panel 1 - friends of cancer research annual meeting pane… · panel 1 participants moderator:...
TRANSCRIPT
![Page 1: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/1.jpg)
Augmenting Randomized Confirmatory Trials for Breakthrough
Therapies with Historical Clinical Trials Data
Panel 1
#FriendsAM18
Wifi: RitzCarlton_CONFERENCE Password: fcr18
![Page 2: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/2.jpg)
Panel 1 Participants
Moderator: Elizabeth Stuart, Johns Hopkins University
• Andrea Ferris, LUNGevity Foundation
• Antoine Yver, Daiichi Sankyo
• Joohee Sul, U.S. FDA
• Pallavi Mishra-Kalyani, U.S. FDA
• Rajeshwari Sridhara, U.S. FDA
• Ruthie Davi, Medidata Solutions
2
#FriendsAM18
![Page 3: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/3.jpg)
Augmenting Randomized Confirmatory Trials for Breakthrough Therapies with Historical Clinical Trials
Data: Landscape Review and Topic Introduction
Elizabeth Stuart, PhD
November 13, 2018
![Page 4: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/4.jpg)
© 2014, Johns Hopkins University. All rights reserved.©2016, Johns Hopkins University. All rights reserved.
Big picture
• Need for non-experimental study designs
• But how do we know when we can “trust” the results from non-randomized studies?
• Growing literature to try to answer this question
• Today will discuss two case studies examining this, within a broader conversation about the possibilities (and issues)
![Page 5: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/5.jpg)
© 2014, Johns Hopkins University. All rights reserved.©2016, Johns Hopkins University. All rights reserved.
The need for non-experimental studies
• Some questions cannot be answered using randomized experiments:• Effects of approved treatments in real-world use• Effects of treatments it would be unethical to deny from
some individuals• Effects of potentially harmful exposures (e.g.,
carcinogens)
• So then left with non-experimental studies where we did not control who received which exposure/intervention• Many designs exist • Today will focus on comparison group designs
![Page 6: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/6.jpg)
© 2014, Johns Hopkins University. All rights reserved.©2016, Johns Hopkins University. All rights reserved.
The challenge of non-experimental studies
• But in non-experimental studies we have to worry about confounding• The people that get one treatment may be
quite different from those that get another
• We can use statistical methods to deal with observed differences
• But to interpret difference in outcomes as due to the treatment need to assume no unobserved confounders that we didn’t adjust for
• So how do we know when to trust the results?
![Page 7: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/7.jpg)
© 2014, Johns Hopkins University. All rights reserved.©2016, Johns Hopkins University. All rights reserved.
Topic of this session
• Two case studies seeing whether using historical comparison data (instead of randomized controls) can yield accurate effect estimates
• In this case, does the average outcome under the control condition match between the historical data and the randomized arm?• (After adjusting for observed characteristics)
• A start at developing a body of literature on this topic
![Page 8: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/8.jpg)
© 2014, Johns Hopkins University. All rights reserved.©2016, Johns Hopkins University. All rights reserved.
Related literature
• This idea of “design replication” or “within study comparison” studies common in other fields, esp. social sciences (esp. Tom Cook)
• Growing body of research learning lessons from that field
• Factors that lead to accurate results:• Consistent measurement of covariates and outcomes• Match groups based on the key confounders, esp. strong
predictors of outcome (e.g., baseline measures)• Match “locally” if possible
The question: Which of these findings carry over to cancer research?
![Page 9: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/9.jpg)
© 2014, Johns Hopkins University. All rights reserved.©2016, Johns Hopkins University. All rights reserved.
Plan for the session
• Brief introductions of panel
• Two case studies (10 min. each)
• Panel discussion (45 min.)
• Audience Q+A
![Page 10: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/10.jpg)
Case Study in Non-Small Cell Lung Cancer
Exploration of whether a Synthetic Control Arm derived from Historical Clinical Trials by Matching of Baseline Characteristics can Replicate the Overall Survival of a Randomized Control Arm
![Page 11: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/11.jpg)
11© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Working GroupRuthie Davi
Medidata
David Lee
Medidata
Larry Strianese
Medidata
Andrea Ferris
LUNGevity Foundation
Antara Majumdar
Medidata
Elizabeth Stuart
Johns Hopkins University
Andrew Howland
Medidata
Pallavi Mishra-Kalyani
U.S. FDA
Joohee Sul
U.S. FDA
Dominic Labriola
Bristol-Myers Squibb
Zhenming Shun
Daiichi Sankyo
Xiang Yin
Medidata
Michael LeBlanc
Fred Hutchinson Cancer Research
Center, University of Washington
Rajeshwari Sridhara
U.S. FDA
Antoine Yver
Daiichi Sankyo
![Page 12: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/12.jpg)
12© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
• Maintaining a concurrent control arm can be difficult due to rarity of the disease or availability of the investigational agent outside the study
• BRAVO study in BRCA+ breast cancer1
• Unusually high rate of censoring in the control arm likely associated with increased availability of PARP inhibitors
• Unlikely to produce data that is interpretable
• Sunitinib Malate in gastrointestinal stromal tumor2
• Large effect on progression free survival (HR 0.33, 95% CI (0.24, 0.47))
• After 84% of placebo patients elected to receive sunitinib malate, effect on overall survival was not observed (HR 0.88, 95% CI (0.68, 1.1))
The Challenge: Recruitment, Retention, and Compliance in Randomized Control
1. Tesaro press release (2017)
2. Sunitinib malate capsule prescribing information (2006)
![Page 13: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/13.jpg)
13© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
A Possible Solution –Synthetic Control Arm
A Synthetic Control Arm is
• Patient level data from multiple historical clinical trials in the same indication
• Carefully selected historical patients
• Who meet eligibility criteria and were assigned to receive the
appropriate standard of care
• With baseline characteristics that statistically match those in the
current-day experimental arm
• For a setting where a randomized control is problematic
![Page 14: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/14.jpg)
14© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Historical Clinical Trials Data
Treatment Arm of Target Trial
![Page 15: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/15.jpg)
15© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Candidate Historical Patients per Eligibility Criteria
Treatment Arm of Target Trial
![Page 16: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/16.jpg)
16© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Candidate Historical Patients per Eligibility Criteria
Well Matched Synthetic Control Arm
Treatment Arm of Target Trial
![Page 17: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/17.jpg)
17© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Pro
po
rtio
n S
urv
ivin
g
Days
Treatment
![Page 18: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/18.jpg)
18© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Pro
po
rtio
n S
urv
ivin
g
Days
Synthetic
Control Arm
Treatment
Arm of
Target Trial
![Page 19: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/19.jpg)
19© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Synthetic Control
Pro
po
rtio
n S
urv
ivin
g
Days
Synthetic
Control Arm
Treatment
Arm of
Target Trial
Randomized
Control of
Target Trial
![Page 20: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/20.jpg)
20© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Candidate Historical Patients per Eligibility Criteria
Well Matched Synthetic Control Arm
Treatment Arm of Target Trial
Randomized Control of Target Trial
![Page 21: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/21.jpg)
21© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Candidate Historical Patients per Eligibility Criteria
Well Matched Synthetic Control Arm
Treatment Arm of Target Trial
Randomized Control of Target Trial
![Page 22: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/22.jpg)
22© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Pro
po
rtio
n S
urv
ivin
g
Days
Synthetic Control
Synthetic
Control Arm
Synthetic Control
Randomized
Control of
Target Trial
Case Study Objective: Assess whether a Synthetic
Control Arm can be created to replicate the outcomes
of a traditional randomized control
![Page 23: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/23.jpg)
23© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Non-Small Cell Lung Cancer Case Study
![Page 24: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/24.jpg)
24© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Building the SCAPropensity Score Matching
![Page 25: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/25.jpg)
25© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Patient level data from multiple previous NSCLC trials
Data Sources
• Project Data Sphere1
• Medidata Enterprise Data Store (MEDS)2
SCA Patient Eligibility Criteria
• Inclusion in a historical clinical trial accessible within this project
• NSCLC at stage III or IV
• Received prior platinum-based chemotherapy
• Men and women ≥ 18 years of age
• ECOG performance status of ≤ 2
• Measurable disease
• Received treatment with docetaxel
1. These analyses are based on research using information obtained from www.projectdatasphere.org, which is maintained by Project Data Sphere, LLC. Neither Project Data Sphere, LLC nor the owner(s) of any information from the web site have contributed to, approved or are in any way responsible for the contents of this work.
2. Includes thousands of previous clinical trials conducted by the pharmaceutical industry for drug or medical product development with patient level data recorded through the Medidata electronic data capture system. Legal agreements permit use in deidentified (i.e., patients and original sponsor of the trial cannot be identified) and aggregated (i.e., every analysis must include data from two or more sponsors) form.
Trial Characteristics
• Open label and blinded phase 2 & 3 trials
• Multinational
• Timespan of starts of trials (2004 to 2013)
• Overall survival measured
![Page 26: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/26.jpg)
26© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Prespecified Propensity Score Matching
Baseline Characteristics for Matching• Age at baseline (continuous)
• Years from cancer diagnosis (continuous)
• Race (White vs Others)
• Sex (Female vs Male)
• Smoking (Current vs Former vs Never)
• Histology (Squamous vs Non-squamous)
• Stage (III vs IV)
• ECOG (0 vs 1 vs 2)
• Prior surgery (Yes/Maybe vs No)
• EGFR/KRAS mutation (Positive vs No/Unknown)
Randomized Control of
Target Trial
Historical Pool for SCA
![Page 27: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/27.jpg)
27© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Matching PerformanceBaseline Comparability
![Page 28: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/28.jpg)
28© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Comparability of SCA & Target Control
• Considerable mismatch of propensity scores before matching
• After matching, SCA and Randomized Control from Target Trial have similar propensity score distributions
Rand Control frm Target Trial
Hist Pool / SCA
![Page 29: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/29.jpg)
29© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Baseline Characteristics Well Balanced After Matching
Age
Years since
cancer diag.
56.857.6
0.70.8
Before Matching
Hist Pool Rand Control from Target Trial
After Matching
Age
Years since
cancer diag.
SCARand Control from Target Trial
![Page 30: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/30.jpg)
30© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Absolute Standardized Differences in Baseline
Characteristics are Small/Negligible After Matching
Propensity score matching has achieved good balance of baseline characteristics between the SCA and the randomized control from target trial.
![Page 31: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/31.jpg)
31© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
SCA ValidationOverall Survival
![Page 32: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/32.jpg)
32© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Before Matching
Overall SurvivalAfter Matching
Hist Pool
Rand Cntrl
Target Trial
SCA Rand Cntrl
Target Trial
Hist Pool Rand Control Target Trial Hist Pool Rand Control Target Trial
Log-Rank P-value 0.03
Hazard Ratio (95% CI) 1.16 (1.02, 1.32)
Log-Rank P-value 0.65
Hazard Ratio (95% CI) 1.04 (0.88, 1.23)
![Page 33: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/33.jpg)
33© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Summary
• In this NSCLC case study, SCA successfully replicated the control arm of the target trial
• Important step in understanding how SCA may mitigate challenges faced with a concurrent control arm in difficult-to-study indications
• Future work• Assessment of whether the treatment effect can be replicated
with the use of SCA
• Exploration of tweaks to the matching methods to reduce the proportion of patients who are not matched
• Expand to additional indications
![Page 34: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/34.jpg)
34© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Thank you.
![Page 35: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/35.jpg)
35© 2018 Medidata Solutions, Inc. – Proprietary and Confidential
Overall Survival – Matched versus Unmatched Patients from Randomized Control
Matched
TargetUnmatched
Target
![Page 36: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/36.jpg)
36
November 13, 2018
Pallavi S. Mishra-Kalyani
Division of Biometrics V, Office of Biostatistics, CDER,
FDA
Use of External Controls:
A Case Study in Melanoma
www.fda.gov
![Page 37: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/37.jpg)
37
Disclosures
• I have no financial relationships to disclose
• This presentation reflects the views of the author and should not
be construed to represent FDA’s views or policies
www.fda.gov
![Page 38: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/38.jpg)
38
Introduction
• Randomized clinical trials (RCT) remain the gold standard for the identification of treatment effect
• However, disease or population characteristics may require a non-randomized study design
• In the case that a randomized control arm is not feasible, an external control arm may be an option for estimating comparative treatment effect
www.fda.gov
![Page 39: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/39.jpg)
39
Use of External Controls
• External controls could be considered to demonstrate:– Natural history of disease
– Contribution of components
– Established efficacy from prior trials (e.g. comparison to a single arm trial)
• Source of data for controls would determine potential use– Registry
– Electronic health records (EHR)
– Prior clinical trials/synthetic control armwww.fda.gov
![Page 40: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/40.jpg)
40
A Case Study in Melanoma• Trial 1:
– Design: Vemurafenib (Vem) vs. Dacarbazine
– Primary endpoints: PFS (IRC assessed) and OS
– First patient enrolled in 2010
• Trial 2: – Design: Cobimetinib (Cobi) + Vem vs. Vem
– Primary endpoint: PFS (investigator assessed)
– First patient enrolled in 2013
Could the data from the Vem arm of Trial 1 be used as an external/synthetic control arm for the Cobi+Vem experimental
arm of Trial 2 instead of concurrent controls (CC)?www.fda.gov
![Page 41: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/41.jpg)
41
A Case Study in Melanoma• In order to compare the two trials:
– Inclusion/exclusion criteria must match
– Consistency of endpoint measurement (INV vs. IRC, follow-up, etc.)
– Methods needed to make sure comparative efficacy is not subject to bias
• Propensity Score Matching– PS calculated using a logistic regression model for
treatment
– Includes baseline covariates measured in both trials:• Age, sex, region, BMI, ECOG, prior adjuvant therapy, disease
stage, LDH, BRAF mutation genotypewww.fda.gov
![Page 42: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/42.jpg)
42
Trial 1: Vem (EC)
N = 337
Trial 2: Vem (CC)
N = 248
Trial 2: Cobi + Vem
N = 247
Age 55.2 (13.8) 55.3 (13.8) 54.9 (14.0)
BMI 26.8 (5.4) 27.1 (5.1) 27.7 (5.9)
Sex: Female 200 (59) 140 (56) 146 (59)
Race: White 86 (26) 26 (10) 25 (10)
Region: N. America 86 (26) 26 (10) 25 (10)
Europe 205 (61) 184 (74) 182 (74)
Other 46 (14) 38 (15) 40 (16)
ECOG: 1 108 (32) 80 (32) 58 (23)
Stage: Unresectable IIIC 13 (4) 13 (5) 21 (16)
M1a 40 (12) 20 (16) 40 (16)
M1b 65 (19) 42 (17) 40 (59)
M1c 220 (65) 153 (62) 146 (59)
Prior Adjuvant Therapy 68 (20) 24 (10) 24 (10)
Prior Brain Metastasis 0 (0) 2 (1) 1 (0)
Elevated LDH 142 (42) 104 (42) 112 (45)
Baseline Population Characteristics
www.fda.gov
![Page 43: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/43.jpg)
43www.fda.gov
OS and PFS Across Trial Arms
![Page 44: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/44.jpg)
44
Results
www.fda.gov
Original Trial All External Controls Matched EC
Vem
N = 248
Cobi + Vem
N = 247
Vem
N = 337
Cobi + Vem
N = 247
Vem
N = 230
Cobi + Vem
N = 230
PF
S
Medians
(95% CI)
6.2
(5.6, 7.5)
9.9
(9.0, NR)
6.9
(6.3, 7.0)
9.9
(9.0, NR)
6.9
(6.6, 7.2)
11.1
(8.6, NR)
HR
(95% CI)
0.51
(0.39, 0.68)
0.53
(0.41, 0.68)
0.57
(0.43, 0.75)
OS
Medians
(95% CI)
17.4
(15.3, 20.5)
22.3
(20.4,NR)
13.6
(12.6, 15.4)
22.3
(20.4,NR)
15.4
(13.6, 19.6)
22.3
(20.3,NR)
HR
(95% CI)
0.70
(0.54, 0.89)
0.57
(0.45, 0.71)
0.65
(0.51, 0.83)
![Page 45: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/45.jpg)
45www.fda.gov
Comparing Original Trial to Matched Analysis
![Page 46: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/46.jpg)
46
Discussion and Future Considerations
• When may EC be most appropriate?
• Data considerations when using EC
– Compatibility of endpoints
– Temporality of data – even small lags may make a big difference (such as for prior adjuvant therapy)
– Treatment/disease/patient population idiosyncrasies
• Future work to include
– Other disease areas
– Combination of several trial arms for SCAwww.fda.gov
![Page 47: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/47.jpg)
47
Acknowledgements
• Susan Jin, FDA
• Raji Sridhara, FDA
• Gideon Blumenthal, FDA
• Marc Theoret, FDA
![Page 48: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/48.jpg)
48
APPENDIX
www.fda.gov
![Page 49: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/49.jpg)
49
Results: Sensitivity to Caliper
www.fda.gov
Caliper = 1 Caliper = 0.025 Caliper = 0.01
Vem
N = 230
Cobi + Vem
N = 230
Vem
N = 184
Cobi + Vem
N = 184
Vem
N = 153
Cobi + Vem
N = 153
PF
S
Medians
(95% CI)
6.9
(6.6, 7.2)
11.1
(8.6, NR)
6.9
(6.6, 7.2)
NR
(8.6, NR)
6.9
(6.6, 7.9)
NR
(11, NR)
HR
(95% CI)
0.56
(0.43, 0.74)
0.53
(0.39, 0.72)
0.51
(0.36, 0.72)
OS
Medians
(95% CI)
15.4
(13.6, 19.6)
22.3
(20.3,NR)
15.4
(13.5, 19.5)
22.1
(19.2, NR)
16.0
(13.4, 19.9)
22.3
(20.3,NR)
HR
(95% CI)
0.65
(0.51, 0.83)
0.64
(0.49, 0.85)
0.64
(0.47, 0.86)
![Page 50: Panel 1 - Friends of Cancer Research Annual Meeting Pane… · Panel 1 Participants Moderator: Elizabeth Stuart, Johns Hopkins University • Andrea Ferris, LUNGevity Foundation •](https://reader034.vdocument.in/reader034/viewer/2022052019/60324d206eeaec67e71b9d15/html5/thumbnails/50.jpg)
Panel 1 Participants
Moderator: Elizabeth Stuart, Johns Hopkins University
• Andrea Ferris, LUNGevity Foundation
• Antoine Yver, Daiichi Sankyo
• Joohee Sul, U.S. FDA
• Pallavi Mishra-Kalyani, U.S. FDA
• Rajeshwari Sridhara, U.S. FDA
• Ruthie Davi, Medidata Solutions
50
#FriendsAM18