TY - JOUR T1 - <strong>Investigation of 18F-FDG-PET radiomic features in satellite breast tissue in breast cancer patients</strong> JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 1242 LP - 1242 VL - 60 IS - supplement 1 AU - Sarah Boughdad AU - Fanny Orlhac AU - Christophe Nioche AU - Anne-Sophie DIRAND AU - Laurence Champion AU - Irene Buvat Y1 - 2019/05/01 UR - http://jnm.snmjournals.org/content/60/supplement_1/1242.abstract N2 - 1242Introduction: A few recent studies (eg Braman et al Breast Cancer Res 2017) suggest that breast tumor close environment might contain useful prognostic information and might help better predict patient outcome. We therefore investigated FDG PET radiomic features (RF) extracted from a spherical region of non-pathological breast tissue in the breast containing the tumor. In particular, we studied the ability of these RF to predict pCR and survival in all breast tumors and as a function of tumor subtype in breast cancer (BC) patients undergoing neoadjuvant chemotherapy (NAC). Methods: BC patients with Luminal B HER2-, Luminal B HER2+ and Triple-Negative (TN) tumors who underwent 18F-FDG PET/CT at initial staging before NAC consisting of taxanes, anthracyclines and trastuzumab in tumors expressing HER2 were studied. FDG PET/CT was performed 60 to 80 min after injection of 3-3.5 MBq/kg of F18-FDG if capillary blood glucose was below 11 mMol/L and after 6 hours of fasting. A 6 mL spherical volume of interest S-VOI was drawn in the breast containing the tumor but away from the breast tumor and outside any abnormal uptake seen on visual analysis. The breast tumor region T-VOI ≥ 3 mL was also automatically segmented (40% of SUVmax). RF were calculated within T-VOI and S-VOI including SUVs (mean, max, peak), Metabolic Tumor Volume (MTV), Total Lesion Glycolysis (TLG), 4 histogram-based indices (HBI) and 31 texture indices (TI). Pearson correlation tests were used to assess the relationship between RF calculated in T-VOI and S-VOI. T-tests for independent samples with Benjamini-Hochberg procedure for False Detection Rate correction were used to compare RF extracted from S-VOI as a function of complete pathological response (pCR), event free survival (EFS) and overall survival (OS). Kaplan-Meier method (KM) was used to characterize the association between RF and EFS or OS. Results: A population of 107 patients with 117 tumors (10 patients had multifocal BC) was analyzed consisting of 50 Luminal B HER2-, 25 Luminal B HER2+ and 42 TN tumors. Median follow-up in the whole cohort was 38.7 months ±14.7 for EFS and 41.6 months ±13.8 for OS. The Pearson correlation coefficients between RF extracted from T-VOI and S-VOI were always &amp;#8804;0.15. In S-VOI, 16 RF were significantly different (p&lt;0.05) between the 3 BC subtypes including SUVmean, SUVmax, SUVpeak, TLG, Homogeneity, Short-Run Emphasis (SRE) and High Grey-level Zone Emphasis (Anova, p&lt;0.05) but no HBI. No significant association between S-VOI RF and pCR was found when analyzing all tumors together or each subtype independently. Similarly, no significant association between RF and EFS was found when pooling all patients together, in Luminal B HER2+ patients and in Luminal B HER2- patients, but 4 TI were related to EFS in TN patients. For instance, Entropy&lt;2.7 in S-VOI was associated with lower EFS than Entropy ≥2.7 (KM log rank test: p&lt;0.02) in TN patients. For OS prediction, several RF extracted from S-VOI regardless BC subtype yielded p&lt;0.05 on t-tests including SUVmax, Homogeneity, SRE and Long-Run Emphasis (LRE). In Luminal B HER2- tumors, SUVmax &gt; 1.35 was associated with higher survival (KM log rank test: p=0.05) whereas in TN tumors only Contrast predicted OS on t-test (p&lt;0.05) with a value above 0.017 more frequent in patients with better OS (KM log rank test: p=0.1). OS in BC patients with Luminal B HER2+ tumors was not studied as no death was reported during the follow up. Conclusions: RF values in a region satellite to the breast tumor are not correlated with the tumor RF but some RF extracted from this region appear to be associated with the BC molecular subtype and patient survival. These findings suggest that in BC patients, 18F-FDG PET radiomic analysis should not focus on the tumor only but also on surrounding tissue. Our results also underline the need for accounting for the subtype of BC in radiomic analysis. ER -