Alotaibi S, Kennedy J, Tellier R, Stephens D, Banwell B. Epstein-Barr Virus in Pediatric Multiple Sclerosis. JAMA. 2004;291(15):1875-1879. doi:10.1001/jama.291.15.1875
Author Affiliations: Department of Pediatrics (Neurology), Al-Sabah Hospital, Shuwaikh, Kuwait (Dr Alotaibi); and Paediatric Laboratory Medicine (Dr Tellier), Department of Paediatrics (Neurology) (Dr Banwell), Brain & Behavior (Ms Kennedy), Population Health Sciences (Mr Stephens), The Hospital for Sick Children, University of Toronto, Toronto, Ontario.
Context Infection with common viruses, particularly Epstein-Barr virus (EBV),
has been postulated to contribute to the pathobiology of multiple sclerosis
(MS). Detailed virological studies in pediatric MS have not been previously
Objective To evaluate whether children with MS are more likely to be seropositive
for EBV or other common viruses than their healthy age-matched peers.
Design, Setting, and Patients Case-control study of viral samples collected from March 1994 to February
2003 from 30 pediatric MS patients, 90 emergency department controls matched
3:1 with the MS patients by year of birth, and 53 healthy control children.
Main Outcome Measures Archived serum samples were analyzed for the presence of IgG antibodies
directed against EBV viral capsid antigens, nuclear antigens, and early antigens,
cytomegalovirus, parvovirus B19, herpes simplex virus, and varicella zoster.
Results Serological evidence for remote EBV infection was present in 83% of
pediatric MS patients compared with 42% of emergency department and healthy
controls (P<.001). Five pediatric MS patients
were negative for all 3 EBV antigens. Pediatric MS patients were less likely
than controls to have been exposed to herpes simplex virus (P = .003), while seropositivity for cytomegalovirus, parvovirus B19,
and varicella zoster did not differ between MS patients and controls.
Conclusion These results suggest an association between EBV infection and pediatric MS.
Multiple sclerosis (MS) is believed to involve a complex interplay between
environmental triggers (such as infections), genetic predisposition, and aberrant
immune cell activation. Epidemiological studies suggest that environmental
exposure to a putative infectious agent must occur during a specific window
of immunological vulnerability in childhood.1
Epstein-Barr Virus (EBV) is of particular interest. Acute symptomatic
infection with EBV (infectious mononucleosis) can be associated with central
nervous system (CNS) demyelination.2 Although
the majority of adult MS patients do not have clinical or serological evidence
of acute mononucleosis at the time of MS diagnosis, nearly 100% demonstrate
serological evidence of remote EBV infection.1,3,4 While
the association of EBV with adult MS is statistically significant, the pathobiological
significance of this observation has been questioned since EBV infects more
than 90% of the healthy adult population of Western societies.5 Infection
with EBV occurs in childhood or adolescence in 50% of individuals6; the remainder contract EBV during early adulthood.
Approximately 5% of all MS patients experience the onset of their disease
prior to age 18 years.7,8 If EBV
infection is involved in the initiation of MS, children with MS should demonstrate
serological evidence of prior EBV exposure at the time of their MS diagnosis,
at an age when the majority of their healthy peers have yet to be exposed
to the virus.
Epstein-Barr virus serological studies were available for 30 of 35 children
with clinically definite MS (defined by 2 separate and well-documented demyelinating
attacks9) enrolled in the Pediatric MS Clinic
at the Hospital for Sick Children (Toronto, Ontario) as of February 2003.
Viral samples were collected from March 1994 to February 2003. Viral serology
was not available for 4 children referred from outside Canada and for 1 child
in whom initial viral results were inconclusive and archived serum was insufficient
Selection of control samples was based on the availability of EBV serological
results and/or archived serum samples stored in the virology department. To
study completely healthy children, we selected samples obtained from bone
marrow transplant (BMT) donors. To control for age, we selected samples from
an emergency department (ED) cohort matched 3:1 for age with each MS patient.
Selection of control samples was performed by searching, using predetermined
search strategies, the Hospital for Sick Children databases for patients entered
between 1993 and the end of 2002. All searches were performed blinded to the
viral serological results. For the BMT controls the following search criteria
were applied: (1) listed as a BMT donor in the database; (2) age between 4
and 18 years; and (3) EBV serology performed. Medical charts were reviewed,
and only those BMT donors documented to be completely healthy were then included.
For the ED cohort, the following search criteria were used: (1) the patient
had been seen in the ED with a presenting diagnosis of rash, pharyngitis,
or abdominal pain, and (2) EBV serology was obtained. Medical charts of the
potential ED controls were reviewed to ensure that the child was documented
to be completely well prior to the acute illness that prompted the ED visit.
Serum samples from all participants were analyzed in the licensed clinical
microbiology laboratory at the Hospital for Sick Children in batches, blinded
to case status. Samples were analyzed using standardized enzyme-linked immunosorbent
assay (ELISA) kits for IgG antibodies directed against EBV capsid (EBV-VCA),
nuclear (EBV-EBNA), and early antigens (EBV-EA) (DiaSorin, Stillwater, Minn).
Archived samples from the MS cohort and ED controls, obtained and stored at
the time of initial EBV sampling, were then retrieved and analyzed for the
presence of IgG antibodies directed against cytomegalovirus (CMV) (Zeus Scientific,
Raritan, NJ), parvovirus B19 (Biotrin International Ltd, Mount Merrion, Co.
Dublin, Ireland), varicella zoster virus (VZV) (Zeus Scientific), and herpes
simplex virus (HSV) (BioChem ImmunoSystems Italia SPA, Casalecchio di Reno,
Italy). Twenty of the control samples originally analyzed for EBV using immunofluorescence
assays were reanalyzed by ELISA to ensure uniform methodology. The HSV ELISA
kit does not discriminate infection with HSV-I from HSV-2. One MS patient
had insufficient serum to analyze for VZV, another insufficient serum for
HSV, and a third patient had no archived serum available. The remaining 27
MS patients and 67 of the ED controls had sufficient archived serum for analysis
of the entire viral panel.
Patients were classified as "remotely infected" if EBV antibodies against
both VCA and EBNA (irrespective of EA) were detected, "recently infected"
if antibodies against VCA and EA (but not EBNA) were detected, and "EBV-naive"
if antibodies against all 3 EBV antigens were absent. Samples were viewed
as uninterpretable if results did not conform to 1 of the 3 possibilities.
Serological test results for CMV, parvovirus B19, VZV, and HSV were recorded
as positive or negative based on interpretive criteria provided by the manufacturer.
Logistic regression analysis was performed comparing remote EBV infection
between MS patients and BMT donors and comparing the number of MS patients
and controls with negative serological results for EBV. Conditional logistic
regression analysis for a matched case-control design was performed comparing
the MS patients with the age-matched ED controls for EBV, CMV, parvovirus
B19, VZV, and HSV. Odds ratios (ORs) and 95% confidence intervals (CIs) were
calculated. Statistical analysis was performed using SAS version 8.2 (SAS
Institute Inc, Cary, NC).
The study was approved by the Research Ethics Board of The Hospital
for Sick Children. Individual consent for analysis of archived specimens was
The mean time from the first MS attack to viral sample acquisition was
1.36 years (Table 1). All MS children
experienced multiple attacks, and although many are now receiving MS-targeted
therapies, none were receiving these treatments at the time of sample acquisition.
None of the MS patients reported a history of symptoms compatible with acute
mononucleosis. The mean age of the MS and matched ED cohorts was similar as
expected (13.40 and 13.37 years, respectively), but the BMT patients were
younger (10.30 years) (Table 2).
There were more girls in the MS and ED cohorts but more boys in the BMT control
As shown in Figure 1, remote
EBV infection was identified in 83% of the MS cohort compared with 42% of
the BMT controls (OR, 7.04; 95% CI, 2.3-21.3; P<.001)
and 42% of the healthy, age-matched ED controls (OR, 8.7; 95% CI, 2.5-30.3; P<.001). Only 17% of the MS patients were seronegative
for EBV, compared with 55% of the BMT donor cohort (OR, 0.17; 95% CI, 0.06-0.5; P<.001) and 36% of the ED controls (OR, 0.27; 95% CI,
0.075-0.987; P = .04). As expected, recent infection
was highest in the ED cohort (22%) in whom serological testing was performed
due to clinically suspected acute EBV infection. Recent infection with EBV
was not found in children with MS, even those sampled at the time of their
first demyelinating episode.
As shown in Figure 2, MS patients
did not differ from controls for the prevalence of antibodies against parvovirus
B19, VZV, or CMV, but were less likely to have been exposed to HSV than the
control cohort (52% vs 88%) (OR, 0.14; 95% CI, 0.04-0.51; P = .003).
Pediatric MS patients are significantly more likely to have experienced
EBV infection than their peers. Our results may be interpreted in several
ways, including the following: (1) infection with EBV initiates or propagates
MS pathogenesis; (2) MS leads to an increased susceptibility to B-cell infection
with EBV; or (3) a common mechanism exists leading to heightened susceptibilities
to early EBV infection and early-onset MS.
The pathogenesis of MS may relate to immune responses to environmental
agents such as viruses encountered during the pediatric-age window of risk.5,10- 14 There
are several features of EBV that make it biologically plausible that it could
play a role in MS. Exposure to EBV results in persistent B-cell infection,
expansion of EBV-transformed B-cell clones, and the production of antibodies
directed against specific EBV viral antigens, as well as lifelong T-cell surveillance
of infected B cells.15 The presence of EBV
antigen–responsive T cells is not inherently pathogenic, as they are
present in a quiescent state in most healthy EBV-positive adults. The pathogenic
potential of these T cells requires activation, possibly through molecular
mimicry. A pentapeptide sequence found in the EBV nuclear antigen shares sequence
homology with an epitope of myelin basic protein, a major component of the
myelin sheath.15 Furthermore, EBV induces B-cell
surface expression of alpha-B crystallin, a protein recently identified as
a major autoantigen constituent that is abnormally expressed in brains of
patients with MS.16 It is thus conceivable
that exposure to EBV may lead to a misdirected host immune response against
self-antigens in the CNS, such as alpha-B crystallin and/or myelin basic protein.
Epstein-Barr virus is not the only virus implicated in MS and clearly
is not a requisite trigger, as evidenced by the 5 EBV-negative pediatric MS
patients. A role for human herpesvirus 6 (HHV-6) has been suggested by studies
of HHV-6 expression in CNS tissue and by the identification of increased HHV-6
antibody titers in serum samples of MS patients.17 Human
herpesvirus 6 variant A infection leads to activation of the EBV genome in
EBV-positive B cells,18 raising the possibility
that multiple viral exposures may act in concert. However, the literature
on HHV-6 is complicated by differences in methodology between studies,19 and by the fact that nearly 100% of the population
is infected with HHV-6 by the age of 2 years. Of greater interest would be
the study of HHV-6 replicative/latency status, which would require molecular
methods such as polymerase chain reaction techniques.20 Such
analyses are planned. Although many viral agents other than EBV, including Chlamydia pneumoniae, have been studied in MS, strong associations
have yet to be documented, owing in part to differences in methodology and
It is possible that the association between EBV infection and MS relates
to increased exposure or susceptibility to EBV infection in MS-affected children,
rather than a causal role for EBV in MS pathogenesis. However, pediatric MS
patients do not seem to have an increased susceptibility or exposure to viruses
in general, as evidenced by the similarity in seropositivity rates between
MS patients and controls for parvovirus B19, CMV, and VZV.
Another important issue is whether EBV infection initiates, rather than
propagates, the immunological processes involved in MS. A study of EBV infection
in 3 million US military personnel demonstrated a strong positive association
between EBV antibodies and MS risk in samples collected more than 5 years
before MS diagnosis.22 Further support for
the role of EBV in the initiation of MS pathogenesis will be sought by studying
EBV serology in children presenting with initial acute CNS demyelination:
children subsequently diagnosed with MS would be expected to show evidence
of remote EBV infection even at the time of their first attack.
An interesting observation in our study was the fact that healthy children
were more likely than pediatric MS patients to be seropositive for HSV. Although
our methodology does not allow discrimination between HSV-1 and HSV-2, most
participants in our study were younger than 16 years and few were likely to
be sexually active. Thus, their positive serological test results most probably
reflect previous infection with HSV-1. It has been hypothesized that HSV-1
immunity is protective against MS.23 If the
sequence of viral exposures in childhood is important in MS pathobiology,
then perhaps pediatric MS patients experience EBV infection without the "protective"
benefit of early HSV-1 infection.
Although demographic data on our control cohorts are limited, it is
unlikely that demographic differences between controls and MS patients would
account for the marked difference in EBV seropositivity. In fact, the demographic
features of our 2 control cohorts were not matched to each other, yet EBV
seroprevalence was identical. Furthermore, the seroprevalence of EBV in our
controls was similar to that of other pediatric control cohorts in North American
studies of EBV.6,24 In addition,
although the sample size of the current study was small, significant results
were obtained. Validation of these results requires a larger, prospective
study of multiple viruses in healthy children and pediatric MS patients.