Updates on Lobular Neoplasms and Papillary, Adenomyoepithelial, and Fibroepithelial Lesions of the Breast

Noninvasive lobular neoplasms include atypical lobular hyperplasia (ALH), classic type lobular carcinoma in situ (C-LCIS), florid lobular carcinoma in situ (F-LCIS), and pleomorphic lobular carcinoma in situ (P-LCIS). ALH and C-LCIS belong to the “high-risk lesions” of the breast, which comprise 1% to 15% of over 1 million image-guided core biopsies in the United States each year.^1–5  The term “high risk” applies to a group of lesions diagnosed on image-guided core biopsy that bear a high rate of upgrade to carcinoma in subsequent excision. Other high-risk lesions include atypical ductal hyperplasia (ADH), radial scar, flat epithelial atypia, papilloma, and mucocele-like lesions.^6–8  ALH and C-LCIS are characterized by a proliferation of monotonous low-to-intermediate grade dyscohesive tumor cells with decreased or absent expression of E-cadherin, which is encoded by the CDH1 gene located on the long arm of chromosome 16. C-LCIS was first described by Foote and Stewart in 1941.⁹  C-LCIS is diagnosed when the terminal ductal lobular unit is expanded and 50% or more of the expanded terminal ductal lobular units are involved by the tumor cells. ALH is diagnosed when the criteria for C-LCIS are not met (Figure 1, A). ALH and C-LCIS are generally clinically occult lesions and often multicentric.^10,11  ALH and C-LCIS are risk factors and nonobligate precursors of invasive mammary carcinoma. Patients with ALH had a 2.56 to 6.79 times increased risk of developing breast cancer (invasive carcinoma or ductal carcinoma in situ [DCIS]) in a study with 2 large cohorts.¹²  The cancer risk increases with increased number of ALH foci.¹²  In another study, the odds ratio of breast cancer risk was 5.2 (95% CI 2.7–10.0) for 1 to 2 foci of ALH versus 8.0 (95% CI 4.5–14.2) for 3 foci of ALH, although this did not reach statistical significance.¹³  The increased cancer risk of ALH is not related to age, indicating the molecular alterations trump other risk factors.¹⁴  Patients with ALH have twice the risk of developing ipsilateral breast cancer than developing contralateral breast cancer and such ipsilateral risk predominance is more pronounced in the first 5 years, supporting that ALH is a risk factor as well as a nonobligate precursor.¹⁵  In the same study, 76.8% of the invasive carcinomas developed in patients with ALH were the ductal type, similar to the rate of ductal type (78.0%) of invasive carcinoma developed in patients with ADH.¹⁵  This is consistent with molecular studies showing shared but evolving genetic changes from ALH, ADH to C-LCIS, and low-grade DCIS, and then to low-grade invasive breast carcinoma.^16–20  Intuitively and proven by studies, patients with C-LCIS have a higher cancer risk than patients with ALH. Patients with C-LCIS have a 10-year risk of approximately 15% to develop invasive carcinoma, which is a 7- to 10-fold increase compared with the general population.^21–25 

The management of ALH and CLCIS diagnosed on core biopsy specimens is still controversial. Immediate surgical removal of any high-risk breast lesion is to identify and excise coexisting carcinoma rather than to decrease subsequent breast cancer risk. The upgrade rate of ALH and C-LCIS varied in studies. These variations are multifactorial, including whether assiduous radiology-pathology (rad-path) correlation was implemented, accuracy of diagnoses, different patient population, breast tissue sampling during the biopsy procedure, and the length of follow-up. Of note, nearly all publications are retrospective studies and biases may exist. In studies with vigorous rad-path correlations, the upgrade rate of ALH and C-LCIS is low.^26–34  A recent study with prospective decisions from multidisciplinary, high-risk breast lesion conferences showed no upgrade rate in 7 C-LCIS cases and no disease progression in 5 patients with C-LCIS and 19 patients with ALH during follow-up (mean follow-up of 336 days for C-LCIS and 668 days for ALH).⁷  These data indicate that the patients with ALH and LCIS diagnosed in core biopsy specimens with rad-path concordance and no concerning pathologic or radiologic features can be followed. However, such recommendation should only be rendered after careful rad-path correlation and a consensus is agreed upon by breast pathologists, imagers, and surgeons.⁷  Chemoprevention with hormonal therapy could be considered in patients with C-LCIS or ALH, which decreases the cancer risk by more than 50%.³⁵  The margin status of C-LCIS in excisional specimens does not need to be reported because a positive margin does not increase recurrence rate.^11,36 

Pleomorphic LCIS (P-LCIS) and florid (F-LCIS) are rare variants of LCIS. P-LCIS has marked nuclear pleomorphism with at least 3 to 4 times the variation (Figure 1, B) in nuclear size. P-LCIS can show apocrine features.³⁷  F-LCIS is defined as a lobular neoplasm with markedly distended acini of involved terminal ductal lobular units with little or no intervening stroma or distended acini/ductile filling 1 HPF, which is equivalent to 40 to 50 cells in diameter (Figure 1, C and D).³⁸  The tumor cells in F-LCIS are low-to-intermediate grade, similar to C-LCIS. Although both LCIS variants are often associated with comedonecrosis, comedonecrosis is not a required diagnostic criterion. The majority of P-LCIS and F-LCIS cases are associated with concurrent C-LCIS.³⁷  While almost all C-LCIS lesions are positive for estrogen receptor (ER) and negative for human epidermal growth factor receptor 2 (HER2), P-LCIS and F-LCIS can be ER negative and are more frequently positive for HER2 overexpression, in up to 41% of P-LCIS cases.^37,39–41  Both of the LCIS variants diagnosed in core biopsy specimens have a high upgrade rate in excision.^37,42–44  Therefore, unlike ALH and C-LCIS, P-LCIS and F-LCIS diagnosed in a core biopsy specimen should be excised. Because F-LCIS has similar cytologic features as C-LCIS but the managements can be different, it is imperative to differentiate F-LCIS from C-LCIS in core biopsy specimens. In a study including 61 P-LCIS and 24 F-LCIS, 75% of the F-LCIS were associated with invasive carcinoma and 4% were associated with DCIS; 77% of the P-LCIS were associated with invasive carcinoma and 7% were associated with DCIS; and the majority of the invasive carcinomas were lobular type.³⁷  Studies show different recurrence rates of P-LCIS and F-LCIS after excision.^{37,39,40,45,46}  Although the true outcome of P-LCIS and F-LCIS is unclear, it is recommended to report positive or close margins to F-LCIS and P-LCIS in excisional specimens. The utility of radiotherapy in patients with the LCIS variants without invasion is unknown.

Both P-LCIS and F-LCIS share genetic alterations with C-LCIS, including gain of 1q, loss of 16q, and the CDH1 gene mutations. However, P-LCIS and F-LCIS show more complex genetic alterations, including a high rate of ERBB2 gene mutations and amplifications and other genetic alterations, including mutations in TP53, FOXA1, ERBB3, PIK1CA, RUNX1, and other genes.^47–49  These molecular studies strongly indicate P-LCIS and F-LCIS are related to C-LCIS but in a more advanced stage.

First described by Hamperl⁵⁰  in 1970, adenomyoepithelioma (AME) is a rare biphasic breast tumor composed of cuboidal to columnar and sometimes inconspicuous epithelial-lined tubules surrounded by a layer of myoepithelial cells arranged in tubular, papillary, solid, or mixed-growth patterns (Figure 2, A through C).⁵¹  The proliferation of epithelial or myoepithelial cells or both underlies a spectrum of AME from benign to malignant with squamous, glandular, spindle, or clear cell morphology.^51,52  The typical presentation of AME is a solitary mass in postmenopausal women detected by palpation or on screening mammograms.⁵³  Benign AMEs are usually small, round, or lobulated with either solid or combined cystic appearance. Immunohistochemistry staining for the epithelial component (such as AE1/AE3 and CAM5.2) and myoepithelial markers (such as p63, calponin, and smooth muscle actin) can be helpful in identifying the biphasic components in cases with complex morphology (Figure 2, D). The majority of AMEs are ER negative, but the tumor can be focally positive for ER staining, which is mainly in epithelial cells (Figure 2, E and F).

Tumors with malignant potential manifest as rapidly enlarging lesions with infiltrative growth pattern, severe nuclear atypia, and increased mitotic rate (Figure 2, G and H). Necrosis is more frequent in malignant AME, but it can be seen in the central zones of benign tumors (Figure 2, I). The malignant component could be composed of luminal, myoepithelial, or both cellular components. Malignant AMEs can resemble poorly differentiated epithelial-myoepithelial carcinoma of the salivary gland when the biphasic differentiation is retained.⁵³  When only the epithelial or myoepithelial component undergoes malignant transformation, it can resemble metaplastic tumors.⁵⁴  Therefore, AME may be considered a nonobligate precursor of some metaplastic breast carcinoma variants, such as low-grade fibromatosis-like, spindle cell, and myoepithelial carcinoma, sharing a similar mutational landscape.⁵⁵  Similar myriad interactions between the luminal epithelium and myoepithelium can be seen in other breast tumors. Morphologic distinction from sclerosing lesions, papillary tumors, adenoid cystic carcinoma, low-grade adenosquamous carcinoma, or other types of metaplastic tumors of the breast can be challenging.⁵²  There has been a significant effort to identify diagnostic and prognostic biomarkers ancillary to the evaluation of AME.⁵⁶ 

A combination of next-generation sequencing modalities has allowed the comprehensive molecular characterization of AMEs (Figure 3). Overall, AMEs display fewer copy number alterations and lower mutational burden than conventional breast carcinoma with a ductal phenotype.⁵⁶  Frequent copy number alterations include losses of 6p22 (19%), 9p21 (CDKN2A, 13%), and 4q31 (INPP4B, 6%) and gains of 12p12.3 (ETV6, 16%) and 5p15 (TERT, 13%). Homozygous deletions and gene amplifications are uncommon, but when present, they correlate with a malignant phenotype.⁵⁶ HER2 amplifications are never encountered in AMEs, but they can be detected within co-occurring intraductal proliferations, suggesting that AME and ductal neoplasia are probably clonally unrelated.^57,58 

Recurrent somatic alterations in members of the MAPK and PI3K-AKT signaling pathways are the proposed genetic drivers for AMEs.^52,56  These hotspot mutations are associated with ER status. For instance, AKT1 E17K is only found in ER-positive AMEs, while HRAS Q61R/K always co-occurs with PIK3CA H1047R, E545K, E542K mutations or small PIK3R1 deletions in ER-negative AMEs.⁵⁶  Experimental data support that the acquisition and persistence of ER-positive luminal or ER-negative basal phenotypes in AMEs are probably facilitated by oncogenic signaling triggered by AKT1 E17K and HRAS Q61R, respectively.^56,59 

The presence of HMGA2 rearrangements in a small subset of ER-positive AMEs lacking hotspot mutations highlights the molecular and morphologic resemblance with epithelial-myoepithelial lesions of the salivary glands.^60,61  Fusion transcripts detected in other tumors with myoepithelial differentiation, such as PLAG1 fusions in pleomorphic adenomas and MYB-MYBL1 in adenoid cystic carcinomas, have not been detected in AMEs by RNA-seq or fluorescence in situ hybridization analysis.^61,62  The HMGA2-WIF1 fusion has been described in a benign AME.⁶¹  AMEs with HMGA2 rearrangement exhibit tubular and papillary growth pattern and bland cytology with an indolent clinical course. This is different in salivary gland tumors where HMGA2 rearrangement is more commonly associated with progression of pleomorphic adenoma to epithelial-myoepithelial carcinoma.^60,61,63 

From a practical perspective, demonstration of HRAS Q61 hotspot mutations can potentially stratify the risks associated with AMEs.⁵⁶  Preliminary data reveal that the immunohistochemical expression of HRAS Q61R mutation in AME, using a highly specific monoclonal (SP174) antibody, correlates with the infiltrative growth, tumor necrosis, and increased mitotic rate.⁶⁴  Contrary to AKT1 and PIK3CA mutations, which can be seen in diverse types of breast tumors, HRAS Q61R appears to be specific for AMEs and is not reported in adenoid cystic carcinoma and papillary lesions of the breast.^65,66 

The majority of AMEs have a benign clinical course, and surgical excision is sufficient for the management of localized disease.⁶⁷  However, recurrence and distant metastasis have been reported even in AMEs without overt features of atypia or malignancy.⁶⁸  Local recurrence may develop, usually within 2 years after resection, owing to satellite nodules away from the main mass.⁵³  Malignant AMEs have a higher potential to recur locally and/or may have metastatic potential, especially those with high nuclear grade.⁶⁹  Complete excision with negative margins is the mainstay of treatment of AMEs.^53,69  Given that AME primarily disseminates hematogenously, axillary lymph node sampling is generally not recommended.⁷⁰  Data about the outcome associated with recurrent or metastatic disease are limited to a few case reports. Chemoradiotherapy and hormonal therapy do not appear to improve the overall survival.⁷¹  So far, the rarity of AME cases and even lower incidence of malignant cases preclude the assessment of targeted therapy.

Papillary neoplasms of the breast include a few categories with different clinical behaviors and treatment modalities. Recent studies have advanced our knowledge in the optimal patient managements.

Intraductal papilloma (IDP) without atypia is a benign neoplasm. ADH or DCIS within IDP may not be readily diagnosed in core biopsy specimens due to limited sampling. When IDP is involved by atypical epithelial proliferation with high nuclear grade, the diagnosis of IDP with DCIS should be made regardless of its extent.³⁸  Studies show mixed upgrade rate to invasive carcinoma or DCIS when IDP is excised.^6,8,72,73  Recent studies have indicated that the upgrade rate of IDP with careful and concordant rad-path correlations is low and these IDPs may not require surgical excision.^6,8,74  IDP involved by ADH or DCIS warrants excision owing to the high upgrade rate.^72,75 

Papillary DCIS refers to DCIS with a pure papillary growth pattern. The nuclear grade of papillary DCIS is usually low to intermediate. Of note, it is the most common type of DCIS in males.⁷⁶  Papillary DCIS can be differentiated from IDP involved with DCIS by lack of residual IDP component. Papillary DCIS is absent of myoepithelial cells along the fibrovascular cores but has myoepithelial cells at the epithelium-stroma interface at the periphery. Although the myoepithelial cells can be scant when the involved ducts are dramatically expanded, immunohistochemistry staining for myoepithelial cells may be helpful in establishing the diagnosis. Papillary DCIS is clinically managed as other types of DCIS.

Encapsulated papillary carcinoma (EPC) is characterized by proliferation of neoplastic epithelial cells with low or intermediate nuclear grade with fibrovascular cores. EPC is generally surrounded by a fibrous capsule (Figure 4, A). Myoepithelial cells are not found within or at the periphery of EPC.^77,78  There are debates on whether it should be considered invasive carcinoma due to the complete loss of myoepithelial cells. However, available outcome studies indicate excellent prognosis of EPC with complete excision.^77,79  When frank invasion through the capsule is identified, the tumor should be staged based on the size of the invasive component only and not including the EPC component. Without tumor invasion through the capsule, patients with EPC have excellent prognosis, and EPC should be managed conservatively as DCIS.^77,79  The majority of EPCs have low-to-intermediate nuclear grade. A recent study showed EPC can have high nuclear grade and increased mitotic activity, and these high-grade EPCs tend to be negative for ER and progesterone receptor.⁸⁰  In this study, 1 of 10 patients with high-grade EPC without any invasive component died of recurrent disease and metastasis after initial surgical procedure and radiotherapy. The remaining 9 patients had no significant events during follow-up.⁸⁰ 

Solid papillary carcinoma (SPC) manifests a solid growth pattern with circumscribed border and delicate fibrovascular cores (Figure 4, B). Without an accompanying invasive component, SPC is also managed conservatively as DCIS.^81,82  Patients with SPC with extravasated mucin may have guarded prognosis and these patients may benefit from close follow-up.⁸²  Invasive carcinoma may have SPC pattern but shows infiltrating growth pattern with irregular borders and ragged contours, which create a geographic jigsaw pattern with desmoplastic stroma (Figure 4, C and D). It may be diagnosed as invasive SPC.⁸³  Vascular invasion can sometimes be identified. Invasive carcinoma with SPC pattern should be clinically managed as other invasive mammary carcinomas with similar stage and biomarker status.

Pure invasive papillary carcinomas (IPCs) of the breast are rare, accounting for less than 1% of invasive breast cancers.⁸⁴  The majority of IPCs are diagnosed in older women.^85,86  IPC usually has a relatively well-demarcated border but displays frankly invasive growth pattern with fibrovascular cores surrounded by neoplastic cells. It is not associated with EPC or SPC. Most of the reported cases show low-to-intermediate nuclear grade.⁸⁶  Owing to its rarity, a diagnosis of IPC should be made cautiously after excluding metastatic carcinomas with papillary architectures, such as ovarian serous carcinoma, thyroid papillary carcinoma, or papillary adenocarcinoma of lung. Most of the metastatic carcinomas are positive for organ-specific biomarkers, such as Wilms tumor suppressor gene-1, paired box gene-8, thyroglobulin, thyroid transcription factor-1, napsin A, and so on. IPC of the breast is negative for these markers but is usually positive for GATA-3, ER, and progesterone receptor.^85,86  Different from IPC, invasive micropapillary carcinoma of the breast is characterized by small, hollow, or morula-like clusters of tumor cells without fibrovascular cores surrounded by empty space. Invasive micropapillary carcinoma generally shows inside-out growth pattern with inverted cell polarity.^87,88  The prognosis of IPC is not clear. It is recommended to be managed as invasive ductal carcinoma of no special type. Table 1 summarizes the clinicopathologic features of the above papillary lesions.

Fibroepithelial lesion (FEL) of the breast is a common entity in the breast characterized by proliferation of both the epithelial and stromal components. It consists of 2 major lesions—fibroadenoma (FA; a benign lesion with several variants) and phyllodes tumor (PT; with 3 different grades—benign, borderline, and malignant).⁸⁹  Periductal stromal tumor has been classified as a variant of PT by the 2019 World Health Organization classification of breast tumors.^38,90 

Clinically, FA is usually asymptomatic. It occurs more frequently in younger patients and is generally smaller compared with PT. FAs are hormone sensitive and therefore may show cyclical changes in size, and increase in size during pregnancy and regress after menopause. PTs are often symptomatic due to the large size and fast growth. Histologically, there are 2 distinct patterns for FEL as follows: the intracanalicular pattern—in which the stromal component compresses and stretches the ductal elements to form slit-like epithelium–lined spaces; and the pericanalicular pattern—in which the stroma grows around the open tubules and acini with occasional compression of the rounded tubules (Figure 5, A). These 2 patterns often coexist in FA. In PT, there is increased stromal cellularity and an exaggerated and accentuated intracanalicular growth pattern resulting in the leaf-like structure (Figure 5, B). These are the 2 key diagnostic features for PT. PTs are further classified as benign, borderline, or malignant based on the following morphologic features of the stromal cells—hypercellularity, cellular atypia, mitosis, presence or absence of stromal overgrowth, necrosis, and whether the tumor border is pushing or infiltrative.^91–94 Table 2 summarizes the clinicopathologic and molecular features of FA and PTs.

Diagnosis of FA or PT is straightforward in most cases, but occasionally, due to the overlapping features, differential diagnosis between FA and PT, especially its subtypes, such as cellular (Figure 5, C), juvenile FA (Figure 5, D), or FA in young patients, can be problematic. One key feature worth mentioning is that in FA, the biphasic proliferation, both stromal cellularity and glandular distribution, is balanced and uniform throughout the lesion. In PT, the proliferation of stroma and distribution of the epithelial component are often heterogeneous and not uniform. Periductal stromal tumor is also a biphasic lesion and is currently classified as a variant of PT. Periductal stromal tumor shows hypercellular stroma hugging benign breast epithelium and it lacks the leaf-like structure of the PT. The border of periductal stromal tumor is not well-demarcated. Diagnosis of FEL is mainly based on morphologic features. For larger PT lesions, generous sampling is critical for the correct diagnosis as critical morphologic features, such as leaf-like structure or high-grade area, can be very focal. Immunohistochemistry analysis for CD34, Ki67, and P53 have been used in some studies to aid the diagnosis.^95–97  However, these have not been extensively validated and have no proven value for routine clinical use.

There has been significant progress in understanding the molecular changes in FEL, which helps not only in better understanding of the relationships among different types of FEL, but also in making more accurate diagnosis.⁹⁸  There are 2 major molecular pathways in the development of FELs. One pathway is driven by mediator complex subunit 12 (MED12) mutations that are associated with TERT mutations; the other pathway is driven by other gene mutations, including P53, PIK3CA, or EGFR, and so on, in MED12 wild-type FELs. MED12 gene is an X-linked gene and its mutations were first reported in uterine leiomyoma.⁹⁹  The MED12 mutations are more frequently seen in FA and benign PT, and less frequently in malignant PT. The other key genetic alteration associated with MED12 pathway is TERT mutation. The TERT mutations are rarely seen in FA, less frequently in benign PT, and with high frequency in malignant PT.¹⁰⁰  One study showed pediatric FELs harbor MED12 mutations but not TERT promoter hotspot mutations, as seen in the adult population.¹⁰¹  Of note, different subgroups of genes seem to be involved in pediatric FELs as well when compared with conventional FA.¹⁰²  Because FAs in pediatric patients tends to have overlapping morphologic features with benign to borderline PT, lack of TERT mutations reassure their nature as FAs.

Treatment choice for FA is surgical excision, while the asymptomatic FA can be clinically and radiologically followed if patients elect to do so. The current National Comprehensive Cancer Network treatment guideline for PT is wide local excision, with the intent of obtaining surgical margins greater than 1 cm. No axillary staging is needed. Cautious consideration of adjuvant radiotherapy and chemotherapy can be given to recurrent and malignant PT. However, clinicians do not consistently follow this guideline. A recent multicenter study has shown many patients are managed outside of current guidelines.¹⁰³  Recent studies show wide margins may not be necessary to control local recurrence in phyllodes tumor.^103–106  Treatment options for pediatric PT are still very controversial, and optimal treatment options should be carefully planned.¹⁰⁷