The TCGA platform was used to obtain the transcriptome RNA sequences and clinical data for 471 KIRP samples. The Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/) platform was used to collect the GEO cohort (GSE2748) KIRP samples. Incomplete clinical information was deleted from samples, and FPKM values in TCGA-KIRP were transformed to Transcripts Per Kilobase Million (TPM) values and utilized for copy number variation (CNV) analysis (Zhao et al., 2021). The transcriptome RNA sequences from the TCGA-KIRP and GSE2748 datasets were combined after correction. R software version 4.1.1 was used to process the raw data.

First and foremost, PRGs were identified from published literature (Man and Kanneganti, 2015; Wang and Yin, 2017; Karki and Kanneganti, 2019). The transcription of 52 PRGs was utilized to identify distinct pyroptosis types and to identify patients for further research using an unsupervised clustering approach. To do the aforementioned study, we utilized the R package “ConsensuClusterPlus” and 1,000 repeats to ensure clustering stability (Wilkerson and Hayes, 2010). Our best guess for how many clusters there should be was arrived at by using the consensus clustering method. Gene set variation analysis (GSVA) was performed on the MSigDB hallmark gene set in order to investigate the variations in PRGs that occur throughout cellular mechanisms. Furthermore, the Kaplan–Meier curves generated by the R packages “survival” and “survminer” were used to investigate the differences in overall survival across subtypes.

We utilized the ESTIMATE method to assess each patient’s immunological and stromal scores. KIRP samples’ fractions of 22 different kinds of human immune cells were also determined using the CIBERSORT technique (Chen et al., 2018). To evaluate the degree of immune cell infiltration in the TME of KIRP, a single-sample gene set enrichment analysis (ssGSEA) approach was utilized in conjunction with the other techniques (Subramanian et al., 2005).

We utilized the R package “limma” to compare DEGs between PRG clusters (Ritchie et al., 2015). The Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were carried out using the “clusterProfiler” program (Gene Ontology Consortium, 2015; Kanehisa et al., 2017).

To objectively measure pyroptosis in individual KIRP patients, a score system was developed. The following is the procedure for developing the scoring system: The DEGs found in various pyroptosis clusters were first standardized across all samples, and then the overlap genes were retrieved. Then, for each gene, we ran a univariate Cox regression analysis. For the following phase of the investigation, a list of genes with a high prognosis was created. The following formula was used to determine pyroptosis scores using principal component analysis (PCA):

PRG Score = Σ (Expi * Coefi)

The clinical characteristics and risk score from the independent prognostic study were used to generate a nomogram using the “rms” program (Park, 2018). In the nomogram scoring approach, a score was given to each variable, and the total score was derived by adding the scores from all variables in each sample. With the help of nomogram calibration plots, it was possible to see the relationship between surviving events and their virtually observed consequences.

The mutation annotation file from the TCGA was developed using the “maftools” R package to determine the genetic alterations of KIRP patients in various categories (Mayakonda et al., 2018). For each patient with KIRP in the two groups, we estimated the tumor mutation burden (TMB) value. Using the “pRRophetic” software, we estimated the semi-inhibitory concentration (IC50) values of medications frequently used to treat KIRP in patients to see if there were any variations in their therapeutic effects (Geeleher et al., 2014).

We began by looking at their CNV and mutation frequency in TCGA-KIRP. PRG mutations were found in 39 of 281 TCGA-KIRP samples, accounting for 13.88 percent of the total. The top four PRGs with the greatest mutation frequencies were TP53 (2%), NLRP1 (1%), NLRP7 (1%), and SCAF11 (1%) (Figure 1A). Further investigating the predictive impact of PRGs, we discovered that most PRGs, such as TP53, GZMB, AIM2, CHMP7, and GSDMC, were substantially linked with the overall prognosis of KIRP patients. Figure 1B depicts the 39 PRG positions on the chromosomal CNV alterations. CNV change was widespread in the 52 PRGs that exhibited mostly copy number amplification, such as CHMP6 and PJVK, whereas CASP9, CASP4, CASP5, CASP1, and IL18 showed considerable copy number decrease (Figure 1C). The levels of PRGs in KIRP patients and renal tissues were examined to see whether the genetic variants mentioned above had an influence on PRG transcription in KIRP patients (Figure 1D). According to our results, variations in CNV may be major factors driving changes in PRG regulator expression. CNV-amplified PRG transcription was considerably greater in KIRP compared to normal renal tissue (e. g., CHMP6 and PJVK) and vice versa (e. g., CHMP3, CHMP7, ELANE, CASP9, NLRP2, and TIRAP). This study found considerable PRG inherited heterogeneity and variance in KIRP patients, indicating that PRG transcription imbalances were important in the onset and development of KIRP.

The patients from two suitable KIRP cohorts (TCGA-KIRP and GSE2748) were included in our study for additional analysis in order to thoroughly recognize the PRG expression levels implicated in carcinogenesis. We created a pyroptosis network map to demonstrate the interactions of PRGs in KIRP patients and their relationship to prognosis in these patients (Figure 2A). We employed a consensus clustering technique to identify patients according to the transcript level of the 52 PRGs to learn more about their expression features in KIRP. Our findings indicated that k = 2 was the best choice for categorizing the whole patients into subgroup A (n = 177) and subgroup B (n = 146) (Figures 2B,C). According to the results of the PCA analysis, the PRG expression patterns of the two subgroups were considerably different (Figure 2D). In addition, comparing the clinicopathological characteristics of various KIRP subtypes revealed substantial disparities in PRG transcription and clinical features (Figure 2E). Cluster A had the highest level of expression for PRGs.

Following that, we used GSVA enrichment to investigate the biologic effects of several PRG groups. There was significant enrichment of immune cell activation pathways in PRG cluster A, including T cell and B cell receptor signaling pathways (Figure 3A). Then, we employed ssGSEA to compare immune cell infiltration in the tumor microenvironment amongst distinct PRG clusters. Surprisingly, we found a considerable variation in immune cell concentration amongst PRG clusters, with cluster A was considerably enriched in immune cells (Figure 3B).

We utilized the “limma” R package to explore 2461 PRG phenotype-associated DEGs, which were designated as PRG signature genes, to learn more about the probable biologic activities of distinct PRG clusters. Surprisingly, it was shown that DEGs are greatly overrepresented in immune-related pathways. DEGs were found to be concentrated in biological processes of BP related to T cell activation and leukocyte cell-cell adhesion. DEGs were considerably enriched in cytokine receptor binding, cytokine binding, and immune receptor activity throughout MF processes (Figure 4A). Additionally, DEGs were considerably enriched in immune-related pathways in KEGG pathway enrichment analyses: cytokine-cytokine receptor interaction, Human T-cell leukemia virus infection, and chemokine signaling pathway (Figure 4B).

Following that, we used univariate Cox analysis to screen for prognostically related DEGs, yielding 659 genes related to survival time. We employed unsupervised cluster analysis to separate the KIRP patients into three gene groups based on the genes that were chosen for inclusion (Figures 5A,B). The survival research showed that patients in subgroup C had the highest overall survival rate, while those in subgroup A had the lowest OS (Figure 5C). In gene cluster A, all PRGs were found to be overexpressed (Figure 5D). This revealed that greater PRG expression might be related to a worse prognosis in KIRP patients. Furthermore, the heatmap revealed that these prognostically associated DEGs were common in subgroup A, which confirmed the findings of the previous study (Figure 5E).

We established the PRG score, a scoring method that evaluates the pyroptosis types of each KIRP patient, to further examine the activities of pyroptosis in KIRP. In order to begin, we separate the patients into two groups: those who will be trained and those who will be tested. LASSO and multivariate Cox analyses were performed on prognostic DEGs associated with pyroptosis subtypes to further select the best prognostic features (Figures 6A,B). In order to depict variations in the characteristics of specific patients, an alluvial diagram was employed (Figure 6C). Between subgroups, we identified a statistically significant variance in PRG score. Additionally, both gene cluster A (Figure 6D) and PRG cluster A had a high PRG score (Figure 6E). On the basis of the median PRG score, patients were split into high and low-risk categories. When the PRG score is increased, the survival duration reduces, and the recurrence rate rises, as shown by the risk distribution chart for the PRG score (Figures 6F,G). Furthermore, we discovered substantial disparities in the expression of PRGs between risk categories (Figure 6H).

In all sets (p < 0.001, Figure 7A), training sets (p = 0.008, Figure 7B), and testing sets (p < 0.001, Figure 7C), those with low scores had considerably better OS than subjects with elevated scores, according to the K–M curves. In all sets, AUC values of 0.948, 0.809, and 0.820 were used to reflect the 1-, 3-, and 5-year survival rates of the PRG score (Figure 7D). Similarly, the AUC in the training set was 0.967, 0.707, and 0.713 (Figure 7E), and in the test set, they were 0.936, 0.916, and 0.959 (Figure 7F). According to this, the PRG score may accurately predict the clinical outcome with KIRP patients. Because the PRG score’s clinical utility in evaluating OS in patients of KIRP is unsatisfactory, a nomogram integrating the PRG score and clinical characteristics was developed to predict survival rates (Figure 7H). The calibration chart revealed that the PRG score performed well, with a good match between the projected and actual survival rates (Figure 7G).

The CIBERSORT technique was utilized to determine the association between the PRG score and immune cell abundance. The PRG score was favorably connected with B cells and dendritic cells resting, macrophages M1, and other cells, as indicated in the scatter diagrams, and negatively correlated with macrophages M0, macrophages M2 (Figure 8A–L). Additionally, we examined the relationship between the genes and the number of immune cells in the body. We discovered that the 13 genes were highly linked with the majority of immune cells, especially GBP1 and macrophages M1, KCNJ5 and macrophages M2 (Figure 8M).

Then, in the TCGA-KIRP cohort, we looked at how the distribution of somatic mutations differed across two PRG score groups. As it turned out, the mutation rates between the two groups varied significantly (64.18 percent and 59.44 percent). TTN, MUC16, MET, KMT2C, KIAA1109, SETD2, USH2A, MUC4, KMT2D, and WDFY3 were the remarkable mutant genes in the various risk groups (Figures 9A,B). TTN, MUC16, and MET mutations were more common in patients with lower PRG scores than in those with higher PRG values. KMT2C and KIAA1109, on the other hand, have the exact opposite mutation levels. KMT2C and KIAA1109 exhibited a greater frequency of missense mutations and a lower rate of frame-shift mutations in patients with high PRG score. We next chose chemotherapeutic medications that are presently used to treat KIRP to assess the sensitivities of individuals in distinct risk categories to these treatments. We discovered individuals with high PRG scores had lower IC50 values with A-443654 (specific inhibitor of Akt), ATRA (all-trans retinoic acid), AZD-2281 (olaparib), AZD8055 (mTOR inhibitor), and BI-D1870 (RSK inhibitor). However, the IC50 values of chemotherapeutics like AZD6244 (MEK inhibitor selumetinib) and BMS-708163 (gamma secretase inhibitor) were much lower in individuals with a lower PRG score. The findings suggested that PRGs were linked to medication sensitivity (Figure 9C–N). While most of the ideas about how to diagnose and treat cancer have come from research on clear cell RCC, the unique characteristics of KIRP may have significance for illnesses.