背景:皮肤T细胞淋巴瘤(CTCL)是高度异质性的,其预后与疾病分期密切相关。肿瘤微环境(TME)是肿瘤组织的重要组成部分,驱动癌细胞生长,programming,和转移。然而,TME在CTCL中的诊断价值尚未得到深入研究。迄今为止,没有一项研究对CTCL中TME的意义进行了综合评价.
方法:使用基于大量RNA测序数据的xCell方法,我们在126例患者中推断了TME中的免疫细胞分数,并评估了免疫细胞的预后重要性.进行共有聚类以确定TME亚型并表征每个亚型的转录组。基于TME亚型,我们使用随机森林算法和逻辑回归建立了疾病进展模型。使用另外的49名患者队列检查模型的功效。最后,我们在16例患者队列中使用免疫化学在蛋白质水平验证了我们的发现.
结果:晚期CTCL患者总体上比早期患者具有更活跃的免疫力。随机森林算法揭示了免疫细胞CD4、巨噬细胞、和树突状细胞(DC)是最有效的预后预测因子。因此,我们基于这些免疫细胞使用逻辑回归构建了风险模型.TME评分可用于有效预测三个数据集中的疾病状况,AUC分别为0.9414、0.7912和0.7665。蛋白质水平的免疫组织化学显示,辅助性T细胞和巨噬细胞标志物CD4和CD68可以成功区分患者的不同CTCL分期,而DC标记langerin显示随疾病进展无变化。
结论:我们发现晚期CTCL与活跃的免疫微环境有关,免疫特征CD4和CD68在预测CTCL疾病进展方面显示出相对较高的准确性。
BACKGROUND: Cutaneous T-cell lymphoma (CTCL) is highly heterogeneous, and its prognosis is closely related to the disease stage. The tumor microenvironment (TME) is an important component of tumor tissue, driving cancer cell growth, progression, and metastasis. However, the diagnostic value of TME in CTCL has not yet been studied in-depth. To date, no study has performed a comprehensive evaluation of the significance of the TME in CTCL.
METHODS: Using xCell methods based on bulk RNA sequencing data, we inferred immune cell fraction in the TME in 126 patients and assessed the prognostic importance of immune cells. Consensus clustering was performed to determine the TME subtypes and characterize the transcriptome of each subtype. Based on the TME subtypes, we established the disease progression model using random forest algorithms and logistic regression. The efficacy of the model was examined using an additional 49-patient cohort. Finally, we validated our finding at the protein level using immunochemistry in a 16-patient cohort.
RESULTS: Patients with advanced CTCL presented with a more active immunity overall than those with early stage. Random forest algorithms revealed that the immune cells CD4, macrophages, and dendritic cells (DCs) were the most effective prognosis predictors. Therefore, we constructed a risk model using logistic regression based on these immune cells. The TME score could be used to effectively predict disease conditions in three datasets with the AUC of 0.9414, 0.7912, and 0.7665, respectively. Immunochemistry at the protein level revealed that helper T cells and the macrophage markers CD4 and CD68 could successfully distinguish different CTCL stages in patients, whereas the DC marker langerin showed no change with disease progression.
CONCLUSIONS: We found advanced-stage CTCL was associated with an active immune microenvironment, and the immune signatures CD4 and CD68 showed a relatively high accuracy in predicting CTCL disease progression.