Introduction

The liver is the most frequently injured abdominal organ, despite its relatively protected location [1–5]. Management of liver injuries has changed significantly over the last two decades, with significant improvement in outcomes. There is now a broad consensus regarding most aspects of management, with the introduction of standard protocols, but in other areas considerable controversy persists. This literature review considers the diagnosis, investigation, and recommendations for the management of trauma to the liver.

Mechanisms of injury

Road traffic accidents and antisocial, violent behaviour account for the majority of liver injuries. Industrial and farming accidents also account for a significant number. There is an interesting difference in incidence throughout the world, with penetrating injuries (gunshot and stab wounds) accounting for the majority in North America and South Africa and blunt injuries representing the majority in Europe and Australasia [6–11].

The liver consists of a relatively fragile parenchyma contained within the Glisson’s capsule, which is thin and does not afford it great protection. Hence, the parenchyma and its vasculature are very susceptible to blunt and penetrating trauma. The vasculature consists of wide-bore, thin-walled vessels with a high blood flow, and injury is usually associated with significant blood loss.

Blunt trauma in a road traffic accident, or fall from a height, may result in a deceleration injury as the liver continues to move on impact. This leads to tears at sites of fixation to the diaphragm and abdominal wall. A well-recognised deceleration injury involves a fracture between the posterior sector (segments VI and VII) and the anterior sector (segments V and VIII) of the right lobe (Fig. 1). This type of injury may be associated with a significant vascular injury due to a tear of the right hepatic vein. In contrast, a direct blow to the abdomen may produce a central crush injury, with an extensive stellate-type laceration involving segments IV, V, and VIII (Fig. 2). With this pattern of injury, an associated major vascular injury may be present, with disruption of the hepatic arteries, portal veins, or the major hepatic veins. This type of injury may also be associated with bleeding from the caudate lobe (segment I). Such an injury may occur with a blow from a fist or weapon, or a central crush in a road traffic, industrial, or farming accident. Overall, blunt trauma more commonly affects the right hepatic lobe, particularly the posterior sector, with the caudate lobe rarely affected [12, 13].

Fig. 1
figure 1

A deceleration-type fracture injury between the anterior and posterior sectors of the right lobe (left) demonstrated on CT scan (right)

Full size image
Fig. 2
figure 2

A central crush injury resulting in a stellate-type laceration

Full size image

Penetrating injuries may be associated with a significant vascular injury. For example, a stab injury may cause major bleeding from one of the three hepatic veins or the vena cava and also from the portal vein or hepatic artery if it involves the hilum. Gunshots may similarly disrupt these major vessels; this disruption may be much more marked than with stab wounds due to the cavitation effect, particularly with bullets from high-velocity weapons.

The connection between the thin-walled hepatic veins and the inferior vena cava (IVC), at the site where the ligamentous mechanism anchors the liver to the diaphragm and posterior abdominal wall, represents a vulnerable area, particularly to shearing forces during blunt injury. Disruption here leads to the “juxtahepatic” venous injuries, which are usually associated with major blood loss and present a particularly challenging management problem.

Grading of liver injuries

The severity of liver injuries ranges from the relatively inconsequential minor capsular tear to extensive disruption of both lobes with associated hepatic vein, portal vein, or vena caval injury. While several classification systems have been devised, the most widely used is that of The American Association for the Surgery of Trauma (Table 1) [14]. Grade I or II injuries are generally considered minor and usually do not require operative treatment, while grade III–V injuries are severe and often require operative management. Significant vascular injuries can be associated with a haematoma or parenchymal laceration (grade III) but usually occur with major parenchymal disruption (grades IV and V). High-grade hepatic injuries are associated with a higher surgical intervention rate and a poorer prognosis, thus emphasising the importance of adherence to an appropriate management strategy [2, 7, 15–17].

Table 1 American Association for the Surgery of Trauma liver injury scale [14]
Full size table

Patient assessment and initial investigation

It is generally accepted that initial resuscitation and management is the same as for any patient with major trauma and should follow the Advanced Trauma Life Support (ATLS) principles of aggressive fluid resuscitation, guided by monitoring of central venous pressure and urinary output [18]. Management should also be directed toward avoidance of any of the sinister triad of hypothermia, coagulopathy, and acidosis, which are associated with significantly increased mortality. Mechanisms to avoid hypothermia are standard now in major centres and include the use of rewarming blankets and heat exchanger pumps for rapid infusion of resuscitation fluids and blood [19]. Avoidance of coagulopathy and acidosis depends on initial good resuscitation and prompt decision making in the next phases.

The next management phase depends largely on the response to resuscitation and the stability of the patient. Liver injury should be suspected in all patients with blunt or penetrating thoracoabdominal trauma but particularly in shocked patients with blunt or penetrating trauma to the right side. If such a patient remains unstable (systolic pressure <90 mmHg) despite adequate resuscitation (2 L of intravenous fluids), immediate operation is indicated to stop bleeding [18]. This message must be very strongly emphasised as it has been well established that delay in surgery and control of bleeding in the unstable patient are associated with a significantly higher mortality [20].

In stable patients, however, surgery is not the immediate priority and appropriate investigation, perhaps leading ultimately to nonoperative management, can be instituted. The main investigative and therapeutic modalities include ultrasonography, CT scanning, and interventional vascular radiological techniques.

Ultrasonography

Ultrasound has become a major investigative modality for abdominal injury. It is particularly useful for detecting injury to parenchymal organs and the presence of free intraperitoneal fluid or blood. Rapid diagnostic information is facilitated by the fact that it is noninvasive, easily accessible, and less costly than other investigations [21–23]. The particular relevance to major liver injury is the focused assessment by ultrasound for trauma (FAST), often performed in the emergency department, which involves a rapid examination of several areas, namely, the pericardial region, right upper quadrant (including Morrison’s pouch), left upper quadrant, and the pelvis, specifically looking for free fluid [24]. This rapid assessment is excellent for evaluation of the unstable patient in the acute setting in the emergency department. A more definitive ultrasound scan to assess the integrity of the liver and other abdominal organs would require a more prolonged period of sonographic assessment by an experienced radiologist and therefore is unsuitable for the unstable patient.

Ultrasound scanning is very accurate for blunt and penetrating abdominal injuries, with specificity reported between 95 and 100% and sensitivity between 63 and 100%. Ultrasound has largely replaced diagnostic peritoneal lavage (DPL) in the initial assessment of blunt truncal injuries [21, 24–44]. DPL is not of value in isolated organ injuries or retroperitoneal injuries and can also result in positive results for intra-abdominal injuries in up to 30% of patients that do not necessarily require surgery [27, 45, 46]. Although FAST provides a rapid assessment of liver disruption and intraperitoneal bleeding, it is a limited scan that is highly operator dependent. It is very important to note that a negative FAST scan does not safely rule out injury [30, 47]. Due to the operator dependence of the modality, different end points, and inconsistent comparative gold standards in the studies, the reported specificities, sensitivities, and overall accuracies are variable [48–52]. It has been demonstrated that up to a quarter of hepatic and splenic injuries, as well as renal, bladder, pancreatic, mesenteric, and gut injuries, can be missed if ultrasound is used as the primary investigative modality in the stable patient. However, while the possibility of false negatives is ever present, the combination of a negative ultrasound scan and normal clinical examination and observations almost excludes liver injury in the event of significant blunt trauma [21, 47, 53–55]. In patients with significant hepatic injury, ultrasound cannot accurately determine the extent of hepatic parenchymal or vascular injury and therefore should not be a substitute for CT scanning if the patient is stable [25, 49, 53].

Computerised tomography

Spiral computerised tomographic (CT) scanning has become the standard evaluation modality for stable patients with an abdominal injury [12, 56–58]. The trend toward nonsurgical management of liver trauma is largely due to the availability of this technique [59]. In addition to increasing the rate of detection of liver lesions following trauma, CT has also helped to improve the understanding of the course of liver injuries [60]. CT has particularly high sensitivity and specificity for detecting liver injuries, which improve with increasing time between injury and scanning, as lacerations and haematomas become better defined (Fig. 1). Hoff et al. [49] reported a sensitivity of 92–97% and a specificity of 98.7%. The development of multislice CT has improved sensitivity, and more rapid imaging allows visualisation of the major vascular structures in different phases following contrast enhancement. In addition, reconstruction can be performed in multiple planes without significant loss of image quality [61].

The type and extent of liver injury can be readily identified by CT, particularly subcapsular and intraparenchymal haematomas, lacerations, and vascular injuries. CT also gives an estimation of the volume of haemoperitoneum and an indication of ongoing haemorrhage and is an essential element in nonoperative management of liver injuries [56, 58, 62]. The measured attenuation on CT allows differentiation between clotted blood (45-70 HU) and active bleeding (30-45 HU) [59, 60]. The highest attenuating collection indicates the presence of the sentinel clot and may allow localisation of the source of bleeding [63]. Active ongoing haemorrhage is visible and can be demonstrated on CT as extravasation of contrast material and is a strong predictor of failure of nonsurgical management [30, 59, 64, 65]. The presence of ongoing haemorrhage on CT has been suggested as an indication for intervention, whether in the form of surgery or interventional radiology [59, 60]. Although detection of free fluid may point toward a possible laparotomy, it is essential to remember that the haemodynamic stability of the patient will dictate the course of treatment, as there can be significant discrepancies found between CT and operative findings [24, 49, 57, 66–69]. The additional information gleaned from a CT scan will help in appropriate discussion with a specialist unit at an early stage, even in the haemodynamically compromised patient [16, 68].

CT is also invaluable in the detection of associated intraperitoneal and retroperitoneal injuries [69, 70]. Concurrent injuries can exist, affecting the spleen in 21% of cases, kidney in 8.9%, bowel in 4%, chest in 53.9%, and associated pelvic fractures in 22.4% of cases [60]. CT can also be utilised in the follow-up of complications as a result of liver trauma. Delayed haemorrhage, bile leaks, and abscess formation can all be assessed with concomitant therapeutic options such as percutaneous drainage of an abscess or biloma [59, 60]. Follow-up by repeat CT scanning is recommended only when clinically indicated and particularly for grade IV or V liver injuries [60]. This is best performed 7-10 days post injury to detect complications [59, 70].

Although CT is the investigative gold standard, it is important to remember that it involves exposure to high levels of ionising radiation and the use of intravenous contrast may compromise renal function [30]. In the majority of hospitals the use of CT requires movement of the patient away from adequate resuscitation facilities to the X-ray department, highlighting the importance of haemodynamic stability in patients with abdominal trauma being considered for CT examination [30].

Interventional vascular radiological techniques

Interventional radiological techniques have become an integral part of the management of abdominal trauma and have added a new dimension to the management of hepatic vascular injuries (Fig. 3) [71–74]. This multidisciplinary approach to the management of complex hepatic injuries is becoming much more important as the role of interventional radiology expands. Denton et al. [75] reported successful use of a combination of arterial embolisation and transhepatic venous stenting in the management of a grade V injury involving the retrohepatic vena cava in a patient whose injury had been temporarily controlled by perihepatic packing. A similar combined surgical and radiological approach with stent placement in a ruptured hepatic vein was reported by Burch in 1997 [76]. Recent more extensive series of angiography for control of hepatic haemorrhage have reported increasing success, with identification and control of bleeding rates ranging from 68 to 87% [71, 77–80]. CT criteria, including grade of hepatic injury, evidence of arterial vascular injury, and the presence of hepatic venous injury, are being used increasingly in the selection of patients who should undergo hepatic angiography and possibly embolisation [56].

Fig. 3
figure 3

CT scan of a left liver lobe injury with a haematoma (left) managed by angiography and embolisation (right)

Full size image

Angiography allows intervention at difficult-to-access locations [81]. This is important in both the pre- and postoperative stages of management [82]. Angioembolisation has been demonstrated to have a high success rate with low risk of rebleeding [77–79]. The increased role of angioembolisation has truncated any initial operative intervention and therefore stabilisation is reached sooner, with associated improved survival [79]. Further selective hepatic arterial embolisation following packing has been used effectively to control recurrent bleeding [75, 83, 84]. Sclafani et al. [84] reported a 93% success rate in the control of bleeding in a series of 60 patients by the use of coil embolisation, while others have demonstrated that the procedure reduces the volume of transfusion and the need for liver surgery [84–86].

In summary, angiography and embolisation or stenting is a very useful adjunctive technique in the stable patient who is being managed nonoperatively or in the patient who either has been stabilised by perihepatic packing or has rebled after a period of initial stability. It is likely that advances in interventional radiology will push the boundaries of nonsurgical management of liver trauma in the future.

Nonoperative management

Hogarth Pringle, in 1908, provided the first description of operative management of liver trauma [87]. Unfortunately, all eight patients died and Pringle recommended conservative nonoperative management of these patients. In the modern surgical literature, nonoperative management was first reported in 1972 and has been one of the most significant changes in the treatment of liver injuries over the last two decades [88–91]. This paradigm shift developed as a result of several factors: (1) the recognition that 50–80% of liver injuries stop bleeding spontaneously, (2) the precedent of successful nonoperative management in children, and (3) the significant development of liver imaging with CT scanning [69, 91–93]. While nonoperative management was initially introduced for minor injuries, it was soon in vogue for more severe injuries (grades III–V) [8, 9, 70, 94–98].

The mechanism of injury influences the management decision-making process. Evidence for the efficacy of nonoperative management of liver trauma accumulated throughout the 1990 s, with success rates ranging from 80 to 100% and documentation of significant reduction in blood transfusion requirements and reduced hospital stay [4, 10, 99–112]. Nonoperative management is usually recommended for stable patients following a stab injury [96, 98, 113–124]. There is increasing experience for the use of nonoperative management of gunshot wounds (which previously would have mandated exploration), and several authors report that these may be treated conservatively, provided there are no other significant injuries [115, 116]. Traditional fears relating to nonoperative management, such as increased sepsis rates due to infection of bile and blood collections, have been proved inaccurate [104]. Nonoperative management is not always successful, as demonstrated by a multicentre study that revealed that fewer than half the patients sustaining a blunt liver injury are suitable for nonoperative management [70].

The selection criteria for nonoperative management are always evolving and conservative approaches are increasingly adopted in the more severely injured. In particular, the threshold for volume of haemoperitoneum demonstrated on CT scan has increased [98, 117, 125]. Important assessment criteria for nonoperative management include (1) haemodynamic stability after resuscitation, (2) absence of signs of other visceral or retroperitoneal injuries that require surgery, and (3) the availability of an effective multidisciplinary team providing good-quality CT imaging, intensive care facilities, and suitably experienced surgeons [8, 9, 16, 70, 88, 96–98, 117]. While there has been considerable debate about the grade of liver injury and the acceptable volume of haemoperitoneum, it is now generally accepted that the ultimate decisive factor in favour of nonoperative management is the haemodynamic stability of the patient, irrespective of the grade of injury or the volume of haemoperitoneum. It is also essential that appropriate clinical and radiological follow-up is arranged [78, 96–98, 102, 118, 121, 124–127].

The failure rate of nonoperative management leading to the necessity to resort to open surgery is significantly higher in grade IV and V injuries compared to grade I–III injuries [70, 104]. However, the necessity to resort to surgical intervention is rarely due to liver-related complications [89]. The most common reason for surgical intervention in patients initially managed nonoperatively is coexisting abdominal injury such as delayed bleeding from the spleen or kidney [102]. Failure of nonoperative management due to delayed liver bleeding is rare (0–3.5%) [70, 97, 101, 102, 104]. Other factors identified as predictors of nonoperative management failure include age, haemoglobin, blood pressure, active extravasation on CT, and the need for blood transfusion [78, 89, 96, 104, 128].

As experience with nonoperative management has increased, it has become apparent that serial follow-up CT scanning is not necessary for patients with grade I–II injuries, provided the patient remains haemodynamically stable [129, 130]. For other patients, it is essential that appropriate clinical and radiological follow-up is arranged. Also, in those with large haematomas or significant vascular injuries (grades IV and V), it is still recommended to look for signs of further bleeding which would merit further investigation and management, e.g., angiography and embolisation or stenting [131].

Nonoperative management has become the standard of care in patients with blunt liver trauma, with a 23.5% reduction in mortality in grade III and grade IV patients [96, 99, 110, 132, 133]. The same technique has also emerged as effective management in appropriately selected patients with liver gunshot injuries [134, 135].

Operative management

Incision and initial control of bleeding

When nonoperative management is not possible, or fails, the surgeon must be prepared to conduct a resuscitative laparotomy. The most widely adopted incision for the patient with liver trauma is a long midline laparotomy, which can be extended to the right chest if a posterior right lobe injury, major hepatic venous injury, or vena caval injury is encountered. Historically, incisions were extended using an oblique incision to open the right thorax and diaphragm, but a median sternotomy may be used to access the chest. An effective alternative, which gives good exposure and avoids a thoracotomy, is a right subcostal extension (Fig. 4). A bilateral subcostal incision is sometimes favoured by hepatobiliary surgeons if there is an obvious penetrating through-and-through liver injury. This allows excellent exposure of the right lobe of the liver, the hepatic veins, and vena cava without having to open the chest or diaphragm; however, it does compromise access to the lower abdomen (Fig. 4).

Fig. 4
figure 4

Incisions for management of liver injuries. A long midline with a right subcostal extension, if necessary, and a bilateral subcostal

Full size image

If a major liver injury is encountered, immediate control of bleeding is an absolute priority because the greatest threat to the patient’s life at this juncture is exsanguination [136, 137]. The liver should immediately be manually closed and compressed (Fig. 5). Tamponade can then be maintained by packs, which can also be manually compressed if bleeding continues. If this still does not control the bleeding, pedicle occlusion (Pringle manoeuvre) should be applied using an atraumatic vascular clamp or noncrushing bowel clamp (Fig. 6). This manoeuvre can be both therapeutic and diagnostic. If a Pringle manoeuvre controls the bleeding it is unlikely that a major hepatic venous or vena caval injury has occurred. Although the recommended occlusion time is controversial, it is generally agreed that up to 1 h can be tolerated. If bleeding continues despite pedicle occlusion, a major vascular injury (hepatic venous or vena caval) is likely and further packing and manual compression should be used [138].

Fig. 5
figure 5

Bimanual compression of the liver to arrest bleeding

Full size image
Fig. 6
figure 6

Pringle manoeuvre for hepatic inflow occlusion

Full size image

These measures for rapid control of bleeding are important and should be maintained to help the anaesthetist achieve restoration of the blood volume and effective intraoperative resuscitation [139]. Attempts to identify and repair hepatic vascular injuries before effective resuscitation has been achieved should be avoided as they will invariably lead to further exsanguination, hypotension, acidosis, and coagulopathy [11, 139].

Damage control surgery

The concept of damage control surgery involves three principle phases [140]. Phase 1 involves initial control of haemorrhage and contamination followed by packing and rapid wound closure. This is followed by further resuscitation and stabilisation in the intensive care unit for a 24–48-h period until normal physiological parameters have been restored (phase 2). Phase 3 consists of re-exploration and definitive repair. The major aim initially, therefore, is to minimise the metabolic insult (particularly hypothermia, coagulopathy, and acidosis) without immediate concern for restoration of anatomical integrity. Implicit in this concept is limiting the operating time (abbreviated laparotomy) and avoidance of prolonged or complex attempts at anatomical reconstruction [79, 140–142]. This obviously includes avoidance of opening body cavities that have not been traumatised (e.g., thoracic cavity) as this increases operating time and increases the burden of injured tissue in an already unstable patient.

The damage control concept is very appropriate for the management of major liver injuries and, in fact, was initially described by Halsted in 1908 for the control of liver bleeding by packing. It was repopularised in the early 1980 s, particularly for patients who had developed a coagulopathy (nonsurgical bleeding), but was more widely adopted throughout the next two decades. Damage control approaches are associated with a significant survival advantage compared to traditional prolonged surgical techniques [79, 143].

The three key factors that interact to produce a deteriorating metabolic situation are hypothermia, coagulopathy, and acidosis. Patients in this condition are at the limit of their physiological reserve and persistence with prolonged and complex surgical repair attempts will cause exceptionally high mortality [140]. Early recognition of hypothermia, coagulopathy, and acidosis is the key to the damage control approach. It is recommended that definitive surgery should cease and a damage control approach be adopted when hypothermia is deteriorating or a temperature of 34°C is reached, when coagulopathy has developed (nonsurgical oozing or prothrombin time greater than 50% above normal), or when acidosis exists (pH <7.2 despite adequate volume resuscitation) [140].

Perihepatic packing

Elder in 1887 suggested that a liver injury causing haemorrhage would invariably be fatal [144]. Pringle’s landmark paper offered an operative strategy by which this blood loss could be temporarily stanched [87]. After the Second World War, Madding et al. [145] showed a reduced mortality using early laparotomy, drainage, and asepsis. Operative strategies then included parenchymal approximation with large stitches, vessel ligation, and resection. Due to mortality falling from 62.5 to 27.7% by these means, more aggressive strategies were adopted, rather than perihepatic packing [145, 146]. This has since been replaced by a return to favour of the art of perihepatic packing, with associated success.

Perihepatic packing will control profuse haemorrhaging in up to 80% of patients undergoing laparotomy and will allow intraoperative resuscitation (resuscitative packing) [138–143, 147–149]. In the management of severe injuries of the liver, packing has emerged as the key to effective damage control [147, 148]. However, more definitive “therapeutic” packing is also a very effective technique, particularly when used judiciously to prevent the cascade of hypothermia, coagulopathy, and acidosis [91, 150–152]. Although perihepatic packing was somewhat discouraged in the era when definitive surgical repair was popular, packing has become increasingly adopted during the last two decades [140, 153]. Packing is particularly useful for more extensive injuries (grade III–IV) but has also been shown to be effective for even the more major vascular injuries (grade IV–V) [15, 153, 154]. This technique is also extremely useful for the general surgeon in a district hospital as it can be life-saving until major surgery can be performed following transfer to a major trauma or hepatobiliary unit [139]. This on-going role of stabilising the patient to ensure safe transfer to another surgical institution is well recognised and its importance should not be underestimated.

The technique of perihepatic packing is straightforward. Following manual closure or approximation of the liver parenchyma, large, folded laparotomy packs are inserted over the diaphragmatic surfaces of the liver to produce a tamponade effect between the liver and the abdominal wall and thoracic cage. Krige et al. [11] recommend a “six pack” technique (Fig. 7). Packs should not be forced deep into liver fractures as this can extend the injury and cause venous tears and increased bleeding (Fig. 7).

Fig. 7
figure 7

Perihepatic packing. Forceful packing into a fracture should be avoided to prevent extension of the injury or venous tears

Full size image

Care must be taken to avoid excessive packing as this can cause vena caval and renal vein compression, leading to abdominal compartment syndrome (ACS). This is a particular risk with infrahepatic packs and these should be avoided if possible. To reduce the risk of ACS, some advocate closing the upper part of the wound to enhance the tamponade effect but leaving the lower two-thirds open and temporarily covered with a silastic sheet sutured to the skin edges [155, 156]. Appropriate perihepatic packing is essential since the efficacy of the procedure significantly impacts patient outcome [157].

Following resuscitation and stabilisation with correction of coagulation and metabolic deficits, packs are removed after approximately 36–48 h [153]. During this period broad-spectrum antibiotics should be given to reduce the risk of sepsis which occurs in 10–30% of cases [9, 156–161]. The exact timing of the removal of packs is controversial, but they should not be removed before 24 h as this is related to rebleeding and leaving them in place for 24 h or more does improve outcome [138, 149, 160–166]. Delayed removal (up to 1 week after injury) is not associated with an increased incidence of organ-specific or systemic complications [167–169]. Due to the risk of rebleeding during pack removal, some authors recommend insertion of plastic sheets between the liver and packs or insertion of an omental pedicle covered by a plastic sterile drape [161]. Alternatively, an absorbable mesh pack in children or a nonstick bowel bag in cases of extensive hepatic capsular disruption has been suggested to prevent rebleeding [170, 171].

Definitive surgical procedures for liver injuries

Following initial control of bleeding by manual compression, Pringle manoeuvre, and perihepatic packing to allow adequate resuscitation, the surgeon must decide on the next phase of management. If bleeding has stopped after careful pack removal, no further intervention is necessary. If bleeding persists, it is important to obtain adequate exposure and visualisation of the injury. Intermittent release of the Pringle clamp combined with effective suction may allow identification of deep bleeding sites and control by direct suture. If the bleeding is too profuse, experienced judgment is invaluable for deciding whether to continue with exploration and attempted repair or whether to opt for definitive perihepatic packing. This decision depends on the stability of the patient and the presence of adverse factors such as coagulopathy. In general, persistent bleeding at this stage will be due to major parenchymal disruption or a major vascular injury and it may be prudent to opt for definitive packing, particularly in an unstable patient with developing coagulopathy. If, however, the patient remains relatively stable with no signs of coagulopathy, a spectrum of operative techniques is available. No single technique is superior and applicable to every patient; therefore, the one selected will depend on the nature of the injury.

Hepatotomy and selective vascular suture or ligation

Combined hepatotomy and selective vascular ligation has emerged as the preferred method of management for major hepatic venous, portal venous, and arterial injuries in many centres [11, 153, 172]. It is performed under Pringle control and involves finger fracture or Kelly clamp extension of the laceration to allow suture or ligation of the bleeding vessels (Fig. 8). Intermittent release of the Pringle clamp allows detection of ruptured hepatic arterial or portal venous branches and their direct suture or clipping.

Fig. 8
figure 8

Hepatotomy and direct suture or ligation. Rapid division of parenchyma may provide immediate access for repair of major bleeding vessels

Full size image

The hepatotomy and ligation technique has been used extensively for control of bleeding and avoidance of packing with good results [11, 139, 153, 173]. For control of major vascular injuries, Pachter et al. [153] recommend a rapid and extensive finger fracture, often through normal parenchyma, to reach the site of injury. However, it is important to emphasise that with a major hepatic venous injury, significant haemorrhage may occur while attempting to extend a deep liver laceration and that this bleeding will not be controlled by a Pringle clamp and increased morbidity may be incurred. In such cases, or for a mashed liver following a high-velocity road traffic accident, hepatotomy should be abandoned and an alternative such as total vascular exclusion or manual closure and definitive packing should be adopted [139].

Hepatorrhaphy

Hepatorrhaphy was one of the earliest and most widely practised techniques for control of major haemorrhage from liver vascular injuries. It involved wide placement of large sutures (liver sutures) in the parenchyma to compress it and tamponade bleeding vessels. However, the compression necessary to stop the bleeding results in a significant risk of parenchymal ischaemia and necrosis and the technique is no longer recommended [174–177].

Selective hepatic artery ligation

Selective hepatic artery ligation was once widely used in liver trauma but has largely been replaced by more effective alternatives, e.g., hepatotomy and suture or manual closure and packing. In dire emergencies, large sutures encompassing the vascular structures of the right lobe can be used [163, 173]. One reason for not using this technique may be when the right or left hepatic artery has been significantly disrupted in a portal hepatic injury. In this situation it can simply be ligated without any risk of ischaemia of the liver lobe, provided the ipsilateral portal vein is intact. If the right hepatic artery is ligated, a cholecystectomy should be performed to avoid ischaemic necrosis [62, 178–181].

Nonanatomical resection (resectional debridement)

This refers to removal of devitalised parenchyma using the line of injury as the boundary of the resection rather than standard anatomical planes [107, 182]. It may be performed at the initial laparotomy or at relaparotomy following packing. It is essential that the patient is haemodynamically stable and not have a coagulopathy. The principle is to limit the extent of parenchymal dissection so that operating time is short and new tissue planes with further potential for bleeding are not created [139, 183]. Ideally, it should be achieved with minimal additional liver resection, but if this is necessary, finger fracture, a Kelly clamp, or ultrasonic dissection may be used [153]. Extensive dissection through uninjured parenchyma should be avoided if possible. In some cases simple completion of an extensive parenchymal avulsion may suffice, e.g., when there has been an avulsion of the posterior sector of the right lobe (segments VI and VII). This type of injury is often associated with a right hepatic vein laceration and completion of the “resection” will allow control and suture of this. In such situations, vascular stapling devices are extremely useful for rapid and secure division of major veins.

Anatomical resection

This involves resection along standard anatomical planes (usually right hepatectomy) after identification and control of the relevant inflow and outflow vessels. This was performed extensively in the 1960s and during the war in Vietnam, where a rapid right hepatectomy with no hilar control was introduced. Almost universally, extensive anatomical resection for trauma was associated with a very high mortality rate [101, 173]. This, plus the fact that the time and magnitude of the surgery goes against the later principles of conservative surgery and damage control, has resulted in anatomical resection being practised rarely and it is now performed in only approximately 2–4% major liver trauma cases [164].

However, anatomical resection does have significant merit as it removes the source of bleeding and sepsis. This is particularly so when a lobe is shattered or there is proximal ductal injury and devascularisation and repair attempts will inevitably fail. In addition, the traditional poor results and lack of enthusiasm for this technique have been contradicted by the results of some recent series, particularly that from Strong et al. [172] who achieved excellent results in a series of 37 patients, 11 of whom (33%) had grade V juxtahepatic venous injuries [172, 184]. These results probably reflect the fact that this procedure was performed in a specialist liver resection and transplantation unit, and while the majority of liver injuries continue to be managed initially in trauma centres or district hospitals, it is likely that more conservative and damage control procedures will remain the most widely practised techniques.

Total vascular exclusion

Total vascular exclusion was initially introduced for elective liver resection and was later used to manage major retrohepatic venous injuries. The technique involves clamping of the portal triad and infra- and suprahepatic IVC and therefore requires experience with mobilisation of the liver as done in liver resection and transplantation (Fig. 9). Excellent results were reported for this technique by Khaneja et al. [185] who used it to manage grade V penetrating injuries with 90% of patients surviving the operation and an overall survival rate of 70%. One major drawback of this technique is the effect of caval clamping which results in decreased venous return, leading to severe hypotension and circulatory collapse in an already hypovolaemic patient. As a result of this and the experience required to perform it, it is unlikely that this technique will become more widespread and perihepatic packing is likely to be preferable.

Fig. 9
figure 9

Total vascular exclusion for management of major vascular injuries

Full size image

Venovenous bypass

Total vascular exclusion can create physiological derangement in venous return that is not compatible with maintaining a cardiac output. This may be overcome by extracorporeal bypass, which involves shunting of blood, via a vortex pump, from the common femoral and mesenteric veins to the axillary or internal jugular veins, as used in liver transplantation [186, 187]. Experience with this is limited and its use is usually restricted to units with specialist transplantation experience.

Atriocaval shunt

The principle of the atriocaval shunt is that caval control is obtained above and below the liver while venous return from the IVC to the right atrium is maintained. Essentially, it was designed to achieve what can now be achieved by total vascular exclusion and venovenous bypass. The technique involves opening the chest via a median sternotomy and passing a shunt (chest drain) down into the IVC via the right atrial appendage (Fig. 10). The supra- and infrahepatic IVC are controlled by tapes and a Pringle clamp is applied, thus producing vascular isolation of the liver. In an attempt to avoid a sternotomy, a balloon shunt, which can be introduced via the saphenofemoral junction, has been developed [188].

Fig. 10
figure 10

Atriocaval shunt. Note the side holes cut in the chest drain in the region of the right atrium

Full size image

The atriocaval shunt, although logical in principle and initially widely adopted in trauma centres in the U.S., has been associated with very poor survival figures: “more authors than survivors”! [9, 139, 153, 173, 189–193]. Buechter et al. [189] reported a 90% mortality rate for ten patients with juxtahepatic venous injuries who were managed with a shunt, compared to 60% mortality for those managed with total vascular exclusion. Similarly, Burch et al. [194] reported a 67% mortality rate when an atriocaval shunt was used compared to a 47% mortality rate with nonshunting alternatives. In addition, the necessity for a thoracotomy adds further injury to an already severely injured patient and goes directly against the later concept of damage control surgery. For these reasons, atriocaval shunting has largely been replaced by the alternatives described earlier [189, 195, 196].

Liver transplantation

Transplantation in a small number of patients with massive liver damage or grade VI avulsion injuries has been reported [197, 198]. The key obstacle to this approach is how to keep the patient alive while waiting for a suitable graft. Several techniques have been introduced for maintaining venous return and splanchnic decompression during the anhepatic phase. These include venovenous bypass and construction of a temporary end-to-side portocaval shunt [199, 200]. However, while liver transplantation may be life-saving for major liver trauma, the logistical problems will mean that it remains a limited option, available only in specialist centres.

Summary

This literature review has shown that the management of injuries of the liver has evolved significantly throughout the last two decades. Nonoperative techniques for the management of grade IV–V injuries in stable patients have been established, although there is a higher failure rate for these injuries compared with grade I–III injuries. Interventional radiological techniques have become more widely used in patients who are being managed nonoperatively or who have been stabilised by perihepatic packing. In unstable patients immediate control of bleeding is critical and the recommended techniques are manual compression, Pringle manoeuvre, and perihepatic packing. In terms of surgical management there has been a definite move away from major, time-consuming procedures toward conservative surgery and damage control. The preferred surgical technique for inaccessible bleeding within a laceration is rapid finger fracture hepatotomy and direct suture or ligation. Prolonged attempts at surgical control and repair should be avoided, and definitive perihepatic packing should be employed at an early stage in the persistently unstable patient or at the first signs of coagulopathy. Packing is also the recommended technique for the inexperienced surgeon, to allow control and stabilisation prior to transfer. Nonanatomical resection (resectional debridement) is recommended when there is unviable parenchyma. Anatomical resection is generally reserved for a devascularised lobe with a major ductal injury. Hepatorrhaphy and selective arterial ligation are no longer recommended. As a result of the high mortality associated with atriocaval shunting, this technique is also no longer recommended and has been replaced largely by perihepatic packing or total vascular exclusion, with or without venovenous bypass.