Ricardo Nalda-Molina
University of Miguel Hernandez de Elche
Objectives:
To evaluate the predictive performance and adequacy of four population pharmacokinetic models (PopPK) of adalimumab in adult patients diagnosed of inflammatory bowel disease (IBD).
Material and methods:
A retrospective observational study was performed, with the following inclusion criteria: Adult patients with ulcerative colitis or Crohn’s disease treated with adalimumab, with at least one trough concentracion (TC) between 2014 and 2018 were included. Four different pharmacokinetic models were evaluated: FDA, 2008 (Mod-A)[1], Ternant et al, 2015 (Mod-B)[2], Sharma et al, 2015 (Mod-C)[3] and Berends et al, 2018 (Mod-D)[4]. The models were implemented in NONMEM® v7.3.
The individual and population predictions of adalimumab concentrations were estimated from the four PopPK models, by calculating the empirical bayesian of estimates (EBEs). Two different data sets were created from the original dataset to evaluate the model adequacy and predictive performance; 1) DATASET-1: To evaluate the model adequacy, all the patients and TC were included, and their population predictions were compared with the observed TC; 2) DATASET-2: To assess the predictive performance, only the patients with two or more TC were included. Only the first TC of these patients were used to estimate the EBE, and the individual prediction that were left out were compared with the observed TCs.
To validate these models, bias and precision of estimated concentrations were calculated as the mean predictive error (MPE) and the mean square predictive error (MSPE) in the population, respectively.
Results:
In the DATASET-1, 171 patients and 245 TCs were included. The mean values (95% CI) of weight, basal albumin and TC were: 69.5 kg (43.5-100), 3.82 g/dL (2.53 : 4.70) and 5.70 mg/L (0.1 : 13.5), respectively. In the DATASET-2, 55 patients and 74 TCs were included. The mean values (95% CI) of weight, basal albumin and TC were 68.9 kg (49.5 : 89.6), 3.77 g/dL (2.24 : 4.74) and 6.93 mg/L (0.505 : 17.74), respectively. 5.85% of patients in DATASET-1 and 3.64% in DATASET-2 developed antibodies anti-adalimumab.
The bias (95% CI) of the predictions for the model adequacy and predictive performance were: Mod-A: -5.26 (-5.95 : -4.57) and -0.906 (-1.99 : 0.175); Mod-B: -2.88 (-3.47 : -2.29) and -0.666 (-1.71 : 0.376); Mod-C: -3.71 (-4.34 : -3.01) and -2.84 (-3.95 : -1.72); Mod-D: -3.06 (-3.66 : -2.46) and -1.77 (-2.89 : -0.643), respectively.
The precision (95% CI) of the predictions for the model adequacy and predictive performance were: Mod-A: 7.61 (6.80 : 8.42) and 4.80 (2.97 : 6.63), Mod-B: 5.52 (4.88 : 6.16) and 4.59 (2.83 : 6.35), Mod-C: 6.26 (5.52 : 7.00) and 5.63 (4.22 : 7.04); Mod-D: 5.67 (4.92 : 6.41) and 5.20 (3.56 : 6.85), respectively.
Conclusions:
The evaluation of the adequacy of the four adalimumab PopPK models in IBD patients shows that all the four models overestimate adalimumab concentrations in the population with a statistic significance<0.05, although Mod-B had a better bias and precision, (i.e. closer to zero). The evaluation of the predictive performance, shows again an overestimation of the adalimumab TC, and MOD-B showed a better prediction performance of the TCs.
References:
[1] MOD-A: Food and Drug Administration (FDA), (2007) Approval letter – Humira. Application Number 125057 / s-0089
[2] MOD-B: Pharmacokinetics of adalimumab in Crohn’s disease. Ternant D, et al. Eur J Clin Pharmacol. 2015 Sep;71(9):1155-7.
[3] MOD-C: Pharmacokinetics and exposure-efficacy relationship of adalimumab in pediatric patients with moderate to severe Crohn’s disease: results from a randomized, multicenter, phase-3 study. Sharma S et al . Inflamm Bowel Dis. 2015 Apr;21(4):783-92.
[4] MOD-D: Explaining Interpatient Variability in Adalimumab Pharmacokinetics in Patients With Crohn’s Disease. Berends SE, et al. Ther Drug Monit. 2018 Apr;40(2):202-211.
Reference: PAGE 28 (2019) Abstr 9107 [www.page-meeting.org/?abstract=9107]
Poster: Drug/Disease Modelling - Other Topics