Multi-institutional prospective observational study of the effectiveness of eribulin in rst-line or second-line chemotherapy for HER2-negative hormone-resistant advanced or metastatic breast cancer: The KBCRN A001: E-SPEC Study

Purpose To investigate the survival impact of eribulin use in rst-line and second-line chemotherapy for patients with endocrine-resistant advanced or metastatic breast cancer (AMBC) in the real-world clinical setting. Methods This multi-institutional prospective cohort study enrolled patients with triple-negative AMBC or estrogen receptor (ER)-positive AMBC refractory to at least one previous endocrine therapy selected at the physician’s discretion. The overall survival from the start of rst-line (OS1) and second-line chemotherapy (OS2) were assessed. Adjusted hazard ratio (HR) between eribulin and the other regimens (oral 5-uorouracil [5-FU] and anthracycline/taxane) was calculated using a stratied proportional hazards model that included prespecied prognostic factors.


Introduction
In recent years, screening mammography has been utilized widely, and the number of patients with early stage breast cancer (BC) has increased. However, despite the increase in early cancer detection, the number of patients with advanced or metastatic breast cancer (AMBC) has not decreased [1]. The 5-year survival rate of patients with AMBC is less than 25%, and the patients are very unlikely to be cured [2,3]. Therefore, the main objectives in the treatment of AMBC are to prolong overall survival (OS) and maintain the quality of life (QOL). Prior to treatment initiation, treatment-related factors, such as the estrogen receptor (ER), progesterone receptor (PgR), human epidermal growth factor receptor 2 (HER2), and menopausal status and clinical symptoms, are carefully evaluated to facilitate shared decision-making by the patients and physicians on which drugs should be used, as indicated in the treatment strategy presented by Hortobagyi [4]. This strategy has been indicated over the years based on the premise that the use of cytotoxic chemotherapy should be delayed as much as possible in ER-positive AMBC unless the patient is in a visceral crisis.
Recently, the effectiveness of immune checkpoint inhibitors and poly-ADP-ribose polymerase (PARP) inhibitors has been demonstrated in some patients with HER2-negative BC [5,6]. However, a limited number of patients with programmed death ligand 1 (PD-L1)-positive triple-negative or BRCA-mutationpositive BC are available to receive targeted therapy. Therefore, chemotherapy is generally indicated for many AMBC patients with endocrine therapy-refractory ER-positive or triple-negative BC. For rst-line chemotherapy, based on available evidence [7,8] and guidelines [2,9], anthracycline or taxane-based regimens are common choices for patients who have not been treated previously with (neo) adjuvant chemotherapy comprising those drugs. A recent clinical trial in Japan demonstrated that rst-line S-1, an oral 5-uorouracil (FU) regimen, was non-inferior to docetaxel or paclitaxel in terms of OS [10]. As secondline chemotherapy, anthracyclines or taxanes are recommended when they are not used as rst-line treatment [2,9]. For subsequent regimens, the European Society for Medical Oncology guideline recommends single-agent capecitabine, vinorelbine, or eribulin therapy as the preferred choice, while factoring in the following aspects: tumor responsiveness to the previous treatment, patient's performance status, possible adverse events of the drug, and QOL.
Eribulin is a novel chemotherapeutic drug that binds to a different microtubule site wherein taxane acts, and it inhibits the microtubule motility in cancer cells [11,12]. In a phase III study (EMBRACE) for AMBC previously treated with anthracyclines and taxanes, eribulin monotherapy signi cantly prolonged OS when compared with the physician's treatment selection [hazard ratio (HR) 0.81, 95% con dence interval (CI) 0.66-0.99; p = 0.041)] [13]. In this study, OS was prolonged despite similar progression-free survival (PFS) between the eribulin and control groups, and a similar trend was observed in Study 301 that compared eribulin with capecitabine [14]. Moreover, as third-line or later chemotherapy in sarcoma patients, eribulin prolonged OS more than the conventional standard treatment of dacarbazine (HR 0.768, 95% CI 0.618-0.954, p = 0.0169) despite no signi cant difference in PFS [15]. The reason for this difference between PFS and OS, as shown in basic research, is that eribulin inhibits epithelialmesenchymal transition (EMT) or angiogenesis [16][17][18]. Furthermore, the time to the occurrence of new fatal metastases, such as to the lung, liver, and brain, was an important factor for a worse prognosis in an exploratory analysis of the pivotal study of eribulin and in our retrospective study [19,20]. Although there are few single-arm phase II studies on the early line use of eribulin [21][22][23][24], the reason of the abovementioned survival bene t and optimal sequences of chemotherapy for patients with AMBC are still unclear. In Japan, off-label eribulin treatment is sometimes used in the early line setting for AMBC in patients without a history of anthracycline and taxane treatment because of the expectation of an OS-prolongation effect. Hence, we considered that we could evaluate the appropriate application of eribulin in the real-world setting.
The primary aim of this study was to investigate the OS of rst-and second-line chemotherapy in the realworld clinical setting among different regimens, and the secondary aims were to evaluate optimal chemotherapy schedule and the clinicopathological factors that contribute to OS prolongation with eribulin use.

Study design and participants
This multi-institutional, prospective, observational cohort study enrolled consecutive patients with AMBC who were treated between July 2015 and July 2017 at 1 of the 27 institutions of the Kyoto Breast Cancer Research Network (KBCRN) that participated in this study. The inclusion criteria were as follows: 1) female sex with histologically or cytologically con rmed AMBC, 2) HER2-negative status de ned according to the American Society of Clinical Oncology (ASCO)/College of American Pathologists guideline in 2014 [25], 3) resistant to endocrine therapy, 4) scheduled to receive rst-or second-line chemotherapy at their physician's discretion, 5) aged between 20 and 75 years, 6) Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) of 0-3, and 7) adequate bone marrow and major organ function at the discretion of the attending physician. The exclusion criteria were as follows: 1) symptomatic metastasis in the central nervous system, 2) allergic/hypersensitive to any ingredients of drugs included in the scheduled treatment, and 3) unsuitable for study participation, as evaluated by the primary physician.
The scheduled observation period was 2.5 years from the last patient's enrolment in the study. Details of the treatment regimen, laboratory data, vital signs, and 5-level EuroQol 5 Dimensions (EQ-5D-5L) scores prior to the start of each chemotherapy as well as the determinants of treatment e cacy were investigated until the end of third-line chemotherapy. Furthermore, the following baseline variables that could affect outcomes were recorded: medical history, complications, smoking status, cancer stage, hormone receptor status, history of adjuvant chemotherapy and adjuvant endocrine therapy, and metastatic sites.
The appropriate ethics committee or institutional review board at each study site approved the nal protocol. This study was conducted in accordance with the Japanese Guidelines for Clinical Research of the Ministry of Health, Labour and Welfare and the Declaration of Helsinki. All participants provided written informed consent prior to study participation. This study is registered with Clinical Trials.gov (NCT 02551263).
De nition of resistance to endocrine therapy In this study, "resistance to endocrine therapy" was de ned by the presence of at least one of the following criteria: 1) both ER-and PgR-negative status on immunohistochemical analysis of the primary or recurrent lesions according to the ASCO/College of American Pathologists guidelines [26]. In cases with discordant results between the primary and recurrent lesions, the result of the recurrent lesion was considered; 2) refractory to at least one endocrine therapy in ER-positive AMBC (subsequent endocrine therapy after initiation of chemotherapy was acceptable at the discretion of the attending physician); and 3) relapse during adjuvant endocrine therapy or within 6 months after the completion of adjuvant endocrine therapy.

De nition of lines of chemotherapy
The lines of chemotherapy were de ned as follows: The rst-line treatment was the rst chemotherapy for hormone-resistant or triple-negative AMBC. However, in cases where induction chemotherapy was administered for hormone receptor-positive AMBC that was endocrine therapy-naive but received maintenance endocrine therapy, the induction chemotherapy was not included in the rst-line treatment. Therefore, rst-line chemotherapy was de ned as the rst chemotherapy for tumors refractory to maintenance endocrine therapy. Only one induction chemotherapy regimen was allowed.
Second-or third-line chemotherapy was de ned as a subsequent chemotherapy administered after disease progression or after adverse event-related withdrawal of the previous chemotherapy. For patients with recurrence during postoperative adjuvant chemotherapy, the chemotherapy cycle immediately before recurrence was de ned as the rst-line treatment, whereas the regimen just after recurrence was de ned as the second-line treatment.

Treatment
After study enrolment, treatment according to the general practice was administered at the attending physician's discretion. Moreover, dose modi cations, chemotherapy schedules, and decisions of treatment discontinuation were based on the physician's evaluation and shared decision-making by patients and physicians. Furthermore, subsequent endocrine therapies after chemotherapy initiation were allowed for hormone receptor-positive patients, although these therapies were not counted in the treatment lines. Molecular targeted drugs, such as palbociclib, abemaciclib, or everolimus, which are used in combination with endocrine therapy, were classi ed as endocrine therapy and not as chemotherapy.

Outcomes
The primary endpoint was the rst-line OS (OS1), de ned as the time from the start of rst-line chemotherapy to death from any cause. The secondary endpoints were the second-line OS (OS2) and third-line OS (OS3), which were de ned as the time from the start of second-and third-line chemotherapy, respectively. Moreover, PFS, time to treatment failure, new metastasis-free survival, safety, and EQ-5D-5L scores were evaluated. The analysis of the utility score assessed with EQ-5D-5L has been presented in another report.

Statistical analysis
The full analysis set included all enrolled patients, excluding those deemed ineligible after enrolment because they had not received the allocated treatment or had no data after treatment. The cumulative survival rate was estimated using the Kaplan-Meier method. The number of events, incidence (/100 person-years; 95% con dence intervals [CI]), and median survival time (95% CI) were calculated for eribulin, oral 5-FU, and anthracycline/taxane-based chemotherapy regimens. Adjusted hazard ratios with 95% CIs for the comparisons of eribulin with oral 5-FU and an anthracycline/taxane regimen were calculated using a strati ed proportional hazards model that included age, body mass index, smoking, menopausal status, cerebrovascular decease, diabetes, TNM classi cation N, TNM classi cation M, histological grade, hormone receptor status, disease-free interval, endocrine therapy, (neo) adjuvant chemotherapy, radiation therapy, ECOG PS at rst-line chemotherapy, metastasis recurrence site, liver metastasis, lung metastasis, and brain metastasis as covariates after strati cation by the timing of registration.
All analyses were conducted using SAS version 9.4 (Cary, NC, USA). The signi cance level was set to p < 0.05 (2-tailed).

Patient characteristics
We enrolled 201 patients, but the data of 21 were excluded from the full analysis set for the following reasons: rst-line chemotherapy was not administered to 16, 2 were ineligible (one case was found to be HER2 positive after enrolment, and the other was enrolled from third-line treatment), 2 withdrew consent, and 1 was registered erroneously. The median age of 180 patients treated with rst-line chemotherapy was 61 years (range 29-75 years), and the baseline patient characteristics are shown in Table 1. Among those, 146 and 101 patients underwent second-and third-line chemotherapy, respectively. The study participant ow is shown in Fig. 1. Eribulin was used for 116 patients as rst-or second-line chemotherapy, whereas oral 5-FU, taxane, and anthracycline were used for 83, 99, and 16 patients, respectively, as rst-or second-line chemotherapy.

Chemotherapy sequences
The sequences of chemotherapy followed in this study are shown in Fig. 2. The frequent rst-and second-line chemotherapy sequences were as follows: taxanes followed by eribulin (n = 33), oral 5-FUbased therapy followed by eribulin (n = 26), eribulin followed by taxanes (n = 21), taxanes followed by oral 5-FU (n = 11), and eribulin followed by oral 5-FU (n = 10). Additionally, the reasons for the selection of eribulin by the participating physician despite not using anthracyclines or taxanes are shown in Supplementary Fig. S1. Eribulin was administered as rst-or second-line therapy for patients who did not receive anthracyclines or taxanes when the main concern was "lower toxicity."
Data pertaining to OS, PFS, and new metastasis-free survival by treatment line and chemotherapy are presented in Supplementary Table S1. The OS1 durations with the three most frequently used rst-to second-line chemotherapy sequences are shown in Supplementary Fig. S2.

Prognostic factors regarding OS in patients who received eribulin as the rst-or second-line chemotherapy
The univariate and multivariate analyses of prognostic factors in patients who were treated with eribulin as the rst-or second-line chemotherapy are shown in Table 2. Smoking, a higher lactate dehydrogenase (LDH) score, and a lower utility score were associated with poor prognosis in patients treated with eribulin as the rst-or second-line treatment. Adverse event-related discontinuation for each drug The proportions of adverse event-related discontinuations for eribulin, oral 5-FU, and anthracycline/taxane are shown in Fig. 5. Patients treated with eribulin or oral 5-FU as rst-line chemotherapy had signi cantly less discontinuation owing to adverse events than those treated with anthracyclines or taxanes (p = 0.038).

Discussion
To the best of our knowledge, this is the rst multicenter prospective cohort study to follow the sequences of chemotherapy, from the rst to third line, that were used for patients with AMBC in routine clinical practice in Japan. Most importantly, this study found that eribulin and oral 5-FU were as effective as anthracycline/taxane, which have long been considered the gold standard in rst-line chemotherapy for AMBC.
As the main purpose was to survey the actual use of chemotherapy and to investigate the effectiveness of eribulin in early line chemotherapy in Japan, all therapeutic agents were selected by shared decisionmaking that was based on patient-physician discussions. The study participants had quite diverse backgrounds and, among those, the patients who received eribulin as rst-line treatment had many poor prognostic factors. For example, in the ER-positive cohort, more patients treated with eribulin relapsed within 2 years; in the ER-negative cohort, more patients had liver and brain metastases than patients who were treated with the other regimens. The reasons for this trend may be as follows: 1) eribulin was more likely to be chosen for patients with early recurrence after adjuvant chemotherapy with anthracyclines or taxanes; 2) there might have been an expectation, based on the EMBRACE trial, that early use of eribulin would prolong OS [13]; and 3) metronomic therapy, such as oral 5-FU, was more likely to be chosen for patients with a slow tumor growth.
The frequent rst-and second-line chemotherapy sequences were taxane followed by eribulin (n = 33), oral 5-FU-based therapy followed by eribulin (n = 26), and eribulin followed by taxane (n = 21). It was suggested that the relatively high use of eribulin as rst-line (n = 46, 26.6%) and second-line (n = 70, 47.9%) chemotherapy was the result of shared decision making by the patients and treating physician in the hope of lowering chemotherapy-related toxicity. These treatment choices may have been based on the differences in the adverse event pro les, such as less frequent gastrointestinal symptoms or peripheral neuropathy, with regard to regimens other than oral 5-FU or taxanes, respectively. The time to the deterioration of global health status was longer for patients with eribulin use than for those with capecitabine use in triple-negative BC in Study 301 [27]. Similarly, the nding of a better health-related QOL in patients treated with S-1 than in those treated with taxanes as rst-line therapy in the SELECT-BC trial [10] might provide an impression to the patient and physician that oral 5-FU could lead to better QOL.
In this study, the lower proportion of adverse events and lower discontinuation rate of eribulin and oral 5-FU as rst-line therapy re ect the better feasibility of these two treatments. For the comparison between eribulin and S-1, further results of a randomized controlled trial (UMIN ID 000021398), (eribulin vs. S-1) with health-related QOL as the primary endpoint, which is currently being conducted in Japan, are awaited.
As rst-line treatment, eribulin was as effective as anthracycline/taxane and oral 5-FU for patients with ER-positive AMBC, whereas the prognosis in the eribulin cohort was worse for ER-negative patients.
However, as described above, there were poor prognostic factors in the patients treated with eribulin as rst-line chemotherapy, especially in the ER-negative cohort. Thus, the possibility of selection bias cannot be ignored even after adjusting for background variables, while considering that OS1 following eribulin in ER-negative patients was shorter than OS1 (2.99 years) that was previously reported in a phase II study [21], which investigated the e cacy of eribulin as rst-line chemotherapy. In contrast, OS1 for oral 5-FUs in this cohort was longer than the OS reported in a large, randomized controlled trial [10] that investigated rst-line chemotherapy in Japan (OS: 2.92 years in the S-1 group). This might indicate that oral 5-FUs were used for patients with less aggressive AMBC in this study. In contrast, OS2 or OS3 following eribulin was similar to those of two large randomized controlled trials (Study 301: 1.33 years, mainly in the second-or third-line chemotherapy; EMBRACE: 1.09 years, mainly in third-line or later chemotherapy) [13,14].
The results of the multivariate analysis of the data of patients who used eribulin as rst-or second-line chemotherapy demonstrated that higher LDH levels (≥ 300 U/L), smoking history, and lower utility scores were associated with shorter OS1. An elevated serum LDH level is known to be a worse prognostic factor in many cancer patients [28]. Tumor cell hypoxia leads to anaerobic glycolysis, which increases the LDH level and, consequently, the lactate level. In turn, lactate leads to local acidosis in the surrounding microenvironment and, as a result, tumor immunity and drug delivery are compromised [29]. Moreover, it has been suggested that smoking affects tumor immunity, such as through high in ltration of regulatory T-cells [30] and alteration of the phenotype of macrophages [31]. The abovementioned factors may diminish the e cacy of chemotherapy. Nevertheless, eribulin reportedly affects the improvement of hypoxia and causes EMT suppression, which is considered one of the reasons for OS prolongation [16][17][18]. Therefore, further large-sample studies are necessary to investigate this contradictory nding.
Miyoshi et al. reported that high baseline absolute lymphocyte counts (≥ 1500) are a predictive factor for eribulin use [32]; however, we found no signi cant interaction of eribulin with the lymphocyte count in the exploratory analysis of this study (data not shown). Therefore, no speci c predictive factors could be identi ed. However, it interesting to note that the utility score became a prognostic factor, and this suggests the importance of patient-reported outcome measures rather than physician-rated PS.
This study has some limitations. There were a limited number of participating institutions, and the sample size was small, leading to lack of power to discern confounding effects of other drugs used in sequence with eribulin. There was a possibility of channeling bias affecting the choice of treatment by the attending physician, as patients expected effectiveness and better feasibility with eribulin based on the results of pivotal studies. The survivor treatment selection bias that possibly exists in patients enrolled from second-line chemotherapy might have contributed to a favorable outcome for those patients. Furthermore, there were differences in the study treatment from the current standard of care because immune checkpoint inhibitors and PARP inhibitors were not approved for clinical use at that time in Japan. However, no other cohort study has followed the actual treatment sequence up to third-line chemotherapy, and this study provides valuable data on the actual status of chemotherapy for patients with AMBC.
In conclusion, we demonstrated the treatment sequences of rst-, second-, and third-line chemotherapy for HER2-negative AMBC in the real-world setting, and eribulin or oral 5-FU was commonly used for rstand second-line chemotherapy. The lower proportion of adverse events and lower discontinuation rate of eribulin and oral 5-FU in early line chemotherapy may be attributable to the better feasibility of these treatments. OS1 for eribulin was shorter in the ER-negative cohort, whereas there was no difference in OS1 between all three regimens in the ER-positive cohort. Eribulin as rst-line chemotherapy was used for patients with poor prognostic factors, whereas oral 5-FU was used for patients with less aggressive AMBC in this study. Therefore, the results of this study should be interpreted cautiously. The e cacy of eribulin in early line chemotherapy should be validated in randomized controlled trials. Figure 1 Flow chart of study participant selection. Abbreviation: ChT, chemotherapy Trends from rst-line to third-line chemotherapy. Taxanes included paclitaxel (PTX) ± bevacizumab (Bev), nab-paclitaxel (nab-PTX), and docetaxel. Oral 5-uorouracil (5-FU) included capecitabine (Cape), S-1, and tegafur/uracil (UFT). *PTX was administered in rst-line chemotherapy followed by PTX+Bev in secondline chemotherapy in four patients, whereas PTX+Bev was administered in rst-line chemotherapy followed by nab-PTX in second-line chemotherapy in one patient. **Nab-PTX second-line chemotherapy followed by PTX+Bev in third-line chemotherapy in two patients, whereas PTX+Bev in second-line chemotherapy was followed by nab-PTX as third-line chemotherapy in two patients. †S-1 in rst-line chemotherapy followed by Cape and UFT in second-line chemotherapy in four and one patients, respectively, whereas Cape in rst-line chemotherapy was followed by S-1 in second-line chemotherapy in one patient. † †S-1 in second-line chemotherapy was followed by Cape in third-line chemotherapy in two patients, whereas Cape in second-line chemotherapy was followed by S-1 in third-line chemotherapy in one patient.