Ottawa Juin 2004 - 1

1
Population PK/PD and the rational design of an
antimicrobial dosage regimen in veterinary
medicine
UMR 181 Physiopathologie Toxicologie
Expérimentales

• Pierre-Louis Toutain
• AAVM Congress - Ottawa June 2004

2
Co-workers

• Academia
• Horse study
• A. Bousquet-Mélou
• M. Doucet
• D. Concordet
• M. Peyrou
• Pig study
• J. del Castillo
• V. Laroute
• D. Concordet
• P. Sanders
• M. Laurentie
• H. Morvan

• Industry
• Horse study
• Vetoquinol (France)
• M. Schneider
• Pig study
• SOGEVAL (France)
• C. Zemirline
• P. Pomie
• VIRBAC (France)
• E. Bousquet
• INTERVET (germany)
• E Thomas

3

• "The design of appropriate dosage regimens may be
  the single most important contribution of
  clinical pharmacology to the resistance problem"
• Schentag et al. Annals of Pharmacotherapy, 30
  1029-1031

4
Dosage regimen and prevention of resistance

• Many factors can contribute to the development of
  bacteria resistance
• the most important risk factor is repeated
  exposure to suboptimal antibiotic concentrations
• dosage regimen should minimize the likelihood of
  exposing pathogens to sublethal drug levels

5
Ranking (Low, Medium, High) of extent of
antibiotic drug use in animal based on duration
and method of administration

• Individual Groups or pens Flocks, herds
• animal of animal of animals
• Duration
• Short lt6 days L M H
• Medium 6-21 d L M H
• Long gt21 days M H H

6
What is the contribution of the kineticist to the
prudent use of antibiotics

• To assist the clinicians designing an optimal
  dosage regimen
• To ensure that the selected antibiotic reach the
  site of infection at an appropriate effective
  concentration, for an adequate duration and for
  all (or most) animals under treatment to
  guarantee a cure (clinical, bacteriological) and
  without favoring antibioresistance

7
The application of population pharmacokinetic
modelling to optimize antibiotic therapy
8
How to ensure that a dosage regimen minimizes the
likelihood of exposing pathogens to sub-clinical
drug levels

• Individual animals
• groups or pens vs flocks/herds
• ? population approach

9
Reminder

• Traditional vs populational PK/PD approaches
• What is PK/PD for antibiotics and how to
  determine a dosage regimen using PK/PD predictors
• see P. Lees presentation

10
Traditional veterinary PK

• Study performed in experimental setting
• elaborate design
• limited number of animals
• rich data
• Data analysis two stages
• 1- modelling individuals ? samples of individual
  estimates
• Cl, Vss, F, t1/2
• 2- statistical analysis
• mean - SD
• search for difference between subgroups (ANOVA),
  for associations (regression)

11
Limits of traditional PK

• Experimental conditions
• may be not representative of the real world
• consider variability as a nuisance
• Data analysis
• variance and covariance often badly estimated and
  explained
• Solution the population approach

12
How to determine a dosage regimen using PK/PD
predictors
13
Dose titration
Dose
Response
Black box
PK/PD
14
The main goal of a PK/PD trial in veterinary
pharmacology

• To be an alternative to dose-titration studies to
  discover an optimal dosage regimen (will be
  presented by P. Lees)

15
Contributions of the PK/PD approach to the
population determination of a dosage regimen

• The separation of PK and PD variabilities

16
PK/PD variabilities for antibiotics

• Consequence for dosage determination

PK
PD
Effect
BODY
Pathogens
Dose
Plasma concentration

• Physiological/constitutional variables
• Breed, sex, age
• Kidney function
• Liver function...

• Clinical covariables
• pathogens susceptibility (MIC)
• disease severity or duration

PK/PD population approach
17
PK/PD predictors of efficacy
Cmax
Concentrations
MIC
Time
24h
18
AUIC an attempt to combine PK and PD properties
of antibiotics
Capacity to eliminate the drug
PK
Dose / Clearance MIC90 or MIC50
AUC MIC

• AUIC
  critical breakpoint value

• Fixed endpoint related to Emax and EC50

PD
Application fluoroquinolones
19
Computation of dose using a PK/PD predictor

• Dose x x
  Clearance (24h)

PD
Breakpoint to be achieved
AUIC 24h
MIC fu x F
bioavailability
Free fraction
PK
20
Computation of dose using a PK/PD predictor

• Dose x x
  Clearance

MIC50 average
PD
Breakpoint to be achieved
PK
AUIC 24h
MIC F
average
(pop)
PK
21
Dispersion of variance around the mean may be the
most relevant parameter to predict a population
dosage regimen for antibiotics
22
Variability and the likelihood of resistance
F
Resistance pathogens of interest
23
Variability and the likelihood of resistance
Resistance zoonotic, commensal
Resistance pathogens of interest
24
Examples of population approaches for antibiotics
in veterinary medicine

• Identification and explanation of PK variability
• marbofloxacin in horse
• Determining drug PK characteristics in tissues
  using sparse sampling
• marbofoxacin in ocular fluid in dog
• Dosage regimen determining
• doxycyclin in pig

25
Marbofloxacin in horses

• A. Bousquet-Mélou et al.

26
Marbofloxacin in horses PK

• A fluoroquinolone
• No marketing authorization in horses
• Conventional PK study
• data analysis using the two-stage approach
• clearance 4.15 0.75 mL/kg/min CV 18
• Vss 1.48 0.3 L/kg
• t1/2 7.56 1.99 h

27
Marbofloxacin in horses PK/PD integration (oral
route)

• Value of efficacy index (AUIC24h) and Cmax/MIC
  calculated from PK parameters obtained after the
  administration of 2 mg/kg BW in 6 horses
• MIC90 0.027 µg/mL (enterobacteriaceae)
• average PK/PD index

AUIC24h 155 21 Cmax/MIC 31 4.5
28
Population PK approach for marbofloxacin in
horses objective

• To measure the interindividual variability of
  systemic exposure to marbofloxacin in horses
• To identify covariates explaining a part of this
  variability

Body clearance
The only determinant of AUC
29
Materials and Methods (1)

• Animals
• patients from the Equine Clinic of the
  Veterinary School
• healthy horses from the Riding School
• Covariates record
• demographic, physiological, disease
• not all covariates presented
• IV administration of marbofloxacin (2 mg.kg-1)
• Nonlinear mixed-effects modelling
• Kinepop software (D. Concordet)

30
Materials and Methods (2)
Sampling design selection

• Number of samples per animal and selection of
  sampling times
• D - optimal design to maximize the precision of
  AUC 0-24h
• previous informations AUC0-24h Mean and
  Standard Deviation
• Bousquet-Melou et al., Equine Vet J, 34, 2002

AUC imprecision
Sampling windows 30min windows centred
around 1.5, 3, 5, 7 and 19.5 h post-administration

Sampling design
31
Materials and Methods (3)

• PK model - biexponential equation
• - parameterisation in volumes of distribution
  and clearances

• Statistical model - lognormal distribution of
  PK parameters

Model 1 no covariate
Model 2 with covariates for body clearance
32
Results conventional vs pop kinetics

• 52 horses, 253 blood samples

10
1
Marbofloxacin (mg/mL)
0.1
0.01
0.001
0
4
8
12
16
20
24
Time (h)
33
Variability model without covariable
34
Variability model with covariables
Without covariable
With covariables
predicted concentrations (mg/mL)
observed concentrations (mg/mL)
35
Variability explicative covariable
Covariables for body clearance expressed in
L.kg-1.h-1
Note dose was 2 mg/kg BW i.e. already scaled to
BW
36
Marbofloxacin the body weight is a covariable
Allometric relationship with an allometric
exponent gt1
37
Discussion

• Marbofloxacin clearance in horses

Population trial
Classical trials
0.233
Mean (L.kg-1.h-1)
0.19 - 0.246
50
CV ()
18 - 21
Carretero et al., Equine Vet J, 34, 2002
Bousquet-Melou et al., Equine Vet J, 34, 2002

• Influence of body weight

In the range of observed weights about 3-fold
variation in body clearance expressed per kilogram
38
Conclusion

• High interindividual variability of marbofloxacin
  body clearance in horses
• Underestimated in classical PK trials
• Influence of body weight
• Consequences on systemic exposure
• Clinical relevance for efficacy and resistance ?
• Current trial
• Multicentric experiment (Montreal, Toulouse,
  Utrecht, Vienna)
• Increased number of covariates
• Further trials
• Assessment of variability of PD origin

39
Population PK/PD determination of a dosage
regimen for an antiobiotic
40
Objectives

• Document, with population PK/PD approach, the
  dosage regimen for antibiotics in pig
• Ultimate goal make recommendations
• to determine a dosage regimen
• to establish MIC breakpoints
• to establish PK/PD predictor breakpoints

41
Population trial (INRA/SOGEVAL/CTPA)J. del
Castillo et al.

• Antibiotic doxycyclin
• Britain (2 settings)
• 215 pigs (30 to 110 kg BW)
• oral (soup)
• pens of 12-15 pigs (unit of treatment)

42
Population trial

• Decision of treatment metaphylaxis
• prevalence of diseasegt10 (tachypnee, body
  temperature gt 40C)
• Treatments
• Doxycyclin (5 mg/kg) or
• Doxycyclin paracetamol (15 mg/kg)
• 2 meals apart from 24h
• Measure of covariables (rectal temperature
  /clinical signs etc.)
• Blood samplings (4 or 5 after the 2nd dose)
• Dosage HPLC (doxy, paracetamolmetabolite)

43
PK Variability
Doxycycline
n 215
44
PK doxycyclin variability analysis
45
Doxycycline sex effect
Sexe 0 Sexe 1
Doxycycline
Time (h)
46
Doxycycline body temperature effect
Doxycycline
Rectal temperature
47
Doxycycline disease effect
healthy diseased
Concentrations (µg/mL)
Time (h)
48
Variability analysis AUC vs. body weight
49
How to make use of PK/PD population knowledge to
predict how well will doxycyclin perform
clinically?
50
The use of MonteCarlo simulation

• Dose selection at the population level
• Determination of breakpoints
• PK/PD
• MIC

51
Material and Method

• PK/PD analysis was performed using Monte Carlo
  simulations
• The method accounts for the variability in PK as
  well as MIC data to determine the probability of
  reaching a target AUC0-24/MIC ratio

52
Data analysis

• PK non linear mixed effect model
• seek to explain the variability by covariables
• Computation of AUC and statistical establishment
  of distribution
• PK/PD MonteCarlo approach to assess the
  distribution of the PK/PD endpoint

53
Dosage regimen application of PK/PD concepts
The 2 sources of variability PK and PD
PK exposure
PD MIC
Distribution of PK/PD surrogates
(AUC/MIC) Monte-Carlo approach
54
AUC distribution
Under-exposure ?
55
Microbiological dataIntervet, Virbac, AFSSA

• Streptococcus suis (n180)
• Actinobacillus pleuropneumoniae (n110)
• Pasteurella multocida (n206)
• Haemophilus (n25)

56
MIC distribution Actinobacillus
pleuropneumoniae (n106)
40
35
30
25
Pathogens
20
INTERMEDIATE
15
10
SUSCEPTIBLE
RESISTANT
5
0
0.25
0.5
1
2
4
8
MIC
(µg/mL)
57
MIC distributionPasteurella multocida (n205)
40
35
30
Pathogens
25
20
15
10
SUSCEPTIBLE
5
0
0.0625
0.125
0.25
0.5
1
2
4
MIC (
m
g/mL)
58
MIC distributionStreptococcus suis (n180)
Bimodal distribution
35
30
25
Pathogens
INTERMEDIATE
RESIST.
20
SUSCEPTIBLE
15
10
5
0
0.0313
0.0625
0.125
0.5
1
2
4
8
16
32
CMI (
g/mL)
m
59
Statistical distribution of PK/PD predictors

• Question what is the percentage of a pig
  population to achieve a given value of the PK/PD
  predictor for a given dose of doxycyclin for a
• Empirical (initial) antibiotherapy (pathogen
  known, MIC unknown but distribution known)
• Targeted antibiotherapy (MIC known)

60
Doctor or Regulator

• In clinical therapy, we would like to give
  optimal dose to each individual patient for the
  particular disease
• individualized therapy (targeted antibiotherapy)
• In new drug assessment / development, we would
  like to know the overall probability for a
  population of an appropriate response to a given
  drug and proposed regimen
• population-based recommendations (empirical
  antibiotherapy)

H. Sun, ISAP-FDA workshop 1999
61
Population PK/PD applications

• Individualisation ? doctor
• Recommandation ? regulator

62
Doxycycline (5 mg/kg) empirical vs targeted
antibiotherapy for Pasteurella multocida
Empirical antibiotherapy
100
Targeted antibiotherapy (MIC 0.25 µg/mL)
80
60
of pigs above the breakpoint
40
20
0
0
24
48
72
96
120
144
168
192
Breakpoint to be achieved (AUC/MIC) (h)
bacteriostatic
63
Doxycycline (5 mg/kg) empirical vs targeted
antibiotherapy for Actinobacillus pleuropneumoniae
100
Empirical (MIC unknown)
80
Targeted (MIC 0.5 µg/mL)
60
of pigs above the breakpoints
40
20
Breakpoint to be achieved (AUC/MIC) (h)
0
0
24
48
72
Bacteriostatic
64
Doxycycline (5 mg/kg) empirical vs targeted
antibiotherapy for Streptococcus suis
100
80
Empirical antibiotherapy
Targeted antibiotherapy (MIC 16 µg/mL)
60
of pigs above the breakpoint
40
20
0
0
24
48
72
96
120
144
168
192
Breakpoint to be achieved (AUC/MIC) (h)
bacteriostatic
65
Population dose determination

• Question what is the doxycycline dose to be
  administered to achieve a given AUC/MIC ratio for
  a given percentage of the pig population ? (e.g.
  90)

66
Doxycycline selection of an empirical (initial)
dose for Pasteurella multocida
Doses
100
5 mg/kg
90
80
10 mg/kg
20 mg/kg
60
of pigs above a given AUC/MIC ratio
40
20
0
0
24
48
72
96
120
144
168
bacteriostatic
AUC/MIC ratio (h)
67
Doxycycline selection of an empirical (initial)
dose for Actinobacillus pleuropneumoniae
Doses
100
5mg/kg
10 mg/kg
80
20 mg/kg
60
of pigs above a given AUC/MIC ratio
40
20
0
0
24
48
72
bacteriostatic
AUC/CMI ratio (h)
68
Doxycycline selection of an empirical (initial)
dose for Streptococcus suis
Doses
100
5 mg/kg
80
10 mg/kg
20 mg/kg
60
of pigs above a given AUC/MIC ratio
40
20
0
0
24
48
72
96
120
144
168
AUC/MIC ratio (h)
69
Determination of MIC breakpoints by standard
developing organizations using population approach
70
Determination of MIC breakpoints

• Current situation
• PK information is badly taken into account
• population approach

71
Determination (or revision) of the clinical MIC
breakpoint for a given drug against a given
pathogen

• Dose fixed (marketing authorization)
• breakpoint to achieve determined
• TgtMIC gt80 of the dosage interval
• or AUC/MIC 100h
• computation of the critical MIC value for which
  TgtMIC (or other PK/PD indices) are in excess of
  90 (or other ) of subjects.

72
Doxycycline (5 mg/kg) MIC breakpoint for
Actinobacillus pleuropneumoniae to achieve a
given AUC/MIC ratio for 90 of pig
MIC 0.0625 µg/mL
100
MIC 0.125 µg/mL
90
80
MIC 0.25 µg/mL
60
of pigs above the breakpoint
40
20
0
0
24
48
72
96
120
144
168
192
216
240
bacteriostatic
Breakpoint AUC/MIC (h)
73
Doxycycline (5 mg/kg) MIC breakpoint for
Streptococcus suis to achieve a given AUC/MIC
ratio
100
MIC 0.5µg/mL
90
MIC 0.125 µg/mL
80
MIC 0.0625 µg/mL
60
of pigs above a given AUC/MIC ratio
40
20
0
0
24
48
72
96
120
144
168
192
Bacteriostatic
Breakpoint AUC/CMI (h)
74
Doxycycline(5 mg/kg) MIC breakpoints for
Pasteurella multocida to achieve a given AUC/MIC
ratio
MIC 0.0625 µg/mL
MIC 0.125 µg/mL
MIC 0.25 µg/mL
100
90
80
60
de pc avec une AUC/CMIgt seuil
40
20
0
0
24
48
72
96
120
144
168
192
AUC/MIC ratio (h)
Bacteriostatic
75
Determination of PK/PD predictor breakpoints

• For drug dosage prediction, not only PK/PD index
  that determine the effect but also its magnitude
  must be determined
• Prospective or retrospective approach using
  clinical data

76
Conclusion

• For practitioners
• to adjust the dosage regimen for a given
  animal (or a given breed)
• flexible dosage regimen
• For drug companies and authorities
• a general framework to propose an empirical
  (initial) dosage regimen
• For standards-developing organizations
• MIC breakpoints

77
Experimental vs population studies
78
Experimental
Population
79
Experimental vs. population approach

• Two questions regarding experimental approach
• What is its validity (clinical relevance)
• What about variability

80
Drug administration, social behavior and the dose

• Experimental
• Individually controlled by the investigator
  (restricted, tubing)
• The nominal dose is guaranteed to all individuals

• Field
• related to individual feeding behavior (fever,
  anorexia)
• group effect (hierarchy, dominance) or other
  behavior
• Dose actually ingested can be much higher or
  much lower than the nominal dose

81
The pathology

• Experimental Field
• Standardised experimental Spontaneous disease
• infectious model

82
Animal selection

• Experimental
• Highly selected (as homogeneous as possible) body
  weight, sex, age...

• Population
• Representative of the target population different
  breed, age, pathological conditions

83
Study design

• Experimental
• experimental, restrictive
• artificial (temperature, light)

• Population
• Observational
• natural (e.g. field)

• Difference
• Power,
• inference space
• interaction with environment behavior

84
Experimental vs population approachthe status
of variability

• Experimental
• viewed as a nuisance that has to be overcome

• Population
• recognized as an important feature that should be
  identified, measured and explained (covariables)

85
Experimental vs population approachAccuracy and
variability

• In current experimental practices, major
  determinant of drug disposition (PK) or of drug
  effect (PD) can be modified, altered or
  suppressed
• GLP is not synonymous to good science

!
86
Advantage of field population kinetics over
classical experimental setting

• Experimental environment
• healthy animals selected for homogeneity
  inter-individual variability is viewed as a
  nuisance
• conditions rigidly standardized
• artificial conditions

• Real world / clinical setting
• patients representative of target population
• variability (inter intra-individual,
  inter-occasion) is an important feature that
  should be identified and measured
• seek for explaining variability by identifying
  factors of demographic pathophysiology

87
Doxycycline concentration variability population
vs experimental trial
1.5
1.0
Number of data points Trial Population
n215 Experimental n15 to 19
DOXYCYCLINE (µg/mL)
0.5
0.0
0
4
6
12
24
Time (h)
88
Doxycycline concentration variability population
vs experimental trial for time 6h
post-administration
1.5
Number of data points 1 Population n215 2
Experimental n16 3 Experimental n64
1.0
DOXYCYCLINE (µg/mL)
0.5
0.0
2
3
0
1