夏 雷 张宏伟 于春江 张明山 任 铭 曲彦明 王浩然 朱明旺 赵殿江 齐雪岭 姚 坤

Abstract : Object: To evaluate the clinical results, especially surgical outcome of the cysticVS with fluid-fluid level.

Methods: Forty-five cystic VS and 86 solid VS were enrolled and the former were divided into fluid-level and non-fluid-level group further. The clinical and neuroimaging features, intraoperative findings, and surgical outcomes of the tree groups were retrospectively compared.

Results: Peritumoral adhesion is significantly greater in the fluid-level group (70.8%) than in the non-fluid-level group (28.6%) and the solid group (25.6%) (P<0.0001).Complete removal was significantly fewer in the fluid-level group (45.8%) than in the non-fluid-level group (76.2%, 16/21) and the solid group (75.6%, 65/86) (p=0.015). Postoperative facial nerve function in the fluid-level group is less favorable than two other groups, Good/satisfactory facial nerve function 1-year after surgery was seen in 50.0% cases of the fluid-level group versus 83.3% cases of the non-fluid-level group (P=0.038).

Conclusions: The cystic VS with fluid-fluid level have more frequency to adhere to surrounding neurovascular structures and less favorable surgical outcome. The possible mechanism of peritumoral adhesion is intratumoral hemorrhage and consequent inflammatory reactions that lead to destruct the tumor–nerve barrier. These findings may be useful in predicting surgical outcome and making surgical strategy preoperatively.

Keywords: cystic vestibular schwannoma · fluid-fluid level · peritumoral adhesion

Cystic vestibular schwannoma (VS) is described as behaving more aggressively, with atypical initial symptoms, preoperative facial palsy, short clinical history, large size, unenlarged internal auditory canal and sudden deterioration because of rapid growth or unpredictable expansion of the cystic component or hemorrhage.

Cystic VS also present a therapeutic dilemma. Observation is not recommended for these tumors as the cystic component expand rapidly, resulting in severe mass effect and hydrocephalus6. As to radiotherapy, apart from the large tumor size at diagnosis and the cystic component, sudden deterioration resulting from expansion of the cystic component or hemorrhage after radiosurgery does not support this option too.

As to surgery, cystic VS also has worse prognosis because of the difficulty in preserving an adequate subarachnoid dissection plane, hypervascularity of the solid components, frequent engulfment of neurovascular structures, unusual cranial nerve displacement, substantially increased risk of accidental lesion of facial nerve and a greater tendency for postoperative bleeding as compared with solid VS.

But the results of other recent studies have not supported above conclusions. So we hypothesize that all cystic VS may not have worse surgical outcome and some undefined factors originating from cystic VS might be responsible for these different clinical and surgical courses. Interestingly, we found a part of cystic VS presented with fluid-fluid level, indicating

intratumoral hemorrhage, as previously reported. The aim of this study is to evaluate the clinical results of the cystic VS with fluid-fluid level, and to investigate whether the presence of fluid-fluid level in cystic VS may be relevant and important for predicting the surgical outcome. We also discussed the possible mechanism of peritumoral adhesion and formation of fluid-fluid level, and its implication for surgical strategy.

Methods

Patient Population

From April, 2008 to March, 2012, 224 cases of VS were admitted. The clinical data of all cases were reviewed retrospectively. The tumors were defined as cystic when the radiological evidence of cyst formation presented and the intraoperative findings revealed significant cystic components. The cases that were neurofibromatosis-Ⅱ or treated previously by radiosurgery or had recurrent VS were excluded from this study. Eventually, 45 cases of cystic VS and 86 solid VS were enrolled.

Preoperative Investigation

All routine preoperative investigations were performed. Special investigations included pure tone audiometry, speech discrimination test and auditory evoked potentials recordings. Neuroradiological investigations included bone window CT, MRI with contrast enhancement. These 45 cases of cystic VS were further divided into two groups according to whether fluid– fluid level had been seen on preoperative MRI or not. In order to detect small fluid-fluid level in cystic VS, we even used thin-slice T2-weighted 3D TSE pulse sequence with DRIVE. The tumor size was defined as its largest diameter in the cerebellopontine angle without considering the intracanalicular component of the lesion, which was measured and record by the operators (MW. Z. and DJ. Z.) on the radiological workstation.

Surgical Procedure

All cases were operated via suboccipital retrosigmoid approach by the senior author (CJ. Y.), who has performed more than 1000 cases of VS resection. Standard microneurosurgical techniques were employed with intraoperative neuroelectrophysiological monitoring with Medelec Synergy (Oxford Instruments Medical, UK) in all cases.

A complete resection was attempted in all cases. The facial nerve runs along the surface of the tumor and usually becomes elongated and thin. It must pay attention to identify it and dissect it along the arachnoidal plane. Sometime,it is difficult to preserve the arachnoidal plane, due to a stronger peritumoral adhesion. Moreover,the facial nerves may be displaced in a different

position, which may depend on the pattern of the cyst in VS. The presence of multiple cysts made it more difficult to identify the course of facial nerve and protect it during tumor dissection. The characteristics of the adhesion of tumor were carefully noted by reviewing the surgery and surgical video records. By the senior author (CJ. Y.), the tumor was considered to be adhesive if a subarachonid plane of dissection did not exist between the tumor capsule and peritumoral neurovascular structures, such as facial nerve, brain stem or PICA, which resulted in discontinuity of facial nerve or deliberately leaving small pieces of tumor remnant adhering to these important neurovascular structures; Otherwise it is nonadhesive if the subarachonid plane of dissection did exist and can be lent to blunt or sharp dissect the tumor capsule from facial nerve, brain stem or PICA totally.

Postoperative Outcome and Follow-up

Postoperative facial nerve function was assessed according to the House-Brackmann (HB) classification system at discharge (about two weeks postoperatively) and 1 year postoperative. All cases were analyzed for complications and postoperative course and followed up by repeated MRI and neurological examination, especially facial nerve function.

Statistical Analysis

With the statistics software of SPSS 12.0, One-way ANOVA analysis was used to determine statistical differences in the duration of symptoms and tumor size between cystic VS with or without fluid-level and solid VS. Chi-square test was used to determine the differences in the rate of multicyst, peritumoral adhesion, total resection and facial nerve outcome between cystic VS with or without fluid-level and solid VS. Significant differences was considered at p<0.05.

Results

Clinical Features

These groups consisted of 66 male and 65 female cases who were 13 to 79 years of age (mean 44.86 years). According to preoperative MRI, 24 cases of cystic VS with fluid-fluid level were regarded as fluid-level group. While, 21 cases without fluid-fluid level were regarded as non-fluid-level group. (Figure 1 A, B) It is noteworthy that thin-slice T2-weighted 3D TSE pulse sequence with DRIVE detected some small fluid-fluid levels in 7 cases, which had not been found in conventional MRI scan. (Figure 2 A, B) Moreover, obvious multicyst was seen in 19 cases (79.2%, 19/24) of the fluid-level group versus 10 cases (47.6, 10/21) of the non-fluid-level group (P=0.027). (Figure 3)

The clinical records of cystic VS were summarized in Table 1. The mean duration of symptoms was 41.49±11.13 (Mean±SE) months in the fluid-level group, 39.38±10.90 months in the non-fluid-level group and 51.92±7.83 months in the solid group(p = 0.644). The mean tumor size was 38.63±1.89 (Mean±SE) mm in the fluid-level group, and 39.14±2.41mm in the non-fluid-level group (p = 0.955). It was 32.42±1.11 mm in the solid group and significantly smaller than that of cystic VS (p=0.004)

Surgical Outcome

In the fluid-level group, 17(70.8%, 17/24) were considered to be adhesive, of which 14 adhered to facial nerve (Figure 4), 2 adhered to brain stem and 1 adhered to PICA. While in the non-fluid-level group, 6(28.6%, 6/21) were considered to be adhesive, of which 5 adhered to facial nerve and 1 adhered to PICA. And in the solid group, 22(25.6%, 22/86) were considered to be adhesive, of which 15 adhered to facial nerve, 5 adhered to brain stem and 2 adhered to both of them simultaneously. Peritumoral adhesion is significantly greater in the fluid-level group (70.8%) than in the non-fluid-level group (28.6%) and the solid group (25.6%) (P<0.0001). Complete removal was achieved in 45.8% (11/24) patients of the fluid-level group which was significantly lower than in the non-fluid-level group (76.2%, 16/21) and the solid group (75.6%, 65/86) (p=0.015). Five (20.8%, 5/24) in the fluid-level group, 1 (4.8%, 1/21) in the non-fluid-level group and 4 (4.7, 4/86) in the solid group had facial nerve broken, in spite of no significant difference (P= 0.058). However, of this case with facial nerve broken in the non-fluid-group, although MRI did not found any fluid-fluid level, there were other obvious hemorrhage appearances, as shown in Figure 5.

Postoperative Course and Follow-up

At discharge, the facial nerve function was evaluated in all cases. Reconstruction of the facial nerve was attempted in five cases with facial nerve broken, and all of these cases were graded as HB VI, even if they achieved some function. Facial nerve grades at discharge are compared between the tree groups, as shown

in Figure 6. Favorable facial nerve function (HB grades ranging from I to ⅡI) was seen in 62.5% (15/24) cases of the fluid-level group, 81% (17/21) cases of the non-fluid-level group and 62.8%(54/86) cases of the solid group(P= 0.273). The surgery-related significant complications included cerebellar hemorrhage in one case of the non-fluid-level group and one case of the solid group, which were re-explored in the immediate postoperative period. The follow-up period ranged from 4 to 54 months, with a mean followup of 26.2 months. Ninety-five cases underwent follow-up examination up to 1 year to evaluate the postoperative 1-year facial nerve function.Long-term facial nerve grades are compared between the tree groups, as shown in Figure 7. Favorable facial nerve function was seen in 50% (8/16) cases of the fluid-level group versus 83.3% (15/18) cases of the non-fluid-level group (P=0.038). It was seen in 67.8%(40/59) cases of the solid group(p=0.116). One case with HB grade V in the fluid-level group underwent Masseteric-facial nerve anastomosis. Another in the solid group underwent hypoglossal-facial nerve anastomosis. One case in the fluid-level group had demonstrable lesions on follow-up MRI scans. Because of very small tumor size and no symptom, she had been followed up for 19 months. Another case in the solid group had remnant lesion on repeated MRI scans and underwent stereotactic radiosurgery twice. In the other cases, no recurrence was found.

Discussion

The result of this study showed that the cystic VS with fluid-fluid level had more frequency to adhere tightly with surrounding neurovascular structures, less favorable surgical outcome.

Criterion of Cystic VS

The incidence of cystic VS varies from 5.7% to 48%, closing to 10% in more recent studies 23. However, the true incidence is debatable and it is depended on the criteria used. Most recent studies defined the tumors as cystic when the intraoperative findings revealed the presence of significant cystic elements, in addition to the preoperative radiological evidence of cyst formation. But, in the study of Charabi et al, they presented stricter criteria which even includes histological verification of the presence of S-100 positive membrane. In our series, we did not include any histological criteria. Histological verification of S-100 positive membrane is to differentiate peritumoral arachnoid cyst with cystic VS. But for experienced neurosurgeons,when they dissect the tumor capsule in the subarachnoid plane, it is not difficult to differentiate cystic wall and arachnoid cyst. Furthermore, Enhancement of the cyst wall is an imaging characteristic that can be used to differentiate cystic wall from arachnoid cysts.

Facial Nerve Outcome of Cystic VS There are still debates in the correlation between cystic VS and postoperative facial nerve outcome. In most studies, reported facial nerve outcomes after cystic VS resection were worse than the results of solid tumors of comparable size. This may be due to a tight adhesion of the nerve to the tumor surface, with difficulties in preserving the arachnoidal plane, unusual position of displacement depended on the pattern of development and growth of the cyst and the low resistance to traction after cyst-evacuation. However, other studies have not found any significant difference in facial nerve outcomes between cystic and solid tumors or different cystic tumor types. Lunardi et al. believed that aspiration of the cyst content facilitates recognition and preservation of the facial nerve and removal of the solid component thus ensuring a very good prognosis in all patients. While, Mehrotra et al. found that in cystic VS, facial nerve preservation was higher due to the faster and earlier decompression of the lesion that facilitated itsearlier identification. However, in those patients in whom the cystic component was in excess of 90%, preservation of the anatomical continuity of the facial nerve was difficult.

Peritumoral Adhesion of Cystic VS

As there have been conflicting views on whether postoperative facial nerve outcomes of cystic VS were significantly different from solid VS, we believe that apart from tumor size, cystic component, surgeon’s experience, some other factors affect the tumor’s adhesion to the surrounding neurovascular structures. The results of our study showed that all cystic VS did not adhere to surrounding neurovascular structures, but cystic VS with fluid-fluid level, indicating intratumoral hemorrhage, had more frequency to adhere to facial nerve, brain stem or PICA, and worse prognosis.

Certainly, adhesion of tumor (degree of adhesion of the tumor to surrounding neurovascular structures) is unable to measure because this is very subjective, judged by neurosurgeon during dissecting arachoidal plane.The rate of total resection is also affected by neurosurgeon’s experience. But the senior neurosurgeon of this study has performed more than 1000 cases of VS resection for 20 years.Deliberate subtotal tumor removal was performed only for adherence of the tumor to the important neurovascular structures, not for hypervascular tumor with risk of bleeding or intraoperative cerebellar swelling. So the criterion of adhesion in this study is: deliberate subtotal tumor removal or discontinuity of facial nerve resulted from disappearance of arachnoid plane. Therefore, in this study, tumor’s adhesion to surrounding neurovascular structures, judged by the same experienced enough neurosurgeon is relatively objective.

Possible Mechanism of Peritumoral Adhesion

According to the results of this study we hypothesis that repeated intratumoral hemorrhages significantly increase the risk of peritumoral adhesion in cystic VS.

VS initially happened some regressive changes which include: fatty-and hyalinedegeneration. These changes probably result in or companied with vascular abnormalities: hypervascularity, malformed vessels, hyalinized vessels, thrombosis, and then haemorrhages and haemosiderin deposition. Because of the vascular abnormalities, the incidence of intratumoral hemorrhage and multicystic formation of VS increased as the tumor grew. Intratumoral hemorrhage introduces massive inflammationary cells infiltration，especially macrophages and release proteonase, including MMS2, which destructs the tumor-nerve barrier proteolytically and increases perituomral adhesion. In this study, extensive lymphocytes, lipophages (foamy macrophages) and hemosiderin-laden macrophages were observed in the samples of the fluid-level group (data were not shown). Further, we found that most cystic VS with fluidfluid level were multicystic, with thin cystic wall on the surface of the tumors. We believe these multicysts were caused by repeated microhemorrhages and then multiple cysts on the surface increase significantly the chance of peritumoral adhesion with surrounding neurovascular structures, especially facial nerve, brain stem and PICA.

Fluid-fluid Level in Cystic VS

Fluid levels in tumours can be caused by haemorrhage from hypervascular neuromas or by collection of liquefacted necrotic material. But necrosis is usually companied with hemorrhage, while microhemorrhage is common in VS. By histologically analysed intracranial schwannomas with spontaneous hemorrhage, it was found that the majority had hypervascular, with dilated, thinwalled vessels and this resulted in hemorrhage and fluid-fluid level within the cystic tumor1. So we believe that the formation of the fluid–fluid level in cystic VS was due to hemorrhage.

CT and MRI scans, if performed in a gravitydependent plane, depict fluid-fluid levels by virtue of their high-contrast resolution. It is noteworthy that in our experience, thin-slice T2-weighted 3D TSE pulse sequence with DRIVE is more sensitive to detect small fluid-fluid levels. Seven cases of the fluid-level group would have gone missed without this technique.

Surgical Strategy

Although many important neurovascular structures may be found to adhere to VS during the resection, neurosurgeons usually focus their attention on the brainstem, the facial nerve, and the vessels in the posterior fossa.

In order to protect these important neurovascular structures, we advocate the strategy of looking for subarachnoid planes that lend themselves to blunt or sharp dissection. If the arachnoid plane is not well developed, we use the onionskin technique of tumor removal;each layer of VS is removed from the inner to outer layer until the outermost layer is reached and removed piecemeal. Finally, the thin capsule is lying over the facial nerve, the brainstem or vessels of the posterior fossa. At that time, we advocate subtotal resection in these areas, leaving portions of thin capsule and devitalization of remnant by low-power coagulation. This will decrease the risk of neurovascular damage, but not increase recurrent rate as shown in our follow-up data.

Limitations of the study

The retrospective collection of data might have introduced a selection bias. The tumors which were initially received radiotherapy were solid. Therefore, many solid tumors were excluded. The surgeon is impossible to blind to fluid-fluid group. For cases with cosmetic requirement, compromised strategy was adopted. As to surgical assessments, it is subjective to judge peritumoral adhesion. This source of

bias was minimized by using data from a single neurosurgeon who would apply a relatively consistent philosophy. In turn, single surgeon experience also limits in generalizing the results. As to follow-up assessments, the 1-year followup is not long enough. Some facial nerve function improving after 1-year were also found.

Conclusions

The cystic VS with fluid-fluid level had more frequency to adhere tightly with surrounding neurovascular structures, less favorable surgical outcome. The possible mechanism of peritumoral adhesion is intratumoral hemorrhage and consequent inflammatory reactions which lead to destruct the tumor-nerve barrier. MRI scan, especially thin-slice T2-weighted 3D TSE pulse sequence with DRIVE is more sensitive to detect small fluid-fluid level. These findings may be useful in predicting surgical outcome and making surgical strategy preoperatively.

Disclosure

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

(参考文献略)

评点：本文作者回顾性分析了三博脑科医院神外二、六病区3年多来切除囊性听神经瘤的手术结果。45例囊性听神经瘤中，其中有21例在术前MRI上有液平现象。术中发现有液平的囊性听神经瘤与周围神经血管结构粘连较重，肿瘤残留率也相应增加。术后，对这21例患者的的面神经功能进行评价，发现短期(2周)面神经功能要劣于无液平者，长期(一年)面神经功能则明显劣于无液平者。对其中5例囊性听神经瘤进行了细致的组织病理学观察发现，肿瘤内有明显的巨噬细胞细胞浸润、含铁血黄素沉着，提示肿瘤内部有微出血及炎症改变。作者提出有液平现象的囊性听神经瘤可能是由于肿瘤内部发生微小出现，随后产生炎症反应，并且波及到肿瘤表面，导致肿瘤与周围神经血管结构粘连较重。这一发现对听神经瘤术前评价和手术结果预测具有一定的指导意义。