SEARCH

SEARCH BY CITATION

Keywords:

  • basal breast cancer;
  • clinical trials;
  • DNA microarrays;
  • triple-negative

Abstract

  1. Top of page
  2. Abstract
  3. Gene expression profiling shows TN BCs have different subtypes
  4. Gene expression profiling shows basal BCs are homogeneous
  5. Multivariate analysis of basal and TN status
  6. Acknowledgements
  7. References

The basal molecular subtype of breast cancer (BC) is defined by the mRNA expression pattern of an intrinsic ∼500-gene set. It is the most homogeneous subtype in transcriptional terms, and one of the most aggressive in prognostic terms. Clinical trials testing new systemic therapeutic strategies have been launched in basal BCs. Although no proof of evidence has yet been reported, basal tumors are currently assimilated to and selected as triple-negative (TN) BCs in these trials because of their frequent immunohistochemical (IHC) negativity for hormone and ERBB2 receptors. Here, we have assessed the degrees of correlation and of homogeneity of the TN phenotype (IHC-based definition) and the basal subtype (gene expression-based definition). We analyzed 172 TN BCs defined by gene expression profile as basal (123 cases) and nonbasal (49 cases). Conversely, 160 tumors were defined as basal by their gene expression profile and included 123 TN and 37 non-TN samples. Uni- and multivariate analyses revealed that TN BCs represent a more heterogeneous group than basal BCs, including basal and nonbasal tumors very different both at the histoclinical and molecular level, notably for mRNA expression of molecules targeted by specific therapies under evaluation in clinical trials. These results call for caution in the interpretation of ongoing trials and selection of patients in future trials. They also warrant the identification of molecular markers for basal BCs more clinically applicable than gene expression profiles. © 2008 Wiley-Liss, Inc.

The treatment of breast cancer (BC) is currently far from being optimal. It relies on a series of histoclinical prognostic and predictive features, all insufficient to capture the heterogeneity of the disease. However, the selection of patients for inclusion in therapeutic trials remains largely based on these imperfect histoclinical features, leading to accrual of heterogeneous series of patients. Molecular classification aims at resolving this heterogeneity and identifying homogeneous entities that are due to specific molecular alterations and sensitive to specific treatments. Patients within the same class would probably have a more uniform clinical outcome, optimizing the design and interpretation of clinical trials, notably for targeted therapies. The importance of patients' selection for an optimized design of the trial or assessment of the drug is well documented for ERBB2 and EGFR antagonists. Trastuzumab is efficient in about 15% of heavily pretreated metastatic BC patients. This moderate response rate would have been missed in an unselected population. Clinical development of EGFR tyrosine kinase inhibitors (gefitinib and erlotinib) was initially carried out in an unselected population of lung cancer patients. Despite an authentic anti-tumor effect (response rate of 10–20%) of these inhibitors in heavily pretreated patients, evaluation in combination with standard chemotherapy in randomized studies in advanced nonsmall cell lung cancers patients did not identify any benefit over chemotherapy alone.1 It was rapidly suspected that the absence of biomarkers able to identify EGFR-dependent tumors impaired the success of this therapy. This hypothesis was confirmed by the identification of activating EGFR mutations that sensitize tumors to treatment in gefitinib-responding patients.2, 3

Genome-wide approaches seem powerful to build classifiers. A pioneering study using DNA microarrays showed that BCs are heterogeneous and contain at least 5 biologically and clinically relevant molecular subtypes, luminal A and B, ERBB2-overexpressing, basal and normal-like, defined by the expression patterns of an intrinsic ∼500-gene set.4, 5 This taxonomy and its histoclinical correlations have been confirmed in large series by the same group,6–9 and by others in various BC forms,10–14 suggesting it is robust and universal.15 Basal subtype represents around 15% of invasive ductal BCs. It is the most homogeneous subtype in transcriptional terms. Basal BCs display specific epidemiological, phenotypic and molecular features including high grade, proliferation, areas of necrosis and distinct genetic alterations.16, 17 They are usually negative for hormone receptors and ERBB2, show high expression of P53, EGFR and KIT, and share many features with BRCA1-related tumors. In contrast to luminal A and ERBB2-overexpressing subtypes—successfully targeted by hormone therapy and trastuzumab respectively—no targeted therapy is available for basal tumors. However, although they respond relatively well to chemotherapy, basal BCs are associated with a poor prognosis. Genome-wide analysis of the basal subtype has led to the identification of potential molecular targets such as CRYAB, EGFR, KIT or RAS,16 and several clinical trials (http://clinicaltrials.gov/) testing new systemic therapeutic strategies in basal BCs have been launched.

Triple-negative (TN) BCs do not express ER, PR and ERBB2 proteins, as detected by immunohistochemistry (IHC). Because most basal BCs do not express these receptors, TN BCs tend to be referred in current clinical practice as basal tumors, although no formal proof has been reported yet that TN BCs are actual tumors of basal subtype. Ongoing therapeutic trials for basal BCs select patients on the basis of IHC triple-negativity. However, gene expression profiles, which are based on many variables, are probably more appropriate to define this subtype than 3 IHC markers if the objective is to identify a molecularly distinct and homogeneous subgroup of tumors that are likely to respond similarly to treatment. TN is the definition currently used in research. TN approximation of basal phenotype is due to the fact that the gold-standard DNA microarray technology is not readily available in daily practice, in contrast to IHC, which is a simpler and more accessible technique. The risk is that TN BCs included in clinical trials are not exactly identical to basal BCs and constitute a less homogeneous group, leading to a falsely negative conclusion of inactivity of the drug in basal BCs.

We have investigated here to what extent the TN phenotype as defined by using IHC correlates with the basal subtype as defined by using gene expression profiles, and how much the TN phenotype and the basal subtype are homogeneous. We identified in our database and in public databases 172 TN BCs (defined as IHC staining less than 10% for hormone receptors and equal to 0–1+ for ERBB2 or 2+ with negative FISH result) with available DNA microarray-based gene expression data. We profiled 145 tumors using Ipsogen Discovery cDNA microarrays (∼9.000 cDNAs; 50 samples) and Affymetrix U133 Plus 2.0 human oligonucleotide microarrays (47.000 transcripts and variants; 95 samples). We additionally included in our analysis data collected from the web (http://bioinformatics.mdanderson.org/pubdata.html) for 27 TN samples profiled by Hess et al. using similar Affymetrix microarrays.18 All samples were extracted from a series of 964 BCs (831 from our own database and 133 from the study of Hess et al.).

Gene expression profiling shows TN BCs have different subtypes

  1. Top of page
  2. Abstract
  3. Gene expression profiling shows TN BCs have different subtypes
  4. Gene expression profiling shows basal BCs are homogeneous
  5. Multivariate analysis of basal and TN status
  6. Acknowledgements
  7. References

The molecular subtype of tumors was defined using the genes common to each respective data set and the Norway/Stanford intrinsic ∼500-gene set as previously described.19 By measuring the correlation of each of the 172 TN tumors with each centroid, 123, 8, 4, 12 and 15 samples were assigned to the basal, luminal A, luminal B, ERBB2-overexpressing and normal-like subtype, respectively. Ten samples could not be attributed any subtype. Thus, a total of 123 samples (71%) were basal and 49 (29%) were nonbasal. Using univariate analyses, we compared the histoclinical features of these 2 groups (Table I). Several differences were identified. Median age of patients with basal tumors (50 years) was inferior to that of patients with other tumors (57 years). Basal tumors exhibited higher pathological grade: 88% were grade III versus only 51% of nonbasal tumors. The pathological tumor size (pT) was higher in basal tumors with 60% being pT2-pT3 versus 46% in the other group. No correlation was found with pathological axillary lymph node status (pN). There were more medullary BCs within the basal tumors, and more lobular BCs within the nonbasal tumors. Differences in clinical outcome were also evidenced, although not significant probably because of the small population size. Five-year metastasis-free survival was 65% in the basal group and 75% in the other group and 5-year overall survival was, respectively, 68 and 78%. Pathological complete response (PCR) after primary chemotherapy was available in 27 Hess' samples: 59% of basal tumors experienced PCR versus 27% in the other group.

Table I. Histoclinical Characteristics of Breast Cancers in the TN Group and in The Basal Group
 Triple negativeBasal
NNon-basal (n = 49)Basal (n = 123)p1Odds ratio (IC 95 %)NNon-triple negative (n = 37)Triple negative (n = 123)p1Odds ratio (IC 95 %)
  • IDC, invasive ductal cancer; ILC, invasive lobular cancer; MBC, medullary breast cancer; MIX, invasive ductal and lobular cancer; pN, pathological axillary lymph node status; pT, pathological tumor size; pCR, pathological complete response.

  • 1

    To assess differences in histoclinical features between groups of patients, Fisher's Exact test was used for qualitative variables with discrete categories, the Wilcoxon test was used for continuous variables, and the log-rank test was used to compare Kaplan-Meier survivals.–

  • 2

    Affymetrix probeset ID: 205225_at for ESR1, 208305_at for PGR, 216836_s_at for ERBB2, 212022_s_at for MKI67, 201596_x_at for KRT18, 201820_at for KRT5, 213680_at for KRT6B, 232541_at for EGFR, 205051_s_at for KIT, 209283_at for CRYAB, 202647_s_at for NRAS.–

  • 3

    Image cDNA clone ID: 725321 for ESR1, 68688 for PGR, 756253 for ERBB2, 428545 for MKI67, 151663 for KRT18, 251200 for KIT, 153805 for CRYAB, 1941427 for NRAS.

Median age17257.5 (31–85)50 (28–75)1.3E–0216050 (33–77)50 (28–75)0.85
Pathological type166  1.3E–04158  0.24
 IDC 37 (82%)101 (83%)   33 (89%)101 (83%)  
 ILC 5 (11%)1 (1%)   0 (0%)1 (1%)  
 MBC 1 (2%)19 (16%)   3 (8%)19 (16%)  
 MIX 2 (4%)0 (0%)   1 (3%)0 (0%)  
pN140  11.03 (0.46–2.31)130  10.92 (0.37–2.27)
 neg 20 (51%)51 (50%)   14 (48%)51 (50%)  
 pos 19 (49%)50 (50%)   15 (52%)50 (50%)  
pT132  0.0792.17 (0.86–5.53)123  0.231.73 (0.62–4.65)
 pT1 15 (41%)23 (24%)   10 (36%)23 (24%)  
 pT2–3 17 (46%)57 (60%)   18 (64%)72 (76%)  
SBR Grade171  1.1E–066.75 (2.93–16.08)159  1.0E–023.39 (1.28–8.91)
 I + II 24 (49%)15 (12%)   12 (32%)15 (12%)  
 III 25 (51%)107 (88%)   25 (68%)107 (88%)  
pCR38  0.150.27 (0.04–1.44)35  11.14 (0.18–8.9)
 yes 3 (27%)16 (59%)   5 (62%)16 (59%)  
 no 8 (73%)11 (41%)   3 (38%)11 (41%)  
5-year MFS14475%65%0.42413266%65%0.885
5-year OS14478%68%0.36813269%68%0.928
ESR123172  2.5E–070.02 (0–0.16)160  2.6E–030.05 (0–0.5)
 poor 36 (73%)122 (99%)   32 (86%)122 (99%)  
 rich 13 (27%)1 (1%)   5 (14%)1 (1%)  
PGR23172  2.0E–030.33 (0.15–0.69)160  0.400.7 (0.3–1.73)
 poor 24 (49%)92 (75%)   25 (68%)92 (75%)  
 rich 25 (51%)31 (25%)   12 (32%)31 (25%)  
ERBB223172  1.7E–060.16 (0.07–0.36)160  2.4E–020.36 (0.14–0.96)
 poor 25 (51%)107 (87%)   26 (70%)107 (87%)  
 rich 24 (49%)16 (13%)   11 (30%)16 (13%)  
MKI6723172  1.5E–054.65 (2.19–10.12)160  0.291.55 (0.63–3.69)
 poor 29 (59%)29 (24%)   12 (32%)29 (24%)  
 rich 20 (41%)94 (76%)   25 (68%)94 (76%)  
KRT1823172  2.7E–100.07 (0.02–0.18)160  0.090.36 (0.1–1.37)
 poor 24 (49%)115 (93%)   31 (84%)115 (93%)  
 rich 25 (51%)8 (7%)   6 (16%)8 (7%)  
KRT523122  2.8E–079.24 (3.57–25.43)99  1.3E–024.51 (1.19–16.87)
 poor 24 (62%)12 (14%)   7 (44%)12 (14%)  
 rich 15 (38%)71 (86%)   9 (56%)71 (86%)  
KRT6B23122  2.5E–056.17 (2.43–16.34)99  1.7E–024.11 (1.1–15.16)
 poor 21 (54%)13 (16%)   7 (44%)13 (16%)  
 rich 18 (46%)70 (84%)   9 (56%)70 (84%)  
EGFR23122  0.191.8 (0.7–4.54)99  10.83 (0.14–3.53)
 poor 13 (33%)18 (22%)   3 (19%)18 (22%)  
 rich 26 (67%)65 (78%)   13 (81%)65 (78%)  
KIT23172  1.1E–033.1 (1.49–6.57)160  0.241.64 (0.71–3.71)
 poor 29 (59%)39 (32%)   16 (43%)39 (32%)  
 rich 20 (41%)84 (68%)   21 (57%)84 (68%)  
CRYAB23172  7.3E–065.53 (2.47–12.7)160  0.790.71 (0.16–2.37)
 poor 24 (49%)18 (15%)   4 (11%)18 (15%)  
 rich 25 (51%)105 (85%)   33 (89%)105 (85%)  
NRAS23172  4.2E–054.27 (2.02–9.22)160  0.151.85 (0.79–4.26)
 poor 30 (61%)33 (27%)   15 (41%)33 (27%)  
 rich 19 (39%)90 (73%)   22 (59%)90 (73%)  

We also included in the analysis several qualitative variables based on the mRNA expression level of hormone receptors (ESR1, PGR), ERBB2, markers of proliferation (MKI67), of luminal epithelial lineage (KRT18), of basal epithelial lineage (KRT5 and KRT6), and potential therapeutic targets of basal BCs: KIT, CRYAB, NRAS and EGFR. The probe sets were chosen by using the same criteria as Reyal et al.20 They are listed in Table I. To compare in a qualitative manner the expression levels between the basal and nonbasal groups of TN tumors, we used the median expression level of these genes (across the whole series to which the tumors belonged) as a cut-off to define relatively rich and relatively poor tumors. For most markers (Table I), basal and nonbasal groups were different with respect to mRNA expression. A higher proportion of basal tumors overexpressed MKI67 (76 vs. 41%) and KRT5/6 (86 vs. 38% for KRT5 and 84 vs. 46% for KRT6B). Conversely, a high proportion of nonbasal tumors expressed hormone receptors (27 vs. 1% for ESR1 and 51 vs. 25% for PGR) and KRT18 (51 vs. 7%) and ERBB2 (49 vs. 13%), in agreement with their luminal A/B and ERBB2-overexpressing subtype, respectively. Interestingly, a high proportion of basal tumors overexpressed CRYAB (85 vs. 51%) and NRAS (73 vs. 39%). Overexpression of CRYAB small heat shock protein constitutively activates the MAPK kinase/ERK (MEK/ERK) pathway, and MEK inhibitors suppress the phenotype induced by CRYAB.21 Basal tumors also exhibited a high expression of KIT (68 vs. 41%), which encodes a tyrosine kinase receptor whose targeting by imatinib is currently being tested in TN BCs. No difference was observed between basal (78%) and nonbasal (67%) groups for EGFR expression.

Gene expression profiling shows basal BCs are homogeneous

  1. Top of page
  2. Abstract
  3. Gene expression profiling shows TN BCs have different subtypes
  4. Gene expression profiling shows basal BCs are homogeneous
  5. Multivariate analysis of basal and TN status
  6. Acknowledgements
  7. References

To reinforce the idea that the TN phenotype is less performing than gene expression profiles to identify a homogeneous group of basal BCs, we next applied the converse approach. From the same series of 964 tumors with available molecular subtype or ER, PR and ERBB2 IHC status, 160 were defined as basal by gene expression profiling. They included 123 (77%) TN and 37 (23%) non-TN samples. As shown in Table I, univariate comparisons showed much less histoclinical and molecular differences between these 2 basal subgroups than there were between basal and nonbasal TN BCs. When significant, the odds ratios were higher when comparing basal and nonbasal TN samples than when comparing TN and non-TN basal samples. This result suggested that basal BCs represent a more homogeneous group than TN BCs. This was confirmed when we compared across all 964 tumors the basal versus nonbasal groups, and the TN versus the non-TN groups. Here again, differences (as assessed by significant odds ratios) were more frequent and/or stronger in the first comparison than in the second one, including differences in expression of potential therapeutic molecular targets (data not shown).

Multivariate analysis of basal and TN status

  1. Top of page
  2. Abstract
  3. Gene expression profiling shows TN BCs have different subtypes
  4. Gene expression profiling shows basal BCs are homogeneous
  5. Multivariate analysis of basal and TN status
  6. Acknowledgements
  7. References

As further indirect confirmation of greater heterogeneity of the TN BC group as compared with the basal BC group, we applied another approach based on multivariate logistic regression analysis and receiver operating characteristic (ROC) curves. A multivariate model was developed in each group to identify a histoclinical predictor of basal versus nonbasal status in the TN group, and TN versus non-TN status in the basal group. The analysis included 7 continuous variables: age, pT, pN, pathological grade, and mRNA expression level of ESR1, PGR and ERBB2. Using a backward stepwise selection procedure to minimize the Akaïke Information Criterion, 4 variables (pN, grade, ESR1 and ERBB2 mRNA) were retained to predict the basal status of TN BCs (n = 120), whereas 5 variables (same plus age) were retained for predicting the TN status of basal BCs (n = 128). Figure 1 shows ROC curves for each predictor in the TN group (solid curve) and in the basal group (dashed curve). Area under curve is greater for the TN group than for the basal group, suggesting that it is “easier” to identify a performing histoclinical predictor that discriminates basal and nonbasal TN BCs than to identify a predictor that discriminates TN and non-TN basal BCs. Indirectly, this result confirms the greater heterogeneity of the TN group as compared with the basal group.

thumbnail image

Figure 1. Receiver operating characteristic (ROC) curves for histoclinical predictor. Solid curve: prediction of basal versus nonbasal status in the TN breast cancer group; dashed curve: prediction of TN vs. non-TN status in the basal subtype group. AUC means area under curve.

Download figure to PowerPoint

In conclusion, using uni- and multivariate analyses, we have shown that TN BCs, when defined by IHC staining less than 10% for hormone receptors and equal to 0–1+ for ERBB2 or 2+ with negative FISH result, are not synonymous with basal BCs. TN BCs represent a heterogeneous group at both the molecular and histoclinical level. They include basal (∼70% of samples) and nonbasal (∼30% of samples) tumors very different at the histoclinical level. Conversely, we have shown that basal BCs represent a more homogeneous group than TN BCs. Currently, efforts are ongoing to identify a simple IHC profile that would reliably identify basal BCs.22–24 This would have probably to involve more markers, including differentiation markers such as cytokeratins, as well as improvement of antibodies and detection. The incomplete concordance between the basal phenotype and the triple negative status has been reported by using various IHC definitions,25–29 including the ER−, ERBB2−, EGFR+ and/or CK5/6+ IHC profile, which is currently considered as the most reliable definition.23, 24 In all studies, the basal status conferred to TN BCs a poor clinical outcome when compared to nonbasal TN tumors25–28 as suggested by our survival results. To our knowledge, our study (172 TN samples) and the very recent study of Kreike et al.30 (97 TN samples) are the first to address the issue of TN versus basal overlap by using the genomic definition of basal phenotype. Although the study of Kreike concluded that TN tumors are synonymous with basal-like tumors, in fact the overlap was not complete, and a majority of, but not all TN tumors had a basal-like phenotype, and conversely, a majority of, but not all basal-like tumors were TN. As discussed by others,31 these results are concordant with ours and confirm the incomplete overlap between TN and basal BCs. These data suggest that within the TN group, the basal subgroup should deserve recognition as a separate entity, and that basal and nonbasal TN BCs likely need different systemic treatments. We have shown that nonbasal tumors, as compared with basal tumors, display relatively lower mRNA expression levels of specific molecular targets, and consequently a low probability of response to specific therapies under evaluation. Altogether, these observations call for caution in the interpretation of ongoing trials for TN patients and warrant improvement in the design of future trials with respect to patients' selection. They also warrant the identification of molecular markers for basal BCs more clinically applicable than gene expression profile.

References

  1. Top of page
  2. Abstract
  3. Gene expression profiling shows TN BCs have different subtypes
  4. Gene expression profiling shows basal BCs are homogeneous
  5. Multivariate analysis of basal and TN status
  6. Acknowledgements
  7. References
  • 1
    Herbst RS,Giaccone G,Schiller JH,Natale RB,Miller V,Manegold C,Scagliotti G,Rosell R,Oliff I,Reeves JA,Wolf MK,Krebs AD, et al. Gefitinib in combination with paclitaxel and carboplatin in advanced non-small-cell lung cancer: a phase III trial–INTACT 2. J Clin Oncol 2004; 22: 78594.
  • 2
    Lynch TJ,Bell DW,Sordella R,Gurubhagavatula S,Okimoto RA,Brannigan BW,Harris PL,Haserlat SM,Supko JG,Haluska FG,Louis DN,Christiani DC, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. N Engl J Med 2004; 350: 212939.
  • 3
    Paez JG,Janne PA,Lee JC,Tracy S,Greulich H,Gabriel S,Herman P,Kaye FJ,Lindeman N,Boggon TJ,Naoki K,Sasaki H, et al. EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science 2004; 304: 1497500.
  • 4
    Perou CM,Sorlie T,Eisen MB,van de Rijn M,Jeffrey SS,Rees CA,Pollack JR,Ross DT,Johnsen H,Akslen LA,Fluge O,Pergamenschikov A, et al. Molecular portraits of human breast tumours. Nature 2000; 406: 74752.
  • 5
    Sorlie T,Perou CM,Tibshirani R,Aas T,Geisler S,Johnsen H,Hastie T,Eisen MB,van de Rijn M,Jeffrey SS,Thorsen T,Quist H, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci USA 2001; 98: 1086974.
  • 6
    Sorlie T,Tibshirani R,Parker J,Hastie T,Marron JS,Nobel A,Deng S,Johnsen H,Pesich R,Geisler S,Demeter J,Perou CM, et al. Repeated observation of breast tumor subtypes in independent gene expression data sets. Proc Natl Acad Sci USA 2003; 100: 841823.
  • 7
    Carey LA,Perou CM,Livasy CA,Dressler LG,Cowan D,Conway K,Karaca G,Troester MA,Tse CK,Edmiston S,Deming SL,Geradts J, et al. Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study. Jama 2006; 295: 2492502.
  • 8
    Hu Z,Fan C,Oh DS,Marron JS,He X,Qaqish BF,Livasy C,Carey LA,Reynolds E,Dressler L,Nobel A,Parker J, et al. The molecular portraits of breast tumors are conserved across microarray platforms. BMC Genomics 2006; 7: 96.
  • 9
    Sorlie T,Wang Y,Xiao C,Johnsen H,Naume B,Samaha RR,Borresen-Dale AL. Distinct molecular mechanisms underlying clinically relevant subtypes of breast cancer: gene expression analyses across three different platforms. BMC Genomics 2006; 7: 127.
  • 10
    Sotiriou C,Neo SY,McShane LM,Korn EL,Long PM,Jazaeri A,Martiat P,Fox SB,Harris AL,Liu ET. Breast cancer classification and prognosis based on gene expression profiles from a population-based study. Proc Natl Acad Sci USA 2003; 100: 103938.
  • 11
    Kapp AV,Jeffrey SS,Langerod A,Borresen-Dale AL,Han W,Noh DY,Bukholm IR,Nicolau M,Brown PO,Tibshirani R. Discovery and validation of breast cancer subtypes. BMC Genomics 2006; 7: 231.
  • 12
    Bertucci F,Finetti P,Rougemont J,Charafe-Jauffret E,Cervera N,Tarpin C,Nguyen C,Xerri L,Houlgatte R,Jacquemier J,Viens P,Birnbaum D. Gene expression profiling identifies molecular subtypes of inflammatory breast cancer. Cancer Res 2005; 65: 21708.
  • 13
    Van Laere SJ,Van den Eynden GG,Van der Auwera I,Vandenberghe M,van Dam P,Van Marck EA,van Golen KL,Vermeulen PB,Dirix LY. Identification of cell-of-origin breast tumor subtypes in inflammatory breast cancer by gene expression profiling. Breast Cancer Res Treat 2006; 95: 24355.
  • 14
    Yu K,Lee CH,Tan PH,Tan P. Conservation of breast cancer molecular subtypes and transcriptional patterns of tumor progression across distinct ethnic populations. Clin Cancer Res 2004; 10: 550817.
  • 15
    Birnbaum D,Bertucci F,Ginestier C,Tagett R,Jacquemier J,Charafe-Jauffret E. Basal and luminal breast cancers: basic or luminous ?. Int J Oncol 2004; 25: 24958.
  • 16
    Cleator S,Heller W,Coombes RC. Triple-negative breast cancer: therapeutic options. Lancet Oncol 2007; 8: 23544.
  • 17
    Adelaide J,Finetti P,Bekhouche I,Repellini L,Geneix J,Sircoulomb F,Charafe-Jauffret E,Desplans J,Parzy D,Schoenmakers E,Viens P,Jacquemier J, et al. Integrated profiling of basal and luminal A breast cancers. Cancer Res, in press.
  • 18
    Hess KR,Anderson K,Symmans WF,Valero V,Ibrahim N,Mejia JA,Booser D,Theriault RL,Buzdar AU,Dempsey PJ,Rouzier R,Sneige N, et al. Pharmacogenomic predictor of sensitivity to preoperative chemotherapy with paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide in breast cancer. J Clin Oncol 2006; 24: 423644.
  • 19
    Bertucci F,Finetti P,Cervera N,Charafe-Jauffret E,Mamessier E,Adelaide J,Debono S,Houvenaeghel G,Maraninchi D,Viens P,Charpin C,Jacquemier J, et al. Gene expression profiling shows medullary breast cancer is a subgroup of basal breast cancers. Cancer Res 2006; 66: 463644.
  • 20
    Reyal F,Stransky N,Bernard-Pierrot I,Vincent-Salomon A,de Rycke Y,Elvin P,Cassidy A,Graham A,Spraggon C,Desille Y,Fourquet A,Nos C, et al. Visualizing chromosomes as transcriptome correlation maps: evidence of chromosomal domains containing co-expressed genes–a study of 130 invasive ductal breast carcinomas. Cancer Res 2005; 65: 137683.
  • 21
    Moyano JV,Evans JR,Chen F,Lu M,Werner ME,Yehiely F,Diaz LK,Turbin D,Karaca G,Wiley E,Nielsen TO,Perou CM, et al. AlphaB-crystallin is a novel oncoprotein that predicts poor clinical outcome in breast cancer. J Clin Invest 2006; 116: 26170.
  • 22
    Livasy CA,Karaca G,Nanda R,Tretiakova MS,Olopade OI,Moore DT,Perou CM. Phenotypic evaluation of the basal-like subtype of invasive breast carcinoma. Mod Pathol 2006; 19: 26471.
  • 23
    Nielsen TO,Hsu FD,Jensen K,Cheang M,Karaca G,Hu Z,Hernandez-Boussard T,Livasy C,Cowan D,Dressler L,Akslen LA,Ragaz J, et al. Immunohistochemical and clinical characterization of the basal-like subtype of invasive breast carcinoma. Clin Cancer Res 2004; 10: 536774.
  • 24
    Yehiely F,Moyano JV,Evans JR,Nielsen TO,Cryns VL. Deconstructing the molecular portrait of basal-like breast cancer. Trends Mol Med 2006; 12: 53744.
  • 25
    Rakha EA,El-Sayed ME,Green AR,Lee AH,Robertson JF,Ellis IO. Prognostic markers in triple-negative breast cancer. Cancer 2007; 109: 2532.
  • 26
    Sasa M,Bando Y,Takahashi M,Hirose T,Nagao T. Screening for basal marker expression is necessary for decision of therapeutic strategy for triple-negative breast cancer. J Surg Oncol 2008; 97: 304.
  • 27
    Tan DS,Marchio C,Jones RL,Savage K,Smith IE,Dowsett M,Reis-Filho JS. Triple negative breast cancer: molecular profiling and prognostic impact in adjuvant anthracycline-treated patients. Breast Cancer Res Treat 2007.
  • 28
    Tischkowitz M,Brunet JS,Begin LR,Huntsman DG,Cheang MC,Akslen LA,Nielsen TO,Foulkes WD. Use of immunohistochemical markers can refine prognosis in triple negative breast cancer. BMC Cancer 2007; 7: 134.
  • 29
    Bidard FC,Conforti R,Boulet T,Michiels S,Delaloge S,Andre F. Does triple-negative phenotype accurately identify basal-like tumour? An immunohistochemical analysis based on 143 ‘triple-negative’ breast cancers. Ann Oncol 2007; 18: 12856.
  • 30
    Kreike B,van Kouwenhove M,Horlings H,Weigelt B,Peterse H,Bartelink H,van de Vijver MJ. Gene expression profiling and histopathological characterization of triple-negative/basal-like breast carcinomas. Breast Cancer Res 2007; 9: R65
  • 31
    Rakha EA,Tan DS,Foulkes WD,Ellis IO,Tutt A,Nielsen TO,Reis-Filho JS Are triple negative tumours and basal-like breast cancer synonymous?. Breast Cancer Res 2007; 9: R80.