RIBBON-1: Randomized, Double-Blind, Placebo-Controlled, Phase III Trial of Chemotherapy With or Without Bevacizumab for First-Line Treatment of Human Epidermal Growth Factor Receptor 2–Negative, Locally Recurrent or Metastatic Breast Cancer

  1. Nicholas J. Robert⇑,
  2. Véronique Diéras,
  3. John Glaspy,
  4. Adam M. Brufsky,
  5. Igor Bondarenko,
  6. Oleg N. Lipatov,
  7. Edith A. Perez,
  8. Denise A. Yardley,
  9. Stephen Y.T. Chan,
  10. Xian Zhou,
  11. See-Chun Phan and
  12. Joyce O’Shaughnessy

+Author Affiliations

  1. From the Virginia Cancer Specialists, US Oncology, Fairfax, VA; University of California, Los Angeles Translational Oncology Research International, Los Angeles; Genentech, South San Francisco, CA; University of Pittsburgh, Pittsburgh, PA; Mayo Clinic, Jacksonville, FL; Sarah Cannon Cancer Center, Nashville, TN; Baylor-Sammons Cancer Center, Texas Oncology, US Oncology, Dallas, TX; Institut Curie, Paris, France; State Medical Academy, Dnepropetrovsk, Ukraine; Bashkirian Republican Clinical Oncology, Ufa, Russia; and Nottingham University Hospital, Nottingham, United Kingdom.
  1. Corresponding author: Nicholas J. Robert, MD, 8503 Arlington Blvd, Ste 400, Fairfax, VA 22031; e-mail: nicholas.robert@usoncology.com.
  1. Presented in part at the 45th Annual Meeting of the American Society for Clinical Oncology, May 29-June 2, 2009, Orlando, FL, and at the 15th Congress of the European Cancer Organisation and 34th Congress of the European Society for Medical Oncology Multidisciplinary Congress, September 20-24, 2009, Berlin, Germany.


Purpose This phase III study compared the efficacy and safety of bevacizumab (BV) when combined with several standard chemotherapy regimens versus those regimens alone for first-line treatment of patients with human epidermal growth factor receptor 2–negative metastatic breast cancer.

Patients and Methods Patients were randomly assigned in 2:1 ratio to chemotherapy plus BV or chemotherapy plus placebo. Before random assignment, investigators chose capecitabine (Cape; 2,000 mg/m2 for 14 days), taxane (Tax) -based (nab-paclitaxel 260 mg/m2, docetaxel 75 or 100 mg/m2), or anthracycline (Anthra) -based (doxorubicin or epirubicin combinations [doxorubicin/cyclophosphamide, epirubicin/cyclophosphamide, fluorouracil/epirubicin/cyclophosphamide, or fluorouracil/doxorubicin/cyclophosphamide]) chemotherapy administered every 3 weeks. BV or placebo was administered at 15 mg/kg every 3 weeks. The primary end point was progression-free survival (PFS). Secondary end points included overall survival (OS), 1-year survival rate, objective response rate, duration of objective response, and safety. Two independently powered cohorts defined by the choice of chemotherapy (Cape patients or pooled Tax/Anthra patients) were analyzed in parallel.

Results RIBBON-1 (Regimens in Bevacizumab for Breast Oncology) enrolled 1,237 patients (Cape cohort, n = 615; Tax/Anthra cohort, n = 622). Median PFS was longer for each BV combination (Cape cohort: increased from 5.7 months to 8.6 months; hazard ratio [HR], 0.69; 95% CI, 0.56 to 0.84; log-rank P < .001; and Tax/Anthra cohort: increased from 8.0 months to 9.2 months; HR, 0.64; 95% CI, 0.52 to 0.80; log-rank P < .001). No statistically significant differences in OS between the placebo- and BV-containing arms were observed. Safety was consistent with results of prior BV trials.

Conclusion The combination of BV with Cape, Tax, or Anthra improves clinical benefit in terms of increased PFS in first-line treatment of metastatic breast cancer, with a safety profile comparable to prior phase III studies.


Advances in the treatment of early-stage breast cancer have improved outcomes for patients. Despite these improvements, however, many women ultimately develop metastatic breast cancer (MBC). MBC is essentially an incurable disease, and the prognosis has changed little over the past decade; the majority of patients succumb to their disease within 2 years of diagnosis.1–3 New and novel treatments for patients with MBC are needed to improve control of disease and prevent symptoms, while minimizing toxicity.

Antiangiogenic therapies such as bevacizumab (BV), an antibody to vascular endothelial growth factor, have proven to be effective in improving outcomes in metastatic disease from several tumor types.4–6 For MBC, combinations of BV and established chemotherapy regimens potentially offer new, more efficacious treatments without the significant toxicity associated with combining multiple cytotoxic agents.7–10 Two prior randomized first-line phase III trials in MBC have shown the effectiveness of taxane (Tax) plus BV combinations in the initial treatment of human epidermal growth factor receptor 2 (HER2) –negative MBC. In the pivotal phase III E2100 trial, when compared with weekly paclitaxel alone, weekly paclitaxel plus BV significantly prolonged progression-free survival (PFS) and increased the objective response rate (ORR), while having limited impact on safety.11,12 Subsequently, the Avastin Plus Docetaxel (AVADO) trial, which investigated the docetaxel plus BV combination, demonstrated similar findings.13 A previous randomized phase III trial (AVF2119g) demonstrated that ORR was higher in patients treated with capecitabine (Cape) plus BV than Cape alone, suggesting that the Cape plus BV combination might improve PFS as first-line treatment for patients with MBC.14

RIBBON-1, an international, multicenter, phase III, randomized, placebo-controlled trial, was designed to evaluate the efficacy and safety of BV in combination with chemotherapy regimens, including non-Tax, such as Cape, used as initial chemotherapy-based treatment of MBC.


Study Population

Patients 18 years of age or older with locally recurrent breast cancer or MBC untreated with chemotherapy were eligible after providing informed consent. Key inclusion criteria included Eastern Cooperative Oncology Group performance status of 0 or 1 and recovery by day 0 from any grade ≥ 3 radiation toxicity. Patients with nonmeasurable disease were eligible. Exclusion criteria included the following: HER2-positive status (unless the patient had experienced progression on trastuzumab therapy or trastuzumab therapy was contraindicated or unavailable); prior adjuvant or neoadjuvant chemotherapy within 12 months; known CNS metastases; blood pressure greater than 150/100 mmHg; unstable angina; New York Heart Association class II or greater congestive heart failure; history of myocardial infarction within 6 months; history of stroke or transient ischemic attack within 6 months; clinically significant peripheral vascular disease; evidence of bleeding diathesis or coagulopathy; history of abdominal fistula, GI perforation, or intra-abdominal abscess within 6 months; history of anaphylactic reaction to monoclonal antibody therapy not controlled with premedication; serious nonhealing wound; inadequate organ function; locally recurrent disease amenable to resection with curative intent; history of other malignancies within 5 years; and pregnancy. If anthracycline (Anthra) was chosen as chemotherapy, patients were also required to have left ventricular ejection fraction ≥ 50% and no prior history of adjuvant Anthra treatment.

Study Design

RIBBON-1 (Regimens in Bevacizumab for Breast Oncology) was an international, multicenter, phase III, placebo-controlled trial. Before random assignment, investigators chose whether a patient would receive Cape or a Tax- or Anthra-based chemotherapy regimen (Fig 1). Patients were randomly assigned 2:1 to receive either chemotherapy plus BV or chemotherapy plus placebo, using a hierarchical dynamic random assignment algorithm via an interactive voice response system. Stratification factors included disease-free interval, which was defined as time from the later of either primary diagnosis or last dose of adjuvant chemotherapy (≤ or > 12 months); history of prior neoadjuvant or adjuvant chemotherapy (yes or no); number of metastatic sites (< or ≥ three sites); and choice of chemotherapy (Tax- or Anthra-based cohort only).

Fig 1.

Fig 1.

CONSORT flow diagram. Tax, taxane; Anthra, anthracycline; Cape, capecitabine; BV, bevacizumab.

BV was administered at 15 mg/kg intravenously (IV) every 3 weeks and was continued until progressive disease (PD), unacceptable toxicity, investigator decision, completion of 48 months of BV, or death, whichever came first. There were no reductions in BV/placebo dose. If an adverse event (AE) necessitated holding BV/placebo, the dose remained unchanged when treatment resumed.

Available chemotherapy choices were Cape 1,000 mg/m2 oral twice daily on days 1 to 14 of each 21-day cycle; docetaxel 75 to 100 mg/m2 IV or paclitaxel protein-bound particles (Abraxane; Abraxis BioScience, Bridgewater, NJ) 260 mg/m2 IV, every 3 weeks; or one of the following regimens: fluorouracil 500 mg/m2, epirubicin 90 to 100 mg/m2, and cyclophosphamide 500 mg/m2 IV; fluorouracil 500 mg/m2, doxorubicin 50 mg/m2, and cyclophosphamide 500 mg/m2 IV; doxorubicin 50 to 60 mg/m2 and cyclophosphamide 500 to 600 mg/m2 IV; or epirubicin 90 to 100 mg/m2 and cyclophosphamide 500 to 600 mg/m2 IV, every 3 weeks. Chemotherapy was continued until PD, unacceptable toxicity, investigator decision, or death, whichever came first, or for Anthra-based agents, eight cycles maximum. Doses could be modified at investigator discretion. If chemotherapy was discontinued before PD, patients could continue treatment with BV monotherapy.

Patients with documented PD in the blinded phase of the study could choose to receive open-label therapy with BV (5 mg/kg/wk dose equivalent) with a second-line chemotherapy agent, chosen by the investigator. No second-line Anthra chemotherapy was allowed.

Study Evaluations

All patients were assessed for tumor response every 9 weeks by clinical examination and radiographic assessments (computed tomography or magnetic resonance imaging); bone scans were repeated every 18 weeks or as clinically indicated. Disease status was assessed by investigators and an independent review committee (IRC; RadPharm, Princeton, NJ) according to Response Evaluation Criteria in Solid Tumors (RECIST) 1.0.15 Radiographs and blinded clinical tumor assessments were provided to the blinded IRC. Patients who discontinued treatment during the blinded-treatment phase for reasons other than PD were observed with tumor assessment every 9 weeks until documented PD. Patients who discontinued study treatment were observed until death or withdrawal of consent. Safety reporting included selected AEs previously associated with BV (grade ≥ 3 venous thromboembolic events, hypertension, bleeding, proteinuria, sensory neuropathy, wound dehiscence, neutropenia, and febrile neutropenia; grade ≥ 2 atrial thromboembolic events or left ventricular systolic dysfunction; and any grade GI perforation or reversible posterior leukoencephalopathy syndrome), all AEs leading to study drug discontinuation, and all serious AEs.

Study End Points

The primary efficacy end point was PFS based on investigator tumor assessments. Secondary efficacy end points included ORR, overall survival (OS), 1-year survival rate, duration of objective response, and PFS based on IRC assessment and safety.

Statistical Methods

For the Cape cohort, the study design required 600 patients (405 PFS events), which yielded an 80% power to detect an improvement in median PFS from 6 to 8 months (hazard ratio [HR], 0.75) at two-sided α = .05. For the Tax/Anthra cohort, 600 patients (415 PFS events) were required, which provided a 90% power to detect an improvement in median PFS from 7 to 10 months (HR, 0.7) at two-sided α = .05.

All enrolled patients were analyzed separately in the two cohorts according to treatment random assignment for efficacy end points. All reported data are based on investigator tumor assessments. PFS was defined as the time from random assignment to first PD or death from any cause. Data for patients who received nonprotocol-specified anticancer therapy before first PD were censored at the time of the last tumor assessment before initiation of nonprotocol-specified therapy. OS was defined as the time from random assignment until death. ORR was defined as the percentage of patients who achieved a complete or partial response confirmed ≥ 28 days after initial documentation of response. Analyses of PFS in each cohort were performed at the two-sided α = .05 level. Time-to-event data were compared between treatment arms using a stratified log-rank test. The Kaplan-Meier method was used to estimate duration of time-to-event data. The 95% CIs for median time to event were computed using the Brookmeyer-Crowley method. The HRs for time-to-event data were estimated using a stratified Cox regression model with separate nuisance hazards for each of the strata described in study design. There were 16 strata in the Tax- or Anthra-based cohort and eight strata in the Cape cohort. One-year survival rate was assessed between treatment arms using the normal approximation method. ORR in patients with measurable disease at baseline was compared using the stratified Mantel-Haenszel χ2 test. Random assignment stratification factors were included in all stratified analyses.

All treated patients were evaluated in safety analyses based on actual treatment received. AEs were cross-tabulated by treatment arm and chemotherapy class (Cape, Tax, or Anthra).


Patient Population

From December 2005 to August 2007, 1,237 patients were enrolled at 232 sites in 22 countries, 615 in the Cape cohort and 622 in the Tax/Anthra cohort (307 in the Tax subgroup and 315 in the Anthra subgroup; Fig 1). The cutoff date for analysis was July 31, 2008, resulting in a median follow-up time of 15.6 months for the Cape cohort and 19.2 months for the Tax/Anthra cohort.

Baseline Patient Demographics and Tumor Characteristics

In each of the Cape and Tax/Anthra cohorts, demographic and baseline characteristics for all randomly assigned patients were well balanced between the placebo and BV treatment arms; no baseline characteristics differed by more than 10% (Table 1). However, because investigators chose the chemotherapy for each patient before random assignment, there were differences between characteristics of the Cape and Tax/Anthra cohorts. Notably, patients receiving Anthra had less treatment for their early-stage breast cancer, including fewer surgeries, less adjuvant Tax history, and no prior Anthra as required per protocol (data not shown).

View this table:

Table 1.

Patient Demographics and Clinical Characteristics

Treatment Exposure

The average numbers of doses of study drug (placebo or BV) received in the Cape arms were 11.2 doses for Cape plus BV and 9.1 doses for Cape plus placebo. The average number of Cape treatment cycles was also higher in the Cape plus BV arm than in the Cape plus placebo arm (10.3 v 8.7 cycles, respectively). The dose-intensity of Cape and study drug, however, was comparable between the two treatment arms. The average numbers of doses of study drug received were 11.4 doses for Tax plus BV and 10.5 doses for Tax plus placebo in the Tax subgroups and 11.4 doses for Anthra plus BV and 10.3 doses for Anthra plus placebo. The dose-intensity of Tax or Anthra chemotherapy and study drug was similar across the two treatment arms. A total of 461 patients were enrolled onto the optional open-label, postprogression phase, including 48.6% of patients (196 of 403 patients) from the placebo arms and 32.4% of patients (265 of 817 patients) from the BV arms.


The combination of BV and Cape or Tax/Anthra-based chemotherapy as first-line therapy resulted in a statistically significant improvement in PFS. In the Cape cohort, median PFS increased from 5.7 months in the placebo arm to 8.6 months in the BV arm (HR, 0.69; 95% CI, 0.56 to 0.84; log-rank P < .001; Fig 2). In the Tax/Anthra cohort, median PFS increased from 8.0 months in the placebo arm to 9.2 months in the BV arm (HR, 0.64; 95% CI, 0.52 to 0.80; log-rank P < .001; Fig 2). For both cohorts, the majority of the events consisted of disease progressions (Cape, 92%; Tax/Anthra, 89%).

Fig 2.

Fig 2.

Kaplan-Meier estimates for progression-free survival (PFS) in the (A) capecitabine (Cape) cohort and (B) taxane (Tax)/anthracycline (Anthra) cohort. PL, placebo; BV, bevacizumab; HR, hazard ratio.

Analyses of PFS across prespecified key clinical subgroups in both the Cape and Tax/Anthra cohorts showed HRs that consistently favored BV-containing arms (Fig 3). In the Tax/Anthra cohort, a planned exploratory analysis of PFS in the Tax and Anthra subgroups also showed consistent improvements in PFS with the combination of BV and either chemotherapy class versus placebo (Tax: 9.2 v 8.2 months, respectively; HR, 0.75; log-rank P = .0547; Anthra: 9.2 v 7.9 months, respectively; HR, 0.55; log-rank P < .001). A sensitivity analysis of the PFS results using an IRC confirmed the primary findings (Cape cohort: 6.2 months with placebo v 9.8 months with BV; HR, 0.68; 95% CI, 0.54 to 0.86; log-rank P = .0011; Tax/Anthra cohort: 8.3 months with placebo v 10.7 months with BV; HR, 0.77; 95% CI, 0.60 to 0.99; log-rank P = .040).

Fig 3.

Fig 3.

Analysis of median progression-free survival for clinically relevant subgroups in the (A) capecitabine (Cape) cohort and (B) taxane (Tax)/anthracycline (Anthra) cohort. PL, placebo; BV, bevacizumab.

The ORR in patients with measurable disease was significantly higher in the BV-containing arm compared with the placebo arm for both chemotherapy cohorts (Cape cohort: 35.4% v 23.6%, respectively; P = .0097; Tax/Anthra cohort: 51.3% v 37.9%, respectively; P = .0054; Table 2). The majority of confirmed responses in both chemotherapy cohorts were partial responses. Among responders, median duration of objective response was longer in the BV-containing arms for both cohorts versus the placebo arms (Cape cohort: 9.2 months [95% CI, 8.5 to 10.4 months] v 7.2 months [95% CI, 5.1 to 9.3 months], respectively; Tax/Anthra cohort: 8.3 months [95% CI, 7.2 to 10.7 months] v 7.1 months [95% CI, 6.2 to 8.8 months], respectively).

View this table:

Table 2.

Objective Response Rate: Randomly Assigned Patients With Measurable Disease at Baseline

There was no statistically significant difference in OS or the 1-year survival rate between treatment arms for either cohort. For the Cape cohort, the estimated HR for OS, based on the stratified analysis, was 0.85 (95% CI, 0.63 to 1.14; P = .27). The 1-year survival rate was higher in the Cape plus BV arm compared with the Cape plus placebo arm (81.0% v 74.4%, respectively; P = .076). For the Tax/Anthra cohort, the stratified HR for OS was 1.03 (95% CI, 0.77 to 1.38; log-rank P = .83), and the 1-year survival rate was 80.7% in the BV-containing arm and 83.2% in placebo-containing arm (P = .44). As expected for first-line treatment of MBC, after discontinuing their assigned study treatment, the majority of patients received additional lines of systemic treatment with either chemotherapy, hormonal agents, or second-line combinations with BV that might have affected OS (Table 3).

View this table:

Table 3.

Subsequent Anticancer Therapy


For all three classes of chemotherapy (Cape, Tax, and Anthra), the incidence of selected AEs and grade 3 to 5 AEs was higher in the BV-containing arms than in the placebo-containing arms (Table 4). Hypertension and proteinuria were consistently increased in the BV-containing arms regardless of the chemotherapy. In the Tax-treated patients, bleeding events and febrile neutropenia occurred with a more than 5% greater incidence in the BV-containing arm. For both the Tax- and Anthra-based chemotherapy subgroups, the incidence of AEs leading to study drug discontinuation was higher in the BV arm than in the placebo arm; most events were considered to be related to study drug (Tax: 24.1% v 7.8%, respectively; Anthra: 14.3% v 4.0%, respectively). In the Cape cohort, there was no difference (both 11.9%). In both the Tax and Anthra subgroups, although the cumulative AEs leading to study drug discontinuation during the blinded phase had a higher incidence in the BV arm, no single AE or organ toxicity accounted for the increase (data not shown).

View this table:

Table 4.

Overview of Safety Results: Safety-Evaluable Patients

Across chemotherapy classes, there was a higher incidence of serious AEs in the BV-containing arm compared with the placebo-containing arm (Table 4). However, subgroup analyses of AEs and serious AEs by patient demographics between the treatment arms compared with the overall population did not reveal any new safety concerns for any of the chemotherapy classes (data not shown). For all patients treated with Anthra-based chemotherapy, left ventricular ejection fraction decreased slightly after treatment. The decline was similar between treatment arms (baseline vlowest postbaseline value: 65.7% v 59.6%, respectively, in the BV arm, compared with 64.7% v 60.8%, respectively, in the placebo arm).

The majority of deaths occurring during the study period were related to disease progression. The incidence of treatment-emergent deaths (within 30 days of last dose of study drug) and all deaths not related to PD was low and balanced across all treatment arms regardless of the chemotherapy (Table 4).


The results of this phase III trial in patients with MBC confirm and expand on the results from two earlier positive trials, E2100 and AVADO, which demonstrated that the combination of BV with first-line paclitaxel or docetaxel improved PFS and ORR. The current trial consisted of two independently powered cohorts and represents two studies evaluating the addition of BV to several first-line chemotherapy options for MBC. In one cohort, the combination of Cape plus BV showed a 31% relative reduction in risk of progression or death and 12% improvement in ORR. In the other cohort, the combination of BV with one of several Tax- or Anthra-based regimens demonstrated a 36% relative reduction in risk of progression or death and 14% improvement in ORR. Although the analyses from these three trials should not be compared with each other because of differences in study conduct, time period of enrollment, geographic distribution of enrollment, and potential differences in patients chosen to receive the respective chemotherapy backbone, the results from these three trials do show a marked consistency in the relative improvement in PFS with similar hazard reductions and overlapping CIs.

Improvements in PFS were seen with each chemotherapy class used. Although the Tax/Anthra cohort presents an analysis of aggregate data from patients receiving different chemotherapies with which BV was combined, exploratory analyses evaluating subgroups split by chemotherapy class received (Tax or Anthra) showed that the hazard reduction was similar regardless of the chemotherapy received. Although the earlier AVF2119g study failed to meet its primary objective of prolonged PFS with the combination of Cape and BV in patients with advanced MBC,6 the positive findings for the Cape cohort of RIBBON-1 suggest that the AVF2119g findings may have resulted from the patients in that study being more heterogeneous and having more advanced MBC rather than from the ineffectiveness of this treatment combination.

Despite the improved PFS seen in both the Cape and pooled Tax/Anthra chemotherapy cohorts of RIBBON-1, the combination with BV did not lead to a statistically significant difference in OS for these patients, which is not an unexpected finding for first-line MBC treatments.16,17 Of note, consistent with the primary objective of showing a PFS benefit, RIBBON-1 allowed cross over to BV in second-line treatment for all patients at first progression. Use of this option by up to half of the patients in the placebo arms in both chemotherapy cohorts may have compromised or detracted from the ability to demonstrate a difference in OS between the two study arms. A prior trial using a cross-over design showed PFS improvements but no OS difference.18 As expected for a first-line MBC population, most patients had extensive exposure to subsequent lines of systemic therapy. The multiple subsequent therapies may have also limited the ability to detect improvement in OS, as has been observed with other large studies in MBC.19

The combination of BV and chemotherapies besides Tax did not result in any significant changes to the AE profile of BV or increase fatal toxicities. As observed in prior studies with BV, patients treated with BV experienced increased rates of hypertension, proteinuria, and bleeding (which included epistaxis and upper and lower GI bleeding). Other toxicities seen in this trial appear to be linked to the chemotherapy received rather than to BV.

In summary, the results from RIBBON-1 demonstrate that the combination of BV with first-line therapy (Cape, Tax, or Anthra) provides clinical benefit in terms of increased PFS and increased ORR with a safety profile for BV consistent with prior phase III trials. These data, together with those from E2100 and AVADO, provide a clear rationale for adding BV to first-line cytotoxic therapy for patients with HER2-negative MBC.


Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a “U” are those for which no compensation was received; those relationships marked with a “C” were compensated. For a detailed description of the disclosure categories, or for more information about ASCO’s conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment or Leadership Position: Xian Zhou, Roche (C); See-Chun Phan, Genentech (C) Consultant or Advisory Role: Nicholas J. Robert, Roche (C); Véronique Diéras, Roche (C); Adam M. Brufsky, Roche (C); Denise A. Yardley, Roche (U); Joyce O’Shaughnessy, Roche (C) Stock Ownership: Xian Zhou, RocheHonoraria: Nicholas J. Robert, Roche; Véronique Diéras, Roche; Adam M. Brufsky, Roche; Joyce O’Shaughnessy, Roche Research Funding: Nicholas J. Robert, RocheExpert Testimony: None Other Remuneration: None


Conception and design: Nicholas J. Robert, Adam M. Brufsky, Stephen Y.T. Chan, Xian Zhou, See-Chun Phan

Administrative support: Nicholas J. Robert, Stephen Y.T. Chan

Provision of study materials or patients: Nicholas J. Robert, Véronique Diéras, John Glaspy, Adam M. Brufsky, Igor Bondarenko, Oleg N. Lipatov, Edith A. Perez, Denise A. Yardley, Stephen Y.T. Chan,Joyce O’Shaughnessy

Collection and assembly of data: Nicholas J. Robert, John Glaspy, Igor Bondarenko, Denise A. Yardley, Stephen Y.T. Chan, See-Chun Phan

Data analysis and interpretation: Nicholas J. Robert, Adam M. Brufsky, Stephen Y.T. Chan, Xian Zhou, See-Chun Phan, Joyce O’Shaughnessy

Manuscript writing: Nicholas J. Robert, Véronique Diéras, Adam M. Brufsky, Edith A. Perez, Stephen Y.T. Chan, Xian Zhou, See-Chun Phan

Final approval of manuscript: Nicholas J. Robert, Véronique Diéras, John Glaspy, Adam M. Brufsky, Igor Bondarenko, Oleg N. Lipatov, Edith A. Perez, Denise A. Yardley, Stephen Y.T. Chan, Xian Zhou, See-Chun Phan, Joyce O’Shaughnessy