Chronic Viral Reactivation and Associated Host Immune Response and Clinical Outcomes in Acute COVID-19 and Post-Acute Sequelae of COVID-19

Abstract

Chronic viral infections are ubiquitous in humans, with individuals harboring multiple latent viruses that can reactivate during acute illnesses. Recent studies have suggested that SARS-CoV-2 infection can lead to reactivation of latent viruses such as Epstein-Barr Virus (EBV) and cytomegalovirus (CMV), yet, the extent and impact of viral reactivation in COVID-19 and its effect on the host immune system remain incompletely understood. Here we present a comprehensive multi-omic analysis of viral reactivation of all known chronically infecting viruses in 1,154 hospitalized COVID-19 patients, from the Immunophenotyping Assessment in a COVID-19 Cohort (IMPACC) study, who were followed prospectively for twelve months. We reveal significant reactivation of Herpesviridae, Enteroviridae, and Anelloviridae families during acute stage of COVID-19 (0-40 days post-hospitalization), each exhibiting distinct temporal dynamics. We also show that viral reactivation correlated with COVID-19 severity, demographic characteristics, and clinical outcomes, including mortality. Integration of cytokine profiling, cellular immunophenotyping, metabolomics, transcriptomics, and proteomics demonstrated virus-specific host responses, including elevated pro-inflammatory cytokines (e.g. IL-6, CXCL10, and TNF), increased activated CD4+ and CD8+ T-cells, and upregulation of cellular replication genes, independent of COVID-19 severity and SARS-CoV-2 viral load. Notably, persistent Anelloviridae reactivation during convalescence (≥3 months post-hospitalization) was associated with Post-Acute Sequelae of COVID-19 (PASC) symptoms, particularly physical function and fatigue. Our findings highlight a remarkable prevalence and potential impact of chronic viral reactivation on host responses and clinical outcomes during acute COVID-19 and long term PASC sequelae. Our data provide novel immune, transcriptomic, and metabolomic biomarkers of viral reactivation that may inform novel approaches to prognosticate, prevent, or treat acute COVID-19 and PASC.

Keywords: Anellovirdae; COVID-19; Chronic Virus; Cytomegalovirus; Epstein-Barr Virus; Herpesviruses; Long-COVID; PASC; Simplexvirus.

Figure 1 –. The Transcriptionally Active Human Virome of the Blood, Upper Airway and Lungs in Patients Hospitalized for COVID-19.

A) A total of 1154 participants were recruited across 20 sites for the IMPACC study, with PBMC, nasal, and EA samples analyzed with RNA-sequencing. B) The number of participants with the respective transcriptomic at each timepoint. The background gray shading indicates the percentage of total participants, with samples analyzed at that timepoint for each transcriptomic assay. C) Heatmap of the percent of samples that had detected reads for various latent viruses for each transcriptomic assay at the different timepoints. D) Smoothed curves demonstrating the proportion of total samples that were positive for the five most common viruses in the nasal and PBMC transcriptomics. Curves were calculated by the proportion of samples positive for each day ± two days (a rolling window approach), followed by a local polynomial regression fitting. E) Spearman correlation of viral reads per million across transcriptomic assays, with viruses hierarchically clustered. Correlation in (E) only visualized if adj.p <= 0.05. Source data are provided as a Source Data file.

Figure 2 –. Clinical Outcomes Associated with Activation of the Human Virome in Severe COVID-19.

A) Percent of participants in the cohort who had detectable viral reads in at least one sample within 40 days of hospital admission for the respective virus and tissue split by each trajectory group (on the left), a measure of COVID-19 severity, and participants in TG4 split by their long-term mortality outcome (on the right). B) Percent of participants in the cohort who had detectable viral reads in at least one sample within 40 days of hospital admission for the respective virus and tissue across age groups (split by quantiles). In (A) and (B) adjusted p-values of <0.05, <0.01, <0.001, and <0.0001 are represented by *, **, ***, and **** respectively. C) Results from generalized linear mixed modeling of various complications, comorbidities, and medication usage evaluating for association with viral transcripts detected within 40 days of hospitalization while controlling for sex, age quantile, and TG. Directionality of the association indicated by color, with a positive association increasingly red and a negative association increasingly blue. (Filled dots represent adj.p < 0.05, p-values adjusted using Benjamini-Hochberg procedure). Source data are provided as a Source Data file.

Figure 3 -. Antibody Response, Proteomics, and Circulating Cellular Immunophenotyping Validates Chronic Viral Reactivation in COVID-19.

A) Relative EBV GP350 IgG Antibody Titers in participants with detected EBV transcripts (EBV+) vs participants with no detected transcripts (EBV−). Adjusted p-values calculated using a Wilcoxon rank-sum test with Benjamini-Hochberg corrections. B) Percent of patients seropositive for CMV at admission in participants with no CMV transcripts in the acute period (CMV−) vs participants with detectable transcripts (CMV+). C) Percent of participants with detectable HSV1 proteins in the plasma for participants with HSV1 detectable in the nasal swabs (HSV1+) vs participants with no detectable transcripts (HSV−). P-values in (B) and (C) calculated using a chi-square test of independence. D) Results of linear mixed effect modeling identifying whole blood cell frequencies significantly associated with detection of viral reads for different viruses while controlling for TG, sex, and age. Directionality of the association indicated by color, with a positive association increasingly red and a negative association increasingly blue. (Dot only shown if adj.p <= 0.05.). Source data are provided as a Source Data file.

Figure 4 –. Activation of the Human Virome Correlates with Changes in Inflammatory Cytokine Expression.

A) Summary heatmap of adjusted p-values colored by direction of significance for associations of cytokines with specific viruses in the upper respiratory tract and blood. Adj.p shown if adj.p<0.01 for either the main effect or time interaction term in the model. Boxplot of largest gamm residuals for each participant by virus and longitudinal 95% confidence interval for B) CXCL10, C) CXCL11, D) IL18, E) IL6, F) IL10, and G) IFNG. Source data are provided as a Source Data file.

Figure 5 –. Viral reactivation associated with shifts in the plasma metabolome.

A) Number of significant metabolites for each chronically infecting virus across tissues and the percentage of significant metabolites that map to each major branch of metabolism. B) Dot plot of metabolic sub-pathways containing significantly different metabolites for the different viruses, with dot size indicating the impact ratio (the percent of detected metabolites from that pathway that were significantly different) and the color indicating the average adjusted p-value of significant metabolites in that pathway. Largest magnitude gamm residual for each participant and longitudinal 95% confidence interval for C) urea, D) erucate, and E) 6-bromotryptophan. Source data are provided as a Source Data file.

Figure 6 –. Chronic Viral Reactivation is Associated with Alterations in Host Gene Expression.

A) Top enriched pathways in the PBMC transcriptomics associated with reactivation of the identified latently infecting viruses. B) Top enriched pathways in the nasal transcriptomics associated with reactivation of the identified latently infecting viruses. Results for (A) and (B) calculated using hypergeometric enrichment of pathways from Reactome, separately on the positive and negative differentially expressed genes. Dot only shown for a pathway if adjusted p-value < 0.01. Source data are provided as a Source Data file.

Figure 7 –. Associations between Virome Activation and PASC.

A) Percent of participants with reactivation of chronically infecting viruses across specific compartments (nasal or PBMC) that fall into PASC associated patient reported outcome (PRO) groups. The virus negative group are participants who had no viral reactivation for any virus in the acute period, whose rate of total PASC is indicated by the gray line and serves as the baseline reference. B) Percent of the PRO groups which had viral transcripts detected for Anelloviridae in the PBMCs, or Enteroviridae RNA detected in the upper airway, in any of their convalescent samples. Significance calculated using chi-square test, * indicates adjusted p-value < 0.05. C) Top hypergeometric enriched pathways (determined via lowest adjusted p-values) from DEGs associated with Anelloviridae using convalescent samples reveals a similar signature of Anelloviridae during acute COVID-19 (shown in Figure 6A). Source data are provided as a Source Data file.