Prognostic value of FGFR2 in ER-positive breast cancer is influenced by the profile of stromal gene expression: an in silico analysis based on TCGA data

Introduction: Fibroblast growth factor receptor 2 (FGFR2) activation is associated with endocrine therapy resistance in luminal breast cancer (BC) in vitro, but clinical evidence remains inconsistent. Given the role of FGFRs in mediating tumour microenvironment (TME) interactions, the prognostic value of FGFR2 may depend on the stromal component. This study aimed to validate the association between FGFR-related profile of the stroma and FGFR2 prognostic value in oestrogen receptor-positive invasive ductal carcinoma (IDC). Material and methods: An in silico gene expression analysis identified 12 stromal factors (FAP, CXCL12, PDGFRA, COL1A1, HSPG2, CCL2, MMP14, S100A4, MMP9, PDGFA, MCAM, IL6) forming an "FGFR-related profile of the stroma". A cohort of 257 ER+ IDC patients from The Cancer Genome Atlas (TCGA) was analysed. Tumours were clustered using k-means based on stromal gene expression, and Cox proportional hazards regression models were used to assess the association between FGFR2 and overall survival (OS). Results: Two clusters of ER+ IDC tumours were identified based on the stromal gene expression profile. While both clusters had similar tumour stages and hormone receptor statuses, multivariable analysis adjusted for clinical factors revealed a significant association between FGFR2 expression and cluster assignment. In Cluster I (high expression of stromal genes), high FGFR2 was linked to poor prognosis, whereas in Cluster II (low expression), high FGFR2 indicated favourable prognosis. FGFR1, FGFR3, and FGFR4 showed no significant prognostic value. Conclusions: Stromal profiles modulate the prognostic significance of FGFR2 in luminal breast carcinoma, highlighting the importance of TME profiling for biomarker assessment and explaining inconsistencies in FGFR2 studies.