e12567 Background: ER+/HER2- phenotype represents 70% of eBC and is itself a heterogenic entity in which neoadjuvant chemotherapy (NAC) can be indicated. Early change in 18F-FDG uptake on PET/CT, using standardized uptake value (SUV) parameter, has been shown as a predictive biomarker of response to NAC. However, this parameter can be defaulted in ER+/HER2- BC. The objective of this work was to evaluate the correlation of textural parameters of initial 18-FDG PET/CT with histologic response in ER+/HER2- eBC. Methods: All patients receiving NAC for ER+/HER2- eBC, with 18F-FDG PET/CT prior NAC at the Integrated Center for Oncology Pays de Loire were included between 2012 and 2023. Twenty-nine metabolic and volumetric parameters and 42 textural features (DOSISOFT software, Planet Onco v2.0) were collected. Pathological response was assessed for each breast tumoral lesion in case of multifocal disease. Results: A total of 161 patients with 179 lesions were included. Medium age was 51 years old. 89% and 10% of patients presented a cT1 or cT2 tumoral lesion, 78% with lymph node involvement, 44% with a grade 3 and 46% with a HER2 low status. 21% of patients achieved a pathological complete response in breast (ypT0) and 11% a complete histologic response (ypT0N0). Examinations were performed using three types of PET/CT systems (GE Healthcare and Philips Healthcare). In order to harmonize radiomics data Combat method was used. Thirteen semi-quantitative (MTV and SUVmax, SUVmean, SUVmean(EVPc), SUVpeak, TLG for weight-normalized and lean-body-mass-normalized, respectively) and 6 textural parameters (entropy, homogeneity, SRE, LRE, LGZE, and HGZE) were retained. For all radiomics features used in this study, no statistical difference was found between ypT0 and non-ypT0 patients (p ≥ 0.05). Conclusions: Selected textural parameters were not associated with histologic response in our cohort of ER+/HER2- eBC. However, many other textural features, not retained in this study because influenced by acquisition parameters and segmentation methods, could be taken into account. An ongoing preliminary work (as resampling, normalization) is needed to improve their robustness before evaluating their interest in predicting histologic response to NAC.
Read full abstract