Robotic-assisted surgery versus open surgery in the treatment of rectal cancer: the current evidence

The aim of this meta-analysis was to comprehensively compare the safety and efficacy of robotic-assisted rectal cancer surgery (RRCS) and open rectal cancer surgery (ORCS). Electronic database (PubMed, EMBASE, Web of Knowledge, and the Cochrane Library) searches were conducted for all relevant studies that compared the short-term and long-term outcomes between RRCS and ORCS. Odds ratios (ORs), mean differences, and hazard ratios were calculated. Seven studies involving 1074 patients with rectal cancer were identified for this meta-analysis. Compared with ORCS, RRCS is associated with a lower estimated blood loss (mean difference [MD]: −139.98, 95% confidence interval [CI]: −159.11 to −120.86; P < 0.00001), shorter hospital stay length (MD: −2.10, 95% CI: −3.47 to −0.73; P = 0.003), lower intraoperative transfusion requirements (OR: 0.52, 95% CI: 0.28 to 0.99, P = 0.05), shorter time to flatus passage (MD: −0.97, 95% CI = −1.06 to −0.88, P < 0.00001), and shorter time to resume a normal diet (MD: −1.71.95% CI = −3.31 to −0.12, P = 0.04). There were no significant differences in surgery-related complications, oncologic clearance, disease-free survival, and overall survival between the two groups. However, RRCS was associated with a longer operative time. RRCS is safe and effective.

The study inclusion criteria were as follows: (1) the study was a comparative study that compared the safety and efficiency between robotic rectal cancer surgery (RRCS) and open rectal cancer surgery (ORCS); (2) the study included quantitative outcome data (e.g., operative time, length of hospital stay, complications, pathological parameters, and survival outcomes); (3) if the same institution and/or authors reported several studies, only the study with the greatest patient population or the highest quality study was included in the analysis; and (4) the study was published in English. We excluded editorials, comments, meeting abstracts, review articles, and non-relevant topic studies. Data extraction. The primary relevant data from all the included studies were extracted by two reviewers (GXL, ZZ). The extracted data included the following: the basic characteristics of the study, including first author, year, and country of publication; the publication journal name; the basic patient characteristics, including age, sex, body mass index (BMI), American Society of Anesthesiologists (ASA) score, case number, and tumor stage; short-term outcomes, including intraoperative data, postoperative data, and oncologic clearance; and long-term outcomes, including disease-free survival and overall survival. All available data were extracted using standard data extraction by one reviewer and were checked by another reviewer. Quality assessment. Two reviewers (GL, YL) independently evaluated the quality of each included study using the modified Newcastle-Ottawa scale (available at http://www.ohri.ca/programs/clinical_epidemiology/ oxford.asp), which is widely used for cohort study assessment. The quality assessment consisted of three major categories: patient selection, comparability of the RRCS and ORCS groups, and outcome assessment 18 . The details of this quality assessment are provided in Table S1. Any disagreement was resolved via discussion among the author group. Statistical analysis. This meta-analysis was performed using the Review Manager software (version 5.2, provided by the Cochrane Collaboration) and Stata software (version 11.0). Dichotomous variables were analyzed using odds ratios (ORs) with 95% confidence intervals (95% CIs). Survival outcomes were estimated using hazards ratios (HRs) and standard errors. Continuous variables were analyzed using mean differences (MDs) and 95% CIs. A fixed-effects model or a random-effects model was applied according to heterogeneity, which was evaluated by the I 2 measure of inconsistency. Heterogeneity was present when the I 2 statistic was greater than 50%, and a random-effects model was adopted. However, when the I 2 statistic value was less than 50%, a fixed-effects model was used. Publication bias was evaluated by a funnel plot of postoperative complications and Egg's and Begg's tests. Sensitivity analysis was performed by excluding the low-quality studies. A P value less than 0.05 was considered to be significant.

Results
Literature search. The PubMed, EMBASE, Web of Science, and Cochrane Library search identified a total of 277 studies. After excluding duplicates using Endnote software, 180 abstracts were carefully reviewed by two independent reviewers using a standard study selection form. After this process, the reviewers identified 10 studies for a comprehensive review. Three studies were excluded according to the inclusion criteria 19-21 , leaving 7 studies that were included in our analysis [22][23][24][25][26][27][28] . The study selection process is shown in Fig. 1. A total of 1074 patients with 498 (9.00%) cases of RRCS and 576 cases of ORCS were analyzed. Information on the studies and the participants is shown in Table 1. The quality assessment score of each included study is also provided in Table 1, and the details of each included study assessment are provided in Table S2.
Ileus. Ileus events were reported in four studies 22,26,27,28 . The incidence of ileus in the RRCS group was 3.81%, compared with 3.26% in the ORCS group. The results showed no significant difference between the two groups (OR = 1.11, 95% CI: 0.47-2.61, P = 0.80), and there was no heterogeneity among the studies (Fig. 3F).
Urinary retention. Three studies reported urinary retention 22,24,28 . The combined data indicated that urinary retention in the two groups was not significantly different (OR: 0.52, 95% CI: 0.10 to 2.77, P = 0.44), and there was no significant heterogeneity (Fig. 3H).
Length of hospital stay. All of the studies 22-28 described length of hospital stay (LOS). The pooled data of the included studies showed that the LOS was significantly reduced in the RRCS group compared with the ORCS group (MD: − 2.10, 95% CI: − 3.47 to − 0.73; P = 0.003). However, the heterogeneity was high (P < 0.00001, I 2 = 92%) (Fig. 4A).

Long-term outcomes. Disease-free survival (DFS). Two studies reported DFS outcomes 25,28 . Kang et al.
reported that the 2-year DFS was 83.5% in the RRCS group and 79.7% in the ORCS group. Ghezzi et al. 25 reported that the 5-year DFS was 73.2% and 69.5% in the RRCS and ORCS groups, respectively. The combined data indicated no differences in DFS between the two arms (HR: 0.84, 95% CI: 0.53-1.35, P = 0.47) and no heterogeneity (P = 0.87, I 2 = 0) (Fig. 6). Overall survival. Ghezzi et al. 25 reported that the 5-year overall survival rate was higher in the RRCS group than the ORCS group, but this difference was not significant (85.0% vs. 76.1%). Future studies should be conducted to assess this index.
Publication bias. A funnel plot analysis of all studies was performed in the meta-analysis of overall postoperative complications between RRCS and ORCS. Visually, all of the studies were within the limits of the 95% CIs (Fig. 7). Moreover, the statistical test indicated no evidence of publication bias (Egg's test P = 0.174, Begg's test P = 0.764).  Table 2.

Discussion
Several meta-analyses have evaluated the efficacy and safety of robotic-assisted surgery versus open surgery in gastric cancer 29,30 , pancreatic disease 31 , renal disease 32 , and bladder cancer 33 . All of these studies have indicated that robotic-assisted surgery is safe and effective. However, no meta-analysis has been conducted to evaluate robotic-assisted surgery compared with open surgery for rectal cancer.
The combined data indicated that compared with the ORCS group, the EBL was significantly lower in the RRCS group. Due to a lower EBL, it is possible to suggest that RRCS may significantly reduce the probability of transfusion. Indeed, our meta-analysis suggested that the transfusion rate was significantly lower in the RRCS group compared with the ORCS group. Thus, reduced transfusion rates may prevent the recurrence of cancer 34 . Patients with cancer who receive more intraoperative blood transfusions are at greater risk for cancer recurrence, and the volume of transfused blood at surgery is an independent risk factor for cancer recurrence 35 . In addition, more EBL may indicate an unfavorable prognosis of patients with cancer 36,37 .
Another advantage of RRCS is its associated shorter LOS. The shorter LOS may be explained by the following rationale: RRCS is a minimally invasive surgery, and it may provide faster wound recovery and reduce postoperative pain. RRCS has a shorter time to flatus passage and a faster recovery to normal diet resumption.
The overall complication rate was similar between the two techniques (7% in the RRCS group and 8% in the ORCS group). This finding also indicated that RRCS is as safe and feasible as ORCS. The most frequently occurring events included anastomotic leakage, wound infection, ileus, and pelvic abscess. Our analysis indicated that the two groups had no differences in those regards.  Anastomotic leakage is a major surgical complication of gastrointestinal surgery 29 , and it represents one of the most dreaded complications of colorectal cancer. The mortality rate is high in patients with colorectal anastomotic leakage that required re-operation and accounts for almost 40% of the deaths after colorectal cancer surgery 29 . The anastomotic leakage rate was 6.49% (31/478) in the RRCS group and 4.5% (25/556) in the ORCS group, which was not a significant difference. The rate (6.49%) was consistent with previous reports for robotic surgery (1.8-12.1%) 5,38,39 and was similar to the rate of 7% that was reported in open rectal cancer surgery based on a multicenter randomized controlled trial 40 .
The present study revealed no difference in oncologic clearance (proximal margin, circumferential resection margin, distal margin, and number of retrieved lymph nodes). Resection margins and number of retrieved lymph nodes have been regarded as quality indicators in rectal cancer surgery 41 because the distal margin, circumferential resection margin, and number of retrieved lymph nodes were important parameters for the evaluation of prognosis in rectal cancer [42][43][44][45] . Several studies have indicated that RRCS can achieve good-quality performance.
The long-term follow-up was reported in two studies. The results indicated that there was no difference in DFS or overall survival between the two groups. Hara et al. reported that the 5-year overall survival and DFS rates were 88.6% and 76.6%, respectively, for patients diagnosed with stage III rectal cancer who had undergone robotic surgery 7 . The 3-year overall survival rate was 97% 46 .
Robotic rectal cancer surgery can achieve promising survival rates. Due to the limited number of studies, more studies are mandatory to establish the value of robotic surgery for rectal cancer in the future.
We should note the disadvantage of RRCS, namely, it requires a longer operative time compared with ORCS. This result is mainly attributed to the docking and preparation times associated with RRCS. A previous study reported that the median operative time for RRCS ranged from 240 to 310 min 46 . With increasing experience, the operative time would be reduced in robotic surgery 29 . It was reported that a surgical team requires approximately 30 cases to become comfortable and proficient with RRCS 28 .
The following limitations of this meta-analysis should be considered. First, the included studies were not randomized controlled trials; some studies were prospective studies, and some were retrospective studies. Thus, the studies may have been biased, and the results should be interpreted with caution. Second, as a novel technique, the cost-effectiveness of RRCS should be considered. Of the included studies, only that of Berten et al. 23 reported the total cost, which was 11214€ for RRCS and 9858€ for ORCS; the combined data for this index were not available. However, a variety of studies have indicated that the cost of robotic colorectal surgery is higher than the cost of laparoscopic colorectal surgery 8,[47][48][49] . The high capital and running costs of robotic systems have precluded their widespread use in many countries 50 . Third, the studies included patients with different basic characteristics and treatments, and these differences may have affected some of the results. Moreover, the surgeries in the included cases were carried out by different surgeons, and the different experiences and techniques of the surgeons may have affected some of the results 34 . Fourth, only two studies reported the survival outcomes after long-term follow-up; more studies are needed to assess the survival outcome as well as the recurrence events. Fifth, the studies included in this meta-analysis were limited to those published in the English language because the authors of the present study were not literate in other languages. Thus, studies published in English may have more frequently supported our hypotheses, and studies reported in other languages may have more frequently refuted our hypotheses.
In conclusion, this meta-analysis suggests that RRCS is safe and effective. RRCS was associated with reduced EBL, less intraoperative transfusion requirements, a shorter time to flatus passage, a shorter time to resumption of a normal diet, and reduced LOS. There were no significant differences in complication rates, oncologic clearance, and survival outcomes between the two groups. However, RRCS was associated with a longer operative time compared with ORCS. Future well-designed, larger, randomized controlled studies should be performed to assess the clinical and financial benefits and oncologic outcomes of RRCS to establish its role in the minimally invasive management of rectal cancer.