Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Monitoring of intracranial pressure (ICP) in acute liver failure (ALF) is controversial as a result of the reported complication risk (∼20%) and limited therapeutic options for intracranial hypertension. Using prospectively collected information from 332 patients with ALF and severe encephalopathy, we evaluated a recent experience with ICP monitoring in the 24 centers constituting the U.S. ALF Study Group. Special attention was given to the rate of complications, changes in management, and outcome after liver transplantation (LT). ICP monitoring was used in 92 patients (28% of the cohort), but the frequency of monitoring differed between centers (P < 0.001). ICP monitoring was strongly associated with the indication of LT (P < 0.001). A survey performed in a subset of 58 patients with ICP monitoring revealed intracranial hemorrhage in 10.3% of the cohort, half of the complications being incidental radiological findings. However, intracranial bleeding could have contributed to the demise of 2 patients. In subjects listed for LT, ICP monitoring was associated with a higher proportion of subjects receiving vasopressors and ICP-related medications. The 30-day survival post-LT was similar in both monitored and nonmonitored groups (85% vs. 85%). In conclusion, the risk of intracranial hemorrhage following ICP monitoring may have decreased in the last decade, but major complications are still present. In the absence of ICP monitoring, however, patients listed for LT appear to be treated less aggressively for intracranial hypertension. In view of the high 30-day survival rate after LT, future studies of the impact of intracranial hypertension should also focus on long-term neurological recovery from ALF. (Liver Transpl 2005;11:1581–1589.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Brain edema and intracranial hypertension are major causes of morbidity and mortality in acute liver failure (ALF).1 The exact role of intracranial pressure (ICP) monitoring in the management of ALF, however, is still debated. While a recent position paper from the American Association for the Study of Liver Diseases2 states that ICP monitoring is especially useful in the decision to exclude patients from emergency liver transplantation (LT), other experts have noted the ability to manage ALF in the absence of such data.3 The main concern with placement of such monitors is the risk of intracranial hemorrhage. In a survey across medical centers in the United States in the early 1990s, an overall prevalence of intracranial bleeding of 20% was noted with the procedure.4 Hemorrhage was more likely when the dura mater was pierced for the placement of subdural or parenchymal transducers. It was recommended that epidural transducers be utilized, as the risk of bleeding appeared to be reduced with such devices.
In the decade that has elapsed since the report,4 3 developments have occurred that call for a reassessment of the experience with ICP monitoring. First, the manufacture of epidural transducers has ceased, as patients in whom monitoring is frequently used (for head trauma) are better served by the more precise measurements of ICP obtained by subdural or parenchymal transducers. Second, newer approaches to the correction of the coagulopathy of ALF have been proposed. Among them, the combination of recombinant factor VII and fresh frozen plasma holds promise as a tool to reduce the complications associated with invasive procedures in ALF.5 Finally, as we enter an era of large therapeutic trials in ALF, one of them recently completed,6 the interpretation of survival results could be affected by the availability (or not) of results of ICP monitoring.
The U.S. ALF Study Group (ALFSG) has been actively maintaining a prospective registry of patients with this condition since its inception in 1998.7 A diverse practice of ICP monitoring was noted among the different centers that constitute the group. With such differing practices, a controlled evaluation of the role of ICP monitoring could not be undertaken. Rather, we examined the experience of the U.S. ALFSG with placement of ICP monitors in adult patients, with special reference to the rate of complications; changes in clinical management and outcomes of patients in whom ICP monitors were placed prior to the performance of LT were also examined. Our results note the continued presence of complications from the procedure.
While the outcome of the transplant procedure was similar in the presence or absence of ICP monitoring, marked differences in management of intracranial hypertension may be of importance for the long-term neurological recovery of patients surviving ALF.
Between January 1998 and March 2004, information from 704 patients with ALF admitted to the 25 centers that constitute the U.S. ALFSG was prospectively collected in a standard protocol form. Inclusion criteria were those defining ALF,8 namely the presence of coagulopathy (prothrombin time >15 sec or international normalized ratio ≥1.5) and any grade of hepatic encephalopathy within 26 weeks of the first symptoms in a patient with acute liver injury and no previous liver disease. We excluded 190 of the 704 patients from our study because the item collecting information on ICP monitoring was not included in the protocol forms at the time of their data collection (prior to April 2000). A further group of 182 patients did not reach grade III-IV hepatic encephalopathy during the data collection period and was also excluded. Therefore, 332 patients with ALF and grade III-IV hepatic encephalopathy were examined.
No standardized management protocol was used by the centers during the collection of the data. Thus, monitoring and therapeutic interventions were implemented by each center according to their own standard of care. Data regarding the use of surrogate markers of intracranial pressure, such as cerebral blood flow/velocity, for helping the indication of ICP were not collected in the forms. The U.S. ALFSG protocol was approved by the institutional review board of all participant institutions. Since patients were by definition encephalopathic, informed consent was obtained from the patient's next of kin or guardian in all cases.
Registry Form Description and Data Collection
Each case report form was divided into 2 parts. The initial form, filled out at study admission, included demographic, epidemiological, clinical, and laboratory data. A second outcome form was completed at death, at discharge, or 3 weeks after study admission if the patient was still in hospital. This form included detailed information about the hospital course and patient outcome. Clinical findings, laboratory data, and monitoring and therapeutic measures employed during the first 7 days of the study were also collected. After completion, each site sent the protocol forms to the University of Texas Southwestern Medical School. Before data entry into the central database, sites were queried about missing or questionable data items, source documents were reviewed, and corrections were made. In addition, annual audits at each site helped to confirm accuracy of data.
Complications and Details of ICP Monitoring
Information regarding complications and details of ICP monitoring was not included in the standard protocol forms. Thus, centers that used ICP monitoring in more than 5 patients were contacted and asked to provide specific information regarding (1) type of ICP transducer used (epidural, subdural, intraparenchymal, or intraventricular), (2) hemostatic preparation used for ICP placement (fresh frozen plasma, platelets, cryoprecipitate, recombinant factor VIIa, or plasmapheresis), (3) number of days with ICP monitoring, (4) major reason for removal of ICP device (medical improvement, death, malfunction of transducer, or other), and (5) complications of ICP monitoring (hemorrhage, infections, or other).
Analysis of Outcome
Due to the nonrandomized nature of our study, we focused the analysis of outcome on the patients undergoing liver transplantation. A survey was sent to all centers, and information regarding the 30-day survival post-LT was collected.
Numerical results are expressed as mean ± SD. Categorical data are expressed as counts (%). Differences of categorical and continuous variables were assessed using the chi-square or Fisher exact test, and the independent samples t test, respectively. The natural logarithms of aspartate aminotransferase and alanine aminotransferase were used in the analysis to reduce the heterogeneity of variance. The Cochran-Mantel-Haenszel test for general association was used, when indicated, to control for differences between centers. P values <0.05 were considered significant.
General Patient Characteristics
Intracranial pressure monitoring was used in 92 (27.7%) out of the 332 patients with ALF and grade III-IV hepatic encephalopathy included in the analysis. General characteristics of patients with ICP monitoring (ICP group) and without (non-ICP group) are shown in Table 1. The distribution of sex and race was similar, but patients with ICP monitoring were significantly younger. The etiology of liver failure was comparable in both groups, although the proportion of viral and indeterminate etiologies tended to be higher in the ICP group. Interval between first symptoms and appearance of encephalopathy, mean arterial pressure, presence of various clinical complications, and laboratory tests were also similar in the ICP group (at the day of placement of the ICP monitoring device) and in the non-ICP group (at the day of development of grade III-IV encephalopathy), except for a higher arterial pH in patients with ICP monitoring.
Table 1. Demographic, Clinical, and Biochemical Characteristics of Patients With and Without ICP Monitoring
No ICP Monitoring
Number (%) or Mean±SD
Number (%) or Mean±SD
Abbreviations: MAP, mean arterial pressure; WBC, white blood cell count; INR, international normalized ratio; NS, not significant; ALT, alanine aminotransferase; AST, aspartate aminotransferase; pCO2, partial pressure of carbon dioxide.
Values refer to the day of placement of the ICP monitoring device or, in patients without ICP monitoring, to the day of grade III-IV hepatic encephalopathy.
At the time of placement of the ICP transducer, 61 of the 92 patients (66.3%) had grade IV encephalopathy. Similarly, 168 of the 240 patients (70%) in the non-ICP group reached grade IV encephalopathy during the study. Forty-eight percent of ICP transducers were placed the first day of the data-collection period, 24% the second day, and 28% the third day or afterward. The progression to grade III-IV encephalopathy in patients without ICP monitoring followed a similar chronological pattern (first day, 58%; second day, 19%; third day or afterward, 23%, not significant).
Factors Associated With the Placement of ICP Monitoring Devices
The frequency of utilization of ICP monitoring differed significantly between individual centers (P < 0.001). As shown in Figure 1, 3 centers used ICP monitoring in 50% or more of the patients, 8 centers used it in 25 to 50% of the patients, and 8 centers used it in less than 25 % of the patients (2 of these centers did not use ICP monitoring for any patient). The different practice of ICP monitoring across centers was more evident in patients who were not listed for LT (P < 0.01) than in patients listed for LT (P = 0.086).
A negative drug-screening test at admission and listing for urgent LT were positively associated with monitoring of ICP, with a higher proportion of patients being listed for LT in the ICP (68 of 92 [73.9%]) than in the non-ICP group (76 of 239 [31.8%], P< 0.001) (Table 1). The association between ICP monitoring and listing for LT was also present after controlling for center (P< 0.001, Cochran-Mantel-Haenszel test for general association).
Complications and Details of ICP Monitoring
Specific information regarding complications and other details of ICP monitoring was obtained from 58 patients (mean age, 35 ± 13.1 yr, 34 [59%] females) who underwent ICP monitoring in 8 centers. Acetaminophen intoxication was the etiology of ALF in 24 patients (41%). Subdural transducers were the most commonly used (37 of 58 [63.8%]), followed by the parenchymal (12 of 58 [20.7%]) and epidural localizations (9 of 58 [15.5%]) (Figure 2A). Intraventricular transducers were not used in any of the centers. The intracranial localization, however, was highly center-dependent, with 6 centers using subdural transducers in 80-100% of the cases, 1 center using a parenchymal device in 91% (10 of 11 patients), and another using only epidural transducers (7 of 7 patients).
Fresh frozen plasma was the most common product used to correct coagulation prior to placement of the ICP transducer (53 of 58 patients, [91%]), but it was used in combination with other products in 59% of the cases (Figure 2B). Other products used were platelets (18 of 58 patients), cryoprecipitate (11 of 58) and recombinant factor VIIa (15 of 58), mostly in combinations between them and/or with fresh frozen plasma. In one center, plasmapheresis was used in 3 patients (in combination with other measures in 2 of them).
The mean duration of ICP monitoring was 4.4 ± 2.3 days (median, 4 days; range, 1 to 10). Improvement of the medical condition was the reason for removal of the ICP transducer in 34 patients (59%), whereas in 19 cases (33%) it was due to the death of the patient (Figure 2C). Malfunction led to the removal of the transducer in 3 patients with subdural devices.
Complications attributed to ICP monitoring were noted in 6 of the 58 patients (10%), all of them consisting in intracranial hemorrhages. Fresh frozen plasma was used in the haemostatic preparation in the 6 patients, and it was combined with recombinant factor VII in 2 of them. Three of the hemorrhages (5.2%) were associated with clinical symptoms. One occurred in a patient who had an epidural transducer placed the same day of undergoing LT; he is currently alive after 2 years of follow-up. The other 2 clinically relevant hemorrhages occurred in patients with subdural transducers; both of them died with severe brain edema, 1 after undergoing LT.
The remaining 3 complications consisted in small subarachnoid or parenchymal hemorrhages noted in computed tomography scans of the brain in 2 patients with subdural transducers and in 1 patient with an intraparenchymal device. Two of these patients survived ALF, 1 after undergoing LT. The third patient died without LT as a consequence of refractory hypotension.
Management of ICP With and Without ICP Monitoring
Because the severity of disease in patients listed for urgent LT is relatively homogenous, this population was chosen to study the impact of ICP monitoring on clinical management (Table 2). After placement of the ICP transducer, the management of patients with ICP monitoring differed significantly from that of patients who did not undergo such monitoring. In particular, monitoring of ICP was associated with a higher proportion of patients being treated with mannitol (35 of 65 [53.8%]) vs. non-ICP group (10 of 74 [15.7%], P < 0.001), barbiturates (14 of 64 [21.9%] vs. 4 of 76 [5.3%], P< 0.01), and vasopressors (39 of 65 [60.0%] vs. 27 of 76 [35.5%], P< 0.01). The proportion of patients treated with such medications during the first 3 days of ICP monitoring (ICP group) or development of stage III-IV encephalopathy (non-ICP group) is shown in Figure 3.
Table 2. Demographic, Clinical, and Biochemical Variables of Patients With and Without ICP Monitoring Who Underwent Liver Transplantation
No ICP Monitoring
Number (%) or Mean±SD
Number (%) or Mean±SD
Abbreviations: MAP, mean arterial pressure; WBC, white blood cell count; ALT, alanine aminotransferase; AST, aspartate aminotransferase; pCO2, partial pressure of carbon dioxide.
37 ± 15
34 ± 12
Days from onset of symptoms to coma
12.8 ± 13.7
16.2 ± 15.9
88 ± 19
90 ± 14
10.3 ± 1.7
10.3 ± 1.4
WBC (× 109/L)
12.0 ± 8.7
12.3 ± 6.2
Platelets (× 109/L)
141 ± 94
124 ± 76
Prothrombin time (s)
35.1 ± 22.4
28.0 ± 14.7
17.5 ± 12.2
17.3 ± 12.5
1,668 ± 2,121
1,282 ± 2,159
2,371 ± 3,531
1,163 ± 1,913
1.69 ± 1.10
2.07 ± 1.91
7.41 ± 0.13
7.48 ± 0.07
pCO2 (mm Hg)
34.0 ± 8.5
34.3 ± 6.1
One hundred and forty-four patients were listed for urgent LT; the proportion of patients removed from the waiting list was similar regardless of ICP monitoring (ICP group, 25 of 68 [37%] vs. non-ICP group, 27 of 76 [36%], not significant). Information regarding the primary reason for removal was available in 14 of 25 patients of the ICP group and 19 of 27 of the non-ICP group: improvement of liver function (5 of 14 vs. 6 of 19), presence of irreversible brain damage (4 of 14 vs. 4 of 19), and medical instability (4 of 14 vs. 5 of 19). Sepsis, comorbidities, and refusal of LT by the family were causes of removal in 5 more patients. In addition, there were 7 patients (4 with ICP monitoring) who were not removed from the list but who did not receive an organ; only 2 of them survived (1 in each group).
Outcome of Patients undergoing Liver Transplantation
In addition to the nonrandomized nature of our study, the strong association between listing for LT and indication of ICP monitoring precludes a valid comparison of global outcomes between the ICP and non-ICP groups. Therefore, we focused the outcome analysis on the more restricted and homogeneous setting of the patients that underwent LT.
Eighty-five patients underwent LT, 45 in the non-ICP group and 40 in the ICP group. Sex, age and etiology of liver failure were similar in both subgroups of patients (Table 2). Most laboratory data within 24 hours prior to LT were also similar in both subgroups. The shorter interval between onset and coma, and the higher platelet counts and aminotransferases, however, could indicate an earlier time point in the course of the disease in the non-ICP group.
The time that patients spent on the waiting list before receiving a donor organ tended to be longer in patients with ICP monitoring (3.3 ± 3.5 days; median, 2 [range 0 to 19]) than in patients without (2.1 ± 2.2 days; median, 1 [range 0 to 10]) (P = 0.08). Information regarding survival post-LT was obtained in 80 of the 85 patients undergoing LT. Importantly, 30-day survival post-LT was similar regardless of ICP monitoring in the global series (ICP group: 34 of 40 [85%] vs. non-ICP group: 34 of 40 [85%], P = 1.0), as well as when controlling for center (P = 0.63, Cochran-Mantel-Haenszel test for general association). The 30-day survival post-LT for specific etiologies was also similar (Figure 4).
Monitoring of ICP in patients with ALF provides two possible important benefits. First, it allows detection and management of an elevated ICP, as intracranial hypertension can be clinically silent.9 In addition, it allows improved decision-making for the emergency liver transplant candidate, as persistent reductions of cerebral perfusion pressure (mean arterial pressure − ICP) signal a high likelihood of ischemic brain injury and risks poor neurological recovery after the procedure.10 While low cerebral perfusion pressures have been associated with good outcomes,11 such results should be viewed as the exception,12 rather than the norm.
The single factor that precludes a wider use of ICP monitoring in ALF is concern with intracranial hemorrhage associated with its use. Indeed, the practices regarding ICP monitoring of the different centers that constitute the U.S. ALFSG differ widely, with some centers managing patients without ICP monitoring at all (Figure 1). Such an approach is not an exception, even among prominent European centers,3 More than a decade ago, a survey of the results of ICP monitoring in the United States showed an incidence of intracranial hemorrhage of 21%.4 In this report, detailed information on 58 cases from the U.S. ALFSF denotes complications in 6 subjects (10.3%), 3 of which were incidental findings at the time of computed tomography scanning. While the prevalence of hemorrhage appears to have decreased, the seriousness of the complication is highlighted by 2 cases in whom intracranial bleeding may have contributed to death. Fatalities were seen in patients in whom the dura mater was pierced with the monitor, a factor associated with increased rates of bleeding in these patients.4 Measures to correct the coagulopathy prior to insertion of the monitor now include the administration of recombinant factor VIIa, whose effects include a prompt correction of the prothrombin time.5 In 1 center's experience, no intracranial bleeding was seen after its use in 15 subjects.13 In our series, 2 bleeds occurred on recombinant factor VIIa. The exact role, dosing and safety of this compound in ALF await evaluation in a controlled trial, as its use can also be associated with an increased rate of thromboembolic events.14
To minimize the risk of hemorrhage, it was recommended that epidural transducers be used to monitor ICP.4 While the accuracy of such transducers is inferior to those where the dura is traversed, a lower incidence of both fatal and nonfatal hemorrhage provided a counterbalance to the technical inferiority of the device. At the present time, the manufacturer of epidural transducers has discontinued production of this monitor; in the current series, the use of epidural transducers was based on the adaptation of existing parenchymal monitors for use in the extradural space (the case of the Camino monitor, Camino Laboratories, San Diego, CA). Hemorrhage can still be seen with epidural placements, as seen in the past and in one of the cases of this series. Nonetheless, fatalities were not seen in the prior4 or current experiences, suggesting that the epidural approach is still the safest one.
A randomized controlled trial of ICP monitoring in ALF has never been performed, and the U.S. ALFSG could not reach a consensus on this matter. We chose to examine the outcome of patients who underwent emergency LT, as both monitored and nonmonitored groups had similar epidemiological and laboratory characteristics at the time of transplant. Of the 85 transplants of patients with ALF and stage III-IV encephalopathy, half were performed with ICP monitoring and half were done without this device. Survival at 30 days was similar in the patients with and without ICP monitoring, regardless of etiology (Figure 4). Such results are in concordance with previous reports of outcomes after ICP monitoring,9, 15 where improved management of intracranial hypertension was not synonymous with improved survival. In fact, other measures instituted for the treatment of ALF do not impact survival, such as the use of prophylactic antibiotics.16 Survival in this condition is mainly dependent on the recovery of liver function, as exemplified by the good results seen with emergency liver transplantation.
Several considerations should be brought forth regarding the applicability of the results of our study. First, all centers participating in this experience were tertiary centers with familiarity with LT; local factors such as center/surgeon knowledge of ICP monitoring as well as organ availability, however, may have influenced the decision of measuring ICP.
Second, our study has limitations related to the nonrandomized assignment of ICP monitoring. It is possible that in transplant candidates, monitoring was undertaken in sicker patients, though the epidemiological and clinical variables were similar prior to the operation. However, we note an important difference in the management of subjects in whom ICP monitors were placed. They received more treatments for elevated ICP values, including mannitol and barbiturates, than those in whom monitoring was not undertaken. The use of vasopressors was also more common, probably as a result of clinical information on cerebral perfusion pressure. We cannot ascertain whether such therapies translated into better neurological outcomes after recovery from ALF, but neurological deficits have been reported in survivors of LT17 and other conditions associated with intracranial hypertension.18 The U.S. ALFSG is currently conducting a long-term outcome study of patients with ALF, evaluating neurological function 1 and 2 years after the transplant procedure or in spontaneous survivors. With an improved operative survival, improvement in outcomes in ALF has new goals, including a reduction in the 90-day mortality (much higher than in other conditions where LT is performed19 and which reflects the impact of multiorgan failure at the time of the procedure) as well as long-term neurological function.
Finally, the role of ICP monitoring in the patients who did not undergo LT could not be explored in our study. This subpopulation is a heterogeneous group that includes less sick patients, patients very ill who can not be considered for LT, and patients not listed due to other medical/social reasons. The role of ICP monitoring in this population, therefore, remains unclear.
In summary, our results note a decrease in the rate of intracranial hemorrhage with ICP monitoring, but in spite of newer tools to overcome the severe coagulopathy of ALF, the risks of the procedure are still considerable. Most patients receiving ICP monitoring in the experience of the U.S. ALFSF are listed for LT, with only a minority of centers using it for nontransplant candidates. Nonetheless, half of transplants were performed without such monitors, with similar 30-day outcomes. As patients with ICP monitoring receive more treatments for intracranial hypertension, the possibility arises that greater neurological deficits may be seen in the subjects not receiving ICP monitoring. Studies of long-term outcome after ALF, currently under way, should provide answers to this important question.
The authors thank Jie Huang, PhD, for assistance in the statistical analysis.