Radiological imaging characteristics of intramammary hematological malignancies: results from a german multicenter study

To assess radiological procedures and imaging characteristics in patients with intramammary hematological malignancies (IHM). Radiological imaging studies of histopathological proven IHM cases from ten German University affiliated breast imaging centers from 1997–2012 were retrospectively evaluated. Imaging modalities included ultrasound (US), mammography and magnetic resonance imaging (MRI). Two radiologists blinded to the histopathological diagnoses independently assessed all imaging studies. Imaging studies of 101 patients with 204 intramammary lesions were included. Most patients were women (95%) with a median age of 64 years. IHM were classified as Non Hodgkin lymphoma (77.2%), plasmacytoma (11.9%), leukemia (9.9%), and Hodgkin lymphoma (1%). The mean lesion size was 15.8 ± 10.1 mm. Most IHM presented in mammography as lesions with comparable density to the surrounding tissue, and a round or irregular shape with indistinct margins. On US, most lesions were of irregular shape with complex echo pattern and indistinct margins. MRI shows lesions with irregular or spiculated margins and miscellaneous enhancement patterns. Using US or MRI, IHM were more frequently classified as BI-RADS 4 or 5 than using mammography (96.2% and 89.3% versus 75.3%). IHM can present with miscellaneous radiological patterns. Sensitivity for detection of IHM lesions was higher in US and MRI than in mammography.

Diffuse large B-cell lymphoma is the most common histologic IHM subtype, followed by extranodal marginal zone lymphoma and follicular lymphoma 3, 5-7, 12, 13 . Other types of breast lymphoma include mucosa associated lymphoma 3, 6, 12 , Burkitt lymphoma 3, 6 , plasmacytoma 1,6 and T-cell lymphoma 12,14 . In addition, there were reports of Hodgkin's disease manifestation of the breast 15,16 . Breast plasmacytoma has been reported to occur in 1.5% of all incident plasmocytoma cases, whereas it accounts for only 0.2% of all malignant breast cancers 1 . Intramammary relapse in leukemia varied from 1.1-6.5% in myeloid leukemia 17,18 . The exact prevalence of breast leukemia is difficult to determine 14 . As reported previously, at the moment, there is a paucity of studies focusing on radiological characteristics of different IHM. Various IHM imaging characteristics have been described in the literature, complicating radiological assessment of a crucial breast disorder 5,9,19,20 . Therefore, the purpose of our study was to assess radiological procedures and imaging characteristics of IHM in a large multicenter sample.

Methods
Data acquisition and patients. This retrospective study was initiated by the department of radiology of the Martin-Luther-university Halle-Wittenberg and has been approved by the institutional ethics committee. All methods were performed in accordance with the relevant guidelines and regulations. Written informed consent from the study patients was not required.
Intramammary hematological malignancies diagnosed at 10 University affiliated breast centers from 1997 to 2012 were retrospectively assessed. Only histologically proven cases of IHM diagnosed by breast biopsy were included. Breast lesions suspicious for IHM without histopathological examination were excluded. Hematological malignancies with pectoral or axillary lymphadenopathy without breast affection, osseous or soft tissue malignant hematological tumors with invasion into the breast were also excluded from the study.
Imaging. All provided breast imaging studies were accumulated in digital format at the Department of Radiology of the University Hospital Halle and were analyzed by two radiologists (S.W. and A.H.) with 7 and 4 years' experience in breast imaging in consensus reading. They were blinded to the patients' information and unaware of histopathological diagnosis. All lesions were classified according to the Breast Imaging Reporting and Data Systems (BI-RADS) 5 th edition categories 21, 22 . Mammography. Mammography was available in 73 patients. It was obtained on different dedicated digital or analog mammographic equipment's. In every case, a standard mediolateral oblique and craniocaudal view was made. The ACR breast density types of mammographic images was classified according to the BI-RADS ® lexicon 21, 22 . Ultrasound. Sonographic images in axial and transverse planes were available for 53 patients. Different linear-array transducers at a center frequency of 8-13 Megahertz (MHz) were used.
Magnetic resonance imaging. Magnetic resonance imaging (MRI) was performed in 28 patients utilizing different 1.5 Tesla scanners. Data of kinetic analysis of contrast enhancement, without diffusion-weighted imaging was available for 24 patients. Time-signal intensity curves were drawn using operator defined region of interest (ROI). The ROI was smaller than the lesion size. The initial signal increase (Initial SI) from the pre-contrast value (SI p ) to the maximum peak within the first 3 min after the administration of contrast medium (SI 1-3 min ) was calculated as reported previously 23 : The post-initial behavior of the signal curve (Post-initial SI) from the maximum peak (SI peak ) to the end of the examination (SI end ) was also analyzed:  Table 2).
The right breast was involved in 46 patients (45%), the left breast in 37 (37%), and bilateral involvement of the breast was evident in 18 patients (18%). A total of 204 lesions were identified in the 101 patients included in this study (average of 2.02 lesions per patient). In 54 patients (53%), the lesions were solitary, 47 patients (47%) presented with multiple IHM lesions. Although no statistically significant (p > 0.05), the number of lesions per patient varied across different IHM: 1.9 in NHL, 1.5 in leukemia, and 2.9 in plasmacytoma.

Radiological Features
Mammographic findings. Mammography was obtained in 73 patients, comprising 131 lesions. Most of them show a breast density type b (52.1%) or breast density type c (37%), as shown in Table 3.
Although not significant (p > 0.05), the frequency of the mammographic patterns was different in several IHMs, which are summarized in Table 4.
Solitary or multiple intramammary masses were the most frequent mammographic finding in the patients (Fig. 1). Overall, 131 masses were identified. They were most commonly round or irregular in shape (101/77.1%) with indistinct margins, as demonstrated in Table 5. Furthermore, most lesions had an equal density (90/68.7%) to the surrounding tissue (Fig. 2). The mean size of the identified lesions was 15.8 ± 10.1 mm (median size 15 mm, range 5-60 mm). There were no significant differences in size between several IHMs (p > 0.05).    Table 4. Mammographic patterns in different intramammary hematological malignancies (IHM). Notes: *There were no significant differences in patterns between the IHM groups; P > 0.05. ** n = number of patients.      Table 6. The majority of lesions were irregular in shape (55 lesions, 57.9%) with complex echo pattern (56 lesions, 58.9%) and indistinct margins (68 lesions, 71.6%), (Figs 3 and 4). Furthermore, most lesions had a non-parallel orientation (55 lesions, 57.9%) and different posterior acoustic features. Most US findings were classified as BI-RADS 5 (38 lesions, 71.7%). Table 7. Overall, 28 patients with 36 intramammary findings were investigated by MRI. Lesions presented either as masses (86.1%) or as non-mass enhancing lesions (13.9%). Non-mass enhancement manifested as diffuse (n = 2), segmental (n = 1) or localized in multiple regions (n = 2). Most IHM masses presented with irregular shape, irregular or spiculated margins, and various enhancement patterns (Fig. 4). All lesions showed an intermediate signal on T2 weighted images compared to breast parenchyma. Table 4 presents IHM MRI findings.

MRI findings. MRI features of IHM are summarized in
Kinetic analysis of contrast enhancement was performed for 24 masses. It showed a rapid initial signal increase (Initial SI) over 100% in comparison with pre-contrast signal intensity in all lesions. In the delayed phase, a plateau was seen in 20 cases (83.3%). In 4 cases (16.7%), a washout phenomenon was noted.

Discussion
Primary diagnosis and patients. The present study evaluates radiological procedures and imaging characteristics of IHM based on a large sample of 101 patients with 204 lesions acquired in 10 University affiliated breast imaging centers. Previous reports regarding IHM reported on 11 to 36 patients 5,14,16,20 . Furthermore, the studies primarily analyzed primary or secondary lymphomas of the breast, whereas other IHM subtypes, such as breast plasmacytoma or leukemia were not reported.
In our study, IHM predominantly manifests women, which is supported by the literature 24,25 . These results were concordant with our data.
Patients median age at IHM diagnosis was 64, ranging from 22-84 years, consistent with earlier reported median age and age ranges 4,8,11,[26][27][28] .  Most case series on smaller patient cohorts documented a propensity for breast lymphoma manifestation in the right breast 4,5,26,29,30 . However, in the present large scale study, no side predominance of breast lymphoma localization was seen. Bilateral breast involvement was evident in 18% of cases.

Mammographic findings.
In the present study, intramammary masses were the most common mammography pattern of IHM. Most of the masses were round or irregular in shape with indistinct margins. There was no significant difference in shape and margin characteristics of lesions between the different types of the IHM. Furthermore, other patterns of IHM were identified: In 21% of cases, breast lymphoma presented on mammography as architectural distortion (focal or global asymmetry), especially in cases with NHL and leukemia. This pattern was not detected in intramammary plasmacytomas. In addition, a combined pattern of architectural distortion and masses was found in 15% of cases. This manifestation has not been reported previously. The frequency of mammographic patterns varied across several IHMs, although not statistically significant.
In concordance to our findings, mammographic findings of breast lymphomas are heterogeneous and lack a pathognomonic presentation, as described by a smaller case series 5 . In general, breast masses were the most common radiologic features observed in IHM 5,20,31 .
In our study collective the mammograms of 14 patients (19.2%) were categorized as BI-RADS category 2 or 3. These results are in concordance with previous reports 14 . This may be related to the fact that breast lesions in several malignant hematological diseases can mimic benign disorders on mammography. Overall, published case Scientific REPORTS | 7: 7435 | DOI:10.1038/s41598-017-07409-z series described comparably small cohorts of 10 to 96 lesions 5,9,14,16,19,20,23,26,29 . Therefore, our study with 131 mammographic findings in 73 patients with IHM is the largest to date.
Ultrasound findings. In our series based on 95 lesions, 55 (57.9%) were irregular in shape with indistinct margins (71.6%). Most lesions had complex echo patterns with a non-parallel orientation and a broad spectrum of posterior acoustic phenomenon.
Contrasting our findings, the literature describes uniform sonographic appearances with hypoechoic round or oval masses, increasing the risk for false negative diagnoses 14,16,20 . However, these studies were based on smaller cases series of 5 to 24 patients.
In agreement with previous reports 14 , ultrasound had a higher proportion of BI-RADS category 5 lesions (71.7%) than mammography (23.3%). This finding suggests that patients with hematological disorders and intramammary lesions should be clarified by both modalities: mammography and ultrasound.
In our study, overall 36 intramammary findings were analyzed by MRI. IHM mainly presented as an oval breast masses (31/36, 86.1%), which was consistent with the results by Rizzo et al. 37 and Liu et al. 38 . On T2 weighted images, all investigated lesions appeared intermediate compared to breast parenchyma, in line with the literature 14,37 .
Furthermore, most of the lesions showed malignant features, such as an irregular shape with irregular margins, complicating the discrimination from other malignant breast tumors.
After intravenous administration of contrast medium, IHM internal enhancement was mostly homogeneous or slightly heterogeneous. As reported previously, rim enhancement and dark internal septation were rarely seen 14,37,38,41 , which is similar to the enhancement pattern of lymphomas in other sites. Overall, MRI findings of IHM differ from those in frequent breast carcinomas, which usually show marked heterogeneous enhancement 42,43 .
In most reports, only qualitative analysis of the enhancement kinetic was performed 14,37 . Studies reported comparable kinetic curve patterns with a rapid enhancement and a plateau in the delayed phase 14,37,38 . However, these findings were obtained by 1 Tesla MRI scanners 14,37 .  Table 7. MRI findings in IHM.
In our study, kinetic analysis of contrast enhancement was performed in 24 lesions. A rapid initial signal increase over 100% in comparison with pre-contrast signal intensity was seen in all cases. The delayed phase showed a plateau phenomenon in most cases (83.3%). These results were in agreement with earlier studies 37,38 . However, this finding is generically expected for all malignant lesions, regardless of histological type, and is therefore not pathognomonic for breast lymphoma 44,45 .
In our study, MRI yielded a higher proportion of BI-RADS category 5 lesions (64.3%) than mammography (23.3%). Presumably, MRI sensitivity for the detection of multifocal or multicentric tumors might be superior to other modalities. Thus, in the case of suspected or confirmed breast lymphoma, MRI could be recommended for a more accurate and comprehensive assessment.

Strengths.
Our study had several strengths. We present the largest report on IHM with 101 patients and a total of 204 lesions. Moreover, blinded two-reader assessment of various modalities ensured a comprehensive characterization of radiological IHM features. Finally, the multi-center setting of our study yields information on IHM imaging findings generalizable to the population, minimizing local influences.
Limitations. There are several limitations to our study. First, the retrospective setting limits analyses. Second, not all imaging modalities of ultrasound, mammography, and MRI were available for every patient. However, this algorithm reflects the typical workflow in clinical practice, where breast lesions are subjected to percutaneous biopsy without complete imaging work-up on all imaging modalities.

Conclusion
In conclusion, our study demonstrated imaging findings of IHM based on the largest sample to date. IHM manifested most frequently as a lobular or irregular mass with indistinct margins at mammography and as hypoechoic, irregular mass with indistinct margins and different posterior acoustic phenomena at ultrasound. On MRI, the identified lesions are variable in shape and show rapid contrast enhancement with post-initial plateau. Furthermore, the imaging features of breast lesions in IHM are more typically malignant in ultrasound and MRI than in mammography.