Introduction

Metastatic breast cancer (MBC) remains a largely incurable disease causing more than 40,000 deaths in the US alone every year [1]. Taxanes (paclitaxel and docetaxel) have similar efficacy as anthracyclines (doxorubicin and epirubicin) without cross-resistance, thus expanding the therapeutic strategies available for treatment of breast cancer [2–4]. Due to their different mechanisms of action, studies combining anthracyclines and taxanes have been successful in prolonging treatment-associated outcomes in patients with metastatic disease [5–8]. Drug resistance (both primary and acquired) is a major problem associated with taxane therapy, either following metastatic or adjuvant treatment [9, 10]. Efficacy of available second and subsequent lines of therapy after anthracyclines and taxanes is poor with declining response rates (10–30%) and overall survival durations between 6 and 12 months [11]. With increasing use of anthracyclines and taxanes for early breast cancer, fewer effective options are available for patients with metastatic disease [10, 12]. The reduced effectiveness of taxanes for patients with disease relapse has led to the development of new microtubule stabilizing agents (MTSA) as antineoplastic drugs with low susceptibility to drug resistance mechanisms.

The epothilones and their analogs are a new class of MTSA anticancer drugs that are active against multidrug- resistant (MDR) cell lines and tumors [13]. Although they stabilize microtubules in a manner similar to taxanes, they are structurally different, and bind to tubulin in a distinct manner that confers sensitivity in taxane-resistant models. Ixabepilone is a semi-synthetic analog of epothilone B demonstrating particularly high anti-microtubule activity and low susceptibility to drug resistance mechanisms, such as over-expression of efflux transporters (e.g., P-glycoprotein and multidrug-resistance protein-1) and class-III isoform of beta-tubulin [14, 15]. It has shown notable efficacy in several phase II trials in patients with locally advanced and metastatic breast cancer. The efficacy achieved with ixabepilone monotherapy in the metastatic setting has varied with the degree of patient pretreatment [16–21]. Besides studying ixabepilone as a single agent, a large randomized phase III trial (CA163-046) compared ixabepilone plus capecitabine to capecitabine alone in patients with locally advanced (stage IIIB) or metastatic (stage IV) anthracycline- and taxane-resistant BC [22]. This study demonstrated the superiority of ixabepilone combination therapy over capecitabine monotherapy, with a 25% reduction in the risk of disease progression, the primary endpoint, (median PFS, 5.8 vs 4.2 months; hazard ratio, 0.75; P = .0003) and a significantly improved ORR (35% vs 14%; P < .0001) [22]. The US Food and Drug Administration requested a sensitivity analysis for PFS, censoring patients receiving subsequent therapy before the date of progression, which indicated a 31% reduction in the risk of disease progression (HR: 0.69 (95%CI, 0.58 to 0.83; P < .0001) [23]. Based on the activity in this trial, ixabepilone is approved in the US in combination with capecitabine for treatment of metastatic or locally advanced breast cancer resistant to an anthracycline and a taxane.

Here, we present an updated report describing the results of overall survival from the CA163-046 phase III study.

Patients and methods

Patients

In this large multinational phase III trial, 752 patients with MBC who were anthracycline pretreated and met strict predefined resistance criteria to anthracyclines and taxanes, were enrolled between September 2003 and January 2006. Anthracycline and taxane resistance was defined as tumor progression during treatment or within 3 months (amended to 4 months for taxane) of last dose in the metastatic setting, or recurrence within 6 months (amended to12 months for taxane) in the neoadjuvant or adjuvant setting; patients with tumors not resistant to anthracyclines were also eligible if they received a minimum cumulative dose of doxorubicin of 240 mg/m2 or epirubicin of 360 mg/m2. These patients were randomized to receive either ixabepilone (40 mg/m2 IV over 3 h Q3w) + capecitabine (1,000 mg/m2 PO BID 1–14 day), or capecitabine (1,250 mg/m2 PO BID 1–14 day) alone. Crossover from capecitabine alone to combination therapy was not permitted. Eligibility and pretreatment evaluation have previously been reported in detail [22].

In brief, study eligibility requirements included women ≥18 years of age with a Karnofsky performance score of 70 to 100 and life expectancy ≥12 weeks. They were allowed to receive up to 3 prior chemotherapy regimens in any setting, with sequential neoadjuvant/adjuvant treatment counting as 1 regimen. Key exclusion criteria included brain metastases, neuropathy grade ≥2, reduced hepatic function, hematologic and renal functions, prior severe hypersensitivity to agents containing polyethoxylated castor oil or hypersensitivity to fluoropyrimidine, and continued treatment with potent cytochrome P450 3A4 inhibitors.

Study design

All patients were randomly assigned to receive ixabepilone plus capecitabine or capecitabine monotherapy. Efficacy (ORR, PFS, and OS) was evaluated on all randomized patients, while safety was evaluated for all treated patients. The primary endpoint of the study was an intent-to-treat analysis of progression-free survival (PFS) defined as the time from randomization to progressive disease (as assessed by an IRRC) or death. Overall survival (OS) was one of the key secondary endpoints. The study had 80% power to detect a hazard ratio of 0.8 using a two-sided 0.05 level log-rank test. This analysis was planned when at least 84% of the patients had died.

Statistical analysis

The analyses of OS were conducted on all randomized patients on an intent-to-treat basis.

The primary OS analysis was stratified by the following factors assigned at randomization: presence of visceral metastases in liver and/or lung, minimum of either doxorubicin 240 mg/m2 or epirubicin 360 mg/m2 and relapse >6 months in adjuvant setting and prior chemotherapy for metastatic disease. A survival distribution for each group was estimated using the Kaplan–Meier product-limit method. A two sided 95% confidence interval (CI) for the median survival was computed by the Brookmeyer and Crowley method. The hazard ratio and its 95% CI of ixabepilone plus capecitabine to capecitabine alone were estimated using a Cox proportional hazards model, stratified by the above factors. A multivariate Cox model was used to estimate the effect of treatment on OS adjusted for pre-specified baseline prognostic factors. Predefined subset analysis of OS was conducted for the randomly assigned population based on potential prognostic factors, some of which are listed in Fig. 3. These analyses were not adjusted for multiple comparisons and are considered exploratory.

Results

Patient population

Patients (median age 53 years) were randomized to receive either ixabepilone plus capecitabine (N = 375) or capecitabine alone (N = 377) (Fig. 1). Patient characteristics were well balanced between the two arms of the study at entry [22] and shown in Table 1. Briefly, 65% of patients had ≥3 metastatic disease sites, and 84% had visceral disease involving the liver and/or lung. An additional 6 patients (1.6%) in the ixabepilone plus capecitabine group and 12 (3.3%) patients in the capecitabine group were still on treatment following the primary database lock.

Fig. 1
figure 1

CONSORT flowchart of CA163-046

Full size image
Table 1 Baseline patient demographics and disease characteristics (adapted from Thomas et al. [22])
Full size table

Efficacy

Overall survival: clinical outcome

A total of 318 (84.8%) patients in the ixabepilone plus capecitabine group and 321 (85.1%) patients in the capecitabine group died by the time of the final analysis. Patients receiving ixabepilone plus capecitabine treatment had a median overall survival of 12.9 months (95%CI: 11.5–14.2) compared to 11.1 months (95%CI: 10.0–12.5) for patients receiving capecitabine alone (HR = 0.9; 95%CI: 077–1.05; P = 0.1936) (Table 2). A separation of the curves between 3 and 24 months of follow-up was observed (Fig. 2). The difference in median overall survival (a 16% increase) favored the combination; however, this difference did not reach statistical significance.

Fig. 2
figure 2

Kaplan–Meier Analysis of overall survival showing median OS (with 95% CI) in each treatment arm

Full size image
Table 2 Analysis of overall survival in patients receiving ixabepilone plus capecitabine vs capecitabine alone
Full size table

A pre-specified secondary analysis using a Cox proportional hazards model was carried out to assess the association of potential prognostic factors with overall survival and to adjust the treatment comparisons for those factors. The estimated HR of ixabepilone plus capecitabine over capecitabine alone was 0.87; 95%CI: 0.74–1.02; P = 0.0803. Factors that were statistically significant from the Cox model in predicting overall survival time were Karnofsky performance status (KPS) (HR 0.63, P = < 0.0001), number of disease sites (HR 0.72, P = 0.0001), estrogen receptor status (HR 1.45, P = < 0.0001), moderate/severe liver dysfunction (HR 0.56, P = 0.0004), and time from diagnosis to randomization (HR 0.79, P = 0.0083).

Approximately 74% of patients in each treatment group received subsequent therapy (chemotherapy, hormonal/immuno/biologic or radiotherapy) following termination of the treatment period; 61% in the combination group received subsequent chemotherapy compared to 64% in the capecitabine group. This imbalance was primarily reflected in subsequent paclitaxel (9 vs. 14%) and docetaxel use (5 vs. 8%) between the treatment arms; however, a post-hoc analysis of survival, censoring the patients at the start of their subsequent taxane therapy did not show a statistically significant difference in the OS (data not shown).

Predefined subset analyses of OS showed consistency with the primary analysis (Fig. 3). The observed HR of OS for ixabepilone plus capecitabine over capecitabine was <1 for nearly all subsets explored; the 95%CI included 1 for all subsets except for patients with KPS 70–80. In these patients, median OS was 10.1 and 7.8 months, respectively, and the HR was 0.75 (95% CI: 0.58, 0.98). In patients with KPS 90–100, median OS was 14.1 months in each group and the HR was 1.01 (95% CI: 0.83, 1.22).

Fig. 3
figure 3

Subset analysis of overall survival shows the overall survival hazard ratios (with 95% CI) for subgroups of patients from the study. Hazard ratios less than 1.00 favor the combination therapy. ER, estrogen receptor, PR, progesterone receptor, HER-2, human epidermal growth factor receptor-2

Full size image

Safety

As observed in the previous report of ixabepilone combination treatment in patients with MBC resistant to anthracyclines and taxanes, treatment-related adverse events were mostly grade 1/2 and generally reversible; the toxicity profile of ixabepilone plus capecitabine combination therapy reflected that of the individual agents. Thirty-three (9%) patients receiving combination therapy died within 30 days of last dose (from any causes), unchanged from the previous report. Table 3 summarizes the incidences of key treatment-related adverse events from patients classified by the treatment arms.

Table 3 Most common treatment-related adverse events and hematologic abnormalities (adapted from Thomas et al. [22])
Full size table

Discussion

Anthracyclines and taxanes are the standard of care in the treatment of breast cancer, both in the locally advanced and in the metastatic setting. Unfortunately, patients who develop progressive disease on anthracycline and taxane therapy have limited proven treatment options. Until recently, capecitabine was the only agent widely approved for this patient population in the US, although phase III studies have not been conducted to determine whether capecitabine achieves a survival advantage for patients with MBC resistant to an A and a T [24, 25].

The primary objective of CA163-046 was to compare PFS benefits of ixabepilone plus capecitabine treatment to capecitabine alone in patients with advanced breast cancer resistant to an anthracycline and taxane. Ixabepilone plus capecitabine demonstrated superior PFS compared to capecitabine alone [22, 23]. Secondary endpoint of ORR also showed significant improvement [22]. Here, we report the results of a prespecified analysis of OS, a secondary efficacy endpoint of the study. We also report the final update of safety since 18 patients were still on treatment at the time of the first publication.

An observed difference in median overall survival (16%) favoring ixabepilone plus capecitabine arm did not reach statistical significance. The median OS for patients in the combination arm was 12.9 months compared to 11.1 months for patients receiving capecitabine only (HR = 0.9; 95%CI: 077–1.05; P = 0.1936). These overall survival results are consistent with those observed in another Phase III study (CA163-048) investigating the effect of ixabepilone plus capecitabine on overall survival in patients with MBC previously treated with or resistant to anthracyclines and taxanes [26].

The impact of the ixabepilone combination on overall survival is similar to that seen in other phase III studies conducted in similar patient populations with metastatic breast cancer previously treated with anthracyclines and taxanes. Although many of the new studies demonstrated statistically significant improvement in response rates or PFS, none demonstrated a statistically significant survival advantage [27–31]. Significant OS benefit has not been seen for any second line, and most first line therapies in MBC trials [32, 33]. A likely reason could be the availability of other treatment options following investigational therapy that exhibit modest efficacy when administered in a sequential fashion.

A clinical benefit was observed in a subset of the total population of our study in symptomatic patients (KPS 70–80). In these patients, an improvement in OS of 2.3 months was seen; a similar result also observed in CA163-048, comparing the same combination to capecitabine alone in women with metastatic breast cancer previously treated with an anthracycline and a taxane [26]. The observed survival advantage seen in this subpopulation classified as the KPS 70–80 from both the phase III studies is the first documented OS advantage in such patients with metastatic breast cancer resistant to or pretreated with anthracyclines and taxane. Patients with suboptimal performance status have, in general, a shorter OS, and therefore, a reduced possibility to avail themselves from other therapies that could influence their survival. In addition, such patients usually have a larger tumor burden, and most systemic treatments appear less effective in the face of larger tumor burden. Since combination therapy produces a higher response rate, it might reduce tumor burden to a greater extent and might allow the use of additional therapies post protocol treatment. The efficacy of ixabepilone in this subpopulation could be used as a starting point for future studies aimed at these patients.

Overall survival versus progression-free survival as meaningful endpoints in MBC

Metastatic breast cancer is increasingly perceived as a chronic disease. However, it remains largely incurable; the current goals of therapy are to ameliorate symptoms, delay disease progression, improve or maintain quality of life (QoL), and prolong overall survival with as little toxicity as possible. The median survival from first evidence of metastasis has increased from 12–18 months to 30–40 months, with 5-year survival rates exceeding 25% on population-based analyses [34–36]. In these patients who respond to subsequent treatment, therapy is generally continued until there is evidence of disease progression or intolerable toxicity [37]. Response to therapy is typically evaluated after 8–12 weeks of treatment. Failure to achieve any response by that time predicts a low likelihood of further response.

Continued treatment in the absence of progression (as opposed to a preset number of treatments followed by re-induction at the time of progression) is associated with longer duration of PFS, and in a Cochran meta-analysis, with prolongation of survival [38]. Comparative studies of continuous chemotherapy approach (chemotherapy until progression) versus interrupted therapy (stopping therapy after an induction period and resuming at the time of progression) show improved TTP with the continuous approach but with more toxicity and no difference in overall survival [39]. Also, continuous therapy approach was associated with better quality of life in one randomized trial [40].

Short remissions and frequent treatment changes are associated with increased anxiety; most patients prefer (in the absence of serious toxicity and side effects) to remain on a single treatment with longer control of the disease.

Overall survival as an endpoint is easy to measure, unambiguous and of great clinical relevance, and for any antineoplastic agent can be demonstrated through a randomized controlled trial. However, use of overall survival as a primary endpoint for a clinical trial has several major drawbacks [31, 41]. Advent of newer therapeutic options has prolonged OS substantially for patients with metastatic breast cancer [42]. Mortality occurs after a relatively long time for most patients; hence, reliable quantitations of statistically significant differences in OS require large numbers of patients and several years. In a recently reported trial where women with MBC were randomized to receive either bevacizumab plus paclitaxel as first line or paclitaxel alone, PFS endpoint was reached within 2 years in contrast to 8 years for overall survival [30, 43]. In addition, in many clinical trials with new drugs, patients assigned to the control arm are either allowed to cross over to receive the investigational agent or receive the drug off-study if they progress. This, unfortunately, dilutes the effect of the investigational agent on OS, and in many instances ethical considerations prevent prohibition of crossover in clinical trials. However, in study 046, patients assigned to capecitabine monotherapy were not allowed to crossover to the combination arm.

With the availability of multiple active drugs and drug regimens, many patients receive multiple sequential treatment regimens between the detection of first metastasis and death. The only way OS could be a true reflection of the activity of a drug is if all subsequent therapies were applied equally to both arms of a trial, something that is virtually impossible today. Increasingly, each episode of treatment represents a smaller fraction of the overall survival duration of a patient. In consequence, even moderate prolongation in PFS with one regimen (50% or even longer) may have minimal impact on the overall survival, and most published studies in metastatic breast cancer are underpowered to detect such modest differences in survival.

Since the assessment of true survival benefits from chemotherapy in MBC can be difficult to obtain, there is a need to develop a surrogate primary endpoint which may deliver definitive results. PFS measures the benefit of the exploratory therapy alone with limited confounding by other interventions. PFS is defined as the time from randomization to objective tumor progression or death; it may be a preferred regulatory endpoint since it includes death. In disease settings where limited accrual to trials is anticipated and follow-up is very long for OS, drugs have been approved on the basis of demonstration of a longer PFS duration in randomized trials. For a drug to receive approval with this end point, the results must be shown convincingly in a randomized trial with the use of an independent, blinded review committee for adjudication, thus limiting the potential biases associated with use of PFS. The observed difference in PFS has to be of clinical benefit and the results must be statistically robust. There must be a minimum of censoring due to patient dropouts and missing data. Approval of ixabepilone by FDA as a combination therapy in patients with metastatic disease resistant to anthracyclines and taxanes was granted based on prolongation of PFS [22]. The study was carefully designed to show a difference in PFS between capecitabine monotherapy and combination therapy. The primary PFS end point was properly defined as radiologic progression or death from any cause. Progression was determined by an independent radiologic review committee, blinded to the identity of the treatment group, which is an important consideration in reducing bias. The primary end point was met with convincing statistical significance and the secondary end points (investigator assessment of PFS and ORR) also supported the primary finding. In addition, all sensitivity analyses done by either the sponsor or the FDA supported the finding of an advantage in PFS, suggesting the robustness of the result across the entire population studied and not driven by a particular treatment group. In addition to ixabepilone approval, recent approvals of bevacizumab (Avastin), trastuzumab (Herceptin), and lapatinib (Tykerb) based on positive PFS with no OS benefit is an indication of FDA’s acceptance of PFS as a registrational endpoint [41].

In conclusion, it can be said that OS remains the most desirable endpoint to achieve for patients with metastatic disease. However, ease and reliability of the measure of PFS as an efficacy endpoint makes it an appropriate choice as a primary endpoint regardless of the surrogacy for OS.

This study convincingly demonstrated that ixabepilone is a new and effective agent which, in combination with capecitabine is more efficacious than capecitabine monotherapy in patients with metastatic disease pretreated or resistant to anthracyclines and taxanes. Ixabepilone needs to find its optimal place in the management of primary and metastatic breast cancer.