Proceedings of the 52nd Annual ASTRO Meeting S53

Conclusions: Despite the use of breast plus boost RT, women with close and positive margins, in combination with grade III his-tology, LVI, and N2/3 stage, experienced significantly higher rates of LR as the first event, compared to counterparts with negativemargins. More definitive surgery should be considered to optimize local control in these patient subsets.

Author Disclosure: P. Truong, None; K. Lupe, None; C. Alexander, None; M. Lesperance, None; S. Tyldesley, None.

112 Triple-Negative Breast Cancer is Associated with Higher Risk of Local Recurrence after 3D-Conformal

External Beam Accelerated Partial Breast Irradiation (3D-APBI)

I. M. Pashtan1, M. Ancukiewicz2, J. Y. Wo1, A. E. Hirsch3, B. L. Smith2, S. N. Powell4, A. Recht5, A. G. Taghian2

1Harvard Radiation Oncology Program, Boston, MA, 2Massachusetts General Hospital, Boston, MA, 3Boston Medical Center,Boston, MA, 4Memorial Sloan-Kettering Cancer Center, New York, NY, 5Beth Israel Deaconess Medical Center, Boston, MA

Purpose/Objective(s): APBI has become an increasingly utilized component of breast conserving therapy. Currently, there are noevidence-driven recommendations regarding its use in relation to biologic subtype. Here, we evaluate whether biologic subtype, asdetermined by estrogen, progesterone, and HER-2 receptors, is associated with local recurrence (LR) following 3D-APBI.

Materials/Methods: From October 2003 through November 2005, 99 patients with Stage I breast cancer were treated on the firstdose-step of a prospective IRB-approved dose-escalation clinical trial. Eligibility criteria included: infiltrating ductal histology orspecial subtype; pT1N0; absence of EIC and LVI; margins $ 2 mm. The median age was 61 years (range, 40-80 years); the mediantumor size was 0.8 cm; 91% had ductal carcinomas. 89% of tumors were ER+; 10% had a triple-negative phenotype; 1% hada HER2 subtype. 47% were histologic grade 1, 42% grade 2, and 10% grade 3; higher tumor grades correlated with triple-negativephenotype (Kendall’s tau = 0.43, p \ 0.0001). 7 patients received chemotherapy and 65 hormonal therapy. Radiation techniquesincluded: mixed photons and electrons, usually with 2 mini-tangents and an en-face electron field delivering about 20% of the dose,for 63 patients; photons alone for 16 patients; and protons for 20 patients. Median FU was 57.8 months (range, 1.6-74.3 months).

Results: Five patients had LR at a median of 47.5 months after APBI (range, 29.5-55.3 months), for a 5-year actuarial LR rate of6% (95% CI, 0.6-11%). Three of these occurred in patients with triple-negative disease (1 same quadrant recurrence, 2 elsewhere),for a 5-year actuarial LR rate of 32.5% (95% CI, 0-57%). 2 occurred in non-triple-negative patients (both elsewhere), for a 5-yearactuarial LR rate of 3% (95% CI, 0-7%) [p = 0.0001, log-rank test]. The hazard ratio for LR with triple-negative phenotype was 15.2(95% CI, 2.5-91). All LR patients had received systemic treatment. On multivariate analysis, triple-negative phenotype was theonly predictor of LR, with borderline statistical significance (p = 0.052, log-rank test) after adjusting for tumor grade. Effect oftumor grade on risk of LR was not significant (p = 0.33, log-rank test) after adjusting for triple-negative status. All LR patientswere successfully salvaged with mastectomy.

Conclusions: In this study, patients with triple-negative breast cancer had a significantly higher LR rate than patients with otherreceptor subtypes following 3D-APBI to 32 Gy. Additional prospective studies are needed to assess the long-term efficacy of thisapproach, and whether patients with triple-negative disease might be better treated with whole-breast irradiation.

Author Disclosure: I.M. Pashtan, None; M. Ancukiewicz, None; J.Y. Wo, None; A.E. Hirsch, None; B.L. Smith, None; S.N. Po-well, None; A. Recht, CareCore Oncology Ltd, F. Consultant/Advisory Board; A.G. Taghian, None.

113 Chemoradiation (CRT) Safety Analysis of ACOSOG Z6041: A Phase II Trial of Neoadjuvant (NEO) CRT

followed by Local Excision (LE) in uT2 Rectal Cancer (RC)

E. Chan1, Q. Shi2, J. Garcia-Aguilar3, B. G. Wolff2, C. S. Johnson4, B. Sanders5, X. Carrero2, M. C. Posner6, D. M. Ota7,

C. R. Thomas8

1Vanderbilt-Ingram Cancer Center, Nashville, TN, 2Mayo Clinic, Rochester, MN, 3City of Hope, Duarte, CA, 4Natalie WarrenBryant Cancer Center, Tulsa, OK, 5St. Francis Cancer Center, Indianapolis, IN, 6University of Chicago, Chicago, IL, 7DukeClinical Research Institute, Durham, NC, 8Oregon Health and Science University Knight Cancer Institute, Portland, OR

Purpose/Objective(s): Radical surgical resection of RC can lead to significant morbidity. Sphincter-sparing approaches, such asLE, result in less morbidity but has a higher local recurrence rate. This phase II major cooperative group study explores neo CRTfollowed by LE for uT2 RC. The specific aim of the present analysis is to report the toxicity results during the neo CRT.

Materials/Methods: Patients (Pts) with uT2N0 RC were treated with external beam radiation (RT) (total dose 50.4 Gy) combinedwith capecitabine (C) (825 mg/m2 M-F wks 1-5) and oxaliplatin (50 mg/m2 on Days 1, 8, 22, and 29) (Original dose = O dose) for 6wks followed by LE 4-8 wks after completion of CRT. Study was subsequently amended secondary to toxicity to lower the C doseto 725 mg/m2 bid (Modified dose = M dose), with an early stopping rule if the $ grade (G) 3 adverse event (AE) rate was $ 30%with the M dose. Planned accrual was 102 pts. The study closed early due to crossing the stopping toxicity boundary at M dose.Frequencies and percentages were reported for AE.

Results: Ninety pts accrued, 62 pts in the O dose and 28 pts in the M dose. All pts that signed consent, received $ 1 cycle oftreatment (Tx), and had available AE data were considered in the safety analysis. 84 pts (O dose, 57; and M dose, 27) were evalu-able for AE. At the O dose, 25 (44%) pts had a $ G3 AE regardless of attribution, all deemed at least possibly related to Tx. 9 (33%)pts treated with the M dose developed a $ G3 AE regardless of attribution. 8 (30%) were deemed at least possibly related to Tx. Themost common $ G3 AE seen with the O dose were GI (n = 16 [28%]), dermatologic (n = 6 [11%]), hematologic (n = 4 [7%]), pain(n = 4 [7%], and metabolic (n = 3 [5%]). The most common $ G3 AE seen with the M dose were hematologic (n = 5 [19%]), GI (n =3 [11%]), dermatologic (n = 2 [7%]), pain (n = 2 [7%], and metabolic (n = 2 [7%]). 48 (84%) pts on the O dose and 20 (74%) pts onthe M dose completed chemotherapy (CT). In wks 1-3, 6 (11%) pts on the O dose and 1 (4%) of pt on the M dose had a CT dosedelay or modification secondary to AE. In wks 4 and 5, 17 (30%) pts on the O dose and 5 (19%) of pts on the M dose had a CT dosedelay or modification secondary to AE. With regard to RT, 47 (82%) pts on the O dose and 27 (100%) pts on the M dose completedRT per protocol. 6 (11%) pts on the O dose did not complete RT as a result of AE. In the O dose group, 30 (53%) pts completed RTwithout interruptions and 12 (21%) had interruption of RT as a result of AE. In the M dose group, 19 (70%) pts completed RTwithout interruptions and 3 (11%) had interruption of RT as a result of AE.

S54 I. J. Radiation Oncology d Biology d Physics Volume 78, Number 3, Supplement, 2010

Conclusions: While neo CRT followed by LE on Z6041 yielded one of the highest pCR rate for a cooperative group setting (ASCO2010), the current regimen is toxic. Our successor trial design aims to increase the pCR rate while minimizing toxicity, thus op-timizing the therapeutic ratio.

Author Disclosure: E. Chan, None; Q. Shi, None; J. Garcia-Aguilar, None; B.G. Wolff, None; C.S. Johnson, None; B. Sanders,None; X. Carrero, None; M.C. Posner, None; D.M. Ota, None; C.R. Thomas, None.

114 Phase II Trial of Neoadjuvant Capecitabine (Cp) with Irinotecan (Ir) followed by Capecitabine-based

Chemoradiotherapy (CRT) for Patients (pts) with Locally Advanced Rectal Cancer (LARC): A HoosierOncology Group Study

H. Cardenes1, S. Sanghani1, M. Burns1, B. Robb1, D. Hinkle1, C. Johnson1, M. Yu1, C. Currie2, P. Loehrer1, E. Chiorean1

1Indiana University Medical Center, Indianapolis, IN, 2Hoosier Oncology Group, Indianapolis, IN

Purpose/Objectives: Fluoropyrimidine-based neoadjuvant CRT is standard therapy for LARC, but high systemic failure ratessuggest the addition of induction chemotherapy to this regimen. Ir and Cp are pro-drugs that require activation by carboxylesteraseenzymes (CES). The purpose of this study was to determine the pathological response rate of this regimen (complete pCR, and non-complete pNCR), and to analyze the gene expression profile of enzymes involved in the metabolism of Cp and Ir.

Materials/Methods: Patients with endoscopic ultrasound (EUS) staged T3/T4 or $ N1 adenocarcinoma of the rectum weretreated with Cp 1000 mg/m2 BID days 1-14, and Ir 200 mg/m2 on day 1 q21 days for 2 cycles, followed by Cp 825 mg/m2

BID days 1-5/week with concurrent RT 50.4 Gy [28, 1.8- Gy fractions]. Surgical resection was performed 4-6 wks after completionof therapy. Gene expression levels or sequencing were used to analyze CES1, CES2, thymidylate synthase, thymidine phosphor-ylase, topoisomerase1 and UGT1A in pre- and post-treatment tumor samples. T-test was used to study the association with re-sponse and toxicity.

Results: Twenty-two pts were enrolled, median age 54 yrs (36-67), M/F 11/11, T3/T4 86%/9% and 91% $ N1. Eighteen pts com-pleted all neoadjuvant therapy (3 pts discontinued for social reasons, and 1 pt due to metastatic disease). The most common tox-icities were diarrhea (86%), fatigue (73%), nausea (73%), alopecia (59%), and rectal pain (36%). The only grade 3 toxicityoccurring in . 1 pt was diarrhea (n = 4), and 1 pt had grade 4 febrile neutropenia. All 18 pts underwent R0 resection. Downstagingwas noted in 16 pts (89%), and 6 (33%) obtained pCR. After a median follow-up of 43 months (95% CI 20-49), 3 pts progressed, 1pt died and the median PFS has not yet been reached. We observed a trend for higher CES2 expression in the pNCR vs. pCR groups(p = 0.07), as well as an increased incidence of grade $ 2 diarrhea associated with higher CES1 levels (p = 0.09).

Conclusions: The combination of Cp+Ir followed by CRT has significant antitumor activity and tolerable toxicity. The effect ofCES on toxicity and efficacy from Cp and Ir needs to be validated in larger study.

Author Disclosure: H. Cardenes, None; S. Sanghani, None; M. Burns, None; B. Robb, None; D. Hinkle, None; C. Johnson, None;M. Yu, None; C. Currie, None; P. Loehrer, None; E. Chiorean, None.

115 Validated Nomograms based on Six Large European Trials for the Prediction of Local Control, Distant

Metastases, and Survival for Locally Advanced Rectal Cancer Patients after LongCourse Chemoradiotherapy

V. Valentini1, R. G. P. M. van Stiphout2,3, K. Bujko4, L. Cionini5, C. Rodel6, J. P. Gerard7, F. Bonnetain8, L. Collette9,

J. F. Bosset10, P. Lambin2

1Department of Radiotherapy, Universita Cattolica S. Cuore, Rome, Italy, 2Department of Radiation Oncology (MAASTRO),GROW, University Medical Centre Maastricht, Maastricht, Netherlands, 3Department of Knowledge Engineering, Faculty ofHumanities and Sciences, Maastricht, Netherlands, 4Department of Radiotherapy, The Maria Sklodowska-Curie MemorialCancer Centre, Warsaw, Poland, 5Department of Radiotherapy, Azienda Ospedaliera Universitaria Pisana, Pisa, Italy,6Department of Radiotherapy and Oncology, University of Frankfurt, Frankfurt, Germany, 7Department of Radiotherapy,Centre Antoine Lacassagne, Nice, France, 8Department of Statistics, Faculte Medecine Dijon, Dijon, France, 9Department ofStatistics, ESTRO, Brussels, Belgium, 10Department of Radiotherapy, CHU Besancon, Besancon, France

Purpose/Objective(s): To develop accurate nomograms to predict local recurrence (LR), distant metastases (DM) and overall sur-vival (OS) within 5 years of follow-up for locally advanced rectal cancer patients treated with long-course chemoradiotherapy. Thisshould allow proper selection in future trials and selection of patients who may benefit most from postoperative (alternative) ad-juvant chemotherapy and intensive follow-up.

Materials/Methods: Clinical data (3253 pts) in a pooled database originating from six trials was analyzed; French trial (697 pts)1,EORTC trial (950 pts)2, German trial (387 pts)3, Polish trial (136 pts)4 and two not already published Italian population-based co-horts of 595 and 488 patients. The common clinical variables for all datasets were: gender, age, tumor location, RT dose, concom-itant chemotherapy (CT), surgery type (local excision, LAR, APR), pT-stage, pN-stage, and adjuvant CT. The prediction modelswere based on multivariate analysis with a support vector machine. Performance of the model was expressed as the Area-Under-the-Curve (AUC) of the Receiver Operating Characteristic (ROC) curves. Nomograms were developed based on a 100-fold ran-domized selection of train and validation sets, with an optimal training set size of 60%. Weights were assigned to each selectedpredictor and converted to a probability for a 5-year outcome using logistic regression.

Results: In the pooled database the occurrences of events were 9,7% LR, 26.4% DM, 67.5% disease free survival (DFS) and 76.5%OS. The accuracy of the developed nomograms was consistent for the validation sets for all outcomes; AUCLR = 0.70+/-0.019,AUCMD = 0.72 +/- 0.01, AUCOS = 0.69+/-0.013. Tumor location, RT dose, surgery type, and pTN-staging were important pre-dictors for all outcomes. Concomitant CT and adjuvant CT were only selected for LR prediction with relative low impact. Ameta-analysis supports these findings with respect to the model accuracy and selected predictors.

Conclusions: The provided nomograms have been validated and are able to accurately predict long-term outcome for rectal cancerpatients after chemoradiotherapy. These nomograms should allow proper stratification in future trials, generation of new