Understanding Clinical Trial Data Through Use of...
Transcript of Understanding Clinical Trial Data Through Use of...
![Page 1: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/1.jpg)
Understanding Clinical Trial Data Through Use of Statistical Graphics
Will BushnellGroup Director GSK Oncology
GlaxoSmithKline
![Page 2: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/2.jpg)
A picture tells a thousand words
Lake Blanche
![Page 3: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/3.jpg)
A picture tells a thousand words
![Page 4: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/4.jpg)
Outline
Introduction
Patient Safety– Exposure– Laboratory data– Hy’s law– AE relative risk
Efficacy data– Waterfall plots– Skyline plots– Forest plots
![Page 5: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/5.jpg)
Review of safety data
Safety data from clinical trials is usually evaluated through simple summary tables and review of individual patient data
– Formal analysis much less developed than for efficacy– Scan tables and patient listings and highlight “important” results in
textual summaries
The safety of a molecule is best understood by understanding data at the individual patient level
Ideal opportunity to use graphical methods– Present concise summaries– Communicate main messages– Increase efficiency of review
![Page 6: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/6.jpg)
Exposure
Before looking at safety we need to answer a key question:
How many for how long and at what dose?– What is the exposure underlying the safety profile– Tabular summaries are useful….
![Page 7: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/7.jpg)
Summary of Exposure
187Median
7Min
500Max
198Mean
225N
Exposure(Days)
![Page 8: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/8.jpg)
Summary of Exposure
![Page 9: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/9.jpg)
A screening tool for AEs
![Page 10: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/10.jpg)
Lab Data
![Page 11: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/11.jpg)
Routine Summary Statistics for Liver Function
2%25%Bilirubin(n=200)
12%35%AST(n=200)
15%40%ALT(n=200)
Elevations > 3XULN
Any Elevation
This tabular summary does not tell the whole story
![Page 12: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/12.jpg)
Consider a Simple Graph
![Page 13: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/13.jpg)
Graphical Style
Graphical equivalent of a “shift table”
Reference lines at clinically important levels
Focus on upper left quadrant of each panel
Position legend in unused part of graph with statement of numbers of patients
May need legend for each panel to account for varying patient numbers
Plot control group last to more readily identify Tx effect
Consider displaying on log scale to account for skewed distribution
![Page 14: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/14.jpg)
Some Relevant Questions
Do ALT and AST track together?
Are there simultaneous elevations in ALT and Bilirubin?
…
![Page 15: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/15.jpg)
Another Simple Scatter Matrix
![Page 16: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/16.jpg)
Hy’s Law
“Hy’s Law” was developed by Hyman Zimmerman as criteria for evaluating drug induced liver injury
The signal for potential for severe drug-related hepatotoxicityhas three components:
– 1. Elevation of transaminases >3 ULN: ALT AST
– 2. Concomitant elevation of Bilirubin
– 3. Absence of other etiology
“Hy’s Law” could be evaluated as part of the scatter matrix by generating different symbols for subjects who meet the criteria
![Page 17: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/17.jpg)
Another version of the Scatter Matrix – Some Hypothetical Data
10^-0.5
1
3
10
32
10^-0.5 1 3 10 32 10^-0.5 1 3 10 32 10^-0.5 1 3 10 32
10^-0.5
1
3
10
32
10^-0.5
1
3
10
32
ALKP (xULN) ALT (xULN) AST (xULN)
Tota
l.Bili
. (xU
LN)
AS
T (x
ULN
)A
LT (x
ULN
)
![Page 18: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/18.jpg)
Patient Profiles
![Page 19: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/19.jpg)
Waterfall Plot for Evaluating Changes from Baseline
![Page 20: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/20.jpg)
Waterfall plot showing best volumetric reduction of brain tumors
Subject20 40 60 80 100 120 140 160 180 200
-100
0
50
100
150
200
250
300
350
400
450
-20-50
Volu
met
ric c
hang
e fr
om b
asel
ine,
%
0
Lin et al, ASCO 2007 N=194
![Page 21: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/21.jpg)
Independent-Reviewer Assessed Percent Reduction in Tumor Measurement at 12 Weeks From Baseline
Hutson et al, ASCO 2007 N=225
![Page 22: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/22.jpg)
Graphical style of “waterfall” plot
A method for looking at tumour shrinkage
Can be adapted for other data
Displays the distribution by looking at the order statistics
Color and patterns to differentiate magnitude of changes
Can display distributional location shifts in a comparative setting
– Beware if randomization is not 1:1
![Page 23: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/23.jpg)
Some examples from regulatory review
![Page 24: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/24.jpg)
A forest plot to make a point
HR=1.0
![Page 25: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/25.jpg)
Skyline Plot, examination of time course of hazard ratio
From FDA statistical review summary of Lapatinib
![Page 26: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/26.jpg)
Summary and Conclusions
Lake Blanche on Sunday
Exposure critical first step in safety analysis
Evaluation of patient safety from clinical trials– summary stats alone of limited value– graphics must provide identification of outliers – careful examination of individual patient data
Waterfall plots valuable for change data
Skyline plots indicate the maturity of an estimate
![Page 27: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/27.jpg)
Acknowledgements
The graphics were developed by a team in the Biostatistics Division of GSK:
Frank ManinoPeter LaneRandall AustinNada BoudiafMike ColopySusan DukeRich HeibergerGurudutt LingashastryDaniel LorandDaniel ParkShi-Tao YehMichael DuranteTheresa CroftsMichelle Casey
![Page 28: Understanding Clinical Trial Data Through Use of ...stat-computing.org/events/2007-jsm/bushnell.pdf · Understanding Clinical Trial Data Through Use of Statistical Graphics Will Bushnell](https://reader031.fdocuments.us/reader031/viewer/2022022422/5a9d79e27f8b9a21688b74c0/html5/thumbnails/28.jpg)
References
1. Zimmerman, HJ, 1978, Drug-Induced Liver Disease, In: Hepatotoxicity, The Adverse Effects of Drugs and Other Chemicals on the Liver, 1st ed., pp. 351-3, Appleton-Century-Crofts, New York.
2. Ratain MJ, Eisen T, Stadler WM, et al. Phase II placebo-controlled randomized discontinuation trial of sorafenib in patients with metastatic renal cell carcinoma. J Clin Oncol 2006;24:2505-2512
3. A. Di Leo, H. Gomez, Z. Aziz, Z. Zvirbule, M. Arbushites,M. Koehler, L. S. Williams, J. Dering,R. S. Finn, Lapatinib With Paclitaxel Versus Paclitaxel as First-Line Treatment for Patients With Metastatic Breast Cancer: A Phase III Randomized, Double-blind Study in 580 Patients, ASCO, 2007
4. Pazopanib (GW786034) is active in metastatic renal cell carcinoma (RCC): Interim results of a phase II randomised discontinuation trial (RDT). TE Hutson, ID Davis, JP Machiels, PL deSouza, BF Hong, S Rottey, KL Baker, T Crofts, L Pandidte, R Figlin. ASCO 20007