Optimizing abbreviated breast MRI protocols for the early assessment of neoadjuvant chemotherapy response in breast cancer

Introduction

Breast magnetic resonance imaging (MRI) is the most accurate imaging modality for assessing breast cancer response to neoadjuvant chemotherapy (NAC) (1,2). In clinical practice, MRI is widely used to evaluate breast tumor response during and after NAC, straining imaging resources and increasing the interpretive workload of radiologists (3). However, limited MRI access, particularly in public healthcare systems, can delay disease management and restrict its standardized use.

Abbreviated MRI protocols, which use fewer or shorter imaging sequences than full-protocol MRI examinations, represent one approach to increasing MRI availability (4,5). A previous study investigated the use of an abbreviated MRI protocol, comprising T2-weighted, pre-contrast T1-weighted, and a single post-contrast T1-weighted sequence, for evaluating treatment response after the completion of NAC before surgery (6). The diagnostic accuracy of this protocol for predicting pathological complete response (pCR) was comparable to that of the full-protocol MRI. Other studies have drawn similar conclusions (7,8). These promising results provide preliminary evidence of the feasibility of abbreviated MRI in the preoperative evaluation of NAC in breast cancer. However, the application of abbreviated MRI protocols to the early assessment of NAC response in breast cancer is more complex and uncertain. The Response Evaluation Criteria in Solid Tumors version 1.1 primarily relies on changes in tumor size to evaluate treatment response, which may not fully capture the early biological effects of therapy. Multiparametric MRI approaches, incorporating both qualitative and quantitative information from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and diffusion-weighted imaging (DWI), are being explored for early NAC response assessment (9-11).

Among imaging modalities, DCE-MRI has the highest accuracy for evaluating tumor response to NAC (12). Both changes in tumor dimensions and enhancement kinetics can be assessed with DCE-MRI (13,14). The time-intensity curve, a commonly used hemodynamic indicator, is relatively straightforward to apply, but it requires acquisition of the full enhancement phase and may not adequately capture the spatial heterogeneity of tumor perfusion (15). Recent research has reported that early-phase parameters, such as the signal enhancement ratio (SER), are associated with histologic grade, proliferation status, molecular subtype, and pCR, suggesting that the acquisition phase can be shortened (16,17). DWI measures the mobility of water molecules in vivo and is sensitive to identifying longitudinal microstructural changes in diffusivity during NAC (18). Apparent diffusion coefficient (ADC) maps derived from multi-b-value acquisitions have shown particular promise, with longitudinal changes in ADC (ΔADC) repeatedly linked to early tumor response (19,20). However, it remains uncertain which pulse sequences and derived biomarkers should be retained in abbreviated MRI protocols to ensure the accurate and reliable early assessment of NAC response. Given the biological heterogeneity and subtype-dependent responses to NAC, imaging biomarkers may perform differently across molecular subtypes and therapies (21); therefore, it is important to determine whether an abbreviated MRI strategy maintains consistent diagnostic performance across subtypes.

This study aimed to evaluate the diagnostic performance of different abbreviated MRI protocols for the early assessment of NAC response in breast cancer after two treatment cycles. It also sought to identify an optimal protocol that balances diagnostic performance and efficiency, and to evaluate its robustness in subgroup analyses. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-2066/rc).

Methods

Study population

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethical Committee of Guangzhou First People’s Hospital (No. S-2023-083-02). All participating hospitals were informed of and agreed to the study. The requirement of individual consent for this retrospective analysis was waived.

Patients were retrospectively identified from the electronic medical record systems at Guangzhou First People’s Hospital (Center 1), The Third People’s Hospital of Honghe Hani and Yi Autonomous Prefecture (Center 2), and the Cancer Hospital, Chinese Academy of Medical Sciences in Shenzhen (Center 3). Consecutive patients with pathologically confirmed primary invasive breast cancer who received standard NAC and underwent breast MRI at both the baseline and after two cycles of NAC were included in the study. The enrollment period was from March 2018 to December 2023 at center 1, and from March 2021 to March 2024 at centers 2 and 3. Patients were excluded from the study if they did not undergo full-protocol MRI examinations at the baseline and/or after either of the two cycles of NAC, did not undergo surgery or had incomplete postoperative pathology, had poor-quality MRI scans, or had missing clinical information. Ultimately, 169 patients from center 1 comprised the primary cohort, while 89 from center 2 and 101 from center 3 served as external validation cohort 1 (EVC1) and external validation cohort 2 (EVC2), respectively. The study flowchart is shown in Figure 1.

Figure 1 Flowchart of patient inclusion and exclusion in the study. Center 1: Guangzhou First People’s Hospital; Center 2: The Third People’s Hospital of Honghe Hani and Yi Autonomous Prefecture; Center 3: Cancer Hospital, Chinese Academy of Medical Sciences in Shenzhen. MR, magnetic resonance; MRI, magnetic resonance imaging; NAC, neoadjuvant chemotherapy; pCR, pathological complete response.

All participants received six or eight standard cycles of NAC before surgery according to the National Comprehensive Cancer Network guidelines (22). Patients received a taxane- or anthracycline-based regimen. For human epidermal growth factor receptor 2 (HER2)-positive tumors, anti-HER2 targeted agents were added to the chemotherapy regimen.

Histopathologic evaluation

Information from biopsy specimens obtained before NAC included the expression of estrogen receptor (ER), progesterone receptor (PR), HER2, and Ki-67 status. ER and PR positivity were defined as the detection of ≥1% positively stained invasive tumor cells; patients with either ER or PR positivity were classified as hormone receptor (HR) positive. HER2 positivity was defined as a score of 3+ or 2+ with positive fluorescent in situ hybridization amplification. Tumor subtypes were categorized as HR-positive/HER2-negative, HER2-positive, and triple-negative breast cancer (TNBC). Patients were categorized as having pCR or no pCR based on post-surgical histopathologic examination findings. In this study, pCR was defined as the absence of residual invasive cancer (ypT0/is) on microscopic evaluation of the resected breast specimen.

MRI examinations and protocols

Breast MRI examinations were performed with patients in the prone position using either a 1.5-T scanner (uMR 560, United Imaging, center 1) or 3.0-T scanners (Ingenia, Philips, center 2; Discovery MR 750W, GE, center 3). The full protocol included axial fat-saturated T2WI, axial turbo spin echo T1-weighted imaging, axial DWI (b=0 and 800 s/mm²) with ADC map reconstruction, and axial fat-suppressed DCE T1-weighted imaging (with fat-saturated sequences in center 1 and Dixon sequences in centers 2 and 3). For DCE, a pre-contrast T1-weighted scan was acquired, followed by the intravenous injection of gadolinium chelate (GD-DTPA, 0.1 mmol/kg). After contrast administration, a brief bolus-timing step was used to initiate the first post-contrast acquisition. Subsequently, center 1 acquired five consecutive post-contrast phases (~60 s per phase), center 2 acquired six phases (~60 s per phase), and center 3 acquired eight phases (~50–60 s per phase). Detailed sequence parameters for each center are summarized in Table S1.

The abbreviated MRI protocols (Protocols 1–6) were retrospectively derived from the standard full protocol, which included T2WI, DCE-MRI with multiple post-contrast phases, and DWI with corresponding ADC maps. T2WI was retained in all abbreviated protocols as a routine non-contrast morphologic reference in the NAC setting for lesion-bed localization and quality control. The abbreviated protocols were then constructed by varying the post-contrast DCE phase and the inclusion of DWI. In the primary cohort, three representative post-contrast DCE phases were selected based on acquisition timing: approximately 20 (C1), 90 (C2), and 270 s (C5) after contrast injection. Protocols 1–3 included T2WI together with a single DCE phase: Protocol 1 with C1, Protocol 2 with C2, and Protocol 3 with C5. Protocols 4–6 were constructed in the same way but also incorporated DWI; that is, Protocol 4 with T2WI + C1 + DWI, Protocol 5 with T2WI + C2 + DWI, and Protocol 6 with T2WI + C5 + DWI. For external validation, the same representative phases (C1, C2, and C5) were collected on MRI scanners, where post-contrast acquisitions occurred at comparable time intervals.

The scanning time for each abbreviated protocol was defined as the sum of the acquisition times of the included sequences. The full protocol was defined as the acquisition time of all sequences, including the entire DCE series. The workflow of the study is illustrated in Figure 2.

Figure 2 Study workflow. (A) Full breast MRI protocol acquisition, including T2WI, DWI with ADC maps, and multiphase DCE imaging (C0–C5). (B) Quantitative MRI variable measurement and selection. Lesion-to-background parenchymal SER and tumor size were derived from C1, C2, and C5 phases; ADC values were measured, and delta changes between time points were calculated. Significant variables were selected using the Mann-Whitney U test. (C) Abbreviated protocol evaluation and comparison. Six proposed abbreviated protocols are illustrated on the left. Diagnostic performance is illustrated with ROC curves, and scan and interpretation times were also assessed. ADC, apparent diffusion coefficient; DCA, decision curve analysis; DCE, dynamic contrast-enhanced; DWI, diffusion-weighted imaging; MRI, magnetic resonance imaging; NAC, neoadjuvant chemotherapy; ROC, receiver operating characteristic; SER, signal enhancement ratio; T2WI, T2-weighted imaging.

Image analysis and assessment

The image sets were reviewed by a dedicated breast imaging radiologist with nine years of experience, who was blinded to the pathological outcomes. Tumor location was provided to ensure measurement consistency. Tumor size and SER were measured at the C1, C2, and C5 post-contrast MRI phases. Tumor size was defined as the longest diameter on axial DCE images (23). SER was defined as the lesion signal intensity divided by that of the normal ipsilateral breast parenchyma (16). An ipsilateral lesion-to-parenchyma ratio measured on the same post-contrast phase was used reduce sensitivity to variability in overall parenchymal enhancement across examinations. Two circular regions of interest (ROIs) (median diameter, 3 mm; range, 2–5 mm) were manually placed to capture the most enhancing part of the lesion and the least enhancing part of the normal parenchyma, avoiding fat (16,24). For ADC measurements, tumor location was identified on post-contrast DCE subtraction images and used to guide ROI placement on the corresponding ADC maps at both the baseline and after two cycles of NAC. The ROIs were placed over visually identified areas of the lowest signal intensity within the tumor, avoiding adjacent fat, fibroglandular tissue, and regions with high T2 signal intensity (e.g., necrosis or seroma) (21). To limit ROI selection bias and improve longitudinal consistency, a standardized small ROI approach was used, and the same placement criteria were applied at both time points.

All quantitative measurements were performed at the baseline (Size0, SER0, and ADC0) and after two cycles of NAC (Size1, SER1, and ADC1). For each metric, the percentage change (Δ%) between the baseline and after two cycles of NAC was calculated to characterize within-patient longitudinal change. The percentage changes were calculated as follows: Δ%Size = (Size1−Size0)/Size0, Δ%SER = (SER1−SER0)/SER0, and Δ%ADC = (ADC1−ADC0)/ADC0. To evaluate reproducibility, a second breast radiologist with 16 years of experience independently repeated the measurements in a randomly selected subset of 60 patients, and interobserver agreement was assessed using the intraclass correlation coefficient (ICC).

Interpretation time assessment

To evaluate interpretation time, 30 patients were randomly selected from the primary cohort and 30 from the EVCs, yielding 60 cases for analysis. Two breast radiologists independently assessed these cases one month after completion of the primary image analysis to minimize recall bias. Each case was interpreted under all six abbreviated protocols in randomized order, with a one-week washout interval between protocols to reduce learning and carryover effects. Interpretation time was defined as the interval from loading the protocol-specific image set on the workstation to completion of the recorded measurements, excluding report generation.

Statistical analysis

The statistical analysis was performed using SPSS 26.0 (IBM, Armonk, NY, USA) and R version 4.1.2. The baseline characteristics of the study sample were summarized and stratified by pCR status. Categorical variables were compared using the chi-squared test or Fisher’s exact test, and continuous variables were compared using the Student’s t-test or the Mann-Whitney U test. Interobserver agreement for size, SER, and ADC measurements was evaluated using the ICC.

Candidate variables were predefined according to the sequences included in each abbreviated protocol. Specifically, Protocols 1–3 included Δ%Size and Δ%SER measured at C1, C2, and C5, respectively, and Protocols 4–6 also included Δ%ADC. For each protocol, a protocol-specific multivariable logistic regression model was developed in the primary cohort and applied to the EVCs. Receiver operating characteristic (ROC) curves and the corresponding areas under the curves (AUCs) were generated to evaluate the predictive performance of each abbreviated protocol, and the AUCs were compared using the DeLong test. Sensitivity (SEN), specificity (SPE), accuracy (ACC), positive predictive value (PPV), and negative predictive value (NPV) were also calculated to describe diagnostic performance.

A decision curve analysis (DCA) was conducted to assess the potential clinical utility of each protocol. Interpretation times were compared between protocols using the Wilcoxon signed-rank test, and analyses were performed separately for each reader. Subgroup analyses were further conducted according to molecular subtype. A two-sided P value <0.05 was considered statistically significant.

Results

Patient characteristics

A total of 359 patients were included in the study: 169 in the primary cohort, 89 in EVC1, and 101 in EVC2. The pCR rates were 30.18% (51/169) in the primary cohort, 33.71% (30/89) in EVC1, and 35.64% (36/101) in EVC2. The clinical and pathologic characteristics of the patients are detailed in Table 1. Across the three cohorts, HER2 status and molecular subtype distributions differed significantly between the pCR and non-pCR groups, with HER2-positive tumors more likely to achieve pCR (P<0.05). Age, menopausal status, HR status, and Ki-67 index showed no significant association with pCR in any of the cohorts (P>0.05).

Evaluation and selection of MRI-based variables

The interobserver agreement for tumor size, SER, and ADC measurements at the baseline and after two cycles of NAC was evaluated between the two radiologists, showing good to excellent consistency (ICCs 0.778–0.921; Table S2).

No significant differences in Δ%SER values were observed across the field strengths at C1, C2, and C5 phases (P>0.05), indicating consistent measurements between the 1.5 and 3.0 T scanners. However, Δ%ADC values showed some variation between the 1.5 and 3.0 T scanners, particularly in EVC 2 (P=0.001, Table S3).

In all three cohorts, lesions achieving pCR demonstrated significantly greater reductions in tumor size and SER (Δ%Size and Δ%SER) at C1, C2, and C5 phases compared to those that did not achieve pCR (all P<0.05). Similarly, Δ%ADC between the baseline and after two cycles of NAC was significantly higher in the pCR group than in the non-pCR group (all P<0.001; Table 2). These variables were subsequently used to construct protocol-based prediction models for pCR.

Performance of abbreviated MRI protocols for early NAC response assessment

Protocol-based models were constructed using multivariable logistic regression. Protocols 1–3 combined Δ%Size and Δ%SER measured at C1, C2, and C5, respectively, whereas Protocols 4–6 also included Δ%ADC. In the primary cohort, the AUC values of the six abbreviated protocols ranged from 0.734 to 0.876. Protocol 4 achieved the best performance, with an AUC of 0.876 [95% confidence interval (CI): 0.820–0.932] and an ACC of 0.840. Similar trends were observed in EVC1 (range, 0.759–0.878, highest Protocol 4) and EVC2 (range, 0.771–0.880, highest Protocol 4).

In relation to the protocols with different enhancement phases, Protocols 1–3 showed inconsistent performance in the primary cohort, with Protocol 3 significantly lower than Protocols 1 and 2 (0.734 vs. 0.813, P=0.010; 0.734 vs. 0.774, P=0.014), while no significant differences were observed among Protocols 1–3 in EVC1 and EVC2 (P>0.05). In relation to the DWI-containing protocols (Protocols 4–6), no significant differences were observed across all cohorts (P>0.05).

When comparing the protocols with and without DWI, Protocols 4–6 consistently outperformed their non-DWI counterparts (Protocols 1–3) across all three cohorts (all P<0.05). The detailed diagnostic performance of each protocol is shown in Table 3, with ROC curves and radar plots presented in Figure 3. ACC, SEN, SPE, PPV, and NPV are summarized in Table S4.

Figure 3 Model performance of the six abbreviated MRI protocols across three cohorts. ROC curves for the primary cohort (A), external validation cohort 1 (B), and external validation cohort 2 (C). Radar plots showing comparative performance metrics in the primary cohort (D), external validation cohort 1 (E), and external validation cohort 2 (F). ACC, accuracy; AUC, area under the curve; MRI, magnetic resonance imaging; NPV, negative predictive value; PPV, positive predictive value; ROC, receiver operating characteristic; SEN, sensitivity; SPE, specificity.

The DCA demonstrated the clinical utility of the protocols (Figure 4). Protocol 4 showed a net benefit exceeding both “treat-all” and “treat-none” across threshold probabilities (Pts) of 0.05–0.81 in the primary cohort, 0.02–0.80 in EVC1, and 0.02–0.93 in EVC2, with a consistent overlap of approximately 0.05–0.80 across all three cohorts. In both EVC1 and EVC2, the non-DWI protocols showed a net benefit mainly within intermediate Pt ranges across cohorts. In EVC1, the DWI-containing protocols remained net-benefit-positive from Pt 0.02 up to approximately 0.77–0.80, whereas the non-DWI protocols started at higher Pt thresholds (0.07–0.17) and ended earlier (0.62–0.71). A similar pattern was observed in EVC2, where the DWI-containing protocols extended the net-benefit-positive coverage up to 0.85–0.93 with lower bounds around 0.01–0.02, compared with 0.62–0.66 and 0.04–0.08 for the non-DWI protocols. Representative cases evaluated with abbreviated protocols are shown in Figure 5.

Figure 4 DCA for the six abbreviated MRI protocols across three cohorts. DCA curves for the primary cohort (A), external validation cohort 1 (B), and external validation cohort 2 (C). The DCA demonstrated the net clinical benefit of each abbreviated MRI protocol across a range of threshold probabilities. In the primary cohort, Protocol 4 showed a higher net benefit compared to the other protocols. In the external validation cohorts, Protocols 4–6 exhibited a relatively greater net benefit than the other protocols. DCA, decision curve analysis; MRI, magnetic resonance imaging.

Figure 5 Representative cases assessed with abbreviated MRI protocols. (A) A 32-year-old woman with invasive breast cancer received NAC and underwent modified radical mastectomy of the left breast. Pathology confirmed pCR. MRI performed before and after two cycles of NAC showed decreases in tumor size (ΔSize_C1: –47%, ΔSize_C2: –43%, ΔSize_C5: –38%), and the lesion-to-background SER (ΔSER_C1: –26%, ΔSER_C2: –8%, ΔSER_C5: –1%), with a marked increase in the ADC value (ΔADC: +64%). Abbreviated Protocols 4–6 correctly predicted pCR in this patient. (B) A 57-year-old woman with invasive breast cancer received NAC followed by modified radical mastectomy of the right breast. Pathological results confirmed that pCR was not achieved. MRI examinations before and after two cycles of NAC showed decreases in tumor size (ΔSize_C1: –61%, ΔSize_C2: –49%, ΔSize_C5: –44%), and the lesion-to-background SER (ΔSER_C1: –33%, ΔSER_C2: –27%, ΔSER_C5: –21%), with a slight increase in the ADC value (ΔADC: +28%). Abbreviated Protocols 4–6 correctly predicted no pCR for this patient. ADC, apparent diffusion coefficient; DWI, diffusion-weighted imaging; MRI, magnetic resonance imaging; NAC, neoadjuvant chemotherapy; pCR, pathological complete response; SER, signal enhancement ratio; T2WI, T2-weighted imaging.

Efficiency of abbreviated MRI protocols

Table 4 summarizes the scan times of the six abbreviated protocols. In the primary cohort, the scan times ranged from approximately 270 s (Protocol 1) to 621 s (Protocol 6), corresponding to a reduction of 11.16–61.37% relative to the full protocol. Similar patterns were observed in the two EVCs, with reductions of 13.46–56.48% in EVC1 and 25.48–65.02% in EVC2.

Table 5 shows the interpretation times of the abbreviated protocols. No significant differences were observed in the interpretation times of the two readers across Protocols 1–3 or across Protocols 4–6 (P>0.05). However, the interpretation times for Protocols 4–6 were consistently longer than those for their corresponding non-DWI counterparts (Protocols 1–3), with statistically significant differences across all comparisons (all P<0.001). These findings indicate that the inclusion of DWI increased the interpretation time, whereas the choice of enhancement phase (C1, C2, or C5) had little effect.

Subgroup analyses of the optimal protocol

Given that Protocol 4 demonstrated the best diagnostic performance with relatively shorter scan and interpretation times, it was considered the optimal abbreviated protocol. We therefore evaluated its performance across molecular subtypes in all three cohorts. Protocol 4 maintained good diagnostic performance across the subtypes, with AUCs ranging from 0.849 to 0.917 for HR+/HER2– tumors, 0.841 to 0.877 for HER2+ tumors, and 0.819 to 0.938 for TNBC.

In addition, we performed subgroup analyses based on the type of NAC regimen. For taxane-based regimens, Protocol 4 demonstrated AUC values ranging from 0.865 to 0.869 across all cohorts, while for anthracycline-based regimens, the AUC values ranged from 0.830 to 0.925. Detailed diagnostic metrics, including ACC, SEN, SPE, PPV, and NPV, are provided in Table 6.

Discussion

This study evaluated the diagnostic value of different abbreviated MRI protocols, focusing on the relative contribution of contrast-enhancement phases and the addition of DWI in predicting early response to NAC in breast cancer patients. Our results showed that extending dynamic contrast acquisition beyond the ~20-s post-contrast phase offered little incremental diagnostic value, whereas incorporating DWI consistently improved performance across cohorts, albeit with longer interpretation times. Notably, all the abbreviated protocols shortened the scan duration by approximately 11–65% compared with the full protocol, highlighting their potential to enhance clinical efficiency.

Previous studies using abbreviated MRI to assess response after NAC have primarily focused on post-treatment imaging and reported diagnostic performance comparable to full protocols, supporting the potential clinical utility of abbreviated MRI (6,25). Extending this application, this study assessed abbreviated protocols at the early treatment stage and demonstrated consistently good diagnostic performance across three independent cohorts. These findings suggest that abbreviated MRI may also be applied for the early monitoring of NAC response, expanding its potential role beyond post-treatment evaluation.

In this study, variables measured on MRI at the baseline and after two cycles of NAC were analyzed in relation to pCR across three cohorts. Δ%Size and Δ%SER across different DCE phases were both significantly correlated with pCR. Lesions achieving pCR showed larger absolute change rates, indicating that greater shrinkage and stronger SER decline were associated with pCR. Similarly, consistent with previous studies (20,26), we found that Δ%ADC were significantly higher in the pCR lesions than in the non-pCR lesions. Additionally, no significant differences in Δ%SER across different DCE phases between the 1.5 and 3.0 T scanners were observed, suggesting that this parameter is consistent across field strengths. Consistent with previous studies (27), Δ%ADC showed some variation between the 1.5 and 3.0 T scanners. However, this variation did not affect its predictive correlation with pCR. In all cohorts, both Δ%SER and Δ%ADC demonstrated consistent predictive power, suggesting that these parameters could be robust biomarkers for early NAC response assessment.

When comparing enhancement phases, no significant differences in diagnostic performance or interpretation time were observed among the protocols based on images acquired at approximately 20, 90, and 270 s after contrast injection. This suggests that extending contrast acquisition beyond the earlier post-contrast phase offers little additional benefit for the early prediction of pCR during NAC. Previous studies have indicated that early enhancement primarily reflects tumor vascularity and cellular activity, whereas delayed phases may better depict residual ductal carcinoma in situ and invasive lobular carcinoma (28,29). Reig et al. similarly reported no significant difference in residual tumor size assessment between early and delayed post-contrast images, further supporting our observation (30). Therefore, earlier enhancement phases may be sufficient for the accurate early prediction of pCR, enabling abbreviated protocols to reduce the scan time and streamline the workflow without compromising diagnostic performance.

Incorporating DWI into the abbreviated protocols substantially improved diagnostic performance, with Protocols 4–6 achieving higher AUCs and greater sensitivity than their non-DWI counterparts, which is consistent with the predictive results reported by Hottat et al. (19). Previous studies have also recommended combining DWI with DCE-MRI in standard protocols, as it provides complementary information on tumor cellularity and vascularity, thereby enhancing diagnostic accuracy (31). Longitudinal ADC changes have likewise shown promise for predicting pCR, and combining them with DCE-derived parameters may further improve predictive performance (21,32). However, the inclusion of DWI increased the interpretation time, revealing a trade-off between improved diagnostic performance and efficiency. In clinical practice, this additional time is likely justified given the substantial diagnostic benefit of DWI, supporting its integration as a key component of abbreviated protocols for the early prediction of pCR.

Previous studies have shown that biological heterogeneity across molecular subtypes may influence imaging biomarkers (29,33). Subtype-related differences in tumor vascularity and cellularity may contribute to variability in contrast enhancement and diffusion characteristics, and the reported predictive value of ADC changes may differ across subtypes (21,28,34). However, our subgroup analyses showed comparable discrimination of the optimal abbreviated protocol across the HR+/HER2–, HER2+, and TNBC subtypes in all three cohorts, with AUC values in a similar range. This consistency across subtypes suggests that the abbreviated approach may be broadly applicable despite known biologic differences. Additionally, we found that the performance of the optimal abbreviated protocol remained stable across different NAC regimens. While slight differences were observed between taxane-based and anthracycline-based regimens, the diagnostic performance of Protocol 4 was consistently strong across all cohorts, highlighting its robustness and potential clinical applicability across diverse treatment protocols. Notably, the confidence intervals were wider in the smaller subgroups, and these findings should be interpreted cautiously.

This study had several limitations. First, this was a retrospective study in which abbreviated protocols were extracted from full MRI examinations; prospective studies are needed to validate the findings of this study and assess the clinical applicability of the optimal protocols. Second, other potential abbreviated MRI protocols like non-enhanced MRI were not explored (35,36). Third, despite supportive interobserver agreement, ROI-based SER and ADC measurements remain susceptible to background parenchymal enhancement and sampling variability, and assessment using automated or volumetric approaches may further strengthen reproducibility. Finally, although multicenter cohorts were included, some molecular subtype subgroups were small, particularly TNBC in the EVCs, resulting in wider confidence intervals; larger cohorts are needed to confirm the stability of model performance across subtypes.

Conclusions

This study demonstrated the value of abbreviated MRI protocols for the early assessment of NAC response in breast cancer, with all protocols showing good diagnostic performance and shorter scan times than the full protocol. Prolonged enhancement phases offered little added value; however, the inclusion of DWI improved diagnostic accuracy, though with an increase in interpretation time. Protocols that integrate DWI with a shorter enhancement phase could provide a promising balance of diagnostic benefit and efficiency in clinical settings.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-2066/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-2066/dss

Funding: This work was supported by National Natural Science Foundation of China (No. 82302314 to W.T., No. 82271938 to X.J., No. 82302153 to Z.L.), Basic and Applied Basic Research Foundation of Guangdong Province (No. 2022A1515110792 to W.T., No. 2023A1515220097 to Y.G., No. 2024A1515010653 to Y.G.), Science and Technology Projects in Guangzhou (No. 2024A03J1030 to W.T., No. 2025A03J4163 to Y.G.), and the Special Fund for the Construction of High-level Key Clinical Specialty (Medical Imaging) in Guangzhou, Guangzhou Key Laboratory of Molecular Imaging and Clinical Translational Medicine.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-2066/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethical Committee of Guangzhou First People’s Hospital (No. S-2023-083-02). All participating hospitals were informed and agreed to the study. Individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Fowler AM, Mankoff DA, Joe BN. Imaging Neoadjuvant Therapy Response in Breast Cancer. Radiology 2017;285:358-75. [Crossref] [PubMed]
2. Álvarez-Benito M. Imaging evaluation of neoadjuvant breast cancer treatment: where do we stand? Eur Radiol 2024;34:6271-2. [Crossref] [PubMed]
3. Morrow M, Waters J, Morris E. MRI for breast cancer screening, diagnosis, and treatment. Lancet 2011;378:1804-11. [Crossref] [PubMed]
4. Kuhl CK. Abbreviated Breast MRI: State of the Art. Radiology 2024;310:e221822. [Crossref] [PubMed]
5. Kim SY, Cho N, Hong H, Lee Y, Yoen H, Kim YS, Park AR, Ha SM, Lee SH, Chang JM, Moon WK. Abbreviated Screening MRI for Women with a History of Breast Cancer: Comparison with Full-Protocol Breast MRI. Radiology 2022;305:36-45. [Crossref] [PubMed]
6. Tang WJ, Chen SY, Hu WK, Li XL, Zheng BJ, Wang ZS, Ding HJ, Chen LX, Zhang QQ, Yu XM, Sui Y, Wei XH, Guo Y. Abbreviated Versus Full-Protocol MRI for Breast Cancer Neoadjuvant Chemotherapy Response Assessment: Diagnostic Performance by General and Breast Radiologists. AJR Am J Roentgenol 2023;220:817-25. [Crossref] [PubMed]
7. Dornelas EC, Kawassaki CS, Olandoski M, Bolzon CL, de Oliveira RF, Urban LABD, Rabinovich I, Elifio-Esposito S. A three-sequence dynamic contrast enhanced abbreviated MRI protocol to evaluate response to breast cancer neoadjuvant chemotherapy. Magn Reson Imaging 2023;102:49-54. [Crossref] [PubMed]
8. Yirgin IK, Engin G, Yildiz Ş, Aydin EC, Karanlik H, Cabioglu N, Tukenmez M, Emiroglu S, Onder S, Yildiz SO, Yavuz E, Saip P, Aydiner A, Igci A, Muslumanoglu M. Abbreviated and Standard Breast MRI in Neoadjuvant Chemotherapy Response Evaluation: A Comparative Study. Curr Med Imaging 2022;18:1052-60. [Crossref] [PubMed]
9. Bitencourt AGV, Pires BS, Calsavara VF, Negrão EMS, Souza JA, Graziano L, Guatelli CS, Makdissi FB, Sanches SM, Tavares MC, Osório CABT, De Brot M, Marques EF, Chojniak R. Prognostic value of response evaluation based on breast MRI after neoadjuvant treatment: a retrospective cohort study. Eur Radiol 2021;31:9520-8. [Crossref] [PubMed]
10. Li Z, Li J, Lu X, Qu M, Tian J, Lei J. The diagnostic performance of diffusion-weighted imaging and dynamic contrast-enhanced magnetic resonance imaging in evaluating the pathological response of breast cancer to neoadjuvant chemotherapy: A meta-analysis. Eur J Radiol 2021;143:109931. [Crossref] [PubMed]
11. Jeh SK, Kim SH, Kang BJ. Comparison of the diagnostic performance of response evaluation criteria in solid tumor 1.0 with response evaluation criteria in solid tumor 1.1 on MRI in advanced breast cancer response evaluation to neoadjuvant chemotherapy. Korean J Radiol 2013;14:13-20. [Crossref] [PubMed]
12. Padhani AR, Hayes C, Assersohn L, Powles T, Makris A, Suckling J, Leach MO, Husband JE. Prediction of clinicopathologic response of breast cancer to primary chemotherapy at contrast-enhanced MR imaging: initial clinical results. Radiology 2006;239:361-74. [Crossref] [PubMed]
13. Rahmat K, Mumin NA, Hamid MTR, Hamid SA, Ng WL. MRI Breast: Current Imaging Trends, Clinical Applications, and Future Research Directions. Curr Med Imaging 2022;18:1347-61. [Crossref] [PubMed]
14. Khalifa F, Soliman A, El-Baz A, Abou El-Ghar M, El-Diasty T, Gimel’farb G, Ouseph R, Dwyer AC. Models and methods for analyzing DCE-MRI: a review. Med Phys 2014;41:124301. [Crossref] [PubMed]
15. Goto M, Ito H, Akazawa K, Kubota T, Kizu O, Yamada K, Nishimura T. Diagnosis of breast tumors by contrast-enhanced MR imaging: comparison between the diagnostic performance of dynamic enhancement patterns and morphologic features. J Magn Reson Imaging 2007;25:104-12. [Crossref] [PubMed]
16. Kim SY, Cho N, Shin SU, Lee HB, Han W, Park IA, Kwon BR, Kim SY, Lee SH, Chang JM, Moon WK. 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. Eur Radiol 2018;28:2986-95. [Crossref] [PubMed]
17. Ramtohul T, Tescher C, Vaflard P, Cyrta J, Girard N, Malhaire C, Tardivon A. Prospective Evaluation of Ultrafast Breast MRI for Predicting Pathologic Response after Neoadjuvant Therapies. Radiology 2022;305:565-74. [Crossref] [PubMed]
18. Gullo RL, Partridge SC, Shin HJ, Thakur SB, Pinker K. Update on DWI for Breast Cancer Diagnosis and Treatment Monitoring. AJR Am J Roentgenol 2024;222:e2329933. [Crossref] [PubMed]
19. Hottat NA, Badr DA, Lecomte S, Besse-Hammer T, Jani JC, Cannie MM. Value of diffusion-weighted MRI in predicting early response to neoadjuvant chemotherapy of breast cancer: comparison between ROI-ADC and whole-lesion-ADC measurements. Eur Radiol 2022;32:4067-78. [Crossref] [PubMed]
20. Jensen LR, Garzon B, Heldahl MG, Bathen TF, Lundgren S, Gribbestad IS. Diffusion-weighted and dynamic contrast-enhanced MRI in evaluation of early treatment effects during neoadjuvant chemotherapy in breast cancer patients. J Magn Reson Imaging 2011;34:1099-109. [Crossref] [PubMed]
21. Partridge SC, Zhang Z, Newitt DC, Gibbs JE, Chenevert TL, Rosen MA, Bolan PJ, Marques HS, Romanoff J, Cimino L, Joe BN, Umphrey HR, Ojeda-Fournier H, Dogan B, Oh K, Abe H, Drukteinis JS, Esserman LJ, Hylton NM. Diffusion-weighted MRI Findings Predict Pathologic Response in Neoadjuvant Treatment of Breast Cancer: The ACRIN 6698 Multicenter Trial. Radiology 2018;289:618-27. [Crossref] [PubMed]
22. Gradishar WJ, Moran MS, Abraham J, Aft R, Agnese D, Allison KH, et al. Breast Cancer, Version 3.2022, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 2022;20:691-722. [Crossref] [PubMed]
23. Zhu C, Chen M, Liu Y, Li P, Ye W, Ye H, Ye Y, Liu Z, Liang C, Liu C. Value of mammographic microcalcifications and MRI-enhanced lesions in the evaluation of residual disease after neoadjuvant therapy for breast cancer. Quant Imaging Med Surg 2023;13:5593-604. [Crossref] [PubMed]
24. Kim SY, Cho N, Choi Y, Lee SH, Ha SM, Kim ES, Chang JM, Moon WK. Factors Affecting Pathologic Complete Response Following Neoadjuvant Chemotherapy in Breast Cancer: Development and Validation of a Predictive Nomogram. Radiology 2021;299:290-300. [Crossref] [PubMed]
25. Spriet J, Miled AB, Mailliez A, Bonnier S, Forestier A, Chauvet MP, Rozwag C, Aoud IE, Ceugnart L. Breast magnetic resonance imaging patterns of tumor regression after neoadjuvant chemotherapy and immunotherapy in early triple-negative breast cancer patients: prediction of pathological response and performance of ultrafast sequences. Breast Cancer Res Treat 2025;212:195-203. [Crossref] [PubMed]
26. Sun S, Zhou J, Bai Y, Gao W, Lin L, Jiang T, You C, Gu Y. Role of oedema and shrinkage patterns for prediction of response to neoadjuvant chemotherapy and survival outcomes in luminal breast cancer. Clin Radiol 2024;79:e1010-20. [Crossref] [PubMed]
27. Partridge SC, Nissan N, Rahbar H, Kitsch AE, Sigmund EE. Diffusion-weighted breast MRI: Clinical applications and emerging techniques. J Magn Reson Imaging 2017;45:337-55. [Crossref] [PubMed]
28. Mann RM, Cho N, Moy L. Breast MRI: State of the Art. Radiology 2019;292:520-36. [Crossref] [PubMed]
29. Huang G, Du S, Gao S, Guo L, Zhao R, Bian X, Xie L, Zhang L. Molecular subtypes of breast cancer identified by dynamically enhanced MRI radiomics: the delayed phase cannot be ignored. Insights Imaging 2024;15:127. [Crossref] [PubMed]
30. Reig B, Lewin AA, Du L, Heacock L, Toth HK, Heller SL, Gao Y, Moy L. Breast MRI for Evaluation of Response to Neoadjuvant Therapy. Radiographics 2021;41:665-79. [Crossref] [PubMed]
31. Baltzer P, Mann RM, Iima M, Sigmund EE, Clauser P, Gilbert FJ, Martincich L, Partridge SC, Patterson A, Pinker K, Thibault F, Camps-Herrero J, Le Bihan D. Diffusion-weighted imaging of the breast-a consensus and mission statement from the EUSOBI International Breast Diffusion-Weighted Imaging working group. Eur Radiol 2020;30:1436-50. [Crossref] [PubMed]
32. Zhang MQ, Liu XP, Du Y, Zha HL, Zha XM, Wang J, Liu XA, Wang SJ, Zou QG, Zhang JL, Li CY. Prediction of pathological complete response of breast cancer patients who received neoadjuvant chemotherapy with a nomogram based on clinicopathologic variables, ultrasound, and MRI. Br J Radiol 2024;97:228-36. [Crossref] [PubMed]
33. Eom HJ, Kim HH, Kim HJ, Choi WJ, Chae EY, Shin HJ, Cha JH. Comparison of diffusion-weighted and contrast-enhanced MRI for monitoring response to neoadjuvant therapy in breast cancer. Eur Radiol 2025;35:8217-27. [Crossref] [PubMed]
34. Wang M, Du S, Gao S, Zhao R, Liu S, Jiang W, Peng C, Chai R, Zhang L. MRI-based tumor shrinkage patterns after early neoadjuvant therapy in breast cancer: correlation with molecular subtypes and pathological response after therapy. Breast Cancer Res 2024;26:26. [Crossref] [PubMed]
35. Du S, Gao S, Zhao R, Liu H, Wang Y, Qi X, Li S, Cao J, Zhang L. Contrast-free MRI quantitative parameters for early prediction of pathological response to neoadjuvant chemotherapy in breast cancer. Eur Radiol 2022;32:5759-72. [Crossref] [PubMed]
36. Cavallo Marincola B, Telesca M, Zaccagna F, Riemer F, Anzidei M, Catalano C, Pediconi F. Can unenhanced MRI of the breast replace contrast-enhanced MRI in assessing response to neoadjuvant chemotherapy? Acta Radiol 2019;60:35-44. [Crossref] [PubMed]