Tolerability of Eribulin and correlation between polymorphisms and neuropathy in an unselected population of female patients with metastatic breast cancer: results of the multicenter, single arm, phase IV PAINTER study

Metastatic breast cancer is an incurable disease, with a median survival ranging from 24 to 48 months, which varies according to biological characteristics, metastatic sites, patients’ age, etc. [1, 2]. Treatments are tailored based patients and tumor characteristics, and the use of new drugs, such as CDK4/6 inhibitors or monoclonal antibodies significantly improved disease control [3, 4]. Despite the development of new agents, single-agent chemotherapy remains an important backbone in the metastatic setting. Capecitabine, vinorelbine and eribulin are the preferred options in patients who have already been treated with anthracyclines and taxanes [2].

Eribulin is a synthetic analog of the natural product halichondrin B. Its cytotoxic effects are mainly due to its ability to interfere with microtubule dynamics by causing the blockage of mitotic spindle formation, mitotic arrest and subsequent cell death by apoptosis [5]. Differently from other microtubule interfering agents, eribulin inhibits the growth phase of the microtubules without any inhibition of the shortening phase [6‐8]. However, many other non-mitotic effects of eribulin on tumor biology have been described, including tumor vascular remodeling, which leads to better tumor perfusion and reduced hypoxia, and interference with epithelial mesenchymal transition which reduces the ability of tumor cells to migrate and invade, both in vitro and in vivo. These latter effects of eribulin seem to be relevant in justifying the clinical activity observed with this drug [9].

Eribulin has been approved by the Food and Drug Administration in 2010 and in Italy in 2012 for the treatment of patients with locally advanced or metastatic breast cancer which has progressed after at least one chemotherapy regimen, including a combination of anthracyclines and taxanes in both the adjuvant and metastatic setting.

The approval of this drug in breast cancer was based on the results of two randomized phase III trials [10, 11]. The first trial EMBRACE [10] enrolled 762 women with locally recurrent or MBC randomly assigned (2:1) to eribulin or a treatment of physician’s choice (TPC). Overall survival was significantly improved in the experimental arm (median 13.1 months, 95% CI 11.8–14.3) compared to TPC arm (10.6 months, 9.3–12.5; hazard ratio 0.81, 95% CI 0.66–0.99; p = 0.041). The second trial, the 301 study [11], evaluated the efficacy and safety of eribulin as first, second or third line monotherapy versus capecitabine in 1102 patients with locally advanced breast cancer or MBC who had been previously treated with anthracyclines and taxanes. The median OS was 15.9 months for eribulin versus 14.5 months for capecitabine (HR 0.879; 95% CI: 0.770–1.003; p = 0.056), and the study failed to demonstrate the superiority of eribulin. A pooled analysis suggested a major benefit in the subgroup of women with HER2 negative disease [12]. Results from phase III trials suggested that eribulin was well tolerated and the most common adverse effects (AEs) were neutropenia, fatigue and peripheral neuropathy. Specifically, this latter AE occurred in 5% of patients enrolled in the EMBRACE study and in 13% of patients enrolled in the 301 study. In a recent meta-analysis, the incidence of all-grade and high-grade peripheral neuropathy after treatment with Eribulin was 27.5% (95% CI: 23.3–32.4%) and 4.7% (95% CI: 3.6–6.2%), respectively [13]. Even though the precise mechanism behind the neurotoxicity caused by microtubule interfering agents (taxanes, vinca alkaloids and eribulin) has not been fully defined yet, preclinical and histological studies suggest that its pathogenesis is mainly a consequence of the interruption of the axonal transport within the neuron, which relies on intact microtubule structures [14, 15]. The prevalence of severe neuropathy is extremely variable, suggesting that individual characteristics might affect susceptibility. For this reason, the study of gene polymorphisms could help in identifying patients at a higher risk of developing neurotoxicity, as has already been demonstrated for taxanes [14, 16].

In this paper we report the results of the PAINTER study which aimed to evaluate the toxicity and quality of life (QoL) in unselected Italian patients with MBC treated with eribulin. The correlation between neurotoxicity and SNPs was also explored.

Starting from May 2014 until June 2018, 180 patients were enrolled in this study by 20 different Italian centers. As depicted in the Flowchart (Fig. 1), 10 patients were excluded from the safety analysis: 2 patients due to major protocol violations (having received eribulin in previous lines of treatment) and 8 patients because they never started treatment with eribulin. Only 159 patients out of 180 were included in the analysis of polymorphisms (molecular analysis set), since the blood samples of 11 patients were not available. Ninety-eight patients (54.4%) were considered for QoL assessment. The median follow-up was 15.4 months.

The main baseline characteristics of patients, according to the analysis sets, are summarized in Table 1. Overall, mean patient age was 59.7, 64% had a Luminal HER-2 negative metastatic breast cancer, ECOG Performance status was 0–1 in 100% of the patients, and the mean number of previous chemotherapy lines for metastatic disease was 5 (from 0 to a maximum of 18). Previous neuropathy was reported in 15.9% and 17% of patients in the safety and molecular analysis sets, respectively.

The median number of eribulin cycles administered per patient was 4.5 (first quartile[Q1]- third quartile[Q3] 3.0–7.0) from a minimum of 1 to a maximum of 23 cycles. Half of the patients received 84.7% of the cycles at the full dose but the treatment was modified for 83 patients (48.8%), mostly before the 3^rd cycle. The main reasons for treatment discontinuation were disease progression (82.9%), loss to follow up (4.7%), medical decision (4.7%) and toxicity (3.5%).

The treatment of MBC is a challenge in oncology and for several years no advances in overall survival have been observed [26]. New targeted therapies which have become available are changing the natural history of MBC, and new goals have been achieved [3, 27, 28]. Among the different types of chemotherapy, which remain a backbone in this setting, eribulin has demonstrated an improvement in OS in patients with Her2-negative MBC after treatment with anthracyclines/taxanes [10]. As stated in the International Consensus Conference on advanced breast cancer, preserving QoL and avoiding treatment related adverse events are important issues for patients with MBC, since it is a chronic and lethal disease [2]. Eribulin represents a therapeutic option for MBC and as its clinical use will increase with time, a better knowledge of its safety profile outside of clinical trials is warranted. Several retrospective studies have been published on the safety of eribulin in a real life setting [29‐31]; however, few prospective studies, aimed at thoroughly evaluating its safety and efficacy in daily clinical practice are available [32, 33]. The PAINTER study was designed to investigate the tolerability of eribulin in a real life setting and to study the association between the onset of neurotoxicity and specific gene polymorphisms. The trial enrolled pretreated MBC patients with characteristics typical of this population: a mean age of 59 years old, 5 median lines of previous chemotherapy for the metastatic setting, up to 18 cycles and mainly Luminal B Her2-negative disease (65%). In the PAINTER study, women received a median of 4,5 cycles of eribulin (range 1–23), and this data are slightly higher than those reported in the EMBRACE and in the TROTTER studies, where patients received 4 lines (range 1–7) and 3 lines (1–10), respectively.

Regarding efficacy, the median OS was 12 months (Q1-Q3: 6.4–21.7 months), in some ways similar to other studies: in EMBRACE the median OS was 13.1 months (95% CI 11.8–14.3) and in the pooled analysis of the 2 phase III studies, the median OS in the ITT population was 15.2 months [12]. The Painter OS is in line with what was observed in real life studies: an OS of 10.1 months (95% CI: 8.1–13), 11.6 months (0.6–33.3 months; 95% CI 8.7–14.5) and 13.53 months (95% CI9.39–17.67 months) [29‐31]. Interestingly in the VESPRY study the median OS was 31.8 months (CI 95% 27.9–34.4) and as the authors reported, compared with both these pivotal trials [29‐31], there was almost a 2.5 fold increase in OS.This clinical benefit can be partially explained by the fact that the patients who received eribulin were not heavily pretreated, as all patients were in the third line of treatment [34]. Also, Gamucci et al., who reported a median OS of 14.3 months (95%CI, 11.7–16.8) pointed out that a significant improvement in response was observed when eribulin was given as third-line treatment (p = 0.02) [35].

Our study confirms that, even in a heavily pretreated setting, eribulin is well tolerated in fact only 3.5% of patients discontinued treatment because of toxicity /severe adverse events. Most adverse events occur within the 6^th cycle; in fact, the risk of interrupting or reducing the dose of eribulin is higher as in the first cycles than in subsequent ones. This is why some patients could continue the treatment for a very long time, up to 23 cycles. Among other G1–G4 toxicities it should be noted that osteo-muscular, abdominal and in tumor site pain was observed in 20% of patients. While this side effect has been described after treatment with other drugs [36], it has never been reported after eribulin treatment.

Most of the enrolled patients had been previously treated with taxanes (97%), a well known neurotoxic drug, similarly to other studies [10, 35]. Interestingly, in our study, 15% of patients reported having neurotoxicity before starting eribulin and this was related to previous treatments; this data are often unreported, but it is very important and must be considered when a patient is evaluated for treatment with eribulin.

Regarding severe toxicity, 38.2% patients experienced at least one severe adverse event; in particular neutropenia in 15.3%of patients, neurotoxicity in 14.7% and asthenia in 5.9%. In our opinion, neuropathy is probably under-reported in other studies [31, 33], while in the PAINTER study the incidence of neurotoxicity was accurately reported thanks to the way the information was collected. In fact, at each cycle a specific question was asked regarding the most frequent toxicities including neurotoxicity. Another possible explanation involves the characteristics of the study population, such as the starting dose of eribulin, previous radiotherapy, and hemoglobin levels at baseline, as identified by Tsurutani et al. as significantly associated with peripheral neuropathy [37]. Anyway, it is clear that eribulin causes peripheral neuropathy; Zhao et al. demonstrated that eribulin-treated subjects (both with breast cancer and liposarcoma) had a significantly increased risk of all-grade (RR, 2.00; 95% CI, 1.70–2.35; p = 0.008) and high-grade (RR, 3.68; 95% CI, 2.30–5.89; p < 0.001) neurotoxicity [38].

A part of the PAINTER study was dedicated to exploring the role of polymorphisms and their relationship with neuropathy. Fifteen SNPs were analyzed and 2 polymorphisms, rs2233335 (T/T) in the NDRG1 gene and rs7214723 (T/T) in the CAMKK1 gene, were associated with eribulin-induced severe neurotoxicity. Regarding rs7001034 in the FZD3 gene and rs242557 in the MAPT gene, an association with neurotoxicity was observed, in particular allele A increased the occurrence of ineuropathy, although it was not confirmed for severe neuropathy. The functions of the involved genes are different; NDRG1 (N-myc downstream regulated gene-1) is a stress response protein involved in multiple cellular pathways, including the endoplasmic reticulum stress response; CAMKK1 (calcium/calmodulin-dependent protein kinase kinase-1) is a serine/threonine kinase that is activated by an increase in intracellular Ca2 + levels and Ca2 + /calmodulin binding; FZD3 is a member of the frizzled gene family, which has been shown to play a role in neurite outgrowth and nerve development; finally, MAPT (microtubule associated protein Tau) is involved in tubulin assembly and polymerization [39‐42]. The different mechanisms that underlie the onset of neuropathy are unknown and many studies are exploring them, even in new drugs, like nab-paclitaxel compared to paclitaxel. This suggests that the selection of the nab-paclitaxel regimen should be individualized based on the clinical context and potentially supported by pharmacogenetic analysis [43].

After three cycles of eribulin, QoL questionnaires showed that the global health status was overall preserved, including social, physical and emotional roles. Patients reported a worsening of fatigue and nausea induced by treatment, while no differences were found in the other 7 symptoms and items investigated. In the metastatic setting, the preservation of a good quality of life is essential and our data show that treatment with eribulin allows patients to maintain good health. The global worsening of health status and symptoms at the end of treatment evaluation, was indeed observed at disease progression.

The PAINTER study offers a wide spectrum of information on the tolerability of eribulin. In fact, the prospective nature of the study allowed the investigators to collect many items based on previous studies. As expected, fatigue, peripheral neuropathy and neutropenia were the most common toxicities, but few patients experienced severe AEs. Toxicities rarely led to drug discontinuation, even if schedule and dosage modifications were common, as can be frequently observed in heavily pretreated patients. Patients reported outcomes show the preservation of a satisfactory global health status during the first three cycles of eribulin and a progressive worsening of symptoms at the end of treatment. To the best of our knowledge, this is the first study that demonstrates a correlation between SNPs and neuropathy in patients treated with eribulin. Ongoing studies are trying to understand the molecular mechanisms at the basis of this association. These data, if confirmed by other studies, will allow a tailored treatment with eribulin, addressing the proper use of the drug while avoiding useless toxicity.