High Versus Low Ligation of the Inferior Mesenteric Artery in Colorectal Cancer Surgery: A Systematic Review and Meta-Analysis

Background and Objectives: This study aimed to compare the effects of high ligation (HL) versus low ligation (LL) in colorectal cancer surgery. Materials and Methods: We performed a comprehensive search using multiple databases (trial registries and ClinicalTrials.gov), other sources of grey literature, and conference proceedings, with no restrictions on the language or publication status, up until 10 March 2021. We included all parallel-group randomized controlled trials (RCTs) and considered cluster RCTs for inclusion. The risk of bias domains were “low risk,” “high risk,” or “unclear risk.” We performed statistical analyses using a random-effects model and interpreted the results according to the Cochrane Handbook for Systematic Reviews of Interventions. We used the GRADE guidelines to rate the certainty of evidence (CoE) of the randomized controlled trials. Results: We found 12 studies (24 articles) from our search. We were very uncertain about the effects of HL on overall mortality, disease recurrence, cancer-specific mortality, postoperative mortality, and anastomotic leakage (very low CoE). There may be little to no difference between HL and LL in postoperative complications (low CoE). For short-term follow-up (within 6 months), HL may reduce defecatory function (constipation; low CoE). While HL and LL may have similar effects on sexual function in men, HL may reduce female sexual function compared with LL (low CoE). For long-term follow-up (beyond 6 months), HL may reduce defecatory function (constipation; low CoE). There were discrepancies in the effects regarding urinary dysfunction according to which questionnaire was used in the studies. HL may reduce male and female sexual function (low CoE). Conclusions: We are very uncertain about the effects of HL on survival outcomes, and there is no difference in the incidence of postoperative complications between HL and LL. More rigorous RCTs are necessary to evaluate the effect of HL and LL on functional outcomes.


Introduction
Colorectal cancer accounts for approximately 10.2% of all cancers and is the third-mostcommon cancer in terms of incidence and the second-most-common cause of cancer-related mortality [1]. Left-sided colorectal cancer accounts for up to two-thirds of all colorectal cancers [2]. Most patients can be treated with radical surgery with or without perioperative chemotherapy and radiotherapy. Surgery has been recommended as the gold-standard treatment for colorectal cancer.
The removal of the tumor and wide resection of the colonic mesentery with vascular ligation have been the standard techniques for the surgical treatment of left-sided colon cancer and rectal cancer [3]. However, currently, there is no worldwide consensus on the optimal level of arterial ligation in terms of oncological outcomes, postoperative morbidity, and functional outcomes. Thus, most surgeons determine the level of ligation based on experience.
In 1908, Moynihan and Miles first introduced two different techniques for the ligation of the inferior mesenteric artery (IMA): high ligation (HL) and low ligation (LL) [4]. HL refers to the ligation of the IMA immediately after it branches off the anterior surface of the abdominal aorta, whereas LL refers to ligation at the level of the superior rectal artery, preserving the left colic artery (LCA).
When determining the level of IMA ligation, oncologic outcomes, functional outcomes, and technical safety should be considered. In general, HL of the IMA is technically difficult, may enable more radical lymphadenectomy [5,6], and may lead to adverse functional outcomes because autonomic nerves around the origin of the IMA may be damaged. In contrast, LL of the IMA may provide an abundant blood supply to the proximal end of the anastomotic site, reduce the risk of autonomic nerve injury, and result in less radical lymph node dissection [7,8].
Therefore, we hypothesized that HL of the IMA may be beneficial for patients' survival compared with LL, because more radical lymphadenectomy may be possible. In addition, LL of the IMA may be better than HL in terms of functional outcomes and postoperative complications due to the preservation of autonomic nerve functions, abundant blood supply, and less radical lymph node dissection.
In this study, we tried to analyze important patient outcomes, such as survival outcomes, postoperative complications, and anastomotic leakage. In addition, in particular, functional outcomes, such as defecatory dysfunction, urinary dysfunction, and sexual dysfunction, were assessed in our review to determine which level of ligation is superior in terms of functional impairment.

Literature Search
This systematic review and meta-analysis was performed according to the protocol published by PROSPERO (registration number: CRD42021241241). This systematic review follows PRISMA guidelines (Tables S1 and S2).
We performed a comprehensive search of several databases, including MEDLINE, EMBASE, Cochrane Library, Scopus, Web of Science, Latin American and Caribbean Health Sciences Literature, and other resources, such as ClinicalTrials.gov (www.clinicaltrials.gov/: accessed on 10 March 2021), the World Health Organization International Clinical Trials Registry Platform search portal (apps.who.int/trialsearch/: accessed on 10 March 2021), and OpenGrey (www.opengrey.eu/: accessed on 10 March 2021). Table S3 presents the search strategy for each database. We also searched the reference lists of the selected studies for Supplemental Studies and contacted their authors for reports of unpublished or published studies, including new or additional studies, or works in progress.
The date of the last search of all databases was 10 March 2021. Reference management software (EndNote version 20, Clarivate Analytics, Boston, MA, USA) was used to identify and remove potentially duplicated records. Two review authors (KK and SA) independently screened all potentially relevant records and classified the studies according to the criteria provided in the Cochrane Handbook for Systematic Reviews of Interventions [9]. Rayyan-a web and mobile application for systematic reviews (available at www.rayyan.ai)-was used for screening. All disagreements were resolved through discussion. We included all parallel-group randomized controlled trials (RCTs) and considered cluster RCTs for inclusion. Crossover studies that were not applicable and nonrandomized studies were excluded. Studies were included regardless of the publication status or language of publication.

Type of Participants
We defined the eligible patient population as all patients undergoing open or laparoscopic anterior resection and low anterior resection for curable colorectal cancer. Trials including patients with massive invasion of colorectal cancer into adjacent organs that could not be resected, synchronous unresectable metastasis or peritoneal metastasis, and those with inoperable disease owing to comorbidities were excluded.

Types of Interventions and Comparators
We compared HL and LL. Concomitant interventions had to be the same in the experimental and comparator groups to establish fair comparisons. HL was defined as IMA ligation immediately after it branches off the anterior surface of the abdominal aorta, whereas LL was defined as IMA ligation at the level of the superior rectal artery, preserving the LCA.

Type of Outcomes
We did not measure the outcomes assessed in this review as eligibility criteria.

Primary Outcomes
Overall mortality (OM) and overall postoperative complications were regarded as the primary outcome measures. OM was defined as the length of time from randomization to death from any cause. Overall postoperative complications were defined as complications occurring within 30 days after surgery that could be classified according to the Clavien-Dindo classification.

Secondary Outcomes
Disease recurrence (DR), cancer-specific mortality (CSM), postoperative mortality, anastomotic leakage, defecatory dysfunction, urinary dysfunction, and sexual dysfunction were regarded as secondary outcome measures. DR was defined as the length of time from randomization to recurrence. CSM was defined as the length of time from randomization to cancer-related death. Postoperative mortality was defined as the number of deaths within 30 days after surgery. Anastomotic leakage was defined as incontinuity at the anastomotic site detected clinically or radiologically within 30 days after surgery. Defecatory dysfunction was assessed using the Fecal Incontinence Quality of Life (FIQL) scale, Jorge-Wexner Incontinence Score (JWIS), Agachan-Wexner Constipation Score (AWCS), or the Gastrointestinal Quality of Life Index (GIQLI). Urinary dysfunction was assessed using the International Consultation on Incontinence Questionnaire-Urinary Incontinence (ICIQ-UI) and International Prostate Symptom Score (IPSS). Sexual dysfunction was assessed using the International Index of Erectile Function (IIEF) and the Female Sexual Function Index (FSFI).
We considered outcomes for defecatory, urinary, and sexual dysfunction measured up to and including 6 months after randomization as short-term, and those beyond 6 months as long-term outcomes.

Assessment of Risk of Bias in Included Studies
Two review authors (KK and SA) independently assessed the risk of bias in each included study. All disagreements were resolved through discussion. We planned to assess the risk of bias of the RCTs using the Cochrane risk of bias tool for randomized trials. The risk of bias domains were "low risk," "high risk," or "unclear risk," and they were evaluated using individual items, as described in the Cochrane Handbook for Systematic Reviews of Interventions [10].

Data Collection and Analysis
Outcome data were extracted as needed to calculate summary statistics and measures of variance. The collected information for the included studies is provided in Table S4. For dichotomous outcomes, we obtained the number of events and their proportions, and the summary statistics with corresponding measures of variance. For continuous outcomes, we obtained the means, standard deviations, or other necessary data. We calculated the hazard ratios (HRs) using the method of Tierney et al. [11] and the corresponding 95% confidence intervals (CIs) for time-to-event outcomes, and analyzed the data using a random-effects model. Review Manager 5 software (The Cochrane Collaboration, Copenhagen, Denmark) was used for statistical analysis. We assessed the impact of heterogeneity on the metaanalysis and interpreted it according to the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions [9]. We expected the characteristics, such as age (younger than 65 years versus older than 65 or 65 years of age), adjuvant therapy (adjuvant therapy versus no adjuvant therapy), and tumor stage (localized versus locally advanced versus advanced), to be heterogeneous and planned to carry out subgroup analyses with an investigation of the interactions limited to primary outcomes. Sensitivity analyses of primary outcomes were only planned for RCTs to explore the influence of the risk of bias (when applicable) on the effect sizes by excluding studies with high or unclear risks. However, we could not perform secondary analyses because there were no relevant data, and the RCTs were scarce. If there were at least 10 studies investigating a particular outcome, funnel plots were used to assess small-study effects.

Summary of Findings Table
We assessed the overall certainty of the evidence (CoE) for each outcome according to GRADE. Two review authors (KK and JHJ) independently rated the CoE for each outcome, and resolved any discrepancies by consensus. We considered the criteria related to internal validity (risk of bias, inconsistency, imprecision, and publication bias) and external validity, such as the directness of results [12].

Search Results
In total, 2403 records were identified through our database searches. A gray literature repository was also found. Two additional records were identified [13,14]. After removing duplicates, initial screening was performed for the titles and abstracts of 1370 records, and 1339 records were excluded. After the initial screening, screening of the full text of 31 articles was performed, and 5 full-text articles were excluded for the following reasons: nonrandomized articles (4 articles) [15][16][17][18] and different interventions (1 article) [19]. Two studies (two articles) were ongoing [20,21]. Finally, 12 RCTs (24 articles) that met the inclusion criteria were included for qualitative synthesis in this review. The assessment process is illustrated in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart ( Figure 1).
A total of 1431 randomized participants were included in the studies. The mean age of the RCTs, participants ranged from 49.9 to 69.0 years. The total number of participants with stage 0/I disease was 445; stage II, 390; stage III, 400; and stage IV, 27. The number of participants by stage was not reported in two RCTs [24,28]. Five RCTs [14,23,25,26,33] excluded patients with stage IV disease, and two RCTs [22,32] did not report the number of patients with stage IV disease. Fiori et al. [23] conducted their study on patients with A total of 1431 randomized participants were included in the studies. The mean age of the RCTs, participants ranged from 49.9 to 69.0 years. The total number of participants with stage 0/I disease was 445; stage II, 390; stage III, 400; and stage IV, 27. The number of participants by stage was not reported in two RCTs [24,28]. Five RCTs [14,23,25,26,33] excluded patients with stage IV disease, and two RCTs [22,32] did not report the number of patients with stage IV disease. Fiori et al. [23] conducted their study on patients with sigmoid colon cancer; Kruszewski et al. [14] included patients with rectosigmoid colon cancer; and all other studies were conducted among patients with rectal cancer. Participants who underwent only laparoscopic surgery were recruited in eight RCTs [22,23,[25][26][27][28]32,33], whereas other RCTs recruited participants who underwent laparoscopic or open surgery [14,24,[29][30][31]. Patients who underwent neoadjuvant chemoradiotherapy or radiotherapy were excluded from eight RCTs [22,[24][25][26]28,[31][32][33]. A total of 294 patients (HL: 149 and LL: 135) underwent adjuvant chemotherapy, and 181 participants (HL: 86 and LL: 95) underwent protective stoma formation (Table 1). (a) HL, high ligation, LL, low ligation, NR, not reported, RCT, randomized controlled trial; a . not defined; b . mean age was based on randomized patients; c . two articles were from 1 RCT; d . median (range). (b) ALND, apical lymph node dissection, AR, anterior resection, CRT, chemoradiotherapy, CTx, chemotherapy, HL, high ligation, HO, Hartmann's procedure, LAR, low anterior resection, LL, low ligation, NR, not reported; a . no significant difference; b . p-value not reported; c . preoperative radiotherapy was excluded; d . Baseline characteristics of Mari et al. [27] were based on randomized patients (n = 214); e . statistically significant; f . Two articles were from 1 RCT.

Excluded Studies
We excluded 4 studies [15][16][17]19] (5 articles) of 18 studies (29 articles) after the evaluation of the full-text articles. Three studies (four articles) were not RCTs: two were nonrandomized studies [16,17] and one was a retrospective study [15] using RCT data collected for other purposes. The intervention and comparator in one excluded study were IMA dissection first versus inferior mesenteric vein dissection, which were different from the intervention and comparator in our study [19] (Table S6).

Risk of Bias of Included Studies
The details of the risk of bias in the included studies are described in Figure 6.   Guo et al. [28] was funded by the Health Project of Jilin Province, China, and Zhou et al. [32] was funded by the Guangzhou Important Special Program of Health Medicine Cooperation and Innovation (Grant number: 201604020005). All authors of the included studies declared no conflicts of interest.

Excluded Studies
We excluded 4 studies [15][16][17]19] (5 articles) of 18 studies (29 articles) after the evaluation of the full-text articles. Three studies (four articles) were not RCTs: two were nonrandomized studies [16,17] and one was a retrospective study [15] using RCT data collected for other purposes. The intervention and comparator in one excluded study were IMA dissection first versus inferior mesenteric vein dissection, which were different from the intervention and comparator in our study [19] (Table S6).

Risk of Bias of Included Studies
The details of the risk of bias in the included studies are described in Figure 6. Guo et al. [28] was funded by the Health Project of Jilin Province, China, and Zhou et al. [32] was funded by the Guangzhou Important Special Program of Health Medicine Cooperation and Innovation (Grant number: 201604020005). All authors of the included studies declared no conflicts of interest.

Excluded Studies
We excluded 4 studies [15][16][17]19] (5 articles) of 18 studies (29 articles) after the evaluation of the full-text articles. Three studies (four articles) were not RCTs: two were nonrandomized studies [16,17] and one was a retrospective study [15] using RCT data collected for other purposes. The intervention and comparator in one excluded study were IMA dissection first versus inferior mesenteric vein dissection, which were different from the intervention and comparator in our study [19] (Table S6).

Risk of Bias of Included Studies
The details of the risk of bias in the included studies are described in Figure 6. Seven studies were classified as having a low risk of bias for random sequence generation [14,24,25,27,[29][30][31][32]. Two studies were classified as having a low risk of bias for allocation concealment [14,27]. Eight studies were rated as having a high risk of bias for blinding participants and personnel because double blinding was impossible owing to the surgical trial nature [14,23,[25][26][27][28][29][30][31]. Other studies were rated as having an unclear risk of bias owing to the lack of information regarding the blinding method [22,24,32,33]. One RCT was classified as having a low risk of bias for blinding of the outcome assessment of subjective outcomes (postoperative complications, anastomotic leakage, defecatory dysfunction, urinary dysfunction, and sexual dysfunction), because the investigators and outcome assessors were blinded in these studies [29,30]. Six studies were classified as having a high risk of bias for selective reporting because the outcomes in the material and Seven studies were classified as having a low risk of bias for random sequence generation [14,24,25,27,[29][30][31][32]. Two studies were classified as having a low risk of bias for allocation concealment [14,27]. Eight studies were rated as having a high risk of bias for blinding participants and personnel because double blinding was impossible owing to the surgical trial nature [14,23,[25][26][27][28][29][30][31]. Other studies were rated as having an unclear risk of bias owing to the lack of information regarding the blinding method [22,24,32,33]. One RCT was classified as having a low risk of bias for blinding of the outcome assessment of subjective outcomes (postoperative complications, anastomotic leakage, defecatory dysfunction, urinary dysfunction, and sexual dysfunction), because the investigators and outcome assessors were blinded in these studies [29,30]. Six studies were classified as having a high risk of bias for selective reporting because the outcomes in the material and methods section were not the same as the actual reported outcomes [14,23,[26][27][28][29][30]. Two studies were classified as having a high risk of bias for other biases because the protocol of the studies was different from that of the published journal [23,26].  We are very uncertain about the effects of HL on improving anastomotic leakage

Effects of Interventions
The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MCID: minimal clinically important difference; RCT: randomized controlled trial; HR: hazard ratio; HL: high ligation; LL: low ligation; RR: risk ratio GRADE working group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of the effect a . Downgraded by one level due to study limitations: allocation was clearly not concealed in most of the included studies, and/or participants were clearly not blinded in the included studies; b . Downgraded by two levels due to imprecision: wide confidence interval crosses the assumed threshold of clinically important difference; c . Downgraded by one level due to inconsistency due to clinically important heterogeneity; d . Not downgraded further due to imprecision: wide confidence intervals attributed to the observed inconsistency (for which we rated down); e . Downgraded by one level due to imprecision: confidence interval crosses the assumed threshold of clinically important difference; f . Downgraded by two levels due to imprecision: small study population or very rare events; g . Downgraded by one level due to publication bias: asymmetry of funnel plot with dominant positive results. The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of the effect a . MCID: 25% improvement (greater than 1 point) from the baseline (HL: number; LL: number). b . Downgraded by one level due to study limitations: high or unclear risk of performance and detection bias. c . Downgraded by one level due to imprecision: optimal information size was not met. d . Downgraded by one level due to imprecision: confidence interval crosses the assumed threshold for a clinically important difference. e . MCID: from Shi et al. [37]. f . MCID: from Lim et al. [38]. g . MCID: from Barry et al. [39]. h . MCID: from Spaliviero et al. [40]. i . MCID: from Krychman et al. [41].
(2) Urinary dysfunction Two hundred and forty-two (HL: 123 and LL: 119) participants from two RCTs [26,27] for the ICIQ-UI were analyzed. There may be little to no difference in urinary dysfunction assessed with the ICIQ-UI between HL and LL (MD: 1.90, 95% CI: 0.82-2.99; I2 = 54%; low CoE) ( Table 4). This was downgraded due to serious imprecision and serious study limitations. Analysis of 196 (HL: 101 and LL: 95) participants from one RCT [27] for the IPSS was performed. HL may aggravate urinary symptoms assessed with the IPSS compared with LL (MD: 4.72, 95% CI: 2.43-7.01; low CoE) ( Table 4). This was downgraded due to serious imprecision and serious study limitation.

(3) Sexual dysfunction
Analyses of 158 (HL: 84 and LL: 74) participants from two RCTs [26,27] for the IIEF-5 and 46 participants (HL: 22 and LL: 24) for the FSFI were performed. HL may reduce sexual function assessed with the IIEF-5 (MD: −5.11, 95% CI: −6.85 to −3.37; I2 = 0%; low CoE) and the FSFI (MD: −5.00, 95% CI: −6.74 to −3.26; low CoE) compared with LL (Table 4). We downgraded the CoE for each questionnaire due to serious imprecision and serious study limitations.  We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of the effect a . MCID: 25% improvement (greater than 1 point) from the baseline (HL: number; LL: number). b . Downgraded by one level due to study limitations: high or unclear risk of performance and detection bias. c . Downgraded by one level due to imprecision: optimal information size was not met. d . Downgraded by one level due to imprecision: confidence interval crosses the assumed threshold for a clinically important difference. e . MCID: from Shi et al. [37]. f . MCID: from Lim et al. [38]. g . MCID: from Barry et al. [39]. h . MCID: from Spaliviero et al. [40]. i . MCID: from Krychman et al. [41].

Subgroup Analysis
In terms of OM, we found an RR of 0.91 (95% CI 0.36-2.32) with localized disease versus an RR of 1.01 (95% CI 0.48-2.12) with locally advanced disease versus an RR of 2.86 (95% CI 0.79-10. 36) with advanced disease. The test for interaction showed no evidence of a difference between the subgroups (p = 0.32, I2 = 11.9 %) (Figure 7). In terms of DR, we found an RR of 0.73 (95% CI 0.36-1.47) with localized disease versus an RR of 1.30 (95% CI 0.89-1.91) with locally advanced disease. The test for interaction showed no evidence of a difference between the subgroups (p = 0.15, I2 = 51.2 %) (Figure 8). Other subgroup analyses were not performed because of the limited data.

Subgroup Analysis
In terms of OM, we found an RR of 0.91 (95% CI 0.36-2.32) with localized disease versus an RR of 1.01 (95% CI 0.48-2.12) with locally advanced disease versus an RR of 2.86 (95% CI 0.79-10. 36) with advanced disease. The test for interaction showed no evidence of a difference between the subgroups (p = 0.32, I2 = 11.9 %) (Figure 7). In terms of DR, we found an RR of 0.73 (95% CI 0.36-1.47) with localized disease versus an RR of 1.30 (95% CI 0.89-1.91) with locally advanced disease. The test for interaction showed no evidence of a difference between the subgroups (p = 0.15, I2 = 51.2 %) (Figure 8). Other subgroup analyses were not performed because of the limited data.

Discussion
Our study showed a very uncertain effect of HL on improving survival outcomes compared with LL. Several studies have demonstrated that apical lymph node metastasis is a prognostic factor for overall survival (OS) and disease-free survival (DFS) [42][43][44]. In general, HL is considered to be oncologically safer than LL because more lymph nodes,

Subgroup Analysis
In terms of OM, we found an RR of 0.91 (95% CI 0.36-2.32) with localized disease versus an RR of 1.01 (95% CI 0.48-2.12) with locally advanced disease versus an RR of 2.86 (95% CI 0.79-10. 36) with advanced disease. The test for interaction showed no evidence of a difference between the subgroups (p = 0.32, I2 = 11.9 %) (Figure 7). In terms of DR, we found an RR of 0.73 (95% CI 0.36-1.47) with localized disease versus an RR of 1.30 (95% CI 0.89-1.91) with locally advanced disease. The test for interaction showed no evidence of a difference between the subgroups (p = 0.15, I2 = 51.2 %) (Figure 8). Other subgroup analyses were not performed because of the limited data.

Discussion
Our study showed a very uncertain effect of HL on improving survival outcomes compared with LL. Several studies have demonstrated that apical lymph node metastasis is a prognostic factor for overall survival (OS) and disease-free survival (DFS) [42][43][44]. In general, HL is considered to be oncologically safer than LL because more lymph nodes,

Discussion
Our study showed a very uncertain effect of HL on improving survival outcomes compared with LL. Several studies have demonstrated that apical lymph node metastasis is a prognostic factor for overall survival (OS) and disease-free survival (DFS) [42][43][44]. In general, HL is considered to be oncologically safer than LL because more lymph nodes, including apical lymph nodes, can be removed during HL. However, although our study showed a very uncertain effect of HL on improving survival outcomes, HR was higher in HL than LL. What should be considered is whether apical lymph node dissection (ALND) was performed during LL and whether ALND was performed appropriately during HL in each RCT. LL of the IMA with ALND was performed in six studies among all of the enrolled studies. Other studies did not report whether ALND was performed during LL. In fact, the Japanese guidelines recommend that D3 lymph node dissection can be performed for T2 or more advanced diseases [45]. Many centers perform LL with ALND on patients with clinically suspected apical lymph node metastasis [46]. Therefore, although seven studies did not report whether ALND was performed, we speculate that ALND was likely performed when apical lymph node metastasis was clinically suspected. Although a single study cannot represent all enrolled studies, Fujii et al. [31] reported that the number of harvested lymph nodes and metastatic lymph nodes around the IMA root was not different between HL and LL. Thus, lymphadenectomy during LL may be performed appropriately in enrolled RCTs. In contrast, the higher HR of HL may suggest the possibility of inappropriate ALND during HL. ALND may be incompletely performed to avoid autonomic nerve injury during HL. In fact, Turgeon et al. [47] reported that the proportion of patients with a number of harvested lymph nodes of < 12 was larger in HL than LL. For the more accurate evaluation of survival outcomes, more rigorous trials in which standardized HL, LL, and the same extent of lymphadenectomy during HL are performed are necessary.
This study showed that HL may not increase postoperative complications. In addition, the effect of HL on postoperative mortality was very uncertain because only two patients died in the LL group. These findings suggest that the level of IMA ligation does not lead to a difference in the postoperative complications and mortality.
A major complication of colon resection for sigmoid and rectal cancers is anastomotic leakage. Our study showed a very uncertain effect of HL on anastomotic leakage. Proponents of LL believe that performing LL maintains a better blood supply to the proximal colonic limb. Komen et al. [16] reported that the blood flow at the proximal colonic limb increased after LL, whereas it was not significantly decreased after HL. Guo et al. [28] reported that the marginal artery stump pressure was significantly higher in patients who underwent LL than in patients who underwent HL. Han et al. [48] reported that the time of perfusion to the colon could be more delayed after HL, but the total intensity of perfusion was similar between HL and LL in perfusion tests using intraoperative indocyanine green angiography. However, none of these studies reported a significant difference in the incidence of anastomotic leakage between HL and LL. These results indicated that, although the blood flow to the proximal colonic limb may be lower after HL than after LL, the relatively low blood flow after the HL of the IMA may be sufficient for anastomotic healing.
Another factor affecting anastomotic healing after colorectal surgery is tension between the proximal and distal colonic limbs. HL may allow tension-free anastomosis to be achieved. Some studies have reported that a much longer colonic length could be gained after HL than after LL [49][50][51]. Therefore, the lack of a difference in the incidence of anastomotic leakage between HL and LL in this study may be explained by the sufficient blood flow to the proximal colonic limb after HL and the ease of tension-free anastomosis.
Interestingly, a large, multi-institutional study with 877 patients conducted in the US showed that LL was not inferior compared with HL in terms of the anastomotic leak rate, locoregional recurrence, DFS, and OS, which is similar to the results of our study [47].
Our study showed that HL may aggravate constipation based on the minimal clinically important difference (MCID). Injury to the superior hypogastric plexus (SHP) around the IMA is common during HL, and nerve injuries may lead to defecatory dysfunction. Denervation of the proximal anastomotic site may lead to colonic hypomotility, inefficient intestinal content transport, and upstream colonic gas retention [23,30]. Long denervation of the SHP during HL may lead to a severe change in the proximal colon compared with that during LL, and this change may lead to feelings of incomplete evacuation and abdominal pain on the left side.
The inferior hypogastric plexus (IHP) is interconnected with SHP via the inferior mesenteric ganglia acting as junctions. The IHP receives pelvic parasympathetic fibers from roots S2-S5 (splanchnic nerves). These nerves are covered by the parietal fascia, pierce the endopelvic fascia, cross the retrorectal space, and form branches into the rectum via the lateral ligaments [52]. Nerve fibers from the IHP innervate the seminal vesicle, prostate, bladder, cervix, and vagina. The IHP is also responsible for penile erection, ejaculation, detrusor contractility, female arousal, and vaginal lubrication. Urinary function depends on the parasympathetic nerves for bladder emptying and the sympathetic nerves for urinary continence. Incontinence and urgency may occur if the IHP is injured. Male sexual function requires the coordination of parasympathetic nerves for erection and sympathetic nerves for ejaculation, while both nerves play a similar role in sexual arousal in females. Thus, injury to these nerves leads to erectile dysfunction and retrograde ejaculation in males, and dyspareunia in females. Injury to the SHP and inferior mesenteric ganglia may lead to the same urinary and sexual dysfunction occurring in cases of injury to the IHP due to interconnectivity. Significant urinary dysfunction assessed by the IPSS, male sexual dysfunction assessed by the IIEF-5, and female sexual dysfunction assessed by the FSFI were reported as long-term outcomes (beyond 6 months), whereas only significant female sexual dysfunction was reported within the 6 months after surgery based on the MCID. Nerve injury can be permanent due to complete nerve fiber transection or can be reversible when stretching or compression occurs [53]. Complete nerve transection may be avoided and permanent urinary and sexual dysfunction may be prevented when LL is performed instead of HL. Late improvement after partial nerve injury may have occurred in the LL group.
HL and LL in colorectal cancer surgery have been investigated in recent reviews [4,13,46,54,55]. Similar to our results, Hajibandeh et al. [13] reported no significant differences in anastomotic leakage, postoperative complications, postoperative mortality, OS, and DFS, and Kong et al. [4] found no significant difference in anastomotic leakage between HL and LL. In contrast, Jonnada et al. [46] reviewed 31 studies and reported that the LL of the IMA was associated with decreased rates of colorectal anastomotic leaks, urinary dysfunction, and overall postoperative morbidity. This study analyzed functional outcomes, including urinary dysfunction, as a dichotomous variable. However, 24 studies among the 31 enrolled studies were nonrandomized studies, and urinary dysfunction was assessed in nonrandomized studies. In addition, only seven RCTs were included in the analysis of the anastomotic leakage. In particular, Yin et al. [54] used LL with high dissection of lymph nodes as an experimental intervention to compare with HL. This study reviewed four RCTs [22,27,28,31] and 13 non-randomized studies, and reported no significant differences in OS, DFS, and systemic recurrence between HL and LL with high dissection of lymph nodes. These results are similar to our results. In addition, this study reported that the LL of the IMA with high dissection of lymph nodes was associated with decreased rates of colorectal anastomotic leaks. Tryliskyy et al. [55] utilized two RCTs [26,27] to assess the genito-urinary function at 9 months following surgery using ICIQ-UI and IIEF, and reported that the LL of the IMA demonstrated significantly better ICIQ-UI and IIEF than HL. We included the same RCTs for long-term genito-urinary function in our study. Our statistical methods for long-term genito-urinary function differed from those used in the study. We adopted the mean difference for ICIQ-UI, since both RCTs employed the same questionnaire (ICIQ-UI short form). Using the method described by Thorlund et al. [56], we converted IIEF to IIEF-5, the more well-known questionnaire. They did not follow rigorous methodologies, such as the Methodological Expectations of Cochrane Intervention Reviews (MECIR), as laid out by the Cochrane Collaboration. One study reported the CoE according to the GRADE [4]. However, the MCID or the reasons for downgrading were not presented. One study reported functional outcomes using questionnaires [55]. However, it was limited to the genito-urinary function using ICIQ-UI and IIEF.
The first advantage of our study is that we followed rigorous Cochrane methodologies and the GRADE approach to assess CoE. Secondly, to the best of our knowledge, our study is the first systematic review to investigate important patient outcomes, including all assessable functional outcomes, using questionnaires. Finally, our study comprehensively reported oncologic outcomes, including CSM, not reported in previous systematic reviews.
The first limitation of our study was that we calculated hazard ratios according to raw data from original studies. However, it was inevitably performed for meta-analysis, because a heterogeneity of the unit of measurement was observed in the enrolled RCTs. Secondly, despite a comprehensive search strategy without any publication or language restrictions, we found only a small number of studies. In this review, these studies were insufficient to generate funnel plots; therefore, the risk of publication bias may have been underestimated. Thirdly, most of the enrolled RCTs had study limitations, because RCTs for surgical interventions cannot be fully blinded. Therefore, the CoE ranged from low to very low according to GRADE, which meant that the true effect may be substantially different from what the review showed. Therefore, additional studies of better quality comparing HL to LL appear to be essential. Future trials should be conducted with higher methodologic standards. Fourthly, few RCTs have reported the functional outcomes assessed using the questionnaires. In particular, discrepancies in the reporting method were observed in four RCTs that reported the result of the FIQL. Fifthly, although HL and LL were performed in the enrolled RCTs, HL and LL may not be standardized. In particular, some of the enrolled RCTs did not report whether ALND was performed; thus, clarifying whether ALND during HL was performed appropriately is necessary. Finally, the enrolled RCTs provided limited data regarding the use of neoadjuvant therapy and the number of participants who underwent laparoscopic surgery, adjuvant therapy, disease stage, protective stoma use, and ALND. Therefore, limited subgroup analyses based on these parameters were performed.

Conclusions
We are very uncertain about the effect of HL on OM, DR, and CSM. Postoperative complications may not be different between HL and LL. We are very uncertain about the effect of HL in terms of postoperative mortality and anastomotic leakage. In addition, LL may be beneficial for short-term and long-term defecatory dysfunction (constipation). LL may be more beneficial for long-term urinary symptoms and sexual function compared with HL. However, these results were based on the low CoE; therefore, more rigorous RCTs are necessary to evaluate the effects of HL and LL on the treatment of colorectal cancer.