Influence of renal replacement therapy on pharmacokinetics in critically ill patients

Best Practice & Research Clinical Anaesthesiology Vol. 18, No. 1, pp. 175 –187, 2004 doi:10.1016/j.bpa.2003.09.002, available online at http://www.sciencedirect.com

11 Influence of renal replacement therapy on pharmacokinetics in critically ill patients Jan Frederik Bugge*

MD, PhD

Consultant in Cardiothoracic Anaesthesia and Intensive Care Department of Anaesthesia, Rikshospitalet, N-0027 Oslo, Norway

Critical illness has a great impact on many pharmacokinetic parameters. An increased volume of distribution often results in drug underdosing, whereas organ impairment may lead to drug accumulation and overdosing. Renal replacement therapy (RRT) in critically ill patients with renal failure may significantly increase drug clearance, requiring drug-dosing adjustments. Drugs significantly eliminated by the kidney are likely to experience substantial removal during RRT, and a supplemental dose—corresponding to the amount of drug removed by RRT—should be administered. Mechanisms of drug removal during RRT are reviewed together with methods for measuring or estimating RRT drug clearances. Approaches for drug-dosing adjustments are suggested and, at the end, the pharmacological principles for antibiotic prescription in the critically ill are discussed. Key words: acute renal failure; dialysis; drug dosing; haemofiltration; haemodiafiltration; intensive care unit; pharmacokinetics.

Critically ill patients often have multi-organ dysfunction, sepsis, or other conditions that require complex drug therapy and may influence drug concentrations through changes in absorption, distribution, metabolism and elimination. However, our knowledge about the influence of critical illness on drug pharmacokinetics is insufficient and drug dosing is frequently based on pharmacokinetic data obtained from healthy or less severely ill patients. The addition of renal replacement therapy (RRT) may further complicate drug therapy, and the principles of drug removal during RRT need to be clearly understood in order to avoid under- or overdosing in the intensive care setting. This chapter provides some principles and practical guidelines for more effective and appropriate drug dosing in the critically ill with acute renal failure (ARF) on RRT. INFLUENCE OF ARF AND CRITICAL ILLNESS ON PHARMACOKINETIC PARAMETERS IN RELATION TO DRUG ELIMINATION BY RRT The effective concentration of a drug is the free fraction at the site of action, and this concentration reflects a complicated interplay between dose, absorption, protein * Fax: þ 47-23073690. E-mail address: [email protected] (J.F. Bugge). 1521-6896/$ - see front matter Q 2003 Elsevier Ltd. All rights reserved.

176 J. F. Bugge

binding, volume of distribution (Vd), and clearance (metabolism and elimination). Critical illness and ARF may affect all of these pharmacokinetic parameters, and most of these drug properties influence drug clearance by RRT. Total body clearance of a drug is the sum of clearances from different sites in the body that may include hepatic, renal and other metabolic pathways. It is the contribution of renal clearance to total body clearance that is the major determinant of the need for dosing adjustments in renal failure. If the renal clearance of a drug is normally less than 25– 30% of total body clearance, impaired renal function is unlikely to have clinically significant influence on drug removal.1 Similarly, drug removal by RRT will have little influence on total body clearance and dosing adjustments do not have to be considered. Drugs significantly eliminated by the kidney are likely to experience substantial removal during RRT, and dosing adjustment is often required. Only the unbound fraction of a drug is available for filtration, and drugs with a high protein binding are poorly cleared by RRT. Many factors may alter the fraction of unbound drug—such as systemic pH, heparin therapy, hyperbilirubinaemia, concentration of free fatty acids, relative concentration of drug and protein, as well as the presence of uraemic products and other drugs that may act as competitive displacers.2 – 4 Most of our knowledge about alterations in protein binding and pharmacokinetics is derived from studies in chronic renal insufficiency. The influence of ARF on protein binding is not well described. Critically ill patients often have low albumin levels which may increase the unbound fraction of many drugs with possible deleterious effects, as documented for phenytoin.5 These patients also often have increased levels of acid a1-glycoprotein (an acute-phase protein) which may increase protein binding of some drugs. Thus, the reported unbound fraction in healthy volunteers and in patients with chronic renal insufficiency may differ substantially from the unbound fraction of drugs in critically ill patients on RRT. The Vd is a mathematical reflection of the volume in which a drug would need to be dissolved to obtain the observed blood concentration, assuming homogenous mixing in the body. A large Vd reflects a drug that is highly tissue bound, and consequently only a small proportion actually resides in the vascular compartment available for clearance by endogenous or extracorporeal routes. The overall response to critical illness includes increased capillary permeability, fluid shifts, and third space losses resulting in large extravascular, interstitial fluid accumulation6, and these changes may increase the Vd of many drugs used in the intensive care setting. Thus, the actual Vd in critically ill patients may differ from values obtained from pharmacological tables, and it shows great interand intra-individual variations.7 A drug with a small Vd (# 1 l kg21) is more likely to be cleared by extracorporeal therapies than a drug with a large Vd ($ 2 l kg21). However, there is a significant difference between intermittent haemodialysis (IHD) and continuous RRT (CRRT). A drug with a large Vd and high clearance during high-flux IHD will rapidly be removed from plasma, but only a small amount of the body’s drug content is removed during one dialysis session, and plasma concentration will be restored between therapies due to re-distribution. CRRT, by its continuous and slower action, has much less influence on the plasma concentration of drugs with large Vd because there is time for continuous re-distribution of the drug from the tissues to the blood. Because of this re-distribution the total amount of drug removed will be greater with continuous than with intermittent RRT (supposing a similar daily extracorporeal urea clearance). Although drug elimination during CRRT is much slower for drugs with a large Vd, the same is true for endogenous (hepatic) elimination that has to clear the same Vd. As a consequence, drug-dosing adjustments to be made during CRRT are much more dependent on

Influence of renal replacement therapy on pharmacokinetics 177

the relative contribution of CRRT to total body clearance of the drug than on the drug’s Vd.8 Renal and hepatic impairment result in decreased drug clearance and increase the risk of drug accumulation and overdosing. Liver failure may increase the contribution of renal clearance and hence the contribution of RRT to total body clearance, making drug-dosing adjustments necessary during RRT for drugs that in healthy individuals mainly undergo hepatic degradation. Similarly, renal failure may alter extrarenal clearance of drugs that are normally excreted by the kidney, and the critically ill patient with severe ARF may have a residual clearance that is often remarkable.9 Thus, drug dosing in the critically ill must take into account the opposing effects of increased Vd (and increased clearance in the early hyperdynamic phase of sepsis) resulting in low serum drug concentrations versus organ dysfunction resulting in high serum drug concentrations. In addition, the intensivist must be aware of the factors influencing the free fractions of drugs and the contribution of RRT to drug elimination.

MECHANISMS OF DRUG REMOVAL DURING RRT Essential to rational drug prescribing in patients undergoing RRT is an understanding of the different methods of solute removal that occur with the various types of treatment. Diffusion, convection and adsorption are the three mechanisms used for solute removal during RRT. The movement of a solute through a membrane from an area of high concentration to an area of lower concentration is called diffusion. Diffusion is the primary method of solute removal during dialysis. Membranes used for conventional IHD are less water permeable and have smaller pores, compared with those used for continuous venovenous haemofiltration (CVVH) or haemodiafiltration (CVVHDF). Convection is the removal of solutes along with the solvent in which they are present, and the rate of ultrafiltration determines the convective clearance of a solute during CRRT. Convection is not influenced by the solute concentration gradient across the membrane and is dependent only on membrane pore size. The diffusivity of a solute is inversely proportional to its molecular weight, and the relative importance of convection increases with increasing molecular weight. Diffusion favours clearance of small solutes whereas middle- and large-sized molecules are removed mainly by convection. Most drugs have a molecular weight # 500 Da, and very few are greater than 1500 Da. Conventional dialysis membranes favour diffusive clearance of low-molecular-weight solutes below 500 Da, whereas the typical high-flux membranes used for CRRT have greater pores (20 – 30 000 Da), making no significant filtration barrier to unbound drugs. As an example: for the removal of the 1448 Da molecule, vancomycin, convection is quantitatively more important than diffusion, and this drug is not substantially removed during IHD.10 Adsorption to filter membranes leads to increased removal from plasma11, and the various filters have different absorptive capacity. Some filter membranes, such as the commonly used polyacrylonitrile, may adsorb a substantial amount of drug to its surface.12,13 However, adsorption is a saturable process, and the influence on drug removal will depend on the frequency of filter changes. In general, with filters lasting approximately 18– 24 hours, as in most CRRT procedures, adsorption probably has minor influence on drug removal. Information about the various filters’ adsorptive capacity for most drugs is lacking. In IHD adsorption of drugs to the filter surface may play a significant role in drug elimination. Filter adsorption is not accounted for in drugdosing guidelines.8,14

178 J. F. Bugge

CRRT is performed mainly as CVVH or CVVHDF. In CVVH solutes are removed only by convection, and drug removal is limited by the ultrafiltration rate. By adding dialysis to filtration, as in CVVHDF, solutes are removed by both convection and diffusion, increasing the removal of small molecules more than middle-sized and large molecules. It also gives the possibility for the two processes to interact in such a manner that solute removal is significantly less than what is expected if the individual components are simply added together. In a study by Brunet et al15, the clearance during CVVHDF of small molecules, such as urea and creatinine, was approximately the sum of the two clearances obtained when CVVHD and CVVH were performed separately. Even with the highest flow rates of 2.5 and 2.0 l/hour of dialysate and ultrafiltration, respectively, the differences between actual and predicted clearances from addition of the two procedures differed less than 10%, indicating minimal interaction between diffusion and convection. For b2-microglobulin, however, combining diffusion and convection did not increase clearance compared to convection alone, indicating that this small protein was removed by convection only. This means that the diffusive clearance of a drug during CVVHDF is difficult to predict and will depend on its molecular weight, blood and dialysate flow rates, and the membrane used. In CVVH and CVVHDF, the method by which replacement fluids are administered influences solute removal efficiency. In post-dilution procedures the replacement fluid is administered distal to the filter, and mass removal per unit volume of effluent fluid is relatively high. For small solutes the effluent concentration may be the same as in plasma water. However, the ultrafiltration rate is limited by the operating characteristics of the system, mainly the haematocrit and the blood flow rate. The upper limit of ultrafiltration rate is approximately 25 –30% of plasma flow rate.16 In predilution procedures, the replacement fluid is administered proximal to the filter, diluting the concentration of solutes in the blood, and thus reducing solute clearances approximately 15 –20%.15 This modest decrease in efficiency can be overcome by increasing the ultrafiltration rate, which, in this setting, is not restricted by haematocrit and blood flow.17

RRT DRUG CLEARANCE AND DOSING ADJUSTMENTS The critically ill patient with renal failure is at risk for drug accumulation and overdose, but also for underdosing that may be life-threatening, such as in the case of insufficient antibiotic treatment. As pointed out earlier, many pharmacokinetic parameters change substantially in the critically ill and our knowledge about these matters are sparse. Ideally, drug dosing should be monitored by measuring plasma drug concentrations to a much greater extent than is practised today, both to secure adequate dosing in the individual patient and to obtain pharmacokinetic knowledge leading to improved drug dosing in the future. An increased Vd and greater than expected elimination either from RRT or from residual endogenous clearance are probably the main reasons for underdosing, whereas underestimated organ dysfunction is the main cause of overdosing. When making drug-dosing adjustments during RRT our goal is to compensate for the drug clearance of RRT (CLRRT). This clearance can be measured as: CLRRT ¼ QE CE =CP where CE and CP are drug concentrations in effluent fluid and plasma, respectively. QE is the effluent flow rate, which is the sum of ultrafiltration flow rate (QUF ) and dialysate

Influence of renal replacement therapy on pharmacokinetics 179

flow rate (QD ). For most drugs, concentration measurements are not available and RRT clearances have to be estimated. The sieving coefficient (S) of a drug is the concentration in ultrafiltrate (CUF ) divided by the concentration in plasma: S ¼ CUF =CP The exact formula for the differences between CPin and correct. For readily filterable unbound drug in plasma and S drug, making

sieving coefficient is S ¼ 2CUF =ðCPin þ CPout Þ; but the CPout are small, making the above equation almost molecules, CUF approximates the concentration of can be estimated by the unbound fraction (fu ) of the

CLCRRT ¼ fu QUF or ClCRRT ¼ fu ðQUF þ QD Þ during CVVH or CVVHDF, respectively. The value of fu is retrieved from pharmacological tables, but as outlined above, the unbound fraction in the critically ill may differ from these values. However, apart from some exceptions and individual variations, Golper and Marx found that, for a large series of drugs frequently used in critically ill patients, the filtered fraction during CRRT correlated well with the unbound fraction in healthy subjects.18 If CRRT is performed in a pre-dilution mode, fu has to be corrected by the dilution factor ¼ plama flow rate/(plasma flow rate þ replacement fluid flow rate). The dialysate flow rate during CVVHDF is low (750 – 2500 ml/hour), allowing almost complete diffusive equilibrium to occur between dialysate and plasma concentrations for small molecules. For these small drugs, fu QD represents an acceptable estimate of diffusive clearance. However, it will always be overestimated, and increasingly overestimated with increasing molecular weight and dialysate flow rate. Vos, Vincent and colleagues19,20 evaluated drug removal during continuous arteriovenous haemodiafiltration (CAVHDF) and derived a mathematical expression of the limiting effect of a drug’s molecular weight on the diffusive mass transfer Kdrel ¼ Kd=Kdc ¼ ðMW=113Þ20:42 where Kd and Kdc are the diffusive mass transfer coefficients for the drug and creatinine, respectively. MW is the drug’s molecular weight and 113 is the molecular weight of creatinine. Incorporating this limiting factor in the clearance equation yields9: CLCRRT ¼ fu ðQUF þ QD Kdrel Þ Using this approach for antimicrobial agents during CVVH (QD ¼ 0), Joos et al21 found that the CCVH clearance deviated less than 15% from estimated, whereas total body clearance was underestimated in the range of 30%. This means that the actual non-CRRT clearance was greater than expected. Kroh et al9 found very good correlations between doses calculated from measured and estimated kinetic data during CVVH, but they did not mention how the non-CRRT clearance was estimated. Data are lacking for CVVHDF, but when using the Kdrel on CAVHDF clearances reported in the literature, Kroh et al9 found very good correlations between observed and estimated clearances (y ¼ 0:004 þ 0:96x). For non-toxic drugs, doses can safely be increased beyond estimates, and a 30% increase is recommended to ensure adequate dosing.7 During IHD, QD is much higher (30 l/hour) than during CVVHDF (750 – 2500 ml/hour) and diffusive equilibrium is never obtained, even for small molecules, making the diffusive saturation (SD ) substantially smaller than fu ; depending on both

180 J. F. Bugge

the molecular weight of the drug and QD : This makes drug clearance estimates difficult to perform. For toxic drugs, and for drugs with a narrow therapeutic range, drug monitoring with measurements of plasma concentrations is mandatory. This method addresses the actual total body clearance in which RRT is only one part, and adjustments can be made as total body clearance is changed by the clinical condition and/or by RRT adjustments. The following dosing formula is often used to achieve the desired peak concentration (CPEAK ) from the actual concentration (CACTUAL ): D ¼ ðCPEAK 2 CACTUAL ÞVd £ Body weight As outlined above, the estimated Vd in this equation may differ from the actual Vd, but with repeated measurements this error is self-correcting. Because of the phenomenon of re-distribution, at least 2 hours should elapse after each dialysis session before samples for drug monitoring are taken. Making estimates is time-consuming and requires a careful search for pharmacological data. Another approach to drug dosing during RRT is to utilize a readily available reference. In a publication by Kroh7, there is a table of normal kinetic data of many actual drugs and suggested dosing adjustment factors for different ultrafiltration rates during CVVH. If using references that utilize glomerular filtration rate (GFR) in its specific recommendations, such as the Bennett tables22, it is useful to regard CRRT as a GFR of 10 –50 ml/minute depending on dialysate and haemofiltration flow rates. However, these tables22 are based on measurements from patients with chronic renal insufficiency and do not take into account the pharmacokinetic changes induced by acute critical illness as discussed above.

CRITICAL ILLNESS AND PHARMACOLOGICAL PRINCIPLES FOR ANTIBIOTIC PRESCRIPTION The pharmacological principles outlined above will be illustrated by the example of antibiotic treatment. Antibiotics are chosen because most critically ill patients undergo antibiotic treatment, several antibiotics have toxic effects, and underdosing may have fatal consequences. The goal of antibiotic treatment is to achieve effective active drug concentrations that result in clinical cure while avoiding or minimalizing drugassociated toxicity.23,24 Understanding the relationship between the pharmacodynamic and the pharmacokinetic properties of antibiotics helps to determine the optimal dosage regimen and to predict which pharmacokinetic parameters should be taken into account when making dosing adjustments. Parameters to consider are peak plasma drug concentrations, the area under the plasma concentration curve (AUC), time above the minimum inhibitory concentration (MIC), and the area under the inhibitory curve (AUC/MIC). Pharmacodynamic properties of antibiotics differ and relate to the time course of drug activity and the mode of action on the microorganisms. Aminogycosides The bactericidal effect of the aminoglycosides is concentration-dependent.25,26 A high peak concentration provides better and faster bacterial killing and is associated with

Influence of renal replacement therapy on pharmacokinetics 181

better clinical response.27 Aminoglycosides exhibit a significant post-antibiotic effect (PAE)25 – 28, which refers to continued suppression of bacterial growth despite no measurable concentration of the antibiotic. The duration of this effect depends on the preceding peak concentration28, and both the requirement for high peak concentrations and the PAE argue for giving the daily dose of aminoglycosides as one single dose. A single daily dose produces better clinical outcome29 – 31 and is also associated with less oto- and nephrotoxicity.32,33 Toxicity is closely related to high trough values, or more precisely the AUC, than to high peaks. The Vd of aminoglycosides is normally 0.2 – 0.3 l kg21, but is increased in critical illness, giving lower peak values than expected from a given dose. Elimination is dependent on glomerular filtration, and plasma half-life increases in accordance with the reduction in GFR. Dose adjustments have to be performed during ARF, and increasing the dosing interval according to the reduction in renal function without reducing each single dose seems to be a reasonable approach, and is supported by clinical investigations.34 RRT has a significant impact on aminoglycoside elimination. The protein binding is low and the sieving coefficient is approximately 0.9 for both diffusion and convection, making CLRRT ¼ 0:9QE : This clearance gives some guidance in estimating the dosing interval, but given the toxicity and the variability in the Vd changes induced by critical illness, monitoring plasma concentrations during aminoglycoside therapy is still mandatory to secure adequate and safe dosing. However, the optimal point of time in the dosing interval to make these measurements remains a matter for discussion. b-Lactam antibiotics Bacterial killing by b-lactam antibiotics is slow and continuous, and is almost entirely related to the time the drug levels in tissue and plasma exceed the MIC. Except for the carbapenems, which exibit some PAE35, other b-lactam antibiotics lack any clinically relevant PAE and bacterial growth resumes as soon as the concentration of the antibiotic is too low.28,36 The best correlate for therapeutic efficacy is the percentage of the dosing interval for which drug concentrations remain above MIC37, whereas the development of resistance seems to be related to the percentage of the dosing interval with levels below MIC.38 Maximal killing effects occur at levels four to five times MIC with no additional effect of higher levels. The logical conclusion would be to keep the concentration at four to five times MIC for long periods by more frequent dosing or by continuous infusion of the antibiotic.25,39 – 41 Elimination of b-lactam antibiotics is related to GFR.42 – 46 Critically ill patients in the early hyperdynamic phase of sepsis often have increased GFR, and standard dosing regimens may lead to underdosing in this phase of the disease where adequate antibiotic treatment is essential for the outcome.44 In ARF the dose has to be reduced and, from the discussions above, it is logical to reduce the individual dose and not extend the dosing interval. RRT increases the elimination of most b-lactam antibiotics depending on their S values, which reflect their levels of protein binding and which differs substantially. During CRRT, ceftriaxone has limited removal while ampicillin, ceftazidime, and meropenem have clearances close to QE : The situation may become a little bit more complex for drugs containing two compounds such as imipenem and cilastatin. Renal failure results in more accumulation of cilastatin than imipenem and this dysequilibrium is attenuated by RRT. In other words, the renal contribution to total body clearance is greater for cilastatin, resulting in a greater impact of renal failure on the drug’s level. However, the S is also greater,

182 J. F. Bugge

resulting in more elimination of cilastatin than of imipenem.47,48 Meropenem is therefore the preferred carbapenem in RRT patients. Similarly, tazobactam accumulates compared to piperacillin when using a fixed piperacillin – tazobactam combination, but because of low toxicity the fixed combination can be used in most critically ill patients on RRT.49 Compared to aminoglycosides, b-lactam antibiotics have much less toxicity and when using estimated RRT clearances to perform drug-dosing adjustments, doses can safely be increased 30% above estimates to secure adequate dosing.7 Fluoroquinolones Bacterial killing by fluoroquinolones is both concentration- and time-dependent50,51, and these antibiotics exhibit PAE.50 – 52 A combination of peak concentrations and time above MIC gives the best correlation of ciprofloxacin efficiency in experimental settings and can be expressed in the AUC/MIC ratio. A ratio above 125 for Gram-negative and above 30 for Gram-positive bacteria correlates with better clinical outcome.53,54 The Vd for ciprofloxacin does not significantly change in the critically ill.55 Renal elimination normally amounts to 60% of total body clearance56, and ciprofloxacin dosing has to be reduced in ARF. Both individual dose reductions and/or extension of the dosing interval seem to be reasonable approaches.55,57 RRT significantly increases ciprofloxacin elimination depending on the RRT mode used, and elimination is less with standard lowflux IHD than with CRRT. Wallis et al55 found an S value of 0.7 during CCVHDF. This value corresponds with the unbound fraction in plasma21,58 and can be used in estimating CRRT clearances (CLCRRT ¼ 0:7QE ). Interestingly, the study of Wallis et al55 showed greater day-to-day variation in total body clearance (36%) than in CRRT clearance (20%), indicating that critically ill patients seldom achieve a condition of steady state. Levofloxacin elimination is nearly completely dependent on intact renal routes of excretion, and dosage has to be substantially reduced in renal failure.59 Malone et al60 found S values of 0.67 and 0.56 during CVVH and CVVHDF, respectively. At ultrafiltration rates between 15 and 20 ml minute l21 CVVH clearance was approximately 26% of total body clearance. By adding a dialysate flow rate of 1 l/hour (CVVHDF) extracorporeal clearance increased to 40% of total body clearance. A loading dose of 500 mg followed by 250 mg every 48 hours is recommended by the manufacturer for patients with a creatinine clearance less than 20 ml/minute. This dosing regimen is also confirmed to provide adequate AUC/MIC ratios in patients with end-stage renal disease on IHD.61 In the study of Malone et al60 500 mg followed by 250 mg every 24 hours provided adequate AUC/MIC ratios during CVVH and CVVHDF, but this dosing regimen will probably have to be further adjusted during more aggressive haemofiltration and/or dialysis. Vancomycin Vancomycin does not exhibit concentration-dependent killing, and exceeding the MIC more than four- to fivefold does not result in increased activity.62 – 64 Recent data suggest that optimal vancomycin therapy may be achieved by sustained therapeutic concentrations without high peaks62,65 and vancomycin therapy given as continuous infusion demonstrates comparable efficacy and tolerance.65 A target concentration of 20 –25 mg l21 seems to be a reasonable goal.65 Vancomycin is poorly protein-bound and distributes into the extravascular space. The Vd is increased in the critically ill, and higher doses are required to achieve

Influence of renal replacement therapy on pharmacokinetics 183

appropriate concentrations in these patients. Vancomycin is almost entirely eliminated by glomerular filtration, and in ARF the dose has to be reduced in proportion to the reduction in GFR. The 1448 Da molecule vancomycin is better eliminated by convection than by diffusion. It is not removed by standard low-flux IHD membranes, but passes readily through the high-flux membranes used in CRRT with an S value of approximately 0.7 for convective clearance. Diffusive clearance is 30 –40% less.66 Fluconazole Fluconazole is a well-tolerated drug for treatment of serious infections caused by Candida species with a dosage up to 12 mg kg21 day21 given as a single dose. Most Candida species are inhibited at concentrations above 6 mg l21 and for Candida albicans the MIC is 1 – 3 mg l21.67,68 Candida crusei, and often also Candida glabrata, is resistant.69 Protein binding of fluconazole is approximately 12%70, but this increases in patients with chronic renal failure.71 Protein binding in the critically ill with ARF is unknown. Vd of fluconazole is 0.8 l kg21 BW corresponding to total body water. In critically ill patients with ARF, Muhl et al72 found a Vd ranging from 0.65 to 1.28 l kg21 BW. The kidney eliminates 80% of the drug and this elimination is proportional to GFR.73 The low molecular weight (306 Da) and low protein binding of fluconazole indicate high extracorporeal diffusive and convective clearances. Muhl et al72 found S values of 0.96 and 0.88 during CVVH and CCVHDF, respectively. With a QE of 2 l during CVVHDF, 70% of the administered dose was eliminated by the extracorporeal circuit in 24 hours. Total body clearance during CVVHDF was approximately twice the clearance observed in patients with normal renal function, and some patients experienced low trough values.72 The higher extracorporeal compared with renal clearances are related to an important tubular re-absorption of fluconazole in the normal kidney. Serious fungal infections have high mortality, and underdosing increases the risk of therapeutic failure. Fluconazole seems to have a wide therapeutic range74,75, and doses should be increased above standard during CRRT. Depending on CRRT mode and QE ; the dose might be increased up to 1000 –1600 mg a day.

SUMMARY Critical illness and ARF change many pharmacokinetic parameters. Increased Vd and increased drug elimination are the main reasons to underdosing, whereas organ dysfunction might lead to drug accumulation and overdosing. Drugs significantly eliminated by the kidney are likely to experience substantial removal during RRT, and a supplemental dose corresponding to the amount of drug removed by RRT should be administered. Clearance by RRT can either be measured or estimated. The high-flux membranes used in CRRT make no filtration barrier to most drugs, and the filtrate concentration can be estimated by the unbound fraction of the drug in plasma. When adding dialysis to filtration, this approach overestimates drug clearance, and a correcting factor should be used. For non-toxic drugs, doses can safely be increased 30% above estimates to ensure adequate dosing. For drugs with a narrow therapeutic margin, monitoring of plasma concentrations is mandatory. Understanding the relationship between pharmacodynamic and pharmacokinetic properties of antibiotics helps to determine the optimal dosage regimen and the pharmacokinetic parameters

184 J. F. Bugge

that should guide dosing adjustments. For some antibiotics CRRT elimination may be surprisingly high.

Practice points † the unbound fraction of a drug in plasma can be used to estimate the sieving coefficient during CRRT † effluent flow rate has a great impact on drug elimination during CRRT † drugs normally eliminated by the kidney are likely to experience substantial elimination by RRT † the Vd of many drugs is increased in the critically ill and this may lead to underdosing † toxic drugs and drugs with a narrow therapeutic margin should be monitored by measuring plasma drug concentrations

Research agenda † pharmacokinetic studies of drugs in the critically ill † drug elimination during CRRT under different modes and various volumes † drug interactions in the critically ill

REFERENCES 1. Levy G. Pharmacokinetics in renal disease. American Journal of Medicine 1977; 62: 461–463. 2. Suh B, Craig WA, England AC & Elliott RL. Effect of free fatty acids on protein binding of antimicrobial agents. Journal of Infectious Diseases 1981; 143: 609–610. 3. Gulyassy PE & Depner TA. Impaired binding of drugs and endogenous ligands in renal disease. American Journal of Kidney Diseases 1983; 2: 578–601. 4. McNamara PJ, Lalka D & Gibaldi M. Endogenous accumulation products and serum protein binding in uremia. Journal of Laboratory and Clinical Medicine 1981; 98: 730–740. 5. Driscoll DF, McMahon M, Blackburn GL & Bistrain BR. Phenytoin toxicity in a critically ill, hypoalbuminemic patient with normal serum drug concentrations. Critical Care Medicine 1988; 16: 1248–1249. 6. Gosling P, Sanghera K & Dickson G. Generalized vascular permeability and pulmonary function in patients following serious trauma. Journal of Trauma 1994; 36: 477– 481. * 7. Kroh UF. Drug administration in critically ill patients with acute renal failure. New Horizons 1995; 3: 748 –759. * 8. Bohler J, Donauer J & Keller F. Pharmacokinetic principles during continuous renal replacement therapy: drugs and dosage. Kidney International 1999; 56(supplement 72): S24–S28. * 9. Kroh UF, Holl TJ & Steinhausser W. Management of drug dosing in continuous renal replacement therapy. Seminars in Dialysis 1996; 9: 161–165. 10. Jeffrey RF, Khan AA, Prabhu P et al. A comparison of molecular clearance rates during continuous hemofiltration and haemodialysis with a novel volumetric continuous renal replacement system. Artificial Organs 1994; 18: 425–428. 11. Rumpf KW, Rieger J, Ansorg R et al. Binding of antibiotics by dialysis membranes and its clinical relevance. Proceedings of the European Dialysis and Transplant Association 1977; 14: 677–688. 12. Kronofol NO, Lau AH & Barakat MM. Aminoglycoside binding to polyacrylonitrile hemofilter during continuous hemofiltration. Transactions of the American Society of Artficial Internal Organs 1987; 33: 300–303. 13. Oborne CA, Buckton G & Barker N. The interaction of albumine and drugs with two hemofiltration membranes. Journal of Clinical Pharmacology and Therapeutics 1994; 19: 119 –125.

Influence of renal replacement therapy on pharmacokinetics 185 * 14. Joy MS, Matzke GR, Armstrong DK et al. A primer on continuous renal replacement therapy for critically ill patients. Annals of Pharmacotherapy 1998; 32: 362– 375. * 15. Brunet S, Leblanc M, Geadah D et al. Diffusive and convective solute clearances during continuous renal replacement therapy at various dialysate and ultrafiltration flow rates. American Journal of Kidney Diseases 1999; 34: 486 –492. 16. Kaplan AA. Continuous arteriovenous hemofiltration and related therapies. In Winchester J (ed.) Replacement of Renal Function by Dialysis. Dordrecht: Kluwer Academic, 1996, pp 392–400. 17. Clark WR & Ronco C. CRRT efficiency and efficacy in relation to solute size. Kidney International 1999; 56(supplement 72): S3– S7. * 18. Golper TA & Marx MA. Drug dosing adjustments during continuous renal replacement therapies. Kidney International 1998; 53(supplement 66): S165–S168. * 19 Vos MC & Vincent HH. Continuous arteriovenous hemodiafiltration: Predicting the clearance of drugs. Contributions to Nephrology 1991; 93: 143 –145. * 20. Vincent HH, Vos MC, Akcahuseyin E et al. Drug clearance by continuous haemodiafiltration (CAVHD). Analysis of sieving coefficients and mass transfer coefficients of diffusion. Blood Purification 1993; 11: 99 –107. * 21. Joos B, Scmidli M & Keusch G. Pharmacokinetics of antimicrobial agents in anuric patients during continuous venovenous hemofiltration. Nephrology Dialysis Transplantation 1996; 11: 1582–1585. * 22. Aronoff GR, Berns JS, Brier ME, et al. Drug Prescribing in Renal Failure: Dosing Guidelines for Adults. Philadelphia: American College of Physicians, 1999. 23. Lipman J, Crewe-Brown HH, Saunders GL & Gaes AG. Subtleties of antibiotic dosing—do doses and intervals make a difference in the critically ill? South African Journal of Surgery 1996; 34: 160 –162. 24. Fry DE. The importance of antibiotic pharmacokinetics in critical illness. American Journal of Surgery 1996; 172(supplement 6A): 20S–25S. 25. Roosendaal R, Bakker-Woudenberg IA, van den Berghe-van Raffe M et al. Impact of the dosage schedule on the efficacy of ceftazidime, gentamicin and ciprofloxacin in Klebsiella pneumoniae pneumonia and septicaemia in leukopenic rats. European Journal of Clinical Microbiology and Infectious Diseases 1989; 8: 878–887. 26. Bakker-Woudenberg IA & Roosendaal R. Impact of dosage schedule of antibiotics on the treatment of serious infections. Intensive Care Medicine 1990; 16(supplement 3): S229–S234. 27. Moore RD, Lietman RS & Smith CR. Clinical response to aminoglycoside therapy: importance of the ratio of peak concentration to minimal inhibitory concentration. Journal of Infectious Diseases 1987; 155: 93–99. 28. Vogelman B, Gudmundsson S, Legett J et al. Correlation of antimicrobial pharmacokinetic parameters with therapeutic efficacy in an animal model. Journal of Infectious Diseases 1988; 158: 831 –847. 29. Marik PE, Havlik I, Monteagudo FS & Lipman J. The pharmacokinetic of amikacin in critically ill adult and paediatric patients: comparison of once- versus twice-daily dosing regimens. Journal of Antimicrobial Chemotherapy 1991; 27(supplement C): 81–89. 30. van der Auwera P, Meunier F, Ibrahim S et al. Pharmacodynamic parameters and toxicity of netilmicin (6 milligrams/kilogram/day) given once daily or in three divided doses to cancer patients with urinary tract infection. Antimicrobial Agents and Chemotherapy 1991; 35: 640–647. 31. Prins JM, Buller HR, Kuijper EJ et al. Once versus thrice daily gentamicin in patients with serious infections. Lancet 1993; 341: 335 –339. 32. Tran Ba Huy P & Deffrennes D. Aminoglycoside ototoxicity: influence of dosage regimen on drug uptake and correlation between membrane binding and some clinical features. Acta Otolaryngologica 1988; 105: 511–515. 33. Zhanel GG & Ariano RE. Once daily aminoglycoside dosing: maintained efficacy with reduced nephrotoxicity? Renal Failure 1992; 14: 1–9. 34. Nicolau DP, Freeman CD, Belliveau PP et al. Experience with a once-daily aminoglycoside program administered to 2,184 adult patients. Antimicrobial Agents and Chemotherapy 1995; 39: 650– 655. 35. Mouton JW, Touzw DJ, Horrevorts AM & Vinks AA. Comparative pharmacokinetics of the carbapenems: clinical implications. Clinical Pharmacokinetics 2000; 39: 185– 201. 36. Mouton JW & den Hollander JG. Killing of Pseudomonas aeruginosa during continuous and intermittent infusion of ceftazidime in an in vitro pharmacokinetic model. Antimicrobial Agents and Chemotherapy 1994; 38: 931–936. 37. Turnidge JD. The pharmacodynamics of beta-lactams. Clinical Infectious Diseases 1998; 27: 10 –22. 38. Fantin B, Farinotti R, Thabaut A & Carbon C. Conditions for the emergence of resistance to cefpirome and ceftazidime in experimental endocarditis due to Pseudomanas aeruginosa. Journal of Antimicrobial Chemotherapy 1994; 33: 563–569. 39. Benko AS, Capelletty DM, Kruse JA & Rybak MJ. Continuous infusion of versus intermittent administration of ceftazidime in critically ill patients with suspected Gram-negative infections. Antimicrobial Agents and Chemotherapy 1996; 40: 691 –695.

186 J. F. Bugge 40. Mouton JW, Vinks AA & Punt NC. Pharmacokinetic-pharmacodynamic modeling of activity of ceftazidime during continuous and intermittent infusion. Antimicrobial Agents and Chemotherapy 1997; 41: 733–738. 41. MacGowan AP & Bowker KE. Continuous infusion of beta-lactam antibiotics. Clinical Pharmacokinetics 1998; 35: 391–402. 42. Lipman J, Wallis SC & Richard C. Low plasma cefepime levels in critically ill septic patients: pharmacokinetic modeling indicates improved troughs with revised dosing. Antimicrobial Agents and Chemotherapy 1999; 43: 2559–2561. 43. Lipman J, Wallis SC, Richard CM & Fraenkl D. Low cefpirome levels during twice daily dosing in critically ill septic patients: pharmacokinetic modeling calls for more frequent dosing. Intensive Care Medicine 2001; 27: 363–370. 44. Pinder M, Bellomo R & Lipman J. Pharmacological principles of antibiotic prescription in the critically ill. Anaesthesia and Intensive Care 2002; 30: 134– 144. 45. Christensson BA, Nilsson-Ehle M, Hutchinson M et al. Pharmacokinetics of meropenem in subjects with various degree of renal impairment. Antimicrobial Agents and Chemotherapy 1992; 36: 1532– 1537. 46. Chimata M, Nagase M, Suzuki Y et al. Pharmacokinetics of meropenem in patients with various degrees of renal function, including patients with end-stage renal disease. Antimicrobial Agents and Chemotherapy 1993; 37: 229–233. 47. Przechera M, Bengel D & Risler T. Pharmacokinetics of imipenem/cilastatin during continuous arteriovenous hemofiltration. Contributions to Nephrology 1991; 93: 131–134. 48. Vos MC, Vincent HH & Yzerman EPF. Clearance of imipenem/cilastatin in acute renal failure patients treated by continuous hemodiafiltration (CAVHD). Intensive Care Medicine 1992; 18: 282–285. 49. Mueller SC, Majcher-Peszynska J, Hickstein H et al. Pharmacokinetics of piperacillin-tazobactam in anuric intensive care patients during continuous venovenous haemodialysis. Antimicrobial Agents and Chemotherapy 2002; 46: 1557–1560. 50. Campoli-Richards DM, Monk JP, Price A et al. Ciprofloxacin. A review of its antibacterial activity, pharmacokinetic properties and therapeutic use. Drugs 1988; 35: 373 –447. 51. Bauernfeind A. Questioning dosing regimens of ciprofloxacin. Journal of Antimicrobial Chemotherapy 1993; 31: 789–798. 52. Chin NX & Neu HC. Post-antibiotic suppressive effect of ciprofloxacin against gram-positive and gramnegative bacteria. American Journal of Medicine 1987; 82: 58 –62. 53. Forrest A, Nix DE, Ballow CH et al. Pharmacodynamics of intravenous ciprofloxacin in seriously ill patients. Antimicrobial Agents and Chemotherapy 1993; 37: 1073–1081. 54. Nix DE, Spivey JM, Norman A & Schentag JJ. Dose-ranging pharmacokinetic study of ciprofloxacin after 200, 300 and 400 mg intravenous doses. Annals of Pharmacotherapy 1992; 26: 8– 10. 55. Wallis SC, Mullany DV, Lipman J et al. Pharmacokinetics of ciprofloxacin in ICU patients on continous veno-venous haemodiafiltration. Intensive Care medicine 2001; 27: 665–672. 56. Vance-Bryan K, Guay DR & Rotschafer JC. Clinical pharmacokinetics of ciprofloxacin. Clinical Pharmacokinetics 1990; 19: 434–461. 57. Bellmann R, Egger P, Gritsch W et al. Pharmacokinetics of ciprofloxacin in patients with acute renal failure undergoing continuous venovenous haemofiltration: influence of concomitant liver cirrhosis. Acta Medica Austriaca 2002; 29: 112–116. 58. Ho¨ffken G, Lode H, Prinzing C et al. Pharmacokinetics of ciprofloxacin after oral and parenteral administration. Antimicrobial Agents and Chemotherapy 1985; 27: 375 –379. 59. Fish DN & Chow AT. Clinical pharmacokinetics of levofloxacin. Clinical Pharmacokinetics 1992; 32: 101–119. 60. Malone RS, Fish DN, Abraham E & Teitelbaum I. Pharmacokinetics of levofloxacin and ciprofloxacin during continuous renal replacement therapy in critically ill patients. Antimicrobial Agents and Chemotherapy 2001; 45: 2949–2954. 61. Sowinski KM, Lucksiri A, Kays MB et al. Levofloxacin pharmacokinetics in ESRD and removal by the cellulose acetate high performance-210 hemodialyzer. American Journal of Kidney Diseases 2003; 42: 342–349. 62. Saunders NJ. Vancomycin administration and monitoring reappraisal. Journal of Antimicrobial Chemotherapy 1995; 36: 279–282. 63. Larsson AJ, Walker KJ, Raddatz JK & Rotschafer JC. The concentration-independent effect of monoexponential and biexponential decay in vancomycin concentrations on the killing of Staphylococcus aureus under aerobic and anaerobic conditions. Journal of Antimicrobal Chemotherapy 1996; 38: 589– 597. 64. Wakanabe T, Ohashi K, Matsui K & Kubata T. Comparative studies of bactericidal, morphological and post-antibiotic effects of arbekacin and vancomycin against methicillin-resistant Staphylococcus auereus. Journal of Antimicrobial Chemotherapy 1997; 39: 471 –476.

Influence of renal replacement therapy on pharmacokinetics 187 65. Wysocki M, Delatour F, Faurisson F et al. Continuous versus intermittent infusion of vancomycin in severe staphylococcal infections: prospective multicenter randomized study. Antimicrobial Agents and Chemotherapy 2001; 45: 2460–2467. 66. Reeves JH & Butt WW. A comparison of solute clearance during continuous hemofiltration, hemodiafiltration and haemodialysis using a polysulfone hemofilter. Asaio Journal 1995; 41: 100– 104. 67. Hughes CE, Bennett RY, Tuna IC & Beggs WH. Activities of fluconazole and ketoconazole against ketoconazole susceptible and resistant Candida albicans. Antimicrobial Agents and Chemotherapy 1988; 32: 209–212. 68. Rogers TE & Galgiani J. Activity of fluconazole and ketoconazole against Candida albicans in vitro and in vivo. Antimicrobial Agents and Chemotherapy 1986; 30: 418–422. 69. Fisher MA, Shen HA, Haddad J & Tarry WF. Comparison of in vivo activity of fluconazole with that of amphotericin B against Candida tropicalis, Candida glabrata and Candida crusei. Antimicrobial Agents and Chemotherapy 1989; 33: 1443–1446. 70. Debruyine D. Clinical pharmacokinetics of fluconazole in superficial and systemic mycoses. Clinical Pharmacokinetics 1997; 33: 52 –77. 71. Arredondo G, Martinez-Jorda R, Calvo R et al. Protein binding of itraconazole and fluconazole in patients with chronic renal failure. International Journal of Clinical Pharmacology and Therapeutics 1994; 32: 361– 364. 72. Muhl E, Martens T, Iven H et al. Influence of continuous veno-venous haemodiafiltration and continuous veno-venous haemofiltration on the pharmacokinetics of fluconazole. European Journal of Clinical Pharmacology 2000; 56: 671– 678. 73. Grant SM & Clissold SP. Fluconazole—a review of its pharmacodynamic and pharmacokinetic properties and therapeutic potential in superficial and systemic mycoses. Drugs 1990; 39: 877–916. 74. Diaz M, Negroni R, Montero-Gei F et al. A Pan-American 5-year study of fluconazole therapy for deep mycoses in the immunocompetent host. Clinical Infectious Diseases 1992; 14(supplement 1): 68–76. 75. Graninger W, Presteril E, Schneeweiss B et al. Treatment of Candida albicans fungemia with fluconazole. Journal of Infection 1993; 26: 133 –146.