Sex-Specific Genetic Susceptibility To Adverse Neurodevelopmental Outcome in Offspring of Pregnancies at Risk of Early Preterm Delivery

Abstract

Objective

To evaluate sex-specific genetic susceptibility to adverse neurodevelopmental outcome (ANO, defined as cerebral palsy [CP], mental, or psychomotor delay) at risk for early preterm birth (EPTB, < 32 weeks).

Study Design

Secondary case–control analysis of a trial of magnesium sulfate (MgSO₄) before anticipated EPTB for CP prevention. Cases are infants who died by the age of 1 year or developed ANO. Controls, matched by maternal race and infant sex, were neurodevelopmentally normal survivors. Neonatal DNA was evaluated for 80 polymorphisms in inflammation, coagulation, vasoregulation, excitotoxicity, and oxidative stress pathways using Taqman assays. The primary outcome for this analysis was sex-specific ANO susceptibility. Conditional logistic regression estimated each polymorphism’s odds ratio (OR) by sex stratum, adjusting for gestational age, maternal education, and MgSO₄-corticosteroid exposures. Holm–Bonferroni corrections, adjusting for multiple comparisons (p < 7.3 × 10⁻⁴), accounted for linkage disequilibrium between markers.

Results

Analysis included 211 cases (134 males; 77 females) and 213 controls (130 males; 83 females). An interleukin-6 (IL6) polymorphism (rs2069840) was associated with ANO in females (OR: 2.6, 95% confidence interval [CI]: 1.5–4.7; p = 0.001), but not in males (OR: 0.8, 95% CI: 0.5–1.2; p = 0.33). The sex-specific effect difference was significant (p = 7.0 10⁻⁴) and was unaffected by MgSO₄ exposure. No other gene–sex associations were significant.

Conclusion

An IL6 gene locus may confer susceptibility to ANO in females, but not males, after EPTB.

Neurodevelopmental abnormalities, including cerebral palsy (CP), autism, and neurodevelopmental delay, are more common in males than females.¹ While a portion of neurodevelopmental delay is attributed to X chromosome genetic abnormalities (e.g., fragile X syndrome),² the sex disparity related to CP and other prematurity-related brain injury are unexplained.

Preterm birth is the leading cause of childhood brain injury.³ Relative to females, preterm males have higher rates of neonatal death, and survivors are more likely to have long-term neurodevelopmental sequelae, including severe disability and CP.^4–9 This sex discrepancy persists even after controlling for perinatal, neonatal, and postnatal factors, suggesting that unmeasured biologic variables explain the neurodevelopmental disadvantage of preterm males.⁹ Neurobiological differences may confer differential susceptibility to fetal and neonatal brain injury.¹⁰ In vitro and in vivo data suggest that sex differences in brain injury patterns and therapeutic response may result from intrinsic differences in cell death pathways.^11–13

Sex-specific genetic predisposition may partly explain the observed neurobiological difference. In a candidate gene association study of term and preterm CPs, several polymorphisms in genes related to inflammation and coagulation were associated with CP in females, but not in males.¹⁴ There is a need for genetic association studies of adverse neuro-developmental outcomes (ANOs) in preterm infants that specifically consider the impact of infant sex.

We have previously shown that genetic variants are associated with ANOs in preterm children.^15–18 These analyses did not evaluate the potential effect of infant sex. Based on the known differences in susceptibility to brain injury among males and females, we hypothesized that there may be sex-specific genetic predisposition for neonatal death, CP, and neurodevelopmental delay. We aimed to evaluate whether there is sex-specific genetic susceptibility to ANO after early preterm birth (EPTB).

Materials and Methods

Subjects

Our subjects were children enrolled in the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network randomized, placebo-controlled, double-masked multicenter clinical trial of magnesium sulfate (MgSO₄) for prevention of CP before anticipated preterm birth. Women with singleton or twin gestations between 24^0/7 and 31^6/7 weeks’ gestation and at risk of imminent preterm delivery were eligible for enrollment. The primary outcome of the randomized clinical trial was a composite of stillbirth or infant death by 1 year of life, and moderate or severe CP at or beyond 2 years of age. Neurodevelopmental delay was included as a secondary outcome. This was assessed using the Bayley Scales of Infant Development II, which were administered at 2 years of age, and included mental and psychomotor developmental indices (abbreviated MDI and PDI). The details of the trial, which was conducted between 1997 and 2004, have been previously reported.¹⁹

Our nested case–control secondary analysis aimed to evaluate whether there is sex-specific genetic susceptibility to ANO, including death, after EPTB. Inclusion criteria were (1) enrollment in the original MgSO₄ randomized controlled trial, (2) neurologic outcome data at the age 2 of years in survivors, and (3) available DNA for genotyping. Cases were defined as live births who died by the age of 1 year or who had abnormal neurodevelopment at the age of 2 years. Both death and abnormal neurodevelopment were included in the composite outcome because death and abnormal neurodevelopment are competing outcomes. Abnormal neurodevelopment was defined as (1) CP or (2) mental or psychomotor delay. Mental and psychomotor delays were defined by scores of < 70 (2 standard deviations below the mean) on the Bayley Scales of Infant Development II MDI and PDI, respectively), which were administered by a trained psychologist or psychometrist. The diagnosis of CP was made by an annually certified pediatrician or pediatric neurologist according to prespecified criteria (gross motor delay, or abnormality in muscle tone, movement, or reflexes). Controls were survivors with normal neurodevelopment, defined as Bayley MDI and PDI ≥85, and no diagnosis of CP, periventricular leukomalacia or grade III/IV intraventricular hemorrhage.

The study group is illustrated in Fig. 1. There were 2,444 fetuses enrolled in the original MgSO₄ trial. We excluded infants with major congenital malformations and/or aneuploidy, and infants lost to follow-up. We also excluded subjects with no available DNA sample. Randomly selected control group children were matched to case group children for self-reported maternal race/ethnicity and infant sex to minimize the chance of misidentifying a race-based or sex-based genotype association. We excluded samples that either failed to genotype and/or had > 30% of genotypes missing, and then excluded the corresponding case or control group subjects, to maintain the matching design and ensure valid and accurate genotyping results. One twin from each pair was randomly excluded to avoid the issue of including related individuals in the analysis.

Fig. 1.

Selection of cases and controls from the original magnesium sulfate randomized trial.³⁴

Selection of Single-Nucleotide Polymorphisms for Analysis

Neonatal DNA was previously evaluated for 80 polymorphisms (33 genes) in inflammation, coagulation, vasoregulation, excitotoxicity, and oxidative stress pathways using Taqman assays^17,18 (Table 1). The candidate gene polymorphisms were selected based on previously described associations with CP and neurodevelopmental delay, or involvement in hypothesized causal pathways. Details and results of these analyses were previously reported.^17,18 For interleukin-6 (IL6) and interleukin-6 receptor (IL6R), single-nucleotide polymorphisms (SNPs) were selected using the HapMap III tagging algorithm to capture the common variation present in each gene. The included IL6 and IL6R polymorphisms define nearly 100% of haplotypes in Caucasians and African Americans. The remaining candidate genes did not have sufficient polymorphisms to define haplotypes.

Table 1.

Genetic variants included in the analysis

Gene	Symbol	Chr:position	RS number
C-reactive protein	CRP	1:159682233	rs1205
Interleukin 1β	IL1β	2:113594867 2:113595829 2:113598107	rs16944 rs1143623 rs4848306
Interleukin 6	IL6	7:22768219 7:22769773 7:22768707 7:22766645 7:22766246 7:22766221 7:22759655 7:22767433 7:22767871 7:22768027 7:22768479 7:22768572 7:22772260 7:22775663 7:22763009	rs1524107 rs1548216 rs1554606 rs1800795 rs1800796 rs1800797 rs1880243 rs2069832 rs2069835 rs2069837 rs2069838 rs2069840 rs2069852 rs11766273 rs12700386
Interleukin 6 receptor	IL6R	1:154368928 1:154382049 1:154370938 1:154389196 1:154436195 1:154418879 1:154394417 1:154426947 1:154400015 1:154422067 1:154380486 1:154400320 1:154438084 1:154404336 1:154404380 1:154430092 1:154442960 1:154389741 1:154369252	rs952146 rs 1386821 rs2054855 rs4075015 rs4240872 rs4537545 rs4601580 rs4845374 rs4845618 rs4845625 rs6427641 rs6687726 rs7514452 rs7549250 rs7549338 rs1126561 8 rs11265621 rs12090237 rs17654071
Interleukin 10	IL10	1:206946634 1:206946407 1:206946897	rs1800871 rs1800872 rs1800896
Interleukin 13	IL13	5:131995964	rs20541
Mannose-binding lectin 2	MBL2	10:54531226 10:54531685 10:54532014	rs1800451 rs7096206 rs11003125
Tumor necrosis factor α	TNFα	6:31543031	rs1800629
Toll-like receptor 1	TLR1	4:38807654	rs5743551
Toll-like receptor 2	TLR2	4:154607126 4:38807654	rs4696480 rs5743551
Toll-like receptor 4	TLR4	9:120475302 9:120475602	rs4986790 rs4986791
Factor II	F2	11:46761055	rs1799963
Factor V	F5	1:169519049	rs6025
Factor VII	F7	13:1 13773159	rs6046
Plasminogen activator inhibitor 1	PAI1	7:100781445 7:100769706	rs7242 rs1799768
Plasminogen activator inhibitor 2	PAI2	18:61564394 18:61570529	rs6098 rs6104
Methyltetrahy-drofolate reductase	MTHFR	1:1 1854476 1:1 1856378	rs1801131 rs1801133
N-methyl-D-aspartate receptor 2B subunit gene	GRIN2B	12:14134843	rs1019385
N-methyl-D-aspartate receptor 3A subunit gene	GRIN3A	9:104335682	rs3739722
N-methyl-D-aspartate receptor 3B subunit gene	GRIN3B	19:1005230	rs2240158
Vasoactive intestinal peptide	VIP	6:153067927	rs17083008
Vasoactive intestinal peptide Receptor 2	VIPR2	7:158851234 7:158821547	rs2098349 rs885861
Brain-derived neurotrophic factor	BDNF	11:27679916	rs6265
Aquaporin 4	AQP4	18:24436314 18:24430529	rs3906956 rs9951307
Transient receptor potential melastatin 7	TRPM7	15:50878630	rs8042919
Transthyretin	TTR	18:29172865	rs1800458
Reelin	RELN	7:103251161 7:103130403	rs362691 rs736707
Receptor for advanced glycation end products	RAGE	6:32146492	rs3134945
Kalirin	KALRN	3:123973586	rs1 1712039
Inducible nitric oxide synthase	NOS2	17:26105932	rs1137933
Nitric oxide synthase 3	eNOS3	7:150689943 7:150690176	rs1800779 rs3918226
Glutathione peroxidase 1	GPX1	3:49395757	rs1800668
Superoxide dismutase 2	SOD2	6:160113872	rs4880

Abbreviations: Chr, chromosome; RS, reference sequence.

Statistics

Demographic and clinical variables for case and control children enrolled in the parent trial were previously compared using chi-square, Fisher’s exact, or Wilcoxon’s rank-sum tests for maternal characteristics and generalized estimating equations were used for neonatal characteristics to account for the correlation between siblings.^17,18 Conditional logistic regression analysis was used to evaluate differences in the association between baseline characteristics and case/control status stratified by child sex with an F test for heterogeneity. This methodology accounted for matching by maternal race and child sex. A nominal p-value of 0.05 was considered statistically significant.

The primary analysis for this secondary study evaluated composite ANO (death by the age of 1 year, CP, and/or mental or psychomotor delay). Further analyses stratified neurodevelopmental outcome by psychomotor delay, mental delay, and death/CP. Conditional logistic regression determined the odds ratio (OR) for each polymorphism by sex stratum. We assumed an additive genetic pattern, in which each minor allele confers additional risk and two copies of the minor allele have twice the effect of one copy. The additive genetic model is commonly used when the genetic model is not known a priori. The analysis adjusted for demographic and clinical variables were significantly different between cases and controls or were known a priori to be associated with neurodevelopmental outcomes following preterm birth. Covariables considered included gestational age at birth, maternal education level, clinical chorioamnionitis, cesarean delivery, singleton gestation, and exposure to MgSO₄ corticosteroids. Maternal race/ethnicity was controlled through matching and accounted for with conditional logistic regression. An interaction term between child sex and genotype tested heterogeneity across strata. Analyses then determined if in utero MgSO₄ exposure modified the association between polymorphism and neurodevelopmental outcome, within sex strata.

After accounting for possible linkage disequilibrium between polymorphisms, the Holm–Bonferroni method was used to adjust for multiple comparisons.²⁰ A p-value of < 7.7 × 10⁻⁴ was considered to be statistically significant. The unadjusted and adjusted p-values are both reported. Exact tests for Hardy–Weinberg equilibrium²¹ were performed on control subjects for each polymorphism. All calculations were performed using SAS software version 9.2 (SAS Institute, Inc, Cary, NC) and R version 2.10.0.

The Institutional Review Boards (IRBs) of all participating centers approved the original trial and all participants gave written informed consent before enrollment. After IRB review, this secondary analysis was determined to be exempted from IRB approval procedures secondary to deidentification of data and study samples prior to analysis.

Results

Four hundred and twenty-four subjects, 211 cases (134 males and 77 females) and 213 controls (130 males and 83 females), were analyzed. Among the cases, there were 44 children who died by the age of 1 year and 167 who had abnormal neurodevelopment at the age of 2 years. Of children with abnormal neurodevelopment, 25 infants had CP, and 142 infants had neurodevelopmental delay.

As previously reported, children with DNA available for analysis were more likely to be born at a later gestational age and more likely to be singleton compared with those without DNA available.^17,18

The demographic and clinical characteristics of our study subjects are shown in Table 2. There is a statistically significant difference in the association between maternal education and case/control status by sex (p = 0.02). For males, maternal education ≥ 12 years was significantly less in cases versus controls. This difference was not observed in females. There was also a statistically significant difference in the association between clinical chorioamnionitis and case/control status by sex (p = 0.03). For females, clinical chorioamnionitis was more common in cases versus controls. This difference was not observed in males. After stratifying by child sex, cases and controls were similar with regard to maternal ethnicity, exposure to MgSO₄ and antenatal corticosteroids, gestational age at delivery, mode of delivery, and singleton gestation. Maternal education was included as a covariable in the logistic regression analysis. Clinical chorioamnionitis was excluded as a covariable secondary to small numbers of affected cases and controls.

Table 2.

Maternal and neonatal characteristics of cases and controls^a, by neonatal sex^b

Characteristic	Males (n = 264)			Females (n = 160)			p-Value for heterogeneity
	Controls (n = 130)	Cases (n = 134)	p-Value	Controls (n = 83)	Cases (n = 77)	p-Value
Ethnicity
African American	67 (51.5)	67 (50)	0.96	32 (38.6)	28 (36.4)	0.96	0.99
Caucasian	41 (31.5)	43 (32.1)		29 (34.9)	28 (36.4)
Hispanic	22 (16.9)	24 (17.9)		22 (26.5)	21 (27.3)
Maternal education (≥12 years)	102 (78.5)	74 (55.2)	< 0.001	51 (61.4)	45 (58.4)	0.66	0.02
Magnesium sulfate exposure	63 (48.5)	74 (55.2)	0.27	37 (44.6)	28 (36.4)	0.31	0.14
Antenatal corticosteroid exposure	127 (97.7)	130 (97.0)	0.75	80 (96.4)	75 (97.4)	0.72	0.63
Clinical chorioamnionitis	18 (13.8)	14 (10.4)	0.41	7 (8.4)	16 (20.8)	0.03	0.03
Gestational age at birth (weeks)	31.0 ± 2.5	29.5 ± 3.3	<0.001	30.8 ± 2.7	29.1 ± 3.1	< 0.001	0.91
Preterm delivery
< 37 weeks	126 (96.9)	131 (97.8)	0.67	82 (98.8)	75 (97.4)	0.49	0.42
< 34 weeks	122 (93.9)	122 (91.0)	0.40	73 (88.0)	74 (96.1)	0.08	0.06
< 28 weeks	13 (10.0)	50 (37.3)	< 0.001	14 (16.9)	31 (40.3)	0.001	0.32
Cesarean delivery	49 (37.7)	50 (37.3)	0.934	26 (31.3)	31 (40.3)	0.24	0.33
Singleton gestation	120 (92.3)	122 (91)	0.72	80 (96.4)	69 (89.6)	0.11	0.25

^aCases are infants who died by the age of 1 year or developed CP, mental, or psychomotor delay by the age of 2 years. Controls were survivors with normal neurodevelopment, defined as Bayley MDI and PDI ≥85, and no diagnosis of CP, periventricular leukomalacia or grade III/IV intraventricular hemorrhage.

^bData presented as mean ± standard deviation or n (%) unless otherwise specified.

Four SNPs failed Hardy–Weinberg disequilibrium in the controls and were excluded from the analysis. A polymorphism in the inflammatory cytokine IL6 gene (rs2069840) was associated with composite ANO in females (OR: 2.6, 95% confidence interval [CI]: 1.5–4.7; p = 0.001), but not in males (OR: 0.8, 95% CI: 0.5–1.2; p = 0.33). The effect difference between males and females was significant (p = 7.0 × 10⁻⁴). MgSO₄ exposure did not modify this association. Although several additional polymorphisms showed a sexspecific difference in association with composite ANO, these gene–sex associations were not significant after correction for multiple comparisons. Genotype frequencies and ORs for cases and controls are reported, by sex, for polymorphisms associated with composite ANO at p ≤ 0.05 for heterogeneity across strata (Table 3).

Table 3.

Genotype frequencies and odds ratios for cases and controls,^a by sex, for composite adverse neurodevelopment^b

Gene	RS	Allele	Males								Females								p-Value for heterogeneity
			Cases			Controls			OR (95% Cl)	p-Value	Cases			Controls			OR (95% Cl)	p-Value
		m/M	MM	Mm	mm	MM	Mm	mm			MM	Mm	mm	MM	Mm	mm
IL6	2069840	G/C	60.5	30.6	8.9	50.8	38.3	10.8	0.8 (0.5–1.2)	0.33	39.4	45.1	15.5	63.3	32.9	3.8	2.6 (1.5–4.7)	0.001	0.0007
IL6	1524107	T/C	74.2	19.6	6.2	77.5	19.1	3.4	1.3 (0.7–2.3)	0.45	83.0	17.0		70.2	24.6	5.3	0.4 (0.1–1.0)	0.05	0.03
IL6	2069852	A/G	75.5	22.3	2.1	81.1	18.9		1.7 (0.8–3.6)	0.15	86.0	14.0		79.0	17.5	3.5	0.5 (0.2–1.4)	0.18	0.05
N0S2	1137933	A/G	58.7	37.2	4.1	72.3	24.4	3.4	1.7 (1.0–2.7)	0.05	67.1	30.0	2.9	61.0	32.5	6.5	0.8 (0.4–1.4)	0.36	0.05
F7	6046	A/G	79.3	19.8	0.9	80.9	17.3	1.8	1.0 (0.5–2.0)	0.93	90.2	8.2	1.6	68.8	28.1	3.1	0.2 (0.1–0.6)	0.004	0.02

Abbreviations: CI, confidence interval; OR, odds ratio; RS, reference sequence.

^aCases are infants who died by the age of 1 year or developed CP, mental, or psychomotor delay by the age of 2 years. Controls were survivors with normal neurodevelopment, defined as Bayley MDI and PDI >85, and no diagnosis of CP, periventricular leukomalacia or grade III/IV intraventricular hemorrhage

^bAdditive genetic model used. All models adjusted for gestational age at birth, maternal education level, and exposure to magnesium sulfate (MgSO4). Maternal race/ethnicity was controlled through matching. Results reported for polymorphisms associated with composite adverse neurodevelopmental outcome at p < 0.05 for heterogeneity across strata. Only the IL6 polymorphism rs2069840 (bold) was significant different between males and females after correction for multiple comparisons

Genotype frequencies and ORs for cases and controls are reported, by sex, for polymorphisms associated with the stratified outcomes of psychomotor delay, mental delay, and death/CP at p ≤ 0.05 for heterogeneity across strata (Table 4). None of the polymorphism effect differences between males and females were significant at the Holm–Bonferroni corrected p-value for these outcomes.

Table 4.

Genotype frequencies and odds ratios for cases and controls,^a by sex, for stratified neurodevelopmental outcomes^b

Gene	RS	Allele	Males								Females								p-Value for heterogeneity
			Cases			Controls			OR (95% Cl)	p-Value	Cases			Controls			OR (95% Cl)	p-Value
		m/M	MM	Mm	mm	MM	Mm	mm			MM	Mm	mm	MM	Mm	mm
Psychomotor delay
IL1β	1143623	G/C	57.9	36.8	5.3	70.7	25.9	3.4	2.1 (0.8–5.4)	0.14	70.6	20.6	8.8	51.4	42.9	5.7	0.6 (0.2–1.8)	0.32	0.04
IL6	2069840	G/C	59.3	33.9	6.8	49.1	36.8	14.0	0.5 (0.2–1.1)	0.10	46.9	43.8	9.4	66.7	27.3	6.1	2.0 (0.7–5.5)	0.20	0.05
IL10	1800896	C/T	27.3	60.0	12.7	44.6	42.9	12.5	5.3 (1.6–17.6)	0.007	51.5	24.2	24.2	35.3	41.2	23.5	0.6 (0.2–1.4)	0.22	0.005
AQP4	9951307	G/A	27.6	58.6	13.8	36.2	46.6	17.2	1.0 (0.5–2.1)	0.91	21.9	62.5	15.6	43.8	50.0	6.3	NA	0.07	0.03
Mental delay
IL6	1880243	A/C	65.3	25.0	9.7	65.8	31.5	2.7	1.0 (0.5–1.9)	0.93	58.8	35.3	5.9	78.4	21.6	0	4.1 (0.9–18.7)	0.07	0.02
IL6	1800796	C/G	76.3	17.1	6.6	74.7	20.0	5.3	1.2 (0.6–2.4)	0.62	77.8	22.2	0	67.7	23.5	8.8	0.3 (0.1–0.9)	0.04	0.02
IL6	2069840	G/C	58.1	31.1	10.8	53.5	35.2	11.3	0.7 (0.4–1.5)	0.37	42.4	42.4	15.2	66.7	27.8	5.6	3.6 (1.1–12.1)	0.04	0.02
IL6R	7549338	C/G	32.9	51.4	15.7	46.5	40.8	12.7	1.9 (1.0–3.7)	0.05	41.9	41.9	16.1	29.4	47.1	23.5	0.6 (0.2–1.9)	0.37	0.04
IL6R	7549250	C/T	22.9	57.1	20.0	48.4	34.4	17.2	1.7 (0.9–3.2)	0.08	30.6	58.3	11.1	25.7	51.4	22.9	0.4 (0.2–1.2)	0.09	0.01
F7	6046	A/G	74.2	25.8	0	81.8	16.7	1.5	2.0 (0.7–5.4)	0.20	96.4	0	3.6	69.0	27.6	3.5	0.1 (0.0–1.5)	0.10	0.03
TLR1	5743551	T/C	45.7	35.7	18.6	39.2	32.4	28.4	0.6 (0.3–1.2)	0.18	29.7	48.6	21.6	54.3	28.6	17.1	2.6 (0.9–7.3)	0.08	0.02
Death/cerebral palsy
IL1β	16944	A/G	34.9	34.9	30.2	37.8	50.0	12.2	1.9 (0.9–4.1)	0.12	29.2	45.8	25.0	34.0	38.3	27.7	0.5 (0.2–1.2)	0.13	0.01
IL1β	4848306	A/G	65.1	23.3	11.6	36.1	50.6	13.3	0.5 (0.2–1.3)	0.15	52.0	36.0	12.0	54.2	37.5	8.3	1.7 (0.7–3.9)	0.23	0.04
IL6	2069840	G/C	67.6	24.3	8.1	51.3	36.3	12.5	0.9 (0.3–2.6)	0.84	33.3	45.8	20.8	62.5	35.4	2.1	4.2 (1.3–13.4)	0.02	0.01
TNFα	1800629	A/G	89.2	10.8	0	69.2	29.5	1.3	0.4 (0.1–1.9)	0.24	52.4	47.6	0	66.7	31.0	2.4	10.4 (1.1–102.6)	0.04	0.01
MBL2	7096206	G/C	67.5	30.0	2.5	69.1	28.4	2.5	0.9 (0.3–2.6)	0.79	79.2	20.8	0	65.9	27.3	6.8	0.2 (0.0–0.9)	0.04	0.05
F7	6046	A/G	82.4	17.6	0	80.3	18.4	1.3	0.3 (0.1–1.5)	0.15	90.9	9.1	0	64.1	30.8	5.1	0.0 (0.0–0.6)	0.03	0.02
RAGE	3134945	A/C	76.7	20.9	2.3	65.5	28.7	5.7	0.4 (0.1–1.5)	0.19	47.8	52.2		66.0	30.0	4.0	7.6 (1.1–54.1)	0.04	0.01

Abbreviations: CI, confidence interval; OR, odds ratio; RS, reference sequence.

^aCases are infants who died by the age of 1 year or developed CP, mental, or psychomotor delay by the age of 2 years. Controls were survivors with normal neurodevelopment, defined as Bayley MDI and PDI ≥85, and no diagnosis of CP, periventricular leukomalacia or grade III/IV intraventricular hemorrhage.

^bAdditive genetic model used. All models adjusted for gestational age at birth, maternal education level, and exposure to magnesium sulfate (MgSO4). Maternal race/ethnicity was controlled through matching. Results reported for polymorphisms associated with composite adverse neurodevelopmental outcome at p < 0.05 for heterogeneity across strata. No loci were significantly different between males and females after correction for multiple comparisons.

Comment

We used a candidate gene approach to identify sex-specific genetic risk factors associated with neurodevelopment at the age of 2 years of offspring delivered from pregnancies at risk for EPTB. An IL6 gene locus may confer susceptibility to ANO in females, but not in males. Our results lend support to the hypothesis that inflammatory cytokine gene polymorphisms influence susceptibility to brain injury in preterm children. Our results also suggest that unique genotype–sex interactions may further modify risk. These results do not explain the observed increased risk of death and neurodevelopmental disability in preterm males, relative to females.

Proinflammatory cytokines in amniotic fluid and cord blood have been strongly associated with short- and long-term neonatal complications after preterm birth.^22–25 The magnitude of the inflammatory response is partly regulated by genetic factors. The IL6 gene, located on chromosome 7, is involved in regulation of innate immunity. While several genes have been linked with neurodevelopmental outcomes after preterm birth, IL6 is a particularly compelling genetic target. IL6 gene variants have been associated with CP in term and near-term children.^26,27 IL6 gene variants have also been associated with CP and neurodevelopmental delay children born preterm.^15,27–29

While many studies have demonstrated an increased risk of neonatal death and long-term neurodevelopmental sequelae in preterm males, relative to females,^4–9 few studies have evaluated how gene–sex interactions might further influence outcomes. A study by Gibson et al reported sexspecific genetic associations with CP; however, preterm and term children were analyzed together and polymorphic sites within the IL6 gene were not evaluated.¹⁴

The IL6 rs2069840 SNP is located within an intron and C is the ancestral allele (G is derived). Data from the HapMap project report a frequency of the derived G allele is ~0.8 in the YRI (sub-Saharan African) population, 0.9 in CHB (Asian) and JPT (Asian) populations, and 0.7 in the CEU (European) population.³⁰ The CC homozygous ancestral genotype is present in most Asians (~0.9) and sub-Saharan Africans (0.7) and approximately half of Europeans (0.5). In this study, ~51% of control males had the CC homozygous ancestral genotype, compared with 60% of cases. Approximately, 63% of control females had the CC homozygous ancestral geno-type, compared with 39% of cases.

In this analysis, the G allele of IL6 was associated with increased risk of ANO in females, but not males. In a genetic analysis of adult cardiovascular disease risk, this polymorphism was differentially associated with disease in men and women (G allele was protective for men in a recessive model).³¹ The effect of rs2069840 on IL6 cytokine expression is not known. Another polymorphism within the promoter of the IL6 gene, rs1800795, has been associated with sexspecific susceptibility to celiac disease in girls³² and with CP in term and near-term males.²⁶ We did not see such an association in this study, at least in part because of the difference in study ascertainment.

The mechanisms underpinning sex-specific disease susceptibilities are poorly understood. Although hormonal differences may explain sex-specific susceptibility to many adult diseases, differences in the immature brain may be more rooted in genetic differences that influence cellular mechanisms.¹⁰ In vitro and in vivo data suggest that sex-specific differences in susceptibility and response to early brain injury may result from intrinsic differences in cell death pathways.^11–13 In these models of cellular apoptosis, males appear to be more vulnerable to glutamate-mediated excitotoxicity, and females appear to be more vulnerable to oxidative stress. Importantly, animal models have also demonstrated differential response to neuroprotection strategies.^33,34

A major strength of this study is that it was nested in a multicenter trial with well-characterized maternal and neonatal outcomes and standardized prospective collection of data and biospecimens. In addition, a critical review of genotype–sex association studies suggested three criteria for appropriate analysis and publication: (1) the genetic effect is based on the same genetic model in both sexes (e.g., additive), (2) different genetic subsets in the two sexes are not compared (e.g., term males and preterm females), and (3) a statistically significant test for gene–sex interaction is presented (defined as a nominal p-value threshold of 0.05).³⁵ Our study fulfills these criteria and also appropriately corrects for multiple comparisons.

Our study has several limitations. Sample size did not permit stratification of the analysis by maternal race/ethnicity. Small sample size and lack of placental pathology also limited our ability to evaluate sex–environment interactions of interest, such as the interaction between fetal genotype and intrauterine infection. Interaction tests are usually underpowered since clinical trials are powered to test main effects. Positive associations could therefore have been missed in this analysis, particularly for the stratified outcomes where numbers were even more limited. This could result in important gene associations being missed. Further, sample size did not permit haplotype analysis for genes in which multiple SNPs were evaluated (particularly IL6 and IL6R). Candidate genes of potential interest need to be further evaluated in other cohorts.

Given correction for multiple corrections, the chance of type I error is very low. However, replication in validation cohorts will be necessary to support our IL6 association. In addition, several candidate gene polymorphisms were not significantly associated with neurodevelopment at the Holm–Bonferroni corrected p-value of < 7.3 × 10⁻⁴, but had p ≤ 0.05, and may be worthy of re-evaluation in future studies.

Our definition of ANO included abnormal neurodevelopment (CP, mental, and/or psychomotor delay) and death, as these are competing outcomes. Similarly, death and CP were analyzed as a combined outcome in a predefined subanalysis as this was the primary outcome of the randomized clinical trial.¹⁹ A child cannot be diagnosed with CP at the age of 2 years if that child has already died. Small numbers limited meaningful analyses of CP or death alone. It should be noted that neurodevelopmental testing at the age of 2 years is of limited predictive value, particularly for subtle deficits. Whether the IL6 polymorphism we describe might have any longer term, sex-specific neurocognitive effects of clinical significance is uncertain, and cannot be extrapolated from these data. An IL6 polymorphism in linkage disequilibrium with rs2069840 may be the actual causative variant associated with ANOs.

We identified a variant in IL6 that may uniquely confer risk of ANO to preterm females. The effect of genotype–sex interactions on neurodevelopmental outcomes after preterm birth remains largely unexplored. Identification of sex-specific genetic susceptibility to ANOs following preterm birth may improve understanding of the pathogenesis and facilitate identification of new neuroprotection strategies.

Appendix A

In addition to the authors, other members of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal–Fetal Medicine Units Network are as follows:

University of Utah, Salt Lake City, UT—E. Clark, K. Anderson, M. Jensen, L. Williams (University of Utah); L. Fullmer (Utah Valley Regional Medical Center); and A. Guzman (McKay-Dee Hospital).

University of Texas Medical Branch, Galveston, TX—G. Hankins, T. Wen, L.A. Goodrum, G.R. Saade, G.L. Olson, H. M. Harirah, and E. Martin.

University of Alabama at Birmingham, Birmingham, AL—J.C. Hauth, A. Northen, T. Hill-Webb, S. Tate, K. Nelson, and F. Biasini.

University of Texas Southwestern Medical Center, Dallas, TX—M.L. Sherman, J. Dax, L. Faye-Randall, C. Melton, E. Flores.

Case Western Reserve University-MetroHealth Medical Center, Cleveland, OH—M. Collin, G. VanBuren, C. Milluzzi, M. Fundzak, and C. Santori.

The Ohio State University, Columbus, OH—J. Iams, F. Johnson, M.B. Landon, C. Latimer, V. Curry, and S. Meadows.

Thomas Jefferson University, Philadelphia, PA—R. Wapner, A. Sciscione, M.M. DiVito, M. Talucci, S. Desai, D. Paul.

University of Tennessee, Memphis, TN—B.M. Sibai, R. Ramsey, W. Mabie, L. Kao, and M. Cassie.

Wayne State University, Detroit, MI—G.S. Norman, D. Driscoll, B. Steffy, and M.P. Dombrowski.

Wake Forest University Health Sciences, Winston-Salem, NC—P.J. Meis, M. Swain, K. Klinepeter, M. O’Shea, and L. Steele.

University of North Carolina at Chapel Hill, Chapel Hill—K.J. Moise, Jr., S. Brody, J. Bernhardt, and K. Dorman.

University ofTexas Health Science Center at Houston, Houston, TX—L.C. Gilstrap, III, M.C Day, E. Gildersleve, F. Ortiz, and M. Kerr.

Columbia University, New York, NY—V. Pemberton, L. Paley, C. Paley, S. Bousleiman, and V. Carmona.

Brown University, Providence, RI—J. Tillinghast, D. Allard, B. Vohr, L. Noel, and K. McCarten.

University of Cincinnati, Cincinnati, OH—M. Miodovnik, N. Elder, W. Girdler, and T. Gratton.

University of Chicago, Chicago, IL—A.H. Moawad, M. Lind-heimer, and P. Jones.

University of Miami, Miami, FL—F. Doyle, C. Alfonso, M. Scott, and R. Washington.

Northwestern University, Chicago, IL—G. Mallett, M. Ramos-Brinson, and P. Simon.

University of Texas at San Antonio, San Antonio, TX—O. Langer, E. Xenakis, D. Conway, and M. Berkus.

University of Pittsburgh, Pittsburgh, PA—T. Kamon (deceased), M. Cotroneo, and C. Milford.

The George Washington University Biostatistics Center—E. Thom, S. Weiner, B. Jones-Binns, M. Cooney, M. Fischer, S. McLaughlin, K. Brunette, and E. Fricks.

National Institute of Neurological Disorders and Stroke, Bethesda, MD—D. Hirtz and K.B. Nelson.

Eunice Kennedy Shriver National Institute of Child Health and Human Development, Bethesda, MD—C. Spong, S. Tolivaisa, D. McNellis, C. Catz, and K. Howell.

MFMU Network Steering Committee Chair (University of Pittsburgh, Pittsburgh, PA)—J. Roberts.