Precision biologic dosing in psoriasis: Development, validation and beta-testing of a therapeutic drug monitoring dashboard

Abstract

Following the advent of biologic therapies for chronic plaque psoriasis, achieving clear/nearly clear skin (disease control) has become a realistic goal. Patients currently continue high-cost biologic therapy indefinitely once they have achieved disease control, which leads to a long-term drug and healthcare burden. Clinical trial data indicate that some individuals can maintain disease control with less frequent doses, especially those who have higher serum drug concentrations, indicating a role for therapeutic drug monitoring (TDM) in guiding dosing decisions. To enable real-world implementation of TDM-guided precision dosing of biologics, we developed, validated, and beta-tested an online interactive TDM dashboard, powered by a pharmacokinetic (PK) model. We searched for population PK models of the exemplar psoriasis biologic risankizumab in PKPDAI and PubMed databases. We externally validated the PK model using an independent multicentre real-world UK psoriasis dataset (n = 50 patients, BSTOP study) prior to embedding it within an online R Shiny interactive dashboard. Two rounds of beta-testing of the TDM dashboard were performed with healthcare professionals to assess usability. A two-compartment PK model with first-order absorption and elimination processes was identified. The PK model was based on Phase I-III trial data and adopted a nonlinear mixed effects approach (13 123 observations from 1899 patients; 71% male; median age, 47; median weight, 97 kg; Suleiman et al., Clin Pharmacokinet 2019). We integrated the model into an online R Shiny interactive dashboard, which generates a personalized dosing interval for risankizumab. The model is used as a Bayesian prior, to predict individual patient parameters using their serum drug concentrations, dose history and other covariates (weight, serum concentrations of antidrug antibody, albumin and creatinine). Individual predictions were externally validated (mean absolute error 0.95 mg/L; mean percentage error 17.6%; root mean square error 1.7 mg/L; R2 0.8). Beta-testing sessions with 29 clinicians (58% female; mean age 41; 33% doctor; 42% nurse) from 8 UK dermatology centres showed above-average usability of the dashboard (mean System Usability Scale score 72/100, rating 'good'). All participants evaluated the dashboard as user-friendly and rated it acceptable or completely acceptable. Mean time to generate a dosing interval using the dashboard was 1.9 min (SD 1.2). Mean time to generate a dosing interval using the dashboard was 1.9 min (SD 1.2). Our study provides a novel pipeline for the implementation of TDM-enabled precision dosing of biologics in real-world practice. Our user-friendly online TDM dashboard has the potential to incorporate other biologic therapies and can be extended across disease contexts (including non-response and other immune-mediated inflammatory diseases) for maximal real-world health benefits and cost saving.