Atypical lobular hyperplasia and lobular carcinoma in situ at core needle biopsy of the breast: An incidental finding or are there characteristic imaging findings?

Abstract

BACKGROUND:

Atypical lobular hyperplasia and classic-type lobular carcinoma in situ, collectively known as lobular neoplasia, are classically described as incidental findings found on breast core-needle biopsy without distinguishing imaging characteristics.

OBJECTIVE:

The purpose of this study was to investigate concordant imaging findings of lobular neoplasia identified at core-needle biopsy after careful radiologic–pathologic correlation.

METHODS:

The pathology database was searched from October 1, 2006 to October 1, 2013 for breast biopsies yielding lobular neoplasia not associated with a coexistent malignancy or other high risk lesion in the biopsy specimen.

RESULTS:

Of the 482 biopsies performed containing lobular neoplasia, 65 cases had lobular neoplasia as the highest risk lesion at core-needle biopsy. Of the 65 total cases in which lobular neoplasia was the highest risk lesion, 18 (28%) cases had concordant imaging correlates. 13 of 18 (72%) cases presented as calcifications on mammography and 5 of 18 (28%) presented on magnetic resonance imaging as a focus (n = 2) or non-mass enhancement (n = 3).

CONCLUSION:

With careful radiologic–pathologic correlation, mammographically detected calcifications and foci or non-mass enhancement on magnetic resonance imaging can be considered concordant imaging findings of lobular neoplasia after breast core-needle biopsy.

Introduction

Atypical lobular hyperplasia (ALH) and classic-type lobular carcinoma in situ (LCIS), collectively known as lobular neoplasia (LN), are proliferative lesions of the breast associated with an increased risk of ipsi- and contralateral breast cancer.

The basic functional unit of the breast, the terminal duct-lobular unit is the anatomical site of origin for lobular neoplasia. The normal lobule consists of multiple glandular acini that are connected to a single terminal duct (Fig. 1).

The term lobular carcinoma in situ was first described in 1941 by Foote and Stewart. They described a pattern of neoplasia that started in the lobule or terminal lobular duct, was multicentric, and could not be identified grossly or clinically [1]. Classic type lobular carcinoma in situ have nucleoli that are rather uniform and lack pleomorphism and/or necrosis which distinguish them from pleomorphic lobular carcinoma in situ (PLCIS) (Figs 2 and 3) [2–4].

ALH was defined in 1978 by Page et al. as atypical epithelium involving lobular units with similar cytology as LCIS; however only involving fifty percent of the lobule (Fig. 4) [2]. The term “lobular neoplasia” has been coined to encompass the full spectrum of lobular lesions [2–4].

LN is classically described as an incidental finding identified on breast core-needle biopsy (CNB) without any distinguishing physical or imaging findings [5]. Literature describes the most common imaging finding of lobular neoplasia as mammographically detected calcifications, occurring in 60–90% of LN lesions found on breast CNB [6–16]. LN found on CNB has also been associated with an architectural distortion, or non-mass enhancement (NME) and foci on magnetic resonance imaging (MRI) albeit at a lesser frequency [6–15]. Literature shows that NME and foci on MRI occur in 5–25% of LN lesions found on CNB (Table 1). However, the frequency is highly dependent on the population as high-risk populations will typically have more MRI studies performed with subsequently more LN lesions seen on MRI.

Calcifications and NME on mammography and MRI, respectively, are increasingly being accepted as concordant imaging findings of LN found at CNB [6–16]. On the other hand, mass lesions that return with LN at CNB are thought to be due to another pathologic process [6–8]. However, little is written in literature about the possibility of LN demonstrating characteristic imaging findings.

The purpose of this study is to investigate the imaging findings of LN found at CNB across imaging modalities and across the pathologic spectrum of LN to identify if there are any distinguishing imaging findings associated with LN. In addition, we describe and give examples of concordant and discordant imaging findings to assist the reader when evaluating radiologic–pathologic concordance. We evaluated 65 cases at our own institution over the last six years and compared these cases with data from current literature to provide a better understanding of the characteristic imaging findings of lobular neoplasia.

Methods

This retrospective study was approved by the Institutional Review Board of Penn State Milton S Hershey Medical Center and was compliant with the Health Insurance Portability and Accountability Act.

The pathology database was searched from October 1, 2006 to October 1, 2013 for ultrasound guided, stereotactic guided, or magnetic resonance imaging guided breast biopsies yielding ALH or LCIS not associated with a coexistent malignancy or other high risk lesion in the biopsy specimen.

All CNBs with concomitant high risk lesions (papilloma, radial scar, atypical ductal hyperplasia, flat epithelial atypia, ductal carcinoma in situ or invasive carcinoma) in the same biopsy were excluded from this study. All cases of PLCIS were excluded. Of the 482 biopsies performed containing LN, only 65 cases had ALH or LCIS as the highest risk lesion at CNB.

All imaging associated with the lesion was carefully reviewed by dedicated breast imagers, (J.M.) with 17 years of experience, (M.W.) with 18 years of experience, (S.S.) with 24 years of experience, or (A.C.) with 5 years of experience. Lesions were correlated across multiple imaging modalities when possible.

Lesions were classified using American College of Radiology BI-RADS lexicon [17].

Histology correlation was reviewed by a single board certified pathologist, (N.W.) with 5 years of breast subspecialty experience. Pathology was classified using the criteria as described in the introduction for ALH and LCIS.

After independent review by the radiologists and by the breast pathologist, radiologic–pathologic correlation was conducted among the radiologists and pathologist to determine if the imaging findings were concordant with the pathologic findings.

Concordant lesions were further divided into incidental or non-incidental findings. Lobular neoplasia was considered incidental if the imaging target was not associated with the LN histologically (i.e. targeted suspicious calcifications were seen associated with benign ducts histologically and lobular neoplasia was not associated with calcifications). Lobular neoplasia was considered non-incidental (i.e. calcifications seen associated intimately with lobular neoplasia) if the target of the biopsy matched histologically with the imaging findings.

Our data was then compared with data from current literature to give a better understanding of the imaging findings of lobular neoplasia that can be expected by the reader (Table 1).

The 95% confidence intervals (CI) were calculated using a modified Wald method.

Results

There were 65 total cases of lobular neoplasia identified after breast core needle biopsy in which lobular neoplasia was the highest risk lesion. After careful radiologic–pathologic correlation, 18 of 65 total cases (28%) were non-incidental concordant findings.

There were two discordant cases of the total 65. One case was an 8 mm mass seen only on MRI. It had medium initial with persistent delayed enhancement. Final pathology demonstrated ALH and extensive columnar cell change. The imaging was thought to be discordant with pathology. Excision yielded ALH on background of fibrocystic change without any explanation for the enhancing mass lesion. The other discordant case was a mass lesion seen on screening mammography and ultrasound in which CNB showed LCIS only. The imaging was thought to be discordant with pathology. Excision revealed DCIS.

Of the 65 total cases, 45 cases (71%) of lobular neoplasia underwent biopsy due to mammographically detected suspicious finding. Eighteen of 65 cases were initially detected on MRI. One case of 65 was initially detected and only visualized on ultrasound.

The age distribution of the 65 cases included three cases 39 years old or younger, 20 cases between 40 and 49 years old, 39 cases between 50 and 75 years old, and 3 cases 76 years old and older.

The indications for the 65 total cases included 40 screening cases in the general population, eight screening cases in a high risk population (National Cancer Institute lifetime risk greater than 20%), two diagnostic cases for clinical abnormality, eight cases for preoperative evaluation with new diagnosis of contralateral carcinoma, and seven cases for preoperative evaluation with new diagnosis of separate distinct ipsilateral carcinoma.

Of the 65 total cases, 43 went on to surgical excision. Thirty five of 43 cases showed consistent pathology at excision with the original CNB. Eight of 43 cases had additional lesions at excision. One of eight cases was a known invasive lobular carcinoma which had a single excision for both carcinoma and LCIS with two separate biopsy sites and separate lesions noted on excisional pathology. The invasive lobular carcinoma was initially detected on mammography as a spiculated mass at the right 6:00 position with CNB showing invasive lobular carcinoma. Preoperative MRI showed NME at the 9:00 position, 1 cm above the initial carcinoma, with CNB demonstrating LCIS. Excision showed two separate lesions corresponding to the invasive lobular carcinoma and LCIS with two respective biopsy sites.

Seven of eight cases were upgraded to low-grade DCIS. Five of seven cases were mammographic calcifications that at excision were associated with the LN and not the DCIS. Therefore, DCIS was thought not to be associated with the initial imaging finding and considered an incidental finding at excision. Another case of seven was a concordant incidental NME. CNB showed stromal fibrosis which could have explained the NME, and therefore LCIS was thought to be an incidental finding. Excision of this lesion showed DCIS and stromal fibrosis. The last case of seven was a discordant mass lesion seen on mammography and ultrasound in which CNB showed LCIS only. The imaging was thought to be discordant with pathology. Excision of the lesion revealed DCIS.

Remaining 22 of 65 cases either had complete mastectomy (n = 11), were followed with imaging for greater than two years with stability (n = 5), followed with imaging for one year with stability but lost to follow-up (n = 5), or lost to follow-up initially (n = 1).

Mastectomies were done in two high-risk patients (National Cancer Institute lifetime risk of 20% or higher), four with new diagnosis of ipsilateral breast cancer, and five with new diagnosis of contralateral breast cancer. Mastectomies confirmed the original CNB findings in all cases with sectioning of the mastectomy, although possibly not as accurate as a directed lesion excision.

Mammographic calcifications

Mammographically detected calcifications were the most frequent radiographic finding associated with LN found on CNB. Of the 46 cases of lobular neoplasia initially detected by mammography, 37 cases were identified after biopsy of suspicious mammographically detected calcifications (Table 1).

Of the cases of LN with non-incidental concordant imaging findings, 13 of 18 (72%), presented as calcifications on mammography. Nine of 13 (69%) LN cases demonstrated morphology of grouped fine pleomorphic calcifications (Fig. 5), 2 of 13 (15%) with grouped amorphous calcifications, 2 of 13 (15%) with grouped coarse heterogenous calcifications.

Thirty two of these 37 cases were detected on screening mammography. Three cases were detected in high-risk screening patients (1 with prior LCIS, and 2 with strong family history). One case, (LCIS) was detected in a patient with new diagnosis of contralateral DCIS with excision confirming both lesions. One case (ALH) was detected in a patient with new diagnosis of ipsilateral IDC separate from the lesion with excision confirming both lesions.

NME and foci seen on magnetic resonance imaging

Five of 18 (28%) LN cases with non-incidental concordant imaging findings presented as either a focus or NME on MRI. All five cases were LCIS. Of the 5 of 18 cases detected with MRI, 2 of 5 (40%) cases presented as a focus; 3 of 5 (60%) cases showed NME in a linear clumped pattern (Figs 6 and 7) with persistent (2 cases) or plateau (1 case) kinetics (Table 1).

Four of these five MRI detected cases were identified on pre-operative MRI performed for women with newly diagnosed breast cancer and one of the five cases was detected on a supplemental screening MRI for a woman at high risk for breast cancer (Table 2).

Mass lesions

Of the total 65 cases, 15 LN cases were associated with a mass lesion detected on imaging.

Eight of 15 cases were solitary masses detected on mammography with six of these also visualized with ultrasound. In all cases LN was an incidental finding and the mass lesions were explained by benign findings as follows: complex fibroadenoma (n = 1), fibroadenoma (n = 3), complex cyst (n = 1), extensive fibrocystic change (n = 3). Seven of these masses were initially detected following screening mammography. One lesion was identified during diagnostic evaluation of a palpable lump which appeared as a mass both on mammography and ultrasound and was found to be a fibroadenoma (with coexistent lobular neoplasia). One of the 15 mass lesions was associated with grouped pleomorphic calcifications. The mass and grouped pleomorphic calcifications were found to be due to fibroadenomatous change and LCIS was an incidental finding. This was initially detected on screening mammography with a mass correlate lesion on ultrasound.

Five of 15 mass lesions were detected on MRI with all cases of LN being an incidental finding and the mass lesions explained by the following pathologic diagnoses: fibroadenoma (n = 1), adenosis (n = 1), extensive fibrocystic change (n = 2), sclerosing adenosis with fibroadenomatous change (n = 1). The one fibroadenoma case had imaging correlates with ultrasound and mammography.

Two of these five MRI cases were performed on screening in high risk patients and the other three were performed to evaluate extent of disease in patients with newly diagnosed breast carcinoma.

Finally, one lesion was only visualized on ultrasound performed for diagnostic evaluation of a palpable lump which appeared as a complicated cyst on ultrasound. ALH was incidentally identified after core needle biopsy was performed of the complicated cyst.

Discussion

Mammographic calcifications

Mammographically detected calcifications represented the most frequent radiographic finding associated with lobular neoplasia identified on CNB. This is in agreement with the current literature which ranges from 60–90% (Table 1) [6–16]. The wide range is due to some studies including only mammographic data while others (such as our study) include mammography, ultrasound and MRI.

Previously it was believed that LN was an incidental finding and that there was no imaging correlate. Lieberman et al. described the phenomenon that with larger bore CNB’s, this incidental finding of LN would increase as larger tissue samples were taken of calcifications associated with benign tissue [5]. However, current literature has shown a significant amount of calcifications to be associated with LN ranging from 8–41% (Table 1) [6–16]. Nagi et al. described that most calcifications are found within benign tissue, and in the study only 8% were associated with LN, and none with LN alone [10]. The majority of current literature which utilizes thorough radiologic–pathologic correlation demonstrates that approximately 20–40% of lobular neoplasia cases are non-incidental being associated histologically with calcifications, which mirrors the findings in our study (Table 1) [6,8,14–16].

In our study, the most common mammographic calcification morphology of all concordant LN (both incidental and non-incidental) was grouped fine pleomorphic calcifications (57%), followed by grouped amorphous calcifications (24%) and grouped coarse heterogenous calcifications (19%).

This is in agreement with current studies including Atkins et al. whom performed a thorough evaluation of the imaging findings in LN. In their study, grouped fine pleomorphic calcifications was the most common concordant morphology (21 of 32, 65%), followed by grouped amorphous calcifications (8 of 32, 25%), fine linear calcifications (2 of 32, 6%), and punctuate calcifications (1 of 32, 3%) [7].

Accordingly there was a similar distribution in concordant non-incidental LN, with 69% showing a morphology of grouped fine pleomorphic calcifications, 15% with grouped amorphous calcifications, and 15% with grouped coarse heterogenous calcifications To our knowledge, there is no literature addressing the morphology of calcifications on imaging which are non-incidental and associated histologically with calcifications.

Histologically, ALH, LCIS, and PLCIS represent distinct entities. However, they can occur simultaneously and may be difficult to differentiate histologically. Radiographically, there have been no distinguishing features along the spectrum of LN described to date. It has been hypothesized that similar to the pathologic spectrum of ALH, LCIS, and PLCIS, there may be imaging correlates across the spectrum of LN. Georgian et al. described LCIS and PLCIS as a pathologic spectrum, with PLCIS having comedocarcinoma calcifications that are similar to DCIS which may give it a radiographic appearance similar to DCIS. They described classic LCIS commonly having calcifications similar to surrounding benign tissue which may lend to a more benign appearance radiographically [18].

However, in our study we found no difference across the spectrum of LN radiographically. Four of 13 (31%, CI: 12-58) of the mammographically detected calcifications represented ALH and 9 of 13 (69%, CI: 42-88) represented LCIS. The difference between ALH and LCIS was not found to be different nor was there a difference seen amongst similar studies in our literature search (Table 3). However the small sample size in our study and other studies limits statistical signficance, which may be considerable in a larger sample size. In addition as stated above, PLCIS was not included in our study, which may have a notable radiographic appearance when compared to ALH and/or LCIS.

NME and foci seen on magnetic resonance imaging

Foci and NME have started to gain acceptance as possible imaging correlates to LN. Murray et al. described how MRI foci and NME may be considered concordant, but that the issue of radiologic–pathologic correlation need to be even further scrutinized due to the lack of clear concordant–discordant criteria [6]. The biggest hurdle is due to the inability to clearly document tissue sample adequacy as seen in calcifications. With calcifications the ability to document the target calcifications in histologic samples makes concordance much simpler. In MRI, pre and post biopsy imaging must be thoroughly scrutinized to evaluate for adequate sampling. In addition, MRI patterns need to be thoroughly evaluated, and if the pattern is not consistent with the pathologic process, then discordance must be entertained. Any mass lesion, or lesion with malignant features should be considered discordant if pathology results return lobular neoplasia [6,19–21]. Common benign concurrent incidental lesions of a focus or NME could include sclerosing adenosis, fibrocystic changes, pseudoangiomatous stromal hyperplasia, and stromal fibrosis [19–21].

In our study, 18 of 65 cases were initially detected on magnetic resonance imaging. After careful radiology–pathology correlation, 5 of 18 (28%) LN cases were shown to have non-incidental concordant MRI findings as either a focus or NME. Two cases (40%) presented as a focus; three cases (60%) demonstrated NME in a linear clumped pattern with persistent (2 cases) or plateau (1 case) kinetics. All five cases were LCIS. This is an agreement with current studies including Murray et al. whom described 17 of 80 MRI NME lesions which returned lobular neoplasia as the highest risk lesion after CNB. In their study, lobular neoplasia was seen in conjunction with other benign pathology including stromal fibrosis, sclerosing adenosis, fibrocystic changes, and pseudoangiomatous stromal hyperplasia. Of note only LCIS cases (without any ALH cases) were seen as non-incidental concordant findings of foci or NME on MRI. However, the sample size of five is small limiting any conclusions. To our knowledge, there is no study which describes the percentage of cases of NME or foci seen on MRI which could be attributed to LN specifically.

Mass lesions

In our study, all mass lesions were found to be due to another benign or pathologic lesion while LN was considered an incidental finding. This is an agreement with most literature which indicate LN is not thought to form mass lesions visualized with mammography, ultrasonography, or MRI.

If a mass lesion is seen and pathology returns lobular neoplasia, the pathology should be considered discordant unless there is a benign finding other than LN that would explain the imaging feature of a ‘mass’ such as fibroadenoma, fibroadenomatous change, sclerosing adenosis, and focal fibrocystic changes, amongst other benign entities. The general consensus is that LN does not form mass lesions.

Furthermore, several studies have shown that mass lesions which returned LN on CNB, and later went onto excisional biopsy almost always show an alternative histology explaining the mass, commonly DCIS or invasive carcinoma [6–8,22,23]. Middleton et al. described 6 of 17 lesions detected as mass lesions which returned as LN on CNB. All were upgraded to DCIS or invasive carcinoma at excision [23]. Therefore, a mass lesion should not be considered concordant if found in association with LN on CNB without an alternative to explain the mass lesion.

Since mass lesions are not considered concordant findings, ultrasound is an uncommon method to detect LCIS. Asymmetry and architectural distortions may be seen better on ultrasound but the mainstay of imaging LCIS is mammography and MRI. If a lesion is detected on ultrasound, whether it be a mass lesion, focal asymmetry or architectural distortion and the CNB returns as LCIS the biopsy should be considered discordant unless there is a concomitant benign finding to explain the lesion.

Conclusion

Although the majority of ALH and LCIS lesions are incidentally identified, 28% of cases in our study demonstrated non-incidental concordant imaging findings. Among those, over half presented as suspicious calcifications on mammography with the remaining identified as NME or a focus on MRI. In a minority of cases, mammographically detected calcifications and foci or NME on MRI are findings that could be considered concordant imaging findings of LN found during breast core needle biopsy after careful radiologic–pathologic correlation.