BLCA is one of the most common and aggressive urological malignancies with great molecular and pathological heterogeneity [1–2]. Although early-stage BLCA could be handled with several inhibitory strategies such as TUR-BT and BCG intravesical perfusion therapy, the rate of recurrence and progression remained high [4–5]. Once the NMIBC progressed into MIBC, the possibility of occurrence of lymph node metastasis would increase sharply, which may easily develop into metastatic BLCA and the survival of patients would become extremely worse [6–7]. Recently, the emergence of ICB therapy shed light of hope on the high-stage BLCA patients, while the immune status possessed great heterogeneity between individuals, which resulted in varied response to ICB from different groups of BLCA patients [29–30]. Therefore, key biomarkers and effective prognosis predictors are urgently warranted for better subgrouping of BLCA patients and more precise clinical decision making. Increasing evidence has indicated that the establishment of key gene signature model based on crucial biological genetic processes, such as hypoxia, immune infiltration, ferroptosis, lipid metabolism and so on, could serve as effective tools for molecular subgrouping and prognosis prediction for multiple cancers [31–33]. Likewise, DNA damage repair, which is a highly-conserved biological process, has been found to be able to form prognosis prediction models based on a range of crucial related genes [34]. Here in this study, a 7-gene signature model based on DDR pathway was established by us in BLCA, which demonstrated great accuracy in prognostic prediction and promising potential in ICB treatment guidance.
Firstly, to evaluate whether the differentially-expressed genes in DDR pathway could effectively subgroup the BLCA, two molecular subgroups of BLCA were derived based on the 232 DDR-related genes (Fig. 1A-D). Through differentially-expressed genes detection (Fig. 2B) and gene pathway enrichment analysis (Fig. 2C), we found that the two molecular subtypes exhibited significantly differed characteristics, including DNA replication, nuclear division, mismatch repair, cell cycle and cellular senescence, which are all important phenotypes and biological functions closely related with the DDR process. Hence, we may conclude that the regulatory roles played by DDR-related genes possessed great heterogeneity among BLCA patients and could distinct molecular subtypes could form dependent on DDR difference. In addition, within the candidate DDR-related genes, combination of a range of crucial genes may help establish a more convenient and specific prognostic model for cancer diagnosis. Recently, multiple studies utilized DDR-related genes to form a range of effective prognostic models for several cancers, such as lung cancer [35], breast cancer [36], glioma [37] and cervical carcinoma [38]. While in BLCA, a range of prognostic models have also been established. Cao, et al developed a 7-gene model based on EMT in BLCA, which demonstrated a good effect in predicting BLCA patients’ prognosis and the derived riskscore may serve as an independent risk factor for BLCA diagnosis [39]. However, this study only explored the prognostic value of the EMT-signature in BLCA, and whether this tool could guide the ICB or chemotherapy selection is still unclear. Moreover, the biological mechanism remained unknown, which limited the further clinical use of this model. Zhou, et al also established a 11-gene signature based on autophagy for prognosis prediction in BLCA, which exhibited an even better efficacy in survival prediction. Nevertheless, a same defect still existed in this study that the inner mechanism explanation lacked [40]. More recently, ferroptosis has been identified as a novel iron-dependent cell death regulatory mechanism and has also been applied as a prognostic modality in BLCA. Notably, the established ferroptosis-based riskscore was found to be associated with the expression of a range of immune-checkpoint proteins in the tumor microenvironment of BLCA [41]. This study suggested that the gene signature may not only serve as a prognosis predictive tool, but also could represent the inner immune status and mark the suitability of BLCA patients for ICB therapy. As mentioned above, DDR is a pivotal modulator in regulating the progression and therapeutic response in cancer. More importantly, DDR pathways are also found to be tightly correlated with the immune activity, which may be utilized as genomic biomarkers for cancer immunotherapy. Qing, et al explored the double-sword function of DDR in the cancer immunotherapy and found that DDR pathway mutations as well as their derived up-regulated neoantigens could determine diverse immune phenotypes [42]. Likewise, another study has also identified the co-mutations in DDR system could effectively predict the outcomes of lung cancer and melanoma patients in response to ICB therapy [43]. For BLCA, the main risk factors include smoking and exposure to deleterious chemicals, which are all powerful driving force for induction of DNA damage [44]. Thus, we may naturally understand the great biological significance of DDR system in BLCA. Correspondingly, multiple studies have revealed the important regulatory roles played by DDR in BLCA. In a randomized controlled trial, the polymorphism of DDR-related genes has been found be associated with the progression of BLCA, especially muscle-invasiveness [45]. Notably, under the treatment of platinum-based chemotherapy, patients with DDR-related genetic alterations have a significantly better prognosis compared with the controls [46–47], which indicated that DDR-related information may assist in the patient selection for multiple therapeutics, including ICB therapy. Nevertheless, whether DDR-related genes could help establish an applicable and effective prognosis prediction model and guide in selection of different therapeutics is still unknown.
Based on the DDR-subgrouping of molecular subtypes in BLCA, we screened out 7 key DDR-related genes, which are all significantly-associated risk factors for BLCA prognosis, to form the final gene signature model. As shown in Fig. 3D-F, the model could effectively predict the overall survival of BLCA patients with a high accuracy in two independent datasets. And the riskscore could serve as an independent risk factor for BLCA. The relatively low number of genes provided with the convenience in developing the rapid detection kit for clinical practice to BLCA. GSEA enrichment analysis enabled us to further identified the biological functions differed by DDR features. As we can see from Fig.A and Fig.C, several gene pathways are significantly upregulated in the high-risk groups, including some classical oncological pathways, such as WNT signaling pathway, growth factor binding and pathway in cancer. Hyperactivation of WNT signaling pathways has been verified as a major engine force for immune escape promotion and therapeutic resistance, which is mainly dependent on enhancement of DDR [48–49]. Oxidative stress-mediated WNT pathway is a potential therapeutic target for optimizing cancer therapy via regulating genomic integrity [50]. In addition, a range of metabolic pathways have also been found to be significantly altered between the high-risk and low-risk group, including purine, fatty-acid, α linolenic-acid and arachidonic-acid metabolism. The synthesis of fatty-acid through fatty-acid synthase could up-regulate DDR and therefore induce cancer cell survival and therapeutic resistance [51]. In turn, specific DDR proteins could also modulate lipid metabolism in tumor cells [52]. In all, the interactive work between cellular metabolism and DDR is an important regulator of biological processes, which may be a valuable target for novel therapeutics design. The difference in the expression of multiple cytokines, soluble ligands and growth regulators between the high-risk and low-risk group further indicate the variation of biological phenotypes differed by DDR features (Fig. 6E).
The immune infiltration analysis showed us that the higher riskscore is positively correlated with the activation of both innate and adaptive immunity. As can be seen in Figure.7, the expression of macrophage has a significantly-positive association with the riskscore across all the algorithms. Furthermore, the immune function analysis indicated that a range of immune pathways mainly concerning antigen-presentation processes, including APC co-stimulation, HLA and MHC-I. Alterations or deficiency in the DDR system could cause increased accumulation of genetic mutations and chromosomal rearrangement, which lead to genomic instability and may promote tumor progression [53]. Nevertheless, this process could also generate a series of neoantigens [54–55] and up-regulate immune-checkpoint proteins, such as PD-L1[56–57], which may exert immune-activating functions and could serve as a synergic functioning target in combination with ICB agents. In BLCA, Zeng, et al found the macrophage infiltration, especially M1 subtype is an effective biomarker for predicting ICB response and determining immune characteristics in the tumor microenvironment. Notably, M1 expression is positively associated with T cell immunity and PD-L1 expression and negatively correlated with immune exclusion [58]. Under the premise that abnormal regulation of DDR system and increased accumulation of neoantigens, macrophages could also exert their antigen-presentation capabilities, which would also active adaptive T cell anticancer immunity [59–60]. However, the cross-presentation of macrophages could also induce immune tolerance. In consistency with the results of the immune infiltration analysis, the immune-checkpoint protein analysis indicated that the high-risk group exhibited a significantly higher expression of a range of immune-checkpoint proteins, including PD-L1, PD-1 and CTLA-4 (Fig. 8). Higher PD-L1 expression in tumor commonly represented an immune-sensitive status, which confers the patients with better response to ICB therapy [61–62]. On the other hand, the over-expression of immune-checkpoint like PD-1 on CD8+ T cell represented an immune-exhausted state, which could be recovered by anti-PD-1 ICB therapy and reinvigorate the anti-tumor immunity [63–64]. In our TIDE scoring analysis, we found that the overall immune escape phenotype and immune exclusion in high-risk group is significantly better than the low-risk group, which indicated that our model-derived high-risk group of BLCA patients may be an ideal subgrouping selection for better responders to ICB therapy. In addition, the PD-1-negative T cell expression is significantly lower in high-risk group than that in the low-risk group (Fig. 10D and F), which further verified the great potential of immune reactivation in high-risk group patients of BLCA.
In addition to the ICB therapy, considering that the standard therapeutics for BLCA are still centered on chemotherapy and molecule-targeted therapy, we also performed a drug sensitivity analysis on a series of commonly-used drugs for BLCA based on the DDR model grouping. Surprisingly, we found that the high-risk group patients displayed a significantly better sensitivity to all the 9 analyzed agents (Fig. 9). Cisplatin is the first metal-based antitumor chemotherapeutic agent, which is still the standard first-line chemotherapy regimen for BLCA. The main anticancer mechanism involves its cytotoxic interaction with the DNA bases and the induction of cellular apoptosis [65–66]. The core of the tumor-killing capability of cisplatin is dependent on its promotion of DNA damage, which could be resisted innately or adaptively by the up-regulation of DDR processes [67]. Wang, et al found that the a DDR key gene BRCA1 could be inhibited by XPC inhibition in BLCA, which could lead to increased chromosomal instability and maintaining cisplatin-induced DNA damage [68]. Moreover, DDR-related genes have been found to be able to effectively predict disease-free survival for muscle-invasive BLCA [69]. The same effect has also been found in Doxorubicin, Mitomycin and Camptothecin, which are mainly dependent on the DNA-cytotoxicity. From these results we may conclude that DDR itself may be a favorable factor for maintaining genomic stability in a homeostasis state. However, under the treatment of DNA-cytotoxicity dependent chemotherapy, the up-regulation of DDR may be an unfavorable factor for forming drug resistance. In all, our 7-gene model based on DDR gene signature could also serve as an effective tool for BLCA patient selection of different therapeutics.