future plans for east - cytel
TRANSCRIPT
Future Plans for East
Cyrus Mehta President, Cytel Inc
Where are we going with East?
Five New Ini*a*ves • Enhance the simula=ons capabili=es by permi@ng external calls to R and SAS
• Condi=onal simula=on of the remainder of the trial given the interim data
• Gatekeeping methods for mul=ple comparisons • Mul=-‐arm group sequen=al designs • Popula=on enrichment designs
R-‐func=on for Weibull Distribu=on
April 23-‐24, 2013 East Webinar 3
External calls to R for Adap=ve Decisions
• East already simulates pa=ent arrivals, pa=ent responses and and drop-‐outs
• Adap=ve decision rules are currently based on condi=onal power
• Permit user specified decision rules: – Bayesian criteria for SSR – More general func=ons for SSR
Implement through calls to R or SAS at each interim analysis for each simulated trial
Different Func=ons for SSR (J & T, 2012)
Condi=onal Simula=on of Remainder of Trial
• At an interim analysis both DMC and Sponsor are very interested in the ques=on: “What is the chance that this trial will succeed”
• Condi=onal power has some limita=ons – no visual display of what could happen – does not give a sense of variability – does not es=mate final treatment effect
Predic=ve Interval Plots I. (Li, Evans, Hajime and Wei, 2009)
Predic=ve Interval Plots II. (Li, Evans, Hajime and Wei, 2009)
Gatekeeping Procedures
Serial and Parallel Gatekeeping
Examples of Serial and Parallel Gatekeeping
Alzheimers: Serial Gatekeeping in East 6.2
• Two co-‐primary endpoints – CGI: Clinical Global Impression – ADAS-‐COG: Alzheimers Disease Assessment Scale and Cogni=ve Subscale
• Two secondary endpoints – SIB: Severe Impairment Babery – MMSE: Mini-‐Mental State Examina=on
Design Parameters
Endpoint Control Mean Treatment Mean
CGI 2.3 2.6
ADAS-‐COG -‐5.9 -‐2.5
SIB -‐10.2 -‐4.5
MMSE -‐0.9 -‐0.6
CGI ADAS-‐COG SIB MMSE
CGI 1.125 1.36 1.84 0.9
ADAS-‐COG 1.36 6.53 4.44 2.16
SIB 1.84 4.44 12.07 2.93
MMSE 0.9 2.16 2.93 2.85
Reference: Reisberg et. al. New England Journal of Medicine, 2003, vol 348, 1333-41
Mul=-‐Arm Group Sequen=al Designs
• Compare D dose groups to a common control without dose-‐response assump=ons
• Standard Approachs include: – Single-‐stage design with closed tes=ng (MCP in East) – Two-‐stage adap=ve design with early stopping or treatment selec=on at stage 1, combining stages with pre-‐specified weights, and closed tes=ng (Posch et al)
• Limita=ons: – correla=on between test sta=s=cs not exploited – Not easy to generalize to mul=ple stages
Generaliza=on of Two-‐Arm GSD
• Monitor the D test sta=s=cs over K looks: – C1, C2, … CK are the K efficacy boundaries – Stop and claim efficacy if one of the test sta=s=cs crosses an efficacy boundary
– Incorporate non-‐binding fu=lity boundaries • Find C1, C2, … CK such that the probability that the max. of D mul=variate normal sta=s=cs exceeds one of the Cj’s is α
Example: Four Arms and Three Stages
Look Cumula*ve Sample Size/Arm
Stopping Boundaries
4-‐Arm Design 2-‐Arm Design
1 25 3.39 3.011
2 50 2.89 2.547
3 75 2.77 1.999
Alterna*ve Hypothesis Power
0.4, 0.4, 0.4 67.6%
0, 0.4, 0.4 58.7%
0, 0.2, 0.4 44.6%
0, 0, 0.4 42.3%
Stopping Boundaries of γ(-4) Spending Function at α=0.05
Power of corresponding two-arm design with 75 patients/arm is 68%
Popula=on Enrichment Designs for Oncology
• Failure rate for late stage oncology trials is almost 60% (Kola and Landis, 2004)
• Two recent scien=fic developments can improve this track record – development of molecularly targeted agents – sta=s=cal methodology of adap=ve trial design applied to =me-‐to-‐event data
• Fact: Some subgroups benefit differen=ally from others when treated with the targeted agent
Oncology Products Approved in the USA for Selected Pa*ent Popula*on
Compound/Target Indica*on (prevalence target)
Crizo*nib (Xalkori®)/ ALK-‐rearrangement • Non-‐small cell lung cancer with ALK-‐rearrangements (5%)
Vemurafenib (Zelboraf®)/ BRAF muta*on • Advanced melanoma with mutant BRAF (30-‐40%)
Trastuzumab (Hercep*n®); Lapa*nib (Tykerb®/ Her2
• Her2 expressing breast cancer (25%) • Her2 expressing metasta=c gastric cancer (20-‐30%)
Aromatase inhibitors (letrozole, exemestane) • ER(+) breast cancer (60-‐70%)
Rituximab (Rituxan®)/ CD20 • CD20(+) B-‐cell lymphomas (90%+) Cetuximab (Erbitux®); Panitumumab (Vec*bix®) / EGFR
• Advanced Head/neck cancer (~100%) • EGFR(+) metasta=c colorectal cancer (60-‐80%) • KRASWT metasta=c colorectal cancer (60%)
DIA Adap=ve Design Scien=fic Working Group
Schema=c Representa=on of Protocol
ALL COMERS
Subgroup 𝑆
Subgroup 𝑆
1⁄3
2⁄3
.5 Treatment .5 Control
.5 Treatment .5 Control
INTERIM
ANALYSIS
𝑛↓0 patients 𝑑↓0 events
𝑑 ↓0 events
Stop for Fu=lity
Con=nue with
S and 𝑆
Con=nue with S only
FINAL ANALYSIS
Perform a closed test
of S
S T RAT I F Y
𝑛↓0 patients
DFCI Annual Biostat Day 5-‐8-‐13
Time Line of S Subgroup
𝑛↓0↑ 𝑑↓0↑ 𝑇↓0↑
Interim Analysis
𝑆↓↑′ cohort
𝑛↓𝑠↑ 𝑑↓𝑠↑ 𝑇↓𝑠↑
Planned Final
Analysis
𝑆↓↑′′ cohort
𝑛↓𝑠 ↑ 𝑑↓𝑠 ↑ 𝑇↓𝑠 ↑
Actual Final
Analysis
Time Axis 0
DFCI Annual Biostat Day 5-‐8-‐13
Time Line of 𝑺 Subgroup
𝑛 ↓0↑ 𝑑 ↓0↑ 𝑇 ↓0↑
Interim Analysis
𝑆↑′ cohort
𝑛 ↓𝑠 ↑ 𝑑 ↓𝑠 ↑ 𝑇 ↓𝑠 ↑
Planned Final
Analysis
Time Axis 0
𝑆 ↓↑′′ cohort Drop the Subgroup if it has low conditional power
𝑺
DFCI Annual Biostat Day 5-‐8-‐13
Hypothesis Testing at Final Analysis (a) If you do not drop 𝑺 at interim
1.96
2.24
R: rejection region for the Intersection hypothesis 𝑯↓𝑺 ∩ 𝑯↓𝑺
2.24 T↓s
T ↓s
Rejection region for the elementary hypothesis
𝑯↓𝑺
For closed testing both 𝑯↓𝑺 and 𝑯↓𝑺 ∩ 𝑯↓𝑺 must be rejected
DFCI Annual Biostat Day 5-8-13
Thank you for par=cipa=ng
• Lots of good discussion • Many ideas for new sopware • Cytel is well on its way. Looking forward to the next 25 years of growth
