Académique Documents
Professionnel Documents
Culture Documents
27
NUMBER
17
JUNE
10
2009
O R I G I N A L
R E P O R T
Purpose
The indications of adjuvant chemotherapy for patients with estrogen receptor (ER) positive breast
cancer are controversial. We analyzed the predictive value of Ki67, HER2, and progesterone
receptor (PR) expression for the efficacy of docetaxel in patients with ER-positive, node-positive
breast cancer.
Patients and Methods
Expression of Ki67, HER2, and PR was measured by immunohistochemistry in tumor samples from
798 patients with ER-positive breast cancer who participated in PACS01, a randomized trial that
evaluated the efficacy of docetaxel. Risk reduction was evaluated using a Cox model adjusted for age,
tumor size, nodal involvement, treatment arm, and biomarkers. The predictive value of biomarkers was
assessed by an interaction test. Disease-free survival (DFS) was the primary end point.
Results
Ki67, HER2, and PR were expressed in 21%, 9%, and 62% of samples, respectively. Hazard ratios
for relapse associated with docetaxel were 0.51 (95% CI, 0.26 to 1.01) in ER-positive/Ki67-positive
tumors and 1.03 (95% CI, 0.69 to 1.55) in ER-positive/Ki67-negative tumors (ratio for interaction:
0.53; 95% CI, 0.24 to 1.16; P .11). Five-year DFS rates were 81% (95% CI, 76% to 86%) and
84% (95% CI, 75% to 93%) in patients with ER-positive/Ki67-negative and ER-positive/Ki67positive tumors treated with docetaxel and 81% (95% CI, 76% to 86%) and 62% (95% CI, 52%
to 72%) in patients with ER-positive/Ki67-negative and ER-positive/Ki67-positive tumors treated
with fluorouracil, epirubicin, and cisplatin. No trend for interaction was observed between
docetaxel and HER2 (ratio for interaction: 0.83; 95% CI, 0.35 to 1.94; P .66), nor between
docetaxel and PR (ratio for interaction: 0.89; 95% CI, 0.47 to 1.66; P .71).
Conclusion
Ki67 expression identifies a subset of patients with ER-positive breast cancer who could be
sensitive to docetaxel treatment in the adjuvant setting.
J Clin Oncol 27:2809-2815. 2009 by American Society of Clinical Oncology
INTRODUCTION
Downloaded from jco.ascopubs.org on January 26, 2015. For personal use only. No other uses without permission.
Copyright 2009 American Society of Clinical Oncology. All rights reserved.
2809
Penault-Llorca et al
PACS 01 trial
PACS 01
biomarker study
Tumor blocks
collected for
1,190 patients
ER immunostaining
population for
present analysis
Fig 1. Study flowchart shows the process for tumor blocks and patient
selection. F, fluorouracil; e, epirubicin; C, cyclophosphamide; D, docetaxel; ER,
estrogen receptor.
Downloaded from jco.ascopubs.org on January 26, 2015. For personal use only. No other uses without permission.
Copyright 2009 American Society of Clinical Oncology. All rights reserved.
study. Events were defined as locoregional relapse, metastatic relapse, contralateral breast cancer, and death, whatever the cause. HRs for relapse were
assessed using a Cox model that included age, tumor stage (T), and nodal stage
(N) as categoric covariates, together with each biomarker. Age was considered
as a binary variable with a cutoff defined at 50 years (stratification variable).
Cutoffs for T and N were, respectively, more than 2 and more than 3. Predictive value of each biomarker was evaluated using a test for interaction between
treatment effect and biomarker expression. To evaluate treatment by biomarker interaction, a Cox model was conducted that included age, T, and N together
with Ki67 treatment interaction, PR treatment interaction, and HER2
treatment interaction. Subpopulation treatment effect pattern plot (STEPP)
analyses were performed to assess docetaxel effect according to Ki67 expression. The STEPP analysis is intended to evaluate the magnitude of the treatment effect as a function of the values of the continuous covariates.18 It was
carried out using the sliding window algorithm described by Bonetti et al,18
which allowed overlapping subpopulations of patients to be defined, ensuring
that each subpopulation contains a large enough number of observations and
reasonably spans the range of the covariate. In the present analysis, five subpopulations were defined (A, B, C, D, E) as listed in Appendix Table A1 (online
only). Absolute effect of treatment of DFS was assessed by Kaplan-Meier
curves and log-rank tests. This article was written in accordance with Reporting Recommendations for Tumor Marker Prognostic Studies guidelines.19
RESULTS
Patient Characteristics
Seven hundred ninety-eight patients were included in the present
analysis. Table 1 reports the patient characteristics for the whole pop-
No.
Ki67-Negative
(n 549)
%
No.
50.0
25.0-66.0
Ki67-Positive
(n 150)
%
No.
49.0
25.0-64.0
%
50.0
26.0-66.0
305
493
38
62
211
338
38
62
62
88
41
59
359
310
38
51
44
5
256
201
31
52
42
6
65
70
6
46
50
4
644
143
82
18
499
50
91
9
71
78
48
52
549
150
79
21
NA
705
73
91
9
507
38
93
7
121
28
77
23
279
454
38
62
176
323
35
65
52
79
40
60
398
400
175
50
50
21
266
283
105
48
52
19
86
64
46
57
43
31
NA
Abbreviations: NA, not applicable; PR, progesterone receptor; F, fluorouracil; E, epirubicin; C, cyclophosphamide; DFS, disease-free survival.
P .01.
HER2 overexpression was defined as immunohistochemistry score 3 or HER2 amplification detected by fluorescent in situ hybridization analysis.
www.jco.org
Downloaded from jco.ascopubs.org on January 26, 2015. For personal use only. No other uses without permission.
Copyright 2009 American Society of Clinical Oncology. All rights reserved.
2811
Penault-Llorca et al
Table 2. Cox Univariate and Multivariate Regression Analysis of the Risk of Relapse According to Ki67, HER2, and PR Expression Together With Age,
Node Status, T Stage, and Treatment Covariates in Patients With ER-Positive Tumors
Treatment Arm, Biomarker, and
Clinical Characteristics
Age, years
50
50
No. of axillary nodes involved
3 nodes
3 nodes
T stage
T1-T2
T3-T4
Treatment with docetaxel
No
Yes
Ki67 expression
20%
20%
HER2 overexpression
Normal
Overexpressed
PR expression
10%
10%
Multivariate Analysis
Univariate Analysis
HR for Relapse
95% CI
HR for Relapse
1
0.7
0.6 to 1.0
.05
1
2.3
1.7 to 3.0
1
2.0
Multivariate Analysis
95% CI
HR for Relapse
95% CI
NA
1
0.7
0.5 to 1.0
.07
.0001
NA
1
2.2
1.5 to 3.1
.0001
1.2 to 3.4
.01
NA
1
1.6
1.2 to 2.4
.006
1
0.8
0.6 to 1.1
.1
NA
1
0.9
0.6 to 1.3
.6
1
1.7
1.2 to 2.4
.002
1
1.6
1.2 to 2.3
.01
1
1.5
1.0 to 2.2
.046
1
1.9
1.3 to 2.9
.002
1
1.6
1.0 to 2.5
.05
1
1.5
0.9 to 2.4
.12
1
1.7
1.2 to 2.3
.001
1
1.6
1.2 to 2.2
.005
1
1.8
1.3 to 2.5
.07
Abbreviations: PR, progesterone receptor; ER, estrogen receptor; HR, hazard ratio; NA, not applicable.
PR-negative tumors, respectively. We then included the three biomarkers in a Cox model together with age, number of axillary node
involved, T stage, and treatment. Ki67 (HR 1.47; 95% CI, 1.01 to
2.16; P .046) and PR expressions (HR 1.79; 95% CI, 1.26 to 2.54;
P .07) were associated with a significantly higher risk of relapse.
Among the clinicopathologic characteristics, a high number of axillary
nodes involved ( three) and a large tumor size (T stage 2) were
Biomarker
Positive (n = 150)
Negative (n = 549)
.11
Overexpressed (n = 73)
Normal (n = 705)
.66
Positive (n = 454)
Negative (n = 279)
.71
Ki67
HER-2
PR
0.5
1.0
1.5
2.0
2.5
3.0
Downloaded from jco.ascopubs.org on January 26, 2015. For personal use only. No other uses without permission.
Copyright 2009 American Society of Clinical Oncology. All rights reserved.
DISCUSSION
The precise role and indications of adjuvant chemotherapy for ERpositive breast cancer are matters of controversy. The present data give
rise to the hypothesis that Ki67 determination could be allowed to
identify a subset of patients with ER-positive breast cancer who would
potentially derive a high benefit from adjuvant docetaxel. Ki67 is a
protein expressed during cell cycle that is correlated with short-term
prognosis20 and has been associated with a higher sensitivity to preoperative chemotherapy.21 Several studies have evaluated the predictive
value of Ki67 for efficacy of first-generation adjuvant chemotherapy in
Disease-Free Survival
(probability)
1.0
0.8
0.6
0.4
0.2
Time (years)
Fig 3. Disease-free survival (DFS) according to treatment and Ki67 expression in
patients with estrogen receptor (ER) positive tumors. Arm A, fluorouracil,
epirubicin, and cyclophosphamide (FEC) arm; arm B, sequential FEC followed by
docetaxel arm.
www.jco.org
100
n = 331
n = 150
n = 244
5-Year Disease-Free
Survival (%)
n = 288
80
n = 242
Arm B
60
Arm A
40
20
10
15
20
25
30
35
40
45
Downloaded from jco.ascopubs.org on January 26, 2015. For personal use only. No other uses without permission.
Copyright 2009 American Society of Clinical Oncology. All rights reserved.
2813
Penault-Llorca et al
the patients included in the trial were analyzed in this biomarker study.
Although the clinical characteristics did not differ with the original
cohort, we cannot exclude the possibility that the analyzed population
could be biased as compared with the original one. Second, although
performed on one of the largest trials that addressed taxane efficacy
(PASC01), the study was not powered enough to successfully address
interaction between biomarkers and treatment efficacy in ER-positive
breast cancer. This limitation is shared with other unplanned analyses
from randomized trials that compared two different chemotherapy
regimens. This is related to the small relative effect of the investigational drug (HR 0.82 in the present study) and to the low number of
events observed in early breast cancer patients (n 175 of 798 in the
present study). The probability of detecting significant interaction
between treatment and biomarker in an unplanned analysis of a single
randomized trial is therefore very low in ER-positive disease. This
consideration highlights the need for pooled analyses to address the
question of predictive biomarkers for taxanes efficacy within ERpositive cancers. Finally, it is important to acknowledge the fragility of
subset analyses performed in unplanned analyses, including the
present one. These analyses, especially those reporting treatment effects in subsets, are indeed associated with a high rate of false positivity.23 This concern highlights the need for validation sets to confirm
that Ki67 could be a robust biomarker candidate to predict docetaxel
efficacy in ER-positive breast cancer. These validation studies are
ongoing using tumor samples from the PACS04 trial, a randomized
trial that addresses docetaxel efficacy in patients with node-positive
breast cancer.
Overall, the present study suggests that Ki67 is a biomarker
candidate for predicting docetaxel efficacy in ER-positive breast
cancer, to be validated in further larger studies. These data confirm
the heterogeneity of ER-positive tumors and make the argument to
perform further trials in well molecularly defined groups to reduce
the inconsistencies among trials.24 Finally, these data, together
with other retrospective studies,3-7,10 could provide a rationale for
performing a prospective trial that would tailor the indications and
choice of adjuvant chemotherapy according to ER, HER2, and
Ki67 immunostaining.
REFERENCES
1. Andre F, Pusztai L: Molecular classification of
breast cancer: Implications for selection of adjuvant
chemotherapy. Nat Clin Pract Oncol 3:621-632,
2006
2. Goldhirsch A, Wood WC, Gelber RD, et al:
10th St. Gallen conference: Progress and promise
Highlights of the international expert consensus on
the primary therapy of early breast cancer 2007. Ann
Oncol 18:1133-1144, 2007
3. Conforti R, Boulet T, Tomasic G, et al: Breast
cancer molecular subclassification and estrogen receptor expression to predict efficacy of adjuvant
anthracyclines-based chemotherapy: A biomarker
study from two randomized trials. Ann Oncol 18:
1477-1483, 2007
4. Berry DA, Cirrincione C, Henderson IC, et al:
Estrogen-receptor status and outcomes of modern
chemotherapy for patients with node-positive breast
cancer. JAMA 295:1658-1667, 2006
5. Regan MM, Viale G, Mastropasqua MG, et al:
Re-evaluating adjuvant breast cancer trials: Assessing hormone receptor status by immunohistochem2814
AUTHOR CONTRIBUTIONS
Conception and design: Frederique Penault-Llorca, Anne-Laure Martin,
Henri Roche
Financial support: Frederique Penault-Llorca
Administrative support: Anne-Laure Martin, Henri Roche
Provision of study materials or patients: Frederique Penault-Llorca,
Christine Sagan, Magali Lacroix-Triki, Yves Denoux, Veronique Verriele,
Jocelyne Jacquemier, Marie Christine Baranzelli, Frederic Bibeau,
Martine Antoine, Nicole Lagarde
Collection and assembly of data: Frederique Penault-Llorca, Christine
Sagan, Magali Lacroix-Triki, Yves Denoux, Veronique Verriele, Jocelyne
Jacquemier, Marie Christine Baranzelli, Frederic Bibeau, Martine
Antoine, Nicole Lagarde, Anne-Laure Martin, Henri Roche
Data analysis and interpretation: Frederique Penault-Llorca, Fabrice
Andre, Christine Sagan, Magali Lacroix-Triki, Yves Denoux, Veronique
Verriele, Jocelyne Jacquemier, Marie Christine Baranzelli, Frederic
Bibeau, Martine Antoine, Nicole Lagarde, Bernard Asselain, Henri Roche
Manuscript writing: Frederique Penault-Llorca, Fabrice Andre,
Henri Roche
Final approval of manuscript: Frederique Penault-Llorca, Fabrice
Andre, Anne-Laure Martin, Henri Roche
Downloaded from jco.ascopubs.org on January 26, 2015. For personal use only. No other uses without permission.
Copyright 2009 American Society of Clinical Oncology. All rights reserved.
Acknowledgment
We thank Vale`re Lounnas for his analyses, Anne Cayre, Christiane Jullien, and Matthieu Roche for technical support, and the French and
Belgium pathologists involved in PACS01 study for providing kindly the tumor blocks.
www.jco.org
Downloaded from jco.ascopubs.org on January 26, 2015. For personal use only. No other uses without permission.
Copyright 2009 American Society of Clinical Oncology. All rights reserved.
2815