Patterns of failure and optimal radiotherapy target volumes in primary intradural extramedullary Ewing sarcoma

To the Editor,

Soft tissue Ewing sarcoma (ES) is a member of the Ewing sarcoma family of tumors that includes osseous Ewing sarcoma and peripheral primitive neuroectodermal tumors (pPNET). When involving the spine, these tumors arise more often in the paravertebral or extradural regions with limited reports of tumors originating from the intradural extramedullary space. Other reported tumors found within the dura, but not involving the cord include central primitive neuroectodermal tumors (cPNET), and more commonly schwannoma, meningioma, neurofibroma, paraganglioma, or ependymoma tumors [1].

Pathologically the distinction between cPNET and pPNET (or ES) can be difficult because they appear similar morphologically as small round blue cell tumors. ES is distinct from cPNET by immunohistochemistry and cytogenetics. ES shares common findings of immunoreactivity for the cell surface glycoprotein CD99 (MIC2 gene product), which is not expressed in cPNET, but found in other primary central nervous system (CNS) tumors including ependymomas, meningeal hemangiopericytomas and leukemic infiltrates. ES tumors also have common translocations involving chromosome 22, most notably t(11;22)(q24;q12). This mutation has not been identified in cPNET [2].

ES is treated with surgery and/or focal radiotherapy in addition to systemic chemotherapy. Previous reports of primary intradural extramedullary ES also suggest standard Ewing sarcoma regimens including focal radiotherapy may be effective in this disease location. In contrast, since as early as 1952 [3], the treatment of cPNET of the spine has included craniospinal radiotherapy due to the 20–32% risk of patients who have occult CNS dissemination at diagnosis [4]. This pattern of failure has not previously been associated with ES, which is traditionally locally invasive and disseminates hematogenously to the lungs and bone marrow.

We previously reported two patients with intradural ES treated with surgery, focal radiation and systemic chemotherapy [5]. In this study we present an update on our previously reported patients, two additional patients from our institution, and a review of the previously reported cases from the literature. The purpose of this report is to describe patterns of failure to determine the optimal radiotherapy volume for primary intradural ES.

Materials and methods

Pediatric and adult patients from our institution and in the literature with pathologically confirmed primary extraosseous Ewing sarcoma or pPNET of the intradural extramedullary spine were included. Patients with osseous, extradural, intramedullary or metastatic disease at diagnosis were excluded. Prior to 1992 the ES translocation analysis and magnetic resonance imaging (MRI) were not standardly performed, so these patients and reports were also excluded.

Patients at our institution were identified using the Mayo Clinic tumor registry. Patient reports from the literature were identified using the PubMed, Medline and Scopus databases with combinations of the keywords Ewing sarcoma, pPNET, intradural, extramedullary and cauda equina. Data was collected from both groups about gender, age at diagnosis, spinal level of tumor, extent of surgery, chemotherapy, radiation volume and dose, recurrence and survival. Authors of published reports with incomplete data were contacted using the listed corresponding addresses. Cases without radiotherapy target volume or dose, follow-up information or did not receive standard Ewing sarcoma systemic chemotherapy drugs were excluded from the analysis. Standard chemotherapy backbones included combinations of vincristine, doxorubicin, cyclophosphamide, ifosfamide, etoposide, and/or dactinomycin. The use of high-dose chemotherapy with autologous stem cell rescue was also included.

Data was analyzed using JMP statistical software. Descriptive statistics included categorical data frequencies and percentages with continuous data expressed as a percent and median. Local failure was defined as recurrence at the primary tumor location. Distant failure was defined as recurrence of disease within the neuraxis at a site other than the primary. Survival was calculated from the time of diagnosis to death or the last patient follow-up. Event-free survival (EFS) was calculated from time of diagnosis to recurrence or death. To estimate date of diagnosis for reports that published events from the end of therapy, 10 months were added to the time period to account for the standard length of treatment. Small sample size and limited follow-up data restricted our use of Kaplan-Meier analysis, therefore only crude data is reported.

Results

Review of the Mayo Clinic tumor registry from 1992 to 2014 identified 25 patients with ES of the spine. Four patients had primary intradural extramedullary disease and met inclusion criteria, of which two we previously reported in 2011 [5]. Here, we will present an update on one of the previously reported patients, review the second and present two additional patients.

Patient 1

We previously reported a 50-year-old man who presented with radiating asymmetric hip and low back pain with weakness and numbness of the medial thighs and groin [5]. MRI identified a non-metastatic primary T10-L1 intradural extramedullary mass that was grossly resected. Pathology was positive for MIC-2 (CD99), CD56, pancytokeratin (AE1/AE3), OSCAR, CAM 5.2 and synaptophysin. Reverse transcription polymerase chain reaction (RT-PCR) was positive for the EWSR1-FLI1 fusion.

In our 2011 report, our patient was stable 26 months after diagnosis with treatment using VDC/IE for a total of six cycles with focal radiotherapy to the thoracolumbar spine to a dose of 50.4 Gy [5]. On regular follow-up, 48 months from diagnosis, an MRI showed a new lesion at the left cerebellopontine angle, near the foramen of Luschka (Figure 1). Biopsy confirmed a CD99 and synaptophysin positive ES consistent with his previous diagnosis. He underwent a gross total resection, but on first follow-up imaging was found to have re-growth of his primary disease. He went on to receive 57 Gy to the area. Two months following completion of radiotherapy a follow-up MRI showed new brain lesions. He was treated with additional radiation of 39.6 Gy to modified whole brain fields prior to gamma knife radiosurgical boost to areas of gross disease, followed by temozolomide and irinotecan. Two months later additional intradural spinal lesions were noted, and he succumbed to his disease 60 months from diagnosis.

Figure 1. Case 1 sagittal T1-weighted MRI at diagnosis showing a mass from T11-L1 (left – previously reported image [6]). Coronal T1-weighted MRI at recurrence showing a new brain mass in the region of the left foramen of Luschka (right).

Patient 2

We previously reported a 60-year-old gentleman who presented with progressive radiating back pain [5]. MRI showed an enhancing mass at L2-3, measuring 3.3 cm in length. A partial resection of the intradural extramedullary mass was performed to spare lumbar nerve function. Immunohistochemistry revealed strong expression of MIC2 with focal immunoreactivity for CAM5.2, synaptophysin and vimentin. An RT-PCR confirmed the FL1-EWS fusion product confirming ES. Imaging for metastasis was negative. He was treated with two cycles of ifosfamide and etoposide alternating with ifosfamide and doxorubicin prior to focal radiation therapy to 50.4 Gy from L1 to L4, followed by an additional cycle of chemotherapy. His four-month follow-up imaging demonstrated a new lesion outside of the initial radiation field extending from L5 to S2. Biopsy confirmed recurrence, and he was started on daily temozolomide while undergoing sacral radiation to 59.4 Gy.

Five months after completing radiation and temozolomide a follow-up scan showed diffuse leptomeningeal metastasis with multiple new posterior fossa, cervical and thoracic spine lesions. He was treated with modified craniospinal axis radiation including the brain and spinal axis to L1. Following radiation he completed three cycles of vincristine, doxorubicin and cyclophosphamide. Follow-up imaging showed interval improvement with residual disease at the T7 level. Eight months later a repeat MRI showed evidence of progressive disease at T7 and from L2 to L5, which were within previous radiation fields. He received palliative radiation and oral etoposide before dying from his disease 47 months from diagnosis.

Patient 3

A 25-year-old young man presented following a subtotal surgical resection for a cervical tumor at an outside institution. He reported initially developing numbness and weakness of the right hand 3–4 weeks prior. Following surgery his symptoms progressed bilaterally and he underwent drainage of a large fluid collection at the surgical site. With minimal symptom improvement he returned to the US and presented to our emergency department. An MRI demonstrated an enlarging intradural mass from C4 to C7 compared to the outside postoperative images, with increasing spinal cord compression. He underwent tumor debulking, instrumentation, and arthrodesis from C3 to T1.

Pathology showed a small blue cell sarcoma with immunostaining positive for CD99, and negative for actin, desmin, MYOD1, chromogranin, synaptophysin, WT-1 and Cam 5.2. FISH studies showed a rearrangement of the ESWR1 locus, consistent with ES. Bone marrow, bone scan, positron emission tomography (PET), brain and spine MRI, and chest computed tomography (CT) negative for metastatic disease.

During his recovery period, he developed progressive neck pain and urinary retention. A repeat MRI revealed increased cord edema and tumor progression. He was started on dexamethasone and given emergent radiation using parallel opposed fields to the tumor bed for 4 fractions before starting his scheduled craniospinal radiation course. This was followed by an additional boost to the tumor area. The total treatment dose was 54 Gy. During radiation he initiated chemotherapy with ifosfamide and etoposide, and then alternated VDC/IE for a total of 12 cycles. There is no evidence of disease 20 months from diagnosis.

Patient 4

A 34-year-old gentleman presented with cauda equina syndrome to an outside facility. Imaging showed an intradural mass at the L4-L5 level with three nodular lesions in the subarachnoid space at S1-S2 and S4-S5 nerve roots. An excisional biopsy of the lumbar mass was performed, leaving residual disease at the sacral level. Pathology showed Ewing sarcoma with immunohistochemical stains positive for CD99, EMA, and synaptophysin. Molecular studies confirmed the EWSIL-FL1 fusion transcript. Metastatic evaluation with PET scan, spine and brain MRI were negative. He was treated with five cycles of VDC/IE and autologous stem cell support. Prior to starting radiation imaging showed complete response. He went on to receive craniospinal radiation to 30 Gy with a boost to 59.4 Gy. The patient completed therapy and has remained stable for three months.

Literature review

Review of the literature identified 37 unique cases of primary intradural extramedullary ES, which combined with the cases from our institution, provided 41 total. Authors of reports with incomplete inclusion data were contacted by email for additional information. Nine of 18 (50%) authors responded. From the combined group with updated information from responding authors, 21 additional cases were then excluded because radiation was not used, radiation volume and/or dose were not available, chemotherapy was not used, non-ES chemotherapy was used, or follow-up information was not available (Supplementary Figure 1, available online at http://www.informahealthcare.com).

Combined results

Twenty cases combined met our study criteria (summarized in Table 1 and Supplementary Table 1, available online at http://www.informahealthcare.com) [5–20]. Twelve patients were male and eight were female with ages ranging from 8 to 70 years old (median age of 31). Median follow-up was 17 months (range of 5–144 months). Seventeen patients (85%) received only focal radiation to the primary site with a median total dose of 50 Gy (range from 30 to 56 Gy). Three patients (15%) were treated with craniospinal axis radiation to a total dose of 30–36 Gy, with a total dose of 45–59.4 Gy to the primary site (Supplementary Table I).

Table 1. Patient characteristics (n = 20).

Variables	No. patients (%)
Gender
Male	12 (60)
Female	8 (40)
Age (years)
0–18	4 (20)
>18	16 (80)
Tumor location
Cervical	3 (15)
Thoracic	0
Thoracolumbar	3 (15)
Lumbar	6 (30)
Lumbosacral	7 (35)
Sacral	0
Thoracolumbosacral	1 (5)
Pathology confirmed
CD99	16 (80)
t(11;22)(24;12)	11 (55)
CD99 or t(11;22)(24;12)	20 (100)
Radiotherapy volume
Focal to primary site	17 (85)
Craniospinal axis	3 (15)

Eight of the 20 (40%) patients developed recurrent disease within the CNS; 47.1% of the patients treated with focal radiation therapy, and no patients treated with craniospinal radiation (Table 2). No patients developed recurrence outside of the CNS. Median time to recurrence was 18 months (range of 8–48 months), with a crude two-year EFS of 57.6%. Of the patients who recurred, one experienced primary site failure only, five had distant craniospinal axis failure and two had both. Five patients died of their disease (62.5% of those who recurred, and 25% overall) at a median of 18 months (range of 12–60 months) from diagnosis. Four of seven patients who experienced distant craniospinal axis recurrence died at a median of 34 months (range of 16–60 months) from diagnosis. One patient, reported by Mobley et al. died following a local recurrence that was not responsive to additional chemotherapy. Metastatic disease was not found on bone scan or chest imaging [15].

Table 2. Combined patient outcomes.

	Radiotherapy volume
	Primary site (%)	CSI	Combined (%)
	n = 17	n = 3	n = 20
Recurrence	8 (47.1)	0	8 (40)
Local only	1 (5.9)	0	1 (5)
Distant CNS only	5 (29.4)	0	5 (25)
Both	2 (11.8)	0	2 (10)
Died of disease	5 (29.4)	0	5 (25)

CSI: craniospinal irradiation.

Discussion

In this report we review 20 cases of intradural extramedullary ES treated with surgery, chemotherapy and radiation with findings of local and distant CNS patterns of failure. The majority of patients we studied were treated with focal radiotherapy to the primary site of disease (85%). Of the patients treated with focal therapy, 47.1% recurred. Three patients were treated with craniospinal radiation, and though the numbers are small, there were no recurrences or deaths in this group with a median follow-up of 24 months (range of 10–31 months). Although longer follow-up is needed, optimal radiotherapy volume appears to favor entire craniospinal axis treatment, similar to the standard treatment for cPNET.

This data conflicts with previous reports recommending the use of focal radiotherapy for the treatment of intradural extramedullary ES. The discrepancy from our findings may be due to the short follow-up at the time of publication for the majority of the published case reports, including our own prior report [5]. The median time to recurrence for the patients in our study was 18 months, but with a median follow-up time of only 17 months, it is possible that our report underestimates the risk of recurrence thus the risk of craniospinal dissemination may even be higher.

There are possible explanations for why this location of ES disseminates differently than when arising outside of the spine. Patients often present with neurologic symptoms secondary to cord or spinal nerve compression. Therefore, the surgical resection takes place emergently instead of the standard approach of neoadjuvant chemotherapy prior to local control with surgery and/or radiation. Additionally, because of the tumor location patients are unable to undergo an oncologic or en bloc resection that attempts to maintain the tumor integrity. Instead, patients have intralesional resections, disrupting the tumor and surrounding tissues, and increasing the probability of microscopic tumor dissemination via the cerebral spinal fluid [21]. Our data suggests that although pathologically ES and cPNET are separate entities, the patterns of failure of primary intradural extramedullary ES and cPNET appear to be similar.

Overall radiation was well tolerated in our four patients. Acute side effects included fatigue, nausea and mild to moderate skin erythema with one case of moist desquamation which improved with topical treatment. Myelosuppression with the development of mucositis was noted with concurrent radiation and chemotherapy. One patient developed dysphagia during craniospinal treatment requiring a liquid diet and later esophageal dilation for strictures. Cardiac changes were noted in one patient with a decreased ejection fraction which resolved with discontinuation of anthracyclines and beta blocker therapy. Despite the short- and long-term side effects, all four patients completed therapy. In the future, use of proton therapy should aid in reducing these side effects.

Limitations of this report include the small sample size and a median follow-up of less than two years. Variation in the reporting of time to event data in the literature led us to estimate the time of treatment as 10 months, which may underestimate the risk of recurrence failure in this population. Also, by only reporting on three patients treated with CSI with a follow-up of 10–31 months, it remains to be seen whether this approach will improve outcomes.

Intradural extramedullary ES appears to have a poor prognosis with only 57.6% crude two-year EFS in patients treated with multimodality therapy. Distant craniospinal axis failure appears to be the predominant pattern of failure (41.2% crude) in patients treated with focal radiotherapy. Based on this review, we recommend considering craniospinal radiation for all patients with intradural extramedullary Ewing sarcoma.

Acknowledgements

This publication was made possible by the CTSA Grant UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS), a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.

Disclosure statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this article.