Risk Factors For Cytogenetically NormalHoloprosencephaly in California: A Population-BasedCase-Control Study

Lisa A. Croen,1* Gary M. Shaw,1 and Edward J. Lammer2

1March of Dimes Birth Defects Foundation/California Department of Health Services, California Birth DefectsMonitoring Program, Emeryville, California

2Division of Medical Genetics, Children’s Hospital, Oakland, California

Holoprosencephaly is a developmental fielddefect manifested by a spectrum of abnor-malities of the forebrain and midface. Ap-proximately 50% of holoprosencephalycases are associated with a cytogenetic ab-normality or a monogenic syndrome. Sug-gested risk factors for the remaining 50% ofcases have been described in case reports,but have not been confirmed in systemati-cally conducted studies. We report the re-sults of a population-based case-controlstudy of holoprosencephaly. Live births, fe-tal deaths, and terminations with a diagno-sis of cytogenetically normal holoprosen-cephaly were identified by the CaliforniaBirth Defects Monitoring Program. Tele-phone interviews were conducted with themothers of 58 cases and 107 live born, non-malformed controls. Women were ques-tioned about their health and reproductivehistories, family demographics, and expo-sures occurring during their pregnancies.Among nonsyndromic cases, increased riskswere observed for females (OR=1.8, 95% C.I.0.9–3.9), foreign-born vs. U.S. or Mexico-born women (OR=3.1, 95% C.I. 1.1–8.6), andwomen with early menarche (OR=2.3,95%C.I. 0.9–5.7). Maternal periconceptional ex-posures associated with increased risks fornonsyndromic holoprosencephaly includedalcohol consumption (OR=2.0, 95% C.I. 0.9–4.5), cigarette smoking (OR=4.1, 95% C.I. 1.4–12.0), and combined alcohol and smoking(OR=5.4, 95% C.I . 1.4–20.0) , insulin-dependent diabetes (OR=10.2, 95% C.I. 1.9–

39.4), medications for respiratory illnesses(OR=2.3, 95% C.I. 0.9–6.0), and salicylate-containing medications (OR=2.5, 95% C.I.0.8–7.9). These findings are consistent withrisk factors identified in some previous re-ports, and identify several new potentialrisk factors that require confirmation in fu-ture studies. Am. J. Med. Genet. 90:320–325,2000. © 2000 Wiley-Liss, Inc.

KEY WORDS: holoprosencephaly; epidemi-ology; risk factors; alcohol;smoking; diabetic embryopa-thy

INTRODUCTION

Holoprosencephaly is a developmental field defect in-volving the brain and face. It results from the incom-plete cleavage of the embryonic forebrain prior to thefourth week of gestation. It is manifested by a spec-trum of abnormalities of the forebrain and midfaceranging from mild (arrhinencephaly, hypotelorism,single central incisor) to severe (alobar holoprosen-cephaly, cebocephaly, cyclopia) [DeMyer et al., 1964].While the occurrence of holoprosencephaly among livebirths and fetal deaths is uncommon (1.2 per 10,000births) [Croen et al., 1996], it is one of the most com-mon defects of the developing brain, occurring in ap-proximately 4 per 1,000 conceptuses [Matsunaga andShiota, 1977].

Approximately 50% of livebirths and fetal deathswith holoprosencephaly are associated with a cytoge-netic abnormality or a monogenic syndrome [Croen etal., 1996; Cohen, 1989]. A few risk factors for cytoge-netically normal or nonsyndromic holoprosencephalyhave been suggested in case reports or small case se-ries, but none have been investigated in a population-based epidemiologic study. Suggested risk factors in-clude maternal history of diabetes [Barr et al., 1983],previous pregnancy loss [Matsunaga and Shiota, 1977],poverty [Roach et al., 1975], prenatal exposure to medi-cations such as retinoic acid [Lammer et al., 1985], sa-

Contract grant sponsor: the Department of Health and HumanServices, Public Health Service, Centers for Disease Control andPrevention; Contract grant number: U50/CCU908501.

*Correspondence to: Lisa A. Croen, Ph.D., California Birth De-fects Monitoring Program, 1900 Powell Street, Suite 1050, Em-eryville, CA 94608.

Received 29 August 1999; Accepted 4 October 1999

American Journal of Medical Genetics 90:320–325 (2000)

© 2000 Wiley-Liss, Inc.

licylates [Khudr and Olding, 1973; Agapitos et al.,1986; Benawra et al., 1980], anticonvulsants [Rosa,1995; Kotzot et al., 1993; Holmes and Harvey, 1994],estrogen/progestin [Stabile et al., 1985], and prenatalinfection with cytomegalovirus [Byrne et al., 1987],toxoplasmosis [Lison et al., 1967], and rubella [Probst,1979].

We conducted a case-control study to examine theassociations between a variety of demographic, repro-ductive, lifestyle, and medical factors and the occur-rence of cytogenetically normal holoprosencephalyamong a population-based sample of California births.

MATERIALS AND METHODS

Cases were ascertained by the California Birth De-fects Monitoring Program (CBDMP) by reviewingmedical records at all hospitals and medical geneticscenters serving the population of 986,197 births occur-ring to women residing in one of eleven Californiacounties [Croen et al., 1991]. Considered for potentialinclusion as cases were live births, fetal deaths ($ 20weeks of gestation), and terminations (at any time dur-ing pregnancy) with a diagnosis of holoprosencephaly,possible holoprosencephaly, arrhinencephaly, probos-cis or single nostril, cyclopia, aprosencephaly, or holo-telencephaly with an actual or expected date of deliv-ery between January 1, 1993 and July 31, 1996. Amedical geneticist (EJL) determined study eligibilityafter review of prenatal ultrasonographic, brain imag-ing, autopsy, and cytogenetic reports. Only infants orfetuses with incomplete cleavage of the cerebral fore-brain were considered to have holoprosencephaly. In-fants or fetuses with a chromosomal anomaly con-firmed by cytogenetics were excluded. Cases were alsoexcluded if the forebrain malformation resulted fromamniotic band disruption, porencephaly, or schizen-cephaly.

Cases were classified as “isolated”, “multiple”, or“syndromic” based upon the presence and pattern ofaccompanying anomalies and family history informa-tion. Cases with no other anomaly, or with craniofacialmalformations that are considered part of the holopros-encephaly field defect were considered “isolated”. Caseswith accompanying major malformations were consid-ered “multiple”. Cases with diagnoses consistent with aknown monogenic disorder or recognizable pattern ofmalformation, or infants with a family history consis-tent with one of the dominant or recessive holoprosen-cephaly syndromes were classified as “syndromic”[Siebert et al., 1990].

Control infants were randomly selected from hospi-tals in the study area in proportion to the hospital’smonthly contribution to the total number of infantsborn alive in the area between January 1993 and July1996. The California Vital Statistics live birth com-puter files for the years 1991–1994 were used to esti-mate the number of births occurring at each hospital ineach month of the study period. Infants were eligible ascontrols if they were born alive, their mothers wereresidents of one of the eleven study counties at the timeof delivery, and they did not have a reportable birthdefect apparent at the time of birth.

Telephone interviews were conducted with the moth-ers of cases and controls in either English (51.5%) orSpanish (48.5%). Women who only spoke languagesother than English or Spanish were excluded (1 caseand 4 controls). Interviews were completed with themothers of 58 (68.2%) of 85 eligible cases and 107(70.4%) of 152 eligible controls an average of 19 monthsfor cases and 13 months for controls after the actual orestimated date of term delivery. Information was un-available for 8.2% of case and 11.2% of control motherswho refused to be interviewed and for 23.5% of case and18.4% of control mothers who could not be located.

The hour and a half interview elicited information onmaternal health and reproductive history, family de-mographics and medical history, and exposures andevents occurring between the three months before con-ception and the end of the index pregnancy. The peri-conceptional period was defined as the three monthsbefore conception through the end of the first trimester.Since women are often not aware that they are preg-nant during the first several weeks of their pregnan-cies, and recall of the timing of exposures is often notprecise, exposures reported during this six month timeperiod were used as a proxy for exposures potentiallyoccurring during the relevant embryologic time period.

The odds ratio (OR) and its 95% confidence interval(CI) was used to estimate potential risks for holopros-encephaly associated with parental and infant charac-teristics. ORs were computed only when there were aminimum of 10 exposed cases and controls combined.

RESULTS

Among the 58 interviewed cases, 35 (60.3%) wereisolated, 13 (22.4%) were multiple, and the remaining10 (17.2%) were syndromic. The syndromic cases in-cluded: suspected trisomy 13—cytogenetic analysiswas not performed for these cases (n44), autosomalrecessive holoprosencephaly (n42), chromosome find-ings of 46,XY,7p+ (n41), central myotubular myopathy(n41). The remaining 2 syndromic cases had previ-ously unrecognized syndromes: holoprosencephalywith split hand, anomalous lung lobation, imperforateanus, unilateral kidney with hydronephrosis, andtibial hypoplasia (n41), and holoprosencephaly withunilateral postaxial polydactyly, absent sternum, andbicornuate uterus (n41). Table I presents informationon the birth status and the brain and facial phenotypesof all interviewed cases.

Selected infant and maternal characteristics for non-syndromic cases (n448) and controls are shown inTable II. Compared to controls, cases tended to be fe-male (OR41.8), and case mothers were more likely tobe Hispanic (OR41.6), born outside the U.S. andMexico (OR43.1), and have less than (OR42.1) ormore than (OR41.9) a high school education. With theexception of maternal birthplace, confidence intervalsaround these estimates included unity. Among the 4cases and 4 controls with Asian mothers, 3 cases had aPhilippine-born mother as compared to only one con-trol. No variability in risk was observed across the pa-ternal race/ethnicity or maternal age categories.

As shown in Table III, ORs of 1.5 or greater were

Risk Factors For Holoprosencephaly 321

identified for number of previous live births $ 3, age atmenarche #11 years of age, gravidity $ 3, and prepreg-nancy BMI > 29 kg/m2. However, confidence intervalsaround these estimates included unity.

Table IV displays frequencies and risk estimates fora variety of exposures occurring during the periconcep-tional period. Increased risks were observed for womenwho drank any alcohol (OR42.0, 95% C.I. 0.9–4.5), orwho smoked any cigarettes (OR44.1, 95% C.I. 1.4–12.0). The majority of drinkers drank less than onedrink per week (72% of cases, 60% of controls). Like-wise, the majority of smokers smoked less than one

pack per day (92% of cases, 62.5% of controls). Drinkingand smoking are often highly correlated behaviors sowe analyzed combined exposures to both. In thissample, alcohol consumption among smokers showed aslightly higher odds ratio (OR41.8, 95% C.I. 0.2–18.8)than among nonsmokers (OR41.3, 95% C.I. 0.5–3.6),although confidence intervals included unity for both

TABLE II. Demographic Characteristics of NonsyndromicHoloprosencephaly Cases and Controls

ParametersCases

(n 4 48)Controls

(n 4 107)OR

(95% CI)

Child sexMale 18 56 1.0 refFemale 30 51 1.8 (0.9–3.9)

Maternal race/ethnicityWhite 6 21 1.0 refHispanic 35 76 1.6 (0.5–4.9)Black 2 4Asian 4 4Other 1 2

Paternal race/ethnicityWhite 7 16 1.0 refHispanic 34 75 1.0 (0.4–3.1)Black 1 8Asian 3 2Other 3 6

Maternal birthplaceU.S.A. 16 49 1.0 refMexico 18 44 1.2 (0.5–3.0)Elsewhere 14 14 3.1 (1.1–8.6)

Maternal age<20 5 13 0.9 (0.2–2.9)20–34 36 83 1.0 ref>34 6 11 1.3 (0.4–4.1)

Maternal education<High school 23 42 2.1 (0.8–6.0)High school graduate 8 31 1.0 ref>High school 17 34 1.9 (0.7–5.7)

TABLE III. Pregnancy and Reproductive Characteristics ofNonsyndromic Holprosencephaly Cases and Controls*

ParametersCases

(n 4 48)Controls

(n 4 107) OR (95% CI)

Number previouslive births0 17 35 1.0 ref1 15 42 0.7 (0.3–1.8)2 5 17 0.6 (0.2–2.2)$3 11 13 1.7 (0.6–5.3)

Prior pregnancy lossa

0 19 48 1.0 ref$1 17 34 1.3 (0.5–3.0)

Age at menarche#11 15 16 2.3 (0.9–5.7)12–14 30 74 1.0 ref$15 3 16 0.5 (0.1–1.9)

Gravidity1 8 25 1.0 ref2 12 34 1.1 (0.3–3.5)3 10 21 1.5 (0.4–5.1)4 18 27 2.1 (0.7–6.4)

Prepregnant BMI(Kg/m2)#29 34 83 1.0 ref>29 10 16 1.5 (0.6–4.0)

Family history ofbirth defectsb

No 43 100 1.0 refYes 3 2

*BMI, Body mass index.aOnly women with prior pregnancies included.bIn first degree relatives.

TABLE IV. Occurrence of Periconceptional* Exposures AmongNonsyndromic Holoprosencephaly Cases and Controls

ParametersCases

(n 4 48)Controls

(n 4 107) OR (95% CI)

Folic acid supplement useNo 12 33 1.0 refYes 35 68 1.4 (0.6–3.3)

Alcohol consumptionNone 29 81 1.0 refAny 18 25 2.0 (0.9–4.5)

SmokingNo 36 98 1.0 refYes 12 8 4.1 (1.4–12.0)

Recreational drug usea

None 38 80 1.0 refAny 10 27 0.8 (0.3–1.9)

Birth-control pill useNo 43 95 1.0 refYes 5 12 0.9 (0.3–3.0)

*Defined as the 6-month period from 3 months before to 3 months afterconception.aIncludes use of marijuana, hash, cocaine, crack, heroin, methadone, crys-tal methedrine, MDA, speedball, LSD, ecstasy, mescaline, PCP, mush-rooms, amyl nitrate, nitrous oxide, and inhalation of glue, paint, hairspray,white out, freon, and gasoline.

TABLE I. Phenotypic Characteristics ofHoloprosencephaly Cases

ParametersIsolated(n 4 35)

Multiple(n 4 13)

Syndromic(n 4 10)

Anatomic typeAlobar 16 5 2Semilobar 13 3 3Lobar 4 3 1Unknown 2 2 4

Facial phenotypeCyclopia 2 2 2Ethmocephaly 1 0 0Cebocephaly 12 3 3Median CL(P) 4 1 1Agnathia 5 1 0Less severe 11 6 2Unknown 0 0 2

Birth typeLive born 29 9 3Stillborn 3 2 2Termination 3 2 5

322 Croen et al.

estimates. Smoking among drinkers was associatedwith a 4-fold increased risk (95% C.I. 0.9–19.4), andsmoking among non-drinkers a 3-fold increased risk(95% C.I. 0.4–20.2). Confidence limits were wide forthese estimates as well. Compared to women who nei-ther smoked nor drank in the periconceptional period,women who both drank and smoked (9 cases and 5controls) were over 5 times as likely to have a babywith holoprosencephaly (OR45.4, 95% C.I. 1.5–20.7).Approximately half of these women consumed severalalcoholic beverages per week. Risks associated with theuse of folic acid containing vitamin supplements, rec-reational drugs, and birth control pills were consistentwith unity.

Risks associated with maternal illnesses and medi-cation use are displayed in Table V. The mothers of 12cases and 15 controls reported having ever been told bya doctor that they had diabetes (OR42.0, 95% C.I. 0.8–5.2). Among those diabetic women who reported takinginsulin during the index pregnancy, a 10-fold risk wasobserved (OR410.2, 95% C.I. 1.9–39.4). Among theeight cases whose mothers took insulin, 6 had alobarholoprosencephaly with an accompanying severe facialmalformation [cyclopia (n41), cebocephaly (n43), pro-boscis (n41), median cleft lip (n41)]. Among womenwith gestational diabetes the risk for holoprosen-cephaly was 2.0 (95% C.I. 0.7–5.3). Women reporting ahistory of anemia had over a 3-fold risk (OR43.4, 95%C.I. 1.0–11.8). Case mothers were twice as likely ascontrol mothers to report a history of respiratory ill-nesses during the periconceptional period (OR42.0,95% C.I.40.9–4.5). The majority of respiratory ill-nesses reported were colds. Among those women whotook medication for their respiratory illnesses, the OR

was 2.3 (95% C.I. 0.9–6.0). Most women took severalmedications that included common over-the-countercold formulas. Women who reported taking aspirin orsalicylate-containing medication for any reason duringthe periconceptional period were at over twice the riskfor delivering an infant with nonsyndromic holopros-encephaly (OR42.5, 95% C.I. 0.8–7.9). Among theeight exposed cases, 4 women reported taking aspirinfor headaches, 2 for colds, 1 for nausea, and 1 for diz-ziness. Three of the eight reported chronic use for atleast 3 months during the periconceptional period.

DISCUSSION

This is the first population-based case-control studyto investigate a variety of potential risk factors for cy-togenetically normal holoprosencephaly. Previouslypublished observations were based on case reports orsmall case series that lacked control populations, andused differing case inclusion criteria with respect tobirth type (e.g., live birth, fetal death, termination), orcase type (e.g., cytogenetically normal, syndromic).Thus, the results of this investigation cannot be di-rectly compared to past findings.

In this sample, we found that cytogenetically normal,nonsyndromic holoprosencephaly cases were nearlytwice as likely to be female as male. Our results areconsistent with a report by Roach et al. [1975] whoobserved a female predominance (M:F410:22) among aclinic-based series of 32 infants with severe holopros-encephaly not associated with a chromosomal abnor-mality. A female excess was also present in two popu-lation-based samples [Rasmussen et al., 1996; Olsen etal., 1997], the first of which included cytogeneticallyabnormal and syndromic cases. Our results contrastwith our earlier published finding of equal risk for fe-males and males among a California sample restrictedto livebirths and fetal deaths with cytogenetically nor-mal holoprosencephaly [Croen et al., 1996]. The inclu-sion of terminations in the present sample did not ac-count for the inconsistency with our previouslypublished findings because the female excess was ob-served among the subgroup of livebirths and fetaldeaths as well. There were also no differences betweenmales and females in the percent whose mothers par-ticipated in the interview.

Our observation of equal risks across the maternalage spectrum is consistent with several previous re-ports [Matsunaga and Shiota, 1977; Roach et al., 1975;Rasmussen et al., 1996; Croen et al., 1996]. Our find-ings of an increased risk associated with maternal dia-betes and use of insulin during pregnancy are consis-tent with the report by Barr et al. [1983] whosummarized the occurrence of 22 cases of holoprosen-cephaly in the infants of diabetic mothers. Our findingof a nearly two-fold increase in risk among women withgestational diabetes is also consistent with the reportby Martínez-Frías et al. [1998] who examined thespecificity between the association of gestational dia-betes and 20 types of congenital anomalies, includingholoprosencephaly. Our observation of an increasedrisk among women who took aspirin or salicylate con-taining medication during the periconceptional period

TABLE V. Odds Ratios and 95% Confidence Intervals forNonsyndromic Holoprosencephaly Associated With Selected

Maternal Illnesses and Medication Use

Parameters Cases Controls ORa 95% CI

Diabetesb 12 15 2.0 0.8–5.2Medication usedc 8 2 10.2 1.9–39.4No medication used 4 13 0.8 0.2–2.8

Gestational diabetesb 10 13 2.0 0.7–5.3High blood pressureb 8 12 1.6 0.5–4.6Anemiab 8 6 3.4 1.0–11.8Skin problemd 7 14 1.1 0.4–3.3Allergiesd 9 20 1.0 0.4–2.6Respiratory illnesse 17 23 2.0 0.9–4.5

Mediation used 13 15 2.3 0.9–6.0No medication used 4 8 1.3 0.3–5.5

Urinary tract infectiond 10 16 1.5 0.6–3.9Genital herpesd 2 1Chlamydiad 2 1Injuryd 4 10 0.9 0.2–3.3Surgeryd 2 1X-ray or scan 8 17 1.1 0.4–2.9Take aspirin/salicylatese 8 8 2.5 0.8–7.9

aReference group for calculation of the odds ratio is persons without aparticular condition. For an odds ratio associated with medication use, thereference group is persons who did not use that particular medication andwho did not have that particular condition.bRefers to ever having a history of that condition not limited to the firsttrimester.cMedication used in the first trimester.dDuring index pregnancy.eDuring periconceptional period of index pregnancy.

Risk Factors For Holoprosencephaly 323

is consistent with a case report of cyclopia in an infantwhose mother had taken aspirin daily throughout thefirst trimester of pregnancy [Benawra et al., 1980].

To follow up on the case report of holoprosencephalyfollowing in utero exposure to retinoic acid [Lammer etal., 1985], a European study was conducted using datafrom regional birth defect registries affiliated with theEURACAT program [De Wals et al., 1991]. Among 16infants with cytogenetically normal holoprosen-cephaly, no mother reported any skin disease or reti-noid use. Similarly, in our study population, no womanreported using isotretinoin or any topical preparationof retinoic acid during her pregnancy. Eight holopros-encephaly cases with maternal use of anticonvulsantshave been reported [Rosa, 1995; Kotzot et al., 1993;Holmes and Harvey, 1994]. However, no woman in ourstudy population reported a history of epilepsy or sei-zures, or use of anticonvulsant medication. While ourdata do not support a causal association between holo-prosencephaly and exposures to retinoids or anticon-vulsants, the low frequency of these exposures (1 in 400pregnancies for retinoic acid; 1 in 250 births for anti-convulsants [Holmes and Harvey, 1994]) coupled withthe relatively small size of our study population re-sulted in low power to detect a modestly elevated risk.

This is the first epidemiologic study to suggest a linkbetween maternal alcohol and cigarette exposure dur-ing the periconceptional period and holoprosencephalyrisk. In contrast to the reports of 5 holoprosencephalycases born to women with heavy maternal alcohol in-take early in their pregnancies [Jellinger et al., 1982;Bonnemann and Meinecke, 1990; Ronen and Andrews,1991], our data suggest that combined low dose expo-sure to alcohol and cigarettes may pose a substantialrisk. While there is experimental evidence to supportan embryonic alcohol effect for holoprosencephaly [Su-lik and Johnston, 1982; Webster, 1983], we are notaware of any experimental data demonstrating the ef-fect of smoking by itself, or the combined effect of smok-ing and drinking on the risk of holoprosencephaly.

This is also the first study to suggest that maternalbirthplace, maternal education level, and young age atmenarche might be associated with an increased riskfor holoprosencephaly. Our observation of increasedrisks associated with respiratory illnesses and medica-tion use for these illnesses during the periconceptionalperiod, and history of maternal anemia have also notbeen previously reported. The biologic significance ofthese associations is not known.

Our findings must be interpreted in the context ofsome important methodologic limitations. First, wewere only able to identify and interview approximately70% of the eligible study population. If the non-interviewed cases and controls differed substantiallyfrom the interviewed study population, our findingscould have been biased. Second, our use of the six-month periconceptional exposure time period may haveresulted in misclassification of exposure, particularlyacute exposures, such as infections or medication use.However, we would not expect the magnitude of thismisclassification to differ for cases and controls, andtherefore the effect would be to bias results toward thenull value. The six-month exposure time period is

likely to more adequately represent long term orchronic exposures, such as cigarette smoking and alco-hol consumption. Third, the study population was verysmall, and most reported associations did not reachstatistical significance. Furthermore, the small samplesize precluded the utility of conducting multivariateanalyses.

Before any of these suggested risk factors can be con-sidered etiologically related to nonsyndromic holopros-encephaly, the associations presented here will requirereplication in a larger study which incorporates moreprecise measurements of exposure.

ACKNOWLEDGMENTS

This study was supported by Cooperative AgreementNo. U50/CCU908501, Dept. of Health and Human Ser-vices, Public Health Service, Centers for Disease Con-trol and Prevention. We are grateful to Kimberly Mor-land for her assistance with dataset development, andto Neachan Harvey for manuscript preparation.

REFERENCES

Agapitos M, Georgian-Theodoropoulou M, Koutselinis A, Papacharalam-pus N. 1986. Cyclopia and maternal ingestion of salicylate. PediatrPathol 6:309–310.

Barr M Jr, Hanson JW, Currey K, Sharp S, Toriello H, Schmickel RD,Wilson GN. 1983. Holoprosencephaly in infants of diabetic mothers. JPediatr 102:565–568.

Benawra R, Mangurten HH, Duffell DR. 1980. Cyclopia and other anoma-lies following maternal ingestion of salicylates. J Pediatr 96:1069–1071.

Bonnemann C, Meinecke P. 1990. Holoprosencephaly as a possible embry-onic alcohol effect: another observation. Am J Med Genet [Letter] 37:431–432.

Byrne PJ, Silver MM, Gilbert JM, Cadera W, Tanswell AK. 1987. Cyclopiaand congenital cytomegalovirus infection. Am J Med Genet 28:61–65.

Cohen MM Jr. 1989. Perspectives on holoprosencephaly: Part I. Epidemi-ology, genetics, and syndromology. Teratology 40:211–235.

Croen LA, Shaw GM, Jensvold NG, Harris JA. 1991. Birth defects moni-toring in California: a resource for epidemiological research. PaediatrPerinat Epidemiol 5:423–427.

Croen LA, Shaw GM, Lammer EJ. 1996. Holoprosencephaly: epidemiologicand clinical characteristics of a California population. Am J Med Genet64:465–472.

De Wals P, Bloch D, Calabro A, Calzolari E, Cornel MC, Johnson Z, LiguticI, Nevin N, Pexieder T, Stoll C, Tenconi R, Tilmont P. 1991. Associationbetween holoprosencephaly and exposures to topical retinoids: resultsof the EUROCAT survey. Paediatr Perinat Epidemiol 5:445–447.

Demyer W, Zeman W, Palmer CG. 1964. The face predicts the brain: di-agnostic significance of median facial anomalies for holoprosencephaly(arhinencephaly). Pediatrics 34:256–263.

Holmes LB, Harvey EA. 1994. Holoprosencephaly and the teratogenicity ofanticonvulsants. Teratology 49:82.

Jellinger K, Gross H, Kaltenback E. 1981. Holoprosencephaly and agenesisof the corpus callosum: frequency of associated malformations. ActaNeuropathol (Berlin) 55:1–10.

Khudur G, Olding L. 1973. Cyclopia. Am J Dis Child 125:120–122.

Kotzot D, Weigl J, Huk W. 1993. Hydantoin syndrome with holoprosen-cephaly: a possible rare teratogenic effect. Teratology 48:15–19.

Lammer EJ, Chen DT, Hoar RM, Agnish ND, Benke PJ, Braun JT, CurryCJ, Fernhoff PM, Grix AW, Lott IT, Richard JM, Sun SC. 1985. Reti-noic acid embryopathy. N Engl J Med 313:837–841.

Lison MP, Armbrust-Figueiredo J, Mega D. 1967. Arhinencephalia: con-siderations a propos a case diagnosed during life. Acta Neurol Psychi-atr (Belgium) 67:25–36.

324 Croen et al.

Martínez-Frías ML, Bermejo E, Rodríguez-Pinilla E, Prieto L, Frías JL.1998. Epidemiological analysis of outcomes of pregnancy in gestationaldiabetic mothers. Am J Med Genet 78:140–145.

Matsunaga EI, Shiota K. 1977. Holoprosencephaly in human embryos:epidemiologic studies of 150 cases. Teratology 16:261–272.

Olsen CL, Hughes JP, Youngblood LG, Sharpe-Stimac M. 1997. Epidemi-ology of holoprosencephaly and phenotypic characteristics of affectedchildren: New York State, 1984–1989. Am J Med Genet 73:217–226.

Probst FP. 1979. The prosencephalies: morphology, neuroradiological ap-pearances and differential diagnosis. Berlin: Springer-Verlag. p. 1–25.

Rasmussen SA, Moore CA, Khoury MJ, Cordero JF. 1996. Descriptive epi-demiology of holoprosencephaly and arhinencephaly in metropolitanAtlanta, 1968–1992. Am J Med Genet 66:320–333.

Roach E, Demyer W, Conneally PM, Palmer C, Merritt AD. 1975. Holo-prosencephaly: birth data, genetic and demographic analyses of 30families. Birth Defects XI:294–313.

Ronen GM, Andrews WL. 1991. Holoprosencephaly as a possible embryonicalcohol effect. Am J Med Genet 40:151–154.

Rosa F. 1995. Holoprosencephaly and antiepileptic exposures. Teratology51:230.

Siebert JR, Cohen MM, Sulik KK, Shaw C-M, Lemire RJ. 1990. Holopros-encephaly: an overview and atlas of cases. New York: Wiley-Liss. p1–397.

Stabile M, Bianco A, Iannuzzi S, Buonocore MC, Ventruto V. 1985. A caseof suspected teratogenic holoprosencephaly. J Med Genet 22:147–149.

Sulik KK, Johnston MC. 1982 Embryonic origin of the fetal alcohol syn-drome: Interrelationship of the developing brain and face. Scan Elec-tron Microsc 1:309–322.

Webster WS. 1983 Some teratogenic properties of ethal acetaldehyde inC57BL/6J Mice: Implications for the study of fetal alcohol syndrome.Teratology 27:231–243.

Risk Factors For Holoprosencephaly 325