Anti-angiogenic agents in metastatic colorectal cancer

Abstract

Colorectal cancer (CRC) is a major public health concern being the third leading cause of cancer mortality in the United States. The availability of better therapeutic options has led to a decline in cancer mortality in these patients. Surgical resection should be considered in all stages of the disease. The use of conversion therapy has made surgery a potentially curative option even in patients with initially unresectable metastatic disease. In this review we discuss the role of various anti-angiogenic agents in patients with metastatic CRC (mCRC). We describe the mechanism of action of these agents, and the rationale for their use in combination with chemotherapy. We also review important clinical studies that have evaluated the safety and efficacy of these agents in mCRC patients. Despite the discovery of several promising anti-angiogenic agents, mCRC remains an incurable disease with a median overall survival of just over 2 years in patients exposed to all available treatment regimens. Further insights into tumor biology and tumor microenvironment may help improve outcomes in these patients.

Key Words: Anti-angiogenic agents; Metastatic colorectal cancer; Targeted agents; Conversion therapy; Colorectal metastasectomy

Core tip: Colorectal cancer is a major health concern and a leading cause of cancer mortality worldwide. New innovations have provided improved survival in recent years. In this review, we outline the novel anti-angiogenic agents and their respective roles in metastatic colorectal cancer. In addition to three agents approved by the Food and Drug Administration, several alternative anti-angiogenic agents hold promise for use in the metastatic setting.

INTRODUCTION

The past decade has seen a significant decline in the incidence rate and cancer mortality in patients with colorectal cancer (CRC) in the United States. The decrease in cancer deaths appears to be due largely to the widespread use of screening colonoscopy and the availability of better treatment options. However, from a public health perspective, CRC remains a major concern, with 136830 new cases estimated to be diagnosed and over 50000 deaths predicted to occur in the United States alone in 2014. Today, CRC is the third leading cause of cancer mortality in the United States, surpassed only by lung cancer, breast cancer in women and prostate cancer in men[1].

Surgery remains the mainstay of treatment in patients with early stage and locally-advanced CRC and should be considered for those with metastatic CRC (mCRC) with liver-only or lung-only metastases. Though only 10%-20% of patients with liver-only metastases are resectable at the time of diagnosis[2], the use of conversion therapy can make up to 61.9% of tumors resectable[3].

Since the discovery of 5-fluorouracil (5-FU) in 1957[4], several chemotherapeutic agents have been approved for the treatment of mCRC, including capecitabine, oxaliplatin, and irinotecan. Insights into the molecular mechanisms of disease led to the discovery of biologic agents targeting tumor vascular endothelial growth factor (VEGF) and epidermal growth factor receptor (EGFR). This review focuses on the anti-angiogenic agents used in the treatment of mCRC.

BIOLOGICAL BASIS OF ANTI-ANGIOGENIC THERAPY

Tumor cells and endothelial cells are inter-dependent for their growth via a carefully regulated system[5]. Pre-clinical studies have shown that implanted tumor cells can only grow to a size of 2-3 mm without neo-vascularization. They can remain dormant for several years or switch to an angiogenic phenotype[6]. Tumor cells with the angiogenic phenotype release growth factors (pro-angiogenic factors) which stimulate endothelial proliferation, migration, and formation of new capillaries. This process is called tumor angiogenesis and leads to tumor perfusion, growth, and metastases[5,7]. Hematopoietic stem cells and circulating endothelial progenitor cells (CEPs), which are bone marrow derived rapidly proliferating cells, are also thought to contribute to tumor angiogenesis[8].

Vascular endothelial growth factor (VEGF) is one of the most extensively studied pro-angiogenic factors. It is produced by normal and certain neoplastic cells (such as CRC cells)[9]. The human VEGF family is primarily composed of 5 glycoproteins (VEGF A, B, C, D, and platelet derived growth factor, or PlGF). These proteins exert their effects by binding to receptor tyrosine kinases (VEGFR1, R2, and R3)[9-11]. VEGF-A is commonly referred to as VEGF or vascular permeability factor (VPF) and is first discovered by Senger et al[12].

Tissue hypoxia (via hypoxia inducible factor), growth factors (e.g., epidermal growth factor, insulin like growth factor-1), and oncogenes (e.g., c-Src proto-oncogene) increase VEGF expression[9,13,14]. VEGF then exerts its angiogenic effects predominantly via VEGFR2; however, the role of VEGFR1 remains unclear[15]. VEGF promotes tumor angiogenesis by increasing permeability of post-capillary venules, which subsequently leads to the leakage of plasma proteins such as fibrinogen and clotting factors into the extracellular matrix (ECM). Fibrinogen is converted to fibrin in the ECM which leads to increased endothelial cell migration and proliferation[16]. VEGF is also an endothelial cell mitogen[13] and causes endothelial cell proliferation by activating members of the MAP kinase and protein kinase C pathways[9]. Other pro-angiogenic factors include hepatocyte growth factor (HGF), axon guidance factors, interleukins (IL-1, 6, 8, and stromal cell derived factor 1), fibroblastic growth factors (FGF 1 and 2), angiopoietins, and pro-angiogenic chemokines[17]. Another important regulator of angiogenesis is the tyrosine kinase with immunoglobulin and epidermal growth factor homology domain 2 (TIE2) expressed primarily on endothelial cells. TIE2 interacts with angiopoietin 1, angiopoietin 2, VEGF, and FGF to cause maturation of immature blood vessels[18].

The use of anti-angiogenic therapy to arrest tumor growth and thereby make these tumors more susceptible to chemotherapy and cell-mediated immunity was first proposed by Folkman[5] in 1971. Angiogenesis inhibitors can be broadly classified into 2 groups, direct and indirect anti-angiogenic agents. Direct angiogenic inhibitors act on endothelial cells of the microvasculature, thus inhibiting their response to angiogenic stimuli. Indirect angiogenic inhibitors on the other hand target pro-angiogenic stimuli either at the level of the ligand (e.g., VEGF inhibition) or at the level of the receptor (e.g., VEGFR inhibition)[19].

ANTI-ANGIOGENIC AGENTS IN THE TREATMENT OF UNRESECTABLE mCRC

Bevacizumab is an IgG1 monoclonal antibody against the VEGF-A ligand that was developed by humanization of the murine anti-human VEGF antibody A.4.6.1[20,21]. It was the first anti-angiogenic agent to be FDA-approved in the treatment of mCRC in combination with chemotherapy[22]. Adverse effects include hypertension, proteinuria, hemorrhage, GI perforation, delayed wound healing, and arterial and venous thromboembolism[23,24]. Hypertension is a common side effect of bevacizumab therapy, with more than half of the patients requiring pharmacologic intervention. It has been hypothesized that VEGF inhibition leads to a decrease in nitric oxide synthase, leading to inhibition of vasodilation. In addition, bevacizumab decreases arteriolar and capillary perfusion leading to increased peripheral vascular resistance and hypertension[25]. The hemorrhage and thrombosis paradox of bevacizumab therapy can be explained by the disruption in hemostasis secondary to VEGF inhibition. VEGF inhibition leads to apoptosis of quiescent endothelial cells, which in turn leads to activation of the extrinsic coagulation pathway. This mechanism lends credence to the prothrombotic properties of bevacizumab. Inhibition of angiogenesis and platelet function are thought to contribute to hemorrhage and impaired wound healing related to bevacizumab therapy. Proteinuria as a consequence of bevacizumab therapy is common and is secondary to renal thrombotic microangiopathy leading to glomerular endothelial injury[25].

Bevacizumab in treatment-naive patients

Bevacizumab has been extensively studied in several clinical trials with favorable results (Table 1). The combination of bevacizumab and bolus 5-FU/leucovorin (LV) chemotherapy was compared to bolus 5-FU/LV alone in treatment-naive mCRC patients in a phase II randomized study by Kabbinavar et al[26] in the year 2003. The addition of low-dose bevacizumab led to higher response rates (40% vs 17%), longer time to disease progression (9 mo vs 5.2 mo), and a longer median overall survival (OS) (21.5 mo vs 13.8 mo) in these patients[26]. In another phase II trial comparing first-line bevacizumab plus chemotherapy (bolus 5-FU/LV) to chemotherapy alone in mCRC patients who were poor candidates for irinotecan therapy, a 3.7 mo progression-free survival (PFS) advantage was noted in the group that received bevacizumab (9.2 mo vs 5.5 mo; HR, 0.50; P = 0.0002). There was a trend toward a longer median OS in the bevacizumab-containing group; however; this difference was not statistically significant (16.6 mo vs 12.9 mo; HR, 0.79; P = 0.16)[27].

Table 1 Bevacizumab in the first-line setting in metastatic colorectal cancer.

Ref.	Regimen	PFS (mo)	P value	OS (mo)	P value
Kabbinavar et al[26]; Phase II	1Bolus 5-FU/LV ± bevacizumab	9 vs 5.2 (TTP3)	NA	21.52vs 13.8	NA
Kabbinavar et al[27]; Phase II	1Bolus 5-FU/LV + bevacizumab vs bolus 5-FU/LV + placebo	9.2 vs 5.5	0.0002	16.6 vs 12.9	0.16
Hochster et al[35]; Phase II (TREE-1)	mFOLFOX6/bFOL/CapeOX	8.7/6.9/5.9 (TTP3)	N/A4	19.2/17.9/17.2 18.2 (overall)	N/A4
Hochster et al[35]; Phase II (TREE-2)	mFOLFOX6 + bevacizumab/bFOL + bevacizumab/CapeOX + bevacizumab	9.9/8.3/10.3 (TTP3)		26.1/20.4/24.6 23.7 (overall)
Hurwitz et al[22]; Phase III	IFL + bevacizumab vs IFL + placebo	10.6 vs 6.2	< 0.001	20.3 vs 15.6	< 0.001
Stathopoulos et al[115]; Phase III	IFL ± bevacizumab	NA	NA	22 vs 25	0.1391
Saltz et al[29]; Phase III	FOLFOX/CapeOX + bevacizumab vs FOLFOX/CapeOX + placebo	9.4 vs 8	0.0023	21.3 vs 19.9	0.077

¹Roswell Park regimen: LV 500 mg/m² over 2 h and FU 500 mg/m² as a bolus midway through the LV infusion;

²Data presented is on patients who received chemotherapy plus low-dose bevacizumab;

³Time to progression;

⁴Comparison between outcomes of TREE-1 and TREE-2 is not possible as they were sequential cohorts. 5-FU: 5-Fluorouracil; LV: Leucovorin; PFS: Progression-free survival; OS: Overall survival; mFOLFOX6: Three fluoropyrimidine regimens-infusional 5FU/LV; bFOL: Bolus FU/LV; CapeOX: Capecitabine; FOLFOX: 5-FU/LV/oxaliplatin; NA: Not available; N/A: Not applicable.

Subsequently, a large randomized phase III trial compared the use of bevacizumab plus irinotecan, bolus 5-FU/LV (IFL) vs IFL plus placebo as front-line therapy. The addition of bevacizumab not only conferred a benefit in median OS (20.3 mo vs 15.6 mo; HR, 0.66; P < 0.001) and PFS (10.6 mo vs 6.2 mo; HR, 0.54; P < 0.001), but also led to higher response rates (44.8% vs 34.8%; P = 0.004) and more durable responses (10.4 mo vs 7.1 mo; P = 0.001)[22]. The results of this trial led to the FDA approval of bevacizumab for use as a first-line agent in mCRC patients in combination with chemotherapy. In a combined analysis of 2 phase II (53, 54) and 1 phase III study[22], patients in the 5-FU/LV/bevacizumab arm had a statistically significant improvement in median OS (17.9 mo vs 14.6 mo; HR, 0.74; P = 0.008) and median PFS (8.8 mo vs 5.6 mo; HR, 0.63; P≤ 0.0001) when compared to the chemotherapy-only arm (patients receiving 5-FU/LV or IFL)[28]. However, in another phase III randomized trial comparing IFL with and without bevacizumab, the addition of bevacizumab did not confer an OS advantage (22 mo in the IFL-bevacizumab arm vs 25 mo in the IFL arm; P = 0.1391)[28]. With the emergence of combination chemotherapy regimens [5-FU/LV/oxaliplatin (FOLFOX), and 5-FU/LV/irinotecan (FOLFIRI)], subsequent studies focused on testing the efficacy and safety of these regimens in combination with bevacizumab. In a randomized phase III study by Saltz et al[29], untreated mCRC patients were randomized to receive either bevacizumab or placebo in combination with chemotherapy (FOLFOX-4 or Cape-OX). Though the effect size was small, a PFS advantage was seen in the bevacizumab-containing arm (9.4 mo vs 8 mo; HR, 0.83; P = 0.0023), however there was no statistically significant difference in median OS between the two groups (21.3 mo vs 19.9 mo; HR, 0.89; P = 0.077). An interesting observation in this study that the authors effectively point out is the similar median treatment duration of patients receiving bevacizumab and placebo (approximately 6 mo), in contrast to the significantly longer PFS (as noted above) in the bevacizumab arm. The early discontinuation of bevacizumab (prior to disease progression) probably explains the absence of a survival advantage in the bevacizumab-containing arm. The authors concluded that continuation of bevacizumab until disease progression is critical for a meaningful clinical benefit[29].

A randomized head-to-head comparison of FOLFIRI with and without bevacizumab has not been done to date. However, sufficient evidence to justify the use of FOLFIRI plus bevacizumab in untreated mCRC patients exists. In a pooled analysis on 29 published trials, patients who received FOLFIRI-bevacizumab had a median PFS of 10.8 mo (95%CI: 8.9-12.8) and a median OS of 23.7 mo (95%CI: 18.1-31.6)[30]. In an open-label, phase IV AVIRI study, patients who received first-line FOLFIRI plus bevacizumab had a PFS of 11.1 mo and a median OS of 22.2 mo[31]. A phase III trial of 285 patients compared efficacy of CapeIri plus bevacizumab with FOLFIRI plus bevacizumab. There was no difference in PFS (10.2 mo vs 10.8 mo; P = 0.74), or median OS (20.0 mo vs 25.3 mo, P = 0.099) between the two groups[32].

After the Gruppo Oncologico Nord Ovest (GONO) group showed that 5-FU/LV/oxaliplatin/irinotecan (FOLFOXIRI) improved response rate (RR), PFS, and OS in treatment-naive unresectable mCRC patients in a phase III randomized trial[33], the addition of bevacizumab to FOLFOXIRI was compared to FOLFIRI plus bevacizumab by the same group of investigators. The latter trial was also a phase III randomized trial (TRIplet plus BEvacizumab, or TRIBE), which showed that patients receiving triplet chemotherapy (FOLFOXIRI) plus bevacizumab had a longer PFS (primary end point; 12.1 mo vs 9.7 mo; HR, 0.75; P = 0.003) and better objective response rate (65% vs 53%; P = 0.006) when compared to those receiving FOLFIRI plus bevacizumab. Though patients in the FOLFOXIRI arm had a longer median OS when compared to those in the FOLFIRI arm, this difference was not statistically significant (31.0 mo vs 25.8 mo; HR, 0.79; P = 0.054). Not surprisingly, patients who received the triplet chemotherapy regimen plus bevacizumab had a significantly higher incidence of grade 3-4 neutropenia, diarrhea, stomatitis, and peripheral neuropathy when compared to the FOLFIRI plus bevacizumab arm[34].

The Three Regimens for Eloxatin Evaluation (TREE) study was initially designed to evaluate the efficacy and safety of Oxaliplatin (Eloxatin) in combination with three fluoropyrimidine regimens- infusional 5FU/LV (mFOLFOX6), bolus FU/LV (bFOL), and Capecitabine (CapeOX). When the trial was nearing completion of accrual, data on the efficacy of bevacizumab in mCRC began to emerge. The study was therefore modified to include 2 sequential cohorts of patients- the initial cohort of patients who did not receive bevacizumab (TREE-1) and a subsequent cohort of patients who received bevacizumab in combination with one of the above three chemotherapy regimens (TREE-2). The incidence of serious (grade 3/4) treatment related AEs in the first 12 wk of therapy in each of the patient groups in the TREE-2 cohort (primary end point) were 59% (mFOLFOX6/bevacizumab), 51% (bFOL/bevacizumab), and 56% (CapeOX/bevacizumab), with neutropenia, diarrhea, and nausea/vomiting being the most common AEs in each of the treatment groups respectively. The respective incidence of grade 3/4 AEs in the TREE-1 cohort were 59% (mFOLFOX6), 36% (bFOL), and 67% (Cape-OX). The overall median OS was nearly 2 years (23.7 mo) in the TREE-2 cohort, and 18.2 mo in the TREE-1 cohort[35].

Two randomized phase III trials (FIRE-3 and CALGB 80405) compared the efficacy of cetuximab vs bevacizumab in combination with chemotherapy in previously untreated KRAS WT mCRC patients. The FIRE-3 study randomized patients with KRAS WT exon 2 tumors to receive either cetuximab plus FOLFIRI or bevacizumab plus FOLFIRI as front-line therapy. Though the objective response (CR/PR; primary end point) and median PFS were similar between the two groups, median OS favored the cetuximab-containing group (28.7 mo vs 25.0 mo; HR, 0.77; P = 0.017)[36]. The CALGB 80405 trial randomized untreated mCRC patients with KRAS WT (codons 12 and 13) tumors to receive either cetuximab or bevacizumab in combination with chemotherapy (FOLFIRI or mFOLFOX6). The OS (primary end point) and PFS were similar in both groups and the authors concluded that either regimen would be an appropriate option in these patients. It is important to note that in contrast to the FIRE-3 study, most patients (73.4%) in the CALGB 80405 study received mFOLFOX6 as their combination chemotherapy regimen[37].

Bevacizumab as maintenance therapy

Maintenance treatment in advanced CRC for the Treatment of Digestive Tumors (MACRO TTD) was the first randomized phase III study undertaken to evaluate the role of bevacizumab alone in the maintenance setting. Patients were randomized to receive either bevacizumab alone vs bevacizumab plus maintenance chemotherapy (Cape-OX), after completion of induction therapy (Cape-OX + bevacizumab, or Cape-OX-B). The primary end point was PFS and the prespecified non-inferiority limit of HR for PFS was set at 1.32. After a median follow-up of 29 mo, median PFS in patient receiving maintenance Cape-OX-B vs Bevacizumab alone was 10.4 mo and 9.7 mo respectively. The HR for PFS was 1.10 with a 95%CI: 0.89-1.35. The study thus did not confirm non-inferiority of bevacizumab maintenance when compared to Cape-OX-B as the upper limit of the 95%CI of HR for PFS exceeded the pre-specified limit of 1.32. However, there was no statistically significant difference in PFS, OS, and response rate between the two arms, with a significantly lower frequency of grade 3-4 sensory neuropathy in the bevacizumab alone group (8% vs 26%; P < 0.0001)[38].

Subsequently, the role of bevacizumab maintenance therapy in patients who had stable disease/partial response (PR)/complete response (CR) after bevacizumab-containing induction chemotherapy was evaluated in a multicenter retrospective analysis of treatment-naive mCRC patients. The study results favored bevacizumab maintenance over no maintenance therapy (PFS: 13 mo vs 8 mo; P < 0.0001). An OS advantage was only seen in those patients who received bevacizumab maintenance after they had an objective response to induction chemotherapy[39].

More recently, the role of bevacizumab plus chemotherapy as maintenance therapy was investigated in the phase III CAIRO3 trial. After completion of six cycles of Cape-OX-B, patients were randomized to either receive maintenance therapy with capecitabine plus bevacizumab (Cape-B) or receive no further therapy. Irrespective of randomization, patients who had first progression (PFS1) received Cape-OX-B until second progression (PFS2). After a median follow-up of 2 years, maintenance therapy conferred a PFS advantage (PFS1: 8.5 mo vs 4.1 mo; P < 0.0001; PFS2: 11.7 mo vs 8.5 mo; P < 0.0001)[40]. In patients with baseline synchronous metastases and resected primary tumor, an OS benefit was noted as well (25 mo vs 18 mo; P < 0.0001)[40].

An ongoing randomized phase III trial (NCT00973609) is evaluating three treatment strategies in mCRC patients. All patients will receive induction (and re-induction) with a 5-FU, oxaliplatin, and bevacizumab-based chemotherapy for a period of 6 mo. Induction therapy will be followed by maintenance therapy with a fluoropyrimidine and bevacizumab (active comparator), or bevacizumab alone (experimental arm) or no maintenance therapy (experimental arm).

Bevacizumab in the second-line setting

A multi-center, randomized phase III E3200 study was pivotal in bevacizumab’s approval in previously treated mCRC patients (Table 2). This study randomized patients who were previously treated with fluoropyrimidine and irinotecan to received FOLFOX-4 plus bevacizumab (group 1), FOLFOX-4 alone (group 2), or bevacizumab alone (group 3). Patients in group 1 had a longer median OS and a better PFS when compared to patients in group 2 (group 1 vs group 2; OS: 12.9 mo vs 10.8 mo; P = 0.0011; PFS: 7.3 mo vs 4.7 mo; P < 0.0001) and group 3 (group 1 vs group 3; OS: 12.9 mo vs 10.2 mo; PFS: 7.3 mo vs 2.7 mo)[41].

Table 2 Phase III trials using bevacizumab in the second-line setting.

Ref.	Regimen	PFS (mo)	P value	OS (mo)	P value
Giantonio et al[41]; (E3200)	FOLFOX4 ± bevacizumab	7.3 vs 4.7	< 0.0001	12.9 vs 10.8	0.0011
Bennouna et al[45]; (ML18147)	Chemotherapy ± bevacizumab	5.7 vs 4.1	< 0.0001	11.2 vs 9.8	0.0062
Masi et al[46]; (BEBYP)	Chemotherapy ± bevacizumab	6.8 vs 5.0	0.010	14.1 vs 15.51	0.0431

¹The lower median OS in the bevacizumab arm was due to intersection of curves; adjusted HR was 0.77 (stratified log-rank P = 0.043) and favored the bevacizumab arm. PFS: Progression-free survival; OS: Overall survival.

Bevacizumab beyond progression: The rationale behind continuing bevacizumab despite progression on bevacizumab-containing chemotherapy is that the mechanisms of resistance to cytotoxic chemotherapy and to bevacizumab differ significantly and may not necessarily occur concomitantly[42,43]. Changes in tumor cell biology and genetic instability via mutations in tumor suppressor genes or of drug targets, contribute to chemotherapy resistance. As anti-VEGF therapy targets the genetically stable tumor microvasculature, emergence of resistance to bevacizumab requires development of alternative proangiogenic signaling[43]. Thus, it is not unreasonable to assume that progression of disease on a combination treatment regimen (cytotoxic chemotherapy plus bevacizumab) may be secondary to resistance to chemotherapy alone and continuation of bevacizumab beyond progression in combination with a different chemotherapy regimen may be an option. This hypothesis was validated by two large observational studies[42,44] and large phase III study (ML 18147)[45]. The BRiTE (Bevacizumab Regimens: Investigation of Treatment Effects and Safety) study was a large observational cohort study undertaken to evaluate the role of bevacizumab continuation beyond disease progression. The study enrolled 1445 mCRC patients who had progression on a first-line bevacizumab-containing treatment regimen. Patients who had received “Bevacizumab Beyond Progression” (BBP: n = 642) had a significantly longer median OS when compared to those who discontinued bevacizumab (no-BBP; n, 531) therapy (median OS: 31.8 mo vs 19.9 mo; HR, 0.49; P < 0.001). As would have been expected, patients in the BBP group had a higher rate of hypertension requiring medication compared to the no-BBP group or to the overall study population (24.6% vs 19.2%), however, the risk of serious AEs including arterial thromboembolic events, grade 3 or 4 bleeding, and GI perforation were similar between the two groups[42].

The ARIES study was another observational study that confirmed that findings of the BRiTE study. In this study, a total of 1105 patients survived longer that 2 mo after first progression, and were included in the modified ITT analysis. The median post-progression survival was higher in the BBP group when compared to the no-BBP group [14.4 mo vs 10.6 mo; multivariable HR (95%CI): 0.84 (0.73-0.97)]. Protocol-specified adverse events were higher in the BBP group vs the no-BBP group (13% vs 8.5%)[44].

In order to validate the results of the BRiTE and ARIES studies, a multinational phase III trial (ML18147) randomized mCRC patients with POD within 3 mo of discontinuation of 1^st line bevacizumab plus chemotherapy, to receive 2^nd line chemotherapy while continuing BBP vs chemotherapy alone. A total of 819 patients were included in the ITT analysis. After a median follow-up of 11.1 and 9.6 mo in the chemotherapy plus BBP group and chemotherapy alone group respectively, the median OS (primary end point) significantly favored the bevacizumab containing arm [11.2 mo vs 9.8 mo; HR (95%CI): 0.81 (0.69-0.94); unstratified log-rank test, 0.0062] (Table 2). Patients receiving BBP had a higher rate of grade 3-5 bleeding (2% vs < 1%), GI perforation (2% vs < 1%), and VTE (5% vs 3%), but as is evident from the frequency of these AE, the difference between the two arms was not considerable. Neutropenia (16% vs 13%), diarrhea (10% vs 8%), and asthenia (6% vs 4%) were the most common grade 3-5 adverse events and were comparable between the two arms. Also, the rate of arterial thromboembolism was not increased in the BBP group when compared to the no-BPP group. Thus, continuation of bevacizumab beyond first progression in mCRC patients significantly improved median OS without substantially increasing serious AEs[45].

The Bevacizumab Beyond Progression (BEBYP) trial (Table 2) was a phase III, prospective, multicenter Italian study that evaluated the efficacy and safety of continuation or reintroduction of bevacizumab after first progression in patients with unresectable mCRC. The sample size was much smaller when compared to the ML18147 trial, but also included patients with POD beyond 3 mo of discontinuation of first-line therapy. PFS was the primary end point and 184 patients were included in the ITT analysis. After a median follow-up of 45.3 mo, the median PFS was noted to be significantly higher in the bevacizumab group when compared to the chemotherapy-only group (6.8 mo vs 5.0 mo; HR, 0.70; stratified log-tank P = 0.010). PFS benefit persisted when patients were stratified based on the bevacizumab-free interval (≤ 3 mo vs > 3 mo). An OS advantage was also noted in the bevacizumab group (adjusted HR, 0.77; stratified log-rank P = 0.043), though responses were comparable between the two arms (17% in the chemotherapy arm vs 21% in the bevacizumab arm; P = 0.573). Consistent with the safety data of the ML18147 trial, grade 3-4 AEs were similar between both arms[46].

Bevacizumab-based chemotherapy in the elderly

The efficacy and tolerability of bevacizumab in the elderly has been studied both in the first- and second-line settings. The BRiTE study was a large observational cohort study of 1953 untreated mCRC patients with 896 patients ≥ 65 years of age. PFS in the elderly patients was similar to their younger counterparts though median OS declined with increasing age[47]. Interestingly however, in another large observational cohort study of 1777 treatment-naive German mCRC patients, those ≥ 75 years of age had a significantly lower PFS and median OS when compared to those < 75 years of age (PFS: 10.5 mo vs 8.9 mo; P =0.00019; OS: 25.8 mo vs 20.8 mo; P <0.0001)[48]. In a multicenter phase II study by the Hellenic Oncology Research Group, the combination of capecitabine, oxaliplatin, and bevacizumab (AVELOX) was proven to be safe and effective in the first-line treatment of elderly patients (≥ 70 years old)[49].

In a pooled analysis of 439 untreated mCRC patients ≥ 65 years old, bevacizumab-based chemotherapy produced a PFS and OS advantage when compared to chemotherapy alone[50]. In another retrospective pooled analysis of 4 RCTs (3 RCTs in the first-line setting and 1 RCT in the second-line setting), the addition of bevacizumab conferred a PFS and OS advantage in elderly patients (≥ 65 and ≥ 70 years old) when compared to chemotherapy alone. Patients receiving bevacizumab had more arterial thromboembolic events; however, there was no increase in ≥ grade 3 adverse events with increasing age[51].

More recently, the safety of bevacizumab-based chemotherapy was studied in a multi-national phase III randomized trial (AVEX) in which 280 patients with a median age of 76 years were randomized to receive bevacizumab plus capecitabine vs capecitabine alone. Progression free survival favored the bevacizumab-containing arm (9.1 mo vs 5.1 mo; P < 0.0001). Overall, the percentage of patients who had any grade treatment related adverse events was similar in both groups (84% vs 81% in the bevacizumab-containing arm vs the chemotherapy alone arm). However, a higher percentage of patients in the bevacizumab plus chemotherapy group had grade 3 or greater treatment-related adverse events when compared to the chemotherapy alone group (40% vs 22%). Not surprisingly, bevacizumab-specific any grade adverse effects such as hypertension, proteinuria, and venous thromboembolism were greater in the bevacizumab containing arm[52].

Ziv-aflibercept (VEGF trap): A human recombinant soluble decoy protein that was engineered by the fusion of the second immunoglobulin (Ig) domain of VEGFR1 and the third Ig domain of VEGFR2 with the constant region (Fc) of human IgG1[53]. The drug binds to VEGF-A, VEGF-B, and placental growth factor (PlGF) with high affinity, thus preventing these ligands from binding to their respective endogenous receptors[54]. This leads to tumor growth and angiogenesis inhibition as shown in in-vitro and in-vivo studies[53]. When compared to bevacizumab, in addition to inhibiting endothelial cell migration, ziv-aflibercept has a much greater binding affinity to VEGF-A and more potent inhibition of VEGFR1 and VEGFR2 activation[54]. Adverse effects include fatigue, headache, hemorrhage, nausea, diarrhea, hypertension, and proteinuria[55-57].

The efficacy of ziv-aflibercept in mCRC patients was evaluated in a large randomized phase III trial (VELOUR). The study included all mCRC patients who progressed after prior oxaliplatin-based therapy for metastatic disease or who relapsed within 6 mo of adjuvant oxaliplatin-based chemotherapy. Prior bevacizumab therapy was not an exclusion criterion, though prior irinotecan therapy was not allowed. Patients who received prior bevacizumab therapy constituted 30.6% of the intent to treat (ITT) population. Patients were randomized to receive FOLFIRI plus ziv-aflibercept (ziv-aflibercept arm) vs FOLFIRI plus placebo (control arm). After a median follow-up of 22.28 mo, patients in the ziv-aflibercept arm had a significantly longer median OS (13.50 mo vs 12.06 mo; P = 0.0032) and PFS (6.90 mo vs 4.67 mo; HR, 0.758, P < 0.0001) when compared to the control arm. Neither prior bevacizumab use nor ECOG PS had an interaction with treatment for OS or PFS. OS and PFS advantage with ziv-aflibercept vs placebo was noted regardless of prior bevacizumab exposure [prior bevacizumab therapy: median OS: 12.5 mo vs 11.7 mo; HR (95%CI): 0.862 (0.673-1.104); median PFS: 6.7 mo vs 3.9 mo; HR (95%CI): 0.661 (0.399-1.095); no prior bevacizumab therapy: Median OS: 13.9 mo vs 12.4 mo; HR (95%CI): 0.788 (0.669-0.927); Median PFS: 6.9 mo vs 5.4 mo; HR (95%CI): 0.797 (0.58-1.096)]. Grade 3 and 4 adverse events that were higher in the ziv-aflibercept arm included hypertension, hemorrhage, thromboembolic events (arterial and venous)[56]. In a post-hoc subset analysis of the VELOUR trial, patients with liver-only metastases had a greater OS and PFS benefit from ziv-aflibercept in comparison to patients with no liver metastasis or liver plus other organ metastases. Prior bevacizumab therapy did not have an influence on treatment effect[58]. Ziv-aflibercept in combination with FOLFIRI is FDA approved for use in the treatment of mCRC patients who have progressed through or following a first-line oxaliplatin-based regimen.

Regorafenib: A biaryl-urea compound which functions as an oral multikinase inhibitor of angiogenic (VEGF R1-3, tyrosine kinase with immunoglobulin-like and epidermal growth factor-like domains, or TIE2), stromal (PDGF-β, fibroblast growth factor receptor 1), and oncogenic (RET, KIT, BRAF) receptor tyrosine kinases[18,59]. The safety and efficacy of regorafenib was first demonstrated in humans in a phase I dose-escalation study which enrolled 53 patients with advanced and refractory solid tumors. The maximum tolerated dose was determined to be 160 mg daily, with a 3 wk on, 1 wk off schedule every 4 wk. More than half of the patients (66%) had either partial response or stable disease per RECIST criteria. The most common drug-related adverse events noted were voice changes, hand-foot syndrome (HFS), mucositis, diarrhea, and hypertension. Most patients (83%) developed at least 1 treatment related AE. The most frequently observed grade 3 or 4 drug-related AEs were hand-foot skin reaction (HFS), skin rash, hypertension, and diarrhea[60].

Another phase I dose-escalation and extended cohort trial enrolled 37 patients with advanced or mCRC refractory to standard therapy and 1 patient with treatment-naive disease who refused standard therapy. Of the 27 patients evaluable for response, 19 had stable disease and 1 had partial response. The median PFS was 107 d (95%CI: 66-161). As in the prior phase 1 trial, most patients (84%) had treatment-related AEs (HFS, skin rash/desquamation, fatigue, fatigue, voice changes, diarrhea), though most of the AEs were grade 3 or lower. HFS was the most common grade 3 or greater treatment related AE. More than half of the patients (66%) required dose reduction or treatment interruption due to AEs with HFS being the most common AE requiring dose reduction[61].

The efficacy and safety of regorafenib in combination with FOLFOX or FOLFIRI chemotherapy in the 1^st or 2^nd line setting was evaluated in a phase Ib trial of 45 mCRC patients. Of the 38 patients evaluable for treatment response, either partial response or stable disease was noted in 33 patients. Median TTP for the study population was 119 d (FOLFOX group: 116 d; FOLFIRI group: 186.5 d). Most patients (71%) had treatment related AEs that were grade 3 or higher, of which neutropenia was the most common AE. Common any grade AEs included diarrhea, mucositis, neutropenia, HFS, alopecia, and fatigue. Interestingly the area under the curve (AUC) of irinotecan and its active metabolite (SN-38) were significantly higher in cycle 2 when compared to cycle 1 prior. Overall, regorafenib was shown to have acceptable tolerability in combination with chemotherapy in this study[62].

The CORRECT trial was a randomized multinational phase III trial evaluating the benefit and tolerability of regorafenib in previously treated mCRC patients after failure of standard licensed therapy. A total of 760 patients were randomized in a 2:1 ratio to receive regorafenib plus BSC or placebo plus BSC. The mean treatment duration was 2.8 mo in the regorafenib arm and 1.8 mo in the placebo arm. Either partial response or stable disease was achieved in 41% of patients in the regorafenib group compared to 15% of patients in the placebo arm (P < 0.0001). Median OS was 6.4 mo vs 5 mo (HR, 0.77; P = 0.0052) and median PFS was 1.9 mo vs 1.7 mo (HR, 0.49; P < 0.0001) in the regorafenib group vs placebo arm, respectively. Though the magnitude of OS benefit with regorafenib vs placebo appears small, the HR of 0.77 would imply a 23% reduction in the risk of death during the study period. When stratified based on the primary site of disease, patients with colon cancer who received regorafenib had a significant OS advantage with an HR of 0.70 and 95%CI: 0.56-0.89, when compared to the placebo arm, however this benefit was not seen in patients with rectal cancer [HR (95%CI), 0.95 (0.63-1.44)]. PFS favored the regorafenib arm in colon, rectal, and colon and rectal cancer subgroups with an HR of 0.55, 0.45, and 0.35 respectively. The apparent lack of an OS advantage despite a PFS advantage in patients with rectal cancer could be attributed to the higher percentage of patients in the placebo group and the smaller proportion of patients in the regorafenib group went on to receive further anti-cancer therapies post-study. Ninety three percent of patients in the regorafenib group vs 61% of those in the placebo group had treatment-related adverse events. Hand-foot syndrome, fatigue, diarrhea, hypertension, skin rash or desquamation were the most frequent toxicities that were ≥ grade 3. Hepatotoxicity with elevated liver transaminases and bilirubin (mostly grade 1-2) was noted to be more common in the regorafenib group when compared to placebo. A fatal case of liver injury in a 62-year-old male with liver metastases was noted in the regorafenib arm 43 d after treatment initiation. Health-related quality of life and health outcomes were measured using standard scoring systems and showed no difference in deterioration in the regorafenib vs placebo arms. Regorafenib monotherapy appears to be a reasonable option in patients with refractory mCRC who have exhausted all other systemic treatment options[63].

Ramucirumab: A human IgG1 monoclonal Ab against the extracellular domain of VEGFR2, thereby preventing the interaction between VEGF and VEGFR2[64]. The efficacy and safety of Ramucirumab was initially reported in a phase I study of 37 advanced solid tumor patients, of whom 6 had a primary CRC (refractory to standard therapy). After at least 12 wk of therapy, three of the six (50%) CRC patients experienced SD for 30 (dose level: 2 mg/kg), 31 (dose level: 4 mg/kg), and 15 wk (dose level: 10 mg/kg) respectively. Overall, 22 patients (60%) developed grade 3-5 AEs, with hypertension, abdominal pain, anorexia, vomiting, increased blood alkaline phosphatase, headache, proteinuria, dyspnea, and deep venous thrombosis being the common serious AE[65]. Subsequently, a phase II study enrolled 42 treatment-naive mCRC patients to receive Ramucirumab (at a dose of 8 mg/kg) in combination with mFOLFOX6 every 2 wk. The combination was shown to be efficacious with a median PFS of 11.5 mo and a median OS of 20.4 mo (Table 3). Neutropenia, hypertension, and neuropathy were the most commonly reported serious (grade 3-4) AEs[66]. The benefit of the addition of Ramucirumab to FOLFIRI in the second-line setting was recently evaluated in a large, randomized double-blind phase III study (RAISE). The trial enrolled a total of 1072 patients who had POD during or after first line therapy with a combination of bevacizumab, fluoropyrimidine, and oxaliplatin, were randomized in a 1:1 design to receive FOLFIRI plus Ramucirumab vs FOLFIRI plus placebo. Patients in the Ramucirumab arm had a longer median OS (primary end-point; 13.3 mo vs 11.7 mo; HR, 0.84; log-rank P = 0.0219) and a longer PFS (5.7 mo vs 4.5 mo; HR, 0.79; log-rank P = 0.0005). The commonly reported serious AE (≥ grade 3) included neutropenia, hypertension, diarrhea, and fatigue[67]. Ramucirumab in combination with FOLFIRI is a promising second-line treatment option in patients with unresectable mCRC.

Table 3 Other anti-angiogenic agents.

Ref.	Regimen (line of treatment)	Sample size	Objective response (%)	PFS (mo)	OS (mo)	Serious AE (grade 3-4)5
Samalin et al[73]; Phase I/II	Sorafenib/irinotecan (NEXIRI) (2nd or later line KRAS mutated)	10 (phase I)	64.9 (DCR)	3.7	8	Asthenia, diarrhea, neutropenia, HFS
		54 (phase II)
Tabernero et al[74]; Phase IIb	Sorafenib/mFOLFOX vs Placebo/mFOLFOX (1st line)	198	NA	9.1 vs 8.7	17.6 vs 18.1	Neutropenia, peripheral neuropathy, HFS
				HR, 0.88	HR, 1.13
				P = 0.46	P = 0.51
Starling et al[116]; Phase I	Sunitinib/FOLFIRI (1st line)	37	57.9	NA	NA	Febrile neutropenia neutropenia, anemia, diarrhea, mucosal inflammation, stomatitis, vomiting, lethargy, pyrexia, thrombotic events
Yoshino et al[117]; Phase I	Sunitinib/mFOLFOX6 (1st line)	12 (6 + 6)3	66.7 in each arm	NA	NA	Neutropenia, leukopenia, thrombocytopenia
Saltz et al[118]; Phase II	Sunitinib (refractory setting)	43 (prior bevacizumab)	2.4	2.2 (TTP; prior bevacizumab)	7.1	Fatigue, diarrhea, nausea, vomiting, and anorexia (most common any grade toxicities)
		41 (no prior bevacizumab)	0	2.5 (TTP; no prior bevacizumab)	10.2
Tsuji et al[75]; Phase II	Sunitinib/FOLFIRI (1st line)	71	36.61/42.32	6.71/ 7.22	NR due to early study closure	Neutropenia, leukopenia, thrombocytopenia diarrhea, nausea decreased appetite and fatigue (most common any grade)
Carrato et al[119]; Phase III	Sunitinib/FOLFIRI vs Sunitinib/placebo (1st line)	768	NA	7.8 vs 8.4 HR 1.095 one-sided stratified log-rank P = 0.807	20.3 vs 19.8 HR, 1.171 P = 0.916	Diarrhea, stomatitis/oral syndromes, fatigue, HFS, neutropenia, thrombocytopenia, anemia, febrile neutropenia
Michael et al[79]; Phase I	Vandetanib/mFOLFOX6 (1st or 2nd line)	9 (100 mg/d dose)	44.44	NA	NA	Diarrhea, nausea and lethargy (most common any grade toxicities)
		8 (300 mg/d dose)	NA	NA	NA
Saunders et al[80]; Phase I	Vandetanib/FOLFIRI	11 (100 mg/d dose)	18.18	NA	NA	Diarrhea, nausea fatigue (most common any grade toxicities; were grade 1-2)
	(1st or 2nd line)	10 (300 mg/d dose)	NA	NA	NA
Yang et al[81]; Phase II	Vandetanib/mFOLFOX6 vs Placebo/mFOLFOX6	32 (100 mg/d dose)4	NA	NA	NA	Diarrhea, nausea, thrombocytopenia, peripheral sensory neuropathy (most common any grade toxicities)
		35 (300 mg/d dose)4
Van Cutsem et al[84]; Phase III	FOLFOX 4/Vatalanib vs FOLFOX4/placebo (2nd line)	855	NA	5.6 vs 4.2	13.1 vs 11.9	Neutropenia, HTN, diarrhea, fatigue, nausea, vomiting, dizziness
				HR, 0.83	HR, 1.0
				P = 0.013	P = 0.957
Hecht et al[85]; Phase III	FOLFOX4/Vatalanib vs FOLFOX4/placebo (1st line)	1168	NA	7.7 vs 7.6	21.4 vs 20.5	Neutropenia, HTN, diarrhea, fatigue, nausea, vomiting
				HR, 0.88	HR, 1.08
				P = 0.118	P = 0.260

¹By independent review;

²Investigator initiated review;

³Six patients received sorafenib 2 wk on, 2 wk off and another 6 patients received sorafenib 4 wk on, 2 wk off;

⁴Progression events (objective/clinical progression/death) in vandetanib 100 mg arm vs placebo: 72% vs 65% (HR, 1.21; 2-sided P = 0.53); vandetanib 300 mg arm vs placebo: 77% vs 65% (HR, 1.41; 2-sided P = 0.25);

⁵In study drug containing arm. DCR: Disease control rate; NA: Not available; NR: Not reached; PFS: Progression-free survival; OS: Overall survival; mFOLFOX6: Three fluoropyrimidine regimens-infusional 5FU/LV; HFS: Hand-foot syndrome.

Famitinib: A small molecule multi-tyrosine kinase inhibitor with predominantly antiangiogenic properties[68]. The drug inhibits VEGFR2 and VEGFR3, PDGFR, stem cell factor receptor c-KIT, FMS-like tyrosine kinase-3 receptor (FLT3), and the proto-oncogene tyrosine-protein kinase inhibitor RET[69,70]. The tolerability of famitinib in patients with advanced solid tumor malignancies was evaluated in a phase I study of 44 patients, including 7 patients with advanced CRC. The most common grade 3-4 toxicities at occurring in the first 8 wk of therapy dose levels of 24, 25, and 27 mg included hypertension, bone marrow suppression leading to leukopenia, neutropenia, thrombocytopenia, and anemia, HFS, hypertriglyceridemia, and proteinuria. The authors recommended a dose of 25 mg for a phase II trial. The efficacy data in patients with advanced CRC was not reported in this study[69]. More recently, the efficacy and safety of famitinib in the third or later line setting was studied in a multi-center phase II, randomized, double-blind study of 154 advanced CRC patients. Patients were randomized in a 2:1 design to receive either famitinib or placebo. Patients who received famitinib have a longer median PFS (primary end point; 2.8 mo vs 1.5 mo; HR, 0.58; P = 0.0034) and a better disease control rate (57.58% vs 30.91%; P = 0.0023) when compared to the placebo arm. The most commonly reported AEs were predominantly grade 1-2 and included neutropenia, thrombocytopenia, hypertension, proteinuria, and HFS. There was no significant difference in serious AEs between the two arms. Famitinib was thus noted to be efficacious and safe in mCRC patients who have failed second or later line therapies[68]. The results will however require further validation with a phase III trial.

OTHER ANTI-ANGIOGENIC AGENTS

Several other antiangiogenic agents have been studied in patients with advanced CRC with disappointing results.

Sorafenib is an orally administered small molecule multi-tyrosine kinase inhibitor which targets the RAF/MEK/ERK pathway in addition to inhibiting several receptor tyrosine kinases including VEGFR2, VEGFR3, PDGR beta, c-KIT, FLT3, and tyrosine kinase colony-stimulating factor 1 receptor (c-Fms)[71]. Adverse effects include HFS, fatigue and diarrhea[72]. In a phase I/II trial evaluating the benefit of sorafenib plus irinotecan in previously treated mCRC patients, the combination was shown to be well tolerated in both phases of the trial. In phase 2, an encouraging response rate of 64.9% was noted with a PFS of 3.7 mo and a median OS of 8 mo[73]. However, in a subsequent phase IIb study of 198 treatment naive mCRC patient, the combination of sorafenib and mFOLFOX4 was shown to offer no PFS or OS advantage over placebo (Table 3)[74].

Sunitinib is an oral multi-tyrosine kinase inhibitor targeting VEGFR1, VEGFR2, VEGFR3, PDGFR alpha and beta, FLT3, stem cell factor receptor, colony stimulating factor receptor, and glial cell line-derived neurotrophic factor[75]. The efficacy and tolerability of sunitinib as monotherapy and in combination with chemotherapy was studied without significant benefit (Table 3). Common side effects include fatigue, HFS, diarrhea, mucositis, hypothyroidism, yellow discoloration of skin, and cardiotoxicity[76].

Vandetanib is an antiangiogenic agent that inhibits VEGFR2 and VEGFR3 in addition to targeting EGFR, and several tyrosine and serine-threonine kinases[77]. Common side effects include diarrhea, rash, dermatitis, nausea/vomiting, hypertension, fatigue, abdominal pain, decreased appetite, and QT prolongation[78]. After early phase trials[79,80] demonstrated safety of vandetanib in combination with chemotherapy in advanced CRC patients, a phase II trial randomized patients to receive chemotherapy plus vandetanib vs chemotherapy plus placebo[81]. In this study the frequency of progression events - defined as objective or clinical progression or death from any cause- were noted to be higher in the vandetanib containing arm when compared to placebo (vandetanib 100 mg arm vs placebo: 72% vs 65%; HR, 1.21; 2-sided P = 0.53; vandetanib 300 mg arm vs placebo: 77% vs 65%; HR, 1.41; 2-sided P = 0.25)[81].

Vatalinib is an orally active antiangiogenic agent that blocks all VEGFR tyrosine kinase mediated signaling by competitively inhibiting the binding of ATP to the receptor kinase[82]. Adverse effects include lightheadedness, ataxia, nausea, vomiting, and hypertension[83]. Despite a tolerable toxicity profile in phase 1 studies[83], Vatalinib showed no survival advantage over placebo in two phase III randomized trials studies in mCRC patients (Table 3)[84,85].

ANTI-ANGIOGENIC THERAPY IN INITIALLY AND POTENTIALLY RESECTABLE mCRC

Carefully selected patients can be cured, if not at least provided with improved survival benefits, with resection of their metastases. Improved 5-year OS after liver resection was found in up to 46% of patients with up to 25% resected patients considered cured[86-90]. The 5-year survival rate of patients treated with pulmonary metastasectomies was found to be 55%-67%[91,92]. The median disease-free survival (DFS) and OS for those who had both hepatic and pulmonary resection has been shown to be 13-19.8 mo and up to 87 mo, respectively[93,94].

Bevacizumab is the only anti-angiogenic agent that has been extensively studied in the setting of resectable (or potentially resectable) mCRC. Small phase II studies have shown that when used either as preoperative therapy or as conversion treatment, bevacizumab in combination with Cape-OX or FOLFOX is associated with improved pathologic response, PFS, and OS in these patients[3,95,96]. Additionally, the combination of Cape-OX and bevacizumab rendered 40% of initially unresectable patients resectable in the BOXER (bevacizumab, oxaliplatin, capecitabine in unresectable liver metastases) study. This regimen provided objective responses in 78% of patients (95%CI: 63% to 89%) with 9% of patients (4 patients) achieving complete radiologic responses. These 4 patients remained in remission for 18-30 mo[97]. However, the sample sizes of these studies are too small to draw meaningful conclusions. Furthermore, the similar response rates (38% vs 38%; OR, 1.00; P = 0.99) between the bevacizumab and placebo arms when added to oxaliplatin-containing chemotherapy, in conjunction with a similar proportion of patients undergoing attempted curative intent metastasectomies (8.4% vs 6.1%) in a large phase III study by Saltz et al[29] argue against the use of bevacizumab in combination with oxaliplatin-based chemotherapy as conversion therapy. Several studies have demonstrated benefit with an irinotecan-containing regimen in combination with bevacizumab. In a retrospective study evaluating histopathologic features of resected liver tissue samples of 42 patients with mCRC who received FOLFOXIRI/Cape-irinotecan (Cape-IRI) with or without bevacizumab in the pre-operative setting, a significantly higher pathological response was noted in patients who received bevacizumab plus chemotherapy vs chemotherapy alone (63% vs 28%; P = 0.033)[98]. In a phase II study evaluating the efficacy and safety of preoperative bevacizumab plus FOLFIRI, patients with resectable liver metastases had a median PFS of 14 mo (95%CI: 11-24 mo), median OS of 38 mo (95%CI: 28-NA mo), an objective response rate of 66.7% (95%CI: 49.8% to 80.9%) and an R0 resection rate of 84.6%[99]. Masi et al[100] showed a conversion rate to R0 resection of 26% and up to 40% in those with liver-only metastatic disease after treatment with FOLFOXIRI and bevacizumab. Osterlund et al[101] also showed that bevacizumab plus cytotoxic chemotherapy was able to convert unresectable patients to resectable candidates in the 1^st- and 2^nd-line setting. Finally, Loupakis et al[34] found a response rate of 53.1% in the FOLFOX and bevacizumab arm compared to 65.1% in the FOLFOXIRI and bevacizumab arm with an odds ratio of 1.64 (95%CI: 1.15-2.35, P = 0.006) in the phase III TRIBE trial. However, there was no difference in the rate of R0 metastasectomy (12% vs 15%, respectively, P = 0.33)[34].

More recently, the OLIVIA trial provided further support for use of FOLFOX or FOLFOXIRI with bevacizumab. It also provided further evidence that while FOLFOXIRI with bevacizumab resulted in increased toxicities, it also offered improved resection rates and PFS compared to the FOLFOX and bevacizumab regimen. Thirty-nine patients with initially unresectable disease were assigned to the FOLFOX with bevacizumab arm and 41 patients received FOLFOXIRI with bevacizumab. The overall resection rate was 49% (95%CI: 32-65) and 61% (95%CI: 45-76), respectively. R0 resection was accomplished in 23% and 49% of patients, respectively. Median overall survival was 32.2 mo in the FOLFOX and bevacizumab group. It has not yet been reached in the FOLFOXIRI and bevacizumab group. Median PFS was 11.5 mo (95%CI: 9.6-13.6) in the FOLFOX and bevacizumab group compared to 18.6 mo (95%CI: 12.9-22.3) in the FOLFOXIRI and bevacizumab group. Most common grade 3-5 toxicities included diarrhea (14% with FOLFOX, 30% with FOLFOXIRI) and neutropenia (35% and 50%, respectively)[102].

Table 4 summarizes some of the currently available clinical data in this patient population.

Table 4 Bevacizumab plus chemotherapy as conversion therapy.

Ref.	Regimen	Rate ofconversion (%)	Overallresponse (%)	Median PFS (mo)	Median OS (mo)
Bertolini et al[95]; Phase II	FOLFOX6 + bevacizumab	61.9	57.1	12.9	22.5
Wong et al[97]; Phase II	CAPE-OX + bevacizumab	40	78 (95%CI: 63-89)	NA1	NA1
Nasti et al[99]; Phase II	FOLFIRI + bevacizumab	N/A	66.7 (95%CI: 49.8-80.9)	14 (95%CI: 11-24)	38 (95%CI: 28 to NA)
Klinger et al[3]; Meta-analysis/phase II	CAPE-OX/FOLFOX + bevacizumab	N/A	38 vs 10 (P < 0.001)	NA2	67 (95%CI: 8.4-125.6)2
Gruenberger et al[96]; Phase II	CAPE-OX + bevacizumab	N/A	73.2	NA	NA
Gruenberger et al[102]; Phase II	FOLFOX/FOLFOXIRI + bevacizumab	49% (FOLFOX), 61% (FOLFOXIRI)	62% (95%CI: 45-77) (FOLFOX), 81% (95%CI: 65-91) (FOLFOXIRI)	11.5 (95%CI: 9.6-13.6) (FOLFOX), 18.6 (95%CI: 12.9-22.3) (FOLFOXIRI)	32.2 (FOLFOX), not yet reached (FOLFOXIRI)
Masi et al[100]; Phase II	FOLFOXIRI + bevacizumab	26	NA	NA	NA
Loupakis et al[34]; Phase III	FOLFOX/FOLFOXIRI + bevacizumab	53.1 (FOLFOX), 65.1 (FOLFOXIRI)	12 (FOLFOX), 15 (FOLFOXIRI)	NA	NA
Saltz et al[29]; Phase III	FOLFOX/Cape-OX + bevacizumab vs FOLFOX/Cape-OX + placebo	8.4 vs 6.1	38 vs 38	9.4 vs 8	21.3 vs 19.9
		P = NA	P = 0.99	P = 0.0023	P = 0.077
Loupakis et al[98]; meta-analysis	FOLFOXIR/Cape-IRI ± bevacizumab	NA	63 vs 28	NA3	NA
			P = 0.033
Osterlund et al[101]; retrospective analysis	FOLFIRI + bevacizumab	9	42%	8.8	18.4

¹Though median PFS and OS were not specifically reported by Wong et al[97], the 12-mo PFS was 50% (95%CI: 34%-64%) and 12-mo OS was 86% (95%CI: 70%-94%);

²The OS in this study was not reported as a single parameter given its sample population. Instead, it was reported as a function of tumor regression grade, or TRG. The median OS of 67 mo cited in this table was found in those patients with lower TRGs (histologically with more fibrosis/necrosis than tumor, or major histological response). This OS decreases to 44 mo (95%CI: 14.1-73.8) in those with higher TRGs (histologically with more tumor than fibrosis/necrosis, or no histological response). Though the median PFS was not reported in this study, the 5-year PFS was 34% in lower TRGs and 9% in higher TRGs;

³Again, the PFS was reported in this study as a function of TRGs. There was a PFS benefit in those with lower TRGs compared to those with higher TRGs such that for every 10 units in the percentage of necrosis, there was a 0.83 HR reduction (95%CI: 0.7-0.99, P = 0.04). NA: Not available; N/A: Not applicable; PFS: Progression-free survival; OS: Overall survival.

Multiple studies have now established that the use of bevacizumab in combination of cytotoxic chemotherapy given preoperatively neither affects the recovery of liver function nor its regeneration. The anti-VEGF activity likely persists after preoperative cessation for at least 6 wk but does not seem to affect postoperative liver recovery. Furthermore, it was found not to increase the rate of complications if discontinued at least 5 wk prior to resection[96,103-111]. In fact, there is evidence that bevacizumab, when added to oxaliplatin-based chemotherapy, may protect against sinusoidal dilatation or sinusoidal obstruction syndrome[3,112,113].

Despite the efficacy of bevacizumab plus chemotherapy in the neoadjuvant setting, it was not found to provide either a PFS or OS benefit when used as adjuvant therapy after liver metastasectomy[114].

CONCLUSION

Anti-angiogenic therapy has assumed a vital role in the management of patients with mCRC. A total of three anti-angiogenic agents are currently approved in the treatment of these patients: bevacizumab, ziv-aflibercept, and regorafenib. The choice of agents differs based on tumor resectability and line of therapy. Patients with potentially resectable liver metastases have been shown to have an improved pathological response with the addition of bevacizumab to neoadjuvant chemotherapy. Studies have refuted concerns about hepatotoxicity and liver regeneration in patients treated with bevacizumab in the neoadjuvant setting.

Bevacizumab in combination with irinotecan-based chemotherapy has also been used as conversion therapy with a resection rates up to 61% in combination with FOLFOXIRI though at the expense of increased toxicities. In patients with treatment-naive unresectable mCRC, the addition of bevacizumab to cytotoxic chemotherapy achieves better and more durable responses, in addition to an advantage in PFS and OS when compared to chemotherapy alone. The beneficial role of bevacizumab in combination with a fluoropyrimidine in the maintenance setting, and the benefits of continuing bevacizumab beyond progression have been confirmed in multiple studies.

The use of bevacizumab in the first-line setting in patients with KRAS WT unresectable mCRC has been challenged by the FIRE-3 and CALGB 80405 studies, and cetuximab-based chemotherapy appears to be a viable option in these patients. More recently, two new anti-angiogenic agents were added to the armamentarium of targeted agents approved for use in mCRC. Ziv-aflibercept improved survival when used in the second-line setting in combination with an irinotecan-based chemotherapy in patients who have failed oxaliplatin-based therapy, and Regorafenib improved survival when compared to placebo in the treatment of patients with refractory mCRC. Another antiangiogenic agent, Ramucirumab has shown to improve survival in the second-line setting when used in combination with chemotherapy, and awaits FDA approval.

Despite these advances, mCRC remains an incurable disease with a median OS of approximately over 2 years in patients exposed to all available treatment regimens. Further insights into tumor biology and tumor microenvironment may help improve outcomes in these patients.