Maternal Vitamin Use, Genetic Variation of Infant Methylenetetrahydrofolate Reductase, and Risk for Spina Bifida

Transcription

1 American Journal of Epidemiology Copyright 1998 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved Vol. 18,. 1 Printed in U.S.A. Maternal Vitamin Use, Genetic Variation of Infant Methylenetetrahydrofolate Reductase, and Risk for Spina Bifida Gary M. Shaw, 1 Rima Rozen, 2 Richard H. Finnell, 3 Cathy R. Wasserman, and Edward J. Lammer 5 Maternal periconceptional use of vitamin supplements containing folic acid substantially reduces the risk of neural tube defects (NTDs) in the offspring. The mechanism underlying this reduction in risk is unknown. Several recent studies have reported an association between homozygosity for a variant form (the C677T genotype) of the 5,10-methylenetetrahydrofolate reductase (MTHFR) gene and risk for NTDs in individuals. It has been hypothesized that maternal folic acid supplementation prevents NTDs by partially correcting reduced MTHFR activity associated with the variant form of the enzyme. Using data from two California case-control interview studies ( birth cohorts), the authors investigated whether an interaction for spina bifida risk existed between infant MTHFR C677T genotype and maternal use of supplements containing folic acid. The authors genotyped the allelic variants of MTHFR in 21 livebom case infants with spina bifida and 503 control infants for whom information on maternal periconceptional vitamin use was available. The percentage of all case infants with the C677T MTHFR mutation, for both homozygous () and heterozygous () genotypes, was slightly higher than that of controls. The C677T genotype was substantially more frequent among both case and control Hispanic infants than among non-hispanic infants. Among all infants whose mothers did not periconceptionally use vitamins containing folic acid, the risk of spina bifida, as measured by the odds ratio, was 1.6 (95% confidence interval (Cl) ) for all infants with the genotype and 2.0 (95% Cl ) for non-hispanic white infants with the genotype, as compared with infants with the genotype. This result indicates a modestly increased risk associated with the C677T genotype. A lower risk estimate (odds ratio = 1.2, 95% Cl 0.-.0) was observed among infants whose mothers periconceptionally used vitamin supplements containing folic acid. This population-based California study found a modestly increased risk of spina bifida among infants who were homozygous for the C677T genotype, but only minimal evidence of an interaction between the C677T genotype and maternal folic acid intake in the occurrence of spina bifida. If this mutant MTHFR genotype plays a role in the association between maternal vitamin use and NTD risk, it may be a small role, or it may be conditional on maternal genotype. Am J Epidemiol 1998; 18:30-7. abnormalities; folic acid; genotype; neural tube defects; spinal dysraphism; vitamins Neural tube defects (NTDs) are a group of congenital anomalies primarily comprised of spina bifida and anencephaly. Periconceptional use of vitamin supplements containing folic acid by pregnant women sub- Received for publication May 1,1997, and in final form vember 1, Abbreviations: C, cytosine; Cl, confidence interval; NTD(s), neural tube defect(s); MTHFR, 5,10-methylenetetrahydrofolate reductase; OR, odds ratio; T, thymine. 1 March of Dimes Birth Defects Foundation, California Birth Defects Monitoring Program, Emeryville, CA. 2 Departments of Human Genetics, Pediatrics, and Biology, McGill University, Montreal Children's Hospital, Montreal, Quebec, Canada. 3 Department of Veterinary Anatomy and Public Health, College of Veterinary Medicine, Texas A&M University, College Station, TX. Community and Family Health, Maternal and Child Health, Washington State Department of Health, Olympia, WA. 5 Division of Medical Genetics, Children's Hospital, Oakland, CA. Reprint requests to Dr. Gary Shaw, California Birth Defects Monitoring Program, 1900 Powell Street, Suite 1050, Emeryville, CA stantially reduces the risk of NTDs (1, 2). The underlying mechanism by which folic acid confers this lower risk, however, is unknown. For identification of such mechanisms, several research efforts currently under way are focusing on candidate genes that encode enzymes involved in folate metabolism or receptors involved in folate transport. Among these is the gene that codes for 5,10-methylenetetrahydrofolate reductase (MTHFR). MTHFR converts 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate, the latter being the major circulatory form of folate and the primary methyl donor for the methylation of homocysteine to methionine. Thus, this enzyme is key for regulating plasma homocysteine concentrations and maintaining an adequate intracellular methionine pool. Adequate amounts of the essential amino acid methionine may be important in protecting embryos against NTDs (3, ). Recently, a human polymorphism at the MTHFR Downloaded from by guest on 12 September

2 MTHFR Genotype and Spina Bifida 31 locus has been described: C677T, a cytosine (C)- to-thymine (T) transition (resulting in an alanineto-valine substitution) at base pair 677. This mutation codes for a thermolabile enzyme with reduced activity, and it has been linked to elevated plasma homocysteine levels in homozygous C677T individuals (5). Several studies now suggest that this genotype depends on a low plasma folate level for the development of hyperhomocysteinemia (6-8). Given this particular enzyme's involvement in the metabolism of folate and homocysteine, this MTHFR mutation clearly warranted further investigation for elucidation of its potential role in the mechanism underlying the reduction in NTD risk by folic acid supplementation. Recently, several studies reported an association between homozygosity for the C677T mutation and an increased risk of NTDs (9-). Another study did not find such an association (1). Relative to individuals who are homozygous for C677 (wildtype), substantially increased risks among persons who are homozygous for C677T have been observed, with odds ratios ranging approximately three- to sevenfold above unity (9 ). Among persons who are heterozygous for C677T, slightly increased NTD risks have been observed (10, 12). However, the studies that have investigated the relation between MTHFR genotype and NTD risk have not considered the risk associated with the mutant genotype in the presence versus absence of maternal folic acid intake from vitamin supplements or diet, and they have often lacked a population-based sample of cases and controls with which to estimate the frequency of the C677T polymorphism and therefore adequately estimate risk (15). To surmount these limitations, we genotyped infants from two large population-based case-control studies of infants with spina bifida (16, 17). The purpose of this study was to explore whether an interaction existed between the infant MTHFR C677T genotype and maternal use of vitamin supplements containing folic acid with regard to risk of spina bifida. We hypothesized that infants who were homozygous for the C677T genotype would be at increased risk for spina bifida because of lower MTHFR enzymatic activity. We further hypothesized that elevations in maternal serum folate levels resulting from periconceptional folic acid supplementation could improve the activity of the poorly functioning MTHFR enzyme, possibly via enzyme stabilization as previously suggested (5). Therefore, this hypothesis predicted that the spina bifida risk for infants who were homozygous for the C677T genotype would be lower in the group whose mothers periconceptionally used vitamin supplements containing folic acid than in the group whose mothers did not. MATERIALS AND METHODS Details on the two population-based case-control studies used in this analysis have been provided elsewhere (16, 17). Briefly, infants or fetuses with any NTD (anencephaly, spina bifida cystica, craniorachischisis, or iniencephaly) were identified through reviews of medical records, including prenatal diagnosis records, at hospitals and genetic clinics for all infants/ fetuses delivered in California (all counties except Los Angeles, Riverside, and Ventura) and whose mothers gave California as their state of residence. Eligible for study were all liveborn infants and fetuses diagnosed with NTDs (including those prenatally diagnosed and electively aborted between February 1989 and January 1991) in the total cohort of 1,052,33 births occurring between January 1987 and December 1988 and between June 1989 and May Control births were randomly selected from each surveillance area hospital in proportion to that hospital's estimated contribution to the total population of infants born alive in a given month from June 1989 to May 1991 (16), or were randomly selected from all infants born alive during 1987 and 1988 (17). Control infants chosen had no major structural malformations (18). Case and control mothers were interviewed in person (16) or by telephone (17), in English or Spanish as appropriate. Women who died prior to interview, who spoke only a language other than English or Spanish, or who had previously had an NTD-affected pregnancy were excluded. For the more recent birth cohort (16), interviews were completed approximately 5 months (an average of.9 months for cases and.6 months for controls) after the actual or estimated date of full term delivery. For the other study, the average amount of time between delivery and interview was approximately 3.7 years (17). Information was not available from about 5 percent of case mothers and percent of control mothers who refused to be interviewed, or from the 8 percent of case mothers and percent of control mothers who could not be located. The interviews elicited information on maternal medical, reproductive, and family history, employment, hobbies, and vitamin use. Women were specifically asked whether they used vitamin or mineral supplements and which type or brand of supplement(s) they used. Information on the timing of vitamin use was obtained slightly differently for the two different studies. In the study involving the more recent birth cohort (16), information on vitamin use during the 3 months before conception or during any of the three trimesters of pregnancy was requested. In the study of the earlier birth cohort (17), information on vitamin use was requested for each month during the -month Downloaded from by guest on 12 September 2018 Am J Epidemiol Vol. 18,. 1, 1998

3 32 Shaw et al. period beginning 1 month before conception and ending 3 months after conception. For the current investigation, we divided women into three categories reflecting the timing of their use of vitamins containing folic acid. "Early use" was defined as preconceptional use in either study or use beginning within the first month postconception in the earlier study (17). "Late use" was defined as use beginning any time within the second or third month after conception for the earlier study (17), or beginning at any time after conception but within the first trimester for the more recent study (16). "nuse" was defined as initiation of vitamin use after the third month of pregnancy or no vitamin use during pregnancy (both studies). Information on daily intake of dietary folate during the 3 months before conception was also available from a standardized food frequency questionnaire (19) for one of the two source data sets (16). Because of our interest in analyzing information about periconceptional vitamin use, only cases and controls whose mothers were interviewed were eligible for this analysis. Of these, only liveborn spina bifida case infants and liveborn controls were included, because DNA was to be isolated from residual newborn screening blood specimens (filter paper), and such samples were only available from liveborn infants. (Blood specimens were not available for any of the anencephalic cases.) These specimens are collected from all liveborn children in California, are coded, and are stored individually. Among the total of 1,000 ascertained NTD cases (all types), 802 had complete maternal interviews. Of these 802 cases, 25 were specifically cases of spina bifida, with the remainder being diagnosed as other NTDs. Of the 25 infants with spina bifida, 311 were liveborn. Among the total of 1,296 ascertained control infants, 1,020 had complete maternal interviews. Of the 311 liveborn spina bifida case infants and the 1,020 control infants, blood specimens were identified for 26 (79 percent) cases and 910 (89 percent) controls. Some specimens were not obtained because: 1) no blood sample remained on the filter paper, 2) the filter paper could not be located, or 3) the information available was insufficient to match infant records. DNA was extracted from the filter papers by means of standard laboratory procedures (20) and was amplified using the polymerase chain reaction. Genotyping for the MTHFR mutation was performed by restriction digestion of polymerase chain reaction products with Hinfl, as described elsewhere (5). All genotyping was performed by persons blinded to subjects' case/control status and to information on maternal vitamin use. To minimize the number of genotypic analyses required, we randomly sampled 25 percent of the 68 control infants whose mothers were "late users" of vitamins containing folic acid and genotyped only this subset of "late user" controls. All control infants whose mothers were considered nonusers or "early users" of vitamins were genotyped. All case infants were genotyped and included in the analyses irrespective of when or whether their mothers had used vitamins. A total of 21 cases and 503 controls were genotyped. Cases and controls were categorized as if they were homozygous for the C677T allele, if they were heterozygous for the C677T allele, and if they were homozygous for the C677 (wild-type) allele. Because we genotyped only a 25 percent sample of the control group, MTHFR genotypic frequencies among controls in the overall data set were estimated by weighting the genotype frequency among "late" vitamin users by a factor of (table 1). We performed analyses to estimate the risk of spina bifida among infants with either the or the genotype, as compared with infants with the genotype, who had mothers who were "nonusers" or "early users" of vitamins containing folic acid. Odds ratios and their 95 percent confidence intervals were computed as estimates of relative risks. Statistical heterogeneity was assessed by the chi-square test for homogeneity, using the recommended criterion of p ^ 0.20 as evidence for statistical heterogeneity (21). RESULTS Frequencies of the MTHFR genotypes in the study population are shown in table 1. The percentage of infants with a C677T allele (both and genotypes) was somewhat higher among cases than among controls. By race/ethnicity, the percentage of infants with the mutation was highest among Hispanic births, and this higher frequency was observed in both case and control infants. To our knowledge, the frequency of the homozygous MTHFR genotype in the Hispanic infants (25 percent) is the highest seen thus far in any population. There were several possible approaches to assessment of the potential association between MTHFR genotype and maternal folic acid intake in the risk for spina bifida. We hypothesized that, among infants with the genotype (as compared with the genotype), the risk for spina bifida would be substantially higher in infants whose mothers did not use vitamins early in pregnancy ("nonusers") than in those whose mothers did. The results, shown in the upper part of table 2, only weakly support this hypothesis. Compared with infants with the genotype, the odds ratios for spina bifida among infants with the Am J Epidemiol Vol. 18,. 1, 1998 Downloaded from by guest on 12 September 2018

4 MTHFR Genotype and Spina Bifida 33 TABLE 1. Frequency of different methylenetetrahydrofolate reductase (MTHFR) genotypes among spina bifida case infanta and population control infants, by race/ethnicity, California, Infant MTHFR genotype* All racial/ethnic groups n-hispanic whites Hispanic whites Blacks Other races/ etnnicmes % Cases (n = 21) Controls (n = % ) C, cytosine; T, thymine. C refers to the 677C genotype and T refers to the 677T genotype. t Percentages among controls were estimated by weighting the genotype frequency by maternal vitamin use status. Genotype frequencies among mothers who began their vitamin use in the first trimester or afterward were weighted by a factor of. All other frequencies were weighted by a factor of 1. (See "Materials and Methods" for more detail concerning the sampling scheme used.) Thus, the percentages shown may not equal the percentages that can be computed from the numbers of control infants given. genotype were 1.2 (95 percent confidence interval (CI) 0.-.0) for infants whose mothers were "early users" of vitamins containing folic acid and 1.6 (95 percent CI ) for infants whose mothers were considered nonusers. Although the difference between the risk estimates pointed in the hypothesized direction, the two estimates were not statistically different from each other (p = 0.76). In addition, among infants who were heterozygous for C677T (genotype ), the risk estimate was minimally increased (odds ratio (OR) = 1.2) for those whose mothers did not use vitamins but was higher for those whose mothers were "early" vitamin users. This difference pointed in the direction opposite that predicted by the hypothesis. Similar analyses conducted separately for each contributing data set did not show substantial differences in risk from those observed when we combined the two data sets. TABLE 2. Odds ratios for spina bifida by infant methylenetetrahydrofolate reductase (MTHFR) genotype and maternal periconceptional use of vitamin supplements containing folic acid, California, MTHFR genotype* Earlyt maternal of controls ORt % C\% * C, cytosine; T, thymine. C refers to the 677C genotype and T refers to the 677T genotype. t Defined as use beginning preconceptionally or beginning within 1 month postconceptjon. " use" included women who started vitamiri use after the third month of pregnancy or did not consume vitamins during pregnancy. $ OR, odds ratio; CI, confidence interval. Referent. A second approach to assessment of the potential NTD risk associated with the C677T genotype and maternal vitamin use is shown in the lower half of table 2. This part of the table shows risk estimates associated with maternal "early" vitamin use, stratified by infant genotype. The hypothesis predicted that the odds ratio for vitamin use in the stratum would be substantially lower than that in the stratum. The differences in odds ratios between the and strata should have reflected the magnitude of spina bifida risk that was potentially alterable by correcting the lower MTHFR activity through maternal folic acid use. Reduced risks for spina bifida were observed for maternal vitamin use irrespective of infant genotype. However, the risk estimates were similar (p = 0.76) for infants who were homozygous () for the C677T mutation (OR = 0.2) and those who were homozygous () for the C677 allele (OR = 0.3). Data were too sparse for adequate assessment of risks associated with the genotype by maternal vitamin use in each racial/ethnic group, with the exception of non-hispanic whites (e.g., among Hispanic case infants with the genotype, no mother used vitamins early in pregnancy). Among non-hispanic whites, the odds ratios for spina bifida among infants Downloaded from by guest on 12 September 2018 Am J Epidemiol Vol. 18,. 1, 1998

5 3 Shaw et al. with the genotype compared with infants with the genotype were 0.6 (95 percent CI ) for those whose mothers used vitamins early in pregnancy and 2.0 (95 percent CI ) for those whose mothers did not use vitamins. This difference in risk estimates pointed in the hypothesized direction, but the two estimates were not statistically different from each other {p = 0.33). Moreover, the risk estimates among non-hispanic white infants who were heterozygous for C677T (genotype ) were above (OR = 1.5, 95 percent CI ) for infants whose mothers used vitamins and less than (OR = 0.7, 95 percent CI ) for infants whose mothers did not use vitamins. This difference pointed in the direction opposite that predicted by the hypothesis. There is an alternative analytical approach with which to explore a potential interaction between genotype and vitamin use. However, this approach did not directly assess our hypothesis, which predicted that spina bifida risk among infants who were homozygous for the genotype would be lower in the group whose mothers used vitamins than in the group whose mothers did not. The alternative approach involved assessing the risk associated with 1) the genotype and no vitamin use, 2) the genotype and no vitamin use, and 3) the genotype and vitamin use, by comparing each group with the infants who had the genotype and maternal vitamin use. The odds ratios for these comparisons were 5.2 (95 percent CI ), 3.3 (95 percent CI ), and 1.2 (95 percent CI ), respectively. The first two risk estimates, 5.2 and 3.3, were not statistically different from each other (p = 0.1), which suggests that the influence of the genotype alone may not substantially increase risk for spina bifida beyond the increased risk associated with no maternal vitamin use. For purposes of comparison with previous studies, we computed risk estimates for being a case infant and having either the or the genotype relative to having the genotype. We computed these risk estimates adjusting for maternal vitamin use by the Mantel-Haenszel procedure (22) (including "late users" data not shown). This adjustment was carried out because controls were sampled for analytical purposes based on maternal vitamin use. Odds ratios of 1.5 (95 percent CI ) for the genotype and 1. (95 percent CI -2.0) for the genotype were observed, suggesting a modestly increased spina bifida risk associated with C677T homozygosity or heterozygosity. The corresponding summary odds ratios for non-hispanic whites were 1. (95 percent CI ) and 1.1 (95 percent CI ), and those for Hispanic infants were 1.6 (95 percent CI ) and 1.8 (95 percent CI -3.3), respectively. Finally, we explored the potential risk of having the C677T genotype among infants whose mothers' dietary intake of folate from food was relatively low (using data from the more recent data set only (16)). Restricting analyses to only those women who were nonusers of vitamins, and using quartiles of control mothers' intake as the cutpoints, we computed risks for the and genotypes relative to the genotype, for each quartile of intake (table 3). The largest odds ratio (OR = 5.2, 95 percent CI ) was observed for the genotype among infants whose mothers had dietary folate intakes in the lowest quartile. A lower risk estimate (OR = 1.7, 95 percent CI ) was observed in the highest folate intake group. However, data were sparse, resulting in low precision and limited statistical evidence for heterogeneity between risk estimates. DISCUSSION Our results suggested some increased risk of spina bifida (OR = 1.5) in infants who were homozygous for the C677T genotype. However, we observed only minimal evidence for an interaction between infant genotype and maternal supplemental vitamin use in the occurrence of spina bifida in this population-based sample of California births. Based on our results and on data from previous investigations, it appears that if infant MTHFR genotype plays a role in the association between maternal periconceptional supplement use and reduced NTD risk, it may be a small role (e.g., it TABLE 3. Odds ratios for spina bifida by quartile of average daily maternal dietary intake of folate in the 3 months before conception, among women who did not use vitamin supplements periconceptionally,* California, Quartile of dietary folate Intake (mg) ! genotypey. of cases of controls OR* % Cl * Assessed by food frequency questionnaire. t MTHFR, methylenetetrahydrofolate reductase; C, cytosine; T, thymine. C refers to the 677C genotype and T refers to the 677T genotype. $ OR, odds ratio; CI, confidence interval. Referent. Downloaded from by guest on 12 September 2018 Am J Epidemiol Vol. 18,. 1, 1998

6 MTHFR Genotype and Spina Bifida 35 has been suggested () that 12 percent of NTDs in the Irish population are attributable to the MTHFR mutation), or its influence may be limited to certain population groups (e.g., the Irish (9, ) or Dutch (10, 11)). Because of sparse data, our ability to investigate whether a genotype-vitamin association was more evident in certain racial/ethnic groups within this heterogeneous study population was limited. We did observe the frequency of C677T homozygosity among Hispanic infants to be the highest found in any population thus far. However, risks for spina bifida associated with the genotype were similar in Hispanic infants and non-hispanic white infants (OR =1.6 vs. OR = 1.). The rationale for exploring lower infant MTHFR enzymatic activity as a possible etiologic mechanism in abnormal neural tube closure was the knowledge that 5-methyltetrahydrofolate, the product of the MTHFR reaction, is required for homocysteine methylation, and it has recently been suggested that mothers with NTD-affected infants metabolize homocysteine less effectively than mothers with non-ntd infants (23). Furthermore, individuals with C677T homozygosity may have higher folate requirements for maintenance of physiologic plasma homocysteine concentrations (6). The benefit of the additional folate may relate to increased stability of the MTHFR enzyme. Thus, specific folate-related enzymes that are involved in the metabolism of homocysteine, especially ones such as MTHFR, clearly warrant further investigation for their possible mechanistic role in the risk of folate-responsive NTDs. In addition to this plausible biologic evidence, studies derived from four data sets have reported three- to sevenfold increased NTD risks associated with being homozygous for the C677T genotype (9-). An Irish study by Whitehead et al. (9) included NTD cases (nearly all spina bifida) and 99 healthy adult controls. Finding 15 cases and six controls to be homozygous for C677T, these investigators reported a 3.5-fold increased NTD risk associated with the mutation. Kirke et al. () extended this study population to 153 NTD cases and a control group comprised of adult females. The genotype was found in 29 cases and 20 controls, revealing an increased NTD risk of 2.6 associated with C677T homozygosity. A similar increased risk for spina bifida patients with C677T homozygosity was reported by van der Put et al. (10). Their study included 55 Dutch patients with spina bifida and 207 unrelated adult Dutch controls from a single general practice in the Hague. Seven cases and 10 adult controls were homozygous for C677T. Ou et al. (12) reported a 7.2-fold increased NTD risk associated with C677T homozygosity. DNA samples from 1 NTD cases from South Carolina were genotyped and compared with the genotypes of 109 unrelated healthy persons from Atlanta, Georgia. Nine case samples and five control samples were found to be homozygous for C677T. In addition to finding a lower magnitude of risk for spina bifida associated with the C677T genotype, our study differed methodologically in several important ways from these previous studies (9-), as well as from another study (1) that also did not find a large increased risk associated with C677T. Relative to these other studies, ours was the only study that investigated the potential effect of an interaction between MTHFR genotype and maternal vitamin supplement use or dietary folate intake on spina bifida risk. Another unique aspect of our study was that it had population-based ascertainment of both cases and infant controls, whereas previous studies did not have complete ascertainment of cases and did not have control groups selected from the same source population as the cases. In fact, the composition of the control groups may explain why at least three (9-12) of the previous studies observed higher NTD risks associated with C677T homozygosity. The frequencies of C677T homozygosity among controls, which included both adults and children, were 6-8 percent in Ireland (9, ) and 5 percent in the Netherlands and Atlanta (10-12). In contrast, we found, as did the British study by Papapetrou et al. (1), a frequency that was approximately twofold higher than this among controls (see figure 1). Thus, either the studies that have observed substantially increased NTD risks associated with C677T homozygosity underreported its true frequency in their control populations (resulting in spuriously elevated risks) or there is significant ethnic variation in the frequency of the C677T allele in the Netherlands and Ireland. Available data from other studies suggest that the genotype is found in approximately percent of the US population (7, 2, 25), 12 percent of the Canadian population (5), 11 percent of the Australian population (26), and 16 percent of the Italian population (27). These estimates are similar to the population frequency identified in non- Hispanic whites in our study, but they are lower than the frequency found in Hispanic infants. Our study may have been limited, as were the previous investigations, by the lack of information on every incident case fetus with spina bifida. Fetuses with spina bifida that do not survive to term, due to either elective abortion or spontaneous termination of pregnancy, may have a higher or lower frequency of the homozygous C677T genotype; such cases, therefore, may have a different risk relation. Downloaded from by guest on 12 September 2018 Am J Epidemiol Vol. 18,. 1, 1998

7 36 Shaw et al. 30 Percent MTHFR 25 Genotype () Neural Tube Defect Cases o FIGURE 1. Prevalence of the 677C^>T methylenetetrahydrofolate reductase (MTHFR) mutation (a cytosine (C)-to-thymine (T) transition at base pair 677) in five previously published studies (9, 10, 12-1) and the current study. (*Prevalence among non-hispanic whites only.) Our study was designed to assess the hypothesis that infant MTHFR genotype is important in conferring the beneficial effect of maternal supplemental vitamin use. Since we did not observe a strong interaction between maternal vitamin use and infant genotype, it is possible that the protective effect of vitamin use relates to correction of the maternal metabolic defect rather than that of the fetus. Several investigators have suggested that maternal MTHFR genotype may have a role in NTD risk (9-11), including one study that examined the combination of maternal and fetal genotypes (11). A disturbance in maternal metabolism could affect fetal development even before the homocysteine remethylation pathway has been activated in the fetus. This proposed influence of the maternal genotype on fetal development has been observed with other metabolic disorders such as phenylketonuria; maternal hyperphenylalanemia is associated with congenital defects in the fetus, in the absence of a mutant fetal genotype (28). A limitation of all of the studies performed thus far is the lack of information on maternal plasma folate concentrations in combination with genotypes. Our study, although the first to consider vitamin intake (and dietary folate intake) in combination with MTHFR genotype data relative to NTD risk, relied on mothers' reports of vitamin use and intake of folate To "m a itehea CM a! CO 5 Q. O 1= CO CO 15 Z3 S 5: a foods. The lack of information on maternal plasma folate may have been particularly important, because the MTHFR mutation alone may not be sufficient for disruption of homocysteine remethylation it may be dependent on an accompanying low folate status for the development of hyperhomocysteinemia (6). Our study contained too few infants whose mothers did not use vitamin supplements and had low dietary folate intakes for us to adequately assess the role of MTHFR genotype in NTD risk among infants whose mothers' plasma folate level may have been low. The mechanism underlying the risk reduction associated with maternal folic acid supplementation remains an important question in investigations of the etiology of NTDs and of the reduced occurrence of other congenital anomalies linked with periconceptional maternal multivitamin intake (29-31). Further study of candidate genes encoding other specific folate-related enzymes or proteins associated with folate absorption and transport may provide important additional lines of inquiry. ACKNOWLEDGMENTS The authors are indebted to Dr. George Cunningham, Dr. Fred Lorey, Terry Kennedy, and John Arnopp for making it CO "5 5 Kirke o Downloaded from by guest on 12 September 2018 Am J Epidemiol Vol. 18,. 1, 1998

8 MTHFR Genotype and Spina Bifida 37 possible to access newborn blood specimens; to Kimberly Morland, Sharon Campleman, and Karen Todoroff for their careful data analysis; and to Carmen Artigas, Robert Barber, and Nelly Sabbaghian for their excellent technical assistance. REFERENCES 1. Medical Research Council Vitamin Study Research Group. Prevention of neural tube defects: results of the Medical Research Council Vitamin Study. Lancet 1991;338: Czeizel AE, Dudas I. Prevention of the first occurrence of neural-tube defects by periconceptional vitamin supplementation. N Engl J Med 1992;327: Mills JL, McPartlin JM, Kirke PN, et al. Homocysteine metabolism in pregnancies complicated by neural tube defects. Lancet 1995;35: sel PG, Klein NW. Methionine decreases the embryotoxicity of sodium valproate in the rat: in vivo and in vitro observations. Teratology 1992;6: Frosst P, Blom HJ, Milos R, et al. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. (Letter). Nat Genet 1995; 10: Jacques PF, Bostom AG, Williams RR, et al. Relation between folate status, a common mutation in methylenetetrahydrofolate reductase, and plasma homocysteine concentrations. Circulation 1996;93: Ma J, Stampfer MJ, Hennekens CH, et al. Methylenetetrahydrofolate reductase polymorphism, plasma folate, homocysteine, and risk of myocardial infarction in US physicians. Circulation 1996;9: Christensen B, Frosst P, Lussier-Cacan S, et al. Correlation of a common mutation in the methylenetetrahydrofolate reductase gene with plasma homocysteine in patients with premature coronary artery disease. Arterioscler Thromb Vase Biol 1997;17: Whitehead AS, Gallagher P, Mills JL, et al. A genetic defect in 5,10-methylenetetrahydrofolate reductase in neural tube defects. QJM 1995;88: van der Put NM, Steegers-Theunissen RP, Frosst P, et al. Mutated methylenetetrahydrofolate reductase as a risk factor for spina bifida. Lancet 1995;36: van der Put NM, van den Heuvel LP, Steegers-Theunissen RP, et al. Decreased methylene tetrahydrofolate reductase activity due to the 677C»T mutation in families with spina bifida offspring. J Mol Med 1996;7: Ou CY, Stevenson RE, Brown VK, et al. 5,10-Methylenetetrahydrofolate reductase genetic polymorphism as a risk factor for neural tube defects. Am J Med Genet 1996;63: Kirke PN, Mills JL, Whitehead AS, et al. Methylenetetrahydrofolate reductase mutation and neural tube defects. (Letter). Lancet 1996;38: Papapetrou C, Lynch SA, Burn J, et al. Methylenetetrahydrofolate reductase and neural tube defects. (Letter). Lancet 1996; 38: Posey DL, Khoury MJ, Mulinare J, et al. Is mutated MTHFR a risk factor for neural tube defects? (Letter). Lancet 1996; 37: Shaw GM, Schaffer D, Velie EM, et al. Periconceptional vitamin use, dietary folate, and the occurrence of neural tube defects. Epidemiology 1995;6: Wasserman CR, Shaw GM, O'Malley CD, et al. Parental cigarette smoking and risk for congenital anomalies of the heart, neural tube, or limb. Teratology 1996,53: Croen LA, Shaw GM, Jensvold NG, et al. Birth defects monitoring in California: a resource for epidemiological research. Paediatr Perinat Epidemiol 1991;5: Block G, Hartman AM, Dresser CM, et al. A data-based approach to diet questionnaire design and testing. Am J Epidemiol 1986; 12: Schwartz El, Khalchitsky SE, Eisensmith RC, et al. Polymerase chain reaction amplification from dried blood spots on Guthrie cards. (Letter). Lancet 1990;336: Selvin S. Statistical analysis of epidemiologic data. 1st ed. New York, NY: Oxford University Press, (Monographs in epidemiology and biostatistics, vol 17). 22. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959;22: Steegers-Theunissen RP, Boers GH, Trijbels FJ, et al. Maternal hyperhomocysteinemia: a risk factor for neural-tube defects? Metabolism 199;3: Chen J, Giovannucci E, Kelsey K, et al. A methylenetetrahydrofolate reductase polymorphism and the risk of colorectal cancer. Cancer Res 1996;56: Ma J, Stampfer MJ, Giovannucci E, et al. Methylenetetrahydrofolate reductase polymorphism, dietary interactions, and risk of colorectal cancer. Cancer Res 1997;57: Wilcken DE, Wang XL. Relevance to spina bifida of mutated methylenetetrahydrofolate reductase. (Letter). Lancet 1996; 37: de Franchis R, Sebastio G, Mandato C, et al. Spina bifida, 677T->C mutation, and role of folate. (Letter). Lancet 1995; 36: Scriver CR, Kaufman S, Eisensmith RC, et al. The hyperphenylalanemias. In: Scriver CR, Beaudet AL, Sly WS, et al, eds. The metabolic and molecular bases of inherited disease. 7th ed. New York, NY: McGraw-Hill Book Company, 1995: Shaw GM, Lammer EJ, Wasserman CR, et al. Risks of orofacial clefts in children born to women using multivitamins containing folic acid periconceptionally. Lancet 1995;36: Shaw GM, O'Malley CD, Wasserman CR, et al. Maternal periconceptional use of multivitamins and reduced risk for conotruncal heart defects and limb deficiencies among offspring. Am J Med Genet 1995,59: Li DK, Daling JR, Mueller BA, et al. Periconceptional multivitamin use in relation to the risk of congenital urinary tract anomalies. Epidemiology 1995;6: Downloaded from by guest on 12 September 2018 Am J Epidemiol Vol. 18,. 1, 1998