Impact of Pre-Existing Chronic Viral Infection and Reactivation on the Development of Long COVID

Abstract

The presence and reactivation of chronic viral infections such as Epstein-Barr virus (EBV), cytomegalovirus (CMV) and human immunodeficiency virus (HIV) have been proposed as potential contributors to Long COVID (LC), but studies in well-characterized post-acute cohorts of individuals with COVID-19 over a longer time course consistent with current case definitions of LC are limited. In a cohort of 280 adults with prior SARS-CoV-2 infection, we observed that LC symptoms such as fatigue and neurocognitive dysfunction at a median of 4 months following initial diagnosis were independently associated with serological evidence of recent EBV reactivation (early antigen-D [EA-D] IgG positivity) or high nuclear antigen IgG levels, but not with ongoing EBV viremia. Evidence of EBV reactivation (EA-D IgG) was most strongly associated with fatigue (OR 2.12). Underlying HIV infection was also independently associated with neurocognitive LC (OR 2.5). Interestingly, participants who had serologic evidence of prior CMV infection were less likely to develop neurocognitive LC (OR 0.52) and tended to have less severe (>5 symptoms reported) LC (OR 0.44). Overall, these findings suggest differential effects of chronic viral co-infections on the likelihood of developing LC and predicted distinct syndromic patterns. Further assessment during the acute phase of COVID-19 is warranted.

Summary: The authors found that Long COVID symptoms in a post-acute cohort were associated with serological evidence of recent EBV reactivation and pre-existing HIV infection when adjusted for participant factors, sample timing, comorbid conditions and prior hospitalization, whereas underlying CMV infection was associated with a decreased risk of Long COVID.

Figure 1.

Schema of EBV-specific antibody responses during acute infection and hypothetical responses during SARS-CoV-2-related reactivation. EBV viral capsid antigen (VCA) IgM and IgG rise fairly early after acute infection, with VCA IgG levels persisting long-term. EBV nuclear antigen (NA) IgM levels rise more slowly following acute infection, at a time when virus changes from the lytic to latent phase of infection. Early antigen-D (EA-D) IgG responses rise early following acute infection but decay, often to low or undetectable levels over many months. The dashed lines represent potential changes to antibody levels following EBV reactivation secondary to an insult such as acute SARS-CoV-2 infection. EBV EA-D IgG responses and perhaps increases in baseline levels of EBV NA IgG may be observed 3–4 months following reactivation.

Figure 2.

Results from covariate adjusted logistic regression analysis of predictors of Long COVID and long COVID symptoms. Demographic, underlying health conditions, HIV and CMV positivity, and EBV serological results as predictors of participants with any persistent symptom (PASC) or greater than 5 symptoms across organ systems compared with those without PASC are shown in (a). Associations between covariates and fatigue, neurocognitive symptoms (sx), cardiopulmonary symptoms, and gastrointestinal symptoms are shown in (b). N = 258 for all models with exception of LC >5 symptoms (N= 153; participants with one to four LC symptoms excluded from the analyses). Cases with missing values were excluded from the regression models. Dots and bars represent odds ratios and 95% confidence intervals. P values from regression analyses are shown adjusted for all covariates listed in the figure. BMI = body mass index; EBV = Epstein Barr Virus; VCA = viral capsid antigen; NA = nuclear antigen; EA-D = early antigen-D.; LC = long COVID.

Figure 3.

Circulating markers of Inflammation grouped by EBV and CMV antibody result. No significant differences in inflammation marker levels were observed within each antibody group (e.g. EA-D IgG + versus EA-D IgG −) by two-sided Kruskal-Wallis testing with Dunn’s correction for multiple comparison (* P <0.05, ** P <0.01). Bars and lines represent mean and standard deviation (all data points are shown). Units are in pg/mL.