8806 Russian patients demonstrate T cell count as better marker of COVID-19 clinical course severity than SARS-CoV-2 viral load

The article presents a comparative analysis of SARS-CoV-2 viral load (VL), T lymphocyte count and respiratory index PaO2:FiO2 ratio as prospective markers of COVID-19 course severity and prognosis. 8806 patients and asymptomatic carriers were investigated in time interval 15 March–19 December 2020. T cell count demonstrated better applicability as a marker of aggravating COVID-19 clinical course and unfavourable disease prognosis than SARS-CoV-2 VL or PaO2:FiO2 ratio taken alone. Using T cell count in clinical practice may provide an opportunity of early prediction of deteriorating a patient’s state.

Subjective symptomatic score. We introduced a so-called "subjective symptomatic score" whose invention was partly inspired by methodology described in works of Calza et al. 39 and Galloway et al. 9 . However, we used our own modification of this factor. The presence of every symptom has a score of + 1 -+ 5 depending on the pronouncedness, its absence 0. Medical personnel was consulted to apply such a system of evaluation and we received the numerical data in the majority of cases. Where numerical data were missing, we transformed qualitative description of symptomatic course into the score discussed, by ourselves. While the symptomatic picture is definitely subjective as it is based on personal perception of COVID-19 symptoms or observations of medical staff and therefore qualitative, the score may give a semi-quantitative indication of COVID-19 course severity. The list of symptoms taken into account are summarized in Table 1.
As one can conclude, asymptomatic carriers had the total score of 0. The maximal possible value is 5 × 20 = 100. When calculating subjective symptomatic score for symptomatic cohort, we did not count asymptomatic values lest the score should be greatly skewed to low values.
Acute respiratory distress syndrome (ARDS) score. We suggest using ARDS score AS to assess COVID-19 respiratory complications. We may define it as where RR is respiratory rate [min -1 ]. The detailed description of this indicator is provided in our another work 34 . An advantage of using AS over any of PaO 2 /FiO 2 ratio, SpO 2 or RR can be explained by the complex nature of AS. AS encompasses more border states of ARDS than any of these three indicators. Therefore, in some cases using AS may be more reliable as it usually highlights suspicious cases of ARDS that some of the three separate indicators may miss.
The formula for AS contains reversed PaO 2 /FiO 2 ratio as the first item (with multiplier 1000 taken for convenience). It explains why we used reversed PaO 2 /FiO 2 ratio in the current study instead of PaO 2 /FiO 2 ratio that is generally used as an indicator of respiratory state of a patient. Indeed, we searched for direct proportionality between VL and COVID-19 respiratory complications. One can expect that high VLs may potentially indicate severe respiratory failure, as it was shown in numerous research papers, e.g. in the works [40][41][42] . Low PaO 2 /FiO 2 ratio values which stand for respiratory failure do not correlate with high VLs as direct proportionality, but high reversed PaO 2 /FiO 2 ratios do. However, using reversed PaO 2 /FiO 2 ratio was a mere convention dictated by the simplicity of formulas in the current study. It can be easily substituted by PaO 2 /FiO 2 ratio.

Results and discussion
Main clinical parameters. Demographics and main clinical parameters of the patients are summarised in Table 2.
SARS-CoV-2 viral load as prospective marker of COVID-19 course severity. Figure 1 demonstrates relationship between SARS-CoV-2 VL and reversed PaO 2 /FiO 2 ratio (the first item in the formula for ARDS score AS, see "Methods" section). We can observe that there is no clear correlation between reversed PaO 2 /FiO 2 ratio and VL for the whole range of VLs. None the less, the almost linear piecewise dependence of reversed PaO 2 /FiO 2 ratio on SARS-CoV-2 VL (several linear fragments) helps to isolate asymptomatic carriers (the leftmost points in Fig. 1) and severest cases (the rightmost points), whereas the majority of VLs corresponds to approximately the same reversed PaO 2 / FiO 2 ratio = 3.6 ± 1.3 mmHg -1 (horizontal plateau in the middle of Fig. 1). That corresponds to direct PaO 2 /FiO 2 ratio range = [204…434 mmHg].  Fig. 1 we see that only very high VL led to the distortion of plateau PaO 2 /FiO 2 = 281 mmHg (to the right of the red vertical line, i.e. further diminishing PaO 2 /FiO 2 ), whereas the majority of VLs were statistically undistinguishable in terms of PaO 2 /FiO 2 ratio.
Leukocyte and B lymphocyte counts in serum as prospective markers of COVID-19 course severity. In Fig. 2 one can see the dependence of total leukocyte count in serum on logarithm of SARS-CoV-2 VL and in Fig. 3 the dependence of B lymphocyte subpopulation counts on logarithm of SARS-CoV-2 VL. Leukocyte count (more exactly, granulocyte count) may be understood as a common indicator of the front 1000  www.nature.com/scientificreports/ line of innate immune system response to SARS-CoV-2 and B cell counts as parts of both innate and adaptive immune response. However, again we do not observe any direct proportionality between VL and overall white blood cell counts, nor B lymphocyte counts, as the correlations are non-linear or there are no correlation at all (for B1 cells, Fig. 3).
Both leukocyte and plasma cell count correlations with SARS-CoV-2 VL have an asymmetric wave shape with maximums around VLs of 10 5 -10 8 RNA equivalents per reaction. These maximums correspond with the majority of symptomatic patients. In asymptomatic carriers, immune response is not pronounced and almost all severest cases were cases of patients with chronic immune problems. This may indicate that the strongest immune response (innate and adaptive) to SARS-CoV-2 is present in COVID-19 symptomatic or at least paucisymptomatic patients without serious immune disorders.
In Fig. 6, dependence of T lymphocyte population size on reversed PaO 2 /FiO 2 ratio is presented.
Here the dependence is much stronger than the dependence on VL. Statistically significant difference in T killer, NKT cell (Fig. 6A), γδT and CD8αα + cell population size (Fig. 6B) gives an opportunity to use this functional relation as a more unequivocal marker of COVID-19 course severity than T cell count-VL dependency. We did not detect any connection of CD8αα + cell count on SARS-CoV-2 VL (Fig. 5D), but observed strong diminishment of CD8αα + lymphocyte population size in pharynx MALT with the growth of respiratory distress severity (Fig. 6B).

Markers of clinical course outcome, mortality and prognosis.
To find a convincing marker of COVID-19 clinical course outcome, including mortality, may be regarded a separate important task in suggesting markers for COVID-19 clinical course predictions.
COVID-19-related mortality was closely connected with T cell population suppression in severest cases. T cell population size diminishment (especially of T killers, Th1 helpers, and CD8αα + lymphocytes) was more closely connected with COVID-19-related mortality (Pearson correlation coefficient C = 0.7411 at p = 0.1217) than PaO 2 /FiO 2 ratio with COVID-19-related mortality (C = 0.6238 at p = 0.2150). But even more importantly, as we detected, strong diminution of T cell populations often (in 58.22% cases) shows deterioration of a patient's conditions 2-5 days before worsening his/her respiratory status measured by PaO 2 /FiO 2 ratio.
The mortality distribution on VLs (Fig. 4, rightmost dark-grey box) did not demonstrate direct proportionality relationship between VL (or its reversed logarithm 1 log 10 VL ) and lethal outcome. While mortality net number was distributed in an asymmetric peak-like mode, not Gaussian mode, the distribution of mortality percentage did not show any clear dependence on VL (Fig. 7A). The approximation dependence However, the mortality percentage distribution on the product of several factors ("complex factor" CF1) did demonstrate a stronger dependence (Fig. 7B). The best fit can be achieved through approximating by biphasic dose-response function: By comparing results of the statistical analyses, we see that the complex factor CF1 is more promising for using as an input variable for predicting lethal outcome than VL. In the defining formula for CF1, we did not include Th1 helper count, as Th1 count was shown to be very strongly correlated with T killer count (Fig. 5A,B). Therefore, of the two variables T killer and Th1 helper counts, a truly independent variable is merely one.
Limitations of the study. The research has serious limitations that cannot be ignored, despite considerable size of the set studied.
1. The range of patients' demographics is extremely broad. Such breadth is a result of initial randomising approach of the study, which was aimed at collecting data from a most representative population set. However, such an approach has as well drawback as advantages. Patients with quite different comorbidities entered the set, as there was no filtration of patients/carriers. In the study, the influence of comorbidities on COVID-19 clinical course was not studied quantitatively and thoroughly, as it may be a task for a separate investigation. Different comorbidities may lead to quite dissimilar COVID-19 clinical course and blur the VL/ARDS score/T cell count relationships. It was not paid due attention in the current work. 2. The main conclusion of the study is T cells counts' being the best predictive markers for COVID-19 clinical course and mortality. Therefore, it may be implied that effective adaptive immune system response plays a major role in positive prognosis and immune impairments deteriorate the prognosis. However, we did not specifically and assiduously investigate patients with immunity disorders (we merely took into consideration the fact of clinically documented immunity diseases), nor immunity deterioration with age. As the set was very broad, many aged persons and people with impaired immunity were included in it, but their immunity functioning (e.g., median levels of T cell serum concentrations in the presence and absence of an acute infectious disease) before COVID-19 was not taken into account. This oversimplifying might distort the results to considerable degree. 3. After the extensive research, it is still unclear whether SARS-CoV-2 causes lymphopenia or, on the contrary, "cytokine storm" (hyperinflammation) as its most probable effect on immunity of a conditionally healthy adult. There are plenty of works supporting either assumption. The data obtained in our research cannot support a suggestion that SARS-CoV-2 causes any significant CD8 + and CD4 + lymphopenia itself, even less so suppression of other T lymphocyte populations. It is highly possible that SARS-CoV-2 may be an exacerbating factor of immune dysfunctions or diseases already present in people with serious disease-or age-related lymphocyte population decrease. E.g. HIV, HBV, HCV, immunosuppressed status after anticancer chemotherapy, organ/tissue transplantation/grafting etc. may cause substantial lymphopenia that SARS-CoV-2 may exaggerate further. We ground our assumption on the following observation. Of all deaths registered for patients with more than twofold T cell count diminution (either of T killer, Th1 helper, NKT cells, γδT cells or CD8αα + cells), 87.4% were also associated with immune dysregulation/diseases not related to SARS-CoV-2 and clinically documented before the pandemic. Therefore, we suggest that the assumption about SARS-CoV-2's resulting in immune dysregulation of a healthy person, should be re-evaluated on a more clinically checked set of SARS-CoV-2 carriers/patients, ideally in randomised controlled clinical trials.
Primary targeting T killer and NKT cell populations instead of CD4 + lymphocyte population hints that SARS-CoV-2 influence on cell immunity is completely dissimilar with HIV-1. Indeed, it is possible that SARS-CoV-2 can act as an inductor or "amplifier" of immune disorders/dysfunction. If this is true, in most patients SARS-CoV-2 does not cause lymphopenia itself, rather only augments it, possibly like influenza or some other respiratory viruse (e.g. parainfluenza viruses) [47][48][49][50][51][52][53] .
However, the role of SARS-CoV-2 in lymphopenia, whether it is primary or secondary, was not studied. Suppression of lymphocyte population size (lymphopenia) needs a further detailed research. We do not know biochemical mechanisms of such lymphopenia, i.e. whether they are related to membrane protein degradation or another distortion of T cell differentiation or activation. 4. SARS-CoV-2 was found to cause a wave-form correlation of leukocyte and plasma cell serum concentrations with VL in the majority of symptomatic patients and typical VLs of 10 5 -10 8 RNA copies per reaction were determined to correspond with the maximum of the wave. This fact was not possible to interpret in the current study within any theory of immune response. 5. It is clear that SARS-CoV-2 targets adaptive immunity much more than innate immunity, and T and NKT cells more than γδT cells. However, homodimeric CD8αα + lymphocyte population (a part of innate immunity) in pharyngeal MALT (and hence very likely in all parts of the upper respiratory tract MALT) is very seriously affected in severe COVID-19 cases and may be the best marker of a patient's respiratory state deterioration. The finding has not yet been fully interpreted and comprehended in this research, partly because insufficiency of our data and partly because the role of CD8αα + cells in human immune system is still not completely clear. 6. We had a rather arbitrary choice of measurement time of VL, PaO 2 /FiO 2 ratio and leukocyte/lymphocyte counts, as specified in Methods section. No longitudinal studies were made in this work. The time of measurement was approximately the same for all carriers/patients that allowed to reduce the dispersion of results. However, we have to remember that dynamics of the parameters studied is very important, as these parameters change with COVID-19 clinical course and, according to the remark of our Reviewer, their alteration may indicate improvement or progression of COVID-19. 7. For the majority of patients, no immune tests have been ever made before the onset of COVID-19 disease (or detecting the asymptomatic course). We may only assume that their immune indicators (B and T cell counts) were within normal range of a conventionally healthy adult before COVID-19. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.