Comparison of perioperative complications and health‐related quality of life between robot‐assisted and open radical cystectomy: A systematic review and meta‐analysis

Abstract To compare postoperative complications and health‐related quality of life of patients undergoing robot‐assisted radical cystectomy with those of patients undergoing open radical cystectomy. A systematic search was carried out according to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses statement. A pooled meta‐analysis was carried out to assess the differences between robot‐assisted radical cystectomy and open radical cystectomy according to randomized and non‐randomized comparative studies, respectively. We identified six randomized comparative studies and 31 non‐randomized comparative studies. Most robot‐assisted radical cystectomy patients were treated with extracorporeal urinary diversion. Robot‐assisted radical cystectomy was associated with longer operative times, and lower blood loss and transfusion rates compared with open radical cystectomy in both randomized comparative studies and non‐randomized comparative studies. There was no significant difference between robot‐assisted radical cystectomy and open radical cystectomy in the rate of patients with any or major complications within 90 days both in randomized comparative studies and non‐randomized comparative studies. Non‐randomized comparative studies reported a lower rate of complications at 30 days, mortality at 90 days and length of stay for patients treated with robot‐assisted radical cystectomy, which were not confirmed in randomized comparative studies. Additionally, there were no differences in postoperative quality of life score assessment at 3 and 6 months between robot‐assisted radical cystectomy and open radical cystectomy. Robot‐assisted radical cystectomy is associated with less blood loss and lower transfusion rates. There is no difference in complications, length of stay, mortality, and quality of life between robot‐assisted radical cystectomy and open radical cystectomy. Data from non‐randomized comparative studies favor perioperative outcomes in robot‐assisted radical cystectomy patients, the failure to confirm in randomized comparative studies, likely due to bias in study design and reporting. Further randomized comparative studies comparing postoperative complications and quality of life between robot‐assisted radical cystectomy with intracorporeal urinary diversion and open radical cystectomy are required to assess potential differences between these two surgical approaches.


Introduction
ORC with pelvic lymph node dissection and urinary diversion is the standard treatment for patients with muscle-invasive bladder cancer and those with very high-risk non-muscle-invasive bladder cancer. 1 ORC continues to be associated with a high rate of mortality and Reporting Items for Systematic Reviews and Meta-Analyses QLQ-C30 = Quality of Life Questionnaire Core 30 QOL = quality of life RARC = robot-assisted radical cystectomy RC = radical cystectomy RCTs = randomized morbidity, such as perioperative UTIs, thrombosis and ileus, which cause prolonged hospital stay and affect the patients' health-related QOL. Indeed, >60% of patients treated with ORC experience at least one perioperative complication, and approximately 20% experience a highgrade complication within 90 days of the surgery. 2 Although progress has resulted from changes in perioperative management, such as the ERAS, a proposed strategy to reduce perioperative complications has been to minimize the invasiveness of the surgical procedures, such as through the performance of a RARC.
Initial non-randomized retrospective studies showed significant advantages to RARC over ORC, such as lower estimated blood loss, lower blood transfusion rates, shorter length of stay and lower rate of postoperative complications. 3 Conversely, early randomized controlled trials failed to find any differences in complications at 90 days between RARC and ORC. 4,5 As the last meta-analysis on this subject was carried out in 2017, 6 a significant body of novel data including both randomized (RCTs) and NRCTs have emerged, adding new evidence to the topic. 2,5,7,8 To assess the differential perioperative complications and health-related QOL outcomes between RARC and ORC, we carried out an up-to-date systematic review and meta-analysis of the literature comparing complications rates, as well as health-related QOL outcomes of patients treated with RARC, with those of patients treated with ORC. We analyzed the data from the RCTs and the NRCTs separately to unmask potential bias arising from study design.

Methods
The protocol has been registered in the International Prospective Register of Systematic Reviews database (PROSPERO: CRD42018108001). The PRISMA checklist is reported in Table S1.

Literature search and inclusion/exclusion criteria
The present systematic review and meta-analysis were carried out according to the PRISMA statement 9 and the Cochrane Handbook for Systematic Reviews of Interventions. 10 A comprehensive literature search using the electronic databases (MEDLINE, Web of Science, Scopus and Cochrane Library) was carried out on 10 August 2018 to retrieve the articles published comparing postoperative complications and health-related QOL of patients treated with RARC with that of those treated with ORC. All full text papers were assessed and excluded with reasons when deemed inappropriate after screening based on the study title and abstract. Two reviewers carried out this process independently. Disagreements were resolved by a third party. The following string terms were used: (bladder cancer) AND (postoperative complication OR intraoperative complication OR perioperative complication OR postoperative morbidity OR postoperative mortality OR quality of life OR length of stay) AND (robot-assisted radical cystectomy OR da Vinci radical cystectomy OR robotic radical cystectomy).
Studies were included if they compared RARC with ORC and reported at least one postoperative complication outcome or health-related QOL assessment between both arms in RCTs or NRCTs that included retrospective observational or cohort studies. If there were multiple articles written by the same group based on a similar patient cohort, only the largest or most recently published study was included. Review articles, editorials, comments and meeting abstracts were excluded. References of included manuscripts were scanned for additional studies of interest.

Data extraction
Two authors independently extracted the required data. The baseline patient characteristics, postoperative complication rate, and health-related QOL assessment data of both RARC and ORC arms in all eligible studies were collected. The outcomes of interest in this meta-analysis were postoperative complication rates within 30 and 90 days, types of complications, and other perioperative parameters, such as operative time, estimated blood loss, blood transfusion rate and length of hospital stay, as well as health-related QOL scores. All discrepancies regarding data extraction were resolved by consensus or finally decided by Delphi consensus with other authors. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Statistical analysis
The WMD and RR with 95% CIs were used as the summary variables for continuous and dichotomous outcomes, respectively. If studies only reported continuous data as the median and range or interquartile range, the mean and SDs were calculated according to Hozo et al. 11 Statistical heterogeneity among studies was calculated using the I 2 statistics. The v 2test and I 2 statistics with significance set at P <0.10 and I 2 <50%, respectively, were used to assess statistical heterogeneity among studies. If there was a lack of heterogeneity, fixed effects models were used for meta-analysis. Random effects models were used in cases of heterogeneity. Statistical analyses were carried out using Review Manager (RevMan-Computer program, Version 5.3 Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014).

Risk of bias
For RCTs, we used the Cochrane Risk of Bias Tool for RCTs. 10 For NRCTs, we used the ROBINS-I tool. 12 Two authors independently assessed the risk of bias in each study. All discrepancies regarding the risk of bias were resolved by consensus between two authors.

Study selection and characteristics
Overall, 525 articles were identified for an initial assessment (Fig. 1); 187 duplicates were removed. Then, 230 and 71 articles were excluded after title and abstract assessment, and full text reading, respectively. Finally, 37 and 34 studies that reported postoperative complications and health-related QOL outcomes between RARC and ORC were included for qualitative and quantitative analyses, respectively. The general characteristics of the eligible studies are summarized in  Table 4. Just three RCTs and three NRCTs reported comparisons of health-related QOL changes before and after RARC and ORC. It was not possible to carry out a meta-analysis of the health-related QOL outcomes due to the heterogeneity of health-related QOL assessment tools between included studies. The risk of bias tables in RCTs and NRCTs are shown in Figure S1 and Table S2, respectively.

Comparison of mortality within 90 days between RARC and ORC
In RCTs, there was no significant difference between RARC and ORC in mortality rates within 90 days (RR 0.82, 95% CI 0.24-2.81, P = 0.75). Conversely, in NRCTs, RARC was  Fig. S2). The v 2 -test and I 2 -test did not show any heterogeneity in all pooled analyses (Fig. S2).

Subgroup analyses of types of complications between RARC and ORC
In RCTs, RARC was associated with higher overall UTI rates (RR 1.46, 95% CI 1.05-2.03, P = 0.03) compared with ORC without heterogeneity (Fig. S3a). In NRCTs, ORC was associated with higher overall thrombolic event rates (RR 0.42, 95% CI 0.21-0.82, P = 0.01) compared with RARC without heterogeneity (Fig. S3b). Figure S3c showed   both arms had significantly improved after surgery compared with those at baseline (RARC/ORC 120/121). Bochner et al. compared health-related QOL at baseline with those at 3 and 6 months postoperatively in RARC and ORC using QLQ-C30. 4 They showed that there were no statistical differences in QLQ-C30 change from baseline to 3 and 6 months after surgery in any of the evaluated domains.
In NRCTs (  25 They reported that the BCI and BIS scores were not statistically different between the RARC and ORC arms at any time point for urinary function (P = 0.11), bowel function (P = 0.58) and body image (P = 0.93), but the sexual function at all time points was better for patients treated with ORC (P = 0.047). Conversely, Li et al. used the BCI and CARE tools to compare health-related QOL between RARC and ORC in 324 patients at baseline, 1, 3, 6 and 12 months after surgery. 34 They found no differences in any of the BCI domains (urinary, sexual and bowel domains) at any time point between both procedures.

Discussion
Bladder cancer carries, due to the advanced age and high smoking rate of most patients, significant cumulative morbidity. RC, the standard of care for patients with muscle-invasive and very high-risk bladder cancer, is associated with a high rate of complications due to the complexity of the surgery itself and the inherent frailty of the patients. 36 Minimal invasive surgeries, such as RARC, promise to reduce postoperative complications and maintain patients' health-related QOL despite its higher cost and need for a learning curve. 46 While we are awaiting more robust data on the oncological equivalence of RARC to ORC, its benefits regarding postoperative complication rates and health-related QOL are intensively debated. In a meta-analysis published in 2015, Novara et al. reported that the rates for any and major (≥3) Clavien grade of complications within 90 days were slightly in favor of RARC. 47 In this updated systematic review and meta-  analysis, our aim was to examine the most up-to-date data on the differential impact of these two procedures on perioperative complications, mortality and health-related QOL, with a focus on the difference between the results of RCTs and NRCTs. Although NRCTs are known to potentially lead to a more biased evaluation and reporting of outcomes compared with RCTs, they are often the basis of decision-making and guideline recommendations, especially in surgical specialties.
In this systematic review and meta-analysis, we analyzed six RCTs and 31 NRCTs comprising 581 and 48 392 patients, respectively, treated with RARC or ORC. Additionally, we carried out meta-analysis of RCTs and NRCTs separately to ensure bias-sensitive analyses. We found that compared with ORC, RARC was consistently associated with lower estimated blood loss, lower transfusion rates and longer operative times in both RCTs and NRCTs. For example, the largest RCT 2 and the largest NRCT 22 both reported that RARC had significantly lower estimated blood loss, lower transfusion rates and longer operative times compared with ORC. This is consistent with the benefit of robotic surgery for other disease entities, such as robotic radical prostatectomy. 48 Another potential benefit of robotic surgery has been the shortened length of stay. However, we did not find any difference in the length of stay between RARC and ORC in RCTs, whereas NRCTs did report a shorter length of stay. This might be due to changes in postoperative pathways and the learning curve with a benefit to more modern pathways, such as ERAS, which is known to shorten the length of stay. 8 Indeed, in many NRCTs, RARC was carried out in a more recent time period than ORC, leading to an experience and practice pattern change bias favoring the more recent technique.
We found that there was no difference in postoperative complications and mortality between RARC and ORC. Interestingly, this was true for complications within 30 and 90 days, and mortality within 90 days in RCTs. However, in NRCTs, the mortality rate within 90 days and the complication rate within 30 days were worse for ORC. This reporting deviation from NRCTs could be due to a selection, reporting and/or detection bias in favor of RARC. For example, several studies included ORC patients with higher comorbidities,  higher BMI and/or more advanced disease compared with those treated with RARC, which led to a selection bias benefiting RARC. 7,16,18,19,26,33,37 Furthermore, when comparing novel interventions with conventional ones, there can be a novelty bias in NRCTs, which is a form of selection bias. 49 There was no significant difference between RARC and ORC in the overall ileus rate in both RCTs and NRCTs. Additionally, postoperative health-related QOL was not significantly different between RARC and ORC in both RCTs and NRCTs.
There are several limitations to be considered in this systematic review and meta-analyses. There were just six RCTs included with a low number of patients, especially compared with NRCTs. However, all NRCTs were observational and retrospective, which can introduce selection bias. The operative time in RARC was reported including setting and console time in almost all eligible studies, therefore, we were not able to compare the exact operative time between console time and ORC. The definition of complications is slightly different among studies or not reported in several studies. The health-related QOL assessment tools were very different between eligible studies. A surgeon's experience and expertise, which are only partially controlled in RCTs, were very different between studies, which might introduce a confounder.
Another limitation is the failure to report and control for the type and approach to the urinary diversion. Due to the heterogeneity regarding urinary diversion among eligible studies, we were not able to compare based on the type of urinary diversions. Furthermore, most urinary diversions described in RCTs and NRCTs of this meta-analysis were carried out using ECUD, which might limit the benefits of RARC compared with ORC. Indeed, ICUD seems to have advantages in the postoperative complication rates and health-related QOL after RARC compared with ECUD. Recently, in a retrospective study, Tan et al. showed the beneficial impact of RARC with ICUD using enhanced recovery protocols resulting in improved postoperative complication rates compared with ORC. 8    prospective study to compare RARC with ICUD to RARC with ECUD or ORC.

Conclusions
RARC leads to less blood loss and lower transfusion rates compared with ORC; ORC is, in turn, associated with shorter operative times. Our systematic review and meta-analyses did not show any difference between RARC and ORC in postoperative complications, mortality, and health-related QOL in RCTs. NRCTs consistently reported better perioperative outcomes for RARC, such as shorter length of stay, mortality rate within 90 days and complication rate within 30 days, which were not confirmed in RCTs. These discrepancies could be due to bias in study design, measurement, selection and reporting. Further RCTs comparing postoperative complications and health-related QOL between RARC with ICUD and ORC are required to assess potential benefits of RARC with the varied forms of urinary diversion.

Supporting information
Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Figure S1. Risk of bias table for RCTs. Figure S2. Forest plot showing the comparison of mortality rates within 90 days between RARC and ORC. Figure S3. Forest plots showing the comparison of (a) rates for UTI, (b) rates for thromboembolisis, and (c) rates for ileus between RARC and ORC. Table S1. PRISMA 2009 checklist. Table S2. Risk of bias assessment for NRCTs (ROBINS-I).