Article Text

Download PDFPDF

Mammary phyllodes tumour: a 15-year multicentre clinical review
  1. Michael Co1,
  2. Clement Chen1,
  3. Julia Y Tsang2,
  4. Gary Tse2,
  5. Ava Kwong1
  1. 1 Department of Surgery, Division of Breast Surgery, Queen Mary Hospital, The University of Hong Kong, Hong Kong SAR, China
  2. 2 Department of Anatomical and Cellular Pathology, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong SAR, China
  1. Correspondence to Dr Ava Kwong, Division of Breast Surgery, Room K1401, The University of Hong Kong, Queen Mary Hospital, Hong Kong SAR, China; avakwong{at}


Aims Phyllodes tumour (PT) is an uncommon fibroepithelial tumour of the breast. It has a spectrum of aggressiveness in biological behaviour with chance of local recurrence and, occasionally, metastasis.

Methods A 15-year retrospective review from a multicentre database in Hong Kong was performed.

Results Clinical and pathological records of 465 patients with 469 PTs between 1998 and 2014 were reviewed. Median age of occurrence was 44 years (range 12–86 years). 281 (59.9%) PTs were benign, 124 (26.4%) were borderline and 64 (13.6%) were malignant. About half of all PTs (239, 51.5%) were between 2 and 5 cm while another 186 (40.1%) were >5 cm in size. Most PT (84.6%) were radiologically benign. Breast-conserving surgery (BCS) was feasible in 384 (82%) patients, whereas 84 (18%) patients had mastectomy. Multivariate analysis found that positive surgical margin (P<0.001) and BCS (P<0.001) were the only significant risk factors for local recurrence, while large tumour size (P=0.008) and malignant PT histotype (P<0.001) were the only significant risk factors for metastasis. Long-term prognosis of benign and borderline PT was excellent. After median follow-up interval of 85 months (range 12–180 months), the disease-specific survival of benign, borderline and malignant PT were 99.6%, 100% and 90.6%, respectively.

Conclusions Local recurrence of PTs occurs irrespective of the tumour grade. Surgical margin is the only amendable factor to reduce the chance of recurrence.

  • breast pathology
  • breast
  • surgical pathology

Statistics from

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.


Phyllodes tumours (PTs) of the breast are rare fibroepithelial tumours occurring in 1 per 100 000 women and account for only 0.5% of all breast neoplasms.1 PT is classically divided into benign, malignant and borderline according to their histopathological features;2 3 its clinical course is often unpredictable, and it is more common to occur in middle-aged women.4 5

Clinically, PT usually presents as a round, mobile and painless lump. It is difficult to distinguish PT (both benign and malignant) from benign lesions based on clinical or radiological features.6 In fact, there are no pathognomonic mammographic or ultrasound features.7 As such, pseudo-angiomatous stromal hyperplasia, fibroadenoma, intraductal papilloma or complex cysts are the main differential diagnoses. Special types of carcinomas, including carcinoma with medullary features or mucinous carcinoma, may be considered as well. Even though imaging plays a fundamental role as a first diagnostic approach, final diagnosis is confirmed only by histological examination.

Given the rarity of PT, only a few large-scale studies had been published in the literature.8 The current series represents another large cohort to further delineate relevant clinical factors affecting outcomes in PT.


A 15-year retrospective review from a multicentre database in Hong Kong was performed. This database included prospectively maintained clinical and pathological data of patients treated with PT in five major university/government hospitals in Hong Kong. Patient demographic and clinical data, radiology and pathology results were retrieved and analysed. The diagnosis of the PT was based on WHO criteria.2 Risk factors of local recurrence were evaluated by univariate analysis and multivariate analysis. Kaplan-Meier survival analysis with log-rank test was performed to evaluate the difference of local recurrence rate between different PT histotypes.


Between 1998 and 2014, the clinicopathological records of 465 women with 469 PTs were retrieved from a combined database from five major breast centres in Hong Kong. Median age of occurrence was 44 years (range 12–86 years). In total, 281 (60.0%) PTs were benign, 124 (26.4%) were borderline and 64 (13.6%) were malignant; none of our patients had documented malignant heterologous sarcomatous element.

All patients presented with symptomatic breast mass, and all underwent triple assessment with breast ultrasound (±mammogram) and core needle biopsy (CNB), followed by multidisciplinary management. More than half (n=239, 51.5%) of all PTs were between 2 and 5 cm while another 186 (40.1%) were >5 cm in size. Most patients (n=397, 84.6%) had benign sonographic features (Breast Imaging Reporting and Data System (BIRADS) 2 or 3), suggestive of benign breast neoplasms. The median tumour size under ultrasound was 27 mm (range 13–130 mm). Only 36 patients (7.7%) had the diagnosis of PT made preoperatively by CNB, while most patients (n=325, 69.3%) were diagnosed as ‘fibroepithelial lesion’ by preoperative CNB. Of note, up to 9% of the current cohort were misdiagnosed as fibroadenoma (n=42) or ‘no evidence of malignancy’ (n=66) by CNB preoperatively. Breast-conserving surgery (BCS) was feasible in 384 (81.9%) lesions, whereas 85 (18.1%) had mastectomy. In total, 8 patients received chemotherapy (including 1 benign PT with coexisting invasive ductal carcinoma and 7 malignant PT) and 47 patients received radiotherapy (34 malignant PTs and 13 borderline PTs) (table 1).

Table 1

Baseline patient demographics and summary of treatment received

Over the 15-year study period, local recurrence occurred in 38 patients (21 in benign, 14 in borderline and 3 in malignant PT). With adequate treatment, there was no difference in the risk of local recurrence between benign, borderline and malignant PT (P=0.221) (figure 1). On the contrary, distant metastasis occurred solely in malignant PT (eight patients, 89%) or PT with malignant transformation (one patient, 11%). (figure 2).

Figure 1

Local recurrence in benign, borderline and malignant phyllodes tumour. Log-rank test, P=0.221.

Figure 2

Distant metastasis in benign, borderline and malignant phyllodes tumour. Log-rank test, P<0.001.

Univariate analysis found that among the factors including age, tumour size, final surgical margin and type of surgery (ie, BCS), only large tumour size involved final surgical margin and BCS were associated with increased risk of local recurrence. Similarly, large tumour size, malignant histotype and BCS were associated with increased risk of distant metastasis (tables 2 and 3).

Table 2

Univariate analysis of risk factors for local recurrence

Table 3

Univariate analysis of risk factors of distant metastasis

Multivariate analysis was performed for local recurrence and distant metastasis using Cox proportional hazards model by forward selection. Initial model included the positive factors from the univariate analysis (ie, tumour size, surgical margin status and surgery type as factors for local recurrence; tumour size, PT histotype and surgery type as factors for metastasis). Eventually, only surgical margin and surgery type remained statistically significant for local recurrence in the model through forward selection by Cox proportional hazards model (P<0.001) (table 4). Similarly, only large tumour size and malignant PT histotype remained statistically significant for metastasis (P=0.008 and <0.001, respectively) (table 5).

Table 4

Multivariate analysis of risk factors of local recurrence using Cox proportional model

Table 5

Multivariate analysis of risk factors of metastasis

After median follow-up interval of 85 months (range 12–180 months), there were seven PT-related mortalities from the current cohort, including one benign PT with malignant transformation and seven malignant PT. The patient who developed malignant transformation of benign PT was initially treated with wide excision but tumour recurred in 4 months where she received completion mastectomy. The patient developed metastasis within 1 year of mastectomy and died of metastatic PT 96 months after the initial operation. Disease-specific survival was 99.6% in benign PT, 100% in borderline PT and 90.6% in malignant PT (figure 3).

Figure 3

Disease-specific survival by different phyllodes tumour histotype.


PT is an uncommon breast neoplasm, accounting for <1% of all benign and malignant breast tumours.1 Only a few large-scale studies had been published in the literature, including the study by a Singapore group in 2005 that has evaluated 335 cases of PT.8

Our current cohort presents one of the largest series in the literature with 469 PTs diagnosed over the 15-year study period. Due to the limitations of our retrospective multicentre database, the total number of benign/malignant breast tumours treated during the same period of time is lacking. As such, we are unable to precisely estimate the proportion of PT new cases among all breast neoplasms treated in the five centres. Nevertheless, if we take Queen Mary Hospital, the largest volume centre of all five participating hospitals in this study, as an example, 121 PTs were treated over the last 15 years compared with 3042 breast cancers operated in the same centre.9 PT is therefore an uncommon disease even in large volume breast centres.

PT of the breast was first characterised as a distinct disease entity by Johannes Muller in 1838 using the term cystosarcoma phyllodes.10 11 PT is classified into benign, borderline and malignant lesions, based on the histological parameters of stromal hypercellularity, cellular pleomorphism, mitotic activity, nature of the margins and stromal pattern.2 Some other authors subdivide PT into benign and malignant and further categorised the malignant group into low and high grades, essentially retaining a three-tiered classification.12 13 However, it is known that diagnosis and classification of PT is not straightforward and is subjected to much interobserver variation.14 15

Unlike fibroadenoma, which is essentially benign in nature, PT, even if it is histologically benign, tends to recur and progress. Malignant transformation occurs in up to 30% of PTs, usually within the stromal component. Presence of heterologous sarcomatous elements was reported to be associated with poor outcome.16 In fact, the relationship between these fibroadenoma and PT is poorly understood; early reports documented the presence of areas of fibroadenoma within an otherwise characteristic PT,10 17 suggesting that fibroadenoma may be the possible precursor or progenitor of PT. It was until more recently that progression of fibroadenoma to PT was better described using molecular evidence.18 19 It is suggested that fibroadenoma might progress not only in a stromal manner into PT, but also in an epithelial direction towards carcinoma in situ. This suggestion is further supported by a recent region-wide study on PT. Out of the 557 patients treated for PT in Hong Kong and Shenzhen, China, 6 patients had coexisting ductal carcinoma;20 coexistence of ductal carcinoma within the epithelial component of PT could have been underdiagnosed and under-reported in the past.

Treatment of PT is mainly surgical, aiming for complete excision with an adequate margin. A large retrospective study was conducted in Singapore on 335 patients in 2005; the authors concluded that the involved margin is associated with an increased risk of local recurrence.8 We routinely advise our patients with involved surgical margin to undergo further excision of margin, but some patients declined further treatment. In the current series, involved final surgical margin is also one among the most important risk factors for local recurrence but not distant metastasis.

One interesting observation from our study is that at least 186 patients presented with large breast tumour of >5 cm; however, wide excision was still feasible in most of these patients. Although surgical decision for or against mastectomy is largely based on the tumour size and patient cup size, preoperative CNB diagnosis also plays an important role. Almost 80% of patients in our cohort had preoperative diagnosis suggestive of benign ‘fibroepithelial lesion’ or even misdiagnosed as ‘fibroadenoma’, resulting in the tendency towards more conservative surgical approach in these patients.

In terms of long-term survival, patients with benign PT demonstrate very good disease-specific survival with <1% of PT-related death in 10 years. However, the situation is completely different in malignant PT, where the 10-year disease-specific survival has significantly dropped to 90%.

We recognise the inherited limitation of this study of being a retrospective study. However, our results were derived from a sizeable cohort from a prospectively maintained database from five major breast centres in the region. Based on the findings of this study, surgical margin is the only amenable factor to reduce local recurrence. Careful preoperative planning to ensure adequate surgery with clear margin is of utmost importance when managing patients with PT. Mastectomy should be performed only when good surgical margin cannot be confidently achieved.


Local recurrence tends to occur in patients with involved surgical margins and in patients who underwent BCS. Malignant PT is associated with metastasis and poor long-term survival outcome.

Take home messages

  • Malignant phyllodes tumour (PT) is associated with increased risk of metastasis, but not local recurrence.

  • Positive surgical margin and breast-conserving surgery are the only identifiable risk factors for local recurrence.

  • Large tumour size and malignant histotype are the only significant risk factors for metastasis.

  • Long-term disease-specific survival in benign and borderline PT is excellent.

Abstract translation

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


We would like to acknowledge Dr Zeng Shengguang (Specialist Breast Surgeon, The University of Hong Kong-Shenzhen Hospital, China) for helping with the translations of the abstract.



  • Handling editor Dhirendra Govender.

  • Contributors MC is responsible for data collection, analysis and manuscript preparation. AK is responsible for data collection and coordination of database, data analysis and manuscript preparation. CC, GT and JYT all contributed to the study by helping with data collection and analysis.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval The University of Hong Kong/Hong Kong Hospital Authority (West Cluster).

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