Fetal ADH2*3,Maternal Alcohol Consumption,and Fetal Growth

Abstract

There is some evidence suggesting the allele for alcohol dehydrogenase 2*3 (ADH2*3) is associated with a protective effect against alcohol-related intrauterine growth retardation (IUGR). This study was conducted to explore the affect of the ADH2*3 allele on fetal growth. Bloodspots (n =1016) belonging to individual infants of a subgroup of the Baltimore-Washington Infant Study (BWIS) were assayed for the presence of the ADH2*3 allele by a polymerase chain reaction (PCR)-based method. Infants genotyped for ADH2*3 were those for whom bloodspots were identified and obtained from the Maryland Newborn Screening Program. The effect of ADH2*3 and maternal alcohol consumption on intrauterine growth was explored by multivariable linear regression analysis. Twenty-six percent of the 306 blood spots belonging to African-American infants were positive for ADH2*3 (4% were homozygous and 22% were heterozygous). Only a small percentage of bloodspots for Caucasian (1.3%) were positive for the ADH2*3 allele. Consequently, further analysis concentrated on gene-exposure interactions for African-American infants. It was found that the incidence of being small-for-gestation-age (SGA) was lower for ADH2*3-positive infants (2.5% versus 8.8%; p =.08). SGA infants had elevated odds for being ADH2*3 negative (OR: 3.15, 95% C.I.: 0.70–14.26) and for being born to mothers that consumed alcohol during pregnancy (OR: 2.31, 95% C.I.: 0.77–6.91). A negative trend between maternal alcohol consumption and mean offspring birthweight was found; however, ADH2*3 did not have a significant impact on mean birthweight for infants born to mothers that drank during pregnancy. These results could be interpreted as possible support for the hypothesis that ADH2 genotype in the infant may impact risk for alcohol-related IUGR. However, this study has limitations in that it is a “nested study of convenience” and involves a relatively small number of infants born to mothers reporting moderate to heavy alcohol use during pregnancy.

Despite warnings about the negative effects of alcohol consumption during pregnancy, a significant proportion of U.S. women report drinking alcohol during pregnancy (MMWR 1997, Ventura et al. 1999). Fetal alcohol exposure is estimated to occur in about 1% or 40,000 pregnancies in the United States (Ventura et al. 1999). Intrauterine growth retardation (IUGR) is associated with fetal alcohol exposure (Jones and Smith 1973; Little 1977; Pytkowicz-Streisguth et al. 1980, 1986; Institute of Medicine [IOM] 1996). However, there is no consensus on whether low to moderate periconceptional maternal alcohol consumption poses a significant risk for IUGR or giving birth to small-for-gestational-age (SGA) infants (Mills et al. 1984; Smith et al. 1986).

The rate at which alcohol is metabolized and cleared from the body may be an important variable that influences the risk for alcohol-related birth defects (McCarver et al. 1997). There is significant variation in human alcohol metabolism rates of which 50% may be genetically determined (Bosron and Li 1986). Polymorphism at the alcohol dehydrogenase 2 (ADH2) allele may account for this variability (Bosron and Li 1986). In humans, ADH2 is involved in the rate-limiting step in the conversion of ethanol to acetaldehyde (Bosron and Li 1986; Bosron, Lumeng, and Li 1988). There are three isozymes of ADH2 encoded by three ADH2 alleles. ADH2*1 encodes for isozyme β₁, ADH2*2 encodes for isozyme β₂, and ADH2*3 encodes for isozyme β₃. Persons expressing the β₂ or β₃ variants are expected to have faster ethanol clearance rates than persons expressing the β₁ variant only (e.g., β₁β₁) (Bosron and Li 1986; Thomasson, Beard, and Li 1995). ADH2 polymorphism has been related to alcohol consumption frequency (Neumark et al. 1988), risk for alcohol dependence (Ehlers et al. 2001; Chao et al. 1997; Shen et al. 1997; Tanaka et al. 1997; Whitfield 1997), risk for alcohol-related liver disease (Tanaka et al. 1997; Whitfield 1997; Yamauchi et al. 1995), and risk for alcoholic brain atrophy (Maezawa et al. 1996). ADH2 is expressed in the fetus beginning in the 2nd trimester of pregnancy (Smith, Hopkinson, and Harris 1971, 1973; Pikkarainen and Raiha 1967).

ADH2*1 is the most common allele among Caucasians and African-Americans with frequencies of 0.85 to 1.00 and 0.66 to 0.85, respectively (Bosron and Li 1986; Bosron, Lumeng, and Li 1988; Iron et al. 1992; Suzuki et al. 1994; McCarver et al. 1997). ADH2*2 is most common among persons of Japanese descent (F: 0.65 to 0.85) and occurs at a frequency of less than 0.05 among Caucasian- and African-Americans (Bosron and Li 1986). ADH2*3 is most common among persons of African-American descent (F: 0.15 to 0.33) and occurs among Caucasian and persons of Southeast Asian descent at a frequency of <0.05 (Bosron and Li 1986; McCarver et al. 1997; Roychoudhury and Nei 1998).

Current evidence suggests that ADH2*3 may protect the fetus against alcohol teratogenicity, possibly as a result of higher fetal or maternal ethanol (ETOH) oxidation afforded by ADH2*3 (McCarver et al. 1997). It is also thought that alcohol may cause teratogencity by disrupting cellular retinoic acid (RA) homeostasis (Whitmire et al. 1995; Ang et al. 1996; Duester 1996). ADH2 isozymes convert retinol to retinal in vitro at different rates (Yang et al. 1994). Therefore, ADH2 polymorphism may have a direct impact on late gestation fetal growth and development through the mechanism of ADH-mediated retinoid metabolism. The negative impact of exogenous retinoids on fetal growth and development has been documented (Lammer et al. 1985). In this study, we tested two hypotheses. We tested the hypothesis that fetal ADH2*3 status (e.g., positive or negative) has a significant influence on population mean infant birthweight and on the odds for being born SGA based on its possible involvement in RA metabolism. We also tested the hypothesis that ADH2*3 in the infant is protective against population mean birthweight deficits and/or decreased odds for being categorized SGA among infants with possible alcohol exposure during gestation.

MATERIALS AND METHODS

Study Population

Bloodspots assayed for ADH2*3 were from Maryland-born infants enrolled in the Baltimore-Washington Infant Study (BWIS), a population-based case-control study of the potential risk factors for congenital heart defects (Ferencz et al. 1993). Parents of cases and controls were interviewed at home by trained personnel within 1 year of the infant’s birth and asked questions concerning medical and demographic/socioeconomic factors as well as questions on cigarette smoking and alcohol consumption. Maternal alcohol consumption and cigarette smoking was indexed to three time periods: 3 months before the last menstrual period and the first 3 months of pregnancy, the 2nd trimester, and the 3rd trimester of pregnancy. Infants were categorized as SGA if their gender- and gestational age–specific birth weights were below the 10th percentile of the population fetal growth curve reported by Brenner, Edelman, and Hendricks (1976) for infants in the United States.

Bloodspots used in this study were retrieved for as many BWIS controls as possible from the Maryland Newborn Screening Program (NSP) (Loffredo and Ewing 1997). Participation in the NSP by parents is voluntary and requires informed parental consent (Panny and Bernhardt 1989). Consent allows blood samples to be used in medical studies once screening tests are completed. Permission to use BWIS blood spots for this study was granted by the Maryland Department of Health and Mental Hygiene (DHMH) and the University of Maryland Institutional Review Board. The study methods used in the BWIS and in our nested study conform to the principles embodied in the World Medical Association Declaration of Helsinki (World Medical Association General Assembly 2002). Infants with chromosomal anomalies, single gene disorders, and genetic syndromes were identified through medical records prior to genotyping and were excluded from the study population.

Identification of ADH2*3-Positive Infants

BWIS control blood spots were genotyped for ADH2*3 using a polymerase chain reaction (PCR)-based method (Groppi, Begueret, and Iron 1990). Investigators were blinded to the identification of bloodspot donors until all genotyping was completed. Bloodspots were prepared for genotyping following previously published methods (Makowski, Aslanzaden, and Hofer 1995). The PCR product was digested with AluI (50 U/μl) in Tris buffer (50 mM Tris-HCl at pH 8.0, 10 mM MgCl₂) for 2 h at 37°C. The digest products were separated by gel electrophoresis, stained with ethidium bromide, and visualized under ultraviolet (UV) light. As a quality control measure, the 236-bp PCR product from 10 ADH2*3-negative, 10 ADH2*3-heterozygous, and 10 ADH2*3-homozygous blood spots were sequenced to confirm that the PCR-based method gave correct genotype information. Nucleotide sequences from around the 236 bp AluI cut site were compared with published sequences (Groppi, Begueret, and Iron 1990) for the ADH2*3 allele.

Data Analysis

At the completion of all genotyping, descriptive population characteristics for infants genotyped for ADH2*3 were generated using SAS (SAS Institute, Carey, North Carolina) from BWIS interview data and medical records. Infants of multiple births, of unknown birthweight, or unknown gestation length were excluded from analysis. Caucasians and infants of other racial background (Southeast Asian, Southwest Asian, and Native American Indian) had comparatively low frequencies for the ADH2*3 allele (e.g., 0.01 and 0.00, respectively). As a result, further analysis concentrated on the interaction of maternal alcohol consumption during pregnancy, infant ADH2*3 status, and the combination of these factors on mean birthweight of African-American infants.

African-American infants were stratified by ADH2*3 status (positive or negative). Heterozygous infants were categorized as ADH2*3 positive to increase statistical power of analyses for ADH2*3 interaction. This was justified given previous findings that alcohol clearance rates do not differ significantly between ADH2*3-homozygous and -heterozygous African-American adults (Thomasson, Beard, and Li 1995). Population characteristics known to influence risk for IUGR were compared between ADH2*3-positive and -negative infants either by t, χ², or Fisher’s Exact Test using SAS (see Table 2). Differences were considered statistically significant if p ≤05. Mean birth-weights were stratified by mother’s drinking status (yes/no) and compared by t test (see Table 3). A difference in mean birth-weights was considered statistically significant if p ≤05.

Risk factors for being born SGA were identified by multivariable logistic regression (see Table 4). Variables entered into the regression equation were infants’ ADH2*3 status (positive, negative), sex of the infant, gestation age of <37 weeks, maternal pregnancy weight gain of <13.6 kg (30 pounds), maternal age at infant birth (<20 years), history of previous pregnancy, history of infant premature birth, less than high school education of the mother, single marital status, any alcohol use by the mother 3 months prior to conception through the end of the 1st trimester (yes/no), any alcohol use by the mother during the 3rd trimester (yes/no), and any cigarette smoking by the mother during pregnancy (yes/no). Regression coefficients were calculated using PROC LOGISTIC in SAS.

The impact of ADH2*3 status on mean birthweight of infants born to mothers that reported drinking during pregnancy was assessed by multivariable linear regression. Infants whose mothers reported they did not drink alcohol during the Critical Period or 3rd trimester of pregnancy were excluded from the analysis. Variables entered into the analysis were infant ADH2*3 status, gender, and gestation age, maternal weight gain during pregnancy, mother’s age, average number of cigarettes smoked by the mother per day during the Critical Period average number of cigarettes smoked by the mother per day during the 3rd trimester, and the average number of drinks consumed by the mother per day during the Critical Period or 3rd trimester. Regression coefficients were calculated using PROC REG in SAS.

RESULTS

A total of 1016 of 1039 blood spots (97.7%) were successfully genotyped for ADH2*3 status (Table 1). The highest percentage of ADH2*3 infants occurred in African-American infants; 1.3% of Caucasian infants were positive (heterozygous), and none of the 19 bloodspots linked to infants of other racial background were typed ADH2*3 positive. Population statistics of infants successfully genotyped for ADH2*3 were virtually identical to population statistics reported for BWIS control infants (N =3572) (Ferencz et al. 1993), suggesting that our nested population provided good representation of the larger BWIS control population. No conflicts in genotype arose when 200 randomly selected bloodspots were typed for ADH2*3 status a second time or when the sequences of PCR products from ADH2*3-negative, -heterozygous, and -homozygous bloodspots were compared with published sequences (Groppi, Begueret, and Iron 1990).

Because of the relatively high frequency for ADH2*3 among African-American infants, further analysis focused on this subpopulation (Table 2). The ratio of ADH2*3-negative, -homozygous, and -heterozygous infants in our subpopulation was very similar to ratios reported previously (McCarver et al. 1997; Bosron and Li 1986; Bosron, Lumeng, and Li 1988). There were no statistically significant differences in mean birthweight or gestation age when compared between ADH2*3-positive versus ADH2*3-negative infants. A significantly higher percentage of ADH2*3-negative infants were born to mothers with less than a complete high school education. Fewer ADH2*3 positive infants were born to families reporting a yearly income of >$30,000. The average number of drinks consumed by the mother per episode did not differ significantly by infant ADH2*3 status for episodes occurring in the Critical Period, 2nd, or 3rd trimesters of pregnancy (data not shown).

There was a negative trend for the mean birthweight of African-American infants with fetal alcohol exposure in the Critical Period through the 3rd trimester of pregnancy (Table 3). However, differences in mean birthweight for infants with or without fetal alcohol exposure were not statistically significant (p ≤.05).

SGA infants had a greater odds of being genotyped ADH2*3 negative as compared to normal and large infants (Table 4). SGA infants were also at higher odds for several other factors known to impact odds for being born SGA, including alcohol consumption by the mother during pregnancy. Of the 22 infants characterized as SGA, 7 were born to mothers that reported consuming alcohol in the 3rd trimester (32%), as compared with 47 (17%) of 238 mothers that reported drinking any alcohol during the 3rd trimester and gave birth to non-SGA infants.

The effect of ADH2*3 status on mean birthweight of infants with fetal alcohol exposure was assessed by multivariable regression analysis. ADH2*3 status was not a significant factor determining the mean birthweight of African-American infants with fetal alcohol exposure during the Critical Period or 3rd trimester of pregnancy. Factors found to have a significant (p ≤05) impact on mean birthweight for infants with fetal alcohol exposure during the Critical Period were infant gestation age and maternal weight gain during pregnancy. Infant gestation age was a significant (p ≤05) determinant of mean birthweight for infants with fetal alcohol exposure during the 3rd trimester. Factors of borderline significance were maternal weight gain during pregnancy (p ≤06) and maternal smoking during the 3rd trimester (p ≤06).

DISCUSSION

The finding that SGA African-American infants were three times more likely to be genotyped ADH2*3 is suggestive of an interaction between fetal growth and ADH2 genotype. Biologically, there is some evidence to suggest that ADH2 genotypes may have different rates for RA metabolism (Yang et al. 1994). Our findings for SGA infants were not statistically significant, however, and may very well represent a finding of no difference in the risk for SGA between ADH2*3-positive and -negative infants in our study population. We also found weak evidence that SGA infants with fetal alcohol exposure in the Critical Period or 3rd trimester of pregnancy were at elevated odds for being ADH2*3 negative. This finding may be indirect evidence of a protective effect of ADH2*3 against increased risk for being born SGA for infants whose mothers drank alcohol during pregnancy. However, this association is complicated by the finding of elevated odds for maternal smoking for SGA infants, which is correlated with alcohol drinking by the mother during pregnancy (Sokol, Miller, and Reed 1980; Haste et al. 1991; IOM 1996) and may explain the increased odds for SGA associated with maternal alcohol consumption in this population. The hypothesis that ADH2*3-positive infants with fetal alcohol exposure were at decreased odds for being born SGA could not be directly tested because of the small number of ADH2*3-positive SGA infants (n = 2) found in our study population. Our finding that the risk for SGA in or population was associated with low maternal weight gain during pregnancy has been reported previously (Abrams and Newman 1991; Lang, Lieberman, and Cohen 1996).

We did not find evidence for a direct impact of ADH2*3 on the mean birthweight of African-American infants in our population. Our initial analyses gave some indication that fetal alcohol exposure was negatively associated with mean infant birthweight. Thus, giving some justification for the more complex multivariable regression analysis of the impact of fetal alcohol exposure and infant ADH2*3 status on birthweight while accounting for other factors known to impact mean infant birth-weight, including gestation age, maternal weight gain, maternal smoking during pregnancy, etc. A priori calculations indicated our study had adequate statistical power to detect a 300-g difference in mean birthweight when infants were stratified by both ADH2*3 and fetal alcohol exposure status (yes/no). Although we did not find evidence of a protective effect against decreased birthweight associated with ADH2*3 among infants with fetal alcohol exposure, we did find that infant gestation age and maternal weight gain were significant factors affecting birth-weight. These findings have been reported by others (Kramer 1987; Kramer et al. 1990) and suggest that the BWIS data could be used with confidence in a study on factors impacting birth-weight despite the fact that the BWIS was not designed for this purpose. Many of the same risk factor data collected in the BWIS would probably also be collected in a study on the factors that impact infant birthweight.

Our nested case-control study provides an example of many of the possible pitfalls associated with studies of “convenience” not initially designed to explore gene-effect interactions. For this study, ideally, one would want the following: (1) a large population of mother-infant pairs of African-American racial background with biological samples from both the mother and infant; (2) detailed information on the birthweight and gestation age of the infant; (3) detailed interview data relevant to infant birth outcome and birthweight; (4) interview data on alcohol consumption levels and frequency of consumption from the mother at various time points during pregnancy; and (5) possibly maternal biological samples at various time points during pregnancy to assess maternal blood-alcohol concentrations. Recognized weaknesses in this nested case-control study include the following: (1) potential recall inaccuracies by the mothers interviewed in the BWIS regarding infant birthweight and gestation age; (2) drinking habits and number of drinks consumed per episode during pregnancy; and (3) the relatively small number of infants born to BWIS mothers that reported consuming any alcohol during the 2nd and 3rd trimesters of pregnancy. These shortcomings are understandable because the BWIS study was developed to explore environmental and familial risk factors for congenital heart defects and did not specifically focus on the interaction of maternal alcohol consumption on fetal growth and development. Although data on birthweight and gestation age were collected, the methods used to determine these measurements were not the methods preferred (e.g., research of birth certificates, hospital records, etc.) for when these factors are the focal point of study. Additionally, BWIS mothers were not recruited based on a history of chronic alcohol abuse and none of the 6949 BWIS cases and controls was diagnosed with fetal alcohol syndrome (FAS) or (FAE) within the first year after birth.

The relatively low alcohol consumption frequencies and consumption levels reported by BWIS mothers were the most significant limiting factor in assessing the impact of ADH2*3 × Fetal Alcohol Exposure interaction on mean birthweight. Currently, it is thought that maternal drinking levels of ≥2 drinks daily or five to six drinks per occasion may be a risk factor for IUGR (Little 1977; Kuzma et al. 1982; Wright et al. 1983). In our study, 25% of the mothers reported drinking during the 2nd or 3rd trimesters of pregnancy. Most of those that drank reported they drank less than once a week (79%) and consumed one to two drinks per occasion (89%). Maternal alcohol consumption was more common for the Critical Period (59% of mothers), with a greater number of mothers reporting that they consumed three or more drinks per occasion (48%). Consumption frequencies for these mothers were 40.7% never, 35.8% <once/week, 18.9% once/week, and 4.6% reported consuming alcohol on a daily basis during the Critical Period.

A previous study (McCarver et al. 1997) found the ADH2*3 genotype was protective against alcohol teratogenicity, namely deficits in Bayley Scales of Infant Development Mental Indices (MDI) and decreased birthweight. Absence of ADH2*3 in the mother or infant was associated with significant MDI deficits, whereas absence of the allele in the mother was associated with significant deficits in mean infant birthweight. This study differed from our study in that the study subjects were specifically recruited based on periconceptional alcohol intake and maternal ADH2*3 genotype, but alcohol consumption levels for mothers during pregnancy were not reported. Based on the findings of McCarver et al. (1997), there is some evidence suggesting that maternal ADH2*3 status may have a protective effect against certain alcohol-related birth defects in the fetus: possibly more of an effect than fetal ADH2*3 status. We did not have knowledge of the ADH2*3 status of the mothers of the infants in this study and therefore could not assess the impact of maternal ADH2*3 status on birth outcomes in our population.