Sudeep Pradhan (1), Daniel F.B. Wright (1), Stephen B. Duffull (1)
(1) School of Pharmacy, University of Otago, Dunedin, New Zealand
Background: Methods to adjust doses of drugs that are renally cleared are based on the assumption of a linear relationship between renal drug clearance (CLR) and glomerular filtration rate (GFR). The theory underpinning this practice is the Intact Nephron Hypothesis (INH) [1]. A recent review suggested that the INH might not be a suitable general model for renal drug clearance, particularly for drugs that are cleared primarily by active tubular processes where a non-linear relationship between CLR and GFR might be expected [2]. To date, the study designs required to detect a deviation from the INH that might influence dose recommendations from renal drug studies have not been explored.
Objective: To evaluate the phase 1 study designs for pharmacokinetic studies in patients with renal impairment recommended by the European Medicines Agency (EMA) [3] and the United States Food and Drug Administration (FDA) [4] for the purpose of delineating a non-linear relationship between CLR and GFR i.e., to test the INH.
Methods: The EMA and the FDA guidelines are essentially similar with main difference being recommended method for describing renal function i.e., using an exogenous marker for measured GFR (mGFR) and a serum creatinine based equation for estimating GFR (eGFR), respectively. Two models were proposed to describe the relationship between CLR and GFR:
1. M1: a linear model (INH scenario)
CLR = THETA(1) * GFR
2. M2: a nonlinear model (non-INH scenario)
CLR = THETA(1) * GFR ^ THETA(2)
where, GFR is mGFR or eGFR for the EMA or the FDA guidelines, respectively; THETA(1) is a linear coefficient parameter and THETA(2) is an exponent parameter. The value of the nonlinear exponent was based on the work of [5].
This is a stochastic simulation estimation study. Virtual subjects were simulated with GFR values stratified to the 4 renal function groups as defined by the respective guidelines. The number of subjects for each simulated study was n = 4, 8, 12, 16, 20, 24, 32, 48, 72, 120, 240, 480, 1080; and they were equally distributed across the renal function groups. Simulations were performed assuming M2 as the “true model” i.e., a non-linear relationship between CLR and GFR, using MATLAB (R2016b) to generate data sets. The simulated data sets were fitted to both the M1 and the M2 using NONMEM (version 7.3). Alpha error was calibrated to be 5% for every simulation setting. The preferred model for each simulated trial was based on the log likelihood ratio test. The simulation and estimation steps were repeated 1000 times for each of the designs tested. Power, relative standard error (RSE) and bias were calculated to evaluate the designs based on the EMA and the FDA guidelines.
Results: Study designs under the EMA guideline with ≥ 8 subjects had ≥ 80% power to correctly detect a non-linear relationship between CLR and GFR. Under the FDA guidelines, ≥ 80% power was achieved only with ≥ 24 subjects. The linear coefficient (THETA(1)) was well estimated when power was ≥ 80% for the designs under both the EMA (n ≥ 8) and the FDA (n ≥ 24) guidelines with a RSE of < 25%. The nonlinear exponent (THETA(2)) was poorly estimated (RSE ≤ 59%) for the designs under both the EMA and FDA guidelines, even when power was ≥ 80%. Of note all designs using eGFR (FDA) yielded a slight bias in the estimate of the nonlinear exponent which was not evident for the mGFR (EMA), where bias was < 20%.
Conclusions: The designs under EMA guideline would require fewer subjects to achieve ≥ 80% power to detect non-linear relationship between CLR and GFR compared to those based on the FDA guideline. This was entirely predicated on the choice of method used to estimate renal function with eGFR being a poor choice compared to mGFR, the latter providing an unbiased measure of GFR [6]. Note however that precision of estimate of the nonlinear exponent was independent of the measure of GFR with a requirement of up to 72 subjects (across the 4 renal function groups) being required to achieve a RSE of < 25%. In essence, therefore, both guidelines perform equivalently for the limiting feature of the design (parameter precision). In conclusion, if the non-INH holds then neither guidelines, EMA guidelines (n=24) and FDA guidelines (n=24) would provide a precise estimate of the true relationship between CLR and GFR.
References:
[1] Bricker NS et al. Am J Med, 28:77-98, 1960.
[2] Pradhan S et al. Eur J Clin Pharmacol, 75(2):147-156, 2018.
[3] European Medicines Agency and Committee for Medicinal Products for Human Use (CHMP),(2016)
[http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2016/02/WC500200841.pdf].
[4] Food and Drug Administration and Center for Drug Evaluation and Research (CDER), (2010) [http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM204959.pdf]
[5] Wright DFB and Duffull SB. Br J Clin Pharmacol, 83:1869-1872, 2017
[6] Kooman JP. J Magn Reson Imaging, 30(6):1341-1346, 2009
Reference: PAGE 28 (2019) Abstr 8833 [www.page-meeting.org/?abstract=8833]
Poster: Study Design