The Colorectal Anastomosis: A Timeless Challenge

Abstract

Colorectal anastomosis is a sophisticated problem that demands an elaborate discussion and an elegant solution.

“Those who forget the past are condemned to repeat it.”

George Santayana, Life of Reason , 1905

The colorectal anastomosis is the apotheosis of the specialty, for inasmuch, as it bears the same name and it distinguishes the colorectal surgeon from their peers. It should come as no surprise that outcomes are considerably better in specialized hands ^{1 2 3 4 5} which cannot be overlooked when the stakes are so high. ⁶ Anastomotic leak is physiologically and functionally devastating if not deadly, ^{7 8 9} and where cancer is concerned, it increases local recurrence and reduces long-term survival. ^{10 11 12}

The early history of distal colon and rectal surgery is a testament to the challenges of achieving adequate resection without compromising function. The first successful sigmoid resection with end-to-end anastomosis was performed in 1823 by Jean Francois Reybard who was subsequently censured by the Paris Academy of Medicine as drastic and dangerous. Throughout the 19th century, resection with stoma and distal closure remained the favored approach for colorectal cancers. Although “cure” rates gradually increased from 20 to 60% with colonic exteriorization, mortality rates followed close behind, upward of 50%. Those who survived, often suffered from permanent functional disability and cutaneous fistulae. Furthermore, the extraperitoneal rectum remained largely inaccessible and most operations were palliative. ¹³

At the turn of the 20th century, combined abdominal and perineal approaches to rectal cancer were described by Czerny and Mayo and eventually standardized by Sir William Miles who improved 1-year survival to 58%, albeit with the well-known attendant morbidity of the abdominoperineal resection (APR). ¹⁴ In 1921, Hartmann introduced a better tolerated option for upper rectal cancers, but it was not until 1948 when Claude Dixon published his seminal work on anterior resection, demonstrating low mortality with improved long-term survival that restorative proctectomy with colorectal anastomosis became a viable option. ¹⁵

Over the next 30 years, Goligher et al and Pollett and Nicholls demonstrated distal resection margins as short as 2 cm were oncologically adequate, ^{16 17} permitting low anterior resection (LAR); and in 1976, Jeffery et al demonstrated that ultra-low resection with coloanal anastomosis was feasible. ¹⁸ Perhaps the biggest revolution in colorectal anastomosis came in 1972 when Ravitch and Steichen introduced surgical staplers, facilitating low colorectal anastomoses with a reported leak rate of 3.6%. ^{19 20} In 1980, Knight and Griffen introduced the double-stapled technique (DST) ²¹ which remains the dominant mode of colorectal anastomosis till now, providing a significant speed advantage, while eliminating contamination, luminal mismatch, and the restrictions of a narrow pelvis.

The field of colorectal surgery continues to evolve, carried forward on waves of new technology, such as laparoscopy and robotics, and novel techniques, such as transanal total mesorectal excision. But it is important to remember that, in spite of the past 75 years of technical advances, the anastomotic leak rate has not improved since Dixon's initial series of 426 patients with 12 leaks (2.8%). Today, the acceptable leak rate ranges from 5 to 10%, providing for the added strains of radiation, chemotherapy, and an aging patient population. Technology has merely changed the face of surgery but not its core tenets. Whether the surgeon's hands are sewing in the pelvis, firing a stapler from the other side of the abdominal wall, or steering at a console on the opposite side of the room, command of physiology, appreciation for anatomy, gentle tissue handling, meticulous dissection, adequate blood flow, and minimal anastomotic tension remain paramount to success. And, in fact, many of Dixon's original considerations, such as use of mechanical bowel preparation, benefit of antibiotics, assessment of perfusion, avoidance of suture line tension, the overuse of proximal diversion, and the role of drains, are continuing to gain clarity.

Thus, the impossible challenge for the modern surgeon is not really creating an anastomosis between the colon and rectum. Rather, it is understanding how, while standing on the shoulders of giants, with an embarrassment of toys, and a profusion of data, we continue to sustain any leaks at all. The aims of this article are two-fold: (1) provide exhaustive review of the scientific evidence supporting various techniques of colorectal anastomosis, and (2) familiarize the surgeon with available technology. In reading the literature, we must keep in mind that study populations, experimental conditions, and even the very definitions of colorectal anastomosis and anastomotic leak are inconstant. We peer at the truth inside the black box through very small holes. But if we make enough holes from a great many perspectives, we may begin to see its form.

Preparing for Anastomosis

Regional Perfusion

The descending colon is supplied by the left colic artery (LAC) arising from the inferior mesenteric artery (IMA) and running cephalad, sometimes parallel to the colon, depending on how it is retracted. It has additional blood supply from the marginal artery of Drummond and the meandering mesenteric artery (the arc of Riolan), both arising from the middle colic artery in the territory of the superior mesenteric artery. The marginal artery is the more prominent of the two. It exists reliably on the left side with documented absence of only 7% at the splenic flexure (Griffith's point) and 8% at the sigmoid. ²² Following high ligation of the IMA, the marginal artery is the major blood supply of the proximal anastomotic limb. Preservation of the LCA enhances blood flow in the marginal artery. Attention should be paid to patency of the blood supply in patients with atherosclerotic disease, as this may influence operative decision making.

The sigmoid colon is supplied by branches of the IMA. The bifurcation patterns of the IMA are variable, but in most patients (type I, 50%) the IMA first gives off an ascending left colic branch and subsequently a sigmoidal trunk (ST) that branches into the sigmoidal arteries before crossing the left iliac artery as the superior hemorrhoidal (rectal) artery (SRA). ²³ Common minor variants include ST bifurcating from the LCA (type II, 25%) and trifurcation of the LCA, ST, and SRA (type III, 20%). The superior rectum is supplied by the superior hemorrhoidal artery. Most societies European Association of Endoscopic Surgeons (EAES), Society of American Gastrointestinal and Endoscopic Surgeons (SAGES), European Society of Coloproctology (ESCP), and American Society of Colon and Rectal Surgeons (ASCRS)) recommend preservation of the IMA and superior hemorrhoidal branch during surgery for diverticular disease to preserve perfusion of the distal anastomotic limb, but agreement is not universal. ²⁴ Some studies suggest superior hemorrhoidal artery preservation may reduce anastomotic leak, although a recent meta-analysis failed to find a benefit. ²⁵

For cancer surgery of the sigmoid and rectum, the debate regarding high versus low ligation of the IMA (with preservation of the LCA; Fig. 1 ) rages on some 114 years, since the techniques were broached by Moynihan and Earnest Miles, respectively, in 1908. ^{26 27} Generally, high IMA ligation is considered a superior oncologic procedure and offers superior mobilization for tension-free anastomosis. Conversely, low ligation is considered less morbid, preserves blood supply to the anastomosis, and potentially avoids risk of nerve injury. These assumptions have not borne out in the literature, however. Good evidence from randomized controlled trials is lacking and existing retrospective cohort studies are confounded by the decades-long march to total mesolectal excision, the minimally invasive approach, and preoperative chemoradiation. Most modern studies fail to show an oncologic advantage to high ligation, neither in terms of lymph node harvest (which is adequate regardless) nor in terms of disease-free and overall survival. ²⁸ And, while cadaveric studies confirm that high ligation provides more reach than low ligation, ²⁹ the additional length is unnecessary in 80% of cases ³⁰ and may even be negated by the need to resect more of the devascularized left colon and mobilize the splenic flexure. The final point of contention is clearing the apical lymph nodes along the proximal IMA, wherein metastases have especially poor prognosis. Although early studies suggested improved survival with more radical lymphadenectomy, later work has failed to replicate this advantage, ^{31 32} especially when considering that apical lymph node metastases are rare (0.3–8.6%) and when extant portend distant metastatic disease that has already escaped the domain of the scalpel.

Fig. 1.

Blood supply of the left colon, depicted with an incomplete meandering marginal artery. ( A ) High ligation of the IMA near its origin abrogates flow from the LCA, limiting arterial supply of the left colon to the marginal artery from the left branch of the MCA. ( B ) Low ligation of the IMA beyond the takeoff of the LCA preserves native blood supply to the left colon, purportedly enhancing perfusion of the proximal limb of a colorectal anastomosis. Image courtesy of Jay Vuitch.

The evidence for low ligation is mounting. ³³ High ligation of the IMA interrupts blood flow to the LAC, and hemodynamic studies show that the concomitant reduction in marginal artery blood flow is reduced by up to 40%, ³⁴ potentially threatening the proximal limb of the anastomosis. The difference is especially stark in older males with peripheral atherosclerotic disease. ³⁵ One can imagine that the watershed area in the splenic flexure that is susceptible to ischemia in low flow states is indeed a tenuous blood supply for a fresh anastomosis in a hemodynamically labile postoperative patient. Whether “this disruption is clinically significant” was challenged by early meta-analyses, ^{36 37} dominated by a large population-based study from the Swedish Colorectal Cancer Registry that showed equivalent symptomatic leak rate of 10% with high and low ligation. ³⁸ But meta-analyses incorporating more modern data, including four randomized controlled trials, show a significant reduction (odds ratio [OR] = 0.75) in anastomotic leak with low ligation versus high ligation. ^{32 39} Low ligation also theoretically avoids inadvertent injury to the hypogastric nerve which may course more than a few centimeters anterior to the aorta. Despite this, no studies have consistently shown reduced morbidity with low ligation in terms of defecatory, urinary, or sexual function. ⁴⁰

Given these data are mostly surgical, societies now advocate for low ligation in the absence of gross lymphadenopathy, with the important caveat that evidence is lacking. In many Asian countries, where surgeons tend to follow the paradigm of more radical lymphadenectomy, a hybrid approach is often utilized, comprising low ligation of the vessel with high lymph node dissection to capture the apical nodes. ^{33 41} In the West, where obesity is rampant and meticulous dissection of the IMA origin is not routinely undertaken, ³¹ the question of high versus low ligation becomes pragmatic. With the medial to lateral dissection utilized by most laparoscopic and robotic surgeons, high ligation is technically easier and provides excellent mobilization for tension-free anastomosis. So long as the surgeon is aware of the anatomic variants and carefully evaluates blood flow to the proximal limb of the anastomosis, either level of ligation can be safe and effective.

The rectum is supplied by (1) the superior hemorrhoidal artery, a continuation of the IMA after it crosses the left iliac artery, (2) the occasionally absent or unilateral middle rectal artery from the internal iliac or pudendal artery, and (3) paired inferior rectal arteries from the internal pudendal artery (terminal branch). Sudeck's point describes the area between last sigmoidal branch and superior hemorrhoidal artery, supplied by anastomotic vessels. This is classically described as a watershed area, but studies have shown that the entire rectum can be supplied by the inferior hemorrhoidal arteries alone. ¹⁵ Terminal branches enter the rectal wall perpendicularly; hence, mesorectal division should be essentially at the same level as rectal division to avoid devascularizing the distal staple line.

Local Perfusion

While diligent attention to regional anatomic principles is important, the ultimate determinant of anastomotic perfusion is local blood flow at the cut margins. Hence, an empiric assessment of local perfusion is warranted. Traditional methods of assessing bowel viability, including color, bleeding (incidental or provocative), palpable pulse, and temperature can be subjective and unreliable. Neither trainees nor experienced surgeons in a large academic hospital were able to predict anastomotic leak with any appreciable accuracy, demonstrating sensitivity of 38% for colonic and 62% for colorectal leaks. ⁴²

Objective measures have been employed since the 1970s, including Doppler ultrasound, ^{43 44} oxygen spectroscopy, ^{45 46} laser Doppler flowmetry, ^{47 48 49 50} and most recently fluorescence angiography. ^{51 52 53 54 55 56} The ultimate question is whether these technologies, beyond providing an additional predictive variable, alter the surgical plan to circumvent anastomotic leak. In a prospective study of 200 consecutive open left-sided colectomies and LARs, blood supply was assessed using a Doppler ultrasound, prompting revision of resection margin in 5%, resulting in an anastomotic leak rate of 2.5% (only 1% clinically detected). ⁴⁴ A quantitative study of intraoperative laser doppler flowmetry in 55 patients undergoing LAR for rectal and distal sigmoid cancer demonstrated that a 10% reduction in blood flow in either the descending colon or rectal stump increased the leak rate to 33%. ⁵⁰ However, compensatory intervention for low blood flow was only triggered in two patients. Considering these modest findings, Doppler ultrasonography is not worthwhile for routine application but could be useful in select cases.

The most recent enthusiasm for intraoperative perfusion assessment centers on near-infrared (NIR) fluorescence angiography using indocyanine green (ICG) dye ( Fig. 2 ). ICG is a water-soluble tricarbocyanine molecule that binds to plasma proteins in the vascular system and undergoes rapid clearance in the bile in a matter of (3–5) minutes. It has an excellent safety profile and is safely used in patients with iodine allergy, despite its 5% iodide content. It absorbs light at 807 nM wavelength and emits fluorescence at 822 nM and is detectable by a growing number of surgical platforms, including SPY/PINPOINT (Stryker), IMAGE1/SPIES/OPAL1 (Karl Storz), and Firefly (Intuitive) in the laparoscopic/robotic realm. The first evidence for intraoperative utility of NIR-ICG in colorectal anastomosis was a prospective single-arm trial in 402 patients that showed laser fluorescence angiography reduced anastomotic leak by 60% from 7.5 to 3.5% versus a historical control from the same institution. ⁵¹ The reduction was most prominent in high-risk groups, including patients older than 70 (11.9–4.3%), and handsewn anastomoses (8.5–1.2%). Similar results were found in robotic LAR, where use of the Firefly technology led to adjusted proximal resection margin in 3 of 16 patients and a reduction in anastomotic leak rate of 6 versus 18% in retrospectively matched controls among a high-risk population with a preponderance of neoadjuvant radiation and ultra-low anastomosis (<5 cm from the anal verge). ⁵²

Fig. 2.

(Left) Robotic-assisted colorectal anastomosis prior to firing the circular end-to-end stapler. (Right) Near infrared light view using Firefly after injection of ICG demonstrates adequate blood flow (bright green) to the proximal and distal limbs of the anastomosis. Photos courtesy: Marco Ferrara, MD. ICG, indocyanine green.

The Perfusion assessment In Laparoscopic Left-sided/Anterior Resection (PILLAR II) trial, a prospective, multicenter, single-arm trial of 139 patients using the PINPOINT platform that demonstrated an extremely low anastomotic leak rate of 1.4%. ⁵⁴ The protocol included preanastomotic laparoscopic assessment of the cut margins and postanastomotic endoscopic assessment which prompted intervention in 8% of patients, none of whom developed subsequent leaks. The overall morbidity was 17% and neither of the two patients who leaked required surgical drainage or diversion. Despite these very promising results, the trial had several notable limitations. First, there was lack of standardization in terms of surgical approach. Second, the patient population comprised mainly benign indications with only 11% receiving neoadjuvant radiation; no patients with ultra-low anastomosis (<5 cm from the anal verge) were included. Finally, most cases that required open conversion were excluded from the analysis. The subsequent PILLAR III randomized controlled trial was concluded early after failing to reach accrual. ⁵⁶ The study population ( N  = 347) was limited to LAR for rectal neoplasm within 10 cm from the anal verge with 65% of patients undergoing neoadjuvant chemoradiation. However, it failed to show a benefit to NIR-ICG with a leak rate of 9.0 versus 9.6% in the control group.

As a final point, regardless of recent data, NIR-ICG technology remains a safe, widely available, multifunctional tool with growing applications across many surgical disciplines. In addition to perfusion assessment, it may be used for tumor localization (in lieu of ink), fluorescence lymph node mapping, and ureteral identification (avoiding indwelling catheters). ICG interpretation does not significantly prolong surgery, especially in the robotic platform, where NIR visualization (Firefly) can readily be toggled at the console. And while it may not reduce anastomotic leak in experienced hands, it behooves the surgeon to employ or at least be familiar with as many adjunctive techniques as possible. ICG is stocked as a lyophilized powder of 25 mg. It is the practice of the authors to dilute it in 10 mL of saline and ask the anesthesiologist to infuse it in aliquots of 2.5 mL, followed by 10 mL flush, delivering a dose of 6.25 mg, or approximately 0.06 mg/kg to the average 100-kg patient. A variety of protocols have been published, ranging from 0.1 to 0.3 mg/kg or 5 to 15 mg bolus. The maximum safe dose is 2 mg/kg. Vascular illumination is rapid, within 5 to 30 seconds and short lived, while colonic illumination occurs within 1 to 2 minutes and lasts 20 to 30 minutes. The authors strongly recommend enabling NIR visualization prior to administration of ICG, particularly in subsequent doses after the initial bolus. Venous congestion can at times interfere with the ability to rely on ICG illumination when evaluating for adequate perfusion. At the current time, the assessment remains qualitative (subjective), but efforts to quantify and standardized assessment for colorectal anastomosis are underway. ^{57 58}

Tension

The concept of minimizing tension predominates the literature surrounding colorectal anastomosis but, to the authors' knowledge, formal evidence is limited to a single study published in 1986 concerning stapled anastomosis in dog intestine. ⁵⁹ Using contrast microangiography, the authors demonstrated that the colon was much more sensitive to shear stress than the small intestine, with marked perfusion deficits at low pressures. Understanding the Law of Laplace (T = Pr for a cylinder), it is expected that the larger caliber colon experiences higher wall tension, thereby impeding blood flow, than the small intestine. Coupled with the passage of bulky solid feces and higher peristaltic forces, it is natural from a purely mechanical perspective that the colon should be at higher risk for anastomotic disruption. Beyond these theoretical concerns, the real problem with the concept of anastomotic tension is the lack of a feasible means to measure it intraoperatively. ⁶⁰ When forming a colorectal anastomosis, the proximal colon should ideally flop into the pelvis spontaneously, or, at minimum, not recoil toward the spleen. But short of these extremes, we are left to judge tension by feel, or worse, visual cues from a haptic free robotic console.

The more appropriate, related concept is reach. Every colorectal surgeon should be intimately familiar and facile with the well-described series of maneuvers used to attain reach ( Fig. 1 ). These include (1) mobilization of the descending colon, (2) mobilization of the splenic flexure, (3) high ligation of the IMA, and (4) high ligation of the IMV. In more extreme circumstances, advanced maneuvers such as (5) creating an ileal mesenteric tunnel or (6) derotation of the ascending colon (Deloyer's procedure) may be necessary. These techniques fall outside the scope of this guide, excepting some brief remarks on the practice of routine splenic flexure mobilization. Once considered surgical dogma in the era of open LAR, splenic flexure mobilization has become far less frequent since the practical challenge of performing it laparoscopically and robotically inadvertently led to the discovery that it is not always necessary. ⁶⁰ Many considerations, such as the superior perfusion of the descending colon, the inferiority of the sigmoid conduit, the oncologic advantage of longer specimen with greater lymph node harvest, and attendant risks of bleeding and splenic injury have not been borne out by objective data. ^{61 62 63} The only certainty is that the surgeon who employs routine splenic flexure mobilization will spend more time, approximately 30 minutes in the operating room. However, inexperience in the mobilization of the flexure can lead to difficulties when it is requisite in an operation. As long as the resection margin is not sacrificed to facilitate reach, the individual surgeon may arrive at their own philosophy.

Microbiome

Interest in the microbiome has surged recently, as mounting evidence from disparate corners of biology and health science are demonstrating broad implications of the host-microbiome interaction in metabolic disease, immunity, infection, wound healing, and cancer treatment. Whether the key to immortality has been hiding in our feces or we are merely sifting through it because we can remains to be seen. After “unlocking the human genome” in some ways failed to deliver on its promises thus far, ⁶⁴ the lens of next generation sequencing has been turned to unlocking the microbiome. With widely available high-throughput technologies to process DNA, RNA, and protein, we have the capacity to characterize not just the microbiota or their environment but describe how they respond to it, communicate with it, and interact with one another in their native state or after injury. Afterall, the mechanics of cutting and carefully reconnecting the bowel are only half of the puzzle; any discussion of colorectal anastomosis without elaboration of the cellular and biochemical dynamics of healing would be incomplete. In the same vein, not recognizing the contributions of pathogenic bacteria in the breakdown of a hard-fought, meticulous surgical anastomosis is a recipe for perpetual frustration.

Intestinal wound healing progresses through much of the same canonical pathway as skin, involving inflammatory, proliferative, and remodeling phases comprising neutrophil and macrophage migration, deposition of new collagen, and tissue organization, respectively. Compared with the small intestine, colon healing is protracted, requiring roughly 4 months to reach 75% initial strength. ⁶⁵ Mucosal disruption triggers migration and hyperplasia of epithelial cells from the margin within 3 days. ⁶⁵ Even sooner, however, within hours of ischemia-reperfusion injury, goblet cells secrete their mucin stores to reconstitute the protective mucus layer of bowel wall, an impermeable barrier to bacterial translocation. A genetically engineered mouse model suggests that without mucin, anastomotic breakdown is inevitable. ⁶⁶ Intestinal bacteria are also integral to the healing process. While early crude studies in canines demonstrated that antibiotic irrigation could prevent anastomotic leak, ⁶⁷ we now understand that commensal bacteria promote reepithelialization through at least two important mechanisms. First, mucosal disruption exposes toll-like receptors to bacteria, triggering release of interleukin (IL)-6 and heat-shock protein which promote epithelial proliferation and angiogenesis. ⁶⁸ Second, commensal organisms also metabolize fiber into short chain fatty acids, the primary colonocyte fuel which in turn stimulate mucin release and cell growth. ^{69 70} In fact, butyrate enemas have been shown to enhance anastomotic strength in rats. ⁷¹ Unfortunately, while fiber deficiency is deleterious to gut immunity, fiber supplementation is not directly beneficial. ⁷²

While some intestinal bacteria are beneficial, others are implicated in the pathogenesis of anastomotic leak. The insult of surgery, not just direct trauma, ischemia, and reperfusion injury but also prolonged fasting, mechanical bowel preparation, oral and parenteral antibiotics, and opioids, induces a shift from a health-promoting to a disease-promoting microbiome. ^{70 73} This “dysbiosis” comprises a loss of diversity and abundance of commensal organisms, such as Bacteroides fragilis (reduced by 90%), with a relative “bloom” of pathogenic strains, such as Pseudomonas aeruginosa and Enterococcus faecalis . ⁷⁴ More than a shift in populations, the physiologic stress prompts a phenotypic switch with expression of virulence factors including the elaboration of collagenase, adhesion molecules, biofilm, and antibiotic resistance that contribute directly to anastomotic dehiscence. ^{75 76 77} Host-derived (human) collagenase activity is especially pronounced in the colon (vs. small bowel) after injury. ⁷⁸ That is to say, in the first days after surgery, the anastomosis is actively degrading itself, relying entirely on the staples or sutures for strength. ⁶⁵

With our growing knowledge, the hope is to design meaningful intervention. Maintaining the gut mucosal barrier has implications for distant organ homeostasis. Early enteral feeding has reduced morbidity, infection rates, and mortality in the critically ill and at-risk (malnourished) populations. ⁷⁴ But our ability to manipulate the microbiome has not progressed beyond killing it indiscriminately with antibiotics. And the burgeoning field of immunonutrition has yet to demonstrate reproducible advantages to oral supplements, such as glutamine, branched-chain amino acids, arginine, omega-3-fatty acids, probiotics, or prebiotics. ⁷⁴ It cannot be overlooked that these readily absorbed elemental substrates may not be reaching their targets in the hindgut.

Our primary tool, bowel preparation, has undergone its own torrid history of controversy. ⁷⁹ The premise behind bowel preparation is bacterial dilution. Nichols et al ⁸⁰ published their landmark randomized controlled trial demonstrating combined mechanical bowel preparation with nonabsorbable oral antibiotics reduced surgical site infections compared with mechanical prep alone. After 30 years of universal acceptance, the benefit of bowel preparation was called into question by a series of studies which omitted oral antibiotics. After two decades of trials, large database reviews, and meta-analyses, we have come full circle to the same truth that Nichols elucidated half a century ago: combined mechanical and oral antibiotic preparation is our best option. Compared with no prep, antibiotics alone, or mechanical prep alone, Nichols' prep reduces surgical site infections, length of stay, readmission rate, and anastomotic leak. It is supported by the latest ASCRS Guidelines with grade IB recommendation based on moderate-quality evidence. ⁸¹ But the idea of washing the colon and indiscriminately killing its flora (90% of which is “beneficial”) should cause some degree of skepticism if not outright alarm. The case is certainly not closed and the hope is that a more sophisticated understanding of the microbiome afforded by next generation sequencing will pave the way for “smart prep” that not only reduces morbidity but enhances symptom-free anastomotic healing even when mechanical disruption occurs. ⁷⁹

Anastomotic Construction

Stapled End-to-End Anastomosis (Standard Double-Stapled Technique)

The most common configuration for colorectal anastomosis in the era of minimally invasive surgery and advanced surgical staplers is end-to-end using a circular stapler. The DST was introduced by Knight and Griffen ²¹ in 1980 to obviate the technical challenge of sewing a purse string in the deep, narrow male pelvis during open LAR. In a cohort of 75 patients who included two-thirds elderly, two-thirds cancer, and one quarter low anastomosis (within 6 cm of the anal verge), the leak rate was limited to 2.7%. Over the following 40 years, DST has been thoroughly validated, widely adopted, modified, and adapted to minimally invasive surgery. But it has seldom been improved. The published anastomotic leak rate commonly ranges from 3 to 20% ^{9 50 54 82} with a median of 7.3%. ⁸³ Depending on the study population, surgical technique, and proximity to the anus, it may be as low as 0.7% ⁸⁴ or exceed 30%. ⁸⁵ The five major risk factors for leak after DST are low anastomosis/tumor location, multiple staple fires for rectal transection, male sex, prolonged operative time, and blood transfusion, ^{83 86} not a single modifiable factor among them. The chief determinant of leak is clearly the capacity to dissect the distal rectum, underscoring the stark demand for excellent surgical technique. Unsurprisingly, individual surgeon has been identified as an independent risk factor for leak. ⁸⁷

The surgeon must be familiar with the DST and its variations. The rectum should be dissected circumferentially in the avascular plane, permitting cephalad traction out of the pelvis to optimize the angle of attack for the stapler. The mesorectum is divided perpendicular to the axis of the rectum, exposing the bare wall of the rectum circumferentially. A reticulating powered stapler, either 45- or 60-mm in length, is used to transect the rectum along a perpendicular axis ( Fig. 3 ). From a right lower quadrant port, a single fire at times is not sufficient to achieve a transverse resection. Subsequent fires should maintain the same perpendicular line to limit overlap, provide even blood supply to the rectal stump, and keep the staple line as short as possible. Alternatively, the rectum may be divided vertically through the right lower quadrant port, a suprapubic port, the Pfannenstiel incision, or a miniature laparotomy. ^{88 89 90} This is often necessary when stapling deep in the pelvis for a low rectal transection. Curved linear cutting staplers or linear noncutting staplers may also be used with the open technique. Three firings of the stapler or in some studies as few as two has been shown to increase the risk of anastomotic failure. ^{83 86} In fact, on October 8, 2021, the U.S. Food and Drug Administration officially reclassified surgical staplers as class II devices (special controls), subject to premarket review, in part due to mounting evidence of increased anastomotic leak rates with multiple stapler firings. ⁹¹

Fig. 3.

Perpendicular transection with the linear powered stapler. Due to the native width of the rectum, a single firing will be required. Photo courtesy: Marco Ferrara, MD.

After mobilization of the proximal colon and distal transection have been accomplished, the proximal colon is typically extracorporealized through either the Pfannenstiel or lower midline incision. It is our practice to use a wound protector and drape the field with towels to limit contamination. It is useful at this juncture to pack a laparotomy sponge in the pelvis for passive hemostasis. The proximal resection margin is selected based on oncologic considerations and viability. After clearing a mesocolic window, the colon is divided sharply, typically over an automatic purse string device. Alternatively, a reusable purse string clamp can be used with a 2–0 monofilament suture (e.g., Nylon) on a straight needle. A third option is hand sewing the purse string once the anvil is installed through the open end of the colon. The speed advantage of automated purse string devices is weighed against their capacity to misfire or fail. No matter which method is employed, it is of the utmost importance to ensure that the purse string captures the full thickness of the layers of the colon wall. Any unrecognized gaps, slippage, or serosal tears can lead to anastomotic failure. Defects or diverticula should be imbricated inside the staple line. Redundant epiploic and mesenteric fat may be trimmed from the anastomotic “landing zone” to reduce the amount of tissue involved in the stapler-anvil interface but care should be taken to avoid devascularizing the colon. Literature search did not identify any publications remarking on this practice. With increasing robotic experience, purely intracorporeal techniques have emerged, avoiding the need to externalize the colon. ^{92 93} The anvil is introduced into the abdominal cavity and secured with a purse string ( Fig. 4 ).

Fig. 4.

Anvil secured to the proximal limb (colon) with purse-string suture during intra-corporeal Double-Stapled Technique. Photo courtesy of Marco Ferrara, MD.

The colon is returned inside the pelvis and the anastomosis completed in open, laparoscopic, or robotic fashion. Ideally, the anvil falls readily into the pelvis and rests against the rectal stump. The mesocolon is oriented medially, using the taeniae as a guide to ensure no inadvertent torsion has occurred. After passing sizers through the rectum and gently dilating the anus, the stapler is brought up to the rectal stump. The trocar is delivered adjacent to or directly through the staple line, with no evidence favoring one practice over the other. The anvil is coupled to the trocar and the stapler is closed, avoiding inadvertent inclusion of epiploica or other organs in the staple line. The stapler is tightened and fired according to the manufacturer's specifications. Intimate familiarity with the nuances of the brand and format of stapler is necessary to avoid disruptions and misfires. One of the more devastating complications is incorporating the vagina into the staple line, causing iatrogenic rectovaginal fistula which may go unrecognized for some time. ⁹⁴ This complication can be avoided by completely dissecting the rectovaginal septum. A sizer may be placed inside the vagina to help delineate the anatomy. Passing the stapler trocar posterior to the rectal staple line may reduce the risk. A digital vaginal examination should also be performed prior to firing the stapler to rule out inadvertent inclusion of vaginal mucosa.

Variations of Stapled End-to-End Anastomosis

Despite the advantages of automation, staplers have failed to improve on historical anastomotic leak rates. This disappointing reality has spurred modifications of the standard DST. One straightforward variation technique is the triple-stapled technique (TST). In lieu of using a purse string, which can be unreliable and cumbersome, the proximal end of the colon is resealed with a stapler after the anvil is passed proximally. The anvil is subsequently delivered through the end of the colon adjacent to the staple line via its detachable spike or with electrocautery. The anastomosis then proceeds normally. TST has excellent outcomes with anastomotic leak rate as low as 1 to 2% in case series, ^{82 95} comparing favorably with 6% leak rate in historical controls using DST. The added cost of an additional staple fire is offset by reduced operating time.

Detractors of the DST point out that size mismatch between the rectal staple line and the circular stapler produces intersecting staple lines and redundant “dog ears” which have each been implicated in the development of anastomotic leak. Although an experimental model of colon anastomosis in pigs demonstrated intersecting staple lines do not significantly limit perfusion, as one might imagine, ⁹⁶ another porcine model has shown that the formation of “dog ears” significantly reduces the bursting pressure in side-to-end colorectal anastomosis. ⁹⁷ Furthermore, while buttressing the double-stapled anastomosis does not increase its bursting pressure, the intersection of the “EEA” and “Endo GIA” staple lines was a specific weak point. ⁹⁸ With enough empiric and experimental evidence to generate interest, several workable solutions have emerged. The simplest stratagem, with some direct evidence to support it, is deploying the trocar off-center to reduce the staple line intersections from two to one. Lee et al ⁸⁴ devised an invagination technique, whereby both potential dog ears were eliminated; in most cases, by impaling them on the trocar spike. This technique yielded the lowest anastomotic leak rate in recently published literature, that is, 0.7%. Out of 128 patients, the one who leaked had two staple line intersections.

Another prominent option is reversion to the single-stapled technique (SST) which employs purse-string sutures in the colon and rectum. This technique has regained favor with the advent of robotic surgery which simplifies suturing in the deep pelvis. ⁹⁹ Although randomized controlled trials are scarce, a meta-analysis comprising 10 studies comparing DST versus SST established equivalence regarding anastomotic leak (11.2 11.7%), stricture, and overall complications. ⁸³ However, DST was significantly faster at 167 versus 204 minutes. Another elegant solution is the omega suture ( Fig. 5 ), first described by Asao et al ¹⁰⁰ in 2002. A horizontal mattress suture placed through the rectal staple line is tightened against the shaft of the stapler trocar, drawing the straight staple line into an omega shape within the cutting circle of the stapler, thereby preventing “dog ears” and intersecting staple lines altogether. Unlike the purse-string technique, it avoids fecal spillage. A new retrospective review of DST versus SST, ¹⁰¹ comprising 30% classical and 70% omega suture, demonstrating a reduction in anastomotic leak from 9 to 3% ( p  = 0.045) and stenosis from 6 to 1% ( p  = 0.037) in favor of SST.

Fig. 5.

Omega stitch for the single-stapled technique. ( A ) A horizontal mattress suture is placed in the staple line. ( B ) The trocar is delivered anterior to the staple line and the suture is secured in front of (not around) the anvil. ( C ) When the suture is tied, the linear staple line is drawn into an omega shape inside of the cut line of the circular stapler, thereby avoiding intersecting staple lines. Image courtesy of Jay Vuitch.

Technical Considerations

The surgeon should be familiar with the various staple devices and their subtle differences. Circular staplers intended for colorectal anastomosis are manufactured by Covidien, a subsidiary of Medtronic (Dublin, Ireland), and Ethicon, a subsidiary of Johnson & Johnson (New Brunswick, NJ). Both companies offer two generations of staplers with incremental improvements in a variety of diameters.

Covidien offers the end-to-end anastomosis (EEA) with the Directional Stapling Technology Series (released 2010) and more recently the EEA with the Tri-Staple Technology (released 2020). The Directional Stapling Technology series includes staples with rectangular wires designed to bend more reliably. It comes in five diameters (21, 25, 28, 31, and 33 mm) and two fixed staple heights, 3.5 and 4.8 mm which crimp down to 1.5 and 2.0 mm, respectively. The Tri-Staple Technology offers graduated open staple heights of 3.0, 3.5, and 4.0 mm from inner to outer row for the medium thickness (purple) load and 4.0 to 5.0 mm for the extra thick (black) load. Both shaft lengths are offered. Both lines offer 22-cm standard and 35-cm XL shaft lengths. Another feature is the tilt-top anvil that streamlines removal (Covidien).

Ethicon offers the PROXIMATE Curved Intraluminal and Echelon Circular Powered staplers. The PROXIMATE series comes in four diameters, that is, 21, 25, 29, and 33 mm, corresponding with internal lumens of 12.4, 16.4, 20.4, and 24.4 mm, and offers adjustable closed staple height from 1.5 to 2.2 cm which can be set by the surgeon to accommodate to a range of tissue thickness. The shaft length is 26 cm (white) and 37 cm (black). The powered version was introduced in 2019, featuring an automated trigger mechanism that obviates the need to apply force and thus potentially reduces motion during firing. It comes in 21, 25, 29, and 31 mm diameters and only the 24-cm shaft length. It also accommodates variable staple heights ranging from 1.5 to 2.2 cm and introduces an “improved gripping surface” and “3D stapling technology” with offset staple legs designed to distribute compression and “reduce potential leak paths” (Ethicon). A propensity-score matched cohort study of consecutive patients undergoing left-sided resection with colorectal anastomosis above 5-cm from the anal verge demonstrated reduction of anastomotic leak rate from 11.8 to 1.7% ( p  = 0.022) with the introduction of the powered stapler. ¹⁰²

Stapler diameter is an important consideration. The association between stapled anastomosis and stricture has been linked to the size of the lumen, ^{103 104 105} although investigation in patients with ileal pouch-anal anastomosis did not show an effect on anastomotic leak, stricture, long-term functional outcomes, or quality of life. ¹⁰⁴ The common stapler diameters used in colorectal anastomosis range from 28 to 33 mm. A review of 473 cases showed a significant difference with 7.1% anastomotic stricture with 28 to 29 mm staplers versus 2.1% with 31 to 33 mm staplers ( p  = 0.007) with no significant difference in anastomotic leak, readmission, or reoperation. ¹⁰⁶ Larger staplers were more often used in men. Stapler brand and style may be a matter of preference, but the surgeon should strive to use the largest stapler the anatomy can accommodate.

Handsewn Anastomosis

Long before staplers revolutionized surgery, sutures were the preeminent and only modality for colorectal anastomosis. Today, suture remains a viable option with equivalent outcomes in experienced hands. A handsewn anastomosis may be conducted in end-to-end or side-to-end configuration using one or two layers of sutures. Slieker et al ¹⁰⁷ published a comprehensive systemic review and meta-analysis of techniques for colorectal anastomosis. Polydiaxanone suture (PDS) is preferred, as it possesses all the superior qualities borne out in animal models and randomized clinical trials: monofilament, slowly absorbable with preserved strength, minimal trauma and inflammation, and low bacterial adherence. The bites may be serosubmucosal or full thickness and should invert the anastomosis which produces fewer adhesions at the cost of mild stenosis and reduces the risk of colorectal anastomotic leak versus the everting technique. ¹⁰⁸ The classic two-layer inverting anastomosis described by Czerny in 1880 involving a full thickness internal layer covered by an outer layer of the Lembert sutures is common in practice today. However, multiple animal models have since shown that single-layer anastomosis is superior in regard to inflammation and perfusion. Furthermore, the only randomized clinical trial performed in humans showed reduced leak with anastomosis below the peritoneal reflection; otherwise, the techniques were equivalent among 92 patients. ¹⁰⁹ A single layer is also more expedient and less expensive. The technique involves serosubmucosal bites that exclude and thus invert the mucosa, essentially identical to what Lembert described in 1826. ¹¹⁰ No randomized controlled trials compare interrupted to continuous suture, although animal models show better serosal apposition and blood flow with equivalent burst pressure for continuous. From a practical consideration, continuous suture is technically easier and faster.

The question of handsewn versus stapled anastomosis has been extensively reviewed in the literature. Multiple randomized controlled trials and meta-analyses fail to show a significant difference in anastomotic leak between stapled and handsewn technique, 13 versus 13.4% leak rate which were clinically detectable in 6.3 versus 7.1%, respectively. ^{107 111} Staplers were statistically faster by an average of 7.6 minutes, while sutures reduced stricture from 8 to 2% (OR = 3.59). ¹¹¹ Although sutured technique theoretically increases pelvic contamination, wound infection rates were not different (5.9 vs. 4.3%). Interestingly, a subgroup analysis of anastomoses performed by residents demonstrated increased leak rate in handsewn technique, underscoring the importance of skill and experience. This has led some to advocate for the use of staplers to diminish the role of the surgeon's performance in undesired clinical outcomes. ¹¹² Although standardization is an important part of quality, so long as patients continue to come in different shapes and sizes, surgical proficiency and sound decision making from the individual surgeon are demanded.

Recent advances in robotic surgical technology have rekindled interest in the sutured anastomosis. While not a handsewn technique in the strictest sense, the robotic platform permits precision suturing with uncanny dexterity, as if the surgeon's hands have been restored to their original domain inside the abdomen. Robotic right hemicolectomy with intracorporeal anastomosis has become fairly commonplace with a variety of described techniques, ¹¹³ including totally “handsewn” sutured anastomosis. ¹¹⁴ Intracorporeal colocolonic anastomosis has also been described, ¹¹⁵ using barbed (directional) suture which eliminates the need for tying knots. Although long-term data are lacking, robotic suturing may be even better suited to the narrow confines of the deep pelvis. When performing robotic “handsewn” colorectal anastomosis, it is the author's preference to perform a single layer EEA using 3–0 absorbable barbed sutures. The colon and rectum are first approximated with stay sutures to ensure proper alignment. The posterior wall of the anastomosis is performed in bidirectional fashion from the middle using full-thickness continuous bites, inverting the mucosa. At the corners, the suture is transitioned to a Connell stitch to maintain mucosal inversion, as has been well-described for open handsewn anastomosis. Reinforcing sutures can be placed at any perceived weak points. Anastomotic testing is performed in standard fashion. Provided that good technical principles are followed, robotic suturing is expected to have equivalent outcomes to hand suturing and may obviate some of the potential pitfalls of stapler utilization.

Side-to-End Anastomosis

The side-to-end anastomosis was devised by Baker ¹¹⁶ in 1950 to overcome the size mismatch between the sigmoid colon and rectum (Baker 1950) and bears his eponym. Either stapled or hand-sewn anastomoses may be configured in an end-to-end or side-to-end fashion, with the antimesenteric side of the colon attached to the end of the rectum ( Fig. 6 ). For a high anastomosis, above the peritoneal reflection or beyond the pelvic brim, a side-to-end configuration may sit more naturally, without placing undue tension on the mesentery. The anvil is passed through the open end of the colon and delivered through the antimesenteric wall of the colon, avoiding the need to make a purse string. The end of the colon is stapled closed, and a stapled anastomosis is conducted in the usual fashion. Alternatively, a colotomy is created in the antimesenteric border of the colon, matching the diameter of the rectum. Typically stay sutures are placed through the lateral “corners” of the anastomosis and the posterior wall is closed in one or two layers, followed by the anterior wall. The side-to-end configuration is generally preferred when tension allows, as it affords a wider anastomosis that avoids stenosis and facilitates the “reservoir capacity” of the rectum. Data comparing side-to-end versus end-to-end anastomosis are sparse. One randomized controlled trial of 77 patients with T1 and T2 rectal cancer demonstrated 5% leak rate in the side-to-end group versus 29.2% leak rate in the end-to-end group. ⁸⁵ Although the difference was statistically significant, the unusually high leak rate casts doubt on the validity of this study.

Fig. 6.

End-to-Side Anastomosis. The anvil is placed into the open end of the colon and delivered through a small colotomy in the antimesenteric border. The colon is closed with a linear cutting stapler, 3 cm from the anvil. The side-to-end anastomosis is completed with a circular stapler. Image courtesy of Jay Vuitch.

The Baker anastomosis has also been proffered as a solution to the bowel dysfunction, frequency, urgency, incontinence, and incomplete evacuation caused by total mesorectal excision with straight coloanal anastomosis, collectively referred to as “low anterior rection syndrome (LARS).” The blind-ending colonic limb beyond the colorectal anastomosis may serve as a potential reservoir for stool. A randomized trial of 44 patients investigated the function of short (3 cm) versus long (6 cm) side limbs. ¹¹⁷ Stool frequency, urgency, incontinence, and medication use were not significantly different, and sphincter function was unaffected. Incomplete evacuation as measured by defecography was 25% in the short limb versus 65% in the long, including one occurrence of fecal impaction in the latter. Three leaks occurred, one in the short arm at the anastomotic junction and two in the long arm at the colonic stump. The authors concluded that a short limb is preferred but the findings are generally regarded as equivocal. ^{107 118}

Compression Anastomosis

Although it may sound like a technology of the future, suture-less anastomosis via compression ring was first conceived of in 1826 by Denans and prototyped in 1892 as Murphy's button. ^{119 120} Nearly 100 years passed before compression anastomosis was seriously attempted, with the introduction of the AKA-2 device (Seidel Medipool, Germany) in 1984 ¹²¹ and Biofragmentable Anastomotic Ring (BAR; Valtrac, Norwalk, CT, USA) in 1985. ¹²² Despite clinical efficacy, neither was adopted. ^{123 124} The AKA-2 was designed for transanal application and demonstrated good clinical outcomes with anastomotic leak rate of 2.4, 2.6, and 3% at anastomosis >10, 5 to 10, and <5 cm from the anal verge in 442 consecutive elective and emergent resections of the left colon and rectum. ¹²⁵ The BAR, which comprises interdigitating plastic rings deployed over purse-string sutures in open ¹²² or transanal fashion, ¹²⁶ also demonstrated good clinical results with low anastomotic leak and stenosis rates of 2.8 and 2.1% in a meta-analysis of 10 randomized controlled trials. These outcomes were not statistically better than conventional (handsewn and stapled anastomoses): 3.7% leak and 4% stenosis. However, return of bowel function and length of stay were improved by 1 day, while the risk of bowel obstruction doubled from 2.2 to 4.2% ( p  = 0.03), ¹²⁷ presumably due to the small internal lumen (≤20 mm).

The NiTi Endoluminal Compression Anastomotic Ring (CAR) 27 (“ColonRing”), which became FDA approved in 2006, is the newest and most promising compression anastomosis device. It comprises a plastic anvil secured in the proximal colon and a barbed metal ring delivered on a transanal platform. The stapled rectum is pierced by a trocar that couples to the exposed shaft of the anvil in the proximal cuff. As the instrument is closed, the cuffs of bowel are apposed, impaled on a circle of metal barbs, and the anastomosis is completed by deploying the ring and cutting the intervening tissue, producing familiar “anastomotic doughnuts” which may be checked for completeness. The metal ring contains superelastic Nitinol springs that exert consistent pressure (7.7 Newtons) independent of tissue thickness as they strive to return to their original shape. ¹²⁶ Theoretically, this is superior to fixed-height staples which deliver variable pressure depending on the tissue thickness. The pressure gradually necroses the tissue between the plastic and metal rings, while permitting anastomotic healing of the colon to the rectum just outside. After approximately 8 days, the necrotic tissue sloughs and the ring detaches, passing in the stool. ¹²⁰

The ColonRing intentionally looks and performs like a modern circular stapler, with a detachable anvil that is secured with a purse-string suture, a knob in the handle that screws the trocar outward and inward, a safety latch, and a long trigger. All things considered, the mechanism of this compression anastomosis is not so different from staples: the tissue is held together by metal until wound healing occurs. Contrary to stapled anastomosis, however, the foreign body is expelled, leading to reduced inflammation, scarring, and improved burst pressure in an animal model. ^{128 129} The ColonRing has performed well in clinical studies, demonstrating equivalence to traditional DST regarding anastomotic leak (2.6%), stricture, operative time, obstruction, and 3-year overall survival. ¹³⁰ One subgroup analysis of patients undergoing LAR within 6 cm of the anal verge showed a leak rate of 5.3 versus 12.1% for stapled anastomosis (not statistically significant) with no grade C leaks among 38 patients. ¹³¹ A large multinational, multicenter registry review of 1,180 patients demonstrated 38 leaks (3.22%) with only four technical failures (0.34%). ¹²⁰ Nonetheless, large randomized clinical trials are warranted and will probably need to show superiority if ColonRing is to gain traction. NiTi also manufactures a Compression Anastomotic Clip (CAC) for side-to-side anastomosis which will not be reviewed.

Magnetic devices are the latest novelty in compression anastomosis. Proof of concept was demonstrated in animal models for intestinal and later colorectal anastomoses, using side-to-side or side-to-end configuration. ¹³² A five-patient clinical trial in humans showed “magnamosis” which may be safe and effective for intestinal side-to-side anastomosis.

The equivalent outcomes among this huge array of anastomotic methods suggest that the apparatus used to connect bowel to bowel is not nearly as important as we might suppose. Regardless of the device, the major risk factors for anastomotic leak persist: low position, multiple staple fires, male sex, prolonged operation, and blood transfusion, all indicative of a difficult dissection with the potential for tension and poor perfusion. While emerging technologies look to eliminate foreign materials from the fusion of disparate intestinal limbs, the challenge of bringing those limbs together remains unchanged, and the specter of the narrow, irradiated male pelvis looms large.

Anastomotic Problems

Anastomotic complications are best discovered immediately. A variety of tools have emerged to facilitate intraoperative testing of gross anastomotic failure, supported by a growing body of literature. The three most common techniques are as follows: (1) confirming complete anastomotic donuts, (2) endoscopic inspection of the staple line, and (3) air leak test which is accomplished by submerging the anastomosis underwater (saline) and checking for bubbles. The anastomotic donuts should be expected for integrity and thickness. While completeness is not a guarantee of success, incompleteness or irregularity may be the first sign of failure. While we recommend performing all three tests with every anastomosis, incomplete anastomotic donuts should generate a high index of suspicion and prompt further investigation if not performed routinely. Intraoperatively flexible sigmoidoscopy is an invaluable tool and should be available to assist with every colorectal surgery. It may be used to guide rectal dissection, ensure adequate margin, or to examine the fresh anastomosis. It was recently validated in a meta-analysis, ¹³³ demonstrating increased identification of anastomotic defect (OR = 5.21), as well as decreased incidence of anastomotic leak (OR = 0.45) and anastomotic bleeding (OR 0.40). Endoscopy also facilitates measurement of the distance between the anastomosis and anal verge which is important for internal housekeeping, complete documentation, and compliance with National Accreditation Program for Rectal Cancer (NAPRC) standards. Air leak test is commonly used and may be performed blindly with a syringe or under direct visualization with an endoscope. Evidence suggests the latter is superior. ¹³⁴ Although routine air leak test has not consistently shown reduction in anastomotic leak, ^{135 136 137} it provides the opportunity for immediate intervention.

Anastomotic Leak

Any red flags during anastomotic testing should prompt consideration of repair, revision, and/or proximal diversion. Classically, bubbling during an air leak test can be used to locate the defect and repair it with suture. Depending on the level of the anastomosis, suture repair may be limited for posterior leaks, necessitating anastomotic revision. A retrospective noninferiority study ¹³⁸ of 2,360 patients with positive air leak test in 119 showed 9% leak rate for primary repair versus 0% leak rate for diversion or reconstruction. This data suggest that suture repair alone is inadequate and should either be combined with proximal diversion or abandoned in favor of formal revision. Proximal diversion permits spontaneous healing even when suture repair is omitted, but formal revision should be considered when the bowel reach permits, to avoid making a stoma out of convenience. The surgeon weighs these considerations against the feasibility of redoing the anastomosis, which can be extremely challenging in the setting of ultra-LAR.

Bleeding

Although rare, profuse bleeding from the staple line may occur immediately or in delayed fashion. When bleeding is identified intraoperatively, the endoscope offers a wide gamut of possibilities, including topical hemostatic agents, epinephrine injection, or endoscopic clips. Pulsatile arterial bleeding is best treated with transabdominal suture or endoscopic clips. Mild oozing is usually self-limited and a welcome sign of good perfusion. Early detection via intraoperative endoscopic surveillance can prevent major hemorrhage, hypotension, need for blood transfusion, and subsequent anastomotic dehiscence.

Adjuncts

Diversion

The choice of proximal diversion is an important one. Unfortunately, it is far from simple. The subject has been reviewed ad nauseum, including two recent publications in this very same journal. ^{139 140} If the goal were merely to reduce the risk of leak, the choice would be simple: divert indiscriminately. More pertinently, even if we were to accept a leak rate as high as 20%, 8 out of 10 patients would derive absolutely no benefit from diversion, not to mention the social implications, psychological tolls, physiologic strains, and surgical dangers. Hence, discerning voices continually clamor about the overuse of proximal diversion. ^{141 142 143} One problem is we have no way of accurately predicting who will leak. ⁴² The other is that the penalty for being wrong can be utterly devastating. ^{7 8 9}

The vast majority of evidence on the subject comes from retrospective studies either investigating the risk factors for leak ^{7 144 145 146 147} or directly comparing leak rates between patients who were and were not diverted, ^{141 148} including several systemic reviews ^{86 124 142 149 150} and meta-analyses. ^{9 151} While a majority of studies correlate lack of diverting stoma with increased risk of leak, many studies show equivalent outcomes, and one very large (72,055 patient) nationwide database review of anterior resection for rectal cancer showed increased incidence of anastomotic leak in laparoscopic surgery with diverting loop ileostomy. ¹⁴⁷ The outcomes following anastomotic leak are far less equivocal, demonstrating a significant advantage in terms of morbidity, reoperation, and even mortality for diversion. ^{9 139 140 151} When the choice of diversion is left to the operating surgeon, an intrinsic selection bias is created ^{6 9 145 148 151 152} as follows: patients selected for diversion have implicitly higher risk, either by intangible, subjective assessment or objective measure, such as positive air leak, blood transfusion, or ultra-low anastomosis. That they have equal or lower leak rate could be testament to the efficacy of diversion.

The question is better answered by randomized controlled trials of which five have been conducted to date. ^{6 153 154 155 156} Their findings are unequivocal and well summarized in a recent meta-analysis, ¹⁵² with an excellent critique on the heterogeneous experimental design, inclusion criteria, randomization, and procedures performed. All trials were exclusive to resection for rectal cancer but differed somewhat on the level and configuration of anastomosis. Regardless, all five demonstrated a significant reduction in clinical anastomotic leak and reoperation, with pooled ORs of 0.292 and 0.219, respectively. Whether this concludes the debate, we leave to the reader.

Occasionally, the choice of diversion is more straightforward: patients with intraoperative complications, immediate anastomotic failure, ¹³⁸ ultra-LAR (<5–6 cm from the anal verge), coloanal anastomosis, ileoanal pouches, or chronic steroid use. ^{139 140} The preferred method of diversion is also fairly clear, while functionally equivalent to a diverting colostomy, loop ileostomy exhibits less wound infection, postoperative ileus, length of stay, prolapse, and hernia. Reversal is considerably easier than for colostomy, requiring laparotomy in only 3.7% of cases. ¹⁴³ Most complications of ileostomy are mild, related to uncontrolled ileostomy output causing skin excoriation, dehydration, electrolyte disturbances, and acute kidney injury, sometimes prompting readmission. ¹⁴⁰ The most common serious complication is small bowel obstruction which occurs in 7.2% of patients after reversal. ¹⁴³ In roughly one-fifth, the “temporary” ostomy is permanent, ^{6 8 139} albeit considerably lower than permanent stoma rate in patients with unprotected leaks (25–68%). ^{6 138 139 144}

Ultimately, the surgeon must take into consideration not only the risk of leak but also the severity of the consequences. A healthy patient with high risk of leak might forego diversion and escape with minimal long-term consequences after suffering a leak. Conversely, an elderly patient with a low risk of leak may not survive one if diversion is deferred. A surgeon should never be afraid to bring up an ostomy if there is doubt, nor should a patient enter the operating room without the express knowledge and consent that an ostomy may be performed. Although the concomitant risks of diverting loop ileostomy are not negligible, especially among the frail, elderly, or nephropenic, ¹⁴⁰ they are heavily outweighed by the devastating consequences of anastomotic leak: emergent laparotomy, sepsis, death, anorectal dysfunction, permanent ostomy, local tumor recurrence, metastasis, and poor long-term survival. ^{6 8 9 10 11 12 141 152} Ultimately, final decision should incorporate the risk tolerance of the surgeon and the patient.

Transanal Tubes

Transanal drainage tubes have been explored as an alternative to diverting stoma for the protection of tenuous anastomoses, such as in LAR. Transanal tubes are placed intraoperatively across the anastomosis, secured at the perianal skin, and connected to a collection bag. The purported mechanism is fecal diversion and pressure reduction. ^{157 158} They have gained some popularity in Asia, where early studies have shown some promise. More recent data are less convincing, however. A network meta-analysis of 26 retrospective cohort studies and 6 randomized controlled trials demonstrated equivalent reduction in anastomotic leak rate for transanal tubes (OR = 0.42) and diverting stoma (OR = 0.51) when compared with no prevention. ¹⁵⁷ However, only diverting stoma significantly reduced reoperation rate (OR = 0.10). These findings were corroborated by a large randomized controlled trial on laparoscopic LAR without neoadjuvant radiation therapy. ¹⁵⁸ The anastomotic leak rate was 6.4% with a drain and 6.8% without which was not significantly altered by presence of a stoma. The drains were associated with anal pain in half of patients and associated with premature removal secondary to patient discomfort, accidental dislodgement, and dysfunction in 7.1%.

Drains

Pelvic drainage was formerly commonplace in colorectal anastomosis, as presacral hematoma or seroma was believed to provide a fertile milieu for bacterial growth which might secondarily trigger anastomotic disruption. ¹⁴⁵ Clinical studies displayed discordant findings, with some, such as the Dutch TME trial, ¹⁴⁵ showing prevention or mitigation of leaks and others showing no effect. Detractors point out that drains may offer a false sense of security or even contribute to anastomotic leak. ¹⁵⁹ A meta-analysis of 11 randomized controlled trials comprising 1,803 patients showed no statistical difference between routine drainage and nondrainage on anastomotic leakage (overall, clinical, and radiologic) or mortality, wound infection, reoperation, and respiratory complications. ¹⁶⁰ As prophylactic drains are being phased out across a wide variety of abdominal procedures, we advocate for meticulous hemostasis and symptom-driven, image-guided drainage postoperatively in lieu of prophylactic drainage.

Temporary percutaneous ileostomy drains have been proposed as a substitute for formal diversion. ^{161 162} An 18- or 20-Fr jejunostomy tube is inserted into the proximal limb of a potential ileostomy. The balloon is inflated to obstruct the lumen and the tube is secured with a purse string and brought out through a laparoscopic port site. A small cohort study of 143 consecutive patients undergoing laparoscopic LAR demonstrated low anastomotic leak rate (1.3%) versus conventional loop ileostomy (5.9%; p  = 0.191) and no complications, with a median decannulation on postoperative day 9. A randomized controlled trial was proposed in 2014, ¹⁶³ but no updates have been reported. A slightly modified version using an endotracheal tube and a single row of staples for temporary occlusion of the efferent limb showed safety and efficacy equivalent to diverting loop ileostomy. ¹⁶² The ileum recanalized spontaneously after approximately 30 days. Ileostomy drains remain experimental ( Table 1 ).

Table 1. Summary of controversial topics in colorectal anastomosis.

Technique	Evidence/recommendation
High vs. low ligation	Equivalent oncologic and functional outcomes. Do what is practical
NIR-ICG	No proven advantage, but not inconvenient robotically
Splenic flexure mobilization	Selective
Microbiome	No actionable targets currently
Rectal transection	Limit stapler firings when able
Intersecting staple lines	Mixed data, limit if feasible
Single vs. double vs. triple staple technique	Equivalent outcomes in experienced hands
Stapled versus sutured anastomosis	Equivalent leak rate. Decreased operating time with staples. Decreased stenosis with sutured
Configuration (end-to-end, side-to-end, J-pouch)	Equivalent leak rates and long-term function/LARS
Stapler diameter	At least 31 mm if feasible
Robotic “handsewn” technique	No data but promising
Compression anastomosis	Potentially equivalent but need RCTs vs. more conventional methods
Intraoperative Flex Sig	Sensitive, reduces leak and bleeding
Positive air leak test	Divert ± repair or redo
Proximal diversion	Indicated for intraoperative complications, low anastomosis, immunosuppression/steroids, or patient unable to tolerate leak
Loop Ileostomy vs. Colostomy	Similar function, ileostomy easier to reverse
Transanal tubes	Inferior to diverting loop ileostomy
Pelvic drainage	Sufficient evidence to say no benefit
Temporary percutaneous ileostomy drain	Experimental

Abbreviations: ICG, indocyanine green; LARS, low anterior rection syndrome; NIR, near infrared; RCT, randomized controlled trial.

Conclusion

After 75 years of refining the colorectal anastomosis, many of the challenges that plagued our predecessors persist. The manner in which we operate has changed considerably but so has the matter on which we operate. Despite our myriad technologic innovations, we are still beleaguered by the arduous dissection of the rectum. And the score, insofar as anastomotic leak rate reflects our progress, remains the same.

In some sense, our focus on the proximate problem, the anastomosis itself has blinded us to the real drivers of anastomotic failure: features of the landscape that make impossible the meticulous dissection and gentle approximation of distant limbs of bowel inside a very narrow tunnel. We have designed automated staplers and intricate configurations that have no bearing on the success of our endeavor but failed to devise practical means to quantify blood flow and tension, tenets we have “understood” for centuries. We have also thus far failed to illuminate the biology of colorectal wound dehiscence.

In another sense, the odds are against us. Many of our most persistent challenges are permanent: the colon is ill suited for healing, the pelvic space is narrow, and the field is contaminated. Moreover, while we have not moved the needle of anastomotic leak rate, we have pushed the boundaries of oncologic outcomes and organ preservation; our patient population gets older and sicker. So, as our methods improve, so too does the horizon of what we dare possible ever stretch out before us. Rather than lose hope, we must recognize that the field is still rife with opportunity for those arriving with fortitude, fresh perspective, and the tenacity to “ask why are we doing it this way and how can we make it better?”