O11 Model-informed precision dosing of biologic therapies in psoriasis: development, validation and clinical translation

Abstract

Biologic therapies have transformed psoriasis management but use fixed dosing despite interindividual variability in response. Conventional therapeutic drug monitoring (TDM) adjusts dosing using serum drug concentrations and population-based thresholds, ignoring prior patient data. Model-informed precision dosing (MIPD) integrates TDM with prior-informed pharmacokinetic (PK) and pharmacodynamic (PD) models and Bayesian forecasting to individualize dosing, enabling data-driven treatment optimization. The aim of this study was to develop a clinician-facing MIPD dashboard for biologics. We (i) identified published population PK/PD models (risankizumab, adalimumab and ustekinumab); (ii) validated models using UK real-world BSTOP cohort, and implemented them within a multidrug R Shiny-based dashboard that uses Bayesian methods, serum drug concentrations and patient-specific covariates to recommend personalized doses; (iii) user tested the dashboard and (iv) evaluated its clinical impact in a multicentre feasibility trial of risankizumab dose minimization in remission, and in silico against routine-care TDM of adalimumab in a national psoriasis service. Population models comprised a two-compartment PK (risankizumab), one-compartment PK with turnover PD (adalimumab) and one-compartment PK with mixture-turnover PD model (ustekinumab). Bayesian models demonstrated good predictive performance (mean prediction error 15% for risankizumab). Clinicians (n = 29) rated the dashboard usability as high (System Usability Scale 74/100), averaging 2 min per calculation. In the feasibility trial (n = 94; clear/nearly clear skin > 12 months on risankizumab; 14 UK centres), clinicians successfully implemented dashboard-guided dosing. Among the dashboard-guided dosing arm, 49% received longer intervals than standard (mean 16 weeks; range 12–23), saving £3326 per patient annually. In silico, 70 of 73 (96%) patients (66% male, median weight 84 kg, median Psoriasis Area and Severity Index 0.8) were recommended longer adalimumab intervals by the dashboard than routine-care TDM (mean 5 vs. 2 weeks; range 1–10 vs. 1–4). We demonstrate end-to-end MIPD translation, from model validation to dashboard usability and impact assessment, using a framework applicable across dermatology. Ongoing work includes ≥ 12-month feasibility assessment (acceptability and practicality), cost-effectiveness modelling and adding further agents to support routine adoption of precision biologic dosing.