Vedolizumab PBPK Modeling Links Disease Severity to Target-Mediated Clearance

Representing the Target as Part of Antibody Disposition

Vedolizumab blocks α4β7 integrin and reduces lymphocyte trafficking to the intestinal mucosa. Researchers at Cincinnati Children’s Hospital Medical Center, the University of Cincinnati and Oita University Hospital examined how changes associated with inflammatory bowel disease could alter its pharmacokinetics. They developed a minimal physiologically based pharmacokinetic (PBPK) model that incorporates target-mediated drug disposition (TMDD).

Implemented in Simcyp version 22.0, the model represents blood, tissue and lymph compartments. It includes antibody transport, cellular uptake, neonatal Fc-receptor recycling, lysosomal degradation and binding to the α4β7 target. This structure allows the rate of target binding and removal to influence the circulating drug concentration.

The team combined published physicochemical and physiological inputs with clinical concentration–time data. They first estimated distribution and clearance parameters using healthy-volunteer studies. A model initialized with endogenous-IgG parameters underpredicted vedolizumab exposure, prompting sequential refinement of parameters governing transport, uptake, target capacity and clearance.

Estimating Disease-Related Target Capacity

The healthy-adult model was then adapted to adults with ulcerative colitis or Crohn’s disease. The investigators estimated an effective target-abundance parameter, Rmax, from clinical pharmacokinetic data for different disease-severity groups. Its fitted values were 0.0027 μM in healthy adults, 0.0048 μM in mild-to-moderate ulcerative colitis, and 0.0075 μM in moderate-to-severe ulcerative colitis or Crohn’s disease.

These Rmax values are effective capacities inferred from clinical drug-concentration profiles. The authors note that the fitted parameter may capture several inflammation-related processes, including factors beyond target expression that affect antibody disposition.

Clinical profiles came from published phase I–III studies, including healthy-adult single doses of 0.2–10 mg/kg and repeated dosing in patients. Virtual trials were matched to available study demographics. Each simulation scenario used 10 repeated trials of 10 virtual individuals, giving 100 simulated people per scenario; that count describes the simulations, not the size of the clinical datasets.

Comparing Predictions with Clinical Pharmacokinetics

For healthy adults, mean predicted-to-observed ratios ranged from 0.81 to 1.11 for peak concentration and from 0.80 to 1.02 for AUC to the last observation. In mild-to-moderate ulcerative colitis, the corresponding ranges were 0.88–1.02 and 0.78–0.99. For moderate-to-severe ulcerative colitis or Crohn’s disease, trough-concentration ratios ranged from 0.67 to 1.49.

The paper also reports an external population assessment using Japanese clinical data and the simulator’s Japanese population library. The main text describes agreement in concentration–time profiles; the detailed Japanese comparisons are in supplementary figures and tables.

Using a maintenance scenario of 300 mg intravenously every 8 weeks, the researchers simulated how higher Rmax would affect exposure. Greater target abundance increased target-mediated removal and lowered trough concentrations. The paper compared these outputs with exposure thresholds drawn from earlier clinical association studies, whose proposed thresholds vary.

What the Mechanistic Fit Can Explain

The study provides a way to examine how disease-associated changes in target binding may contribute to variability in antibody exposure. It also separates a physiological explanation for a concentration profile from the direct measurement of that mechanism in an individual patient.

Parameter identifiability remains a constraint when plasma profiles are the main source of information. Albumin, vascular permeability, protein catabolism and tissue distribution were not all modeled as separate disease-dependent processes, and several could be reflected in the effective Rmax estimate. Human organ-specific distribution measurements were also unavailable. The reported work evaluated pharmacokinetic fits and simulated maintenance exposure; prospective testing of model-guided dose changes was outside the study.