Secondary Schizophrenia (111 page)

Brown et al., 2004 [41]

Elevated maternal interleukin-8,

CHDS/PDS

p
=
.04

second trimester

Buka et al., 2001 [21]

Herpesvirus type 2 (HSV2) IgG

NCPP
∗∗

p
=
.02
∗∗∗

antibody, at delivery of neonate

Babulas et al., 2006 [25]

Maternal genital/ reproductive

CHDS/PDS

RR
=
5.0, 95% CI
=
2.0–3.7, p
=
.001

infections, periconceptional

∗
OR
=
Odds ratio, RR
=
Rate ratio. P-values are provided when no OR or RR are reported;
∗∗
National Collaborative Perinatal Project;
∗∗∗
Outcome is psychosis.

the exposure, since the majority of individuals who
records. Below, we review the findings of these studies.

were in utero during influenza epidemics would not
These findings are summarized in
Table 22.1.

have been exposed to this virus; such misclassification
would bias any association toward the null
[5].
Further
Influenza

methodological limitations included the sole reliance
Our group examined the relationship between mater-on hospital registry diagnoses, which could result in
nal influenza and risk of schizophrenia utilizing
diagnostic misclassification, and selection biases from
archived maternal serum samples from mothers in the
loss to follow-up. Similar study designs have been
cohort of the Child Health and Development Study
used to assess relationships between other infections
(CHDS)
[10, 11]
, which enrolled pregnant women who
and risk of schizophrenia. These studies suggest that
were members of the Kaiser Permanente Medicare
exposure to epidemics of certain respiratory infections
Care Plan (KPMCP) from 1959 to 1967 in Alameda
[6, 7],
measles, varicella-zoster
[8],
and polio
[9]
were
County, California. In the Prenatal Determinants of
associated with later schizophrenia, mostly in the
Schizophrenia (PDS) study, we ascertained and diag-second trimester.

nosed schizophrenia using a structured diagnostic
interview and longitudinal psychiatric records among
members of this birth cohort
[12].
(This cohort was
Serologic studies

also used in our studies of toxoplasmosis, cytokines,
In order to surmount the limitations of previous stud-maternal genital/reproductive infection, and maternal
ies, our group and others have capitalized on sev-respiratory infection, described below).

eral methodological advantages made possible by the
To characterize influenza infection in the cohort,
use of large birth cohorts that have been followed
we analyzed the maternal sera corresponding to cases
up for schizophrenia in adulthood. These advantages
and matched controls in the cohort for influenza anti-include the use of maternal sera acquired during preg-body. Because the gestational timing of the availabil-nancy, permitting the use of biomarkers to define
ity of serum samples in each pregnancy did not per-exposure status, prospectively collected data on infec-mit the assessment of a fourfold rise in antibody titer
tious exposures, and, in most cases, rigorous diag-between serial samples (the “gold standard” for diag-

280

noses using research-based interviews and psychiatric
nosis of influenza infection), we utilized a method to
Chapter 22 – Infection and schizophrenia

characterize the presence of influenza using only a sin-Although that study differed from our work in that
gle serum sample. For this purpose we validated an
prenatal sera were not available, the source of IgG to
antibody titer of > 20 against known influenza infec-toxoplasma in neonatal blood should have been infection in the cohort using serial samples from preg-tion that occurred in the mother either during or prior
nancies that had serum samples from each trimester,
to pregnancy, as would have been the case in our study.

and who were neither cases nor controls in the study.

Toxoplasma may increase liability to schizophre-The validity parameters, including sensitivity, speci-nia via reactivation of toxoplasma oocysts. After acute
ficity, and positive predictive value, were excellent (see
infection, toxoplasma oocysts exist in a dormant state
[13]
for description). We found that maternal exposure
in which they are sequestered in affected organs,
during the first half of pregnancy was associated with
including the brain
[16, 17].
Host immunosuppres-an increased risk of schizophrenia in adult offspring
sion or other factors can lead these cysts to rupture
(OR
=
3.0, 95% CI
=
0.98–10.1, χ2
=
3.8, p
=
.052).

and become transformed from the inactive bradyzoite
For first trimester exposure to influenza, we observed a
form to the active tachyzoite form. Tachyzoites pro-sevenfold increased risk of schizophrenia in offspring
liferate and invade cells and potentially cross the pla-

(OR
=
7.0, 95% CI
=
0.7–75.3, χ2
=
3.0, p
=
.08)
[13].

centa and infect the fetus
[16, 20].
Another possibility
There was no increased risk of schizophrenia for expo-is that elevated toxoplasma IgG may itself be terato-sures occurring in the second half of pregnancy, for the
genic, or that the cytokine response which suppresses
second trimester as a whole, or for the third trimester
toxoplasma from becoming reactivated may damage
[13].

the fetal brain
[18].

In gestational influenza exposure, the mechanism
of teratogenesis is not clearly understood because
the virus does not appear to cross the placenta or
Herpes simplex virus

blood-brain barrier and invade the fetal brain
[14].

In a study from the Collaborative Perinatal Project
One possible mechanism is that maternal IgG anti-

(CPP), an association was observed between IgG levels
bodies elicited by influenza, rather than the infection
to herpes simplex virus type 2 (HSV-2) and risk of psy-itself, cross the placenta and react with fetal brain
chotic illness in 27 offspring (p
=
.04)
[21].
These off-antigens by molecular mimicry, thereby disturbing
spring had diagnoses of schizophrenia, affective psy-fetal brain development and increasing vulnerability
choses, and other psychotic disorders. No association
to schizophrenia
[15].
Another plausible mechanism
was observed in this study between psychotic disoris that the effect is mediated by maternal cytokines
ders and IgG antibody to herpes simplex virus type
(discussed later), which have been associated with
1 (HSV-1), cytomegalovirus (CMV), rubella virus,
periventricular leukomalacia, cerebral palsy, and pre-human parvovirus B19, chlamydia trachomatis, or
term birth.

human papillomavirus type 16. We were not able to
replicate this finding, however, in a larger investigation on schizophrenia and other schizophrenia spec-

Toxoplasmosis

trum disorders, the vast majority (85%) of whom had
The plausibility of Toxoplasma gondii (T. gondii) as
either schizophrenia or schizoaffective disorder based
a risk factor for neurodevelopmental schizophrenia
on the PDS study
[22].
This negative finding persisted
is supported by its association with CNS anomalies
whether HSV-2 IgG antibody was categorized as a
and both subtle and severe fetal and childhood neu-dichotomous (OR
=
1.13, 95% CI
=
0.51–2.55, p
=

rocognitive disturbances
[16, 17].
Using bioassays on
0.76) or as a continuous variable (OR
=
0.82, 95% CI
=

archived maternal sera, we demonstrated that elevated
0.09–7.75, p
=
0.86) for subjects who were seroposi-toxoplasma IgG antibody titer (> 1:128) was associ-tive to HSV-2. For the latter analysis, we also analyzed
ated with an increased risk of schizophrenia in adult
the IgG levels in both seropositive and in seronega-offspring (OR
=
2.6, 95% CI
=
1.0–4.8, χ2
=
3.8,
tive subjects, as in the study from the CPP cohort;
p
=
.05)
[18]
. For toxoplasma IgM, there was no ele-again, no effects were observed. Differences between
vation in maternal sera for any of the cases or con-the findings of the two studies may have resulted
trols, suggesting that active infection is unlikely. This
from the fact that our study focused on schizophre-work has recently been replicated in a Danish sam-nia and schizophrenia spectrum disorders, and had a
281

ple with neonatal blood from filter paper samples
[19].

considerably larger sample, which may have provided
Organic Syndromes of Schizophrenia – Section 3

greater stability of odds ratios and confidence inter-two months of pregnancy
[27]
. We also capitalized
vals. However, an even larger study of HSV-2 IgG antion prospective data collected in childhood and ado-body in neonatal filter paper blood samples from Den-lescence in nearly every subject in the cohort, which
mark, which utilized assay methodology identical to
allowed us to demonstrate a marked decline in IQ in
that in the CPP study, also did not report an asso-nearly every subject who later developed schizophre-ciation with “schizophrenia and related disorders,”

nia spectrum disorders; in contrast, only one-third
although no measures of effect were reported in that
of exposed subjects who remained free of the disor-paper
[19]
. Because this work is ongoing in each of
der at the time of follow-up
[27]
exhibited a decline
the birth cohorts described above, it would be prema-in IQ. This finding provides validation of the strong
ture to conclude at this time that there is no associ-association between prenatal rubella and schizophre-ation between prenatal herpes viruses and the risk of
nia, which has since been associated with premorbid
schizophrenia.

cognitive decline in population-based samples
[32].

Whereas the mechanisms that may increase the risk
Maternal genital/reproductive

of schizophrenia are unclear, rubella has long been
documented to cross the placenta, enter the fetal
infections

brain, and disrupt development of the central ner-It has long been known that maternal geni-

vous sytem
[33],
resulting in inhibition of mitosis,
tal/reproductive (G/R) infections during pregnancy
which leads to diminished neurons and total brain
increase the risk of congenital neurological disorders
size
[34, 35],
diminished replication of oligodendro-in offspring
[23, 24]
. We demonstrated that pericon-cytes, and a consequent deficit of myelin
[36],
as well
ceptional genital/reproductive infection was related
as a pro-inflammatory response, which increases levels
to a marked elevation in schizophrenia risk among
of cytokines and other immune-mediating molecules,
offspring (RR
=
5.0, 95% CI
=
2.0–3.7, p
=
.001)
leading to ischemic damage
[35, 37].

[25]
. Unlike the infections previously described,
these microbes directly infect the uterus and an effect
Cytokines

on schizophrenia may arise from infection of the
early embryo. In that study, HSV was not reported
The cytokines are a family of soluble, polypeptide
in any of the mothers of cases who later developed
proteins that are known to mediate host responses
schizophrenia.

to a broad array of infections
[38]
and are known
to lead to adverse reproductive outcomes
[39, 40].

We found a marked and significant increase in the
Rubella

mean level of serum interleukin-8 (IL-8) in the moth-Like T. gondii, rubella is a known central nervous sys-ers of schizophrenia cases, as compared to mothers
tem teratogen. In the Rubella Birth Defects Evalua-of controls (p
=
0.02); levels of this cytokine were
tion Project (RBDEP) we had a unique opportunity to
nearly twice as high in mothers of schizophrenia
relate in utero rubella infection to adult schizophre-offspring than controls
[41]
. There were no differ-nia
[26, 27].
A key advantage of the RBDEP cohort
ences with regard to maternal IL-1
β
, IL-6, and TNF-was that all mothers and offspring in the cohort were
α
between schizophrenia cases and controls. Unlike
prospectively documented with prenatal rubella infec-other cytokines, interleukin-8 (IL-8) has been associ-tion by clinical examination
[26, 27, 28, 29]
and sero-ated with chorioamnionitis in infants born at term,
logic testing
[28]
. Such a cohort was ideal for inves-and maternal and neonatal serum IL-8 levels are signif-tigating the role of a prenatal infection and risk of
icantly correlated with one another
[39].
IL-8 belongs
schizophrenia. Hence, we administered a psychiatric
to a subclass of the cytokine superfamily known
diagnostic assessment on members of this cohort duras chemokines
[42, 43]
and is especially important
ing early-middle adulthood
[27].
We found that the
for neutrophil attraction and activation
[42, 44, 45].