Robotic radical hysterectomy is superior to laparoscopic radical hysterectomy and open radical hysterectomy in the treatment of cervical cancer

Objective Cervical cancer (CC) continues to be a global burden for women, with higher incidence and mortality rates reported annually. Many countries have witnessed a dramatic reduction in the prevalence of CC due to widely accessed robotic radical hysterectomy (RRH). This network meta-analysis aims to compare intraoperative and postoperative outcomes in way of RRH, laparoscopic radical hysterectomy (LTH) and open radical hysterectomy (ORH) in the treatment of early-stage CC. Methods A comprehensive search of PubMed, Cochrane Library and EMBASE databases was performed from inception to June 2016. Clinical controlled trials (CCTs) of above three hysterectomies in the treatment of early-stage CC were included in this study. Direct and indirect evidence were incorporated for calculating values of weighted mean difference (WMD) or odds ratio (OR), and drawing the surface under the cumulative ranking curve (SUCRA). Results Seventeen 17 CCTs were ultimately enrolled in this network meta-analysis. The network meta-analysis showed that patients treated by RRH and LRH had lower estimated blood loss compared to patients treated by ORH (WMD = -399.52, 95% CI = -600.64~-204.78; WMD = -277.86, 95%CI = -430.84 ~ -126.07, respectively). Patients treated by RRH and LRH had less hospital stay (days) than those by ORH (WMD = -3.49, 95% CI = -5.79~-1.24; WMD = -3.26, 95% CI = -5.04~-1.44, respectively). Compared with ORH, patients treated with RRH had lower postoperative complications (OR = 0.21, 95%CI = 0.08~0.65). Furthermore, the SUCRA value of three radical hysterectomies showed that patients receiving RRH illustrated better conditions on intraoperative blood loss, operation time, the number of resected lymph nodes, length of hospital stay and intraoperative and postoperative complications, while patients receiving ORH demonstrated relatively poorer conditions. Conclusion The results of this meta-analysis confirmed that early-stage CC patients treated by RRH were superior to patients treated by LRH and ORH in intraoperative blood loss, length of hospital stay and intraoperative and postoperative complications, and RRH might be regarded as a safe and effective therapeutic procedure for the management of CC.


Search strategy
We retrieved PubMed, Cochrane Library and EMBASE databases to obtain literature relevant to this study, and relevant articles were also reviewed manually in case of the omission of any potentially relevant literature. The literature search was limited to the English language and ended in June 2016. The search terms included a combination of key words and free words as follows: (1) cervical cancer, cervical carcinoma, cervical neoplasms, uterine cervical cancer, neoplasm, uterine cervical, cervix neoplasms, cancer of the cervix, cervical cancers, uterine, neoplasms and cervical; (2) surgery, surgical procedures, operative, operative surgical procedures, and operative procedures; (3) hysterectomy and radical hysterectomy; (4) randomized, randomized controlled trial, placebo, double-blind method, controlled clinical trial (CCT), and cohort study.

Inclusion and exclusion criteria
The inclusion criteria were as follows: (1) study design must be CCTs; (2) the interventions were RRH, LRH and ORH; (3) study subjects should be patients with early-stage CC aging from 15-85 years, body mass index (BMI, kg/m 2 ) ranging was from 15-45, the type of histological cell should be squamous or adenocarcinoma, and patients suffering from early-stage CC were at the FIGO Stage I and II; (4) the outcomes of studies included estimated blood loss (ml), hospital stay (days), intraoperative complications, number of pelvic lymph nodes removed, operative time (min) and postoperative complications. The exclusion criteria were as follows: (1) patients previously undergone radiotherapy, chemotherapy and neoadjuvant therapy; (2) patients with celiaca; (3) pregnant or lactating patients; (4) studies lacking complete literature data; (5) non-CCTs; (6) duplications; (7) conference reports, meta-analysis and summaries; (8) non-English references.

Data extraction and quality assessment
Two researchers independently carried out data extraction on the basis of a unified data collection form. Any dispute appearing during data extraction was resolved through discussion with multiple researchers. The quality of all included studies was assessed by researchers according to the Physiotherapy Evidence Database scale (PEDro) [15]. The total scores of PEDro were 11 points, score ! 4 points was regarded as high quality and score < 4 points was deemed as low quality [16]. The assessment consisted of a judgment of "yes," "no," or "unclear" for each domain to indicate a low, high, or unclear risk of bias, respectively. If one or no domain was deemed "unclear" or "no," the study was classified as having a low risk of bias. If four or more domains were deemed "unclear" or "no," the study was classified as having a high risk of bias. If two or three domains were deemed "unclear" or "no," the study was regarded as having a moderate risk of bias [17]. The Review Manager 5 (RevMan 5.2.3, Cochrane Collaboration, Oxford, UK) statistical computing software was used to carry out quality assessment and investigation of publication bias.

Statistical analysis
Firstly, traditional pairwise meta-analyses were performed for studies that compared different treatment arms directly. Our results reported the pooled estimates of odds ratios (ORs) or weighted mean difference (WMDs) and 95% confidence intervals (CIs). Heterogeneity among the studies was tested using the Chi-square test and I-square tests [18]. Random effect model was employed for the condition that the comparison results showed I 2 > 50% and P h < 0.05. Otherwise, fixed effect model was used for experiments. Secondly, R version 3.2.1 statistical computing software and network package were used to draw the network graphs, with each node representing different interventions, node size representing sample size, and the thickness of lines between the nodes indicating the number of included studies. Thirdly, Bayesian network meta-analyses were performed in order to compare different interventions with each other. Each analysis was performed based on the non-informative priors for effect sizes and precision. Convergence and lack of auto correlation were examined and confirmed after four chains and a 20,000-simulation burn-in phase; ultimately, direct probability statements were derived from an additional 50,000-simulation phase [19]. Furthermore, the node-splitting method was adopted in order to evaluate the consistency of the model, which separated evidence on a particular comparison into direct and indirect evidence [20]. To provide assistance in the interpretation of ORs, the surface under the cumulative ranking curve (SUCRA) was used in order to calculate the probability of each intervention being the most effective diagnostic method based on a Bayesian approach using probability values, and the larger the SUCRA value, the better the rank of the intervention [21,22]. Cluster analyses SUCRA values were conducted in order to group and rank the treatments according to their similarity with respect to two outcomes [21]. All computations were carried out by R (V.3.

Network evidence results suggesting more patients received RRH and LRH while less received ORH in the treatment of early-stage CC
The following three radical hysterectomies were included in this study: RRH, LRH and ORH. We found that a large number of patients underwent ORH and LRH, while the least number of patients underwent RRH (Fig 2).

The main results of network meta-analysis of intraoperative and postoperative outcomes of RRH, LRH and ORH in the treatment of earlystage CC
The network meta-analysis showed that patients treated by RRH and LRH had lower estimated blood loss than those treated by ORH (WMD = -399.52, 95% CI = -600.64~-204.78; WMD = 25% 50% 75% 100% 1. Eligibility criteria were specified 2. Subjects were randomly allocated to groups(in a crossover study,subjects were randomly allocated an order in which treatments were received) 3. Allocation was concealed 4. The groups were similar at baseline regarding the most important prognostic indicators 5. There was blinding of all subjects 6. There was blinding of all therapists who administered the therapy 7. There was blinding of all assessors who measured at least one key outcome 8. Measurements of at least one key outcome were obtained from more than 85% of the subjects initially allocated to groups 9. All subjects for whom outcome measurements were available received the treatment or control condition as allocated, or where this was not the case, date for at least one key outcome were analyzed by "intention to treat" 10. The results of between-group statistical comparisons are reported for at least one key outcome 11. The study provides both point measurements and measurements of variability for at least one key outcome  Table 2).

Inconsistency test of network meta-analysis of intraoperative and postoperative outcomes of RRH, LRH and ORH in the treatment of earlystage CC
The node-splitting method was used in order to test for inconsistencies for the six outcomes, and found that was consistent with the direct evidence and the indirect evidence so that we should use consistency model (all P > 0.05) ( Table 3).

RRH has the highest SUCRA values in estimated blood loss (ml), operative time (min), number of pelvic lymph nodes removed, intraoperative complications, hospital stay (days) and postoperative complications in the treatment of early-stage CC
As shown in Table 4, the SUCRA value of cumulative probability sorting of intraoperative and postoperative of three radical hysterectomies on early-stage CC showed that RRH had the highest   (min) and hospital stay (days), more number of pelvic lymph nodes removed and less intraoperative and postoperative complications among three radical hysterectomies.

Based on cluster analysis results, RRH had better intraoperative and postoperative clinical outcomes in the treatment of early-stage CC
Cluster analysis of SUCRA values based on estimated blood loss (ml), operative time (min), number of pelvic lymph nodes removed, intraoperative complications, hospital stay (days) and postoperative complications showed that patients treated by RRH showed better intraoperative and postoperative clinical outcomes in the treatment of early-stage CC, while patients treated by ORH had the worst conditions (Fig 4).

Assessment of publication bias of intraoperative and postoperative outcomes of RRH, LRH and ORH in the treatment of early-stage CC
The results of assessment of publication bias showed symmetrical distribution, indicating no small sample effect or publication bias in this network meta-analysis All the scattered points were of hypodispersion in the funnel, and red lines were symmetrical on both sides, indicating that the bias of reference applied in our study was small (Fig 5).

Discussion
In this study, we evaluated three different approaches for hysterectomies in the treatment of CC by summarizing clinical data in a pairwise meta-analysis. Our clinical data supports previous findings and suggests that RRH and LRH have better intraoperative and postoperative outcomes compared to ORH in the treatment of CC. Currently, an increasing number of gynecologists opt for the RRH approach in order to decrease postoperative morbidity [12]. Robot assisted surgery allows greater visualization of the instrument by means of binocular vision, using seven degrees of freedom of the instrument with greater flexibility, and the motion of the damping control is more accurate in 2005. [35]. LRH is performed routinely all around the world, due to the advances in minimally invasive surgery [12]. A recent meta-analysis showed that a comparison of RRH and PRH was not practicable as a result of insufficiency in studies that assessed appropriate "radical" hysterectomy merely for uterine CC [38]. Many clinicians believe that the RRH is associated with a lower incidence of postoperative morbidity compared to the traditional relative humidity, with similar clinical efficacy and safety [12]. The network meta-analysis showed that patients treated by RRH and LRH had lower estimated blood loss compared to patients undergoing ORH. Patients treated by RRH and LRH had shorter hospital stays than ORH. Compared with ORH, patients undergoing RRH treatment demonstrated lower postoperative complications. A previous meta-analysis showed that LRH and RRH were similar in terms of operating time, length of hospital stay, and number of pelvic lymph nodes resected, and RRH presented an overwhelming advantage and less blood loss against LRH with respect to complications [12]. Compared with ORH, RRH indicated lower blood loss and shorter length of hospital stays [35]. It is quite difficult to draw comprehensive conclusions from different studies about operative time, blood loss, and number of   lymph nodes, however, the overall consensus is that a minimally invasive technique seems to be the best laparoscopic radical hysterectomy to treat CC [33]. Soliman et al reported that RRH is associated with shortened hospital stay and reduced blood loss, nevertheless, the LRH and LRH all showed longer operation time than the laparotomy [27].
The results of the cluster analysis showed that the SUCRA value of RRH is higher than that of LRH and ORH as seen in   Radical hysterectomies for early-stage CC with the other two surgical groups, the robotic group showed postoperative parameters that reduced postoperative and 24-hour pain scores, shortened the length of hospital stay, and reduced the time to full diet resumption [27]. The results of the study proved that comparable surgical outcomes of patients receiving RRH of traditional laparoscopic approach in the treatment of early-stage CC, with lower intraoperative blood loss and early complication rates [24]. Chen CH et al reported that robotic surgery is verified to have a lower proficiency plateau and relatively shorter learning curve than traditional approaches [27]. Blood loss, rate of blood loss and length of hospital stay are similar for laparoscopy and robotics, and are significantly reduced as compared with laparotomy [39]. The data suggested that robotic surgery is a workable and potentially optimal option to treat CC with favorable short-term surgical outcomes [27]. However, significant differences for the number of RRH, LRH and ORH on the direct comparison of various interventions and the sample size of each intervention our present network meta-analysis had also limitation and advantage: (1) the sample size of each intervention, which might influence the overall results of the study; (2) in this research, our study showed the significant difference of RRH, LRH and ORH on hysterectomy in the treatment of earlystage CC; (3) due to lack of sufficient summarized studies to evaluate the long-term clinical outcomes between different treatment methods, we only focused on the comparisons of the short-term clinical outcomes using network meta-analysis.

Conclusion
In conclusion, these results of our meta-analysis indicate that patients with early-stage CC treated by RRH had better clinical outcomes of intraoperative blood loss, length of hospital stay and intraoperative and postoperative complications than LRH and ORH, which has a certain guiding significance for the clinical use and treatment of early-stage CC.