Host cell proteins (HCPs) are process-related impurities that may persist in therapeutic protein products and can impact product quality or trigger unwanted immunogenicity responses. The immunogenicity potential of individual HCPs can be assessed using in silico approaches or T cell-based assays. Recently, LC–MS/MS-based major histocompatibility complex (MHC)-associated peptide proteomics (MAPPs) has emerged as a powerful approach for identifying naturally processed and presented peptides, thereby providing insights into mapping the potential immunogenic epitopes of protein therapeutics. In this study, we applied MAPPs workflow to evaluate the immunogenicity potential of six frequently identified CHO-derived HCPs with varying molecular weights and degrees of homology to their human counterparts, including PLBL2, carboxypeptidase, complement C1r subcomponent, S100A5, S100A11, and beta-hexosaminidase. A pooled HCP mixture demonstrated peptide repertoires comparable to individually loaded proteins, highlighting the method’s capability for multiplexed HCP immunogenicity assessment. Overall, this LC–MS/MS-based approach provides a robust framework for evaluation of HCP-specific T cell epitopes and expands immunogenicity risk assessment strategy beyond in-silico predictions and cell-based functional assays.
We explored the impact of chronological age, proton pump inhibitors (PPIs) in older adults, and Type II diabetes in older adults on pH, buffer capacity, lipase activity, pepsin activity, individual bile acids, and osmolality in antral contents during fasting to evaluate whether drug product disposition is likely to differ from healthy adults. Antral contents were aspirated from 20 adults (Reference Group, RG), 14 older adults (Test Group 1, TG1), 7 older adults on PPIs (Test Group 2, TG2), and 6 diabetic older adults (Test Group 3, TG3). All subjects were free of additional conditions likely to confound the luminal characteristics of interest and were living with up to mild frailty. Median TG2 pH (7.84) was significantly higher than the median RG pH (1.75). Median buffer capacity across all groups was 10.38–19.85 mmolNaOH/L/ΔpH. Mean lipase activity was 96.72-124.47 Utrib/mL. Mean TG2 pepsin activity (32,500 USPunits/mL) was significantly lower than mean RG pepsin activity (176,669 USPunits/mL). Based on medians, glycochenodeoxycholic was the most abundant bile acid in TG2; glycocholic was the most abundant bile acid in all other groups. Median total bile acid content across all groups was 98.1–896.4 μM. Median osmolality was 257–328 mOsmol/kg. Chronological ageing is unlikely to affect drug disposition during fasting via changes in the physicochemical characteristics of antral contents. Caution should be exercised when fasted older adults are on PPIs and the drug or the drug product is sensitive to gastric pH and/or pepsin activity. Power-based sample-size estimations might allow for clarifying the impact of Type II diabetes on fasting gastric pH.
Peptide based therapeutics such as GLP-1 analogs are an increasingly important therapeutic modality for metabolic diseases. As these therapeutics often mimic endogenous counterparts, assessing their immunogenicity during clinical development is critical for understanding safety and efficacy profiles and meeting regulatory expectations. Here, we describe a sample pretreatment strategy using activated charcoal to deplete residual drug and mitigate matrix interference, enabling development of drug-tolerant neutralizing anti-drug-antibody (nAb) assays. This approach was optimized and validated using Survodutide, a half-life extended GLP-1R/GCGR dual-agonist currently in clinical development. The optimized charcoal depletion improved assay drug tolerance by approximately 100-fold, facilitating reliable nAb detection in clinical samples. Ultimately, the method was successfully implemented in several validated nAb assays to support Survodutide clinical development. This platform approach has the potential for broad application across peptide therapeutics where drug tolerance is a critical challenge.
Kidney transplant recipients (KTRs) frequently receive vancomycin for severe infections, but data on vancomycin pharmacokinetics (PK) and nephrotoxicity in this population remain limited. This retrospective cohort study included hospitalized adult KTRs who received vancomycin at the University of Michigan between 2007 and 2024, with a frequency-matched nontransplant comparison cohort. Population PK analysis was performed using nonlinear mixed-effects modeling. Nephrotoxicity was assessed using KDIGO criteria and 2009 vancomycin guideline definition. Risk factors were evaluated using multivariable logistic regression. The PK analysis included 1539 patients (469 KTRs, 1070 nontransplant). A one-compartment model adequately characterized vancomycin PK. Estimated clearance (CL) was substantially lower in KTRs than in nontransplant patients (1.5 vs. 3.9 L/h), and both creatinine clearance (CLcr) and KT status were significant covariates. The nephrotoxicity analysis included 1115 patients (303 KTRs, 812 nontransplant). KTRs had a higher crude incidence of nephrotoxicity under the KDIGO criteria (38.6
PyDarwin is an open-source Python package integrating machine learning with NONMEM for automated model selection; however, its utility in model-informed precision dosing (MIPD) remains unvalidated. This study aims to evaluate pyDarwin for population pharmacokinetics (PopPK) model development in MIPD compared with expert-developed models, using paediatric tobramycin as a representative case. The model-building dataset included 450 plasma concentrations from 63 children with cystic fibrosis (CF). One group manually developed a model using stepwise covariate selection, while another applied pyDarwin within a predefined 27-dimensional model space, using a fitness function that incorporated a penalty based on 24-h area under the curve (AUC24h). External validation was performed using 1011 concentrations from 144 children, including evaluation using all therapeutic drug monitoring (TDM) data and clinically relevant MIPD scenarios (a priori [AP], Bayesian 1, Bayesian 2). Both models included weight-based scaling, a maturation function, and estimated glomerular filtration rate as covariates on clearance, yielding comparable clearance estimates. The pyDarwin-selected model was a two-compartment model with fixed allometric scaling exponents, whereas the manual model was a one-compartment model with estimated exponents. Both models showed comparable probability of target attainment for steady-state AUC24h. In external validation using all TDM data, the pyDarwin-selected model demonstrated better predictive performance for CF patients than manual model (prediction-based MPE -14.6
