1. Pk/Pd modelling : Clinical Implications
JW Mouton
Dept Medical Microbiology
Canisius Wilhelmina Hospital
Nijmegen, The Netherlands

2. infections are treated with the same
dosing regimen irrespective of the absolute
susceptibility of the micro-organism

3. Susceptible =
MIC = .016 mg/L
MIC = 2.0 mg/L

4. Severity Infection Immunesystem Pharmacokinetics Pharmacodynamics
conc. vs time
Conc.
Time 25
0.0
0.4
Pharmacodynamics
conc. vs effect 1 10 -4 -3
Conc (log)
Effect
PK/PD
effect vs time
Time
Effect 1 25
Severity Infection
Immunesystem

5. Predictors of Clinical Response
Forrest, 1993

6. AUIC =125 : a magical number??
PK models : 125
Animal studies granulocytopenic : 100
Animal studies immunocompetent : 40
Human studies: peak/MIC ratio: > 1:10, AUC/MIC

7. There is a clear relation between pharmacodynamic index and effect.
At some concentration there is a maximum effect
Ideally, a dose should be chosen to obtain a maximum effect

8. Example Target Controlled Dosing Ixacin
Patient 60 yr, pneumonia and suspected bacteraemia/sepsis
Ixacin 400 mg IV q8h
Gram negative rod, E-test MIC=0.01 mg/L
Adjust dose to 100 mg/day
Mouton & Vinks, PW 134:816

9. Use of Pharmacodynamics in setting breakpoints
Known concentration effect relationship
microbiology
clinical effect
Standard dose used clinically
pharmacokinetic profile
patient/disease specifics

11. Breakpoints based on AUC/MIC =100

12. Quinolones : dose optimalization
Three studies have shown AUC/MIC predictive for outcome
One prospective study Peak/MIC more predictive
To peak or not to Peak?

13. For most quinolones, because of the relatively long half-life, there is a strong correlation between AUC/MIC and Peak/MIC

14. Survival linked to Peak/MIC when ratio > 10/1
Pharmacodynamics of fluoroquinolones against Pseudomonas infection in neutropenic rats Drusano et.al., AAC, 1993, 37:
Survival linked to Peak/MIC when
ratio > 10/1
Survival linked to AUC/MIC when
ratio < 10/1

15. ! ! ! Proteinbinding in animals/humans Duration of treatment
Extended dosing interval
Immune system
Severity of infection
Bacteriological vs Clinical cure

16. T>MIC: how long?? breakpoints
Free fractions of the drug (Fu)?
The same for all micro-organisms?
The same for all beta-lactams?
Efficacy vs emergence of resistance
The same for all infections?
Value in combination therapy?
Variance pk in population?
Static dose vs maximum effect?
If maximum effect, which max effect?

17. Same all beta-lactams? Same all m.o.?
T> MIC for static effect
Drug Enterobacteriaceae S. pneumoniae
Ceftriaxone (F) 38 (34-42) (37-41)
Cefotaxime (36-40) (36-40)
Ceftazidime (27-42) (35-42)
Cefpirome (29-40) (33-39)
MK (20-39)
Meropenem 22 (18-28)
Imipenem 24 (17-28)
Linezolid (33-59)

18. Predicted efficacy of ticarcillin and tobramycin
pseudomonas
Mouton ea., aac 1999

19. Concentration-time profile of beta-lactam
Vd = 20 L, Ka = 1.2 h-1, Ke = 0.3 h-1

20. Monte Carlo Simulation of beta-lactam
Vd = 20 L, Ka = 1.2 h-1, Ke = 0.3 h-1, VC=20% 4h
10h
Mouton, Int J Antimicrob Agents april 2002

21. Monte Carlo Simulations in pk/pd (1)
Estimates of pk parameter values and a measures of dispersion (usually SD)
Simulate pk curves based here-on

22. Target Attainment Rates
Pharmacokinetics of piperacillin
Population pk using npem2
‘validation’using classical pk analysis (winnonlin)
Use parameter estimates for Monte Carlo Simulation
Using sd’s only
Using correlation matrix
Obtain TARS at various T>MICs

24. Monte Carlo Simulations in pk/pd (2)
Estimates of pk parameter values and a measures of dispersion (usually SD)
Obtain covariance matrix or correlation matrix
Simulate pk curves based here-on :
The result will be SMALLER 95% CI because of correlation between parameter estimates

26. Needed for TAR Pharmacokinetic profile of the drug
Variation of PK parameters
Estimates of VC
Population analysis, correlation matrix

27. Cumulative Fraction of Response
TAR at each MIC
Distribution of MICs
Multiply, determine fraction and cum fraction
Beware !!!!! For prediction, a ‘true’ MIC distribution is needed, NOT a biased collection of strains!!!!!

28. After Drusano et al

29. Pharmacodynamic properties of beta-lactams
Efficacy dependent on time above mic
Relatively concentration independent
Target for concentration profiles optimizing time > mic
Avoid high peak concentrations

30. Target Concentration continuous infusion
Maximum effect time-kill at 4 x MIC
Maximum effect in vitro model 4 x MIC (Mouton et al 1994)
Effect in endocarditis model 4 x MIC (Xiong et al 1994)
Effect in pneumonia model dependent on severity of infection (Roosendaal et al 1985,86)

31. Efficacy in Relation to MIC
continuous infusion ceftazidime, K. pneumoniae rat pneumonia
Roosendaal et al. 1986, AAC 30:403; Roosendaal et al. 1985, JID 152:373

32. Continuous Infusion Pharmacokinetic Considerations
Protein binding
Linear relationship between clearance and dose
Linear relationship between protein binding and dose
Third compartment effects (CNS)

33. Dose Calculations continuous infusion
Total Clearance estimate
Volume of Distribution
Elimination rate constant
TBC = Ke x Vd

34. Normogram Continuous Infusion
Mouton & Vinks, JAC 1996

35. Example Target Controlled Dosing cefticostix
Patient 60 yr, UTI and suspected bacteraemia/sepsis
Cefticostix 1 g IV q8h
Gram negative rod, E-test MIC=0.12 mg/L
Adjust dose to 30 mg/day CI based on patient clearance
Mouton & Vinks, PW 134:816

36. Example Cost Efficiency
Continuous Infusion Beta-lactam

37. Therapeutic Drug Monitoring continuous infusion
Clearance estimate
Variable clearance (ICU)
Non-linear clearance

38. Ceftazidime concentrations ICU patients

39. Piperacillin concentrations continuous vs intermittent