, which was supported by modeling and approved by the MHRA, was also approved by the Committee for Medicinal Products for Human Use [8]. Whilst this demonstrates how ICH M15 represents a formalization of the principles that regulators currently follow during assessment, other case studies (e.g., Altuvoct [9]) show discrepancies between regulatory decisions. It is expected that formal adoption of ICH M15 will be supportive of more consistent regulatory evaluation. In addition, the two case studies show how the regulatory policy that is aligned with ICH M15 can be used to support the proportionate use of quantitative methods to link complex modeling approaches with regulatory decisions and enable patient access to medicines.

FIGURE 1.

Model risk matrix illustrating the relative positioning of Sugemalimab and Lecanemab under the ICH M15 framework.

TABLE 1.

Summary of model influence, consequence of wrong decision, and model risk (ICH M15).

Product	Model influence	Consequence of a wrong decision	Overall model risk (ICH M15)
Sugemalimab (Eqjubi)	Moderate to high. Quantitative modeling played a pivotal role in dose justification through exposure matching and informed regulatory assessment beyond descriptive analysis	Clinically meaningful. Inappropriate dose selection could plausibly result in loss of efficacy or increased toxicity, particularly when extrapolating exposure across body‐weight ranges in the absence of a defined exposure–response relationship	Higher model risk. Driven by the combination of significant model influence and a meaningful consequence of a wrong decision that require conservative assumptions, transparent reporting and integration with the totality of evidence
Lecanemab (Leqembi)	Moderate to low. Population PK modeling was used to characterize variability and assess the need for dose adjustment. However, key regulatory decisions relied mainly on clinical efficacy and safety data	Limited. Modest exposure misprediction was unlikely to result in immediate loss of efficacy or unacceptable toxicity due to a relatively wide therapeutic index and extensive clinical safety monitoring	Lower model risk. Results from limited model influence and a lower anticipated consequence of a wrong decision which allowed for the use of modeling for descriptive and characterization purposes rather than critical decision making

TABLE 2.

Examples of how ICH M15 principles could be applied across different therapeutic modalities.

Therapeutic modality	Typical quantitative methods	Example regulatory questions	Model influence and consequence considerations (ICH M15)
Small molecules	Population PK, exposure–response, PBPK	Dose selection, DDIs, special populations	Model influence may be high when replacing clinical studies (e.g., DDI waivers). Consequence of wrong decision depends on therapeutic index and disease severity
Monoclonal antibodies	Population PK, PK/PD, exposure–biomarker models	Dose justification, covariate effects, extrapolation	Often moderate model influence, with consequence driven by exposure sensitivity and safety margins, as illustrated by the present case studies
Cell therapies	PK‐like cellular kinetics, QSP	Dose selection, persistence, variability	Model influence typically lower; consequence of wrong decision may be high due to irreversible effects which require conservative assumptions
Gene therapies	Translational PK/PD, durability models	Dose justification, long‐term expression	Models often support interpretation instead of decision‐critical use; consequence of wrong decision may be high taking into consideration the limited reversibility
Pediatric development	Population PK, extrapolation frameworks	Dose selection, bridging	Model influence could be high when supporting extrapolation; consequence depends on developmental sensitivity and safety margins

4. Conclusions

Quantitative medicine is an important factor in modern drug development and regulatory decision‐making, particularly when applied with clarity and transparency. Sugemalimab and Lecanemab case studies demonstrate how the ICH M15 framework can be applied in diverse therapeutic areas and regulatory decisions. These examples demonstrate that the value in the systematic alignment of context of use, model influence and decision impact. As quantitative approaches are increasingly used to support regulatory decisions, ICH M15 provides a common language to support their responsible and effective integration into regulatory and translational decision‐making.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.