Background
During conventional multi-port laparoscopic colonic resection (MLC), the camera and surgical instruments are inserted through 4 − 5 trocars. The resected colon part is extracted by an additional minilaparotomy (i.e. low Pfannenstiel or midline incision). Laparoscopic colonic surgery increasingly became the new standard for colorectal resection [
1,
2]. There is evidence that incisional hernias are less frequent using the total-laparoscopic approach instead of open abdominal surgery [
3]. For caesarean section, wound length was found as a risk factor for surgical-site complications [
4].
Therefore, newer approaches and advances of the minimal invasive surgery aim to minimise the total length of incisions even further, which in turn may reduce the morbidity of the abdominal wall, such as wound pain, wound infection and hernia formation. This implies the expectation of a faster recovery in the early postoperative phase. One way to achieve this aim is to minimise the number of incisions used. Single-incision laparoscopic colectomy (SILC) uses only one umbilical port site [
5,
6]. However, the likely limitations of SILC include an additional learning curve and advanced laparoscopic skill requirements [
7], because triangulation is missing, when all instruments are oriented intraabdominally in the same direction [
8,
9].
There are several meta-analyses published [
10‐
15] comparing SILC to MLC, none of which included randomised controlled trials exclusively, but predominantly observational studies such as case-matched studies. Most of these reviews noted substantial heterogeneity in some of their outcomes [
10,
11,
13], which might reflect the differences in study design, surgical technique, patient selection, postoperative care or even the incomplete learning curve among the different studies. The potential bias of the results due to the low quality of the included studies was also addressed by several reviews [
10‐
12]. It is important to assess the efficacy and safety of SILC by preparing this systematic review based on only RCTs. Including only RCTs minimises the heterogeneity and potential bias mentioned above that might be introduced into the analysis by the inclusion of observational studies.
The aim of this work is the assessment of the available evidence. This includes an investigation of the methodological quality and results of previously published meta-analyses comparing SILC to MLC in adult patients. Furthermore and as previously specified in the protocol to this systematic review [
16], we compared SILC to MLC in adult patients, in whom elective colectomy is indicated because of malignant or benign disease, by performing a systematic review and meta-analysis of randomised controlled trials.
Methods
Previously published meta-analyses on this topic (i.e. SILC vs. MLC) were systematically identified from the same literature searches as described below. Meta-analyses were eligible, if they examined SILC in the treatment of malignant or benign diseases of colon or rectum. The methodological quality of these meta-analyses was assessed by using AMSTAR (‘A Measurement Tool to Assess Systematic Reviews’), which contains 11 single items and give a maximum score of 11 points [
17]. The appraisal of meta-analyses was independently done by two reviewers.
We conducted this systematic review according to a pre-specified protocol [
16]. The protocol describes the surgical procedures studied, the eligible patient groups, as well as the pre-specified methods (i.e. criteria for considering studies for this review, search strategy, data collection and analysis). Thus, unless stated otherwise, the present systematic review was performed according to the protocol [
16].
The search was conducted from 2008 to March 2016. Electronic literature searches were performed in the databases CENTRAL, MEDLINE and EMBASE. For the search in two clinical trial registries, the following terms were used: ‘single-incision laparoscopic colectomy’, ‘single-port laparoscopic colectomy’, ‘single AND colectomy’, ‘single AND incision AND colon’, ‘single AND incision AND colectomy’, ‘single AND port AND colon’, ‘single AND port AND colectomy’, ‘transumbilical AND colectomy’, ‘transumbilical AND colon’, ‘notes AND colectomy’, ‘notes AND colon’. In addition, a manual search of several potentially relevant systematic reviews and meta-analyses on this topic [
10‐
15,
18] was carried out to identify additional trials.
Eligible studies were selected independently by two authors according to the previously specified criteria (i.e. RCT, SILC and MLC as intervention, adult patients, in whom elective colectomy was indicated because of either malignant or benign disease). Primary outcomes were previously defined as local complications (intra- and postoperative events) and mortality. Secondary outcomes were defined as conversion rate to laparoscopic, hand-assisted laparoscopic or open surgery, estimated blood loss, operative time, number of patients with R0 resection, tumour-free resections or both, number of lymph nodes harvested, postoperative pain intensity, general complications, resumption of bowel function, length of hospital stay, quality of life or fatigue, cosmetic results and disease-free survival.
We contacted authors of potentially eligible studies to obtain any missing information. The study by Poon et al. [
19] presented their results as median, but they kindly provided mean and standard deviations for their reported continuous outcomes upon request. Therefore, no imputation of missing data was relevant. The risk of bias assessment was performed using the criteria described in the Cochrane Handbook for Systematic Reviews of Interventions [
20].
We intended to explore reasons for heterogeneity (Chi2 test with significance being set at
P value < 0.05) in the studies using subgroup and sensitivity analysis, but this was not possible due to the low number of studies. This is also the reason, why the assessment of potential publication bias using a funnel plot would not have been meaningful. In cases of substantial statistical heterogeneity we did not pool the results. Analysis was conducted using Review Manager Version 5.3 [
21].
Discussion
We assessed 6 previous meta-analyses, which all tried to overcome the current sparseness of high-quality data by including observational studies such as case-matched studies. We assessed the methodological quality of these reviews and found it to be rather low. Therefore, the results and conclusions of these reviews may be affected by substantial methodological bias. Furthermore, we noticed that their reported results were inconsistent. While three reviews [
10,
11,
15] reported that they could not find any difference concerning the conversion rate, one review [
12] reported a higher conversion rate in the SILC group and another review reported a lower conversion rate in the SILC group [
14]. We also noticed that several eligible studies were not included by several of these meta-analyses, although these primary studies were published at the time of literature search.
In the our systematic review and meta-analysis we identified two randomised controlled trials [
19,
23], including 82 participants with malignant diseases, 41 in each of the two treatment. There were no patients with benign diseases included in this review. Based on these trials we found insufficient evidence to clarify whether single-incision laparoscopic colectomy (SILC) leads to less local complications (including both intraoperative and postoperative events) or lower mortality. Due to the small number of included studies, lack of event occurrence, as well as substantial heterogeneity in one outcome (number of lymph nodes harvested), meta-analyses were conducted only for two (operative time and length of hospital stay) of the 12 regarded secondary outcomes. Besides a significantly shorter hospital stay of one day, there was no statistically significant difference between SILC and MLC observed.
The total length of scar was not reported in either of the included studies. However, according to the description of the operation procedures, the reduction of total incision length in the SILC group is only a few centimetres. After adjustment for multiplicity, the reported postoperative pain intensity by Poon et al. [
19] was statistically significantly reduced in the SILC group at the first day after the operation. The pain intensity of SILC was 1.64 points lower on average compared to the MLC with a 95 % confidence interval of 0.67 − 2.61. Since the confidence interval is relatively wide, it is not clear whether this difference is clinically important [
68]. Since none of the studies reported cosmetic results, quality of life, or fatigue, and the reported reduction of pain intensity in the SILC group may be clinically unimportant, there is no evidence to investigate any potential dependency between post-operative comfort and total length of scars. Further randomised controlled trials are necessary to replicate reported results and to resolve inconsistencies between the studies.
The quality of the evidence was low, due to the sparse data and because the results from one of the two included studies were of a high risk of bias. Thus, the main limitation of this systematic review is the limited number of patients included in the meta-analysis for primary outcomes and very limited available results on secondary endpoints. However, this is not a limitation of our work but due to a lack of evidence and therefore not remediable.
The results of randomised and non-randomised studies sometimes differ [
69] and non-randomised studies produce, on average, effect estimates that indicate more extreme benefits of the effects of health care than randomised trials [
70], which is why it is not surprising that our review, including only RCTs, differs from these other reviews in some of the outcomes. Since these reviews predominantly present non-significant or heterogeneous results, our results are mostly in agreement with at least one of the reviews results. Thus, the inclusion of observational data does not lead to more reliable clinical recommendations, but instead leads to heterogeneity of results and increases risk of bias, due to the very low quality of the included studies. Clearly, more data are not necessarily better data. Therefore, our review presents the most reliable evidence currently available, by means of randomised controlled trials.
Although the available data are very sparse, so that there is a possibility that the lack of findings are due to a lack of evidence of effect and not due to a lack of effect itself, studies evaluating single incision laparoscopic surgery in different application areas showed similar results. A recent patient- and assessor-blinded randomised multi-centre trial [
71], as well as a review including 659 patients from nine RCTs investigating single incision versus multi-incision laparoscopic cholecystectomy [
72], could not show any benefit of single-incision laparoscopic cholecystectomy in postoperative pain, operating time, hospital stay and complication rate. The only significant benefit single-incision laparoscopic cholecystectomy showed in those studies was a better cosmetic result. Thus, one should consider the possibility that small advantages of SILC might be clinically irrelevant. Also, the available data enclosed in our review did not include a sufficient follow-up period to assess any long-term benefit or harm, so that a potentially negative effect of SILC cannot be excluded.
Although some surgeons will rate the present meta-analysis with only two included studies as not very valuable, it is important to describe how little data on SILC exist so far. Neither our review, including randomised controlled studies, nor the reviews, which also included observational studies, could confirm the safety. The lack of high-quality studies precludes a confirmation of safety, both in the short-term and in the oncological long-term. Nevertheless, surgeons increasingly practice SILC, which can be seen by the increasing number of published articles on this topic. The number of comparative studies published in the last four years has almost tripled [
10,
11,
13‐
15], not counting the case reports and series published during this time. Hence, the main purpose of our systematic review is to remind the surgical community, currently deciding whether or not to use this new method that safety and effectiveness of SILC are yet to be confirmed. Also, this review acts as a warning sign that SILC should only be performed in a research setting. It is inevitable to wait for the results of further RCTs to be published.
Conclusion
The currently available study results are too sparse to detect (or rule out) relevant differences between SILC and MLC. The quality of the current evidence is low, and the additional analysis of non-randomised data attempts, but does not solve this problem. For colorectal cancer patients, it is essential to assess oncologic outcomes (e.g. disease-free survival) in the long-term. For some complications (e.g. incisional hernia), a longer follow-up time is also necessary. SILC should still be considered as an experimental procedure, since the evidence of well-designed randomised controlled trials is too sparse to allow any recommendation.
Acknowledgments
We thank the trial search coordinator (TSC) of CCCG for developing the Search strategy. We are very grateful to Elke Hausner for supporting us by screening the titles and abstracts of our electronic search from 2008 to 2015. We thank Dr. Poon, Dr. Mingoli, Dr. Zapolskiy and Dr. Panis for providing additional information on their trials.