Wannee Kantasiripitak (1), Erwin Dreesen (1), Iris Detrez (1), Sebastian Stefanovic (2), Séverine Vermeire (3), Marc Ferrante (3), Thomas Bouillon (4), David Drobne (2), Ann Gils (1)
(1) Therapeutic and Diagnostic Antibodies, Department of Pharmaceutical and Pharmacological Sciences, University of Leuven, Belgium, (2) Department of Gastroenterology and Hepatology, University Medical Centre Ljubljana, Slovenia, (3) Department of Gastroenterology and Hepatology, University Hospitals Leuven, Belgium, (4) Drug Delivery and Disposition, Department of Pharmaceutical and Pharmacological Sciences, University of Leuven, Belgium
Objectives: Golimumab (GLM, Simponi®) is a fully human anti-tumour necrosis factor-alpha monoclonal antibody for the treatment of moderately to severely active ulcerative colitis (UC). Endoscopic remission (ER; a decrease in Mayo endoscopic subscore [MES] from 3 [severe disease] or 2 [moderate disease] at baseline to 1 [mild disease] or 0 [inactive disease] at week 14) has been focused as a therapeutic target in patients with UC as it is associated with improved long-term clinical outcomes. 1 Higher serum trough concentrations (TC) of GLM six weeks after start of therapy were associated with a greater proportion of patients achieving ER.2
Our aims are to develop a population pharmacokinetic (popPK) model and an exposure-response model that links GLM exposure metrics (TC and area under the concentration-time curve [AUC]), derived from the popPK model, to probabilities of transitioning between MES states from baseline to week 14.
Methods: GLM concentration-time data of 56 patients with UC (414 peripheral venepuncture [VP] samples and 296 dried blood spot [DBS] samples) were obtained from 2 study centres (University Hospitals Leuven, Belgium and Ljubljana University Medical Centre, Slovenia).3–5 Serum and DBS concentrations were fitted simultaneously by estimation of a population conversion factor that related individually predicted serum and DBS concentrations.6 A popPK model was developed in NONMEM (version 7.4). A first-order conditional estimation method with interaction was employed to obtain PK parameter estimates. Residual error models were tested for serum and DBS concentrations separately. The stepwise covariate modelling approach was employed to obtain the final covariate model (forward α = 0.010, backward α = 0.001). The developed PK model was used to derive GLM exposure metrics (i.e., TC and AUC).
A logistic regression exposure-response model was implemented to describe the relationship between these GLM exposure metrics and the response.7 We assumed that only ordered transitions can occur (i.e., patients going from states 3 to combined 1 and 0 transitioned through state 2 and vice versa) and that the transition probabilities between states can be inversed (e.g. P3à2=1-P2à3). Sampling importance resampling (SIR) was adopted for parameter uncertainty estimation.8
Results: Data were described by a two-compartment model with linear absorption and elimination. The estimated PK parameters (typical value [relative standard error]) were absorption rate constant (ka: 0.495 1/day [15%]), apparent total body clearance (CL/F: 0.417 L/day [9%]), apparent volume of distribution in the central compartment (Vc/F: 8.82 L [8%]), apparent volume of distribution in the peripheral compartment (Vp/F: 3.85 L [43%]), apparent intercompartmental clearance (Q/F: 0.469 L/day [20%], and conversion factor (4.14 [3%])). The residual error was best described using combined additive and proportional error models for VP and DBS samples separately. Median values of the estimated parameters from the SIR were in good agreement with the NONMEM point estimates, and the 95% confidence intervals were narrow, indicating acceptable precision. The CL/F was 31% higher in patients when antibodies to GLM were present. Patients who had previously received biologicals had a 4-fold higher Vp/F. The unexplained interindividual variability remained large after the introduction of the two covariates (57% and 221% for CL/F and Vp/F, respectively). A total of 14/40 patients (35%, 16/56 had no endoscopy data available) achieved ER. A GLM TC at week 14 was the best predictor of ER (lowest objective function value). A GLM TC at week 14 of 0.5 mg/L [48%] corresponded to a 50% probability of going from MES 3 to 2. A GLM TC at week 14 of 4.1 mg/L [36%] corresponded to a 50% probability of going from MES 2 to 1 or 0. Targeting the patients in our cohort (55% and 45% at MES 3 and 2 at baseline, respectively) to the TC of 3.2 mg/L at week 14 predicted a 41% probability of achieving ER:
0.55x[(3.2/0.5)/(1+(3.2/0.5))x(3.2 /4.1)/(1+(3.2 /4.1))] + 0.45x[(3.2/4.1)/(1+(3.2/4.1))]
= 0.55×0.38 + 0.45×0.44 = 0.41.
Patients with MES at baseline of 3 and 2 had a 38% and 44% chance for achieving ER at w14, respectively.
Conclusion: Our popPK and exposure-response models allow dose selection for targeting a certain GLM exposure and the associated probability of ER.
References:
[1] Peyrin-Biroulet L et al. Am. J. Gastroenterol (2015) 110, 1324–1338.
[2] Adedokun OJ et al. J Crohns Colitis (2017) 11, 35–46.
[3] Detrez I et al. AAPS J (2018) 21, 10.
[4] Detrez I et al. J Crohns Colitis (2016) 10, 575–581.
[5] Stefanovic S et al. ECCO (2018) P632.
[6] Kothare PA et al. AAPS J (2016) 18, 519–527.
[7] Dreesen E et al. Br J Clin Pharmacol (2019) doi:10.1111/bcp.13859
[8] Dosne AG et al. J Pharmacokinet Pharmacodyn (2017) 44, 509–520.
Reference: PAGE 28 (2019) Abstr 9103 [www.page-meeting.org/?abstract=9103]
Poster: Drug/Disease Modelling - Other Topics