Identification and validation of a novel pyroptosis-related lncRNAs signature associated with prognosis and immune regulation of hepatocellular carcinoma

Pyroptosis is an inflammatory form of cell death triggered by certain inflammasomes. However, research concerning pyroptosis-related lncRNAs in hepatocellular carcinoma (HCC) remains scarce. This study aims to explore the prognostic pyroptosis-related long non-coding RNAs (lncRNAs) of HCC patients. Data of 373 HCC patients were obtained from the TCGA database. The entire cohort was randomly divided into a training cohort and a validation cohort in a 2:1 ratio. Pyroptosis-related lncRNAs were identified by the Pearson correlation analysis with reported pyroptosis-related genes. LASSO Cox regression was used to construct the signature. A prognostic signature consisting of nine pyroptosis-related lncRNAs was identified, and patients with lower risk scores had a better prognosis than those with higher risk scores. Multivariate Cox regression analysis showed that the signature was an independent risk factor for prognosis in both the training and validation cohorts. In the training cohort, the area under the signature curve reached 0.8043 at 1-year, 0.7878 at 2-year, and 0.8118 at 3-year; in the validation cohort, it reached 0.7315 at 1-year, 0.7372 at 2-year, and 0.7222 at 3-year. Gene set enrichment analysis (GSEA) suggested associations between the signature and several immune-related pathways. The expression of multiple immune checkpoints was also increased in the high-risk group, including PD-1, PD-L1, CTLA4, B7-H3, VSIR, LAG3, and TIGIT. A novel pyroptosis-related lncRNA signature, which may be associated with tumor immunity and potentially serve as an indicator for immunotherapy, has been identified to precisely predict the prognosis of HCC patients.


Identification of pyroptosis-related lncRNAs.
Thirty-three pyroptosis-related genes were obtained from Ye's reports (Supplementary Material 1) 10 . Pearson correlation test was used to calculate the correlations between lncRNAs and pyroptosis-related genes. Based on the cut-off criteria of Pearson correlation coefficient > 0.3, these lncRNAs were considered as candidate pyroptosis-related lncRNAs.
Construction and validation of the prognostic pyroptosis-related lncRNAs signature. The cohort was randomly divided into a training cohort and a validation cohort in a 2:1 ratio. The data from the training cohort were used to construct the prognostic pyroptosis-related lncRNAs signature, while the other cohort was used for validation. Univariate Cox regression analysis was used to identify the prognostic pyroptosis-related lncRNAs. Subsequently, the least absolute shrinkage and selection operator (LASSO) Cox regression was used to construct the signature by R packages (glmnet and survival) as follows: risk score = expression of lncRNA1 × β1lncRNA1 + expression of lncRNA2 × β2lncRNA2 + …expression of lncRNAn × βnlncRNAn. The two cohorts were further divided into the low-risk and high-risk groups, respectively. Survival analysis and time-dependent ROC curves were performed to investigate the prognostic value. Moreover, multivariate Cox regression of available patient characteristics was performed to reconfirm the prognostic value of the signature. Additionally, a nomogram was constructed to predict the patient prognosis more precisely.
The mRNA-lncRNA co-expression network. In order to better demonstrate the associations between pyroptosis-related genes and pyroptosis-related lncRNAs, a co-expression network was constructed. A Sankey diagram was used to illustrate the mRNA-lncRNA relationships.
Gene set enrichment analysis (GSEA) and subsequent functional enrichment analysis. The tumor tissues were divided into the low-risk group and high-risk group based on risk scores. Differentially expressed genes between groups were identified by the "DEseq2" package with cut-off criteria of false discovery rate < 0.05 and |log 2 foldchange| > 1. After that, these differentially expressed genes were uploaded for GSEA analysis (http:// www. broad insti tute. org/ gsea) 11 . Immunological analysis. The abundance of tumor-infiltrating immune cells in HCC tissues was investigated using the Tumor IMmune Estimation Resource (TIMER) database (https:// cistr ome. shiny apps. io/ timer/) 12 . Statistical analysis. R 3.3.0 and Statistical Package for Social Sciences 23.0 (SPSS Inc., Chicago, IL, United States) were used for statistical analysis. One-way analysis of variance (ANOVA) with homogeneous variances was used to analyze differences in immune cell components between the normal and HCC tissues, while Welch's ANOVA was applied when variances were heterogeneous. Kaplan-Meier curve was used for survival analysis, where a Log-rank test was adopted for comparison between groups.
Ethics approval and consent to participate. Extra informed consent is not essential for the data were all obtained from public database. The authors cannot assess to any identifying characteristics, which do not distort scientific meaning.
Research involving human participants and/or animals. This article does not contain any studies with human participants or animals performed by any of the authors.

Results
Cohort studies of the training cohort and validation cohort. A total of 373 HCC patients were enrolled. However, two patients were excluded from this study due to incomplete clinical data (Fig. 1a). Thus, 371 patients were finally included in this study and randomly divided into the training cohort and the validation cohort in a 2:1 ratio. The characteristics of HCC patients in the two cohorts are shown in Table 1, with no statistically significant differences in all characteristics.
Identification of prognostic pyroptosis-related lncRNAs. Pearson correlation test was performed between lncRNAs and 33 pyroptosis-related genes, and then 5372 lncRNAs were preliminarily identified with PCA analysis in supplement material 2. Subsequently, in the training cohort, the prognostic value was examined by the univariate Cox regression, and 1030 prognostic pyroptosis-related lncRNAs were chosen for further analysis. Using LASSO Cox regression analysis, 9 lncRNAs (AC019080. Then, a co-expression network between the pyroptosis-related genes and pyroptosis-related lncRNAs was constructed to confirm their relationships (Supplementary Material 4). As shown in Fig. 3a and supplement material 5, 24 pyroptosis-related genes and 9 pyroptosis-related lncRNAs are included in the network. It is worth noting that MKLN1-AS and AL031985.3 might be the most likely pyroptosis-related lncRNAs. In addition, the close correlation between pyroptosis-related genes and pyroptosis-related lncRNAs is indicated by the Sankey diagram (Fig. 3b). These findings suggest that the 9 pyroptosis-related lncRNAs might play important roles in HCC.
Validation of the prognostic pyroptosis-related lncRNA signature. The prognostic signature was subsequently validated in the training cohort and validation cohort. The patients in the two cohorts were divided into the low-risk group and high-risk group based on the median risk score. The characteristics of the two groups in the training cohort are compared in Table 2. The high-risk group is proved to be associated with a poorer T stage (p = 0.031), pathologic stage (p = 0.013), and histologic grade (p < 0.001). Figure 4a-d shows that the probability of death is higher in the high-risk group than in the low-risk group in both cohorts. Kaplan-Meier curves of OS show consistent results in both cohorts (Fig. 4e, f), indicating that the survival of HCC-TCGA patients in the high-risk group is significantly worse than that in the low-risk group. A time-dependent receiver operating characteristic (ROC) was performed to investigate values in predicting the patient prognosis (Fig. 4g, h).  The barplots comparing the 9 prognostic pyroptosisrelated lncRNAs between HCC tissues and adjacent normal tissues. HCC, hepatocellular carcinoma. *P < 0.05; **P < 0.01; ***P < 0.001.    (Table 3) and the validation cohort ( Table 4). The results show that the risk score could be an independent factor to predict the patient prognosis in the two cohorts.

Scientific Reports
Nomogram based on the signature for predicting the OS of HCC patients. In order to develop a more accurate model for prognosis prediction, univariate Cox regression analysis was performed in both cohorts (Fig. 5a). The pyroptosis-related lncRNA signature (hazard ratio 2.013, 95%; confidence interval 1.731-2.341), hepatitis B (hazard ratio 0.357, 95% confidence interval 0.221-0.576), and tumor stage (hazard ratio 2.448, 95% confidence interval 1.689-3.548) were risk factors for the prognosis of HCC patients. Proportional hazard assumption analysis was subsequently performed, and the results showed the variables did not violate the proportional hazard assumption (Supplementary Material 6). Then, a nomogram, including hepatitis B, tumor stage, and risk scores, was constructed to explore the probability of the lncRNA signature in predicting 1-, 3-, and 5-year survival in HCC patients with a C-index of 0.721. In this combined nomogram, the risk score model exerts the most excellent weight in predicting the 1-year, 3-year, and 5-year survival probability with a C-index of 0.711 (Fig. 5b). The findings identify that this novel pyroptosis-related lncRNA signature can act as a promising prognostic model for HCC patients.
GSEA and functional enrichment analysis. GSEA analysis was performed to explore the potential biological functions involved in the signature, using the differentially expressed genes between the low-and highrisk groups (Supplementary Material 7). Figure 6a-i shows the top 9 immune-associated signaling pathways, including the CTLA4 pathway, antigen presentation folding assembly and peptide loading of class I MHC, autoimmune thyroid disease, inflammatory pathway, antigen processing and presentation, IL5 pathway, cytokines and inflammatory response, TCR signaling, and TH1TH2 pathway. The results show that the pyroptosis-related lncRNA signature may be associated with multiple immune-related pathways.
Associations between the signature and immune infiltration. In order to explore the roles of the pyroptosis-related lncRNA signature in the immune microenvironment of HCC patients, the associations between the signature and immune infiltration cells were further investigated using the ESTIMATE algorithm (Supplementary Material 8). The proportions of various immune cells are shown in Fig. 7a, b, and the correlations between these immune cells in the low-risk and high-risk groups are shown in Fig. 7c, d. Among the immune cells, the CD4 + memory activated T cells, regulatory T cells (Tregs), M0 macrophages, and neutrophils were all increased in the high-risk group, while the CD4 + memory resting T cells, monocytes, and M2 macrophages were decreased in the high-risk group compared with the low-risk group (Fig. 7e). In addition, the associations between the signature and other immune checkpoints, including PD-1, PD-L1, CTLA4, B7-H3, VSIR, LAG3, and TIGIT, were also investigated. As shown in Fig. 7f, the immune checkpoints were significantly up-regulated in the high-risk group compared with the low-risk group. To conclude, the results indicate that the   www.nature.com/scientificreports/ pyroptosis-related lncRNA signature may be associated with immune-related mechanisms and the response of HCC patients to immunotherapy.

Discussion
In this study, prognostic pyroptosis-related lncRNAs in HCC were comprehensively investigated. A novel prognostic signature consisting of 9 pyroptosis-related lncRNAs (AC019080.5, AP003392.4, MKLN1-AS, AL031985.3, PCCA-DT, AC007128.1, LNCSRLR, AL445228.3, AC023157.2) was developed using LASSO Cox regression analysis and validated by the validation cohort. The subsequent functional analysis confirms the associations between the signature and multiple immune-related pathways. The results suggest that the identified signature has potential values in predicting the patient prognosis and managing immunotherapy for HCC patients. Several studies have investigated the roles of lncRNAs in pyroptosis 13 . It is reported that lncRNA KCN-Q1OT1 inhibits the miR-21 expression and subsequently promotes the caspase-1-dependent pyroptosis during cataract formation 14 . It is found that LncRNA MALAT1 increases the level of NLRP3 by up-regulating ELAVL1 expression, thus leading to cell pyroptosis in diabetic nephropathy 15 . It has been proven that LncRNA ANRIL up-regulates NLRP3 and IL-1β by increasing the expression of BRCC3, thus activating pyroptosis in uric acid   17 . In terms of cancer, lncRNA RP1-85F18.6 has been identified to promote  In the co-expression network, MKLN1-AS and AL031985.3 might be the most likely pyroptosis-related lncR-NAs in the TCGA-HCC cohort. Studies have suggested that MKLN1-AS aggravates the progression of HCC by sponging miR-654-3p 21 . However, no studies have reported the potential role of AL031985.3 in HCC. Because of the great significance of MKLN1-AS, AL031985.3, and 7 other lncRNAs in HCC, future studies are encouraged to reveal the underlying mechanisms of these candidate lncRNAs in HCC biology, especially cell pyroptosis.
Our study shows significant relationships between the signature and many immune-associated signaling pathways. Patients with higher risk scores display a higher expression of multiple immune checkpoints, including PD-1, PD-L1, CTLA4, B7-H3, VSIR, LAG3, and TIGIT. Additionally, PD-1, a member of the B7/CD28    25 . Here, this study suggests that patients with higher risk scores may benefit more from immunotherapy than those with lower risk scores. However, this study has certain limitations. Firstly, cohorts in this study were mainly based on the TCGA database. Thus, real-world data are needed for further validations of the pyroptosis-related lncRNA signature in HCC patients. Secondly, this study failed to investigate the underlying immune-related mechanisms of the identified lncRNAs. Moreover, further studies should focus on the potential of the pyroptosis-related lncRNA signature as an indicator of immunotherapy.

Conclusion
In summary, a novel pyroptosis-related lncRNA signature has been identified to precisely predict the prognosis of HCC patients. The signature is robustly associated with the tumor immunity, providing a personalized prediction model for the prognosis and immunotherapeutic response of HCC patients.