Background
Triple-negative breast cancer (TNBC) is an aggressive subtype of breast cancer with more frequent and earlier relapses compared to other subtypes. Achieving a pathologic complete response (pCR) after neoadjuvant chemotherapy is strongly associated with improved survival. Dynamic contrast-enhanced (DCE) breast MRI is the most accurate imaging tool for assessing treatment response, but its performance after neoadjuvant chemoimmunotherapy (NACI) remains unclear. With rising pCR rates from the KEYNOTE-522 trial, accurately identifying responders on MRI could open the door to less invasive surgical strategies.
Question
Can post-treatment DCE breast MRI, alone or combined with clinical and radiomic features, accurately predict pCR in TNBC patients after NACI?
Design
Prospective, multicenter cohort with external validation (Aug 2021–Jun 2024).
Participants
259 women with biopsy-proven stage II–III TNBC, treated with the KEYNOTE-522 NACI regimen. All underwent post-treatment DCE breast MRI before surgery.
Methods
Post-treatment DCE breast MRI scans performed using 1.5-T scanners across multiple vendors.
Images independently interpreted by radiologists; radiologic complete response (rCR) defined as absence of enhancement in the tumor bed.
Surgical specimens evaluated for pCR (ypT0/is ypN0) and classified by the Residual Cancer Burden (RCB I–III) system.
Results
In the external test set (n=84, pCR rate 63%):
rCR achieved AUC 0.83, with PPV 90%.
False positives occurred in 10% of rCR cases, all with minimal residual disease (RCB I–II).
A three-factor model (rCR + baseline Ki-67 >30% + node-negative status):
AUC 0.88.
False-positive rate reduced to 3.4% (only 1 case, RCB I).
In non-rCR cases, adding radiomic features from residual enhancement reduced the false-omission rate from 37% to 20%, improving accuracy further.
Recent Literature
Mohamed et al., 2024: Multiparametric MRI-based radiomic model (baseline + after 2 NAC cycles) achieved AUC 0.905 (training) and 0.802 (test).
Janssen et al., 2024: Meta-analysis of 1,658 TNBC patients showed PPV 73–75%, compared to 90% in Ramtohul et al., possibly reflecting higher pCR rates with NACI (63% vs 34%).
Teaching Points
In TNBC after NACI, post-treatment breast MRI showing rCR can yield very high PPV, especially when combined with node-negative status and Ki-67 >30%, supporting selection for potential surgery omission trials.
Residual enhancement after therapy is not always viable cancer. Radiomics can help reduce false positives by distinguishing fibrosis, inflammation, or necrosis from true tumor.
- Ramtohul T, Lollivier D, Spriet J, Jin M, Djerroudi L, Gaillard T, Bonneau C, Loirat D, Bello-Roufai D, Kirova Y, Loap P, Malhaire C, Vincent Salomon A, Bidard FC, Tardivon A, Maillez A, Wallaert L, Ceugnart L, Nhy C, Cabel L. Posttreatment MRI to Predict Pathologic Complete Response of Triple-Negative Breast Cancer to Neoadjuvant Chemoimmunotherapy. Radiology. 2025 Jul;316(1):e243824. doi: 10.1148/radiol.243824. PMID: 40590696.
- Onishi N. Breast MRI for Assessing Triple-Negative Breast Cancer after Neoadjuvant Chemoimmunotherapy. Radiology. 2025 Jul;316(1):e251503. doi: 10.1148/radiol.251503. PMID: 40590697; PMCID: PMC12314758.
- Mohamed RM, Panthi B, Adrada BE, Boge M, Candelaria RP, Chen H, Guirguis MS, Hunt KK, Huo L, Hwang KP, Korkut A, Litton JK, Moseley TW, Pashapoor S, Patel MM, Reed B, Scoggins ME, Son JB, Thompson A, Tripathy D, Valero V, Wei P, White J, Whitman GJ, Xu Z, Yang W, Yam C, Ma J, Rauch GM. Multiparametric MRI-based radiomic models for early prediction of response to neoadjuvant systemic therapy in triple-negative breast cancer. Sci Rep. 2024 Jul 12;14(1):16073. doi: 10.1038/s41598-024-66220-9. PMID: 38992094; PMCID: PMC11239818.
- Janssen LM, den Dekker BM, Gilhuijs KGA, van Diest PJ, van der Wall E, Elias SG. MRI to assess response after neoadjuvant chemotherapy in breast cancer subtypes: a systematic review and meta-analysis. NPJ Breast Cancer. 2022 Sep 19;8(1):107. doi: 10.1038/s41523-022-00475-1. PMID: 36123365; PMCID: PMC9485124.
