PT - JOURNAL ARTICLE AU - Pedrero Piedras, Roberto AU - García-Pérez, Francisco Osvaldo AU - SINISTERRA SOLIS, FABIO ANDRÉS AU - Quiñones-Capistran, Christian Aníbal AU - Pitalua Cortes, Quetzali AU - Soldevilla-Gallardo, Irma AU - Michel Sanchez, Emiliano AU - Bargallo Rocha, Juan AU - Cabrera Galeana, Paula AU - Porras Reyes, Fanny TI - <strong>18F-FDG PET/CT radiomic features in the evaluation of molecular breast cancer subtypes : a potential imaging biomarker of intra-tumoral heterogeneity</strong> DP - 2022 Aug 01 TA - Journal of Nuclear Medicine PG - 4009--4009 VI - 63 IP - supplement 2 4099 - http://jnm.snmjournals.org/content/63/supplement_2/4009.short 4100 - http://jnm.snmjournals.org/content/63/supplement_2/4009.full SO - J Nucl Med2022 Aug 01; 63 AB - 4009 Introduction: Radiomic analysis is the method in which quantitative features can be extracted from radiological images to generate a set of relevant data for the characterization of biological tissues. In recent years, it has been used in multiple neoplastic models, showing potential in intra-tumoral heterogeneity evaluation. In breast cancer, the most frequent malignancy worldwide, the textural analysis has been studied with MRI, ultrasound and mammography and has shown its usefulness characterizing correctly molecular phenotypes and predicting a worst prognostic, however there is limited information about radiomics with PET/CT and its potential as a noninvasive biomarker identifying tumoral heterogeneity.Methods: We analyzed retrospectively 100 women from 2013 to 2021 diagnosed in our institution with locally advanced breast cancer, whom had been staged initially with [18]F-FDG PET CT. The textural and metabolic analysis was performed with LIFEx 7.1.0 freeware. 3D-Regions of interest (3D-ROI) of the same volume were drawn in neoplastic and healthy contralateral breast tissue to stablish comparison. For each 3D-ROI, 73 features were calculated, including metabolic features shape, first-order statistics, gray- level co-occurrence matrix (GLCM), gray-level run-length matrix (GLRLM), gray-level size-zone matrix (GLZM) and gray-level dependence matrix (NGLDM). The values obtained were compared between molecular subtypes running normality tests, descriptive statistics, T test and graphics using the IBM SPSS software, version 26. The statistical significance was established at p&lt;0.05.Results: The average age of our population was 51.6 ± 11.36 years. Between healthy and neoplastic tissue existed statistically significant differences in morphological, first and second order variables. In the molecular subtype analysis, we found the following: comparing Luminal A and B phenotypes, the radiomic conventional and metabolic features showing statistically significant differences were Conventional TLG (p=0.025), Discretized TLG (p=0.025), SUVbwstd (p=0.047), Discretized SUVpeak sphere (p=0.046), Shape volume (p=0.034), Shape compacity (p=0.023), GLRM GLNU (p=0.045), NGLDM Coarseness (p=0.007) and GLZLM SZLGE (p=0.002). For Luminal A and HER-2 enriched groups, there were differences statistically significant in Conventional SUV (p&lt;0.05), Shape volume (p=0.013), Shape compacity (p=0.01), GLCM Contrast variance (p=0.002), GLCM entropy (p=0.012), GLCM dissimilarity (p=0.001), GLRLM RP (p=0.008), NGLDM Contrast (p=0.02) and GLZM SZE (p=0.04). Finally, between Luminal A and triple negative phenotypes, the characteristics showing statistically significant differences were Conventional SUVbw, Discretized SUVbw, Shape volume, GLCM, GLZLM GLU and SZHGE (all of them with p&lt;0.05).Conclusions: Our study shows that certain radiomic characteristics derived from textural analysis are very typical of certain molecular subtypes, this finding was observed more strongly in the group of luminal A versus triple negatives, these findings could imply that in patients whose radiomic characteristics are discordant with histological information, additional examinations are necessary to rule out the coexistence of other phenotypes, the same principle can be applied in the scenario of recurrent disease with suspected changes in receptor expression and this, aditionally the advent of new radiotracers could be the future of next generation molecular imaging.