S682 I. J. Radiation Oncology d Biology d Physics Volume 72, Number 1, Supplement, 2008

Conclusions: Using a genome-wide approach, Syne1 was identified as a candidate tumor suppressor gene undergoing progressivepromoter hypermethylation with increasing stage of breast cancer at presentation. Further studies will be conducted to elucidate themechanisms and function of Syne1, its role in tumor progression, and its utility as a biomarker in breast cancer.

Author Disclosure: A. Rimner, None; A.Y. Ho, None; N. Ahuja, None; K.E. Schuebel, None; S.B. Baylin, None; W.L. Gerald,None; T.A. Chan, None.

3140 TGF-beta1 Gene Polymorphism and Quality of Life in Prostate Cancer Patients Treated with

Brachytherapy Seed Implantation

A. Meyer1, T. Dork1, N. Bogdanova1, M. Brinkhaus1, B. Wiese1, M. Bremer1, R. Baumann1, J. H. Karstens1, S. Machtens2

1Hannover Medical School, Hannover, Germany, 2Marien-Krankenhaus Bergisch-Gladbach, Bergisch-Gladbach, Germany

Purpose/Objective(s): Transforming growth factor beta1 (TGF-beta1) can act as a tumor suppressor by inhibiting cellular pro-liferation and as a stimulator of tumor progression by promoting cellular differentiation in the early phase of cancer development.A polymorphic TGF-beta1 variant, Leu10Pro (L10P), has previously been implicated in prostate cancer risk and in radiation-in-duced side-effects. We investigated whether the prevalence of the TGF-beta1 L10P polymorphism is increased in a hospital-basedseries of patients treated with I-125 brachytherapy (permanent seed implantation) for early stage prostate cancer vs. controls andwhether carriers are at increased risk for treatment-related side effects.

Materials/Methods: A series of 445 consecutive patients treated for early-stage prostate cancer receiving definitive I-125 brachy-therapy between 10/2000 and 10/2007 at our institution and a comparison group of 457 male healthy controls were screened for thepresence of the TGF-beta1 L10P (869T.C) polymorphism. Outcome regarding morbidity was assessed prospectively and com-pared between carriers vs. non-carriers with the International Prostate Symptom Score (IPSS-15), a Quality-of-Life-index (QoL)and the International Index of Erectile Function (IIEF-15) with its subgroups (IIEF-5 and EF).

Results: The Leu/Leu genotype was found in 150 prostate cancer patients (34%) vs. 180 controls (39%), the Pro/Pro genotype in75 patients (17%) vs. 65 controls (14%) and the Leu/Pro genotype in 220 patients (49%) and 212 controls (46%). TGFbeta-1genotype frequencies were not statistically significant different between the two groups, although there was a trend towards anincreased prevalence of the L10P substitution among cases with a per allele odds ratio of 1.19 (95% CI 0.99 - 1.44; p = 0.08). Aftera mean follow-up of 18 months there were no statistically significant differences regarding IPSS (p = 0.27), QoL (p = 0.79), IIEF-15score (p = 0.98), IIEF-5 score (p = 0.85), and EF score (p = 0.51), respectively.

Conclusions: The TGF-beta1 polymorphism L10P is not strongly associated with prostate cancer risk. After 18 months, there wasno evidence for an increased adverse radiotherapy response in heterozygote or rare homozygote carriers. Further studies of can-didate gene variants for radiosensitivity will be needed and might have important clinical implications for prostate cancer treatment.

Author Disclosure: A. Meyer, None; T. Dork, None; N. Bogdanova, None; M. Brinkhaus, None; B. Wiese, None; M. Bremer,None; R. Baumann, None; J.H. Karstens, None; S. Machtens, None.

3141 Automatic Segmentation of Functional Images for Radiotherapy Treatment Planning

M. Hatt1, N. Boussion1,2, C. Roux1, O. Pradier2, D. Visvikis1

1INSERM U650 LaTIM, Brest, France, 2Institut de Cancerolgie, Brest, France

Purpose/Objective(s): Automatic volume delineation methodologies previously developed for PET have been essentially thresh-old-based. We have developed a robust and automatic segmentation algorithm based on a combination of statistical and fuzzy mod-eling that was previously shown to perform accurately, independent of image noise and contrast characteristics, on a phantom withspherical lesions and uniform activity distributions. In the present study we investigate its ability to delineate inhomogeneous andnon-spherical tumors increasing its potential for use in radiotherapy treatment planning.

Materials/Methods: We compared two approaches; namely the current state-of-the-art adaptive thresholding (two differentmethods) and our FLAB (Fuzzy Locally Adaptive Bayesian) method. Twenty tumors and their simulated counterparts (generatedin 3D using PET images of patient lesions as models) were used in the evaluation. Voxels misclassifications and volume errors werecomputed with respect to the ground-truth of the simulated tumors. Results, including variability, observed with the simulated data-sets were also compared to the results obtained on the real tumors images.

Results: The two threshold-based methods lead to large errors with 20.5% and 29.8% mean volume errors respectively. Significantvariability was observed in the results, with large dependence upon noise, contrast or lesion size (standard deviations 20.7 and 19.6,respectively). On the other hand FLAB lead to the best results with low errors (8.5% mean volume error, standard deviation 10%)and good accuracy in volume delineation even when dealing with heterogeneous activity distribution and in the presence ofnecrotic lesions.

Conclusions: FLAB outperformed current state of the art methodologies for the complex task of delineating realistic, non sphericaltumors. Furthermore it is able to accurately deal with non-uniform activity distributions: it can identify regions of interest withvariable activity concentrations within the external contour of tumors providing a solution for future applications such as ‘‘dosepainting’’ in radiotherapy treatment planning.

Author Disclosure: M. Hatt, None; N. Boussion, None; C. Roux, None; O. Pradier, None; D. Visvikis, None.

3142 In Vitro 3D Modeling of Host Organ-specific Whole Tumor Responses to Radiation

A. Volgin, L. LeRoux, D. Maxwell, D. Schellingerhout, A. Thitai-Kumar, Y. He, J. Gelovani, D. L. Schwartz

UT M.D. Anderson Cancer Center, Houston, TX

Purpose/Objective(s): To model organ-specific tumor responses to radiation in vitro using multicellular tumor spheroids contain-ing stromal cells derived from different host tissue sites.

Proceedings of the 50th Annual ASTRO Meeting S683

Materials/Methods: We used organ-derived microvascular endothelial cells (MVEC) isolated from skin and lung from trans-genic H-2Kb-tsA58 mice whose tissues harbor temperature-sensitive SV40 large T antigen. These MVEC lines stably retain theirphenotype with serial passaging at 33oC and maintain morphologic characteristics specific to their site of origin at 37oC. Lungand skin derived MVEC were matched with isogenic fibroblasts isolated from lung or skin, and then admixed with LLC1 murinelung cancer cells stably expressing a HRE-eGFP-CMVdsRed reporter for localization and detection of HIF-1 hypoxia signaling.MVEC were labeled with CellTracker Blue CMAC. Spheroids were composed of 1x104 LLC1 cells, 1x104 organ-specificfibroblasts, and 1x103 organ-specific MVEC. Spheroids were treated with 0, 5 or 10 Gy 24 hours after plating. Photomicrographswere taken over the next 48 hours via an Olympus IX81/FV1000 Fluoview confocal system at a fixed depth of 200 mm nearmidplane of spheroids. Automated image analysis software was developed to perform multi-channel quantification of fluores-cence intensity as a function of depth along spheroid radius. MVEC survival 48 hours post-radiation was quantified as BlueCMAC signal/mm2. MVEC signal was localized relative to LLC1 HIF-1 signaling in the central and peripheral portions ofspheroids.

Results: Radioresistant LLC1 growth and MVEC survival was seen in site-of-origin (lung) spheroids relative to spheroidscontaining stroma derived from mismatched host tissue (skin). Surviving MVEC geographically co-localized with LLC1 HIF-1signaling following treatment. Corroborative immunohistochemical characterization of spheroids is currently ongoing and willbe presented.

Conclusions: We demonstrate reproducible fabrication of lung and skin derived organotypic spheroids which recapitulate differ-ential tumor-stromal cell interactions specific to host organ sites. Hypoxic pro-angiogenic tumor cell signaling plays a critical rolein determining stromal responses and tumor resistance to radiation across host locations. This in vitro 3D spheroid reporter systemwill facilitate study of these mechanisms and allow for high-content screening of novel radiosensitizing drugs.

Author Disclosure: A. Volgin, None; L. LeRoux, None; D. Maxwell, None; D. Schellingerhout, None; A. Thitai-Kumar, None; Y.He, None; J. Gelovani, None; D.L. Schwartz, None.

3143 Magnetic Resonance Spectroscopy and Definition of Cancer Rich Targets in the Prostate

G. H. Goldin1, C. Vargas2, M. Bridges3

1University of Miami Miller School of Medicine, Miami, FL, 2University of Florida, Jacksonville, FL, 3Mayo Clinic, Jacksonville,FL

Purpose/Objective(s): The efficiency of radiation oncology lies in our ability to target malignant tissue and spare normal tissue.The purpose of an imaging tool for radiation oncology planning is different than its use for diagnosis. Reducing target boost vol-umes is related to the negative predictive value of a test and the spatial resolution for a given cancer volume. Magnetic resonancespectroscopy (MRS), an imaging modality which employs the ability to detect the altered concentrations of citrate, choline, andcreatine present in prostate cancer, has proved more informative than traditional MRI. The value of MRS to define cancer richtargets with high spatial resolution and percent of involvement is not clear.

Materials/Methods: Two hundred twenty-eight different zones were evaluated by means of MRS and transrectal ultrasoundguided biopsies in 19 intermediate or high risk prostate cancer patients. The MRS 3D information was acquired based on3mm voxels prior to the biopsies. Biopsies were then done within 48 hours of the MRS study based on a 12 zone grid. Thelocation and percentage invasion for each core was determined as well as the Gleason score. The location was then determinedand evaluated independently based on the MRS information. Cancer targets were defined by the percent involvement for eachcore.

Results: A total of 35.1% (80/228) positive cores were found during the 12 zone prostate biopsies. For 12 zones the Sensitivity (ST)was 65.4%, specificity (SP) 76.2%, positive predictive value (PPV) .6, and negative predictive value (NPV) 0.8. The ability of theMRS to detect cancer volumes $1%, $30%, $50%, $70% was 65% (52/80), 65.3% (31/48), 62.5% (20/32), and 70% (7/10),respectively. However, a negative MRS had a NPV for cancer targets with an involvement $30%, $50%, $70% of 88%,91%, and 98%, respectively. MRS could have decreased the number of zones treated by 61.5% (88/228). MRS and positivebiopsies could have decreased the number of zones treated by 49.1% (116/228).

Conclusions: MRS was shown in our study, and in previous studies, to aid in the identification of cancer-rich regions in theprostate. The identification of prostate cancer is becoming increasingly important in cancer treatment as more targeted ther-apies are emerging, but also remains vital for staging and evaluating for eradication of disease. MRS can be employed todefine target volumes with a low cancer burden, and to define prostate boost targets with a high spatial resolution. Moreextensive research needs to be done to evaluate the utility of MRS in assessing the biological aggressiveness of a malignancy.This could help provide stage-specific treatment guidelines and thereby lessen the high economical burden that the diseasehas created for our healthcare system. Such information could also prove invaluable to the patient by offering more accurateprognosis.

Author Disclosure: G.H. Goldin, None; C. Vargas, None; M. Bridges, None.

3144 In Vivo Detection of Tumor Response to Radiotherapy and New Radiosensitization Agent using

Quantitative Non-invasive High Frequency Ultrasound

G. J. Czarnota1, J. Lee1, A. Giles2, R. Karshafian2, B. Banihashemi3

1Dept of Radiation Oncology, Sunnybrook Health Sciences Centre, Toronto, ON, Canada, 2Imaging Research, MedicalBiophysics, University of Toronto, Toronto, ON, Canada, 3Dept of Radiation Oncology, University of Toronto, Toronto, ON,Canada

Background: Imaging of acute tumor response during cancer therapy may help guide optimization and individualization of treat-ment regimens. Morphologic changes associated with apoptosis such as cell shrinkage, membrane blebbing and nuclear conden-sation/fragmentation can change the characteristics of high frequency ultrasound backscatter signal.