Key Concept: Macro– and microvascular blood flow provide the necessary factors to enable anastomotic healing. Avoid using the sigmoid colon, when possible, and fully mobilize the splenic flexure to provide a tension–free low colorectal anastomosis.

For the normal healing process of an anastomosis to take place, it must have ample tissue perfusion to deliver the influx of inflammatory cells, growth factors, and oxygen. Ample tissue perfusion of a healing anastomosis is determined by the macrovascular and microvascular anatomy as well as the arterial tissue oxygen saturation [5].

Currently the list of identified risk factors for anastomotic leaks is extensive, and their exact relationship to, and significance in, anastomotic leaks is hard to define. You are therefore faced with a complex array of risk factors, all with varying degrees of importance, affecting to some degree one or more components of anastomotic healing, and you must decide which one is at play in a particular patient.

Key Concept: Certain inherent risk factors are present with every operation, though identifying and targeting modifiable risk factors, when possible, may mitigate the development of leaks.

Weakly associated patient-related risk factors include age, tobacco and alcohol use, and obesity [20]. Obesity is not felt to be a risk factor at all in right-sided resections; however, there tends to be a more significant association with rectal resections [21]. A combination of risk factors may yield even more significance. This is likely why a higher American Society of Anesthesia (ASA) classification (usually scores of 3 or more) is one of the more consistently found risk factors for leak [20]. Likewise, patients with Charleston Comorbidity Index scores of 3 or more have also been found to have significantly higher anastomotic leaks [17]. These and other types of scoring systems might be more useful indicators of a patient’s risk of anastomotic leak.

Key Concept: Nutrition is a potentially modifiable risk factor that can often be improved prior to surgery. Its impact on anastomotic leak is somewhat variable, but poor nutrition is commonly associated with higher rates of leak, specifically for right–sided resections.

Certain preventive measures can be taken even before the operation is begun to reduce the risk for leak, primarily in the elective setting. Wexner suggests that these preventive measures are more commonly encountered preoperatively in a right colectomy and intraoperatively in a left colectomy [22]. This statement has been further validated by other studies, including one multicenter prospective trial [23]. Identifying those patients who are of poor nutritional status and treating them preoperatively may reduce the risk for and the morbidity and mortality from anastomotic leaks [23]. Generally, poor nutritional status has been defined in the literature as weight loss ≥10 %, serum albumin <3.5, and serum proteins <5.5 g/dL [24–27]. It is important to define what level of severity of malnutrition requires preoperative nutritional support. Most studies have used a weight loss of ≥10 % to define malnutrition [28]. However, according to Jie et al., this only represents the cutoff value for which complications are seen to increase [28]. The nutritional risk score (NRS) may be a better indicator of severe malnutrition for which preoperative nutrition is indicated. Parameters which are used in the NRS include weight loss in the last 3 months, decrease in food intake, body mass index, severity of disease, and age. Thus, a 65-year-old patient with colon cancer who has greater than 15 % weight loss in the last 3 months and needs a right colectomy would have a score of 5. With this scoring system, Jie found that in those patients with a NRS ≥5 and a lower GI resection, preoperative nutrition decreased the complication rate from 45 % in the control group to 27 % in those patients with preoperative nutrition [28]. An in-depth discussion regarding the impact of nutritional support can be found in Chap. 3 by Dr. Maykel.

Key Concept: Patients with malignancy and inflammatory bowel disease often require surgical resection and are commonly immunosuppressed or taking immunosuppression agents preoperatively. You need to have a thorough understanding of how this can impact postoperative morbidity including anastomotic leaks.

Key Concept: In addition to the pathophysiology of the disease itself, the Crohn’s patient is likely to present with other risk factors for anastomotic leaks including preoperative malnutrition and immunosuppressive medication use.

Multiple studies have shown a high risk of intra-abdominal septic complications in patients with Crohn’s disease [39, 40]. Other risk factors which are much more specific to Crohn’s disease include a hand-sewn anastomosis, end-to-end anastomoses, histologic positive margins, penetrating-type disease, and the need for sigmoid resection [20, 30, 34, 41]. We recommend stapled side-to-side anastomoses when performing an ileocolostomy and to highly consider performing proximal diversion in cases where both small bowel and sigmoid resections are needed. Grossly histologic negative margins are almost impossible to achieve in a patient with Crohn’s disease without causing unnecessary bowel resection. A grossly negative margin, as indicated by the soft, thin mesentery at the point of resection, is adequate.

Key Concept: Radiation alone may not increase the risk of leak, and does not mandate the need for diversion, typically, unless when combined with chemotherapy.

Neoadjuvant radiotherapy in the pelvis may result in increased risk of anastomotic leak; however, this belief has not been definitively established in the literature [42]. A study that looked at 1,338 patients with rectal cancer over a 30-year period was unable to find any significant association between neoadjuvant radiotherapy and anastomotic leaks. Other factors such as location and size of tumor were found to be of more significance. This has been confirmed in other reports, including a fairly large randomized trial [43]. There is no need for routine diversion in patients who have received neoadjuvant radiotherapy alone. If chemotherapy is added to the radiation, a higher risk is likely, and a diverting stoma is recommended. As will be further discussed later, proximal diversion does not prevent a leak but reduces its impact.

It is a commonly held belief that an anastomosis created in the setting of peritonitis will be at increased risk of anastomotic dehiscence [48, 49]. However, previous animal studies and other clinical studies have failed to show any increase in risk of leaks in the setting of peritonitis [49, 50]. There may be a difference between purulent versus fecal peritonitis (e.g., Hinchey III vs. IV) and the risk of leak, as previously reported by Biondo and colleagues [50]. In a follow-up study, they were able to perform a primary resection and anastomosis with a respectable 5.7 % leak rate without the use of proximal diversion in patients with purulent peritonitis. They excluded those with fecal peritonitis, as well as ASA IV, and unstable or immunocompromised patients. Of the remaining 208 patients, 50 % of the patients had peritonitis, of which half of these had diffuse peritonitis. Peritonitis was not found to be an independently associated risk factor for anastomotic leaks [47]. At present it does not appear that purulent peritonitis alone is a risk factor for anastomotic leak [47, 50–52].

One of the reasons why emergency surgery is felt to be associated with higher leaks is that these operations are performed on the unprepped colon. The “loaded colon” has been reported to have up to a threefold increase in anastomotic leaks [53]. This is contradictory to the most recent studies on mechanical bowel preparation that have concluded that it can be safely omitted. Methods such as intraoperative colonic lavage have been shown to have a positive effect on anastomotic integrity and collagen metabolism and can allow for a primary anastomosis to be performed without diversion in emergency operations for colonic obstruction [49, 54]. Until additional evidence to the contrary emerges, it is recommended that colonic lavage be performed when distal colon and rectal anastomoses are created in the “loaded colon.” In contrast, elective operations to remove the left colon or rectum can be safely performed with only enemas (without a complete bowel preparation) to empty the stool.

Since the healing anastomosis is extremely dependent upon adequate perfusion, episodes of hypotension, and especially those requiring vasopressors, should be an absolute indication for preventive measures. In fact, shock was one of the only two risk factors for which proximal diversion would be needed, according to an AAST multicenter trial, in the cases of traumatic injuries to the colon [55].

In the majority of emergency or diverticular operations—in the absence of either fecal peritonitis or shock—a primary resection and anastomosis can be performed safely. Lower rates of anastomotic leak, re-interventions, and other wound infections are seen with proximal diversion when other risk factors are present. As previously stated, intraoperative lavage is recommended for emergency resections involving the impacted left colon to allow primary anastomosis [20].

The site of the anastomosis has been the most consistent and significant risk factor for anastomotic leak [56]. The further distal an anastomosis is created, the higher the risk of leak. An ileocolic anastomosis has a leak rate of 2–3 % compared to a 10–17 % leak rate in coloanal anastomosis. Even in rectal anastomoses, a significant difference can be seen the closer the staple line gets to the anal verge [56]. The highest risk of anastomotic leak can be seen for anastomoses at and below 5–8 cm from the anal verge [23].

There are several proposed reasons for the difference in leak rates between proximal and distal locations including:

Key Concept: One way you can reduce the risk of anastomotic failure during the operation is by limiting intraoperative blood loss and the time it takes to perform the operation.

While both of these parameters have been confirmed to decrease leak rate on multivariate analysis [64, 65], the actual significance is not completely clear, as this has not been uniform across the literature [66, 67]. Furthermore, the actual amount of blood loss or duration of operation that matters is much less clear. Operative times found to be of significance ranged from 120 to 270 min [17, 24, 66, 67], while meaningful operative blood loss has been defined as that which requires blood transfusion—a highly variable definition [17]. What seems clearer is that increased operative time leads to more exposure of the patient to tissue trauma and bacteria [25] and correlates with hypothermia in most patients. Primary hypothermia correlates with increased infectious complications and hospital length of stay [68]. Even more, increased operative times and greater amounts of blood loss are surrogate markers for the degree of difficulty of the surgery. You must be mindful of these objective indicators of a more difficult operation in order to accurately decide whether or not to perform proximal diversion. While occasionally viewed as a “failure,” it is never wrong to err on the side of diversion.

Key Concept: Adverse events during the operation, even if not directly involving the anastomosis, can increase the risk of anastomotic leak.

Trencheva et al. defined an intraoperative complication as injury to the bowel, other organs, or blood vessels. In addition, stapling device malfunction, hypotension, oxygen saturation less than 90 % for more than 5 min, pH less than 7.3, and even blood loss requiring intraoperative blood transfusion were also classified as an intraoperative complication. In their series, any patient with an intraoperative complication was four times as likely to have an anastomotic leak [17]. On one hand, this may again be a surrogate for a more difficult operation. More appropriately, this highlights the degree of interconnectivity among all aspects of an operation and the impact one problem can have on another.

Key Concept: Although oncologically sound, TME results in a lower anastomosis and the potential for loss of blood supply that may increase the leak rate.

During the widespread adoption of total mesorectal excision, there was a substantial rise in leak rates from the previously reported 9 to ~23 % [69]. Over the next 4 years following this initial study, the leak rate eventually did return to the level seen before TME [69]. A study looking at laparoscopic resections showed that the addition of TME more than doubled the leak rate for upper rectal cancer [67]. TME for high rectal cancer may result in insufficient blood supply to the posterior portion of the proximal rectum, which is further evidenced by the fact that tumor-specific mesorectal excisions have lower rates of anastomotic leaks [15, 70].

Key Concept: Avoiding tension has been classically viewed as one of the fundamental principles of a healing anastomosis and remains a significant preventative measure in reducing anastomotic leak.

The colon seems to be especially effected by applied tension, even more so than the small intestine. Blood flow did not return to preoperative levels until the seventh postoperative day following experimentally applied tension in animal studies [71]. Mobilization of the splenic flexure has commonly been used to decrease tension in the rectal anastomosis. Karanja et al. found a 9 % leak rate with splenic flexure mobilization compared to 22 % without mobilization [72]. Additional evidence supporting its actual significance is limited. The anastomosis is subjected to other sources of tension during peristalsis and defecation that result in radial tension for which splenic flexure mobilization would seem to provide less benefit. The major importance in splenic flexure mobilization may ultimately be improved blood supply of the descending colon (rather than the sigmoid) when used for the colorectal anastomosis.

Key Concept: Drains may be useful in low extraperitoneal anastomosis, but are not typically indicated when the anastomosis is intraperitoneal.

The use of drains has been extensively debated over the last decade, largely because they offer both real and theoretical benefits and risks, yet are only one of many factors that play a role in a proper anastomosis. Furthermore, they are widely variable in their use, type, and rationale for placement. There is extensive evidence that draining an intraperitoneal anastomosis is of no benefit [69]. In the case of gross contamination or abscess at the time of resection, the intraperitoneal anastomosis should be created in a less hostile location and after extensive contamination control. Draining the pelvic anastomosis may be of some benefit. The pelvis does seem to be unique as compared to the abdominal cavity in that fluid is much more likely to accumulate in the most dependent area of the pelvis around the anastomosis and the non-peritonealized pelvic floor fails to absorb fluid efficiently [47, 73]. However, it is unclear how effectively our drains remove this fluid or what impact this has on the healing anastomosis [17, 47, 74]. The other proposed benefit of draining the pelvic anastomosis is detection of an anastomotic leak. On one hand, studies have shown that drains have very poor detection rates and that other clinical signs are more likely to appear before any change in drain effluent [61]. Others have shown some benefit in leak detection where drains detected a leak in 80 % of patients and in 40 % of cases this preceded any other clinical signs [75]. In the most recent meta-analysis, there was not enough sufficient evidence to indicate that drains are able to prevent anastomotic leaks or other anastomotic complications; however, the authors acknowledged the need for more randomized control trials specifically looking at lower rectal anastomosis [66]. It is our practice to use drains selectively in cases where build up of fluid in the pelvis begins during the case despite good hemostasis or after a very bloody operation where every vessel could not have been controlled. In general, we remove the drains once the effluent is less than 15 cc/day or if clear.

Key Concept: Laparoscopic approaches with experienced surgeons may result in a decrease in leak rates.

A study using the Nationwide Inpatient Sample population database found a decrease in the rate of anastomotic leak along with a corresponding decrease in wound infection following laparoscopic surgery, despite an increased leak rate with conversion to open surgery [76]. The recently published Danish nationwide cohort study demonstrated an increase risk of leak with minimally invasive approaches; however, this study was performed during the period when the laparoscopic method was first being used [77]. While no definite conclusion can be drawn, this latter study does highlight the potential impact that inexperience can play in leak rates. Interestingly anastomotic leaks have been diagnosed earlier following laparoscopic surgery, and as a result of the primary operation being laparoscopic, laparoscopic management of the leak was possible [67]. A possible cause of low rectal anastomotic leaks in totally laparoscopic cases may be the use of multiple firings of the endoscopic stapler to transect the rectum. Crossing staple lines and poor perfusion always put an anastomosis at risk, regardless of the surgical approach.

Key Concept: Wrapping the anastomosis with a well-vascularized pedicle of omentum has been associated with decreased leaks in small studies.

Animal studies have proven the unique ability of the omentum to adhere to and effectively bridge the anastomosis [78, 79] and allow for absorption of fluid [80]. Some have cautioned its use secondary to the likely negative impact in cases where the omental pedicle is devascularized [80]. The most recent meta-analysis found a significant reduction in clinical anastomotic leak only. Issues involving blinding and a small number of patients within each study limited the strength of the conclusion. Omentoplasty should be left up to the surgeon’s personal experience.

Key Concept: Staged resections may decrease the overall morbidity and leak rate for extensive disease, while simultaneous resection is generally safe in carefully select patients.

Synchronous liver metastases are present in 23–51 % of newly diagnosed patients [81], and liver resection remains the best option for those patients with resectable disease [82–84]. Staged procedures have been the traditional approach [83]. Most studies evaluating safety and efficacy are retrospective and therefore suffer from selection bias resulting in more extensive liver resections in the group of patients with a staged approach to extensive liver resections [83]. These studies do show that limited liver resections can be performed at the same time as the colon resections with equivalent morbidity and mortality [82–84]. With more extensive liver resections and in those 70 years or older, the morbidity and mortality significantly increase, favoring a staged approach [82]. Age and extensive liver resections seem to be the main factors to consider when deciding upon a staged versus simultaneous resection. Leak is just one of the many causes of postoperative morbidity and mortality. The only study focused on anastomotic leak after simultaneous liver resection showed an operative time greater than 8 h was the most significant risk factor for anastomotic leak regardless of the extent of the liver resection. The majority of leaks occurred with rectal resections (36 % leak rate). Patients with colonic resections were observed to have a 13 % leak rate, but this was still higher than those patients who only underwent a colonic resection [82].

Key Concept: Several methods are available for investigating the integrity of the anastomosis. Whatever method you choose, it should be, in general, a routine part of your practice for all left–sided anastomosis.

Following injection of fluorescent dye, a mounted camera with an infrared filter is used to view the resected ends or newly created anastomosis. Special software can be used to compare two different areas of perfusion. A study showed that using the IC-View system® (Pulsion Medical System, Munich, Germany) changed the site of resection in 16 % of patients [97]. Other available forms of fluorescence angiography include the Spy Elite® (Lifecell, Bridgewater, NJ) for open surgeries and the Firefly^TM used in the DaVinci® Si^TM Surgical System Robot (Intuitive Surgical, Sunnyvale, VA).

Studies assessing the effectiveness of air leak testing have shown mixed results [98–100], though typically demonstrate usefulness and no downside. In a large retrospective cohort [101], untested anastomoses had twice the rate of clinical leak than those that were tested. Patients who underwent suture repair after a positive air leak test had a clinical leak rate of 12.2 % compared to 3.8 % for patients with a negative air leak test. Patients who underwent anastomotic revision or proximal diversions after a positive air leak test had a 0 % clinical leak rate [101]. This study provides significant evidence for the use of intraoperative leak testing. A diverting ostomy should always be a consideration with positive air leaks.

Li et al. [102] looked at the selective versus routine use of endoscopic examination in bowel resections. Endoscopic examination with air leak testing was performed on the pre-resected bowel, the rectal stump, and the post-anastomotic bowel. This study showed a nonsignificant increase in leaks (5.1 % vs. 0.9 %) with selective versus routine endoscopic examination. The endoscope compared to the proctoscope provides better visualization of the anastomosis and likely a better assessment of its integrity.

Using a 22 French Foley, a mixture of sterile water and blue dye is injected intraluminally, while the bowel proximal to the anastomosis is clamped. It takes a volume of 180–240 mL to adequately distend the anastomosis. A study using this method found that the dye test allowed for the easier detection and localization of leaks compared to air leak testing [103].

Key Concept: These devices are either early in their experience or have not demonstrated a marked benefit to reducing leaks.

Methods used to bolster the staple line with bioabsorbable Seamguard® (W.L. Gore & Associates, Flagstaff, AZ) or meshed AlloDerm® (Lifecell, Bridgewater, NJ) have not improved the anastomotic strength [110, 111]. Clinical data evaluating the use of such tissue-bolstering devices for the colorectal anastomosis is limited. Staple line reinforcement has not had the same success in the colorectal anastomosis as is seen in the gastric bypass or sleeve, where there primary purpose is reduction in bleeding [112].

Anastomotic leaks account for a quarter of all deaths following colorectal surgery [45]. These mortality rates have changed very little over the last three decades despite the continuing improvements in critical care management. The mortality and morbidity from anastomotic leaks are greatly influenced by the duration of time before a diagnosis is made and the source of infection controlled [45]. Unfortunately many patients will be discharged home before a diagnosis is made, and others will be treated with more conservative therapies, both of which delay definitive therapy and extend the duration of infection and sepsis. Aside from morbidity and mortality, patients with anastomotic leaks can have a significant decrease in their quality of life, which is mainly due to the high rates of a permanent stoma (up to 72 % in some studies), especially when end ostomies are performed instead of proximal diversion [113].