Roeland E. Wasmann (1), Elin M. Svensson (1,2), Stein J. Schalkwijk (1), Roger J. Brüggemann (1), Rob ter Heine (1)
(1) Department of Pharmacy, Radboud Institute for Health Sciences, Radboud University Medical Center, Nijmegen, The Netherlands, (2) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives:
Choosing the correct size descriptors for your pharmacokinetic model is important, especially for drugs with a narrow therapeutic index (e.g. aminoglycosides),. Like humans, body size descriptors come in many shapes and sizes, with total body weight (TBW) and fat-free mass (FFM) currently being most accepted [1, 2].
Recently, estimation of normal fat mass (NFM) has been advocated [3]. The NFM size descriptor is the sum of the predicted FFM (based on height, weight and sex) and the estimated relative contribution of fat mass. This method is increasingly employed [4-15].
NFM is estimated using equation 1, where Ffat reflects the contribution of fat relative to FFM. NFM=FFM+Ffat*(TBW-FFM) [eq.1]
Ffat can be considered a drug-specific parameter, suggesting that it can be used for extrapolation of the pharmacokinetics to populations with different body sizes [3]. It remains unclear whether NFM (with estimation of Ffat) can be reliably estimated in typical pharmacokinetic studies although it can be highly relevant in dose selection. Therefore, we investigated the identifiability of NFM in typical pharmacokinetic studies.
Methods:
As a best case scenario, we simulated pharmacokinetic data from a 1-compartment model with first order elimination, a clearance (CL) of 0.693 L/h, volume of distribution (V) of 1 L, inter-individual variability (IIV) of 30% on CL and V, and proportional residual error of 15%. Rich time-concentration data after a single bolus of 1 mg were simulated over a time span of 3 half-lives (n=8 samples). We investigated the identifiability of Ffat in 16 virtual drugs consisting of all possible combinations of Ffat of 0, 0.5, 1 or 5 for CL and V.
We chose a balanced study population containing individuals with a wide range of body sizes: each pharmacokinetic study consisted of three arms (50% male): 1/3 non-obese (BMI 18.5-30 kg/m2), 1/3 obese (BMI 30-40 kg/m2) and 1/3 morbidly obese (BMI >40 kg/m2). Real body size data for each population were randomly sampled from the NHANES database [16].
First, for every 16 virtual drugs we simulated 1000 large studies containing 10,000 individuals each, this served as a reference confirming that Ffat could be identified at all. Next, we simulated 1000 typical studies containing 30 subjects per study. After simulation, we re-estimated all parameters (allometric coefficients fixed) using PsN 4.7.0 and NONMEM 7.3.0 with FOCE-I. We calculated median estimates and the 95% estimation interval (EI) for each drug and each of the two study sizes (large and typical). Feasibility was tested by comparing the 95%EI with specific feasibility criteria for each value of Ffat. Feasibility test was passed when the 95%EI was within the bounds of a criterion of -0.25-0.25, 0.25-0.75, 0.75-1.25, and >2 for an Ffat of 0, 0.5, 1, and 5, respectively.
For sake of simplicity, here, we report only identifiability of Ffat in typical studies of four virtual drugs where Ffat combinations were equal for both CL and V.
Results:
For the large studies with 10,000 subjects we found an unbiased estimated of Ffat with a high precision and all within the feasibility criteria. The identifiability of the estimated Ffat on CL in a typical PK studies was poor with median [95% EI] of 0.0 [-0.4-0.9], 0.5 [-0.1-2.1], 1.0 [0.1-4.4] and 5.4 [1.3-3.3*108] for the drugs with Ffat of 0, 0.5, 1 and 5 for CL. Although the observed bias was minimal, a high imprecision was observed. Poor identifiability was observed for Ffat on V with a median [95% EI] of 0.0 [-0.3-0.5], 0.5 [-0.1-1.6], 1.1 [0.3-3.0] and 5.2 [1.5-85] for a simulated Ffat of 0, 0.5, 1 and 5, respectively. Similar results were observed for the other virtual drugs.
Conclusions: We have shown that the identifiability of Ffat is excellent in (unreasonably) large studies. High imprecision, however, was observed for estimates of Ffat in more realistic studies. This could have consequences for dosing drugs with a narrow therapeutic index, especially at extreme weights. Therefore, NFM must be used with caution and, when used, one should consider the power of a study to reliably estimate NFM.
References:
[1] McLeay SC, Morrish GA, Kirkpatrick CM, Green B. The relationship between drug clearance and body size: systematic review and meta-analysis of the literature published from 2000 to 2007. Clin Pharmacokinet. 2012;51(5):319-30. doi:10.2165/11598930-000000000-00000.
[2] Janmahasatian S, Duffull SB, Ash S, Ward LC, Byrne NM, Green B. Quantification of lean bodyweight. Clin Pharmacokinet. 2005;44(10):1051-65.
[3] Holford NHG, Anderson BJ. Allometric size: The scientific theory and extension to normal fat mass. Eur J Pharm Sci. 2017;109S:S59-S64.
[4] Allegaert K, Olkkola KT, Owens KH, Van de Velde M, de Maat MM, Anderson BJ et al. Covariates of intravenous paracetamol pharmacokinetics in adults. BMC Anesthesiol. 2014;14:77.
[5] Cortinez LI, Anderson BJ, Holford NH, Puga V, de la Fuente N, Auad H et al. Dexmedetomidine pharmacokinetics in the obese. Eur J Clin Pharmacol. 2015;71(12):1501-8.
[6] Cortinez LI, Anderson BJ, Penna A, Olivares L, Munoz HR, Holford NH et al. Influence of obesity on propofol pharmacokinetics: derivation of a pharmacokinetic model. Br J Anaesth. 2010;105(4):448-56.
[7] Dorlo TP, Huitema AD, Beijnen JH, de Vries PJ. Optimal dosing of miltefosine in children and adults with visceral leishmaniasis. Antimicrob Agents Chemother. 2012;56(7):3864-72.
[8] Hopkins AM, Wojciechowski J, Abuhelwa AY, Mudge S, Upton RN, Foster DJ. Population Pharmacokinetic Model of Doxycycline Plasma Concentrations Using Pooled Study Data. Antimicrob Agents Chemother. 2017;61(3).
[9] McCune JS, Bemer MJ, Barrett JS, Scott Baker K, Gamis AS, Holford NH. Busulfan in infant to adult hematopoietic cell transplant recipients: a population pharmacokinetic model for initial and Bayesian dose personalization. Clin Cancer Res. 2014;20(3):754-63.
[10] Rhodin MM, Anderson BJ, Peters AM, Coulthard MG, Wilkins B, Cole M et al. Human renal function maturation: a quantitative description using weight and postmenstrual age. Pediatr Nephrol. 2009;24(1):67-76.
[11] Salem AH, Giranda VL, Mostafa NM. Population pharmacokinetic modeling of veliparib (ABT-888) in patients with non-hematologic malignancies. Clin Pharmacokinet. 2014;53(5):479-88.
[12] Smythe W, Khandelwal A, Merle C, Rustomjee R, Gninafon M, Bocar Lo M et al. A semimechanistic pharmacokinetic-enzyme turnover model for rifampin autoinduction in adult tuberculosis patients. Antimicrob Agents Chemother. 2012;56(4):2091-8.
[13] Tham LS, Wang LZ, Soo RA, Lee HS, Lee SC, Goh BC et al. Does saturable formation of gemcitabine triphosphate occur in patients? Cancer Chemother Pharmacol. 2008;63(1):55-64.
[14] Wright DF, Stamp LK, Merriman TR, Barclay ML, Duffull SB, Holford NH. The population pharmacokinetics of allopurinol and oxypurinol in patients with gout. Eur J Clin Pharmacol. 2013;69(7):1411-21.
[15] Zvada SP, Denti P, Geldenhuys H, Meredith S, van As D, Hatherill M et al. Moxifloxacin population pharmacokinetics in patients with pulmonary tuberculosis and the effect of intermittent high-dose rifapentine. Antimicrob Agents Chemother. 2012;56(8):4471-3.
[16] Centers for Disease Control and Prevention (CDC), National Center for Health Statistics (NCHS). National Health and Nutrition Examination Survey Data. Hyattsville, MD: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention,. 2018. www.cdc.gov/nchs/nhanes. Accessed 14-02-2018.
Reference: PAGE 27 (2018) Abstr 8443 [www.page-meeting.org/?abstract=8443]
Poster: Drug/Disease Modelling - Other Topics