Density gradient qualitative evaluation
Patients were stratified according to their clinical status and the specimens were grouped in accordance with the time of sampling i.e. within 24 h from admission (early phase COVID-19) or 14 days after discharge. As healthy controls we took samples from blood donors at the time of the donation. As expected, the typical mononuclear population from control blood donors, isolated Lymphoprep™, was recovered at the interface between the density medium and the sample (Fig. 1A). By contrast the vast majority of PBMCs derived from patients with an early phase of COVID-19 floated quite differently. Some samples from this first COVID-19 cohort (within 24 h from hospitalisation) showed a thicker and fluffier band of mononuclear cells, mostly lower than usual, sometimes stickier, often decorated underneath by a dispersed fine crown of red blood cells or with erythrocyte aggregates (Fig. 1B). A number of times the density medium also appeared rather cloudy or less transparent than usual. The buoyant density of the PBMCs was restored in samples collected 14 d post-discharge (recovered COVID- 19 cohort; Fig. 1C), although a slightly thicker band was often observed. Comparison between one sample at day 14 post-discharge and four samples collected at the admission, processed simultaneously and under the same conditions, are shown in figure 1D. For a better appreciation, magnification of a 14 d post-discharge sample (recovered COVID-19 cohort) and a sample taken within 24 h from admission (early phase COVID- 19 cohort) are displayed in figure 1E, while figure 1F is a closer view of two samples taken at the time of admission into hospital. Occasionally a dual band one right at the interface between the sample and the medium and a second within the density medium itself was also observed (Fig. 1, panels G, H and I).
Population dynamics study
To assess the differential count of cell populations (neutrophils, lymphocytes, monocytes) in the band isolated on the density gradient, an automated haematology cell analyser was used. To have an estimate of the variation in cell composition in the collected fraction also cell ratios were determined.
The neutrophils in the samples of early phase of COVID-19 (taken within 24 h from hospitalisation) heavily outnumbered in percentage those of the convalescent patients (14 d after discharge) as well as those from healthy controls (Fig. 2A). In patients from the 24 h cohort, the amount of neutrophils found in the PBMC band was of 35.4±27.2% (mean±SD; t-test versus healthy controls P<0.0001), with a median value of 28.8% (IQR 11.6-56.1), thus indicating that the number of neutrophils varied extensively among the different COVID patients, probably depending from the severity of the disease at the time of hospitalisation. The neutrophil data distribution was less spread out in the cohort of patients who had the samples taken 14 days after being released from hospital. The mean value in the discharged patients group was 7.4±9.2% (t-test versus blood donor controls P=0.0469; median of 3.8%; IQR 2.5-7.9) while a range much narrower displayed by the healthy controls i.e. 5.4±5.2% (with a median of 3.3 %; IQR 2.3-6.8).
Conversely, the percentage of lymphocytes showed an opposite trend across the different cohorts as compared to what it was observed for the neutrophil populations. Indeed, at admission, the average percentage of the lymphocytes was completely reverted compared with the post-discharge samples and those from blood donors, displaying mean values of 47.3±2.4% (t-test vs healthy controls P<0.0001; median 46.9%; IQR 28.2-67.4) the former and, 73.0±10.8% (t-test vs blood donor controls P=0.8120; median 74.7%;IQR 70.1-80.1) and 72.6±8.3% (median 74.3%; IQR 68.0-78.2) the latter two, respectively (Fig. 2B). Thus, the percentage of lymphocytes, collected post-gradient separation, returned to normal levels at 14 d post-discharge as suggested by the comparison with the values found for the blood donors.
Although still statistically meaningful, the differences in the percentage of monocytes among the three groups were less remarkable (Fig. 2C). The pattern of monocyte distribution, expressed as mean percentage was 17.3±10.9% (Welch’s t-test vs healthy controls P=0.0034; median 15.0%; IQR 8.4-23.0) for the early phase COVID cohort (i.e. samples within 24 h from hospital admission) and 19.6±6.9% (t-test vs blood donor controls P=0.0447; median 18.9%; IQR 15.2-23.2) for the COVID recovered group, whereas healthy controls showed a percentage of 22.0±7.5% (median 20.5%; IQR 16.7-24.6).
To determine the efficiency of the enrichment in the PBMC fraction from the samples of the three different cohorts, we measured the total cell counts that somehow provide the yield of the separation method. The total number of cell recovered were 22.6*106±21.5 cells (t-test vs blood donor controls P=0.2935; median 15.4; IQR 8.8- 29.9) recovered at 24 h, 24.9*106±15.5 cells (t-test vs healthy controls P=0.7595: median 22.4; IQR 15.8-32.8) collected at 14 d post-discharge and 25.5*106±10.9 cells (median 24.2; IQR 18.4-32.2) retrieved from the healthy controls, respectively (Fig. 2D).
The enrichment in terms of total number of cells recovered was not too dissimilar in the three groups and an optimal yield of the mononuclear cells should not account for more than 5% of granulocytes. To assess the effectiveness of the PBMC recovery after the isolation and to evaluate the distribution pattern of the prevailing cell types post-separation through density gradient, we determined the neutrophil-to-lymphocyte ratio (NLR-PBMC, Fig. 2E) and lymphocyte-to-monocyte ratio (LMR-PBMC, Fig. 2F) in the samples of the three cohorts. Although the three different patient cohorts were all statistically different (Fig. 2E), the cells collected from the 24 h group were highly unbalanced towards neutrophils. The NLR-PBMC of 2.1±4.2 (SD; early phase COVID- 19 versus blood donors t-test P<0.0001) demonstrated that, in the early phase of the disease, the PBMC fraction was heavily contaminated by neutrophils, while the ratios of the other two groups were much more close to each other. The post-separation cells collected from the 14 d post-discharge cohort displayed a mean of 0.13±0.21 (14 d discharged versus blood donors t-test P=0.0165) whereas the cell ratio from the blood donor group had values ranging from 0.08±0.08. The LMR-PBMC (Fig. 2F) of the control cohort showed a mean of 3.7±1.3, while the mean ratio of cells collected from COVID patients at 24 h were 3.5±2.7 (Welch’s t-test vs blood donors P=0.6203) and those from patients 14 days after dismissal were 4.3±2.2 (t-test vs healthy controls P=0.0132). Thus indicating that the distribution pattern of mononuclear cells in the density band was similar and the collected fractions contained similar amount of lymphocytes and monocytes. However it is worth to mention that in the present context the above NLR-PBMC and LMR-PBMC are calculated on recovered cells and provide only a parameter of the cell distribution within the enriched fractions following the separation medium centrifugation.
Flow cytometry analysis
Whole blood leftovers were used to analyze leukocyte population of COVID-19 patients (N=13) and healthy blood donors (N=10) by flow cytometry. Cell populations were assessed after red blood cell lysis. Representative forward scatter (FSC) versus side scatter (SSC) dot plots from the three cohorts are depicted in Figure 3A, B and C. In particular, FSC parameter is proportional to the cells size, while SSC reflects the internal cell complexity (i.e. nuclear morphology and granularity). Leukocyte morphological properties from patient samples taken 24 h from hospitalisation (Fig. 3B) clearly exhibited a shift down in the granulocyte population distribution, probably indicating a lack/loss of granularity, when compared to those from both healthy controls (Fig. 3A) and patient at discharge (Fig. 3C). In addition, a diminished number of lymphocytes and an alteration of the monocyte population was observed in the early phase of infection. The post-dismissal specimen (Fig. 3C) showed that the cell distribution pattern returned almost back to normal, though with still a slight increased number of events in the granulocyte region.
The box-and-whisker chart (Fig. 3D) of the median channel intensity (MCI) in the forward scatter (FSC) indicated that there are some significant differences in the cell sizes, indeed the MCI was lower in the 24 h cohort (159221 MCI, Mann–Whitney U test vs blood donors P=0.0070) and tended to raise again in the 14 d discharged patient group (165465 MCI, Mann–Whitney U test vs healthy blood donor controls P=0.0236) to levels close to normality (healthy controls median=168870 MCI), although not completely back to normal. However, the most striking difference was observed instead in the granularity (Fig. 3E), where the internal macromolecular complexity of the cells had a massive impact despite the fact that the side scatter (SSC) is measured at a ninety degree angle on the laser light and its signal is weaker than the FSC. The cells from early phase of COVID-19 patients (within 24 h) showed a marked decrease in the SSC compared with the blood donors (U test vs blood healthy controls P=0.0001). The internal complexity was only partially restored 14 d after hospital-dismissal (Mann–Whitney test vs blood donors P=0.0378).
Morphological inspection of COVID-19 samples
Blood smears and PBMCs from density gradient spotted onto glass slides were fixed and stained with May–Grünwald Giemsa for a simple cytology evaluation. In figure 4, (panels A-D) we showed the blood smears from four different COVID-19 patients within 24 h from hospitalisation. Those samples confirmed the presence of excessive number of mix-shape neutrophils in peripheral blood. Some of these neutrophils had either a ribbon-shaped or a horseshoe-shaped nucleus suggesting that these are immature forms. Neutrophils are still present in blood smears (Fig. 4E, F and G), from specimen taken after 14 days post-discharge, though mostly in the mature form. The panels H and I in figure 4 are examples of the mononuclear cell enrichment prior to the last centrifugation to remove cell debris and blood residues
Blood cell counts
Almost all the COVID patients and the blood donors had a complete blood count (CBC) test done at the same time or on the same day at which the samples for biobanking were taken. Therefore we have retrieved and retrospectively analysed these data to compare them with the results obtained from the gradient centrifugation.
Early phase COVID patients showed elevated levels of neutrophils i.e. 73.3±14.4% (mean±SD; Welch’s t-test vs blood donors P<0.0001), with a median value of 76.0% (IQR 62.9-84.7) whereas the frequency of neutrophils 14 d post-discharge was 56.4±9.8% (t-test vs healthy controls P=0.4141; median 57.6%; IQR 50.4-62.5) and in the whole blood of healthy donors was 57.6±8.5% (median value of 57.7%; IQR 51.2- 63.8; Fig. 5A).
The lymphocytes detected at 24 h showed a downward trend (Fig.5B), confirming the inversion previously seen in the enriched populations (Fig. 2B). These leukocytes displayed a mean of 18.2±12.2% (t-test vs blood donors P<0.0001; median 16.1%; IQR 8.1-23.1) early in the SARS-Cov-2 infected individuals which rose to 30.7±7.6% (t-test vs healthy controls P= 0.8089; median 30.0%; IQR 25.6-36.7) in the discharged patients and was comparable to the values, 31.0±8.6% (median 31.5%; IQR 24.2-36.2), found in the healthy blood volunteers.
The whole blood monocytes showed fewer fluctuations (Fig. 5C) than the enriched fractions in figure 3C. The populations were more compact in all three cohorts and displayed mean percentage values of 7.3±3.3 (Welch’s t-test vs healthy controls P= 0.0055; median 6.8%; IQR 4.9-9.5) at 24 h from admission, with a slight increase in the discharged patients to 8.7±1.9% (t-test vs blood donors P=0.5707; median 8.7%; IQR 7.3-10.2) and 8.5±1.8% (median 8.3%; IQR 7.0-9.6) in the blood donors.
A white blood cell (WBC) count measures the number of leukocytes in the blood and showed minor differences (Fig. 5D). As expected the percentage of circulating white blood cells was slightly higher in the early phase of the infection 7.1±3.2% (t-test vs blood donors P=0.0077; median 6.7%; IQR 4.8-8.8) and in the recovered patients (14 d post-dismissal) 7.0±2.7 (t-test vs healthy controls P= 0.0010; median 6.4%; IQR 5.5-8.0) and dropped to 6.0±1.5% (median 5.7%; IQR 5.0-6.8) in the healthy blood donor controls.
Since the neutrophil-to-lymphocyte ratio (NLR) and lymphocyte-to-monocyte ratio (LMR) are predictive of inflammatory dysfunctions and have prognostic significance of disease severity and outcome for several infections, malignancies and beyond, we report here mean±SD and median with IQR of these biological markers determined in our cohorts (Fig. 5E and F). COVID patients within 24 h from hospitalisation (panel 5E) showed NLR levels of 6.9±5.6 (Welch’s t-test vs healthy controls P<0.0001; median 4.9; IQR 2.9-10.7), which decreased to 2.2±1.2 (t-test vs blood donors P=0.8274; median 1.9; IQR 1.4-2.5) in those measured 14 d post-recovery. The latter cohort displayed values similar to the healthy blood donor controls, whose levels were 2.1±0.8 (median 1.8; IQR 1.4-2.7). Conversely LMR levels were higher in the blood donors i.e. 3.7±1.1 (median 3.6; IQR 2.8-4.8) and in recovered patients at 14 days post-discharged 3.8±1.9 (t-test vs healthy controls P= 0.9123; median 3.5; IQR 2.9- 4.2), while in the early COVID were 2.8±2.8 (Welch’s t-test vs blood donors P=0.0065; median 2.4; IQR 1.4-3.3). Thus both biological marker levels, i.e. NLR and LMR, differed between the groups and were altered in the early phase of the SARS-Cov-2 infection.