Introduction

Management of intracranial pressure (ICP) is at the core of neurocritical care. Monitoring and normalizing the ICP reduces secondary neurological injury and its associated morbidity and mortality. This review of the literature aims to do the following: introduce rudimentary principles in the anatomy and physiology of ICP management, discuss commonly encountered pathologies that lead to increased ICP, summarize diagnostic approaches being used in the evaluation of ICP, and present evidence-based treatment paradigms that drive clinical care in intensive care units today. Lastly, the authors present an example of an algorithm that is used within our system of hospitals for managing patients with changes in ICP.

Detection of Elevated Intracranial Pressure

Invasive ICP Monitoring

Invasive ICP monitoring with external ventricular drains (EVD) and intraparenchymal monitors (IPM) continue to be the gold standards for measurement. While invasive ICP monitoring is commonplace in the management of traumatic brain injury (TBI), the recent “Benchmark Evidence from South American Trials: Treatment of Intracranial Pressure” (BEST:TRIP) [1] trial caused some to re-evaluate its use. BEST:TRIP [1], a randomized control trial published in 2012, showed that aggressive ICP monitoring using IPMs in patients with severe TBI did not improve long-term outcomes compared to management based on clinical signs and neuroimaging [1]. The study setting of Bolivia and Ecuador in which pre-hospital resuscitation is not as robust as other countries is a point of contention against its generalizability [1, 2]. Post-ICU-stay care is also different from other countries, which likely played a part in the relatively high 14-day mortality rate of 35% [1]. Despite being a well-designed, robust trial, these limitations diminish its results in countries with well-established ICP monitoring protocols. Additionally, those in the ICP monitoring had lower therapeutic intensity, so ICP management may be a cost-effective and less labor-intensive means to manage patients. Neurocritical care in North America and Europe has not moved away from invasive ICP monitoring as a standard and the authors themselves have argued for continued monitoring of these patients.

EVDs are considered the gold standard and provide the clinician a therapeutic modality of cerebrospinal fluid (CSF) drainage if needed. Recently, Bales showed that the choice of ICP monitoring may not be frivolous, highlighting that severe TBI patients may benefit more from IPM rather than EVD placement [3]. This is contradictory to previous studies showing no benefit [4] or EVD superiority [5]. Regardless of which invasive ICP monitor is used, the complications of placement must be considered. Complications include ventriculostomy-associated infections, procedural hemorrhage, and monitor misplacement [6].

Non-invasive Ultrasonography of Optic Nerve Sheath

With the increasing use of ultrasound (US) in hospital settings, bedside point-of-care diagnostics such as optic nerve sheath diameter (ONSD) measurements have become more commonplace. The space between the optic nerve and its sheath is contiguous with the subarachnoid space, allowing for changes in ICP to be transmitted to the retrobulbar sheath quickly, causing its expansion [7, 8]. Recent meta-analyses have confirmed the utility of ONSD to detect elevated ICP in patients presenting with TBI, with the areas under the curve (AUC) of 0.981 [9] and 0.932 [10•]. Each meta-analysis included between 320 and 350 patients and most studies were single-center, prospective studies of TBI patients in emergency departments with known high rates of intracranial hypertension (13.6–83%). Note, some studies did have concerns for bias including improper blinding. Most studies defined hypertension as an ICP > 20 mmHg, although two defined it higher at > 25 mmHg. All studies measured the ONSD 3 mm behind the globe with 2–3 measurements taken in each eye; however, the threshold ONSD measurement for denoting elevated ICP ranged from 0.48 cm [7] to 0.63 cm [11]. Recent studies have tried to overcome variability by different means. Agrawal et. al. have shown that coronal axis insonation of the ONSD had less variability amongst measurements, however, was not able to replicate the ICP estimation efficacy of axial insonation [12]. Du et. al. built on ONSD by using eyeball transverse diameter (ETD) to form a ratio: ONSD/ETD, which was more predictive of increased ICP than ONSD alone [13]. Similarly, they also found that a predecessor study [14] had different ONSD/ETD ratio thresholds for optimal diagnosis of increased ICP (0.25 [13] vs. 0.29 [14]).

While the utility in TBI looks promising, ONSD had poor performance in predicting intracranial hypertension in the largest series to date of patients presenting with acute liver failure [15]. The AUC was only 0.59 (85% CI 0.37–0.79) for ONSD, however, ICP calculated from transcranial Doppler (TCD) flow velocities using the estimated cerebral perfusion pressure (CPP) technique showed good correlation and excellent negative predictive value [16]. The ability to perform non-invasive assessments in these patients is paramount given high rates of cerebral edema and intracranial hypertension with concurrent severe coagulopathy limiting safety of invasive ICP monitoring. In general, however, the accuracy of non-invasive ICP measurements using TCDs has been variable across studies, with none reaching the accuracy necessary to replace invasive monitoring [17].

Although ONSD-US has shortcomings, Rajajee suggested internal standardization to find an institutional-specific ONSD value correlating with ICP above 20 mmHg to overcome technique and operator variability [7]. After standardization, OSND-US in TBI can be very powerful in clinical practice due to its accessibility at bedside, speed of obtaining information, and accuracy in diagnosing intracranial hypertension.

While ONSD is most often studied with respect to its correlating with instantaneously measured ICP, it can also be used to predict which patients may have ICP elevations during their admission. In a group of severe TBI patients, Robba found that an initial admission ONSD measurement significantly correlated to the following: mean ICP while in the intensive care unit (ICU), time of ICP above 20 mmHg while in the ICU, and ICU mortality [18]. However, further confirmatory work needs to be done in order to employ ONSD as a prognosticating index.

Pupillometry

The diagnostic value of pupil size and reactivity have been discussed as qualitative neurological exam findings, however, a more quantitative approach through the use of a pupillometer also is effective in providing evidence of increased ICP. Chen et. al. introduced the Neurological Pupil index (NPi) as a measure that incorporated pupil size, latency, constriction velocity, and dilation velocity to successfully determine, and even predict by an average of about 16 h before, the presence of intracranial hypertension [19]. NPi was able to separate the study cohort into 3 groups: normal NPi (3–5) which had a mean peak ICP of 19.6 mmHg, abnormal NPi (< 3) with a peak mean of 30.5 mmHg, and unreactive NPi with peak mean of 33.8 mmHg. These differences in mean peak ICP were statistically significant when considering ICP and NPi trends although the study did not address the notion of single-point abnormal NPi values that were associated with normal ICP values [19]. Although the temporal relationship of isolated NPi values is still being investigated, recent studies have provided encouraging evidence showing NPi improvement as early as 2 h after administration of hyperosmolar therapy, highlighting pupillometry as a non-invasive method to validate response of hyperosmolar intervention [20, 21].

Electroencephalography

The advent of multimodality monitoring in neurocritical care has put forth avenues of utilizing continuous electroencephalography (cEEG) to analyze patient data. Newey et al. demonstrated a case in which the presence of cerebral edema had been tipped off by cEEG asymmetries that evolved into burst suppression before any clinical signs manifested in the patient [22]. Also, pathologies that increase ICP can also cause seizures, which in itself can lead to cerebral hypermetabolism, in turn, increasing cerebral blood flow (CBF) and increasing ICP. Therefore, proactively identifying and halting epileptic activity should be a goal with patients that have intracranial pathologies with associated tenuous ICP. Isolated cEEG recordings may not be sensitive in determining increases in ICP until it is significantly elevated to the point of significantly limiting CBF [23], however, utilizing quantitative cEEG parameters may be beneficial to detect smaller changes in CBF, and thus, subclinical increases in ICP. Hartings et al. was able to show that timing of cortical spreading depression (CSD) was loosely tied to blood pressure instability and increased ICP; however, the study was unable to state whether CSDs were causing the increased ICP or if they were resulting from increased ICP [24]. Today, cEEG is not strictly used as a non-invasive ICP monitor due to its lack of sensitivity, however, it can provide valuable clues to signal increased ICP when used in the right context.

Treatment of Elevated ICP

Hyperosmolar and Pharmacological Therapies

When faced with focal edema from intracranial pathology, hyperosmolar therapy (HOT) is used principally to draw interstitial fluid intravascularly for clearance, decreasing cerebral edema, and intracranial ICP. HOT has shown benefit in TBI-related edema, stroke-related edema, and perihematomal edema in intracranial hemorrhage (ICH) [25,26,27]. Mainstays of hyperosmolar therapy are mannitol and hypertonic saline (HTS). Agent choice depends on patient comorbidities and characteristics, e.g., volume status and serum tonicity [27, 28], with an aim to minimize the consequent side effects of HOT. Gu et al. published a meta-analysis of 12 randomized controlled trials (RCT) comparing HTS and mannitol. Overall, the meta-analysis did not show superiority of either HTS or mannitol in magnitude of ICP reduction (p = 0.149; 11 of 12 studies); however, control of ICP reduction, or time of treatment effect, was significantly (p = 0.044; 8 of 12 studies) longer with HTS than mannitol. There was no significant difference between mortality (relative risk of 0.78, p = 0.216; 6 of 12 studies) or neurological function using the Glasgow Outcome Scale (relative risk of 1.17, p = 0.258; 5 of 12 studies). The included studies were heterogeneous in terms of sample size, sample criteria (e.g., TBI and non-TBI neurosurgical patients), study design (e.g., crossover-control trials), and definitional thresholds. The authors tried to limit this by selectively comparing proportions of the 12 RCTs based on the studied variable and the provided data, however, there were still noted differences [29]. Mangat et al. used a small sample of 25 case-control matched pairs of HTS and mannitol (note 1 patient from HTS cohort received 23.4% HTS boluses) to show that HTS had a statistically significant (p = 0.0003) shorter period of cumulative intracranial hypertension (15.5% of monitoring period) when compared to mannitol (36.5% of monitoring period) [30]. A subsequent 2019 study using the same matched patient data further showed HTS usage led to significantly less days with low CPP compared to mannitol with a cumulative burden trending towards HTS superiority. The study also looked at cumulative burden of high ICP/low CPP concomitance; this was found to be significantly less in the HTS group than the mannitol group [31••].

Intravenous (IV) glyburide is being investigated as a method of preventing post-large hemispheric stroke edema by inhibiting SUR1 receptors, whose upregulation and activation in ischemia leads to microvascular instability and subsequent edema [32, 33]. The “Glyburide Advantage in Malignant Edema and Stroke” (GAMES-RP) trial [34] showed that intravenous glyburide administration within 10 h of symptoms onset in anterior circulation strokes with large infarct volumes (lower limit of 82 cm3) that were not considered to be futile (upper limit of 300 cm3) showed benefit in reducing the rate of cerebral edema-related deaths [34, 35••]. IV Glyburide administration did not decrease the rate of malignant edema between placebo and intervention cohorts; however, in the subgroup of patients that developed malignant edema, glyburide administration decreased the rate of midline shift and degree of worsening seen on the NIH stroke scale. However, IV glyburide was associated with a higher rate of decompressive craniectomies although the difference amongst the arms was not significant [35••]. Post hoc subgroup analysis of GAMES-RP participants that were ≤ 70 years old showed a significant functional benefit on the Barthel Index and a non-significant benefit on the modified Rankin scale [36].

Corticosteroids are mainstays in treating mass effect and increased ICP from intracranial neoplastic tumors [37]. Tumors increase intracranial tissue and also evoke inflammatory cascades to produce peritumoral edema, increasing interstitial fluid. Steroids limit the interstitial vasogenic edema and decrease mass effect until definitive surgical intervention, if possible, is performed [38,39,40]. Corticosteroids were also tested in acute head injury in the “Corticosteroids Randomisation after Significant Head injury” (MRC-CRASH) trial, in which intravenous methylprednisolone was given within 8 h of head injury in patients with resulting GCS < 14. The results at 2 weeks showed that mortality was greater in the corticosteroid cohort compared to the placebo cohort [41]. At 6 months, the corticosteroid group again had greater mortality compared to placebo [42]. Given these results, steroids are not recommended in the setting of traumatic brain injury.

Hypothermia for Intracranial Hypertension After Traumatic Brain Injury

Hypothermia aims to reduce cerebral metabolism and CBF, subsequently reducing ICP. While prophylactic hypothermia after TBI has been extensively studied in the “National Acute Brain Injury Study: Hypothermia” (NABISH) [43] trials and the recently published “Prophylactic Hypothermia Trial to Lessen Traumatic Brain Injury” (POLAR) RCT with little success, the specific use of “late-rescue” hypothermia to treat intracranial hypertension after TBI has garnered less attention [44••]. EuroTherm3235, an open-label RCT compared use versus non-use of hypothermia prior to administration of hypertonic solutions as part of a staged treatment of intracranial hypertension after TBI [45]. All subjects were closed-TBI patients with ICP of >20 mmHg for at least 5 min that was refractory to stage 1 treatments (defined as mechanical ventilation, sedation, head of bed to 30°, analgesia, CPP optimization, +/− ventriculostomy, or +/− surgical removal of space-occupying lesions). Those randomized to the hypothermia group received therapeutic temperature modulation (TTM) with initial target of 35 °C but lowered incrementally down to as low as 32 °C until ICP was below 20 mmHg; hypertonic solutions were then administered if hypothermia was ineffective. Patients in the control group received hypertonic solutions once stage-1 therapies had failed to bring the ICP to ≤ 20 mmHg. Once hypothermia was started, it was maintained for at least 48 h and then re-warmed using a standardized protocol if ICPs were under control. The primary outcome was scored using the Extended Glasgow Outcome Scale (GOS-E) at 6 months.

The trial was designed to enroll 600 patients but was stopped early after enrolling and randomizing 387 subjects due to worse outcomes in the hypothermia group. Note, the groups were well matched in baseline characteristics. The common odds ratio for the GOS-E score was 1.53 (95% CI 1.02–2.3), indicating a worse outcome in the hypothermia group. A favorable GOS-E (5–8) was seen in 26% of the hypothermia group and 37% in the control group. Notably, patients in the hypothermia group received less stage-3 treatments (stage-3 treatments included barbiturate therapy, decompressive craniectomy, or further surgical intervention), mainly barbiturates, than those in the control group (44% vs 54%). Measured ICPs, however, were similar between groups throughout the study.

This trial gave fairly convincing evidence that application of hypothermia to 32-35 °C likely has more risks than benefits for treatment of intracranial hypertension in TBI. However, many practitioners still use hypothermia in the following contexts, which were not studied in Eurotherm3235-Trial:

a) Use only after hypertonic solutions have failed,

b) For more severe ICP elevations (> 25–30 mmHg), or

c) Only after barbiturate coma has failed.

The use of hypothermia in these clinical situations remains unclear, especially for utilization in patients with severe, refractory elevations in ICP. At our center, we continue to utilize TTM management to 33 °C for those patients with ICP > 25 mmHg who have failed hypertonic solutions in order to reduce the use of barbiturate-induced coma. However, there is no data to support the use of hypothermia prior to barbiturate use, vice versa, or at all. Figure 1 demonstrates how we integrate hypothermia into the step-wise management of ICP elevation at our institution (Fig. 1a, b).

Fig. 1
figure 1figure 1

a New York University-Langone ICP Management Algorithm that is used within our system that describes the steps taken to initially assess patients suspected of having increased intracranial pressure in addition to therapies used to treat non-intubated patients. b New York University-Langone ICP Management Algorithm describing the steps taken to manage intubated patients with suspected increased intracranial pressure, of note, therapies can be used concurrently with therapies used in a

Full size image

Eurotherm3235-Trial lends evidence that early use of therapeutic hypothermia in the staged management of ICP elevation is not beneficial. Furthermore, there is mounting evidence that prophylactic hypothermia does not benefit patient outcomes and does not meaningfully impact levels of intracranial hypertension. Combined with the earlier NABISH trials, the question of prophylactic hypothermia for TBI has been answered: as currently practiced, it does not help patients.

Surgical Decompression for Intracranial Hypertension After Traumatic Brain Injury

The “Randomised Evaluation of Surgery with Craniectomy for Uncontrollable Elevation of ICP” (RESCUEicp) Trial was designed to answer whether TBI patients benefited from decompressive craniectomy in the setting of refractory intracranial hypertension [46]. RESCUEicp was designed to overcome the study design criticisms of its predecessor, “Decompressive Craniectomy” (DECRA) trial, which enrolled patients with diffuse, bilateral injury and defined intracranial hypertension as ICP > 20 mmHg for at least 15 min in a 1-h period [47]. Please see Table 1 for a comparison of the DECRA and RESCUEicp trials. RESCUEicp included TBI patients with more commonly encountered mass lesions and had a sample size of 398 patients with well-balanced baseline demographics in both groups. RESCUEicp defined refractory intracranial hypertension as ICP elevation > 25 mmHg for 1 to 12 h [48]. Patients randomized into the surgical intervention arm underwent a decompressive craniectomy no later than 4 to 6 h after randomization, with the decision on craniectomy laterality left to the discretion of the surgeon. The primary outcome was measured using an ordinal analysis of the GOS-E at 6 months with a pre-specified sensitivity analysis for “favorable outcome” on the GOS-E at 6 months.

Table 1 Comparison of two recent trials in decompressive craniectomy for ICP elevation after TBI
Full size table

The ordinal analysis demonstrated a difference in GOS-E scores at 6 months and the pre-specified sensitivity analysis showed good outcomes in 45.4% in the surgical group as compared with 32.4% in the medical group (p = 0.01). Mortality was considerably lower in the surgical group but this was counterbalanced by an increased number of vegetative patients in this group. Interestingly, the European majority of enrolling centers led to almost 2/3 of the total craniectomies being bilateral, reflecting European practice patterns. Barbiturates were utilized in 87% of the medical group and in 9% of those in the surgical group. Importantly, over 1/3 of patients in the medical group ended up needed a surgical decompression thereby diluting the treatment effect seen in the study.

Given these results, the practitioner is faced with a dilemma very similar to that for patients presenting with malignant MCA infarct: will this patient “benefit” from surgery? Surgery is often life-saving but comes at the expense of increased levels of severe neurological impairment. Therefore, in-depth conversations with the family to ascertain the patient’s known wishes, as well as an open dialog about the expected functional outcomes, are essential to guide proper management based on individual patient preference.

Determining Optimal Cerebral Perfusion Pressure Thresholds

Ziai [49•] analyzed patient outcomes and ICP and CPP readings of 499 patients from the “Clot Lysis: Evaluating Accelerated Resolution of Intraventricular Hemorrhage Phase-III” (CLEAR-III) trial [50] to determine optimal ICP and CPP targets for patients presenting with hypertensive intraventricular hemorrhage. ICP elevations were not infrequent, with around 10% of q4 hour ICP readings above 20 mmHg and almost 2% above 30 mmHg. Given that ICP targets are mainly derived from the TBI literature, this important observational study found that sustained, unmitigated ICP elevations even as low as 18 mm Hg are associated with early mortality. The authors noted different ICP thresholds based on age, with those over 55 having lower ICP thresholds than younger patients (18 vs 22 mmHg). Low CPPs were correlated with both mortality and poor functional outcomes with a clear “dose” response for the amount of time spent at low CPPs.

Conclusion

Since the advent of the neurocritical care unit, elevated ICP management has been a central and essential point of goal-directed therapy. Significant progress has been made in evolving from invasive ICP monitoring to more non-invasive methods, however, gold standard continues to be invasive monitors such as EVDs which can also therapeutically drain CSF, if needed. Multiple avenues of treatment have also come forth in the care of these critical patients, and evidence has shown benefit and efficacy. Further work continues to provide clarity in the appropriate therapeutic treatment of intracranial hypertension.