A Subset of Malignant Phyllodes Tumors Express p63 and p40

Abstract

Breast phyllodes tumors are rare fibroepithelial neoplasms of variable grade, and one key differential of malignant phyllodes on core biopsy is sarcomatoid carcinoma. p63 is reported to be sensitive and specific for sarcomatoid carcinoma, with rare expression in phyllodes in limited series. The p63 deltaNp63 isoform, p40, is postulated to be more specific for squamous differentiation but has not previously been evaluated in breast phyllodes or sarcomatoid carcinoma. Tissue microarrays containing 34 unambiguous phyllodes tumors (10 benign, 10 borderline, 14 malignant), 13 sarcomatoid carcinomas, and 10 fibroadenomas were labeled by immunohistochemistry for p63, p40, CD34, and cytokeratins AE1/AE3, 34betaE12, and CK8/18. No borderline phyllodes tumor, benign phyllodes tumor, or fibroadenoma labeled with p63, p40, or cytokeratin. However, p63 labeled 57% malignant phyllodes tumors and 62% sarcomatoid carcinomas, and p40 labeled 29% malignant phyllodes (focal) and 46% sarcomatoid carcinomas. Among established markers, cytokeratins labeled 21% malignant phyllodes tumors (focal) and 100% sarcomatoid carcinomas. CD34 labeled 57% malignant phyllodes tumors and no sarcomatoid carcinomas. Focal p63, p40, and cytokeratin labeling can be seen in malignant phyllodes tumors but not in lowergrade fibroepithelial lesions, and immunoreactivity with these markers alone is not diagnostic of sarcomatoid carcinoma on core needle biopsy. In the differential diagnosis of malignant phyllodes, p40 is a more specific but less sensitive marker of sarcomatoid carcinoma than p63. These results are consistent with the sarcoma literature in which p63 labeling has been increasingly reported and suggest caution in classifying malignant spindle cell tumors of the breast on core biopsy.

Phyllodes tumors (PTs) are rare fibroepithelial breast neoplasms of variable grade, including benign phyllodes (BP), borderline phyllodes (BLP), and malignant phyllodes (MP) tumors, all of which are typically treated by complete wide excision.^1,2 One key differential diagnosis of MP tumors on core needle biopsy is sarcomatoid (metaplastic) carcinoma (SC), which is typically treated by complete excision with sentinel node sampling but can alternatively be treated by neoadjuvant chemotherapy.³ Nuclear p63 labeling has been reported to be a highly sensitive and specific marker of SC^4–7 with only rare expression in PT in limited series.⁸ However, our consultation experience suggests that both p63 and cytokeratin labeling in MP with classic histologic features is more common than otherwise reported in the literature.

p40 is the deltaNp63 isoform of p63⁹ and has been postulated to be a more specific marker of squamous or sarcomatoid differentiation.^10–12 The expression of p40 in PT or SC of the breast has not been reported. Here, we systematically study the expression of the p63 and p40 proteins in a series of breast fibroepithelial lesions with classic histologic features compared with SC, in the context of expression of cytokeratins (conventional markers of SC^8,13–17) and CD34 (a conventional marker of PT^15–17).

MATERIALS AND METHODS

Tissue Microarrays and Case Selection

This study was approved by the Institutional Review Board of the Johns Hopkins Medical Institutions. We evaluated a set of previously described tissue microarrays (TMAs)^18–20 constructed from archived formalin- fixed, paraffin-embedded tissues from 34 PTs, 10 fibroadenomas (FAs), and 13 SC. Each TMA consisted of 99 cores measuring 1.4mm in diameter, including 9 cores of benign normal control organs. To minimize sampling error and to approximate the size of a breast core needle biopsy sample, multiple cores per tumor were included on the TMAs (5 cores per PT and SC; 2 cores per FA), harvested from different regions of the tumor. One core per case included the presence of benign breast lobules as an internal control. Clinicopathologic data were collected, including patient age, race, resection type, tumor laterality, tumor grade and stage, hormone receptor and Her2 status, margin status, and clinical outcome.

The PT consisted of 14 MPs, 10 BLPs, and 10 BPs, which were subdivided into these categories on the basis of tumor circumscription, the degree of stromal cellularity, the presence or absence of stromal overgrowth, the degree of stromal nuclear atypia and pleomorphism, and the presence of stromal mitosis as evaluated on the original whole-tumor sections using established criteria.^1–2 The PTs selected for inclusion on the TMA were consecutive cases that had sufficient tissue available for TMA construction, as well as unambiguous, classic histologic features of the given subtype. Hematoxylin and eosin–stained whole sections of all tumor slides were reviewed on all cases at the time of TMA construction. Any tumor that was difficult to classify (ie, had feature of >1 subtype) was purposely excluded from the TMA. FAs were included in the analysis as they are in the differential diagnosis of BP.

Immunohistochemistry and Expression Scoring

The TMAs were labeled by immunohistochemistry (IHC) for p63, p40, CD34, and cytokeratins AE1/AE3, high–molecular weight cytokeratin (34betaE12), and CK8/18. The antibody source, specifications, and incubation timings are described in Table 1. In brief, IHC was performed on formalin-fixed paraffin-embedded tissue sections (4 μm thickness). Staining for all antibodies was done using automated machines (Ventana benchmark; Ventana Medical Systems Inc., Tucson, AZ). Slides were deparaffinized and hydrated, and heat-induced antigen retrieval was performed. Incubation with primary antibody using optimal conditions was followed by development of immunostaining and counterstaining as per manufacturer’s instructions.

TABLE 1.

Antibody Sources, Specifications, and Incubation Times

Antibody Name	Clone	Dilution	Antibody Incubation Time (min)	Source and Catalog #
Anti-pan cytokeratin (Pan)	AE1/AE3/PCK26	Prediluted	16	Ventana 760-2595
Cam5.2/CK8 &18	B22.1&B23.1	Prediluted	20	Cell Marque Ref# 818M-98
HMW CK	34betaE12	Prediluted	32	Cell Marque Ref# 334M-88
Anti-CD34	QBEnd/10	Prediluted	32	Ventana 790-2927
P40	Rabbit polyclonal	Optimized 1:2000	32 with amplification step	EMD Millipore/Merck
P63	4A4	Prediluted	32 with amplification step	Biocare Medical

The p63 and p40 nuclear protein expression was scored by labeling intensity (none, weak [W], moderate [M], or strong [S]) and percentage labeling (0% to 100%). Any labeling >1% was considered positive, with focal labeling defined as 1% to 5% labeling. The myoepithelial cells in the benign breast lobules and in the epithelial component of the PT served as internal controls for the p63 and p40 labeling. Cytoplasmic and membranous CD34 and cytokeratin expression was scored by percentage labeling, with any labeling considered positive and <5% labeling considered focal. The presence of blood vessels served as internal controls for CD34 labeling. The epithelial components of the benign breast lobules and of the PT served as internal controls for the cytokeratins. The IHC scoring was evaluated manually and reviewed by 2 board certified pathologists (A.C.-M. and P.A.).

RESULTS

Clinicopathologic Characteristics

The clinicopathologic characteristics of the patients with breast SC and fibroepithelial lesions are listed in Table 2. Briefly, the SC (n=13) had a mean tumor size of 4.7 cm (range, 1.6 to 13 cm) and were all triple negative for estrogen receptor, progesterone receptor, and human epidermal growth factor 2 (Her2) expression. The majority of SC (92%) were Elston grade III. Follow-up data were available on all 13 patients. Neoadjuvant therapy was administered in 23% of cases, and adjuvant therapy was administered in 69% of cases. No patients developed local recurrence, and 3 patients (23%) developed distant metastases. Two of the metastatic patients and a third with complications due to chemotherapy died of disease (23% of the total).

TABLE 2.

Clinicopathologic Features of Patients With Breast SCs, PTs, and FAs

	SCs	MP	BLP	Benign Phyllodes	FA
Number (n)	13	14	10	10	10
Age (y; mean, range)	64.5 (43–89)	48.3 (30–67)	45.9 (20–76)	38 (18–48)	26.2 (13–51)
Race (n)
White	7	5	4	5	4
Black	5	3	3	4	4
Other/NR	1	6	3	1	2
Resection (n)
Lumpectomy	5	8	10	10	10
Mastectomy	8	6	0	0	0
Tumor laterality
Left	9	8	5	3	7
Right	4	6	5	7	3
Tumor size (cm; mean, range)	4.7 (1.6–13)	7.6 (2.5–20)	5.2 (1–12)	2.1 (1–2.5)	3.2 (1–6.3)
Tumor Elston grade (n [%])		N/A	N/A	N/A	N/A
Grade I	0
Grade II	1 (8)
Grade III	12 (92)
Positive LN (if sampled) (n [%])	3/10 (30)	0/2 (0)	N/A	N/A	N/A
Triple negative for ER/PR/Her2 (n [%])	13 (100)	N/A	N/A	N/A	N/A
Final margins (n [%])
Positive	1 (8)	3 (21)	6 (60)	6 (60)	6 (60)
Negative	12 (92)	11 (79)	4 (40)	4 (40)	1 (10)
Not reported	0	0	0	0	3 (30)
Subset of cases with available follow-up data and status at last follow-up
Number	13	6	9	9	9
Mean follow-up (mo)	18.4 (0.5–53)	39.3 (15–110)	38.7 (7–112)	47.9 (0.5–129)	1 (0.3–6)
Neoadjuvant treatment (n [%])	3 (23)	0	0	0	0
Adjuvant treatment (n [%])	6 (26)	2 (33)*	1 (11)†	0	0
Status at last follow-up (n [%])
No evidence disease	9 (69)	3 (50)	9 (100)	9 (100)	9 (100)
Local Recurrence	0	0	0	0	0
Metastasis	3 (23)‡	3 (50)§	0	0	0
Died due to disease	3 (23)	3 (50)	0	0	0

^*Radiation.

^†Radiation, for concurrent ductal carcinoma in situ separate from the PT.

^‡23% of SCs developed metastasis, which consisted of 1 to the lung, 1 to the liver and lung, and 1 widespread. Two patients with metastasis died due to disease.

^§50% of MP with follow-up developed metastases, which consisted of 2 to the lung and 1 to the brain. All 3 patients who had metastases died due to disease.

ER indicates estrogen receptor; LN, lymph node; NR, not reported; PR, progesterone receptor.

Briefly, the MP (n=14) had a mean tumor size of 7.6 cm (range, 2.5 to 20 cm). Follow-up data were available on 6 patients (50%). No patient received neoadjuvant therapy, but 2 patients received adjuvant radiation therapy for the MP. No patient developed a local recurrence; however, 3 patients developed distant metastasis and died due to disease. The limited number of cases with follow-up precludes statistical comparison of the clinical course and outcomes between the patients with SC and MP. None of the patients with BLP, BP, or FA with follow-up data had a local recurrence, metastasis, or death due to disease. One patient with BLP received adjuvant radiation therapy for concurrent ductal carcinoma in situ within the same excision, but spatially separate from the BLP.

p63 and p40 IHC

The IHC profiles of breast SC, PT, and FA are summarized in Table 3, and the labeling profiles of the individual MP cases are detailed in supplemental material (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/PAS/A226). Nuclear p63 labeling of malignant stromal cells was seen in 57% (8/14) MPs, and half of these exhibited moderate to strong labeling (Figs. 1A–C). The remaining half demonstrated weak intensity of nuclear labeling. The majority (63%) of MPs with stromal p63 positivity exhibited focal (1% to 5%) nuclear labeling, and no MP demonstrated >30% cells labeling for p63. Nuclear p40 labeling of the malignant stromal cells was seen in 14% (2/14) MPs (Fig. 1D); both of these cases also showed p63 labeling, and both demonstrated stronger p63 labeling intensity than p40 intensity. As the number of cases with clinical follow-up was small, there were no associations between p63 and p40 expression and survival or metastasis. For instance, of the 3 patients who died due to metastatic MP (cases MP10, MP13, and MP14), 1 had no p63 labeling, 1 had <5% labeling, and 1 had 20% labeling. No nuclear p63 or p40 labeling was seen in any of the stromal cells in the BLP, BP, or FA cases (internal controls present and intact).

TABLE 3.

IHC Labeling for p63, p40, Cytokeratins, and CD34 in Breast SCs, PTs, and FAs

Tumor Type	Total Number	p63 Positivity (n [%])	p40 Positivity (n [%])	Cytokeratin Positivity (n [%])	CD34 Positivity (n [%])
SC	13	8 (62)	6 (46)	13 (100)	0 (0)
MP	14	8 (57)	2 (14)	3 (21)	8 (57)
BLP	10	0 (0)	0 (0)	0 (0)	10 (100)
Benign Phyllodes	10	0 (0)	0 (0)	0 (0)	10 (100)
FA	10	0 (0)	0 (0)	0 (0)	10 (100)

FIGURE 1.

Focal stromal p63 and p40 labeling is seen in a subset of MP tumors, but not seen in borderline or BP tumors. This MP tumor (MP case 2) contains regions of characteristic leaf-like architecture (A) as well as stromal overgrowth with highly atypical stromal cells (B), which are focally positive for p63 (C) and p40 (D).

Nuclear p63 labeling was seen in 62% (8/13) SCs (Table 3), all with moderate to strong labeling intensity (Figs. 2A–F) and a percentage labeling range from 20% to 100%. Nuclear p40 labeling was seen in 46% (6/13) SCs (Figs. 2G, H). Three (38%) SCs with p63 positivity were negative for p40 (Fig. 2I), and 1 SC that was negative for p63 did demonstrate focal weak p40 labeling. The intensity and distribution of p63 labeling was stronger than the p40 labeling in the majority (57%) of SC cases that labeled for both. The sensitivity and specificity of p63 in diagnosing SC versus MP was 62% and 43%, respectively; the sensitivity and specificity of p40 in diagnosing SC versus MP was 46% and 86%, respectively.

FIGURE 2.

p40 is more specific but less sensitive than p63 in the labeling of breast SCs. SCs of the breast (A–C) are in the differential diagnosis of MP tumors on core needle biopsy. Sixty-two percent of cases label with p63 (D–F), but only 46% of cases label with p40 (G and H); 3 p63-positive cases in this series were negative for p40 (I).

Cytokeratin IHC

Focal cytokeratin expression was seen in the malignant stromal cells in 21% (3/14) MPs (Table 3), all of which showed p63 labeling (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/PAS/A226). Two cases exhibited labeling with all 3 cytokeratin antibodies, and 1 case labeled with CK8/18 only. No cytokeratin labeling was seen in any of the stromal cells of the BLP, BP, or FA cases (internal controls present and intact). Cytokeratin immunoreactivity was seen in 100% (13/13) SCs (Table 3), and 62% (8/13) demonstrated labeling with all 3 antibodies.

CD34 IHC

At least focal CD34 labeling of stromal cells was seen in 57% (8/14) MPs (Table 3), including 4 cases with p63 positivity (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/PAS/A226). The majority (75%) of MPs with CD34 positivity showed only focal (ie, <5%) labeling; in contrast, 100% of BLP, BP, and FA cases demonstrated diffuse stromal CD34 labeling. No CD34 labeling was seen in the SC tumor cells (internal controls present and intact).

DISCUSSION

Breast PTs are rare fibroepithelial lesions subclassified into benign, borderline, and malignant subtypes, ^1,2 with variable risks for local recurrence (ranging from <20% for BP to >25% in MP²¹) and of distant metastasis (ranging from <5% in BLP to >25% in MP¹). The differential diagnosis of MP on core needle biopsy of the breast includes other high-grade neoplasms, most notably breast SC (metaplastic carcinoma). High-grade SC is a rare but generally aggressive subtype of invasive carcinoma³ that has a worse overall survival than conventional triple-negative invasive ductal carcinomas. ^22–24 SCs are characterized by heterologous elements or divergent differentiation such as chondroid, osseous, pleomorphic/sarcomatoid, spindled, and/or squamous differentiation.^3,25–28 Like MP, they also display atypical spindled cells with nuclear pleomorphism and abundant mitoses. However, MP is typically managed by complete surgical excision, whereas SC may be treated with neoadjuvant chemotherapy before surgery that may include sentinel lymph node sampling. Thus, accurate distinction between these 2 lesions on core needle biopsy is crucial. Incorrect diagnosis of an MP as an SC could result in exposure of a patient to preoperative chemotherapy, which has known toxicity but no proven benefit for an MP tumor. In addition, unnecessary sentinel lymph node sampling carries a small but real risk of lymphedema.

The p63 gene is related to the tumor suppressor p53²⁹ and regulates development of epithelium and epithelial structures.^30,31 IHC for the p63 protein has been reported to be specific to epithelial and myoepithelial cells,^32,33 to serve as a sensitive and specific marker of SC in the breast^4–7 and elsewhere, and initially reported to have limited expression in soft tissue neoplasms.³⁴ However, a recent study by Bishop et al, ¹² reported nuclear p63 labeling in both stromal neoplasms (30%) as well as in benign reactive stromal proliferations (30%). They and others^11,12 report that p40, the deltaNp63 isoform of p63, is a more specific marker of squamous differentiation or sarcomatoid differentiation in spindled lesions.

Stromal p63 labeling in PT has been evaluated in several previous studies, with p63 positivity documented in 1 study⁸ but the majority reporting no labeling.^4,5,7 However, the studies vary widely in the number of PTs analyzed, ranging from only 5 to 109.^7,8 In addition, the grade of PT varied widely, with the number of MPs studied ranging from only 3 to 9.^4,7 No previous study has evaluated p40 expression in breast PT or SC. Stromal cytokeratin expression in PT has also been investigated, with cytokeratin positivity documented in 2 studies^7,35 but the majority reporting no expression.^8,13–15 However, these studies also vary widely in both the number and spectrum of cytokeratin antibodies evaluated, ranging from 2 to 7,^7,14,15 as well as the type of PT analyzed, ranging from 3 to 9 MPs.^7,15

Not surprisingly, the most recent study documenting cytokeratin positivity in PT included a broad panel of 6 cytokeratins and evaluated a large number (109) of PTs, including 9 MPs.⁷ Thus, one explanation for the range in published rates of cytokeratin or p63 positivity in PTs is that the expression is focal and limited to MP, as seen in our results.

Our consultation and clinical experience suggests that both p63 and cytokeratin labeling in MP is more common than the literature would suggest and was the impetus to perform this formal study. In 1 such illustrative case (Fig. 3), a core needle biopsy of a breast mass revealed highly atypical spindled cells with focal cytokeratin positivity (Figs. 3A, B), p63 positivity, and a focal epithelioid component (Fig. 3C). One consideration was to diagnose the lesion as an SC. Instead, we chose to classify the lesion as a malignant spindle cell neoplasm, favoring SC on the basis of the cytokeratin positivity; however, we felt that an MP could not be excluded with certainty, so a complete excision was recommended. The excisional specimen revealed histologic features of a classic MP (Figs. 3D–J) but with aberrant p63 and cytokeratin labeling (Figs. 3K, L).

FIGURE 3.

Case example of a MP tumor with aberrant cytokeratin and p63 labeling seen in consultation. A needle biopsy of a large breast mass (A–C) contained highly atypical spindled to epithelioid cells (A, H&E) that were positive for cytokeratin (B, AE1/ AE3) and p63 (not shown). The lesion had 1 focal area suggestive of an associated epithelial component (C), which may have also been an entrapped benign duct. The tumor was diagnosed as a malignant spindle cell neoplasm, favor metaplastic carcinoma on the basis of the cytokeratin expression; however, an MP tumor could not be excluded with certainty, and excision was recommended. At mastectomy, the lesion displayed classic characteristics of a PT including multiple regions of leaf-like architecture (D–G) with low-grade to intermediate-grade features, as well as abundant zones of stromal overgrowth (H) with highly atypical malignant cells (I), diagnostic of an MP tumor. Similar to the core biopsy, the atypical stromal cells (J) displayed focal positivity for p63 (K) and cytokeratin (L, CK8/18), best interpreted as aberrant expression in an otherwise classic MP tumor.

On the basis of such distinctive cases, in the current study we systematically analyze the expression of p63, p40, and cytokeratins in a series of breast SCs and classic PTs. We used TMAs, which, despite sampling of multiple cores per case, have the disadvantage of a limited size of tumor sampling. However, the aggregate tumor size of the samples on these TMAs approximate that of a core needle biopsy specimen, which is the setting in which the potential for misdiagnosis is greatest and which we sought to approximate.

Our results indicate that focal p63, p40, and cytokeratin labeling can be seen in MP tumors but not in BLP, BP, or FA, and immunolabeling with any of these markers alone is not diagnostic of SC in the breast. Although the p40 isoform of p63 is a more specific (but less sensitive) marker of SC than p63, the presence of labeling in 14% of MPs limits its utility in clinical practice. CD34 immunolabeling may help resolve some cases; however, not all MPs demonstrate CD34 positivity. Other newer markers with differential expression in MP and SC, such as Sox10¹⁸ and GATA3,¹⁹ may have additional utility in making this distinction between these entities on core needle biopsy.

In summary, focal p63, p40, and cytokeratin labeling can be seen in MP tumors, and immunolabeling with any of these markers alone should not be used to diagnose SC on breast core needle biopsy. Future studies are needed to determine the biological basis for p63, p40, and cytokeratin positivity in MP. One possibility is that this expression may be related to aberrant myoepithelial differentiation in a subset of these lesions. Our results are consistent with the more recent published literature in which p63 labeling has been increasingly observed in sarcomas¹² and suggest caution in classification of malignant spindle cell tumors of the breast on core needle biopsy.