The role of vascular endothelial growth factor inhibitors in the treatment of epithelial ovarian cancer

Abstract

Advanced epithelial ovarian, fallopian tube and primary peritoneal cancers (EOC) are a leading cause of gynaecological cancer-associated mortality and angiogenesis plays a key role in their growth. Vascular endothelial growth factor inhibitors (VEGFi) disrupt angiogenesis and improve the response rate, progression-free survival and in some cases, overall survival, when administered with and following cytotoxic chemotherapy, irrespective of the platinum sensitivity of EOC. Recent data have identified new indications for VEGFi in EOC: repeated exposure to VEGFi in the first- and then second-line treatment has sustained clinical efficacy; combinations of VEGFi with poly (ADP-ribose) polymerase inhibitors (PARPi) have proven effective as first-line or second-line maintenance regimens. However, recent trial data have not shown improved outcomes with combinations of VEGFi and immune checkpoint inhibitors. There remains a critical need to optimise patient selection for these effective yet somewhat toxic and expensive treatments. The search continues for validated biomarkers to optimise the use of VEGFi, of which the most promising at present is plasma Tie2. Based upon these studies, we propose a model of care incorporating VEGFi into the treatment of EOC, highlighting the need to change from the prescription of single courses of VEGFi, to allow use and re-use as clinically indicated.

Introduction

Ovarian cancer is the leading cause of gynaecological mortality in developed countries and accounts for 180,000 deaths per year worldwide [1]. Due to the occult symptomatology in early disease and the lack of a reliable screening instrument, the majority of the epithelial ovarian, fallopian tube and primary peritoneal cancers (henceforth ovarian cancer or EOC) are diagnosed at an advanced stage. While response rates to initial treatment with surgery and platinum-based doublet chemotherapy are high, the optimum use of maintenance therapies such as vascular endothelial growth factor inhibitors (VEGFi) and poly (adenosine diphosphate-ribose) polymerase inhibitors (PARPi) has not been fully defined [2].

Angiogenesis is a key requirement for tumour growth in EOC [3]. A range of growth factors stimulate and regulate the process, driving an ‘angiogenic switch’ that promotes blood vessel development within solid tumours [4, 5]. Of these, vascular endothelial growth factor (VEGF) is the pro-angiogenic growth factor with the greatest clinical validity [6].

VEGF is released by tumour cells under the influence of hypoxia and signals in a paracrine fashion through VEGF receptor-2, to increase vascular endothelial permeability and leakage of plasma proteins [6]. It then upregulates delta-like ligand-4 expression facilitating the formation of new endothelial tip cells that are differentiated from stalk cells through NOTCH expression and signalling. This process, known as sprouting angiogenesis, requires anastomosis and then canalisation following which, perfusion can occur. The growth of metastatic disease is reliant on the formation of new blood vessels [7], and tumours use sprouting angiogenesis, along with other less VEGF-dependent methods to acquire nutrition and oxygen, including vessel co-option, vascular mimicry, vessel splitting or recruitment of endothelial progenitor cells [6], to provide a supportive tumour microenvironment.

The inhibition of VEGF is attractive in EOC because of its critical role in angiogenesis, the potential for VEGF autocrine circuits to exist in ovarian cancer cells [8] and because VEGF increases vascular permeability, which contributes to the formation of ascites, in vivo. The first VEGF inhibitor to reach the clinic was the humanised anti-VEGF monoclonal antibody, bevacizumab, which induces ‘vascular normalisation’ in tumour vessels by pruning nascent vessels and restoring pericyte coverage of endothelial cells [9]. This reduces tumour interstitial pressure [10], leads to the more efficient delivery of chemotherapeutic agents and supports infiltration of immune effector cells into tumour tissue [11].

Bevacizumab is largely well tolerated but can occasionally cause severe side effects including haemorrhage (36% mucocutaneous bleeding of any grade, 1% grade 3 or higher), thrombosis (11% any grade), delayed wound healing (5% any grade) and gastrointestinal perforation [12–14]. The rate of any-grade perforation was 11% in earlier trials of heavily pre-treated patients [13], and this complication was reduced to 1–3% in first-line maintenance trials [14, 15] which excluded patients with bowel wall disease or obstruction. More commonly, hypertension and proteinuria are class-specific toxicities affecting all VEGFi, whereas orally administered small molecule VEGF-receptor tyrosine kinase inhibitors (TKI) have more off-target effects (Table 1) and thus are associated with a greater prevalence of gastrointestinal and dermatological toxicities.

Table 1.

Vascular endothelial growth factor inhibitors that have been assessed in clinical trials for use in epithelial ovarian cancer (EOC) and their molecular targets, common significant toxicities, and efficacy as a monotherapy treatment for recurrent EOC in both platinum-sensitive and platinum-resistant settings.

Drug	Target	Most common grade 3–5 adverse eventsa	Efficacy as single agent in recurrent EOC		Discontinuation due to AE
			Platinum-sensitive	Platinum-resistant
Monoclonal antibody
Bevacizumab [13, 52]	VEGF-A	Gastrointestinal perforation (all grades: 11.4%), hypertension (9.1%), abdominal digestion (4.5%), abdominal pain (4.5%), dyspnoea (4.5%), fatigue (4.5%)	PFS 4.7 months, ORR 21% (n = 62, 41.9% p-s)	PFS 4.4 months, ORR 15.9% (n = 44)	3.2–18.2% (with 3 treatment-related deaths)
Ramucirumab [71]	VEGFR-2	Headache (10%), fatigue (3.3%), hypertension (3.3%), thrombocytopenia (3.3%)	PFS 3.9 months, ORR 6.7% (n = 15)	PFS 3.2 months, ORR 4.4% (n = 45)	11.7%
Receptor tyrosine kinase inhibitor
Apatinib [70]	VEGFR-2 ATP-binding site	Hand-foot syndrome (17.2%), hypertension (10.4%), thrombocytopenia (3.4%)	PFS 5.1 months, ORR: 41.4% (n = 29, 38% p-s)		8.7%
Cediranib [68]	VEGFR-1/2/3, c-Kit	Hypertension (27%), fatigue (20%), diarrhoea (14%), headache (6%)	PFS 7.2 months, ORR 23% (n = 39)	PFS 3.7 months, ORR 0% (n = 35)	25.7% (1 treatment-related death)
Nintedanib [83]	VEGFR-1/2/3, PDGFR, FGFR1/3	Liver enzyme elevation (15%), diarrhoea (7%), abdominal pain (7%)	PFS-6 11.5%, ORR 7.4% (n = 27, bevacizumab-resistant pop.,18.5% p-s)		nr
Pazopanib [132]	VEGFR-1/2/3, PDGFR, c-Kit	Fatigue (11%), GGT elevation (11%), diarrhoea (8%), ALT elevation (8%)	PFS-6 17%, response rate 18% (n = 17)		28%
Sorafenib [133]	VEGFR-2/3, b-Raf, c-Raf, PDGFR, c-Kit, Flt-3	Dermatologic (19.7%), metabolic (14.1%), constitutional (4.2%), gastrointestinal (4.2%)	PFS-6 24%, ORR 3.4% (n = 71, 29.6% p-s)		12.7%
Sunitinib [69, 134]	VEGFR-1/2/3, PDGFR, c-Kit, Flt3	Lymphopenia (16.6%), leucopenia (13.3%), anaemia (10%), diarrhoea (10%), fatigue (10%), granulocytopenia (10%), hand-foot syndrome (10%), thrombocytopenia (10%)	PFS 4.1 months, response rate 3.3% (n = 30, 73% p-s)	PFS 4.8 months, ORR 16.7% (n = 36)	5.4–16.7%
Fusion protein
Aflibercept [72]	VEGFR-trap	Hypertension (27.5%), proteinuria (7.3%), asthenia (4.6%), headache (4.6%), fatigue (3.7%)	nd	PFS 13.3 weeks, ORR: 4.6% (n = 109)	21.9% (2 treatment-related deaths)

AE adverse event, ALT alanine aminotransferase, ATP adenosine triphosphate, EOC epithelial ovarian cancer, FGFR fibroblast growth factor receptor, Flt3 fms-like tyrosine kinase 3, GGT gamma-glutamyl transferase, mPFS median progression-free survival, mTTP median time to progression, nd no data, nr not reported, ORR objective response rate, PDGFR platelet-derived growth factor receptor, PFS median progression-free survival, PFS-6 six month progression-free survival, pop. population, p-s platinum-sensitive, VEGF vascular endothelial growth factor, VEGFR vascular endothelial growth factor receptor.

^aAdverse event prevalence listed in descending frequency as reported in trials of single-agent therapy in the recurrent setting.

Recent trials have demonstrated the clinical potential of PARPi to exploit vulnerabilities in EOC cells that harbour defects in the homologous-recombination pathway through synthetic lethality. Like VEGFi, these agents have been used as maintenance therapy in EOC but there remain questions around the best sequence and selection of maintenance therapies if we are to maximise the benefit for our patients. This review focuses on the role of VEGFi in current clinical practice; the implication of recent phase II/III trials in EOC assessing re-use after progression and the combination with PARP inhibitors and other novel therapies, and concludes with a suggested pathway for the sequencing of systemic maintenance treatments in EOC.

Current use of VEGF inhibitors

First-line treatment

The utility of VEGFi has been firmly established in first-line therapy from the results of two large randomised phase III clinical studies, GOG-0218 [15] and ICON7 [14] (Table 2). Both trials randomly allocated patients to receive carboplatin and paclitaxel with or without concurrent and maintenance bevacizumab for 15 or 12 months, respectively. GOG-0218 also included a concurrent bevacizumab arm with chemotherapy without maintenance treatment. Maintenance therapy demonstrated a median progression-free survival (PFS) increase of 3.8 months in GOG-0218, and 4.1 months (using restricted mean survival time due to the presence of non-proportional hazards) in ICON7 for the cohort of patients at high risk of progression, namely FIGO stage IV disease, inoperable stage III or sub-optimally debulked stage III with more than 1.0 cm residual disease after surgery. This effect on PFS was demonstrated across both trials despite ICON7 utilising a lower dose regimen of 7.5 mg/kg instead of 15 mg/kg.

Table 2.

Major clinical trials assessing the efficacy of vascular endothelial growth factor inhibitors in advanced ovarian cancer.

Trial name	Phase	Population	Interventional arm	Control arm	Outcome	Intervention	Control	HR (95% CI)	p-value
First-line therapy - monotherapy
GOG-0218 [15, 87] (n = 1873)	III	FIGO incompletely resected stage III or any stage IV EOC	(1) TC + bevacizumab 15 mg/kg 3-weekly, then placebo maintenance	TC + placebo then placebo maintenance	Concurrent-only arm: PFS	11.2 mo	10.3 mo	0.908 (0.795–1.040)	0.16
					Concurrent-only arm: OS	40.8 mo	41.1 mo	1.06 (0.94–1.20)	0.34
			(2) TC + bevacizumab 15 mg/kg 3-weekly, then bevacizumab maintenance for 16 cycles		Maintenance arm: PFS	14.1 mo	10.3 mo	0.717 (0.625–0.824)	<0.001
					Maintenance arm: OS	43.4 mo	41.1 mo	0.96 (0.85–1.09)	0.53
ICON7 [14, 16] (n = 1528)	III	FIGO stage I–IIA high risk or advanced stage IIB–IV EOC	TC + bevacizumab 7.5 mg/kg 3-weekly, then 12 cycles maintenance	TC	ITT: PFS (RMST)	19.9 mo (29.2 mo)	17.5 mo (27.7 mo)	0.93 (0.83–1.05)	0.25
					ITT: OS (RMST)	58.0 mo (45.5 mo)	58.6 mo (44.6 mo)	0.99 (0.85–1.14)	0.85
					High risk: PFS (RMST)	16.0 mo (20.0 mo)	10.5 mo (15.9 mo)	0.73 (0.61–0.88)	0.001
					High risk: OS (RMST)	39.7 mo (39.3 mo)	30.2 mo (34.5 mo)	0.78 (0.63–0.97)	0.03
ROSiA [19] (n = 1021)	IIIb	FIGO stage I–IIA high risk or advanced Stage IIB–IV EOC	TC + bevacizumab 15 mg/kg or 7.5 mg/kg 3-weekly up to 24 months	N/A (historical comparison)	ITT: PFS	25.5 mo	N/A	N/A	N/A
ANTHALYA [32] (n = 95)	II	FIGO stage IIIC or IV EOC deemed ineligible for primary complete debulking	Neo-adjuvant TC + bevacizumab 15 mg/kg, then IDS, then adjuvant TC + bevacizumab continued as maintenance for 12 months	Neo-adjuvant TC, then IDS, then adjuvant TC + bevacizumab 15 mg/kg continued as maintenance for 12 months	ITT: CRR at IDS	58.6%	51.4%	nr	nr
AGO-OVAR 12 [22, 23] (n = 1366)	III	FIGO stage IIB–IV	TC + nintedanib 200 mg BD, then nintedanib maintenance up to 120 weeks	TC + placebo then placebo maintenance	ITT: PFS	17.6 mo	16.6 mo	0.86 (0.75–0.98)	0.029
					ITT: OS	62.0 mo	62.8 mo	0.99 (0.83–1.17)	0.86
AGO-OVAR16 [21, 24] (n = 940)	III	FIGO stage II–IV EOC treated with surgical debulking and received at least 5 cycles of taxane-platinum chemotherapy	Pazopanib 800 mg daily up to 24 months	Placebo	ITT: PFS	17.9 mo	12.3 mo	0.766 (0.643–0.911)	0.0021
					ITT: OS	59.1 mo	64.0 mo	0.960 (0.805–1.145)	0.6431
First-line therapy - combination regimens
PAOLA-1 [33] (n = 806)	III	FIGO stage III/IV EOC with CR/PR after surgery and first-line platinum-taxane chemotherapy and bevacizumab	Bevacizumab 15 mg/kg 3-weekly for 15 months + olaparib 300 mg BD for 24 months	Bevacizumab 15 mg/kg 3-weekly for 15 months + placebo	ITT: PFS	22.1 mo	16.6 mo	0.59 (0.49–0.72)	<0.001
					HRD incl BRCA mutations: PFS	37.2 mo	17.7 mo	0.33 (0.25–0.45)	nr
					HRD excl BRCA mutations: PFS	28.1 mo	16.6 mo	0.43 (0.28–0.66)	nr
					HRD-negative/unknown: PFS	16.9 mo	16.0 mo	0.92 (0.72–1.17)	nr
IMagyn050 [117] (n = 1301)	III	FIGO Stage III/IV EOC with primary surgery and gross residual disease, or NACT and interval surgery	TC + bevacizumab 15 mg/kg + atezolizumab 1200 mg 3-weekly, then bevacizumab + atezolizumab maintenance for 16 cycles	TC + bevacizumab 15 mg/kg + placebo 3-weekly, then bevacizumab + placebo maintenance for 16 cycles	ITT: PFS (co-primary with OS)	19.5 mo	18.4 mo	0.92 (0.79–1.07)	0.28
					PD-L1 positive: PFS	20.8 mo	18.5 mo	0.80 (0.65–0.99)	0.038
Platinum-sensitive recurrence - monotherapy
OCEANS [34, 39] (n = 484)	III	First recurrence of EOC and disease progression ≥ 6 months after first-line chemotherapy	CG + bevacizumab 15 mg/kg 3-weekly until PD	CG + placebo	ITT: PFS	12.4 mo	8.4 mo	0.484 (0.388–0.605)	<0.0001
					ITT: OS	33.6 mo	32.9 mo	0.95 (0.77–1.18)	0.65
GOG-0213 [bevacizumab component] [35] (n = 674)	III	Recurrent EOC and disease progression ≥6 months with clinical complete response after first-line platinum chemotherapy	TC + bevacizumab 15 mg/kg 3-weekly until PD	TC	ITT: PFS	13.8 mo	10.4 mo	0.628 (0.534–0.739)	<0.0001
					ITT: OS	42.2 mo	37.3 mo	0.829 (0.683−1.005) a0.823 (0.680–0.996)	0.056 a0.0447
AGO-OVAR 2.21 [44] (n = 682)	III	First recurrence of EOC and disease progression ≥6 months after first-line chemotherapy	C/PLD + bevacizumab 10 mg/kg 2-weekly C1-6, then 15 mg/kg 3-weekly until PD	CG + bevacizumab 15 mg/kg 3-weekly until PD	ITT: PFS	13.3 mo	11.6 mo	0.81 (0.68–0.96)	0.012
					ITT: OS	31.9 mo	27.8 mo	0.81 (0.67–0.98)	0.032
Platinum-sensitive recurrence - combination regimens
ICON6 [46, 47] (n = 456)	III	First recurrence of EOC and disease progression ≥6 months after first-line chemotherapy	(1) Platinum-doublet chemotherapy + concurrent cediranib 20 mg daily, then placebo maintenance until PD	Platinum-doublet chemotherapy + placebo concurrent, then placebo as maintenance until PD	Concurrent-only arm: PFS	9.9 mo	8.7 mo	nr	nr
			(2) Platinum-doublet chemotherapy + concurrent cediranib 20 mg daily, then cediranib as maintenance until PD		Maintenance arm: PFS	11.0 mo	8.7 mo	0.56 (0.44–0.72)	0.0001
					Maintenance arm: OS (RMST)	27.3 mo (34.2 mo)	19.9 mo (29.4 mo)	0.85 (0.66–1.10)	0.21
NCI-8348 [93, 94] (n = 90)	II	Recurrent EOC and disease progression ≥6 months after platinum chemotherapy	Olaparib 200 mg BD + cediranib 30 mg daily	Olaparib 400 mg BD	ITT: PFS	16.5 mo	8.2 mo	0.50 (0.30–0.83)	0.006
					ITT: OS	44.2 mo	33.3 mo	0.64 (0.36–1.11)	0.11
AVANOVA2 [89] (n = 97)	II	Recurrent EOC and disease progression ≥6 months after platinum chemotherapy	Niraparib 300 mg daily + bevacizumab 15 mg/kg	Niraparib 300 mg daily	ITT: PFS	11.9 mo	5.5 mo	0.35 (0.21–0.57)	<0.0001
					HRD-positive: PFS	11.9 mo	6.1 mo	0.38 (0.20–0.72)	nr
					HRD-negative: PFS	11.3 mo	4.2 mo	0.40 (0.19–0.85)	nr
NRG-GY004 [95] (n = 565)	III	Recurrent EOC and disease progression ≥6 months after platinum chemotherapy	(1) Olaparib 300 mg BD until PD	Platinum-doublet chemotherapy	Monotherapy arm: PFS	8.2 mo	10.3 mo	1.20 (0.93–1.54)	nr
			(2) Olaparib 200 mg BD + cediranib 30 mg daily until PD		Combination arm: PFS	10.4 mo	10.3 mo	0.856 (0.66–1.11)	0.08
Platinum-sensitive recurrence - re-challenge
MITO16b [45] (n = 405)	III	FIGO stage IIIB-IV EOC treated with bevacizumab in the first-line, with first recurrence ≥6 months after platinum chemotherapy	Platinum-doublet chemotherapy + bevacizumab 15 mg/kg 3-weekly (or 10 mg/kg 2-weekly) until PD	Platinum-doublet chemotherapy	ITT: PFS	11.8 mo	8.8 mo	0.51 (0.41–0.65)	<0.0001
					ITT: OS	26.7 mo	27.1 mo	0.99 (0.73–1.39)	<0.98
Platinum-resistant recurrence or platinum-refractory
AURELIA [53] (n = 361)	III	Recurrent EOC and disease progression within 6 months after ≥4 cycles platinum chemotherapy	Non-platinum chemotherapy + bevacizumab 15 mg/kg 3-weekly (or 10 mg/kg 2-weekly) until PD	Non-platinum chemotherapy	ITT: PFS	6.7 mo	3.4 mo	0.48 (0.38–0.60)	<0.001
					ITT: OS	16.6 mo	13.3 mo	0.85 (0.66–1.08)	<0.174
MITO-11 [61] (n = 74)	II	Recurrent EOC and disease progression during first-line chemotherapy, or relapsed within 6 months of last platinum treatment	Weekly paclitaxel + pazopanib 800 mg daily	Weekly paclitaxel	ITT: PFS	6.35 mo	3.49 mo	0.42 (0.25–0.69)	0.0002
					ITT: OS	19.1 mo	13.7 mo	0.60 (0.32–1.13)	0.056

The primary outcome of each trial is in bold.

AUC area under the curve, BD twice daily, C/PLD carboplatin and pegylated-liposomal doxorubicin chemotherapy, CG carboplatin and gemcitabine chemotherapy, CRR complete resection rate, EOC epithelial ovarian cancer, FIGO International Federation of Gynecology and Obstetrics, HR hazard ratio, HRD homologous-recombination deficiency, IDS interval debulking surgery, ITT intent-to-treat, mo months, N/A not applicable, nr not reported, OS overall survival, PD progressive disease, PFS progression-free survival, RMST restricted mean survival time, TC taxane and carboplatin chemotherapy.

^aAdjusted analysis due to incorrect stratification data.

A benefit to overall survival (OS) was also observed within the high risk of progression subgroup in ICON7 of an additional 4.8 months [16], although this was not observed in GOG-0218 in either the pre-specified or exploratory subgroup analysis using the ICON7 high-risk definition. However, the stage IV cohort in GOG-0218 had a significant 10.2 month OS benefit over the control arm and a hazard ratio (HR) of 0.75 (95% CI 0.59–0.95), replicating the benefit seen in ICON7 in those at the highest risk of progression. The effect of first-line treatment on OS may well be obscured in these studies due to crossover and the influence of post-progression treatments in EOC [17, 18]. Bevacizumab did not provide any benefit as an adjuvant treatment for patients with FIGO stage I–II or completely resected stage III disease within ICON7, consistent with the notion that micro-metastatic disease does not require neo-angiogenesis [3].

The duration of exposure to VEGFi appears to be an important factor. The non-proportional hazards for progression observed in ICON7, which peaked in favour of the bevacizumab arm at approximately the end of planned maintenance therapy, and diminished thereafter once VEGFi were withdrawn, hints that the biological effect may be related to continued exposure to VEGF blockade. This is also supported by GOG-0218, which demonstrated the futility of concurrent bevacizumab administration during chemotherapy without maintenance, and also reported an equivalent maximal separation of the PFS curves at the completion of protocol bevacizumab administration and convergence nine months thereafter.

Extended bevacizumab maintenance therapy is thus the logical next point for evaluation. Bevacizumab was well tolerated in first-line therapy, with 62% patients receiving all planned 18 cycles in ICON7 and 14% ceasing early due to insufficient therapeutic response. Only 7.9% of patients stopped bevacizumab during the maintenance phase for adverse events or inter-current illness. Concerns regarding increased gastrointestinal perforation with extended therapy have not been demonstrated, with no significant increase in events observed by extending therapy to 24 months in a single-arm trial [19]. In this trial, 33% of patients discontinued bevacizumab for disease progression, and 17% for unacceptable toxicity (mostly haematuria and proteinuria), demonstrating the tolerability of extended therapy. In view of these studies, the anticipated benefit of extended exposure is that PFS will be improved; however, the data from the AGO-OVAR 17/BOOST trial evaluating 15 versus 30 months of bevacizumab has recently reported no significant improvement to PFS [20]. This is an interesting result which confirms that the optimal duration in first-line maintenance therapy remains the standard-of-care 15 months.

Although the concept of extending first-line bevacizumab appears to be supported by available data, the same hypothesis is not consistently supported by data arising from studies of VEGF-receptor TKI. Pazopanib, administered for 24 months as maintenance therapy, demonstrated a statistically significant 5.6 month PFS advantage over placebo [21], and 120 weeks of nintedanib maintenance in AGO-OVAR 12 [22] led to a modest but statistically significant 1.0 month PFS advantage [23]. On the other hand, both of these TKI were associated with significant toxicity; reduced OS amongst East-Asian patients receiving pazopanib [24] and reduced response rate and the likelihood of optimal debulking surgery when nintedanib was combined with neoadjuvant chemotherapy [25]. A similar study of sorafenib in the neo-adjuvant setting was terminated after the first four patients enrolled experienced intolerable increases in toxicity and morbidity [26]. Improved understanding and hence better management of diarrhoea [27] related to TKI is important as there is a clear role for VEGFi as first-line maintenance in patients at high risk of progression.

The advent of PARPi as first-line maintenance has complicated the selection of post-chemotherapy treatment strategies, with several trials showing dramatic PFS improvement particularly in patients with BRCA1/2 mutations and/or homologous-recombination deficiency (HRD) [28–30]. Due to a lack of head-to-head comparative clinical trials between PARPi and VEGFi, clinicians face challenges in determining the optimal treatment strategy. However, there are some important differences that enable treatment decisions: whilst the trials for PARPi required either complete or partial responses to surgery and chemotherapy before their administration, VEGFi are administered simultaneously with chemotherapy and do not have this limitation on patient eligibility. Furthermore, concurrent chemotherapy and VEGFi are postulated to enhance systemic delivery of chemotherapy via normalisation of tumour vasculature and restoration of pressure gradients [31]. Conceivably, this synergistic effect could allow for more effective chemotherapy delivery to tumours when given pre-operatively, and in patients who were deemed initially to have unresectable disease a complete resection rate of 58.6% was observed after neo-adjuvant bevacizumab and platinum chemotherapy in a single-arm trial [32].

Even after a response to systemic therapy, the rationale for switching from bevacizumab to PARPi after chemotherapy in the absence of a known BRCA1/2 mutation or HRD is unclear given the lack of PFS benefit in the concurrent-only arm of GOG-0218 and the only modest benefit of niraparib in the absence of BRCA1/2 mutations or HRD [30]. Such a strategy would use both maintenance options in the first-line setting, reducing their availability in the recurrent disease setting. Thus, the combination of VEGFi with chemotherapy to obtain maximal benefit from chemotherapy, and the addition of PARPi to the maintenance phase alongside bevacizumab would retain the relative advantages of both treatments, and this has been evaluated in PAOLA-1 [33]. While PARPi are currently substituted for bevacizumab in patients with germline or somatic BRCA1/2 mutations after first-line treatment [28], it is likely that the combination described in PAOLA-1 will become the standard maintenance regimen in this biologically defined group of patients, as described below.

Second-line, platinum-sensitive recurrence

Combining VEGFi with standard of care platinum-based chemotherapy has been established in three pivotal clinical trials in women who have relapsed platinum-sensitive EOC. OCEANS [34] and GOG-0213 [35] demonstrated the efficacy of the addition of bevacizumab to the previously established treatments of carboplatin and gemcitabine [36] or carboplatin and paclitaxel [37] chemotherapy, respectively, with maintenance bevacizumab administered until disease progression, in women with their first platinum-sensitive recurrence. Both trials demonstrated a statistically significant benefit in median PFS of between 3.4 and 4.0 months with the addition of bevacizumab, with a HR of 0.48 and 0.63 for OCEANS and GOG-0213, respectively. Hypertension and proteinuria were more common in bevacizumab-treated patients; however, discontinuation due to these toxicities occurred in 2.4–3.6% of patients in OCEANS [34]. More patients in the bevacizumab arm of GOG-0213 discontinued due to an adverse event (25% versus 11%), noting the study was not placebo-controlled. An OS advantage of 5 months was even observed in GOG-0213 [35], despite the recognised influence of crossover and post-progression treatments which can obscure OS differences in EOC [38]. Additionally, the OCEANS study was not powered sufficiently to detect OS differences between arms [39].

The PFS and OS advantage of secondary cytoreductive surgery, which was established in the DESKTOP III study [40, 41], has been challenged in trials that incorporated anti-angiogenic agents. Only 23% of patients were exposed to bevacizumab in DESKTOP III [42], whereas 84% of patients received bevacizumab with second-line chemotherapy in the surgical component of the GOG-0213 trial [43], which did not detect a difference in OS with secondary cytoreduction in contrast to DESKTOP III. Despite not restricting entry using the more restrictive AGO score in GOG-0213 to identify patients with resectable disease, the rates of complete gross resection and thus risk of relapse were comparable. Accepting the less restrictive eligibility criteria used in GOG-0213, the systemic treatment differences between these two studies suggest that bevacizumab may obviate the advantage associated with secondary surgery, though potential differences in rates of BRCA1/2 mutations and/or post-progression therapies including PARPi may also explain these discordant findings. Given the good tolerance of systemic therapy, this approach has the potential to limit unnecessary surgery in the non-curative context.

Prior bevacizumab therapy does not contraindicate usage in the recurrent setting. AGO-OVAR 2.21 [44] compared bevacizumab with either carboplatin and pegylated-liposomal doxorubicin (PLD) or carboplatin and gemcitabine, and revealed a median PFS and OS advantage with the PLD-based regimen. In this study, 48% of patients had received previous anti-angiogenic therapy, whilst OCEANS and GOG-0213 were conducted in largely bevacizumab-naïve patient populations. Subgroup analysis showed consistent treatment effect regardless of prior anti-angiogenic exposure, alluding to the potential efficacy of re-use of VEGFi which was formally evaluated in MITO16b (see below) [45].

Other VEGFi have also been investigated in the recurrent platinum-sensitive setting with evidence of a clinical effect. Cediranib concurrent with platinum-doublet chemotherapy and then as maintenance until progression, was evaluated in ICON6 [46]. A median PFS benefit was demonstrated over cytotoxic chemotherapy alone despite the premature termination of the study due to discontinued drug development [47], with a numerical trend towards but statistically non-significant OS advantage. In keeping with first-line trials, maintenance therapy was needed to observe benefit and there was non-proportionality of the survival curves [47], re-emphasising a duration-based biological effect from VEGFi.

Platinum-resistant setting

The standard of care therapy for platinum-resistant (relapse within six months of platinum-based therapy) and platinum-refractory (progressive disease during platinum-based therapy) EOC is single-agent non-platinum chemotherapy including weekly paclitaxel [48, 49], topotecan [50] or PLD [51]. Bevacizumab initially demonstrated modest activity as monotherapy in the recurrent setting [52] with 21% of patients achieving a clinical response in a cohort of both platinum-sensitive and resistant patients. However, a similar study [13] saw an elevated level of gastrointestinal perforations and treatment-related deaths, particularly in patients with radiological bowel involvement or more than three prior regimens of chemotherapy.

The AURELIA [53] study sought to explore the benefit of bevacizumab in combination with standard chemotherapy of physician’s choice in the platinum-resistant setting, excluding patients who had clinical or radiological evidence of bowel involvement or previous bowel obstruction or gastrointestinal perforation. This study revealed a significant improvement in median PFS with the addition of bevacizumab and an improved objective response rate (ORR) from 11.8 to 27.3% (p = 0.001). Greater median exposure to chemotherapy was seen in the bevacizumab arm, reflecting the longer PFS. Again, increased hypertension and proteinuria were observed with bevacizumab, and 2.2% of patients had a grade 2 or higher gastrointestinal perforation, though deaths unrelated to disease progression were equal in each arm. Despite insufficient power to investigate an OS difference, there was a trend towards an increased median OS which did not meet statistical significance. However, survival data were confounded as 40% of patients in the control arm crossed over to bevacizumab after progression.

An exploratory analysis of patients on AURELIA [54] who received bevacizumab on the experimental arm or after optional crossover, compared to those who never received bevacizumab, revealed a reduced risk of death in those who received bevacizumab (HR 0.68, 95% CI 0.52–0.90, p = 0.01). These findings were biased however by non-randomisation at crossover and the likely propensity for deteriorating patients not to receive crossover therapy. Subgroup analysis showed that patients treated with paclitaxel specifically had the greatest increase in median PFS from the addition of bevacizumab, from 3.9 to 10.4 months (HR 0.46, 95% CI 0.30–0.71) [55], suggesting that combination with paclitaxel is particularly synergistic [56]. There were no additional safety concerns regarding the frequency of bevacizumab toxicities, with gastrointestinal perforation occurring in 2.2% of bevacizumab-exposed patients versus none in the control arm, demonstrating comparable safety in this higher risk platinum-resistant population.

Other combinations involving VEGFi have been evaluated in smaller studies. Bevacizumab has shown activity when combined with metronomic oral cyclophosphamide [57] and the vascular disrupting agent fosbretabulin [58], though combinations with other targeted therapies including everolimus [59] and erlotinib [60] were not active. The MITO-11 study [61] compared paclitaxel alone with the addition of pazopanib, and this demonstrated a similar magnitude of PFS benefit to AURELIA, although this finding was not confirmed in a similar trial which included patients with platinum-sensitive disease [62]. Similarly, sorafenib plus topotecan was active [63], sorafenib with bevacizumab showed modest clinical activity [64] and apatinib with etoposide also showed activity in platinum-resistant disease [65]. However, nintedanib showed no activity when combined with cyclophosphamide [66].

Responses have also been observed with VEGF inhibitor monotherapy, including cediranib [67, 68], sunitinib [69] and apatinib [70]. However, not all VEGFi show activity, with less promising results using ramucirumab [71] and aflibercept [72].

Aside from survival, symptom-based endpoints are also clinically relevant in the palliative platinum-resistant context. VEGFi reduce vascular permeability which contributes to malignant ascites and cancer lymphoedema. Control of malignancy-related symptoms was improved in AURELIA, such that the subgroup of patients with ascites before treatment required fewer paracenteses in the bevacizumab arm (2% versus 17%) [53]. Patient-reported outcomes also showed a greater improvement in abdominal and gastrointestinal symptoms by week 8/9 (21.9% versus 9.3%, p = 0.002) [73]. Taken in conjunction with other reports of intraperitoneal bevacizumab and intravenous aflibercept as effective and safe treatments for malignant ascites [74, 75], these findings support the use of VEGFi to improve survival and symptom control in platinum-resistant EOC.

New indications

Re-use of VEGF inhibitors

Sustained inhibition of angiogenesis appears to be a clinically valid strategy as angiogenesis is driven by upregulation of VEGF and other pro-angiogenic molecules [76] rather than acquired genetic mutations [77, 78]. Conversely, withdrawal of VEGFi also leads to rapid revascularisation within the empty sleeves of the basement membrane and thus regrowth of tumours [79], hence continued exposure to VEGFi as tolerated would be biologically justified if we are to maximise PFS.

Treatment beyond progression has been demonstrated as clinically efficacious in other tumour types, notably metastatic colorectal cancer [80, 81] where repeated exposure to bevacizumab across multiple chemotherapy lines prolonged OS. As discussed earlier, a substantial proportion of the AGO-OVAR 2.21 trial patient population had received prior anti-angiogenic therapy with no suggestion of reduced efficacy of bevaicuzmab [44]. In view of these data, a formal evaluation of the re-use of VEGFi was strongly warranted.

MITO16b [45] evaluated re-use of VEGFi in women who had received bevacizumab for first-line maintenance, randomly allocating women at the time of first recurrence of platinum-sensitive disease to platinum-based chemotherapy with or without bevacizumab. The results showed a significant 3.0-month improvement in the primary endpoint of investigator-assessed median PFS in favour of repeated exposure to bevacizumab. Notably, this effect was preserved regardless of whether patients developed recurrent disease whilst on first-line maintenance bevacizumab or after the completion of first-line maintenance treatment, supporting the notion that resistance to VEGFi is from physiological adaptations rather than clonally-acquired evolution of tumour cells. Additional proteinuria and hypertension were seen with the bevacizumab arm, but there was no increase in the rate of gastrointestinal perforations nor chemotherapy interruption, and toxicity led to maintenance bevacizumab discontinuation in only 9% of patients. Median OS was not significantly different between groups, although this was not the primary endpoint and would likely be influenced by post-progression treatments. An apparent lack of efficacy in the subgroup with BRCA1/2 mutations may have been influenced by the unplanned, retrospective and incomplete collection of mutation status, and further studies are required to guide patient selection, acknowledging the availability of PARPi for EOC associated with BRCA1/2 mutations.

With the increasing use of PARPi in first-line maintenance, re-challenge with PARPi has not been validated as an effective strategy [82] and the results of OrEO/ENGOT Ov-38 (NCT03106987) are awaited, leaving bevacizumab the sole maintenance therapy that can be justified for re-use. This effect appears limited to bevacizumab, with nintedanib unable to replicate these results in a bevacizumab-resistant population [83]. The results of JGOG3023 (UMIN000017247) will further evaluate the efficacy of re-challenge with bevacizumab in the platinum-resistant setting [84].

Combination with PARP inhibitors

PARPi have proven to be a remarkable advance in the treatment of EOC. The rationale for combination with VEGFi is potential synergism from hypoxia-induced downregulation of the homologous-recombination repair genes BRCA1 and RAD51, increasing the synthetic lethality of PARPi [85]. Non-hypoxic mechanisms have also been described for cediranib that were affected by homology-directed DNA repair through platelet-derived growth factor receptor inhibition [86]. However, the extent to which these mechanisms pertain to the clinic is not clear as combinations of VEGF-receptor TKI with PARPi do not appear markedly more effective than bevacizumab-PARPi combinations, implying that the dominant effect is additive.

Once bevacizumab [16, 87] and olaparib [28] had become established in first-line maintenance therapy, it was appropriate to conduct the randomised phase III PAOLA-1 study [33]. This trial assessed the addition of 24 months of olaparib maintenance therapy in women with newly-diagnosed FIGO stage III or IV high grade serous or endometrioid ovarian cancer (or other non-mucinous epithelial histological subtype if associated with a germline BRCA1/2 mutation), if they had no evidence of disease or achieved a clinical complete or partial response after first-line platinum-taxane-bevacizumab therapy. Fifteen months of bevacizumab (15 mg/kg) maintenance was continued in both arms. A 19.5-month increase in median PFS was observed in the olaparib-bevacizumab combination arm in the cohort of patients with either a tumour BRCA1/2 mutation or HRD. However, patients with no tumour BRCA1/2 mutation and a negative or unknown HRD status had no PFS improvement with the addition of olaparib to bevacizumab. Subgroup analysis showed patients who only had a partial response to chemotherapy did not have the same likelihood of PFS advantage as the entire cohort (HR 0.85, 95% CI 0.61–1.18), raising uncertainty over the benefit of prescribing PARPi in patients who attain only partial responses to first-line remission-induction therapy; a finding that was also seen in first-line single-agent PARPi trials [30]. Thus, in patients who have a partial response, single-agent maintenance bevacizumab would be a reasonable option that retains PARPi as maintenance for subsequent recurrent disease, which is inevitable in the context of radiologically evaluable residual disease after completion of primary multimodality treatment.

The lack of a comparator, olaparib-only arm in PAOLA-1 limits analysis of whether the benefit in the presence of BRCA1/2 mutations and/or HRD is driven by PARPi alone, as observed in first-line maintenance trials of PARPi [28–30], or requires the interaction of bevacizumab. SOLO1, which assessed olaparib as first-line maintenance against placebo for patients with a BRCA1/2 mutation reported a HR of 0.30 [28], which approximates to the HR of 0.31 reported in PAOLA-1 for the subgroup of patients with a tumour BRCA1/2 mutation. However, the reported median PFS from SOLO1 of 56 months [88] is notably longer than the 37.2 months in PAOLA-1 for this same subgroup, acknowledging the differing trial populations. The results of AVANOVA2 which incorporated a monotherapy PARPi control arm [89], at least suggest additive benefit with bevacizumab regardless of HRD, albeit in the platinum-sensitive recurrent setting as discussed below. Additionally, concurrent bevacizumab improved the ORR to chemotherapy by 19% in ICON7 [14], which subsequently would increase the number of patients eligible for maintenance PARPi. Hence bevacizumab is likely to still play a role in first-line treatment by increasing the sensitivity to neoadjuvant chemotherapy, even for patients with a BRCA1/2 mutation.

The efficacy of dual PARPi and VEGFi has also been assessed for platinum-sensitive recurrences as either maintenance after chemotherapy, or as a ‘chemotherapy-sparing’ treatment strategy, given PARPi have efficacy comparable to non-platinum chemotherapy in specific settings [90–92]. The latter treatment strategy has been reported in the AVANOVA2 phase II study [89] which randomised patients to receive niraparib with or without bevacizumab at relapse without intervening chemotherapy. This showed a benefit to PFS with the combination in the intent-to-treat population, with a HR of 0.35 (95% CI 0.21–0.57, p < 0.0001) irrespective of HRD or chemotherapy-free interval. This finding supports the notion that VEGFi should not be unnecessarily restricted by HRD, unlike PARPi.

The use of cediranib with PARPi has also been examined. A phase II trial [93] of olaparib with or without cediranib for the treatment, rather than maintenance, of platinum-sensitive recurrences, reported a significant doubling of the median PFS [94]. However, the lack of a chemotherapy control arm made this data difficult to interpret for clinical use; hence the phase III NRG-GY004 trial was constructed to compare platinum-based chemotherapy with either olaparib monotherapy or the olaparib–cediranib combination. This study was reported at the 2020 ASCO Annual Meeting [95] as a negative result with no improvement in PFS with either oral treatment irrespective of BRCA1/2 mutation status. Unfortunately, the trial was not powered to detect non-inferiority of the oral treatment arms and 32% of patients on the standard chemotherapy arm had non-protocol (predominantly PARPi maintenance) therapy, which may have obscured any benefit. In fact, response rates were 71.3% to standard chemotherapy and 69.4% for olaparib–cediranib and the median PFS was 10.3 and 10.4 months, respectively. The germline BRCA1/2 mutation population had a superior median PFS of 18.0 vs 10.5 months with a HR of 0.55 (95% CI 0.32–0.94) for olaparib–cediranib over chemotherapy, suggesting a subgroup of patients who may experience greater benefit from this combination. Regardless, establishing non-inferiority would provide an alternative option to chemotherapy, and thus further studies are required to evaluate this strategy, particularly against the modern standard of chemotherapy and bevacizumab. The role of olaparib–cediranib as maintenance after chemotherapy for platinum-sensitive recurrence is also under evaluation in the phase III trial ICON9 (NCT03278717).

Dual VEGFi and PARPi may also have a role in overcoming acquired platinum resistance. A single-arm study of cediranib and olaparib in a germline BRCA1/2 wild-type, platinum-resistant cohort observed an ORR of 15.3% and median PFS of 5.1 months in 60 patients [96], which is comparable to single-agent non-platinum chemotherapy. Another randomised phase II study reported a 2.6-month increase in median PFS compared to weekly paclitaxel in a heavily pre-treated population, but this did not meet significance [97]. This combination is being evaluated further in the COCOS (NCT02502266) and OCTOVA (NCT03117933) trials. However, the addition of anti-angiogenic therapy to PARPi following prior resistance to the latter does not appear to overcome acquired PARPi resistance, with scarce responses observed in a phase II study testing olaparib and cediranib post-progression on PARPi [98]. Taking these data together, the role of VEGFi/PARPi may be restricted to platinum-sensitive disease but further evaluation is required.

Combination with immuno-oncology agents

The seminal correlation of the presence of CD3⁺ tumour-infiltrating T cells with PFS and OS in advanced EOC has prompted the investigation of immunotherapeutic approaches in treatment [99, 100]. Increased programmed cell death-ligand 1 (PD-L1) expression on human ovarian tumour cells is correlated with a poorer prognosis, inversely with intra-epithelial CD8⁺ T lymphocyte count [101], and promotes peritoneal dissemination of mouse ovarian cells [102]. However, immune checkpoint inhibitors (ICI) targeting the programmed cell death protein 1 (PD-1)/PD-L1 pathway have achieved only modest response rates ranging from 10 to 16% [103–105], rising to 31% for combined PD-1 receptor/cytotoxic T-lymphocyte-associated protein (CTLA) 4 blockade; [106] a lack of efficacy that may be related in part to the impact of p53 mutations on the cytoplasmic DNA sensing machinery, cGAS-STING-TBK1-IRF3 [107]. Mutations involving ARID1A in clear cell EOC, which result in chromatin remodelling defects that render the tumour more antigenic [108], may afford greater susceptibility to ICI in comparison to high-grade serous EOC, and account for reports describing clinical responses in clear cell tumours [109].

Several relationships between VEGF in the tumour microenvironment (TME) and immunosuppression have prompted the exploration of combination immunotherapeutic-angiogenic approaches to the treatment of EOC. Elevated VEGF levels induce immunosuppression through inhibition of cytotoxic T lymphocyte trafficking and function, inhibition of dendritic cell maturation and thus T-cell mediated response, and recruitment of immunosuppressive T_reg cells and tumour-associated macrophages to the TME [11], making VEGF blockade an important consideration to improving immune control of tumour cells. The combination of VEGFi with ICI has already been successful in other solid tumours including hepatocellular carcinoma [110], renal [111, 112], lung [113] and endometrial [114] cancers. Within EOC, a single-arm phase II clinical trial combining nivolumab, a PD-1 receptor monoclonal antibody, with bevacizumab [115] reported response rates of 29%, rising to 40% in the platinum-sensitive cohort. Another study utilising atezolizumab, a PD-L1 inhibitor, and bevacizumab, found only a modest response rate of 15% in the platinum-resistant setting [116].

The phase III IMagyn050 trial [117] investigated atezolizumab versus placebo in the first-line setting along with standard surgery, platinum-taxane chemotherapy and bevacizumab. This trial was negative for the primary endpoint of PFS improvement in either the intent-to-treat population or for PD-L1 expressing tumours, though the subgroup of patients with ≥5% PD-L1 expression on immune cells may potentially benefit from atezolizumab (HR 0.64, 95% CI 0.43–0.96). Overall survival may be more markedly influenced than PFS by ICI through alteration of the tumour microenvironment [118], and this will be evaluated when mature OS data is reported. Further assessment of this same regimen in the recurrent disease setting is forthcoming in the AGO-OVAR 2.29 trial [119].

In an attempt to maximise activity, PARPi have been added to the combination of VEGFi and ICI. Although these regimens are inducing anti-tumour activity in early phase clinical trial evaluation [120], further development will be challenging without biomarkers to guide patient selection. Otherwise, they may remain active but too expensive for widespread use. Various clinical trials of combinations are currently in recruitment.

Optimising patient selection

Whilst there is an established relationship between BRCA1/2 mutations, HRD and the benefit of PARPi either as monotherapy or in combination with other therapies such as in PAOLA-1, current commercially available genome sequencing approaches may not always predict individual response to treatments [121]. With the complexity of treatment pathways in EOC increasing and the availability of re-use of VEGFi in clinical practice, optimising patient selection using validated biomarkers for VEGFi is needed to individualise treatments for patients, particularly when balanced against the benefit from PARPi or their combination. Prolonged exposure to VEGFi increases the cumulative risk of toxicities including hypertension and proteinuria, and biomarkers for VEGF pathway inhibition would help prevent unnecessary exposure to futile therapy and the administration of alternative efficacious treatments. There is also significant healthcare cost associated with long durations of maintenance treatment and in healthcare utilisation for dispensing and administration of intravenous formulations. However, the cost of bevacizumab has fallen significantly since coming off patent, and because of the availability of bio-similars.

The 121 isoform of VEGF-A has been the most studied potential biomarker given its presence in circulating peripheral blood samples [122]. Unfortunately while modestly prognostic, pre-treatment plasma VEGF-A concentrations did not predict response to VEGFi retrospectively in several tumour types [123], nor prospectively in metastatic breast cancer [124]. In EOC, retrospective analyses from GOG-0218 failed to demonstrate a relationship between plasma VEGF-A levels and bevacizumab efficacy [125], although tumour VEGF-A expression considered alongside tumour micro-vessel density may be potentially predictive [126]. Taken together, these studies concluded that plasma VEGF-A₁₂₁ cannot be used as a predictive biomarker for VEGFi.

The search for other biomarkers is ongoing, and plasma tunica internal endothelial cell kinase (Tie2) has been identified as a generic tumour vascular response biomarker for bevacizumab in colorectal cancer [127], and in ovarian cancer where a 25% reduction in plasma Tie2 within 9 weeks of starting bevacizumab predicted PFS advantage [128]. In contrast, a 40% or greater increase in Tie2 concentration above the nadir predicted progressive disease; an effect that was additive with Ca-125 [129]. As a similar benefit was seen in biliary tract cancer where patients were treated with cediranib (manuscript in preparation) these studies imply that changes in plasma Tie2 concentrations are the first vascular response biomarker for generic VEGFi. These characteristics posit the absence of a decrease in circulating Tie2 levels after VEGF blockade as a potential biomarker to predict non-responders to treatment. This will be tested in the prospective study, VALTIVE1 (NCT04523116) [130].

Clinical factors can also determine which patients will derive the most benefit from VEGFi. Alongside FIGO stage, which restricts benefit to those with advanced sub-optimally debulked EOC, symptoms of malignant bowel obstruction or clinical bowel involvement delineate patients at greater risk of perforation who should avoid VEGFi. Conversely, patients with malignant ascites derive greater benefit from VEGFi, with fewer paracenteses [75] and greater palliation of malignant abdominal symptoms after treatment [73]. Within GOG-0218, ascites at baseline was a negative prognostic marker, but this subgroup had an improved PFS with the addition of bevacizumab, from 10.4 to 15.2 months (adjusted HR 0.71, 95% CI 0.62–0.81, p < 0.001), whilst those without ascites did not see any significant PFS or OS benefit with bevacizumab [131]. This defines a subgroup of patients with ascites at diagnosis who will gain the most benefit from the prescription of VEGFi.

The future standard of care

Based upon the clinical studies now reported, a proposed schematic of the optimal systemic treatment pathway for EOC at first diagnosis is shown at Fig. 1, and for recurrent disease in Fig. 2. Given the consistent efficacy of VEGFi regardless of platinum sensitivity, these therapies now feature as treatment options across all phases of EOC treatment in combinations utilising chemotherapy or PARPi.

Fig. 1. Optimal first-line systemic treatment of advanced epithelial ovarian cancer.

At initial diagnosis, patients proceed to IPS if deemed an appropriate surgical candidate for complete surgical resection, or receive neo-adj chemotherapy with the combination of C/T, with the addition of bev. Patients with an interval response after neo-adj chemotherapy may then proceed to DPS, with completion chemotherapy and bev. Patients who have had IPS and are FIGO stage IV or FIGO stage III with high-risk factors for progression (more than 1.0 cm residual disease) should have combination C/T with bev. If a complete response or no evidence of disease is seen at the end of treatment, then maintenance therapy should be guided by the presence of BRCA1/2 mutations and HRD, with the addition of a PARPi along with maintenance bev in patients with evidence of BRCA1/2 mutations or HRD. Patients who do not have HRD (or this is unknown) may either continue bev maintenance, or consider a change to maintenance PARPi if they have a complete response. Patients without evidence of a complete response to chemotherapy should continue on bev therapy only. *The use of pre-operative bev will be reported as part of the ICON 8B study. Adj adjuvant, bev bevacizumab, C/T carboplatin/paclitaxel, DPS delayed primary surgery, FIGO International Federation of Gynecology and Obstetrics, g/s BRCA mut germline/somatic BRCA1/2 mutation, g/s BRCA wt germline/somatic BRCA1/2 wild-type, HRD homologous-recombination deficiency, IPS immediate primary surgery, PARPi poly (adenosine diphosphate-ribose) polymerase inhibitor.

Fig. 2. Optimal systemic treatment of progressive or recurrent epithelial ovarian cancer.

a Patients who develop a platinum-sensitive recurrence should be evaluated for the possibility of secondary cytoreductive surgery on the basis of DESKTOP III eligibility criteria. Operated patients should then proceed to platinum-based chemotherapy after surgery, or consider platinum-based combination chemotherapy if not eligible for surgery. However, the addition of bevacizumab to platinum-based chemotherapy may negate the benefit to progression-free survival seen from cytoreductive surgery. Bevacizumab should be given concurrently with platinum chemotherapy and a taxane, gemcitabine or pegylated-liposomal doxorubicin depending on patient factors and suitability, with maintenance bevacizumab continued thereafter until progression. Bevacizumab treatment should not be withheld even if patients have received bevacizumab in the first-line setting. Patients who receive chemotherapy and have a CR or PR should consider a PARPi if the patient is PARPi-naïve or did not develop progressive disease whilst on a PARPi in the first-line setting, particularly if there is a known BRCA1/2 mutation. Another option for PARPi-naïve patients, who are not candidates for platinum-based chemotherapy, is treatment with a PARPi combined with a VEGFi. b Patients with platinum-resistant or platinum-refractory disease should be offered treatment with weekly paclitaxel or pegylated-liposomal doxorubicin, in combination with bevacizumab. PARPi are not efficacious in the platinum-resistant setting. Participation in a well-designed clinical trial is also appropriate given the limited efficacious options available to patients. C/G carboplatin/gemcitabine, C/PLD carboplatin/pegylated-liposomal doxorubicin, C/T carboplatin/paclitaxel, CR complete response, PARPi poly (adenosine diphosphate-ribose) polymerase inhibitor, PLD pegylated-liposomal doxorubicin, PR partial response, VEGFi vascular endothelial growth factor inhibitor.

Now that use and re-use have been validated clinically, and the healthcare costs of delivery have fallen significantly, underutilisation of VEGFi is an important concern that must be addressed through greater freedom in the funding and availability of VEGFi for treatment of EOC by respective funding bodies. Consideration should also be given to longer-term treatment with VEGFi in the first-line setting based upon biomarkers such as Tie2, and the evaluation of novel combinations of VEGFi and DNA repair inhibitors to maintain disease control in the post-PARPi era.

Author contributions

AM and GJ contributed to conceptualisation and outline. AM, GJ, AC and RM contributed to the manuscript preparation and review.

Funding

The author(s) received no specific funding for this work.

Data availability

Not applicable.

Ethics approval and consent to participate

No ethics approval required for this review of previously published literature.

Consent to publish

Not applicable.

Competing interests

GJ and AC receive research funding from AstraZeneca. AM and RM declare no competing interests.