To update the recommendations of the Spanish Society of Neurology’s Stroke Study Group on the diagnosis and treatment of malignant middle cerebral artery (MCA) infarction and malignant cerebellar infarction, including the role of decompressive craniectomy.
DevelopmentA systematic review of the literature was conducted on PubMed, selecting clinical trials and meta-analyses evaluating malignant infarction as an independent variable. A series of PICO (patient, intervention, comparison, outcome) questions were developed to identify practical aspects of the diagnosis and the medical and surgical treatment of these patients. Based on the results, we developed a series of recommendations in response to each PICO question.
ConclusionsWe propose evidence-based definitions of malignant MCA infarction and malignant cerebellar infarction. Clinical trials of such anti-oedema therapies as osmotherapy and hypothermia have not been shown to reduce rates of mortality or disability. In patients aged < 60 years with malignant MCA infarction, the number needed to treat to avoid a death with decompressive craniectomy is 2, and the number needed to treat to achieve survival with a modified Rankin Scale score ≤ 3 is 4. In patients aged 60-75 years, decompressive craniectomy reduces mortality but is associated with severe sequelae, and therefore is not recommended for routine use. Suboccipital decompressive craniectomy should be considered either as a primary indication or after failure of a ventricular drain, in patients with extensive cerebellar infarction and neurological deterioration due to brainstem compression.
Actualizar las recomendaciones del Grupo de Estudio de Enfermedades Cerebrovasculares de la Sociedad Española de Neurología sobre el diagnóstico y tratamiento del infarto maligno de la arteria cerebral media (ACM) e infarto maligno cerebeloso, incluyendo el papel de la craniectomía descompresiva.
DesarrolloSe ha realizado una revisión sistemática de la literatura en Pubmed seleccionando ensayos clínicos y meta-análisis evaluando el infarto maligno como variable independiente. Se han elaborado preguntas tipo PICO (Patient, Intervention, Comparison, Outcome) para identificar aspectos prácticos del diagnóstico y del tratamiento médico y quirúrgico en el paciente con infarto maligno. A partir de los resultados se han redactado unas recomendaciones en base a cada una de las preguntas PICO.
ConclusionesSe propone una definición basada en la evidencia del infarto maligno de ACM e infarto maligno cerebeloso. Los ensayos clínicos realizados con medidas antiedema como la osmoterapia o la hipotermia no han demostrado reducir la mortalidad o la discapacidad. El número necesario de pacientes a tratar (NNT) para evitar una muerte mediante la craniectomía descompresiva en infartos de la ACM en pacientes <60 años es 2, y de 4 para conseguir una supervivencia con escala de discapacidad de Rankin modificada (ERm) ≤3. En pacientes entre 60-75 años, la craniectomía descompresiva disminuye la mortalidad, pero a expensas de secuelas graves por lo que no se recomienda de forma rutinaria. La craniectomía suboccipital descompresiva debe considerarse, bien como indicación primaria bien tras el fracaso de la derivación ventricular, en pacientes con infarto cerebeloso extenso y deterioro neurológico por compresión troncoencefálica.
Extensive ischaemic stroke is associated with high rates of disability and mortality, but malignant middle cerebral artery (MCA) infarctions represent a discrete entity in which mortality due to the mass effect of cerebral oedema and secondary herniation can be as high as 60%–80% in the absence of appropriate treatment. The publication of clinical trials of decompressive craniectomy in these patients represented a paradigm shift in the management of this disease, with a marked reduction in mortality and disability rates.1 Despite this, some uncertainty remains regarding the indication of this treatment, the eligible patient profiles, and the most appropriate time for surgery.
In recent years, new clinical trials have studied decompressive craniectomy to treat malignant infarction in older patients and with wider therapeutic windows. Other clinical trials have analysed new therapeutic approaches from a medical perspective, such as therapeutic hypothermia or the administration of drugs to prevent or reduce the development of cerebral oedema.
In cerebellar infarcts, secondary oedema occurs in a space that is anatomically delimited by 2 rigid structures: the squamous part of the occipital bone, below, and the cerebellar tentorium, above. As a result of these anatomical characteristics, the pressure vector in the event of mass effect is primarily forwards. Therefore, oedema secondary to cerebellar infarction can cause compression of the pons, hydrocephalus secondary to obstruction of the fourth ventricle, and even infratentorial herniation.2 Few data are available on the incidence of severe oedema as a complication of cerebellar infarction, although approximately 20% of patients are estimated to develop radiological signs of oedema accompanied by clinical deterioration.3,4 Infarct size is the main predisposing factor for oedema, although such other factors as haemorrhagic transformation or inadequate collateral circulation are also involved. Extensive cerebellar infarctions are associated with considerable disability and mortality. However, given the scarcity of quality studies into the medical and surgical treatment of these patients, there is considerable uncertainty regarding the most appropriate management.
Therefore, the Spanish Society of Neurology’s Stroke Study Group, as part of its clinical practice guidelines for the acute treatment of ischaemic stroke, has updated its recommendations on the diagnosis and treatment of malignant MCA infarction and cerebellar infarction.
These recommendations are intended to guide and support diagnostic and therapeutic decision-making by healthcare professionals attending patients with ischaemic stroke.
This article evaluates the clinical and imaging criteria for the diagnosis of malignant ACM infarction and cerebellar infarction, assesses the efficacy and indications of medical treatment, and analyses indications for surgical treatment and its limitations.
MethodsA series of PICO (patient, intervention, comparison, outcome) questions were formulated about malignant MCA infarction and its treatment: 1.1) How is malignant infarction diagnosed in patients with ischaemic MCA stroke?; 1.2) How is malignant infarction diagnosed in patients with cerebellar infarction?; 2.1) Does treatment with osmotherapy, therapeutic hypothermia, hyperventilation/sedation, or corticosteroids decrease rates of disability or mortality in patients with malignant MCA infarction?; 2.2) Does osmotherapy decrease rates of disability or mortality in patients with malignant cerebellar infarction?; 3) Is intracranial pressure (ICP) monitoring indicated in patients with suspected malignant infarction?; 4.1) Is decompressive hemicraniectomy indicated in patients with malignant MCA infarction?; 4.2) Is suboccipital decompressive craniectomy indicated in patients with malignant cerebellar infarction?; 5.1) What is the therapeutic window for hemicraniectomy in patients with malignant MCA infarction?; 5.2) What is the therapeutic window for suboccipital decompressive craniectomy in patients with malignant cerebellar infarction?; 6) Does laterality of involvement affect the prognosis of malignant MCA infarction?
A systematic review was conducted on PubMed of articles published up to June 2024. Details of the search strategy are included in the supplementary material.
Subsequently, we reviewed the articles identified, with selection prioritising randomised clinical trials, meta-analyses, and systematic reviews.
Evidence was classified as level A (high-quality evidence from more than one randomised clinical trial, meta-analyses of high-quality clinical trials, or data from one or more randomised clinical trials corroborated by high-quality registry studies), level B (moderate-quality evidence, based on one or more randomised clinical trials; one or more non-randomised, observational, or high-quality registry studies; meta-analyses of moderate-quality clinical trials; or meta-analyses of non-randomised studies), or level C (limited evidence, with data from observational studies or registry studies presenting methodological limitations in design or execution).5
Grades of recommendation were classified as class I (strong recommendation: benefits far greater than risk), class IIa (moderate recommendation: benefit outweighs risk), class IIb (weak recommendation), class III-no benefit (benefit is equal to risk), and class III-harm (risk outweighs benefit).
ResultsHow is malignant infarction diagnosed in patients with ischaemic middle cerebral artery stroke?Given the lack of a single international consensus definition of malignant MCA infarction, we reviewed the inclusion criteria in clinical trials of patients with malignant infarction. In all clinical trials, the fundamental criterion for this diagnosis was presence of extensive neurological deficit associated with decreased level of consciousness.
The definition of extensive neurological deficit varies between studies, with the majority of trials using the National Institutes of Health Stroke Scale (NIHSS) to assess the neurological deficit, differentiating between the dominant and non-dominant hemispheres. All trials include the concept of decreased level of consciousness as a key finding in the neurological examination, and exclude patients with bilateral mydriasis or Glasgow Coma Scale (GCS) scores < 6. In the DECIMAL clinical trial,6 the inclusion criterion for malignant infarction was NIHSS ≥ 16, with a score ≥ 1 in item 1a (level of consciousness). The HAMLET clinical trial7 included patients with NIHSS scores ≥ 16 for right-hemisphere stroke and ≥ 21 for left-hemisphere stroke, including a gradual decrease in level of consciousness with a GCS score ≤ 13, for right-sided stroke, or ≤ 9 for the components eye opening and motor response, for left-hemisphere stroke. The DESTINY trial8 included patients with NIHSS scores > 18 for right-hemisphere lesions and > 20 for left-hemisphere lesions, with a score of ≥ 1 for item 1a. The DESTINY II trial9 used less strict criteria than its predecessor: NIHSS ≥ 15 in stroke affecting the non-dominant side and ≥ 20 if the dominant side was affected, with a score ≥ 1 for item 1a. The trial by Zhao et al.10 did not include a cut-off score for neurological deficit, with the exception of altered level of consciousness (GCS ≤ 9) and alterations in the eye opening and motor response items. Slezins et al.11 defined neurological deficit in malignant infarction as NIHSS ≥ 16, regardless of which hemisphere was involved. Similarly, the HeADDFIRST trial used a score of NIHSS ≥ 18, without distinguishing between hemispheres, with a level of consciousness of < 2 on item 1a. Trials of therapeutic hypothermia for malignant infarction included patients with NIHSS ≥ 15 for stroke in the non-dominant hemisphere and ≥ 20 for stroke in the dominant hemisphere, and scores ≥ 1 for item 1a.12,13
We also reviewed the radiological criteria used in clinical trials to define malignant MCA infarction. Only 2 trials included patients with malignant infarctions evaluated according to volume on MRI diffusion-weighted sequences, with both trials using a minimum volume of 145 cm3.6,11 The remaining clinical trials included patients with infarct volumes of at least two-thirds of the MCA territory, including the basal ganglia.7–10,14–16 Two clinical trials defined malignant infarctions as those greater than 50% of the MCA territory.6,11 The DESTINY and DESTINY II trials,8,9 as well as the trial by Zhao et al.,10 specifically allowed for the possibility of including patients with concomitant infarction of the posterior or the anterior cerebral artery. The HAMLET trial7 excluded patients with complete infarction of an entire cerebral hemisphere, but not those with concomitant posterior or anterior cerebral artery involvement.
RecommendationA patient with malignant MCA infarction is defined as a patient meeting all of the following criteria (level of evidence A, class I recommendation):
- •
Severe neurological deficit, with an NIHSS score ≥ 15 for lesions in the non-dominant hemisphere or ≥ 20 for lesions in the dominant hemisphere.
- •
Progressive decrease in level of consciousness with scores ≥ 1 for NIHSS item 1a.
- •
Extensive acute infarction (neuroimaging) affecting at least two-thirds of the MCA territory, including the basal ganglia, as measured with head CT or MRI (level of evidence A) with oedema with mass effect (level of evidence B); or with a volume > 145 cm3 measured with diffusion-weighted MRI (level of evidence B). Patients may display ipsilateral infarction of the anterior and/or posterior cerebral artery territory. Midline shift is not an essential criterion.
Given the lack of a universally accepted definition of malignant cerebellar infarction, we reviewed the criteria used in the different studies analysed.
Definitions of impairment secondary to oedema in cerebellar infarction vary between studies, and are based on clinical findings, neuroimaging evidence, or a combination of the 2. Thus, candidates for surgical treatment after cerebellar infarction are defined or selected according to such disparate criteria as progressive neurological deterioration due to large cerebellar infarction,18 clinically significant cerebellar infarction,19 brainstem compression with associated obstructed hydrocephalus,2 expansive cerebellar infarction with decreased level of consciousness and radiological evidence of posterior fossa compression,19 GCS < 12 at admission,3 GCS ≤ 13,19 deterioration on GCS (scores ≥ 2),3,20 acute hydrocephalus, rapid deterioration to coma, infarct volume > 5 cm3,21 infarct volume ratio of 0.25-0.33,22 neuroimaging findings showing tight posterior fossa,23 or scores ≥ 4 on the Jauss scale,24 among others. Thus, none of these criteria can be treated as more than a reference or non-validated approach to defining the condition.
From a clinical perspective, decreased level of consciousness is considered the most reliable indicator of cerebellar oedema,2 with the GCS being the most frequently used tool to assess this sign. However, no specific cut-off score or rate of deterioration is universally accepted as an indicator.
In addition to the decrease in level of consciousness, and in the absence of a primary brainstem lesion, pontine compression secondary to oedema may give rise to such other clinical manifestations as oculomotor alterations, changes in breathing pattern, or arrhythmias; these manifestations should also be considered as part of therapeutic decision-making.
From a radiological viewpoint, some authors associate the development of oedema with posterior inferior cerebellar artery infarction. However, clinical deterioration depends more on infarct volume than on involvement of any specific vascular territory.2,3
Head CT findings of displacement or effacement of the fourth ventricle are considered the key radiological marker for identifying mass effect secondary to cerebellar infarction, followed by obliteration of the quadrigeminal cistern, brainstem compression, hydrocephalus, downward tonsillar herniation, and upward transtentorial herniation.3
Jauss’ study group2,25 proposed a scale based on CT findings, scored from 0 to 9 points. The scale considers 3 indicators: compression of the fourth ventricle (0: no compression; 1: unilateral compression; 2: midline shift; 3: not visible), compression of the quadrigeminal cistern (0: no compression; 1: mild with asymmetrical compression ipsilateral to the infarction; 2: moderate with evidence of bilateral compression; 3: severe bilateral compression with obscured trigeminal cistern), and dilatation of the inferior horn of the lateral ventricle (0: no dilatation; 1: mild; 2: moderate; 3: severe). Scores of 0-3 are considered to indicate mild or absent mass effect, scores of 4-6 indicate moderate mass effect, and scores of 7-9 indicate severe mass effect.
In a series of 67 patients, Wang et al.26 identified growth of the oedema volume on serial CT studies conducted over the first 48 hours as an independent predictor of the need for decompressive craniectomy (20.1 mL/day vs 8.01 mL/day). Fabritius et al.27 studied a series of 51 patients to analyse the value of CT perfusion study in the early identification of patients who subsequently developed malignant cerebellar oedema. They found that this patient group displayed greater deficit volumes on all perfusion sequences and lower posterior circulation–Acute Stroke Prognosis Early CT Score from perfusion maps. A cut-off point of 22 mL (40% of a cerebellar hemisphere) for cerebellar blood flow deficit volume presented the best levels of sensitivity/specificity (100% and 90%, respectively). However, these data are merely exploratory, as the series included only 9 patients with malignant cerebellar infarction.
Few studies have addressed the value of MRI for predicting cerebellar oedema and evaluating brainstem displacement or damage to critical structures. In this regard, Koh et al.3 found that MRI or CT detection of hydrocephalus, brainstem deformity, and basal cistern compression may predict clinical deterioration, whereas vertical displacement of the amygdalae or aqueduct did not. Although these results may be relevant for selecting candidates for surgery, the correlation with clinical deterioration is low, with radiological signs of cerebellar infarction with mass effect being associated with clinical deterioration in fewer than 50% of cases.
RecommendationMalignant cerebellar infarction is defined clinically as infarction with progressive impairment of level of consciousness or with the appearance of progressive signs of brainstem compression. Deterioration must be recognised, through monitoring of level of consciousness and of the appearance of new signs of brainstem involvement. The GCS is a useful tool for this assessment. However, no cut-off score or rate of deterioration is universally accepted as a reliable predictor for therapeutic decision-making. (Level of evidence B, class IIa recommendation).
Extensive acute infarction (neuroimaging). Though their role as a predictor of clinical deterioration is not well established, we must assess the appearance of radiological markers of mass effect secondary to extensive cerebellar infarction (displacement or effacement of the fourth ventricle, quadrigeminal cistern compression, brainstem deformity, hydrocephalus/dilatation of the inferior horn of the lateral ventricle, downward tonsillar herniation, and upward transtentorial herniation) or oedema growth on serial CT studies; this information may support therapeutic decision-making. (Level of evidence B, class IIb recommendation).
Does treatment with osmotherapy, therapeutic hypothermia, hyperventilation/sedation, or corticosteroids decrease rates of disability or mortality in patients with malignant middle cerebral artery infarction?Below, we review the published studies of different medical treatments for patients with malignant infarction.
OsmotherapySeveral clinical trials and meta-analyses address the use of various osmotic agents to treat ischaemic stroke.
We identified 4 studies and one meta-analysis of intravenous mannitol treatment of ischaemic stroke. Generally, these studies present poor methodological quality, as they are not randomised, controlled clinical trials, do not include a target population exclusively composed of patients with malignant infarction, or provide few clinical data.
The study conducted by Bereczki et al.28 included 805 patients with stroke, which was ischaemic in 666 cases. Mannitol treatment was administered to 546 patients within 72 hours of stroke onset; 259 did not receive the treatment. This was a non-randomised study in which treatment was indicated according to clinical judgement. Only 21% of patients presented impaired level of consciousness at treatment onset, and no data are provided on infarct volume or midline shift; therefore, this is not a specific study of malignant infarction or cerebral oedema. The study reports 30-day mortality rates of 25% in the mannitol group and 16% in the control group (P = .006) and 12-month rates of 38% vs 25% (P < .001), although baseline characteristics were not well balanced.
A 2007 Cochrane meta-analysis29 included 3 clinical trials with a total of 226 patients, of which only one trial included patients with probable ischaemic stroke (without CT confirmation: the trial was conducted in 1978) and treated with mannitol.30 The study does not report specific mortality rates in patients with ischaemic stroke, and no clinically significant differences were detected between the mannitol group and controls.
The study by Schwarz et al.31 included 9 patients with stroke (8 ischaemic strokes) under ICP monitoring, with a total of 30 episodes of ICP crises, who were randomly assigned to receive intravenous mannitol or hypertonic saline hydroxyethyl starch (HS-HES). Both treatments were associated with significant decreases in ICP (P < .01), with HS-HES achieving a greater decrease; cerebral perfusion pressure increased significantly after administration in the mannitol group but remained stable in the HS-HES group. Rates of mortality and disability are not reported. The same study group later reported 8 patients with 22 ICP crises resistant to mannitol, who were treated with hypertonic saline solution; this treatment achieved a significant decrease in ICP at 35 minutes, with an increase in cerebral perfusion pressure.32 Once more, the authors do not report mortality or disability data.
In 2011, Diringer et al.33 published a randomised clinical trial assessing haemodynamic changes after osmotherapy in patients with extensive ischaemic stroke displaying midline shift > 2 mm and clinical deterioration. The study included 9 patients who were randomly allocated to receive mannitol or hypertonic saline solution, and underwent brain positron emission tomography imaging studies before and 1 hour after treatment; a trend towards increased cerebral blood flow contralateral to the infarction was observed for mannitol (P = .09) but not for saline treatment. The authors do not report mortality or disability data.
A 2004 Cochrane meta-analysis of glycerol treatment34 included 11 clinical trials, with a total of 945 patients, comparing this treatment against placebo to reduce cerebral oedema. These clinical trials present several limitations, as many studies were performed prior to the widespread availability of head CT; therefore, it was not possible to confirm the presence of cerebral ischaemia in many patients. The authors of the meta-analysis conclude that glycerol treatment is not recommended in patients with ischaemic stroke.
Preclinical trials have shown glibenclamide to be capable of reducing cerebral oedema and improving survival in extensive ischaemic stroke. In this regard, a double-blind, randomised, phase 2 clinical trial was conducted to compare intravenous glibenclamide against placebo,35 including 86 patients with acute stroke displaying MRI lesions of 82-300 cm3. No significant differences were observed in the primary outcome measure (modified Rankin Scale [mRS] score of 0-4 at 90 days, not requiring decompressive craniectomy), which was observed in 41% of patients in the glibenclamide group and 39% of controls (odds ratio [OR]; 0.87; [95% confidence interval (CI), 0.32-2.32]; P = .77). In secondary subanalyses,36,37 no differences were observed in rates of haemorrhagic transformation or malignant oedema, although glibenclamide treatment was associated with a reduction in deaths due to cerebral oedema (2.4% vs 22.2%; P = .01) and less midline shift (P < .01). A multicentre, randomised phase 3 trial (252LH301, CHARM) has since been conducted, and was terminated early in 2023 according to a decision by the sponsor.
HypothermiaThe study published in 2016 by Su et al.13 reports an open (blinded raters), randomised, controlled clinical trial of mild hypothermia (33-34 °C). The study included 33 patients aged 18-80 years with malignant infarction after stroke of < 48 hours’ progression, with infarction greater than two-thirds of the MCA territory and decreased level of consciousness, who were not eligible for decompressive craniectomy. Sixteen patients were treated with endovascular hypothermia with saline solution (mild hypothermia for 24-72 hours), and 17 were assigned to the control group. The hypothermia group showed a greater number of adverse events (P < .001), particularly bradycardia, electrolyte disorders, gastrointestinal bleeding, and hyperglycaemia. The 3-month mortality rate was 50% in the hypothermia group and 41% in the control group (P = .73); mRS scores of 1-3 at 90 days were observed in 43.8% in the hypothermia group vs 23.5% in the control group (P = .28).
Subsequently, a randomised clinical trial analysed moderate hypothermia (33.0 °C ± 1.0 °C) in 50 patients with malignant infarction undergoing decompressive craniectomy.12 The study included patients aged 18-60 years, with intravascular or surface systems used to induce hypothermia for 72 hours after craniectomy. The primary outcome measure was mortality at 14 days, with rates of 19% in the hypothermia group and 13% in the control group (P = .70); the trial was stopped due to a significant increase in adverse events in the treatment group (80% at one year, vs 43% in the control group; P = .005). No differences were observed in rates of disability.
Schneider et al.38 published a trial of decompressive craniectomy and hypothermia (48 hours; target temperature 33-34 °C) in 53 patients with malignant infarction. Patients were compared against 58 historical controls who underwent decompressive craniectomy in the DECIMAL, DESTINY, and HAMLET trials. In the primary outcome measure, no benefit was observed in the percentage of patients with mRS scores of 0-3 at 12 months (25% in the hypothermia group and 41% in controls; OR: 0.66 [95% CI, 0.38-1.13]). Furthermore, the hypothermia group showed a lower percentage of patients scoring 0-4 on the mRS (40%, vs 72% among controls; OR: 0.53 [95% CI, 0.37-0.76]) and lower survival (49% vs 79%; OR: 0.60 [95% CI, 0.44-0.82]).
The clinical trial by Park et al.39 included 20 patients with malignant infarction treated with mild hypothermia after decompressive craniectomy, whose progression was compared against a historical cohort of 27 patients treated with decompressive craniectomy only. The study included patients with an infarct volume greater than two-thirds of the MCA territory and GCS scores < 9. A trend towards lower in-hospital mortality was observed in the hypothermia group (15.0%, vs 40.7% in controls; P = .056), with hypothermia being an independent predictor of survival in the multivariate analysis (OR: 6.21 [95% CI, 1.04-37.05]; P = .045). No significant differences were observed in disability at 12 months (P = .35). The overall number of adverse events is not reported, but treatment was discontinued for this reason in 3 patients.
In another open, single-centre, randomised trial, Shang et al.40 studied 60 patients aged < 70 years with ischaemic stroke of < 12 hours’ progression; 30 were treated with hypothermia (32-35 °C) and 30 received conventional treatment. Significant differences were observed in the NIHSS, Barthel Index, and GCS at 72 hours (P < .05), but no medium- or long-term disability or mortality data are reported. In a randomised clinical controlled trial of 34 patients, Fan et al.41 compared decompressive craniectomy against craniectomy plus hypothermia (endovascular or local with head surface cooling), finding no significant differences in mortality or disability; a higher rate of adverse events was observed in the endovascular hypothermia group.
Finally, Els et al.42 conducted a randomised, controlled clinical trial of 25 patients with ischaemic stroke and infarct volume greater than two-thirds of the MCA territory who were treated with craniectomy plus hypothermia (35 °C; n = 12) or craniectomy only (n = 13). No adverse events were observed in the hypothermia group, with an overall mortality rate of 12%. Mortality rates were similar in both groups, with a trend towards higher NIHSS scores at 6 months in the hypothermia group.
Hyperventilation/sedationWe identified no clinical trials of other treatments for managing intracranial hypertension in malignant infarction, such as hyperventilation and barbiturates. Schwab et al.43 studied 60 cases of barbiturate treatment in patients with extensive ischaemic stroke and severe intracranial hypertension despite previous treatment with osmotherapy and hyperventilation. Eighty-three percent presented an initial improvement of ICP, although survival was only 8%. Twenty-five percent presented severe adverse events related to barbiturate treatment.
CorticosteroidsWe identified no studies specifically assessing the efficacy and/or safety of corticotherapy in patients with malignant infarction. A 2011 meta-analysis by Sandercock and Soane,44 on the effect of corticotherapy in ischaemic stroke in general, included 8 studies of a total of 466 patients; the majority were performed in the 1970s and 1980s and include little information on methodological quality or the risk of bias. No differences were observed in the risk of mortality at one year (OR: 0.87 [95% CI, 0.57-1.34]) or in functional status. The adverse effects reported included gastrointestinal bleeding, infections, and hyperglycaemia.
Recommendations- •
No specific recommendations can be made on the efficacy or safety of osmotherapy in patients with malignant MCA infarction and intracranial hypertension. Although the use of these agents seems reasonable in this context, insufficient evidence is available to recommend or discourage their use (expert recommendation). Insufficient evidence is available to recommend a specific drug for osmotherapy.
- •
Outside of clinical trials, glibenclamide is not indicated for the treatment of malignant MCA infarction (level of evidence B, grade IIb recommendation).
- •
Hypothermia is not recommended for the treatment of malignant MCA infarction (level of evidence B, grade III recommendation).
- •
Corticosteroids and barbiturates are not recommended for the treatment of malignant MCA infarction (level of evidence C, grade III recommendation).
We identified no studies analysing the efficacy or safety of treatment with osmotic agents in cerebellar infarction. The studies identified assess hemispheric infarcts only,31,32 or offer global data without specifying the territories involved in the description of the series or in the results section.28 Therefore, we are unable to establish any conclusions beyond the recommendations for hemispheric infarction.
RecommendationOsmotic agents. No specific recommendations can be made on the efficacy or safety of osmotherapy for malignant cerebellar infarction.
Is intracranial pressure monitoring indicated in patients with suspected malignant infarction?A literature search identified no high-quality or low-quality clinical trial assessing the efficacy or safety of ICP monitoring in extensive ischaemic stroke with mass effect or impaired level of consciousness.
The 1996 study by Schwab et al.45 analysed 48 patients with extensive ischaemic stroke and intracranial hypertension, who did not undergo craniectomy, and who were monitored with ipsilateral epidural ICP probes. The survival rate was only 19%, with the cause of death being transtentorial herniation in 39 cases. In all patients, clinical signs of herniation were observed prior to detection of intracranial hypertension, and this monitoring did not correlate adequately with changes in head CT results.
Poca et al.46 assessed 19 patients with malignant infarctions and intraparenchymal ICP monitoring prior to craniectomy, observing that despite mean midline shift of 6.7 mm, ICP was below 20 mmHg in all cases, even in 2 patients with signs of brainstem compression and pupillary alterations.
Jeon et al.47 monitored ICP in 25 patients following craniectomy for malignant infarction, observing correlations between ICP and midline shift (P = .009), pupillary alterations (P = .015), GCS score (P = .025), and mortality at 6 months.
Recommendation- •
ICP monitoring is not indicated for decision-making regarding decompressive hemicraniectomy in patients with malignant MCA infarction (level of evidence B, grade IIb recommendation). The indication should be based on clinical and radiological criteria (level of evidence A, grade I recommendation) (Table 1).
Table 1.Summary of the main studies assessing the medical treatment of malignant infarction.
STUDY METHODS/DESIGN PARTICIPANTS/SELECTION EXPERIMENTAL INTERVENTION EVALUATION NOTES RESULTS (DHC vs MT) Bereczki et al29(2007) • Prospective• Control group• Non-randomised trial • 805 patients • 0-72 hours after stroke onset• No severity limit • Intravenous mannitol vs control Mortality at 30 days and at 1 year • The sample included patients with stroke in general, with or without increased intracranial pressure• Significant baseline differences between groups, with greater severity in the mannitol group • Mortality at 30 days: 25% with mannitol vs 16% in the control group (P = .006)• Mortality at 1 year: 38% vs 25% (P < .001)• Multivariate analysis: no effect or harm (with different models) Bereczki et al29(2007) Meta-analysis • 3 trials• 226 patients with stroke included• 166 patients with ischaemic stroke analysed, 36 treated with mannitol • Intravenous mannitol vs control or placebo Mortality, disability • No specific assessment of mortality due to ischaemic stroke• As the trial of ischaemic stroke was performed in the pre-CT era, neuroimaging confirmation is not available • 33% of patients receiving mannitol showed clinical improvement, vs 34% in the control group (OR: 1.02 [0.42-2.5]) Swartz et al31(1998) • Randomised• Controlled trial • 30 episodes of increased intracranial pressure (in 9 patients) • Intravenous mannitol vs hypertonic saline solution • Change in intracranial pressure• No clinical data • Decrease of 11.4 mm Hg with hypertonic saline solution vs 6.4 with intravenous mannitol• Significant increase in cerebral perfusion pressure with mannitol Diringer et al33(2011) • Randomised• Controlled trial • 9 patients with stroke displaying midline shift > 2 mm and clinical deterioration • Intravenous mannitol vs hypertonic saline solution • Changes in cerebral perfusion• No clinical data • Brain PET scan before and 1 hour after treatment • Trend towards increased cerebral blood flow contralateral to the infarction with mannitol (P = .09)• No differences with saline solution. Righetti et al.34(2004) • Meta-analysis • 11 randomised clinical trials of glycerol vs control• 482 patients with glycerol vs 463 controls • Glycerol vs control • Mortality, disability • Many trials were performed in the pre-CT era• Glycerol is not recommended in patients with acute stroke • Non-significant reduction in the risk of death: OR: 0.78 (0.58-1.06)• Good clinical progression: OR: 0.73 (0.37-1.42) Seth et al35(2016) • Randomised• Controlled• Double-blind trial • 86 randomised patients• 18-80 years• < 80-300 cm3 lesion on DWI • Glibenclamide vs placebo • Disability (mRS 0-4 at 90 days) • Terminated early • mRS ≤ 4 at 90 days: 41% vs 39% (OR: 0.8P ≤ .77)• mRS ≤ 3 at 12 months: 50% vs 22.2% (P = .10)• Survival at 30 days: 88% vs 47% (P = .02). Osmotherapy Therapeutic hypothermia STUDY METHODS/DESIGN PARTICIPANTS/SELECTION INTERVENTION EVALUATION NOTES RESULTS (DHC vs MT) Su et al13(2016) • Randomised trial• Controlled• Open, rater blinded trial • 33 patients• Infarction > 2/3 of MCA territory; > 48 hours' progression• Altered level of consciousness and NIHSS > 14 (non-dominant side affected) or > 19 (dominant side affected)• Not eligible for craniectomy • Hypothermia (24-72 hours' duration) vs control• Target temperature 33-34°C• Endovascular injection of saline solution • Mortality and mRS of 1-3 at 90 days • Mortality: 50% for hypothermia vs 41% for controls (P = .73)• mRS 1-3: 43.8% for hypothermia vs 23.5% for controls (P = .28) Neugebauer et al12(2019) • Randomised trial• Controlled• Open • 50 patients• 18-60 years• Altered level of consciousness and NIHSS > 14 (non-dominant side affected) or > 19 (dominant side affected).• Infarction > 2/3 of MCA territory; > 48 hours' progression; treated with craniectomy • Moderate hypothermia vs controls• Target temperature 33°C (± 1.0°C)• Endovascular or surface hypothermia Primary outcome measure: mortality at 14 daysRate of adverse events at 14 days and at 12 months Terminated early due to safety concerns • 14-day mortality of 19% for hypothermia vs 13% for controls (P = .70)• Increase in severe adverse events at 14 days (P = .26) and 12 months (P = .005)• No differences in disability Schneider et al38(2017) • Clinical trial• Historical control group with decompressive craniectomy • 53 patients treated with hypothermia• 18-60 years• Malignant MCA infarction; > 48 hours' progression; treated with craniectomy • Hypothermia (target temperature 33-34°C) Primary outcome measure: Disability (mRS 0-3 at 12 months)Secondary outcome measures: mortality and mRS 0-4 at 12 months • mRS score ≤ 3 at 12 months: 25% for hypothermia vs 41% for controls; OR: 0.66 (95% CI, 0.38-1.13)• mRS score at ≤ 4 at 12 months: 40% vs 72%; OR: 0.53 (0.37-0.76)• Survival at 12 months: 49% vs 79%; OR: 0.60 (0.44-0.82) Park et al39(2018) • Clinical trial• Historical control group with decompressive craniectomy • 20 patients treated with hypothermia• Infarction > 2/3 of MCA territory• GCS < 9 • Hypothermia (34°C) with surface system Primary outcome measure: Death during hospitalisation • Death during hospitalisation: 15.0% for hypothermia vs 40.7% for controls; (P = .056).• No differences in disability P = .36)• Multivariate analysis: beneficial effect for mortality; OR: 6.21 (1.04-37.05); P = .045 Zhang et al40(2019) • Randomised trial• Controlled • 30 patients treated with hypothermia and 30 controls• Massive stroke; < 12 hours' progression• < 70 years • Hypothermia (target temperature: 32-35°C); 72 hours duration NIHSS, Barthel Index, and GCS at 72 hours of treatment • Medium-/long-term disability/mortality not assessed • Significant differences in NIHSS, Barthel Index, and GCS at 72 hours in the hypothermia group (P < .05). Els et al42(2006) • Randomised• Controlled • 12 patients treated with hypothermia plus craniectomy vs 13 with craniectomy only• Infarction > 2/3 of MCA territory • Hypothermia (target temperature 35°C) Mortality and disability (NIHSS) • No differences in morality (15% vs 8.3%; ns)• Trend towards better NIHSS scores at 6 months (10 vs 11; P < .08) DHC: decompressive hemicraniectomy; GCS: Glasgow Coma Scale; MCA: middle cerebral artery; mRS: modified Rankin Scale; MT: medical treatment; NIHSS: National Institutes of Health Stroke Scale.
The main clinical trials of decompressive craniectomy are summarised in Table 2.
Summary of the main studies assessing the surgical treatment of malignant middle cerebral artery infarction.
| Study | Methods | Design | Participants/selection | Intervention | Evaluation | Notes | Results (DHC vs MT) |
|---|---|---|---|---|---|---|---|
| DECIMAL6 (2007) | ● Prospective | ● CDHC + medical treatment vs medical treatment only | ● 38 patients | ● Extensive hemicraniectomy + durotomy | Primary outcome: % with mRS ≤ 3 at 6 months | ● Terminated early (slow recruitment and need for joint analysis with the other 2 European trials) | ● mRS ≤ 3 at 6 months: 25% vs 5.6% (P = .18) |
| ● Randomised | ● < 55 years | ● < 6 hours after randomisation | Secondary outcomes: | ● mRS ≤ 3 at 12 months: 50% vs 22.2% (P = .10) | |||
| ● Controlled | ● > 50% MCA territory | ● < 30 hours after stroke onset | ● Survival | ● 52.8% reduction in mortality (P < .0001) | |||
| ● Open | ● Volume > 145 cm3 | ● mRS ≤ 3 | |||||
| ● Blinded raters | ● > 1 point on NIHSS item 1a | ● Barthel Index > 85 | |||||
| ● NIHSS | |||||||
| ● QoL (SIS 2.0) at 12 months | |||||||
| DESTINY8 (2007) | ● Prospective | ● CDHC + medical treatment vs medical treatment only | ● 30 patients | ● Hemicraniectomy ≥ 12 cm (including frontal, temporal, parietal, and occipital bones) + durotomy | Primary outcome: % with mRS ≤ 3 at 6 and at 12 months | ● Terminated early (due to results of the pooled analysis) | mRS ≤ 3 at 6 months: 47% vs 27% (P = .23) |
| ● Randomised | ● < 60 years | ● Osmolality 315-320 mOsm, mechanical ventilation, blood pressure control, maintenance of normothermia and normoglycaemia | Secondary outcomes: | ● mRS ≤ 3 at 12 months: 50% vs 22.2% (P = .10) | |||
| ● Controlled | ● > 2/3 MCA territory | ● < 6 hours after randomisation | ● mRS ≤ 4 | Survival at 30 days: 88% vs 47% (P = .02). | |||
| ● Open | ● > 1 point on NIHSS item 1a | ● 12-30 hours after stroke onset | ● Barthel Index at 6 and at 12 months | ||||
| ● Mortality at 30 days | |||||||
| HAMLET7 (2009) | ● Prospective | ● HCDHC + medical treatment vs medical treatment only | ● 64 patients | ● Hemicraniectomy ≥ 12 cm (including frontal, temporal, parietal, and occipital bones) + durotomy | Primary outcome: % with mRS ≤ 3 at 12 months | ● Terminated early (data monitoring committee judged that there would be no significant differences with the pre-established sample size) | ● No effect on primary endpoint |
| ● Randomised | ● < 60 years | ● Osmolality 315-320 mOsm; mechanical ventilation; blood pressure control; maintenance of normothermia, normoglycaemia, and normovolaemia | Secondary outcomes: | ● Reduction in mortality (38% [95% CI, 15-60]) | |||
| ● Controlled | ● > 2/3 MCA territory | ● < 3 hours after randomisation | ● mRS ≤ 4 at 12 months | ||||
| ● Open | ● Evidence of mass effect | ● < 96 hours after stroke onset | ● BI at 12 months | ||||
| ● Altered level of consciousness (GCS < 13) | ● Mortality at 12 months | ||||||
| ● Depressive symptoms (MADRS) at 12 months | |||||||
| ● QoL (SF-36 and VAS) at 12 months | |||||||
| DEMITUR (2009) | Prospective | ● DHC + medical treatment vs medical treatment only | ● 151 patients | ● Hemicraniectomy | Primary outcome: % with mRS ≤ 3 at 6 and at 12 months | Not published | Not published |
| ● Randomised | ● 40–80 years | ● < 48 hours after stroke onset | Secondary outcomes: | ||||
| ● ● mRS ≤ 4 at 12 months | |||||||
| ● ● Barthel Index > 85 at 6 and at 12 months | |||||||
| ● NIHSS at 6 and at 12 months | |||||||
| ● QoL (SIS 2.0) at 6 and at 12 months | |||||||
| Slezins et al.11 (2012) | ● Prospective | ● DHC + medical treatment vs medical treatment only | ● 28 patients | ● Hemicraniectomy ≥ 12 cm (including frontal, temporal, parietal, and occipital bones) + durotomy | ● mRS ≤ 4 at 12 months | ● No differences in NIHSS or GCS | |
| ● Randomised | ● > 18 years | ● < 48 hours after stroke onset | ● Survival with mRS ≤ 4: 45% in craniectomy group vs 0% in medical treatment group | ||||
| ● Controlled | ● ≥ 50% MCA territory or volume > 145 cm3 | ● Survival: 45.5% vs 7.69% (P = .06) | |||||
| ● NIHSS > 15 | |||||||
| Zhao et al.10 (2012) | ● Prospective | ● DHC + medical treatment vs medical treatment only | ● 47 patients | ● Hemicraniectomy ≥ 12 cm (including frontal, temporal, parietal, and occipital bones) + durotomy | Primary outcome: mRS at 6 months | ● Subgroup analysis of patients aged > 60 years (range, 60-80) | Total group: |
| ● Randomised | ● ≤ 80 years | ● < 48 hours after stroke onset | Secondary outcomes: | ● Lower % with mRS > 4 at 6 months: 33.3% vs 82.6% (P = .001) | |||
| ● Controlled | ● Decreased level of consciousness (GCS eye and motor score ≤ 9) | ● Mortality at 6 and at 12 months | ● Lower mortality: | ||||
| ● Open | ● ≥ 2/3 MCA territory (with or without ACA/PCA involvement) | ● mRS at 12 months | ● 6 months: 12.5% vs 60.9% (P = .001) | ||||
| ● Blinded raters | ● Development of space-occupying oedema | ● 12 months: 16.7% vs 69.6% (P < .001) | |||||
| ● Lower % with mRS > 4 at 12 months: 25.0% vs 87% (P < .001) | |||||||
| Patients aged > 60 years: | |||||||
| ● Lower mortality: | |||||||
| ● 6 months: 12.5% vs 61.5% (P = .016) | |||||||
| ● 12 months: 18.8% vs 69.2% (P = .01) | |||||||
| ● Lower % with mRS > 4 | |||||||
| ● 6 months: 31.2% vs 92.3% (P < .002) | |||||||
| ● 12 months: 37.5% vs 100% (P < .001) | |||||||
| HeADDFIRST16 (2014) | ● Prospective | ● DHC + medical treatment vs medical treatment only | ● 26 randomised patients (75 established in study design) | ● Extensive hemicraniectomy (defined by anatomical limits) + durotomy | ● Mortality | ● Mortality | |
| ● Randomised | ● 18-75 years | ● < 96 hours after stroke onset | ● Functional status | ● 21 days: 21% vs 40% (ns) | |||
| ● NIHSS ≥ 18 | ● Caregiver burden | ● 180 days: 36% vs 40% (ns) | |||||
| ● ≥ 2 on NIHSS item 1a | ● Patient perception | ● No data for the remaining endpoints | |||||
| ● ≥ 50% MCA territory | |||||||
| DESTINY II9 (2014) | ● Prospective | DHC + medical treatment vs medical treatment only | ● 112 patients | ● Hemicraniectomy ≥ 12 cm + durotomy | Primary outcome: survival with mRS ≤ 4 at 6 months | ● mRS ≤ 4 at 6 months: 38% vs 18% (OR: 2.91; P = .04) | |
| ● Randomised | ● ≥ 61 years | ● < 48 hours after stroke onset | Secondary outcomes: | ● Mortality at 6 months: 33% vs 70% | |||
| ● Open | ● Extensive MCA infarction | ● Survival | ● No patient had mRS ≤ 2 | ||||
| ● Controlled | ● NIHSS > 14 (non-dominant hemisphere) or > 19 (dominant hemisphere) | ● NIHSS | ● All secondary endpoints showed significant improvements in the DHC group. | ||||
| ● Multicentre | ● mRS | ● Retrospective consent: 63% vs 53% | |||||
| ● Barthel Index | |||||||
| ● QoL (SF-36, EQ-5D, VAS) | |||||||
| ● Depression (HDRS) | |||||||
| ● Adverse reactions | |||||||
| ● Retrospective question (to patients, or to caregivers in patients with comprehension difficulties): In hindsight, would patients consent to the treatment received? | |||||||
| HeMMI49 (2015) | ● Prospective | DHC + medical treatment vs medical treatment only | ● 29 patients randomised (151 established in study design) | ● Hemicraniectomy ≥ 12 cm (including frontal, temporal, parietal, and occipital bones) + durotomy | Primary outcome: mRS ≤ 3 at 6 months | ● Terminated early due to slow recruitment, publication of other trials, and recommendation for pooled data analysis | ● mRS ≤ 3 at 6 months: 23.1% vs 38.4% (ns) |
| ● Randomised | ● 18–65 years | ● Osmolality 300–320 mOsm; mean arterial pressure > 90 mmHg; maintenance of normothermia and normoglycaemia | ● Mortality at 6 months: 38.5% vs 54.5% (ns) | ||||
| ● Controlled | ● GCS 6-14 (non-dominant hemisphere) or 5-9 (dominant hemisphere) | Secondary outcomes: | ● mRS < 4 at 6 months: 45.5% vs 46.2% (ns) | ||||
| ● Single-centre | ● ≥ 2 on NIHSS item 1a | ● Survival at 6 months | |||||
| ● ≥ 50% of the MCA territory | ● ● mRS< 4 at 6 months | ||||||
| Li et al.51 (2018) | DHC + medical treatment vs medical treatment only | ● > 60 years | ● < 48 hours after stroke onset | ● mRS ≤ 3 at 12 months | ● mRS ≤ 3 at 12 months: 25% vs 16.1% (ns) |
| Study | Methods | Design | Participants/selection | Intervention | Evaluation | Notes | Results (DHC vs MT) |
|---|---|---|---|---|---|---|---|
| Pooled analysis | |||||||
| Vahedi et al.48 (2007) | Pooled analysis | ● Analysis of individual data from patients from the 3 trials (DECIMAL, DESTINY, HAMLET) who met pre-established eligibility criteria and were enrolled before 1 November 2005: | ● 93 patients | ● Hemicraniectomy ≥ 12 cm (including frontal, temporal, parietal, and occipital bones) + durotomy | Primary outcome: % with mRS ≤ 4 at 12 months | ● mRS ≤ 4 at 12 months: 75% vs 24% (RR 51%) | |
| ● DECIMAL: 38 patients | ● 18-60 years | Secondary outcomes: | ● mRS ≤ 3 at 12 months: 43% vs 21% (RR 23%) | ||||
| ● DESTINY: 32 patients | ● < 48 hours after stroke onset | ● mRS ≤ 3 | ● Survival at 12 months: 78% vs 29% (RR 50%) | ||||
| ● HAMLET: 34 patients | ● Mortality at 12 months | ● NNT for mRS ≤ 4: 2 | |||||
| ● NNT for mRS ≤ 3: 4 | |||||||
| ● NNT for survival: 2 |
| Reviews | |||
|---|---|---|---|
| Cochrane71 (2012) | Cochrane methodology | Studies evaluated: | ● Surgical decompression reduced the risk of death and of death + severe disability (mRS > 4) in selected patients aged ≤ 60 years with massive hemispheric infarction and oedema. |
| ● Electronic search | ● DECIMAL (2007) | ● Given that survival is associated with significant disability, surgery should be the treatment of choice only if this is accepted. | |
| ● Search of other sources | ● DESTINY (2007) | ● As the trials were terminated early, overestimation of effect size cannot be ruled out. | |
| ● Evaluation of primary and secondary endpoints | ● HAMLET (2009) | ||
| ● Evaluation of risk of bias | Studies pending evaluation (not published at the date of review): | ||
| ● Analysis of practical implications | ● DEMITUR | ||
| ● Analysis of research implications | ● HEADDFIRST | ||
| ● Very specific conclusions | Studies underway at the time of publication of the review: | ||
| ● "Plain language” summary | ● DESTINY II | ||
| ● HeMMI |
ACA/PCA: anterior cerebral artery/posterior cerebral artery; BI: Barthel Index; DHC: decompressive hemicraniectomy; GCS: Glasgow Coma Scale; HDRS: Hamilton Montgomery-Åsberg Depression Rating Depression Rating Scale; MADRS: Scale; MCA: middle cerebral artery; mRS: modified Rankin Scale; NIHSS: National Institutes of Health Stroke Scale; NNT: Number Needed to Treat; QoL: Quality of Life; SF36: Short Form-36 Health Survey; SIS: Stroke Impact Scale; VAS: Visual Analog Scale.
In 2007, the first data were published on the efficacy of decompressive craniectomy, with combined data from the first 3 clinical trials of the treatment of malignant infarction with hemicraniectomy (HAMLET, DESTINY, and DECIMAL trials).48 The study used a pooled analysis, enabling greater statistical power, and demonstrated the superiority of the surgical procedure over the best medical treatment. The analysis included 93 patients aged between 18 and 60 years and treated within the first 48 hours after stroke. The primary outcome measure was disability at one year after stroke, classified as favourable (mRS ≤ 4) or unfavourable (> 4). Patients undergoing craniectomy showed significantly better clinical outcomes than those receiving medical treatment, both for the primary outcome measure (mRS ≤ 4 in 75% vs 24%; absolute risk reduction [ARR]: 50% [95% CI, 34%-69%]) and for secondary outcome measures: mRS ≤ 2 (43% vs 21%; ARR: 23% [95% CI, 5%-41%]) and survival (78% vs 29%; ARR: 50% [95% CI, 33%-67%]). The number needed to treat (NNT) was 2 to achieve survival with mRS ≤ 4, 4 for survival with mRS ≤ 3, and 2 to avoid a death (Table 3).
Summary of the main studies assessing the surgical treatment of malignant cerebellar infarction.
| Study | Methods | Design | Participants/selection | Intervention | Evaluation | Notes | Results |
|---|---|---|---|---|---|---|---|
| Prospective studies | |||||||
| Jauss et al.2 (1999) | ● Prospective | SOC vs VD vs medical treatment | ● 84 patients | Treatment based on clinical judgement: | mRS at 3 months: | Results by level of consciousness: | Somnolence/stupor (n = 29) |
| ● Multicentre | ● 18-80 years | ● Medical treatment | ● Somnolence/stupor | ● Surgical superior to medical treatment | |||
| ● Observational | ● Cerebellar infarction with mass effect on neuroimaging | ● VD | ● ≤ 2: good outcome | ● Coma | Coma (n = 19) | ||
| ● Non-randomised trial | ● < 5 days since onset | ● SOC + durotomy (± resection of the posterior atlas arch ± resection of necrotic tissue) | ● > 2: poor outcome | Methodological shortcomings | ● Cannot be compared between treatment groups (all underwent surgery) | ||
| ● Non-blinded raters | ● Exclusion criteria: acute coma, quadriparesis, “other indicators” of basilar occlusion, previous mRS > 2 | ● Unclear conclusions and contradictions between abstract and results | ● Good progression in 50% (all underwent surgery) | ||||
| ● Levels of significance not specified | |||||||
| Case-control studies | |||||||
| Kim et al.22 (2016) | ● Retrospective analysis | Preventive SOC vs non-preventive SOC | ● 84 patients with cerebellar infarction | Preventive SOC if: | mRS at discharge and at 12 months: | All members of the control group are assumed to have received non-surgical treatment, but the authors do not specify whether patients undergoing non-preventive SOC were excluded from this group. | mRS ≤ 2 in preventive SOC (n = 28) vs controls (n = 56) |
| ● Prospectively collected database | ● GCS ≥ 9 | ● ≤ 2: favourable | Discordant mortality data at discharge and at 12 months in the control group | ● At discharge: 64.5% vs 48.2% (P = .048) | |||
| ● Multicentre | ● No clinical deterioration at 72 hours (no changes in GCS) | ● > 2: unfavourable | ● At 12 months: 66.7% vs 51% (P = .03) | ||||
| ● Cases-controls | ● Infarct volume ratio 0.25-0.33 | Mortality in preventive SOC (n = 28) vs controls (n = 56) | |||||
| ● Raters blinded | Control group: selected by propensity score matching | ● At discharge: 1/28 vs 5/56 | |||||
| ● At 12 months: 2/28 vs 4/56 | |||||||
| Retrospective study | |||||||
| Chen et al.17 (1992) | ● Retrospective | Surgical treatment (VD + SOC) | ● 11 patients with cerebellar infarction | 100% SOC (uni- or bilateral) + VD ± resection of necrotic tissue | ● GCS at 7 days after surgery | ● Candidates for surgery selected based on clinical judgement due to ineffectiveness of medical treatment with imminent fatal outcome | ● Survival: 11/11 (100%) |
| ● Single-centre | ● Barthel Index “post-discharge” (16–60 months) | ● Improvement in GCS at 7 days: 10/11 | |||||
| ● Barthel Index | |||||||
| ● 100: 2/11 | |||||||
| ● 60-95: 6/11 | |||||||
| ● < 60: 3/11 | |||||||
| Hornig et al.25 (1994) | ● Retrospective | SOC vs medical treatment | ● 52 patients with massive cerebellar infarction, defined by presence of at least one of the following: | Treatment based on clinical judgement: | mRS at discharge: | ● 3 stages of clinical deterioration as proposed by Heros et al. | ● Medical treatment (n = 16)/SOC (n = 36) |
| ● Single-centre | ● Fourth ventricle displacement | ● SOC (± VD ± tonsillectomy) | ● ≤ 2: “minor stroke” | ● Stage 1–2 | |||
| ● Obstructive hydrocephalus | ● Medical treatment | ● > 2: “major stroke” | ● No differences between treatment groups | ||||
| ● Total or partial obliteration of the basal cisterns | ● Stage 3 | ||||||
| ● Medical treatment | |||||||
| ● Independent: 0/3 | |||||||
| ● Disability: 1/3 | |||||||
| ● Death: 2/3 | |||||||
| ● Surgical treatment | |||||||
| ● Independent: 9/23 | |||||||
| ● Disability: 9/23 | |||||||
| ● Death: 5/23 | |||||||
| ● Poorer prognosis (major stroke or death; mRS > 2) | |||||||
| ● Age > 60 years (OR: 10; P = .043) | |||||||
| ● Signs of brainstem involvement (OR: 7.5; P = .08) | |||||||
| ● Stage 3 (OR: 3.5; P P = .089) | |||||||
| Mathew et al.72 (1995) | ● Retrospective analysis | Surgical treatment (VD/SOC) vs conservative treatment | ● 50 patients with cerebellar infarction | Treatment based on clinical judgement: | Functional recovery according to clinical judgement (no assessment criteria specified): | ● Selection bias: referral from different centres without established criteria (judgement of each referring centre) | ● Limitation of therapeutic effort (n = 3) |
| ● Prospectively collected database | ● Medical treatment | ● Good | ● Non-standardised assessment | ● Conservative treatment (n = 34) | |||
| ● Single-centre | ● Surgical treatment (VD/SOC) | ● Moderate disability | ● Good recovery: 33/34 | ||||
| ● Patients with haemorrhage or infarction (analysed independently) | ● Death | ● Surgical treatment (n = 13) | |||||
| ● Moderate disability: 1/13 | |||||||
| Death: 4/13 | |||||||
| Raco et al.20 (2003) | ● Retrospective | SOC vs VD vs medical treatment | ● 44 patients | Treatment based on clinical judgement: | Functional recovery at discharge based on the GOS: | Results by treatment group: | ● Medical treatment (n = 25) |
| ● Single-centre | ● 9-83 years | ● Medical treatment only | ● Good recovery | ● Medical treatment | ● Good: 20/25 | ||
| ● Observational | ● Exclusion criteria: brainstem infarction | ● VD + medical treatment | ● Moderate disability | ● VD | ● Moderate disability: 4/25 | ||
| ● Consecutive patients | ● SOC + medical treatment | ● Severe disability | ● SOC | ● Severe disability: 0/25 | |||
| ● Raters blinded | ● Vegetative state | Methods: | ● Death: 1/25 | ||||
| ● Death | ● Levels of significance not specified | ● VD (n = 13) | |||||
| ● Good: 10/13 | |||||||
| ● Moderate disability: 2/13 | |||||||
| ● Severe disability: 0/13 | |||||||
| ● Death: 1/13 | |||||||
| ● SOC (n = 4) | |||||||
| ● Good: 0/4 | |||||||
| ● Moderate disability: 1/4 | |||||||
| ● Severe disability: 1/4 | |||||||
| ● Death: 2/4 | |||||||
| Kudo et al.73 (2007) | ● Retrospective | SOC vs VD | ● 25 patients | Treatment based on clinical judgement: | Postoperative functional status (GOS) | ● Non-standardised assessment: GOS determined in the “postoperative” phase | ● VD (n = 5) |
| ● Observational | ● VD | ● Good | ● Groups established according to “first treatment” performed | ● Good: 1/5 | |||
| ● Multicentre | ● SOC + durotomy (± resection of necrotic tissue) | ● Moderate disability | ● Moderate disability: 0/5 | ||||
| ● Severe disability | ● Severe disability: 3/5 | ||||||
| ● Death | ● Death: 1/5 | ||||||
| ● SOC (n = 20) | |||||||
| ● Good: 10/20 | |||||||
| ● Moderate disability: 6/20 | |||||||
| ● Severe disability: 2/20 | |||||||
| ● Death: 2/20 | |||||||
| Jüttler et al.24 (2009) | ● Retrospective | SOC vs VD | ● 56 patients undergoing surgery | Treatment based on clinical judgement: | ● mRS “at follow-up” (median, 8.2 years): | Results by treatment group: | ● No differences in survival or mRS scores (either at discharge or “at follow-up”) between treatment groups |
| ● Single-centre | ● ≥ 4 on Jauss scale | ● VD | 3 cut-off scores: | ● VD | ● Age and mRS at discharge are independent predictors of functional progression (mRS ≤ 3) | ||
| ● Observational | ● Exclusion criteria: basilar occlusion | ● SOC + durotomy (± resection of the posterior atlas arch ± resection of necrotic tissue) | ● ≤ 1 vs > 1 | ● SOC | |||
| ● Consecutive patients | ● ≤ 2 vs > 2 | ||||||
| ● ≤ 3 vs > 3 | Methods: | ||||||
| ● Scale for the Assessment and Rating of Ataxia | ● Non-standardised assessment (at 8.2 years; mRS cut-off scores) | ||||||
| Mostofi et al.21 (2009) | ● Retrospective | Surgical treatment (SOC/VD) vs medical treatment | ● 53 patients | Treatment based on clinical judgement: | ● GCS at 4 weeks | Results by treatment group and GCS at admission: | ● Coma |
| ● Single-centre | ● MICI: volume > 5 cm3, hydrocephalus, or brainstem compression | ● Medical treatment | ● Coma | ● Surgical treatment (n = 9): 5-point improvement on GCS | |||
| ● Observational | ● Exclusion criteria: brainstem infarction | ● SOC and/or VD | ● Surgical treatment/medical treatment | ● Medical treatment (n = 6): 2-point deterioration on GCS | |||
| ● Non-coma | ● Non-coma | ||||||
| ● Surgical treatment/medical treatment | ● Surgical treatment (n = 16): 2-point improvement on GCS | ||||||
| Methods: | ● Medical treatment (n = 22): no change in GCS | ||||||
| ● Non-standardised assessment (GCS at 4 weeks) | |||||||
| ● Levels of significance not specified | |||||||
| Pfefferkorn et al.55 (2009) | ● Retrospective | Bilateral SOC | ● 57 patients | 100% bilateral SOC ± VD ± resection of necrotic tissue | ● mRS | ● Non-standardised assessment: performed “at the end of follow-up” (mean [SD]: 4.7 [2.9] years) | At the end of follow-up (4.7 years): |
| ● Single-centre | ● Acute space-occupying cerebellar infarction with clinical deterioration and one of the following: | ● mRS classification: | ● SF-36 compared to historical controls (healthy German population) | ● 5/52 lost to follow-up | |||
| ● Observational | ● Signs of brainstem compression | ● ≥ 4: unfavourable | ● 21/52 (40%) deceased | ||||
| ● Imminent transforaminal or transtentorial herniation | ● ≤ 3: favourable | ● 31/52 survivors | |||||
| ● Obstructive hydrocephalus | ● SF-36 | ● Unfavourable (mRS ≥ 4): 4/31 (7.6%) | |||||
| ● Patient’s retrospective opinion about whether surgery was the correct decision (yes/no) | ● Favourable (mRS ≤ 3): 27/31 (52%) | ||||||
| Brainstem infarction with unfavourable progression (mRS ≥ 4): OR: 9.05 (P = .001). | |||||||
| Tsitsopoulos et al.18 (2011) | ● Retrospective | Bilateral SOC + VD + resection of necrotic tissue + resection of the posterior atlas arch | ● 10 patients | ● 100% bilateral SOC | ● GCS and RLS at discharge | ● Non-standardised assessment: performed in the “long term” (57.6 months) | At discharge: |
| ● Single-centre | ● Bilateral cerebellar infarction with: | ● “Long term” (57.6 months) mRS: | ● Improvement in GCS (preoperative vs discharge): mean (SD) of 8.9 (3.3) vs 12.6 (3.6) | ||||
| ● Observational | ● Decreased level of consciousness | ● ≥ 3: unfavourable | Long term (57.6 months): | ||||
| ● Jauss radiological criteria | ● < 3: favourable | ● Favourable (mRS < 3): 6/10 | |||||
| ● Unfavourable (mRS ≥ 3): 4/10 | |||||||
| All patients with unfavourable prognosis presented brainstem infarction | |||||||
| Tsitsopoulos et al.19 (2011) | ● Retrospective | Bilateral SOC + VD + resection of necrotic tissue ± resection of the posterior atlas arch | ● 32 patients | ● 100% bilateral SOC | ● GCS and RLS at discharge | Non-standardised assessment: | At discharge: |
| ● Single-centre | ● Unilateral cerebellar infarction with: | ● mRS at 6 months and in the “long term” (67.5 months) | ● 6-month assessment partially deduced from interview questions about estimation of functional status at 6 months | ● Improvement in GCS (preoperative vs discharge): 9 vs 13.6 (P < .05) | |||
| ● Observational | ● Decreased level of consciousness (GCS ≤ 13) | ● ≥ 3: unfavourable | ● In the “long term” (67.5 months) | At 6 months: | |||
| ● Plus one of the following: | ● < 3: favourable | ● Survival: 28/32 (87.5%) | |||||
| ● Compression of the fourth ventricle | ● Favourable (mRS < 3): 19/32 (59.3%) | ||||||
| ● Supratentorial herniation | Long term (67.5 months): | ||||||
| ● Obliteration of the basal cisterns | ● Survival: 22/32 (68.8%) | ||||||
| ● Individualised decision in some cases: decreased level of consciousness | ● Favourable (mRS < 3): 17/32 (77% of survivors) | ||||||
| ● Exclusion criteria: | ● Unfavourable (mRS ≥ 3): 5/32 (22.7% of survivors) | ||||||
| ● Coma (GCS 3-4) | Number of days receiving ventilation and preoperative GCS are independent predictors of unfavourable prognosis (mRS ≥ 3). | ||||||
| ● Basilar artery thrombosis | Age > 70 years is not associated with poor 6-month or long-term outcomes. | ||||||
| ● Bilateral infarction | |||||||
| ● Brainstem infarction | |||||||
| Lindeskog et al.74 (2019) | ● Retrospective | SOC + VD ± resection of necrotic tissue ± resection of the atlas arch | ● 22 patients | ● 100% SOC | ● mRS classification: | ● Functional assessment at 12 months | At 12 months: |
| ● Single-centre | ● Cerebellar infarction with oedema and: | ● Candidates for surgery selected according to clinical judgement | ● ≥ 4: unfavourable | ● Non-standardised mRS classification | ● Favourable (mRS < 4): 54% (12/22) | ||
| Observational | Decreased level of consciousness (GCS ≤ 13) | ● < 4: favourable | ● Unfavourable (mRS ≥ 4): 46% (10/22) | ||||
| ● Previous functional independence | ● Death: 7/22 (32%) | ||||||
| ● No contraindication for surgery | Brainstem infarction and bilateral cerebellar infarction are associated with unfavourable prognosis. | ||||||
| Tartara et al.58 (2018) | ● Retrospective | Small unilateral suboccipital craniotomy* + resection of necrotic tissue + extensive CSF drainage | ● 11 patients | ● 100% treated surgically | ● GCS score at discharge | At discharge (preoperative GCS vs GCS at discharge): | |
| ● Multicentre | * Defined as smaller than the dimensions of the infarcted tissue. | ● Exclusion criteria: | ● mRS at 6 months: | ● Mean (SD) of 9.27 (2.05) vs 14.9 (0.32) | |||
| ● Observational | ● Bilateral infarction | ● > 2: unfavourable | At 6 months (mRS rating): | ||||
| ● Brainstem infarction | ● ≤ 2: favourable | ● Favourable: 9/11 | |||||
| ● Excellent (mRS ≤ 1): 4/11 | |||||||
| ● Unfavourable: 2/ 11 | |||||||
| ● Death: 1/11 | |||||||
| Prognostic factors: | |||||||
| ● Patients with favourable prognosis were significantly younger (62.11 [7.64] vs 76.5 [4.95] years; P < .05). | |||||||
| Hernández-Durán et al.59 (2020) | ● Retrospective | Necrosectomy only* | ● 34 patients | ● 100% treated surgically | ● GOS, mRS, and Barthel Index at discharge and at 30 days | ● Subgroup analysis of results by age group (≤ 60 vs > 60 years) | Prognosis: |
| ● Single-centre | * Microsurgical resection via a burr-hole. | ● Inclusion criteria: cerebellar infarction with mass effect and: | ● Favourable | ● Favourable: 26/34 (76%) | |||
| ● Observational | ● > 3 cm diameter (regardless of clinical status) in patients with basal cistern effacement or obliteration of the fourth ventricle | ● GOS ≥ 4 | ● Unfavourable: 8/34 (24%) | ||||
| ● Consecutive patients | ● > 3 cm diameter and clinical deterioration due to brainstem compression or basal cistern effacement or obliteration of the fourth ventricle | ● mRS ≤ 2 | ● Death: 7/21 (21%) | ||||
| ● Exclusion criteria: | ● Barthel Index > 90 | No differences between age groups in mortality or functional status (GOS, mRS, Barthel Index) | |||||
| ● Non-reactive pupils > 60 minutes | ● Unfavourable | ||||||
| ● Brainstem infarction | ● GOS ≤ 3 | ||||||
| ● Basilar occlusion not susceptible to recanalisation | ● mRS ≥ 3 | ||||||
| ● Advance healthcare directives | ● Barthel Index ≤ 85 | ||||||
| Kim et al.60 (2023) | ● Retrospective | Navigation-guided burr-hole aspiration | ● 14 patients | ● 100% treated surgically | ● Analysis of the “effectiveness” of the surgical technique | ● Study lacks a control group or historical data with other surgical techniques that may enable comparison and support the conclusions drawn. | The authors conclude that the technique is safe, less invasive than other surgical alternatives, and presents shorter operation times (time from skin incision to catheter insertion of 28 minutes). |
| ● Single-centre | ● Inclusion criteria: acute cerebellar infarction treated with navigation-guided burr-hole aspiration | ● Analysis: | No complications related to the technique are reported. | ||||
| ● Observational | ● Pre- and postoperative GCS | ||||||
| ● Infarct volume before and after surgery and at 7 days | |||||||
| ● Postoperative complications | |||||||
| Hernández-Durán et al.57 (2024) | ● Retrospective | Suboccipital decompressive craniectomy vs suboccipital craniotomy + necrosectomy | ● 92 patients | ● 49 patients: necrosectomy | ● Primary: percentage of patients with favourable outcomes (mRS 0-3) at discharge and at 3 months | ● VD alone (n = 4) is not included in the analysis. | Patients undergoing necrosectomy present higher rates of favourable outcomes at discharge than those undergoing SOC only (65.3% vs 27.9%; P < .01). |
| ● Multicentre | ● Inclusion criteria: patients undergoing surgery to treat space-occu(2023)pying cerebellar infarcts, from 5 tertiary-level hospitals in Germany | ● 43 patients: SOC | ● Secondary: postoperative complications, mortality, and hospitalisation time | This difference was also observed at 3 months (65.3% vs 41.7%; P = .03). | |||
| ● Observational | No significant differences were observed in rates of postoperative complications, infection, or CSF leaks. | ||||||
| Won et al.53 (2024) | ● Retrospective | Surgical vs conservative treatment | ● 142 patients | ● 50% (n = 71) received surgical treatment (posterior fossa decompression: SOC or craniotomy with necrosectomy). | ● Primary outcome: percentage of patients with favourable outcomes (mRS 0-3) “at follow-up” | ● Patients undergoing VD alone were included in the medical treatment group. | Overall, surgery was not associated with better prognosis than medical treatment. |
| ● Multicentre | ● Inclusion criteria: patients with a diagnosis at discharge of cerebellar infarction after radiological confirmation, from 5 tertiary-level hospitals in Germany | ● 50% (n = 71) received conservative treatment (including ventriculostomy via external ventricular drain placement). | ● Secondary outcomes: | ● Raters not blinded | When results were stratified by infarct volume: | ||
| ● Cohorts | ● Exclusion criteria: concurrent supratentorial infarction, radiologically | ● Mortality at discharge and at 12 months | ● Patients with brainstem infarction not excluded | ● Surgical treatment was superior to medical treatment for larger infarcts (≥ 35 mL). | |||
| ● Observational | inapparent stroke, insufficient clinical or radiological data | ● Identification of cut-off values associated with infarct volume | ● Conservative treatment was superior to surgical treatment for smaller infarcts (< 25 mL). | ||||
| ● 531 patients selected | |||||||
| ● After propensity score matching, 71 patients were included in each group. | |||||||
GCS: Glasgow Coma Scale; GOS: Glasgow Outcome Scale; mRS: modified Rankin Scale; RLS: Reaction Level Scale; SOC: suboccipital craniotomy; VD: ventricular drain.
Subsequently, in 2012, Slezins et al.11 published a randomised clinical trial of 28 patients aged > 18 years with malignant infarction of < 48 hours’ progression, of whom 11 underwent craniectomy and 13 received medical treatment. Survival was 45% in the craniectomy group and 7.7% in the medical treatment group (P = .06), with mRS ≤ 4 in 100% of survivors in the former group and 0% in the latter.
The same year, Zhao et al.10 published a randomised clinical trial of 47 patients aged ≤ 80 years and treated in the first 48 hours after stroke, reporting similar results to those of the 2007 pooled analysis. The primary outcome measure was 6-month disability as measured by the mRS, with scores ≤ 4 considered to indicate good functional outcomes; this was achieved in 66.7% of the craniectomy group and 17.4% of the medical treatment group (P < .001), with an ARR of 49.3% and a NNT of 2. At 12 months, ARR was 62% for disability and 52.9% for mortality. ARR for mRS < 4 was 16.5% (NNT: 6).
The HeADDFIRST trial included 66 patients aged 18-75 years with malignant infarction of less than 96 hours’ progression.16 Of these 66 patients, only 25 met criteria for randomisation; 10 received medical treatment and 15 received surgical treatment. After withdrawal of one patient from the surgical treatment group, data from the remaining 24 were analysed, with no significant differences in the primary outcome measure (mortality at 21 days), with 40% (4/10) in the medical group and 21% (3/14) in the surgical group. Six-month mortality rates were 40% in the medical group and 36% in the surgical group (not significant). Due to the low mortality rate in the medical group as compared to other trials, the results are less comparable to other studies. These differences may partly be explained by older age of the sample, the longer therapeutic window, and differences in inclusion criteria (eg, midline shift ≥ 7.5 mm and the lack of requirement for impaired level of consciousness).
Finally, the HeMMI trial, published in 2015, was terminated early due to slow recruitment and the publication of other clinical trials. At 6 months, the researchers observed mortality rates of 38.5% and 54.5% in the surgical and in the medical groups, respectively (not significant), with disability (mRS ≤ 4) in 45.5% and 46.2%, respectively (not significant).49
Several meta-analyses have evaluated the efficacy of decompressive craniectomy to treat malignant infarction. The meta-analysis by Alexander et al.1 included 7 randomised clinical trials with low risk of bias. The results showed that craniectomy was associated with a 39% reduction in the risk of death, and increases of 13% in the number of patients with mRS scores of 2-3, 22% in the number of patients with mRS scores of 4, and 4% in the number of patients with mRS scores of 5 (P < .001 in all cases).
Regarding prognostic factors, an observational study of 96 patients undergoing craniectomy found that baseline infarct volume of 270 cm3 predicted poor functional outcomes with sensitivity of 81% and specificity of 74%. However, these data have not since been validated.50
Hemicraniectomy in patients older than 60 yearsThe initial evidence on decompressive craniectomy was demonstrated in patients younger than 60 years. Subsequent clinical trials used broader age ranges, or specifically included patients older than 60 years.
The clinical trial with the largest number of patients of advanced age is the DESTINY II trial, published in 2014, which included 112 patients aged > 60 years (range, 61-82) with malignant infarction within 48 hours of stroke onset.15 The primary endpoint was survival without severe disability (mRS ≤ 4) at 6 months, which was observed in 38% of the craniectomy group and 18% of the control group (P = .04). This difference was mainly due to lower mortality in the surgical group (33%, vs 70% in the control group). However, no patient scored 0-2 on the mRS; 7% of patients treated with craniectomy and 3% of controls scored 3; and the majority of patients were dependent, with mRS scores of 4 (32% of the craniectomy group vs 15% of controls) or 5 (28% vs 13%). At 12 months, a larger percentage of patients from the craniectomy group had mRS scores of 4-5 versus 6 (P < .001); however, scores of 0-55 on the Barthel Index were observed in 89% of survivors from the craniectomy group, versus 62% in the control group (P = .002).
The trial by Li et al.51 specifically included patients aged > 60 years with progression times < 48 hours; however, no differences were observed in the primary outcome measure (mRS ≤ 3 at 12 months: 25% in the surgery group and 16.1% in the control group; not significant).
The trial by Zhao et al.10 included patients aged ≤ 80 years, and reports subgroup analyses of patients aged > 60 years. Of the 47 patients enrolled, 29 were aged > 60 years (16 underwent craniectomy and 13 received medical treatment). In these patients, mortality at 6 and 12 months was lower in the surgical group (6 months: 12.5% vs 61.5%; P = .016; 12 months: 18.8% vs 69.2%; P = .01). The percentage of patients with mRS > 4 was also lower in the surgical group (6 months: 31.2% vs 92.3%; P < .002; 12 months: 37.5% vs 100%; P = .001). However, the likelihood of having mRS scores of 4-6 was only 12.5% lower in the surgical group (87.5% vs 100%; NNT: 8; P = .048).
The HeADDFIRST trial included patients aged up to 75 years, but does not report subgroup data for patients older than 60 years.16
Only limited conclusions may be drawn from the trial by Slezins et al.11 (patients aged > 18 years; range, 49-81) due to its small sample size. Of the 28 patients randomised, only 5 from the craniectomy group and one from the medical treatment group survived. All patients aged > 60 years in the craniectomy group died during follow-up.
A recent meta-analysis confirmed that craniectomy is associated with greater likelihood of survival in patients aged > 60 years (risk ratio: 1.93 [95% CI, 1.45-2.59]; P < .001).52
Surgical techniqueWith regard to the surgical technique, most clinical trials of decompressive craniectomy report large craniectomies (diameter ≥ 12 cm) with durotomy (Table 2).
Recommendations- •
Decompressive craniectomy is indicated in patients aged > 60 years meeting criteria for malignant MCA infarction, given that this treatment increases the likelihood of survival and reduces the rate of severe disability; however, it is associated with a high likelihood of moderate sequelae (mRS > 3 in 40%-50%). Therefore, it is advisable to provide patients and their families with detailed information and assess their expectations (level of evidence A, grade I recommendation).
- •
In patients aged 60-75 years who meet criteria for malignant MCA infarction, decompressive craniectomy decreases mortality rates but is associated with severe sequelae; therefore, it is not recommended for routine use, but may be indicated in very select cases (level of evidence B, grade IIb recommendation).
- •
Craniectomies performed to treat malignant MCA infarction should be large (≥ 12 cm), and durotomy should also be performed (level of evidence A, grade I recommendation).
We identified no randomised clinical trial comparing suboccipital decompressive craniectomy against medical treatment (Table 3). The great majority of the articles identified report observational, retrospective, single-centre or collaborative studies; we identified 2 case-control studies with controls assigned by propensity score matching,21,53 one prospective study,2 and one meta-analysis.54
Patients with acute onset of coma, quadriparesis,2 imaging evidence of established brainstem infarction, or basilar artery occlusion are excluded from the different studies. These markers have been considered as exclusion criteria for surgery3,18,21 or predictors of unfavourable outcomes.18,22,25,55
Studies consistently suggest that patients undergoing decompressive surgery present lower mortality and functional improvements. However, due to the designs of the different studies, we are unable to draw robust conclusions.
In the only prospective multicentre study published,2 patients were non-randomly assigned to 3 treatment groups: medical, ventriculostomy, and craniotomy with evacuation. Patients were allocated to each group according to level of consciousness, CT evidence of mass effect, or brainstem signs, which constitutes a limitation for comparing results across treatment groups. The authors also analyse the results in 3 subgroups according to level of consciousness after clinical deterioration: awake/drowsy, somnolent/stupor, and coma. The authors note in their conclusions that the surgical treatment was not superior to medical treatment in the first 2 groups. Of the patients with coma or progressing to coma who were treated with craniotomy, 50% presented good functional outcomes (mRS ≤ 2).
The only case-control study available is that published by Kim et al.22 This was a retrospective study with propensity score matching. The main objective was to analyse the role of preventive decompressive craniectomy, which was performed before clinical deterioration and indicated according to clinical and neuroimaging criteria (presence of cerebellar infarction, initial GCS ≥ 9, lack of clinical deterioration in the first 72 hours, and infarct volume ratio of 0.25-0.33). The surgical group showed better functional status (mRS ≤ 2) at discharge (P = .048) and at 12 months (P = .03), and lower rates of mortality at 12 months (P < .05). Preventive suboccipital decompressive craniectomy (OR: 4.815; P = .009) and absence of brainstem infarction (OR: 2.862; P = .033) were independently associated with good functional prognosis.
In a meta-analysis of 11 studies, including a total of 283 patients meeting pre-defined inclusion criteria, Ayling et al.54 analysed the percentage of patients with moderate-severe disability (mRS of 3-5) and rates of mortality and severe adverse events after suboccipital decompressive craniectomy in the treatment of cerebellar infarction. They reported rates of moderate-severe disability of 28%, mortality of 19.9%, a combined rate of unfavourable outcome and death of 48%, and a rate of adverse reactions related to decompressive surgery of 22.9%. Although these data indicate high overall rates of morbidity and mortality, the procedure is considered an acceptable treatment option, given the natural course of the disease, with mortality rates estimated at up to 80%.
Despite a lack of statistically significant results and broad, overlapping confidence intervals in the sensitivity analysis, the variables associated with lower risk of unfavourable progression were craniectomy performance before 48 hours and preoperative GCS score ≥ 9. Variables associated with lower mortality were preoperative GCS score ≥ 9, concomitant external ventricular drain insertion, and debridement of infarcted tissue.
More recently, Won et al.53 published the results of another retrospective, multicentre, case-control study with propensity score matching, performed at 5 tertiary-level hospitals in Germany, analysing functional outcome at one year as determined with the mRS (favourable: mRS of 0-3; unfavourable: mRS of 4-6). The study design has 2 important differences with regard to other studies. Firstly, the conservative treatment group (71 patients in the final analysis) included patients undergoing ventricular drain placement, whereas the surgical group only included patients undergoing suboccipital craniectomy with or without resection of necrotic tissue. Secondly, the authors did not exclude patients with established brainstem infarction, who accounted for 21.1% of patients (n = 15) in each group.
No differences between groups were observed for the primary outcome (percentage of patients with favourable functional outcomes in each group). In the secondary analysis (probability of favourable functional outcomes, as a function of infarct volume, GCS at admission, and treatment modality), patients with larger infarct volume (≥ 35 mL) showed better outcomes at one year (mRS of 0-33) with surgical treatment than with conservative treatment (38 [61%] vs 3 [25%]; OR: 4.8; P = .03). On the other hand, conservative treatment was associated with better outcomes at one year in patients with infarct volume < 25 mL (2 [34%] vs 218 [74%]; OR: 0.2; P = .047). No differences were observed between groups in patients with infarct volumes of 25-35 mL.
The authors conclude that overall, surgical treatment seems to be beneficial in patients with larger infarct volume (≥ 35 mL), whereas conservative treatment seems to be beneficial in patients with smaller infarcts (< 25 mL).
Regarding surgical technique, several studies analyse ventricular drainage alone and suboccipital decompressive craniectomy.2,3,20,24,25 Few studies enable comparison of results between the 2 techniques; different authors report a lack of difference between the techniques,24 better functional outcomes and lower mortality rates in the decompressive craniectomy group,55 and better prognosis in the ventricular drain group.20 However, given the low overall number of patients assigned to undergo ventricular drain placement only, the differences in patients’ preoperative status, and the overlap between treatment groups, robust conclusions cannot be drawn. As a more practical method, numerous authors suggest a stratified, rather than exclusive, approach, in which both techniques could be applied sequentially as a function of clinical presentation and the predominance of signs of hydrocephalus or brainstem compression on neuroimaging studies.20,56 Thus, some authors propose therapeutic algorithms based on level of consciousness, neuroimaging findings, and the clinical response observed after ventricular drain insertion, which help decide whether to proceed directly or sequentially to craniectomy.
Hernández-Durán et al.57 compare suboccipital decompressive craniectomy to suboccipital craniotomy with concurrent necrosectomy in a retrospective, multicentre study conducted at 5 tertiary-level hospitals in Germany. The primary outcome measure was functional status (favourable: mRS scores 0-3; unfavourable: mRS scores 4-6) at discharge and at 3 months. The study included 92 patients (49 underwent necrosectomy and 43 underwent craniectomy only). Patients in the necrosectomy group showed a higher rate of favourable outcomes at discharge than those in the craniectomy only group (63% vs 27.9%; OR: 4.9; P < .001). This difference persisted at 3 months (65.3% vs 41.7%; OR: 2.7; P = .03). No differences were observed in rates of mortality or postoperative complications.
Recent studies have proposed less invasive techniques, such as small unilateral suboccipital craniotomy (defined as less than the dimensions of the infarcted tissue),58 and necrosectomy only,59,60 performed with microsurgical techniques or burr-hole aspiration, both of which are associated with good functional outcomes in their respective trials.
Recommendation- •
Suboccipital decompressive craniectomy should be considered either as a primary indication or after failure of a ventricular drain, in patients with extensive cerebellar infarction and neurological deterioration due to brainstem compression (level of evidence B, class I recommendation).
- •
External ventricular drain placement should be considered in patients with obstructive hydrocephalus secondary to cerebellar infarction (level of evidence B, class I recommendation).
- •
Craniotomy with concurrent necrosectomy may lead to better functional outcomes (level of evidence C, class IIa recommendation).
The majority of patients in clinical trials of decompressive craniectomy were treated within 48 hours of symptom onset (Table 2), with wider therapeutic windows in a lower percentage of patients.
The DECIMAL and DESTINY trials included patients treated in the first 30 hours after onset (n = 38 and n = 30, respectively).6,61 The third clinical trial establishing evidence on craniectomy was the HAMLET trial (n = 64), which included patients treated within a therapeutic window of up to 96 hours after stroke onset; 25 patients (39%) were treated beyond 48 hours of progression. In the analysis of pre-established subgroups of patients treated < 48 and > 48 hours after onset, the former group showed a significant reduction in the absolute risk of death or mRS scores 5-6 (59% and 30%, respectively). On the other hand, the group treated beyond 48 hours after onset showed no significant differences with respect to the control group (ARR of 8% for both outcome measures; not significant).
The 2007 pooled analysis of data from the 3 clinical trials (HAMLET, DECIMAL, and DESTINY) only included patients treated within 48 hours of symptom onset, with the results described above (NNT of 2 for survival with mRS ≤ 4 and 4 for survival with mRS ≤ 3).48 These data were later supported by a meta-analysis.1,52
Subsequently, several studies and clinical trials evaluated patients treated with therapeutic windows extended beyond 48 hours (up to 96-100 hours).
The HeADDFIRST trial included 26 patients treated up to 96 hours after stroke onset, with progression times > 48 hours in 14.16 No significant differences in the primary outcome variable were observed overall, and the study did not include a subgroup analysis of patients treated beyond 48 hours.
Dasenbrock et al.62 analysed a national database of patients treated between 2002 and 2011 in the United States, with 1301 patients undergoing craniectomy; this is the largest study published to date. Surgery was performed beyond 48 hours after symptom onset in 44.2% of cases. Time of surgery was not associated with in-hospital mortality, but longer delays were associated with greater likelihood of dependence (OR: 1.12 [95% CI, 1.03-1.23]; P = .02). Similarly, surgery performed after more than 72 hours was associated with greater dependency than earlier procedures (OR: 1.52 [1.07-2.16]; P = .02), but not with higher mortality.
A meta-analysis published in 2020 analyses the clinical outcomes of decompressive craniectomy as a function of the timing of the intervention,63 comparing data from patients treated before and after 48 hours of progression. In 7 studies reporting a total of 1508 patients treated with craniectomy, procedures performed beyond 48 hours after stroke onset were not associated with greater risk of dependence or death (mRS 4-6), with an OR of 1.1 (95% CI, 0.89-1.38). Signs of cerebral herniation and absent pupillary reflex are associated with poorer prognosis.
The EDEMA scale was proposed with a view to predicting which patients are at risk of developing malignant infarction, and facilitating triage for monitoring and potential surgery.64 The scale includes basal cistern effacement, glycaemia > 15 mg/dL, fibrinolysis and/or thrombectomy, midline shift, and history of stroke; a cut-off score of > 7 presents an area under the curve of 0.76, with a positive predictive value of 93%. This scale has recently been validated.65
Recommendations- •
When decompressive craniectomy is indicated in patients with malignant MCA infarction, the procedure should be performed within 48 hours of stroke onset (level of evidence A, grade IIa recommendation).
- •
Decompressive craniectomy may be indicated in patients with malignant MCA infarction with progression times of 48-96 hours, if they meet criteria for this indication and present no signs of cerebral herniation, such as absent pupillary reflex (level of evidence B, grade IIb recommendation).
Neurological deterioration secondary to oedema due to cerebellar infarction shows an incidence peak on days 2-4 of progression,2,22,25 and may be followed by rapid unfavourable progression, developing coma within 24 hours, according to some authors.25 Furthermore, better functional prognosis is reported in patients undergoing surgery within 48 hours54 and even 24 hours,18 in patients receiving surgical treatment before progression to coma,2 and in patients with higher GCS scores.18,54
Kim et al.22 even recommend considering “preventive” surgery before clinical deterioration is established, although this recommendation is restricted to patients without brainstem infarction and with an infarct volume ratio of 0.25-0.33.
Taylor et al.66 also consider intervening prior to the clinical deterioration: seeking to identify objective parameters or measures that would enable early identification of patients at high risk of clinical deterioration and needing surgery, they conducted a volumetric analysis of the baseline CT studies of 86 patients with cerebellar infarction, retrospectively analysing the volumes of the infarct, the cerebellum, and the posterior fossa in each patient, and designed a scale (Cerebellar Ischaemia Stroke Score). Scores were significantly associated with subsequent need for neurosurgical interventions.
Therefore, despite the limitations inherent to retrospective study designs, these results support early surgery, prior to the onset of advanced neurological deterioration, based on neuroimaging findings and the detection of early changes in neurological status. Although a cut-off point of 48 hours’ progression time has been established for the purpose of analysing prognosis, this limit is arbitrary and cannot be extrapolated to individual patients.
Recommendation- •
Therapeutic window for suboccipital decompressive craniectomy. Although it is not possible to establish a precise therapeutic window, surgery should be performed early, preferably within 48 hours of onset, according to the detection of early changes in neurological status and neuroimaging findings, and before onset of advanced clinical deterioration or coma (level of evidence B, class I recommendation).
A meta-analysis of the initial clinical trials demonstrating the efficacy of hemicraniectomy in malignant infarction included a pre-defined subgroup analysis to identify variables that may influence or modify the efficacy and/or safety of the procedure.48 Although there was not a specific analysis comparing left- and right-sided stroke, the study does analyse the presence or absence of aphasia as a predictor of disability (mRS ≤ 4 at one year after stroke). In both cases, surgery presented very favourable results compared to medical treatment, with no inconsistencies between trials. Patients with aphasia presented an ARR of 44.2% (OR: 0.14 [95% CI, 0.04-0.50]; P < .001) for disability at one year, compared to 58.2% (OR: 0.06 [0.01-0.31]; P < .001) in patients without aphasia. A recent meta-analysis comparing mortality and disability in left and right malignant MCA infarction67 found no association between the hemisphere affected and short-term (OR: 1.00 [0.69-1.45]) or long-term progression (OR: 1.01 [0.767-1.33]), concluding that the available evidence does not support ruling out craniectomy based on stroke laterality.
Regarding quality of life after malignant infarction and craniectomy of the affected hemisphere, the majority of studies present poor methodological quality, limiting the robustness of their conclusions. Van Middelaar et al.68 conducted a systematic review on quality of life in patients undergoing decompressive craniectomy after malignant infarction, reviewing 17 studies with a total of 459 patients, including 3 randomised clinical trials. However, only 5 studies (reporting 81 patients) assessed the impact of the hemisphere affected. Four of these 5 studies found no significant differences in quality of life as a function of the hemisphere affected; one study found poorer quality of life in patients with right-hemisphere infarctions.69 Furthermore, 4 of these studies were unable to adequately assess quality of life in some patients with aphasia, requiring the participation of family members to evaluate quality of life. Quality of life data were similar in observational studies and in the 3 randomised clinical trials analysed.6,16,70
Recommendation- •
There is no evidence that the laterality of malignant MCA infarction significantly influences functional outcomes after decompressive craniectomy (level of evidence C, grade IIb recommendation).
This study has received no funding.
The authors have no conflicts of interest to declare.
The following is Supplementary data to this article:
