The present study was a prospective study conducted in the Oncology Teaching Hospital, Medical City (Baghdad, Iraq) from April 2022 to December 2022. The study protocol was approved by the Ethics Committee of the Institutional Review Board at the College of Medicine, University of Baghdad (approval no. 1458; Baghdad, Iraq).

Inclusion criteria were as follows: i) Consecutive female patients referred for breast MRI, either for screening purposes or for further characterization of a previously identified lesion; ii) presence of a pure non-mass enhancement (NME) lesion on MRI; iii) absence of any associated mass or focal mass-like enhancement and iv) available image-guided core biopsy for histopathological evaluation. Excisional biopsy was performed when indicated by core biopsy findings or at the discretion of the treating physician.

Exclusion criteria: 1- Patients without a definitive histopathological diagnosis (e.g., biopsy not performed, inadequate tissue sample). 2- Cases with borderline findings on core biopsy (such as atypical ductal hyperplasia, lobular neoplasia, or other indeterminate lesions) were excluded if no subsequent surgical pathology report was available.3- Patients lacking sufficient demographic or clinical information in medical records were excluded from final analysis.

Histopathological diagnoses were classified according to the World Health Organization (WHO) Classification of Breast Tumours, 5th Edition (2019) (12).

For each area of NME the data collected are according to the features mentioned in the 5th edition of Breast Imaging Reporting and Data System (BI-RADS) 2013 (12).

Initially, subtracted images were reviewed to identify the NME lesion and evaluate its morphological features, including maximal size, distribution (focal, linear, segmental, regional or diffuse) and internal enhancement pattern (homogeneous, heterogeneous, clumped or clustered ring) and the dynamic curve. For dynamic curve, two components were assessed: i) The initial phase representing the SI observed within the first 2 min, divided into 3 groups as per American College of Radiology BI-RADS 5th edition: Slow, when <50% increase in SI, medium, when 50–100% increase in SI and rapid, when >100% increase in SI occurs within the first 2 min; and ii) the delayed phase represents the enhancement pattern after 2 min or when the curve starts to change. It is grouped as: Persistent, when there is continuous signal increase >10% over time, plateau, when SI does not change over time after its initial rise and remains flat, and washout, when SI decreases >10% after its highest point from its initial rise indicating a suspicious finding.

A superconducting 1.5T MR imaging device (Magnetom Aera; Siemens Healthineers) was used for the breast MRI. All patients were examined in the prone position by applying bilateral 16-channel breast coils and using scout view sagittal protocol localization and T1-weighted pulses.

Fast spin-echo was utilized to collect axial non-fat saturated T1-weighted imaging (T1WI) with the following image parameters: i) Repetition time (TR), 426 msec; ii) echo time (TE), 4.6 msec; iii) slice thickness, 4 mm; iv) field of view (FOV), 380×380 mm; and v) matrix, 307×512. The settings for axial non-fat-suppressed T2-weighted turbo spin-echo were TR, 6,030 msec; TE, 71 msec; FOV, 400×400 mm; matrix, 384×512; and 4 mm slice thickness.

To get axial short T2 Transverse Dixon fat and water, the settings were TR, 9,440 msec; TE, 72 msec; slice thickness, 3 mm; inter-slice gap, 1 mm; FOV, 450×450 mm and matrix, 307×512.

Dynamic contrast study was performed by applying fluoroscence-3D T1WI spectral attenuated inversion recovery sequence with injection of a 0.2 mmol/kg gadopentetate dimeglumine bolus by an automated injector at 3–5 ml/sec via an 18–20 gauge IV cannula in the antecubital vein. The parameters applied were: i) TR, 5.08 msec; ii) TE, 2.39 msec; iii) 1.8 mm slice thickness; iv) no inter-slice gap; v) FOV, 360×360 mm; and vi) matrix, 307×512. Next, a 20 ml saline bolus infusion at 3–5 ml/sec was administered.

The dynamic series consisted of one pre-contrast and five post-contrast acquisitions; each repeated every 90 sec.

Initially, the subtracted images of each examination were evaluated by RadiAnt DICOM viewer 2022.2 (Medixant) to detect the NME lesion and assess its size, distribution and the internal enhancement pattern. Subsequently, the non-subtracted images with pre-contrast T1W image and the four post-contrast images were evaluated. A region of interest was manually placed on the area of highest enhancement within the NME on the first post-contrast series, then time-intensity plots of dynamic images were generated using computer-assisted diagnosis software.

The percentage of the initial enhancement was calculated as follows: (SIpost - SIpre/SIpre) ×100, where SIpre and SIpost are the SIs on the pre- and first post-contrast images, respectively (13).

The SPSS software for windows (version 26; IBM Corp.) was used for all statistical analyses. Observational data was presented in the form of frequencies and percentages. Associations between categorical variables were assessed using the χ² or Fisher's exact test, as appropriate. To estimate malignancy likelihood for lesions exhibiting a specific enhancement pattern (Type I with initial rise), the positive predictive value (PPV) was calculated as the proportion of test-positive cases that were confirmed malignant. As sensitivity and specificity assess overall MRI performance and not sub-classification, applying them to Type I curves, typically benign, can misrepresent diagnostic value. PPV further captures the significance of atypical malignant presentations within this curve type. It reflects real-world decision-making more accurately in this context. A two-sided P<0.05 was considered to indicate a statistically significant difference.

A total of 38 patients met the enrollment criteria; their ages ranged from 26 to 75 years, with a mean age of 45±11.45 years. A total of 12 patients (31.6%) were in the postmenopausal period. The majority of the patients 24 (63.2%), presented with a breast lump, whereas 10 individuals (26.3%) were scheduled for screening. The initial mammography revealed microcalcifications in 12 patients (41.4%) and focal asymmetry in 17 patients (58.6%). Furthermore, 7 patients (18.4%) were referred for further imaging due to the presence of dense breast tissue. The definitive histopathological diagnosis was corroborated through core biopsy in 6 cases (15.8%) and via surgical excision in 32 cases (84.2%). Histopathological evaluation confirmed the presence of malignancy in 26 cases, representing 68.4% of the total, while the remaining 12 cases, accounting for 31.6%, were classified as benign (Fig. 1). Half of the benign lesions were inflammatory (acute and granulomatous mastitis), 4 (33.3%) exhibited fibrocystic changes (Fig. 1A), one of which exhibited atypical ductal hyperplasia and the remaining 2 (16.7%) were benign intraductal papilloma. Among the malignant lesions examined, 12 (46.2%) were classified as ductal carcinoma in situ (DCIS; Fig. 1B), while invasive ductal carcinoma (IDC) accounted for 14 (53.8%; Fig. 1C). Details are presented in Table I.

The homogeneous pattern of NME was significantly more frequent in benign lesions (8/12; 66.7%) compared with that in malignant lesions (8/26, 30.8%) (P=0.038). By contrast, heterogenous and clumped patterns were more frequently observed in malignant tumors 10 (38.5%) and 7 (26.9%), respectively compared with 2 (16.7%) of benign lesions, although the difference was not significant (P>0.05; Table II). There was no significant difference between benign and malignant tumors in terms of the enhancement distribution (P>0.05); however, segmental enhancement was the most frequent enhancement encountered in malignant lesion 10 (38.5%) with a PPV of 83.3% compared with 2 (16.7%) of the benign lesions followed by the regional type 9 (34.6%) with a PPV of 64.3%. When comparing invasive and in situ carcinomas, regional enhancement was significantly associated with IDC, occurring in 8 (57.1%) of IDC cases vs. 1 (8.3%) of DCIS cases (P=0.014). The dynamic curve of malignant lesions was significantly different compared with that of the benign ones in terms of initial upslope and delay phase, P=0.001 (Table III; Fig. 2, Fig. 3, Fig. 4). Benign lesions exhibited slow initial upslope in 7 (58.3%) of the cases while five (41.7%) have medium initial slope. By contrast, rapid initial slope was the feature of 13 (50%) of malignant tumors (P=0.001; Fig. 3). A persistent delay phase was the significant feature of 9 (75%) of the benign lesions compared with only 7 (26.9%) malignant lesions (P=0.005; Fig. 2). None of the benign lesions exhibited rapid initial or washout delay phase.

To further evaluate the 16 cases with persistent delay phase, kinetic analysis of the initial phase was performed. Among the benign lesions, a slow initial phase was observed in 6 cases (66.7%), while 3 (33.3%) exhibited medium initial phase. However, a slow initial phase was also observed in 3 (42.9%) and a medium in 4 (57.1%) of the malignant lesions (P=0.615) (Table IV; Fig. 2C and D). Among the 12 lesions with a plateau delayed phase, 1 out of the 3 benign lesions exhibited slow initial phase (Fig. 3A and B; however, none of the malignant lesions depicted slow initial phase, yet the difference was not significant (P=0.159; Table IV).

As illustrated in Table V, the PPV of the persistent curve is relatively notably high at 43.8%. However, the PPV was diminished by ~10% (33.3%) upon the inclusion of the initial phase upslope of <50%. By contrast, adding initial upslope >50% to plateau curve increased the PPV from 75 to 81.8%. Although the plateau curve achieved 100% PPV enhanced with the addition of initial upslope >100% (Table V), the detection rate was 5 of 9 malignant cases confirmed by histopathology in this category (Table III).