Results
Preoperative Data
One hundred patients with intramedullary ependymomas underwent 102 operations. The mean age at clinical presentation was 44 ± 15 years (range 8–74 years), with an average clinical history of 35 ± 49 months. All but 3 patients were adults in this series. Eight patients were affected by neurofibromatosis Type 2 (NF2). Forty-seven ependymomas affected the cervical cord, 52 the thoracic cord, and 3 the conus. Fifty-eight percent of patients demonstrated a syrinx.
Table 2 lists the percentages for individual neurological symptoms to provide an overview of the clinical spectrum. There was a trend toward more severe neurological deficits for thoracic tumors compared with cervical tumors. Overall, 11.2% of patients were unable to walk, and 4.0% were incontinent at presentation. During the overall study period, preoperative McCormick grades increased significantly. Since 1991, 79% of patients presented in grade I compared with 48% prior to 1990. Cervical ependymomas were associated with a significantly lower McCormick grade compared with thoracic tumors (Table 3), even though there was no significant difference in patient preoperative histories.
Management
GTR was attempted in every patient. Overall, 88 operations (86.3%) resulted in GTR, while 7 tumors were removed subtotally, 5 partially, and 2 underwent a biopsy. Thirteen operations involved recurrent tumors. The GTR rate increased continuously from Period A (72%) and Period B (84%) to Period C (100%) (chi-square test: p = 0.03). The rates varied also according to surgical experience: surgeons in Group A completely removed 71.4%, while those of Group B resected 80% and those of Group C removed 90.1% (chi-square test: p = 0.03). Comparing tumors with and without syringomyelia revealed no significant difference in GTR rates (84% and 89% with and without syringomyelia, respectively). All tumors but 1 anaplastic ependymoma were classified as WHO Grade II.
In a multivariate analysis, independent factors predicting a GTR were determined (Table 4). The most important factors were a low preoperative McCormick grade and first surgery on the tumor. Other less important predictors were absence of arachnoid scarring between spinal cord and dura, a short clinical history, and an experienced surgeon.
Short-term Outcome
Complications were encountered in 17 instances (16.7%) (Table 5). This rate was irrespective of groups of surgeons or study period subgroups.
A transient, postoperative neurological deterioration was observed for 40.1% of patients. This rate did not correlate with surgical experience (43.8% for surgeons in Group A, 33.3% for Group B, and 43.7% for Group C; differences not significant) or the amount of tumor resection. A transient neurological deterioration was observed postoperatively for 43.2% of patients after GTR compared with 25% of patients after partial tumor removal (chi-square test: not significant), and 97% of affected patients had regained their preoperative neurological status after 6 months, with exceptional patients requiring a year or more for recovery.
Overall, permanent surgical morbidity was determined to occur at a rate of 27.5%. This resulted in a higher McCormick grade for 19.6% of patients. Looking at the different groups of surgeons, permanent morbidity did not decline with growing experience and remained unchanged throughout the overall study period. Significant statistical differences were detected according to the preoperative neurological state, spinal level of the ependymoma, and the presence of tumor hemorrhages. The lowest morbidity was associated with ependymomas of the cervical cord and McCormick Grade I. Surgery on 91 tumors in patients still able to walk resulted in a permanent loss of walking abilities after 15 operations (16.5%). Although not statistically significant, permanent morbidity increased for patients older than 60 years of age. There was no difference for tumors associated with or without syringomyelia (Table 6).
According to a multiple regression analysis, a low spinal level, advanced age, and a long history were the most influential independent risk factors for permanent surgical morbidity, apart from tumor hemorrhages (Fig. 3) and surgery on a recurrent tumor (Table 7).
(Enlarge Image)
Figure 3.
Case of a 65-year-old man with an ependymoma at the T8–11 level, with a 15-month history of local pain followed by slight gait ataxia and sphincter problems. A–E: The cystic tumor with evidence of hematomyelia due to a hemorrhage into the accompanying syrinx is shown in sagittal (A) and axial (B) T2-weighted MR images. The hemorrhage had caused no sudden aggravations of symptoms. Comparison of T1-weighted images before (C) and after gadolinium application (D and E) reveals patchy contrast uptake of the tumor. Postoperatively, the patient's neurological state worsened for 5 months, when he had reached his preoperative state again. He did not develop a neuropathic pain syndrome. The postoperative T1-weighted MR image with contrast (F) 2 years later showing no tumor remnant or recurrence after GTR, with slight adhesions between cord and dura on the axial T2-weighted image (G). No pain syndrome or myelopathy have developed.
The postoperative clinical results for the first 12 months are presented in Table 8. As a general rule, the individual scores declined postoperatively and improved toward the end of the 1st postoperative year. Overall, the preoperative McCormick grade remained the same for 69.1% of patients, while 6.2% showed an improvement. Due to the presence of more severe neurological deficits in patients with thoracic ependymomas, their neurological scores after 1 year remained worse compared with patients with cervical ependymomas. Whereas the postoperative courses for sensory functions, dysesthesias, and pain were quite similar, the differences between both tumor locations for motor function, gait, and bladder control had increased at that time point in favor of cervical ependymomas.
Long-term Outcome
Mean follow-up was 77 ± 91 months. The overall recurrence rate for benign ependymomas after 10 years was 7.4%, with all recurrences occurring within the first 2 years after surgery. The recurrence rate correlated significantly with the amount of tumor resection (4.2% after complete vs 18.5% after incomplete resections within 20 years; log-rank test: p = 0.03) (Fig. 4).
(Enlarge Image)
Figure 4.
Kaplan-Meier analysis of tumor recurrence rates for intramedullary ependymomas related to the amount of tumor resection (log-rank test: p = 0.03).
Seven patients died during follow-up. Two deaths were related to surgery within 4 and 6 weeks, respectively, while 1 patient, aged 64 years, died of complications related to his postoperative paraplegia 1 year later. One patient died at 48 years of age from cranial tumors associated with NF2 23 years after her spinal operation, and 3 patients died from unrelated causes 3, 6, and 26 years later at ages 51, 76, and 82 years, respectively. The corresponding survival rate for 20 years was 92% according to Kaplan-Meier analysis.
Apart from tumor recurrences, the quality of life after surgery for an intramedullary ependymoma may be compromised by a neuropathic pain syndrome or a postsurgical myelopathy.
A neuropathic pain syndrome was observed after 37.0% of operations and always started a few weeks or months after surgery. No significant differences for any subgroups could be detected. The rate of patients affected by neuropathic pain syndromes remained stable throughout the overall study period. In a multivariate analysis, preoperative neuropathic pain, permanent surgical morbidity, a low preoperative McCormick score, advanced age, and no tumor hemorrhage were the strongest predictors for this syndrome, while GTR, female sex, and postoperative tethering of the cord to the dura had less predictive power (Table 9).
Four percent of operations were followed by clinical signs of a myelopathy unrelated to a recurrent tumor growth. This phenomenon was seen exclusively in patients with postoperative fixation of the spinal cord to the dura and affected 10% of those with radiological evidence of postoperative cord tethering (Fig. 5).
(Enlarge Image)
Figure 5.
Case of a 54-year-old woman with an ependymoma at the C5–7 level, with a 2-year history of dysesthesias, pain, and a progressive tetraparesis. A: Sagittal T1-weighted MR image with contrast showing a solid tumor with contrast enhancement and irregular demarcation posteriorly and at both tumor poles. B: T2-weighted MR image demonstrating a more clearly defined tumor mass with hemosiderin caps on both tumor poles related to small hemorrhages. C: Postoperative sagittal T2-weighted MR image showing the complete resection with preservation of the gliosis containing the hemosiderin. D: Axial T2-weighted MR image showing the remaining cord tissue after resection and some hemosiderin close to the midline. Postoperatively, the patient regained her neurological functions after 5 months but developed a neuropathic pain syndrome and neurological signs of a progressing myelopathy 15 months after surgery. E and F: The corresponding T1-weighted MR image with contrast (E) and the axial T2-weighted MR image (F) 15 months after surgery demonstrating no tumor recurrence but profound cord atrophy and adhesions between cord and dura.