Posttreatment MRI to Predict Pathologic Complete Response of Triple-Negative Breast Cancer to Neoadjuvant Chemoimmunotherapy
Menée sur 175 et 84 patientes ayant reçu une chimio-immunothérapie néoadjuvante pour un cancer du sein triple négatif (âge moyen : 49 ans et 52 ans), cette étude multicentrique évalue la précision d'une IRM, réalisée après le traitement néoadjuvant, pour prédire la réponse pathologique complète
Abstract : Posttreatment MRI accurately predicted pathologic complete response following neoadjuvant chemoimmunotherapy in participants with early triple-negative breast cancer, demonstrating a low false-discovery rate.
Background : Neoadjuvant chemoimmunotherapy (NACI) has substantially improved pathologic complete response (pCR) rates in early triple-negative breast cancer (TNBC). However, the predictive accuracy of posttreatment MRI remains unexplored.
Purpose : To assess the performance of posttreatment MRI in the prediction of pCR in participants with TNBC treated with NACI.
Materials and Methods : In this prospective multicenter study (August 2021–June 2024), women with early TNBC were recruited from three centers (training set: Institut Curie; test set: Institut Godinot and Institut Oscar Lambret). Post-NACI dynamic contrast-enhanced MRI scans from multiple vendors were analyzed. Radiologic complete response (rCR)—defined as no enhancement in the tumor bed—was evaluated for predicting pCR. A multivariable logistic regression model incorporating rCR, nodal involvement, and Ki-67 index was developed and externally validated. In cases with residual enhancement (non-rCR), a radiomic score using shape and first-order features was tested.
Results : A total of 175 women were included in the training set (mean age, 49 years ± 11 [SD]) and 84 women in the external test set (mean age, 52 years ± 12). The rCR at MRI was predictive of pCR, with an area under the receiver operating characteristic curve (AUC) of 0.83 (95% CI: 0.75, 0.92). The combined model (rCR + nodal status + Ki-67) yielded an AUC of 0.88 (95% CI: 0.81, 0.96) in the test set. In node-negative patients with Ki-67 greater than 30%, the rCR false-discovery rate (ie, the proportion of rCR cases that were actually non-pCR or residual disease missed at breast MRI) was 3.6% (two of 56) in the training set and 3.5% (one of 29) in the test set; all cancers were limited to residual cancer burden I. In non-rCR cases, a model incorporating the radiomics score and lesion count achieved an AUC of 0.80 (95% CI: 0.69, 0.90).
Conclusion : Posttreatment rCR at MRI demonstrated strong predictive value for pCR in early TNBC following NACI.
Radiology , résumé, 2025