Dosing of Gentamicin in Patients With End-Stage Renal Disease Receiving Hemodialysis

http://www.jclinpharm.orgPharmacology

The Journal of Clinical

DOI: 10.1177/0091270006292987 2006; 46; 1259 J. Clin. Pharmacol.

Mette Maja B. Teigen, Stephen Duffull, Lily Dang and David W. Johnson Dosing of Gentamicin in Patients With End-Stage Renal Disease Receiving Hemodialysis

http://www.jclinpharm.org/cgi/content/abstract/46/11/1259 The online version of this article can be found at:

Published by:

http://www.sagepublications.com

On behalf of: American College of Clinical Pharmacology

can be found at:The Journal of Clinical Pharmacology Additional services and information for

http://www.jclinpharm.org/cgi/alerts Email Alerts:

http://www.jclinpharm.org/subscriptions Subscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://www.jclinpharm.org/cgi/content/abstract/46/11/1259#BIBLSAGE Journals Online and HighWire Press platforms):

(this article cites 34 articles hosted on the Citations

© 2006 American College of Clinical Pharmacology. All rights reserved. Not for commercial use or unauthorized distribution. at UNIV LIBRARY AT IUPUI on April 16, 2007 http://www.jclinpharm.orgDownloaded from

http://www.accp1.org

http://www.jclinpharm.org/cgi/alerts

http://www.jclinpharm.org/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://www.jclinpharm.org/cgi/content/abstract/46/11/1259#BIBL

http://www.jclinpharm.org

J Clin Pharmacol 2006;46:1259-1267 1259

The aim of this study was to evaluate dosing schedules ofgentamicin in patients with end-stage renal disease andreceiving hemodialysis. Forty-six patients were recruitedwho received gentamicin while on hemodialysis. Eachpatient provided approximately 4 blood samples at vari-ous times before and after dialysis for analysis of plasmagentamicin concentrations. A population pharmacokineticmodel was constructed using NONMEM (version 5). Theclearance of gentamicin during dialysis was 4.69 L/h andbetween dialysis was 0.453L/h. The clearance betweendialysis was best described by residual creatinine clearance(as calculated using the Cockcroft and Gault equation),

which probably reflects both lean mass and residual clear-ance mechanisms. Simulation from the final populationmodel showed that predialysis dosing has a higher proba-bility of achieving target maximum concentration (Cmax)concentrations (>8 mg/L) within acceptable exposure limits(area under the concentration-time curve [AUC] values >70and <120 mg·h/L per 24 hours) than postdialysis dosing.

Keywords: Aminoglycosides; hemodialysis; end-stagerenal disease; pharmacokinetics

Journal of Clinical Pharmacology, 2006;46:1259-1267©2006 the American College of Clinical Pharmacology

Dosing of Gentamicin in Patients WithEnd-Stage Renal Disease Receiving

Hemodialysis

Mette Maja B. Teigen, BPharm, MClinPharm, Stephen Duffull, MPharm, PhD,Lily Dang, BPharm(Hons), and David W. Johnson, MBBS, FRACP, PhD

The mortality rate among patients with end-stagerenal disease (ESRD) in the Western world is high

(16%-22%),1,2 with infection being the second mostcommon cause of death in this patient group. Sep-ticemia accounts for approximately 75% of this infec-tion rate in the United States, and mortality fromsepsis is higher when compared to the general popu-lation, even when stratifying for other risk factors.3

Surprisingly, after more than 50 years of clinicalexperience with aminoglycosides, many aspects withregard to their use remain unresolved. When pre-scribing aminoglycoside antibiotics to patients withESRD, it appears to be common practice to administerthe drug after the patient has received hemodialysis.

Emerging data and theoretical considerations suggestthis may not be the most beneficial manner of admin-istration.4 Over recent years, a more complete under-standing of the pharmacokinetic-pharmacodynamicrelationships of aminoglycosides has led to a globalchange in clinical practice from multiple to once-dailydosing. Relevant pharmacokinetic-pharmacodynamicaspects of aminoglycosides include bactericidal activ-ity that is linked to Cmax (efficacy improves withhigher peak concentrations),5-8 adaptive resistance thatis reversed by an adequate interval of low drug con-centration between doses,9-12 and exposure-dependenttoxicity.13-16

Most once-daily dosing regimens are based onachieving an equivalent exposure (area under theconcentration-time curve [AUC]) to that which wouldhave been achieved with divided doses; thus, it issuggested that the risk of toxicity should be no higherwhen the dose is administered once daily.17,18 Thepharmacokinetic-pharmacodynamic behavior ofaminoglycosides that led to longer dose intervals withhigher Cmax values also strongly supports adminis-tration of these agents before hemodialysis. Theoverall pharmacokinetic-pharmacodynamic goal

From the School of Pharmacy, University of Queensland, Brisbane, Australia(Ms Teigen, Dr Duffull, Ms Dang); Diakonhjemmet Hospital Pharmacy,Oslo, Norway (Ms Teigen); the Department of Renal Medicine, Universityof Queensland at Princess Alexandra Hospital, Brisbane, Australia(Dr Johnson); and the School of Pharmacy, University of Otago, Dunedin,New Zealand (Dr Duffull). Submitted for publication February 19, 2006;revised version accepted July 13, 2006. Address for correspondence:Stephen Duffull, MPharm, PhD, School of Pharmacy, University of Otago,Dunedin, New Zealand; e-mail: [email protected]: 10.1177/0091270006292987



1260 • J Clin Pharmacol 2006;46:1259-1267

when prescribing aminoglycosides in hemodialysispatients is to achieve a concentration-time profilethat approaches the profile seen after once-daily dos-ing in patients with normal renal function. In doingthis, it is believed that the benefits of aminoglyco-sides when used in patients with normal renal func-tion can be conferred to those receiving hemodialysis.It is hypothesized that dosing of aminoglycosidesprior to, rather than after, hemodialysis will allow fora higher dose to be administered for any given levelof exposure and hence provide higher peak concen-trations.4 Predialysis dosing has been supported basedon simulations from a semimechanistic model describ-ing the influence of ESRD and hemodialysis on thepharmacokinetics of aminoglycosides.19 The subse-quent dialysis session provides clearance of the drugthat approaches that of patients with normal renalfunction and leads to lower concentrations prior toadministering the next dose so that exposure (AUC)is reduced. It also remains possible that the amino-glycoside could be administered during the initialstage of dialysis to decrease waiting time. This con-cept has been tried previously with vancomycin.20

Only one published article that described the pro-file of aminoglycosides when administered predialysiswas found. In this study of arbekacin in 10 patients,4

dosing prior to dialysis appeared to produce an AUCof approximately 40% that of the AUC when the samedose was given postdialysis. The aim of the currentstudy was to evaluate dosing schedules of gentamicinin patients with ESRD and receiving hemodialysis.

METHODS

The study consisted of three phases: (1) data col-lection (prospective and retrospective), (2) modeldevelopment and evaluation using the collectedpharmacokinetic data, and (3) investigation of dos-ing schedules.

Study Population

The study patients were enrolled prospectively fromthe Princess Alexandra Hospital, a tertiary referralhospital in Brisbane, Australia. Patients were eligi-ble if they were enrolled in the hemodialysis programand were receiving gentamicin to treat a suspectedor proven infection. Because of recruitment difficul-ties, some patients were also enrolled retrospectively.Patients were studied on the dosing they happenedto be receiving at the time they presented (regardlessof time of dosing in relation to dialysis), and they wereeligible for inclusion both on first and subsequentdoses of gentamicin. Patients were excluded if they

had been on the dialysis care program for less than 1month.

Data Collection

Collected data included dose, dose interval, the tim-ing of all doses, the infusion time, the time of bloodsamples, and the subsequent concentrations of gen-tamicin, start and stop times for hemodialysis ses-sions, blood flow to the dialyzer, dialyzer filterthickness, surface area, dialysate rate, and amount offluid to be removed. Demographic data collectedincluded age, sex, weight (immediately after dialysiswhen possible), height, and a predialysis creatinineconcentration. Creatinine clearance (CLCR) was approx-imated with the Cockcroft and Gault formula21 usinga predialysis plasma concentration of creatinine andideal body weight,22 as determined by height.23 Thisis an easily available approximation, which hasnumerous limitations that are especially relevant inthis patient group.24-27 Thus, the CLCR descriptor isnot intended to reflect true residual renal functionbut rather to provide an overall descriptor of thepatients’ functional status. It is likely that CLCR is amarker of lean mass and nonrenal mechanisms forclearance of creatinine in this setting.

Ethics

The study was reviewed and approved by the PrincessAlexandra Hospital Research Ethics Committee beforepatient involvement commenced. Prospective patientsprovided written informed consent before joining thestudy.

Pharmacokinetic Sampling

Four blood samples were taken from most prospec-tively enrolled patients. The first blood sample for thestudy was taken by venipuncture at approximately 30minutes after the end of the infusion of the studydose of gentamicin. This blood sample represents anapproximation to Cmax. Three further blood sampleswere collected from each patient to establish plasmaconcentrations of the aminoglycoside antibiotic andclearance within and between dialysis sessions. Typ-ically, this involved a blood sample at the beginningof dialysis, a blood sample at the end of dialysis(both taken from the dialysis machine), and an inter-dialytic blood sample taken prior to the next dialysissession. Any additional blood samples that were takenas a part of routine clinical care were also used fordata analysis. For the retrospective component of the

TEIGEN ET AL



study, blood samples were available from the comput-erized patient records as a part of usual clinical care.

Gentamicin plasma concentrations were determinedusing the Bayer Advia Centaur assays. The between-run coefficient of variation has been estimated as7.4% at 2.2 mg/L and 5.9% at 6.5 mg/L. The lowestlevel of detection was 0.3 mg/L.

Population Analysis

A model was developed using the first-order condi-tional estimation method (FOCE) with the interactionoption in NONMEM (version 5). The model fit wasevaluated by considering the value of the objectivefunction, parameter estimates, their between-subjectvariability (BSV), and visual inspection of diagnosticplots. Pharmacokinetic parameters for the patientpopulation were estimated, such as interdialytic clear-ance (termed nonhemodialysis CL [CLNHD]), clearanceduring hemodialysis (CLHD), apparent volume of dis-tribution (Vd), BSV of the estimates, and variance ofthe residual uncertainty. It was assumed that the elim-ination of gentamicin during dialysis was mediated bya first-order process. The NONMEM covariance com-mand was used to obtain standard errors of theestimates. Model fit was assessed by the likelihoodratio test, where it is assumed that the difference of2 NONMEM objective function values for nestedmodels is asymptotically and approximately χ2 dis-tributed. Under this assumption, the likelihood ratiotest at α = .05 significance level was used as a guideto statistical significance for model discriminationand a reduction of more than 3.84 units in the objec-tive function is considered significant for models with1-parameter difference (χ2, P < .05). Hemodialysiswas considered an essential covariate, and thereforea base hemodialysis model was established beforeother covariates were considered.

Inclusion of covariates into the model was basedon 4 criteria: (1) the parameter values produced bythe covariate model had to be biologically plausible;(2) inclusion of the covariate into the model led to asignificant drop in the objective function of 3.84points or more; (3) the inclusion of the covariate pro-duced a drop in the unexplained between-subject vari-ance; and (4) the observed change in the parametervalues as a result of the covariate inclusion wereclinically significant (eg, >20% change in the para-meter value given the range of covariate values). Themodel was specified as an ordinary differentialequation, with a time-related switch that adjustedclearance due to start and stop times for hemodialy-sis. Volume of distribution was also allowed to incre-ment during the interdialytic interval associated with

an accumulation in extracellular fluid. The statisticalmodel for the data was given by additive, propor-tional, or combined errors (combined error shown):

yij = f(θi, xij)eε1ij + ε2ij

where yij is the jth observed concentration for the ithindividual and f(θi, xij) is the model-predicted concen-tration, ε1ij is the proportional residual error, and ε2ij isthe additive residual error. It is assumed that ε1ij and ε2ijare independent and identically distributed of theform ε ~ N(0, σ 2). The statistical model for BSV was

θi = θeηi

where ηi represents a vector of differences betweenthe ith individual parameter estimates from the pop-ulation parameter values. It is assumed that η areindependent and identically distributed with meanzero and variance-covariance (Ω).

Evaluation of Dosing Schedules

The desirable targets when administering gentam-icin to adult patients with normal renal function areCmax greater than 10 mg/L and a 24-hour AUC between70 and 120 mg·h/L.17 For the purpose of this patientpopulation, a lower Cmax target of 8 mg/L was con-sidered a success, thus accounting for a necessarydose reduction in patients with ESRD when admin-istering aminoglycoside antibiotics. The value of8 mg/L is also the concentration target for conven-tional multiple daily dosing.17 A maximum 24-hourAUC value of 120 mg·h/L was chosen, as this repre-sents the upper limit proposed by Begg et al,17 ratherthan the more usual 100 mg·h/L for patients whohave normal to moderately impaired renal function.

Therefore, treatment success was based on achiev-ing a success in each of the following 3 criteria,assuming a 48-hour dialysis schedule:

1. Cmax ≥8 mg/L2. AUC ≥140 mg·h/L/48 h3. AUC ≤240 mg·h/L/48 h

The final covariate model was used to simulatevarious dosing schedules. MATLAB (version 6.0.0.88,release 12) was used to perform simulations for 1000virtual patients for each proposed dosing schedule.Three doses administered over a treatment period of6 days were simulated for each virtual patient, wherehemodialysis was administered every 48 hours. Wedid not include a 72-hour interdialytic period once perweek, as would commonly be the case in clinical

GENTAMICIN IN PATIENTS WITH ESRD RECEIVING HEMODIALYSIS

PHARMACOKINETICS 1261



1262 • J Clin Pharmacol 2006;46:1259-1267

practice. An extended break would not affect the inter-pretation of the simulations because it would serveonly to increase the success rate across all dosing reg-imens for 1 dose interval. The primary outcome mea-sures were peak concentration of aminoglycoside(Cmax) and AUC. Treatment success in 1 of the crite-ria was denoted by an indicator variable set to 1, andtreatment failure was set to 0. With 3 administereddoses and 3 indicator variables, an overall successfor a virtual patient was provided by the product ofthe 9 indicator variables. Success rates for the sepa-rate criteria for individual doses were also consid-ered. The best dosing schedule was one that providedthe highest likelihood of overall success (in all cate-gories on all doses). Execution variability was incorpo-rated into the simulation model to account for variabletimes of starting hemodialysis in relation to the dose(up to 4 hours delayed) and also variability regard-ing the duration of the infusion (15-60 minutes).

A postdialysis dosing schedule was simulated, giv-ing 3 commonly used postdialysis doses (160 mg asthe first dose and 120 mg as the second and third dose)within 4 hours after the termination of each dialysissession every 48 hours. Consequently, success rateswere compared between the proposed best dosing reg-imen and the empirical (postdialysis) dosing schedule.

RESULTS

Forty-six patients were included in the NONMEManalysis, 20 were enrolled prospectively, and 26patients were enrolled retrospectively. The patients’ages ranged from 18 to 83 years, with a mean of 57.3

years; there were 23 women and 23 men. Character-istics of the study subjects are shown in Table I.High-flux polysulfone membranes were the most com-monly used dialyzer (used in approximately 59% ofpatients), followed by low-flux polysulfone mem-branes (used in 41% of patients). Filter types areprovided in the appendix. The mean blood flow ratewas 14.6 L/h and dialysate flow rate was similar forall patients in the study (≈30 L/h).

Population Analysis

A zero-order input 1-compartment model with log nor-mal BSV of CL and Vd and a combined proportionaland additive error provided the best fit to the data andwas chosen as the best base model. Hemodialysis wasthe most appropriate first step to improve model fit,and the incorporation of hemodialysis resulted in adrop in the objective function value of 69 points (seeTable II for parameter values). Between-subject vari-ability was only included for CLNHD and Vd, becausethe data set did not support its inclusion for CLHD. Amore mechanistic model, which included a factor fordialysis clearance taking into account differences indialyzer filter thickness, surface area of the mem-brane, intrinsic clearance, and blood flow was notexplored, because no random effect could be esti-mated for CLHD and hence no improvement in modelfit could occur. We did however assess the influenceof adding the filter type (high- or low-flux) into themodel that had a BSV term for CLHD. This resulted ina 9% increase in CLHD for the high-flux filters com-pared to low-flux. However, this was not statisticallysignificant, and filter type was not retained in themodel. The indicator covariate CLCR was the onlyinfluential covariate (for CLNHD) and resulted in a dropin the objective function value by a further 24.1 points.The CLCR indicator variable explained 35% of thebetween-subject variance in CLNHD and also reducedthe between-subject variance in Vd by 53% (see Table IIfor parameter values with different models).

Including weight or ideal body weight into themodel alone or combined as influential factors onVd or CLNHD did not improve model fit. An attemptwas made to include fluctuations in extracellularfluid volume as a function of the changing wetweight as a covariate for the Vd, but the data did notsupport this.

The final model contained HD as a covariate forCLHD and centered CLCR as a covariate for CLNHD.The mean value of interdialytic gentamicin clear-ance was 0.453 L/h, and the mean value for gentam-icin clearance during dialysis was 4.69 L/h. The

TEIGEN ET AL

Table I Patient Characteristics

Patient Characteristic Mean ±± SD(n = 46) (range) or %

Male 50Age, y 57.3 ± 17.3 (18-83)Height, cm 164.7 ± 11.6 (135-195)Weight, kga 72.4 ± 17.2 (42.1-100.5)CLCR, L/hb 0.53 ± 0.2 (0.26-1.24)No. of blood samples 4.6 ± 2.2 (1-10)

Dialysis detailsBlood flow rate, L/h 14.6 ± 2.2 (10.8-15.3)Duration of dialysis, h 4.1 ± 0.5 (2-5)Interdialytic period, h 47 ± 19.9 (20-102)Time since starting 22.4 ± 29.5 (1-144)

dialysis, mo

a. Dry weight.b. Calculated using the Cockcroft and Gault formula using ideal bodyweight.



observed versus predicted concentrations for the baseand final model are shown in Figure 1.

Of the study patients, 38 had postdialysis dosingand an evaluable peak concentration taken. Theremaining patients who had postdialysis dosing didnot have a blood sample taken near (within 30 min-utes of) the actual time of the Cmax (see Figure 2 for thedistribution of observed “Cmax” values). Four patients(11% of these patients) attained a peak concentra-tion ≥8 mg/L, and 2 of these patients had an AUCabove the maximum target of 240 mg·h/L/48 h. Bothpatients had estimated values of CLNHD and CLCR lowerthan the average values for the population, and theadministered doses were 120 mg and 180 mg post-dialysis. Only one patient received predialysis dosing(specifically 240 mg gentamicin immediately beforedialysis), and this patient achieved an observed peakconcentration of 8.5 mg/L, and an AUC of 89.6 mg·h/L.Only 4 patients in the study had AUC values above themaximum target. Thirty-four patients had low values ofthe observed “Cmax” values of between 2 and 7 mg/L.

Dose Individualization

Multiple dosing schedules were evaluated based onthe final covariate model. The best dosing scheduleidentified for gentamicin through the simulationswas 300 mg as a first dose, 240 mg as a second dose,and 220 mg as a third dose, immediately beforehemodialysis occurring every 48 hours, with up to 4hours flexibility between the dose and hemodialy-sis. This dosing schedule achieved treatment successfor all 3 criteria for all 3 doses in 23.4% of patientsand resulted in a satisfactory peak concentration inmore than 90% of patients for all 3 doses, with 85%of patients achieving a maximum value of AUC



Table II Parameter Values Estimated With the Base, Base Hemodialysis (HD),and Final Covariate Model

Parameter Base Model Base HD Model Final Model (SE%)

CL, L/h 0.604 — —CLNHD, L/h — 0.388 0.453 × CLCR/0.53 (0.9)CLHD — 4.74 4.69 (0.86)V, L 22.5 24.3 23.5 (0.91)*ΩCL 0.271 0.406 0.264 (0.59)*ΩV 0.0637 0.0537 0.0256 (0.45)Σ proportional** 0.133 0.0625 0.0804 (1.77)

additive** 3.54 ×10–10 0.0495 0.00659 (3.05)

CL, clearance; CLNHD, nonhemodialysis clearance; CLHD, clearance during hemodialysis; Vd, volume of distribution; *Ω refers to the lognormal varianceof η (measure for BSV) corresponding to CLNHD (CL base model) and Vd. **variance.

A

Ob

serv

ed c

on

cen

trat

ion

s(D

V)

B

0

2

4

6

8

10

12

14

0 4 8 10

Predicted concentrations

Ob

serv

ed c

on

cen

trat

ion

s(D

V)

0

2

4

6

8

10

12

14

0 6 10 12

Predicted concentrations2 4 8

2 6

Figure 1. Observed versus population predicted concentrationfor (A) base model and (B) final model.



1264 • J Clin Pharmacol 2006;46:1259-1267

(AUCmax) below 240 mg·h/L at all times throughthe 6-day course of treatment (see Table III andFigure 3).

None of the simulated patients receiving the post-dialysis dosing schedule (160 mg, 120 mg, and 120 mgas first, second, and third dose, respectively) achievedoverall success, and the main reason for this was thelow peak concentrations achieved with this dosingregimen (see Table III). This regimen was better atmeeting the AUC criteria than the peak concentrationcriterion; thus, unacceptable aminoglycoside expo-sure appears to be relatively infrequent with this dos-ing schedule.

It is worth noting that although we provide dosingsuggestions for up to 3 days postdose that these doseswould need to be adjusted to meet the patient’spharmacokinetic and clinical needs. We include thedoses to highlight 2 points, to show that (1) doseswhen given predialysis are larger (almost twice thedose) and that (2) minimal accumulation occurs whendosing is given predialysis. This latter point is shownas the second and subsequent doses are not greatly

reduced from that of the first dose when the dosesare administered predialysis.

DISCUSSION

The model developed in this study describes the phar-macokinetics of gentamicin in hemodialysis patientsand was used to explore dosing regimens aiming toimprove the efficacy-toxicity profile when adminis-tered to patients with ESRD receiving hemodialysis.

Although it appears to be common practice todose aminoglycosides after patients have receivedhemodialysis, simulations from the model indicatethat attainment of suitable peak concentrations isvery poor in this circumstance. However, it appearsthat these regimens may provide a suitable exposurelevel. Simulations of a postdialysis dose of gentam-icin of 160 mg yielded no patients who attained aCmax greater than 8 mg/L on the first dose. This dosehowever provided a reasonable distribution of AUCvalues, which suggests that doses that will provideappropriate Cmax values are likely to result in over-exposure. The only dose from which a small fractionof patients actually met the peak concentration treat-ment criterion was for the third dose and essen-tially arose because of significant accumulation thatoccurred over the 3 dose periods. Whenever doseslower than the 160 mg/120 mg/120 mg scenario areadministered, even poorer results may be anticipatedin terms of efficacy than with the simulated post-dialysis regimen.

The estimated value of gentamicin clearance dur-ing dialysis (4.7 L/h) is similar to values of clearanceobserved in patients with normal renal function.28 Inaddition, in our study we found a 51% BSV in gen-tamicin clearance, which is comparable to 45% for

TEIGEN ET AL

Figure 2. Observed peak concentrations with postdialysis dos-ing (n = 38).

Table III Percentage of Success Rate Achieved With the Simulated Dosing Schedules

Success in AllAUCmin AUCmax Criteria on All Doses

Criteria Cmax ≥≥88 mg/L ≥≥140 mg·h/L/48h ≤≤240 mg·h/L/48h (overall success)

Predialysis dosing1. 300 mg 97 77 92 232. 240 mg 97 75 853. 220 mg 91 70 89

Postdialysis dosing1. 160 mg 0 89 93 02. 120 mg 0.7 84 773. 120 mg 6.1 82 76

Cmax, maximum concentration; AUCmin, minimum value of the area under the concentration time curve; AUCmax, maximum value of the area under theconcentration time curve.



patients with normal renal function.28 In 2 other stud-ies evaluating dialyzer membranes with increasedpermeability (high-flux membranes), only moderatelyhigher values than this (5.5 L/h and 7 L/h) are reported,corresponding to a half-life of 2.2 hours or more duringdialysis.29,30 In a typical patient with normal renal

function and weighing 72.4 kg (the mean value inthis study population), an elimination half-life of 2.5hours would be expected, which relates to a clear-ance in such a patient to be approximately 5 L/h. Xuanet al recently found a gentamicin clearance of 4.32L/h in a pharmacokinetic evaluation of 939 adulthospitalized patients on aminoglycoside therapy.31

Thus, hemodialysis appears to offer almost normalclearance conditions in patients with ESRD duringthe dialysis period. In essence therefore, the dialy-sis period will provide the most close to “normal”pharmacokinetic-pharmacodynamic profile for gen-tamicin in patients with ESRD. From this study, fora typical 4-hour hemodialysis session, the clearanceof gentamicin during dialysis (18.8 L/4 h) is approx-imately equivalent to the average total clearance dur-ing the interdialytic interval (19.9 L/44 h).

Based on predialysis dosing simulations from thisstudy, it was found that an appropriate dosing strategyof gentamicin for patients who dialyze 3 times a weekto be 300 mg predialysis as the first dose and then sub-sequent predialysis doses of 240 mg and 220 mg assecond and third dose, respectively. This regimen pro-duced a desirable peak concentration (above 8 mg/L)in 97% of patients for the first and second dose and a91% success for patients for the third dose withoutunacceptably high exposure (see Table III). This illus-trates how higher doses (300 mg, 240 mg, 220 mg)can be administered predialysis, because of the subse-quent clearance provided by dialysis (similar togentamicin clearance in patients with normal renalfunction), resulting in a lower total exposure thanwhen administering lower doses (160 mg, 120 mg, 120mg) postdialysis. The doses discussed here are onlystarting conditions, and it is recommended to monitorgentamicin concentrations from the first dose andadjust the dose accordingly using Bayesian dose indi-vidualization methodology in conjunction with knowl-edge of the patient’s current clinical picture. It shouldalso be noted that the optimal exposure levels have yetto be determined for this patient population, and it hasbeen assumed in this study that exposure levels forpatients with some level of intact renal function are areasonable surrogate for those with ESRD.

The final model included the indicator for residualrenal function, provided by CLCR, as a covariate, andthis was the only covariate that improved model fit.Creatinine clearance as a measure for residual renalfunction is expected to be less accurate in hemodial-ysis patients compared to patients with normal renalfunction or less advanced renal impairment.25 Inhemodialysis patients, plasma creatinine is a reflectionof dialysis adequacy and the amount of body muscle,



A

B

Cmax 1. dose predialysis dosing

50

6 8 10 12 140

100

150

200

250

50

0 100 200 3000

100

150

200

250

50

6 8 10 12 140

100

150

200

250


50

6 8 10 12 16140

100

150

200

300

250


AUC 1. dose predialysis dosing

400

50

0 100 200 3000

100

150

200

250

500400

50

0 100 200 3000

100

150

200

300

250

500400



Figure 3. Histogram of the simulated criteria values comparedto target values (vertical lines) for predialysis dosing for (A) max-imum concentration (Cmax) and (B) area under the concentration-time curve (AUC) for a dose of 300 mg for the first dose, 240 mgfor the second dose, and 220 mg for the third dose.



1266 • J Clin Pharmacol 2006;46:1259-1267

as well as residual renal function. In addition, tubu-lar secretion of creatinine may contribute to an over-estimation of glomerular filtration, which may bemore significant as renal dysfunction progresses.25-27

The observed range for the estimated CLCR in thisstudy is high (range, 0.27-1.24 L/h; mean, 0.53 L/h).It is likely that estimation of CLCR by urine collectionwould have resulted in lower values, as urine pro-duction in a considerable proportion of patients isoften minimal. Nevertheless, the covariate CLCR wasa clinically and statistically significant predictor ofinterdialytic gentamicin clearance. The variation inCLCR explained a substantial amount of the observedvariability in clearance between dialysis sessions.Whether CLCR in this setting is predominantly acharacterization of creatine producing body mass orsome level of residual renal function is unknown.Caution should be taken if this relationship is extrap-olated to other types of dialysis settings.

The developed model in this study was empiricallybased (although has reasonable biological grounds)and included a switch that turned clearance due todialysis on and off according to onset and offsettimes of hemodialysis. A semimechanistic modelthat was developed previously,19 which accounts forhemodialysis factors that vary between patients,such as blood flow rates, filter thickness, and surfacearea of the membrane, was not implemented in thismodel. The application of this model depended onthe ability to estimate the random variability betweenpatients in their value of CLHD. Unfortunately, thiswas not supported by the data, and hence the morecomplex semimechanistic model could not be assessedfor its ability to explain the random variability asso-ciated with CLHD. This more complex model remainsa candidate to achieve better predictions of thewithin-dialytic clearance of gentamicin.

In summary, this study provides strong support forpredialysis dosing of aminoglycosides in patients onhemodialysis. Hemodialysis provides a replacementfor normal renal function in terms of gentamicin clear-ance so that the concentration-time curve approachesthat of patients without ESRD for at least a portion ofthe dose interval. In patients with ESRD and hemodial-ysis who are at high risk of death from infection, pre-dialysis dosing of gentamicin in the setting of suspectedor confirmed sepsis would seem to represent a ratio-nal, efficacious, and safe strategy. Target concentrationintervention strategies that employ Bayesian method-ology are warranted to dose individual patients receiv-ing gentamicin who have ESRD and are receivinghemodialysis.

We wish to acknowledge assistance from medical and nursingstaff in the hemodialysis units, pathology, and pharmacy depart-ments at Princess Alexandra Hospital through the course of thestudy.

Financial disclosure: We also wish to acknowledge the Universityof Queensland for financial support through an ECR Grant (2003).

APPENDIXFilter Types Used for Patients in This Study

Filter Type Number of Patients

Fresenius F10 HPS Low Flux 2.4 11Fresenius F7 HPS Low Flux 1.6 3Fresenius F7 Low Flux 1.6 1Fresenius F8 HPS Low Flux 1.8 3Fresenius F8 Low Flux 1.8 1Fresenius FX80 2Fresenius FX80 S 1.8 3Fresenius HF80 1.8 10Fresenius HF80 S 1.8 1Gambro Allwal GFS/GEH Plus 11 1

(Haemophan)Gambro Polyflux 140 H 3Gambro Polyflux 170H 5Gambro Polyflux 210H 2

REFERENCES

1. Goodkin DA, Mapes DL, Held PJ. The dialysis outcomes andpractice patterns study (DOPPS): how can we improve the care ofhemodialysis patients? Sem Dialysis. 2001;14:157-159.2. McDonald SP, Russ GR, Kerr PG, Collins JF. ESRD in Australiaand New Zealand at the end of the millennium: a report from theANZDATA registry. Am J Kidney Dis. 2002;40:1122-1131.3. Sarnak MJ, Jaber BL. Mortality caused by sepsis in patientswith end-stage renal disease compared with the general popula-tion. Kidney Int. 2000;58:1758-1764.4. Matsuo H, Hayashi J, Ono K, et al. Administration of amino-glycosides to hemodialysis patients immediately before dialysis:a new dosing modality. Antimicrob Agents Chemother. 1997;41:2597-2601.5. Moore RD, Smith CR, Lietman PS. The association of amino-glycoside plasma levels with mortality in patients with gram-negative bacteremia. J Infect Dis. 1984;149:443-448.6. Moore RD, Smith CR, Lietman PS. Association of aminoglycosideplasma levels with therapeutic outcome in gram-negative pneu-monia. Am J Med. 1984;77:657-662.7. Moore RD, Lietman PS, Smith CR. Clinical response to amino-glycoside therapy: importance of the ratio of peak concentrationto minimal inhibitory concentration. J Infect Dis. 1987;155:93-99.8. Vogelman B, Gudmundsson S, Leggett J, Turnidge J, Ebert S,Craig WA. Correlation of antimicrobial pharmacokinetic parame-ters with therapeutic efficacy in an animal model. J Infect Dis.1988;158:831-847.

TEIGEN ET AL



9. Barclay ML, Begg EJ. Aminoglycoside adaptive resistance:importance for effective dosage regimens. Drugs. 2001;61:713-721.10. Barclay ML, Begg EJ, Chambers ST. Adaptive resistance fol-lowing single doses of gentamicin in a dynamic in vitro model.Antimicrob Agents Chemother. 1992;36:1951-1957.11. Barclay ML, Begg EJ, Chambers ST, Peddie BA. The effect ofaminoglycoside-induced adaptive resistance on the antibacterialactivity of other antibiotics against Pseudomonas aeruginosa invitro. J Antimicrob Chemother. 1996;38:853-858.12. Barclay ML, Begg EJ, Chambers ST, Thornley PE, Pattemore PK,Grimwood K. Adaptive resistance to tobramycin in Pseudomonasaeruginosa lung infection in cystic fibrosis. J Antimicrob Chemother.1996;37:1155-1164.13. Moore RD, Smith CR, Lietman PS. Risk factors for the devel-opment of auditory toxicity in patients receiving aminoglycosides.J Infect Dis. 1984;149:23-30.14. Moore RD, Smith CR, Lipsky JJ, Mellits ED, Lietman PS. Riskfactors for nephrotoxicity in patients treated with aminoglycosides.Ann Intern Med. 1984;100:352-357.15. Giuliano RA, Verpooten GA, Verbist L, Wedeen RP, De BroeME. In vivo uptake kinetics of aminoglycosides in the kidney cor-tex of rats. J Pharmacol Exp Ther. 1986;236:470-475.16. Verpooten GA, Giuliano RA, Verbist L, Eestermans G, De BroeME. Once-daily dosing decreases renal accumulation of gentamicinand netilmicin. Clin Pharmacol Ther. 1989;45:22-27.17. Begg EJ, Barclay ML, Duffull SB. A suggested approach to once-daily aminoglycoside dosing. Br J Clin Pharmacol. 1995;39:605-609.18. Barclay ML, Duffull SB, Begg EJ, Buttimore RC. Experience ofonce-daily aminoglycoside dosing using a target area under theconcentration-time curve. Aust NZ J Med. 1995;25:230-235.19. Dang L, Duffull S. Development of a semi-mechanistic modelto describe the pharmacokinetics of gentamicin in patients receiv-ing hemodialysis. J Clin Pharmacol. 2006;46:662-673.20. Ariano RE, Fine A, Sitar DS, Rexrode S, Zelenitsky SA. Adequacyof a vancomycin dosing regimen in patients receiving high-fluxhemodialysis. Am J Kidney Dis. 2005;46:681-687.

21. Cockcroft DW, Gault MH. Prediction of creatinine clearancefrom serum creatinine. Nephron. 1976;16:31-41.22. Pesola GR, Akhavan I, Madu A, Shah NK, Carlon GC. Predictionequation estimates of creatinine clearance in the intensive care unit.Intensive Care Med. 1993;19:39-43.23. Devine D. Case study number 25 gentamicin therapy. DrugIntell Clin Pharm. 1974;8:650-655.24. Bauer JH, Brooks CS, Burch RN. Clinical appraisal of creati-nine clearance as a measurement of glomerular filtration rate. AmJ Kidney Dis. 1982;2:337-346.25. Shemesh O, Golbetz H, Kriss JP, Myers BD. Limitations of cre-atinine as a filtration marker in glomerulopathic patients. KidneyInt. 1985;28:830-838.26. Milutinovic J, Cutler RE, Hoover P, Meijsen B, Scribner BH.Measurement of residual glomerular filtration rate in the patientreceiving repetitive hemodialysis [abstract]. Kidney Int. 1975;8:185-190.27. Luke DR, Halstenson CE, Opsahl JA, Matzke GR. Validity ofcreatinine clearance estimates in the assessment of renal function.Clin Pharmacol Ther. 1990;48:503-508.28. Kirkpatrick CM, Duffull SB, Begg EJ. Pharmacokinetics of gen-tamicin in 957 patients with varying renal function dosed oncedaily. Br J Clin Pharmacol. 1999;47:637-643.29. Amin NB, Padhi ID, Touchette MA, Patel RV, Dunfee TP,Anandan JV. Characterization of gentamicin pharmacokinetics inpatients hemodialyzed with high-flux polysulfone membranes.Am J Kidney Dis. 1999;34:222-227.30. Vercaigne LM, Ariano RE, Zacharias JM. Bayesian pharma-cokinetics of gentainicin in a haemodialysis population. ClinPharmacokinet. 2004;43:205-210.31. Xuan D, Nicolau DP, Nightingale CH. Population pharmacoki-netics of gentamicin in hospitalized patients receiving once-dailydosing. Int J Antimicrob Agents. 2004;23:291-295.





Dosing of Gentamicin in Patients With End-Stage Renal Disease Receiving Hemodialysis

Documents

Transcript of Dosing of Gentamicin in Patients With End-Stage Renal Disease Receiving Hemodialysis