Expression levels of ICDRGs in LUSC
504 normal and colorectal cancer tissue samples' transcriptomes and clinical data were acquired from TCGA. This study includes 452 colorectal cancer patients with clinical information and gene expression profiles. Wilcoxen evaluated the expression of 28 ARGs. (|log2(FC)|, Pvalue < 0.05). Finally, 14 ICDRGs with high expression in tumors were obtained, and the expression correlation of ICDRGs was studied according to the expression level of ICDRGs in tumors (Fig. 1A). Based on the expression levels of survival-related ICDRGs in the TCGA database, two distinct regulatory patterns were identified using an unsupervised clustering method(Fig. 1B). A total of 258 cases were classified into ICDRGs group 1 and 272 cases were classified into ICDRGs - group 2 (Fig. 1C-D). Based on immune cell dysfunction-related genes, we analyzed two types of immune cell dysfunction-related scores using the GSVA method. The immune cell dysfunction-related scores of group 1 were higher than those of group 2 (Fig. 1E). In order to evaluate the survival difference between the two groups, we analyzed patients from different group ( TCGA ) in different ranges, and analyzed the complete survival information(Fig. 1F).
Immune Cell Infiltration Evaluation
Additionally, the TCGA samples' 22 immune cell subtypes were examined in 2 subtype clusters using the CIBERSORT method. The findings revealed that memory B cells, T cells CD4 memory resting, follicular helper cells, T cells gamma delta, macrophages M0, macrophages M2, resting mast cells, and eosinophils were immune infiltrated by two subtype clusters, which accounted for a sizable portion of immune cell infiltration (Fig. 2A). Additionally, compared to other subtype groups like group2, which had more pure tumors, group1 had higher immune, stromal, and estimated scores (Fig. 2B). Cancer has responded very well to immune checkpoint inhibition, which inhibits inhibitory signals of T cell activation. Figure 2C shows the two clusters' HLA family gene expression. Higher HLA gene expression is found in group1. Immunological cells express immunological checkpoints, which can control the level of immune activation. They are crucial in stopping the development of autoimmunity. In group1, there is more expression of PDCD1 and LAG3.
Enrichment Analysis
A total of 730 Differentially expressed genes(DEGs) are found between group 1 and group 2 using the limma method under the filtering conditions of FDR q value 0.01 and absolute value of logFC > 0.5 (Fig. 3, Supplementary Data Table 1). Gene enrichment analysis in difference analysis was carried out to look into the biological pathways and processes connected to groups. Figure 4A demonstrates that the majority of biological processes connected to immune response activation and molecule synthesis are linked. Additionally, the exterior side of the plasma membrane and the immunoglobulin complex are significantly enriched in the cellular component (Fig. 4B). These really relate to the body's active immune metabolic processes, which show the body's immune reaction as cancer spreads. We discovered through looking at Fig. 4C that the genes in Molecular function are primarily enriched in antigen binding and immune receptor activation. Finally, pathway enrichment analysis revealed that the Estrogen Signaling Pathway and Phagosome were the key enrichments for the differentially expressed genes of group 1 and group 2 (Fig. 4D).
Construction and validation of the prognostic model
The LASSO approach was then used to build risk models, which finally resulted in the discovery of 13 prognostically important genes (Fig. 5A-C). Then, based on training (TCGA, n = 530), test set1 (GSE30219, n = 307), and test set2 (GSE37745, n = 196) datasets acquired from TCGA LUSC and GEO dataset, respectively, risk score-based models using these genes were built. According to the findings of the survival study, greater risk scores in the training and test sets were associated with worse survival (P < 0.0001) (Figs. 5D,E,G,H,J,K). The sensitivity of the prognostic model was evaluated using time-dependent ROC curves. In accordance with the areas under the ROC curves (AUCs) results, the training set's 1-, 3-, and 5-year AUCs were 0.707, 0.655, and 0.792, respectively (Fig. 5F), while the test sets' corresponding AUCs were 0.638, 0.66, and 0.684 (Fig. 5I) and 0.609, 0.642, and 0.648, respectively (Fig. 5L).
Independent prognostic ability of risk model
We further looked at the relationship between the risk model and clinicopathological characteristics of TCGA-LUSC using univariate regression analysis and multivariate Cox regression analysis to assess the independent predictive capacity of the developed risk model. Grade, Age, TNM Stage, Stage, and Risk Model were registered as clinically relevant variables for analysis. As a consequence, the prognosis of LUSC patients may be predicted independently by Grade, Age, and Risk Model (Fig. 6A-B). As a therapeutically useful quantitative technique tool for estimating the mortality of specific BC patients, we created a nomogram by integrating independent prognostic indicators (Fig. 7A). Each patient is given a total score that is calculated by adding the results of each prognostic parameter. The prognosis of patients declines as the overall score rises. Performance of the modal diagram is comparable to that of the ideal model (Fig. 7B-D). The risk model predicts patient survival more correctly than age, according to receiver operating curve analysis (Fig. 7E). Nomogram fared better than other single factors, according to the c-exponential curve of various variables across time in the TCGA cohort (Fig. 7F). In addition, the DCA curve showed that the nomogram's net benefit curve in terms of age was stable and reliable compared to other clinical parameters (Fig. 7G).
Risk model can predict how chemotherapy work
Eight chemotherapeutic medicines' IC50 discrepancies were examined using the "pRRophetic" package to forecast their sensitivity to drug therapy. The drug sensitivity data for Sorafenib, Gefitinib, Bleomycin, Bosutinib, Etoposide, Lenalidomide, Camptothecin, and Methotrexate in the risk model of LUSC are displayed in Figs. 8A-H, respectively. The statistical findings showed that the IC50 values for chemotherapeutic medicines are greater in high-risk individuals.
Gene expression level verification via quantitative reverse transcription PCR (qRT‑PCR)
Through protein-protein interaction analysis, we obtained the core network of risk model genes, consisting of AKR1B1, LOX, SERPINA1, SERPINA5 and GPC3. In order to verify the validity of identified genes, qRT-PCR was used to measure the expression levels of those genes. According to Fig. 9, all mRNA transcripts were found to be expressed at levels that were noticeably greater in the LUSC tumor cell lines (A549) than in the nearby normal cell lines (BEAS-2B).