Skip to main content
NCBI home page
Heliyon logoLink to Heliyon
. 2024 Mar 18;10(6):e28174. doi: 10.1016/j.heliyon.2024.e28174

Landscape of infiltrated immune cell characterization in COVID-19

Wei Zheng a, Yuan Zhang b, Guichuan Lai b,⁎⁎, Biao Xie b,⁎
PMCID: PMC10965775  PMID: 38545143

Abstract

Purpose

Although the role of SARS-CoV-2-specfic immune cells has been revealed, a comprehensive understanding of immune patterns remains unknown.

Methods

In this work, unsupervised consensus clustering analysis was used to classify 240 coronavirus disease 2019 (COVID-19) patients into different immune subtypes. Next, we performed differentially expressed analysis between different immune subtypes. Functional enrichment and pathway analyses were employed to reveal the biological significance of these differentially expressed genes (DEGs). Besides, we compared feature score of some DEGs between whole blood and lung tissues. Then, we utilized the “GSVA” algorithm to construct an immune cell infiltrating (ICI) tool based on the categories of these DEGs. Finally, we developed a nomogram associated with severity of COVID-19.

Results

As a result, we identified two immune subtypes, and 238 DEGs which mainly participated in some immune-related functions and the COVID-19 pathway. Most importantly, the 238 DEGs could reflect the characterization of immune patterns in lung tissues. ICI scores were markedly negative associated with immune scores. It was worth noting that ICI score was a strong indicator for severity of COVID-19 and could accurately predict the severity of COVID-19.

Conclusion

Our findings could provide more valuable strategies for the management of COVID-19.

Keywords: Coronavirus disease 2019, Immune infiltrating cells, Immune response, Gene expression, Viral infection, Severity

1. Introduction

Coronavirus disease 2019 (COVID-19) is a special pneumonia caused by novel coronavirus and infectious patients mainly experience fever, dry cough, and dizziness [1]. As a major public health event, an increasing number of people are facing the crisis brought by COVID-19 on their health and life [2]. COVID-19 is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). In view of the high similarity between SARS-CoV transcriptome and SARS-CoV-2, the data of immune regulation, signaling pathways and pro-inflammatory cytokines obtained in SARS-CoV infection can be effectively applied to the study of COVID-19. The application of this data provides a useful platform for future research on drugs and vaccines [3]. Although nowadays available SARS-Cov-2specific vaccines have been developed to protect against the virus, not all patients can benefit from these vaccines due to the heterogeneity of vaccine responses. A recent review pointed that genomic biomarkers were strongly associated the severity of COVID-19, indicating omics methods were necessary for predicting the progression of COVID-19 patients [4]. In addition, the study found neprilysin, Angiotensin-converting enzyme 2 (ACE2) and Carbonic anhydrase are potential factors associated with the severity of COVID-19 [5].

Like tumors, immune microenvironment also influences the development of COVID-19. For example, it was found that the abundance of innate lymphoid cells was related to adverse outcomes, leading to disease tolerance in COVID-19 in an observational study [6]. Kent et al. found that accumulation of T cell immunity contributed to reduction of continual infection, especially for infection with SARS-CoV-2 variants [7]. In summary, cellular immunity determines the degree of infection and closely correlates with the severity of COVID-19 [8]. These findings can support the fact that SARS-CoV-2-specific innate immune microenvironment impacts the progression.

Despite the fact that SARS-CoV-2-specific immune cells are crucial for identifying the disease severity, many patients continue to have a poor prognosis. Increasing evidence suggests that immune subtyping can identify different clinical features, including the severity and stage of COVID-19, and guide COVID-19 treatment [9,10]. Hence, a tool used for assessing the immune patterns of COVID-19 needs to be developed.

In this work, we firstly classified 240 COVID-19 patients into different immune subtypes based on 22 immune cells. Then, we determined potential proteins differentially distributed in these immune subtypes. According to these potential proteins, we used the gene set variation analysis (GSVA) algorithm to construct an immune cell infiltrating (ICI) tool, and further investigated the relationship between ICI tool and immune microenvironment, and some important clinical features, including age, hospital-free days post 45 (HFD45), ventilator free days (VFDs) and severity. We aimed to evaluate immune patterns of COVID-19 patients and to construct a novel tool associated with these immune patterns.

2. Materials and methods

2.1. Datasets used in this study

In this work, we collected RNA sequence and corresponding clinical data of four datasets including GSE152641 [11], GSE157103 [12], GSE171110 [13] and GSE196822 [14] from GEO (http://www.ncbi.nlm.nih.gov/geo/) database. In this work, we chose samples based on the following criteria: 1. Patients who have been confirmed as COVID-19 without other infectious diseases were considered as enrolled samples. 2. Whole blood was used as the source of RNA extraction. 3. Patients with complete gene expression were included in this study. Therefore, we excluded these bacterial co-infection patients in GSE196822 according to these criteria. All gene expressions of the four datasets were count-based measurement. Next, we transformed the count data into transcript per million (TPM) and the final format was presented by log2 (TPM+1). Finally, the four datasets were merged and batch effects were corrected by “sav” packages [15]. The principal component analysis (PCA) method was used to evaluate whether the batch effects have been corrected. The detailed information of enrolled samples was displayed in Table S1 and the workflow design was shown in Fig. 1.

Fig. 1.

Fig. 1

Open in a new tab

The workflow design in this study. Unsupervised consensus clustering analysis was used to identify different immune subtypes, and the “GSVA” algorithm was used to construct an immune cell infiltration (ICI) tool to accurately predict the severity of COVID-19.

2.2. Assessment of immune microenvironment

“CIBERSORT” (https://cibersort.stanford.edu/) algorithm was used to calculate 22 representative immune cells based on the LM22 profile and normalized expression matrix [16]. The CIBERSORT algorithm is a deconvolution method for characterizing cell composition of complex tissues from their gene expression profiles, which has been widely used to evaluate the immune microenvironment of disease [[17], [18], [19]]. The LM22 contains 547 genes that distinguish 22 human hematopoietic cell phenotypes. After running this algorithm, we chose the samples with p < 0.05 for further analyses. Besides, we performed the “ESTIMATE” algorithm to quantify the immune microenvironment of COVID-19 patients.

2.3. Unsupervised consensus clustering analysis

To classify COVID-19 patients into different immune subtypes, we used the unsupervised consensus clustering analysis via “ConsensusClusterPlus” package on basis of proportion of 22 immune cells [20]. The parameter setting mainly included 80% of resampling proportion, 1000 repetitions to effectively reduce sampling error and improve the stability of results; Maximum clustering numbers to seven, and in the clustering process, the “pam” algorithm is used to optimize the clustering results. At the same time, the “Euclidean” distance was used as a distance measure to better measure the similarity between immune cell ratios. Subsequently, the optimal clustering number was determined through the lowest proportion of ambiguous clustering value (PAC). The proportion of PAC algorithm is a method used to infer the optimal number of clusters (K) in consensus clustering. It determines the flatness of the middle segment in the empirical cumulative distribution function (CDF) plot of the consensus index value. A lower value of PAC indicates that the optimal K value is more pronounced and thus helps in determining the optimal number of clusters [21].

2.4. Functional enrichment and pathway analyses

The “limma” package was used to screen the differentially expressed genes (DEGs) among different subtypes with the criteria of “|log2FC| > 1”and “p-adjust <0.05” [22]. To reveal the biological functions of these candidate genes in COVID-19 pathogenesis, “clusterprofiler” package was used to conduct “GO” enrichment and “KEGG” pathway analyses [23]. Differential functions and pathways were selected via “BH” method with an adjusted p value < 0.05.

2.5. Comparison with feature score of DEGs in lung tissues

We downloaded GSE150316 and GSE183533 datasets from GEO database. Next, we respectively collected 52 COVID-19 patients in GSE150316 and 31 COVID-19 patients in GSE183533. Then, CIBERSORT algorithm was used to calculate the fraction of 22 immune cells in the two datasets. Likewise, we performed unsupervised consensus clustering analysis with the same parameter settings and selective criteria. The “edgeR” package was used to screen the DEGs among different subtypes with the criteria of “|log2FC| > 1” and “FDR <0.05” [24]. Finally, to compare the results more clearly, we conducted the PCA and the principal component 1 was used to reflect the feature score of certain gene sets. We defined PC1(A) as feature score of DEGs from whole blood, and termed PC1 (B) as feature score of DEGs from lung tissues in GSE150316 and PC1 (C) as feature score of DEGs from lung tissues in GSE183533. The Spearman analyses were used to reveal the correlation of PC1 (A) with PC1 (B) and PC1 (C).

2.6. Development of an ICI tool

We utilized the “GSVA” algorithm to construct an ICI score to quantify the characterization of these DEGs associated with immune patterns by “GSVA” package [25]. In this study, we defined the genes positively correlated with cluster signatures as gene signature α while the genes negatively correlated with cluster signatures were termed as gene signature β. The ICI score of each patient was ultimately calculated according to the following formula: ICI score = GSVAα-GSVAβ.

2.7. Construction of nomogram and calibration plots

In GSE157103, we compared the ability of ICI score and some clinical features with fewer missing samples for identifying the severity of COVID-19 using “pROC” package. Then, the valuables with higher recognizable power were employed to construct a nomogram associated with the probability of ICU by “rms” package. Finally, calibration curve was further to assess the identifiable value of ICI score.

2.8. ICI score and SARS-Cov-2-specific vaccines

To evaluate the relationship between ICI score and SARS-Cov-2-specific vaccine, we downloaded GSE173488 and GSE200274 from GEO database. In GSE173488, eight COVID-19 patients stimulated with SARS-CoV-2 spike protein (S-protein), eight COVID-19 patients stimulated with lipopolysaccharide (LPS) and eight COVID-19 patients without stimulation were selected into this study. Additionally, six unvaccinated COVID-19 patients and six vaccinated COVID-19 patients stimulated with SARS-CoV-2 spike protein, lipopolysaccharide and unstimulation in GSE200274 were introduced into this work.

2.9. Statistical analysis

In this work, R4.0.3 was used to finish analyses above. The Wilcoxon analysis was required for differences between different immune subtypes and clinical groups. The Spearman analysis was applied to reveal the correlation of feature score of 238 DEGs with feature score of DEGs from lung tissues, correlation of ICI scores with immune cells, immune scores, stromal scores and some clinical features. The ROC was used to evaluate the identifiable power of this presently constructed ICI score in severity of COVID-19. Finally, two-side and p < 0.05 were regarded as a significant difference.

3. Results

3.1. Relationship between immune-related cells and severity of COVID-19

We explored the relationship between the level of infiltration of 22 immune cells and the occurrence of ICU using one-way logistic regression and multifactorial logistic regression used in this study. The unifactorial results found that the infiltration levels of CD8 + T cells, activated memory CD4+ T cells, resting NK cells, and Monocytes were protective factors for the severity of COVID-19, while the infiltration level of Neutrophils was a risk factor for the severity of COVID-19. However, with multifactorial results we only found that only the level of CD8+ T cells infiltration was a protective factor for COVID-19 severity (Table S2).

3.2. Immune subtypes of COVID-19

PCA indicated that the batch effects caused by merging four datasets from different platforms have been weakened (Fig. 2A and B). Based on the fraction of 22 immune cells, we classified 240 COVID-19 patients into two immune subtypes (Fig. 2C, Fig S1, and Table S3). Totally, we found that 13 immune cells were differentially distributed between cluster 1 and 2. In particular, patients in cluster 1 had a higher immune infiltrating proportion of memory B cells, CD8+ T cells, resting memory CD4+ T cells, resting NK cells, Monocytes and Eosinpphils but lower ratio of plasma cells, naïve CD4+ T cells, follicular helper T cells, gamma delta T cells, activated Dendritic cells as well as Neutrophils than those patients in cluster 2 (Fig. 2D). There was no statistical difference in stromal scores between cluster 1 and 2 (Fig. 2E). But the immunity scores were found to be higher in cluster 1 (Fig. 2F). Next, we screened out 238 DEGs comprising of 75 up-regulated genes and 163 down-regulated genes between cluster 1 and 2 (Fig. 3A, Table S4). The biological activities of the 238 DEGs mainly included some immune-related functions, such as T cell activation, T cell receptor signaling pathway, B cell activation, MHC protein complex and immune receptor binding (Fig. 3B). Most importantly, we found that the 238 DEGs were involved in Coronavirus disease-COVID-19 except some pathways associated with immunity (Fig. 3C). To further investigate the roles of these genes related to immune patterns in COVID-19, we divided 240 COVID-19 patients into three clusters according to expressions of 238 DEGs (Fig. 3D, Fig S2, and Table S5). During this process, 16 immune cells were found to be differentially expressed in three groups. Neutrophils, naïve B cell, naïve CD4+ T cells, gamma delta T cells, follicular helper T cells, activated Dendritic cells and M0 macrophages were more enriched in cluster 3 while memory B cells, CD8+ T cells, resting memory CD4+ T cells, Tregs, resting NK cells and Monocytes were more gathered in cluster 2 (Fig. 3E). In addition, we found that patients in cluster 2 had the highest immune scores while patients in cluster 1 had the highest stromal scores (Fig. 3F). In contrast, the lowest immune and stromal scores occurred in cluster 3 (Fig. 3F).

Fig. 2.

Fig. 2

Open in a new tab

Immune correlation analysis of two molecular subtypes of COVID-19. (A) PCA for GSE152641, GSE157103, GSE171110 and GSE196822 before removing the batch effects. (B) PCA for GSE152641, GSE157103, GSE171110 and GSE196822 after removing the batch effects. (C) Consensus matrix of two immune subtypes. (D) Comparison of 22 immune cells between two immune subtypes. (E) Box plots of stromal scores between two immune subtypes. (F) Box plots of immune scores between two immune subtypes(E) and Data in (D–F) were analyzed by Wilcoxon test; ns, no significance; *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001.

Fig. 3.

Fig. 3

Open in a new tab

Differential gene expression analysis reveals immune-related patterns and distinct clusters in COVID-19 patients. (A) Volcano plot of 238 DEGs. (B) GO enrichment analysis of 238 DEGs. (C) KEGG pathway analysis of 238 DEGs. (D) Consensus matrix of three immune patterns. (E) Comparison of 22 immune cells among three immune patterns. (F) Comparison of stromal and immune scores among three immune patterns. Data in (E-F) were analyzed by Wilcoxon test; ns, no significance; *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001.

3.3. Comparison of severity of different subtypes

To investigate whether there was a statistical association between patients with different immune subtypes and the severity of the disease, we found a statistically significant difference using the chi-square test with a greater proportion of ICUs occurring in patients with subtype 2 than in those with subtype 1 (p = 0.0027) (Fig S3).

3.4. Comparison with feature score of DEGs in lung tissues

After calculating 22 immune cells of these patients using CIBERSORT algorithm, we totally obtained 47 COVID-19 patients in GSE150316 and 20 COVID-19 patients in GSE183533 according to p-value <0.05. The results of unsupervised consensus clustering analysis showed that 47 patients were divided into cluster 1 (28 samples) and cluster 2 (19 samples) in GSE150136, and 20 patients were also classified into cluster 1 (13 samples) and cluster 2 (7 samples) in GSE183533. Following by differentially expressed analyses between cluster 1 and cluster 2, 69 DEGs in GSE150316 and 1281 DEGs in GSE183533 were selected for subsequent analyses. After performing the Spearman analysis, we found that PC1 (A) was strongly correlated with PC1 (B) (Fig. 4A) and PC1 (C) (Fig. 4B), indicating that 238 DEGs in this study could also better reflect characterization of immune patterns in lung tissues.

Fig. 4.

Fig. 4

Open in a new tab

The Spearman analysis showing consistent immune patterns based on gene expression in lung tissues across different patient cohorts. (A) The correlation of feature score of 238 DEGs with feature score of 69 DEGs in GSE150316. (B) The correlation of feature score of 238 DEGs with feature score of 1281 DEGs in GSE183533.

3.5. Calculation and characterization of ICI scores

The 74 DEGs positively correlated with the cluster signatures were considered as gene signature α and the 164 DEGs negatively correlated with the cluster signatures were considered as gene signature β. Finally, we acquired an ICI tool to quantify the characterization of these DEGs based on the following formula: ICI score = GSVAα-GSVAβ Next, the results of CIBERSORT suggested that Tregs, CD8+ T cells, resting memory CD4+ T cells, activated memory CD4+ T cells and resting NK cells were negatively related to ICI scores (Fig. 5A). However, a stronger positive correlation between ICI scores and gamma delta T cells, Neutrophils and M0 Macrophages was noticed (Fig. 4A). Interestingly, we found that ICI scores were negatively associated with immune scores though the correlation of ICI scores with stromal scores had no statistical significance (Fig. 5B, C, D, E, F, G, H, I, J, K).

Fig. 5.

Fig. 5

Open in a new tab

Associations between ICI scores and immune cell infiltration. (A) The correlation of ICI scores with 22 immune cells in five datasets. (B) The correlation of ICI scores with stromal scores in meta cohort. (C) The correlation of ICI scores with immune scores in meta cohort. (D–G) The correlation of ICI scores with stromal scores in GSE152641, GSE196822, GSE171110 and GSE157103. (H–K) The correlation of ICI scores with immune scores in GSE152641, GSE196822, GSE171110 and GSE157103.

3.6. ICI score and clinical features

In GSE157103, we found that ICI scores were positively associated with acute physiologic assessment and chronic health evaluation (APACHE II) scores, C-reactive protein (crp), sequential organ failure assessment (SOFA) scores and age, but negatively associated with HFD45 and VFDs (Fig. 6A, B, C, D, E, F). ICI scores were positively related to age in GSE152641 but no differential correlation was found in GSE196822 (Fig. 6I–K). Meanwhile, the ICI scores were not statistically differential in gender and diabetes groups (Fig. 6G, H, J, L). In terms of severity of COVID-19, patients in intensive care unit (ICU) had more ICI scores than those not in ICU (Fig. 7A). In addition, severe patients had higher ICI scores compared to those mild patients (Fig. 7B).

Fig. 6.

Fig. 6

Open in a new tab

Associations of ICI scores with clinical features and demographics in COVID-19 patients. (A–F) The correlation of ICI scores with APACHE II score, crp, HFD45, SOFA scores, VFDs and age in GSE157103. (G) Distribution of ICI scores between patients with diabetes and patients without diabetes in GE157103. (H) Distribution of ICI scores between female patients and male patients in GE157103. (I) The correlation of ICI scores with age in GSE152641. (J) Distribution of ICI scores between female patients and male patients in GSE152641. (K) The correlation of ICI scores with age in GSE196822. (L) Distribution of ICI scores between female patients and male patients in GSE196822. Data in (G, H, J and L) were analyzed by Wilcoxon test; ns, no significance.

Fig. 7.

Fig. 7

Open in a new tab

Associations of ICI scores with clinical features in COVID-19 patients. (A) Comparison of ICI scores between ICU patients and non-ICU patients in GSE157103. (B) The association between ICI score and progression of COVID-19 in GSE196822. Data in (A) were analyzed by Wilcoxon test; ****p < 0.0001.

3.7. ICI score was a strong indicator for identifying the severity of COVID-19

To evaluate the power of ICI score in identifying the severity of COVID-19 patients when considering other complete clinical features, ROC was used to reflect the recognizable abilities of these indicators. As a result, the top three variables (HFD45, VFDs and ICI score) with higher identifiable value were utilized to construct a nomogram associated with probability of ICU (Fig. 8A). The nomogram analysis indicated that ICI score was more important in predicting the probability of ICU compared to HFD45 and VFDs, which was lined with ROC analysis (Fig. 8B). Finally, the calibration plot suggested that a higher fitting degree was found between predicated probability and actual probability (Fig. 8C).

Fig. 8.

Fig. 8

Open in a new tab

Evaluation of the ICI score as a predictor of COVID-19 severity using a nomogram and ROC analysis. (A) ROC of some clinical indicators and ICI score for identifying the ICU patients. (B) Nomogram for predicting the probability of ICU. (C) Calibration curves of the nomogram for predicting the probability of ICU.

3.8. The role of ICI score in selection of potential SARS-Cov-2-specific vaccine

ICI scores were significantly decreased in patients stimulated with LPS and patients stimulated with LPS had lower ICI scores than patients stimulated with S-protein and receiving no stimulation (Fig. 9A). Noteworthy, ICI scores were gradually reduced in vaccinated patients stimulated with LPS compared with unvaccinated patients (Fig. 9B). However, no statistical differences were discovered between vaccinated patients stimulated with S-protein or without stimulation and corresponding unvaccinated patients (Fig. 9C and D).

Fig. 9.

Fig. 9

Open in a new tab

Relationship Between ICI Scores and SARS-Cov-2 Specific Vaccines. (A) The distribution of ICI scores in patients stimulated with LPS, S-protein and without stimulation. (B) Box plot of ICI scores between vaccinated patients stimulated with LPS and unvaccinated patients stimulated with LPS. (C) Box plot of ICI scores between vaccinated patients stimulated with S-protein and unvaccinated patients stimulated with S-protein. (D) Box plot of ICI scores between vaccinated patients without stimulation and unvaccinated patients stimulated without stimulation.

4. Discussion

Although an increasing number of studies have focused on immune compositions of COVID-19 patients, lack of comprehensive studies on immune patterns of COVID-19 still exists. In this study, we constructed a novel ICI tool, which can quantify the immune patterns and accurately predict the severity of COVID-19.

Firstly, Tregs, gamma delta T cells, CD8+ T cells, resting memory CD4+T cells, resting NK cells, Neutrophils and Monocytes were associated with the immune patterns and ICI scores in this work. Lower level of Tregs could aggravate the damage in severe COVID-19 patients and Tregs played an immune-related role in the disease progression of COVID-19 [26,27]. It was demonstrated that gamma delta T cells served as a determining function in severity of COVID-19, and thus were often considered as the main indicator for COVID-19 symptoms [28,29]. Most importantly, their promising potential in targeting to infection of virus has been explored. CD8+ T cells exerted a protective ability that can rapidly clean and kill the virus [30]. Additionally, some studies have shown that higher CD8+ T cells were related to a reduced risk for COVID-19 and more CD8+ T cells were predominated in mild patients not in severe patients [31,32]. More recently, a machine learning analysis suggested resting NK cells were the most significant predictor of severity of COVID-19 [33]. Neutrophils were often regarded as a monitor for severe symptom of COVID-19 and are also more likely to be targeted therapeutic cells [34,35]. Loss of non-classical Monocytes appeared in severe COVID-19 patients and Monocytes indirectly enabled the development of severe respiratory failure [36,37]. Except for these important immune cells associated with severity of COVID-19, we also found a strong correlation of immune scores with immune patterns, and ICI scores. Available evidences have proved that low immunity was responsible for severity because of the reduced T cell numbers in severe patients, leading to the formation of immune suppressive microenvironment [[38], [39], [40], [41], [42], [43]]. Overall, the above findings could strongly support the relationship between severity of COVID-19 and immune patterns, as well as ICI score from the immune microenvironment sight.

DEGs correlated with immune patterns participated in various immune-related activities and pathways, indicating the influence of indispensable immune regulation on COVID-19. Most importantly, we found that these DEGs were involved in the Coronavirus disease-COVID-19 pathway, highlighting their roles in the development of COVID-19. The results of GO and KEGG emphasized the necessity of these candidate proteins for COVID-19. A similar study conducted differential expression analysis of RNA-Seq data to identify all genes involved in the progression of COVID-19, revealed the correlation between these DEGs and COVID-19 phenotypes, and provided support for our findings [44]. In addition, the authors also observed the characteristics of coronavirus disease-COVID-19, T cell receptor signaling and ribosomal pathways when paying attention to the combined effects of COVID-19 and systemic lupus erythematosus molecular subtypes, which also confirmed that specific regulatory pathways were significantly associated with COVID-19 [45]. To verify whether these 238 DEGs derived from whole blood tissues could reflect immune profile in lung tissues, we compared the feature score of DEGs in whole blood with feature score of DEGs in lung tissues through PCA algorithm. The results showed a strongly positive correlation between feature score of DEGs in whole blood with feature score of DEGs in lung tissues, suggesting that our DEGs could also reflect the immune characterization in lung tissues. In disease researches, “GSVA” algorithm was usually used for calculating the scores of pathways, proportion of immune cells and scores of certain gene sets [[46], [47], [48]]. Interestingly, “GSVA” algorithm was utilized to predict melanoma patient’ response to immune checkpoint inhibitors [48]. These studies can favor the prevalence of “GSVA” algorithm in diseases. Therefore, ICI score originated from these DEGs based on “GSVA” algorithm was used to integrate these proteins and quantify the effect caused by immune patterns.

In this study, the comparison of ICI score with some clinical traits indicates that ICI score may be a stronger indicator for predicting the severity of COVID-19. Firstly, as we know, high APACHE II scores, crp and SOFA scores, low HFD45 and ventilator free days reflected the severity of disease. Secondly, most studies have discovered that age was a risk factor for COVID-19 [49,50]. It was worth noting that older individuals expressed uncoordinated adaptive immunity when infected by COVID-19 [51]. Thirdly, ICI scores were statistically up-regulated in severe patients and ICU patients. Finally, we found that ICI score can independently and accurately identify the severity of COVID-19 considering incorporation with other clinical features, accounting for the severity of COVID-19 to a great extent. In a word, we thought that this presently developed ICI tool may be a successful substitute for some important clinical biomarkers in recognizing the severity.

Our study initially found that the newly constructed immune marker ICI has the potential to predict the progression of COVID-19 and the response to COVID-19-specific vaccines, and in the clinical setting, these findings can be effectively translated to some extent into the management of COVID-19 outbreaks, for example, an increasing number of studies have utilized gene expression to predict COVID-19 severity, which can provide a theoretical basis for the theoretical basis for prognostic management of COVID-19 patients [[52], [53], [54]]. In addition, the prediction of vaccine reactions can also be recognized [55,56].

There were still some limitations though we evaluated the characterization of immune patterns in COVID-19 and developed a corresponding signature through bioinformatics methods. First, the present ICI tool just predicted the severity of COVID-19 but not the prognosis due to the absence of survival data. Then, the datasets used for the merger are from different countries, resulting in data from different source platforms. This can therefore lead to non-harmonized clinical information, e.g. information on severity associated with COVID-19 is missing in GSE152641, GSE171110. Finally, it is worth noting that according to related research, ACE2 is expressed in multiple human organs, which means that the COVID-19 virus can infect other organs other than the lungs [57]. This finding suggests that we need to consider the effects of other organs when studying the immune patterns of COVID-19 and assessing the severity of the disease. However, the specific effects and mechanisms still need further research and exploration.Future studies need to integrate a large transcriptome cohort of COVID-19 and use artificial intelligence to investigate the relationship further systematically between gene expression, immune microenvironment and COVID-19 progression and treatment, to provide theoretical reference for the prognostic management of COVID-19 patients. Since we can only get the correlation between ICI-related markers and COVID-19 progression and vaccine response in the current study, and the causal relationship cannot be confirmed, we can use the most popular causal inference methods (e.g., Mendelian randomization) to genetically infer whether the SNP loci of the immune-related genes we obtained are causally related to the COVID-19 progression and vaccine response on this basis. progression and vaccine response causally.

5. Conclusions

In this work, we identified several immune patterns of COVID-19 and developed an ICI tool associated with immune microenvironment. Besides, this constructed ICI tool can accurately predict the severity of COVID-19. We believe this study can help to form more preventive strategies and provide targeted therapies for COVID-19.

Ethics declarations

Review and/or approval by an ethics committee was not needed for this study because the study did not involve human or animal subjects.

Consent for publication

Not applicable.

Funding

This research was funded by the National Youth Science Foundation Project, China grant number (82204159), and Science and Technology Research Program of Chongqing Municipal Education Commission, China (Grant No. KJQN202300423).

Data availability statement

The results shown here are based upon data generated by GEO (http://www.ncbi.nlm.nih.gov/geo/) and CIBERSORT (https://cibersort.stanford.edu/) databases. The login numbers of the GEO database are GSE152641, GSE157103, GSE171110 and GSE196822.

CRediT authorship contribution statement

Wei Zheng: Writing – original draft, Visualization, Formal analysis, Conceptualization. Yuan Zhang: Writing – original draft, Visualization, Formal analysis, Conceptualization. Guichuan Lai: Writing – review & editing, Supervision, Data curation, Conceptualization. Biao Xie: Writing – review & editing, Supervision, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Footnotes

Appendix A

Supplementary data to this article can be found online at.https://doi.org/10.1016/j.heliyon.2024.e28174

Contributor Information

Guichuan Lai, Email: 2020111425@stu.cqmu.edu.cn.

Biao Xie, Email: kybiao@cqmu.edu.cn.

Appendix A. Supplementary data

The following are the Supplementary data to this article.

Multimedia component 1
mmc1.docx (18.1MB, docx)
Multimedia component 2
mmc2.xlsx (54.6KB, xlsx)
Multimedia component 3
mmc3.xlsx (12KB, xlsx)
Multimedia component 4
mmc4.xlsx (13KB, xlsx)
Multimedia component 5
mmc5.xlsx (1.4MB, xlsx)
Multimedia component 6
mmc6.xlsx (14KB, xlsx)

References

  • 1.Dhochak N., Singhal T., Kabra S.K., et al. Pathophysiology of COVID-19: why children fare better than adults? Indian J. Pediatr. 2020 Jul;87(7):537–546. doi: 10.1007/s12098-020-03322-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Koelle K., Martin M.A., Antia R., et al. The changing epidemiology of SARS-CoV-2. Science. 2022 Mar 11;375(6585):1116–1121. doi: 10.1126/science.abm4915. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Zolfaghari Emameh R., Nosrati H., Eftekhari M., Falak R., Khoshmirsafa M. Expansion of single cell transcriptomics data of SARS-CoV infection in human bronchial epithelial cells to COVID-19. Biol. Proced. Online. 2020;22:16. doi: 10.1186/s12575-020-00127-3. Published 2020 Jul 23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Gorog D.A., Storey R.F., Gurbel P.A., T, et al. Current and novel biomarkers of thrombotic risk in COVID-19: a consensus statement from the international COVID-19 thrombosis biomarkers colloquium. Nat. Rev. Cardiol. 2022 Jul;19(7):475–495. doi: 10.1038/s41569-021-00665-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Zolfaghari Emameh R., Falak R., Bahreini E. Application of system biology to explore the association of neprilysin, angiotensin-converting enzyme 2 (ACE2), and carbonic anhydrase (CA) in pathogenesis of SARS-CoV-2. Biol. Proced. Online. 2020;22:11. doi: 10.1186/s12575-020-00124-6. Published 2020 Jun 19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Silverstein N.J., Wang Y., Manickas-Hill Z., et al. Innate lymphoid cells and COVID-19 severity in SARS-CoV-2 infection. Elife. 2022 Mar 11;11 doi: 10.7554/eLife.74681. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Kent S.J., Khoury D.S., Reynaldi A., et al. Disentangling the relative importance of T cell responses in COVID-19: leading actors or supporting cast? Nat. Rev. Immunol. 2022 Jun;22(6):387–397. doi: 10.1038/s41577-022-00716-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Moga E., Lynton-Pons E., Domingo P. The robustness of cellular immunity determines the fate of SARS-CoV-2 infection. Front. Immunol. 2022 Jun 27;13 doi: 10.3389/fimmu.2022.904686. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Ren X., Wen W., Fan X., et al. COVID-19 immune features revealed by a large-scale single-cell transcriptome atlas. Cell. 2021 Apr 1;184(7):1895–1913.e19. doi: 10.1016/j.cell.2021.01.053. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Karaderi T., Bareke H., Kunter I., et al. Host genetics at the intersection of autoimmunity and COVID-19: a potential key for heterogeneous COVID-19 severity. Front. Immunol. 2020 Dec 22;11 doi: 10.3389/fimmu.2020.586111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Thair S.A., He Y.D., Hasin-Brumshtein Y., et al. Transcriptomic similarities and differences in host response between SARS-CoV-2 and other viral infections. iScience. 2020;24(1) doi: 10.1016/j.isci.2020.101947. Published 2020 Dec 16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Overmyer K.A., Shishkova E., Miller I.J., et al. Large-Scale multi-omic analysis of COVID-19 severity. Cell Syst. 2021;12(1):23–40. e7. doi: 10.1016/j.cels.2020.10.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Lévy Y., Wiedemann A., Hejblum B.P., et al. CD177, a specific marker of neutrophil activation, is associated with coronavirus disease 2019 severity and death [published correction appears in iScience. 2022 Dec 24;26(1):105715] iScience. 2021;24(7) doi: 10.1016/j.isci.2021.102711. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Banerjee U., Rao P., Reddy M., et al. A 9-gene biomarker panel identifies bacterial coinfections in culture-negative COVID-19 cases. Mol Omics. 2022;18(8):814–820. doi: 10.1039/d2mo00100d. Published 2022 Sep. 26. [DOI] [PubMed] [Google Scholar]
  • 15.Leek J.T., Johnson W.E., Parker H.S., et al. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 2012 Mar 15;28(6):882–883. doi: 10.1093/bioinformatics/bts034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Newman A.M., Liu C.L., Green M.R., et al. Robust enumeration of cell subsets from tissue expression profiles. Nat. Methods. 2015 May;12(5):453–457. doi: 10.1038/nmeth.3337. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Kim Y., Kang J.W., Kang J., et al. Novel deep learning-based survival prediction for oral cancer by analyzing tumor-infiltrating lymphocyte profiles through CIBERSORT. OncoImmunology. 2021 Mar 29;10(1) doi: 10.1080/2162402X.2021.1904573. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Tian X., Xu W., Wang Y., et al. Identification of tumor-infiltrating immune cells and prognostic validation of tumor-infiltrating mast cells in adrenocortical carcinoma: results from bioinformatics and real-world data. OncoImmunology. 2020 Jun 23;9(1) doi: 10.1080/2162402X.2020.1784529. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Han X., Cao W., Wu L., et al. Radiomics assessment of the tumor immune microenvironment to predict outcomes in breast cancer. Front. Immunol. 2022 Jan 3;12 doi: 10.3389/fimmu.2021.773581. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Wilkerson M.D., Hayes D.N. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics. 2010 Jun 15;26(12):1572–1573. doi: 10.1093/bioinformatics/btq170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Șenbabaoğlu Y., Michailidis G., Li J.Z. Critical limitations of consensus clustering in class discovery. Sci. Rep. 2014 Aug 27;4:6207. doi: 10.1038/srep06207. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Ritchie M.E., Phipson B., Wu D., et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015 Apr 20;43(7) doi: 10.1093/nar/gkv007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Wu T., Hu E., Xu S., et al. clusterProfiler 4.0: a universal enrichment tool for interpreting omics data. Innovation. 2021 Jul 1;2(3) doi: 10.1016/j.xinn.2021.100141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Robinson M.D., McCarthy D.J., Smyth G.K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010 Jan 1;26(1):139–140. doi: 10.1093/bioinformatics/btp616. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Hänzelmann S., Castelo R., Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinf. 2013 Jan 16;14:7. doi: 10.1186/1471-2105-14-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Qin C., Zhou L., Hu Z., et al. Dysregulation of immune response in patients with coronavirus 2019 (COVID-19) in Wuhan, China. Clin. Infect. Dis. 2020 Jul 28;71(15):762–768. doi: 10.1093/cid/ciaa248. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Sadeghi A., Tahmasebi S., Mahmood A., et al. Th17 and Treg cells function in SARS-CoV2 patients compared with healthy controls. J. Cell. Physiol. 2021 Apr;236(4):2829–2839. doi: 10.1002/jcp.30047. [DOI] [PubMed] [Google Scholar]
  • 28.von Massow G., Oh S., Lam A., et al. Gamma delta T cells and their involvement in COVID-19 virus infections. Front. Immunol. 2021 Oct 29;12 doi: 10.3389/fimmu.2021.741218. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Sanz M., Mann B.T., Chitrakar A., et al. Defying convention in the time of COVID-19: insights into the role of γδ T cells. Front. Immunol. 2022 Aug 11;13 doi: 10.3389/fimmu.2022.819574. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.ette A., Crotty S. Adaptive immunity to SARS-CoV-2 and COVID-19. Cell. 2021 Feb 18;184(4):861–880. doi: 10.1016/j.cell.2021.01.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Peng Y., Mentzer A.J., Liu G., et al. Broad and strong memory CD4+ and CD8+ T cells induced by SARS-CoV-2 in UK convalescent individuals following COVID-19. Nat. Immunol. 2020 Nov;21(11):1336–1345. doi: 10.1038/s41590-020-0782-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Schulien I., Kemming J., Oberhardt V., et al. Characterization of pre-existing and induced SARS-CoV-2-specific CD8+ T cells. Nat Med. 2021 Jan;27(1):78–85. doi: 10.1038/s41591-020-01143-2. [DOI] [PubMed] [Google Scholar]
  • 33.García-Hidalgo M.C., Peláez R., González J., et al. Genome-wide transcriptional profiling of pulmonary functional sequelae in ARDS- secondary to SARS-CoV-2 infection. Biomed. Pharmacother. 2022 Oct;154 doi: 10.1016/j.biopha.2022.113617. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Masso-Silva J.A., Moshensky A., Lam M.T.Y., et al. Increased peripheral blood neutrophil activation phenotypes and neutrophil extracellular trap formation in critically ill coronavirus disease 2019 (COVID-19) patients: a case series and review of the literature. Clin. Infect. Dis. 2022 Feb 11;74(3):479–489. doi: 10.1093/cid/ciab437. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Tian J., Yuan X., Xiao J., et al. Clinical characteristics and risk factors associated with COVID-19 disease severity in patients with cancer in Wuhan, China: a multicentre, retrospective, cohort study. Lancet Oncol. 2020 Jul;21(7):893–903. doi: 10.1016/S1470-2045(20)30309-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Schulte-Schrepping J., Reusch N., Paclik D., et al. Severe COVID-19 is marked by a dysregulated myeloid cell compartment. Cell. 2020 Sep 17;182(6):1419–1440.e23. doi: 10.1016/j.cell.2020.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Pence B.D. Severe COVID-19 and aging: are monocytes the key? Geroscience. 2020 Aug;42(4):1051–1061. doi: 10.1007/s11357-020-00213-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Tian Y., Carpp L.N., Miller H.E.R., et al. Single-cell immunology of SARS-CoV-2 infection. Nat. Biotechnol. 2022 Jan;40(1):30–41. doi: 10.1038/s41587-021-01131-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Brodin P. Immune determinants of COVID-19 disease presentation and severity. Nat Med. 2021 Jan;27(1):28–33. doi: 10.1038/s41591-020-01202-8. [DOI] [PubMed] [Google Scholar]
  • 40.Bonifacius A., Tischer-Zimmermann S., Dragon A.C., et al. COVID-19 immune signatures reveal stable antiviral T cell function despite declining humoral responses. Immunity. 2021 Feb 9;54(2):340–354. doi: 10.1016/j.immuni.2021.01.008. .e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Zhang J.Y., Wang X.M., Xing X., et al. Single-cell landscape of immunological responses in patients with COVID-19. Nat. Immunol. 2020 Sep;21(9):1107–1118. doi: 10.1038/s41590-020-0762-x. [DOI] [PubMed] [Google Scholar]
  • 42.Vabret N., Britton G.J., Gruber C., et al. Immunology of COVID-19: current state of the science. Immunity. 2020 Jun 16;52(6):910–941. doi: 10.1016/j.immuni.2020.05.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Mathew D., Giles J.R., Baxter A.E., et al. Deep immune profiling of COVID-19 patients reveals distinct immunotypes with therapeutic implications. Science. 2020 Sep 4;369(6508) doi: 10.1126/science.abc8511. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Di Salvatore V., Crispino E., Maleki A., Nicotra G., Russo G., Pappalardo F. Computational identification of differentially-expressed genes as suggested novel COVID-19 biomarkers: a bioinformatics analysis of expression profiles. Comput. Struct. Biotechnol. J. 2023;21:3339–3354. doi: 10.1016/j.csbj.2023.06.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Zhao X., Zhang M., Jia Y., et al. Featured immune characteristics of COVID-19 and systemic lupus erythematosus revealed by multidimensional integrated analyses. Inflamm. Res. 2023;72(9):1877–1894. doi: 10.1007/s00011-023-01791-3. [DOI] [PubMed] [Google Scholar]
  • 46.Chen Y., Feng Y., Yan F., et al. A novel immune-related gene signature to identify the tumor microenvironment and prognose disease among patients with oral squamous cell carcinoma patients using ssGSEA: a bioinformatics and biological validation study. Front. Immunol. 2022 Jul 6;13 doi: 10.3389/fimmu.2022.922195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Gong J., Li R., Chen Y., et al. HCC subtypes based on the activity changes of immunologic and hallmark gene sets in tumor and nontumor tissues. Brief Bioinform. 2021 Sep 2;22(5) doi: 10.1093/bib/bbaa427. [DOI] [PubMed] [Google Scholar]
  • 48.Wu C.C., Wang Y.A., Livingston J.A., et al. Prediction of biomarkers and therapeutic combinations for anti-PD-1 immunotherapy using the global gene network association. Nat. Commun. 2022 Jan 10;13(1):42. doi: 10.1038/s41467-021-27651-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Tabata S., Imai K., Kawano S., et al. Clinical characteristics of COVID-19 in 104 people with SARS-CoV-2 infection on the Diamond Princess cruise ship: a retrospective analysis. Lancet Infect. Dis. 2020 Sep;20(9):1043–1050. doi: 10.1016/S1473-3099(20)30482-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Laxminarayan R., B C.M., G V.T., et al. SARS-CoV-2 infection and mortality during the first epidemic wave in Madurai, south India: a prospective, active surveillance study. Lancet Infect. Dis. 2021 Dec;21(12):1665–1676. doi: 10.1016/S1473-3099(21)00393-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Rydyznski Moderbacher C., Ramirez S.I., Dan J.M., et al. Antigen-specific adaptive immunity to SARS-CoV-2 in acute COVID-19 and associations with age and disease severity. Cell. 2020 Nov 12;183(4):996–1012. doi: 10.1016/j.cell.2020.09.038. .e19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Peterson D.R., Baran A.M., Bhattacharya S., et al. Gene expression risk scores for COVID-19 illness severity. J. Infect. Dis. 2023;227(3):322–331. doi: 10.1093/infdis/jiab568. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Krishnamoorthy S., Li G.H., Cheung C.L. Transcriptome-wide summary data-based Mendelian randomization analysis reveals 38 novel genes associated with severe COVID-19. J. Med. Virol. 2023;95(1) doi: 10.1002/jmv.28162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Pairo-Castineira E., Rawlik K., Bretherick A.D., et al. GWAS and meta-analysis identifies 49 genetic variants underlying critical COVID-19 [published correction appears in Nature. 2023 Jul;619(7971): e61] Nature. 2023;617(7962):764–768. doi: 10.1038/s41586-023-06034-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Tan Y., Lu S., Wang B., et al. Single-cell transcriptome atlas reveals protective characteristics of COVID-19 mRNA vaccine. J. Med. Virol. 2023;95(1) doi: 10.1002/jmv.28161. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Machado T.L., Santos A.C., Azamor T., et al. CLEC5A expression can be triggered by spike glycoprotein and may be a potential target for COVID-19 therapy. J. Med. Virol. 2023;95(1) doi: 10.1002/jmv.28427. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Aghamohamadi N., Shahba F., Zarezadeh Mehrabadi A., et al. Age-dependent immune responses in COVID-19-mediated liver injury: focus on cytokines. Front. Endocrinol. 2023;14 doi: 10.3389/fendo.2023.1139692. Published 2023 Aug 15. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Multimedia component 1
mmc1.docx (18.1MB, docx)
Multimedia component 2
mmc2.xlsx (54.6KB, xlsx)
Multimedia component 3
mmc3.xlsx (12KB, xlsx)
Multimedia component 4
mmc4.xlsx (13KB, xlsx)
Multimedia component 5
mmc5.xlsx (1.4MB, xlsx)
Multimedia component 6
mmc6.xlsx (14KB, xlsx)

Data Availability Statement

The results shown here are based upon data generated by GEO (http://www.ncbi.nlm.nih.gov/geo/) and CIBERSORT (https://cibersort.stanford.edu/) databases. The login numbers of the GEO database are GSE152641, GSE157103, GSE171110 and GSE196822.

Articles from Heliyon are provided here courtesy of Elsevier

RESOURCES