Drug therapy in kidney disease

Imagine an 80-year-old 50-kg woman with liver impairment to whom you want to give, let us say, morphine. You know that the dose should be reduced in liver disease. You know that it is metabolized by the liver and that it has an increased pharmacodynamic action in liver impairment, with a risk of hepatic encephalopathy. However, you are unsure whether the patient's liver impairment is moderate or severe. After all, you know that liver function tests do not correlate well with the degree of functional impairment. So you reduce the dose somewhat and monitor the patient's progress carefully, observing conscious level and measuring liver function from time to time. It's not easy and you don't feel at all confident.

Now suppose you want to give morphine to a patient with renal impairment. You know that it has an active metabolite that is eliminated by the kidneys and that you should reduce the dose in moderate to severe renal impairment. So you measure the creatinine clearance and make a good stab at working out an appropriate dose. Well, actually you don't measure the creatinine clearance at all, because nobody does that these days. What you do is ask the laboratory for renal function tests, and the answer comes back that the serum creatinine concentration is 240 µmol/l and the eGFR (estimated glomerular filtration rate) is 18 ml/min/1.73 m². Now, given such an apparently accurate assessment of the patient's renal function, you feel really confident about your dosage calculation. But how confident should you feel?

Glomerular filtration rate gives a reasonably good estimate of overall renal function, but it is not practical to measure it routinely, and so creatinine clearance is used as a surrogate. That in turn also provides practical problems, and serum creatinine concentration is generally used as a surrogate for that. But the serum creatinine concentration does not bear a linear relation to the glomerular filtration rate or the creatinine clearance – indeed it changes very little as glomerular filtration rate falls to about 50 ml/min (Figure 1) [1]. Equations have therefore been derived to estimate the creatinine clearance from the serum creatinine concentration, taking into account this non-linear relation and the effects of age, weight, and sex. The most commonly used of these is the Cockcroft–Gault equation [2]:

Because this formula is inaccurate at low creatinine concentrations, correction factors have also been introduced to deal with creatinine concentrations below 85 µmol/l [3] or 60 µmol/l [4].

Figure 1.

The mean relation between serum creatinine concentration and glomerular filtration rate (GFR) in men, measured using iothalamate [solid line, adapted from reference 1]. The dotted line is the hypothetical relation for patients with burns.

In an attempt to provide a more accurate estimate of glomerular filtration rate, the data from the Modification of Diet in Renal Disease (MDRD) study have been analysed [1] and a modified MDRD equation [5] derived:

However, this equation is also inaccurate. For example, it underestimates glomerular filtration rate at low values [6] and there is therefore a risk that its use might be associated with unnecessary referrals for investigation of end-stage kidney disease.

In this issue of the Journal we publish evidence that these equations are not accurate in critically ill patients with burns [7]. In such patients the serum creatinine concentration is reduced, for various reasons, not all associated with altered renal function – reduced creatinine production (due to reduced muscle mass and reduced cellular secretion), reduced renal tubular creatinine secretion, and hypermetabolism. The authors suggested that a reduced volume of distribution might also contribute, but that would not affect the serum creatinine concentration at steady state. Thus, serum creatinine in these patients can be within the reference range over a wider range of values of glomerular filtration rate than one would normally expect, as shown in the hypothetical relation in Figure 1. Nevertheless, the data suggest that the Cockcroft–Gault equation gives a reasonable prediction of creatinine clearance at values below 100 ml/min, and better than the prediction given by the modified MDRD equation.

Other papers published in this issue of the Journal deal with the palatability of angiotensin receptor antagonists in children with acute or chronic kidney diseases [8] and the effects of gastric emptying on the pharmacokinetics of oral mycophenolic acid in renal allograft recipients [9]. In the former, pulverized candesartan was significantly more palatable than four other pulverized forms of angiotensin receptor antagonists. In the latter, impaired gastric emptying in some patients with stable renal allografts reduced the C_max of oral mycophenolic acid and delayed the t_max but did not affect overall absorption. However, in neither case was it clear that kidney disease was directly involved in the effects shown.

The introduction of the glomerular filtration rate, as estimated by the MDRD equations, has highlighted the problem of estimating renal function at the bedside. No method is perfect and there is no standard classification of degrees of severity of renal impairment. When adjusting drug dosages in patients with renal impairment, several principles should be remembered. Firstly, protein binding can alter [10], either because of altered affinity of serum albumin in chronic renal insufficiency or because of hypoalbuminaemia in the nephrotic syndrome. Secondly, the pharmacodynamic actions of drugs can be affected by impaired renal function; for example, impaired platelet aggregation in renal insufficiency can alter the effects of antiplatelet drugs [11]. Finally, the studies on which changes in drug dosage in renal insufficiency are based usually involve the use of creatinine clearance, rather than estimated glomerular filtration rate. The failure of the MDRD study to yield clear results about the efficacy of diet in the management of chronic renal insufficiency did not mitigate the potential usefulness of the data that were collected during the study. However, estimation of glomerular filtration rate, based on the MDRD data, does not have a place in adjustment of drug dosages in patients with renal insufficiency. For this purpose it is probably best to go on using the Cockcroft–Gault equation.

Glossary

Box 1

British Journal of Clinical Pharmacology

Editor-in-Chief

The British Pharmacological Society is seeking a clinically qualified member of the Society, with suitable experience, to appoint as Editor-in-Chief of the British Journal of Clinical Pharmacology from the start of 2008, when Jeff Aronson steps down. The position will carry an honorarium (unless the incumbent is a Trustee of the Society).

The Editor-in-Chief will be expected to provide strong leadership to the Journal over the next 5 years and to maintain close and effective links with the staff of the British Pharmacological Society and with the Editors and publishers of the British Journal of Clinical Pharmacology. The person appointed will have excellent English language skills.

Individuals who want to be considered should apply to the Chairman of the Clinical Pharmacology Section Committee, Robin Ferner (r.e.ferner@bham.ac.uk), who can provide further details.

Closing date: 30 June 2007