Risk of Anastomotic leakage with

non-steroidal anti-inflammatory

drugs in colorectal surgery

Shahid Sadoughi university of medical science

British Journal of Surgery

2012; May

Background

With the implementation of multimodal analgesia regimens in fast-track surgery programmes, non-steroidal anti-inflammatory drugs (NSAIDs) are being prescribed routinely. However, doubts have been raised concerning the safety of NSAIDs in terms of anastomotic healing.

Methods• Data on patients who had undergone

primary colorectal anastomosis at two teaching hospitals between January 2008 and December 2010 were analysed retrospectively. Exact use of NSAIDs was recorded. Rates of anastomotic leakage were compared between groups and corrected for known risk factors in both univariable and multivariable analyses.

• A total of 795 patients were divided into four groups

according to NSAID use: no NSAIDs (471 patients),

use of non-selective NSAIDs (201), use of selective

cyclo-oxygenase (COX) 2 inhibitors (79), and use of

both selective and non-selective NSAIDs (44). The

overall leak rate was 9·9 per cent (10·0 per cent for

right colonic, 8·7 per cent for left colonic and 12·4 per

cent for rectal anastomoses).

• Known risk factors such as smoking and use of steroids were not significantly associated with anastomotic leakage. Stapled anastomosis was identified as an independent predictor of leakage in multivariable analysis (odds ratio (OR) 2·22, 95 per cent confidence interval 1·30 to 3·80; P = 0·003).

Methods

• Patients on NSAIDs had higher anastomotic leakage rates than those not on NSAIDs (13·2 versus 7·6 per cent; OR 1·84, 1·13 to 2·98; P = 0·010). This effect was mainly due to non-selective NSAIDs (14·5 per cent; OR 2·13, 1·24 to 3·65; P = 0·006), not selective COX-2 inhibitors (9 per cent; OR 1·16, 0·49 to 2·75; P = 0·741). The overall mortality rate was 4·2 per cent, with no significant difference between groups (P = 0·438).

Conclusion:

• Non-selective NSAIDs may be associated with anastomotic leakage. Copyright © 2012 British Journal of Surgery Society Ltd. Published by John Wiley & Sons, Ltd.

• Anastomotic leakage is a severe complication after colorectal surgery. Peritonitis and septicaemia lead to reoperations, admission to the intensive care unit and a profoundly increased mortality rate1. Furthermore, anastomotic leakage is a risk factor for local recurrence of colorectal cancer, and has a significant impact on disease-free and overall survival

Introduction

• Several risk factors for anastomotic leakage have been identified, including patient characteristics, neoadjuvant chemoradiation therapy and surgery-related factors

Introduction

• Recently, non-steroidal anti-inflammatory drugs (NSAIDs) have been suggested as a potential factor in the development of anastomotic leakage8–15. NSAIDs are increasingly being used in fast-track surgery programmes to optimize pain control and to reduce opioid usage16,

Introduction

Introduction

• Addition of NSAIDs to morphine is estimated to have an opioid-sparing effect of 30 per cent, with a subsequent decrease in morphine-related side-effects

• However, inhibition of cyclo-oxygenase (COX) interferes with normal processes important for wound repair and healing of intestinal anastomoses20. Experimental studies have shown inferior anastomotic healing following treatment with NSAIDs

Introduction :

Introduction

• Two retrospective clinical studies reported a potential detrimental effect of NSAIDs on anastomotic leakage14, 15. However, patient numbers in these studies were small and the contribution of NSAIDs alone was not clearly defined. The present study was designed to investigate the role of both selective and non-selective COX inhibitors in anastomotic leakage following colorectal surgery.

Methods

• In this retrospective study all patients undergoing colorectal surgery in the Orbis Medical Centre, Sittard, and the Atrium Medical Centre, Heerlen, between January 2008 and December 2010 were registered in a database to investigate the hypothetical relationship between NSAIDs and anastomotic leakage. This study was approved by a medical ethics committee for both hospitals. Indications for colorectal surgery included malignancy and benign diseases. Surgery was always performed by, or under the supervision of, one of seven experienced colorectal surgeons.

Methods• All patients were treated according to a fast-

track surgery protocol17 implemented in both hospitals since 2007. Patients routinely received epidural anaesthesia combined with paracetamol. The use of NSAIDs and/or selective COX-2 inhibitors was not standardized. These were prescribed according to the preference of the attending physician, in the absence of contraindications. The most prescribed non-selective NSAID was diclofenac

Methods

• Meloxicam and celecoxib were the most commonly used selective COX-2 inhibitors. No changes in surgical teams or postoperative protocols were implemented during the study period.

• Patient characteristics, operative procedures, post- operative care and complications were extracted from the database. Co-morbidities such as cardiovascular disease (myocardial infarction, arrhythmia, hypertension), pulmonary disease (chronic obstructive pulmonary disease) and diabetes mellitus were registered only when the patient was currently being treated for the condition by a specialist or general physician. Malignancy was registered when a colorectal malignancy was the indication for surgery.

Methods

Methods

• The study population was divided into groups according to the use of NSAIDs. A distinction was made between non-selective NSAIDs and selective COX-2 inhibitors. Four groups were identified: no use of NSAIDs, use of non-selective NSAIDs, use of selective COX-2 inhibitors, and consecutive or concomitant use of non-selective NSAIDs and selective COX-2 inhibitors. Only use within the first 5 days after surgery was included in the analysis.

• Complications were assessed according to the Dindo–Clavien classification of surgical complications21. The primary endpoint was anastomotic leakage defined as clinical and radiological signs of anastomotic leakage as confirmed by reoperation or occurrence of an enterocutaneous fistula. Sensitivity analysis was also performed, including abscess formation in the quadrant of the anastomosis requiring radiological drainage.

Methods

Statistical analysis

• Baseline characteristics and clinical outcomes of the study groups were compared using one-way ANOVA or Kruskal–Wallis test for continuous variables (depending on data distribution) and χ2 test for categorical variables. The incidence of anastomotic leakage after colorectal surgery in the various groups was compared by means of multivariable logistic regression models using the Wald test for statistical significance. Clinically relevant variables were included as co-variables in the models, whereas adjustment for other baseline characteristics and interaction terms was done in a two-step approach

• Final models were obtained by stepwise removal of interaction terms that were not statistically significantly associated with anastomotic leakage. Assumptions of the logistic regression models were checked with addition of statistically significant quadratic terms. Multicolinearity and influential outliers were checked using variance inflation factors and Cook's distances. Odds ratios (ORs) were calculated with 95 per cent confidence intervals (c.i.). P < 0·050 was considered statistically significant. Descriptive analyses were made using SPSS® version 16.0 statistical software (SPSS, Chicago, Illinois, USA).

Statistical analysis

Results

• A total of 795 patients were included in the study. Of these, 324 (40·8 per cent) used NSAIDs in the immediate postoperative period; 201 (25·3 per cent of all patients) used only non-selective NSAIDs, 79 (9·9 per cent) used only selective COX-2 inhibitors, and 44 patients (5·5 per cent) used both. In this combination group 41 patients used both classes of NSAID concomitantly after changing from recovery to general surgical wards. In the remaining patients the chronic use of celecoxib was overlooked as patients also received diclofenac.

Results

• Patients who took NSAIDs used them for a mean(s.d.) of 2·8(1·3) days in the first 5 days after operation, starting 0·9(1·0) days after surgery. Only 25 patients started NSAID use more than 3 days after operation, of whom five first took them from the fourth day.

Results

• Patient and perioperative characteristics, including known risk factors for anastomotic leakage, are compared between groups in Tables1 and 2 . There were no significant differences in patient characteristics between the four groups except for a higher frequency of pulmonary disease in the combination group (Table1).

Results• In univariable analysis pulmonary

disease seemed to be a risk factor for anastomotic leakage (OR 1·91, 95 per cent c.i. 1·04 to 3·52; P = 0·043), but multivariable analysis showed no significance (OR 1·53, 0·78 to 2·99; P = 0·214) (Table3). Other known risk factors such as smoking and use of steroids were not significantly associated with anastomotic leakage in univariable analysis (Table3).

Results

• Perioperative characteristics and predefined clinically relevant variables for anastomotic leakage were no different between groups, except for greater use of laparoscopic techniques and epidural analgesia in the combined NSAIDs/COX-2 group (Table2). However, in univariable and multivariable analyses neither factor was associated with development of clinically significant anastomotic leakage (Table3).

Results

• In univariable analysis three factors were significantly associated with a higher incidence of anastomotic leakage: pulmonary disease, NSAID use (total use of non-selective NSAIDs and/or selective COX-2 inhibitors) and stapled anastomoses (Table3). In multivariable logistic regression analysis both NSAID use and stapled anastomoses remained independent risk factors for anastomotic leakage (Table3).

Results

• Seventy-nine (9·9 per cent) of 795 patients had clinically relevant anastomotic leakage, 12·4 per cent with rectal anastomoses, 8·7 per cent after left colonic surgery and 10·0 per cent after right colonic surgery, with no significant difference between types of surgery. The occurrence of anastomotic leakage and abscess formation in the quadrant of the anastomosis was 11·7 per cent.

Results

• The anastomotic leak rate was 7·6 per cent among patients not using NSAIDs. A significant increase in anastomotic leakage was found in patients using any NSAID (13·2 per cent; OR 1·84, 1·13 to 2·98; P = 0·010). Stratified for non-selective NSAIDs and selective COX-2 inhibitors, the most profound effect on anastomotic leakage was found in the group that used non-selective NSAIDs (14·5 per cent; OR 2·13, 1·24 to 3·65; P = 0·006) (Fig.1).

Results

• Anastomotic leakage was no higher in patients who received selective COX-2 inhibitors than in those who did not use NSAIDs (9 versus 7·6 per cent; OR 1·16, 0·49 to 2·75; P = 0·741). Anastomotic leakage rates were highest in the combined group, but this was not statistically significant (16 per cent; OR 1·86, 0·75 to 4·61; P = 0·181).

Results

• Figure 1. Rates of anastomotic leakage in relation to use of non-steroidal anti-inflammatory drugs (NSAIDs): no NSAIDs (471 patients), non-selective NSAIDs (201), selective cyclo-oxygenase (COX) 2 inhibitors (79), and both selective and non-selective NSAIDs (44). *P = 0·006 versus no NSAIDs (multivariable logistic regression model using the Wald test for statistical significance

Results

• Sensitivity analysis including both anastomotic leakage and abscess formation in the quadrant of the anastomosis produced similar results. The combined rate of leakage and abscess among patients who used any NSAID was significantly greater than that in patients who did not use NSAIDs (14·8 versus 9·6 per cent; OR 1·71, 1·06 to 2·77; P = 0·033), again with the highest percentage among those who took non-selective NSAIDs (15·0 per cent).

Results• A significant relationship was found between

the duration of NSAID use within the first 5 days after surgery and anastomotic leakage. Use of any NSAID for 3 days or more was associated with a higher rate of anastomotic leak than use for only 1 or 2 days (16·6 versus 10·0 per cent respectively; OR per day of NSAID use 1·24, 1·08 to 1·43; P = 0·003) (Fig.2). Selective COX-2 inhibitors were used for a mean of 2·3 days within the first 5 days (total duration 4·3 days), compared with 2·9 (total duration 5·3) days for non-selective NSAIDs.

• There was no difference in complications, as recorded according to the Dindo–Clavien classification21, between treatment groups (Table4). The overall mortality rate was 4·2 per cent, with no significant difference between groups (P = 0·438).

Discussion

• Reported anastomotic leak rates vary considerably and range from 1 to 30 per cent owing to lack of a standard definition22. When both anastomotic leakage and abscess formation in the quadrant of the anastomosis were included, the leak rate in the present study was 11·7 per cent, in accordance with the Dutch Surgical Colorectal Audit register of 201023.

• The significantly higher rate of anastomotic leakage in patients using NSAIDs is in line with previous findings14, 15. Anastomotic leak rates were highest in the combined group, but the small group size limits interpretation of this finding.

• The mechanism by which NSAIDs may increase anastomotic leakage is still unclear. Inhibition of COX by NSAIDs affects leucocyte function, induces apoptosis and decreases crypt survival. It also reduces production of vascular endothelial growth factor and angiogenesis, and interferes with collagen formation and cross-linking24–29. As this study was not designed to investigate the mode of action of NSAIDs, the contribution of these factors remains unclear.

• In contrast to the profound detrimental effect of selective COX-2 inhibitors described by Holte and colleagues14, no significant impact of selective NSAIDs was found in the present investigation. In this study, selective COX-2 inhibitors were used for a mean of 2·3 days within the first 5 days (total duration 4·3 days), compared with 2·9 (total duration 5·3) days for non-selective NSAIDs. This difference in duration of use may provide an explanation for the observed effects. In multivariable analysis, each day of NSAID use increased the risk of anastomotic leakage by 24 per cent.

• An unexpected finding was the significantly higher anastomotic leakage rate when a stapling device was used rather than a fully handsewn anastomosis. As this study was not designed to investigate this as an endpoint, the result must be interpreted with caution. There was no influence of NSAIDs on leakage from stapled versus handsewn anastomoses.

• The main limitation of this study is its non-randomized and observational nature. Possible confounding may have occurred when patients received more NSAIDs because of higher pain levels caused by greater surgical trauma, which in itself may have led to an increased risk of anastomotic leakage. Some patients received NSAIDs because of failure of the epidural anaesthesia or allergic/hypersensitive reactions to opioid analgesics.

• Interestingly, the duration of morphine use was similar in all groups, with or without NSAID use. NSAIDs are increasingly being prescribed after colorectal surgery to reduce morphine-related effects on gastrointestinal motility and hasten recovery. However, the evidence supporting this practice is very limited and often difficult to quantify18, 19.