Does Prior Infection with Varicella-Zoster Virus Influence Risk of Adult Glioma?

Abstract

To evaluate a possible association between varicella-zoster virus infection and glioma, the authors asked adults with glioma (n = 462) whose tumors were diagnosed between August 1, 1991, and March 31, 1994, and age-, sex-, and ethnicity-matched controls (n = 443) about their histories of chickenpox or shingles. Cases were significantly less likely than controls to report a history of either chickenpox (odds ratio = 0.4, 95% confidence interval (CI) 0.3-0.6) or shingles (odds ratio = 0.5, 95% CI 0.3-0.8). To obtain serologic support for these findings, the authors conducted double-blind enzyme-linked immunosorbent assays for immunoglobulin G antibodies to varicella-zoster virus among 167 self-reporting subjects for whom blood samples were available. Cases and controls reporting no history of chickenpox were equally likely to test positive (73% vs. 75%), but among those reporting a positive history, cases were less likely than were controls to test positive (71% vs. 85%). Despite the misclassification, an odds ratio of 0.6 was obtained using either serologic data (95% CI 0.3-1.3) or reported history of chickenpox (95% CI 0.3-1.1) in this subgroup of subjects. This suggests that adults with glioma were less likely than controls either to have had prior varicella-zoster virus infection or to have an immunoglobulin G antibody response adequate to indicate positivity. Since either explanation suggests novel mechanisms for brain tumor pathogenesis, these findings require corroboration and elaboration.

Full text

American Journal of Epidemiology
Copyright © 1997 by The Johns Hopkins University School of Hygiene and Public Health
All rights reserved
Vol 145, No 7
Printed in US A.
A BRIEF ORIGINAL CONTRIBUTION
Does Prior Infection with Varicella-Zoster Virus Influence Risk of Adult
Glioma?
Margaret Wrensch,
1
Adriana Weinberg,
2
John Wiencke,
1
Helen Masters,
2
Rei Muke,
1
Geoffrey Barger,
3
and
Manon Lee
1
To evaluate a possible association between varicella-zoster virus infection and glioma, the authors asked
adults with glioma (n = 462) whose tumors were diagnosed between August 1,1991, and March 31,1994, and
age-,
sex-, and ethnicity-matched controls (n - 443) about their histories of chickenpox or shingles. Cases
were significantly less likely than controls to report a history of erther chickenpox (odds ratio = 0.4, 95%
confidence interval (Cl) 0.3-0.6) or shingles (odds ratio = 0.5, 95% Cl 0.3-0.8). To obtain serologic support for
these findings, the authors conducted double-blind enzyme-linked immunosorbent assays for immunoglobuhn
G antibodies to varicella-zoster virus among 167 serf-reporting subjects for whom blood samples were
available. Cases and controls reporting no history of chickenpox were equally likely to test positive (73% vs.
75%),
but among those reporting a positive history, cases were less likely than were controls to test positive
(71
% vs. 85%). Despite the misclassification, an odds ratio of 0.6 was obtained using either serologic data
(95%
Cl 0.3-1.3) or reported history of chickenpox (95% Cl 0.3-1.1) in this subgroup of subjects. This
suggests that adults with glioma were less likely than controls erther to have had pnor varicella-zoster virus
infection or to have an immunoglobulin G antibody response adequate to indicate positivity. Since either
explanation suggests novel mechanisms for brain tumor pathogenesis, these findings require coiToboration
and elaboration. Am J Epidemiol 1997;145:594-7.
chickenpox; glioma; herpesvirus 3, human
Editor's note: A companion article by Wrensch et
al. appears on page 581 of this issue.
It has long been speculated that infectious agents or
an immunologic response to these agents may play a
role in causing, promoting, or preventing brain tumors
or other cancers (1, 2). Certain viruses are known to
induce glioma in test animals (1). There are intriguing
but inconclusive epidemiologic data linking simian
virus 40 infection with increased incidence of glioblas-
toma multiforme and medulloblastoma (3). Bithell et
al.
(4) reported a statistically significant excess of
children with medulloblastoma born to mothers who
Received for publication February 20, 1996, and In final form
August 2, 1996
Abbreviation: Cl, confidence interval.
1
Department of Epidemiology and Biostatlstics, School of Med-
icine,
University of California, San Francisco, CA
2
Pediatric Infectious Diseases, Diagnostic Virology Laboratory,
University of Colorado Hearth Sciences Center, Denver, CO.
3
Department of Neurology, School of Medicine, Wayne State
University, Detroit, Ml.
Reprint requests to Dr. Margaret Wrensch, Box 0560, University
of California School of Medicine, San Francisco, CA 94143-0560.
had had chickenpox during pregnancy, but the results
were based on only three observed cases, with 0.3
cases expected.
With regard to other infectious agents, Schuman et
al.
(5) showed that astrocytoma patients were signifi-
cantly more likely than controls to have antibodies to
Toxoplasma gondii using the Sabin-Feldman dye test,
and they showed that animals exposed to Toxoplasma
can develop glioma. Ryan et al. (6) could not confirm
this finding in a more recent Australian study; how-
ever, they did not report histologic types for their
subjects, and Schuman et al.'s findings were confined
to astrocytoma.
In the ongoing San Francisco Bay Area Adult Gli-
oma Study, subjects were asked about their histories of
chickenpox and shingles. Both are caused by varicella-
zoster virus, a herpesvirus which is known to have
nervous system involvement (7), and we thought they
might be comparatively memorable infections. Cases
were significantly less likely than controls to report a
history of both chickenpox (odds ratio = 0.4, 95
percent confidence interval (Cl) 0.3-0.6) and shingles
(odds ratio = 0.5, 95 percent Cl 0.3-0.8) (8). In this
report, we present serologic support for the finding
594
by guest on July 12, 2011aje.oxfordjournals.orgDownloaded from

Varicella-Zoster Virus and Adult Brain Cancer 595
that glioma cases were less likely than controls either
to have had varicella-zoster infection or to have anti-
bodies to this virus.
MATERIALS AND METHODS
Details on subject recruitment and interview have
been given elsewhere (8). Briefly, 462 eligible adults
newly diagnosed with glioma in any of six San
Francisco Bay Area Counties between August 1, 1991,
and March 31, 1994, participated; 443 age-, sex-, and
ethnicity-matched controls were obtained through ran-
dom digit dialing. Participation rates were 82 percent
for cases and 63 percent for controls. In-person struc-
tured interviews asked about personal and familial
medical history, demographic factors, and potential
brain tumor risk factors.
Blood sample collection was undertaken partway
through the study. Up to 30 ml of blood was collected
from 187 cases and 169 controls in heparinized green-
topped tubes. Whole blood was stored at — 70°C and
was transported frozen using dry ice.
Because the blood specimens were obtained primar-
ily for polymorphism analyses, serologic studies were
conducted only on a sample of available blood speci-
mens.
The subsample originally included all subjects
with a negative history of chickenpox and a random
sample of 43 percent of those with a positive history.
Upon specimen retrieval, insufficient blood remained
for serologic studies from 16 glioma cases with a
negative chickenpox history. Consequently, to pre-
serve the observed association between chickenpox
history and glioma, an additional 13 glioma cases with
a positive chickenpox history were deleted from sero-
logic studies.
Serologic analysis was performed using the Vari-
cella STAT enzyme-linked immunoassay (BioWhit-
taker, Walkersville, Maryland) according to the man-
ufacturer's instructions, with modified criteria of
interpretation. Briefly, test sera and controls were di-
luted at 1:20 and added to antigen-coated wells in a
microtiter plate. Bound antibodies were revealed with
peroxidase-conjugated anti-human immunoglobulin G
and colorimetric substrate. Absorbances measured
with a spectrophotometer were used to calculate the
varicella index for each serum sample by dividing the
absorbance of the test serum well by the absorbance of
the manufacturer's positive control. Based on a previ-
ous study (9), a varicella index £ 0.9 indicated the
absence of specific antibodies; a varicella index S: 1.2
indicated immunity; and a varicella index greater than
0.9 but less than 1.2 was reported as borderline, be-
cause it did not correlate with immunity nor did it
exclude past infection. As performed, this test has 87
percent sensitivity and 91 percent specificity in com-
parison with cell-mediated immunity (9). Serologic
analysis was conducted blind to both case-control sta-
tus and reported history of chickenpox.
All odds ratios comparing cases with controls were
adjusted for age using the SAS Logistic procedure (10).
RESULTS
History of chickenpox among all subjects and
those sampled for blood and serologic analysis
Table 1 compares the odds ratios for a reported
history of chickenpox among all subjects, as well as
among subjects from whom a blood sample was ob-
tained, those selected for serologic study, and those for
whom adequate blood specimens remained for sero-
logic study. Among subjects from whom blood was
obtained, a significant negative association between
chickenpox history and glioma persisted (table 1). As
TABLE 1. History of chickenpox among glioma cases and controls, San Franctsoo Bay Area Adult
Glioma Study, 1991-1995
All subjects
All subjects from whom blood
was obtained
Subsample selected for
serologic studies^
Subsample with sufficient blood
available for serologic
studies^
Ca
No.wSti
posUve
history
267
138
59
45
No.wtth
negative
history
114
49
49
33
Controls
No.wBh
positive
history
348
141
61
61
No wBh
negative
hfstoiy
66
28
28
28
OR'.t
0.4
0.5
0.5
0.6
95%
a*
0.3-0.6
0.3-0.8
0.3-0.9
0.3-1.1
* OR, odds ratio; Cl, confidence interval,
t Adjusted for individual year of age.
i See text for explanation of sampling strategy. These subsamples were specifically chosen to have similar
odds ratios for a reported history of chickenpox as that found in all subjects from whom blood was obtained.
Am J
Epidemiol
Vol. 145, No. 7, 1997
by guest on July 12, 2011aje.oxfordjournals.orgDownloaded from

596 Wrensch et al.
stated above, the subsamples selected for serologic
analysis were chosen to be representative of the per-
sons from whom blood was obtained and to preserve
the observed negative association between chickenpox
history and glioma.
Serologic results for immunoglobulin G
antibodies to varicella-zoster virus
The odds ratio for the presence of immunoglobulin
G antibodies to varicella-zoster virus for glioma cases
versus controls was 0.6 (table 2). This is the same
magnitude of association as that found for reported
history of chickenpox, despite rather substantial mis-
classification (table 3); the overall sensitivity of a
reported chickenpox history using immunoglobulin G
as the standard was 65 percent (84/129), and the
specificity was 43 percent (13/30). The proportions of
cases and controls without a history of chickenpox
who were antibody-positive were very similar (73
percent and 75 percent, respectively). However, 85
percent of history-positive controls but only 71 percent
of history-positive cases were antibody-positive.
Prior chemotherapy and radiation did not appear to
explain the lower prevalence of immunoglobulin G
antibodies to varicella-zoster virus among glioma
cases versus controls. Twelve percent (2/17) of
antibody-negative and 27 percent (15/56) of antibody-
positive glioma patients reported prior chemo-
therapy, while 82 percent (14/17) of antibody-
negative and 89 percent (50/56) of antibody-positive
patients reported radiation therapy. Two antibody-
negative cases received chemotherapy 6 or 11 days
before the blood drawing. Seven antibody-positive
cases had chemotherapy on the day of blood drawing;
the other eight had chemotherapy 4-81 days before
blood collection. The average number of days between
the last radiation treatment and the blood drawing was
150 for antibody-negative cases and 140 for antibody-
positive cases.
DISCUSSION
Although there was substantial misclassification be-
tween self-reported history of chickenpox and the
presence of immunoglobulin G antibodies to varicella-
zoster virus, the nature of the misclassification was
such that the odds ratio for glioma cases versus con-
trols for history of chickenpox was very similar to the
odds ratios for positive antibodies to varicella-zoster
virus.
This appears to be the first time the inverse
association with varicella-zoster virus antibodies has
been reported; because power was limited, we choose
to interpret the finding cautiously.
However, because the serologic data support the
statistically significant results for reported histories of
chickenpox and shingles (8), it is worth considering
the implications of the finding of a negative associa-
tion between glioma and varicella-zoster virus infec-
tion or antibodies to varicella-zoster virus. One possi-
bility for an effect of this size is confounding;
however, given the general ubiquity of the varicella-
zoster virus, it is difficult to imagine what the con-
founding factors might be. Possible effects of increas-
ing age on loss of antibody were accounted for through
age adjustment with logistic analyses.
Another possibility is that having a brain tumor
depresses serum immunoglobulin levels, such that it
would be more difficult to detect any immunoglobulin
G in cases compared with controls. On the contrary,
one recent study (11) found serum immunoglobulin G
levels to be significantly higher in glioma cases than in
controls. The comparability of that study group to the
present series is unclear; neither the origin of the
control group nor the treatments received by cases
were reported. The data from this current study indi-
cate that prior radiation or chemotherapy treatments
would not be likely explanations for reduced levels of
immunoglobulin G among the glioma cases.
People with glioma may be less likely than controls
to have a history of a wide variety of infections, but
the evidence is scant. In a report by Schlehofer et al.
TABLE 2. Presence of Immunoglobultn G antibodies to varicella-zoster virus among gDoma
controls, San Francisco Bay Area Adult Glioma Study, 1991-1995
and
ImmunoglobuBn Q
antfcocttesto
varicella-rosier
virus
Negative
Positive
Borderline
Total
Ce
Mo.
17
56
5
78
ises
%
21.8
71.8
6.4
100
No
13
73
3
89
Controls
%
14.6
82.0
3.4
100
OR'.t
(95%CI»)
1.0
0.6(0.3-1.3)
* OR, odds ratio; Cl, confidence interval.
fOdds ratios for positive versus negative antibodies to varicella-zoster virus (borderline results were
considered missing data), adjusted for incividuai year of age. If borderline results are considered positive, OR -
0.6, 95% Cl 0.3-1.3; if borderline results are considered negative, OR
•=•
0.5, 95% Cl
0.3-1.1.
Am J Epidemiol Vol. 145, No. 7, 1997
by guest on July 12, 2011aje.oxfordjournals.orgDownloaded from

Varicella-Zoster Virus and Adult Brain Cancer 597
TABLE 3. Presence of Immunoglobulin G antibodies to varicella-zoster virus by reported history of chickenpox, San Francisco
Bay Area Adult Qlloma Study, 1991-1995
Immunogtobuln Q
antibodies to
varicela-zoster
vfrus
Negative
Positive
Borderline
Total
NegaUva history ot cNckerpox
Qfloma cases
No.
7
24
2
33
%
21.2
72.7
6.1
100
No.
6
21
1
28
Controls
%
21.4
75.0
3.6
100
Alls»t>jects
No.
13
45
3
61
%
21.3
73.8
4.9
100
Positive history of chictenpox
Qioma cases
No.
10
32
3
45
%
22.2
71.1
6.7
100
No.
7
52
2
61
Controb
%
11.5
85.2
3.3
100
Afl subjects
No.
17
84
5
106
%
16.0
79.3
4.7
100
(12),
adults with newly diagnosed primary brain tu-
mors were less likely than were population-based con-
trols to report colds or infections during the 5 years
prior to interview. The result showed a dose response,
with a lower odds ratio (odds ratio = 0.3, 95 percent
CI 0.1-0.8) for brain tumors appearing among those
reporting three or more colds or infections per year
than among those reporting 1-2 colds or infections per
year (odds ratio = 0.8). The results also were very
similar for men and women. Data on prior infections
were not reported separately for glioma and meningi-
oma. Schlehofer et al. interpreted this finding as indi-
cating that general activation of the immune system
might play a role in influencing these tumors. The
implication seems to be that cancer cells, in general,
might be more readily destroyed with a heightened
immune system. In partial support of this contention,
Abel et al. (13) reported a decreased history of prior
colds and several childhood infections, including
chickenpox, among newly diagnosed cancer patients
(stomach, colorectal, breast, and ovarian carcinoma)
compared with controls. They thoroughly discussed
the previous literature on infections and cancer risk.
Hypotheses have included either that infections de-
crease the chance of cancer development or kill exist-
ing cancer cells; alternatively, infections may be less
likely to arise in individuals who are more susceptible
to cancer or who are developing cancer. Reporting
bias by cases, diminishing the reported severity of
their prior illnesses, was also suggested. However, this
bias could not explain the present serologic findings.
Another, more speculative explanation is that if a virus
or viruses with some cross-reactivity to varicella-
zoster virus influence glioma development, individu-
als with stronger immunity to varicella-zoster virus
might be less susceptible to glioma.
Clearly, the role of viral and other infections in brain
tumor etiology requires further epidemiologic investi-
gation. This is especially important given the ex-
tremely poor prognosis of most brain cancers and the
paucity of available knowledge with which to develop
meaningful preventive strategies. Our results suggest
that self-reported history of chickenpox is an unreli-
able indicator of serologic positivity. In addition to the
need for replication of these results, it would be valu-
able to compare glioma cases and controls for anti-
bodies to herpes simplex virus, cytomegalovirus, and
Epstein-Barr virus. Herpes simplex virus is another
herpesvirus that establishes latency in the nervous
system, while cytomegalovirus and Epstein-Barr virus
are herpesviruses that infect the central nervous sys-
tem without establishing latency.
ACKNOWLEDGMENTS
This work was supported by grants RO1CA52689 and
RO3CA57220 from the National Institutes of Health.
REFERENCES
1 Johnson RT. Cerebral tumors. In: Viral infections of the
nervous system New York, NY- Raven Press, 1982.293-310.
2.
Schoenberg BS. Nervous system. In: Schottenfeld D,
Fraumeni JF Jr. Cancer epidemiology and prevention. Phila-
delphia, PA: WB Saunders Company, 1982:968-83.
3.
Geissler E SV40 and human brain tumors. Prog Med Virol
199O;37:211-22.
4.
Bithell JF, Draper GJ, Gorbach PD. Association between
malignant disease in children and maternal virus infections. Br
MedJ 1973;1.706-8.
5 Schuman LM, Choi NW, Gullen WH. Relationship of central
nervous system neoplasms to Toxoplasma gondii infection.
Am J Public Health 1967;57:848-56
6. Ryan P, Hurley SF, Johnson AM, et al. Tumours of the brain
and presence of antibodies to Toxoplasma gondii. Int J Epi-
demiol 1993,22:412-19.
7.
Taylor-Robinson D, Caunt AE. Varicella virus. New York,
NY: Springer-Verlag, 1972.
8. Wrensch M, Lee M, Muke R, et al. Familial and personal
medical history of cancer and nervous system conditions
among adults with glioma and controls Am J Epidemiol
1997,145:581-93.
9 Weinberg A, Hayward A, Masters H, et al. Comparison of two
methods for detecting varicella-zoster virus antibody with
varicella-zoster virus cell-mediated immunity J Clin Micro-
biol 1996;34:445-6.
10.
SAS Institute, Inc. SAS/STAT user's guide, version 6. 4th ed.
Vol 2. Cary, NC. SAS Institute, Inc, 1990.
11.
Manjula S, Aroor AR, Raja A, et al. Serum immunoglobulins
in brain tumours. Acta Neurochir
1992;
115:103-5.
12.
Schlehofer B, Blettner M, Becker N, et al. Medical risk factors
and the development of brain tumors. Cancer 1992;69:2541-7.
13.
Abel N, Becker R, Angerer R, et al. Common infections in the
history of cancer patients and controls. Cancer Res Clin Oncol
1991;
117:339-44.
Am J Epidemiol Vol. 145, No. 7, 1997
by guest on July 12, 2011aje.oxfordjournals.orgDownloaded from