Comparison of tumor microenvironment scores and gene expression profiles in patients with lung adenocarcinoma(LUAD) of different genders
Gene expression data and corresponding clinical information were downloaded from TCGA database, including 304 cases of female LUAD patients and 247 cases of male LUAD patients. The estimate score is ranged from -2358.46 to 4889.83 among female patients with LUAD and from -2328.69 to 4818.63 among male patients with LUAD. The immune score ranged from -943.17 to 3229.35 among female patients with LUAD and from -541.75 to 3441.78 among male patients with LUAD. The stromal score is ranged from -1790.23 to 2097.27 among female patients with LUAD and from -1786.94 to 1722.70 among male patients with LUAD. The estimate, immune and stromal scores of women were higher than those of male (p<0.05) (Figure 1A). The samples were divided into high-and low- score groups by the medium value. Up-regulated genes and down-regulated genes were showed in Figure 1B. By interaction of DEGs in immune and stromal groups, 269 up-regulated genes and 35 down-regulated genes were screened in female patients, and 340 up-regulated genes and 28 down-regulated genes were screened in male patients (Figure 1C).
Associations of the scores with age, TNM stage and prognosis in LUAD patients of different genders
To explore the relationship between estimate, immune or stromal scores and age, TNM stage or survival, the samples were divided into high-score and low-score groups. The analysis indicated that for female patients with LUAD, T1 stage had higher estimate scores than that of T3 stage (p=0.01), and N1 stage had higher estimate score than that of N0 stage (p=0.015) and N2 stage (p=0.018) (Figure 2A). However, in male patients with LUAD, estimate score had no obvious relationship with the T or N stage (Figure 2D). Whether it was a male or female patient with LUAD, high immune score was related to older age (Figure 2B, 2E). Immune score of T1 stage was higher than that of T3 stage in female patients (Figure 2B), but in male patients, immune score of T1 stage was higher than that of T2 and T4 stage (Figure 2E). In addition, N1 stage had higher immune scores than that of N0 stage (p=0.038) and N2 stage (p=0.011) (Figure 2B) for female patients, and clinical stage I had higher immune scores than that of stage III for male patients (p=0.037) (Figure 2E). For female patients with LUAD, the stromal scores of N1 stage were higher than that of N0 (p=0.016) (Figure 2C), and stromal scores of M0 stage or clinical stage I were higher than that of M1 stage (p=0.044) or clinical stage IV (p=0.024) (Figure 2F). To study the correlations of overall survival with estimate, immune or stromal scores, we performed Kaplan-Meier survival analysis (Figure 2G). The results indicated that high estimate score had better overall survival for female patients with LUAD (p=0.039), but they had no significant relationship with prognosis of male patients (p=0.039). The overall survival was not associated with immune or stromal score whether it was a male patient or a female patient (p>0.05).
GO and KEGG analysis of immune-related DEGs
To further explore the functions of immune-related DEGs, we performed GO and KEGG analysis on these genes. The top 5 enriched BPs in female patients with LUAD based on GO analysis were leukocyte proliferation, lymphocyte proliferation, mononuclear cell proliferation, T-cell activation, and lymphocyte differentiation (Figure 3A), and the top 5 enriched BPs in male patients with LUAD based on GO analysis were T-cell activation, leukocyte migration, leukocyte proliferation, lymphocyte proliferation, and regulation of immune effector process (Figure 3E). Circular plot demonstrated the functional interactions between the BP, CC or MF pathways and genes as extracted from GO (Figure 3B, 3F). Moreover, the top 5 enriched BPs in female patients with LUAD based on KEGG were viral protein interaction with cytokine and cytokine receptor, cytokine-cytokine receptor interaction, chemokine signaling pathway, hematopoietic cell lineage, and staphylococcus aureus infection (Figure 3C), and top 5 enriched BPs in male patients with LUAD based on KEGG were cytokine-cytokine receptor interaction, viral protein interaction with cytokine and cytokine receptor, hematopoietic cell lineage, chemokine signaling pathway, and graft-versus-host disease (Figure 3G). Circular plot demonstrated the functional interactions between the BP, CC or MF pathways and genes as extracted from KEGG (Figure 3D, 3H).
PPI network and COX regression analysis of immune-related DEGs
The STRING tool was used to plot PPT networks of immune-related DEGs, which were regenerated by Cytoscape (Figure 4A, 4B). The top 30 genes with most number of adjacent nodes in female and male patients with LUAD were showed in Figure 4C and 4D. We performed univariate Cox regressiom analysis to evaluate the prognostic value of immune-related DEGs in female (Figure 4E) and male (Figure 4F) patients with LUAD (genes with p<0.05 displayed in forest plot). Two-way venn diagram comparing the key genes in female (Figure 4G) and male (Figure 4H) cohorts. CCR2, LCP2, and PTPRC were selected as key prognostic factors of female patients with LUAD. BTK and CCR2 were selected as key prognostic factors of male patients with LUAD.
The expression level of these key TIICs-related genes and their prognostic value in LUAD patients
We further revealed the expression level of these key TIICs-related genes and their prognostic value in LUAD patients. According to the result, the expression level of CCR2 in tumor tissues of female LUAD patients was not significantly different from that in normal tissue (p>0.05) (Figure 5A and 5B), but female LUAD patient with high level of CCR2 exhibited a better overall survival (p=0.001) (Figure 5C). The expression level of LCP2 in tumor tissues of female LUAD patients was lower than that in normal tissues (p<0.05) (Figure 5D), yet there was no difference in the expression of LCP2 in paired tumors and adjacent normal tissues (p>0.05) (Figure 5E). Interestingly, female LUAD patient with a high level of LCP2 exhibited a better overall survival (p=0.033) (Figure 5F). The expression level of PTPRC in tumor tissues of female LUAD patients was lower than that in normal tissues (p<0.05) (Figure 5G and 5H), and female LUAD patient with high level of PTPRC exhibited a better overall survival (p=0.006) (Figure 5I). The expression level of BTK in tumor tissues of male LUAD patients was lower than that in normal tissues (p<0.05) (Figure 5J and 5K), and male LUAD patient with high level of BTK exhibited a better overall survival (p=0.035) (Figure 5L). The expression level of CCR2 in tumor tissues of male LUAD patients was lower than that in normal tissues (p=0.035) (Figure 5M), yet there was no difference in the expression of CCR2 in paired tumors and adjacent normal tissues (p>0.05) (Figure 5N). Interestingly, male LUAD patient with a high level of CCR2 exhibited a better overall survival (p=0.011) (Figure 5O). Associations of these key genes expression with age and TNM stage in LUAD patients of different genders were showed in Figure S1.
CIBERSORT for estimating TIICs components in female and male LUAD microenvironment
We further selected 22 types of immune cell for research on immune landscape of TME in LUAD. The selected major immune cell types included B cells, T cells, NK cells, neutrophils, dendritic cells, mast cells, macrophages, et al., which participated in the regulation process of innate immunity and adaptive immunity. Based on CIBERSORT algorithm, we first analyzed the distribution characteristicsof the 22 types of immune cells in each LUAD sample (i.e., 270 female patients and 227 male patients) (Figure 6A and 6B). Additionally, the correlation between various TIICs in female and male LUAD patients varied from weak to moderate (Figure 6C and 6D). The correlation matrix indicated that CD8 T cells and M0 macrophages had a strong negative correlation with memory resting CD4 T cells (Cor=-0.47) and resting mast cells (Cor=-0.41) in female LUAD patients. CD8 T cells, M2 macrophages, and resting dendritic cells had a strong positive correlation with memory activated CD4 T cells (Cor=0.49), monocytes (Cor=0.42), and resting mast cells (Cor=0.4) in female LUAD patients (Figure 6C). As shown in Figure 6D for male LUAD patients, there was a significant negative correlation between CD8 T cells and memory resting CD4 T cells(Cor=-0.41), and also a significant positive correlation between CD8 T cells and memory activated CD4 T cells (Cor=0.47).
Difference analysis of TIICs in LUAD tumor and adjacent normal tissues
To compare the differences in immune cell infiltration between female and male patients with LUAD, we analyzed the differences in the distribution of TIICs in the TME of female and male patients with LUAD (Figure 7). Compared with paracancerous tissues in female patients, 8 TIICs (memory B cells, plasma cells, memory activated CD4 T cells, follicular helper T cells, regulatory T cells, gamma delta T cells, M1 macrophages, and resting dendritic cells) had a higher proportion in cancerous tissues, and 6 TIICs (memory resting CD4 T cells, resting NK cells, monocytes, M2 macrophages resting mast cells, eosinophils, and neutrophils) had a lower proportion in cancerous tissues (p<0.05) (Figure 7A). Similarly in male LUAD patients, 5 TIICs (plasma cells, memory activated CD4 T cells, follicular helper T cells, regulatory T cells, and M1 macrophages) accounted for a higher proportion in cancerous tissues than that in paracancerous tissues, and the proportion of 4 TIICs (resting NK cells, monocytes, resting mast cells, and neutrophils) in cancerous tissues was lower than that in paracancerous tissues (p<0.05) (Figure 7B). Female patients with LUAD had a higher proportion of memory B cells, while the percentage of T cells CD4 naïve and resting NK cells was lower in female patients (Figure 7C).
In order to further research the influence of the proportions of TIICs on the prognosis of female and male LUAD patients, univariate Cox regression analysis was conducted on the 22 TIICs for female and male patients respectively. The proportion of activated dendritic cells was identified as risk factor for female LUAD patients (Table S1). The proportion of gamma delta T cells, activated NK cells and activated mast cells were identified as risk factors for male LUAD patients (Table S2).
Impact of the key identified genes on TIICs
We further analyzed the effects of the main differentially expressed genes in female (Figure S2) and male (Figure S3) patients with LUAD on immune cell infiltration. For female LUAD patients, the relative proportions of memory B cells (p = 0.004), CD4 memory resting T cells (p = 0.014), resting dendritic cells (p = 0.049) and resting mast cells (p = 0.011) were significantly upregulated in high CCR2 expression group compared with the low CCR2 expression group (Figure S2A). M0 macrophages (p = 0.019) and activated mast cells (p = 0.006) were present in lower proportions in the high CCR2 expression group than in the others (Figure S2A). The relative proportions of CD 8 T cells (p = 0.019), CD4 memory activated T cells (p = 0.005), M1 macrophages (p = 0.009) and neutrophils (p = 0.027) were significantly upregulated in high LCP2 expression group compared with the low LCP2 expression group (Figure S2B). Plasma cells (p = 0.002) and M0 macrophages (p = 0.038) were present in lower proportions in the high LCP2 expression group than in the others (Figure S2B). The relative proportions of memory B cells (p = 0.035), CD4 memory activated T cells (p = 0.013), and eosinophils (p = 0.016) were significantly upregulated in high PTPRC expression group compared with the low PTPRC expression group (Figure S2C). Activated NK cells (p = 0.044) and activated mast cells (p = 0.008) were present in lower proportions in the high PTPRC expression group than in the others (Figure S2C).
For male LUAD patients, the relative proportions of memory B cells (p = 0.022), CD8 T cells (p = 0.034), CD4 memory activated T cells (p < 0.001), monocytes (p < 0.001), resting dendritic cells (p = 0.002), and eosinophils (p < 0.001) were significantly upregulated in high BTK expression group compared with the low BTK expression group (Figure S3A). Plasma cells (p = 0.024), follicular helper T cells (p = 0.031), T cells regulatory (Tregs) (p = 0.039), activated NK cells (p = 0.013), and M0 macrophages (p < 0.001) were present in lower proportions in the high BTK expression group than in the others (Figure S3A). The relative proportions of CD8 T cells (p = 0.008), CD4 memory activated T cells (p < 0.001), monocytes (p = 0.006), and M1 macrophages (p < 0.001) were significantly upregulated in high CCR2 expression group compared with the low CCR2 expression group (Figure S3B). Gamma delta T cells (p = 0.031) and M0 macrophages (p < 0.001) were present in lower proportions in the high CCR2 expression group than in the others (Figure S3B).
Validation of TCGA results with GEO database
In order to verify the prognostic value of the identified genes from TCGA, we used GSE72094 as a validation cohort. Patients were divided into high expression groups and low expression groups respectively according to gene expression. We first compared the survival curves of female LUAD patients and male LUAD patients, which showed that the overall survival rate of female patients is better than that of male patients (Figure 8A). Kaplan-Meier survival curves further confirmed that male LUAD patients with high expression levels of CCR2 (Figure 8B) and BTK (Figure 8C) were present in a better overall survival than those with low expression levels of CCR2 and BTK. Similarly, Kaplan-Meier survival curves further verified that female patients with high expression levels of CCR2 (Figure 8D) LCP2 (Figure 8E), and PTPRC (Figure 8F) were present in a better overall survival than those with low expression levels of CCR2 LCP2, and PTPRC.