Inflammatory response-based subtyping and potential therapeutic strategies for triple-negative breast cancer

Abstract

Objective: Inflammatory response plays a crucial role in the development and treatment of cancer. However, the role of inflammatory response in triple-negative breast cancer (TNBC) remains unclear. Based on the heterogeneity of the inflammatory response, we classified TNBC, elucidated its subtype features, and revealed potential therapeutic strategies. Methods: We established inflammatory response subtyping based on the RNA sequencing data of TNBCs derived from a cohort at the Fudan University Shanghai Cancer Center (FUSCC). Next, we explored the features and potential therapeutic strategies for each subgroup by analyzing transcriptome data. Using a machine-learning method, we validated and generalized the TNBC inflammatory response subtypes in an external dataset. Results: A total of 360 TNBC samples and 88 normal tissues were collected from a cohort at FUSCC. Patients with TNBC were divided into four inflammatory response groups (IRGs) based on the expression of inflammatory response genes: high inflammatory response gene expression with pronounced pyroptosis phenotype and high immune cell infiltration (IRG 1), low inflammatory response gene expression and low immune cell infiltration (IRG 2), ITGB8 specific inflammatory response with a predominant proliferation phenotype (IRG 3), and low M1/M2 ratio with a marked angiogenesis phenotype (IRG 4). Relapse-free survival (RFS) was better in IRG 1 and 2 and worse in IRG 3 and 4. Owing to their poor prognosis, we mainly focused on IRG 3 and IRG 4 to investigate potential treatment strategies. ITGB8 was highly expressed in IRG 3; thus, targeting ITGB8 may be a potential therapeutic strategy for patients in IRG 3. IRG 4 had a lower M1/M2 ratio and a marked angiogenesis phenotype; therefore, therapeutic strategies, such as anti-angiogenesis or M2 to M1 repolarization of macrophages, could be recommended for these patients. Additionally, we validated and generalized the TNBC inflammatory response subtyping in an external dataset using a machine-learning method. Conclusion: TNBC patients with different inflammatory response subtypes have different characteristics and may need subtype-specific treatment strategies.