This study aims to identify the relationship between immune subgroup and prognosis of lung cancer samples based on gene expression data from multiple immune checkpoints. The gene expression data from TCGA and GEO databases were divided into high - medium - low expression groups according to the gene expression levels of 47 immune checkpoints. The gene expression of the three expression groups from the two databases are of high consistency. In other words, genes with either high or low expression in TCGA database are also similarly expressed in GEO database. It indicates a relatively consistent distribution of genes in the collected samples of the two databases. Therefore, the analysis results can be mutually verified in subsequent studies.
Then by using univariate COX regression analysis to screen for genes associated with patients’ overall survival, we obtained nine genes significantly related to prognosis from TCGA database and 11 prognosis-related genes from GEO database. Among them, CD40LG, a leukocytic coding gene of the cluster of differentiation 40 ligand - CD40L, is significantly correlated with better prognosis in both databases (HR < 1, log rank p < 0.05) [21]. CD40L, which is a type II membrane-related glycoprotein and a member of the tumor necrosis factor (TNF) superfamily [22], is conducive to regulate the immune response and inhibit tumor growth. It produces direct growth-inhibiting effect through apoptosis in CD40L over-expressed malignant tumor cells (such as breast cancer) [23]. The correlation of ICGs expression level in both databases is positive and is showing an obvious aggregation effect, indicating a possibility of coordinated expression relationship between ICGs.
To make clear the relationship between ICGs and other biomarkers for immune checkpoint treatment, we analyzed the relationship between ICGs and tumor mutational burden (TMB), MMRs, and neoantigens, respectively. TMB is a reliable indicator for predicting clinical efficacy of PD-1 inhibitor [24]. The more mutant genes in the tumor tissue, the more likely abnormal proteins will be largely produced. These abnormal proteins are able to activate the body's anti-tumor immune response, improving the sensitivity of immunotherapy. So it is more suitable for patients with high TMB level to continue immunotherapy with PD1 inhibitors [25]. Through analyzing the relationship between ICGs and TMB, we found that the expression of TMB is strongly inversely correlated with that of CD40LG, CD200R1, TNFRSF14 and TNFSF14 (R < 0 and FDR < 0.005). That is, the high expression of CD40LG, CD200R1, TNFRSF14 and TNFSF14 corresponds to the low expression of TMB. Among them, TNFSF14 is a potential cause of unfavourable prognosis. It is inferred that the high expression of TNFSF14 corresponds to the low tumor mutational burden, so the high expression of TNFSF14 may not be suitable for the treatment of immune checkpoint inhibitors. MMRs is an intracellular mismatch repair mechanism. Its loss of function will lead to irreparable DNA replication errors, thus, leading to the high generation of gene mutations [26]. The 5 MMRs gene mutations evaluated in this study are positively correlated with ICGs expression. Among them, CD70, which is also significantly correlated with prognosis which might be a potential predictor of MMR gene mutations. New proteins produced by gene mutation that may activate the immune system and further trigger the immune system to attack cancer cells are called neoantigens [27]. Our study found that CD40LG, CD200R1, TNFSF14 and TNFRSF14 were also significantly negatively correlated with the expressions of neoantigens (R < 0 and FDR < 0.05), which was in line with the extremely correlation of negative expression between TMB and the four genes mentioned above. Our experiment showed that TNFSF14, the unfavourable prognostic gene, was of high expression in patients. While the high expression of TNFSF14 corresponds to low level neoantigens. So in this case, it was not appropriate for patients to be treated with personalized neoantigens [28]. In general, the greater total number of gene mutations (the higher the TMB) in the tumor tissues of patients, the more neoantigens they carry [29]. According to our results, the expressions of CD40LG, CD200R1, TNFSF14 and TNFRSF14 were negatively correlated with TMB and neoantigens. This indicated TMB and neoantigens share the same variation trend, which coincides with what was reported in literature.
In order to reveal the relationship between ICGs and adaptive immune-resistance pathway genes, we analyzed the correlation between CD8A, GZMB, CD68, NOS2 and ICGs. Most of them were found out to be positively and significantly correlated. We obtained the same results in TCGA and GEO databases. This indicated that adaptive immune pathway genes might have a certain regulatory effect on ICGs expressions.
In the relationship between ICGs and clinical features, we found that the expression levels of ICGs in early tumor samples were higher than that in advanced tumor samples, CD200R1 in particular. In the above analysis, we found that CD200R1 were highly correlated with the overall survive of TCGA, and negatively correlated with TMB and neoantigens. In addition, in correlation with the adaptive immune-resistance pathway genes, we found CD200R1 were positively correlated with CD8A, CD68 and GZMB, suggesting that CD200R1 might be the possible cause of expression dysregulation of adaptive immune-resistance pathway genes. Then we analyzed the relationship between IDO1, CD274, CTLA4 and the H/L expression groups and prognosis of CD200R1. based on TCGA database our analysis results showed that prognosis will be the best when it is the group with high expression of CD200R1 and low expressions of IDO1, CD274 and CTLA4, while prognosis will be the worst when it is the group with CD200R1 low expression and low expressions of IDO1, CD274 and CTLA4. Moreover, there was a significant difference between of two groups. It was also confirmed in GEO database that the group with high expression of CD200R1 and low expressions of IDO1, CD274 and CTLA4 had the best prognosis. However, in the previous univariate COX analysis, we did not find a direct correlation between such ICGs as IDO1, CD274 and CTLA4 and patient prognosis. Therefore, integrating the expression levels of PD-l1, IDO1, CTLA4 and CD200R1 could be better evaluated for the immune status of lung cancer patients and explore the prognostic information that cannot be acquired by a single immune checkpoint gene.
However, here are some limitations existing in this study. For example, there is insufficient large-scale lung cancer samples for calculation and analysis, and most of the samples are no more than 1000 [30]. Our research results still need to be verified by more experiments and clinical observation, and on this basis, to develop and design potential immune checkpoint inhibitors [31].