Early phases of COVID-19 are characterized by a reduction of lymphocyte populations and the presence of atypical monocytes

Background Severe acute respiratory syndrome coronavirus 2 is a recently discovered pathogen responsible of coronavirus disease 2019 (COVID-19). The immunological changes associated with this infection are largely unknown. Methods We evaluated the peripheral blood mononuclear cells profile of 63 patients with COVID-19 at diagnosis and the presence of association with inflammatory biomarkers and 28-days mortality. Results Lymphocytopenia was present in 51 of 63 (80.9%) patients. This reduction was mirrored also on CD8+ lymphocytes (128 cells/uL), natural killer cells (67 cells/uL) and natural killer T cells (31 cells/uL). Monocytes were preserved in total number but displayed a subpopulation composed mainly of cells with a reduced expression of both CD14 and HLA-DR. A direct correlation was found between serum values of IL-6 and the frequency of Th2 lymphocytes (R=0.17; p=0.04) but not with the monocytes count (R=0.01; p=0.60). Patients who died in the 28 days from admission (N=10, 15.9%), when compared to those who did not, displayed lower mean values of CD3+ (p=0.028) and CD4+ cells (p=0.042) and higher mean percentages of CD8+/CD38+/HLA-DR+ lymphocytes (p=0.026). Conclusions The early phases of COVID-19 are characterized by lymphocytopenia, predominance of Th2 lymphocytes and less immunocompetent monocytes, which include atypical mononuclear cells.

were also employed as healthy controls. The study was approved by the Ethical Committee Milano Area 2 (#358_2020) and was conducted in accordance with the Helsinki Declaration. Inflammatory biomarkers assessment and clinical data collection 1 0 8 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020 Procalcitonin (PCT), ferritin and interleukin-6 (IL-6) were measured with electro-  For each patient, the P/F ratio (partial pressure of oxygen [PaO 2 ]/fraction of inspired oxygen 1 1 5 [FiO 2 ]), was calculated at the admission to the hospital. The 28-days mortality was collected 1 1 6 from electronic medical records. Two different methods were used for viral detection. The first one consisted in Seegene Inc rRT-PCR targeting three viral genes (E, RdRP and N). Left-over peripheral blood samples were processed within 24 hours from collection for the cell subsets. For this purpose, we employed the monoclonal antibodies (BD Biosciences, San optimal dilution for a given staining of 100 µL volume of whole blood was determined by 1 3 2 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 6, 2020. (SSC), and the data were analysed using FACSSuite and FlowJo softwares (BD Biosciences). An automatic standard compensation was applied for each acquisition. Internal quality 1 3 9 assurance procedures included BD cytometer setup and tracking beads, according to the  Descriptive statistics were performed for all the variables assessed in the study population. The Spearman test and linear regression were used to examine correlations whereas Kruskall- Wallis test and multiple t test were employed to assess differences among groups. A p value 1 4 5 <0.05 was deemed statistically significant. All the analysis was performed with GraphPad  Sixty-three patients with confirmed SARS-CoV-2 infection were enrolled, on average 5.3 1 5 0 days (SD 2.9) after symptoms onset. The median P/F ratio at admission was 170 mmHg. Hypoxemia was severe (P/F < 100) in 13 patients (20.6%), moderate (P/F 100-200) in 13 1 5 2 (20.6%) and mild (P/F 200-300) in 23 (36.5%), while the remaining 14 patients had a P/F > 1 5 3 300 mmHg. Overall, the 28-days mortality rate was 15.9% (10/63). Demographic and clinical  Table 1. Overall, all the 1 5 5 inflammatory markers considered were above the reference intervals. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10. 1101/2020 White blood cells total count and subpopulations 1 5 7 At diagnosis of SARS-CoV-2 infection, white blood cells (WBC) count, monocytes and 1 5 8 neutrophils median values were within the reference intervals employed in our Centre. Instead, lymphocyte counts were well below the lower reference limit. Overall, 1 6 0 lymphocytopenia was found in 51 of the 63 patients (80.9%). Consequently, the NLR was of CD3+ and CD8+ cytotoxic lymphocytes (CTLs) were below the lower reference limit and 1 6 6 also CD4+ lymphocytes were reduced, but still in the lower part of the reference interval like 1 6 7 CD19+ lymphocytes (Table 2). Likewise, the two groups of cytotoxic lymphocytes belonging 1 6 8 to the innate immunity, NK and NKT cells, appeared reduced in number and below the lower 1 6 9 reference limits ( Figure 1 and Table 2). Regarding the activation status of T lymphocytes,  The T CD4+ lymphocyte subclasses are presented in Figure 1. The relative median 1 7 6 frequencies of Th1, Th2, Th17 and Treg lymphocytes were respectively 13.12% (IQR 10.80-1 7 7 17.60), 27.6% (IQR 21.67-36.70), 7.2% (IQR 3.59-12) and 5.1% (IQR 3.61-6.8), 11% of CD4+ cells for Th1, Th2, Th17 and Treg lymphocytes, respectively. Consequently, CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020. As stated above, NK cells median values were reduced and below the lower reference value. CD56 bright CD16-(immature) and NK CD56 dim CD16+ (mature) frequencies were 3.7% (IQR  The median frequencies of monocyte subpopulations are displayed in figure 1 C. Overall, 1 9 5 relative subpopulations were 3.3% (IQR 1.66-5.7), 3.7% (IQR 0.8-8.7) and 77.2% (IQR 1 9 6 70.28-85.9) for non-classical monocytes (CD14 dim CD16+), intermediate monocytes  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020. Therefore, we modified flow cytometer parameters with reduced photomultiplier tube values 2 0 8 in order to analyse better these peculiar monocytes (high-SC). We performed this analysis in represented a variable fraction of total monocytes (range 2-14%).

1 2
To better understand the differences in the phenotypic signature between normal and high-SC HLA-DR on high-SC monocytes was lower than MFI on monocytes of healthy donors 2 1 7 (p<0.0001). Regarding HLA-DR, we also found a reduction on monocytes with normal  We also assessed the relationship between the different variables evaluated in our cell  Variables associated with death 2 2 8 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020. Finally, we stratified all the results presented above based on 28-days mortality. Patients who 2 2 9 died within 28 days from admission had higher age, LDH values, CD4/CD8 ratio and 2 3 0 percentages of CD38+/HLA-DR+ CTL compared to those alive (Table 3). Instead, they had 2 3 1 lower P/F ratio, CD3+ and CD4+ lymphocyte counts. In our cohort of patients hospitalized for COVID-19, we observed a reduction of circulating 2 3 4 lymphocytes, both in terms of total count and specific subpopulations. Precisely, all cytotoxic 2 3 5 lymphocytes (CTLs, NK cells and NKT cells) were below the lower reference limits. Moreover, the number of activated CD8+ T cells was slightly increased, whereas the T CD4+ were not correlated to monocyte count but instead to T cells with a Th2 phenotype. Finally, 2 4 1 patients who died within 28 days from admission presented lower counts of CD3+ and CD4+ 2 4 2 cells and higher percentages of activated T cells when compared to those alive. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020.  Regarding lymphocytes, Zhou and colleagues observed that CD4+ T lymphocytes were 2 7 4 rapidly activated to become pathogenic Th1 cells, secreting proinflammatory cytokines. The of IL-6. (Zhou et al., 2020) In contrast, in our study the T helper ratio was oriented toward a 2 7 7 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020. Th2 polarization and Il-6 levels were directly proportional to this population. This is an 2 7 8 expected finding, considering that IL-6 shift the Th1/Th2 balance of an immune response 2 7 9 towards Th2, by promoting early IL-4 expression and by rendering CD4 T cells unresponsive those data highlight that a worst outcome is associated with a depleted lymphocytes 2 8 8 compartment, partially activated and thus unable to eliminate/abrogate an inflammatory  It should be noted that samples for this study were collected at diagnosis, on average after 5 2 9 1 days from the appearance of symptoms. Considering that the mean incubation time of 2 9 2 COVID-19 after infection has been estimated to be 5 days, we can assume that on average 2 9 3 our patients were in the 10th day of infection. For this reason, our results could be considered 2 9 4 a faithful description of the immunological changes occurring in the early phases of the 2 9 5 disease and could complement well the data provided by others. At the same time, we 2 9 6 acknowledge that our research has limitations. First, it is a single centre study, conducted on a 2 9 7 small number of patients. Second, it involved only patients diagnosed in hospital, therefore 2 9 8 with a severe disease. Third, we did not investigate the presence and the features of immune  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10.1101/2020 1 5 Overall, based on our results, we can hypothesize that COVID-19 causes a reduction of the In a similar manner, our data might suggest that, in the early phases of COVID-19, the virus 3 1 6 can elicit an inflammatory response leading to a reduction in the number of both cytotoxic  In conclusion, we present here the first description of immunologic features at diagnosis of a 3 2 1 cohort of Italian COVID-19 patients that both confirm already available evidence and add prospective studies are needed to provide a precise picture of the immunologic profile during 3 2 7 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 6, 2020.  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 6, 2020.  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 6, 2020. . https://doi.org/10. 1101/2020