Benefit from adjuvant taxanes and endocrine responsiveness in breast cancer

ARTICLE IN PRESS THE BREAST The Breast 16 (2007) S127–S131 www.elsevier.com/locate/breast

Review

Benefit from adjuvant taxanes and endocrine responsiveness in breast cancer Miguel Martina,, John Mackeyb, Charles Vogelc a

Servicio de Oncologı´a Me´dica, Hospital Universitario San Carlos, 28040 Madrid, Spain Department of Oncology, University of Alberta, Cross Cancer Institute, 11560 University Avenue, Edmonton, Canada T6G 1Z2 c Breast Cancer for Aptium Oncology, Boca Raton, FL, USA

b

Abstract Several phase III trials have been coincident in showing a benefit with adjuvant taxanes in node-positive breast cancer (BC). These trials provide level I evidence of the efficacy of these drugs. The absolute 5-year DFS gain obtained with the taxane-containing regimens in the positive studies ranged from 4% to 7%. However, the regimens including taxanes are more toxic than the conventional anthracyline-containing combinations. Therefore, the pre-treatment identification of the small proportion of patients who actually benefit from taxanes is of major importance. Several attempts have been made to identify the biological peculiarities of the BC patients who benefit most with taxanes, largely based on retrospective subset analyses. Hormone receptor (HR) status, one of the critical determinants of BC biology, had been suggested to be a factor modulating response to adjuvant chemotherapy in general and taxanes in particular. Unfortunately, none of the first-generation adjuvant taxane trials was designed to address the question of the differential efficacy of taxanes in the specific subgroups of HR positive and HR negative BC patients. Indirect (retrospective, unplanned) analysis suggest that the magnitude of benefit from taxanes is somewhat more in the HR negative subset, but the benefit also exists, and is clinically relevant, in patients with HR positive cancers. Therefore, the predictive value of HR status in the selection of patients for adjuvant taxane therapy is low. The predictive value of HER2 status in addition to HR status to select patients for the same purpose remains controversial. r 2007 Elsevier Ltd. All rights reserved. Keywords: Breast cancer; Adjuvant taxanes; Endocrine responsiveness

Introduction Several randomized clinical trials and a large metaanalysis have demonstrated that postsurgical chemotherapy significantly decreases the risk of recurrence and death in women with node-positive operable breast cancer (BC).1 CMF-like regimens were able to reduce the annual odds of recurrence and death of operable BC patients by 24% (+/ 3%) and 14% (+/ 4%), respectively. In the late 1970 and 1980s, anthracycline-containing combinations were tested in prospective randomized adjuvant trials and Corresponding author. Tel.: +34 91 3303546.

E-mail addresses: [email protected] (M. Martin), [email protected] (J. Mackey), [email protected] (C. Vogel). 0960-9776/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.breast.2007.09.001

found to be slightly but significantly superior to CMF-like regimens. The anthracycline regimens (5-fluorouracil, doxorubicin, cyclophosphamide (FAC), 5-fluorouracil, epirubicin, cyclophosphamide (FEC), doxorubicin, cyclophosphamide (AC) and others) resulted in an additional reduction in annual odds of relapse and death of 12% (+/4%) and 11% (+/5%), respectively, compared with CMF,2 and became the standard adjuvant treatment of node-positive operable BC. In the 1990s, the taxanes, docetaxel and paclitaxel, were incorporated to the standard armamentarium for metastatic BC and were soon tested in the adjuvant setting. The taxanes are partially non crossresistant with anthracyclines3,4 and, therefore, several regimens using taxanes and anthracyclines (either in combination or in sequence) were tested in the adjuvant setting. The results of nine pure adjuvant trials comparing

ARTICLE IN PRESS S128

M. Martin et al. / The Breast 16 (2007) S127–S131

anthracycline-containing combinations with taxanecontaining combinations (the so-called ‘‘first-generation taxane trials’’) have been reported.5–11 Six of them were clearly positive in terms of disease-free survival (DFS) in favour of the taxane arm.5–11 The ECOG 2197 trial did not find any significant DFS difference between AT (doxorubicin plus docetaxel) and AC (doxorubicin and cyclophosphamide) after a median follow-up of 53 months, but the number of DFS events was lower than expected.9 Finally, the two more recently reported trials (the TAXIT 216 and the BIG 2-98) found a difference in favour of the taxane arms of borderline significance for the primary endpoint (p values of 0.0576 and 0.051, respectively)12,13 and probably will become positive when more DFS events occur. Overall, these trials provide level I evidence of the efficacy of taxanes in the adjuvant treatment of nodepositive BC. The absolute 5-year DFS gain obtained with the taxane-containing regimens in the positive studies ranged from 4% to 7%. On the other side of the coin, the regimens including taxanes (particularly docetaxel) are clearly more toxic than the conventional anthracylinecontaining combinations, particularly in terms of myelosuppression, asthenia, and peripheral neuropathy. While myelosuppression and its complications can be life threatening,14 in aggregate the additional taxane side effects may produce a transient but significant deterioration of the quality of life while women undergo therapy.15 Because only 4–7% of patients obtain a DFS benefit with adjuvant taxanes (when compared to the less toxic anthracycline-containing combinations), the pre-treatment identification of that subset of patients would be of major importance for the individualization of therapy. Several attempts have been made to identify the biological peculiarities of the BC patients who benefit most with taxanes, largely based on retrospective subset analyses. Hormone receptor status, one of the critical determinants of BC biology, had been suggested to be a factor modulating response to adjuvant chemotherapy in general16 and taxanes in particular.17 We review here the relationship between the hormone receptor status of primary breast cancers and the benefit obtained with taxanes in the reported adjuvant clinical trials. Methodological problems None of the first-generation adjuvant taxane trials included among their endpoints a prospectively planned subgroup analysis of outcomes based on HR status. This fact constitutes a major methodological mistake, considering our current understanding of BC as a biologically heterogeneous disease. Despite the recognition in the 1970’s of the biological relevance of the presence of functional hormonal receptors (HR) in breast tumors, HR status had not been commonly used in the randomized phase III adjuvant chemotherapy trials as a stratification factor. Consequently, the vast majority of adjuvant randomized phase III chemotherapy trials included a

relatively unselected population of BC patients, and used the number of axillary lymph nodes as the primary variable in the stratification of patients; only one of the nine recently reported first-generation adjuvant taxane trials used hormone receptor status as a stratification factor.9 This oversight makes retrospective analysis of subsets based on hormonal status somewhat unreliable. Additionally, the definition of ‘hormone responsiveness’ and hormone receptor status is not a simple issue. HRs were first determined using ligand binding assays (LBA), a quantitative method, and the magnitude of benefit from endocrine therapy was shown to be related to the quantity of ER produced by the tumor. Because of reduced cost, and ease of specimen handling, and gradual reduction in the mean size of primary tumors, immunohistochemical (IHC) testing of hormone receptors largely replaced the quantitative LBA, and is now the standard method in clinical diagnostic laboratories. While IHC performs reasonably well in differentiating HR-positive versus HR-negative cases and minimizes the likelihood of falsenegative results, it is markedly less reliable for quantifying HR, because it is not an intrinsically quantitative method.18 Albain et al.16 shown that quantification of estrogen receptors (ER) by IHC (using Allred0 s score) may be of value in helping to identify patients with ER-positive tumors who may benefit from adjuvant FAC chemotherapy. Furthermore, there is a clinically relevant intercenter variability in the determination of HR. To achieve reliable HR results using ICH, assays should be performed in a central reference laboratory, assessed by experienced pathologists, and, with two notable exceptions, this has not been the case in the first-generation taxane trials. These three problems (lack of prospectively planned analysis of powered HR-based subsets, lack of stratification according to HR status, and lack of central determination and quantification of HR) limit the confidence one can place in the search for a relationship between HR status and benefit from adjuvant taxanes. Results from first generation clinical trials Table 1 summarizes the results of the nine reported firstgeneration adjuvant taxane trials, in terms of relative risk (RR) of recurrence (non-taxane/taxane arms) and the relationship with HR status. We note that only one trial (the ECOG 2197) stratified patients according to HR status and only the BCIRG 001and the GEICAM 9906 trials performed a central determination of HR. In general, the RR of recurrence with taxanes was numerically lower within the group of HR negative patients, but patients with HR positive tumors also obtained a clinically relevant reduction in recurrences with the taxane arms. The CALGB 9344 trial was the first-taxane study reporting results according to HR status. In an unplanned subset analysis, the addition of paclitaxel was more clearly beneficial in patients with hormone HR-negative or

ARTICLE IN PRESS M. Martin et al. / The Breast 16 (2007) S127–S131

S129

Table 1 Relative risk of recurrence according to hormonal receptor status in first generation taxane trials Study

Design

No pt

Stratification by hormone receptor status

RR of recurrence (non taxane/taxane arm), hormone receptor positive pt

RR of recurrence (non taxane/taxane arm), hormone receptor negative pt

CALGB 93445,17 NSABP B286 BCIRG 0017 PASC 018 GEICAM 990610 USO 973511 ECOG 21979 TAXIT 21612 BIG 2–9813

AC versus AC-paclitaxel AC versus AC-paclitaxel FAC versus TAC FEC versus FEC-docetaxel FEC versus FEC-weekly paclitaxel AC versus TC AT versus AC E-CMF versus E-T-CMF A-CMF versus AC-CMF versus AT-CMF versus A-T-CMF

3121 3060 1491 1999 1248 1016 2778 997 2887

No No* No No No No Yes No No

0.88 0.77 0.72 0.81 0.79 0.71 0.99*** NR** NR

0.75 0.90 0.69 0.79 0.76 0.64 1.21*** NR** NR

P values are not shown since these were unplanned and often underpowered subset analyses. RR: relative risk. A: doxorubicin. C: cyclophosphamide. E: epirubicin. T: docetaxel. F: 5-fluorouracil. NR: not reported. *Patients aged 450 with hormone receptor positive tumors and all patients who were 50 years or older regardless of hormone receptor status were given tamoxifen concomitantly with chemotherapy; tamoxifen use was a stratification factor. **The biostatistician of the group refused to perform subset analyses, since they were not planned in the protocol. ***RR of AT vs AC; RR41 favors the taxane arm.

unknown HR tumors than in those with HR-positive tumors (hazard ratio of 0.72 versus 0.91, respectively, in the first published report of the trial). However, the differences between the effects in these subsets were not significant after adjustment for multiple comparisons. Berry et al.17 analyzed the influence of HR status on the results of three consecutive CALGB phase III adjuvant trials, including the CALGB 9344 trial. The first one (CALGB 8541) compared three different dose–intensity regimens of CAF, the CALGB 9344 trial compared AC with AC followed by paclitaxel and the CALGB 9741 compared sequential versus concomitant administration and conventional every 3 weeks with dose-dense every 2 weeks administration of AC and paclitaxel. The authors found that the three trials independently and consistently showed the same finding: the benefit of the best chemotherapy arm was clearly more marked in the ER negative subset of patients, although this was particularly apparent in the dose-dense trial with relatively short follow-up. The reductions in the RR of recurrence in ER negative versus ER positive patients were 21% versus 9% (CALGB 8541), 25% versus 12% (CALGB 9344) and 24% versus 8% (CALGB 9741). Comparing the results of the worse arm of the first trial (the low-dose CAF regimen of the CALGB 8541) with those of the best arm of the last trial (the dose-dense arm of the CALGB 9741), ER negative patients had an improvement in 10-year DFS of 25.8% versus 10% for ER positive patients. However, this analysis raises several concerns. A retrospective, unplanned identification of these subgroups by means of multiple comparisons can lead to spurious treatment correlations. No stratification according to HR status was done in the trials, raising the possibility of imbalance between treatment arms. HR were not determined centrally or quantitatively, and did not use uniform methodology over the 14 years (1985–99) of

accrual spanned by these trials. The populations of patients treated with adjuvant hormonal therapy evolved during these trials to include premenopausal women, not all women with ER positive cancers received hormonal therapy, and the type and duration of hormonal therapy were not protocol specified in CALGB 9741. The follow-up of the three trials is very different and relatively short (about 5 years) in the CALGB 9741. This could be relevant since the DFS curves of HR positive patients from the prior trial (CALGB 9344) diverged in favour of the paclitaxel arm only after 5 years. While the authors conclude that the benefit of chemotherapy in general and taxanes in particular is mostly seen in ER negative patients, multiple methodologic concerns undermine the relevance of these conclusions and their applicability to current practice. The results of the NSABP B28 trial6 provide a demonstration of the weaknesses of unplanned retrospective analysis. With a design very similar to that of CALGB 9344 (4 cycles of AC versus 4 cycles of AC followed by 4 cycles of paclitaxel), the results were qualitatively different: HR positive patients had a bigger reduction of the RR of recurrence with paclitaxel than HR negative patients (0.77 and 0.90, respectively). The BCIRG 001 compared 6 cycles of the FAC regimen with the same number of cycles of TAC as adjuvant postsurgical therapy in node-positive BC patients.7 At a median follow-up of 55 months, the 5-year DFS was significantly better with TAC than with FAC (75% versus 68%, respectively, p ¼ 0.001). This trial is unique since the determination of HR was done centrally and was available in all patients. The relative reduction in risk of relapse was very similar in HR positive and HR negative patients (0.72 and 0.69, respectively). Retrospectively it has been determined that the rate of chemotherapy-induced amenorrhea

ARTICLE IN PRESS S130

M. Martin et al. / The Breast 16 (2007) S127–S131

in premenopausal TAC-treated patients was higher than FAC, adding yet another confounding variable. Unfortunately, HR status was not a stratification factor in this trial. The interim results of the Spanish trial GEICAM 9906 were presented at the 2005 San Antonio BC Symposium.11 Female patients with node-positive operable BC were randomized after surgery to receive 6 cycles of FE90C chemotherapy or 4 cycles of the same FEC regimen followed by 8 weekly administration of paclitaxel. After a median follow-up of 46 months, FEC followed by paclitaxel produced a statistically significant reduction in the risk of recurrence (hazard ratio 0.63, po0.0008). In this trial, a central determination of HR was available in 889 patients (71% of total). Among them, the RR of recurrence with paclitaxel was also very similar in HR positive and HR negative patients (0.79 and 0.75, respectively). The results of the French PACS 001 study were reported by Roche et al.8 at San Antonio BC Symposium in 2004. Patients with node-positive BC were randomized to receive 6 cycles of FE100C or 3 cycles of FE100C followed by 3 cycles of docetaxel. At a median follow-up of 60 months, the respective 5-year DFS rates were 78.4% (FECdocetaxel) and 73.2% (FEC), for a hazard ratio of 0.82 (p ¼ 0.012). The RR of recurrence for HR positive and HR negative patients with docetaxel were again very similar (0.81 and 0.79, respectively). However, an unspecified number of premenopausal patients with HR positive tumors did not receive tamoxifen after chemotherapy, potentially confounding this analysis. The US Oncology 9735 trial compared 4 cycles of AC to 4 cycles of TC (docetaxel plus cyclophosphamide) in patients with node-negative and node-positive disease. The projected 5-year DFS was superior with TC (86% versus 80%, HR for recurrence 0.67 p ¼ 0.015). The RR of recurrence of HR positive and negative patients were 0.71 and 0.64, respectively.11 The study ECOG 2197 compared doxorubicin plus cyclophosphamide (AC) and doxorubicin plus docetaxel (AT) in patients with node positive (1–3 axillary lymph nodes) and high-risk node-negative BC (65%).9 At a median follow-up of 59 months, there were many fewer recurrences than expected in the control arm and no statistically significant differences in terms of DFS or OS between the two treatments were found. Patients with HR negative tumors treated with AT had a 21% reduction in RR of recurrence, while patients with HR positive tumors had practically the same RR of recurrence with both treatments. These results are particularly interesting, since HR was a stratification factor in the trial, although no prospective and powered subset analysis were planned in the protocol. Unfortunately, no central determination of HR status was done. The follow-up of the study is still short for this low-risk population and a long-term benefit in HR positive patients cannot be ruled out. Finally, two studies presented in meetings did not communicate subgroup analysis.12,13 Of particular interest,

the biostatistician of the Italian trial TAXIT 216 refused to do unplanned, retrospective subset analysis based on HR status and this anecdote is illustrative of the low credibility that these analyses have among biostatisticians—(it seems that the clinicians are much more prone to accept them). Overall, this review suggests that patients with HR negative tumors may derive somewhat more benefit from adjuvant taxanes than the HR positive counterparts, although the benefit in the latter group seems to be clinically relevant as well. Therefore, allocating taxane treatment on the basis of HR status alone is not justified by the currently available data. Hormone receptor status, HER2 status and benefit from taxanes The combination of HER2 amplification/overexpression and HR status (the two most relevant molecular markers identified in BC so far) defines four groups of BC patients with different biological features. Hayes et al.19 have analyzed the correlation of HER2/HR status and outcome in the CALBG 9344 trial comparing AC to AC followed by paclitaxel. The study had a double randomization process, first to three different doses of doxorubicin (60, 75 and 90 mg/m2) and then to paclitaxel versus no paclitaxel. They hypothesized than HER2 status might predict benefit from higher doses of doxorubicin or from paclitaxel. Tumor blocks were collected from 2800 out of the 3121 patients included in the trial. Two sets of 750 patients each were randomly selected from the overall population and 643 and 679 tissue specimens were available in sets 1 and 2. The analyses were done first in set 1 and confirmed in the combination of sets 1 and 2. There was no interaction at all between doxorubicin dose and HER2 status. On the other hand, the interaction between HER2 and outcome with paclitaxel was statistically significant. Patients with HER2 positive tumors had a benefit in 5-year DFS with paclitaxel of 22%, versus only 2% in HER2 negative patients (p ¼ 0.0013). This was seen in set 1 and in the combination of sets 1 and 2. A further subset analysis combining HER2 and HR status showed that the benefit of adding paclitaxel to AC was greater in HER2 positive-HR negative patients (31% absolute increase in 5-year DFS). The subsets of HER2 negative-HR negative and HER2 positive-HR positive patients still obtained some benefit (8–9% absolute increase in 5-year DFS) while the remaining patients with HER2 negative-HR positive tumors did not benefit. The authors considered the latter analysis only hypothesis generating and will proceed to further validation in the remaining cases in CALGB 9344 and other paclitaxel trials. The main weakness of this analysis is that the two study arms of CALGB 9344 differed in two variables (paclitaxel or not and duration of chemotherapy—4 versus 8 cycles). The better outcome of HER2 positive patients in the paclitaxel arm could be due to either paclitaxel itself or to the longer duration of chemotherapy in the paclitaxel arm.

ARTICLE IN PRESS M. Martin et al. / The Breast 16 (2007) S127–S131

To confirm these findings, the GEICAM recently did a similar analysis of the GEICAM 9906 trial comparing 6 cycles of FEC to 4 cycles of FEC followed by 8 weekly administration of paclitaxel. In this trial, the duration of both treatments was similar and, overall, FEC followed by paclitaxel was superior to FEC in 5-year DFS. Tissue microarrays were available from 889 patients (71%). After a central determination of HR and HER2 (by FISH), the superiority of FEC-paclitaxel over FEC was present only in the HER2 negative subgroups (particularly in the HER2 negative-HR negative or triple negative subgroup) (unpublished data). These findings suggest that the duration of chemotherapy, rather that the use of paclitaxel itself, is the most likely explanation for the CALGB 9344 results. In a similar vein, the French PASC 01 trialists compared 6 cycles of FEC to 3 cycles of FEC followed by 3 cycles of docetaxel, and report the DFS benefit of the sequential FEC-docetaxel arm is mainly confined to the basal-like subtype of patients (HER2 negative, HR negative, EGFR1 positive, Pcadherin positive).20 The BCIRG group has recently performed a subset analysis based on HER2 and HR in the BCIRG 001 data set (unpublished results). All subgroups (HR+/her2+, HR+/her2, HR/her2+ and HR/her2) obtained a clinically relevant reduction of risk of recurrence with TAC. Conclusions Unfortunately, none of the first-generation adjuvant taxane trials was designed to address the question of the differential efficacy of taxanes in the specific subgroups of HR positive and HR negative BC patients. Indirect (retrospective, unplanned) analysis suggest that the magnitude of benefit from taxanes is somewhat more in the HR negative subset, but the benefit also exists, and is clinically relevant, in patients with HR positive cancers. Therefore, the predictive value of HR status in the selection of patients for adjuvant taxane therapy is low. The predictive value of HER2 status in addition to HR status to select patients for the same purpose remains controversial. References 1. Early Breast Cancer Trialists’ Collaborative Group. Polychemotherapy for early breast cancer: an overview of the randomized trials. Lancet 1998;352:930–42. 2. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomized trials. Lancet 2005;365:1687–717. 3. Valero V, Hohnes F, Walters RS, et al. Phase II trial of docetaxel: a new, highly effective anti-neoplastic agent in the management of patients with anthracycline-resistant metastatic breast cancer. J Clin Oncol 1995;13:2886–94. 4. Ravdin PM, Burris HA, Cook G, et al. Phase II trial of docetaxel in patients with anthracycline-resistant or anthracenedione-resistant breast cancer. J Clin Oncol 1995;13:2879–85.

S131

5. Henderson IC, Berry DA, Demetri GD, et al. Improved outcomes from adding sequential paclitaxel but not from escalating doxorubicin dose in an adjuvant chemotherapy regimen for patients with nodepositive primary breast cancer. J Clin Oncol 2003;21:976–83. 6. Mamounas EP, Bryant J, Lembersky BC, et al. Paclitaxel after doxorubicin plus cyclophosphamide as adjuvant chemotherapy for node-positive breast cancer: results from NSABP-B 28. J Clin Oncol 2005;23:3686–96. 7. Martin M, Pienkowski T, Mackey J, et al. Adjuvant docetaxel for node-positive breast cancer. N Engl J Med 2005;352:2302–13. 8. Roche H, Fumoleau P, Spielmann M, et al. Sequential adjuvant epirubicin-based and docetaxel chemotherapy for node-positive breast cancer patients: the FNCLCC PACS 01 trial. J Clin Oncol 2006; 24:5664–71. 9. Goldstein LJ, O’Neill A, Sparano JA, et al. E2197: phase III of AT vs. AC in the adjuvant treatment of node-positive and high-risk nodenegative breast cancer. J Clin Oncol 2005;23(Suppl 16S):6S. 10. Martin M, Rodriguez-Lescure A, Ruiz A, et al. Multicenter, randomized phase III study of adjuvant chemotherapy for node positive breast cancer comparing 6 cycles of FE90C versus 4 cycles of FE90C followed by 8 weekly paclitaxel administrations: interim efficacy analysis of GEICAM 9906 trial. Breast Cancer Res Treat 2005;94(Suppl 1):S20 (abstract 39). 11. Jones SE, Savin MA, Holmes FA, et al. Phase III trial comparing doxorubicin plus cyclophosphamide with docetaxel plus cyclophosphamide as adjuvant therapy for operable breast cancer. J Clin Oncol 2006;24:5381–7. 12. Bianco AR, de Matteis A, Manzione L, et al. Sequential epirubicindocetaxel-CMF as adjuvant therapy of early breast cancer: results of the Taxit216 multicenter phase III trial. J Clin Oncol 2006;24 (Suppl 18S):8S. 13. Crown JP, Francis P, Di Leo A, et al. Docetaxel given either concurrently or sequentially to anthracycline-based adjuvant therapy for patients with node-positive breast cancer: comparison with nontaxane combination chemotherapy. First results of the BIG 2-98 trial at 5 years median follow-up. J Clin Oncol 2006;24(Suppl 18S):7S. 14. Brain EGC, Bachelot T, Serin D, et al. Life-threatening sepsis associated with adjuvant doxorubicin plus docetaxel for intermediaterisk breast cancer. J Am Med Assoc 2005;293:2367–71. 15. Martin M, Lluch A, Segui MA, et al. Toxicity and health-related quality of life in breast cancer patients receiving adjuvant docetaxel, doxorubicin, cyclophosphamide (TAC) or 5-fluorouracil, doxorubicin and cyclophosphamide (FAC): impact of adding primary prophylactic granulocyte-colony stimulating factor to the TAC regimen. Ann Oncol 2006;17:1205–12. 16. Albain K, Barlow W, O’Malley F, et al. Concurrent (CAFT) versus sequential (CAF-T) chemohormonal therapy (cyclophosphamide, doxorubicin, 5-fluorouracil, tamoxifen) versus T alone for postmenopausal, node-positive, estrogen and/or progesterone receptor-positive breast cancer; mature outcomes an new biological correlates on phase III intergroup trial 0100 (SWOG-8814). Breast cancer Res Treat 2004;88(Suppl 1) abstract 37. 17. Berry DA, Cirrincione C, Craig Henderson I, et al. Estrogen-receptor status and outcomes of modern chemotherapy for patients with nodepositive breast cancer. J Am Med Assoc 2006;295:1658–67. 18. Schnitt SJ. Estrogen receptor testing of breast cancer in current clinical practice: what’s the question? J Clin Oncol 2006;24: 1797–9. 19. Hayes DF, Thor A, Dressler L, et al. HER2 predicts benefit from adjuvant paclitaxel after AC in node-positive breast cancer: CALGB 9344. J Clin Oncol 2006;24(Suppl 18s):5s (abstract 510). 20. Jacquemier J, Pernault-Lorca F, Mnif H, et al. Identification of a basal-like subtype and comparative effect of epirubicin-based chemotherapy and sequential epirubicin followed by docetaxel chemotherapy in the PACS 01 breast cancer trial. 33 markers studied on tissue-microarrays (TMA). J Clin Oncol 2006;24(Suppl 18s):5s (abstract 509).