Introduction
lung carcinoma (LC) is known as one of the most prevalent forms of carcinoma and is the underlying reason for cancer mortality in China [
1]. In the US, it possesses a 5-year survival rate of 14% [
2]. Delayed diagnosis, particularly after metastasis, is one of the major causes for its poor prognosis for LC [
3]. Approximately 85% of all cases of LC have non-small cell LC (NSCLC), including varying histological types, such as LUAD, squamous cell carcinoma, and large cell LC, with LUAD being the most common type of NSCLC [
4]. Presently, there are no reliable biomarkers available for predicting prognosis in patients suffering from LUAD, which, in turn, requires the need for the identification of suitable prognostic markers.
Alternative splicing (AS) is a critical post-transcriptional process that involves the formation of protein isoforms with varied structural and functional characteristics. AS has a considerable contribution in the modification of > 95% of human genes and is thus, commonly used to explore proteome diversity and cellular complexity [
5‐
7]. Numerous investigations have revealed the association between AS abnormal expression and the pathogenesis of several cancers, including LUAD [
8‐
11]. The splice factor SRSF1 has been shown to modulate PTPMT1 alternative splicing to regulate LC cell radioresistance [
12]. Also, whole-genome analysis of AS events in LC has resulted in the identification of several candidate splicing factors, which might act as therapeutic targets of LUAD, and help in disease prognosis of patients via the construction of gene signatures [
13,
14], demonstrating the role of AS in LC.
AS is directly linked with tumorigenesis and critically involved in the formation of TME, which includes tumor cells, tumor-associated fibroblasts, immune/inflammatory cells, microvessels, stromal tissues, and many cytokines and chemokines [
15‐
19]. The prognosis of cancer patients is known to be directly related to the immune cell count in the TME, which can act as a useful prognostic marker [
20‐
22]. However, except for some preliminary studies on LC-related AS events [
13,
14] and immunological microenvironment [
23,
24], there is a lack of sufficient data on the immunological relevance of AS events.
Here, we integrated TME and AS events to conduct a first-of-its-kind analysis of prognostic variables in patients with LUAD. Firstly, the immune and stromal scores of patients were procured with LUAD through exploring public databanks and applying the ESTIMATE algorithm. Next, the Kaplan–Meier (K–M) plots were generated by using the obtained data to investigate prognostic variations between higher and lower score groups (stromal or immune). Subsequently, we searched for DEAS through comparative analysis of AS events in a higher and lower score group (stromal or immune). Then, we created two OS- and PFS-based prognostic signatures and determined that prognostic indicators were independent of other pathophysiological markers. Finally, we discerned distinct AS-based LUAD clusters. We studied the relationship between AS-based clusters and the pathophysiological factors and other immunological characteristics, allowing an improved understanding of the prognosis of patients with LUAD.
Discussion
Immunotherapy has revolutionized the treatment of cancers over recent decades and possesses a considerable role in LUAD treatment. A randomized Phase III trial comparing atezolizumab with docetaxel in cases with earlier treated progressive NSCLC, which revealed that atezolizumab outperformed docetaxel in terms of OS. Across programmed death-ligand 1 and histological subcategories, atezolizumab was found to improve the survival rate. Patients who received atezolizumab had fewer treatment-associated adverse events of grade 3 or 4 than those who received docetaxel [
37]. Durvalumab indicated clinically substantial enhancements in OS and PFS than the standard of care in strongly pretreated patients with metastatic NSCLC. Durvalumab + tremelimumab indicated numerical enhancements in OS and PFS relative to standard of care [
38]. Immunotherapy has given people with LC novel hope, but it is not appropriate for all of them. To enhance the prognosis of LC patients, more research is required to identify immune-associated prognostic markers and to develop innovative therapeutic options. The immune-associated DEAS events were employed to build a model of risk score that accurately predicted the outcomes of LUAD patients in our investigation.
Herein, the ESTIMATE algorithm was employed to determine scores (immune as well as stromal) of LUAD generated from the TCGA database from the microenvironment's perspective. We then used K–M curves to predict the prognosis of LUAD and indicated that cases with a greater immune/stromal score exhibited a higher chance of survival. We also evaluated immune/stromal score-associated DEASs and then chose the optimal DEASs linked with the survival using LASSO CRA by comparing the profiles of transcriptional expression within LUAD cases with higher against lower immune/stromal scores. In addition, the final prognostic signature was developed, demonstrating that it is capable of accurate prediction. The LRG, in particular, had a higher chance of surviving than the HRG. Furthermore, the risk score can be used for predicting LUAD patient survival as an independent factor. This characteristic, when taken as a whole, holds a lot of promise for predicting LUAD patients' survival.
The biological functions, such as adherens junction, cell–matrix adhesion, lysosome, and others, were found to be closely linked with the development, growth, and progression of tumors, based on GO and KEGG enrichment analysis; Adherens junctions are significantly linked with the invasive and migratory potential of cancerous cells [
39], for instance, E-cadherin is one of the key constituents of adherens junction, and a tumor suppressor and its loss is linked with a bad prognosis in a wide spectrum of cancers, including prostate cancer [
40] and neck and head cancer [
41]. Cell–matrix adhesion has also been linked to the progression of many cancers [
42‐
44]. Lysosomes considerably contribute to the cell's degradation, and defects in them can lead to unregulated cell proliferation [
45]. All of the above shows that the biological effects related to AS events are inseparable from the occurrence and development of tumors.
We also performed an analysis of unsupervised cluster to divide the TCGA LUAD cohort into three subgroups. The patients with a good prognosis had the greatest immune, stromal, and ESTIMATE scores, as well as the greatest level of immune cell infiltration, according to our findings. Immune cells are key prognostic factors in patients with LUAD in earlier research [
46]. For instance, the increase in several targetable immune checkpoint molecules is linked to EMT and the microenvironment of inflammatory tumors [
47]. Öjlert et al. discovered that a higher immune score and great estimations of numerous adaptive immune cell types were linked to higher PFS in LUAD patients [
23]. Jones et al. discovered that the immunological signatures of cytotoxic lymphocytes and T-cell trafficking were linked to the prognosis of female malignancies like LUAD, showing that immune cell infiltration in the TME influenced treatment and survival results in LUAD cases [
48].
In the current study, we also evaluated the prognostic value of DEASs. An OS-prognostic signature based on 5 DEASs (MKL1|62,349|AP, ICAM3|47,503|RI, SOD2|78,301|AT, CLASP2|63,869|AP, and EXOC6|12,541|AP) and PFS-prognostic signature based on 6 DEASs (DNMT3A|52,857|AT, NCK1|66,941|AP, SOD2|78,301|AT, IFRD1|81,447|AP, SH3KBP1|88,642|AP, and ABI1|11,048|ES) was constructed. Interestingly, among the splicing events, one overlapping AS event (SOD2|78,301|AT) was discovered with substantial variations based on OS and PFS concurrently, implying that SOD2|78,301|AT is the most plausible independent prognostic factor. Its parental gene SOD2 is a crucial enzyme for scavenging ROS generated by mitochondria, which is necessary for cellular homeostasis [
49]. Reported studies have been revealed that SOD2 has been linked with the energy metabolism of colorectal cancer [
50], oxidative stress in endocrine cancer [
51], and stem cell reprogramming in breast cancer [
52]. Other parental genes in the two signatures, in addition to this one, have been identified to perform a considerable task in the existence and progression of many cancers to differing degrees, as well as the prognosis of cancer patients. An elevated expression of MKL1 indicates an unfavorable prognosis in papillary thyroid cancer patients and enhances nodal metastases [
53]. CLASP2 has been linked to the EMT and early progression of bladder cancer [
54] and has also been shown to predict bladder cancer prognosis [
55]. In several malignancies, DNMT3A has been shown to influence cellular apoptosis [
56], cisplatin resistance [
57], tumor growth, and metastasis [
58]. NCK1-AS1 enhances the expression of NCK1 to aid carcinogenesis and chemo-resistance in ovarian cancer, according to Chang et al. [
59]. Elevated expression of IFRD1 indicates a low survival rate in patients suffering from human colon cancers, according to Lewis et al. [
60]. ABI1 has been linked to the EMT induction in prostate carcinogenesis [
61], ovarian cancer metastasis [
62], and neuroblastoma development, invasion, and metastasis in several investigations [
63]. These risk variables were nearly all linked to cancer patients’ prognosis, even though there is little connected study on LUAD, which will require more research in the future.
In summary, a cluster model and two prognostic nomogram models which were well established to predict the survival (OS and PFS) of LUAD patients from different angles. Two prognostic nomogram models, which combine risk score models and clinical variables, allow clinicians to estimate the OS and PFS of each LUAD patient by entering the score of each parameter. However, unsupervised clustering can analyze data samples based on the inherent characteristics and find a natural grouping among the data [
64,
65]. The unsupervised consensus analysis based on DEAS identified three discernable clusters with different prognosis, and significantly associated with immune and stromal scores (as shown in the figure). This implies that we can roughly estimate the prognosis of patients based on the level of tumor immune infiltration to verify and supplement the results of nomogram.Furthermore, it was found that splicing-derived new epitopes expand the potential of the immunotherapy target space [
66]. The study of Peng et al. also showed that abnormal splicing events contribute to tumor progression through the influence of immune response [
67], which suggests that we should consider splicing events and immunity together to evaluate the prognosis of patients.
There is no denying that our study still has significant flaws. First and foremost, an independent external validation cohort to confirm the performed DEAS-based predictive risk score signature would be preferable. Secondly, some experimental validation would have been beneficial to further confirm our data. These are also our future research directions, and we will verify these results in advanced research settings.
In conclusion, our study developed a risk score model on the basis of 11 prognostic DEASs for predicting survival rate in LUAD patients. Notably, this study also shed light on the complexity as well as the diversity of the immune microenvironment in LUAD patients that might explore the lack of therapeutic success in these individuals.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.