Current management in the treatment of intramedullary ependymomas in children

Current management of pediatric intramedullary ependymoma is extrapolated from adult series since large studies in children are unavailable. This has led us to share our experience with this rare tumor and compare it to the literature and to review and highlight important aspects of current management and point out inconsistencies. This is a retrospective analysis of patients with intramedullary ependymoma managed at our institution between 2004 and 2021. During the study period, 5 patients were treated for intramedullary ependymoma. Cases of myxopapillary ependymoma were excluded. The mean age of our cohort was 11.2 years. We identified 4 cases of grade II ependymoma and 1 case of grade III ependymoma. Gross tumor removal (GTR) was achieved in two patients (40%) of patients. One patient was treated with radiotherapy for recurrence and two patients received chemotherapy. There were no cases of recurrence among patients treated with GTR, but in all patients treated with STR. Eighty percent of patients either improved or stayed stable neurologically. During follow-up (mean 73 months), 2 patients died of disease. GTR and tumor grade remain the key prognostic factor of long-term tumor-free survival. Many questions prevail regarding outcomes, correct use of adjuvant therapy, and prognostic factors.


Introduction
Intramedullary spinal cord tumors represent a rare disease in adults and even more so in children [1][2][3]. They account for 2-4% of all central nervous system tumors [4] with specific age and pathology distribution. Astrocytomas and ependymomas (EP) represent the most common entities [4][5][6]. While EP are more common in adults [5,6], they are exceedingly rare in children [1,2,7]. Although a multitude of relatively large surgical series and reviews are available for the management of intramedullary EP in adults [1,5,[8][9][10], there are just a few case series in the literature reporting treatment results and long-term follow-up in children [11][12][13][14][15][16][17][18][19]. Additionally, studies often do not distinguish between extramedullary and intramedullary EP [7,14,20], which are distinct both from surgical as well as biological points of view. The purpose of this study is to report our experience with 5 children harboring intramedullary ependymoma treated in the past 17 years. Additionally, we want to highlight and review key aspects of current management and provide a brief case-illustrated literature review.

Patient population
Patients treated for intramedullary EP between 2004 and 2021 at our institution were identified in a prospectively collected database. Patients harboring myxopapillary EP were excluded. Patients presenting with recurrent tumors after undergoing treatment elsewhere were only included if additional surgery for recurrence was performed at our institution.

Clinical data and outcome assessment
Analyzed data comprised basic demography (gender, age) at presentation, duration, and type of symptoms. Neurological status was assessed preoperatively, postoperatively, and at follow-up visits using the modified McCormick scale [21,22] (Table 1). Follow-up visits were performed postoperatively at 6 weeks, 3 months, 6 months, 1 year, and annually thereafter. Permanent surgery-related neurological deficit was defined as progression on the modified McCormick scale by at least one grade on a 3-month follow-up. Surgical mortality was defined as any death within 30 days of surgery. At each follow-up visit, patients were also classified according to their disease status: complete remission with no residual tumor, stable disease, or progressive disease. Disease progression was defined as tumor recurrence after gross-total resection (GTR), progression of residual tumor by more than 25% [23] after less extensive resection, or metastatic disease distant from the primary surgery site.

Neuroimaging
Each patient underwent detailed magnetic resonance imaging (MRI) preoperatively, postoperatively, and during follow-up visits. Details noted on MRI included tumor location and extent, presence of tumor-associated cyst/syrinx, and spinal deformity. The extent of resection (EOR) was assessed by an experienced neuroradiologist (not involved in the study) on postoperative MRI as GTR (no residual tumor apparent), near-total resection (NTR, slight contrast uptake at the margins of resection cavity or more than 95% resection), subtotal resection (STR, 80-95% resection), and partial resection (PR, less than 80% resection) [24]

Surgical technique
Surgery was performed using standard microsurgical techniques under intraoperative monitoring. Either osteoplastic laminotomy or standard laminectomy extending one segment above and below the tumor margins was performed after the correct spine level was verified under fluoroscopy. Following durotomy, spinal dorsal columns were split at midline over the whole extent of the tumor, and the tumor was identified. The goal of resection was GTR when the tumor margins could be safely identified. Motor evoked potentials and D-wave were used to guide the resection extent. If a clear tumor margin could not be identified, resection of clearly pathological tissue was performed using an ultrasonic aspirator unless significant deterioration of motor evoked potentials occurred. A significant increase of transcranial threshold stimulation (20% of baseline value), a decrease of amplitude (< 50%), or a change in latency of motor evoked potentials warned the surgeon against pursuing further resection. Further increase in threshold stimulation (50% of baseline value) and decrease of amplitude (< 20%) were signals for immediate stop of resection. Similarly, more than 50% amplitude decrease of D-wave was deemed as an indication not to pursue further resection.

Literature search for similar case series
A comprehensive literature search was performed using the PubMed database with the keywords "intramedullary ependymoma" and "spinal ependymoma." Available case series of pediatric intramedullary EP or "adult series," where pediatric cases were separately reported, are listed in Table 4. Additionally, references to relevant articles were examined to identify relevant articles that might have been missed in Mild motor or sensory deficit, maintains functional independence 3 Moderate motor or sensory deficit, limitation, independent with external aids 4 Severe motor or sensory deficit, fully dependent on external help 5 Paraplegic or quadriplegic the initial search. Individual case reports/series reporting on less than 3 patients have been excluded.

Patient and preoperative tumor characteristics
During the study period, a cohort of 5 patients (all male) with a median age of 12 years (range, 2 to 17 years) was identified. The main presentation included motor impairment in 3 patients and pain in 2 patients. The median history length was 15 days (range, 1 day to 3 months). According to the modified McCormick scale, each of the 5 grades was represented by one patient preoperatively. Except for one patient (case 5), all patients had their primary surgery at our hospital. The most common tumor location was the thoracic spinal cord. In two patients, the thoracic segment alone was affected, and, in another case, both the cervical and thoracic segment were involved. One case presented with a tumor localized in the conus medullaris, and yet another patient presented with a holocord tumor (extension of the tumor from the cervicomedullary junction to the tip of the conus). The median tumor extent (defined as the spinal segments involved) was 7 segments (range, 3 to 20 segments). The presence of a peritumoral cyst was noted in 3/5 (60%) of cases. Additional clinical and imaging details are presented in Table 2.

Surgery and surgical complications
Osteoplastic laminotomy was performed in three patients as the surgical approach and laminectomy in two patients. GTR was achieved in two patients ( Fig. 1a-d); the remaining patients received STR due to a decline in intraoperative monitoring. There was no surgical mortality. Regarding neurological status (Table 3), in the immediate postoperative period, one patient declined by one grade on the modified McCormick scale, the remaining patients were either stable (2) or improved by one grade (2). Other surgical complications included liquorrhea necessitating revision in two patients. Postoperative spinal deformity after laminectomy was observed in two patients after a follow-up of 6 and 12 months.

Neurological outcome
Long-term neurological outcome was available in all patients (Table 3). Improvement was observed in three patients (modified McCormick grade at last follow-up of 1, 1, and 2), one patient remained stable (preoperative grade 5), and one patient declined (preoperative grade 4) due to disease progression and dissemination after early postoperative improvement.

Histology
Tumor histology revealed an EP in all patients. Aside from one patient (grade III), all others were initially classified as having a grade II tumor. One recurrent grade II EP was reclassified as grade III after recurrence and additional resection due to increased proliferation activity (initially 6% increased to 10%) and nuclear atypia. Proliferation activity using the Ki67 marker was assessed in two other tumors and revealed 3% (case 1) and 15% (case 5).

Survival
None of the two patients, who underwent GTR, experienced tumor recurrence after a follow-up of 12 and 179 months. Contrary, all patients after STR experienced tumor recurrence after a mean progression-free survival (PFS) of 11.3 months (see Fig. 2). All recurrent tumors underwent additional surgery (all STR). Adjuvant therapy was not initiated in one patient (case 2) due to poor McCormick grade, and he died 15 months after primary surgery. Case 4 (anaplastic EP) received only chemotherapy. He was treated according to the protocol in the CCLG Infant ependymoma study for intracranial EP [25], and radiotherapy was withheld in this patient considering his age and excellent response to repeated resection and chemotherapy. He remains with stable disease after 70 months. The last patient (case 5) experienced disease dissemination including multiple intracranial and spinal metastases. Despite multiple resections, chemotherapy, and comprehensive craniospinal radiotherapy, he died 88 months after primary surgery. At the last follow-up (median 70 months; range 12-179 months), two patients were alive with no evidence of disease, one was alive with stable disease (interestingly grade III histology), and two patients died due to disease progression.

Discussion
The goal of this case series is to share our experience and results in the treatment of this rather rare entity of spinal cord tumors and compare it to the available literature to provide an up-to-date case illustrated review.

Patient presentation
The natural history of spinal cord tumors is slowly progressive, and symptoms usually develop inconspicuously. Delay in diagnosis for many months has therefore not been uncommon [13,26]; however, widespread availability of  [27]. Our data suggest a similar trend with only one patient exceeding more than 3 months of history length. Typical presentation includes motor deficit and pain, as nicely illustrated by the presented patients. Acute and dramatic presentation (case 2) is rather uncommon for intramedullary tumors and is reported only sporadically [28][29][30].

Importance of complete tumor removal and factors affecting it
Comprehensive treatment of spinal EP starts with surgery. The goal of surgery should always be GTR, however, not at the cost of neurological deterioration. Thus, intraoperative monitoring aiming to maximize safety has become the standard of intramedullary surgery [9,31]. An overview of case-series reporting outcomes of patients with intramedullary EP separately is given in Table 4. EP is generally more resectable than their more common counterparts astrocytomas, which grow more diffusely and where a clear tumor to spinal cord margin is often not identified [5,32]. Nevertheless, a rather wide range of GTR has been reported in the literature, ranging from 25 to 100% [11][12][13][15][16][17][18][19]. For adults, this number has been more consistently reported between 70 and 100% [1,5,[8][9][10]. Our rate of GTR (40%) falls into the lower part of the spectrum, while small numbers certainly influence the results. STR was performed in á priori "complicated" patients: a paraplegic patient with a holocord tumor, a high-grade EP, and a recurrent tumor. The decline of intraoperative monitoring and the decision not to pursue further resection is unsurprising in these patients. Certain factors such as reoperation [13], tumor radiation [13,26,32], absence of tumor syrinx [35], tumor grade [13,[36][37][38], tumor extent [1,36], and lower spinal localization [13,17,38,39] have been proposed by some authors as independent factors that may negatively influence the possibility of achieving GTR and thus being prognostic factors for outcome. For example, Oh and colleagues [39] have proposed that tumor localization may be an independent risk factor for PFS and overall survival (OS). His group has analyzed that a GTR rate of 75.9% was achieved for upper spinal cord EP, whereas only 48.3% for the lower spinal cord EP, which was not explained by other factors such as tumor grade.

Surgical outcome: neurological outcome and postoperative complications
Numerous studies have consistently reported that a good preoperative neurological status is the single best predictor for good long-term postoperative neurological status [1,2,4,8,9,26,32,37]. Immediate transient neurological decline after surgery is rather common [32,36] and related to dorsal myelotomy and surgical manipulation necessary for tumor resection [9]. It is usually completely reversible for most patients in the first 6 months following surgery [5,32,36]; however, patients and their families should be made aware Fig. 1 a, [26,36,40]. Spinal deformity is the most reported postoperative complication next to liquorrhea. Both have developed in two of our patients. Literature shows that particularly young children are more susceptible to the development of spinal deformities than adults [26,32,41]. It is estimated that around 5-30% of patients will develop spinal deformity, which needs orthopedic intervention [26,32,40,41]. Consequently, in recent years, osteoplastic laminotomy is the preferred goal over complete laminectomy to lower the risk of spinal deformity [13,40,41]. Other factors, such as preoperative scoliosis, thoracolumbar localization of the tumor, a younger age, and the number of surgeries, are factors, which have been correlated with an increased risk of spinal deformity [41].

Long-term outcome: recurrence rate and overall survival
Most studies (ours included) have relatively short followups not reaching adulthood. Lundar et al. [14] recently reported the first ultra-long follow-up of children with spinal ependymomas over a span of 8 decades. In their cohort of 33 children (16 grade II/III, 17 myxopapillary), they found relatively good survival rates with 85% and 65% of patients being alive after 10 and 20 years after diagnosis. However, they also showed relatively high rates of recurrence rate, particularly among myxopapillary EP. They report cases that presented with recurrence as long as 20 years after the initial surgery supporting the notion that lifelong follow-up is needed in such patients, even after GTR. Nonetheless, care should be taken to not look at different tumor grades together, as myxopapillary tumors seem to exhibit different biological characteristics and almost exclusively grow extramedullary [14,18,42].
GTR usually prevents tumor recurrence and is the only factor consistently associated in the literature with increased OS and/or PFS [2,4,13,17,38], in contrast to patients with STR [4,13,17,38]. Our two cases of GTR remained tumor-free in the last-follow up. However, one of them was followed only for one year (the other one for 7.5 years). Following STR, recurrence was observed in our series rather soon; two patients experienced recurrence within the first 6 months and the last patient in 24 months. The fact that two of those three cases were anaplastic EP should be taken into consideration, as they show a more aggressive clinical course [43].Yet another factor associated with OS is the tumor grade. Safaee et al. [38] showed that anaplastic EP were associated with a decreased OS in comparison to lower grade EP, but interestingly not with a decreased PFS. The decreased possibility of achieving GTR in those tumors might be a potential confounding factor, but that hypothesis did not reach statistical significance in their analysis [13,38]. The question remains, whether age and gender are prognostic factors for OS and/or PFS. While some authors found a positive association of female gender with outcome [44], others did not [11,38]. A similar situation exists with the effect of age [11,38,44].

The current role of radio-and chemotherapy
Current clinical practice is imperfect and orientates itself on studies performed on intracranial EP and/or adult EP patients [7,13]. Good long-term results have been reported repeatedly for children in which GTR has been achieved Fig. 2 Overview of treatment and outcomes without radiotherapy for grade II EP and is therefore currently not recommended [7,13,17,40]. STR with RT has been shown in some studies in adults to be superior to STR alone but remains a topic of controversy [1,10,45,46]. Any potential benefit must be juxtaposed to potential damage conferred on the developing osseous and nervous systems in this age group [2,41,47]. In practice, radiotherapy is usually reserved for cases of anaplastic EP or inoperable tumors, even though there is a lack of data proving efficacy for both adults and children due to the small number of cases [1,7,10]. Despite the good response encountered in one of our patients to chemotherapy, the role of chemotherapy itself in the treatment of intramedullary EP is unclear. Chemotherapy is usually reserved for cases of recurring and/or anaplastic EP as a way of improving the prognosis by all means possible [48][49][50]. However, the evidence is poor, due to the small number of patients and the heterogenicity of applied regimens reported in the literature [7,10]. Larger patient cohorts will be ultimately needed to evaluate the potential benefit of chemotherapy regimens designed for intracranial EP for spinal EP, such as seemingly encountered in one of our patients [25].

Limitations of WHO grading and molecular markers
There exists lasting controversy regarding the clinical utility and reproducibility of grading EP according to the WHO [51]. As more research concerning pathogenesis and molecular markers is performed, newer classification systems, with the aim to increase clinical utility and diagnostic accuracy, have been proposed [52,53]. The recent 5th edition of the WHO classification moves away from a predominantly histopathologic classification to integrating these novel findings [54]. For EP, it currently recognizes 10 different types that are distributed across 3 major anatomical compartments (supratentorial, posterior fossa, and spinal cord), in contrast to only 4 entities in 2016 [33,54]. For the "spinal compartment," the subtypes subependymoma (SPE) and myxopapillary EP (MPE), as histopathologically diagnosed entities, have been retained and two "new" classes have been added [54]. A novel NMYC amplification containing spinal EP (SP-MYCN) and spinal EP (SP-EPN) has been introduced, while the term classic and anaplastic EP has been dropped [54,55]. Notably, the former seems to grow distinctly extramedullary, while the latter intramedullary [34,56]. As classification systems are evolving, it will be interesting to see, whether we can identify more subtypes, particularly within the SP-EPN type.

Study limitations
The main limiting factor, along with other publications [11][12][13][15][16][17][18][19] reporting on childhood spinal cord EP, is the small number of patients enrolled in this study alongside the usual shortcomings of a retrospective study design. The incidence of one EP patient every 3-4 years falls in line with other single-institution series [15,17,26]. Therefore, our data remains descriptive, as statistical analysis of such a small cohort is not feasible. Although our treatment and follow-up protocol remained consistent during the study period, important clinical data could have been missed.

Conclusion
Intramedullary ependymomas are extremely rare tumors in childhood. In our series, we encountered no cases of recurrence among patients treated with GTR, but in all patients treated with STR. This highlights the importance of primary surgical management, given the current absence of effective adjuvant therapy.