Article Text

Download PDFPDF

Expression of p40 and laminin 332 in metaplastic spindle cell carcinoma of the breast compared with other malignant spindle cell tumours
  1. Timothy M D'Alfonso,
  2. Dara S Ross,
  3. Yi-Fang Liu,
  4. Sandra J Shin
  1. Department of Pathology and Laboratory Medicine, Weill Cornell Medical College, New York, New York, USA
  1. Correspondence to Dr Timothy M D'Alfonso, New York-Presbyterian Hospital/Weill Cornell Medical College, 525 East 68th Street, Starr 1031E, New York, NY 10065, USA; tid9007{at}med.cornell.edu

Abstract

Aims To determine the use of p40 and laminin 332 (LN332) immunostains for diagnosing metaplastic carcinoma by studying the expression of these and other routine markers in spindle cell metaplastic carcinomas and other malignant spindle cell tumours.

Methods We identified cases of spindle cell metaplastic carcinoma (n=36) and other atypical/malignant spindle cell tumours, including 20 phyllodes tumours (14 borderline, six malignant) and 23 spindle cell sarcomas (three primary to breast). Immunohistochemical staining was performed for p40 and two LN332 chains, β3 (kalinin B1) and γ2 (lamC2). The expression of these markers was compared with p63 and cytokeratins.

Results p40 and p63 expression was seen in 21 of 36 (58.3%) and 33 of 36 (91.7%) metaplastic carcinomas, respectively. No phyllodes tumours showed stromal expression of p40 or p63. One of 23 (4.3%) sarcomas showed focal weak p63 staining. LamC2 and kalinin B1 expression was seen in 28 of 36 (77.8%) and 26 of 36 (72.2%) metaplastic carcinomas, respectively. LamC2 and kalinin B1 each showed positive stromal cell expression in two of 20 (10%) phyllodes tumours. No sarcomas showed staining with lamC2. Kalinin B1 staining was seen in 17 of 23 (73.9%) sarcomas, including two of three primary breast sarcomas. Cytokeratin expression was seen in 32 of 36 (88.9%) metaplastic carcinomas and diffuse staining was most often seen in 34βE12 and CK5.

Conclusions The diagnostic value of relatively novel markers p40 and LN332 was found to be less than that of routinely used markers (p63 and cytokeratins). p40 proved to be a specific marker but lacked the sensitivity of p63, while LN332 showed staining in a significant proportion of phyllodes tumours and sarcomas.

  • BREAST CANCER
  • IMMUNOHISTOCHEMISTRY
  • BREAST

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Metaplastic carcinoma is a type of triple-negative breast carcinoma that shows non-glandular differentiation in the form of spindle, squamous or heterologous components. It is important that spindle cell (sarcomatoid) metaplastic carcinoma is distinguished from other atypical/malignant spindle cell proliferations of the breast such as phyllodes tumour and the less common primary mammary sarcoma. Sentinel lymph node sampling and chemotherapy (including neoadjuvant therapy) may be offered to patients diagnosed with metaplastic carcinoma, while other atypical and malignant spindle lesions of the breast are most often treated by surgery alone.1–4 A diagnosis of metaplastic carcinoma can be confirmed by showing immunohistochemical expression of various cytokeratins and/or p63. More recently, the use of other immunohistochemical markers with myoepithelial/basal reactivity for diagnosing metaplastic carcinoma has been investigated.3 ,5–9

In the current study, we sought to analyse the staining characteristics of relatively newer antibodies with myoepithelial reactivity, p40 and laminin 332 (LN332), in metaplastic spindle cell carcinomas. p40 is the ΔN isoform of p63, which in two reports has shown variable (0%–60%) expression in metaplastic spindle cell carcinomas.10–12 LN332, formerly laminin 5, is an extracellular matrix protein expressed by myoepithelial cells and a component of basement membranes of ducts and lobules. LN332 contains α3, β3 and γ2 chains and the expression of β3 (kalinin B1) and γ2 (lamC2) chains is specific to laminin 332 and not present in other laminins.13 LN332 has properties promoting tumour invasion; however, the clinical significance of LN332 expression in breast carcinoma is uncertain.14–16 LN332 expression is seen in a high proportion of triple negative breast carcinomas and was observed in 90% of spindle cell metaplastic carcinomas in one prior study.14 We aimed to compare these immunohistochemical markers with more established diagnostic markers including p63 and cytokeratins to determine their potential diagnostic use for distinguishing metaplastic spindle cell carcinoma from other malignant spindle cell tumours arising in the breast.

Materials and methods

Case selection

Thirty-six cases of metaplastic carcinoma with predominantly spindle cell growth (>90%) were identified from our institution's surgical pathology and breast pathology consultation files. Slides were reviewed by two breast pathologists (TMD and SJS) and the diagnoses were confirmed. The diagnosis of metaplastic carcinoma was based on spindle cell reactivity with cytokeratin antibodies and/or p63 or the presence of invasive or in situ ductal carcinoma in association with the tumour. The nuclear grade of spindle cells was assessed and the presence of other elements within the tumour such as squamous differentiation and heterologous elements was recorded. All tumours included in the study were negative for oestrogen receptor, progesterone receptor and human epidermal growth factor receptor 2 (HER-2) (triple-negative), either by report or by staining done at our institution.

Two control groups of atypical/malignant spindle cell tumours were studied to compare staining patterns with those seen in metaplastic carcinomas. One group of 20 mammary phyllodes tumours consisted of 14 borderline phyllodes tumours and six malignant phyllodes tumours. Phyllodes tumours were graded based on established criteria which take into account the degree of stromal cell cellularity, nuclear atypia, mitotic figures, tumour circumscription and stromal overgrowth.17 The second group consisted of 23 sarcomas with spindle cell morphology originating from various sites, including three tumours arising in the breast. Tumours classified as primary mammary sarcomas showed neither cytokeratin nor p63 expression, lacked an epithelial (squamous or glandular) component, were not associated with ductal carcinoma in situ (DCIS) and did not contain fibroepithelial structures (epithelial-lined clefts).

Histological material was available in the form of formalin-fixed paraffin embedded (FFPE) tissue from excision specimens, with the exception of one case which was a needle core biopsy specimen. This study was conducted under an institutional review board-approved protocol.

Immunohistochemical staining

Immunohistochemical staining for p40, p63, lamC2 and kalinin B1 was performed on selected blocks from each case. Metaplastic carcinomas and phyllodes tumours were also stained with selected cytokeratin antibodies. Immunohistochemical staining was accomplished using the Bond III Autostainer (Leica Microsystems, Illinois, USA) (table 1) on whole tissue sections. FFPE tissue sections were first baked and deparaffinised. Heat or enzymatic-induced antigen retrieval was performed. Sections were then subjected to sequential incubation with the endogenous peroxidase block, primary antibody, postprimary (equivalent to secondary antibody), polymer (equivalent to tertiary antibody), diaminobenzidine and haematoxylin. Finally, sections were dehydrated in 100% ethanol and mounted in Cytoseal XYL (Richard-Allan Scientific, Kalamazoo, Michigan, USA). Appropriate controls were included. For p40, p63 and laminin 332 immunostains, normal breast glandular tissue (myoepithelial cells) served as an internal positive control.

Table 1

Antibodies, retrieval methods, dilutions and sources of immunohistochemical stains used in this study

The intensity of staining (none, weak, moderate and strong) of neoplastic cells and the proportion of neoplastic cells showing staining were recorded. Staining was considered positive if ≥1% of cells showed reactivity. The staining distribution was categorised as focal (<25% of cells staining), regional (25%–75% cells) or diffuse (>75% cells).

Results

Clinical characteristics of studied patients

Thirty-six patients with metaplastic carcinoma were women ranging in age from 19 to 98 years (median, 70 years). All presented with a unifocal mass (right, 22; left, 14). Tumours ranged in size from 0.7 to 10 cm (mean, 3.18 cm). Axillary lymph nodes were sampled at the time of surgery for nine of 36 (25%) patients. One of these patients was found to have one of seven lymph nodes involved by macrometastatic metaplastic carcinoma.

Patients with phyllodes tumours (n=20) were women ranging in age from 19 to 93 years (median, 41 years). All tumours were unifocal and ranged in size from 1.5 to 13.5 cm (mean, 3.98 cm). The mean size for borderline phyllodes tumours was 4.17 cm and was 3.17 cm for malignant phyllodes tumours.

Patients with non-mammary sarcomas (n=20) included 10 men and 10 women ranging in age from 13 to 89 years (median, 64.5 years). Non-mammary sarcomas included undifferentiated pleomorphic sarcomas with spindle cell growth (n=9), dedifferentiated liposarcoma (n=4), leiomyosarcoma (n=2), synovial sarcoma (n=3), endometrial stromal sarcoma (n=1) and low-grade fibromyxoid sarcoma (n=1). Tumours ranged in size from 2.5 to 23 cm (mean, 9.9 cm). Primary mammary sarcomas occurred in women aged 23, 65, and 77 years. Two tumours were spindle cell sarcomas, not otherwise specified (NOS) and measured 3.5 and 7 cm. The remaining tumour was a primary high-grade angiosarcoma that measured 4 cm. None of the sarcomas, mammary or non-mammary, occurred in patients who had previously received radiation therapy.

Histological features of metaplastic carcinomas

Spindle cell differentiation was observed in all 36 studied cases. Squamous differentiation was seen, at least focally, within eight of 36 (22.2%) tumours. Heterologous elements were identified in three tumours (two chondroid, one liposarcoma). The nuclear grade of spindle cells was categorised as intermediate in 31 (86.1%) and high in 5 (13.9%) tumours. Three tumours (8.3%) with intermediate nuclear grade also showed low-grade ‘fibromatosis-like’ features within the tumour.18 Invasive ductal carcinoma of no special type was a component of the tumour in two cases and constituted <20% of the tumour in each case. In four other cases, DCIS coexisted with the metaplastic spindle cell carcinoma. Lymphovascular invasion was not identified in any of the studied cases.

Immunohistochemical staining results

p63 and p40

Nuclear p63 expression was seen in 33 of 36 (91.7%) metaplastic carcinomas. Of these, p40 staining was seen in 21 tumours (64%; 58.3% overall; figure 1, table 2). The remaining 12 were only p63 positive. All tumours negative for p63 expression were also negative for p40. The distribution and intensity of staining for p63 and p40 were variable among tumours. For p63, diffuse staining was seen in more than half of cases and staining intensity was typically moderate to strong. In contrast, p40 staining was weaker in intensity (strong 4, moderate 13 and weak 4) with no characteristic staining distribution (equally diffuse or focal). In tumours showing squamous differentiation, the squamous component showed diffuse staining with both p63 and p40 (figure 2). None of 20 phyllodes tumours showed stromal expression of p63 or p40. Among sarcomas, one of 23 (4.3%) sarcomas (high-grade pleomorphic sarcoma of retroperitoneum) showed focal weak staining with p63. None of the three mammary sarcomas showed p63 staining and no sarcomas, regardless of site, showed staining with p40. The sensitivity and specificity of p63 for establishing a diagnosis of metaplastic carcinoma versus other malignant spindle cell tumours studied were 91.7% and 97.7%, respectively. The sensitivity and specificity of p40 were 58.3% and 100%, respectively.

Table 2

p63 and p40 immunohistochemical staining results and distribution of staining within studied tumours

Figure 1

(A) Metaplastic carcinoma showing spindle cells with storiform and fascicular growth (H&E, ×200). Diffuse staining with p63 (B) and p40 (C) was seen in the tumour with strong and moderate staining intensities, respectively (B and C, ×200). (D) A borderline phyllodes tumour showed no staining with p40. p40 highlights myoepithelial cells in the glandular component of the tumour (×200). In the same tumour as A–C, strong and diffuse cytoplasmic and membranous staining was observed with kalinin B1 (E) and LamC2 (E and F, ×200). Inset E: Benign duct showing kalinin B1 staining in myoepithelial cells and weak staining in periductal stroma (×400). Higher power of cells seen in F in inset (×400).

Figure 2

(A) Metaplastic carcinoma showing spindle and squamous differentiation (H&E, ×200). (B) p40 highlights cells with squamous differentiation, while (C) p63 stains squamous and spindle cell components of this tumour (B and C, ×200).

Laminin 332

LamC2 expression was seen in 28 of 36 (77.8%) metaplastic carcinomas (figure 1). Similarly, kalinin B1 staining was observed in 26 of 36 (72.2%) metaplastic carcinomas. The distribution and intensity of staining for lamC2 and kalinin B1 were variable among tumours (table 3). Most tumours showed moderate to strong staining intensity with each LN332 antibody (lamC2, 89.2%; kalinin B1, 65.4%), whereas weak staining was seen in the remaining tumours. Twenty-one (58.3%) tumours showed staining with both laminin 332 chains, whereas in 12 (33.3%) tumours, positive staining with only one antibody was seen. Of these, seven showed staining with lamC2 and were negative with kalinin B1 and five were positive for kalinin B1, but were negative for lamC2.

Table 3

Laminin 332 and cytokeratin immunohistochemical staining results and distribution of staining within studied tumours

LamC2 expression was seen in two of 20 (10%) phyllodes tumours. One was a borderline phyllodes tumour that showed regional staining with moderate intensity. The second was a malignant phyllodes tumour that showed focal weak staining. Kalinin B1 staining was observed in two of 20 (10%) phyllodes tumours, each of which was borderline phyllodes tumour. One tumour, which also showed moderate regional staining with lamC2, showed a similar staining pattern with kalinin B1. The second showed focal moderate staining, but was negative for lamC2. None of the 23 sarcomas showed expression of lamC2. Kalinin B1 staining was observed in 17 of 23 (73.9%) sarcomas, including two of three primary sarcomas of the breast (angiosarcoma—focal staining with moderate intensity, primary spindle cell sarcoma—regional staining with moderate intensity). For the remaining sarcomas, there was a variable distribution and intensity of staining with kalinin B1. The sensitivity and specificity of lamC2 for diagnosing metaplastic carcinoma were 77.8% and 95.4%, respectively. For kalinin B1, the sensitivity and specificity were 72.2% and 55.8%, respectively.

Cytokeratins

Cytokeratin expression was seen in 32 of 36 (88.9%) metaplastic carcinomas (CKAE1/AE3: 13/33 (39.4%); CK7: 13/33 (39.4%); 34βE12: 24/32 (75%); CK5: 24/34 (70.6%); table 3). Twenty-two of 36 (61.1%) metaplastic carcinomas showed diffuse staining with at least one cytokeratin. 34βE12 and CK5 most often showed diffuse staining (20 and 17 tumours, respectively). The metaplastic carcinomas that were negative for cytokeratin stains showed positive staining with p63. Due to limited material available for study, only 34βE12 and CK7 were evaluated in our cohort of phyllodes tumours. We found that all 20 cases were negative for both 34βE12 and CK7.

Comparison of LN332 and p40 with routinely used markers in metaplastic carcinoma

All metaplastic carcinomas that showed staining for LN332 or p40 also showed positive staining with p63 and/or cytokeratins (table 4). All (21/21) metaplastic carcinomas showing p40 staining showed positive staining for both p63 and cytokeratins. For lamC2 and kalinin B1, most positive cases (85% and 76.9%, respectively) also showed p63 and cytokeratin staining.

Table 4

p40 and LN332 immunohistochemical expression in spindle cell metaplastic carcinomas compared with p63 and cytokeratin expression

Discussion

In most instances, a needle core biopsy is the first tissue sampling of a patient's breast mass. Some of these biopsies reveal a tumour consisting of mostly spindle cells with atypical or frankly malignant cytomorphology. When such a lesion is encountered, pathologists first need to exclude spindle cell metaplastic carcinoma as this tends to be the most common among spindle cell tumours arising in the breast19 and more importantly, such tumours require management and treatment options not offered for patients with other spindle cell tumours (sarcomas or phyllodes tumours).

Due to significant morphological overlap among spindle cell tumours, pathologists rely heavily on immunohistochemical markers p63 and cytokeratins to investigate the possibility of spindle cell metaplastic carcinoma. p63, a member of the p53 family normally expressed in myoepithelial cells, has been shown to be immunoreactive in most metaplastic carcinomas and is a highly specific marker.6 ,11 ,13 ,20 p40 is the ΔN isoform of p63 which is more specific for carcinomas with squamous differentiation when compared with p63.10 In the current study, we strove to determine whether p40 had similar utility in diagnosing metaplastic carcinoma as p63. Similar to established findings in the literature, we found p63 to be a highly sensitive and specific marker for metaplastic carcinoma. However, our data also indicate that p40, while also highly specific (100% in our study), lacked the sensitivity of p63 (58.3% vs 91.7%).

Two other studies have investigated p40 expression in mammary metaplastic carcinoma. In a study using tissue microarrays as a surrogate for core biopsy tissue sample size, Cimino-Mathews et al reported p40 immunoreactivity in six of 13 (46%; spindle cell) metaplastic carcinomas and two of 14 (14%) malignant phyllodes tumours, whereas p63 staining was observed in 62% of metaplastic carcinomas and 57% of malignant phyllodes tumours. Importantly, most malignant phyllodes tumours in their analysis showed only focal (1%–5% cells) labelling with p63 and p40, and none showed >30% of cells staining. We found this observation to be intriguing as in our experience, none of 20 phyllodes tumours studied in whole tissue sections showed any immunoreactivity for p63 or p40. It would be of interest to know whether these staining patterns would be similar in whole tissue sections of these tumours. Similar to our findings, the authors found p40 to be a specific marker for metaplastic carcinoma, but it lacked the sensitivity of p63. In another study examining expression of p40 and p63 in various benign and malignant breast lesions, p63 staining was seen in nine of 36 (25%) metaplastic carcinomas. Among spindle cell metaplastic carcinomas in their study, 0 of five cases showed p63 staining.12 In this study, two different rabbit polyclonal p40 antibodies were used and showed staining in 27.8% and 72.2% of metaplastic carcinomas overall. In the spindle cell group, three of five tumours showed staining with one antibody while none of the five tumours showed staining with the second antibody. The low frequency of p63 staining was attributed to the use of tissue microarrays and potential heterogeneity of p63 expression within the tumours. While the latter reason is theoretically feasible, we found over half of our studied cases of spindle cell metaplastic carcinoma to diffusely stain with p63. Of note, a mouse monoclonal p40 antibody was used in our study and Cimino-Mathews11 used a rabbit polyclonal p40 antibody, each of which offered similar staining results in metaplastic carcinomas.

Immunohistochemical expression of LN332 has been documented in a high proportion of triple-negative breast carcinomas13 ,15 and its expression in metaplastic carcinomas has been reported in one previous study.14 Carpenter et al reported LN332 to be expressed in 24 of 25 (96%) metaplastic carcinomas overall and in nine of 10 (90%) metaplastic carcinomas with spindle cell morphology. In contrast, p63 was positive in eight of 10 (80%) and CK5/6 in six of 10 (60%) spindle cell metaplastic carcinomas. Staining for LN332 was observed in none of two undifferentiated sarcomas of the breast, one malignant phyllodes tumour and one mammary angiosarcoma. However, weak p63 staining was seen in a ‘small percentage of cells’ in one of the two sarcomas. In our study, we also found both LN332 antibodies to be expressed in a high proportion of metaplastic carcinomas (lamC2, 77.8%; kalinin B1, 72.2%). However, these markers were less specific than p63, p40 and cytokeratins, showing staining in 10% of phyllodes tumours with each antibody. Furthermore, 74% of sarcomas showed staining with kalinin B1, including two of three primary sarcomas of the breast.

Finally, the expression of cytokeratins in metaplastic carcinoma in our study was consistent with what has been reported.3 ,13 ,21 We found high molecular weight cytokeratins 34βE12 and CK5 to be the most frequently expressed in metaplastic carcinomas, and these two antibodies most often showed a diffuse staining pattern in the tumours. We additionally studied cytokeratin expression (34βE12 and CK7) in phyllodes tumours due to the reported focal (1%–5% cells) positivity in some tumours in a recent study of 109 phyllodes tumours by Chia et al22 We found that no phyllodes tumours showed stromal expression of 34βE12 or CK7, each of which showed patchy expression in 21.8% and 28.3% of phyllodes tumours, respectively, in their study.22

A relative strength of our study is the opportunity to evaluate immunohistochemical staining patterns in whole tissue sections of excisions rather than needle core biopsy samples. We believe that the distribution of staining of immunostains is critically important in the scenario of limited diagnostic material (ie, needle core biopsy) where the diagnosis heavily relies on immunohistochemical results and misdiagnosis can have significant management consequences. For p63 and p40, most (85% and 62% of positive cases, respectively) showed >25% staining in metaplastic carcinomas and as such, this staining would likely be seen in smaller core biopsy samples. Cimino-Mathews et al reported that a subset of malignant phyllodes tumours shows focal p63 and p40 labelling in core biopsy samples and cautioned that such staining should not prompt a diagnosis of metaplastic carcinoma.11 While we agree that focal expression of these markers should be interpreted with caution, we did not have the same experience in our cohort of 20 phyllodes tumours all of which were negative for both markers.

One limitation of our study is that 20 of 23 sarcomas studied were from non-mammary sites. Some of these non-mammary sarcomas, however, such as undifferentiated pleomorphic sarcoma, leiomyosarcoma and liposarcoma, may rarely arise as primary tumours in the breast and/or axilla. We suspect that the immunohistochemical staining patterns of these markers in non-mammary sarcomas would be similar for sarcomas with similar morphology/differentiation arising in the breast, but this would need additional study. A potential bias of our study is that the inclusion of metaplastic carcinomas was based on p63 and/or cytokeratin staining and all tumours in the study showed staining with at least one of these markers. Using p63 and cytokeratin stains as the ‘gold standard’ for diagnosing metaplastic carcinoma does not allow us to identify metaplastic carcinomas that may be positive for LN332 or p40, while negative for the traditional markers.

We concluded from the data herein that for different reasons, p40 and LN332 did not demonstrate enough diagnostic value to replace or add to the panel of p63 and cytokeratins currently used to investigate the possible diagnosis of spindle cell metaplastic carcinoma. Akin to conclusions drawn from other investigators, we found that p40 was as specific as p63, but fell short in comparison when assessing sensitivity. Moreover, LN332 stains were sensitive for spindle cell metaplastic carcinoma but lacked specificity as each was immunoreactive in 10% of phyllodes tumours studied. Additionally, kalinin B1 showed staining in three quarters of sarcomas studied.

In summary, we aimed to provide more information on the diagnostic use of known (p63 and cytokeratins) and relatively novel (p40 and LN332) immunostains in the work-up of malignant spindle cell tumours. Although p40 appeared promising initially, we found that its diagnostic value in this clinical scenario to be less than that of p63. Therefore, we do not recommend substituting p63 with p40 or adding p40 as a first-line immunostain. LN332 was also disappointing in its inability to demonstrate high specificity for spindle metaplastic carcinomas and therefore should not be used in the needle core biopsy setting. Lastly, we confirmed p63 and cytokeratins, particularly 34βE12 or CK5, to be useful in diagnosing spindle cell metaplastic carcinomas. None of our 20 phyllodes tumours were positive for p63 and only one of 23 sarcomas (non-mammary) demonstrated focal, weak staining. A first-line immunopanel in the work-up of atypical or malignant spindle cell tumours encountered in the needle core biopsy should at a minimum include p63 and multiple cytokeratins such as pan-cytokeratin, 34βE12, CK7 and CK5.

Take home messages

  • p40, ΔN isoform of p63, is highly specific for metaplastic carcinoma, but lacked the sensitivity of p63 in our analysis.

  • Laminin 332, formerly laminin 5, is expressed in a high proportion of metaplastic carcinomas but lacks specificity, staining a significant number of phyllodes tumours and sarcomas.

  • Traditional markers used to diagnose metaplastic carcinoma, p63 and cytokeratins, particularly high molecular weight cytokeratins, were most specific and sensitive in our study.

  • We do not recommend the routine use of p40 or laminin 332 for establishing a diagnosis of metaplastic carcinoma.

References

Footnotes

  • DSR was a resident in the Department of Pathology and Laboratory Medicine at the time of this study.

  • Correction notice This article has been corrected since it was published Online First. The provenance and peer review statement has been amended.

  • Handling editor Cheok Soon Lee

  • Contributors All authors are justifiably credited with authorship. TMD: Conception, design, analysis and interpretation of data, drafting of the manuscript and final approval given; DSR: acquisition of data, analysis and interpretation of data and final approval given; Y-FL: immunohistochemical workup and staining, drafting of the manuscript and final approval given; SJS: conception, design, analysis and interpretation of data and final approval given.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.