Main

To excise or not to excise is the question posed when certain non-malignant but high-risk breast lesions are diagnosed on core needle biopsy after imaging detection. The appropriate post-core needle biopsy management of these lesions causes debate, with a variety of practices including surgical excision, close-clinical follow-up with short-term repeated imaging and possible consideration for risk-reducing medication. Currently, high-risk lesions are diagnosed in fewer than 9% of core needle biopsy cases, but volumes will likely increase as advances in breast imaging techniques increase lesion detection.1, 2, 3, 4

Core needle biopsy diagnoses such as atypical ductal hyperplasia, flat epithelial atypia, atypical lobular hyperplasia, lobular carcinoma in situ, and radial scar are some the non-malignant lesions that are considered high risk for carcinoma. Patients with radial scar and atypical ductal hyperplasia were found to have a two-fold and up to five-fold increased risk of breast cancer over the general population, respectively.5, 6, 7 Increased risk of ipsilateral breast cancer was seen in patients with atypical lobular hyperplasia and lobular carcinoma in situ.8 And histologic and genetic similarities have been described between flat epithelial atypia and carcinoma.9, 10, 11

Notably, these high-risk lesions are not only markers of future carcinoma, but also indicators of concurrent carcinoma missed due to biopsy sampling. Upgrade rates to ductal carcinoma in situ or invasive carcinoma from core needle biopsy to excision in the literature range from 0 to 62% for atypical ductal hyperplasia, 0 to 21% for flat epithelial atypia, 0 to 67% for atypical lobular hyperplasia, 0 to 60% for lobular carcinoma in situ, and 0 to 16% for radial scar.12, 13, 14, 15, 16, 17, 18, 19, 20, 21 Given these data, it is not surprising that the management of high-risk breast lesions seen on core needle biopsy is controversial. More defined upgrade rate data are needed and may prevent over- and undertreatment.

The present study evaluates excision results for patients with non-malignant high-risk lesions, including atypical ductal hyperplasia, flat epithelial atypia, atypical lobular hyperplasia, lobular carcinoma in situ, and radial scar without atypia diagnosed on core needle biopsy at our institution. Additionally, we investigate the predictive role of clinicopathological features including patient age, Breast Imaging Reporting and Data System (BI-RADS) score, calcification span, mass size, and needle gauge on management and excision diagnosis. This is one of the largest studies elaborating on variables that predict benign diagnosis on excision, in addition to establishing baseline upgrade rates. We subsequently present the most comprehensive literature review to date, and conclude with a summary of our statistically significant findings with recommendations, providing a reference for the management of high-risk breast lesions diagnosed on core needle biopsy.

Materials and methods

After obtaining IRB approval, our pathology database was searched using the terms atypical ductal hyperplasia, flat epithelial atypia, atypical lobular hyperplasia, lobular carcinoma in situ, and radial scar on breast core needle biopsy cases from 1 January 2003 to 30 June 2014. Both imaging-targeted lesions and incidental pathologic findings were included. Radiologists at our institution used smaller 11- to 14-gauge needles from 2003 through 2005, and then larger 9-gauge needles were introduced and used almost consistently after 2007. Over the duration of the study, four subspecialized breast pathologists classified cases. Original slides were not re-reviewed for the purposes of this study. However, per institutional quality assurance policy, every breast core needle biopsy case was reviewed by a second pathologist within 1 week of original sign out. In addition, diagnoses were confirmed by a second pathologist during tumor board for excisional biopsy cases upgraded to ductal carcinoma in situ and/or invasive carcinoma.

For analysis, cases were assigned to appropriate high-risk entity categories (Figure 1). We excluded atypical ductal hyperplasia bordering on ductal carcinoma in situ, pleomorphic lobular carcinoma in situ, other variants of lobular carcinoma in situ such as florid/distended lobular carcinoma in situ with central necrosis, and radial scar associated with any type of epithelial atypia. Core needle biopsy cases with juxtaposed known synchronous ipsilateral breast carcinoma were also excluded. For flat epithelial atypia, cases from 2003 diagnosed using historical names such as ‘columnar cell change with atypia’, ‘columnar cell hyperplasia with atypia’, or ‘columnar alteration with prominent apical snouts and secretions’ were included. Only pure flat epithelial atypia cases were included in the flat epithelial atypia category for this study. If flat epithelial atypia was seen with lobular neoplasia, we tabulated the case under either atypical lobular hyperplasia or lobular carcinoma in situ. If flat epithelial atypia was seen with atypical ductal hyperplasia, then we tabulated the case under atypical ductal hyperplasia. Cases with focal and incidental lobular neoplasia seen with atypical ductal hyperplasia were included in the atypical ductal hyperplasia analysis. Other mixed atypical lesions were excluded.

Figure 1
figure 1

High-risk breast lesions. (a) Atypical ductal hyperplasia with low-grade nuclear atypia demonstrating ‘roman bridge’ and micropapillary architectural complexities. (b) Flat epithelial atypia, often detected due to luminal microcalcifications, consists of a monolayer of cytologically low-grade columnar-shaped cells with luminal ‘tomb-stone’ snouts lining a distended duct. (c) Atypical lobular hyperplasia consisting of uniform, dishesive small cells. (d) Lobular carcinoma in situ with small dishesive cells filling and expanding ductules. (e) Radial scar is a stellate lesion with a central elastic stroma and ducts radiating out, entrapped in a background of fibroelastosis, often mimicking a low-grade invasive ductal carcinoma, as seen on this core needle biopsy. (a, c, e: hematoxylin and eosin × 20; b, d: hematoxylin and eosin × 40)

Full size image

From the electronic medical record, clinical and radiologic data for core needle biopsy cases were recorded, including patient age, Breast Imaging Reporting and Data System (BI-RADS) score, imaging findings (ie mass or calcification), calcification span, mass size, whether or not residual calcifications were noted after core needle biopsy, and needle gauge. Only cases with radiographic findings were included in this study. During the 14.5 year duration of the study, all cases were reviewed during weekly radiology–pathology concordance conferences. Electronic record review of radiology–pathology concordance comments revealed <1% (3 of 462) cases were disconcordant. All three cases with histologic findings not explained by imaging findings were diagnosed as radial scar on biopsy, and two of the three cases went to excision while the third was lost to follow-up.

Our current institution guidelines include recommendation for excision in the setting of atypical ductal hyperplasia, lobular carcinoma in situ, atypical lobular hyperplasia, flat epithelial atypia, and radial scar. More recently, recommendations for excision or surveillance were made on a case-by-case basis after radiology–pathology correlation conference and with patient input.

For the analysis, upgrades were defined as ductal carcinoma in situ, pleomorphic lobular carcinoma in situ, and invasive mammary carcinoma on excision. Residual high-risk lesions such as atypical ductal hyperplasia, lobular carcinoma in situ, atypical lobular hyperplasia, and flat epithelial atypia on excision were not included as upgrades.

For categorical variables, statistical comparisons between groups were made using χ2 or Fisher's exact test, where appropriate. For normally distributed continuous variables, comparisons were made using t-test. Multivariate analysis was conducted using stepwise logistic regression. All tests of significance were two-tailed, and significance was set at P<0.05. Statistical analysis was performed with SPSS statistical software (Statistical Package for the Social Sciences, IBM Corp., Version 22.0, Armonk, NY, USA).

Results

Over 14.5 years, 5750 core needle biopsy cases were collected, including 2589 stereotactic, 2913 ultrasound-guided, and 248 MRI-guided biopsies. A total of 462 (8%) high-risk breast lesion core needle biopsy cases were identified, including 249 (4.3%) atypical ductal hyperplasia, 72 (1.3%) flat epithelial atypia, 50 (0.9%) atypical lobular hyperplasia, 37 (0.6%) lobular carcinoma in situ, and 54 (0.9%) radial scar. The average patient age in years at index diagnosis was 55.0 for atypical ductal hyperplasia, 52.2 for flat epithelial atypia, 52.6 for atypical lobular hyperplasia, 54.2 for lobular carcinoma in situ, and 53.2 for radial scar.

Excision results were available for 333 cases, including 78, 77, 76, 64, and 46% of lobular carcinoma in situ, atypical ductal hyperplasia, flat epithelial atypia, atypical lobular hyperplasia and radial scar cases, respectively (Table 1). No excision results were available for 129 cases, including 22, 23, 24, 36, and 54% of lobular carcinoma in situ, atypical ductal hyperplasia, flat epithelial atypia, atypical lobular hyperplasia, and radial scar cases, respectively. Of 129 cases that did not undergo excision, 67 (52%) were lost to follow-up, while 62 (48%) were not excised for other reasons. Many cases lost to follow-up either had no further documentation in the medical record after the biopsy of interest, or had subsequent major illness diagnosed or emergency department visit documented. Cases with no excision that were not lost to follow-up had imaging follow-up at 6 months or later describing no progression of the targeted lesion of concern. Nine of 62 (15%) of no excision cases had surgeon documentation of patient decision to not proceed with excision.

Table 1 Management and excision outcomes of high-risk lesions diagnosed on CNB
Full size table

Follow-up results were further analyzed by entity. Of 57 atypical ductal hyperplasia cases that were not excised, 34 (61%) were lost to follow-up, while 23 (40%) were not excised, including 14 cases with imaging follow-up, 7 cases not excised due to patient input, and 2 cases treated with risk-reducing medication. Of 17 flat epithelial atypia cases that were not excised, 8 (47%) were lost to follow-up, while 9 (53%) were not excised including 8 cases with imaging follow-up and 1 case not excised due to patient input. Of 18 atypical lobular hyperplasia cases that were not excised, 7 (39%) were lost to follow-up, while 11 (61%) were not excised including 10 cases with imaging follow-up and 1 case not excised due to patient input. Of eight lobular carcinoma in situ cases that were not excised, 6 (75%) were lost to follow-up, while 2 (25%) were not excised, but had imaging follow-up. Finally, of 29 radial scar cases that did not undergo excision, 12 (41%) were lost to follow-up, while 17 (58%) had documentation of recommendation for imaging surveillance (specifically 6-month follow-up) after radiology-pathology consensus conference.

Univariate analysis was performed overall and by entity to assess for differences between the lost to follow-up and no excision groups and yielded no significant differences in patient age and radiographic features, including needle gauge, BI-RADS score, calcification vs mass targeting and lesion size; however, our analysis is limited by low case numbers.

Of 333 excision cases, the diagnosis was benign in 98 (29%), residual high risk in 179 (54%), and upgraded to carcinoma in 56 (17%) (Table 1). Upgraded cases included 42 ductal carcinoma in situ, 9 invasive ductal carcinoma, 3 invasive lobular carcinoma, 1 pleomorphic lobular carcinoma in situ, and 1 isolated tumor cells. Most invasive tumors had low to intermediate modified Bloom and Richardson grade and scores, estrogen receptor (ER) positive, progesterone receptor (PR) positive, and human epidermal growth factor receptor 2 (HER2) protein negative by immunohistochemistry and/or fluorescent in situ hybridization with low proliferation indices (Table 2). By entity, the upgrade rate was 18% (35/192) for atypical ductal hyperplasia, 11% (6/55) for flat epithelial atypia, 9% (3/32) for atypical lobular hyperplasia, 28% (8/29) for lobular carcinoma in situ, and 16% (4/25) for radial scar.

Table 2 Characteristics of invasive tumors seen on excision of high-risk lesions diagnosed on CNB
Full size table

Univariate analysis was utilized to elaborate on differences in age and radiological features based on excision status (excision vs no excision) for the 462 high-risk lesion cases, and excision diagnosis (upgrade vs no upgrade, and benign vs not benign) for the 333 cases that went to excision (Table 3).

Table 3 Comparison of cases based on excision status and excision diagnosis
Full size table

Core needle biopsy diagnosis was the only significantly different variable between patients who received excision and patients with no excision (P<0.001) (Table 3, left). On multivariable logistic regression analysis, compared with atypical ductal hyperplasia, lesions diagnosed as radial scar on core needle biopsy were less likely to be excised (OR=0.26, P<0.0001, CI=0.14–0.47). A trend towards younger patients (<50 years old) with more concerning imaging (BI-RADS score ≥4B) and targeting of calcifications or larger masses (≥1 cm), with residual calcifications after biopsy was noted among cases that went to excision. However, patient age, BI-RADS score, calcification targeting, mass targeting, and needle gauge were not statistically significant predictors of surgical management (excision vs no excision). Variables with excessive missing values (calcification span, residual calcification on mammogram post-core needle biopsy, and mass size) were excluded from regression models.

Comparing upgrade vs no upgrade (high-risk or benign) groups yielded no significantly different variables (Table 3, middle). A trend towards larger mass size (≥1 cm) and presence of residual calcification was appreciated among upgraded cases. BI-RADS scores and patient age were similar between the upgrade and no upgrade groups.

The data were reorganized to evaluate differences between benign vs not benign (high risk or upgrade) groups (Table 3, right). Based on logistic regression modeling, radial scar was less likely (vs atypical ductal hyperplasia) to be high risk or carcinoma on excision (OR=0.29, P=0.014, CI=0.11–0.77). Cases that were not benign on excision (vs benign) were more likely to have residual calcifications (27% vs 17%, P=0.013), and mass size greater than 1 cm (50% vs 17%, P=0.001). On post hoc subgroup analysis, atypical ductal hyperplasia cases that were high risk or upgrade (vs benign) on excision were more likely to have larger masses (≥1 cm) (P=0.025). No other significant difference was found between benign and not benign groups within the separate atypical ductal hyperplasia, flat epithelial atypia, atypical lobular hyperplasia, lobular carcinoma in situ, or radial scar subgroups on post hoc subgroup analysis. However, among cases that were benign on excision, there was a trend towards lower BI-RADS scores (<4B), smaller mass sizes (<1 cm), and no targeting of calcifications or residual calcifications.

Of note, the upgrade rate was highest for lobular carcinoma in situ (28%), but this rate decreased to 7% (1/14) for incidental lobular carcinoma in situ, and lobular neoplasia (lobular carcinoma in situ or atypical lobular hyperplasia) diagnosed incidentally (vs targeted) on core needle biopsy was significantly less likely to upgrade on excision (5% vs 39%, P=0.002) (Table 4).

Table 4 Incidental lobular neoplasia is less likely to upgrade on excision
Full size table

Discussion

Optimal management for high-risk breast lesions diagnosed on core needle biopsy remains controversial. Institutional differences in excision rates underscore the lack of consensus.22, 23, 24 The wide range of published upgrade rates to ductal carcinoma in situ and/or invasive mammary carcinoma on excision partly contributes to the problem. Here we present the most comprehensive literature review of reported upgrade rates for atypical ductal hyperplasia, atypical lobular hyperplasia, lobular carcinoma in situ, flat epithelial atypia, and radial scar, along with a summary of the key findings from our statistical analysis with management recommendations (Tables 5, 6, 7, 8).

Table 5 Literature review of upgrade rates of ADH diagnosed on CNB
Full size table
Table 6 Literature review of upgrade rates of lobular neoplasia diagnosed on CNB
Full size table
Table 7 Literature review of upgrade rates of FEA diagnosed on CNB
Full size table
Table 8 Literature review of upgrade rates of radial scar without atypia diagnosed on CNB
Full size table

The estimated rate of atypical ductal hyperplasia diagnosis on core needle biopsy has been reported at 3%, and atypical ductal hyperplasia was diagnosed in 4.3% of core needle biopsy cases at our institution.25 On literature review, 23% of atypical ductal hyperplasia cases were upgraded on average, with a range of 0–62% (Table 5).12, 13, 17, 19, 22, 24, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60 In our study, the upgrade rate for patients with atypical ductal hyperplasia diagnosis on core needle biopsy was 18%. No significant decrease in frequency of upgrade for atypical ductal hyperplasia was seen when a larger needle was used for biopsy. Consistent with published findings, atypical ductal hyperplasia lesions associated with a smaller targeted mass (<1 cm) in our study were statistically more likely to be benign on excision.34 However, 17% (4/23) of atypical ductal hyperplasia core needle biopsy cases involving targeting of smaller masses (<1 cm) still upgraded to carcinoma. In agreement with generally accepted practice guidelines, we conclude that atypical ductal hyperplasia seen on core needle biopsy merits surgical excision.

The estimated rates of lobular neoplasia diagnosis on core needle biopsy have been reported at approximately 1%.25 Our institutional diagnosis rate was 0.9% for atypical lobular hyperplasia and 0.6% for lobular carcinoma in situ. Literature review yielded a 9% mean upgrade rate for atypical lobular hyperplasia (range 0–67%), and an 18% mean upgrade rate for lobular carcinoma in situ (range 0–60%) (Table 6).13, 14, 18, 19, 21, 28, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83 At our institution, the upgrade rate for atypical lobular hyperplasia was 9%, and the upgrade rate for lobular carcinoma in situ was 24%. Recent studies examining incidentally diagnosed lobular neoplasia reported rates below 2%.78, 82 In agreement with these findings, radiology–pathology correlation analysis of our excised lobular neoplasia cases revealed that incidental atypical lobular hyperplasia and lobular carcinoma in situ had an upgrade rate of 5%, significantly lower than the 39% upgrade rate for targeted lobular neoplasia (P=0.002). Larger needle size in the setting of lobular neoplasia did not decrease upgrade frequency. Similar to the National Comprehensive Cancer Network 2015 Guidelines, we support surgical excision when atypical lobular hyperplasia and lobular carcinoma in situ are diagnosed on core needle biopsy.84

In our study, the rate of pure flat epithelial atypia diagnosis on core needle biopsy was 1.3% with an upgrade rate of 11%. On literature review, approximately 8% of pure flat epithelial atypia diagnosed on core needle biopsy was upgraded to carcinoma, with a range of 0–21% (Table 7).19, 20, 22, 24, 30, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104 In our study, no statistically significant radiologic differences were seen between upgraded and not upgraded flat epithelial atypia core needle biopsy groups; however, residual calcification after biopsy was associated with upgrade. For pure flat epithelial atypia, as stated in the World Health Organization Working Group, careful radiological–pathological correlation is necessary, and in the absence of residual calcifications or other indications for excision, surveillance may be acceptable.105

The estimated rate of radial scar diagnosis on core needle biopsy has been reported at 1–2%, and our rate was 0.9%.25 On literature review, a 7% average upgrade rate was reported with a range of 0–16% when radial scar was diagnosed on core needle biopsy (Table 8).2, 3, 15, 16, 19, 23, 24, 44, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118 Our institutional upgrade rate of 16% was higher than the reported average, likely due to our small case volume (n=25). No statistically significant radiologic differences were seen between upgraded and not upgraded radial scar core needle biopsy groups in our study. Previous research has shown that incidental and small radial scar lesions (<5 mm) were less likely to be upgraded on excision.16, 110, 115 There is no consensus for the management of radial scar without atypia. It may be prudent to recommend conservative excision for radial scar without atypia with radial scar size greater >5 mm. After careful radiological–pathological correlation, watchful surveillance may be acceptable for incidental and small radial scar seen on core needle biopsy. Radial scar with atypia should be excised.

With the exception of mass size ≥1 cm for atypical ductal hyperplasia, no other variables, including BI-RADS score, target lesion (mass vs calcification), mass size, calcification span, needle gauge, or residual calcification seen on mammogram post-core needle biopsy were predictive of upgrade to carcinoma excision for any other entity in our study. However, our analysis carries biases inherent to retrospective reviews, including insufficient case numbers to power statistical regression analysis, and incompletely available data. Heterogeneity in study design, including variations in size, inclusion and exclusion criteria, imaging modality, biopsy technology, outcome metrics and inclusion of other histopathology in the samples studied, precluded combination of the studies collected into an effective meta-analysis.

This is the largest single-institution analysis of a variety of high-risk breast lesions diagnosed on core needle biopsy, including management recommendations for atypical ductal hyperplasia, flat epithelial atypia, atypical lobular hyperplasia, lobular carcinoma in situ, and radial scar. For years teams have published upgrade analyses of high-risk breast lesions diagnosed on core needle biopsy in an effort to prevent over- or undertreatment, and management remains problematic. We agree with the general opinion in the literature that a multicenter controlled study evaluating high-risk breast lesions is needed.25, 119, 120, 121 Research in molecular differences between groups of high-risk lesions may offer further risk stratification. Because ductal carcinoma in situ is being considered for a ‘downgrade’ to the diagnosis ‘borderline breast lesion’ with approved trials randomizing patients to active monitoring and standard treatment, similar studies could be performed with high-risk breast lesions such as atypical ductal hyperplasia and lobular carcinoma in situ.122 Long-term studies will be important to better characterize the risks and benefits of excision.