A Pretreatment Model Integrating IDEAL-IQ–Derived Peritumoral Adipose Metrics for Predicting Pathological Complete Response in Breast Cancer

Abstract Background Peritumoral adipose tissue may reflect tumor–microenvironment interactions relevant to treatment response. This study evaluated whether IDEAL-IQ-derived fat-fraction metrics provide additional pretreatment information for pathological complete response (pCR) after neoadjuvant chemotherapy (NAC). Methods A total of 73 patients with breast cancer who underwent pretreatment MRI with IDEAL-IQ were included. Peritumoral fat fraction (FFp), contralateral fat fraction (FFh), and their difference (dFF = FFh − FFp) were measured. Clinical, IDEAL-IQ, and combined logistic regression models were assessed using receiver operating characteristic analysis, calibration, decision-curve analysis, DeLong tests, and 1,000 bootstrap resamples. Results Among the 73 patients, 42 (57.5%) achieved pCR. FFp was significantly lower than FFh overall. Compared with the non-pCR group, patients who achieved pCR had higher FFp and lower dFF. In the combined model, HER2 positivity and lower dFF were independently associated with pCR. The combined model had an AUC of 0.911, compared with 0.856 for the clinical model and 0.719 for the dFF-only model. The AUC improvement over the clinical model was not statistically significant (difference, 0.055; P = 0.061), whereas the improvement over the dFF-only model was significant (difference, 0.192; P = 0.002). The optimism-corrected AUC was 0.890, with a Brier score of 0.140 and calibration slope of 0.824. Conclusions IDEAL-IQ–derived peritumoral adipose metrics, particularly dFF, were associated with pCR and may provide information complementary to a parsimonious clinical model. However, the increase in AUC after adding dFF was not statistically significant, and the incremental value of dFF requires external validation.