Main

The clinical behavior of phyllodes tumors of the breast is difficult to determine.1, 2, 3, 4, 5 Histologic grade,6, 7, 8 stromal overgrowth,9, 10 tumor necrosis and heterologous stromal elements,11 as well as a combination of morphologic parameters,12 have been traditionally used as predictors of clinical outcome. Nevertheless, specific parameters that can independently define recurrent likelihood are not universally agreed upon.2 A substantial number of reports have concluded that it is the adequacy of surgical margins that is most important,5, 6, 12, 13, 14 and that histologic features have an inconsistent predictive role.

We recently described the pathologic parameters of a large series of phyllodes tumors,15 and found that grade, as well as the individual histologic components of stromal atypia, stromal hypercellularity and permeative microscopic borders were correlated with recurrent disease. On multivariate analysis however, it was margin status, as well as pseudoangiomatous stromal hyperplasia that were independently predictive of recurrence.

Beyond morphology, investigators have studied the role of biological markers in phyllodes tumors, with p53 perhaps being the most widely evaluated.1, 16, 17, 18, 19 p53 expression has been described as being associated with malignant histological features, but does not appear to predict recurrent likelihood.1, 17 Recently, c-kit (CD117) expression was also observed in malignant phyllodes tumors, although its impact on recurrence was not addressed.20, 21

In this study, we employed p53 and CD117 immunohistochemistry on our previously described phyllodes tumors using the tissue microarray technique, and correlate the results with histological parameters and clinical outcome.

Materials and methods

Patients and Tumors

These cases of breast phyllodes tumors were diagnosed between January 1992 and December 2002 at the Department of Pathology, Singapore General Hospital. Patient details and tumor laterality were determined from accession forms, while tumor size was obtained from the surgical pathology reports. Histologic review and classification into benign, borderline and malignant categories, as previously defined,15 relied on a constellation of findings, namely degree of stromal hypercellularity, cytologic atypia, mitotic activity, stromal overgrowth and nature of the borders (circumscribed vs permeative).

Patient follow-up was obtained from casenotes and the Singapore Cancer Registry, National Disease Registries Office.

Tissue Microarray Construction

The arrays were constructed with the 2 mm punch on the Beecher arrayer.22 The array layout in grid format was designed using Microsoft Excel. Hematoxylin and eosin (H&E)-stained sections of phyllodes tumors were reviewed and the area of interest marked out on the slide. Using a marker pen, the corresponding region was circled on the archival ‘donor’ paraffin block. The samples were then arrayed on to a ‘recipient’ blank block. The 2 mm punch was selected as it allowed easy manipulation during punching; with the larger individual sample size offering more lesional tissue for assessment. Difficulties encountered in the construction of tissue microarrays were overcome as earlier described.22

A total of 59 phyllodes tumors were subjected to more than one punch per case: 52 cases with two microarray cores; four cases with three microarray cores; and three cases with more than three microarray cores (two with four cores and one with six cores) each.

Immunohistochemistry

Immunohistochemical staining was assessed in both epithelial and stromal components. The proportion of cells stained, as well as the intensity of staining, was evaluated. For staining intensity, 0, 1+, 2+ and 3+ defined nil, weak, moderate and strong reactivity, respectively. p53 staining was assessed in the cell nuclei, while CD117 was evaluated in the cytoplasm and cytoplasmic membrane. Any unequivocal staining, regardless of intensity and proportion of cells stained, was regarded as positive. For cases with more than one microarray constructed per case, staining was scored separately and the results compared. The finding that yielded the highest scores was used in the analysis. The degree of agreement of stromal immunostaining in these cases was ascertained.

Sections (4 μm) were cut from the tissue microarray blocks and fished onto coated slides (POLYSINE, Menzel-glaser) in a similar orientation to facilitate faster evaluation. The slides were then baked in the oven overnight at 55°C to enhance adhesion of the sections to the slides. Deparaffinization in xylene and through graded alcohol followed. For the p53 antibody, sections were subjected to pressurized superheating in Milestone T/T Mega for 4 min at 120°C in 0.01 M citrate buffer, then cooled for 4 min under running water, then stained with the p53 antibody (D07, Dako M7001, 1:70 dilution) using the Envision ChemMate Kit on the Dako Autostainer. For CD117, sections underwent pressurized superheating in Milestone T/T Mega, for 4 min at 115°C, in Dako S3307 pH9 buffer and then were cooled for 4 min under running water. The sections were then stained with the CD117 antibody (c-kit, Dako A4502, 1:250 dilution) using the Envision ChemMate Kit on the Dako Autostainer. After that, the sections were counterstained with hematoxylin, dehydrated and mounted in depex.

A breast carcinoma known to be positive for p53, and a CD117 positive case of gastrointestinal stromal tumor, were used as positive controls; with mast cells also serving as an internal positive for CD117. Negative controls were accomplished by omitting the primary antibody with each batch run.

Statistical Analysis

Statistical analysis was carried out using the software SPSS for windows release 11.5. Patient age, histologic parameters and recurrences were correlated with p53 and CD117 immunostaining results using the χ2 test. The Mann–Whitney U-test was used to compare means between variables. A P-value of <0.05 was considered a significant result.

Results

Patients and Tumors

A total of 335 women were diagnosed with phyllodes tumors during the study period, with ages ranging from 16 to 69 years (mean 41 years, median 42 years). Details are as previously documented.15 Briefly, tumor size for the series ranged from 0.9 to 25 cm (mean 5.4 cm, median 4 cm). Histologic classification revealed 250 (75%) benign, 54 (16%) borderline and 31 (9%) malignant lesions. Other morphologic parameters are shown in Tables 1 and 2.

Table 1 Clinicopathologic features of phyllodes tumors correlated against p53 stromal immunostaining
Full size table
Table 2 Clinicopathologic features of phyllodes tumors correlated against CD117 stromal immunostaining
Full size table

Tissue Microarrays and Immunohistochemistry

Of 335 cases of phyllodes tumors, the maximum number of microarrays that could be scored was 289. In some cases, the microarray cores were incomplete, with donor cores that were too thin or not able to sit snugly within the recipient blocks, resulting in either loss of tissue or incomplete sections when cutting.

Results of immunohistochemical staining of p53 and CD117 on the tissue microarrays are detailed in Tables 3 and 4, respectively.

Table 3 Association of p53 immunostaining with grade of phyllodes tumors
Full size table
Table 4 Association of CD117 immunostaining with grade of phyllodes tumors
Full size table

For 59 cases that were subjected to duplicate, triplicate and more microarray cores, 50 (85%) and 54 (92%) revealed agreement of p53 and CD117 immunohistochemical findings, respectively, when positive vs negative stromal staining was compared.

p53

The number of microarrays suitable for evaluation of p53 immunohistochemical staining on epithelial, myoepithelial and stromal components was 278, 251 and 289, respectively. The quantum of luminal epithelium available for analysis in each microarray was variable, sometimes demonstrating only a short stretch; and because myoepithelial cells were not always discerned in tandem with luminal epithelium, the number of microarrays in which myoepithelial staining could be documented was fewer. Table 3 shows the distribution of p53 immunohistochemical results, including staining intensities, in histologically benign, borderline and malignant phyllodes tumors. The proportion of epithelial cells decorated with p53 ranged from 1 to 10% (mean 4%, median 5%) (Figure 1a). Myoepithelial cell staining ranged from 1 to 30% (mean 7%, median 5%), while stromal cells stained from 1 to 80% (mean 15%, median 5%) (Figure 1b and c). Borderline and malignant phyllodes tumors were more likely to reveal stromal staining of moderate than weak intensity. Statistically significant correlations between p53 stromal immunostaining and morphologic parameters are depicted in Table 1. No correlation of stromal positivity with patient age, tumor size or margin status was observed. There was also no association between p53 immunoreactivity of either epithelium or myoepithelium with any histologic features. Stromal p53 positivity was correlated with luminal epithelial and myoepithelial p53 immunoexpression (P<0.001).

Figure 1
figure 1

(a) p53 immunostaining of the epithelial component in a phyllodes tumor, showing focal and patchy nuclear positivity. (b) Stromal cells of a phyllodes tumor in a tissue microarray revealing strong nuclear staining. (c) Higher magnification of p53 immunostaining within stromal cells of a phyllodes tumor.

Full size image

CD117

For CD117, 267 and 273 microarrays could be assessed for epithelial and stromal staining characteristics (Figure 2a and b). No reactivity was observed in myoepithelial cells. Like for p53, there were microarrays in which epithelium was absent, thereby resulting in more cases disclosing results for CD117 immunostaining of stromal cells. Table 4 shows the distribution of CD117 immunohistochemical findings, including the staining intensities in histologically benign, borderline and malignant lesions. Unusually, CD117 stromal staining in benign phyllodes tumors tended to be of moderate than weak intensity, when compared against borderline and malignant ones. Statistically significant correlations between CD117 stromal staining and morphologic parameters are depicted in Table 2. There was no association of CD117 stromal staining with margin status, nor epithelial CD117 reactivity with histologic features.

Figure 2
figure 2

(a) CD117 decorates the epithelium of a phyllodes tumor, with cytoplasmic and cytoplasmic membrane staining. (b) CD117 cytoplasmic positivity in stromal cells of a phyllodes tumor.

Full size image

Recurrences

A total of 43 (13%) women suffered 57 recurrences during the follow-up period (mean and median follow-up duration of 30.3 and 20.4 months, respectively). The relationship between clinicopathologic parameters and onset of first local recurrence has been previously discussed.15 Of note, there was no statistically significant association of p53 immunostaining (epithelium, myoepithelium or stroma) with recurrent disease. On the other hand, CD117 stromal immunopositivity was correlated with recurrence (P=0.001), with cases showing stromal reactivity for CD117 being more likely to recur than those that were negative (Table 5).

Table 5 Correlation between CD117 stromal immunohistochemical staining and recurrence of phyllodes tumors
Full size table

Discussion

Biological markers have often been used as adjuncts to morphology in predicting behavior of tumors. p53 is a tumor suppressor gene that is widely studied in neoplastic processes. Located on chromosome 17p13.1, it encodes for a 53 kDa nuclear phosphoprotein that is expressed in all normal cells at low levels.23 The wild-type (normal) p53 gene is involved in cell cycle regulation as well as apoptosis. As the half-life of wild-type p53 is short, immunohistochemistry is believed to highlight expression of mutant p53 protein that is more stable with a longer half-life. In breast phyllodes tumors, the role of p53 has been fairly well investigated,1, 16, 17, 18, 19 with some authors suggesting a possible prognostic function,16, 18, 19 while its predictive utility has not been validated by others.1, 17 It is stated that p53 immunohistochemical expression in stromal cells can be relied upon as an adjunctive tool in the diagnosis of malignancy in phyllodes tumors.24, 25 In our study, p53 staining was correlated with grade, with contributions from stromal hypercellularity and overgrowth, but did not reveal an association with recurrent disease, corroborating the conclusions of other authors, and supporting its potential use as a confirmatory marker of malignancy in these tumors.

Interestingly, p53 immunohistochemical positivity was also noted in a proportion of luminal epithelial and myoepithelial cells in our series. Studies of p53 expression in phyllodes tumors have focused primarily on stromal cells,14, 16, 17, 18, 25 with observation of perithelial accentuation of stromal cell staining.14, 18 Feakins et al1 mentioned that the epithelial component in their cases usually stained negatively, while in another two reports, p53 reactivity was described in the breast epithelium in conjunction with stromal staining,26, 27 and it was postulated that dual expression of p53 protein in both the epithelium and stroma reflected the existence of an interaction between these components,26 akin to the conclusion of Sawyer et al,28 who believed that both epithelium and stroma of phyllodes tumors were participants in the neoplastic process. Our finding of an association between p53 stromal staining with that in luminal and myoepithelium lends credence to this purported relationship, although the presence of myoepithelial staining and its possible role in this process is unclear.

CD117, also known as c-kit, is a membrane-bound tyrosine kinase receptor. Dimerization and autophosphorylation of CD117 is known to inhibit apoptosis via the phosphatidylinositol-3 kinase/Akt system and potentiate cell proliferation via the Ras/MAP kinase pathway and JAK/STA signalling.29 As CD117 overexpression is characteristically observed in gastrointestinal stromal tumors,30 it has served as a therapeutic target by drugs (Glivec) used to manage patients with these tumors. More recently, CD117 has also been found in the stromal cells of malignant phyllodes tumors, and it was postulated that its overexpression may be instrumental in the growth of these tumors.20 Although gain-of-function mutations of CD117 have been reported in gastrointestinal stromal tumors, mast cell neoplasms and germ cell tumors,29 no activating mutations of CD117 have been described in breast phyllodes tumors.20, 31

In our study, CD117 protein expression within stromal cells was significantly associated with myriad morphologic parameters, including grade and recurrent disease. It can be argued that the strength of these correlations is diminished by the fact that only 17 cases were CD117 stromal positive on immunohistochemistry in our series. Nevertheless, its association with grade has been corroborated by other authors.20, 21 There was also a correlation of CD117 with p53 stromal staining, which suggests the possibility of these two proteins cooperating in potentiating cell cycle progression. After all, mutant forms of the p53 protein are known to interfere with cell growth suppressor effects of wild-type p53 and CD117 is associated with cell growth. If the role of CD117 in malignancy and recurrence of phyllodes tumors is validated, its expression can be advantageously harnessed for therapeutic purposes.

While studies on p53 and c-kit have predominantly dealt with protein expression using immunohistochemistry, it will be useful to validate these findings at the molecular level in order to determine specific genetic abnormalities in the responsible genes. Sequencing of the p53 gene carried out by various authors has discovered mostly mutations,16, 24 although one study disclosed a wild-type gene sequence in a malignant phyllodes tumor.32 Proliferative activity using Mib-1 and S-phase fraction,19, 27 microvessel density,25 CD34 and factor XIIIa stromal positivity33 are also believed to be potentially helpful in predicting phyllodes tumor behavior, although their clinical relevance remains to be further ratified.

The use of tissue microarrays has revolutionized translational research by allowing the rapid screening of numerous tumor samples simultaneously to validate candidate genes as potential prognostic markers.22 In this study, tissue microarray technology was used to evaluate the immunohistochemical expression of p53 and CD117 in a large series of breast phyllodes tumors. If standard histological sections were used, the number of sections requiring assessment would be much greater, incurring not only substantially more time, but also additional cost of antibodies and reagents because of the larger sections and greater number of slides. Previous reports reviewing the accuracy and reliability of tissue microarray vs standard sections on breast cancers have concluded that results derived from tissue microarrays are representative.34, 35 In this current study, when cases constructed in duplicate, triplicate and more microarray cores were compared, 85 and 92% of cases revealed agreement of p53 and CD117 immunohistochemical findings, respectively, when positive vs negative immunostaining of stromal cells was evaluated. We compared only stromal immunostaining as this was the cellular component under main scrutiny in these tumors. The lesser degree of immunostaining concordance for stromal p53 may be attributable to tumor heterogeneity, possibly linked to tumor size, suggesting that for phyllodes tumors that achieve relatively large dimensions, a recommendation to increase the number of punches/arrays commensurate with the lesional size may be warranted. The fact that p53 stromal staining displayed less concordance among the duplicate/triplicate/more cores than CD117 may also indicate that p53 immunoexpression tends to be more patchy and variable in phyllodes tumor stroma. An additional advantage of utilizing the tissue microarray method of evaluating this series of phyllodes tumors for p53 and CD117 protein expression is that it will allow expeditious selection of cases for further molecular studies.

In conclusion, p53 and CD117 protein expression by stromal cells in phyllodes tumors is correlated with histologic parameters such as grade, implying a possible role of their being used as adjunctive markers of malignancy in these tumors. The association of p53 immunostaining of the epithelium, myoepithelium and stroma affirms a link between these cellular compartments, a relationship that awaits further clarification. Although the number of CD117 stroma-positive cases is small, its expression appears to be associated with recurrent likelihood, and this needs to be investigated. The tissue microarray technique can be effectively applied to large numbers of phyllodes tumors to study other biological markers that may be of potential predictive utility.