Int. J. Radiation Oncology Biol. Phys., Vol. 52, No. 4, pp. 1017–1024, 2002 Copyright © 2002 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/02/$–see front matter

PII S0360-3016(01)02725-0

CLINICAL INVESTIGATION

Pediatric Tumors

PROTON RADIOTHERAPY IN MANAGEMENT OF PEDIATRIC BASE OF SKULL TUMORS EUGEN B. HUG, M.D.,*†‡§ REINHART A. SWEENEY, M.D.,‡ PAMELA M. NURRE, B.S., R.T.(T.),‡ KITTY C. HOLLOWAY, R.T.,(R.)(T.),‡ JERRY D. SLATER, M.D.,‡ AND JOHN E. MUNZENRIDER, M.D.* *Department of Radiation Oncology, Massachusetts General Hospital, Boston, MA; †Harvard Cyclotron Laboratory, Cambridge, MA; Departments of ‡Radiation Medicine and §Pediatrics, Loma Linda University Medical Center, Loma Linda, CA Purpose: Primary skull base tumors of the developing child are rare and present a formidable challenge to both surgeons and radiation oncologists. Gross total resection with negative margins is rarely achieved, and the risks of functional, structural, and cosmetic deficits limit the radiation dose using conventional radiation techniques. Twenty-nine children and adolescents treated with conformal proton radiotherapy (proton RT) were analyzed to assess treatment efficacy and safety. Methods and Materials: Between July 1992 and April 1999, 29 patients with mesenchymal tumors underwent fractionated proton (13 patients) or fractionated combined proton and photon (16 patients) irradiation. The age at treatment ranged from 1 to 19 years (median 12); 14 patients were male and 15 female. Tumors were grouped as malignant or benign. Twenty patients had malignant histologic findings, including chordoma (n ⴝ 10), chondrosarcoma (n ⴝ 3), rhabdomyosarcoma (n ⴝ 4), and other sarcomas (n ⴝ 3). Target doses ranged between 50.4 and 78.6 Gy/cobalt Gray equivalent (CGE), delivered at doses of 1.8 –2.0 Gy/CGE per fraction. The benign histologic findings included giant cell tumors (n ⴝ 6), angiofibromas (n ⴝ 2), and chondroblastoma (n ⴝ 1). RT doses for this group ranged from 45.0 to 71.8 Gy/CGE. Despite maximal surgical resection, 28 (97%) of 29 patients had gross disease at the time of proton RT. Follow-up after proton RT ranged from 13 to 92 months (mean 40). Results: Of the 20 patients with malignant tumors, 5 (25%) had local failure; 1 patient had failure in the surgical access route and 3 patients developed distant metastases. Seven patients had died of progressive disease at the time of analysis. Local tumor control was maintained in 6 (60%) of 10 patients with chordoma, 3 (100%) of 3 with chondrosarcoma, 4 (100%) of 4 with rhabdomyosarcoma, and 2 (66%) of 3 with other sarcomas. The actuarial 5-year local control and overall survival rate was 72% and 56%, respectively, and the overall survival of the males was significantly superior to that of the female patients (p ⴝ 0.002). Of the patients with benign tumors, 1 patient (giant cell tumor) had local failure at 10 months. The other 8 patients continued to have local tumor control; all 9 patients were alive at last follow-up (actuarial 5-year local control and overall survival rate of 89% and 100%, respectively). Severe late effects (motor weakness and sensory deficits) were observed in 2 (7%) of 29 patients. Conclusion: Proton RT for children with aggressively recurring tumors after major skull base surgery can offer a considerable prospect of tumor control and survival. Longer follow-up is necessary to assess the real value of protons, in particular with regard to bone growth and cosmetic outcome. © 2002 Elsevier Science Inc. Protons, Base of skull tumors, Particle therapy, Conformal therapy, Pediatric malignancies.

INTRODUCTION Base of skull tumors pose a considerable challenge to the pediatric oncology team of neurosurgeons, radiation oncologists, and medical oncologists. A tumor primarily arising within the skull base or extending into it by way of secondary involvement is in close proximity to critical structures, such as the cranial nerves, arteries, brain parenchyma, and brainstem (1). Surgical resection is considered the primary treatment modality for many histologic features. Despite far-reaching

advancements in microscopic and image-guided neurosurgical procedures, complete resection often remains elusive. Immediate postoperative scans may reveal gross residual disease. Tumor adherent to a critical structure can result in positive resection margins. Piecemeal resection indicates the high likelihood of microscopic residual disease or of seeding into the operative site. In the adult patient, postoperative radiotherapy (RT) has been used successfully for several benign histologic features. However, the results using conventional RT methods

Reprint requests to: Eugen B. Hug, M.D., Department of Radiation Oncology, Dartmouth-Hitchcock Medical Center, One Medical Center Drive, Lebanon, NH 03756. Tel: (603) 650-6614; Fax: (603) 650-66616; E-mail: [email protected]

Acknowledgments—The authors thank Martin Fuss, M.D. for his valuable contributions and Bill Preston for editorial comments. Received Jun 4, 2001, and in revised form Sep 25, 2001. Accepted for publication Sep 27, 2001. 1017

1018

I. J. Radiation Oncology

● Biology ● Physics

have been disappointing for many semimalignant and malignant mesenchymal tumors, such as osteogenic sarcomas, chordomas, and chondrosarcomas. The development of 3-dimensional (3D)-planned, conformal irradiation offers the opportunity to deliver the high radiation doses required for these tumors, while respecting the critical normal-tissue tolerance dose levels. Proton RT has repeatedly demonstrated its ability to safely treat adult skull base patients and achieve higher local control and survival rates than previously reported (2, 3). Challenges in the treatment of adult patients are magnified in the pediatric and adolescent population. Severe structural, functional, and cosmetic impairments and deficits can supervene after RT. The severity of late effects depends on the age of the patient at the time of treatment, the amount of normal organ structure irradiated, and the dose delivered. Well-documented untoward sequelae include intellectual and sociobehavioral deficits after brain irradiation and cosmetic defects due to impaired development of facial structures. Both high and moderate radiation dose levels can cause severe late effects. In contrast to adult patients, in whom most treatment-related side effects will become manifest within 2– 4 years, the full expression of late effects in children may occur 5–10 years after treatment or even later (4, 5). Thus, especially in the pediatric patient, it is of paramount importance to optimally conform the high-dose radiation area around the target volume, while at the same time minimizing the volume of normal tissue exposed to even moderate radiation dose levels. This study reviewed a group of 29 pediatric patients with mesenchymal base of skull tumors, all treated under the guidance of one of us (E.B.H.). Treatments were conducted at the two North American institutions currently offering proton RT for skull base tumors: Harvard Cyclotron Laboratory, in cooperation with Massachusetts General Hospital (MGH/HCL), in Boston, MA, or Loma Linda University Medical Center (LLUMC), Loma Linda, CA. Local control, RT-associated side effects, and survival were analyzed. METHODS AND MATERIALS Patient and tumor characteristics Twenty-nine pediatric and adolescent patients diagnosed with mesenchymal neoplasms invading the skull base were treated with fractionated proton RT or combined proton and photon RT between July 1992 and April 1999. Patient age at treatment ranged from 1 to 19 years, and 14 patients were female and 15 male. Nineteen patients (66%) were white, 3 patients (10%) were black, and 7 patients (24%) were Hispanic. The symptoms before diagnosis varied according to the tumor location within the skull base and the absence or presence of intracranial extension. Headaches were by far the most frequently reported symptom (15 of 28 patients), followed by diplopia, reduced gag reflex, neck and/or occipital pain, nausea, and vomiting. The duration between symptom onset and diagnosis varied between ⬍1 and 41 months (mean 8, median 3). One patient had a documented

Volume 52, Number 4, 2002

prior diagnosis of enchondromatosis (Ollier’s disease), predisposing him to the development of chondrosarcoma. The varying degrees of local aggressiveness, sensitivity to RT, and malignant potential among the different tumor entities did not permit a summarized, pooled analysis of all 29 patients. We therefore classified tumors into two groups of either benign or malignant mesenchymal neoplasms, depending on their known clinical behavior of remaining either localized (despite a tendency toward aggressive local recurrence) or their potential for regional or distant spread. The malignant subgroup included 20 patients with chordoma (n ⫽ 10), low-to-intermediate grade chondrosarcoma (n ⫽ 3), rhabdomyosarcoma (n ⫽ 4), myxoid sarcoma (n ⫽ 1), malignant fibrous histiocytoma (n ⫽ 1), and epithelioid sarcoma (n ⫽ 1). The 9 benign pathologic tumors included giant cell tumor (n ⫽ 6), chondroblastoma (n ⫽ 1), and angiofibroma (n ⫽ 2). The patient and tumor parameters according to tumor histologic features are summarized in Table 1. All patients had undergone either surgical resection (median number of operations 2) or biopsy. All pathologic specimens were verified at the treating institution. Fifteen patients (52%) were referred for proton RT as part of their primary therapeutic management, and 14 patients (48%) were referred for recurrent disease. A comparison of preoperative and postoperative (pre-RT) images for each patient was used to evaluate the presence of gross residual disease. If a persistent, enhancing, soft-tissue region either within or next to the surgical resection bed was demonstrated on the postoperative images, it was assumed to be evidence of gross residual disease. According to this criterion, 28 patients (97%) in our study displayed radiographically visible residual or recurrent disease before RT. All but 1 patient had intracranial tumor extension. In 25 patients (86%), the tumor abutted or involved critical neurologic areas, with the brainstem and/or upper cervical spinal cord the most frequently reported structure. Treatment characteristics Treatments were conducted at two proton accelerator facilities in the United States under the guidance of one of us (E.B.H.). Seventeen patients were treated at MGH/HCL, and 12 were treated at LLUMC. Before initiating proton treatment, patients were fitted with immobilization devices (custom masks or vacuumassisted rigid bite blocks). All patients underwent imaging using a combination of thin-cut, 512 ⫻ 512 matrix CT imaging with i.v. contrast and gadolinium-enhanced MRI. Critical target and nontarget structures were delineated, and the gross tumor volume (GTV) and clinical target volume (CTV) were delineated according to International Commission on Radiation Units and Measurement (ICRU-50) criteria (6). The GTV was designed to include any enhancement, as per CT or MRI. The extent of the CTV depended on the malignant potential of the respective tumor histologic findings. It included an envelope of at least 0.5 cm around the GTV, depending on its anatomic structures and natural

Proton RT for pediatric skull base tumors

● E. B. HUG et al.

1019

Table 1. Patient and treatment characteristics of 29 pediatric and adolescent patients with mesenchymal tumors of the skull base Patient

Treatment

Histologic findings (No. of patients)

Mean AGE at RT (y) (range)

Gender (M/F)

Primary/recurrent tumor (n)

Median time between symptoms and diagnosis (mo)

Overall (29) Malignant (20) Chordoma (10) Chondrosarcoma (3) Rhabdomyosarcoma (4) Others* (3) Benign (9) Giant cell (6) Angiofibroma (2) Chondroblastoma (1)

12 (1–19) 10 (1–19) 11 (2–19) 15 (14–19) 3 (1–8) 10 (5–13) 15 (10–16) 15.5 (10–16) 15 (14–16) (13)

14/15 11/9 5/5 1/2 2/2 3/0 4/5 2/4 2M 1F

15/14 12/8 5/5 2/1 4/0 1/2 3/6 3/3 0/2 0/1

2.99 (0–41) 3.91 (0–24) 6.7 (1–24) 5.5 (0–17) 0.7 (0–1) 9 (1–24) 3 (1–41) 2.6 (1–10) 20.7 (1–41) 10

Protons and photons vs. protons only (n)

Median dose (CGE)

13/16 9/11 4/6 3/0 4/0 2/1 4/5 2/4 0/2 0/1

70 (45–78.6) 71.1 (50.4–78.6) 73.7 (70–78.6) 70.0 (69.6–70.2) 50.4 (50.4–59.6) 71.6 (66.6–75.8) 60.4 (45–71.8) 60.4 (57.6–64.5) 45 71.8

Abbreviations: RT ⫽ radiotherapy; M/F ⫽ male/female; CGE ⫽ cobalt Gray equivalent. Numbers in parentheses are the range. * Myxoid sarcoma, epithelioid sarcoma, malignant fibrous histiocytoma.

barriers. As a minimum, the postoperative tumor bed, representing the initial, preoperative disease extent, was also included. In general, no attempt was made to include the entire surgical access route. The details of the fractionated proton RT technique and delivery systems have been previously published for each of these facilities (8, 9). In brief, 3D treatment planning was performed using the Massachusetts General Hospital planning system (10) at both institutions. Beam angles and field configurations were optimized using beam’s-eye-view mode. Appropriate modulation wheels were introduced to spread out the Bragg peak, and custom apertures and boluses were made to ensure lateral and in-depth field conformity. Digital reconstructed radiographs from planning CT data were created to show the target position with respect to bony landmarks in multiple planes. These digital reconstructed radiographs were compared with orthogonal X-ray films obtained daily in the treatment position for each field to verify alignment. Patients were treated at LLUMC in the supine position using a 360° gantry system and at MGH/ HCL in the supine or seated position using a fixed horizontal beam arrangement. Proton treatments at MGH/HCL were offered 4 d/wk using the 160-MeV proton beam line. Megavoltage photons were used 1 d/wk because of the unavailability of the proton beam on that day. Treatments at LLUMC were conducted 5 d/wk using proton energies of 155 and 200 MeV. The proton dose was reported at either institution as the CGE, defined as the physical proton dose multiplied by a biological effectiveness factor (relative to 60 Co) of 1.1 (relative biological effectiveness factor) (7). In general, patients at both facilities received 1 treatment daily at 1.8 –2.0 CGE dose per fraction. Depending on the age of the patient and degree of ma-

turity, general anesthesia was used for the planning and daily treatments. Thirteen patients (45%) underwent proton RT alone, and 16 patients (55%) received a combination of megavoltage photon and proton RT. In general, photon radiation doses were kept at ⬍30% of the total prescribed dose. They ranged between 7.2 and 36 Gy (median 14.8). Two chordoma patients received preoperative photon doses of 17 and 20 Gy for rapidly growing disease followed by additional postoperative irradiation. The total prescribed doses depended on the primary histologic findings and ranged from 45 (for angiofibromas) to 78.6 CGE (for chordomas) with median tumor dose of 71 CGE for malignant and 60 CGE for benign histologic features (Table 1). The dose constraints for normal tissues were as follows: maximum permissible dose to optic nerves and optic chiasm, 60 CGE; brainstem and spinal cord surface, 64 CGE; and brainstem and spinal cord center, 53 CGE. Figure 1 depicts the isodose distribution for an 8-year-old patient with parameningeal rhabdomyosarcoma treated with proton RT. The prescribed dose to the CTV and the GTV was 41.4 CGE and 50.4 CGE, respectively. The tumor originated in the ethmoid sinuses and extended into the cribriform plate and had invaded through the right medial orbital wall to extend into the right orbit. The CTV included the primary site with margin, orbit minus the globe, and 2 cm of the base of the skull in all areas with intracranial extensions. Ninety percent of the lens received ⬍4.5 CGE total dose fractionated over the treatment course. Figure 2 illustrates the dose distribution of proton RT only, with concurrent chemotherapy, in a 13-year-old boy with high-grade malignant fibrous histiocytoma involving the superior nasal cavity, ethmoid sinuses, and cribriform

1020

I. J. Radiation Oncology

● Biology ● Physics

Volume 52, Number 4, 2002

outcome and to determine the significance levels. The minimal level of significance accepted was p ⱕ0.05. RESULTS

Fig. 1. Planning CT scan of an 8-year-old boy with parameningeal rhabdomyosarcoma. Transverse section with CTV, GTV, and nontarget optic nerves, lens, and pituitary gland contoured. Color display of dose distribution starting at 20 CGE (blue) with prescribed dose levels of 41.4 and 50.4 CGE to the CTV and GTV, respectively.

plate. Using the sharp distal dose falloff, as well as patch combinations, an arrangement was designed to permit target coverage of the ethmoid sinus region between both optic nerves. With doses of 50.4 CGE to the CTV and 66.6 CGE to the GTV, the dose to the optic chiasm was limited to 38 CGE maximal dose and ⱕ3 CGE to 50% of the chiasm volume. The right optic nerve received a maximal dose of 48 CGE and ⱕ20 CGE to 50% volume. The dose to the left optic nerve was 54 CGE (maximal dose) and 22 CGE (50% volume), respectively. Statistical analysis The actuarial control and survival rates were analyzed according to the product-limit (Kaplan–Meier) method (11). Stepwise Cox proportional hazard regression analysis (12) was performed to determine the univariate factors affecting

No patient was lost to follow-up. The observation time ranged from 13 to 92 months (mean 40), and 40% of patients were followed for ⱖ4 years. Twenty-seven patients (93%) were followed by serial CT and MRI. Local control was defined as stability or decrease of the tumor in the follow-up imaging studies. Two adolescent patients had no tumor progression at 3 years after treatment. They then refused additional imaging, but both were clinically stable 51 and 77 months after proton RT at the time of analysis. No patient died of intercurrent disease or treatment-related factors. Disease-specific survival, therefore, equated with overall survival. The patterns of failure, local control, and overall survival, as well as treatment-related side effects according to tumor histologic features, are summarized in Table 2. Malignant skull base neoplasms Ten patients with chordoma, 3 with low-to-intermediate grade chondrosarcoma, 4 with rhabdomyosarcoma, and 3 with other sarcomas (myxoid sarcoma, epithelioid sarcoma, and malignant fibrous histiocytoma) comprised the malignant group (see Table 1 for characteristics and treatment details). Sixty percent were treated for primary disease and 40% were referred from outside centers after recurrence. Radiographically distinct gross residual or recurrent tumor in the base of skull was present in 19 (95%) of 20 patients. All patients but 1 (rhabdomyosarcoma) had intracranial extension of the tumor, and 17 (85%) of 20 patients had tumor involvement or abutment of critical, normal, and dose-limiting structures such as the optic nerve, optic chiasm, or brainstem. Radiation doses ranged between 50.4 (rhabdomyosarcoma) and 78.4 CGE (mean dose 69). Follow-up ranged from 13 to 86 months (mean 37, median 30).

Fig. 2. Planning CT scan of a 13-year-old boy with malignant fibrous histiocytoma. (A) Transverse and (B) coronal section with CTV and nontarget optic nerves contoured. Color display of dose distribution starting at 20 CGE (blue) with prescribed dose levels of 50.4 and 66.6 CGE to CTV and GTV, respectively.

Proton RT for pediatric skull base tumors

● E. B. HUG et al.

1021

Table 2. Treatment results in 29 pediatric and adolescent patients with mesenchymal tumors of the skull base Patterns of failure* (n)

Outcome (%)

Histologic findings (No. of patients)

Local

Surgical access

Distant

Malignant (20) Chordomas (10) Chondrosarcomas (3) Rhabdomyosarcomas (4) Others† (3) Benign (9) Giant cell (6) Angiofibromas (2) Chondroblastoma (1)

5 4 0 0 1 1 1 0 0

1 1 0 0 0 0 0 0 0

3 1 0 2 0 0 0 0 0

Local control 15/20 (75) 6/10 (60) 3/3 (100) 4/4 (100) 2/3 (66) 8 (89) 5 (83) 2 (100) 1

Overall control

Overall survival

11 (55) 4 (40) 3 (100) 2 (50) 2/3 (66) 8 (89) 5 (83) 2 (100) 1

13 (65) 6 (60) 3 (100) 2 (50) 2/3 (66) 9 (100) 6 (100) 2 (100) 1

* No regional failures noted. † Myxoid sarcoma, epithelioid sarcoma, malignant fibrous histiocytoma.

Five patients (25%) experienced local failure; 4 patients with chordoma and 1 with epithelial cell sarcoma. One patient with chordoma had failure in the surgical access route and 3 patients developed distant metastasis (1 patient with chordoma and 2 with rhabdomyosarcoma). In terms of tumor histologic features, local control was maintained in 6 (60%) of 10 patients with chordoma, 3 (100%) of 3 patients with chondrosarcoma, 4 (100%) of 4 patients with rhabdomyosarcoma, and 2 (66%) of 3 patients with other sarcomas (Table 2). The actuarial 5-year local control rate for the entire group was 72% (Fig. 3). Seven patients had died of progressive disease at the time of analysis. Of these, 3 patients had developed distant metastasis despite local tumor control and died of progressive metastatic disease within 2–18 months after diagnosis.

Four additional patients died of uncontrolled local tumor progression despite various salvage attempts of repeated resection and chemotherapy. At 5 years, the rate of actuarial survival was 56% (Fig. 3). The patterns of failure according to tumor histologic features are summarized in Table 2. In this study, no correlation was found between the outcome parameters and tumor size (i.e., size of GTV, duration of symptoms, or treatment for primary vs. recurrent disease). However, in female patients, disease progressed in significantly more cases (7 of 9 patients) than in males (2 of 11) (p ⫽ 0.002). Benign skull base neoplasms Nine patients underwent proton RT for recurrent disease (7 patients) or for primary, aggressive tumors (2 patients)

Fig. 3. Actuarial rates of local control and overall survival of 20 pediatric patients with malignant skull base tumors after high-dose, conformal proton or combined proton/photon RT.

1022

I. J. Radiation Oncology

● Biology ● Physics

that had been subtotally resected, leaving known residual unresectable disease. Radiation doses were tailored according to the known histologic radiation sensitivity and ranged from 45 CGE (angiofibroma) to 71.8 CGE (chondroblastoma). At a follow-up period after completion of proton RT ranging from 12 to 92 months (median 46), 8 (89%) of 9 patients have continued tumor control, and all patients (100%) were currently alive at last follow-up. One patient with giant cell tumor had local failure at 10 months. He underwent surgery for salvage, but the tumor could not be completely resected. He was alive with growing tumor at last follow-up. No patient experienced distant failure. The 5-year actuarial local control and overall survival rate for the 7 patients with clinically comparable histologic findings (giant cell tumor or chondroblastoma) was 85% and 100%, respectively. The 2 patients with recurrent angiofibroma continued to have locally control with receding tumor 13 and 29 months after treatment.

RT-related side effects The acute side effects during RT were within the expected range. For most patients, the side effects consisted of temporary epilation over the treatment area, skin erythema, occasional headaches, fatigue, and loss of appetite. The development of oropharyngeal mucositis, resulting in dysphagia depended on the amount of posterior pharyngeal wall included in the treatment fields. These symptoms were treated symptomatically. No treatment breaks were required for patients who underwent proton RT alone. One patient with high-grade malignant fibrous histiocytoma who received concurrent chemotherapy required a 2-week treatment interruption and hospitalization for oral cavity infection and mucositis. This infection coincided with the blood count nadir. One additional patient developed a delayed acute reaction with severe headaches that occurred within 4 weeks of proton RT completion. After a short course of steroid medication, the severe headaches resolved within several weeks. All acute side effects completely subsided within the expected time frame after the completion of proton RT. Severe late effects were observed in 2 (7%) of 29 patients. One patient who presented with a most aggressive, rapidly growing, chordoma required two additional surgical resections for posterior fossa tumor regrowth after initial resection. She then underwent pre- and postoperative combined photon and proton RT to 75.4 CGE. Cerebellar and brainstem parenchymal damage, documented by serial MRI scans, developed within 16 months after treatment, and resulted in right-sided motor weakness and ataxia. This patient subsequently died of progressive disease. One additional patient developed temporal lobe damage within 6 months after a prescribed dose of 57 CGE for giant cell tumor, resulting in unilateral, upper extremity sensory deficit. We correlated the radiographic MRI changes with the isodose distribution. The highest isodose line encompassed by the MRI changes was the 72 CGE line for the chordoma

Volume 52, Number 4, 2002

patient and the 50 CGE line for the patient with giant cell tumor. Eight patients (27%) with intra- and parasellar tumors developed pituitary insufficiencies and required hormonal replacement therapy. The follow-up included evaluation of the patients’ current performance status. However, formal neurocognitive testing was only recently performed in some patients. Repeated interviews of patients and parents did not reveal any significant drop in the performance status of any of the patients exhibiting tumor control. Current status Of the 29 patients in this study, 19 were alive with controlled tumor, 3 were alive with tumor growth, and 7 had died of their disease at the time of analysis. Seventeen of the 22 surviving patients were in school or at work full time, 4 worked part time, and 1 received disability secondary to recurrent tumor growth. DISCUSSION The rationale for proton RT in the treatment of pediatric patients is based on two concepts, first to permit higher radiation doses in aggressive, “radioresistant” tumors, and second to decrease the sequelae of treatment by substituting protons for photons in tumors requiring relatively modest doses of radiation. Our patient population and its distinction between malignant and benign skull base tumors incorporated both considerations. As a result of its Bragg peak effect and its finite range, proton RT can allow high doses of radiation to be delivered near critical structures (8). The principal advantage of proton irradiation is based on its physical characteristics, enabling it to conform irradiation to targets of any size or shape and covering microscopic, as well as macroscopic, disease as clinically indicated. Verhey et al. (13) compared physical dose distributions for 3D photon and proton plans used in stereotactic radiosurgery and demonstrated that the dose conformity differential between protons and 3D photon plans increases with increasing target complexity, irregularity, and size. The bony skull base creates distinctly separate anatomic compartments (orbits, paranasal sinuses, inner and middle ear, infratemporal fossa, cavernous sinus, sella, and anterior, middle, and posterior fossa). Macroscopic and microscopic target volumes may have to include and outline these compartments, depending on the extent of involvement. This generally leads to highly irregular and complex target configurations, confounded by the presence of nearby critical structures. Proton RT has been used extensively for “radioresistant” skull base tumors in adult patients during the past 25 years (2, 3, 16). Very few reports are available on the topic of pediatric skull base tumors. Benk et al. (14) reported on 18 children, 4 –18 years of age, with chordomas of the base of skull and cervical spine treated between 1981 and 1990 with fractionated, combined photon and proton RT at MGH/HCL to prescribed doses ranging between 55.8 and 75.6 CGE

Proton RT for pediatric skull base tumors

(median dose 69). At 5 years, the rate of actuarial overall survival was 68% and the disease-free survival rate was 63%— comparable to the present series. Three children developed distant metastasis. In the present series, 1 girl with local tumor control developed multiple, bilateral pulmonary metastasis at 22 months and succumbed within 2 months. Despite the limitations of this inhomogenous group of tumors and limited patient numbers, we attempted a statistical analysis to identify early trends. All 5 female children and adolescents with chordomas had local or distant failure compared with 1 of 5 males. In the large series at the Massachusetts General Hospital of adult patients with chordomas, female patients fared significantly worse than did males (2). Possible explanations were discussed previously (15), but thus far the causes are unknown. Estrogen and progesterone receptor studies did not reveal a difference (16). We share the opinion of Borba et al. (17) that chordomas in the pediatric population can in general behave more aggressively than in adults. Children appear to have a relative short history of symptoms, a short interval to progression after initial surgery, and a documented tendency toward the early development of metastatic disease. The present series comprised a high-risk group, selected for proton RT because of complex and extensive tumor configuration. Nearly all patients had gross tumor abutting critical, normal structures at time of RT. Debus et al. (18) reported previously dose-limiting normal structure involvement as a significant prognostic factor of failure after RT. In our series, all patients with local failure had tumor abutment of the optic nerve, optic chiasm, or brainstem. In adult patients with skull base chondrosarcomas, proton RT can achieve excellent and lasting tumor control. In the MGH/HCL series, 98% of 229 patients with low-grade chondrosarcomas remained with actuarial tumor control at 5 and 10 years after combined photon and proton RT (2). We are not aware of any publication specifically addressing the topic of skull base chondrosarcomas in children. Our 3 chondrosarcoma patients continued to have local control and were alive and well at last follow-up. It is our hope that the pediatric results will continue to parallel the excellent results of proton irradiation obtained in adult chondrosarcoma patients. After conventional photon RT of between 45 and 55 Gy, local control rates for giant cell tumors range from 70% to 85% (19, 20). Tumors of pathologically benign histologic morphology can be locally aggressive. Once a tumor has declared its aggressive potential after the first recurrence, comprehensive and definitive treatment is indicated to avoid repeated surgical procedures with their risks and morbidities. However, if RT fails, surgical salvage is rarely successful. To maximize the chances of permanent tumor control and with the consideration of the ability to conform the isodoses, we delivered relatively high radiation doses for benign tumors (angiofibroma, 45 CGE; giant cell tumor, median dose 60.4 CGE). Of 9 patients, 8 (89%) were alive with local tumor control. The primary rationale for the use of protons for angiofi-

● E. B. HUG et al.

1023

bromas of the infratemporal fossa and parameningeal rhabdomyosarcomas was to minimize the amount of facial structures (bones, teeth, lacrimal gland, temporomandibular joint, lens) exposed to potentially damaging radiation doses. Target dose levels of 45–55 Gy could certainly be delivered with 3D conformal photon techniques, including intensitymodulated therapy. In a recent editorial, Tarbell et al. (21) has emphasized that the “physical advantage of protons for single beams extends to multibeam treatments, including intensity modulation,” given the dose distribution properties of protons. To conform the dose to the target by using photons and multiple field arrangements, increasing volumes of surrounding normal tissues are exposed to irradiation in the low to mid-dose range, reflecting the entrance and exit dose contributions from a multitude of treatment field angles (21). A dose conformity comparison of conformal protons and modern, conformal photon techniques for pediatric skull base tumors would certainly be desirable. However, that is beyond the scope of this clinical review analysis and requires a separate project. The principal disadvantage of 3D photon conformal RT for pediatric patients—the increase of integral volume receiving radiation— has recently been quantified by Fuss et al. (22) in a comparative study on optic pathway gliomas. The advantages of proton vs. 3D photon plans became increasingly apparent with increasing target size and tumor complexity. Even in small tumors, conformity of 3D photon irradiation came at the expense of a larger amount of normal tissue receiving moderate to low doses. Lin et al. (23), also for the Loma Linda University proton group, compared photon and proton plans of posterior fossa treatment and analyzed the resulting doses to the temporal lobes. With equal coverage of the posterior fossa, doses to 50% and 10% of the temporal lobe volume were restricted to 2% and 67% by use of protons vs. 56% and 100% by 3D photons, respectively. These consideration have guided us in our recommendations to use proton radiation—particularly in the very young child—for histologically benign skull base tumors that require only moderate dose levels for successful tumor control. Side effects of RT in this series were within acceptable ranges, considering anatomic location, critical normal tissue involvement in most patients, and the high doses required for the malignant tumor group. One patient developed Grade 3 cerebellar and brainstem damage, having previously required four major base of skull resections overall for rapid tumor regrowth. Debus et al. (18) in a detailed analysis of brainstem tolerance after combined proton and photon RT for skull base chordomas reported that patients who required ⬎2 major neurosurgical procedures before RT were at increased risk of radiation-induced damage of the brainstem. The authors postulated that repeated surgical procedures with disruption of the microvasculature could lead to decreased radiation tolerance of brain parenchyma. Santoni et al. (24) analyzed a similar patient population treated at the same institution with combined protons and photons for skull base chordomas. They reported severe and

1024

I. J. Radiation Oncology

● Biology ● Physics

symptomatic temporal lobe injury in 8 of 92 patients. Most damage to temporal lobes manifested itself within 2 years. Pituitary insufficiency in 8 patients of this series was not surprising and reflected the limitations of any conformal RT modality. Conformal RT to the skull base reduces the dose to the hypothalamic-pituitary axis and to the pituitary gland—if the target volume is close, but still distinct, from the sella. However, if the sella itself is part of the target volume because of tumor involvement, this will invariably result in irradiation of the pituitary gland. This was the case in our patients. Careful review of the tumor extent before RT by the radiation oncologist plays an important role in the initial counseling of the patient and parents about the anticipated side effects of RT. It was beyond the scope of this review to evaluate the effects of proton RT on the neurocognitive functions. An accurate assessment can only be performed in a prospective manner on a larger patient population. No untoward effects were seen in a limited study of adult patients with skull base tumors (25).

Volume 52, Number 4, 2002

In summary, the results of this clinical review suggest that conformal RT, in this series delivered by use of proton RT, can offer children with aggressive, recurrent, or unresectable skull base tumors the prospect of lasting tumor control and survival. Longer follow-up is necessary to assess the real value of protons, in particular with regard to bone growth and cosmetic outcome. It is our recurring experience that young children in particular will undergo repeated, subtotal resections for aggressive regrowth without the hope of definitive tumor control and that RT is not offered for fear of its severe side effects. Today’s conformal RT modalities can achieve excellent dose conformity with reduced exposure of normal tissues to even low and moderate amounts of RT. It appears appropriate to reevaluate the role of definitive RT in the overall treatment management recommendations for pediatric skull base tumors. We hope that these results will serve as the bases for hypothesis development in future prospective trials of conformal RT.

REFERENCES 1. Hug EB, Slater JD. Proton radiation therapy for pediatric malignancies: Status report. Strahlenther Onkol 1999; 175(Suppl. 2):89 –91. 2. Munzenrider JE, Liebsch NJ. Proton therapy for tumors of the skull base. Strahlenther Onkol 1999;175(Suppl. 2):57– 63. 3. Hug EB, Loredo LN, Slater JD, et al. Proton radiation therapy for chordomas and chondrosarcomas of the skull base. J Neurosurg 1999;91:432– 439. 4. Donahue B. Short- and long-term complications of radiation therapy for pediatric brain tumors. Pediatr Neurosurg 1992; 18:207–217. 5. Merchant TE, Wang MH, Haida T, et al. Medulloblastoma: Long-term results for patients treated with definitive radiation therapy during the computed tomography era. Int J Radiat Oncol Biol Phys 1996;36:29 –35. 6. International Commission of Radiation Units and Measurements. Prescribing, recording and reporting proton beam therapy. Bethesda: ICRU; 1993. 7. Urano M, Verhey LJ, Goitein M, et al. Relative biological effectiveness of modulated proton beams in various murine tissues. Int J Radiat Oncol Biol Phys 1984;10:509 –514. 8. Slater JM, Archambeau JO, Miller DW et al. The proton Treatment Center at Loma Linda University Medical Center: Rationale for and description of its development. Int J Radiat Oncol Biol Phys 1992;22:383–389. 9. Munzenrider JE, Austin-Seymour M, Blitzer PJ, et al. Proton therapy at Harvard. Strahlenther Onkol 985;161:756 –763. 10. Goitein M, Abrams M. Multidimensional treatment planning: Beam’s eye view, back projection and projection through CT sections. Int J Radiat Oncol Biol Phys 1983;9:789 –797. 11. Kaplan E, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457– 481. 12. Cox DR, Oakes N. Analysis of survival data. New York: Chapman and Hall; 1988. 13. Verhey LJ, Smith V, Serago CF. Comparison of radiosurgery treatment modalities based on physical dose distributions. Int J Radiat Oncol Biol Phys 1998;40:497–505. 14. Benk V, Liebsch NJ, Munzenrider JE, et al. Base of skull and cervical spine chordomas in children treated by high-dose irradiation. Int J Radiat Oncol Biol Phys 1995;31:577–581. 15. Halperin EC. Why is female sex an independent predictor of

16.

17. 18. 19. 20. 21.

22.

23.

24.

25.

shortened overall survival after photon/proton radiation therapy for skull base chordomas? Int J Radiat Oncol Biol Phys 1997;38:225–230. Munzenrider JE, Liebsch NJ, Efird JT. Chordomas and chondrosarcomas of skull base: Treatment with fractionated x-ray and proton radiotherapy. In: Johnson JT; Didolkar M, editors. Head and neck cancer. Vol. III. New York: Elsevier Science; 1993. p. 649 – 654. Borba LAB, Al-Mefty O, Mrak RE, et al. Cranial chordomas in children and adolescents. J Neurosurg 1996;84:584 –591. Debus J, Hug EB, Liebsch NJ, et al. Brainstem tolerance to conformal radiotherapy of skull base tumors. Int J Radiat Oncol Biol Phys 1997;39:967–975. Chen ZX, Gu DZ, Yo ZH, et al. Radiation therapy of giant cell tumor of bone: Analysis of 35 patients. Int J Radiat Oncol Biol Phys 1986;12:329 –334. Schwartz LH, Okunieff PG, Rosenberg A, et al. Radiation therapy in the treatment of difficult giant cell tumor. Int J Radiat Oncol Biol Phys 1989;17:1085–1088. Tarbell NJ, Smith AR, Adams J, et al. The challenge of conformal radiotherapy in the curative treatment of medulloblastoma (editorial). Int J Radiat Oncol Biol Phys 2000;46: 265–266. Fuss M, Hug EB, Schaefer RA, et al. Proton radiation therapy (PRT) for pediatric optic pathway gliomas: Comparison with 3D planned conventional photons and a standard photon technique. Int J Radiat Oncol Biol Phys 1999;45:1117–1126. Lin R, Hug EB, Schaefer RA, et al. Conformal proton radiation therapy of the posterior fossa: A study comparing protons with three-dimensional planned photons in limiting dose to auditory structures. Int J Radiat Oncol Biol Phys 2000;48: 1219 –1226. Santoni R, Liebsch N, Finkelstein DM, et al. Temporal lobe (TL) damage following surgery and high-dose photon and proton irradiation in 96 patient affected by chordomas and chondrosarcomas of the base of the skull. Int J Radiat Oncol Biol Phys 1998;41:59 – 68. Glosser G, McManus P, Munzenrider JE, et al. Neuropsychological function in adults after high dose fractionated radiation therapy of skull base tumors. Int J Radiat Oncol Biol Phys 1997;38:231–239.