目的：由于缺乏有效的治疗手段，三阴性乳腺癌（TNBC）的预后非常差。代谢重编程是肿瘤发生、癌症诊断、预后和治疗的重要特征。对TNBC中的代谢模式进行分类对于对抗异质性和靶向治疗至关重要。方法：将来自TCGA 的115例TNBC患者合并成一个虚拟队列，并通过其他验证集进行验证，发现有差异表达的基因（DEGs）。为了识别可靠的代谢特征，我们对五个独立数据集采用相同的程序进行验证，以验证已鉴定的TNBC亚型，这些亚型在预后、代谢特征、免疫浸润、临床特征、体细胞突变和药物敏感性方面有所不同。结果：总体而言，TNBC可以分为两个代谢上有明显差异的亚型。C1 有免疫检查点基因表达高，免疫和基质评分高，对PD-1抑制剂治疗敏感。另一方面，C2代谢途径涉及到碳水化合物、脂质和氨基酸代谢有很大的变异。更重要的是，C2缺乏免疫标志物，病理分期晚，免疫浸润少，预后差。有趣的是，C2 PIK3CA、KMT2D 和 KMT2C 的突变频率较高，并且显示了 PI3K 和血管生成通路的显著激活。作为最终的结果，我们创建了一个100个基因分类器，可可靠地区分TNBC亚型，而AKR1B10则是C2亚型的潜在生物标志物。结论：总之，我们鉴定了两个具有不同代谢表型的亚型，为TNBC异质性提供了新的见解，并为治疗策略提供了理论基础。版权所有 © 2023 Li, Wu, Zhuang, Geng, Zeng, Wang, Wang, Ruan, Zheng, Liu 和 Gao。

Objective: Due to a lack of effective therapy, triple-negative breast cancer (TNBC) is extremely poor prognosis. Metabolic reprogramming is an important hallmark in tumorigenesis, cancer diagnosis, prognosis, and treatment. Categorizing metabolic patterns in TNBC is critical to combat heterogeneity and targeted therapeutics. Methods: 115 TNBC patients from TCGA were combined into a virtual cohort and verified by other verification sets, discovering differentially expressed genes (DEGs). To identify reliable metabolic features, we applied the same procedures to five independent datasets to verify the identified TNBC subtypes, which differed in terms of prognosis, metabolic characteristics, immune infiltration, clinical features, somatic mutation, and drug sensitivity. Results: In general, TNBC could be classified into two metabolically distinct subtypes. C1 had high immune checkpoint genes expression and immune and stromal scores, demonstrating sensitivity to the treatment of PD-1 inhibitors. On the other hand, C2 displayed a high variation in metabolism pathways involved in carbohydrate, lipid, and amino acid metabolism. More importantly, C2 was a lack of immune signatures, with late pathological stage, low immune infiltration and poor prognosis. Interestingly, C2 had a high mutation frequency in PIK3CA, KMT2D, and KMT2C and displayed significant activation of the PI3K and angiogenesis pathways. As a final output, we created a 100-gene classifier to reliably differentiate the TNBC subtypes and AKR1B10 was a potential biomarker for C2 subtypes. Conclusion: In conclusion, we identified two subtypes with distinct metabolic phenotypes, provided novel insights into TNBC heterogeneity, and provided a theoretical foundation for therapeutic strategies.Copyright © 2023 Li, Wu, Zhuang, Geng, Zeng, Wang, Wang, Ruan, Zheng, Liu and Gao.