Contrast-enhanced MRI after neoadjuvant chemotherapy of breast cancer: lesion-to-background parenchymal signal enhancement ratio for discriminating pathological complete response from minimal residual tumour

To retrospectively investigate whether the lesion-to-background parenchymal signal enhancement ratio (SER) on breast MRI can distinguish pathological complete response (pCR) from minimal residual cancer following neoadjuvant chemotherapy (NAT), and compare its performance with the conventional criterion. 216 breast cancer patients who had undergone NAT and MRI and achieved pCR or minimal residual cancer on surgical histopathology were included. Clinical-pathological features, SER and lesion size on MR images were analysed. Multivariate logistic regression, ROC curve and McNemar's test were performed. SER on early-phase MR images was independently associated with pCR (odds ratio [OR], 0.286 [95% CI: 0.113-0.725], p = .008 for Reader 1; OR, 0.306 [95% CI: 0.111-0.841], p = .022 for Reader 2). Compared with the conventional criterion, SER 1.6 increased AUC (0.585-0.599 vs. 0.709-0.771, p=.001-.033) and specificity (21.9-27.4% vs. 80.8-86.3%, p <.001) in identifying pCR. SER 1.6 and/or size 0.2 cm criterion showed the highest specificity of 90.4%. SER on early-phase MR images was independently associated with pCR, and showed improved AUC and specificity compared to the conventional criterion. The combined criterion of SER and size could be used to select candidates to avoid surgery in a future study. Compared with conventional criterion, SER 1.6 criterion increased AUC and specificity. Simple measurement of signal intensity could differentiate pCR from minimal residual cancer. SER 1.6 and/or size0.2cm criterion showed the highest specificity of 90.4 %. The combined criterion could be used for a study to avoid surgery.