Systematic review and meta-analysis of the association between Epstein-Barr virus, multiple sclerosis and other risk factors

Abstract

Background: Epstein-Barr virus (EBV) infection is thought to play a central role in the development of multiple sclerosis (MS). If causal, it represents a target for interventions to reduce MS risk.

Objective: To examine the evidence for interaction between EBV and other risk factors, and explore mechanisms via which EBV infection may influence MS risk.

Methods: Pubmed was searched using the terms 'multiple sclerosis' AND 'Epstein Barr virus', 'multiple sclerosis' AND EBV, 'clinically isolated syndrome' AND 'Epstein Barr virus' and 'clinically isolated syndrome' AND EBV. All abstracts were reviewed for possible inclusion.

Results: A total of 262 full-text papers were reviewed. There was evidence of interaction on the additive scale between anti-EBV antibody titre and HLA genotype (attributable proportion due to interaction (AP) = 0.48, p < 1 × 10^-4). Previous infectious mononucleosis (IM) was associated with increased odds ratio (OR) of MS in HLA-DRB1*1501 positive but not HLA-DRB1*1501 negative persons. Smoking was associated with a greater risk of MS in those with high anti-EBV antibodies (OR = 2.76) but not low anti-EBV antibodies (OR = 1.16). No interaction between EBV and risk factors was found on a multiplicative scale.

Conclusion: EBV appears to interact with at least some established MS risk factors. The mechanism via which EBV influences MS risk remains unknown.

Keywords: Epstein–Barr virus; Multiple sclerosis; clinically isolated syndrome; infectious mononucleosis; meta-analysis; systematic review.

Figure 1.

PRISMA flow charts with details of publications retrieved via searches, abstracts screened, full-text articles assessed and used in analyses.

Figure 2.

(a) Forest plot demonstrating OR_MS for HLA⁺EBNA^hi persons. (b) Forest plot demonstrating OR_MS for HLA⁺EBNA^lo persons. (c) Forest plot demonstrating OR_MS for HLA^–EBNA^hi persons. (d) Bar chart demonstrating evidence of interaction between HLA-DRB1*1501 genotype and EBNA antibody titre on an additive, but not multiplicative scale. The dotted line represents the null (OR = 1). (e)–(h) Forest plots demonstrating estimates of interaction – AP, RERI, synergy index, and multiplicative interaction respectively – for studies with individual-level data available. The reference group (with OR = 1) is HLA^–EBNA^lo individuals for all panels. MIT: multiplicative interaction term.

Figure 3.

(a) Forest plot of studies examining the relationship between previous infectious mononucleosis and MS. (b) Funnel plot demonstrating no clear evidence of publication bias in these studies. (c) Forest plot demonstrating OR_MS for HLA⁺IM⁺ persons. (d) Forest plot demonstrating OR_MS for HLA^–IM⁺ persons. (e) Forest plot demonstrating OR_MS for HLA⁺IM^– persons. (f) Bar chart demonstrating lack of evidence of interaction between HLA-DRB1*1501 genotype and prior IM on an additive, but not multiplicative scale. The dotted line represents the null (OR = 1). (g)–(j) Forest plots demonstrating estimates of interaction – AP, RERI, synergy index, and multiplicative interaction respectively – for studies with individual-level data available. The reference group (with OR = 1) is HLA^–IM^– individuals for all panels. MIT: multiplicative interaction term.

Figure 4.

(a) Forest plot demonstrating OR_MS for EBNA^hiSmoking⁺ persons. (b) Forest plot demonstrating OR_MS for EBNA^loSmoking⁺ persons. (c) Forest plot demonstrating OR_MS for EBNA^hiSmoking^– persons. (d) Bar chart demonstrating lack of evidence of interaction between smoking status and EBNA titre on an additive, but not multiplicative scale. (e)–(h) Forest plots demonstrating estimates of interaction – AP, RERI, synergy index, and multiplicative interaction respectively – for studies with individual-level data available. The reference group (with OR = 1) is EBNA^loSmoking^– individuals for all panels. MIT: multiplicative interaction term.

Figure 5.

(a) Combined forest plot with meta-analysis of EBV seropositivity in children and adults with MS. Odds ratios represent the odds ratio for EBV seropositivity given a diagnosis of MS (i.e. odds of EBV seropositivity among people with MS/odds of EBV seropositivity among controls). (b) Funnel plot demonstrating evidence of publication bias in publications examining EBV seropositivity and MS.