Drug dosing and estimates of kidney function

Serum (or plasma) creatinine, creatinine clearance (CrCl), estimated creatinine clearance (eCrCl) and estimated glomerular filtration rate (eGFR) are used to assess kidney function in order to detect and manage kidney disease and to optimize the dosing of renally excreted medicines.

Despite improvements in assay performance following the introduction of isotope dilution mass spectrometry (IDMS)-traceable creatinine methods, including enzymatic methods, serum creatinine results can still vary by ± 13% between laboratories at levels of 85 µmol/L. ¹ In addition to methodological issues, there are other limitations to serum creatinine as a measure of kidney function. For example, serum creatinine can vary by about 10% depending upon meat intake; ² creatinine is secreted in the proximal tubule and at low urine flow rates can be reabsorbed; ² extra-renal excretion of creatinine can also occur. ³ Creatinine is not an ideal filtration marker.

CrCl is calculated using methods that do not measure the ‘true’ creatinine, but which estimate its concentration in blood and a timed urine collection. To avoid collecting a timed urine, numerous equations and nomograms have been devised to estimate CrCl from serum creatinine. In estimating CrCl, it is assumed that creatinine excretion is constant and equal to creatinine production, which itself is proportional to muscle mass and can, in turn, be estimated from an individual’s age, gender and weight. eCrCl will overestimate CrCl in obese or oedematous patients and in those with reduced creatinine production. eCrCl also assumes that serum creatinine concentration is at steady state.

eCrCl is probably most commonly calculated using the Cockcroft and Gault (C&G) equation first described in 1976 ⁴ using serum creatinine measured with a kinetic Jaffe assay on a Technicon Autoanalyser II. In principle, the technique used to determine the patient’s serum creatinine should be the same as the one used to derive the equation from which the eCrCl (or eGFR) are calculated: ² this can clearly not be the case with modern use of the C&G equation. Other equations have now been proposed including the Modification of Diet in Renal Disease (MDRD) study and Chronic Kidney Disease Epidemiology (CKD-EPI) collaboration equations^5,6 that estimate GFR normalized to a body surface area (BSA) of 1.73 m². All equations are valid only for the population in which they were derived and some have been produced for more specific purposes, e.g. the Wright formula for cancer patients. ⁷ Wright et al. ⁷ recognized that such equations were also serum creatinine method dependent.

The C&G equation has been and is used in pharmacokinetic studies. ⁸ It is therefore included in drug dosage information despite the fact that the serum creatinine method used to derive it is no longer available and despite the fact that IDMS-traceable creatinine assays tend to give lower results and therefore a higher eCrCl than the earlier non-IDMS traceable creatinine assays. ⁹

In 2005, the Department of Health ¹⁰ recommended the use of the MDRD equation to detect chronic kidney disease (CKD) in the UK. Since then eGFR has been routinely reported and the relative merits of using eGFR and the C&G eCrCl to guide drug dosing have been debated in the literature. On occasions, articles arguing for eGFR ¹¹ have appeared in the same edition of the same journal as articles supporting the continued use of the C&G eCrCl. ¹²

In this edition of the Annals, Chin et al. ¹³ compare the performance of C&G, MDRD and CKD-EPI equations in predicting gentamicin clearance. The authors conclude that both the CKD-EPI and C&G equations provide reasonable estimates of gentamicin clearance in most situations but that body size should be accommodated when using estimating equations at extremes of body size. This is consistent with published guidance. ¹⁴ It is clear that creatinine and its derivative equations are affected strongly by muscle mass. ¹⁵ Yet, how many laboratories are capable of reporting BSA-adjusted eGFRs and how many clinicians will perform this calculation themselves in the frail or the obese?

What does this mean in clinical practice? Drug dosing can clearly be heavily influenced by the serum creatinine method and the equation used to estimate CrCl/GFR. This is especially true in the frail or the obese. Accurate dosing of gentamicin is important to optimize antimicrobial activity and minimize risks of harm. This is complicated by the multitude of different ways in which gentamicin is used clinically (e.g. different dosing regimens for endocarditis, gram-negative sepsis and surgical prophylaxis). Extended-interval gentamicin dosing (EIGD) regimens are in common use; a region wide ‘Yorkshire Hartford Gentamicin Regimen’ (http://nww.lhp.leedsth.nhs.uk/common/guidelines/detail.aspx?id = 1944), for example, uses the C&G (total body weight version) to initially assess suitability for EIGD. Beyond this initial assessment, CrCl calculators do not have a role, as therapeutic drug monitoring is used to modify dose. The importance of weight and obesity in gentamicin dosing are not well appreciated. Gentamicin does not distribute well into adipose tissue, so an obese patient might be given a larger dose than is ideal if their actual (total) body weight is used in dosing.

Approximately, 10% of drugs are cleared entirely by renal pathways ¹⁶ and for some drugs, e.g. carboplatin, blood concentrations are not monitored so dosage is determined using measured and/or estimated GFR. ¹⁷ For patients to receive the correct dose of a renally cleared drug, where eCrCl/eGFR is used to guide dosing, laboratories responsibilities extend beyond merely supplying serum creatinine results and an eGFR. Laboratories should be properly engaged with colleagues prescribing drugs to ensure that the most appropriate equation is used to calculate the eCrCl or eGFR including an assessment as to whether body weight/surface area adjustments are appropriate.

Serum creatinine is an excellent example of why we must be engaged with our users to ensure that the number we generate is understood and used appropriately whether to detect/monitor CKD or acute kidney injury or to ensure that a patient receives the correct dose of a renally cleared drug. Only if we do so, can we play our part in ensuring that we achieve optimum patient outcomes. Chin et al. should be commended for their meticulous validation of estimating equations in the clinical context of drug dosing.