3.1 Identification of dysregulated pyroptosis- and immune-related lncRNAs

The transcriptome data downloaded from the TCGA database was divided into mRNA and lncRNA datasets. We identified 59 dysregulated pyroptosis-related genes and 316 dysregulated immune-related genes. We screened out 279 differentially expressed PyrolncRNAs (DE-Pyro lncRNAs) from 797 PyrolncRNAs (Fig. 1A) and 379 differentially expressed immune-related lncRNAs (DE-Imm lncRNAs) from 1194 immune-related lncRNAs (Fig. 1B). Then, Venn analysis between the DE-Pyro lncRNAs and DE-Imm lncRNAs was performed, generating 245 differentially expressed pyroptosis- and immune-related lncRNAs (Fig. 1C).

Fig. 1

Identification of differentially expressed PyrolncRNAs (DE-Pyro lncRNAs) and immune-related lncRNAs (DE-Imm lncRNAs) in bladder cancer. A Volcano plot of DE-Pyro lncRNAs. Blue dots: down-regulation. Orange dots: up-regulation. B Volcano plot of DE-Imm lncRNAs. Blue dots: down-regulation. Orange dots: up-regulation. C Venn analysis between the DE-Pyro lncRNAs and DE-Imm lncRNAs

3.2 Prognostic model construction

60 Pyro-Imm lncRNAs associated with the prognosis of bladder cancer patients through univariate Cox regression analysis were obtained (Fig. 2A). Then, LASSO regression analysis screened seven Pyro-Imm lncRNAs (Fig. 2B). Ultimately, Multivariate Cox regression revealed three Pyro-Imm lncRNAs (MAFG-DT, AC024060.1, AC116914.2) to construct a prognostic model. Risk score = (0.5185 × MAFG-DT expression value) + (− 0.3744 × AC024060.1 expression value) + (− 0.5498 × AC116914.2 expression value). The lncRNA-mRNA co-expression network was visualized in Fig. 2C. AC024060.1 and AC116914.2 were protective factors and MAFG-DT was a risk factor in the Sankey diagram (Fig. 2D).

Fig. 2

Establishment of a risk score model. A, B LASSO regression was performed based on 60 Pyro-Imm lncRNAs obtained by univariate Cox regression analysis. C The co-expression network of lncRNAs and mRNAs. Blue color: Pyro-Imm lncRNAs. Green color: Pyroptosis-related mRNAs. Purple color: immune-related mRNAs. D Sankey diagram showing the correlations between prognostic Pyro-Imm lncRNAs, mRNAs, and risk type

3.3 Evaluation of the prognostic risk model

To evaluate the predictive signature, patients were randomly separated into two cohorts (first internal or second internal cohort) and divided into low- and high-risk groups based on their median risk score. Regardless of internal cohorts or overall dataset, Kaplan-Meier analysis revealed that the OS rate in the low-risk group was significantly higher than that in the high-risk group. Here, we take the result of the first internal cohort as a representative, and the result of the second internal cohort or overall cohort was presented in the Supplementary figure 1. The risk score for the low-and high-risk groups in the first internal is shown in Fig. 3A respectively. Figure 3B displayed the survival stats of these cases. The Kaplan–Meier analysis revealed that the OS rate in the low-risk group was significantly higher than that in the high-risk group (Fig. 3C). In the time-dependent ROC curve, the first internal cohort’s area under the curve (AUC) at 1, 3, and 5 years was 0.741, 0.74, and 0.805 (Fig. 3 D).

Fig. 3

Evaluation of the risk score model. Risk scores and survival status in the first internal cohort (A, B). Kaplan-Meier tests in the first internal cohort (C). Time-dependent ROC analysis of risk score at 1, 3, and 5 years in the first internal cohort (D). AUC area is under the curve

3.4 Evaluation of the risk signature as an independent prognostic factor for bladder cancer

The HRs (95% CI) of risk score were 1.521 (1.328–1.742) in the univariate Cox regression analysis (p-value < 0.001, Supplementary figure 2 A) and 1.380 (1.196–1.593) in the multivariate Cox regression analysis (p-value < 0.001, Supplementary figure 2 A, B). Multi-index ROC analysis showed that the AUC of risk score was 0.731, higher than the clinicopathological characteristics (age, sex, tumor stage, T stage, and N stage) (Supplementary figure 2 C). Additionally, principal component analyses (PCA) showed that patients with different risk scores could be better distinguished based on the three Pyro-Imm lncRNAs (Supplementary figure 2 F), compared with the whole genes (Supplementary figure 2 D) and the overlapping 245 Pyro-Imm lncRNAs (Supplementary figure 2 E).

3.5 The risk model was associated with prognosis in different clinicopathological features

Bladder cancer patients were stratified by the clinicopathologic features, including age, sex, tumor stage, grade, T stage, and N stage. According to Kaplan-Meier survival analysis, the age, male, high grade, tumor stage III-IV, T stage, and N stage were associated with higher survival probability in the low-risk group compared with that in the high-risk group (Supplementary figure 3 A–I).

3.6 Relationships between Pyro-Imm lncRNAs and clinicopathological features and nomogram construction

In the risk prognosis model, the 3 Pyro-Imm lncRNAs were related to clinicopathological features. MAFG-DT was associated with survival status, age, and N stage (Supplementary figure 4 A, B, C). AC024060.1 was related to survival status and grade (Supplementary figure 4 D, E). AC116914.2 was connected to the survival status, grade, tumor, T, and N stages (Supplementary figure 4 F–J). Besides, containing the risk score and clinicopathological factors, the nomogram could predict the 1, 3, and 5 year prognosis of bladder cancer patients (Supplementary figure 5, A). The predicted survival rates were consistent with the actual OS rates at the time of 1, 3, and 5 year, demonstrating the nomogram’s strong predictive ability (Supplementary figure 5 B).

3.7 Gene set enrichment analysis

GSEA revealed that KEGG pathways, including the WNT signaling pathway, cell cycle, DNA replication, focal adhesion, and ECM (extracellular matrix) receptor interaction were significantly enriched in the high-risk group, while no signaling pathways were enriched in the low-risk group (Fig. 4 A). The GO analysis showed that the high-risk group enriched in cell adhesion via plasma membrane adhesion molecules, cell adhesion mediator activity, positive regulation of cell cycle G2/M phase transition, positive regulation of cell division, and WNT protein binding, whereas no items were enriched in the low-risk group (Fig. 4 B).

Fig. 4

Gene Set Enrichment Analysis. A KEGG pathway analysis showed five pathways were enriched in the high-risk group. B Five GO items were enriched in the high-risk group. (ECM represents extracellular matrix)

3.8 Immune infiltration and immune checkpoints

We investigated the correlations between immune cells and risk score. The results showed that macrophages, plasmacytoid dendritic cells (pDCs), T helper type 1 cells (Th1), and regulatory T cells (Treg) were significantly higher in the high-risk group, compared with the low-risk group (Fig. 5 A). We further evaluated the relationships between the expression value of immune checkpoint genes and the two risk groups. The immune checkpoint genes, including TNFRSF14, TNFRSF15, TNFRSF25, LGALS9, BTNL2, HHLA2, CD40, CD40LG, CD160, ADORA2A, TMIGD2, and IDO2 were highly expressed in low-risk group, while TNFRSF4, PDCD1LG2, NRP1, CD44, and CD276 were substantially expressed in high-risk group (Fig. 5 B). However, the expression of PDL-1 and CTLA4 was no difference between high-risk and low-risk group. The results indicate that the two risk groups' immune responses varied and may react differently to immunotherapy.

Fig. 5

Immune infiltration and immune checkpoints. A The infiltration levels of 16 immune cells in the low-risk and high-risk groups. B The expression value of 17 immune checkpoints in the low-risk and high-risk groups. *p < 0.05, **p < 0.01, ***p < 0.001, ns: not significant

3.9 Relationship between the risk model and bladder cancer treatment

We analyzed the relationship between the risk model and the efficacy of general chemotherapeutic and targeted drug treatment for bladder cancer. The results showed that the IC50 values of cisplatin, docetaxel, paclitaxel, imatinib, and pazopanib were lower in the high-risk group, whereas the IC50 values of methotrexate, vinorelbine, and axitinib were higher in the high-risk group (Fig. 6).

Fig. 6

The efficacy of bladder cancer treatment in low-risk and high-risk groups. The IC50 of cisplatin A, docetaxel B, methotrexate C, paclitaxel D, vinorelbine E, axitinib F, imatinib G, and pazopanib H were compared between low-risk and high-risk groups. (IC50 represents half-maximal inhibitory concentration)