An Idea for a Mid to Long Term Strategy for Pradaxa

An Idea for a Mid to Long Term Strategy for Pradaxa

Background – Experience after the Pradaxa SPAF Launch

Confidential - Internal use only 2

• Patients on older OACs are accustomed to and many appear to like and expect frequent

interactions with the Health Care system and its staff

• Switches from older OACs to NOACs are uncommon in the marketplace

• Most new AF patients still are prescribed an older OAC

• Prescribers often want to know the extent of anticoagulation each patient is receiving via their

current anticoagulant

• External experts and the media have expressed concerns about a “one size fits all” OAC, since dose

adjusted OACs may have optimized outcomes

• Xarelto competes effectively against Pradaxa despite inferior ROCKET data

• Eliquis is perceived to have the best outcomes data of all NOACs

• FDA has indicated that only path towards approval of 110 mg dose would be Modeling &

Simulation results followed by a PK/PD study

Modeling % Simulation – „Study“ Design


Results

Distribution of Doses and Exposure

Dabigatran Dose (A) Reference (B) Titration

75 mg bid --- 25.5%

110 mg bid --- 29.9%

150 mg bid 100% 44.6%


• Majority of patients eligible for 150 mg bid

• Significant amount assigned to 75 mg bid

Ctrough,ss

N=5000 Median P10 – P90

(A) Reference 97.1 48.6 – 199.4

(B) Titration 77.0 48.6 – 101.6

• Shift in exposure

• Median Ctrough,ss ~21% reduced

• Minimum exposure levels maintained

• 90th percentile significantly reduced

Distribution of Dabigatran Doses Distribution of Dabigatran Exposure

Results

Model Predicted Outcome

Ischemic Stroke/SEE Major Bleeding

Mean* 90% CI$ Mean* 90% CI$

(A) Reference 1.26 1.01 – 1.55 4.38 3.91 – 4.89

(B) Titration 1.34 1.08 – 1.63 3.49 3.08 – 3.96


*risk of event within median RE-LY duration [~20 months], not annualized; $ Clopper-Pearson (Exact); §Range 10th percentile – 90th percentile

Titration vs. Reference

• Risk of ischemic stroke/SEE events comparable (Relative Risk 1.06)

• Risk of major bleeding events significantly reduced (Relative Risk 0.8)

Absolute Event Rates, not annualized

(B) Titration vs. (A) Reference Rel. Risk 90% CI

Ischemic Stroke/SEE 1.06 (0.76 – 1.50)

Major Bleeding 0.80 (0.66 – 0.97)

Relative Risk

Results

Comparison to Warfarin (observed RE-LY data)



(B) Titration (n=5000) 1.34 1.08 – 1.63 3.49 3.08 – 3.96

(C) Warfarin (n=4597) 1.68 1.38 - 2.02 5.83 5.27 - 6.43



(B) Titration vs. (C) Warfarin Rel. Risk 90% CI


Major Bleeding 0.60 (0.50 – 0.72)

Relative RiskAbsolute Event Rates, not annualized

Titration vs. Warfarin

• Only RE-LY warfarin patients from centers that contributed PK measurements were considered (center

matching)

• Risk of ischemic stroke/SEE events reduced (Relative Risk 0.8)


• Comparison needs to be handled with care. A potential bias can not be excluded due to the type of analysis.

COMPARISON OF OBSERVED DABIGATRAN VS. OBSERVED

WARFARIN OUTCOMES FROM RE-LY

Impact of Dabigatran Dose Adjustment on Clinical Outcome in AF Patients


Results – Clinical Outcome

“Dabigatran Titration vs Warfarin” – Safety Dataset


Treatment Titration

DE 110 mg 1081

DE 150 mg 2005

Warfarin 3961

Total 7047

Number of patients(DE110 + DE 150)

Optimally treated (=titrated) dabigatran patients reveal a promising efficacy and

safety profile compared to matched warfarin group.


“Titration vs RE-LY” – Safety Dataset


Treatment Titration RE-LY

DE 110 mg 1081 5983

DE 150 mg 2005 6059

Warfarin 3961 5998

Total 7047 18040

Number of patients

Titration patients have an efficacy profile that is comparable to the RE-LY DE 150 mg group and a safety profile

that is comparable to the RE-LY DE 110 mg group

Potential Mid to Long Term Strategy

• Discuss Modeling & Simulation data with FDA and explore potential path forward to approval of

such an approach:

• PK/PD study only acceptable?

• Trial with bleeding endpoints required?

• “Real world trial” using electronic health records of titration strategy as alternative? (e.g.

within regulated system such as KP which would be more feasible than “RE-LY 2” from a

budget and timeline perspective)

• Explore timing of such results compared to that of (1) globally available lab test, (2) a point of care

testing device, and (3) the Pradaxa antidote

If data acceptable to regulators: Re-Launch Pradaxa with titration strategy (resulting in significantly

fewer strokes and significantly fewer bleeding events compared to warfarin), Point of Care testing

device and availability of Antidote


Open Questions

• Would we exclude patients from Pradaxa use in AF because of insufficient exposure?

• If yes, which percentage of the overall population would be excluded?

• Given a ≈40% intra-individual variability of DE concentrations, can a simple titration regimen be designed

with a low need for multiple titrations?

• Will some of the assumptions change over time with more familiarity in the Marketplace on NOACs without

the need for titration?

• Which extent of new data would be needed to reflect a titration strategy in the label (including differences

between countries)?

• Would this strategy endanger the superiority label vs. warfarin?

• What would be the implication of such a strategy for indications other than SPAF?


Potential Mid to Long Term Strategy - Alternative

• BI has been approached by Astra Zeneca about a potential co-development and co-promotion of

their Direct Thrombin Inhibitor which is about to enter Phase III

• The AZ approach (received after the titration idea was developed at BI) is very similar to the

aforementioned potential BI strategy:

• Titration of dose, e.g. one week after first dosing

• Have specific lab test available at time of launch

• Have reversal agent available at time of launch

• AZ DTI is expected to be a once-daily therapy with low peak to trough ratio (exposure testing

potentially independent of timing of last dose) and low intra-individual variability


Proposed Concrete Next Steps

1. Prepare meeting request and discuss Modeling data as well as potential PK/PD study with FDA

2. Conduct Market Research to test potential acceptance of titration strategy in Marketplace

3. Develop models assessing cost sensitivity to payers under different assumptions

4. Conduct Challenge Meeting to further define Pros and Cons of titration strategy including

MAPOR, Regulatory, Marketing, Medicine, IPM, etc.

5. Have VC with Astra Zeneca after US label of apixaban is available to further exchange thoughts

on a potential collaboration


BACKUPs


A Clinical Trial Simulation Analysis

May 08, 2012

Impact of Dabigatran Dose Adjustment on Clinical

Outcome in AF Patients

Objectives

• Evaluate the impact of dose adjustment (=“titration”) on clinical outcome in AF patients by clinical

trial simulations.

• Dose adjustment to 75, 110 and 150 mg BID based on trough concentration at steady-state

(Ctrough,ss) observed after one week dosing with DE 150 mg BID.

• Identification of optimal cut-off concentrations for dose adjustment.

• Comparison to warfarin data


Study Design


Methods

• Cut-off values between 0 ng/mL and 250 ng/mL (step size 10 ng/mL) were evaluated to identify

optimal cut-off concentrations for dose adjustment, resulting in 352 different combinations.

• 500 clinical trials with 5000 patients each were simulated for each of the 352 cut-off combinations.

• Patient characteristics were bootstrapped from RE-LY database.

• Ctrough,ss, ischemic stroke/SEE and major bleeding event rates were calculated.


0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

150

160

170

180

190

200

210

220

230

240

250

0102030405060708090

100110120130140150160170180190200210220230240250

-35.0 -32.5 -30.0 -27.5 -25.0 -22.5 -20.0 -17.5 -15.0 -12.5 -10.0 -7.5 -5.0 -2.5 0.0

Results

Impact on Major Bleeds


Cut-off 2

Cut

-off

1

% Change Compared to Reference (150 mg BID)

Ctrough,ss

[ng/mL]

Dosing

< Cut-off1 150 mg BID

Cut-off1

-< Cut-off 2

110 mg BID

> Cut-off 2 75 mg BID

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

150

160

170

180

190

200

210

220

230

240

250

0102030405060708090

100110120130140150160170180190200210220230240250

0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5

Results

Impact on Ischemic Stroke/SEE


Cut-off 2

Cut

-off

1

% Change Compared to Reference (150 mg BID)

Ctrough,ss

[ng/mL]

Dosing


Cut-off1

-< Cut-off 2

110 mg BID


0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

150

160

170

180

190

200

210

220

230

240

250

0102030405060708090

100110120130140150160170180190200210220230240250

-6.0 -5.5 -5.0 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0

Results

Impact on Major Bleeds AND Ischemic Stroke/SEE


Cut-off 2

Cut

-off

1

Ctrough,ss

[ng/mL]

Dosing


Cut-off1

-< Cut-off 2

110 mg BID


= %Change Major Bleeds + 2x %Change Ischemic Stroke/SEE

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

150

160

170

180

190

200

210

220

230

240

250

0102030405060708090

100110120130140150160170180190200210220230240250

-6.0 -5.5 -5.0 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0

Results

Impact on Major Bleeds AND Ischemic Stroke/SEE


Cut-off 2

Cut

-off

1

Ctrough,ss

[ng/mL]

Dosing

< 90 150 mg BID

90 -<140 110 mg BID

>140 75 mg BID

= %Change Major Bleeds + 2x %Change Ischemic Stroke/SEE

Study Design


Results


Dabigatran Dose (A) Reference (B) Titration

75 mg bid --- 25.5%

110 mg bid --- 29.9%

150 mg bid 100% 44.6%


• Majority of patients eligible for 150 mg bid

• Significant amount assigned to 75 mg bid

Ctrough,ss

N=5000 Median P10 – P90

(A) Reference 97.1 48.6 – 199.4

(B) Titration 77.0 48.6 – 101.6

• Shift in exposure

• Median Ctrough,ss ~21% reduced

• Minimum exposure levels maintained

• 90th percentile significantly reduced


Results

Model Predicted Outcome



(A) Reference 1.26 1.01 – 1.55 4.38 3.91 – 4.89

(B) Titration 1.34 1.08 – 1.63 3.49 3.08 – 3.96



Titration vs. Reference

• Risk of ischemic stroke/SEE events comparable (Relative Risk 1.06)


Absolute Event Rates, not annualized

(B) Titration vs. (A) Reference Rel. Risk 90% CI


Major Bleeding 0.80 (0.66 – 0.97)

Relative Risk

Results




(B) Titration (n=5000) 1.34 1.08 – 1.63 3.49 3.08 – 3.96

(C) Warfarin (n=4597) 1.68 1.38 - 2.02 5.83 5.27 - 6.43



(B) Titration vs. (C) Warfarin Rel. Risk 90% CI


Major Bleeding 0.60 (0.50 – 0.72)

Relative RiskAbsolute Event Rates, not annualized

Titration vs. Warfarin

• Only RE-LY warfarin patients from centers contributed to PK measurements were considered (center

matching)

• Risk of ischemic stroke/SEE events reduced (Relative Risk 0.8)


• Comparison needs to be handled with care. A potential bias can not be excluded due to the type of analysis.

Results - Overview

Titration versus Warfarin/Reference


Summary

Clinical Trial Simulation Analysis

• Extensive and comprehensive clinical trial simulation analyses were performed to investigate the impact of

dabigatran dose titration on the outcome in AF patients.

• Dabigatran Ctrough,ss values of 90 ng/mL and 140 ng/mL were identified as promising cut-off values to

assign dabigatran doses of 150 mg bid, 110 mg bid and 75 mg bid.

• Compared to a reference treatment (=DE 150 mg bid), dose adjustment showed a significant reduction of

major bleeding events (RR 0.8), while the ischemic stroke protection was maintained (RR 1.06).

• Compared to warfarin treatment, dose adjustment showed a significant reduction of ischemic stroke/SEE (RR

0.8) and major bleeding events (RR 0.6).


COMPARISON OF OBSERVED DABIGATRAN VS. OBSERVED

WARFARIN OUTCOMES FROM RE-LY

Impact of Dabigatran Dose Adjustment on Clinical Outcome in AF Patients


Background

• The comparison of the projected dabigatran versus the observed warfarin outcome is not considered as ideal

• Solution: Comparison of observed dabigatran versus observed warfarin outcomes from RE-LY in a matching

cohort

• Difficulty: Identification of the appropriate warfarin comparison group

• ~26% of dabigatran treated patients are expected to be assigned to the 75 mg bid dose and are

consequently not considerable for further analysis

• These patients are expected to be patients at higher risk for outcome events due to their demographic

characteristics

• It is therefore important to remove this population also from the warfarin group to ensure a fair

comparison between both groups


Methods

1. All dabigatran treated patients from RE-LY with valid Ctrough,ss concentrations (n=8458) were assigned to their optimal

target dose (75, 110, 150), based on their measured trough concentrations in RE-LY assuming dose proportionality. If

more than 1 concentration was available, the median concentration was calculated.

2. A classification model was developed in dabigatran treated patients based on the propensity score method to identify

with baseline characteristics the patients assigned to “DE 75 mg bid” versus “not DE 75 mg bid”.

3. The developed classification model was applied to center matching warfarin patients from RE-LY to identify those

warfarin patients that are comparable to “optimal DE 75 mg” patients. Those patients were removed from the

comparison with the observed DE group.

4. Selection of patients for statistical comparison from RE-LY safety dataset

a) Dabigatran: only patients who were treated “optimally” based on their trough concentration were included (i.e.

predicted optimal dose = received dose in RE-LY)

b) Warfarin: only patients assigned to “not DE 75 mg bid” were included.


Results

Database

• From the 8458 dabigatran treated RE-LY patients with valid Ctrough,ss values, 25.6%, 25.8% and 49.6% were

assigned to 75 mg, 110 mg and 150 mg bid as their “optimal” dose.

• From these 8458 patients, 3086 patients (36.5%) received the “optimal” dabigatran dose in RE-LY.

• A propensity score classification model was successfully developed.

• 1340 of 5301 center matching warfarin patients (25.3%) were identified as “75 mg patients” and excluded

• In depth comparison of patient characteristics of excluded dabigatran and excluded warfarin patients

showed no relevant difference in any of the relevant factors (e.g. CHADS2, CHADS VASc, HAS−BLED, age, etc.)

• Dabigatran and warfarin patients included in the analysis showed also no relevant difference in any of the

relevant patient characteristics



“Dabigatran Titration vs Warfarin” – Safety Dataset


Treatment Titration

DE 110 mg 1081

DE 150 mg 2005

Warfarin 3961

Total 7047

Number of patients(DE110 + DE 150)

Optimally treated (=titrated) dabigatran patients reveal a promising efficacy and

safety profile compared to matched warfarin group.


“Titration vs RE-LY” – Safety Dataset


Treatment Titration RE-LY

DE 110 mg 1081 5983

DE 150 mg 2005 6059

Warfarin 3961 5998

Total 7047 18040

Number of patients

Titrated patients show comparable efficacy profile as DE 150 mg and comparable

safety profile as DE 110 mg group from RE-LY.

Summary

• A complex propensity score classification model was successfully developed and allowed a matched

warfarin comparison.

• Optimally treated (=titrated) dabigatran patients revealed a promising efficacy and safety profile

compared to matched warfarin group:

• Efficacy profile comparable to DE 150 mg group from RE-LY

• Safety profile comparable to DE 110 mg group from RE-LY


Conclusion

• Comprehensive clinical trial simulation analyses identified promising cut-off values to assign optimal

dabigatran doses of 150, 110 and 75 mg bid.

• Optimally treated (=titrated) dabigatran patients revealed a promising efficacy and safety profile compared to

matched warfarin group.

• Future clinical studies might be required to confirm results.

• Dose titration based on exposure is a promising approach to significantly reduce major bleeding events while

maintaining the stroke protection in AF patients.


backup



Yearly Event Rates – Safety Dataset

DE110 DE150 Warfarin

N 1081 2005 3961

primary stroke/SE 0.70 0.81 1.32

ischemic stroke (incl. uncertain) 0.55 0.76 0.92

Haemorrhagic stroke 0.05 0.00 0.34

Death 1.90 1.68 1.90

Vascular death 1.50 1.22 1.35

MI (incl. silent) 0.85 0.65 0.48

NCB (stroke/se, PE, MI (incl.

silent), death, MBE)5.59 4.42 5.59

stroke/se, MBE, all cause death 4.89 3.88 5.21

stroke/se, MBE 3.50 2.47 3.80

MBE 2.85 1.68 2.96

life-threatening MBE 1.30 0.60 1.47

GI MBE 0.90 0.49 0.82

ICH 0.15 0.14 0.66

any bleeds 15.38 14.23 19.11


Optimally treated DE patients and matching Warfarin patients

Results - cut-off: 90 ng/mL and 140 ng/mL



Results - cut-off: 90 ng/mL and 140 ng/mL




(A) Reference (n=5000) 1.26 (1.01 – 1.55) 4.38 (3.91 – 4.89)

(B) Titration (n= 5000) 1.34 (1.08 – 1.63) 3.49 (3.08 – 3.96)

(C) Warfarin (n= 4597) 1.68 (1.38 - 2.02) 5.83 (5.27 - 6.43)


*risk of event within median RE-LY duration [~20 months], not annualized; $ Clopper-Pearson (Exact)


Rel. Risk 90% CI Relative Risk 90% CI$

Titration vs Reference 1.06 (0.76 – 1.50) 0.80 (0.66 – 0.97)

Titration vs Warfarin 0.80 (0.58 – 1.11) 0.60 (0.50 – 0.72)

Background

Dose-Exposure-Response Relationship

Dose-Exposure Relationship

• Pharmacokinetic was thoroughly characterized in > 11.000 AF patients, OS patients, and healthy

subjects; relevant covariates were identified.

• Good predictive performance of dabigatran exposure.

Exposure-Response Relationship

• Exposure-response relationship was intensively characterized in >8000 AF patients from RE-LY

showing a significant relationship between dabigatran trough concentrations at steady-state

(Ctrough,ss) and major bleeding events and ischemic stroke/SEE events; relevant covariates were

identified.

• Bleeding and ischemic stroke/SEE model showed good predictive performance of outcome events

from remaining ~3500 RE-LY patients not included in analysis!!! (NEW)

Good predictive performance of available models justifies projection of “what-if” scenarios.


A Clinical Trial Simulation Analysis-add on-

May 25, 2012



Translational Medicine

Background

• In the original analysis, all patients were kept in the analysis, even if they had low exposure levels

which might have lead to insufficient stroke protection.

• How many patients had low exposure levels even under 150 mg BID?

• What is the impact if these patients are excluded from the analysis?

• What is the impact if these patients are up-titrated to 220 mg BID?


Methods

• Patients from the original analysis with low exposure (Ctrough,ss < 50 ng/mL at 150 mg BID) were

in addition

a) Excluded

b) Up-titrated to 220 mg BID


Scenario Description

Reference all patients receive 150 mg BID

Titration 1 patients receive 150, 110 or 75 mg BID based on Ctrough,ss at 150 mg BID (cut-offs 90

& 140 ng/mL)

Titration 2 as Titration 1, patients with Ctrough,ss < 50 ng/mL excluded

Titration 3 as Titration 1, patients with Ctrough,ss < 50 ng/mL uptitrated to 220 mg bid

Results

Model Predicted Outcomes and Exposure

Ischemic Stroke Major Bleeding Ctrough,ss [ng/mL]

n Mean* 90% CI$ Mean* 90% CI$ Median P10-P90

Reference 5000 1.256 1.02-1.54 4.377 3.93-4.88 97.1 48.6-199.4

Titration 1 5000 1.345 1.10-1.64 3.490 3.09-3.94 77.0 48.6-101.6

Titration 2 4450 1.354 1.10-1.67 3.702 3.26-4.20 79.3 61.3-103.4

Titration 3 5000 1.332 1.09-1.63 3.545 3.14-4.00 77.0 57.2-101.6


*risk of event within median RE-LY duration [~20 months], not annualized; $ Wilson confidence interval

Reference: all patients 150 mg BID; Titration 1: original analysis; Titration 2: exclusion of patients; Titration 3: up-titration 220 mg BID

Results

Model Predicted Outcomes


Titration 1: original analysis; Titration 2: exclusion of patients; Titration 3: up-titration 220 mg BID

Results

Distribution of doses

Dabigatran Dose (bid) Reference Titration 1 Titration 2 Titration 3

0 mg (excluded) --- --- 11.0% ---

75 mg --- 25.5% 25.5% 25.5%

110 mg --- 29.9% 29.9% 29.9%

150 mg 100% 44.6% 33.6% 33.6%

220 mg --- --- --- 11.0%


• 11% of patients < 50 ng/mL, resulting in exclusion or up-titration.

Reference: all patients 150 mg BID; Titration 1: original analysis; Titration 2: exclusion of patients; Titration 3: up-titration 220 mg BID

Summary

• About 11% of the patients are predicted to have low exposure levels (Ctrough,ss < 50 ng/mL) after

treatment with DE 150 mg BID

• Exclusion of these patients (Titration 2) results in increased ischemic stroke/SEE (RR:1.01) and

major bleeding rates (RR:1.06) compared to original titration (Titration 1).

• Reason: the excluded patients are in general younger and healthier and have consequently a

lower risk for outcome events. If excluded a population at higher risk is generated.

• Up-titration of these patients (Titration 3) results in minor reduction in ischemic stroke/SEE (RR:

0.99) and in minor increase of major bleedings (RR: 1.02) compared to the original titration

analysis (Titration 1).


Conclusion

• AF patients with low exposure have a minor impact on the overall outcome of the titration analysis.

• Up-titration or exclusion of patients with low exposure did not improve the outcome significantly.

• Threshold of low exposure (Ctrough,ss < 50 ng/mL) resulting in up-titration/exclusion was chosen

based on educated guess. A systematic review of the optimal cut-off value could be performed, but

is not expected to improve the outcome significantly.


A Clinical Trial Simulation Analysis-back up slides-

July 11, 2012



Global Translational Medicine

Pharmacometrics

Questions prepared

• What is the impact of intra-individual variability on the dose selection?

• What is the impact of the 10-16 h sampling window on the dose selection?

• How is the predictive performance of the dose-exposure-response model?

• Why is the projected percentage of patients on 75 mg BID so high?


WHAT IS THE IMPACT OF INTRA-INDIVIDUAL VARIABILITY ON

THE DOSE SELECTION?


Intra-individual Variability

Background

• The intra-individual pharmacokinetic variability was estimated based on the RE-LY dataset using a

population PK modeling approach.

• An intra-individual variability of 26.6 %CV on bioavailability was estimated.

• Results not officially reported yet.



Methods

• Adapted RE-LY population PK model was used to predict the influence of intra-individual

variability on dose selection.

• Baseline demographics from 9521 AF patients from RE-LY were included. For each patient, 10

repeated measurements (Ctrough,ss). Overall, 500 clinical trials with 5000 patients each were

bootstrapped and simulated.

• For each patient at each replicate, the median Ctrough,ss across the 10 measurements was

calculated, assuming to represent the “true Ctrough,ss” value and consequently the “true target

dose”.

• Subsequently, the median of the first two up to ten trough concentrations was calculated and the

respective dose assigned.

• The assigned dose was compared to the true target dose.



Results – Example Patient


75 mg BID

110 mg BID

150 mg BID




75 mg BID

110 mg BID

150 mg BID




75 mg BID

110 mg BID

150 mg BID


Results


#Correctly dosed

patients [%]

1 76 0

2 82 0

3 84 9

4 88 1

5 89 5

6 91 9

7 92 8

8 95 2

9 95 7

10 100 0


Results – by dose group


# Correctly dosed patients [%]

75 mg 110 mg 150 mg

1 81 8 58 0 85 2

2 87 5 69 4 87 7

3 88 0 73 8 90 7

4 90 7 79 9 92 3

5 91 3 82 2 93 6

6 93 4 86 5 94 8

7 93 8 87 8 95 6

8 96 0 92 1 97 0

9 96 2 92 9 97 4

10 100 0 100 0 100 0


Summary

• Intra-individual variability might influence dose selection.

• If only one measurement is taken per patient, >76% of patients are dosed correctly.

• At least 2 measurements are required to achieve an accuracy > 80%.

• Analysis by dose group showed, that the target doses DE 75 mg and DE 150 mg are relatively

robust against intra-individual variability.

After one measurement, already >82% of patients are dosed correctly.

• The DE 110 mg dose group was identified as more sensitive towards intra-individual variability.


WHAT IS THE IMPACT OF THE 10-16 H SAMPLING WINDOW ON

THE DOSE SELECTION?


Sampling Window

Background


Sampling Window

Methods

• RE-LY population PK model was used to predict the influence of the 10-16 h sampling window on

dose selection.

• Baseline demographics from 9521 AF patients from RE-LY were included. For each patient, steady-

state concentrations 10, 11, 12, 13, 14, 15 and 16 hours after the last dose were simulated.

• Overall, 500 clinical trials with 5000 patients each were simulated. Patients were randomly

bootstrapped. Also one time after dose value was randomly picked.

• The assigned dose was compared to the “true target dose” (= based on 12 h value)


Sampling Window



Sampling Window



Sampling Window



Sampling Window

Results


TAD

[h]

Correctly dosed

patients [%]

10 79 5

11 90 1

12 100 0

13 91 1

14 83 3

15 76 3

16 70 0

Sampling Window



Time after

dose [h]

Correctly dosed patients [%]

75 mg 110 mg 150 mg

10 100 0 69 4 74 6

11 100 0 85 7 87 4

12 100 0 100 0 100 0

13 85 9 82 5 100 0

14 74 0 66 5 100 0

15 63 6 51 3 100 0

16 54 8 37 8 100 0

Sampling Window

Summary

• The time after dose of the Ctrough,ss value might influence dose selection.

• The time window between 10 – 14 hours results in a stable selection of the correct dose (>80%

correctly dosed patients).

• A more narrow time window of 11 – 13 hours would increase the precision to >90% correctly

dosed patients.

• Analysis by dose group showed, that the 110 mg BID target dose was most sensitive towards the

sampling window.


HOW IS THE PREDICTIVE PERFORMANCE OF THE DOSE-

EXPOSURE-RESPONSE MODEL?


Predictive Performance of Models

Background: Population PK Model RE-LY

• RE-LY PK data was analyzed in a

Population Pharmacokinetic modeling

approach.

• 9522 patients contributing 27706 dabigatran

plasma concentrations.

• Of the covariates identified, only renal

function (creatinine clearance - CRCL)

showed a major effect on dabigatran

exposure.

CentralComp.

Absorption/Bioavailability

ClearanceCreatinineClearanceAgeSexOthers

PeripheralComp.

IntercompClearance

BodyWeightHemoglobin

VerapamileAmiodaroneProton PumpInhibitors

Liesenfeld et al., J Thromb Haemost, 2011, 9: 2168–2175

19 December 2011 - Internal Use Only


Results: Population PK Model RE-LY

72

Validation 2:

• Adolescent patients (1160.89)

Dabigatran Ctrough,ss [ng/mL]

0 50 100 150 200 250

50mg bid-n=1011

150mg bid-n=1108

225mg bid-n=785

300mg qd-n=715

Observed Predicted

Validation 1:

• Three phase II trials

• Adult atrial fibrillation and orthopedic surgery patients

(Median and 10th – 90th percentile)



Background: Exposure Response Model

• The relationship between dabigatran trough concentrations at steady-state and major clinical endpoints (major

bleeding events, ischemic stroke/SEE) was investigated using the RE-LY database. In addition, the impact of

covariates was investigated. Data from 8458 patients with trough concentrations were included. Only events

on treatment were considered for analysis.

• A significant correlation between dabigatran trough concentration and the clinical endpoints major bleeding

events and ischemic stroke/SEE was established.

• In addition, various covariates (e.g. age) were identified having an influence on the outcome events.

Confidential - Internal use only 73Source: Draft Exposure Response Manuscript


Result: Exposure Response Model

• The predictive performance of the final logistic regression models for major bleeding and ischemic stroke/SEE

was tested by projecting the event rates in the dabigatran patients without plasma concentration

measurements (N= 3,642).

• For these patients the trough plasma concentrations were predicted using the population PK model from RE-

LY, the individual stroke and bleed rates were calculated based on the final logistic regression models.

• Patients were sorted ascending according to their calculated risk and divided into deciles. For each of the 10

validation subgroup the mean predicted risk for an event and for no event was calculated and compared with

observed rates.

• The predictive performance was evaluated by the Hosmer and Lemeshow test.

• For major bleedings as well as for ischaemic stroke/SEE the P-value was > 0.05, indicating no significant

difference between the predicted and observed event rates.

Good predictive performance!


Source: Draft Exposure Response Manuscript


Result: Exposure Response Model

No significant difference is observable between the observed and predicted event rates.


The figure outlines the predicted risk

(mean and 95% CI) for each of the

10 validation subgroups and the

observed event rates. The graph

underlines the good predictive

performance of the models.



Summary

• The dose-exposure (=population PK) model as well as the exposure-response model have

shown an excellent descriptive as well as predictive performance.

• Important covariates influencing the dose-exposure-response relationship have been

identified.

• The excellent performance of the models allows prediction of untested scenarios.


WHY IS THE PROJECTED PERCENTAGE OF PATIENTS ON 75 MG

BID SO HIGH?


75 mg BID Dose

Background and Results

Dabigatran Dose (B) Titration

75 mg bid 25.5%

110 mg bid 29.9%

150 mg bid 44.6%


• A shift to 75mg bid was proposed for Ctrough,ss

concentrations >140 ng/mL at steady-state after 150

mg bid dosing.

• 25.5% of patients were predicted to be titrated to

the 75 mg bid dose.

• In the DE 150 mg bid arm from RE-LY, the 75th

percentile was calculated at 145 ng/mL, i.e. 25% of

samples are above 145 ng/mL. 27% of samples are > 140 ng/mL.

• This fits well with model prediction that 25.5% of

patients are >140 ng/mL

Projected Dose Distribution

Observed Ctrough,ss in RE-LY(based on population PK dataset)

Dabigatran Dose Median 75th percentile

110 mg bid 65.3 ng/mL 102.0 ng/mL

150 mg bid 93.8 ng/mL 145.0 ng/mL

WHAT IS THE IMPACT ON OUTCOME EVENTS IF THE 110 MG

BID DOSE IS OMITTED?


Cut-Off Ctrough,ss [ng/mL]

0 20 40 60 80 100 120 140 160 180 200 220 240 260

%C

hang

e M

ajor

Ble

eds

+ 2x

%C

hang

e Is

chem

ic S

troke

/SE

E

-8

-6

-4

-2

0

2

Results – 75 mg and 150 mg


Confidential - Internal use only 80Optimal cut-off at 110 ng/mL

Results


Dose (A) Ref. (B) Titrat. 1 (C) Titrat. 2

75 mg bid --- 25.5% 41.1%

110 mg bid --- 29.9% 0.0%

150 mg bid 100% 44.6% 58.9%


Ctrough,ss

N=5000 Median P10 – P90

(A) Reference 97.1 48.6 – 199.4

(B) Titration 1 77.0 48.6 – 101.6

(C) Titration 2 74.2 48.6 – 106.9


Reference: all patients 150 mg BID; Titration 1: original analysis (DE 150, 110, 75); Titration 2: only 75 mg BID and 150 mg BID

Results


Ischemic Stroke Major Bleeding Ctrough,ss [ng/mL]

n Mean* 95% CI$ Mean* 95% CI$ Median P10-P90

Reference 5000 1.256 1.02-1.54 4.377 3.93-4.88 97.1 48.6-199.4

Titration 1 5000 1.345 1.10-1.64 3.490 3.09-3.94 77.0 48.6-101.6

Titration 2 5000 1.352 1.11-1.65 3.460 3.06-3.91 74.2 48.6–106.9


*risk of event within median RE-LY duration [~20 months], not annualized; $ Wilson confidence interval


Results




Results




Results





Results


#Correctly dosed

patients [%]

1 86.62 90.23 91.84 93.55 94.36 95.67 96.08 97.49 97.610 100.0




#Correctly dosed patients

[%]

75 mg 150 mg

184.6 88.0

289.7 90.4

390.1 92.8

492.6 94.1

593.1 95.1

694.8 96.1

795.1 96.7

896.9 97.8

997.0 98.0

10100.0 100.0

Sampling Window

Results - 75 mg and 150 mg


TAD

[h]

Correctly dosed

patients [%]

10 79 5

11 90 1

12 100 0

13 91 1

14 83 3

15 76 3

16 70 0

Sampling Window

Results – by dose group - 75 mg and 150 mg


Time after

dose [h]

Correctly dosed patients [%]

75 mg 150 mg

10 100.0 80.811 100.0 90.912 100.0 100.013 88.5 100.014 77.9 100.015 68.9 100.016 60.6 100.0

backup


Cut-Off Ctrough,ss [ng/mL]

0 20 40 60 80 100 120 140 160 180 200 220 240 260

%C

hang

e M

ajor

Ble

eds

-40

-35

-30

-25

-20

-15

-10

-5

0

%C

hang

e Is

chem

ic S

troke

/SE

E

0

2

4

6

8

10

12

14

16

18

20

Results – 75 mg and 150 mg



Distribution Time after Dose RE-LY


distribution of time after dose for trough value

Min P1 P5 P25 Median Mean P75 P95 P99 Max N

DOSE10.0 10.3 11.2 12.4 13.3 13.4 14.4 15.6 16.0 16.0 6306

110

150 10.0 10.3 11.1 12.3 13.3 13.3 14.4 15.6 16.0 16.0 6271

All 10.0 10.3 11.1 12.3 13.3 13.4 14.4 15.6 16.0 16.0 12577


110 mg bid



150 mg bid



Results: Prediction of RE-LY



Results: Prediction of RE-LY



Background: Exposure Response Model



Results




Group Comparison RR LCL UCL PValue

Titration 2 Warfarin 0.80716 0.58366 1.11625 0.19442


Reference Warfarin 0.74985 0.53863 1.04389 0.08699

Titration 2 Reference 1.07643 0.76521 1.51424 0.67222


Titration 2 Titration 1 1.00446 0.71837 1.40447 0.97925

Group Comparison RR LCL UCL PValue



Reference Warfarin 0.75096 0.63102 0.89369 0.00120



Titration 2 Titration 1 0.99140 0.80636 1.21892 0.93472

Major Bleeding events

Ischemic Stroke/SEE events


Background: Exposure Response Model (Major Bleeds)




Background: Exposure Response Model (Ischemic Stroke)


Taken from draft Exposure Response Manuscript

RE-LY® Special AnalysisModelling vs. CHADS2 ≥ 3 or EU Label Recomm. Dosing

CHADS2 ≥ 3 an Modelling Data Sets- HR (95% CI) -CHADS2 ≥ 3 an Modelling Data Sets- HR (95% CI) -

CHADS ≥ 3 Analysis (safety set [ss]) (ss)

Event 110 (n=1966) vs. W

(n= 1931)

150 (n=1979) vs. W

(n=1931)

110 vs. 150 DE (150/110) vs. W DE (tailored dosing)vs. W (matched)

Stroke 0.71 (0.50-1.02)

0.52 (0.35-0.77)

1.37 (0.90-2.07)

0.62 (0.46-0.84) 0.58 (0.41-0.83)

Ischemic Stroke 0.99 (0.66-1.47)

0.67 (0.43-1.05)

1.47 (0.94-2.30)

0.83 (0.58-1.19) 0.75 (0.50-1.11)

Haemorrhagic stroke 0.09 (0.02-0.39)

0.14 (0.04-0.47)

0.65 (0.11-3.88)

0.12 (0.04-0.31) 0.05 (0.01-0.38)

MI (incl. Silent) 1.01 (0.59-1.72)

1.58 (0.97-2.27)

0.64 (0.39-1.04)

1.29 (0.83-2.02) 1.49 (0.95-2.35)

MBE 0.81 (0.64-1.02)

1.08 (0.86-1.34)

0.75 (0.60-0.95)

0.94 (0.77-1.14) 0.70 (0.56-0.88)

Life Threatening Bleed 0.72 (0.52-1.02)

0.86 (0.62-1.19)

0.84 (0.59-1.20)

0.79 (0.60-1.05) 0.57 (0.41-0.81)

ICH 0.13 (0.05-0.34)

0.33 (0.17-0.63)

0.41 (0.14-1.15)

0.23 (0.13-0.41) 0.21 (0.10-0.45)

GI MBE 1.28 (0.87-1.88)

1.81 (1.26-2.60)

0.71 (0.50-0.99)

(1.54 1.10-2.15) 0.78 (0.51-1.18)

Death 0.84 (0.64-1.09)

0.89 (0.68 -1.16)

0.94 (0.71-1.24)

0.86 (0.69-1.08) 0.91 (0.71-1.18)

CHADS2 ≥ 3 subgroup from RE-LY- event rate differences for pre-defined Net Clinical Benefit (NCB -CHADS2 ≥ 3 subgroup from RE-LY- event rate differences for pre-defined Net Clinical Benefit (NCB -

CHADS ≥ 3 Analysis (safety set; adjudicated)

Event 110 vs. W 150 vs. W

NCB (stroke, SEE, PE, MI, death, MBE) -2.4 (-4.6 to -0.2)

-1.0(-3.3 to 1.3)

NCB (stroke, SEE, PE, MI (incl. silent), death, MBE)

-2.2 (-4.5 to 0.0)

-0.7(-3.0 to 1.5)

NCB (stroke, SEE, death) -1.8 (-3.6 to -0.1)

-2.4 (-4.1 to -0.7)

NCB (stroke, SEE, PE, MI, vasc. Death) -1.6 (-3.3 to 0.2)

-1.7 (-3.4 to –0.0)

NCB (stroke, SEE, PE, MI (incl. silent), vasc. Death)

-1.4 (-3.1 to 0.3)

-1.4 (-3.2 to 0.3)

NCB (ischem. Stroke, SEE, PE, MI, TIA, hospitalization, death)

-6.2(-9.3 to -3.0)

- 1.5 (-4.7 to 1.6)

NCB (ischem. Stroke, SEE, PE, MI (incl. silent), TIA, hospitalization, death)

-6.1 (-9.2 to -3.0)

-1.4 (-4.6 to 1.7)

EU label recommended dosing and Modelling Data Sets - HR (95% CI) -EU label recommended dosing and Modelling Data Sets - HR (95% CI) -

EU label recommendeddosing

Modelling Data (safety set)

Event DE according EU label vs. W DE (tailored dosing)vs. W (matched)

Stroke 0.74 (0.60-0.91) [ITT] 0.58 (0.41-0.83)

Ischemic Stroke 0.91 (0.72-1.15) [ITT] 0.75 (0.50-1.11)

Haemorrhagic stroke 0.22 (0.11-0.44) [ITT] 0.05 (0.01-0.38)

MI (incl. Silent) 1.14 (0.83-1.55) [ITT] 1.49 (0.95-2.35)

MBE 0.85 (0.73-0.98) [safety set] 0.70 (0.56-0.88)

Life Threatening Bleed 0.72 (0.58-0.91) [safety set] 0.57 (0.41-0.81)

ICH 0.28 (0.17-0.45) [safety set] 0.21 (0.10-0.45)

GI MBE 1.23 (0.96-1.59) [safety set] 0.78 (0.51-1.18)

Death 0.86 (0.75-0.98) [safety set] 0.91 (0.71-1.18)

Patient numbers EU label analysis: ITT: EU label (n= 6004) W (n=6022) Safety Set: EU label (n= 5981) W (n=5998)

Documents

An Idea for a Mid to Long Term Strategy for Pradaxa