Amide proton transfer-weighted magnetic resonance imaging for the diagnosis of triple-negative breast cancer: a comparative study with diffusion-weighted imaging and dynamic contrast-enhanced magnetic resonance imaging

Introduction

Breast cancer is the most common malignant tumor (excluding nonmelanoma skin cancer) in women, accounting for 32% of all new cancers in women (1). It is highly heterogeneous (2), is associated with a high mortality rate, and thus poses a considerable threat to women’s health and quality of life. Based on immunohistochemical receptor status, breast cancer can be classified into four molecular subtypes: luminal A, luminal B, human epidermal growth factor receptor 2 (HER2)-enriched, and triple-negative breast cancer (TNBC) (3). TNBC is associated with aggressive behavior and poor prognosis, and thus there is an urgent need for a means to accurately and noninvasively characterize breast tumors in the preoperative setting.

In clinical practice, breast imaging commonly consists of mammography (MG), ultrasound (US), and magnetic resonance imaging (MRI). MG and US form the basis of cancer screening, but their sensitivity is limited in dense breast tissue (4). Dynamic contrast-enhanced MRI (DCE-MRI) remains the gold standard for preoperative diagnosis and staging, and with its high sensitivity, provides valuable morphological, hemodynamic, and semiquantitative/quantitative information. However, it requires gadolinium contrast agent injection, which carries the risk of allergic-like reactions, and, more importantly, the risk of nephrogenic systemic fibrosis (NSF) in patients with severe renal insufficiency (5). Diffusion-weighted imaging (DWI) provides noncontrast functional information but is limited in its ability to discriminate between subtypes due to influences from b-values, necrosis, and fibrosis (6). Thus, there is a need for noncontrast techniques that can accurately characterize the biological features of tumors potentially identify specific molecular subtypes.

Chemical exchange saturation transfer (CEST) is an emerging molecular MRI technique that can achieve noninvasive quantitative analysis of various endogenous metabolites in the human body (7-10). Conventional amide proton transfer-weighted (APTw) imaging, which targets amide protons at 3.5 ppm, has shown diagnostic value in brain tumors (11-13). However, this conventional approach typically only provides information for a single parameter—the asymmetric magnetization transfer ratio (MTRasym)—which can be confounded by overlapping signals from semisolid macromolecules [e.g., magnetization transfer (MT)] and nuclear Overhauser enhancement (NOE) effects, potentially leading to inaccurate quantification (14).

Multipool CEST analysis addresses these limitations by fitting the Z-spectrum with a five-pool Lorentzian model to separate signals from five core proton pools: amide (+3.5 ppm and intracellular protein/peptide-NH), Amine (+2.0/+2.75 ppm and small-molecule-NH₂), MT (approximately −1 ppm and semisolid macromolecules), (NOE; −3.5 ppm and nonexchangeable protons), and direct water saturation (DS; approximately 0 ppm) (15). Twelve parameters are derived through stepwise signal purification. Among these, MTRasym (3.5 ppm) reflects the raw asymmetric signal difference at ±3.5 ppm but remains confounded by MT, NOE, and DS (11,16); MTR referenced to the exchange effect (MTRrex) removes DS and MT via five-pool fitting (16); apparent exchange-dependent relaxation (AREX) further normalizes MTRrex via water T₁ to exclude T₁ effects (17); and downfield NOE-suppressed AREX (DNS-AREX) subtracts the scaled NOE contribution to achieve the purest amide-specific signal (18). Two pH indices [pH-weighted and amine concentration-independent detection (AACID)] assess tissue acid-base status through the differential pH sensitivity of amide and amine protons (19). Because AREX is mathematically derived from MTRrex (AREX = MTRrex/T₁, water), these parameters are collinear and cannot be directly compared (17). The formulae and biological correlates of these parameters are provided in Table 1 and the Methods section.

Although APTw imaging has been examined in the context of breast lesions (20,21), its diagnostic value of fully corrected multipool parameters for molecular subtyping remains unclear. Therefore, this study aimed to preliminarily evaluate the ability of multipool CEST to differentiate between benign from malignant breast lesions and to identify TNBC and compared it with DWI and DCE-MRI. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0550/rc).

Methods

Patients

This prospective single-center study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments and was approved by the Ethics Committee of Shanghai Tenth People’s Hospital. Written informed consent was obtained from all participants. From September 2023 to August 2025, patients with clinically suspected breast lesions (based on abnormal imaging or physical examination) were consecutively enrolled at Shanghai Tenth People’s Hospital. The patient inclusion criteria were as follows: (I) no contraindications to MRI; (II) no prior treatment before MRI examination; and (III) histopathological confirmation by biopsy or surgical specimen. Meanwhile, the exclusion criteria were as follows: (I) poor APTw image quality due to motion artifacts or B0/B1 field inhomogeneity (assessed during postprocessing of the CEST acquisition); (II) unclear or inconclusive histopathological diagnosis; (III) lesion size <1 cm (small lesions are susceptible to partial volume effects that may compromise quantitative APTw measurements). Finally, 148 patients with 148 breast lesions were included in the analysis (Figure 1) and underwent MRI examination within 2 weeks prior to biopsy or surgery.

Figure 1 Flowchart of patient enrollment. APTw, amide proton transfer-weighted imaging; DCE, dynamic contrast-enhanced imaging; DWI, diffusion-weighted imaging; HER2-enriched, human epidermal growth factor receptor 2-enriched; MRI, magnetic resonance imaging; TNBC, triple-negative breast cancer.

MRI scanning sequences

All examinations were performed on a 3.0 T MRI scanner (Ingenia CX, Philips, Amsterdam, the Netherlands) equipped with a dedicated 16-channel breast coil. The patients were placed in the prone position, with their feet entering the scanner first. The scanning sequences and their parameters are described below:

• T1-weighted imaging (T1WI) was performed under the following parameters: repetition time (TR) =646 ms, echo time (TE) =8.44 ms, slice thickness =3 mm, field of view (FOV) =240 mm × 369 mm, acquisition matrix =268×368, number of excitations (NEX) =1, and scanning time =2 minutes and 35 seconds.
• T2-weighted imaging (T2WI) was performed under the following parameters: TR =1,500 ms, TE =123 ms, slice thickness =3 mm, FOV =240 mm × 369 mm, acquisition matrix =268×336, NEX =1, and scanning time =2 minutes and 11 seconds.
• DWI was performed under the following parameters: TR =4,000 ms, TE =52 ms, slice thickness =3 mm, FOV =369 mm × 234 mm, acquisition matrix =144×90, b values =0 and 800 s/mm², NEX =1, and scanning time =2 minutes 48 seconds. The b-values of 0 and 800 s/mm² were selected based on established breast DWI protocols that balance adequate diffusion weighting with acceptable signal-to-noise ratio.
• APTw (multi-pool CEST) was performed with a single-slice two-dimensional (2D) acquisition being performed at the level of the largest tumor diameter identified on preliminary localizer images. A total of 64 frequency offsets were acquired, including +300 ppm, 0 ppm, and 31 symmetric pairs ranging from ±100 to ±0.25 ppm (specifically: ±100, ±30, ±25, ±20, ±15, ±10, ±7.5, ±7, ±6.5, ±6, ±5.5, ±5, ±4.75, ±4.5, ±4.25, ±4, ±3.75, ±3.5, ±3.25, ±3, ±2.75, ±2.5, ±2.25, ±2, ±1.75, ±1.5, ±1.25, ±1, ±0.75, ±0.5, and ±0.25 ppm). The +300 ppm offset was used as the unsaturated reference image, as it is sufficiently distant from all known proton exchange frequencies to provide a baseline water signal unaffected by saturation transfer effects. APTw was conducted under the following acquisition parameters: TR =5,000 ms, TE =6.9 ms, saturation time =0.8 s, saturation power =2 µT, slice thickness =6 mm, FOV =180 mm × 320 mm, acquisition matrix =92×159, and total scan time =5 minutes 25 seconds.
• DCE-MRI was performed under the following parameters: TR =3.9 ms, TE =1.87 ms, flip angle of T1 mapping =5/15°, slice thickness =3 mm, acquisition matrix =269×154, and FOV =320 mm × 210 mm. Two mask phases were scanned before contrast agent injection, gadolinium-based contrast agent (GBCA) was injected through the cubital vein with a high-pressure injector at 3.0 ml/s flow rate and a dose of 0.2 mmol/kg, a 20-ml saline flush was then administered, and dynamic enhanced scanning started 35 second after contrast injection, with 50 phases being scanned and a scanning time of 8 minutes and 5 seconds.

Image analysis and postprocessing

All images were uploaded to the IntelliSpace Portal workstation (Philips) on the Ingenia CX system. APTw, DWI, and DCE images were analyzed by two radiologists with 3 and 15 years of experience, respectively, who were blinded to the histopathological results. APTw images were postprocessed via a MATLAB R2022b software package (MathWorks, Natick, MA, USA), and manual region-of-interest (ROI) delineation was performed with the open-source software MRIcroGL.

Multipool CEST parameter calculation

The Z-spectrum was fitted with a five-pool Lorentzian model, and 12 parameters were derived (Table 1). The five raw pool signals were obtained from direct fitting amplitudes and included amide, amine, MT, NOE, and DS. These parameters characterize intracellular protein enrichment, small-molecule metabolism, tissue structural integrity, lipid distribution, and DS, respectively (22).

The remaining seven parameters were progressively corrected metrics that were designed to reduce the influence of confounding factors in CEST imaging. MTRasym (3.5 ppm) (hereafter referred to as MTR3.5) is a basic CEST parameter that reflects mixed extracellular pH and protein content but remains influenced by DS, MT, T1, and NOE effects (11). It was calculated as follows:

MTRasym(MTR3.5)=[Z(−3.5)−Z(+3.5)]Z(−3.5)×100%[1]

The MTRrex is a parameter that reduces the influence of DS and MT while remaining affected by T1 and NOE contributions (16). It was calculated according to the following equation:

MTRrex=1Zlab(Δω)−1Zref(−Δω)[2]

MTRrex was determined by subtracting inverse z-spectra to achieve a compensation for spillover effects and MT correction.

The AREX is a parameter that further corrects for T1 relaxation effects and retains information related to proton exchange rate and solute concentration (17). It was calculated using the following equation:

AREX=MTRrexT1,water[3]

The DNS-AREX is a parameter designed to remove NOE-related confounding effects and improve specificity for amide proton transfer (18). It was calculated as follows:

DNS-AREX=AREX(+Δω)−rrNOE×AREX(−Δω)[4]

The amide/amine ratio (AAR) is a parameter that reflects intracellular acid-base tendency (23). It was calculated using the following equation:

AAR=Amide−AREXAmine−AREX[5]

Where amide-AREX and Amine-AREX are the AREX-corrected values for the amide and amine pools, respectively, representing the proton exchange rate and solute concentration after T₁ relaxation correction.

Two additional pH indices were evaluated. The pH-weighted parameter, defined as MTRasym (3.0 ppm), provides a rapid qualitative assessment of extracellular acidosis (24). The AACID is a parameter that enables accurate intracellular pH quantification by reducing the influence of amide and amine concentration variations (19). It was calculated as follows:

AACID=k×ln(Amide−AREXAmine−AREX)+b[6]

Because AREX is mathematically derived from MTRrex (AREX = MTRrex/T1, water), these two parameters are share a collinear relationship. To avoid statistical redundancy, they were not compared directly in our analysis, nor were they included simultaneously in any regression models. Only MTRrex-amide (i.e., MTRrex measured at the amide frequency offset of +3.5 ppm) and DNS-AREX were retained for amide-transfer analysis.

With the delayed DCE-MRI and T2-weighted images as anatomical guides, a single ROI was manually drawn on the APTw source images (corresponding to the same slice) to encompass the solid, enhancing tumor component. Areas of obvious necrosis, cysts, calcifications, and peripheral edema were carefully excluded. The same ROI was then copied to the corresponding slices of the apparent diffusion coefficient (ADC) and parametric DCE maps [volume transfer constant (Ktrans), rate constant (Kep), and extracellular extravascular volume fraction (Ve)] for consistent measurement. The workstation software automatically extracted the mean (ADC), DCE-MRI parameters (Ktrans, Kep, and Ve), and all 12 CEST-derived metabolic values (including amide, NOE, amine, MT, DS, pH-weighted, AACID, MTRrex-amide (MTRrex measured at the amide frequency offset of +3.5 ppm), AREX-amide (AREX measured at the amide frequency offset of +3.5 ppm), DNS-AREX, AAR, and MTR3.5(as defined above)). To ensure measurement stability, the extraction process was performed three consecutive times, and the mean value was used for analysis.

Statistical analysis

Statistical analysis was performed with SPSS version 26.0 (IBM Corp., Armonk, NY, USA) and MedCalc Statistical Software (MedCalc Software Ltd., Ostend, Belgium). The interobserver reliability for the quantitative measurements was assessed via the intraclass correlation coefficient (ICC), with values greater than 0.75 indicating good reliability. The normality of all continuous data was first assessed with the Shapiro-Wilk test. As the data were confirmed to be nonnormally distributed, nonparametric tests were applied for all subsequent group comparisons. The Mann-Whitney U test was used for comparisons between two groups, and the Kruskal-Wallis H test with the post hoc Dunn test was used for comparisons across breast cancer molecular subtypes. Bonferroni correction was applied separately for different comparisons. For the comparison between benign and malignant lesions, all 12 APTw-derived parameters were tested; the significance threshold was set at P<0.004 (0.05/12). For the differentiation of TNBC from non-TNBC, only four prespecified APTw parameters (DNS-AREX, MTRrex-amide, pH-weighted, and MTR3.5) were evaluated, with the significance threshold was set at P<0.0125 (0.05/4). For the comparison of four molecular subtypes, post hoc pairwise comparisons were adjusted via Bonferroni correction (P<0.0083 for six pairwise comparisons, as indicated in the Results). Receiver operating characteristic (ROC) curve analysis was used to evaluate diagnostic efficacy, with the area under the curve (AUC) comparisons being performed via the DeLong test. For multivariate analysis, binomial logistic regression models were constructed, and the predictive probability values from these models were used as combined diagnostic indices. Internal validation was performed via bootstrap resampling with 1,000 iterations to estimate 95% confidence intervals (CIs) for AUC values and to assess model stability.

Due to the mathematical relationship between MTRrex and AREX (AREX = MTRrex/T₁, water), these two parameters are collinear. Therefore, they were not compared directly or included simultaneously in any regression model. Only MTRrex-amide (as the semiquantitative metric) and DNS-AREX (as the fully corrected metric) were retained for amide-transfer analysis.

Results

Clinical data

This study initially enrolled 195 patients with suspected breast lesions. Among these, 47 cases were excluded: 21 due to motion artifacts compromising APT image quality, 19 due to unclear histopathological diagnosis, and 7 due to lesion size <1 cm. Finally, 148 patients comprising 148 breast lesions were included in this study. The clinical characteristics are summarized in Table 2.

Consistency test

Interobserver agreement was excellent, with ICCs ranging from 0.805 to 0.925.

Differences in APTw, DWI, and DCE quantitative parameters between benign and malignant lesions

Among 12 CEST-derived parameters, after Bonferroni correction was applied for multiple comparisons (threshold P<0.004), only the Amide value remained significantly higher in malignant lesions than in benign lesions (P=0.002). pH-weighted values showed a nominal difference (P=0.046) that did not reach statistical significance after correction (Table 3 and Figure 2). Malignant lesions also demonstrated significantly lower ADC values and higher Ktrans and Kep values (P<0.05).

Figure 2 Comparison of APTw parametric maps between benign and malignant breast lesions. (A) A 44-year-old woman with fibroadenoma: contrast-enhanced T1WI showed a well-defined mass with moderate enhancement. (B) Corresponding amide map, with an amide value of 2.73%. (C) Corresponding pH-weighted map, with a pH-weighted value of 1.37%. (D) A 63-year-old woman with invasive ductal carcinoma (TNBC): contrast-enhanced T1WI showed an irregular mass with heterogeneous enhancement. (E) Corresponding amide map, with an amide value of 6.08%. (F) Corresponding pH-weighted map, with a pH-weighted value of 3.54%. APTw, amide proton transfer-weighted imaging; T1WI, T1-weighted imaging; TNBC, triple-negative breast cancer.

The AUC of Amide was 0.722, higher than that of Ktrans (0.687), while the AUCs of ADC and Kep were 0.861 and 0.830, respectively (Figure 3). The combination of amide, ADC, Ktrans, and Kep yielded the highest AUC of 0.932 (Figure 4), representing a significant improvement over any single parameter (P<0.05).

Figure 3 ROC curves of parameters with statistically significant differences (amide, ADC, Ktrans, and Kep) for differentiating benign from malignant breast lesions. ADC, apparent diffusion coefficient; Kep, rate constant; Ktrans, volume transfer constant; ROC, receiver operating characteristic.

Figure 4 ROC curves for the combined application of statistically significant parameters in distinguishing benign and malignant breast lesions. DCE, dynamic contrast-enhanced imaging; DWI, diffusion-weighted imaging; ROC, receiver operating characteristic.

The performance of APTw, DWI, and DCE parameters in differentiating between four molecular subtypes of breast cancer

For the CEST-derived parameters, DNS-AREX, MTRrex-amide, pH-weighted, and MTR3.5 values differed significantly across the four molecular subtypes of breast cancer (P<0.05) (Table 4). Post hoc pairwise comparison with Bonferroni correction revealed that TNBC had significantly higher DNS-AREX values than did luminal A (P<0.001), while no significant differences were observed between the TNBC and luminal B, TNBC and HER2-enriched subtypes. Moreover, no significant differences were found between subtypes for ADC, Ktrans, Kep, or Ve values.

The performance of APTw, DWI, and DCE parameters in differentiating TNBC from non-TNBC

After Bonferroni correction for multiple comparisons (threshold P<0.0125), only DNS-AREX showed a statistically significant difference between TNBC and non-TNBC. The median DNS-AREX value was 8.02% [interquartile range (IQR): 3.65–32.17%] in TNBC, while it was 0.83% (IQR: −2.30 to 8.98%) in non-TNBC (P=0.008). MTRrex-amide was higher in TNBC (10.25%, IQR: 8.25–13.20%) than in non-TNBC (6.81%, IQR: 3.67–10.68%), but this difference did not reach statistical significance after correction (P=0.015). Similarly, pH-weighted and MTR3.5 showed no significant differences (Table 5). DNS-AREX achieved an AUC of 0.762 for this differentiation, with a sensitivity of 91.7% and a specificity of 68.5% (Figure 5). ADC and Ktrans, Kep, Ve values showed no significant difference between TNBC and non-TNBC (Figures 6,7).

Figure 5 ROC curve for DNS-AREX in differentiating TNBC from non-TNBC breast lesions. DNS-AREX, downfield nuclear Overhauser effect-suppressed apparent exchange-dependent relaxation; ROC, receiver operating characteristic; TNBC, triple-negative breast cancer.

Figure 6 Multiparametric MRI findings in a 52-year-old woman with TNBC. (A) For DCE-MRI, the Ktrans, Kep, and Ve values were 110.29 (min⁻¹), 452.20 (min⁻¹), and 259.39 (min⁻¹), respectively. (B) The APTw-derived DNS-AREX was 35.07(%), indicating elevated mobile protein content. (C) DWI showed marked hyperintensity, indicating restricted water diffusion. (D) The ADC map showed low signal (ADC =1.07×10⁻³ mm²/s), confirming high cellularity. ADC, apparent diffusion coefficient; APTw, amide proton transfer-weighted imaging; DCE-MRI, dynamic contrast-enhanced magnetic resonance imaging; DNS-AREX, downfield nuclear Overhauser effect-suppressed apparent exchange-dependent relaxation; DWI, diffusion-weighted imaging; Kep, rate constant; Ktrans, volume transfer constant; MRI, magnetic resonance imaging; TNBC, triple-negative breast cancer; Ve, extracellular extravascular volume fraction.

Figure 7 Multiparametric MRI findings in a 76-year-old woman with non-TNBC (luminal A) breast cancer. (A) For DCE-MRI, the Ktrans, Kep, and Ve values were 123.92 (min⁻¹), 287.35 (min⁻¹), and 431.26 (min⁻¹), respectively. (B) The APTw-derived DNS-AREX value was 9.98(%). (C) DWI showed marked hyperintensity. (D) The ADC map showed a slightly hypointense signal (ADC =1.60×10⁻³ mm²/s), indicating mild restriction of water diffusion, which is consistent with moderate cellularity. ADC, apparent diffusion coefficient; APTw, amide proton transfer-weighted imaging; DCE-MRI, dynamic contrast-enhanced magnetic resonance imaging; DNS-AREX, downfield nuclear Overhauser effect-suppressed apparent exchange-dependent relaxation; DWI, diffusion-weighted imaging; Kep, rate constant; Ktrans, volume transfer constant; MRI, magnetic resonance imaging; TNBC, triple-negative breast cancer; Ve, extracellular extravascular volume fraction.

Discussion

Diagnostic value of APTw, DWI, and DCE for benign and malignant breast lesions

Our results suggest that among the 12 examined CEST-derived metabolic parameters, amide values were significantly higher in malignant lesions than in benign lesions, demonstrating modest but statistically significant diagnostic efficacy (AUC =0.722), which is consistent with previous studies on APTw imaging of the breast and other sites (21,25-27). This finding can be attributed to the increased intracellular protein concentration and altered microenvironment in malignant tumors, which enhance the chemical exchange between amide protons and water protons.

Our study further demonstrated that malignant lesions exhibited significantly higher Ktrans and Kep values but significantly lower ADC values compared to benign lesions. These findings are consistent with previous reports (21,28). When APTw was combined with ADC and DCE parameters, the diagnostic performance was significantly improved (AUC =0.932). These combined models demonstrated diagnostic efficacy comparable to, and even numerically higher than, conventional DCE-MRI alone. Therefore, for patients with contraindications to gadolinium contrast agents, a noncontrast MRI regimen combining APTw and DWI may serve as a potential alternative, although this speculation remains to be validated in larger studies.

Diagnostic value of APTw, DWI, and DCE for molecular subtypes of breast cancer and TNBC identification

The primary exploratory aim of this study was to assess the utility of APTw imaging in molecular subtyping, particular the identification of TNBC, as it is known for its aggressive behavior, high cellularity, early relapse, metastasis, and poorer survival (29-31). The high cellularity of TNBC, in particular, may contribute to increased mobile protein content and enhanced amide proton transfer, leading to higher APT signals.

Few studies have explored the inconsistency of APTw imaging in distinguishing the different molecular subtypes of breast cancer. Our findings align with those of Zhang et al. and Xu et al. (32,33), who reported higher APT values in TNBC compared to luminal and HER2-enriched subtypes under 3D APTw imaging. However, other studies (34,35) found no significant differences in APT signals among molecular subtypes, likely due to the small sample size.

We also observed that the CEST-derived parameter DNS-AREX was significantly elevated in TNBC compared to non-TNBC lesions, demonstrating a modest discriminative ability (AUC =0.762). In contrast, DWI and DCE-MRI parameters (ADC, Ktrans, Kep, and Ve) showed no significant differences across subtypes, which is consistent with previous reports that highlight the limited utility of these techniques in molecular subtyping (36-38). DNS-AREX is the purest amide proton transfer metric after multistep correction for MT, T₁, and downfield NOE and primarily represents the concentration of mobile proteins/peptides. TNBC is characterized by high cellularity, a high nucleus: cytoplasm ratio, and active protein synthesis, which contribute to an increased mobile protein load within the cells. This enhances the chemical exchange between amide protons and water protons, resulting in elevated DNS-AREX signals. Therefore, DNS-AREX may serve as a noninvasive imaging biomarker for identifying TNBC.

Considerations for clinical implementation

The APTw imaging sequence used in this study required approximately 5 minutes 25 seconds for single-slice acquisition, which is clinically feasible within a standard breast MRI protocol. Postprocessing was performed with a MATLAB-based tool, which currently requires offline processing. Although 3 T MRI scanners are becoming increasingly available, multipool CEST analysis software remains vendor-specific and not yet widely disseminated. Future development of automated, cloud-based postprocessing platforms could facilitate broader clinical adoption. The noncontrast nature of APTw imaging makes it particularly valuable for patients with renal insufficiency or gadolinium allergy, and thus it may serve as a complementary or alternative functional imaging biomarker in breast MRI protocols.

Limitations

This study involved certain limitations that should be addressed: (I) the small sample size, particularly for benign (n=43) and HER2-enriched (n=16) lesions, may limit statistical power and generalizability of the results. Of note, breast MRI is predominantly performed in a high-risk population (39), and this might have influenced the spectrum of lesions included. Larger multicenter studies are needed to validate our findings; (II) the single-slice 2D CEST acquisition may not fully capture intratumoral heterogeneity. Future implementations could incorporate accelerated 3D techniques to enable whole-tumor coverage within clinically feasible scan times; (III) the precise biological interpretation of certain multipool CEST parameters, such as DNS-AREX, which are still in the development stage, remain to be further validated through correlation with histopathological and genomic data.

Conclusions

Multipool CEST-derived parameters demonstrated promising diagnostic value for differentiating benign from malignant breast lesions and, more importantly, for identifying TNBC, achieving superior performance compared to DWI and DCE-MRI. These findings indicate that APTw imaging could serve as a noninvasive imaging biomarker for molecular subtyping, aiding in preoperative planning and patient stratification in breast cancer management.

Acknowledgments

The authors thank Mr. Xiance Zhao, engineer at Philips Healthcare, for his valuable guidance and technical assistance in the acquisition and post processing of the APTw data.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0550/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0550/dss

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0550/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. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This prospective single-center study was approved by the Ethics Committee of Shanghai Tenth People’s Hospital, and written informed consent was obtained from all participants.

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. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin 2024;74:12-49. [Crossref] [PubMed]
2. Bertucci F, Birnbaum D. Reasons for breast cancer heterogeneity. J Biol 2008;7:6. [Crossref] [PubMed]
3. Goldhirsch A, Wood WC, Coates AS, Gelber RD, Thürlimann B, Senn HJ. Panel members. Strategies for subtypes--dealing with the diversity of breast cancer: highlights of the St. Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2011. Ann Oncol 2011;22:1736-47.
4. Elmore JG, Barton MB, Moceri VM, Polk S, Arena PJ, Fletcher SW. Ten-year risk of false positive screening mammograms and clinical breast examinations. N Engl J Med 1998;338:1089-96. [Crossref] [PubMed]
5. Fraum TJ, Ludwig DR, Bashir MR, Fowler KJ. Gadolinium-based contrast agents: A comprehensive risk assessment. J Magn Reson Imaging 2017;46:338-53. [Crossref] [PubMed]
6. Okazawa A, Iima M, Kataoka M, Okumura R, Takahara S, Noda T, Nishi T, Ishimori T, Nakamoto Y. Diagnostic Utility of an Adjusted DWI Lexicon Using Multiple b-values to Evaluate Breast Lesions in Combination with BI-RADS. Magn Reson Med Sci 2024;23:438-48. [Crossref] [PubMed]
7. Ward KM, Aletras AH, Balaban RS. A new class of contrast agents for MRI based on proton chemical exchange dependent saturation transfer (CEST). J Magn Reson 2000;143:79-87. [Crossref] [PubMed]
8. Zhou J, Hong X, Zhao X, Gao JH, Yuan J. APT-weighted and NOE-weighted image contrasts in glioma with different RF saturation powers based on magnetization transfer ratio asymmetry analyses. Magn Reson Med 2013;70:320-7. [Crossref] [PubMed]
9. Lin G, Zhuang C, Shen Z, Xiao G, Chen Y, Shen Y, Zong X, Wu R. APT Weighted MRI as an Effective Imaging Protocol to Predict Clinical Outcome After Acute Ischemic Stroke. Front Neurol 2018;9:901. [Crossref] [PubMed]
10. Debnath A, Gupta RK, Singh A. Evaluating the Role of Amide Proton Transfer (APT)-Weighted Contrast, Optimized for Normalization and Region of Interest Selection, in Differentiation of Neoplastic and Infective Mass Lesions on 3T MRI. Mol Imaging Biol 2020;22:384-96. [Crossref] [PubMed]
11. Zhou J, Tryggestad E, Wen Z, Lal B, Zhou T, Grossman R, Wang S, Yan K, Fu DX, Ford E, Tyler B, Blakeley J, Laterra J, van Zijl PC. Differentiation between glioma and radiation necrosis using molecular magnetic resonance imaging of endogenous proteins and peptides. Nat Med 2011;17:130-4. [Crossref] [PubMed]
12. Zhou J, Zhu H, Lim M, Blair L, Quinones-Hinojosa A, Messina SA, Eberhart CG, Pomper MG, Laterra J, Barker PB, van Zijl PC, Blakeley JO. Three-dimensional amide proton transfer MR imaging of gliomas: Initial experience and comparison with gadolinium enhancement. J Magn Reson Imaging 2013;38:1119-28. [Crossref] [PubMed]
13. Joo B, Han K, Ahn SS, Choi YS, Chang JH, Kang SG, Kim SH, Zhou J, Lee SK. Amide proton transfer imaging might predict survival and IDH mutation status in high-grade glioma. Eur Radiol 2019;29:6643-52. [Crossref] [PubMed]
14. Zhang L, Zhao Y, Chen Y, Bie C, Liang Y, He X, Song X. Voxel-wise Optimization of Pseudo Voigt Profile (VOPVP) for Z-spectra fitting in chemical exchange saturation transfer (CEST) MRI. Quant Imaging Med Surg 2019;9:1714-30. [Crossref] [PubMed]
15. Zhou J, Payen JF, Wilson DA, Traystman RJ, van Zijl PC. Using the amide proton signals of intracellular proteins and peptides to detect pH effects in MRI. Nat Med 2003;9:1085-90. [Crossref] [PubMed]
16. Zaiss M, Bachert P. Exchange-dependent relaxation in the rotating frame for slow and intermediate exchange -- modeling off-resonant spin-lock and chemical exchange saturation transfer. NMR Biomed 2013;26:507-18. [Crossref] [PubMed]
17. Zaiss M, Windschuh J, Paech D, Meissner JE, Burth S, Schmitt B, Kickingereder P, Wiestler B, Wick W, Bendszus M, Schlemmer HP, Ladd ME, Bachert P, Radbruch A. Relaxation-compensated CEST-MRI of the human brain at 7T: Unbiased insight into NOE and amide signal changes in human glioblastoma. Neuroimage 2015;112:180-8. [Crossref] [PubMed]
18. Zaiss M, Windschuh J, Goerke S, Paech D, Meissner JE, Burth S, Kickingereder P, Wick W, Bendszus M, Schlemmer HP, Ladd ME, Bachert P, Radbruch A. Downfield-NOE-suppressed amide-CEST-MRI at 7 Tesla provides a unique contrast in human glioblastoma. Magn Reson Med 2017;77:196-208. [Crossref] [PubMed]
19. McVicar N, Li AX, Gonçalves DF, Bellyou M, Meakin SO, Prado MA, Bartha R. Quantitative tissue pH measurement during cerebral ischemia using amine and amide concentration-independent detection (AACID) with MRI. J Cereb Blood Flow Metab 2014;34:690-8. [Crossref] [PubMed]
20. Meng N, Wang X, Sun J, Han D, Bai Y, Wei W, Wang Z, Jia F, Wang K, Wang M. A comparative study of the value of amide proton transfer-weighted imaging and diffusion kurtosis imaging in the diagnosis and evaluation of breast cancer. Eur Radiol 2021;31:1707-17. [Crossref] [PubMed]
21. Zhuang L, Lian C, Wang Z, Zhang X, Wu Z, Huang R. Breast-lesion assessment using amide proton transfer-weighted imaging and dynamic contrast-enhanced MR imaging. Radiol Oncol 2023;57:446-54. [Crossref] [PubMed]
22. Zhang L, Xu C, Li Z, Sun J, Wang X, Hou B, Zhao Y. Chemical exchange saturation transfer (CEST) magnetic resonance imaging (MRI) quantification of transient ischemia using a combination method of 5-pool Lorentzian fitting and inverse Z-spectrum analysis. Quant Imaging Med Surg 2023;13:1860-73. [Crossref] [PubMed]
23. Mancini L, Casagranda S, Gautier G, Peter P, Lopez B, Thorne L, McEvoy A, Miserocchi A, Samandouras G, Kitchen N, Brandner S, De Vita E, Torrealdea F, Rega M, Schmitt B, Liebig P, Sanverdi E, Golay X, Bisdas S. Correction to: CEST MRI provides amide/amine surrogate biomarkers for treatment‑naïve glioma sub‑typing. Eur J Nucl Med Mol Imaging 2022;49:2454. [Crossref] [PubMed]
24. Harris RJ, Cloughesy TF, Liau LM, Prins RM, Antonios JP, Li D, Yong WH, Pope WB, Lai A, Nghiemphu PL, Ellingson BM. pH-weighted molecular imaging of gliomas using amine chemical exchange saturation transfer MRI. Neuro Oncol 2015;17:1514-24. [Crossref] [PubMed]
25. Zhang N, Kang J, Wang H, Liu A, Miao Y, Ma X, Song Q, Zhang L, Wang J, Shen Z, Xu X. Differentiation of fibroadenomas versus malignant breast tumors utilizing three-dimensional amide proton transfer weighted magnetic resonance imaging. Clin Imaging 2022;81:15-23. [Crossref] [PubMed]
26. Tian S, Chen A, Li Y, Wang N, Ma C, Lin L, Wang J, Liu A. The combined application of amide proton transfer imaging and diffusion kurtosis imaging for differentiating stage Ia endometrial carcinoma and endometrial polyps. Magn Reson Imaging 2023;99:67-72. [Crossref] [PubMed]
27. Yang L, Wang L, Tan Y, Dan H, Xian P, Zhang Y, Tan Y, Lin M, Zhang J. Amide Proton Transfer-weighted MRI combined with serum prostate-specific antigen levels for differentiating malignant prostate lesions from benign prostate lesions: a retrospective cohort study. Cancer Imaging 2023;23:3. [Crossref] [PubMed]
28. Li L, Wang K, Sun X, Wang K, Sun Y, Zhang G, Shen B. Parameters of dynamic contrast-enhanced MRI as imaging markers for angiogenesis and proliferation in human breast cancer. Med Sci Monit 2015;21:376-82. [Crossref] [PubMed]
29. Borri F, Granaglia A. Pathology of triple negative breast cancer. Semin Cancer Biol 2021;72:136-45. [Crossref] [PubMed]
30. Kala C, Athar M, Kala S, Khan L, Jauhari RK, Satsangi A. Clinical and Cyto-Morphological Characterization of Triple Negative Breast Cancer. J Cytol 2019;36:84-8. [Crossref] [PubMed]
31. Makhlouf S, Wahab N, Toss M, Ibrahim A, Lashen AG, Atallah NM, Ghannam S, Jahanifar M, Lu W, Graham S, Mongan NP, Bilal M, Bhalerao A, Snead D, Minhas F, Raza SEA, Rajpoot N, Rakha E. Evaluation of tumour infiltrating lymphocytes in luminal breast cancer using artificial intelligence. Br J Cancer 2023;129:1747-58. [Crossref] [PubMed]
32. Zhang N, Shao X, Xu L, Zhu W, Wang H, Luo R, Yang C, Ye X, Zeng M, Chen C, Yue X, Bi Z, Lu X. Three-dimensional turbo-spin-echo amide proton transfer-weighted and intravoxel incoherent motion imaging mri for triple-negative breast cancer: a comparison with molecular subtypes and histological grades. BMC Cancer 2025;25:465. [Crossref] [PubMed]
33. Xu M, Shan D, Zhang R, Li J, Guo L, Chen X, Qu J. Differentiation of breast cancer subtypes and correlation with biological status using functional magnetic resonance imaging: comparison with amide proton transfer-weighted imaging and diffusion-weighted imaging. Quant Imaging Med Surg 2025;15:6102-17. [Crossref] [PubMed]
34. Liu Z, Wen J, Wang M, Ren Y, Yang Q, Qian L, Luo H, Feng S, He C, Liu X, Wu Y, Luo D. Breast Amide Proton Transfer Imaging at 3 T: Diagnostic Performance and Association With Pathologic Characteristics. J Magn Reson Imaging 2023;57:824-33. [Crossref] [PubMed]
35. Yu T, Li L, Shi J, Gong X, Cheng Y, Wang W, Cao Y, Cao M, Jiang F, Wang L, Wang X, Zhang J. Predicting histopathological types and molecular subtype of breast tumors: A comparative study using amide proton transfer-weighted imaging, intravoxel incoherent motion and diffusion kurtosis imaging. Magn Reson Imaging 2024;105:37-45. [Crossref] [PubMed]
36. Yuan C, Jin F, Guo X, Zhao S, Li W, Guo H. Correlation Analysis of Breast Cancer DWI Combined with DCE-MRI Imaging Features with Molecular Subtypes and Prognostic Factors. J Med Syst 2019;43:83. [Crossref] [PubMed]
37. Li X, Fu P, Jiang M, Zhang J, Tan L, Ai T, Li X. The diagnostic performance of dynamic contrast-enhanced MRI and its correlation with subtypes of breast cancer. Medicine (Baltimore) 2021;100:e28109. [Crossref] [PubMed]
38. Meyer HJ, Wienke A, Surov A. Diffusion-Weighted Imaging of Different Breast Cancer Molecular Subtypes: A Systematic Review and Meta-Analysis. Breast Care (Basel) 2022;17:47-54. [Crossref] [PubMed]
39. Strigel RM, Burnside ES, Elezaby M, Fowler AM, Kelcz F, Salkowski LR, DeMartini WB. Utility of BI-RADS Assessment Category 4 Subdivisions for Screening Breast MRI. AJR Am J Roentgenol 2017;208:1392-9. [Crossref] [PubMed]