Proton therapy is increasingly used to treat pediatric brain tumors. However, the response of both tumors and healthy tissues to proton therapy is currently under investigation. One way of assessing this response is magnetic resonance (MR) diffusion tensor imaging (DTI), which can measure molecular mobility at the cellular level, quantified by the apparent diffusion coefficient (ADC). In addition, DTI may reveal axonal fiber directional information in white matter, quantified by fractional anisotropy (FA). Here we report use of DTI to assess tumor and unexposed healthy brain tissue responses in a child who received proton therapy for juvenile pilocytic astrocytoma.
A 10-year-old boy with recurrent juvenile pilocytic astrocytoma of the left thalamus received proton therapy to a dose of 50.4Gy (RBE) in 28 fractions. Functional magnetic resonance imaging was used to select beam angles for treatment planning. Over the course of the 7-year follow-up period, magnetic resonance imaging including DTI was done to assess response. The MR images were registered to the treatment-planning computed tomography scan, and the gross tumor volume (GTV) was mapped onto the MR images at each follow-up. The GTV contour was then mirrored to the right side of brain through the midline to represent unexposed healthy brain tissue.
Proton therapy delivered the full prescribed dose to the target while completely sparing the contralateral brain. The MR ADC images obtained before and after proton therapy showed that enhancement corresponding to the GTV had nearly disappeared by 25 months. The ADC and FA measurements confirmed that contralateral healthy brain tissue was not affected, and the GTV reverted to clinically normal ADC and FA values.
Use of DTI allowed quantitative evaluation of tumor and healthy brain tissue responses to proton therapy.
Juvenile pilocytic astrocytomas (JPAs) are World Health Organization grade I brain tumors that typically occur in children or young adults. These tumors commonly show marked enhancement and high glucose metabolism on imaging [1, 2]. The primary treatment for JPAs is surgery with radiation reserved for unresectable, progressive disease [3, 4]. Given its ability to spare the healthy tissues beyond the range of the beam, proton therapy is increasingly being used to treat pediatric brain tumors, including JPAs [5–10]. However, the biological response of tumors and healthy tissue to proton therapy—as opposed to conventional photon (X-ray) therapy—remains under investigation.
Diffusion tensor imaging (DTI) can be used to characterize treatment-related changes in tissue microstructures, quantified by the apparent diffusion coefficient (ADC) and fractional anisotropy (FA). The ADC is a magnitude of molecular mobility, and it is inversely related to the tumor or tissue cellularity. The FA reveals information on axonal fiber direction and the degree of anisotropy in the white matter. Decreases or losses of FA are related to disrupted microstructures. Both ADC and FA have been used to quantitatively measure the response of tumors to radiation therapy as well as radiation-induced changes in white matter [11–15]. Here we report responses of tumor and healthy brain tissue to proton therapy, based on quantitative analysis of DTI, in a child with JPA. This report serves as the proof of principle for use of this technique for sparing uninvolved brain, while also providing valuable information on the use of magnetic resonance imaging (MRI) for treatment planning and novel insights into tumor response as assessed by DTI.
The patient, a 10-year-old boy, presented to the University of Texas MD Anderson Cancer Center with a recurrent JPA of the left thalamus. Five years earlier, his daily headaches had prompted a brain computed tomography (CT) scan, which had revealed hydrocephalus and a mass, and a brain MRI scan showed a golf ball–sized lesion near the ventricles with associated hydrocephalus. A biopsy confirmed the diagnosis of JPA, approximately 90% of which was successfully resected.
Given the patient's age and the potential for disease progression, chemotherapy with carboplatin and vincristine was begun but was stopped after the patient had an anaphylactic reaction to the carboplatin. The patient was then followed with serial MRI. The disease remained stable, and there was no progression for 3 years. At that time evidence of tumor regrowth led the patient's family to self-refer to the University of Texas MD Anderson Cancer Center. The patient was evaluated and proton therapy was recommended. Functional and volumetric MRI were performed to aid in treatment planning, and neuropsychiatric and neurocognitive tests were done to provide a pretreatment baseline of cognitive functioning.
Proton therapy was delivered via a passive scatter technique with a cumulative dose of 50.4 Gy (RBE) delivered in 28 fractions. Treatment was well tolerated; the patient experienced slight fatigue but no significant hair loss or skin reaction. Follow-up visits, including MRI and neuropsychological and neurocognitive testing, were initially scheduled for every 3 months and subsequently scaled back to every 6 months and then every year; the patient has been followed for nearly 7 years after completion of therapy.
Proton Therapy Treatment Planning
Treatment plans were developed from CT images obtained with 120-kV X-rays on a 16-slice scanner (LightSpeed RT 16, GE Healthcare, Waukesha, Wisconsin). During treatment simulations and delivery, the patient was immobilized with a Precise Bite positioner attached to a thermoplastic mask with a polyurethane foam headrest (CIVCO Medical Solutions, Orange City, Iowa). The gross target volume (GTV) was defined based on magnetic resonance (MR) images registered with the planning CT images. The clinical target volume was defined as the GTV plus a 3-mm expansion, and the planning target volume was the clinical target volume plus a 2-mm to 3-mm expansion. The range uncertainty was assumed to be 3.5% of the proton beam range plus 3 mm. Passive scattered proton therapy was delivered in a dose of 50.4 Gy (RBE) to the clinical target volume in 28 fractions; the RBE value was assumed to be 1.1. For such tumors, typical dose constraints for brainstem, optic nerves, and chiasm are less than 54 Gy (RBE), respectively which is easily achievable given the prescription dose. Great care is taken, however, to avoid beam geometries in which the end of range would be in critical healthy tissues given the potential for higher biologically effective doses distal regions. For the hippocampus and receptive speech areas, the dose was reduced as much as reasonably achievable while maintaining reasonable target coverage. Treatment fields were selected to avoid significant dose to brain tissues associated with functions such as receptive speech, on the left side, assisted by functional MRI (Figure 1A and 1F), which resulted in a 3-field plan: Field 1 was a left oblique superior-inferior vertex field, field 2 was a left anterior oblique field, and field 3 was a slightly right (5° from anterior) anterior oblique field (Figure 1B, 1D, and E). The treatment plan was created with an Eclipse treatment planning system (TPS version 8.1, Varian Medical Systems, Palo Alto, California) with a proton module. Treatment was delivered using a Hitachi Probeat proton therapy system (Hitachi, Ltd, Tokyo, Japan).
Magnetic Resonance Diffusion Tensor Imaging Data Acquisition and Analysis
The patient underwent baseline MRI, pretreatment functional MRI scanning, and follow-up scans. Response was assessed from the conventional brain MRI features: T1-weighted and T2-weighted fluid attenuation inversion recovery, post-contrast T1-weighted images, and DTI. All conventional MR scans were obtained with a GE 1.5T system (GE Healthcare) with an 8-channel phase array brain coil, maximum gradient strength of 33 mT/m, and slew rate of 120 T/millisecond (ms). Brain functional MRI, with a focus on speech and motor tasks, was done with a GE 3.0T system with an 8-channel brain coil, maximum gradient strength of 44 mT/m, and slew rate of 150 T/ms. DTI settings were a 22-cm field of view, matrix size 128 × 128, 27 diffusion directions, 5.0-mm slice thickness, and b values of 1200 sec/mm2; the ADC and FA maps were generated on the scanner right after the DTI data had been acquired. The MR images were registered to the treatment planning CT scan, and the GTV was mapped onto each MR image. The GTV contour was then mirrored to the right side of the brain through the midline (Figure 2) to represent healthy brain tissue. The ADC and FA values were measured within the GTV and mirrored GTV.
Cognitive Tests and Correlation
The patient was seen by a specialist in pediatric neuropsychology before treatment and at follow-up visits as part of routine care. Age-appropriate tests were used to assess intellectual functioning, language, verbal and nonverbal memory, and attention. Coordination, fine motor speed, and tactile perception were also tested.
As illustrated in Figure 1, the proton dose distributions near the center of the target volume on the planning CT scans (left) and on the MRI-fluid attenuation inversion recovery scans (right) demonstrate that the target volume received the full prescribed dose and the contralateral brain was nearly completely spared. The maximum point doses to brainstem, optic chiasm, receptive speech, and right hippocampus were 52.39, 51.14, 2.90, and 0.01 Gy (RBE), respectively. Successful treatment of the tumor was further confirmed on a series of MRI ADC images from pretreatment to long-term follow-ups. Shown in Figure 3A through 3C are examples from pretreatment and at 10 and 25 months after proton therapy. By 25 months after completion of therapy, the enhancement corresponding to the GTV had nearly disappeared (Figure 3C). The corresponding post-contrast T1-weighted images are displayed in Figure 3D through 3F. These qualitative findings correspond to quantitative measurements of ADC (Figure 4) and FA (Figure 5), which demonstrate that contralateral normal brain tissue was not affected by proton therapy and that the tumor volume had reverted to normal ADC and FA values 5 years after completion of treatment.
Proton therapy was well tolerated, with no significant side effects aside from mild fatigue. At approximately 7 months after treatment, a slight increase in lesion size visible on MRI was suggestive of possible pseudoprogression. However, after that single measurement, which was not associated with clinical symptoms or signs, the lesion remained stable and subsequently became smaller.
Performance on neuropsychological and neurocognitive tests showed little change over time, remaining average overall. Among all the tests, the patient was strongest in working memory and processing speed, 2 skills that are often negatively affected as late sequelae of brain radiation; a nonsignificant decline in math calculation was noted, but overall, the patient's cognitive function remained stable relative to baseline.
This study presents the imaging, clinical, and cognitive outcomes for a child with recurrent JPA who received proton therapy after undergoing surgery and chemotherapy. The decrease in ADC within the GTV to clinically normal levels over time (Figure 4) indicates that the tissue in the tumor had become less “watery” and had increased membrane density and white matter cellularity. The corresponding increase in FA in the GTV over time (Figure 5) reflects growth of the neural microstructure, which could result from remyelination or increases in the number of axons or fiber coherence. Lesion enhancement was observed before treatment in the post-contrast T1-weighted image, and it eventually faded by the end of 25-month posttreatment period as demonstrated in Figure 3.
The unique physical characteristics of proton beams allow delivery of prescribed doses precisely to the target volume while sparing healthy tissues beyond the range of the proton beam. Although proton therapy does have some physical and biological uncertainties, it nevertheless seemed quite effective in the case described here. Eaton and Yock  recently reviewed the use of proton therapy for benign or low-grade pediatric brain tumors and confirmed that proton therapy generally provides better sparing to the critical structures important to cognitive development and other neurological functions, and minimization of the integral dose delivered reduces the risk of developing a second malignancy (relative to that from photon treatments).
To date, 2 other groups have reported clinical outcomes for children with low-grade gliomas. In 2002, Hug et al  at Loma Linda University Medical Center reported a series of 27 pediatric patients with low-grade astrocytoma (14 [52%] of whom had JPA) who received proton therapy with target doses between 50.4 and 63.0 Gy (RBE) (mean = 55.2 Gy [RBE]). This group reported that proton therapy was generally safe and effective for pediatric low-grade astrocytoma, with a local control rate of 78% and, at a mean follow-up time of 39 months, an overall survival rate of 85%. Of the 14 patients with JPA, 4 had local failure (3 hemispheric and central gliomas and 1 brainstem glioma). In the other study, Greenberger et al  at the Massachusetts General Hospital reported outcomes for 32 pediatric patients with low-grade glioma (19 [59%] with JPA) treated with proton therapy to a median dose of 52.2 Gy(RBE) (range 48.6–54 Gy[RBE]). This study had a considerably longer follow-up time; the progression-free survival rates were 89.7% at 6 years and 82.8% at 8 years, and the overall survival rate was 100%. Neither of these 2 studies included MRI analysis to quantify the responses of the tumor or normal tissues.
Khong et al  reported using DTI in an attempt to correlate losses of white matter anisotropy with declines in intelligence quotient in 30 patients with childhood cancers (medulloblastoma and acute lymphoblastic leukemia). They found that changes in FA correlated significantly with intelligence quotient and suggested that FA might be a clinically useful biomarker for assessing neurotoxicity after treatment for cancer. Others have linked FA and treatment-induced white matter injury as well [12, 14]. Farjam et al  used the diffusion abnormality index, which is derived from histograms of ADC, to assess tumor response after whole-brain radiation therapy. That group found that the utility of the diffusion abnormality index for this purpose depended on the treatment regimen used, and they recommended that DTI be used to assess responses of tumor and normal tissues after conventional X-ray–based radiation therapy.
In the current study, findings from neuropsychological and neurocognitive testing suggest that proton therapy did not have a lasting negative effect in this patient. The patient continued to perform in the average range on all assessed skills, and any low or below-average results, perhaps due to tumor effects, improved upon subsequent testing. However, these findings are limited to a single case and need to be confirmed in larger numbers of patients.
We conclude that proton therapy was effective for treating JPA in the patient described. In this proof-of-principle report, DTI was found to allow quantitative evaluation of tumor and normal brain tissue responses. Cognitive testing revealed stable cognitive function before and after treatment. Further study in additional patients is warranted.
ADDITIONAL INFORMATION AND DECLARATIONS
Conflicts of Interest: The authors have no conflicts of interest to disclose.