Surgical Treatment for Endometrial Cancer, Hysterectomy Performed via Minimally Invasive Routes Compared with Open Surgery: A Systematic Review and Network Meta-Analysis

Simple Summary Keyhole surgery has replaced open surgery as the gold standard of care in the surgical treatment of cancer of the womb. Previous reviews comparing keyhole and open surgery exclusively analysed data from randomised control trials. We present a comprehensive review using randomised and non-randomised trials to compare keyhole surgery and open surgery. This review investigates benefits, complications and long-term outcomes in terms of survival after treatment of cancer of the womb, and it shows that keyhole surgery lessened blood loss and the length of hospital stay compared to open surgery. Among the keyhole methods, robotic surgery decreased some complications while rendering the return of cancer less likely. Abstract Background: Total hysterectomy with bilateral salpingo-oophorectomy via minimally invasive surgery (MIS) has emerged as the standard of care for early-stage endometrial cancer (EC). Prior systematic reviews and meta-analyses have focused on outcomes reported solely from randomised controlled trials (RCTs), overlooking valuable data from non-randomised studies. This inaugural systematic review and network meta-analysis comprehensively compares clinical and oncological outcomes between MIS and open surgery for early-stage EC, incorporating evidence from randomised and non-randomised studies. Methods: This study was prospectively registered on PROSPERO (CRD42020186959). All original research of any experimental design reporting clinical and oncological outcomes of surgical treatment for endometrial cancer was included. Study selection was restricted to English-language peer-reviewed journal articles published 1 January 1995–31 December 2021. A Bayesian network meta-analysis was conducted. Results: A total of 99 studies were included in the network meta-analysis, comprising 181,716 women and 14 outcomes. Compared with open surgery, laparoscopic and robotic-assisted surgery demonstrated reduced blood loss and length of hospital stay but increased operating time. Compared with laparoscopic surgery, robotic-assisted surgery was associated with a significant reduction in ileus (OR = 0.40, 95% CrI: 0.17–0.87) and total intra-operative complications (OR = 0.38, 95% CrI: 0.17–0.75) as well as a higher disease-free survival (OR = 2.45, 95% CrI: 1.04–6.34). Conclusions: For treating early endometrial cancer, minimal-access surgery via robotic-assisted or laparoscopic techniques appears safer and more efficacious than open surgery. Robotic-assisted surgery is associated with fewer complications and favourable oncological outcomes.


Introduction
Endometrial cancer (EC) is the sixth most common cancer in women and the 15th most common cancer overall.Its worldwide incidence is 8.7/10,000 (Age-standardized rate), with more than 417,000 new cases diagnosed in 2020 [1].EC risk increases with age, and the highest rates are reported in women aged 75-79 [2].A European Cancer Registry study demonstrated an overall survival of 76% in women diagnosed with EC in the years 2000-2007 [3].
EC typically presents early with postmenopausal bleeding but may also present with persistent intermenstrual and heavy vaginal bleeding associated with features of anovulation [4].Diagnosis is accomplished through a histological evaluation of an endometrial sample, and surgery is the first-line treatment.The extent of surgery depends on histopathological features such as type, grade and International Federation of Gynaecology and Obstetrics (FIGO) stage.For early-stage EC (FIGO stage 1-2), total hysterectomy with bilateral salpingo-oophorectomy (BSO) is the standard treatment [5], and minimally invasive surgery (MIS) has become the preferred mode.Sentinel node dissection and omental biopsy may be considered in high-risk disease.In stage 2 disease, total hysterectomy with BSO is adequate, but a radical hysterectomy may be required to achieve margin-free resection.Exclusive open surgery (OS) is advocated in advanced disease (FIGO stages 3 and 4), where primary debulking surgery can be considered if associated morbidity and quality of life are acceptable.Palliative surgery also has a role in symptomatic women with advanced EC [5].
The rate of EC is estimated to be increasing, and as the awareness of associated symptoms increases, most women are expected to present with early stages of cancer, for which surgical treatment is often curative.The surgical management of EC, therefore, is an important area of clinical care and research.Current guidelines recommend MIS as the preferred route for early-stage EC (FIGO Stage 1 and 2) based on evidence from randomised control trials (RCTs) demonstrating low post-operative morbidity with comparable oncological outcomes [5,6].These recommendations are associated with the widespread adoption of MIS in clinical practice.A recent Cochrane review has shown that laparoscopic surgery (LRS) is associated with similar overall survival and recurrence rates with reduced post-operative morbidity.Quality of life (QOL) was better in the LRS group for the first three years; however, after four years, QOL was similar in both groups [7].RCTs have also demonstrated robotic surgery (RS) to be non-inferior to either standard LRS or OS; however, the available evidence is limited on long-term outcomes of the RS approach [5,[8][9][10][11].
RCTs typically only report a handful of possible risks and outcomes.Further useful clinical information may be available from non-randomised studies.In this regard, there have not been sufficient attempts to capture all reported risks associated with the different surgical approaches available for hysterectomy indicated in early-stage EC, which includes evidence from non-randomised studies.Such data help patients with EC and their clinicians in shared decision making regarding surgical treatment while also informing healthcare providers to align their services.
With this background, our aim was to systematically collate the published evidence to determine the comparative surgical and oncological outcomes related to three different surgical treatment options for early stages of EC.Therefore, we systematically reviewed the published evidence from randomised and non-randomised studies reporting clinical and oncological outcomes of both MIS (laparoscopic or robotic) and OS in treating early-stage

Risk of Bias of Included Studies
The Newcastle Ottawa Quality Assessment Scale (NOS) was used to assess the risk of bias.The studies were evaluated using the following criteria: selection, comparability and exposure.A maximum of four stars was awarded for selection, two for comparability and three for outcomes, with a maximum of nine stars.NOS was used to assess the quality of both randomised and non-randomised studies.The studies were categorised into low risk if they scored 7-9 stars, moderate risk if they scored 5-6 stars and high risk if they scored 0-4 stars.Fifteen studies were in the high-risk group, and the majority (80) were in the low-risk group.

Synthesis of Results
The primary outcomes extracted from the studies include duration of operation, length of stay in hospital, intra-operative complications (e.g., blood loss), incidence of additional treatments (e.g., blood transfusion), post-operative complications (e.g., fever, infection, ileus), complications of uncertain timing (e.g., VTE), total complications, total intraoperative and post-operative complications and oncological outcomes of disease-free survival and recurrence.
The ten binary outcomes (blood transfusion, fever, infection, ileus, VTE, total intraoperative complications, total complications, total post-operative complication, recurrence and disease-free survival) were assessed and reported with OR and the corresponding 95% credibility interval (CrI) calculated from absolute numbers or percentages.The four continuous outcomes (blood loss, duration of operating time, length of hospital stay and total number of lymph nodes dissected) were assessed and reported with mean difference (MD) and their computed variances.For studies where only median values and ranges (or interquartile ranges) of continuous outcomes were reported, the results were transformed into means and variances [14].
The network meta-analysis (NMA), a random-effect model, was expanded with a Bayesian method that allowed the inclusion of direct and indirect comparisons of the surgical techniques used, allowing for a better understanding of the data [15][16][17].The simultaneous inference of the evidence, considering the three surgical interventions, was facilitated by a data structure that could be regarded as a k-comparison to synthesise the available evidence.In line with this, the research question: "What is the prevalence of complications associated with three surgical techniques used among EC patients?"was developed and answered via the following distinct aims.The prevalence of peri-operative and oncological complications associated with RS versus LRS versus OS, as well as their rating, assessment of performance and clinical effectiveness defined by the rate of complications associated with each surgical method, were investigated.The Markov Chain Monte Carlo (MCMC) simulation was applied to estimate the posterior distributions of the model parameters and generate the results.The convergence of the MCMC process was assessed by evaluating the trace plots, and the consistency assumption was checked by performing a node-split analysis, which evaluates every comparison of interest using a separate model [18].
For the interpretation of the Bayesian NMA results, Forest plots, Rankograms and surface under the cumulative ranking (SUCRA) plots were used [19,20].Based on an interim empirical evaluation of the last decade, a trend of moving away from OS to LRS and RS was observed within the pooled studies.As a result, to assess the possible time trend of the outcomes, a meta-regression of each identified outcome based on the time period of the study was performed.The study year of all publications included was utilised as the midpoint of the study duration using the following formula: Study year = (study start year + study end year)/2 The estimated regression coefficients of the study years were examined to report time trends based on current empirical evidence reported from practitioners.

Study Characteristics
The initial search yielded 74,322 references, from which 59,783 duplicates were removed, and the remaining 14,539 records were screened to select 874 relevant publications to assess the abstracts for eligibility.A total of 194 studies [9,13,, with 245,408 women, were included in the review (Table 1), and out of them, a total of 99 [9,23,24,26,30,32,36,39,42,43,47,49,51-54,56,57,59,63,67,68,70,71,76-79,81,82,84,85,87-90, 94,96,100,102,105,110-114,119,122-124,128,132,134-136,138,139,143,145,153,154,159,164,165, 167-173,177,178,180-184,186-188,190,192,194-196,198,200-202,208-210,212-219], comprising five RCTs and 94 cohort studies, were included in the NMA.The types of studies are detailed in the study characteristics and included 181,716 women.The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) diagram (Figure 1) illustrates the process of elimination.Detailed characteristics and a quality analysis of a subset of the studies included in the current systematic review and meta-analysis are listed in Table 1.The NMA utilised 14 outcomes, while those for other outcomes reported in fewer than ten studies were omitted (Table 2).Reviews and Meta-analyses) diagram (Figure 1) illustrates the process of elimination.Detailed characteristics and a quality analysis of a subset of the studies included in the current systematic review and meta-analysis are listed in Table 1.The NMA utilised 14 outcomes, while those for other outcomes reported in fewer than ten studies were omitted (Table 2).

Duration of Operating Time
There were statistically significant increases in the duration of operating time of 18.95 min (95% CrI: 7.68-30.20)with LRS and 29.00 min (95% CrI: 13.66-44.23)with RS compared with OS, respectively.This suggests that patients undergoing LRS or RS had a significantly longer duration of operation than those undergoing OS.The difference between RS and LRS is not statistically significant (MD = 10.05 min, 95% CrI: −5.60-25.48).

Duration of Operating Time
There were statistically significant increases in the duration of operating time of 18.95 min (95% CrI: 7.68-30.20)with LRS and 29.00 min (95% CrI: 13.66-44.23)with RS compared with OS, respectively.This suggests that patients undergoing LRS or RS had a significantly longer duration of operation than those undergoing OS.The difference between RS and LRS is not statistically significant (MD = 10.05 min, 95% CrI: −5.60-25.48).
When RS was compared with LRS, there were statistically significant differences in two binary post-operative outcomes: ileus (OR = 0.40, 95% CrI: 0.17-0.87)and total intra-operative complications (OR = 0.38, 95% CrI: 0.17-0.75).These results suggest that patients undergoing RS had a significantly lower incidence of ileus and total intra-operative complications than those undergoing LRS.The incidence of other binary outcomes was not significantly different between RS and LRS.

Length of Hospital Stay
Compared with OS, there was a statistically significant reduction in the length of hospital stay in women who underwent LRS and RS with mean differences of −3.54 days (95% CrI: −4.22-−2.87)and −3.79 days (95% CrI: −4.79-−2.79),respectively.This suggests that patients undergoing MIS had a significantly shorter length of stay in hospital than those undergoing OS.The difference between RS and LRS is not statistically significant (MD = −0.25 days, 95% CrI: −1.26-0.77).
3.4.Oncological Outcomes (Figures 2 and 3, Tables 2-4) There was a significant reduction in the binary outcomes of cancer recurrence (OR = 0.64, 0.47-0.84)with LRS compared to OS.The incidence of disease-free survival was not significantly different between LRS and OS.
When RS was compared with OS, there was a significantly higher disease-free survival (OR = 3.29, 95% CrI: 1.46-8.36)associated with this method, but recurrence was not significantly different between RS and OS.
Compared with LRS, RS was associated with significantly higher disease-free survival (OR = 2.45, 95% CrI: 1.04-6.34),but the other oncology outcomes appear to be similar between the two approaches.
3.5.Ranking Analysis (Figure 2 and Tables 2 and 3) Ranking analysis indicates that OS is the best technique when the duration of operation or total lymph nodes are considered, LRS is the best technique when incidences of infection, VTE and recurrence are considered, and RS is the best technique when blood loss, length of stay in hospital, disease-free survival and incidences of blood transfusion, fever, ileus, total complications, total intra-operative complications and total post-operative complications are considered.

Meta-Regression Analysis (Figure 4 and Table 4)
A meta-regression analysis was conducted for each outcome in line with the study timelines to study the possible time trend on the outcomes.
The study year did not significantly affect comparisons among the three surgical techniques on most outcomes.However, time trend was significant in four cases: comparison between LRS vs. OS on duration of operation (estimated regression coefficient −2.3596 (p = 0.0493)), comparison between RS vs. LRS on blood loss (estimated regression coefficient 15.8981 (p = 0.0003)), comparison between LRS vs. OS on blood transfusion (estimated regression coefficient −0.0915 (p = 0.0160)) and comparison between RS vs. LRS on total complications (estimated regression coefficient 0.1415 (p = 0.0357)) (Figure 4).These data suggest that some differences between techniques appear to reduce in magnitude with time.For example, the initial longer duration of operation between LRS and OS became smaller over time.Similarly, the difference in blood loss between RS and LRS also reduced over time.
Conversely, the difference in a lower incidence of blood transfusion between LRS and OS increased over time.Although the earlier studies reported a lower incidence of total complications in patients undergoing RS vs. LRS, the more recent studies reported contrastingly lower total complication rates with LRS than RS.The superiority of RS in disease-free survival, when compared with either OS or LRS, is of key interest.RS also demonstrated improved peri-operative outcomes, including a reduction in blood loss, length of stay in hospital, total intra-operative complications and incidences of blood transfusion, fever, ileus, total post-operative complications and total  The superiority of RS in disease-free survival, when compared with either OS or LRS, is of key interest.RS also demonstrated improved peri-operative outcomes, including a reduction in blood loss, length of stay in hospital, total intra-operative complications and incidences of blood transfusion, fever, ileus, total post-operative complications and total complications according to the ranking analysis.However, the recommendation of RS as the best option for hysterectomy for early EC should be made with caution since the meta-regression analysis demonstrated a possible increase in total complications associated with the RS route compared to LRSs.There has been an increasing number of surgeons learning and performing hysterectomies via the RS route in recent years as opposed to the highly experienced specialists as in early studies, and thus increasing numbers of RS cases may explain these findings.Nevertheless, future studies should focus on further examining this trend.

Laparoscopic Surgery
The LRS approach appears to be the best technique to consider when the reduction of incidences of infection, VTE, and recurrence is desired according to the ranking analysis.However, OS faired best when reducing the duration of operation or increasing the total number of lymph nodes to be harvested.This information is vital for service planning and future directions in the management of EC.The main disadvantage of LRS is that the duration of surgery appears to reduce with time, according to our meta-regression analysis, possibly due to increasing skills in LRS.

Comparison with Existing Literature
This study presents a comprehensive systematic review including 194 manuscripts and an NMA comprising 99 papers, demonstrating the significant superiority of MIS for early-stage EC, compared with OS in multiple aspects.Although the duration of surgery was slightly longer, MIS was associated with significantly lower rates of complications during and after surgery, in conjunction with a possible superiority in oncological outcomes, compared with those who underwent OS.MIS approaches were also associated with a reduced duration of hospital stay, which relates to reduced health service costs.Our findings are consistent with the Cochrane review [7] on LRS for EC and large RCTs such as LACE [21] and LAP2 [22].OS was associated with a higher incidence of post-operative complications such as Fever, infection and ileus compared with MIS.Previous Cochrane reviews and other RCTs [23] have shown comparable rates of recurrence and disease-free survival with the OS or MIS approach, yet our review has shown that MIS (LRS or RS) is associated with lower recurrence and better disease-free survival.The reasons for these observations are unclear but may be due to increasing expertise in MIS over recent years and possible selection bias, where high-risk ECs were preferentially treated by OS.However, with the recent advent of robotic surgery, high-risk patients, for example, with morbid obesity have been particularly assigned to undergo surgery via MIS route, and thus this data may be a true reflection of superior outcomes.

Strengths and Limitations
The main strength of this study is the inclusion of all relevant published data from both RCTs and cohort studies.By including data from cohort studies that were excluded by previous systematic reviews, our study represents, to our knowledge, the most comprehensive summary to date of peri-and post-operative and oncological outcomes associated with surgical treatment for early EC.This data thus could provide a solid foundation for developing core outcomes for hysterectomy for EC.The inclusion of observational data limits causal inferences and inevitably is subjected to carryover confounding bias.Nonstandardised outcome reporting limited the number of studies that could be included in the analysis.We did not formally investigate selection bias.
It is also important to note that trends in endometrial cancer staging surgery have changed over time (with improved pre-operative imaging and molecular subtyping influencing the extent of surgery [6]).In particular, the advent of sentinel lymph node biopsy has resulted in far fewer systematic pelvic and para-aortic lymph node dissections, thus reducing associated risks and morbidity.However, this practice varies considerably between institutions.Facilities for sentinel lymph node biopsy are not universally available, and not all institutions have the expertise to perform laparoscopic lymph node dissections.These variations in practice and techniques (e.g., new minimal access approaches including natural orifice transvaginal endoscopic surgery (NOTES), laparo-endoscopic single site (LESS) surgery and robotic single-site surgery) will introduce heterogeneity to the studies.Therefore, discussion regarding the risks of surgery should be tailored to the patient, surgeons and the cancer unit, considering the anticipated complexity of the planned operation and locally available expertise and resources when assessing the relevance of the outcomes reported here.The majority of studies included compared outcomes for different surgical approaches within the same institution(s), and we have included a large number of studies from a large geographical area.Thus, it is reasonable to assume heterogeneity within each surgical approach for the findings to be generalisable.

Conclusions
MIS, via either the robotic or the laparoscopic route, appears to be a safer and more efficacious approach when compared with OS for the treatment of early EC.The MIS approach is associated with fewer complications with favourable oncological outcomes.Time trends on outcomes, identified in our meta-regression analysis, provide vital information for policymakers and researchers to use in future-proofing health services and clinical trials.

Figure 1 .
Figure 1.The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) diagram.

Figure 1 .
Figure 1.The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) diagram.

Figure 2 .
Figure 2. Results of the node-split analysis checking consistency and assumptions.The effect sizes and 95% credible intervals from direct comparison, indirect comparison and the network combining the two are shown in Figure 2. The p-values in this context were used to test the consistency between direct and indirect comparisons.Figure 2 demonstrates that the consistency assumption is generally satisfied for 11 outcomes.The remaining three outcomes, fever, disease-free survival and totalintraoperative complications, were not shown due to insufficient data.(LRS: laparoscopic surgery, OS: open surgery, RS: robotic surgery).
and 95% credible intervals from direct comparison, indirect comparison and the network combining the two are shown in Figures 2. The p-values in this context were used to test the consistency between direct and indirect comparisons.
Figure 2 demonstrates that the consistency assumption is generally satisfied for 11 outcomes.The remaining three outcomes, fever, disease-free survival and total-intraoperative complications, were not shown due to insufficient data.(LRS: laparoscopic surgery, OS: open surgery, RS: robotic surgery).

Figure 3 .
Figure 3. Shows the forest plots of the 14 outcomes, providing the pooled estimates of the effect size of each surgery technique compared to open surgery (OS).(LRS: laparoscopic surgery, OS: open surgery, RS: robotic surgery).

Figure 3 .
Figure 3. Shows the forest plots of the 14 outcomes, providing the pooled estimates of the effect size of each surgery technique compared to open surgery (OS).(LRS: laparoscopic surgery, OS: open surgery, RS: robotic surgery).

Figure 4 .
Figure 4. Forest plots of the Bayesian network meta-analyses for each of the 14 outcomes.

Figure 4 .
Figure 4. Forest plots of the Bayesian network meta-analyses for each of the 14 outcomes.

Table 1 .
Characteristics of the 194 studies included in the qualitative synthesis.

Table 1 .
Characteristics of the 194studies included in the qualitative synthesis.r Year of Study Sample Size Type of Study Comparator

Table 2 .
League table showing pairwise comparisons among the three surgical techniques (LRS: laparoscopic surgery, OS: open surgery, RS: robotic surgery) for the 14 outcomes from the Bayesian network meta-analysis.Rows represent the references, and columns represent the comparators.95% credible intervals are included in parentheses.The clinically significant differences are shown in bold.

Table 3 .
SUCRA (surface under the cumulative ranking) scores of the three surgical techniques (LRS: laparoscopic surgery, OS: open surgery, RS: robotic surgery) for the 14 outcomes from the Bayesian network meta-analysis.The significant differences are shown in bold.

Table 4 .
Results of meta-regressions on study year for the 14 outcomes.The significant differences are demonstrated in bold.