Molecular pathology of phyllodes tumours of the breast—much more than MED12

Abstract

Phyllodes tumours of the breast present challenges in their diagnosis, classification and management. Further understanding of the molecular changes underpinning these tumours may lead to more precise classification and potential treatment options. Similar to fibroadenomas, MED12 is the most frequently mutated gene in phyllodes tumour. However, in addition, there is a spectrum of molecular alterations from benign to malignant phyllodes tumours with increasing genomic complexity, high level copy number alterations and aberrations of cancer driver genes in malignant phyllodes tumours. This review summarizes the molecular pathology of phyllodes tumours, the use of these data in developing a model of phyllodes tumour pathogenesis, and how molecular pathology might be applied to aid diagnosis and guide treatment in this rare tumour type.

Phyllodes tumours can develop from fibroadenomas bearing MED12 variants by the development of pTERT alterations (“MED12 pathway”) or de novo (“MED12 independent pathway”). Grade progression is associated with increasing genetic complexity including cancer driver gene aberrations. Molecular alterations can be useful in assigning grade and distinguishing phyllodes tumours from histological mimics.

Abbreviations

FFPE

formalin fixed, paraffin embedded

ROR

risk of recurrence

Introduction

Phyllodes tumours are relatively rare tumours which form part of the spectrum of fibroepithelial tumours of the breast and are graded as benign, borderline and malignant phyllodes tumours based on several histological features. ¹ , ² In Australia, they account for 0.2% of all breast cancers with only ~32 malignant cases per year diagnosed in 2019, equating to 0.2–0.3 per 100,000. ³ This is similar to the incidence reported in England (2 per million women per year), ⁴ USA (average annual age‐adjusted incidence 2.1 per 1 million women) ⁵ and in the Netherlands (0.3 per 100,000 person years). ⁶ In contrast, phyllodes tumour accounted for 6.92% of breast cancers in a Singaporean institutional study. ⁷

Distinction of a phyllodes tumour from its mimics and grading of phyllodes tumour are essential for appropriate patient management. ⁸ Unfortunately, this is also a recognized challenge amongst pathologists, ⁹ , ¹⁰ especially on core biopsy specimens, due to tumour heterogeneity, overlapping histological criteria, inter‐observer variation in interpretation of these criteria, and in most centres, low case volumes of phyllodes tumours. Therefore, consistent histological classification of phyllodes tumours is difficult in practice and may not strictly correlate with clinical outcome. ⁹ , ¹⁰ , ¹¹ , ¹²

Over the last decade, identification of recurrent molecular alterations in fibroepithelial tumours has allowed greater understanding of the pathogenesis of fibroepithelial lesions and provided additional evidence for grading, risk stratification, and in differentiating phyllodes tumour from other entities including fibroadenoma and metaplastic carcinoma, and presented the possibility of targetable molecular alterations. This review aims to provide an overview of the molecular alterations in phyllodes tumours and their potential clinical utility.

Molecular Alterations in Phyllodes Tumours

Germline Molecular Alterations

Of the tumours that occur in patients with germline TP53 mutations (Li‐Fraumeni syndrome), malignant phyllodes tumours have the greatest incidence relative to the general population with 78x relative frequency. ¹³ Other reported pathogenic or likely pathogenic germline variants in patients with phyllodes tumours include mutations of BRCA1, RB1, CHEK2, APC, ATM, HOXB13, MITF, RAD 51D, BARD1, MSH6, POT1, RAD51C, SDHA, MUTYH, BLM, WRN, NTHL1, POLE, and RECQL4. ¹⁴ , ¹⁵ In a cohort of 234 women with phyllodes tumours, pathogenic or likely pathogenic germline variants occurred in approximately 14% of patients, with just of a half of these being mutations associated with an autosomal dominant pattern of inheritance, similar to that observed in patients with breast carcinoma. ¹⁴ The prevalence of pathogenic or likely pathogenic variants increased with increasing grade of phyllodes tumour, although this difference did not reach statistical significance. ¹⁴ Nevertheless, one might argue that this frequency is sufficient to warrant reflex germline testing in patients with phyllodes tumours.

Somatic Alterations

While some have reported similar molecular alterations in both the stromal and epithelial components of fibroepithelial lesions, ¹⁶ , ¹⁷ most studies have observed molecular alterations to be confined to the mesenchymal component of the tumour, ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² consistent with this compartment being the dominant neoplastic component of the lesion. Molecular heterogeneity within a single phyllodes tumour has been described, with differences in copy number alterations and mutations between different areas of the same tumour, ²³ and between pre‐operative core biopsy tissue and subsequent excisions. ²⁴ Phyllodes tumours are microsatellite stable and have a relatively low median tumour mutational burden with reported median tumour mutational burdens ranging between 2.5 mutations/MB and 2.7 mutations/MB with 1.5% to 3.3% of cases harbouring tumour mutational burdens of at least 10 mutations/MB. ²⁵ , ²⁶ , ²⁷

MED12 Mutation in Phyllodes Tumours

MED12 is the most frequently mutated gene in phyllodes tumours (53%) and variants have been reported in all grades of phyllodes tumour ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ as well as in fibroadenomas. ¹⁹ More than one MED12 mutation can exist within a single tumour ²⁸ and MED12 mutations can differ between fibroepithelial lesions from an individual patient. ²⁸ , ⁴⁴ The product of the MED12 gene, mediator complex subunit 12, forms part of a protein complex known as mediator coactivator complex which regulates transcription either by interacting with other transcriptional regulators such as CDK8 kinase, similar to the mechanism seen in uterine leiomyomas, or directly in the case of oestrogen receptors alpha and beta. ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ The majority of MED12 mutations in phyllodes tumours affect exon 2, especially codon 44, ³¹ and result in loss of function of this mediator complex ⁵¹ , ⁵² , ⁵³ from altered positioning of the MED12 “activation helix,” which is necessary for activating CDK8. The mutations themselves do not prevent MED12 from binding CDK8 but diminish the kinase activity by misplacing this helix. ⁵⁴

Genes Altered in Phyllodes Tumours

The true prevalence of mutations of alterations for genes (including MED12) is difficult to define due to the relative rarity of phyllodes tumours limiting cohort sizes, the use of a candidate gene approach in many studies focusing on likely altered genes based on the cohort composition, and different testing methodologies with a range of sensitivities and specificities. To estimate the overall frequency of alterations of individual genes, we identified published studies from PubMed using the search term “phyllodes molecular” and raw data extracted from relevant published papers and/or supplementary material associated with the papers. As the frequency of alterations is likely to be artifically skewed in reports with small numbers, a cut‐off of at least 50 cases tested was required for a gene to be included in the gene frequencies reported below.

When phyllodes tumours are regarded as a whole, the most frequently altered genes are MED12 (53%), TERT promoter region (pTERT, 47%), CDKN2A/CDKN2B (44%), MTAP (28%), RARA (20%), FLNA (19%), TP53 (17%), KMT2D (16%), MYC (15%), SETD2 (15%), NOTCH2 (13%), TSC2 (12%), NF1 (12%), ARID1B (11%), ZBED4 (11%), EGFR (11%), SYNE1 (10%), RB1 (10%), DNAH11 (10%), ATM (9%), USH2A (9%), PCLO (9%), NRAS (9%), PIK3CA (8%), ADAMTS18 (8%), BRAF (8%), BRCA2 (7%), NTRK (6%), PTEN (6%), ERBB3 (6%), PDGFRB (6%), STK11 (6%), KIT (6%), CHEK2 (5%) and RUNX1 (5%) ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ (see supplementary data).

The gene alterations consisted of single nucleotide variants alone (MED12, FLNA, SETD2, ARID1B, ZBED4, SYNE1, PCLO, NRAS, ADAMTS18, BRCA2, ERBB3, CHEK2, RUNX1), copy number alterations alone (MTAP, MYC, KIT), mutations and copy number alterations (CDKN2A/CDKN2B, RARA, KMT2D, NOTCH2, TSC2, NF1, EGFR, DNAH11, ATM, USH2A, PIK3CA, PTEN, PDGFRB), mutations and gene fusions (BRAF, STK11), as well as genes affected by mutations, copy number alterations and fusions (pTERT, TP53, RB1, NTRK). ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³

A single study has suggested a higher prevalence of NF1, KMT2D, and RB1 mutations in metastatic tumours compared with primary tumours, as well as molecular differences associated with metastatic site, such as the presence of pTERT promoter, MED12 and RB1 mutations in lung metastases but not at other sites, and more frequent NF1 and KMT2D mutations in non‐lung metastases compared with lung metastases, although these differences did not reach statistical significance. ²⁷ It is notable that liposarcomatous differentiation in malignant phyllodes tumours is not associated with MDM2 and CDK4 amplification in contrast to well‐differentiated liposarcoma and de‐differentiated liposarcoma. ⁵⁵

Molecular Events as the Basis of a Model of Phyllodes Tumour Pathogenesis

Phyllodes tumours are postulated to rise through two pathways, one associated with alterations of MED12 and the other independent of MED12. This hypothesis is based on the following recurrent genetic alterations observed in fibroepithelial tumours (Figures 1 and 2, supplementary data).

1. MED12 mutations are frequently present in fibroadenomas and phyllodes tumours suggesting that MED12 mutations are early events in the development of fibroepithelial tumours. ²⁸

2. The prevalence of MED12 mutations decreases with increasing grade of phyllodes tumours with 66%, 57% and 33% variants reported in benign, borderline and malignant phyllodes tumours, respectively, ²⁰ , ²² , ³¹ , ³³ , ³⁶ , ⁴⁰ . ³¹ , ³³ , ³⁶ , ³⁹

3. pTERT alterations are exceedingly rare in fibroadenomas, ²² , ²⁴ , ³⁴ are more prevalent in malignant (56%) and borderline phyllodes (59%) tumours compared with benign phyllodes (34%) tumours ²² , ³⁶ and can be acquired in recurrent phyllodes tumours. ²⁸

4. MED12 mutations are associated with pTERT alterations in phyllodes tumours. ³⁴ , ³⁶

5. There is greater genetic complexity with increasing grade of phyllodes tumour. ¹⁸ , ²² , ³¹ , ³² , ³⁶ , ⁵⁶

6. Alterations of cancer driver genes TP53, RB1, EGFR, CDKN2A/CDKN2B, NF1, PTEN, NRAS, MTOR, ARID1B, PIK3CA and ERBB4 are rare in fibroadenoma and more prevalent in malignant/borderline phyllodes tumours compared with benign phyllodes tumours. ²⁰ , ²² , ³¹ , ³² , ³⁶

7. MED12 wild‐type fibroepithelial lesions show more frequent TP53 and PIK3CA variants. ³⁶

Figure 1.

Molecular pathology of phyllodes tumours. Molecular alterations in phyllodes tumours differ with histological grade. Malignant tumours have greater complexity and increased numbers of altered genes including cancer driver genes; however, the prevalence of MED12 mutations is lower compared with benign and borderline tumours. * indicates genes with <50 cases tested in the published literature. ** indicates genes with no cases tested in the published literature.

Figure 2.

Potential pathways of phyllodes tumour pathogenesis. Phyllodes tumour can develop from MED12‐bearing fibroadenomas by development of pTERT alterations (so‐called “MED12 pathway”). Phyllodes tumours can also develop without a pre‐existing fibroadenoma (so‐called “MED12 independent pathway”). Progression of phyllodes tumours from benign to borderline and malignant phyllodes tumour can occur through accumulation of genetic abnormalities which may be present in a subclone of the original tumour.

On the basis of these observations, it is hypothesized that a proportion of phyllodes tumours arise from fibroadenomas. These phyllodes tumours harbour MED12 mutations and progress from fibroadenomas to phyllodes tumours by the gain of pTERT alterations. ³⁴ , ³⁵ pTERT alterations are postulated to enable tumours to progress through a greater number of cell cycles leading to stromal proliferation in phyllodes tumours and increasing the likelihood of developing cancer driver mutations. ²² In phyllodes tumours without pre‐existing MED12 mutations, the de novo development of pTERT alterations and other cancer‐related genes is an alternative pathway to neoplasia. ²² , ³⁴ , ³⁵ This model is further supported by the observation of higher frequency of MED12 mutations in phyllodes tumours with areas morphologically resembling fibroadenomas compared with phyllodes tumours without fibroadenoma‐like areas. ³⁵ While the frequency of pTERT mutations or amplifications is similar in phyllodes tumours regardless of the presence of fibroadenoma‐like areas, EGFR, RB1, TP53 and NF1 alterations were more common in phyllodes tumours without fibroadenoma‐like areas, reaching statistical significance in the case of EGFR alterations. ³⁵ However, these findings have not been replicated in a subsequent study of malignant phyllodes tumours with and without fibroadenoma‐like areas, ⁵⁷ and both studies are limited by small cohort sizes (n = 16).

Case studies have also supported a linear genetic progression from fibroadenomas to phyllodes tumour and in grade progression of phyllodes tumours whereaccumulation of additional mutations occuron the background of shared genetic alterations. ⁴⁴ , ⁵⁸ However, it has also been demonstrated that recurrent tumours do not necessarily harbour the same genetic alterations as their primary tumours, ⁵⁹ , ⁶⁰ and it has been hypothesized that grade progression in recurrences results from subclones that are not easily identified in the primary tumour. ⁵⁹ , ⁶⁰

Methylation Profiles

Methylation profiles of phyllodes tumours have been demonstrated to differ from those of normal breast tissue and non‐phyllodes breast tumours, ¹⁸ , ⁶¹ including differential methylation of genes relating to epithelial‐mesenchymal transition, KRAS signalling pathways and PRC2 targets. ⁶¹ In a small series of fibroepithelial lesions including 41 phyllodes tumours and 17 fibroadenomas, methylation profiles derived from the Illumina Infinium MethylationEPIC BeadChip did not show a clear distinction between fibroadenomas and phyllodes tumours or between benign, borderline and malignant phyllodes tumours. ¹⁸ However, using a much larger data set derived from the same methylation microarray and the Infinium HumanMethylation 450 BeadChip, including 33 breast tumours analysed by the authors and 817 breast tumours from publicly available databases, Meyer et al. ⁶¹ identified 53 differentially methylated regions between malignant and non‐malignant phyllodes tumours. The most significantly differentially methylated region encompassed the promoter and part of the gene body of HSD17B8, a gene involved in oestrogen metabolism, ⁶² and the degree of methylation was inversely correlated with gene expression, suggesting a role for methylation in regulating expression of this gene.

Gene Expression Profiles

Expression patterns of 105 breast cancer‐related genes have been demonstrated to differ between different types of fibroepithelial lesions. ⁶³ Not surprisingly, hypoxia related genes, including VEGFA, PNP, GAL, FLVCR2, NDRG1, FABP5 and DDIT4, and proliferation related genes, including MKI67, CDC20, CCNB1, CENPF, BIRC5 and NDC80, were highly expressed in malignant phyllodes tumours compared with fibroadenomas and benign phyllodes tumours. Epithelial and luminal related genes were more highly expressed in fibroadenomas compared with benign phyllodes tumours, juvenile fibroadenomas, borderline phyllodes tumours and malignant phyllodes tumours in decreasing order of expression, perhaps related to the increasing stromal composition. ⁶³ Most malignant phyllodes tumours have similar gene expression profiles to basal‐like/claudin‐low breast carcinomas. ⁶³ High expression of genes involved in the negative regulation of apoptosis, negative regulation of the cell cycle, and in blood vessel, morphogenesis was also observed in malignant phyllodes tumours. ⁶³ Likewise, Gatalica et al. ³⁸ observed overexpression of genes involved in angiogenesis in malignant phyllodes tumours and Li et al. ⁶⁴ also showed malignant phyllodes tumours were distinct from other fibroepithelial lesions in their expression of genes related to hypoxia and angiogenesis, cancer‐associated intracellular signalling cascades, apoptosis regulation, epigenetic regulation, matrix remodelling and metastasis. ⁶⁴ In contrast, fibroadenomas, cellular fibroadenomas and benign phyllodes tumours had generally similar gene expression profiles, with differences in gene expression between benign and borderline phyllodes tumours in only a small number of genes. ⁶⁴

Relevance of Molecular Alterations in Clinical Practice

Molecular Distinction of Phyllodes Tumour from Fibroadenoma and in Grading of Phyllodes Tumours

Using a 16‐gene panel developed from genomic profiling of breast fibroepithelial lesions ²⁰ to interrogate a large cohort of tumours, including 258 conventional fibroadenomas, 45 cellular fibroadenomas, 322 benign phyllodes tumours, 117 borderline phyllodes tumours and 54 malignant phyllodes tumours, Md Nasir et al. ³⁶ demonstrated a higher number of mutations in phyllodes tumours compared with fibroadenomas in line with other studies, ²⁰ , ²⁴ including mutations of pTERT, FLNA and SETD2. The presence of pTERT mutation was identified as the most sensitive and specific marker differentiating phyllodes tumours from fibroadenomas, similar to previous studies reporting the absence ²² or rarity ²⁴ , ³⁴ of pTERT alterations in fibroadenomas.

In another study using the same 16‐gene panel on core biopsies of 211 fibroepithelial lesions (167 fibroadenomas and 44 phyllodes tumours), 27 cases (12.8%) had molecular findings discordant with the original histological diagnosis. On histological review, the diagnosis of 5 cases was revised (2 cases of fibroadenoma reclassified as benign phyllodes tumour, 3 cases of phyllodes tumour reclassified as a fibroadenoma), amounting to 18.5% of the reviewed cases and 2.4% of the overall cohort. ²⁴

Ng et al. ⁶⁵ further interrogated the data obtained by Md Nasir et al. ³⁶ to investigate the feasibility of using molecular alterations to refine diagnoses in benign fibroepithelial lesions, whereby a diagnosis of benign phyllodes tumour would warrant complete excision. ⁸ There were no statistically significant differences in copy number variations between conventional fibroadenomas, cellular fibroadenomas and benign phyllodes tumours. While benign phyllodes tumours differed from the combined group of conventional and cellular fibroadenomas by having a higher prevalence of MED12, pTERT, RARA, FLNA, SETD2 and RB1 alterations, the clinical utility was limited by the relatively low incidence of non‐MED12 mutations. When histological review was undertaken due to the presence of cancer driver mutations (TP53, RB1, NF1, PTEN, PIK3CA, EGFR, BCOR, ERBB4, MAP3K1, IGF1R) or the presence of more than 2 mutations in non‐cancer driver genes (20% of the overall cohort), the original diagnosis was upgraded from fibroadenoma to benign phyllodes tumour or from benign phyllodes tumour to borderline phyllodes tumour in only 4.7% of the reviewed cases leading to a clinically significant change in diagnosis in only approximately 1% of the overall cohort.

Another study using a gene expression assay comprising five genes (ABCA8, APOD, CCL19, FN1 and PRAME) developed to distinguish between fibroadenomas and phyllodes tumours on formalin fixed, paraffin embedded (FFPE) core biopsies was found to correlate with histological diagnoses in 213 of 230 cases (92.6%), with a sensitivity of 82.9% and specificity of 94.7%. ⁶⁶

Nevertheless, the absence of a tight concordance between histological diagnosis and molecular findings suggests that routine molecular testing of fibroepithelial lesions would be generally of low diagnostic yield, including the detection of pTERT variants and the use of surrogates such as p53 and RB1 immunohistochemistry in differentiating between fibroadenoma and benign phyllodes tumours, where pTERT, TP53 and RB1 variants occur in 33%, <3% and <5% of benign phyllodes tumours, respectively (supplementary data). Thus, any testing currently available may not justify the additional resources required except in select cases where there is a diagnostic dilemma, such as demonstrated in a case report where the tumour had MED12 and pTERT mutations but no cancer driver alterations, resulting in a diagnosis of borderline phyllodes tumour being rendered over malignant phyllodes tumour. ⁶⁷ However, one might argue this is more of an academic exercise as it may not significantly influence patient management.

Molecular Alterations Aid Distinction between Phyllodes Tumours and Non‐fibroepithelial Tumours

The marked stromal proliferation of phyllodes tumour can raise the differential diagnoses of metaplastic carcinoma, fibromatosis or sarcoma when the typical architecture and epithelial component of phyllodes tumour are absent or sparse. While the distinction between phyllodes tumour and soft tissue tumours may not necessarily alter treatment, metaplastic carcinoma differs from these entities in terms of chemotherapy regimens and recommended surgical margin clearances and therefore the distinction from metaplastic carcinoma is of particular importance. ⁸ In this situation, demonstration of a MED12 variant is most useful. In contrast to phyllodes tumours, MED12 mutations are not identified in metaplastic breast carcinoma and sarcoma, ⁶⁸ and only rarely occur in fibromatosis. ³⁹ In addition, RARA and pTERT mutations are not seen in metaplastic carcinoma. ⁶⁹ The utility of molecular testing in cases where there is morphological and immunophenotypical overlap between phyllodes tumours, metaplastic carcinoma and breast sarcoma has been demonstrated in case reports with the presence of MED12 and/or pTERT variants establishing the diagnosis of malignant phyllodes tumour. ⁶⁷ , ⁷⁰ , ⁷¹

Molecular Alterations as a Marker of Recurrence

Several studies have suggested that molecular alterations can be determinants of disease recurrence in phyllodes tumour in addition to traditional parameters such as tumour grade and margin status. In a small group of 19 patients with phyllodes tumours, the presence of high‐level amplifications and homozygous deletions, including in regions encompassing MDM4, RAF1, EGFR, PDZD2 and CDKN2A, were only detected in tumours of patients who subsequently developed recurrent disease (local recurrence, metastasis or death) including histologically benign phyllodes tumours. ⁷² The presence of MED12 mutations has been associated with improved disease‐free survival in a cohort of 97 phyllodes tumours with a median follow up of 30 months, ³³ but this has not been replicated in other studies. ³⁹ , ⁵⁷ Tsang et al. ³¹ in a cohort of 49 benign, 25 borderline and 14 malignant phyllodes tumours, reported the presence of RARA mutation was associated with recurrent disease but this was no longer statistically significant when cases with positive surgical margins were excluded. However, recurrence was significantly associated with the absence of epigenetic pathway alterations (defined as variants of ARID1B, ATRX, BCOR, CHD4, CHD8, KMT2C, KMT2D, SETD2, SMARCA4 and TET3) regardless of surgical margin status. In a retrospective study including 33 phyllodes tumours, gene expression assessed by the PAM50 risk of recurrence (ROR) score was more predictive of clinical outcome than morphological classification of phyllodes tumours; however, the number of recurrent events was limited to 7 cases. ⁶³

Molecular Alterations Identify Potential Targets for Therapy

Although MED12 is involved in oestrogen mediated pathways, the presence of MED12 alterations is not correlated with ER protein expression. ⁷³ In addition, ER alpha, the well‐established biomarker indicating potential response to antioestrogen treatment, is only expressed in the epithelial component and not in the neoplastic stromal component of phyllodes tumours, ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶ while ER beta is expressed in the stromal component. ⁷³ , ⁷⁵ The therapeutic significance of ER beta receptor expression is unknown due to uncertainties regarding its biological function in breast cancer, the presence of multiple isoforms and non‐standardized methodologies for detection. ⁷⁷ , ⁷⁸ , ⁷⁹

As systemic treatment options for advanced phyllodes tumours are limited, ⁸⁰ it is hoped that detection of molecular alterations in phyllodes tumour would allow the use of targeted therapies as in other tumour types. ⁸¹ , ⁸² , ⁸³ Targetable variants identified in phyllodes tumours include EGFR amplifications and CDKN2A/CDKN2B deletions, ¹⁸ , ²⁰ , ²³ , ³⁵ variants of AKT1, EGFR, ERBB2, ERBB3, FGFR1, FGFR2, FGFR3, BRAF, NF1, NRAS, KRAS, PIK3CA, BRCA1, BRCA2 and PTEN, ²⁰ , ²³ , ²⁵ , ²⁶ , ³⁵ , ⁸⁴ , ⁸⁵ and fusions involving NTRK, BRAF and FGFR3. ²⁵ , ²⁷ A very small proportion of malignant phyllodes tumours have a high tumour mutational burden which may indicate response to immunotherapy. ²⁶ Documented incidences of successful targeted therapy include a patient treated with larotrectinib for a pathogenic NTRK1 fusion, ²⁷ complete pathological response in a lung metastasis following treatment with a personalized multi‐epitope peptide neoantigen nano‐vaccine developed based on whole exome sequencing, RNA sequencing and new antigen prediction, ⁸⁶ and a patient with a high tumour mutational burden demonstrating response to pembrolizumab. ⁸⁷ In addition, certain genetic alterations may indicate lack of response to treatment; for example, in a retrospective study, the presence of CDKN2A and pTERT alterations was associated with reduced sensitivity to chemotherapy and CDKN2B alterations associated with absence of response to chemotherapy, although these relationships did not reach statistical significance. ⁸⁷ Nevertheless, we have seen a clinical response to pazopanib in a patient with metastatic malignant phyllodes tumour with high VEGF expression that was likely driven by a pTERT variant, demonstrating its downstream effects of signalling on hypoxia and angiogenesis. ⁸⁸ Models such as patient‐derived xenografts or organoids as with other tumours might help in defining optimal therapies for patients, although limited by the known heterogeneity. ⁸⁹

Conclusions

Over a decade of molecular research on phyllodes tumours has provided insights into their pathogenesis. While molecular testing is not currently part of routine pathological assessment, molecular profiling can be a useful adjunct in the clinically important distinction from non‐phyllodes tumours such as metaplastic carcinoma through the detection of MED12, pTERT and RARA variants, and to refine histological grading by identification of cancer driver genes in very selected cases. As these genes are not generally targeted in routine molecular panels for common tumours such as lung carcinomas, access to such testing may be limited in many pathology laboratories.

Genetic testing in patients with phyllodes tumour is also a valid consideration given the similar prevalence of germline mutations of autosomal dominant cancer susceptibility genes, including TP53, to the breast carcinoma population. ¹⁴

Recently, less stringent criteria for the diagnosis of malignant phyllodes tumour have been proposed based on the malignant behaviour of some histologically borderline tumours, ¹² which will be incorporated into the 6^th Edition of the WHO Breast Tumour Classification Blue Book. It will be interesting to note the correlation of molecular alterations with revised histological grade, especially cancer driver genes alterations which are present in a proportion of tumours currently classified as borderline phyllodes tumours.

Most studies have used a candidate gene approach to detect known DNA sequence variants and potentially structural variants. Therefore, information outside of these targets, including epigenetic and expression data, is limited. Agnostic approaches analysing the whole genome/transcriptome may provide a more comprehensive picture of the molecular changes in phyllodes tumours as technologies become more accessible in the future, including mutational signatures that might shed light on the aetiology and cell of origin. Multi‐institution research collaborations will help overcome the relatively low frequency of phyllodes tumours, especially malignant phyllodes tumours, and lead to more robust data regarding the link between molecular alterations, response to targeted therapies and clinical outcomes.

Author contributions

JMBP performed literature review, wrote and revised the manuscript. KLG wrote and reviewed the manuscript. PHT and SBF conceptualized, wrote and reviewed the manuscript. All authors approve the final version of the manuscript, figures and supplementary material for publication.

Funding information

SBF is in receipt of a National Health and Medical Research Council grant: GNT1193630.

Conflict of interests

The authors declare no conflicts of interest in relation to the subject matter discussed in this manuscript.

Supporting information

Data S1. Prevalence of alterations by gene and phyllodes tumour grade in published studies.

Data availability statement

The authors will provide all data on request.