Population pharmacokinetics of piperacillin in plasma and
subcutaneous tissue in patients on continuous renal replacement therapy
MatsBuea,b,c,*,TomásSoud,Anna Sophie L.Okkelsb,Pelle Hanberga,c,Anders Thorstedd, Lena E.Fribergd,TorbenL.Anderssone,Kristina Öbrink-Hansenf,SteffenChristensenb
aDepartmentofOrthopaedicSurgery,HorsensRegionalHospital,Sundvej30,8700Horsens,Denmark
bDepartmentofAnaesthesiaandIntensiveCareMedicine,AarhusUniversityHospital,PalleJuul-JensensBoulevard99,8200AarhusN,Denmark
cOrthopaedicResearchUnit,AarhusUniversityHospital,PalleJuul-JensensBoulevard99,8200AarhusN,Denmark
dDepartmentofPharmaceuticalBiosciences,UppsalaUniversity,Box591,75124Uppsala,Sweden
eDepartmentofClinicalBiochemistry,AarhusUniversityHospital,PalleJuul-JensensBoulevard99,8200AarhusN,Denmark
fDepartmentofInfectiousDiseases,AarhusUniversityHospital,PalleJuul-JensensBoulevard99,8200AarhusN,Denmark
ARTICLE INFO
Articlehistory:
Received7September2019
Receivedinrevisedform11January2020 Accepted13January2020
Keywords:
PK/PD Modelling β-lactams Microdialysis Tissuedistribution
Continuousrenalreplacementtherapy
ABSTRACT
Objectives:Piperacillinisaβ-lactamantimicrobialfrequentlyusedincriticallyillpatientswithacute kidney injurytreatedwithcontinuousrenal replacementtherapy(CRRT).However, dataregarding piperacillintissueconcentrationsinthispatientpopulationarelimited.Aprospectiveobservational studywasconductedoffreepiperacillinconcentrationsduringasingle8-hdosingintervalinplasma(8 samples) and subcutaneoustissue (SCT) (13 samples),in 10patients treatedwith CRRTfollowing piperacillin4ggivenevery8hasintermittentadministrationover3min.
Methods: A population pharmacokinetic model was developed using NONMEM 7.4.3, tosimulate alternative administration modes and dosing regimens. SCT concentrations were obtained using microdialysis.Piperacillinconcentrationswerecomparedtotheclinicalbreakpointminimuminhibitory concentration (MIC) for Pseudomonas aeruginosa (16 mg/l), with evaluation of the following pharmacokinetic/pharmacodynamicstargets:50%fT>1MIC,100%fT>1MIC,and100%fT>4
MIC.
Results:SCTconcentrationsweregenerallylowerthanplasmaconcentrations.Forthetargetof50%free time(fT)>1MICand100%fT>1MIC,piperacillin4gevery8hresultedinprobabilityoftarget attainment(PTA)>90%inbothplasmaandSCT.PTA>90%forthetargetof100%fT>4MICwasonly achievedforcontinuousinfusion.
Conclusions:Piperacillin4gevery8hislikelytoprovidesufficientexposureinbothplasmaandSCTto treatP.aeruginosainfectionsincriticallyillpatientsonCRRT,giventhattargetsof50%fT>1MICor100%
fT>1MICareadequate.However,ifamoreaggressivetargetof100%fT>4 MICisadopted, continuousinfusionisneeded.
©2020TheAuthor(s).PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases.
ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc- nd/4.0/).
Introduction
Septicshockisacommoncauseofacutekidneyinjury(AKI) incriticallyillpatients(deMendoncaetal.,2000;Bagshawetal., 2007).Piperacillin/tazobactamisaβ-lactam/β-lactamaseinhibi- torcombinationcommonlyusedbothforempiricalandtargeted treatment in theintensive care unit(ICU setting. It exhibits a broadspectrumantibacterialactivityagainstbothgram-positive and gram-negative bacteria, including Pseudomonas aeruginosa (Pinder and Lipman, 2002).The efficacy is related to thetime duringwhichthefreedrugconcentrationismaintainedabovethe
* Corresponding author at: Department of Orthopaedic Surgery, Horsens RegionalHospital,Sundvej30,8700Horsens,Denmark.
E-mailaddresses:matsbue6@rm.dk(M.Bue),tomas.sou@farmaci.uu.se(T.Sou), asokkels@hotmail.com(A.S.L. Okkels),pehanb@rm.dk(P.Hanberg),
anders.thorsted@farmbio.uu.se(A.Thorsted),lena.friberg@farmbio.uu.se (L.E. Friberg),t.l.a@stofanet.dk(T.L. Andersson),
kristina.Obrink.Hansen@auh.rm.dk(K.Öbrink-Hansen), steffen.christensen@auh.rm.dk(S.Christensen).
https://doi.org/10.1016/j.ijid.2020.01.010
1201-9712/©2020TheAuthor(s).PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).
ContentslistsavailableatScienceDirect
International Journal of Infectious Diseases
j o u r n a l h o m ep a g e : w w w . e l s e v i e r . c o m / l o c a te / i j i d
minimum inhibitory concentration(fT >MIC). For piperacillin, achievementof50%fT>1MICinplasmahasbeenassociated withclinicalcure(Drusano,2004;Seyleretal.,2011);however, the exact association between piperacillin concentrations and clinicalefficacyremainsuncertain.Sincemostbacteriaresidein theinterstitialspaceof solidtissues, information ontarget site concentrations is needed in order to evaluate the existing treatment targets more closely (Drusano, 2004). Microdialysis hasevolvedasapromisingtooltocontinuouslysamplethefree fraction of antimicrobials from the interstitial fluid of the investigated tissue (de la Pena et al., 2000; Joukhadar and Muller,2005).Incontrasttoexistingmethodsfortheassessment of tissue concentrations, microdialysis has the potential to generatedynamicantimicrobialpharmacokinetic(PK)datafrom targetsites.
Continuousrenalreplacementtherapy(CRRT)isusedin5–10%
ofcriticallyillpatientstreatedin theICU(Uchinoet al.,2005).
During CRRT, plasma–water and solutes with low molecular weight,includingmanytypesofantimicrobials,areremovedfrom the circulation. In particular, water-soluble drugs with low albuminbinding, suchas piperacillin, maybe removed (Seyler etal.,2011;Robertsetal.,2012).Still,dataregardingpiperacillin tissueconcentrationsincriticallyillpatientstreatedwithCRRTare limited(Ulldemolinsetal.,2014;Vargheseetal.,2014).Therefore, wesetouttoassessthePKoffree(unbound)piperacillinduringa single8-hdosingintervalinplasmaandsubcutaneoustissue(SCT) using microdialysis in critically ill patients treated with CRRT followingpiperacillin4 ggivenevery8h(q8h)as intermittent administration(IA) over3min.ByestablishingapopulationPK model,alternativedosingregimensandmodesofadministration weresimulated,andtheirprobabilityoftargetattainment(PTA) wasevaluated.
Methods
This prospective observational study was conducted in the DepartmentofAnaesthesiaandIntensiveCareMedicine,Aarhus UniversityHospital,Denmark,betweenDecember2017andJune 2018. Chemicalanalyses wereperformed in theDepartment of ClinicalBiochemistry,AarhusUniversityHospital.
Patientpopulation,CRRT,andpiperacillin
Patients treated with intravenous piperacillin/tazobactam(4 g/0.5g)q8hgivenasanIAover3minandrequiringCRRTwere eligibleforthestudy.CRRTtreatmentandpiperacillin/tazobactam therapywereinitiatedindependentofthecurrent study,atthe discretionofthetreatingphysician.Aminimumoftwodosesof piperacillin/tazobactamweregivenbeforestudyinclusion,withall priorpiperacillin/tazobactamdosesrecordedforincorporationin thedataset.
Exclusioncriteriawerepregnancy, plateletsbelow5109/l, internationalnormalized ratio(INR) above 5,known allergy to benzylpenicillinorpiperacillin,lactate>4mmol/l,andtheneedfor noradrenalin>0.50mg/kg/minor theequivalentdoses of vaso-
pressors.Thefollowingdatawerecollectedandregisteredforeach enrolledpatientonthedayofinclusion:age,sex,weight,typeof anticoagulant required for CRRT, admission diagnosis, urine output,microbiology,plasmacreatinine,plasmabilirubin,plasma albumin, plasma urea, Simplified Acute Physiology Score III (SAPS3),andCRRTsettings.
ThePrismaflexCRRTsystemwasused(BaxterInternationalInc., Deerfield,IL,USA)andallpatientsweretreatedusingtheM100 filter.AllCRRTsettingswereprescribedatthediscretionofthe treatingphysicianfollowing standardproceduresin thedepart- ment, including continuous veno-venous haemodiafiltration
(CVVHDF),citrateanticoagulation,and theaimfora prescribed effluentdoseof30ml/kg/h.
Microdialysis
Briefly,microdialysisisacatheter-basedtechniquethatfollows adiffusionofwater-solublemolecules,includingantimicrobials, fromtheinterstitialspaceofaccessibletissuesacrossasemiper- meablemembranelocatedatthetipof thecatheter(Bueetal., 2018a,b;Tottrupetal.,2019).Duetocontinuousperfusionofthe microdialysiscatheter,equilibriumwillneveroccur.Accordingly, thefreeconcentrationofthedialysatewillonlyrepresentafraction of the actual free concentration in the tissue. This fraction is referredtoas‘relativerecovery’andcanbedeterminedbyvarious calibration methods, which is imperative when assessing free tissueconcentrations(BouwandHammarlund-Udenaes,1998;de laPenaetal.,2000;Joukhadarand Muller,2005).Inthisstudy, benzylpenicillinwasusedasaninternalcalibratorforpiperacillin (Robertsetal.,2009).Nopatientreceivedbenzylpenicillinbefore orafterinclusioninthestudy.Therelativerecoveryofpiperacillin wascalculatedfromthelossoftheinternalcalibratoracrossthe microdialysis membrane, using the retrodialysis by calibrator method(BouwandHammarlund-Udenaes,1998;Hanbergetal., 2018).Allcatheterswerecalibratedindividually.Detaileddescrip- tions of microdialysis can be found elsewhere (Muller, 2002;
JoukhadarandMuller,2005;Hanbergetal.,2018).Thisstudyused equipment from M Dialysis AB (Stockholm, Sweden). CMA 63 catheterswereused(membranelength30mmwith20kDacut- off), and CMA 107precisionpumps produced a flow rate of 2
ml/min.
Samplingprocedures
Richmicrodialysisandbloodsamplingwereperformedduring a single 8-h dosing interval. The microdialysis catheter was insertedintheSCToftheupperarmaminimum30minpriortoa new piperacillin/tazobactam administration. The catheter was fixedtotheskinwithasinglesuturetopreventdisplacement.The microdialysis system was perfused with 0.9% NaCl containing benzylpenicillin at a concentration of 4 mg/l (provided by the
Pharmacyat Aarhus UniversityHospital). Single blood and SCT trough samples were obtained prior to the administration of piperacillin/tazobactam (time 0), after which dialysates were collectedevery20minforthefirst3handevery60mininthe following5h,providing13dialysatesintotal.Bloodsampleswere drawnfromaperipheralarterialcatheterat10,20,30,60,120,240, and480minpost-dose,resultinginatotalofeightbloodsamples perpatient.
Blood samples were kept for a maximum of 3 h at room temperaturebeforebeingcentrifugedat3000gfor10min.Plasma aliquots were frozen and stored at 80 C until analysis. All dialysateswereimmediatelyfrozenondryiceforamaximumof10 hbeforebeingtransferredandstoredat 80Cuntilanalysis.
Quantificationofpiperacillinandbenzylpenicillinconcentrations
Unbound piperacillin concentrations in microdialysates and plasma and benzylpenicillin concentrations in microdialysates weresimultaneouslyquantifiedwithastandardassayusingultra- highperformanceliquidchromatography(UHPLC)intheroutine hospitallaboratoryfortherapeuticdrugmonitoringinhospitalized patients. The UHPLC system consisted of an eluent pump, autosampler, column compartment, and a UV detector(Agilent 1290 Infinity; Agilent Technologies, Waldbronn, Germany), equippedwitha Poroshell 120,EC-C18 2.1 100mm, 2.7-mm
column(Agilent).BeforeUHPLCanalysis,plasmawascentrifuged
ina96-wellultrafilterplatewitha30-kDamolecularmasscut-off retaining protein (mainly albumin) with unspecifically bound antimicrobials, leaving only thefree fractionof piperacillin for analysis. Five microliters of microdialysate orplasma was then mixed with 20 ml phosphate buffer pH 3.0 (50 nM NaH2PO4, adjustedwitho-phosphoric acid,85%).Aftermixing,10mlwas
injectedintotheUHPLCsystem.Calibrationstandardsof1.56,3.13, 6.26, and 25.0 mg/ml piperacillinand 1.56, 3.13,6.25, and 25.0
mg/mlofbenzylpenicillinin15ml0.9%NaClwerepreparedand
analysed.Nosignificantmatrixeffectonthecalibrationcurveswas observed.Chromatographywasconductedat40C,withagradient of acetonitrile and phosphate buffer (50 nM NaH2PO4, 5%
acetonitrile,pH3.0).Theconcentrationofacetonitrilewaselevated from20%to30%overatimespanof4min,andthetotaltimeof analysiswas4minwithapostruntimeof1min.Piperacillinand benzylpenicillinwere detected at 210 nm. The lower limits of quantificationwerefoundtobe0.5mg/mlforpiperacillinand0.3
mg/mlforbenzylpenicillin.Inter-runimprecisions(percentcoef- ficientsofvariation(CV%)were15.3%at4.5mg/mland8.2%at15.6 mg/mlforquantificationofpiperacillin,and13.3%at2.6mg/mland
14.4%at12.0mg/mlforquantificationofbenzylpenicillin.Dilution experimentsusinghigh-concentrationsamplesshowedlinearity in measurement response within 0.5 mg/ml to 500 mg/ml for
piperacillin,andwithin0.5mg/mlto60mg/mlforbenzylpenicillin.
All samples were diluted before bulk analyses as appropriate, basedonpilotexperiments.
MIC
The clinicalbreakpoint MIC for planktonic P.aeruginosa (16 mg/l) defined by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (EUCAST, 2019) was used to evaluate thefollowing pharmacokinetic/pharmacodynamic(PK/
PD)targets:50%fT>1MIC,100%fT>1MIC,and100%fT>4 MIC.P.aeruginosaisaproblematicpathogenfrequentlyseeninthe ICUsetting,andtheclinicalbreakpointMICmayreflectaworst case scenario regarding bacterial susceptibility in empirical antimicrobialtherapy(Shorr,2009).
Pharmacokineticmodelling
All available samples were included and analysed using populationPKmodellingin NONMEM7.4.3(ICONDevelopment Solutions,Hanover,MD,USA)(Boeckmannetal.,2013),withthe supportofPerl-Speaks-NONMEM(PsN)andPiraña(Keizeretal., 2013). The first-order conditional estimation method with interactionwas usedforparameterestimation. Rversion3.5(R FoundationforStatisticalComputing,Vienna,Austria)wasusedfor datamanagement,andthexpose4packagewasusedtosupport modeldiagnosticsandgraphicalevaluationoftheresults(Keizer etal.,2013).
Thelikelihoodratiotest(LRT)wasusedtoevaluatestatistical significancefortheinclusionofadditionalparametersinnested models,withtheobjectivefunctionvalue(OFV)assumedtobeChi- squaredistributed.AreductioninOFV(DOFV)of3.84betweentwo
nestedmodelswithone parameterdifference wasconsidereda statisticallysignificantdifferenceatthe5%significancelevel,with a p-value of 0.05. Model selection was guided by statistical goodness-of-fit,graphicalevaluationofresidualdiagnosticplots, and simulation-based prediction-corrected visual predictive checks (VPC)(Bergstrandet al.,2011), andplausible parameter estimateswithacceptableprecision(Nguyenetal.,2017).
Initially, a model was developed to describe the plasma samples,forwhichone-,two-,andthree-compartmentdisposition modelswereevaluated.Thereafter,the modelwas extended to includemicrodialysisobservationsusingmethodologydescribed
previously(Tunbladetal.,2004).Theobservationsweretreatedas obtained,i.e.,asdialysateconcentrationattheendofthecollection intervalaftercorrectionforrecovery.Duringmodeldevelopment, thecentral,peripheral,oraseparatecompartmentwasevaluated toreflectthecollectionsiteofthemicrodialysisdata.Todescribe thevariabilitybetweenpatients,inter-individualvariability(IIV) terms were included if significant, assuming log-normally distributedparameters.
To explain some of the observed random variability, the following patient characteristics were evaluated in a covariate analysis:age,weight,andserumalbumin.Covariateselectionwas guidedbygraphicalevaluationandtheLRT,withprimaryfocuson the clearance parameter. Since piperacillin is predominantly eliminatedrenally(about70%)(Andersenetal.,2018),thereisa physiological relationship between creatinine clearance and piperacillinclearance. However,sincethepatientsinthis study were anuric and on dialysis,it was not possibleto evaluate a covariateeffectofcreatinineclearanceonelimination.Clearance andvolumeparameterswerescaledtobodyweightinlinewith allometry, whereweightwas normalized to70 kgfora typical adultandscaledusingafixedexponentof0.75forclearanceand1 forvolume.Categoricalcovariateswereincludedasashiftinthe typical value from the most common category. The covariate search was performed by stepwise covariate model building supported by PsN, starting with forward addition followed by backward elimination, with a significant p-value of 0.05 for forwardadditionand0.01forbackwardelimination(Jonssonand Karlsson,1998).
Simulations
Alternativedosingregimensandmodesofadministrationwere assessedatsteadystate(definedasthethirddayofdosing)with respecttothePTAforthethreePK/PDtargets. Piperacillindaily dosesof16,12,and8gwereassessed,administeredbycontinuous infusion(CI),extendedinfusion(EI),orintermittentadministration (IA).TheEIdosingregimenswere4gq6h(infusionover3h),4g q8h(infusionover4h),and4gq12h(infusionover6h).TheIA dosingregimenswere4gq6h,4gq8h,4gq12h,2gq4h,and2g q6h.Apopulationof100000individualsweresimulatedforeach regimen with the final model, with values for any identified covariate sampled from the observed values in the study population,oracorrespondingpopulationdistribution.
Results
Patientcharacteristicsandpiperacillinconcentrations
Tenpatientswereincludedinthestudy;theircharacteristics are summarized in Table 1. No microdialysis or piperacillin/
tazobactamadverseeventswereobserved.Relativerecoverycould notbedeterminedforoneofthemicrodialysiscatheters.Sincethe dialysatepiperacillinconcentrationsfromthiscatheterresembled those of theother catheters,the mean valueof theremaining relativerecoverieswasappliedforthiscatheter.Relativerecovery (meanstandarddeviation(SD))was0.230.12,andtroughfree piperacillinconcentrationswere76.2933.42mg/mlinplasma
and65.0843.27mg/mlinSCT.
Pharmacokineticmodelling
Ingeneral,freepiperacillininSCTwaslowerthanthatobserved inplasma,asshowninFigure1.Theparameterestimatesofthe final PKmodel,inconjunctionwiththeuncertainties andinter- individualvariabilityoftheparameters,aresummarizedinTable2.
Theplasmasamplesweredescribedbyatwo-compartmentmodel
Table1
Patientbaselinecharacteristics(n=10).a Patient Age
(years)
Sex Weight (kg)
AC Admission
diagnosis
Urine output
Microbiologyb Creatinine (mmol/l)
Bilirubin (mmol/l)
Albumin (g/l)
Urea (mmol/l)
SAPS3 Prescribed effluentdose (ml/kg/h)c
1 81 F 65 Citrate Pneumonia Anuria Klebsiella
pneumoniae1
131 9 30 6.6 54 45
2 71 M 86 Citrate PIVSD Anuria NF 259 20 33 11.8 65 37
3 78 F 74 Heparin Pulmonary
embolism
Anuria NF 85 22 22 14.3 73 31
4 70 M 98 Citrate Sepsis Anuria Enterococcusfaecalis2 244 21 27 12.8 68 40
5 72 M 60 Citrate Cardiac
arrest
Anuria NF 103 35 19 5.8 70 55
6 65 M 91 Citrate Aorta
dissection
Anuria Staphylococcusaureus2 191 334 26 16.2 65 37
7 62 F 58 Citrate Pneumonia Anuria InfluenzaAvirus3 109 41 15 7.5 - 38
8 80 M 70 Citrate Pneumonia Anuria InfluenzaAvirus4 140 15 26 9.2 91 44
9 46 M 84 Citrate Aorta
dissection
Anuria NF 649 9 33 24.1 51 40
10 64 F 84 Citrate Cardiac
arrest
Anuria InfluenzaBvirus3 186 10 35 16.4 91 34
Median 70.5 79 163 20.5 26.5 12.3 68 39
AC,anticoagulant;SAPS3,SimplifiedAcutePhysiologyScoreIII;F,female;M,male;PIVSD,postinfarctventricleseptumdefect;NF,notfound.
aAlldatawereregisteredonday1.
b Theoriginofthepathogensisindicatedbysuperscriptnumbers:1,pleuraleffusion;2,trachealsecretion;3,respiratorysecretion;4,throatswab.
cCRRTwasnotinterruptedduringthestudyperiodinanyofthepatients(prescribeddose=delivereddose).
Figure1.Overviewofthefreemicrodialysisconcentrations(closedcircles)andplasmaconcentrations(opencircles)availableforeachofthe10individualsincluded.The predictedfreeconcentration–timecoursesfromthemodelareshownforbothplasma(darksolidlines)andsubcutis(lightsolidlines)compartments.Theobserved microdialysisconcentrations(theaverageconcentrationoverthecollectioninterval)areplottedatthemid-pointofthesamplingintervalforillustration.Bodyweightandage aregivenforeachpatient.
Table2
Finalparameterestimatesandvariancesfromthepopulationpharmacokineticmodellinganalysis,includingparameteruncertaintyandshrinkageininter-individualand residualvariability.
Parameter Parameterdescription Estimate (RSE%)[SHR%]
CL(l/h) Eliminationclearance 3.30 (3.9)
Vc(l) Centralvolumeofdistribution 6.77 (8.3)
Q(l/h) Inter-compartmentalclearance 15.4 (21)
Vp(l) Peripheralvolumeofdistribution 10.0 (9.2)
fpl,sc Scalingfactorbetweenplasmaandsubcutis 0.661 (11)
CV%CL Inter-individualvariabilityinCL 10.9% (25)[6.6]
CV%Q Inter-individualvariabilityinQ 54.8% (29)[9.1]
CV%Vp Inter-individualvariabilityinVp 26.0% (33)[12]
CV%fpl,sc Inter-individualvariabilityinfpl,sc 35.7% (17)[0]
ERR(%) Proportionalresidualerror 16.6% (15)[7.9]
CV%,coefficientofvariationinpercent;RSE,relativestandarderrorinpercent;SHR,shrinkageinpercent(onstandarddeviationscale).