Ann Surg Oncol (2011) 18:2851–2857 DOI 10.1245/s10434-011-1665-8
ORIGINAL ARTICLE – BREAST ONCOLOGY
Impact of Breast Cancer Molecular Subtypes on Locoregional Recurrence in Patients Treated with Neoadjuvant Chemotherapy for Locally Advanced Breast Cancer Michael O. Meyers, MD1, Nancy Klauber-DeMore, MD1, David W. Ollila, MD1, Keith D. Amos, MD1, Dominic T. Moore, MPH3, Amy A. Drobish, MA3, Emily M. Burrows, MPH3, E. Claire Dees, MD2, and Lisa A. Carey, MD2 1
Division of Surgical Oncology, Department of Surgery, University of North Carolina School of Medicine at Chapel Hill, Chapel Hill, NC; 2Division of Medical Oncology, Department of Medicine, University of North Carolina School of Medicine at Chapel Hill, Chapel Hill, NC; 3The Lineberger Comprehensive Cancer Center, University of North Carolina School of Medicine at Chapel Hill, Chapel Hill, NC
ABSTRACT Background. Gene expression studies have identified distinct breast cancer subtypes, including luminal A, luminal B, Her2-enriched, and Basal-like, which differ in survival. The impact of subtypes on locoregional recurrence (LRR) after neoadjuvant chemotherapy for locally advanced breast cancer is unknown. Methods. A total of 149 patients with stage II and III breast cancer with known ER, PR, and HER2 who underwent neoadjuvant chemotherapy from 1991 to 2005 were analyzed. We used clinical assays to distinguish luminal A (ER or PR?/HER2-, n = 55), luminal B (ER or PR?/ HER2?, n = 25), HER2 (ER and PR-/HER2?, n = 20), and Basal-like (ER, PR, and HER2-, n = 49) subtypes. Covariates associated with LRR were evaluated by logistic regression and differences between subtypes tested using Wald v2. Results. Median follow-up was 55 months. Forty-nine (33%) patients had breast conservation (BCT) with radiation, 82 (55%) had a mastectomy with radiation, and 18 (12%) had a mastectomy alone. Eighty-eight (59%) were clinically node positive. A pathologic complete response was seen in 39 (26%) patients. LRR was identified in 11 (7%) patients: 2 after BCT (4%) and 9 after mastectomy (9%). LRR rates by subtype are as follows: luminal A 2 of
Ó Society of Surgical Oncology 2011 First Received: 26 April 2010; Published Online: 26 March 2011 M. O. Meyers, MD e-mail:
[email protected]
55 (4%), luminal B 1 of 25 (4%), Her2 1 of 20 (5%), and basal-like 7 of 49 (14%). Compared with all other subtypes, basal-like patients were more likely to have a LRR (7/49 (14%) vs. 4/100 (4%), p = 0.03). Conclusions. Molecular subtype predicts LRR with basallike patients more likely to develop LRR. These patients may be candidates for investigation with novel chemotherapy regimens and radiation sensitizing agents, which may offer improvement in local control.
Management of patients with locally advanced breast cancer has evolved significantly during the past 20 years. It is now accepted that women with locally advanced breast cancer often are managed with chemotherapy as the initial treatment as opposed to surgery. In women who have a significant clinical response to chemotherapy, it also is recognized that breast conservation (BCT) often can be achieved and that they are not obligated to mastectomy.1–3 We and others have demonstrated that this approach is both feasible and safe. It also appears that local recurrence rates of patients who achieve breast conservation are acceptable and comparable to those who undergo BCT who present with smaller tumors and are operated on as the initial form of therapy.1 Along with changes in the paradigms of clinical management of breast cancer is increasing recognition that this is a diverse disease that encompasses a number of distinct clinical, pathologic, and biologic variables. Combinations of these factors influence both the prognosis and management of patients diagnosed with breast cancer.4 More recently, DNA microarray analysis has been used to
2852
describe phenotypic differences in gene expression patterns and distinct molecular subtypes of breast cancer have been described.5,6 Several studies have shown a difference in both prognosis and response to chemotherapy amongst subtypes.7–10 These subtypes include two primary types of estrogen receptor (ER)-negative tumors: basal-like (human epidermal growth factor type 2 negative) and HER2-enriched, and at least two types of ER-positive tumors (luminal A and luminal B).8,9 In addition to this gene expression data derived from small sets of frozen tumor samples, immunohistochemical (IHC) surrogates to identify breast tumor subtypes using paraffin-embedded tumor have been validated.11,12 This has allowed evaluation of outcomes in larger sets of data using commonly acquired pathologic variables. Previous work has demonstrated that molecular subtypes of breast cancer predict not only survival and distant recurrence but locoregional recurrence as well.13–16 However, little is known about the impact in patients who undergo neoadjuvant chemotherapy for locally advanced breast cancer. In this paper, we sought to evaluate variables associated with locoregional recurrence in patients treated with neoadjuvant chemotherapy in whom IHC data were available to determine molecular subtypes.
METHODS Definition of Breast Cancer Molecular Subtypes As previously noted, the original descriptions of molecular subtypes in breast cancer were based on gene expression profiles using DNA microarray analysis. However, IHC surrogates have been described that are validated against the original gene expression profiles. This has been done by comparing DNA microarray results to IHC for ER, HER2, HER1, and cytokeratin 5/6 in a single dataset of breast cancers.11 This allowed identification of combinations of IHC markers that closely matched the gene expression patterns. These IHC surrogates were validated in a large dataset, allowing expansion of analysis of molecular subtypes to paraffin-embedded tissue. The IHC subtypes described were luminal (ER?/HER2-), HER2? and basal (ER-/HER2-/cytokeratin? and/or HER1?). In this study, we used a modified definition that creates four subtypes by taking into account progesterone receptor status and separating HER2-positive tumors into two groups.12 The four subtypes are: luminal A (ER and/or PR?/HER-), luminal B (ER and/or PR?/HER2?), basal (ER-/PR-/HER2-) and HER2-enriched (HER2?/ER-/ PR-). Whereas additional markers can enrich for basallike subtype, the triple negative proxy is accurate approximately 75% of the time.17
M. O. Meyers et al.
Patient Population This is a single-institution retrospective review of an institutional review board-approved prospective breast cancer database at the University of North Carolina at Chapel Hill. All patients who underwent neoadjuvant chemotherapy for clinical stage II and III breast cancer from 1991 to 2005 are included. Patients were categorized according to the sixth edition of the AJCC staging manual. Locally advanced tumors were defined as clinically node positive, regardless of T size and those that were at least T2 in whom breast conservation was not considered to be a primary option. This database is updated every 6 months and includes clinical, radiographic, and pathologic data as well as follow-up information. The neoadjuvant regimens received by the patients in this database are reflective of the clinical trials and standard approaches of the institution during that time period, resulting in some heterogeneity of the regimens. However, the majority of patients received anthracycline and taxane-based regimens. For this study, patients who received neoadjuvant chemotherapy and had IHC data available for ER, PR, and HER2 are included. Patients presenting with stage IV disease or with inflammatory breast cancer (T4D) were excluded. Postoperative radiotherapy was utilized at the discretion of the treating physicians. A locoregional recurrence was defined as follows: (1) any chest wall recurrence in those who underwent mastectomy; (2) any in-breast recurrence in those achieving breast conservation; and (3) any regional lymph node basin (axillary, supraclavicular, or infraclavicular) recurrence. A recurrence in more than one of these locations simultaneously was counted as a single recurrence for the purposes of data collection and reporting. Statistical Analysis Fisher’s exact test was used to test for general association within contingency table. For continuous covariates, the Wilcoxon rank-sum test (using van der Waerden or normal scores) was used for two-group comparisons. The Kruskal-Wallis one-way analysis of variance test (also using normal scores) was used to evaluate differences in responses across several categories. Genetic subtype was evaluated with two variables. One subtype variable had subtypes in four categories (luminal A, luminal B, HER2, or basal). The other subtype variable was dichotomized as basal or not basal. Logistic regression was used to evaluate the association of covariates of interest on the probability of a locoregional recurrence. Cox regression was used to examine the association of covariates of interest on time to first locoregional recurrence. Only univariable models were reported, based in large part by the relatively small number
Impact of Breast Cancer Molecular Subtypes
2853
of locoregional recurrences (n = 11). The Kaplan–Meier method was used to estimate the time to local recurrence. The log-rank test was used to test for the significance between basal and not basal estimated time to event curves. Analyses were performed with SAS statistical software, Version 9.2 (SAS Institute Inc., Cary, NC).
RESULTS A total of 149 patients who underwent neoadjuvant chemotherapy for locally advanced breast cancer who had complete IHC data for ER, PR, and HER2 were identified. All were diagnosed and treated between May 1997 and April 2005; 85% were diagnosed after June 2000. Fifty-five (37%) were luminal A, 25 (17%) were luminal B, 20 (13%) were HER2-enriched, and 49 (33%) were basal-like. All were followed for at least 24 months after diagnosis, and the median follow-up of survivors was 55 months. Patient demographic and tumor data are reported in Table 1. All patients were stage II (45%) or stage III (55%) at presentation. The majority of patients were T2 (37%) or T3 (54%), and 60% were node-positive.
Patients in this analysis were treated with a variety of chemotherapy regimens (Table 2). However, most received an anthracycline (93%) or a taxane (81%) and 74% received both. Three received chemotherapy that did not contain a taxane or anthracycline. In HER2? patients, 38% received trastuzumab in addition to cytotoxic chemotherapy, reflecting a change in treatment paradigm during the study period (Table 2). One hundred patients were treated with mastectomy and 49 with BCT. Patients with a basallike subtype were more likely to have BCT when compared with all other subtypes in dichotomized fashion (p = 0.01) but not when compared independently to all four subtypes (p = 0.08). Most (88%) received radiotherapy as part of their treatment, including all with BCT. The proportion of patients who received radiation was no different among molecular subtypes. Pathologic analysis of tumors revealed a pathologic complete response rate of 26%. This ranged from 5% in luminal A patients to 50% in HER2 patients (p = 0.002). The proportion of patients who had a pathologic T1 tumor after neoadjuvant chemotherapy was different among subtypes; luminal A patients were less likely to have a pathologic T1 tumor (38%) compared with other subtypes (69–85%; p \ 0.0001).
TABLE 1 Demographic and pretreatment stage data Luminal A (n = 55)
Luminal B (n = 25)
HER2 (n = 20)
Basal (n = 49)
All (n = 149)
p All 4 subtypes
p basal/ not basal
0.11
0.003
Age (year) B35
1 (2%)
3 (12%)
6 (30%)
8 (16%)
18 (12%)
[35
54 (98%)
22 (88%)
14 (70%)
41 (84%)
131 (88%)
Median age
51 ± 9
48 ± 11
41 ± 11
47 ± 10
48 ± 11
0.34
0.003
White
39 (71%)
16 (64%)
14 (70%)
24 (49%)
93 (62%)
0.19
0.21
Black Other
13 (24%) 3 (5%)
7 (28%) 2 (8%)
6 (30%) 0
23 (47%) 2 (4%)
49 (24%) 7 (14%)
T1
1 (2%)
0
1 (5%)
2 (4%)
4 (3%)
0.87
0.56
T2
23 (42%)
8 (32%)
5 (25%)
19 (39%)
55 (37%)
T3
23 (42%)
15 (60%)
12 (60%)
21 (43%)
81 (54%)
T4
8 (15%)
2 (8%)
1 (5%)
5 (10%)
16 (11%)
N0
25 (45%)
10 (40%)
4 (20%)
21 (43%)
60 (40%)
0.55
0.31
N1
23 (42%)
11 (44%)
10 (50%)
18 (37%)
62 (42%)
N2
7 (13%)
4 (16%)
4 (20%)
7 (14%)
22 (15%)
N3
0
0
2 (10%)
3 (6%)
5 (3%)
IIa
11 (20%)
1 (4%)
2 (10%)
9 (18%)
23 (15%)
0.66
0.13
IIb
15 (27%)
13 (52%)
4 (20%)
13 (27%)
45 (30%)
IIIa IIIb
20 (36%) 9 (16%)
9 (36%) 2 (8%)
11 (55%) 1 (5%)
19 (39%) 5 (10%)
59 (40%) 17 (11%)
IIIc
0
0
2 (10%)
3 (6%)
5 (3%)
Race
T Stage
N stage
AJCC stage
2854
M. O. Meyers et al.
TABLE 2 Treatment received as a function of molecular subtype Luminal A (n = 55)
Luminal B (n = 25)
HER2 (n = 20)
Basal (n = 49)
All (n = 149)
Anthracycline
50 (91%)
24 (96%)
19 (95%)
46 (94%)
139 (93%)
Taxane (with or without anthracycline) Trastuzumab
43 (78%)
20 (80%)
20 (100%)
38 (78%)
121 (81%)
NA
6 (24%)
11 (55%)
NA
NA
BCT
15 (27%)
7 (28%)
5 (25%)
23 (47%)
49 (33%)
Mastectomy
40 (73%)
18 (72%)
15 (75%)
36 (53%)
100 (67%)
49 (89%)
21 (84%)
18 (90%)
43 (88%)
131 (88%)
Chemotherapy
Surgery
Radiotherapy received
A locoregional recurrence was identified in 11 patients (7%). Median time to recurrence was 14.2 (range, 7–33) months (Table 3). There were two recurrences in patients who achieved BCT (4%) and nine with mastectomy (9%; p = 0.32). Seven basal-like patients (14%) developed a locoregional recurrence compared with 4–5% for the other three subtypes (p = 0.02; Fig. 1). Seven occurred without a simultaneous distant recurrence and four were detected at the same time as a distant recurrence. In the seven patients with an isolated regional recurrence, five eventually developed distant recurrences, which were recognized between 2 and 57 months after detection of locoregional recurrence. The two patients who only had an LRR were in-breast recurrences in basal-like patients. In the four patients with a simultaneous local and distant recurrence, two were basal-like, and one each was luminal A and HER2-enriched. In the patients who achieved BCT who developed a local recurrence, both were basal-like subtype and both developed local recurrences in the breast after receiving postoperative radiation therapy (Table 4). There were no regional lymph node recurrences identified in the BCT patients. Of the nine patients who required mastectomy who developed a locoregional recurrence, two were luminal A, one was luminal B, and one was HER2-enriched. The remaining five were basal-like. The luminal B and HER2-enriched patients both received trastuzumab, the use of which did not yield any difference in LRR rate: 2/17 (12%) in those receiving vs. those not receiving (0/28) trastuzumab (p = 0.14). The majority of recurrences were in the chest wall, with eight having this as a component of
FIG. 1 Kaplan–Meier curve comparing locoregional recurrence-free survival between basal-like subtype and all other molecular subtypes by log-rank test
local failure. Five regional lymph node recurrences were identified, but only one was isolated. In the mastectomy patients developing a local recurrence, eight (88%) received postoperative radiotherapy. The one patient who did not receive radiation was a basal-like patient with a node-negative T2 tumor who developed a chest wall recurrence. Type of chemotherapy was not associated with a difference in LRR rate. Those receiving an anthracycline and a taxane (n = 111) had a 6% incidence of locoregional recurrence (n = 7), whereas those who received only an
TABLE 3 Local and regional recurrence rates as a function of surgery type and molecular subtype Luminal A (n = 55)
Luminal B (n = 25) 0/7
HER2 (n = 20) 0/5
Basal (n = 49)
All (n = 149)
p
BCT (n = 49)
0/15
2/24 (8%)
2/49 (4%)
0.99
Mastectomy (n = 100)
2/40 (5%)
1/18 (6%)
1/15 (7%)
5/25 (20%)
9/100 (9%)
0.06
Any (n = 149)
2/55 (4%)
1/25 (4%)
1/20 (5%)
7/49 (14%)
11/149 (7%)
0.04
Impact of Breast Cancer Molecular Subtypes
2855
TABLE 4 Location of locoregional recurrence and incidence of radiation therapy as a function of surgery type and molecular subtype Lum A
Lum B
HER2
Basal
All
Included in XRT field
Chest wall
2
1
1
4
8
7a
Axillary LN
0
1
0
1
2
1
Supraclavicular LN
0
0
0
3
3
2b
2 sites
NA
1
NA
3
4
In breast
0
0
0
2
2
Axillary LN
0
0
0
0
0
NA
Supraclavicular LN
0
0
0
0
0
NA
Mastectomy (n = 9)
BCT (n = 2)
a
2
One patient with a chest wall recurrence did not receive post-mastectomy radiation
b
One patient with a combined supraclavicular and chest wall recurrence who received post-mastectomy radiation did not specific records as to whether the supraclavicular basin was included in the radiation field
TABLE 5 Covariates associated with locoregional recurrence in univariable logistic regression and Cox regression modeling Covariate
Logistic regression OR (95% CI)
Molecular subtype (basal/not basal)
4 (1.1–14.4)
Cox regression p
HR (95% CI)
p
0.03
5.3 (1.4–20)
Number ? axillary LNs Posttreatment T stage
1.2 (1.1–1.3) 1.9 (1.1–3.2)
0.001 0.02
1.1 (1.06–1.2) 1.9 (1.1–3.3)
0.02
Pathologic tumor size
1.2 (1.01–1.5)
0.04
1.2 (1–1.4)
0.05
Posttreatment N stage
2.7 (1.5–4.8)
0.001
2.6 (1.5–4.4)
0.0005
Presence of LN extracapsular extension
5.7 (1.9–16.8)
0.002
4.1 (1.8–9.8)
0.001
\0.0001 0.01
Only univariable model results are reported due to the relatively small effective sample size of locoregional recurrences (n = 11) for the logistic regression analysis, and deaths for the Cox regression analysis (= 10)
anthracycline (n = 28) had a 14% recurrence rate (n = 4; p = 0.23). A logistic regression analysis of clinical and pathologic variables associated with recurrence was performed (Table 5). A number of previously recognized variables, including number of positive axillary lymph nodes, presence of lymph node extracapsular extension, posttreatment pathologic T and N stage, and pathologic tumor size were associated with the development of a locoregional recurrence. Molecular subtype also was associated with a local recurrence, with a basal-like subtype more likely to develop a local recurrence (p = 0.03). Importantly, pretreatment clinical staging, including both clinical T and N staging, was not associated with the development of a local recurrence, suggesting that staging after neoadjuvant chemotherapy is more prognostic than pretreatment assessment. Although pathologic tumor size and T stage were covariates associated with locoregional recurrence, the presence of a pathologic complete response was not associated. Given the limited number of recurrences in this data set, multivariable modeling was not undertaken because it would be of limited utility with such a small number of patients.
DISCUSSION In this series of women with locally advanced breast cancer who underwent neoadjuvant chemotherapy, we have demonstrated that rates of locoregional recurrence are significantly variable depending on molecular subtype as determined by immunohistochemistry for ER, PR, and HER2. The local recurrence of all patients in this series was only 7%. However, a basal-like subtype had a locoregional recurrence rate of 14% compared with 4–5% for all other subtypes with most of these occurring in women who had undergone a mastectomy. Although molecular subtype has proven to be a powerful predictor of survival and distant recurrence, the data are more limited as to the impact on locoregional recurrence.18 Several authors have reported their results evaluating this question, although comparisons among these reports are challenging because of differences in patient populations.13–16 Some included only patients who underwent BCT, whereas others included mastectomy. In addition, one trial included only African-American patients. Three of these reports have suggested similar rates of locoregional
2856
recurrence between varying molecular subtypes.13,14,16 In one series, the rate of in-breast relapse was 17% in both basal-like patients and all other patients at 5 years, with the basal-like patients more likely to have a regional lymph node recurrence (6%) than others (1%).13 In another series, the local recurrence rate was almost identical for basal-like patients compared with other subtypes (13 vs. 12%). Yet another series found no significant difference in locoregional recurrence between molecular subtypes regardless of whether they were treated with BCT or mastectomy in black women.14 The most recent report to examine this identified a significant difference in locoregional recurrence in patients of varying molecular subtype. In 793 women who underwent BCT between July 1998 and December 2001,15 those women with HER2-enriched or basal-like subtype were more likely to have a locoregional recurrence (8.4 and 7.1%, respectively) compared with very low rates of locoregional recurrence for the luminal A (0.8%) or luminal B (1.5%) subtypes. The most recent series examining this has not yet been published but was recently presented. In a cohort of 644 patients treated with BCT after neoadjuvant chemotherapy, those with basal-like and HER2-enriched subtypes had a higher incidence of locoregional recurrence, particularly in those patients who did not have a pathologic complete response to chemotherapy and those who had four or more positive nodes.19 Although there are some similarities between our data set and those of previously published series, several critical differences exist. The most significant of these is that our data includes only patients treated with neoadjuvant chemotherapy for locally advanced breast cancer. With the exception of data presented only in abstract form, all of the previous reports have included patients who underwent surgery first. Second, two of these reports included only those patients who underwent BCT and did not include those who underwent mastectomy as ours has.13,15 Third, only the most recent of these previous series included patients treated exclusively during the most recent era, all having been treated since 1998.14,15 The earlier work included patients treated from 1980 to 2003 and 1987 to 1997, respectively.13,16 Inclusion of patients from previous eras of treatment may yield higher rates of locoregional recurrence compared with those treated more recently due to several factors. The evolution of systemic therapy has continued, which likely has an effect on local control just as it does on systemic recurrence. The use of trastuzumab in appropriate HER2? patients may magnify the differences between HER2? and HER2-/basal like patients. The use of radiation, particularly the use of postmastectomy radiation, as well as radiation to regional lymph node basins, has expanded and may have an impact. Breast imaging has evolved, particularly with use of high-quality ultrasound and breast magnetic resonance imaging, which
M. O. Meyers et al.
likely impacts in-breast recurrence in those achieving BCT. These factors, among others, likely serve to decrease rates of locoregional recurrence in patients treated in the recent past relative to previous eras, making it difficult to draw absolute comparisons between data and explaining, at least in part, the differences seen between series. Another consideration of this work is that the majority of the patients in our series who developed a locoregional recurrence received postoperative radiation therapy. There is some question about the relative benefit of radiation between molecular subtypes of breast cancer. Although a clear picture has not emerged, data have shown that those with a basal-like subtype derived less benefit in local control from postmastectomy radiation compared with other subtypes.20 Taken together with the data from Nguyen et al. in patients treated with BCT, which demonstrated that the basal-like and HER2-enriched subtypes have a higher rate of in-breast recurrence than other subtypes, this suggests that radioresistance may play a role in this finding.15 There are several limitations of our data that preclude drawing definitive conclusions. The first of these is the modest number of patients in this data set and the small number who developed a locoregional recurrence. Confirmation of these findings in a larger data set will be important. The second limitation is the differing systemic therapy between patients in this cohort and the impact that this had on local and regional recurrence. That said, the low local recurrence rate observed in this series likely approximates what would be expected with current treatment paradigms. Another consideration is the number of patients in our data who developed a synchronous or metachronous distant recurrence in association with the locoregional recurrence. In this regard, there were no significant differences between molecular subtypes as to the incidence of synchronous as opposed to metachronous development of a distant recurrence or absence of distant recurrence. It can be rationally argued that in patients with a simultaneous distant recurrence, the presence of a locoregional recurrence becomes clinically less relevant and is less likely to influence overall outcome and treatment decisions. However, with improvement in systemic therapy for patients with metastatic breast cancer, locoregional recurrences in these patients may become increasingly clinically relevant. Acknowledging these limitations, we have shown that basal-like subtype is associated with the development of a locoregional recurrence after neoadjuvant chemotherapy. This is similar to previous findings that molecular subtype is an important prognostic variable for survival in breast cancer. This appears to be independent of pretreatment T or N stage, because these were not associated with a locoregional recurrence in this data set. Similarly, this is independent of whether a complete pathologic response
Impact of Breast Cancer Molecular Subtypes
was seen to neoadjuvant chemotherapy. Taken together, these findings suggest that the development of a locoregional recurrence is predicated more by biology of disease, as evidenced by molecular subtype, than by standard clinical and pathologic risk factors. This suggests that attention to novel locoregional therapy, in addition to focusing on advances in systemic therapy, may be important in basal-like patients to try and decrease local recurrence rates. Although improvements in systemic therapy alone are likely to impact this, investigation of radiosensitizing agents, radiation dose escalation, and other novel techniques may prove to be important, particularly as treatment of metastatic breast cancer continues to improve. REFERENCES 1. Chen AM, et al. Breast conservation after neoadjuvant chemotherapy: the MD Anderson cancer center experience. J Clin Oncol. 2004;22(12):2303–12. 2. Cance WG, et al. Long-term outcome of neoadjuvant therapy for locally advanced breast carcinoma: effective clinical downstaging allows breast preservation and predicts outstanding local control and survival. Ann Surg. 2002;236(3):295–302; discussion 302-3. 3. Bonadonna G, et al. Primary chemotherapy in operable breast cancer: eight-year experience at the Milan Cancer Institute. J Clin Oncol. 1998;16(1):93–100. 4. Anders CK, Carey LA. Biology, metastatic patterns, and treatment of patients with triple-negative breast cancer. Clin Breast Cancer. 2009;9(Suppl 2):S73–81. 5. Dairkee SH, Puett L, Hackett AJ. Expression of basal and luminal epithelium-specific keratins in normal, benign, and malignant breast tissue. J Natl Cancer Inst. 1988;80(9):691–5. 6. Perou CM, et al. Molecular portraits of human breast tumours. Nature. 2000;406(6797):747–52. 7. Rouzier R, et al. Breast cancer molecular subtypes respond differently to preoperative chemotherapy. Clin Cancer Res. 2005;11(16): 5678–85.
2857 8. Sorlie T, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci USA. 2001;98(19):10869–74. 9. Sorlie T, et al. Repeated observation of breast tumor subtypes in independent gene expression data sets. Proc Natl Acad Sci USA. 2003;100(14):8418–23. 10. Sotiriou C, et al. Breast cancer classification and prognosis based on gene expression profiles from a population-based study. Proc Natl Acad Sci USA. 2003;100(18):10393–8. 11. Nielsen TO, et al. Immunohistochemical and clinical characterization of the basal-like subtype of invasive breast carcinoma. Clin Cancer Res. 2004;10(16):5367–74. 12. Carey LA, et al. Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study. JAMA. 2006;295(21):2492–502. 13. Haffty BG, et al. Locoregional relapse and distant metastasis in conservatively managed triple negative early-stage breast cancer. J Clin Oncol. 2006;24(36):5652–7. 14. Ihemelandu CU, et al. Treatment and survival outcome for molecular breast cancer subtypes in black women. Ann Surg. 2008;247(3):463–9. 15. Nguyen PL, et al. Breast cancer subtype approximated by estrogen receptor, progesterone receptor, and HER-2 is associated with local and distant recurrence after breast-conserving therapy. J Clin Oncol. 2008;26(14):2373–8. 16. Dent R, et al. Triple-negative breast cancer: clinical features and patterns of recurrence. Clin Cancer Res. 2007;13(15 Pt 1):4429–34. 17. Carey LA, et al. Potential predictive markers of benefit from cetuximab in metastatic breast cancer: an analysis of two randomized phase 2 trials. Proc AACR/CTRC San Antonio Breast Cancer Symposium, 2009. Abstract 2014. 18. Parker JS, et al. Supervised risk predictor of breast cancer based on intrinsic subtypes. J Clin Oncol. 2009;27(8):1160–7. 19. Yu TK, et al. Local-regional control according to surrogate markers of breast cancer subtypes and response to neoadjuvant chemotherapy in breast cancer patients undergoing breast conserving therapy. Proc AACR/CTRC San Antonio Breast Cancer Symposium, 2009. Abstract 957. 20. Kyndi M, et al. Estrogen receptor, progesterone receptor, HER-2, and response to postmastectomy radiotherapy in high-risk breast cancer: the Danish Breast Cancer Cooperative Group. J Clin Oncol. 2008;26(9):1419–26.