Subtype-Specific Metabolic Patterns in Staging Breast Cancer: Insights from Conventional and Parametric PET Radiomics.

Background: Breast cancer (BC) is a biologically heterogeneous disease, with molecular subtypes differing fundamentally in terms of aggressiveness, metabolic behavior, and therapeutic response. Conventional 18F-FDG PET provides semi-quantitative assessment of glucose-analogue uptake using the standardized uptake value (SUV), whereas dynamic PET with Patlak analysis can estimate kinetic parameters such as the net influx rate constant (Ki) and the volume of distribution (Vd). However, the interpretation of these parameters in breast tissue remains insufficiently defined, and data across BC subtypes are limited. This proof-of-concept study explored the feasibility of deriving preliminary reference ranges for non-pathological breast tissue and describing subtype-specific patterns of kinetic parameters. Methods: Six breast cancer patients affected by Luminal (n = 4) or triple negative BC (TNBC, n = 3) and six control patients underwent dynamic 18F-FDG PET/CT. Patlak kinetic analysis generated parametric Ki and Vd maps. Lesion-level first-order radiomics features were extracted for each breast lesion. Non-lesional breast volumes (contralateral breast and breast excluding disease) were characterized to derive exploratory physiological reference ranges. Results: In control whole-breast tissue, median values were SUV 0.28 g/mL, Ki 0.056 mL/min/100 mL, and Vd ~9%. Non-lesional breast tissue in BC remained statistically concordant with controls. Lesion analysis suggested a possible metabolic gradient across subtypes: Luminal B with low proliferation activity exhibited lowest Ki (0.34 mL/min/100 mL), and highest Vd (~39%); Luminal A showed intermediate Ki (1.93 mL/min/100 mL) and Vd (~27%) values; TNBC showed higher Ki (3.02 mL/min/100 mL) with a greater proportion of zero-valued Vd voxels. Conclusions: This exploratory study suggests that parametric maps may be feasible for describing physiological breast tissue kinetic ranges and for exploring subtype-associated metabolic differences in BC. The observed Ki and Vd patterns may reflect differences in glucose phosphorylation and tracer distribution across tumor phenotypes, but the biological meaning of Patlak-derived parameters requires further methodological and histopathological validation. Larger prospective studies are needed before these parameters can be considered clinically applicable imaging biomarkers.