Bevacizumab biosimilar BEVZ92 versus reference bevacizumab in combination with FOLFOX or FOLFIRI as first-line treatment for metastatic colorectal cancer: a multicentre, open-label, randomised controlled trial

Articles

Bevacizumab biosimilar BEVZ92 versus reference bevacizumab in combination with FOLFOX or FOLFIRI as first-line treatment for metastatic colorectal cancer: a multicentre, open-label, randomised controlled trial Alvaro Romera, Sergiy Peredpaya, Yaroslav Shparyk, Igor Bondarenko, Giovanni Mendonça Bariani, Kathia Cristina Abdalla, Enrique Roca, Fábio Franke, Felipe Melo Cruz, Anita Ramesh, Vikas Ostwal, Pradeep Shah, Sajeed Abdul Rahuman, Alexandra Paravisini, Camino Huerga, Ana Del Campo García, Susana Millán

Summary

Background BEVZ92 is a proposed biosimilar to bevacizumab. The two molecules have similar physicochemical and functional properties in in-vitro and preclinical studies. In this clinical study, we compared the pharmacokinetic profile, efficacy, safety, and immunogenicity of BEVZ92 with reference bevacizumab as a first-line treatment in patients with metastatic colorectal cancer. Methods We did a randomised, open-label trial at 15 centres in Argentina, Brazil, India, Spain, and Ukraine. Eligible patients were aged 18 years or older, had metastatic colorectal cancer with at least one measurable non-irradiated lesion for which first-line chemotherapy was indicated and Eastern Cooperative Oncology Group (ECOG) performance status of 2 or less, had not received previous treatment for advanced disease, and whose bone marrow, hepatic, renal, and coagulation markers were all within normal ranges. Patients were randomly assigned (1:1) to either BEVZ92 or reference bevacizumab (5 mg/kg on day 1 of each cycle every 2 weeks) in combination with fluorouracil, leucovorin, and oxaliplatin (FOLFOX) or fluorouracil, leucovorin, and irinotecan (FOLFIRI). Randomisation was done via a web service based on a stochastic minimisation algorithm and was stratified by chemotherapy regimen (FOLFOX vs FOLFIRI), previous adjuvant therapy (yes vs no), ECOG performance status (0–1 vs 2), and study site. The primary endpoint was the area under the concentration-versus-time curve after a single infusion (AUC0–336h) and at steady state (AUCss)—ie, at cycle 7— in the assessable population, which comprised all treated patients for whom serum concentration measurements were available during the first seven cycles. Bioequivalence was established if the 90% CIs for the ratio of BEVZ92 to reference bevacizumab of the geometric means for AUC0–336h and AUCss were within the acceptance interval of 80–125%. Secondary endpoints included objective response, clinical benefit, and progression-free survival in the intention-to-treat population and immunogenicity and safety profiles in all treated patients. This trial is registered with ClinicalTrials.gov, number NCT02069704, and is closed to new participants, with follow-up completed. Findings 142 patients were randomly assigned, 71 to the BEVZ92 group and 71 to the reference bevacizumab group. Two participants assigned to BEVZ92 did not receive treatment (one withdrew consent, the other had a serious intestinal obstruction before starting treatment); therefore, the treated population comprised 69 patients in the BEVZ92 group and 71 in the reference bevacizumab group. The geometric mean ratio of AUC0–336h in the BEVZ92 versus the control group was 99·4% (90% CI 90·5–109·0) and of AUCss was 100·0% (90·2–112·0). Objective response (35 [49%] of 71 vs 40 [56%] of 71), clinical benefit (62 [87%] vs 65 [92%]), and progression-free survival (median 10·8 months [95% CI 7·4–11·5] vs 11·1 months [95% CI 8·0–12·8]) were similar in the BEVZ92 and reference bevacizumab groups. No relevant differences were noted between the safety profiles of the two study treatments. Neutropenia was the most common grade 3 or 4 adverse event reported in the BEVZ92 (14 [20%] of 69 patients) and reference bevacizumab (19 [27%] of 71 patients) groups. Serious adverse events occurred in 19 (28%) patients in the BEVZ92 group and 21 (30%) in the reference bevacizumab group. Two patients died because of bevacizumab-related serious adverse events: a sudden death in the BEVZ92 group and a serious large intestinal perforation in the reference bevacizumab group. The occurrence of anti-drug antibodies was low and similar in both treatment groups (two patients in the BEVZ92 group and one in the reference bevacizumab group). Interpretation Our results suggest that BEVZ92 and reference bevacizumab are pharmacokinetically bioequivalent and have no appreciable differences in safety profiles as first-line treatment in combination with FOLFOX or FOLFIRI in patients with metastatic colorectal cancer. Funding mAbxience Research SL.

Lancet Gastroenterol Hepatol 2018 Published Online September 24, 2018 http://dx.doi.org/10.1016/ S2468-1253(18)30269-3 Centro Oncológico de Rosario, Rosario, Santa Fe, Argentina (A Romera MD); Municipal Healthcare Facility Kharkiv Regional Clinical Oncology Center, Kharkiv, Ukraine (S Peredpaya MD); Danylo Halytsky Lviv National Medical University, Lviv State Oncology Regional Hospital 2a, Lviv, Ukraine (Y Shparyk MD); Dnipropetrovsk State Medical Academy, Dnipropetrovsk, Ukraine (Prof I Bondarenko MD); Instituto do Câncer do Estado de São Paulo Otávio Frias de Oliveira, São Paulo, Brazil (G Mendonça Bariani MD); Fundaçáo Pio XII—Hospital de Câncer de Barretos, Barretos, Brazil (K C Abdalla MD); Hospital de Gastroenterologia Dr Bonorino Udaondo, Buenos Aires, Argentina (E Roca MD); Hospital de Caridade Ijuí, Cacon, Brazil (F Franke MD); Instituto Brasileiro de Controle do Câncer, São Paulo, Brazil (F Melo Cruz PhD); Saveetha Medical College and Hospital, Chennai, India (Prof A Ramesh MD); Department of Medical Oncology, Tata Memorial Centre, Parel, Mumbai, India (V Ostwal MD); M S Patel Cancer Centre, Shree Krishna Hospital, Karamsand, India (P Shah MD); Regional Cancer Center & Medical College, Thiruvananthapuram, India (S A Rahuman MD); and mAbxience Research SL,Madrid, Spain (A Paravisini MD, C Huerga BPharm, A Del Campo García BPharm, S Millán PhD)

Copyright © 2018 Elsevier Ltd. All rights reserved. www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

1

Articles

Correspondence to: Ana Del Campo Garcia, mAbxience Research SL, Manuel Pombo Angulo 28 Street, 28050 Madrid, Spain ana.delcampo@ mabxience.com

Research in context Evidence before this study Bevacizumab was first approved for first-line treatment of patients with metastatic colorectal cancer in combination with regimens containing intravenous fluorouracil-based therapy in 2005. Since then, many studies of its pharmacology, efficacy, safety, and immunogenicity have been published. The proposed bevacizumab biosimilar BEVZ92 has shown high structural and functional similarity to reference bevacizumab in preclinical studies. As per guidelines from the US Food and Drug Administration, the European Medicines Agency, and WHO, the clinical development of BEVZ92 focused on confirming the similarity in terms of pharmacokinetics, efficacy, safety, and immunogenicity between the biosimilar and the reference product in an appropriately sensitive population, rather than on demonstration of clinical efficacy per se. We searched PubMed and the European Medicines Agency website with the terms “European public assessment report”, “bevacizumab”, “colorectal neoplasms”, “pharmacokinetics”, “efficacy”, “phase III clinical trial”, “meta-analysis”, “review”, and “biosimilar” for clinical and pharmacokinetic studies of bevacizumab in metastatic colorectal cancer published in English up to July 1, 2011.

Introduction A biosimilar is a drug that is highly physically, chemically, and biologically similar to an already available biological drug (the reference product).1 Interest in biosimilars is increasing because they could provide safe and efficacious treatment options for lower costs than the equivalent reference drug. The approval of a biosimilar involves the same strict standards of quality, safety, and efficacy that apply to any other biologic.1 Regulatory guidelines require that biosimilars show similarity to the reference medicine on the basis of a stepwise totality-ofevidence approach in structure, function, pharmaco­ kinetics, clinical efficacy, and safety. Bevacizumab is a recombinant humanised monoclonal IgG1 antibody that selectively binds with high affinity to all isoforms of human VEGF. It was approved by the US Food and Drug Administration in 2004 and by the European Medicines Agency in 2005 as first-line treatment for metastatic colorectal cancer in combination with chemotherapy.2,3 The efficacy of bevacizumab in combi­ nation with various chemotherapy regimens in metastatic colorectal cancer is supported by several clinical studies,4–7 and the use of bevacizumab is now the  standard of care.8 Two of the most common chemotherapy regimens used with bevacizumab are fluorouracil, leucovorin, and oxaliplatin (FOLFOX) and fluorouracil, leucovorin, and irinotecan (FOLFIRI). Since then, bevacizumab has been approved for use in combination therapies to treat other cancers, including metastatic or recurrent non-small-cell lung cancer, metastatic renal cell cancer, glioblastoma (not in Europe, 2

We identified pharmacokinetic data for reference bevacizumab in ten clinical trials in patients with solid tumours. Pivotal studies, meta-analyses, and publications from high-impact journals were selected and analysed. Added value of this study To our knowledge, ours is the first study of a biosimilar in which a population of patients with metastatic colorectal cancer—rather than a healthy population—was used as a model to generate evidence of pharmacokinetic equivalence and confirmation of similar safety, immunogenicity, and efficacy data with the reference product. This allowed for intrapatient assessment of the relationships between all variables. Implications of all the available evidence Our results show that BEVZ92 and reference bevacizumab are highly similar in terms of pharmacokinetics, efficacy, immuno­genicity, and safety. When used in the same way as the reference product, BEVZ92 is expected to produce the same benefits with no differences in safety, and could provide an affordable alternative for patients with metastatic colorectal cancer.

but in more than 70 countries worldwide), metastatic breast cancer (only in Europe), cervical cancer, and recurrent epithelial ovarian, fallopian tube, and primary peritoneal cancers.2,3 BEVZ92 is a proposed biosimilar of bevacizumab that was developed according to the guideline-recommended strategy9–12 for biosimilars. These guidelines suggest focusing on the quality attributes of the products as the first step. BEVZ92 was similar to reference bevaci­ zumab in an extensive physicochemical and functional character­isation, including in terms of primary structure, higher order structure, biological activity, and binding affinity to VEGF (unpublished data). In view of the minimal differences in vitro between BEVZ92 and the reference bevacizumab, the next step was a confirmatory trial, including assessment in an appropriately sensitive population of the similarity of pharmacokinetics and efficacy in the absence of meaning­­ful differences in safety and immunogenicity. We tested bioequivalence in a clinical setting in a pop­ ulation of patients with metastatic colorectal cancer receiving first-line treatment with the approved dose of 5 mg/kg (within the linear dose range of bevacizumab) in com­ bination with FOLFOX or FOLFIRI.13 Patients receiving first-line treat­ment are a more homogeneous population and have better prognosis than those receiving second-line and subsequent lines of treatment, and the chemotherapy regimens used not only reflect standard clinical practice, but are not expected to affect the pharmacokinetic profile of bevacizumab or BEVZ92.2,3

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

Articles

Methods

Study design and participants We did an open-label, randomised controlled trial (BEVZ92-A-01-13) at 15 hospitals in Argentina, Brazil, India, Spain, and Ukraine (appendix p 2). Eligible participants were aged 18 years or older, had metastatic colorectal cancer with at least one measurable nonirradiated site of disease according to the Response Evaluation Criteria in Solid Tumors (RECIST; version 1.1)14 for which first-line chemotherapy was indicated, had Eastern Cooperative Oncology Group (ECOG) performance statuses of 2 or less, had not received previous chemotherapy for advanced or meta­ static disease, and had bone marrow, hepatic, renal, and coagulation all within normal ranges (appendix). Adjuvant chemo­therapy or adjuvant chemoradiotherapy were permitted. Participants with active bleeding, clinically signi­ ficant cardiovascular disease, serious non-healing wounds, ulcers, bone fractures, proteinuria, or un­ controlled hypertension, diabetes, infection, or seizures, and those who had undergone major surgery within 4 weeks of randomisation were excluded. The full inclusion and exclusion criteria are detailed in the appendix (pp 3–5). Patients could be withdrawn at the investigators’ or sponsor’s discretion at any time for reasons such as serious adverse event, protocol deviation (eg, dosing regimen, failure to comply with clinic visit schedule), coexisting disease, worsening of condition, use of prohibited drugs, pregnancy, and clinically significant abnormal laboratory results. This study was done in accordance with the World Medical Association Declaration of Helsinki,15 the European Union GCP Directive CPMP/ICH/135/95,16 the guidelines for similar biological medicinal products containing monoclonal antibodies,10 and local laws and regulations relevant to the use of new therapeutic agents in each country of conduct. Ethical approval was sought and granted at each centre. All patients provided written informed consent before any study-specific procedures were done.

Randomisation and masking Study investigators enrolled participants, who were then randomly assigned (1:1) to receive either BEVZ92 or reference bevacizumab. Randomisation was done by stochastic minimisation as described by Pocock and Simon,17 with the minimisation probability parameter of 0·80, and was stratified by chemotherapy regimen (FOLFOX vs FOLFIRI), previous adjuvant therapy (yes vs no), ECOG performance status (0–1 vs 2), and study site, to ensure reasonable balance between the randomised treatments overall and within factors and sites, while maintaining the unpredictability of treatment allocations at individual sites. The electronic data capture system, randomisation system, and randomisation sequence were managed by a statistical third party provider. Investigators launched the electronic data

capture system, which would call up the randomisation feature (via a web service), and they would receive back the randomisation information. The study was open label, so neither investigators nor participants were masked to treatment assignment. However, drug concentration analysis for the primary pharmacokinetic endpoint was done by staff at a central laboratory, who were unaware of treatment assignment. The pharmacokinetic analysis dataset was masked by the study statistician and, to prevent unblinding, the pharmacokineticist in charge of data interpretation could not access any unblinded data. Generation of the anti-drug antibody (ADA) data in the bioanalytic laboratory was not masked, but the ADA profiling staff were unaware of treatment assignment. Any data transferred to the pharmaco­kineticist for ADA profiling was treated in the same way to prevent unblinding.

See Online for appendix

Procedures BEVZ92 or reference bevacizumab was given at the initial intravenous dose of 5 mg/kg on day 1 of each cycle every 2 weeks, in combination with the chosen chemo­therapy regimen.8 Each cycle of BEVZ92 or reference bevacizumab plus FOLFOX or FOLFIRI was repeated until progressive disease or unacceptable toxicity occurred, or until patients withdrew consent. Dose reduction of BEVZ92 or reference bevacizumab for adverse events was not permitted: therapy was either delayed or permanently discontinued. The first oxali­platin dose was fixed at 85 mg/m² but could be reduced thereafter. Dose reductions were allowed for fluoro­uracil, oxaliplatin, or irinotecan according to the indications in the relevant Summary of Product Characteristics. For pharmacokinetic assessments, blood samples were collected from each patient at cycles 1 and 7: before infusion, at the end of infusion, and 1 h, 2 h, 6 h, 24 h, 48 h, 72 h, 120 h, 168 h, and 240 h after the end of infusion. Samples were also taken before infusion and at the end of infusion at cycles 2, 5, and 8 to monitor serum bevacizumab concentrations during treatment. Serum concentrations were measured with a validated ELISA processed by Covance Laboratories (Harrogate, UK; appendix p 6). Tumour lesions were assessed at baseline and every 8 weeks until progressive disease or unacceptable toxicity was noted, or until patients withdrew consent. Anti-tumour activity was assessed by CT or MRI, and the best overall response for each patient at the end of the study was classified as complete response, partial response, stable disease, or progressive disease according to RECIST. For safety assessments, vital signs were measured, ECOG performance status was established, and physical and neurological examinations and haematology, serum biochemistry, and coagulation tests were done at screening visit and at designated timepoints throughout the study (on day 1 of every cycle and 30 days after the

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

3

Articles

177 patients assessed for eligibility

35 not recruited 26 did not meet inclusion criteria 8 declined to participate 1 lost to follow-up

142 randomly assigned

71 allocated to BEVZ92

71 allocated to reference bevacizumab

2 did not receive BEVZ92 1 withdrew consent 1 serious adverse event before treatment

71 received reference bevacizumab

69 received BEVZ92

69 discontinued BEVZ92 24 disease progression 14 withdrew consent 10 investigators’ decision 10 adverse events 4 protocol deviations 6 deaths 1 underwent surgery on primary tumour

0 remained on treatment at trial completion

69 analysed for secondary endpoints

14 excluded from primary analysis 5 disease progression 4 deaths 2 adverse events 1 withdrew consent 1 investigators’ decision 1 protocol deviation

55 analysed for primary endpoint

68 discontinued BEVZ92 29 disease progression 12 withdrew consent 8 investigators’ decision 6 adverse events 5 protocol deviations 6 deaths 2 underwent curative surgery

3 remained on compassionate treatment

71 analysed for secondary endpoints

10 excluded from primary analysis 2 disease progression 2 deaths 2 adverse events 2 investigators’ decision 1 withdrew consent 1 pharmacokinetic sample unavailable for cycles 1 and 7

61 analysed for primary endpoint

Figure 1: Trial profile

end of study treatment). A 12-lead electrocardiogram was done at screening, the end of treatment, and whenever clinically indicated. Treatment-emergent adverse events (TEAEs) were recorded during the study and up to 30 days after the end of study treatment, coded according to the Medical Dictionary for Regulatory Activities (version 16.0), and graded on the basis of the US National Cancer Institute’s Common Terminology for Adverse Events (version 4.0). For immunogenicity assessments, a screening assay was done to detect the presence of ADAs at baseline. 4

This assay was followed by a confirmatory analysis based on the percent signal inhibition in which positive samples were further characterised (eg, ADA titres were measured). Four blood samples (5 mL each) were gathered: one each at baseline, cycle 5, cycle 8, and 12 months after first drug administration (pre-dose). The concentration of anti-BEVZ92 and anti-referencebevacizumab antibodies was measured with an electro­ chemiluminescence detection assay (Meso Scale Discovery, Harrogate, England) and processed by Covance Laboratories (appendix p 7).

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

Articles

Outcomes Our primary endpoint was the truncated area under the concentration-versus-time curve calculated from start of the first infusion until start of the second infusion (AUC0–336h) and over a dosage interval at steady state (AUCss).18 Secondary pharmacokinetic endpoints were maximum (Cmax) and minimum (Ctrough) serum concen­ trations after one infusion and at steady state, the elimination rate constant of the terminal phase after multiple infusions (Kel), elimination half-life after multiple infusions (t1/2), time to maximum serum concentration after a single infusion (tmax), total clearance, and volume of distribution. Other secondary endpoints were objective response (ie, a complete or partial response), clinical benefit (ie, a complete response, partial response, or stable disease of any duration at any time during the study), progression-free survival (ie, the time from randomisation until the first evidence of progression as defined by RECIST or death from any cause), safety, and immunogenicity (as measured by assessment of ADAs).

Statistical analysis For the pharmacokinetic similarity assessments, we followed regulatory guidelines on bioequivalence,9–13,18,19 where­­­­­­by two treatments are judged not to be different from one another if the 90% CI of the ratio of a log-transformed exposure measure (AUC or Cmax) falls completely within the range 80–125%. On the basis of previously published pharmacokinetic data for reference bevacizumab in metastatic colorectal cancer,20,21 we assumed a conservative inter-participant coefficient of variation for AUC of around 35%. A sample size of 51 patients, in a parallel-group design, could show bio­ equivalence with a nominal power of 90% and an α of 0·05, if the 90% CIs for the ratio of BEVZ92 to reference bevacizumab of the geometric means for AUC0–336h and AUCss were within the acceptance interval of 80–125%.12,18,19 Pharmacokinetic parameters were assessed by standard non-compartmental analysis based on application of the trapezoidal rule for measurements of AUC.22 The analysis was done in Phoenix WinNonlin (version 6.2.1) in the assessable patient population, which comprised all randomly assigned and treated patients for whom serum concentration measurements during the first seven cycles were available to allow calculation of steady-state pharma­co­kinetic parameters. Complementary to the main pharma­cokinetic analysis, we did a population pharma­ cokinetic analysis with NONMEM (version 7.3).23 The population pharma­ co­ kinetic model of bevacizumab and parameter estimates we used were described by Han and colleagues.24 By using this model, we aimed to describe the data in patients with metastatic colorectal cancer, compare parameter estimates from the population pharmaco­ kinetic model (including variability) with bevacizumab reference values, and to do simulations to design further

All treated patients

Assessable patients

BEVZ92 (n=69)

Reference bevacizumab (n=71)

BEVZ92 (n=55)

Reference bevacizumab (n=61)

Mean (SD; range)

56·3 (12·9; 29–83)

56·7 (11·6; 33–78)

57·0 (12·8; 29–83)

57·0 (12·0; 33–78)

<65

49 (71%)

51 (72%)

39 (71%)

42 (69%)

≥65

20 (29%)

20 (28%)

16 (29%)

19 (31%)

Age, years

Sex Female

30 (43%)

32 (45%)

23 (42%)

28 (46%)

Male

39 (57%)

39 (55%)

32 (58%)

33 (54%)

White

50 (72%)

55 (77%)

38 (69%)

47 (77%)

Asian

14 (20%)

12 (17%)

12 (22%)

10 (16%)

Black

2 (3%)

3 (4%)

2 (4%)

3 (5%)

Other

3 (4%)

1 (1%)

3 (5%)

1 (2%)

Race

TNM stage IVa

30 (43%)

32 (45%)

26 (47%)

29 (48%)

IVb

38 (55%)

38 (54%)

28 (51%)

31 (51%)

1 (1%)

1 (1%)

1 (2%)

1 (2%)

Missing

Months since diagnosis of metastatic colorectal cancer n

67

69

53

61

Mean (SD)

13·0 (14·7)

13 (17·5)

12·4 (14·5)

13·2 (17·9)

Median (IQR)

4·5 (1·6–21·4)

4·3 (1·6–17·7)

3·6 (1·6–21·0)

4·3 (1·6–21·0)

ECOG performance status 0

7 (10%)

18 (25%)

6 (11%)

17 (28%)

1

57 (83%)

43 (61%)

45 (82%)

35 (57%)

2

5 (7%)

10 (14%)

4 (7%)

9 (15%) 10 (16%)

Number of target lesions 1

9 (13%)

11 (15%)

8 (15%)

2

29 (42%)

31 (44%)

25 (45%)

27 (44%)

≥3

31 (45%)

29 (41%)

22 (40%)

24 (40%)

Extent of disease Liver only

7 (10%)

5 (7%)

7 (13%)

4 (7%)

62 (90%)

66 (93%)

48 (87%)

57 (93%)

Surgery

55 (80%)

53 (75%)

45 (82%)

46 (75%)

Radiotherapy

14 (20%)

15 (21%)

12 (22%)

12 (20%)

Chemotherapy

24 (35%)

23 (32%)

20 (36%)

18 (30%)

Other Previous treatment*

Data are n (%) unless otherwise specified. TNM=tumour, node, metastasis. ECOG=Eastern Cooperative Oncology Group. *Recorded at screening visit.

Table 1: Baseline characteristics in all treated patients and assessable patients

studies. We selected this model over other population models25 because it includes more homo­geneous studies and was externally validated. Efficacy analyses were done in the intention-to-treat population (ie, all randomly assigned patients irrespective of treatment received). The response assessments were reported by the investigators. Safety analyses were done in all patients who received at least one dose of the study treatment. We used SAS (versions 9.3 and 9.4) for the descriptive statistics for the efficacy, safety, and immu­no­genicity analyses. This study is registered with ClinicalTrials.gov, number NCT02069704.

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

5

Articles

Role of the funding source

A Bevacizumab serum concentration (ng/mL)

125 000

BEVZ92 Reference bevacizumab

100 000 75 000

Results

50 000

Between Oct 29, 2014, and Nov 13, 2015, 177 patients were screened and 142 were randomly assigned, 71 to each group (figure 1). Two participants assigned to BEVZ92 did not receive treatment (one withdrew consent, the other had a serious intestinal obstruction before starting treatment). Therefore, the treated population comprised 69 patients in the BEVZ92 group and 71 in the reference bevacizumab group (figure 1). The baseline characteristics of treated patients and assessable patients were well balanced between groups (table 1). The mean age of

25 000 0

B Bevacizumab serum concentration (ng/mL)

1 000 000

100 000

BEVZ92

Cycle 1

C Bevacizumab serum concentration (ng/mL)

200 000

n

55

61

AUC0–336 h (ng/h/mL)

16 500 000

16 600 000

Ratio of geometric least squares means (90% CI)

99·4% (90·5–109·0)

··

n

55

58

AUC (ng/h/mL)

35 900 000

35 700 000

Ratio of geometric least squares means (90% CI)

100·0% (90·2–112·0)

··

Cycle 7 (steady state)

175 000 150 000 125 000 100 000

Secondary endpoints

75 000

Cycle 1

50 000

n

55

61

25 000

Cmax (ng/mL)

120 000

123 000

Ratio of geometric least squares means (90% CI)

97·9% (89·8–107·0)

··

Ctrough (ng/mL)

344

349 (n=60)

0

D

Cycle 7 (steady state)

1 000 000 Bevacizumab serum concentration (ng/mL)

Reference bevacizumab

Primary endpoint

10 000

100 000

10 000 0

48

96

144

192

240

288

336

Time since infusion (h)

Figure 2: Serum bevacizumab concentrations during cycle 1 on a standard (A) and semi-logarithmic (B) scale and during cycle 7 on a standard (C) and semi-logarithmic (D) scale in all assessable patients in the BEVZ92 and reference bevacizumab groups

6

Employees of the study funder had roles in study design, data analysis and interpretation, and writing of the report, and had access to the raw data. They had no role in data collection. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

n

55

58

Cmax (ng/mL)

195 000

200 000

Ratio of geometric least squares means (90% CI)

97·3% (88·9–107·0)

··

Total body clearance (L/h) 0·00936

0·00934

Ratio of geometric least squares means (90% CI)

100·0% (90·6–111·0)

··

Ctrough (ng/mL)

69 600 (n=54)

69 300 (n=58)

t1/2 (h)

294 (n=49)

289 (n=52)

Kel, (1/h)

0·00236 (n=55)

Volume of distribution (L) 4·06 (n=55)

0·00240 (n=61) 3·86 (n=61)

AUC=area under the curve (concentration vs time). Cmax=maximum serum  concentration. Ctrough=minimum serum concentration. Kel=elimination rate constant. t1/2=elimination half-life.

Table 2: Pharmacokinetic endpoints in the assessable patient population

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

Articles

Complete response

100

Reference bevacizumab (n=71)

1 (1%)

3 (4%)

Partial response

34 (48%)

37 (52%)

Stable disease

27 (38%)

25 (35%)

Progressive disease

4 (6%)

2 (3%)

Unevaluable

5 (7%)

4 (6%)

Objective response*

35 (49% [37–61])

40 (56% [44–68])

Clinical benefit†

62 (87% [77–94])

65 (92% [83–97])

Data are n (%) or n (% [95% CI]). The best overall responses among all post-baseline assessments of each patient, including unscheduled assessments, are included. *Patients with a complete or partial response at any time during the study (including unscheduled assessments; patients who did not undergo any post-baseline tumour assessments were deemed non-responders); exact two-sided 95% CIs were calculated with the Clopper-Pearson method. †Patients with a complete response, partial response, or stable disease at any time during the study (including unscheduled assessments and assessments after end of treatment).

Table 3: Response to treatment in the intention-to-treat population

participants was 56·5 years (SD 12·2), 62 (44%) were female, and 105 (75%) were white. 40 (29%) participants were aged 65 years or older (table 1). Most patients in both groups had an ECOG performance status of 1 and had previously undergone surgical treatment (table 1). At trial completion all patients had discontinued study treatment, except for three participants in the bevaci­zumab group who were still receiving treatment under a compassionate programme. Median exposure to study treatment was 6·1 months (95% CI 4·0–9·3) in the BEVZ92 group and 7·1 months (4·5–10·7) in the control group; median time to study end was 10·2 months (5·7–12·5) versus 11·8 months (7·7–13·1). The median cumulative doses were 65 mg/kg (IQR 45–99) for BEVZ92 and 71 mg/kg (45–115) for refer­ ence bevacizumab. BEVZ92 and reference bevacizumab had very similar median concentration–time pharmacokinetic profiles after one infusion and at cycle 7 when steady state was reached (figure 2). The geometric mean ratio was 99·4% (90% CI 90·5–109·0) for AUC0–336h and 100·0% (90·2–112·0) for AUCSS (table 2). The bio­equivalence of BEVZ92 to reference bevacizumab was therefore shown because the 90% CIs for the ratio of geometric means for the primary endpoints were both within the predefined bioequivalence acceptance range of 80–125%. Secondary pharmacokinetic end­points were also similar between treatment groups (table 2). The two-compartment drug disposition of bevacizumab in our population was well described by the model used, with good agreement between observations and predictions (appendix p 9). Population pharmaco­kinetic estimates of clearance, central volume of distribution, and t1/2 for a typical 70 kg patient were 9·0 mL/h, 2·9 L, and 19·6 days, respectively, for BEVZ92. Individual model-derived parameters showed similar median values for BEVZ92 and reference bevacizumab (clearance 9·0 vs

n

Events Treatment

69 41 71 42

80 Progression-free survival (%)

BEVZ92 (n=71)

Median (95% CI)

BEVZ92 10·8 (7·4–11·5) Reference bevacizumab 11·1 (8·6–12·8)

60

40

20

0

0

Number at risk BEVZ92 69 Reference 71 bevacizumab

3

6

9

12

15

18

21

24

4 11

2 10

1 7

1 4

Time (months) 57 65

39 50

25 31

10 18

Figure 3: Progression-free survival in the BEVZ92 and reference bevacizumab groups BEVZ92 (n=69)

Reference bevacizumab (n=71)

Any TEAE irrespective of causality 66 (96%)

71 (100%)

Grade ≥3 TEAE

44 (64%)

49 (69%)

TEAE leading to discontinuation

13 (19%)

6 (8%)

Any treatment-related TEAE*

63 (91%)

70 (99%)

Grade ≥3 treatment-related TEAEs 37 (54%)

44 (62%)

Any serious† TEAE

21 (30%)

Fatal TEAEs Any bleeding event Grade ≥3 bleeding events

19 (28%) 8 (12%)

5 (7%)

14 (20%)

19 (27%)

1 (1%)

2 (3%)

Data are n (%). TEAE=treatment-emergent adverse event. *Related to any study treatment, including fluorouracil, leucovorin, and oxaliplatin (FOLFOX) or fluorouracil, leucovorin, and irinotecan (FOLFIRI) regimens. † Any untoward medical occurrence that at any dose results in death,a life-threatening event, hospitalisation or prolongation of hospitalisation, substantial or persistent disability, a congenital anomaly or birth defect, or any other medically important condition.

Table 4: Participants with at least one TEAE in the safety population

9·3 mL/h, central volume 3201 vs 3213 L, intercom­ partmental clearance of 18·1 vs 20·9 mL/h, and peripheral volume of 2875 vs 2750 L). According to RECIST, efficacy was similar between treatment groups: objective response was 49% (95% CI 37–61) in the BEVZ92 group and 56% (44–68) in the control group (table 3). Clinical benefit was noted in 87% (95% CI 77–94) of patients in the BEVZ92 group and 92% (83–97) in the control group (table 3). Median progression-free survival was 10·8 months (95% CI 7·4–11·5) in the BEVZ92 group and 11·1 months (8·6–12·8) in the control group (figure 3). The proportion of patients reporting at least one TEAE was similar in both groups (tables 4, 5). The most

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

7

Articles

BEVZ92 (n=69)

Reference bevacizumab (n=71)

Grade 1–2

Grade 3–4

Grade 1–2

Grade 3–4

Diarrhoea

25 (36%)

6 (9%)

32 (45%)

6 (8%)

Nausea

27 (39%)

2 (3%)

29 (41%)

2 (3%)

9 (13%)

14 (20%)

7 (10%)

19 (27%)

Neutropenia Vomiting

20 (29%)

20 (28%)

2 (3%)

Anaemia

16 (23%)

2 (3%)

21 (30%)

3 (4%)

Asthenia

17 (25%)

1 (1%)

18 (25%)

1 (1%)

7 (10%)

7 (10%)

13 (18%)

6 (8%)

Fatigue

14 (20%)

1 (1%)

14 (20%)

4 (6%)

Decreased appetite

15 (22%)

Hypertension

Leucopenia Peripheral neuropathy

··

··

17 (24%)

··

7 (10%)

7 (10%)

13 (18%)

3 (4%) 7 (10%)

19 (28%)

5 (7%)

22 (31%)

Thrombocytopenia

6 (9%)

3 (4%)

12 (17%)

1 (1%)

Paraesthesia

5 (7%)

5 (7%)

6 (8%)

2 (3%)

Epistaxis

7 (10%)

··

10 (14%)

Dysgeusia

4 (6%)

··

11 (15%)

Stomatitis

5 (7%)

Abdominal pain

5 (7%)

Skin hyperpigmentation

7 (10%)

Weight loss

3 (4%)

·· ··

6 (8%)

2 (3%)

··

13 (18%)

1 (1%)

··

7 (10%)

··

6 (9%)

··

7 (10%)

··

Increased aspartate aminotransferase

3 (4%)

··

9 (13%)

··

Proteinuria

6 (9%)

1 (1%)

5 (7%)

Palmar-plantar erythrodysaesthesia

3 (4%)

1 (1%)

6 (8%)

1 (1%)

··

·· 2 (3%)

Oral candidiasis

··

Urinary tract infection

··

··

1 (1%)

2 (3%)

Hypotension

··

··

1 (1%)

2 (3%)

Drug hypersensitivity

··

··

1 (1%)

2 (3%)

Sepsis

··

··

··

2 (3%)

2 (3%)

Data are n (%). The table shows all grade 1–2 adverse events that occurred in 10% or more of patients and all grade 3–4 events that occurred in 3% or more of patients.

Table 5: Common treatment-related treatment-emergent adverse events

frequent (ie, occurring in >40% of patients) TEAEs were diarrhoea and nausea in the BEVZ92 group and diarrhoea, nausea, and peripheral neuropathy in the reference bevacizumab group (appendix pp 10–11). Grade 3 or 4 TEAEs occurred in 44 (64%) participants in the BEVZ92 group and 49 (69%) in the reference bevacizumab group (table 4). Neutropenia was the most common grade 3 or 4 TEAE in both groups (14 [20%] vs 19 [27%]; table 5). Most TEAEs were judged by investigators to be related to study treatment (either with bevacizumab or chemotherapy; table 4). Most of these events were events commonly noted with chemotherapy, and occurred at similar frequencies in both groups (table 5). Other events often described in patients treated with bevacizumab, such as hypertension, proteinuria, thrombocytopenia, and bleeding, were also reported in similar proportions in both groups (tables 4, 5). Bleeding events were mostly mild or moderate in intensity: only one grade 3–4 event (lower gastrointestinal haemorrhage) was noted in the BEVZ92 group, and only two (subdural haematoma and tumour haemorrhage) were noted in the reference bevacizumab group (table 4). 8

19 patients reported an adverse event leading to discontinuation of treatment, 13 in the BEVZ92 group and six in the control group (table 4). Discontinuation was associated with bevacizumab alone or with bevacizumab and chemotherapy by investigators in nine patients (five in the BEVZ92 group and four in the reference bevacizumab group). The bevacizumab-related TEAEs leading to discontinuation in the BEVZ92 group were surgical wound bleeding, wound dehiscence, aortic thrombosis, aortic dissection, and thrombo­embolism, in one patient each; only the thrombo­embolism was judged to be serious. In the control group, the bevacizumabrelated TEAEs leading to discon­tinuation were hyper­ tension, portal vein thrombosis, abdominal abscess, and hypotension and dystonia, in one patient each. Only the case of hypo­tension and dystonia was considered serious. Serious adverse events were reported in similar proportions of patients in the BEVZ92 group (19 [28%]) and the reference bevacizumab group (21 [30%]; table 4). Common treatment-related serious adverse events included neutropenia (one patient in the BEVZ92 group vs four patients in the control group), diarrhoea (one vs four), anaemia (one vs three), leucopenia (zero vs two), sepsis (zero vs two), and hypotension (zero vs two). Other treatment-related serious adverse events during the study included an asymptomatic thromboembolism in the right trunk of the pulmonary artery in a patient in the BEVZ92 group, and a large intestinal perforation and a superior vena cava thrombosis (in two different patients) in the reference bevacizumab group. Death caused by a TEAE occurred in eight patients in the BEVZ92 group and five in the control group (table 4). Only two patients died because of a serious adverse event related to bevacizumab: one sudden death judged to be related to both BEVZ92 and FOLFIRI, and one large intestinal perforation that was deemed unlikely to be related to reference bevacizumab. No other significant adverse events occurred, and we noted no other important differences between the two groups in terms of safety outcomes (data not shown). All 140 patients who received treatment were analysed for ADAs. Only two patients allocated to the BEVZ92 group developed de-novo ADAs after administration of study drugs (one detected at cycle 8, and one at the 12-month study visit). In the reference bevacizumab group, only one patient had positive results at cycle 8. This patient also had positive results at baseline, but titration did not result in boosted ADA concentrations. Among these ADA-positive patients, we noted no effects on pharmacokinetics, efficacy, or safety (data not shown).

Discussion In this randomised, open-label study, we showed that BEVZ92 was bioequivalent to reference bevacizumab because the 90% CIs for the ratio of geo­metric means for AUC0–336h and AUCSS were both within the predefined bio­ equivalence acceptance range of 80–125%. Additionally,

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

Articles

the efficacy, safety, and immuno­genicity of BEVZ92 and reference bevacizumab were similar. We did our study in patients with metastatic colorectal cancer being treated with FOLFOX or FOLFIRI, mimick­ ing the use of bevacizumab in clinical settings, instead of in healthy volunteers. An advantage of this innovative, simplified clinical development approach is that safety, immunogenicity, and efficacy data can be gathered from every participant, which allows for intrapatient assess­ ment of the relationships between all variables. Additionally, patients were receiving first-line treatment, and thus less heterogeneity was expected because patients had better prognosis than those receiving second-line or later treatment, and because they were less likely to have received previous therapies. Further­ more, the 5 mg/kg dose of bevacizumab allows accurate assessment of the pharma­cokinetic similarity of BEVZ92, because it lies within the linear section of the established dose–response curve (1–10 mg/kg),26 and does not interact with the chemotherapy regimen used.9–13 Individual model-derived pharmacokinetic para­meters obtained from the implemented population analysis also showed similar values, when factors known to affect the pharmacokinetics of bevacizumab were accounted for, showing the robustness of our bio­equivalence results. The binding specificity, affinity or avidity, and binding kinetics to VEGF are the main determinant factors of the clinical activity of bevacizumab. BEVZ92, as a potential biosimilar to bevacizumab, has an identical primary structure, similar post-translational modifications and biochemical properties, the same binding specificity and affinity to the same target antigens (ligand binding and neutralisation assays), and similar binding kinetics to the neonatal Fc receptor as reference bevacizumab (un­ published data). Therefore the plasma drug con­ cen­ tration should be the sole deter­minant of clinical activity. To confirm similar clinical behaviour, the selected endpoints should be sensitive to the detection of productrelated differences in pharmacokinetics, safety, immuno­ genicity, and efficacy. As recommended by regulatory guidelines,10 measurement of response is probably the best endpoint for assessments of efficacy in a biosimilar study, because it is more sensitive for detection of differences between the proposed biosimilar and the reference product. Overall survival, the preferred efficacy endpoint in oncology, might not be suitable to establish bio­ similarity because it can be affected by various factors that are not attributable to the activity of the bio­ similar or the reference product.10 Furthermore, objective response was selected as the primary endpoint in the confirmatory efficacy study of ABP 215, the first bevaci­ zumab biosimilar approved by the European Medicines Agency and the US Food and Drug Administration.27 Objective response is not always associated with longterm improvements in patient outcomes and does not closely correlate with progression-free survival in patients

with metastatic disease.28 Because patients in our trial received study treatment and were followed up until disease progression, unacceptable toxicity, death, or withdrawal of consent, we were able to compare objective response and progression-free survival between treat­ ment groups as secondary efficacy endpoints. In terms of response, BEVZ92 had similar efficacy to reference bevacizumab in combination with FOLFOX or FOLFIRI, with an objective response in 49% of patients in the BEVZ92 group and 56% of patients in the reference bevaci­ zumab group. Although care should be taken when comparing data from trials of different designs, these results are in line with those of previous publications about reference bevacizumab plus different chemotherapy regimens as first-line treatment of meta­ static colorectal cancer, in which the frequency of objective responses varies from 26% to 57·9%.29 In this clinical setting, the frequency of objective response to reference bevacizumab in combination with chemo­ therapy was 39·2% in a meta-analysis30 published by Wang and colleagues, with a pooled objective response of 49% in combination with oxaliplatin-based regimens31 and 51% in combination with FOLFIRI.32 Progressionfree survival was similar in the BEVZ92 and reference bevacizumab groups in our trial. Our results were similar to those published previously for reference bevaci­ zumab, which had a median progression-free survival of 10·3 months in combination with FOLFOX and 10·2 months in combination with FOLFIRI.33 Bevacizumab treatment is associated with an increase in the frequency of events that commonly occur with VEGF inhibitors, including hypertension, proteinuria, bleeding, wound-healing complications, gastrointestinal perfor­ations, and thromboembolic events.34 In our study, the well characterised safety profile of bevacizumab in metastatic colorectal cancer3 was shown by the similar safety results in both the BEVZ92 and reference bevacizumab groups. An imbalance was noted between groups in the number of patients who discontinued treatment because of adverse events (13 [19%] patients in the BEVZ92 group vs six [8%] in the control group), but discontinuation was related to chemotherapy (FOLFOX or FOLFIRI) or to the underlying disease in more than 50% of patients (60% in the BEVZ92 arm and 40% in the reference bevacizumab group). When only bevacizumabrelated adverse events were included, the proportion of patients who discontinued treatment was similar in both groups (five [7%] in the BEVZ92 group vs four [6%] in the control group). Overall, BEVZ92 had a similar safety profile to reference bevacizumab, with no relevant differences in the nature, severity, or frequency of adverse events. Bevacizumab is a low-immunogenic molecule, and our results accord with previous immunogenicity data. The incidence of treatment-emergent anti-bevacizumab antibodies in clinical studies of reference bevacizumab in colon carcinoma is 0·6%.2 Our study also showed a

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

9

Articles

low frequency of de-novo ADA development, with no relevant differences between groups, even though we used more sensitive and specific assays than were used in previous studies of reference bevacizumab. Further­ more, ADA responses did not seem to affect pharmaco­ kinetics, safety, or efficacy, and thus their clinical significance remains unknown.2 Our study had several limitations. We used an openlabel design, which was chosen because of the characteristics of the distinctive brand appearance of the reference bevacizumab vials. Because the primary endpoint was pharmacokinetic, we thought that blinding on-site clinicians was unnecessary because they would not have access to any pharma­cokinetic results. However, the staff involved in the pharmacokinetic analysis, input database, and ADA profiling were masked to minimise any potential bias. Other potential disadvantages of this open-label design, such as issues with participant retention or detection or reporting bias, do not seem to have affected the study, because results were similar in both groups. The study sample size was adequate for the assessment of the primary pharmacokinetic endpoint, but was not sufficiently powered to show potential differences in efficacy. However, the results of the study in terms of efficacy, safety, and immunogenicity in patients treated with BEVZ92 in combination with FOLFOX or FOLFIRI were similar to those in patients treated with reference bevacizumab in combination with FOLFOX or FOLFIRI, and were consistent with the published data for reference bevacizumab in combination with fluoropyrimidine in first-line treatment of patients with metastatic colorectal cancer. BEVZ92 development has followed a stepwise approach to generate data in support of a demonstration of biosimilarity. Our results show the bioequivalence of BEVZ92 with the reference product, bevacizumab, and suggest no appreciable differences in terms of efficacy, immunogenicity, or safety profiles in a representative clinical model. The extensive structural and functional characterisations showed a low residual uncertainty and, as a scientific matter, this comparative clinical study in patients with metastatic colorectal cancer is considered a targeted and selective approach to identify whether there are clinically meaningful differences between the proposed biosimilar and the reference product. Although additional studies are being considered for global distribution of BEVZ92 because of changes to manu­ facturing sites and to satisfy local regulatory require­ ments, the results of this study and of the full in-vitro comparability assessment confirm BEVZ92 as a biosimilar of the reference bevacizumab The development of effective and safe biosimilars provides the public with greater access to treatment options, and could lower health-care costs through competition and increase access to life-saving drugs. Renowned oncology societies, including the American Society of Clinical Oncology35 and the European Society 10

for Medical Oncology,36 support the use of biosimilars. Furthermore, the arrival of biosimilars will ease the investigation of new combinations (such as with PD-L1 inhibitors in cancer) by helping to reduce the costs of innovation. One bevacizumab biosimilar has been approved (ABP 215), and several other potential bio­ similars will come onto market in the near future. Access to BEVZ92 would increase the range of biological drugs available and could contribute positively to oncology and the sustainability of health-care systems. Contributors AP and SM conceived and designed the study. AR, SP, YS, IB, GMB, KCA, ER, FF, FMC, AR, VO, PS, and SAR enrolled patients and collected data. AP, ADCG, and SM analysed and interpreted data. CH analysed data. AP wrote and revised the report, which was critically revised for important intellectual content by SM. All authors approved the final version of the Article. Declaration of interests AP, CH, ADCG, and SM are employees of mAbxience Research SL. All other authors declare no competing interests. Acknowledgments We thank Sulabhchandra Bhamare (Curie Manavta Cancer Center, Nashik, India), Meenakshisundaram Manickavasagam and Suresh Sudalaindi (Sri Ramachandra Hospital, Chennai, India), Yuriy Vinnyk (Municipal Healthcare Facility Kharkiv Regional Clinical Oncology Center, Kharkiv, Ukraine), Arinilda Campos Bragagnoli (Fundaçáo Pio XII—Hospital do Câncer de Barretos, Barretos, Brazil), Gustavo Werutsky (Hospital São Lucas da PUC RS, Porto Alegre, Brazil), and Aldo Dettino (AC Camargo Cancer Center, São Paulo, Brazil) for their help with enrolment and data collection; Dimitri Stamatiadis (MAIA consulting, Cointrin, Switzerland) for writing assistance; and Bernando de Miguel Lillo for pharmacokinetic assistance. References 1 European Medicines Agency. Biosimilars in the EU. Information guide for healthcare professionals. http://www.ema.europa.eu/ docs/en_GB/document_library/Leaflet/2017/05/WC500226648.pdf (accessed June 11, 2018). 2 Food and Drug Administration. Avastin (bevacizumab) prescribing information. https://www.accessdata.fda.gov/drugsatfda_docs/ label/2009/125085s0169lbl.pdf (accessed June 11, 2018). 3 European Medicines Agency. Avastin summary of product characteristics. http://www.ema.europa.eu/docs/en_GB/document_ library/EPAR_-_Product_Information/human/000582/ WC500029271.pdf (accessed June 11, 2018). 4 Hochster HS, Hart LL, Ramanathan RK, et al. Safety and efficacy of oxaliplatin and fluoropyrimidine regimens with or without bevacizumab as first-line treatment of metastatic colorectal cancer: results of the TREE study. J Clin Oncol 2008; 26: 3523–29. 5 Hurwitz H, Fehrenbacher L, Novotny W, et al. Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med 2004; 350: 2335–42. 6 Kabbinavar F, Hurwitz HI, Fehrenbacher L, et al. Phase II, randomized trial comparing bevacizumab plus fluorouracil (FU)/leucovorin (LV) with FU/LV alone in patients with metastatic colorectal cancer. J Clin Oncol 2003; 21: 60–65. 7 Saltz LB, Clarke S, Diaz-Rubio E, et al. Bevacizumab in combination with oxaliplatin-based chemotherapy as first-line therapy in metastatic colorectal cancer: a randomized phase III study. J Clin Oncol 2008; 26: 2013–19. 8 National Comprehensive Cancer Network. Clinical practice guidelines in oncology. Guidelines for colon cancer and rectal cancer 2018. https://www.nccn.org/professionals/physician_gls/ pdf/colon.pdf (accessed June 11, 2018). 9 Food and Drug Administration. Scientific considerations in demonstrating biosimilarity to a reference product. Guidance for industry. http://www.fda.gov/downloads/drugs/ guidancecomplianceregulatoryinformation/guidances/ucm291128. pdf (accessed June 11, 2018).

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

Articles

10 European Medicines Agency. Guideline on similar biological medicinal products containing monoclonal antibodies—nonclinical and clinical issues. http://www.ema.europa.eu/docs/en_ GB/document_library/Scientific_guideline/2012/06/WC500128686. pdf (accessed June 11, 2018). 11 European Medicines Agency. Guideline on similar biological medicinal products containing biotechnology-derived proteins as active substance: non-clinical and clinical issues. http://www.ema.europa.eu/docs/en_GB/document_library/ Scientific_guideline/2015/01/WC500180219.pdf (accessed June 11, 2018). 12 WHO Expert Committee on Biological Standardization. Guidelines on evaluation of similar biotherapeutic products (SBPs). Geneva: World Health Organization, 2013. 13 WHO Expert Committee on Biological Standardization. Guidelines on evaluation of monoclonal antibodies as similar biotherapeutic products (SBPs). http://www.who.int/biologicals/expert_ committee/mAb_SBP_GL-ECBS_review_adoption-2016.10.2611.7post_ECBS-Clean_Version.pdf (accessed July 27, 2018). 14 Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009; 45: 228–47. 15 World Medical Association. Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA 2013; 310: 2191–94. 16 European Union. GCP Directive CPMP/ICH/135/95. http://www.ema.europa.eu/docs/en_GB/document_library/ Scientific_guideline/2009/09/WC500002874.pdf (accessed June 11, 2018). 17 Pocock SJ, Simon R. Sequential treatment assignment with balancing for prognostic factors in the controlled clinical trial. Biometrics 1975; 31: 103–15. 18 Food and Drug Administration. Clinical pharmacology data to support a demonstration of biosimilarity to a reference product. Guidance for industry. http://www.fda.gov/downloads/drugs/ guidancecomplianceregulatoryinformation/guidances/ucm397017. pdf (accessed June 11, 2018). 19 European Medicines Agency. Guideline on the investigation of bioequivalence http://www.ema.europa.eu/docs/en_GB/document_ library/Scientific_guideline/2010/01/WC500070039.pdf (accessed June 11, 2018). 20 Zhi J, Chen E, Major P, et al. A multicenter, randomized, open-label study to assess the steady-state pharmacokinetics of bevacizumab given with either XELOX or FOLFOX-4 in patients with metastatic colorectal cancer. Cancer Chemother Pharmacol 2011; 68: 1199–206. 21 Food and Drug Administration. Avastin—drug approval package, study no 5. https://www.accessdata.fda.gov/drugsatfda_docs/ nda/2004/STN-125085_Avastin_BioPharmr.pdf (accessed June 11, 2018). 22 Gabrielsson J, Weiner D. Non-compartmental analysis. Methods Mol Biol 2012; 929: 377–89.

23 Beal S, Sheiner LB, Boeckmann A, Bauer RJ (Eds). NONMEM User’s Guides (1989–2011). Ellicott City, MD: Icon Development Solutions, 2011. 24 Han K, Peyret T, Marchand M, et al. Population pharmacokinetics of bevacizumab in cancer patients with external validation. Cancer Chemother Pharmacol 2016; 78: 341–51. 25 Lu JF, Bruno R, Eppler S, Novotny W, Lum B, Gaudreault J. Clinical pharmacokinetics of bevacizumab in patients with solid tumors. Cancer Chemother Pharmacol 2008; 62: 779–86. 26 European Medicines Agency. Scientific discussion. http://www.ema.europa.eu/docs/en_GB/document_library/ EPAR_-_Scientific_Discussion/human/000582/WC500029262.pdf (accessed June 11, 2018). 27 Thatcher N, Thomas M, Paz-Ares L, et al. Randomized, double-blind, phase 3 study evaluating efficacy and safety of ABP 215 compared with bevacizumab in patients with non-squamous NSCLC. Proc Am Soc Clin Oncol 2016; 34: abstr 9095. 28 Socinski MA, Curigliano G, Jacobs I, Gumbiner B, MacDonald J, Thomas D. Clinical considerations for the development of biosimilars in oncology. MAbs 2015; 7: 286–93. 29 Strickler JH, Hurwitz HI. Bevacizumab-based therapies in the first-line treatment of metastatic colorectal cancer. Oncologist 2012; 17: 513–24. 30 Wang M, Zheng X, Ruan X, et al. Efficacy and safety of first-line chemotherapy plus bevacizumab in patients with metastatic colorectal cancer: a meta-analysis. Chin Med J 2014; 127: 538–46. 31 Petrelli F, Coinu A, Ghilardi M, Cabiddu M, Zaniboni A, Barni S. Efficacy of oxaliplatin-based chemotherapy + bevacizumab as first-line treatment for advanced colorectal cancer: a systematic review and pooled analysis of published trials. Am J Clin Oncol 2015; 38: 227–33. 32 Petrelli F, Borgonovo K, Cabiddu M, et al. FOLFIRI-bevacizumab as first-line chemotherapy in 3500 patients with advanced colorectal cancer: a pooled analysis of 29 published trials. Clin Colorectal Cancer 2013; 12: 145–51. 33 Bendell JC, Bekaii-Saab TS, Cohn AL, et al. Treatment patterns and clinical outcomes in patients with metastatic colorectal cancer initially treated with FOLFOX-bevacizumab or FOLFIRI-bevacizumab: results from ARIES, a bevacizumab observational cohort study. Oncologist 2012; 17: 1486–95. 34 Hurwitz HI, Tebbutt NC, Kabbinavar F, et al. Efficacy and safety of bevacizumab in metastatic colorectal cancer: pooled analysis from seven randomized controlled trials. Oncologist 2013; 18: 1004–233. 35 Lyman GH, Balaban E, Diaz M, et al. American Society of Clinical Oncology statement: biosimilars in oncology. J Clin Oncol 2018; 36: 1260–65. 36 Tabernero J, Vyas M, Giuliani R, et al. Biosimilars: a position paper of the European Society for Medical Oncology, with particular reference to oncology prescribers. ESMO Open 2016; 1: e000142.

www.thelancet.com/gastrohep Published online September 24, 2018 http://dx.doi.org/10.1016/S2468-1253(18)30269-3

11