Why do anastomoses leak?

Anastomotic leak (AL) is perhaps the most dreaded complication of intestinal surgery. In an era of enhanced recovery programs, patients undergoing colon resection are typically prepared for a short hospital stay and rapid return to regular activities and/or work. Instead, patients who develop a leak may be hospitalized for weeks or months, undergo countless invasive procedures, reoperations in a hostile operative field, creation of an intestinal stoma and endure multiple complications or worse. Patients often experience prolonged periods of disability which can cause major stress and hardship for the patient and their family, and create great strain on the relationship with their surgeon.

Since anastomotic leak is often conceptualized to result from technical factors (although typically unspecified and unknown), the patient and surgeon are left to wonder “what went wrong”. Although there are many risk factors that have been associated with AL, it is not often possible to know whether or how these factors actually were responsible for the leak. For decades and generations, the fundamental assumption has been that AL occurs owing to mechanical factors (e.g. tension on the anastomosis, error in technique) and consequently there has been considerable emphasis on what can make the anastomosis “stronger”.

Yet this paradigm has not appeared to serve us well as AL rates have remained largely unchanged for decades; studies comparing single layer vs two layer, sutures vs staples, inversion vs eversion, mechanical buttresses/adjuncts etc. have generally failed to demonstrate any difference in leak rates and the vexing problem of AL appears to remain largely unsolved. Are we sure we have the right paradigm as to the cause of leaks after intestinal anastomosis? The aim of this review is to provide an overview of traditional and emerging concepts on the pathogenesis of AL after bowel resection.

It really is about technique

Technique has long been considered to play a critical role in maintaining anastomotic integrity. However, exactly what constitutes optimal technique has continued to be elusive and uncertain. A systematic review of clinical and experimental studies published over four decades looking at suture material, suture format, single vs two layer construction, interrupted versus continuous suture and anastomotic configuration generally failed to demonstrate any factors consistently associated with lower leak rates.¹ It is generally accepted that an anastomosis under tension has a higher chance of developing a leak, and that adequate mobilization of the bowel (e.g. splenic flexure mobilization) is necessary to create a tension free anastomosis. However, multiple studies have failed to show an advantage to routine mobilization of splenic flexure, suggesting a case by case approach.² Controversy still exists about the impact of handsewn versus stapled anastomosis. A 2012 Cochrane database review found insufficient evidence to demonstrate superiority of stapled over handsewn techniques in colorectal anastomosis surgery, regardless of the level of anastomosis.³

Several studies have also compared the anastomotic leak rates between intracorporeal versus extracorporeal anastomosis. A systemic review and meta-analysis found no difference in AL after minimally invasive right colectomy between these two techniques.⁴ Laparoscopic colectomy has been associated with reduced odds of developing an anastomotic leak following colectomy, when controlling for patient disease and procedure related factors.⁵ The type of reconstruction after a colon resection has also been evaluated. A multicenter randomized clinical trial comparing colonic J pouch to straight stapled colorectal reconstruction after low anterior resection for rectal cancer found no difference in the incidence of anastomosis leak.⁶

Broadly speaking, it would be hard to argue that technique “doesn’t matter”; indeed, studies have shown that surgeon characteristics such as specialty training or even individual surgeons operating in the same setting may impact leak rates.⁷ But it is equally true that there is very little if any evidence to suggest that a particular technical consideration holds the key to reducing leak rates, assuming that basic technique is fundamentally sound.

Is it the blood supply?

Perfusion is a key component of anastomotic healing and may be a modifiable risk factor to prevent AL. In the ideal setting, perfusion is able to be confirmed with reliability. However, traditional perfusion assessment, such as pulsation of the mesentery, bowel coloration and active bleeding are subjective and not reliable. In the era of laparoscopy, these methods can be very limited in their applicability. In retrospective and phase II clinical trials, ICG fluoroscopy has been shown to reliably predict anastomotic perfusion and has been associated with a decrease in AL.¹⁰

However, the randomized trials published to date have yielded only mixed results. The PILLAR III trial reported no benefit with ICG, noting similar anastomotic leak rates in patients undergoing low anterior resection for rectal cancer (9.0% in perfusion group compared to 9.6% in control group), with a mean anastomotic height of 5.2cm in both groups.¹¹ On the other hand, De Nardi observed a lower anastomotic leak rate in an ICG group compared to the control group (6 vs 9%, p<0.05).¹² The FLAG trial reported a lower anastomotic leak rate in the ICG group (14.4%) as compared to controls (25.7%).413 The leak rate was similar in the subset with higher anastomoses.

ICG fluoroscopy represents a qualitative approach to perfusion and more quantitative assessment tools that measure oxygen tension, for example, may better indicate adequacy of the anastomotic blood supply. Further, anastomotic leaks may be multifactorial without a single causative factor in all cases. The IntAct trial which includes 25 sites and 880 patients in Europe will evaluate the microbiome in the setting of ICG fluoroscopy. This study may answer some questions regarding the interplay between these two key players in the pathogenesis of anastomotic leak.¹⁴

Is it the microbiome’s fault?

The intestinal microbiome consists of the trillions of organisms and millions of unique microbial genes that inhabit the gastrointestinal tract. It is now widely appreciated that this vast ecosystem has enormous metabolic potential, and that the host microbe interaction plays a critical role in maintaining intestinal homeostasis. Consequently, given the direct proximity to a newlyconstructed anastomosis, it is logical to assume that perturbations of intestinal microbes would play a key role in healing. Perhaps the first investigation on the role of bacteria on anastomotic healing was in 1956 when dogs whom underwent an intestinal anastomosis were subjected to direct anastomotic injection of intraluminal tetracycline.¹⁵ Despite mesenteric devascularization antibiotics completely prevented anastomotic dehiscence, whereas 80% of control animals that underwent devascularization without antibiotics developed leakage and peritonitis. Twenty years later, Patel demonstrated that increasing exposure of anastomotic tissue to kanamycin and erythromycin in dogs was associated with increasing anastomotic tensile strength.

Despite these early observations associating antimicrobials with improved anastomotic healing, it has not been until recently that a more mechanistic approach to the role of microbes has been explored. Our group has demonstrated that certain bacteria, specifically Enterococcus faecalis, Pseudomonas aeruginosa, Serratia marcescnes, and Proteus mirabilis, preferentially colonize anastomotic tissue in murine models of a left sided colon anastomosis. We have discovered that when certain environmental conditions exist that are known to increase risk for leak, such as intestinal ischemia and a high fat diet (i.e. obesity), collagenase producing strains of these microbes predominate and can cause anastomotic dehiscence.¹⁶

The mechanism underlying how these collagenolytic microbes can promote anastomotic leak is likely multifactorial. Intestinal anastomotic healing is complex and thought to follow the characteristic inflammatory, proliferative, and remodeling phases seen in other tissues. During the proliferative phase, fibroblasts proliferate and begin collagen deposition, leading to a peak of collagen content on postoperative day six. Matrix metalloproteinases (MMP) then subsequently reorganize and remodel the extracellular matrix and the immature collagen fibrils. This cadence, which occurs significantly faster in the gut compared to skin, results in reacquisition of near maximum intestinal strength by the end of the first postoperative month. We have found that bacterial collagenase has both direct and indirect effects on this normal healing process. First, bacterial collagenases can directly degrade both collagen type I and type IV, the primary collagen subsets found within intestinal tissue. Second, collagenolytic organisms can bind to and activate MMP’s and other proteases, creating a supra-physiologic proteolytic environment.¹⁷ Together, the cumulative effects of collagenolytic organisms colonizing anastomotic tissue clearly causes anastomotic leak in rodents.

The role of these organisms in promoting anastomotic leak in humans is less clear. While organisms such as E. faecalis and P. aeruginosa are commonly found in drainage fluid following an anastomotic leak, they are also found to be colonizing intestinal tissue during the perioperative period in patients that do not develop a clinical anastomotic leak.18 These incongruent results between rodents and humans is an area of active investigation. While human and mouse microbiomes are up to 90% similar in terms of which phyla are present, significant differences in bacterial abundance, community structure and metabolite production exist. Therefore, while E.faecalis may cause leak in rodents, alternative species or the community as a whole, may play the key role in humans. In one of the few studies on alterations of the microbiome in human anastomotic leak, van Praagh compared the microbiota in 123 patients undergoing a stapled colorectal anastomosis.19 They found that low microbial diversity and high abundance of Bacteroidaceae and Lachnospiraceae families were significantly associated with anastomotic leak. Functional analysis was not performed in thus study, and thus collagenolytic potential of these bacterial communities is unknown.

Taken together, the influence of microbes on intestinal healing is emerging, and further human based studies using both stool and direct anastomotic sampling are needed. If it is shown that collagenase producing organisms can degrade healing anastomotic tissue in humans, it would set the stage for targeted therapies to reduce the bacterial collagenolytic potential during the perioperative period. This may include perioperative dietary modifications that prevent the emergence of these collagen destroying strains, or selective decontamination only in patients that are colonized with these microbes. Further, the relationship and interactions between technique, perfusion and microbiome will need to be explored, as the pathogenesis of AL seems likely to be multifactorial, at least in some circumstances.