Significant MRI indicators of malignancy for breast non-mass enhancement

Abstract

OBJECTIVE:

To explore and evaluate new malignant predictors of breast non-mass enhancement lesions using the new BI-RADS MRI lexicon.

METHODS:

A dataset involving 422 consecutive women underwent breast 3.0 T MRI between January 2014 and July 2016 was assembled for this study. Each case was retrospectively reviewed by 3 radiologists. Eighty-four lesions that present non-mass enhancement in 79 patients were identified in the study. Dynamic contrast-enhanced MRI features were analyzed using univariate and multivariate analyses to identify significant indicators of malignancy.

RESULTS:

Of 84 non-mass enhancement lesions, 52 (61.9%) were malignant and 32 (38.1%) were benign. Segmental distribution (P = 0.015 from univariate analysis; OR = 4.739, P = 0.008 from multivariate analysis), cluster ring enhancement (P = 0.017 from univariate analysis; OR = 3.601, P = 0.032 from multivariate analysis), time-intensity curve of plateau (P = 0.002 from univariate analysis; OR = 3.525, P = 0.027 from multivariate analysis) and phase to peak (P = 0.06 from univariate analysis; OR = 6.327, P = 0.015 from multivariate analysis) were significantly different between malignant and benign lesions.

CONCLUSIONS:

This study demonstrated that segmental distribution, clustered ring enhancement, and short time to peak could act as new malignant predictors for breast non-mass enhancement detected on 3.0 T MRI.

Keywords

Breast non-mass enhancement magnetic resonance imaging breast imaging reporting and data system

1 Introduction

In 2003, the American College of Radiology (ACR) developed the Breast Imaging Reporting and Data System (BI-RADS) lexicon for MRI, because of the wide application of breast magnetic resonance imaging (MRI) [1]. The BI-RADS lexicon can be used to predict benign and malignant disease, eliminate ambiguity, allow automated data collection, and facilitate concise communication with referring physicians and radiologists across facilities [2 –4]. With application of BI-RADS MRI lexicon, suspicious areas are classified as focus, mass, or non-mass-like enhancement. Focus is a tiny dot of enhancement smaller than 5 mm that does not clearly represent a space-occupying lesion. The term mass here applies to a 3-dimensional space-occupying lesion ≥ 5 mm, usually spherical or ball-like, which may displace or retract surrounding breast tissue. Non-mass-like enhancement is enhancement of an area that is neither a mass nor a blood vessel, and does not has space-occupying effect.

In the general population, the prevalence of non-mass-like enhancement was much lower than mass enhancements (13% versus 76%) [5]. However, 57% of non-palpable invasive cancers presented with non-mass-like enhancements in image manifestation [6]. In this way, non-mass-like enhancement has posed a great challenge to the effective use of breast MRI [7]. Previous studies have shown that the BI-RADS has great value in predicting benign or malignancy for masses based on characteristics of lesions. However, the MRI characteristics of malignancy for non-mass-like enhancement still remain unclear [7 –10].

The hardware and software development in breast MRI enabled simultaneous acquisition of higher spatial and higher temporal resolution images, which improved detail visualization of MRI manifestations [11 –13]. The BI-RADS MRI lexicon was revised accordingly, and the fifth edition was launched in 2013. New terms were added to improve the description of lesions seen on current pulse sequences, while some terms that were difficult to assess or had a poor inter-observer agreement were removed [14]. In the latest edition, “non-mass enhancement” (NME) replaced the term “non-mass-like enhancement.” For the distribution patterns, the term “ductal” was eliminated, and the term “linear” was described as enhancement in a line, with or without branching. The internal enhancement patterns were classified as homogeneous, heterogeneous, clumped, and cluster ring. Some terms, such as stippled, punctuate, reticular, and dendritic enhancement, were removed [4, 15].

So far, few studies have explored the characteristics of NME using the fifth edition BI-RADS MRI lexicon [10, 16]. The purposes of our study were to analyze the dynamic contrast enhancement (DCE) features of NME as detected by 3.0 T MRI and to identify significant predictors of malignancy according to the latest BI-RADS MRI lexicon.

2 Materials and methods

2.1 Study population

This study was approved by the medical ethics committee of the Second Affiliated Hospital of Xi’an Jiaotong University. Because this is a retrospective study, and all the cases used in this study were collected from the server of PACS, written informed content from each patient was waived. Between January 2014 and July 2016, 422 consecutive women underwent breast 3.0 T MRI in our hospital, and met the following criteria: 1) Insufficient or equivocal mammographic or sonographic findings; 2) Preoperative staging; 3) Nipple discharge; 4) Palpable breast mass but non-definitive mammography or sonography. After a review of all MR images by consensus among three radiologists, the patients who met the following indications were enrolled in the study. (a) MR examination for analysis in the study was performed before treatment. (b) The lesions were pathologically diagnosed by means of surgery or core needle biopsy. (c) MRI showed NME lesions.

2.2 MRI protocol

All breast MRIs were performed on 3.0 T system (Signa HDxt; GE Medical Systems, Milwaukee, WI, U.S.) using a dedicated breast phased-array coil on patients in the prone position. MR sequences was as follows: Fast spin-echo (FSE) T₁-weighted imaging on the transverse plane, short T₁ inversion recovery (STIR) imaging on the transverse plane, and T₂-weighted imaging with fat suppression on the sagittal plane, and T₁-weighted fat-suppressed DCE MRI sequences before and after contrast administration. The DCE MR images were taken before and 64 s, 128 s, 192 s, 256 s, and 318 s after the injection of 0.2 mmol/kg body weight of gadolinium-DTPA (Magnevist, Schering, Germany) at a rate of 2.0 ml/s followed by 15 ml saline solution. T₁-weighted 3D fast spoiled gradient-recalled echo sequence with parallel imaging (VIBRANT) sequence on the transverse plane (TR/TE/TI = 4.4/2.1/125 ms, flip angle 14°, matrices = 416×320, section thickness/spacing = 1.6/0 mm, and field of view = 350×350 mm) was used.

2.3 MRI interpretation

All MR images were independently reviewed by two radiologists (F.LL. and J.X. with 3 and 7 years’ experience in breast imaging, respectively) based on the fifth edition of BI-RADS MRI lexicon [14]. Disagreements were evaluated by a third radiologist (Y.QX. with 7 years’ experience in breast imaging and 25 years’ experience in MRI, respectively). All of them were unaware of any clinical information, including any mammographic, sonographic, or pathological results.

First, NME lesions were identified from background parenchymal enhancement (BPE) on post-contrast axial images acquired at 128 s. Second, the distribution patterns were classified as focal, linear, segmental, regional, multiple regions, and diffuse on post-contrast axial images, maximum intensity projection (MIP) images, and sagittal multi-planar reconstruction (MPR) images acquired at 128 s. Third, the internal enhancement patterns were classified as homogeneous and heterogeneous; heterogeneous enhancement with “cobblestone pattern” or with varied cluster ring component detected on any image of 5 post-enhanced phases were then classified into the clumped or cluster ring pattern, respectively [4 , 16].

A time-intensity curve (TIC) was produced using a GE Medical Systems Functool workstation. The region of interest (ROI) was manually placed over the most markedly enhanced area on the post-contrast image acquired at 128 s. ROI size varied according to the size of lesion, and it was always greater than 3 pixels. The TIC was classified as persistently enhancing (type I), plateau (type II), and washout (type III). The criterion for the plateau pattern was the delayed signal intensity changed equal or less than 10%. The criterion for the washout pattern was the delayed signal intensity decreased more than 10% [17]. Furthermore, the following semi-quantitative parameters were calculated: 1) initial signal enhancement ratio (SER_intial): SER_intial = (SI_early-SI_pre)/SI_pre, where SI_early and SI_pre represented the signal intensity at the first post-contrast phase and the pre-contrast phase, respectively. 2) Peak of enhancement (PE): PE = SIpeak – SIpre, where SIpeak was the maximum signal intensity. 3) Phase of peak (PP): the phase with maximal signal intensity.

2.4 Statistical analysis

Data were analyzed using the SPSS19.0 software package (SPSS Inc, Chicago, IL, U.S.). Probability values of less than 0.05 were considered significant. SER_intial and PE were summarized by their mean and standard deviation, and malignant and benign lesions were compared using the independent sample T-test. For analysis of the difference in frequency of each MR finding between benign and malignant lesions, Fisher’s exact test or chi-square test was used. The factors that were significantly associated with malignancy in univariate analysis were then entered together into multivariate analysis using the logistic regression model, and odds ratios (ORs), 95% confidence intervals (CIs), and P values were calculated.

3 Results

3.1 Lesion histology and type

Here, 84 lesions presenting NME in 79 patients (age range 33–79 years; mean age 51.4 years) were enrolled in the study. The pathological diagnosis was established by core biopsy (n = 10), excisional biopsy (n = 7), or examination of lumpectomy or mastectomy specimens (n = 67). Among the 84 NME lesions, 52 (61.9%) were malignant and 32 (38.1%) were benign (see Table 1 for pathological subtypes). Among the 52 malignancies, 24 (46.2%) were invasive ductal carcinomas (IDC), 20 (38.5%) ductal carcinomas in situ (DCIS), 5 (9.6%) invasive lobular carcinomas (ILC), 2 (3.8%) inflammatory breast cancers, and 1 (1.9%) papillary carcinoma. Among the 32 benign lesions, 16 (50.0%) were fibrocystic hyperplasia, including 7 combined with ductal epithelium hyperplasia, 6 combined with ductal ectasia and focal metaplasia apocrine, 3 combined with focal fibroma-like hyperplasia; 12 (37.5%) were intraductal papilloma; and 4 (12.5%) were chronic purulent inflammation.

Table 1
Histological composition of malignant and benign non-mass enhancement lesions

Malignant (n = 52) Benign (n = 32)

IDC 24 Fibrocystic hyperplasia 16

DCIS 20 Intraductal papilloma 12

ILC 5 Chronic purulent inflammation 4

Inflammatory breast cancer 2

Papillary carcinoma 1

Malignant (n = 52)		Benign (n = 32)
IDC	24	Fibrocystic hyperplasia	16
DCIS	20	Intraductal papilloma	12
ILC	5	Chronic purulent inflammation	4
Inflammatory breast cancer	2
Papillary carcinoma	1

IDC, invasive ductal carcinoma; DCIS, ductal carcinoma in situ; ILC, invasive lobular carcinoma.

3.2 Distribution patterns

Table 2 shows the frequencies of distribution patterns. Among the NME lesions, segmental (Fig. 1) was the most frequent distribution pattern (35/84, 41.7%), followed by regional (Fig. 2) (18/84, 21.4%), linear (Fig. 3) (10/84, 11.9%) and multiple regions (10/84, 11.9%). The most frequent pattern among malignancy was segmental distribution (27/52, 51.9%). The positive predictive value (PPV) of segmental distribution for malignancy was 77.1% (27/35). The frequency of segmental enhancement in malignancy was significantly higher than in benign lesions (P = 0.015).

Table 2
Distribution patterns of non-mass enhancement of malignant and benign lesions

Distribution pattern Malignant (n = 52) Benign (n = 32) P-value

Focal 3 (5.8%) 4 (12.5%) 0.419

Linear 5 (9.6%) 5 (15.6%) 0.495

Segmental 27 (51.9%) 8 (25.0%) 0.015

Regional 8 (15.4%) 10 (31.3%) 0.085

Multiple-region 6 (11.5%) 4 (12.5%) 1.000

Diffuse 3 (5.8%) 1 (3.1%) 1.000

Distribution pattern	Malignant (n = 52)	Benign (n = 32)	P-value
Focal	3 (5.8%)	4 (12.5%)	0.419
Linear	5 (9.6%)	5 (15.6%)	0.495
Segmental	27 (51.9%)	8 (25.0%)	0.015
Regional	8 (15.4%)	10 (31.3%)	0.085
Multiple-region	6 (11.5%)	4 (12.5%)	1.000
Diffuse	3 (5.8%)	1 (3.1%)	1.000

Fig.1

51-year-old woman with IDC. (a) Transverse the third contrast-enhanced image of right breast shows segmental cluster ring enhancement in the outer quadrant. (b) Image of the same slice showing ROI to measure TIC of the lesion. (c) The TIC is a plateau curve with a peak of enhancement of 1700.05 in the third phase.

Fig.2

50-year-old woman with DCIS. (a) Transverse the third contrast-enhanced image of right breast shows regional heterogeneous enhancement with spiculation. (b) Image of the same slice showing ROI to measure TIC of the lesion. (c) The TIC is a plateau curve with a peak of enhancement of 1950.15 in the third phase.

Fig.3

47-year-old woman with fibrocystic hyperplasia. (a) Transverse the third contrast-enhanced image of left breast shows linear clumped enhancement. (b) Image of the same slice showing ROI to measure TIC of the lesion. (c) The TIC is a persistent curve with a peak of enhancement of 1620.12 in the fifth phase.

3.3 Internal enhancement patterns

Table 3 shows the frequencies of internal enhancement patterns. Among the NME lesions, cluster ring enhancement was the most frequently observed internal enhancement (Fig. 1) (29/84, 34.5%), followed by heterogeneous enhancement (26/84, 31.0%) (Fig. 2) and clumped enhancement (16/84, 19.0%) (Fig. 3). The most frequently observed internal enhancements among malignant lesions were cluster ring enhancement (23/52, 44.2%). The PPV of cluster ring enhancement in the diagnosis of malignancy was 79.3% (23/29). The frequency of cluster ring enhancement in malignancy was significantly higher than in benign lesions (P = 0.017).

Table 3
Internal enhancement patterns of non-mass enhancement of malignant and benign lesions

Internal enhancement Malignant (n = 52) Benign (n = 32) P-value

Homogeneous 5 (9.6%) 8 (25.0%) 0.070

Heterogeneous 14 (26.9%) 12 (37.5%) 0.309

Clumped 10 (19.2%) 6 (18.8%) 0.905

Cluster ring 23 (44.2%) 6 (18.8%) 0.017

Internal enhancement	Malignant (n = 52)	Benign (n = 32)	P-value
Homogeneous	5 (9.6%)	8 (25.0%)	0.070
Heterogeneous	14 (26.9%)	12 (37.5%)	0.309
Clumped	10 (19.2%)	6 (18.8%)	0.905
Cluster ring	23 (44.2%)	6 (18.8%)	0.017

3.4 TIC features and semi-quantitative parameters of DCE

Table 4 shows the frequencies of TIC features. The plateau pattern (type II) was observed significantly more frequently in malignancy than in benign lesions (P = 0.002). The PPV of the plateau pattern diagnosing malignant NME lesions was 74.1% (40/54). The frequency of the persistent pattern (type I) in benign lesions was significantly higher than in malignant lesions (P = 0.000). The PPV of the persistent pattern diagnosing benign lesions was 66.7% (18/27). The highest enhanced intensity of most malignant lesions appeared at the third phase (192 s after injection of contrast agent). However, the highest enhanced intensity of most benign lesions usually appeared at the fifth phase (318 s after injection of contrast agent). There were significant differences (P = 0.006 and P = 0.003, respectively).

Table 4
DCE features of non-mass enhancement of malignant and benign lesions

DEC feature Malignant (n = 52) Benign (n = 32) P-value

TIC 0.000

Type I 9 (17.3%) 18 (56.3%) 0.000

Type II 40 (76.9%) 14 (43.8%) 0.002

Type III 3 (5.8%) 0 0.284

PP 0.003

Phase 1 (64 s) 0 0

Phase 2 (128 s) 11 (21.2%) 3 (9.4%) 0.160

Phase 3 (192 s) 19 (36.5%) 3 (9.4%) 0.006

Phase 4 (256 s) 12 (23.1%) 10 (31.3%) 0.403

Phase 5 (318 s) 10 (19.2%) 16 (50.0%) 0.003

DEC feature	Malignant (n = 52)	Benign (n = 32)	P-value
TIC			0.000
Type I	9 (17.3%)	18 (56.3%)	0.000
Type II	40 (76.9%)	14 (43.8%)	0.002
Type III	3 (5.8%)	0	0.284
PP			0.003
Phase 1 (64 s)	0	0
Phase 2 (128 s)	11 (21.2%)	3 (9.4%)	0.160
Phase 3 (192 s)	19 (36.5%)	3 (9.4%)	0.006
Phase 4 (256 s)	12 (23.1%)	10 (31.3%)	0.403
Phase 5 (318 s)	10 (19.2%)	16 (50.0%)	0.003

DCE, dynamic contrast enhancement; TIC, time-intensity curve; PP, phase of peak.

The mean SER_intial of malignant and benign NME lesions were 2.21±1.19 and 1.86±0.94, respectively (Fig. 4). The mean PE of malignant and benign NME lesions were 1812.11±286.56 and 1790.36±421.01, respectively (Fig. 4). There were no significant differences between malignant and benign lesions (P = 0.183 and P = 0.792, respectively).

Fig.4

Box plots of initial signal enhancement ratio (SER_intial), and peak of enhancement (PE) for malignant and benign non-mass enhancement lesions. The horizontal lines inside the boxes represent the median, the boxes represent the interval between the 25th and 75th percentiles, the whiskers represent the 10th and 90th percentiles, and the dot beyond the whiskers represent outliers beyond the 10th and 90th percentiles. SER_intial and PE do not differ significantly between malignant and benign non-mass enhancement lesions (P = 0.183 for SER_intial and P = 0.792 for PE).

3.5 Multivariate analysis

In multivariate analysis, segmental distribution, cluster ring enhancement, type II TIC and the third phase to peak were still found to be significant predictors of malignancy in non-mass enhancement lesions (Table 5). The probability of malignancy for patients with segmental enhancement, cluster enhancement, type II TIC, or the third phase to peak were 4.739, 3.601, 3.525, and 6.327 times that of patients without these features, respectively (P = 0.008, 0.032, 0.027, and 0.015).

Table 5
Results of multivariate analysis

MRI features Malignant Benign B Wald P-value OR 95% CI

(n = 52) (n = 32)

Segmental 27 8 1.556 7.100 0.008 4.739 1.509–14.83

Non-segmental 25 24

Cluster ring 23 6 1.281 4.587 0.032 3.601 1.115–11.631

Non-cluster ring 29 26

Type II TIC 40 14 1.260 4.922 0.027 3.525 1.158–10.730

Non-type II TIC 12 18

3rd phase 19 3 1.845 5.867 0.015 6.327 1.422–28.150

Non-3rd phase 33 29

MRI features	Malignant	Benign	B	Wald	P-value	OR	95% CI
Segmental	27	8	1.556	7.100	0.008	4.739	1.509–14.83
Non-segmental	25	24
Cluster ring	23	6	1.281	4.587	0.032	3.601	1.115–11.631
Non-cluster ring	29	26
Type II TIC	40	14	1.260	4.922	0.027	3.525	1.158–10.730
Non-type II TIC	12	18
3rd phase	19	3	1.845	5.867	0.015	6.327	1.422–28.150
Non-3rd phase	33	29

TIC, time-intensity curve.

4 Discussion

In our study, the DCE features of breast NME lesions detected on 3.0T MRI were analyzed to identify significant predictors of malignancy based on the fifth BI-RADS MRI lexicon. The study revealed that segmental distribution, clustered ring internal enhancement, the plateau pattern of TIC, and short time to peak were significant indicators of malignancy for breast NME.

NME was defined as an enhancing area that is not a mass and is different from BPE. It may extend over a small or large region in the breast and may contain interspersed fat [4]. It should be distinguished from BPE. Typical BPE is bilateral, symmetrical, and diffuse in distribution. The degree of enhancement is usually minimal or mild, with slow initial and persistent kinetic curve [18]. These features generally pose no difficulty to interpretation. However, BPE sometimes presents with an asymmetric, focal, or regional distribution. Enhancement was occasionally moderate or marked in degree and rapid early contrast agent uptake. It was unusual for BPE to display plateau or washout kinetic curves; however, these more suspicious kinetic curves have been reported [2 , 19–21] in benign mastopathic conditions such as focal fibrocystic areas and sclerosing adenosis, and these underlying benign histologic entities may be present within uninvolved normal enhancing tissue. When BPE presented with a more focal, asymmetric, or regional distribution, it was difficult to distinguish from NME. In our study, regardless of distribution, if enhancement was symmetric or showed slow initial enhancement under than 50% with persistent kinetic curve, it was considered BPE. All patients were followed up for at least 6 months, and none of the enhancements that had been classified as BPE became positive lesions.

Segmental distribution was the most frequently observed NME distribution pattern (41.7%). It was also the most common distribution pattern among malignancy (51.9%). As reported in the literature, the PPV of segmental enhancement was the highest for malignant NME, ranging from 67% to 100% [11 , 23]. In this study, the frequency of segmental enhancement in malignancy was significantly higher than in benign lesions (P = 0.015), and the PPV diagnosing malignancy was 77.1%. According to the results of multivariate analysis, the probability of malignancy for patients with segmental enhancement was 4.739 times that of patients without it (P = 0.008). Our results were closely consistent with those published previously. Other distribution patterns such as focal, linear, regional, and multiple region, were detected in both malignant and benign lesions with no significant differences. Some researchers have reported the PPV of ductal distribution to be 20% – 50% [11, 24]. Thomassion-Naggara et al. reported the overall linear and ductal distributions to be 33.33% in malignancies and 46.67% benign lesions [20]. All of them analyzed NMEs based on the previous editions of BI-RADS lexicon. In the fifth edition, the ductal distribution was eliminated, instead included in the term “linear”. Cho et al. reported that the linear pattern was most common in malignant NMEs with the PPV of 95.1% based on the fifth edition. However, their research was performed on a patient with newly diagnosed breast cancer, which was different from ours. Diffuse NME was first reported in inflammatory breast cancer, with cluster ring internal enhancement, skin thickening, and nipple retraction.

In the fifth edition BI-RADS lexicon, the internal enhancement of NME was revised further, with addition of the term cluster ring and deletion of the term stippled, punctuate, reticular, and dendritic. Our study demonstrated that the most frequent internal enhancement both for NMEs and malignant lesions was cluster ring (34.5% and 44.2%, respectively). The frequency of cluster ring was significantly different between malignant and benign lesions (P = 0.017). The PPV for malignancy was 79.3%, which was higher than Uematsu’s result (77%) [11] but lower than Tozaki’s (96%) [16]. These differences may be attributed to any of three causes: the various sample size (124 cases in [11], 61 cases in [16] and 84 cases in our study), different internal enhancement classification based on different editions of BI-RADS lexicon, and different methods of assessment. Tozaki et al. found two types of clustered ring enhancement [16]. They called the ringlike enhancement pattern that appeared within 100 s a subtype of clumped enhancement and the ringlike enhancement pattern produced by washout as truly clustered ring enhancement. Uematsu et al. found that all clustered ring enhancement detected by axial images in the late dynamic phase and 14% of clustered ring enhancement were detectable solely in sagittal images [11]. Both Uematsu and Tozaki concluded that clustered ring enhancement should be evaluated in the late dynamic phase. However, our results demonstrate that axial images combined with 1.4–3 mm sagittal MPR images could help improve the detection of clustered ring enhancement. Furthermore, all lesions, including clustered ring enhancements, could be detected in the third phase (192 s after injection contrast agent) of dynamic contrast enhancement, which differed significantly from the previous study.

We first analyzed the DCE features of NME combined TIC and semi-quantitative parameters. Our study demonstrated that the TIC patterns and the phase of peak differed significantly between malignant and benign lesions. The PPV of the plateau curve indicating malignancy was 74.1%. In most malignant NME lesions, the time of peak was 192 s; while a kinetic curve showing persistent enhancement continuing throughout the entire time period was commonly observed in benign lesions. There difference was significant (P = 0.003). The probability of malignancy for patients with type II TIC or the third phase to peak was 3.525 and 6.327 times that of patients without these features, respectively (P = 0.027 and 0.015). Bluemke et al. set both plateau and washout patterns as indicators of malignancy, which yielded a sensitivity of 63.2% and a specificity of 65.4% [26]. We saw the washout pattern rarely among NME lesions, and only 3 malignant lesions showed it. One reason for this might be that the NME lesions, especially those with heterogeneous internal enhancement (including clumped and cluster ring enhancement), usually contain fewer enhanced components interspersed between the enhancing tissues, which induced a partial volume effect on the measurement of mean signal intensity. Other studies also indicated that the rapid enhancement and washout pattern were unreliable for diagnosis of DCIS [19] and ILC [19, 27], appearing as a non-mass-like enhancement. There was also considerable overlap between malignant and benign lesions for the plateau pattern. Hence, enhancement kinetics for NME should be interpreted with caution and used by combination with distribution and internal enhancement patterns. A kinetic curve reaching the peak around 3 min was a significant indicator for malignant NME lesions (P = 0.006), with an overall PPV of 83.3%. The initial signal enhancement ratio and peak enhancement were, however, not significantly different (P = 0.183 and P = 0.792, respectively). Because there is a normal breast parenchyma in the NME lesion, the tumor can easily derive nutrition from the normal parenchyma, and there is less tumor neovascularization. Therefore, dynamic enhancement parameters for the differential diagnosis NME lesions had limited value.

Our study has a limitation. Our hospital is one of the area’s key provincial hospitals. Most of patients who underwent breast MRI in our hospital were clinically suspected of malignancy with unclear mammography or ultrasound diagnosis or suspected of breast cancer after preoperative evaluation. Some NMEs showed suspicious lesions found by clinical or other examination, and some NMEs involved extra lesions incidentally detected by MRI. In addition, the asymmetrically distributed NMEs which showed slow initial and persistent enhancement were considered BPE. This may be why there were fewer benign lesions than malignant lesions. The difference in size between malignant and benign lesions is considerable and may have caused statistical selection bias. Further research should be performed in the screening population.

5 Conclusions

The combination of axial and sagittal imaging, 3D construction technical (MPR and MIP) and TIC can facilitate accurate interpretation of non-mass enhancement features in 3.0T MRI. Segmental distribution, clustered ring enhancement, and short time to peak were here found to be significantly associated with malignancy.

6 Declarations

6.1 Abbreviations

MRI, magnetic resonance imaging; ACR, American college of radiology; BI-RADS, breast imaging reporting and data system; NME, non-mass enhancement; DCE, dynamic contrast enhancement; FSE, fast spin-echo; STIR, short T₁ inversion recovery; VIBRANT, T₁-weighted 3D fast spoiled gradient-recalled echo sequence with parallel imaging; BPE, background parenchymal enhancement; MIP, maximum intensity projection; MPR, multi-planar reconstruction; TIC, time-intensity curves; ROI, regions of interest; SERintial, initial signal enhancement ratio; SI, signal intensity; PE, peak of enhancement; PP, phase of peak; OR, odds ratio; CI, confidence interval; IDC, invasive ductal carcinoma; DCIS, ductal carcinoma in situ; ILC, invasive lobular carcinoma; PPV, positive predictive value.

6.2 Ethics approval and consent to participate

This study was conducted in accordance with the Helsinki Declaration and was approved by the medical ethics committee of the second affiliated hospital of Xi’an Jiaotong University with the number 2017002. Because this is a retrospective study, and all the cases used in this study were collected from the server of PACS, written informed content from each patient was waived.

6.3 Consent for publication

Not applicable.

6.4 Availability of data and materials

The data sets collected and/or analyzed during the current study are available from the corresponding author upon reasonable request.

6.5 Competing interests

The authors and involved institutions have no conflicts of interest.

6.6 Funding

This work was supported by the National Natural Science Foundation of China (grant number 30701008).

6.7 Authors’ contributions

QXY participated in its design, MRI assessment, manuscript editing, and approval for important intellectual concepts. XJ and LLF carried out MRI assessment, participated in the literature research, and helped to draft the manuscript. LZ participated in the literature research and manuscript editing. XQZ participated in MRI imaging data acquisition and data analysis. QW participated in the design of the study and helped performed statistical analysis. XC developed the study concept and participated in its design, MRI assessment, statistical analysis, manuscript drafting and editing, and approval for important intellectual concepts. All authors read and approved the final manuscript.

Footnotes

Acknowledgments

We would like to thank Lei Deng, M.D., for helping us prepare our figures. We would like to thank LetPub () for providing linguistic assistance during the preparation of this manuscript.

References

American College of Radiology (ACR), Breast imaging reporting and data system (BI-RADS). 4th edition. American College of Radiology, Reston (VA); 2003.

Agrawal

, Su

M.Y.

, Nalcioglu

, Feig

S.A.

and Chen

J.H.

, Significance of breast lesion descriptors in the ACR BI-RADS MRI lexicon, Cancer 115 (2009), 1363–1380.

Sohns

, Scherrer

, Staab

and Obenauer

, Value of the BI-RADS classification in MR-mammography for diagnosis of benign and malignant breast tumors, Eur Radiol 21 (2011), 2475–2483.

Rao

A.A.

, Feneis

, Lalonde

and Ojeda-Fournier

, A pictorial review of changes in the BI-RADS fifth edition, Radiographics 36 (2016), 623–639.

Schnall

M.D.

, Blume

, Bluemke

D.A.

, DeAngelis

G.A.

, DeBruhl

, Harms

, et al., Diagnostic architectural and dynamic features at breast MRI imaging: Multicenter study, Radiology 238 (2006), 42–53.

Bartella

, Liberman

, Morris

E.A.

and Dershaw

D.D.

, Nonpalpable mammographically occult invasive breast cancers detected by MRI, AJR Am J Roentgenol 186 (2006), 865–870.

Gutierrez

R.L.

, DeMartini

W.B.

, Eby

P.R.

, Kurland

B.F.

, Peacock

and Lehman

C.D.

, BI-RADS lesion characteristics predict likelihood of malignancy in breast MRI for masses but not for nonmasslike enhancement, AJR Am J Roentgenol 193 (2009), 994–1000.

Newell

, Nie

, Chen

J.H.

, Hsu

C.C.

, Yu

H.J.

, Nalcioglu

and Su

M.Y.

, Selection of diagnostic features on breast MRI to differentiate between malignant and benign lesions using computer-aided diagnosis: Differences in lesions presenting as mass and non-mass-like enhancement, Eur Radiol 20 (2010), 771–781.

Wilhelm

, McDonough

M.D.

and DePeri

E.R.

, Malignancy rates of non-mass like enhancement on breast magnetic resonance imaging using American College of Radiology Breast Imaging Reporting and Data System descriptors, Breast J 18 (2012), 523–526.

10.

Yabuuchi

, Matsuo

, Kamitani

, Setoquchi

, Okafuji

, Soeda

, et al., Non-mass-like enhancement on contrast-enhanced breast MR imaging: Lesion characterization using combination of dynamic contrast-enhanced and diffusion-weighted MR images, Eur J Radiol 75 (2010), e126–e132.

11.

Uematsu

and Kasami

, High-spatial-resolution 3-T breast MRI of nonmasslike enhancement lesions: An analysis of their features as significant predictors of malignancy, AJR Am J Roentgenol 198 (2012), 1223–1230.

12.

Kuhl

C.K.

, Breast MR imaging at 3T, Magn. Reson. Imaging Clin N Am 15 (2007), 315–320.

13.

Chatterji

, Mercado

C.L.

and Moy

, Optimizing 1.5-Tesla and 3-Tesla dynamic contrast-enhanced magnetic resonance imaging of the breasts, Magn Reson Imaging Clin N Am 18 (2010), 207–224.

14.

American College of Radiology (ACR), Breast imaging reporting and data system. 5th edition, American College of Radiology. Reston (VA); 2013.

15.

Shin

, Phalak

, Hamame

and Whitman

G.J.

, Interpretation of breast MRI utilizing the BI-RADS fifth edition lexicon: How are we doing and where are we headed? Curr Probl Diagn Radiol 46 (2017), 26–34.

16.

Tozaki

and Fukuda

, High-spatial-resolution MRI of non-mass like breast lesions: Interpretation model based on BI-RADS MRI descriptors, AJR Am J Roentgenol 187 (2006), 330–337.

17.

Kuhl

C.K.

, Mielcarek

, Klaschik

, Leutner

, Wardelmann

, Gieseke

, et al., Are signal time course data useful for differential diagnosis of enhancing lesion in dynamic breast MR imaging?, Radiology 211 (1999), 101–110.

18.

Giess

C.S.

, Yeh

E.D.

, Raza

and Birdwell

R.L.

, Background parenchymal enhancement at breast MR Imaging: Normal patterns, diagnostic challenges, and potential for false-positive and false-negative interpretation, RadioGraphics 34 (2014), 234–247.

19.

Kuhl

C.K.

, The current status of breast MR imaging. Part I. Choice of technique, image interpretation, diagnostic accuracy, and transfer to clinical practice, Radiology 244 (2007), 356–378.

20.

Thomassin-Naggara

, Trop

, Chopier

, David

, Lalonde

, Darai

, et al., Non-masslike enhancement at breast MR imaging: The added value of mammography and US for lesion categorization, Radiology 261 (2011), 69–79.

21.

Chen

J.H.

, Liu

, Baek

H.M.

, Nalcioglu

and Su

M.Y.

, Magnetic resonance imaging features of fibrocystic change of the breast, Magn Reson Imaging 26 (2008), 1207–1214.

22.

Yuen

, Uematsu

, Masako

, Uchida

and Nishimura

, Segmental enhancement on breast MR images: Differential diagnosis and diagnostic strategy, Eur Radiol 18 (2008), 2067–2075.

23.

Morakkabati-Spitz

, Leutner

, Schild

, Traeber

and Kuhl

C.K.

, Diagnostic usefulness of segmental and linear enhancement in dynamic breast MRI, Eur Radiol 15 (2005), 2010–2017.

24.

Liberman

, Morris

E.A.

, Kim

C.M.

, Kaplan

J.B.

, Abramson

A.F.

, Menell

J.H.

, et al., MR imaging findings in the contralateral breast of women with recently diagnosed breast cancer, AJR Am J Roentgenol 180 (2003), 333–341.

25.

Cho

Y.H.

, Cho

K.R.

, Park

E.K.

, Seo

B.K.

, Woo

O.H.

, Sung

B.C.

, et al., Significance of additional non-mass enhancement in patients with breast cancer on preoperative 3T dynamic contrast enhanced MRI of the breast, Iran J Radiol 13 (2016), e30909.

26.

Bluemke

D.A.

, Gatsonis

C.A.

, Chen

M.H.

, Chen

M.H.

, DeAngelis

G.A.

, DeBruhl

, et al., Magnetic resonance imaging of the breast prior to biopsy, JAMA 292 (2004), 2735–2742.

27.

Szabo

B.K.

, Aspelin

, Wiberg

M.K.

and Bone

, Dynamic MR imaging of the breast. Analysis of kinetic and morphologic diagnostic criteria, Acta Radiol 44 (2003), 379–386.

Significant MRI indicators of malignancy for breast non-mass enhancement

Abstract

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1 Introduction

2 Materials and methods

2.1 Study population

2.2 MRI protocol

2.3 MRI interpretation

2.4 Statistical analysis

3 Results

3.1 Lesion histology and type

Table 1 Histological composition of malignant and benign non-mass enhancement lesions Malignant (n = 52) Benign (n = 32) IDC 24 Fibrocystic hyperplasia 16 DCIS 20 Intraductal papilloma 12 ILC 5 Chronic purulent inflammation 4 Inflammatory breast cancer 2 Papillary carcinoma 1

5 Conclusions

6 Declarations

6.1 Abbreviations

6.2 Ethics approval and consent to participate

6.3 Consent for publication

6.4 Availability of data and materials

6.5 Competing interests

6.6 Funding

6.7 Authors’ contributions

Footnotes

Acknowledgments

References

Table 1
Histological composition of malignant and benign non-mass enhancement lesions

Malignant (n = 52) Benign (n = 32)

IDC 24 Fibrocystic hyperplasia 16

DCIS 20 Intraductal papilloma 12

ILC 5 Chronic purulent inflammation 4

Inflammatory breast cancer 2

Papillary carcinoma 1