Early model-based precision-dosing at home to guide adalimumab therapy
Running title: early prediction of adalimumab levels
Paul A.G. de Klaver1, Ron J. Keizer2, Rob ter Heine3, Frank Hoentjen4, Paul J. Boekema5, Inge Kuntzel6, Tiny Schaap7, Annick de Vries8, Karien Bloem9, Theo Rispens10, Lisa Smits11, Luc J.J. Derijks12
Author’s institutional affiliations and email address:
1 Máxima Medical Center, Department of Pharmacy and Clinical Pharmacology, Veldhoven, the Netherlands, p.deklaver@mmc.nl
2 InsightRx Inc, San Francisco, CA, US, ron@insight-rx.com
3 Radboud University Medical Center, Radboud Institute for Health Sciences, Department of Pharmacy, Nijmegen, the Netherlands, R.terHeine@radboudumc.nl
4 Radboud University Medical Center, Department of Gastroenterology, Nijmegen, the Netherlands, Division of Gastroenterology, Department of Medicine, University of Alberta, Edmonton, Canada, hoentjen@ualberta.ca
5 Máxima Medical Center, Department of Gastroenterology, Veldhoven, the Netherlands, P.Boekema@mmc.nl
6 Máxima Medical Center, Department of Rheumatology, Eindhoven, the Netherlands, I.Kuntzel@mmc.nl
7 Biologics Laboratory, Sanquin Diagnostic Services, Amsterdam, The Netherlands, j.schaap@sanquin.nl.
8 Biologics Laboratory, Sanquin Diagnostic Services, Amsterdam, The Netherlands, annick.devries@sanquin.nl
9 Biologics Laboratory, Sanquin Diagnostic Services, Amsterdam, The Netherlands, K.Bloem@sanquin.nl
10 Biologics Laboratory, Sanquin Diagnostic Services, Amsterdam, The Netherlands,
Department of Immunopathology, Sanquin Research, Amsterdam, The Netherlands, and Landsteiner Laboratory, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands, T.Rispens@sanquin.nl
11 Radboud University Medical Center, Department of Gastroenterology, Nijmegen, the Netherlands, Division of Gastroenterology, Lisa.Smits@radboudumc.nl
12 Máxima Medical Center, Department of Pharmacy and Clinical Pharmacology, Veldhoven, the Netherlands, l.derijks@mmc.nl
Ethical statements:
Data availability statement: raw data were generated at Máxima Medical Center and Radboud University Medical Center. Derived data supporting the findings of this study are available from the corresponding author [PK] on request.
Funding statement: n.a.
Conflict of interest disclosure: the authors declare that there is no conflict of interest
Ethics approval statement: the study was approved by the local ethics committee.
Patient consent statement: all patients provided written informed consent.
Clinical trial registration: the trial was registered in the Netherlands Trial Register with trial registry number NTR 7692 (www.trialregister.nl).
Keywords: adalimumab, model based precision dosing, Inflammatory bowel disease, rheumatology
Total word count excluding summary: 2144
Underdosing of adalimumab can result in non-response and poor disease control. In this study we investigated the prediction of adalimumab levels with population pharmacokinetic model-based Bayesian forecasting early in therapy. This way underexposed non-responders can possibly be identified early to optimise disease control.
Methods
A literature study was performed to identify adalimumab pharmacokinetic models. With data from a previous pharmacokinetic adalimumab study a model was evaluated retrospectively. In the prospective phase, a fit-for-purpose evaluation of the model was performed for rheumatologic and inflammatory bowel disease patients with peak, trough and control adalimumab samples obtained by a volumetric absorptive microsampling technique and administration data from an electronic needle container. Steady state adalimumab levels were predicted from peak and trough levels collected after the first adalimumab administration. Predictive performance was calculated with mean prediction error (MPE) and normalized root mean square error (RMSE).
Results
An existing pharmacokinetic model was selected with external validation for the prospective phase. Thirty-six patients (22 rheumatologic and 14 IBD) were included in our study. After stratification for absence of anti-adalimumab antibodies, the calculated MPE was -2.6% and normalised RMSE 24.0%. Concordance between predicted and measured adalimumab serum levels falling within or outside the therapeutic window was 75%. Three patients (8.3%) developed detectable levels of anti-adalimumab antibodies.
Conclusion
This prospective study demonstrates that adalimumab levels at steady state can be predicted from early samples. This concept enables early precision dosing at home to guide therapy.
“Clinical trial registry number: Netherlands Trial Register, NTR 7692”
Keywords: model-based precision-dosing adalimumab
Introduction
Adalimumab is a fully human recombinant IgG1k monoclonal antibody against Tumor Necrosis Factor (TNF) alpha. It is approved for moderate to severe inflammatory bowel disease (IBD) and the rheumatic diseases rheumatoid arthritis (RA), psoriatic arthritis (PsA), and ankylosing spondylitis (SpA) with poor response to conventional immunosuppressants. Adalimumab is administered subcutaneously. For RA, PsA and SpA the licensed dose is 40 mg every other week, without induction therapy. For IBD the licensed induction dose is either 80 mg followed by 40 mg after two weeks or 160 mg followed by 80 mg after two weeks, the latter induction scheme being used more frequently in clinical practice. The licensed maintenance dose is 40 mg every other week.
Up to 30% of patients with IBD do not respond to initial treatment with TNFα antagonists. It is important to differentiate between true primary non-responders (pharmacodynamic failure) and underexposed non-responders (pharmacokinetic failure), to intervene early in latter cases and adjust dosage to the individual patient. This serves patient satisfaction, disease control and drug expenses.
Target adalimumab trough-levels can range from 5-12 mg/L and therapeutic drug monitoring (TDM) can be performed in routine clinical practice, most often reactively during the maintenance phase of therapy. Population pharmacokinetic models have been developed and could theoretically be used for serum level prediction at steady state and therefore early dosage prediction, but these models have not yet reached clinical practice.
In the current study, we investigated the feasibility of predicting adalimumab levels with population pharmacokinetic model-based Bayesian forecasting early in therapy. This can be used to identify underdosed non-responders as soon as possible to optimise disease control in clinical practice.
Materials and Methods
2.1 Population pharmacokinetic model selection
A 3-step-approach as described by ter Heine et al was used. For step 1, identification of models, a PubMed search for a population pharmacokinetic adalimumab model was performed and FDA registration data were evaluated. In step 2, an expert panel of pharmacometricians and clinical pharmacologists retrospectively evaluated the predictive performance of the pharmacokinetic models with data from a published study with IBD patients in Máxima Medical Center using Nonlinear Mixed Effects Modelling (NONMEM) version 7.4, executed through the Pirana workbench. Final model selection was based on Goodness of fit evaluation in line with best practice. Step 3 in this strategy is described below as the prospective observational cohort study.
2.2 Study design and population
This multicentre prospective observational cohort study aimed to collect data from 40 patients ≥ 18 years with IBD or RA, SpA and PsA starting with adalimumab from March 2019 up to August 2020.
Patients were recruited from Rheumatology and Gastroenterology departments of Máxima Medical Center, Veldhoven/Eindhoven, the Netherlands and Gastroenterology department of Radboud University Medical Center (UMC), Nijmegen, the Netherlands. Adalimumab was dosed according to label and local clinical care pathways.
Pregnancy, known allergy for adalimumab or excipients and previous adalimumab use were exclusion criteria. For each drop-out a new patient was recruited. Patients weight, gender, date of birth and indication for treatment with adalimumab were collected.
The workflow is shown in figure 1.
2.3 Sampling
Sampling was done with a volumetric absorptive microsampling (VAMS) method. All patients were provided with 3 sampling sets for capillary blood microsampling at home. A sampling set consists of two 20 microliter MitraTM microsamplers (Neotyrex, Torrance, USA) and a BD microtainer 2 mm contact-activated lancet (BD, Dublin, Ireland). Patients were asked to perform capillary sampling at home 5 days, 13 days and 12 weeks after first adalimumab administration (Figure 1). Patients could receive sampling reminders for each sampling moment per email or text message on request. Samples were returned and stored under refrigerated conditions until analysis at Sanquin Diagnostic Services (Amsterdam, the Netherlands). Patients completed the trial upon returning the third sample.
2.4 Drug administration monitoring
All patients were required to use a an electronic needle container (HealthBeacon Injection Care Management SystemTM, Health Beacon Ltd, Dublin, Ireland). Electronic needle containers were provided as part of standard care to all patients in this study. The electronic needle container is a device intended to monitor and improve compliance for patients on therapy with injectables. It reports the date and time a syringe is dropped in the device after use. The electronic needle container reports were automatically sent to Máxima Medical Center with secure mail.
2.5 Measurement of adalimumab and anti-adalimumab antibody concentrations