efficacy analysis and graphical representation in … sde presentations/efficacy analysis... ·...
TRANSCRIPT
Efficacy analysis and graphical representation in Oncology trials - A case study
Anindita Bhattacharjee Vijayalakshmi Indana
Cytel, Pune
The views expressed in this presentation are our own and do not necessarily represent the views of Cytel Statistical Software & Services Limited
1
Agenda
Ø Oncology endpoints
Ø A Case Study Ø Analysis Ø Graphical Representation
Ø Take away points
2
Oncology endpoints Ø Early phase – maximum tolerated dose/
recommended phase 2 dose, biological drug activity
Ø Late phase – Clinical benefit
Ø Endpoint choice depends on – indication, line of
therapy, available treatment options, etc.
3
Time to Event (Survival) Endpoints
4
Endpoint Definition Advantages Disadvantages
Overall Survival (OS)
Randomization until death
Precise and easy to measure – most reliable
May involve larger studies
Progression Free Survival (PFS)
Randomization until progression/ death
Smaller sample size
Not precisely measured
Oncology endpoints
Response and Symptom Endpoints
5
Endpoint Definition Advantages Disadvantages Objective Response Rate (ORR)
Proportion of responders (Complete or Partial)
Assessed earlier and in smaller studies
Not a direct measure of benefit
Symptom Endpoints
Patient’s quality of life (QOL)
Patient perspective of direct clinical benefit
Data are frequently missing or incomplete
Oncology endpoints
Time to Event data: Concepts
X
�
�
Event Death, disease occurrence, disease recurrence, recovery, or other experience of interest
Censoring When a subject does not have an event of interest during the
observation interval
Time (months)
1 2 3 4 5 6 7 8
Patie
nts
6
Analysis Timepoint
Event
Censored
Censored
X� Censored
Time to Event data: Concepts
X
�
�
Time (months)
1 2 3 4 5 6 7 8
Patie
nts
7
Analysis Timepoint
Event
Censored
Censored
X� Censored
Prevent Loss of information
Retain original sample size – as decided in the hypothesis in Protocol
Case Study – Efficacy Analysis
Ø Protocol
Ø Analysis Dataset Ø Derivation
Ø Graphical Analysis Ø Primary and secondary endpoints
8
Protocol
9
Title Ø A Phase III randomized lung cancer study, two
arms
Primary Endpoint Ø Progression Free Survival (PFS)
Secondary Endpoint Ø Objective response rate(ORR)
PFS (with event)
Rando-mization
Treatment Start
Disease Progression
Death
Time to First Event occurring
Randomization Date(RANDT)
Progression Date (PDDT)
PFS (in days) = (PDDT - RANDT + 1) Censor = 0
10
PFS (with censoring)
Rando-mization
Treatment Start
Last TA Discontinued Study
Time till last tumor assessment indicating lack
of progression
Randomization Date(RANDT)
Last TA (TADT)
PFS (in days) = (TADT - RANDT + 1) Censor = 1
11
Analysis dataset (ADPFS)
12
Unique Subject Identifier
Treatment Group
PFS Date
PFS Time (months)
Censoring Flag
Disposition/ Response/ Tumor Assessment
Primary Endpoint
Ø Estimates the probability of survival to a given time
using the proportion of patients who have survived
to that time
Ø Accounts for censoring
14
Kaplan-Meier Survival Analysis Method:
20
60
40
80
Surv
ival
Pro
babi
lity
(%)
Trt 1 : (N=42) Trt 2 : (N=43) Number of events Trt 1: 26 Trt 2: 39 Kaplan Meier PFS Trt 1: 14.29 months Trt 2: 5.95 months
42 43
14 1
18 9
Trt 1 Trt 2
30 20
15 3
9 1
Trt 1 Trt 2
Patients at risk # 0 6 12 18 24 30 36 Time (months)
Median Survival Time
6 months
proc lifetest data=adpfs method=km outsurv=kmsurv; time pfstime*censor(0); strata trt; run;
100
14 months
20
60
40
80
Surv
ival
Pro
babi
lity
(%)
Trt 1 : (N=42) Trt 2 : (N=43) Number of events Trt 1: 26 Trt 2: 39 Kaplan Meier PFS Trt 1: 14.29 months Trt 2: 5.95 months
42 43
14 1
18 9
Trt 1 Trt 2
30 20
15 3
9 1
Trt 1 Trt 2
Patients at risk # 0 6 12 18 24 30 36 Time (months)
At month 18 Trt 1=15 Trt 2=3
At month 30 Trt 1=9 Trt 2=1
100
At month 6 Trt 1=30 Trt 2=20
At month 0 Trt 1=42 Trt 2=43
Secondary endpoints: Ø Objective Response Rate can be analyzed using a
Waterfall plot
17
Ø Depicts increase or decrease in rate for a parameter
of interest.
Waterfall Plot
18
PR (Partial Response)
PD(Progressive Disease)
Subjects
% c
hang
e fr
om b
asel
ine
(mea
sura
ble
lesi
on)
Decrease in best percentage change from baseline 53.13% (17) Increase in best percentage change from baseline 40.63% (13)
Waterfall Plot
19
% c
hang
e fr
om b
asel
ine
(mea
sura
ble
lesi
on)
Subjects
Decrease in best percentage change from baseline 28.21% (11) Increase in best percentage change from baseline 64.10% (25)
Waterfall Plot
20
Subjects Subjects
Decrease in best percentage change from baseline 53.13%(17) 28.21%(11) Increase in best percentage change from baseline 40.63%(13) 64.10%(25)
Take away points
Ø Censoring algorithm
Ø Latest tumor evaluation
Ø Last contact date
Ø Randomization date
Ø Data checks – raise flag
Ø Missing data (e. g. missing PFS)
Ø Cross check across Tables, Listings and graphs
Ø Heavy censoring
23
References
24
Ø Guidance for Industry Clinical Trial Endpoints for the Approval of Cancer Drugs
and Biologics
Ø FDA's Richard Pazdur: Drug Approval Entails Evaluation of Clinical Benefit, Not
Just Endpoints
Ø Oncology Clinical Trials Successful Design, Conduct and Analysis – W.M. Kevin
Kelly, Susan Halabi
Ø Thomas R. Fleming, Mark D. Rothmann, and Hong Laura Lu - Journal
Of Clinical Oncology - Issues in Using Progression-Free Survival When Evaluating
Oncology Products - J Clin Oncol 27:2874-2880, 2009