Surgery and proton radiation therapy for pediatric base of skull chordomas: Long-term clinical outcomes for 204 patients

Abstract

Background

Data on clinical outcomes for base of skull (BOS) chordomas in the pediatric population is limited. We report patient outcomes after surgery and proton radiotherapy (PRT).

Methods

Pediatric patients with BOS chordomas were treated with PRT or combined proton/photon approach (proton-based; for most, 80% proton/20% photon) at the Massachusetts General Hospital from 1981 to 2021. Endpoints of interest were overall survival (OS), disease-specific survival, progression-free survival (PFS), freedom from local recurrence (LC), and freedom from distant failure (DC).

Results

Of 204 patients, median age at diagnosis was 11.1 years (range, 1–21). Chordoma location included 59% upper and/or middle clivus, 36% lower clivus, 4% craniocervical junction, and 1% nasal cavity. Fifteen (7%) received pre-RT chemotherapy. Forty-seven (23%) received PRT, and 157 (77%) received comboRT. Median total dose was 76.7 Gy (RBE) (range, 59.3–83.3). At a median follow-up of 10 years (interquartile range, 5–16 years), 56 recurred. Median OS and PFS were 26 and 25 years, with 5-, 10-, and 20-year OS and PFS rates of 84% and 74%, 78% and 69%, and 64% and 64%, respectively. Multivariable actuarial analyses showed poorly differentiated subtype, radiographical progression prior to RT, larger treatment volume, and lower clivus location to be prognostic factors for worse OS, PFS, and LC. RT was well tolerated at a median follow-up of 9 years (interquartile range, 4–16 years). Side effects included 166 patients (80%) with mild/moderate acute toxicities, 24 (12%) patients with late toxicities, and 4 (2%) who developed secondary radiation-related malignancies.

Conclusion

This is the largest cohort of BOS chordomas in the literature, pediatric and/or adult. High-dose PRT following surgical resection is effective with low rates of late toxicity.

Key Points.

1. Postoperative proton-based RT demonstrates survival outcomes in pediatric base of skull chordomas with PFS of 64% at 20 years.

2. Poorly differentiated type, pre-RT progression, lower clival location, and larger size are associated with worse survival outcomes.

3. Long-term follow-up is necessary for tumor surveillance and monitoring for late side effects.

Importance of the Study.

This study is the largest series of pediatric chordomas of the base of the skull (BOS) treated with surgery and proton radiation therapy (PRT). It exceeds the largest published studies of this disease by nearly an order of magnitude with long follow-up significantly greater than the median time to recurrence. In the absence of a prospective trial, this data suggests that high-dose PRT following maximal safe surgical resection is effective in the management of pediatric BOS chordoma with 10- and 20-year OS and PFS rates of 78% and 69%, and 64% and 64%, respectively, while maintaining an acceptable side effect profile. Lifelong clinical and imaging follow-up in centers with chordoma expertise and multidisciplinary care team is recommended to evaluate for disease control and severe treatment toxicities among survivors.

Chordomas are a rare form of bone tumor, thought to arise from embryonic notochordal remnants and accounting for only 1%–4% of bone cancers.¹ They occur most frequently in the sacrum, followed by the skull base and the mobile spine, and are typically present between 50 and 60 years of age.^1,2 Chordomas are rare in patients younger than 40, and fewer than 5% of cases are present in children and adolescents.^1,3 In this age group, chordomas most commonly originate from the skull base, followed by the mobile spine and the sacrum. Chordomas are slow growing, but locally invasive and destructive; historically reported chordoma survival rates are low, with median survival around 6 years, and 5-, 10-, and 20-year survival rates rapidly declining from 67.6% to 39.9%, and 13.1%, respectively.¹ Inferior outcomes have been reported in younger patients, with children ≤ 5 years of age often experiencing particularly poor outcomes.^4,5

Complete resection is associated with improved outcomes.^6,7 However, due to the proximity to critical anatomical structure, en bloc or even complete surgical excision of the often tumor remains elusive. Patients often receive postoperative radiotherapy, though traditional X-ray radiotherapy offers only palliative benefits and does not improve median survival, almost certainly due to the lower doses feasible (50–60 Gy) with X-ray treatment as seen in older series.⁸ Overall recurrence rates with X-ray radiotherapy were historically reported to be 80%–100%.^8–10 A few adult photon series have recently published encouraging early outcomes in BOS chordomas treated with modern photon techniques including intensity-modulated RT (IMRT) [5-year OS and local control (LC) rates of 85.6% and 65.3%, respectively] and stereotactic radiosurgery (SRS) [5-year OS and PFS rates of 83% and 54.7%, respectively] that allow for dose escalation beyond 60 Gy.^11,12

With regards to pathology, historically, pediatric chordomas had been subdivided into 3 categories: Conventional, chondroid, and poorly differentiated; the fourth chordoma subtype (dedifferentiated chordoma) has only been reported in the adult setting.¹³ The current (fifth edition) WHO classification considers chondroid chordoma as a variant of conventional chordoma.¹⁴ The majority of chordomas are conventional. Nuclear immunostaining for brachyury is considered the most sensitive and specific tool to establish the pathological diagnosis of chordoma. A rare subtype, called poorly differentiated chordoma is characterized by the loss of expression of (SMARCB1)/(Integrase integrator 1) (INI-1) and is seen almost solely in children.

To date, there are no large prospective studies or randomized trials for chordomas and radiation therapy (RT), and therefore, there is limited guidance for tumor management. Given that 30%–40% of chordomas originate at the base of the skull, numerous critical structures such as cranial nerves, brainstem, optic chiasm, and spinal cord make high-dose delivery without toxicity difficult with conventional 3D RT.^7,15,16 However, charged particle RT has marked a positive shift in treatment outcomes for chordomas. Proton-beam therapy allows for precise delivery of high-dose radiation to specific targets, while avoiding delivery of radiation to the surrounding normal tissues. Proton radiation therapy (PRT), or mixed proton and photon therapies, results in improved LC rates, ranging from 67.4%–87.5% at 3 years, 46%–73% at 5 years, and 54% at 10 years.^17–20 Of note, one study demonstrated a 5-year LC and overall survival (OS) of 60% for 10 children and adolescents with chordoma, when treated with proton or mixed radiotherapy after surgical resection.²¹ Late treatment-related side effects can require exceptionally careful consideration when treating a child with high-dose RT, making proton therapy a widely accepted modality to reduce morbidity in the pediatric cancer population.^17,22 In this study, we report on the clinical outcomes of a large cohort of children with base of skull (BOS) chordoma treated with surgery and proton-based RT. To the best of our knowledge, this study is the largest series of pediatric BOS chordoma patients and the largest overall series for BOS chordoma reported in the literature, and the first to report long-term survival outcomes and treatment toxicities.

Methods

Patient Selection

Between 1981 and 2021, 204 pediatric patients (age ≤ 21 years) were treated with proton-based radiation therapy for BOS chordoma at the Massachusetts General Hospital. Individual clinical information consisted of age, gender, ethnicity of the patient, location and extension of the tumor, type and duration of presenting symptoms and signs, and detailed longitudinal follow-up data. Surgical information documented date and type of surgical intervention, while RT data documented the type and dose of PRT given and any systemic agent administered. Patients were excluded if they had received prior radiation to the head or neck area that would compromise the ability to deliver the prescribed treatment. All patients had histopathological confirmation of BOS chordoma by a pathologist at our institution. This retrospective study protocol was approved by a standing institutional review board.

Pre-RT Treatment

All patients underwent surgical intervention (resection or biopsy) before delivery of RT. Surgical intervention for this study was characterized as gross total resection (GTR), defined as complete removal of all visible tumor tissue, subtotal resection (STR), defined as debulking of the tumor or near total resection of the tumor with any amount of unresected tumor tissue left behind, or biopsy, defined as minimal removal of tumor mass. The extent of the tumor removal was estimated by the operating surgeon and a comparison of preoperative and postoperative pre-RT imaging scans (MRI Brain, CT Head, or both). Following surgery, patients underwent adjuvant RT with proton-beam radiation targeted at the residual tumor and adjacent areas that were at high risk for microscopic disease. Based on the pathological subtype, pre-RT chemotherapy was administered in select cases.

Radiation Treatment

All patients underwent 3D planning with a contrast-enhanced CT scan (unless contrast was contraindicated). A CT myelogram, which is a useful tool in RT-planning to better define the spinal canal in treatment position, was performed in cases where tumor extended into the upper cervical spine, where gross tumor remained in close proximity to the brainstem and/or spinal cord, and in the presence of metallic hardware. All MRIs (preoperative and postoperative) were also used to guide treatment. Depending on the age of the patient, sedation was performed for the planning and daily treatments. The preoperative tumor volume as well as the tumor bed and any residual gross disease accounting for anatomical shifts and changes were contoured as gross tumor volume (GTV), whereas the clinical target volume included the target volume and an additional margin for microscopic extension of disease of 2 mm accounting for anatomical constraints. Target dose to the GTV was typically 76 Gy(RBE) and 54 Gy(RBE) to the clinical target volume. The dose to target volumes was delivered in such a manner as to attempt to respect constraints on adjacent normal structures. These constraints are as follows: The maximum dose to the surface of the spinal cord or brainstem should not exceed 67 Gy (RBE), and the maximum dose to the center of the spinal cord or brainstem should not exceed 55 Gy (RBE). The maximum dose to the optic chiasm and optic nerves should not exceed 62 Gy (RBE) (Supplemental Table). Dose was prescribed in Gray relative biological equivalents (Gy [RBE]) using the RBE value of 1.1.

For patients treated before 2001, when the Harvard Cyclotron was the treatment facility, proton treatments were offered 4 day/week using the 160-MeV proton-beam line. Megavoltage photons were used 1 day/week because clinical treatments were available only 4 days per week. When we transitioned to the Francis H. Burr Proton Center of the Massachusetts General Hospital, we continued this historical trend of delivering approximately 20% of the total dose with photons using a 3D conformal technique or intensity-modulated RT (IMRT). This was in part to allow for improved conformality and skin sparing. Once proton-beam scanning was available (2011), which offers a lower skin dose and improved conformality, a transition was made to use protons only. The prescribed total dose was administered in daily proton fractions of 1.8 or 2.0 Gy (RBE), photon fractions of 1.8 or 2.0 Gy each, 5 fractions per week, with the exception of 11 patients that received twice daily treatments.

Outcomes

The outcomes of interest for this study were overall survival (OS), disease-specific survival (DSS), progression-free survival (PFS), freedom from local recurrence (LC), freedom from distant failure (DC), and toxicity. The OS and DSS were measured from the initial diagnosis to the date of death from any cause and the date of death from disease, respectively. The PFS, LC, and DC were measured from the completion of RT to the date of the first local, distant, synchronous, or iatrogenic failure. For analysis of tumor volume as a prognostic factor tumor, tumor volume was defined as the GTV, which represented the tumor bed from the initial tumor volume and any gross disease present at the time of RT. Acute and chronic toxicities were also documented. Neuro-ophthalmology, endocrinology, and audiology examinations were obtained if performed.

Statistical Analysis

Actuarial analysis of OS, DSS, and PFS was performed using univariate and multivariable Cox proportional hazards regression. The analysis of LC and DC was performed using univariate and multivariable competing risks regression with death event as a competing risk. The selection of variables for multivariable analyses was based on prior published and otherwise communicated experience augmented by exploratory univariate and multivariable analyses of our data, and by the number of events for each outcome of interest. Wilcoxon rank-sum test was used to compare distributions of continuous variables between groups. The statistical analyses were done using Stata programming software (StataCorp. 2021. Stata Statistical Software: Release 17. College Station, TX: StataCorp LLC.)

Results

Patient, Tumor, and Treatment Characteristics

Patient and clinical characteristics are outlined in Table 1. Presenting symptoms and clinical findings at the time of diagnosis are described in Table 2. Of the 204 patients <21 years of age included in the study cohort, roughly half were female (n = 113, 55%). The median age at diagnosis was 11.1 years (range, 1–21). Our cohort consisted of 82 international patients (40%). Two patients were also diagnosed with tuberous sclerosis complex. Within the cohort, 119 (59%) patients had chordomas presenting in the upper and/or middle clivus, 74 (36%) in the lower clivus, 9 (4%) in the craniocervical junction, and 2 (1%) in the nasal cavity. Sixty-eight tumors (33%) extended to the brainstem, whereas extension to the upper cervical spine was found in 71 cases (35%). Patients were predominantly diagnosed with conventional chordoma (n = 131, 64%), 55 (27%) were chondroid, and 18 (9%) with poorly differentiated/ atypical chordomas. One hundred thirty-five tumors were not tested for brachyury or INI-1. Sixty-five patients had a tumor confirmed to be brachyury-positive, whereas only 4 tumors were negative for brachyury (2 conventional, one chondroid, and one poorly differentiated chordoma). Twenty-five tumors retained INI-1 expression, while 8 tumors were INI-1 negative (all poorly differentiated chordomas). All patients underwent at least one surgical resection (range, 1–8), with 32 (16%) patients who underwent GTR, 162 (80%) STR, and 10 (4%) patients that received a biopsy only. Surgical spinal stabilization with metallic hardware was performed in 43 (21%) patients. Ninety-five (47%) patients developed postoperative complications including newly acquired cranial nerve deficits (21%), CSF leak (18%), incision dehiscence (6%), and formation of pseudo meningocele (4%). Fifteen (7%) patients received pre-RT chemotherapy. Eleven of those patients had a poorly differentiated/atypical chordoma, 3 with a conventional type, and one patient had a mixed chondroid and cellular chordoma. The majority of patients (n = 202, 99%) received only postoperative radiotherapy (RT), though 2 (1%) patients with conventional chordoma also received pre-op RT experimentally. Prior to RT, 56 (28%) patients were found to have tumor regrowth in a pre-RT imaging scan. The median interval time between the last surgery and start of RT was 6.1 months (range, 1.1–57.4). Forty-seven (23%) patients received only PRT, and 157 (77%) received combined proton/photon RT (comboRT). The median total dose of radiation was 76.7 Gy (RBE) (range, 59.3–83.3), delivered in 1.2–2.2 Gy (RBE) daily fractions (1.2 Gy for BID). The median percentage of total dose delivered with protons was 74% (range 38%–100%). The median treatment duration for patients receiving standard fractionation was 57 days (range 43–78). Dosimetry to critical structures is presented in Supplemental Table.

Table 1.

Patient and Clinical Characteristics

Characteristic	N=204
Median age at diagnosis, years (range)	11.1 (0.8-21)
< 5, n (%)	28 (14%)
Gender, n (%)
Female	113 (55%)
Male	91 (45%)
Race, n (%)
Black	12 (6%)
White	163 (80%)
Hispanic	14 (7%)
Asian	7 (3%)
Other	8 (4%)
Home region, n (%)
USA	122 (60%)
International	82 (40%)
Location of chordoma, n (%)
Upper and/or middle clivus	119 (59%)
Lower clivus	74 (36%)
Cranio-cervical junction	9 (4%)
Nasal Cavity	2 (1%)
Extension site of chordoma, n (%)
Posterior Fossa	101 (49.5%)
Brainstem	68 (33%)
C-spine	71 (35%)
Nasopharynx	74 (36%)
Dorsum sella	45 (22%)
Cavernous sinus	44 (22%)
Occipital Condyle	41 (20%)
CPA	14 (7%)
Intradural	9 (4%)
Met at diagnosis, n (%)	8 (4%)
Pathology, n (%)
Conventional chordoma	131 (64%)
Chondroid subtype	55 (27%)
Poorly differentiated/Atypical	18 (9%)
Extent of resection, n (%)
GTR	32 (16%)
STR	162 (80)
Biopsy only	10 (4%)
Number of surgeries, n (%)
1	70 (34%)
2	67 (33%)
>2	67 (33%)
Hydrocephalus, n (%)	16 (8%)
CSF Leak, n (%)	37 (18%)
Spinal Stabilization, n (%)	43 (21%)
Pre-RT Chemo, n (%)	15 (7%)
Tumor progression prior to RT, n (%)	56 (28%)
RT Timing, n (%)
Pre- and post-op	2 (1%)
Post-op	202 (99%)
RT modality, n (%)
Proton radiotherapy	47 (23%)
Combo-radiotherapy	157 (77%)
Median Duration of RT, days (range)	56 (20-78)
BID RT, n (%)	11 (5%)
Median time from diagnosis to RT, months (range)	7.1 (1.5-59.2)
Median time from most recent surgery to RT, months (range)	6.1 (1.1-57.4)
Median RT Dose, Gy RBE (range)	76.7 (59.3-83.3)

Abbreviations: RT: Radiation Therapy, BID: Twice-Daily, GTR: Gross total resection, STR: Subtotal resection, CSF: Cerebrospinal Fluid, C-spine: cervical spine

Table 2.

Presenting Symptoms & Clinical Signs

2A. Presenting Symptom, n (%)	N=204
Headaches	91 (45%)
Neck pain	55 (27%)
Neuro-ophthalmologic
Diplopia	66 (32%)
Eye drop	8 (4%)
Visual Difficulties	9 (4%)
Abnormal eye movements	24 (12%)
Airway obstruction
Persistent nasal congestion	20 (10%)
Breathing difficulty	18 (9%)
Snoring	20 (10%)
Sleep apnea	14 (7%)
Ear
Decreased hearing	32 (16%)
Ear pain	162 (80%)
Dysphonia	28 (14%)
Dysphagia	14 (7%)
Drooling	5 (3%)
Facial Numbness	4 (2%)
Seizures	2 (1%)
Learning Disability	5 (3%)
Short-term memory issues	2 (1%)
Systemic symptoms
Nausea/Vomiting	23 (5%)
Dizziness	6 (3%)
Anorexia	8 (4%)
Failure to thrive	6 (3%)
Extremity Weakness	18 (9%)
Ataxia/ Gait abnormality	25 (12%)
Asymptomatic/Incidental finding	13 (6%)

2B. Clinical Finding on presentation, n (%)	N=204
Cranial Nerve Deficit
Optic	2 (1/%)
Oculomotor	16 (8%)
Trochlear	2 (2%)
Trigeminal	1 (1%)
Abducens	74 (36%)
Facial	3 (2%)
Vestibulocochlear	5 (3%)
Glossopharyngeal	9 (4%)
Vagus	7 (3%)
Accessory	3 (2%)
Hypoglossal	38 (9%)
Cervical Dystonia	11 (5%)
Hypotonia	4 (2%)
Hyperreflexia	13 (6%)
Dysesthesias	4 (2%)
Cerebellar signs	13 (6%)
Hemiparesis	7 (3%)
Quadriparesis	2 (1%)
Papilledema	4 (2%)
Adenoid hypertrophy	10 (5%)
Mass found in exam	11 (5%)

Disease Control

At a median follow-up time of 10 years (interquartile range, 5.4–16.4 years) from the date of diagnosis, 56 (27%) patients recurred. One patient was lost to follow-up after RT. Forty-two (21%) patients relapsed within the first 3 years after the completion of treatment. One patient with conventional chordoma of the upper clivus relapsed within the radiation field 14 years after receiving radiotherapy. With regards to first relapse, 33 patients relapsed locally, 15 patients had a distant failure while 2 patients had synchronous relapse (local and distant). For 6 patients, first relapse was in a surgical track. The most common site of distant metastasis was the spine (12/22) followed by lung (6/22), neck, and other locations (eyelid, calf). The median time from the end of RT to the first relapse was 19 months with very wide range of 13 days to 25 years. The median OS was 25.9 years, with 5-, 10-, and 20-year OS at 83.6%, 77.6%, and 63.9%, respectively (Figure 1A). The median DSS was not reached and the 5-, 10-, and 20-year DSS were 85.3%, 79.1% and 68.7%, respectively (Figure 1B). The median PFS was 25.4 years, with 5-, 10-, and 20-year PFS of 73.7%, 69.0%, and 63.6%, respectively (Figure 1C). The cumulative incidence of local failure at 5, 10, and 20 years were 19.5%, 23.8%, and 25.3 %, respectively. The cumulative incidence of distant failure at 5, 10, and 20 years were 9.0%, 11.4%, and 12.3 %, respectively (Figure 1 D and E).

Figure 1.

(A, B, and C) Kaplan–Meier plots of OS, DSS, and progression-free survival (PFS). The 5-, 10-, and 20-year overall survival (OS) were 83.6%, 77.6% and 63.9%, respectively. The 5-, 10-, and 20-year disease-specific survival (DSS) were 85.3%, 79.1% and 68.7%, respectively. The 5-, 10-, and 20-year PFS were 75.3%, 71%, and 69.6%, respectively. (D and E) Cumulative incidence of local failure and distant failure. The cumulative incidence of local failure at 5, 10, and 20 years were 19.5%, 23.8%, and 25.3 %, respectively. The cumulative incidence of distant failure at 5, 10, and 20 years were 9.0%, 11.4%, and 12.3 %, respectively.

Forty-three of the fifty-six patients who relapsed underwent salvage therapy including surgery, systemic treatment and/or radiotherapy. Salvage surgical resection was performed in 26 patients, whereas 11 patients received systemic therapy only. At the time of the analysis, 47 (23%) patients had died of disease. Of those, eleven patients relapsed distantly despite local tumor control. Thirty- four patients died of uncontrolled local tumor progression despite several salvage attempts. Two additional patients died of progressive disease with no available data about recurrence. Survival outcome of the 56 patients with relapse was calculated using the date of first relapse (local, distant, iatrogenic) as time zero. The median survival for all relapsed patients was 19 months with OS at 2-, 5-, and 10-year at 49.8%, 23.3%, and 8.0%, respectively. Forty-five of those patients who relapsed succumbed to their disease. Two patients from the relapsed cohort died due to complications associated with RT including multiple radiation-induced strokes and symptomatic brainstem injury. Overall, 6 patients died due to other causes including secondary tumors, RT-related symptomatic injury to brainstem and cerebellum, RT-induced stroke, acute respiratory failure, and recurrent meningitis.

In multivariable analyses, poorly differentiated tumor histology, lower clivus origin of the tumor, larger target tumor volume, and pre-RT tumor progression were found to be associated with worse OS, DSS, PFS, LC, and DC outcomes (Figure 2). Spinal stabilization was found to be a significant prognostic factor on a univariate analysis; however, it was nonsignificant on the multivariate analysis when it was adjusted for location, pathology, tumor size, and pre-RT tumor progression. Tables 3 and 4 summarizes those findings. Younger age and extent of the resection were not associated significantly with disease control. A re-run of our analysis showed that the extent of resection continued to not be associated with worse outcomes even if GTR/NTR patients were grouped together. Note that the extent of resection is significantly correlated with tumor volume with the median tumor volume on GTR about 5 times smaller than the median tumor volume on partial resection or biopsy (rank-sum P < .0001). Also, age is significantly associated with pathology with 25% of patients younger than 5 years had poorly differentiated/atypical pathology versus 6% of patient older than 5 years (rank-sum test P = .0007 with age as a continuous variable).

Figure 2.

Kaplan–Meier plots of OS stratified by pathology, DSS stratified by tumor volume, PFS stratified by tumor location, and PFS stratified by pre-RT tumor progression. All 3 prognostic factors are strongly associated with the respective outcome with log-rank P value < .0001.

Table 3.

Multivariable Cox Regression Analysis of OS, DSS, and PFS

Prognostic factor	OS		DSS		PFS
	HR (95%CI)	P value	HR (95%CI)	P value	HR (95%CI)	P value
Lower clivus origin	3.1 (1.7–5.8)	0.001	2.7 (1.4-5.1)	0.004	3.4 (1.9-6.1)	<0.0001
Poorly differentiated histology	5.6 (2.8-11.3)	<0.0001	5.7 (2.8-12.0)	<0.0001	5.3 (2.6-11.0)	<0.0001
Pre-RT tumor progression	2.0 (1.1-3.8)	0.02	1.8 (0.9-3.5)	0.07	1.9 (1.1-3.4)	0.03
Tumor volume (per 10cc)	1.1 (1.1-1.2)	<0.0001	1.1 (1.1-1.2)	<0.0001	1.2 (1.1-1.2)	<0.0001

Abbreviations: OS=Overall survival, DSS= Disease-specific survival, PFS= Progression-free survival, HR = hazard ratio

Table 4.

Multivariable Competing Risks Regression Analysis of LC and DC

Prognostic factor	LC		DC
	SHR (95%CI)	P value	SHR (95%CI)	P value
Lower clivus origin	3.6 (1.8-7.2)	<0.0001	2.5 (0.9-6.5)	0.07
Poorly differentiated histology	2.7 (0.9-8.0)	0.07	6.3 (2.4-16.6)	<0.0001
Pre-RT tumor progression	2.2 (1.1-4.4)	0.02	1.9 (0.8-4.8)	0.2
Tumor volume (per 10 cc)	1.1 (1.1-1.2)	0.001	1.2 (1.1-1.3)	0.001

Abbreviations: LC= Local control, DM= Distant control, SHR = subhazard ratio

Toxicities

Acute side effects during RT mostly consisted of skin erythema (61%) over the treatment area, otitis (11%), oropharyngeal mucositis (12%) associated with dysphagia, anorexia (15%), nausea (20%), and headaches (5%). No patients experienced grade 2+ acute toxicities. Thirty-eight (19%) patients had an uneventful RT course.

Late toxicities are summarized in Table 5. Of the 67 patients who had at least one formal endocrinology evaluation, pituitary abnormalities consisted of growth hormone deficiency (n = 39), hypothyroidism (n = 27), ACTH deficiency (n = 17), and hypogonadotropic hypogonadism (n = 16). Most of those patients received hormone replacement therapy. Thirty patients underwent audiology evaluation, and they were found to have unilateral (n = 10) or bilateral (n = 20) hearing loss which was management with ear tube placement, hearing aids or cochlear implant in 27 cases. Evidence of radiographic and/or symptomatic radiation-induced injury to critical structures was found in 24 patients. The median time from the completion of treatment to RT injury was 59.3 months (range, 4.6–286.9). Four patients developed symptomatic brainstem injury that required hospitalization. One of those patients subsequently developed right sided hemiparesis, that was attributed to radiation damage in the right cerebellar hemisphere and the medulla, and eventually passed away. The remainder of these patients died of progressive disease. Asymptomatic temporal lobe injury was documented in 4 patients. The earliest and the latest case of radiographic temporal lobe injury were documented 3- and 12 years following RT, respectively. Upper cervical osteomyelitis attributed to radiation was documented in 3 patients. Radiation-induced vasculopathy and/or stroke were found in 7 patients. Notably, one patient developed moya-moya arteriopathy 7 years following RT. One patient developed bilateral internal carotid artery occlusion and a stroke resulting in hemiparesis and ipsilateral facial droop 29 years after RT. During the follow-up period, 4 patients suffered from in-field secondary malignancies, including a meningioma of the middle cranial fossa, intrasellar osteosarcoma, glioblastoma of the brainstem, and spindle cell sarcoma of the cranial base, occurring 22.3-, 23.9-, 6.1-, and 2.1- years after treatment, respectively. The latter 2 patients remained chordoma-free, but they succumbed to their RT-induced tumors.

Table 5.

Late Toxicities

Toxicities, n	N=204
Endocrine abnormalities, n (%) *
GH Deficiency	39 (58%)
On replacement therapy	30 (77%)
Hypothyroidism	27 (40%)
On replacement therapy	26 (96%)
Adrenal Deficiency	17 (25%)
On replacement therapy	17 (100%)
Sex Hormone Deficiency	16 (24%)
On replacement therapy	14 (88%)
Auditory toxicities, n (%)
Recurrent otitis	27 (13%)
Hearing loss **	30 (15%)
Management (tubes, hearing aids, implant)	27 (90%)
Brainstem injury (radiographic and symptomatic)	4 (2%)
Temporal lobe injury (radiographic only)	4 (2%)
Vascular damage	7 (3%)
Osteomyelitis	3 (1.5%)
Palate fibrosis	1 (0.5%)
RT-induced esophageal stricture	1 (0.5%)
Secondary tumors within RT field	4 (2%)

Abbreviations: RT = radiation therapy, GH: growth hormone

^*Sixty-seven patients had at least one formal endocrinology evaluation

^**Unilateral or bilateral

Discussion

Pediatric skull base chordoma are rare, locally destructive tumors surrounded by critical anatomical structures. Based on the apparent physical advantage of charged particle RT and early outcomes in a few published case series, postoperative PRT is recommended to achieve better LC while decreasing late toxicities. Benk et al.²³ published the early experience of Harvard Cyclotron with 18 children treated with a mix of protons and photons for chordomas of the skull base and the upper cervical spine.

Twenty-seven years later, we review our institutional experience with treating 204 pediatric patients with skull base chordomas aiming to identify prognostic factors for tumor progression and patient survival and to evaluate the use of high-dose proton-based RT for this understudied disease. Our cohort exceeds the sum up of all previously published series combined.^23–29 Excellent outcomes were achieved with surgery and proton-based radiation with a median overall survival of 25.9 years. Our results show effective tumor control after postoperative PRT for pediatric chordomas despite the inclusion of patients with poorly differentiated chordoma, a subtype with inferior outcomes not seen in older adults. Previous smaller proton studies with short follow-up have shown estimates of 5-year OS and PFS ranging from 64% to 86% and 56% to 82%, respectively.^23–29 In spite of differences in irradiation techniques, including the transition from passive scattering protons and 3D conventional RT to pencil beam scanning and IMRT era, respectively, our analysis showed that the year of treatment was not significant for any tumor control outcomes. Despite the encouraging results for primary tumors, we found that the 2-, 5-, and 10-year overall survival (OS) for the 56 relapsed cases were 49.8%, 23.3%, and 8.0%, respectively. For patients who do not respond to surgery and radiation, there remains a need for novel targeted therapies. The limited salvage treatment options and the poor survival rates following tumor recurrence indicate that newly diagnosed chordoma patients should receive upfront referral to centers with chordoma expertise, and protons should strongly be considered above other RT modalities as more data exists for proton therapy for this disease and higher doses with better sparing of healthy tissues is achieved with PRT.

Multiple studies have looked at the factors predicting relapse and survival outcomes in adult patients, but the literature on pediatric population is limited. Chondroid histology has been found to be a worse prognostic factor in some pediatric series; however, other studies in adult population have shown that chondroid and conventional chordomas carry similar prognosis.^30,31 In our study, we did not find a statistically significant difference in disease progression and mortality between chondroid and conventional chordomas. However, we found that poorly differentiated or previously known in the literature as atypical chordomas are strongly associated with worse survival outcomes, as previously suggested.³² For these patients, chemotherapy is effective and often administered prior to definitive surgery and radiation with the most common chemotherapeutic agents used vincristine, doxorubicin, and cyclophosphamide alternating with ifosfamide and etoposide. Tazemetostat, a selective EZH2 inhibitor, has also shown activity for poorly differentiated chordomas.^33,34

This is the first manuscript to document this poor outcome on poorly differentiated chordomas; however, inferior outcomes are noted in case series and clinical practice. Complete resection of pediatric chordomas remains challenging and varies from 0% to 70%.^23,27,28,35 Extent of resection was not significantly associated with survival and tumor control when adjusted for other risk factors, especially lower clivus location, pathology, tumor volume, and the number of surgeries; however, target tumor volume, which was based on residual tumor and extent of initial tumor volume, was significantly associated with both survival and tumor control. Thus, our data still supports maximal resection and we believe that GTR is highly important for improved outcomes for this disease based on previously published outcomes.^27,36 Pre-RT tumor progression, sometimes due to delay in seeking treatment or RT avoidance after surgery, was strongly related with worse outcomes, suggesting that post-op radiation should be strongly considered for these patients, despite their young age. Further studies are needed to determine which patients may safely avoid or delay radiation. Multidisciplinary tumor management, including consultations with experienced skull base neurosurgeon, pediatric oncologist, and radiation oncologist, should be recommended at the time of diagnosis.

Initial tumor volume and location have been previously described as prognostic factors for chordomas. Studies by Berson et al³⁷ and Austin-Seymour et al.³⁸ showed that the risk of tumor progression increases with larger tumor volume. Similar to adult studies,^39–41 our multivariable analysis showed that tumor volume is a significant prognostic factor. Considering tumor location, several studies have concluded that chordomas arising from the upper cervical spine are associated with worse tumor control. Jahangiri et al.⁴² showed that presence of chordoma in the lower third of the clivus carry a high risk of recurrence. Eighty-three patients (41%) in our population had a tumor originating from the lower clivus or the cranio-cervical junction. Those patients were found to have worse outcomes compared to patients with chordomas of the upper/middle clivus. One can deduce that chordomas in the upper half of the clivus are more amenable to complete resection with an endoscopic endonasal approach in contrast to lower clival tumors that usually require a multi-staged open procedure with a high risk of morbidity. Six patients (3%) developed a recurrence along the path of the surgical pathway underscoring the importance of management by a surgeon with expertise in chordoma.

The presence of spinal stabilization has been reported to be a significant risk factor for local failures in chordoma patients.^43,44 For that reason, some proton institutions do not treat chordoma patients with metallic hardware in place.²⁵ In our institution, sometimes we have purposely included a portion of photon treatment for patients with metallic hardware. Some previously published studies suggested that hardware placement and extent of resection are significant prognostic factors for spinal and skull base chordomas, respectively.^27,45,46 In a multivariable actuarial analysis of our data hardware placement and extent of resection are not directly associated with analyzed outcomes in pediatric skull base chordomas when adjusted for pathology, pre-RT progression, tumor location, and size. It is important to note that this series includes only patients with hardware placed for skull based tumors in a location allowing for more beam angles than the lower/mobile spine and this finding cannot be extrapolated to patients with mobile spine chordomas.

At a median follow up of 9 years (interquartile range, 4–16 years) after RT, 24 patients developed radiographic and/or symptomatic radiation injury to critical structures. This relatively high number of complications is not usually seen in other chordoma series as the follow-up time is short. Four (2%) patients in our population developed symptomatic brainstem injury. This rate of brainstem toxicity is consistent with published data from other proton series in pediatric brain tumors treated with lower doses than the ones used for chordomas.⁴⁷ Radiation-induced secondary malignancies were occurred in 4 patients at 22.3-, 23.9-, 6.1-, and 2.1- years after treatment, respectively. All those patients were treated with passively scattered protons. Modern pencil beam scanning techniques provide the greatest opportunity to further reduce the risk of secondary tumor from protons.

The main limitation of this study was its retrospective nature which is inevitable when reporting outcomes for rare diseases due to lack of prospective series. Most of the patients were treated with a combined proton/photon therapy which is a less common treatment approach in modern era and is used as a strategy to mitigate hardware effects. Additionally, the majority of the patients were not local and given the rarity of pediatric chordoma, the post-treatment care varied greatly; some patients did not come to the attention of a physician until after they developed a recurrence or a serious late treatment effect. Moreover, endocrinology and auditory evaluations were only performed in subsets of patients. Nonetheless, this study stands as the largest to date with significant long-term follow up.

Conclusions

Our 40-year-experience shows that pediatric skull base chordomas can be successfully treated by an interdisciplinary combined-modality strategy integrating maximal safe resection followed by high-dose proton-based radiotherapy. In view of the high doses delivered and the surrounding structures, proton RT offers excellent tumor control with an acceptable toxicity profile. While the results of more studies with sufficiently long follow-up are necessary, the low rate of secondary tumors is encouraging and supports the use of protons for skull base chordomas. Magnetic Resonance Angiography should be considered in addition to MRIs for follow-up and patients that receive RT to the pituitary gland or hypothalamus should have annual endocrine panel or be followed by a neuro-endocrinologist. The enrollment of chordoma patients in the Natural History Study of Rare Solid Tumors (NCT03739827) can also provide a better understanding of the biological behavior of chordomas and guide therapeutic management.⁴⁸ Advances in genetics and molecular analysis may also lead to development of targeted therapies.^49,50 Novel therapeutic agents for progressive disease, multi-institutional collaborations and prospective clinical trials are needed to advance the field.

Funding

The work in this paper was supported by the George and Marie Vergottis Scholarship Award at Harvard Medical School and the Robert Small and Christine Olsen charitable fund. The work in this paper was supported by the George and Marie Vergottis Scholarship Award at Harvard Medical School and The Robert Small and Christine Olsen charitable fund.

Part of this study was presented at the 2020 American Society for Radiation Oncology (ASTRO) Annual Meeting and the 2021 American Radium Society (ARS) Annual Meeting.

JetPub Scientific Communications LLC assisted the authors in the preparation of this manuscript, in accordance with Good Publication Practice (GPP3) guidelines.

Conflict of interest statment

The authors declare that there is no conflict of interest.

Authorship

Conception and design: MII, SM. Collection and assembly of data: MII, AN, DK, AT, TU, SL, GPN, YH, AS, MP, KB, SR, NL, HK, MBC, AJ, SM. Data analysis and interpretation: MII, AN, DK, KB, SR, MP, SM. Manuscript writing: All authors. Final approval of manuscript: All authors. Author Responsible for Statistical Analysis Name & Email Address, Andrzej Niemierko PhD, aniemierko@mgh.harvard.edu