The unidentified participants were randomly enrolled from the electronic databases of the UConn John Dempsey Hospital in Farmington, Connecticut, from 2020 to 2022. Here, we leverage the hospital database to build a clinical study on patients infected with COVID-19 from March 2020 to December 2022, as well as a group of non-COVID-19 controls who were annual physical examination patients with no symptoms of SARS-CoV-2 infection within 2 weeks before the examination. Please note that it is possible that there could be asymptomatic COVID-19 individuals in the group of non-COVID-19 controls. An automated Sysmex XN-3000 analyzer was used to measure the peripheral blood white blood cell (leukocyte) counts.

Patient demographics, including survival, age, sex, COVID-19 vaccination status, corticosteroid use, and comorbid conditions, were recorded during hospital admissions and summarized in Table 1 . Comorbidities include chronic lung diseases, cardiovascular diseases, diabetes, and cancer. Clinical hematology testing was performed one or several times after hospital admissions, and we summarized patients’ first-time results of leukocyte and subtype counts in peripheral blood ( Figures 1 , 2 ). The concentrations of neutrophils, immature granulocytes, lymphocytes, and monocytes were calculated based on the total leukocyte counts (leukocyte count × percentage of subtype/100). According to the Centers for Disease Control and Prevention (CDC) genomic surveillance for SARS-CoV-2 variants (Lambrou et al., 2022), we approximately divided the patients into three groups: 1) ancestral strain infection from March to June 2020; 2) Delta variant infection from July to December 2021; and 3) Omicron variant infection from January to December 2022.

To further understand the physiological and immunopathological differences among the ancestral SARS-CoV-2 strain, the Delta variant, and the Omicron variant, we analyzed the 2020, 2021, and 2022 data separately. The COVID-19 patients were divided into two groups - COVID-19 survivors and decedents. Sex, age, COVID-19 vaccination status, corticosteroid usage, and comorbid conditions were compared statistically between non-COVID-19 controls and COVID-19 patients each year. The two-sample t-test was used to compare continuous variables. The chi-square test was used to compare categorical variables. Multiple linear regression analysis was performed to examine the relationships between leukocytes and their subtypes with COVID-19 disease outcomes during the three years following adjustment for age, sex, COVID-19 vaccination, corticosteroid use, and comorbid conditions.

Multiple logistic regression analysis was conducted to examine the associations of mortality with leukocytes and each subtype after adjustment for sex, age, COVID-19 vaccination status, corticosteroid use, and comorbid conditions. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were calculated by exponentiating the regression coefficients of each cell type. Based on the multiple logistic regression model, we calculated and plotted the predicted mortality probabilities and corresponding 95% CIs. Specifically, for each cell type, we generated a sequence of cell values across the observed range and constructed a new dataset comprising all possible combinations of these cell values, categorical covariates, and mean age. Predicted probabilities were then obtained for each combination using the prediction function in the logistic regression model. For each cell value, we calculated the mean predicted mortality probability and derived the 95% CI using the quantile function. All statistical tests were 2-sided, and a P-value less than 0.05 was considered statistically significant. All statistical analyses were performed using GraphPad Prism version 9.0.0 and R version 4.4.1.

Compared with the non-COVID-19 controls, the average age was significantly higher for the COVID-19 survivors and decedents ( Table 1 ), showing that patients hospitalized for COVID-19 are older than patients hospitalized for other issues. This is consistent with the previous studies showing that older people, especially those with comorbidities, might be more likely to be infected and have poorer prognosis (Attaway et al., 2021).

Compared with the non-COVID-19 controls, the survivors and decedents were more likely to be male. However, the sex difference was non-significant for the Delta variant group due to the small sample sizes ( Table 1 ). The results are consistent with previous studies (Garg et al., 2020; Shrock et al., 2020; Takahashi et al., 2020; Webb Hooper et al., 2020) showing that males might be more vulnerable to COVID-19 than females. This may be because of differences in ACE2 expression (Peng et al., 2021; Viveiros et al., 2022), hormonal influences (Selek et al., 2021; Sciarra et al., 2023), immune response variations (Klein and Flanagan, 2016; Takahashi and Iwasaki, 2021; Zhao et al., 2021), and other biological or behavioral factors (Friske and Spanagel, 2023).

Vaccination status correlates with COVID-19, where COVID-19 patients have a significantly lower vaccination rate than non-COVID-19 controls ( Table 1 ). Corticosteroid is commonly used in several respiratory diseases, including COVID-19. As expected, COVID-19 patients have a much higher percentage of corticosteroid use than non-COVID-19 controls ( Table 1 ). There is also a higher percentage of COVID-19 patients with comorbidities than non-COVID-19 controls ( Table 1 ), suggesting a correlation between COVID-19 and other diseases.

To assess whether leukocyte responses differ between infection of ancestral SARS-CoV-2 and variants, we analyzed the concentrations of peripheral blood leukocytes and subtypes from COVID-19 patients and non-COVID-19 controls from various aspects. We performed multiple linear regression analyses to compensate for the effects of age, sex, COVID-19 vaccination, corticosteroid use, and comorbid conditions. We then compared the average leukocyte and subtype concentrations between COVID-19 survivors, decedents, and non-COVID-19 controls ( Figure 1 ), where the ancestral strain, Delta, and Omicron variant groups were analyzed separately. In addition, we compared leukocyte differences between the three variant groups ( Figure 2 ) in survivors or decedents. We then performed multiple logistic regression for each group separately to determine whether leukocytes or subtypes are associated with COVID-19 mortality ( Table 2 ).

We found that in all three groups, total leukocyte and neutrophil concentrations were significantly higher in COVID-19 survivors and further higher in COVID-19 decedents compared with the non-COVID-19 controls ( Figures 1A, F, K ). The higher leukocyte concentration is mainly contributed by the higher neutrophil concentration ( Figures 1B, G, L ) because neutrophils are the most abundant leukocytes in human blood. Multiple logistic regression analyses showed that total leukocytes and neutrophils were positively associated with COVID-19 mortality ( Table 2 , Supplementary Figure 1A–B, F-G, K-L ). These imply that the elevation of neutrophil levels may be correlated with COVID-19 pathogenesis and prognosis.

When comparing the three groups, we found that the Delta and Omicron variant survivors had significantly higher total leukocyte and neutrophil concentrations than the ancestral strain survivors ( Figures 2A,B ). Moreover, Omicron variant survivors also have significantly higher monocyte concentrations than the ancestral strain survivors ( Figure 2E ). Omicron but not Delta variant decedents showed significantly higher total leukocyte concentrations than the ancestral strain decedents ( Figure 2F ). This is due to the higher monocyte ( Figure 2J ) but not neutrophil concentration ( Figure 2G ).

The increase of immature granulocytes in peripheral blood happens in many inflammatory diseases, suggesting emergency granulopoiesis and cytokine storm (Polidoro et al., 2020; Combadiere et al., 2021; Malengier-Devlies et al., 2021). As expected, COVID-19 also showed a higher concentration of immature granulocytes in peripheral blood ( Figures 1C, H, M ). The COVID-19 decedents showed significantly higher peripheral blood immature granulocytes than survivors during the ancestral strain and Omicron variant infection ( Figures 1C, M ) but not in the Delta variant infection ( Figure 1H ). Consistent with this result, multiple logistic regression analyses showed that immature granulocyte concentrations are positively associated with COVID-19 mortality in the ancestral strain and Omicron variant infection but not the Delta variant infection ( Table 2 ; Supplementary Figure 1 ). When comparing the three groups, we found that the Delta variant survivors had significantly higher immature granulocyte concentrations than the other two groups ( Figure 2C ), suggesting that the Delta variant may induce more emergency granulopoiesis. The decedents showed no difference in immature granulocyte concentrations between groups ( Figure 2H ).

Compared with the non-COVID-19 controls, the COVID-19 patients had fewer lymphocytes in all three groups, especially for patients who died from COVID-19 ( Figures 1D, I, N ). The results are consistent with the previous studies (Huang et al., 2021; Zhang et al., 2021; Belaid et al., 2022). However, the Omicron variant decedents did not show a further lower concentration of lymphocytes compared to survivors ( Figure 1N ). In addition, lymphocyte concentrations were negatively associated with COVID-19 mortality, although the association was non-significant for those infected with the Omicron variant ( Table 2 ; Supplementary Figure 1D, I, N ), suggesting suppression of regulatory immunity may contribute to a worse prognosis in the ancestral strain and Delta variant but not Omicron variant-caused COVID-19. This was further supported by the results that the Omicron variant decedents ( Figure 2I ) had higher lymphocyte concentrations than those in the Delta variant decedents. The suppression of regulatory immunity might be due to the administration of corticosteroids to the patients.

We found that COVID-19 survivors had significantly higher monocyte concentrations than non-COVID-19 controls ( Figures 1E, J, O ). However, the situation was different for the decedents. Compared with the non-COVID-19 controls, the monocyte concentration in the decedents was significantly higher in the Omicron variant infection ( Figure 1O ) but significantly lower in the ancestral strain infection ( Figure 1E ) and the Delta variant infection ( Figure 1J ). Additionally, monocyte concentrations were negatively associated with COVID-19 mortality, although the association was less significant in the Omicron variant infection ( Table 2 ; Supplementary Figure 1 ), suggesting the failed induction of monocyte immunity might be involved in the prognosis of the ancestral SARS-CoV-2 and Delta variant but not Omicron variant caused COVID-19. By comparing the groups, we found the Omicron variant survivors ( Figure 2E ) and decedents ( Figure 2J ) had significantly higher monocyte concentrations than those in the ancestral strain and Delta variant infection, implying a possibility that the Omicron variant may have less capacity to suppress monocyte immunity.