RAGMA - GEICAM Madrid, Spain Clinical Research: A critical ...Problems with BOLERO-2 Trial •...
Transcript of RAGMA - GEICAM Madrid, Spain Clinical Research: A critical ...Problems with BOLERO-2 Trial •...
RAGMA - GEICAM Madrid, Spain
Clinical Research: A critical appraisal
Eitan Amir MD PhD Princess Margaret Cancer Centre
20 June 2015
Overview
• Early breast cancer:
– Neoadjuvant endpoint selection – the utility of pCR
– Meta-analyses to identify efficacy and toxicity
• Metastatic breast cancer:
– Optimal endpoint selection
– Toxicity of new cancer drugs
– Statistical methods and bias –informative censoring
• Efficacy-effectiveness gap
Early Breast Cancer
0
20%
40%
60%
80%
100%
2 4 6 8 Year
P<0.001 cPR
cNR
pCR
2 4 6 8
cPR
cNR
pCR P<0.001
Wolmark N: CDC, 2000
Pathologic response predicts overall survival
pCR and prognosis
B-18 DFS by response B-18 OS by response
Figure 6 Trial-level correlation between treatment effect on pathological complete response and event-free survival or overall
survival Each circle corresponds to one randomised comparison and the size of the circle represents the sample size.
A=GeparQuat...
Patricia Cortazar , Lijun Zhang , Michael Untch , Keyur Mehta , Joseph P Costantino , Norman Wolmark , Hervé Bonn...
Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis
The Lancet, Volume 384, Issue 9938, 2014, 164 - 172
http://dx.doi.org/10.1016/S0140-6736(13)62422-8
Treatment effect on pathologic complete response (pCR) versus (A) disease-
free survival (DFS) and (B) overall survival (OS).
Berruti A et al. JCO 2014;32:3883-3891
©2014 by American Society of Clinical Oncology
Winer EP, SABCS 2010
When would differences in pCR lead to improvement in long-term outcome?
Likelihood of distant relapse in patients with residual cancer burden (RCB) -0
(pathologic complete response), RCB-I, RCB-II, or RCB-III
Symmans W F et al. JCO 2007;25:4414-4422
Summary
• pCR should be used for hypothesis generation, not hypothesis testing.
• Should very seldom be the basis for drug registration except in exceptional situations:
– Large magnitude benefit in metastatic trials
– Require confirmation in adequately powered adjuvant trials with definitive endpoints.
Meta-analysis
• Method for pooling “similar” data from different sources.
• Primary advantage is to provide increased power and regression of results to the mean. – Valuable in the assessment of rare events (e.g. certain
toxicities)
– Can allow for more formal assessment of effect in subgroups which are underpowered in individual trials
• Main negative consequence is an increase in the alpha error (higher risk of false positive results).
Adjuvant Endocrine Therapy for Breast Cancer
Cardiovascular Events
ITA 1.22 [0.59-2.54]
ABCSG8/ARNO 1.51 [0.25-9.02] IES 1.15 [0.92-1.43]
N-SAS BC03 0.67 [0.11-4.03]
ATAC 1.24 [0.95-1.61]
Subtotal 1.15 [0.93-1.41]
BIG 1-98 1.43 [1.01-2.03]
Upfront AI versus tamoxifen
Subtotal 1.30 [1.06-1.61]
Subtotal 1.37 [1.05-1.79]
Switch versus tamoxifen
0.5 0.7 1 1.5 2
Study or Group OR [95% CI]
P for overall effect = 0.01
P for overall effect = 0.20
TEAM 1.37 [1.05-1.79]
P for overall effect = 0.02
Switch versus AI
TOTAL 1.26 [1.10-1.43] P for overall effect < 0.001
Increased with tamoxifen Increased with AI
Amir E et al. J Natl Cancer Instit 2011
0 5 10
Yrs
Breast Cancer Recurrence: All Women
Bone Recurrence
888 events
Nonbone Recurrence
18% Relative Reduction <1% Relative Reduction
Coleman R, et al. SABCS 2013. Abstract S4-07.
1947 events
50
40
30
20
10
0 0 5 10
Yrs
Bo
ne
Re
cu
rre
nc
e, %
17709 women
5.2%
4.0%
8.4%
10-yr gain 1.5% (SE: 0.6)
Log-rank 2P = .0009
17709 women
50
40
30
20
10
0
10.2%
10.0%
15.1%
10-yr gain 0.1% (SE: 0.8)
Log-rank 2P = .71
Dis
tan
t R
ec
urr
en
ce
Ou
tsid
e B
on
e, %
Bisphosphonates No Bisphosphonates
6.9%
15.0%
0 5 10
Yrs
Breast Cancer Recurrence:
Postmenopausal Women
508 events 1056 events
Coleman R, et al. SABCS 2013. Abstract S4-07.
11036 women 11036 women
34% Relative Reduction
10-yr gain 2.9% (SE: 0.8)
Log-rank 2P < .00001
8.8%
5.9% 5.1%
3.2%
8% Relative Reduction
9.2%
8.4%
14.3%
13.3%
10-yr gain 0.9% (SE: 1.0)
Log-rank 2P = .24
Bone Recurrence Nonbone Recurrence
Bisphosphonates No Bisphosphonates
50
40
30
20
10
0
50
40
30
20
10
0
Bo
ne R
ecu
rren
ce, %
Dis
tan
t R
ecu
rren
ce
Ou
tsid
e B
on
e, %
0 5 10
Yrs
General Assumptions in Clinical Trials
• Hypotheses tested usually address an overall or ‘average’ treatment effect in the study population.
• No assumption of homogeneity of effect across (unstratified) subgroups.
• Direction, not magnitude, of the treatment effect is expected be the same in subgroups.
Summary
• Meta-analysis can be an effective tool to determine effects among trials where there is insufficient power among individual studies.
• Results based on full datasets are more robust than pooling of data from subgroups.
– Results of pooled subgroup analyses have more validity if similar outcomes observed in other settings (metastatic vs. early-stage) and when a robust biological rationale is present.
Metastatic Breast Cancer
Endpoint Selection
• It is expensive and time consuming to conduct trials with overall survival as the primary endpoint
• Use of surrogate endpoints is reasonable if “surrogacy” has been established
• For surrogacy to be established: – There should be strong and consistent correlation
between the surrogate and definitive endpoints.
– A surrogate endpoint should also predict the net effect of treatment on the clinical outcome
Ann Intern Med. 1996;125(7):605-613. doi:10.7326/0003-4819-125-7-199610010-00011
Intra-tumoral heterogeneity
Treatment-related toxicity
Safety and Tolerability of New Cancer Drugs
• New anti-cancer drugs are associated with an increase in all safety and tolerability endpoints:
– Death (OR = 1.40; P < 0.001)
– Discontinuation (OR = 1.33; P < 0.001)
– Grade 3 & 4 AEs (OR = 1.52; P < 0.001)
Niraula S et al. J Clin Oncol 2012
Problems with BOLERO-2 Trial
• Everolimus is toxic:
– SAEs due to treatment: 11% vs. 1%
– Early discontinuation due to SAE or withdrawn consent: 24% vs. 6%
– Both worse than for many palliative chemotherapy regimens
• Analysis was biased by informative censoring leading to over-estimation of benefit of everolimus.
Arnoud J. Templeton , Olga Ace , Eitan Amir , Francisco Vera-Badillo , Alberto Ocana , Gregory R. Pond , Ian F. T...
Influence of censoring on conclusions of trials for women with metastatic breast cancer
European Journal of Cancer, 2015
http://dx.doi.org/10.1016/j.ejca.2014.12.016
Endpoints
• Overall survival need to remain a prominent endpoint in trials of metastatic breast cancer.
• Improvements in progression-free survival without clear improvement in overall survival especially if there is increased toxicity likely do not reflect benefit to patients.
• Care needs to be applied to the interpretation of trials analysed with informative censoring.
Efficacy-effectiveness gap
Definition
• Differences in outcomes between patients treated in ideal settings (i.e. RCTs) and those treated in the real world.
• Explained by multiple factors
– Patient characteristics likely explain some of the effect.
Treweek S et al. Eur J Cancer. 2015; 51(8): 907-14.
Solution?
• Validation of RCTs using well-designed observational studies is desirable to ensure patient safety and improved outcomes.
• Large scale population-based initiatives such as the American Society of Clinical Oncology CancerLinQ aims to provide a platform for real-world quality and comparative effectiveness analyses.