Volume 96 Number 2S Supplement 2016 significantly benefitted from adjuvant chemoradiation on multivariable analyses (HR Z 0.62, 95% CI Z 0.40-0.96, P Z 0.033) and those with tumors 4 cm significantly benefitted from adjuvant chemotherapy (HR Z 0.71, 95% CI Z 0.53-0.96, P Z 0.028). The same results were seen in the margin-negative cohort. In the margin-positive cohort, patients with tumors >4 cm significantly benefitted from both adjuvant chemoradiation (HR Z 0.54, 95% CI Z 0.32-0.92, P Z 0.024) and radiation (HR Z 0.52, 95% CI Z 0.29-0.93, P Z 0.028). In margin-positive patients with tumors <4 cm, a trend towards improved overall survival was seen in those who received adjuvant chemotherapy, but the patient cohort was too small for this to be significant. Conclusion: Our results suggest that size is the most important determinant of the management of tumors invading the chest wall: tumors 4 cm should be treated with adjuvant chemotherapy whereas tumors >4 cm should be treated with adjuvant CRT if margin-negative and adjuvant chemoradiation or radiotherapy if margin-positive. These findings demonstrate that tumors staged as T3N0 with chest wall invasion should be managed differently from other stage IIB tumors. Author Disclosure: S. Gao: None. C.D. Corso: None. F.C. Detterbeck: None. D. Boffa: None. R.H. Decker: None. A.W. Kim: None.
154 The Role of Postoperative Radiation Therapy in pN2 Non-Small Cell Lung Cancer: An Analysis of the National Cancer Data Base A. Herskovic, P. Christos, E. Mauer, and H. Nagar; New York-Presbyterian Hospital/Weill Cornell Medicine, New York, NY Purpose/Objective(s): The role of postoperative radiotherapy (PORT) in the treatment of pathologic N2 (pN2) nonesmall cell lung cancer (NSCLC) remains controversial. A heterogeneous set of trials and metaanalyses conducted over a long period of time have come to differing conclusions. We investigated national practice patterns for patients with pN2 NSCLC treated with surgery and multiagent chemotherapy and the impact of modern PORT on OS using the National Cancer Database (NCDB). Materials/Methods: Patients with known histologies of pN2 NSCLC who underwent surgery with negative margins and received adjuvant multiagent chemotherapy from 2004-2013 were identified from the NCDB and stratified by the use of PORT (45 Gy). Patients also had to have survived at least 98 days after chemotherapy initiation to be included, so that patients were only included if they were expected to survive long enough to receive adjuvant therapy. To prevent receipt of concurrent chemoradiation, patients in the PORT group received radiation 84 days from the initiation date of chemotherapy. Landmark analysis was set at 120 days following the initiation date of adjuvant chemotherapy to prevent survival bias in the non-PORT group, as this was the median time to initiation of PORT. Multivariable proportional hazards modelling was used to examine factors associated with receiving PORT and the association of treatment and mortality adjusting for demographic, socioeconomic and clinicopathologic factors. Results: A total of 2,815 patients were identified with median follow-up of 32.5 months. Improved survival was associated with female sex, lower Charlson-Deyo comorbidity index, race, smaller tumor size, squamous histology, lobectomy as surgery performed and receipt of PORT. Prior to landmark analysis, the HR showed a benefit to PORT. This benefit remained after landmark analysis was completed, with a median overall survival was 51.7 months in the PORT group versus 44.2 months in the no PORT group (adjusted HR 0.82, [95% CI Z 0.71 to 0.93]; P < 0.005). Factors associated with receipt of PORT were facility location, CharlsonDeyo comorbidity index and grade (all P < 0.05). Conclusion: Improved survival is associated with receipt of PORT for patients with N2 NSCLC treated with complete resection and multiagent chemotherapy. Evidence from randomized trials should better delineate the role of PORT in this patient population. Author Disclosure: A. Herskovic: None. P. Christos: None. E. Mauer: None. H. Nagar: None.
Oral Scientific Sessions
S69
155 The Differential Impact of Postoperative Radiation Therapy for Completely Resected Stage IIIA (N2) Non-Small Cell Lung Cancer: Based on the Risk Prediction Model for Locoregional Recurrence W. Feng, X.L. Fu, X.W. Cai, Q. Zhang, W. Yu, and H.Q. Chen; Shanghai Chest Hospital, Shanghai Jiao Tong University, Shanghai, China Purpose/Objective(s): The value of postoperative radiotherapy (PORT) for completely resected non-small cell lung cancer (NSCLC) remains controversial because the effect on survival has been inconclusive. Growing evidence suggests that PORT using the modern 3D-CRT technique has a favorable effect on survival of patients with pN2 disease. After complete resection and adjuvant chemotherapy (CT), 20%-40% of cases have a risk of locoregional recurrence (LRR). Identification of factors that predict for LRR after surgery may help in precise PORT strategies. In this study, we aimed to establish a clinical risk prediction model for LRR and evaluate the efficacy of PORT based on the risk stratification. Materials/Methods: Data were analyzed for all consecutive patients between 2005 and 2012 with pathologic T1-3N2M0 NSCLC treated with complete resection with negative surgical margins and no neoadjuvant RT and/or CT. The Prognostic Index (PI) was built first in the non-PORT group, using significant prognostic factors in cumulative incidence of LRR on the Cox regression model. Then, based on these factors, a prognostic scoring system was developed that could be applied in the clinical setting to separate this cohort of patients into two risk groups based on the median of the PI scores. We hypothesized that the degree of survival benefit between treatment arms (PORT group and non-PORT group) would vary among the different risk groups. Results: Three hundred fifty-seven cases (70 patients in PORT group and 287 in non-PORT group) met the inclusion criteria and entered the analysis. The median follow-up was 55.3 months (range, 23.9-132) for living patients. In non-PORT group, the 5-year Kaplan-Meier LRR rate as first event was 31.3%. On multivariate analysis, heavy cigarette smoking history (HR Z 2.4, b Z 0.9, P Z 0.001), cN2 status (HR Z 1.7, b Z 0.5, P Z 0.03) and number of involved lymph nodes >4 (HR Z 2.2, b Z 0.8, P Z 0.001) were independently significant factors predicting LRR. The PI equation was built including the three significant categorical variables and coefficients based on their level of significance: PI Z (0.9 smoking history) + (0.5 clinical N status) + (0.8 number of involved lymph nodes). For the low risk subset (PI score <3.5, N Z 190), PORT tended to reduce 5-yr Kaplan-Meier LRR rate (2.8% for PORT vs 15.9% for nonPORT, P Z 0.1); but no significant differences in overall survival (OS) were noted (5-yr OS: 42.5% for PORT vs 44% for non-PORT, P Z 0.5). For the high risk subset (PI score 3.5, N Z 167), PORT significantly improved not only local control (5-yr Kaplan-Meier LRR rate: 10.6% for PORT vs 51.5% for non-PORT, P < 0.001), but also actual survival (5-yr OS: 55.7% for PORT vs 22.5% for non-PORT, P < 0.001). Conclusion: This risk model for prediction of LRR after surgery may provide clinicians with a better framework for recommending PORT for completely resected stage IIIA(N2) NSCLC. Further studies to validate this prediction model are still needed in order to refine the precise indications for PORT and how to integrate PORT with adjuvant chemotherapy. Author Disclosure: W. Feng: None. X. Fu: None. X. Cai: None. Q. Zhang: None. W. Yu: None. H. Chen: None.
156 Prediction of Treatment Response of Cervical Nodes Using IVIMDWI G. Sanguineti,1 S. Marzi,2 L. Marucci,1 A. Farneti,1 F. Piludu,2 and A. Vidiri2; 1Regina Elena National Cancer Institute, Rome, Italy, 2 Reginal Elena Cancer Institute, Rome, Italy Purpose/Objective(s): To investigate the predictive role of Intravoxel Incoherent Motion Diffusion-Weighted Imaging (IVIM-DWI) parameters on cervical nodal response both at baseline and during chemoradiotherapy (CRT) of head and neck squamous cell carcinoma (HNSCC).
S70
International Journal of Radiation Oncology Biology Physics
Materials/Methods: Patients were included in the present prospective study if they had pathologically confirmed HNSCC of various subsites, had at least one positive lymph node (LN) in the neck and were planned to receive concomitant chemo-IMRT. In addition to standard imaging, patients underwent two serial IVIM-DWIs: before and at mid-treatment. Three contiguous sections of the largest LN of each patient were contoured on each scan, using T2-weighted images as a guide for the lesion location. Apparent diffusion coefficient (ADC), perfusion fraction f and pure diffusion coefficient D were estimated. Lymph node failure was defined as the pathological evidence of residual disease at completion surgery after CRT for presumed residual disease or the development of a nodal failure during the follow after a complete clinical response. Also in the latter case pathological confirmation was required. The ManneWhitney test for unpaired samples was performed to compare the two patients groups, responders, and nonresponders. Optimal thresholds for pre-treatment values and early changes of all parameters were evaluated using the Receiver Operating Characteristic (ROC) Curves. Results: Thirty-four patients were accrued from October 2011 to May 2014. The primary tumor sites were: oropharynx (N Z 15), nasopharynx (N Z 13), hypopharynx (N Z 5), and unknown (N Z 1). Median follow up is 21.2 months, with patients with no regional failure having a minimum follow up of 12 months. Ten LNs (29.5%) were found to have pathology proven residual disease at neck dissection after CRT (N Z 6) or during the follow up (N Z 4). Baseline IVIM-DWI parameters were extracted for patients at baseline imaging, while, due to susceptibility artifacts, 4 patients had missing mid-treatment values. Patients with controlled LNs showed significantly lower pre-treatment ADC and D values than patients having regional failure (P Z 0.038 and P Z 0.034, respectively). Half-way through RT, patients having regional failure showed significantly higher D (P Z 0.025) and exhibited borderline lower f values and larger reductions in f compared to baseline (P Z 0.060 and P Z 0.105, respectively). Using optimal thresholds from the ROC analysis, the negative predictive values (NPV) were 86% for both ADC and D at baseline and 100% and 95% for D and f at mid-treatment, while the positive predictive values were 58%, 54%, 50%, and 67%, respectively. Conclusion: IVIM-DWI parameters at both baseline and mid-treatment allow remarkably high NPV values that are promising in the setting of evaluation of nodal response after CRT and thus deserve further investigation. Author Disclosure: G. Sanguineti: None. S. Marzi: None. L. Marucci: None. A. Farneti: None. F. Piludu: None. A. Vidiri: None.
MRIs. The tumors were contoured based on multi-parameter MRI, including T1 Arterial, T1 Venous, and ADC images, on pre- and post-nCR MRIs. A statistical analysis was performed to identify regions of interest (ROI) and spatial trends between pre- and post-nCR MRIs. To detect statistical changes within a ROI, a t-test was used to compare the ADC spectrum of smaller ROI samples (ROIs) within the tumor to the overall distribution of ADC of the tumor ROI (ROIt). For each voxel, the ROIs is defined as N by N by N/3 region around the voxel. The P value of the t-test is mapped out to define the statistically similar regions, along with the outlier regions with higher or lower ADC mean than the ROIt mean. The statistically similar region is defined to be voxels within (1-P) < 0.0001; this translates to the distribution within the ROIs having a very similar mean and standard deviation as the ROIt distribution. For an evenly mixed combination of 2 means, the t-test will fail to isolate outliers due to the lack of variation between different ROIs. Additional textures were calculated in the identified ROIs, first order features to show the change in combination of cellularity and fibrosis, and GLCM features to show the fine variations of the texture. Results: The ROIs identified with a statically similar manner between preand post-nCR ADC are different between the poor (G3) and moderate to good response (G1 and G2) cases. The mean percent of the total volume for G3 is 65 +/- 13% (pre-nCR) and 51 +/- 19% (post-nCR). For G2, 30 +/7% (pre) and 38 +/- 8% (post); and for G1, 39 +/- 9% (pre) and 38 +/- 8% (post). Conclusion: Spatially, the ADC distribution for a G3 response tumor does not vary greatly in contrast to G1 or G2 response tumors which demonstrate greater inhomogeneity in both pre and post-nCR. These data reflect the texture of ADC maps acquired prior and post nCR could be used to assess tumor response and to improve the selection of patients for possible therapeutic intensification. Author Disclosure: D. Schott: None. W.A. Hall: None. T. Gilat Schmidt: None. E. Dalah: None. K. Oshima: None. E.S. Paulson: None. P.M. Knechtges: None. B.A. Erickson: None. A. Li: None.
157 Correlation of ADC Texture With Treatment Response for Chemoradiation Therapy of Pancreatic Cancer D. Schott,1 W.A. Hall,2 T. Gilat Schmidt,3 E. Dalah,1 K. Oshima,4 E.S. Paulson,1 P.M. Knechtges,5 B.A. Erickson,1 and A. Li1; 1Department of Radiation Oncology, Medical College of Wisconsin, Milwaukee, WI, 2 Medical College of Wisconsin Department of Radiation Oncology, Milwaukee, WI, 3Department of Biomedical Engineering, Marquette University, Milwaukee, WI, 4Department of Pathology, Medical College of Wisconsin, Milwaukee, WI, 5Department of Radiology, Medical College of Wisconsin, Milwaukee, WI Purpose/Objective(s): Imaging texture analysis has been shown to reveal tumor heterogeneity, which extends beyond what is visible to the human eye. The purpose of this study is to perform a texture analysis on apparent diffusion coefficient (ADC) maps obtained from diffusion weight MRI of pancreatic cancer and to correlate the texture with treatment response. Materials/Methods: The MRI and pathological response data collected from 18 patients with resectable or borderline resectable pancreatic head cancer were analyzed. This set of patients all received neoadjuvant chemoradiation (nCR) before tumor resection. Pathological response to nCR was graded from the surgical specimen as G0, G1, G2, and G3 for complete, moderate, minimal, and poor response, respectively. The patient pool includes 4 patients for G1, 9 patients for G2, and 8 patients for G3 responses. Of the 18 patients, 17 have pre-nCR MRIs and 11 have post-nCR
158 Radiomic Features Prognostic for Recurrence in Human PapillomavirusePositive Oropharyngeal Cancer B.A. Altazi,1,2 G.G. Zhang,1 A.O. Naghavi,1 E.G. Moros,1 and J.J. Caudell1; 1H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, 2University of South Florida, Tampa, FL Purpose/Objective(s): We previously reported a multi-radiomic logistic regression models that were highly prognostic for HPV+ cervical cancer treatment outcomes. We hypothesized that applying the same models may be prognostic in HPV+ oropharyngeal cancer for locoregional recurrence (LRR) or distant metastasis (DM). Materials/Methods: A cohort of 50 patients diagnosed with HPV+ oropharyngeal cancer treated between 2007 and 2013 with definitive radiotherapy was identified. Patient, tumor, treatment, and outcome data were abstracted from the chart. Metabolic Tumor Volume (MTV) in patient’s primary site was delineated on pretreatment PET/CT by a board certified radiation oncologist. Eighty quantitative image features were computed for each volume. Radiomic features were extracted using four calculation methods; Co-occurrence Matrix (COM), Run Length Matrix (RLM), Gray Level Size Zone Matrix (GLSZM) in addition to shape/ geometric based features (SGBF). Two Standard Uptake values were measured; SUVPeak and SUVMax were all measured from MTV. The correlation between the extracted features, SUV and treatment outcomes were tested using logistic regression modeling and assessed using Area Under Receiver Operator Curve (AUROC). All the mentioned models were cross validated using Leave-on-out-cross-validation (LOOCV) resampling technique. Results: The models consisted of 2 to 3 Radiomic features for each outcome. LRR model consisted of (Intensity contrast and Low Gray-Level Run Emphasis). DM models consisted of (size zone variability and smallarea emphasis) and (surface/Area and Volume). Multivariable models for binary outcomes (DM and LRR) were fit using logistic regression models.