Simultaneous population pharmacodynamic modelling of the growth hormone and insulin-like growth factor-I effects after deep subcutaneous administration of Lanreotide Autogel® in acromegalic patients. Application of Nonparametric estimation method in NONMEM
María J. Garrido1, Josep-María Cendrós2, Joaquim Ramis2, Concepción Peraire2, Rosendo Obach2, Iñaki F. Trocóniz1
1, Pharmacy and Pharmaceutical Technology; School of Pharmacy; University of Navarra; Pamplona; 2, Drug Metabolism, Pharmacokinetic and Immunology Department; Ipsen Pharma S.A; Sant Feliu de Llobregat; Barcelona ; Spain
Background: Acromegaly arises from excessive growth hormone (GH) production, which stimulates abnormally high secretion of insulin-like growth factor (IGF-I). Synthetic somatostatin analogues (SSA) inhibit GH and subsequently IGF-I secretion. However a quantitative description of the relationship between the changes in IGF-I levels as a function of SSA-induced GH decrease is not available.
Objectives: To develop a population pharmacodynamic model describing the relationship between serum levels of the SSA lanreotide Autogel® (lan ATG) and its effects on GH and IGF-I in acromegalic patients.
Methods: A phase II multicentre, randomised study was conducted in 108 patients who may or may not have been previously treated with SSAs. Patients received placebo or lan ATG at doses of 60, 90, or 120 mg once every four weeks by deep subcutaneous injection. This 52-week study had four phases: (i) washout (weeks -12 to 0) for previously treated patients, (ii) double-blind, placebo controlled (weeks 0 to 4): single dose of lan ATG or placebo, (iii) single-blind, fixed-dose (weeks 4 to 20): four injections of lan ATG, and (iv) open-label dose titration of lan ATG (weeks 20 to 52). Lanreotide (Cmin) levels, GH, and IGF-I in serum were measured simultaneously at pre-treatment and weeks 4, 13, 16, and 52. GH levels were described as a function of the observed Cmin, and IGF-I levels were modelled as a function of the predicted GH, using the population approach with NONMEM VI and NONPARAMETRIC option.
Results: No evidence of hysteresis loops between Cmin, GH, and IGF-I were found in the data. The GH vs Cmin relationship was described with an inhibitory sigmoidal EMAX model, and the IGF-I vs GH relationship was modelled with an EMAX model. Despite the fact that the goodness of fit plots and visual predicted checks suggested that the model was adequate, further model validation based on posterior predicted checks [(PPC) (using the percentage of patients with normalized levels of the two hormones as main descriptor of the data] revealed clear bias. Results from the PPC were greatly improved when the final population model was re-run using the NONPARAMETRIC method and simulations were performed accordingly. Results from model-based simulations provided a simulated percentage of patients with normalized levels of GH, IGF-I, and both hormones of 66, 51, and 40 %, respectively after 16 weeks of treatment with lan ATG 120 mg, which were consistent with clinical observations.
Conclusion: A pharmacodynamic model that can describe simultaneously the reduction of GH and IGF-I levels after lan ATG administration was constructed. Model evaluation confirms (i) the existence of non-symmetric ETA distributions that could be handled properly, and (ii) the efficacy of lan ATG on GH and provides a tool to handle the related changes in IGF-I.