Cerebral hemodynamic and metabolic profiles in fulminant hepatic failure: Relationship to outcome


  • Shushma Aggarwal,

    Corresponding author
    1. Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh Medical Center, Pittsburgh, PA
    • Department of Anesthesiology, University of Pittsburgh Medical Center, Room C220, 230 Lothrop Street, Pittsburgh, PA 15213

    Search for more papers by this author
    • Telephone: 412-647-6465; FAX: 412-647-6290

  • Walter Obrist,

    1. Department of Neurosurgery, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • Howard Yonas,

    1. Department of Neurosurgery, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • David Kramer,

    1. Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh Medical Center, Pittsburgh, PA
    Current affiliation:
    1. Mayo Clinic Transplant Center, Jacksonville, FL
    Search for more papers by this author
  • Yoogoo Kang,

    1. Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh Medical Center, Pittsburgh, PA
    2. Thomas Jefferson University, Jefferson Medical College, Philadelphia, PA
    Search for more papers by this author
  • Victor Scott,

    1. Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh Medical Center, Pittsburgh, PA
    2. Med-Pro, Phoenix, AZ
    Search for more papers by this author
  • Raymond Planinsic

    1. Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author


The purpose of this retrospective study was to examine the potential role of cerebral hemodynamic and metabolic factors in the outcome of patients with fulminant hepatic failure (FHF). Based on the literature, a hypothetical model was proposed in which physiologic changes progress sequentially in five phases, as defined by intracranial pressure (ICP) and cerebral blood flow (CBF) measurements. Seventy-six cerebral physiologic profiles were obtained in 26 patients (2 to 5 studies each) within 6 days of FHF diagnosis. ICP was continuously measured by an extradural fiber optic monitor. Global CBF estimates were obtained by xenon clearance techniques. Jugular venous and peripheral artery catheters permitted calculation of cerebral arteriovenous oxygen differences (AVDO2), from which cerebral metabolic rate for oxygen (CMRO2) was derived. A depressed CMRO2 was found in all patients. There was no evidence of cerebral ischemia as indicated by elevated AVDO2s. Instead, over 65% of the patients revealed cerebral hyperemia. Eight of the 26 patients underwent orthotopic liver transplantation—all recovered neurologically, including 6 with elevated ICPs. Of the 18 patients receiving medical treatment only, all 7 with increased ICP died in contrast to 9 survivors whose ICP remained normal (P < 0.004). Hyperemia, per se, was not related to outcome, although it occurred more frequently at the time of ICP elevations. Six patients were studied during brain death. All 6 revealed malignant intracranial hypertension, preceded by hyperemia. In conclusion, the above findings are consistent with the hypothetical model proposed. Prospective longitudinal studies are recommended to determine the precise evolution of the pathophysiologic changes. (Liver Transpl 2005;11:1353–1360.)

Cerebral edema and herniation are common causes of mortality in patients with fulminant hepatic failure (FHF).1, 2 It is estimated that 2,300 to 2,800 cases of FHF occur per year in the United States.3 Most of them are young, and timely treatment could delay or reverse their deterioration.

Under normal physiologic conditions, cerebral blood flow (CBF) is tightly coupled with the cerebral metabolic rate for oxygen (CMRO2), such that the oxygen demands of the brain are met.4, 5 Hence, the extraction of oxygen across the cerebral vascular bed remains relatively constant. However, CBF and oxidative metabolism can become uncoupled under pathophysiologic conditions. During ischemia, CBF fails to meet the oxidative metabolic demands,6 whereas in hyperemia (luxury perfusion), CBF exceeds the brain's metabolic requirement.7

Changes in cerebral hemodynamics and metabolism have been widely reported in patients with FHF.8 The findings suggest that hemodynamic and metabolic changes occur in progressive phases which, if not reversed, could result in brain death. In the earliest phase, CBF is low and coupled with the low metabolic demands of the brain.9 As the disease progresses, CBF becomes excessive (uncoupled) in relation to the metabolic demand,10 pronounced cerebral vasodilation occurs,11, 12 cerebrovascular autoregulation is impaired,13 intracranial pressure (ICP) increases,10, 14–16 systemic blood pressure decreases,17 and cerebral swelling develops.18–21

There is a clear need to characterize these physiologic changes and their progression in order to facilitate clinical management of FHF patients. A delay or reduction in cerebral swelling could potentially provide time for medical treatment to restore liver function or for the diseased liver to be replaced.

Based on the previous literature it is hypothesized that cerebral hemodynamic and metabolic changes undergo 5 sequential phases, as shown in Fig. 1: phase 1, low CBF and normal ICP; phase 2, high CBF and normal ICP; phase 3, high CBF and high ICP; phase 4, low CBF and high ICP; and phase 5, brain death. This model assumes that the CBF changes precede and contribute to an elevated ICP and brain swelling.

Figure 1.

Definition of the 5 phases according to the level of ICP and PaCO2-adjusted CBF (aCBF). High and low cutoff values are shown, as defined in the text.

In accordance with this model, a retrospective analysis was performed on 26 comatose patients with FHF in whom multiple measurements of both ICP and CBF were obtained. The aim of the study was to determine the usefulness of the above classification in predicting outcome. Particular attention was paid to the relationship of CBF to intracranial hypertension and brain swelling.


FHF, fulminant hepatic failure; CBF, cerebral blood flow; CMRO2, cerebral metabolic rate for oxygen; ICP, intracranial pressure; AVDO2, cerebral arteriovenous oxygen difference; CHMP, cerebral hemodynamic and metabolic profile; PaCO2, arterial carbon dioxide tension; aCBF, adjusted CBF; aAVDO2, adjusted AVDO2; MAP, mean arterial pressure; OLTx, orthotopic liver transplantation; CT, computed tomography.

Patients and Methods

This study was approved by the Institutional Review Board of the University of Pittsburgh Medical Center. Medical records were reviewed of 72 consecutively admitted patients with a diagnosis of FHF, there being no overlap with previously published studies.10, 22 FHF was diagnosed as a sudden, severe impairment of liver function, characterized by progressive jaundice, coagulopathy, and encephalopathy in previously healthy persons with no known underlying liver disease.23, 24

Because depth of coma is an important index of the severity of hepatic failure,23 only patients with grade 3 or 4 encephalopathy were selected for study; all were intubated and mechanically ventilated. After excluding patients who did not receive multiple CBF studies and ICP monitoring, a sample of 26 patients was obtained, 8 men and 18 women, with a mean age of 38 ± 9 years (range, 14–64 years). The etiology of their FHF upon admission was paracetamol overdose in 10, viral hepatitis in 9, postpartum hepatitis in 1, halothane hepatitis in 1, and undetermined in 5 (Table 1).

Table 1. Demographic Data on 26 Patients With FHF
Med TxMed TxOLTx
n = 9n = 9n = 8
  • *

    Death during hospitalization.

  • Living at the time of hospital discharge.

Age (yrs)38 ± 738 ± 639 ± 5
 Paracetamol overdose343
 Viral hepatitis342
 Postpartum hepatitis 1 
 Halothane hepatitis1  
 Unknown etiology2 3

All patients were admitted to the intensive care unit, where they underwent the following procedures in accordance with standard practice at this institution. A medical history and physical examination were obtained upon admission. Monitored variables included electrocardiogram, blood pressure via an indwelling intraarterial catheter, hemodynamic profile by a pulmonary artery catheter, urine output, arterial blood gas tensions and acid-base balance, electrolytes, liver function tests (serum levels of alanine aminotransferase, aspartate aminotransferase, bilirubin, alkaline phosphatase, ammonia, and albumin), coagulation tests (prothrombin time, activated partial thromboplastin time, and platelet count), and blood urea nitrogen and creatinine. In addition, a copper level, toxicology screen, and liver biopsy were obtained. Particular attention was paid to prothrombin time, serum creatinine, and arterial pH, since these were considered prognostic indicators.25

Upon admission, all patients revealed some degree of hepatic encephalopathy: grade 1, a confused state with altered mood and behavior (no patients); grade 2, sleepiness with slow arousability (5 patients); grade 3, coma responsive only to painful stimuli (17 patients); and grade 4, coma unresponsive to painful stimuli (4 patients). The 5 patients with admission grade 2 deteriorated rapidly to grade 3 or 4, and therefore were entered into the study.

Computed tomography (CT) scans of the brain were obtained to identify any focal lesions or global abnormalities such as atrophy or swelling. A neuroradiologist rated the presence of cerebral swelling as none, mild, moderate, or severe. The first CT scan was performed within 24 hours after admission to the intensive care unit, and additional scans were performed only when clinically indicated and if the patient's condition was stable enough for transport to the CT scanner.

Cerebral Hemodynamic and Metabolic Profiles

Six cerebral hemodynamic and metabolic variables were examined: ICP, CBF, CMRO2, arteriojugular venous oxygen content difference (AVDO2), cerebral vascular resistance, and cerebral perfusion pressure. Collectively, these were termed the cerebral hemodynamic and metabolic profile (CHMP). The first profile was obtained within 24 hours of admission to the intensive care unit.

CBF Measurements

The patients underwent 2 to 5 CBF studies. In every subject, the first CBF study was performed using the stable xenon-enhanced CT method (Xe-CT) (XeScan, Diversified Diagnostic Products, Houston, TX).26 All subsequent studies were performed at bedside by the intravenous Xe-133 clearance technique.27 For the Xe-CT method, CBF was measured in 10 regions of interest at 2 scanning levels, as described previously.22 The flow values in these regions were averaged to generate a mean global CBF estimate for each study.

For the Xe-133 method, brain clearance of intravenously injected Xe-133 (35 mCi in 5 mL of saline) was monitored by 10 extracranial detectors, 5 over each hemisphere. The clearance curves were analyzed by a 2-compartment model, which yielded CBF-15, an estimate of mean blood flow for the gray and white matter compartments.27 Global CBF was determined by averaging the values of the 10 monitored regions. Blood flow estimates obtained by the 2 techniques (Xe-CT and Xe-133) are comparable and highly correlated.28, 29

Because of the wide variation in arterial carbon dioxide tension (PaCO2), comparisons of CBF within and between patients were based on CBF values adjusted to the mean PaCO2 of the sample (28.4 ± 5.9 mm Hg). Based on the literature,30, 31 a correction factor of 3% per millimeter Hg PaCO2 was employed, which assumes normal CO2 reactivity—a condition generally found in FHF patients.12, 16, 22 The adjusted CBF (aCBF) = CBF / (1 − 0.03 × [28.4 – PaCO2]), expressed in mL/100 g/min.

AVDO2 and CMRO2 Determinations

AVDO2 was determined in all patients at the time of each CBF measurement. A jugular venous catheter was inserted into the right internal jugular vein and advanced cephalad to the jugular bulb, its correct position being verified by a lateral skull roentgenogram. Blood samples were drawn simultaneously from the jugular venous and intraarterial catheters at the time of each CBF study. Like CBF, AVDO2 varies with arterial CO2 tension, but in a reciprocal manner. AVDO2 was therefore adjusted to the average PaCO2 (28.4 mm Hg) using the same correction factor applied to CBF. The adjusted AVDO2 (aAVDO2) = AVDO2 × (1 − 0.03 × [28.4 – PaCO2]), expressed in vol%. CMRO2 was calculated as the product of the global CBF and AVDO2 (CMRO2 = CBF × AVDO2 / 100, expressed in mL/100 g/min).

ICP and Mean Arterial Pressure (MAP)

ICP was measured continuously in every patient with an epidural fiber optic monitor (Ladd Research Laboratories, Burlington, VT), which was inserted in the operating room, as previously described.10 MAP was continuously monitored from a peripheral artery. The difference between MAP and ICP yielded cerebral perfusion pressure (CPP = MAP – ICP). Cerebral vascular resistance (CVR) was calculated as the ratio of perfusion pressure and blood flow (CVR = CPP / CBF).

Patient Management

The management of patients with FHF involved attempts to maintain ICP below 25 mm Hg, AVDO2 between 4.5 and 8 vol%, and MAP above 80 mm Hg. If present, intracranial hypertension was treated sequentially with hyperventilation, diuretics (furosemide or mannitol), hypothermia, and barbiturates, each modality being added only if the previous measure was ineffective. Deviations in AVDO2 were corrected by adjustment of the patient's ventilation (increased for low and decreased for high AVDO2). Maintaining MAP above 80 mm Hg was usually a challenge, despite adequate replacement of fluid volume. Vasopressors (dopamine and/or epinephrine) were required in most cases. The vasopressor used depended on the severity of hypotension.

Antibiotic and antifungal prophylaxis was routinely used, guided by serial cultures. Electroencephalograms were recorded in patients suspected of having a seizure disorder, and appropriate anticonvulsant medication was administered. Intermittent or continuous hemodialysis was performed as indicated.

All patients were considered potential candidates for orthotopic liver transplantation (OLTx). The principal criteria used were progression of encephalopathy, liver biopsy showing necrosis in more than 30% of hepatocytes, an increasing normalized prothrombin ratio,24 hemodynamic stability, and dependence on renal dialysis. The decision to transplant was made jointly by Surgery, Critical Care Medicine, and Anesthesiology.

Definition of FHF Phases

Figure 1 presents the criteria used to classify the 5 FHF phases. As shown, normal or “low” ICP was defined as ≤25 mm Hg, and elevated or “high” ICP as <25 mm Hg.14 The criteria for CBF were derived from 42 previously studied normal control subjects.31 A “low” CBF was defined as 2 standard deviations below the normal mean, and a “high” CBF as within (± 2 standard deviations) or above the normal range. In the present study, adjustment to the mean PaCO2 (28.4 mm Hg) yielded cutoff values of <30 (low aCBF) and ≥30 mL/100 g/min (high aCBF).

Definition of Hyperemia

In coma, low CBFs are usually coupled to a depressed CMRO2.31 The finding of a normal aCBF was therefore considered “relative” hyperemia (30–44 mL/100 g/min); i.e., in excess of metabolic demand. aCBF values that exceeded the normal mean by 2 standard deviations were considered “absolute” hyperemia (>44 mL/100 g/min).

Because AVDO2 represents the balance between CBF and metabolism (AVDO2 = CMRO2 / CBF),31 it provides an independent estimate of hyperemia. Thus, values less than 2 standard deviations below the normal mean32 would represent hyperemia. For PaCO2-adjusted values (aAVDO2), relative hyperemia was defined as 4.0–5.0 vol%, and absolute hyperemia as <4.0 vol%.

It should be noted that adjustment of the observed CBF and AVDO2 values to the mean PaCO2 of the sample does not affect the definition of hyperemia, since the norms for these variables were adjusted to the same degree. Also, the adjusted and observed CBF and AVDO2 values yield identical CMRO2s, since the same adjustment was applied reciprocally to each.

Statistical Analysis

Except for a Pearson product-moment correlation between CBF and AVDO2, nonparametric statistics were employed throughout. The frequency of occurrence (number of patients) was entered into 2 × 2 tables, where Fisher's Exact Probability Test was applied. This yielded 1-tailed P values. Because of the small number of patients receiving OLTx, the relationship between phase and outcome was examined in the medically treated group only. P < 0.05 was considered statistically significant.


Of the 26 patients (Table 1), 9 died without undergoing OLTx, either because they were hemodynamically unstable, prohibiting transplantation, or because they died while waiting for a donor liver to become available. These patients received medical treatment only. Nine patients did not require OLTx and recovered neurologically after medical treatment. Eight patients underwent OLTx and all recovered neurologically.

The 26 patients had a total of 76 CHMP, averaging approximately 3 CHMP per patient. As shown in Table 2, phase 1 had a total of 29 CHMPs; phase 2, 24; phase 3, 10; phase 4, 7; and phase 5, 6. The number of profiles differed among phases because some patients remained in the same phase throughout the course of their illness while others changed phases. On the first examination, 10 patients were observed to be in phase 1, 8 in phase 2, and 8 in phase 3. Seventy-five of the 76 profiles were obtained within 6 days of establishing an FHF diagnosis (1 study was performed on the eighth day).

Table 2. Cerebral Hemodynamic and Metabolic Variables During Different Phases of FHF in 26 Patients
PhaseNo. of PatientsCHMP (n)ICP (mm Hg)aCBF (mL/100 g/min)aAVDO2 (vol%)CMRO2 (mL/100 g/min)CVR (mm Hg/mL/100 g/min)MAP (mm Hg)CPP (mm Hg)
  1. NOTE. Values are mean ± standard deviation, with medians in parentheses.

1162916 ± 421 ± 35.9 ± 1.21.22 ± 0.23.08 ± 0.588 ± 972 ± 9
2142418 ± 441 ± 83.4 ± 1.31.32 ± 0.41.80 ± 0.487 ± 968 ± 12
391038 ± 845 ± 93.2 ± 1.11.36 ± 0.41.53 ± 0.597 ± 1159 ± 16
45736 ± 820 ± 35.3 ± 1.41.06 ± 0.32.32 ± 0.682 ± 1046 ± 13
56684 ± 177 ± 12.0 ± 1.00.13 ± 0.5−1.578 ± 17−6

Means and standard deviations for the individual components of CHMP are also shown in Table 2. Median values are presented in parenthesis. In the normal ICP phases (1 and 2), ICP averaged 16 ± 4 and 18 ± 4 mm Hg, respectively; in the high ICP phases (3, 4, and 5), the averages were 38 ± 8, 36 ± 8, and 84 ± 17 mm Hg, respectively. Patients in the later 3 phases were intensively treated for intracranial hypertension, as described above. In the low aCBF phases (1 and 4), average aCBF values were 21 ± 3 and 20 ± 3 mL/100 g/min, respectively; in the high aCBF phases (3 and 4), they averaged 41 ± 8 and 45 ± 9 mL/100 g/min, respectively.

aAVDO2 was lower in the high aCBF phases (2 and 3), where it averaged 3.4 ± 1.3 and 3.2 ± 1.1 vol%, respectively; in the low aCBF phases (1 and 2), aAVDO2 averaged 5.9 ± 1.2 and 5.3 ± 1.4 vol%, respectively (Table 2). As expected, these findings reveal an inverse relationship between CBF and AVDO2, the lower aAVDO2s being associated with higher aCBFs, indicative of hyperemia. A product-moment correlation of −0.70 was obtained between aCBF and aAVDO2 (P < 0.001). CMRO2 was less than 50% of normal in all 5 phases (normal = 3.3 ± 0.4 mL/100 g/min).32

Cerebral vascular resistance was also lower in the high aCBF phases (2 and 3), which is consistent with cerebral vasodilatation. MAP was essentially unchanged in all 5 phases. However, maintaining MAP greater than 80 mm Hg was difficult and required vasopressors (dopamine and epinephrine infusion) in phases 3, 4, and 5. Cerebral perfusion pressure was lower than 80 mm Hg in all phases. It progressively decreased from phase 1 to 4 (72 ± 9, 68 ± 12, 59 ± 16, and 46 ± 13 mm Hg, respectively), and was absent in phase 5.

Relationship of Treatment Modality and Phase to Outcome

Of the 18 patients who received medical treatment only (Table 3A), 11 never progressed beyond phases 1 or 2. Nine of these 11 patients survived. In contrast, all 7 of the patients who reached phases 3 or 4 at some time during their illness died. This relationship between phase and outcome was statistically significant (P < 0.002, 1-tailed).

Table 3A. Outcome Findings Related to Treatment Modality (Med Tx vs. OLTx) in Four Phases of FHF
PhasesMed Tx*OLTx
n = 9n = 9n = 18n = 8n = 0n = 8
  • *

    For the medically treated group (Med Tx), outcome was significantly related to phase (P < 0.002, Fisher exact probability test).


The 8 patients who received OLTx survived. It is noteworthy that 6 of the 8 surviving OLTx patients had reached phases 3 or 4, which was associated with death in all 7 of the medically treated group who reached these phases (P < 0.001, 1-tailed).

Relationship of Changes in Phase to Outcome

During their stay in the intensive care unit, 8 patients in the medically treated group moved from lower to higher phases (Table 3B). Seven of these 8 patients died. In contrast, 10 patients showed either a decrease or no change in phase. Eight of these patients survived. This relationship between phase changes and outcome was statistically significant (P < 0.008, 1-tailed).

Table 3B. Outcome Findings Related to Changes in Four Phases of FHF (Med Tx only)*
Change in PhaseLivedDiedTotal
n = 9n = 9n = 18
  • *

    P < 0.008, Fisher exact probability test.

Decrease or no change8210

Brain Death (Phase 5)

Of the 9 patients that died, 6 had CHMP studies at the time of brain death. Five of these 6 patients had CT scans just preceding or during brain death, and all showed moderate to severe brain swelling. Five of the 6 patients also developed absolute hyperemia prior to brain death (aCBF > 44 mL/100 g/min, aAVDO2 < 4.0 vol%); the sixth case revealed a relative hyperemia.

Prevalence of Ischemia

Because of the consistently low CMRO2 observed, ischemia was defined by AVDO2; i.e., CBF relative to cerebral metabolic rate. For the PaCO2-adjusted A-V difference (aAVDO2), none of the patients exceeded the upper limit of normal (10 vol%), indicating an absence of ischemia. The highest unadjusted value of 6.9 vol% at a PaCO2 of 40.0 mm Hg was within the normal range.32 It should be noted that 8 patients with very low CBFs (<20 mL/100 g/min) all had low or normal, but not elevated, aAVDO2s.

Prevalence of Hyperemia

Hyperemia as defined by aCBF was observed in 65.4% of the patients (17 of 26). Of the 17 patients, 6 had relative (30–40 mL/100 g/min) and 11 absolute hyperemia (>44 mL/100 g/min); 34.6% of these patients never developed hyperemia. Hyperemia as defined by aAVDO2 was observed in 80.8% of the patients (21 of 26). Of the 21 patients, 7 had relative (4.0–5.0 vol%) and 14 absolute hyperemia (<4.0 vol%); 19.2% of these patients never developed hyperemia.

Relationship of Hyperemia to Outcome and ICP

As presented above, the finding of hyperemia in this sample was as an ubiquitous sign. The occurrence of hyperemia per se (at any time during the illness) was not significantly related to outcome or to the development of intracranial hypertension. This was true for both absolute and relative hyperemia, defined either by aCBF or aAVDO2.

Hyperemia at the Time of ICP Elevations

Table 4A presents concurrent aCBF and ICP findings in patients with and without ICP elevations (>25 mm Hg). For patients with increased ICP, the study with the highest pressure was chosen. For patients whose ICP never exceeded 25 mm Hg, their last study was chosen for comparison. As the table shows, there was a tendency (P < 0.095, 1-tailed) for elevated ICPs to be associated with hyperemia (aCBF ≥ 30 mL/100 g/min) at the time of the study.

Table 4. Concurrent ICP and aCBF* and aAVDO2 Findings in FHF Patients
n = 13n = 13n = 26
  • *

    P < 0.095, Fisher exact probability test.

  • P < 0.042, Fisher exact probability test.

  • aCBF ≥ 30 indicates hyperemia.

  • §

    aAVDO2 ≤ 5.0 indicates hyperemia.


Table 4B presents a similar analysis for aAVDO2 and ICP. In this case, a significant relationship (P < 0.042, 1-tailed) was found, such that elevated ICPs were associated with hyperemia (aAVDO2 <5.0 vol%) at the time of the study.


The present study is based on a retrospective analysis of 26 FHF patients in which 5 phases of cerebral hemodynamic and metabolic changes were postulated. The relationship of these phases to outcome was examined, with emphasis on the concordance of CBF changes and intracranial hypertension. Although the results are generally consistent with the hypothesis of sequential phases, the limitations of the study preclude definitive conclusions.

Limitations of the Study

The small number of cases did not permit an adequate (multivariate) statistical analysis. Although sequential observations were made in all patients, the variable number and irregular timing of the examinations provided only limited information on the time course of the disease process. Also, the effects of treatment were difficult to evaluate in the absence of regularly timed measurements. A possible sampling bias might have been introduced by restricting the sample to comatose patients with both ICP monitoring and multiple CBF examinations.

In order to evaluate the representativeness of the sample, a comparison was made between the 26 cases studied and the 46 who were excluded for failure to meet the inclusion criteria, which were: (1) grades 3–4 encephalopathy, (2) both ICP monitoring and CBF-AVDO2 measurements, and (3) multiple (2 or more) CBF studies. The reason for exclusion were: 21 patients with coma grades 1–2 only; 9 without paired ICP and CBF observations (8 did not have surgery for ICP monitor placement because of the calculated risk); 10 with insufficient time for multiple studies between admission and OLTx or death (4 <24 hours and 6 <48 hours); 4 who rapidly recovered consciousness (to coma grades 1–2); and 2 with serious nonhepatic disease (fungal pneumonia and human immunodeficiency virus–positive). Generalization can thus be made only to the population that met the inclusion criteria.

Summary of Physiologic Findings

In spite of these limitations, a number of interesting observations were made. Outcome (death vs. survival) was significantly associated with the presence or absence of intracranial hypertension (phases 3 and 4 vs. 1 and 2) in the medically treated group. Patients who showed an increase in phase were more likely to die. These findings clearly indicate the deleterious role of an elevated ICP. The fact that all 6 patients studied in brain death (phase 5) had malignant intracranial hypertension (ICP = 51–110 mm Hg) agrees with this concept.

The role of CBF changes was more difficult to assess. As reported previously,10 there was little or no evidence of cerebral ischemia (elevated AVDO2). Cerebral hyperemia, on the other hand, was found in 65 to 81% of the patients at some time during the course of their illness. AVDO2s were below the lower limit of normal, indicating an uncoupling of CBF and metabolism (luxury perfusion).

Hyperemia per se was, however, a poor predictor of outcome. This lack of relationship can probably be explained by the confounding effects of treatment. ICP therapies, which were frequently administered, are known to reduce CBF,10, 21, 31, 33, 34 thereby limiting the adverse effects of hyperemia. Of particular note is the occurrence of severe (absolute) hyperemia just preceding brain death in 5 of the 6 patients studied (the sixth patient had a relative hyperemia). In 5 of these cases with CT scans, the hyperemia was associated with brain swelling.

When the temporal relationship of hyperemia to intracranial hypertension was examined, a definite trend was observed. CBFs at the time of maximum ICP elevation were in the hyperemic range, in contrast to the lower values recorded in patients whose ICP remained normal. The relationship was stronger for hyperemia defined by AVDO2 (CBF relative to metabolism).

Liver Transplantation, ICP, and Outcome

The 8 patients who underwent OLTx survived. Six of these survivors had an elevated ICP (phases 3 and 4), which is in sharp contrast to those who received medical treatment only. All 7 of the latter patients who developed intracranial hypertension died. This suggests that liver transplantation can reverse the course of ICP elevation and associated brain swelling.


In accordance with the previous literature, the present study provides evidence that intracranial hypertension and associated brain swelling are primary factors in the death of FHF patients. A sequence of events was proposed in which cerebral hemodynamic and metabolic alterations contribute to the rise in ICP, similar to that observed in acute brain trauma.31, 35 Because of its limitations, however, the present study could not provide definitive evidence on the role of hemodynamic factors. Nevertheless, the findings are consistent with such a hypothesis, and indicate a need for further, more systematic investigation.

A fundamental question is whether cerebral hyperemia, found in 80% of the patients, precedes and contributes to the rise in ICP. As shown in Table 2, patients with hyperemia can have either a normal ICP (phase 2), or an elevated ICP (phase 3). Assuming a transition from phase 2 to 3, what is the mechanism responsible for this change? Clearly, there has been a compromise in brain compliance; i.e., an increase in brain stiffness.36 Aside from an elevated cerebral blood volume, hyperemia may enhance the development of vasogenic edema,21, 37 which would then increase ICP. Recent evidence38 suggests that inflammation (cytokines) may contribute to intracranial hypertension in FHF by inducing cerebral hyperemia.

A further question arises concerning the assumed transition between phases 3 and 4 (Table 2). Whereas both phases have an elevated ICP, the values for CBF are quite different: hyperemia in phase 3, and reduced flow in phase 4. As shown in the table, perfusion pressure drops to 46 mm Hg in phase 4. Given the likelihood of impaired cerebrovascular autoregulation,13, 39, 40 this could account for the reduction in CBF.

The above speculations on pathophysiology support the need for extended research aimed at elucidating the mechanisms responsible for decompensation of FHF patients. Although the present study does not provide specific insight into mechanisms, the results argue for a prospective investigation involving regularly timed observations of physiologic variables.