Myoepithelium
To distinguish between benign and malignant breast glandular proliferations, an understanding of myoepithelial cells and the immunohistochemical markers used to identify them is important.
Invasive carcinomas lack surrounding myoepithelial cells, whereas the majority of benign lesions and in-situ carcinomas retain the myoepithelial cell layer, to varying degrees.
The first step in determination of whether a glandular proliferation within the breast is benign or malignant is made at a low magnification. Benign small glandular proliferations usually attempt to recapitulate the normal lobular architecture of the breast. The prime example of this is sclerosing adenosis. As its name implies, it represents a proliferation of glands (adenosis), which are compressed or sclerosed. At low magnification, even though the proliferated glands may appear to have an infiltrative architecture, they retain 'lobulocentricity', that is, they resemble a lobule (albeit a larger and more sclerosed one!) in terms of gland distribution.
Sometimes, it may not be easy to identify individual myoepithelial cells. This usually happens in sclerosing lesions e.g. sclerosing adenosis and radial sclerosing lesions. As ductal carcinoma in-situ (DCIS) distends the duct system, it may cause compression, attenuation and focal loss of peripheral myoepithelial cells. Invasive ductal carcinoma, on the other hand, completely lacks a myoepithelial layer.
Immunohistochemistry is useful to identify myoepithelial cells in these cases. The most commonly used myoepithelial markers include smooth muscle actin (SMA) , smooth muscle myosin heavy chain (SMMHC) , calponin , p63 and high molecular weight cytokeratin e.g. cytokeratin 5 and 14.
These markers must be interpreted in association with the findings on routine staining, as otherwise they can be misleading. They also vary in their specificity.
Smooth muscle actin is a very sensitive myoepithelial marker but lacks specificity. It is also expressed by myofibroblasts, which are commonly present surrounding invasive glands and ductal carcinoma in-situ. To avoid falling into the trap of misinterpreting myofibroblasts for myoepithelial cells, pay close attention to exactly where the antibody is being expressed, in relation to the epithelial cells. Myofibroblasts usually have a linear staining pattern, whereas myoepithelial cells often bulge towards the lumen, adopting a triangular profile and tend to interdigitate at the periphery of epithelial islands. If you see a predominantly linear staining pattern and reactivity in the surrounding stroma, beware of invasion. In this situation, use a more specific myoepithelial marker e.g. p63 or smooth muscle myosin. In the context of ductal carcinoma in-situ and invasive ductal carcinoma, smooth muscle actin may also stain proliferated blood vessels around the lesion, mimicking myoepithelial staining, if the blood vessels are in close proximity to the neoplastic epithelium. p63 and high molecular weight cytokeratin are usually reassuring in this situation.
Smooth-muscle myosin heavy chain, p63 and calponin are more specific markers of myoepithelial cells than smooth muscle actin. It is often useful to employ a panel of three or more of these myoepithelial stains in difficult cases, as they vary in their sensitivities. SMMHC and calponin are cytoplasmic stains, whereas p63 decorates the myoepithelial nucleus. Once again, it is important to correlate myoepithelial marker expression with the appearances on routine staining, as apparently specific myoepithelial markers may be expressed by some carcinomas e.g. up to 12% of invasive carcinomas have p63 positive cells. In this context, p63 is seen most often in high grade and metaplastic carcinomas. To avoid this pitfall, concentrate on the layer expressing p63. If these are true myoepithelial cells, they will stain at the periphery only and not through the tumour island. The nuclei expressing the marker will also resemble those of the malignant cells, if it is aberrant tumour cell expression which you are witnessing. Remember also that some types of breast cancer show myoepithelial differentiation. These tumours include adenoid cystic carcinoma, metaplastic carcinoma, low grade adenosquamous carcinoma, myoepithelial carcinoma and malignant adenomyoepithelioma.
Misinterpretation of positive staining is one problem and overlooking areas of negative microinvasive tumour, in the context of DCIS (when distracted by that which is positive) is another! Always pay close attention to the negatively stained areas (especially in the context of periductal inflammation) of a slide with DCIS.
Please remember that the myoepithelial layer surrounding DCIS may not always be complete. As the neoplastic intraductal population expands the duct system, gaps begin to appear in the surrounding myoepithelial layer. The significance of the gaps is not clear but in some cases they may facilitate the progression to invasive carcinoma. Use of a myoepithelial panel is important in this context as some markers e.g. p63 may be very inconspicuous. Always be aware of the appearances on routine staining and learn to interpret the myoepithelial stains in this context to avoid over diagnosis of invasive carcinoma. Very occasionally, rounded tumour nests (resembling DCIS) lack evidence of myoepithelial cells, despite evaluation of multiple markers. This finding should raise concern for nested invasive carcinoma, especially if there is no myoepithelial cell layer attenuation in the background DCIS.
Some benign lesions may also present difficulty with interpretation of myoepithelial markers. Up to 12% of radial sclerosing lesions uncomplicated by carcinoma, show prominent myoepithelial attenuation. If in this context you spot a central tubule without an apparent myoepithelial layer, compare it to those with a myoepithelial layer. If it looks the same as the obviously benign tubules and is present amongst them, it should be interpreted as benign.
Displaced epithelial cell clusters following core biopsy may also appear to lack a myoepithelial layer. This is common following needling of papillary lesions. The luminal epithelial cells may get displaced, separated from their myoepithelial supporting layer, simulating invasion. Beware! The cluster will be confined to the biopsy site if artifactual and (if the biopsied lesion was a benign papilloma) will resemble the pre-existing benign lesion.
Finally some rare benign lesions such as microglandular adenosis, by definition, lack a myoepithelial layer, but are surrounded by an intact basement membrane.