Erwin Dreesen (1), Sophie Berends (2), David Laharie (3), Geert D’Haens (4), Séverine Vermeire (5), Ann Gils (1), Ron Mathôt (2)
(1) Department of Pharmaceutical and Pharmacological Sciences, KU Leuven, Leuven, Belgium, (2) Hospital Pharmacy, Amsterdam UMC, Amsterdam, Netherlands, (3) Department of Hepato-gastroenterology and Digestive Oncology, Haut-Lévêque Hospital, Bordeaux UMC, Bordeaux, France, (4) Department of Gastroenterology and Hepatology, Amsterdam UMC, Amsterdam, Netherlands, (5) Department of Gastroenterology and Hepatology, University Hospitals Leuven, Leuven, Belgium.
Objectives: Infliximab (IFX) is a monoclonal antibody that binds and inhibits tumour necrosis factor alpha. Twenty years after approval by the US Food and Drug Administration, IFX remains a commonly used agent for inducing and maintaining disease remission in patients with moderate to severe Crohn’s disease (CD). While regulatory approval and initial success of IFX were based on controlling clinical disease activity, therapeutic goals have evolved, now aiming for endoscopic remission (ER) in addition to clinical remission (CR) [1]. However, ER is found in only one-third of patients with CR, indicating undertreatment at the label-recommended dose [2].
Strategies for treatment optimisation have been explored, including pharmacokinetic (PK; trough concentration) and pharmacodynamic (PD; faecal calprotectin [fCal]) monitoring. However, evidence for the benefits of PK and PD monitoring remains scarce [3]. With the currently available flowchart algorithms for PK(/PD) monitoring, it is difficult to hit the target rapidly and precisely [4]. In addition, the relationship between IFX concentrations, fCal concentrations, and ER has not been investigated before, making it even more difficult to dose optimise patients towards ER.
We aimed to develop a population (pop)PK/PD model to characterise the IFX dose-exposure-biomarker-response relationship in patients with CD (biomarker: fCal; response: ER).
Methods: Analyses were performed using data of a phase 4 dose-escalation study (TAILORIX) [3]. Patients underwent standard IFX induction therapy with 5 mg/kg infusions at week (w)0, 2 and 6. From w14 through w54, the IFX dose was irreversibly doubled in 48% of patients based on one of three monitoring algorithms. Endoscopies were performed at w0, 12 and 54. IFX and fCal concentrations were measured at w0, 2, 6, 12, 14, and every four weeks thereafter until w54.
Three models were developed sequentially: A 2-compartment popPK model linking IFX dose to IFX exposure, an indirect response popPK-PD model describing the inhibitory effect of IFX concentrations on fCal concentrations (incl. baseline and drug effect model with IFX affecting Kin based on an inhibitory Emax model), and a first-order Markov PD model linking fCal to transitions between states of ER (CD endoscopic index of severity <3), no ER, and dropout.
Empirical Bayes estimates of individual parameters from the popPK and popPK-PD models were used to predict individual IFX and fCal concentrations that were then linked to the individual observed fCal and ER time courses, resp. A stepwise forward inclusion (α=0.01) backward elimination (α=0.001) covariate modelling procedure was used.
The models were used to simulate 5, 7.5, and 10 mg/kg IFX induction or maintenance dosing with standard dosing intervals (n=50,000 patients per dose). Covariates were sampled from the original dataset. All modelling and simulation were performed using NONMEM 7.4.
Results: A total of 1,329 IFX concentration measurements were used for popPK modelling. IFX clearance (typical value 0.277 L/day [4% relative standard error]) increased with increasing fCal concentrations, decreasing albumin concentrations, increasing CD activity index, and presence of anti-drug antibodies. The four covariates explained 11.3% of the interindividual variability in IFX clearance. Baseline fCal increased with increasing C-reactive protein and decreasing platelet count. No covariate relationships were tested for EC50 and Kout given high eta shrinkage (>20%). Lower fCal increased the probability of attaining ER and decreased the probability of losing ER. Probability of dropping out given an earlier state of absence of ER increased with time.
Simulations of 150,000 patients receiving 5, 7.5, or 10 mg/kg IFX (1:1:1) resulted in a flat dose-response curve with large interindividual variability. The predicted fraction of patients achieving ER at w12 was 45.1% [30.3-60.5] (median [interquartile range]) when on 5 mg/kg IFX (~46.4% observed in data). Simulations of 10 mg/kg induction doses predicted 47.5% [32.0-62.6] of patients achieving ER. A similar observation was done during maintenance therapy.
Conclusions: Large variability in IFX PK and PD resulted in highly variable probabilities of achieving ER between patients. Our models may be used to guide combined PK and PD monitoring to allow targeting of personally optimal IFX and fCal concentrations that are associated with a predefined probability of ER (eg targeting a 64% chance of ER at w12 ~ 100 μg/g fCal at w6).
References:
[1] Peyrin-Biroulet L et al. Am J Gastroenterol, 110:1324–38, 2015.
[2] Baars J et al. Inflamm Bowel Dis, 18:1634–40, 2012.
[3] D’Haens G et al. Gastroenterology, 154 (5), 1343-1351.e1, 2018.
[4] Dreesen E et al. Clin Gastroenterol Hepatol, 18 (3), 637-646.e11, 2020.
Reference: PAGE () Abstr 9408 [www.page-meeting.org/?abstract=9408]
Poster: Oral: Drug/Disease Modelling