Mechanisms of liver injury

Blunt and penetrating trauma are the two principal mechanisms of liver injury. Road traffic accidents account for the majority of blunt injuries, whereas knife and gunshot wounds constitute the major cause of penetrating injuries. In the UK, blunt trauma predominates by a ratio of approximately 2:1 as documented in a large Scottish epidemiological study.¹ Whilst this is typical for other European centres,² it differs from the experience in South Africa, where penetrating injuries account for 66% of liver trauma,³ and in North America, where up to 86% of liver injuries are penetrating wounds.^4,5

Two types of blunt liver trauma have been described – deceleration (shearing) trauma and crush injury. Deceleration injuries occur in road traffic accidents and in falls from a height where there is movement of the liver relative to its diaphragmatic attachments.⁶ Crush injuries are caused by direct trauma to the liver area. The two types of injury may coexist but tend to produce somewhat different types of liver injury. Deceleration or shearing injuries create lacerations in the hepatic parenchyma, typically between the right posterior section (segments 6 and 7) and the right anterior section (segments 5 and 8), which can extend to involve major vessels. In contrast, a direct blow to the abdomen may lead to a crush injury, with damage to the central portion of the liver (segments 4, 5 and 8). Compression between the right lower ribs and the spine may also cause bleeding from the caudate lobe (segment 1). Blunt trauma can rupture Glisson’s capsule and can also lead to subcapsular or intraparenchymal haematoma formation. Penetrating injuries are usually associated with gunshot or stab wounds, with the former usually resulting in more tissue damage due to the cavitation effect as the bullet traverses the liver substance.

Injury to the hepatic veins and juxtahepatic vena cava can occur as a result of shearing stress in blunt trauma. It is worth noting that there may not be initial exsanguinating haemorrhage as the weight of the liver may provide some compression.

Classification of liver injury

The severity of liver trauma ranges from a minor capsular tear, with or without parenchymal injury, to extensive disruption involving both lobes of the liver with associated hepatic vein or vena caval injury. The American Association for the Surgery of Trauma has adopted for general use the classification of liver injury described initially in 1989 by Moore and colleagues, and revised subsequently in 1994⁷ (Table 17.1). The hepatic injury grade is calculated from assessment of the liver injury using information derived from radiological study, operative findings or autopsy report. Where there are multiple injuries to the liver, the grade is advanced by one stage. Grade I or II injuries are considered minor; they represent 80–90% of all cases and usually require minimal or no operative treatment. Grade III–V injuries are generally considered severe and may require surgical intervention, while grade VI lesions are regarded as incompatible with survival. Schweizer et al. have described a protocol-based liver trauma management system employing this classification system that permits lesser injuries to be treated non-operatively and allows more appropriate selection of patients for operative treatment.⁸

The initial assessment and management of an injured patient should proceed according to the Advanced Trauma Life Support (ATLS) guidelines of the American College of Surgeons Committee on Trauma. The initial focus of attention is on the patient’s airway, breathing and circulation. The airway is secured, intravenous access established and fluid resuscitation commenced.

The role of ‘aggressive’ high-volume fluid replacement in trauma victims has been questioned, with evidence suggesting that excessive fluid replacement is associated with adverse outcome.⁹ As this evidence came from an American series that included a large proportion of relatively young, previously fit adults suffering from penetrating trauma to the torso, with ready access to trauma centres, the results may not necessarily be applicable to practice in other countries.

Diagnosis of liver injury

In penetrating abdominal trauma, hepatic injury should be considered in any patient with a wound to the abdomen. Hepatic injury should also be considered in patients with penetrating low thoracic wounds and also in posterior penetrating wounds below a coronal plane at the tips of the scapulae.

Patients with major hepatic injury may present with profound clinical shock and abdominal distension. Hypotension resistant to fluid resuscitation combined with gross abdominal distension is an indication for immediate laparotomy. The operative management options for patients in this situation will be discussed in detail subsequently. Emergency room thoracotomy with cross-clamping of the descending thoracic aorta is a dramatic intervention, but even in centres where this technique is advocated the outcome is poor.

In Feliciano et al.’s series of 1000 patients with liver trauma treated during a 5-year period, 45 patients underwent emergency room thoracotomy for control of haemorrhage related to their liver injury and all died.⁴ Similarly, in an 11-year review of 783 patients who sustained liver trauma in Scotland, 11 patients underwent an unsuccessful laparotomy or thoraco-laparotomy in the emergency room.¹

In less dramatic situations, with a patient who is haemodynamically stable or responds to fluid resuscitation, appropriate investigations can be employed to obtain more information regarding the liver injury and to ascertain whether there is coexisting intra-abdominal visceral injury. During the initial survey a detailed clinical history is taken. Particular attention is paid to the mechanism of a road traffic accident, with supplemental information from ambulance crew, witnesses or police being used to piece together a picture of the accident. Speed of vehicle, position of occupant in vehicle, use of seatbelts, employment of airbag restraint systems and a history of ejection of the patient from the vehicle are important items of information. Conscious patients may complain of abdominal pain. Shoulder tip pain may arise from blood in the subdiaphragmatic space causing phrenic nerve irritation.

As resuscitation proceeds, a detailed physical examination is carried out. On inspection, attention is paid to the presence of anterior abdominal wall bruising, which may indicate compression from a seatbelt, and flank bruising, which may indicate retroperitoneal extravasation of blood. Signs of localised or generalised peritonitis are recorded in the conscious patient. In this context it should be noted that although there is evidence that the use of opiate analgesia will not significantly obscure physical signs in patients with acute abdominal pain, these findings have not been confirmed in abdominal trauma patients where the situation may be complicated by head injury, alcohol intoxication or the requirement for assisted ventilation.

Baseline investigations consist of a full blood count (for haemoglobin and haematocrit), serum urea and electrolytes, serum amylase, a coagulation screen, and blood for crossmatching. An erect chest radiograph and a plain abdominal film can be taken if the patient is sufficiently stable. In the context of diagnosing liver injury, features that may be of relevance include fractures of the lower ribs, elevation of the right hemidiaphragm and loss of the psoas shadow suggesting retroperitoneal bleeding. Retroperitoneal perforation of the duodenum may give rise to soft tissue shadowing in the right upper quadrant, loss of the psoas shadow and occasionally extraluminal gas may be noted.

Following initial assessment, patients who are conscious but have haemodynamic instability resistant to fluid resuscitation and with clinical signs of peritonitis should undergo laparotomy. In patients who are haemodynamically stable and have suspected liver injury, further diagnostic tests may be undertaken at this stage to define the nature of the injuries. An ideal test will establish the presence and extent of any liver injury together with providing information on concomitant visceral injury.

Formerly, diagnostic peritoneal lavage (DPL) was the procedure of choice for the quick diagnosis of haemoperitoneum, particularly in patients with an impaired level of consciousness and equivocal physical signs. However, DPL is invasive and a positive result for blood provides no information regarding either the site or the nature of the injury, and in the context of liver injury may lead to patients undergoing surgery where they may be better treated non-operatively.

An alternative investigation that has been advocated in initial trauma evaluation is focused assessment with sonography for trauma (FAST).¹⁰ This involves ultrasonographic assessment of the pericardium, right upper quadrant including Morrison’s pouch, left upper quadrant and pelvis. This evaluation is not designed to identify the degree of organ injury, but rather the presence of blood. A large meta-analysis of the use of emergency ultrasonography for blunt abdominal trauma reported sensitivity rates ranging from 28% to 97% and specificity rates close to 100%.¹¹

Rozycki et al. demonstrated a significant correlation between haemoperitoneum in the right upper quadrant and injury to the liver, and suggested that adherence to a pre-agreed protocol increased the reliability of ultrasound assessment of abdominal trauma.¹² Other centres have also reported that ultrasound is a reliable ‘first’ test for the assessment of a patient with suspected liver trauma.¹³ However, an important cautionary note comes from the study carried out by Richards et al.¹⁴ In a series of 1686 abdominal ultrasound scans for trauma, 71 patients had bowel or mesenteric injury and 30 patients had a negative ultrasound scan (43% false-negative rate). Limitations of FAST include operator dependence, poor assessment of the retroperitoneum, unreliable detection of pneumoperitoneum and difficulty in scanning obese patients or those with overlying wounds.

Computed tomography (CT) is the ‘gold standard’ investigation for the evaluation of a patient with suspected liver trauma (Fig. 17.1). The use of intravenous contrast may help in the detection of non-viable parenchyma. CT has high sensitivity and specificity for detecting liver injuries; these attributes increase as the time between injury and scanning increases, as haematomas and lacerations become better defined. Specific CT features of liver trauma have been reported by a number of authors. Fang et al. described intraparenchymal ‘pooling’ of intravenous contrast that correlated strongly to the presence of ongoing haemorrhage.¹⁵ Yokota and Sugimoto documented ‘periportal tracking’ to consist of a circumferential area of low attenuation around the portal triad.¹⁶ Periportal tracking is thought to represent blood or fluid within the condensation of the Glissonian sheath around the portal structures and indicates the presence of injury to structures in the portal triad. If the sign is present in the periphery of the liver it may alert the clinician to the presence of a peripheral bile duct injury that in turn may present as a bile leak. Addition of oral contrast does not appear to add to the diagnostic yield of CT in the assessment of liver injury.¹⁷

In order to maintain a balanced perspective, it is worthwhile considering some of the limitations of CT in the assessment of liver trauma. The CT-defined grade of injury may differ from the grade of liver injury found at operation, with the predominant tendency being to overdiagnose the grade of injury on CT as compared with subsequent operative findings. Croce et al. concluded that CT should not be used in isolation to estimate blood loss and that CT may not provide an accurate assessment of the extent of a liver laceration in some areas of the liver – specifically in the vicinity of the falciform ligament.¹⁸

Bearing the above limitations in mind, CT will help define the extent of the liver injury and will be of value in the detection of injury to other intra-abdominal viscera, in particular pancreatic injury. CT will allow the liver injury to be graded and thus will provide objective information that is mandatory if non-operative treatment is to be contemplated. Further refinements now permit accurate three-dimensional image reconstruction, and technical modifications such as helical CT combined with intravenous contrast allow demonstration of the biliary tree (CT cholangiography) or vascular anatomy (CT angiography).

Some authors recommend performing a whole-body CT as the standard diagnostic tool during the early phase for patients with polytrauma, advocating that this will alter treatment in up to 34% of patients with blunt trauma.¹⁹ A 30% reduction in mortality using this approach has also been reported.²⁰ Other arguments in favour of an imaging survey are the reduction in time from admission to intervention and consistency in managing haemodynamically unstable patients.²¹ However, at present the logistics of such an approach are not universally applicable as it requires a CT scanner in, or very close to, the emergency department.

Management of liver injury: selection of patients for non-operative management

The feasibility of non-operative management of patients with intra-abdominal solid-organ injury was first established in paediatric surgery but was subsequently extended to adult practice. Richie and Fonkalsrud described successful conservative management of four patients with liver injury in an era before the availability of CT.²⁶ Further indirect evidence for the feasibility of a non-operative approach came from a report published by White and Cleveland²⁷ in the same year. They reported a consecutive series of 126 patients with liver trauma, all of whom underwent laparotomy. Interestingly, 67 patients in this series (53%) had placement of a drain to the subhepatic space as their only liver-related surgical intervention at laparotomy. Subsequent studies have recognised that 50–80% of liver injuries stop bleeding spontaneously and this has led to a non-operative approach for blunt liver trauma in selected patients.

Non-operative management of liver trauma is now a well-established treatment option. Trunkey’s group in Portland, Oregon, first defined in 1985 the following criteria for the selection of patients for non-operative management:

Farnell et al. extended the threshold of haemoperitoneum to 250 mL and described specific liver injuries suitable for non-operative management.²⁹ Feliciano suggested subsequently that any blunt hepatic injury, regardless of its magnitude, should be managed without operation if the patient was haemodynamically stable and had a haemoperitoneum of less than 500 mL.³⁰ The degree of liver injury amenable to successful non-operative management has gradually extended over recent years, and most authors now believe that the ultimate decisive factor in favour of non-operative management is haemodynamic stability of the patient at presentation or after initial resuscitation, irrespective of the grade of liver injury on CT or the amount of haemoperitoneum.^31,32

A 22-month prospective study from Memphis of the initial non-operative treatment of haemodynamically stable blunt hepatic trauma patients compared outcome to a matched cohort of blunt hepatic trauma patients treated operatively.³³ The study reported that of 136 patients with blunt trauma, 24 (18%) underwent emergency surgery. Of the remaining 112 patients, 12 (11%) failed conservative management (for causes not related to the liver injury in seven) and the remaining 100 patients were treated successfully without operation. Of these, 30% had minor injuries (grades I and II) but 70% had major injuries (grades III–V). This study concluded that non-operative management was safe for haemodynamically stable patients and that this was independent of the CT-delineated grade of the liver injury. The blood transfusion requirement and the incidence of abdominal complications were lower in the non-operatively treated group.

Reporting a single institutional experience, Boone et al. stated that 46 (36%) of 128 consecutive patients with blunt liver trauma were successfully treated non-operatively, including 23 patients with grade III and IV injuries.³¹ A review of 495 patients from the published literature noted a success rate for non-operative treatment of 94%.³⁴ This was accomplished with a mean transfusion rate of 1.9 units, a complication rate of 6% and a mean hospital stay of 13 days. There were no liver-related deaths, nor were there any missed enteric injuries.

The current consensus view is that successful selection of patients for conservative treatment after blunt abdominal trauma cannot be carried out by CT alone, but that an overall assessment of suitability for such an approach must take into account the findings of careful repeated clinical examination and the results of close monitoring of haemodynamic and haematological parameters. If non-operative management is selected, haemodynamic instability is the predominant indication for intervention early in the clinical course whilst intervention (often radiological or endoscopic) may be required later for management of bile leak or intrahepatic collections.

If a non-operative strategy is selected it should be borne in mind that the risk of hollow-organ injury increases in proportion to the number of solid organs injured³⁵ and that there is a small but significant risk of delayed haemorrhage. However, it appears that the natural course of liver injuries is more analogous to that of lung or kidney injuries, rather than splenic injuries, in that any deterioration is usually gradual, with a fall in haemoglobin level or an increase in fluid requirement, rather than acute haemodynamic decompensation. Therefore, with close supervision, patients who fail with an initial non-operative approach can be detected early and treated appropriately.

Although non-operative management of haemodynamically stable patients with liver trauma has become the standard of care over the past decade, the role of in-hospital follow-up CT to monitor the injury remains controversial. Demetriades et al. reported that follow-up CT at a mean of 10 days after surgical intervention showed a 49% incidence of liver-related complications, most of which required subsequent intervention.³⁶ However, other authors suggest there is little evidence that follow-up CT provides additional information and rarely changes management.³⁷ In the author’s practice, in-hospital follow-up scan is not employed routinely unless the patient develops relevant symptoms or signs, but a follow-up scan 4–6 weeks later is undertaken to ensure resolution of the injury.

The management policy for abdominal gunshot injuries in most centres continues to be a mandatory laparotomy, regardless of the clinical presentation;³⁸ however, several studies have reported successful non-operative management of selected liver gunshot injuries.^39,40 In the study by Omoshoro-Jones et al., 26.6% of patients who presented with liver gunshot injuries were managed non-operatively, with an overall success rate of 94% and a morbidity rate of 36%, of which 3% were liver related.³⁹ This approach is associated with the risk of failure to detect concomitant intra-abdominal visceral injury and therefore should only be considered in specialist centres with experience in management of liver trauma and appropriate facilities to deal with any complications that arise.

Operative management of liver injury