Proton Therapy for Low-Grade Gliomas: Results From a Prospective Trial

Abstract

BACKGROUND:

In this prospective study, the authors evaluated potential treatment toxicity and progression-free survival in patients with low-grade glioma who received treatment with proton radiation therapy.

METHODS:

Twenty patients with World Health Organization grade 2 glioma who were eligible for radiation therapy were enrolled in a prospective, single-arm trial of proton therapy. The patients received proton therapy at a dose of 54 Gy (relative biological effectiveness) in 30 fractions. Comprehensive baseline and regular post-treatment evaluations of neurocognitive function, neuroendocrine function, and quality of life (QOL) were performed.

RESULTS:

All 20 patients (median age, 37.5 years) tolerated treatment without difficulty. The median follow-up after proton therapy was 5.1 years. At baseline, intellectual functioning was within the normal range for the group and remained stable over time. Visuospatial ability, attention/working memory, and executive functioning also were within normal limits; however, baseline neurocognitive impairments were observed in language, memory, and processing speed in 8 patients. There was no overall decline in cognitive functioning over time. New endocrine dysfunction was detected in 6 patients, and all but 1 had received direct irradiation of the hypothalamic-pituitary axis. QOL assessment revealed no changes over time. The progression-free survival rate at 3 years was 85%, but it dropped to 40% at 5 years.

CONCLUSIONS:

Patients with low-grade glioma tolerate proton therapy well, and a subset develops neuroendocrine deficiencies. There is no evidence for overall decline in cognitive function or QOL.

INTRODUCTION

World Health Organization (WHO) grade 2 low-grade glioma (LGG) most commonly occurs in young adults. The efficacy of treatment for LGG is weighed heavily against potential adverse effects of therapies that can significantly impact patients’ lives, because the majority of these patients survive for >5 years, and substantial numbers live for well over a decade.^1–4 Thus, therapeutic strategies must consider not only survival and acute toxicities but also late effects that may affect quality of life (QOL). Although radiation therapy improves progression-free survival (PFS) in patients with LGG,⁵ cranial irradiation harbors known risks of neurocognitive and neuroendocrine dysfunction.^6–8

Proton radiation therapy offers an alternative to photon-based radiation therapies as a means of markedly reducing excess dose to surrounding normal tissues, thereby potentially mitigating the known adverse effects on neurocognitive and neuroendocrine function for patients with brain tumors. Patient-perceived QOL arguably can be the most important outcome when assessing treatment toxicity. Here, we report on a prospective study using proton therapy in the management of patients with LGG who underwent rigorous evaluation for potential treatment-associated morbidity and survival outcomes.

MATERIALS AND METHODS

Patients

Patients were defined as eligible if they had a pathologically confirmed, WHO grade 2 LGG and indications for radiation therapy. Indications included at least 1 of the following: 1) progressive or recurrent disease by imaging; 2) persistence, recurrence, or progression of symptoms attributable to the tumor; or 3) the presence of 1 or more other high-risk factors for early progression, including age ≥40 years, a mindbomb E3 ubiquitin protein ligase 1 (MIB-1) level ≥3%, or tumor size ≥6 cm. Additional inclusion criteria were age ≥18 years, a Karnofsky performance status ≥70, no comorbidities to suspect survival for <5 years, no obvious baseline cognitive deficits or other medical history to potentially compromise neurocognitive assessment, and no prior cranial irradiation. Informed consent was obtained from all patients for participation in this institutional review board-approved protocol.

Treatment

All patients received fractionated proton therapy. Magnetic resonance imaging (MRI) fusion was used to assist in target definition for all patients. The clinical target volume was defined as the composite of the T2-hyperintense tumor, any T1-enhancing disease, the abutting surgical bed, and a 1.5-cm expansion beyond that with respect of normal anatomy. A passive scattering technique was used with a beam-specific planning target volume equivalent, both of which are standard in our institutional proton therapy planning. These include 3.5% computed tomography density variation correction and an additional distal 1 mm for beam range uncertainty. Lateral margins were typically 8 mm to cover penumbra and set-up uncertainty. Compensator smearing of 3 mm was applied. The dose delivered was 54 gray (relative biological effectiveness) (Gy[RBE]) at 1.8 Gy(RBE) per fraction over 6 weeks. A sample proton plan and comparative photon plans are illustrated in Figure 1. Dosimetric analyses between proton and photon planning has been previously reported.⁹ Daily treatments typically involved 2 beams rotated from a total of 3 or 4 fields. Each field provided full coverage of the target volume.

Figure 1.

Dosimetric plans of proton therapy versus photon therapy for a low-grade glioma of the left temporal lobe are shown. Equivalent tumor target dose coverage is achieved but markedly less radiation is delivered to nontarget tissues with proton therapy.

Assessments

History and physical examination, brain MRI, neurocognitive function, neuroendocrine function, emotional well being, and other QOL measures were performed for each patient at baseline, defined as within 8 weeks of initiating radiation therapy, and at 3 months, 6 months, 12 months, 24 months, 36 months, 48 months, and 60 months after the completion of radiation therapy. Neuropsychological testing was not performed at 3 months, and an abbreviated test battery was used at the 6-month follow-up to minimize potential learned behavior. Patients were removed from protocol at the time they developed clear disease progression or if desired by the patient. Disease progression was defined by surgical confirmation after radiographic suspicion and sometimes with clinical symptoms.

All neurocognitive evaluations were performed by a neuropsychologist or by trained psychometricians under the supervision of the neuropsychologist. Assessed domains included intellectual functioning, visuospatial ability, attention and working memory, processing speed, executive functions, language, and memory (verbal and visual). A previously reported summary measure (the Clinical Trials Battery) also was included. Tests for each neurocognitive domain are listed in Table 2, and test scores (z scores) were averaged within each domain.¹⁰ Performances were considered impaired by z scores less than or equal to 21.5 relative to normative data, according to conventional neuropsychological practice.

TABLE 2.

Neurocognitive and Quality-of-Life Outcomes

Domain	Tests	Baseline Score: Mean ± SD (Range)	Average Score Change per Year: Average ± SE	P
Intellectual	WAIS-III Full Scale IQ	0.47 ± 0.56 (−0.47, −1.40)	0.07 ± 0.04	.1400
Visuospatial	WAIS-III Perceptual Organization Index	0.54 ± 0.69 (−0.60, −2.33)	0.13 ± 0.05	.0187
Language	WAIS-III Verbal Comprehension Index, Boston Naming Test, Auditory Naming Test	−0.50 ± 2.19 (−5.72, −1.00)	0.07 ± 0.09	.4462
Attention and working memory	WAIS-III Working Memory Index and Spatial Span; Continuous Performance Test: Inattention Score and Vigilance Score	0.24 ± 0.49 (−0.37, −1.58)	0.04 ± 0.04	.3292
Processing speed	WAIS-III Processing Speed Index; Trail Making Test A	0.06 ± 0.83 (−1.86, −1.33)	0.10 ± 0.07	.1679
Executive function	Trail Making Test B; Controlled Oral Word Association Test F-A-S; Wisconsin Card Sorting Test; Continuous Performance Test Impulsivity Score	−0.18 ± 0.62 (−1.18, −0.77)	0.12 ± 0.06	.0501
Verbal memory	HVLT-R: Total Recall, Delayed Recall, and Retention	−0.72 ± 1.19 (−2.67, −0.93)	0.04 ± 0.07	.5316
Visual memory	BVMT-R: Total Recall and Delayed Recall	−0.81 ± 1.41 (−3.00, −1.05)	−0.003 ± 0.06	.9644
Clinical trials battery	HVLT-R Total Recall; WMS-III Trails A and Trails B; Controlled Oral Word Association Test F-A-S	−0.35 ± 0.78 (−1.57, −1.13)	0.11 ± 0.06	.0742
Emotionala	Beck Anxiety Inventory	8.9 ± 8.0 (0–25)	−0.50 ± 0.36	.1870
	Beck Depression Inventory	12.71 ± 9.85 (0–31)	−0.05 ± 0.54	.9212
Quality of life	FACT-G Total Score	77.0 ± 18.4 (39–102)	0.41 ± 0.58	.4919
	FACT-Fatigue Score	32.7 ± 14.8 (8–52)	1.05 ± 0.44	.0265
	FACT-Br Total Score	131.0 ± 28.5 (84–174)	1.47 ± 0.89	.1154

Abbreviations: BVMT-R, Brief Visual Memory Test-R; FACT, Functional Assessment of Cancer Therapy; FACT-Br, Functional Assessment of Cancer Therapy-Brain; FACT-G, Functional Assessment of Cancer Therapy-General; HVLT-R, Hopkins Verbal Learning Test-R; SD, standard deviation; SE, standard error; WAIS-III, Weschler Adult Intelligence Scale, third edition; WMS, Weschler Memory Scale.

^aThree patients were not assessed by Beck Inventories at baseline.

To assess hypothalamic-pituitary (HP) axis function, laboratory studies were performed after overnight fasting. Serum levels of prolactin, insulin-like growth factor-1, thyroid-stimulating hormone, free thyroxine, morning cortisol, and cortisol after cosyntropin stimulation were obtained in all patients. The gonadal axis was assessed by morning testosterone, luteinizing hormone, and follicle-stimulating hormone levels in men and by luteinizing hormone and follicle-stimulating hormone levels in postmenopausal and reproductive age women without regular menstrual cycles; in the latter, estradiol levels also were measured.

Standardized self-rating forms were used to assess emotional functioning and perceived QOL at each time point (the Beck Depression Inventory, the Beck Anxiety Inventory, and the Functional Assessment of Cancer [FACT]-General, FACT-Brain, and FACT-Fatigue measures). Patient employment was also assessed as a surrogate of performance status and QOL.

Acute toxicities and late effects (>3 months after completing radiation) were assessed and scored according to Common Terminology Criteria for Adverse Events (version 3.0). Any symptom that was not clearly attributable to another cause was reported as potential radiation-associated toxicity. Neurocognitive, anxiety, depression, and neuroendocrine outcomes were reported separately by specific assessments.

Statistics

PFS and overall survival (OS) were defined from the first date patients started radiation therapy. PFS was measured to the date of disease progression or otherwise was censored at the last follow-up for progression-free patients who remained alive. OS was measured to the date of death or was censored at the latest follow-up. PFS and OS rates were estimated using the Kaplan-Meier method. Mixed linear models were used to analyze the repeated measures of Weschler Adult Intelligence Scale-third edition z scores, Beck Inventory scores, and FACT scores and were estimated using the restricted maximum-likelihood method. In these analyses, patient-specific intercepts and slopes are assumed to be random effects with unstructured covariance, whereas standard variance components are specified within patients. The fixed-slope effect is used to assess for decline in neurocognitive and QOL outcomes, but only negative trends are considered to be of clinical significance. The cumulative risk of developing new HP deficiency was estimated by treating disease progression considered as a competing risk of failure. Data analysis was performed using the SAS 9.3 statistical software package (SAS Institute Inc., Cary, NC) with P values based on a 2-sided hypothesis test.

RESULTS

Twenty patients with LGG were enrolled between October 2007 and May 2010. Demographic, clinical, and pathologic features of all patients are listed in Table 1. Eight patients had new tumor diagnoses, and 10 patients were followed after surgery and subsequently enrolled at the time of radiographic and/or clinical progression. The remaining 2 patients had recurrent disease after prior temozolomide. No patients received concurrent chemotherapy with proton radiation.

TABLE 1.

Baseline Patient Characteristics

Characteristic	No. of Patients (%)
Age: Median [range], y	37.5 [22–56]
Sex
Men	13 (65)
Women	7 (35)
Histology
Astrocytoma	7 (35)
Oligoastrocytoma	9 (45)
Oligodendroglioma	4 (20)
Dominant location
Frontal	9 (45)
Temporal	5 (25)
Frontotemporal	3 (15)
Parietal	2 (10)
Occipital	1 (5)
Tumor laterality
Right	12 (60)
Left	8 (40)
Greatest tumor dimension: Median [range], cm	6.3 [1.4–10.0]
MIB-1 index: Median [range], %	3.3 [<1 to 12]
1p/19q Status
Intacta	17 (85)
Loss	2 (10)
Unknown	1 (5)
Surgery
Gross total resection	4 (20)
Subtotal resectionb	12 (60)
Biopsy only	4 (20)
Symptoms at study baseline
None	3 (15)
Seizure under control	2 (10)
Seizure under control and other symptoms	4 (20)
Seizure	5 (25)
Seizure and other symptoms	6 (30)
Indication for radiation
Newly diagnosed high-risk	8 (40)
Persistent symptoms	6
MIB-1 ≥3%	5
Tumor size ≥6 cm	3
Age ≥40 y	3
Recurrent/progressive disease	12 (60)
Radiographic only	3
New symptoms with radiographic change	8
New symptoms, no radiographic change	1
Time from initial diagnosis to starting radiation: Median [range]	18 mo [30 d to 13 y]
Tumor volume: Median [range], cc	175 [23–404]

Abbreviation: MIB-1, mindbomb E3 ubiquitin protein ligase 1.

^aIt was assumed that 1p/19q was intact if it was not tested in astrocytoma.

^bThree of these patients underwent a second subtotal resection for disease progression or recurrence.

The median follow-up at the time of data cutoff was 5.1 years among the 9 patients who remained alive without progression (range, 3.3–5.2 years). Another 8 patients remained alive with progressive disease at a median follow-up of 4.9 years (range, 3.8–5.9 years). The PFS rate at 1 year, 3 years, and 5 years was 100%, 85%, and 40%, respectively (Fig. 2). The OS rate at 1 year, 3 years, and 5 years was 100%, 95%, and 84%, respectively (Fig. 3). Of the 11 patients who progressed, 2 had the presence of necrosis mixed with recurrent disease. It was unclear whether the necrosis was treatment related or secondary to the disease progression.

Figure 2.

Progression-free survival is illustrated for the study cohort of 20 patients with low-grade glioma.

Figure 3.

Overall survival is illustrated for the study cohort of 20 patients with low-grade glioma.

For neurocognitive function, 8 patients exhibited baseline impairment before radiation in 1 or more of the language, visual or verbal memory, or processing speed domains (Table 2). Performances in all neurocognitive domains remained stable or improved marginally over time for all patients. The median follow-up for neurocognitive measures was 3.2 years, which included 11 patients who were removed from protocol at the time of disease progression.

At baseline, HP dysfunction included the growth hormone axis and central hypothyroidism. The cumulative risk of developing new HP deficiency was 15%, 25%, and 30% at 1 year, 3 years, and 5 years, respectively. At the last follow-up, there were an additional 9 new hormonal deficits affecting 6 patients (Table 3). Three patients developed central hypothyroidism, 4 patients developed new adrenal insufficiency (with 1 case resolved 4 years later), and 2 men developed central hypogonadism. There were no cases of new deficiencies of growth hormone or female HP gonadal axes, and there was no new hyperprolactinemia. Two of the 6 patients who received >30 Gy(RBE) to the pituitary developed a new HP deficiency that was deemed potentially secondary to radiation exposure. A third patient developed transient central adrenal insufficiency and abnormal thyroid tests that most likely were caused by high-dose steroids. Among the 14 patients who received no more than 30 Gy(RBE) to any part of the pituitary, 2 developed a new endocrine deficiency that potentially was attributable to radiation. Three years after radiation, a third patient developed hypogonadism, which probably was not related to treatment, because exposure to the pituitary was <1 Gy(RBE).

TABLE 3.

Neuroendocrine Outcomes

	No./Total No. of Patients (%)
Axis	Baseline Deficits	New Deficitsa	Total Deficits
Growth hormone	4/20 (20)	0/16 (0)	4/20 (20)
Thyroid	2/20 (10)	3/18 (17)	5/20 (25)
Corticosteroid	0/20 (0)	4/20 (20)	4/20 (20)
Gonadal, men	0/13 (0)	2/13 (15)	2/13 (15)
Gonadal, women	0/7 (0)	0/7 (0)	0/7 (0)
Prolactin, women	0/7 (0)	0/7 (0)	0/7 (0)

^aCrude rates are shown.

Of the 17 patients who were assessed for emotional well being by the Beck Depression Inventory and the Beck Anxiety Inventory at baseline, 1 patient was severely depressed at baseline, and none reported severe anxiety. On average, there was no change to the emotional well being of patients over time. QOL determined by patient-reported questionnaires demonstrated no significant decline over time (Table 2). With regard to employment, 9 of 11 patients who were working full time at baseline before radiation therapy were continuing to work full time at the last follow-up or at the time of disease progression. One patient who was doing well electively stopped working, and 1 patient reduced work to part time because of symptoms associated with disease recurrence. Three patients who were employed part time at baseline were continuing to work part time or full time at the last follow-up. Of 2 patients who were on disability before radiation, 1 returned to full-time work. One of 4 patients who was not working at baseline before radiation decided to take on full-time employment.

Acute and late toxicities did not include any grade 4 or 5 events. All toxicities were within the expectations of cranial irradiation, including most common acute toxicities of fatigue, alopecia, headaches, and scalp erythema (Table 4). All other reported toxicities were continuations of symptoms that were present before radiation therapy. Late toxicities were most commonly persistent headaches, persistent fatigue, persistent alopecia, and new neurologic deficits. No late toxicities were unexpected (Table 5).

TABLE 4.

Acute Toxicities During Treatment and Within 3 Months of Completing Radiation

	No. of Events
CTCAE Event	Grade 1	Grade 2	Grade 3
Fatigue	8	10	2
Erythema	13	3	1
Headache	10	4	1
Seizure	NA	5	0
Taste/smell alteration	1	2	NA
Anorexia	1	1	0
Alopecia	17	0	NA
Cognitive deficit	1	0	0
Nausea/vomiting	4	0	0
Paresthesia	2	0	0
Dizziness	2	0	0
Anxiety	2	0	0
Insomnia	2	0	0
Motor dysfunction	2	0	0
Photophobia	1	0	0
Itching	1	0	0

Abbreviations: CTCAE, Common Terminology Criteria for Adverse Events; NA, not applicable.

TABLE 5.

Late Effects >3 Months After Radiation Completion

CTCAE Event	Grade 1	Grade 2	Grade 3
Fatigue	10	4	3
Seizure	NA	3	1
Headache	12	3	0
Insomnia	3	0	0
Alopecia	11	1	NA
Hearing	3	1	0
Tinnitus	NA	1	0
Dizziness/imbalance	5	1	0
Paresthesia	2	1	0
Temperature sensitivity	3	0	0
Weakness, cranial nerve	3	0	0
Weakness, peripheral	2	0	0
Stuttering/twitching	1	0	0
Cognitive dysarthria	1	0	0
Visual disturbance other than loss	3	0	0
Vision loss	1	0	0
Nausea	3	0	0
Taste alteration	1	0	NA
Proprioception deficit	1	0	0

Abbreviations: CTCAE, Common Terminology Criteria for Adverse Events; NA, not applicable.

DISCUSSION

Proton therapy remains a relatively novel radiation therapeutic option for patients because of its limited availability and high cost. The identification of appropriate patient populations for proton therapy is challenged by the limited resources, cost of studies, and difficulty in capturing useful endpoints of late toxicity. The benefits of proton therapy over other forms of radiation are primarily 2-fold. Radiation dose escalation may be beneficial for some diseases and can be achieved with proton therapy but is not feasible with other radiation modalities. Second, a reduction in radiation exposure to surrounding, nontarget, normal tissues may decrease the risk of radiation-associated adverse effects, and it is for this potential benefit that patients with LGG are considered for treatment with protons. These patients are often young, highly functioning adults who typically survive with their disease for many years and, thus, represent a particularly compelling patient population in which to prioritize QOL preservation. Our results are among the first detailed, prospective data on outcomes with proton therapy for patients with glioma and provide a uniquely rigorous longitudinal assessment of potential domains of radiation treatment effects with a relatively long median follow-up of 5 years. Our data support the excellent tolerance of treatment and relatively low treatment-related toxicities within the first 5 years after therapy.

Previous reports on the use of proton therapy for the treatment of LGG demonstrated feasibility and tumor control comparable to those achieved with photon therapy but provided limited or no data on toxicities of treatment.¹¹ Treatment approaches have also evolved in the last decade with technical advancements in radiation delivery, improved tumor definition by routine use of fusion magnetic resonance-based imaging, and increased understanding and standardization in the clinical management of patients with LGG. Our data concur with an early toxicity report of 19 patients with LGG who received proton therapy; those patients had a median imaging follow-up of 5 months, and the expected acute toxicities of predominantly alopecia and fatigue were reported.¹² In addition, we provide in-depth data on longer term and more comprehensive endpoints.

Neurocognitive dysfunction as a result of cranial irradiation may be the most feared sequelae of radiation therapy. The few studies that addressed this issue in patients with LGG demonstrated that cognitive dysfunction was secondary to both radiation^6,7,13,14 and other factors, such as the disease itself, surgery, or medical therapy.^14,15 The concern of additional treatment toxicity among physicians and patients alike has led to the common practice of postponing radiation for patients with LGG until there is clear evidence of tumor progression. We performed comprehensive assessments of all major domains of cognitive function. Our results suggest that proton radiation therapy is not associated with neurocognitive decline in at least the first 5 years after treatment, even for patients with pre-existing cognitive deficits that likely were incurred from their tumor. Because our patients were removed from study at the time of disease progression, it is less likely that any new neurocognitive deficits would be attributable to disease. In fact, the assessments all of evaluated domains of intelligence, visuospatial function, language, attention/working memory, and processing speed revealed slightly positive trends over time, although none were statistically significant.

A recent report from the Radiation Therapy Oncology Group 9802 trial on cognitive function in patients with LGG who underwent surveillance after surgery; or received radiation alone; or received radiation plus combined procarbazine, lomustine, and vincristine chemotherapy indicated that there was no decrement or difference in Mini-Mental Status Examination scores between the 3 cohorts through 5 years.¹⁶ Other LGG studies similarly have reported no decline in neurocognitive assessment after photon-based radiation therapy, but those results may have been limited by insensitivity of tests, practice effects, or inadequate follow-up.^17,18 In the current study, we used a more comprehensive battery of neuropsychological tests than has been previously reported and detected no cognitive decline attributable to proton therapy.

Pituitary dysfunction is a well recognized sequelae of cranial irradiation that becomes increasingly more prevalent with years after radiation therapy.¹⁹ Secondary hypopituitarism, specifically among patients with LGG who receive radiation, also has been documented.²⁰ Our base line screening results of HP deficits affecting 6 individuals demonstrate that this patient population is often hormonally compromised before irradiation. In addition, not all new deficits may be related to radiation. New medications or other health-related factors may contribute to these changes. Nonetheless, approximately 33% of our patients who received >30 Gy(RBE) to some part of the pituitary developed a new hormonal deficit that potentially was attributable to radiation versus only 2 of the 14 patients (14%) who were exposed to lower pituitary doses, consistent with an association between lower radiation exposure and a reduced risk of secondary hypopituitarism. Although protons always dosimetrically spare nontarget brain from excess radiation better than photon radiation, sparing of the pituitary varies specifically based on tumor location. Among the 14 patients who received pituitary doses <30 Gy(RBE), not all patients required the use of protons for dose sparing. Conversely, among the remaining 6 patients who received greater pituitary doses, this was not preventable even with the use of protons because of tumor target proximity (abutting or involving) to the pituitary. It should also be acknowledged that the full extent of new endocrine deficiencies in our study probably has not yet been captured given the spread of latencies to the onset of deficit and, thus, is expected to increase with time.²¹ The magnitude of potential benefit of protons in enabling the prevention of neuroendocrine deficits requires an understanding of the technical abilities and limitations of both photon and proton dosimetry.

Patient-perceived QOL is among the most important outcomes. Our assessments integrated a combination of both objective tests and patient questionnaires. Patients discovered effective ways of coping with their disease, as indicated by the few reported symptoms of depression and anxiety and also based on FACT questionnaires evaluating general well being, fatigue, and brain-specific queries. Most patients maintained or increased their preirradiation baseline level of employment, which is another indirect measure of stable or improved performance status. These findings are encouraging given concerns raised by older data from Surma-aho et al,⁷ who reported QOL decrements based on cognitive function and Karnofsky performance scores in patients with LGG. Those data are not entirely applicable to today’s approaches, because many of their patients received radiation to whole-brain fields at doses of 60 Gy.

Aside from the use of proton instead of photon radiation therapy, the therapeutic intervention in our patients was similar to that in other typical LGG patients. Our OS rate of 84% at 5 years is in keeping with trends in other recent LGG studies of increasingly longer survival, perhaps reflecting the collective advances in care for these patients, but also further fueling the parallel importance of optimizing long-term QOL.^1,2 It is important to emphasize the noninferior therapeutic survival outcomes with proton therapy to give merit to considering treatment options that aim to reduce toxicity. With the recent report of a survival benefit with the inclusion of chemotherapy in the management of patients with newly diagnosed LGG,² new questions arise regarding whether there should be greater emphasis for normal brain tissue sparing by radiation to reduce cumulative treatment toxicity, some of which may not be realized for years after treatment.

An obvious limitation in our study is the lack of randomization of patients between proton versus photon therapy. The utility of this study relies on the ability to place our data in context with other investigations. We used commonly accepted indications for radiation therapy and radiation dose schedules to minimize potential sources of differences in our outcomes. Our study includes a relatively small number of patients. Further investigation would benefit from a multicenter approach to achieve increased patient numbers.

Conclusion

Our results with proton therapy demonstrate feasibility of delivery, preservation of cognitive function, and maintenance of QOL. Larger studies that include the integration of standardized, contemporary chemotherapy regimens with randomization of proton versus photon therapy would be useful to further characterize potential differences in radiation late effects, such as effects on neuroendocrine function.