Randomized Phase II Trial of Anthracycline-free and Anthracycline-containing Neoadjuvant Carboplatin Chemotherapy Regimens in Stage I–III Triple-negative Breast Cancer (NeoSTOP)

This article is featured in Highlights of This Issue, p. 903

Triple-negative breast cancer (TNBC), which is defined by the lack of expression of estrogen receptor (ER) and progesterone receptor (PgR) and absence of ERBB2 (HER2) overexpression and/or gene amplification, accounts for 15% of all breast cancers in the United States. TNBC is associated with inferior long-term outcomes compared with other breast cancer subtypes (1, 2). Systemic chemotherapy reduces the risk of death and recurrence in patients with TNBC and is recommended for patients with TNBC with stage I (T > 1 cm)-III disease (3, 4). Pathologic response to neoadjuvant chemotherapy is a very good surrogate for long-term outcomes. Attainment of pathologic complete response (pCR) is associated with excellent long-term outcomes, and conversely, presence of residual disease is associated with high risk of recurrence in patients with TNBC (1, 5, 6).

Recent studies have demonstrated that addition of neoadjuvant carboplatin to anthracycline plus taxane regimen improves pCR rate in TNBC (7, 8). However, this improvement in pCR rate comes at the cost of increase in toxicity, and long-term benefits from the addition of neoadjuvant platinum to anthracycline plus taxane combinations in TNBC are not yet clear (9, 10). Thus, there is a growing interest in exploration of anthracycline-free carboplatin plus taxane–based neo/adjuvant chemotherapy regimens in TNBC. Anthracycline and cyclophosphamide-devoid chemotherapy regimens are also attractive in the curative setting, because exposure to these agents is associated with small, but serious late risks (secondary leukemia/myelodysplastic syndrome and cardiomyopathy; refs. 11–13). Neoadjuvant carboplatin plus taxane regimens have demonstrated encouraging efficacy with a favorable toxicity profile in TNBC. WSG-ADAPT-TN trial demonstrated a pCR rate of 45.9% with a 12-week carboplatin plus nab-paclitaxel regimen, and an 18-week carboplatin plus docetaxel regimen also yielded a high pCR rate (55%; refs. 14, 15). Importantly, patients achieving pCR with platinum-taxane regimens demonstrate excellent 3-year outcomes without adjuvant anthracycline (15–17).

We hypothesized that carboplatin-containing anthracycline-free and anthracycline-based neoadjuvant chemotherapy regimens will yield comparable pCR rates in patients with stage I–III TNBC.

NeoSTOP (Neoadjuvant Study of Two Platinum Regimens in Stage I–III TNBC, NCT02413320) was a phase II, randomized, open-label trial. Participants were enrolled at two academic and three community sites within the Kansas City metropolitan area and the state of Kansas. This study was conducted in accordance with the U.S. Common Rule and the International Ethical Guidelines for Biomedical Research Involving Human Subjects. The study was approved by the Human Subjects Committee at the University of Kansas Medical Center (Westwood, KS) or the local institutional review board at each participating site, and all patients provided written informed consent.

Eligible patients were females aged 18 years or older with stage I (T size > 1 cm), II, or III TNBC who had not undergone definitive breast surgery and had not received systemic chemotherapy. Patients had to have adequate hematologic, hepatic, and renal function, and Eastern Cooperative Oncology Group performance status of 0 or 1. Patients were excluded if they had ongoing grade 2 or higher peripheral neuropathy, had ejection fraction <50% on ECHO or MUGA, or had inflammatory breast cancer.

Triple negativity was defined as ER and PgR IHC nuclear staining of ≤10% and HER2 negativity per 2013 American Society of Clinical Oncology/College of American Pathologists guidelines (18). For patients with clinically/radiologically suspicious axillary lymph node/s, histologic confirmation by core needle biopsy or fine-needle aspiration was required prior to enrollment.

Patients were randomized (1:1, stratified by nodal status) to one of two study arms (Supplementary Fig. S1; study schema): arm A, CbP → AC (carboplatin AUC6 every 3 weeks × 4 + paclitaxel 80 mg/m² every week × 12, followed by doxorubicin 60 mg/m² + cyclophosphamide 600 mg/m² every 2 weeks × 4) or arm B, CbD (carboplatin AUC6 + docetaxel 75 mg/m² every 21 days × 6).

Myeloid growth factor support (pegfilgrastim or equivalent) was administered with all four cycles of dose-dense AC in arm A and with all six cycles of CbD in arm B. Myeloid growth factor support was permitted during the CbP phase of arm A, but was not mandated. In terms of hematologic parameter requirements for chemotherapy administration, absolute neutrophil count (ANC) ≥ 1,000/μL and platelet count ≥ 50,000/μL were required for the weekly paclitaxel administration during the CbP phase of arm A. For the dose-dense AC phase of arm A and for all six cycles of arm B, ANC ≥ 1,000/μL and platelet count ≥ 75,000/μL were required for cycle administration. During the CbP phase, skipped doses of paclitaxel were not made up prior to start of dose-dense AC.

Patients underwent definitive breast surgery after the conclusion of neoadjuvant chemotherapy. Axillary lymph node sampling at time of surgery was required in all patients, but the extent of axillary surgery was determined by the treating physician. Subsequent irradiation and postoperative adjuvant chemotherapy were also determined by the treating physician, and patients were followed for recurrence and survival. All patients were asked to complete a six-question travel and time analysis survey once during neoadjuvant treatment (information collected from this survey was used for cost analysis; ref. 19). Germline BRCA1/2 testing was offered to patients on the basis of standard clinical practice/National Comprehensive Cancer Network guidelines, and commercially available tests were used for germline testing.

Pathologic response was determined locally, without central pathologic review. pCR was defined as the absence of residual invasive disease in the breast and axilla, with or without ductal carcinoma in situ (ypT0/isN0). Residual cancer burden (RCB) for all patients was calculated centrally (using parameters reported on individual patient pathology reports) using the classification by Symmans and colleagues (20). Patients achieving pCR (RCB 0) or near-pCR (RCB I) were assessed within the group RCB 0+I. Common Terminology Criteria for Adverse Events (CTCAE) version 4.03 was used for toxicity assessment.

Histopathologic determination of stromal tumor-infiltrating lymphocyte (sTIL) density was assessed using a single hematoxylin and eosin–stained invasive tumor section according to criteria described previously (21). The histopathologist who performed slide reviews (R. Salgado) was blinded to randomization and outcome information. sTIL density was reported as a percentage estimate in increments of 5%.

This was a randomized phase II study with primary endpoint of estimation of pCR in breast and axilla in the two study arms. Secondary endpoints included RCB and toxicity; cost and survival analysis were exploratory endpoints. Estimated pCR rate in each treatment arm was 50%, with top bound constraints for the SEs at 7.5% and the top bound constraint for the SE of the difference between pCR rates of the two arms at 12%, yielding a required sample size of 45 patients per treatment arm (22). To ensure 45 evaluable patients in each treatment arm, overall sample size was inflated to 100 patients. The pCR rates and RCB 0+I rates were estimated for both treatment arms and 95% exact binomial confidence bounds were calculated using the exact two-sided binomial test. The study was not powered to show superiority or equivalence of pCR in the two arms. Treatment-related toxicity and treatment delivery completion were assessed for each arm. Overall frequencies and percentages were summarized for demographic and clinicopathologic characteristics. Categorical variables were compared between treatment arms using Fisher exact test. Continuous variables were compared using Mann–Whitney U-test. Logistic regression analysis was used to examine the effect of multiple variables on attainment of pathologic response. Event-free survival (EFS) was defined as the time from diagnosis to first recurrence (invasive ipsilateral breast, invasive local/regional, or distant) or to breast cancer–related death. Overall survival (OS) was defined as time from diagnosis to death as a result of any cause. Patients were censored on the date of last contact if an event had not been observed. Survival analysis was descriptive; survival curves were assessed by the Kaplan–Meier method and groups compared by the log-rank test. All analyses were conducted using SPSS Statistics version 25 (IBM Corporation).

Cost analysis: cost minimization analysis was utilized because the efficacy of the two regimens was hypothesized to be similar (23). Direct and key indirect costs of neoadjuvant chemotherapy treatment were estimated, as were costs of patients' transportation and loss of productivity associated with time spent for receipt of chemotherapy treatments. Direct costs were computed for all patients on an intention-to-treat basis. The cost of outpatient care associated with neoadjuvant treatment was calculated using the 2018 Medicare reimbursement rates for chemotherapeutic agents (including administration costs), chemotherapy premedication drugs, myeloid growth factors, provider office visits, and laboratory tests. Standard costs for treatment-related febrile neutropenia hospitalization and blood transfusion were used (24). Patient transportation costs (for chemotherapy treatment visits) were calculated using the 2018 IRS mileage rate, and lost productivity (for time spent receiving chemotherapy infusion and travelling to chemotherapy treatment visits) was calculated using the 2018 median hourly wage for Kansas. Cost of care was reported as the mean cost for each treatment arm because, compared with median, arithmetic mean is a more informative measure of total healthcare costs and is also typically utilized for cost minimization assessment (25, 26). The mean combined cost of each regimen was compared using unpaired t test.

Between 9 July 2015 and 18 May 2018, 101 patients were enrolled and randomly assigned to arm A (n = 49) or arm B (n = 52). One patient randomized to arm A did not receive any study treatment (physician decision) and is not included in the intention-to-treat analysis (Supplementary Fig. S2). Baseline demographic and tumor characteristics of the study population are summarized in Table 1. Median age was 51 years, 30% of patients had axillary lymph node–positive disease, and 16% had ER/PgR expression between 1% and 10%. Seventeen percent of the study population carried a deleterious BRCA1/2 mutation. Baseline demographic and tumor characteristics were balanced between the two arms (Table 1).

Two patients (one each in arms A and B, respectively) developed progressive disease during neoadjuvant chemotherapy and are counted as no pCR in the intent-to-treat analysis. Two patients in arm A did not undergo breast surgery (development of distant metastases while on neoadjuvant chemotherapy, n = 1 and patient decision, n = 1) and are counted as no pCR. Rate of pCR was 54% [95% confidence interval (CI), 40%–69%] in arm A and 54% (95% CI, 40%–68%) in arm B. RCB 0+I rate was 67% in both arms (Fig. 1). Among patients with RCB I–III disease, 30% of patients in arm A and 38% in arm B received adjuvant chemotherapy (P = 0.752).

On univariate logistic regression analysis, grade III histology was associated with higher likelihood of achieving a pCR (P = 0.025). Presence of gBRCA mutation (P = 0.052) and sTILs ≥ 20% (P = 0.096) showed a trend toward higher pCR (Table 2). On multivariate logistic regression analysis, no factors were associated with pCR.

sTILs information was available for 90 of 98 patients who underwent definitive surgery (tumor specimen was unavailable/inadequate for sTILs evaluation in n = 8). Because median sTIL density and pCR were similar for the two arms (Table 1), both arms were combined for assessment of sTILs and pCR. Numerically higher sTILs were noted for patients with pCR compared with those with residual disease (median sTILs 20% vs. 5%; P = 0.25; Supplementary Fig. S3). A total of 43% of patients demonstrated sTIL density ≥ 20% (48% in arm A and 40% in arm B). sTIL density of ≥20% was associated with a trend toward higher pCR (pCR = 67% vs. 49% in patients with sTILs ≥ 20% and <20%, respectively; P = 0.098).

Figure 2 shows Kaplan–Meier curves for EFS and OS by study arm (A and B) and by pCR status (C and D). At a median follow-up of 38 months (range, 6–59 months) EFS and OS appeared similar in the two arms (A and B). Arms A and B were combined for assessment of pCR and survival outcomes. Patients achieving pCR demonstrated numerically better EFS and OS compared with patients without pCR (Fig. 2C and D). Estimated 3-year EFS was 100% and 81% (95% CI, 69–93%) in patients with and without pCR, respectively (log-rank P < 0.003). Estimated 3-year OS was 100% and 86% (95% CI, 75%–96%) for patients with and without pCR, respectively (log-rank P < 0.003). On multivariate Cox regression analysis [factors analyzed: study arm, nodal status, tumor–node–metastasis (TNM) stage, sTILs, pCR, and grade], only pCR status was associated with EFS (log-rank P < 0.001) and OS (log-rank P = 0.003).

Grade 3/4 adverse events were more common in arm A than in arm B (73% vs. 21%; P < 0.0001), with the most notable differences seen in the rates of neutropenia, febrile neutropenia, and anemia (Table 3). All episodes of anemia were grade 3. In arm A, 44% (21/48) of patients received filgrastim support during the CbP phase and 17% (8/48) of patients received red blood cell transfusion (all during the dose-dense AC phase).

In arm A, 72% of patients completed 10 or more doses of paclitaxel, and 85% completed all four doses of carboplatin. Ninety-two percent of arm B patients completed all six cycles of CbD treatment (Supplementary Fig. S4).

One patient in arm A developed secondary acute myeloid leukemia 23 months after completion of neoadjuvant chemotherapy and succumbed to the leukemia at age 53. No treatment-related deaths occurred in arm B.

The mean total cost of neoadjuvant treatment (treatment costs plus patient transportation and lost productivity costs) was lower in arm B compared with arm A (arm B = $33,148 and arm A = $36,720; P = 0.02).

The NeoSTOP study demonstrates that a non-anthracycline two-drug regimen of CbD yields pCR and RCB 0+I rates similar to the four-drug regimen of CbP followed by AC, but with a much more favorable toxicity profile, higher treatment completion rates, and lower health care costs. We also show that EFS and OS appear broadly similar in the two arms.

Anthracyclines and cyclophosphamide, although highly active for treatment of breast cancer, carry established small, but serious long-term risks (secondary leukemia/myelodysplastic syndrome and cardiomyopathy; refs. 11–13). Indeed, in our study there was one fatal case of secondary leukemia in the anthracycline arm. The carboplatin/docetaxel regimen minimizes these serious long-term toxicities. Rates of pCR noted in this trial with CbP followed by AC (54%) are in line with what have been reported in previous randomized trials with this regimen (pCR rate of 54% and 58% in CALGB 40603 and BrighTNess trials, respectively; refs. 8, 9). A pCR rate of 54% with six cycles of CbD is also consistent with previous reports and appears to be numerically higher than those reported in several trials using classical neoadjuvant anthracycline-taxane combinations, in which 28%–39% of patients with TNBC achieved pCR (15, 27–29). Compared with some contemporary neoadjuvant trials, NeoSTOP included a lower risk population (17% with stage I disease), which could be impacting the observed pCR rates. However, sensitivity analysis after exclusion of patients with stage I disease showed no change in pCR rates in the two arms (arm A = 54% and arm B = 50%). Because this study enrolled patients at a few institutions within the United States, the proportion of patients with node-positive disease in NeoSTOP (30%) is slightly lower compared with other contemporary larger international trials (43% in BrighTNess; ref. 8). In our study, patients who achieved pCR (on either arm) did not receive further adjuvant chemotherapy, and patients with residual disease received adjuvant chemotherapy per recommendation of treating physicians. Our observed excellent EFS and OS among patients achieving pCR and poor outcomes among those with residual disease are in line with published literature (1, 5, 6). Given the small trial size, these survival results should be viewed with caution. A larger neoadjuvant study of CbD regimen (N = 190) has previously reported that patients who achieve pCR demonstrate excellent 3-year recurrence-free survival (90%) and OS (94%) without adjuvant anthracycline, thus indicating that patients achieving pCR with this regimen can safely avoid anthracycline (16).

Although it provides intriguing data in support of a neoadjuvant carboplatin/taxane regimen, our study does have some limitations. Although randomized, this is a small phase II trial and was not designed as a noninferiority study. The logical next step would be to confirm these findings in a phase III trial. However, an appropriately powered phase III trial to show noninferiority of the two chemotherapy regimens would require a sample size exceeding 2,500 patients. In the current era, where the majority of clinical research investigations in TNBC are focused on newer agents (immune checkpoint inhibitors, agents targeting the PI3K/AKT pathway, etc.), a trial evaluating equivalence of two clinically available regimens of conventional chemotherapy is unlikely to be appealing to patients or engage traction within funding agencies. This study was also not designed to address merits of inclusion or exclusion of neoadjuvant carboplatin, and, thus, cannot provide any insights into that question. However, in situations where the decision to include carboplatin in the neoadjuvant treatment has been made, our study does provide evidence for effectiveness of an anthracycline-free two-drug alternative with seemingly similar efficacy, but a better toxicity profile compared with the four-drug option. This regimen also provides an effective alternative for patients who need neoadjuvant chemotherapy, but are not candidates for anthracyclines.

The tumor immune microenvironment plays an important role in prognosis and response to chemotherapy for TNBC. Multiple studies have confirmed the association between increasing sTILs in pretreatment tumor tissue and higher pCR plus improved long-term outcomes, and this observation spans across different chemotherapy regimens among several trials (30–34). In our study, although there was a numeric association between sTILs and pCR, this difference was not statistically significant, probably due to the small sample size. Interestingly, in a setting of platinum-based chemotherapy, we noted a quite robust pCR rate of almost 70% in patients with baseline sTIL density of ≥20%. These findings suggest that baseline sTILs can identify patients with high likelihood of pCR who can potentially be considered for treatment deescalation trials (e.g., omission of immunotherapy and deescalation of chemotherapy regimen/duration). Additional translational studies on pretreatment tumor tissue and blood to explore biomarkers predictive of pathologic response are ongoing.

The toxicity profile of CbP followed by AC observed in our study is comparable with what has been previously reported with this regimen. In the recent BrighTNess trial, the rate of grade 3–4 anemia was 17% (during the CbP phase of treatment; rate during the entire treatment not reported), and the rate of febrile neutropenia was also 17%. Addition of carboplatin to weekly paclitaxel is known to be associated with reduced number of paclitaxel doses administered (8, 9). In our trial, the carboplatin and paclitaxel phase of arm A was to be completed within a period of 12 weeks per protocol. Paclitaxel dose administration with this schedule can be improved by extending the allowable time beyond 12 weeks. The BrighTNess trial allowed administration of carboplatin and paclitaxel to extend over 16 weeks, which led to a higher proportion of paclitaxel doses administered. Tolerance of the CbD regimen used in this study was good and was consistent with previous reports (15, 16). Cost analysis suggests that CbD regimen is associated with lower health care cost compared with CbP followed by AC. We report better tolerability of CbD over CbP followed by AC using conventional CTCAE toxicity; however, patient-reported outcomes were not collected in the study, thus we are unable to provide any insights on patient perspective regarding toxicities of the two regimens.

Given its good tolerance, the CbD chemotherapy backbone is also well suited for treatment escalation trials where additional agents (e.g., checkpoint inhibitors) are combined with chemotherapy. Recently, results from the KEYNOTE-522 study show that the addition of pembrolizumab (an anti-programmed death 1 mAb) to carboplatin plus paclitaxel followed by anthracycline-based chemotherapy significantly improved pCR from 51.2% to 64.8% in patients with stage II or III TNBC (35). Given the similar pCR noted with CbP followed by AC and CbD regimens in our study, the more tolerable CbD regimen can also serve as an excellent alternative neoadjuvant backbone for immunotherapy combinations. An ongoing phase II trial is assessing six cycles of CbD plus pembrolizumab in stage I–III TNBC (NCT03639948).

In summary, we provide evidence that for patients with early-stage TNBC, carboplatin plus docetaxel regimen achieves encouraging pCR and survival rates, which appear similar to those of taxane-carboplatin plus AC, yet with lower patient and financial toxicity. This regimen provides a safe alternative for patients who are not candidates for anthracyclines and should also be explored further for neoadjuvant chemotherapy deescalation strategies.

A. O'Dea reports personal fees from PUMA Biotechnology, Pfizer, Astra Zeneca, and Daiichi Sankyo outside the submitted work. M. Hoffmann reports personal fees from Pharmacyclics outside the submitted work. A.K. Godwin reports grants from NIH/NCI and NIH/NIGMS during the conduct of the study. Q. Khan reports grants from Novartis outside the submitted work. No disclosures were reported by the other authors.

P. Sharma: Conceptualization, resources, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. B.F. Kimler: Formal analysis, writing-review and editing. A. O'Dea: Investigation, writing-review and editing. L. Nye: Investigation, writing-review and editing. Y.Y. Wang: Data curation, writing-review and editing. R. Yoder: Data curation, formal analysis, project administration, writing-review and editing. J.M. Staley: Data curation, formal analysis, investigation, writing-review and editing. L. Prochaska: Investigation, writing-review and editing. J. Wagner: Investigation, writing-review and editing. A.L. Amin: Investigation, writing-review and editing. K. Larson: Investigation, writing-review and editing. C. Balanoff: Investigation, writing-review and editing. M. Elia: Investigation, writing-review and editing. G. Crane: Investigation, writing-review and editing. S. Madhusudhana: Investigation, writing-review and editing. M. Hoffmann: Investigation, writing-review and editing. M. Sheehan: Investigation, writing-review and editing. R. Rodriguez: Data curation, investigation, writing-review and editing. K. Finke: Data curation, writing-review and editing. R. Shah: Data curation, investigation, writing-review and editing. D. Satelli: Investigation, writing-review and editing. A. Shrestha: Investigation, writing-review and editing. L. Beck: Investigation, writing-review and editing. R. McKittrick: Investigation, writing-review and editing. R. Pluenneke: Investigation, writing-review and editing. V. Raja: Investigation, writing-review and editing. V. Beeki: Investigation, writing-review and editing. L. Corum: Investigation, writing-review and editing. J. Heldstab: Investigation, writing-review and editing. S. LaFaver: Investigation, writing-review and editing. M. Prager: Formal analysis, investigation, writing-review and editing. M. Phadnis: Formal analysis, writing-review and editing. D.P. Mudaranthakam: Resources, funding acquisition, writing-review and editing. R.A. Jensen: Resources, funding acquisition, writing-review and editing. A.K. Godwin: Resources, funding acquisition, investigation, writing-review and editing. R. Salgado: Formal analysis, investigation, writing-review and editing. K. Mehta: Formal analysis, investigation, methodology, writing-review and editing. Q. Khan: Data curation, investigation, methodology, writing-review and editing.

This work was supported by the University of Kansas Cancer Center pilot grant, the Cancer Center Support Grant to the University of Kansas Cancer Center (P30 CA168524 to Biospecimen Repository Core Facility), the Breast Cancer Research Foundation (17-194 to R. Salgado), and the National Institute of General Medical Sciences (P20 GM130423 to A.K. Godwin).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.