NAC is one of the preferred treatment strategies for early breast cancer because it has various advantages including improved breast conservation and enhanced assessment of tumour response. However, several side effects are associated with neoadjuvant agents such as paclitaxel. We conducted this prospectively registered multicentre study with patient-relevant outcomes to explore whether eribulin followed by an anthracycline confers fewer side effects than paclitaxel, with emphasis on grade 1 or higher peripheral neuropathy. To our knowledge, this is the first prospective randomised controlled trial comparing eribulin and paclitaxel focusing on neuropathy. A significant difference in PSN was observed between eribulin and paclitaxel. However, the trial review committee of the JONIE group halted this trial because a similar study at MD Anderson had closed early due to lower pCR rates that crossed a futility stopping boundary in the eribulin group [21]. pCR rates in the current study indicated that the efficacy of eribulin was lower than that of paclitaxel, affecting prognosis.
Neuropathy should be evaluated if taxane is administered as the control because neuropathy accompanying taxanes causes severe long-term damage to quality of life, and there is currently no effective standard therapy [22, 23]. Thus, it is important to alleviate neuropathy from adjuvant chemotherapy, which occurs in certain patients, most of whom are survivors. Neuropathy was evaluated as one of several AEs in two previous trials, but to our knowledge there have been no prospective randomised controlled trials comparing neuropathy caused by eribulin with that of paclitaxel [14, 15].
In this study, the frequency and severity of neuropathy were significantly better in group E than group P. PNQ was evaluated for 4 years, and PSN was significantly better in group E. Some patients suffered from severe neuropathy (PNQ score 4 or 5; interfering with activities of daily living) in group P, which continued for several years, consistent with previous reports [24]. It remains unclear as to why some patients have long-term severe neuropathy in perioperative chemotherapy, but a previous report indicated that neuropathy induced by paclitaxel in Japanese breast cancer patients was associated with older age and a polymorphism in the ABCB1 transporter [25]. Therefore, paclitaxel alternatives should be considered, of which eribulin could be a candidate.
Our study showed a significant difference in the response rate on imaging but no significant difference in pCR between patients who received paclitaxel or eribulin (29.8% and 20.7%, respectively; P=0.289). This contrasts with the findings of the MD Anderson trial [21], in which pCR was 27% in the paclitaxel group and 5% in the eribulin group. These differences may be because of the low number of cases receiving eribulin in the MD Anderson trial [21]. Another study comparing eribulin with paclitaxel in combination with anthracycline was performed by the National Surgical Adjuvant Breast and Bowel Project (NSABP) [26]. Although the prognosis was not reported, they also revealed a lower pCR rate of 17% for eribulin compared with 26% for paclitaxel (not significant). Although there was no significant difference in the pCR rate between eribulin and paclitaxel followed by anthracycline, the efficacy for early breast cancer was commonly lower in the eribulin arm in the three trials. Nonetheless, the differences between the results of these previous studies and the current study are marked. One explanation is that they were conducted in different patient populations and the differences could be attributed to pharmacoethnicity, but there is currently no reported evidence of such differences for eribulin. Furthermore, while the US Food and Drug Administration endorses pCR as a surrogate endpoint for breast cancer to enable acceleration of drug approval, a systematic review of pCR as a surrogate endpoint for DFS and overall survival in randomised trials for neoadjuvant therapy of early breast cancer questioned its reliability and suggested that it was not robust as a primary endpoint [27]. Hence, the use of pCR to assess efficacy may be misleading, and a future study examining the effect of neoadjuvant eribulin on other equivalent endpoints should be considered. Thus, caution should be exercised when comparing pCR rates between studies.
Regarding prognosis, despite there being several more patients with nodal metastases in group E at baseline, the Kaplan–Meier curve was superior in group E in this study, although not significant. However, according to Lim et al., DFS and overall survival was better for paclitaxel, but this was also not significant [15]. Because the prognostic data was not sufficient, further investigations are necessary. Eribulin was reported to promote an antitumour immune response [28], and thus it could contribute to inconsistencies in tumour shrinkage and prognosis.
No significant differences between the two groups were reported by Abraham et al. regarding other AEs [14]. However, toxicity was greater in the eribulin arm, especially neutropenia, according to Lim et al. [15]. Haematological toxicity was also more frequent and more severe in group E than group P in this study, and several patients had febrile neutropenia, which must be treated seriously. Nonetheless, eribulin caused fewer allergic reactions, and thus may represent a treatment option for patients who cannot tolerate paclitaxel.
There are several limitations of this study. The number of enrolled patients did not reach the planned sample size, and thus the statistical interpretation of the data was limited. Furthermore, the trial was stopped early, and hence the analyses of the prespecified endpoints are underpowered. The evaluation of neuropathy is subjective and thus has the potential to introduce bias, particularly because the study was not blinded. In addition, eribulin induced more severe haematological events than paclitaxel, and therefore the choice to administer eribulin would need to be justified on a patient-by-patient basis in consideration of these effects. Nonetheless, a strength of this study is that there were notable differences in the AEs between eribulin and paclitaxel.