Child’s Nerv Syst (1999) 15: 17–28 © Springer-Verlag 1999

Trimurti D. Nadkarni Harold L. Rekate

Received: 15 October 1998

T. D. Nadkarni · H. L. Rekate (½) 1 Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona, USA Mailing address: 1 c/o Neuroscience Publications, Barrow Neurological Institute, 350 West Thomas Road, Phoenix, AZ 85013-4496, USA e-mail: [email protected] Tel.: +1-602-406-3593 Fax: +1-602-406-4104

REVIEW PAPER

Pediatric intramedullary spinal cord tumors Critical review of the literature

Abstract The objective of this review was to analyze the literature on the management of intramedullary spinal cord tumors to determine whether enough information was available for treatment guidelines to be established. Using standard computerized search techniques, databases containing medical literature were queried for keywords related to intramedullary spinal cord tumors, beginning in 1966. Of the 445 articles published in English and with potential relevance, only 75 articles were included in the final analyses. Based on the strength of their recommendations for the treatment of this controversial condition, articles were divided into class I, class II and class III data. There were no class I studies related to any aspect of the treatment of intramedullary spinal cord tumors. Based on this critical review of literature, gross total removal of

Introduction

Intramedullary spinal cord tumors are uncommon at any age and quite rare in children. These tumors account for 4–10% of central nervous system (CNS) tumors, essentially reflecting the volume of the spinal cord relative to the volume of the remainder of the CNS [10, 18, 19, 33, 66, 71]. Of all spinal cord tumors in children, 25–35% occur within the parenchyma of the spinal cord [9, 10, 71]. Astrocytomas represent 60% of childhood intramedullary tumors, most of which have a low histological grade. Ependymomas account for about 30% of intramedullary tumors, and 4% are developmental tumors such as teratomas, der-

an ependymoma confirmed by immediate postoperative magnetic resonance imaging and adjunctive treatment for high-grade tumors using radiotherapy, with or without chemotherapy, can be recommended as standards of therapy. With the strength of a guideline, radiotherapy should be withheld after gross total removal of intramedullary ependymomas and radical resection of low-grade intramedullary astrocytomas. The use of intraoperative ultrasonography and evoked potentials, important surgical adjuncts, can also be considered guidelines. The radical resection of intramedullary low-grade astrocytomas is an option. Key words Spinal cord tumor · Intramedullary · Child/children · Astrocytoma · Ependymoma · Radiotherapy · Surgery

moids, and epidermoid cysts [18, 59, 71]. These tumors can be focal or holocord, involving almost the entire spinal cord from the cervicomedullary junction to the conus medullaris. Such tumors usually have a solid component and rostral and caudal cysts. Intratumoral cysts can occur within the solid portion. The biological behavior of most intramedullary spinal cord tumors in children is slowly progressive, and their clinical presentation is insidious. The characteristic pattern of clinical presentation varies according to the anatomical site of the tumor, irrespective of tumor type. Occasionally, clinical diagnosis is difficult – because their evolution is insidious, their true nature may not be recognized until many years have passed. Operative techniques for

18

intramedullary spinal cord tumors have been reviewed elsewhere [10, 30, 52, 70, 71]. This study used techniques from evidence-based medicine to assess the following issues related to the treatment of pediatric intramedullary spinal cord tumors: (1) the feasibility of complete resection of astrocytomas and ependymomas, (2) the correlation between surgeon-defined extent of resection and postoperative neuroimaging, (3) the need for surgical adjuncts (e.g., ultrasonic aspiration, ultrasonography, the surgical laser, and evoked potential monitoring), (4) the relationship between extent of resection and outcomes, (5) the utility of treatment adjuncts (e.g., radiation therapy and chemotherapy), and (6) the incidence of postoperative spinal deformity and its treatment (laminotomy or laminectomy).

for educational purposes or for guiding the design of future studies. Options represent decisions that the physician may employ when faced with clinical decisions.

Results

Using the keywords, 445 articles with potential relevance were identified. Scanning of the titles and online abstracts excluded all but 74 articles, 17 of which dealt primarily or exclusively with children. There were no class I studies related to any aspect of the treatment of intramedullary spinal cord tumors. Feasibility of complete resection

Methods This critical review of the literature was derived from recognized databases of peer-reviewed scientific publications. All relevant articles were classified with respect to the strength of their recommendation(s). The methodology was derived from that described by the Joint Section on Trauma and Critical Care of the American Association of Neurological Surgeons and the Brain Trauma Foundation in their Guidelines for the Management of Severe Head Injury [38]. Using standard computerized search techniques, medical literature databases were queried for keywords related to intramedullary spinal cord tumors, beginning in 1966. The keywords used were spinal cord tumors, intramedullary, pediatric, child/children, astrocytoma, ependymoma, radiotherapy, and surgery. The titles of all identified articles were scanned for relevance, and potentially relevant articles published in English were obtained from the library of St. Joseph’s Hospital and Medical Center in Phoenix or by interlibrary loan and read at least cursorily. Several articles that were not identified in the databases but were referenced in the articles obtained were also relevant and included in the analysis. Articles discussing treatment were divided into class I (randomized control trials), class II (prospective reviews or retrospective data comparing two definable groups), and class III studies (everything else) to assess their utility for developing a consensus on the treatment of pediatric intramedullary spinal cord tumors. An article could contain both class II and class III data with reference to different aspects of management. Treatment recommendations were categorized as standards, guidelines, or options as follows. Standards represent accepted principles of patient management that reflect a high degree of clinical certainty. Typically, standards are based on class I evidence. Strong class II data may also form the basis of a standard, especially if an issue does not lend itself to randomized prospective trials, as in the case of intramedullary spinal cord tumors. Where standards exist, unless a case is somehow unusual or treatments have advanced significantly, clinicians must act in conformity with these standards. Guidelines represent a particular strategy or range of management strategies that reflect moderate clinical certainty. Guidelines are usually based on class II evidence or a preponderance of class III evidence. In this situation, the optimal choice is uncertain. If a clinician decides that the guideline is not the best approach, he or she must inform the patient of the rationale underlying the choice of a different treatment and then document the reasons. Otherwise, the clinician should follow the guidelines. The outcomes associated with options are uncertain. Options are usually based on class III evidence and are much less helpful, except

Forty-four studies were concerned with the extent of resection of intramedullary spinal cord tumors. There were 20 class II studies and 24 class III studies. Of these, 12 articles reported exclusively pediatric series. Epstein’s group [18, 19, 21, 29] consistently achieved gross total resection of low-grade astrocytomas. In their latest series of pediatric tumors, the gross total resection rate was 73% [29]. Garrido and Stein [28] achieved gross total resection of three of five low-grade astrocytomas. In their other two patients, removal was almost total (99%). Twelve other studies have reported the safe gross total removal of intramedullary astrocytomas in more than 50% of the patients [5, 8, 9, 12, 14, 18, 19, 21, 28, 29, 36, 73]. These series indicate that radical surgery for low-grade astrocytomas is feasible without incurring significant surgical morbidity. Furthermore, Constantani et al. [9] showed that intramedullary tumors can be radically and safely removed in children less than 3 years of age and in patients who had previously undergone biopsy. In five series [3, 16, 31, 35, 64], however, no patient underwent gross total resection. Furthermore, in 22 series [2–4, 7, 26, 31–33, 35–37, 39, 51, 52, 56, 59, 61, 64–66, 68, 70] gross total resection was reported for less than 50% of the patients. Thus, overall, low-grade astrocytomas and malignant gliomas have been treated predominantly by subtotal resection. Radical surgery is also feasible for malignant astrocytomas of the spinal cord [8, 70]. The same institute that previously advocated radical excision [8] has more recently reversed its opinion for patients with anaplastic astrocytomas, which are unlikely to improve and risk becoming worse [21]. All other workers, including those publishing on anaplastic tumors, advocated limited surgery for these poor-prognosis tumors [14, 16, 26, 32, 35, 37, 39, 45, 46, 49, 54, 56, 59, 61, 65–68]. The probability of achieving total resection appears to be significantly higher with ependymomas than with lowgrade astrocytomas. Five series [22, 24, 28, 52, 55] led to reports of gross total resection in all patients treated for

19

ependymomas, and another 13 series [4, 12, 14, 23, 29, 31, 32, 36, 48, 51, 67, 70, 73] to reports of gross total resection in more than 50% of their patients. Only for seven series did the authors report gross total resection in less than 50% of the patients with ependymomas [2, 3, 7, 33, 57, 59, 72]. Samii and Klekamp [67] concluded that the proportion of completely resected tumors was higher among patients with associated syrinxes. They found that intramedullary tumors, astrocytomas, or ependymomas that infiltrated spinal cord tissue and thus did not demonstrate cleavage planes were less often associated with syrinxes. Four articles [4, 21, 48, 73] discussed the major difficulties associated with surgeries (with or without biopsy) that followed radiotherapy. These articles suggested that reoperation on these patients was technically more difficult in terms of surgical planes. They also reported increased rates of postoperative morbidity.

Ultrasonic aspiration Nineteen articles (five concerning children) referred to the use of the ultrasonic aspirator for tumor resection [2, 4, 5, 8–10, 12, 13, 18, 19, 21–23, 29, 36, 55, 61, 71, 72]. Ferrante et al. [23] achieved total tumor resection in 9 of 11 cases (82%) with the use of the ultrasonic aspirator after 1980, as against a total resection rate of 65% (22 of 34) before that date without this instrument. This difference is not statistically significant at the 0.05 level. Also, the improvement in outcome, if any, cannot be attributed solely to the use of the aspirator, because the increased experience of the surgeon could also be involved. In contrast, gross total resection was achieved consistently without the use of the ultrasonic aspirator in five series [24, 28, 52, 54, 70]. Kalangu and Couto [40] also radically resected 20 intramedullary tumors using a two-stage technique without either an ultrasonic aspirator or laser.

Extent of resection and postoperative imaging Laser Essentially, the extent of resection has been defined by the operating surgeon’s estimates, a method with well-known pitfalls. Xu et al. [73] have defined several criteria for determining total tumor resection. First, the subarachnoid spaces obliterated by tumor compression are re-opened and cerebrospinal fluid (CSF) issues from them. Second, the rostral and caudal syrinxes communicate with the tumor bed. Third, the surface of the tumor bed is smooth and white or yellowish. Fourth, a bulging spinal cord sinks into the thecal sac after the tumor mass has been removed and the CSF drained. Finally, pulsations of the spinal cord rostral and caudal to the tumor bed are restored. Magnetic resonance (MR) imaging clearly defines the exact location and extent of the tumor and any associated cysts and other lesions associated with tumors. Only recently, however, have some surgeons used intraoperative ultrasonography and immediate postcontrast MR imaging to confirm the extent of residual tumor. In 13 series [8, 9, 14, 21, 22, 29, 48, 55, 59, 61, 67, 68, 73] the extent of resection was confirmed radiologically with postoperative MR imaging. Intraoperative ultrasonography has been used to confirm gross total removal in three series [8, 29, 68]. Four series [9, 22, 29, 48] documented the extent of resection with immediate postoperative MR imaging. Only two studies have correlated postoperative findings on MR imaging with the surgeon’s estimate of extent of resection [59, 73]. Basically, surgeons’ estimates are quite good in the case of ependymomas but of limited value when dealing with astrocytomas. Adjuncts to surgery The literature was also analyzed with reference to the use of various surgical adjuncts.

Fourteen articles reported the use of the CO2 surgical laser (light amplification by the stimulated emission of radiation) as an adjunct to the removal of intramedullary spinal cord tumors [2, 4, 8, 10, 12–14, 19, 21, 22, 34, 36, 71, 73]. Herrmann et al. [34] used the CO2 laser mounted on the operating microscope to perform excisions in 15 patients. They concluded that this tool was sufficiently safe and reliable for total removal of intramedullary tumors to be attempted with it, particularly those previously considered unresectable by the standard surgical technique. Ultrasonography In 16 articles it was reported that intraoperative ultrasonography improved the surgeon’s ability to define the extent of resection of intramedullary spinal cord tumors [8–10, 12, 13, 20–23, 29, 36, 42, 61, 62, 71, 73]. This modality was also useful for determining the extent of the tumor relative to rostral or caudal syrinxes. Evoked potential monitoring Twenty articles reported the intraoperative use of evoked potential monitoring for intramedullary tumors [2, 8–10, 12, 13, 15, 19, 21, 23, 29, 43, 47, 48, 54, 55, 58, 61, 70, 71]. One well-designed prospective study [43] of intramedullary spinal cord dissection led to the conclusion that postoperative neurological deficits were always predicted by significant intraoperative changes in somatosensory evoked potentials (SSEPs), which reflect the functional integrity of the dorsal columns. In this study numerous falsepositive recordings were observed but no false-negative

20

recordings. McCormick et al. [54, 55] found that intraoperative monitoring of SSEPs had limited value, and they use it primarily for investigative purposes. Steinbok et al. [71] no longer monitor evoked potentials, because they found that its use did not influence their intraoperative decision making. None of these three groups provided data or rationales for their conclusions in these reports [17, 54, 55]. Two other series [9, 47] monitored both SSEPs and motor evoked potentials (MEPs), which monitor the function of the corticospinal tracts. MEP monitoring is not yet universally available, but these two articles from the same center report that this modality is more sensitive than SSEPs for surgery involving the substance of the spinal cord. Outcomes The literature on intramedullary spinal cord tumors indicates that the histological grade of a tumor [29, 37, 39, 46, 56, 65, 72] and the patient’s preoperative neurological condition [5, 21, 23, 36, 37, 67] are statistically significant prognostic indicators. The impact of treatment is less clear, except in the case of ependymomas.

Almost all patients undergoing gross total resection of intramedullary spinal cord tumors experienced some postoperative deterioration of neurological function. Typically, this transient aggravation of a neurological deficit was followed by recovery within a few days (during hospitalization) or months [4, 29, 34, 36, 55, 67]. Epstein’s group [29] reported minor deterioration in all of their patients, but almost all (88.6%) also recovered to baseline or better within 1–3 months of physical therapy. In four other series [4, 29, 34, 67] it was also reported that all patients deteriorated transiently but subsequently recovered. Cristante and Hermann [14] found a clear relationship between the volume of the solid component (and thus the length of dorsal myelotomy required) and the severity of transitory functional impairment of dorsal column function. Innocenzi et al. [36] reported that 60% of their patients with astrocytomas and 50% of those with ependymomas showed some degree of transient neurological impairment after surgery. At discharge the proportion of patients who had worsened was higher in children with astrocytomas than in those with ependymomas. In six series [9, 14, 22, 52, 61, 70] there was more prolonged and sometimes permanent postoperative deterioration. Long-term outcomes

Operative mortality rates Three series reported significant postoperative mortality rates [23, 31, 32]. Ferrante et al. [23] have had an overall operative mortality rate of 13% (six patients), but no operative deaths since 1980. The six patients who died had transient complications related to the surgical wound (four CSF leaks and two infections). Four of these patients had a cervical tumor with its lower limit at C5. Guidetti et al. [32] reported an operative mortality rate of 5.4% (seven patients). Two patients died 10 and 15 days after surgery, respectively, from autopsy-confirmed pulmonary emboli. Three patients died from respiratory paralysis and two from purulent meningitis. In another series by Guidetti [31], seven patients (10%) with cervical spinal cord tumors died. Early in the postoperative period, two patients died from pulmonary embolism and two from respiratory paralysis related to ascending postoperative myelopathy. Two patients died of complications arising from purulent meningitis. The final patient had an intramedullary hemangioma and died intraoperatively from ventricular fibrillation while under hypothermia. Immediate postoperative deterioration Seventeen papers discussed postoperative deterioration [4, 5, 9, 14, 22, 23, 28, 29, 31, 32, 34, 36, 52, 55, 61, 67, 70]. Only two of these dealt primarily with children [10, 36].

The incidence of clinical improvement after surgery is higher in patients undergoing total resection than in patients undergoing partial resection [73]. In the former, surgical outcome is related to the patient’s preoperative clinical condition [37]. Thus, early surgery before the onset of profound neurological deficits has been recommended. In no series has a paraplegic or severely impaired patient shown any neurological recovery, regardless of the extent of resection [5, 10, 19, 22, 24, 36]. In contrast, patients with minimal deficits can improve significantly after operation [28]. Consequently, withholding surgical intervention from nonambulatory patients has been advocated [5]. In 44 children undergoing surgery, 32 of whom underwent gross total resection, two-thirds had mild to moderate deficits at a mean follow-up of 45 months. This is the largest series with gross total resection of low-grade tumors in children. It presents a compelling argument that radical surgery can improve the outcomes of patients with low-grade tumors [29]. The minimal residual deficits noted in this reported included foot drop, increased tone, or dysesthesias. Although these deficits did not greatly hamper these children’s overall functional ability, they were unable to participate in vigorous activities or sports. Nine series have led to the conclusion that the extent of resection for low-grade astrocytomas does not significantly influence prognosis [12, 14, 31, 35–37, 56, 59, 67]. Extended symptom-free survival has been possible after surgery even when resection has been incomplete [7, 66, 68].

21

Three studies on the role of radical resection of malignant tumors have failed to show any benefit associated with this modality [8, 12, 13]. Most of the patients experience rapid neurological deterioration leading to death. Cohen et al. [8] reported on 19 patients with malignant astrocytomas who underwent radical resection who had a median postoperative survival of 6 months. Radically resected anaplastic astrocytomas recur 7–10 months after surgery [14]. The outcome for malignant tumors has been dismal in most series. In 13 reports, no patient with an anaplastic tumor survived beyond 2 years [7, 8, 12–14, 21, 33, 37, 39, 49, 51, 65, 71], and leptomeningeal dissemination is frequent [8, 12, 13, 33, 35, 39, 45, 49, 71]. Despite aggressive chemotherapy and radiotherapy, glioblastomas invariably progress [9, 31, 32, 49, 52, 70]. Only four series of patients with high-grade tumors have included long-term survivors [9, 35, 59, 61]. Przybylski et al. [61] had four patients with anaplastic astrocytomas who survived for 14–18 years. Only one of these patients had undergone total excision, and three had undergone radiotherapy. One patient with a glioblastoma multiforme still survives 12 years after total excision and radiotherapy. None had chemotherapy. O’Sullivan et al. [59] reported that two children with a high-grade astrocytoma were alive 10 and 16 years, respectively, after diagnosis. The extent of resection was not specified, although both also received radiotherapy. Constantini et al. [9] reported that one child with an anaplastic astrocytoma was doing well 11 years after the first resection. Huddart et al. [35] have achieved a 5-year survival rate of 33% in patients with high-grade tumors treated with partial resection and radiotherapy. A maximal survival of 3 years after treatment of spinal cord glioblastomas multiforme has been reported [32, 46]. Alivisi et al. [3] reported a mean survival of 8.7 years for three patients with glioblastomas. Well-designed class II studies have found substantial benefits associated with total resection of intramedullary ependymomas [4, 12, 14, 23, 32, 36, 67, 72] The actuarial survival for ependymomas was 219 months for patients undergoing gross total resection, as opposed to 130 months for those undergoing subtotal resections [36]. Whitaker et al. [72] reported that patients with complete excisions had a significantly better survival rate than those with incomplete excisions (P < 0.025). Recurrence rates Samii and Klekamp [67] suggested that the biology of the tumor is important in determining long-term postoperative outcomes, because even though most astrocytomas in their series were partially removed, only 18% had a clinical recurrence. Epstein’s group [30] also reported an 18% recurrence rate after a mean follow-up of 54 months in a series in which 73% of the patients underwent gross total resection. For low-grade astrocytomas, some authors report no

significant difference in recurrence rates between patients with gross total resections and those with subtotal resections of astrocytomas [29, 59, 68]. Even Epstein’s group, the most forceful advocates of gross total removal of intramedullary astrocytomas, failed to find significantly different recurrence rates between subtotal and gross total resections (P = 0.13) [29]. Four articles, however, have reported no recurrences in patients who underwent total resection of low-grade astrocytomas [18, 29, 61]. Przybylski et al. [61] reported that no patient treated with complete resection relapsed during a follow-up that ranged from 5 to 14 years. In contrast, 9 of 13 patients treated with less than total resection did relapse. This difference is significant (P = 0.029). Extensive surgery sometimes yields good results and long-term stability for patients with low-grade tumors [18, 21, 29, 70, 73]. Even holocord tumors can be radically excised, leaving patients with minimal deficits and an improved prognosis [18, 19, 29]. Samii and Klekamp [67] reported a lower recurrence rate for ependymomas with than those without syrinxes (7% vs 21%), but the finding was not statistically significant. In the case of ependymomas, a large number of class III studies [22, 24, 28, 29, 31, 48, 51, 52, 70, 73] support the class II studies and no data argue against gross total removal if it is feasible. For ependymomas, the time to recurrence appears to depend upon the volume of residual disease [28]. Adjuvant therapy We also attempted to analyze the usefulness of adjunctive therapies in the management of intramedullary spinal cord tumors. Radiotherapy Thirty-two studies dealt with the role of radiation therapy in the management of intramedullary spinal cord tumors [3, 8, 9, 12, 16, 17, 23, 24, 26, 31–33, 35–37, 39, 44, 46, 48, 49, 51, 53, 56, 59–61, 64–66, 68, 69, 72]. Of these studies, 15 found that radiotherapy was effective in the management of benign tumors and their recurrences [17, 26, 27, 31, 35, 39, 46, 48, 49, 56, 59, 60, 65, 69, 72]. The information in only four of these papers could be classified as class II data, and all of these strongly advocated the use of radiotherapy [26, 46, 56, 59]. Kopelson et al. [46] recommended radiotherapy not only for partially resected tumors but also after gross total resection. However, no comparisons showing that patients with total resections benefit from this treatment support their conclusion. In a class II study, Reimer and Onofrio [65] concluded that there was a statistically significant difference in survival (P = 0.02) between patients treated with

22

subtotal resection (100% at 84 months) and those treated with biopsy and radiation therapy (42.7% at 84 months). The authors, however, do not discuss the cause of death in their patients. O’Sullivan et al. [59] suggested that radiotherapy without resection can achieve long-term control of spinal astrocytomas or ependymomas in children. In fact, radiotherapy achieved local control of the tumor in 26 cases (84%) despite either grossly incomplete or no resection in 25 of these cases. Four patients (13%), 2 of whom died, developed a second malignant tumor within the field of radiation. Minehan et al. [56] reported improved survival in patients undergoing postoperative radiation therapy. However, the outcomes of patients who did not receive radiotherapy were negatively biased – 4 of 12 postoperative survivors did not receive radiotherapy because of their poor neurological condition. Another study provides indirect evidence of the efficacy of radiotherapy [26], because higher radiotherapeutic doses improved tumor control and increased the length of survival. Of patients who received less than 40 Gy, 77% died of recurrent tumor. Of those who received more than 40 Gy, however, 83% were alive 4.1–28.9 years after treatment. Seven studies revealed a favorable role for radiotherapy for intramedullary ependymomas [7, 17, 27, 46, 48, 60, 72], while five found a favorable role for radiotherapy for astrocytomas [35, 39, 46, 56, 65]. In five other series [9, 18, 19, 22, 29], radiotherapy was not used after gross total resection of either low-grade astrocytomas or ependymomas. Radiotherapy has been used to treat malignant tumors because other forms of treatment have failed to retard the disastrous effects of tumor growth. Among high-grade tumors, 33% remained controlled locally 3 years after radiotherapy; all patients had residual spinal tumor before irradiation [35]. This finding suggests that radiotherapy may help control the disease. The combination of aggressive surgical treatment, chemotherapy, and total neuraxis radiation has also been recommended [8, 36, 71]. Chemotherapy Chemotherapy has been reported as a palliative measure for malignant tumors in eight studies, three of which dealt with children. Occasionally, a patient’s symptoms have improved (reduced local pain) and shown a sustained objective response [72]. Both intravenous and intrathecal routes have been used. Vincristine, cyclophosphamide, and chloroethylcyclohexylnitrosourea (CCNU) have been used with some success [8–10, 24, 33, 54, 71, 72]. As stated, the outcome for these tumors has been poor despite adjunctive treatment, and no class II studies that would permit the selection of one form of treatment over another are available.

Incidence of spinal deformity and its treatment The incidence of spinal deformity after multilevel laminectomy varied from 25% to 41% [16, 65, 75] and correlated with the patient’s age (the younger the patient, the more likely its occurrence) and the level of laminectomy (100% for multiple-level cervical laminectomies, 36% for thoracic laminectomies, nil for lumbar laminectomies) [71, 74, 75]. The incidence did not correlate with either gender or the number of laminae removed [25, 75]. The presence of preoperative spinal deformities and the magnitude of a neurological deficit were additional risk factors [14]. Of 32 cases reported by Reimer and Onofrio [65], 13 patients developed a spinal deformity that was apparent radiographically a mean of 33 months after laminectomy. Eight patients (62%) needed orthopedic intervention 8 months to 12.3 years after laminectomy. In O’Sullivan’s series of 22 children, 15 (68%) reportedly developed kyphosis or kyphoscoliosis that affected spinal cord function [59]. Of these 15 children, 7 (47%) required spinal fusion. The frequency of spinal fusion increased with the proximal location of the tumor (cervical in 5 of 6, thoracic in 2 of 8, and no lumbar fusions). DeSousa et al. reported postoperative scoliosis in 10 of 81 patients [16]: 5 patients developed scoliosis, and preoperative scoliosis progressed in the other 5. Lunardi et al. [51] reported that 15 of 25 patients harbored a spinal deformity: 6 deformities developed postoperatively, while the other 9 represented progressions of preoperative spinal deformity. Conservative treatment of deformities consisted of postoperative external immobilization for 5 years or until growth was completed. In one series [29], 25 children (56%) had scoliosis at their initial presentation. Almost half (48%) required bracing and two-thirds (58%) needed fusion, even though all underwent a laminotomy instead of a laminectomy. It is unclear whether postoperative bracing is indicated after multiple-level laminectomies in children, but in one study [19] patients were managed with postoperative bracing for several years. None of these patients developed new spinal deformities. In a mean follow-up of 45 patients for 3.4 years, progressive postoperative kyphoscoliosis occurred in 3 of 20 patients who had an osteoplastic laminotomy and in 9 of 25 patients who had a simple laminectomy [10]. This difference is not statistically significant. All patients who progressed after laminoplasty had recurrent disease. Over 10 years, orthopedic correction appeared necessary in half of these patients. Thus, osteoplastic laminotomy did not prevent the postoperative evolution of spinal deformity. The almost complete lack of data comparing the late outcomes of replacing or not replacing laminar arches precludes any conclusion on the benefit of laminotomy.

23

Table 1 Summary of treatment recommendations for pediatric spinal cord tumors (MRI magnetic resonance imaging, SSEPs somatosensory evoked potentials, MEPs motor evoked potentials) Level of recommendation

Treatment recommendation

Standards

Resect ependymomas totally Reoperate on ependymomas if postoperative MRI shows unexpected residual tumor Withhold radiation therapy if gross total resection is achieved

Guidelines

Withhold radiation therapy if radical or total resection of low-grade astrocytomas is achieved Follow extent of resection with intraoperative ultrasonography Monitor SSEPs to improve safety of surgery Treat malignant astrocytomas with postoperative irradiation

Recommended options

Attempt total resection of intramedullary spinal cord tumors if a cleavage plane exists Reoperate for recurrences of ependymomas and low-grade astrocytomas in ambulatory patients Use ultrasonic aspiration as surgical adjunct Monitor MEPs to improve safety of surgery Withhold radiation therapy for low-grade astrocytomas before tumor progression is clear Use osteoplastic laminotomy to decrease postoperative spinal deformity

Discussion

The literature reviewed here reflects the effects of the advances in technology over the past 30 years. The widespread availability of contemporary neuroradiological techniques such as computerized tomography and MR imaging has improved our understanding of the extent of spinal cord involvement and the anatomical relationship of tumor with the spinal cord itself. Other advances, including ultrasonic aspiration, the laser, evoked potential monitoring, and intraoperative ultrasonography, have increased the likelihood that radical surgery can be performed with reasonable safety. The era of mere decompression and occasional biopsy has been replaced with more radical attempts at excision. However, the potential survival benefit associated with radical resection must be balanced against the increased risk of morbidity (Table 1). Extent of resection and outcomes Surgical exploration is mandatory for any child with a progressively symptomatic intramedullary tumor. The extent of resection is a key issue in the surgical management of these tumors. Logic and the principles of neuro-oncology as they relate to low-grade tumors support the hypothesis that as much tumor should be removed as possible with-

out untoward functional cost. If an intramedullary tumor has been totally resected, no other treatment is required. This should be viewed as a standard for ependymomas and a guideline for low-grade astrocytomas. Although the efficacy of radical surgery has been established, information to suggest the duration of remission or the likelihood of permanent cure, which can be known only after many years of follow-up and retrospective analysis, is still limited. Unpredictable tumor biology, especially for low-grade astrocytomas, remains an important hurdle to be overcome. The insidious biological behavior of low-grade intramedullary spinal cord tumors is an important concern, because many patients with partially resected tumors achieve prolonged survivals with good neurological function even without adjuvant therapy. Unfortunately, histological analysis is still inadequate to define which patients are likely to survive long term without a recurrence. The principles of oncology suggest that a gross totally excised tumor should not recur, as is usually the case for ependymomas. The evidence, however, is far less compelling for astrocytomas. In fact, some studies [12, 14, 31, 35–37, 56, 59, 67] have led to the conclusion that surgery for lowgrade astrocytomas should be quite limited. In at least part of their course through the spinal cord, low-grade astrocytomas are locally infiltrating. In some areas the surgeon can define a plane; in others it becomes impossible to delineate the interface. Radical resection then greatly depends on the operating neurosurgeon’s surgical judgment and skill, which can improve with experience. Malignant varieties of astrocytomas tend to be more vascular and hemorrhagic and lack a distinguishable plane between the tumor and spinal cord. The tumor is often exophytic in an eccentric position at the dorsal or lateral surface of the spinal cord [70]. Occasionally, juvenile pilocytic astrocytomas also have a clearly demarcated plane. Most data favor an attempt at total resection if the astrocytoma has a cleavage plane. Radical surgery is compatible with good functional outcome and recovery, and profound permanent deficits are rare. The data comparing recurrence and survival against extent of resection in patients with low-grade astrocytomas are contradictory. None of the papers reviewed showed statistically significant differences between patients with low-grade astrocytomas treated with gross total removal and similar groups treated with alternative forms of treatment, such as biopsy or subtotal resection followed by radiation. Enthusiasm for attempting gross total removal of low-grade astrocytomas is relatively recent. Very late outcomes for this group have yet to be published in large enough numbers to obtain statistical significance. Proponents of radical resection suggest that children undergoing radical resection of low-grade astrocytomas should be observed and receive no adjuvant therapy, and that a recurrence could be safely treated with radical surgery. Overall, insufficient data exist to recommend a standard or guideline on the extent to which intramedullary spinal

24

cord astrocytomas should be removed. If a realistic cleavage plane exists, attempts at resection should be viewed as an option and other approaches can still be justified. When a good cleavage plane cannot be discerned or the dissection leads to a decrease in evoked potentials, subtotal removal is an acceptable outcome. For malignant astrocytomas, there is no benefit to reexploration and adjunctive treatment is definitively necessary as a standard. If the tumor is a low-grade astrocytoma there are two options, because of its indolent nature. The patient can be observed for symptomatic recurrence, or adjunctive therapy with radiation therapy can be pursued. Reoperation for a recurrent low-grade astrocytoma is also an option. In ependymomas, the surgical plane is usually well defined and obvious, and surgeons’ approaches should be more uniform than when astrocytomas must be resected. The relationship between length of survival and extent of resection of ependymomas is clear [14, 22, 23, 31, 32, 36, 67]. Gross total resection of ependymomas may be feasible without surgical morbidity and offers the best neurological outcomes (functional results) and long-term survivals free of recurrence. When an ependymoma has been totally removed, tumor recurrence is exceptional [7, 12, 14, 22, 23, 29, 31, 36, 48, 55, 67, 70, 72]. For the treatment of intramedullary ependymomas, the strength of the published data therefore supports a standard of care. The goal of surgery for intramedullary ependymomas should be to remove the tumor in its entirety. What should be done if residual tumor is found on postoperative imaging studies after an attempt to resect an intramedullary spinal cord tumor completely? The answer depends on two factors. If the tumor cannot be fully removed because the surgical plane cannot be defined or because the evoked potential decreased intraoperatively, there is little reason to believe that reexploration would significantly improve the extent of resection. The surgeon may then elect to use adjunctive treatments or to monitor the patient. If, however, residual tumor is found unexpectedly on a postoperative imaging study and more extensive surgery is possible, the decision, to some extent, depends on the cell type and the growth pattern of the tumor. If surgery was discontinued only because the tumor was “missed,” the patient should be returned to surgery to complete the resection. If the growth pattern is static or indolent, neurological deficits may improve or remain static for many years. The absence of scarring from previous surgery or radiation bestows the best chance of achieving a successful radical excision. Scar tissue and the possible changes caused by radiation make secondary surgical procedures difficult. Nonetheless, a dissection plane can still be developed around a tumor and total excision achieved in some patients who have undergone previous treatments [18, 19].

Resection confirmed by postoperative imaging Follow-up MR imaging is important not only to document residual tumor but also to detect asymptomatic recurrence. Constantani et al. [9] reoperated on patients with substantial residual tumor on MR imaging or tumor recurrence. Cristante and Herrmann [14] suggested that minute residues of astrocytomas with their ill-defined margins and patchy gadolinium enhancement on postoperative studies may be beyond the resolution of MR imaging. In the case of low-grade astrocytomas, even when MR imaging confirms gross total removal, there is little question that tumor fragments remain. Surgical adjuncts Ultrasonic aspiration Ultrasonic aspiration is useful for rapidly removing the core of intramedullary tumors with little attendant risk to adjacent vital structures. It is capable of discrete removal of a broad range of tissue with minimal distortion of adjacent tissues not in direct contact with the tip. Laboratory studies have demonstrated that long-term function and blood flow in the adjacent neural tissue is maintained beyond a 1-mm radius of the vibrating tip [11]. The instrument is used to debulk the tumor from within until the white matter interface is reached [10, 29, 71]. The ultrasonic aspirator is reported to be very useful in the radical resection of intramedullary spinal cord tumors, but such radical resection can also be achieved using more traditional microsurgical techniques. The use of ultrasonic aspiration should be considered an option. Laser As a surgical tool, lasers capitalize on precision that minimizes trauma, both thermal and mechanical, to surrounding structures. This precision of incision is exploited for myelotomy, whereas precision of vaporization and coagulation is used to remove tumor remnants at the glia-tumor interface. After an intramedullary tumor has been rapidly debulked by an ultrasonic aspirator, a neurosurgical laser may be used to remove residual fragments. The laser, however, also has some disadvantages [10]. It is time consuming to debulk a large volume of tumor. The resultant char also makes it difficult to recognize the glia-tumor interface and mandates frequent interruptions of the ongoing dissection while blackened tissues are gently removed with a small-caliber suction instrument. Overall, enthusiasm for the use of the surgical laser is diminishing because the procedure is lengthy. Its use can also be considered an option.

25

Ultrasonography Ultrasonography has been used for the intraoperative localization (level and extent) of the solid component of tumors, especially holocord tumors. The laminectomy and dural incision can thus be limited and executed exactly to expose the solid portion of the tumor accurately. Ultrasonography can be used to monitor the extent of resection, to confirm complete tumor removal, and to define areas of cyst and to ensure their drainage. Occasionally, artifacts within the operative bed limit the usefulness of this surgical adjunct. Ependymomas and astrocytomas have different echo patterns. Ependymomas are uniformly echogenic and are centrally located, uniformly expand the spinal cord, and are typically associated with syrinxes. Astrocytomas have a variable echo pattern and expand the spinal cord asymmetrically. Thus, sonographic images vary according to different tumor types and hence can help to determine the surgical strategy for different tumors [20, 42]. The use of ultrasonography should be considered a guideline. Evoked potential monitoring Since the intraoperative monitoring of SSEPs relies on the integrity of the dorsal columns, preoperative abnormalities of the patient’s position sense often preclude recording of these potentials. The theoretical possibility that severe abnormalities of the corticospinal tracts and other more anterior parts of the spinal cord could be present without affecting the signal recorded from the dorsal columns suggests that this form of monitoring may not be sensitive enough to detect damage to the intramedullary spinal cord. Although not specific, SSEPs are still very sensitive to damage to the spinal cord. They alert surgeons to potential damage to the spinal cord unless a procedure is modified or abandoned. Decreases in the amplitude of sensory evoked potentials have been noted during myelotomy, placement of pial sutures, use of laser for more than 20 s, and dissection of tumor in a specific location. In such cases, the manipulations are interrupted and the electrical activity is permitted to recover before the procedure is resumed [10]. SSEPs should be monitored to minimize the risk to the patient during tumor removal, and this can be regarded as a guideline. A combination modality of evoked potentials seems to be necessary to monitor axonal integrity in a compromised spinal cord. Theoretically, monitoring of both SSEPs and MEPs provides sensitive information regarding function in real time and can be updated every few seconds. These potentials are sensitive to anesthesia and surgical manipulation and require an experienced electrophysiologist to monitor and interpret [10, 29]. Overall, MEPs appear to be more predictive of postoperative outcome. Epstein does not push dissection beyond a 50% decrease in the ampli-

tude of MEPs. If such a decline occurs, manipulation of the spinal cord in this area is halted to allow the potentials to recover [10, 47, 58]. At present the limited clinical availability of this modality and the limited number of controlled studies on intraoperative monitoring of MEPs make its use an option. Adjuvant therapy Until recently, most spinal cord tumors were treated by biopsy followed by radiation therapy. There is compelling evidence that this form of treatment has been successful in a large number of cases. No controlled trials have compared radical surgery with biopsy and radiation therapy, however, nor is it likely that such studies will ever be done. Even studies in which patients who have undergone extensive surgery are randomized to early or late radiation therapy are difficult, if not impossible, to perform because of the indolent course of the disease. In most patients, the brief, devastating course of high-grade tumors does not seem to be altered by the combined use of radical surgery, radiotherapy and chemotherapy. Their prognosis is uniformly dismal, and their postoperative course tends to be disastrous with rapid growth and seeding of the tumor. The role of radiotherapy in the treatment of pediatric intramedullary spinal cord tumors remains the most controversial aspect of the management protocol. Individual neurosurgeons’ strong biases against radiotherapy have further weakened the case for radiotherapy. Radiotherapy is not curative but is effective in yielding a long-term, symptom-free survival. It also has potential serious complications, especially in children. It definitely has a role in the therapeutic armamentarium in treatment of spinal cord intramedullary tumors. Kopelson [44] has discussed the tolerance of the spinal cord for radiotherapy. Tumoricidal doses can be given to patients with astrocytomas (4500–5000 rad in 180- to 200-rad fractions) and those with ependymomas (4000– 4500 rad in 180- to 200-rad fractions) with the expectation of no increased risk of chronic radiation myelopathy, even though long spinal segments (often involving thoracolumbar segments) are irradiated. Radiation therapy should be withheld in the case of ependymomas and low-grade astrocytomas that have undergone gross total resection; this statement represents a standard for ependymomas and a guideline for low-grade astrocytomas. Radiation should be used after surgery for malignant astrocytomas, and this recommendation also represents a guideline. Radiation therapy should be used to treat patients at the time of recurrence unless total surgical removal is attempted. This strategy represents a guideline. Withholding radiation therapy from patients with low–grade astrocytomas until a recurrence is clear is an option.

26

Chemotherapy is essentially used only as a salvage procedure for high-grade tumors. Cohen et al. [8] has advised the following chemotherapy for malignant astrocytomas: 1,3-bis(2-chloro-ethyl)-1-nitrosurea (BCNU) or multiagent therapy consisting of prednisone, 1-(2-chloroethyl)1-nitrosourea (CCNU) and vincristine or a protocol combining eight drugs given within 1 day (“8-in-1” therapy). Children with anaplastic cerebral gliomas may benefit from the addition of CCNU, vincristine, and prednisone after radiotherapy. Use of this type of chemotherapy has been extended to children with anaplastic tumors [33]. None of the presently available chemotherapy agents or regimens have altered the outcome of patients with malignant intramedullary spinal cord tumors, however. The use of chemotherapy for high-grade tumors can be recommended as an option. All other aspects of surgical decision making in the management of intramedullary spinal cord tumors should be considered options. The options should be discussed as such with the patients, and the reasons for selection of any of these options should be documented in the medical record. Laminectomy vs laminotomy for spinal deformity Spinal deformities often occur in patients with spinal cord tumors [10, 16, 18, 65, 71]. In children such deformity is a well-documented complication of laminectomy. Kypho-

sis is the most common postoperative spinal deformity and results from vertebral body subluxation [10, 19, 25, 50, 71, 74, 75]. Spinal deformity appears to be the result of many factors. Laminectomy can lead to the loss of posterior supporting elements. Paraspinal muscles may become weak as the result of a neoplasm, and spinal irradiation can affect vertebral endplates [6, 10, 29, 41, 63, 71]. Spinal deformity may be unpreventable in pediatric patients with intramedullary spinal cord tumors. Osteoplastic laminotomy permits the replacement of bone, which is the nidus for subsequent osteogenesis, posterior fusion, and protection against future postoperative spinal instability [1, 63]. The efficacy of laminotomy in achieving this goal has not been established [29]. Without stronger supportive information from clinical studies, laminotomy should be considered an option. Spinal deformity remains a major complication of intramedullary spinal cord tumors and their surgical treatment. No strategy is completely effective in preventing this complication, but enthusiasm for the use of osteoplastic laminotomy seems to be increasing. Symptoms include spinal pain, limitation of spinal movement, torticollis, or obvious spinal deformity – kyphosis or scoliosis. In severe cases, kyphosis can compress the spinal cord and cause progressive myelopathy. A spinal deformity can occur months or years after surgery. Consequently, patients should be followed up throughout childhood and adolescence for the development or aggravation of spinal deformity, and this recommended follow-up is a guideline.

References 1. Abbott R, Feldstein N, Wisoff JH, Epstein FJ (1992) Osteoplastic laminotomy in children. Pediatr Neurosurg 18: 153–156 2. Ahyai A, Woerner U, Markakis E (1990) Surgical treatment of intramedullary tumors (spinal cord and medulla oblongata). Analysis of 16 cases. Neurosurg Rev 13: 45–52 3. Alvisi C, Cerisoli M, Giulioni M (1984) Intramedullary spinal gliomas: long-term results of surgical treatments. Acta Neurochir (Wien) 70: 169–179 4. Brotchi J, Dewitte O, Levivier M, Balériaux D, Vandesteene A, Raftopoulos C, Flament-Durand J, Noterman J (1991) A survey of 65 tumors within the spinal cord: surgical results and the importance of preoperative magnetic resonance imaging. Neurosurgery 29: 651–657

5. Brotchi J, Noterman J, Baleriaux D (1992) Surgery of intramedullary spinal cord tumours. Acta Neurochir (Wien) 116: 176–178 6. Cattell HS, Clark GL Jr (1967) Cervical kyphosis and instability following multiple laminectomies in children. J Bone Joint Surg [Am] 49: 713–720 7. Chigasaki H, Pennybacker JB (1968) A long follow-up study of 128 cases of intramedullary spinal cord tumours. Neurologia 10: 25–66 8. Cohen AR, Wisoff JH, Allen JC, Epstein F (1989) Malignant astrocytomas of the spinal cord. J Neurosurg 70: 50–54 9. Constantini S, Houten J, Miller DC, Freed D, Ozek MM, Rorke LB, Allen JC, Epstein FJ (1996) Intramedullary spinal cord tumors in children under the age of 3 years. J Neurosurg 85: 1036–1043 10. Constantini S, Epstein FJ (1996) Intraspinal tumors in infants and children. In: Youmans JR (ed) Neurological surgery. Saunders, Philadelphia, pp 3123–3133

11. Constantini S, Epstein FJ (1996) Ultrasonic dissection in neurosurgery. In: Wilkins RH, Rengachary SS (eds) Neurosurgery. McGraw-Hill, New York, pp 607–608 12. Cooper PR (1989) Outcome after operative treatment of intramedullary spinal cord tumors in adults: intermediate and long-term results in 51 patients. Neurosurgery 25: 855–859 13. Cooper PR, Epstein F (1985) Radical resection of intramedullary spinal cord tumors in adults. Recent experience in 29 patients. J Neurosurg 63: 492–499 14. Cristante L, Herrmann H-D (1994) Surgical management of intramedullary spinal cord tumors: functional outcome and sources of morbidity. Neurosurgery 35: 69–76

27

15. Deletis V (1993) Intraoperative monitoring of the functional integrity of the motor pathways. In: Devinsky O, Beric´ A, Dogali M (eds) Electrical and magnetic stimulation of the brain and spinal cord. Raven Press, New York, pp 201–214 16. DeSousa AL, Kalsbeck JE, Mealey J Jr, Campbell RL, Hockey A (1979) Intraspinal tumors in children. A review of 81 cases. J Neurosurg 51: 437–445 17. Di Marco A, Griso C, Pradella R, Campostrini F, Garusi GF (1988) Postoperative management of primary spinal cord ependymomas. Acta Oncol 27: 371–375 18. Epstein F, Epstein N (1981) Surgical management of holocord intramedullary spinal cord astrocytomas in children. Report of three cases. J Neurosurg 54: 829–832 19. Epstein F, Epstein N (1982) Surgical treatment of spinal cord astrocytomas of childhood. A series of 19 patients. J Neurosurg 57: 685–689 20. Epstein FJ, Farmer J-P, Schneider SJ (1991) Intraoperative ultrasonography: an important surgical adjunct for intramedullary tumors. J Neurosurg 74: 729–733 21. Epstein FJ, Farmer J-P, Freed D (1992) Adult intramedullary astrocytomas of the spinal cord. J Neurosurg 77: 355– 359 22. Epstein FJ, Farmer J-P, Freed D (1993) Adult intramedullary spinal cord ependymomas: the result of surgery in 38 patients. J Neurosurg 79: 204–209 23. Ferrante L, Mastronardi L, Celli P, Lunardi P, Acqui M, Fortuna A (1992) Intramedullary spinal cord ependymomas – a study of 45 cases with longterm follow-up. Acta Neurochir (Wien) 119: 74–79 24. Fischer G, Mansuy L (1980) Total removal of intramedullary ependymomas: follow-up study of 16 cases. Surg Neurol 14: 243–249 25. Fraser RD, Paterson DC, Simpson DA (1977) Orthopaedic aspects of spinal tumours in children. J Bone Joint Surg [Br] 59: 143–151 26. Garcia DM (1985) Primary spinal cord tumors treated with surgery and postoperative irradiation. Int J Radiat Oncol Biol Phys 11: 1933–1939 27. Garrett PG, Simpson WJK (1983) Ependymomas: results of radiation treatment. Int J Radiat Oncol Biol Phys 9: 1121–1124 28. Garrido E, Stein BM (1977) Microsurgical removal of intramedullary spinal cord tumors. Surg Neurol 7: 215–219

29. Goh KYC, Velasquez L, Epstein FJ (1997) Pediatric intramedullary spinal cord tumors: is surgery alone enough? Pediatr Neurosurg 27: 34–39 30. Greenwood J Jr (1967) Surgical removal of intramedullary tumors. J Neurosurg 26: 276–282 31. Guidetti B (1967) Intramedullary tumours of the spinal cord. Acta Neurochir (Wien) 17: 7–23 32. Guidetti B, Mercuri S, Vagnozzi R (1981) Long-term results of the surgical treatment of 129 intramedullary spinal gliomas. J Neurosurg 54: 323– 330 33. Hardison HH, Packer RJ, Rorke LB, Schut L, Sutton LN, Bruce DA (1987) Outcome of children with primary intramedullary spinal cord tumors. Child’s Nerv Syst 3: 89–92 34. Herrmann H-D, Neuss M, Winkler D (1988) Intramedullary spinal cord tumors resected with CO2 laser microsurgical technique: recent experience in fifteen patients. Neurosurgery 22: 518–522 35. Huddart R, Traish D, Ashley S, Moore A, Brada M (1993) Management of spinal astrocytoma with conservative surgery and radiotherapy. Br J Neurosurg 7: 473–481 36. Innocenzi G, Raco A, Cantore G, Raimondi AJ (1996) Intramedullary astrocytomas and ependymomas in the pediatric age group: a retrospective study. Child’s Nerv Syst 12: 776–780 37. Innocenzi G, Salvati M, Cervoni L, Delfini R, Cantore G (1997) Prognostic factors in intramedullary astrocytomas. Clin Neurol Neurosurg 99: 1–5 38. Joint Section on Trauma and Critical Care of the American Association of Neurological Surgeons and the Brain Trauma Foundation (1995) Guidelines for the management of severe head injury. American Association of Neurological Surgeons, Park Ridge, Ill 39. Jyothirmayi R, Madhavan J, Nair MK, Rajan B (1997) Conservative surgery and radiotherapy in the treatment of spinal cord astrocytoma. J Neurooncol 33: 205–211 40. Kalangu KKN, Couto M-T (1996) Radical resection of intramedullary spinal cord tumors without cavitron ultrasonic aspirator or CO2 laser: a “two stage” technique. Surg Neurol 46: 310–316 41. Katzman H, Waugh T, Berdon W (1969) Skeletal changes following irradiation of childhood tumors. J Bone Joint Surg [Am] 51: 825–842 42. Kawakami N, Mimatsu K, Kato F (1992) Intraoperative sonography of intramedullary spinal cord tumour. Neuroradiology 34: 436–439

43. Kearse LA Jr, Lopez-Bresnahan M, McPeck K, Tambe V (1993) Loss of intraoperative somatosensory evoked potentials during intramedullary spinal cord injury predicts postoperative neurologic deficits in motor function. J Clin Anesth 5: 392–398 44. Kopelson G (1982) Radiation tolerance of the spinal cord previously damaged by tumor and operation: long term neurological improvement and time-dose-volume relationships after irradiation of intraspinal gliomas. Int J Radiat Oncol Biol Phys 8: 925–929 45. Kopelson G, Linggood RM (1982) Intramedullary spinal cord astrocytoma versus glioblastoma. The prognostic importance of histologic grade. Cancer 50: 732–735 46. Kopelson G, Linggood RM, Kleinman GM, Doucette J, Wang CC (1980) Management of intramedullary spinal cord tumors. Radiology 135: 473–479 47. Kothbauer K, Deletis V, Epstein FJ (1997) Intraoperative spinal cord monitoring for intramedullary surgery: an essential adjunct. Pediatr Neurosurg 26: 247–254 48. Lee TT, Gromelski EB, Green BA (1998) Surgical treatment of spinal ependymoma and post-operative radiotherapy. Acta Neurochir (Wien) 140: 309–313 49. Linstadt DE, Wara WM, Leibel SA, Gutin PH, Wilson CB, Sheline GE (1989) Postoperative radiotherapy of primary spinal cord tumors. Int J Radiat Oncol Biol Phys 16: 1397–1403 50. Lonstein JE (1977) Post-laminectomy kyphosis. Clin Orthop 128: 93–100 51. Lunardi P, Licastro G, Missori P, Ferrante L, Fortuna A (1993) Management of intramedullary tumours in children. Acta Neurochir (Wien) 120: 59–65 52. Malis LI (1978) Intramedullary spinal cord tumors. Clin Neurosurg 25: 512– 539 53. Marcus RB Jr, Million RR (1990) The incidence of myelitis after irradiation of the cervical spinal cord. Int J Radiat Oncol Biol Phys 19: 3–8 54. McCormick PC, Stein BM (1990) Intramedullary tumors in adults. Neurosurg Clin N Am 1: 609–630 55. McCormick PC, Torres R, Post KD, Stein BM (1990) Intramedullary ependymoma of the spinal cord. J Neurosurg 72: 523–532 56. Minehan KJ, Shaw EG, Scheithauer BW, Davis DL, Onofrio BM (1995) Spinal cord astrocytoma: pathological and treatment considerations. J Neurosurg 83: 590–595

28

57. Mørk SJ, Løken AC (1977) Ependymoma. A follow-up study of 101 cases. Cancer 40: 907–915 58. Morota N, Deletis V, Constantini S, Kofler M, Cohen H, Epstein FJ (1997) The role of motor evoked potentials during surgery for intramedullary spinal cord tumors. Neurosurgery 41: 1327–1336 59. O’Sullivan C, Jenkin RD, Doherty MA, Hoffman HJ, Greenberg ML (1994) Spinal cord tumors in children: long-term results of combined surgical and radiation treatment. J Neurosurg 81: 507–512 60. Peschel RE, Kapp DS, Cardinale F, Manuelidis EE (1983) Ependymomas of the spinal cord. Int J Radiat Oncol Biol Phys 9: 1093–1096 61. Przybylski GJ, Albright AL, Martinez AJ (1997) Spinal cord astrocytomas: long-term results comparing treatments in children. Child’s Nerv Syst 13: 375–382 62. Raghavendra BN, Epstein FJ, McCleary L (1984) Intramedullary spinal cord tumors in children: localization by intraoperative sonography. AJNR Am J Neuroradiol 5: 395–397

63. Raimondi AJ, Gutierrez FA, Di Rocco C (1976) Laminotomy and total reconstruction of the posterior spinal arch for spinal canal surgery in childhood. J Neurosurg 45: 555–560 64. Rauhut F, Reinhardt V, Budach V, Wiedemayer H, Nau H-E (1989) Intramedullary pilocytic astrocytomas – a clinical and morphological study after combined surgical and photon or neutron therapy. Neurosurg Rev 12: 309–313 65. Reimer R, Onofrio BM (1985) Astrocytomas of the spinal cord in children and adolescents. J Neurosurg 63: 669–675 66. Rossitch E Jr, Zeidman SM, Burger PC, Curnes JT, Harsh C, Anscher M, Oakes WJ (1990) Clinical and pathological analysis of spinal cord astrocytomas in children. Neurosurgery 27: 193–196 67. Samii M, Klekamp J (1994) Surgical results of 100 intramedullary tumors in relation to accompanying syringomyelia. Neurosurgery 35: 865–873 68. Sandler HM, Papadopoulos SM, Thornton AF Jr, Ross DA (1992) Spinal cord astrocytomas: results of therapy. Neurosurgery 30: 490–493 69. Schwade JG, Wara WM, Sheline GE, Sorgen S, Wilson CB (1978) Management of primary spinal cord tumors. Int J Radiat Oncol Biol Phys 4: 389– 393

70. Stein BM (1983) Intramedullary spinal cord tumors. Clin Neurosurg 30: 717– 741 71. Steinbok P, Cochrane DD, Poskitt K (1992) Intramedullary spinal cord tumors in children. Neurosurg Clin North Am 3: 931–945 72. Whitaker SJ, Bessell EM, Ashley SE, Bloom HJG, Bell BA, Brada M (1991) Postoperative radiotherapy in the management of spinal cord ependymoma. J Neurosurg 74: 720–728 73. Xu Q-W, Bao W-M, Mao R-L, Yang G-Y (1996) Aggressive surgery for intramedullary tumor of cervical spinal cord. Surg Neurol 46: 322–328 74. Yasuoka S, Peterson HA, Laws ER Jr, MacCarty CS (1981) Pathogenesis and prophylaxis of postlaminectomy deformity of the spine after multiple level laminectomy: difference between children and adults. Neurosurgery 9: 145–152 75. Yasuoka S, Peterson HA, MacCarty CS (1982) Incidence of spinal column deformity after multilevel laminectomy in children and adults. J Neurosurg 57: 441–445