18F-FDG PET/CT for prediction of response in breast cancer

Abstract

Breast cancer remains one of the most heterogeneous malignancies, with marked variability in biology, therapeutic sensitivity, and clinical outcomes. As treatment strategies evolve toward individualized approaches, early and accurate assessment of response has become critical for optimizing outcomes and minimizing toxicity. Recent Findings: ¹⁸F-FDG PET/CT provides a biologically grounded, non-invasive measure of tumour metabolism, heterogeneity, and early treatment adaptation. Baseline metrics such as SUVmax, metabolic tumour volume (MTV), and total lesion glycolysis (TLG)—reflect proliferative drive and aggressiveness, while early changes (ΔSUV, ΔMTV/TLG after 1–2 cycles) predict pathological complete response (pCR) with high negative predictive value. PET-derived nomograms integrating clinical, molecular, and metabolic data outperform clinicopathologic models alone. Radiomic and artificial-intelligence (AI) analyses further refine prediction by quantifying spatial heterogeneity and enabling subtype-specific modelling. Joint EANM/SNMMI guidelines and NCCN recommendations increasingly endorse ¹⁸F-FDG PET/CT for staging and response monitoring in high-risk or locally advanced disease. ¹⁸F-FDG PET/CT has transitioned from staging to precision-response prediction, particularly in HER2-positive and triple-negative breast cancer. Integration into AI driven nomograms supports adaptive, patient-tailored decisions that minimize toxicity and cost while maximizing benefit. Prospective multicentre validation aligned with EANM/SNMMI/NCCN guidance will consolidate PET’s role in adaptive oncology.

HIGHLIGHTS • ¹⁸F-FDG PET/CT enables early, non-invasive prediction of therapy response in breast cancer. • Metabolic parameters (SUVmax, MTV, TLG) and ΔSUV predict pCR and survival across subtypes. • Radiomics and AI enhance prediction by quantifying tumour heterogeneity and resistance. • PET-based nomograms outperform clinicopathologic models for pCR and recurrence prediction.