Neural Tube Defects on the Texas-Mexico Border:What We’ve Learned in the 20 Years since the

Brownsville Cluster

Lucina Suarez,1,* Marilyn Felkner,1 Jean D. Brender,2 Mark Canfield,1 Huiping Zhu,3 andKatherine A. Hendricks4

1Prevention and Preparedness Division, Department of State Health Services, Austin, Texas2School of Rural Public Health, Texas A&M Health Science Center, College Station, Texas

3Dell Pediatric Research Institute, Department of Nutritional Services, University of Texas, Austin, Texas4Bacterial Special Pathogens Branch, Division of High Consequence Pathogens and Pathology, Centers for Disease Control and

Prevention, Atlanta, Georgia

Received 8 May 2012; Revised 11 July 2012; Accepted 15 July 2012

We reviewed the published findings from the Texas Neural Tube Defect Project, a 6-year case-control study(1995–2000) of neural tube defects (NTDs) on the Texas-Mexico border. In this review, we highlight whatwas learned about environmental, genetic, and nutritional factors (i.e., those related to the folate and othermetabolic pathways) and the novel putative risk factors that emerged from this study of Mexican Americanwomen living on the Texas-Mexico border. Our investigations of the micronutrients and metabolic pathwaysinvolved confirmed the findings of other researchers that increased folate intake has a protective effect andthat low serum B12, high serum homocysteine levels, and obesity independently contribute to risk. Studiesof this population also have implicated hyperinsulinemia and low ferritin, metabolic risk factors, whichrequire additional study to elucidate their physiologic mechanism. Environmental contaminants such asheavy metals, pesticides, and polychlorinated biphenyls (PCBs), which were of community concern, did littleto explain NTD risk. Studies in this folic acid deficit-population also revealed several novel risk factors,namely, diarrhea, stress, fumonisins, and the combination of nitrosatable drug exposure with high nitrate/nitrite intake. In conclusion, the 23 studies among the Mexican American women living along the Texas-Mexico border have demonstrated the multifactorial nature of NTDs and that a population deficient in folicacid will be vulnerable to a variety of insults whether brought on by individual behaviors (e.g., obesity) orthrough the surrounding environment (e.g., fumonisins). Birth Defects Research (Part A) 00:000–000,2012. � 2012 Wiley Periodicals, Inc.

Key words: neural tube defects; Mexican American; folic acid; fumonisins; environment; case-control study

INTRODUCTION

In April 1991, a physician in Brownsville, Texas,reported to the state health department that three anence-phalic babies had been delivered in one hospital within36 hours. Active surveillance revealed that, in fact, at thishospital, six babies with neural tube defects (NTDs) hadbeen delivered within 6 weeks. For a county with 6000births each year, two or fewer cases would have beenexpected in a year. Active surveillance of all 1991Cameron County (Brownsville, Texas) births showed theprevalence for that year to be 29 per 10,000. This was thehighest recorded prevalence in the United States sincethe 1970s (Suarez et al., 2000), and although comparable

to the then current prevalence in neighboring Mexico (36per 10,000), it was fourfold to fivefold higher than whatwas being reported elsewhere within the United States(�6 per 10,000) (Suarez et al., 2000).In response to this cluster, we implemented active sur-

veillance to identify NTD occurrences in all 14 Texascounties that bordered Mexico, launched a folic acid

The authors have no conflicts of interest to declare.

*Correspondence to: Lucina Suarez, Department of State Health Services,1100 West 49th Street, Austin, Texas 78756.E-mail: lucina.suarez@dshs.state.tx.usPublished online in Wiley Online Library (wileyonlinelibrary. com).DOI: 10.1002/bdra.23070

Birth Defects Research (Part A): Clinical and Molecular Teratology 00:000�000 (2012)

� 2012 Wiley Periodicals, Inc. Birth Defects Research (Part A) 00:000�000 (2012)

intervention to prevent recurrences, and commenced apopulation-based, case-control study to identify causesfor the high prevalence. At the time, people living in theborder communities expressed concern that the NTDswere caused by pollution; the hypotheses publiclyexpressed most often were agricultural pollution, indus-trial pollution from the unregulated industries (maquila-doras) immediately across the border, and the consump-tion of contaminated fish caught from nearby rivers.While none of these potential risks could be ignored,there were other credible risk factors related to theunique aspects of the border community. These includedthe potential for pesticide exposures from the widespreadagricultural industry, and the predominantly MexicanAmerican population (93%) with high rates of povertyand health conditions such as obesity and diabetes; acorn-based diet; substandard housing; and use of folkmedicines and unique access to over-the-counter pre-scription medications in Mexican pharmacies.

Active surveillance showed that the annual prevalenceof NTDs in the border counties, with the exception of ElPaso county (10 per 10,000), was 16 to 18 per 10,000 dur-ing 1993 to 1995, twice as high as in other areas of theUnited States (Hendricks et al., 1999). Surveillance alsoshowed high prevalence of craniorachischisis, a rareNTD subtype involving both the brain and spinal cord(0.5 per 10,000) (Johnson et al., 2004). In this article, wereview the published findings from the 6-year case-con-trol study (1995–2000), highlighting what was learnedabout environmental, genetic, and nutritional factors (i.e.,those related to the folate and other metabolic pathways)and the novel putative risk factors that emerged fromthis study of Mexican American women living on theTexas-Mexico border.

BACKGROUND

The multicomponent project (Texas Neural Tube DefectProject), including surveillance, a folic acid intervention,and a population-based case-control study, was primarilyfunded through a cooperative agreement with the Cen-ters for Disease Control and Prevention, and secondarilyby the State of Texas and the Environmental ProtectionAgency. The heightened public and political awarenessof the Brownsville anencephaly cluster in 1991 and thesubsequent discovery of the endemic high prevalence ofNTDs along the whole of the Texas-Mexico border weresignificant factors in providing for these funds. Multiplepublished studies from the Texas Neural Tube DefectProject describe the methods and materials which wesummarize here (Suarez et al., 2000; Hendricks et al.,2001; Brender et al., 2002; Felkner et al., 2003).

Methods included active surveillance of 21 hospitals,39 birthing centers, 4 genetic clinics, 74 ultrasound cen-ters, 4 abortion clinics, and 150 midwives to identifyanencephaly, spina bifida, and encephalocele defects.Considerable effort was made to identify NTDs prena-tally and at birth. Fifty percent of cases were live born;15%, stillborn; 2%, miscarriages; and 33%, induced abor-tions. Twenty-four percent of cases were <20 weeks ges-tation. Cases with chromosomal and known geneticcauses were excluded. For the case-control study, con-trols were randomly selected from normal live births,and were frequency matched to cases by hospital andyear.

To cover the 14 counties along the 1000-mile border,field teams were located in Harlingen, Laredo, and ElPaso to actively search for cases in area facilities. Womenwere contacted in the facilities and consent was obtainedto interview them 2 months later (after delivery or preg-nancy termination). Women enrolled in the study werevery poor (over half had family incomes less than$15,000 per year), were generally Spanish speaking (48%completed the interview solely in Spanish), and under-educated (half had <12 years of schooling). The inter-view, in English or Spanish, was conducted in personand took 2 hours to administer and covered an extensivematernal health history (health conditions, medicationstaken including whether prescribed, over the counter, orhome remedies), reproductive history, family demo-graphics, medical history, use of nutritional supplements,maternal tobacco use, maternal and paternal alcohol con-sumption, maternal and paternal use of street drugs andinhalants, environmental exposures, residential history,tortilla consumption, fish consumption, maternal and pa-ternal occupational history and exposures, stress andsocial support, knowledge and awareness of birth defectsand their relation to vitamin use, interviewer assessmenton the quality of the interview, and a 98-item food fre-quency questionnaire that was specifically tailored to theborder population.Before the interview, women were asked to focus on

the 3 months before conception and 3 months after con-ception; questions about exposures were specific for eachmonth during this periconceptional period. Using themedical record, the date of conception was establishedfor each mother. Interviewers collected blood and urinespecimens from the mother at the time of the interviewand tested water sources for nitrates. Blood spots werealso collected from the baby/fetus, the mother, and thefather for genetic analyses. In a subset of cases and con-trols, cord blood and placenta tissue was collected at thetime of delivery or termination. Due to the interest inpotential tortilla contaminants, the interviewers collectedtortilla samples from a subsample of study participanthomes. At the 1-year anniversary of the index conception(anniversary month was 2–4 months after interviewvisit), staff collected from participants blood and urinespecimens to measure environmental contaminant bio-markers (pesticides, polychlorinated biphenyls [PCBs],heavy metals, paranitrophenols, sphingosine, and sphin-ganine). This time period was chosen to account for sea-sonal environmental exposures (e.g., agricultural pesti-cides). Participants were paid $20 for the interview and$20 for providing specimens, with an additional $15 forthe environmental biomarker visit.Table 1 summarizes the a priori hypotheses, the expo-

sures of interest and how they were measured, and thesignificant findings.

STUDY PARTICIPANT CHARACTERISTICS

Of the 225 Mexican American case-women and 378control-women identified for study, 184 case-women(82%) and 225 control-women (60%) completed inter-views (Hendricks et al., 2001). Of those who completedinterviews, 85% of case-mothers and 83% of control-mothers provided specimens, and 55% of case-mothersand 53% of control-mothers provided specimens for theenvironmental biomarkers study done 1-year postconcep-

2 SUAREZ ET AL.

Birth Defects Research (Part A) 00:000�000 (2012)

tion. The good participation rates were likely due tothe bilingual interviewers, the in person interview andthe intense follow-up, and case coordination providedto the women from the onset of contact in the hospitalor facility.

As noted above, case-mothers and control-motherswere generally poor (65% of case-mothers and 57% ofcontrol-mothers had annual incomes under $15,000),undereducated (51% of case-mothers and 49% of con-trol-mothers had <12 years of schooling), and about halfof participants were Mexico-born (Brender et al., 2002;Felkner et al., 2003). Characteristics that would havehad a significant effect on the health of offspring werethe high prevalence of obesity (29% of case-mothers and18% of control-mothers were obese) (Brender et al.,2002) and the rare use of preconceptional folic acid-con-taining vitamins. Only 2% of case-mothers and 3% ofcontrol-mothers used folic acid-containing vitaminsdaily during the 3 months before conception (Suarezet al., 2000). Additionally, median serum and red bloodcell folate levels did not differ significantly betweencase-mothers and control-mothers (serum folate ng/mL11.2 vs 11.4; red blood cell [RBC] folate ng/mL 339 vs360) (Suarez et al., 2003b).

Findings Related to the Individual:Micronutrients, Metabolic Pathways,

and Genetic Factors

Folic Acid and Folate. In 2000, we published prelimi-nary information that demonstrated a modest protectiveeffect of combined folate intake from supplements anddiet among this population (Suarez et al., 2000). Mothersconsuming 1.0 mg per day or greater had an adjustedodds ratio (OR) of 0.73 (95% CI, 0.31–1.72) compared tothose consuming <0.4 mg per day; there was no signifi-cant difference in those consuming between 0.4 and 0.9mg per day (adjusted OR, 0.96; 95% CI, 0.54–1.71). Areanalysis of the completed case-control study showed astronger trend with mothers consuming 0.4 to 0.9 mg/day having an OR of 0.8, and those consuming 1.0 mg/day or more having an OR of 0.6 compared to those withintakes under 0.4 mg/day (Suarez et al., 2003b).Although an earlier California study had suggested thatfolic acid might be less effective in Hispanics (Shawet al., 1995), we thought it far more likely that Hispanicwomen obtained most of their combined intake from die-tary folates in the polyglutamated form, in which onlyabout 50% of the folate is bioavailable rather than themore effective folic acid in the monoglutamated form(which is 100% bioavailable). In our own study, that sofew Mexican American women on the border consumedfolic acid-containing vitamins (2%) during periconceptionlimited what could be determined about the use of sup-plemental folic acid and its effect on NTDs. With the sys-tematic examination of hypotheses in Table 1, we believethat findings demonstrate that risk in this populationwas mediated by host susceptibility factors such asgenetic or acquired variability in the intake, uptake, ormetabolism of folic acid or other factors in the metabolicpathway. Insufficient use of folic acid supplements leftthis economically disadvantaged and medically under-served population vulnerable to these multiple insults onthe developing embryo (e.g., fumonisins, stress).

Serum B12, Homocysteine, Methionine

Although the intake of folic acid and folate seemed tohave a modest effect on NTDs in this border population,and no clear relation was observed between serum orRBC folate levels and risk, examination of B12 serum lev-els indicated a strong, consistent relation with NTD risk(Suarez et al., 2003b). Based on quintiles of serum B12

values in control-mothers, the ORs from lowest to thefourth quintile were 3.0 (95% CI, 1.4–6.3), 1.6 (95% CI,0.7–3.6), 1.7 (95% CI, 0.8–3.8), and 1.1 (95% CI, 0.5–2.6)compared to the highest quintile. Very few of the womenwere overtly deficient in B12 but requirements for B12-mediated end products in rapidly dividing neural tubecells may exceed the normal maternal requirements forB12. B12 is an essential cofactor for methionine synthase,which catalyzes the simultaneous conversion of 5-methyltetrahydrofolate to tetrahydrofolate (THF) and homocys-teine to methionine. A deficiency of either folate or B12

results in a paucity of the products THF and methionineand a buildup of homocysteine. Reduced levels of methi-onine interfere with the synthesis of lipids, nucleic acids,and proteins (Lee and Herbert, 1999); high levels of ho-mocysteine seem to be damaging to the neural tube(Brauer and Tierney, 2004).To explore the effects of methionine and homocys-

teine on NTD risk, we then performed two additionalanalyses. In the 2009 study of serum homocysteine(Felkner et al., 2009), mothers in the upper quintileshad increased ORs of NTDs (ORs were 1.7, 1.3, 2.8, and2.4, respectively, from next lowest to highest quintile)compared to mothers in the lowest quintile. Moreover,high serum homocysteine levels were associated withNTD risk even when serum B12 or RBC folate levelswere high (Felkner et al., 2009), implying that excesshomocysteine might play an independent role in the de-velopment of NTDs, but not resolving whether theoffending agent was high homocysteine or low methio-nine. At this point in the continuing investigation, se-rum samples from the case-control study had beendepleted so that they were no longer representative ofthe study population. This constrained further investi-gation of the folate metabolic pathway to measuresbased on calculated nutrients estimated from mothers’recollection of dietary intake. Despite the lack of a bio-logic measure, our study of methionine, based on infor-mation from the food frequency questionnaire, revealeda protective effect for NTDs (adjusted OR, 0.66; 95% CI,0.30–1.45) for the highest quartile of methionine intakecompared to the lowest quartile). However, the CIincluded 1 and the test for trend across quartiles wasnot statistically significant (p 5 0.24) (Graham et al.,2010). Closer examination of the modifying effects of se-rum B12 levels provided another clue. With mothers inthe lowest quartile of methionine intake and lowestquintile of serum vitamin B12 level serving as the refer-ent, increasingly higher levels of B12 were protectiveeven when mothers had relatively low methionineintakes. Increasingly higher levels of methionine intakewere protective even when mothers had relatively lowlevels of B12.

3NEURAL TUBE DEFECTS ON TEXAS-MEXICO BORDER

Birth Defects Research (Part A) 00:000�000 (2012)

Tab

le1

SummaryofTexas

Neu

ralTubeDefectProject

Hypotheses

andPublish

edFindings

Exposu

rehypotheses

Studiedexposu

rePublication

Mainfindings(O

R;95%

CI)

Sam

ple

size

Micronutrients

andMetabolicPathways

Folicacid

supplemen

tsan

ddietary

folate

Recalledmaternal

dietary

intakeoffolicacid

from

FFQ

andrecalled

consu

mptionoffolicacid

supplemen

tsfrom

MQ

Nutritional

Supplemen

tssection.

Suarez

etal.,20

00Nofolicacid

supplemen

ts(aOR,1.12;0.22–5

.78).

148cases

>1.0mg/day

intakeoffolicacid

from

supplemen

tsan

ddiet(aOR,0.73;0.31–1

.72).

158controls

Only

2%ofwomen

consu

med

supplemen

ts.

B12an

dfolate

levels

Maternal

postpartum

serum

B12an

dfolate

andRBC

folate

levels.

Suarez

etal.,20

03b

Low

serum

B12(cOR,3.0;

1.4–

6.30).

157cases

Low

serum

folate

(cOR,1.0;

0.5–1.9).

186controls

Low

RBC

folate

(cOR,1.3;

0.6–2.7).

Homocysteine(impairedfolate

B12metab

olism

)Maternal

postpartum

serum

homocysteinelevels.

Felkner,20

09Highserum

homocysteine(cOR,2.4;

1.0–5.7).

103cases

Riskeffect

indep

enden

tofserum

B12orfolate

or

RBC

folate.

139controls

Methionineintake

Recalledmaternal

dietary

intakeofmethioninefrom

FFQ.

Graham

,2010

Highdietary

methionineintake(aOR,0.66;0.3–1.45).

184cases

225controls

Diarrhea

Recalledep

isodes

ofdiarrhea

from

MQ

Maternal

Healthsection.

Felkner

etal.,2003

Periconceptional

diarrhea,an

yep

isodes

(aOR,3.3;

1.5–7.1).

184cases

225controls

�2ep

isodes

compared

with�1(cOR,8.4;

2.9–24

.6).

H.pylori

Maternal

postpartum

serum

IgG

H.pylori

antibodies.

Felkner

etal.,20

07H.pyloriseropositivity(cOR,1.4;

0.8–2.4).

95cases

147controls

CacheValleyvirus

Maternal

postpartum

serum

CVV

neu

tralizing

antibodies.

Edwards,

1997

CVV

serum

neu

tralizationactivity:nonedetectedin

case

orcontrolwomen

.74

cases

Irondefi

cien

cies

Maternal

postpartum

serum

ferritin

levels;

hem

oglobin,hem

atocrit,an

dRBC

indices;recalled

maternal

dietary

intakeofiron,vitam

inC,

calories,an

dmeatfrom

FFQ;recalled

anem

ia,

pica,

heavymen

struation,gravidityan

dinterpregnan

cyinterval

(other

ironstatus

indicators)from

MQ

Maternal

Healthan

dNutritional

Supplemen

tssections.

Felkner

etal.,20

05Low

serum

ferritin

(cOR,1.8;

1.0–3.3).

158cases

Indep

enden

tofserum

folate

andB12.

189controls

Low

hem

oglobin,hem

atocrit,RBC

indices:no

association.

Dietary

intakes:noassociation

Pica(cOR,5.0;

1.0–24

.2).

Other

ironstatusindicators:noassociation.

Insu

lininsensitivity

Maternal

postpartum

serum

insu

lin;recalled

height

andweightforBMIfrom

MQ

Rep

roductive

History

section.

Hen

drickset

al.,20

01Hyperinsu

linem

ia(cOR,1.91;1.21–3

.01).

149cases

Hyperinsu

linem

iaad

justed

forobesity(aOR,1.75;

1.09–2

.82).

178controls

Obesityad

justed

forhyperinsu

linem

ia(aOR,1.45

;0.84–2

.51).

Dietingbeh

aviors

Recalledvariousdietingbeh

aviors

from

MQ

Nutritional

Supplemen

tsection.

Suarez

etal.,2012

Preconceptional

dietingto

lose

weight(cOR,1.9;

1.1–3.3).

184cases

225controls

Indep

enden

tofBMI.

Mitigated

by1.0mg/day

folate

intake.

Genetic

Factors

Folate

pathway

gen

esAlleles

ofMTHFR

gen

e,an

dFR-a

from

infantan

dmother

bloodsp

ots.

Barber

etal.,20

00MTHFRgen

einfantheterozy

gosity

(cOR,1.8;

0.47–6

.8).

101cases

139controls

MTHFRgen

einfanthomozy

gosity

(cOR,1.8;

0.35–9

.4).

MTHFRgen

emother

(cOR,0.6;

0.30-1.33)

Folate

receptorgen

e:little

ornoassociation.

Transcriptionfactorgen

esAlleles

oftran

scriptionfactorgen

esfrom

infant

bloodsp

ots.

Zhuet

al.,2003a

ZIC

2an

dZIC

3tran

scriptiongen

es:noassociation.

101cases

139controls

Promotorhap

lotype;

PDGFgen

eAlleles

ofpromotorhap

lotypePDGFgen

esfrom

infantan

dmother

bloodsp

ots.

Zhuet

al.,2004

Promotorhap

lotypePDGFRA

lowesttran

scription

activitymothers(cOR,2.2;

1.0–4.6).

122case

mothers

127controlmothers

Promotorhap

lotypePDGFRA

atleastonelow

activityallele

infants

(cOR,1.5;

0.8–3.1).

43case

infants

124controlinfants

Tab

le1

SummaryofTexas

Neu

ralTubeDefectProject

Hypotheses

andPublish

edFindings(continued

)

Exposu

rehypotheses

Studiedexposu

rePublication

Mainfindings(O

R;95

%CI)

Sam

ple

size

Homocysteineremethylation

enzy

mepolymorphisms(M

TR,

MTRR)

Alleles

ofhomocysteineremethylationen

zyme

polymorphisms(M

TR,MTRR)from

infantan

dmother

bloodsp

ots.

Zhuet

al.,20

03b

MTRmutantinfant(cOR,1.9;

0.9–3.9).

122case

mothers

MTRmutantmother

(cOR,1.2;

0.7–

2.1).

127controlmothers

MTRR

mutantinfant(cOR,2.6;

1.3–5.3).

43case

infants

MTRR

mutantmother

(cOR,1.9;

1.1–3.1).

124controlinfants

Both

mutants

infants

(cOR,5.1;

1.7–15

.4).

Both

mutants

mother

(cOR,2.1;

1.0–4.7).

EnvironmentalandOccupationalRiskFactors

PCBs

Maternal

postpartum

serum

PCBlevels.

Suarez

etal.,20

05PCBcongen

ers(cOR<1ornoassociation).

87cases

101controls

Pesticides

Recalledvariousreported

sources

ofpesticide

exposu

reat

homean

dwork

from

MQ

Environmen

talsection.

Brender

etal.,20

10Pesticideexposu

rein

homeoryardallNTDs(aORs,

2.0;

1.2–

3.0).

184cases

225controls

Exposu

reto

3–5pesticidesources

compared

to0

exposu

resan

encephaly(aOR,2.7;

1.1–

6.8).

Exposu

reto

3–5pesticidesources

compared

to0

exposu

ressp

inabifida(aOR,2.7;

1.1–6.8).

Mercu

ry,lead

,arsenic

(heavy

metals)

Maternal

1-yearpostconceptionurinemercu

ryan

darsenic

levels;

maternal

postpartum

serum

lead

levels;

recalled

occupational

exposu

resfrom

job

titles

from

MQ

Occupationsection;allheavy

metalsin

drinkingwater

from

1-year

postconceptionpublicwater

supply

record

s;distance

from

metal

airem

issionsfrom

EPA

record

s.

Brender

etal.,20

06Exposu

reto

burningtreatedwood(cOR,4.2;

1.1–

16.0).

184cases

225controls

20other

exposu

res:

noassociation.

Solven

tsRecalledoccupational

exposu

refrom

jobtitles

(recalledjobhistory);potential

solven

texposu

res

from

maternal

hobbiesan

dother

reported

activitiesat

homefrom

MQ

Occupational

and

Environmen

talsections.

Brender

etal.,20

02Exposu

reto

solven

tsat

work

orwithhobbies(cOR,

2.5;

1.3–

4.7).

184cases

225controls

Fumonisins

Recalledmaternal

dietary

intakeofcorn

tortillas

from

MQ

Environmen

talsection;corn

tortilla

fumonisinslevels;

maternal

postpartum

serum

SA:SO

ratio.

Missm

eret

al.,20

06Fumonisinsexposu

re18

4cases

increasestherisk

ofNTD,proportionateto

dose,up

toathresh

old

level.ORssh

ownarehighestrisk

effect.

225controls

Moderatetortillasconsu

mption(aOR,2.4;

1.1–5.3).

SA:SO

ratio(aOR,4.4;

1.2–15

.5).

Nitrates

Recalledmaternal

dietary

intakeofnitritesan

dnitratesfrom

FFQ,recalled

nitrosatable

druguse

from

MQ

Maternal

Healthsection;usu

alsourceof

drinkingwater

nitrate

levels.

Brender

etal.,2004

Highnitrosatable

druguse

andhightotalnitrite

intake(aOR,7.5;

1.8–

45).

184cases

225controls

Usu

aldrinkingwater

sourcewithnitrate

above3.5

mg/L(cOR,1.9;

0.8–4.6).

Alcohol,tobacco,recreational

drugs

Recalledexposu

reto

cigarette

smoke,

alcoholic

drinks,

anduse

ofstreet

drugsfrom

MQ

Tobacco,

Alcohol,an

dStreetDrugsan

dInhalan

tssections.

Suarez

etal.,20

08Passivecigarette

smoking(cOR,2.6;

1.6–

4.0).

175cases

Smokeexposu

re<halfpack/day

(cOR,2.2;

1.0–

4.8).

221controls

Smokeexposu

re�halfpack/day

(cOR,3.4;

1.2–

10.0).

Alcoholintakeorstreet

drugs:

noassociation.

Hyperthermia

Recalledfever,febrile

illnesses,

exposu

resto

external

heatsources,hyperthermia-inducingactivities

from

MQ

Maternal

Healthan

dEnvironmen

tal

sections.

Suarez

etal.,20

04First

trim

esterfever

(cOR,2.9;

1.5–5.7).

175cases

Exposu

reto

heatdev

ices

(hottubs,

saunas,or

electric

blankets)

(cOR,3.6;

1.1–

15.9).

221controls

Stress

Recalledstressfullife

even

ts,social

andem

otional

support

indices

from

MQ

Stressan

dSocial

Support

section.

Suarez

etal.,20

03a

1ormore

stressfullife

even

ts1yearpreceding

conception(cOR,2.9;

1.8–4.7).

Associationindep

enden

tofknownrisk

factors.

184cases

225controls

OR,oddsratio;CI,confiden

ceinterval;FFQ,FoodFrequen

cyQuestionnaire;MQ,Mother

Questionnaire;aO

R,ad

justed

oddsratio;RBC,redbloodcell;cO

R,crudeoddsratio;CVV,Cache

Valleyvirus;

BMI,bodymassindex;MTHFR,methylentetrah

ydrofolate

reductase;

FR-a,folate

receptor-a;PDGF,platelet-derived

growth

factors;PDGFRA,platelet-derived

growth

factorre-

ceptoralpha;

PCB,polych

lorinated

biphen

yl;NTDs,

neu

raltubedefects;EPA,Environmen

talProtectionAgen

cy;SA:SO,sp

hingan

ine:sp

hingosine.

Intestinal Parasites, Helicobacter pylori,and Diarrhea

Our hypotheses regarding inadequate folate and B12

encompassed not only intake but also the role of the gas-trointestinal ecosystem and malabsorption, recognizingthe unique aspects of the border environment—the sub-standard sanitation and the frequent travel to Mexico.We had initially hypothesized that intestinal parasitesmight interfere with B12 metabolism and absorption, butinformation on a subset of our study participants showedthat Diphyllobothrium latum and Giardia lamblia (gastroin-testinal parasites known to interfere with B12 absorption)were virtually nonexistent in the study population. Tofollow-up on the malabsorption hypothesis, we looked ata bacterial pathogen, Helicobacter pylori, which has beenlinked to B12 insufficiency (Felkner et al., 2007). We wereable to test a portion of the case-mother (42%) and con-trol-mother (39%) serum samples for antibodies to H.pylori. Results did not indicate a significant role for H.pylori in NTD risk for the total population; ORs were 1.4(95% CI, 0.8–2.4) for mothers with seropositivity com-pared to those testing negative. However, results didsuggest that this risk might merit further investigationamong the young (<20 years old; OR,2.4; 95% CI, 0.7–7.0)and poorly educated (<7 years of schooling; OR, 2.3; 95%CI, 0.5–10.0). Periconceptional diarrhea, defined as threeor more loose stools per day, without regard to a specificpathogen, emerged as a more intriguing risk factor forthis population (Felkner et al., 2003). Sixteen percent ofcase-mothers and 5% of control-mothers reported havingpericonceptional diarrhea for an OR of 3.7 (95% CI, 1.8–7.6). Mothers who reported two or more episodes had an8.4-fold increase in NTD risk (95% CI, 2.9–24.6) comparedto mothers with one or no episodes. Neither fever norother established risk factors explained the associationwith diarrhea (adjusted OR, 3.3; 95% CI, 1.5–7.1).

Iron-Related Nutritional Factors

Although the majority of our metabolic studies focusedon the micronutrients in the folate pathway, which hasbeen definitively linked to NTD risk, we also analyzedfactors related to iron status (Felkner et al., 2005). Thisanalysis was prompted by the 1991 NTD Brownsvillecluster investigation of 24 cases and 23 normal controlbirths, which showed that case-mothers had significantlylower hematocrit levels than control-mothers. This factorwas one of the few that differed significantly betweencase-mothers and control-mothers in the initial clusterinvestigation. In our Texas-Mexico border study, wefound case-mothers had higher odds of having minimalor no iron stores (serum ferritin <30 lg/L) compared tocontrol-mothers (OR, 1.8; 95% CI, 1.0–3.3) (Felkner et al.,2005). Also, mothers who reported a history of picabehavior had 5 times the NTD risk than those without ahistory (OR, 5.0; 95% CI, 1.0–24.2), a possible manifesta-tion of poor bioavailability of dietary iron. Unlike micro-nutrients in the folate pathway, which rebound relativelyquickly postpartum, pregnancy has both a greater acuteand chronic impact on iron levels, weakening our abilityto impute causality to this association. This study didserve to bring attention to poverty fostered factors thatmight increase risks for poor outcomes in addition toNTDs.

Obesity and Hyperinsulinemia

Juxtaposed against low nutrient levels, obesity isanother nutrient-related risk factor that is now generallyaccepted as an NTD risk factor. In this population, wefound that both maternal obesity and hyperinsulinemiacontributed to the risk of NTDs in offspring (Hendrickset al., 2001). Mothers who were obese (body mass index[BMI] �30) had a 1.6 OR (95% CI, 0.9–2.9) compared tonormal weight mothers (BMI 5 18.5–24.9), whereas thosewho had high (11–15 lIU/mL) or very high (>15 lIU/mL) serum insulin levels had ORs of 1.8 and 1.9, respec-tively, compared to those with normal levels (6–10 lIU/mL). Adjusting insulin ORs for obesity did little tochange the effect (adjusted OR, 1.8; 95% CI, 1.1–2.8 for in-sulin >10 lIU/mL). Alternatively, adjusting the obesityOR for insulin levels only slightly reduced the estimatedeffect (adjusted OR, 1.5; 95% CI, 0.8–2.5 for BMI �30).The twofold increase in risk associated with hyperinsu-linemia would have had a large impact on the occurrenceof NTDs in Mexican Americans because the prevalenceof diabetes and obesity was high. Fifty-four percent ofthe control-mothers had elevated insulin levels (>10lIU/mL). These findings should lead others to considerwhether populations at high risk of diabetes mellitusshould be screened for hyperinsulinemia before concep-tion, and whether a population genetically disposed todiabetes mellitus might require higher than recom-mended levels of folic acid to prevent NTDs.Because of studies suggesting that lowered weight gain

and maternal weight reduction in early pregnancy mightbe risk factors, we also examined the association of self-reported maternal dieting behaviors and NTDs (Suarezet al., 2012). Mexican American mothers who reportedbeing on a diet to lose weight during the 3 months beforeconception had a twofold risk of NTDs. Surprisingly, thiselevated OR associated with dieting did not differ amongnormal, overweight, or obese mothers but the effect dis-appeared among mothers consuming at least 1.0 mg/dayof folate. Obviously, although ‘‘dieting’’ may have a neg-ative impact through restriction of micronutrients, theoverriding factor is still sufficient folic acid in the diet.

Genetic Factors

To date, using offspring bloodspots, four studies of therole of gene variants in the risk of NTDs among thisTexas-Mexico border population have been completed.The first of these exploring the association between folatemetabolism genes (methylentetrahydrofolate reductase[MTHFR]) and folate receptor gene (FOLR1) (Barberet al., 2000) showed nonstatistically significant elevatedORs for the C677T polymorphism in the MTHFR gene(OR, 1.8; 95% CI, 0.47–6.8 for heterozygosity and OR, 1.8;95% CI, 0.35–9.4 for homozygosity) relative to the wild-type homozygotes. The number of children samples wassmall (24 cases, 93 controls) and the analyses of themother samples showed opposite results (decreasedORs). The risk allele in the MTHFR C677T polymorphismhas a higher prevalence in our study population com-pared to other populations. This polymorphism hasshown interactions with folate and other micronutrientstatus in other studies (Shaw et al., 1998; Volcik et al.,2003), however, due to the limited sample size, we werenot able to evaluate the potential interaction and its effecton NTD risk in our study. In a separate study of tran-

6 SUAREZ ET AL.

Birth Defects Research (Part A) 00:000�000 (2012)

scription factors, no associations were observed for thetranscription candidate genes ZIC2 and ZIC3 in this pop-ulation (Zhu et al., 2003a). However, we observed anassociation between a functional promoter haplotype ofplatelet-derived growth factor receptor alpha (PDGFRA)and NTD risk (Zhu et al., 2004). Mothers with two copiesof low promotor/transcription alleles had an elevatedOR (OR, 2.2; 95% CI, 1.0–4.6) compared to mothers withtwo copies of the high PDGFRA transcription activity al-leles. These results were inconsistent in that no increasedOR was observed for one copy nor were these resultsreflected in the infant samples (low-low was 1.2). Perhapsthe more intriguing study was of genes encoding homo-cysteine remethylation enzymes (MTR, MTRR), whichshowed that infant and mothers with either the MTR orMTRR mutant gene had increased risk of NTDs com-pared to the wild type (ORs ranged from 1.2–2.6) (Zhuet al., 2003b); and that infants and mothers with bothmutants (MTR and MTRR) had even higher risks (ORranged from 2.1–5.1).

Findings Related to the Environment,Occupational or External Factors

Polychlorinated Biphenyls, Heavy Metals, Pesticides,Solvents. Given the level of concern by the border com-munities over the high prevalence of NTDs and the pre-sumed cause of environmental contamination, we alsoexamined the risks associated with exposures to PCBs,heavy metals, and pesticides. In the 2005 study (Suarezet al., 2005), we found no association between PCB serumlevels and risk of NTDs. ORs for various PCB congenerswere below one or compatible with the null. In the 2010study that focused on pesticide exposures, case-motherswere more likely than control-mothers to report multiplesources of pesticide exposure (Brender et al., 2010). Assources of pesticide exposure opportunities increased,risk of anencephaly and spina bifida in offspringincreased (p for trend, respectively, 0.001 and 0.028).Although case-mothers were more likely to report livingwithin 0.25 miles of cultivated fields than control-moth-ers, this result may have generated through differentialrecall. Although to produce the observed OR, differencesin recall between case-mothers and control-motherswould have had to be extreme. Results for biomarkersamples (blood and urine), including for chlorinatedhydrocarbon insecticides and methyl parathion, werebelow detectable levels for the majority of mothers.

In a separate study of heavy metals, median levels ofbiomarkers for arsenic, cadmium, lead, and mercury didnot differ significantly between case-mothers and control-mothers (Brender et al., 2006). There were also no signifi-cant associations between NTD risk and occupationalexposures to heavy metals (cadmium, lead, mercury, andarsenic), heavy metals in drinking water supply, or closeproximity to facilities with air emissions to heavy metals.However, these exposures were infrequent in cases andcontrols. The small proportion of the study populationthat participated in the biomarker portion of the studyand the small sample sizes for drinking water contami-nants further constrained the robustness of our findings.The only environmental finding of note was an increasedOR for maternal periconceptional exposures to burningtreated wood (OR, 4.2; 95% CI, 1.1–15.5). One of themore significant public health findings was the relatively

high levels of urinary mercury found in both case-moth-ers and control-mothers. Twenty-three percent of Mexi-can American mothers in the study had urinary mercurylevels of 5.62 lg/L or greater, compared to 5% of Mexi-can American women in National Health and NutritionExamination Survey. During the study period, a facecream manufactured in Mexico was found to containmercurous chloride, which had poisoned users in theborder population. Although women in the study deniedusing the face cream, the high levels of mercury found inour study participants may have been linked to the useof such products.In our study of parental occupation and NTDs in off-

spring (Brender et al., 2002), assignment of occupationalexposures was based on detailed questions about workhistory including job titles and materials handled and in-formation from the available occupational health and tox-icological literature. In that study, case-mothers were 2.5times more likely (95% CI, 1.3–4.7) than control-mothersto engage in work activities or hobbies that involved sol-vent exposure. Other maternal jobs associated with NTDsin offspring included work as a cleaner (OR, 9.5; 95% CI,1.1–82.2) and health care occupations (analyses restrictedto those in patient care and/or exposure to chemical/physical agents; OR, 3.0; 95% CI, 1.0–9.0). With the excep-tion of health care occupations (6% of case-mothers), lessthan 5% of case-mothers and control-mothers worked inthe occupational groups of interest. That few womenwere employed in the occupational groups of interestand no industrial hygienist reviewed the work historiesin exposure assignment were limitations of this study.

Fumonisins

Two years before the 1991 NTD cluster in Brownsville,Texas, an outbreak of equine leukoencephalomalacia(ELEM; liquefaction of the white matter of the brain inhorses) occurred nationwide (Hendricks, 1999). In Texas,40 ELEM clusters were reported in less than 2 months.The corn harvest that year was implicated and livestockwere taken off feed containing corn. Although the cornharvest was thought to be contaminated with mold,humans and non-equine animals continued to eat cornfrom the same harvest. The corn contaminant that led tothe ELEM cluster was a class of mycotoxins called fumo-nisins, produced by the molds Fusarium verticillioides.Cornmeal samples collected in the United States duringthe NTD cluster had relatively high fumonisins levels.Because corn tortillas are a basic food staple among Mex-ican Americans living on the border, we believed that theELEM and NTD clusters and the high NTD prevalencemight have the same link to fumonisins. To measurefumonisins exposure, we collected detailed informationabout tortilla consumption during the interview and col-lected corn tortilla samples from the home. We also col-lected maternal serum to measure the ratio of sphinga-nine to sphingosine, a fumonisins biomarker (Missmeret al., 2006).Our study published in 2006, was the first epidemio-

logic evidence that fumonisins may increase the risk ofneural tube defects (Missmer et al., 2006). We found thatmoderate (301–400 tortillas) compared to low (<5100 tor-tillas) consumption of corn tortillas during the first tri-mester was associated with increased ORs of having anNTD-affected pregnancy (OR, 2.4; 95% CI, 1.1–5.3), after

7NEURAL TUBE DEFECTS ON TEXAS-MEXICO BORDER

Birth Defects Research (Part A) 00:000�000 (2012)

adjusting for confounders. No increased risks wereobserved at intakes higher than 400 tortillas (OR, 0.8;95% CI, 0.4–1.6) for 401–800 and OR, 1.0; 95% CI, 0.5–2.3)for >800. Furthermore, increasing levels of fumonisin ex-posure, as measured by the postpartum sphinganine:sphingosine (SA:SO) ratio, were associated with increas-ing ORs for NTDs, except for the highest exposure(SA:SO >0.35). An analysis of maternal fumonisin expo-sure imputed from tortilla samples taken from the homealso showed an inverted U-shaped pattern with NTDrisk. These findings suggested that fumonisin exposureincreases the NTD risk proportionate to dose, up to athreshold level, at which point fetal death may be morelikely to occur. Laboratory studies of fumonisins haveprovided insight into a potential mechanism between ex-posure and NTD occurrence. Fumonisins inhibit the bio-synthesis of sphingolipids (Wang et al., 1991), whichinterferes with the uptake of 5-methyltetrahydrofolateand decreases total folate binding. Mice embryos sub-jected to fumonisins develop NTDs in vitro (Flynn et al.,1997; Sadler et al., 2002) and in vivo (Gelineau-van Waeset al., 2005). Moreover, administration of folate reversesthis effect. Clearly, fumonisin contamination of corn con-sumed by humans represents a potential risk for NTDoccurrences, especially in populations with inadequatefolate intake who are reliant on corn-based foods.

Nitrates, Nitrites, and Nitrosatable Drugs

A consideration of other environmental contaminantsled us to a comprehensive examination of nitrite expo-sure including dietary nitrates and nitrites, nitrates indrinking water, and exposure to nitrosatable drugs(Brender et al., 2004). We calculated the daily intake ofdietary nitrates and nitrites from the Food FrequencyQuestionnaire, estimating the average amounts of nitratesand nitrites in foods from the published literature. Alsofrom the published literature, we classified drugsreported in the mother questionnaire as nitrosatable (ornot). Interviewers determined the usual periconceptionaldrinking water source (bottled or tap water) and, if avail-able, collected water samples from that same source,which were measured for nitrates. Overall, intake of die-tary nitrates and nitrites were not positively associatedwith NTD risk. In contrast, mothers who took drugs clas-sified as nitrosatable were 2.7 times more likely (95% CI,1.4–5.3) to have an NTD-affected pregnancy than motherswho did not (Brender et al., 2004). Mothers whose drink-ing water nitrates measured 3.5 mg/L or greater (medianof control-mothers’ drinking water) were 1.9 times morelikely (95% CI, 0.8–4.6) to have an NTD-affected preg-nancy than mothers with lower water nitrate levels.Because endogenous nitrosation involves the reaction ofnitrosatable compounds with nitrites, we stratified nitro-satable drug exposure by dietary nitrite and total nitriteintakes. Nitrate levels in water as well as estimated dailydietary intake of nitrites and total nitrites substantiallymodified NTD risk associated with this drug exposure.Higher intakes of total dietary nitrites (dietary nitrites15% dietary nitrates) were associated with higher risksfor NTDs if nitrosatable drugs were taken during thepericonceptional period. NTD ORs associated with nitro-satable drug exposure were 0.9 (95% CI, 0.3–3.5), 2.7(95% CI, 0.8–11), and 7.5 (95% CI, 1.8–45), respectively,for the lowest, middle, and highest tertile of total nitrite

intake. These findings were corroborated in a recentstudy of nitrosatable drug exposure and NTDs that ana-lyzed data from the National Birth Defects PreventionStudy population (Brender et al., 2011).

Passive Cigarette Smoke

Evidence on the association between cigarette smokingand NTDs has historically been mixed and generallyinconclusive. Before the study of cigarette smoking in theTexas-Mexico border population, little consideration hadbeen given to the effects of secondhand cigarette smoke.In our study in 2008, we showed that nonsmoking moth-ers exposed to secondhand smoke during the first trimes-ter had an NTD OR of 2.6 (95% CI, 1.6–4.0) compared tothose who neither smoked nor were exposed to second-hand smoke (Suarez et al., 2008). Compared to the refer-ent, the OR among mothers who smoked less than half apack a day during the first trimester was 2.2 (95% CI,1.0–4.8) and 3.4 (95% CI, 1.2–10.0) for those smoking ahalf a pack or more a day. This study indicated to usthat secondhand smoking had the same effect as smokinga half pack or less. Although it is possible that cigarettesmoking or exposures to cigarette smoke is a surrogatefor other risk factors (e.g., nutritional factors), adjustmentfor age, education, BMI, or folate intake had a negligibleeffect on these findings. Interestingly, other studies haveshown a positive correlation between cigarette smokingand homocysteine levels (Chrysohoou et al., 2004; Ganjiand Kafai, 2003).

Fever, Febrile Illnesses, and Heat Exposures

We also confirmed that maternal hyperthermiaincreased the risk for NTD-affected offspring in this pop-ulation (Suarez et al., 2004). The OR associated withmaternal fever in the first trimester compared to no feverwas 2.9 (95% CI, 1.5–5.7). Mothers taking fever-reducingmedications showed a lower risk effect (OR, 2.4; 95% CI,1.0–5.6) than those who did not (OR, 3.8; 95% CI, 1.4–10.9). First trimester maternal exposures to heat devicessuch as hot tubs, saunas, or electric blankets were associ-ated with an OR of 3.6 (95% CI, 1.1–15.9). The proportionof control-mothers experiencing fever or febrile illnessesin the first trimester was 15%, indicating that the attribut-able risk due to hyperthermia may be significant.

Stress

Given the Texas-Mexico border environment, we hadalso hypothesized that this Mexican American populationwould be subject to more chronic stressors than otherpopulations, which would act to intensify underlyingrisks. In a 2003 study, we investigated whether maternalstress increased NTD risk (Suarez et al., 2003a). We meas-ured maternal stress by counting the number of jobchanges, residential moves, and major injuries occurringthe year before conception. Mothers who experiencedone or more stressful life events during the year beforeconception had increased ORs for NTDs (OR, 2.9; 95%CI, 1.8–4.7) (Suarez et al., 2003a). Furthermore, as thenumber of life events increased, so did the OR for NTDs.Mothers experiencing one event had 2.7 times the NTDrisk of those with no events, and mothers with two ormore events had 8.3 times the risk. The estimated effectswere not modified or confounded by age, education,

8 SUAREZ ET AL.

Birth Defects Research (Part A) 00:000�000 (2012)

country of birth, income, obesity, vitamin supplements,dietary folate, cigarette smoking, or alcohol consumption.The mechanism by which stress might cause NTDs wasnot obvious. Stress may indirectly affect neural tube de-velopment by inducing high-risk behaviors (e.g., womenmight smoke or drink more, or eat poorly). Alternatively,it is possible that factors such as personality, lifestyle, orenvironment predispose women to both stress andNTDs. As to a biologic pathway, psychologic indices ofmaternal stress have been shown to be correlated withincreased levels of circulating adrenocorticotropin andcortisol (Wadhwa et al., 1996). Cortisol levels in themother are highly correlated with fetal concentrations(Gitau et al., 1998). Elevated cortisol levels during preg-nancy induce oral clefts in animals (Fraser and Fainstat,1951; Lahti et al., 1972). Similar to our findings, a Califor-nia study had also shown that stressful life events experi-enced periconceptionally were related to NTD risk in loweducated women (OR, 2.6) but not in highly educatedwomen (Carmichael and Shaw, 2000).

COMMENTARY

The 1991 Brownsville Texas cluster of NTDs remainsan important and historic public health event, occurringthe same year that the British Medical Research Councilpublished the finding that folic acid prevents the recur-rence of NTDs. Although the border community wasconcerned about environmental causes and contaminants,on the international front it was unequivocally confirmedthat folic acid prevents occurrences of NTDs (Czeizel andDudas, 1992). At the national level, based on this andother evidence, the United States Public Health Serviceproceeded to recommend that all women of childbearingage take supplemental folic acid (United States PublicHealth Service, 1992). Thus, this case-control study ofNTDs in Mexican Americans living along the Texas-Mex-ico border was begun with the recognition that folic acidwas an effective intervention and that a lack of it in suffi-cient quantities caused NTDs.

This study was one of the most comprehensive investi-gations of NTDs conducted to date. Active surveillance,including for cases <20-weeks gestation and a good par-ticipation rate resulted in many cases and controls. Thatthis large sample size consisted of a minority populationknown to be at high risk for NTDS further enhanced thevalue of this inquiry. The availability of biomarkers fromboth postpartum and season of conception added rich-ness to data. In general, the most significant limitationswere those inherent in most studies of rare birth defects,the necessity of doing a retrospective study to obtain suf-ficient power for statistical testing, thus encumbering thestudy with the potential for recall bias and postpartumbiomarkers that may not represent the periconceptionalstatus. Although our sample size was comparativelylarge, it was insufficient to test all the potential interac-tions and confounders that might have been warrantedin an investigation of so many variables. The difficulty inobtaining biomarkers including blood spots and theirultimate exhaustibility particularly reduced this option inhypotheses relying on biomarkers. Other limitations werespecific to the hypothesis and were covered in detail ineach article.

What have we learned in the 20 years since theBrownsville cluster? First of all, environmental contami-

nants such as heavy metals, pesticides, and PCBs thatwere a concern of the community did not explain theNTD risks to any degree. Second of all, we confirmedthat some previously established risk factors (e.g., folate,B12, obesity, and hyperthermia) were also causal factorsin this Mexican American population. Our investigationsof the micronutrients and metabolic pathways involvedconfirmed the findings of other researchers that increasedfolate intake has a protective effect and that low serumB12, high serum homocysteine levels, and obesity inde-pendently contribute to risk. However, this work on theTexas-Mexico border has served to further understandthe folate metabolic pathway by proffering a modelwhere high levels of homocysteine and low levels of vita-min B12 impact NTD risk by reducing methionine. In thismodel, adequate vitamin B12 compensates for low methi-onine intake through methionine production from homo-cysteine. Conversely, if vitamin B12 intake is low, butenough methionine is consumed, the production of me-thionine using vitamin B12 can be bypassed, and the die-tary methionine can directly methylate proteins, lipids,and DNA. We have also proposed that low micronutrientlevels resulting in NTDs may result from malabsorptionrather than inadequate consumption. In addition, wehave implicated other metabolic risk factors that requireadditional study to elucidate their physiologic mecha-nism—hyperinsulinemia and low ferritin.Third, our studies in this folic acid-deficit population

revealed several novel risk factors, namely, diarrhea,fumonisins, and the combination of nitrosatable drug ex-posure with high nitrate/nitrite intake. As noted earlier,the combined effect of nitrate/nitrite and nitrosatabledrug exposure has been replicated with data from theNational Birth Defects Prevention study (Brender et al.,2011). To date, no other epidemiologic study has corrobo-rated the findings related to diarrhea. Nor has there beenadditional evidence that a population with high preva-lence of NTDs also consumes large amounts of poten-tially contaminated staple products or has elevated bio-markers for fumonisin exposure. However, the body ofliterature related to fumonisins has grown substantiallyin other regards. At the population level, studies con-ducted since our published findings have documentedintake of foods commonly contaminated by fumonisinsand have found that variations in season, geography, anddietary habits place certain subpopulations at risk of con-suming fumonisins at much higher levels than had beenintended under the current regulatory recommendations(Torres et al., 2007; Dvorak et al., 2008; Gong et al., 2008;Burger et al., 2010; van der Westhuizen et al., 2010; DeGirolamo et al., 2011; Sun et al., 2011). Considerable pro-gress has been made in identifying the appropriate bio-marker and evaluating its correlation with individual die-tary intake (Dvorak et al., 2008; Gong et al., 2008; van derWesthuizen et al., 2010; Xu et al., 2010), concluding thaturinary FB1 is the biomarker of choice for further studyof the link between fumonisins and NTDs. Additionalanimal studies have corroborated that exposure to fumo-nisins disrupts maternal sphingolipid metabolism result-ing in NTDs and fetal death (Voss et al., 2009), andimpaired male reproductive capacity (Ewuola and Egbu-nike, 2010).In conclusion, the collective findings from the 23 stud-

ies among the Mexican American women living alongthe Texas-Mexico border demonstrate the multifactorial

9NEURAL TUBE DEFECTS ON TEXAS-MEXICO BORDER

Birth Defects Research (Part A) 00:000�000 (2012)

nature of NTD formation. A population deficient in folicacid will be vulnerable to a variety of insults whetherbrought on by individual behaviors (e.g., obesity) orthrough the surrounding environment (e.g., fumonisins).The public health solution to the frequent occurrence ofNTDs on the border requires vigilance in assuring thatpoor women are supplementing with folic acid, that theyare educated to avoid cigarette smoke and heat expo-sures during preconception, and that they consult withtheir physicians about the use of drugs or medications;and that the environment (food, water, air) is monitoredand controlled from harmful contaminants.

ACKNOWLEDGMENTSAn enormous number of individuals contributed to the

Texas Neural Tube Defect Project, including the regionalinterviewers located in the Texas Department of Healthregional offices (Harlingen, El Paso, Laredo); the Laborand Delivery and Medical Records staff of the 21 deliv-ery hospitals in the study area; scientists and laborator-ians at the University of Texas Health Science CenterLaboratory, Division of Laboratories at the Centers forDisease Control and Prevention; scientists in the BirthDefects Branch at CDC; laboratorians and scientists at theU.S. Food and Drug Administration, the National Insti-tute of Environmental Health Sciences, and the CDCNational Center for Environmental Health Laboratory;individuals at the Texas Commission on EnvironmentalQuality; scientists at the Institute of Biosciences andTechnology, Texas A&M University Health Science Cen-ter; and collaborators at the University of Texas HealthScience Center in Houston. We thank the participatingfamilies who provided their invaluable time, biologicsamples, and information so that we could explore thecauses of neural tube defects and identify ways to pre-vent these defects.

REFERENCES

Barber R, Shalat S, Hendricks K, et al. 2000. Investigation of folate path-way gene polymorphisms and the incidence of neural tube defects ina Texas Hispanic population. Mol Genet Metab 70:45–52.

Brauer PR, Tierney BJ. 2004. Consequences of elevated homocysteine dur-ing embryonic development and possible modes of action. CurrPharm Des 10:2719–2732.

Brender J, Suarez L, Hendricks K, et al. 2002. Parental occupation andneural tube defect-affected pregnancies among Mexican Americans. JOccup Environ Med 44:650–656.

Brender JD, Felkner M, Suarez L, et al. 2010. Maternal pesticide exposureand neural tube defects in Mexican Americans. Ann Epidemiol20:16–22.

Brender JD, Olive JM, Felkner M, et al. 2004. Dietary nitrites and nitrates,nitrosatable drugs, and neural tube defects. Epidemiology 15:330–336.

Brender JD, Suarez L, Felkner M, et al. 2006. Maternal exposure to ar-senic, cadmium, lead, and mercury and neural tube defects in off-spring. Environ Res 101:132–139.

Brender JD, Werler MM, Kelley KE, et al. 2011. Nitrosatable drug expo-sure during early pregnancy and neural tube defects in offspring:National Birth Defects Prevention Study. Am J Epidemiol 174:1286–1295.

Burger HM, Lombard MJ, Shephard GS, et al. 2010. Dietary fumonisin ex-posure in a rural population of South Africa. Food Chem Toxicol48:2103–2108.

Carmichael SL, Shaw GM. 2000. Maternal life event stress and congenitalanomalies. Epidemiology 11:30–35.

Chrysohoou C, Panagiotakos DB, Pitsavos C, et al. 2004. The associationsbetween smoking, physical activity, dietary habits and plasma homo-cysteine levels in cardiovascular disease-free people: the ‘ATTICA’study. Vasc Med 9:117–123.

Czeizel A, Dudas I. 1992. Prevention of the first occurrence of neural tubedefects by periconceptional vitamin supplementation. N Engl J Med327:1832–1835.

De Girolamo A, Pascale M, Visconti A. 2011. Comparison of methods andoptimisation of the analysis of fumonisins B1 and B2 in masa flour,an alkaline cooked corn product. Food Addit Contam Part A ChemAnal Control Expo Risk Assess 28:667–675.

Dvorak NJ, Riley RT, Harris M, McGregor JA. 2008. Fumonisin mycotoxincontamination of corn-based foods consumed by potentially pregnantwomen in southern California. J Reprod Med 53:672–676.

Ewuola EO, Egbunike GN. 2010. Effects of dietary fumonisin B1 on theonset of puberty, semen quality, fertility rates and testicular morphol-ogy in male rabbits. Reproduction 139:439–445.

Felkner M, Hendricks K, Suarez L, Waller DK. 2003. Diarrhea: a new riskfactor for neural tube defects? Birth Defects Res A Clin Mol Teratol67:504–508.

Felkner M, Suarez L, Liszka B, et al. 2007. Neural tube defects, micronu-trient deficiencies, and Helicobacter pylori: a new hypothesis. BirthDefects Res A Clin Mol Teratol 79:617–621.

Felkner M, Suarez L, Canfield MA, et al. 2009. Maternal serum homocys-teine and risk for neural tube defects in a Texas-Mexico border popu-lation. Birth Defects Res A Clin Mol Teratol 85:574–581.

Felkner MM, Suarez L, Brender J, et al. 2005. Iron status indicators inwomen with prior neural tube defect-affected pregnancies. MaternChild Health J 9:421–428.

Flynn TJ, Stack ME, Troy AL, Chirtel SJ. 1997. Assessment of the embryotoxicpotential of the total hydrolysis product of fumonisin B1 using culturedorganogenesis-staged rat embryos. Food Chem Toxicol 35:1135–1141.

Fraser FC, Fainstat TD. 1951. Production of congenital defects in the off-spring of pregnant mice treated with cortisone; progress report.Pediatrics 8:527–533.

Ganji V, Kafai MR;Third National Health and Nutrition Examination Sur-vey. 2003. Demographic, health, lifestyle, and blood vitamin determi-nants of serum total homocysteine concentrations in the thirdNational Health and Nutrition Examination Survey, 1988–1994. Am JClin Nutr 77:826–833.

Gelineau-van Waes J, Starr L, Maddox J, et al. 2005. Maternal fumonisinexposure and risk for neural tube defects: mechanisms in an in vivomouse model. Birth Defects Res A Clin Mol Teratol 73:487–497.

Gitau R, Cameron A, Fisk NM, Glover V. 1998. Fetal exposure to maternalcortisol. Lancet 352:707–708.

Gong YY, Torres-Sanchez L, Lopez-Carrillo L, et al. 2008. Associationbetween tortilla consumption and human urinary fumonisin B1 levels ina Mexican population. Cancer Epidemiol Biomarkers Prev 17:688–694.

Graham A, Brender JD, Sharkey Jr, et al. 2010. Dietary intake and neuraltube defects in Mexican-American women. Birth Defects Res A ClinMol Teratol 88:451–457.

Hendricks K. 1999. Fumonisins and neural tube defects in South Texas.Epidemiology 10:198–200.

Hendricks KA, Nuno OM, Suarez L, Larsen R. 2001. Effects of hyperinsu-linemia and obesity on risk of neural tube defects among MexicanAmericans. Epidemiology 12:630–635.

Hendricks KA, Simpson JS, Larsen RD. 1999. Neural tube defects alongthe Texas-Mexico border, 1993–1995. Am J Epidemiol 149:1119–1127.

Johnson KM, Suarez L, Felkner MM, Hendricks K. 2004. Prevalence ofcraniorachischisis in a Texas-Mexico border population. Birth DefectsRes A Clin Mol Teratol 70:92–94.

Lahti A, Antila E, Saxen L. 1972. The effect of hydrocortisone on the clo-sure of the palatal shelves in two inbred strains of mice in vivo andin vitro. Teratology 6:37–41.

Lee GR, Herbert V. 1999. Nutritional factors in the production and func-tion of erythrocytes. In: Lee GR, editor. Wintrobe’s Clinical Hematol-ogy. Philadelphia: Lippincott, Williams & Wilkins.

Missmer SA, Suarez L, Felkner M, et al. 2006. Exposure to fumonisinsand the occurrence of neural tube defects along the Texas-Mexicoborder. Environ Health Perspect 114:237–241.

Sadler TW, Merrill AH, Stevens VL, et al. 2002. Prevention of fumonisinB1-induced neural tube defects by folic acid. Teratology 66:169–176.

Shaw GM, Rozen R, Finnell RH, et al. 1998. Maternal vitamin use, geneticvariation of infant methylenetetrahydrofolate reductase, and risk forspina bifida. Am J Epidemiol 148:30–37.

Shaw GM, Schaffer D, Velie EM, et al. 1995. Periconceptional vitamin use,dietary folate, and the occurrence of neural tube defects. Epidemiol-ogy 6:219–226.

Suarez L, Cardarelli K, Hendricks K. 2003a. Maternal stress, social sup-port, and risk of neural tube defects among Mexican Americans. Epi-demiology 14:612–616.

Suarez L, Felkner M, Brender JD, et al. 2008. Maternal exposures to ciga-rette smoke, alcohol, and street drugs and neural tube defect occur-rence in offspring. Matern Child Health J 12:394–401.

10 SUAREZ ET AL.

Birth Defects Research (Part A) 00:000�000 (2012)

Suarez L, Felkner M, Brender JD, Canfield MA. 2012. Dieting to loseweight and occurrence of neural tube defects in offspring of Mexi-can-American women. Matern Child Health J 16:844–849.

Suarez L, Felkner M, Hendricks K. 2004. The effect of fever, febrile ill-nesses, and heat exposures on the risk of neural tube defects in aTexas-Mexico border population. Birth Defects Res A Clin Mol Tera-tol 70:815–819.

Suarez L, Gilani Z, Felkner M, et al. 2005. Exposure to polychlorinatedbiphenyls and risk of neural-tube defects in a Mexican Americanpopulation. Int J Occup Environ Health 11:233–237.

Suarez L, Hendricks K, Felkner M, et al. 2003b. Maternal serum B12 levelsand risk for neural tube defects in a Texas-Mexico border population.Ann Epidemiol 13:81–88.

Suarez L, Hendricks KA, Cooper SP, et al. 2000. Neural tube defectsamong Mexican Americans living on the US-Mexico border: effects offolic acid and dietary folate. Am J Epidemiol 152:1017–1023.

Sun G, Wang S, Hu X, et al. 2011. Co-contamination of aflatoxin B1 andfumonisin B1 in food and human dietary exposure in three areas ofChina. Food Addit Contamin Part A Chem Anal Control Expo RiskAssess 28:461–470.

Torres OA, Palencia E, Lopez de Pratdesaba L, et al. 2007. Estimatedfumonisin exposure in Guatemala is greatest in consumers of low-land maize. J Nutr 137:2723–2729.

United States Public Health Service. 1992. Recommendations for the useof folic acid to reduce the number of cases of spina bifida and otherneural tube defects. MMWR Recomm Rep 41:1–7.

van der Westhuizen L, Shephard GS, Rheeder JR, Burger HM. 2010. Indi-vidual fumonisin exposure and sphingoid base levels in rural popu-

lations consuming maize in South Africa. Food Chem Toxicol48:1698–1703.

Volcik KA, Shaw GM, Lammer EJ, et al. 2003. Evaluation of infant meth-ylenetetrahydrofolate reductase genotype, maternal vitamin use, andrisk of high versus low level spina bifida defects. Birth Defects Res AClin Mol Teratol 67:154–157.

Voss KA, Riley RT, Snook ME, Waes JG. 2009. Reproductive and sphingo-lipid metabolic effects of fumonisin B(1) and its alkaline hydrolysisproduct in LM/Bc mice: hydrolyzed fumonisin B(1) did not causeneural tube defects. Toxicol Sci 112:459–467.

Wadhwa PD, Dunkel-Schetter C, Chicz-DeMet A, et al. 1996. Prenatalpsychosocial factors and the neuroendocrine axis in human preg-nancy. Psychosom Med 58:432–446.

Wang E, Norred WP, Bacon CW, et al. 1991. Inhibition of sphingolipidbiosynthesis by fumonisins. Implications for diseases associated withFusarium moniliforme. J Biol Chem 266:14486–14490.

Xu L, Cai Q, Tang L, et al. 2010. Evaluation of fumonisin biomarkers in a cross-sectional studywith two high-risk populations in China. Food Addit Con-tam Part A ChemAnal Control Expo Risk Assess 27:1161–1169.

Zhu H, Junker WM, Finnell RH, et al. 2003a. Lack of association betweenZIC2 and ZIC3 genes and the risk of neural tube defects (NTDs) inHispanic populations. Am J Med Genet A 116A:414–415.

Zhu H, Wicker NJ, Shaw GM, et al. 2003b. Homocysteine remethylationenzyme polymorphisms and increased risks for neural tube defects.Mol Genet Metab 78:216–221.

Zhu H, Wicker NJ, Volcik K, et al. 2004. Promoter haplotype combina-tions for the human PDGFRA gene are associated with risk of neuraltube defects. Mol Genet Metab 81:127–132.

11NEURAL TUBE DEFECTS ON TEXAS-MEXICO BORDER

Birth Defects Research (Part A) 00:000�000 (2012)