Eur J Clin Pharmacol (2014) 70:1403–1404 DOI 10.1007/s00228-014-1740-8

LETTER TO THE EDITORS

Between-subject variability: should high be the new normal? Hesham S. Al-Sallami & Song Lim Cheah & Shiou Yii Han & Joel Liew & Jin Lim & Mary Anne Ng & Hayneil Solanki & Run Jie Soo & Victoria Tan & Stephen B. Duffull

Received: 30 July 2014 / Accepted: 18 August 2014 / Published online: 5 September 2014 # Springer-Verlag Berlin Heidelberg 2014

The pharmacokinetics of numerous drugs are said to be predictable [1–5]. This is often seen as an advantage and implies the ease of dosing or dose adjustments of these drugs. However, predictability requires both accuracy and precision. In the context of pharmacokinetics, precision refers to the ability to achieve a specified target concentration in different individuals. Precision is the inverse of between-subject variance (BSV), i.e. the greater the BSV, the less precise/predictable a parameter is across a patient population. BSV in pharmacokinetic (PK) parameters is usually quantified by the coefficient of variation (CV%). The current convention is that BSV in PK parameters is considered “low” (CV%≤10 %), “moderate” (CV%∼25 %), or “high” (CV%>40 %) [6]. We contend that the average CV% in PK parameters in the population is normally high. In other words, we hypothesise that a CV% of 40 % is actually normal and a CV%≤10 % is abnormally low. A literature review of population PK studies from various data sources was conducted. We reviewed the range of BSV values of PK parameters reported for patient populations of preselected drug classes. Drug classes studied included psychotropics, immunosuppressants, cardiovascular drugs, and antibiotics. Estimates of clearance (CL), volume of distribution (V), absorption rate constant (ka), and their corresponding CV% were recorded. PK studies in healthy volunteers were excluded from the review. A total of 182 studies involving 95 drugs were found (see “Appendix”). For the purpose of illustration, we report on the values of CL and V only. We extracted BSV values from the

final PK models that accounted for covariates and report here as CV%. The mean CV% in CL was 40.3±24 % and in V was 51.3±40.4 %. The mean CV% in CL in predominately renally cleared drugs was 31 % and in those predominately hepatically cleared drugs was 47.4 %. For the nonintravenously administered drugs, the BSV refers to the between-subject variability in apparent oral clearance. Additionally, drugs with low bioavailability (<70 %) had a mean CV% in CL of 45.5 % whereas drugs with high bioavailability had a mean value of 35.4 %. Age, sex, weight, and renal function were among the most significant covariates reported across the drug classes. It therefore seems that according to the current convention, it is the norm that drugs show moderate to high BSV. We believe the current convention needs to be recalibrated to consider that a low BSV in CL is <25 %, normal BSV in CL is 25–50 %, and high BSV in CL is >50 %. Clinically, this means that a normal level of variability in CL would result in a 4- to 5-fold variability in steady state average plasma concentrations and, therefore, for all drugs with a low therapeutic index, monitoring of plasma concentration or response and dose-individualisation will be essential. We have shown that a CV% in PK parameters of ∼40 % should be considered normal. This is not intended to be alarming, rather a reflection of the actual variability inherent in patient populations. Using the loaded term “predictability” in describing the PK of a drug is not helpful, and instead the BSV in PK parameters of the drug needs to be stated.

Electronic supplementary material The online version of this article (doi:10.1007/s00228-014-1740-8) contains supplementary material, which is available to authorized users.

References

H. S. Al-Sallami (*) : S. L. Cheah : S. Y. Han : J. Liew : J. Lim : M. A. Ng : H. Solanki : R. J. Soo : V. Tan : S. B. Duffull School of Pharmacy, University of Otago, Dunedin, New Zealand e-mail: [email protected]

1. Bereznicki LR, Peterson GM (2010) New antithrombotics for atrial fibrillation. Cardiovasc Ther 28(5):278–86 2. Young JJ, Kereiakes DJ, Grines CL (2000) Low-molecularweight heparin therapy in percutaneous coronary intervention:

1404 the NICE 1 and NICE 4 trials. National investigators collaborating on enoxaparin investigators. J Invasive Cardiol 12:Suppl E–E14 3. Miceli JJ, Wilner KD, Hansen RA, Johnson AC, Apseloff G, Gerber N (2000) Single- and multiple-dose pharmacokinetics of ziprasidone under non-fasting conditions in healthy male volunteers. Br J Clin Pharmacol 49(Supply 1):5S–13S

Eur J Clin Pharmacol (2014) 70:1403–1404 4. Benet LZ, Zech K (1994) Pharmacokinetics—a relevant factor for the choice of a drug? Aliment Pharmacol Ther 8(Suppl 1):25–32 5. Peck AW (1991) Clinical pharmacology of lamotrigine. Epilepsia 32(Spply 2):S9–12 6. Rowland M, Tozer TN (2011) Variability. Clinical pharmacokinetics and pharmacodynamics: concepts and applications, 4th edn. Lippincott Williams & Wilkins, Philadelphia, pp 333–55