Subcategorization of suspicious non-mass-like enhancement lesions(BI-RADS-MRI Category4)

A total of 122 cases of breast NMLE lesions confirmed by postoperative histology were collected in our hospital from July 2013 to September 2022. Among the 45 cases of benign lesions, there were 7 cases of intraductal papilloma, 9 cases of abscess formation, 8 cases of chronic inflammation, one case of sclerosing adenosis, and one case of fat necrosis, and the rest were proliferative changes with or without fibromatous structure formation and ductal dilatation with periductitis. Among the 77 cases of malignant lesions, there were 29 cases of intraductal carcinoma in situ (5 cases with early invasion), 20 cases of invasive ductal carcinoma, 9 cases of invasive lobular carcinoma, 6 cases of invasive ductal-lobular carcinoma, 4 cases of invasive intraductal carcinoma, 3 case of intraductal papillary carcinoma, and one case of mucinous carcinoma. All of the cases had a single lesion. Inclusion criteria: 1. There was no treatment history before the MRI examination, and the surgical treatment was performed within 2 weeks after the MRI examination. 2. The imaging data were complete, and the dynamic contrast-enhanced scan showed NMLE. 3. The postoperative pathological data were complete. All patients were female, aged 29 to 78 years, with an average of 49.67 ± 11.12 years.

This study was approved by the Ethics Committee of The Eighth People’s Hospital of Jinan and all methods were also performed in accordance with the relevant guidelines and regulations under the committee supervision. Informed consent was obtained from these patients.

All patients underwent preoperative, multiphase dynamic noncontrast, contrast-enhanced, and diffusion-weighted imaging (DWI) MRI scans. A GE 1.5 T HDe superconducting MR imaging system and a 4-channel breast dedicated surface coil were used. Single-shot echo-planar imaging (EPI) for DWI examination was performed using the following settings: a repetition time (TR) = 8400 ms, an echo time (TE) = 93.8 ms, b-values = 0, 800 s/mm², matrix = 128 × 128, and the number of excitations (NEX) = 2. The fat-suppressed T2-weighted imaging (T2WI) fast spin-echo (FSE) sequence was obtained using the following settings: chemical frequency-selective fat saturation, TR = 4660 ms, TE = 89.2 ms, matrix = 320 × 256, and NEX = 2. Both the DWI and T2WI sequences were transaxial scans with consistent positioning. The field of view was 320 mm × 320 mm, the slice thickness was 4 mm, and the interval was 1 mm, with 32 slices covering the entire breast. Dynamic contrast-enhanced scanning was performed using a VIBRANT sequence, with TR = 4.7 ms, TE = 2.2 ms, matrix = 320 mm × 320 mm, and slice thickness = 1.0 mm, covering the breast and axilla. A total of 12 scans were performed, with a single-scan time of 41 s, of which the first was a pre-scan, and at the end of the first scan, a bolus injection of the contrast agent gadolinium-diethylenetriamine pentaacetic acid (Gd-DTPA) at 0.1 mmol/kg was delivered via the cubital vein at a flow rate of 2 mL/s.

All scanned images were sent to the ADW4.3 workstation for post-processing. The morphological characteristics of the lesions were determined based on the dynamic contrast-enhanced images, and the distribution and internal enhancement characteristics of the lesions were observed and recorded according to the BI-RADS method (2013 edition). The T2WI signal characteristics of the lesions were recorded using the pectoralis major muscle as a reference. Time-intensity curve (TIC) measurement: The region of interest was placed in the area with the most significant enhancement, and the measurement was repeated three times. The curve that was most likely to indicate malignancy was selected as the TIC of the lesion, and its early enhancement rate at 1 min was calculated. Referring to the Fischer score an early enhancement rate of less than 50% was defined as grade 0, 50%-100% as grade 1, and > 100% as grade 2. Apparent diffusion coefficient (ADC) measurement: The region of interest was determined according to the images from dynamic contrast-enhanced examination. Three regions of interest were selected for the measurement, and the mean value was taken. Receiver operating characteristic (ROC) curves were used to determine the optimal diagnostic cut-off value for benign lesions, with those less than the optimal cut-off value considered to be malignant. Determination of vascular signs: The subtraction images with the most obvious enhancement were used for maximum intensity projection reconstruction to observe whether there were abnormally increased and thickened vessels in and around the lesion. Blood vessels with a length ≥ 3 cm and a maximum diameter ≥ 2 mm were used as the screening criteria [3], and a difference in blood vessel count ≥ 2 between the two breasts was considered an asymmetrical increase in breast blood supply; if one or more vessels entered the lesion, they were considered the feeding vessels. If one or both of the above criteria were satisfied, then the adjacent vascular sign was considered positive. Acquisition of the characteristics and parameters of all lesions was completed by two collaborating physicians with many years of work experience in breast MRI.

Of the 122 lesions, there were 14 cases of linear distribution, 68 cases of segmental distribution (including distribution along the duct), 23 cases of local distribution, 11 cases of regional distribution, 4 cases of multiple regional distribution, and 2 cases of diffuse distribution. There were 6 cases of homogeneous enhancement, 55 cases of heterogeneous enhancement, 56 cases of clustered enhancement, and 9 cases of clustered ring enhancement. There were 3 cases of hypointense T2WI signal, 36 cases of isointense signal, 51 cases of slightly hyperintense signal, and 32 cases of hyperintense signal. There were 7 cases with ductal dilatation, 8 cases with adjacent skin thickening, and 5 cases with nipple retraction. See Table 1 for details.

The ADC values of the 81 lesions ranged from 0.63 to 2.03 × 10⁻³ mm²/s; with significantly higher values for benign lesions than for malignant lesions (1.37 ± 0.43 VS 1.14 ± 0.25, t = 3.82, P < 0.001); the optimal cut-off value of ROC analysis was 1.40 × 10⁻³ mm²/s, the area under the curve was 0.673, the sensitivity was 85.71 (95% confidence interval (CI): 75.9–92.6%), and the specificity was 46.67% (95% CI: 31.7–62.1%). See Fig. 1. The early enhancement rate ranged from 20 to 196%. In particular, the early enhancement rate was < 50% in 3 cases, 50 – 100% in 40 cases, and > 100% in 79 cases. There were 26 cases of inflow curve(type I), 80 cases of plateau curve (type II), and 16 cases of washout curve (type III). For details, see Table 1.

The distribution, T2WI signal, internal enhancement pattern, early enhancement rate, enhancement curve, ADC value, and vascular sign of the lesions were included in the logistic multivariate analysis. It was found that the distribution (odds ratio (OR) = 8.70), internal enhancement pattern (OR = 6.29), ADC value (OR = 4.56), and vascular sign (OR = 2.84) of the lesions were closely related to the benignity and malignancy of the lesions, and the differences had statistical significance (P < 0.05). The above signs were included in the multimodal scoring: 4 point for a segmental lesion, and 0 points for lesions with other distributions;3 point for clustered enhancement, and 0 points for other enhancement types; 2point for ADC with a value less than 1.40, otherwise 0 points; and 1 points for the presence of vascular signs, otherwise 0 points. ROC analysis revealed that its optimal diagnostic cut-off value was 5. ROC analysis revealed that the diagnostic specificity and sensitivity were 87.01% and 82.22%, respectively. All lesions were subjected to multimodal scoring, and the details are shown in Table 2 and Fig. 2. Lesions with 1–6 points were considered BI-RADS category 4 lesions, and the positive predictive values of subtypes 4a, 4b, and 4c lesions were 15.79%, 31.25%, and 77.78%, respectively. For details, see Table 3 and Figs. 3 and 4.