Contrast-enhanced Mammography for Early Prediction of Neoadjuvant Therapy Response in Breast Cancer

See also article by Pires-Gonçalves et al.

Roberto Lo Gullo, MD, is a member of the Department of Radiology at Columbia University Vagelos College of Physicians and Surgeons, New York, NY. He is the director of International Research Fellows and Observership Program at Columbia University Irving Medical Center. His research interests focus on breast MRI and the application of artificial intelligence in breast imaging to develop imaging biomarkers for precision medicine. He is the recipient of two Research Trainee Prize awards from RSNA and a research fellow award from the New York Roentgen Society.

Janice Sung, MD, is the Associate Division Chief of Breast Imaging and an associate professor of radiology at Columbia University Irving Medical Center. Her research focuses on contrast-enhanced mammography and MRI biomarkers. Dr Sung serves on national guideline committees and has authored numerous peer-reviewed publications advancing breast imaging science.

Katja Pinker-Domenig, MD, is Chief of the Division of Breast Imaging for the Department of Radiology and professor of radiology at Columbia University Vagelos College of Physicians and Surgeons (VP&S) and adjunct professor at the Department of Radiology at the Medical University of Vienna, Vienna, Austria. She is an expert in translational and clinical breast imaging. Her research interests focus on functional breast imaging with MRI and CEM and the application of AI in oncologic imaging to develop imaging biomarkers for precision medicine.

Neoadjuvant therapy (NAT) has become a standard approach for treating locally advanced breast cancer, offering the dual benefits of downstaging tumors to enable breast-conserving surgery and providing an in vivo test of chemosensitivity and a surrogate marker of survival. NAT is also increasingly being considered for patients presenting with smaller tumors who have a clear indication for NAT at diagnosis, such as those with triple-negative breast cancer or human epidermal growth factor receptor 2 (HER2)–positive breast cancer. A key end point of NAT is pathologic complete response (pCR)—the absence of residual invasive cancer in the breast and lymph nodes at surgery—which is associated with improved long-term outcomes (1).

Early prediction of pCR during NAT is highly desirable, as it could enable clinicians to tailor treatment by escalating or switching therapy in nonresponders, thereby minimizing unnecessary toxicity. However, early response assessment is challenging using clinical breast examination and conventional breast imaging. This has spurred interest in advanced imaging modalities capable of providing both morphologic and functional information to better evaluate tumor response.

Contrast-enhanced MRI is the most sensitive imaging tool for breast cancer detection and assessment of lesion extent and is therefore the preferred modality for monitoring response to NAT. By visualizing tumor-associated neovascularity, MRI provides a tumor-specific assessment and has demonstrated superiority over clinical examination and conventional breast imaging in predicting treatment response and residual disease (1). However, MRI is costly, not universally accessible, and may be poorly tolerated by some patients due to factors such as claustrophobia. These limitations underscore the need for additional neovascularity-based tools that can match MRI’s performance in early response assessment while offering improved accessibility and patient tolerability.

Contrast-enhanced mammography (CEM) has emerged as a promising alternative imaging modality that could fulfill this role. CEM augments digital mammography with functional neovascularity-based imaging by using an intravenous iodinated contrast agent and dual-energy image acquisition. The result is a set of conventional mammographic images plus "recombined" images highlighting contrast material uptake in the breast, analogous to enhancement observed with MRI. In essence, CEM can depict tumor neovascularity using a standard mammography unit (2).

Given the ability of CEM to visualize both tumor morphology and neovascularity, the authors of a study in Radiology: Imaging Cancer hypothesized that CEM could be used to monitor breast cancer response to chemotherapy in real time, similar to MRI (3). In this study by Pires-Gonçalves et al, CEM was performed at baseline and after the first cycle of NAT to measure changes in tumor size and enhancement in 36 individuals with breast cancer selected from a tertiary cancer center. Two radiologists measured the longest dimension of the lesion on recombined images. A key metric was the CEM-derived change in lesion dimension (CLD), defined as the reduction in the tumor’s largest dimension on contrast-enhanced images, after the initial cycles of chemotherapy. The cutoff for CLD was 20.93%.

Based on assessment of CEM-derived measurements after the end of the first cycle of NAT, 11 participants (30.5%) exhibited CLD greater than or equal to 20.93% and 25 participants (69.5%) exhibited CLD of less 20.93%. A CLD greater than or equal to 20.93% was statistically significantly associated with pCR (P < .001). The sensitivity, specificity, positive predictive value, and negative predictive value for the CLD cutoff to predict pCR were 73%, 88%, 73%, and 88%, respectively; in comparison, RECIST criteria achieved sensitivity, specificity, positive predictive value, and negative predictive value of 45%, 88%, 62%, and 79%, respectively (3).

These findings suggest that early changes observed at CEM are strong predictors of treatment success. Patients who ultimately achieved pCR showed significantly greater reductions in tumor size on early-treatment CEM compared with those with residual disease at surgery. In particular, a large early decrease in lesion size at CEM served as an early imaging biomarker of pCR, specifically in hormone receptor (HR)–negative cancers, including triple-negative and HER2-positive subtypes, compared with HR-positive cancers (P = .001). These subtypes tend to respond more rapidly and completely to neoadjuvant chemotherapy, and accordingly, lesion size at CEM often diminishes dramatically in responders. In contrast, HR-positive/HER2-negative tumors, which less commonly achieve pCR, showed weaker association between early imaging response and final pathology, consistent with findings from previous studies (4).

From a clinical standpoint, early use of CEM during NAT could aid in stratification of patients. For example, an HR-negative tumor showing marked reduction in enhancement and size at CEM after a few treatment cycles is likely on track to achieve pCR, potentially supporting continuation of the current regimen or consideration of less extensive surgery if confirmed. Conversely, minimal early tumor reduction at CEM might prompt changes in treatment, such as switching therapies, opting for mastectomy, or introducing investigational agents in clinical trials. While response-adaptive strategies continue to evolve, the findings in this study by Pires-Gonçalves et al provide important evidence supporting CEM as a tool for early response prediction, paralleling prior MRI-based approaches. Although the concept of midtreatment imaging to predict pCR is well established, MRI studies have led the way. However, this work shows that CEM, a simpler and more accessible modality, can offer similar early treatment feedback. Future studies could explore the integration of artificial intelligence and machine learning to further enhance and refine predictive models for NAT response, as has been done with MRI (5).

A natural question is how CEM compares with the MRI reference standard for monitoring NAT. Fundamentally, both modalities assess the same phenomenon—tumor enhancement following contrast agent uptake—but using different imaging techniques. Several studies have directly compared CEM and breast MRI in the NAT setting, with encouraging results for CEM. Tumor size measurements at CEM have shown strong correlation with MRI measurements at all treatment time points. In a prospective study of women undergoing NAT, Iotti et al (6) reported nearly identical accuracy between CEM and MRI in distinguishing responders from nonresponders, with both modalities agreeing on response categorization for 45 of 46 patients. Notably, in that study CEM demonstrated higher sensitivity and specificity than MRI for assessing complete response. Other recent prospective trials have similarly shown that CEM performs comparably to MRI in therapy response assessment, with no significant differences in predicting pCR (7,8).

Beyond its diagnostic performance, CEM offers practical advantages that support its broader adoption for therapy monitoring. CEM is performed on standard mammography equipment, which is widely available and less expensive than MRI (9). MRI requires dedicated scanners and specialized breast coils and has limited availability, particularly in resource-constrained settings where routine access to breast MRI for all patients undergoing NAT may not be feasible. Other advantages of CEM are improved patient comfort and tolerability, as some patients cannot tolerate MRI due to claustrophobia, back pain when lying flat, or contraindications such as metallic implants or other MRI-incompatible devices.

While early results supporting use of CEM for NAT response prediction are promising, larger studies with longer-term outcomes are needed to further clarify its clinical role. In this study, the Youden index was used to determine the optimal threshold to identify pCR. However, additional studies with larger datasets may help identify the precise cutoff of CLD that best predicts pCR. Several important questions remain, including how predictive values might vary across different chemotherapy regimens or newer treatments like immunotherapy. Additionally, integrating CEM findings with other emerging biomarkers, such as tumor genomics or blood-based markers like circulating tumor DNA, could further improve response prediction. Future research may also explore radiomics and advanced quantitative analysis of CEM beyond lesion diameter to refine predictive models. Initial radiomics studies suggest that the degree of contrast enhancement reduction at CEM may carry prognostic value (10).

In summary, this study by Pires-Gonçalves et al provides evidence that CEM can serve as an early indicator of neoadjuvant treatment efficacy, approaching the performance of MRI while offering practical advantages in cost, speed, accessibility, and patient comfort. These findings have important implications for the broader radiology and oncology communities, suggesting that a widely available imaging modality could be used not only for diagnosis and staging but also to guide treatment decisions. As we move toward personalizing breast cancer treatment, tools like CEM will play an essential role. This study brings us closer to a future where early response monitoring is accessible to all patients undergoing NAT, allowing care teams to adjust treatment strategies as needed and optimize patient outcomes.