Intraperitoneal Perfusion Therapy of Endostar Combined with Platinum Chemotherapy for Malignant Serous Effusions: A Meta-analysis

Malignant serous effusions (MSE) are one of the common complications in patients with advanced cancer, with serious impact on systemic antitumor treatment and quality of life, suggesting a poor prognosis for patients. Currently there are no standard treatment patterns or reference guide for malignant serous effusions (Barni et al., 2001), making treatment a difficult task. Usually treatment strategy of MSE is to perform cavity puncture for drainage of the fluid and perfuse into the cavity with drugs to inhibit MSE generation, but the overall effect is poor. Many studies have confirmed that vascular endothelial growth factor (VEGF) and matrix metalloproteinase (MMPs) are two important factors that are involved in the formation of MSE. Compared with benign effusions, VEGF was significantly increased in MSE (Verheul et al., 2000; Zhou et al., 2009), in which MMPs can also result in ascites formation by promoting


Introduction
Malignant serous effusions (MSE) are one of the common complications in patients with advanced cancer, with serious impact on systemic antitumor treatment and quality of life, suggesting a poor prognosis for patients. Currently there are no standard treatment patterns or reference guide for malignant serous effusions (Barni et al., 2001), making treatment a difficult task. Usually treatment strategy of MSE is to perform cavity puncture for drainage of the fluid and perfuse into the cavity with drugs to inhibit MSE generation, but the overall effect is poor. Many studies have confirmed that vascular endothelial growth factor (VEGF) and matrix metalloproteinase (MMPs) are two important factors that are involved in the formation of MSE. Compared with benign effusions, VEGF was significantly increased in

Intraperitoneal Perfusion Therapy of Endostar Combined with Platinum Chemotherapy for Malignant Serous Effusions: A Meta-analysis
Rong Liang & , Hai-Ying Xie & , Yan Lin, Qian Li, Chun-Ling Yuan, Zhi-Hui Liu, Yong-Qiang Li* VEGF release (Becker et al., 2006;Tamsma, 2007), the research progress in above mechanism prompted anti-VEGF become a new therapeutic strategy for MSE. Study found that intraperitoneal perfusion of bevacizumab (A kind of anti-VEGF humanized monoclonal antibody) shows curative effect either as monotherapy or combined with other chemotherapeutic drugs (Kobold et al., 2009). However, because bevacizumab is an expensive imported targeted drug with difficult clinical application, it is necessary to seek more economical and effective anti-VEGF alternative drugs.
Endostar (Chemical name: recombinant human endostatin) is a new anti-tumor drug targeting to vessels developed by China. Endostar shows potent inhibition on neovascularization (Liu et al., 2015), and its mechanisms are associated with decreasing the expression of VEGFR-2, MMPs, TGF-β1, HIF-1α and bFGF (Ling et al., 2007;Lu et al., 2008;Wu et al., 2014). When combined with cisplatin, Endostar showed enhanced anti-cancer effect in esophageal cancer and lung cancer (Xu et al., 2014;Fan et al., 2015). Furthermore, Endostar inhibits ascites formation and prolongs survival in malignant ascites mouse models (Wei et al., 2015) and acquired 45%~100% effective rates in MSE patients when Endostar was admmonistrated through intraperitoneal perfusion combined with platinum (Yan et al., 2012). However, the related studies are single center RCTs with small sample size, and among each study, the treatment programs were confused and had no reference standard and consistent observation index. This study aims to perform a meta-analysis to evaluate effectiveness and safety of intraperitoneal perfusion of Endostar combined with platinum in malignant serous effusions.

Inclusion criteria of literature
The literatures selected for inclusion met the following criteria: published randomized controlled trial of Endostar combined with platinum intraperitoneal perfusion therapy in malignant serous effusions. The experimental group received Endostar plus platinum and control group received platinum alone. No chemotherapy, radiation or interventional therapy was performed simultaneously. The full text provided effective and safety indicators such as objective response rate [complete response (CR) + partial response (PR)], Karnofsky performance status (KPS) (behavioral state) score, adverse reactions and other outcome indicators.

Exclusion criteria of literature
Exclusion criteria were as follows (1) Review, case reports, animal experiments and basic research; (2) Subject information was not complete; (3) Repeated reports; (4) Non-random studies without control group.

Literature search
An electronic search was performed on scientific literature published in the databases of PubMed, EMBASE, web of science, Wanfang, CNKI, VIP and CBM. Meanwhile, manually search was performed on reference literatures of reviews. The search was performed using the following English retrieval words: Endostar; recombinant human endostatin; hydrothorax; ascites; pericardial effusion; pleural effusion; peritoneal effusion; serous effusion. The retrieval time was up to December 8, 2014.

Data extraction and quality evaluation
The quality of included literature was evaluated using Risk of bias evaluation criteria of Cochrane Handbook for systematic reviews of interventions (Version 5.3), and the detailed content was as follows: (1) Random allocation scheme; (2) Allocation concealment; (3) Blind method; (4) The data integrity; (5) There was no selective reporting of results; (6) Other sources of bias. The quality evaluation was carried out independently by two researchers, and disagreement between the reviewers was settled by discussion. The entries are divided into three grades: full, unclear and incomplete. There were 6 "full" in entries was classified as "A" grade, there were 1 or more "unclear" was classified as "B" grade, and there are 1 or more "incomplete" was classified as "C" grade.

Statistical analysis
Meta-analysis was performed using RevMan 5.3 software provided by Cochrane collaboration network. If homogeneity exist among studies then fixed-effect model was applied. If there is obvious inter-study heterogeneity, random-effect model was applied to analyze the sources of heterogeneity and to judge the publication bias through the funnel plot. The pooled effects of this study were expressed using RR and 95% confidence interval (CI). P<0.05 was considered the statistically significant difference.

Literature selection
A total of 535 literatures were identified by first screening. Firstly, through extensive reading the title and abstract, reviews, case reports, duplicated reports, animal experiments, and papers with the irrelevant subject or without conformity to the inclusion criteria were excluded, with 58 remaining literatures. After careful reading of the text, 33 literatures were further excluded, including non randomized trials, studies without control group, patients with systemic chemotherapy and intervention measures inconsistent with criteria. Eventually, 25 RCTs remained for meta-analysis, including 1523 patients, with 749 cases in the experimental group and 774 cases in the control group. The flow chart of literature screening process is shown in Figure 1, and the basic characteristics of the study are shown in Table 1.

Methodological quality assessment of the study
All studies referred to the random grouping, data integrity, the non-selective reporting. Among all 25 studies, 8 studies described using a random number table for grouping, 4 studies using random digit grouping method, 1 study by envelope method, and other 12 studies did not describe the detailed grouping methods. All studies did not describe the implementation of the blind method, and other sources of bias were not clear. The quality of all  the literatures was strictly evaluated as "B" grade, with the moderate risk of bias (Table 2).

Meta-analysis
Response rate of effusions: Meta-analysis was carried out using fixed-effect model and showed no statistical heterogeneity (P=0.96, I 2 =0%) between the 25 studies.
Life quality improvement rate: There were no statistical heterogeneity (P=0.8, I 2 =0%) among 18 studies by meta-analysis using fixed-effect model. The results showed that the difference in life quality improvement rate between the two groups was statistically significant   (Figure 3).
Rate of adverse effects: Meta-analysis using the fixed-effect model showed that 16 studies reported the occurrence of nausea and vomiting, with no statistically heterogeneity among results of different studies (P=0.9, I 2 =0%). The results showed that the difference in nausea and vomiting between the two groups was not statistically significant (35% vs. 34%, RR=1.01, 95%CI: 0.87-1.18, P=0.88) (Figure 4).
There were 4 studies which reported occurrence of renal damage, and the damage extent was I -II degree. There was no statistically heterogeneity among different studies. (I 2 =0%, P=0.92). Meta-analysis was performed using fixed-effect model, and showed that the difference in renal damage between the two groups was not statistically significant (18% vs. 20%, RR=0.86, 95%CI: 0.43-1.74, P=0.68) (Figure 6).   Subgroup analysis of response rate: Subgroup analysis on control rate of effusions showed that Endostar at dose of 45 mg/times (RR=1.76), treatment interval of 7 days (RR=1.67), treatment duration ≤2 weeks (RR=1.58), treatment duration >2 weeks (RR=1.61), combined cisplatin subgroups (RR=1.65) all had RR values close to or higher than the summery results (RR=1.63) (Table 3).

Publication bias
The funnel plot analysis was performed on the included studies with the efficiency as the index. The results showed that the scattered points were distributed on both sides of the line and were close to the top of the funnel. The distribution was almost symmetric (Figure 7). The results indicate that there is little possibility of publication bias.

Discussion
Since the clinical application of Endostar, combined treatment with Endostar and platinum has been widely used in malignant serous effusions, achieving satisfactory results. However, due to little sample size and inconsistency in research design, interventions measures and observation index, there are many inconclusive problems on Endostar combined with platinum MSE therapy about dose, treatment interval and period, curative effect and adverse reactions. This study performed comprehensive quantitative analysis to explore the value of Endostar in the treatment of MSE.
A total of 25 RCTs were included in this study, with 749 cases in the experimental group and 774 cases in the control group. Meta-analysis showed that Endostar combined with platinum group has higher response rate (76%) than platinum single drug group (47%) (1.63 fold, P < 0.00001) and higher quality of life improvement rate (69%) than platinum single drug grouop (44%) (1.57 fold, P<0.00001). This indicates that the effusions control efficiency and the patients' quality of life improvement was better than in experiment group than the control group. There was no significant difference between experiment group and control group in the incidence of nausea and vomiting (35% vs. 34%), white blood cell reduction (38% vs. 38%) and renal function damage (18% vs. 20%) (P>0.05). In a retrospective study of Ma Qian (Ma, 2014) including 43 cases of malignant hydrothorax and ascites patients, the response rate of bevacizumab combined with cisplatin group was 1.66 times of the cisplatin group (80% vs. 48%, P < 0.05). Therefore, combinations of Endostar or bevacizumab with cisplatin are superior to cisplatin monotherapy in the therapeutic efficacy of MSE, with similar efficiency between Endostar and bevacizumab. However, Endostar is more readily available and cheaper than bevacizumab in clinical practice, and is worth of clinical application.
There is no unified standard in Platinum drug type combined with Endostar, optimal dose, optimal treatment interval and duration of Endostar in the treatment of MSE. This study investigated these questions respectively. In included studies, combined Endostar and platinum group (Endostar dose: 30-60mg/time) can achieve 53-83% response rates in MSE, with highest RR value in 45 mg dose group (6 RCTs, RR=1.76), second highest RR value in 30 mg dose group (5 RCTs, RR=1.7), and third highest RR value in 60 mg dose group (8 RCTs, RR=1.62). The RR values of 30 mg dose group and 45 mg dose group were slightly higher than the aggregate results (1.63), but no obvious dose effect relationship was found among these data. This suggests that Endostar ar 45mg/time may be the most appropriate dose. A single large dose of Endostar is rare in clinical practice. In our results only 2 studies were included with single dose of Endostar more than 90 mg/time, with no significant increase in response rate (RR=1.51), so is not recommended routinely. Compared with the RR value of pooled results (RR=1.63), RR value with treatment interval of 3 days (3 RCT, RR=1.51) was significantly decreased and RR value with treatment interval of 1 week (19 RCT, RR=1.67) was significantly increased. Therefore Endostar treatment interval of 1 week is usually recommended in clinical practice. RR value in treatment time ≤2 weeks (6 RCTs, RR=1.58) was similar to that in treatment time >2 weeks (Most study was 4 weeks, 13 RCTs, RR=1.61). This indicates that treatment time length may have no significant effect on response rate, and 2-4 weeks is appropriate and is usually recommended, adjusted based on severity of illness, tolerance and the compliance of the individuals. The RR value of Endostar combined with cisplatin (22 RCTs, RR=1.65) was similar to that of summary results (25 RCTs, RR=1.63). The therapeutic effect of Endostar combined with second generation platinum (2 RCTs combined with nedaplatin, 1 RCT combined with carboplatin) does not seem to be superior to that of cisplatin, therefore in clinical, cisplatin treatment is preferred due to lower costs and adverse reactions similar to second generation platinum.
There are some limitations in this study: 1 The majority of the literatures only referred to the random words, and the specific random methods was not described, so the random method may not be sufficient. 2 All RCTs did not describe the allocation concealment and blind method, which may lead to the bias in intervention implementation or the outcome measure, thereby reducing the reliability of the results. 3 The inconsistency in treatment dosage and time of platinum and Endostar may affect the outcome. 4 Tumor location in included patients was not completely consistent, and the initial treatment and retreatment conditions of effusion patients was not clear, so stratified analysis can not be applied in the measured observation index. 5 Most of the trials did not perform follow-up, and were terminated when the therapeutic effects were observed, so it is lack of long-term efficacy such as duration of efficacy and overall survival data. We hope the future clinical research can further improve the quality of method, especially the implementation of random scheme and blind method, and also can optimize the above test methods and experiment design.