Validation of an endoscopic anastomotic grading score as an intraoperative method for assessing stapled rectal anastomoses

Anastomotic leak is a dreaded complication of colorectal surgery. An endoscopic grading score of the perianastomotic mucosa has been previously developed at our institution (UCI) to assess colorectal anastomotic integrity. The objective of this study is to validate the UCI anastomotic score and determine its impact in anastomotic failure. As a follow-up study of the UCI grading score implementation during 2011 to 2014, patients undergoing stapled colorectal anastomoses after sigmoidectomy or proctectomy at a single institution from 2015 to 2018 were retrospectively reviewed. Patients were grouped into three tiers based on endoscopic appearance (grade 1, circumferentially normal mucosa; grade 2, ischemia/congestion < 30% of circumference; grade 3, ischemia/congestion > 30% of circumference). On the basis of endoscopic mucosal evaluation, grade 1 anastomosis was observed in 299 patients (94%), grade 2 anastomosis in 14 patients (4.4%), and grade 3 anastomosis in 5 patients (1.6%). All grade 3 classifications were immediately and successfully revised intraoperatively with reclassification as a grade 1 anastomosis. The anastomotic leak rate of the follow-up study period from 2015 to 2018 was 6.4% which was lower compared to the anastomotic leak rate of 12.2% in the original study period from 2011 to 2014 (p = 0.07). Anastomotic leak rate for the entire patient series was 8.5%. A grade 2 anastomosis was associated with higher anastomotic leak rate compared to a grade 1 anastomosis (35.7% vs. 7.4%, p < 0.05). None of the five grade 3 anastomoses resulted in an anastomotic leak upon revision. This study further validates the anastomotic grading score and suggests that its systematic implementation can result in a reduction in anastomotic leaks.

A plethora of intraoperative methods for detecting AL have been proposed and tested, including mechanical air leak testing, microperfusion evaluation, and endoscopic visualization [8][9][10].At our institution, we have developed and implemented an intraoperative endoscopic visualization method for assessing anastomotic integrity and estimating the risk of AL [11].We have previously reported our findings in a retrospective cohort study including 106 newly constructed colorectal anastomoses, each of which was stratified into three tiers of mucosal grades, based on the level of ischemia and congestion visible in the perianastomotic mucosa [11].Multivariate logistic regression analysis identified that increasing grade tiers, consistent with worsening ischemia, trauma, and/or congestion of the perianastomotic mucosa, was associated with increased risk of AL [11].
Limitations of our prior study included the relatively small sample size, which potentially limited the overall power of the study, as well as the novel application of the endoscopic visualization method [11].As such, the objective of this study was to perform a follow-up study for validation of the University of California, Irvine (UCI) grading score over the next 3 years following its systemic implementation during 2011-2014.We hypothesize that systematic implementation of the UCI grading score can assist surgeons in predicting the risk and decreasing the rate of AL following rectal anastomoses.

Methods
As a follow-up study from the prior prospective application of the UCI grading score from 2011 to 2014 [11], patients undergoing a stapled colorectal anastomosis from 2015 to 2018 at UCI Medical Center were reviewed.Consecutive patients were identified after accessing an all-inclusive UCI internal database that contained all patients who had undergone minimally invasive or open low anterior resections, sigmoidectomies, or colostomy reversal by six colorectal surgeons over the study period.As per the original study, patient clinical data was retrospectively reviewed from the day of surgery with 60 days postoperative follow-up.Exclusion criteria included patients under the age of 18 or patients with incomplete data, including undocumented UCI grading score, during chart review.Approval for this study was acquired through the institutional review board at UCI Medical Center.
As described in the original study protocol, the colorectal surgeons at our institution continued the same method in their operative approaches, with medial to lateral dissection of the mesocolon and splenic flexure mobilization [11].Dissection included high ligation of the inferior mesenteric artery as well as ligation of the inferior mesenteric vein at the inferior edge of the pancreas adjacent to the ligament of Treitz [11].Medial to lateral dissection of the mesocolon and splenic flexure mobilization was performed laparoscopically, while robotic dissection was reserved for total mesorectal excision in the pelvis.During proximal transection of the colon, the operating surgeons assessed the cut mucosal edges for brisk bleeding to ensure that there was adequate perfusion prior to construction of the colorectal anastomosis.
Patients were classified into three tiers based on flexible endoscopic score, which had been previously developed and implemented at UCI.As described in the original study, this grading system was based on the degree of perianastomotic mucosal ischemia or venous congestion, each equally treated as indicators of compromised perfusion to the anastomosis [11].Grade 1 was defined as circumferentially normal mucosa on either side of the colorectal anastomosis.Grade 2 was defined as mucosal ischemia or congestion less than 30% of the circumference on either side of the colorectal anastomosis.Grade 3 was defined as mucosal ischemia or congestion greater than 30% of the circumference on either side of the colorectal anastomosis or any degree of ischemia/ congestion on both sides of the anastomosis (Fig. 1) [11].Scoring of the colorectal anastomosis was performed immediately at the time of the colorectal anastomosis construction by flexible high definition (HD) white-light sigmoidoscopy.The grade of each anastomosis was determined semiobjectively by each operating surgeon, each of whom was either involved with the initial development and implementation of the grading system or thoroughly trained in its use.On the basis of the grading score, decisions for intraoperative modification of the anastomosis were left to the clinical discretion of the operating surgeon.
Patients were stratified into groups based on endoscopic grade, which were grade 1 and grade 2 for comparison.All grade 3 anastomoses had been successfully revised immediately with reclassification as a grade 1 anastomosis.However, grade 3 anastomoses were not included in the grade 1 cohort for comparison given their immediate reconstruction.Data from the original study from 2011 to 2014 was combined with data acquired during the follow-up study period from 2015 to 2018 for final analysis.Cross-comparisons between study periods (2011-2014 versus 2015-2018) were also performed.Demographics and clinical outcomes were analyzed.Demographics included age, gender, body mass index (BMI), American Society of Anesthesiologists (ASA) Fig. 1 Illustration of three-tiered grades of the UCI endoscopic mucosal grading system.White arrows illustrate location of mucosal irregularity in grade 2 and grade 3 [11] score, diabetes, hypertension, smoking status, cardiovascular disease, diagnosis, and exposure to pelvic radiation.Clinical outcomes included AL, symptomatic leak, pelvic abscess, intraoperative revision, any postoperative intervention, operative time, estimated blood loss (EBL), use of drain, diverting ileostomy, and level of colorectal anastomosis.The primary outcome was AL.As in the original study, AL was defined as a disruption in the staple line that was identified via radiologic imaging (water soluble contrast enema, computed tomography imaging) or via endoscopic assessment.Symptomatic leak was defined as AL with clinical symptoms.Any endoscopic, percutaneous, or operative mediation in the management of an AL was defined as an intervention.
Pearson chi-square testing and Fisher's exact testing for categorical variables and unpaired Student's t test for continuous variables were used to perform the univariate analysis for demographic and clinical outcomes data.A multivariate logistic regression analysis adjusted to anastomotic level was also performed to assess clinical outcomes.Statistical significance was set at p < 0.05.Data analysis was carried out using Stata 16 software (StataCorp, 2019, Stata Statistical Software: Release 16, College Station, TX, USA).

Results
From 2011 to 2018, a total of 318 patients were identified to have undergone a stapled colorectal anastomosis with intraoperative flexible endoscopic evaluation and mucosal grading.Of these patients, 106 patients (33.3%) were evaluated from 2011 to 2014 during the original study period, and 212 patients (66.7%) were reviewed from 2015 to 2018 during the follow-up study period.On the basis of mucosal evaluation during routine intraoperative endoscopy after construction of the anastomosis, a grade 1 anastomosis was observed in 299 patients (94%), grade 2 anastomosis in 14 patients (4.4%), and grade 3 anastomosis in five patients (1.6%).A total of 5 patients were excluded during the follow-up study period because of incomplete documentation of UCI grade.
For comparison, demographics, comorbidities, and clinical characteristics are depicted in Table 1.The average age of the grade 1 anastomosis and grade 2 anastomosis cohorts was 57.1 and 56.1 years, respectively.Female gender was more common in the grade 1 anastomosis group compared to the grade 2 anastomosis group (51.2% vs. 14.2%, p = 0.02).There were no significant differences between the  2. Mean operative time and mean EBL were similar.Compared to the grade 1 anastomosis, the grade 2 anastomosis group had a significantly higher percentage of drain use vs. 49.2%,p = 0.03) and use of a diverting ileostomy (64.3% vs. 35.5%,p = 0.03).In terms of level of colorectal anastomosis, grade 1 anastomosis had a higher percentage of anastomoses measured greater than 10 cm from the anal verge (48.2%), while grade 2 had a higher percentage of anastomoses measured less than 5 cm from the anal verge (42.9%).Grade 2 anastomoses experienced a higher percentage of intraoperative revision compared to grade 1 anastomoses (21.4% vs. 1.0%, p < 0.01).Of the three patients with a grade 1 anastomosis that was intraoperatively revised, two patients underwent revision because of a positive mechanical air leak test, while the remaining patient underwent revision because of an incomplete anastomotic rectal doughnut.
Of the three patients with a grade 2 anastomosis that was intraoperatively revised, all three patients were found to be potentially ischemic on the basis of the endoscopic evaluation of the operating surgeon, with all three patients undergoing suture reinforcement.Of the five patients with grade 3 anastomosis, all underwent complete anastomosis takedown and were reconstructed.None of the patients that underwent intraoperative revision developed an AL.
Postoperative outcomes are illustrated in Table 3.The overall AL rate of the entire patient cohort was 8.5% (27/318).The AL rate for grade 1 anastomosis was 7.4% (22/299), and the AL rate for grade 2 anastomosis was 35.7% (5/14).No ALs occurred in patients with grade 3 anastomosis, which were revised into grade 1 anastomoses during the operation.Compared to the grade 1 anastomosis group, the grade 2 anastomosis group had a significantly higher percentage of patients experiencing postoperative AL (35.7% vs. 7.4%, p = 0.001) and symptomatic leak (21.4% vs. 4.7%, p = 0.007).There were no significant differences between the two groups in terms of pelvic abscess development.Compared to the grade 1 anastomosis group, the grade 2 anastomosis group had a higher percentage of patients undergoing intervention after surgery (21.4% vs. 6.4%, p = 0.03).
Compared to the original study period from 2011 to 2014, the follow-up study period from 2015 to 2018 had a lower clinical AL rate (6.4% vs. 12.2%, p = 0.07).The follow-up study period from 2015 to 2018 also had a lower rate of intraoperative revision (0.9% vs. 8.5%, p < 0.01), compared to the original study period from 2011 to 2014.

Discussion
In this updated study, we validated a mucosal grading system developed and implemented at our institution over the follow-up study period from 2015 to 2018.Using multivariate analysis, we confirm that grade 2 classification, which is presumably associated with worse vascular perfusion at the newly constructed anastomosis site, correlated with risk of AL.We found that the associated risk of clinical AL was higher in grade 2 compared to grade 1 classification, despite the grade 2 group having a statistically higher percentage of diverting stomas.Using this updated retrospective data set, we reinforce our previous conclusion that grade 2 or grade 3 classification should elicit consideration for anastomotic revision [11].
Our endoscopic mucosal grading system validates the use of routine intraoperative endoscopy for anastomotic evaluation.No real consensus for anastomotic inspection after anastomosis construction has yet been established in the surgical community [12].At the time of the implementation of our grading system from 2011 to 2014, methods for leak identification, such as intraoperative endoscopy, had not been shown to make a statistically significant difference in AL complications [8,13,14].However, since implementation of our grading system, additional studies have demonstrated the effectiveness of routine intraoperative flexible endoscopy [12,15].In a prospective series of 415 consecutive patients undergoing colorectal or ileoanal  after performing intraoperative flexible sigmoidoscopy on all patients, concluding that intraoperative flexible sigmoidoscopy should be performed routinely [15].Additionally, in a meta-analysis assessing the impact of endoscopy on AL, Aly et al. found that patients undergoing intraoperative endoscopy had a significantly reduced rate of AL compared to conventional methods [12].In our study, we found that intraoperative examination of the newly constructed anastomosis was simple, safe, and feasible, as there were no associated complications from intraoperative inspection.We also confirmed that grade 2 classification was associated with increased risk of AL, by almost 6.99-fold risk, compared to grade 1 classification.This represented an increase by almost twofold from the prior study from 2011 to 2014, where we initially reported 4.09-fold associated risk increase of AL [11].Grade 2 classifications were also associated with increased risk of AL despite the grade 2 classifications having a higher percentage of fecal diversion by 28.8% compared to grade 1 classification.Our new analysis added an additional sample size of 212 patients over a 3-year period that provided additional power, with consistent implementation of the grading system by the operating surgeons.While there was a large amount of grade 1 anastomotic classifications (299 patients), we were able to identify four additional grade 2 classifications and one additional grade 3 classification.This likely occurred because our operating surgeons were aware of the clinical implications of higher AL rates associated with a grade 2 or grade 3 classification.Therefore, they likely spent more time controlling for various technical risk factors associated with AL, such as anastomotic tension or vascular supply, to avoid classification as a grade 2 or grade 3 [6,16,17].
Our study supports the systematic implementation of an endoscopic mucosal grading system, which can potentially be associated with a lower AL rate.In comparing study periods, we found that our follow-up study from 2015 to 2018 had a lower leak rate of 6.4% compared to the leak rate of 12.2% in the original study from 2011 to 2014.Although not statistically significant, this reduction represents a clinical improvement in the AL rate.This decrease in the AL rate may have occurred with our risk stratification from the mucosal grading system for the colorectal anastomosis, which is known to be a high risk anastomosis for leakage [4,18,19].Risk stratification provided by the mucosal grading system recognizes the underlying pathophysiology of anastomotic viability, where mucosal ischemia/congestion can contribute to lack of anastomotic healing and increased risk of leak [6,17,19].However, this grading system reflects the spectrum of ischemic changes/congestion that occur with a newly constructed anastomosis, which may not be readily seen prior to anastomosis construction without use of a high-definition endoscope.In our follow-up study period, we revised fewer anastomoses, likely because of increased experience with AL prevention techniques [11,20].
To our knowledge, our mucosal grading system has not been widely implemented at other institutions, although other endoscopic visualization methods have been described in the literature.Kryzauskas et al. describe an endoscopic visualization method where they perform simultaneous transabdominal and transanal indocyanine green (ICG) fluorescence imaging [21].Although they do not report actual implementation of our technique, they postulate that adding ICG imaging to our described mucosal endoscopic evaluation method should theoretically optimize the grading system's ability to predict leak [21].In a case series of 52 patients undergoing construction of a stapled colorectal anastomosis, Vallicelli et al. used the same intraoperative endoscopic evaluation method with ICG imaging of the colorectal anastomosis [22].While ICG imaging is useful for anastomotic evaluation, our endoscopic mucosal grading system is simpler and less costly to perform without the need for adjunctive medication, while still offering the benefit of a semiobjective assessment of the anastomosis.Given the importance of perianastomotic evaluation, the movement towards direct anastomotic inspection is occurring.
Patient demographics, such as gender and smoking status, were also noteworthy in our follow-up analysis of the mucosal grading system.Numerous previous studies have shown that female gender is associated with a lower AL rate [3,23,24].Previously, it has been postulated that this is due to a wider gynecoid female pelvis versus the more narrow android male pelvis, which makes for a more difficult dissection [3,25].On univariate analysis, our study showed that gender differences were identifiable at the actual time of anastomotic construction, as female individuals were less likely to have a grade 2 anastomosis compared to male ones.However, upon multivariate regression analysis, gender was not ultimately identified as an independent risk factor for AL.Additionally, smoking history was also found to be more commonly associated with a grade 2 anastomosis in comparison to a grade 1 anastomosis.In the literature, smoking has been shown to have a four times greater chance of AL, as smoking has an impact on wound healing through oxidative stress and mucosal perfusion [26].
Several limitations should be noted when considering our study.As a result of the retrospective design of our study, inherent biases, such as selection bias and misclassification, exist within the study.Small sample sizes, especially for grade 2 and grade 3 anastomotic classifications, may have continued to impact the overall power of the followup study, although this was improved with the addition of 212 more patients from the original study from 2011 to 2014.Although each operating surgeon was involved with or trained in the endoscopic mucosal grading system, the inherent subjectivity when grading a colorectal anastomosis continues to be present.

Conclusion
This study validates that implementation of a simple endoscopic grading score system can potentially lead to a reduction in ALs and quantifies the associated risk of AL with different level of ischemia/congestion.We confirmed that grade 2 and grade 3 classification, which were indicative of perfusion compromise to the anastomosis, were associated with increased risk of AL.Further studies are warranted to test this grading system at a national level.

Table 4
Comparison of demographics and perioperative character-

Table 5
Adjusted logistic regression analysis of key perioperative clinical outcomes and anastomotic leak risk factors a Subgroup logistic regression analysis of patients with < 5 cm anastomosis level compared to endoscopic grade score