Lymphopenia in severe coronavirus disease-2019 (COVID-19): systematic review and meta-analysis

Objective Clinical and laboratory biomarkers to predict the severity of coronavirus disease 2019 (COVID-19) are essential in this pandemic situation of which resource allocation must be urgently prepared especially in the context of respiratory support readiness. Lymphocyte count has been a marker of interest since the first COVID-19 publication. We conducted a systematic review and meta-analysis in order to investigate the association of lymphocyte count on admission and the severity of COVID-19. We would also like to analyze whether patient characteristics such as age and comorbidities affect the relationship between lymphocyte count and COVID-19. Methods Comprehensive and systematic literature search was performed from PubMed, SCOPUS, EuropePMC, ProQuest, Cochrane Central Databases, and Google Scholar. Research articles in adult patients diagnosed with COVID-19 with information on lymphocyte count and several outcomes of interest, including mortality, acute respiratory distress syndrome (ARDS), intensive care unit (ICU) care, and severe COVID-19, were included in the analysis. Inverse variance method was used to obtain mean differences and its standard deviations. Maentel-Haenszel formula was used to calculate dichotomous variables to obtain odds ratios (ORs) along with its 95% confidence intervals. Random-effect models were used for meta-analysis regardless of heterogeneity. Restricted-maximum likelihood random-effects meta-regression was performed for age, gender, cardiac comorbidity, hypertension, diabetes mellitus, COPD, and smoking. Results There were a total of 3099 patients from 24 studies. Meta-analysis showed that patients with poor outcome have a lower lymphocyte count (mean difference − 361.06 μL [− 439.18, − 282.95], p < 0.001; I2 84%) compared to those with good outcome. Subgroup analysis showed lower lymphocyte count in patients who died (mean difference − 395.35 μL [− 165.64, − 625.07], p < 0.001; I2 87%), experienced ARDS (mean difference − 377.56 μL [− 271.89, − 483.22], p < 0.001; I2 0%), received ICU care (mean difference − 376.53 μL [− 682.84, − 70.22], p = 0.02; I2 89%), and have severe COVID-19 (mean difference − 353.34 μL [− 250.94, − 455.73], p < 0.001; I2 85%). Lymphopenia was associated with severe COVID-19 (OR 3.70 [2.44, 5.63], p < 0.001; I2 40%). Meta-regression showed that the association between lymphocyte count and composite poor outcome was affected by age (p = 0.034). Conclusion This meta-analysis showed that lymphopenia on admission was associated with poor outcome in patients with COVID-19.


Introduction
Coronavirus disease 2019 (COVID -19) has been declared by the World Health Organization (WHO) as a global public health emergency due to its pandemicity [1]. Since its first emergence in Wuhan, China, more than 450,000 cases and 20,000 deaths have been recorded globally due to COVID-19 [2]. While most patients with COVID- 19 have mild influenza-like illness and may be asymptomatic, a minority of patients will develop severe pneumonia, acute respiratory distress syndrome (ARDS), multi-organ failure (MOF), and death [3]. Clinical and laboratory biomarkers [4] to predict the mortality and severity of COVID-19 are essential in this pandemic situation of which resource allocation must be urgently prepared especially in the context of respiratory support readiness.
Since the first descriptive study in China regarding the COVID-19 infection [5], lymphocyte count has been a marker of interest. It has been associated with severe COVID-19 [6,7], and non-survivors of COVID-19 were reported to have a significantly lower lymphocyte count than survivors [7]. Whether lower lymphocyte count and lymphopenia could really be predictor of severity of COVID-19 was our main interest, since this laboratory tools are readily available even in the remote areas. Therefore, in the present study, we conducted a systematic review and meta-analysis in order to investigate the association of lymphocyte count on admission and the severity of COVID-19. We would also like to analyze whether patient characteristics such as age and comorbidities affect the relationship between lymphocyte count and COVID-19.

Eligibility criteria
We included research articles concerning adult patients diagnosed with COVID-19 that has information on lymphocyte count at admission, and clinical grouping or outcome of clinically validated definition of severe COVID-19, death, or ICU care. We exclude review articles, non-research letters, commentaries, case reports, animal studies, original research with samples below 20 or case reports and series, non-English language articles, and studies in pediatric populations (≤ 17 years old).

Search strategy and study selection
We systematically searched PubMed, SCOPUS, Eur-opePMC, ProQuest, Cochrane Central Databases, and Google Scholar with the search terms "COVID-19" OR "SARS-CoV-2" AND "Lymphocyte" (Table S1). After initial search, duplicates were excluded. Two independent authors (IH and RP) screened title and abstracts for potentially relevant articles. The full-text of the potential articles was assessed by applying inclusion and exclusion criteria. The literature search was finalized on March 25, 2020. The study was carried out in accordance with the declaration of Helsinki and with the term of local protocol. This is a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-compliant systematic review and meta-analysis

Data extraction
Data extraction was performed independently by two authors (IH and RP). We used standardized forms that included author, year, study design, age, gender, cardiac comorbidities, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, smoking, lymphocyte count, lymphopenia, mortality, ARDS, ICU care, and severe COVID-19.
The outcome of interest was composite poor outcome that comprised of mortality, ARDS, ICU care, and severe COVID-19. Mortality and ICU care was defined as death and admittance to ICU during inhospital care, respectively. ARDS was defined according to the criteria from the World Health Organization (WHO) interim guidance for severe acute respiratory infection (SARI) in COVID-19, which includes the acuity of symptom onset, chest X-ray and origin of pulmonary infiltrates, and oxygenation impairment [8]. Severe COVID-19 was defined as patients who had any of the following features at the time of, or after, admission: (1) respiratory rate ≥ 30 breaths per min, (2) oxygen saturation ≤ 93% (at rest), (3) ratio of partial pressure of arterial oxygen to fractional concentration of oxygen inspired air (PaO2 to fiO2 ratio) ≤ 300 mmHg, or (4) specific complications, such as septic shock, respiratory failure, and or multiple organ dysfunction [9].

Statistical analysis
The meta-analysis of studies was performed using Review Manager 5.3 (Cochrane Collaboration) and Stata version 16. To pool continuous variables, we used an inverse variance method to obtain mean differences (MDs) and its standard deviations (SDs). Maentel-Haenszel formula was used to calculate dichotomous variables to obtain odds ratios (ORs) along with its 95% confidence intervals (CIs). We used random-effects models for pooled analysis regardless of heterogeneity. All P values were two-tailed, and statistical significance was set at ≤ 0.05. Subgroup analysis was performed for lymphopenia cutoff point at ≤ 1100 cells/μL. Sensitivity analysis using a leave-one-out method was performed to single out the cause of heterogeneity. Regression-based Egger's test was used to assess smallstudy effects for continuous variables and Harbord's test for binary outcome. Restricted maximum likelihood randomeffects meta-regression was performed for age, gender, cardiac comorbidity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease (COPD), and smoking.

Subgroup analysis for age
Since the composite poor outcome was affected by age, we performed subgroup analysis by using 55 years old as cutoff point.

Publication bias
Funnel plot analysis showed asymmetrical shape for lymphocyte count and composite poor outcome (Fig. 4). The funnel plot was symmetrical for lymphopenia and severe COVID-19. Regression-based Egger's test showed statistically significant small-study effects (p = 0.018) for the lymphocyte and composite poor outcome. Trimand-fill method did not impute any study. Regressionbased Harbord's test showed no evidence of small-study effects (p = 0.086) for lymphopenia and severe COVID-19 outcome.

Discussion
This meta-analysis showed that lower lymphocyte count was associated with increased mortality, ARDS, need for ICU care, and severe COVID-19. The association seemed to be stronger in younger patients compared to older patients. Although the definition of lymphopenia differed among studies, a subgroup analysis using ≤ 1100 cells/μL cut-off point has showed a consistent outcome in four studies [10][11][12]19]. We set a cut-off point of ≤ 1100 μL because there were 4 studies using it as a cutoff point. There were only 2 studies for ≤ 1000 μL, and 1 study for < 1200μL and ≤1500 μL, respectively. This subgroup analysis aimed to determine the magnitude of odds ratio at a specific cutoff point (not because of its superiority over the other cutoff points). Based on the meta-regression result, subgroup analysis of age group by using 55 years old as the cutoff point was performed. By analyzing the bubble plot chart, the center of bubble plot is approximately 52 to 55 years old. Hence, we chose 55 as the cutoff point to ensure the number of studies is almost equal in the left side and the right side of the bubble plot. If the number of studies was too small, the pooled effect estimate will be less reliable. Interestingly, we found that the association between lymphopenia and severe COVID-19 was stronger in younger patients compared to older patients. This was a novel finding which, as far as we know, has not been discussed in previous literature. Although changes in the number and composition of lymphocytes are considered as hallmark of immunosenescence [30], it could not fully explain this association. One possible hypothesis is that the aging of the immune system could contribute to a relatively "non-reactive" immune state, thereby causing a relatively stable reduced lymphocyte count, while in younger populations, the highly active lymphocyte kinetics may be influenced by a wide range of insults and comorbidities, thus contributing to a relatively higher mean difference between younger populations. This is further reflected by the sensitivity analysis which showed that upon removal of Wang et al. study, heterogeneity can be reduced to 0% for the ICU care outcome. This heterogeneity was attributed to the mean/median age; there were 3 studies for the ICU care outcome, Cao et al. Pre-existing cardiac disease has been shown to increase mortality in patients with COVID-19 [20]; in this metaanalysis, cardiac comorbidity was not found to affect the association between lymphocyte count difference and composite poor outcome. Angiotensin-converting enzyme (ACE) inhibitor and angiotensin-receptor blocker (ARB) have been hypothetically suggested to exacerbate COVID-19 due to increase in angiotensin II level [31]. These drugs are frequently used in patients with diabetes and hypertension, which was associated with poor outcome [32,33]. Although we did not have data on hypertensive medications in the present study, meta-regression showed that  hypertension and diabetes did not significantly affect the lymphocyte count difference between poor and good outcome.
Our understanding of the pathogenesis of lymphocyte reduction in COVID-19 might possibly be enlightened by studies of other similar beta-CoV infection, including severe acute respiratory syndrome (SARS)-CoV and Middle East respiratory syndrome (MERS)-CoV [34]. Peripheral T lymphocytes, both CD4+ and CD8+, are rapidly reduced in acute SARS-CoV infection hypothetically due to lymphocyte sequestration in specific target organs [35]. Although MERS-CoV and SARS-CoV are structurally similar, they bind to different receptors to facilitate entry. SARS-CoV attaches to angiotensinconverting enzyme 2 (ACE2) to enter the host cells, while MERS-CoV attaches to a different receptor, namely dipeptidyl peptidase 4 (DPP4) [36]. Although the mechanism of significant lymphocyte reduction in severe COVID-19 remains unclear, there are hypothesis other than lymphocyte infiltration and sequestration in the lungs, gastrointestinal tracts, and or lymphoid tissues: (1) lymphocytes express the ACE2 receptor and may be a direct target of SARS-CoV-2 infection [37], and (2) an increase of pro-inflammatory cytokines in COVID-19, especially IL-6, could induce further lymphocyte reduction [34].

Implication for clinical practice
Lymphopenia can be used as a marker for poor prognosis in COVID-19 and in younger patients in particular. Lymphopenia defined as lymphocyte count ≤ 1100 cells/μL is associated with threefold risk of poor outcome.

Limitation
The limitation of this systematic review and metaanalysis is the presence of publication bias. This is apparent in the lymphocyte count and composite poor outcome. Most of the articles included in the study were published at preprint server of which are not yet peerreviewed. Data curation from preprint server is crucial due to the novel and emergent nature of COVID-19; most of the studies are not yet published in journals. Most of the studies were exclusively from China; thus the possibility of the same patients reported more than once is high and may represent inaccurate scientific records. The included studies were also mostly retrospective in design. We encourage further studies to create prognostic model that include lymphopenia.

Conclusion
This meta-analysis showed that lymphopenia on admission was associated with poor outcome in patients with COVID-19.