Phase IIIb study of the bevacizumab biosimilar candidate BI 695502 plus mFOLFOX6 in metastatic colorectal cancer

Globally, colorectal cancer (CRC) is the third most common cancer in men and the second most common cancer in women, with over 1.8 million new cases reported in 2018 [1]. VEGF plays a key role in tumor growth and metastasis through the promotion of angiogenesis and vascularization [2]. VEGF promotes tumor angiogenesis through several mechanisms, including enhanced endothelial cell proliferation and survival, increased migration and invasion of endothelial cells, and increased permeability of existing vessels [3]. Bevacizumab is a recombinant humanized monoclonal antibody directed against VEGF [4], which was licensed internationally for first- and second-line treatment of metastatic CRC (mCRC) in 2004 [5] and 2005 [5,6]. In preclinical studies, administration of bevacizumab to xenotransplant models of colon cancer in nude (athymic) mice caused reduction of microvascular growth and inhibition of metastatic disease progression [5].

Three pivotal clinical trials support the use of bevacizumab as a treatment of mCRC. The first of these studies demonstrated that the addition of bevacizumab to irinotecan, bolus fluorouracil, and leucovorin (IFL) significantly improved progression-free survival (PFS), overall survival (OS), and response rates in patients with previously untreated mCRC versus placebo plus IFL [7]. In the second study, bevacizumab plus oxaliplatin-based chemotherapy significantly improved PFS compared with oxaliplatin-based chemotherapy alone in patients with mCRC [8]. In the third pivotal study (BEAT), use of bevacizumab plus chemotherapy in routine clinical practice produced a median PFS of 10.8 months and median OS of 22.7 months [9]. Numerous other studies have provided evidence of the benefits of administering bevacizumab, as either a first- or second-line treatment, in combination with chemotherapy [4,7–19]. 5-Fluorouracil, leucovorin, and oxaliplatin (mFOLFOX) is an established chemotherapy for CRC and is recommended in both European and United States treatment guidelines [20,21]. Several of the bevacizumab studies have shown that the drug is effective when combined with FOLFOX as first-line treatment of mCRC [9,15–19,22,23].

Key patents relating to bevacizumab reference product (RP) expired in the US in 2019 and will expire in the European Union (EU) in 2022, introducing potential opportunities for biosimilar products to reduce the costs associated with anti-VEGF therapy. Several bevacizumab biosimilar products are in development, including ABP 215 (Mvasi^®; Amgen), which has been approved in the US and the EU [24,25], and PF 06439535 (Zirabev^®; Pfizer), which has been approved in the EU [26]. BI 695502 was developed as a bevacizumab biosimilar candidate, and in a Phase I trial, three-way bioequivalence was demonstrated between BI 695502, US-licensed bevacizumab RP, and EU-approved bevacizumab RP in healthy male subjects [27].

Here, we report a study that was conducted to assess the safety, efficacy, pharmacokinetics, and immunogenicity of BI 695502 in combination with mFOLFOX6 in patients with previously untreated mCRC.

Materials & methods

Study design & treatment

This Phase IIIb, multicenter, single-arm, open-label trial of BI 695502 was conducted in the USA, Japan, Spain and Ukraine. Study treatment was administered in 14-day cycles. On day 1 of each cycle, BI 695502 5 mg/kg was administered by intravenous (IV) infusion. This was followed by mFOLFOX6 chemotherapy, comprising IV infusion of oxaliplatin 85 mg/m² and leucovorin 400 mg/m² (both administered over 2 h on day 1), and 5-fluorouracil (5-FU; IV bolus of 400 mg/m² on day 1, then IV infusion of 2400 mg/m² over 46–48 h [days 1–2]). If oxaliplatin was stopped during the study, patients continued to receive 5-FU + leucovorin with BI 695502 until disease progression. Treatment was continued until disease progression (according to Response Evaluation Criteria in Solid Tumors [RECIST] version 1.1), death, unacceptable toxicity, or end of study. Patients were considered as having completed treatment if they received ≥8 cycles.

Clinic visits were scheduled for the administration of study treatment and for follow-up 30 days and 18 weeks after the last dose of BI 695502. Patients were subsequently monitored every 3 months until death or end of study.

Due to a manufacturing issue (particles detected in a batch of BI 695502), from 21 December 2017 onward the remaining patients were switched from BI 695502 to commercially available bevacizumab RP (Avastin^®; Roche, Welwyn Garden City, UK; Genentech, CA, USA). All data presented in this manuscript relate to treatment administered before this switch was implemented.

Patients

Previously untreated patients aged ≥18 years (≥20 years in Japan), with histologically confirmed mCRC that was not amenable to surgery, were eligible to enter the study. All study participants were required to have ≥1 measurable lesion (RECIST 1.1), an Eastern Cooperative Oncology Group (ECOG) performance score ≤1, and investigator-assessed life expectancy of ≥12 months. Key exclusion criteria were prior systemic therapy for metastatic disease, prior treatment with monoclonal antibodies or small-molecule inhibitors of VEGF, previous malignancy other than CRC in the last 5 years, thrombotic or hemorrhagic event during the previous 6 months, and a history of poorly controlled hypertension. A full list of inclusion and exclusion criteria is provided in the Supplementary Data.

End points

The primary study objective was to evaluate the safety and tolerability of BI 695502 in combination with mFOLFOX6 and as maintenance therapy (when applicable). The primary end point was the percentage of patients with any of the following prespecified adverse events (AEs): anaphylactic, hypersensitivity or infusion-related reactions; arterial or venous thromboembolic events; gastrointestinal (GI) perforations; hypertension; proteinuria; hemorrhage; wound-healing complications including abscess and fistulas; posterior reversible encephalopathy syndrome; or ovarian failure. This list was designed to encompass the principal safety considerations arising from clinical experience with bevacizumab RP. Incidence rates for all individual AEs were determined, with AEs being coded according to the Medical Dictionary for Regulatory Activities (MedDRA, version 20). The investigators determined whether each AE was related to study treatment. AEs potentially relating to immunogenicity (anaphylactic, hypersensitivity, or infusion-related reactions in patients with antidrug antibodies [ADAs]) were also analyzed.

Secondary (efficacy) end points were PFS, objective response rate (ORR, consisting of complete response or partial response [PR]), duration of response (DOR), time to progression (TTP), and OS. Objective response was assessed according to RECIST 1.1, and all images were reviewed centrally.

Blood samples for determining the presence/absence of ADAs and neutralizing ADAs (nADAs) were taken before study drug infusion (cycles 1, 3, 5, 9, and 13). Pharmacokinetic assessments of BI 695502 were based on plasma concentrations, measured using blood samples taken before (cycles 1, 3, 5, 9, 13, 17, 21 and 27) study drug infusion.

Statistical analysis

All patients treated with ≥1 dose of study treatment (treated set) were included in evaluations of both safety and efficacy. AEs are presented using descriptive statistics, with 95% confidence intervals (CIs). Exploratory subgroup analyses were performed to investigate the potential effects of age, sex, race, region, and ECOG performance status on the primary analysis. Efficacy end points were assessed by descriptive analyses, and Kaplan–Meier methodology was used where appropriate. As this was an exploratory trial, no formal hypothesis testing was performed. The sample size was determined based on an expectation that 30–50% of patients would experience a primary end point AE. With an expected AE rate of 30–50%, recruitment of 120 patients was predicted to generate a 95% CI width for the primary end point AE rate of between 16.2% and 17.6%.

Results

Patient disposition

A total of 182 patients were screened at 46 trial centers. Of these, 123 patients were enrolled and treated (Figure 1). Most study participants (96 patients, 78.0%) received at least 8 cycles. The most common reasons for discontinuation before the end of cycle 8 were AEs (11 patients, 8.9%), progressive disease (eight patients, 6.5%), and withdrawal by patient (four patients, 3.3%). Due to the manufacturing issue with BI 695502, 43 (35.0%) patients were switched to bevacizumab RP.

Patient demographics & baseline characteristics

Key patient demographics and characteristics are presented in Table 1. The majority of patients were <65 years of age (65.9%), Caucasian (66.7%), male (55.3%), and had stage IV cancer at diagnosis (81.3%). The ECOG performance score was zero in 54.5% of patients. Eighty-three patients (67.5%) were taking at least one concomitant medication, most commonly magnesium oxide (13 patients, 10.6%), lisinopril and ondansetron (12 patients each, 9.8%), and paracetamol (11 patients, 8.9%).

Safety

In the primary safety analysis, at least one prespecified AE occurred in 72 patients (58.5% [95% CI: 49.7–66.9]; Table 2). The most commonly occurring prespecified AE categories were hypertension (reported in 28.5% of patients), hemorrhage (22.8%), and anaphylactic, hypersensitivity, or infusion-related reactions (18.7%).

Subgroup analyses of the primary end point are shown in Figure 2. For most subgroups, results were consistent with the overall population. However, there were some variations according to race (rates of prespecified AEs: White, 49%; Asian, 82%; Black or African–American patients, 83%), geographic region (Eastern Europe, 30%; Asia, 83%; Western Europe, 69%), and ECOG performance status at baseline (0, 70%; ≥1, 44%).

At least one treatment-emergent AE of any type occurred in 121 patients (98.4%). As shown in Table 3, the most common AEs were nausea, fatigue, and peripheral sensory neuropathy. The most common system organ classes were GI disorders (95 patients, 77.2%), nervous system disorders (79 patients, 64.2%), and general disorders and administration site conditions (71 patients, 57.7%). Treatment-related AEs were reported in 89 patients (72.4%). Hypertension, fatigue, and epistaxis were the most frequent treatment-related AEs (Table 3). The most common system organ classes for treatment-related AEs were GI disorders (35 patients, 28.5%), vascular disorders (31 patients, 25.2%), and general disorders and administration site conditions (31 patients, 25.2%). Grade 3 or 4 AEs occurred in 77 patients (62.6%), while treatment-related grade 3 or 4 AEs affected 31 patients (25.2%).

Thirty-three patients (26.8%) had at least one serious AE (SAE). Treatment-related SAEs occurred in 12 patients (9.8%); those occurring in more than one patient were pulmonary embolism (three patients, 2.4%) and deep vein thrombosis (two patients, 1.6%). A minority of patients (n = 21, 17.1%) experienced AEs leading to discontinuation of study treatment. None of the AEs were fatal.

Efficacy

Approximately half of the patients (n = 61, 49.6%) either died (n = 16, 13.0%) or experienced disease progression (n = 45, 36.6%) during the trial. The median PFS was 10.5 months (95% CI: 9.4–11.8 months; Figure 3). The observed ORR was 61.0% (95% CI: 52.1–69.1%). At baseline, the median sum of diameters of target lesions was 77.9 mm (range 11.4–262.6 mm). Subsequent assessments showed ongoing reduction up to cycle 17, at a median value of 40.4 mm (range 0.0–212.1 mm).

The median DOR was 9.1 months with a lower bound of the 95% CI of 7.3 months (upper bound: not calculable). Median TTP was 11.1 months (95% CI: 9.5–12.9 months). A total of 41 patients (33.3%) died; 40 of the deaths were due to progressive disease, and one had an unknown cause. The median OS time was 19.4 months (95% CI: 16.7–21.1 months).

Immunogenicity

One patient had a numerically confirmed positive ADA result in a baseline sample (day 1, cycle 1), but with very low titer (2.0) and all subsequent samples for this patient were negative for ADAs. There were no nADA-positive samples, and no AEs potentially related to immunogenicity were reported.

Pharmacokinetics

The mean plasma concentration 90 minutes after the end of infusion of BI 695502 on day 1, cycle 1 was 94.3 μg/ml. As expected, preinfusion trough levels increased as the study progressed until cycle 9, with no substantial changes after this time (Figure 4). Three preinfusion samples showed high levels of BI 695502, one at cycle 5 and two at cycle 13; two of these samples were from the same patient.

Discussion

In this study, 59% of patients experienced at least one of the prespecified AEs for the primary analysis (anaphylactic, hypersensitivity or infusion-related reactions; arterial or venous thromboembolic events; GI perforations; hypertension; proteinuria; hemorrhage; wound-healing complications including abscess and fistulas; posterior reversible encephalopathy syndrome; or ovarian failure). Based on previous published studies of bevacizumab RP, around 30% of patients are expected to have experienced such AEs [8,9,18]. Although the proportion of patients with a prespecified AE in this study appears to be higher than anticipated, considerably lower percentages of patients were affected by other clinically significant types of AEs. For example, treatment-related grade 3 or 4 AEs affected a quarter of patients, with only 10% of patients reporting treatment-related SAEs and about 17% of patients experienced AEs leading to discontinuation. Considerably lower incidences of prespecified AEs were observed in the Eastern European patient population compared with the Asian and Western European populations. These data were only able to be generated as the trial design stratified patients by geographic region. Overall, the data from this study indicate that BI 695502 has acceptable tolerability.

The current study shows that when administered with mFOLFOX6 as first-line therapy for mCRC, BI 695502 was associated with median PFS of 10.5 months and median OS of 19.4 months. Cross-study comparisons should always be conducted with caution due to differences in patient populations, treatment regimens, and assessment methods; however, we consider it notable that the efficacy of BI 695502 in the present study appears comparable with previous studies of bevacizumab RP. Historical data from six studies in which bevacizumab RP was administered with FOLFOX as first-line treatment for mCRC show median PFS ranging between 10.1 and 11.4 months, and median OS ranging between 21.0 and 28.9 months [16–19,22,23]. Other studies suggest that outcomes are broadly similar when bevacizumab RP is administered with other chemotherapy regimens as first-line treatment for mCRC. In one study of KRAS wild-type advanced or metastatic CRC, bevacizumab RP was administered in combination with either FOLFOX or leucovorin, fluorouracil, and irinotecan (FOLFIRI); the median PFS was 10.6 months and median OS was 29.0 months [28]. In another study of bevacizumab plus IFL in patients with previously untreated and unselected mCRC, median PFS was 10.6 months and median OS was 20.3 months [7]. In the observational BEAT study, bevacizumab was administered with a variety of chemotherapy regimens, including FOLFOX, FOLFIRI, capecitabine plus oxaliplatin (XELOX), and fluoropyrimidines; median PFS and median OS were 10.8 months and 22.7 months, respectively [9]. In our study, 41% of patients had RAS mutation-positive tumors, which tend to have a worse prognosis than RAS wild-type [29], making comparison with recent, more selective studies challenging [28]. It is notable that although median PFS was within historical range in this study, median OS was on the lower side.

It is relevant to consider the safety findings of the current study in the context of data from previous studies of bevacizumab RP. In two studies comparing bevacizumab with the VEGF tyrosine kinase inhibitor, cediranib, diarrhea, fatigue, nausea, hypertension and peripheral neuropathy were among the most frequently reported AEs [15,18]. Hypertension was one of the AEs highlighted in the Phase II PEAK study, with a ≥5% higher incidence with bevacizumab RP versus panitumumab (anti-EGFR), both in combination with FOLFOX [19,22]. Furthermore, in another study, hypertension was the predominant mild-to-moderate AE with bevacizumab [28]. Grade 3–4 hypertension (6%), bowel perforation (2%), sepsis (1%), impaired wound healing (1%), and grade 3–4 proteinuria (1%) were noted as bevacizumab-related toxicities in the Phase II TREE-2 study [17]. In the BEAT study, bleeding, GI perforation, arterial thromboembolism, hypertension, proteinuria, and wound-healing complications were identified as serious or severe AEs of interest [9]. SAE incidence rates between 33% and 44% have been reported with bevacizumab, slightly higher than the rate of 27% in the present study [9,15,19,22]. The choice of chemotherapy influences the tolerability profile of anti-VEGF therapy and, as noted above, this should be carefully considered when reviewing outcomes from different studies. However, parallels between the previous data and the current results suggest that BI 695502 has a similar safety profile to that of bevacizumab RP.

BI 695502 was previously shown to have similar pharmacokinetics and safety to bevacizumab RP in healthy volunteers [27]. Plasma concentration–time profiles were similar with BI 695502 and both US- and EU-sourced bevacizumab RP among healthy volunteers in the previous study, and based on all available pharmacokinetic data, the three drugs were shown to be bioequivalent [27]. In the present study, the mean plasma concentration of BI 695502 90 minutes after infusion of 5 mg/kg on day 1, cycle 1 was 94.3 μg/ml. This timepoint was later than the time at which the maximum observed drug concentration in plasma (C_max) would have occurred. Mean plasma concentration of BI 695502 (94.3 μg/mL) was higher than the C_max values observed for BI 695502 and bevacizumab RP (US- and EU-sourced) among the healthy volunteers (24.0, 23.8 and 26.1 μg/ml, respectively) [27], which is attributable to the difference in doses (5 mg/kg in the current study vs 1 mg/kg in the study of healthy volunteers). Furthermore, bevacizumab pharmacokinetic parameters show linear changes over the dose range 1–20 mg/kg [30], and it is likely this would also be the case with BI 695502.

The strengths of this study include the inclusion of patients from several continents, and receipt by most patients of at least eight treatment cycles. Administration of chemotherapy that is commonly used in mCRC, and the fact that the study participants' baseline characteristics are representative of the intended target population of patients, increase the relevance of these findings to routine clinical practice. One limitation of this study was the lack of a comparator group. However, this trial and the wider clinical development program of BI 695502 was designed to fulfill the regulatory requirements for the development of biosimilars set by the US FDA [31]. When therapeutic equivalence has been demonstrated in one indication, the extrapolation of indications allows for the biosimilar to be used in other indications for which the RP is approved. As such, there is no regulatory requirement to generate further, comparative clinical data in other indications once therapeutic equivalence has been demonstrated. For BI 695502, pharmacokinetic bioequivalence was previously demonstrated in a Phase I, randomized trial [27]. The current study was designed to confirm safety and efficacy of BI 695502 in patients with mCRC, while another trial was designed for regulatory requirements to confirm equivalent safety and efficacy between BI 695502 and bevacizumab RP in the most sensitive indication (patients with non-small-cell lung cancer). Another limitation was the occurrence of a manufacturing issue with BI 695502 that caused patients to be switched to bevacizumab RP, although all data presented in this manuscript relate to treatment administered prior to this switch. A final limitation to interpreting the results of this study is that data on tumor ‘sidedness’ (location of the primary tumor), which may be an additional confounding factor, was not collected [32–34].

Conclusion

This study provides evidence of acceptable safety and efficacy of BI 695502 combined with mFOLFOX6 as first-line therapy in patients with mCRC. Consideration of our data alongside those from previous studies in this setting suggests that BI 695502 may be comparable to bevacizumab RP with respect to tolerability, efficacy, and pharmacokinetics. These findings may help to inform the development of other bevacizumab biosimilars, and more specifically their use in mCRC.