Comparative analysis of preoperative contrast-enhanced cone beam breast CT (CE-CBBCT) and MRI for differentiating pathological complete response from minimal residual disease in breast cancer.

Abstract

Rationale and objectives: To evaluate the performance of contrast-enhanced cone-beam breast CT (CE-CBBCT) using visual, quantitative, and combined models in distinguishing pathological complete response (pCR) from minimal residual disease (MRD) after neoadjuvant therapy (NAT), and to compare its diagnostic efficacy with MRI.

Materials and methods: This study enrolled 65 female patients who underwent both CE-CBBCT and MRI after NAT and were classified as having either pCR or MRD. Univariate and multivariate logistic regression analyses were performed to identify independent visual and quantitative features from CE-CBBCT and MRI associated with pCR. Model performance was assessed and compared using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), DeLong's test, and McNemar's test. The bootstrap method was employed to assess the stability of each model.

Results: Multivariate analysis identified fine and branched calcification morphology on CE-CBBCT (visual model: odds ratio [OR] = 4.500; combined model: OR = 4.527), enhanced degree (ΔHU, quantitative model: OR = 1.036; combined model: OR = 1.035), radiographic complete response (rCR; visual model: OR = 0.103; combined model: OR = 0.097), and delayed-phase MRI enhancement ratio (ER_dpMRI; quantitative model: OR = 5.048; combined model: OR = 5.583) as independent predictors of pCR. The CE-CBBCT combined model demonstrated a significantly higher AUC than the visual model (0.805 vs. 0.698, p = 0.017) and performed comparably to the MRI combined model (0.805 vs. 0.819, p = 0.811). In the HER2-enriched subgroup, the CE-CBBCT combined model exhibited higher specificity than MRI (0.857 vs. 0.714, p = 0.011) for identifying pCR.

Conclusion: The combination of calcification morphology and ΔHU on CE-CBBCT improved accuracy in discriminating pCR from MRD, achieving performance comparable to MRI. Notably, the CE-CBBCT combined model showed superior specificity to MRI within the HER2-enriched subgroup, suggesting its potential utility in reducing overtreatment in this patient population.

Clinical trial number: Not applicable.