MATERIAL AND METHODS

Eight cases with a histologically confirmed diagnosis of pure DCIS and 15 with a histologically confirmed diagnosis of pure IDC of the breast were retrieved retrospectively from the pathology files of the Hospital São João, University of Porto, Porto, Portugal. The histological sections were reviewed by two of the authors and the diagnoses of pure DCIS and pure IDC were reconfirmed. In each case, smears obtained by previous FNABs were available. Only those cases with a Papanicolaou stained slide with cytological features consistent with a diagnosis of ductal carcinoma were included in our present study.

The histological and cytological samples were reviewed by two of the authors, and the most representative Papanicolaou stained cytology slide was chosen for the immunocytochemical analysis, as described previously.¹⁶ Briefly, the coverslips of Papanicolaou stained slides were removed. Heat induced antigen retrieval in a wet bath, using the Dako antigen retrieval solution was performed for 20 minutes in all samples. Immunocytochemistry was performed according to the streptavidin–biotin–peroxidase technique using the antibody 4A4 raised against p63 (1/200 dilution; Neomarkers, Freemont, California, USA), as described elsewhere.¹⁷ Only cells showing strong and distinctive nuclear immunoreactivity for p63 were considered positive. Cytoplasmic and membranous staining were considered non-specific. Positive (histological sections of a sclerosing papilloma with myoepithelial hyperplasia) and negative (omission of primary and secondary antibodies) controls were included in each slide run. All controls gave appropriate results.

In our present study, we specifically searched for p63 positive cells overlying malignant cell clusters and for p63 positive myoepithelial cells admixed with malignant cells. We compared their frequency in histologically confirmed DCIS and IDC. Statistical analysis was performed with Statview 5.0 software. Statistical differences between p63 expression in myoepithelial cells overlying malignant cell clusters and admixed with malignant cells, and the final histological diagnosis were calculated using Fisher's exact test. A test was considered significant only if the two sided p value was less than 0.05. We also calculated the sensitivity and specificity of the presence of p63 positive myoepithelial cells as a marker for DCIS. Sensitivity was defined as the ratio between: the number of DCIS cases with p63 positive myoepithelial cells (true positive; TP)/TP + the number of DCIS cases without p63 positive myoepithelial cells (false negative; FN). Specificity was defined as the ratio between: the number of IDC cases without p63 positive myoepithelial cells (true negative; TN)/TN + the number of IDC cases with p63 positive myoepithelial cells (false negative; FN). Both sensitivity and specificity were expressed in percentages.

RESULTS

Table 1 summarises the cytological and histopathological data of all cases. The review of the pure DCIS and IDC cases revealed that there was no invasive component in the DCIS cases and no in situ component in the IDC cases.

Cytological analysis of p63 stained slides showed a distinctive nuclear staining pattern in myoepithelial cells; a faint background was seen in three cases, although this did not impair the evaluation of the slides.

p63 positive myoepithelial cells overlying malignant cell clusters were seen in all DCIS cases (fig 1A) and in 60% (nine cases) of IDCs (fig 1B). Most importantly, the presence of myoepithelial cells was significantly more frequent in DCIS than in IDC cases (p = 0.0375). The sensitivity of p63 positive myoepithelial cells for the diagnosis of DCIS was 100% (8/(8 + 0) = 1 × 100 = 100%) and its specificity was 40% (6/(6 + 9) = 0.40 × 100 = 40%). When stratified by grade, p63 positive myoepithelial cells were seen in 100% of grade I, 40% of grade II, and 50% of grade III IDC cases.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

p63 immunocytochemistry highlighting myoepithelial cells overlying tridimensional malignant clusters. (A) Photomicrography showing a fine needle aspiration of a ductal carcinoma in situ with the typical distribution of p63 positive myoepithelial cells. (B) Similar findings in an invasive ductal carcinoma. Note the presence of intracytoplasmic lumens/vacuoles (arrows). Streptavidin–biotin–peroxidase/diaminobenzidine stain; original magnification: A, ×200; B, ×400.

A remarkable feature seen in our present study was the presence of scattered (< 5%) p63 positive neoplastic cells in eight cases (three cases of DCIS and five of IDC) (fig 2A). In two of these cases, up to 5% of the neoplastic cells also showed positivity for anti-p63 antibody applied to histological sections (fig 2B), whereas in six no p63 positive neoplastic cells were seen in the histological samples.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

p63 positive invasive ductal carcinoma not otherwise specified. (A) Note the presence of malignant cells with p63 positive nuclei. (B) Histological section of the same case showing a grade III invasive ductal carcinoma with scattered p63 positive cells. Both streptavidin–biotin–peroxidase/diaminobenzidine stain; original magnification, ×200.

DISCUSSION

Currently, the treatment of breast carcinomas is tailored according to a constellation of clinical and pathological findings. The presence or absence of invasion is one of the major characteristics that determine whether or not a patient will be submitted to axillary dissection or even to neoadjuvant chemotherapy.³

Several groups have searched for specific cytological criteria that may guide pathologists in determining the presence or absence of invasion.^2,3,6–10 Several criteria have been described, including tumour cell dissociation, the presence of tubular/glandular structures, cytoplasmic lumens/cytoplasmic vacuoles, fibroblastic proliferation, infiltration of fat/mammary stroma by neoplastic cells, and the presence of myoepithelial cells overlying malignant cell clusters.^2,3,6–10 McKee and colleagues² emphasised that to diagnose invasion, pathologists need look for the presence of stromal fragments or fat tissue invaded by neoplastic cells, because these findings are seen only in IDC and not in DCIS; in the absence of these criteria, the presence of cytoplasmic lumens/vacuoles and tubule formation suggest invasion.² It should be noted that these criteria are not seen in all cases, and even in the series of McKee et al,² only 72% of IDCs displayed definitive stromal or fat tissue invasion; moreover, in our opinion, the identification of stroma/fat tissue invasion in cytological preparations is very far from straightforward.

In contrast, the identification of certain cytological features may suggest the absence of invasion, pointing towards a diagnosis of DCIS. In the series of McKee et al,² the most important criterion to suggest DCIS was the presence of myoepithelial cells overlying tumour cell clusters. However, myoepithelial cells are frequently misidentified, because in FNAB smears some of their usual cytomorphological features are distorted and they may be confounded with apoptotic cells, stromal cells, epithelioid macrophages, or even cancer cells with a spindle cell phenotype.^2,6,7,10,11

“Most importantly, p63 does not crossreact with secretory cells, breast stromal cells, smooth muscle cells, endothelial cells, pericytes, and neural cells”

In this study we evaluated p63 expression in FNABs from eight cases of DCIS and 15 cases of IDC. Based upon our results and on those previously reported,^2,7,9 myoepithelial cells overlying malignant cell clusters were significantly more frequent in DCIS than in IDC cytological samples (100% of DCIS cases v 60% of IDC cases; p = 0.0375); however, in contrast to published results,^2,7,9 we found these cells in 60% of IDC samples (specificity, 40%). This discrepancy is most probably related to the high sensitivity of p63 for myoepithelial cells, but we cannot exclude the possibility that it could be related to the small number of samples evaluated in our study. Another possible explanation could be the presence of an associated DCIS in those cases diagnosed as IDC, although even after a review of the histological samples of all IDC cases included, no DCIS component was found. However, we cannot rule out the possibility that the neoplastic areas selected at gross examination, which were submitted to histological analysis, and reviewed herein failed to identify the foci of DCIS. It should be noted that because many IDCs are associated with DCIS, the specificity of p63 positive myoepithelial cells overlying malignant cell clusters may putatively be lower in diagnostic cytopathology practice than that obtained in our present study.

It has been shown that p63 is a reliable myoepithelial cell marker^12–15 and that it overcomes the problem of myoepithelial cell cytoplasm fragmentation during fine needle aspiration, because it is expressed in the nuclei of myoepithelial cells in contrast to all of the other myoepithelial markers,^11–15,18 which decorate either the cytoplasm or the cytoplasm and nucleus of myoepithelial cells. Most importantly, p63 does not crossreact with secretory cells, breast stromal cells, smooth muscle cells, endothelial cells, pericytes, and neural cells.^12,13,15 However, we must emphasise that scattered p63 positive neoplastic cells might be seen in up to 34.8% of cases. According to our previous result,¹⁴ these might reflect a partial myoepithelial/basal cell differentiation of some breast carcinomas.^12,15 It is noteworthy that these p63 positive cells did not pose a problem in the final diagnostic interpretation.

In conclusion, the presence of myoepithelial cells overlying malignant cell clusters is a very frequent finding in DCIS,² but it is also found in up to 60% of IDC samples. We stress that the identification of myoepithelial cells is easy using p63 antibodies in Papanicolaou stained slides. Our results indicate that the presence of myoepithelial cells either overlying malignant cell clusters or admixed with neoplastic cells is not a useful criterion to rule out invasion (or suggest the presence of DCIS) in diagnostic cytopathology practice.