CLINICAL DESIGN PHASE IIICLINICAL DESIGN PHASE IIICLINICAL DESIGN PHASE III CLINICAL DESIGN PHASE III CLINICAL TRIALS AND ENPOINTSCLINICAL TRIALS AND ENPOINTS
Evandro de Azambuja, MD, PhDJules Bordet InstituteJules Bordet InstituteBREAST data centerBrussels, Belgium
ESMO Young Oncologist Committee
[email protected] @https://www.br-e-a-s-t.org
I h fli fI h fli fI have no conflict of I have no conflict of interest to declareinterest to declareinterest to declareinterest to declare
OUTLINEOUTLINE
• Oncology drug developmentgy g p
• Why drug development fails in phase III trials?
• Classic Phase III Study Design
• The targeted therapy era and challenges forThe targeted therapy era and challenges for phase III trials
Cli i l d i t i h III t i l• Clinical endpoints in phase III trials
• Surrogate endpoints in phase III oncology trialsg p p gy
• Novel strategies to move forward in drug developmentdevelopment
ONCOLOGY DRUG DEVELOPMENTONCOLOGY DRUG DEVELOPMENT
Oncology Drug DevelopmentOncology Drug DevelopmentTHE GOOD & BAD NEWSTHE GOOD & BAD NEWSTHE GOOD & BAD NEWSTHE GOOD & BAD NEWS
THE GOOD NEWSTHE GOOD NEWS THE BAD NEWSTHE BAD NEWS
• Given priority ratings by the FDA
• Faster review of marketing applications
• Estimated true cost of a drug reaching the market is US$0.8‐1.0 Billion2
• For oncology Drugs:
• The FDA reviewed oncology drugs, on average, 6 months faster than other drugs1
• Longer US clinical development times
• ~1 year longer from of clinical testing to US regulatory approval
M di t iti ti i 7 8 6 3• Median transition time is 7.8 vs 6.3yrs for non oncology drugs
WHYWHY
• Difficulties in recruitment
• Longer times needed to establish efficacy (especially for Overall Survival)
• Targeted era Mean clinical development and regulatory approval times for Targeted era
1. DiMasi, J Clin Oncol 2007; 25(2):2092.Walker, Nature Reviews, Drug Development Jan 2009, 8:15-17
new oncology therapies approved by FDA from 1990 to 2005
Oncology Drug DevelopmentOncology Drug Development
TRANSITION FROM PHASE I TO PHASE IIITRANSITION FROM PHASE I TO PHASE III
• 50% of oncology drugs that entered phase III NEVER make it to US regulatory approval1
• Success rates can be increased by:– Longer development timeframe for drugs1 Clinical phase transition probabilities for investigational
• Longer development time in period 1993 to 1997 meant approximately 1/5 drugs reached regulatory
h d d l6370
se
oncology compounds from the firms by pharmaceutical sales (2005) by period which compound first entered clinical testing
• But shorter drug development time in the period 1993 to 2002 meant approximately only 1/ 4 reached regulatory approval
22
38
30
40
50
60
f Dru
gs in
Pha
str
ials
(%
)
– Higher response rates in earlier stages of development2
• Balance of competing factors
22
0
10
20
<5% 5-10% >10%
Effic
acy
of III t
(%)• Balance of competing factors <5% 5 10% >10%Response Rate in Phase I trials (%)
1. DiMasi, J Clin Oncol 2007; 25(2):2092.Sekine I. et al., Annals of Oncology 2002; 13(8):1306
Why do Drugs fail at Phase III Development?Why do Drugs fail at Phase III Development?Why do Drugs fail at Phase III Development?Why do Drugs fail at Phase III Development?
d d• Wrong study design
• Wrong doseo g dose
• Wrong schedule
• Wrong combination
• Wrong comparatorWrong comparator
• Wrong endpoint
• Wrong tumour type
• Weak phase 2 dataWeak phase 2 data
CLASSIC PHASE III STUDY DESIGNSCLASSIC PHASE III STUDY DESIGNS
Trial Designs: Parallel DesignTrial Designs: Parallel Design
Parallel DesignParallel Design
• Each patient receives one i d f hiPOPULATION treatment in a random fashion
• Simple and easy to implement
l bl d
POPULATION
Assessment & ScreeningAssessment & Screening
Eligible patients consent • Applicable to acute conditions
• The interpretation of the results i t i htf d
g p
Allocation/RandomisationAllocation/Randomisation
TREATMENT A TREATMENT B is straightforward
• Good if inter‐patient variability exists
TREATMENT A TREATMENT B
AssessmentAssessment
existsData AnalysisData Analysis
Trial Designs: Cross OverTrial Designs: Cross Over
• Same patients receive different treatments at different dosing Cross OverCross Over gperiods
• Patients serve as their own l h
POPULATIONPOPULATION
controls: removes the inter‐patient variability from the comparison between treatments
Assessment & ScreeningAssessment & Screening
Eligible patients consent Eligible patients consent
Allocation/Randomisation between treatments
• Typically requires a much smaller sample size than a parallel design
/
TREATMENT ATREATMENT A TREATMENT BTREATMENT B
WASH OUT sample size than a parallel design
• Should only be used for chronic diseases
TREATMENT BTREATMENT B TREATMENT ATREATMENT A
WASH OUT
• Need to be careful with ”carryover” effect and washout
AssessmentAssessment
Data AnalysisData Analysis
period
Adapted from Hoering A. et al Clin Can Res 2008; 14:4358
Trial Designs: 2 X 2 Factorial DesignTrial Designs: 2 X 2 Factorial Design
2 x 2 Factorial Design2 x 2 Factorial Design
• Attempts to evaluate two or more treatments in a single experimentll l h• Allows to evaluate the interaction
effect between treatments
M f t i l t i l l k ffi i t
Trt A Control
Trt B a b
Control c d • Many factorial trials lack sufficient power to look at the interaction effecta = A +B effect
• 2x2x2 or more complex design: require large sample size and the
b = B + Control
c = A + Control
d C l C lq g p
results are more difficult to interpret.
d = Control + Control
Adapted from Hoering A. et al Clin Can Res 2008; 14:4358
The Targeted Therapy Era: The Targeted Therapy Era: Challenges in the Development of newChallenges in the Development of newChallenges in the Development of new Challenges in the Development of new
Therapeutic AgentsTherapeutic Agents
• Change in type of cancer therapies developed and the clinical practice associated with these therapies
• Greater than 90% of all new oncology agents in clinical trials are targeted therapies
Th t h h i f ti• These agents have new mechanisms of action
• New agents pose major challenges for drug development– Predominately cytostatic drugs rather than cytotoxic y y g y
– Often produce less toxic side effects than conventional cytotoxic agents and therefore have a broad therapeutic margin
– Continued beyond progression in some casesContinued beyond progression in some cases
• Conventional trial design is a potential reason for the high percentage of drug failure in Phase III clinical trials
• Modification to current trial design needs to take into consideration these challenges
Example of Trial Design FailureExample of Trial Design FailurePhase III Equivalence Study of BAY 12-9566 in advanced pancreatic cancer
GEMCITABINERA
FOLLOW UP AND ASSESSMENT
Patients with Advanced Pancreatic
GEMCITABINEANDOM
First Interim Analysis
SecondInterim Analysis
Trial Terminated
Pancreatic cancer
BAY-12 9566
MISED
Analysis Analysis
TRIAL CONTINUED
TRIAL CONTINUED
• BAY 12‐9566: Metalloproteinase inhibitor with efficacy in pancreatic cancers, SCLC,
D
ovarian and breast• Trials in pancreatic and SCLC prematurely terminated due to methodological problems• First interim analysis: Termination rule: <6/30 patients on BAY 12‐9556 were
i f t 2 thprogression free at 2 months– Analysis failed to demonstrate this
• Second Interim Analysis: Completed after 227 patients enrolledG it bi ffi i ifi tl b tt– Gemcitabine efficacy significantly better
– Immediate trial termination• Trials in lung cancer and ovarian cancer also terminated
Parulekar W.R. et al Annal Oncol 2002; 13(supple 4): 139
Example of Trial Design FailureExample of Trial Design Failure
• What were the problems:What were the problems:– Inadequate number of patients at first interim analysis
Interim analysis conducted too early– Interim analysis conducted too early
– Trial design: consider pilot phase II or integrative Phase II/III designII/III design
– Trial arm design: Gemcitabine vs. Gemcitabine + BAY 12‐95569556
– Wrong study population: may have demonstrated benefit in patients with low tumour burden rather than advanced pdisease as cytostatic
IMPROVEMENTS IN PHASE III DESIGNIMPROVEMENTS IN PHASE III DESIGN
Improvement in Phase III Study DesignImprovement in Phase III Study Design
1. Enriching the patient population– Pre‐screening
– Trial designs based around a pre‐screened population• Example HER‐2 positive patients
2. Testing patients with an earlier stage disease1. Adjuvantj
2. Neo‐adjuvant
3. Cancer chemopreventionp
Novel Strategies in Drug Development: Novel Strategies in Drug Development: Th NTh N Adj t S ttiAdj t S ttiThe NeoThe Neo‐‐Adjuvant SettingAdjuvant Setting
Biological Window pCR
C
LapatinibPaclitaxel
RA S 24.7%
pCR
C PaclitaxelAND
SUR
Invasive Operable HER2+ BC
C
TrastuzumabPaclitaxel
DOM
RGE
29.5%
HER2+ BCT>2cmLVEF ≥ 50%
CTrastuzumabLapatinib
MIZ
ERY 51.3%
≥ 50%N=450
CTrastuzumabPaclitaxel
ZE
Y
+ 12 weeks6 weeks
Baselga J et al, SABCS 2010
CLINICAL ENDPOINTSCLINICAL ENDPOINTS
Clinical Endpoints Definition: Clinical Endpoints Definition: Colorectal CancerColorectal CancerColorectal CancerColorectal Cancer
Comparability between studies
D fi i i DFSPETACC‐31 MOSAIC2
Definition DFS Stage III CRC Irinotecan/ 5‐FU/FA vs 5‐FU/FA
Stage IIFOLOFOX vs 5‐FU/FA
Locoregional recurrence E E
Distant Metastases E E
Second primary, same cancer E E
d hSecond primary, other cancer E I
Death from same cancer E E
Death from other cancer E EDeath from other cancer E E
Non‐cancer related death E E
Treatment related death E E
Loss to follow up C C
Failed to meet primary endpoint
Met Primary endpointendpoint
DFS=Disease free survival; E=Event; I=Ignored; C=Censored
1.Van Cutsem E., Proc Am Soc Clin Oncol 2005;23:10902. Andre T, N Eng J Med 2004; 350:2343
Clinical Endpoints Definition: Clinical Endpoints Definition: BreastBreast CCancerancer
Definition of DFS1 BIG‐1982 MA‐173 ATAC4 IES5 ARNO6
Breast Breast CCancerancer
Local/Regional recurrence ✔ ✔ ✔ ✔ ✔
Distant Metastasis ✔ ✔ ✔ ✔ ✔Distant Metastasis ✔ ✔ ✔ ✔ ✔
Death from any cause ✔ ✔ ✔
Invasive contralateral breast cancer ✔ ✔ ✔ ✔ ✔
Second primary invasive cancer (non breast) ✔
Ipsilateral DCIS ✔ ✔
Contralateral DCIS ✔ ✔
Ipsilateral LCIS ✔
Contralateral LCIS ✔Contralateral LCIS ✔
DFS=Disease free survival
1.Hudis CA et al, J Clin Oncol, 2007; 25:2127-21321.Hudis CA et al, J Clin Oncol, 2007; 25:2127 21322.Thurliman B et al, N Eng J Med, 2005; 353:2747
3.Goss PE et al, N Eng J Med, 2003; 349:1793-18024. Baum M et al, Lancet, 2002; 359:2131-2139
5. Coombes RC et al, N Eng J Med, 2004; 350:1081-10926. Jakesz R et al, Lancet, 2005; 366:455-462
Problems with Clinical Problems with Clinical EEndpoint ndpoint DDefinitionsefinitions
No formal consensus of definitions
Not consistently
used
Formal definitions/expert
opinions exist only for a fewopinions exist only for a few
cancer sites
Clinical Endpoints in Phase III Trials: Clinical Endpoints in Phase III Trials: OverallOverall SSurvivalurvivalOverall Overall SSurvivalurvival
Confounder:Lines of Rx
Easy to measure& interpret Confounder:
Cause deathSimple to measure
Extended
Cause death
GOLD standard
Extended Follow‐up
Advantages Disadvantages
Alternatives to Conventional Alternatives to Conventional EEndpointsndpoints
SURROGATE END POINTMetastatic trials PFS instead of Overall survivalAdjuvant trials DFS instead of Overall survival
SURROGATE END POINTMetastatic trials PFS instead of Overall survivalAdjuvant trials DFS instead of Overall survival
h h d
ADVANTAGES OF SURROGATE END POINTSADVANTAGES OF SURROGATE END POINTS
l d l
DISADVANTAGES OF SURROGATE ENDPOINTSDISADVANTAGES OF SURROGATE ENDPOINTS
• Reach the surrogate endpoint more rapidly
Ad t f t ti l d d t
• Misleading results
• Advantage of potential reduced cost of trials
Validated Surrogate Validated Surrogate EEndpointsndpoints
• A validated surrogate endpoint: the surrogate must reliably predict the overall effect on the clinical outcomepredict the overall effect on the clinical outcome
Greatest potential for valid surrogate endpointGreatest potential for valid surrogate endpoint
Reliably predicts endpointDisease
Surrogate End Point True clinical End Point
• Surrogate‐endpoint validation requires a pooled or meta‐analysis of clinical trial dataanalysis of clinical trial data. – At least three trials with 500 patients each are preferable to one
– Reasonable heterogeneity in the trials is desirable, as it minimizes theReasonable heterogeneity in the trials is desirable, as it minimizes the impact of any one particular treatment or trial
Adapted from Fleming T.R. et al., Ann Intern 1996; 125(5):605
Reason for Failure of Surrogate Reason for Failure of Surrogate EEndpointsndpointsA Example CRC and CEA
? IMPACT on Clinically relevantClinically relevant
outcome Disease
B Example Intervention
False positiveDisease
B Example Intervention
False negative
C Example Cytostatic Agent Intervention
False negativeDisease
InterventionD Example RR &Toxicity OS
Miss effectDisease
InterventionD Example RR &Toxicity OS
Adapted from Fleming T.R. et al., Ann Intern 1996; 125(5):605
TimeTimeSurrogate End Point True clinical End Point
Validation Trial: Advanced Validation Trial: Advanced CColorectal olorectal CCancerancer
5 FU/LV
10 Historical Trials 3 Registration Trials
5‐FU/LVvs
5‐FU(1744 pts)
5‐FU/LV vs
Ralitrexed(1345 pts)
5‐FU/LVvs
FOLFOX or FOLFIRI
(1263 pts)(1263 pts)
FDA APPROVAL OF VALIDATED SURROGATE ENDPOINT PFS IN ADVANCED COLORECTAL TRIALS
Buyse M. et al.,J Clin Oncol 2007;25:5218
Other ExamplesOther Examplespp
TRIALS IN TUMOUR TYPE ENDPOINT VALIDATED VALIDATION FDA APPROVAL
Advanced colorectal cancer1 PFS for OS ✔ ✔
Adjuvant colorectal cancer2 3 year DFS for 5yr OS ✔ ✔
Advanced head and neck cancer3
PFS for OS ✔ ✗
Adjuvant lung cancer DFS for OS ✔ ✗
Metastatic breast cancer PFS for OS ✗ ✗
1.Buyse M. et al.,J Clin Oncol 2007;25:52182. Sargent D et al., J Clin Oncol 2005; 23:8664
3. Michiels S et al., Lancet Oncology 2009; 10:3414. Michiels S et al., ASCO 2011
Summary: Consideration when Designing Summary: Consideration when Designing Clinical Clinical TTrialsrials
• Properties of new therapy to be investigated
• Role of a biomarkers prior to designing the phase III
trial (this can be difficult)
• Novel trial design to test a hypothesis• Novel trial design to test a hypothesis
• Endpoint choice and use of surrogate endpoints –p g p
the need for standardisation!
AcknowledgementsAcknowledgementsgg
Lina Pugliano, MDResearch FellowResearch Fellow
Back upBack up
Clinical Trial Design for Biomarkers: Clinical Trial Design for Biomarkers: Randomise AllRandomise AllRandomise AllRandomise All
Randomise ALLRandomise ALL Targeted DesignTargeted Design Strategy DesignStrategy Design
POPULATION
g gg g
POPULATION
gy ggy g
POPULATIONPOPULATION
Assessment Marker (M)Assessment Marker (M)
Eligible patients consent Eligible patients consent
Assessment Marker (M)Assessment Marker (M)
M‐M‐ M+M+
Assessment Marker (M)Assessment Marker (M)
Eligible patients consent Eligible patients consent
Allocation/Randomisation
Allocation/RandomisationAllocation/Randomisation
TREATMENT A TREATMENT B
EligibleEligible
Allocation/RandomisationAllocation/Randomisation
MM
TREATMENT ATREATMENT A
MM Allocation/Randomisation
Tx NOT based on M
Tx based on M
AssessmentAssessment
M+M+ M‐M‐ M+M+M‐M‐
M+M+ M+M+TREATMENT BTREATMENT BTREATMENT ATREATMENT A
M+M+ M‐M‐ M+M+M‐M‐TxATxATx ATx A Tx BTx B
Data AnalysisData Analysis AssessmentAssessment
Data AnalysisData Analysis
AssessmentAssessment
Data AnalysisData Analysis
Adapted from Hoering A. et al Clin Can Res 2008; 14:4358
Clinical Trial Design for BiomarkersClinical Trial Design for BiomarkersTable 1: Randomised Phase III trials for Targeted Agents
Randomise All Targeted Design Strategy DesignRandomise All Targeted Design Strategy Design
Best Used • If treatment works for M-
as well as M+ patients
• When the treatment helps
only M+ patients
• Generally inferior to
other designs
• Large marker prevalence
• Cut points of M+ and M-
• Small marker prevalence
• Cut point of M+ and M+ is
not well established well established
Cautions • If the treatment is harmful
for M patients
• Small benefit of treatment
in M
• Does not compare
efficacy of 2 treatmentsfor M- patients in M-
• Cut point of M+ and M- not
well established
efficacy of 2 treatments
Disadvantages • Generally requires higher
sample size
• Requires highest number
of patients for screening
• Generally requires
higher sample size
Adapted from Hoering A. et al Clin Can Res 2008; 14:4358
Consensus Agreement: Adjuvant Consensus Agreement: Adjuvant CColorectal Trialsolorectal Trials
Definition DFS DFS RFS TTR TTF CSS OSDefinition DFS DFS RFS TTR TTF CSS OS
Locoregional recurrence E E E E I I
Distant Metastases E E E E I I
Second primary, same cancer E I I E I I
Second primary other cancer E I I E I ISecond primary, other cancer E I I E I I
Death from same cancer E E E E E E
Death from other cancer E E C E C E
Non‐cancer related death E E C C C E
Treatment related death E E C E C E
Loss to follow up C C C C C CDFS=Disease free survival; RFS:= Relapse free survival; TTR=Time to recurrence; TTF=Time to treatment failure; CSS=Cancer specific survival; OS=Overall Survival;E=Event; I=Ignored; C=CensoredE=Event; I=Ignored; C=Censored
Cornelis J.A. et al., JNCI 2007; 99(13):998
Proposed Standard Proposed Standard DDefinitions: Breast efinitions: Breast CCancerancer
Proposed Standard Definitions1 OSDFS‐DCIS
IDFS DDFS DRFS RFS RFI BCFI DRFI
Invasive ipsilateral breast tumour recurrence ✔ ✔ ✔ ✔ ✔✔ ✔ ✔ ✔ ✔
Local/Regional recurrence ✔ ✔ ✔ ✔ ✔
Distant recurrence ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔
Death from breast cancer ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔
Death from non breast cancer cause ✔ ✔ ✔ ✔ ✔ ✔
Death from unknown cause ✔ ✔ ✔ ✔ ✔ ✔Death from unknown cause ✔ ✔ ✔ ✔ ✔ ✔
Invasive contralateral breast cancer ✔ ✔ ✔
Ipsilateral DCIS ✔ ✔✔ ✔
Contralateral DCIS ✔ ✔
Second primary invasive cancer (non breast) ✔ ✔ ✔Notes: Lobular carcinoma in situ not included as an event in these definitions as it is not generally considered to be a direct precursor of breast cancer‘Contralateral invasive breast cancer’ is preferred to ‘second primary breast cancer’ as it is less ambiguousIpsilateral invasive breast cancer are presumed to be a recurrenceSecond non‐breast primary cancers should NOT include squamous or basal skin cancers or new in situ carcinomas of any site
OS O ll i l DFS DCIS Di f i l d t l i i it IDFS I i di f i l i i DDFS Di t t di fOS=Overall survival; DFS‐DCIS=Disease free survival –ductal carcinoma in situ; IDFS=Invasive disease free survival‐invasive; DDFS=Distant disease freesurvival; DRFS=Distant relapse free survival; RFS=Recurrence free survival; RFI=recurrence free interval; BCFI=Breast cancer free interval; DRFI=Distant recurrence free interval.
1.Hudis CA et al, J Clin Oncol, 2007; 25:2127-2132