Clinical Pharmacology & Therapeutics最新文献

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are increasingly prescribed for type 2 diabetes and weight management. However, conflicting evidence from preclinical, clinical, and pharmacovigilance studies suggests potential neuropsychiatric effects. This study examined associations between GLP-1RA use and initiation of a range of neuropsychiatric medications using real-world prescription data. A Prescription Sequence Symmetry Analysis was conducted using Australia's Pharmaceutical Benefits Scheme 10% random sample between July 1, 2013 and December 31, 2024. Individuals with an incident dispensing of both a GLP-1RA and a neuropsychiatric medication were included. Outcomes comprised antidepressants, medication for substance use disorder (SUD), antipsychotics, psychostimulants, antidementia medications, antiparkinsonian medications, antiepileptics, and antimigraine agents. Adjusted sequence ratios (aSRs) with 95% confidence intervals (CIs) were calculated using a one-year exposure window, with sensitivity analyses varying various time windows. Among 2,033 individuals, semaglutide was the most common GLP-1RA (50.6%), followed by exenatide (27.5%) and dulaglutide (21.9%). GLP-1RA initiation was inversely associated with initiation of antidepressants (aSR: 0.85; 95% CI: 0.76-0.94) and medications for SUD (aSR: 0.70; 95% CI: 0.51-0.88). No significant associations were observed for other neuropsychiatric outcomes. GLP-1RA use was inversely associated with subsequent initiation of antidepressants and medications for SUD, while no associations were identified for other neuropsychiatric marker medications. These findings highlight the need for further studies to clarify the nature and magnitude of these potential neuropsychiatric associations with GLP-1RAs.

Clopidogrel is first-line for secondary stroke prevention, but its effectiveness is affected by CYP2C19 genetic variation. An American Heart Association (AHA) scientific statement concluded that CYP2C19-guided P2Y12 inhibitor (P2Y12i) therapy is evidence-supported, though real-world application depends on implementation factors (e.g., timely, understandable results). This prospective observational cohort quality improvement project used data from three comprehensive stroke centers within one health system, each offering rapid CYP2C19 testing. Eligible patients had CYP2C19 testing ordered, a P2Y12i prescribed (clopidogrel or ticagrelor), or a P2Y12i indication. Consensus recommendations were clopidogrel for CYP2C19 normal/rapid/ultrarapid metabolizers (NM/RM/UM) and ticagrelor for intermediate/poor metabolizers (IM/PM). The primary target population was patients with acute ischemic stroke or TIA (AIS/TIA) prescribed a P2Y12i. The primary outcome was the proportion of discharge ticagrelor prescriptions (among P2Y12i users) with AIS/TIA, compared between IM/PM and NM/RM/UM groups. Secondary endpoints included test turnaround time (TAT), result availability before discharge, and alignment of discharge P2Y12i with consensus recommendations (PGx-aligned prescribing). The evaluation period was September 2024-August 2025. Results were available for 467 patients; 451 (97%) had results before discharge, with a median TAT of 5.2 hours (interquartile range 3.3-9.6). Of 337 patients prescribed a P2Y12i at discharge, 296 had AIS/TIA. Clinicians more often selected ticagrelor over clopidogrel for IM/PMs than NM/RM/UMs (89 of 102 [87%] vs. 11 of 194 [5.7%]; RR 15.4; 95% CI: 8.6-27.4; P < 0.0001). PGx-aligned prescribing occurred in 272 (92%) of 296 patients. Rapid CYP2C19 testing efficiently translates into tailored P2Y12i prescribing in clinical practice.

Model-informed drug development is increasingly used to support chimeric antigen receptor (CAR)-T-cell therapy programs. However, most population pharmacokinetic (PK) CAR-T-cell models available in literature are variations of an empirical piecewise-linear model, which accurately describes the observed data but lacks a mechanistic foundation. This limits its use for simulations of scenarios beyond the observed data. This work presents a mechanism-based population PK framework that aims to bridge existing empirical and mechanistic CAR-T-cell modeling approaches. The model was developed using a clinical database of 473 patients with various relapsed or refractory lymphomas (large B cell, follicular, marginal zone, or mantle cell) receiving axicabtagene ciloleucel or brexucabtagene autoleucel. The framework incorporates two CAR-T-cell compartments alongside a latent kinetic-pharmacodynamic tumor compartment, utilizing a continuous, mechanistically-driven approach to characterize the observed PK profiles. A covariate search identified five significant covariates impacting PK (P < 0.001). Among these, only product type/mantle cell lymphoma (MCL) disease type was deemed to have clinically relevant effects on exposure. Despite its mechanistic basis, the model remains parsimonious, requiring only seven structural parameters, and no additional data beyond post-infusion CAR-T-cell peripheral blood concentrations and covariate information routinely collected in clinical trials. With its mechanistic foundation but parsimonious structure, this model balances data-driven practicality with the integration of underlying biological processes. Its ability to identify relevant covariates could make it a valuable tool in supporting drug development decisions.

External control arms (ECAs) have emerged as important tools to provide benchmark evidence on comparator effectiveness and thereby support regulatory decision making in situations where randomized controlled trials are difficult to conduct. However, how key methodological elements for ECAs are documented and implemented in practice and how residual uncertainty is managed at the time of approval remain unclear. Regulatory approvals of new medical products in Japan between 2019 and 2024 were systematically screened to identify cases employing ECAs. Among 674 eligible approvals screened from 835 approvals identified, 23 cases using ECAs were included and evaluated using a structured framework of 20 key attributes derived from regulatory guidance and methodological recommendations across planning, selection, design, analysis, and reporting domains. The post-marketing requirements associated with each approval were also analyzed to assess how residual uncertainty was addressed following approval. ECAs were predominantly used in settings characterized by rare diseases and small patient populations. Although attributes related to study planning and data-source selection were relatively well described in public documentation, there was substantial heterogeneity in the documentation of design and analysis elements. In particular, explicit specifications of estimands and approaches to handling intercurrent events and missing data were inconsistently described. Residual uncertainty was frequently addressed through post-marketing requirements, often involving studies of efficacy considerations and long-term surveillance, supporting a lifecycle-based approach to uncertainty management. These findings highlight opportunities to enhance transparency and consistency in the documentation of ECA-based evidence and to further develop structured approaches to uncertainty management in regulatory practice.

This study aims to design and validate a large language model (LLM) framework for systematic extraction of medication dosing across multiple therapeutic classes using electronic health record (EHR) data. Manual dose annotations were completed for 4295 medications across nine therapeutic classes. Five publicly available LLMs (Mistral-small, Llama 3-70B, Nova lite, DeepSeek, and Claude 3.5 Sonnet) were tested through iterative prompt engineering. Discrepancies between manual and model-derived doses were assessed using R² and classification accuracy. Among 2146 training and 2149 testing samples, Claude achieved high-performance metrics (R² = 99.32%, accuracy = 92.36%). DeepSeek had R² of 97.17% and accuracy of 90.31%. After prompt optimization, both models showed minor improvements (Claude R² = 99.34%, accuracy = 92.78%; DeepSeek R² = 98.89%, accuracy = 91.01%). Nova lite, Llama, and Mistral-small performance metrics improved after prompt re-engineering (ΔR² = 9.23%, 12.65%, and 39.03%, respectively). Models had modest agreement with an intra-class correlation of 0.73 (CI = 0.72-0.74). In testing data, Claude and DeepSeek demonstrated high-performance metrics R² = 99.74%, accuracy = 93.45% and R² = 95.58%, accuracy = 91.73%, respectively. Performance improvements were observed across all models following prompt refinement; however, differences may be attributed to model-specific responses to the prompting strategy used in this study. LLMs show high accuracy for automated medication dose extraction from EHR data. This study can serve as an initial attempt to demonstrate LLMs' capacity to extract medication dosing information from real-world data. Utilization of LLMs can augment clinicians in clinical practice through medication reconciliation and enhance research endeavors.

Commercial laboratories often have differing approaches to pharmacogenomic testing, which detects the presence of germline genetic variants associated with drug response. Setting standards for clinical laboratories and defining criteria for clinicians to consider when choosing a testing laboratory can help maximize the benefits of pharmacogenomics. Here, we propose specific laboratory practices that should be discontinued or established and followed as standard practice. We also propose ideal practices that would advance pharmacogenomics in clinical care.

Although sodium-glucose cotransporter-2 inhibitors (SGLT2i) are recommended for patients with type 2 diabetes mellitus (T2DM) and chronic kidney disease (CKD), concerns regarding urogenital infections persist. Therefore, we evaluated the risk of urinary tract infection (UTI) and genital tract infection (GTI) after SGLT2i initiation in this population. Using nationwide Korean claims data, we emulated a target trial within an active-comparator, new-user cohort design. Adults with T2DM and CKD who initiated SGLT2i or dipeptidyl peptidase-4 inhibitors (DPP-4i) between 2016 and 2023 were included. After inverse probability of treatment weighting, Cox models estimated hazard ratios (HRs) and 95% confidence intervals (CIs). Among 41,801 eligible initiators, 6385 initiated SGLT2i and 35,416 initiated DPP-4i. Compared with DPP-4i initiation, SGLT2i initiation was not associated with increased UTI risk (HR, 0.99; 95% CI, 0.83-1.17); instead, it was associated with a modestly increased GTI risk (HR, 1.31; 95% CI, 1.09-1.58). Severe UTI and GTI requiring hospitalization or emergency department visits did not differ between groups. Moreover, the findings were robust across per-protocol analyses, Fine-Gray competing-risks models, propensity score matching, and alternative outcome definitions. Subgroup and landmark analyses showed no consistent increase in UTI risk, whereas excess GTI risk clustered early and was most pronounced among women aged 65-79 years. Overall, SGLT2i initiation was not associated with increased UTI risk and was associated with only a modest, early increase in GTI risk, without an increase in severe urogenital infections. These findings support clinically indicated use of SGLT2i with targeted early counseling, particularly in older women.

Anakinra, an interleukin-1 receptor antagonist, shows promise for reducing inflammatory complications of premature birth. However, the pharmacokinetics (PK) of anakinra in neonates has not been characterized, limiting evidence-based dosing. This study aimed to develop a population PK model for anakinra to guide dosing in preterm neonates. Data from 25 preterm neonates (median gestational age: 26.3 weeks) from a Phase I/IIa trial of intravenous anakinra (Anakinra Pilot, NCT05280340) were analyzed, with 4-8 samples collected per neonate across 0-20 days postnatal age (PNA). A population PK model was developed and dosing simulations were performed incorporating a pharmacodynamic (PD) direct effect model (I_max, IC₅₀) and target exposure (AUC_0-24h). The data were best described by a one-compartment model with linear elimination and a baseline component for endogenous IL-1Ra. Allometric scaling was applied to clearance (CL) and volume of distribution (V), with a maturation function using PNA on CL to account for early age-related pharmacokinetic changes. Estimated CL and V were 3.97 L/h/70 kg (0.15 L/h/0.9 kg) and 28.9 L/70 kg (0.37 L/h/0.9 kg). At birth, CL was ~45% of the maximal early postnatal value, achieving ~100% by day 10 PNA. Simulations suggest that an optimal intravenous dosing regimen of 0.6 mg/kg 12 hourly at birth, 0.8 mg/kg 12 hourly on PNA day 1, 1.0 mg/kg 12 hourly from PNA days 2-3 and 1.15 mg/kg 12 hourly day 4 onward. This model delivers the first critical understanding of small protein PK in premature neonates, establishing a foundation for evidence-based dosing strategies to guide future prospective clinical trials.

The regulatory framework for QTc assessment has been optimized against false negatives while ignoring the false positives that silently remove efficacious medicines from development. False positives arise from three independent, compounding sources. First, a structural source: sensitive tests applied in low-prior populations generate predominantly false positives as a mathematical inevitability. Second, a measurement artifact: patch-clamp electrophysiology systematically biases hERG IC₅₀ estimates, creating artificially elevated priors. Third, biomarker non-specificity: QTc captures autonomic, thermal, and heart-rate-driven changes that carry no torsadogenic risk. Clinical TdP case series confirm that torsade requires large, concentration-dependent QTc prolongation well above regulatory detection thresholds. The multichannel hypothesis is not supported by the three primary reference agents. Furthermore, competitive equilibrium binding K_i measures the thermodynamic quantity that physiology reflects; it predicts QTc prolongation more consistently than IC₅₀ and yields a more sensitive and specific assessment. We propose a Bayesian weight-of-evidence framework (a quantitative New Approach Methodology suitable for FDA ISTAND qualification) anchored on the hERG K_i margin. The framework updates sequentially with in vivo and clinical QTc data, propagates uncertainty, and produces a posterior probability of clinically meaningful QTc prolongation. This posterior gates a staged assessment, reserving high-sensitivity studies for compounds where the prior warrants them. For peptides and proteins, a Beta-Binomial analysis of class-level evidence from approved therapeutics is sufficient to extinguish the hERG-driven torsade prior without compound-specific assessment. These elements place false positive and false negative costs on the same ledger, providing infrastructure to move cardiac safety assessment from qualitative judgment to a reproducible, decision-rule-based methodology.