Preoperative risk factors associated with anastomotic leakage after colectomy for colorectal cancer: a systematic review and meta-analysis

ABSTRACT Introduction: anastomotic leak (AL) after colectomy for colorectal cancer (CRC) is a life-threatening complication. This systematic review and meta-analysis aimed to evaluate the preoperative risk factors for AL in patients submitted to colectomy. Methods: the bibliographic search covered 15 years and 9 months, from 1st January 2005 to 19th October 2020 and was performed using PubMed, Cochrane Library, Scopus, Biblioteca Virtual em Saúde, Europe PMC and Web of Science databases. The inclusion criteria were cross-sectional, cohort and case-control studies on preoperative risk factors for AL (outcome). The Newcastle-Ottawa scale was used for bias assessment within studies. Meta-analysis involved the calculation of treatment effects for each individual study including odds ratio (OR), relative risk (RR) and 95% confidence intervals (95% CI) with construction of a random-effects model to evaluate the impact of each variable on the outcome. Statistical significance was set at p<0.05. Results: cross-sectional studies were represented by 39 articles, cohort studies by 21 articles and case-control by 4 articles. Meta-analysis identified 14 main risk factors for AL in CRC patients after colectomy, namely male sex (RR=1.56; 95% CI=1.40-1.75), smoking (RR=1.48; 95% CI=1.30-1.69), alcohol consumption (RR=1.35; 95% CI=1.21-1.52), diabetes mellitus (RR=1.97; 95% CI=1.44-2.70), lung diseases (RR=2.14; 95% CI=1.21-3.78), chronic obstructive pulmonary disease (RR=1.10; 95% IC=1.04-1.16), coronary artery disease (RR=1.61; 95% CI=1.07-2.41), chronic kidney disease (RR=1.34; 95% CI=1.22-1.47), high ASA grades (RR=1.70; 95% CI=1.37-2.09), previous abdominal surgery (RR=1.30; 95% CI=1.04-1.64), CRC-related emergency surgery (RR=1.61; 95% CI=1.26-2.07), neoadjuvant chemotherapy (RR=2.16; 95% CI=1.17-4.02), radiotherapy (RR=2.36; 95% CI=1.33-4.19) and chemoradiotherapy (RR=1.58; 95% CI=1.06-2.35). Conclusions: important preoperative risk factors for colorectal AL in CRC patients have been identified based on best evidence-based research, and such knowledge should influence decisions regarding treatment.

carried out a systematic review and meta-analysis with the aim of determining the preoperative risk factors associated with the evolution of the condition in patients submitted to colectomy for CRC.

Protocol and registration
This systematic literature review formed part Table 3).

Eligibility criteria and selection of studies
Cross-sectional, cohort, case-control, and randomized controlled studies relating to the risk factors of AL in patients submitted to colectomy (right hemicolectomy, extended right hemicolectomy, high left segmental colectomy, left colectomy, sigmoid colectomy, subtotal colectomy, total colectomy, high anterior resection, low anterior resections, ultra-low anterior resection) for CRC were considered eligible for inclusion.
The exclusion criteria were studies published in languages other than English, Portuguese, Spanish or French, literature reviews, systematic reviews, metaanalyses, studies without original data, case reports, case series, animal studies, and grey literature.Figure 2. Fluxograma dos estudos incluídos.

Dias
Preoperative risk factors associated with anastomotic leakage after colectomy for colorectal cancer: a systematic review and meta-analysis  [9]   mean risk differences between patients with and without AL were calculated.Cochran's Q test and I 2 statistics were employed to quantify the variability (heterogeneity) among the results of the selected studies with the significance level set at p<0.10.The degree of heterogeneity was interpreted according to the range of I 2 as follows: 0 -40%, likely not important; 30 -60%, moderate; 50 -90%, significant; and 75 -100% substantial.In consideration of the considerable heterogeneity identified among the studies, meta-analyses were conducted using the random-effects model to evaluate the impact of each variable on the outcome.The statistical significance of the differences between groups in the univariate analysis was set at p<0.05.

Overall features of the studies
The bibliographic search resulted in 64 studies being selected for inclusion in the review (Figure 2;    suggested a moderate risk of bias, and 0 -3 a high risk of bias.In the present study, an average score of 6 or above was considered satisfactory 5 .

Statistical analysis
All analyses were performed using RevMan software version 5.

Age
Forty-six studies had analyzed age as a risk factor for AL, but only six 11,14,22,42,56,64 found a statistically significant difference (p<0.05) in advanced age (elderly patients) as a possible risk factor.In the meta-analysis, 3,727 patients were evaluated.Participants aged 60 years or less were compared to those aged over; an RR of 0.79 (95% CI: 0.58-1.08)was found, and only one 22 of the studies showed a statistically significant difference.Thus, it is understood that age is not a risk factor for AL.

Sex
Among the 53 studies that analyzed sex as a possible predictor of outcome, 22 revealed significant Dias Preoperative risk factors associated with anastomotic leakage after colectomy for colorectal cancer: a systematic review and meta-analysis showed a statistically significant difference.Out of the 34 cross-sectional studies, only 13 established higher risk of AL in male patients.Since one article 21 separated data on colonic (sigmoid) from rectal surgeries, its inclusion in the systematic review had been done considering this division.

Dias
Preoperative risk factors associated with anastomotic leakage after colectomy for colorectal cancer: a systematic review and meta-analysis

Smoking habits
Fourteen studies included in the literature survey evaluated smoking as a risk factor for AL and, of these, nine 3,11,13,20,26,31,42,51,55 showed significant differences between smokers and non-smokers.
Smoking was assessed in nine cross-sectional studies involving 6,268 patients, but only two studies 13,31 reported a statistical difference.A further five cohort studies addressed tobaccoism and two 51,55 showed significant differences between the groups.

Alcohol consumption
Thirteen studies included in the systematic review investigated alcoholism as a risk factor for AL although a significant difference between alcohol users and non-users was observed in only one cohort study 51 .
The consumption of alcohol was examined by nine cross-sectional studies in the meta-analysis, and in this assessment a statistical difference was observed in only one study 15 .

Tumor location
Regarding the tumor site, 15 articles evaluated this variable in relation to the occurrence of AL.Of this sample, nine studies 6,9,10,14,16,18,44,46,60 showed a statistically significant difference (p<0.05).Of these, four studies 10,18,44,46 suggested cancer located in the middle and lower rectum to be an independent risk factor for the occurrence of AL.
This meta-analysis analyzed the tumor site by comparing tumors on the right side (cecum, right colon, hepatic flexure and transverse colon) to the left side (splenic flexure, left colon and sigmoid).
As the tumor location is a variable reported in the literature as an important risk factor for AL, a metaanalysis was performed comparing the left and right sides even though only three transversals studies 6,8,9 have contemplated this variable in a feasible way.The meta-analysis of these three studies, which analyzed a total of 20,277 patients, did not show a statistically significant difference; the RR found was 0.89 (95% CI: 0.74, -1.07).
Tumors in the upper rectum were considered as those located 5cm above the anal margin; Tumors below 5cm were considered to be in the lower rectum.
Four cross-sectional studies were accepted for the metaanalysis, two of which had a statistically significant difference 10,14 .However, the result of the meta-analysis showed RR 1.73 (95% CI: 0.95-2.03),not conceiving the tumor site in the lower or upper rectum as a risk factor for AL.

Diabetes mellitus
Twenty-five studies included in the review investigated diabetes as a risk factor for AL, of which six 11,13,29,41,42,49 established a significantly higher risk of the outcome in diabetic patients compared with their non-diabetic counterparts.Meta-analysis of 16 studies confirmed diabetes as a risk factor (Table 2; Figure 3).

Lung disease
Seven studies investigated lung disease as a risk factor for AL, three of which 9,13,43 demonstrated significant differences between patients with and without the condition.Meta-analysis of five cross-sectional studies revealed a significant association between lung disease and increased risk of the outcome (Table 2).

COPD
Six studies reported statistical differences between patients with and without COPD but in only one case-control study 63 was the difference significant.In the Dias Preoperative risk factors associated with anastomotic leakage after colectomy for colorectal cancer: a systematic review and meta-analysis meta-analysis of six cross-sectional studies that examined COPD as a risk factor for AL, one study 15 presented a weighting of 98.9% by virtue of the large number of participants (72,055) involved (Table 2).

CAD
Eight studies compared individuals with and without CAD, and four 2,7,20,63 reported significant differences between the two groups regarding the evolution of AL.Of the six cross-sectional studies (Table 2) included in the meta-analysis, only one 2 was able to demonstrate the association between CAD and increased risk of AL.

CKD
The possibility of CKD as a risk factor for AL was investigated in six studies but no significant differences between patients with and without the disease were found in the univariate analysis of these reports.However, metaanalysis of four cross-sectional studies 8,15,16,21 revealed that CKD was a predictor of the outcome (Table 2).

Previous abdominal surgery
Among the 13 studies comparing the development of AL in individuals who had or had not been submitted to abdominal surgery prior to colectomy for CRC, none showed statistical differences between the groups.Of the four cohort studies submitted to metaanalysis (Table 2), only one 56 demonstrated a significant association between previous abdominal surgery and higher risk of AL.

CRC-related surgical emergency
Twelve studies investigated CRC-related surgical emergency as a risk factor for AL and, of these, five studies 6,8,9,57,63 demonstrated statistical differences between individuals that had required this procedure and those that had not.Meta-analysis of five cross-sectional studies 6,8,9, 13,19 established a significant association between CRC-related surgical emergency and higher risk of the outcome (Table 2; Figure 5).

Neoadjuvant chemotherapy
Among the 13 studies that investigated neoadjuvant chemotherapy as predictor of AL, three 23,34,63 demonstrated statistical differences between individuals that had been submitted to the treatment and those that had not.Meta-analysis of five studies 44,55-58 confirmed neoadjuvant chemotherapy as a risk factor.(Table 2; Figure 6).

Dias
Preoperative risk factors associated with anastomotic leakage after colectomy for colorectal cancer: a systematic review and meta-analysis

Neoadjuvant radiotherapy
Three 30,47,57 of the 10 studies that investigated neoadjuvant radiotherapy as a risk factor for AL demonstrated statistical differences between individuals that had been submitted to the treatment and those that had not.Of the four cohort studies considered in the meta-analysis (Table 2; Figure 7), three 47,56,57 reported significant associations between radiotherapy and higher risk of AL.Several studies have demonstrated significant differences between males and females regarding the risk of AL and our meta-analysis confirmed that males exhibit a higher risk than females.Since men have a narrower pelvis, dissection of the tissues is more difficult and may cause postoperative complications.Furthermore, hormonal differences may influence intestinal microcirculation and, consequently, healing of the anastomosis 68 .
Our meta-analysis confirmed that smoking is a significant predictor of AL in CRC patients.Adequate tissue perfusion is essential for healing, and this seems to be particularly relevant for surgeries involving low rectal anastomosis 69 .The association between smoking and AL may be explained by four possible mechanisms, namely nicotine-induced vasoconstriction, cellular hypoxia caused by carbon monoxide, tissue hypoxia resulting in decreased collagen deposition and increased platelet adhesion and aggregation 70 .
A multicentre study 51 has demonstrated that alcohol abuse is an independent risk factor for the evolution of AL (OR = 1.63; 95%CI = 1.23-2.15;p=0.001).Individuals who consume more than 35 drinks per week have a significantly higher risk of developing AL compared with those who abstain from alcoholic consumption.The probable causes are subclinical heart failure, immunosuppression and low hemostatic function 70 , all of which impair wound healing.However, the negative effect of alcohol on wound healing has yet to be proven 51 .
The meta-analysis did not verified tumor height as a risk factor for AL.Presumably, this occurred because many studies had shown divergences regarding

Thirteen
studies investigated chemoradiotherapy as a risk factor for AL, and four of these 2,28,56,57 demonstrated statistical differences between individuals that had been submitted to the treatment and those that had not.Of the ten cross-sectional studies considered in the meta-analysis (Table 2), two 2,27 reported significant association between chemoradiotherapy and higher risk of AL.

DISCUSSION
The physiopathological reasons for increased risk of AL in patients submitted to surgical treatment of CRC have yet to be elucidated, although it is recognized that this life-threatening condition is multifactorial 4 .The literature review presented herein disclosed 16 potential risk factors for AL, and 14 of these were confirmed in the subsequent meta-analysis.
Advanced age is no longer considered a contraindication for CRC surgery 67 and our metaanalysis verified that older adults presented no increased risk of developing AL.Thus, it is accepted that primary the tumor height classification, which prevents the aggregation of these data in a meta-analysis.In addition, few studies have found this variable as an independent risk factor for the occurrence of AL, probably due to the lack of specific studies in this area.More studies are needed to elucidate this subject.
A meta-analysis performed by Rojas-Machado et al. 71 showed that diabetes mellitus is a risk factor (OR = 1.60; 95%CI = 1.12 -2.13) for AL, as confirmed by the results of our study.However, the association between the disorder and AL remains controversial since a large prospective study was unable to demonstrate that the presence of diabetes increased the rate of AL 72 .Nevertheless, the mortality rate among diabetic patients who developed AL was more than four-fold higher in comparison with their non-diabetic counterparts.
The impact of CAD on the development of anastomotic leak is unclear.An early study performed by Fawcett et al. 73  A number of retrospective studies 27,47,61,74,75 have established that neoadjuvant radiotherapy with or without concurrent chemotherapy is a strong predictor of AL, a finding that is confirmed by the meta-analyses presented herein.However, according to Park et al. 75 , while chemoradiotherapy was a risk factor for AL in a subgroup of patients who did not receive a protective stoma after low rectal anterior resections for CRC, when all the patients submitted to the surgery were analyzed together, chemoradiotherapy did not appear as a risk factor.In a systematic review, McDermott et al. 68 reported increased rates of AL and mortality in patients One limitation of our study was that some risk factors and their associations with the development of AL have been poorly investigated in patients colectomized for treatment of CRC.Additionally, some potential predictors studied by other authors had to be excluded from the meta-analysis since the variables could not be aggregated owing to heterogeneity between, and lack of standardization of, the same risk factors.

Dias
Preoperative risk factors associated with anastomotic leakage after colectomy for colorectal cancer: a systematic review and meta-analysis of a study submitted to and approved by the Research Ethics Committee of Santa Casa de Belo Horizonte under the protocol CAAE 36476320.2.0000.5138.The Ethics Committee waived the requirement of written informed consent since the systematic review and metaanalysis were based entirely on data published in the literature.The review was carried out in accordance with the checklist included in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) protocol and was registered on the International Prospective Register of Systematic Reviews (PROSPERO) database (https://www.crd.york.ac.uk/prospero) under protocol CRD42020219325.Information sources and search strategy The bibliographic search, which covered the 15 year and nine months period from 1st January 2005 to 19th October 2020, was performed on 17th November 2020 and updated on 1st December 2021 without new entries.Studies were retrieved from the PubMed, Cochrane Library, Scopus, Biblioteca Virtual em Saúde, Europe PMC and Web of Science databases using combinations of DeCS and MeSH descriptors (Figure 1; Relevant publications were selected by two researchers (VED and LVP) on the basis of the eligibility criteria by reading the titles and, subsequently, the detailed abstracts.Duplicate studies were eliminated and the full texts of the remaining articles were analyzed to select the studies to be included in this review.In cases of disagreement about the inclusion of a publication, consensus was attained by discussion or through mediation with the help of a third researcher (LBRG).The PRISMA 2020 flow chart shown in Figure 2 outlines the phases of the selection of studies.
anastomosis may be performed on senior patients without exposing them to increased risk of AL provided that they do not exhibit other comorbidities.It is likely that the improvement in operative techniques over the years, the spread of laparoscopy and the greater preand intraoperative care of elderly patients justify this result.Besides, preoperative selection of elderly patients may have resulted in the selection of the most apposite ones.The surgeons are probably more careful while choosing elderly patients for surgery treatment.These facts could explain why age isn't considered as a risk factor for AL.
who underwent colon anastomosis for chronic radiation enteritis, suggesting that the operating surgeon should consider a previous history of irradiation in order to assess whether anastomosis is safe.In view of the data presented in that review and the results of our metaanalysis, surgeons should seriously consider performing a protective stoma during anastomosis in CRC patients undergoing neoadjuvant chemoradiotherapy, radiotherapy and chemotherapy.Our study highlights the importance of further research in two main areas: (i) elucidation of the histo-and patho-physiological basis of AL in order to properly define and easily recognize this potentially fatal condition; (ii) understanding the underlying mechanisms by which the various risk factors influence the evolution of AL so that an optimal number of factors could be selected and ranked to facilitate decision-making about the management of risk development of the condition.

Table 1 -
Articles included in the systematic review and their respective Newcastle-Ottawa Scale (NOS) quality assessment scores.

Table
DiasPreoperative risk factors associated with anastomotic leakage after colectomy for colorectal cancer: a systematic review and meta-analysis 4 (Cochrane, London, UK).Treatment effects, including odds ratio (OR), relative risks (RR) and 95% confidence intervals (95%CI), were computed for each selected study and, in cases where mean values and standard deviations for a given risk factor were provided,