Analysis of Efficacy/Safety Ratio of Angiogenesis-inhibitors based Therapies in Ovarian Cancer: A Comprehensive Meta-analysis

1. Introduction

Significant improvements have been made in the management of ovarian cancer, but this disease continues to have a huge burden on health care especially in countries with a low human development index. According to GLOBOCAN 2020 statistics, ovarian cancer remains the 3^rd in incidence and the 2^nd cause of death among the genital neoplasms. In central and eastern Europe, ovarian cancer has the highest incidence rate in the World (10,7/100.000) and a mortality rate of 5,6/100.000[1]. In present, the guidelines promote a combination of cytoreductive surgery and combination chemotherapy of platinum agents and taxanes. Although most patients will experience remission under therapy, 80% will relapse within 18 months[2]. The relapse is usually diagnosed in advanced stages, despite the initial response to platinum-based chemotherapy, and is showing increased chemo-resistance, leading in most cases to death[3,4]. Thus, new agents like angiogenesis inhibitors and PARP inhibitors were introduced as treatment options. Regardless of initial factors that leads to the neoplastic transformation of cells, due to the increased metabolic need and secondary tissular hypoxia, cancers require neo-angiogenesis for their progressive growth and metastasis, cancer cells being able stimulate nearby normal cells to produce angiogenesis signaling molecules[5,6,7]. The angiogenesis in solid tumors is a well-known fact and targeting the pathways that regulate angiogenesis was suggested as potential therapeutic approach[8,9]. After the initial paper by Folkman in 1971[8] that shifted the therapeutic paradigm from targeting the tumor cell to an anti-angiogenetic approach, a new field of study in oncology emerged. Over the years numerous discoveries have been made regarding the identification of angiogenetic factors, the regulation of the tumoral angiogenesis process and the development of new drugs that could interfere with pathological tumor-related vessel formation. After the establishment of VEGF as the key mediator in pathological angiogenesis[10] targeting the VEGF/VEGF-R signaling axis has become the pivot of research in the development of antiangiogenetic agents. Clinical trials (number ranging in the thousand) demonstrated the benefits of the anti-angiogenesis drugs. Numerous molecules have been developed and have received regulatory approval for cancer therapy[11,12,13,14] and secondary for the treatment of ocular neovascular diseases which share molecular pathways with cancer neo-angiogenesis[13,15,16,17,18,19,20]. These therapies are credited with improved progression-free survival (PFS) and over-all survival (OS), but those results are inconstant. Due to the high costs, but also the frequency and severity of therapy-specific adverse effects, in low-income countries there is a tendency for careful consideration of efficacy-safety ratio. In Romania, these therapies are fully refunded through national health care programs, and as a result, in the Oncology Institute of Bucharest, around 50 patients with ovarian cancers are treated each year with anti-angiogenesis drugs. Although we observed improved survival, some patients experience severe adverse reactions and extreme alterations of QoL leading us to the conclusion that more in depth analysis of security profiles are warranted. In Romania, we see no reduction of incidence for ovarian cancer[21], thus the therapeutic options for these cases continues to be of concern.

The aim of this study is to provide a comprehensive review of efficacy-safety ratio of angiogenesis-inhibitors based therapies in ovarian cancer, thus providing a better evaluation and selection tool for clinicians.

2. Materials and Methods

2.1. Search Strategy

A systematic literature search, according to PRISMA guidelines[22], was performed in multiple international databases - PubMed, Web of Science, Embase, Cochrane Library and ASCO and ESMO abstract database, from establishment up to May 1st, 2022. The following combinations of search terms was used: (1) (ovar or ovarian or ovary) and (neoplasm or cancer or carcinoma or malignant or tumor or tumor); (2) vascular endothelial growth factor, angiogenesis inhibitor, VEGF, VEGFR, VEGF-R, anti-VEGF, VEGF-target, anti-angiogenic, anti-angiogenesis, antiangiogenetic; (3) bevacizumab, avastin, cediranib, AZD2171, recentin, aflibercept, VEGF trap, AVE0005, zaltrap, sorafenib, nexavar, pazopanib, Votrient, trebananib, AMG386, nintedanib, BIBF 1120, tyrosine kinases inhibitors, TIE, AXL, FLT, sunitinib, SU11248.

2.2. Study Selection

References were prepared by using Mendeley Reference Manager [23] After removing duplicate records an extensive title and abstract screening process was performed and records with irrelevant focus were removed.

Inclusion criteria were adult patients with confirmed ovarian cancer regardless of histology, regardless of treatment settings (first-line, recurrence or maintenance), receiving anti-angiogenetic drugs. The experimental arm needed to be compared to standard chemotherapy or placebo, with PFS and/or OS being reported by hazard ratio with 95%CI. PFS, OS and adverse effects were our outcomes of interest, but we did not limit the inclusion if a study didn’t report all three. The study design of trials included were Phase II/III randomized controlled. Non-randomized controlled studies or Phase I trials were also excluded due to the fact that were not considered high-quality statistical data. Book chapters, reviews and case-reports were also excluded for the same reason. Studies using cyclophosphamide as control arm were excluded due to the fact that the superiority of carboplatin/paclitaxel over cyclophosphamide regimens was long-time established by prior trials[24].

Population, Intervention, Comparator, Outcomes and Study designs (PICOS) structure shown in Table 1 was used for considering studies for the review[25].

The remaining articles were obtained as full text and were independently reviewed by two investigators, and discrepancies were discussed between all authors. The literature search flow chart is presented in Figure 1.

2.3. Data Extraction

From each trial selected the following data was retained: name of first author, year of clinical trial, study phase, patient selection criteria, drug regimens on experimental and controlled arms of study, sample size, different bias factors, and outcomes intended for use in meta-analysis (progression-free survival, overall survival and adverse events – type, total number of events, number and type of grade 3 or higher adverse events).

2.4. Risk of Bias Assessment

Risk of bias was assessed on six domains used by the Cochrane Collaboration’s tool[26]: selection bias (random sequence generation and allocation concealment), performance bias, detection bias, attrition bias, reporting bias, and other sources of bias. Details of bias risk is presented in Figure 2a,b. Risk was categorized as low, high or unclear.

2.5. Statistical Analysis

Pooled hazard-ratios (HRs) for overall survival (OS) and progression-free survival (PFS) were calculated in RevMan 5.4.1 software[27]. Risk ratios (RR) for adverse events were also calculated with 95% CI. I² statistic was used in order to see the statistical heterogeneity of the studies. When a high heterogeneity between studies was observed we used a random model for statistical analysis. For I²<40% we used a fixed model.

3. Results

3.1. Literature Search

After the literature review, we identified n=9754 records concerning the utilization of antiangiogenetic therapies for ovarian cancer. Another 54 additional records were found after reference review. Duplicate records (n=2183) were removed and after title and abstract screening 7514 records were excluded: irrelevant focus (n=5127), non-randomized controlled/ Phase I studies (n=443), reviews or case reports (n=1908), other reasons (n=39). The remaining 119 records were obtained as full text articles and assessed. Another 95 studies were excluded due to lacking of outcomes of interest or being previous versions or exploratory outcomes of studies already included. In the end 25 randomized controlled trials (RCTs) were included.

3.2. Studies Characteristics

A total of 15487 patients met the inclusion criteria (adults with confirmed ovarian cancer, with therapies including anti-angiogenetic agents and compared to other drug regimens without such agents or placebo). The 25 RCTs have a publication date between 2011 and 2021, and they evaluated 7 inhibitors of angiogenesis (bevacizumab – 7, pazopanib – 5, trebananib – 4, nintedanib – 3, sorafenib – 2, cediranib – 2, aflibercept – 1). In the 2019 Tewari study there were 2 experimental arms which were both included in our study, the difference being in the maintenance therapy (1 arm using bevacizumab - BEVm, 1 arm using placebo - PLm). A similar situation was encountered in the Ledermann 2016 study, but we did not use separately the experimental arms due to the lack of hazard ratios for PFS and OS between the control and PLm arms of the study. The general characteristics, the references and the summary of outcomes from the included trial are listed in Table 2 and Table 3.

3.3. Progression-free Survival and Overall Survival

In our study the anti-angiogenetic regimens were credited with a significant improvement of PFS when compared to control arm (HR 0.72, 95% CI 0.66–0.78, I2 = 77%, P < 0.00001). The Z-test for overall effect was significant (Z=7,29; P < 0.00001). Improved OS was also observed, but the results had a lower P value (HR 0.95, 95% CI 0.91–0.99, I2 = 0, P=0.02). Improvement of progression was observed in all patients, regardless of treatment settings: first-line (HR 0.83, 95% CI 0.77-0.90, P<0.00001), recurrent cancer (HR 0.61, 95% CI 0.54-0.68, P<0.00001) or maintenance (HR 0.82, 95% CI 0.67-1.00, P=0.04). For the recurrent cancers, the PFS improvement affected both platinum-sensitive and platinum-resistant tumors (HR 0.58, 95% CI 0.49-0.69, P<0,00001, respectively HR 0.56, 95% CI 0.42-0.75, P<0,00001) - Figure 3. When analyzing the OS in different treatment setting, the results were more heterogenic – significant improvements were observed in recurrent cancers (both in P-S R and P-R R – HR 0.88, 95% CI 0.79-0.99, P=0,03, respectively HR 0.78, 95% CI 0.65-0.94, P=0.01). However, when employed as first-line or maintenance therapy, no improvement was observed – Figure 4.

Different action mechanisms of the antiangiogenetic drugs may have different effects. We tried to establish that by evaluating PFS and OS for 3 distinct action mechanisms – VEGF inhibitors (bevacizumab, aflibercept), VEGF-R inhibitors (pazopanib, cediranib, nintedanib, sorafenib) and angiopoietin inhibitors (trebananib). All three groups experienced improvements on PFS: VEGF inhibitors – HR 0.66, 95% CI 0.56–0.79, P < 0.00001; VEGF-R inhibitors HR 0.74, 95% CI 0.66–0.83, P < 0.0001; angiopoietin inhibitors HR 0.82, 95% CI 0.69–0.99, P = 0.04 – Figure 5. No significant differences in OS between subgroups and control groups were observed however – Figure 6.

In order to evaluate the impact of heterogeneity in PFS/OS results we performed a sensitivity analysis, by visual inspection of comparison-adjusted funnel plot of 25 included RCTs. We found no evidence of asymmetry in the funnel plot, indicating absence of publication bias.

3.4. Adverse Events

We analyzed the adverse effects reported for the included studies. For the Karlan and Tewari studies which included two experimental arms, the safety analyses were done by comparing the two experimental arms together with the control arm. Our analysis included 31 adverse events that are considered to be associated with anti-angiogenetic therapy for which RR were calculated (Supplementary Materials – Figures S1-S31). Hypertension was found to have a higher risk for the patients on angiogenesis inhibitors (RR 3.64, 95% CI 1.76-7.50, P=0,0005, I2=99%). Similar results were found for hemorrhagic events (RR 1.94, 95% CI 1.00-3.76, P=0.05, I2=77%) and for thromboembolic events (RR 1.47, 95% CI 1.12-1.92, P=0.006, I2=2%). However, only arterial thromboembolism was statistically linked to the anti-angiogenetic therapy (RR 2.13, 95% CI 1.41-3.23, P=0.0004, I2=17%). Same is not true for venous thromboembolism (RR 1.14, 95% CI 0.82-1.57, P=0.44, I2=43%). Significant increased risk was also seen for proteinuria (RR 4.52, 95% CI 2.12-9.62, P<0.00001, I2=87%), gastro-intestinal perforations (RR 2.86, 95% CI 1.36-6.04, P=0.006, I2=0%), infections (RR 1.28, 95% CI 1.07-1.53, P=0.008, I2=0%), ascites (RR 2.06, 95% CI 1.65-2.57, P<0.00001, I2=35%), neutropenia (RR 1.27, 95% CI 1.09-1.48, P=0.002, I2=92%), leucopenia (RR 1.21, 95% CI 1.05-1.39, P=0.009, I2=59%), thrombocytopenia (RR 1.98, 95% CI 1.56-2.52, P<0.00001, I2=86%), nausea (RR 1.27, 95% CI 1.12-1.45, P=0.0002, I2=79%), vomiting (RR 1.28, 95% CI 1.10-1.51, P=0.002, I2=57%), diarrhea (RR 2.05, 95% CI 1.60-2.62, P<0.00001, I2=91%), fatigue (RR 1.14, 95% CI 1.02-1.28, P=0.03, I2=83%), dyspnea (RR 1.20, 95% CI 1.05-1.37, P=0.008, I=0%), hypomagnesemia (RR 1.63, 95% CI 1.15-2.32, P=0.006, I2=46%), hypokalemia (RR 1.87, 95% CI 1.53-2.28, P<0.00001, I2=0%), headache (RR 1.63, 95% CI 1.36-1.95, P<0.00001, I2=58%), abdominal pain (RR 1.12, 95% CI 1.04-1.22, P=0.005, I2=11%). No significant correlations were seen between the following symptoms and the anti-angiogenetic therapy: reversible posterior leukoencephalopathy syndrome (RR 1.57, 95% CI 0.31-7.81, P=0.58, I2=0%), pyrexia (RR 0.88, 95% CI 0.73-1.08, P=0.22, I2=4%), wound related issues (RR 1.36, 95% CI 0.92-2.01, P=0.12, I2=12%), anemia (RR 0.94, 95% CI 0.82-1.07, P=0.33, I2=76%), anorexia (RR 1.54, 95% CI 0.96-2.49, P=0.07, I2=85%), constipation (RR 0.99, 95% CI 0.87-1.14, P=0.93, I2=45%), alopecia (RR 0.98, 95% CI 0.92-1.03, P=0.43, I2=11%), rash (RR 1.16, 95% CI 0.89-1.52, P=0.27, I=78%), back pain (RR 0.89, 95% CI 0.66-1.20, P=0.45, I2=60%), pain (RR 1.02, 95% CI 0.85-1.22, P=0.85, I2=66%). Grade 3 or higher adverse effects risk was increased in the therapeutic arms of our study (RR 1.43, 95% CI 1.20-1.72, P<0.00001) – Figure 7.

4. Discussion

4.1. Interpretation of Results

Our study demonstrated that anti-angiogenetic drugs can significantly improve PFS in ovarian cancer patients and increase the risk of common adverse events of all grades. The benefits regarding OS are more uncertain, being observed only in specific types of patients and tumors. We observed contradictory results for OS – pooled OS for all trials showed a significant improvement of survival favoring the angiogenesis inhibitors, but could not find a significant improvement when studying different treatment setting or types of anti-angiogenetic drugs separately. None of the therapeutic categories (VEGF blockade, VEGF-R inhibitors and angiopoietin inhibitors) were associated with improved OS. As first line therapy, we could prove no significant correlation with improved OS, possible reasons for this being: (1) most cases of ovarian cancer that benefit from first line chemotherapy are advanced (among the 6 trials included studying first-line therapy only one enrolled stage I-II cases[28]); (2) Three of the 25 RCTs [29,30,31] allowed patients in the control arm to receive anti-angiogenetic salvage therapy after disease progression tush diminishing the OS differential between experimental and control arms; (3) High-risk ovarian cancer is more likely to have a maximal response when treated with anti-angiogenesis agents as first-line.

The ICON7 trial[28]considered high-risk cancers as stage IV FIGO, inoperable cancer, or sub-optimally resected (> 1cm residual disease) stage III FIGO. The TRINOVA 3[53] GOG-0218[51], and the AGO-OVAR 12[49] shared the same definition for high-risk tumors. However, the OS between these trials were unconcording: ICON7 showed that bevacizumab in high-risk tumors is associated with improved OS (HR0.78, 95%CI 0.63–0.97, P=0.01)[28]. On the contrary, AGO-OVAR 12 showed that improved OS is obtain in standard chemotherapy rather than in nintedanib treated group (HR 1.14, 95%CI 0.89–1.45)[49]. GOG-0218 trial showed that advanced high-risk disease favored the concurrent use of bevacizumab with chemotherapy followed by bevacizumab maintenance arm (HR 0.72, 95% CI 0.53–0.97)[51]. The 2022 network meta-analysis by Helali et al. supported the use of bevacizumab, concurrently with chemotherapy, followed by bevacizumab maintenance until progression for chemo-naïve disease was associated with the highest probability of OS and PFS [54]. More over the same network meta-analysis ranked the effects of different anti-angiogenetic agents in high-risk chemo-naïve disease (Bevacizumab concurrent + maintenance > Nintedanib concurrent + maintenance > Tebananib concurrent + maintenance > Standard of care chemotherapy Carboplatin/ Paclitaxel). All mentioned trials showed no OS benefit favoring the usage of anti-angiogenetic agents in non-high-risk chemo-naïve disease. OS in the non-high-risk chemotherapy-naïve cancers can vary, thus making it difficult to propose any clinical recommendations for the usage of angiogenesis inhibitors in this disease setting, but most studies infer a probable lack of efficacy of angiogenesis inhibitors when used in the chemo-naive non-high-risk ovarian carcinoma[54].

In recurrent treatment settings, our study proved the benefits of anti-angiogenesis agents for both platinum-sensitive and platinum-resistant disease. For these categories both PFS and OS were improved, so angiogenesis inhibitors can be a treatment option for these patients. Helali et al.[54] ranked the angiogenesis inhibitors by probability of benefit in recurrent epithelial ovarian cancer. For platinum-resistant group they found that the concurrent chemotherapy and pazopanib has the best chance of improving OS, followed by combination chemotherapy with sorafenib. For platinum-sensitive group they found that the addition of anti-angiogenetic drugs to standard chemotherapy does not improve OS, but the association of bevacizumab or cediranib to chemotherapy in maintenance stage can improve PFS. Helali et al. also suggested that PARP inhibitors in addition to chemotherapy are the best option for platinum-sensitive recurrent disease[54]. Other studies suggested that PARP inhibitors in combination with antiangiogenetic agents may be a better therapeutic option than monotherapy with antiangiogenetic agents in platinum-sensitive cancers [55]

When analyzing the OS benefit of antiangiogenetic drugs used as maintenance therapy, we found no improvements.

The OS benefits are difficult to evaluate due to the fact that most trials included did not reported OS based on PFI (platinum free interval) and also the low sample size in some studies (MITO 11[47], TRIAS[33]) may influence results.

The results of our study suggest that PFS is not a viable surrogate evaluation of response for OS. The FDA approval process of various cancer drugs for solid tumors use surrogate endpoints (PFS) rather than clinical outcomes (OS). The definition of FDA for clinical outcomes is a direct measure of benefit of an intervention in a trial. A predictive substitute for clinical outcomes is called a surrogate endpoint. PFS is the most commonly used surrogate endpoint for OS for solid cancer[56,57], and its use is growing every day. The FDA has approved many cancer drugs based on surrogate endpoint data. Among those receiving regulatory approval, 57% of cancer drugs did not demonstrate an OS benefit[58]. Given the results of our study we consider that the usage of a surrogate endpoint for clinical outcomes (PFS) in the process of drug approval may need to be reconsidered, as PFS proved not to be a reliable substitute for OS. The suboptimal predictive value of PFS as a surrogate for OS was highlighted by several different articles and studies [54,59,60,61].

The absence of predictive biomarkers for anti-angiogenic agent therapy response makes it difficult to do a proper selection of patient categories that benefit most from anti-angiogenetic drugs. A series of markers have been suggested as response predictors: high Ang1/low Tie2 serum values was associated with improved PFS in experimental bevacizumab arm of ICON 7 patients[62]. The ICON 7 trial also validated the gene signature proposed by Gourley et al. in order to identify the angiogenic molecular subtypes of ovarian tumors[63]. In the serum samples from ICON 7 trial, Collinson et al.[64] identified several biomarkers predictive for response: fms-like tyrosine kinase-4 (FLT4), mesothelin, and α1-acid glycoprotein (AGP). This resulted in identifying a subgroup of patients likely to benefit from antiangiogenetic therapy, patients with a positive signature profile had a median PFS improved by 5,5 months (p=0.001). The GOG-218 trial was the source of a number of articles identifying several potential predictors for response to anti-angiogenetic therapy: plasmatic concentrations of VEGF and VEGF-R2 [65], tumor VEGF-A expression[66], micro-vessel density (MVD)[67], IL-6 levels[68]. Although the identification of these predictive biomarkers is encouraging, they need to be validated by more extensive trials in order to be incorporated into guideline recommendation for practicians. Oxidative stress related genes and factors have also been suggested to play an important role in ovarian cancer as a predictive factor for response to angiogenesis inhibitors therapy in ovarian cancer[69,70,71]. It is known that the OS affects chemoresistance through specific point mutations of key redox enzymes[72]. Oxidative stress factors have been identified in circulation among ovarian cancer patients[73]. However, the role of oxidative stress related prediction of response to therapy factors still needs to be investigated by trials.

Since for advanced ovarian cancer the treatment is mostly palliative in nature, and in the context of the myriad of potentially severe adverse effects and dubitable OS benefits, the question of quality of life (QoL) in patients treated with anti-angiogenetic agents should be an important indicator of therapeutic success. In present there are very few studies focusing on QoL in patients treated with angiogenesis inhibitors [74,75]. This aspect needs to be evaluated further by clinical trials. Our study demonstrated that some adverse effects that can significantly affect the QoL are more frequent when angiogenesis inhibitors are used. The anti-angiogenetic agents are associated with higher incidence of gastro-intestinal perforations, infections, thromboembolic and hemorrhagic events.

When debating the efficacity-toxicity ratio of angiogenesis inhibitors, one more concern is the cross results when employed in conjunction with other another class of targeted therapies like poly-(ADP-ribose) polymerase (PARP) inhibitors, especially since an ever-increasing number of trials have been published testing this therapeutic combination. This combination of drugs has a reported synergic effect, since angiogenesis inhibitors lead to hypoxia, which has been shown to induce homologous recombination repair deficiency (HRD) through the downregulation of homologous recombination repair genes. This effect result in an increased sensitivity of tumor cells to PARP inhibitors[76,77,78]. Liu et al.[43] showed a significant improvement of PFS when treating the recurrent platinum-sensitive ovarian cancer with cediranib combined with olaparib, when compared to olaparib monotherapy. The combination also increases the OS in patients without germline BRCA1/2 mutation. There is also data that suggested an improvement of PFS after the addition of olaparib in the maintenance phase in patients with advanced ovarian cancer which have received standard first-line therapy including bevacizumab – Phase II study conducted by Ray-Coquard in 2019[55]. This combination of angiogenesis inhibitors with PARP inhibitors can represent a new direction for future research. Another small Phase I study (Lorusso et al. from 2020) focused on the association between bevacizumab and a different PARP inhibitor (rucaparib) and found no safety concerns about the combination[79]. The randomized Phase II study MITO 25 will investigate further maintenance rucaparib with/without bevacizumab in patients with newly diagnosed stage III-IV ovarian cancer[80]. Mirza et al.[81] published a Phase I study on the combination of niraparib and bevacizumab in platinum-sensitive epithelial ovarian cancer which showed no safety concerns.

Anti-angiogenetic therapies, although constantly improving PFS, show suboptimal clinical effect[82,83]. This may be explained by the triggering of selective survival of hypoxic cells in the center of tumoral mass[7]. Moreover, the disruption of a given angiogenic pathway may provoke compensatory reactions through compensatory secretion of alternative angiogenic factors[84,85,86,87,88,89,90], leading to resistance to single-target therapeutic approaches. We, therefore see an unmet need for novel strategies in order to compensate for the shortcomings of current therapeutic modalities by employing concurrent therapies that target multiple angiogenetic pathways.

4.2. Comparison with Other Studies

In our study consists of 25 RCTs thus being the largest and more recent meta-analysis and systematic review done on the subject of anti-angiogenetic drugs and their usage in ovarian cancer, including 15487 patients. Over time a number of systematic review and meta-analysis demonstrating the outcome variabilities were performed on angiogenesis inhibitor randomized controlled trials[54,91,92,93,94,95,96,97,98,99,100,101]. One of the latest (Guo)[92] included 22 RCTs published between 2011 and 2019, with 11254 patients meeting the inclusion criteria. Our study included 3 other studies (Burger, du Bois 2014, du Bois 2016). Helali et al.[54] although included 23 RCTs focused on epithelial ovarian cancer omitting studies like Liu and Lederman 2016[42,43] included in our study, in which the sample included serous or endometrioid histology. They included the Matulonis 2019 (KEYNOTE 100) study[102] on the effects of pembrolizumab (a PD-1 inhibitor) in ovarian cancer, which we consider outside of the scope of this paper. The Hall et al. 2020[103] study was not included due to the fact that cyclophosphamide was used as treatment line rather than carboplatin/paclitaxel - the superiority of carboplatin/paclitaxel of over cisplatin/cyclophosphamide has been demonstrated by previous trial[24] and the usage of cyclophosphamide may falsely alter the PFS results in favor of anti-angiogenetic drugs.

Angiogenesis inhibitors are associated with an increased PFS over all patient groups and all categories of drugs used, results similar to previous studies. In our study the OS was only improved by the studied therapy for recurrent ovarian cancer, not influencing cases in first line or maintenance settings. The improved OS observed in other studies when using VEGF inhibitors or angiopoietin inhibitors was not observed in our study.

Most of previous meta-analysis regarding angiogenesis inhibitors in ovarian cancer patients, focused on survival parameters, toxicity being ignored. For a better understanding of the efficacy-safety ratio of angiogenesis inhibitors, we also analyzed the therapy-specific adverse effects. By calculating the RR for 31 therapy-specific we observed increased risk for hypertension, hemorrhagic and thromboembolic events, proteinuria, gastro-intestinal perforations, infections, ascites, neutropenia, leucopenia, thrombocytopenia, nausea, vomiting, diarrhea, fatigue, dyspnea, hypomagnesemia, hypokalemia, headache, abdominal pain. For all grade ≥ 3 adverse events, the risk was higher in angiogenesis inhibitors groups. There were no significant increased risks found in the following adverse events: venous thromboembolism, anemia, pyrexia, anorexia, rash, pain, back pain. Two of the adverse effects specific to the angiogenesis inhibitors (wound related issues and reversible posterior leukoencephalopathy syndrome were found not to be significantly associated with the therapy in our study.

4.3. Strengths and Limitations

The strengths of this study are the broad inclusion of 15487 patients treated with seven different angiogenesis inhibitors which allowed us to perform a comprehensive analysis of survival parameters (PFS, OS) and therapy-specific toxicity. To the best of our knowledge this is the largest meta-analysis on the usage of anti-angiogenetic drugs in ovarian cancer to this date. The RCTs included enrolled all stages of ovarian cancer and different treatment settings, thus allowing us to evaluate the role of the anti-angiogenetic therapy as first-line, as maintenance or as secondary therapy for recurrent disease. Furthermore, the subgroup analyses of different mechanism anti-angiogenetic drugs, may allow to make prediction on the type of anti-angiogenetic therapy more likely to obtain the maximum benefit. As a side effect of the broad inclusion criteria regardless of tumor type, stage of disease, drug regimens, number of previous lines of therapy, response to previous therapy, duration of follow-up and so on, the heterogeneity was increased, giving raise to the possibility of different results when focusing on more narrow patient types or more specific drug regimens (our study included VEGF- blockade, VEGF-R inhibitors and angiopoietin inhibitors). The high heterogeneity across studies is the most important limitation of our analysis. Another limitation of our study is the fact that our meta-analysis is based on published data, rather than on actual patient data. Also, some of the studies included had an increased risk of bias due to issues with randomization or blinding.

5. Conclusions

Our study showed that anti-angiogenetic agents can improve the PFS in ovarian cancer regardless of treatment setting or type of drug used (VEGF inhibitors, VEGF-R inhibitors or angiopoietin inhibitors), thus being an option in treatment of ovarian cancer. Second, due to the fact that OS improvements are only seen in high-risk chemo-naive cancers and in recurrent disease, we believe that the usage of angiogenesis inhibitors needs to be administrated with prudence outside of these setting and after careful consideration of each case prognostic and response factors. Thirdly, we consider that further studies are required in order to better understand which categories benefit the most from angiogenesis inhibitors, and which are only subjected to the risk of adverse effects and diminished QoL. The need for identifying and validating biomarkers accurately predicting the response to therapy is a good future direction for study. Fourthly, for the platinum-sensitive recurrent disease the role of concurrent therapy with angiogenetic and PARP inhibitors needs to be explored further. And last: we consider that the usage of a surrogate endpoint for clinical outcomes (PFS) in the process of drug approval may need to be reconsidered as PFS proved not to be a viable substitute for OS.