A pilot study of multimodal MRI for the preoperative assessment of triple-negative breast cancer

Objectives: To evaluate the diagnostic value of multimodal MRI by comprising quantitative mapping, diffusion-weighted imaging (DWI), and dynamic contrast-enhanced MRI (DCE-MRI) in differentiating triple-negative breast cancer (TNBC) from non-TNBC, and to investigate the correlation between significant quantitative mapping parameters and the Ki-67 proliferation index.

Methods: This prospective study enrolled 120 patients with invasive breast cancer (30 TNBC and 90 non-TNBC) who underwent preoperative quantitative mapping, DWI, and DCE-MRI. DCE-MRI parameters (RE, ME, MRE, T0, TTP, WI, WO, BOE), DWI metrics (ADCmin, ADCmean, rADCmin, rADCmean), and quantitative mapping parameters (Native T1, Enhanced T1, ΔT1, ΔT1%, T2) were compared between groups. Diagnostic performance was evaluated using receiver operating characteristic (ROC) analysis and area under the curve (AUC) for both individual significant parameters and a combined predictive model. Pearson’ s correlation coefficient was used to analyze the relationship between quantitative mapping parameters and the Ki-67 index.

Results: Compared to the non-TNBC group, TNBC lesions exhibited significantly lower RE, ME, WI, WO, and ΔT1% (p < 0.05), but higher Enhanced T1 and T2 values (p < 0.05). The combined model integrating DCE-MRI and quantitative mapping achieved the highest diagnostic accuracy (AUC = 0.977), outperforming DCE-MRI alone (AUC = 0.951) and quantitative mapping alone (AUC = 0.864), with sensitivity and specificity of 93.3% and 92.2%, respectively. In TNBC, the Ki-67 index showed a weak positive correlation with Enhanced T1 (r = 0.387, p < 0.05) and T2 (r = 0.381, p < 0.05).

Conclusion: Parameters including RE, ME, WI, WO, Enhanced T1, ΔT1%, and T2 serve as valuable imaging biomarkers for the preoperative identification of TNBC. Integrating quantitative mapping with DCE-MRI significantly enhances diagnostic performance, offering a robust approach for differentiating TNBC from non-TNBC subtypes.

1 Introduction

Breast cancer remains the most prevalent malignancy among women in China, with mortality rates ranking fourth, posing a significant public health threat (1). Due to high biological heterogeneity, breast cancer exhibits considerable variation in clinical management, treatment response, and overall survival across diverse molecular subtypes (2, 3). Triple-negative breast cancer (TNBC), characterized by the absence of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) expression, represents the most aggressive subtype, accounting for 15% to 20% of all breast cancer cases in China (4). Globally, TNBC constitutes approximately 10% to 15% of all breast cancer diagnoses (5). Recent epidemiological analyses have further highlighted significant racial and regional disparities in TNBC incidence, particularly within the United States where specific demographic cohorts and geographical regions exhibit distinct prevalence patterns (5). Given the lack of targeted therapeutic options, early and accurate diagnosis is crucial because patients with TNBC do not benefit from endocrine therapy or HER2 targeted agents and rely predominantly on chemotherapy (2, 6). Currently, immunohistochemical assessments are primarily derived from preoperative biopsy, which yields limited tissue samples and thus hampers the comprehensive evaluation of tumor heterogeneity (7).

Conventional magnetic resonance imaging (MRI) relies on subjective interpretation of morphological features such as tumor shape, signal intensity, and enhancement patterns, which limits objectivity. With the increasing demand for precision oncology, there is a growing need for advanced, quantitative imaging modalities capable of providing reliable biomarkers for objective clinical assessment. Diffusion-weighted imaging (DWI) is a non-invasive functional MRI technique that evaluates tissue microstructure by measuring the restricted diffusion of water molecules, quantified through the apparent diffusion coefficient (ADC). Dynamic contrast-enhanced MRI (DCE-MRI) assesses tumor microvascular architecture and perfusion by tracking the dynamic uptake and washout of contrast agents, typically analyzed using pharmacokinetic models. Both techniques have been widely applied in breast cancer for diagnosis, differential diagnosis, molecular subtyping, and monitoring response to neoadjuvant chemotherapy (8–12).

T2 mapping is an emerging quantitative MRI technique that enables voxel-wise measurement of T2 relaxation times, reflecting tissue free water content. Elevated free water content correlates with prolonged T2 values, providing insights into lesion composition (13). Furthermore, T2 mapping can generate color-coded parametric maps through pixel-by-pixel reconstruction, enhancing visualization and reducing observer bias (14). Similarly, T1 mapping allows specific assessment of longitudinal relaxation times at the voxel level (15). Native T1 refers to pre-contrast T1 values, while Enhanced T1 denotes post-contrast measurements, which are typically shortened due to gadolinium uptake. Initially developed for cardiac imaging, T1 mapping is increasingly being explored for oncological applications (16, 17). The high histological heterogeneity of breast cancer results in complex tissue compositions, necessitating more sensitive and quantitative imaging tools.

In the rapidly evolving landscape of oncology, precision medicine is being revolutionized by upgraded theragnostic strategies. For instance, multivariate nano-gold is being explored for the upgraded management of hepatocellular carcinoma (18). Similarly, harnessing the unique architecture of solid lipid nanoparticles has been proposed as a state-of-the-art strategy to combat colon cancer through improved drug delivery (19). Other emerging strategies include cabazitaxel-loaded redox-responsive nanocarriers based on tocopheryl chitosan and hyaluronic acid, designed to enhance anti-tumor efficacy (20). Furthermore, multifunctional nanosponges offer significant potential by integrating targeted drug delivery with imaging capabilities (21). These advancements in targeted therapy underscore the parallel imperative for equally precise, non-invasive diagnostic methods like multiparametric MRI in breast oncology.

Accurate noninvasive preoperative identification of TNBC is a critical step toward personalized medicine, enabling timely and appropriate therapeutic decision making. Therefore, this study aims to evaluate whether quantitative MRI techniques, including DWI, DCE-MRI, and T1/T2 mapping, can effectively capture the microstructural and physiological characteristics of TNBC. By constructing and comparing multiple diagnostic models, we seek to assess their individual and combined predictive performance in distinguishing TNBC from non-TNBC, thereby providing valuable insights for the development of personalized treatment strategies in breast cancer.

2 Materials and methods

2.1 Patients

This single-center prospective study was approved by the Institutional Review Board, and the requirement for informed consent was waived. From September 2022 to December 2024, we enrolled 245 consecutive patients with invasive breast cancer confirmed by core needle biopsy or surgical pathology. All subjects underwent preoperative multimodal MRI examinations, incorporating T1 and T2 mapping, diffusion-weighted imaging (DWI), and dynamic contrast-enhanced MRI (DCE-MRI) sequences. Exclusion criteria were as follows: (1) prior neoadjuvant chemotherapy, radiotherapy, or endocrine therapy (n = 5); (2) lesions manifesting exclusively as non-mass-like enhancement (n = 40); (3) insufficient clinical or pathological data (n = 46); (4) poor image quality or severe motion artifacts (n = 12); (5) neoplasms measuring less than 1 cm in the longest axis (n = 10); and (6) multifocal or multicentric tumors (n = 12). Consequently, a total of 120 patients with 120 lesions were included in the final analysis.

The sample size for this pilot study was determined based on feasibility and precedent from similar exploratory radiomics studies in breast cancer (9). A post-hoc power analysis (α = 0.05, β = 0.20) indicated that with 30 TNBC and 90 non-TNBC cases, our study possessed sufficient statistical power (power > 80%) to detect effect sizes (Cohen’s d) of 0.8 or greater for key MRI parameters, such as relative enhancement, maximum enhancement, and Enhanced T1 values. This represents a clinically relevant difference in this exploratory context.

2.2 Materials, equipment, and methods

2.2.1 Image acquisition

All breast MRI examinations were performed on a 3.0-T superconducting MR scanner (Ingenia 3.0 T, Philips Healthcare, Best, The Netherlands) with patients placed in the prone position. Both breasts were positioned within a dedicated 16-channel phased-array breast surface coil. The contrast medium employed was gadobenate dimeglumine (MultiHance, Bracco, Milan, Italy).

The comprehensive imaging protocol included the following sequences: T1-weighted imaging [repetition time (TR) = 584 ms, echo time (TE) = 8 ms, matrix 296 × 292, slice thickness 4 mm, number of excitations (NEX) = 1.5]; T2-weighted spectral adiabatic inversion recovery fat-suppressed imaging (TR = 4000 ms, TE = 70 ms, matrix 312 × 255, slice thickness 4 mm, NEX = 2); diffusion-weighted imaging (DWI) [TR = 6391 ms, TE = 77 ms, matrix 128 × 111, slice thickness 4 mm, NEX = 2, 2 b-values (0 and 800 s/mm²)] (22); native T1 mapping using a Look-Locker sequence (TR = 2.4 ms, TE = 1.13 ms, matrix 144 × 136, slice thickness 6 mm, NEX = 1); T2 mapping (TR = 1000 ms, TE = 9.8 ms, matrix 108 × 97, slice thickness 6 mm, NEX = 1); and dynamic contrast-enhanced MRI (DCE-MRI) [TR = 4.4 ms, TE = 2.1 ms, matrix 220 × 313, slice thickness 1.4 mm, NEX = 1]. This sequence comprised 8 dynamic acquisitions with a total duration of 6 minutes and 33 seconds, including one pre-contrast and seven post-contrast phases. Gadobenate dimeglumine was administered intravenously at a dose of 0.2 mmol/kg and an injection rate of 2.5 mL/s, followed by a 20 mL saline flush. Image acquisition for the dynamic series was initiated simultaneously with the start of the contrast agent injection. The temporal resolution for each dynamic phase was approximately 49 seconds. A total of 8 dynamic phases were acquired over a duration of 6 minutes and 33 seconds (23).

The quantitative consistency of T1 and T2 mapping was ensured by the manufacturer’s integrated calibration algorithm, with the system undergoing regular quality control using standardized phantoms. During DCE-MRI acquisition, patients were instructed to breathe freely and smoothly. No breath-hold maneuvers were performed, as the prone position effectively minimizes respiratory motion artifacts.

2.2.2 Image analyses and data processing

Two expert radiologists with 10 and 15 years of breast imaging experience analyzed all cases independently while remaining unaware of pathological findings. To evaluate measurement reproducibility, inter-observer agreement was assessed using intraclass correlation coefficients (ICC) in a randomly selected subset of 30 cases. The ICC values indicated excellent agreement for ADCmin (0.989), ME (0.979), ADCmean (0.977), Enhanced T1 (0.953), and RE (0.916), and good agreement for T2 (0.891) and Native T1 (0.881) (Supplementary Figure S1). Any disagreements in morphological assessments were settled via consensus-based discussion. Semi-quantitative and quantitative data were determined by calculating the mean values of the measurements obtained by both radiologists.

Morphological characteristics were analyzed based on the 2013 guidelines of Breast Imaging Reporting and Data System established by the American College of Radiology. The assessment includes the following: (1) mass features, including shape and spiculation; (2) mass enhancement patterns; (3) surrounding features, such as foci, peripheral vascularity, axillary lymph node, and alterations in the skin and nipple.

ADC value measurement: The ADC image was automatically generated upon transmission of the original DWI image to the Philips IntelliSpace Portal (ISP) workstation. Three layers exhibiting the largest radial tumor diameter, excluding regions with cystic degeneration, necrosis, and hemorrhage, were selected. A region of interest (ROI) measuring approximately 40–50 mm² was positioned at the area within the tumor contour where the ADC grey level displayed the most significant visual decrease. For each tumor, a single region of interest (ROI) was delineated on the slice exhibiting the largest axial diameter. The lowest ADC value among all ROIs was designated as ADCmin. The ADC map was used to localize the tumor’s maximal cross-sectional area, where an optimally sized oval or circular ROI was carefully delineated within the lesion boundaries to calculate the ADCmean. An ROI of identical size was positioned in the corresponding area of the contralateral breast to measure the ADC value of normal breast tissue. The relative ADCmin (rADCmin; rADCmin = ADCmin/ADC of normal breast) and relative ADCmean (rADCmean; rADCmean = ADCmean/ADC of normal breast) were subsequently calculated.

DCE-MRI measurement: DCE-MRI data were processed using the Philips ISP workstation, where semi-quantitative parameters were obtained through T1-weighted perfusion analyses. The lesion size was measured in the second phase, post-enhancement within the DCE-MRI sequence. The lesion’s maximal cross-sectional width was measured in the coronal, sagittal, and axial planes, with the largest value recorded as the lesion size. Concurrently, an ROI was placed at the area of the lesion showing the most pronounced enhancement in the same phase. The semi-quantitative parameters were calculated by averaging triplicate measurements across different tissue planes. The derived metrics comprised relative enhancement (RE), maximum enhancement (ME), maximum relative enhancement (MRE), T0 time, time to peak (TTP), wash-in rate (WI), wash-out rate (WO), brevity of enhancement (BOE), and time-intensity curve (TIC). TIC patterns were categorized into three types according to enhancement dynamics: type I (inflow type), type II (plateau type), and type III (outflow type).

Quantitative mapping measurement: Native T1, Enhanced T1, and T2 mapping images were transmitted to the Philips ISP workstation to generate quantitative pseudo-color maps. Native T1, Enhanced T1, and T2 mapping values were measured within the same ROI as defined in the ADC image. Measurements were repeated across three layers and averaged to calculate ΔT1 (ΔT1 = Native T1 - Enhanced T1) and ΔT1% (ΔT1% = ΔT1/Native T1) (24–26). The synthetic MRI sequences (Look-Locker for T1, multi-echo spin-echo for T2) incorporated the manufacturer’s integrated calibration algorithm to ensure quantitative consistency across scans without requiring external phantom calibration.

2.2.3 Histopathological analyses

The criteria for estrogen receptor (ER) and progesterone receptor (PR) positivity required at least 1% positively stained nuclei in 10 high-power microscopic fields (27). HER2 positivity was determined as an immunohistochemical (IHC) HER2 score of 3 or an IHC HER2 score of 2 or greater with gene amplification confirmed by fluorescence in situ hybridization (28). The Ki-67 index was calculated as the ratio of positively stained cells to the total cell count, assessed by evaluating 1000 cells in the area of highest staining density within the tumor. For clinical correlation analyses, a cutoff of 20% was used to define high versus low proliferation, consistent with the St. Gallen International Expert Consensus on the primary therapy of early breast cancer (29).

2.3 Statistical analyses

Statistical analyses were performed with MedCalc (version 22.0), SPSS (version 26.0), and Prism (version 10.4). Statistical significance was set at p < 0.05. For categorical variables, either the Chi-square test or Fisher’s exact test was applied as appropriate. Continuous data following a normal distribution were expressed as mean ± standard deviation, whereas non-normally distributed data were expressed as median (range). Differences in DWI, DCE-MRI, and quantitative mapping parameters between the TNBC and non-TNBC groups were assessed using the independent samples t-test (for normally distributed data with homogeneous variance) or the Mann-Whitney U test (for skewed distributions or uneven variance). Receiver operating characteristic (ROC) analyses were used to determine the diagnostic performance of individual parameters in distinguishing TNBC from non-TNBC. Furthermore, logistic regression was utilized to examine the predictive capability of the combined parameters. To assess the risk of overfitting and validate the combined model (DCE-MRI + quantitative mapping), we performed 5-fold cross-validation. The stability of sensitivity and specificity estimates was further evaluated using bootstrapping with 1000 resamples. Pearson correlation analyses were used to calculate the relationship between quantitative mapping parameters and the Ki-67 index.

3 Results

3.1 Clinicopathological and MRI characteristics of invasive breast cancer

A total of 120 patients (mean age, 51.4 ± 9.8 years; range, 32–76 years) were included in the final analysis. Among them, 30 patients were diagnosed with TNBC, and 90 patients were diagnosed with non-TNBC. The clinicopathological characteristics of the two groups are summarized in Table 1. TNBC lesions were more frequently located in the upper inner quadrant and were more likely to exhibit a round or oval shape compared to the irregular shapes often seen in non-TNBC (p < 0.05). Subgroup analysis of TNBC morphology revealed that the round or oval shape was predominant across all pathological subtypes. Specifically, 94.7% (18/19) of invasive ductal carcinomas, 100% (3/3) of invasive lobular carcinomas, and 100% (8/8) of other histological subtypes manifested as round or oval masses (Supplementary Table S1). Histopathologically, the TNBC group had a significantly higher proportion of Grade III tumors (66.7% vs. 36.7%, p = 0.009) and a higher Ki-67 proliferation index (68.77% ± 18.30% vs. 42.98% ± 21.05%, p < 0.001) than the non-TNBC group. No significant differences were observed in age, menopausal status, or tumor size between the two cohorts (p > 0.05), indicating comparable baseline characteristics.

Table 1

Table 1. Comparison of clinicopathologic and MRI characteristics between TNBC and non-TNBC groups.

3.2 Comparison of MRI parameters between TNBC and non-TNBC groups

Quantitative MRI analysis revealed significant disparities in hemodynamic and mapping parameters between the subtypes, as detailed in Table 2 and illustrated in Figures 1, 2. Regarding DCE-MRI metrics, the TNBC group exhibited significantly lower values for Relative Enhancement (RE) (169.26% ± 29.04% vs. 198.36% ± 36.33%, p < 0.001), Maximum Enhancement (ME) (183.56 ± 54.13 vs. 325.93 ± 129.49, p < 0.001), Wash-in rate (WI) (1.34 ± 0.41 s^-1 vs. 2.33 ± 0.68 s^-1, p < 0.001), and Wash-out rate (WO) (0.22 ± 0.11 s^-1 vs. 0.40 ± 0.27 s^-1, p < 0.001) compared to the non-TNBC group. In contrast, no statistically significant differences were observed in diffusion-weighted imaging parameters, including ADCmin (0.69 ± 0.17 × 10^–3 mm²/s vs. 0.64 ± 0.21 × 10^–3 mm²/s, p = 0.473) and ADCmean (0.93 ± 0.17 × 10^–3 mm²/s vs. 0.90 ± 0.23 × 10^–3 mm²/s, p = 0.423). However, quantitative mapping parameters showed marked differentiation; the TNBC group demonstrated significantly higher Enhanced T1 values (487.54 ± 75.57 ms vs. 417.09 ± 68.05 ms, p < 0.001) and T2 values (99.36 ± 13.23 ms vs. 83.33 ± 10.04 ms, p < 0.001). Furthermore, ΔT1% was significantly lower in TNBC (0.66 ± 0.05 vs. 0.70 ± 0.05, p < 0.001).

Table 2

Table 2. Comparison of DWI, DCE-MRI and quantitative mapping parameters between TNBC and non-TNBC groups.

Figure 1

Figure 1. Representative images of a Triple-Negative Breast Cancer (TNBC) lesion (Invasive Ductal Carcinoma, Ki-67: 60%) in the left breast. The tumor exhibits a round shape with smooth margins and heterogeneous enhancement. (A) Axial DCE-MRI image. (B) Relative Enhancement (RE) map. (C) Maximum Enhancement (ME) map. (D) Wash-in rate (WI) map. (E) Wash-out rate (WO) map. (F) ADC map showing restricted diffusion. (G) Native T1 map. (H) Enhanced T1 map. (J) T2 map. (K) Time-Intensity Curve (TIC) showing an outflow pattern (Type III).

Figure 2

Figure 2. Representative images of a non-TNBC lesion (Luminal B subtype, Invasive Ductal Carcinoma, Ki-67: 30%) in the left breast. The tumor exhibits an irregular shape with spiculation and heterogeneous enhancement. (A) Axial DCE-MRI image. (B) Relative Enhancement (RE) map. (C) Maximum Enhancement (ME) map. (D) Wash-in rate (WI) map. (E) Wash-out rate (WO) map. (F) ADC map. (G) Native T1 map. (H) Enhanced T1 map. (J) T2 map. (K) Time-Intensity Curve (TIC) showing a plateau pattern (Type II).

3.3 Diagnostic efficacy of DCE-MRI parameters

To assess the discriminative power of hemodynamic features, Receiver Operating Characteristic (ROC) analyses were performed for the significant DCE-MRI parameters, as summarized in Table 3 and Figure 3A. Among the individual metrics, Wash-in rate (WI) and Maximum Enhancement (ME) demonstrated robust diagnostic potential, achieving AUC values of 0.894 (95% CI: 0.825–0.943) and 0.887 (95% CI: 0.816–0.937), respectively. Relative Enhancement (RE) and Wash-out rate (WO) showed moderate diagnostic utility with AUCs of 0.723 and 0.707. Notably, the integration of these four parameters into a combined DCE-MRI model yielded superior performance, achieving an AUC of 0.951 (95% CI: 0.896–0.982) with a sensitivity of 93.3% and specificity of 85.6% (Table 4).

Table 3

Table 3. Diagnostic efficacy of DCE-MRI (RE, ME, WI, WO) and quantitative mapping (Enhanced T1, ΔT1%, T2) for TNBC.

Figure 3

Figure 3. Receiver operating characteristic (ROC) curves evaluating the diagnostic performance of different MRI parameters and models for distinguishing TNBC from non-TNBC. (A) ROC curves of individual DCE-MRI parameters (RE, ME, WI, WO) and their combination. (B) ROC curves of quantitative mapping parameters (Enhanced T1, ΔT1%, T2) and their combination. (C) Comparison of diagnostic performance among the DCE-MRI model, Quantitative Mapping model, and the Combined Model (DCE-MRI + Quantitative Mapping). The Combined Model achieved the highest Area Under the Curve (AUC = 0.977).

Table 4

Table 4. Diagnostic efficacy of DCE-MRI, quantitative mapping and combined model for TNBC.

3.4 Diagnostic efficacy of quantitative mapping parameters

The diagnostic utility of quantitative mapping parameters was similarly evaluated (Table 3; Figure 3B). T2 mapping emerged as the most effective individual predictor in this category, with an AUC of 0.820 (95% CI: 0.739–0.884), followed by Enhanced T1 (AUC = 0.756) and ΔT1% (AUC = 0.721). The combined quantitative mapping model enhanced the overall specificity to 90.0% (95% CI: 81.9–95.3), although sensitivity remained moderate at 70.0% (95% CI: 50.6–85.3). The resulting AUC for the combined mapping model was 0.864 (95% CI: 0.789–0.919), indicating that while mapping parameters are highly specific, they are most powerful when used in conjunction with other modalities.

3.5 Comparison of diagnostic models

The overall diagnostic performance of the individual and combined models is presented in Table 4 and Figure 3C. The integrated model, which combined both DCE-MRI and quantitative mapping parameters, achieved the highest diagnostic accuracy. This comprehensive model yielded an AUC of 0.977 (95% CI: 0.932–0.996), significantly outperforming either modality alone. It demonstrated balanced and high-performance metrics, with a sensitivity of 93.3% (95% CI: 77.9–99.2) and a specificity of 92.2% (95% CI: 84.6–96.8). To validate the robustness and generalizability of these findings, a 5-fold cross-validation was performed, which resulted in a mean AUC of 0.962 (SD: 0.015), confirming the stability of the combined diagnostic model.

3.6 Correlation with Ki-67 index

The results indicated a significant positive correlation between Enhanced T1 values and the Ki-67 proliferation index (Figure 4A; r = 0.387, 95% CI: 0.031–0.656, p = 0.035). Notably, both Enhanced T1 and T2 values showed a progressive increasing trend concomitant with higher Ki-67 indices. In contrast, no statistically significant correlation was observed between ΔT1% and the Ki-67 index (Figure 4B; r = -0.332, 95% CI: -0.618–0.032, p = 0.073), suggesting that relative T1 changes may be less sensitive to proliferative activity than absolute relaxation times. Similarly, T2 values exhibited a positive correlation with Ki-67 levels (Figure 4C; r = 0.381, 95% CI: 0.024–0.652, p = 0.038).

Figure 4

Figure 4. Scatter plots illustrating the correlation between quantitative mapping parameters and the Ki-67 proliferation index in the TNBC group. (A) Correlation between Enhanced T1 values and Ki-67 index. (B) Correlation between ΔT1% and Ki-67 index. (C) Correlation between T2 values and Ki-67 index. Each dot represents an individual tumor lesion. Pearson’s correlation coefficients (r) and p-values are indicated in the respective plots.

4 Discussion

The selection of complex treatment regimens for breast cancer primarily depends on hormone receptor and HER2 expression status. Patients with hormone receptor-positive tumors are eligible for endocrine therapy, whereas those with HER2-positive tumors benefit from anti-HER2 targeted therapy. In contrast, patients whose tumors lack expression of both hormone receptors and HER2, commonly referred to as triple-negative breast cancer (TNBC), have limited therapeutic options and generally exhibit a poorer prognosis (30). Breast cancer is characterized by substantial tumor heterogeneity. Currently, molecular subtyping research largely relies on dynamic contrast-enhanced MRI (DCE-MRI), diffusion-weighted imaging (DWI), and their derived sequences (1, 22). This study innovatively incorporated novel mapping sequences into the existing framework. By systematically analyzing quantitative mapping, DCE-MRI, DWI, and integrated models, we identified RE, ME, WI, WO, Enhanced T1, ΔT1%, and T2 as significant predictive biomarkers. Furthermore, the multimodal integration of DCE-MRI with quantitative mapping demonstrated enhanced preoperative diagnostic accuracy, holding promise for guiding more precise clinical decision-making.

4.1 Morphological and pathological characteristics

First, we evaluated morphological characteristics on DCE-MRI. Our results demonstrated that TNBC predominantly exhibited a round or oval morphology, which was more prevalent than in non-TNBC. Concurrently, the incidence of spiculation in TNBC was significantly lower, consistent with prior studies (31, 32). This suggests that TNBC often mimics the imaging features of benign breast tumors, posing challenges for accurate preoperative diagnosis. Previous studies have indicated that spiculation is associated with the expression of estrogen receptor (ER) and progesterone receptor (PR) genes (33–35). This implies that the absence of ER and PR expression contributes to the smooth tumor margins typical of TNBC. Moreover, TNBC has been reported to lack a pronounced fibrotic connective tissue response (32). The reduced occurrence of spiculation in the TNBC group may be attributed to limited interstitial reaction and diminished fibrous connective tissue proliferation resulting from cancer cell infiltration.

Heterogeneous enhancement and necrosis were more frequently observed in TNBC, with a higher proportion of rim enhancement compared to non-TNBC. Additionally, increased peritumoral vascularity was noted, aligning with previous findings of elevated microvascular density at the tumor periphery. This phenomenon is likely driven by rapid cancer cell proliferation leading to central fibrosis and necrosis (36–38). This hypothesis is further supported by the significantly higher Ki-67 index observed in the TNBC group in our cohort. Pathologically, the smooth, non-spiculated tumor margins are consistent with a reduced desmoplastic reaction, associated with ER/PR negativity and a pushing, expansive growth pattern driven by high proliferative activity. Thus, the round or oval shape with minimal spiculation represents a radiological manifestation of TNBC’s distinct pathobiological behavior.

4.2 Hemodynamic features and paradoxical perfusion

Beyond morphological characteristics, we evaluated semi-quantitative hemodynamic parameters derived from DCE-MRI. No significant intergroup difference was observed in time-intensity curve (TIC) type classification (p > 0.05). Currently, there is no consensus regarding the typical TIC pattern of TNBC, and existing studies indicate that differences in TIC types often lack statistical significance (22, 39, 40). Semi-quantitative DCE-MRI parameters provide a more accurate assessment of intratumoral microvascular density and vascular permeability, offering insights beyond TIC classification alone (41, 42).

Tumor aggressiveness is generally positively correlated with intratumoral hemodynamics. High-grade malignancies typically exhibit increased perfusion and vascular permeability, leading to elevated peak enhancement and rapid wash-in. Contrary to expectation, our study found that TNBC lesions showed significantly lower values for semi-quantitative parameters (RE, ME, WI, WO). This seemingly paradoxical finding may point to a distinct tumor microenvironment in TNBC. We propose that this pattern stems from an immature, hyperpermeable, and architecturally disorganized vasculature. Such vessels facilitate rapid extravasation of contrast agent into the interstitial space and equally rapid clearance, preventing the accumulation necessary to reach high enhancement peaks during standard acquisition. This interpretation aligns with evidence that TNBCs often display elevated quantitative perfusion parameters, such as K^trans, which mark hyperpermeable yet functionally inefficient vasculature (43, 44). Furthermore, the lack of HER2 expression in TNBC has been hypothesized to impair local vascular endothelial growth factor (VEGF) production, potentially contributing to this abnormal vascular phenotype (43, 45).

4.3 Diagnostic value and limitations of diffusion-weighted imaging

In contrast to the DCE-MRI findings, our DWI analysis revealed no statistically significant differences in ADC metrics between TNBC and non-TNBC groups (p > 0.05). This aligns with several previous reports (11, 46, 47). The lack of significant ADC differentiation may be attributed to TNBC’s complex microstructure, where opposing effects neutralize the net ADC value. Specifically, the characteristic high cellularity of TNBC tends to restrict water diffusion and lower ADC values (10). Conversely, the frequent presence of necrosis and edema (observed as rim enhancement) tends to increase water diffusivity. These competing factors likely cancel each other out. Furthermore, the use of a single intermediate b-value of 800 s/mm² in our protocol represents a standardized clinical compromise. While maintaining optimal signal-to-noise ratio, it may lack the sensitivity to disentangle these microstructural subtleties compared to multi-b-value protocols or advanced models like intravoxel incoherent motion (IVIM) (22).

4.4 Value of quantitative mapping and pathological basis

The integration of quantitative T1 and T2 mapping into a multimodal MRI protocol adds a novel, objective dimension to preoperative TNBC assessment. Our study demonstrates that the TNBC group exhibited decreased ΔT1% but increased Enhanced T1 and T2 values. Mechanistically, VEGF expression is differentially regulated by HER2 and ER status (45, 48, 49). In TNBC, the absence of both receptors may lead to relative deficiencies in VEGF-driven angiogenesis stability, resulting in comparatively poorer blood supply retention. This explains the smaller post-contrast reduction in T1 relaxation time (higher Enhanced T1, lower ΔT1%). Additionally, elevated T2 values in TNBC likely reflect increased free water content, attributable to characteristically high cellular density and frequent microscopic necrotic foci and edema (50, 51).

4.5 Synergistic diagnostic performance

The study demonstrated that while DCE-MRI alone achieved high sensitivity and quantitative mapping alone showed high specificity, integrating these modalities yielded substantially improved performance. The combined model maintained high sensitivity at 93.3% while elevating specificity to 92.2%. This synergistic model outperforms single-modality approaches, demonstrating significant promise in minimizing both false positive and false negative diagnoses (14, 39). This highlights its potential for clinical application in distinguishing aggressive breast cancer subtypes.

4.6 Correlation with proliferation marker Ki-67

Our study identified significant, albeit weak, positive correlations between the Ki-67 proliferation index and both Enhanced T1 and T2 values in TNBC. The correlation with T2 may be explained by TNBC pathophysiology, where rapid proliferation leads to a hypoxic and necrotic microenvironment, increasing tissue water content and T2 signal (51). While previous studies grouped tumors by Ki-67 expression to show relationships with T2, our analysis provides a direct quantitative assessment. However, given the modest correlation strength, mapping parameters alone are likely insufficient for precise individual Ki-67 prediction but serve as valuable supportive biomarkers (52, 53).

4.7 Limitations and future prospects

This study has several limitations. First, its single-center design and relatively small sample size, particularly within the TNBC subgroup, may limit generalizability. However, post-hoc power analysis indicated sufficient statistical power to detect significant differences in key parameters. Future multi-center studies with larger cohorts are needed for validation. Second, manual lesion delineation introduces subjectivity. Future implementation of AI for automated segmentation could improve precision (54). Third, the lack of direct spatial correlation between MRI features and histopathology limits definitive attribution. Future studies incorporating MRI-guided biopsy could enable precise radiologic-pathologic correlation. Finally, conventional mono-exponential DWI may lack sensitivity. Future work using multi-b-value acquisitions or advanced models (IVIM, DKI) could provide more nuanced characterization (10).

5 Conclusion

In summary, this study demonstrates that the integration of quantitative T1 and T2 mapping with conventional DCE-MRI parameters provides a highly effective and noninvasive tool for the preoperative differentiation of TNBC. The combined predictive model achieves optimal diagnostic performance by capturing the distinct pathophysiological features of TNBC, including its unique tumor microenvironment and perfusion patterns. The ability to accurately identify TNBC preoperatively has profound clinical implications. It allows for the stratification of patients who are most likely to benefit from aggressive systemic therapies, such as the prioritization of neoadjuvant chemotherapy, which is crucial for achieving pathological complete response and improving long term prognosis. While the promising results from this pilot study require validation in larger cohorts, this integrated imaging framework represents a tangible advance toward personalized, precision medicine in breast cancer management.

Data availability statement

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

Author contributions

SL: Conceptualization, Data curation, Writing – review & editing. RL: Formal Analysis, Visualization, Writing – original draft. HL: Investigation, Writing – original draft. ZZ: Investigation, Writing – original draft. XW: Formal Analysis, Visualization, Writing – original draft. LT: Formal Analysis, Visualization, Writing – original draft. XH: Investigation, Writing – original draft. YL: Investigation, Writing – original draft. PL: Investigation, Writing – original draft. JY: Writing – review & editing. PW: Writing – review & editing. XL: Conceptualization, Writing – review & editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication. Fujian Province Natural Science Foundation (2022J011501 and 2023J011880).

Conflict of interest

Authors JY and PW were employed by company Philips Healthcare.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc.2026.1714620/full#supplementary-material

References

1. Orlando AAM, Clauser P, Zarcaro C, Ferraro F, Curatolo C, Marino MA, et al. MRI insights in breast imaging. Curr Med Imaging. (2024) 20:e15734056274670. doi: 10.2174/0115734056274670240205090722

PubMed Abstract | Crossref Full Text | Google Scholar

2. Dass SA, Tan KL, Selva RR, Mokhtar NF, Adzmi ERM, Rahman WFWA, et al. Triple negative breast cancer: A review of present and future diagnostic modalities. Med (Kaunas). (2021) 57:62. doi: 10.3390/medicina57010062

PubMed Abstract | Crossref Full Text | Google Scholar

3. Chen H, Min Y, Xiang K, Chen J, and Yin G. DCE-MRI performance in triple negative breast cancers: comparison with non-triple negative breast cancers. Curr Med Imaging. (2022) 18:970–6. doi: 10.2174/1573405618666220225090944

PubMed Abstract | Crossref Full Text | Google Scholar

4. Johansson A, Trewin CB, Fredriksson I, Reinertsen KV, Russnes H, Ursin G, et al. In modern times, how important are breast cancer stage, grade and receptor subtype for survival: a population-based cohort study. Breast Cancer Res. (2021) 23:17. doi: 10.1186/s13058-021-01393-z

PubMed Abstract | Crossref Full Text | Google Scholar

5. Zhang W, Bai Y, Sun C, Lv Z, and Wang S. Racial and regional disparities of triple negative breast cancer incidence rates in the United States: An analysis of 2011–2019 NPCR and SEER incidence data. Front Public Health. (2022) 10:1058722. doi: 10.3389/fpubh.2022.1058722

PubMed Abstract | Crossref Full Text | Google Scholar

6. Nounou MI, Elamrawy F, Ahmed N, Abdelraouf K, Goda S, and Syed-Sha-Qhattal H. Breast cancer: conventional diagnosis and treatment modalities and recent patents and technologies. Breast Cancer (Auckl). (2015) 9:17–34. doi: 10.4137/BCBCR.S29420

PubMed Abstract | Crossref Full Text | Google Scholar

7. Ma Y, Liu A, O’Connell AM, Zhu Y, Li H, Han P, et al. Contrast-enhanced cone beam breast CT features of breast cancers: correlation with immunohistochemical receptors and molecular subtypes. Eur Radiol. (2021) 31:2580–9. doi: 10.1007/s00330-020-07277-8

PubMed Abstract | Crossref Full Text | Google Scholar

8. Suo S, Zhang D, Cheng F, Cao M, Hua J, Lu J, et al. Added value of mean and entropy of apparent diffusion coefficient values for evaluating histologic phenotypes of invasive ductal breast cancer with MR imaging. Eur Radiol. (2019) 29:1425–34. doi: 10.1007/s00330-018-5667-9

PubMed Abstract | Crossref Full Text | Google Scholar

9. Tezcan S, Ozturk FU, Uslu N, and Akcay EY. The role of combined diffusion-weighted imaging and dynamic contrast-enhanced MRI for differentiating Malignant from benign breast lesions presenting washout curve. Can Assoc Radiol J. (2021) 72:460–9. doi: 10.1177/0846537120907098

PubMed Abstract | Crossref Full Text | Google Scholar

10. Mori N, Inoue C, Tamura H, Nagasaka T, Ren H, Sato S, et al. Apparent diffusion coefficient and intravoxel incoherent motion-diffusion kurtosis model parameters in invasive breast cancer: Correlation with the histological parameters of whole-slide imaging. Magn Reson Imaging. (2022) 90:53–60. doi: 10.1016/j.mri.2022.04.003

PubMed Abstract | Crossref Full Text | Google Scholar

11. Choi BB. Effectiveness of ADC difference value on pre-neoadjuvant chemotherapy MRI for response evaluation of breast cancer. Technol Cancer Res Treat. (2021) 20:15330338211039129. doi: 10.1177/15330338211039129

PubMed Abstract | Crossref Full Text | Google Scholar

12. Xie T, Zhao Q, Fu C, Grimm R, GuY, and Peng W. Improved value of whole-lesion histogram analyses on DCE parametric maps for diagnosing small breast cancer (≤ 1 cm). Eur Radiol. (2022) 32:1634–43. doi: 10.1007/s00330-021-08244-7

PubMed Abstract | Crossref Full Text | Google Scholar

13. Yuan C, Jin F, Guo X, Zhao S, Li W, and Guo H. Correlation analyses of breast cancer DWI combined with DCE-MRI imaging features with molecular subtypes and prognostic factors. J Med Syst. (2019) 43:83. doi: 10.1007/s10916-019-1197-5

PubMed Abstract | Crossref Full Text | Google Scholar

14. Sun SY, Ding Y, Li Z, Nie L, Liao C, Liu Y, et al. Multiparameter MRI model with DCE-MRI, DWI, and synthetic MRI improves the diagnostic performance of BI-RADS 4 lesions. Front Oncol. (2021) 11:699127. doi: 10.3389/fonc.2021.699127

PubMed Abstract | Crossref Full Text | Google Scholar

15. Sabouri S, Chang SD, Savdie R, Zhang J, Jones EC, Goldenberg SL, et al. Luminal water imaging: A new MR imaging T2 mapping technique for prostate cancer diagnosis. Radiology. (2017) 284:451–9. doi: 10.1148/radiol.2017161687

PubMed Abstract | Crossref Full Text | Google Scholar

16. Serai SD. Basics of magnetic resonance imaging and quantitative parameters T1, T2, T2*, T1rho and diffusion-weighted imaging. Pediatr Radiol. (2022) 52:217–27. doi: 10.1007/s00247-021-05042-7

PubMed Abstract | Crossref Full Text | Google Scholar

17. Adams LC, Ralla B, Jurmeister P, Bressem KK, Fahlenkamp UL, Hamm B, et al. Native T1 mapping as an in vivo biomarker for the identification of higher-grade renal cell carcinoma: correlation with histopathological findings. Invest Radiol. (2019) 54:118–28. doi: 10.1097/RLI.0000000000000515

PubMed Abstract | Crossref Full Text | Google Scholar

18. Halder D, Gupta A, Chakraborty S, Roy P, and Sen KK. Upgraded theragnostic strategies of multivariate nano gold in the management of hepatocellular carcinoma. Nano-Struct Nano-Objects. (2024) 40:101342. doi: 10.1016/j.nanoso.2024.101342

Crossref Full Text | Google Scholar

19. Gupta A, Jadhav SR, Colaco V, Saha M, Ghosh A, Sreedevi A, et al. Harnessing unique architecture and emerging strategies of solid lipid nanoparticles to combat colon cancer: A state-of-the-art review. Int J Pharm. (2025) 675:125562. doi: 10.1016/j.ijpharm.2025.125562

PubMed Abstract | Crossref Full Text | Google Scholar

20. Jha A, Kumar M, Goswami P, Bharti K, Manjit M, Gupta A, et al. Cabazitaxel-loaded redox-responsive nanocarrier based on d-alpha-tocopheryl-chitosan and hyaluronic acid for improved anti-tumor efficacy in DMBA-induced breast cancer model. Mater Adv. (2024) 5:7789–808. doi: 10.1039/D4MA00556B

Crossref Full Text | Google Scholar

21. Jadhav SR, Gupta A, Colaco V, Saha M, Moumita S, Sreedevi A, et al. Multifunctional nanosponges in cancer therapy: Integrating targeted drug delivery and theranostic potential. Int J Pharm X. (2025) 10:100421. doi: 10.1016/j.ijpx.2025.100421

PubMed Abstract | Crossref Full Text | Google Scholar

22. Zhang J. Study on Magnetic Resonance Diagnosis Strategy of Breast Lesions and Efficacy Evaluation of Neoadjuvant Chemotherapy. Beijing: Chinese PLA General Hospital (2010).

Google Scholar

23. Matsuda M, Tsuda T, Ebihara R, Toshimoria W, Okadaa K, Takedaa S, et al. Triple-negative breast cancer on contrast-enhanced MRI and synthetic MRI: A comparison with non-triple-negative breast carcinoma. Eur J Radiol. (2021) 142:109838. doi: 10.1016/j.ejrad.2021.109838

PubMed Abstract | Crossref Full Text | Google Scholar

24. Baidya KE, Sharma N, Sharma R, Bakhshi S, Kandasamy D, and Mehndiratta A. T1 mapping as a surrogate marker of chemotherapy response evaluation in patients with osteosarcoma. Eur J Radiol. (2022) 148:110170. doi: 10.1016/j.ejrad.2022.110170

PubMed Abstract | Crossref Full Text | Google Scholar

25. Yuan J, Wen Q, Wang H, Wang J, Liu K, Zhan S, et al. The use of quantitative T1-mapping to identify cells and collagen fibers in rectal cancer. Front Oncol. (2023) 13:1189334. doi: 10.3389/fonc.2023.1189334

PubMed Abstract | Crossref Full Text | Google Scholar

26. Yang R, Chen Z, Pan J, Yang S, and Hu F. Non-contrast T1rho dispersion versus Gd-EOB-DTPA-enhanced T1mapping for the risk stratification of non-alcoholic fatty liver disease in rabbit models. Magn Reson Imaging. (2024) 107:130–7. doi: 10.1016/j.mri.2024.01.013

PubMed Abstract | Crossref Full Text | Google Scholar

27. Allison KH, Hammond M, Dowsett M, McKernin SE, Carey LA, Fitzgibbons PL, et al. Estrogen and progesterone receptor testing in breast cancer: ASCO/CAP guideline update. J Clin Oncol. (2020) 38:1346–66. doi: 10.1200/JCO.19.02309

PubMed Abstract | Crossref Full Text | Google Scholar

28. Wolff AC, Hammond M, Allison KH, Harvey BE, Mangu PB, Bartlett JM, et al. Human epidermal growth factor receptor 2 testing in breast cancer: American society of clinical oncology/College of American pathologists clinical practice guideline focused update. Arch Pathol Lab Med. (2018) 142:1364–82. doi: 10.5858/arpa.2018-0902-SA

PubMed Abstract | Crossref Full Text | Google Scholar

29. Curigliano G, Burstein HJ, Winer EP, Gnant M, Dubsky P, Loibl S, et al. De-escalating and escalating treatments for early-stage breast cancer: the St. Gallen International Expert Consensus Conference on the Primary Therapy of Early Breast Cancer 2017. Ann Oncol. (2017) 28:1700–12. doi: 10.1093/annonc/mdx308

PubMed Abstract | Crossref Full Text | Google Scholar

30. Breast Cancer Professional Committee of Chinese Anti Cancer Association. Guidelines and norms for diagnosis and treatment of breast cancer of China Anti-Cancer Association (2021 edition). China Oncol. (2021) 31:609–80. doi: 10.19401/j.cnki.1007-3639.2021.10.013

Crossref Full Text | Google Scholar

31. Yetkin DI, Akpinar MG, Durhan G, and Demirkazik FB. Comparison of clinical and magnetic resonance imaging findings of triple-negative breast cancer with non-triple-negative tumours. Pol J Radiol. (2021) 86:e269–76. doi: 10.5114/pjr.2021.106137

PubMed Abstract | Crossref Full Text | Google Scholar

32. Navarro Vilar L, Alandete Germa´n SP, Medina Garci´a R, Blanc García E, Camarasa Lillo N, and Vilar Samper J. MR imaging findings in molecular subtypes of breast cancer according to BIRADS system. Breast J. (2017) 23:421–8. doi: 10.1111/tbj.12756

PubMed Abstract | Crossref Full Text | Google Scholar

33. Huang J, Lin Q, Cui C, Fei J, Su X, Li L, et al. Correlation between imaging features and molecular subtypes of breast cancer in young women (≤30 years old). Jpn J Radiol. (2020) 38:1062–74. doi: 10.1007/s11604-020-01001-8

PubMed Abstract | Crossref Full Text | Google Scholar

34. Moffa G, Galati F, Collalunga E, Rizzo V, Kripa E, D’Amati G, et al. Can MRI biomarkers predict triple-negative breast cancer? Diagnostics (Basel). (2020) 10:1090. doi: 10.3390/diagnostics10121090

PubMed Abstract | Crossref Full Text | Google Scholar

35. Long N, Ran C, Sun J, Hao C, Sui Y, Li J, et al. Correlation study between the magnetic resonance imaging features of breast cancer and expression of immune molecular subtypes. Eur Rev Med Pharmacol Sci. (2020) 24:11518–27. doi: 10.26355/eurrev_202011_23793

PubMed Abstract | Crossref Full Text | Google Scholar

36. Galati F, Rizzo V, Moffa G, Caramanico C, Kripa E, Cerbelli B, et al. Radiologic-pathologic correlation in breast cancer: do MRI biomarkers correlate with pathologic features and molecular subtypes? Eur Radiol Exp. (2022) 6:39. doi: 10.1186/s41747-022-00289-7

PubMed Abstract | Crossref Full Text | Google Scholar

37. Szep M, Pintican R, Boca B, Perja A, Duma M, Feier D, et al. Multiparametric MRI features of breast cancer molecular subtypes. Med (Kaunas). (2022) 58:1716. doi: 10.3390/medicina58121716

PubMed Abstract | Crossref Full Text | Google Scholar

38. Jeh SK, Kim SH, Kim HS, Kang BJ, Jeong SH, Yim HW, et al. Correlation of the apparent diffusion coefficient value and dynamic magnetic resonance imaging findings with prognostic factors in invasive ductal carcinoma. J Magn Reson Imaging. (2011) 33:102–9. doi: 10.1002/jmri.22400

PubMed Abstract | Crossref Full Text | Google Scholar

39. Zhu GM, Dong YD, Zhu RT, Tan YM, Chen DC, Yang G, et al. Value of multimodal MRI functional imaging in predicting molecular subtypes of breast cancer. Gansu Med J. (2025) 44:227–231, 255. doi: 10.15975/j.cnki.1004-2725.2025.03.009

Crossref Full Text | Google Scholar

40. Song L, Li C, and Yin J. Texture analyses using semiquantitative kinetic parameter maps from DCE-MRI: preoperative prediction of HER2 status in breast cancer. Front Oncol. (2021) 11:675160. doi: 10.3389/fonc.2021.675160

PubMed Abstract | Crossref Full Text | Google Scholar

41. Du S, Gao S, Zhang L, Yang X, Qi X, and Li S. Improved discrimination of molecular subtypes in invasive breast cancer: Comparison of multiple quantitative parameters from breast MRI. Magn Reson Imaging. (2021) 77:148–58. doi: 10.1016/j.mri.2020.12.001

PubMed Abstract | Crossref Full Text | Google Scholar

42. Shi JW, Xu JX, Zhou HY, and Wang KX. Relationship between quantitative parameters of dynamic contrast-enhanced magnetic resonance imaging and differentiation degree and molecular subtypes of breast cancer. J Imaging Res Med Appl. (2024) 8:44–7. doi: 10.3969/j.issn.2096-3807.2024.20.013

Crossref Full Text | Google Scholar

43. Derakhshan F and Reis-Filho JS. Pathogenesis of triple-negative breast cancer. Annu Rev Pathol. (2022) 17:181–204. doi: 10.1146/annurev-pathol-042420-093238

PubMed Abstract | Crossref Full Text | Google Scholar

44. Sun Z, Hu S, Shao L, Jin L, Lv Q, Li Y, et al. A pilot study of low-dose CT perfusion imaging (LDCTPI) technology in patients with triple-negative breast cancer. J Xray Sci Technol. (2019) 27:443–51. doi: 10.3233/XST-180465

PubMed Abstract | Crossref Full Text | Google Scholar

45. Linderholm BK, Hellborg H, Johansson U, Elmberger G, Skoog L, Lehtiö J, et al. Significantly higher levels of vascular endothelial growth factor (VEGF) and shorter survival times for patients with primary operable triple-negative breast cancer. Ann Oncol. (2009) 20:1639–46. doi: 10.1093/annonc/mdp062

PubMed Abstract | Crossref Full Text | Google Scholar

46. Park SH, Choi HY, and Hahn SY. Correlations between apparent diffusion coefficient values of invasive ductal carcinoma and pathologic factors on diffusion-weighted MRI at 3. 0 Tesla. J Magn Reson Imaging. (2015) 41:175–82. doi: 10.1002/jmri.24519

PubMed Abstract | Crossref Full Text | Google Scholar

47. Gigli S, Amabile MI, David E, Luca AD, Grippo C, Manganaro L, et al. Morphological and semiquantitative kinetic analyses on dynamic contrast enhanced MRI in triple negative breast cancer patients. Acad Radiol. (2019) 26:620–5. doi: 10.1016/j.acra.2018.06.014

PubMed Abstract | Crossref Full Text | Google Scholar

48. Sharma P, Chida K, Wu RR, Tung K, Hakamada K, Ishikawa T, et al. VEGFA gene expression in breast cancer is associated with worse prognosis, but better response to chemotherapy and immunotherapy. World J Oncol. (2025) 16:120–30. doi: 10.14740/wjon1993

PubMed Abstract | Crossref Full Text | Google Scholar

49. Li Q, Xiao Q, Yang M, Chai Q, Huang Y, Wu P, et al. Histogram analyses of quantitative parameters from synthetic MRI: Correlations with prognostic factors and molecular subtypes in invasive ductal breast cancer. Eur J Radiol. (2021) 139:109697. doi: 10.1016/j.ejrad.2021.109697

PubMed Abstract | Crossref Full Text | Google Scholar

50. Kazama T, Takahara T, Kwee TC, Nakamura N, Kumaki N, Niikura N, et al. Quantitative values from synthetic MRI correlate with breast cancer subtypes. Life (Basel). (2022) 12:1307. doi: 10.3390/life12091307

PubMed Abstract | Crossref Full Text | Google Scholar

51. Zhang Q, Zhao Y, Nie J, Long Q, Wang X, Wang X, et al. Pretreatment synthetic MRI features for triple-negative breast cancer. Clin Radiol. (2024) 79:219–26. doi: 10.1016/j.crad.2023.10.015

PubMed Abstract | Crossref Full Text | Google Scholar

52. Matsuda M, Kido T, Tsuda T, Okada K, Shiraishi Y, Suekuni H, et al. Utility of synthetic MRI in predicting the Ki-67 status of oestrogen receptor-positive breast cancer: a feasibility study. Clin Radiol. (2020) 75:398.e1–8. doi: 10.1016/j.crad.2019.12.021

PubMed Abstract | Crossref Full Text | Google Scholar

53. Gao W, Yang Q, Li X, Chen X, Wei X, Diao Y, et al. Synthetic MRI with quantitative mappings for identifying receptor status, proliferation rate, and molecular subtypes of breast cancer. Eur J Radiol. (2022) 148:110168. doi: 10.1016/j.ejrad.2022.110168

PubMed Abstract | Crossref Full Text | Google Scholar

54. Yang H, Wang W, Cheng Z, Zheng T, Cheng C, Cheng M, et al. Radiomic machine learning in invasive ductal breast cancer: prediction of Ki-67 expression level based on radiomics of DCE-MRI. Technol Cancer Res Treat. (2024) 23:15330338241288751. doi: 10.1177/15330338241288751

PubMed Abstract | Crossref Full Text | Google Scholar