Proceedings of the 50th Annual ASTRO Meeting S539

Materials/Methods: Twenty-three patients received 5FU-based chemotherapy and concomitant radiotherapy for primary colorec-tal or anal canal malignancies. Thirteen patients received supine hypofractionated pelvic or pelvic-inguinal IMRT with simulta-neous integrated tumor boost to 4,950-5,500 cGy in 22-25 fractions. Ten patients received prone 3D conformal external beamradiotherapy to the whole pelvis, followed by a tumor boost for cumulative doses of 5,040-5,580 cGy in 28-31 fractions. All pa-tients were localized either daily or twice weekly using mega-voltage CBCT. The couch shifts for the supine and prone patientswere recorded and compared. Grades of diarrhea were recorded according to NCI CTCAE v3.0, and compared against the volumeof SB receiving doses greater than defined levels, from 5 to 55 Gy in 5 Gy steps (V5-V55).

Results: The shift distributions for the prone and supine patients did not suggest any difference in terms of setup uncertainty.However, random errors in the order of 7-8 mm, and up to 1.5 cm for some patients, emphasize the need for daily localizationfor both setups. The average ratio of mean SB V5-V30 values for patients in the supine to prone position was 2.6 (range,1.4-3.5, p \ 0.05). However at higher doses, V40-V55, the average ratio was 1.0 (range, 0.01-2.5). Four prone patients(40%), and 7 supine patients (54%) developed diarrhea of grade 2 or 3 (p = 0.3). No patients experienced diarrhea greaterthan grade 3.

Conclusions: Data show that while the prone position provides better SB volume sparing at low doses (\35 Gy) in 3D-CRT, thesupine hypofractionated IMRT approach offers similar volume sparing for doses greater than 40 Gy. The absence of a statisticallysignificant difference in diarrhea incidence suggests that high SB doses (.40 Gy) are linked to acute toxicity, in agreement withrecent studies showing V45 as the best predictor. These initial results suggest that hypofractionated IMRT in supine position isa reasonable alternative clinical approach for rectal and anal canal cancer. Due to significant interfractional set-up uncertainty, dailyimage-guidance is recommended.

Author Disclosure: O. Gayou, None; B. Karlovits, None; M. Miften, None; M. Wong, None; A. Kirichenko, None; B. Leicher,None; D. Medich, None; D. Parda, None.

2828 Dose-delivery Confirmation of SBRTof the Head of Pancreas Cancer, using a Non-rigid Image Registration

B. Liu, M. Garofalo, F. Lerma

University of Maryland Medical Center, Baltimore, MD

Purpose/Objective(s): Evaluate the total dose delivered to head of pancreas cancer tumors under image-guided stereotacticbody radiation therapy (SBRT), using a deformable image registration of cone beam CT (CBCT) to planning fan beam CT(FBCT).

Materials/Methods: Five Pancreatic cancer patients were treated by SBRT under image guidance. The prescription dose is 30 Gyin 6 fractions. A CTV is constructed by the union of CTVs drawn from 4D-CT at 10 phases; and two PTVs are constructed from theCTV by 7 mm and 5 mm expansions for treatment planning and coverage evaluation, respectively. The SBRT plan made from theplanning FBCT is copied to the CBCT acquired in each delivery fraction after rigid-body registration. B-Spline based deformableregistration is performed from each CBCT to the planning CT; and the SBRT dose calculated from each CBCT scan is deformed tothe planning CT accordingly, in order to evaluate the total delivered dose. Anatomy landmarks were contoured on CBCT and de-formed onto planning CT to evaluate the accuracy of the deformable registration.

Results: Landmark contours deformed onto FBCT generally deviate from the contours drawn from FBCT by less than 1 mm,with maximum deviation of 3 mm, yielding a 1 mm accuracy in dose-mapping reconstruction. V95, which is the fraction of thePTV that receives at least 95% of planning dose, is 100% in treatment planning, whereas the actual V95 as evaluated by on-lineCBCT is 96.5%, with a standard deviation of 2.2% from fractions to fractions. The uniformity index D5/D95 is 1.07 in treat-ment planning, whereas the mean uniformity index is 1.16, with a standard deviation of 0.05 from fractions to fractions. D5 andD95 are respectively dose to 5% PTV volume and dose to 95% PTV volume, where the PTV is the PTV expanded by 5 mmfrom CTV.

Conclusions: On-line CBCT is used to evaluate the actual dose delivered to pancreatic cancer. A 7 mm treatment planning marginensures target coverage and dose uniformity. Thus, smaller margins may be acceptable and are presently investigated for this bodysite.

Author Disclosure: B. Liu, None; M. Garofalo, None; F. Lerma, None.

2829 Optimizing the Cardiac and Pulmonary Dose: A Comparison of IMRT Photon and 3-D Proton Treatment

Planning for Distal Esophageal Cancer

E. M. Crowley1, L. A. Kachnic2, H. J. Mamon3, J. A. Adams1, N. C. Choi1, T. S. Hong1

1Massachusetts General Hospital, Boston, MA, 2Boston University Medical Center, Boston, MA, 3Dana Farber Cancer Institute,Boston, MA

Purpose/Objective(s): For distal esophageal cancer, intensity-modulated radiation therapy (IMRT) has been suggested asa method to decrease radiation dose to the heart and lungs. However, even with IMRT, it can be difficult to achieve all tumor do-simetric goals and critical normal tissue dose constraints while maintaining dose homogeneity. In this study, we perform a dosevolume histogram (DVH) comparison of 3D passively scattered proton treatment plans and photon IMRT plans for 8 patientswith distal esophageal cancer.

Materials/Methods: Eight patients with distal esophageal/GE junction cancer were identified. The clinical tumor volume(CTV) receiving the initial 45 Gy was defined by the CT-based gross tumor volume (GTV) with a 3-5 cm cranial-caudaland a 1-2 cm radial expansion. A cone down boost of 5.4 Gy was added to the GTV plus a customized margin yielding a finaldose of 50.4 Gy. For photon IMRT plans, planning tumor volume (PTV) expansions were individualized based on 4D-CT. Forproton plans, residual motion was assessed with 4D-CT to determine the dosimetric ‘‘smearing’’ necessary to account for thechanging tissue heterogeneity (lung vs. soft tissue) due to diaphragmatic excursion. The treatment planning goals were, in orderof priority: (1) spinal cord max dose \45 Gy; (2) .95% of the PTV covered by the prescription isodose line; (3) lung dose

S540 I. J. Radiation Oncology d Biology d Physics Volume 72, Number 1, Supplement, 2008

restricted to a combined V20 \30% and a mean lung dose of 20 Gy; (4) heart dose restricted to V40 \20%; and (5) global hotspot of less than 15%.

Results: DVH results were obtained for 16 proton and IMRT distal esophageal plans generated for 8 patients. Overall, the protonplans achieved comparable CTV coverage, but with superior dose homogeneity. The average global maximum dose was 106%with protons and 114% with IMRT. Proton plans also yielded lower lung dose across V5, V10, V20, V30, and V40. This differencewas most striking in the low dose region, with a combined lung V5 of 31.6% and 69.8% for protons and IMRT respectively. In spiteof the superior lung sparing, cardiac dose was also lower using protons, with lower cardiac V10, V20, V30, V40, and V50. Spinalcord planning constraints were achieved for all patients.

Conclusions: In this preliminary planning study comparing 3D passively scattered protons and IMRT with our institutional mo-dality-specific motion solutions, protons consistently yielded more homogeneous plans with greater cardiac and pulmonary spar-ing, particularly in the low dose region. Studies are needed to determine if these dosimetric improvements are clinically relevant.

Author Disclosure: E.M. Crowley, None; L.A. Kachnic, None; H.J. Mamon, None; J.A. Adams, None; N.C. Choi, None; T.S.Hong, None.

2830 Dose Escalation with Proton or Photon Radiation Treatment for Pancreatic Cancer

M. Bouchard, T. M. Briere, R. A. Amos, S. Beddar, C. H. Crane

U.T. M.D. Anderson Cancer Center, Houston, TX

Purpose/Objective(s): Dose escalation for pancreatic cancer may offer better disease control for selected patients. Proximity oforgans at risk (OAR) could impair the feasibility of dose escalation using conventional 3D conformal radiation therapy (3D-CRT).The purpose of this study is to define which pancreatic tumor locations are safe for dose escalation (72 Gy) according to the fol-lowing treatment modalities: 3D-CRT, IMRT or protons.

Materials/Methods: We used free-breathing CT and 4D-CT data sets showing the same 3 cm pancreas head tumor. A 3 mm CTV,a 7 mm superior-inferior expansion for the ITV and a 5 mm 3D expansion for the PTV were added. Then, the same PTV was trans-lated to the body and tail. For each location, we generated 3 plans: 3D-CRT (18 MV beams), IMRT (6 MV beams) and proton plans(passive scattering technique). The PTV in photon and proton planning is different, but the same area was planned to receive 72 Gy(or CGE) in 36 fractions with at least 95% receiving 99% of the prescription dose. For all plans, we set the following constraints:liver, mean dose \30 Gy; stomach, \54 Gy; small bowel, V50-54 Gy = 2% and V45-54 Gy = 25%; spinal cord \45 Gy; duo-denum\60 Gy and V45-60 Gy = 33%. For the proton plans, we verified each beam dose distribution. We evaluated the distancebetween GTV and isodoses. We compared DVHs for target coverage and OAR sparing.

Results: To achieve objective coverage for the head location, the distance between the duodenum and the GTV should be morethan 15 mm (60 Gy isodose distribution with isocenter in the middle of GTV, radial directions) with IMRT while 18mm anteriorand 26 mm to the right for proton plans. The stomach needed to be more than 19 mm (54 Gy, all directions) with IMRT and protons.For tumors located in the body or tail of pancreas, the small bowel within 20 mm of the GTV precludes the possibility of doseescalation for IMRT and proton plans. With respect to all OAR constraints, the ITV coverage at 72 Gy for the patient’s head tumor,the most likely clinical situation, were as follows: 3D-CRT = 59%; IMRT = 92%; proton plans = 94%.

Conclusions: The knowledge of the different dose distributions according to treatment modality gives an indication of which pan-creatic cancer cases benefit from dose escalation according to patient anatomy and tumor location. IMRT allows more a conformaldose distribution in the high dose regions while proton therapy reduces low dose bath irradiation to the body. Uncertainty marginsneeded for proton planning precludes its full potential for higher dose area and intensity-modulated proton therapy might be a so-lution.

Author Disclosure: M. Bouchard, None; T.M. Briere, None; R.A. Amos, None; S. Beddar, None; C.H. Crane, None.

2831 IMRT for Anal and Anorectal Cancers: Incorporation of In Vivo Dosimetry in Radiotherapy Plan

Verification and Dose Correction

R. Nordal, R. Popple

University of Alabama, Birmingham, AL

Purpose/Objective(s): Anal tumors present radiation planning challenges in delivering curative doses to the primary tumor andlymph nodes while sparing multiple avoidance structures. Intensity modulated radiotherapy (IMRT) has the potential to meet thesechallenges. The objective was to determine the accuracy of target dose distributions obtained from radiotherapy planning software,and the need for dose modification based on measured doses in the high dose region.

Materials/Methods: 14 patients (pts) with anal and anorectal tumors received IMRT and concurrent chemotherapy. IMRT planswere generated using the Varian Eclipse system. Treatment was delivered with dynamic multileaf collimation. Radiation plansemployed simultaneous integrated boost and moderate dose escalation and treatment acceleration. IMRT planning employed 7 co-planar axial fields, and either a 2 or 3 dose level-fractionation scheme. (Total and daily doses were 5400/216 and 4500/180 [n = 8],or 5940/212.1, 5440/194.3, and 4760/170 [n = 6]). Actual doses were measured at locations within the anorectum during radio-therapy delivery. Radiation doses were measured using thin layer dosimetry (TLD), with confirmation using a metal oxide semi-conductor field effect transistor (MOSFET) detector system in selected cases.

Results: One to 5 measurement locations in the anorectum spanned the region of the primary tumor. Measured doses were com-pared with the prescription point dose from the radiotherapy plan (including error estimation for the plan dose due to setup errorand organ motion). TLD measurements are subject to approximately +/� 5% error, plus additional variation between annealingbatches. The accuracy of the MOSFET system is +/� 4% per the manufacturer. Concordance was observed between calculatedand measured doses, and between TLD and MOSFET measurements for most patients. Correction of monitor units (MU) de-livered was considered for pts with measured dose discrepancy .7% (5% TLD error) + 2%(estimated error for plan point dose