1 Hôpitaux Universitaires de Genève, Geneva, Switzerland; 2 EORTC Data Center, Brussels; 3 Institut J. Bordet, Brussels, Belgium; 4 Netherlands Cancer Institute, Amsterdam, The Netherlands; 5 Institut Bergonié, Bordeaux, France; 6 NCIC Clinical Trials Group, Kingston, Canada; 7 Hospitais da Universidade, Coimbra, Portugal; 8 Centre Henri Becquerel, Rouen, France; 9 University Hospital, Leuven, Belgium
Received 9 September 2002; revised 15 October 2002; accepted 19 November 2002
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Locally advanced and/or inflammatory breast cancer (LABC) is a heterogeneous disease. Molecular markers may help to understand this heterogeneity. This paper reports the results of a study assessing the potential prognostic or predictive value of HER-2, p53, cyclinD1, MIB1, ER and PgR expression by immunohistochemistry from patients included in an EORTCNCICSAKK trial.
Patients and methods:
A total of 448 patients with a cytological or histological diagnosis of LABC were randomised into a trial comparing two anthracycline-based neoadjuvant regimens. Chemotherapy was followed by standard locoregional therapy. Survival was comparable in both arms. We collected and analysed centrally paraffin-embedded tumour specimens from 187 (72.5%) of 258 patients that had a histological diagnosis.
Results:
Of the patients included in this molecular marker study 114 relapsed and 91 died. In the multivariate analysis p53 positivity was associated with a shorter progression-free survival [hazard ratio (HR) = 1.96; 95% CI 1.332.91; P = 0.0008) and a shorter overall survival (HR = 1.98; 95% CI 1.283.06; P = 0.002). PgR positivity predicted for a longer overall survival (HR = 0.54; 95% CI 0.350.83; P = 0.0045).
Conclusions:
p53 was an independent factor predicting for survival. In order to clarify whether p53 is a pure prognostic and/or a predictive factor, a phase III trial is being conducted (EORTC 10994/BIG 00-01 study) using functional assay in yeast from frozen tumour samples.
Key words: breast cancer, locally advanced/inflammatory breast cancer, neoadjuvant chemotherapy, p53
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The results of a large intergroup (EORTCNCICSAKK) phase III trial conducted on 448 patients presenting with locally advanced and/or inflammatory breast cancers were recently reported [7]. Patients were randomised between two different neoadjuvant anthracycline chemotherapy regimens. After six cycles of chemotherapy, locoregional treatment was planned, followed by tamoxifen for 5 years. After a median follow-up of 5.5 years, no significant difference in terms of progression-free survival (PFS) or overall survival (OS) was found between the two chemotherapy arms. Patients with inflammatory breast cancer had a shorter median PFS than patients with locally advanced breast cancer (24 months and 44 months, respectively). However, both subgroups were heterogeneous, some patients having progressed very early and others remaining without progression several years after diagnosis. Beyond the clinical presentation (locally advanced versus inflammatory breast cancer) we attempted to understand this heterogeneity with the help of molecular markers. Therefore, we collected tumour specimens from patients included in the EORTCNCICSAKK study and performed a central immunohistochemical analysis of the following markers: estrogen receptor (ER) and progesterone receptor (PgR), MIB-1, HER-2, p53 and cyclin D1. This paper reports the results of a retrospective assessment of the potential prognostic or predictive value of these markers in relation to clinical response, PFS and OS. By definition "prognostic factors are associated with either the metastatic and/or growth rate potential of the primary tumour and predictive factors are associated with relative sensitivity and/or resistance to specific therapeutic agents" [8]. Some factors can be both prognostic and predictive. We will elaborate on this in the discussion.
![]() |
Patients and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Central pathology review
Formalin-fixed or Bouin Hollande-fixed, paraffin-embedded tumour samples from the primary tumour were obtained prior to the initiation of neoadjuvant chemotherapy from participating EORTC institutions. Tumours were provided as unstained 5-µm tissue sections or as tumour blocks, in which case the slides were prepared centrally. The tumours were graded on haematoxylin and eosin (H&E) slides according to the Elston modified scoring scheme of Bloom and Richardson [9] by a reference pathologist (S.D.B.) blinded to the clinical outcome of the patient.
Immunohistochemical analysis
All immunohistochemical analyses were performed in a single reference laboratory, using the unstained 5-µm tissue sections. The general protocol for immunohistochemical staining was performed according to a standardised method previously described by Clahsen et al. [10]. This standardised method has been used in several retrospective studies on tissue sections from patients treated within EORTC trials.
Except for HER-2, an antigen retrieval procedure was applied: tissue sections were pre-incubated in 10 mM sodium citrate solution and incubated three times for 5 min in a 600 W microwave oven. Slides were then pre-incubated for 15 min in phosphate-buffered saline (PBS) with 5% bovine serum albumin (BSA) and incubated for 1 h at room temperature with the following primary antibodies: mouse monoclonal HER-2 Abs CB11 (Biogenex, undiluted), mouse monoclonal p53 Abs DO-7 (Dako, 1/100 dilution), mouse monoclonal Cyclin D1 Abs DCS-6 (Novocastra, 1/40 dilution), mouse monoclonal estrogen receptors Abs 1D5 (Dako, 1/100 dilution), mouse monoclonal progesterone receptors Abs PGR-1A6 (Biogenex, 1/20 dilution), mouse monoclonal Ki-67 antigen Abs MIB-1 (Immunotech, 1/20 dilution). Sections were washed with PBS and reacted with streptavidinbiotin peroxidase reagents (Dako) and diaminobenzidine chromogen. Sections were finally counterstained with haematoxylin (5 min).
In all series and for each antibody, positive controls (known cases of breast carcinoma) and negative controls (primary antibody omitted) were included.
All slides were scored independently by two investigators (S.D.B. and H.B.) without patient information. The method of counting and scoring the immunostaining and the definition of a cut-off value were established for each antibody according to previous studies before starting the analyses; all details are described below. All discrepancies in scoring between the two investigators were resolved by consensus.
p53 overexpression. Scoring was done using a semiquantitative system according to a method previously described [10]. Mean nuclear staining intensity (MSI) was evaluated and could vary from 0 (none), 1 (weak), 2 (moderate) to 3 (strong). Percentage of positive tumour cell nuclei (PPN) was estimated and values were given: 0 (0%), 1 (125%), 2 (2650%), 3 (5175%) and 4 (>75%). The addition of these two values (MSI and PPN) results in the p53 staining score (range 27). Expression of p53 was considered as positive for scores 4 [10].
HER-2 overexpression. As previously described, a tumour was considered as positive if a clear circonferential membranous positivity was found in the tumour cells either focally or throughout the tumour, as defined in previous studies [11]. Tumour cells which showed a granular cytoplasmic staining were considered to have normal HER-2 expression. The proportion of HER-2 positive cells was determined semi-quantitatively as the percentage of invasive tumour cells ranging from 0% to 100% from the entire tissue section.
MIB-1 positivity. Only clear nuclear staining in tumour cells was considered positive. The percentage of positively stained cells was calculated from the number of brown nuclei seen in a sample of 200 tumour cells [12]. We selected the threshold of 20% (<20%, negative; 20%, positive), as used in a previous EORTC immunohistochemical study [10].
ER and PgR status. Only clear nuclear staining in tumour cells was considered positive. The proportion of ER and PgR-positive cells was determined as the percentage of invasive tumour cells ranging from 0% to 100% from the entire tissue section. The threshold of 10% positivity was chosen as the cut-off value, according to a current consensus recommendation [13]. The staining intensity was scored from 0 to 3, in a similar fashion as for p53 (see above).
Cyclin D1 expression. Only clear nuclear staining in tumour cells was considered positive. Rarely, a weak cytoplasmic staining was observed; this was considered to be non-specific. The percentage of positive cells and the staining intensity were determined as described above. We chose a cut-off value of 10% (<10%, negative tumours; 10%, positive tumours), as used in previous studies [14].
Statistical methods
Life tables were constructed to examine PFS and OS, using the KaplanMeier method [15] and were compared using the log-rank test [16]. A proportional hazards regression model [17] with stratification for treatment [18] was used in the univariate and multivariate analysis. A step-down (backward) variable selection procedure was used to fit the multivariate model [19]. A two-sided test was used at the 5% level of significance.
In the statistical analysis of prognostic value, we planned to analyse molecular marker expression by immunohistochemistry both as continuous and dichotomous variables. For each marker the definitions of positivity and negativity were selected before we performed the analysis based on previous studies, as described in the previous section. The correlation between the response status and the predictive value of each biological factor has been analysed in a univariate and multivariate logistic regression model [17, 19]. All analyses testing prognostic significance of the different factors identified have been performed with SAS software. The survival data and response status were provided directly from Software for the Management and Analysis of Randomised Trials (SMART). The correlation between the response status and the predictive value of each biological factor has been analysed in a univariate analysis.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
Univariate analysis for PFS and OS
A first univariate analysis was performed on the following parameters analysed as dichotomous variables: tumour grade (I/II versus III), ER status [percentage of cells stained (%CS) <10 versus 10], PgR status (%CS <10 versus
10), MIB-1 percentage (%CS <20 versus
20), HER-2 overexpression (%CS = 0 versus >0), p53 overexpression (score <4 versus
4) and cyclin D1 overexpression (% CS <10 versus
10). This analysis demonstrated that two factors were significantly associated with a shorter PFS and OS: ER negativity and p53 positivity (Table 3). PgR negativity was associated with a shorter PFS and OS; this difference was statistically significant for OS (P = 0.007), but not for PFS (P = 0.23). Progression-free survival and OS curves according to immunohistochemically defined p53 status are shown in Figures 1 and 2.
|
|
|
|
All the factors found to be significant at the 0.1 level in the univariate analysis were included in the multivariate Cox regression model for their relation to PFS and OS. Consequently, grade and HER-2 were not analysed in the first multivariate analysis, and HER-2 and cyclin D1 were not analysed in the second multivariate analysis.
In both analyses, p53 positivity (tumour with a score 4 or with a high percentage of positive cells) was found to be an independent prognostic factor associated with a statistically significant higher risk of progression and death (Tables 5 and 6).
|
|
Predictive factors of a clinical complete response to chemotherapy
A complete clinical response to neoadjuvant chemotherapy was observed in 48 of 179 patients (27%). In the univariate analysis, cyclin D1-negative tumours were associated with a lower rate of complete clinical response (P = 0.051) (Table 7). No statistically significant correlation between grade or immunohistochemical status of ER, PgR, MIB-1, HER-2 or p53 and a clinical complete response to chemotherapy was found (Table 7).
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
In early breast cancer ER and PgR status are weak prognostic factors [22] and well established predictive factors of response to tamoxifen and other hormonal manipulations [23]. In locally advanced and/or inflammatory breast cancers the prognostic or predictive value of hormone receptors remains controversial [21, 24]. In our study, PgR status was found to be an independent factor predicting for OS (Tables 5 and 6). In the multivariate analysis for OS, ER status lost its significance (Tables 5 and 6). This is most probably because of its close correlation with PgR (in the multivariate analysis for OS, ER was removed from the model in favour of PgR). The fact that PgR is an independent factor for OS, but not for PFS seems paradoxical. We hypothesised that we missed its prognostic information for PFS due to the relatively small number of patients analysed.
In our series HER-2 did not predict for either PFS or OS in the univariate analysis. The potential pure prognostic value of HER-2 remains controversial, as shown in a recent meta-analysis [25]. Regarding its predictive value, data in the literature suggest that HER-2 overexpression is associated with anthracycline sensitivity and that this sensitivity increases with dose escalation [2]. In our study patients received chemotherapy regimens with a high dose of anthracyclines in both treatment arms. Therefore, our study design is not optimal to evaluate the hypothesis of an association between HER-2 overexpression and anthracycline sensitivity. Similarly, there are data in the literature that suggest that HER-2-positive tumours may be less sensitive to tamoxifen than HER-2-negative tumours [2]. In our study all patients received tamoxifen and the results do not suggest an inverse association between HER-2 overexpression and tamoxifen sensitivity.
In this study p53 was a strong independent factor predicting for PFS and OS. We first performed a multivariate analysis with molecular markers as dichotomous variables using a predefined score detailed in the methods section. In this multivariate analysis p53 was the only factor associated with an increased risk of progression (HR = 1.96) and the strongest factor associated with an increased risk of death (HR = 1.98) (Table 5). We then performed a second multivariate analysis, the factors being analysed as continuous variables. Again p53 emerged as the only factor predicting for a shorter PFS and the strongest factor predicting for a shorter OS (Table 6).
The risk of false positive and false negative results is higher when p53 is assessed by the immunohistochemical method as compared with molecular biology methods (e.g. denaturating gradient gel electrophoresis, genomic p53 sequencing method) [6, 26]. With immunohistochemistry there is a risk of false negative cases related to the type of p53 mutation. With many p53 gene mutations, p53 protein half-life is increased and immunohistochemistry detects the protein in the nucleus, but when p53 gene mutations encode unstable proteins (nonsense mutations, splicing mutations) immunohistochemistry remains negative (null mutations). These mutations have been found to account for 2547% of all p53 mutations in series of patients with early breast cancers [27, 28]. These specific mutations were probably uncommon in our series and this may explain our observation: p53 by immunohistochemistry was a strong independent prognostic factor in this subset of high risk tumours. We could hypothesise that the use of a more sensitive method (e.g. denaturating gradient gel electrophoresis, genomic p53 sequencing method) could only have increased the independent prognostic value of p53 assessed by immunohistochemistry.
In our series we did not find a correlation between p53 status and clinical response to chemotherapy (Table 7). Amongst other possibilities two reasons may explain this observation. First the number of events (number of clinical complete responses) was small in our series and we may have missed a correlation by a lack of statistical power. Secondly, as discussed previously, immunohistochemistry is not the best method to assess p53. In five trials conducted in the neoadjuvant setting where p53 status was assessed by immunohistochemistry, no correlation was found between p53 status and clinical response to chemotherapy [6, 2932]. While in one clinical study [5], recently updated [6], where p53 was assessed by temporal temperature gradient gel electrophoresis (TTGE) and genomic sequencing, specific p53 mutations were associated with resistance to anthracyclines. Pathologically assessed complete response may be a better surrogate for chemotherapy efficacy than clinical response. Therefore biological markers predicting for a pathological complete response (pCR) would be of great value. However, in view of the small number of patients who achieved a pCR, we decided not to evaluate the possible correlation between p53 status (or other markers) and pathological response.
Our results do not allow us to conclude whether p53 is a pure prognostic, or a mixed prognostic and predictive factor. Preclinical data suggest that p53-mutated tumours are less sensitive to anthracyclines [3, 4], but remain sensitive to taxanes [3335]. Moreover, the correlation between p53 status and pCR, that we were unable to address, needs to be evaluated. In order to test this hypothesis a large intergroup phase III clinical trial is being conducted (EORTC, SAKK, Swedish and Angloceltic groups) under the auspices of the Breast International Group (EORTC 10994/BIG 00-01 study). Patients with large operable or locally advanced breast cancer are randomised to receive six cycles of neoadjuvant chemotherapy, either an anthracycline-based regimen or a docetaxel-based regimen, followed by locoregional treatment and antihormonal treatment when indicated. In this study, p53 status will be determined by using a functional assay in yeast that detects functionally important p53 mutations [36]. This test gives direct information regarding p53 function (detects biologically important mutations) and is theoretically more sensitive than sequencing because it is insensitive to contamination of samples with normal tissue. We will measure the correlation between p53 assessment by immunohistochemistry method and functional test as part of a side study. The functional test will be performed from frozen tumour samples taken by double trucut biopsy or by single incisional biopsy, as we have shown that the material from both biopsies gives identical results [37]. These frozen samples will also be assessed by cDNA microarray technology, with the hope of identifying a gene expression profile that predicts exquisite sensitivity to a taxane-based regimen.
![]() |
Acknowledgements |
---|
![]() |
Footnotes |
---|
Study participants are listed in the Acknowledgements.
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2. Yamauchi H, Stearns V, Hayes DF. When is a tumor marker ready for prime time? A case study of c-erbB-2 as a predictive factor in breast cancer. J Clin Oncol 2001; 19: 23342356.
3. Lowe SW, Ruley HE, Jacks T, Housman DE. p53-dependent apoptosis modulates the cytotoxicity of anticancer agents. Cell 1993; 74: 957967.[ISI][Medline]
4. Lowe SW, Bodis S, McClatchey A et al. p53 status and the efficacy of cancer therapy in vivo. Science 1994; 266: 807810.[ISI][Medline]
5. Aas T, Borresen AL, Geisler S et al. Specific p53 mutations are associated with de novo resistance to doxorubicin in breast cancer patients. Nature Med 1996; 2: 811814.[ISI][Medline]
6. Geisler S, Lonning PE, Aas T et al. Influence of TP53 gene alterations and c-erbB-2 expression on the response to treatment with doxorubicin in locally advanced breast cancer. Cancer Res 2001; 61: 25052512.
7. Therasse P, Mauriac L, Jassem J et al. Final results of a randomised phase III trial comparing CEF (cyclophosphamide, epirubicin, 5FU) with a dose intensified EC + G-CSF (epirubicin, cyclophosphamide + filgrastim) in locally advanced breast cancer. An EORTCNCICSAKK multicenter study. J Clin Oncol 2003; In press.
8. Hayes DF, Trock BJ, Harris AL. Assessing the clinical impact of prognostic factors: when is statistically significant clinically useful? Breast Cancer Res Treat 1998; 52: 305319.[CrossRef][ISI][Medline]
9. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology 1991; 19: 403410.[ISI][Medline]
10. Clahsen PC, van de Velde CJ, Duval C et al. p53 protein accumulation and response to adjuvant chemotherapy in premenopausal women with node-negative early breast cancer. J Clin Oncol 1998; 16: 470479.[Abstract]
11. Allred DC, Swanson PE. Testing for erbB-2 by immunohistochemistry in breast cancer. Am J Clin Pathol 2000; 113: 171175.[ISI][Medline]
12. Wintzer HO, Zipfel I, Schulte-Mönting J et al. Ki-67 immunostaining in human breast tumors and its relationship to prognosis. Cancer 1991; 67: 421428.[ISI][Medline]
13. Goldhirsch A, Glick JH, Gelber RD et al. Meeting highlights: International Consensus Panel on the Treatment of Primary Breast Cancer. Seventh International Conference on Adjuvant Therapy of Primary Breast Cancer. J Clin Oncol 2001; 19: 38173827.
14. McIntosh GG, Anderson JJ, Milton I et al. Determination of the prognostic value of cyclin D1 overexpression in breast cancer. Oncogene 1995; 11: 885891.[ISI][Medline]
15. Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457481.[ISI]
16. Mantel N. Evaluation of survival data and the two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966; 50: 163170.[Medline]
17. Cox DR. Regression models and life-tables. J R Stat Soc B 1972; 34: 187202.[ISI]
18. Breslow N. Comparison of survival curves. In Buyse ME, Staquet MJ, Sylvester RJ (eds): Cancer clinical trials methods and practice. Oxford, UK: Oxford Medical Publications 1988; 382406.
19. Simon R, Altman DG. Statistical aspects of prognostic factor studies in oncology. Br J Cancer 1994; 69: 979985.[ISI][Medline]
20. Faille A, Cremoux P, Extra JM et al. p53 mutations and overexpression in locally advanced breast cancers. Br J Cancer 1994; 69: 11451150.[ISI][Medline]
21. Honkoop AH, van Diest PJ, de Jong JS et al. Prognostic role of clinical, pathological and biological characteristics in patients with locally advanced breast cancer. Br J Cancer 1998; 77: 621626.[ISI][Medline]
22. Allred DC. Prognostic and predictive factors in breast cancer by immunohistochemical analysis. Mod Pathol 1998; 11: 155168.[ISI][Medline]
23. Early Breast Cancer Trialist Collaborative Group. Tamoxifen for early breast cancer: an overview of the randomised trials. Lancet 1998; 351: 14511467.[CrossRef][ISI][Medline]
24. Rubens RD, Armitage P, Winter PJ et al. Prognosis in inoperable stage III carcinoma of the breast. Eur J Cancer 1977; 13: 805811.[ISI][Medline]
25. Trock BJ, Yamauchi H, Brotzman M et al. C-erbB2 as a prognostic factor in breast cancer: a meta-analysis. Proc Am Soc Clin Oncol 2000; 19: 97a (Abstr).
26. Sjögren S, Inganäs S, Norberg T et al. The p53 gene in breast cancer: prognostic value of complementary DNA sequencing versus immunohistochemistry. J Natl Cancer Inst 1996; 88: 173182.
27. Berns EM, Foekens JA, Vossen R et al. Complete sequencing of TP53 predicts poor response to systemic therapy of advanced breast cancer. Cancer Res 2000; 60: 21552162.
28. Chappuis PO, Estreicher A, Dieterich B et al. Prognostic significance of p53 mutation in breast cancer: frequent detection of non-missense mutations by yeast functional assay. Int J Cancer 1999; 84: 587593.[CrossRef][ISI][Medline]
29. MacGrogan G, Mauriac L, Durand M et al. Primary chemotherapy in breast invasive carcinoma: predictive value of the immunohistochemical detection of hormonal receptors, p53, c-erbB-2, MiB1, pS2 and GST. Br J Cancer 1996; 74: 11201125.[ISI][Medline]
30. Makris A, Powles TJ, Dowsett M et al. Prediction of response to neoadjuvant chemoendocrine therapy in primary breast cancer. Clin Cancer Res 1997; 3: 593600.[Abstract]
31. Linn SC, Pinedo HM, van Ark-Otte J et al. Expression of drug resistance proteins in breast cancer in relation to chemotherapy. Int J Cancer 1997; 71: 787795.[CrossRef][ISI][Medline]
32. Rozan S, Vincent-Salomon A, Zafrani B et al. No significant predictive value of c-erbB-2 or p53 expression regarding sensitivity to primary chemotherapy or radiotherapy in breast cancer. Int J Cancer 1998; 79: 2733.[CrossRef][ISI][Medline]
33. Fan S, Cherney B, Reinhold W et al. Disruption of p53 function in immortalized human cells does not affect survival or apoptosis after taxol or vincristine treatment. Clin Cancer Res 1998; 4: 10471054.[Abstract]
34. Lanni JS, Lowe SW, Licitra EJ et al. p53-independent apoptosis induced by paclitaxel through an indirect mechanism. Proc Natl Acad Sci USA 1997; 94: 96799683.
35. Wahl AF, Donaldson KL, Fairchild C et al. Loss of normal p53 function confers sensitization to Taxol by increasing G2/M arrest and apoptosis. Nature Med 1996; 2: 7279.[ISI][Medline]
36. Flaman JM, Frebourg T, Moreau V et al. A simple p53 functional assay for screening cell lines, blood, and tumors. Proc Natl Acad Sci USA 1995; 92: 39633967.
37. Bonnefoi H, Ducraux A, Movarekhi S et al. p53 as a potential predictive factor of response to chemotherapy: feasibility of p53 assessment using a functional test in yeast from trucut biopsies in breast cancer patients. Br J Cancer 2002; 86: 750755.[CrossRef][ISI][Medline]