bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/2023.07.13.548803

Applying Hierarchical Generalized Additive Models to Non-Compartmental Analysis of Pharmacokinetic Data

Woodward, A. P.
View original source ↗Export metadata ↓

Abstract

Non-compartmental analysis (NCA) is a popular strategy for obtaining estimates of pharmacokinetic parameters, while requiring both minimal structural assumptions, and limited input by the analyst. As typically applied, its scope and depth are constrained by its statistical simplicity. Embedding the NCA within a hierarchical generalized additive model (HGAM) may facilitate the simultaneous analysis of data from multiple subjects, estimation of covariate effects in one stage, and implementation of censored responses, similarly to the capabilities of nonlinear multilevel models as widely applied in pharmacometrics. HGAM is an interesting extension to multilevel linear models that allows the effects of predictors to be implemented as smooth functions, which has been widely implemented in various disciplines to nonlinear trends, including for longitudinal data. This approach extends the capability of previous implementations of spline-based methods applied to NCA, within an accessible workflow in open software. Application of HGAM to two example datasets, one describing oral drug administration, and one describing IV and oral drug administration with categorical covariates and censoring, illustrates the overall approach, including parameter estimation, visualization and model checking, and uncertainty quantification. A Bayesian approach to estimation facilitates interpretable expressions of the uncertainty in individual parameters, population parameters, and functions of parameters such as contrasts.

Explore related subjects

pharmacology and toxicologyPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Woodward, A. P.. 2023-07-15. Applying Hierarchical Generalized Additive Models to Non-Compartmental Analysis of Pharmacokinetic Data. https://doi.org/10.1101/2023.07.13.548803

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Peptide Sequence Determines the Efficacy of Single-Dose AAV-Vectorized GLP-1R Agonists in Diet-Induced Obese Mice

Background: Glucagon-like peptide-1 receptor (GLP-1R) agonists such as semaglutide produce substantial weight loss and glycemic improvement, but their long-term efficacy depends on sustained adherence to chronic dosing. Adeno-associated virus (AAV) gene therapy could, in principle, provide durable GLP-1R agonist (GLP-1RA) exposure from a single administration, but the degree of peptide engineering required for a vectorized GLP-1RA transgene to be efficacious in vivo is unknown. Methods: We generated four self-complementary AAV (scAAV) constructs encoding: native human GLP-1(7-36) (GLP-1), a DPP-IV-resistant Gly8-GLP-1(7-36) variant (Gly8), Exendin-4 (Ex4), or an Exendin-4 human IgG4 Fc fusion protein (Ex4-Fc). All four constructs were validated to secrete functional GLP-1R agonists in vitro using a conditioned-media potency assay. Subsequently, these constructs were packaged into PHP.eB capsids and administered as a single intravenous dose (5x1012 vg/kg) to diet-induced obese (DIO) mice alongside a vehicle control. Body weight, body composition, food intake, fasting plasma insulin, and oral glucose tolerance were assessed over a 3-week period. At the end of the study, tissues were harvested for assessments of plasma/liver lipids, tissue transgene transcript and peptide levels. Results: HEK293 cells transfected with the four AAV constructs all secreted active peptides/proteins as assessed by a cell-based GLP-1R activation assay, although the effective concentration of the GLP-1 constructs in the media spanned nearly four orders of magnitude 72 hours after transfection. In diet-induced obese (DIO) mice, Ex4 and Ex4-Fc vectors, but not GLP-1 or Gly8 vectors, produced significant reductions in body weight of up to 22.6% of starting body weight out to 21 days after a single dose. In addition, reductions in food intake, fat mass, and fasting insulin relative to vehicle were also observed for Ex4 and Ex4-Fc. Furthermore, Ex4-Fc improved oral glucose tolerance while reducing plasma cholesterol and liver triglycerides relative to vehicle. Although transcript levels did not differ appreciably among the four constructs in either brain or liver, measurable peptide product in plasma and brain lysate tracked the rank order of the in vitro supernatant screen, with GLP-1 and Gly8 groups indistinguishable from endogenous background. Conclusions: PHP.eB scAAV Ex4 and Ex4-Fc showed robust efficacy on multiple metabolic endpoints including weight loss and glucose tolerance. In contrast, vectorized GLP-1 and Gly8 showed no in vivo efficacy despite comparable levels of viral transcript in the brain and the liver. Vectorized GLP-1 and Gly8 also performed poorly compared to exendin-4-based constructs in vitro, suggesting that conditioned-media potency screening is a valid early filter for construct selection in vectorized-peptide AAV development. Engineering peptides to improve translation and secretion efficiency, in addition to circulating plasma stability, could further optimize vectorized GLP-1RA therapeutics.

pharmacology and toxicology↗
bioRxiv2026

Preclinical Evaluation of a Novel BOC-Modified Triazoloacridinone (TRA-01AR) in Combination with 5-Fluorouracil in Colorectal Cancer

5-Fluorouracil (5-FU) remains a cornerstone of colorectal cancer (CRC) therapy, yet its clinical efficacy is limited by resistance, systemic toxicity, and unfavorable pharmacokinetics. Strategies that enhance tumor sensitivity while maintaining tolerability are therefore of considerable interest. Here, we evaluated 5-FU in combination with two triazoloacridinones, C-1305 and a novel BOC-modified derivative, TRA-01AR, using integrated cellular, transcriptomic, bioenergetic, and in vivo approaches. Both compounds induced cytotoxic effects across cancer cell models and produced overlapping but distinct transcriptional responses, primarily affecting DNA replication and cell cycle pathways. Bioenergetic profiling revealed a shift toward oxidative phosphorylation accompanied by reduced glycolysis, indicating modulation of cellular metabolism relevant to chemotherapy response. Interestingly, TRA-01AR exhibited delayed cytotoxicity relative to C-1305, although the underlying mechanism remains to be defined. In vivo, the combination of 5-FU with C-1305 significantly reduced tumor weight in HT-29 xenografts, whereas 5-FU combined with TRA-01AR did not result in a statistically significant reduction compared to 5-FU alone. TRA-01AR was associated with a more favorable systemic profile and increased intratumoral blood flow compared with C-1305. Collectively, these findings indicate that triazoloacridinones modulate transcriptional and metabolic programs relevant to the 5-FU response. While C-1305 demonstrated greater antitumor activity under the tested conditions, TRA-01AR exhibited distinct biological effects, including improved tolerability and vascular modulation. Further studies are required to clarify TRA-01AR's mechanism of action and assess its potential relevance in CRC-specific therapies.

pharmacology and toxicology↗
bioRxiv2026

Molecular Mechanisms of Quizartinib Resistance in FLT3 Mutants

Quizartinib, a type II FLT3 inhibitor approved for FLT3-ITD-positive acute myeloid leukemia (AML), is frequently limited by acquired resistance, which arises in up to 40% of patients with disease progression and is driven largely by the D835Y and F691L kinase-domain mutations. To define the underlying mechanisms, we performed triplicate 200-ns all-atom molecular dynamics simulations of wild-type FLT3, FLT3-D835Y, and FLT3-F691L in complex with Quizartinib, combining conformational analysis, MM-GBSA binding free-energy calculations, and binding-pocket characterization. Both mutations destabilized the inactive DFG-out conformation required for type II inhibitor binding, with mean binding free-energy reductions of 5 kcal/mol for D835Y and 10 kcal/mol for F691L relative to wild-type. D835Y disrupted the activation-loop electrostatic network and reduced Cys694 hydrogen-bond occupancy from 85% to 45%, whereas F691L abolished key van der Waals contacts and opened a 120 cubic Angstrom hydrophobic void at the gatekeeper position. Both mutants sampled a wider range of conformations and lost 14-17% of binding-pocket volume. These predictions were corroborated by CCK-8 viability assays and Western blotting of p-FLT3 (Y591) in Ba/F3 cells, which reproduced the predicted resistance hierarchy (F691L > D835Y > WT) at the cellular level. Together, these results indicate that D835Y and F691L confer resistance through distinct electrostatic and hydrophobic mechanisms that converge on destabilization of the inactive kinase conformation, providing a structural basis for the design of next-generation FLT3 inhibitors.

pharmacology and toxicology↗