Results

Features of classification and radiological diagnosis

We distinguished 2 main groups after analysis of CT/MRI in comparison with intraoperative data and results of biopsy:

1. Hematomas (subacute, chronic) without MR signs of vascular malformation (n=191);

2. Cavernous angiomas verified by MRI with signs of hemorrhage or without hemorrhage (n=131).

Hematomas without MR signs of vascular malformations (n=191)

Features of MRI and CT scans in this group varied depending on duration of hemorrhage. For example, subacute hemorrhage (2 weeks—1.5 months) was characterized by hyperintense CT focus and T1/T2WI scans with ring-shaped focal hyperintense signal and perifocal edema (Fig. 1a—c).

Chronic hemorrhage (1.5 months—1 year, mean 2.5 months) manifested by ring-shaped contrast accumulation on CT scans and MR (TI/T2WI) focus of uniform hyperintense signal (Fig. 2a—c).

In this group, vascular malformation was confirmed intraoperatively and after biopsy. Malformation was confirmed only in 44% of cases. In other patients (56%), malformation was not detected. Apparently, it was due to small dimensions of malformation and/or its destruction after hemorrhage. CAs were histologically confirmed only in a third of malformations. The majority was AVMs and undifferentiated vascular malformations.

Cavernous angiomas verified by MRI (n=131)

There was heterogeneously changed signal on TI and T2WIs. Isointense signal on TIWI with perifocal ring of hypointense signal on T2WI (Fig. 3a)

were less common. CT revealed inhomogeneous focus of hyperintense signal (Fig. 3b).

In case of hemorrhage, typical CA with adjacent focus of changed signal corresponding to subacute or chronic hematoma (hemorrhage outside of malformation) was detected. In another variant, area of changed signal was characterized by several (≤2—3) adjacent to each other hyperintense foci (hemorrhage into malformation) (Fig. 4a, b).

In all cases, CA was confirmed after histological examination.

Clinical manifestations

In most cases (92%), the disease was clearly manifested after hemorrhage. Neurological symptoms of hemorrhage (with extremely rare exceptions) developed on the background of intact consciousness. More than half of the patients (64%) experienced the first hemorrhage after no previous symptoms. In other patients, various precursors preceded the first hemorrhage for several years: headache, dizziness, transient neurological symptoms or combination of these signs. Presumably, these symptoms may be explained by microhemorrhage and/or thrombosis inside the malformation.

We identified two clinical types of hemorrhage depending on the nature and rate of neurological impairment:

1. Stroke-like course: acute or subacute symptoms of brain stem lesion. Neurological aggravation developed over a short period of time (up to 1 week) and was followed by further stabilization. Timing of stabilization ranged from few days to a month. Subsequent gradual regression of symptoms was followed by partial or (less often) complete remission.

2. Pseudotumorous course: slow progression of symptoms of brainstem lesion for several weeks (sometimes months). Nevertheless, stabilization and even partial regression of symptoms ensued in long-term period (after a month or later).

Structure of focal syndrome was determined by localization of lesion and included signs of damage to the nuclei of cranial nerves and pathways. Syndromes of pontine lesion prevailed due to the most frequent localization of hematomas and/or CA in this zone of the brainstem (Table 1).

Clinical course of disease was characterized by certain features in each of these groups. For example, hemorrhage was the first and single in the majority (77%) of patients with hematomas without MR signs of malformations. Large hematoma often occupying the base and tegmentum of the brain stem determined extremely severe state of most patients (mean KS score 30—50). Recurrent hemorrhages were characteristic for patients with MRI-confirmed CA. The majority of patients (92%) have suffered several hemorrhages by the moment of surgery. However, their certain limit should be emphasized. In our sample, there were 3—4 hemorrhages with clear symptoms (even in case of long duration of disease). Five recurrent hemorrhages were observed only in 1% of cases that was the maximum in our patients (Tables 2, 3).

Probably, this phenomenon is caused by thrombosis and sclerosis of malformation cavities. There were no clear symptoms of hemorrhage in 8% of patients with CA. Gradual neurological deterioration in these cases was most likely due to changes in cavernoma per se associated with impaired venous outflow and enlargement. Patients with CA were characterized by less severe neurological symptoms compared to those with hematoma. KS score less than 60—70 was rare even in acute period of hemorrhage and progressive neurological deterioration (Table 4).

Features of surgical treatment

We assessed the following factors in selecting patients for surgical treatment: clinical severity; volume of hemorrhage; number of previous hemorrhages; surgical availability of CA, i.e. its proximity to the brainstem surface. The main indications for surgery were the following:

1. Subacute or chronic brainstem hematomas accompanied by severe neurological symptoms;

2. Brainstem CA followed by hemorrhage and severe neurological deficit;

3. Surgically available (adjacent to the brainstem surface) CA outside the hemorrhage stage, but with recurrent previous hemorrhages followed by progression of symptoms of brainstem lesion.

All operations were performed under intraoperative neurophysiological monitoring (INM). It was either direct stimulation of the brainstem nuclei in approach through the 4th ventricle or transcranial motor pathway stimulation during manipulations near the brainstem. Both methods were combined in patients with extensive hematomas of the base and tegmentum of the brainstem. Somatosensory evoked potentials were used less frequently for INM due to complexity and low informative value of this method.

INM required an appropriate anesthetic management at the stages of initiation and continuation of anesthesia. During intubation, we used depolarizing muscle relaxants and moderate dose of non-depolarizing muscle relaxants. Further anesthesia maintenance was carried out by infusion of propofol (20—50 ml/h) or fentanyl (10 ml/hour).

Removal of hematoma and/or malformation was made within the closest contact of hematoma or malformation with the brainstem surface. We planned surgical approach as a line passing through the point on the brainstem surface closest to the malformation or hematoma and through their center.

Midline suboccipital craniotomy with an approach through the 4th ventricle was more common because the majority of hematomas and vascular malformations were subependymal and localized near pontine tegmentum. Dissection of rhomboid fossa was always carried out within the "safe" zones (suprafacial, infrafacial triangle, etc.). However, mapping of the fourth ventricle floor was an important stage of surgery. This procedure ensured navigation in the brainstem structures in case of severe hemorrhage-related deformation and manipulations away from the cranial nerve nuclei. It should be noted that this method is also not reliable enough, especially for facial nerve function monitoring. In our sample, early postoperative facial nerve paralysis developed in one third of cases despite preserved "M-response" in intraoperative electrical stimulation after resection of malformation. Perhaps, contraction of facial musculature was caused by stimulation of distal root, while nucleus per se and/or proximal root were intraoperatively damaged.

This approach was the most acceptable even for large hematomas occupying almost the entire brainstem cross section due to its simplicity and less traumatic nature compared to other approaches.

We have performed suboccupital craniotomy only in patient's prone position for the last 9 years in order to prevent complications such as intraoperative massive air embolism. We did not dissect the lower parts of cerebellar vermis at the intracranial stage and used telovelar approach so that the axis of surgical manipulations was as close to the optimal one as possible (i.e. passed through the above-mentioned points). This approach implied dissection of the fissure between the upper parts of amygdala and cerebellar tongue followed by separation of the vascular layer up to superior medullary velum. These manipulations ensured exposure of almost the entire half of the rhomboid fossa on the side of lesion. However, we preferred an approach from the ventrolateral surface of the brainstem if deep hematoma did not adhere to the brainstem surface, especially to the fourth ventricle floor, and symptoms of brainstem lesion were minimal. This approach was useful to avoid postoperative neurological deterioration, in particular symptoms of pontine tegmentum nuclei lesion. Far-lateral, retrolabyrinthine, presigmoid, subtemporal and anterior transpetrous approaches were predominantly used. Retrosigmoid suboccipital approach was rarer in our practice. Indeed, angle of attack of this approach did not allow dissection of the hematoma cavity or CA for total resection. Presigmoid or transmastoid retrolabyrinthine approach was more common for lesion of ventrolateral parts of the pons and medulla oblongata. Subtemporal craniotomy combined with anterior transpetrous approach was used for lesion of pontomesencephalic zone.

Subtentorial-supracerebellar or suboccipital transtentorial approach was used for removal of hematoma and/or vascular malformation of the midbrain. Pterional craniotomy and approach through the anterior parts of Sylvian fissure were applied in certain cases if hematoma or CA adhered to the brainstem surface within the interpeduncular cistern.

In all cases, selection of surgical approach was strictly individualized and determined by MRI data.

Intracranial stage of removal of hematoma or malformation implied a small incision of the brainstem either in the area of maximal lesion due to hemorrhage or CA attachment of the brain stem surface or in the area associated with minimal functional damage after incision. We used safe approaches through the brainstem described by various authors. Debridement of hematoma (liquid part and clots) was performed through this incision. In case of chronic hematomas, we tried to resect totally its capsule. In some patients, dissection of immature capsule was associated with advanced trauma of surrounding brain matter. Careful coagulation of residual capsule fragments was necessary to prevent subsequent hemorrhages. CA was fragmented and resected in pieces. It was unacceptable to resect cavernous malformation in “en bloc” fashion due to high risk of damage to surrounding brain matter. Dissection of malformation into the fragments was almost never accompanied by bleeding due to low blood flow in the cavernoma. We observed intraoperative bleeding associated with removal of hematoma or CA in only 6% of cases. This complication was caused by micro-AVM. Intensive arterial bleeding could be stopped only after complete coagulation and dissection of abnormal vessels.

A thorough coagulation of vessels in glial tissue around CA or hematoma is always considered as an important and necessary surgical stage. However, it should be remembered that CA may be combined with venous malformations in 20—30% of cases [5, 14—16]. It is believed that occlusion of the venous component of malformation can result delayed neurological symptoms due to venous infarction [9, 12]. Considering our experience, we cannot confirm this statement. In our sample, we had to coagulate veins within the malformation several times. However, severe neurological deterioration was observed only in 2 patients with lesion of tegmentum of the midbrain.

Early outcomes of surgical treatment

Early postoperative clinical improvement was common after debridement of hematoma. Overall postoperative KS score was increased among the entire sample of patients with hematomas and no MR signs of malformation (from 50 scores prior to surgery to 60 scores after surgery). On the contrary, patients with CA were characterized by reduced KS score from 70 to 50 points immediately after surgery (Fig. 5a, b).

Early postoperative regression of neurological symptoms was noted in 77% of patients with subacute hematomas and in 59% of patients with chronic hematomas. Neurological aggravation was observed only in 8% and 18% of cases, respectively. Postoperative outcomes were less favorable in patients undergoing surgery for CA with hemorrhage. Deterioration of symptoms occurred in 33% of patients, regression of symptoms — 27% of cases. Resection of CA outside the hemorrhage stage resulted unfavorable early postoperative outcome in the majority of cases. Deterioration of symptoms occurred in 73% of patients, regression — only in 5% of patients (Fig. 6).

Previous recurrent hemorrhages in patients with CA did not significantly affect the risk of postoperative aggravation of symptoms of brainstem lesion. However, these events reduced the probability of regression of symptoms from 36% to 16%. Apparently, the latter is explained by irreversible changes in the brainstem structures associated with recurrent hemorrhages.

Long-term outcomes of surgical treatment

Gradual clinical improvement in long-term postoperative period was observed in most patients (85%). This was true for patients with early postoperative regression of symptoms and for those with initial neurological aggravation or no changes early after surgery (in these cases, improvement was usually observed in 3—4 weeks after surgery) (Fig. 7).

Mean duration of regression of neurological symptoms was 4.7 months (from 2 weeks to 2 years).

The most significant regression of symptoms was found in patients with hematomas. Significant clinical improvement was confirmed in 6 months after surgery compared to preoperative and early postoperative status. Up to 57% of patients became actually healthy and returned to good quality of life (KS score 80—100), while 70% of patients were bedridden and needed intensive therapy prior to surgery (KS score ≤50). Surgery for CA associated with hemorrhage was followed by subtotal regression of symptoms in 35% of patients (KS score 90—100). Importantly, hemorrhagic component prevailed over the malformation in all these patients and preoperative KS score was 50—60. Improvement was less significant if malformation tissue prevailed over the volume of hematoma (preoperative KS score 60—70 and 80 scores in the follow-up). Thus, the main important aspect in patients with CA undergoing surgery outside the hemorrhage stage is significant clinical improvement through the follow-up period compared to early postoperative period. However, recovery takes place, at best, only up to preoperative level.

Recurrent hemorrhages in long-term postoperative period

Overall incidence of recurrent hemorrhages was 15%. Mean hemorrhage-free postoperative period was 3 years (from 2 weeks to 4 years). Redo surgery was required in the majority of patients (92%).

Recurrent hemorrhages occurred in 20% of patients with hematoma within 2 weeks—1 year (mean 6 months) after surgery. The following aspects caused hemorrhages:

1. Development of CA after debridement of subacute hematoma (14% of patients with subacute hematomas), while there were no signs of malformation according to MRI data, intraoperative data and histological examination of specimens (Fig. 8a, b, c);

2. Hematoma capsule remnants and their abnormal vessels. In our sample, these recurrent hemorrhages occurred in 23% of patients.

Recurrent hemorrhage and redo surgery were followed by deterioration of neurological deficit with subsequent partial regression.

Case report

A 62-year-old patient K. suffered from pulsating headache in cervical-occipital region for a long time. Unsteady gait and diplopia occurred in April 2006. These symptoms were characterized by gradual aggravation for 2 months. Left-sided numbness and weakness in the left arm and leg occurred for this time. MRI revealed a hematoma of tegmentum of the midbrain (Fig. 9a).

There was no further neurological deterioration until surgical treatment in January 2007. Chronic right-sided hematoma of tegmentum of the midbrain was resected. The walls of hematoma were thin and their total resection was impossible. There were no any obvious intraoperative signs of malformation, but multiple small vessels (telangiectasias) in adjacent brain matter were observed. Postoperative regression of left-sided pyramidal symptoms was associated with deterioration of the oculomotor nerve function on the right (ptosis). Early postoperative neurological impairment was followed by significant regression in long-term follow-up. There was only mild diplopia in upward gaze. Control MRI in 2 months after surgery revealed no residual hematoma or malformation (Fig. 9b). Recurrent hemorrhage developed in 3 months after surgery. The complication was accompanied by dysfunction of the oculomotor nerve on the right, hypoesthesia on the right side of the face and unsteady gait. Redo surgery was refused considering small volume of hemorrhage and stable clinical state. Medication resulted regression of X-ray signs of hemorrhage and neurological improvement over 2 months. Acute clinical deterioration occurred in August 2008 (non-systemic dizziness, unsteady gait, diplopia and left-sided numbness). These symptoms gradually increased for the following time. Control MRI revealed subacute right-sided hemorrhage in tegmentum of the midbrain. Redo surgery was performed in August 2008. Biopsy verified typical hematoma besides the hematoma capsule. Redo surgery was followed by neurological deterioration (impairment of left-sided superficial sensitivity, disturbances of deep sensitivity up to the level of wrist and ankle joints, dynamic ataxia in the right extremities). There was no further regression of symptoms. Moreover, the course of disease acquired an undulating-progressive nature characterized by the episodes of deterioration as a result of hemorrhages and subsequent partial regression of neurological symptoms. However, general condition gradually worsened. Weakness in the left hand and extrapyramidal symptoms (tremor and muscle tone changes) appeared. The patient ceased to care for herself in 4 years after redo surgery. MRI confirmed enlargement of CA of the right midbrain (Fig. 9c). Redo surgery was refused considering severe clinical state, progression of neurological symptoms and large dimensions of CA occupying the whole right half of the brainstem.

In the group of MRI-verified CA, recurrent postoperative hemorrhages occurred only in 2% of patients within 2—36 months (mean 15 months). It is noteworthy that all these patients were operated on the background of hemorrhage from malformation, and only hematoma was resected. Residual CA caused recurrent hemorrhage. Unfortunately, redo surgery was not followed by total resection of malformation. However, hemorrhage did not occur again in the majority of these patients. Active growth of malformation followed by recurrent hemorrhages was observed only in 2 cases that required redo surgery.

Case report

A 49-year-old patient S. The disease manifested in January 2009. Increased blood pressure was followed by headache in the cervical-occipital region, nausea, dizziness and unsteady gait. Acute neurological symptoms aggravated for the following week. Right-sided numbness including the face has occurred for this time. MRI revealed a pontine hematoma (Fig. 10a).

Symptoms persisted for 2 months and regressed later. Dizziness, unsteady gait and right-sided numbness have improved within 3 months. Chronic hematoma and CA of tegmentum of the pons and midbrain on the left were resected in September 2009. Resection was performed through the lateral supracerebellar approach. Total resection of malformation was impossible through this approach. We only coagulated the caudal fragments of malformation. Early postoperative period was characterized by improved right-sided sensitivity and gaits. However, ptosis of the left eyelid was observed which regressed within a month after surgery. Total postoperative regression of symptoms lasted for 3 months. There was only mild of numbness of the right arm and leg after this period. Control MRI in 6 months after surgery revealed residual malformation within tegmentum of the pons (Fig. 10b). The patient's condition was stable for 10 years. She went back to her previous social activity. Deterioration has been noted since April 2018, when the patient paid attention to the changes in handwriting. There was impairment of numbness in the right arm and leg in May. These symptoms were characterized by gradual progression. In September, increased blood pressure resulted dizziness, unsteady gait, weakness in the right hand and leg, impaired deep and temperature sensitivity in the right hand. Control MRI revealed recurrent hemorrhage and enlarged malformation in the left half of the pons (Fig.10c). Redo surgery was performed through suboccipital craniotomy and telovelar approach to achieve total resection of malformation. This procedure was followed by neurological deterioration. Horizontal gaze paresis, impaired deep sensitivity up to the level of interphalangeal joints of the right hand and foot, ataxia in the left extremities were diagnosed.

Thus, incorrect surgical approach at the first surgery did not allow total resection of malformation that was followed by further growth and recurrent hemorrhages in long-term postoperative period. Redo surgery was more traumatic due to predominance of malformation over hemorrhage that led to neurological deterioration.

Importantly, postoperative recurrent hemorrhages were absent in patients who underwent resection of CA outside the stage of hemorrhage. At the same time, partial resection of malformation took place in 11% of cases.

Outcomes of conservative treatment (n=193)

Conservative management was chosen in the following cases:

1. Significant regression of neurological symptoms by the moment of examination or their initial minimal severity;

2. Small hematomas (MRI data);

3. High risk of postoperative neurological deterioration;

4. No substrate for neurosurgery (hemorrhage resorption by the moment of consultation).

Usually, two or even three factors were combined to make a decision in favor of conservative treatment.

The majority (79%) of patients had CA without MR sings of hemorrhage at admission. In 22% of cases, no clinical symptoms of hemorrhage were observed throughout the disease. Acute symptoms of hemorrhage regressed almost completely within 2.3 months in 78% of cases and KS score was 90—100 by the moment of examination. Single hemorrhagic event was observed in 83% of cases, recurrent hemorrhage — only in 27% of patients.

Considering significant regression of neurological symptoms, conservative management was preferred in 8% of patients with moderate chronic hematomas (2—3 mm3) despite their persistence. KS score was 80 by the moment of examination. Slow improvement of neurological symptoms has continued for 1.5 years of the follow-up. However, there was no complete regression of neurological syndrome in any patient. Persistent residual syndrome of neurological deficit did not significantly hinder social adaptation in any case (KS score 90).

In 13% of cases, primary consultation was carried out after complete resorption of hemorrhage (after 8—12 months). Thus, only hypointense focus was observed in T2WIs (hemosiderin). Subtotal remission was observed in these patients (KS score 90—100).

Recurrent hemorrhages were observed in 9% of patients throughout the entire follow-up period. Conservative therapy was effective in most of them (66%) and resulted complete regression of minor neurological symptoms within a month. These were patients with CA as a rule. Recurrent hemorrhage required surgical intervention in one third of the patients (28%). These were patients who had initially been counselled at the stage of hemorrhage resorption. CA was intraoperative revealed in all patients. Subsequent biopsy confirmed this diagnosis.

Radiosurgery was applied in 6% of patients with MRI-verified CA. This approach did not essentially affect the further course of disease. Mean hemorrhage-free period was 3 years after radiosurgery in all patients. Mean follow-up period was 6 years (range 4—10). Clear symptoms of recurrent hemorrhage with subsequent partial regression were observed in these patients throughout the follow-up period in addition to multiple minor hemorrhages (Fig. 11a, b).