Annals of Oncology 29: 178–185, 2018 doi:10.1093/annonc/mdx690 Published online 24 October 2017

ORIGINAL ARTICLE

Ten-year results of intense dose-dense chemotherapy show superior survival compared with a conventional schedule in high-risk primary breast cancer: final results of AGO phase III iddEPC trial

V. Mo¨bus1*, C. Jackisch2,H.J.Lu¨ck3, A. du Bois4, C. Thomssen5, W. Kuhn6, U. Nitz7, A. Schneeweiss8, J. Huober9, N. Harbeck10, G. von Minckwitz11, I. B. Runnebaum12, A. Hinke13, G. E. Konecny14, M. Untch15 & C. Kurbacher16, on behalf of the AGO Breast Study Group (AGO-B)†

1Department of Gynecology and Obstetrics, Klinikum Ho¨chst, Frankfurt; 2Department of Gynecology and Obstetrics, Sana Klinikum Offenbach GmbH, ; 3Gynecologic Oncology Practice, Hannover; 4Department of Gynecology & Gynecologic Oncology, Klinikum -Mitte, Essen; 5Department of Gynecology, Martin-Luther University -Wittenberg, Halle (Saale); 6Department of Gynecology and Obstetrics, University of , Bonn; 7Breast Center Niederrhein, Evangelic Hospital Bethesda, Mo¨nchengladbach; 8National Centre of Tumor Diseases, University of , Heidelberg; 9Department of Gynecology and Obstetrics, University of , Ulm; 10Department of Gynecology and Obstetrics, University of , Munich; 11German Breast Group, Neu- Isenburg; 12Department of Gynecology, University of , Jena; 13WiSP Research Institute, Langenfeld, ; 14David Geffen School of Medicine, University of California Los Angeles, Los Angeles, USA; 15Department of Gynecology and Obstetrics, Helios Klinikum -Buch, Berlin; 16Medical Center, Bonn-Friedensplatz, Bonn, Germany

*Correspondence to: Prof. Dr Volker J. Mo¨bus, Department of Gynecology and Obstetrics, Klinikum Frankfurt Ho¨chst, Academic Hospital of the Goethe University Frankfurt, Gotenstrasse 6-8, D-65929 Frankfurt, Germany. Tel: þ49-69-300-599-45; Fax: þ49-69-300-599-46; E-mail: [email protected] †The investigators (physicians and staff) at the participating institutions (3 recruited patients) on behalf of the AGO Breast Study Group are given in the Appendix. Note: This study was previously presented in part at the 35th Annual San Antonio Breast Cancer Symposium (SABCS), 4–8 December 2012.

Background: Primary breast cancer (BC) patients with extensive axillary lymph-node involvement have a limited prognosis. The Arbeitsgemeinschaft fuer Gynaekologische Onkologie (AGO) trial compared intense dose-dense (idd) adjuvant chemotherapy with conventionally scheduled chemotherapy in high-risk BC patients. Here we report the final, 10-year follow- up analysis. Patients and methods: Enrolment took place between December 1998 and April 2003. A total of 1284 patients with 4 or more involved axillary lymph nodes were randomly assigned to receive 3 courses each of idd sequential epirubicin, paclitaxel and cyclophosphamide (iddEPC) q2w or standard epirubicin/cyclophosphamide followed by paclitaxel (EC ! P) q3w. Event- free survival (EFS) was the primary end point. Results: A total of 658 patients were assigned to receive iddEPC and 626 patients were assigned to receive EC ! P. The median duration of follow-up was 122 months. EFS was 47% (95% CI 43% to 52%) in the standard group and 56% (95% CI 52% to 60%) in the iddEPC group [hazard ratio (HR) 0.74, 95% CI 0.63–0.87; log-rank P ¼ 0.00014, one-sided]. This benefit was independent of menopausal, hormone receptor or HER2 status. Ten-year overall survival (OS) was 59% (95% CI 55% to 63%) for patients in the standard group and 69% (95% CI 65% to 73%) for patients in the iddEPC group (HR ¼ 0.72, 95% CI 0.60–0.87; log-rank P ¼ 0.0007, two-sided). Nine versus two cases of secondary myeloid leukemia/myelodysplastic syndrome were observed in the iddEPC and the EC ! P arm, respectively. Conclusion: The previously reported OS benefit of iddEPC in comparison to conventionally dosed EC ! P has been further increased and achieved an absolute difference of 10% after 10 years of follow-up. Key words: high-risk early breast cancer, dose-dense, intense dose-dense

VC The Author 2017. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For Permissions, please email: [email protected]. Annals of Oncology Original article Introduction Totally, 1154 assessable patients had to be recruited and followed for a median period of 5 years in order to achieve 80% power to identify an im- For adjuvant treatment of primary breast cancer (BC), anthracy- provement from 60% to 67% in EFS after 5 years with 5% type I error cline and taxane-based chemotherapy regimens are considered (one-sided). Some over-recruitment was allowed to increase the statis- standard of care [1]. In addition, dose-dense (dd) [2–4], intense tical validity of the prospectively planned subgroup comparisons in the dd (idd) regimens [5] or tailored dd regimens [6] have shown su- strata with 4–9 and 10 positive ALN (558 and 598 patients were perior clinical outcomes when compared with conventionally required to have 80% power for an anticipated improvement from 60% to 70% and 45% to 55%, respectively). dosed chemotherapy, however, clinical follow-up has been rela- Time to event distributions were estimated using the Kaplan–Meier tively short in these studies [median 5 (range 3–7) years] and method [8], and compared between treatment and prognostic groups long-term clinical outcome data are missing [2, 4, 5, 6]. using the log-rank test [9]. For multivariable analysis, a Cox proportional Hematological toxicity is more pronounced with dd regimens, hazard model was applied. Either Fisher’s exact test or an exact version of but grade 3/4 neutropenia and febrile neutropenia can safely be the Cochran–Armitage trend test was used to compare toxicity scores. All prevented by primary prophylaxis with granulocyte-colony stim- tests except for the primary hypothesis were two-sided and of explorative ulating factors (G-CSF). Primary prophylaxis even leads to less nature. This includes all subgroup analyses and multivariable models. treatment-related deaths in comparison to conventionally dosed chemotherapy [2, 7]. Here, we report the 10-year follow-up results for idd versus Results conventionally dosed chemotherapy. Patient characteristics

Patients and methods A total of 1284 patients were recruited between November 1998 and April 2003 in 165 centers in Germany (iddEPC N ¼ 658, Patients EC ! P N ¼ 626). Six hundred forty-one (97%) and 611 (98%) patients were assessable for the primary end point (supplemen Main inclusion criteria were: women with histologically confirmed pri- tary Figure S1, available at Annals of Oncology online). Three pa- mary BC stage II and IIIA with 4positive axillary lymph nodes (ALN); tients in the iddEPC arm were considered non-eligible (cardiac age between 18 and 65 years, M0 status, and R0 resection of the primary arrhythmia, liver metastasis present before randomization, severe tumor and axilla with a minimum of 10 ALN removed. Additional eligi- bility criteria have been previously published [5]. wound healing complication). All three patients were excluded The ethics committees of all participating institutions approved the from the analysis of the primary end point. study. All patients provided written informed consent. The treatment arms were well balanced with respect to demo- graphic and prognostic factors (Table 1). In both study arms, the Randomization and masking median number of positive ALN was eight and 42% of the pa- tients had 10 involved ALN. Randomization was stratified according to institution, number of posi- tive ALN (4–9 versus 10) and menopausal status. Computer-generated randomization lists were used for each stratum and were balanced by Treatment block randomization with randomly varying block sizes of 2, 4 and 6. After central review of eligibility, random assignment was done by fax. All planned cycles of chemotherapy were administered to 91% of Patients and treating physicians could not be masked to allocation be- patients in the conventional arm, and to 84% in the idd arm. The cause of the nature of the interventions, and investigators were not predominant causes for treatment discontinuation in the iddEPC masked since the outcomes (relapse, death) were objectives. arm were toxicity (65% of withdrawals) and patients’ preference (20%). Differences in dose reduction and treatment delays have Treatment been reported [5]. Relapse during treatment occurred more fre- Idd treatment consisted of sequential administration of epirubicin (E) quently under conventional dosing (11 versus 3 patients). (150 mg/m2 i.v. as bolus infusion) q2w for three cycles, followed by pacli- taxel (P) (225 mg/m2 i.v.) q2w for three cycles, followed by cyclophos- phamide (C) (2500 mg/m2 i.v.) q2w for three cycles. By definition, the Hematological toxicity iddEPC regimen was dd and used a higher total dose per cycle. Patients Hematological toxicity was more pronounced in the idd arm and received filgrastim subcutaneously days 3–10 of each cycle. Women in the incidence was highest during treatment with cyclophospha- the iddEPC arm were additionally randomized to receive or not epoetin alfa during the entire chemotherapy period. The standard treatment con- mide and modest under paclitaxel. The difference between arms sisted of 4 cycles of EC (90/600 mg/m2 iv.) q3w followed by 4 cycles of was significant (P < 0.0001) with respect to all three peripheral paclitaxel (175 mg/m2) q3w (EC!P), without primary growth factor blood cell lines. Overall, at least one episode of febrile neutro- support (Figure 1). Post-study treatment recommendations were exten- penia was recorded in 5% of the patients (iddEPC arm 7% versus sively published [5]. EC ! P 2%, P < 0.0001). Only 6 patients (1%) in the EC ! P arm compared with 127 patients (20%) in the iddEPC arm Statistical aspects received red blood cell transfusions (P < 0.0001, Fisher’s exact test). The sub-randomization plus/minus Epoetin-alpha had no The primary end point was event-free survival (EFS) defined as locore- gional or distant relapse, contralateral invasive BC, second primary can- effect on EFS or OS in the iddEPC arm [10]. Detailed results will cer occurrence or death for any reason. Secondary end points included be published separately. No treatment-related deaths were overall survival (OS), toxicity and quality of life (QoL). observed.

Volume 29 | Issue 1 | 2018 doi:10.1093/annonc/mdx690 | 179 Original article Annals of Oncology Epirubicin (E) Paclitaxel (P) Cyclophosphamide (C) 150 mg/m2 225 mg/m2 2500 mg/m2 q2w × 3 q2w × 3 q2w × 3

Arm 1 + TAM

G-CSF (Filgrastim) ± Epoetin-a R

EC 90/600 mg/m2 Paclitaxel 175 mg/m2 q3w × 4 q3w × 4

Arm 2 + TAM

Figure 1. Trial design.

Table 1. Baseline characteristics of patients

Intense dose-dense Conventional chemotherapy sequential chemotherapy (n 5 643) (n 5 612)

Age, median (range) 51 (28–67) 51 (21–72) Premenopausal 306 (48%) 292 (48%) Postmenopausal 336 (52%) 320 (52%) Performance status (ECOG) >0 113 (18%) 116 (19%) Breast-conserving surgery 274 (43%) 308 (50%) Extent of axillary surgery Level I/II 505 (78%) 474 (77%) Level III 138 (22%) 138 (23%) Tumor stage pT1 181 (28%) 199 (33%) pT2 362 (56%) 330 (54%) pT3 96 (15%) 80 (13%) pT4 4 (1%) 3 (0%) Lymph nodes, median Examined 20 20 Positive 8 8 Number of positive nodes 4–9 376 (58%) 358 (58%) 10 267 (42%) 254 (42%) Grading G1 15 (2%) 11 (2%) G2 310 (49%) 293 (48%) G3 310 (49%) 302 (50%) Receptor status ER and/or PR þ 499 (78%) 467 (77%) ER and PR 139 (22%) 139 (23%) HER2 status Positive 162 (25%) 150 (25%) Negative 371 (58%) 346 (57%) Not done 107 (17%) 115 (19%) Endocrine treatment (ER/PRþ), postmenopausal patients only Tamoxifen 5 years 93 (38%) 99 (42%) AI after 2–3 years of Tamoxifen 46 (19%) 42 (17%) AI 5 years 20 (8%) 21 (8.5%) Extended adjuvant therapy 69 (28%) 74 (30%) (10 years TAM only or 5 years TAM ! AI sequential) No hormonal treatment started 10 (4%) 12 (4%) Radiotherapy Whole breast RT 277 (49%) 308 (54%) Whole breast plus boost 204 (36%) 226 (39%) Internal mammary nodes 210 (37%) 201 (35%) Supra- and infraclavicular 467 (81%) 483 (84%) Axillary field 305 (53%) 294 (51%) Postmastectomy chest wall 311 (54%) 279 (49%)

180 | Mo¨bus et al. Volume 29 | Issue 1 | 2018 Annals of Oncology Original article Table 2. Non-hematological toxicity in the iddEPC and EC fi P treatment arm

Arms iddEPC (n 5 588) EC fi P(n 5 550)

NCI-grade 0 1 2 3 4 n.e. 0 1 2 3 4 n.e.

Nauseaa 78 (12%) 232 (37%) 262 (42%) 47 (7%) – 10 (2%) 91 (15%) 267 (45%) 193 (32%) 33 (6%) – 10 (2%) Vomitingb 217 (34%) 157 (25%) 194 (31%) 31 (5%) 5 (1%) 25 (4%) 248 (42%) 184 (31%) 121 (20%) 17 (3%) 5 (1%) 19 (3%) Diarrheab 357 (57%) 111 (18%) 105 (17%) 19 (3%) 2 (0%) 35 (6%) 408 (69%) 85 (14%) 62 (10%) 5 (1%) – 34 (6%) Stomatitisb 172 (27%) 227 (36%) 148 (24%) 51 (8%) 5 (1%) 26 (4%) 317 (53%) 177 (30%) 62 (10%) 10 (2%) – 28 (5%) Sensoryb 132 (21%) 261 (41%) 179 (28%) 41 (7%) – 16 (3%) 195 (33%) 236 (40%) 115 (19%) 26 (4%) – 22 (4%) Paina 108 (17%) 178 (28%) 227 (36%) 96 (15%) 8 (1%) 12 (2%) 151 (25%) 149 (25%) 202 (34%) 74 (12%) 1 (0%) 17 (3%) Consciousnessa 426 (68%) 119 (19%) 41 (7%) 2 (0%) 1 (0%) 40 (6%) 449 (76%) 84 (14%) 21 (4%) 2 (0%) 1 (0%) 37 (6%) Feverb 348 (55%) 89 (14%) 143 (23%) 8 (1%) 2 (0%) 39 (6%) 456 (77%) 55 (9%) 48 (8%) 1 (0%) 1 (0%) 33 (6%) Infectionb 279 (44%) 113 (18%) 120 (19%) 79 (13%) 1 (0%) 37 (6%) 344 (58%) 106 (18%) 93 (16%) 28 (5%) 1 (0%) 22 (4%) Skinb 298 (47%) 192 (31%) 93 (15%) 11 (2%) 6 (1%) 29 (5%) 373 (63%) 134 (23%) 48 (8%) 5 (1%) – 34 (6%) Alopecia 17 (3%) 12 (2%) 594 (94%) – – 6 (1%) 10 (2%) 11 (2%) 561 (94%) – – 12 (2%)

aCochran–Armitage test: P ¼ 0.0005 (nausea), P ¼ 0.0009 (pain), P ¼ 0.0013 (consciousness). bCochran–Armitage test: P < 0.0001. n.e., not evaluable.

Non-hematological toxicity tumor stage, number of positive ALN, tumor grade and hormone receptor status were prognostically relevant for EFS. All param- Non-hematological toxicity occurred significantly more often in eters identified by univariate analyses retained independent prog- the iddEPC arm, but the overall incidence of grade 3/4 toxicities nostic relevance in a multivariable Cox model. was low and clinically acceptable (Table 2). The impact of treatment arm on EFS in the major subgroups is After 10 years of follow-up, no grade 3 congestive heart failure shown in Figure 3. Chemotherapy intensification had a more was observed. One patient died due to cardiac failure after sur- pronounced effect in the subgroup with 10 infiltrated ALN gery for aortic aneurysm. After a median follow-up of 10 years, (P ¼ 0.00039). A trend toward greater efficacy for iddEPC in pa- 0.7% of the patients in the idd arm compared with 0.2% in the tients with less favorable tumor biology was also observed. conventionally scheduled group reported persisting grade 1 neur- Benefit from iddEPC was independent of HER2 and estrogen re- opathy. No persisting grade 2/3 neurotoxicity was observed. ceptor (ER) expression. Secondary neoplasms have been observed in 46 patients (iddEPC N ¼ 27, EC ! P N ¼ 19). Acute myeloid leukemia or myelodysplastic syndrome (MDS) were reported in 11 patients during 10-year follow-up (0.9%) (iddEPC N ¼ 9, EC ! P N ¼ 2; Discussion P ¼ 0.065). Dose-dense adjuvant chemotherapy has become one of the stand- Efficacy of chemotherapy ard treatment options for high-risk BC patients [11]. Several studies with different designs have independently shown that dd After a median follow-up period of 10 years (range 0.1– chemotherapy results in a superior EFS [2–6] and OS [2–5]in 158 months), 604 events and 446 deaths have been recorded. EFS comparison to conventionally dosed chemotherapy. Meta- (282 versus 322 events) and OS (201 versus 245 events) both sig- analyses [12, 13] also persistently confirm the superiority of dd nificantly favored the iddEPC arm. At 10 years, EFS rates for the chemotherapy in comparison with standard chemotherapy. iddEPC versus EC ! P were 56% (95% CI 52% to 60%) versus Long-term follow-up data are necessary to robustly validate the 47% (95% CI 43% to 52%), respectively [HR ¼ 0.74, 95% CI clinical relevance of dd/dose intensified chemotherapy and to (0.63–0.87); log rank test P ¼ 0.00014)], after adjusting for help establish their role for the treatment of patients with inter- major prognostic baseline parameters in a multivariable model mediate/high risk primary BC as a standard of care. Here, we pro- (Figure 2A). Ten-year survival rates were 69% (95% CI 65% to vide the first long-term clinical outcome data of an intense dd 73%) in the iddEPC compared with 59% (95% CI 55% to 63%) chemotherapy regimen. in the EC ! P arm [HR ¼ 0.72 (95% CI 0.60–0.87); log rank test: At 10 years follow-up, EFS and OS continue to be significantly P ¼ 0.0007] (Figure 2B). In patients with >10 positive ALN OS superior for the idd approach, with an absolute difference in OS rates were 62% versus 48% (HR ¼ 0.66; P ¼ 0.0016), whereas in of 10%. As the 5-year OS data demonstrated an absolute survival patients with four to nine positive ALN OS rates were 74% versus difference of 5% (82% versus 77%) in favor of iddEPC [5], the 66% (HR ¼ 0.77; P ¼ 0.061), respectively. survival difference has actually increased with longer clinical follow-up. Given the fact that 42% of our patients had 10 posi- Prognostic and predictive factors. The results of univariate ana- tive ALN (median 8 positive ALN), a 10-year survival rate of 69% lyses are listed in Table 3. At univariate analyses treatment arm, in the iddEPC arm, represents, to the best of our knowledge, the

Volume 29 | Issue 1 | 2018 doi:10.1093/annonc/mdx690 | 181 Original article Annals of Oncology

A 1.0

0.8 iddEPC 10-year EFS-rate:56%

0.6

0.4 Logrank test: p=0.00014, one-sided (proportion) HR: 0.74, 95%-CI, 0.63 to 0.87

Event - free survival Event EC→P 10-year EFS-rate: 47% 0.2 iddEPC: n=641, 282 events, median = NA → 0 EC P: n=611, 322 events, median = 95.1

Time (months) 01020 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 Patients at risk iddEPC n=641 622 567 525 491 465 448 423 399 383 354 289 207 123 54 9 0 EC→P n=611 579 508 452 413 375 360 339 313 292 263 215 153 104 43 12 0

B 1.0

iddEPC: 10-year overall survival: 69% 0.8

0.6 logrank test: p=0.0007, two-sided HR: 0.72; 95%-CI, 0.60 to 0.87 0.4 (prportion) → Overall survival Overall EC P: 10-year overall survival: 59%

0.2 iddEPC: n=641, 201 events, median = NA → 0 EC P: n=611, 322 events, median = NA

Time (months) 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 iddEPC n=641 636 619 592 569 537 518 492 475 450 426 347 256 154 68 14 0 EC→P n=611 602 574 551 513 479 456 430 407 386 348 280 195 130 50 14 0 Figure 2. Event-free (A) and overall survival (B) by treatment arm.

Table 3. Univariate and multivariable analysis of event-free survival

Univariate Multivariable (reduced model) Multivariable (reduced model) P-value P-value HR (95% CI)

Treatment arm, intense dose-dense versus conventional 0.00014a <0.001 0.71 (0.60–0.84) Menopausal status 0.31 – – pT 2/3 versus pT 1 <0.001 <0.001 1.43 (1.18–1.72) 10 versus 4–9 positive nodes <0.001 <0.001 1.59 (1.35–1.89) Grading, G3 versus G1/2 <0.001 0.017 1.23 (1.04–1.47) Hormone receptor statusb <0.001 0.0056 0.76 (0.62–0.92) HER2 0.33 – – Baseline hemoglobin, < versus 12 g/dl 0.17 – – CEA, normal range versus elevatedc <0.001 – – CA 15-3, normal range versus elevatedc 0.062 – –

aOne-sided. bAt least one positive receptor status versus both negative. cBefore the first cycle of chemotherapy. –: not in model (in case of tumor markers due to high frequency of missing values).

182 | Mo¨bus et al. Volume 29 | Issue 1 | 2018 Annals of Oncology Original article Event-free survival Subgroup Hazard ratio, 95% CI Premenopausal Postmenopausal

HR pos. HR neg.

Her2 pos. Her2 neg.

Triple-negative Non-triple-negative

4-9 lymph nodes ≥10 lymph nodes

Total

0.5 0.7 1 1.5 2 Favours iddEPC Favours EC -> P Figure 3. Hazard plot of treatment effect in major subgroups. highest survival rate that has been reported for such a high-risk [14] used a lower than usually recommended dose of epirubicin group of patients to date. Importantly, this survival rate was in both arms corresponding to only 50% of the dose used in the achieved without upfront therapy with aromatase inhibitors, standard Canadian CEF regimen [16]. The ‘dd’ arms of the UK bisphosphonates or trastuzumab. Moreover, the superiority of TACT2 trial [15] and the Canadian MA21 [17] trial were a 50% the iddEPC regimen was independent of menopausal, estrogen split of dd and standard dosed chemotherapy, thus hindering an receptor and HER2 status. explicit interpretation of their negative results. However, not all dd trials have reported improved outcomes. In this context it is also interesting to note that in retrospective Especially differences in the number of involved nodes (4, 5, 7, subset analyses only the Italian GIM-2 [4], the tailored dd 15) and apparent differences in terms of number, total applied Panther [6] and the German AGO-EPC [5] trial, but not the dose and schedule of chemotherapy (2, 4, 7, 14, 15) may explain CALGB C9741 trial demonstrated superiority of dd chemother- these inconsistent results. apy also in the subset of ER-positive patients. Especially in the Superiority of dd/idd chemotherapy regimens has only been AGO-EPC trial the high risk situation of the patients may effect- observed in studies that have accrued high-risk primary BC pa- ively override the interaction between dose density and HER2 tients. The CALGB C9741 [2], GIM-2 [4] and AGO-EPC [5] and/or ER status. were positive trials which recruited patients with a median num- Nine cases (1.4%) of secondary leukemia/MDS were reported ber of three, five and up to eight positive ALN, respectively. In in the iddEPC arm versus two cases (0.3%) in the conventional contrast, the UK TACT2 trial [15] and the NSABP B-38 [7] arm. The percentage of 1.4% in the iddEPC arm is consistent accrued patients with a median number of one or two positive with the published data of the 10-year follow-up analysis of the nodes and failed to demonstrate superiority of dd chemotherapy. Canadian MA5 trial [16]. Praga et al. [19] reviewed 19 adjuvant Eighty-seven % of patients in the UK TACT2 trial were N0 or N1, trials with epirubicin and cyclophosphamide and patients had an none of these lower risk patients have been recruited in the AGO- 8-year cumulative probability of secondary leukemia/MDS rang- iddEPC trial. ing between 0.37% and 4.97%. Independently of the low-risk profile we cannot exclude that In summary, the iddEPC regimen leads to a relative risk reduc- the TAC regimen used in the NSABP B-38 trial may be more ef- tion for relapse and death of 26% and 28%, respectively. The fectiveincomparisontotheAC/ECfollowedbypaclitaxelregi- iddEPC regimen encompasses the concepts of dose density, dose men used in the CALGB C9741, GIM-2 and AGO-EPC trials. escalation and sequential therapy. This schedule may increase ef- When these trials were recruiting patients, four cycles of AC or ficacy both when tumor growth follows Gompertzian kinetics EC followed by four cycles of paclitaxel were a modern stand- requiring higher density and when partially resistant clones are ard of care in lymph node positive disease. The recently pub- present requiring higher doses. It is the first trial, which reports lished 10-year follow-up of the E1199 trial showed superiority 10-year survival and toxicity data for a dd chemotherapy regi- between P1 versus P3 only for the triple negative subgroup, but men. The absolute survival benefit of 10% appears to be quite re- no longer for the largest subgroup of ERþ/HER2 patients markable. idd chemotherapy with epirubicin, paclitaxel and [18].Thedifferenceinefficacymaybeoverestimatedforthis cyclophosphamide is an effective and safe regimen. Consistent subgroup. with other reports on dd trials [2, 4, 6] we also observed no treat- Other differences in the trial design may help to explain some ment-related death despite a significantly higher hematologic of the contradictory findings. For example, the GONO-MIG trial toxicity in the dd arms,

Volume 29 | Issue 1 | 2018 doi:10.1093/annonc/mdx690 | 183 Original article Annals of Oncology Independently of known molecular BC subtypes, dd chemo- from Hoffmann-La Roche and AMGEN. CK has reported hono- therapy remains an important treatment option for high-risk BC raria from AMGEN, Teva, Hoffmann-La Roche, Hexal, patients. Future research will need to focus on identifying bio- Novartis, Celgene and Axios. He received research funding from markers that predict treatment response and will need to evaluate Novartis, MSD Sharp & Dohme, Teva, AMGEN and Hexal. All combination therapies with novel targeted agents to further im- remaining authors have declared no conflicts of interest. prove treatment outcomes for patients diagnosed with high-risk primary BC. References 1. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Acknowledgements Comparisons between different polychemotherapy regimens for early breast cancer: meta-analyses of long-term outcome among 100.000 We wish to thank Mrs Christina Weißmu¨ller for her help and women in 123 randomised trials. Lancet 2012; 379: 432–444. assistance in collecting the follow-up data and preparing the 2. Citron ML, Berry DA, Cirrincione C et al. Randomized trial of dose- manuscript, and Mrs Heidi Eustermann for performing statis- dense versus conventionally scheduled and sequential versus concurrent tical analyses. We thank all the patients and their families who combination chemotherapy as postoperative adjuvant treatment of participated in this trial. node-positive primary breast cancer: first report of intergroup trial C9741/Cancer and Leukemia Group B Trial 9741. J Clin Oncol 2003; 21: 1431–1439. 3. Sparano JA, Wang M, Martino S et al. Weekly paclitaxel in the adjuvant Funding treatment of breast cancer. N Engl J Med 2008; 358(16): 1663–1671. 4. Del Mastro L, De Placido S, Bruzzi P et al. Flourouracil and dose-dense This work was supported by Bristol-Myers-Squibb; Amgen; chemotherapy in adjuvant treatment of patients with early-stage breast Johnson & Johnson and Pharmacia. A drug supply of epoetin cancer: an open-label, 22 factorial, randomized phase 3 trial. Lancet alfa was provided by Johnson & Johnson. There is no grant 2015; 385(9980): 1863–1872. number applicable. 5. Moebus VJ, Jackisch C, Lueck HJ et al. Intense dose-dense chemotherapy with epirubicin, paclitaxel and cyclophosphamide compared with con- ventionally scheduled chemotherapy in high-risk primary breast cancer: mature results of an AGO phase III study. J Clin Oncol 2010; 28: Disclosure 2874–2880. 6. Foukakis T, von Minckwitz G, Bengtsson NO et al. Effect of tailored VM has reported honoraria from AMGEN, Celgene and dose-dense chemotherapy vs standard 3-weekly adjuvant chemotherapy Hoffmann-La Roche and is a consultant to Celgene and on recurrence-free survival among women with high-risk early breast MyeloTherapeutics. CJ has reported honoraria and accommo- cancer. A randomized clinical trial. JAMA 2016; 316(18): 1888–1896. dations from AMGEN. HL is a consultant to Hoffmann-La 7. Swain S, Tang G, Geyer C et al. Definitive results of a phase III adjuvant Roche, Celgene, AstraZeneca, Novartis and MSD and he has dis- trial comparing three chemotherapy regimens in women with operable, closed speaker’s bureau from Hoffmann-La Roche, Novartis, node-positive breast cancer: the NSABP B-38 trial. J Clin Oncol 2013; 31: AstraZeneca and Celgene. AdB is a consultant to AstraZeneca, 3197–3204. 8. Kaplan EL, Meier P. Nonparametric estimation from incomplete obser- Pharmamar, Roche/Genentech, Mundipharma and Pfizer. CT vations. J Am Stat Assoc1958; 53(282): 457–481. has reported honoraria from AMGEN, AstraZeneca, Celgene, 9. Peto R, Peto J. Asymptotically efficient rank invariant test procedures Janssen, Novartis, Pfizer, Hoffmann-La Roche and TEVA. He is (with discussion). J R Stat Soc A 1972; 135(2): 185–207. a consultant to AMGEN, AstraZeneca, Celgene, Janssen, 10. Moebus V, Schneeweiss A, du Bois A et al. Ten year follow-up analysis of Novartis, Pfizer, Hoffmann-La Roche and TEVA. UN has re- intense dose-dense adjuvant ETC (epirubicin, paclitaxel and cyclophos- ported honoraria from Celgene, Hoffmann-La Roche, AMGEN, phamide) confirms superior DFS and OS benefit in comparison to con- TEVA, Bayer, Genomic Health, Agenda Nanostring, Pfizer and ventional dosed chemotherapy in high-risk breast cancer patients with 4 positive lymph nodes. Cancer Res 2012; 72(Suppl; abstr S3-4): 24s. Novartis. She is a consultant to Hoffmann-La Roche and 11. NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines) Celgene and she received research funding from Genomic Breast Cancer. Version 3, 2014; http://www.nccn.org/patients (17 June Health, TEVA, Bayer, Roche and AMGEN. She has disclosed ac- 2017, date last accessed). commodations from Hoffmann-La Roche and Pfizer. AS 12. Bonilla L, Ben-Aharon I, Vidal L et al. Dose-dense chemotherapy in non- received honoraria, research funding and accommodations metastatic breast cancer: a systematic review and meta-analysis of random- from Hoffmann-La Roche and Celgene. NH has received hono- ized controlled trials. J Natl Cancer Inst 2010; 102(24): 1845–1854. 13. Petrelli F, Cabiddu M, Coinu A et al. Adjuvant dose-dense chemotherapy raria from Hoffmann-La Roche, Novartis, Celgene, NanoString in breast cancer: a systematic review and meta-analysis of randomized Technologies, AMGEN, Pfizer and Genomic health. She is a trials. Breast Cancer Res Treat 2015; 151(2): 251–259. consultant to Roche/Genentech, Novartis, Celgene, 14. Venturini M, Del Mastro L, Aitini E et al. Dose-dense adjuvant chemo- AstraZeneca, Sandoz, Pfizer and Hexal. She received research therapy in early breast cancer patients: results from a randomized trial. funding from Roche/Genentech, Novartis, Boehringer J Natl Cancer Inst 2005; 97(23): 1724–1733. Ingelheim and Pfizer. GvM has received research funding from 15. Cameron D, Morden J, Canney P et al. Accelerated versus standard epi- AMGEN, Pfizer, GlaxoSmithKline, Sanofi, Novartis, Celgene, rubicin followed by cyclophosphamide, methotrexate, and fluorouracil or capecitabine as adjuvant therapy for breast cancer in the randomised Teva, Boehringer Ingelheim, AstraZeneca, Myriad Genetics and UK TACT2 trial (CRUK/05/19): a multicentre, phase 3, open-label, rand- Hoffmann-La Roche. IR has disclosed honoraria from Celgene omised, controlled trial. Lancet Oncol 2017; 18(7): 929–945. and AstraZeneca. He is a consultant to Martin and he received 16. Levine M, Pritchard K, Bramwell V et al. Randomized trial comparing accommodations from Celgene. AH has reported honoraria cyclophosphamide, epirubicin, and fluorouracil with cyclophosphamide,

184 | Mo¨bus et al. Volume 29 | Issue 1 | 2018 Annals of Oncology Original article methotrexate, and fluorouracil in premenopausal women with node- Hospital, Halle/Saale); B. Bru¨ ckner (University Hospital, Bonn); R. positive breast cancer: update of National Cancer Institute of Canada Gros (Klinikum Idar-Oberstein GmbH); H. Stehle (Marienhospital, Clinical Trials Group Trial MA5. J Clin Oncol 2005; 23: 5166–5170. ); K. Lobodasch (DRK Krankenhaus, -Rabenstein); 17. Burnell M, Levine MN, Chapman JA et al. Cyclophosphamide, epirubi- T. Dinkelacker (Helfenstein Klinik, Geislingen), E. Klo¨pper cin, and fluorouracil versus dose-dense epirubicin and cyclophospha- (Allgemeines Krankenhaus, Celle); C. Karg (Kreiskrankenhaus mide followed by paclitaxel versus doxorubicin and cyclophosphamide Waiblingen); M. Stibora (Marienhospital ); W. Weise followed by paclitaxel in node-positive or high-risk node-negative breast (University Hospital, ); F. Melchert (University Hospital, cancer. J Clin Oncol 2010; 28(1): 77–82. ); S. Mohrmann (University Hospital, Du¨ sseldorf), C. Klatt 18. Sparano JA, Zhao F, Martino S et al. Long-term follow-up of the E1199 (Sta¨dtisches Klinikum, Delmenhorst); M. Schmidt (University phase III trial evaluating the role of taxane and schedule in operable Hospital, ); M. Glados (Praxis fu¨r Ha¨matologie und breast cancer. J Clin Oncol 2015; 33(21): 2353–2360. Internistische Onkologie, Coesfeld); H. Mickan (Sta¨dtischesKlinikum, 19. Praga C, Bergh J, Bliss J et al. Risk of acute myeloid leukemia and myelo- Esslingen); A. Pollmanns (Ev. Krankenhaus, Mu¨ lheim/Ruhr); T. Beck dysplastic syndrome in trials of adjuvant epirubicin for early breast can- (Sta¨dtisches Klinikum, Rosenheim); W. Schro¨der (University cer: correlation with doses of epirubicin and cyclophosphamide. J Clin Oncol 2005; 23: 4179–4191. Hospital, ); H. Klingemann (Kreiskrankenhaus Goslar); G. Ko¨hler (University Hospital, Greifswald); W. Freier (Onkologische Schwerpunktpraxis, ); S.-T. Graßhoff (Klinikum Quedlinburg); U. So¨ling (Praxis fu¨r Ha¨matologie und Onkologie, Appendix ); E. Keil (Helios Klinikum, Berlin-Buch); C. Ho¨ß (Kreisklinik The authors thank the investigators (physicians and staff) at the Ebersberg); U. Kullmer (Asklepios Klinik, Lich); A. Niesel participating institutions (3 recruited patients) on behalf of the (Kreiskrankenhaus Peine); H. Bodenstein (Klinikum Minden); G. AGO study group: Bartzke (Kreiskrankenhaus Rottweil); C.M. Schlotter (Klinikum S. Kahlert (University Hospital, Mu¨ nchen-Großhadern); N. Hauser Ibbenbu¨ ren); W. Wiest (St. Vincenz- und Elisabeth- Hospital, Mainz); (University Hospital, Ulm); F. Marme´ (University Hospital, W. Ernst (St. Elisabeth Krankenhaus, Mayen); O. Brudler/B. Heinrich Heidelberg); L. Kiesel (University Hospital, Mu¨ nster); D. Wallwiener (Ha¨mato- onkologische Gemeinschaftspraxis, ); J. (University Hospital, Tu¨ bingen); B. Rautenberg (Bethesda Hackmann (Marien-Hospital, ), W. Niedner Krankenhaus, ); R. Hofmann (University Hospital, (Bethanienkrankenhaus Moers); G. Hoffmann (St. Josefs Hospital, Hannover); J. Barinoff (Horst-Schmidt Kliniken, ); E. Wiesbaden); W. Knapp (St. Josefs Stift, Delmenhorst); G. Dresemann Stickeler (University Hospital, Freiburg); H. Eidtmann (University (Franz-Hospital, Du¨ lmen); C. Karl (St. Antonius Hospital, Eschweiler); Hospital, ); V. Mu¨ ller (University Hospital, ); G. Emons F.-J. Klemm (Sta¨dtisches Klinikum Friedrichshafen); G.C. Schliesser (University Hospital, Go¨ttingen); P. Mallmann (University Hospital, (Praxis fu¨r Ha¨matologie und Onkologie, Gießen); M. Butterwegge Ko¨ln); R. Hackenberg (Sta¨dtisches Klinikum, ); C. Hielscher (Marienhospital Osnabru¨ ck); H. Meden (Diakoniekrankenhaus (Klinikum Stralsund); G. Ra¨ber (DRK Kliniken Ko¨penick, Berlin); H. Rotenburg, Wu¨ mme); M. Kirschbaum (Caritasklinik St. Theresia, Sommer (University Hospital; Mu¨ nchen-Maistrasse); P. Schmidt- Saarbru¨ cken); T. Schwenzer (Klinikzentrum Mitte, ); H.-J. Rhode (Allgemeines Krankenhaus Barmbek, Hamburg); B. Jahns Voigt (Westpfalz-Klinikum, Kaiserslautern); U. Weiß (Klinikum ¨ (Paracelsus Klinik, Henstedt-Ulzburg); A. von der Assen Lippe-Lemgo);G.Onder (St. Johannis Krankenhaus, Landstuhl); W. (Franziskushospital Harderberg, Georgsmarienhu¨ tte); S. Dietterle Langer (Kreiskrankenhaus Aurich); M. Glaubitz (St. Franziskus (Carl-Thiem-Klinikum, ); M. Kaufmann (University Hospital, Hospital, Ahlen); M. Beha (Klinikum St. Marien, Amberg); A. Coumbos Frankfurt/M.); C. Strunk (St. Barbara Klinik, ); R. Goebel (Ev. (University Hospital Benjamin Franklin, Berlin); Y. Ko (University Krankenhaus, ); S. Paepke (Technical University Hospital, Hospital, Bonn); J.-U. Deuker (Vinzenz-Krankenhaus, Hannover); O. Mu¨ nchen); A. Paulenz (Klinikum Ernst von Bergmann, ); A. Pru¨ mmer (Klinikum Kempten); U. Wagner (University Hospital Junker-Stein (Sta¨dtisches Klinikum, Frankfurt/M.); U. Rhein (SRH ); A. Kohlstedt (St. Elisabeth Krankenhaus, -Bad); P. Zentralklinikum, Suhl); K. Bremer (Augusta-Kranken-Anstalt, Schreiber (Klinikum Salzgitter); V. Jovanovic (Sta¨dtisches Klinikum ); W. Bauer (Schwarzwald-Baar-Klinikum, Villingen- ); K. Drzewiecki (Harz-Klinikum Wernigerode); H. Guba Schwenningen); P. Hohlweg-Majert (Klinikum Nordstadt, Hannover); (Sta¨dtisches Klinikum Zwickau); H. Seipt (Sta¨dtisches Klinikum D. Kramer (Krankenhaus Siloah, ); H.-G. Meerpohl (St. Friedrichstadt, ); M. Stauch (Praxis fu¨r Ha¨matologie, Vincentius Kliniken, ); K. Diedrich (University Hospital, Kronach); C. Maintz (Praxis fu¨r Ha¨matologie, Wu¨ rselen); B. Morenz Lu¨ beck); P. Dall (Sta¨dtischesKlinikum,Lu¨ neburg); H. Ko¨lbl (University (Sta¨dtisches Klinikum Magdeburg).

Volume 29 | Issue 1 | 2018 doi:10.1093/annonc/mdx690 | 185