S602 I. J. Radiation Oncology d Biology d Physics Volume 69, Number 3, Supplement, 2007

(0.5 cm)3 voxel using interpolated cell survival curves for CHOA88 and U373MG (from Garcia LM et al, 2007) or the L-Q modelwith a = 0.3 Gy�1 and b = 0.03 Gy�2. The cell survival data were chosen because they exhibit an initial decrease in SF with in-creasing dose, followed by a small increase in SF around 100 cGy–mimicking the hyper-radiosensitivity response–then an expo-nential decrease. The volume-average biologic effect (VBE) for the overall 35-cm cube, central 5-cm cubic core and the peripheralvolume (overall cube less core) was then calculated from the SF for each voxel. Thus, smaller values of VBE represent a greaterchange in the cell population; a VBE of 1 signifes no change.

Results: While the volume-average central and peripheral doses were invariant with the number of beams for each course of RT,the VBE for the central core increased with an increasing number of beam pairs (Table 1). However, the number of beams hadminimal impact on the peripheral VBE for both the conventional and hypofractionated courses and for all dose-biologic effectcurves. The minimal differences in peripheral VBE between the 4- and 6-field plans were further reduced when the dose at isocenterwas adjusted to obtain the same central VBE.

Conclusions: Purely increasing the number of beams has minimal impact on the average biologic effect in the ‘‘normal tissue’’periphery in this model system, despite the non-linearity of the various dose response curves.

Impact of Number of Orthogonal Beam Pairs on Volume-Average Biologic Effect (VBE)

Thirty 2-Gy Fractions Five 8-Gy Fractions

Dose-Effect Model

2 Field 4 Field 6 Field 2 Field 4 Field 6 Field

U373MG

Central VBE

9.53 � 10�4 6.17 � 10�4 5.02 � 10�4 3.21 � 10�3 1.71 � 10�3 1.24 � 10�3

Peripheral VBE

0.885 0.894 0.880 0.905 0.915 0.903

CHOA88

Central VBE

7.47 � 10�6 9.77 � 10�5 7.33 � 10�5 1.29 � 10�3 1.09 � 10�3 1.01 � 10�3

Peripheral VBE

0.800 0.804 0.789 0.838 0.844 0.831

LQ Model

Central VBE

5.85 � 10�5 1.90 � 10�5 1.39 � 10�5 2.06 � 10�4 7.54 � 10�5 5.36 � 10�5

Peripheral VBE

0.863 0.868 0.854 0.888 0.895 0.881

U373MG

Central VBE

6.17 � 10�4 6.17 � 10�4 6.17 � 10�4 1.71 � 10�3 1.71 � 10�3 1.71 � 10�3

(Uniform Central VBE)

Peripheral VBE

0.880 0.894 0.884 0.901 0.915 0.906

Author Disclosure: J.P. Kirkpatrick, Varian Medical Systems-supported research, B. Research Grant; Z. Wang, Varian MedicalSystems-supported research, B. Research Grant.

2726 Modeling of Normal Tissue Complication Probability in Liver Irradiation

A. Tai, L. Grossheim, B. Erickson, A. X. Li

Medical College of Wisconsin, Milwaukee, WI

Purpose/Objective(s): The ability to predict normal tissue complication probability (NTCP) is essential for NTCP-based treat-ment planning. The purpose of this work is to estimate the Lyman model parameters based on published clinical data collectedfrom several fractionation regimens.

Materials/Methods: The NTCP data of radiation induced liver disease (RILD) using radiation therapy for hepatocellular carci-noma (HCC) were selected to analyze. The data were collected from 5 institutions for tumor sizes in the range of of 8–10 cmand for patients with liver cirrhosis of Child-Pugh grade A. The dose per fraction ranged from 1.5 Gy to 6 Gy. A representativeDVH with the parameter Veff = 43% was used in the data fitting. The biologically equivalent dose (BED) was calculated byD*(1+d/a/b+f*N), where D, d and N are prescription dose, dose per fraction and total number of fraction, a and b are linear-qua-dratic (LQ) model parameters, and f is a fitting parameter. This BED expression can be understood by a modified LQ model withtwo components corresponding to radiosensitive and radioresistant functional subunits in the normal liver tissue. Using the pres-ently obtained parameters, we considered the data from both external beams and Y-90 microsphere brachytherapy based on theconcept of equivalent uniform dose (EUD).

Results: The fitting result is showed in the Figure below. There are 5 parameters in the model: TD50, m, n, a/b and f. In the Figure,BED has been normalized to the regimen used by the Michigan group (61.5 Gy at 1.5 Gy per fraction). Because there are too fewclinical data points, we fixed two parameters n and a/b to be 1.0 and 2.0 Gy, respectively. The extracted parameters from the fittingare TD50 = 40.6 ± 7.7 Gy, m = 0.36 ± 0.08, f = 0.162 ± 0.048. For Y-90 microsphere therapy, BED is calculated by replacing N =(trep+64)/trep, where trep is the sublethal damage repair half-time and 64 is the half-life of Y-90 decay in hour. With a typicalDVH of normal liver used in the literature, it was found that EUD should be smaller than 30 Gy, for a reasonable trep .1.5 hours,for the Y-90 microsphere therapy. This EUD value is below the threshold for RILD, explaining why RILD was not observed formicrosphere therapy.

Conclusions: A new expression of LQ/BED is proposed to consider different fractionations and/or modality for analyzing NTCPof liver irradiation. This approach allows to reasonably fit clinical NTCP data from various treatment schemes with the Lymanmodel, and also provides an explanation why no RILD was observed in Y-90 therapy. The Lyman parameters generated presentlymay be used to predict NTCP for treatment planning of innovative liver irradiation, such as stereotactic body radiation therapy(SBRT).

Proceedings of the 49th Annual ASTRO Meeting S603

Author Disclosure: A. Tai, None; L. Grossheim, None; B. Erickson, None; A.X. Li, None.

2727 Recombinant Human Keratinocyte Growth Factor Modulates Inflammatory Changes in Mouse Oral

Mucosa During Fractionated Irradiation

C. Richter, J. Jaal, M. Kuschel, W. Doerr

University of Technology Dresden, Dresden, Germany

Purpose/Objective(s): Oral mucositis is a severe and frequently dose-limiting side effect of radio(chemo)therapy for head andneck tumours. Keratinocyte Growth Factor, synthesized mainly by mesenchymal cells, predominantly modulates epithelial prolif-eration and differentiation processes. A significant reduction of radiation-induced oral mucosal reactions by recombinant humanKeratinocyte Growth Factor (Palifermin) has been demonstrated in a number of experimental studies and in clinical investigations.The underlying mechanisms, however, currently remain unclear. Aim of the present study was to quantify the impact of Paliferminon radiation-induced inflammatory changes during daily fractionated irradiation in mouse tongue.

Materials/Methods: Daily fractionated irradiation, 10 � 3 Gy/2 weeks, was given to the snouts of the animals. A single subcu-taneous injection of Palifermin at a dose of 15 mg/kg was administered on day �1, before the first radiation fraction at day 0; thisschedule has previously been shown to significantly reduce the response of the mucosa to irradiation. Groups of 3 mice per daywere sacrificed from day 0 to 16. The immunohistochemical staining intensity of Intercellular Adhesion Molecule-1 (ICAM-1)–aprotein associated with inflammatory changes–was determined in the endothelium of tongue blood vessels using an arbitrary,semiquantitative score. Moreover, vasodilatation, as determined by the relative area in the histological sections covered by thelumen of blood vessels (denoted as vascular area) was determined in five randomly chosen microscopic fields per tongue.

Results: Fractionated irradiation resulted in a significant increase in the ICAM-1 staining signal in tongue blood vessels during thewhole study period. These changes were observed in small, subepithelial vessels as well as in bigger blood vessels within thetongue muscle tissue. Moreover, significant vasodilatation was found in irradiated tongues. A single, pre-irradiation administrationof Palifermin largely prevented these inflammation-associated changes in endothelial ICAM-1 expression and vascular area.

Conclusions: A single administration of Palifermin before (day�1) the onset of fractionated irradiation resulted in a pronounced,long-lasting inhibition of inflammatory changes in mouse oral mucosa. This investigation demonstrates a novel mechanism of Pal-ifermin for the amelioration of radiation effects in normal tissues, independent of the modulation of epithelial cell proliferation anddifferentiation.

Author Disclosure: C. Richter, None; J. Jaal, None; M. Kuschel, None; W. Doerr, AMGEN Inc., Thousand Oaks, CA, U.S.A.,C. Other Research Support.

2728 The In Vitro and In Vivo Radiosensitization Effect of Cloforabine

B. Xu, M. Cariveau, M. Stackhouse, X. Cui, W. Waud, W. Parker, J. Secrist

Southern Research Institute, Birmingham, AL

Combination of radiotherapy and chemotherapy has emerged as the dominant form of cancer adjuvant treatment in recent years.Several drugs, such as cisplatin, 5-fluorouracil (5-FU) and gemcitabine have been shown to increase tumor sensitivity to radiother-apy. Clofarabine, a newly approved drug for the treatment of pediatric leukemia, is a second-generation nucleoside analogue thatmay also have the radiosensitizing potential. The active form of clofarabine, clofarabine 50-triphosphate, a potent inhibitor of DNApolymerase -a, -e, and ribonucleotide reductase, can block DNA synthesis and possibly inhibit DNA repair. Therefore, we hypoth-esized that clofarabine could work synergistically with radiotherapy to increase tumor cell responses to ionizing radiation (IR). Wereport here that, low doses of clofarabine can prolong the existence of radiation-induced g-H2AX nuclear focus formation, whilehigh doses of clofarabine can induce DNA double strand breaks, suggesting that clofarabine can interfere with DNA damage path-ways. A significant decrease in clonogenic survival was observed in irradiated cells treated with clofarabine, demonstrating a strong