Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
A fistula is an abnormal communication between two epithelialized surfaces. Over the past half century, the mortality associated with gastrointestinal fistulas has decreased from 40% to 60% to approximately 15% to 20% of patients. This improvement in prognosis is attributable to advances in fluid and electrolyte/acid-base knowledge and therapy, blood administration, critical care, antibiotic regimens, and nutritional management. Formerly, malnutrition and electrolyte imbalance were the major causes of death in affected patients. Currently, mortality is principally attributable to uncontrolled sepsis and its associated malnutrition. Sepsis is responsible for 80% of all deaths in fistula patients.
Spontaneous fistulas account for 15% to 25% of gastrointestinal fistulas and causes include radiation, inflammatory bowel disease, diverticular disease, appendicitis, ischemic bowel, perforation of gastric or duodenal ulcers, pancreatic and gynecologic malignancies, and intestinal actinomycosis or tuberculosis. The majority (75% to 85%) of gastrointestinal fistulas are iatrogenic as a result of technical complications of surgical procedures and trauma. Etiologies include anastomotic dehiscence, intraoperative injury to the bowel or its blood supply, erosion from indwelling tubes, retention sutures or prosthetic mesh, and misplacement of a suture through the bowel during abdominal closure. Other complications that may cause a fistula include intraperitoneal bleeding and abscess formation with or without suture line dehiscence. Fistulas may develop after percutaneous abscess drainage.
The critical tenets in successful management of gastrointestinal fistulas are recognition of the fistula, control of infection and further contamination, restoration of fluid and electrolyte losses, and reestablishment of a positive nutritional balance before undertaking major definitive corrective procedures.
Gastrointestinal fistulas are a consequence of perforations that communicate with adjacent organs or intestine (internal fistulas) or externally with the abdominal wall (enterocutaneous fistulas [ECFs]). Enteroatmospheric fistulas are a specific type of ECF that occur in an open abdomen. Several factors make gastrointestinal fistulas complex and potentially lethal. First, the patients are usually systemically ill. Sepsis is a recognized antecedent risk factor for the development of a gastrointestinal fistula, and the high metabolic requirement of the septic state can prevent spontaneous closure. Sepsis is often secondary to the factor leading to the fistula itself. Malnutrition is also common and results both from the hypermetabolic state of the septic, postoperative patient and from the large volume of protein-rich fluid produced and lost through the fistula. Fluid and electrolyte abnormalities (hypovolemia, hypokalemia, hypomagnesemia, metabolic acidosis) are common and result from the sustained loss of intestinal fluid. Such losses are not limited to ECF because internal fistulas, such as enterocolic fistulas, bypass normal intestinal continuity and may overwhelm the absorptive capacity of the recipient organ. Malabsorption and malnutrition from bacterial overgrowth may occur in gastrocolic or enterocolic fistulas. Local wound excoriation and discomfort from the intestinal effluent can thwart potential abdominal wall reconstruction and recovery after operation to repair a fistula. Finally, operating on a fistula before control of sepsis and nutritional optimization can lead to increased mortality and operative failure.
The majority of gastric and duodenal fistulas occur after surgical, endoscopic, or interventional procedures. Anastomotic or suture line failures account for 80% to 85% of all such fistulas. Before 1950, greater than 60% mortality was observed in patients with gastric and duodenal fistulas, but in the 21st century the incidence has decreased to less than 3% and the mortality rate has decreased to less than 15%. Comparisons of sutured versus stapled anastomoses show no obvious superiority of either. Postoperative leaks from gastric staple or suture lines after ulcer surgery accounted for most perforations in the past. However, the decline in gastric resection for ulcer disease, along with the broad application of new endoscopic and laparoscopic techniques for other diseases, contributes to other newer causes of perforation, albeit at a lower incidence.
Any of the available gastric operations for morbid obesity may result in gastric staple line disruption in the early or late postoperative period. Early anastomotic or staple line leaks in this patient population are highly morbid and often lethal. For gastric bypasses, the 10% to 30% incidence rate of internal fistula formation after simple stapling has been reduced to 3% to 6% by either gastric division after stapling or up to three applications of the stapler without division.
In a series of 318 partial gastrectomies, Pickleman reported a 1.3% anastomotic leak rate, all from the gastrojejunostomy. After total gastrectomy with Roux-en-Y esophagojejunostomy, anastomotic leaks occurred in 4.8%. A perforation rate of 1.5% has been reported after vertical banded gastroplasty for morbid obesity, and leak rates as high as 6% have been reported after gastric bypass, again from the gastrojejunostomy. The risk of leakage at gastric staple lines may be increased with the use of cautery to control bleeding at the stapled edge, intersecting staple lines within an anastomosis, and the use of a thin tissue stapler on a thickened or edematous gastric wall, which may cause overcompression, tearing, and devascularization. In such tissue a handsewn closure may be preferable.
Duodenal stump leakage has declined because of the decreased use of antrectomy for ulcer disease. Duodenal stump leakage is more common after difficult gastric resections. In a high-risk patient, morbidity and mortality can be decreased and possibly prevented by placement of a duodenostomy tube along with closed suction drains external to the duodenum.
The ongoing extension of laparoscopic techniques to gastric surgery has not eliminated the risk of perforation or fistula formation. The incidence of esophageal or gastric perforation during fundoplication ranges from 0.3% to 1.9%, with a large retrospective review of 2453 procedures by Perdikis et al. showing an overall incidence of 1%. Laparoscopic fundoplication may also result in delayed gastric perforation along the greater curvature from inadvertent thermal or cautery injury during division of the short gastric arteries. If the diaphragmatic crura are not approximated adequately, the fundoplication can herniate into the chest during postoperative straining, vomiting, or heavy lifting, with subsequent gastric ischemia and perforation. Laparoscopic revision of a previous fundoplication requires more gastric traction and division of adhesions, with a 3% risk for gastric laceration. The laparoscopically placed adjustable silicone gastric band, positioned around the proximal part of the stomach for the treatment of morbid obesity, has also resulted in gastric perforation in less than 1% of patients.
Laparoscopic cholecystectomy may produce duodenal injury if the duodenum and gallbladder are densely adherent to one another as a result of either direct cutting action or cautery and thermal injury. Laparoscopic cholecystectomy may also result in colonic injury by the same mechanisms. In addition, improperly insulated instruments may cause electrical arcing to the duodenum, small bowel, or colon with resultant perforation. These injuries are usually apparent within 24 to 72 hours and fistulas are rare.
The capacity and compliance of the stomach make endoscopic examination routine, with a low incidence of injury. However, endoscopic polypectomy or attempts at tumor removal with a snare, cautery, or endomucosal resection may cause either immediate full-thickness perforation or deep penetration with thermal injury and subsequent delayed perforation and fistula. Percutaneous endoscopic gastrostomy tube placement has also resulted in perforation, either from dislodgement of the tube before complete gastric adhesion to the abdominal wall or from trauma during placement. Tube insertion may perforate the adjacent jejunum or transverse colon and result in a persistent gastrojejunal or gastrocolic fistula, even after the gastrostomy tube has been removed. Gastrostomy tube placement may cause a persistent gastrocutaneous fistula that enlarges through erosion or infection of fascia and skin. These fistulas may be difficult to control, with continued drainage of gastric fluid onto the surrounding skin. The substitution of a larger gastrostomy tube will not control the leakage and usually results in enlargement of the opening. Persistent drainage may require either tube removal or placement of a smaller tube, along with direct or nasogastric suction until the tract contracts down around the tube. Surgical closure is required for a persistent gastrocutaneous fistula that does not respond to such measures. Endoscopic clipping of the gastric opening has been described and is effective in a select group of affected patients.
Because many endoscopic duodenal procedures involve the second portion of the duodenum, perforation is usually retroperitoneal. Failure to recognize an injury or a delay in treatment markedly increases morbidity and mortality. Perforation after endoscopic retrograde cholangiopancreatography (ERCP) with ampullary sphincterotomy for stone extraction or biliary stent placement is one of the more frequent postendoscopic indications for urgent surgical intervention. Repair of the distal bile duct, as well as repair of the duodenum, may be required. Controlled leaks confined to the retroperitoneum can often be monitored with very close clinical observation in stable patients. Retroperitoneal perforation is more common during therapeutic ERCP, with an incidence of 0.6% to 1.8% and a mortality rate of up to 25%. Delayed duodenal perforation from the biliary stent itself may be caused by partial extrusion and impingement of the end of the stent on the distal second or proximal third portion of the duodenum, with eventual erosion and perforation. Proximal stent migration into the common bile duct may cause a choledochoduodenal fistula to subsequently form if the stent reenters the duodenum away from the papilla. Similarly, pancreatic duct stents may produce a pancreaticogastric fistula with proximal migration of the stent into the gastric antrum. Other procedures at risk for the development of duodenal perforation include endoscopic polyp or tumor removal, push enteroscopy, endoscopic ultrasound with transduodenal biopsy, and endoscopically assisted transgastric jejunal feeding tube placement.
Crohn disease is a rare cause of gastrocolic, duodenocolic, or duodenocutaneous fistulas. Primary gastric or duodenal involvement is reported in less than 1% of patients with Crohn disease; duodenocutaneous fistulas may develop from the first or second portion of the duodenum. However, most gastric or duodenal fistulas are internal and result from involvement of primary Crohn disease of the transverse colon or, more commonly, from recurrence at the ileocolic anastomosis after previous resection. Those with gastrocolic fistulas have a 40% incidence of vomiting, which may be feculent; duodenocolic fistulas are often asymptomatic, with only a 4% incidence of vomiting, which is not usually feculent.
Neoplastic causes of internal fistulas are uncommon. Gastrocolic fistulas have resulted from gastric ulcer erosion and invasion of the transverse colon by gastric adenocarcinoma or lymphoma. In rare instances, primary hepatic flexure or transverse colon adenocarcinoma may invade and create a fistula to the duodenum or stomach.
Small intestinal fistulas can arise in a number of ways. The small intestine's length, as well as its elaborate anatomy, predisposes it to association in a variety of diseases. Any surgical procedure involving the abdomen can result in iatrogenic injury to the small intestine and later fistula formation. The development of a fistula between the small intestine and an internal structure can be a life-threatening event, as with exsanguination from an aortoenteric fistula. Other fistulas, particularly enteroenteric fistulas may be asymptomatic. External small intestinal fistulas (enterocutaneous fistulas, or ECFs) are the most prevalent of small intestinal fistula. ECFs most commonly follow postoperative complications and are often the result of technical errors at the time of an abdominal procedure. ECF mandates careful management to avoid further complicating the well-being of the patient. Of 35 fistulas originating in the jejunum or ileum reported in one study, 75% drained externally. The ileum is the most common site of origin of an ECF. ECF can be classified according to the daily volume of drainage. A high-output fistula drains 500 mL/day or more of fluid. In general, a high-output fistula is associated with greater morbidity and mortality. Polk et al. found that patients with high-output fistulas had a greater incidence of malnutrition and fluid and electrolyte disturbances. Mortality was increased, and the rate of fistula closure was low. In contrast, excellent results with high-output fistulas were reported in Graham's series in which 35 of 39 consecutive patients underwent spontaneous fistula closure with a 3% mortality rate. High-output fistulas usually originate from a proximal portion of the small intestine. Independent conditions, such as previous intestinal irradiation, intraabdominal sepsis, or the presence of diseased or ischemic intestine, can also cause external fistulas. Enteroenteric or enterocolic fistulas develop almost exclusively from the transmural inflammation associated with Crohn disease.
Webster and Carey proposed five general mechanisms for small intestinal fistula formation :
Congenital . A rare form of congenital small bowel fistula involves failure of the vitellointestinal duct to obliterate, resulting in an ECF to the umbilicus. The diagnosis should be suggested by the appearance of fecal material at the umbilicus after postnatal slough of the umbilical cord.
Trauma . Traumatic injury to the small intestine that results in fistula formation usually occurs from an internal source, such as a swallowed fish bone, toothpick, magnet, or metallic object. Erosion of these objects into an adjacent loop of small intestine results in an internal enteroenteric fistula. Major penetrating trauma without damage-control laparotomy rarely results in fistula formation because these cases are explored surgically and the intestinal injuries repaired. Patients treated with damage-control laparotomy techniques have an increased risk for delayed formation of intestinal fistulas caused by prolonged exposure and desiccation of multiple intestinal loops.
Infection . An abscess or invasive intestinal infection may erode through the intestine and create a fistula. Amebiasis, tuberculosis, coccidioidomycosis, actinomycosis, and salmonellosis may cause intestinal fistulas. Intestinal perforation at the ileum from tuberculosis and typhoid fever is still occasionally seen in the Third World. Actinomyces is a rare cause of after appendectomy.
Perforation or Injury with Abscess . Perforation of the intestinal wall by tumor, inflammation, or operative injury may result in the local formation of an abscess. A fistula may develop if this abscess subsequently erodes into an adjacent structure. An ECF rarely develops spontaneously—most develop after an abdominal operation. Most ECFs develop as a result of injury to the small intestine during surgery. They also arise from exposure of the bowel to an abdominal defect or prosthetic mesh used to repair such defects. Abdominal wall dehiscence with evisceration and strangulation of a hernia with infarction and perforation have been implicated in the development of external fistulas. In most large series, 60% to 90% of ECFs were caused by operative complications. In addition, ECFs are caused by leakage from an intestinal anastomosis or enterotomy closure. Fistulas may also develop as a result of percutaneous drainage of an intraabdominal abscess.
Inflammation, Irradiation, or Tumor . The small intestine and an adjacent structure can become densely adherent from chronic inflammatory conditions, abdominal radiation injury, or tumor erosion. Subsequent degeneration of the common wall results in fistula formation. Inflammatory bowel disease, particularly Crohn disease, is well known to create fistulas in this fashion. In Crohn disease the disease makes fistula formation after anastomosis more likely. Although a spontaneous external fistula can develop as a direct result of Crohn disease, most occur only after a previous operation has caused the affected intestine to adhere to the abdominal wall. Postoperative fistulas in the setting of Crohn disease are as likely to develop after simple exploration, bypass, or appendectomy as after primary resection. Fistula formation after laparotomy is usually an early complication, especially when arising from an anastomosis, whereas a late fistula generally indicates recurrent Crohn disease. Fistula formation is particularly apt to occur after irradiation of a pelvic malignant lesion. Fistulas that arise secondary to radiation injury rarely, if ever, close spontaneously. Laparoscopy has been found to decrease the incidence of fistula in Crohn disease.
Acute intraoperative perforations are best handled by maintaining a strong index of suspicion for technical errors, recognizing the injury before the end of the procedure, and immediately repairing, suturing, or reinforcing weakened tissues. The tendency for potential injuries must be recognized and overcome, especially during prolonged laparoscopic cases. Serosal injuries should be carefully examined. Intraluminal instillation of methylene blue and saline or direct endoscopic examination can demonstrate a small perforation or provide reassurance that an area of concern is not a full-thickness injury. During repeat laparotomy for an open abdomen secondary to damage-control laparotomy, the urge to break up interloop adhesions to search for interloop abscesses and reaffirm “normal” small intestinal anatomy should be suppressed because the dense inflammation between the viscera leads to the development of serosal injuries and possible future fistulas.
Postoperatively, unrecognized perforations or leaks that develop at suture or staple lines are manifested as instability or failure to improve as expected. A gastrointestinal fistula can be obvious in some patients and extremely difficult to identify in others. Fistula formation is frequently heralded by fever and abdominal pain until gastrointestinal contents discharge through an abdominal incision. Spontaneous fistulas from neoplasm or inflammatory disease usually develop in a more indolent manner. ECFs often have intestinal contents or gas exiting from a drain site or through the abdominal incision after an operation. The drainage fluid is usually typical of intestinal contents, with obvious bile staining, and intestinal gas may accompany the effluent. At times the initial fistula drainage may appear clear rather than yellow or green, and the fistula may be misdiagnosed as a seroma or wound infection. At other times a heavy purulent component may also mask the enteric communication and instead suggest a wound infection. If the drainage persists and the diagnosis is uncertain, the patient may be given activated charcoal or indigo carmine by mouth and the drainage inspected for these substances.
Gastrointestinal fistulas are classified by their anatomic characteristics and are either internal or external (enterocutaneous). Typically the name of a fistula is derived from the involved and connected organs or structures. Examples include gastrocolic, jejunoileal, and aortoenteric fistulas. The anatomy of a fistula will suggest the cause and help to predict whether spontaneous closure will occur. Fistulas can be classified physiologically in terms of output over a 24-hour period. They can be classified as low (<200 mL/day), moderate (200 to 500 mL/day), and high (>500 mL/day). An accurate measure of fistula output, as well as the chemical makeup of the effluent, assists in preventing and treating metabolic deficits and correcting ongoing fluid, electrolyte, and protein losses. The anatomic and etiologic factors are more important in predicting spontaneous closure than the actual output of the fistula. The underlying disease process helps to forecast both the closure rate and mortality.
ECFs are the most common type of small intestinal fistula and are usually recognizable. In contrast, internal fistulas that communicate between the intestine and another hollow viscus may not be suspected for some time because the symptoms may be minimal or mimic the underlying disease process.
Fluid and electrolyte disturbances occur commonly in patients with ECFs. The salivary glands, stomach, duodenum, pancreas, liver, and small intestine secrete 8 to 10 L/day of a fluid rich in sodium, potassium, chloride, and bicarbonate. The degree of volume loss and electrolyte imbalance depends on the anatomic location of the fistula and may exceed 3000 mL/day. Duodenal fistulas are particularly prone to volume and electrolyte loss. High-output duodenal or jejunal fistulas continue to carry a mortality rate of approximately 35%. A distal fistula, such as one arising from the terminal ileum in a patient with Crohn disease, is associated with smaller fluid losses because of proximal absorption.
The most common abnormalities seen are hypovolemia, hypokalemia, and metabolic acidosis. Hypokalemia occurs primarily from potassium loss in the fistula effluent; hypovolemia contributes by causing renal retention of sodium in exchange for potassium secretion. Sepsis contributes to the hypovolemic state by raising the metabolic rate and increasing insensible water loss. Metabolic acidosis results from the loss of pancreatic juice rich in bicarbonate and is more common with proximal fistulas. Gastric fistulas may cause a hypokalemic, hypochloremic metabolic alkalosis secondary to the loss of a large volume of hydrochloric acid.
Patients with fistulas causing fluid and electrolyte abnormalities have a higher mortality rate. Advances in critical care, invasive monitoring, and aggressive fluid and electrolyte management can reduce this early mortality considerably, as evidenced by data from the Massachusetts General Hospital.
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here