Protocol based invasive intracranial pressure monitoring in acute liver failure: feasibility, safety and impact on management

Acute liver failure (ALF), defined as hepatic encephalopathy and coagulopathy within 26 weeks of illness onset, is a rare but potentially life-threatening illness [1]. Patients with ALF may develop diffuse cerebral edema, thought to result from an elevated serum level of ammonia, which easily crosses the blood-brain barrier then reacts with glutamate to form glutamine within astrocytes [2]. The large quantity of intracellular glutamine is then thought to result in an osmotic shift of fluid into astrocytes and cerebral edema [2]. Elevated cerebral blood flow (hyperemia), possibly as a result of cerebral vasodilatation and impaired autoregulation, may also contribute to cerebral edema and intracranial hypertension [2]. Intracranial hypertension is an important cause of death in patients with ALF [1‐7]. Appropriate management of cerebral edema, in conjunction with emergency liver transplantation (LT) in select patients, may result in favorable clinical outcomes [1, 8‐20]. Since the bedside physical exam and radiographic imaging are poor predictors of elevated intracranial pressure (ICP) [12, 21], invasive ICP monitoring (IICPM) remains the gold standard for detection and management of intracranial hypertension [1, 2, 10‐14, 22, 23]. However, ALF patients, by definition are coagulopathic, and at risk for life-threatening intracranial hemorrhage following monitor insertion [1, 24, 25]. Several prior studies have reported a high incidence of intracranial hemorrhage with IICPM [1, 11, 16, 22]. Most patients in older studies of IICPM in ALF were, however, monitored with subdural/epidural monitors [11, 12, 16, 22, 25‐29], Some studies have included patients monitored with external ventricular drains (EVDs) [11]. The ALF Study Group (ALFSG) recommends against the use of EVDs because of the high risk for hemorrhage [1]. Epidural/subdural bolts are rarely used today because of concerns about technical accuracy and overestimation of ICP [29, 30]. Also in these earlier studies, recombinant Factor VIIa (rFVIIa), which is more effective than fresh frozen plasma (FFP) as a reversal agent [1, 14, 31, 32], was not consistently utilized prior to insertion. Intraparenchymal monitors cause less tissue disruption and are associated with a lower risk of hemorrhage compared to EVDs [10, 11, 14, 16, 23]. Some centers have described a relatively low risk of complications with the consistent use of intraparenchymal monitors following administration of rFVIIa [10, 14, 23].

There is a need for more data on the safety and clinical value of IICPM using contemporary devices and reversal agents. The objective of our study was to evaluate the feasibility, safety and impact on clinical management of protocol-driven use of IICPM in a consecutive series of patients with ALF managed at a large liver referral center.

In 2011 our institution put in place a protocol for the use of IICPM in ALF. Protocol components included written clinical criteria for invasive monitoring and standardizing the use of agents to reverse coagulopathy prior to insertion. In view of the potential risk associated with the intervention, we created a database to prospectively track complications associated with IICPM use, the frequency with which intracranial hypertension was detected, changes in management following detection of intracranial hypertension and clinical outcomes.

Approval was obtained from the University of Michigan Institutional Review Board (HUM00052556) in 2011 for a quality control database to monitor the complications associated with, and the utility of, invasive monitoring in patients with ALF. Subsequent approval was obtained for this study, to use the database to analyze the outcomes of patients with ALF who underwent IICPM between June 2011 and October 2016 (HUM00120468).

Thirty-seven consecutive admissions for ALF between July 2011 and October 2016 were evaluated for use of IICPM. Thirteen (35%) patients subsequently did not undergo IICPM. The most common reason for exclusion was an improved neurological examination finding of grade 2–3 encephalopathy when left off sedation, in seven patients (54%). No patient developed clinical decline or new pupillary asymmetry during the period of sedation interruption. Other reasons were the absence of all brainstem function in two patients (15%), rapidly worsening multi-organ failure leading to death within <24 hours in 2 patients (15%) and surrogate refusal of consent in 2 patients (15%). Therefore, a total of 24 patients (65%) underwent IICPM. The median number of days from symptom onset to ICP monitor placement was 5 (IQR 2–10). The distribution of baseline variables and outcomes in patients who did and did not undergo IICPM is depicted in Table 3. Tables 4 and 5 depict the clinical details and outcomes of the 24 patients who underwent IICPM, in survivors of the inpatient admission and deceased patients, respectively.

Our study demonstrates that protocol-based IICPM use in patients with ALF is both feasible and safe. Only one patient (4%) suffered hemorrhage, which was a 5-mm asymptomatic subdural hemorrhage in a patient who had a good outcome. Other notable findings are that over half of monitored patients developed at least one episode of sustained intracranial hypertension requiring TIL >1, and about one in five developed refractory intracranial hypertension requiring extreme measures for ICP control (TIL-4). Our study also demonstrates that excellent outcomes are possible, including inpatients not deemed eligible for LT. This is illustrated by the 15-year-old patient deemed ineligible for LT who required TIL-4 care (barbiturate coma and hypothermia) before demonstrating spontaneous liver recovery and eventual excellent outcome (GOS 5 at 6 months).

Table 6 lists the studies that have described the use of IICPM in ALF [10‐14, 16, 21‐23, 26‐29, 37]. Our study is distinct for several reasons. All data points were prospectively collected to monitor the safety and quality of a potentially risky procedure. While larger multi-center studies have reported the use of a mix of devices (including subdural/epidural catheters and EVDs) and reversal strategies [11, 16, 22], this is the largest reported series from a single center with protocol-driven use of intraparenchymal catheters and rFVIIa in all patients. Most important, unlike prior studies, every patient who underwent IICPM received follow-up imaging, thereby providing a reliable estimate of the rate of intracranial hemorrhage. The low rate of hemorrhage in our study, with no symptomatic bleeds, may, therefore, be particularly noteworthy.

We believe that protocol-driven reversal of coagulopathy is essential in this regard. Considering the relatively short 2-hour half-life of rFVIIa, our protocol placed a heavy emphasis on monitor placement within an hour of a dose sufficient to reliably achieve normalization of the INR [1, 31, 39‐41], without repeat laboratory testing which may delay placement. The use of rFVIIa may achieve normalization of the INR without reversal of all of the elements of the coagulopathy associated with ALF. Our protocol therefore included a requirement that FFP be transfused at the time of monitor insertion (Table 1) in all cases [1]. A recent meta-analysis did not demonstrate a reduction in transfusion requirements with the use of rFVIIa in the setting of hepatic surgery [42], however, and rFVIIa is not universally accepted as the standard of care for the reversal of coagulopathy prior to invasive procedures in the setting of ALF. [43] In particular, the efficacy of rFVIIa may be degraded by acidosis, with one study suggesting a loss of efficacy of over 90% with a drop in the pH from 7.4 to 7.0 [44]. It is therefore relevant that the median pH in patients who underwent IICPM, and therefore received rFVIIa, was 7.39 in our study - rFVIIa may be less effective, and hemorrhage risk possibly higher, in a patient population with a higher incidence of severe, uncorrected acidosis at the time of the invasive procedure. It is also important to note that the low incidence of procedural hemorrhage in our study may primarily reflect the other elements of coagulopathy reversal that were part of the protocol, such as those directed at platelet count and function (platelet transfusion to goal >50,000/cm³ plus use of DDAVP in all cases) and adequate fibrinogen levels (goal >100 mg/dL), rather than the use of rFVIIa. A comprehensive approach to the coagulopathy of ALF may be critical to mitigate the risk of procedural bleeding.

Our study is also unique in providing information on the therapeutic intensity level directed at ICP management, which specifically provides a measure of the impact of IICPM on clinical management. It is important to note that all escalation in therapeutic intensity above TIL-1 in our study occurred exclusively on the basis of the measured ICP, rather than the clinical examination, which was already very poor, or imaging, which was only repeated in the first post-procedure day to look for hemorrhage.

Our tracking of long-term outcomes confirms that patients may survive ALF with excellent outcome despite the occurrence of intracranial hypertension, but some may also be left with long-term cognitive dysfunction from potentially preventable secondary brain injury. Another interesting observation in our study is that, in contrast to earlier studies of ALF [3‐5, 45], and consistent with some more recent studies [1, 17, 46], intracranial hypertension was not the primary cause of death - 87% of monitored patients died of progressive multi-organ failure (following effective control of ICP) and only 9% died as a direct consequence of intracranial hypertension. It is possible that the use of IICPM may decrease the frequency of death from intracranial hypertension, while overall survival remains dependent on timely emergency LT or spontaneous recovery of liver function.

The major argument in favor of invasive monitoring is that IICPM frequently reveals the presence of otherwise unsuspected sustained intracranial hypertension requiring therapy. Patients in grade-4 encephalopathy are all deeply comatose with a GCS of 3–4 T, as seen in our study, providing limited opportunity for detection of raised ICP through further neurological decline until cerebral herniation and pupillary asymmetry occurs. In our study, CT findings such as sulcal effacement were poorly predictive of intracranial hypertension, a finding consistent with other studies in patients with ALF [12, 21]. It is also impractical to perform serial imaging with CT in patients with ALF, who frequently develop severe hemodynamic and respiratory instability. The use of IICPM permits both escalation of therapy in patients who do not respond to first line measures (and may progress to cerebral herniation/brain death without more aggressive measures), and withholding of treatment measures with potentially serious side effects in patients whose measured ICP is normal (46% of monitored patients in our study). This includes avoidance of the potential side effects of mannitol (hypovolemia, worsening renal function), hypertonic saline (acidosis, circulatory overload), deep sedation (hypotension, prolongation of mechanical ventilation and ICU length of stay), hypothermia (worsening coagulopathy, infection) and barbiturates (prolonged sedation, hypotension, ileus, pneumonia), among others. The impact of IICPM on clinical management was most evident in the 21% of monitored patients requiring extreme measures such as barbiturate coma or hypothermia (TIL-4), which successfully controlled ICP in 4/5 patients. Less invasive monitoring tools such as jugular venous oximetry and noninvasive tools such as optic nerve ultrasound and transcranial Doppler may have a future role in the assessment of ICP and cerebral perfusion in this population, given the risks and expense associated with IICPM [47‐55]. Their ability to accurately detect elevated ICP and direct appropriate changes in management is uncertain, however, and their role in the management of ALF is currently investigational. Our group is currently studying the accuracy of optic nerve ultrasound and transcranial Doppler compared to IICPM for the prediction of intracranial hypertension in ALF.

The primary limitations of our study are a small sample size and its descriptive nature. Our study cannot provide definitive evidence favoring the routine use of IICPM in patients with ALF. Since ALF is an uncommon condition, and IICPM is only performed in a subset of major liver centers, a randomized clinical trial focused on functional outcomes, with an adequate sample size, will likely be challenging to conduct. Clinical trials of monitoring tools are also inherently limited by the fact that outcomes are determined by the actions (or lack thereof) taken in response to the monitoring tool. In the BEST-TRIP clinical trial of IICPM following traumatic brain injury (TBI), statistically significant differences in therapies used to manage ICP may have played a significant role in determining outcomes, beyond simply the use of IICPM [56]. While the ALFSG has pooled information from multiple centers to provide valuable data from large patient samples, retrospective comparisons of outcomes in patients who do and do not receive IICPM in these studies may be limited by potential confounding from selection bias as well as the absence of a common protocol and the consequent heterogeneity in patient selection, devices, reversal strategies and therapies used to control ICP [11, 16, 22]. Illustrating the potential for selection bias, patients who underwent IICPM in our study had greater disease severity, with worse GCS (p = 0.002), as a consequence of the protocol eligibility criteria. While the 2-hour period of sedation interruption may have resulted in a transient increase in cerebral metabolic demand, several patients (7 of 37 screened) were in fact found to be at a better neurological grade than initially suspected, following sedation interruption, and, as a consequence, did not undergo monitor placement. None of these patients subsequently died as a result of intracranial hypertension. We therefore believe that sedation interruption was necessary, to restrict the risks of the intervention to as few patients as possible - those with the highest risk of death from elevated ICP.

Protocol-based use of invasive intracranial pressure monitoring following acute liver failure is feasible, is associated with a low incidence of serious complications and has a significant impact on management.