2013-10-22_dia webinar

23
10/21/2013 1 Planning and Implementing Large Cardiovascular Trials Cardiovascular Trials Paul Strumph; Lexicon Zoran Antonijevic; Cytel This is a Solution Provider Webinar brought to you by DIA in cooperation with Cytel, Inc and Lexicon Pharmaceuticals, Inc. Disclaimer The views and opinions expressed in the following PowerPoint slides are those of the individual presenter and should not be attributed to Drug Information Association, Inc. (“DIA”), its directors, officers, employees, volunteers, members, chapters, councils, or Special Interest Area Communities or affiliates. These PowerPoint slides are the intellectual property of the individual presenter and are protected under the copyright laws of the United States of America and other countries. Used by permission. All rights reserved. Drug Information Association, DIA and DIA logo are registered trademarks or trademarks of Drug Information Association Inc. All other trademarks are the property of their respective owners.

Upload: cytel

Post on 27-Jun-2015

58 views

Category:

Health & Medicine


0 download

TRANSCRIPT

Page 1: 2013-10-22_DIA Webinar

10/21/2013

1

Planning and Implementing Large Cardiovascular TrialsCardiovascular Trials

Paul Strumph; Lexicon

Zoran Antonijevic; Cytel

This is a Solution Provider Webinar brought to you by DIA in cooperation with Cytel, Inc and Lexicon Pharmaceuticals, Inc.

Disclaimer

The views and opinions expressed in the following PowerPoint slides are those of the individual presenter and should not be attributed to Drug Information Association, Inc. (“DIA”), its directors, officers, employees, volunteers, members, chapters, councils, or Special Interest Area Communities or affiliates.

These PowerPoint slides are the intellectual property of the individual presenter and are protected under the copyright laws of the United States of America and other countries. Used by permission. All rights reserved. Drug Information Association, DIA and DIA logo are registered trademarks or trademarks of Drug Information Association Inc. All other trademarks are the property of their respective owners.

Page 2: 2013-10-22_DIA Webinar

10/21/2013

2

• Background

• Medical Considerations

St t & D i

Outline

• Strategy & Design– Strategy #1: Address Pre & Post-Marketing Objectives

Simultaneously

– Strategy #2: Address Pre & Post-Marketing Objectives Separately

• Sample Size and Duration of ProgramSample Size and Duration of Program

Background

Page 3: 2013-10-22_DIA Webinar

10/21/2013

3

• 2008: Diabetes Mellitus — Evaluating Cardiovascular Risk in New Antidiabetic Therapies to Treat Type 2 Diabetes:

FDA Guidance

1. The upper bound of two-sided 95% confidence interval of the risk ratio of treatment over control is less than 1.8 using integrated Phase 2 and 3 data. Data from a separate controlled safety study can be integrated with data from Phase 2 and 3 trials.

2 If pre marketing data result with the upper limit in the range2. If pre-marketing data result with the upper limit in the range between 1.3 and 1.8 then a post-marketing study needs to be conducted such that the upper limit based on integrated data that includes this post-marketing study is less than 1.3.

• In response to this guidance sponsors developing

Cardiovascular Outcomes Trials (CVOT)

diabetes products are now conducting large CVOTs.– The cost of these trials is often measured in $100s M.

• The FDA now makes similar requirements for a qnumber of other indications.

Page 4: 2013-10-22_DIA Webinar

10/21/2013

4

• This presentation will begin with medical considerations to put in into a broader context.

• Cardiovascular trials are event-based. Therefore, strategy

Presentation Flow

Cardiovascular trials are event based. Therefore, strategy and design will be discussed in terms of numbers of events.

• We will then describe how selected strategies/designs translate into numbers of patients and length of development.

• In order to demonstrate these strategies we are using numerous assumptions that are to our knowledge mostnumerous assumptions that are to our knowledge most representative of the “real world” scenarios.

• Still, every development program will have its specifics, and we recommend that your focus is more on concepts than on assumptions.

Medical Considerations

Page 5: 2013-10-22_DIA Webinar

10/21/2013

5

• Current Situation

Medical Considerations

• CV Outcome Landscape

• Acceleration of NDA filing

• Serving two masters

Current Situation• Worldwide epidemic of T2DM (IDF)

– 2011 – 366 million; 2030 – expect 552 million

• Regulatory guidance for demonstration of CV safety in USRegulatory guidance for demonstration of CV safety in US and EU (post-Avandia meta-analysis):– Large scale cardiovascular outcomes trials (CVOTs)

– ~5,000-10,000 subjects, 3-5 years, @ cost: $200-400 million

CV Outcomes study is Largest and Highest Cost Study

Must be designed correctly

Page 6: 2013-10-22_DIA Webinar

10/21/2013

6

Cardiac Safety Research Consortium (CSRC)

• https://www.cardiac-safety.org/papers

• Designs and statistical approaches to assess CV risk of new type 2 diabetes therapies in development. Leader: Mary Jane Geiger (In Process)

CV Outcomes Trial Landscape: Current designs

MACE‐primary endpoint Start Drug Size Population NCT

ORIGIN –CV mort 9/2003 Insulin Glargine 12,500 “At risk CVD” prior MI,stroke, revasc 00069784

TECOS (M4) UA‐hosp 12/2008 Sitagliptin 14,000 Preexisting CV D 00790205

M3 = CV death, non fatal MI, non fatal stroke

p g p g

ACE (M3) 02/2009 Acarbose 7,500 Prev MI, ACS, UA, stable Angina 00829660

EXAMINE (M3) 09/2009 Alogliptin 5,400 ACS within 15 to 90 days 00968708

CANVAS (M3) 11/2009 Canagliflozin 4,300 History of or a high risk CVD 01032629

AleCardio (M3) 02/2010 Aleglitazar 7,000 ACS 2‐6 w prior to randomization 01042769

SAVOR TIMI‐53 (M3) 04/2010 Saxagliptin 16,500 EstablCVD and/or multiple RF 01107886

ELIXA (M4)  UA‐hosp 06/2010 Lixisenatide 6,000 ACS within 180 days 01147250

EXSCEL (ND) 06/2010 Exenatide LAR 9,500 No CV Inclusion listed 01144338

C‐SCADE 8 (M3) 07/2010 Empagliflozin 7,000 Prior MI, UA, revasc 01131676

CAROLINA (M4) UA‐hosp 10/2010 Linagliptin 6,000 Pre‐existing CVD 01243424

LEADER (M3) 11/2010 Liraglutide 8,750 Concomitant CVD 01179048

REWIND (M3) 07/2011 Dulaglutide 9,600 Established CVD 01394952

ITCA 650 (M4) UA‐hosp 01/2012 Exenatide ITCA650 2,000 Hx of CVD, Stroke or PAD 01455896

Adapted from Gore O, McGuire D. Diabetes & Vascular Disease Research. 9(2) 85-888. 2012

Selected populations vary, as does MACE definition

Page 7: 2013-10-22_DIA Webinar

10/21/2013

7

• Events = Annual Event rate × Subjects ×

Strategies to accelerate NDA filing

years

• Accelerate NDA filing date by:– Increasing subjects

– Increasing annual event rate

Accelerate NDA filing – latent period considerations

Secondary PreventionScandinavian Simvastatin Study Survival Group

Primary PreventionAFCAPSPrevention of Coronary Heart Disease with Pravastatin

In Men with Hypercholesterolemia

Page 8: 2013-10-22_DIA Webinar

10/21/2013

8

• Does the diabetic population selected

bl th ti t

Accelerate NDA filing – Increasing Annual Event Rate

resemble the patients who would use the drug

– ACS within 15-90 days

– CKD stage 4

• NDA safety exposure filing requirements

Serving Two Masters

• NDA CV Safety requirements

Page 9: 2013-10-22_DIA Webinar

10/21/2013

9

Strategy & Design

1. Address pre-marketing and post-marketing objectives simultaneously.

One meta analysis that includes CVOT that starts in

Strategy

– One meta analysis that includes CVOT that starts in parallel with Phase 3 trials

– Pre-marketing requirement addressed at interim analysis

2. Address pre-marketing and post-marketing objectives separately. j p y1. Meta-Analysis + CVOT pre-marketing. Another, post-

marketing CVOT to address 1.3

2. Meta-Analysis of Phase 2/3 trials only for pre-marketing objectives. CVOT post-marketing

Page 10: 2013-10-22_DIA Webinar

10/21/2013

10

• Event-based trials.

• Sample size considerations.

Design Considerations

– 122 events to achieve the 1.8 requirement, assuming HR=1.0, Power 90%, two-sided α=0.05

– 611 events to achieve the 1.3 requirement, assuming HR=1.0, Power 90%, two-sided α=0.05

• Rare Events

There is a lot of ncertaint regarding the “tr e”• There is a lot of uncertainty regarding the “true” HR, and de-risking is highly recommended– Interim analyses and sample-size re-assessment

• Meta-analysis

Strategy #1Address Objectives Simultaneously

Page 11: 2013-10-22_DIA Webinar

10/21/2013

11

• Address pre-marketing (1.8) and post-marketing (1.3) objectives simultaneously. – Sequential testing, once 1.8 achieved test for 1.3.

Example Strategy #1

• Meta-analysis includes data from all controlled trials: Phase 2b, Phase 3, CVOT.– CVOT initiated in parallel with Phase 3.

• 1.8 requirement addressed with 2 (could have more) interim analyses. Assume HR=1.0.– First interim not to be scheduled before efficacy studies

fi li dare finalized– Second interim is also serving as final analysis for 1.8.

• 1.3 to be assessed at two additional analyses; another interim for 1.3 and final for 1.3. Assume HR=1.0.

Expected Number of Events After Incorporating Interim Analyses (1.8)

Sample Size (Events) and Probability of Stopping at Interim

Design Look Expected # of 

EventsInterim 

(at 75% events)

Final

No Interims 122 122

O’Brien‐

Fleming

93 (69%) 124 103

Pocock 101 (82%) 135 107

Page 12: 2013-10-22_DIA Webinar

10/21/2013

12

• Type I error for each analysis (1.8 or 1.3) controlled by α-spending functions

The o erall t pe I error controlled b seq ential

Final Design Chart

• The overall type I error controlled by sequential testing– Proceed to testing 1.3 only if 1.8 achieved.

HR<1.8 (75%)

HR<1.3 (15%) HR<1.3 (100%)

HR<1.8 (100%)

HR<1.3 (20%) HR<1.3 (50%)

Begin0

IA2 124

Final 613

IA1 93

IA3 307

• What if HR≠1.0?

Sensitivity Analysis:Probability of Stopping for Efficacy

“True” HR

NI Margin

IA1E=93

IA2E=124

IA3E=307

FinalE=613

Total

What if HR≠1.0?

HR Margin E 93 E 124 E 307 E 613

0.8 1.8 94% 5% >99%

1.3 0% 2% 88% 10% 100%

0.9 1.8 85% 12% 97%

1.3 0% 0% 60% 39% >99%

1 0 1 8 69% 21% 90%1.0 1.8 69% 21% 90%

1.3 0% 0% 25% 65% 90%

1.1 1.8 52% 25% 78%

1.3 0% 0% 7% 47% 54%

Page 13: 2013-10-22_DIA Webinar

10/21/2013

13

• One can consider the uSSR to protect against HR > 1.

• Under this strategy an uSSR for 1.8 is of limited

Additional Considerations:Unblinded Sample Size Re-Assessment

(uSSR)

gyoperational impact, because the enrolment is ongoing.

• Similarly, an uSSR can be implemented for 1.3.

HR<1.8 (75%)

HR<1.3 (15%) HR<1.3 (100%)

HR<1.8 (100%)

HR<1.3 (20%) HR<1.3 (50%)

Begin0

IA2 124

Final 613

IA1 93

IA3 307

SSR

• Can be incorporated as an extension of previously described design

Additional Considerations: Superiority

• Proceed to testing once 1.8 & 1.3 achieved.

• When should superiority be considered?

HR<1.8HR<1.3 HR<1.3

Sup Sup Sup……

Begin FinalIAx IAy

Page 14: 2013-10-22_DIA Webinar

10/21/2013

14

Additional Considerations: Superiority

Sample Size (Events) for Superiority, Assuming no Interims

HR # Events

0.75 508

0.80 845

0 85 15920.85 1592

0.90 3787

• Population definition: ITT or treated

• Futility assessments: to be considered,

Additional Considerations:General

particularly for superiority.

• Filing & preserving CVOT integrity.

Page 15: 2013-10-22_DIA Webinar

10/21/2013

15

Strategy #2

Address Objectives Separately

• Meta-Analysis + CVOT pre-marketing.

• Another, post-marketing CVOT to address 1.3

Address Objectives Separately

• Accrual of events begins from the beginning for post-marketing trial

HR<1.8 (75%)

HR<1.3 (100%)

HR<1.8 (100%)

HR<1.3 (50%)

Begin0

Final 124

Final 613

IA1 93

IA1 307

Begin0

Page 16: 2013-10-22_DIA Webinar

10/21/2013

16

• Meta-Analysis + CVOT pre-marketing. Another, post-marketing CVOT to address 1.3

Meta Analysis Including a CVOT trial

• Disadvantages: – You lose events used to address pre-marketing

requirements

– White space, re-initiation of enrolment

– Overall: increased costs, delays in submission and achieving post-marketing goalsachieving post marketing goals

• Advantages– Easier to design and implement

• Meta-Analysis of Phase 2/3 trials only for pre-marketing objectives. CVOT post-marketing

Ke Challenge ho to ass re s fficient po er?

Meta Analysis Only

• Key Challenge: how to assure sufficient power?– Efficacy studies smaller in size

– Low expected event rate (<1%)

– To be illustrated in next section

• Could extend efficacy studies, however:Still t h ffi i t– Still may not have sufficient power

– Delays in approval

– Studies must remain blinded

Page 17: 2013-10-22_DIA Webinar

10/21/2013

17

Sample Size and Duration of Program

• So far everything has been presented in terms of events

• How does this translate into number of patients and development times?

Event-Based Trials

development times?

• Projected times of interim and final analyses.

• Impact of combining with meta-analysis.

• To enrich, or not to enrich?

• Monitoring event rates: as planned or not? How to manage?

• Tools for predicting enrolment and accrual of events critical during both planning and implementation phase

Page 18: 2013-10-22_DIA Webinar

10/21/2013

18

Simulation OverviewTrial Design parameters• sample size• events model/parameters• patient visits schedule• milestones

Enrolment parameters / data• countries / sites data • enrolment / dropout rates• realized enrolment / events data• Activation plan

Si l iSimulationInput

SimulateSi l ti SimulateScenario

SimulationResults

Simulate site activationSimulate patient enrolment Simulate patient eventsSimulate patient dropouts

• Tables / plots o Enrolment predictiono Events predictiono Dropout predictiono Milestones prediction

Enrolment Simulation Model:

• Poisson model with Gamma Prior for rates

Priors based on initial estimates of site enrolments

Simulation Models

• Priors based on initial estimates of site enrolments

• Bayesian updates of site enrolment rates based on realized enrolments so far

Events Simulation Model :

• Piecewise Exponential with Gamma Prior for hazard rates

P i b d i iti l ti t f h d t• Priors based on initial estimates of hazard rates as per the protocol

• Bayesian updates of hazard rates based on known events and withdrawals until latest follow-up of each patient

Page 19: 2013-10-22_DIA Webinar

10/21/2013

19

• Number of countries: 28

• Number of sites: 557

Number of patients: 12000

Case Study: Assumptions

• Number of patients: 12000

• Enrolment period ~ 24 months

• Total trial duration ~ 36 months

• MACE rate 4% (high risk population)

• Interim looks: 3

• Events distribution model: exponentialEvents distribution model: exponential

• Milestones: 93, 124, 307, and 613 MACE events

• Trial start: October 2013

• Four studies in normal risk T2DM population.– Assume annual MACE rate of 0.6%

Efficacy Studies Assumptions

Study Study Design  N Enrolled 

Total/arm 

Arms ST 

A NCE vs. 

Placebo 

240/120 ‐Placebo

‐NCE

24wk

B NCE vs. 

Placebo 

240/120 ‐Placebo

‐NCE

24wk

C NCE vs.  600/300  ‐NCE 52wk 

Glimepiride 

/

‐Glimeperide

D NCE vs. TZD  600/300 ‐NCE

‐TZD

52wk

Page 20: 2013-10-22_DIA Webinar

10/21/2013

20

No. MilestonePr. of 

Achieving 

Pr. of Achieving Milestone 

Milestones & Probability of Achieving Them

Milestone (CVOT only)

(CVOT + Phase 2/3 Efficacy)

1 6000 patients enrolled by 31 December 20140.128 0.37

2 12000 patients enrolled by 31 December 20150.688 0.858

3 93 MACE events by 31 December 2014 0.596 0.772

44 124 MACE events by 28 February 2015 0.714 0.814

5 307 MACE events by 30 September 2015 0.656 0.77

6 613 MACE events by 30 June 2016 0.666 0.752

N MilFrom Phase 2/3 Effi

Reduced f

Contribution from Efficacy Trials

No. Milestone 2/3 Efficacy trials

target for CVOT

1 93 MACE events by 31 December 2014 6 87

2 124 MACE events by 28 February 2015 6 118

3 307 MACE events by 30 September 2015 12 295y p

4 613 MACE events by 30 June 2016 18 595

Page 21: 2013-10-22_DIA Webinar

10/21/2013

21

Enrolment Milestones Probability Distribution

Event Milestones Probability Distribution

Page 22: 2013-10-22_DIA Webinar

10/21/2013

22

Enrolment Prediction

• Given assumptions that we used it would be extremely unlikely that pre-marketing requirement can be achieved without a CVOT.

Simulations Discussion

without a CVOT.

• Since a small number of events that can be expected from efficacy trials one could consider using events from CVOT only (although there are other reasons that may favor meta-analysis).

• Similar simulations can be implemented to assess the impact of enrichment on timelinesimpact of enrichment on timelines.

• Such simulations are also necessary to predict timing of interim and final analysis, and to re-assess timelines if enrolment and/or event accrual are not as planned.

Page 23: 2013-10-22_DIA Webinar

10/21/2013

23

• Requirement: 2500 of any length; 1300-1500 > 12 months; 300-500 > 18 months.

• These requirements can be accomplished within sample

Additional Considerations:NDA Safety Exposure Filing Requirements

These requirements can be accomplished within sample sizes required to meet the CV requirement.

• However, some planning is required. In our case 93, and even 124 events would be reached before the 18 months requirement is met.

• Clearly meta-analysis gives you best chance to minimize time to filing One can also consider initiating one or severaltime to filing. One can also consider initiating one or several efficacy studies prior to initiation of CVOT.

• Previously described simulations can be very useful here.

• Large CVOT studies are causing large increases in cost and risk in drug development in diabetes, and in a number of other indications.

Th fi l i i hi h d l

Summary

• The first planning step is to assess which development strategies are reasonable, and/or most efficient in terms of number of events.

• As the next step simulations should be used to determine what this means in terms of number of patients, and length of development.

• Usually a number of iterations are needed to determine best strategies for number of trials, sample sizes, sequencing, timing of analysis…

• Once the strategy and design are finalized, simulations are necessary for predicting, and re-predicting the timing of analyses