This study introduces a novel drug-disease modeling framework designed to assess the benefit-risk balance of antibody-drug conjugates (ADC) in oncology. The framework integrates dose levels, pharmacokinetics, tumor growth dynamics, progression-free survival (PFS), and dose-adjusted adverse events. We demonstrated this through its application to tusamitamab ravtansine (Tusa), an ADC targeting Carcinoembryonic Antigen-Related Cell Adhesion Molecule 5 in non-squamous non-small cell lung cancer (nsq NSCLC). We developed our model using phase I trial safety data from 254 patients (doses: 5-190 mg/m2) and efficacy data from 88 nsq NSCLC patients (dose 100 mg/m2). This model accurately predicted phase III outcomes for the Tusa arm via an iterative simulation. Using phase III baseline characteristics, simulations of Tusa doses comparing three dose levels (80, 100, and 120 mg/m2 every 2 weeks) revealed a critical trade-off: while higher doses increased response rates, they also substantially increased corneal toxicity without improving survival. These findings demonstrate how early-phase data can inform optimal dose selection by quantifying benefit-risk. This robust framework and methodology is generalizable beyond Tusa, offering value to support dose selection and trial decision-making in oncology drug development.

Clinical pharmacology and pharmacometrics are central to understanding patient response to new and existing therapies, yet access to training in Africa remains limited, particularly in francophone countries where language is a barrier. We report the design and implementation of a 12-week French-language online course combining asynchronous lessons with live interactive sessions. The inaugural cohort enrolled 72 students from Senegal (n = 37), Tunisia (n = 24), and the Democratic Republic of the Congo (n = 11). While only one-third completed the full program, all survey respondents judged the course useful, highlighting clear objectives and adequate preparation time. Challenges related to perceived workload, time-zone difference in scheduling, and internet connectivity were reported by participants. A second iteration of the course hosted by the Tunisia Chapter of Pharmacometrics Africa has just completed and will gather further feedback and attempt to address lessons learnt. This proof-of-concept hands-on educational program demonstrates that delivering training in French through an online format can effectively expand access to pharmacometrics education in Africa. Strengthening local institutional partnerships to take on technical support will be critical to improving participant retention and ensuring program sustainability. Moreover, this model could be adapted to other languages and extended to additional regions, thereby promoting more inclusive global capacity building.

Introduction: Normalised prediction distribution errors (npde) are used to graphically and statistically evaluate continuous responses in non-linear mixed effect models. Here, our aim was to extend npde for categorical data and to evaluate their performance. We applied our approach to a real case-study describing the evolution of severe onychomycosis (toenail infection) in a trial comparing two treatment groups. Methods: Let V denote a dataset with categorical observations. The null hypothesis H0 is that observations in V can be described by a model M. Residuals called npde can be adapted to categorical observations using jittering techniques. Their theoretical standard normal distribution can be evaluated through the Kolmogorov-Smirnov test. We evaluated the performance in terms of power through a simulation and compared it to a Chi-square. We illustrated the test and graphs on a real case-study. Results: npd were able to detect misspecifications in the structural model and model parameter value. As expected, the power to detect model misspecifications increased both with the difference in the shape of the probability, and with the sample size. Chi-square test performed better but npd could be readily applied in all type of design. Based on the toe-nail data, graphs reveal a huge discrepancy of the base model, and a good adequation for the best model we found. Conclusions: npde can be extended to categorical data, particularly in clinical settings with unbalanced design and graphs can be useful to evaluate the model as well as the covariate effects.

ABSTRACT Selecting the most relevant models for model‐informed precision dosing (MIPD) remains challenging, particularly when numerous population pharmacokinetic (popPK) models are available, making external evaluation of all candidates time‐ and resource‐consuming. We aimed to develop and assess an adaptable relevance‐based scoring methodology, illustrated with tacrolimus in adult transplant recipients. Candidate criteria were derived from published guidelines and external evaluations. A 23‐item relevance score was developed, covering training dataset, study design, evaluation methodology, covariates, and parameter precision. For tacrolimus, a systematic literature review identified 28 models which were scored for relevance and externally evaluated using an independent dataset from Rennes University Hospital (86 patients). Predictive performances were assessed at individual and population levels using bias and imprecision metrics. Predicted and observed AUCs were compared against therapeutic thresholds. Models were then ranked using an external evaluation score (maximum 14 points) based on predefined acceptance limits for bias and imprecision. Correlations between relevance‐based and external scores and ranks were assessed using Pearson and Spearman correlations. The mean relevance score was 22.3/56 (range: 6–38). Higher scores were associated with larger and richer datasets, prospective or multicentric designs, inclusion of clinically relevant covariates, and precise parameter estimates. Individual predictions showed acceptable bias and imprecision, whereas population predictions were consistently inaccurate. AUC concordance (56%–94%) was mainly limited by overestimation. Highest external score was 6/14. Correlation between relevance‐based and external scores was moderate (Pearson = 0.56; Spearman = 0.61). This work provides a structured basis for popPK model selection and identified tacrolimus models suitable for MIPD applications.

Introduction: Joint models are increasingly used in clinical trials. An important part of model building is to properly assess the descriptive and predictive ability of these models. Normalised prediction discrepancies (npd) and normalised prediction distribution errors (npde) have been developed to evaluate graphically and statistically non-linear mixed effect models for continuous responses. In this work, we propose to use a combined test to evaluate joint models. Methods: Prediction discrepancies (pd) are defined as the quantile of the observation within its predictive distribution and obtained by Monte-Carlo simulations. The pd for unobserved (censored) event times are imputed in a uniform distribution based on the model prediction of the probability of censoring, using a similar method as the one developed to handle data under the lower quantification limit (LOQ). We propose to combine the p-values of the tests on longitudinal data and on time-to-event (TTE) data, adjusted with a Bonferroni correction. We performed simulation studies based on a joint model characterising the relationship between prostate specific antigen biomarker (PSA) and survival in prostate cancer patients to evaluate the type I error and power of npd/npde to detect different types of model misspecifications. Results: For all types of misspecifications, the type I error of the combined test was found to be close to the expected 5 Conclusions: npd can be readily extended for event data by imputing the pd for censored event under the model. The test showed an adequate type I error, and was quite sensitive to alternative models tested.

Abstract We propose two new tests to detect drug effects (DE) in trials of one to very few patients followed during two periods (before and after initiation of a treatment). Both methods use longitudinal natural history data to inform the estimation of each patient’s DE. The first method uses a non-linear mixed effect model (NLMEM) reflecting an expected natural history with a hypothetical drug effect, to estimate the Conditional Distribution of the Drug Effect (CDDE). The second method trains a Pareto Depth Analysis (PDA) algorithm, a machine learning–based approach based on outlier detection, that we implement using data simulated under the NLMEM. We evaluated the two tests with a simulation study. We used data from the PROSPAX study in Autosomal Recessive Cerebellar Ataxias (ARCAs, to derive a NLMEM for the Scale for the Assessment and Rating of Ataxia score. The CDDE method provided controlled type I error and, in some scenarios, adequate corrected power, though sensitivity analyses showed vulnerability to misspecification. The PDA method demonstrated lower statistical power except with high score precision. These results highlight different strategies for quantifying treatment effects in ultra-rare, patient-specific trials. They can inform methodological design for future ARCA precision therapies.

Background and Objective: Uncertainty in non-linear mixed effect models is often assessed using the Fisher information matrix to derive the standard errors of estimation. The bootstrap is an alternative to the asymptotic method, with different approaches to handle the different levels of individual and population variabilities. The simplest method is the Case bootstrap where the entire vector of individuals is resampled, but this approach does not take into account the hierarchical nature of non-linear mixed effect models (NLMEM). Methods: We propose here a non-parametric bootstrap, cNP, to preserve the structure of the original data. We resample interindividual random effects from the conditional distribution of the individual parameters, obtained as a by-product of the SAEM algorithm, and residuals from their distribution. cNP was implemented in the saemix package for R along with the case, parametric (Par), and non-parametric (NP) residual bootstraps. Coverage rates were compared in a simulation study using sigmoid Emax models, with rich, sparse and unbalanced designs, and 3 levels of residual variability. Results: The asymptotic method tended to produce lower than theoretical coverages for the variance terms. Bootstraps provided more adequate coverage, but none of the approaches maintained coverage when the residual error increased. Overall, the new cNP and the Case provided better coverage than the classical NP. Conclusion: The new conditional non-parametric bootstrap can be used when it is important to preserve the structure of the original dataset, such as the number of observations or the repartition of covariates as it does not require stratification.

Background and Objectives: Longitudinal data are increasingly collected in clinical trials to provide information on treatment action and disease evolution. The trajectory of continuous biomarkers such as target hormone concentrations or viral loads can then be modelled in relationship to the occurrence of events such as recovery or hospitalisation. Other studies may include repeated measurements of discrete pain scores, number of episodes (count) or occurrence of events (survival). Non-linear mixed-effect models (NLMEM) can handle individual differences in trajectories while modelling the underlying population evolution and are the natural choice for their analysis. The saemix package for R is one of the few open-source solutions and the most flexible. In this paper, we extend it to accommodate a variety of models for non-Gaussian data. Methods: The saemix package estimates parameters through the Stochastic Approximation Expectation-Maximisation (SAEM) algorithm. Within the package, non-Gaussian models are specified by their log-likelihood functions, affording maximal control over model formulation. We extend estimation algorithms as well as exploratory and diagnostic plots for non-Gaussian data. Bootstrap approaches were implemented to estimate parameter uncertainty. To evaluate the performance of saemix, we performed a simulation study based on the toenail dataset, containing repeated binary data from a randomised clinical trial. Results: saemix showed good performance to recover the true parameter values in the simulation study, and was stable across different starting values for the parameters. An algorithm jointly searching for covariate and interindividual variability model was also implemented to build the covariate model and applied to categorical and survival-type data.

In the context of clinical research, computational models have received increasing attention over the past decades. In this systematic review, we aimed to provide an overview of the role of so-called in silico clinical trials (ISCTs) in medical applications. Exemplary for the broad field of clinical medicine, we focused on in silico (IS) methods applied in drug development, sometimes also referred to as model informed drug development (MIDD). We searched PubMed and ClinicalTrials.gov for published articles and registered clinical trials related to ISCTs. We identified 202 articles and 48 trials, and of these, 76 articles and 19 trials were directly linked to drug development. We extracted information from all 202 articles and 48 clinical trials and conducted a more detailed review of the methods used in the 76 articles that are connected to drug development. Regarding application, most articles and trials focused on cancer and imaging-related research while rare and pediatric diseases were only addressed in 14 articles and 5 trials, respectively. While some models were informed combining mechanistic knowledge with clinical or preclinical (in-vivo or in-vitro) data, the majority of models were fully data-driven, illustrating that clinical data is a crucial part in the process of generating synthetic data in ISCTs. Regarding reproducibility, a more detailed analysis revealed that only 24

Parallel designs with an end-of-treatment analysis are commonly used for randomised trials, but they remain challenging to conduct in rare diseases due to small sample size and heterogeneity. A more powerful alternative could be to use model-based approaches. We investigated the performance of longitudinal modelling to evaluate disease-modifying treatments in rare diseases using simulations. Our setting was based on a model describing the progression of the standard clinician-reported outcome SARA score in patients with ARCA (Autosomal Recessive Cerebellar Ataxia), a group of ultra-rare, genetically defined, neurodegenerative diseases. We performed a simulation study to evaluate the influence of trials settings on their ability to detect a treatment effect slowing disease progression, using a previously published non-linear mixed effect logistic model. We compared the power of parallel, crossover and delayed start designs, investigating several trial settings: trial duration (2 or 5 years); disease progression rate (slower or faster); magnitude of residual error ( σ =2 or σ =0.5); number of patients (100 or 40); method of statistical analysis (longitudinal analysis with non-linear or linear models; standard statistical analysis), and we investigated their influence on the type 1 error and corrected power of randomised trials. In all settings, using non-linear mixed effect models resulted in controlled type 1 error and higher power (88

In early clinical trials, incorporating biological mechanisms of drug action in model-based drug development may improve Phase I success rates compared to approaches neglecting established mechanisms. Our goal is to investigate how pharmacokinetics (PK) knowledge is introduced in dose-finding methods and assess the performance of Bayesian designs incorporating PK data to estimate toxicity and robustness to misspecifications. Following a literature review, three approaches to integrate PK data into toxicity estimation were selected. The first approach assumes a normal distribution for the Area Under the Curve (AUC). The second method estimates a population PK model from longitudinal concentration data to compute the AUC for each patient. The third considers latent PK profiles to measure drug exposure. Different scenarios were implemented reflecting assumptions about the maximum tolerated dose (MTD) position and misspecifications in PK exposure measures or the PK model. Dose-finding methods were compared using the probability of correct MTD selection and the estimated probability of toxicity at each dose. PK dose-finding designs performed well in terms of accurate MTD selection and were at least as effective as a method without PK. They were robust to underlying PK model misspecification and incorrect exposure measure. Additionally, these methods can assess the dose-toxicity curve.

IntroductionWhole-body dynamic (WB4D) positron emission tomography (PET) imaging data using radiolabeled analogs of drugs are mostly analyzed using descriptive approaches, with no relationship to traditional pharmacokinetic studies based on blood sampling. Here, we build a pharmacokinetic (PK) model from WB4D PET data obtained using a microdose of radiolabeled glyburide ([11C]glyburide) in humans, aiming to describe the biodistribution of this drug and compare estimated pharmacokinetic parameters with the parameters obtained in standard PK studies.MethodsThe present work analyzes data acquired over 40 min after injection of [11C]glyburide in 16 healthy subjects using non-linear mixed-effect models (NLMEM). In 10 subjects, a second PET acquisition was performed after rifampicin administration, which may cause a drug-drug interaction and inhibit the liver uptake transport of glyburide. Arterial blood, liver, kidneys, pancreas, and spleen kinetics were modeled using NLMEM. The model-building strategy involved selecting the structural model using baseline [11C]glyburide PET data and then selecting the covariate model (rifampicin, age, and gender) and refining the structure of the interindividual variability model using both administration periods. Model selection was based on the corrected Bayesian information criterion and implemented in Monolix software.ResultsThe final model included seven compartments, with two compartments each for the Liver and kidneys to account for within-tissue exchanges. Rifampicin decreased the Liver distribution by 261%.DiscussionThe estimated central volume of distribution (V = 3.6 L) and elimination rate (k = 0.8 h-1) were consistent with the known pharmacokinetics of glyburide, which is a promising first step in leveraging microdose data to study the WB4D biodistribution.RegistrationEudraCT identifier no. 2017-001703-69

The development of new treatments for rare neurological diseases (RNDs) may be very challenging due to limited natural history data, lack of relevant biomarkers and clinical endpoints, small and heterogeneous patient populations, and other complexities. A systematic approach is needed for comparing various design and analysis strategies to identify "optimal" approaches for a clinical trial in a chosen RND with the given resource constraints. For this purpose, we propose a pharmacometrics-informed clinical scenario evaluation framework (CSE-PMx), which includes some important research hallmarks relevant to RND clinical trials: a disease progression model for simulating individual longitudinal outcomes, the choice of a suitable randomization method for trial design, and an option to perform subsequent statistical analysis with randomization tests. We illustrate the utility of CSE-PMx for an exemplary randomized trial to compare the disease-modifying effect of an experimental treatment versus control in patients with Autosomal-Recessive Spastic Ataxia Charlevoix Saguenay (ARSACS). In the considered example, our simulation evidence suggests that a nonlinear mixed-effects model (NLMEM) with a population-based likelihood ratio test analysis is valid, robust, and more powerful than some conventional methods such as two-sample t-test, analysis of covariance (ANCOVA), or a mixed model with repeated measurements (MMRM). Our proposed framework is very flexible and generalizable to clinical research in other rare disease indications.

The objective of this work was to review the modelling and simulation methods used in two selected case-studies from the Invents project, with a focus on longitudinal data modelling, simulation methods and extrapolation approaches (across populations, across indications or across similar drugs) applied during the later phases of drug development. We included publications based on relevance to the case-studies involved in Invents, extending the search to articles reporting results of modelling, simulation, or extrapolation approaches for the two drugs secukinumab and tocilizumab through keyword-based literature searches. When outcomes were measured repeatedly during the study, the focus of the analyses was usually along the lines of improvement at a given time-point, with other time-points included in secondary analyses. When PK (pharmacokinetic) or PK/PD (pharmacodynamic) models were developed, they were generally simple, with compartmental models for PK and direct or indirect response models for PD. Clinical questions addressed in the studies selected in this review were mainly centered on efficacy (57%, including exposure-response relationships and efficacy concerning patient-reported outcomes) and safety (34%). Besides the few studies investigating PK, other clinical questions focused on finding predictors of response, either biomarkers related to changes in the outcome under study, molecular characteristics of certain groups of patients, or prognostic factors at baseline. Finally, a couple of studies were more methodological. We found almost no paediatric study, despite modelling and simulation analyses being put forward in the Paediatric Investigation Plan, which could be due to the time needed to perform these studies, lack of interest in obtaining paediatric approval, waivers or failure to publish. Despite the clear focus on longitudinal modelling and simulation in our search, we found very few examples of published PK or PK/PD models for the two case-studies considered, and almost no extrapolation approaches for paediatric or disease applications.

Levofloxacin is a valuable antibiotic in the treatment of bone and joint infections. Due to the known risk of treatment failure and bacterial resistance, the development of tools facilitating pharmacokinetic/pharmacodynamic parameter monitoring to inform precision dosing is needed. Therefore, we assessed the use of Bayesian estimation to predict AUC0-24 and Cmax of levofloxacin under various sampling scenarios realistic in clinical routine. Furthermore, we developed a free web-based application allowing model-informed precision dosing. All published population pharmacokinetic models of ofloxacin and levofloxacin in bone and joint infections were researched and their predictive performance was compared using a real-life data cohort. We used simulated data to validate the robustness of various scenarios for Bayesian estimation of AUC0-24 and Cmax with up to three samples, including the potential impact of an incorrectly reported sampling time at peak concentration. Relative bias and relative root mean square error were estimated to assess accuracy and precision, respectively. One of the three published levofloxacin models showed negligible mean relative prediction error (0.02 ± 0.09). We modified it to a closed form approximation to facilitate the implementation in the application. The 2-sample scenario (T0h-T3h) allowed accurate and precise AUC0-24 estimations for 500 mg q12h and 750 mg q24h dosing regimens. None of the tested scenarios allowed a satisfactory estimation of Cmax. We finally developed and validated a free web-based application (https://levoshiny.iame-research.center/) using the selected model. The Shiny application will be useful in clinical practice to individualize dose regimens based on the AUC0-24 obtained using the proposed limited sampling strategy.

The advent of digital twins in pharmacology presents transformative potential for precision medicine, enabling personalized treatment optimization through dynamic computational simulations of drug interactions at molecular, cellular, and patient levels. These advanced virtual replicas of a patient's biological system are designed to predict individual therapeutic responses with high fidelity, thereby moving beyond the one-size-fits-all paradigm. This paper explores the concept of digital pharmacological twins, detailing how they can integrate heterogeneous data, including multi-omic, pharmacokinetic, pharmacodynamic, clinical, and environmental information, and employing a synergy of advanced mechanistic and machine learning models. Using illustrative examples from ongoing international initiatives, this work highlights the methodological frameworks necessary for developing and validating such comprehensive predictive tools. We underscore the critical importance of model interoperability, robust data integration strategies, and rigorous validation to ensure clinical utility. Ultimately, digital pharmacological twins promise to enhance therapeutic efficacy, minimize adverse drug reactions, and accelerate the translation of pharmacological science into tangible patient benefits.

Mycophenolic acid (MPA), administered as its prodrug Mycophenolate mofetil (MMF), is widely used to prevent rejection after organ solid transplantation. Few population pharmacokinetic studies of MPA in lung transplantation are available, so the present study was designed to describe the population pharmacokinetics (PK) of MPA and develop a maximum a posteriori Bayesian estimator of MPA area under the curve (AUC0-12h) in lung transplant (LT) recipients in the early post-transplant period. We conducted a single-center retrospective study in the Plessis-Robinson hospital (France), including LT recipients receiving MMF in whom at least one concentration-time profile was available during the first 3 months post-transplantation. MPA was measured before intake and 0.5, 1, 2, 4 ± 6 hours post-administration. Patients were divided into an index group (N=72) and a validation group (N=46). A population PK model and a Bayesian estimator with three sampling times were built with the index dataset and externally evaluated with the validation dataset. Analyses were performed using nonlinear mixed-effects models with Monolix® software. The PK model was built using 97 MPA profiles in the index dataset. MPA PK was best described using a two-compartment model with a lag time and first-order absorption and elimination. The significant covariates on the MPA apparent clearance were creatinine, body weight and ciclosporin co-administration and the posttransplantation time on the apparent volume of the central compartment. The model was successfully evaluated in the 60 MPA profiles from the validation dataset. The best Bayesian estimator included samples just before intake, 1 and 4 hours post-administration. The prediction of AUC was unbiased in both datasets and had a precision around 20 %. This is the first Bayesian estimator allowing the prediction of AUC in LT patients without cystic fibrosis in the early post-transplant period, providing a tool to improve the therapeutic drug monitoring of MPA.