Transsulcal approach supported by navigation-guided neurophysiological monitoring for resection of paracentral cavernomas

2.1. Patients and nature of lesions 

Between June 2000 and July 2007, 17 patients harboring a paracentral cavernoma underwent surgery via transsulcal approach with the aid of frameless neuronavigation and neurophysiological intraoperative monitoring. 6 patients were male and 11 were female, and the patients ranged in age from 17 to 65 years (mean±S.D., 41.24±13.52 years).

There are eight cavernomas located in the right and nine in the left hemisphere. The size of lesions varied between 8mm up to 45mm in diameter (mean±S.D., 23.06±10.29mm). The depth of the cavernomas ranged from 5mm up to 40mm. The space-occupying lesions were located in the frontal lobe rostral to the precentral gyrus in 2 subjects, in the precentral gyrus in 3 subjects, and in the postcentral gyrus in 5 subjects. Seven lesions extended into more than one cerebral gyrus. Three subjects had multiple lesions. In these 3 subjects, only the paracentral lesions were removed surgically, and other supratentorial or infratentorial lesions were left untreated in this study. In one patient the cavernoma was associated with a venous malformation. Four patients with cavernomas suffered from acute hematomas.

All patients had various symptoms including headache, hemiparesis, hemihypesthesia or epileptic seizures. Eight patients suffered from seizures. Of these eight patients, 3 patients with simple partial seizures, 2 patients with complex partial seizures, and 3 patients had a history of generalized seizures. All patients had at least two seizures per year despite adequate antiepileptic therapy. In these patients, EEG revealed an epileptic focus with a topographic relation to the cavernoma. In cases of multiple cavernomas, Video-EEG-Monitoring was used to identify the seizure focus.

The details of clinical symptoms in all patients are summarized in Table 1.

Table 1. Summary of clinical features of 17 patients with paracentral cavernoma.

	Case number	Age (yr)/sex	Presenting symptoms	Size (mm)	Side	Location	Follow-up (month)	Outcome at follow up
	1	65/F	Seizures, right sided hemiparesis	30	L	Frontal+precentral	12	No seizures, hemiparesis improved
	2	46/F	Seizures, minor paresis of the right leg	18	L	Postcentral	13	No seizures, no deficits
	3	29/M	Left sided central facial palsy paresthesia of the left arm and face	26	R	Precentral	15	Symptom-free, no deficits
	4	60/F	Seizures, right sided hemiparesis and slight aphasia	32	L	Frontal+precentral	28	No seizures, hemiparesis unchanged, no new deficit
	5	27/F	Headache, paresthesia of the right arm	8	L	Postcentral	31	Symptom-free, no deficits
	6	32/M	Headache, hemihypesthesia, problems with speaking	25	L	Precentral+postcentral (multiple)	30	Symptom improved, hemihypesthesia, no new deficits
	7	17/F	Transient paresis of the left arm, left sided hemihypesthesia, seizures	34	R	Precentral+postcentral	30	No seizures, symptom-free, no deficits
	8	30/F	Seizures, transient left sided hemiparesis	25	R	Postcentral	2	One seizure within first 2weeks after operation, and then symptom-free, no deficits
	9	29/F	Seizures, hemihypesthesia of the left arm, hemiparesis	10	R	Postcentral (cavernoma+VM)	35	No seizures, no deficits
	10	39/M	Asymptomatic concerning cavernoma	10	R	Postcentral	19	No deficits
	11	38/M	Dysaethesia of the right hand and face, dysarthria	12	L	Precentral	19	Symptom-free, no deficits
	12	55/F	Hemorrhage twice, headache	12	L	Frontal	31	Headache, no deficits
	13	46/F	Left sided hemiparesis	30	R	Frontal+precentral (multiple)	34	Leg paresis improved, no deficits
	14	40/M	Right sided hemiparesis, aphasia	30	L	Precentral+postcentral	11	Symptom-free, no deficits
	15	35/F	Seizures, headache	45	R	Frontal+precentral (multiple)	40	No seizures, symptom-free, no deficits
	16	54/F	Seizures	25	R	Frontal	1	No seizures, symptom-free, no deficits
	17	57/M	Seizures	20	L	Precentral	81	No seizures, symptom-free, no deficits

Abbreviations: yr: years; M: male; F: female; L: left; R: right; VM: venous malformation.

2.2. Image guidance system and image-guided sonography 

The detail technical features of the VectorVision2 (BrainLAB, Heimstetten, Germany) neuronavigation system have been reported previously [2], [3], [4]. Following MRI or CT data acquisition and presurgical planning, the processed patient data were transferred onto the neuronavigation system. The patient registration was performed using standard fiducials.

During the period between June 2000 and March 2005, a conventional intraoperative ultrasound (Aplio, phased array probe, 4.8–8.5MHz, Toshiba, Tokyo, Japan) was used in 9 subjects to aid neuronavigation, and from March 2005 to July 2007, a new image-guided sonography was used in the remaining 8 patients. The technological integration of navigated ultrasound with our novel neuronavigation platform has been described previously [5], [6]. Briefly, a high-end ultrasound device (IGSonic, phased array probe, 5–7.5MHz, BrainLAB) is directly integrated into the VectorVision2 neuronavigation system by a navigation software (VV Cranial, BrainLAB). The navigation system displays the preoperative MRIs as well as the intraoperative ultrasound scan on a touch screen monitor. For a fusion and contrast of the presurgical MRI/CT images versus in-life ultrasound images during surgery, the presurgical image shown in gray-scale was overlaid with real time B-mode ultrasound images shown in green color. Landmarking of the vascular anatomy using color and power duplex techniques was performed in some patients, the procedure was described previously [7].

2.3. Functional mapping and neurophysiological intraoperative monitoring 

We have applied various techniques of functional mapping and neurophysiological intraoperative monitoring (IOM) over the past decade in an effort to ameliorate the technological support of our surgery in eloquent areas. We updated our equipment within the course of our study continuously from somatosensory evoked potential (SEP), motor evoked potential (MEP) to cortical mapping and preoperative functional MRI (fMRI), recently. IOM of SEP, MEP and cortical mapping for the localization of primary motor cortex (M1) were conducted under general anesthesia. fMRI was performed in the five most recent cases of 17 patients.

2.3.1. Evoked potentials (SEP and MEP) 

Evoked potentials were measured in 16 patients for intraoperative functional monitoring and M1 localization. To localize the central sulcus, a four-contact linear subdural strip electrodes (AD-TECH Co., WI, USA) was placed on the cortex (Fig. 1, Fig. 2). A series of recordings were made from the cortical surface by moving the electrode to find and verify the localization of somatosensory cortex by using a referential montage with median or tibial nerve stimulation, until the highest amplitude of N20/P25 or P40/N50 and shortest latency was recorded. The central sulcus is localized by means of SEP phase reversal. An Endeavor CR IOM system (Viasys Healthcare, Nicolet Biomedical, WI, USA) was used for all examinations.

View Large Image Download to PowerPoint

Fig. 1. (A) Intraoperative positioning of the patient head and the incision on the skin. (B) Intraoperative photograph of the cortical surface prior to resection showing the four-contact strip electrode (white arrow) overlying the motor cortex. (C) The SEP recordings during resection. A well developed N20–P25 can be seen in Contacts 1, 2, 3, and a SEP phase reversal can be seen in Contact 4. (D) CT scan after the operation. (E–G) Intraoperative view. (E) Opening of the sulcus just anterior to the cavernoma, the central gyrus shown by yellow open arrow was outlined prior to surgery on the basis of functional MRI (not shown). (F) Dissection into the precentral sulcus (white arrow) is continued using bipolar coagulator towards the lesion located deeply in the precentral gyrus, and (G) The cavernoma (white arrow) seen in the depth of the sulcus. (H, I, L and M) Intraoperative navigation; (J) image-guided sonography and (K) MRI reconstruction in the same image plane, the white arrow shows the cavernoma and the green strip indicated by yellow open arrow is central sulcus; in this case, the intraoperative nonnavigated US scan before opening of the dura (N), and during resection (O) (near sagittal view) provides a superior image quality as compared to the navigated US image of our first generation (J). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

View Large Image Download to PowerPoint

Fig. 2. A combination of preoperative fMRI and neurophysiological intraoperative monitoring, including SEP, MEP and cortical mapping in a patient with multiple intracranial lesions. Axial MR (A), functional MR (B and C) images and sagittal MR (F), functional MR (G) images reveals a cavernoma (white arrow) and hand motor cortex (yellow areas); ICG-angiography (H and I) showed an avascular area (H, white arrow) with moderate mass effect; (J) intraoperative SEP recording of the inversion of a postcentral negative and a precentral positive peak, the N20–P25 phase reversal was observed between electrodes 1–2 and 3–4 (white arrow); (D and E) stimulation of the cortex with an OJEMANN cortical stimulator (white arrow); (K) ECoG monitoring and (L) a four-contact strip electrode overlying motor cortex (white arrow). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

MEP was continuously recorded to monitor motor function. Muscle action potentials were recorded from the forearm flexor, thenar and quadriceps muscle, contralateral to the side of stimulation. Neurophysiological monitoring was performed through out surgery (Fig. 3Q–S). Baseline readings were obtained prior to skin incision and for intradural surgery after opening of the dura mater. Critical SEP and MEP changes were defined as decreases in amplitude of more than 50% of baseline values or increases in latency of more than 10% of baseline values.

View Large Image Download to PowerPoint

Fig. 3. (A and B) MR images show a cavernoma (white arrow) in the precentral gyrus. (C and D) Postoperative CT/MRI images showed total removal of the lesion as compared with the preoperative MRI. Intraoperative screen shot of neuronavigation and image-guided sonography before dura opening (E–J) and during ongoing resection (K–P). Note that the previously largely hemorrhagic cavity (H) has shrinked after removal of the blood and the cavernoma (N). Furthermore, the image quality is much superior when compared to our first version of navigated ultrasound (seen in Fig. 1J). (Q–S) Intraoperative neurophysiological monitoring. (Q) The central sulcus is localized by means of SEP phase reversal between electrodes 3 and 4; (R) SEP and (S) MEP intraoperative monitoring; (T–X) an operative view during removal of the lesion with transsulcal approach. Opening of central sulcus with microscissors (T), the lesion was made accessible (U), and then excised (V). (W) Surrounding hemosiderin-stained gliotic tissue on the side of frontal lobe was removed and (X) the overview of the 10mm-incision.

2.4. Surgical procedure 

After induction of general anesthesia, patients were positioned supine with mild head extension and rotation to the contralateral side of the lesion. The image-guided excision of the cavernoma was performed in three steps. First, the skin incision and the osteoplastic craniotomy were made according to the exact location and the size of the cavernoma after transcranial visualization of the cavernoma and adjacent sulcus by navigation. Usually a linear incision of 5–6cm was adequate for a small lesion less than 1.5cm (Fig. 4M). As for some big lesion a flap was performed to adapt the whole lesion. Secondly, after the craniotomy had been completed, the central sulcus and precentral gyrus or postcentral gyrus, as well as the cavernoma were identified with the aid of the navigation system again (Fig. 1H–M). Then the functional area was confirmed by neurophysiological test. As for the patients with epilepsy, subdural electrode strips were placed on the brain cortex to define the epileptogenic areas as well as to compare with the pre-operation assessment. Thirdly, the sulcus opening site was chosen to give the shortest and safest corridor to remove the lesion and minimize the potential injury of neural function (Fig. 4A–I).

View Large Image Download to PowerPoint

Fig. 4. (A–I) Intraoperative view. (A and B), a sharp incision of the superficial arachnoidal layer and underlying arachnoidal bands (B, blue open arrow) was made under high magnification. (C) the sulcus was opened, note: veins (blue open arrow) running along and down into the deep sulcus should be preserved. (D) Hemosiderin-stained tissues (blue open arrow) on the surface were the first signs of cavernoma; they suggested the entrance and direction of dissection. The lesion exposed (E) was then dissected (F) and removed (G) (white arrow in E, F and G shows the cavernoma). After removal of the lesion, the veins (blue open arrow) were preserved (H) and the photography of the incision (I). Preoperative MR images in axial (J), coronal (K), and sagittal (L) plane showed a subcortical cavernoma in the right central area (white arrow) and the adjacent sulcus just anterior to the cavernoma. (M) Position of the patient head and the incision. (N) Postoperative CT scan and (O) photography after operation showing the patient without neurological deficit. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

To verify the actual location and adjust the trajectory of dissection toward the lesion during surgery, the navigation guide was applied throughout surgery, and revised by the intraoperative ultrasound (Fig. 1, Fig. 3, Fig. 5). The malformation was removed, including the surrounding hemosiderin-stained gliotic tissue on the side of the frontal lobe and the postcentral gyrus, to eliminate its potential epileptogenic effect, but keep a small proportion on the side of precentral gyrus (Fig. 3W). This maneuver was undertaken with special precautions to avoid injury to adjacent functional structures. Eventually, following closure of dura, cranioplasty and wound closure were performed by standard neurosurgical techniques. In each patient, an early postoperative CT scan was obtained to rule out any local hemorrhage.

View Large Image Download to PowerPoint

Fig. 5. (A–F) Intraoperative screen shot during ongoing resection using neuronavigation and image-guided sonography. The blue crosses indicated the position of the tip of ultrasound probe: axial (A), coronal (B), and sagittal (C) view. (D) Intraoperative ultrasound scan with duplex window (the area highlighted in light blue depicts the entire hyperechogenic lesion with no display of any flow in and around the lesion). (E) MRI reconstruction in the same image plane as the ultrasound scan. (F) Note: the brain shift (white arrows) in the green overlay technique, which is nicely exposed if the cavernoma outline of the MRI (green line) is compared to the hyperechogenic lesion. (G) and (H) Preoperative MR images showed a well-defined lesion (white arrow) in the right precentral gyrus; intraoperative photograph showed the incision on the skin (I) and central sulcus (J, white arrow). (K), recording of SEP to contralateral median nerve stimulation. Phase reversal of N20–P25 is observed between electrodes 2 and 3; (L), postoperative CT scan. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

2.5. Opening of the sulcus 

In all 17 patients, transsulcal approaches (transcentral sulcus in 6, transprecentral sulcus in 6 and transpostcentral sulcus in 5 patients) were used to reach the lesions. The initial incision was made based on the individual shape of the sulcus and the function of surrounding cortex. Usually hemosiderin-stained tissues on the surface were the first signs of cavernoma would suggest the entrance and direction of dissection (Fig. 4D). If the sulcus above the lesion was tight due to brain edema or adhesion of the arachnoid membranes, a relatively normal part of the sulcus was dissected first along the cortical artery, followed by extending the space to the lesion.

On the brain surface, the subarachnoid membrane was cut with a minimal tension on the left by applying gentle retraction onto a cottonoid sponge placed on the adjacent tissue. Subsequently, a sharp incision of the superficial arachnoidal layer and underlying arachnoidal bands was made under high magnification. In opening of the sulcus, microscissors were used as much as possible to avoid cortical injuries due to retraction and coagulation by bipolar forceps [8], [9]. After the pial surface covering the lesion was reached, the lesions were then excised under bipolar low-current coagulation. Great care was taken to maintain the patency of the sites identified as M1 by fMRI, SEP phase reversal or cortical mapping. A self-retaining retractor placed over a cottonoid strip was used in the deep area but its application was limited to a short duration as appropriate. Larger and deeper lesions required longer arachnoidal incisions to avoid any traction at the extremities of arachnoidal incision and minimize compression on the exposed cortical surface within the sulcus.

4.1. Ultrasonography-assisted image-guided neuronavigation 

Precise identification of anatomical landmarks and avoidance of vital structures are crucial during presurgical planning and during surgery, this is evidenced by the reduced morbidity following the application of navigation system [13], [14]. Because most image guidance relies on preoperative digital images, intraoperative loss of CSF and the effect of cortical displacement can affect the accuracy of the navigation system [3], [4], [15], [16], [17]. Supplementing preoperative digital images with intraoperative ultrasound offered one solution to overcome this problem of brain shift during surgery. A simple image overlay [6], [18] or a display of sonographic and corresponding tomographic scans side by side [19] in the system for image-guided ultrasound can identify any tissue shift in a two-dimensional plane and provide an estimate of brain shift. This is helpful for surgeon to adjust to the altered brain shape during surgery. Moreover, integration of the ultrasound into the navigation system with MRI or CT images obtained a priori help to better interpret the intraoperative sonographic data [5], [7], [20].

Intraoperative CT/MRI are other navigation modalities that can help to adapt surgery to the brain shift and are currently being applied [21], [22], [23], [24]. Compared to ultrasound images, CT and MRI images possess higher resolution. However, these technologies are costly and require cumbersome procedures. They are rigid and difficult to adopt for continuous real-time imaging during surgery. In comparison, ultrasound performs true real-time imaging, and it is time- and cost-effective and flexible, can be relatively easily made to address various emergences in surgery [6], [7], [20], [25].

4.2. Evaluation of functional mapping with combined methods, localization of the primary motor cortex 

Surgical removal of brain lesions located within or near the motor cortex frequently results in postoperative deterioration of motor function. fMRI and intraoperative neurophysiological monitoring have been used to help identify and preserve the eloquent area during surgical resection of brain tumors. When appropriately applied, these techniques may identify M1 and are helpful to improve neurosurgical outcomes in certain situations [26], [27], [28], [29], [30].

However, it has been well recognized that using these techniques alone was sometimes not sufficient for the accurate localization of M1 and may result in postoperative motor deficits [31], [32], [33]. For example, SEP is a simple and useful technique, and its principle is based on the fact that a somatosensory potential is recorded from the sensory cortex and that its mirror image is recorded from the motor cortex [34]. But its utility is limited in the context of large central and postcentral tumors. Shinoura et al. [35] found even though phase reversal was successfully detected, SEP was not completely reliable in localizing the M1 in the context of an underlying lesion, because reorganization of brain function could produce a shift in the SEP. Still other studies have suggested that intraoperative cortical mapping may have low sensitivity for the detection of motor function in the area besides the lesion [36], with the risk of induction of epilepsy and postepileptic paresis. All these data suggest that the use of SEP or brain mapping alone was insufficient and a combination of multiple methods should be applied for the accurate localization of M1 rather than use of one method alone [35], [37].

In this report, lesions located in the paracentral area were removed completely without further neurological deficits by using continuous functional monitoring. A combination of preoperative fMRI and IOM, including SEP, MEP and cortical mapping, has been used to prevent compromise of M1 during resection of cavernomas. In our practice, besides the updating of the new techniques, we try to gather more information about the motor functional area from various methods. All these data were analyzed and compared with each other. For example, SEP phase reversal allows intraoperative localization of the central sulcus but yields no functional information [37], and in recent years we use direct stimulation of the motor cortex to ensure intraoperative identification of motor areas. In most of the cases, the data of fMRI, SEP and intraoperative cortical mapping are highly concordant and complementary. However, the functional areas detected by these techniques can be different due to functional reorganization of the M1 [38], [39], [40]. When this happened, or SEP/cortical mapping provided equivocal information on the cortical representation of motor territories, we usually assume that the location of M1 detected by fMRI was correct, and performed surgery based on that information. This was based on the evidences that fMRI using alone was more reliable than SEP or brain mapping for the detection of M1 in proximity to a lesion [35].

4.3. Transsulcal approach 

The surface of the human brain cortex comprises several convoluted folds, the gyri, separated by slanted spaces, the sulci, and displays significant individual variation in morphology. As anticipated, transcortical and remote approaches always involve traumatic injury to surrounding brain tissue, thus, the transsulcus approach is the best way to be considered to minimize collateral damage, as working through these natural tunnels leads to safer surgery [41], [42], [43].

Generally the transsulcal approach should be considered according to the location of lesions and the spatial relationship between lesions and functional areas adjacent to the lesions. If the facial area is close to the lesion, dissection should be started from the facial area to the hand area to avoid retraction to cortex in the hand area.

Besides the precise location of the lesion and avoidance of functional areas for determining the entrance and direction of dissection, it is especially important to perform basic microsurgical techniques [8]. It is of critical importance to take advantage of the natural sulci of the brain and to continue the furthest possible dissection in the natural pathways. Usually, the veins integrated in the arachnoid of the sulcus, are freed by sharp dissection with the least coagulation, so that even small veins running along and then down into the deep sulcus can be preserved (Fig. 4C and H). A microsiphon over the cotton on the superficial cortex is suitable for transient and minimal brain retraction. A feasible part of the sulcus, with relatively less edema and adhesion between arachnoid membranes, should be located and opened in the first place. Then, the working space in the sulcus can be extended deeply and widely to the lesions. Self-retractors, if necessary, could be applied on the postcentral gyrus, not on the precentral gyrus only during the manipulation at the deep part of the sulcus, and they should be loosened frequently.

In summary, as a minimally invasive technique, transsulcal microsurgical approach based on anatomic and functional information allow surgeons to reach and remove lesions deep in the paracentral area with minimal brain tissue loss. In this series, this approach was successfully conducted without any neurological damage in all patients.

4.4. Treatment of concomitant epilepsy associated with cavernoma 

Epileptic seizures are the most frequent concomitant symptom of supratentorial cavernomas (40–70%) [44], [45]. In this series, 8/17 patients had clinical symptom of seizures.

The exact mechanisms of epilepsy associated with cavernoma are complex with no available consensus. There is no definite evidence that a space-occupying mass effect account for the high epileptogenicity of this vascular malformation [46]; rather, the microhemorrhages leading to a hemosiderin fringe and hemin deposits in the tissue surrounding the cavernoma are believed to be responsible for cavernoma associated epilepsy. Iron can induce epilepsy in different ways; in addition, reactive glial proliferation and altered synaptic activity may be epileptogenic [47].

The optimal surgical management of symptomatic epileptic seizures caused by cavernomas is still a matter of debate [48]. Most studies are in agreement that early operation, which reduces the preoperative duration of epilepsy, is associated with a better outcome [44], [49], [50], [51]. But whether microsurgical lesionectomy, including the removal of only the cavernoma alone, or a more extensive resection is necessary to achieve this aim remains controversial [47].

Based on the understanding of the main cause of epileptogenesis of cavernomas is pathologic changes in the adjacent brain tissue, many authors believe that a satisfactory therapeutic outcome requires complete elimination of the epileptogenic foci, with an extended lesionectomy including perilesional tissue if the ECoG reveals focal epileptic activity or a hemosiderin fringe is detected [44], [45], [52]. And the determination of surgical resection boundaries cannot rely merely on imaging techniques, but that functional examinations such as the EEG or MEG should be applied. However, some other studies failed to find any substantial evidence that the additional excision of the hemosiderin-stained tissue around the cavernoma provided better results than merely resection of the cavernoma [53], [54], and more invasive and costly studies to find and remove the epileptogenic cerebral parenchyma seem justified only after lesionectomy fails [55].

Because of these contradictory reports, neurosurgeons often are in a dilemma when dealing with cavernomas: on the one hand, they use a minimal surgical approach to avoid unnecessary damage to brain tissue; on the other hand, resections should be extensive enough to remove the total of epileptogenic tissue.

In our opinion, in contrast to other structural lesions such as dysembryoplastic neuroepithelial tumors or gangliogliomas, the vascular malformation itself is usually not responsible for the epilepsy in patients with cavernomas [47]. The complete removal not only of the cavernoma itself but also of the surrounding hemosiderin-stained brain tissue improves long-term outcome of seizures in symptomatic epilepsy, which has been confirmed by many studies [56]. But considering the risk/benefit of procedures in the paracentral area, we often remove the hemosiderin-stained tissue or surrounding epileptogenic brain tissue delineated by new electrophysiological investigation techniques on the side of frontal lobe or the postcentral gyrus, and keep a small proportion on the side of precentral gyrus. On the other hand, evaluating lesions in the vicinity of the central region with regard to symptomatic epilepsy, ECoG might be useful to confirm and monitor cortical seizure activity.