Background: Phyllodes tumours (PT) are rare, fibroepithelial lesions that account for less than 1% of all breast cancers. PTs are classified as benign, borderline, or malignant, which have poor long-term prognosis[1]. However, a lack of understanding of PT biology makes it difficult to accurately categorise PTs, particularly differentiating between borderline and malignant. PTs also present other clinical challenges, with benign PTs often being misdiagnosed as fibroadenoma[2], while malignant PTs are difficult to distinguish from spindle-cell metaplastic breast cancer and show a rare, but fascinating capacity for multi-lineage sarcoma differentiation[3]. Given that aberrations in DNA methylation occur early in tumorigenesis, methylation profiling could aid in characterisation, risk stratification and uncovering novel biology of this rare, understudied cancer.
Aims: We aim to characterise the DNA methylation landscape of PTs and identify biomarkers that can improve classification of PT disease risk as well as delineate fibroadenoma and metaplastic breast cancer from phyllodes tumours.
Methods: We performed methylome profiling using the MethylationEPIC array on diagnostic PT patient biopsies (n=29). Samples from each risk category were defined by the Singapore General Hospital classification and compared to discover differentially methylated regions (DMRs). Samples were also compared to multiple related cancer types (n=409) to define the methylation landscape of PTs and scRNA-seq was performed to interrogate functional significance of these changes.
Results: We developed a methylation signature which provides insight into the biology and cellular composition of PTs. We also identified loci that distinguish benign PTs from fibroadenomas, and malignant PTs from metaplastic breast cancer. Finally, we identified DMRs (adj.P<0.05) delineating malignant PTs from all other samples and validated these findings in an independent cohort.
Conclusion: Our study is the first detailed characterisation of the PT methylome. We have identified methylation signatures associated with disease risk that have potential to better inform clinical decisions.