Safety and COVID-19 Symptoms in Individuals Recently Vaccinated with BCG: a Retrospective Cohort Study

Abstract

Bacille Calmette-Guérin (BCG) induces long-term boosting of innate immunity, termed trained immunity, and decreases susceptibility to respiratory tract infections. BCG vaccination trials for reducing SARS-CoV-2 infection are underway, but concerns have been raised regarding the potential harm of strong innate immune responses. To investigate the safety of BCG vaccination, we retrospectively assessed coronavirus disease 2019 (COVID-19) and related symptoms in three cohorts of healthy volunteers who either received BCG in the last 5 years or did not. BCG vaccination is not associated with increased incidence of symptoms during the COVID-19 outbreak in the Netherlands. Our data suggest that BCG vaccination might be associated with a decrease in the incidence of sickness during the COVID-19 pandemic (adjusted odds ratio [AOR] 0.58, p < 0.05), and lower incidence of extreme fatigue. In conclusion, recent BCG vaccination is safe, and large randomized trials are needed to reveal if BCG reduces the incidence and/or severity of SARS-CoV-2 infection.

Keywords: BCG; Bacille Calmette-Guérin; COVID-19; SARS-CoV-2; non-specific effects; off-target effects; trained immunity.

Graphical abstract

Figure 1

Schematic Overview of the Study The dataset consists of 430 adult volunteers of which 266 volunteers received the BCG vaccine in the last 5 years and 164 individuals did not receive the BCG vaccine (control group). Information about disease and any symptoms between February 27^th to April 30^th 2020, as well as other metadata including recent travel history and previous contact with SARS-CoV-2 infected individuals were collected via two digital surveys that were sent to participants of the 300BCG, 500FG, and BCG-booster cohorts. Participants that received a BCG vaccination before 2015 and participants from the BCG booster cohort that received a BCG vaccine between February 27^th and April 30^th were excluded from analysis. Circulating immune parameters were assessed in individuals from the 300BCG and 500FG cohorts. The BCG-induced trained immunity response was assessed in the 300BCG cohort by comparing cytokine levels upon ex vivo stimulation of peripheral blood mononuclear cells (PBMCs) 3 months after BCG vaccination to the levels before vaccination. Annotation: ResponseQ1 = response survey 1, containing information on disease and symptoms between February 27^th to March 31^st 2020; ResponseQ2 = response survey 2, containing information on disease and symptoms between April 1^st to April 30^th 2020. For information on timing of BCG vaccination see Figure S1.

Figure 2

BCG Vaccination Is Not Associated with Increased Disease or Severity of COVID-19-Related Symptoms (A) Percentage of participants with reported positive COVID-19 diagnosis (diagnosis made by PCR/physician based on clinical symptoms) in the BCG-vaccinated (n = 266) and control group (n = 164) (p = Fisher’s exact test p value). (B) Percentage of SARS-CoV-2-related hospital admission in the BCG-vaccinated (n = 266) and control group (n = 164). (C) Percentage of self-reported sickness between February 27^th and April 30^th 2020 in the BCG-vaccinated (n = 266) and control group (n = 164) (p = chi-square test p value; _adjp = logistic regression adjusted p value). (D) The average number of reported days of sickness in the BCG-vaccinated and control group (n = 106, p = Wilcoxon rank-sum test p value). (E) Percentage of individuals that reported any symptom in the BCG-vaccinated (n = 266) and control group (n = 164) (p = chi-square test p value). (F) Percentage of individuals that reported the indicated specific symptom between February 27^th and April 30^th 2020 in the BCG-vaccinated (n = 266) and control group (n = 164) (p = chi-square test p value). See also Figure S2.

Figure 3

Self-Reported Sickness and Symptoms during the SARS-CoV-2 Pandemic Do Not Correlate with Circulating Immune Parameters Circulating concentrations of IL-6, IL-8, IL-10, and IL-18 were assessed before BCG vaccination, normalized, and measured on a log2-scale as normalized protein expression values. Differences between self-reported sickness (A) or symptoms (B) and circulating immune parameters were assessed (n = 392; p = Wilcoxon rank-sum test p value). Boxes represent middle 50% of the data with a center line for the median, the lines extending the boxes roughly represent the 95% confidence interval for comparing medians, outliers are shown as separate dots. See also Figure S4.

Figure 4

Association between BCG-Induced Immune Memory and Self-Reported Sickness and Symptoms during the SARS-CoV-2 Pandemic (A) Inter-individual variability of immune memory responses. Trained immunity was assessed by the fold change in ex vivo S. aureus-induced TNF-α (n = 195) and S. aureus- and M. tuberculosis (Mtb)-induced IL-6 (n = 195 and n = 193, respectively) cytokine production 3 months after BCG vaccination as compared to levels before vaccination. The adaptive immune memory response was assessed by the fold change in Mtb-induced IFN-γ production (n = 189). Fold changes in cytokine production are shown in boxplots, red line indicates the non-responder/responder cut off (1.2 for trained immunity, 1.5 for adaptive immunity). Boxes represent middle 50% of the data with a center line for the median, the lines extending the boxes roughly represent the 95% confidence interval for comparing medians, outliers are shown as separate dots. (B–D) Individuals with a fold change of 1.2 or higher were considered responders, individuals with a fold change smaller than 1.2 were considered non-responders. Proportion of non-responders and responders that reported sickness (B; S. aureus-induced IL-6, non-responder, n = 128; responder, n = 67, Mtb-induced IL-6, non-responder, n = 112; responder, n = 70), S. aureus-induced TNF-α, non-responder, n = 94; responder, n = 101) or any symptoms (C; S. aureus-induced IL-6, non-responder, n = 128; responder, n = 67), Mtb-induced IL-6, non-responder, n = 112; responder, n = 70) are shown (p = chi-square test p value). (D) Percentage of S. aureus-induced IL-6 cytokine fold change non-responders and responders that reported a specific symptom (non-responder, n = 128; responder, n = 67; p = Fisher’s exact test p value). (E–G) Percentage of Mtb-induced IFN-γ cytokine fold change non-responders and responders that reported sickness (E) or symptoms (F) (non-responder, n = 93; responder, n = 96; p = chi-square test p value). (G) Percentage of Mtb-induced IFN-γ cytokine fold change non-responders and responders that reported a specific symptom (non-responder, n = 93; responder, n = 96; p = Fisher’s exact test p value). See also Figures S5 and S6.

Figure 5

Summary of Main Results Analyzed for the First 4 Weeks of the COVID-19 Outbreak in the Netherlands, in which More Individuals Responded to the Survey (A) Percentage of participants with positive COVID-19 diagnosis in the BCG-vaccinated (n = 309) and control group (n = 197) (p = Fisher’s exact test p value). (B) Percentage of self-reported sickness between February 27^th and March 31^st 2020 in the BCG-vaccinated (n = 309) and control group (n = 197) (p = chi-square test p value; _adjp = logistic regression adjusted p value). (C) The average number of reported days of sickness in the BCG-vaccinated and control group (n = 102; p = Wilcoxon rank-sum test p value). (D) Percentage of individuals that reported any symptom in the BCG-vaccinated and control group (n = 506; p = chi-square test p value). (E and F) Trained immunity was assessed by the fold change in ex vivo S. aureus-induced TNF-α (n = 208) and S. aureus- and M. tuberculosis (Mtb)-induced IL-6 (n = 208 and n = 206, respectively) cytokine production 3 months after BCG vaccination as compared to levels before vaccination. Individuals with a fold change of 1.2 or higher were considered responders, individuals with a fold change smaller than 1.2 were considered non-responders. Percentage of non-responders (S. aureus-induced TNF-α, n = 98; S. aureus-induced IL-6, n = 133; Mtb-induced IL-6, n = 119) and responders (S. aureus-induced TNF-α, n = 110; S. aureus-induced IL-6, n = 75; Mtb-induced IL-6, n = 87) that reported sickness (E), or symptoms (F) are shown (p = chi-square test p value).