将K-M生存分析显著的TCGA数据库的肿瘤类型包括UVM、LGG、PRAD、LIHC、LUAD、KIRC、HNSC、MESO的基因表达数据和临床数据输入DeepSurv模型,共4360个样本。将样本分为训练集(80%,3488个样本)和测试集(20%,872个样本)进行预测。参数设置3层隐藏层,每层由全连接层、Batch normalization、激活层(ReLU)、dropout层堆叠组成,学习率为1e−2,L2正则化参数为1e−4。采用适应性矩估计(Adaptive Moment Estimation, Adam)算法,根据每个参数的梯度情况来动态调整学习率,它维护了每个参数两个额外的变量:一阶矩估计和二阶矩估计,用来估计梯度的均值和方差。同时,采用Cox比例风险回归模型和随机森林生存分析模型与DeepSurv进行对比,Cox比例风险模型基于风险比例假设,即各个预测因子对风险的影响在时间上是恒定的。Cox风险回归模型如下:
Figure 1. Expression of necroptosis genes in scRNA-seq pan-cancer data. (a) Cellular annotation visualization of NSCLC, downscaled plots of high and low groupings after scoring of necroptosis genes, and mountain range plots of different cell cluster scores. (b) Cellular annotation visualization of COAD, downscaled plots of high and low groupings after scoring of necroptosis genes, and maps of different cell cluster scores mountains. (c) Cellular annotation visualization of ESCA, downscaled plots of high and low groupings after scoring of necroptosis genes, and maps of different cell cluster scores mountains. (d) Cellular annotation visualization of THCA, downscaled plots of high and low grouping after necroptosis gene scoring, and different cell cluster scoring mountain range plots--图1. scRNA-seq泛癌数据中坏死性凋亡基因的表达情况。(a) NSCLC的细胞注释可视化,坏死性凋亡基因打分后高低分组的降维图,以及不同细胞簇得分山峦图。(b) COAD的细胞注释可视化,坏死性凋亡基因打分后高低分组的降维图,以及不同细胞簇得分山峦图。(c) ESCA的细胞注释可视化,坏死性凋亡基因打分后高低分组的降维图,以及不同细胞簇得分山峦图。(d) THCA的细胞注释可视化,坏死性凋亡基因打分后高低分组的降维图,以及不同细胞簇得分山峦图--Figure 2. Machine learning model to identify pan-cancer and necroptosis-related genes. (a)~(c) LASSO regression model screening features with model AUC value of 0.74; (d)~(f) Random forest model based on Boruta feature selection with model AUC value of 0.78; (g) SVM-RFE model screening 56 features; (h) Venn diagram of LASSO, RF and SVM-RFE model screening features taken as intersection--图2. 机器学习模型识别泛癌与坏死性凋亡相关基因。(a)~(c) LASSO回归模型筛选特征,模型AUC值为0.74;(d)~(f) 基于Boruta特征选择的随机森林模型,模型AUC值为0.78;(g) SVM-RFE模型筛选出56个特征;(h) 将LASSO、RF和SVM-RFE模型筛选特征取交集的韦恩图--3.3. 在泛癌症水平上的坏死性凋亡景观的描述
Figure 3. Differential expression of necroptosis genes in normal and cancer samples. (a) Differential expression violin plots of normal and cancer samples in 33 cancer types; (b) Heatmap of differential GSVA scores in normal and tumor samples--图3. 坏死性凋亡基因在正常样本和癌症样本的表达差异。(a) 正常样本和癌症样本在33种癌症类型种的差异表达小提琴图;(b) 正常样本和肿瘤样本的GSVA评分差异热图--Figure 4. NCPS.Sig gene set enrichment analysis. (a-b) GO enrichment analysis; (c) GSEA analysis of gene sets in the apoptotic pathway--图4. NCPS.Sig基因集富集分析。(a-b) GO富集分析;(c) GSEA分析基因集在凋亡通路的表现--图4. NCPS.Sig基因集富集分析。(a-b) GO富集分析;(c) GSEA分析基因集在凋亡通路的表现Figure 4. NCPS.Sig gene set enrichment analysis. (a-b) GO enrichment analysis; (c) GSEA analysis of gene sets in the apoptotic pathway--图4. NCPS.Sig基因集富集分析。(a-b) GO富集分析;(c) GSEA分析基因集在凋亡通路的表现--图4. NCPS.Sig基因集富集分析。(a-b) GO富集分析;(c) GSEA分析基因集在凋亡通路的表现
Figure 5. Tumor immune properties correlated with necroptosis score. (a-b) Radar plots of the correlation of NCPS.Sig score with TMB and MSI; (c) Correlation of NCPS.Sig score with 22 immune cells in pan-cancer. Red color represents positive correlation, blue color represents negative correlation, and larger points represent higher correlation. p-value < 0.05 is *, p-value < 0.01 is **, and p-value < 0.001 is ***--图5. 与坏死性凋亡评分相关的肿瘤免疫特性。(a-b) NCPS.Sig评分与TMB和MSI的相关性雷达图;(c) NCPS.Sig得分与泛癌的22种免疫细胞相关性。红色代表正相关,蓝色代表负相关,点越大代表相关性越高。p值 < 0.05为*, p值 < 0.01为**,p值 < 0.001为***--图5. 与坏死性凋亡评分相关的肿瘤免疫特性。(a-b) NCPS.Sig评分与TMB和MSI的相关性雷达图;(c) NCPS.Sig得分与泛癌的22种免疫细胞相关性。红色代表正相关,蓝色代表负相关,点越大代表相关性越高。p值 < 0.05为*, p值 < 0.01为**,p值 < 0.001为***Figure 5. Tumor immune properties correlated with necroptosis score. (a-b) Radar plots of the correlation of NCPS.Sig score with TMB and MSI; (c) Correlation of NCPS.Sig score with 22 immune cells in pan-cancer. Red color represents positive correlation, blue color represents negative correlation, and larger points represent higher correlation. p-value < 0.05 is *, p-value < 0.01 is **, and p-value < 0.001 is ***--图5. 与坏死性凋亡评分相关的肿瘤免疫特性。(a-b) NCPS.Sig评分与TMB和MSI的相关性雷达图;(c) NCPS.Sig得分与泛癌的22种免疫细胞相关性。红色代表正相关,蓝色代表负相关,点越大代表相关性越高。p值 < 0.05为*, p值 < 0.01为**,p值 < 0.001为***--图5. 与坏死性凋亡评分相关的肿瘤免疫特性。(a-b) NCPS.Sig评分与TMB和MSI的相关性雷达图;(c) NCPS.Sig得分与泛癌的22种免疫细胞相关性。红色代表正相关,蓝色代表负相关,点越大代表相关性越高。p值 < 0.05为*, p值 < 0.01为**,p值 < 0.001为***
Figure 6. Prognostic value of the NCPS.Sig score. (a) Forest plot of pan-cancer COX analysis of NCPS.Sig. (b)~(i) Survival analysis plot of NCPS.SIG scores in pan-cancer--图6. 坏死性凋亡评分的预后价值。(a) NCPS.Sig的泛癌COX分析森林图。(b)~(i) NCPS.Sig打分在泛癌的生存分析曲线图--图6. 坏死性凋亡评分的预后价值。(a) NCPS.Sig的泛癌COX分析森林图。(b)~(i) NCPS.Sig打分在泛癌的生存分析曲线图Figure 6. Prognostic value of the NCPS.Sig score. (a) Forest plot of pan-cancer COX analysis of NCPS.Sig. (b)~(i) Survival analysis plot of NCPS.SIG scores in pan-cancer--图6. 坏死性凋亡评分的预后价值。(a) NCPS.Sig的泛癌COX分析森林图。(b)~(i) NCPS.Sig打分在泛癌的生存分析曲线图--图6. 坏死性凋亡评分的预后价值。(a) NCPS.Sig的泛癌COX分析森林图。(b)~(i) NCPS.Sig打分在泛癌的生存分析曲线图
Figure 7. Necroptosis-associated pan-cancer prognostic modeling. ROC plots of the NCPS.Sig gene set in the DeepSurv model, the RF survival model, and the COX risk regression model--图7. 坏死性凋亡相关的泛癌预后模型。NCPS.Sig基因集在DeepSurv模型、RF生存模型以及COX风险回归模型的ROC曲线图--4. 讨论
References
Bedoui, S., Herold, M.J. and Strasser, A. (2020) Emerging Connectivity of Programmed Cell Death Pathways and Its Physiological Implications. Nature Reviews Molecular Cell Biology, 21, 678-695. >https://doi.org/10.1038/s41580-020-0270-8
Qin, X., Ma, D., Tan, Y., Wang, H. and Cai, Z. (2019) The Role of Necroptosis in Cancer: A Double-Edged Sword? Biochimica et Biophysica Acta (BBA)-Reviews on Cancer, 1871, 259-266. >https://doi.org/10.1016/j.bbcan.2019.01.006
Mariño, G., Niso-Santano, M., Baehrecke, E.H. and Kroemer, G. (2014) Self-Consumption: The Interplay of Autophagy and Apoptosis. Nature Reviews Molecular Cell Biology, 15, 81-94. >https://doi.org/10.1038/nrm3735
Frank, D. and Vince, J.E. (2018) Pyroptosis versus Necroptosis: Similarities, Differences, and Crosstalk. Cell Death&Differentiation, 26, 99-114. >https://doi.org/10.1038/s41418-018-0212-6
Kono, H. and Rock, K.L. (2008) How Dying Cells Alert the Immune System to Danger. Nature Reviews Immunology, 8, 279-289. >https://doi.org/10.1038/nri2215
Gong, Y., Fan, Z., Luo, G., Yang, C., Huang, Q., Fan, K., et al. (2019) The Role of Necroptosis in Cancer Biology and Therapy. Molecular Cancer, 18, Article No. 100. >https://doi.org/10.1186/s12943-019-1029-8
Seifert, L., Werba, G., Tiwari, S., Giao Ly, N.N., Alothman, S., Alqunaibit, D., et al. (2016) The Necrosome Promotes Pancreatic Oncogenesis via CXCL1 and Mincle-Induced Immune Suppression. Nature, 532, 245-249. >https://doi.org/10.1038/nature17403
Ruan, J., Mei, L., Zhu, Q., Shi, G., and Wang, H. (2015). Mixed Lineage Kinase Domain-Like Protein Is a Prognostic Biomarker for Cervical Squamous Cell Cancer. International Journal of Clinical and Experimental Pathology, 8, 15035-15038.
Zhu, F., He, L., Peng, K., Liu, Y. and Xiong, J. (2013) Low Expression of Mixed Lineage Kinase Domain-Like Protein Is Associated with Poor Prognosis in Ovarian Cancer Patients. OncoTargets and Therapy, 6, 1539-1543. >https://doi.org/10.2147/ott.s52805
Zhou, Z., Wu, J., Ma, W., Dong, F. and Wang, J. (2022) Pan‐Cancer Analyses of Necroptosis‐Related Genes as a Potential Target to Predict Immunotherapeutic Outcome. Journal of Cellular and Molecular Medicine, 27, 204-221. >https://doi.org/10.1111/jcmm.17634
Lei, Y., Tang, R., Xu, J., Wang, W., Zhang, B., Liu, J., et al. (2021) Applications of Single-Cell Sequencing in Cancer Research: Progress and Perspectives. Journal of Hematology&Oncology, 14, Article No. 91. >https://doi.org/10.1186/s13045-021-01105-2
Lv, Z., Han, J., Li, J., Guo, H., Fei, Y., Sun, Z., et al. (2022) Single Cell RNA-Seq Analysis Identifies Ferroptotic Chondrocyte Cluster and Reveals TRPV1 as an Anti-Ferroptotic Target in Osteoarthritis. eBioMedicine, 84, Article 104258. >https://doi.org/10.1016/j.ebiom.2022.104258
Li, X., Su, L., Chen, W., Liu, H., Zhang, L., Shen, Y., et al. (2022) Clinical Implications of Necroptosis Genes Expression for Cancer Immunity and Prognosis: A Pan-Cancer Analysis. Frontiers in Immunology, 13, Article 882216. >https://doi.org/10.3389/fimmu.2022.882216
Katzman, J.L., Shaham, U., Cloninger, A., Bates, J., Jiang, T. and Kluger, Y. (2018) DeepSurv: Personalized Treatment Recommender System Using a Cox Proportional Hazards Deep Neural Network. BMC Medical Research Methodology, 18, Article No. 24. >https://doi.org/10.1186/s12874-018-0482-1
Wu, Y., Huang, Y., Zhou, C., Wang, H., Wang, Z., Wu, J., et al. (2022) A Novel Necroptosis-Related Prognostic Signature of Glioblastoma Based on Transcriptomics Analysis and Single Cell Sequencing Analysis. Brain Sciences, 12, Article 988. >https://doi.org/10.3390/brainsci12080988
Palmeri, M., Mehnert, J., Silk, A.W., Jabbour, S.K., Ganesan, S., Popli, P., et al. (2022) Real-World Application of Tumor Mutational Burden-High (TMB-High) and Microsatellite Instability (MSI) Confirms Their Utility as Immunotherapy Biomarkers. ESMO Open, 7, Article 100336. >https://doi.org/10.1016/j.esmoop.2021.100336
Luo, W., Xiang, W., Gan, L., Che, J., Li, J., Wang, Y., et al. (2022) Bulk and Single-Cell Transcriptome Profiling Reveal Necroptosis-Based Molecular Classification, Tumor Microenvironment Infiltration Characterization, and Prognosis Prediction in Colorectal Cancer. Journal of Translational Medicine, 20, Article No. 235. >https://doi.org/10.1186/s12967-022-03431-6
Moujalled, D., Strasser, A. and Liddell, J.R. (2021) Molecular Mechanisms of Cell Death in Neurological Diseases. Cell Death&Differentiation, 28, 2029-2044. >https://doi.org/10.1038/s41418-021-00814-y
Song, J., Zhang, X., Yin, Y., Guo, M., Zhao, X., Wang, L., et al. (2023) Loss of RPA1 Impairs Peripheral T Cell Homeostasis and Exacerbates Inflammatory Damage through Triggering T Cell Necroptosis. Advanced Science, 10, Article 2206344. >https://doi.org/10.1002/advs.202206344
Park, H., Kim, H., Park, S., Hwang, S., Hong, S.M., Park, S., et al. (2021) RIPK3 Activation Induces TRIM28 Derepression in Cancer Cells and Enhances the Anti-Tumor Microenvironment. Molecular Cancer, 20, Article No. 107. >https://doi.org/10.1186/s12943-021-01399-3
Seong, D., Jeong, M., Seo, J., Lee, J., Hwang, C.H., Shin, H., et al. (2020) Identification of MYC as an Antinecroptotic Protein That Stifles RIPK1-RIPK3 Complex Formation. Proceedings of the National Academy of Sciences, 117, 19982-19993. >https://doi.org/10.1073/pnas.2000979117
He, C., Liu, Y., Huang, Z., Yang, Z., Zhou, T., Liu, S., et al. (2021) A Specific RIP3
+Subpopulation of Microglia Promotes Retinopathy through a Hypoxia-Triggered Necroptotic Mechanism. Proceedings of the National Academy of Sciences, 118, e2023290118. >https://doi.org/10.1073/pnas.2023290118
Han, X., Li, B., Bao, J., Wu, Z., Chen, C., Ni, J., et al. (2022) Endoplasmic Reticulum Stress Promoted Acinar Cell Necroptosis in Acute Pancreatitis through CathepsinB-Mediated AP-1 Activation. Frontiers in Immunology, 13, Article 968639. >https://doi.org/10.3389/fimmu.2022.968639
Sethi, J.K. and Hotamisligil, G.S. (2021) Metabolic Messengers: Tumour Necrosis Factor. Nature Metabolism, 3, 1302-1312. >https://doi.org/10.1038/s42255-021-00470-z
Garrido-Martin, E.M., Mellows, T.W.P., Clarke, J., Ganesan, A., Wood, O., Cazaly, A., et al. (2020) M1
hotTumor-Associated Macrophages Boost Tissue-Resident Memory T Cells Infiltration and Survival in Human Lung Cancer. Journal for ImmunoTherapy of Cancer, 8, e000778. >https://doi.org/10.1136/jitc-2020-000778
Urbano, P.C.M., Aguirre-Gamboa, R., Ashikov, A., van Heeswijk, B., Krippner-Heidenreich, A., Tijssen, H., et al. (2018) TNF-α-Induced Protein 3 (TNFAIP3)/A20 Acts as a Master Switch in TNF-Α Blockade-Driven IL-17A Expression. Journal of Allergy and Clinical Immunology, 142, 517-529. >https://doi.org/10.1016/j.jaci.2017.11.024