Currently, scientific research on the role of AS events in STAD still has many unanswered questions owing to the dearth of available large-sample public AS profiles and the paucity of systematic analysis referring to their clinical significance and deep molecular function. These bottlenecks have prevented cancer researchers from effectively recognizing the widespread applicability of AS events in STAD. Exploration of AS patterns broadens our vision and our understanding in traditional transcriptome molecular biomarkers. In this project, we adopted several biomedical computational approaches, which integrate the AS event profiles and clinical information of STAD patients to mine prognosis-related AS and construct splicing prognostic signatures that could stratify STAD patients into subgroups with distinct survival outcomes. Moreover, the SF-AS network could provide further insights into regulatory mechanisms in patients with STAD from the perspective of splicing.
Gastric cancer is a highly heterogeneous malignant tumor. Therefore, single drug is not significantly useful for various types of gastric cancer. Classical cytotoxic therapy cannot be fully effective because of the presence of patients resistant to specific drugs. At present, the diagnosis and treatment of gastric cancer rely on histopathological diagnosis and definite classification. Therefore, in addition to targeted treatment with trastuzumab, we need to develop new targeted drugs to provide better treatment for patients. Potential biomarkers can be mined and used to predict patient outcomes, and treatment strategies can be developed for specific tumor parameters.
The next-generation sequencing technology developed in recent years adopts the whole- genome sequencing method, which has great advantages in exploring alternative splicing. Previously, several studies conducted SpliceSeq analyses to generate alterative splicing profiles for some types of cancer, as well as to construct prognostic signatures for cancer prognosis monitoring, including non-small cell lung cancer [18], colorectal cancer [19], and esophageal cancer [20]. This computational bioinformatics analysis could open up different perspectives on the clinical application and potential pathological mechanism of AS on a macro level. Previously, several studies have proposed transcriptomic signatures related with epithelial-to-mesenchymal transition and diagnosis of gastric cancer [21, 22]. The present in-depth study further explored alterations of transcriptomes used as prognostic predictors and could broaden our horizons in the clinical significance of transcriptomic signatures.
Given the multitude of AS events impacted by their own pre-mRNAs, the downstream functional impact is partly used to describe the molecular function of AS alteration events. In the PPI network analysis, UBA52, STAT3 and PLK4 were the hub genes. Previous studies have shown that UBA52 and STAT3 are all considered to be related molecules involved in the biological process of STAD. For example, bioinformatics analysis has verified the correlation between UBA52 and GC progress and metastasis [23]. STAT3 signaling drives transcription activation of EZH2 and mediates poor prognosis in gastric cancer [24]. STAT3 promotes the increased expression of lncRNA HAGLROS, which leads to further progress of gastric cancer [25]. These findings also pave the way for future clinical applications. Functional enrichment analysis showed that in STAD, the main molecular function of AS event gene related to prognosis is to bind to GTPase, so it may provide selective advantages for cancer cells by regulating GTPase. Increased RhoA activity leads to poorer survival outcomes for the Lauren diffuse type of gastric adenocarcinoma (DGA), and inhibition of RhoA can correct the drug resistance of DGA [26]. RacGAP1 is closely associated with malignant progression and poor survival [27]. Leptin promotes GC migration through the Rho/ROCK mechanism [28]. RASSF6 partially regulates the effect of mir-181a-5p on GC progression through MAKP pathway [29]. It is worth considering that, in gastric cancer cells, RhoA promotes cell proliferation and RhoC stimulates cell migration and invasion, while RhoB functions contrary to RhoA and/or RhoC [30]. Therefore, targeted GTPase therapy is also being explored. For example, ALEX1 functions in gastric cancer through the PAR-1/Rho GTPase signaling pathway, becoming a new target for tumor inhibition [31]. RhoA-mediated Fbxw7 regulates the apoptosis of tumor cells and other phenotypes in gastric cancer [32]. Similarly, Gastrokine 1's inhibition of gastric cancer progression may also be dependent on RhoA [33]. Our findings suggest that a group of AS events play a biological role in the alteration of GTPase in STAD.
The highlight of the current study was that we proposed prognostic signatures based on AS events for monitoring the prognosis of STAD patients. Recently, some prognostic signatures in STAD have been proposed. Zhang H et al. found that the efficacy of postoperative adjuvant chemotherapy for gastric cancer was affected by the degree of neutrophil infiltration of the tumor [34]. Jiang Y et al. developed an immune score GC classifier that can effectively predict the recurrence and survival of patients with gastric cancer, which plays a good role in complementation of the prognosis judgment for the TNM staging system [35]. The clinical management of STAD patients still needs to be improved, and the above mentioned molecular biomarkers have broad prospects. In order to facilitate clinical practice, we selected a group of AS events using the LASSO Cox regression model, and the prognostic model proposed on which showed satisfactory results. Of course, this also requires a separate cohort for validation.
A large number of AS events are programmed by finite SFs in cells [36]. The altered profile of AS events in multiple tumor types emphasizes the important mechanism of splicing factors in cancer which is disordered splicing [37]. It is increased believed that changes of SFs in STAD can be involved in tumorigenesis and progression through various mechanisms [38–40]. The splicing correlation network analysis has also found out the larger regulated nodes, indicating that they occupy a significant position in the SF-AS network. QKI, which is recognized as a tumor suppressor in a wide range of cancers, is highly connected in the network, which can play a significant part in the prognosis induced by splicing events [41–43]. But the role of QKI in STAD has not been fully discussed yet. Our study indicates that the level of QKI expression is significantly correlated with the survival rate of patients with SATD, and it can become an important target for drug design in the future. Nevertheless, our algorithm suggested deregulated AS events as a hallmark of STAD. However, there are some limitations inevitably affecting the reliability of the study. Firstly, we didn’t use a separate cohort for more validation. Secondly, more functional experiments are needed to further investigate the impact of dysregulated AS events and SFs on carcinogenesis.
In conclusion, the current study has found out a phenomenological relationship between AS events and prognosis in STAD patients, which is the base of unscrambling the functional contribution of AS events in STAD. These findings are conducive to develop new genomic models for clinical cancer management. In addition, the further identification of predictive splicing factors for prognosis and the construction of SF-AS networks will pave the way for further exploration of splicing related mechanisms.