Abstract
Patients with impaired kidney function are at risk of excessive drug exposure and toxicity when standard doses of renally cleared drugs are used. Dose adjustment to normalise drug exposure between patients may be required to optimise the safety and effectiveness of treatment. Renal drug dosing in clinical practice generally assumes a linear relationship between dose and glomerular filtration rate (GFR). The theory underpinning this assumption is the intact nephron hypothesis (INH). The INH suggests that kidney disease is the result of a reduction in the number of complete (intact) nephrons. Under this model, renal drug clearance is assumed to be a linear function of glomerular function while tubular handling is not accounted for. The INH is a widely accepted model for renal drug handling even though it was not originally developed for this purpose.
There is concern that the INH may overly simplify renal drug handling such that it may not, in all cases, be a suitable model for renal dose adjustment. The exception will be drugs with low protein binding that are handled almost entirely by glomerular filtration (e.g. gentamicin, vancomycin). Here the renal clearance of the drug would be expected to approximate GFR and the INH should hold as a suitable model to describe renal handling. However, there is increasing evidence to suggest that the assumption of a linear relationship between GFR and renal clearance for drugs that are mainly secreted by the tubules may not be reasonable. This represents a large number of commonly prescribed drugs such as metformin, ACE inhibitors, some beta-lactam antibiotics, furosemide, and some anticancer drugs (e.g. methotrexate, cisplatin).
Renal dose adjustment generally recommended in clinical practice hinges on the INH being true. However, this assumption has not been thoroughly investigated. The standard phase I renal drug studies assessing the need for dose adjustment in patients with kidney disease in drug development also rely on the INH being true. This means that dosing information on product labels are predicated on the INH. Equally, it is not known if the use of estimated GFR (such as Cockcroft-Gault) based on creatinine as a renal biomarker, itself subject to tubular secretion, might distort the relationship that exists between drug clearance and GFR.
In this thesis, it is hypothesised that the INH may not be a suitable model for renal drug handling and hence dose prediction for drugs that are cleared predominantly by tubular secretion. This raises two falsifiable predictions; 1) for drugs that are secreted predominantly by tubular secretion the relationship between renal drug clearance and GFR will be non-linear, and, 2) renal drug study design will influence the ability to capture the non-linear relationship. The aims of this thesis are to investigate the INH in the context of renal drug handling and to propose study designs for testing the INH.
The first prediction was explored by performing a systematic literature review and two model-based analyses. The evidence-base explored suggests that the INH may not be a suitable model for renal drug handling. Additionally, the studies that did not support the INH were robust in some design components in comparison to those supporting the INH. The INH was further explored by performing model-based analyses using data from a renal drug study using probes for glomerular and tubular function. The renal drug study design appeared to be inadequate to inform about the appropriateness of the INH as a model for renal drug handling.
The second prediction was investigated by evaluating the performance of different renal drug study designs. The European Medicines Agency (EMA) based renal drug study designs performed better compared to those based on the Food and Drug Administration (FDA) in terms of power calculated as the probability of correctly identifying the “true” relationship between renal drug clearance and GFR over the misspecified relationship under a given study design. The performance of the study design was dependent on the method used to determine GFR. However, the performance of both the designs were poor in terms of parameter precision when standard number of subjects (n = 24) were enrolled. An optimal design methodology was used to construct a composite study design for the purpose of testing the INH that provided an estimate of parameter with adequate precision with the standard number of subjects (n = 24).
Three difference approaches were identified and explored for the purpose of testing the INH using data from a renal drug study. The three approaches were based on; 1) comparison of renal drug clearance with respect to GFR, 2) linear regression of renal drug clearance against GFR, and, 3) the time course of amount of drug excreted in urine, where subjects with a range of renal function were included. The utility of the three approaches to test the INH were successfully demonstrated.
The findings in this thesis suggest that the INH may not be a suitable general model for renal drug handling. An empirical non-linear model proposed in this thesis can be used to account for both non-linear (non-INH) and linear (INH) renal drug handling. Moreover, appropriately designed studies are required for accurately estimating the change in renal drug clearance with respect to change in renal function and thus improve the prediction of required dose adjustment. This research represents a step towards the safe and effective use of medicines in patients with kidney disease.