Transcript Lee 1 - Division of Quantitative Sciences

Topics in Clinical Trials (1) - 2012
J. Jack Lee, Ph.D.
Department of Biostatistics
University of Texas
M. D. Anderson Cancer Center
1: Introduction to Clinical Trials
•
•
•
•
•
•
•
•
Why do we need them?
What is a clinical trial?
History of clinical trials
Clinical trial phases
Major aspects of clinical trials
Ethics of clinical trials
Study protocol
Examples
Why do we need clinical trials?
A researcher tried jalapenos on a stomach ulcer
patient, and the ulcer went away.
The researcher published an article:
“Jalapenos Cure Stomach Ulcers.”
The next patient subjected to the same treatment
died. The researcher published a follow-up article:
“More Detailed Study Reveals That
Jalapenos Cures 50% of Stomach Ulcers.”
Is erlotinib (an EGFR inhibitors)
effective in treating lung cancer?
Overall response rate:

Chemo – 30%
Erlotinib – 25%
Response rate by EGFR mutation
(a predictive marker)
EGFR Mutation
No (90%)
Yes (10%)
Chemo
30%
30%
Erlotinib
20%
70%
Data:
Signal + Noise
Data Analysis:
Extract the signal and
filter out the noise
Statistics:
Find a needle in a haystack
Finding A Needle In A Haystack
Data Tsunami
Statistics and Medical Research
Premise: Advance in medical practice is based
on research
Research validity is based on scientific credibility
Scientific Method:
QuestionDesignStudyDataAnalysisConclusion
 New Question
Good design + Good analysis  Valid Inference
Statistics is needed in every step
Sound statistics is the basis of evidence based
medicine (EBM)
Statistician is a guardian of science
Triumph of Medical Research and Clinical Trials
Discovery and Approval of STI571/Gleevec (Glivec)
1960-1980s:


Identify chronic myelogenous leukemia (CML) is associated with chromosome translocation of
the abl proto-oncogene, on chromosome 9 in non-CML patients and translocated to the
chromosome 22 in patients with CML.
abl oncogene is activated by this translocation, which leads to the formation of an altered
protein containing a piece of the Abl protein joined to a piece of a second protein, BCR. It is
this fused protein product, called BCR-ABL, which is abnormally expressed in about 95 percent
of CML patients.
1990:



Several labs showed that bcr-abl alone causes leukemia in mice.
STI571 was formulated and shown to inhibit the growth of BCR-ABL-expressing cells.
The preliminary results of this early study showed that 31 out of 31 patients who received at
least 300 milligrams daily had their blood counts return to normal. In nine of the 20 patients
who were treated for five months or longer, no cells with the Philadelphia chromosome could
be found.
2001:



February: Novartis submitted the New Drug Application for STI571, now known as Gleevec, to
FDA for the treatment of the late phases of CML.
April: Results of a larger study of STI571 in 83 patients were reported in New England Journal
of Medicine. In the 54 chronic-phrase CML patients who were treated with doses of 300
milligrams or more, normal blood counts were restored in 53, and in 29 of the 54 patients, the
Philadelphia chromosome disappeared. Most side effects were mild.
May: U.S. Food and Drug Administration approved the sale of STI571/Gleevec for CML.
Capdeville, Nature Review 2002
Vemurafenib for melanoma with BRAF V600E mutation
FDA Approval
August, 2011
Crizotinib for lung cancer with EML4-ALK translocation
FDA Approval
August, 2011
What is a clinical trial?
A clinical trial is a prospective study evaluating the
effect and value of intervention(s) in human beings
under pre-specified conditions.
A controlled clinical trial is a prospective study
comparing the effect and value of intervention(s)
against a control in human beings.
The clinical trial is the most definitive tool for
evaluation of the applicability of clinical research.
It represents a key research activity with the
potential to improve the quality of health care and
control costs through careful comparison of
alternative treatments.
A properly planned and executed clinical trials is a
powerful experimental technique for assessing the
effectiveness of an intervention.
Type of Research
Case Report
Observational Study



Retrospective
Cross-sectional
Prospective
Clinical Trial


Single arm vs. multi-arm trials
Controlled clinical trials
 Historical control
 Concurrent control



Randomization
Blinding
Single center vs. multi-center trials
History of Clinical Trials
Daniel 1:12-16. A diet experiment (Royal food and wine versus
vegetables and water) for 10 days
17th century: Lancaster (captain of a ship of East Indian Shipping
Co.) conducted an experiment to examine the effect of lemon juice
on scurvy for sailors. Ships with lemons were free of scurvy
compared to ships without lemons having scurvy.
Lind (1753) – Study of 5 tx for scurvy in 10 pts (2/tx) plus 2 pts
without tx (control).

First two pts given orange and lemons recovered quickly and was fit for
duty after 6 days, compared favorably with all other patients.
Smallpox experiment (1721) at the Newgate prison in Great Britain.
Voluntary inmates were inoculated and were free from smallpox.
Modern Clinical Trials (post WWII)
Medical Research Council (UK): treatment for common cold (1944),
tuberculosis (1948)
National Institutes of Health (US): Cancer trials (1960’s), Coronary
Drug Project (1965)
Observational studies lead to clinical trials

Cholesterol:
 Framingham Heart Study (1951): high cholesterol, high blood pressure,
smoking, and diabetes correlated with cardiovascular disease
 Scandinavian Simvistatin Survival Study (1994): statin reduced mortality

Hormone Replacement Therapy (HRT)
 Observation studies show that HRT reduced cardiovascular mortality and
morbidity
 HERS trial: HRT has no benefit but a risk for thrombosis (Grady, 1998)
 WHI: HRT has no CV benefit but has a harmful effect due to blood clotting

Beta-carotene
 Numerous epidemiology study showed benefit for protecting against cancers
 ATBC trial and CARET trial showed a harmful effect among current smokers.
Both trials stopped early.
 PHS showed that beta-carotene has no effect on mostly non-smokers
Clinical Trial Phases
Phase I: clinical pharmacology and toxicity
Phase II: Initial Assessment of Efficacy
Phase III: Full-scale Evaluation of
Treatment Efficacy
Phase IV: Postmarking Surveillance
Ref:
1.
Nottage M and Siu LL. Principles of clinical trial design. Journal of
Clinical Oncology 20:42s-46s, 2002.
2.
Lee JJ. Clinical trial design for anticancer therapies. In: The Cancer
Handbook, 2nd Ed. Ed(s) Alison M. Wiley: London, UK, 1330-44,
2007.
Phase I: clinical pharmacology and toxicity






1st experiment in human for new drug,
schedule, or combination
Primary concern: Safety
Goal: define the maximum tolerated dose
(MTD) in a dose-escalation study
Typically required 15-30 patients
Algorithm-based design: 3+3
Model-based design: continual reassessment
method (CRM)
Phase II: Initial Assessment of Efficacy




At MTD or recommended phase II dose,
examine the efficacy and refine the toxicity
profile of the agent
Goal: IIA – screen out ineffective drugs
IIB – sending promising agents to Phase III
Typically, require 30-100 patients in a multistage design
Commonly used design: Gehan’s design,
Simon’s two-stage design, randomized phase
II design
Phase III: Full-scale Evaluation of Treatment Efficacy




Compare new treatment with standard
treatment in a rigorous manner, e.g. a doubleblind randomized placebo-controlled study
Goal: define the ‘best’ treatment, which has an
implication of changing the current practice in
treating patients.
Typically required hundreds or even tens of
thousands patients in primary cancer
prevention trials.
Long-term, expensive
Phase IV: Postmarking Surveillance

Goal: monitoring the adverse effects, longterm morbidity and mortality after the
treatment being used in a large number of
patients and patients being followed up for a
long period of time
Examples
Pilot study (feasibility): N  18
Phase I (toxicity): 20  N  40
Phase II (efficacy): 30  N  100
Phase III (confirmatory): N > 100
Primary Prevention Trials: N > 10,000
e.g. BCPT (Tamoxifen): N=16,000
(13,388)
PHS (aspirin, b-carotene): N=22,071
Phase I
Pt. Characteristics
Phase II
Phase III
sicker
healthier
heavily pre-treated/
Untreated
failed std. tx
Treatment
poor prognosis
good prognosis
heterogeneous
homogeneous
2nd, 3rd, or 4th line
front-line therapy
therapy
Drug Development
Treatment of Choice
Two Big Enemies Against Finding the Truth and
Effective Weapons to Fight Against Them
Bias
Variation
To tackle bias
 Randomization
 Blinding
 Intent to treat principle
To tackle variation
 Control or reduce variation
 Increase sample size
Randomized Control Trials
Advantages of randomization



Remove the potential bias in treatment
assignment - conscious or subconscious
Randomization tends to produce comparable
groups - known or unknown prognostic
variables
Validity of statistical tests of significance is
guaranteed
Randomization
1st introduced by Fisher in 1926 in agriculture
research
1st clinical trial used randomization – Amberson
et al. (1931)


Matching 24 pts with pulmonary tuberculosis into 2
comparable groups of 12 each
Flip a coin to decide which group received sanocrysin,
a gold compound
Streptomycin trial by British Medical Research
Council (1948)

1st to use random numbers in allocation pt to
experimental or control groups
Random Number Table
2
0
1
7
4
2
2
8
2
3
1
7
5
9
6
6
3
8
6
1
0
2
1
0
8
6
1
0
5
1
5
5
9
2
5
2
4
4
2
5
7
4
4
9
0
4
1
9
0
3
0
4
1
0
3
3
5
3
7
0
1
1
5
4
4
8
6
3
9
4
6
0
9
4
4
9
5
7
3
8
9
4
7
0
4
9
3
1
3
8
6
7
2
3
4
2
2
9
6
5
4
0
8
8
7
8
7
1
3
7
1
8
4
8
6
4
0
6
5
7
2
2
1
5
7
8
1
5
6
9
8
4
3
2
5
2
3
2
5
4
1
5
1
2
5
4
0
2
0
1
3
7
3
8
3
7
1
2
9
3
9
3
2
9
1
2
1
8
2
7
3
0
3
0
5
5
9
1
8
7
5
0
5
7
5
8
5
1
4
9
3
6
1
2
5
3
9
6
4
0
4
5
0
4
7
7
9
7
3
6
1
4
9
9
4
5
5
2
9
5
6
9
8
5
0
3
8
3
5
1
8
7
8
5
5
6
2
2
3
7
4
4
9
1
9
9
4
9
8
9
3
9
9
4
6
0
4
8
4
9
0
6
7
7
6
4
7
2
5
9
2
6
0
8
5
1
2
5
5
7
1
6
2
3
9
1
0
2
1
9
9
6
4
7
5
9
8
9
6
5
2
7
8
4
3
0
9
2
6
3
3
7
2
6
2
4
2
3
6
6
0
4
5
0
6
5
0
4
6
5
6
5
8
2
4
2
7
0
5
1
5
5
0
4
6
1
4
7
8
8
8
3
9
9
3
4
8
2
3
7
3
2
7
0
1
7
7
2
0
3
6
1
6
6
2
6
2
4
7
1
2
2
7
7
8
8
3
3
1
7
7
8
0
8
9
2
7
3
4
9
Blindness
Amberson et al. (1931)

Pts were not aware of whether they received IV
injection of sanocrysin or distilled water
Single-blinded study

Either pts or physicians are blinded to the tx allocation
Double-blinded study

Both pts and physicians are blinded to the tx
allocation
Key: In general, it will be desirable that
whosoever evaluates the outcome should be
blinded to the treatment assignment
Controlled Trials
No treatment control


e.g.: standard practice is observation after surgery
Observation versus ‘adjuvant’ therapy
Different types of controls


Placebo
Active control
 use standard therapy
 use another new therapy

‘sham’ treatment control
 sham surgery or accupuncture
Reciprocal control



Tx A: smoking cessation counseling, no dietary
intervention
Tx B: dietary intervention, no smoking cessation
counseling
Two endpoints: Smoking cessation and weight loss
Primary Hypothesis and Endpoint
Each clinical trial must have a primary
question.
The primary question, as well as any
secondary or subsidiary questions, should
be carefully selected, clearly defined, and
stated in advance.
Hypothesis generation versus hypothesis
testing
Definition of primary endpoint(s)
Definition of secondary endpoints
Primary Objective
Define one question the investigators are
most interested in answering and is
capable of being adequately answered.
Define the primary endpoint

Toxicity, efficacy (response/survival), QOL
Define the type of study:


Hypothesis testing or estimation,
Superiority or equivalence trials
The sample size is based on.
Secondary Objectives
Different endpoints
Subgroup hypotheses



Prospectively defined
Based on reasonable expectations
Limited in number
Hypothesis testing vs. hypothesis
generating
Hunting expedition vs. fishing expedition
Multiplicity Issues
Examples of post-hoc subset analyses
In the International Study of Infarct
Survival-2 (ISIS-2), the treatment effect
seemed to differ by astrological sign:




More adverse effect of aspirin on mortality for
patients born under Gemini (5/21-6/21) or Libra
(9/23-10/23) than others (P < 0.00001).
Is the data convincing?
Do you believe it?
What is the one question you may ask to discern
whether the claim is true or not?
Response Variables
Dose limiting toxicities (DLT), complications
Response, incidence of a disease, total mortality,
death from a specific cause
Overall survival, time to progression, time to
cancer
Blood pressure, biomarkers, PSA, CD4 count
Quality of life
Cost and ease of administrating the intervention
In general, a single response variable should be
identified to answer the primary question.
General rules for response variable
Define the questions prospectively and specifically

Gleevec can increase the response rate from 50% to
80% in patients with chronic CML
The primary response variable can be assessed in
all participants and as completely as possible

Informative drop-out or lost to f/u due to toxicity
Participation generally ends when the primary
response variable occurs

Off-drug, off-study, extended f/u
Response variables should be unbiased assessed

Hard, objective endpoints vs. soft, subjective endpoints
Intent-to-treat (ITT) Principle
Unlike animal studies, investigator cannot
dictate what a participant should do in a
clinical trial.
A participant may forget to take the pills,
receive dose reduction due to toxicity, drop
out from the study at any point or lost to f/u.
Use only full compliers in the analysis? Use all
subjects?
ITT compares ‘intervention strategies’ and not
‘interventions.’
Ethics of Clinical Trials
Do no harm. Clinical trials must be reasonably
safe to participants and have a favorable riskbenefit ratio.
Proper informed consent is essential.
Rationale for randomization (clinical
equipoise) must be sufficiently justified.
Finder fee for recruitment can result in bias.
Unreasonable compensation to participants
can also cause problems.
Equal opportunities to be recruited and
treated, e.g.: race and gender.
Individual ethics and group ethics must be
balanced.
Protect Human Subjects
Nuremburg Code (1947): In response to Nazi
atrocities of using concentration camp prisoners
for human experiments.

voluntary consent, avoid unnecessary suffering,
subject/scientist free to end experiment
Declaration of Helsinki (1964, 1989)

informed consent
Belmont Report (1979)

risk/benefit assessment, fair selection of subjects
Human Subject Committee (HSC), Institutional
Review Board (IRB), Ethics Committee,
Surveillance Committee
STATISTICAL PRINCIPLES FOR
CLINICAL TRIALS (ICH E9)
International Conference on Harmonisation
encompassing three main ICH regions:
Europe, Japan and the U.S.A.
An effort to establish clinical trial standards by
integrating inputs from government
(regulatory authority), industry, and academia.






A comprehensive guideline was developed in 1998
and published in Lewis, Statistics in Medicine
18:1903-1942,1999.
E10: Choice of Control Group in Clinical Trials
E9: Statistical Principles for Clinical Trials
E8: General Considerations for Clinical Trials
http://www.emea.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09
/WC500002928.pdf
http://www.ich.org/products/guidelines/efficacy/efficacy-single/article/statisticalprinciples-for-clinical-trials.html
CONTENTS
1. Introduction . .
1.1 Background and Purpose
1.2 Scope and Direction
2. Considerations for overall clinical development
2.1 Trial Context
2.1.1 Development Plan
2.1.2 Confirmatory Trial
2.1.3 Exploratory Trial
2.2 Scope of Trials
2.2.1 Population
2.2.2 Primary and Secondary Variables
2.2.3 Composite Variables
2.2.4 Global Assessment Variables
2.2.5 Multiple Primary Variables
2.2.6 Surrogate Variables
2.2.7 Categorized Variables
2.3 Design Techniques to Avoid Bias
2.3.1 Blinding
2.3.2 Randomization
3. Trial design considerations
3.1 Design Configuration
3.1.1 Parallel Group Design
3.1.2 Cross-over Design
3.1.3 Factorial Designs
3.2 Multi-centre Trials
3.3 Type of Comparison
3.3.1 Trials to Show Superiority
3.3.2 Trials to Show Equivalence or Non-inferiority
3.3.3 Trials to Show Dose-response Relationship
3.4 Group Sequential Designs
3.5 Sample Size
3.6 Data Capture and Processing
CONTENTS (cont.)
4. Trial conduct considerations
4.1 Trial Monitoring and Interim Analysis
4.2 Changes in Inclusion and Exclusion Criteria
4.3 Accrual Rates
4.4 Sample Size Adjustment
4.5 Interim Analysis and Early Stopping
4.6 Role of Independent Data Monitoring Committee (IDMC)
5. Data analysis considerations
5.1 Prespecification of the Analysis
5.2 Analysis Sets
5.2.1 Full Analysis Set
5.2.2 Per Protocol Set
5.2.3 Roles of the Different Analysis Sets
5.3 Missing Values and Outliers
5.4 Data Transformation
5.5 Estimation, Confidence Intervals and Hypothesis Testing
5.6 Adjustment of Significance and Confidence Levels
5.7 Subgroups, Interactions and Covariates
5.8 Integrity of Data and Computer Software Validity
6. Evaluation of safety and tolerability
6.1 Scope of Evaluation
6.2 Choice of Variables and Data Collection
6.3 Set of Subjects to be Evaluated and Presentation of Data
6.4 Statistical Evaluation
6.5 Integrated Summary
7. Reporting .
7.1 Evaluation and Reporting
7.2 Summarizing the Clinical Database
7.2.1 Efficacy Data
7.2.2 Safety Data
Glossary
Study Protocol
Every well-designed study required a
protocol.
Protocol is a written agreement between
investigators, participants, and the
scientific community.
Protocol is a comprehensive operational
manual. It specifies the standard operation
procedure (SOP).
Examples
Intergroup Lung Study
Study Objectives
To evaluate the efficacy of 13cRA in
reducing the incidence of second
primary tumors (SPTs)
To evaluate quantitative and
qualitative toxicity of 13cRA 30
mg/day
To compare overall survival of
patients treated with 13cRA vs.
patients treated with placebo
Intergroup Lung Study
Patient Eligibility
Stage I non-small lung cancer, currently free of
disease, between 6 wks and 36 mos from resection
No prior or concurrent therapy for lung cancer
other than surgery
Normal hepatic, renal and bone marrow function
Fasting triglycerides < 320 mg/dL
Life expectancy > 12 mos & Zubrod 0-2
No cancer history within last 5 years
No synchronous lesions
No history of high dose Vitamin A or beta-carotene
Intergroup Lung Study
R
E
G
I
S
T
R
A
T
I
O
N
8 Week
Run-in
R
A
N
D
O
M
I
Z
A
T
I
O
N
T
R
E
A
T
M
E
N
T
13cRA
Placebo
3 Years
F
O
L
L
O
W
U
P
4 Years
O
F
F
S
T
U
D
Y
Study Design
Randomized, placebo controlled trial to compare
the efficacy of 13cRA vs. placebo in reducing SPT
N=1,379 registered; 1,242 randomized
Annual SPT rate:
3% in the placebo group
SPT rate reduced by 50% in the 13cRA group
3 years of accrual + 4 years of f/u or
4 years accrual + 3 years of f/u
Total study duration = 7 years from 1st patient
randomized
Planned Interim Analyses based on the
O’Brien-Fleming Boundary
Primary endpoint: Time to SPT
Early stopping if Ho: No difference between
arms is rejected
No early stopping based on futility
Level of Significance
Time since 1st pt randomized*
1st Analysis:
0.0005
3 years
2nd Analysis:
0.014
5 years
Final Analysis:
0.045
7 years
* The 1st patient was randomized on 2/5/93.
Trial History
1,486 pts were enrolled from 12/8/92 to 4/9/97
1,304 were randomized from 2/5/93 to 6/23/97
Among them, 1,265 were randomized and eligible
Major Analyses




3/96:
7/97:
7/98:
2/00:
1st interim analysis
2nd interim analysis
3rd interim analysis
4th and final analysis
1,486 Registered
ECOG, 23.8%; CALGB, 22.2%; SWOG, 15.1%; RTOG, 14.4%;
NCCTG, 12.5%; MDACC-CCOP, 6.3%; MDACC/AFFILIATES, 5.7%
1,444 Registered and eligible (planned 1,379)
1,304 Completed run-in and randomized (planned 1,241)
1,265 Randomized and eligible
1,166 Properly randomized and eligible
Placebo
N = 577
Year 0-1
14 deaths + 26 censored
Years 1-3
83 deaths + 152 censored
Years 3-5
33 deaths + 159 censored
Isotretinoin
N = 589
537
Year 0-1
32 deaths + 35 censored
522
302
Years 1-3
76 deaths + 140 censored
306
110
Years 3-5
28 deaths + 174 censored
104
Consort Diagram: Colin et al., JAMA 276: 637-639 1996.
Lung Intergroup Trial: Lippman et al. JNCI 93:605-618, 2001.
Table 1. Characteristics of randomized, eligible patients by study arm*
Characteristics
Placebo (N = 577)
Isotretinoin (N = 589)
Total (N = 1166)
66.0 (34-81)
64.1 (8.9)
65.0 (31-86)
64.3 (8.6)
65 (31-86)
64.2 (8.8)
N (%)
N (%)
N (%)
Gender
Female
Male
251 (43.5%)
326 (56.5%)
248 (42.1%)
341 (57.9%)
499 (42.8%)
667 (57.2%)
Race
White
Black
Other
525 (91.0%)
41 (7.1%)
11 (1.9%)
552 (93.7%)
31 (5.3%)
6 (1.0%)
1077 (92.4%)
72 (6.2%)
17 (1.5%)
Histology
Squamous
Non-Squamous
191 (33.1%)
386 (66.9%)
187 (31.8%)
402 (68.2%)
378 (32.4%)
788 (67.6%)
T-Stage
T1
T2
315 (54.6%)
262 (45.4%)
316 (53.7%)
273 (46.4%)
631 (54.1%)
535 (45.9%)
Smoking Status
Current
Former
Never
231 (40.0%)
301 (52.2%)
45 (7.8%)
225 (38.2%)
319 (54.2%)
45 (7.6%)
456 (39.1%)
620 (53.2%)
90 (7.7%)
Age
Median (Range)
Mean (S.D.)
*S.D. = standard deviation.
Event Charts
Calendar event chart, Interval event chart,
Goldman’s event chart
Graphical tools to track and plot multiple timed
event data at the individual level
Highly effective in monitoring clinical trial
Very useful for assessing covariate effect
Complementary to survival plots
(Lee, Hess, and Dubin: Extension and application of event
charts. The American Statistician, 54: 63-70, 2000)
0
500
1000
1500
Calendar Event Chart for ID91-025
12/8/92
10/18/94
8/27/96
Study Date
7/8/98
5/17/00
0
500
1000
1500
Interval Event Chart for ID91-025
0
2
4
Follow-up Time
6