Phyllodes tumors of the breast: clinicopathological analysis of 106 cases from a single institution

Abstract

BACKGROUND:

Phyllodes tumors (PT) are uncommon biphasic tumors, accounting for less than 1% of all breast primary neoplasms. They form a wide variety of tumors ranging from benign to malignant. Several histological features are used to grade PT into 3 categories: benign (grade I), borderline (grade II) and malignant (grade III) tumors. The aim of our study was to analyse histolopathological, radiological and clinical features of PT from an experience of a single center.

METHODS:

It was a retrospective study including 106 patients diagnosed with phyllodes tumors on surgical specimens at the department of pathology, of Hassan II university hospital (Fez, Morocco), from 2009 to 2016.

RESULTS:

The mean age was 33.81 years (range of 13–66 years), and the mean age increases with the tumor grade (mean ages of 32.32, 32.87 and 33.65 years respectively for grade I, II and III PT) (p = 0.023); 78 patients (73.58%) had benign PT, 20 (18.86%) had borderline PT and 8 (7.54%) patients were diagnosed with malignant PT. Mostly, the tumor size was <5 cm (63.2%), with BI-RADS 3 (51 patients, 48.11%). The tumor size and the radiological suspicion (ACR/BI-RADS) increased with the tumor grade (p < 0.001). Mitosis count, cellular atypia and stromal cellularity increased with the tumor grade (p < 0.001). Also, the presence of necrosis is associated with malignant PT (p < 0.001). Before surgery, patients had undergone core needle biopsies (CNB) for diagnostic purpose, and the overral sensitivity of this diagnostic procedure was 71.83%. The sentivity of the CNB decreased from grade I PT to grade III PT (from 56.81% to 37.5%), however its specificity increased from grade I to grade III PT (from 59.25% to 100%).

CONCLUSION:

Phyllodes tumors of the breast are rare neoplasms with a wide range of clinicopathologic presentations. The core needle biopsy has a good diagnostic sensitivity compared to definitive diagnosis on surgical specimens. There was a statistically significant association between the histological grade of PT and tumor size, radiological suspicion, mitotic count, cellular atypia, stromal cellularity, and tumor necrosis.

Background

Phyllodes tumors (PT) are uncommon biphasic tumors, accounting for less than 1% of all breast primary tumors [1,2]. PT show fibroepithelial components consisting of bilayered benign epithelium disposed within hypercellular stromal component [2,3]. These tumors have variable clinical course according to a number of histological features of the stromal component that are used to grade them into 3 categories: benign, borderline and malignant PT. The grading of these tumors is problematic and challenging as histologic criteria are semi-quantitative and relies on the subjective appreciation by the pathologist [2]. The grading system commonly used is based on stromal cellularity, atypia, mitotic count, the nature of tumor margins and the presence of a malignant heterologous component. The exact importance of each histologic factor in determining the grade of the tumor is another controversial and unresolved issue [1–4]. The aim of our study was to report clinicopathologic features of PT from experience of a single institution over a period of 8 years (2009–2016).

Materials and methods

Patients selection

We retrospectively included in the study all patients diagnosed with PT from January 2009 to December 2016, at the department of pathology of the Hassan II university hospital (Fez, Morocco). All the available clinical, radiological and histological data were retrieved from the pathology request forms and the patients’ medical records. The radiological aspects (mammography and ultrasonography) of the lesions are classified according to the American College of Radiology Breast Imaging Reporting And Database System (ACR/BI-RADS).

Histological analysis

All tissue samples have been fixed in 10% buffered formalin, embedded in paraffin and stained with hematoxylin-eosin-saffron (HES). The histological diagnosis of PT was based on the World Health Organisation (WHO) classification (2003) criteria such as stromal cellularity, stromal overgrowth, atypia, mitotic count, the nature of margins and the presence of the malignant heterologous component. The 3 groups of PT are: benign (grade I), borderline (grade II) and malignant (grade III) (Fig. 1).

Statistical analysis

The statistical analysis was performed to compare clinical, radiological and histological features among the 3 groups of PT. Differences in the distribution of variables were evaluated using the Fisher exact test or chi-square test, as appropriate, and age means have been compared by using the ANOVA (one-way analysis of variance). All statistical analyses were performed by using SPSS 20.0 version software for Windows (SPSS, Inc., Chicago, IL, USA). P  value was considered statistically significant at P < 0.05.

Results

Clinical and radiological characteristics

All included patients were females, the mean age of the 106 patients was 33.81 years (range of 13–66 years). A total of 78 patients (75.68%) had benigh PT, 20 patients (18.86%) had borderline PT and 8 patients (7.54%) were diagnosed with malignant PT. Patients with grade III tumors were older than those with lower grade tumors, as the mean ages for benign, borderline and malignant PT were 32.32, 32.87 and 33.65 years respectively, (p = 0.023) (Table 2). In 77 patients, the radiological data (mammography and/or breast ultrasound) have been retrieved. The ACR/BI-RADS score increased with the histological grade of the tumor (p < 0.0001) (Tables 1 and 2). In fact, the majority of grade I tumors were classified as BI-RADS 3 (45 out of 52 patients, 86.53%) whereas most grade III PT were BI-RADS 5 (6 out of 7 patients, 85.71%).

Pathological features

The tumor size was <5 cm in 63.2%. Most grade I PT tumors measured less than 5 cm (61 out of 77 patients, 79.22%) while grade III tumors were more than 10 cm (6 out of 8 cases, 75%). It appears that the tumor size is significantly associated with a higher grade (p < 0.0001). Also, the statistical analysis showed that histological features like mitotic count, cellular atypia, stromal cellularity and tumor necrosis, are correlated to higher grades tumors (p < 0.0001) (Table 2).

We retrieved histological results of 71 patients with prior core needle biopsies (CNB) before surgery. On biopsies PT showed various histological features, from adenofibroma to grade III tumors (Table 3). Among 13 cases diagnosed as adenofibromas (ADF) on CNB, 12 were difinitively dignosed as grade I PT on surgical specimens. On CNB, sometimes grade I tumors shows histological aspects partially similar to adenofibroma, leading to the ambiguous diagnostic term benign “fibroepithelial tumor” (FET). In our study, all cases of FET (7 patients) prooved to be grade I PT on surgical specimens. Also, on CNB 11 cases out of 36 initially diagnosed as grade I PT have been upgraded as grade II tumors on surgical specimens. Five cases out of 12 diagnosed on CNB as grade II PT have been diagosed as malignant PT on surgical specimens. In fact, 25 cases (56.81%) of grade I PT were correctly diagnosed and 19 cases (43.18%) were diagnosed as ADF or benign FET. Grade II PT were usually diagnosed on biopsy as grade I tumors, grade III tumors were mostly correctly diagnosed (Tables 3 and 4). In sum, the overral diagnostic sensitivity of the CNB was 71.83%. The diagnostic sentivity of CNB decreased from grade I PT to grade III PT (from 56.81% to 37.5%), however the specificity increased from grade I to grade III PT (from 59.25% to 100%).

Treatment and outcome

Of 106 patients, 84 have been treated by lumpectomy (74.24%) and the remaining 22 patients (20.75%) have undergone total mastectomy; they have all negative margins. Three patients with grade III PT and “near positive” margins (1–2 mm) have undergone adjuvant radiotherapy. Available follow-up data showed that only 3 patients had local recurrence, and metastasis or death have not been recorded. One patient with initial grade I PT treated by lumpectomy, had local recurrence as grade II tumors 3 months later, a total mastectomy has been performed for here. Another patient with grade II PT had also local recurrence as grade II PT 2 years after the initial diagnosis. She had a prior surgical treatment as lumpectomy. The third patient with grade III PT, had local recurrence as a high grade PT 3 years later. She had undergone total mastectomy with adjuvant radiotherapy.

Discussion

Phyllodes tumors (PT) are uncommon fibroepithelial tumors of the breast representing less than 1% of all primary breast tumors [4,5]. They constitutes a wide spectrum of tumors, ranging from benign to malignant type. Attempt has been made to establish histological grading system dividing PT into 3 gategories: benign PT (garde I) with potential of local recurrence, borderline PT (Grade II) that has a potential of local recurrence and a low risk of metastasis, and malignant (Grade III) with a highest potential of local recurrence and distant metastasis [2–5]. This histological grading system relies on the degree of stromal cellularity, cellular atypia, mitotic count, stromal overgrowth and the nature of the tumor borders (pushing or infiltrative). Other histological aspects like necrosis or hyperplasia has been reported [3]. The relevance and the prognostic value of these histological criteria have been subject to several studies. Our study assess the prognostic value of histological criteria used to grade PT tumors. We found that most PT were grade I, in 78 patients (75.58%), whereas grade II and grade III were relatively rare, in 20 cases (18.86%) and 8 cases (7.54%) respectively. These findings are similar to those generally reported in the litterature [2,6–10]. However, the predominance of grade II and III tumors has been reported in some previous series [11–13].

The PT affect patients mostly between 35 and 55 years, with the malignant type occuring later compared to benign type [2–6]. The overral mean age in our series was 33.81 years, with statistical difference between the 3 histological types (p < 0.001), the malignant potential increased as patients get older. Some studies showed that PT tumors occurred earlier in Asian and Latino-White patients, with high level of recurrence, compared to Caucasian patients [12,14]. Another study by Bampers et al. on 50 African-Americans patients showed that this ethnic group had more aggressive PT [15]. But these findings about ethnic differences are controversial as other studies did not confirm this fact [12].

The diagnostic value of different imaging modalities is another challenging and unresolved issue. Discrimination between benign and malignant PT or between PT and fibroadenoma, is not easy by several radiological techniques. One study by E. Yilmaz et al.. suggested that although masses of high density at mammography, circumscribed border associated with posterior acoustic enhancement and internal cystic areas at sonography should suggest the diagnosis of phyllodes tumors rather than large sized fibroadenomas, and there was a substantial overlap in the mammographic and sonographic characteristics of these two tumors [16]. Another study using the magnetic imaging resonance (MRI) of PT pointed out that some MRI findings like irregular cyst wall, tumor signal intensity (SI) lower than or equal to normal tissue SI on T2-weighted images, and low apparent diffusion coefficient (ADC) correlated significantly with histologic grade and could be used to help determine the histologic grade of phyllodes breast tumors [17]. Elastography was also another imaging technique that could help to differentiate PT from adenofibroma [18]. Our study showed that the radiological suspicion of malignancy increased from grade I to grade III PT.

We found that tumor size increased with the grade of the tumor, emphasizing that the size is another feature of agressiveness. It was widely reported that malignant PT are larger and the tumor size is correlated with a worse prognosis. This appears logic as malignant tumors have a higher mitotic count and grow more rapidly, thus producing larger amount of tumoral tissue.

Patients with breast lesions often underwent CNB for a diagnostic purpose. A number of previous studies have reported diverse aspects of the CNB diagnostic value in fibro-epithelial lesions. Our current study found that the diagnostic sensitivity of CNB decreased from grade I PT to grade III PT (from 56.81% to 37.5%) and the specificity increased from grade I to grade III PT (from 59.25% to 100%). Zhou et al. have assessed the diagnostic performance of CNB in identifying breast phyllodes tumors in 128 patients [19]. They have reported a low sensitivity of this diagnostic procedure (13.3%). The majority of their patients have been diagnosed as fibroadenoma or benign fibroepithelial tumors. They have reported higher CNB diagnostic specificity that increases according to tumor grades, like our current study. Also, at CNB initial assessment, the diagnosis of fibroadenoma or fibroepithelial lesions (FEL) was frequently rendered by pathologists. In our series, 20/71 patients (28.16%) had this type of diagnosis (13 cases diagnosed as fibroadenomas and 7 as FEL). Previous stydies on such types of diagnosis (fibroadenomas or FEL) on CNB showed that a few proportion of patients that underwent surgical excision showed a definitive diagnosis of PT (37,5% to 38%) [20,21].

The histological criteria used for grading PT have been subject to several studies [2–8]. The grading system is based on a constellation of some histological features of the stromal component, such as hypercellularity, atypia, mitosis, margin aspects (pushing or infiltrating) [1,3]. However, each of these criteria does not suffice alone to predict accurately the behavior of PT [2]. In our study, marked cellular atypia, stromal cellularity and higher mitotic count were correlated to agressiveness of the tumor (p < 0.0001). In the literature, it has been widely reported that tumor behavior, such as local recurrence or distant metastasis, is significantly correlated to tumor size, stromal atypia, mitosis count or margins status [2,22–24]. Recently, Tan PH et al. have suggested a nomogram based on histological criteria and surgical margins to predict the clinical behaviour of PT [2]. They called it AMOS criteria as it was based on atypia, mitoses, overgrowth and surgical margins.

Our current study showed that only 3 patients had local recurrences (2 patients with grade I and II, had local recurrences as grade II PT and 1 patient with grade III PT recurred as a high grade PT), we have not recorded cases of death or distant metastases. These 3 patients had initially negative margins on lumpectomy. Predictive factors of local recurrences are mainly positive margins or histological factors used to grade PT, such as mitotic count, cellular atypia or infiltrative margins [2,3,7].

Conclusion

Phyllodes tumors of the breast are rare neoplasms with a wide range of clinicopathologic presentations. The core needle biopsy has a good diagnostic sensitivity compared to definitive diagnosis on surgical specimens. The current study showed that there was a statistically significant association between the histological grade of PT and tumor size, radiological suspicion, mitotic count, cellular atypia, stromal cellularity, and tumor necrosis.