Surgery for spontaneous intracerebral hemorrhage

A PubMed search for articles published from inception to July 2019 was performed by using the terms “Spontaneous Intracerebral Hemorrhage” [Mesh] AND “Surgery” [Mesh], which returned 261 articles. Also, the reference lists of the most recent guidelines on the management of ICH were scrutinized [9]. The author’s database was also searched for additional articles.

The mechanisms responsible for brain injury within the cerebral hematoma and the surrounding tissues are multiple and complex, which includes the primary effects of blood into the brain parenchyma and the secondary effects of hemoglobin breakdown and its products. Initially, there is the direct effect of acute hemorrhage into the brain parenchyma, causing disruption and mass effect within the cerebral tissue. This primary brain injury is followed by the interruption of bleeding in approximately two thirds of patients. However, in the remaining one third of patients, hematoma continues to expand in the first 24 h, which contributes to additional mass effect, midline shift [10], leading to further neurological deterioration and an increased risk of unfavorable outcome [11, 12].

The hyperacute management of ICH is focus on patients’ airway, breathing, and circulation stabilization, followed by the prevention of hematoma expansion. Several therapies attempting to reduce hematoma expansion have been studied, such as early aggressive blood pressure control [4, 5], the administration of tranexamic acid [6], and the use of recombinant activated factor VII [6, 7]. The use of recombinant activated factor VII reduced hematoma growth but did not decrease mortality or improve functional outcome [7]. Likewise, the early use of tranexamic acid was associated with a significant reduction in hematoma expansion, but did not improve functional outcome at 90 days [6]. Blood pressure control in the acute phase has modest effect in reducing hematoma growth; however, a preplanned pooled analysis of individual patient data obtained from the two largest trials of blood pressure lowering, the INTERACT2 [4] and the ATACH-II trials [5], demonstrated that achieving and maintaining a systolic blood pressure around 120–130 mmHg in the first 24 h is safe and might be associated with improved functional outcome [13].

Hematoma volume and location are the two main predictors of outcome related to the hematoma itself [11, 13, 14]. Hematomas greater than 30 ml are statistically associated with unfavorable outcome [15, 16]. The combination of hematoma volume greater than 60 ml with a GCS lower than 8 has a predicted 30-day mortality greater than 90% [16]. Acute hematomas greater than 150 mL usually leads to death due to the abrupt increase in intracranial pressure and consequently the reduction in cerebral perfusion pressure below critical levels [16].

Regardless hematoma volume, hemorrhages occurring in the posterior fossa (specially the cerebellum) may be life-threatening because the infra-tentorial space is smaller and less complacent than supratentorial area [17, 18]. Infra-tentorial hemorrhages could cause acute hydrocephalus due to fourth-ventricle compression and also lead to direct brainstem herniation [18]. Therefore, in posterior fossa hematoma evacuation may be considered as lifesaving option in patients with larger hematomas, brainstem compression, hydrocephalus, or clinical deterioration, though robust data is limited [17].

Additional to the physical effects of the initial and expanding hemorrhage, there are the effects of persistent hematoma and its blood products leading to a complex cascade of events (Fig. 1) [10, 19, 20].

The majority of ICH patients may not require surgery; however, there is a beneficial hypothesis for early surgical removal of an intraparenchymal hematoma. This benefit is based on the assumption that clot removal would restore the cerebral architecture, reducing mass effect and correcting or avoiding midline shift, and therefore it would improve cerebral perfusion by decreasing intracranial pressure. Additionally, hematoma drainage could prevent or at least reduce the cascade of secondary brain injury (Fig. 1) due to the deleterious effects of hemoglobin and its products into the brain. However, the surgical removal of a blood clot within the brain is not free of risks. In order to reach the hematoma that usually takes deep brain structures, a large layer of healthy cerebral tissue needs to be dissected, usually under general anesthesia. Additionally, postsurgical complications, such as hemorrhages and infections, are not uncommon in this clinical scenario, which increase the rates of mortality and unfavorable outcome [8, 21].

Several surgical approaches exist, which include (a) the insertion of external ventricular drain (EVD) for intraventricular hemorrhage (IVH) management and intracranial pressure (ICP) monitoring, (b) craniotomy for hematoma drainage (Figs. 2, 3, and 4), (c) decompressive craniectomy with or without hematoma drainage, and lastly (d) the use of minimally invasive the use of minimally invasive approaches (Fig. 5).

Intraventricular hemorrhage occurs in approximately 45% of patients with ICH, and it is an independent predictor of unfavorable outcome [22]. Intraventricular hemorrhage can interfere with the normal flow of cerebrospinal fluid, which may cause acute hydrocephalus, and in severe cases can lead to intracranial hypertension. Patients with acute hydrocephalus due to IVH or large intraparenchymal hematomas with mass effect associated with impaired level of consciousness (i.e., GCS ≤ 8) may require the urgent placement of an EVD, which allows for cerebrospinal fluid drainage and ICP monitoring [9, 23]. The goals for ICP and cerebral perfusion pressure (CPP) do not differ from those for traumatic brain injury, which suggests keeping an ICP < 20 mmHg (in more recent guideline 22 mmHg) and a CPP > 60 mmHg [23].

In severe cases, the large volume of blood in the ventricular system can cause drain malfunction and frequent catheter obstruction; therefore, the use of procedures to improve clot clearance have been tested. A phase III trial, the randomized, multicenter, multiregional, placebo-controlled CLEAR III trial [24] compared the use of low intraventricular dose (1 mg every 8 h, to a maximum of 12 doses) of recombinant tissue plasminogen activator (r-tPA) with placebo (i.e., normal saline) for patients with small spontaneous ICH (i.e., volume less than 30 ml) and an IVH obstructing the third or fourth ventricles.

The study investigators were solicited to clear as much IVH as possible, until the third and fourth ventricles were opened; or the IVH mass effect was relieved; or 80% of intraventricular clot was removed; to a maximum of 12 r-tPA doses. The intraventricular clot volumes were analyzed by a core laboratory using semi-automated segmentation and Hounsfield thresholds.

Five hundred participants, who routinely received an EVD, were included from 73 sites between 2009 and 2014. The primary favorable outcome defined as a 6-month modified Rankin scale (mRS) of 0–3 was not significantly different between the r-tPA and saline groups [r-tPA group 48% vs saline 45%; risk ratio (RR) 1.06 (95% CI 0.88–1.28; p = 0.554)]. The treatment with r-tPA was associated with 11% lower case fatality [46 (18%) vs saline 73 (29%), hazard ratio 0.60 (95% CI 0.41–0.86), p = 0.006), to the cost of an 8% increase in the proportion of patients in a vegetative state (i.e., mRS = 5); [42 (17%) vs 21 (9%); RR 1.99 (95% CI 1.22–3.26), p = 0.007]. Complications such as ventriculitis, symptomatic hemorrhage, and serious adverse events were not higher in the r-tPA group.

Eighty-two patients (33%) in the treatment group vs 24 patients (10%) in the control group achieved the endpoint of 80% intraventricular clot removal. A pre-specified secondary analysis showed a significant relation between the amount of clot removed [per clot remaining (mL) as measured by normalized AUC] and both mRS ≤3 [adjusted OR 0.96 (95% CI 0.94–0.97); p < 0.0001], and case fatality [adjusted HR of death per mL of time-weighted clot volume remaining 1.03 (95% CI 1.02–1.04); p < 0.0001]. One of the reasons why treatment was not effective may be explained by the fact that only one third of patients in the intervention group achieved the goal of clot removal.

Therefore, despite the association between the amount of clot removal and improved chances of mRS ≤ 3 (secondary analysis), the use of intraventricular r-tPA in patients with IVH obstructing the third or fourth ventricles did not improve 6-month functional outcome (primary outcome) when compared with placebo [24], and might increase the rates of survivorship with severe disability [25].

The use of dual EVD insertion, with and without thrombolytic therapy [26], and the combination of intraventricular fibrinolysis with lumbar drainage [27] have also been tested. The first was shown to increase clot resolution for large IVH (> 40 ml), with and without thrombolytic therapy [26]. The second significantly reduced the shunt dependency for hydrocephalus after IVH [27].

Another possible approach to manage IVH secondary to spontaneous intracerebral hemorrhage is the clot removal by neuroendoscopy in combination with EVD placement. Neuroendoscopy is minimally invasive and has high rates of clot evacuation with small proportions of surgical complications. A meta-analysis of 11 studies, which included only 5 randomized clinical trials, found the neuroendoscopy + EVD was superior than the EVD + r-tPA approach in terms of mortality, effectiveness of IVH evacuation, favorable functional outcome, and also the need for ventriculoperitoneal shunt [27, 28]. However, despite these interesting preliminary results, the efficacy of neuroendoscopic + EVD insertion for the treatment of IVH remains unclear [9]. Additionally, no definitive evidence concerning the preference between neuroendoscopy vs. EVD alone to treat IVH exists, because of limited data published to date [28].

Although the role of open surgery to treat patients with spontaneous ICH remains controversial, the use of craniotomy for supratentorial hematoma drainage is the most common strategy applied in most centers and also the most studied approach so far (Figs. 2, 3, and 4) [29, 30].

The first controlled study dated from the early 1960s [31], when McKissock and colleagues reported a prospective controlled trial of 180 patients randomized to craniotomy for hematoma evacuation vs. conservative management. Forty-six (51%) patients in the conservative group vs. 58 (65%) patients in the surgical group died. The authors were “unable to demonstrate any benefit from surgery in regard either to mortality or morbidity” [31]. Additionally, patients who were hypertensive had their mortality rate increased by surgery compared with conservative management. It is important to mention that since this early study, conservative management did not mean withholding life support. McKissock and colleagues stated “we would stress that conservative treatment involves more than ‘doing nothing’; nursing care of a high standard, constant medical supervision, and control of cerebral edema and pulmonary complications are implicit in the term” [31].

Decades have passed, but the role of craniotomy for hematoma evacuation remains a topic of hot debate, despite the publication of numerous studies (Table 2) [32‐48], including two well-designed, well-powered (10% absolute increase in favorable outcome in the surgical group), multicenter, multinational, randomized clinical trials [40, 44].

The Surgical Trial in Intracerebral Hemorrhage (STICH) [40] was the first well-powered, multicenter, multinational, randomized clinical trial to compare the benefits of early hematoma drainage with initial conservative management. One thousand and thirty-three (1033) patients with lobar or ganglionic spontaneous supratentorial hematoma were enrolled from 83 centers in 27 countries, to undergo early hematoma evacuation (within 24 h of randomization and within 72 h of ictus) or conservative management (i.e., best medical care with delayed surgery if necessary). Delayed hematoma evacuation was allowed in the conservative group if necessary, in case of delayed neurological worsening.

The study inclusion criteria included the following: (a) confirmation of a spontaneous supratentorial intracerebral hemorrhage by noncontrast CT head performed within 72 h of initial symptoms; (b) a hematoma diameter ≥ 2 cm; (c) GCS ≥ 5; and finally (d) clinical uncertainty, i.e., the responsible neurosurgeon was unsure about the clinical benefits of either treatment. Exclusion criteria included the following: (a) hemorrhage due to a vascular abnormality (e.g., cerebral aneurysm or an arteriovenous malformation); (b) hemorrhage due to tumors or trauma; (c) posterior fossa hemorrhage (i.e., cerebellar hemorrhage or supratentorial hemorrhage extending into the brainstem); (d) if the surgery could not be performed within 24 h of randomization; (e) if the patient was physically or mentally disabled before hemorrhage.

The primary outcome was death or disability according to the extended Glasgow outcome scale (eGOS—Table 2) assessed by structured postal questionnaires at 6 months and evaluated by blinded intention to treat analyses. The authors divided the patients in two groups of estimated prognoses (good and poor) according to the following equation:

A score > 27.672 was used as a cutoff point for a good prognosis. Therefore, patients predicted of poor outcome according to the above described prognosis-based methodology, a favorable prognosis was considered if eGOS = 4–8 was achieved, while for those patients with a predicted good outcome, a favorable outcome included eGOS = 5–8.

At 6 months, 51 patients (5%) had been lost to follow-up. No overall benefit in functional outcome was found with early hematoma drainage, since 122 (26%) patients progressed to a favorable outcome in the surgical group vs. 118 (24%) patients in the initial conservative treatment group (odds ratio 0·89 [95% CI 0·66–1·19], p = 0·414) [40]. Additionally, mortality rate was similar in both groups [36% surgery vs. 37% conservative; OR 0.95 (0.73–1.23), p = 0.707].

Consequently, a second study was performed by the same group of investigators to test the hypothesis that patients with superficial hematomas within 1 cm from cortical surface could benefit from early hematoma removal (early surgery versus initial conservative treatment in patients with spontaneous supratentorial lobar intracerebral haematomas - STICH II) [44]. The study was also an international, multicenter, prospective, randomized trial, which included only patients with superficial hematomas within 1 cm from the cortical surface of the brain. Patients with IVH, hematoma < 10 ml or > 100 ml, comatose patients (i.e., motor GCS < 5 and eye GCS < 2 at randomization), and patients admitted beyond 48 h of ictus were excluded. The same strategy to assess and dichotomized the primary outcome described above was used (i.e., death or disability by the extended Glasgow outcome scale assessed by structured postal questionnaires at 6 months and evaluated according to the prognosis-based outcome).

A total of 601 patients were included from 78 centers in 27 countries (307 in the early surgery and 294 in the conservative group), with an excellent follow-up at 6 months [589 out of 601 (98.0%) patients were available for follow-up at 6 months]. Nor overall benefit in functional outcome [62% unfavorable outcome in the surgical group vs. 59% in the initial conservative treatment group (absolute difference 3.7% (95% CI − 4.3 to 11.6), odds ratio 0.86 (0.62 to 1.20); p = 0.367)], neither mortality benefit was detected [18% in the surgical group vs. 24% in the conservative group (OR 0.71, 95% CI 0.48 to 1.06; p = 0.095)].

When the STICH trials results are combined in a meta-analysis with other 13 studies (sample size of 3366) [31‐36, 38‐41, 44], patients with predicted poorer prognosis, delayed clinical deterioration, or superficial lobar ICH without IVH may have a potential survival benefit [OR 0.74 (95% CI 0.64–0.86; p < 0.0001)] [44]. However, there is a substantial heterogeneity in the quality of studies (p = 0.0002), since the trials included have different patients’ populations and used multiple surgical strategies (e.g., craniotomy, endoscopic surgery, stereotactic ± plasminogen activator), limiting the validity of these results.

In summary, the two largest well designed, well-powered randomized clinical trials comparing early hematoma evacuation by craniotomy vs. initial conservative management did not show functional outcome or mortality benefit with early hematoma evacuation (Tables 1 and 2). Early craniotomy for hematoma evacuation cannot be recommended as routine care for patients suffering from supratentorial ICH, especially in deep hemorrhages and in small lobar hemorrhages with preserved level of consciousness. However, craniotomy for hematoma drainage is an important life-saving measure in critical situations, such as large hematomas with mass effect and midline shift leading to altered level of consciousness or when delayed neurological deterioration occurs due to hematoma expansion [21]. The ideal patients who would benefit from early surgery is still to be determined.

The practice of open craniotomy is not without risks and complications, because it requires a large bone flap, the exposition of the brain tissue, which is dissected, retracted, and manipulated in order to reach the location of hemorrhage (Fig. 4). Healthy brain tissue is damaged during this process. Instead, the application of alternative approaches has been tested in this clinical scenario, which includes the use of minimally invasive techniques, which has the theoretical benefit of producing minimum surgical trauma to the normal brain tissue manipulated throughout the process of hematoma drainage (Fig. 5).

The first controlled trial of minimally invasive surgery for ICH was performed in the 1980s and compared the use of endoscopic hematoma evacuation with conservative management [32]. In neuroendoscopy, an endoscope that measures approximately 5 to 8 mm in diameter with a miniature high-definition video camera attached is introduced through a burr hole created in the skull. The neuroendoscope navigates across normal brain tissue into the hemorrhage. Once the hematoma is reached, the blood clot can be aspirated by the endoscope ± the assistance of fluid or thrombolytic irrigation. Auer et al. [32] were the first to report a study that included 100 spontaneous ICH patients, who presented with focal deficits ± altered level of consciousness; cerebral hematoma ≥ 10 ml, and who were treated within 48 h of ictus. The authors used a rigid 6-mm endoscope tube, which was utilized to continuously rinsed the hematoma cavity with what the authors described as “artificial cerebrospinal fluid at body temperature through one channel at a pressure between 10 and 15 mmHg” [32]. Then, at regular time intervals, the mixture of blood and artificial CSF was suctioned through a separate endoscope channel.

Outcome assessment was performed 6 months after hemorrhage by a scale similar to the modified Rankin scale. Lower mortality (42 vs. 70%, p < 0.01) and higher rates of favorable outcome (40 vs. 25%, p < 0.01) were achieved by the surgical group; however, these results were limited to patients with subcortical hemorrhages, who were alert or somnolent perioperatively. The outcome was not improved by surgery in stuporous or comatose patients, neither in putaminal or thalamic hemorrhages. Although these promising results were achieved by a first-generation endoscope without CT guidance, they still need to be replicated in a well-powered randomized clinical trial.

More recently, Dr. Vespa and colleagues published the ICES trial (Intraoperative Computed Tomography–guided Endoscopic Surgery for Brain Hemorrhage) [46], a pilot multicenter randomized controlled trial funded by the National Institutes of Neurological Disorders and Stroke, which tested the safety and effectiveness of CT-guided endoscopic drainage of ICH. The trial included adult patients with supratentorial ICH within 48 h of ictus, who had a hematoma volume > 20 ml + GCS > 5 + NIHSS > 5. Fourteen patients underwent intraoperative computerized tomographic image–guided endoscopic surgery, which resulted in instant decrease of hematoma volume by 68 ± 21.6% (interquartile range 59–84.5), within 29 h hemorrhage ictus. The surgical procedures were very quick [1.9 h (interquartile range 1.5–2.2 h)], with only one surgical complication described (i.e., a peri-operative surgical bleed). Compared with the medical group from the MISTIE trial [47], the surgical group had a non-significant higher rate of favorable neurological outcome by mRS at 12 months (42.9% versus 23.7%; p = 0.19); however, the study was not powered to assess functional outcome and mortality.

Additional to neuroendoscopy, a second mode of minimally invasive surgery for ICH is the stereotactic or image-guided placement of a catheter inside the hematoma, followed by the intra-hemorrhage thrombolysis, with the ultimate goal of improving hematoma lysis and drainage. Usually, a catheter is left in place inside the hematoma, where frequent small amounts of a recombinant tissue-type plasminogen activator (r-TPa) are delivered in order to drain the clot over a period of days.

The minimally invasive catheter evacuation followed by thrombolysis (MISTIE) technique can be technically challenging for surgeons of variable levels of experience. A phase II study has demonstrated the importance of ideal catheter placement in order to achieve optimal hematoma evacuation [45]. In order to guarantee the accuracy of catheter insertion and the efficacy of hematoma drainage, studies using the MISTIE technique need to unify protocols of neurosurgeon training about the best selection of multiple surgical trajectories for catheter insertion in different hematoma locations (Fig. 5) [49].

This minimally invasive technique seems to be safe [47], feasible, efficacious [49], and reproducible [49, 50], and it is also associated with reduction in hematoma volume and peri-hematoma edema [51]. However, recent data arising from randomized controlled trials did not shown significant benefit of this technique when compared with conservative management [47].

The phase 2 MISTIE study was a randomized, controlled, open-label, phase 2 trial performed in 26 centers across North America and Europe [45]. Adult patients with spontaneous ICH + hematomas volume ≥ 20 ml were randomly allocated to conservative management or MISTIE + alteplase protocol (0.3 mg or 1.0 mg every 8 h for up to nine doses). According to the study protocol, neurosurgeons had to follow a 10-step procedure, with the ultimate goal to achieve a decrease in clot size to less than 15 ml. A rigid cannula was inserted through a burr hole, followed by clot aspiration through a 10-ml syringe. The procedure was stopped when a resistance was felt. Thereafter, the rigid cannula was replaced by a tunneled soft catheter under image guidance, with positioning confirmed by follow-up CT scan. After at least 6 h of catheter placement, the administration of alteplase in a dose of 0.3 mg or 1.0 mg diluted in 1 ml of saline was performed, followed by 3 ml of flush, every 8 h. The catheter was locked by an hour after alteplase infusion. Thrombolytic administration was stopped when residual hematoma was ≤ 15 ml, or when the maximum nine doses of alteplase were given, or in case of hemorrhagic complication, defined as a persistent decrease ≥ 2 points on the motor component of GCS, associated with an increase in the hematoma volume confirmed by CT scan. A total of 96 patients were included (54 in the intervention group and 42 in the conservative group). Thirty-day mortality [9.5%, (95% CI 2.7–22.6) vs. 14.8%, (6.6–27.1), p = 0.542], symptomatic bleeding [2.4%, (0.1–12.6) vs. 9.3%, (3.1–20.3), p = 0.226], and cerebral infections [2.4%, (0.1–12.6) vs. 0%, (0–6.6), p = 0.438] were not different between intervention and conservative groups, respectively. Only asymptomatic hemorrhage was more frequent in the intervention group [22.2%; (95% CI 12.0–35.6) vs. 7.1%; (1.5–19.5); p = 0.051) [45].

This pilot phase 2 study showed that intracerebral hemorrhage can be aimed and drained safely using serial thrombolytic injections through a stereotactically targeted catheter; therefore, a phase 3 trial was carried out.

The MISTIE III trial was an open label, phase 3 trial carried out at 78 hospitals in North America, Europe, Australia, and Asia [47]. The procedure for catheter placement and alteplase injection followed the same steps described above, except the dose of alteplase dose that was limited to 1.0 mg every 8 h to a maximum of nine doses. Adult patients with spontaneous supratentorial ICH + hematoma volume ≥ 30 ml + GCS ≤ 14 or NIHSS ≥ 6, and hematoma stability (hematoma expansion < 5 ml) for at least 6 h after diagnostic CT scan. A total of 506 patients were randomized (255 to MISTIE group vs. 251 to conservative management). The primary outcome was the percentage of patients with favorable functional outcome according to the mRS (0 to 3) at 12 months. The primary outcome was adjusted to ICH stability size, age, GCS, stability IVH size, and hematoma location. Although, MISTIE led to a mean reduction in hematoma size by 69% (SD 20) compared with 3% in the conservative treatment, no outcome benefit was found. At 12 months, 110 patients (45%) in the MISTIE group vs. 100 patients (41%) in the conservative group achieved a favorable outcome [adjusted risk difference 4% (95% CI − 4 to 12); p = 0.33]. The number of serious adverse events, such as symptomatic bleeding and cerebral infections, was similar between the two groups. The main conclusion of the study was that MISTIE is safe, but it does not improve long-term functional outcome. The authors performed a meta-analysis including only multisite trials of MISTIE in which functional outcome was evaluated by mRS or extended Glasgow Outcome Scale assessed at 180 days. No significant benefit of MISTIE was found (OR 0.61, 95% CI 0.29–1.26).

In summary, according to a large randomized, controlled, open-label, blinded endpoint phase 3 trial combined with a meta-analysis that compared minimally invasive surgery with thrombolysis vs. conservative management, despite being safe, it did not show long-term functional outcome benefit (Table 2) [47]. Therefore, MISTIE cannot be recommended as routine care in patients suffering from supratentorial ICH.

According to the American Heart Association/American Stroke Association Guidelines for the Management of Spontaneous Intracerebral Hemorrhage [9] and the European Stroke Organization (ESO) guidelines for the management of spontaneous intracerebral hemorrhage [54], for the majority of patients with spontaneous supratentorial hemorrhage, the benefit of surgical evacuation is not well established (Class IIb; Level of Evidence A) [9], with no supporting evidence for routine surgery (moderate quality, weak recommendation) [54]. However, surgery may be lifesaving for patients with a GCS score 9–12 (moderate quality, weak recommendation) [54], or patients with delayed neurological deterioration (Class IIb; Level of Evidence C) [9].

Decompressive craniectomy with or without hematoma evacuation may reduce mortality in patients with putaminal ICH, especially in those in coma with large hematomas leading to significant midline shift, or also in patients with refractory intracranial hypertension (Class IIb; Level of Evidence C) [9].

Regarding the use of minimally invasive surgical approach, i.e., stereotactic or endoscopic aspiration with or without thrombolytic, its effectiveness remains uncertain (Class IIb; Level of Evidence B) [9].

Patients with posterior fossa hemorrhage with acute hydrocephalus, brainstem compression, or worsening in neuro status, surgery should be performed as soon as feasible (Class I; Level of Evidence B) [9].