Deep immune profiling reveals early-stage and highly coordinated immune responses in mild COVID-19 patients

While immunopathology has been widely studied in severe COVID-19 patients, immunoprotective factors in non-hospitalized patients have remained largely elusive. We systematically analyzed 484 peripheral immune cell signatures, various serological parameters and TCR repertoire in a longitudinal cohort of 63 mild and 15 hospitalized patients versus 14 asymptomatic and 26 control individuals. Within three days following PCR diagnosis, we observed coordinated responses of CD4 and CD8 T cells, various antigen presenting cells and antibody-secreting cells in mild, but not hospitalized COVID-19 patients. This early-stage SARS-CoV-2-specific response was predominantly characterized by substantially expanded clonotypes of CD4 and less of CD8 T cells. The early-stage responses of T cells and dendritic cells were highly predictive for later seroconversion and protective antibody levels after three weeks in mild non-hospitalized, but not in hospitalized patients. Our systemic analysis provides the first full picture and early-stage trajectory of highly coordinated immune responses in mild COVID-19 patients.

We established the longitudinal Predi-COVID cohort 16 in Luxembourg during the first wave of the 118 current pandemic with the aim to gain a deep and systematic understanding of the early antiviral 119 immune response across the full spectrum of COVID-19 disease phenotypes (Fig. 1a). All patients 120 received a diagnosis of COVID-19 confirmed by the SARS-CoV-2 PCR test through the national 121 healthcare system and were then included into Predi-COVID with a delay of maximum 3 days after 122 clinical PCR diagnosis. According to available clinical metadata, patients were stratified into 123 asymptomatic (n=14), mild to moderate (n=63; referred to as "mild" patient group in Fig. 1a and   124 thereafter), and hospitalized (n=15) subgroups for further analyses. All patients (n=92) were 125 sampled on the day of inclusion (defined as "day 1") and three weeks after inclusion ("day 21"). 126 The group of mild COVID-19 patients also contained 11 patients with self-reported shortness-of-127 . CC-BY 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(which was not certified by peer review)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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint breath symptoms that could not be confirmed by a physician and therefore could not be classified 128 as moderate patients following the NIH guideline. We also included control individuals (n=26) from 129 patients' households, who were sampled on day 1 and day 14. While the age range was not 130 different for asymptomatic and mild patients as compared to household controls (Supplementary   131   Table 1, Fig. 1b), hospitalized patients were older than mild patients (median, ~57 vs. ~38 years of 132 age) and controls (Fig. 1b). The BMI was not different among any of the analyzed groups 133 (Supplementary Table 1). In general, more males were included in each of the patient subgroups 134 (between 57% and 69%), while only around 30% of household controls were male participants 135 (Supplementary Table 1). No comorbidity information was available for hospitalized patients and 136 household controls. For the other patient groups, the prevalence of comorbidities (asthma, chronic 137 hematologic disease, obesity and uncomplicated diabetes) was higher among mild than among 138 asymptomatic patients (5% to ~8% among mild vs. none among asymptomatic patients) 139 (Supplementary Table 1).

141
A whole blood count analysis was used to further characterize COVID-19 patients on day 1. We 142 found no significant difference between asymptomatic and mild patients in any of the tested 17 143 general blood count parameters (Supplementary Table 1, Supplementary Fig. 1). However, 144 hospitalized patients showed a remarkable difference compared to mild or asymptomatic patients 145 as demonstrated in the principal component analysis (PCA) plot based on the analysis of whole-146 blood-count parameters (Supplementary Fig. 1a). In line with the reports by others 17 , the 147 frequency of lymphocytes was substantially decreased in hospitalized patients compared to the 148 two other patient groups (Supplementary Fig. 1b), while CRP was significantly elevated in 149 hospitalized patients only (Supplementary Fig. 1c). Furthermore, hospitalized patients showed an 150 increased number of white blood cells (WBC) (Supplementary Fig. 1d). This shift was mainly 151 reflected by an increase in the number and frequency of both monocytes and granulocytes 152 (Supplementary Fig.1e-h). Both the red blood cell count and the hematocrit were modestly but 153 significantly decreased, accompanied by a slightly higher number of platelets in hospitalized 154 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint patients ( Supplementary Fig. 1i-j). Although significantly enhanced, the number of platelets was 155 still within the normal range in most of the hospitalized patients (Supplementary Fig.1j). 156

157
As expected, on day 1 we did not observe a significant increase in IgG antibody levels to SARS 158 CoV-2 S, CoV-2 RBD, CoV-2 NTD and CoV-2 N in the groups of asymptomatic and mild patients 159 compared to controls (Fig. 1c-f). In contrast, while only 48% and 43% of the mild patients showed 160 slightly increased IgG levels, 93% and 75% of the hospitalized patients already displayed 161 significantly enhanced IgG titers to CoV-2 S and N antigens respectively. Three weeks later, we 162 observed a significant increase in IgG levels to all four antigens also in the mild patient group 163 compared to household controls together with a further enhancement of IgG antibody levels in 164 hospitalized patients. The positivity rate for IgG antibodies to CoV-2 S, CoV-2 RBD, CoV-2 NTD 165 and CoV-2 N reached up to 89% among mild and 100% among hospitalized patients at day 21. 166 Asymptomatic patients had a lower positivity rate for IgG against CoV-2 S/S-RBD (67%) and CoV-167 2 N (75%) than mild patients at day 21 ( Fig. 1c-e). IgG antibodies against CoV-2 NTD were in 168 general much lower in asymptomatic and mild patients, only reaching very high levels of 100% 169 positivity in the hospitalized group at day 21 (Fig. 1f). Next, we tested the functional capacity of the 170 induced antibodies against SARS-CoV-2 in a surrogate virus neutralization assay. The assay 171 analyzes the capacity of antibodies to inhibit the binding of labelled recombinant ACE2, the human 172 receptor for SARS-CoV-2, to CoV-2 S or CoV-2 RBD in a multiplex high-throughput format. In line 173 with the serology findings on day 1 (Fig. 1c, d), hospitalized patients already showed blocking 174 antibodies that interfered with ACE2 binding to CoV-2 S or CoV-2 RBD at this early stage (Fig. 1g,   175 h). On day 21, also mild patients had developed a significant inhibitory serologic capacity to block 176 ACE2 binding to CoV-2 S or CoV-2 RBD relative to household controls. Such an increase in ACE2 177 blocking was not seen in the asymptomatic group (Fig. 1g, h), which was also lower in IgG titers 178 against CoV-2 S or CoV-2 RBD (Fig. 1c, d). In line with this notion and as reported by others 18 , 179 IgG antibody titers against both CoV-2 S and CoV-2 RBD, were highly correlated (spearman 180 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint r=0.89 and 0.82 for CoV-2 S and RBD, respectively) with the inhibitory capacity of sera across all 181 patient categories of COVID-19 severity (Fig. 1i, j).  Supplementary Fig. 3a). 197 Interestingly, at day 21, ICOS + CD8 T cells were the only significantly changed immune subset with 198 a decrease in the frequency among CD8 T cells from asymptomatic patients ( Supplementary Fig.   199 3b, c). Although primarily known for its critical role in CD4 T cell and lymph node germinal center 200 formation, studies in ICOS-deficient patients have also indicated a role for ICOS in CD8 effector 201 functions during primary antiviral immunity 19,20 .

203
We next used PCA to show that deep immune profiling was only able to partition hospitalized 204 patients at day 1 from all other groups investigated, but not mild COVID-19 patients from any 205 household control (day 1 and day 14) (Fig. 2a). Then we asked whether specific immune subsets 206 were differentially present in mild COVID-19 patients compared to age-matched household 207 controls at day 1 (Fig. 2b). We observed differences in the frequency of several CD8 T cells 208 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint subsets, such as Ki67 + , CD38 + , and HLADR + CD38 + , representing proliferating, activated and 209 antigen-specific responsive CD8 T cells respectively that were significantly enhanced in mild 210 COVID-19 patients (Fig. 2c-e). The profile included an increase of both Tbet-dependent 211 (Tbet + Ki67 + ) and -independent (Eomes + Ki67 + ) responsive CD8 T cells (Fig. 2f-h). Also the fraction 212 of proliferating CD4 T cells, especially, Th1-responsive (Tbet + Ki67 + ) CD4 T cells was already 213 enhanced early on in mild patients on day 1 (Fig. 2i, j, Supplementary Fig. 4a). In parallel, the 214 frequency of antigen presenting cells (APCs) and antibody secreting cells, such as mature 215 dendritic cells (HLADR + CD38 high DCs) and short-lived plasmablasts (CD27 + CD38 high ), was 216 increased in mild patients (Fig. 2k, l, m, Supplementary Fig. 4b, 4c). Notably, using the unique 217 power of our longitudinal cohort design and the simultaneous comprehensive analysis of both 218 cellular immune subsets and serological responses, allowed the successful prediction of anti-219 SARS-CoV-2 antibody responses at day 21. The frequency of activated CD38 + CD8 T cells and of 220 mature DCs measured at day 1 was highly predictive for the degree of the serological titers of anti-221 SARS-CoV-2 N IgG at day 21 among asymptomatic and mild patients (Fig. 2n, o). On the 222 contrary, it is noteworthy that neither the frequency of CD38 + cells among CD8 T cells nor of 223 mature DCs at day 1 was significantly correlated to the even higher titers of SARS-CoV-2 N IgG 224 (Fig. 2l, m) in the group of hospitalized COVID-19 patients at day 21 (Fig. 2p, q). This finding 225 indicates that the progression and deterioration of COVID-19 is averted only in the presence of a 226 highly coordinated interplay of early cellular innate and adaptive immune responses, which are 227 strongly correlated to the subsequent production of protective antibody titers against SARS-CoV-2.

229
Now switching to day 21 after inclusion, we followed the evolution of the early cellular immune 230 response. On day 21, mild COVID-19 patients were characterized by an enhanced cytotoxic CD8 231 T cell (GZMB + ) response, especially of terminally differentiated responsive CD8 T cells (CD45RO -232 Ki67 + ) (Supplementary Fig. 5a-c). CD4 T cells also showed similar changes after three weeks.

233
The frequencies of CD4 T cells expressing CD57 and GZMB as well as of CD45RO and CD57 234 double-positive CD4 T cells were also significantly enhanced in mild patients compared to 235 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint household controls (Supplementary Fig. 5d-f). Notably, the frequency of GZMB + CD4 cytotoxic T 236 cells showed a trend to be elevated (p=0.053, Kruskal-Wallis test including multiple-group 237 correction) already on day 1 (Supplementary Fig. 5e). The CD57 expressing CD4 T cells 238 detected on day 21 appeared to be mainly cytotoxic effector cells since the percentage of 239 GZMB + CD57 + cells was also significantly enhanced among CD4 T cells (Supplementary Fig. 5g). 240 Although CD57 is regarded and known as a T-cell senescence marker, CD57 + T cells, similar to  asymptomatic, but not in hospitalized patients ( Fig. 2n-q). This finding prompted us to further 249 analyze this aspect. Thus, we asked whether any additional early immune signatures observed in 250 mild patients were significantly different from hospitalized patients and determined the immune 251 signatures that were significantly upregulated or downregulated in mild versus hospitalized COVID-252 19 patients on day 1 (Fig. 3a). As shown in the volcano plot analyses, major differences were 253 present primarily among innate immune cells, such as monocytes, dendritic cells (DC) and natural 254 killer (NK) cells, and to a lesser extent also among B and T cells. Compared with hospitalized 255 patients, mild COVID-19 patients showed a much higher frequency (~40% in mild patients vs.

256
~10% in hospitalized patients) of non-classical monocytes (ncMono, HLADR + CD38 -) 22 . The non-257 classical monocytes were not only higher in frequency among the mild patients, but expressed also 258 higher levels of critical functional markers, such as CD86/CD80 double-positivity (Fig 3b-d), PD-L1 259 and CD13 (Supplementary Fig. 4d, Supplementary Fig.6a, b). Similar to monocytes, the 260 frequency of antigen-presenting cells (APC) such as plasmacytoid DC (pDC) and myeloid DC 261 (mDC) was significantly higher in mild patients versus hospitalized ones (Fig. 3a, d-f). It is 262 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint noteworthy that the frequency of ncMono and mDC (Fig.3b, f) was also slightly lower in mild 263 patients versus household controls, indicating a disease severity-related effect and further 264 supporting the involvement of both cell types in early protective immune responses of COVID-19. 265 Although mature DC were higher in both mild and hospitalized patients (Fig. 2l), the frequency of 266 CD86 -CD80 + functional cells (Fig. 3g, h) and of CD13 + cells (Supplementary Fig. 6c) among total 267 DC was decreased in hospitalized patients only, thus indicating a reduction in phagocytic and hospitalized, but not mild patients (Supplementary Fig. 6d, e). Furthermore, the frequency of NK 274 cells was also significantly decreased only in hospitalized, but not in mild patients relative to 275 household controls (Fig 3i, j). In line with the overall compromised innate immune cell profile,   CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint We next sought to understand whether the impaired innate immune cell response in hospitalized 290 patients was paralleled by early deviated CD8 T cell profiles. Although the intensity (MFI) of ICOS 291 on CD8 T cells was decreased (Supplementary Fig. 6e), the frequency of ICOS + CD8 T cells was 292 unchanged (Supplementary Fig. 3c) and the frequency of CD40L + and PD-1 + GZMB + cells among 293 CD8 T cells was even significantly enhanced in hospitalized patients on day 1 (Supplementary 294 Fig. 6j, k). Furthermore, since the frequency of CD8 T cells expressing other key functional 295 markers, such as Ki67 and CD38, as well as the frequency of CD8 T cells co-expressing HLA-DR 296 and CD38, Tbet and Ki67, as well as Eomes and Ki67 was not decreased in hospitalized patients 297 ( Fig. 2c-g), the functional antiviral capacity of CD8 T cells was most likely equally robust in 298 hospitalized and mild COVID-19 patients. Overall, these data indicate that the major deficiencies 299 observed in hospitalized COVID-19 patients on day 1 were impaired innate immune cells and APC 300 functions rather than adaptive T-and B-cell functions.

302
The differences in cellular immune signatures described above between mild and hospitalized 303 patients were only significant on day 1, but not after three weeks on day 21. This indicates that  To gain further insight into the coordinated early immune response of COVID-19, we analyzed 24 311 different cytokines, chemokines and growth factors on both day 1 and day 21 in sera of all patient 312 and control groups. Interestingly, at day 1, we observed increased levels of interferon gamma-313 inducible protein 10 (IP10/CXCL10) in hospitalized patients and in the mild patient group (Fig. 4a,   314 b). Unexpectedly, a similar regulation was found for the type I-interferon IFNb, which was 315 previously reported to be undetectable in severe COVID-19 patients at around 10 days after 316 symptom onset 24 . Both hospitalized and mild patients showed a significant increase of IFNb 317 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint compared to controls at day 1 (Fig. 4c). While the levels of IP10 and IFNb showed only a 318 temporary and reversible increase among mild patients, declining to normal levels at day 21, this 319 was not the case in hospitalized patients, where both IP10 and IFNb levels remained elevated after 320 three weeks (Fig. 4c). These results point to a crucial and dynamic role of IP10 and IFNb, which is 321 tightly regulated during the early stage of protective immune responses in COVID-19 patients. This 322 notion is also supported by the fact that levels of IP10 and IFNb were significantly correlated with 323 the frequency of mature DCs among all the analyzed patients at day 1 of the Predi-COVID study 324 (Fig. 4d, e). On day 21, none of the 24 circulating immune analytes showed a significant change in 325 the mild patient group versus household controls at day 14 ( Supplementary Fig. 7a). We also 326 could not observe any significant change in the tested cytokine/chemokine levels among 327 asymptomatic patients, neither on day 1 nor on day 21 (Supplementary Fig. 7b, c). The increase 328 of IP10 in mild and hospitalized COVID-19 patients appears to be independent of IFNg, which was 329 only elevated in the hospitalized patient group (Fig. 4c, f, g) and remained significantly higher in 330 hospitalized patients, but still within the normal range for most of the patients (Fig. 4g). Overall, the 331 IP10 and IFNb signature in mild COVID-19 patients was characterized by early dynamic changes 332 with a strong increase at day 1 and a contraction to normal levels at day 21, while the levels in 333 hospitalized patients remained high.

335
With regard to other circulating immune factors, we found a significant increase in plasma levels of 336 eosinophil chemotactic protein (eotaxin-1/CCL11), vascular endothelial growth factor A (VEGFA), 337 IL-6 and IL-10 only in hospitalized patients at day 1 ( Fig. 4h-k). The enhanced levels of IL6 and 338 the regulatory cytokine IL10 in hospitalized versus mild patients on day 1 were still mostly seen 339 within the normal range (Fig. 4j, k). The increased level of IL10 was in line with an increased 340 percentage of FOXP3 + Tregs in both mild and hospitalized patients at day 1 (Fig. 4l). In 341 conclusion, on day 21 after inclusion into the study, most of the analyzed cytokines/chemokines 342 that were elevated early in mild patients had returned to normal levels and waned in hospitalized 343 patients. Only IFNg, IFNb and IP10/CXCL10 remained higher in hospitalized patients than mild 344 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint patients and controls on day 21 (Supplementary Fig. 7d). The only exception was the Th2 345 cytokine IL5, which was not different on day 1, but modestly increased in mild patients at day 21, 346 while hospitalized patients exhibited very low IL5 levels at this point (Supplementary Fig. 7e).  (Fig. 5c). Notably, the SARS-CoV-2-specific T-cell clonotypes 361 were substantially expanded (~6 times higher than in household controls) among mild patients 362 already at day 1, as reflected by clonal breadth and depth 28 , and maintained at day 21 ( Fig. 5d, 363 e). These results indicate a key functional role of early-responsive SARS-CoV-2-specific T cells in 364 mild COVID-19 patients. Completely unexpected, mainly CD4 SARS-CoV-2-specific T cells were 365 expanded among mild patients at day 1, with an average frequency of CD4 SARS-CoV-2-specific 366 TCR clonotypes that was six times higher than that of CD8 T cells among mild patients ( Fig. 5f-h). 367 This finding was in line with a trend for increased frequency of GZMB + cells among CD4 Tconv 368 cells, but not among CD8 T cells in mild patients versus controls on day 1 (Supplementary Fig.   369 5a, e). At day 21, CD4 SARS-CoV-2 specific TCR clonotypes continued to dominate over CD8 370 clonotypes to a similar extent in mild patients (Fig. 5g, h). The expansion of CD4 or CD8 SARS-

371
CoV-2-specific T cells was highly correlated with the frequency of responsive ICOS + Ki67 + cells 372 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint among total CD4 T cells or total CD8 T cells in both asymptomatic and mild patient groups at day 1 373 (Fig. 5i, j). These data highlight a crucial role of early-expanding SARS-CoV-2-specific T cells,  CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint patients. In contrast to that published work, our current study now provides not only information on 401 the number and frequency of a wide spectrum of immune subsets in peripheral blood, but also on 402 the functionality of individual immune cell types. We found both early-stage Tbet-dependent and - CC-BY 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.  CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint relative to controls, although without significant difference between the two groups, was reported in 455 a small scale pilot study 46 and anti-VEGF medication has been investigated to treat COVID-19 in a  We first would like to acknowledge the active involvement of all the anonymous participants in the 505 Predi-COVID cohort. We are also thankful for the excellent support of the recruitment team of the 506 Predi-COVID cohort from CIEC of LIH. The Predi-COVID study is supported by the Luxembourg 507 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint National Research Fund (FNR) (Predi-COVID, 14716273) and the André Losch Fondation. We 508 also highly appreciate the expert support of the IBBL processing and biorepository teams. F.Q.H.   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 September 2, 2021. 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint . CC-BY 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 September 2, 2021. 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint 644 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint   . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint 666 . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint  . CC-BY 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 September 2, 2021. ; https://doi.org/10.1101/2021.08.31.21262713 doi: medRxiv preprint    . CC-BY 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.

776
. CC-BY 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.