CPT: Pharmacometrics & Systems Pharmacology最新文献

In silico clinical trials (ISCT) are computational frameworks that employ mathematical models to generate virtual patients and simulate their responses to new treatments, treatment regimens, or medical devices via simulations mirroring real-world clinical trials. ISCTs are an important component of the model-informed drug development (MIDD) framework for optimizing therapies, treatment personalization, informing regulatory decisions, and accelerating overall drug development by enhancing R&D productivity. However, the emergence of complex models, such as quantitative systems pharmacology (QSP) models, presents significant challenges for their effective implementation. Guidelines for conducting ISCTs have been published to address these challenges, focusing on algorithms and credibility frameworks for generating plausible virtual patients and calibrating virtual populations. However, it is not straightforward to apply existing workflows to models where parameter distributions and correlations are estimated using nonlinear mixed effects (NLME) population fitting approaches, a common practice in the pharmaceutical industry when individual-patient-level data is available. Here, we illustrate a modeling workflow for conducting ISCTs with NLME models, detailing key considerations, methods, and challenges at each step. We demonstrate the practical implementation of this workflow through two examples to showcase its broad applicability: (1) a simple model predicting tumor growth in response to chemotherapy and (2) a more complex mechanistic QSP model of hepatitis B virus infection that captures the physiological mechanisms underlying treatment response with standard-of-care therapies.

Ibrutinib, a Bruton's tyrosine kinase (Btk) inhibitor, is a key therapy for chronic lymphocytic leukemia (CLL). In clinical practice, adverse events, such as hypertension, frequently necessitate dose reductions or treatment discontinuation. Emerging evidence suggests that reduced doses may retain clinical efficacy while mitigating toxicity. The synergistic ibrutinib–venetoclax combination remains understudied at low doses, particularly for ibrutinib. This study aimed to explore dose optimization strategies, with/without venetoclax, in treatment-naïve (TN) and relapsed/refractory (R/R) CLL using mechanism-based, model-informed approaches to characterize the relationship between systemic ibrutinib exposure and efficacy and safety biomarkers. We leveraged data from phase 1b/2 and 3 studies, including plasma concentrations, leukocyte and lymphocyte counts, lymph node and spleen size measurements, and blood pressure. A previously developed semi-mechanistic population pharmacokinetic-pharmacodynamic (PKPD) framework was re-evaluated, extended by integrating additional biomarkers and identifying differences between TN and R/R patients, and used to simulate alternative dosing strategies. The model successfully captured the temporal dynamics of all biomarkers simultaneously. We quantified a 76% longer phospho-Btk half-life and a 43% shorter peripheral CLL cell half-life in TN versus R/R patients, with no evidence of ibrutinib resistance in TN patients. Dose reductions based on response depth or toxicity preserved comparable response rates and progression-free survival to standard dosing. Ibrutinib de-escalation schedules with venetoclax resulted in a ≤ 5% reduction in peripheral blood measurable residual disease compared to standard dosing at 2 years. This PKPD framework supports dose individualization to improve tolerability without sacrificing treatment outcomes, offering a path toward more personalized, effective CLL management.

Assessing the risk or benefit of an inhaled substance is challenging due to the variety of forms it can take (gas, vapor, particle, or droplet) and the complex biological processes involved in its uptake and retention. Physiologically based kinetic (PBK) models offer an alternative to in vivo experiments. However, PBK models for inhalational uptake are to date either designed for gases/vapors or airborne particulates, often with only low regional compartmentalization. The here-presented, newly developed model combines both applications. Its mechanisms are an amalgamation of PBK and non-PBK models integrated into a multicompartmental description of the human lung to include the relevant uptake and clearance processes in the different lung regions, of which macrophage-mediated dissolution is novel to PBK modeling. The model was designed to use a minimal number of substance-specific input parameters, which can be derived from in vitro assays or in silico methods. Model predictions for hypothetical substances with varying physicochemical properties were performed, along with rudimentary sensitivity analyses to identify the most important parameters for gases/vapors and particles. This novel PBK model combines all these aspects and provides in silico predictions of systemic and local lung concentrations, reducing uncertainty in risk assessments and supporting drug development. It serves as a valuable tool to translate nominal ambient air doses into effective localized doses within the lung.

Amlitelimab is a fully human, nondepleting, anti-OX40 ligand monoclonal antibody being investigated for the treatment of moderate-to-severe atopic dermatitis (AD) in adults and adolescents. Population pharmacokinetic (PopPK) and pharmacokinetic/pharmacodynamic-Eczema Area and Severity Index (PopPK/PD-EASI) models were used to inform dosing regimen selection for amlitelimab phase 3 trials. The PopPK model was developed using phase 1 (healthy volunteers) and phase 2 (participants with AD) trial data, including individual exposure variables from the STREAM-AD phase 2b trial following subcutaneous amlitelimab doses ranging from 62.5 to 250 mg given every 4 weeks (Q4W). The PopPK model was used to compute exposures for an extended dosing regimen of 250 mg Q12W (with 500 mg loading dose [+LD]). The PopPK/PD-EASI model was developed from phase 2 trials to predict treatment responses (EASI values) with selected dosing scenarios. Finally, the dose for individuals with lower body weight (i.e., < 40 kg) was determined. Utilizing the PopPK model, the amlitelimab 250 mg Q12W + LD computed exposures were between the exposures of 62.5 mg Q4W and 250 mg Q4W + LD efficacious doses in the STREAM-AD trial. Using the PopPK/PD-EASI model, the simulated efficacy for dosing scenarios of 250 mg Q12W + LD regimen from initiation or 250 mg Q4W + LD for 24 weeks followed by Q12W to Week 60 was similar to continuous 250 mg Q4W. Simulations identified that a twofold dose reduction would allow participants < 40 kg to achieve amlitelimab exposures within the range observed in participants ≥ 40 kg on 250 mg Q4W or Q12W. These results support evaluation of a Q12W dosing regimen for adults and adolescents in phase 3 trials.

Datopotamab deruxtecan (Dato-DXd) is a trophoblast cell surface antigen 2 (TROP2)-directed antibody-drug conjugate (ADC) developed for the treatment of advanced or metastatic solid tumors. Dato-DXd demonstrated promising antitumor activity and a manageable safety profile in the first-in-human Phase 1 study TROPION-PanTumor01. The present work established population pharmacokinetic models for Dato-DXd and DXd in patients with advanced solid tumors, using data from three clinical studies. The final Dato-DXd model was a two-compartment model with both linear (CL_linDatoDXd) and nonlinear elimination. CL_linDatoDXd was the major elimination pathway for Dato-DXd doses ≥ 4 mg/kg, whereas nonlinear clearance was necessary to capture the concentration-dependent clearance at lower doses. Body weight was added as a mechanistic covariate with a fixed allometric exponent of 0.75 (estimated value: 0.80) on clearance and estimated exponents on volumes of distribution. The final model for DXd was a one-compartment model with linear clearance (CL_DXd). The release of DXd was equal to the total elimination rate of Dato-DXd and was found to be time-dependent within and between cycles, as previously observed for other ADCs. For a typical 66 kg male patient, CL_linDatoDXd was 0.386 L/day, central volume of distribution V_cDatoDXd was 3.06 L, CL_DXd was 2.66 L/day, and V_cDXd was 25.1 L. Covariate analysis identified body weight as the most influential covariate on Dato-DXd and DXd exposure. These findings support the weight-based dosing strategy and indicate no dose adjustments are warranted based on the covariates evaluated.

Bispecific antibodies can be broadly divided into two categories: those that are pharmacologically active as either dimers bound to a single target or as trimers bound to both targets, and those that are only active as trimers. Dose selection of trimer-based bispecifics poses a unique challenge, as toxicity increases with dose, but efficacy does not. Instead, trimer-driven bispecifics have a bell-shaped efficacy curve, for which both under- and over-dosing can cause a decrease in efficacy. To address the challenge of dose selection for trimer-based bispecifics, we develop a semi-mechanistic pharmacokinetic/pharmacodynamic model of one such bispecific, teclistamab. By introducing variability in key patient-specific parameters, we find that the currently selected phase II recommended dose of 1.5 mg/kg administered subcutaneously weekly falls within the calculated optimal range for maximizing concentration of the pharmacologically active trimer for a broad population. We next explore different strategies for patient stratification based on pre-treatment levels of measurable biomarkers. We discover that significantly more variability across subpopulations is predicted when the drug is administered every 2 weeks as compared to weekly administration, and that higher doses generally result in more interpatient variability. Further, the pharmacologically active trimers are predicted to be maximized at different doses for different subpopulations. These findings underscore the potential for model-supported patient stratification based on measurable biomarkers, offering a middle ground between population-level approaches and fully personalized medicine.

Pembrolizumab is an immune checkpoint inhibitor that has been approved for more than 20 different indications and has shown great survival benefits in various types of cancer. However, the reported benefits of pembrolizumab in patients' quality of life (QoL) have been inconsistent across different studies and different types of cancer. As oncology drug development increasingly emphasizes patient-centered care, patient-reported outcomes (PROs), particularly patient-reported QoL, are recognized as important clinical endpoints. To characterize the effects of pembrolizumab on patient-reported QoL, we conducted a model-based meta-analysis (MBMA) of published clinical trials evaluating pembrolizumab across different types of cancer. The longitudinal EORTC QLQ-C30 GHS/QOL data were extracted in our analysis as QoL scores. A population model was developed to characterize the longitudinal QoL trajectories and quantify both treatment toxicity and efficacy. Out of more than 300 screened studies, only 20 reported longitudinal EORTC QLQ-C30 QoL data. Among these, 8 studies reported no between-group differences in QoL outcomes between pembrolizumab and control arms. However, our modeling revealed that pembrolizumab was associated with greater toxicity but improved long-term QoL. Notably, our approach identified treatment effects on QoL that were not detected by traditional statistical analyses in the original publications. In summary, our study demonstrates that MBMA combined with population modeling enables more accurate evaluation of longitudinal PROs data, overcoming the limitations of conventional methods. This approach offers a robust framework for integrating patient-centered outcomes into oncology drug development and supports the broader use of PROs data in regulatory and clinical decision-making.

Binding in cis-configuration to the PD-1 receptor (PD-1) and IL-2$��\mathrm{\beta \gamma }�$ receptor (IL-2R$��\mathrm{\beta \gamma }�$) has been shown to lead to differentiation of CD8 T cells to better effectors, which is anticipated to drive efficacy of the immune-targeted cytokine eciskafusp alfa, or PD1-IL2v. Here we present a geometrically driven mathematical formulation informed by in vitro and early clinical data which enables prediction of doses at which cis-binding is at its highest, and explains observed differences in concentration-time profiles for patients who had recent exposure to other anti-PD-1 molecules compared with those who had not. Furthermore, binding in cis-configuration is expected to follow a “bell-shaped” relationship with drug concentration such that high concentrations may lead to reduced benefit/risk ratio compared with concentrations around the peak of the bell-shape. Model simulations identify patient cohorts for whom the upper limit of the pharmacological dosing range may be defined by either undesirable off-tumor target engagement or a decrease in on-tumor cis-binding.

Conventional exposure–response (E–R) analyses, such as logistic regression and time-to-event analysis using summary exposure metrics, are often conducted in oncology using data from the pivotal trial(s) with a single dose level to support dosing decisions. However, these E–R analyses, affected by multiple confounding factors, may mischaracterize true E–R relationships, potentially limiting their utility in dosing decisions. This study investigates potential mischaracterization in such analyses influenced by the following time-dependent confounding factors: exposure accumulation, dose modification patterns, and event onset time. We used a simulation-based approach to evaluate two E–R scenarios: ER1, where event time generated with a Weibull distribution is not affected by drug exposure, and ER2, where the response is driven by drug exposure via a joint PK-tumor size model. Our analyses indicate that when using time-dependent exposure metrics (e.g., average concentration till event/censoring), exposure accumulation tends to induce an inverse E–R trend, while dose modifications (interruptions/reductions) likely induce a positive E–R trend. Simulations suggest that employing static exposure metrics (e.g., first-cycle or steady-state) minimizes these biases. Additionally, adopting an Emax model aligned with the underground truth in ER2 in the E–R analyses could reduce bias. When significant dose modifications are present, including relevant data from a dose-range study and employing modified methods for time-dependent exposure derivation may help mitigate bias. Overall, using multiple exposure metrics (including static ones) to assess E–R consistency and interpreting the potential causal effects with totality of evidence (including dose–response results) should be implemented to better inform dosing decisions.

A subcutaneous formulation of nivolumab was evaluated in the phase III study CheckMate 67T (NCT04810078). The co-primary pharmacokinetic exposure endpoints, time-averaged serum concentration over the first 28 days (C_avgd28), and steady-state trough concentration (C_minss) were determined through population pharmacokinetic analysis as compared to conventional non-compartmental analysis (NCA). We proposed a model-based approach to determine subcutaneous and intravenous nivolumab exposures in CheckMate 67T, leveraging extensive prior pharmacokinetic data across various tumor types. The robustness of this model-informed drug development (MIDD) approach was assessed via clinical trial simulations. Concentration–time profiles in randomly sampled patients with renal cell carcinoma receiving second-line systemic therapy were simulated for subcutaneous/intravenous nivolumab. Population pharmacokinetic parameters, sampled from the joint parameter uncertainty distribution, were applied in simulations based on the CheckMate 67T design. Two population pharmacokinetic approaches—the PRIOR subroutine ($PRIOR) in NONMEM and a pooled analysis with historical nivolumab pharmacokinetic data—were used to analyze the simulated data. Results were compared to NCA using intensive and conventional sampling schemes. Analyses showed that model-based analysis provided more accurate area under the curve estimates than NCA. Model-predicted exposure measures, including C_avgd28, maximum serum concentration after the first dose, and minimum serum concentration at day 28, were also consistent across both population pharmacokinetic approaches, with minimal differences in geometric means. In conclusion, both the $PRIOR and pooled population pharmacokinetic methods yielded more accurate results compared to conventional NCA. The MIDD approach was validated as a robust and feasible method to support non-inferiority assessment based on clinical trial simulations.