Practical Aspects of Nonoperative Management

Stable patients with high-grade liver injuries (grade III and above) are admitted to the intensive care unit for close hemodynamic monitoring and serial abdominal examination. This must include careful palpation of the liver to assess for new tenderness or any increase in liver size; both are indicative of intrahepatic hemorrhage. Many of these patients have associated injuries, such as traumatic brain injury or lung contusions, which also require intensive care. Monitoring bladder pressures for early recognition of intraabdominal hypertension is indicated in patients who have received massive fluid resuscitation and in those who have a firm, distended abdomen (Fusco et al, 2001; Yang et al, 2002).

The duration of ICU monitoring depends on the grade of injury. In general, patients with isolated liver injuries who remain stable are transferred to ward beds after 36 to 48 hours and can be discharged within a week. Routinely obtained laboratory studies include a complete blood count, coagulation profile, and also liver function tests, although these are less important.

The question of routine follow-up imaging is controversial (Alonso et al, 2003). Although routine follow-up imaging is not indicated, most patients with high-grade injuries undergo subsequent CT scans in search of complications such as biloma or hepatic necrosis or because of associated injuries. Fever, leukocytosis, unexplained drops in hematocrit, and a rising bilirubin are obvious indications for repeat imaging (Pachter et al, 1996).

Faliure of Nonoperative Management

Failure of nonoperative management is an uncommon but potentially life-threatening complication that should be recognized early and addressed aggressively (Galvan & Peitzman, 2006). Ongoing bleeding is the major cause of failure. Evidence of ongoing hemorrhage includes hypotension, falling hematocrit, an increase in liver size, or the onset of new or increased pain in the right upper quadrant. If the patient is stable, a repeat CT scan will demonstrate the bleeding, which can then be controlled by angiographic embolization (Misselbeck et al, 2009). The patient in shock is urgently taken to the operating room for laparotomy. The complications of nonoperative management are listed in Box 102.1.

Adjuncts to Nonoperative Management

Adjunctive modalities have a crucial role in the nonoperative management of hepatic trauma. These modalities include angiography, image-guided drainage, endoscopic retrograde cholangiopancreatography (ERCP), and laparoscopy (see Chapters 18, 19, 21, and 28; Carrillo et al, 2010).

Angiography is by far the most important adjunct. A “blush” of intravenous contrast on CT scan (i.e., active extravasation) typically results from injury to a branch of the hepatic artery (Fig. 102.2). This blush is an indication for selective angiographic embolization of the bleeding vessel to prevent pseudoaneurysm formation and delayed hemorrhage (Demetriades et al, 2006; Eubanks et al, 2003).

FIGURE 102.2 Bleeding from a hepatic artery treated by embolization (circled areas).

A blush on CT scan is by no means the only indication for angiography. Evidence suggests that liberal use of angiography for any high-grade liver injury is safe and effective, because significant injury to lobar and segmental vessels may occur without a blush, and if left untreated, this may lead to delayed hemorrhage (Mohr et al, 2003).

CT-guided percutaneous drainage plays a key role in nonoperative management, because bile collections are common around the injured liver (Castagnetti et al, 2006). Such drainage obviates the need for laparotomy in the majority of cases. Persistent high-output bile drainage from a percutaneously placed drain indicates injury to a segmental or lobar bile duct. In these patients, ERCP may be used to delineate the site of injury and to decompress the biliary tree by endoscopic papillotomy and stenting (Lubezky et al, 2006); however, in the absence of ductal isolation, adequate external drainage should be sufficient for fistula closure. Image-guided percutaneous drainage is also useful in addressing perihepatic or intrahepatic abscesses.

Delayed laparoscopic lavage is emerging as another useful adjunct to conservative management (Carrillo et al, 2001; Franklin et al, 2007). This minimally invasive procedure is indicated when the patient fails to thrive, with persistent low-grade fever and other signs of ongoing systemic inflammatory response. Port placement is similar to a laparoscopic cholecystectomy with a single subhepatic lateral port. The injured liver is inspected, perihepatic collections are drained and cultured under direct vision, and the infected “rind” of clotted blood and bile on the surface of the liver is gently and bluntly teased off the capsule. The area of injury is typically obvious and is carefully avoided. The liver is irrigated with the suction irrigator, and drains are placed under direct vision in the relevant spaces above or below it.

Other Complications of Nonoperative Management

Rarely, a false aneurysm of a branch of the hepatic artery (Schouten van der Velden et al, 2009) will decompress into an injured biliary radical in a delayed fashion, causing hemobilia (Green et al, 2001; see Chapters 104 and 105). The patient presents weeks or months after the injury with upper gastrointestinal bleeding, jaundice, and colicky abdominal pain as a result of blood and clots in the biliary tree that are subsequently passed through the ampulla of Vater. Angiographic embolization (see Chapters 28 and 83) of the source of bleeding is the treatment of choice (Mohr et al, 2003).

Occasionally, a devascularized area of liver will become necrotic. Small areas of hepatic necrosis are difficult to diagnose and do not require treatment; however, large areas of necrosis may lead to a systemic inflammatory response or may become infected. On rare occasions, hepatic necrosis may require resectional debridement and wide drainage.

The most devastating complication of nonoperative management is missed hollow viscus injury. This complication is very uncommon (Miller et al, 2002), but when it occurs, the mortality rate is high. The diagnosis is particularly difficult, because there is often free fluid in the peritoneal cavity as a result of the liver injury. The abdomen may be firm and difficult to examine, because the patient is intubated and sedated. The signs of a missed hollow viscus injury are therefore subtle, and timely diagnosis requires a high index of suspicion. The decision not to operate should be critically reevaluated with any signs of hyperdynamic physiology. Any adverse change in the abdominal findings or in the expected clinical trajectory should prompt a repeat CT scan or laparotomy in search of a missed injury. Failure to achieve timely source control results in the patient’s death.

Nonoperative Management of Penetrating Hepatic Trauma?

It was only natural that following the dramatic success of nonoperative management for blunt hepatic trauma in the 1990s, attempts would be made to extend the concept to penetrating injuries. Although conservative management of stab wounds to the abdomen, including the liver, has been practiced for years, application of this approach to gunshot wounds is controversial. In 1994, Renz and Feliciano reported 32 patients with right thoracoabdominal gunshot wounds, of whom 7 had CT-documented liver injuries that were successfully treated nonoperatively. Five years later, Demetriades and colleagues (1999) reported 16 patients with penetrating liver injuries, 5 of whom needed a delayed laparotomy; 1 patient died. Clearly, nonoperative management of hepatic gunshot wounds must be applied carefully and selectively and is not currently the standard of care. Success hinges on complete CT delineation of the entire bullet trajectory within the liver and the ability to reliably exclude injury to adjacent hollow organs, especially the right colon (Demetriades et al, 2006).

Superficial, Low-Grade Injuries

Bleeding from a superficial laceration is controlled by local hemostatic techniques, such as temporary packing, while the surgeon is addressing other injuries, or argon beam coagulation. If an argon beam coagulator is not available, cauterization of the bleeding liver by placing the electrocautery at its highest level and arching the current adjacent to the liver parenchyma without touching it can often achieve the same result. For deeper injuries, local hemostatic agents placed into the laceration followed by manual or pack compression of the edges is a useful adjunct. A portion of healthy omentum stuffed into the laceration and sutured to the liver capsule can serve the same purpose. If manual compression alone controls the bleeding, approximating the laceration with heavy chromic catgut “liver sutures” on a large blunt needle is another useful technique.

Packing of Deep Liver Lacerations

Packing is the workhorse of surgery for hepatic trauma and is especially useful for high-grade lacerations or parenchymal disruptions that are not amenable to the local hemostatic measures described above (Nicol et al, 2007). The technique of packing is often poorly executed, but the goal is to effect hemostasis by compressing the bleeding parenchyma between packs placed above and below it. Packing only over the injury without creating an opposing pressure vector often fails.

It is important to note that packing is most effective for venous injuries. For this reason, we advocate immediate postoperative angiography with selective embolization for high-grade liver injuries with arterial hemorrhage that cannot be directly addressed by suture-ligation of the bleeders (Gaarder et al, 2007).

Although assertive packing with laparotomy pads is essential to achieve hemostasis, sometimes overenthusiasm results in overpacking, which can result in hypotension as a result of vena caval compression (Meldrum et al, 1995). The problem should be recognized immediately, and the packs removed gradually, to identify the degree of packing that achieves hemostasis without the adverse effects.

It is important to spend time observing the packs in situ, because ineffective packing takes time to declare itself, as the blood seeps through multiple layers of cloth with very high absorptive capacity. Failure to wait a few minutes before closing the abdomen often results in an urgent reoperation.

Omentum can sometimes be stuffed into a liver defect and serve the same purpose as a laparotomy pad. Although it has the advantage of not requiring removal, rarely is there enough omentum to provide effective tamponade for a large laceration. In desperation, the senior author of this chapter (W.P.S.) once placed several large pieces of Vicryl mesh into a massively bleeding gunshot wound of the right lobe of the liver in a patient with an associated infrahepatic vena caval injury. The Vicryl packs were never removed, and the patient survived. Although this method cannot be routinely recommended, it is nevertheless an example of improvisations that are sometimes called for in a difficult situation.

Other Techniques for Hemostasis of Deep Liver Lacerations

If packing fails to control the bleeding, inflow occlusion is the next step. The index finger is placed through the foramen of Winslow posterior to the hepatoduodenal ligament, and a window is created in the gastrohepatic ligament. An umbilical tape is then passed through the window in the gastrohepatic ligament and out through the foramen of Winslow, thereby surrounding the portal triad, which is then occluded en masse with either a Rummel tourniquet or a vascular clamp. The occlusion time should be kept to a minimum: 15 minutes of hepatic inflow occlusion is safe in elective hepatic resection (Tralhão et al, 2009). Intermittent hepatic inflow occlusion has been used safely with a total inflow occlusion time in excess of 120 minutes (Ishizaki et al, 2006), although no level 1 data are available to define the safe duration of hepatic inflow occlusion in trauma patients.

If the bullet or knife tract is superficial, a tractotomy can expose the injured vessel for direct suture ligation (Badger et al, 2009). Tractotomy can be performed using the Endo GIA (gastrointestinal anastomosis) stapler (Covidien, Boulder, CO) with a vascular stapler load, gradually dividing the roof of the tract in a piecemeal fashion so that the deep bleeders in the tract can be individually suture ligated or doubly clipped with a hemoclip. If the injury involves the left lateral segment, a stapled segmentectomy, using either the Endo GIA or a TA (thoracoabdominal; Covidien) stapler, is an expedient solution (Schemmer et al, 2007).

More often, the problem arises from a deep tract or through-and-through injury to the main mass of the right lobe. Filling the tract with omentum, Flowseal (Baxter Healthcare, Zurich, Switzerland), or Avitene (Davol, Warwick, RI) will often achieve hemostasis. Balloon tamponade is another useful technique (Poggetti et al, 1992), wherein a balloon device is improvised from a red rubber (Robinson) catheter inserted into a 1-inch Penrose drain. The distal end of the drain is tied closed, and the proximal end is secured snugly around the catheter with a pair of heavy silk ties. The balloon device is inserted through an appropriate stab incision in the right upper quadrant abdominal wall and into the tract and is inflated with saline injected through the catheter. A closed suction drain is placed in the subhepatic space. The balloon is deflated and removed either at planned relaparotomy, 48 to 72 hours after the original operation, or percutaneously if the abdomen was closed.

Sometimes a bleeding vessel can be exposed and suture ligated at the base of a deep liver laceration, using the technique of finger-fracture hepatotomy with selective vascular ligation under inflow occlusion (Hirshberg & Mattox, 2006); however, attempts to reach a deep vessel are often ill advised, as they lead to continued blood loss and physiologic deterioration and may even aggravate the bleeding. Packing, when successful, is the best approach, and immediate postoperative angiographic embolization (see Chapters 28 and 83), if available, can be a lifesaving adjunct.

Resectional debridement of the injured tissue is an option for extensive destruction of the periphery of the liver; however, the surgeon must always keep in mind that the one and only major goal of the operation is to achieve hemostasis as rapidly as possible; other considerations are secondary and must wait. Resectional debridement, if necessary, is more frequently used to address devascularized liver tissue during a delayed reoperation rather than bleeding in the initial procedure.

Bleeding from the Hepatic Veins and Juxtahepatic Vena Cava

The large retrohepatic veins are a low-pressure system; when injured, they are often contained by the triangular ligaments of the liver. An injury to the retrohepatic cava or the major hepatic veins can therefore manifest either as bleeding through a hole in the liver, if the suspensory ligaments remain intact, or as bleeding around the liver, if they are disrupted (Buckman et al, 2000); bleeding through the liver stops with effective packing of the hole in the liver, but bleeding around the liver does not.

Juxtahepatic vena caval hemorrhage should be suspected if a large volume of dark blood gushes around or through a high-grade liver injury. Posterior and inferior traction on the falciform ligament may compress the liver mass onto the retrohepatic vena cava and reduce the bleeding. The next step is packing, although bleeding around the liver may not stop with packing despite heroic efforts. In these situations, when packing and the Pringle maneuver fail to stop the bleeding, an early decision to achieve control of the hepatic outflow tract—that is, the second vascular pedicle—is critical. Delay of the decision ensures failure and patient death as a result of continued massive blood loss.

Vascular isolation of the injured liver is a formidable surgical challenge (see Chapter 91A; Yellin et al, 1971), not only because of the technical complexity involved but also because cross-clamping the suprahepatic and infrahepatic vena cavae (Krige et al, 1990) is poorly tolerated by the hypovolemic patient. Furthermore, although creating a sling to occlude the infrahepatic vena cava can usually be accomplished relatively quickly after a Kocher maneuver, dissection of the infradiaphragmatic suprahepatic cava is difficult, particularly for the emergency surgeon who operates infrequently in this area.

Several options are available for achieving outflow control, and by far the most well-known is the atriocaval shunt (Fig. 102.3; Schrock et al, 1968). After performing a median sternotomy, a No. 32 chest tube is inserted through the purse-stringed right atrial appendage and threaded through the right atrium and juxtahepatic vena cava to a position just above the renal veins. At the same time, a second team isolates the inferior vena cava (IVC) in its suprarenal infrahepatic segment, and a side hole is fashioned at the level of the right atrium. Rummel tourniquets are tightened around the vena cava and the stent within the pericardium and just above the renal veins. Using a 9.5-mm endotracheal tube with a side hole 17 cm from the tip obviates the need to isolate the suprarenal cava, because inflating the balloon of the tube occludes the caval lumen. This maneuver shunts venous return to the heart around the injured segment and controls most of the bleeding, enabling the surgeon to visualize and repair the hole in the vein by completely mobilizing the right lobe of the liver and rotating it medially. This is a difficult and uncommonly used technique, as evidenced by the popular adage that “the number of articles published about atriocaval shunt far exceeds the number of patients who survived it” (Fig. 102.3).

FIGURE 102.3 An atriocaval shunt. IVC, inferior vena cava.

Another option for outflow control is complete hepatic vascular isolation (see Chapter 91A) by clamping the inferior vena cava above and below the liver, in conjunction with simultaneous aortic cross-clamping (Khaneja et al, 1997), or use of a centrifugal pump to return blood to the heart (Baumgartner et al, 1995; Horwitz et al, 1995) to reduce the risk of hypovolemic arrest after caval occlusion. Even hepatic explantation and back-bench repair of hepatic vein avulsion followed by hepatic autotransplantation has been reported (Boggi et al, 2006). Despite all of these elegant technical concepts, which are of very limited utility in the real clinical situation, the mortality rate for unstable patients with juxtahepatic vena caval injury unresponsive to packing remains extremely high (Phelan et al, 2001).

Urgent Reoperation

The three indications for urgent reexploration of a patient with liver injury are 1) continued bleeding, 2) abdominal compartment syndrome, or 3) a missed injury. At laparotomy, the liver packs are removed, and the injury is closely inspected for hemostasis. Thorough exploration of the abdomen in search of other sources of bleeding is then carried out.

The most common causes of bleeding found at urgent relaparotomy for trauma are inadequate hemostasis of the original injury, another source of bleeding in the vicinity, iatrogenic trauma, or diffuse coagulopathy (Hirshberg et al, 1993). In cases of rebleeding, mortality is extremely high, and the prognosis is especially bleak, if no discrete source of hemorrhage is found.

The liver encountered at urgent reoperation is often “hostile” and is very different from the normal organ. It is grossly swollen to twice or more its normal size with a tense capsule that tears easily. The edematous parenchyma does not hold clips or ligatures well, and it is often impossible to properly mobilize such a swollen, friable liver without adding an iatrogenic insult to the original injury. Hemostatic liver sutures tend to tear through the capsule and lose their hemostatic effect. This hostile environment represents a very bad turn of events in an already precarious situation.

Some patients develop intraabdominal hypertension following the initial operation, as previously discussed. In most cases, an urgent decompressive laparotomy with temporary abdominal closure, leaving the abdomen “open,” is the only way to save a rapidly deteriorating patient. Occasionally, this problem can be managed by catheter drainage of the peritoneum, if the patient is relatively stable.

Planned Reoperation

Planned reoperation is an integral part of the damage-control sequence for treatment of severe liver trauma. In general, the patient is reexplored 36 to 48 hours after the initial operation, once hemodynamic stability has been achieved. Although such a practice is not supported by data, trauma surgeons are reluctant to remove liver packs sooner than 36 hours after the initial operation, because earlier removal often results in rebleeding and a need to repack the liver. Packs should rarely, if ever, be left in any patient for more than 72 hours (Scott et al, 2006), because the risk of infection increases sharply. If the packed liver injury is associated with damage control of a bowel injury, intestinal reconstruction should be performed, and the packs should be removed last.

Extreme care should be taken to remove liver packing, which should be done slowly and only after flooding the right upper quadrant with saline. A Toomey syringe is used for continuous pulse irrigation and blunt atraumatic separation of the packs off the liver capsule, assisted by the jet of saline. Bleeding should prompt the surgeon to stop and address the bleeding point with the argon beam coagulator or another local hemostatic technique. Special care should be taken to avoid stripping the liver capsule. The area should be widely drained, and resectional (nonanatomic) debridement of areas of obvious necrosis should be considered, if hemostasis has been achieved and the patient is otherwise stable. Ongoing oozing from the unpacked areas should prompt consideration of repacking; it is important to admit defeat early and repack rather than continue with attempts at futile hemostasis in a hostile liver.

Extrahepatic Biliary Injury

Most extrahepatic biliary injuries result from penetrating trauma (Posner & Moore, 1985) and are discovered at exploratory laparotomy. The gallbladder is the most common structure injured, and cholecystectomy definitively solves the problem (Posner & Moore, 1985); however, it may be unwise in an unstable coagulopathic patient with an associated liver injury, because it creates another source of hemorrhage from the liver bed. A tube cholecystostomy or closure with an absorbable stitch and drainage of a small hole in the gallbladder are therefore better options in a damage-control situation.

Small stab wounds to the extrahepatic biliary ducts can be managed by primary closure with absorbable suture and external drainage (Bade et al, 1989). The use of a T-tube for a common bile duct (CBD) injury is rarely a practical option in a healthy young adult with a normal CBD. Major injuries to the extrahepatic bile ducts require Roux-en-Y choledochojejunostomy or hepaticojejunostomy for definitive repair (Degiannis et al, 2001).

Penetrating trauma to the biliary tree is often associated with injuries to vascular structures such as the portal vein, hepatic artery, and vena cava (Jurkovich et al, 1995). Injuries to adjacent structures—such as the duodenum, pancreas, stomach, and liver—are also common. The hemodynamic stability of the patient and the complexity of the other injuries will govern management of the bile duct injury.

Roux-en-Y choledochojejunostomy and hepaticojejunostomy are the definitive treatments of choice, as they are for bile duct injuries complicating cholecystectomy; however, more often than not, the patient is unstable because of associated injuries that preclude a complex biliary reconstruction. Damage-control drainage of the bile duct injury by temporary T-tube drainage of a complex laceration—or via an infant feeding tube tied into the proximal duct, if the duct has been completely transected—and continued resuscitation in the ICU with subsequent elective biliary reconstruction under controlled conditions is a much better option. Although temporary ligation of the proximal transected duct is another option in a damage-control situation, drainage is generally preferred.

Blunt injury to the gallbladder as a result of a “blowout” (Zellweger et al, 2005) or adhesive tear is uncommon, but the treatment principles are similar to those for penetrating injury. Rarely, the surgeon may encounter avulsion of the common bile duct from the duodenum. If this is an isolated injury, the duodenum can be oversewn, and a Roux-en-Y choledochojejunostomy can be constructed (Melton et al, 2003). Regardless of the type of biliary management, wide drainage is essential.