Cerebral cavernous malformations (CMs), also termed as cavernous angiomas, cavernous hemangiomas, or cavernomas, are vascular malformations consisting of thin hyalinized vascular channels without interposed brain tissue. In the pediatric age group, CMs have been reported to be one of the main causes of brain hemorrhage along with ruptured arteriovenous malformations (AVMs)19,20). In addition, the risk for seizures in patients with a previous history of epilepsy due to CMs is 2 to 4 times higher than that in patients without a previous history of epilepsy18,23). Patients with CMs who are under the age of 40 years are 5.6 times more likely to have a lifelong disability due to seizures than older patients24), especially when the seizures have started during childhood1,2).

Treatment options may vary depending on the patient's presenting symptoms and the location of CMs. Because radiosurgery for CMs does not seem to have a real advantage over conservative or medical therapy, surgical resection of the whole lesion seems to be the ideal treatment for symptomatic lesions1). Recent advances in neuronavigation systems have made targeting of small-sized CMs and approaching the deep-seated regions easier. Early diagnosis and complete resection of lesions can prevent neurological deterioration from repeated bleeding. Furthermore, in patients presenting with seizures, early surgical resection would prevent the epileptogenic focus to enlarge and involve surrounding brain tissue which might include the hemosiderin border. However, seizure outcomes with respect to resection of the hemosiderin border still remains unclear in children with supratentorial CMs3).

The purpose of this study was to investigate the clinical features and outcomes of pediatric CMs treated with surgical resection by using neuronavigation systems, intraoperative neuromonitoring, and TTRS.

We retrospectively reviewed the clinic-radiological data of 29 children (under 18 years) who underwent surgical resection of supratentorial CMs between January 2004 and June 2012. The diagnosis of CM was confirmed pathologically in all patients. Preoperative MRI was available in all patients, and lesion location, number of lesions, and whether the lesions were located in an eloquent area or not were evaluated. We treated all patients with craniotomy and lesionectomy. There were 10 (34%) patients with multiple lesions, and lesionectomy was selected based on the correlation of the lesion with the presenting symptoms. If lesion localization was not achievable, the lesion with an overt hemorrhage or the largest lesion that caused a mass effect was removed. In patients with acute symptomatic hemorrhage, we waited until the liquefaction of hematoma and improvement of neurological deficit occurred unless the symptoms worsen or neurological deficits progress. We did not remove the hemosiderin-stained area surrounding the core lesion; we only removed the vascular malformation.

We used intraoperative ultrasonography in the earlier cases (n=4). A frameless navigation system (BrainLab, Feldkirchen, Germany) has been applied in 24 patients since 2006. The navigation system helped us to choose the shortest pathway to a deeply located lesion and to remove the lesion precisely with mini-craniotomy (Fig. 1). We used the transparent tubular retraction system (TTRS) in 7 patients who had the lesion in a deep-seated region to minimize the incidence of retraction injury to the normal parenchyma and secure an adequate working space for gross removal of the lesion (Fig. 2). We have previously reported the feasibility of TTRS in patients with deep-seated tumors13). When the lesion was located in an eloquent area, especially in the primary motor cortex or descending pyramidal tracts, intraoperative monitoring of motor evoked potentials (MEP) and somatosensory evoked potentials (SSEPs) were performed. In all cases, immediate postoperative CT was performed to detect hemorrhagic complications. MRI was taken postoperatively to detect residual lesions as well as late complications. In seizure patients, residual hemosiderin-stained areas were assessed on MRI for confirming the role of hemosiderin in seizure patients.

Neurological examinations were performed pre- and post-operatively. Extended Glasgow outcome scale (GOS-E) was used to assess surgical outcomes31). In thirteen patients who presented with seizures, Engel's classification was used to measure the surgical outcome, including the functional status of a seizure patient9). The average follow-up period was 27 months (range, 1 month to 8 years). In seizure patients, the average follow-up period was 37.7 months. Except for one 14-month-old boy who was lost to follow-up from 3 months after surgery, all other seizure patients were followed up for at least 20 months.

Of the 29 children with supratentorial CMs, 18 were males. Their age at the time of surgery varied from 9 months to 18 years, and the mean age was 9.4 years. The most common presenting symptom was seizures; 13 patients (45%) presented with seizures (3 cases of simple partial seizures and 10 cases of generalized tonic-clonic seizures). Patients' demographics who presented with seizure are shown in Table 1. The second most common symptom that was reported by 8 patients (28%) was mild neurological symptoms such as headache, dizziness, nausea and vomiting. Only two patients (7%) presented with focal neurological deficits including weakness and delayed speech. In 6 patients (21%), CMs were incidentally detected on MRI, which was mostly performed after minor head trauma. Regardless of the severity of symptoms, variable amount of overt hemorrhage was detected in 7 patients (24%) during their initial imaging study. Evaluation of the lesion number on MRI revealed that 19 patients (66%) had a single lesion, and 10 patients (34%) showed multiple lesions. Evaluation of the lesion location on MRI revealed that in 26 patients CMs were detected only in the supratentorial area (90%) and in 3 patients, CMs were detected in both the supra- and infratentorial area (10%). In 10 cases (34%), CMs were located in the eloquent areas.

Post-operative imaging showed that 26 of 29 patients had complete resection of the target lesion, which accounts for 90% of the cases. A residual lesion was found in 3 patients. One of these 3 patients had a lesion in the left precentral gyrus and had subtotal resection due to change of monitoring during surgery. The other two had a deep-seated lesion in the caudate nucleus, that one of them seemed to be removed totally on visual inspection, but a residual lesion was detected on postoperative MRI and the other had subtotal resection due to change of evoked potential during surgery. Except for these 3 (10%) patients, complete resection was performed in all patients and was confirmed by postoperative MRI. Of the 13 patients who presented with seizures, all had hemosiderin-stained areas surrounding the lesion.

There was no perioperative mortality or severe morbidity that showed worse neurological deficits than pre-operative neurological status; however, some minor complications occurred during the follow-up. 3 (10%) patients had transient weakness or sensory changes immediately after surgery, but they were recovered completely. Assessment of GOS-E revealed one patient with a score of 6 and another with a score of 7, representing mild disabilities. Their symptoms were constant after surgery compared to their symptoms before surgery.

Of the 13 patients who presented with seizures, 10 patients did not develop seizures postoperatively during follow-ups of 20-99 months (Engel class IA). Two patients developed generalized tonic-clonic (GTC) seizures 2 months after surgery, but there was no seizure thereafter without antiepileptic medication (Engel class IC). The remaining one patient with a lesion in the right frontal lobe presented with GTC seizures at the time of diagnosis; he underwent surgery and did not experience seizure for 18 months. He stopped taking antiepileptics, but after 14 months he had simple partial seizures (Engel class IB). He restarted antiepileptics. Assessment of GOS-E revealed all of the 13 patients with score 8 (Table 2).

Cerebral cavernous malformations (CMs) are vascular malformations consisting of thin hyalinized vascular channels without interposed brain tissue. CMs have been known to be present in about 0.4-0.8% of the population18); and one-fourth of CMs occur in the pediatric population7).

CMs are known to have a frequency of symptomatic hemorrhage ranging from 0.1% to 2.7% per lesion-year and 0.013% to 16.5% per patient-year29). In children, CM-related hemorrhagic events are approximately 2-3 times more frequent than in adults1). CMs have been reported to be one of the main causes of brain hemorrhage along with ruptured arteriovenous malformations (AVMs)19,20) in the pediatric age group. Considering the increased risk of re-hemorrhage, surgery should be performed in all symptomatic CMs that are located in eloquent regions because it appears to be a safe option under the modern monitoring conditions32). Clinical results in children who were treated with surgical therapy for CMs of the CNS are generally excellent or good in majority of pediatric series8,11,16,17,19,25). In addition, surgery led to the complete cure of epilepsy (i.e., discontinuation of anticonvulsant therapy) in 31% of the children with preoperative seizures, while better therapeutic control of seizures was achieved in more than 48% of cases1). Given the hemosiderin-related epileptogenesis26,30), it is generally agreed that the hemosiderin-stained areas or gliotic tissue should be removed in order to achieve complete relief from seizures, and complete removal of the hemosiderin-stained area may improve the long-term outcome4,5,10,28,31,34). However, even without performing total removal of hemosiderin-stained areas, we could achieve a satisfactory seizure control outcome. It assumed that additional resection of hemosiderin-stained areas appears to be unnecessary in children. We have previously reported that early surgical removal of superficial CMs or of CMs that are not critically located but do cause seizures may provide a real curative treatment for epilepsy33), thus preventing psychosocial disability in patients on long-term medical therapy and avoiding the risk of neurological deficits due to growth and hemorrhage in the CMs. In children, the growth behavior of CMs is more aggressive than in adults. Also, the risk of hemorrhage or frequency of developing seizures is higher than that of adults1,15,18,21). Considering that the hemosiderin rim would grow larger and take a role in epileptogenesis, early removal of CM with or without resection of the hemosiderin rim would be helpful especially in pediatric group, too. In addition, children have enhanced capacity for brain plasticity compared to adults, and they have capacity to recover from brain injuries or radical surgery14). Therefore, more aggressive surgical resection of CMs would be required in pediatric patients.

In the present study, 10 patients presented with multiple CMs. In cases of multiple CMs, surgery of the symptomatic lesion should be performed. In the illustrated case of this study shows the benefit of early surgery under modern monitoring conditions. Though there were some residuals left, regarding its size and locations, we believe that early surgery prevented the child to have disastrous neurological deteriorations from rebleeding of the CMs. Though multiple CMs may occur in more than 30% of sporadic CMs12,22), autosomal dominant familial CMs should be always considered. The hemorrhagic risk of familial CMs is believed to be higher than that of sporadic CMs6). Although genetic testing was not performed in the present study, this may have substantial benefits in managing patients with multiple CMs.

Today, the widespread use of MRIs, functional MRIs, intraoperative neuromonitoring, neuronavigation systems, brain mapping, and awake craniotomy can minimize the postoperative morbidity in patients with CMs in the most eloquent locations27,32). For an incidental CM in an eloquent location, surgery is not advisable due to a still significant risk of postoperative deterioration32). In our series, with the assistance of neuronavigation system, TTRS, and also intraoperative neuromonitoring, a satisfactory outcome and seizure control was achieved in all patients with CMs. Since patients with cerebral CMs located in an eloquent region who present with seizures can be cured with surgical treatment32), detailed evaluation of the lesions that correlate with the presenting symptoms and vigilant surgical resection must be considered.

For a symptomatic solitary CM, the treatment of choice is complete microsurgical excision preceded by careful anatomical and functional evaluation. If the lesion is located in the eloquent area, surgical intervention may be very challenging for surgeons. With the assistance of neuronavigation systems, intraoperative neuromonitoring, and TTRS, the lesion could be targeted more accurately and excised more safely with minimization of normal tissue damages. Furthermore, based on the satisfactory seizure outcomes achieved, early resection is recommended for CMs presenting with seizures but additional resection of hemosiderin-stained areas appears to be unnecessary in children.