Prediction of nipple-areolar complex involvement by breast cancer: role of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI)

Skin-sparing and nipple-sparing mastectomies were considered as alternative techniques for modified radical mastectomy. In patients who are candidates for nipple-sparing mastectomy, preoperative assessment of the nipple-areolar complex (NAC) is essential for adequate surgical planning. Breast MRI is highly sensitive for cancer detection and has an important role in disease staging. The aim of this study was to estimate the role of DCE-MRI in predicting malignant NAC invasion by underlying breast cancer and assess the best predictors on MRI that can suspect malignant NAC invasion. Out of the 125 patients with breast cancer, 33 patients (26.4%) showed malignant NAC invasion. On basis of multivariate analysis, abnormal nipple enhancement, tumor nipple enhancement, tumor nipple distance ≤ 2 cm, and abnormal and asymmetric nipple morphology were all significant predictors of malignant NAC invasion (P < 0.001) with abnormal unilateral nipple enhancement as the most important independent MRI predictor of malignant NAC invasion (odds ratio = 61.07, 95% CI 12.81–291.22, P < 0.001). When combining more than positive suspicious MRI features, DCE-MRI had 66.6% sensitivity, 76% specificity, 50% PPV, 86.4% NPV, and 73.6% accuracy in prediction of malignant NAC invasion. DCE-MRI could predict malignant NAC invasion with abnormal unilateral nipple enhancement as the most important independent MRI predictor.


Background
Breast cancer is considered as the second commonest cause of cancer-related death in women [1]. Great efforts were performed in order to introduce more conservative surgeries for the treatment of breast cancer [2]. Skinsparing mastectomy (SSM) and nipple-sparing mastectomy (NSM) were considered as alternative techniques for modified radical mastectomy (MRM) [3]. They provide better cosmetic outcomes resulting in improved body image and more patient satisfaction with more cosmetic results obtained by NSM due to preservation of the NAC [4]. In SSM, the recurrence rate was equivalent to that of MRM; however, in NSM, a small amount of ductal tissue was left behind the nipple, and this could result in an increased risk of local recurrence [5].
In patients who are candidates for NSM, preoperative assessment of the nipple-areolar complex (NAC) is essential for adequate surgical planning [6]. Several clinical and pathological factors are suspected to be associated with occult invasion of the NAC by the underlying breast cancer such as tumor size, multicentric tumor, central location of the tumor, and proximity of the tumor to the nipple base [7]. Mammography and breast ultrasound can be used to predict the malignant invasion of the NAC by underlying breast cancer; however, they have limited sensitivity and specificity. Breast MRI is highly sensitive for cancer detection and has an important role in disease staging. It is considered the imaging modality of choice for the assessment of malignant invasion of the NAC [8].
Several previous studies investigated the role of breast MRI in the prediction of malignant invasion of the NAC [6][7][8][9][10][11][12][13]. However, there is still debate on what is the most important MRI predictor of malignant NAC invasion by underlying breast cancer.

Aim of the study
The aim of this work was to estimate the role of DCE-MRI in predicting malignant NAC invasion by underlying breast cancer and assess the best predictors on MRI that can suspect malignant NAC invasion.

Patient's demographic data
This retrospective study was approved by our institution's ethics committee. The medical records and the preoperative MRI studies of 125 female patients with primary breast cancer and prepared for surgery (MRM, SSM, or NSM) were reviewed from the period of August 2018 till September 2020. Their age ranged from 28 to 70 years (mean ± SD = 44.8 ± 8.96). Out of the 125 female patients, 77 patients underwent MRM, 30 patients underwent SSM, and 18 patients underwent NSM. After reviewing the pathological reports, we found that 33 cases (26.4%) showed malignant NAC invasion and 92 cases (73.6%) showed free NAC. We had 9 cases with DCIS, 109 cases with IDC, and 7 cases with invasive lobular carcinoma.

Inclusion criteria
Inclusion criteria included female patients with primary breast cancer and prepared for surgery (MRM, SSM, or NSM).

Exclusion criteria
Exclusion criteria included patients who were referred to other hospitals and their pathological reports were not available, patients who received neoadjuvant chemotherapy, and patients with advanced disease stage (stage IV) because they were not amenable for surgery.

MRI technique
MRI breast for 125 female patients was performed using 1.5 T MR imaging unit (Philips Ingenia). All patients were examined in the prone position by using a dedicated breast coil. All patients underwent the following: (A) localizing sagittal protocol (scout view); (B) axial non-fat-suppressed T1W fast spin-echo images, with the following parameters: TR/TE 450/14 ms, slice thickness 3 mm, field of view (FOV) 300-360 mm, and matrix 307 × 512; (C) axial non-fat-suppressed T2W turbo spinecho images with the following parameters: TR/TE 4500/97 ms, slice thickness 3 mm, and matrix 384 × 512; (D) axial STIR images with the following parameters: TR/TE 7000-9000/70 ms, TI 150 ms, slice thickness 3-4 mm, inter-slice gap 1 mm, FOV 300-360 mm, and matrix 307 × 512; and (E) dynamic MR images were obtained in the axial plane with fat suppression. The sequence used was FLASH 3-D GRE-T1WI with the following parameters: TR/ TE 4-8/2 ms, flip angle 20-25, slice thickness 2 mm, no inter-slice gap, FOV 300-360 mm, and matrix 307 × 512. Dynamic MR images were obtained after injecting a bolus of gadopentetate dimeglumine at a dose of 0.2 mmol/kg using an automated injector at a rate of 3-5 ml/s. This was followed by a bolus injection of saline (total of 20 ml at 3-5 ml/s).
Image post-processing includes (A) image subtraction obtained by subtracting each of the pre-contrast images from each post-contrast series image and (B) maximum intensity projection (MIP) images obtained through each orthogonal plane, producing sagittal, coronal, and axial projection.

Image interpretation
Using a secondary workstation (Phillips Advantage windows workstation with functional tool software), MR images were analyzed by two radiologists (ME, DE) with breast imaging experience for 14 and 9 years. They were blinded to the clinical and pathological data of the patients. The two radiologists joined and reached a correspondence for controversial cases.
The assessed signs on dynamic MR images were (A) malignant mass pattern (mass lesion or non-mass enhancement); (B) tumor size (TS) (maximum diameter of the lesion, if multiple lesions present we took the maximum diameter of the largest lesion); (C) nipple morphology (NM) (normal or retracted nipple) was assessed on axial T1WI and axial STIR image; (D) symmetry of nipple morphology was assessed on MIP images; (E) tumor nipple enhancement (TNE) (presence of enhancement between tumor and nipple base assessed on early subtraction MR images); (F) tumor nipple distance (TND) (measured from the tumor margin to the base of the nipple on early subtraction MR images); (G) abnormal asymmetric nipple enhancement assessed on early subtraction MR images; and (H) thickening of the periareolar skin when compared with the contralateral NAC, it was assessed on early subtraction MR images.

Final diagnosis
The pathological reports after surgery were reviewed and the following data were collected: (1) presence of malignant NAC invasion (malignant NAC invasion was defined on pathology as the presence of IDC, DCIS, or invasive lobular carcinoma within the retro-areolar tissue), (2) histologic tumor type, (3) tumor grade, (4) presence of lymphovascular metastasis, (5) presence of lymph node metastasis, and (6) hormone receptor status (ER, PR, Her 2, and Ki67).

Statistical analysis
Data were fed to the computer and analyzed using IBM SPSS Corp., released 2013, IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp. Qualitative data were described using number and percent. Quantitative data were described using median (minimum and maximum), interquartile range for non-parametric data, and mean and standard deviation for parametric data after testing normality using the Shapiro-Wilk test. The significance of the obtained results was judged at the (0.05) level. Student t test was used to compare 2 independent groups. The diagnostic performance (accuracy) of a test to discriminate diseased cases from nondiseased cases is evaluated using receiver operating characteristic (ROC) curve analysis. Sensitivity and specificity were detected from the curve. PPV, NPV, and accuracy were calculated through cross-tabulation. The multivariate logistic regression model was done to determine the best combined parameters for the prediction of malignant NAC invasion by generating AUC with 95% confidence intervals.

Results
This retrospective study included 125 female patients with primary operable breast cancer. After reviewing the pathological reports, we found that 33 cases (26.4%) showed malignant NAC invasion and 92 cases (73.6%) showed free NAC (Fig. 1).
On univariate analysis, tumor size, tumor nipple distance, abnormal nipple enhancement, nipple morphology, symmetry of nipple morphology, tumor nipple enhancement, and periareolar skin thickening were all important predictors of malignant NAC invasion (Fig.  2a-c). The mean TS for cases of malignant NAC invasion was 5.5 ± 2.34 cm versus 3.55 ± 1.67 cm for free NAC (P < 0.001). The mean TND for cases of malignant NAC invasion was 2.28 ± 1.94 cm versus 3.89 ± 2.63 cm for free NAC (P = 0.032). Abnormal nipple enhancement was observed in 19 (57.6%) out of 33 cases with malignant NAC invasion; however, it was observed only in 2 (2.2%) cases out of 92 cases with free NAC (P < 0.001). Abnormal nipple morphology was observed in 21 (63.6%) out of 33 cases with malignant NAC; however, it was observed only in 6 (6.5%) cases out of 92 cases with free NAC (P < 0.001). TNE was observed in 24 (72.7%) out of 33 cases with malignant NAC invasion; however, it was observed in 22 (23.9%) out of 92 cases with free NAC. Regarding pathological features, tumor type, HER 2 status, and PR state were significant predictors for malignant NAC invasion; P values were 0.006, < 0.001, and 0.014, respectively (Table 1).

Discussion
The incidence of malignant NAC invasion by underlying breast cancer varies in different studies from 5.6 to 24.6% [14][15][16]. One study showed that malignant invasion of the NAC by underlying breast  cancer was found in 37 (21.8%) out of 170 mastectomy specimens [16]. Another study showed that malignant NAC invasion was found in 71 (35.5 %) out of 200 patients [3]. In this work, malignant NAC invasion by underlying breast cancer was reported in 26.4% of patients. Several previous studies investigated the role of DCE-MRI in predicting malignant NAC invasion by underlying breast cancer [1,3,[7][8][9][10][11][12][13][14][15][16]. It was stated that both tumor nipple distance and abnormal enhancement extending from the tumor to the nipple base were the most important predictors of malignant NAC invasion on MRI [9,10]. One stated that TND with cutoff value of 10 mm had high sensitivity and specificity in predicting malignant NAC [3]. Another study stated that Continuity to NAC and unilateral NAC enhancement were all important MRI predictors of malignant NAC invasion [12]. This is in agreement with our study where we found that MRI predictors such as abnormal nipple morphology, asymmetry of nipple morphology, abnormal unilateral nipple enhancement, TNE, and TND ≤ 2cm were all significant predictors of malignant NAC invasion.
In the present study, abnormal unilateral nipple enhancement was the most important independent MRI predictor of malignant NAC invasion (odds ratio = 61.07, 95% CI 12.81-291.22, P < 0.001) (specificity = 97.8% and diagnostic accuracy = 87.2%). This in agreement with Liao et al. [7] who stated that abnormal unilateral nipple enhancement was the most significant independent image predictor of malignant NAC invasion (odds ratio = 4.86, 95% CI 1.76-13.80, P ≤ .01). Also, our results are in agreement with a study performed by Lee et al. [17], and they found that MR images displaying inhomogeneous and diffuse enhancement in areas of thickened skin and the parenchyma of the NAC were indicative of NAC invasion. Another study performed multivariate logistic regression analysis for pathologic diagnosis of NAC involvement and concluded that NAC enhancement and NAC enhancement thickness were the two most important factors related to NAC invasion (P < .001) [18], and this is in agreement with our results.
The sensitivity and specificity of DCE-MRI in the prediction of malignant NAC invasion is variable among different studies. One study stated that the combination of more than one suspicious MRI feature had 60.5% sensitivity, 87.5% specificity, and 84.6% accuracy in predicting malignant NAC invasion [12]. Another study performed interobserver agreement about the ability of DCE-MRI to predict malignant NAC invasion; the first observer found that DCE-MRI had 71.4% sensitivity, 81.6% specificity, and 80.8% accuracy; and the second observer found that it had 78.6% sensitivity, 88.1% specificity, and 87.4% accuracy [7]. In this study combination of more than two suspicious MRI features, it could result in 66.6% sensitivity, 76% specificity, and 73.6% accuracy in predicting malignant NAC invasion.
The limitations of this study were that its retrospective nature, no interobserver agreement was performed, and also the MRI examinations were performed on 1.5 Tesla MRI.

Conclusion
DCE-MRI could predict malignant NAC invasion with abnormal unilateral nipple enhancement as the most important independent MRI predictor.