1 exercise 14 dose titration and selection of a therapeutic dose for a new quinolone in pigs

41
1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

Upload: isai-willford

Post on 28-Mar-2015

224 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

1

Exercise 14Dose titration and selection of a therapeutic dose for a

new quinolone in pigs

Page 2: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

2

Objectives of the exercise

1. To y understand the limits of the classical dose titration to select a dose.

2. To understand how to determine a dose for an antibiotics using PK/PD concepts

3. To analyse qualitative clinical results using a logistic model to compute a probability of cure (POC)

4. From a dose titration trial, to be able to establish a breakpoint value for the selected PK/PD index that guarantee a POC of 90% in pigs.

5. To understand the concept of Target Attainement Rate ( TAR)6. To select a dosage regimen to achieve a TAR of 90% for the

established PK/PD index value in the pigs population7. To perform a sensitivity analysis to identify the most critical factors

of variability (PK or PD) influencing the TAR.

Page 3: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

3

Designs for dose titration and the determination of an

effective dose

Page 4: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

4

The simple parallel design

Page 5: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

5

The parallel designThe parallel design

• The null hypothesis– placebo = D1 = D2 = D3

• The statistical linear model– Yj = wj + j

• Conclusion– D3 = D2 > D1 > Placebo

Placebo Dose

Response

1 2 3

*

*

NS

Selected dose

Page 6: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

6

The parallel designThe parallel design

orDose (log)

Re

sp

on

se

Dose (log)

Re

sp

on

se

Average population

Dose (log)

Res

po

nse

Page 7: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

7

An alternative to the classical dose titration for the selection of a rational dosage regimens

for an antibiotic is the so-called PK/PD approach

Page 8: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

8

Three PK/PD indices are used to predict clinical

and bacteriological outcomes

• For quinolone, – AUC/MIC is the appropriate index– Its numerical values (or breakpoint value)

that is predictive of a good clinical efficacy is typically of 125h;

• it is equivalent to say that the goal is to find a dose achieving an avrage plasma concentration over 24h that is 5 times higher than the targeted MIC.

Page 9: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

9

The first task of the project

• we are developing a new antibiotic for metaphylaxis (control) in pigs,

• The first task will be to establish, experimentaly, the breakpoint value of AUC/MIC that is predictive of a probability of cure rate (POC) of 90% in pig.

Page 10: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

10

Step 1: Establishing the breakpoint value of AUC/MIC for the new antibiotic that is predictive of a probability of cure rate (POC) of 90% in pig

Page 11: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

11

Design of the trial

• A classical parallel design with 4 dose levels (Dose0, Dose1, Dose 2 & Dose3)

• An infectious challenge with a pathogen having a given MIC

• blood samples should be collected to establish individual AUCs

• Clinical endpoint : cure or not cure

Page 12: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

12

Data analysis

• Possibility to establish a conventional dose titration

• Model the relationship between the measured AUCs scaled by the MIC of the pathogen actually used for the infectious challenge and the porobability of cure (POC); – i.e to consider AUC/MIC as generic explicative

variable and the POC as the dependent variable;

Page 13: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

13

Individual results obtained in 40 pigs during a dose-titration trial.

Page 14: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

14

Logistic regression

• When an all-or-none effect is observed due to the selected binary endpoint (cured or not cured for antibiotics), the concentration-response curve is described by a quantal dose-response curve

Page 15: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

15

Logistic regression

This logistic model represents the probability at which a concentration (or AUC/MIC) of a drug produces the all-or-none effect

Page 16: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

16

Logistic vs. Emax model

• Despite the fact that the graph of a S-shaped model looks like the one of an Emax model, a quantal dose-response curve differs of the Emax model because it does not relate dose to intensity of effect but to the probability an effect.

Page 17: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

17

A logistic model

where POC is the probability of cure (from 0 to 1)θ1 is a parameter for baseline (here a placebo effect)θ2 is an index of sensitivity (a slope).X is the independent variable i.e. AUC/MIC ratio

any other quantitative index predicting drug efficacy can be used as X

XePOC

211

1

Page 18: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

18

A logistic model

• If θ2 equal 0, the response is independent of X and the POC correspond to the placebo effect

XePOC

211

1

Page 19: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

19

A logistic model

• For a logistiv curve, the equivalent of the EC50 (or ED50) is named EL50;– The EL50 can be estimated by the ratio –θ1/θ2,

• it is now the median effective level of the independent variable (e.g. dose, AUC/MIC..) for which 50% of the subjects are cure,

• The slope of the curve now represents the dispersion (variance) of the threshold in a population.

XePOC

211

1

Page 20: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

20

Fit data with the following equation

MICAUCbaePOC

1

1

the placebo effect is given by 1/(1+ea), the POC 50% by an AUC/MIC ratio equal to a/b the POC 90% by a AUC/MIC ratio equal to (a+2.1972)/b.

Page 21: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

21

Enter your data and code it

Page 22: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

22

Build a user model of the logistic model • MODEL • remark ******************************************************• remark Developer: PL Toutain• remark Model Date: 04-01-2011• remark Model Version: 1.0• remark ******************************************************• remark• remark - define model-specific commands • COMMANDS • NFUNCTIONS 1• NPARAMETERS 2• PNAMES 'a', 'b'• NSECONDARY 3• SNAMES 'Placebo', 'POC50', 'POC90'• END• remark - define temporary variables• TEMPORARY • index=x• END • remark - define algebraic functions• FUNCTION 1• F= 1/(1+exp(a-b*x))• END• remark - define any secondary parameters• SECONDARY • Placebo=1/(1+exp(a))• POC50=a/b• POC90=(a+2.1972)/b• END • remark - end of model• EOM

Page 23: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

23

The command block

• COMMANDS

• NFUNCTIONS 1

• NPARAMETERS 2

• PNAMES 'a', 'b'

• NSECONDARY 3

• SNAMES 'Placebo', 'POC50', 'POC90'

• END

Page 24: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

24

The TEMPORARY block

• TEMPORARY

• index=x

• END

Page 25: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

25

The FUNCTION block

• FUNCTION 1

• F= 1/(1+exp(a-b*x))

• END

Page 26: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

26

The SECONDARY block

• SECONDARY

• Placebo=1/(1+exp(a))

• POC50=a/b

• POC90=(a+2.1972)/b

• END

Page 27: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

27

Run the model

Page 28: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

28

Estimated parameters

Page 29: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

29

Conclusion of the first step

• The placebo effect was estimated to 30% and the POC90% (our breakpoint value) to 109.77 rounded to 110h

• It should be stressed that what is determined at this step is not a dose for pivotal clinical trials but a target value for a PK/PD surrogate that can be used for predicting clinical efficacy

Page 30: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

30

Step 2: determination of a dose to achieved a Target

Attainement Rate of 90% for the breakpoint value of the

PK/PD index

Page 31: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

31

The selection of a dosage regimen

• The selection of a final dosage regimen intended for use across a population of animals needs to account for interanimal variability.

• We expect to find a dosage regimen able to achieve a TAR of 90% for the breakpoint value of the PK/PD index as determined at the preceeding step (110h) – i.e. to find the dose for which the breakpoint value of

the PK/PD index is actually achieved in at least 90% of the pigs in a population.

– For that we have to take into account the different factors of variabilty able to influence final dose

Page 32: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

32

Equation giving the daily dose

• Dose(per day) = the daily maintenance dose at steady state;

• CL = the plasma clearance (in mL/hour);– It is because the AUC/MIC target is expressed in hours per 24h that the clearance should be

expressed per hour and not per day to compute a daily dose with this equation (see attached reprint by Toutain et al explaining why)

• (AUC/MIC)target = the AUC/MIC breakpoint value to achieve (110 h in our example)

• MIC= a given MIC as MIC90 or any other MIC.

• fu = is the free fraction = the fraction of unbound drug in the plasma or serum (where values range from 0 to 1). For the sake of simplicity fu has been set at a constant value of 1 in this tutorial

• F = the absolute bioavailability (where values range from 0 to 1). Equation given the dose for a given AUC/MIC breakpoint ; here AUC/MIC was established in vivo and no correction is required but if you pick up the BP value of the PKPD indice from the litterature, you have to consider fu.

fuF

MICClearance

dayDose

%

MIC

AUC(mL/hour)

target

Page 33: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

33

Solve the equation

• With point estimates

• With distribution using MCS

Page 34: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

34

Stochastic determination of a dose to achieve a

TAR of 90% for the PK/PD index ;

Page 35: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

35

Build the excel sheet

Page 36: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

36

Numerical values to use for simulation

• CL values was assumed to be log-normally distributed with a geometric mean of 60 ml/kg/h and a GSD of 1.1,

• Bioavailability (F ) was assumed to conform to a uniform distribution i.e. all values had the same occurrence probability in the range of 0.3 to 0.7.

• MIC distribution had 5 classes (0.1µg/mL with P=0.1; 0.2µg/m with P=0.2; 0.5µg/mL with P=0.3; 1µg/ml with P=0.3 and 2µg/ml with P=0.2)

Page 37: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

37

Results

Page 38: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

38

The MCS and the sensitivity analysis

• A major advantage of the PK/PD framework is that it permits the separate evaluation of the principal sources of PK and PD variability.

• PK variability reflects interanimal differences for plasma clearance, bioavailability… while PD variability is mainly associated to differences in microbial susceptibility ( MICs of the pathogen).

Page 39: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

39

Our questionment

• For the dosage regimen previously calculated, it may be of interest to identify the impact of the various model factors on the final dose.

• Why the dose is so high?– is it due to the clearance variability (and I can

do nothing) or the erratic bioavailability (I can improve my formulation) or to the MICs (and I can recommend or not susceptibility testing)

Page 40: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

40

The MCS and the sensitivity analysis

• Sensitivity analysis is the study of how the variability in the model output (the computed dose) can be apportioned, quantitatively, to different sources of variation of the model input (here CL, F and MIC distribution).

Page 41: 1 Exercise 14 Dose titration and selection of a therapeutic dose for a new quinolone in pigs

41

Sensitivity chart