MoSG

Beio

MdPo

Rpo

CT

Kc

IattrsBtr

htmio(chsT

F

A

R

0d

aternal Exposure to Herpes Simplex Virus and Riskf Psychosis Among Adult Offspring

tephen L. Buka, Tyrone D. Cannon, E. Fuller Torrey, Robert H. Yolken, and the Collaborative Studyroup on the Perinatal Origins of Severe Psychiatric Disorders

ackground: Viral exposure during gestation is thought to be a risk factor for schizophrenia. Previous studies have indicated that prenatalxposure to herpes simplex virus type 2 (HSV-2) may be a risk for the subsequent development of schizophrenia in some populations. In this

nvestigation, we tested a large and diverse population to assess the risk of psychoses among offspring of mothers with serological evidencef HSV-2 infection.

ethods: We conducted a nested case-control study of 200 adults with psychoses and 554 matched control subjects (matched for city andate of birth, race/ethnicity, gender, and parent history of treatment for mental disorder) from three cohorts of the Collaborative Perinatalroject (Boston, Providence, and Philadelphia). We analyzed stored serum samples that had been obtained from these mothers at the endf pregnancy for antibodies directed at HSV-2, using type-specific solid-phase enzyme immunoassay techniques.

esults: Offspring of mothers with serologic evidence of HSV-2 infection were at significantly increased risk for the development ofsychoses (odds ratio [OR] � 1.6; 95% confidence interval [CI] � 1.1–2.3). This risk was particularly elevated among women with high ratesf sexual activity during pregnancy (OR � 2.6; 95% CI � 1.4 – 4.6).

onclusions: Maternal exposure to herpes simplex virus type 2 is associated with an increased risk for psychoses among adult offspring.

hese results are consistent with a general model of risk resulting from enhanced maternal immune activation during pregnancy.

ey Words: Herpes simplex virus, pregnancy complications, psy-hoses, schizophrenia, viral exposure

n recent years, the search for the origins of schizophreniaand related psychotic illnesses has focused on two generaldomains of potential etiologic significance. Family, twin, and

doption studies have demonstrated a genetic component tohese pervasive neuropsychiatric disorders (1). First-degree rela-ives of patients with schizophrenia are at 7 to 10 times greaterisk for schizophrenia than the general population. However,pecific genes of large effect have not yet been identified (2).ecause the concordance rate for schizophrenia among identical

wins is less than 50%, environmental factors must also play aole in the pathogenesis of these disorders (3).

Among the multiple potential environmental risk factors thatave been examined, a number of investigations have indicatedhat exposures during the perinatal or early childhood periodsay represent a second general domain of risk. These have

ncluded winter/spring birth (4), birth in an urban area (5),bstetrical complications (6), extreme famine during pregnancy7,8), as well as infections during pregnancy (9 –11) and earlyhildhood (12). Most previous investigations of infectious agentsave involved ecologic designs, for example, examining rates ofchizophrenia several years after the 1957 A2 influenza epidemic.he initial study on this topic indicated increased rates of

rom the Brown University Department of Community Health (SLB), Provi-dence, Rhode Island; Harvard School of Public Health (SLB), Boston,Massachusetts; Departments of Psychology, Psychiatry and Biobehav-ioral Sciences, and Brain Research Institute (TDC), University of California(UCLA), Los Angeles, California; Stanley Medical Research Institute (EFT),Chevy Chase, Maryland; and Stanley Division of Developmental Neuro-virology (RHY), Johns Hopkins School of Medicine, Baltimore, Maryland.

ddress reprint requests to Stephen L. Buka, Sc.D., Brown University Depart-ment of Community Health, 121 South Main Street, Providence, RI02806; E-mail: Stephen_Buka@brown.edu.

eceived May 25, 2007; revised August 29, 2007; accepted September 12,

2007.

006-3223/08/$34.00oi:10.1016/j.biopsych.2007.09.022

schizophrenia among cohorts exposed to this epidemic duringthe second trimester of pregnancy (13), followed by bothpositive (14) and negative replications (15). Other infectiousagents that have been reported as associated with psychosesinclude measles virus (16), varicella zoster virus (16), Toxo-plasma gondii (17), rubella virus (18), and the organisms thatcause viral or bacterial meningitis (19). As a whole, this body ofresearch suggests that prenatal and perinatal infections and otherenvironmental insults that adversely affect infant brain develop-ment may increase the likelihood of psychosis in later life, mostlikely in genetically susceptible individuals (20,21).

Despite these interesting leads, there is a relative paucity ofstudies that have investigated the potential link between specificviral exposures during gestation and later schizophrenia usingserological assays. This is understandable given the logisticaldifficulties in assessing prenatal serum levels in a subject popu-lation 20 to 30 years before their adult diagnostic outcome isknown. One context for pursuing this link is provided by the U.S.National Collaborative Perinatal Project (NCPP), which enrolledand prospectively followed more than 55,000 pregnancies be-tween 1959 and 1966, with stored maternal serum samples. In aprevious investigation of 27 individuals with mixed diagnoses ofpsychotic disorders and 54 control subjects, we reported thatadult offspring with any form of psychotic illness had signifi-cantly elevated maternal levels of immunoglobulin G (IgG) andimmunoglobulin M (IgM) class total immunoglobulins prior todelivery and, in particular, maternal antibodies to herpes simplexvirus type 2 (HSV-2) (11). Further, elevated cytokine levelsduring pregnancy, in particular maternal levels of tumor necrosisfactor (TNF)-alpha, also predicted psychosis among adult off-spring (22).

Parallel work with a California-based prenatal cohort (23)with a sample of approximately 70 cases of psychosis has alsoprovided serologic evidence of increased immune activity duringpregnancy in relation to subsequent psychoses, as reflected bysignificantly elevated levels of the cytokine interleukin-8 (24).

Specific infectious agents identified with this sample have in-

BIOL PSYCHIATRY 2008;63:809–815© 2008 Society of Biological Psychiatry

coededfaif

ffHolteitw

M

asopinrpsm�

C

towssstapwV71roisapsotit(

810 BIOL PSYCHIATRY 2008;63:809–815 S.L. Buka et al.

w

luded influenza (25) and Toxoplasma gondii (26). In contrast tour study, there was no observed association between maternalxposure to HSV-2 and psychosis in this cohort (27). Theseiscrepant results may be due to population differences inxposure and susceptibility to HSV-2 and/or differences in theiagnostic composition of the study samples. In particular, riskor exposure and re-exposure to HSV-2 has been shown to bessociated with a woman’s sexual history and behavior, includ-ng number of sexual partners, age at first sexual intercourse,requency of intercourse, and contraceptive practices (28,29).

The current investigation combines prenatal serological andollow-up diagnostic information on three NCPP cohorts tourther investigate the hypothesis that maternal exposure toSV-2 is associated with the development of schizophrenia andther psychoses, with the largest sample available to date. Inight of the overall pattern of results from prior work in this area,he study had two primary aims: 1) to examine whether serologicvidence of maternal HSV-2 infection was associated with anncreased risk for psychoses among offspring; and 2) whetherhis risk was particularly elevated among offspring of pregnantomen at increased risk for re-exposure to HSV-2 infection.

ethods and Materials

The study sample was drawn from the Boston, Providence,nd Philadelphia cohorts of the NCPP. The NCPP was a large-cale, prospective, multisite study of pregnant women and theirffspring, selected to be representative of patients receivingrenatal care at each study site (30). The combined cohortsncluded 25,025 surviving offspring of a sample of 19,471 preg-ant women. Data from examinations and interviews wereecorded by trained staff beginning at the time of registration forrenatal care, using standardized protocols. Maternal serumamples were collected at registration, approximately every 2onths following registration, and at delivery and stored at20°C.

ase SeriesDue to the great cost implications, it was not feasible to follow

he entire cohort of offspring into adulthood to determine onsetf psychotic illness. Accordingly, a nested case-control approachas employed. In both the New England and Philadelphia study

ites, offspring with psychoses were identified through a two-tage diagnostic assessment procedure. For New England, intage 1, subjects with possible psychotic illness were identifiedhrough personal interviews and/or record linkage with psychi-tric treatment facilities. In a first interview study, as describedreviously (30), 928 subjects were selected for follow-up and 693ere interviewed (75%) using the Diagnostic Interview Scheduleersion III (31). In a second study (32), we selected an additional75 subjects for follow-up and interviewed 574 (74%). In total,267 follow-up interviews were completed, of which 29 subjectseported clinically relevant psychotic symptoms. Through anngoing study of offspring of parents with and without mentalllness, we completed an additional 350 interviews, of which 30ubjects reported psychotic symptoms (33). Finally, record link-ge efforts identified a final group of 80 subjects with a history ofsychiatric treatment for a psychotic illness. In total, this firsttage identified 139 subjects who reported psychotic symptomsr whose treatment records indicated a psychotic disorder. Forhe second stage, these 139 subjects were recontacted andnterviewed by a trained diagnostic interviewer using the Struc-ured Clinical Interview for DSM-IV (34). Trained diagnosticians

Ph.D. and M.D. level) then completed best-estimate consensus

ww.sobp.org/journal

diagnoses according to DSM-IV criteria, based on interview dataand medical chart review. Diagnostic interviews were completedfor 89 subjects and medical charts alone were available for theremaining 50 subjects. These yielded 102 confirmed diagnoses ofpsychoses of which 89 newly identified cases, not previouslyanalyzed, were included for the current study. These included 39schizophrenic psychoses, including 33 schizophrenia and 6schizoaffective disorder-depressed type; 7 other nonaffectivepsychoses; and 43 affective psychoses, including 14 schizoaffec-tive disorder-bipolar type, 22 bipolar disorder with psychoticfeatures, 4 major depressive disorder with psychosis, and 3 typenot specified.

For Philadelphia, stage 1 screening utilized a citywide psychi-atric database that identified 339 study offspring who had beentreated and diagnosed with some form of psychotic disorder.Psychiatric records were located and reviewed for 144 of theseindividuals. Six experienced diagnosticians performed the chartreviews. Diagnostic procedures followed DSM-IV criteria. Of the144 charts reviewed, 111 individuals received a confirmed diag-nosis of schizophrenia or other psychoses. These included 69schizophrenic psychoses, including 60 schizophrenia and 9schizoaffective disorder-depressed type, and 42 affective psycho-ses, including 7 schizoaffective disorder-bipolar type, 11 bipolardisorder with psychotic features, 21 major depressive disorderwith psychosis, and 3 type not specified. The reliability ofdiagnoses was assessed at both sites by having records indepen-dently diagnosed by two experienced clinicians. The interraterreliability for primary diagnoses at the New England site was100% and at the Philadelphia site, 93%.

Control SeriesFor each of the 200 subjects with psychoses, 3 control subjects

were selected, matched for study site, gender, race/ethnicity,date of birth, and parental history of treatment for mentaldisorder. Treatment history for mental disorder was based onmaternal report during the original cohort study that the child’smother or father “had any nervous problem which requiredhospital care or psychiatric treatment.” Control subjects wereselected from the entire surviving study cohort (n � 25,025), withthe exception of 54 subjects from New England that served ascontrol subjects in our previous report on this topic (11). Withineach study site, cases and control subjects were grouped intoeight strata (gender [male/female] � race/ethnicity [Caucasian/African American] � parental history of mental illness [yes/no])and within strata sorted by month of birth. Potential controlsubjects closest for month of birth within each stratum wereselected. Human subjects approval was granted by reviewgroups at Harvard University; the University of Pennsylvania; theCity of Philadelphia; the University of California, Los Angeles;National Institute of Child Health and Human Development(NICHD); the Johns Hopkins School of Medicine; and localpsychiatric facilities. Written consent was obtained from allinterviewed study participants.

Processing of Sera and Antibody MeasurementsFor each study participant, a maternal serum sample was

obtained from the National Institutes of Health (NIH) studyrepository for the last specimen collected during pregnancy(usually on the day of delivery). These late pregnancy sampleswere available for over 98% of study participants. Levels ofspecific IgG class antibodies to surface glycoproteins of HSV-2were measured by solid-phase enzyme immunoassay using

previously described methods (11,35). Assay kits for these mea-

sCasaamuu

S

tcmgpcomaerctteeceucaidrga“

1

T

SMCWMAM

M

a

r at ag

S.L. Buka et al. BIOL PSYCHIATRY 2008;63:809–815 811

urements were purchased from Focus Technologies, Cyprus,alifornia. Sera for matched sets of cases and control subjects,long with standard control samples, were run on each plate. Aample was considered to be seropositive for HSV-2 if it gener-ted a sample/control ratio �1.1 (36). This cutoff value was used,s this has been shown to correlate with other establishedeasures of HSV-2 infection (37). All samples were analyzednder code with the laboratory performing the studies beingnaware of the clinical status of the study individuals.

tatistical AnalysisTo test our primary study hypothesis, we conducted analyses

hat took into account the matching of approximately threeontrol subjects per case to test for differences in rates ofaternal HSV-2 seropositivity between the case and controlroups. For this, we utilized the categorical definitions of sero-ositivity defined above (sample/control ratio � 1.1). Usingonditional logistic regression models, we calculated rates anddds ratios of seropositivity to HSV-2 for cases compared withatched control subjects, treating each matched set of one case

nd matched control subjects as an individual stratum or randomffect (38). We tested for potential effect modification by bothepeating these analyses for subgroups (e.g., for African Ameri-an and Caucasian samples) separately and by modifications tohe above models that permitted us to simultaneously estimatehe effects for each subgroup and to test for subgroup differ-nces. Effect modification was examined according to race/thnicity, diagnosis (schizophrenic psychoses vs. affective psy-hoses),1 and two variables reflecting likelihood of HSV-2xposure–frequency of sexual intercourse during pregnancy andse of contraceptives prior to conception. Frequency of inter-ourse during pregnancy was obtained from a maternal interviewdministered at the first prenatal visit. The interviewer wasnstructed to ask, “How many times have you had intercourseuring the past month?” We created a categorical variableeflecting high frequency of intercourse (5� times per month orreater than once per week). Questions on contraception werelso administered at the first prenatal visit. Mothers were asked,Did you usually use a contraceptive” during the period prior to

Seven cases classified as other nonaffective psychoses were omitted

able 1. Demographic and Clinical Characteristics of Cases and Control Su

Cases(n � 200)

n %

tudy Site (Philadelphia) 111 55.5ale 115 57.5

aucasian 75 37.5inter/Spring Birth (December–April) 75 37.5other Married 157 78.5ny Maternal or Paternal History of Mental Illnessb 33 16.7other Completed High School 74 37.6

Mean (SD)

aternal Age at Delivery (years) 24.7 (6.4)

Categorical variables analyzed by chi-square, quantitative variables analnd standard deviations of study variables for the entire surviving New Eng

aDegrees of freedom � 1.bEither parent reporting mental health treatment at study enrollment o

from this analysis due to limited sample size.

conception? Women reporting a high level of intercourse withouta history of contraceptive use were considered at greatest risk forexposure or re-exposure to HSV-2.

Results

Table 1 summarizes demographic characteristics for the 200subjects with psychoses and the 544 matched control subjects.There were no significant differences between the two groupsfor any of the sample characteristics examined. These include thefive matching variables (study site, race/ethnicity, date of birth,gender, and parent history of treatment for mental disorder) andseveral other potential confounding factors (e.g., maternal mar-ital status, educational attainment, and age at delivery). As caseshad slightly higher levels of parental treatment for mentaldisorder and lower levels of maternal educational attainment,these variables were included as covariates in subsequent anal-yses. The case series was largely comparable with the entiresurviving cohort of 25,025 subjects, from which the sample wasdrawn with regard to season of birth, maternal marital status, andmaternal age. Just over half of the cases were from the Philadel-phia site (n � 111) and accordingly, were more likely to beAfrican American and of lower maternal education than the studycohort as a whole.

The results of the assays for maternal antibodies to HSV-2 areshown in Table 2. Among the 200 subjects with psychoses, 62mothers (31.5%) were seropositive for HSV-2, in contrast to23.7% of the control subjects. This corresponds to a relative riskof subsequent psychoses associated with HSV-2 seropositivity of1.6 (95% confidence interval [CI] � 1.1–2.3, p � .03). A statisti-cally significant elevated risk for subsequent psychoses associ-ated with maternal exposure to HSV-2 was observed for theAfrican American sample only, with an odds ratio (OR) of 1.6,95% CI � 1.0–2.4, p � .04. A comparable odds ratio (1.4) wasobserved for the Caucasian sample; however, this was notstatistically significant due to both limited sample size and theconsiderably lower rate of seropositivity among this group. In aformal test for interaction, the odds ratios for African Americansversus Caucasians did not differ significantly. Similarly, a statis-tically significant elevated risk for subsequent psychoses associ-ated with maternal exposure to HSV-2 was observed for thesample with schizophrenic psychoses (and their matched control

Random ControlSubjects(n � 544)

Chi-Squarea p

Total Sample(n � 25,025)

n % %

302 55.5 .01 .96 37.2316 58.1 .02 .88 50.4192 35.3 .29 .58 57.6200 36.8 .04 .87 39.9413 75.9 .54 .46 80.2

68 12.8 1.83 .16 9.6235 43.9 2.45 .12 48.0

Mean (SD) t test p

24.4 (5.8) .57 .57 24.6 (6.0)

by t test (cases vs. control subjects). Total Sample refers to the rates, means,and Philadelphia Collaborative Perinatal Project cohorts.

e 7 follow-up.

bjects

yzedland

subjects) (OR � 1.8, 95% CI � 1.1–3.0, p � .02) but not for those

www.sobp.org/journal

wasd

sultHw1eosslriiwh

crTr

D

s

T

AAWSA

a

T

P

F

a

812 BIOL PSYCHIATRY 2008;63:809–815 S.L. Buka et al.

w

ith affective psychosis (OR � 1.3, 95% CI � .7–2.4). Again, informal test for interaction, the odds ratios associated with

chizophrenic psychoses versus affective psychoses did notiffer significantly.

As shown in Table 3, there was no association between HSV-2eropositivity and offspring psychoses among subjects whosually used contraception prior to conception and those withower frequency of intercourse (�5 times per month). In con-rast, the odds ratio for subsequent psychoses associated withSV-2 seropositivity was 2.0 (95% CI � 1.2–3.2) among subjectsho did not regularly use contraception and 2.6 (95% CI �.4–4.6) among those with frequent intercourse. There wasvidence of statistical effect modification in relation to frequencyf intercourse but not for use of contraception. The odds ofubsequent psychosis associated with HSV-2 seropositivity weretatistically greater (p � .03) among those with higher versusower rates of intercourse. The strength of association (oddsatios) between HSV-2 seropositivity and offspring psychosesncreased from a low of .7 for those with a low frequency ofntercourse with a history of contraceptive use to 4.3 for thoseith a high frequency of intercourse during pregnancy without aistory of contraceptive use.

All of the analyses reported in Tables 2 and 3 that examinedategorical cut-points reflecting seropositivity to HSV-2 wereeconducted using the quantitative data on HSV-2 IgG values.he results of these analyses were consistent with the categoricalesults.

iscussion

In the largest case series to date, these study results demon-trate a statistically significant association between serologic

able 2. Rates of HSV-2 Seropositivity Among Cases and Control Mothers a

Cases

TotalSample

Seropositivea

n %

lla 200 62 31.5frican American 123 54 43.9hite 77 8 10.4

chizophrenic Psychoses 108 39 36.1ffective Psychoses 85 21 24.7

CI, confidence interval; HSV-2, herpes simplex virus 2; OR, odds ratio.aCutoff � 1.1 for plate-specific value.bMatched for city of birth, season of birth, gender, race/ethnicity, and pa

nd parental treatment for mental disorder.

able 3. Potential Effect Modification by Risks for Re-exposure to HSV-2; Ra

Cases

TotalSample

Seropo

n

rior Use of ContraceptionYes 89 27No 111 35

requency of Intercourse (at Enrollment)Not high (�5/month) 106 28High (5�/month) 88 31

CI, confidence interval; HSV-2, herpes simplex virus 2; OR, odds ratio.aCutoff � 1.1 for plate-specific value.bMatched for city of birth, season of birth, gender, race/ethnicity, and pa

nd parental treatment for mental disorder.

ww.sobp.org/journal

evidence of HSV-2 infection and the subsequent development ofschizophrenia and other psychoses among adult offspring. Thisrisk was slightly, but not significantly, elevated among AfricanAmericans and in relation to schizophrenic rather than affectivepsychoses. These results parallel our previous investigation ofthis topic (11), which also found a significant association be-tween maternal seropositivity to HSV-2 and offspring psychosis,using a separate sample of subjects drawn from the NationalCollaborative Perinatal Project. These effect sizes are comparablewith those reported for specific genetic polymorphisms that havebeen linked to schizophrenia (39).

In the only other investigation of this topic using storedmaternal sera, Brown et al. (27) reported no association betweenHSV-2 and offspring psychoses. Differences in sample demo-graphics, diagnostic composition, and limited sample size maypartially explain these seemingly divergent results. For example,in the current investigation, the association between HSV-2exposure and subsequent psychoses was more pronouncedamong African American than Caucasian subjects. This maylargely be accounted for by the greater prevalence of HSV-2seropositivity among African Americans in the United States (40)and in the current sample. With a base rate of HSV-2 seroposi-tivity of only 7.2%, the power to detect differences betweenCaucasian cases and control subjects in the current sample andwith other predominantly white samples is limited.

Serum samples for this investigation were drawn at the end ofpregnancy, typically as the mother was admitted for labor anddelivery. The available assay methods (for IgG antibodies) areable to detect a prior history of exposure to HSV-2 but not currentexposure or reactivation of the latent virus. The overall preva-lence of HSV-2 infection in the population was substantially

iveryb

Control Subjects

OR (95% CI)Total

Sample

Seropositivea

n %

550 134 23.7 1.6 (1.1, 2.3)350 111 31.7 1.6 (1.0, 2.4)194 14 7.2 1.4 (.6, 3.7)303 75 24.8 1.8 (1.1, 3.0)226 46 20.4 1.3 (.7, 2.4)

l history of treatment for mental disorder. Adjusted for maternal education

f HSV-2 Seropositivity Among Case and Control Mothers at Deliveryb

Control Subjects

OR (95% CI)

a

TotalSample

Seropositivea

n %

.3 244 67 27.5 1.1 (.6, 1.9)

.5 300 58 19.3 2.0 (1.2, 3.2)

.4 326 89 27.3 1.0 (.6, 1.6)

.2 195 34 17.4 2.6 (1.4, 4.6)

l history of treatment for mental disorder. Adjusted for maternal education

t Del

renta

tes o

sitive

%

3031

2635

renta

hefIipnpswiFwfocptmHapccdfarsaohwvvmdlst

tt(tsppbsnsuHHdpiarovo

S.L. Buka et al. BIOL PSYCHIATRY 2008;63:809–815 813

igher than the rate of psychosis. It is thus unlikely that a pastxposure to HSV-2 would, by itself, impact on the developingetus in a fashion that could result in subsequent psychoses.nstead, we hypothesized that those women with prior HSV-2nfection (as evidenced by elevated IgG antibodies in lateregnancy) who were also exposed to conditions during preg-ancy that could result in re-exposure to the virus might be atarticularly elevated risk for psychosis among surviving off-pring. We identified two variables that have been associatedith or serve as proxies for re-exposure to HSV-2: frequency of

ntercourse (41) and failure to use barrier contraceptives (42).requency of intercourse was reported directly by the pregnantomen in mid pregnancy and provided the strongest evidence

or effect modification of the association between HSV-2 andffspring psychosis (OR � 2.6, 95% CI � 1.4–4.6). Use of barrierontraceptives during pregnancy was not assessed; however, aroxy measure of contraceptive practices (history of contracep-ive use prior to pregnancy) showed modest evidence of effectodification (OR � 2.0, 95% CI � 1.2–3.2). Of note, offspring ofSV-2 positive women with a history of prior contraceptive usend who reported a lower frequency of intercourse duringregnancy were at no elevated risk for subsequent psychoses. Inontrast, HSV-2 seropositive women without a history of contra-eptive use and who reported a high frequency of intercourseuring pregnancy were at approximately fourfold increased riskor offspring psychoses. This pattern of results was observed forll race/ethnic groups and study sites (results not shown). Theseesults are consistent with a previous report that prenatal expo-ure to maternal genital and reproductive infections was associ-ted with an increased risk for schizophrenia among adultffspring (43). While we interpret these findings in line with theypothesis that women with a history of HSV-2 infection andho are re-exposed during pregnancy are at particularly ele-ated risk for offspring psychosis, this interpretation must beiewed as speculative both due to the imprecise nature of theeasures of sexual practices during pregnancy and the lack ofirect evidence of re-exposure. It is also possible that otherifestyle issues or unmeasured variables that are associated withexual practices prior to and during pregnancy may account forhe observed effect modification.

There are several possible pathophysiological pathwayshrough which maternal exposure to HSV-2 infection might leado the subsequent development of psychosis among offspring11). It is possible that some infants were directly infected withhe herpes virus during pregnancy or delivery. Since samplesuitable for viral culture or DNA analyses were not collected asart of the study, the rate and timing of viral shedding cannot berecisely determined (44). However, the potential relationshipetween herpes infection and subsequent schizophrenia is plau-ible in light of the tropism of HSV-2 for the infant centralervous system (45). While it is also possible that some HSV-2eropositive mothers infected their children after birth, this isnlikely in light of the low incidence of postnatal infection ofSV-2 during childhood (46). It is also possible that antibodies toSV-2 could cross the placenta and directly cause damage to theeveloping brain by means of molecular mimicry (47). Theossibility that antibodies to HSV-2 are solely an indicator of

ncreased maternal sexual activity or exposure to another sexu-lly transmitted disease which could adversely affect the fetus isendered unlikely by our previous null findings for antibodies tother sexually transmitted pathogens such as human papillomairus (48). Similarly, there was no association between frequency

f intercourse or contraceptive methods and offspring psychoses

among HSV-2 negative women in the current sample. Finally,the results may also be consistent with an epigenetic mechanism,with genetic alterations secondary to the virus or possiblysecondary to other exogenous exposures resulting from in-creased heterogeneity of sexual partners. While plausible, spe-cific epigenetic mechanisms of this type have yet to be identified.

These results add to the set of infectious conditions duringpregnancy that have been found to be associated with risk foroffspring psychoses (e.g., influenza, toxoplasmosis) as well asrisks involving largely unestablished mechanisms that may reflectinfectious processes (e.g., urbanicity and season of birth). Giventhis long and varied list, it is unlikely that any specific infectiousagent acts as an independent teratogen affecting the developingfetal brain. We speculate that the pathophysiological processunderlying the current results may not be specific to the herpesvirus, per se, but rather may result from general enhancedmaternal immune activation, which has been shown in animalsto result in abnormalities of both dopaminergic activity andcognitive function in the offspring, consistent with deficits ob-served in patients with schizophrenia (49). This interpretation isconsistent with our previous findings (22) and others (24) ofelevated prenatal cytokine levels among psychotic offspring.Similarly, animal work has demonstrated decreased neuronalsurvival in rat cortical cell cultures exposed to high levels ofproinflammatory cytokines, consistent with this model (50).Future reports from this team will address this general model,with estimates of the extent to which cytokine levels and othermarkers of general maternal immune activation may account forthe present results.

The 200 subjects with schizophrenia represent .80% of the fullstudy cohort (n � 25,025) and likely do not include all affectedsubjects. Case ascertainment involved personal screening inter-views and record linkage with treatment facilities for the entiresample. The majority of cases (particularly among the Philadel-phia sample) were identified through a recorded history of localpsychiatric service utilization, which missed cases who weredeceased or had not yet come to treatment at the time offollow-up, who had been treated at facilities that were no longeroperating or did not cooperate with the chart reviews, or who,because of emigration or changes in social class, utilized psychi-atric facilities other than those whose patient rolls were screened.Our use of a case-control (rather than a cohort analytic) designmitigates the potential biasing effects of incomplete ascertain-ment to the extent that the cases used in the study are represen-tative of the overall pool of affected individuals in the birthcohort. One indication that the cases included in this study arelikely to be representative is that they were found to be compa-rable in terms of demographics and obstetric history with the 121probands whose charts were not available for review and werethus not included in the current analyses. Of note, the controlsample for this study included all remaining members of thecohort, not simply unaffected control subjects. This helps miti-gate the concern that the observed differences between casesand control subjects could be attributable to patterns of HSV-2antibodies in healthy control subjects.

Due to changes in maiden names, men are somewhat over-represented in the linkage sample. Participants of lower eco-nomic status are likely to be overrepresented in publicly fundedtreatment facilities (included in the record linkage) and personsof higher socioeconomic status are somewhat more likely tohave continued participation in the longitudinal study andcompleted personal screening interviews as adults. However, the

tight matching of cases to control subjects, along with additional

www.sobp.org/journal

sactiws

osTRewtmmohcc

NpF

tcrMaB

goGMPt

t

814 BIOL PSYCHIATRY 2008;63:809–815 S.L. Buka et al.

w

tatistical adjustment for additional family socioeconomic char-cteristics, suggest that the major study results are unlikely to beonfounded due to case ascertainment procedures. It may be thathe risk for subsequent schizophrenia associated with maternalnfection with HSV-2 may be restricted to a subset of personsith psychoses, namely those who were identified with our

tudy procedures; however, this seems unlikely.In conclusion, the current investigation replicates our previ-

us finding that maternal exposure to HSV-2 is a risk factor forubsequent schizophrenia and other psychoses among offspring.his finding has potential clinical and public health implications.ecent clinical trials suggest that the glycoprotein D vaccine hasfficacy for the primary prevention of HSV-2, especially amongomen who are seronegative for both herpes simplex virus

ype 1 (HSV-1) and HSV-2 (51). Similarly, screening, heightenedonitoring, and preventive interventions involving contraceptiveethods for women with a known history of exposure to HSV-2r other high-risk groups may be warranted. Finally, maternalistory of HSV-2 exposure may assist in developing targetedlinical and medication strategies for adult patients with psy-hotic illness.

Supported by the Stanley Medical Research Institute and theational Institute of Mental Health with additional supportrovided by Garen and Sharilyn Staglin and the Staglin Musicestival for Mental Health.

We are indebted to the support and ongoing participation ofhe members of the New England Family Study from which theurrent sample was drawn and the dedication and efforts of ouresearch staff. In particular, we thank Jo-Ann Donatelli, KathycGaffigan, Anne Peters, and Mikhail Salganik for data man-gement and analysis and Ann Cusic, Inna Ruslanova, andogdana Krivogorsky for laboratory technical support.

In addition to the authors, the following persons are investi-ators in the Collaborative Study Group on the Perinatal Originsf Severe Psychiatric Disorders and participated in this study: J.oldstein, M. Salganik, L. Seidman, and M. Tsuang, Harvardedical School; T. Hadley, C. Bearden, and I. Rosso, University ofennsylvania; and D. Dhavale, Stanley Medical Research Insti-

ute.The authors report no biomedical financial interests or po-

ential conflicts of interest.

1. Faraone SV, Matise T, Svrakic D, Pepple J, Malaspina D, Suarez B, et al.(1998): Genome scan of European-American schizophrenia pedigrees:Results of the NIMH Genetics Initiative and Millennium Consortium.Am J Med Genet 81:290 –295.

2. Moldin SO (1997): The maddening hunt for madness genes. Nat Genet17:127–129.

3. Tsuang MT, Faraone SV (1995): The case for heterogeneity in the etiol-ogy of schizophrenia. Schizophr Res 17:161–175.

4. Torrey EF, Miller J, Rawlings R, Yolken RH (1997): Seasonality of births inschizophrenia and bipolar disorder: A review of the literature. SchizophrRes 28:1–38.

5. Mortensen PB, Pedersen CB, Westergaard T, Wohlfahrt J, Ewald H, MorsO, et al. (1999): Effects of family history and place and season of birth onthe risk of schizophrenia. N Engl J Med 340:603– 608.

6. Cannon M, Jones PB, Murray RM (2002): Obstetric complications andschizophrenia: Historical and meta-analytic review. Am J Psychiatry 159:1080 –1092.

7. St. Clair D, Xu M, Wang P (2005): Rates of adult schizophrenia followingprenatal exposure to the Chinese famine of 1959 –1961. JAMA 294:557–562.

8. Susser E, Hoek HW, Brown A (1998): Neurodevelopmental disordersafter prenatal famine: The story of the Dutch Famine Study. Am J Epide-

miol 147:213–216.

ww.sobp.org/journal

9. Brown AS, Cohen P, Harkavy-Friedman J, Babulas V, Malaspina D, Gor-man JM, Susser ES (2001): A.E. Bennett Research Award. Prenatal rubella,premorbid abnormalities, and adult schizophrenia. Biol Psychiatry 49:473– 486.

10. Brown AS, Begg MD, Gravenstein S, Schaefer CA, Wyatt RJ, Bresnahan M,et al. (2004): Serologic evidence of prenatal influenza in the etiology ofschizophrenia. Arch Gen Psychiatry 61:774 –780.

11. Buka SL, Tsuang MT, Torrey EF, Klebanoff MA, Bernstein D, Yolken RH(2001): Maternal infections and subsequent psychosis among offspring.Arch Gen Psychiatry 58:1032–1037.

12. Rantakallio P, Jones P, Moring J, Von Wendt L (1997): Association be-tween central nervous system infections during childhood and adultonset schizophrenia and other psychoses: A 28-year follow-up. Int JEpidemiol 26:837– 843.

13. Mednick SA, Huttunen MO, Machon RA (1994): Prenatal influenza infec-tions and adult schizophrenia. Schizophr Bull 20:263–267.

14. Limosin F, Rouillon F, Payan C, Cohen JM, Strub N (2003): Prenatalexposure to influenza as a risk factor for adult schizophrenia. Acta Psy-chiatr Scand 107:331–335.

15. Crow TJ, Done DJ (1992): Prenatal exposure to influenza does not causeschizophrenia. Br J Psychiatry 161:390 –393.

16. Torrey EF, Rawlings R, Waldman IN (1988): Schizophrenic births and viraldiseases in two states. Schizophr Res 1:73–77.

17. Torrey EF, Bartko J, Lun ZR, Yolken RH (2006): Antibodies to Toxoplasmagondii in patients with schizophrenia: A meta-analysis. Schizophr Bull33:729 –736.

18. Brown AS, Susser ES (2002): In utero infection and adult schizophrenia.Ment Retard Dev Disabil Res Rev 8:51–57.

19. Abrahao A, Focaccia R, Gattaz W (2005): Childhood meningitis increasesthe risk for adult schizophrenia. World J Biol Psychiatry 6:44 – 48.

20. Torrey EF, Yolken RH (2000): Familial and genetic mechanisms in schizo-phrenia. Brain Res Rev 31:113–117.

21. Weinberger DR (1995): From neuropathology to neurodevelopment.Lancet 346:552–557.

22. Buka SL, Tsuang MT, Torrey EF, Klebanoff MA, Wagner RL, Yolken RH(2001): Maternal cytokine levels during pregnancy and adult psychosis.Brain Behav Immun 15:411– 420.

23. Susser ES, Schaefer CA, Brown AS, Begg MD, Wyatt RJ (2000): The designof the prenatal determinants of schizophrenia study. Schizophr Bull26:257–273.

24. Brown AS, Hooton J, Schaefer CA, Zhang H, Petkova E, Babulas V, et al.(2004): Elevated maternal interleukin-8 levels and risk of schizophreniain adult offspring. Am J Psychiatry 161:889 – 895.

25. Brown AS, Begg MD, Gravenstein S, Schaefer CA, Wyatt RJ, Bresnahan M,et al. (2004): Serologic evidence of prenatal influenza in the etiology ofschizophrenia. Arch Gen Psychiatry 61:774 –780.

26. Brown AS, Schaefer CA, Quesenberry CP Jr, Liu L, Babulas VP, Susser ES(2005): Maternal exposure to toxoplasmosis and risk of schizophrenia inadult offspring. Am J Psychiatry 162:767–773.

27. Brown AS, Schaefer CA, Quesenberry CP Jr, Shen L, Susser ES (2006): Noevidence of relation between maternal exposure to herpes simplexvirus type 2 and risk of schizophrenia. Am J Psychiatry 163:2178 –2180.

28. Frenkel LM, Garratty EM, Shen JP, Wheeler N, Clark O, Bryson YJ (1993):Clinical reactivation of herpes simplex virus type 2 infection in seropos-itive pregnant women with no history of genital herpes. Ann Intern Med118:414 – 418.

29. Brown ZA, Benedetti J, Ashley R, Burchett S, Selke S, Berry S, et al. (1991):Neonatal herpes simplex virus infection in relation to asymptomaticmaternal infection at the time of labor. N Engl J Med 324:1247–1252.

30. Buka SL, Tsuang MT, Lipsitt LP (1993): Pregnancy/delivery complicationsand psychiatric diagnosis. A prospective study. Arch Gen Psychiatry 50:151–156.

31. Robins LN, Helzer JE, Croughanj, Ratcliff KS (1981): The NIMH DiagnosticInterview Schedule: Its history, characteristics and validity. Arch GenPsychiatry 38:381–389.

32. Buka SL, Satz P, Seidman L, Lipsitt L (1998): Defining learning disabilities:The role of longitudinal studies. Thalamus Relat Syst 16:14 –29.

33. Buka SL, Goldstein JM, Seidman LJ, Zornberg G, Donatelli JA, Denny LR,Tsuang MT (1999): Prenatal complications, genetic vulnerability, andschizophrenia: The New England Longitudinal Studies of Schizophre-

nia. Psychiatr Ann 29:151–156.

3

3

3

3

3

3

4

4

4

4

S.L. Buka et al. BIOL PSYCHIATRY 2008;63:809–815 815

4. First MB, Spitzer RL, Gibbon M, Williams JB (1996): Structured ClinicalInterview for DSM-IV Axis I Disorders–Patient Edition (SCID-I/P, Version 2.0).Washington, DC: American Psychiatric Press.

5. Jones-Brando LV, Yolken RH (1997): Laboratory diagnosis of viral infec-tions. In: Galasso GJ, Whitley RJ, Merigan TC, editors. Antiviral Agents andHuman Viral Diseases, 4th ed. Philadelphia: Lippincott-Raven Publishers,129 –173.

6. Turner KR, Wong EH, Kent CK, Klausner JD (2002): Serologic herpestesting in the real world: Validation of new type-specific serologic her-pes simplex virus tests in a public health laboratory. Sex Transm Dis29:422– 425.

7. Wald A, Ashley-Morrow R (2002): Serological testing for herpes simplexvirus (HSV)-1 and HSV-2 infection. Clin Infect Dis 35(suppl 2):S173–S182.

8. Brown H, Prescott R (1999): Applied Mixed Models in Medicine. New York:John Wiley & Sons.

9. Gilbody S, Lewis S, Lightfoot T (2007): Methylenetetrahydrofolate re-ductase (MTHFR) genetic polymorphisms and psychiatric disorders: AHuGE review. Am J Epidemiol 165:1–13.

0. Kimberlin DW (2004): Neonatal herpes simplex infection. Clin MicrobiolRev 17:1–13.

1. Guinan ME, Wolinsky SM, Reichman RC (1985): Epidemiology of genitalherpes simplex virus infection. Epidemiol Rev 7:127–146.

2. Vontver LA, Hickok DE, Brown Z, Reid L, Corey L (1982): Recurrentgenital herpes simplex virus infection in pregnancy: Infant outcomeand frequency of asymptomatic recurrences. Am J Obstet Gynecol143:75– 84.

3. Babulas V, Factor-Litvak P, Goetz R, Schaefer CA, Brown AS (2006): Pre-

natal exposure to maternal genital and reproductive infections andadult schizophrenia. Am J Psychiatry 163:927–929.

44. Brown ZA, Selke S, Zeh J, Kopelman J, Maslow A, Ashley RL, et al. (1997):The acquisition of herpes simplex virus during pregnancy. N Engl J Med337:509 –515.

45. Lellouch-Tubiana A, Fohlen M, Robain O, Rozenberg F (2000): Immuno-cytochemical characterization of long-term persistent immune activa-tion in human brain after herpes simplex encephalitis. Neuropathol ApplNeurobiol 26:285–294.

46. Kaplan KM, Fleisher GR, Paradise JE, Friedman HN (1984): Social rele-vance of genital herpes simplex in children. Am J Dis Child 138:872– 874.

47. Nahmias AJ, Nahmias SB, Danielsson D (2006): The possible role oftransplacentally-acquired antibodies to infectious agents, with molec-ular mimicry to nervous system sialic acid epitopes, as causes of neuro-mental disorders: Prevention and vaccine implications. Clin Dev Immu-nol 13:167–183.

48. Dillner J, Kallings I, Brihmer C, Sikstrom B, Koskela P, Lehtinen M, et al.(1996): Seropositivities to human papillomavirus types 16, 18, or 33capsids and to Chlamydia trachomatis are markers of sexual behavior.J Infect Dis 173:1394 –1398.

49. Gilmore J, Jarskog LF, Vadlamudi S, Lauder J (2004): Prenatal infec-tion and risk for schizophrenia: IL-1beta, IL-6, and TNF-alpha inhibitcortical neuron dendrite development. Neuropsychopharmacology29:1221–1229.

50. Jarskog LF, Lauder JM, Lieberman JA, Gilmore JH (2001): Cytokine ef-fects on cortical neuron MAP-2 immunoreactivity: Implications forschizophrenia. Biol Psychiatry 50:743–749.

51. Stanberry LR, Spruance SL, Cunningham AL, Bernstein DI, Mindel A,Sacks S, et al. (2002): Glycoprotein-D-adjuvant vaccine to prevent geni-

tal herpes. N Engl J Med 347:1652–1661.

www.sobp.org/journal