S610 I. J. Radiation Oncology d Biology d Physics Volume 75, Number 3, Supplement, 2009

Conclusions: In our experience, prone positioning on a ‘‘belly board’’ allows electromagnetic localization and tracking technologyto be used on all obese patients. Prostate motion from breathing is small and acceptable. As a result, high dose IMRT for prostatecancer can be delivered accurately and reliably to obese men.

Author Disclosure: M.D. Logsdon, None; J.K. Bareng, None; L. Olson, None; A. Ryan, None; S.W. Lee, None.

2956 Determining a Dosimetric Correlate for Acute Esophagitis in Patients with Advanced Non-small Cell Lung

Cancer Treated with Intensity Modulated Radiation Therapy (IMRT)

G. N. Gan1, D. E. Heron2,1, E. Brandner2, J. Flickinger2,1, R. Smith2, S. Huq2, S. Bahri2

1University of Pittsburgh School of Medicine, Pittsburgh, PA, 2University of Pittsburgh Cancer Institute, Department ofRadiation Oncology, Pittsburgh, PA

Purpose/Objective(s): Acute radiation esophagitis is a common complication in the treatment of lung malignancies which canlead to treatment interruption and undesirable morbidity. We report our experience treating locally-advanced non-small celllung cancers using chemoradiotherapy via IMRT and the associated acute treatment-related toxicities, and we determined potentialdosimetric parameters predictive for acute esophagitis.

Materials/Methods: This retrospective, multi-site study assessed 37 Stage IA-IV lung cancer patients who received IMRT. Allpatients were scanned during free breathing using coached retrospective 4D-CT and 21 had phase-based gated treatment delivery.A median dose of 70.0 Gy (range: 50 - 81.9 Gy) over 35 fractions (range 25 - 39 fractions) was delivered to the tumor using in-homogeneity correction. The esophagus was contoured and the mean dose and the volume of esophagus receiving 30, 50, 60 and 70Gy (V30, V50, V60, V70, respectively) was calculated for each patient. Patients were graded for acute and chronic toxicities using theCTCAE 3.0 and logistic regression was performed to test associations between dosimetric parameters and esophageal toxicity.

Results: All simulated patients completed their course of chemoradiotherapy treatment. Twenty-eight patients (76%) received ra-diation treatment with concurrent chemotherapy. The most common acute toxicities were nausea, esophagitis, and fatigue observedin 11, 26, and 18 patients respectively. Only 3 patients (8%) developed Grade 3 radiation pneumonitis confirmed clinical and ra-diographically and required treatment with oral steroids. Acute Grade 1, 2 and 3 nausea was observed in 5, 4 and 2 patients; acuteGrade 1, 2 and 3 esophagitis was observed in 8, 17 and 1 patients and acute Grade 1, 2 and 3 fatigue was observed in 9, 7 and 2patients, respectively. Patients who experienced any esophagitis had a mean dose in excess of 28.7 Gy to the esophagus (Odds ratio:1.09, 95% confidence interval: 1.02 - 1.18, p = 0.01). Univariate logistic regression correlated acute Grade $2 esophagitis with V60

(p = 0.045) and mean esophageal dose (p = 0.049). We did not identify significant differences in acute esophagitis between patientsreceiving concurrent chemotherapy and radiation versus those receiving radiation alone (p = 0.058). We were not able to correlateany parameters with late/chronic esophagitis.

Conclusions: The mean dose to the esophagus and the volume of esophagus receiving higher doses of irradiation, but not chemo-therapy, are associated with the development of esophagitis. Maintaining a mean doses to the esophagus below 28.7 Gy and re-stricting the volume of esophagus receiving 60 Gy or more can help to minimize treatment associated esophagitis.

Author Disclosure: G.N. Gan, None; D.E. Heron, None; E. Brandner, None; J. Flickinger, None; R. Smith, None; S. Huq, None; S.Bahri, None.

2957 The a/b Ratio: Dose-range Dependent or Model Limitation?

C. Zhang, N. A. Mayr, S. S. Lo, L. Lu, K. Li, J. Z. Wang

The Ohio State University, Columbus, OH

Purpose/Objective(s): The Linear-quadratic (LQ) model has been widely used in modeling the cell-killing by radiation therapy.However, its application for high dose irradiation has been challenged by many investigators. Recently a published study has dem-onstrated that the a/b ratio of the LQ model is dose-range dependent, especially when single-dose irradiation exceeds 10 Gy. Wehave previously proposed a generalized LQ model (gLQ) to address the high dose dilemma of the LQ model. In the current study,we employed both the standard LQ model and our gLQ model to analyze the in vitro cell irradiation data of large dose ranges andinvestigate the dose-range dependence of the a/b ratio.

Materials/Methods: In vitro data of three cell-lines (U373MG, CP3 and DU145) published by Garcia et al. (IJROBP 2007;67:587) were used. The cell-lines were irradiated with single doses up to 14 Gy in steps of 0.5 Gy. The gLQ model accountedfor the reduction of cellular sublethal damage due to re-irradiation, an important component of cell-killing in the high-dose domain.Both, the LQ and the gLQ were used to analyze the three datasets. The minimum c2 method was adopted to fit the data and toevaluate the figure of merit for the two models. Errors of the model parameters were determined by propagating the c2

min. Thea/b ratios from both models were investigated across different dose ranges from 0 Gy to varying final doses.

Results: The a/b ratio derived from the LQ model was dose-range dependent: when the final dose varied from 5 Gy up to 14 Gy, thea/b ratio increased from 2.8 to 9.3 Gy for U373MG, from 0.67 to 3.3 Gy for CP3, and from 3.8 to 19.6 Gy for DU145 cell-line.However, the a/b ratios obtained from the gLQ model were almost constant within the data uncertainties across the dose ranges withfinal dose $5 Gy: 1.0 Gy, 0.027 Gy, and 1.8 Gy for the three cell-lines, respectively. The variances of the a/b ratio generated fordifferent dose ranges with gLQ vs. LQ fitting were: 0.049 vs. 4.7 Gy2 for U373MG, 0.0015 vs. 0.67 Gy2 for CP3, and 0.30 vs. 22Gy2 for DU145. The gLQ model provided an improved fit over the LQ model across the entire dose range: the reduced c2

min of gLQvs. LQ model were 1.5 vs. 3.2 for U373MG, 1.5 vs. 2.7 for CP3, 4.4 vs. 7.2 for DU145. The standard error bars of a/b ratios obtainedfrom the gLQ fitting were much smaller than those of the LQ fitting, and they decreased more rapidly with increasing final dose.

Conclusions: Our results indicate that the dose-range dependence of the a/b ratio is due to the inconsistency of the standard LQmodel in high-dose domain. The gLQ model provides a more consistent interpretation of the cell irradiation data for large doseranges with much more stable model parameters. Therefore, the gLQ model is useful to extend our clinical experience accumulatedfrom conventional low-dose fractionation to high dose irradiation schedules, including SRS, SBRT, and HDR.

Author Disclosure: C. Zhang, None; N.A. Mayr, None; S.S. Lo, None; L. Lu, None; K. Li, None; J.Z. Wang, None.