Randomized Phase III Postoperative Trial of Platinum-Based Chemotherapy Versus Capecitabine in Patients With Residual Triple-Negative Breast Cancer Following Neoadjuvant Chemotherapy: ECOG-ACRIN EA1131

Abstract

PURPOSE

Patients with triple-negative breast cancer (TNBC) and residual invasive disease (RD) after completion of neoadjuvant chemotherapy (NAC) have a high-risk for recurrence, which is reduced by adjuvant capecitabine. Preclinical models support the use of platinum agents in the TNBC basal subtype. The EA1131 trial hypothesized that invasive disease-free survival (iDFS) would not be inferior but improved in patients with basal subtype TNBC treated with adjuvant platinum compared with capecitabine.

PATIENTS AND METHODS

Patients with clinical stage II or III TNBC with ≥ 1 cm RD in the breast post-NAC were randomly assigned to receive platinum (carboplatin or cisplatin) once every 3 weeks for four cycles or capecitabine 14 out of 21 days every 3 weeks for six cycles. TNBC subtype (basal v nonbasal) was determined by PAM50 in the residual disease. A noninferiority design with superiority alternative was chosen, assuming a 4-year iDFS of 67% with capecitabine.

RESULTS

Four hundred ten of planned 775 participants were randomly assigned to platinum or capecitabine between 2015 and 2021. After median follow-up of 20 months and 120 iDFS events (61% of full information) in the 308 (78%) patients with basal subtype TNBC, the 3-year iDFS for platinum was 42% (95% CI, 30 to 53) versus 49% (95% CI, 39 to 59) for capecitabine. Grade 3 and 4 toxicities were more common with platinum agents. The Data and Safety Monitoring Committee recommended stopping the trial as it was unlikely that further follow-up would show noninferiority or superiority of platinum.

CONCLUSION

Platinum agents do not improve outcomes in patients with basal subtype TNBC RD post-NAC and are associated with more severe toxicity when compared with capecitabine. Participants had a lower than expected 3-year iDFS regardless of study treatment, highlighting the need for better therapies in this high-risk population.

INTRODUCTION

Triple-negative breast cancers (TNBC) lack expression of estrogen or progesterone receptor and human epidermal growth factor receptor 2 (HER2), and account for approximately 15% of all invasive breast cancers.¹ Their prevalence is also higher among young and Black women.² TNBC carries a higher likelihood of distant recurrence and death compared with other breast cancer subtypes.³ Despite its aggressive biology, approximately 40% of patients have highly chemotherapy-sensitive tumors, achieving pathologic complete response (pCR; lack of residual invasive cancer in breast and axilla) at surgery following anthracycline- and taxane-containing neoadjuvant chemotherapy (NAC).^4-6 For an individual patient, pCR is associated with improved survival.^5,7 Conversely, patients with evidence of residual invasive disease (RD) at surgery following NAC are more likely to have a recurrence and die from breast cancer,^5,8 with reported 3-year event-free survival rates of 60%-70%.^9,10 Adjuvant capecitabine for six to eight cycles in patients with residual TNBC post-NAC improved invasive disease-free survival (iDFS) and overall survival (OS) in the CREATE-X⁹ trial and became a standard of care.

CONTEXT

• Key Objective

• Patients with triple-negative breast cancer (TNBC) and residual invasive disease (RD) after completion of neoadjuvant chemotherapy (NAC) have a high risk for recurrence, which is reduced by adjuvant capecitabine. The EA1131 trial hypothesized that invasive disease-free survival would not be inferior but improved in patients with basal subtype TNBC treated with adjuvant platinum compared with capecitabine.

• Knowledge Generated

• EA1131 results show that platinum agents do not improve outcomes in patients with TNBC RD post-NAC, regardless of intrinsic subtype, and are associated with more severe toxicity when compared with capecitabine. Participants had a lower than expected 3-year invasive disease-free survival regardless of study treatment, highlighting the need for better therapies in this high-risk population.

• Relevance

• These findings should discourage the adjuvant use of platinum agents in patients with residual TNBC after NAC outside of a trial. For now, the use of adjuvant platinum agents in unselected patients with TNBC remains investigational, and capecitabine remains the standard therapy.

Most TNBC are basal subtype based on the PAM50 gene expression signature,¹ and this has been associated with worse recurrence-free survival (RFS) and OS in patients with RD after NAC.¹¹ Basal subtype TNBC shares many phenotypical and genomic similarities with BRCA1-mutated breast cancers¹²: The intrinsic genomic instability present in some TNBC cells (most notably basal subtype) and BRCA1/2-mutant breast cancers¹³ results from deficient DNA repair mechanisms¹⁴ and may provide greater sensitivity to platinum agents. Indeed, preclinical models show that basal subtype TNBC is particularly sensitive to cisplatin.¹⁵ Additionally, in p53-deficient tumors (about 90% of TNBC), p63 is coordinately expressed with TAp73, antagonizing p73 transcriptional activity.¹⁶ Platinum agents are potent inhibitors of p63 expression,^17,18 resulting in released antagonism of p73 activity, and permitting the expression of proapoptotic Bcl-2 family members, leading to apoptosis. Several clinical trials^19,20 have reported an increase in pCR with the addition of platinum to anthracycline- and taxane-containing NAC in patients with TNBC, but toxicity is increased and the impact on long-term outcomes remains unclear.

The EA1131 trial (ClinicalTrials.gov identifier: NCT02445391) was originally designed to test the hypothesis that adjuvant platinum chemotherapy would improve iDFS compared with observation in patients with clinical stage II-III TNBC who had basal subtype RD in the breast after NAC. Once CREATE-X⁹ results became available, shortly after trial activation, EA1131 was amended to replace observation with capecitabine as the control arm. Five patients randomly assigned before this amendment are excluded from the efficacy analysis (Fig 1).

FIG 1.

CONSORT diagram. All patients (308 basal subtype and 88 nonbasal subtype) enrolled after Protocol Amendment 3, June 2016 (which incorporated capecitabine as the control arm instead of observation based on CREATE-X trial⁷ results), were included in efficacy analyses. A total of 391 patients started protocol therapy and 382 of them submitted AE data and were included in toxicity analysis. Patients enrolled to platinum and capecitabine arms but not treated (n = 21, 16 basal subtype) were followed for recurrence and survival and were included in the intention-to-treat efficacy analysis. Patients enrolled to observation arm were excluded from the report. AE, adverse event; PD, disease progression.

PATIENTS AND METHODS

Patient Eligibility

Eligible patients had clinical stage II or III TNBC (defined as estrogen or progesterone receptor 1+ or 2+ immunohistochemistry intensity in ≤ 10% cells and HER2-negative by immunohistochemistry and/or in situ hybridization) at diagnosis. At least one full cycle of taxane with or without anthracycline-containing NAC had to be completed; neoadjuvant investigational agents were allowed if therapy completed at least 30 days before random assignment. Patients with synchronous bilateral breast cancers or multifocal breast cancers were eligible if all tumors were TNBC, and at least one fulfilled all eligibility criteria. Patients must have completed definitive resection of the primary breast tumor within 24 weeks of random assignment. Radiation therapy (RT), when recommended, could have been given before or after study treatment. Patients must have had a contiguous focus of residual invasive cancer in the breast measuring ≥ 1 cm in diameter, with > 20% of nucleated cells being malignant, regardless of nodal disease. The RD was submitted for PAM50 analysis (Veracyte, South San Francisco, CA) for TNBC subtype determination (basal v nonbasal). All patients had to have adequate renal, hepatic, and hematologic function. No chemotherapy or investigational agents after surgery and no capecitabine or platinum chemotherapy before surgery was allowed. Enrollment in other clinical trials was allowed after Protocol (online only) therapy completed.

The ECOG-ACRIN Cancer Research Group coordinated the study in collaboration with the Alliance for Clinical Trials in Oncology, NRG Oncology, and the SWOG Cancer Research Network. Local institutional review boards approved the Protocol, and patients provided written informed consent before screening. This trial was funded by the US National Cancer Institute.

Treatment Plan

Patients were randomly assigned to receive capecitabine 1,000 mg/m² twice daily, days 1-14, every 3 weeks, for a total of six cycles, or a platinum agent (treating physician choice of cisplatin 75 mg/m² or carboplatin AUC 6 on day 1), once every 3 weeks, for a total of four cycles. The capecitabine dose and number of cycles chosen reflect common practice in the United States, as higher doses are usually not well tolerated among non-Asians.^21-25 Because of recent exposure to NAC and RT, we were concerned that more than four cycles of platinum chemotherapy would not be tolerated because of cumulative myelosuppression and neuropathy. Clinical or laboratory assessments were completed in time for each treatment cycle; routine toxicity-related dose modifications were mandated.

RT before or after study treatment completion was required for all patients after breast-conservation surgery. Postmastectomy RT was required for patients with primary tumors > 5 cm or ≥ 4 axillary lymph nodes involved and allowed at the discretion of the treating physician for all other patients.

Statistical Considerations

The primary end point was iDFS (time from random assignment to the earliest disease recurrence [locoregional or distant], invasive contralateral cancer, second primary cancer, or death) in patients with basal subtype TNBC. Recruitment of 775 patients was planned; assignments were made using permuted blocks within strata with dynamic balancing within institutions and a 1:1 random assignment ratio to platinum or capecitabine arms. Stratification included clinical stage at diagnosis (II or III), residual primary tumor (ypT) diameter after NAC (1-3 cm or > 3 cm), platinum agent (cisplatin or carboplatin), neoadjuvant anthracycline exposure (yes or no), and use of RT (yes or no).

We used a noninferiority design with superiority alternative (hybrid design²⁶) to test the hypothesis that in patients with basal subtype TNBC, platinum would improve iDFS compared with capecitabine. The 4-year iDFS rate in the capecitabine arm was assumed to be 67% (based on CREATE-X⁹). The primary analysis for the primary objective tested the null hypothesis of inferiority of platinum, defined as a hazard ratio (HR) for platinum versus capecitabine of 1.154 (noninferiority margin), corresponding to a 4-year iDFS rate of ≤ 63% in the platinum arm. After enrolling 775 patients (562 expected to have basal subtype TNBC) over 4 years, with an additional 3 years of follow-up, and full information with 196 iDFS events, the study would have 83% power, using a one-sided type I error rate of 0.025, to reject the null hypothesis of inferiority of platinum if the 4-year iDFS rate for platinum was at least 74% (ie, HR = 0.754, the alternative hypothesis for the noninferiority test). If the hypothesis of inferiority of platinum was rejected in the primary analysis, the secondary hypothesis of no difference versus superiority of platinum would be tested. Because of the closed testing procedure, no multiple-comparison adjustment was required for the two hypothesis tests. The Jennison-Turnbull repeated CI (RCI) method,²⁷ using O'Brien-Fleming boundaries,²⁸ was used to monitor for early stopping for efficacy (ie, noninferiority of platinum) or futility (ie, inferiority of platinum); futility was also monitored using the conditional power method (probability of eventually rejecting the null of inferiority). At each interim analysis, if the conditional power of the assigned treatment analysis was < 10% and the lower boundary of the two-sided 95% RCI was > 0.754 for the platinum versus capecitabine arms, the trial would then be stopped for futility. Conversely, if the upper boundary of the two-sided 95% RCI was < 1.154 for the platinum versus capecitabine arms, the trial would be stopped for efficacy (Appendix Fig A1, online only).

Comparisons between arms were intention-to-treat analyses among all patients. The Kaplan-Meier method²⁹ was used to estimate distributions for iDFS, RFS (time from random assignment to recurrence [locoregional or distant] or death), and OS. Cases with incomplete follow-up or without adequate disease evaluations were censored at the date last documented to be free of iDFS events. Cox proportional hazards models, stratified by the factors at random assignment, were used to estimate HRs, and two-sided 95% RCI was reported for iDFS and 95% CIs were reported for RFS and OS.

RESULTS

At the fifth interim analysis (58% information time, 113 iDFS events), the HR for platinum versus capecitabine = 1.09 (95% RCI, 0.62 to 1.90) (Appendix Fig A1), and grade 3 and 4 toxicities were more common in the platinum arm. Because of the low conditional power (6%), at its regular meeting on March 19, 2021 the Data and Safety Monitoring Committee recommended stopping the trial as it was unlikely that further accrual would be able to show noninferiority or superiority of the platinum arm. Data cutoff for this report was April 7, 2021.

Four hundred ten patients were enrolled between June 2016 and March 2021 (Fig 1), and 308 (78%) had basal subtype TNBC (primary analysis population). Study arms were well balanced (Table 1). At diagnosis, most tumors were high-grade, cT2, and cN0/N1; at surgery, most residual tumors were ypT1-2N0 (median residual breast tumor diameter 2.5 cm, interquartile range: 1.6-4.1 cm). Median interval between surgery and treatment initiation was 126 days (interquartile range: 103-149 days, range 35-201 days). Among patients randomly assigned to platinum, 88% received carboplatin and 12% received cisplatin.

TABLE 1.

Baseline Characteristics in Patients With Residual Basal Subtype Triple-Negative Breast Cancer^a

Efficacy

After a median follow-up of 20 months, there were 120 iDFS events (61% of the total information) among 308 patients with basal subtype TNBC (93 distant recurrences, 15 locoregional recurrences, five invasive second primary cancers without recurrence, and seven deaths without recurrence or a second primary cancer). The 3-year iDFS among 308 patients with basal subtype TNBC treated with platinum was 42% (95% CI, 30 to 53) versus 49% (95% CI, 39 to 59) with capecitabine (Table 2, Fig 2A). The iDFS HR for platinum versus capecitabine was 1.06 (95% RCI, 0.62 to 1.81). No clinical factors identified a subset of patients who benefited from platinum (Appendix Table A1, online only). The 3-year RFS was 46% (95% CI, 35 to 57) in the platinum arm and 49% (95% CI, 39 to 59) in the capecitabine arm (HR = 0.99; 95% CI, 0.67 to 1.45, Table 2, Fig 3A). The 3-year OS was 58% (95% CI, 45 to 68) in the platinum arm and 66% (95% CI, 56 to 74) in the capecitabine arm (HR = 1.13, 95% CI, 0.71 to 1.79, Table 2, Fig 3B).

TABLE 2.

Efficacy End Points

FIG 2.

Kaplan-Meier estimates of iDFS. Three-year iDFS (A) patients with basal subtype TNBC were similar across both arms, and (B and C) patients with basal subtype TNBC had poor prognosis (worse iDFS) regardless of treatment arm compared to patients with nonbasal subtype. (B) Patients with nonbasal subtype TNBC appear to have better iDFS when treated with capecitabine than with a platinum agent, but this finding is not statistically significant. HR, hazard ratio; iDFS, invasive disease-free survival; RCI, repeated CI; TNBC, triple-negative breast cancer.

FIG 3.

Kaplan-Meier estimates of RFS and OS. Three-year (A) RFS and (B) OS for patients with basal subtype triple-negative breast cancer were similar across both treatment arms and lower than originally anticipated. HR, hazard ratio; OS, overall survival; RFS, recurrence-free survival.

Among 88 patients with nonbasal subtype TNBC (22% of total patients), 24 iDFS events occurred (19 distant recurrences, two with invasive second primary cancers without recurrence, and three deaths without recurrence or a second primary cancer). The 3-year iDFS for patients treated with platinum was 46% (95% CI, 25 to 65) versus 69% (95% CI, 45 to 83) with capecitabine; the HR for platinum versus capecitabine was 1.94 (95% CI, 0.69 to 5.45) (Table 2, Fig 2B). Patients with basal subtype TNBC had worse iDFS than patients with nonbasal subtype TNBC (HR = 1.71; 95% CI, 1.10 to 2.67) (Fig 2C).

Adverse Events

The adverse event profile in both treatment arms was consistent with known toxicities of these agents (Table 3). Common toxicities seen with capecitabine included myelosuppression, nausea, diarrhea, and palmar-plantar erythrodysesthesia; myelosuppression and nausea were the most common toxicities with platinum. Peripheral sensory neuropathy rates were similar between arms.

TABLE 3.

Select Adverse Events

Although the overall incidence of toxicity was similar between arms, grade 3 and 4 toxicities (mainly anemia and leukopenia) were more frequent with platinum (26%; 95% CI, 20 to 33) compared with capecitabine (15%; 95% CI, 10 to 21). No grade 5 toxicities were observed; metastatic cancer progression was the cause of death in most patients. Eight patients developed second primary cancers (unrelated to treatment) during the study period.

Drug Exposure and Discontinuation

Among 368 patients off study treatment, 82% patients completed four cycles of platinum, and 79% patients completed six cycles of capecitabine. Dose reductions were more frequent in the platinum arm (Appendix Table A2, online only). Main reasons for early therapy discontinuation were disease progression for capecitabine and adverse events for platinum agents (Fig 1).

DISCUSSION

Despite a sound hypothesis supported by preclinical and clinical data, EA1131 showed that adjuvant platinum agents are unlikely to be noninferior or superior to capecitabine at improving iDFS for patients with residual TNBC after NAC. Furthermore, irrespective of treatment arm, a much higher than expected event rate was observed in this high-risk population.

Addition of platinum agents to anthracycline- and taxane-containing NAC for TNBC has consistently shown increases in pCR rates in clinical trials.^19,20,30 For this reason, many oncologists have incorporated platinum agents in NAC regimens. However, these neoadjuvant trials were not powered to address event-free survival or OS benefit and reported increased toxicity rates with platinum addition. Although individual patients who achieve pCR have a better prognosis, the association between improvements in pCR and improvements in long-term outcome is weak. Importantly, the magnitude of pCR improvement needed to result in long-term outcome improvement remains unclear.³¹ Given these limitations, EA1131 was designed to address the role of adjuvant platinum agents in a patient population expected to have a higher residual risk (patients with ≥ 1 cm invasive RD at surgery), thereby sparing those with highly chemotherapy-sensitive disease the added toxicity of upfront platinum therapy. Unfortunately, EA1131 results showed that platinum agents were associated with more severe toxicities and were not more effective than capecitabine.

There may be several explanations for these results. Platinum agents do not appear to offer a differential benefit in the metastatic setting for most patients with TNBC. The phase III TNBC Trial³² randomly assigned patients with metastatic TNBC to first-line carboplatin or docetaxel. Patients with germline BRCA1/2 mutations had improved response rates to carboplatin, but for patients without BRCA1/2 mutations, carboplatin was not better than docetaxel, regardless of basal subtype or homologous recombination deficiency (HRD) score (unweighted sum of three independent DNA-based measures of genomic instability in the tumor [loss of heterozygosity, telomeric allelic imbalance, and large-scale transitions]). A possible explanation for the lack of a differential response in metastatic TNBC is that BRCA1/2 gene methylation, which is associated with functional deficiencies in homologous recombination, is decreased in metastatic tumors compared with primary tumors.³³ It is then conceivable that RD after NAC could functionally mirror the genomic or epigenomic make-up of metastatic disease, which could explain why a platinum agent was not more effective than other drugs in the postoperative setting. However, platinum agents could still be effective in the neoadjuvant setting (in terms of pCR) in combination with other chemotherapy agents, as tumors more sensitive to DNA-damaging agents likely achieve a higher frequency of pCR and would therefore be excluded from EA1131. At the same time, the recently reported TBCRC030 phase II trial showed no difference in pCR rates between NAC with single-agent paclitaxel or cisplatin in patients with TNBC, regardless of HRD score.³⁴ These data are consistent with other trials showing that HRD does not predict response to a specific chemotherapy in the neoadjuvant setting.^19,30 This suggests that HRD is an imperfect biomarker to detect functional deficiencies in homologous recombination, and that prior exposure to different chemotherapy agents in the metastatic or neoadjuvant setting (which is not considered in preclinical models) may still cause potential clonal selection of homologous recombination-proficient cells in the residual tumor.

In EA1131, an effort to quickly optimize treatment while minimizing toxicity led us to select patients with the highest risk of recurrence: basal subtype (78% of all EA1131 participants) and RD post-NAC completion. Patients with basal subtype TNBC had a higher-than-expected observed risk of recurrence than previously reported in other treatment-escalation clinical trials,^9,19,20 regardless of the treatment received. This is consistent with a GEICAM/2003-11_CIBOMA/2004-01 trial preplanned analysis, showing that adjuvant capecitabine improvement in DFS is greater in the nonbasal phenotype.³⁵

The long-term outcome discrepancy between the EA1131 and CREATE-X trials⁹ may be partially explained by differences in patient selection: 42% of CREATE-X⁹ participants had marked treatment responses to NAC, with much less volume of RD at study entry. Furthermore, EA1131 included a diverse (US-based) population (70% of participants were White, 20% Black, and 10% Hispanic, instead of 100% Asian in CREATE-X⁹), which may also explain expected differences in capecitabine tolerance and need for dose modifications.²²

It is possible that the number of treatment cycles and time to treatment initiation negatively influenced outcomes observed across both study arms. At the same time, although EA1131 allowed fewer cycles of capecitabine than CREATE-X,⁹ we anticipated that a higher number of allowed cycles in the platinum or capecitabine arms would not have been feasible from a toxicity standpoint in patients already treated with NAC. The median time to protocol therapy initiation (126 days) may also have contributed to the poor outcomes observed in EA1131,³⁶ although CREATE-X allowed therapy to start up to 120 days after surgery (actual data not reported). Nonetheless, EA1131 results underscore the acute and unfulfilled need for better strategies for a high-risk cohort of patients.

EA1131 provides a richly annotated biobank to explore and develop further hypothesis for future clinical trials of new agents in this setting. Planned studies in the RD will evaluate genomic markers, gene expression, immune markers, circulating markers, including analyses of circulating tumor cells and cell-free DNA (cfDNA), and BRCA-mutation status. These planned studies may help us identify specific subsets of patients at risk, those who could potentially benefit from systemic intervention, and provide early evidence of clinical benefit from therapies for RD after NAC in TNBC (potential role of postsurgical cfDNA positivity to identify a high-risk cohort of TNBC after NAC³⁷). Currently, two planned adjuvant clinical trials in patients with residual TNBC post-NAC completion will use cfDNA positivity as an integral biomarker of patient selection and/or treatment allocation postoperatively: ASPRIA, a phase II trial of sacituzumab govitecan-hziy and atezolizumab, and PERSEVERE, a phase II randomized trial that will stratify patients based on cfDNA positivity and assign them a postoperative targeted therapy matched to the patient's tumor genomic sequencing.

EA1131 tested the hypothesis of a differential benefit from platinum drugs as post-NAC for residual basal subtype TNBC. Our results suggest that a benefit is unlikely regardless of intrinsic subtype. As such, these findings have an immediate impact in clinical practice. As such, these findings have an immediate impact in clinical practice, as they should discourage the adjuvant use of platinum agents in patients with residual TNBC after NAC outside of a clinical trial. Only a few other trials assessing the role of platinum in TNBC have survival as a primary objective, such as NRG BR-003 (ClinicalTrials.gov identifier: NCT02488967), a phase III randomized trial scheduled to complete accrual during 2021, that tests the addition of carboplatin to postsurgical anthracycline- and taxane-containing chemotherapy in chemotherapy-naive patients with pathologic stage II or III TNBC. For now, the use of adjuvant platinum agents in unselected patients with TNBC remains investigational and capecitabine remains the standard therapy. Finally, EA1131 lays the important groundwork for the next generation of studies addressing the critical need for more active and effective therapies for patients with high-risk TNBC.

APPENDIX

FIG A1.

EA1131 interim analysis monitoring (futility and efficacy). The graph outlines the range of possible outcomes for EA1131, designed to demonstrate noninferiority with superiority alternative (hybrid design). The primary analysis for the primary objective tested the hypothesis that platinum chemotherapy was not inferior to capecitabine. The null hypothesis of inferiority of platinum was defined as an HR = 1.154 (noninferiority margin). If noninferiority was demonstrated, the secondary hypothesis of no difference versus superiority of platinum would then be tested. At each interim analysis, if the conditional power (ie, probability of eventually rejecting the null of inferiority) of the assigned treatment analysis was < 10% and the lower boundary of the two-sided 95% RCI was > 0.754 (ie, the alternative hypothesis for the noninferiority test), the trial would then be stopped because of futility. As the HR for platinum versus capecitabine was 1.09 (95% RCI, 0.62 to 1.90) and the conditional power was 6%, the trial was stopped by the Data Safety Monitoring Committee because of futility, since it was unlikely that the trial would be able to show noninferiority or superiority of the platinum arm. HR, hazard ratio; RCI, repeated CI.

TABLE A1.

Subgroup Analysis for iDFS in All Randomly Assigned Patients

TABLE A2.

Treatment Cycles and Dose Modifications

AUTHOR CONTRIBUTIONS

Conception and design: Ingrid A. Mayer, Fengmin Zhao, Carlos L. Arteaga, William F. Symmans, Amye J. Tevaarwerk, Sofia F. Garcia, Lynne I. Wagner, Joseph A. Sparano, Antonio C. Wolff, Kathy D. Miller

Financial support: Lynne I. Wagner

Administrative support: Ingrid A. Mayer, William F. Symmans, Joseph A. Sparano, Kathy D. Miller

Provision of study materials or patients: Ingrid A. Mayer, Carlos L. Arteaga, Brian L. Burnette, Della F. Makower, Kimberly A. Morley, Chirag R. Jani, Bernard Tawfik, Lisa E. Flaum, Erica L. Mayer, William M. Sikov, Joseph A. Sparano, Antonio C. Wolff, Kathy D. Miller

Collection and assembly of data: Ingrid A. Mayer, Fengmin Zhao, William F. Symmans, Brian L. Burnette, Amye J. Tevaarwerk, Della F. Makower, Margaret Block, Kimberly A. Morley, Chirag R. Jani, Shabana J. Dewani, Bernard Tawfik, Lisa E. Flaum, Angela M. DeMichele, Antonio C. Wolff, Kathy D. Miller

Data analysis and interpretation: Ingrid A. Mayer, Fengmin Zhao, Carlos L. Arteaga, Ben H. Park, Brian L. Burnette, Amye J. Tevaarwerk, Karen L. Smith, Craig Mescher, Erica L. Mayer, William M. Sikov, Eve T. Rodler, Lynne I. Wagner, Angela M. DeMichele, Antonio C. Wolff

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Randomized Phase III Postoperative Trial of Platinum-Based Chemotherapy Versus Capecitabine in Patients With Residual Triple-Negative Breast Cancer Following Neoadjuvant Chemotherapy: ECOG-ACRIN EA1131

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).