Simon Yauw

The burden of anastomotic leakage (AL) has been well described in literature and includes higher morbidity and mortality, greater recurrence of cancer and increased costs (1, 2). This burden has not decreased significantly in the past decades (3, 4). An important reason is that leakage aetiology is still poorly understood (3). In this thesis we aimed to increase knowledge of (disturbed) healing of small- and large-bowel anastomoses. We investigated the detrimental effects of NSAIDs—common painkillers used in gastro-intestinal surgery—on ileal and right-colonic anastomoses.

Anastomotic healing physiology: a complex cascade of cellular reactions and inflammatory mediators – Chapter 2
A literature review was conducted to provide an overview of relevant cellular interactions occurring during the anastomotic healing cascade. Knowledge of anastomotic healing was combined with knowledge of other types of wound healing (e.g. skin, bone and intestinal ulcers) and general cellular physiology.
The healing of multilayer intestinal tissue is highly complex and requires an interplay among infiltrating platelets, mast cells, granulocytes, macrophages, dendritic cells, lymphocytes, fibroblasts with resident inflammatory cells, mesenchymal stem cells, neurons, endothelial, epithelial, mesothelial and smooth muscle cells. It is essential for optimal tissue repair that all cell types exert the right action at the right time. Inflammatory activity, involving proteolytic processes, is necessary for host defence and tissue reorganisation. The subsequent cessation of inflammation and timely transition to proliferation is a physiologically challenging process. Biological substances, e.g. bile, bacteria and inflammatory products, and mechanical forces such as peristalsis and intraluminal pressure interact with the fresh anastomosis from the beginning and greatly challenge the balance and order of cellular reactions. When certain drugs or risk factors additionally disturb cellular and tissue processes, the anastomosis becomes at risk for leakage.
Cellular and molecular mechanisms might be potential therapeutic targets, e.g. growth factors or hormones, but the complexity and interaction of the different systems impedes the development of such therapeutic strategies.

Review of experimental research on anastomotic leakage: reporting and quality of animal studies must improve – Chapter 3
Numerous animal studies on intestinal anastomosis have been conducted with regard to different aspects of healing and leakage. However, the clinical impact of animal data on diagnosis, prevention and treatment of AL is unknown. Therefore, and in view of the current societal call to replace, reduce and refine animal experiments, we conducted a systematic review to examine the quality of animal research related to anastomotic healing and leakage. A total of 1,342 animal studies on AL were identified, with exponential publication rates in recent years. Of 350 articles studying experimental therapies, 298 (85%) reported a positive effect on anastomotic healing. Given the very small number of these therapies are studied or used clinically, the translatability and quality of these studies is questionable. On average, 44.7% of relevant study characteristics were not reported, in particular details on anastomotic complications (31.6%), use of antibiotics (75.7%), sterile surgery (83.4%) and postoperative analgesia (91.4%). Anastomotic complications were only reported as a binary outcome (leak or no leak), despite current advice to grade AL according to severity and clinical consequences (3, 5).
The proportion of studies with randomisation, blinding of surgery and blinding of primary outcome assessment has increased in the past two decades but remains insufficient, being included in only 62.4%, 4.9% and 8.5% of publications respectively.
This systematic animal review demonstrates that animal models vary widely in terms of species, method to compromise healing, intestinal segment and outcome measure used. Animal research on AL is of poor quality and is increasing in volume each year, which contradicts societal aims. Both reporting and study quality must improve for better transfer of animal data to the clinic.

Diclofenac causes leakage in the ileum and proximal colon, but not in the distal colon of rats – Chapter 4
Multiple studies have reported an association between NSAID use and AL, which raises questions about the causal role of NSAIDs in leakage (6–11). Important questions include the proportion of leak rate attributable to NSAIDs, how relevant the NSAID subtype, dose and time of administration are, and which bowel segment is most susceptible for NSAID-induced AL.
Clinical cohort studies on NSAID-related leak risk mostly address colorectal surgery (7, 8, 12). Rat experiments by our group showed that NSAIDs disturb healing of the ileal, and not the distal, colonic anastomosis (6, 13). We were interested in the proximal colon, whose morphology and susceptibility to pharmacologic NSAID activity is thought to be more similar to the distal colon, but intraluminal content, i.e. liquidity, microflora and bile acids, corresponds more to the terminal ileum.
In Chapter 4 we demonstrated that diclofenac causes leakage in the proximal colon (73%) and not in the distal colon of rats. Similar to ileum anastomoses, healing of proximal colon anastomoses is most disturbed when diclofenac is administered from the day of surgery. Leakage rates were lower if administration was postponed to postoperative day 1 or day 2. Disturbed healing of the ileal and proximal colonic anastomoses and not the distal anastomosis suggests a role for bowel content in diclofenac-induced leakage (14).

Initiation of inflammation and adhesion by serosal edge abrasion before anastomosing does not attenuate leakage by diclofenac – Chapter 5
Several studies report that NSAIDs suppress intraperitoneal adhesion formation, a phenomenon that theoretically weakens intestinal anastomoses (15–18). We designed a rat model of proximal colon anastomosis enhancing the inflammatory response and promoting anastomotic scar tissue to study if increased fibrosis can prevent diclofenac-induced leakage. In Chapter 5 we evoked an additional inflammatory and adherent process by surgically abrading the serosal edges of the anastomosis before restoring continuity. Explorative histologic analysis showed that serosal abrasion leads to dense scar formation at the serosal edge, compared to loose connective tissue in control anastomoses. However, this dense scar formation did not result in a stronger anastomosis, nor did it reverse diclofenac-induced leakage. This finding suggests that anastomotic healing and strength does not depend on adhesion formation on the outside of the anastomosis. It seems that the detrimental effect of diclofenac on anastomotic healing cannot be reversed by fibrosis-inducing measures, possibly due to the magnitude of this effect.

Biliary excretion of diclofenac affects healing of ileum anastomosis in rats – Chapter 6
NSAIDs that are excreted by the hepatic route (diclofenac, carprofen, celecoxib) cause more leakage than those excreted by the renal route (naproxen) (6, 13, 19). NSAID-induced intestinal mucosal damage has been attributed to biliary excretion of NSAIDs (20–27). We hypothesised that altered bile from NSAIDs is associated with leakage.
In Chapter 6 the relevance of bile composition in diclofenac-induced leakage was studied in 138 rats and 72 rats as bile donors. By cannulating the bile duct and duodenum of rats, bile was drained and replaced by bile from donor rats. The leak rate was 28% after replacement of bile with ‘diclofenac bile’ from donors, and 6% (p=0.089) after replacement with control bile. After oral diclofenac administration, leakage was 76% with ‘diclofenac bile’ and 47% (p=0.127) when bile was replaced with control bile. After intramuscular diclofenac administration 67% leaked, and 25% (p=0.060) leaked when bile was drained. After intramuscular diclofenac, 50% leaked when bile was first diverted and then returned, and 20% leaked (p=0.117) if bile was replaced with control bile. Taken altogether, leakage was significantly more severe in ‘diclofenac bile’ groups. HPLC and LCMS analyses demonstrated that diclofenac metabolite levels in bile peak within two hours, indicating that altered bile reaches the intestine shortly after diclofenac administration. The results demonstrated that altered bile composition by diclofenac administration disturbs the healing of ileal anastomoses in rats. It was not determined whether metabolites or diclofenac itself are responsible for the effects observed.

Microbial glucuronidase inhibition, aimed at preventing reactivation of diclofenac metabolites in the gut, reduces diclofenac-induced anastomotic complications in rats – Chapter 7
In experimental studies on diclofenac-induced small-intestinal mucosal damage it has been shown that reactivation of the biliary metabolite diclofenac-acyl-glucuronide by bacterial glucuronidase in the gut harms intestinal mucosa. The beta-glucuronidase inhibitor Inh1, developed by Wallace et al., prevented this toxic process. We aimed to test the glucuronidase inhibiting effect on anastomotic healing. In Chapter 7, 90 rats were used to study whether Inh1 reduces AL in diclofenac-treated rats. Administration of Inh1 resulted in a drop in the leak rate from 89% to 44% (p=0.094) and significantly reduced leak severity (p=0.029). Analysis of plasma levels of diclofenac revealed no changes in groups receiving Inh1. The finding that administration of a glucuronidase inhibitor reduces signs of leakage indicates that cleavage of diclofenac-acyl-glucuronide metabolites to diclofenac aglycones disturbs ileal anastomotic healing. Next to the human glucuronidase located in the brush border of the gut, β-glucuronidase is mainly produced by bacteria. In this way gut microbes are involved in the pathology of diclofenac-induced leakage.

Surgical stress causes a marked shift in microbial composition and diversity, though effects of diclofenac were not observed – Chapter 8
We wanted to further investigate the gut bacteria in diclofenac-induced AL. For this reason we explored the extent to which the operation and diclofenac administration alter the microbiome. It appeared that the surgery alone drastically alters the microbiome in different segments of the gut. A short period of postoperative diclofenac administration did not lead to additional changes. Either diclofenac has no effect on the diversity of the microbiome, or the effect was masked by an overwhelming effect of the surgery. In this study we did not determine a change in virulence of the microbiome which has been associated with AL in other animal models. It was concluded that the microbiota composition significantly changes following bowel resection and anastomosis. This study did not further contribute to understanding the diclofenac effect on AL, such as mediation via microbiome changes.

Conclusions
This thesis provides an overview of relevant cellular interactions occurring during anastomotic healing, illustrating that this process is insufficiently understood. It demonstrates that experimental research on AL is of low to moderate quality. In particular there is a need for better reporting of study methods and outcomes and for more profound measures to reduce bias.
The thesis provides evidence that diclofenac affects anastomotic healing differently along the gut, and that biliary excretion of diclofenac is relevant in causing anastomotic complications. A luminal intervention targeted at microbial metabolism of diclofenac molecules reduces the effect of diclofenac. The finding that such luminal factors impact anastomotic healing is novel and provides directions for future research and clues for targeted treatment.
Surgical stress, such as bowel resection and anastomosis, significantly alters the gut microbiome. Specific changes in this major luminal factor may make anastomoses more susceptible to leakage in general, or specifically when diclofenac is used, although the latter suggestion could not be demonstrated in this thesis.