Low Blood Folates in NTD Pregnancies are OnlyPartly Explained By Thermolabile5,10-Methylenetetrahydrofolate Reductase: LowFolate Status Alone May Be the Critical Factor

Anne M. Molloy,1* James L. Mills,2 Peadar N. Kirke,3 Dorothy Ramsbottom,4 Joseph M. McPartlin,1Helen Burke,3 Mary Conley,2 Alexander S. Whitehead,4 Donald G. Weir,1 and John M. Scott5

1Department of Clinical Medicine, Trinity College Dublin, Dublin, Ireland2National Institute of Child Health and Human Development, Bethesda, Maryland3Health Research Board, Dublin, Ireland4Department of Genetics and Biotechnology Institute, Trinity College Dublin, Dublin, Ireland5Department of Biochemistry, Trinity College Dublin, Dublin, Ireland

Thermolabile 5,10-methylenetetrahydrofo-late reductase (MTHFR) is the first folate-related variant to be associated with an in-creased risk of neural tube defects (NTDs).The variant causes high plasma homocyste-ine levels and reduced red cell folate (RCF)levels, both of which have also been linkedto an increased risk of NTDs.

We examined the relationship between fo-late status and presence of the common mu-tation MTHFR C677T in 82 NTD-affectedand 260 control mothers. Homozygosity forthe TT genotype was associated with verylow folate status among both the cases(n = 13) and the controls (n = 21). However,after exclusion of TT homozygotes, only 10%of the remaining 240 controls had RCF lev-els less than 200 mg/L compared with 29% ofthe 69 cases (odds ratio, 3.67; 95% confidenceinterval, 1.88–7.18; P < 0.001), and those withRCF less than 150 mg/L had eight timeshigher risk of NTD than subjects with levelsover 400 mg/L. Plasma homocysteine levelsof non-TT cases were also higher than thoseof controls (P = 0.047).

This study shows that homozygosity forthe C677T MTHFR variant cannot accountfor reduced blood folate levels in manyNTD-affected mothers. Thus, a strategy ofgenetic screening of all childbearing women

for this variant would be ineffective as amethod of primary prevention of NTDs. Thedata suggest that low maternal folate statusis itself the major determinant of NTD risk,or else that other folate-dependent geneticvariants confer risk through the reductionof folate levels. These results emphasize theimportance of a food-fortification programas a population strategy for reducing the oc-currence of NTDs. Am. J. Med. Genet.78:155–159, 1998. © 1998 Wiley-Liss, Inc.

KEY WORDS: neural tube defects; red cellfolates; plasma folate; homo-cysteine; vitamin B12; 5,10-methylenetetrahydrofolatereductase

INTRODUCTION

Periconceptional folic acid supplementation can ef-fectively prevent the majority of neural tube defects(NTDs), although most women carrying affected fe-tuses do not have blood folate levels in the deficientrange. This suggests a causal role for folate-dependentenzymes. In a large, nested, case-control study, weshowed lower blood folate status [Kirke et al., 1993]and higher plasma homocysteine levels [Mills et al.,1995] in NTD-affected pregnancies, thus strengtheningthe argument that an abnormality in folate metabolismis involved. We later demonstrated that a woman’s riskof having an NTD-affected child is inversely correlatedwith pregnancy red cell folate (RCF) levels [Daly et al.,1995]. The possibility of a future population screeningprogram for NTD risk factors has now focused researchon the discovery of genetic abnormalities in folate-related enzymes that might account for the observedmetabolic changes.

Contract grant sponsor: National Institute of Child Health andHuman Development; Contract grant sponsor: Health ResearchBoard of Ireland.

*Correspondence to: Dr. Anne Molloy, Department of Biochem-istry, Trinity College Dublin, Dublin 2, Ireland.

Received 3 September 1997; Accepted 26 December 1997

American Journal of Medical Genetics 78:155–159 (1998)

© 1998 Wiley-Liss, Inc.

Between 5% and 15% of normal subjects are homozy-gous for a common thermolabile mutation (C677 → T)of the enzyme 5,10-methylenetetrahydrofolate reduc-tase (MTHFR), which causes reduced enzyme activity[Kang et al., 1991; Frosst et al., 1995; de Franchis etal., 1995; Papapetrou et al., 1996; Kirke et al., 1996].This variant is associated with raised plasma homocys-teine levels [Kang et al., 1988; Engbersen et al., 1995;Kluijtmans et al., 1996; Harmon et al., 1996]. We haverecently shown that it is also associated with lowerthan expected RCFs in both pregnant and nonpregnantwomen [Molloy et al., 1997]. Homozygosity for thisvariant (TT) has been reported to be more common inindividuals with spina bifida and accounts for approxi-mately 13% of the population attributable risk [White-head et al., 1995; van der Put et al., 1995; Kirke et al.,1996; Ou et al., 1996]. There is also evidence of a higherthan expected prevalence of the TT genotype amongmothers of NTD cases [Whitehead et al., 1995; van derPut et al., 1995]. The altered blood folate levels ob-served in NTD mothers in our studies [Kirke et al.,1993; Mills et al., 1995] and other studies [van der Putet al., 1995] may therefore be due, at least in part, tothermolabile homozygous status.

The aim of this study was to ascertain to what extenthomozygosity for the C677T variant explained the re-duced blood folate levels and elevated plasma homocys-teine in maternal blood during NTD-affected pregnan-cies and thus to determine whether genetic screeningof all childbearing women for this variant would beeffective as a method of primary prevention of NTDs.

SUBJECTS AND METHODS

The subjects and methods for this analysis have beendescribed in detail elsewhere [Kirke et al., 1993; Millset al., 1995; Daly et al., 1995]. Briefly, blood sampleswere collected prospectively from 56,049 women at-tending their first antenatal clinic in one of the threemain Dublin maternity hospitals between March 1986and March 1990. This accounted for approximately70% of women who delivered in these hospitals duringthat period. The blood, collected into potassium ethyl-ene diamine tetracetic acid (K EDTA), was logged andprocessed appropriately to give a 1% ascorbic acid ly-sate sample suitable for analysis of RCF; a plasmasample for analysis of vitamin B12, folate, and homo-cysteine, and a sample of whole blood for genotypeanalysis. The blood samples were placed in labeledpolypropylene tubes and stored at −20°C. When preg-nancy outcomes were known, it was found that 137women who had babies affected by NTDs attendedtheir first antenatal clinic within the blood collectionperiod. Using hospital records, a systematic sample of417 controls was taken from births without NTDs ofgestation 23 weeks or over in the same hospital andduring the same period as the cases. From the storedbank of samples, some blood was retrieved for 84(61.3%) of the 137 cases and 266 (63.8%) of the 417controls. Of these samples, however, no plasma wasavailable for 3 cases and 19 controls, and there was noblood for genotype analysis on 2 cases and 5 controls.Microbiological methods were used to determine

plasma vitamin B12, folate, and RCF [Kelleher andO’Broin, 1991; O’Broin and Kelleher, 1992]. Plasma ho-mocysteine was measured using high-performance liq-uid chromatography with fluorescence detection [Ub-bink et al., 1991]. After analysis of plasma vitamin B12and folate, there was insufficient plasma to completethe homocysteine analysis on 5 cases and 7 controls.Results of all of these analyses have previously beenpublished as prospective studies on folate status inmothers whose pregnancy outcome was an infant witha NTD [Kirke et al., 1993; Mills et al., 1995; Daly et al.,1995]. The stored whole-blood fractions were thawedand processed to provide crude cell lysates suitable forpolymerase chain reaction (PCR) amplification using avariation of the Smith et al. [1992] protocol. DNAsamples were genotyped for the 677C → T mutation byPCR and restriction analysis as previously described[Whitehead et al., 1995]. For all analyses, the sampleswere coded so that laboratory staff were unaware of thestatus of the samples being assayed.

Statistical Analysis

The distributions of all blood analytes were not nor-mal; therefore, comparisons for all continuous vari-ables were carried out using nonparametric methods.Because of the skewed data, medians are presented inthe Tables I and II. The level of significance was set atP 4 0.05. Fisher’s exact test was used to compare pro-portions. The statistical analyses were performed us-ing InStatTM for Macintosh (GraphPad software, ver-sion 2.01, 1993). Analysis of relative risk was carriedout using the methods described in detail by Daly et al.[1995], which assumes an underlying risk of 1.9 NTDsper 1,000 births during that time period.

RESULTSGenotype analysis of MTHFR showed an excess of

thermolabile (TT) homozygotes in the NTD mothers(15.9%; n 4 13) compared with the control mothers(8.0%; n 4 21), This was of borderline significance(P 4 0.054). The data for all blood analytes weregrouped according to MTHFR thermolabile genotypeand analyzed for (a) effect of genotype and (b) case ver-sus control status. The results for RCF and plasmafolate are shown in Table I.

Both cases and controls showed the same effect ofgenotype, with TT homozygotes having lower RCF andplasma folate concentrations than the correspondingCC (homozygous wild-type) or CT (heterozygous) sub-jects. These differences were significant in the controlsbut not in the cases, perhaps because of small numbers.Nevertheless, the magnitude of the effect was similar,because 22% of controls and 20% of cases who had RCFlevels less than 200 mg/L were found to have the TTvariant.

Analysis of case versus control status by genotypeshowed a highly significant reduction of both RCF andplasma folate among the CC case mothers comparedwith CC control mothers (P < 0.001 and P 4 0.003 forRCF and plasma folate, respectively; Table I). LowerRCF and plasma folate concentrations were also evi-dent in CT case mothers (P 4 0.014 and P 4 0.085,respectively). TT cases were similar to TT controls.

156 Molloy et al.

Although there was evidence of a trend toward lowerRCF and plasma folate levels from CC through CT toTT genotypes in both cases and controls, the CC andCT groups did not differ significantly from each otherin either the cases or the controls. A similar analysisdemonstrated that this was also true in the case ofplasma homocysteine concentrations. Therefore, for allfurther analyses, the CC and CT subgroups were com-bined.

The results for plasma homocysteine are shown inTable II. The homocysteine concentration of TT homo-zygotes was significantly higher than in the combinedCC/CT subgroups in the controls (P 4 0.026), but it didnot reach significance in cases (P 4 0.104). However,the case/control difference previously demonstrated forthe total group [Mills et al., 1995] was still evidentafter the TT homozygotes were excluded (P 4 0.047).There was no effect of genotype on the plasma vitaminB12 level in cases or in controls, and the previouslyreported observation that vitamin B12 was signifi-cantly lower in cases was still present (data not shown)[Kirke et al., 1993].

It has been shown previously that NTD risk is in-versely related to maternal RCF levels [Daly et al.,1995]. To determine whether the TT genotype ac-counted for this relationship, the thermolabile (TT) ho-mozygotes were excluded, and a risk analysis by RCFwas carried out on the remaining 69 cases and 240controls. This analysis showed that the relation be-tween inverse risk of NTD and RCF was similar to thatreported previously for the overall group [Daly et al.,

1995], in that there remained an eightfold difference inrisk between those with RCF less than 150 mg/L andthose with levels of 400 mg/L or higher. Taking RCFlevels less than 200 mg/L as an arbitary cutoff point,10% of the remaining 240 controls were in this lowcategory compared with 29% of the 69 cases (odds ratio[OR], 3.67; 95% confidence interval [CI], 1.88–7.18;P < 0.001).

DISCUSSION

This study shows that maternal folate-related riskfactors for NTDs are not explained by thermolabileMTHFR, and it suggests that low maternal folate sta-tus itself may be the most important risk factor forNTD (whether from low intake or low absorption orother folate-related genetic mechanisms that act by re-ducing folate levels). These findings indicate that apolicy of genetic screening for the MTHFR variantwould not identify a large proportion of high-risk moth-ers. They also emphasize, in view of the probable mul-tiplicity of mechanisms giving rise to low maternal fo-late status, that food fortification may be the onlypopulation strategy of benefit in the effort to eliminateNTDs.

The MTHFR TT variant selects out those with verylow folate status among both the cases and the con-trols. Taking 200 mg/L as an arbitrary cutoff point toclassify low RCF status, 23% of controls and 20% ofcases who had RCF levels less than 200 mg/L in the

TABLE I. Median (95% Confidence Intervals [CI] of the Median) Red Cell Folate (RCF) andPlasma Folate (PF) Concentrations in Maternal Blood During NTD-Affected Pregnancies

(Cases) and Normal Pregnancies (Controls) Stratified By MTHFR Thermolabile Genotype*

Genotype Cases Controls Mann-Whitneymedian (95% CI) (no.) median (95% CI) (no.) cases versus

mg/L mg/L controls

RCFTT 211 (117–336) (13) 228 (184–324) (21) P 4 0.395CT 240 (190–343) (35) 315 (272–344) (121) P 4 0.018CC 268 (223–301) (34) 337 (306–353) (119) P < 0.001TT versus CTa P 4 0.366 P 4 0.039TT versus CCa P 4 0.172 P 4 0.005

PFTT 2.84 (1.89–5.58) (13) 2.78 (2.18–3.85) (20) P 4 0.606CT 3.26 (2.34–5.33) (34) 4.59 (3.97–5.22) (108) P 4 0.085CC 3.58 (2.58–4.77) (33) 5.12 (4.39–5.75) (114) P 4 0.003TT versus CTa P 4 0.877 P 4 0.010TT versus CCa P 4 0.893 P 4 0.004

*TT, MTHFR thermolabile homozygotes; CT, heterozygotes; CC, wild-type homozygotes.aMann-Whitney U-test.

TABLE II. The Effect of MTHFR TT Genotype on Median (95% Confidence Intervals [CI] ofthe Median) Plasma Homocysteine (Hcy) Concentrations in Maternal Blood During

NTD-Affected Pregnancies (Cases) and Normal Pregnancies (Controls)*

Genotype Cases Hcy Controls Hcy Mann-Whitneymedian (95% CI) median (95% CI) cases versus

mmol/L (no.) mmol/L (no.) controls

TT 10.36 (7.94–13.11) (11) 9.22 (8.12–10.87) (18) P 4 0.57CT/CC 8.22 (7.80–8.66) (64) 7.82 (7.48–8.03) (217) P 4 0.047TT versus CT/CCa P 4 0.104 P 4 0.026

*TT, represents MTHFR thermolabile homozygotes; CT, heterozygotes; CC, wild-type homozygotes.aMann-Whitney U-test.

Thermolabile MTHFR, Maternal Folate, and NTDs 157

initial analysis were later found to have the TT vari-ant. However, whereas the contribution of TT to lowRCF status was significant in the controls (OR, 4.5;95% CI, 1.65–12.24; P 4 0.006), this was not true forthe cases, even though the proportion of TT homozy-gotes was enriched in case mothers. Many NTD moth-ers who were not TT homozygotes had RCF levels lowerthan 200 mg/L. After exclusion of the TT homozygotes,only 10% of the remaining 240 controls had RCF levelsless than 200 mg/L compared with 29% of the 69 cases(OR, 3.67; 95% CI, 1.88–7.18; P < 0.001). Thus, it isclear that homozygosity for the C677T variant ofMTHFR does not account for much of the associationbetween low folate and NTDs.

Several reports, using human and animal data, haveemphasized the role of homocysteine in the cause ofNTDs [Steegers-Theunissen et al., 1994, 1995; vanAerts et al., 1994; Mills et al., 1995; van der Put et al.,1996] and have suggested mechanisms involving dis-ruption in homocysteine-related pathways. These re-ports suggest that embryonic homocysteine accumula-tion or a lack of methyl groups normally generatedthrough homocysteine-linked pathways are causallyimportant in NTDs. Until now it has not been clearwhether the raised plasma homocysteine seen in NTD-affected mothers [Mills et al., 1995] and in spina bifidafamilies [van der Put et al., 1995, 1996] was the resultof homozygosity for the thermolabile variant. Thisstudy shows that plasma homocysteines are also raisedin non-TT-affected cases, which is consistent with thelower blood folate status observed in these mothers.

In our study, the proportion of TT homozygotes washigher among NTD mothers than among control moth-ers (P 4 0.054). A higher incidence of thermolabile(TT) homozygotes has been reported among mothers ofNTD cases in some studies [Whitehead et al., 1995; vander Put et al., 1995, 1996] but not all [de Franchis et al.,1995; Papapetrou et al., 1996]. One study found anadditive effect for the genotype, giving the highest riskfor spina bifida if both mother and child were TT ho-mozygotes [van der Put et al., 1996]. The data pre-sented here do not enable us to predict a mechanism forthe increased risk of TT homozygous mothers havingan NTD-affected offspring, because the risk may be im-posed merely by self-selection into an already estab-lished risk group, namely that of low folate status.

It has been suggested that heterozygosity for theC677T mutation, which results in intermediate en-zyme activity, may also contribute to the risk for NTDs[Posey et al., 1996]. A small but significant reduction inthe serum folate of heterozygotes was seen in a largepopulation study of the C677T variant, which includedmore than 250 subjects in each of the CT and CC ge-notype categories [Harmon et al., 1996]. This study ofcase and control mothers and our previous report,which includes nonpregnant women [Molloy et al.,1997], shows that, although there is a trend towardlower blood folate levels in C677T heterozygotes, thedifference between wild-type (CC) and CT heterozy-gotes is not significant. This is also true for plasmahomocysteine levels in the cases and controls reportedhere. Thus, if heterozygosity for the C677T mutationdoes emerge as a causal risk factor for NTD, then it

may not be acting through reduced embryonic folatesor increased homocysteine.

In conclusion, we have shown that reduced blood fo-late is a common finding in NTD pregnancies, regard-less of MTHFR C677T genotype. Although the MTHFRC677T variant is an important cause of low folate sta-tus and a risk factor for NTDs, these results confirmthe need to screen new candidate genes for possiblemutations. They also emphasize that low maternal fo-late status may in itself be a risk factor for NTDs. Fu-ture investigations into NTDs should focus on geneticand biochemical conditions that are likely to result inlow folate status.

ACKNOWLEDGMENTS

We thank the Coombe, National Maternity, and Ro-tunda Hospitals for making possible the collection ofsamples. The Health Research Board Ireland, and theNational Institute of Child Health and Human Devel-opment provided financial support.

REFERENCES

Daly LE, Kirke PN, Molloy AM, Weir DG, Scott JM (1995): Folate levelsand neural tube defects: Implications for prevention. JAMA 274:1698–1702.

de Franchis R, Sebastio G, Mandato C, Andria G, Mastroiacovo P (1995):Spina bifida, 677T → C mutation and role of folate. Lancet 346:1703.

Engbersen AMT, Franken DG, Boers GHJ, Stevens EMB, Trijbels FJM,Blom HJ (1995): Thermolabile 5,10-methylenetetrahydrofolate reduc-tase as a cause of mild hyperhomocysteinemia. Am J Hum Genet 56:142–150.

Frosst P, Blom HJ, Milos R, Goyette P, Sheppard CA, Matthews RG, BoersGJH, den Heijer M, Kluijtmans LAJ, van den Heuvel LP, Rozen R(1995): A candidate genetic risk factor for vascular disease: A commonmutation in methylenetetrahydrofolate reductase. Nature Genet 10:111–113.

Harmon D, Woodside JV, Yarnell JWG, McMaster D, Young IS, McCrumEE, Gey KF, Whitehead AS, Evans AE (1996): The common ‘‘thermo-labile’’ variant of methylenetetrahydrofolate reductase is a major de-terminant of mild hyperhomocysteinaemia. Q J Med 89:571–577.

Kang SS, Wong PWK, Susmano A, Sora J, Norusis M, Ruggie N (1991):Thermolabile methylenetetrahydrofolate reductase: An inherited riskfactor for coronary artery disease. Am J Hum Genet 48:536–545.

Kang SS, Zhou J, Wong PWK, Kowalisyn J, Strokosch G (1988): Interme-diate homocysteinemia: A thermolabile variant of methylenetetrahy-drofolate reductase. Am J Hum Genet 43:414–421.

Kelleher BP, O’Broin SD (1991): Microbiological assay for vitamin B12performed in 96-well microtitre plates. J Clin Pathol 44:592–595.

Kirke PN, Mills JL, Whitehead AS, Molloy A, Scott JM (1996): Methylene-tetrahydrofolate reductase mutation and neural tube defects. Lancet348:1037–1038.

Kirke PN, Molloy AM, Daly LE, Burke H, Weir DG, Scott JM (1993):Maternal plasma folate and vitamin B12 are independent risk factorsfor neural tube defects. Q J Med 86:703–708.

Kluijtmans LAJ, van den Heuvel LPW, Boers GHJ, Frosst P, StevensEMB, van Oost BA, den Heijer M, Trijbels FJM, Rozen R, Blom HJ(1996): Molecular genetic analysis in mild hyperhomocysteinemia: Acommon mutation in the methylenetetrahydrofolate reductase gene isa genetic risk factor for cardiovascular disease. Am J Hum Genet 58:35–41.

Mills JL, McPartlin JM, Kirke PN, Lee YJ, Conley MR, Weir DG, Scott JM(1995): Homocysteine metabolism in pregnancies complicated by neu-ral tube defects. Lancet 345:149–151.

Molloy AM, Daly S, Mills JL, Kirke PN, Whitehead AS, Ramsbottom D,Conley MR, Weir DG, Scott JM (1997): Association of thermolabilevariant of 5,10-methylenetetrahydrofolate reductase with low red-cellfolates: Implications for folate intake recommendations. Lancet 349:1591–1593.

158 Molloy et al.

O’Broin S, Kelleher B (1992): Microbiological assay on microtitre plates offolate in serum and red cells. J Clin Pathol 45:344–347.

Ou CY, Stevenson RE, Brown VK, Schwartz CE, Allen WP, Khoury MJ,Rozen R, Oakley GP Jr, Adams MJ Jr (1996): 5,10-Methylene-tetrahydrofolate reductase genetic polymorphism as a risk factor forneural tube defects. Am J Med Genet 63:610–614.

Papapetrou C, Lynch SA, Burn J, Edwards YH (1996): Methylenetetrahy-drofolate reductase and neural tube defects. Lancet 348:58.

Posey DL, Khoury MJ, Mulinare J, Adams MJ Jr. Ou C-Y (1996): Is mu-tated MTHFR a risk factor for neural tube defects? Lancet 347:686.

Smith CA, Gough AC, Leigh PN, Summers BA, Harding AE, MaraganoreDM, Sturman SG, Schapira AH, Williams AC, Spurr NK, Wolf CR(1992): Debrisoquine hydroxylase gene polymorphism and susceptibil-ity to Parkinson’s disease. Lancet 339:1375–1377.

Steegers-Theunissen RPM, Boers GH, Blom HJ, Nijhuis JG, Thomas CMG,Borm GF, Eskes TK (1995): Neural tube defects and elevated homo-cysteine levels in amniotic fluid. Am J Obstet Gynecol 172:1436–1441.

Steegers-Theunissen RPM, Boers GHJ, Trijbels FJM, Finkelstein JD,Blom HJ, Thomas CMG, Borm GF, Wouters MGAJ, Eskes TKAB(1994): Maternal hyperhomocysteinemia: A risk factor for neural-tubedefects? Metabolism 43:1475–1480.

Ubbink JB, Vermaak WJH, Bissbort S (1991): Rapid high-performanceliquid chromatographic assay for total homocysteine levels in humanserum. J Chromatogr 565:441–446.

van Aerts LAG, Blom HJ, Deabreu RA, Trijbels FJM, Eskes TK, CopiusPeereboom-Stegeman JHG, Noordhoek J (1994): Prevention of neuraltube defects by and toxicity of L-homocysteine in cultured postimplan-tation rat embryos. Teratology 50:348–360.

van der Put NMJ, Steegers-Theunissen RPM, Frosst P, Trijbels FJM, Es-kes TKAB, van den Heuvel LP, Mariman ECM, den Heyer M, Rozen R,Blom HJ (1995): Mutated methylenetetrahydrofolate reductase as arisk factor for spina bifida. Lancet 346:1070–1071.

van der Put NMJ, van den Heuvel LP, Steegers Theunissen RPM, TrijbelsFJM, Eskes TKAB, Mariman ECM, den Heyer M, Blom HJ (1996):Decreased methylene tetrahydrofolate reductase activity due to the677C → T mutation in families with spina bifida offspring. J Mol Med74:691–694.

Whitehead AS, Gallagher P, Mills JL, Kirke PN, Burke H, Molloy AM,Weir DG, Shields DC, Scott JM (1995): A genetic defect in 5,10-methylenetetrahydrofolate reductase in neural tube defects. Q J Med88:763–766.

Thermolabile MTHFR, Maternal Folate, and NTDs 159