The confusing status of colistin dosing
Click Here to Manage Email Alerts
In the current era of increasing prevalence of multidrug-resistant gram-negative bacillary infections, previously shelved last-line therapies, such as the polymyxin agents colistin and polymyxin B, have made an obligatory comeback. Polymyxin B is emerging as the polymyxin with the most favorable and predictable pharmacokinetics, and it more rapidly achieves therapeutic serum levels than does colistin. Additionally, recent reports suggest that the nephrotoxicity potential of polymyxin B might be less than that of colistin. Despite these favorable characteristics, polymyxin B is not available in many parts of the world, and current and past clinical use of colistin has been much greater than polymyxin B in the United States and worldwide. Furthermore, there is much less information available on the pharmacology and clinical outcomes with polymyxin B. Because it has been the preferred polymyxin, this article will focus on dosing of colistin.
‘Target’ serum concentration
Although it is fortunate that older options such as colistin are still available, it is also problematic because the diminished use of this agent also caused a recession in research regarding its pharmacokinetics (PK) and pharmacodynamics (PD). The reintroduction of colistin and the lack of data surrounding optimal dosing has since inspired various investigations into uncovering the ideal dosing strategy. Unfortunately, although we are now armed with more robust PK/PD data, the optimal dosing strategy has yet to be determined owing to several factors, including differing package insert recommendations, high interpatient variability, and unpredictability of colistin serum concentrations.
Before one can understand the importance of being able to accurately predict the colistin serum concentration, it is necessary to have an appreciation for the PD target that most accurately predicts success with this agent. Most experts use a target colistin steady state concentration (Css,avg) of 2 mg/L when evaluating dosing strategies. Unfortunately, based on PD modeling, this might not be an ideal target concentration. Although such a concentration was sufficient to obtain a bactericidal effect in a mouse thigh model for colistin-susceptible isolates (ie, MIC 2 mg/L), significantly greater concentrations were needed to obtain a bactericidal effect in a mouse lung model. In fact, in many instances, even at maximum tolerated doses, bacteriostasis could not be achieved in isolates considered to be “susceptible” to colistin. Unfortunately, the target serum concentration of colistin cannot be further increased much beyond 2 mg/L because clinical data demonstrate that rates of nephrotoxicity significantly increase when concentrations are greater than 2.4 mg/L. Therefore, dosing regimens targeting concentrations of 2 mg/L are typically considered to be the maximum tolerated regimens.
Dosing colistin to reach concentrations within its narrow therapeutic window is further complicated by the fact that colistin is available only for IV administration as its inactive prodrug, colistin methanesulfonate (CMS), which is then hydrolyzed in vivo to the active drug product. Once formed, colistin is eliminated by nonrenal mechanisms; conversely, CMS undergoes rapid renal elimination. The conversion of CMS to colistin and simultaneous rapid CMS renal elimination make it nearly impossible to precisely estimate the concentration of formed colistin that will be achieved in each patient with any given dose of CMS regardless of dosing strategy.
Three approaches to dosing
Currently, there are three different approaches to colistin dosing: 1) that of the U.S. package insert; 2) that of the European Medicines Agency (EMA) package insert; and 3) dosing per the PK equation established by Nation and colleagues. It is important for clinicians to appreciate how well each of those dosing recommendations perform with regard to achieving the “target” Css,avg of 2 mg/L.
Before assessing the maintenance dose strategies with the various recommendations, it is important to note that given the slow rate of conversion from CMS to colistin and colistin’s subsequent long half-life, a loading dose should be strongly considered for patients receiving IV CMS to avoid delays of up to 48 hours in reaching therapeutic concentrations. We recommend 5 mg/kg, with a cap of 300 mg, as a loading dose.
The three available maintenance dose strategies have slight variations based on renal function and patient weight, resulting in a range of doses and, undoubtedly, considerable confusion for the provider unfamiliar with colistin. For example, consider a patient with a weight of 70 kg and a creatinine clearance (CrCl) of 80 mL/min (see Table 1). Using the U.S., EMA and Nation and colleagues’ dosing algorithms, doses of 350 mg of colistin base activity (CBA), 270 mg CBA, and 340 mg CBA, respectively, are recommended (in which 1 million international units [IU] equals 30 mg of CBA). Using the calculations derived by Nation and colleagues from patient-level data in a PK study of critically ill patients, these colistin doses would be predicted to result in Css,avg of 3.11 mg/L (U.S.), 2.4 mg/L (EMA), and 3.02 mg/L (Nation et al). Although, on average, each of these dosing recommendations would be predicted to achieve the target Css,avg of 2 mg/L, it is important to appreciate that the attainment of this serum concentration is not guaranteed due to the complicating factors regarding CMS and colistin PK described earlier. In fact, Nation and colleagues reported that even at a given dose and CrCl, colistin concentrations varied up to 10-fold among patients.
Although the above-mentioned scenario discusses patients with clearances at 80 mL/min, it is important to recognize that in patients with clearances greater than 80 mL/min, all three equations underperform, and resulting Css,avg concentrations are suboptimal. The primary concern in patients with CrCl above 80 mL/min is that they will often renally eliminate CMS faster than it is converted to active colistin, which therefore prevents the achievement of therapeutic serum concentrations. Nation and colleagues performed their PK analysis of 162 critically ill patients receiving colistin to determine achievable Css,avg in those receiving U.S.– or EMA–approved doses. The model demonstrated that for patients with CrCl of at least 80 mL/min, the highest Css,avg that could be achieved with sufficient reliability (90%) using the U.S. or EMA package inserts was 0.5 mg/L, regardless of the dosing regimen used. Additionally, the target Css,avg of 2 mg/L could be reliably achieved in only 20% to 30% of the time in this patient population. The dosing recommendations provided by Nation and colleagues do not perform much better, with roughly 40% of patients with a CrCl of at least 80 mL/min able to achieve a Css,avg of 2 mg/L. To put these figures into perspective, consider a patient with a CrCl of 130 mL/min in whom the goal target concentration is 2 mg/L. If the goal was to give a similar likelihood of target attainment that is achieved across the other dosing recommendations by Nation and colleagues, an enormously high dose of 560 mg of CBA would be required.
Table 2 lists the average serum concentrations that would be predicted, using the equation derived from Nation and colleagues, for each dosing recommendation at various CrCl levels. As the table suggests, similar exposures would be predicted using any of the equations for clearances 30 to 80 mL/min. However, when the clearance drops below 30 mL/min, the U.S. package insert, owing to greater-than-necessary dose reductions, significantly underperforms, with the Css,avg predicted to fall well below the “target” of 2 mg/L. This is demonstrated nicely in the publication by Nation and colleagues, who show that although Css,avg of 2 mg/L is predicted to be achieved in more than 80% of patients with CrCl less than 30 mL/min using the EMA package insert, only around 10% to 30% would achieve this same target using the U.S. package insert, depending on the weight of the patient. An additional point of consideration is that the estimations given in this review are based on an assumed body weight of 70 kg. However, there is no uniform recommendation for which dosing weight to use in patients, and the pharmacokinetic analysis by Nation and colleagues suggests there is no impact of weight on dosing. If weight-based dosing is used, with either ideal body weight (in men shorter than 69 in. or women shorter than 71 in.) or with actual body weight in patients who weigh less than 70 kg, the U.S. package insert will underperform. Conversely, if either total body weight or adjusted body weight is used in obese patients, there is potential for overexposure and an increased risk for toxicity.
Need for therapeutic drug monitoring
It has become increasingly obvious that no matter which dosing strategy is used, the variability in interpatient PK prevents the reliable attainment of predictable serum concentrations, particularly in patients with normal renal function. Given the narrow therapeutic index of colistin, the wide interpatient variability and the fact that the recommended dosing regimens predict average concentrations well into the toxic range, colistin is an agent that requires therapeutic drug monitoring (TDM). Currently, assays for colistin serum concentrations are not readily available outside of research laboratories, and there are logistical issues related to minimizing continued conversion from CMS to colistin after samples are obtained that would need to be addressed for TDM to become possible. However, based on the evidence, we are unlikely to readily (and safely) achieve therapeutic targets without TDM, and even with this technology probably cannot attain adequate levels in patients with normal renal function.
In conclusion, based on currently available evidence, any of the three dosing recommendations described here are reasonable for patient care, because minor differences between them would be outweighed by interpatient variability. One important exception is with the U.S. package insert dosing in cases in which a patient’s CrCl is less than 30 mL/min. If the U.S. package insert dosing is to be used, we recommend an alternate regimen of 1.25 mg/kg q12h for patients with CrCl ranging from 10 to 29 mL/min, and 1.5 mg/kg/day for those with a CrCl less than 10 mL/min. This dosing regimen will provide serum levels similar to those of the U.S. package insert in patients with normal renal function and, most importantly, more appropriate doses similar to those in the EMA package insert for patients with a CrCl less than 30 mL/min. Overall, the estimated exposures with this dosing regimen will be similar to those of the EMA– and PK–based regimens across the entire renal function spectra. For simplicity’s sake, however, it would make the most sense for clinicians to follow the EMA package insert recommendations.
- References:
- Cheah SE, et al. J Antimicrob Chemother. 2015;doi:10.1093/jac/dkv267.
- Colo-Mycin M [package insert]. Rochester, MI: Monarch Pharmaceuticals;2006.
- European Medicines Agency. Annex III: Amendments to relevant sections of the summary of product characteristics and the package leaflets. Available at: http://www.ema.europa.eu/docs/en_GB/document_library/Referrals_document/Polymyxin_31/WC500176332.pdf. 2014. Accessed April 19, 2017.
- Li J, et al. J Antimicrob Chemother. 2004;53:837-840.
- Nation RL, et al. Clin Infect Dis. 2017;doi:10.1093/cid/ciw839.
- Nation RL, et al. Clin Infect Dis. 2016;doi:10.1093/cid/civ964.
- Plachouras D, et al. Antimicrob Agents Chemother. 2009;doi:10.1128/AAC.01361-08.
- Sorlí L, et al. BMC Infect Dis. 2013;doi:10.1186/1471-2334-13-380.
- Zavascki AP, et al. Antimicrob Agents Chemother. 2017;doi:10.1128/AAC.02319-16.
- For more information:
- Jessica K. Ortwine, PharmD, is a clinical pharmacist in infectious diseases at Parkland Hospital and an assistant professor of medicine at the University of Texas Southwestern Medical Center in Dallas, Texas.
- Jason M. Pogue, PharmD, is a clinical pharmacist in infectious diseases at Sinai-Grace Hospital and an assistant professor of medicine at Wayne State University School of Medicine in Detroit, Michigan.
- Keith S. Kaye, MD, MPH, is a professor of medicine and the director of clinical research in the division of infectious diseases at the University of Michigan Medical School in Ann Arbor, Michigan, and an Infectious Disease News Editorial Board member.
- Donald Kaye, MD, MACP, is a professor of medicine at Drexel University College of Medicine, associate editor of the International Society of Infectious Diseases’ ProMED-mail, section editor of news for Clinical Infectious Diseases and an Infectious Disease News Editorial Board member.
Disclosures: Kaye, Kaye, Ortwine and Pogue report no relevant financial disclosures.