Abstract
Background
Retrospective analyses of NSABP B20 and SWOG 8814 showed a large benefit of chemotherapy in patients with ER-positive tumors and high OncotypeDX Recurrence Score (RS≥31). However, it might be possible that both studies may be contaminated by non-luminal tumors, especially in high-risk RS group.
Methods
We conducted simulations in order to obtain a better understanding of how the NSABP B20 and SWOG 8814 results would have been if non-luminal breast cancer would have been excluded. Simulations were done separately for the node-negative and node-positive cohorts.
Results and conclusion
The results of the simulations suggest that the non-luminal tumors are augmenting the apparent benefit of chemotherapy, but do not appear to be responsible for the entire effect. These simulations could provide information about the potential influence of contamination by unexpected tumor subtypes on the future results of TAILORx and RxPONDER clinical trials
Key words: basal-like, ER-positive, Her2-enriched, luminal A, luminal B, PAM50, OncotypeDX Recurrence Score
introduction.
Adjuvant chemotherapy has been widely used in the treatment of estrogen receptor (ER) and/or progesterone receptor (PR)-positive breast cancer. Based on the Early Breast Cancer Trialists Collaborative Group meta-analysis, the addition of adjuvant chemotherapy to tamoxifen reduces the risk of breast cancer relapse and mortality in hormone receptor-positive disease by ∼30% and 20%, respectively, but without considering subgroups [1]. The indication of adjuvant chemotherapy includes women with negative axillary lymph nodes and tumors above 0.5 cm at very low absolute risk of recurrence [2]. Routine use of adjuvant chemotherapy for women with positive axillary lymph node(s) is recommended [3, 4].
Oncotype DX recurrence score in NSABP B20 and SWOG 8814
Retrospective analyses of NSABP B20 and SWOG 8814 suggested that not all patients with ER-positive disease benefit from the addition of chemotherapy. Both studies support the use of Oncotype DX 21-gene recurrence score (RS) to define patients requiring multi-agent chemotherapy. In NSABP B20, a trial for node-negative breast cancer [5], patients with high RS (≥31) tumors had a large benefit in distant recurrence-free interval from cyclophosphamide, methotrexate, and fluorouracil or methotrexate and fluorouracil chemotherapy [relative risk (RR) = 0.26, 95% confidence interval (CI) 0.13–0.53]. Patients with low-RS (<18) tumors did not benefit from chemotherapy (RR = 1.31, 95% CI 0.46–3.78). Patients with intermediate RS tumors (18–30) did not appear to derive a large benefit from chemotherapy (RR = 0.61, 95% CI 0.24–1.59). In SWOG 8814, a trial for node-positive breast cancer [6], it was found that patients with high RS (≥31) tumors had a significant benefit in disease-free survival (DFS) from chemotherapy [hazard ratio (HR) = 0.59, 95% CI 0.35–1.01]. There was no cyclophosphamide, doxorubicin (adriamycin), and fluorouracil chemotherapy benefit in patients with low-RS (<18) tumors (HR = 1.02, 95% CI 0.54–1.93). Patients with intermediate RS tumors (18–30) did not have an apparent benefit (HR = 0.72, 95% CI 0.39–1.31).
methods
intrinsic subtypes of breast cancer
Breast cancer is a biologically heterogeneous disease and molecular characterization studies over the last decade have identified four main intrinsic subtypes (i.e. luminal A, luminal B, HER2-enriched and basal-like) with significant differences in terms of their incidence, risk factors, prognosis and response to therapies [7, 8]. Among them, nonluminal subtypes (i.e. HER2-enriched and Basal-like) show consistent higher pathological complete response (pCR) rates following neoadjuvant multi-agent chemotherapy compared with luminal disease (i.e. luminal A and B) [9, 10].
The pathological-based biomarkers (e.g. ER, PR, HER2 and Ki67 or grade) currently used in the clinical setting either alone or in combination, do not fully recapitulate the intrinsic subtypes of breast cancer [11]. This poses the question if the chemotherapy benefit observed in NSABP B20 and SWOG 8814 studies might be due to the presence of the generally chemotherapy sensitive nonluminal subtypes? Indeed, gene expression analyses using the research-based PAM50 50-gene classifier have revealed that, although the luminal subtypes predominate within ER-positive breast cancer, all the intrinsic subtype can be identified in ER-positive disease; for example a re-analysis of a combined cohort of 337 patients with ER+/HER2-unknown and node-negative early breast cancer treated without adjuvant systemic therapy (i.e. no chemotherapy and no endocrine therapy), showed that 18.7% of tumors were of the nonluminal subtype [12]. In this cohort, intrinsic subtyping and a microarray-based version of Oncotype DX RS were found significantly associated with survival outcome (Figure 1A and B). Interestingly, 29.1%, 7.7% and 0% of the patients in the Oncotype DX RS high-, intermediate and low-risk groups had nonluminal breast cancer, respectively (Figure 1B).
Figure 1.
Relapse-free survival of patients with ER-positive/HER2-unknown early breast cancer treated without adjuvant systemic therapy. (A) Based on the intrinsic subtypes; (B) based on microarray-based Oncotype DX RS groups.
Further supporting this hypothesis of subtype heterogeneity within ER+ disease, ER status determination has evolved over time. Of note in the NSABP B20, and in a subset of patients from SWOG 8814, ER status was not determined by the current immunohistochemical (IHC)-based methods but by the ligand-binding assay (LBA). This is important since both methods of determination of ER levels are different, and a 14%–22% ER negativity by IHC has been observed among ER-positive tumors by LBA [13]. Similar discordant rates have been observed between two different IHC-based assays for ER status [14]. For example a recent ring study of the two central pathology laboratories of the ALTTO phase III trial (BIG 2-06/NCCTG N063D) that used different IHC-based assays for ER determination observed a ∼12% discordance rate between local and central, and a 15% discordance rate between the two central laboratories [15]. Finally, NSABP B20 and SWOG 8814 did not take into account HER2 status, and HER2 positivity enriches for the presence of the HER2-enriched subtype [11, 16, 17]. In fact, recent studies with central pathology determination of ER and HER2 and gene expression analyses have shown that the nonluminal contamination within ER+/HER2-negative disease represents ∼5.0% [18, 19].
Overall, these data suggest that NSABP B20 and SWOG 8814 may be contaminated by nonluminal tumors, especially in high-risk RS group, which may have more contamination from nonluminal disease. Thus, how would the NSABP B20 and SWOG 8814 results according to Oncotype Dx RS be if nonluminal tumors/patients had been excluded?
simulation of chemotherapy benefit in ER-positive breast cancer
We evaluated distant relapse-free survival (DRFS) data from a previously published combined cohort of 1227 patients with ER+ breast cancer treated with adjuvant tamoxifen-only (551 node-negative and 676 node-positive) [20]. To simulate the addition of chemotherapy effect, it was assumed that among patients who had a DRFS event before 5 years, 43% of the basal-like group, 35% of HER2-enriched group, 17% of luminal B group and 3% of luminal A group could be distant relapse-free at 5 years of follow-up if chemotherapy would have been administered; these estimated chemotherapy effects/times were based on the pCR rate of each intrinsic subtype following neoadjuvant anthracycline/taxane-based chemotherapy [9]. Patients who attain pCR following neoadjuvant chemotherapy have outstanding survival outcomes at 5 years [21., 22., 23.].
Simulations were done separately for the node-negative and node-positive cohorts (Table 1). We used nine different set of assumptions regarding the distribution of the intrinsic subtypes in each ER-positive breast cancer population. The distribution of the intrinsic subtypes within each type of ER-positive population (except the all luminal A and all basal-like cases) was based on a cohort of 337 patients with ER+/HER2-unknown tumors with complete Oncotype DX RS and intrinsic subtype information [12]. Under each set of assumptions, first, a random sample set of 1000 patients was drawn with replacement. Second, a survival advantage as described above was added in 50% of the patients of each random sample set to simulate the results if chemotherapy would have been administered. Finally, the HR of tamoxifen plus chemotherapy versus tamoxifen and the 5-year DRFS percentages for both treatments were calculated. Each random sample set was repeatedly drawn 10 000 times.
Table 1.
Simulation results under different scenarios and distribution of the intrinsic subtypesa
| Type of ER-positive population | Distribution of subtypes | Node-negative |
Node-positive |
||||
|---|---|---|---|---|---|---|---|
| Geometric mean of HR | Mean 5-year DRFS % |
Geometric mean of HR | Mean 5-year DRFS % |
||||
| Tamoxifen-only | Tam. + Chemo. | Tamoxifen-only | Tam. + Chemo. | ||||
| All LumA | LumA = 100% | 0.98 (0.62, 1.54) | 96% (93%, 97%) | 96% (94%, 98%) | 0.98 (0.76, 1.27) | 85% (81%, 88%) | 85% (82%, 88%) |
| All basal-like | Basal-like = 100% | 0.52 (0.42, 0.64) | 66% (62%, 70%) | 81% (78%, 83%) | 0.51 (0.42, 0.63) | 63% (58%, 67%) | 79% (76%, 81%) |
| Fan et al. | Basal-like = 5.9%, HER2-E = 12.8%, LumA = 43.9%, LumB = 37.4% | 0.82 (0.61, 1.11) | 88% (85%, 90%) | 90% (88%, 93%) | 0.85 (0.68, 1.05) | 74% (71%, 78%) | 79% (76%, 82%) |
| Fan et al. (with no contamination) | Basal-like = 0%, HER2-E = 0%, LumA = 54%, LumB = 46% | 0.91 (0.66, 1.26) | 91% (88%, 93%) | 92% (90%, 94%) | 0.91 (0.73, 1.12) | 77% (73%, 80%) | 80% (76%, 83%) |
| Low RS | Basal-like = 0%, HER2-E = 0%, LumA = 90.4%, LumB = 9.6% | 0.96 (0.64, 1.45) | 95% (92%, 97%) | 95% (93%, 97%) | 0.96 (0.75, 1.24) | 83% (79%, 86%) | 84% (81%, 87%) |
| Interm. RS (contaminated) | Basal-like = 3.1%, HER2-E = 4.6%, LumA = 67.7%, LumB = 24.6% | 0.88 (0.63, 1.24) | 91% (89%, 94%) | 93% (91%, 95%) | 0.9 (0.72, 1.13) | 79% (75%, 82%) | 82% (78%, 85%) |
| Interm. RS (with no contamination) | Basal-like = 0%, HER2-E = 0%, LumA = 73.3%, LumB = 26.7% | 0.94 (0.65, 1.34) | 93% (90%, 95%) | 94% (91%, 95%) | 0.93 (0.74, 1.18) | 80% (76%, 84%) | 82% (79%, 85%) |
| High RS (contaminated) | Basal-like = 9.0%, HER2-E = 20.1%, LumA = 19.1%, LumB = 51.8% | 0.78 (0.60, 1.03) | 84% (80%, 87%) | 88% (85%, 90%) | 0.8 (0.66, 0.98) | 70% (66%, 74%) | 77% (73%, 80%) |
| High RS (with no contamination) | Basal-like = 0%, HER2-E = 0%, LumA = 26.9%, LumB = 73.1% | 0.89 (0.67, 1.19) | 88% (85%, 90%) | 90% (87%, 92%) | 0.88 (0.72, 1.07) | 72% (68%, 76%) | 76% (73%, 80%) |
RS, recurrence score; HR, hazard ratio; DRFS, distant relapse-free survival; HER2-E, HER2-enriched.
Simulations were done separately for the node-negative and node-positive cohorts. In each case, 10 000 replicates were analyzed. In each replicate, the sample size was 1000. Summary data of the geometric mean of HR (tamoxifen + chemotherapy versus tamoxifen-only), mean 5-year DRFS rates and their 2.5th and 97.5th percentiles out of the 10 000 replicates were obtained for each case. The distribution of the intrinsic subtypes within each type of ER-positive population was based on a cohort of 337 ER+/HER2-unknown patients with complete microarray-based Oncotype DX RS and intrinsic subtype information. In three scenarios, contaminating nonluminal subtypes (i.e. basal-like and HER2-enriched) were removed. In the low Oncotype DX RS group, no potential contamination issue was observed in the dataset evaluated.
results and discussion
As shown in our simulations (Table 1), as the percentage of patients with nonluminal disease increases, the apparent effectiveness of chemotherapy also increases. A higher percent of patients with high-risk RS tumors presumably have nonluminal disease, thus the contamination of nonluminal patients on the chemotherapy effect could be large in this group of patients. The geometric mean of the HR of tamoxifen plus chemotherapy versus tamoxifen increased from 0.78 to 0.89 for node-negative and 0.80 to 0.88 for node-positive cohorts in our simulations, after removing the nonluminal cases. In patients with intermediate risk RS tumor, the percentage of nonluminal cases is lower. There was some contamination of nonluminal cases, but the influence on the chemotherapy effect was small. The geometric mean of the HR of tamoxifen plus chemotherapy versus tamoxifen increased from 0.88 to 0.94 for node-negative and 0.90 to 0.93 for node-positive cohorts in our simulations, after removing the nonluminal subtype. For patients with low-risk RS tumors, there were no nonluminal cases and, thus, no contamination of nonluminal subtype on the chemotherapy effect was analyzed.
TAILORx and RxPONDER are phase III clinical trials using the Oncotype DX assay to identify patients for whom adjuvant endocrine therapy alone is sufficient treatment due to their inherent prognosis with endocrine therapy alone, due to their lack of benefit from chemotherapy, or both [24, 25]. TAILORx is designed for node-negative patients and enrolled its final patient in 2010. RxPONDER is designed for patients with one to three positive nodes and is still open to accrual.
The primary objective of TAILORx [24] is to determine whether adjuvant hormonal therapy is not inferior to adjuvant chemotherapy plus hormonal therapy for patients with ER-positive/HER2-negative breast cancer with intermediate Oncotype Dx RS values between 11 and 25. In this study, after the Oncotype DX RS is determined, patients are assigned or randomized based on the RS. Patients with RS 10 or less are assigned to receive hormonal therapy alone. Patients with RS 11–25 are randomized to receive chemotherapy plus hormonal therapy versus hormonal therapy alone. Patients with RS 26 or higher are assigned to receive chemotherapy plus hormonal therapy. The primary end point is DFS. A difference in the 5-year DFS rate from 90% with chemotherapy to 87% or lower on hormonal therapy alone (corresponding to a HR for hormonal versus chemotherapy + hormonal of 1.322) would be considered unacceptable. The accrual goal for the randomized cohort is 6860 patients. The study accrued more than 11 000 patients and was closed in October 2010 and the final analyses are expected in 2015.
The primary objective of RxPONDER [25] is to test whether the difference of chemotherapy compared with no chemotherapy depends directly on RS score in ER-positive/HER2-negative breast cancer with one to three positive axillary nodes. If the difference depends on RS, an optimal cut point for recommending chemotherapy will be determined. In this study, after knowing the results of Oncotype DX RS, patients with RS 25 or less are randomized to receive chemotherapy plus endocrine therapy versus endocrine therapy alone. The primary end point is DFS. In the final analysis, the interaction of treatment and RS will be tested first. If it is statistically significant and a point of equivalence can be obtained in the range 0–25 from a Cox regression model with the interaction term, the upper limit of a one-sided 95% CI for the point of equivalence marks the RS for which there is a significant benefit of chemotherapy. If this upper limit is within 0–25, then a clinically significant effect of chemotherapy can be expected for RS values above this cutoff point. Below the cutoff point, there will be no statistically significant chemotherapy benefit. If the interaction of treatment and RS is not statistically significant, the overall effect of chemotherapy will be tested in all patients. The study plans to randomize 4000 patients via screening more than 9000 patients. The study is still accruing patients.
In our dataset evaluated here, the percentages of luminal A, luminal B, HER2-enriched and Basal-like in patients with RS of 11–25, were 85.7%, 12.7%, 1.6% and 0%, respectively. For patients with RS ≤25 tumors, the percentages of luminal A, luminal B, HER2-enriched and basal-like were 89.6%, 9.4%, 0.9% and 0%, respectively. Thus, the potential contamination of nonluminal subtype on the chemotherapy effect in patients with RS 11–25 or RS ≤25 could be small.
Our study has a few limitations. First, our simulations in the dataset of 1227 patients with ER-positive disease were based on the outcome of all patients, regardless of the RS scores. Thus, our simulations were only looking at the effect of different composites of percentages of the four intrinsic subtypes. Second, the chemotherapy effect was based on pCR as a surrogate marker of ‘non-relapse’ despite that the risk of relapsing in patients with pCR is not exactly zero [21]. In addition, the pCR rates of each intrinsic subtype were obtained from a cohort of patients with ER-positive and ER-negative tumors; however, less clear are the pCR rates of the intrinsic subtypes within ER-positive/HER2-negative disease. Finally, our simulations only considered the relative percent of patients who were cured with chemotherapy at the 5-year time point. Thus, we did not consider the potential delay in time to distant recurrence due to chemotherapy.
To conclude, we conducted simulations in order to obtain a better understanding of how the NSABP B20 and SWOG 8814 results would have been if nonluminal breast cancer would have been excluded. The results of the simulations suggest that the nonluminal tumors are augmenting the apparent benefit of chemotherapy, but does not appear to be responsible for the entire effect. This finding supports the current St Gallen International Expert Consensus of administering adjuvant multi-agent chemotherapy in addition to endocrine therapy in luminal B tumors [26]. However, our data suggest that the magnitude of benefit from chemotherapy in this subset of patients with luminal disease is likely lower than the one reported in the Oxford meta-analysis [1]. These simulations could also provide information about the potential influence of contamination by unexpected tumor subtypes on the future results of TAILORx and RxPONDER clinical trials.
disclosure
CMP is equity stock holder of BioClassifier LLC and University Genomics. CMP and MCUC have filed a patent on the PAM50 assay. Uncompensated advisory role of AP for Nanostring Technologies. All remaining authors have declared no conflicts of interest.
Contributor Information
R.D. Gelber, Email: gelber@jimmy.harvard.edu.
C.M. Perou, Email: chuck_perou@med.unc.edu.
references
- 1.Collaborative G., Early Breast Cancer Trialists' Comparisons between different polychemotherapy regimens for early breast cancer: meta-analyses of long-term outcome among 100000 women in 123 randomised trials. Lancet. 2012;379:432–444. doi: 10.1016/S0140-6736(11)61625-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lehmann B.D., Bauer J.A., Chen X. Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies. J Clin Invest. 2011;121:2750–2767. doi: 10.1172/JCI45014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hayes D.F. Targeting adjuvant chemotherapy: a good idea that needs to be proven! J Clin Oncol. 2012;30:1264–1267. doi: 10.1200/JCO.2011.38.4529. [DOI] [PubMed] [Google Scholar]
- 4.Slamon D.J., Leyland-Jones B., Shak S. Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med. 2001;344:783–792. doi: 10.1056/NEJM200103153441101. [DOI] [PubMed] [Google Scholar]
- 5.Paik S., Tang G., Shak S. Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol. 2006;24:3726–3734. doi: 10.1200/JCO.2005.04.7985. [DOI] [PubMed] [Google Scholar]
- 6.Albain K.S., Barlow W.E., Shak S. Prognostic and predictive value of the 21-gene recurrence score assay in postmenopausal women with node-positive, oestrogen-receptor-positive breast cancer on chemotherapy: a retrospective analysis of a randomised trial. Lancet Oncol. 2010;11:55–65. doi: 10.1016/S1470-2045(09)70314-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Perou C.M., Sorlie T., Eisen M.B. Molecular portraits of human breast tumours. Nature. 2000;406:747–752. doi: 10.1038/35021093. [DOI] [PubMed] [Google Scholar]
- 8.TCGA Comprehensive molecular portraits of human breast tumours. Nature. 2012;490:61–70. doi: 10.1038/nature11412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Perou C.M., Parker J.S., Prat A. Clinical implementation of the intrinsic subtypes of breast cancer. Lancet Oncol. 2010;11:718–719. doi: 10.1016/S1470-2045(10)70176-5. [DOI] [PubMed] [Google Scholar]
- 10.Usary J., Zhao W., Darr D. Predicting drug responsiveness in human cancers using genetically engineered mice. Clin Cancer Res. 2013;19:4889–4899. doi: 10.1158/1078-0432.CCR-13-0522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Prat A., Perou C.M. Deconstructing the molecular portraits of breast cancer. Mol Oncol. 2011;5:5–23. doi: 10.1016/j.molonc.2010.11.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Fan C., Prat A., Parker J. Building prognostic models for breast cancer patients using clinical variables and hundreds of gene expression signatures. BMC Med Genomics. 2011;4:3. doi: 10.1186/1755-8794-4-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cheang M.C.U., Treaba D.O., Speers C.H. Immunohistochemical detection using the new rabbit monoclonal antibody SP1 of estrogen receptor in breast cancer is superior to mouse monoclonal antibody 1D5 in predicting survival. J Clin Oncol. 2006;24:5637–5644. doi: 10.1200/JCO.2005.05.4155. [DOI] [PubMed] [Google Scholar]
- 14.Prat A., Ellis M.J., Perou C.M. Practical implications of gene-expression-based assays for breast oncologists. Nat Rev Clin Oncol. 2012;9:48–57. doi: 10.1038/nrclinonc.2011.178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.McCullough A., Dell'Orto P., Reinholz M. Central pathology laboratory review of HER2 and ER in early breast cancer: an ALTTO trial [BIG 2-06/NCCTG N063D (Alliance)] ring study. Breast Cancer Res Treat. 2014;143:485–492. doi: 10.1007/s10549-013-2827-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Prat A., Bianchini G., Thomas M. Research-based PAM50 subtype predictor identifies higher responses and improved survival outcomes in HER2-positive breast cancer in the NOAH study. Clin Cancer Res. 2014;20:511–521. doi: 10.1158/1078-0432.CCR-13-0239. [DOI] [PubMed] [Google Scholar]
- 17.Prat A., Carey L., Adamo B. Molecular features and survival outcomes of the intrinsic subtypes within HER2-positive breast cancer. JNCI. 2014 doi: 10.1093/jnci/dju152. 106: dju152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Dowsett M., Sestak I., Lopez-Knowles E. Comparison of PAM50 risk of recurrence score with oncotype DX and IHC4 for predicting risk of distant recurrence after endocrine therapy. J Clin Oncol. 2013;31:2783–2790. doi: 10.1200/JCO.2012.46.1558. [DOI] [PubMed] [Google Scholar]
- 19.Gnant M., Filipits M., Greil R. Predicting distant recurrence in receptor-positive breast cancer patients with limited clinicopathological risk: using the PAM50 Risk of Recurrence score in 1478 postmenopausal patients of the ABCSG-8 trial treated with adjuvant endocrine therapy alone. Ann Oncol. 2014;25:339–345. doi: 10.1093/annonc/mdt494. [DOI] [PubMed] [Google Scholar]
- 20.Prat A., Parker J.S., Fan C. Concordance among gene expression-based predictors for ER-positive breast cancer treated with adjuvant tamoxifen. Ann Oncol. 2012;23:2866–2873. doi: 10.1093/annonc/mds080. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Cortazar P., Zhang L., Untch M. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet. 2014;384:164–172. doi: 10.1016/S0140-6736(13)62422-8. [DOI] [PubMed] [Google Scholar]
- 22.Esserman L.J., Berry D.A., DeMichele A. Pathologic complete response predicts recurrence-free survival more effectively by cancer subset: results from the I-SPY 1 TRIAL—CALGB 150007/150012, ACRIN 6657. J Clin Oncol. 2012;30:3242–3249. doi: 10.1200/JCO.2011.39.2779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Carey L.A., Dees E.C., Sawyer L. The triple negative paradox: primary tumor chemosensitivity of breast cancer subtypes. Clin Cancer Res. 2007;13:2329–2334. doi: 10.1158/1078-0432.CCR-06-1109. [DOI] [PubMed] [Google Scholar]
- 24.Sparano J.A., Paik S. Development of the 21-gene assay and its application in clinical practice and clinical trials. J Clin Oncol. 2008;26:721–728. doi: 10.1200/JCO.2007.15.1068. [DOI] [PubMed] [Google Scholar]
- 25.Ramsey S.D., Barlow W.E., Gonzalez-Angulo A.M. Integrating comparative effectiveness design elements and endpoints into a phase III, randomized clinical trial (SWOG S1007) evaluating oncotype DX-guided management for women with breast cancer involving lymph nodes. Contemporary Clinical Trials. 2013;34:1–9. doi: 10.1016/j.cct.2012.09.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Goldhirsch A., Winer E.P., Coates A.S. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol. 2013;24:2206–2223. doi: 10.1093/annonc/mdt303. [DOI] [PMC free article] [PubMed] [Google Scholar]