ESTRO 37 Abstract book

ESTRO 37

S409

Purpose or Objective To implement a classifier based on texture analysis on magnetic resonance imaging (MRI) only, and on MRI in association with positron emission tomography (PET), to predict the pathological response to neoadjuvant chemo- radiotherapy (CRT) in locally advanced rectal carcinoma (LARC). Material and Methods This was a retrospective study in which patients with the following inclusion criteria were considered: a) histologically diagnosed LARC, b) pre-treatment MRI and PET/CT, c) neoadjuvant treatment consisting of 46-55 Gy in 23-30 RT fractions alone or in association with either infusional 5-FU or oral Capecitabine, d) total mesorectal excision. Overall, 47 patients were included in this study, and underwent texture analysis. Patients with complete (tumour regression grade, TRG=1) or near complete (TRG=2) regression were defined as responders (pCR+), while patients with moderate (TRG=3) to no regression (TRG=5) were considered as non-responders (pCR-). From diffusion-weighted imaging, we computed the apparent diffusion coefficient (ADC) maps. Tumours were semi- automatically segmented on T2-weigthed (T2-w) MRI, ADC maps and PET/CT acquisitions (standardised uptake value, SUV). From the segmented tumours, the following quantitative features were computed from each image acquisition (T2-w, ADC and PET images): a) 22 second order texture parameters, derived from Haralick analysis, and b) first order parameters: median, mean, percentiles (10 th , 25 th , and 75 th ). Multivariate logistic regression was performed to identify features most correlated with TRG. In a first model, only MR parameters were investigated; in a second model, both MR and PET features were considered in the logistic regression. Results Among the 47 enrolled patients, 26 were classified as non-responders, and 21 as responders after total mesorectal excision. Parameters included in the multivariate regression considering only MR features were "Homogeneity ADC" and "Cluster prominence ADC" (p=0.034). The area under the receiver operating characteristic curve (AUC) was 0.70 (95% confidence interval (95%CI) = 0.55-0.83), sensitivity and specificity in detecting non-responders were 62% and 76%, respectively. When the regression was fed with both MR and PET parameters, the following features were considered in the model: "10 th percentile PET", "10 th percentile T2-w", "Homogeneity ADC", "Homogeneity PET", "Information measure of correlation T2-w" (p=0.002). AUC was equal to 0.83 (95%CI=0.69-0.93), sensitivity and specificity were 76% and 75%, respectively. Conclusion Texture analysis could provide additional information in assessing the response to neoadjuvant treatment in LARC patients. When pre-treatment MRI and PET were used in the regression, both 10 th percentile and homogeneity showed statistical significance in the prediction of TRG. These preliminary results, if confirmed on larger dataset, could be useful to personalize the oncological pathway for patients, in particular delaying or advancing surgery, according to the prediction of treatment response.

PO-0791 Poor prognostic and staging value of tumor deposits in rectal cancer with neoadjuvant chemoradiation Y. Wang 1 , J. Zhang 1 , L. Yang 1 , W. Deng 1 , L. Shen 1 , L. Liang 1 , M. Zhou 1 , W. Yang 1 , R. Hu 1 , J. Zhu 1 , Z. Zhang 1 1 Fudan University Shanghai Cancer Center, Department of Radiation Oncology, Shanghai, China Purpose or Objective Tumor deposits (TDs) are associated with poor prognosis in colorectal cancer. However, the significance in locally advanced rectal cancer (LARC) following neoadjuvant chemoradiotherapy (neo-CRT) and surgery is unclear. We aimed to evaluate the prognostic significance of TDs in LARC patients after neo-CRT, to verify the applicability of the N1c category in those tumors, and to explore the appropriate methods of N staging for those patients. Material and Methods Between 2006 and 2014, 495 LARC patients following neo- CRT and surgery were retrospectively analyzed. Clinicopathological features, overall survival (OS), disease-free survival (DFS), distant metastasis free survival (DMFS) and local recurrence free survival (LRFS) were recorded. Univariate and multivariate analysis were performed using the Kaplan-Meier method and Cox proportional hazards regression model in all patients and lymph nodes negative patients. In addition, we use three methods of N categories to evaluate the impacts of the counts of TDs on staging. 1, oN, N staging without TDs. 2, n1N, N staging according to the N1c category of the 8 th edition AJCC staging system. 3, n2N, N staging including the counts of TDs, considering one TD as one positive lymph node. Results Of the 495 patients, TDs were found in 88 (17.8%) cases. TDs positive patients were related to poorer differentiation, worse tumor regression, positive vascular or neural invasion, more advanced cN, ypT and ypN categories. Kaplan-Meier method and univariate analysis showed that TDs positive patients had worse OS, DFS and DMFS compared with TDs negative patients, both in all patients and lymph nodes negative patients (all P<0.01, Figure 1 a-b, d, e-f, h). In multivariate analysis, TDs were independent poor prognostic factors of OS (HR 1.774, 95%CI 1.124-2.800, P=0.014), DFS (HR 1.638, 95%CI 1.135- 2.363, P=0.008) and LRFS (HR 2.072, 95%CI 1.040-4.127, P=0.038) in all patients, but not DMFS (HR 1.433, 95%CI 0.939-2.187, P=0.095). In lymph nodes negative patients, TDs were also independent poor prognostic factors of OS (HR 2.522, 95%CI 1.267-5.021, P=0.008), DFS (HR 2.463, 95%CI 1.472-4.121, P=0.001) and DMFS (HR 2.338, 95%CI 1.286-4.251, P=0.005), but not LRFS (HR 2.109, 95%CI 0.701-6.343, P=0.184) (Table 1 a, b). Then three multivariate analyses showed that oN was not an independent prognostic factor, but n1N and n2N were independent poor prognostic factors of OS (n1N ， P=0.003; n2N, P=0.003), DFS (n1N ， P=0.001; n2N, P=0.001) and DMFS (n1N ， P<0.001; n2N, P<0.001) ， but not LRFS (n1N ， P=0.202; n2N, P=0.079) (Table 1c). And there were no differences in survival between TD(+)LN(-) and LN(+)TD(-) patients with the same counts of TDs and positive lymph nodes (all P>0.1).