International Journal of Radiation Oncology � Biology � PhysicsS720

ranks will receive bonuses. In this study, we present an analysis of factors

that predict for PG scores.

Materials/Methods: PG scores were analyzed over a time span of 7

months including 71 patients and 1409 survey questions for an academic

institution with two satellite centers. Multiple variables (i.e., age, perfor-

mance status, diagnosis, treatment venue, presence of pain, concurrent

chemotherapy) were analyzed for correlations with scores using the In-

dependent T Test, Pearson Chi Square, and Fisher Exact Test.

Results: 26 patients (36%) filled out all questions with scores of all 5, 16

patients (22%) had all 5s but incomplete survey, 39 patients (55%) had at

least one non-5 score. 4 patients had a single non-5 score while 35 patients

had multiple non-5 scores. For complete surveys, the median score was

104 (range 84 to 105), the average score was 102.5. Pain, increasing age,

and treatment venue were predictive for non-5 scores.

Conclusions: PG scores are becoming a vital metric as department reim-

bursement is tied into these scores. Patients with unsatisfactory scores

tended to give non-5 scores across multiple questions, patients are seldom

dissatisfied in a single area. We found several modifiable (treatment venue)

and non-modifiable (older age and presence of pain) factors related to poor

scores. While there is some potential methods for improving PG scores

there are limitations, departments that see a disproportionate number of

elderly palliative cases will likely have lower scores.

Author Disclosure: A.B. Patel: None. T. LaCouture: None. K. Hunter:

None. A. Tartaglia: None. G.J. Kubicek: None.

3294Neurocognitive Outcomes in Relationship to Hippocampal andBrain Doses After Partial Brain Proton Radiation Therapy inChildrenA. Mahajan,1,2 D. Grosshans,1 D. Ris,2 M. Chintagumpala,2 F. Okcu,2

M. McAleer,1 B. Moore,1 H. Stancel,2 C. Minard,2 D. Guffey,2

and L. Kahalley2; 1M.D. Anderson Cancer Center, Houston, TX, 2Baylor

College of Medicine, Houston, TX

Purpose/Objective(s): We evaluated the relationship between IQ and

hippocampal RT doses in pediatric patients with brain tumors treated with

proton radiation therapy (PRT).

Materials/Methods: IQ scores were abstracted for patients treated with

partial brain PRT who underwent serial neurocognitive testing. Hippo-

campal delineation was performed according to RTOG guidelines. The

PRT dosimetric data were compiled. The relationships between IQ and 1)

median hippocampal doses and 2) percentage of hippocampus receiving

50, 30 and 10 GyRBE were evaluated using general linear mixed models

while controlling for mean brain-GTV dose, CTV volume, tumor location,

and age-at-RT.

Results: Data were collected for 25 patients (13 males, 12 females)

receiving PRT from 2007-2012. The mean interval between first-last IQ

testing was 2.1 y (range 0.9-6.0). The median age-at-PRT was 9.4 y (range

1.7-15.4 y). Tumors were supratentorial in 17 patients (9 glial, 5 cranio-

pharyngioma, 2 germinoma, 1 meningeal tumors) and infratentorial in 7

patients (3 anaplastic ependymoma, 2 medulloblastoma, 2 glial tumor). 17,

2 and 6 tumors were midline, right and left, respectively. The median dose

was 50.4 GyRBE (range 45.0-60.0 GyRBE). The median CTV volume was

43.3 cc (range 12.3-234.4 cc). The median mean brain-GTV dose was 9.4

GyRBE (range 3.9-19.7 GyRBE). The median mean right and left hip-

pocampal doses were 17.0 and 16.3 GyRBE (range 0.2-54.8 and 0.0-46.2

GyRBE), respectively. After controlling for the factors mentioned above,

patients who received any right hippocampal dose >50 GyRBE exhibited

significant IQ decline (6.4 points/y, p Z 0.01). Patients who received 30

GyRBE to �10% of the right hippocampal volume also experienced sig-

nificant IQ decline (4.0 points/y, p<0.01), while patients with less right

hippocampal dose/volume maintained stable IQ over time (p Z 0.67). Left

hippocampal dose/volume was not associated with IQ decline in this

cohort.

Conclusions: Preliminary findings suggest right hippocampal dose is

associated with steeper IQ decline post-PRT in this cohort of patients.

Author Disclosure: A. Mahajan: None. D. Grosshans: None. D. Ris:

None. M. Chintagumpala: None. F. Okcu: None. M. McAleer: None. B.

Moore: None. H. Stancel: None. C. Minard: None. D. Guffey: None. L.

Kahalley: None.

3295Patterns of Failure Following Proton Therapy in Medulloblastoma:LET Distributions and RBE Associations for RelapsesR. Sethi,1 D. Giantsoudi,2 M. Raiford,2 I. Malhi,2 A. Niemierko,2

O. Rapalino,3 P. Caruso,1 T.I. Yock,2 N.J. Tarbell,2 H. Paganetti,2

and S. MacDonald2; 1Harvard Medical School, Boston, MA,2Massachusetts General Hospital, Boston, MA, 3Massachussetts General

Hospital, Boston, MA

Purpose/Objective(s): The pattern of failure in medulloblastoma patients

treated with proton radiation therapy is unknown. For this increasingly

used modality, it is important to ensure that outcomes are comparable to

modern photon series. It has been suggested this pattern may differ from

photons due to variations in linear energy transfer (LET) and relative

biological effectiveness (RBE). In addition, the use of matching fields for

delivery of craniospinal (CSI) radiation may influence patterns of relapse.

We seek to report the patterns of failure following protons, compare to the

available photon literature, and determine the LET and RBE values in

areas of recurrence.

Materials/Methods: Retrospective review of patients with medulloblas-

toma treated with proton radiation therapy at *** between 2002 and 2011.

We documented the locations of first relapse. Discrete failures were con-

toured on the original planning computed tomography scan. Monte Carlo

calculation methods were used to estimate the proton LET distribution.

Models were used to estimate RBE values based on the LET distributions.

Results: One hundred and nine patients were followed for a median of

38.8 months (range, 1.4 to 119.2). Sixteen (16) patients experienced

relapse. Relapse involved the supratentorial compartment (n Z 8), spinal

compartment (n Z 11) and posterior fossa (n Z 5). Eleven failures were

isolated to a single compartment; six failures in the spine, four failures in

the supratentorium and one failure in the posterior fossa. The remaining

patients had multiple sites of disease. One isolated spinal failure occurred

at the spinal junction of two fields. None of the 70 patients treated with an

involved-field-only boost failed in the posterior fossa outside of the tumor

bed. We found no correlation between Monte Carlo-calculated LET dis-

tribution and regions of recurrence.

Conclusions: The most common site of failure in patients treated with

protons for medulloblastoma was outside of the posterior fossa. The most

common site for isolated local failure was the spine. We recommend

consideration of spinal imaging in follow-up and careful attention to dose

distribution in the spinal junction regions. Development of techniques that

do not require field matching may be of benefit. We did not identify a

direct correlation between lower LET values and recurrence in medullo-

blastoma patients treated with proton therapy. Patterns of failure do not

differ from patients treated with photon therapy.

Author Disclosure: R. Sethi: None. D. Giantsoudi: None. M. Raiford:

None. I. Malhi: None. A. Niemierko: None. O. Rapalino: None. P.

Caruso: None. T.I. Yock: None. N.J. Tarbell: K. Advisory Board; Spouse

is on medical advisory board of ProCure. NJT was on medical advisory

board until 2008. H. Paganetti: None. S. MacDonald: None.

3296Impact of Molecular Subtype and Craniospinal Irradiation (CSI)Dose on Relapse of MedulloblastomaA. Wang,1 S. Partap,1 K. Yeom1, M. Martinez,2 H. Vogel,1

S. S. Donaldson,1 P. G. Fisher,1 S. Perreault,3 Y.-J. Cho,1 and I. C. Gibbs1;1Stanford University/Lucile Packard Children’s Hospital, Stanford, CA,

Volume 90 � Number 1S � Supplement 2014 Poster Viewing Abstracts S721

2Temple University, Philadelphia, PA, 3University of Montreal-CHU

Sainte-Justine, Montreal, QC, Canada

Purpose/Objective(s): To evaluate the impact of craniospinal irradia-

tion (CSI) dose and molecular sub-type on relapse rates and patterns

of relapse of medulloblastoma (MB) managed based on clinic-patho-

logical factors.

Materials/Methods: One hundred fifteen patients with medulloblastoma

were treated at our institution with CSI to doses of 18 Gy to �36 Gy

determined by known clinic-pathological factors. Molecular subtyping

data was done by nanostring assay (WNT, SHH, and non-WNT/non-

SHH (NWNS)) and were available for 55 patient tumors. After

exclusion of cases with incomplete data and infants treated without

CSI, a cohort of 36 patients (24 classic, 7 large cell/anaplastic (LC/A),

3 desmoplastic, 1 extensive nodularity (EN), and 1 other) remained as

the final cohort for the present analysis. Patients were followed clin-

ically and radiographically at 1-12 monthly intervals. A CSI dose

schema based on prognostic implications of molecular subtype was

devised. Accordingly, the proposed CSI dose: �36 Gy for M+ and

Group 3; �18 Gy for non-disseminated WNT; 23.4 Gy all others. CSI

dose was categorized as higher, lower, or similar to the proposed CSI

schema. Relapse rates and patterns were assessed and correlated with

CSI dose.

Results: The final cohort included 36 subjects with median age 8.7

years (3.3 e 16.9 years) and median follow-up of 4.1 years (0.2 -13.7

years). Molecular subtypes were as follows: WNT (2), SHH (6),

NWNS (28). CSI dose was 18 Gy (11), 23.4 Gy (9), and �36 Gy

(16). All but 2 also received chemotherapy. CSI dose was divergent

from the proposed dose schema in 18 cases (14-lower dose, 4-higher

dose) and similar in 18. There were 14 relapses (8- classic, 5-LC/A,

1-EN) with median time to relapse of 18.5 months (0.43-62 months).

Relapse by molecular subtype: WNT (0), SHH (1), NWNS (13).

Seven of 14 tumors treated to a lower dose relapsed (1 SHH; 6

NWNS); only 1 (NWNS) of 4 patients with higher divergent dose

relapsed. Six of the 18 patients treated to similar dose relapsed, all

were NWNS (group 3); 4 of 6 had M3 disease. Patterns of failure:

local (9), spinal (11), brain (9), local, brain and spine (5). There were

13 deaths. Eleven of 14 patients who relapsed died; 3 patients are

alive 0.59, 14, 28 months after relapse, respectively. Two patients

developed sarcoma and died without relapse of MB. Three illustrative

cases will be presented: 1) non-disseminated WNT tumor treated to

36 Gy; 2) non-disseminated NWNS treated to 18 Gy; 3) non-

disseminated SHH treated to 18 Gy.

Conclusions: Molecular subtyping may help to avoid under-treating some

NWNS tumors but innovative strategies for metastatic NWNS tumors are

needed. We confirm that histological subtypes may still hold some pre-

dictive value as 5 of 7 LC/A relapsed and no desmoplastic tumor relapsed.

Although the sample of WNT tumors was small, no relapse occurred and

only 1 SHH relapsed. Thus, de-escalation, <18 Gy, for non-metastatic

WNT and SHH tumors should be explored cautiously.

Author Disclosure: I.C. Gibbs: None. S. Partap: None. K. Yeom: None.

M. Martinez: None. H. Vogel: None. S.S. Donaldson: None. P.G.

Fisher: None. S. Perreault: None. Y.J. Cho: None.

3297Incidence of Optic Nerve Involvement in Pediatric IntracranialGerm Cell TumorsJ.Y. Chin,1 R.V. Sethi,2 C.P. Goebel,3 O. Rapalino,3 N.J. Tarbell,3

T.I. Yock,3 and S.M. MacDonald3; 1Harvard Radiation Oncology

Program, Boston, MA, 2Harvard Medical School, Boston, MA,3Massachusetts General Hospital, Boston, MA

Purpose/Objective(s): Optic nerve (ON) involvement for intracranial

germ cell tumors (GCT) is rare and only reported in the literature in case

reports. ON involvement has important implications for radiation (RT)

planning as this region may not be included in craniospinal (CSI) and

whole ventricle (WV) volumes. In addition, brain MRIs may not optimally

image the ON and orbital MRIs, if obtained at diagnosis, may better

characterize ON involvement. We investigated the incidence of ON

involvement in patients with intracranial GCTs in a large cohort of

patients.

Materials/Methods: All patients with intracranial pure germinoma and

nongerminomatous GCTs treated at our institution with proton radiation

therapy between 1998 and 2013 were included.

Results: We identified 78 patients with intracranial GCTs treated with

proton radiation therapy at our institution: 45 pure germinoma and 33

nongerminomatous GCT. Of these patients, 4 (5.1%) had ON

involvement based on pre-treatment MRI. Of the 4 patients with ON

involvement at diagnosis, 3 had localized suprasellar tumors and 1 had

disseminated disease. Median age at diagnosis was 16 years, and 3/4

patients were male. The histology was pure germinoma in 3/4 patients,

and NGGCT in one patient. All were treated with induction chemo-

therapy for 3-6 cycles with near complete (n Z 2/4) to complete (n Z2/4) response, followed by proton radiation therapy with craniospinal

irradiation (n Z 3/4) or whole-ventricular irradiation (n Z 1/4) and

involved-field radiation therapy.

Conclusions: Optic nerve involvement meaningfully impacts radiation

field planning in the treatment of intracranial GCTs. While relatively rare

overall, it may be more common than previously recognized. It is

important to look for ON involvement and adjust fields appropriately to

ensure full radiation dose to this area of disease involvement.

Author Disclosure: J.Y. Chin: None. R.V. Sethi: None. C.P. Goebel:

None. O. Rapalino: None. N.J. Tarbell: K. Advisory Board; ProCure

(spouse). T.I. Yock: None. S.M. MacDonald: None.

3298Diffusion Tensor Imaging (DTI) and Proton MR Spectroscopy (MRS)Detected Region Specific Injury to White Matter in Children TreatedWith Cranial IrradiationK.J. Redmond,1 E.M. Mahone,2 H. Sair,1 T. Vannorsdall,1 S.A. Terezakis,1

L.R. Kleinberg,3 T.R. McNutt,1 M. Wharam,1 and A. Horska1; 1Johns

Hopkins University, Baltimore, MD, 2Kennedy Krieger Institute,

Baltimore, MD, 3Johsn Hopkins University, Baltimore, MD

Purpose/Objective(s): Diffusion tensor imaging (DTI) is a method to

characterize micro-structural changes due to neuro-pathology or treatment.

The purpose of this study was to examine white matter integrity as

measured by DTI and proton MR spectroscopy (MRS) and their associa-

tions with region specific radiation dose and neuropsychological func-

tioning in children treated with cranial irradiation.

Materials/Methods: A total of 20 patients and 34 age- and sex-matched

controls participated. MRS and DTI at 1.5T and neuropsychological

assessment were conducted at baseline, 6, 15, 27 months (times 1-4)

following RT and compared to a single visit in controls. The neuro-psy-

chological battery included: motor dexterity, visual attention/processing

speed, visual and verbal working memory. White matter regions including

the midbrain, frontal lobes, temporal lobes, and genu/splenium of corpus

callosum were contoured. For each region, apparent diffusion coefficient

(ADC) and fractional anisotropy (FA) were recorded. Concentrations of

neuro-metabolites were expressed as ratios N-acetylaspartate (NAA)/cre-

atine (Cr) and choline (Cho)/Cr. Linear mixed effects (LME) regression

models were used to examine associations among ADC, FA, metabolite

ratios, RT dose to contoured structures, and performance on testing, and to

compare findings for patients versus controls.

Results: For patients: Male, n Z 14. Primary site: infratentorial, n Z 7;

supratentorial, n Z 11; leukemia, n Z 2. Mean age at RT: 11.9 years

(range 5.4-18.9). Mean prescription dose: 44.2 Gy (range 2-59.4 Gy). LME

analyses across all regions demonstrated significant differences between

patients and controls in NAA/Cr and ADC, with the greatest difference at

time 4 (p Z <0.01 and <0.0001, respectively). At time 4, the ADC