Association of immunological features with COVID- 19 severity: a systematic review and meta-analysis

Zhicheng Zhang Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology Guo Ai Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology Liping Chen Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology Shunfang Liu Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology Chen Gong Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology Xiaodong Zhu Huang Gang Central Hospital Chunli Zhang Huang Gang Central Hospital Hua Qin Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology Junhui Hu UCLA: University of California Los Angeles Jinjin Huang (  zczhang@tjh.tjmu.edu.cn ) Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology https://orcid.org/0000-0003-1630-0385


Background
Coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection has been spreading all over the world [1]. Till September 09, 2020, the SARS-CoV-2 has infected over 27 million patients and caused over 890,000 deaths [2]. The severity of COVID-19 may be strongly related to immune status of patients, but this is poorly understood. Therefore, it is necessary to explore the association of immunological features with COVID-19 severity, which may help to identify immune markers of disease severity for effective triage of COVID-19 patients.
Two investigators developed the search strategy and one investigator conducted the primary systematic search for all studies meeting the predetermined inclusion criteria. The titles and abstracts of the retrieved articles were screened for duplicates and relevance to the topic. A second investigator checked study eligibility, quality assessment, and data extraction, for validity and consistency. Full-text reports of the identi ed citations were reviewed by both the primary and secondary investigators in order to select the nal studies. Any discrepancy was resolved by consensus, and if necessary, by consultation with the third investigator.
Quality assessment Quality assessments of the studies were carried out based on the Newcastle-Ottawa Scale (NOS). The total NOS score ≥ 7 indicated a good research quality of the included study.

Data synthesis and analysis
Data entry and analysis were carried out with Review Manager 5.3 (The Cochrane Collaboration, Oxford, England). Heterogeneity of effect estimates within each group of studies were assessed by Q test and I 2 statistic, where I 2 > 50% or p < 0.05 indicated heterogeneity and the random-effects model was used.
When I 2 ≤ 50% or p ≥ 0.05, the xed-effects model was used. For continuous data, we calculated mean differences (MD) and 95% con dence intervals (CI) between severe cases and mild cases. To investigate the potential publication bias, we visually examined the funnel plots. For robustness of results, we performed sensitivity analysis by removing one study each time through sensitivity analysis.

Results
Search results and characteristics of included studies  Table 1 presents the characteristics of the 21 included studies, with 758 severe cases and 1275 mild cases of COVID-19 reported. All but one prospective study [9] of the studies included in this meta-analysis were retrospective studies, which were mostly performed in China. All studies were deemed of high quality with 7 or more NOS scores and details can be found in Table 2.  Fig 2g).

Publication bias
We assessed the publication bias of the literature by the funnel plots in all included studies of each indicator, respectively. Funnel plot analysis did not detect obvious publication bias as the shape of all funnel plots did not reveal any evidence of obvious asymmetry (Fig 3).

Discussion
It is necessary to explore the host immune response to SARS-CoV-2, which may help to identify immune markers of disease severity for effective triage of COVID-19 patients [25]. Our study mainly compared the level differences of immune cells, cytokines and chemokines between mild and severe patients with COVID-19.

The variations of immune cells levels are inconsistent in different reports. Most of our included studies found signi cant lower levels of immune cells (CD8 + T, CD4 + T, CD3 + T, B and NK cells) in severe cases
compared with mild cases [3,11,15]. Only two studies reported no signi cant decrease in CD8 + T cell level [4,19], while one study reported higher levels of B cell [14] in severe cases. Synthesizing all the collected evidence, our meta-analysis results found that the levels of immune cells (CD8+ T, CD4+ T, CD3+ T, B and NK cells) were signi cantly lower in severe cases compared with mild cases, but Treg cell level and CD4 + /CD8 + ratio showed no signi cant differences.
The mechanism underlying the association between the reduction of immune cells levels and COVID-19 severity remain to be determined. CD8 + T cells exert their effects mainly through two mechanisms, including cytolytic activities against target cells and secretion of cytokines [20]. CD4 + T cells could activate the CD8 + T cell response to acute respiratory virus infection [25]. SARS-CoV-2 and associated autoimmune antibodies may lead to growth inhibition and apoptosis of hematopoiesis [26], which may decrease the production and maturation of immune cells [6].
With regard to cytokines, the conclusions of different studies are also inconsistent. With the exception of one study on IL-6 [6] and another study on TNF-α, most of our included studies found that of IL-6 and TNF-α levels were signi cantly higher in severe cases compared with mild cases [4,16,20,27]. Some of our included studies found no signi cant differences in the levels of IL-2, IL-4, IL-5, and IFN-γ, while an nearly equivalent number of studies of each indicator found that they were signi cantly higher in severe cases. Synthesizing all the collected evidence, our meta-analysis results found that IL-5, IL-6, IL-10 and TNF-α levels were signi cantly higher in severe cases compared with mild cases. However, the levels of IL-2, IL-4, IFN-γ, Treg cell and CD4 + /CD8 + ratio showed no signi cant differences.
In severely infected individuals, SARS-CoV-2 could induce excessive cytokine response, such as IL-6, IL-10, and TNF-α surge, known as cytokine storm. Cytokine storm could contribute to acute respiratory distress syndrome (ARDS) or multiple-organ dysfunction, leading to physiological deterioration and death [28]. Cytokines such as IL-10, IL-6, and TNF-α are also involved in T cell reduction. IL-6 contributes to host defense via stimulation of acute phase responses [29]. TNF-α is a pro-in ammatory cytokine that can promote T cell apoptosis [30]. Patients requiring ICU admission have signi cantly higher levels of IL-6, IL-10, and TNF-α. Further, the levels of IL-6, IL-10, and TNF-α inversely correlate with CD4 + and CD8 + T cell counts [31]. This fact is strengthened by our meta-analysis results.
SARS-CoV-2 infection is a potent inducer of proin ammatory chemokines that may be involved in the defense against viral infections [24]. Some studies reported higher concentrations of GM-CSF [5], IP-10 [5,23,24] , MCP-1 [23,24] , eotaxin [7] and RANTES [23] between severe cases and mild cases. However, other studies have not showed signi cant differences in the concentrations of GM-CSF [7,24], IP-10 [7], RANTES [24] , MCP-1 [7] , and eotaxin [24] . Synthesizing all the collected evidence, our meta-analysis results found that MCP-1, IP-10 and eotaxin levels were signi cantly higher in severe cases compared with mild cases. However, the levels of GM-CSF and RANTES showed no signi cant differences. Through its binding to the chemokine receptor 3, IP-10 activates and recruits leucocytes, including T cells and monocytes, thereby perpetuating in ammation [32]. MCP-1-mediated migration of monocytes from the blood stream through the vascular endothelium is essential for routine immune surveillance of tissues, as well as in response to in ammation [33]. Abnormally elevated MCP-1, IP-10 and eotaxin levels may help to determine the severity of SARS-CoV-2 infections and may be predictors of clinical symptoms.

Limitations
Several limitations of our study should be considered. First, the number of studies and participants was not large enough for publication bias analysis of most indicators. Second, most of the included studies in this meta-analysis were retrospective. Third, the overall generalizability of the meta-analysis results should be interpreted with caution as most of included studies were conducted in China. It would be better to include as many studies with a broad geographic scope, to gain a more comprehensive understanding of immunological features of COVID-19 patients. Declarations