Surgical treatment of hemorrhagic brainstem cavernous malformations
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.193825
Source of Support: None, Conflict of Interest: None
Context: Microsurgery is considered to be the optimal treatment for brainstem cavernous malformations (BCMs); however, the high surgery-related morbidity requires further assessment of therapeutic protocols.
Keywords: Brainstem; cavernous malformation; microsurgery
Cavernous malformations (CMs) of the central nervous system (CNS) are rare lesions that are pathologically defined as clusters of dilated sinusoidal channels surrounded by a single layer of endothelium without intervening brain parenchyma. Although most of the CMs are located in supratentorial areas, approximately 20% arise in the brainstem and may be associated with a higher hemorrhagic rate.,,,, The optimal treatment of CMs is microsurgical gross total resection, and the general prognosis is favourable. However, due to the proximity of the lesion to critical neural structures and the complex blood supply existing in the region, a high surgery-related morbidity may occur that requires further focus.,,,,,,,,,,,,, Although the primary goal of surgery is to remove the lesion to prevent further haemorrhage, rather than aim for neurological recovery, efforts on preventing neurological deficits caused by surgical intervention should be emphasized.
Previously, we published a study involving a series of 37 patients with BCMs who underwent surgery between September 2003 and August 2007. Since then, 120 patients with BCMs underwent surgical treatment in our hospital from September 2007 to August 2014. In the present study, we evaluated the clinical features and functional outcome of these patients. We have also attempted to elaborate on our institutional experience using a standardized surgical strategy. The appropriate surgical indications, optimal selection of approaches, and application of advanced supplementary techniques are also discussed.
At admission, the patients underwent a routine examination that included a clinical history and neurological examination. The modified Rankin score (mRS) was used to evaluate the neurological status. Preoperatively, cranial computer tomography (CT) and magnetic resonance imaging (MRI), including the T1-weighted, T2-weighted, and gadolinium enhanced T1-weighted sequences were performed. Diffusion tensor tractography (DTT) was utilized for patients who underwent elective surgery.
The episode of hemorrhage was defined as an acute, new-onset or worsening neurological deficit corresponding to the location of hemorrhage on CT and MRI imaging.,,
Our institutional surgical indications for brain stem hemangiomas were as follows: (1) Acute and subacute hemorrhage with severe neurological deficits and significant mass effect; (2) repeated hemorrhages significantly influencing the quality of life; (3) the lesions were found to be either exophytic or surgically accessible following the preoperative evaluation.
Surgical timing, approach, and techniques
Surgeries were preferentially performed during the subacute phase. Nevertheless, for patients with severe progressive neurological deficits, emergency surgery was also considered. The aim of surgery was gross total resection with focus on reduction in the postoperative neurological deficits. Somatosensory evoked potentials (SEP), motor evoked potentials (MEP), and brainstem auditory evoked potentials (BAEP) were monitored routinely, and the electrophysiological monitoring of the cranial nerves was performed according to location of the lesion. A stimulation probe was applied for cranial nerve nuclei and corticospinal tract mapping.
The selection of the surgical approach was tailored on an individual basis and standard skull base craniotomies were performed with the application of neuronavigation. A classic two-point method  and trans-safe entry zone approaches were advocated. In an attempt to identify the safe zones, anatomical landmarks were rechecked with the results of preoperative DTT and intraoperative monitoring. A low-power bipolar coagulation along with a sharp dissection between the lesion and the surrounding hemosiderin-stained gliotic tissue was utilized to excise the lesion. An en bloc resection and an elaborate exploration of the wall of hematoma cavity was performed for increasing the chances of complete resection. The CO2 laser at the power of 3 W was applied in the last 28 cases to incise the surface of the brainstem and shrink the lesion away from the eloquent hemosiderin-stained gliotic tissue. The presence of any developmental venous anomalies (DVA) were noted.
Follow-up and clinical outcome
The postoperative complications and new-onset or worsened neurological deficits were recorded, and mRS was assessed 3 days after the surgery. The follow-up was performed at 6 months and once in a year, thereafter. During the follow-up period, all patients underwent a neurological assessment and cranial MRI. Changes in neurological deficits (improved, stable, or aggravated) and mRS were assessed by the senior author at the outpatient center. Experts of neuroimaging processed the radiological images to assess for any evidence of recurrence or rehemorrhage. The treatment instituted following the recurrence was noted. Patients with a follow-up of less than 18 months were excluded from the outcome analysis.
Data collection and statistical analysis
The following preoperative variables were documented: Age and gender of the patient, the location of the tumor, the number of hemorrhages, the size and depth of the lesion, the presence of a developmental venous anomaly (DVA), and the preoperative mRS. Univariate and multivariate regression analysis were utilized to analyse the statistical significance between the candidate variables and a poor outcome (mRS: 3–5). The Student's t-test was used to compare continuous variables and the Pearson chi-square test was used to compare categorical variables. The result of multivariate regression analysis was assessed in terms of the odds ratio (OR). All statistical analyses were processed by the Statistical Package for the Social Sciences software (version 19.0 [IBM Corp. Armonk, New York, USA]), and P < 0.05 was considered to be statistically significant.
From September 2007 to August 2014, 120 consecutive patients with BCMs underwent surgical treatment in our hospital. The demographics and clinical features of these patients are presented in [Table 1]. There were 56 male and 64 female patients with a mean age of 40.29 ± 14.78 years (range: 4–69 years). All the lesions were divided into groups based upon their location into the mesencephalon (n = 18), pontomesencephalon (n = 16), pontine (n = 58), pontomedullary (n = 14), and medulla oblongata (n = 14) types. The most common presentation was sudden onset of cranial nerve deficits (n = 127). The other symptoms and neurological signs included ataxia (n = 45), dizziness (n = 28), headache (n = 41), hemiparesis (n = 52), hemiparesthesia (n = 44), cognitive deficits (n = 1), disturbances of consciousness (n = 3), and respiratory problems (n = 5) [Table 2]. Ten patients were treated in other hospitals before being admitted in our department The procedures included stereotactic radiotherapy in 7 and surgical resection in 3.
All the patients showed symptoms of one or more hemorrhagic episodes (58 patients presented with a single episode of hemorrhage, 43 patients with two episodes, and 19 patients with three or more episodes). Assuming that all CMs had been present since birth, these patients had experienced a total of 204 hemorrhages during 4835 cumulative years of life. Hence, the calculative annual hemorrhage rate was 4.2%. Sixty-two patients suffered from multiple hemorrhagic episodes with the interval between the hemorrhages ranging from 0.8 to 13 years, and the annual rehemorrhage rate was 42.9% (84 hemorrhages/195.8 patient-years).
Preoperative radiological findings
The mean size of the lesions was 2.04 ± 0.71cm in diameter (range: 0.5–3.6cm). A total of 110 lesions were located superficially, whereas the others were deep-seated. The DVAs were observed in 19 cases (15.8%) on the preoperative MRI.
A total of 117 patients underwent surgical treatment in the subacute phase; however, 3 patients presenting with life-threatening symptoms were taken up for emergency surgery. The following approaches were adopted [Table 3] and [Figure 1]: Subtemporal supratentorial (n = 7), subtemporal transtentorial (n = 7), supracerebellar infratentorial (n = 15), orbitozygomatic (n = 2), presigmoid (n = 4), retrosigmoid (n = 31), suboccipital (n = 41, including telovelar in 4, transvermis in 27, and telovelar associated with splitting the inferior part of the vermis in 7), and far lateral (n = 11) [Figure 1]. Gross total resection was achieved in 116 cases, and subtotal resection was performed in 4 cases due to the firm consistency of the lesion and the critical structures involved [Figure 2] and [Figure 3]. Intraoperatively, DVAs were observed in 42 cases (35%), and all caput medusae were preserved carefully. Part of the distal veins of DVAs (n = 13) were observed to be tightly adherent to the lesions. These abnormal veins were considered to be responsible for drainage and recurrence of the lesion. We coagulated these abnormal veins, and neither the symptoms of venous infarction nor the abnormal presentations on postoperative neuroimaging occurred.
Postoperatively, fifty-three new-onset or worsened neurological deficits occurred [Table 4]. The most common deficits were cranial nerves deficits (n = 21), ataxia (n = 12), and hemiparesthesia (n = 11). The mean mRS at 3 days after surgery in our series was 2.73 ± 0.83.
In our series, 2 patients underwent the second surgery due to cerebellar swelling, and 2 patients suffered from postoperative hemorrhage requiring its evacuation. One patient had developed repeated postoperative hemorrhage three times during the hospitalization. During the first 2 operations, the hematoma was evacuated and no residual CM was observed. In the third operation, we observed that abnormal veins were responsible for the rehemorrhage. Therefore, we removed the hematoma as well as abnormal veins while leaving the venous caput medusae intact. As a result, the patient recovered without rehemorrhage albeit with severe neurological deficits. One patient suffered from hydrocephalus, which was probably caused by adhesions within and obstruction of the fourth ventricle; a ventriculoperitoneal shunt was performed in this patient in the postoperative period.
Other perioperative complications included cerebrospinal fluid (CSF) leak (n=1), intracranial infection (n = 3), pneumonia (n = 4), and aspiration (n = 2). A gastrostomy tube was placed as a routine procedure in 10 patients who suffered from dysphagia preoperatively; 2 patients presented with new onset of lower cranial nerves deficits. Five patients underwent a tracheostomy due to respiratory deficits or aspiration pneumonia, and 2 of them needed continuous ventilatory support for their impaired respiratory function. Two patients died. These included 1 patient with postoperative herniation caused by cerebellar haemorrhage; and another patient who suffered from impaired respiratory function preoperatively. Although emergency surgery was performed, he eventually died due to postoperative pneumonia.
Clinical outcome and risk factors
During the follow-up period, 6 patients had a follow-up of less than 18 months, and 8 patients were lost to follow-up. Therefore, data from 106 patients (88.3%) were available for outcome analysis. At a mean follow-up of 50.7 ± 26.5 months (range: 18–90 months), the neurological status showed improvement in 71 (67.0%) patients and remained stable in 24 (22.6%) [Figure 4], [Table 4] and [Table 5]. The postoperative new-onset or worsened symptoms showed improvement or had resolved in 58.5% patients, remained stable in 32.1% patients, and, had aggravated in 9.4% patients [Table 4]. The mean preoperative mRS score was 2.51 ± 0.90, while it was 2.73 ± 0.83 postoperatively, and 1.71 ± 0.98 at the last available follow-up [Table 5]. During the follow-up period, the residual lesions re-bled in 2 cases, requiring a reoperation at 18 months and 30 months after surgery. Forty patients (33%) showed permanent morbidities [Table 5], and 11 of them had new-onset permanent morbidities, including facial pain (n = 1), diplopia (n = 2), dizziness (n = 1), tremor (n = 1), numbness of limb (n = 4), and hemiparesis (n = 2).
After analysing 8 candidate variables, including age, gender, location, number of hemorrhages, size and depth of the lesion, presence of DVA, preoperative mRS, and application of CO2 laser in the univariate analysis, the pontine location, multiple hemorrhages, presence of a deep-seated lesion, and the preoperative mRS were considered to be the potential predictive factors [Table 6]. The results from the multivariate regression analysis showed that preoperative mRS was an independent predictive factor for the long-term neurological outcome (P = 0.003) [Table 7].
CMs are low-flow and low-pressure vascular malformations that may occur throughout the neuraxis. In 1928, Dandy first resected a CM located in the pontomedullary region, and since then, more than 1800 cases of BCMs have been reported in the literature.,,,,,, It is well-acknowledged that microsurgical resection is the optimal treatment for this disease and the primary goal of surgery is to remove the lesion to prevent further hemorrhage rather than to aim for neurological recovery. However, due to the involvement of critical structures and the complex blood supply in the vicinity of the lesion, the issue of preventing the neurological deficits caused by surgical intervention has always been on the forefront.
Patients with hemorrhagic BCMs typically present with a relapsing and remitting course correlating with the natural history of this disease. Literature reports on the nature of BCMs are numerous and often not in agreement with each other. Studies on patients who had received conservative treatment showed a low annual rehemorrhage rate, ranging from 4.4% to 5.1%. In contrast, rehemorrahge rate in most surgical series is variable, ranging from 15% to 60% [Table 8].,,,, In our series, hemorrhage and rehemorrhage rate were 4.2% and 42.9% per patient per year, respectively. We consider that the rehemorrhage rate reported in literature has been overestimated, and the wide variation of rehemorrhage rates, which is caused by selection bias, warrants further investigation. Moreover, there is consensus that patients have less chances of recovery and do not attain their preoperative neurological status after multiple hemorrhages. In our series, similar findings were noted and patients with a history of multiple hemorrhages suffered from a poorer preoperative neurological status. 25% of them had a poor long-term neurological outcome compared to 14% of patients with a single episode of hemorrhage. Therefore, the major goal of the surgery in these cases is to prevent further rehemorrhage, while ensuring minimal surgery-related morbidy rather than aiming for further neurological recovery.
Standardized surgical strategy
Due to the benign natural course and the high surgery-related morbidity, only patients with multiple hemorrhages or those experiencing progressive neurological deterioration would benefit from surgical treatment, and asymptomatic BCMs should not be subjected to surgical intervention.,,,,,, According to our experience, the surgical indications should be established based on the followings reasoning: (1) BCMs are reported to have a higher risk of recurrent bleeding and progressive neurological deficits.,,, (2) It is evident from the literature review that rehemorrhage significantly increases the pre-existing neurological deficits. (3) It has also been observed in our series that multiple hemorrhages made surgical dissection more difficult. (4) With the application of advanced radiological techniques, such as an intraoperative monitoring, and meticulous surgical techniques, some deep-seated lesions are no longer inaccessible. Therefore, the surgical indications for operating upon these lesions in our department are as follows: (1) There is usually evidence of acute and subacute hemorrhage with severe neurological deficits and significant mass effect. (2) Repeated hemorrhages significantly influence the quality of life and, therefore, should be considered as a major indication for intervention. (3) The lesions must be exophytic or surgically accessible after the preoperative evaluation.
The optimal surgical timing has been debated. Some authors having recommended operating upon brain stem CMs in the subacute stage,,, whereas others have stated that early surgery could promptly relieve the mass effect on brainstem nuclei and tracts, and thus improve the neurological status. We preferred performing surgery during the subacute stage because the plane of cleavage was well-defined and the lesion could be easily dissected. Nevertheless, for patients with severe and progressive neurological deficits, emergency surgery should be considered.
In general, the optimal approach is from the most damaged zone or through the zone of least functional importance so as to reduce neurological deficits caused by surgical intervention. The classic two-point method introduced by Spetzler et al., established a worldwide acceptable principle in choosing the approaches. However, in most of the cases, the cranial nerve nuclei and white matter tracts are displaced by the lesions, and therefore, anatomical landmarks and the surgeon's experience are no longer reliable parameters to guarantee a safe resection. In the present series, we applied advanced neuroimaging and electrophysiological monitoring techniques to accurately locate the critical structures, both preoperatively and intraoperatively. Therefore, we advocate that an optimal approach and a safe parenchymal entry zone should be tailored on an individual basis by integrating the results of the preoperative conventional MRI, DTT, intraoperative anatomical landmarks as well as electrophysiological findings.
Surgical techniques and DVA
The general microsurgical techniques have already been mentioned in several studies.,,, In our series, it was notable that 7 patients who had undergone stereotactic radiotherapy previously, had lesions that were found to be tightly adherent to the parenchyma during surgery, which resulted in significant difficulty during their dissection. In addition, the CO2 laser was adopted in the last 28 cases, and according to our experience, the benefits were as follows: (1) The CO2 laser could coagulate capillaries within the pia and create a sharp bloodless incision on the surface of brainstem. Therefore, the thermal injury caused by coagulation using the bipolar cautery could be avoided. (2) The application of CO2 laser allowed the lesion to be vaporized away from its hemosiderin wall in a touch-free manner, which eliminated undue traction of the critical neural tissues surrounding the lesion. However, the neurological outcome in our series was not able to demonstrate a statistical difference between the traditional techniques of lesion excision and the excision utilizing the application of CO2 laser, which is possibly due to the small sample size within the two groups, the one wehre excision was undertaken utilizing the conventional techniques and the other where CM excision was done utilizing the CO2 laser. Further investigations with a larger set of patients are necessary to unequivocally demonstrate the beneficial effects of laser.
DVAs are congenital anomalies of normal venous drainage, consisting of a number of dilated medullary veins converging into a single large draining vein, typically presenting with the caput medusae appearance.,, It has been reported that BCMs have been associated with DVAs in 16.6–100% cases [Table 8],,, and, in most studies, preservation of the associated DVA was considered a primary goal in order to avoid venous infarction or haemorrhage.,, However, from our intraoperative observation, the occurrence of an associated DVAs was not as high as has been reported (n = 42, 35% in the present series). Some abnormal veins were observed to be closely adherent to the BCMs, and were considered to be responsible for the venous drainage of the BCMs. We defined them as the distal branches of DVAs, which could be coagulated. Otherwise, trying to preserve them might pose a risk of postoperative hemorrhage. The typical appearance of caput medusae was seen in the the main trunks of DVA, which are regarded as being benign but appear as an abnormal constellations of veins draining the normal brainstem tissue. We advocate that every attempt should be made to preserve these structures.
Functional outcome and predictive factors
In the last two decades, the long-term outcome of surgical treatment of BCMs has achieved significant progress in large centers [Table 8].,,,,,,,,,,,,,,,,,,, Previous studies attributed the poor preoperative neurological status, multiple preoperative hemorrhages, age ≥50 years, and deeply-located lesions to be the risk factors for a poor functional outcome. However, in our series, the preoperative mRS was considered to be the exclusive independent predictive factor responsible for the long-term neurological outcome. By adhering to a standard surgical strategy, the patients with a favorable preoperative mRS could achieve a better neurological outcome. This ensured that the patients who had undergone successful surgery could experience neurological recovery to reach a neurological status that was closely resembling their neurological status prior to their rehemorrhage.
Surgery-related neurological deficits in BCMs could be reduced by adopting a standard surgical strategy based on appropriate surgical indications, the optimal selection of the approach, and by application of adjunctive techniques like electrophysiological monitoring and CO2 laser excision. Based on our institutional experience, we recommend that an optimal approach and safe parenchymal entry zone should be tailored according to individual patients by integrating the results of preoperative conventional MRI and DTT with the intraoperative anatomical landmarks, brain stem nuclei mapping as well as electrophysiological findings.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6], [Table 7], [Table 8]