key: cord-1016775-w89rv1n0
authors: Goubet, A.-G.; DUBUISSON, A.; GERAUD, A.; DANLOS, F.-X.; TERRISSE, S.; ALVES COSTA SILVA, C.; DRUBAY, D.; TOURI, L.; PICARD, M.; MAZZENGA, M.; SILVIN, A.; DUNSMORE, G.; HADDAD, Y.; PIZZATO, E.; LY, P.; FLAMENT, C.; MELENOTTE, C.; SOLARY, E.; FONTENAY, M.; GARCIA, G.; BALLEYGUIER, C.; LASSAU, N.; MAEURER, M.; PIACENTINI, M.; GRAJEDA-IGLESIAS, C.; NIRMALATHASAN, N.; APRAHAMIAN, F.; DURAND, S.; KEPP, O.; FERRERE, G.; THELEMAQUE, C.; LAHMAR, I.; FAHRNER, J.-E.; MEZIANI, L.; AHMED-BELKACEM, A.; SAIDANI, N.; LA SCOLA, B.; RAOULT, D.; GENTILE, S.; CORTAREDONA, S.; IPPOLITO, G.; LELOUVIER, B.; ROULET,
title: Prolonged SARS-CoV-2 RNA virus shedding and lymphopenia are hallmarks of COVID-19 in cancer patients with poor prognosis
date: 2021-05-05
journal: nan
DOI: 10.1101/2021.04.26.21250357
sha: 5ae8f3614e7d4c275a1c7c900df2d74db98505d0
doc_id: 1016775
cord_uid: w89rv1n0

Patients with cancer are at higher risk of severe coronavirus infectious disease 2019 (COVID-19), but the mechanisms underlying virus-host interactions during cancer therapies remain elusive. When comparing nasopharyngeal swabs from cancer and non-cancer patients for RT-qPCR cycle thresholds measuring acute respiratory syndrome coronavirus-2 (SARS-CoV-2) in 1063 patients (58% with cancer, 89% COVID-19+), we found that malignant disease favors the magnitude and duration of viral RNA shedding concomitant with prolonged serum elevations of type 1 IFN that anticorrelated with anti-RBD IgG antibodies. Chronic viral RNA carriers exhibited the typical immunopathology of severe COVID-19 at the early phase of infection including circulation of immature neutrophils, depletion of non-conventional monocytes and a general lymphopenia that, however, was accompanied by a rise in plasmablasts, activated follicular T helper cells, and non-naive Granzyme B+FasL+, EomeshighTCF7high, PD-1+CD8+ Tc1 cells. Virus-induced lymphopenia worsened cancer-associated lymphocyte loss, and low lymphocyte counts correlated with chronic SARS-CoV-2 RNA shedding, COVID-19 severity and a higher risk of cancer-related death in the first and second surge of the pandemic. Lymphocyte loss correlated with significant changes in metabolites from the polyamine and biliary salt pathways as well as increased blood DNA from Enterobacteriaceae and Micrococcaceae gut family members in long term viral carriers. We surmise that cancer therapies may exacerbate the paradoxical association between lymphopenia and COVID-19-related immunopathology, and that the prevention of COVID-19-induced lymphocyte loss may reduce cancer-associated death.

Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) is a novel beta-185 coronavirus that has caused a worldwide pandemic of the human respiratory illness 19, resulting in a severe threat to public health and safety worldwide. Because of age, gender, 187 cancer-associated risk factors, metabolic syndrome and side effects induced by their specific 188 therapies (such as cardiomyopathy, systemic immunosuppression and cellular senescence), 189 cancer patients appear more vulnerable to severe infection than individuals without cancer 190 (Derosa et al., 2020) . Indeed, a high hospitalization and mortality rates of SARS- infection were heralded in patients with malignancy in several studies across distinct 192 geographical sites (Albiges et al., 2020; Assaad et al., 2020; Luo et al., 2020; Rugge et al., 193 S1 ). In the Cancer_FR1_TR study, 33%, 21% and 46% presented with localized, locally 248 advanced and metastatic disease, respectively that were equally susceptible to severe COVID-249 19 (Figure S1F-G). 250

Given that cycle threshold (Ct) values of the first RT-qPCR test may be correlated with the 251 clinical characteristics of the patients (Shlomai et al., 2020; Westblade et al., 2020) , we 252 performed a longitudinal follow-up of Ct values by RT-qPCR. We targeted several genes 253 coding for the envelope, the nucleocapsid and/or the replication-transcription complex (RdRP, 254 Orf1a, subgenomic RNA of the SARS-CoV-2 (Corman et al., 2020) (Wölfel et al., 2020) ) to 255 assess the duration of the nasopharyngeal SARS-CoV-2 RNA shedding, starting at COVID-256 19 diagnosis for up to 6 months as per-protocol indications ( Figure S2A ). The duration of 257 viral shedding was defined as the number of days from the first positive to the first negative 258 RT-qPCR, after longitudinal monitoring with an interval inferior to 40 days, to reduce bias in 259 viral shedding estimation. This time lapse of 40 days corresponded to the median of SARS-260

CoV-2 virus carriage in the cancer population ( Figure 1B -C, Table S1 ). In parallel, a similar 261 and systematic COVID-19 protocol with longitudinal RT-qPCR testing was applied to health 262 care workers (HCW) at Gustave Roussy. Health care workers had a mean age of 35 years 263 (range: 19-61), were mostly females (male versus female: 13% versus 87%), and presented 264 with one or two co-morbidities in 27% and 4%, respectively, thereby significantly diverging 265 from the cancer population diagnosed with COVID-19. Starting from 50 COVID-19 positive 266 cancer patients and 100 HCW, we conducted RT-qPCR in 210 and 200 nasopharyngeal 267 swabs, respectively ( Figure S2 ). However, applying the exclusion criteria detailed in Figure  268 S2, we could compare the median length of SARS-CoV-2 RNA detection in 35 cancer 269 patients (Cancer_FR1_TR) and 45 HCW using 168 and 118 samples, respectively. Patients 270 with cancer exhibited prolonged nasopharyngeal RNA virus shedding ( Figure 1B , median of 271 40 days (range: 6-137) for patients with cancer compared to 21 days (range: 7-53) for HCW, 272 Figure 1C , log-rank test p-value <0.0001). This difference persisted after adjusting for age, 273 gender and co-morbidities (Cox multivariate analysis, adjusted hazard ratio [95% confidence 274 interval] =2.88 [1.42;5/85], p=0.00291, Figure 1C ). To further validate the differences 275 observed in the duration of viral RNA shedding between Cancer_FR1_TR and HCW, we 276 analyzed another cohort of patients diagnosed with COVID-19 in a general hospital from 277 positivity in cancer individuals compared with cancer free COVID-19 patients (8 days versus 282 6 days, log-rank test p-value, p=0.03), taking into account that >70% were treated with 283 hydroxychloroquine and azithromycine, a combination regimen reducing viral shedding 284 (Lagier et al., 2020) . Moreover, the proportion of patients with a viral shedding above 16 days 285 (corresponding to the 90 th percentile of the viral shedding in cancer-free patients) was higher 286 in cancer patients ( Figure 1D , p<0.0015). A second independent validation was achieved in a 287 third series of 66 patients with cancer extracted from a cohort of 252 cancer individuals living 288 in Canada and diagnosed with COVID-19 (Cancer_CA), for whom a longitudinal SARS-289

CoV-2-specific RT-qPCR (using Orf1 and E gene probe sets (Boutin et al., 2020) ) follow-up 290 had been carried out (Elkrief et al., 2020) ( Figure 1A , Table S1 ). Here again, we observed that 291 26% of cancer patients were still PCR positive after 40 days from diagnosis by RT-qPCR 292 ( Figure 1E ). Such a long term PCR detection of viral RNA could indicate stable subgenomic 293 RNA contained within double membrane vesicles or presence of a replicative mucosal viral 294 strain. Hence, we confirmed in three independent series of cancer patients a prolongation of 295 RNA virus shedding previously described in case reports in hematological or immuno-296 compromised patients (Avanzato et al., 2020; Aydillo et al., 2020; Choi et al., 2020; 297 Helleberg et al., 2020) . 298 299 Hence, we focused on the differential characteristics of cancer patients presenting with Long 300 term Viral RNA Shedding (LVS), defined by a positive RT-qPCR duration ≥ 40 days (median 301 of RT-qPCR duration in Cancer_FR1_TR ( Figure 1C) ), compared to those experiencing Short 302 term Viral RNA Shedding (SVS), defined by a positive RT-qPCR duration < 40 days 303 henceforth (Table S1 ). The increased susceptibility to develop a LVS was independent of 304 initial symptomatology, observed in 33% of Canadian (CA) to 40% of French (FR1_TR) 305 asymptomatic and 27% (CA) to 56% (FR1_TR) of symptomatic cancer patients ( Figure 1F) . 306

There was a higher propensity to LVS in hematological malignancies compared to solid 307 cancers (86% versus 43%, respectively (p=0.04, Figure 1G , Table S1 ) and in advanced 308 disease (p=0.011) in FR1_TR cohort ( Figure 1H , Table S1 ) but less so, in the CA cohort. 309 Importantly, the LVS phenotype was associated with COVID-19 severity, notably an 310 increased risk to develop a moderate form (defined by thoracic CT scan, hospitalization and 311 oxygen requirement < 9 L/min) in Cancer_FR1_TR (p=0.032) ( Figure 1I ). This trend was 312 confirmed in a third series of French patients from the clinical routine (CR) managed outside 313 the translational ancillary study at Gustave Roussy (called henceforth "Cancer_FR1_CR"; 314 Table S1, Figure S3) , where 20% cancer patients were diagnosed with LVS and exhibited 315 more severe COVID-19 infections (Figure 1I, p=0.011) . Again, the hospitalization rates and 316 transfer to intensive care units were increased in LVS compared with SVS patients in 317

Cancer_FR1_TR (p=0.0018, Table S1 ) and Cancer_FR2, respectively (p=0.02, Table S1 ). 318

Finally, the FR2 and Canadian series of LVS cancer patients also tended to exhibit more 319 severe manifestations of COVID-19 compared with SVS Canadian cancer patients ( Figure 1I , 320 bottom). 321 322 Of note, the duration of viral RNA shedding correlated with "viral load", i.e Ct values 323 at diagnosis, in that cancer patients with LVS experienced lower Ct values at diagnosis than 324 SVS cancer patients in most cohorts for which the data were available ( Figure 1J ). 325

Importantly, cancer patients doomed to develop LVS presented with lower Ct values at 326 diagnosis than those prone to become SVS in Cancer_FR1 and Cancer_FR2 cohorts ( Figure  327 1K). The dynamic course of the infection was significantly different in SVS and LVS, as 328 indicated by the Ct values that remained high for a prolonged period of time in LVS patients 329 compared with SVS ( Figure 1L ). Of note, Ct values at disease onset were significantly 330 anticorrelated with duration of viral RNA shedding in cancer patients using either N or 331

Orf1ab/RdRP gene-specific probe sets. 332

The redundancy analysis (RDA) is an extension of the principal component analysis 333 (PCA) aimed at identifying viral components which depend on other known covariates such 334 as clinical parameters. RDA revealed that, within 30 days from diagnosis, 18% of the 335 variance of the biological parameters are explained by 10 components adjusted for the major 336 clinical parameters for COVID-19 in Cancer_FR1_TR ( Figure 1M) (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 348   349   Intrigued by these findings, we addressed the question as to whether and how  350   prolonged viral RNA shedding would impact on Cancer_FR1_TR patients with respect to  351 COVID-19-related immunological alterations previously reported for cancer-free infected 352 individuals (Arunachalam et al., 2020; Chua et al., 2020; Kaneko et al., 2020; Laing et al., 353 2020; Mathew et al., 2020; Silvin et al., 2020; Takahashi et al., 2020) . More than 80 354 phenotypic markers were quantified on circulating leukocytes by means of high dimensional 355 spectral flow cytometry, complemented by multiplex ELISAs to detect serum chemokines, 356 cytokines and growth factors. These parameters were recorded within or after the first 20 days 357 of inclusion in the Cancer_FR1_TR protocol, for 25 COVID-19 + cancer patients that were 358 divided into LVS versus SVS subgroups, in comparison to 43 COVID-19 negative cancer 359 patients ("controls" or "Ctls") matched for age, gender, co-morbidities, cancer types and 360 tumor extension (Table S2 ). Asymptomatic individuals and cancer patients enrolled at the 361 recovery phase of COVID-19 (meaning that they became PCR negative) were analyzed 362 separately. Within the first 20 days from diagnosis, LVS presented increased proportions of 363 monocytes among circulating leukocytes ( Figure S4A , left panel), and a parallel drop in 364 CD169 -HLA-DR + within conventional monocytes ( Figure S4A , middle panel) and in non-365 conventional monocytes (CD16  + CD14 low/-, Figure S3A right panel) compared to SVS, 366 cancer controls, asymptomatic or recovered patients, as reported (Carvelli et al., 2020; Silvin 367 et al., 2020) . Polymorphonuclear cells (PMN) tended to increase in LVS, specifically 368 immature CD101

neutrophils, compared with SVS, convalescent and 369 controls ( Figure 2A -B, upper and lower panels, Figure S4B ). 370 371 Importantly, the most significant phenotypic traits distinguishing LVS from SVS 372 featured among the reported hallmarks of severe COVID-19 in cancer-free subjects 373 (Arunachalam et al., 2020; Chua et al., 2020; Kaneko et al., 2020; Laing et al., 2020; Mathew 374 et al., 2020; Silvin et al., 2020; Takahashi et al., 2020) (Figure 2A ). In accordance with the 375 reported defects in germinal center formation in secondary lymphoid organs of severe 376 COVID-19 (Kaneko et al., 2020) , LVS cancer patients exhibited increased recirculation of 377 activated CXCR5 + PD-1 high CD4 + follicular T helper cells (TFH) expressing ICOS and CD38 378 ( Figure 2C , left panel), as well as a marked rise in plasmablasts (defined as CD19 low CD27 hi 379 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 release of, and exposure to, type 1 IFN above levels measured in SVS, controls and recovered 385 individuals ( Figure 2E ). Type 1 IFN levels anticorrelated with titers of neutralizing anti-S1 386 RBD antibodies ( Figure 2F ). This landscape of immune profiling was corroborated by non-387 supervised hierarchical clustering of innate and cognate immunotypes and serum cytokine 388 concentrations analyzed within 30 days from diagnosis. This method allowed to segregate a 389 small cluster of individuals characterized by low Ct values (< 25), and moderate/severe 390 complications of COVID-19, that included metastatic cancer carriers with LVS or SVS 391 ( Figure S5 ). This cluster was separated from the others by typical signs of viral infection, 392

including abundant circulating CD38 + HLA-DR + CD8 + T cells, plasmablasts, activated TFH 393 cells and high serum IFNα2a levels ( Figure S5 ). Likewise, while many inflammatory 394 cytokines, chemokines or alarmins (such as IFNγ, CXCL10, IL-4, IL-6 and calprotectin) were 395 elevated in symptomatic COVID-19 individuals compared with controls, asymptomatic and 396 recovered patients, none of them could predict LVS, except a drop in the IFNγ/IFNα2a and 397 CCL11/CXCL10 ratios whose significance remains unclear (p=0.016 and p=0.0019, 398 respectively) ( Figure S4D -I). Interestingly, innate and cognate immunotypes performed in 399 convalescent patients and controls segregated at random in the non-supervised hierarchical 400 clustering ( Figure S6 ). 401

Altogether, the high dimensional flow cytometry of blood immune subsets indicated 402 that LVS cancer patients harbored the immunological hallmarks of severe COVID-19 at 403 diagnosis. 404 405 Virus-associated lymphopenia predicted shorter overall survival in the first and second 406 surge of the pandemic 407 408 Lymphocyte loss is a feature of severe COVID-19 in non-cancer patients (Laing et al., 409 2020; Mathew et al., 2020) . Not surprisingly, blood absolute lymphocyte counts (ALC) at 410 diagnosis anticorrelated with the duration of PCR positivity in Cancer_FR1_TR and 411

Cancer_FR1_CR cohorts ( Figure 3A ). However, although the ALC before the COVID-19 412 pandemic (blood drawn from December 2019 to mid-March 2020) were already somewhat 413 lower in LVS than in SVS cancer patients, the ALC during the outbreak dramatically dropped 414 in cancer patients doomed to develop LVS (in both Cancer_FR1_TR and Cancer_FR1_CR 415 cohorts), more so than in individuals prone to SVS ( Figure 3B , left panel). The extent in ALC 416 reduction was more severe in patients presenting LVS than SVS ( Figure 3C ). Of note, ALC 417 recovered in both patient groups regardless of the LVS/SVS status, supporting the contention 418 that reduced ALC at COVID-19 diagnosis is induced by the virus rather than by the cancer 419 ( Figure 3B , left panel). In accord with the finding that LVS correlates with high viral load at 420 symptom onset ( Figure with Ct values predicted cancer-related overall survival in univariate analyses across all 433 cancer stages (local, locally advanced or metastatic) ( Figure 3D , (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. ; https://doi.org/10.1101/2021.04.26.21250357 doi: medRxiv preprint ( Figure 3E ). Of note, treatment retardation could not explain the high mortality of patients 445 presenting with a high viral load and low ALC (Table 1) . 446

We confirmed these predictors (ALC < 800 & Ct < 25) of poor survival during the 447 second surge of the pandemic (between May 5 th , 2020 to November 25 th , 2020) in 116 new 448 COVID-19 cancer patients ("Validation", Cancer_FR1 and Cancer_FR3, Figure 1A Multiple and non-exclusive mechanisms could account for virus-associated lymphopenia 462 (Arunachalam et al., 2020; Campbell et al., 2020; Mathew et al., 2020; Moole and 463 Papireddypalli, 2020; Sheng et al., 2017; Song et al., 2020) . To further investigate this 464 deleterious virus-induced lymphocyte loss, we searched for the most robust correlates 465 between ALC and immunological, metabolic or pathogenic cues in the Cancer_FR_TR cohort 466 as well as non-cancer COVID-19 patients that we previously reported (Silvin et al., 2020) . Secondly, we performed the serum metabolome determined by untargeted and targeted 486 mass spectrometry-based metabolomics analyzing more than 221 metabolites in 31 cancer 487 patients from Cancer_FR1_TR, as well as in a previously described cohort of 66 cancer-free 488 COVID-19 + patients for validation (Silvin et al., 2020) . The non-supervised hierarchical 489 clustering of the serum metabolome clearly contrasted LVS from LVS patients ( Figure S9 ). 490

The Volcano plot aimed at identifying significant differences between LVS and SVS patients 491 pointed out the biliary salt metabolic pathway segregating SVS from LVS serum ( Figure 5A were decreased in LVS (compared to controls and SVS, Figure S10A -B, left panels) and were 499 associated with lymphocyte loss ( Figure S10A and S10B, right panels). 500

Another metabolic pathway pertaining to polyamines with high biological significance for 501 age-related immunosenescence (Alsaleh et al., 2020; Puleston et al., 2014; Zhang et al., 2019) 502 was also strongly associated with the duration of RT-qPCR positivity, ALC and disease 503 severity ( Figure 5E -G, Figure S9 ). In particular, the N1, N8-diacetylspermidine that 504 anticorrelated with ALC ( Figure 5F , right panel) increased in the serum of LVS patients (but 505 not SVS, Figure 5F , left panel), in accordance with its marked rise in severe COVID-19 in 506 cancer-free individuals ( Figure 5G , left panel) where high levels coincided with the 507 lymphocyte drop ( Figure 5G , right panel). Of note, the tryptophane/kynurenine or lactic acid 508 metabolites were not relevant in our study ( Figure 5A , Figure S9 ). 509

Thirdly, endotoxemia was shown to correlate with the cytokine storm during COVID-510 19 (Arunachalam et al., 2020) and might cause activation-induced lymphocyte cell death. 511

Assuming that the gut permeability could be altered during COVID-19-associated intestinal 512 dysbiosis (Yeoh et al., 2021) , we studied the circulating microbial populations associated with 513 whole leukocytes by sequencing blood rDNA using next-generation sequencing of V3-V4 514 variable regions of the 16S rRNA bacterial gene as previously described (Païssé et al., 2016) . 515

Although we failed to observe significant quantitative differences in blood bacterial load 516 between SVS (n=14) and LVS (n=15) patients, the linear discriminant analysis effect size 517 indicated significant taxonomic differences in the bacteria family members between the two 518 groups ( Figure 6A -B). The DNA from Enterobacteriaceae (mainly composed of Escherichia 519

Shigella genus) was overrepresented in leukocytes of LVS compared with SVS patients 520 ( Figure 6A -B, C left panel). The circulating Enterobacteriaceae-related DNA markedly 521 anticorrelated with CCL22 (a hallmark of SVS, Figure 2A ), but was strongly associated with 522 the increase of exhausted CD8 + T lymphocytes ( Figure To interrogate viral-host interactions during the COVID-19 pandemic in cancer 533 patients, we studied 1106 patients, among them 59% were cancer bearers 534 (FR1+FR2+FR3+CA), and 1063 COVID-19 positive ( Figure 1A ). We used high dimensional 535 flow cytometry to perform deep immune profiling of innate, B and T cells and measurements 536 of 51 soluble markers, with temporal analysis of immune changes during infection in one 537 cohort that was further explored by blood metabolomics and metagenomics. This longitudinal 538 immune analysis was linked to virologic and oncological data ( Figure S5 -S6). Using this 539 approach, we made several intriguing observations. 540 541 First, 51%, 20% and 26% of cancer patients in FR1_TR, FR1_CR and CA, 542 respectively, still shed SARS-CoV-2 RNA after day 40 from symptoms onset (versus 2% in 543 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. ; https://doi.org/10.1101/2021.04.26.21250357 doi: medRxiv preprint replication-competent viral strains between 10 and 20 days after symptom onset has been 545 documented in some persons with severe COVID-19, mostly in immunocompromised cases 546 (Kampen et al., 2020) . However, about 90% of their specimens no longer yielded replication-547 competent viruses after day 15 from symptom onset (Ge et al., 2020; Li et al., 2020) . 548

Prolonged shedding of influenza, parainfluenza, rhinovirus, seasonal coronavirus and 549 respiratory syncytial virus has previously been detected in immunosuppressed patients (El 550 Ramahi and Freifeld, 2019; Geis et al., 2013; Lehners et al., 2016 Lehners et al., , 2013 Milano et al., 2010) . 551

Cancer dissemination, cancer therapies and virus-induced lymphopenia might cause an 552 immunodeficiency that eventually jeopardizes virus clearance. The proposed mechanisms by 553 which lymphopenia occurs in COVID-19 (often shared with cancer dissemination) (Péron et 554 al., 2013) include virus-induced atrophy of secondary lymphoid organs (Buja et al., 2020; Lax 555 et al., 2020; Xu et al., 2020) , disappearance of germinal centers (Kaneko et al., 2020), direct 556 pro-apoptotic activity of the virus related to ACE2-dependent or ACE2-independent entry 557 into lymphocytes (Pontelli et al., 2020) , T cell demise consecutive to activation and 558 exhaustion (Channappanavar and Perlman, 2017; Park, 2020) , senescence (Derosa et al., 559 2020; Westmeier et al., 2020) , and antiproliferative effects of lactic acid (Fischer et al., 2007) . 560

However, in our study, we found that lymphocyte loss was correlated with decrease of 561 secondary biliary salts in LVS patients, most likely associated with increased gut permeability 562 that lead to bacterial translocation, as we observed increased circulating DNA for 563

Micrococcaceae and Enterobacteriaceae family members. Moreover, the transformation of 564 spermidine into N1, N8 diacetylspermidine was linked to decreased ALC, in accordance with 565 the role of spermidine in preventing ageing-related loss of lymphocyte fitness (Alsaleh et al., 566 2020; Puleston et al., 2014; Zhang et al., 2019) . 567 568 Second, prolonged viral RNA carriage was associated with signs of immunopathology 569 (exacerbated T cell responses, extrafollicular TFH, and plasmablast recirculation, exhausted 570 PD-1 + Tc1 cells, sustained serum type 1 IFN levels), likely maintaining a positive feed-back 571 loop for the expression of the interferon-signaling genes product ACE2 (Ziegler et al., 2020) 572 and pro-inflammatory interactions between airway epithelia and immune cells (Chua et al., 573 2020) . 574 575 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. ; https://doi.org/10. 1101 /2021 Third, prolonged SARS-CoV-2 RNA shedding after day 40 might precede the 576 aggravation of both COVID-19 and malignant disease. Indeed, virus and/or cancer-induced 577 lymphopenia and T cell exhaustion may jointly enfeeble tumor immunosurveillance (Rd et al., 578 2011) . Interestingly, SARS-CoV-2 virus -induced immunopathology was accompanied by 579 increased blood levels of IL-8 ( Figure S4G ) and VEGF (Takahashi et al., 2020) , which are 580 well known pro-angiogenic and pro-tumorigenic growth factors, predicting failure to cancer 581 immunotherapy (Sanmamed et al., 2017) . Of note, patients with high initial viral loads or 582 LVS tended to accumulate poor prognosis-related parameters than SVS or patients with 583

higher Ct values in both cohorts (Table 1 & Table S1 ), being older (66 versus 56 years old, 584 p=0.08), more metastatic at diagnosis of infection (72% versus 29%, p=0.011), and increased 585 hospitalization rates (83% versus 23%, p=0.001). As a result, virus-induced lymphopenia 586 markedly predicted early death of patients, within the first 2-3 months post-COVID-19 587 diagnosis in the first and second surge of the pandemic (in more than 200 patients) and call 588 for caution to administer chemotherapy or steroids at the acute phase of the viral infection that 589 exacerbate immunosuppression. 590 591 These observations call for a careful follow up of cancer patients, in particular those 592 bearing hematological and metastatic malignancies, during the second wave of Given the non-consensual efficacy of vaccines against influenza virus in vulnerable 594 individuals suffering from cancer-, virus-and age-associated lymphopenia (Bitterman et al., 595 2018; Péron et al., 2013) , passive immunization of high affinity neutralizing monoclonal 596 antibodies against SARS-CoV-2 at COVID-19 onset might be envisaged. This could be 597 combined with therapeutic stimulation of lymphopoiesis (for instance with rIL-7, G-CSF, 598 inhibitors of indole amine 2,3 deoxygenase), to achieve an immunological tonus that is 599 compatible with anticancer treatments (Cheng et al., 2020; Francois et al., 2018; Laterre et al., 600 2020) . Clinical trials are underway to evaluate rIL-7 against COVID-19, but may benefit from 601 patients stratification based on Ct values, duration of viral RNA shedding and ALC 602 (NCT04407689; NCT04426201; NCT04379076). 603 604 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. A., Cuneo, A., Romano, A., Giuliani, N., Galimberti, S., Corso, A., Morotti, A., 963

Falini, B., Billio, A., Gherlinzoni, F., Visani, G., Tisi, M.C., Tafuri, A., Tosi, P., 964

Lanza, F., Massaia, M., Turrini, M., Ferrara, F., Gurrieri, C., Vallisa, D., Martelli, M., 965

Derenzini, E., Guarini, A., Conconi, A., Cuccaro, A., Cudillo, L., Russo, D., 966

Ciambelli, F., Scattolin, A.M., Luppi, M., Selleri, C., Ortu La Barbera, E., Ferrandina, 967 C., Di Renzo, N., Olivieri, A., Bocchia, M., Gentile, M., Marchesi, F., Musto, P., 968

Federici, A.B., Candoni, A., Venditti, A., Fava, C., Pinto, A., Galieni, P., Rigacci, L., 969

Armiento, D., Pane, F., Oberti, M., Zappasodi, P., Visco, C., Franchi, M., Grossi, 970 P.A., Bertù, L., Corrao, G., Pagano, L., Corradini, P., 2020. Clinical characteristics 971 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. < 25, p=0.0738) . Chi-square tests with *p<0.05, **p<0.01, 1139 ***p<0.001, ****p<0.0001. among CXCR5 + PD-1 + non-naive CD4 + (C, left panel) , plasmablasts defined as 1167 CD19 low CD38 high CD27 + within the CD19 + gate (C, right panel), double negative IgD -CD27 -1168 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. pandemic ("PER", between -4 and +7 days of the disease diagnosis by RT-qPCR), between 1187 210 and 12 days before the symptom onset of COVID-19 ("PRE") or within the recovery 1188 period (between 0 and 123 days after negative RT-qPCR) ("POST") at Gustave Roussy, with 1189 calculation of the reduction between "PRE" and during COVID-19 (C). One patient defined 1190 as outlier (at 215%) by ROUT method was excluded from the LVS group for the analysis. included, according to ALC and Ct value at diagnosis. Refer to Table 1 for patient  1197 characteristics. E. Multivariate Cox regression analysis stratified for the cohort and adjusted 1198 for age, ECOG status, gender and metastatic and/or hematological status of cancer patients 1199 from the Discovery (1st surge) cohort (Cancer_FR1+Cancer_FR3). F. Kaplan Meier curve 1200

and Cox regression analysis of overall survival of cancer patients from Validation (2nd surge) 1201 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. ; https://doi.org/10.1101/2021.04.26.21250357 doi: medRxiv preprint cohort (Cancer_FR1+Cancer_FR3), all stages included, according to ALC and Ct value at 1202 diagnosis. Refer to Table 1 (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 5, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 (GzB) + PD-1 + in Eomes hi TCF-1 hi non-naïve CD8 + measured in blood (E). F. Idem as in A. validation cohorts presenting cycle threshold below (Ct<25) or above 25 (Ct>25) and with (< 1283 800/mm 3 ) or without (> 800/mm 3 ) lymphopenia at diagnosis (refer to Figure 3D (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Determinants of the 680 outcomes of patients with cancer infected with SARS-CoV-2: results from the 681

Autophagy in T cells from aged donors 685 is maintained by spermidine and correlates with function and vaccine responses

Rapid viral diagnosis and ambulatory management of suspected 691 COVID-19 cases presenting at the infectious diseases referral hospital in Marseille

2020: A respiratory virus snapshot

Type 2 diabetes influences bacterial tissue compartmentalisation in human 697 obesity

Systems biological assessment of immunity to mild versus severe COVID-19 704 infection in humans

High mortality rate in cancer patients with symptoms 713 of COVID-19 with or without detectable SARS-COV-2 on RT-PCR

Case Study: Prolonged Infectious SARS-CoV-2 718 Shedding from an Asymptomatic Immunocompromised Individual with Cancer

Shedding of Viable 723

SARS-CoV-2 after Immunosuppressive Therapy for Cancer

Dimensionality reduction for visualizing single-cell data using 727 UMAP

Influenza vaccines in immunosuppressed adults with cancer. Cochrane Database 730 of Systematic Reviews

Comparison of SARS-CoV-2 detection from combined 733 nasopharyngeal/oropharyngeal swab samples by a laboratory-developed real-time RT-734 PCR test and the Roche SARS-CoV-2 assay on a cobas 8800 instrument

The emerging spectrum of cardiopulmonary pathology of the coronavirus 739 disease 2019 (COVID-19): Report of 3 autopsies from Houston, Texas, and review of 740 autopsy findings from other United States cities

Bacterial metabolism of bile acids promotes 745 generation of peripheral regulatory T cells

Association of COVID-19 inflammation with 751 activation of the C5a-C5aR1 axis

Pathogenic human coronavirus infections: causes and 754 consequences of cytokine storm and immunopathology

Effect of Recombinant Human

COVID-19) and Lymphopenia: A Randomized Clinical Trial

Persistence and Evolution of SARS-CoV-2 in an Immunocompromised 769 Host

773 and risk factors associated with COVID-19 severity in patients with haematological 972 malignancies in Italy: a retrospective, multicentre, cohort study

CD4 lymphopenia to identify end-of-life 977 metastatic cancer patients

Infection of human lymphomononuclear cells by SARS-CoV-2

Autophagy is a 989 critical regulator of memory CD8(+) T cell formation

Clinical outcomes of 992 coronavirus disease 2019 (COVID-19) in cancer patients with prior exposure to 993 immune checkpoint inhibitors

Cancer immunoediting: integrating immunity's roles in cancer 996 suppression and promotion

Determinants of COVID-19 1003 disease severity in patients with cancer

SARS-CoV-2 infection in the Italian Veneto 1006 region: adverse outcomes in patients with cancer

Changes in serum interleukin-8 (IL-8) levels reflect 1012 and predict response to anti-PD-1 treatment in melanoma and non-small-cell lung 1013 cancer patients

Trust is good, control is better: technical considerations in blood microbiome 1017 analysis

Hepatic 1020 inflammation caused by dysregulated bile acid synthesis is reversible by butyrate 1021 supplementation

Nasopharyngeal viral load predicts hypoxemia and disease outcome in admitted 1024 COVID-19 patients

Elevated 1033 Calprotectin and Abnormal Myeloid Cell Subsets Discriminate Severe from Mild 1034 COVID-19

Microbial bile acid metabolites modulate 1037 gut RORγ+ regulatory T cell homeostasis

IgA dominates the early neutralizing 1043 antibody response to SARS-CoV-2

Sex differences in immune responses that underlie COVID-19 1051 disease outcomes

SARS-CoV-2 Viral Load Predicts Mortality in Patients with and without 1056 Cancer Who Are Hospitalized with COVID-19. Cancer Cell

Impaired 1061 Cytotoxic CD8+ T Cell Response in Elderly COVID-19 Patients

Virological assessment of hospitalized patients with COVID-2019

Gut microbiota composition reflects disease severity and 1076 dysfunctional immune responses in patients with COVID-19

Polyamines Control eIF5A Hypusination, TFEB Translation, and 1081 Autophagy to Reverse B Cell Senescence

HCA Lung Biological Network. Electronic address: lung-1094 network@humancellatlas.org, HCA Lung Biological Network

Receptor ACE2 Is an Interferon-Stimulated Gene in Human Airway Epithelial Cells 1096 and Is Detected in Specific Cell Subsets across Tissues

PCae(22:4) PCae(20:2)

:1) PCae

We are thankful to the Spectral flow cytometry facility team of Gustave Roussy. We thank 605 the ET-EXTRA team (Biological Resource Center (NF 96-600) and the microbiology team for technical help.

 Table 1  1319 and Table S1 ). Patients diagnosed for COVID-19 from April 10 th , 2020 to May 4 th , 2020 were 1320 included in the Discovery cohort and patients from May 5 th , 2020 to November 25 th , 2020 1321 were included in the Validation cohort. Trial design. Cancer patients were screened for 1322 SARS-CoV-2 virus carriage by nasopharyngeal sampling at every hospital visit. The presence 1323 of SARS-CoV-2 RNA was detected by RT-qPCR assay in a BSL-2 laboratory. Asymptomatic 1324 and symptomatic patients (i.e presenting with fever (t°>38°C) and/or cough and/or shortness 1325 of breath and/or headache and/or fatigue and/or runny nose and/or sore throat, 1326 anosmy/agueusia) with a positive SARS-CoV-2 RT-qPCR test, shifted to the interventional 1327 phase (tailored experimental approach with Hydroxychloroquine and Azithromycin therapy in 1328 symptomatic SARS-CoV-2 positive subjects). Asymptomatic or symptomatic patients with 1329 negative SARS-CoV-2 RT-qPCR test continued their standard of care anti-cancer treatments. 1330Repeated RT-qPCR for SARS-CoV-2 on nasopharyngeal swabs and blood samples were 1331 performed to monitor the status for SARS-CoV-2 and the immune response, respectively, in 1332 COVID-19 positive and negative patients. The COVID-19 severity was defined based on 1333 oxygen, imaging and hospitalization criteria. Patients with a mild COVID-19 disease had 1334 limited clinical symptoms not requiring scan or hospitalization; patients with a moderate 1335 COVID-19 disease were symptomatic with a dyspnea and radiological findings of pneumonia 1336 on thoracic scan requiring hospitalization and a maximum of 9L/min of oxygen; severe 1337 patients had respiratory distress requiring intensive care and/or more than 9L/min of oxygen. 1338Samples for translational research. Whole blood was used for high dimensional spectral 1339 flow cytometry analyses. Serum samples were used to monitor the concentrations of 1340 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 19. Of note, >75% received hydroxychloroquine and >96% received azithromycine (Table  1364 S1) (Amrane et al., 2020; Lagier et al., 2020) . This study was approved by the IHU 1365Méditerranée Infections review board committee (Méditerranée Infection N°: 2020-021). 1366 1367 4/ Third series of cancer patients from Canada (Cancer_CA). We used 66 individuals from 1368 the clinical cohort previously reported (Elkrief et al., 2020) for whom data were available 1369 (Table S1 ). This study was conducted across eight Canadian institutions in Quebec and 1370British Columbia and was approved by the institutional ethics committee at each site (Ethics 1371 number: MP-02-2020-8911 and H20-00892). 1372 1373 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

We used the clinical 1374 cohort previously reported (Albiges et al., 2020) (Table S1) (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Patients from each cohort were classified using the same criteria. The cut-off value of 25 for the cycle threshold was based on the median calculated on 1427Cancer_FR1_TR and the mean calculated on Cancer_FR1_TR+CR. 1428 RT-PCR for subgenomic RNA (sgRNA) for SARS-CoV-2. We used the protocol 1429 previously described by Wölfel et al. in Nature, 2020 . Briefly, the oligonucleotide sequence 1430 of the leader specific primer was as follows: sgLeadSARSCoV2-F; 1431 CGATCTCTTGTAGATCTGTTCTC, and the oligonucleotide sequence of the E primer was 1432 as follows: E_Sarbeco_R; ATATTGCAGCAGTACGCACACA. Briefly, 5 uL of RNA (>21 1433 ng) were used for the sgRNA RT-PCR assay with Superscript III one-step RT-PCR system 1434 with Platinum Taq Polymerase (Invitrogen, Darmstadt, Germany) with 400 nM concentration 1435 of each primer. Thermal cycling was set up as described. Finally, RT-PCR products for 1436 sgRNAs were analyzed on agarose gel 2%. 1437Evaluation of SARS-CoV-2 RNA shedding. The duration of viral shedding was defined as 1438 the number of days from the first positive to the first negative RT-qPCR, after longitudinal 1439 monitoring. In order to prevent an overvaluation of this duration, we considered in this 1440 analysis only patients with an interval below 40 days between the last positive RT-qPCR and 1441 the first negative RT-qPCR. Six patients had one negative RT-qPCR followed by positive RT-1442 qPCR. We extend the duration to the second negative RT-qPCR for 3 patients with a cycle 1443 threshold below 35 for the gene coding replication-transcription complex and within 6 days 1444 after the first negative result. 1445Absolute Lymphocyte Count (ALC). The absolute lymphocyte count was measured for the 1446 clinical routine using the Sysmex XN (Sysmex, Belgium). Values "PRE" were collected 1447 between 210 and 12 days before the symptom onset of COVID-19, values at diagnosis of the 1448 infection were collected between -4 and +7 days of the disease diagnosis by RT-qPCR, values 1449 "POST" were collected at the recovery time or later, meaning between 0 and 123 days after 1450 the first negative RT-qPCR. For the interpretation, the cut-off value for ALC was the median 1451 found in patients with high viral load at diagnosis (ALC=800/mm 3 ). In parallel, we 1452 considered this value as relevant according to the common terminology criteria for adverse 1453 events where grades of lymphopenia were assigned as following: grade 1 ALC < lower limit 1454 of normal to 800/mm3, grade 2 ALC < 800-500/mm3, and grade 3 ALC < 500-200/mm3. 1455Blood tests. Sampling. Blood samples were drawn from patients enrolled in ONCOVID at 1456 Gustave Roussy Cancer Campus (Villejuif, France). Whole human peripheral blood was 1457 collected into sterile vacutainer tubes. Spectral flow cytometry. One hundred and twenty-one 1458 whole blood samples from 88 patients (Supplementary Material Figure 1 ) was mixed at a 1:1 1459 ratio with Whole Blood Cell Stabilizer (Cytodelics), incubated at room temperature for 10 1460 min and transferred to -80°C freezer to await analysis. These samples were secondarily 1461 thawed in a water bath set to +37°C. Cells were fixed at a ratio 1:1 with Fixation Buffer 1462 (Cytodelics, ratio 1:1) and incubated for 10 min at room temperature. Red blood cells were 1463 lysed by addition of 2 mL of Lysis Buffer (Cytodelics, ratio 1:4) at room temperature for 10 1464 min. White blood cells were washed with 2 mL of Wash Buffer (Cytodelics, ratio 1:5). Cells 1465were resuspended in 100 µL extra-cellular antibody cocktail and incubated at room 1466 temperature for 15 min, then washed in Flow Cytometry Buffer (PBS containing 2% of fetal 1467 bovine serum and 2 mM EDTA). For intra-cellular labelling, a step of permeabilization was 1468 performed using 200 µL of eBioscience Foxp3 kit (ThermoFischer); cells were then incubated 1469for 40 min at +4°C, washed in Perm Buffer (ThermoFischer) and resuspended in intra-cellular 1470 antibody cocktail. After incubation, cells were washed in Flow Cytometry Buffer and 1471 resuspended to proceed to the acquisition. All antibodies used are listed in Supplemental 1472Material Table 2 . Samples were acquired on CyTEK Aurora flow cytometer (Cytek 1473 Biosciences). 16S rDNA metagenomic profiling. DNA from blood was isolated and 1474 amplified in a strictly controlled environment at Vaiomer SAS (Labège, France) using a 1475 stringent contamination-aware approach as discussed previously (Anhê et al., 2020; Lluch et 1476 Lluch et al., 2015 Païssé et al., 2016; Schierwagen et al., 2020) . The microbial populations based on 1477 rDNA present in blood were determined using next-generation sequencing of V3-V4 variable 1478 regions of the 16S rRNA bacterial gene as previously described (Lluch et al., 2015) . For each 1479 sample, a sequencing library was generated by addition of sequencing adapters. The joint pair 1480 length was set to encompass a 467 base pairs amplicon (using Escherichia coli 16S as a 1481 reference) with a 2 × 300 paired end MiSeq kit V3 (Illumina, San Diego, CA, USA). The 1482 detection of the sequencing fragments was performed using the MiSeq Illumina® technology. 1483Targeted metagenomic sequences from microbiota were analysed using the bioinformatic 1484 pipeline from the FROGS guideline (Escudié et al., 2018) . Briefly, the cleaning was done by 1485 removing amplicons without the two PCR primers (10% of mismatches were authorised), 1486 amplicons with at least one ambiguous nucleotide ('N'), amplicons identified as chimera 1487 (with vsearch v1.9.5), and amplicons with a strong similarity (coverage and identity ≥ 80%) 1488 with the phiX (library used as a control for Illumina sequencing runs). Clustering was 1489 produced in two passes of the swarm algorithm v2.1.6. The first pass was a clustering with an 1490 aggregation distance equal to 1. The second pass was a clustering with an aggregation 1491 distance equal to 3. Taxonomic assignment of amplicons into operational taxonomic units 1492 (OTUs) was produced by Blast+ v2.2.30+ with the Silva 134 Parc databank. To assess if the 1493 richness of microbiota was adequately captured by metagenomic sequencing, a rarefaction 1494 analysis was performed. To ensure a low background signal from bacterial contamination of 1495 reagents and consumables, two types of negative controls consisting of molecular grade water 1496 were added in an empty tube separately at the DNA extraction step and at the PCR steps and 1497 amplified and sequenced at the same time as the extracted DNA of the blood samples. The 1498 controls confirm that bacterial contamination was well contained in our pipeline and had a 1499 negligible impact on the taxonomic profiles of the samples of this study as published before 1500 (Anhê et al., 2020; Lluch et al., 2015; Païssé et al., 2016; Schierwagen et al., 2020) . One 1501 sample has been excluded of the analyses for aberrant profile. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Panel (Genalyte Inc) is designed to detect antibodies to five SARS-CoV-2 antigens: 1529 nucleocapsid, Spike S1 RBD, Spike S1S2, Spike S2 and Spike S1 with in a multiplex format 1530 based on photonic ring resonance technology (Sterlin et al., 2020) . This system detects and 1531 measure with good reproducibility changes in resonance when antibodies bind to their 1532 respective antigens in the chip. The instrument automates the assay. Briefly, 10µl of each 1533 serum samples were added in a sample well plate array containing required diluents and 1534 buffers. The plate and chip are loaded in the instrument. First the chip is equilibrated with the 1535 diluent buffer to get baseline resonance. Serum sample is then charged over the chip to bind 1536 specific antibodies to antigens present on the chip. Next, chip is washed to remove low 1537 affinity binders. Finally, specific antibodies of patients are detected with anti-IgG or -IgA or -1538 IgM secondary antibodies. 1539 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.Press). Briefly, serum samples were mixed with ice-cold extraction mixture (methanol/water, 1541 (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint this version posted May 5, 2021. after 20 days (>20d) of symptoms with those experiencing short term viral RNA shedding 1639 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint this version posted May 5, 2021.(SVS, orange dots) or long term viral RNA shedding (LVS, purple dots) and RT-qPCR 1640 negative COVID-19 patients in the convalescent phase (Recovery, green dots)). Box plots 1641 display a group of numerical data through their 3 rd and 1 st quartiles (boxe), mean (central 1642 band), minimum and maximum (whiskers). Each dot represents one sample, each patient 1643 being drawn 1-3 times. Statistical analyses used one-way ANOVA with Kenward-Roger 1644 method to take into account the number of specimen/patient: *p<0.05, **p<0.01, ***p<0. 

immunophenotyping and viral parameters. 1666The heatmap shows z score-normalized concentration of parameters. Each column represents 1667 a patient and each row a parameter. The color gradient from blue up to red indicates 1668 increasing gradients of concentrations. In order to assess the immune characteristics after 1669 COVID-19 events in convalescent vs controls, we compared only alive patients and we 1670 excluded all the dead ones in this analysis. 1671 1672 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint this version posted May 5, 2021. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint this version posted May 5, 2021. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint this version posted May 5, 2021. C. D.All rights reserved. No reuse allowed without permission. 3 5 (6) 3 (14) 2 (13) 0 (0) 0 (0) Comorbidities -no(%)°C OPD 6 (7) 2 (9) 1 (7) 1 (4) 2 (10) 

Radiotherapy 8 (7) 2 (6) 3 (14) 1 (3) 2 (9) Surgery 8 (7) 3 (8) 2 (9) 3 (10) 0 (0) Hormonal therapy 11 (10) 4 (11) 0 4 (14) 3 (13) Immunotherapy 12 (11) 4 (11) 1 (5) 4 (14) 3 (13)Others 11 (10) 2 (6) 2 (9) 0 (0) 5