Details of the study sample have been provided elsewhere [8
]. Briefly, there were 727 case-parent triads in which the case individual had a CTRD (). The most common diagnoses among the cases were tetralogy of Fallot (38.4 percent), D-transposition of the great arteries (20.9
percent), and ventricular septal defect (20.1
percent). There was a predominance of males among the cases (59.2
percent), and the majority of parents were reported to be white (72.2
Characteristics of study cases and parents, Children's Hospital of Philadelphia, 1997–2007.
DNA samples were available for 1991 members of the study triads (i.e., 537 complete triads and 190 case-parent pairs). Genotype call rates for these 1991 samples ranged from 96% to 98% for each variant; the proportion of samples that provided discrepant results on repeat genotypes ranged from 0% to 0.8%; and the proportion of triads with genotype combinations that were incompatible with Mendelian inheritance ranged from 0.7% to 2.5% (n
families) per variant. On the basis of these results, all of the genotypes were considered to be of sufficiently high quality to include in the subsequent statistical analyses. However, all genotype data from families that included at least one genotype combination that was incompatible with Mendelian inheritance were omitted from all analyses (n
= 225 samples from 75 triads with a Mendelian inconsistency for one or more variant). In addition, all genotype data from individual samples that failed (i.e., no genotype called) or provided discrepant results on repeat genotyping for four or more of the genotyped variants were omitted from all analyses (n
samples). After the above-mentioned exclusions, 652 families (90%) were available for analysis, and the number of useable genotypes for each of the variants ranged from 1685 to 1715.
In our previous SNP-by-SNP analyses of these data [8
], three of the ten folate metabolic variants were found to be associated with CTRD at P
(unadjusted) < .05: MTR
A2756G (maternal effect, P
= .04), CBS
844ins68 (inherited effect, P
= .05), and MTHFR
A1298C (inherited effect, P
= .002). However, only the association with the inherited MTHFR
A1298C genotype remained significant when the false discovery rate was controlled at 0.05 (unadjusted P
= .0021 < .0025).
MDR-P was used to evaluate inherited gene-gene interactions and heterogeneity across the seven CTRD component phenotypes. Using this approach, the only model achieving significance was the one-locus model for the MTHFR A1298C variant (adjusted P = .02). There was no evidence of heterogeneity in the association of this variant across the seven CTRD component phenotypes, and no other one-, two-, three-, or four-locus model had an adjusted P value < .05 ().
MDR-P results (adjusted P values) for all 2-locus models of the inherited genotype and CTRD, Children's Hospital of Philadelphia, 1997–2007.
The case-only approach was used to assess the association of CTRD with maternal gene-gene interactions (). Using this approach, two maternal gene-gene combinations were associated with CTRD with unadjusted P
values less than .05: MTHFR
844ins68 (unadjusted P
= .01) and MTHFR
844ins68 (unadjusted P
= .02). Based on the data presented in , the MTHFR
677 TT genotype is over-represented and the MTHFR
1298 CC genotype is under-represented among case-mothers with the CBS
NN genotype. As these two MTHFR
variants are in strong linkage disequilibrium [8
], and given prior evidence that homocysteine levels are influenced by a MTHFR
844ins68 interaction [32
], only the MTHFR
interaction was evaluated using log-linear analyses.
Case-only results (unadjusted P values) for all pairwise combinations of maternal genotypes and CTRD, Children's Hospital of Philadelphia, 1997–2007.
Maternal MTHFR and CBS 844ins68 genotype combinations, Children's Hospital of Philadelphia, 1997–2007.
In the log-linear analyses, the maternal CBS 844ins68 IN and II genotypes were combined, due to the small number of II genotypes (n = 4), and the two resulting categories (NN versus IN + II) were used to stratify the data. An unrestricted model, which provides effect estimates for heterozygotes and for rare homozygotes, was fitted to the maternal MTHFR C677T genotype data. For these analyses, data from mothers who were genotyped for the MTHFR C677T but not the CBS 844ins68 variant were excluded (n = 2). Mothers who were CBS NN and MTHFR 677 TT had a 1.85-fold (95 percent confidence interval: 1.13, 3.02) higher risk of having a child with a CTRD as compared to those who were CBS NN and MTHFR 677 CC. This association was not seen in mothers with the CBS IN or II genotypes (). The unadjusted P-value for the LRT of heterogeneity of the effect of the maternal MTHFR C677T genotype across categories defined by the maternal CBS 844ins68 insertion genotype was 0.02, but this association was not significant at the 0.05 level, after applying the Bonferroni correction for the evaluation of all 44 maternal gene-gene combinations (P = .02 > .001).
Relative risk of CTRD for maternal MTHFR C677T/CBS 844ins68 genotype combinations estimated by log-linear regression, Children's Hospital of Philadelphia, 1997–2007.