Risk coefficient model of necroptosis-related lncRNA in predicting the prognosis of patients with lung adenocarcinoma

Model algorithms were used in constructing the risk coefficient model of necroptosis-related long non-coding RNA in identifying novel potential biomarkers in the prediction of the sensitivity to chemotherapeutic agents and prognosis of patients with lung adenocarcinoma (LUAD). Clinic and transcriptomic data of LUAD were obtained from The Cancer Genome Atlas. Differently expressed necroptosis-related long non-coding RNAs got identified by performing both the univariate and co-expression Cox regression analyses. Subsequently, the least absolute shrinkage and selection operator technique was adopted in constructing the nrlncRNA model. We made a comparison of the areas under the curve, did the count of the values of Akaike information criterion of 1-year, 2-year, as well as 3-year receiver operating characteristic curves, after which the cut-off value was determined for the construction of an optimal model to be used in identifying high risk and low risk patients. Genes, tumor-infiltrating immune cells, clinical correlation analysis, and chemotherapeutic agents data of both the high-risk and low-risk subgroups were also performed. We identified 26 DEnrlncRNA pairs, which were involved in the Cox regression model constructed. The curve areas under survival periods of 1 year, 2 years, and 3 years of patients with LUAD were 0.834, 0.790, and 0.821, respectively. The cut-off value set was 2.031, which was used in the identification of either the high-risk or low-risk patients. Poor outcomes were observed in patients belonging to the high-risk group. The risk score was the independent predictor of the LUAD outcome (p < 0.001). The expression levels of immune checkpoint and infiltration of specific immune cells were anticipated by the gene risk model. The high-risk group was found to be highly sensitive to docetaxel, erlotinib, cisplatin, and paclitaxel. The model established through nrlncRNA pairs irrespective of the levels of expression could give a prediction on the LUAD patients’ prognosis and assist in identifying the patients who might gain more benefit from chemotherapeutic agents.

Identifying necroptosis-related lncRNAs. The necroptosis-related gene set M24779.gmt was obtained from http:// www. gseam sigdb. org/ gsea/ index. jsp, the website for Gene Set Enrichment Analysis (GSEA). Subsequently, it was added to the genes that are necroptosis-related from the prior reports. Using the Pearson correlation analysis and co-expression strategy, lncRNAs with a co-expression correlation coefficient > 0.4 and p-value < 0.001 were defined as nrlncRNAs. We utilized the "limma" R package for the purpose of performing analysis of the differential expression of the acquired nrlncRNAs. The nrlncRNAs showing a log fold change (FC) > 1.0 along with false discovery rate (FDR) < 0.05 were identified as differentially expressed nrlncRNAs (DEnrlncRNAs).
Construction of DEnrlncRNA pairs. DEnrlncRNAs were cyclically singly paired, and the parameter value was defined as α value of 0 or 1. The lncRNA pair value was 1 in the case where the lncRNA A expression was more than that of a sample of lncRNA B; else, it was 0. Then, the created 0-or-1 matrix was subjected to additional screening. www.nature.com/scientificreports/ Risk model establishment for risk score assessment. For the purpose of screening lncRNAs related to patient survival from the nrlncRNA pool (p < 0.05), a univariate Cox proportional hazard regression analysis was performed. Then, for 1000 cycles, the least absolute shrinkage and selection operator (LASSO) regression analysis was conducted through the use of a p-value < 0.05 together with cross-validation of 10-folds. To prevent overfitting, stimulation was set up 1000 times for each cycle in a random manner. For the purpose of conducting Cox proportional hazard regression analysis, we chose pairs that contained a frequency of more than 100 times, and the best lncRNA pairs were chosen for the construction of the Cox risk coefficient model. By risk score calculation, we computed the AUC value of every model and drew the 1-, 2-, and 3-year receiver operating characteristics (ROC) curves that are dependent on time of the model.
Groups of low and high-risk were constructed based on the optimal fitting of the Akaike information criterion (AIC).
Constructed risk model validation. The Kaplan-Meier analysis was conducted in investigating the survival differences between patients in both groups. With the aid of the "survival" and "survminer" R packages, the survival curve was drawn for visualization. The Chi-square test was performed in finding out the relationship between the model and clinical factors. We then carried out the Wilcoxon rank-sum test for assessing the associations between several subgroups of clinical indicators and the risk score. To obtain a clear comprehension of the data, we utilized "limma" and "ggpubr" in R packages. To validate that the model can be utilized on LUAD patients as a clinical prognostic predictor that is independent, Cox multivariate and univariate regression analyses were executed on clinical correlation features as well as the risk score. To envisage this data, the R "survival" package was put into use. For the purpose of comparing the accuracy of the risk score and clinically relevant features in predicting survival profiles and outcomes, we made a comparison of the ROC curves acquired from a follow-up lasting 1 year with ROC curves of indicators that are clinically relevant in the same chart.

Tumor-Infiltrating immune cells correlation analysis. Current techniques were utilized in calcu-
lating the status of the immune infiltration in the TCGA samples, including XCELL (http:// xCell. ucsf. edu/) 15 , QUANTISEQ (http:// icbi. at/ quant iseq) 16 , TIMER (version 2.0; http:// timer. cistr ome. org/) 17 , MCPCOUNTER, CIBERSORT (http:// ciber sort. stanf ord. edu/) 18 , EPIC (http:// epic. gfell erlab. org) 19 , and CIBERSORT-ABS to perform analysis of the correlation between infiltration of immune cells and the risk score. We assessed the differences existing between the high-risk and low-risk groups in their tumor-infiltrating immune cell content by the Wilcoxon signed-rank test. Furthermore, we did an analysis on the spearman correlation to find out the relationship between the risk score and the infiltration levels of the immune cells. The "limma", "scales, " "ggplot2, " as well as "ggtext" packages in R were utilized in data visualization.
Immunosuppressive molecules expression analysis related with ICIs. The "limma" and "ggpubr" packages in R were utilized to find out if there were substantial differences in gene expression that are ICI-related between the two groups. After this, visualization of data was performed.
Chemotherapeutic agents correlation analysis. The conventional chemotherapeutic drugs, including erlotinib, cisplatin, paclitaxel, docetaxel, gefitinib, individually or in combination, were selected in determining if there existed a difference in chemotherapeutic agents response based on LUAD patients belonging to both groups. We utilized the drug's half-inhibition rate (IC50) as an index in measuring the sensitivity to the drug. The "limma, " "ggpubr, " "ggplot, " and "pRRophetic" packages in R were used in data analyses and visualization. This study utilized R software (version 4.0.0: http:// www.r-proje ct. org) for conducting statistical analyses, and Supplementary Table S7 provides the specific functions of the R package.
Ethics approval and consent to participate. Not applicable, data was collected from public data repositories.
Guidelines statement. All experimental protocols were performed in accordance with the relevant guidelines and regulations and adhered to the Declaration of Helsinki.

Results
Necroptosis-related lncRNAs of patients with LUAD. As shown in the flow chart ( Fig. 1), the TCGA database was used in retrieving the LUAD transcriptome data. Data that contained no duplicates and follow-up time were not included in the present research, and 497 LUAD samples and 54 normal tissue samples adjacent to the tumor were considered. From GSEA and previous reports, we acquired a 67 necroptosis-related genes (nrgenes) profile (Supplementary Table S1). Co-expression analysis was conducted between lncRNAs and nr-genes that were known, and differentially expressed lncRNAs (|LogFC|> 1.0 and p < 0.05) between cancer and normal samples were identified. There was a total of 484 identified nrlncRNAs (Supplementary Table S2), and 140 obtained DEnrlncRNAs; of the 140, there were 127 upregulated and 13 downregulated DEnrlncRNAs ( Fig. 2A Table S5). Next, 26 DEnrlncRNAs pairs were identified for the risk coefficient model building by performing the LASSO regression analysis. These 26 DEnrlncRNA pairs were then analyzed by performing multivariate as well as univariate Cox regression analysis (Fig. 3A,B) to obtain the risk factor for each necroptosis-related lncRNA pair ( Table 1).

Assessment of the prognostic predictive performance of the risk model. The 26 prognostic DEn-
rlncRNA pairs that were selected were utilized in constructing patients' receiver operator characteristic (ROC) curves during the periods of 1, 2, and 3 years (Fig. 4A). The areas under the curve (AUC) during the periods of 1, 2, and 3 years were found to be 0.834, 0.790, and 0.821, respectively, which also contained a predictive significance. The 1-year area under the curve (AUC) was calculated of 0.834, which was the largest AUC (Fig. 4B). The cut-off value was computed based on the best fit and found to be 2.031 (Fig. 4C). This cut-off value was used in differentiating between high-and low-risk groups of patients with LUAD. Accordingly, the low-risk group contained 338 patients who participated in the group, whereas the high-risk group contained 126 patients.

Correlation analysis of clinical features with the aid of the risk model. R was used in analyzing
the correlation between risk-subgroup patients and risk scores (Fig. 5A). We obtained the relationship of the patients on their risk coefficient score as well as the survival status (Fig. 5B). Subsequently, through the use of the survival status from both groups, the construction of the Kaplan-Meier curve was achieved (Fig. 5C). The findings indicated that the patients' rate of survival in the low-risk group was substantially greater in comparison to that of the group of high-risk (p < 0.001). Moreover, patients belonging to the low-risk group presented a survival time that was significantly longer in comparison to that in the high-risk group (p < 0.001).
For the purpose of examining the correlation between clinical features and LUAD risk, we conducted multiple chi-square tests. The Wilcoxon signed-rank test assisted us in obtaining a heatmap (Fig. 6A). From the findings,    (Fig. 7A). However, the risk score (p < 0.001, HR = 1.235, 95% CI [1.186-1.286]) was the only factor whose presentation was a prognostic predictor that was independent by performing the multivariate Cox regression analysis (Fig. 7B). We did a comparison on the ROC curve of clinical features and the risk coefficient score during the period of 1 year (Fig. 7C). The result showed that the patients' risk score (AUC = 0.834) and stage (AUC = 0.709) had the highest predictive efficacy.

Immune-cell infiltration and risk coefficient model correlation.
We assessed if there was a relationship between the tumor immune microenvironment and the model (Fig. 8). The findings are recorded in Supplementary Table S6. By performing Spearman correlation analysis, we established a positive correlation existed on the tumor-infiltrating immune cells when compared to the high-risk group, including common lymphoid progenitors, resting mast cells, CD4 + T cells, macrophage M1, uncharacterized cells, macrophage M0, as well as neutrophils ( Supplementary Fig. S1).
The correlation between the risk coefficient model and chemotherapeutic agents. Moreover, we investigated the correlation between the risk coefficient model and the sensitivity to chemotherapeutic agents. To assess the drugs' efficacy, we utilized IC50. Lower IC50, indicating that the sensitivity was higher. According to these findings, we established that there was a correlation between the high-risk group and the www.nature.com/scientificreports/ higher sensitivity to cisplatin (Fig. 10A), docetaxel (Fig. 10B), erlotinib (Fig. 10C), and paclitaxel (Fig. 10E). There was no significant difference in the sensitivity to gefitinib in both groups (Fig. 10D).

Discussion
Recent research reports have set up signatures based on lncRNAs for the purpose of evaluating the prognosis of patients with cancer. lncRNA-related models of LUAD, such as immune-related lncRNAs [20][21][22] , autophagy-related lncRNAs [23][24][25] , pyroptosis-related lncRNAs 26 , and methylation-driven lncRNAs 27 , were reported in previous studies. In this study, we constructed the models of risk coefficient, which were essential in the assessment of the LUAD patients' prognosis based on the nrlncRNA pairs. In the present research, we firstly obtained LUAD patients' nr-gene and lncRNAs data from GSEA and TCGA, did an analysis of the differential co-expression in establishing the DEnrlncRNAs, and then performed lncRNA pairs validation through cyclically single pairing them together with a matrix of 0-or-1. Second, we acquired each sample's risk coefficient of patients with LUAD and then constructed a risk coefficient model by performing multivariate regression, LASSO regression analyses, as well as Cox multivariate and univariate regression analyses. Third, we computed each ROC's AUC value in obtaining the optimal model fit and then obtained the critical value based on optimal fitting of the Akaike information criterion (AIC), which was utilized in identifying the difference existing in the high and low-risk groups. The novel model had the clinical practicability benefit in differentiating between the cases belonging to both groups.
We performed the correlation analyses in assessing the efficacy and accuracy of the constructed risk coefficient model, including tumor-infiltrating immune cells, survival, genes, clinical characteristics, and chemotherapeutic agents, and found that the model algorithm worked well. www.nature.com/scientificreports/ To find out the correlation between the risk scores and tumor-infiltrating immune cells, we utilized seven techniques that are generally accepted to estimate the immune infiltrating cells, including TIMER, XCELL, CIBER-SORT, QUANTISEQ, EPIC, MCPCOUNTER, as well as CIBERSORT-ABS, and found out that the relationship between the tumor-infiltrating immune cells, such as CD4 + T cells, resting mast cells, common lymphoid progenitors, uncharacterized cells, macrophage M0, macrophage M1, and neutrophils and the highrisk group was positive (p < 0.01). Wu et al. 28 proposed that LINC00665 played a critical role in enhancing the infiltration levels of macrophages, dendritic cells, and inhibited regulatory T cells to avoid exhaustion. T-cell. Xu et al. 29 established that HK3 enhanced the infiltration of macrophages and monocytes that presented the antigens of the cell surface and regulated the debilitating T cells' critical genes (PD1 and CTLA4), thus having an effect on the process of immune escape.
In cancer immunotherapy, it was found that necroptosis participated highly in the immunity of the antitumor. Good performance was evident from the prognostic signature that was necroptosis-related on the basis of four genes (EZH2, TLR4, TRAF2, and PGAM5) in the prediction of the prognosis of patients with stomach adenocarcinoma 30 . Immune checkpoint inhibitors (ICIs) include anti-cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) and anti-programmed cell death protein 1/programmed cell death ligand 1 (PD-1/PD-L1). Recent studies have identified novel immune checkpoint targets such as lymphocyte activation gene 3 (LAG-3), T cell immunoglobulin and ITIM domain (TIGIT), T cell immunoglobulin and mucin-containing domain 3 (TIM-3), hepatitis A virus cellular receptor 2 (HAVCR2) gene and the TIM-3 ligand galectin-9 (Gal-9), etc. Cytotoxic T lymphocyte-associated protein 4 (CTLA-4) expressed by T cells is recognized as a key immune checkpoint for autoimmunity and cancer therapeutic targets. CTLA-4 is a member of the immunoglobulin-associated receptor family, which Suppresses T cell activation and responsible for all aspects of T cell immune regulation. The generation of specific monoclonal antibodies illustrates the controlling role of CTLA-4 in T cell responses. CTLA-4 can mediate negative regulation of T cell activation by competing with the co-stimulatory receptor CD28 for binding to its co-ligands B7.1 and B7.2. It can also be regulated by promoting Treg development and function. After activation, CTLA-4 expression is induced on CD4 + Foxp3− (forkhead box P3) and CD8 + Foxp3− conventional T cells, while CTLA-4 constitutively expressed by CD4 + Foxp3 + Treg cells 16 . We believe that there are differences between various immune cells and immune-related phenotypes. According to the findings of the present research, we observed the expression levels of CTLA4 were elevated in samples from patients belonging to the high-risk group, which can be utilized as a potential therapeutic target.
Tumor microenvironment changes may be linked to the development of immune-targeted drug resistance, making it crucial to discover sensitive drugs for clinical therapy. The correlation analysis of chemotherapeutic agents showed that the sensitivity to cisplatin, docetaxel, erlotinib, and paclitaxel was greater in the group of high-risk in comparison to that of the group of low-risk. It was shown that necroptosis induction in the immune checkpoint blockade (ICB) and tumor microenvironment may have a synergistic impact on enhancing a longterm tumor rejection 7 . The phosphorylation or lack of caspase 8 was essential for Paclitaxel-triggered necroptosis in lung adenocarcinoma cells. When epithelial-mesenchymal transition (EMT) was triggered, a novel lncRNA, called lncCRLA, was markedly upregulated, which inhibited RIPK1-induced necroptosis by interfering with the RIPK1-RIPK3 interaction by binding to the RIPK intermediate domain 31 .
Necroptosis is a form of regulated cell death regulated by RIP1, RIP3, and MLKL. Inducing necroptosis in mice with orthotopic pancreatic cancer increased the survival time and attenuated tumor growth, stroma, and metastasis 32 . lncRNAs have been established to affect tumor cell growth from several previous research, which is crucial for clinical therapy as well as patient prognosis 33 . For example, via the mechanism of inhibiting www.nature.com/scientificreports/ miR-150-5p, lncRNA LINC00673 regulates the invasion, epithelial-mesenchymal transition, migration, as well as the proliferation of non-small cell lung cancer 34 , while LINC00472 inhibited EMT via binding to YBX1 and affecting the cell's mechanical features, and as a result, obstructing its invading and metastasizing ability 35 . The lncRNA MIF-AS1 enhanced the proliferation of tumor cells while reducing apoptosis in digestive system cancer 36 . Experiments have shown that the lncRNA that is necrosis-related has been shown from several experiments to target miR-873; moreover, RIPK1/RIPK3 plays a role in the regulation of the cardiomyocyte necroptosis 37 .
Another study found that when HCC expresses lncRNA LINC00176, miRNAs, such as miR-9 and miR-185, are produced and downregulate the mRNAs they target, which enhances the necroptosis of liver cancer cells 38 .
The novel insights about nrlncRNAs could assist us in gaining a clear understanding of the LUAD mechanism, which could have a crucial role in the treatment. However, in the present research, there were some disadvantages together with limitations. For initial analysis, there was insufficient raw data and thus more clinical data are needed. The sample size was small. The risk factor model lacked external data validation, which reduced the reliability of the model, so further validation needs to be performed.
Finally, the present research showed that nrlncRNAs and the risk coefficient might be utilized in predicting the prognosis of patients with LUAD and assist in identifying those patients who might benefit from chemotherapeutic agents.  www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.