The role of tumor angiogenesis as a therapeutic target in colorectal cancer

Abstract

Introduction

Angiogenesis is a complex process regulated by several pro- and anti-angiogenic factors, thus the loss of its fine equilibrium plays a key role in colorectal cancer (CRC) development and progression. Therapeutic agents targeting VEGF/VEGFR signaling, the main regulator of this process, proved to be effective across different treatment lines in metastatic CRC (mCRC) and contributed greatly to improve patients’ survival in recent years.

Areas covered

This review aimed to summarize the actual body of knowledge available on the VEGF pathway in CRC, including currently available anti-angiogenic drugs and treatment challenges, mechanisms of resistance, promising predictive biomarkers and future perspectives.

Expert commentary

Angiogenesis inhibition in subsequent lines of treatment is a valid strategy in the continuum of care of mCRC patients. In this scenario, the availability of multiple agents warrants to tailor therapy to an individualized approach. However, the validation of predictive biomarkers to aid therapeutic decisions remains an issue. Intrinsic and adaptive resistance to anti-angiogenic agents comprises distinct and intertwined processes, eventually leading to treatment failure and disease progression. The expanding knowledge on the mechanisms underlying the angiogenesis pathway, different potential treatment targets and mechanisms of tumor resistance, may lead to promising new perspectives in this field.

1. Introduction

The advent of anti-angiogenic therapy has represented a major breakthrough in the treatment of metastatic colorectal cancer (mCRC). In the past decade, several new targeted agents have been added to the therapeutic armamentarium for the treatment of this malignancy, the majority of which are anti-angiogenic drugs. Since bevacizumab, an anti-VEGF-A monoclonal antibody (mAb), was first approved in 2004, other anti-VEGF signaling inhibitors (i.e. regorafenib, aflibercept, and ramucirumab) became part of clinical practice. Together with the development of anti-EGFR mAbs (panitumumab and cetuximab) and the improvement in cytotoxic chemotherapy regimens as well as the integration of multimodality treatment approaches, this has led to an increase of the overall survival (OS) for mCRC patients, that have now reached over 30 months.

Innate and acquired resistance to anti-angiogenic drugs limits the efficacy of these compounds and the majority of patients will eventually experience a disease progression during treatment. Therefore, many efforts have been made to identify predictive biomarkers to better select those patients who will more likely respond to selected agents, in order to guide therapeutic choices and optimize treatment sequences and patients’ benefit.

In this review, we aimed to summarize the actual body of knowledge on VEGF signaling pathway in CRC, including currently available anti-angiogenic drugs and treatment challenges, mechanisms of resistance, and the development of predictive biomarkers.

2. Preclinical data

2.1. Role of VEGF in colorectal cancer

Angiogenesis is a complex process regulated by various pro- and anti-angiogenic factors, which is crucial for tissue growth and development. The loss of its fine equilibrium is one of the hallmarks of cancer [1]. Vascular endothelial growth factor (VEGF) is a key regulator of angiogenesis both in physiological processes and pathological events, such as cancer development. The VEGF family comprises five members: VEGF-A, -B, -C, -D, and placental growth factor (PlGF) [2]. VEGF-A is the most widely recognized and major player in tumor angiogenesis. It undergoes alternative splicing, which leads to several isoforms that differ in terms of distribution and function [3]. Although many elements may influence the VEGF pathway, hypoxia remains the main factor that regulates angiogenesis. Hypoxia induces VEGF expression through transcription factor hypoxia inducible factor-1 (HIF-1) and HIF-2. Moreover, low oxygen tension promotes VEGF upregulation by an increase in the stability of its mRNA51 [4]. Three different receptor tyrosine kinases (RTKs) are involved in VEGF signaling: VEGFR-1, −2, and −3. VEGF-A interacts mainly with VEGFR-2, which is normally expressed by endothelial cells, but it can also bind to VEGFR-1. VEGF-B and PlGF bind to VEGFR-1, while VEGF-C and -D bind to VEGF-2 and VEGFR-3, although with a higher affinity for the latter [5].

VEGFR-2 signaling is implicated in the normal and pathological regulation of endothelial cells survival, proliferation and migration. VEGFR-1 promotes monocyte and macrophage migration to the tumor and it plays a critical role in tumor progression and metastasization in CRC [6]. VEGF-C and D, through VEGFR-3, are crucial for lymphangiogenesis and have been related to the development of metastases in many cancers, including CRC [7]. Tumor vessels, driven by VEGF-A overexpression, show various abnormalities, such as tortuosity, fragility, and lack of pericytes. These dysfunctional and leaky vessels contribute to interstitial hypertension, which eventually limits the delivery of antitumor agents to their target [8].

Even though VEGF is the main pathway which regulates angiogenesis, other factors are involved. Recently, Ang-1 and Ang-2 have been shown to be fundamental angiogenesis regulator. They are ligands for the receptor tyrosine kinase (RTK) Tie-2 with opposite function: Ang-1 activates Tie-2 stabilizing the mature vessels, whereas Ang-2 acts as its inhibitor, disrupting the normal vasculature. Higher expression of Ang-2 has been associated with metastatic growth and poorer survival outcomes in CRC [9]. Low pre-therapeutic serum Ang-2 levels have been associated with a significantly better response rate (RR), a prolonged median progression-free survival (PFS) and OS in mCRC patients treated with bevacizumab-based therapy [10,11]. Moreover, the VEGF pathway showed a tight crosstalk with different crucial pathways promoting angiogenesis, such as fibroblast growth factor (FGF), platelet-derived growth factor (PDGF), and epidermal growth factor (EGF) pathways.

Therapeutic agents targeting VEGF/VEGFR signaling, the main regulator of this process, proved to be effective across different treatment lines in patients with mCRC. However, preexisting or rapidly developing resistance mechanisms to anti-angiogenic drugs can limit their activity and eventually promote disease progression.

2.2. Targeting VEGF signaling and mechanism of resistance

Different approaches have been used to disrupt VEGF signaling, such as mAbs against the ligand or receptor, and tyrosine kinase inhibitors which are able to block VEGFR activation and signaling. To date, several VEGF pathway targeting agents are available for mCRC, as registered drugs include small molecule tyrosine kinase inhibitors (TKI) targeting VEGFR and antibodies targeting VEGF and VEGFR-2 (Figure 1 and Table 1) [12].

Figure 1.

Diagram of angiogenesis pathways and the role of anti-angiogenetic drugs available for metastatic colorectal cancer.

Abbreviations: VEGF: Vascular endothelial growth factor; VEGFR: Vascular endothelial growth factor receptor; PlGF: Placental growth factor; Ang: angiopoietin; PDGF: Platelet-derived growth factor; PDGFR: Platelet-derived growth factor receptor; FGF: Fibroblast growth factor; FGFR: Fibroblast growth factor receptor; Dll4: Delta Like Canonical Notch Ligand 4; EGF: Epidermal growth factor; EGFR: Epidermal growth factor receptor.

Table 1.

Approved anti-angiogenic agents for the treatment of CRC.

Agent	Type of molecule	Target	Setting of approval in CRC	Dose
Bevacizumab	IgG1 monoclonal antibody	VEGF-A	First- and second-line mCRC in combination with 5-FU-based chemotherapy; second-line in combination with fluoropyrimidine-irinotecan- or fluoropyrimidine-oxaliplatin-based regimen, after progression to a first-line bevacizumab-based regimen	5 mg/kg i.v. every 2 weeks 7.5 mg/kg i.v. every 3 weeks
Aflibercept	Fusion protein of the extracellular domains of human VEGFR-1 and -2 with the Fc portion of a human IgG1	VEGF-A, VEGF-B, and PlGF	Second-line mCRC in combination with FOLFIRI after an oxaliplatin-based regimen	45 mg/kg i.v. every 2 weeks
Ramucirumab	Fluman IgG1 monoclonal antibody	VEGFR-2	Second-line mCRC in combination with FOLFIRI after a first-line bevacizumab-, oxaliplatin- and fluoropyrimidine-based regimen	8 mg/kg i.v. every 2 weeks
Regorafenib	Multitarget kinase inhibitor (small molecule)	VEGFR1–3, Angiopoietin-1 receptor, PDGFR, FGFR cKIT, RET, BRAF	mCRC previously treated with fluoropyrimidine-, oxaliplatin-, and irinotecan-based regimens, an anti-VEGF therapy, and an anti-EGFR therapy, if KRAS wild-type	160 mg/day p.o.

mCRC: metastatic colorectal cancer; IgG1: immunoglobuline 1; VEGFR: vascular endothelial growth factor receptor; EGFR: epidermal growth factor receptor; PDGFR: platelet-derived growth factor receptor; FGFR: fibroblast growth factor receptor; PlGF: placental growth factor; KRAS: Kirsten rat sarcoma viral oncogene homolog; 5-FU: 5-fluorouracil; FOLFOX: fluorouracil, folinic acid, oxaliplatin; FOLFIRI: fluorouracil, leucovorin, irinotecan; i.v.: intravenous; p.o.: per os.

Bevacizumab (Avastin^®, Genentech) was the first anti-angiogenic drug to be approved by the US FDA for the treatment of mCRC in 2004, changing the landscape of mCRC therapeutic strategy. It is a humanized mAb that binds to and neutralizes all human VEGF-A [13] and is approved in combination with chemotherapy both in the first- and second-line setting.

Aflibercept (Zaltrap^®, Sanofi and Regeneron Pharmaceuticals, Inc.) is a recombinant fusion protein combining the second Ig-like domain of VEGFR-1 and the third Ig-like domain of VEGFR-2 with a human IgG1 Fc fragment, which acts as a soluble decoy receptor that binds with high-affinity VEGF-A, VEGF-B, and PIGF [14], thus inhibiting downstream signaling. Of note, this broader coverage of the VEGF pathway has been suggested as a possible mechanism, which may play a role in overcoming tumor resistance to bevacizumab therapy [15]. Aflibercept is currently registered as second-line treatment in combination with FOLFIRI (irinotecan, fluorouracil, and leucovorin) [16].

Ramucirumab (Cyramza^®, Eli Lilly and Company) is a fully humanized mAb that selectively binds to the extracellular domain of VEGFR-2 [17]. VEGFR-2 is largely considered the primary VEGF family receptor driving angiogenesis. Following binding of VEGF-A, VEGFR-2 can form either a homodimer or heterodimer complex with VEGFR-1, resulting in intracellular tyrosine phosphorylation, which leads to downstream effects culminating in angiogenesis, increase in vascular permeability and promotion of endothelial cell migration, proliferation, and survival [18]. Ramucirumab received FDA approval in 2015 for the second-line treatment of mCRC [19].

Regorafenib (Stivarga^®, Bayer) is a novel oral multikinase inhibitor that blocks the activity of several protein kinases, including VEGFR-1, VEGFR-2, VEGFR-3, Tie-2, cKIT, RET, BRAF PDGFR, and FGFR [20]. It has been approved by FDA in 2012 for the treatment of patients with mCRC who have progressed after all other available standard treatments [21].

Despite clear evidence of activity across different treatment lines, the mechanism of action of VEGF-A and VEGFR inhibitors has not been fully elucidated. In fact, although angiogenesis is fundamental for cancer progression, it has proven to crosstalk with numerous other pathways and to be highly adaptable. For this reason, many patients do not benefit from anti-angiogenic therapies or develop resistance during treatment. Evasive resistance to angiogenesis inhibitors comprises distinct and intertwined mechanisms: activation and/or upregulation of different pro-angiogenic signals (such as FGF, PDGF, and Ang-1); recruitment of various bone marrow-derived cells (BMDCs), which differentiate into endothelial cells, pericytes and pro-angiogenic monocytes, such as tumor-associated macrophages; enhancing and increasing pericyte coverage which protects tumor blood vessels; increased invasiveness of tumor cells, leading to vessels co-option [22–24]. Predictive biomarkers are needed to identify the emergence of resistance mechanisms during treatment, potentially allowing a dynamic modulation and sequencing of available therapies in order to overcome resistance and optimize patients’ outcomes.

3. Therapeutic agents

To date, four anti-angiogenic agents have been approved by the FDA and the European Medicines Agency (EMA) for the treatment of mCRC in different lines: bevacizumab, aflibercept, ramucirumab, and regorafenib (Table 1). Main clinical trials results for each drug in different settings will be reviewed in the following sections (Tables 2 and 3).

Table 2.

Main first-line bevacizumab trials results.

Trial (phase)	Line	Treatment arms (n.)	Primary endpoint	PFS	OS	ORR (%)
AVF2107g (III) [26]	1st	ILF + bevacizumab (n.402) ILF + placebo (n.411)	OS	10.6 m 6.2 m [HR 0.54; p < 0.001]*	20.3 m 15.6 m [HR 0.66; p < 0.001]*	44.8 34.8 [p = 0.004]
BICC-C (III)	1st	FOLFIRI + bevacizumab (n.57) FOLFIRI (n.60)	PFS	11.2 m 8.3 m [p = 0.28]**	28.0 m 19.2 m [p = 0.037]**	57.9 53.3 ***
NO16966 (III) [27]	1st	XELOX + bevacizumab (n.350) FOLFOX4 + bevacizumab (n.349) XELOX + placebo (n.350) FOLFOX4 + placebo (n.351)	PFS	9.4 m 8.0 m [HR 0.83; 95% CI, 0.72–0.95; p = 0.0023]	21.3 m 19.9 m [HR 0.89; 95% CI, 0.76–1.03; p = 0.0769]	47 49 [p = 0.31]
TRIBE (III) [29,30]	1st	FOLFOXIRI + bevacizumab (n.57) FOLFIRI + bevacizumab (n.57)	PFS	12.1 m 9.7 m [HR 0.75; 95% CI, 0.62–0.90; p = 0.003]	31.0 m 21.8 m [HR 0.79; 95% CI, 0.63–1.00; p = 0.054]	65 53 [p = 0.006]
AVEX (III) [35]	1st	Capecitabine + bevacizumab (n.140) capecitabine (n.140)	PFS	9.1 m 5.1 m [HR 0.53; 95% CI, 0.41–0.69; p < 0.0001]	20.7 m 16.8 m [HR 0.79; 95% CI, 0.57–1.09; p = 0.18]	19 10 [p = 0.04]
CAIRO-3 (III) NCT00442637 [36]	1st	Capecitabine + bevacizumab manteinance (n.279) observation (n.278)	PFS2	11.7 m 8.5 m [HR 0.63; 95% CI, 0.53–0.77; p < 0.0001]	21.6 m 18.1 m [HR 0.83; 95% CI, 0.68–1.01; p = 0.06]	Not evaluated
FIRE-3 (III) NCT00433927 [38]	1st	FOLFIRI + bevacizumab (n.295) FOLFIRI + cetuximab (n.297)	ORR	10.3 m 10.0 m [HR 1.06; 95% CI, 0.88–1.26; p = 0.55]	25.0 m 28.7 m [HR 0.77; 95% CI, 0.62–0.96; p = 0.017]	58 62 [p < 0.18]
PEAK (III) NCT00819780 [40]	1st	FOLFOX + Bevacizumab (n.143) FOLFOX + Panitumumab (n.142)	PFS	10.1 m 10.9 m [HR 0.87; 95% CI, 0.65–1.17; p = 0.35]	24.3 m 34.2 m [HR 0.62; 95% CI, 0.44–0.89; p = 0.009]	53.5 57.8 ***
CALGB/SWOG 80,405 (III) NCT00265850 [41,42]	1st	FOLFIRI/FOLFOX6 + bevacizumab (n.559) FOLFIRI/FOLFOX6 + cetuximumab (n.578)	OS	10.6 m 10.5 m [HR 0.95; 95% CI, 0.84 to 1.08; p = 0.45]	29.0 m 30.0 m [HR 0.88; 95% CI, 0.77 to 1.01; p = 0.08]	55.2 59.6 [p = 0.08]

CI: confidence interval; HR: hazard ratio; m: months; n: number of patients; PFS: progression free survival; ORR: objective response rate; OS: overall survival.

^*95% CI = not reported.

^**HR = not reported.

^***p = not reported.

Table 3.

Approved anti-angiogenic agents: main trials results after the first-line.

Trial (phase)	Line	Treatment arms (n.)	Primary endpoint	PFS	OS	ORR (%)
E3200 (III) [50]	2nd	FOLFOX4 + bevacizumab (n.286) FOLFOX4 + placebo (n.291) Bevacizumab (n.243) prematurely closed due to futility	OS	7.3 m 4.7 m [HR 0.61; 95% CI, 0.51–0.73; p < 0.0001]	12.9 m 10.8 m [HR 0.75; 95% CI, 0.63–0.89; p = 0.0011]	22.7 8.6 [p < 0.0001]
TML (ML1847) (III) [54]	2nd	FOLFOX/FOLFIRI + bevacizumab (n.409) FOLFOX/FOLFIRI + placebo (n.411)	OS	5.7 m 4.1 m [HR 0.68; 95% CI, 0.59–0.68; p < 0.0001]	11.2 m 9.8 m [HR 0.81; 95% CI, 0.69–0.94; p = 0.0006]	5.4 3.9 *
VELOUR (III) [16]	2nd	FOLFIRI + aflibercept (n.612) FOLFIRI + placebo (n.614)	OS	6.9 m 4.7 m [HR 0.76; 95% CI, 0.66–0.87; p < 0.0001]	13.5 m 12.0 m [HR 0.82; 95% CI, 0.71–0.94; p = 0.032]	19.8 11.1 [p < 0.0001]
RAISE (III) [19]	2nd	FOLFIRI + ramucirumab (n.536) FOLFIRI + placebo (n.536)	OS	5.7 m 4.5 m [HR 0.79; 95% CI, 0.70–0.90; p = 0.0005]	13.5 m 12.0 m [HR 0.82; 95% CI, 0.71–0.94; p = 0.0219]	13.4 12.5 [p = 0.6336]
CORRECT (III) [21]	After 2nd	regorafenib + BSC (n.505) placebo + BSC (n.255)	OS	1.9 m 1.7 m [HR 0.49; 95% CI, 0.42–0.58; p < 0.0001]	6.4 m 5.0 m [HR 0.77; 95% CI, 0.64–0.94; one-sided p = 0.0052]	1 0.4 [p =0.19]

C: confidence interval; HR: hazard ratio; m: months; n: number of patients; PFS: progression free survival; ORR: objective response rate; OS: overall survival.

^*p = not reported.

3.1. First-line treatment

The 1998 phase II AVF2107 trial was the first to demonstrate an improvement in RR, median PFS (mPFS), and OS (mOS) from the combination of bevacizumab with fluorouracil (FU) and leucovorin (LV) versus FU/LV alone in a small cohort of 104 previously untreated mCRC patients [25]. In the same year, a phase III trial (AVF2107g) randomized a total of 813 patients and confirmed the efficacy of bevacizumab in association with the standard chemotherapy doublet at that time, FU/LV/irinotecan (ILF) with a significant improvement in OS (20.3 vs. 15.6 months, hazard ratio (HR) 0.66; p < 0.001), mPFS (10.6 vs. 6.2 months, HR 0.54) and RR (44.8% vs. 34.8%) compared to chemotherapy alone. This clinical benefit was accompanied by a relatively modest increase in side effects with an absolute increase of 10% in the overall incidence of grade 3 and 4 adverse events, involving hypertension, diarrhea, and leukopenia [26]. Based on these results, in March 2004, the FDA approved bevacizumab in combination with chemotherapy in the first-line treatment of mCRC.

Moving from these data, the randomized NO16966 phase III trial, initially designed to compare the standard FOLFOX-4 (oxaliplatin, fluorouracil, and leucovorin) regimen to XELOX (oxaliplatin and capecitabine), was amended to a 2 × 2 factorial design to incorporate bevacizumab. The study met its two coprimary endpoints showing the non-inferiority of XELOX compared to FOLFOX-4 in the first-line setting and a significant improvement in mPFS with bevacizumab compared to placebo when combined with an oxaliplatin-based chemotherapy (9.4 vs. 8.0 months, HR 0.83; 95% CI, 0.72–0.95; p = 0.0023) [27].

More recently, a phase III randomized trial by the GONO (Gruppo Oncologico del Nord Ovest) group demonstrated the superiority of the FOLFOXIRI regimen (FOLFOX plus irinotecan) over an irinotecan-based doublet as first-line treatment of clinically selected mCRC patients [28]. Thereafter, the phase III randomized TRIBE study showed a significant advantage for patients treated with FOLFOXIRI plus bevacizumab when compared with FOLFIRI plus bevacizumab, with a still manageable toxicity profile [29,30]. Although no randomized comparison of FOLFOXIRI with or without bevacizumab is available, an estimate of the impact of adding bevacizumab to the triple chemotherapy backbone can be evaluated by comparing outcomes of patients treated in the aforementioned trials: significantly longer mPFS and OS were observed in the FOLFOXIRI plus bevacizumab group compared with the FOLFOXIRI group [31]. Of note, in pre-planned subgroup analyses based on RAS and BRAF status, treatment with FOLFOXIRI plus bevacizumab in the TRIBE study showed an unprecedented mOS of 19.0 months (HR 0.54; 95% CI, 0.24–1.20) in the BRAF-mutated population, with a treatment effect not significantly different across molecular subgroups (p interaction = 0.52) [30].

The topic of the best chemotherapy backbone to be associated with bevacizumab has been addressed by the global randomized phase II MAVERICC trial, comparing first-line treatment with either modified FOLFOX (mFOLFOX-6) or FOLFIRI plus bevacizumab. Consistently with previous findings, PFS and OS were comparable in both treatment arms [32]. Similar results were observed in a randomized phase III Japanese trial, demonstrating the non-inferiority of FOLFIRI plus bevacizumab versus mFOLFOX-6 plus bevacizumab as first-line treatment in mCRC [33]. Subgroup analyses of randomized trials, however, suggest an advantage from an irinotecan-based first-line therapy in patients progressed after an oxaliplatin-based adjuvant treatment [34].

The phase III randomized AVEX trial was the first to specifically demonstrate the efficacy and safety of the addition of bevacizumab to capecitabine in an elderly population of patients (≥70 years) with mCRC, considered not optimal candidates for first-line combined chemotherapy. The addition of bevacizumab improved PFS, primary endpoint (9.1 vs. 5.1 months, HR 0.53; 95% CI, 0.41–0.69; p < 0.0001), as well as ORR (19% vs. 10%; p = 0.04), with no significant difference in OS. No new safety signals were identified in elderly patient populations, and the safety profile was consistent with that reported by other studies [35].

Evidence of inferior survival outcomes for patients discontinuing bevacizumab early during treatment in the NO16966 trial suggested the possible value of a continuous angiogenesis inhibition during mCRC treatment [27]. The CAIRO-3 trial was the first prospective study to investigate this topic. The study randomized patients with stable disease, partial response or complete response after six cycles of induction chemotherapy with capecitabine, oxaliplatin, and bevacizumab (CAPOX-B), to receive a maintenance treatment with bevacizumab and capecitabine versus observation. Upon first disease progression, patients in both arms were to receive reinduction with CAPOX-B until second progression (PFS2) which was the primary endpoint. The study showed that maintenance treatment significantly prolonged mPFS2 compared to observation (11.7 vs. 8.5 months; HR 0.67; 95% CI, 0.56–0.81; p < 0.0001) without compromising patients’ quality of life [36]. In a recently presented post-hoc analysis with updated follow-up, the benefit of maintenance treatment was observed regardless of RAS/BRAF mutational status, mismatch repair (MMR) status and primary tumor sidedness [37].

With the introduction of monoclonal antibodies targeting the epidermal growth factor receptor (anti-EGFR) in the treatment of RAS wild-type mCRC, the issue of the optimal sequencing between anti-VEGFs and anti-EGFRs in this population of patients has been lately a matter of intense debate.

The FIRE-3 trial was the first head-to-head comparison of the anti-EGFR agent cetuximab versus bevacizumab in combination with FOLFIRI chemotherapy in the first-line treatment of patients with KRAS exon 2 wild-type mCRC. In the intention-to-treat analysis, no difference was detected in terms of investigators-assessed ORR, nor mPFS between the two arms, while mOS was significantly longer in the FOLFIRI plus cetuximab than in FOLFIRI plus bevacizumab group (28.7 vs. 25.0 months, HR 0.77; 95% CI, 0.62–0.96; p = 0.017) [38]. In a recently presented post-hoc analysis of the extended KRAS/NRAS exons 2–4 wild-type subgroup the OS benefit was most pronounced in the cetuximab arm (33.1 vs. 25.0 months, HR 0.70; 95% CI, 0.54–0.90; p = 0.0059). After centralized radiological review, FOLFIRI plus cetuximab was shown to induce a superior ORR, a higher frequency of early tumor shrinkage and a greater depth of response, which correlated with the observed OS benefit in the cetuximab arm [39]. These findings are in line with the results of the PEAK study, a randomized phase II trial comparing panitumumab versus bevacizumab in combination with FOLFOX in the first-line setting. This study, although not designed to demonstrate the superiority of one treatment over the other, showed a survival benefit with the use of panitumumab both in terms of PFS and OS in the extended RAS WT population [40]. Regarding the same topic, however, results from another randomized head-to-head phase III study, the CALGB 80,405 [41,42], as well as a meta-analysis including results from these three trials (FIRE-3, PEAK, and CALGB 80,405) [43], were inconsistent in supporting the superiority of the use of one targeted agent over the other.

During the last year, primary tumor sidedness has emerged as a novel prognostic and predictive biomarker in mCRC. Right- (RC) and left-sided CRCs (LC) are characterized by distinct clinical and pathological features with RC patients more likely to be women, older and have mucinous, signet ring cells or undifferentiated histology together with higher TNM stage at diagnosis [44]. From a molecular point of view, RC is characterized by high prevalence of BRAF mutation, CpG island methylation (CIMP-High) and microsatellite instability (MSI), while LC by chromosomal instability and marked aneuploidy [45]. These different molecular characteristics translate into differential clinical outcomes with RC displaying a markedly worse prognosis [46]. Data from the CALGB/SWOG 80,405 trial confirm the prognostic role of primary tumor location with a mOS of 33.3 months for LC compared with 19.4 months for RC (HR 1.55; 95% CI, 1.32–1.82; p < 0.001). When looking at treatment arms, however, no significant OS difference was observed in patients receiving bevacizumab (32.6 vs. 29.2 months in LC vs. RC, respectively; HR 0.88; 95% CI, 0.62–1.25; p = 0.5) [47]. A meta-analysis of three prospective trials FIRE-3, CALGB/SWOG 80,405 and PEAK confirmed a significant benefit from first-line anti-EGFR treatment in RAS wild-type LC. By contrast, in RAS wild-type RC the bevacizumab-based treatment was numerically associated with longer PFS and OS (HR for survival 1.3; 95% CI, 0.97–1.74; p = 0.081) [48]. Similar results were obtained from a pooled analysis of data from randomized CRYSTAL, PRIME, PEAK, FIRE-3, CALGB 80,405, and 20,050,181 trials [49]. Despite the lack of a prospective validation, and the limitations related to the nature of unplanned retrospective subgroup analyses, these data are consistent across different randomized trials, thus supporting the rationale to avoid the use of anti-EGFRs in the first-line setting for RC, when other therapeutic alternatives are available.

Main first-line bevacizumab trials are reported in Table 2.

3.2. Second-line treatment

3.2.1. Bevacizumab

The first evidence of benefit from second-line treatment with an anti-angiogenic agent in mCRC derived from the phase III E3200 trial by Giantonio et al. [50]. This study enrolled 829 anti-angiogenesis naïve patients previously treated with fluoropyrimidine and irinotecan, who were randomized to receive treatment either with FOLFOX-4 associated with or without bevacizumab, or bevacizumab alone. Of note, accrual to bevacizumab monotherapy arm was prematurely closed following an interim analysis showing inferior survival compared with the other arms. The study met its primary endpoint with a significant improvement in OS from chemotherapy plus bevacizumab compared to chemotherapy alone (mOS 12.9 vs. 10.8 months, respectively, HR 0.75; p = 0.0011). Secondary endpoints PFS and ORR were also significant favoring the FOLFOX-4 plus bevacizumab arm (mPFS 7.3 vs. 4.7 months, HR 0.61; p < 0.0001; ORR 22.7% vs. 8.6%, p < 0.0001).

The role of continuing bevacizumab treatment beyond the first-line progression was then explored in two retrospective observational cohort studies: the BRiTE and the ARIES studies [51–53]. Both reported a significant improvement in post-progression OS and PFS from treatment with bevacizumab beyond progression switching to standard second-line chemotherapy versus no bevacizumab beyond progression. Based on these data, the phase III randomized TML (ML18147) study was conducted to explore the efficacy of bevacizumab combined with standard second-line chemotherapy after progression to a first-line bevacizumab-containing regimen in mCRC patients. The trial met its primary endpoint, with a mOS of 11.2 months in the bevacizumab beyond progression arm versus 9.8 months in the chemotherapy alone arm (HR 0.81; 95% CI, 0.69–0.94; p < 0.0062). mPFS was 5.7 versus 4.1 months (HR 0.68; 95% CI, 0.59–0.78; p < 0.0001) in the bevacizumab arm and in the chemotherapy alone arm, respectively. No significant differences in ORR were observed. The benefit from treatment with bevacizumab beyond progression was independent of first-line PFS (≤ or > 9 mo) and tumor KRAS mutational status [54]. Similar results later came from the Italian phase III BEBYP trial [55].

Interestingly, in RAS wild-type patients, two phase II head-to-head trials investigating the efficacy of bevacizumab beyond progression versus cetuximab, the PRODIGE 18 trial [56], and panitumumab, the SPIRITT trial [57], as second-line treatment, did not show any significant difference between the continuation of bevacizumab beyond progression and the switch to an anti-EGFR agent in this setting.

Taken together these data demonstrated a benefit from continuing bevacizumab beyond first progression, suggesting that resistance mechanisms to anti-angiogenic treatment are not always associated with resistance to chemotherapy and that continuing anti-angiogenic inhibition in subsequent lines of treatment is a valid strategy in mCRC.

3.2.2. Aflibercept

The international, prospective, randomized, double-blind, phase III VELOUR trial evaluated the activity and safety of aflibercept in combination with FOLFIRI in the second-line treatment of mCRC patients who had progressed after a previous oxaliplatin-containing regimen [16]. A total of 1226 were randomized to receive treatment with either aflibercept or placebo in association with FOLFIRI. Almost 30% of the patients enrolled in the study had received prior bevacizumab in the first-line setting. The study met its endpoints as the addition of aflibercept to FOLFIRI compared to the placebo arm significantly improved OS (primary endpoint: 13.5 vs. 12.06 months, HR 0.817; 95% CI 0.713–0.937; p = 0.0032), mPFS (secondary endpoint: 6.90 vs. 4.67 months, HR 0.758; 95% CI 0.661–0.869; p < 0.0001) and RR (19.8% vs. 11.1%, p < 0.0001). Pre-planned subgroup analysis showed that the survival benefit was consistent in all patients subgroups and independent of bevacizumab pretreatment [58] or timing of first-line disease progression [59]. However, poorer outcomes were observed in the subgroup of patients who relapsed during or within six months from completion of adjuvant treatment [60]. Interestingly, looking at the Kaplan–Meier survival curves of the trial, a progressive increase in the separation of the curves can be observed past the median time point, although less evident in the subgroup of patients previously treated with bevacizumab, suggesting a sustained survival benefit over time from the use of aflibercept [58]. The evaluation of survival data over the time course of the study in the ITT population showed a progressive increase of median difference of OS and estimated probabilities of survival between the two treatment arms favoring FOLFIRI plus aflibercept (28.0% vs. 18.7% of patients alive at 24 months in the aflibercept and placebo arm, respectively) [61]. As expected a higher rate of grade 3–4 adverse events was observed in the aflibercept arm (83.5%) compared to the placebo arm (62.5%). In addition to main anti-angiogenic related side effects (i.e. hypertension, bleeding, thromboembolic events), an increase in chemotherapy-related adverse events was also observed in the aflibercept arm, with particular reference to neutropenia and grade 3/4 diarrhea, leading to a higher treatment discontinuation rate in the FOLFIRI plus aflibercept group (26.8% vs. 12.1% in the placebo group). These safety data raised concerns on the use of FOLFIRI-aflibercept combination in elderly patients, which were addressed by a recently published age-based post-hoc analysis of the trial. Efficacy and safety of treatment were evaluated by study arm and age (≥ or < 65 years, n = 443 and 783, respectively) confirming a significant benefit both in OS and PFS from the treatment with FOLFIRI plus aflibercept regardless of patients’ age with comparable toxicity profiles in both age groups [62]. In order to avoid serious complications and treatment discontinuation, a careful monitoring and early management of adverse events during treatment with aflibercept are recommended.

Notably, in contrast with bevacizumab, the combination of aflibercept with mFOLFOX-6 in previously untreated mCRC failed to improve the efficacy of treatment and was associated with higher toxicity in the first-line phase II AFFIRM trial [63]. Thus, to date, the combination of aflibercept with oxaliplatin-based regimens is not supported.

3.2.3. Ramucirumab

The randomized, double-blind, multicenter, phase III RAISE study evaluated the combination of ramucirumab plus FOLFIRI in mCRC patients who had progressed to a first-line treatment with bevacizumab, oxaliplatin and a fluoropyrimidine. A total number of 1072 patients were enrolled in the trial and randomized to receive either ramucirumab or placebo in combination with FOLFIRI as second-line treatment. A significant improvement in mOS (13.3 vs. 11.7 months in the placebo arm, HR 0.84; 95% CI, 0.73–0.97; p = 0.0219) and mPFS (5.7 vs. 4.5 months, HR 0.79; 95% CI 0.70–0.90; p = 0.0005) was reported in the ramucirumab arm. The toxicity profile was consistent with expected adverse events and characterized by an increase of grade 3 or worse neutropenia (38% of patients in the ramucirumab group vs. 23% in the placebo group), hypertension (11% vs. 3%), and fatigue (12% vs. 8%) [19]. Pre-planned subgroup analyses showed that the observed survival benefit was consistent across different subgroups of patients according to age (≥65 vs. <65 years), tumor KRAS mutational status and time to first disease progression (TTP ≥6 or <6 months) [64]. As previously reported in gastric cancer, recently published exposure-response analyses suggest that a higher ramucirumab exposure is significantly associated with treatment efficacy both in terms of PFS and OS [65].

These results, alongside data from the VELOUR trial, support the effectiveness of maintaining a prolonged inhibition of the VEGF pathway across treatment lines in mCRC.

Of note, similarly to aflibercept, the combination of ramucirumab with mFOLFOX-6, though evaluated in this case as a second-line treatment after progression to a first-line irinotecan-containing regimen, failed to achieve significant improvement in PFS and OS in mCRC patients in a recent phase II randomized trial [66].

3.3. Anti-angiogenic agents beyond the Second-line

3.3.1. Regorafenib

Regorafenib is the first multikinase inhibitor which demonstrated activity in mCRC and is, to date, the only approved by regulatory authorities for the treatment of patients with advanced CRC after failure of all standard systemic anti-cancer therapies including an anti-VEGF and anti-EGFR agent in eligible patients [67].

After initial promising evidence of activity in phase I studies, two large randomized phase III, placebo-controlled trials, the CORRECT [21] and CONCUR [68] studies, investigated the efficacy of regorafenib in heavily pretreated mCRC patients progressed after all available standard treatments.

The CORRECT trial randomized 760 patients to receive best supportive care plus either regorafenib (n = 505) or placebo (n = 255). The study met its primary endpoint, with a mOS of 6.4 vs. 5.0 months in the regorafenib and placebo arms, respectively (HR 0.77; 95% CI, 0.64–0.94; one-sided p = 0.0052). A statistically significant improvement in mPFS was also observed favoring the regorafenib treatment arm (1.9 vs. 1.7 months, HR 0.49; 95% CI, 0.42–0.58; p > 0.0001). The survival benefit was consistent in all patients subgroups. Treatment with regorafenib resulted in a disease control rate of 41% but radiological response according to RECIST criteria was observed only in 5 (1%) of patients, suggesting that the effect of regorafenib contributes to disease stabilization rather than inducing tumor shrinkage. Ninety three percent of patients in the experimental arm experienced adverse events, 54% grade 3/4. Most frequent toxicities among grade 3 or worse adverse events were hand-food syndrome (17%), fatigue (10%), diarrhea (7%), hypertension (7%), and rash or desquamation (6%); other relevant adverse events were oral mucositis, anorexia, voice changes, and liver function abnormalities [21]. The incidence of adverse events peaked during the first cycle of treatment and toxicities were manageable with early intervention, dose reduction or interruption [69].

The CONCUR trial was the second to demonstrate a survival benefit and an improvement in disease control from regorafenib treatment in advanced mCRC. This study enrolled 204 Asian patients who progressed after at least two previous lines of therapy. Trial design and baseline patients characteristics were similar to those of the CORRECT study. The safety profile of regorafenib in the Asian population was also consistent with previously reported results [68].

Efficacy and safety results of regorafenib as single agent in advanced mCRC were subsequently confirmed in two large open-label, single arm, real-world population studies within compassionate or expanded access programs, the CONSIGN and the REBACCA studies [70,71].

Although statistically significant, the absolute incremental benefit from regorafenib treatment is rather modest (0.2 months PFS and 1.4 months in terms of OS in the CORRECT study) and the majority of patients did not derive any advantage from treatment despite being exposed to potential toxicities. The current effort in finding potential predictive biomarkers in order to refine patients selection and optimize the use of regorafenib is thus critical at this point.

Interestingly, moving from the observation that conventional radiological response evaluation criteria in solid tumors (RECIST) may be inadequate for the assessment of response to anti-angiogenic agents, results from the RadioCORRECT post-hoc analysis of a cohort of patients with mCRC treated within the CORRECT trial, recently showed that the development of lung metastases cavitation at the eight week CT scan predict a favorable outcome to regorafenib. Lung metastases cavitation may thus represent an early novel radiological marker of favorable outcome deserving consideration in this setting [72].

More recently, two phase II trials investigated the use of regorafenib in earlier treatment lines and in combination with chemotherapy regimens [73,74]. Results from these trials showed, respectively, that regorafenib in combination with mFOLFOX6 as first-line treatment did not improve ORR over historical controls, while that the addition of regorafenib on an intermittent schedule (one week on, one week off) to FOLFIRI as a second-line in patients who progressed after an oxaliplatin-based first-line therapy was tolerable, and resulted in a statistically significant improvement of PFS compared to FOLFIRI alone. Of note, data on regorafenib in patients previously untreated with anti-VEGFRs or anti-EGFRs are available from subgroup analyses of the CONCUR trial and show a greater magnitude of OS benefit in this group of patients compared to those who received any prior targeted therapy. Altogether, these data support the rationale of testing regorafenib in earlier treatment settings.

Currently, several regorafenib trials are ongoing, both in combination with other agents and as monotherapy in different treatment lines and patient settings, not only for the treatment of metastatic disease but also in the adjuvant and neoadjuvant setting (NCT02425683, NCT02664077, NCT02910843), while poor accrual led to early termination of a phase III study of regorafenib after liver metastasectomy (NCT01939223).

Finally, results from ongoing studies investigating safety and efficacy of different regorafenib schedules and dose approaches, such as the phase II RE-ARRANGE (NCT02835924) and ReDOS (NCT02368886) trials, alongside additional data from recently completed real-life studies such as the observational CORRELATE trial (NCT02042144), are awaited in order to improve tolerability and clinical management of this agent.

3.3.2. Nintedanib

Nintedanib, an indolinone derivate, is a new oral triple angiokinase inhibitor, which is able to block with high affinity multiple kinase receptors involved in the angiogenic pathway. It binds to all three VEGFR subtypes, PDGFRs, and FGFRs, inhibiting the intracellular signaling cascade. This triple inhibition may play a role in overcoming compensatory angiogenesis and mechanism of escape to single or dual angiogenesis inhibitors and support the rationale of testing nintedanib in patients with intrinsic or acquired resistance to other anti-VEGF agents [75].

The LUME-Colon 1, a global randomized phase III trial, investigated the safety and efficacy of best supportive care plus either nintedanib (200 mg twice daily) or placebo after failure of all standard treatments in 768 mCRC patients. Nintedanib significantly improved mPFS with 1.5 versus 1.4 months in the placebo arm (HR 0.58; p < 0.0001), without determining significant difference in OS (6.4 vs. 6.1 months in nintedanib and placebo arm, respectively, HR 1.01; p = 0.086). Disease control rates were 26 and 11% in the two arms respectively, favoring nintedanib (odds ratio 2.96; p < 0.0001). The safety profile was manageable, and most frequent grade 3 or worse adverse events were liver enzymes increase (16 vs. 8% in the placebo arm) and fatigue (9 vs. 6%) [76].

A phase I/II randomized trial exploring nintedanib combined with mFOLFOX-6 in the first-line setting showed interesting results in terms of PFS and a manageable safety profile [77]. Finally, an ongoing phase I/II study is evaluating safety and efficacy of nintedanib plus capecitabine in refractory mCRC patients (NCT02393755).

Despite signals of nintedanib activity in mCRC, the LUME-Colon1 trial did not meet both its co-primary endpoint. Analyses of potential predictive biomarkers might help to optimize a possible future role of nintedanib in the treatment scenario of mCRC.

4. Potential biomarkers

Since the approval of bevacizumab, anti-VEGF signaling drugs have changed the treatment paradigm for patients with mCRC. Extensive efforts and research have been made to identify predictive biomarkers for anti-angiogenic therapies in the last decade, yet no predictive marker is currently available for use in clinical practice [78]. Because of the lack of predictive biomarkers, clinical benefits may be modest and toxicities can be substantial. In addition, considering the high cost of these therapies, predictive markers are not only a clinical necessity but are becoming more and more an economic requirement.

Multiple candidates are in various phases of development including tissue, serum, and imaging biomarkers. However, the complexity of the angiogenesis pathway and the overlap between the various angiogenic factors present a significant challenge to biomarker discovery [79,80].

Many studies focused on plasma VEGF-A levels as a potential predictive biomarker, in particular for bevacizumab. Nevertheless, results have been largely inconsistent so far and it has been highlighted that pretreatment levels of VEGF-A may serve as a prognostic rather than predictive biomarker [81,82]. Various regulators of angiogenic pathways (such as VEGFRs and PlGF) have been studied in different type of cancer, including CRC. For instance, low serum levels of Ang-2 have been associated with favorable outcomes in terms of RR, PFS, and OS [11]. Despite their promising results, however, none of these have been validated either across different studies or in prospective analysis.

Kopetz et al. investigated the changes in plasma cytokines and angiogenic factors (CAFs) in a phase II trial in previously untreated mCRC patients who underwent FOLFIRI plus bevacizumab treatment. In this study interleukin-8 levels at baseline were associated with a shorter PFS (11 vs. 15.1 months, p = 0.03) and several other CAFs, including FGF, hepatocyte growth factor (HGF), stromal-derived factor-1 (SDF-1) and macrophage chemoattractant protein-3 (MCP-3), were found to be increased prior to evidence of disease progression, suggesting that they may predict resistance to anti-VEGF therapy [83]. Bates et al. analyzed CRC tumor samples from the phase III E3200 trial to test the predictive value of VEGF_165b, a VEGF splice isoform. Patients with a low level of VEGF_165b seemed to benefit more from bevacizumab treatment, however, survival results were not statistically significant [84]. More recently, Carbone et al. demonstrated that patients affected by Homeobox B9 (HOXB9)-negative tumors had a significantly longer PFS compared with those with HOXB9-positive tumors when treated with a first-line bevacizumab-containing regimen (18.0 vs. 10.4 months, p = 0.048). HOXB9 is a highly conserved homeobox transcription factor gene which drives neoplastic transformation and tumor progression through an anti-apoptotic process and tumor cell invasion. The authors demonstrated, with preclinical and clinical data, that transcription factor HOXB9 modulates resistance of pancreatic and colorectal cancer to bevacizumab modulating a complex network of alternative pro-inflammatory and pro-angiogenic secreted factors [85]. A prospective validation of these promising results is highly anticipated.

Another emerging key player in the angiogenesis regulatory pathways is the protein apelin (APLN), which is lately gaining interest because of its pro-angiogenic properties. Apelin signaling participates in several physiological functions and interacts at different levels with mechanisms regulating cell growth, survival and apoptosis. High APLN levels have been observed in colorectal cancer cell lines and patients derived samples [86]. Recently published data from Zuurbier et al. demonstrated that APLN mRNA levels are significantly associated with response in tumor-derived endothelial cells from patients receiving bevacizumab treatment. APLN levels were low in patients who benefitted from bevacizumab and high in non-responders (p = 0.0001) [87].

Finally, NOTCH1 expression was discovered to be a detrimental prognostic factor in mCRC patients treated with chemotherapy plus bevacizumab [88].

All these potential biomarkers, however, still need validation. In fact, promising initial findings frequently fail to be validated in prospective trials, as recently observed for LDH serum levels in patients treated with first-line FOLFIRI plus bevacizumab in an Italian phase II trial [89].

The entry into clinical practice of novel anti-angiogenic agents has increased the interest to identify specific biomarkers for each compound.

Wirapati and colleagues presented at the ASCO annual meeting 2017 data on RAS, BRAF and sidedness as biomarkers in patients treated with aflibercept in the VELOUR trial. Results of this pre-planned subgroup analysis showed no significant interactions according to RAS and BRAF status, although a trend for better outcomes was observed in BRAF mutants treated with aflibercept in comparison with the control arm (mOS 10.3 vs. 5.5 months, HR 0.42; 95% CI, 0.16–1.09; p = 0.08). Primary tumor sidedness did not affect treatment efficacy [90].

Tabernero et al. assessed the correlations of baseline marker levels (VEGF-C/D and sVEGF-1/2/3 in plasma; VEGFR-2 immunohistochemistry in tumor tissue) with clinical outcomes in the RAISE study. Only VEGF-D levels were found to be statistically significant: higher circulating levels of this soluble factor (≥115 pg/mL), in fact, were associated with improved ramucirumab efficacy in comparison with placebo both in terms of PFS (6.0 vs. 4.2 months, HR = 0.62, p < 0.0001) and OS (13.9 vs. 11.5 months, HR = 0.73, p = 0.00229) [91].

A retrospective analysis of circulating tumor DNA (ctDNA) from liquid biopsies collected from patients enrolled in the CORRECT study was performed to assess efficacy of regorafenib in patients bearing KRAS, PIK3CA and BRAF mutations. Results of this analysis, conducted on a subgroup of about 350 patients, were consistent with previous data and showed that the benefit on PFS and OS from regorafenib was irrespective of KRAS and PIK3CA mutational status, while analysis based on BRAF mutational status was not feasible due to the small number of BRAF-mutated patients [92]. More recently, Khan and colleagues demonstrated that the combination of dynamic contrast-enhanced (DCE)-MRI and ctDNA predict the duration of anti-angiogenic response to regorafenib in RAS mutant mCRC patients and may contribute to improving patient management [93].

Different single nucleotide polymorphisms (SNPs) in VEGF signaling pathway have been investigated. Results from a meta-analysis including 158 SNPs and 1348 patients enrolled in five randomized phase III trials suggested that VEGFA rs699946 and VEGFR-2 rs11133360 were associated with improved PFS in bevacizumab-treated patients [94]. Unfortunately, in a prospective study, Loupakis et al. were not able to validate retrospective findings on different candidate SNPs of VEGF/VEGFR pathway genes [95].

The complexity of the angiogenesis process and the relationship with other pathways make it challenging to find a predictive biomarker which can be used in clinical practice. Therefore, more efforts and future studies are warranted to reach this complex goal.

5. Conclusion

As discussed in the previous sections, it is well known that angiogenesis plays a major role in CRC development, and several agents targeting this complex signaling pathway proved to be manageable and effective in mCRC treatment. To date, anti-angiogenic agents are the only targeted agents that have demonstrated to confer a survival advantage in all treatments lines in mCRC. Data from clinical trials and several meta-analyses [96,97], support the rationale of a continuous angiogenesis inhibition strategy across the treatment sequence in the continuum of patients care.

Bevacizumab is registered for first- and second-line treatment in combination with chemotherapy. In the second-line setting, two other anti-angiogenic agents are available as well, aflibercept and ramucirumab, both in combination with FOLFIRI. Regorafenib, on the other hand, represents a therapeutic opportunity for patients progressing after standard chemotherapy and targeted therapies agents. While the availability of multiple anti-angiogenic options enriches the therapeutic armamentarium and treatment opportunities for our patients, the need to define the optimal treatment strategy and drugs sequence is critical.

In the first-line setting, a currently open issue is the choice of the best sequencing between anti-VEGFs and anti-EGFRs in RAS wild-type patients, as both agents are registered in combination with chemotherapy in the first- and second-line setting. Although some evidence seems to support the administration of an anti-EGFR followed by second-line anti-VEGF as the most effective strategy, the optimal sequence has yet to be fully defined. Similarly, no clear data are available to support the choice between a standard sequence of doublet chemotherapy plus anti-VEGF in first- and second-line versus an upfront triplet chemotherapy plus bevacizumab in eligible patients. Hopefully, results of ongoing trials investigating different treatment strategies, such as the phase III TRIBE-2 study (a comparison between first line FOLFOXIRI plus bevacizumab followed by maintenance and re-introduction of FOLFOXIRI-bevacizumab at disease progression versus the sequence of FOLFOX plus bevacizumab followed by second-line FOLFIRI plus bevacizumab) [98] and the STRATEGIC-1 study (investigating the optimal treatment sequence of the available treatment modalities in RAS wild-type mCRC) will help guide future therapeutic decisions.

The second-line scenario offers additional challenges, as multiple anti-angiogenic agents have demonstrated similar efficacy results in this setting. Although no formal cross-trial comparison is possible, it is interesting to note that mPFS and mOS improvements were numerically similar for all three anti-angiogenic agents evaluated in this setting (bevacizumab, aflibercept, and ramucirumab), while a significant improvement in ORR was observed only for aflibercept. Some differences in the study design, patient characteristics and toxicity profile, nevertheless, can help direct the clinical decisions when personalizing treatment choices (Supplementary Table 1). The chemotherapy regimen administered as first-line treatment (as both aflibercept and ramucirumab are registered only in combination with FOLFIRI), the efficacy of bevacizumab first-line therapy, tolerance to previous biologic treatment, treatment goal (i.e. the need for a dimensional response), should all be considered alongside a global evaluation of patient’s clinical conditions and comorbidities, disease characteristics and patient’s preferences when choosing the optimal treatment sequence for each patient.

In this treatment scenario, a crucial effort is directed toward the identification of reliable predictive biomarkers in order to aid clinical management of patients and identify subgroups more likely to benefit from different anti-angiogenic strategies. This would allow to optimize outcomes and maximize treatment efficacy, avoiding exposure of patients to potentially ineffective therapies. Although no predictive biomarker to anti-angiogenic agents is currently available in clinical practice, the growing knowledge on the complex mechanisms underlying the angiogenesis pathway, different potential target for treatments and mechanisms of tumor resistance, is opening new promising perspectives in this field.

6. Expert commentary

Tailoring chemotherapy and biologic treatment to an individualized approach is at the base of current clinical practice, and the availability of multiple anti-angiogenic agents in different treatment settings allow different options to personalize patients’ treatment. The different spectrum and targets of each agent may justify their rationale sequential use based on the possible mechanism of dynamic resistance and tumor escape from anti-angiogenic drugs. Unfortunately, although the emergence of several soluble biomarkers, such as FGF2, PlGF, and HGF, have been suggested to correlate with tumor progression during anti-VEGF treatment and could potentially drive subsequent treatment choices in the modulation of angiogenesis inhibition, no evidence of the clinical applicability of said strategy is currently available. Additionally, several multitargeted TKI agents with extensive anti-angiogenic properties failed to demonstrate activity in mCRC treatment, underlining that mechanisms of intrinsic and adaptive resistance to angiogenesis inhibition operates on multiple levels and have yet to be fully understood. Targeting alternative targets such as pericytes and cancer-associated fibroblasts, as well as antagonizing vessel co-option and vasculogenic mimicry have shown encouraging results in preclinical models, and further development of these strategies will be of interest in the next future. The role of tumor microenvironment in angiogenesis inhibitors resistance is another promising field of research. Myeloid-derived suppressor cells (MDSC), also known as CD11b + Gr1 + myeloid cells, comprising mainly neutrophils but also tumor-associated macrophages and dendritic cells with immunosuppressive and tumor-promoting capacities, have been demonstrated to contribute to resistance to anti-angiogenic therapy. In addition, it has been shown that tumor-infiltrating Th-17 cells and an intact interleukine-17 (IL-17) signaling pathway in the tumor microenvironment are responsible for mediating resistance to anti-angiogenic agents. IL-17 has a pro-inflammatory role by inducing cytokine expression from tumor associated fibroblasts, such as G-CSF, which is crucial for the mobilization and recruitment of CD11b + Gr1 + myeloid cells to the tumor microenvironment promoting VEGF-independent tumorigenesis [99]. Although several studies have been conducted to better understand these mechanisms, still little is known and the bench-to-beside process is rather arduous at this point.

In an era in which tumor molecular characterization has definitively entered common clinical practice and has a crucial part in decision-making in mCRC treatment, it has to be underlined that anti-VEGFRs are effective regardless of tumor RAS mutational status, and are the only approved targeted agents available for the treatment of RAS-mutated tumors. No predictive value on the efficacy of bevacizumab and novel VEGF inhibitors, in fact, has been demonstrated for KRAS mutations both in older studies [100], and recent aforementioned subgroup analyses of randomized trials. Notably, the combination of FOLFOXIRI plus bevacizumab represents the most promising first-line treatment option in clinically eligible BRAF mutated mCRC patients [30]. On the other hand, while only few data are available on safety and efficacy of anti-EGFRs in elderly (>70 years of age) mCRC patients, bevacizumab in combination with capecitabine proved to be effective as first-line treatment in this complex setting of patients [35] and is currently considered a standard in this patients population. Subgroup analyses from all trials investigating anti-angiogenic agents confirm feasibility, safety and efficacy of angiogenesis inhibition in the elderly population, as well as recently presented data from the randomized phase II PRODIGE 20 study, investigating tolerance and efficacy of first-line bevacizumab combined with different chemotherapy regimens (LV/5FU, FOLFOX or FOLFIRI) in patients aged 75 years and older [101].

Another topic of debate is related to maintenance during anti-angiogenic therapy. Many studies have been conducted to elucidate the cost-effectiveness of the maintenance therapy in advanced CRC patients. Among these, four studies have investigated bevacizumab and three have investigated anti-EGFR agents. No relevant benefit in OS has been demonstrated, even though a remarkable prolongation of the PFS1 has been observed [102]. Recently, an update form the CAIRO3 trial, showed that the subgroup of patients who benefit the most from anti-VEGF maintenance therapy are those harboring RAS/BRAF wild-type or V600E BRAF-mutated tumors [37]. Indeed, in RAS/BRAF wild-type group those who underwent maintenance therapy had a better median OS in comparison with the control group (25.7 vs. 19 months; HR = 0.68, p = 0.047). The V600E BRAF-mutated group showed similar results (15.8 vs. 13.6 months, HR = 0.32, p = 0.007), whereas in RAS mutant patients the benefit from maintenance therapy did not reach statistical significance. To date, maintenance treatment is a valid opportunity in the continuum of patents care in order to delay disease progression while allowing patients periods of less intense treatments, and should be discussed in a personalized treatment strategy. Additionally, specific molecular subtypes, such as BRAF-mutated tumors, generally characterized by a poorer outcome and more aggressive disease, seem to have a greater benefit from this treatment approach, supporting the crucial role of angiogenesis-inhibition in these tumors.

In recent years, different classifications of CRC based on gene expression signatures have been proposed. However, inconsistencies among these previous studies have been highlighted. For this reason, in 2015 an international consortium integrated these data and developed the Consensus Molecular Subtypes (CMS), which classify CRC into four distinct subgroups [103]. This is an unsupervised gene expression signature which has been shown to be prognostic and related to clinical outcomes, confirming the clinical relevance of the intrinsic biological processes implicated in each CMS. CMS1 (microsatellite instability immune) tumors are associated with high tumor mutational load (TML), microsatellite instability (MSI), hypermethylation status (CIMP+), BRAF mutation, female gender and right-sided primary. Notably, Lenz et al. recently reported that CMS1 may be a predictive biomarker of response to bevacizumab. In fact, looking at outcomes of patients enrolled in CALGB 80,405 trial according to tumor CMS profile, CMS1 treated in the bevacizumab arm showed longer median PFS (8.7 vs. 5.7 months, p = 0.01) and OS (22.5 vs. 11.7 months, p = 0.02) in comparison with the cetuximab arm. On the other hand, opposite results were observed for CMS2 (canonical) tumors, which are characterized by chromosome instability (CIN) due to high grade of somatic copy number alterations (SCNA), epithelial differentiation and marked upregulation of WNT and MYC pathways and are more frequently left sided [104]. These findings suggest that the CMS classification is not only highly prognostic but may also be predictive of anti-angiogenic treatment efficacy, warranting further exploration. Compelling evidence suggests that right-sided CRCs benefit from bevacizumab-based therapy while left-sided CRCs have better outcomes with cetuximab-based treatment in the first-line setting. However, the underlying causes of this phenomenon are yet to be clarified. It is now believed that tumor sidedness is the surrogate of embryological and molecular differences which drive cancer development, as demonstrated by CMS classification. Of note, several previous reports have pinpointed the immune-modulating effects of bevacizumab in different types of cancer, including CRC (reviewed elsewhere [105]). Since CMS1 has been associated with MSI, diffuse immune infiltrates and immune-associated signatures, this provides a sound rationale to the efficacy of bevacizumab in CMS1 and therefore in right-sided CRC, which deserves to be further explored.

7. Five-year view

The development of novel angiogenesis inhibitors and new more complex anti-angiogenetic strategies have been an intensive field of research in the latest years, and will hopefully contribute to broadening the horizons of current available therapies in the next future.

Several drugs have been abandoned due to lack of benefit or unfavorable toxicity profiles, the latest of which is the bispecific anti-VEGF-A and anti-ANG-2 monoclonal antibody vanucizumab, after negative results of the phase II McCAVE study [106]. Nevertheless, final results from ongoing trials are awaited regarding safety and efficacy of promising novel anti-angiogenic agents. Among those, famitinib, a multi-target RTKI, which previously showed antitumor activity in pre-treated mCRC patients [107], and fruquitinib, a TKI targeting VGFR-1, −2, and −3, which is currently under investigation in a phase III randomized, double-blind, placebo controlled trial in refractory mCRC patients in the Chinese population (NCT02314819). Additionally, other agents with anti-angiogenic activities are currently starting to be tested in mCRC, for example apatinib, a RTK inhibitor with a broad spectrum of targets that selectively inhibits VEGFR-2 (NCT03190616, NCT03271255).

New alternative strategies to enhance the efficacy of available agents are under study as well. Encouraging results from a phase II trial of the autophagy inhibitor hydroxychloroquine (HCQ) administered in combination with FOLFOX and bevacizumab as front-line treatment were recently presented [108]. A total of 37 patients were enrolled in the study, among those 28 were evaluable for response showing an ORR of 68% with an 11% of CR rate. Responses were independent of tumor molecular profile (i.e. KRAS, TP53, BRAF, and PIK3CA mutations). The 1 year OS rate was 74%, and mOS had not been reached at the time of data presentation. Most common grade 3 or worse adverse events included neutropenia (31%), fatigue (11%), thromboembolism (9%), and cardiac events (9%). HCQ-related toxicities included G1–3 insomnia (26%), G1–3 anxiety (20%), G1 visual disturbances (11%), and G3 allergic reactions (3%). Based on these data further evaluation in a randomized controlled trial is planned.

Moreover, new trials are already ongoing or expected to be planned, to explore the role of already approved anti-angiogenic agents in different treatment settings, or with different chemotherapy combinations or different administration schedules (i.e. regorafenib).

In this expanding scenario of potential treatment options, once again, the identification of predictive biomarkers and optimal treatment sequences will be of paramount importance in years to come.

Finally, the extraordinary advancements made in oncology in the past decade provided us with an increased number of more effective drugs to treat our patients, however, the elevated cost of many of these new compounds has raised concerns about the economic sustainability of the newly available treatments. For this reason, extensive efforts have been made to develop biosimilar drugs, which will help to lower healthcare costs and increase access to vital therapies. A biosimilar drug is a complex molecule which is highly similar to a biologic drug and has no clinically meaningful differences in terms of safety and effectiveness from the reference product. Both FDA and EMA follow strict and rigorous standard to ensure the safety and effectiveness of the biosimilar. On 14 September 2017 both FDA and EMA approved Mvasi (bevacizumab-awwb, Amgen Inc.) as a biosimilar to Avastin (bevacizumab, Genentech Inc.) for the treatment of mCRC, non-squamous non-small cell lung cancer, glioblastoma, metastatic renal cell carcinoma, and cervical cancer. The approval was based on comparisons of extensive structural and functional product characterization, preclinical and clinical data, between Mvasi and U.S.-licensed Avastin demonstrating that Mvasi is highly similar to Avastin and that there are no clinically meaningful differences between the two products. In next years targeted agents biosimilars will become part of our daily clinical practice, although currently more time is needed to overcome skepticism against these drugs in the health stakeholders.

Key issues.

• Therapeutic agents targeting VEGF/VEGFR signaling, proved to be effective across different treatment lines in patients with mCRC regardless of disease molecular status. However, preexisting or rapidly developing resistance mechanisms to anti-angiogenic drugs can limit their activity and eventually promote disease progression.

• Bevacizumab is registered for first- and second-line treatment in combination with chemotherapy. A benefit from continuing bevacizumab beyond first progression has been demonstrated to improve outcomes in mCRC patients. Maintenance treatment with bevacizumab plus capecitabine is a valid strategy in order to delay disease progression while allowing patients periods of less intense treatments.

• Both aflibercept and ramucirumab have been registered for the second-line treatment in combination with FOLFIRI. Of note, these drugs are effective also in patients pretreated with bevacizumab-based therapy in the first-line.

• Regorafenib is the first multikinase inhibitor which demonstrated activity in mCRC and is, to date, the only approved for the treatment of patients with advanced CRC after failure of all standard systemic anti-cancer therapies.

• Continuing angiogenesis inhibition in subsequent lines of treatment is a valid strategy in mCRC. However, the availability of multiple anti-angiogenic agents in different treatment settings warrants to tailor therapy to an individualized approach.

• Extensive efforts and research have been made to identify predictive biomarkers for antiangiogenic therapies in the last decade, yet no predictive marker is currently available for use in clinical practice.

• Emerging evidence suggests that right-sided CRCs benefit from bevacizumab-based therapy while left-sided RAS wildtype CRCs have better outcomes with anti-EGFRs-based treatment in the first-line setting.