Figure includes all samples (baseline plus samples during follow-up).
Error bars indicate standard error.
VCA indicates viral capsid antigen; EBNA, Epstein-Barr nuclear antigen.
Epstein-Barr virus titer levels were missing for 1 control.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Levin LI, Munger KL, Rubertone MV, et al. Temporal Relationship Between Elevation of Epstein-Barr Virus Antibody Titers and Initial Onset of Neurological Symptoms in Multiple Sclerosis. JAMA. 2005;293(20):2496–2500. doi:10.1001/jama.293.20.2496
Author Affiliations: Division of Preventive
Medicine, Walter Reed Army Institute of Research, Silver Spring, Md (Dr Levin);
Departments of Nutrition (Ms Munger and Dr Ascherio) and Epidemiology (Drs
Spiegelman and Ascherio), Harvard School of Public Health, Boston, Mass; Channing
Laboratory, Department of Medicine, Brigham and Women’s Hospital and
Harvard Medical School, Boston, Mass (Dr Ascherio); Army Medical Surveillance
Activity, US Army Center for Health Promotion and Preventive Medicine, Washington,
DC (Dr Rubertone); US Army Physical Disability Agency, Washington, DC (Dr
Peck); and Virolab Inc, Berkeley, Calif (Dr Lennette).
Context Infection with Epstein-Barr virus (EBV) has been associated with an
increased risk of multiple sclerosis (MS), but the temporal relationship remains
Objective To determine whether antibodies to EBV are elevated before the onset
Design, Setting, and Participants Nested case-control study conducted among more than 3 million US military
personnel with blood samples collected between 1988 and 2000 and stored in
the Department of Defense Serum Repository. Cases were identified as individuals
granted temporary or permanent disability because of MS. For each case (n = 83),
2 controls matched by age, sex, race/ethnicity, and dates of blood sample
collection were selected. Serial samples collected before the onset of symptoms
were available for 69 matched case-control sets.
Main Outcome Measures Antibodies including IgA against EBV viral capsid antigen (VCA), and
IgG against VCA, nuclear antigens (EBNA complex, EBNA-1, and EBNA-2), diffuse
and restricted early antigens, and cytomegalovirus.
Results The average time between blood collection and MS onset was 4 years (range,
<1-11 years). The strongest predictors of MS were serum levels of IgG antibodies
to EBNA complex or EBNA-1. Among individuals who developed MS, serum antibody
titers to EBNA complex were similar to those of controls before the age of
20 years (geometric mean titers: cases = 245, controls = 265),
but 2- to 3-fold higher at age 25 years and older (cases = 684,
controls = 282; P<.001). The risk of
MS increased with these antibody titers; the relative risk (RR) in persons
with EBNA complex titers of at least 1280 compared with those with titers
less than 80 was 9.4 (95% confidence interval [CI], 2.5-35.4; P for trend <.001). In longitudinal analyses, a 4-fold increase
in anti-EBNA complex or anti–EBNA-1 titers during the follow-up was
associated with a 3-fold increase in MS risk (EBNA complex: RR , 3.0;
95% CI, 1.3-6.5; EBNA-1: RR, 3.0; 95% CI, 1.2-7.3). No association was
found between cytomegalovirus antibodies and MS.
Conclusion These results suggest an age-dependent relationship between EBV infection
and development of MS.
Elevations of levels of serum antibodies to Epstein-Barr virus (EBV)
occurring several years before diagnosis have been characteristically found
in diseases probably caused by EBV, such as Burkitt lymphoma1 and
nasopharyngeal carcinoma,2 and in Hodgkin disease.3 Anti-EBV antibodies are elevated in individuals with
multiple sclerosis (MS),4,5 and
a premorbid increase has been reported in 2 studies,6,7 but
both relied on a single blood sample from each study participant. We therefore
conducted a larger prospective investigation using serial blood samples collected
several years before onset of MS.
The source population for the current study is more than 3 million US
military personnel whose blood samples are stored at –30°C in the
Department of Defense Serum Repository.8 This
repository contains more than 30 million serum specimens from active-duty
and reserve personnel of the US military collected at entry and, on average,
every 2 years thereafter since 1988. The research protocol was approved by
the institutional review boards of Harvard School of Public Health and Walter
Reed Army Institute of Research, which waived the need for informed consent
to use archived blood products or medical records.
Cases were identified by searching the computerized database of the
US Army Physical Disability Agency for active-duty personnel granted temporary
or permanent disability because of MS and by reviewing medical records. We
classified cases as “confirmed MS” if there was a history of 2
or more attacks (occurrence of symptoms of neurological dysfunction lasting
more than 24 hours), a diagnosis of MS made by a neurologist, and a positive
magnetic resonance imaging (MRI) result or if the final diagnosis in the record
was specified as definite MS, clinical definite MS, or laboratory-supported
definite MS.9 Cases were classified as “probable
MS” if they did not meet the criteria for confirmed MS but had at least
2 of the following: history of 2 or more attacks, positive MRI result, and
diagnosis of MS made by a neurologist. These criteria (confirmed or probable
MS) were met by 118 cases, 83 of whom had at least 1 serum sample collected
before onset of MS symptoms (defined as the earliest neurological symptom
ever reported) and were included in the study. For each of these 83 cases,
we identified the earliest available serum sample (baseline sample) plus up
to 2 additional samples collected before onset of MS and the first sample
collected after onset of MS. For each of the 83 cases, we randomly selected
2 controls matched on age (±1 year), sex, race/ethnicity (white, black,
Hispanic, or other), and dates of blood collection (±30 days). For
serial samples, each blood sampling date was matched to within 30 days. Serial
serum samples before MS onset were obtained for 69 case-control sets, including
40 with 2 samples and 29 with 3 samples.
Race/ethnicity was provided by the Army Medical Surveillance Activity,
based on categories defined by the Department of Defense, independently from
the investigators. We included this variable as a matching factor because
of its association with risk of MS, and possible relationship with age at
infection and antibody response.
Serum samples from MS cases and controls were sent to the laboratory
in triplets containing the case and the 2 matched controls in random order
without identification of case-control status. Immunoglobulin G and IgA antibodies
to EBV viral capsid antigen (VCA) and anti–early antigen complex (diffuse
[EA-D] and restricted [EA-R]) were determined by indirect immunofluorescence10,11; IgG antibodies against the EBV nuclear
antigen (EBNA) complex and 2 of its individual members, EBNA-1 and EBNA-2,
were determined by anticomplement immunofluorescence.12 Immunoglobulin
G antibody titers against cytomegalovirus (CMV) were also determined to assess
the specificity of any association that may be found between MS and EBV serology.13
Geometric mean antibody titers (reciprocal of the dilution) in serum
samples collected at baseline were compared between cases and controls using
generalized linear models.14 Conditional logistic
regression was used to estimate the relative risk (RR) of MS associated with
mean serum levels of specific antibody titers. To reduce the within-person
random variation, in these analyses we used for each MS case the geometric
mean antibody titer from all the available serum samples collected before
MS onset, and for each control the geometric mean of the corresponding matched
samples. To explore dose-response relationships, in these conditional logistic
regression models the antibody titers were initially treated as categorical
variables, with each doubling of titers (eg, 20, 40, 80) as a separate category.
However, the lowest and highest categories had to be collapsed in some analyses
because of small numbers.
To take advantage of the longitudinal design of the study, we further
examined whether an increase in antibody titers within person during the follow-up
was associated with an increased risk of MS. For each antibody, we conducted
a conditional logistic regression analysis restricted to the 69 case-control
sets with more than 1 serum sample available. In these models, we used an
indicator (with value 0 or 1) for a 4-fold or greater increase in titers during
the follow-up as the independent variable and case status as the dependent
variable. Because age is strongly related to risk of MS and to exposure to
EBV, we further examined whether the relationship between anti-EBV antibody
and MS was modified by age at blood collection, by conducting stratified analyses
and by adding an interaction term (equal to the product between antibody titers
and an indicator variable for age 20 years or younger vs 21 years or older)
to the conditional logistic regression models. All P values
are 2-tailed and significant at P<.05. We used
SAS version 8.2 (SAS Institute Inc, Cary, NC) for all analyses.
Baseline characteristics of cases and controls are shown in Table 1. For cases, the mean (SD) age at MS onset
was 27 (5.5) years (range, 18-41 years). The diagnosis of MS was confirmed
in 53 (64%) and probable in 30 (36%). Mean (SD) time between baseline blood
collection and MS onset was 4.0 (2.4) years (range, <1-11 years). Three
of 83 cases and 7 of 166 controls were EBV negative (VCA IgG <1:20) at
baseline; the 3 seronegative cases converted before MS onset. The baseline
geometric mean serum antibody titers to VCA (IgG), EBNA complex, EBNA-1, and
EA-D were significantly higher among EBV-positive individuals who later developed
MS than among their matched controls, whereas there were no significant differences
in antibodies to other EBV antigens or CMV (Table
2). Similar results were observed in analyses restricted to cases
with blood samples collected at least 5 years before the onset of MS (Table 2).
Because the incidence of MS increases sharply between the ages of 20
and 30 years, we examined whether the serum titers of antibodies to EBV changed
with age. Among individuals who developed MS, but not among controls, we observed
a sharp and significant increase in mean serum titers of antibodies to EBNA
complex in early adulthood followed by a plateau (Figure 1). Titers to EBNA complex of cases were similar to those
of controls younger than 20 years, but 2- to 3-fold higher at age 25 years
or older (Figure 1). The difference
in geometric mean titers between cases and controls at age 25 years or older
was highly significant (P<.001, using a generalized
linear model). Results were similar for antibodies to EBNA-1. Modest increases
with age were also seen for mean antibody titers to EBNA-2 and EA-R, but not
VCA IgG, EA-D, or CMV (data not shown).
To examine whether this increase in antibody titers with age was explained
by a shorter interval between blood collection and MS onset, we conducted
a regression analysis among MS cases using antibodies to EBNA complex or EBNA-1
as the dependent variable, and, simultaneously, age at blood collection and
the time interval between blood collection and MS onset as the independent
variables. In this regression model, age at blood collection was significantly
and positively associated with mean titers of antibodies to EBNA complex (P = .04) and EBNA-1 (P = .007),
whereas there was no relationship between these antibody titers and the time
interval between blood collection and MS onset.
The risk of MS increased with increasing serum levels of antibodies
to EBNA complex and less strongly to VCA IgG. Compared with individuals with
the lowest antibody titers for EBNA complex (<40) and VCA (<160), the
RR was 35.9 (95% confidence interval [CI], 4.0-322; P for
trend <.001) for individuals in the highest category of EBNA complex and
8.7 (95% CI, 0.93-82; P for trend = .009)
for individuals in the highest category of VCA. To obtain more stable RR estimates,
we repeated the analyses using as the reference category titers less than
320 for VCA, and titers less than 80 for EBNA complex (Figure 2). Positive associations were also found with EBNA-1 (P for trend = .003) and EA-D (P for trend = .05), whereas no significant associations were
found for VCA IgA, EBNA-2, EA-R, and CMV (data not shown).
In within-person analyses, a 4-fold increase in EBNA complex titers
between the sample collected at baseline (typically at time of entry into
the Army) and a subsequent serum sample was associated with a 3-fold increase
in risk of developing MS (RR, 3.0; 95% CI , 1.3-6.5; P = .007); this association was stronger among individuals
with the first blood sample collected at or before age 20 years (RR, 18;
95% CI, 2.2-138; P = .006). Similar results
were obtained for EBNA-1, whereas no significant overall associations were
found for other EBV antibodies or antibodies to CMV (Table 3).
These results confirm those obtained in a smaller study of women with
MS.6 Although the date of onset of MS is difficult
to establish accurately, and many patients at the time of clinical onset have
multiple silent MRI lesions,15 the observation
that anti-EBV antibody titers among cases compared with controls were already
significantly elevated 5 or more years before the onset of MS suggests that
the increased antibody response to EBV is not a consequence of MS, but rather
may be an early event in the pathological process that leads to demyelination
and clinical disease. In particular, elevated risks were found for EBNA complex
and EBNA-1. A significant increase in anti–EBNA-1 titers 5 or more years
before the onset of MS was also found in a study in Sweden,7 although
in that study an opposite association was reported for anti-VCA titers; the
reason for this difference is unclear.
The pattern of antibody response that we observed among individuals
who developed MS is different from the pattern observed in immunocompromised
hosts or in chronic infectious mononucleosis, in which there are elevated
anti–EBNA-2 and reduced anti–EBNA-1 titers,12 and
from that observed in Burkitt lymphoma, where there are prediagnostic elevations
of anti-VCA but not anti-EBNA antibodies.1 Rather,
the elevation of titers to EBNA complex and EBNA-1 suggests a more severe
or more recent primary infection or reactivation of infection accompanied
by a vigorous cellular immune response.16-18 Anti-VCA
and anti-EBNA IgG elevation in prediagnostic serum samples has been associated
with risk of Hodgkin disease3 and nasopharyngeal
carcinoma,19 but in the latter, the strongest
predictors of risk are IgA antibodies to VCA.20
The age-related increase in serum titers of anti-EBNA and anti–EBNA-1
antibodies among individuals with MS was a striking and unexpected finding.
The fact that this increase occurred between the late teens and the mid to
late 20s, independently from the age of MS onset, supports the hypothesis
of an age of vulnerability for the acquisition of MS.21 The
incidence of infectious mononucleosis peaks at this age, but since most participants
in our study were already EBNA-1 seropositive at the time of first blood collection,
a newly acquired EBV infection is an unlikely cause of this antibody response.
More likely, the antibody response is due to either infection with a separate
microorganism or other factors that alter the immune response to EBV22 or, more speculatively, to infection with a strain
of EBV different from that originally carried by the host. There is increasing
evidence that coinfection with multiple EBV strains, either acquired sequentially
or simultaneously, is common even in healthy individuals,23 but
little is known about serological response or other consequences of coinfection.
It has been suggested that multiple infections in early childhood may
reduce the risk of MS by downregulating autoimmune responses that could be
triggered by infection with the same or related microbes later in life.24,25 This hypothesis, often called the
“hygiene hypothesis,” has also been invoked to explain more generally
a positive relationship between incidence of autoimmune and allergic diseases
and level of sanitation.26 A confirmed prediction
of this hypothesis is an increased risk of MS among individuals with a history
of infectious mononucleosis, which is a strong marker of late age for EBV
infection.27 A key question, however, is whether
there is a specific role for late infection with EBV in triggering MS. If
so, the hygiene hypothesis would predict a low MS risk among EBV-uninfected
individuals. In contrast, in the absence of a specific role of EBV, the lack
of anti-EBV antibodies would only be relevant as a marker of low exposure
to infection in childhood,28 and EBV-uninfected
individuals would be predicted to have a high MS risk. Consistent with the
first formulation, the risk of MS among EBV-seronegative individuals is several
fold lower than among EBV-positive individuals.29 Innate
resistance to both EBV and MS could be invoked to explain this association,
but this explanation is virtually ruled out by the recent finding of an 8-fold
higher risk of MS among children infected with EBV than among those not infected.30 This strong association also provides evidence to
counter the explanation that the increase in anti-EBNA titers in our study
is a consequence of a change in the immune system occurring years before the
clinical onset of MS.
Overall, the results of our investigation therefore support a specific
role of EBV as a risk factor for MS. Because of the central role of EBV infection
in the hygiene hypothesis, we suggest that it should be called the “EBV
hypothesis” or the “EBV variant of the hygiene hypothesis”
of MS, to differentiate it from the more general version of the hygiene hypothesis
that refers to asthma and other immune conditions, but does not seem to include
MS. Similar epidemiological evidence relates EBV infection to systemic lupus
erythematosus,31 suggesting that EBV may be
a risk factor for autoimmune diseases.
Corresponding Author: Alberto Ascherio,
MD, DrPH, Harvard School of Public Health, Nutrition Department, 665 Huntington
Ave, Boston, MA 02115 (Alberto.Ascherio@channing.harvard.edu).
Author Contributions: As principal investigator,
Dr Ascherio had full access to all of the data in the study and takes responsibility
for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Levin, Rubertone,
Acquisition of data: Levin, Munger, Rubertone,
Peck, Lennette, Ascherio.
Analysis and interpretation of data: Levin,
Munger, Peck, Spiegelman, Ascherio.
Drafting of the manuscript: Levin, Ascherio.
Critical revision of the manuscript for important
intellectual content: Levin, Munger, Rubertone, Peck, Lennette, Spiegelman,
Statistical analysis: Munger, Peck, Spiegelman,
Obtained funding: Levin, Lennette, Ascherio.
Administrative, technical, or material support:
Levin, Peck, Lennette, Ascherio.
Study supervision: Ascherio.
Financial Disclosures: None reported.
Funding/Support: This study was supported by
grant NS42194 from the National Institute of Neurological Disorders and Stroke.
Preliminary work was supported by a pilot grant from the National Multiple
Role of the Sponsor: The funding organizations
did not have a role in the design and conduct of the study; in the collection,
management, analysis, and interpretation of the data; or in the preparation,
review, or approval of the manuscript.
Disclaimer: The views expressed are those of
the authors and should not be construed to represent the positions of the
Department of the Army or Department of Defense.
Acknowledgment: We thank Walter Willett, MD,
DrPH, and Nancy Mueller, DSc, for their expert advice, and Eilis O’Reilly,
MSc, and Elsa Jiménez for technical assistance.