Download A Study of Risk Factors for Acquisition of Epstein

MAJOR ARTICLE
A Study of Risk Factors for Acquisition
of Epstein-Barr Virus and Its Subtypes
Craig D. Higgins,1 Anthony J. Swerdlow,1 Karen F. Macsween,2 Nadine Harrison,3,a Hilary Williams,2,a
Karen McAulay,2 Ranjit Thomas,2 Stuart Reid,2 Margaret Conacher,2 Kathryn Britton,2,a and Dorothy H. Crawford2
1
Section of Epidemiology, Institute of Cancer Research, Surrey, and 2Clinical and Basic Virology, University of Edinburgh, and 3University Health
Centre, University of Edinburgh, Edinburgh, United Kingdom
(See the editorial commentary by Pagano, on pages 469–70.)
Background. Risk factors for primary infection with Epstein-Barr virus (EBV) and its subtypes have not been
fully investigated.
Methods. Questionnaires and serum samples from a total of 2006 students who entered Edinburgh University
in 1999–2000 were analyzed to examine risk factors for EBV seropositivity, both overall and by EBV type.
Results. The prevalence of EBV seropositivity was significantly increased among females, older students, those
who had lived in tropical countries, those with siblings, and those who were sexually active, particularly if they
had had numerous sex partners. Risk was lower (1) among students who always used a condom than among those
who had sexual intercourse without one and (2) among female oral-contraceptive users than among sexually active
nonusers. Risk factors for type 1 EBV infection were similar to those for EBV overall. No associations were found
between nonsexual risk factors and type 2 infection. Sexual activity increased the risk of type 2 infection, but the
increase in risk with number of sex partners was less consistent than for type 1 infections. Dual infection was
uncommon, but the patterns of risk appeared to be similar to those of type 1 infection.
Conclusion. This study provides further evidence that EBV may be sexually transmitted and some suggestion
that the risk factors for type 1 and type 2 infection differ.
Epstein-Barr virus (EBV) is a tumorigenic herpes virus
that causes infectious mononucleosis (IM) and is associated with a variety of human tumors of both lymphoid (Burkitt lymphoma, Hodgkin disease, and B
lymphoproliferative disease) and epithelial cell (nasopharyngeal carcinoma) origin. Primary EBV infection
usually occurs subclinically during childhood, and
Received 2 June 2006; accepted 25 August 2006; electronically published 10
January 2007.
Potential conflicts of interest: none reported.
Presented in part: British Association of Health Service in Higher Education
meeting, London, 9–11 July 2001; Congress of the Institute of Biomedical Science,
Birmingham, 25–27 September 2001.
Financial support: UK Medical Research Council (MRC; grant G9826804 to
D.H.C., A.J.S., and N.H.); MRC Clinical Training Fellowship (grant G84/5443 to
H.W.).
a
Present affiliations: Primary Care Division, Scottish Executive Health Department, Edinburgh, United Kingdom (N.H.); Bristol Haematology and Oncology
Centre, Bristol, United Kingdom (H.W.); Sir Alastair Currie Cancer Research UK
Laboratories, Molecular Medicine Centre, University of Edinburgh, Western General
Hospital, Edinburgh, United Kingdom (K.B.).
Reprints or correspondence: Mr. Craig Higgins, Section of Epidemiology, Sir
Richard Doll Bldg., Institute of Cancer Research, Cotswold Rd., Sutton, Surrey,
SM2 5NG, UK ([email protected]).
The Journal of Infectious Diseases 2007; 195:474–82
2007 by the Infectious Diseases Society of America. All rights reserved.
0022-1899/2007/19504-0004$15.00
474 • JID 2007:195 (15 February) • Higgins et al.
thereafter the virus establishes a latent infection of B
lymphocytes that persists for life [1]. In nonindustrialized countries, almost all children over the age of 2
years are infected with EBV [2], but, in areas of the
world where standards of living are high, many escape
childhood infection. These individuals are then susceptible to primary EBV infection later in life when IM
may occur [3], most commonly at ages 15–25 years in
high socioeconomic groups [4]. Although usually selflimiting, IM may cause prolonged illness and fatigue and
appears to be a risk factor for Hodgkin disease [5, 6].
Although 2 childhood peaks of EBV seroconversion
are known to occur—at ages 2–4 and 15 years [7, 8]—
only a few studies have attempted to identify the risk
factors that are responsible. Indicators of high socioeconomic status have been found to be associated with
increased likelihood of persistent EBV seronegativity
into the teenage years [9–11], but inconsistent results
have been found for other risk factors. EBV is produced
in the saliva of persistently infected individuals, and
transmission almost certainly occurs orally through
close salivary contact during kissing, but the virus has
also been reported in both male and female genital
secretions, suggesting that sexual transmission may occur [12–
14]. In a recent seroepidemiological study of 11000 students
entering Edinburgh University, sexual intercourse was identified as a significant risk factor for both EBV seropositivity and
IM [15].
Two distinct types of EBV (1 and 2, also called A and B)
that show 70%–85% sequence homology were identified in the
1980s [16, 17], but no epidemiological studies have been conducted examining whether they share risk factors. Several small
studies suggest that type 1 EBV is more prevalent than type 2
and that type 2 is more common in Africa than elsewhere [18].
Most of these studies analyzed EBV types in tumor material or
tumor-derived cell lines, and there have been few studies in
healthy populations. One study found type 2 EBV to be more
common in homosexual than heterosexual men [19], and dual
infection with types 1 and 2 has mainly been identified in
conjunction with HIV infection [20–22].
There is no specific treatment for IM, but a vaccine against
EBV-specific gp350 aimed at preventing it is in clinical trials
[23]. A vaccination strategy has yet to be formulated, however.
This requires the identification of susceptible groups and risk
factors for EBV seroconversion and IM.
We conducted a study of 12000 university entrants, collecting
questionnaires and serum samples to identify sexual and nonsexual risk factors for EBV seroconversion overall and, for the
first time, separately by EBV type.
SUBJECTS, MATERIALS, AND METHODS
Study population. All students enrolling at the Edinburgh
University Health Centre in October 1999 and October 2000
for courses lasting 4 or more years were invited to participate
in the study. Those agreeing were asked to sign a consent form
and complete a questionnaire that asked about demographic,
behavioral, medical, and sexual characteristics before coming
to the university. Participants were also asked to provide a blood
sample to be tested for IgG antibodies in plasma and EBV type,
by polymerase chain reaction (PCR) of peripheral-blood mononuclear cells (PBMCs). The study was approved by the Lothian
Ethics Committee.
EBV serological analysis. Serum or plasma samples were
initially tested for EBV-specific IgG antibodies to EB viral capsid
antigen (VCA) by routine ELISA (Sigma). Negative results were
checked by indirect immunofluorescence (IF) [24]. Equivocal
serum samples were also tested for IgG antibodies to EB viral
nuclear antigens (EBNA) by anticomplimentary IF [25]. Seronegativity was defined as VCA IgG negative (by ELISA and
IF) and EBNA negative (where tested).
Type 1 and 2 detection. DNA was extracted from PBMCs
by use of the Easy DNA Kit (Invitrogen) as per the manufacturer’s instructions. The type of EBV (1 or 2) in each PBMC
sample was determined by nested PCR amplification across a
type-specific region of the EBNA 3C gene [26].
Statistical methods. Statistical analyses were done to identify associations between possible risk factors and the seroprevalence of EBV antibodies overall and separately for type 1,
type 2, and dual infections. Prevalence ratios (PRs) rather than
odds ratios were used as a measure of the relative risk, because
of the high level of EBV positivity among the students and,
hence, violation of the rare-disease assumption [27]. A multivariate Poisson model with scale parameter adjusted by x2
was used to obtain PRs and 95% confidence intervals (CIs)
adjusted for confounding factors [28]. Analyses were conducted
using the Stata statistical package (version 8.0; StataCorp) [29].
RESULTS
Approximately 4000 students registered with the Edinburgh
University Health Centre during 1999–2000, some three-quarters of whom were approached to take part in the study. The
remainder were not approached because of the chaotic nature
of the clinic during the registration procedure, which processed
thousands of new students during the first few days of term.
A total of 2006 students (63% female; average age, 18 years 8
months) were recruited into the study, of whom 1496 (75%)
tested positive for EBV-specific IgG antibodies (table 1). Seropositivity was significantly higher (P ! .001) among female
(79%) than among male (68%) students, among those 19 years
and older (80%) than among those younger (71%), and among
those who had ever lived in a tropical country (81%) than
among those who had not (73%). Analyses of the specific continents in which each student had lived showed little variation
in seroprevalence across continents. Greater seroprevalence was
found among students who were born in a tropical country
than in those who had simply lived in one, with the greatest
seroprevalence being among those born in Africa (94%), followed by South America (85%), other developing countries
(83%), and Southeast Asia (79%).
Analyses of ethnicity were only possible for white versus
nonwhite students, because of limited numbers in the latter
group, and showed no evidence of a difference in seroprevalence. Seropositivity was significantly greater among students
who had at least 1 sibling (75%) than among those with none
(66%), although risk was not related to the number of siblings
(Ptrend p .40), sex of the sibling, or the age(s) of the sibling(s)
relative to their proband.
There was no evidence that the average degree of crowding
(number of rooms per person) in the students’ home from
birth to starting university was associated with EBV risk. Risk
was also unrelated to the number of different vaccinations the
student had ever received or to any particular allergy (asthma,
eczema, hay fever, food, or other) or childhood infectious disease common in the United Kingdom (e.g., measles and chickRisk Factors for EBV and Its Subtypes • JID 2007:195 (15 February) • 475
Table 1.
Risk of Epstein-Barr virus (EBV) seropositivity, by demographic and other characteristics.
Risk factor
Adjusteda
Crude
No. (%)
EBV negative
EBV positive
PR (95% CI)
Ptrend
PR (95% CI)
239 (32.0)
271 (21.5)
509 (68.1)
987 (78.5)
1.0
1.15b (1.09–1.22)
1.0
1.14b (1.08–1.21)
344 (29.0)
166 (20.3)
844 (71.0)
652 (79.7)
1.0
1.12b (1.07–1.18)
1.0
39 (34.2)
469 (24.8)
75 (65.8)
1419 (75.2)
1.0
1.14e (1.00–1.31)
1.0
1.14e (1.01–1.29)
442 (26.9)
68 (18.6)
1199 (73.1)
297 (81.4)
1.0
1.11b (1.05–1.18)
1.0
Ptrend
Sex
Male
Female
Age starting Edinburgh University
!19 years
⭓19 years
c
1.07 (1.02–1.18)
d
Siblings
No
Yes
Ever lived in tropical country
No
Yes
Degree of crowding in house
1.11c (1.04–1.18)
d,f
Less than median
250 (24.9)
753 (75.1)
1.0
1.0
Median or greater
Total no. of different vaccinations since birth
None
1–4
⭓5
237 (25.5)
694 (74.5)
0.99 (0.94–1.05)
0.99 (0.94–1.04)
8 (14.6)
101 (26.2)
401 (25.6)
47 (85.5)
284 (73.8)
1165 (74.4)
1.0
0.86 (0.74–1.01)
0.87 (0.75–1.01)
All
510 (25.4)
1496 (74.6)
.36
.41
1.0
0.87 (0.75–1.02)
0.88 (0.76–1.02)
NOTE. CI, confidence interval; PR, prevalence ratio.
a
b
c
d
e
f
Adjusted for ever sexually active (yes/no).
P ! .001.
P ! .01.
Data missing for siblings (n p 4) and crowding in house (n p 72).
P ! .05.
No. of rooms per person.
enpox). There was also no relation to risk for anthropological
measurements (height and weight), behavioral factors (e.g., alcohol consumption and cigarette smoking), and schooling
(number of schools attended and whether boarding).
A multivariate model that included the significant factors
shown in table 1 (sex, age, siblings, and residence in a tropical
country) showed a significant independent effect of each. Analyses separately for those students who had been sexually active
before coming to the university and those who had not suggested that the effects of these significant factors were not modified by prior sexual behavior.
Sexual activity before coming to the university was significantly associated with increased risk of EBV seropositivity
(P ! .001), and risk increased significantly (P ! .001 ) with the
number of sex partners (table 2). There was no evidence that
these increased risks were influenced by other sexual or nonsexual factors (e.g., smoking and oral contraceptive use). Risk
was lower among students who reported always using a condom
than among those who reported intercourse without one
(P p .02), a result that was borderline significant (P p .05)
after adjustment for number of sex partners. There was no
clear evidence that risk depended on the sex of the partner,
476 • JID 2007:195 (15 February) • Higgins et al.
although seroprevalence was nonsignificantly lower among homosexual students than among other sexually active students.
We were unable to examine the effect of oral genital sex because
we did not specifically ask about this in the questionnaire.
Analysis of the factors in table 2 separately for males and
females produced no decisive evidence that their effects were
sex dependent, although the protective effect of condom use
was apparent only in males (P ! .05 ), and the difference in effect
between males and females for this variable was borderline
significant (P p .06). EBV seroprevalence was significantly (P
! .001) lower among current oral contraceptive users (60%)
than among sexually active nonusers (91%), a result that remained significant after adjustment for other factors (e.g., number of sex partners, condom use, and age).
Of the 1496 EBV-positive serum samples from students, successful typing of EBV type 1, type 2, or dual infection was
accomplished for 617 (41%); the remaining 879 were tested,
but typing was not successful. There was no age difference
between those successfully typed and those not successfully
typed, but there was some suggestion (P p .05 ) that typing was
more successful among females (43%) than among males
(38%). Among those successfully typed, type 1 infection was
Table 2. Risk of Epstein-Barr virus (EBV) seropositivity in relation to various sexual
behavioral characteristics.
No. (%)
Risk factor
EBV negative
EBV positive
PR (95% CI)
315 (36.5)
192 (17.0)
547 (63.5)
941 (83.1)
1.0
1.31c (1.24–1.38)
315 (36.5)
547 (63.5)
1.0
96 (22.8)
70 (15.0)
326 (77.3)
397 (85.0)
23 (9.8)
212 (90.2)
1.22c (1.14–1.30)
1.34c (1.26–1.43)
c
1.42 (1.31–1.54)
315 (36.5)
547 (63.5)
1.0
Heterosexual
172 (16.3)
883 (83.7)
1.32c (1.25–1.30)
Homosexual
13 (25.5)
38 (74.5)
1.17 (0.99–1.39)
3 (17.7)
14 (82.4)
1.30 (0.99–1.70)
315 (36.5)
547 (63.5)
1.0
56 (13.2)
369 (86.8)
1.37 (1.28–1.46)
111 (18.6)
487 (81.4)
1.28 (1.21–1.37)
Sexually active
No
Yes
No. of sex partnersb
0
1
2–4
⭓5
Bisexual
Condom useb
No sexual activity
Never or seldomd
Always
a
.001
b
Sex of partner
No sexual activity
d
P
b
.27
.02
c
c
NOTE. CI, confidence interval; PR, prevalence ratio.
a
The P value for no. of sex partners is for trend; those for other risk factors are for heterogeneity.
Tests exclude the non–sexually active group.
b
Data missing for sexually active (n p 11 ), no. of sex partners (n p 20 ), sex of partner (n p 21), and
condom use (n p 32).
c
P ! .001.
d
Analyses restricted to heterosexual students only.
detected in 78% of students, type 2 in 17%, and dual infection
in 5%.
As with EBV seropositivity overall, the prevalence of type 1
infection was significantly (P ! .001 ) greater among female
(48%) than among male (33%) students and among students
with siblings (44%) than among those without (28%) (P !
.05), even after adjustment for sexual activity (table 3). Type 1
infection was also greater among older students and those who
had lived in a tropical country, although both of these results
were borderline significant after adjustment for sexual activity.
As with EBV risk overall, these significant findings were not
modified by prior sexual behavior.
In contrast, based on smaller numbers, type 2 risk was similar
in males and females and was reduced, although not significantly so, among students with siblings (table 3). Dual-infection
risks broadly mirrored those for type 1 infection but were generally less significant because of the smaller numbers.
The effect of sexual behavior on the risks of type 1, type 2,
and dual infection are shown in table 4. The risk of type 1
infection was significantly (P ! .001 ) increased among students
who had been sexually active prior to coming to the university,
with the risk increasing with the number of sex partners
(Ptrend p .006), and was lower among those who had homosexual compared with other sexual relationships. The latter
result remained significant after adjustment for number of sex
partners. The protective effect of condom use on type 1 infection was not significant. Analyses of type 1 risks separately for
males and females showed that the increase in risk with the
number of sex partners was more consistent in females
(Ptrend p .02) than males (Ptrend p .08). The lower risk of type
1 infection among homosexual students than among other sexually active students was apparent, although not significant, for
both sexes. The protective effect of condom use on type 1
infection was apparent among males (PR, 0.83) but not materially among females (PR, 0.96), although neither of the 2
PRs, or the difference between them, was significant. As with
EBV risk overall, current oral contraceptive use was significantly
protective (P ! .01) against type 1 infection, even after adjustment for other sexual and nonsexual factors.
Although associated with sexual activity, the risk of type 2
infection did not appear to be lower among homosexual students than among other sexually active students or to increase
consistently with the number of sex partners. The lower risk
of type 2 infection among those who always used a condom
than among those who had sexual intercourse without one was
not significant. Sex-specific analyses of these factors for type 2
infection were inconclusive because of small numbers.
The PRs of dual infection associated with sexual activity were
Risk Factors for EBV and Its Subtypes • JID 2007:195 (15 February) • 477
Table 3.
Risk of Epstein-Barr virus (EBV) type 1 infection, type 2 infection, and dual infection, by demographic and other characteristics.
EBV seropositive
EBV
negative,
no. (%)
No. (%)
Male
239 (54.5)
143 (33.2)
1.0
Female
271 (38.9)
336 (48.3)
1.45 (1.25–1.69)
!19 years
344 (49.7)
273 (39.5)
1.0
61 (8.8)
1.0
14 (2.0)
1.0
⭓19 years
166 (38.2)
206 (47.4)
1.10 (0.96–1.27)
44 (10.1)
1.05 (0.72–1.53)
19 (4.4)
2.16 (1.09–4.26)
Risk factor
Type 1
a
PR (95% CI)
Type 2
Ptrend
No. (%)
a
PR (95% CI)
Type 1 and 2
Ptrend
No. (%)
PRa (95% CI)
Ptrend
Sex
b
40 (9.3)
1.0
65 (9.3)
1.01 (0.69–1.47)
9 (2.1)
24 (3.5)
1.0
1.71 (0.83–3.51)
Age started Edinburgh University
c
Siblingsd
No
39 (57.4)
19 (27.9)
Yes
469 (44.4)
459 (43.5)
1.56c (1.10–2.20)
1.0
96 (9.1)
9 (13.2)
No
442 (47.1)
384 (40.9)
1.0
Yes
68 (35.5)
95 (50.8)
1.23 (1.04–1.46)
8 (33.3)
10 (41.7)
1.0
1.0
1 (1.5)
0.69 (0.36–1.32)
32 (3.0)
84 (9.0)
1.0
28 (3.0)
21 (11.1)
1.24 (0.79–1.96)
1.0
2.06 (0.29–14.68)
Ever lived in tropical country
Total no. of different vaccinations since birth
.68
None
1.0
5 (2.7)
0.89 (0.35–2.26)
2 (8.3)
1.0
.86
4 (16.7)
1.0
.68
1–4
101 (44.5)
98 (44.1)
1.06 (0.65–1.74)
18 (8.1)
0.49 (0.17–1.37)
5 (2.3)
0.27 (0.05–1.37)
⭓5
401 (45.5)
371 (42.1)
1.01 (0.63–1.63)
83 (9.4)
0.57 (0.22–1.47)
26 (3.0)
0.35 (0.09–1.46)
510 (45.3)
479 (42.5)
All
NOTE. CI, confidence interval; PR, prevalence ratio.
a
b
c
d
Adjusted for ever sexually active before coming to university (yes/no).
P ! .001.
P ! .05.
Data missing for siblings (n p 3).
105 (9.3)
33 (2.9)
Table 4.
Risk of Epstein-Barr virus (EBV) type 1 infection, type 2 infection, and dual infection, by various sexual characteristics.
EBV seropositive
EBV
negative,
no. (%)
Risk factor
Type 1
No. (%)
PR (95% CI)
Type 2
P
a
No. (%)
PR (95% CI)
Type 1 and 2
P
a
No. (%)
PR (95% CI)
P
a
b
Sexually active
No
315 (59.6) 167 (31.6) 1.0
36 (6.8) 1.0
11 (2.1)
c
d
Yes
192 (32.4) 310 (52.4) 1.66 (1.44–1.91)
69 (11.7) 1.71 (1.17–2.51)
21 (3.6)
No. of sex partnersb
.006
.47
None
315 (59.6) 167 (31.6) 1.0
36 (6.8) 1.0
11 (2.1)
96 (40.0) 108 (45.0) 1.43c (1.19–1.71)
70 (29.8) 134 (57.0) 1.81c (1.52–2.15)
1
2–4
⭓5
23 (20.7)
65 (58.6) 1.85c (1.49–2.31)
b
Sex of partner
315 (59.6) 167 (31.6) 1.0
Heterosexual
172 (31.2) 295 (53.4) 1.69 (1.47–1.96)
f
Always
7 (2.9)
1.41 (0.55–3.59)
22 (9.4) 1.38 (0.83–2.28)
18 (16.2) 2.38 (1.39–4.09)
9 (3.8)
5 (4.5)
1.84 (0.77–4.40)
2.17 (0.76–6.15)
13 (52.0)
3 (30.0)
.99
36 (6.8)
c
1.0
d
66 (12.0) 1.76 (1.20–2.58)
8 (32.0) 1.01 (0.59–1.74)
6 (60.0) 1.90c (1.02–3.53)
3 (12.0) 1.76 (0.58–5.39)
0 (0.0)
…
.17
315 (59.6) 167 (31.6) 1.0
c
111 (32.7) 169 (51.4) 1.63 (1.38–1.91)
.65
11 (2.1)
1.0
19 (3.4)
1.66 (0.80–3.45)
1 (4.0) 1.92 (0.25–14.52)
1 (10.0) 4.81 (0.64–36.29)
.77
36 (6.8)
56 (27.1) 119 (57.5) 1.82c (1.52–2.18)
.44
1.0
29 (12.1) 1.78 (1.11–2.83)
.10
No sexual activity
Homosexual
Bisexual
Condom useb
No sexual activity
Never or seldomf
e
1.0
1.71 (0.83–3.51)
1.0
.17
11 (2.1)
1.0
25 (12.1) 1.77e (1.09–2.89)
7 (3.4)
1.63 (0.64–4.14)
37 (11.3) 1.65e (1.07–2.56)
12 (3.7)
1.76 (0.78–3.95)
NOTE. CI, confidence interval; PR, prevalence ratio.
a
b
c
d
e
f
P values for no. of sex partners are for trend; those for other risk factors are for heterogeneity. Tests exclude the non–sexually active group.
Data missing for sexually active (n p 6), no. of sex partners (n p 11), sex of partner (n p 10), and condom use (n p 16).
P ! .001.
P ! .01.
P ! .05.
Analyses restricted to heterosexual students only.
broadly similar to those for type 1 infection but, being based
on much smaller numbers, there was no apparent reduction
in risk among homosexual students than among heterosexual
or bisexual students or among condom users. Sex-specific analyses of dual infection were difficult to interpret because of small
numbers.
DISCUSSION
Risk factors for primary EBV infection have not been fully
investigated, and the present study was undertaken in a large
student population to elucidate these and to identify risk factors
separately for EBV types 1 and 2. Among university entrants
agreeing to participate, 75% were seropositive for EBV-specific
IgG antibodies. Approximately 30% of students who were asked
to participate declined to do so. This nonparticipation rate is
reasonably low for a study involving venepuncture and recruitment from a noninstitutionalized, healthy population. It
is unlikely that nonparticipation could have been directly related to EBV status, because students were unaware of their
EBV status at the time of recruitment. It is possible, however,
that nonparticipation may have been associated with factors
that are themselves associated with EBV risk—for example,
personality and, hence, sexual behavior—and so our estimate
of 75% seroprevalence should be treated with some caution.
Similar studies conducted 30 years ago in university entrants
and US army recruits [30–32] found 45%–69% to be EBV
seropositive, so the present study provides some evidence that
childhood and early adolescent infections may have increased
over time. Morris and Edmunds [33] reported a decrease in
general practitioner consultations for IM from 1970 to 1999 in
England and Wales, particularly among young children, but it
was apparent in young adults as well. Curiously, Morris and
Edmunds also observed a dramatic increase in hospitalizations
due to IM during the 1990s, so the patterns of change in EBV
transmission are unclear. Investigation of large historical serum
banks might elucidate trends in EBV seroprevalence over time.
Few studies have attempted to identify risk factors for childhood seroconversion. Only age and indicators of low socioeconomic status have been unequivocally identified [9–11].
We found an increased seroprevalence among females that
appeared to be confined to type 1 infection. This might indicate
greater childhood susceptibility to type 1 virus among females
or could be a consequence of adolescent sexual behavior: increased risk might occur among females because, on average,
they form relationships with men older than themselves who
are thus more likely to be EBV (in this case, type 1) positive.
Risk Factors for EBV and Its Subtypes • JID 2007:195 (15 February) • 479
We were unable to adjust our results for the age of sex partner,
but we could take account of whether the student had ever
been sexually active; this did not change the results, suggesting
that childhood, rather than sexual, factors are the reason for
increased seropositivity among females. Previous studies have
found conflicting evidence regarding sex differences in EBV
prevalence, though none, to our knowledge, have examined
type 1 and type 2 separately. Two childhood studies in Africa
[34, 35] found little difference in EBV seroprevalence between
boys and girls but found that high titers (11:160) were more
common in girls. European studies [10, 36] have generally
found no difference in EBV seroprevalence between males and
females at any age, although one study [37] found higher seroprevalence among teenage girls than among teenage boys.
The seroprevalence of EBV infection among residents of temperate countries who had previously lived in the tropics has
not been examined before. The finding of increased seroprevalence among such students accords with the much greater
seropositivity among children in tropical than in nontropical
countries [2]. It is noteworthy, however, that, among our students, type 1 as well as type 2 risk was increased, despite type
2 infections being more prevalent in Africa [18]. We found the
greatest seroprevalence among students who were born in a
tropical country, suggesting that risks are greatest among those
who are resident in tropical countries early in life. One potential
explanation could be that transmission can occur during child
birth from young mothers shedding virus in their cervix.
The greater risk of EBV type 1 infection in students with
siblings than in those without is likely due to transmission of
type 1 virus from an infected brother or sister, although we
found no evidence that the risk increased with the number of
siblings or with having older siblings. Previous smaller studies
of the association between the presence of siblings and EBV
seroconversion have not distinguished between type 1 and type
2 and have produced conflicting results: several found no association [9, 11, 36], whereas one found increased seroprevalence among subjects with siblings [38]. The confinement of
our results to type 1 infection suggests a greater susceptibility
to sibling transmission of this type. Prospective studies that
examine the serological status of all siblings within a family,
from infancy through the teenage years, could elucidate the
transmission patterns of the virus within a family setting.
As demonstrated in an earlier study of part of this cohort
[15], sexual activity is a strong risk factor for EBV positivity,
with the risk increasing with the number of sex partners. The
present analyses suggest that this is particularly so for type 1
infection. These data suggest the possibility that EBV may be
transmitted by sexual intercourse, and, given that the virus has
been demonstrated in both male and female genital secretions
[12–14], this is certainly a possibility. However, these findings
alone cannot distinguish between transmission of virus directly
480 • JID 2007:195 (15 February) • Higgins et al.
in genital fluids and transmission of salivary virus by kissing
during sexual intercourse, particularly since a greater number
of sex partners accords with a greater number of oral contacts.
We therefore asked sexually active participants about condom
use and found a lower seroprevalence among those who always
used condoms than among sexually active nonusers or irregular
users, suggesting that infection could be acquired by the sexual
route. However, this difference could be due to differences in
other kinds of sexual activity (e.g., oral contact might be more
intrusive with more risky sexual behavior), particularly because
the seroprevalence among those who always used condoms
remained greater than that among sexually inactive students.
Alternatively, students reporting that they always used condom
may have been overstating the truth, diluting the protective
effect in our results. The protective effect of condom use was
apparent, although not significant, for both type 1 and type 2
infection. Sex-specific analyses suggested that the protection
might apply only to males, but this could reflect sex differences
in the reliability of reporting of condom use.
Our unexpected finding of a lower seroprevalence among
females who were currently taking oral contraceptives than
among sexually active nonusers remained significant after adjustment for other factors. Therefore, this could indicate hormonal involvement in susceptibility to EBV transmission.
The distribution of EBV types found in the present study
(type, 1 77%; type 2, 17%; and dual infection, 5%) was similar
to that found in smaller studies in Europe and the United States.
Given that most infections are with type 1 EBV, it is unsurprising that type 1 risk factors were similar to those for EBV
overall. Type 1 infection prevalence appeared to be greater in
heterosexual and bisexual study participants but, in contrast to
the findings of a previous study [19], there was no evidence
that type 2 infection was more common among homosexual
than heterosexual men. The suggestion that dual infection is
restricted to people who are immunocompromised—particularly those with HIV infection [20–22]—could not be examined
directly in our study, because ethics approval was not granted
for HIV testing. However, 5% of the subjects in our study had
dual infection, and it is unlikely that 5% of the study participants had HIV infection or another major immunocompromised condition; thus, other risk factors are probably involved
in dual infection, which need investigation in a larger study.
Several methodological issues need to be considered in the
interpretation of the present findings. This was a retrospective
cross-sectional study, and the quality of information on risk
factors was, therefore, dependent on the accuracy of the recall
of events that took place years earlier. This could produce bias
if the accuracy of recall was different between EBV-positive and
-negative subjects, but, given that the students were unaware
of their serological status, only nondifferential misclassification
seems likely. This would lead to the underestimation of risks
and the potentially failure to detect true associations but should
not lead to the detection of false-positive findings [39]. The
main outcome measures in the present study, EBV serological
status and EBV type, were determined using validated laboratory methods and, therefore, are unlikely to be inaccurate.
We were unable to type virus for 59% of the EBV-positive serum
samples that were collected in the study. The factors that determine successful typing are unknown but might relate to EBV
load. We found some evidence that typing was more successful
among female than among male students, but there is no obvious reason why overrepresentation of females in the typespecific analyses would distort the risk-factor findings we have
reported.
In conclusion, the present study provides further evidence
that EBV may be sexually transmitted and new evidence that
the risk factors for type 1 and type 2 EBV infection may differ.
In light of findings that IM is caused significantly more often
by EBV type 1 than type 2 infection [40], along with in vitro
evidence of the different biological properties of the 2 types
[41], it seems increasingly likely that infection by these types
has different predisposing factors.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Acknowledgments
We thank the staff of the University of Edinburgh Health Centre and
all the student volunteers for their help with this study. We also thank B.
Armstrong, K. Valentine, A. Glover, G. Edgar, and S. Morris, for phlebotomy; K. Matthews, D. Burns, and J. Douglas, for technical assistance; H.
Nguyen, for data entry and student recruitment; C. Maguire, for student
recruitment; and E. Edgar, for student recruitment and secretarial help.
References
1. Crawford DH. Biology and disease associations of Epstein-Barr virus.
Philos Trans R Soc Lond B Biol Sci 2001; 356:461–73.
2. De-Thé G, Geser A, Day NE, et al. Epidemiological evidence for causal
relationship between Epstein-Barr virus and Burkitt’s lymphoma from
Ugandan prospective study. Nature 1978; 274:756–61.
3. Niederman JC, McCollum RW, Henle G, et al. Infectious mononucleosis—clinical manifestations in relation to EB virus antibodies.
JAMA 1968; 203:205–9.
4. Henle W, Henle G. The virus as the etiologic agent of infectious mononucleosis. In: Epstein MA, Achong BG, eds. The Epstein-Barr virus.
Berlin, Heidelberg, New York: Springer-Varlag, 1979:279–320.
5. White PD, Thomas JM, Amess J, et al. Incidence, risk and prognosis
of acute and chronic fatigue syndromes and psychiatric disorders after
glandular fever. Br J Psychiatry 1998; 173:475–81.
6. Hjalgrim H, Askling J, Rostgaard K, et al. Characteristics of Hodgkin’s
lymphoma after infectious mononucleosis. N Engl J Med 2003; 349:
1324–32.
7. Pereira M, Blake, JM, Macrae AD. EB virus antibody at different ages.
Brit Med J 1969; 4:526–7.
8. Crawford DH. Epstein-Barr virus. In: Zucherman AJ, Banatvala JE,
Pattison JR, et al., eds. Principles and practice of clinical virology. 5th
ed. London: John Wiley & Sons, 2004:123–46.
9. Dan R, Chang RS. A prospective study of primary Epstein-Barr virus
infections among university students in Hong Kong. Am J Trop Med
Hyg 1990; 42:380–5.
10. Hesse J, Ibsen KK, Krabbe S, Uldall P. Prevalence of antibodies to
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
Epstein-Barr Virus (EBV) in childhood and adolescence in Denmark.
Scand J Infect Dis 1983; 15:335–8.
Hallee TJ, Evans AS, Niederman JC, Brooks CM, Voegtly H. Infectious
mononucleosis at the United States Military Academy: a prospective
study of a single class over four years. Yale J Biol Med 1974; 47:182–95.
Sixbey JW, Lemon SM, Pagano JS. A second site for Epstein-Barr virus
shedding: the uterine cervix. Lancet 1986; 2:1122–4.
Israele V, Shirley P, Sixbey JW. Excretion of the Epstein-Barr virus from
the genital tract of men. J Infect Dis 1991; 163:1341–3.
Näher H, Gissmann L, Freese UK, Petzoldt D, Helfrich S. Subclinical
Epstein-Barr virus infection of both the male and female genital tract—
indication for sexual transmission. J Invest Dermatol 1992; 98:791–3.
Crawford DH, Swerdlow AJ, Higgins C, et al. Sexual history and Epstein-Barr virus infection. J Infect Dis 2002; 186:731–6.
Dambaugh T, Wang F, Hennessy K, Woodland E, Rickinson A, Kieff
E. Expression of the Epstein-Barr virus nuclear protein 2 in rodent
cells. J Virol 1986; 59:453–62.
Sample J, Young L, Martin B, et al. Epstein-Barr virus types 1 and 2
differ in their EBNA-3A, EBNA-3B, and EBNA-3C genes. J Virol
1990; 64:4084–92.
Gratama JW, Ernberg I. Molecular epidemiology of Epstein-Barr virus
infection. Adv Cancer Res 1995; 67:197–255.
van Baarle D, Hovenkamp E, Dukers NHTM, et al. High prevalence
of Epstein-Barr virus type 2 among homosexual men is caused by
sexual transmission. J Infect Dis 2000; 181:2045–9.
Luxton JC, Williams I, Weller I, Crawford DH. Epstein-Barr virus
infection of HIV-seropositive individuals is transiently suppressed by
high-dose acyclovir treatment. AIDS 1993; 7:1337–43.
Falk KI, Zou J-Z, Lucht E, Linde A, Ernberg I. Direct identification
by PCR of EBV types and variants in clinical samples. J Med Virol
1997; 51:355–63.
Sculley TB, Apolloni A, Hurren L, Moss DJ, Cooper DA. Coinfection
with A- and B-type Epstein-Barr virus in human immunodeficiency
virus–positive subjects. J Infect Dis 1990; 162:643–8.
Denis M, Haumont H, Bollen A. Vaccination against Epstein-Barr virus
(EBV): report of phase II studies using recombinant viral glycoprotein
gp350 in healthy adults [abstract 14.01]. In: Program and abstracts of
the 11th Biennial Conference of the International Association for Research on Epstein-Barr Virus and Associated Diseases. Regensburg,
Germany: 2004.
Henle G, Henle W. Immunofluorescence in cells derived from Burkitt’s
lymphoma. J Bact 1966; 91:1248–56.
Reedman BM, Klein G. Cellular localization of an Epstein-Barr virus
(EBV)-associated complement-fixing antigen in producer and nonproducer lymphoblastoid cell lines. Int J Cancer 1973; 11:499–520.
Brooks JM, Croom-Carter DSG, Leese AM, Tierney RJ, Habeshaw G,
Rickinson AB. Cytotoxic T-lymphocyte responses to a polymorphic
Epstein-Barr virus epitope identify healthy carriers with coresident viral
strains. J Virol 2000; 74:1801–9.
Hughes K. Odds ratios in cross-sectional studies. Int J Epidemiol
1995; 24:463–4, 468.
Barros AJD, Hirakata VN. Alternatives for logistic regression in crosssectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Med Res Methodol 2003; 3:21.
StatCorp. Stata statistical software: release 8.0. College Station, TX:
Stata Corporation, 2003.
Niederman JC, Evans AS, Subrahmanyan L, McCollum RW. Prevalence, incidence and persistence of EB virus antibody in young adults.
N Engl J Med 1970; 282:361–5.
Sawyer RN, Evans AS, Niederman JC, McCollum RW. Prospective
studies of a group of Yale University freshmen. I. Occurrence of infectious mononucleosis. J Infect Dis 1971; 123:263–70.
Infectious mononucleosis and its relationship to EB virus antibody: a
joint investigation by university health physicians and P.H.L.S. laboratories. Brit Med J 1971; 4:643–6.
Morris MC, Edmunds WJ. The changing epidemiology of infectious
mononucleosis? J Infect 2002; 45:107–9.
Risk Factors for EBV and Its Subtypes • JID 2007:195 (15 February) • 481
34. Kafuko GW, Henderson BE, Kirya BG, Munube GMR. Epstein-Barr
virus antibody levels in children from the West Nile district of Uganda.
Lancet 1972; 1:706–9.
35. Biggar RJ, Gardiner C, Lennette ET, Collins WE, Nkrumah FK, Henle
W. Malaria, sex, and place of residence as factors in antibody response
to Epstein-Barr virus in Ghana, West Africa. Lancet 1981; 2:115–8.
36. Crowcroft NS, Vyse A, Brown DWG, Strachan DP. Epidemiology of
Epstein-Barr virus infection in pre-adolescent children: application of
a new salivary method in Edinburgh, Scotland. J Epidemiol Community Health 1998; 52:101–4.
37. Morris MC, Edmunds WJ, Hesketh LM, et al. Sero-epidemioloigical
patterns of Epstein-Barr and herpes simplex (HSV-1 and HSV-2) viruses in England and Wales. J Med Virol 2002; 67:522–7.
482 • JID 2007:195 (15 February) • Higgins et al.
38. Lai PK, Mackay-Scollay EM, Alpers MP. Epidemiological studies of
Epstein-Barr herpesvirus infection in western Australia. J Hyg (Lond)
1975; 74:329–37.
39. Rothman KJ, Greenland S. Modern epidemiology. 2nd ed. Philadelphia:
Lippincott-Raven, 1998.
40. Crawford DH, Macsween KF, Higgins CD, et al. A cohort study among
university students: identification of risk factors for Epstein-Barr virus
seroconversion and infectious mononucleosis. Clin Infect Dis 2006;
43:276–82.
41. Rickinson AB, Young LS, Rowe M. Influence of Epstein-Barr virus
nuclear antigen EBNA2 on the growth phenotype of virus-transformed
B cells. J Virol 1987; 61:1310–7.