Main

Intraventricular hemorrhage (IVH), characterized as bleeding into the ventricular system of the developing brain, is one of the leading morbidities for very LBW (VLBW) preterm neonates (1,2). IVH ranges in severity from grade I to the most severe grade IV. The incidence for IVH grades I–IV is around 27% in neonates weighing <1500 g (1). Approximately, 45–85% of premature infants with the more severe grades of IVH (grades III–IV) incur cognitive disabilities such as CP and mental retardation, whereas infants with milder grades of IVH (grades I–II) are at risk for developmental delays (3,4). Risk factors for IVH include LBW, early GA, male gender, maternal smoking, preterm premature rupture of membranes (PPROM), chorioamnionitis, early onset sepsis, RDS, patent ductus arteriosus (PDA), and pneumothorax (2,5–8). However, many of these have not been consistently shown as risk factors for IVH and they do not entirely explain the etiology and pathogenesis of this complex disorder (9).

Recently, genetic factors have been implicated in the risk for developing IVH in both term and preterm infants (10–15). Twin studies suggest that shared genetic and environmental risk factors explain 41.3% of the risk for developing IVH after controlling for gender, GA, and birth weight (16). Previous studies of genetic association with IVH have focused on genes related to either inflammation and infection or complement and coagulation. IL 1 β-511T (rs16944), IL4-590T (rs2243250), IL6-174C (rs1800795), and TNF- α-308 (rs1800629) all associated with IVH (10,17,18). In addition, IL10-1082A (rs1800896) is associated with an increased risk of periventricular leukomalacia (PVL), a condition that often occurs in conjunction with IVH (19,20). The coagulation factor V (FV) Leiden mutation (rs6025), a coagulation factor II (FII), prothrombin polymorphism (G20210A, rs1799963), and a coagulation factor XIII (FXIII) missense mutation (Val34Leu, rs5985) have been implicated in several studies as risk factors for the development of IVH (12,14,15,21,22). In addition, two genes integrin beta-3 (ITGB3) and estrogen receptor-alpha (ESR1) have been associated with IVH (13,23). However, many of these associations have not been tested in multiple independent populations or have not consistently replicated across studies.

To determine whether genetic associations previously identified replicate, we examined 10 candidate genes for association with IVH susceptibility: IL1 β, IL4, IL6, IL10, TNF, FII, FV, ITGB3, and ESR1. We chose either the same single nucleotide polymorphism (SNP) previously associated with IVH or a SNP in high linkage disequilibrium (LD) with the associated SNP and tested associations in 271 preterm infants weighing <1500 g.

MATERIALS AND METHODS

Study population.

Premature infants (delivery before 37 wk of gestation) admitted to the NICU at the University of Iowa Children's Hospital between 2000 and 2009 were recruited to examine preterm birth (PTB) and neonatal complications of prematurity. Blood or buccal swabs from infants and their parents have been collected and banked. Informed consent was obtained from participating families, and the study was approved by the University of Iowa Institutional Review Board (200506792) for sample recruitment and to access the associated clinical information necessary for this study. This population is a subset of one that has been described previously but evaluated for PTB with respect to the progesterone receptor and genes affecting cholesterol metabolism (24,25).

The first analysis consisted of 271 unrelated infants born <32 wk of gestation and weighing <1500 g. There were 48 sets of twins in this study, one twin was chosen from each pair. The twin with the most severe case of IVH was selected, if both twins had the same IVH status one twin was randomly selected. Cases were defined as infants with IVH grades I–IV, and controls were those without documented IVH. IVH grade was confirmed by ultrasonography with grade I defined as blood in the periventricular germinal matrix, grade II as blood in the ventricular system without ventricular dilatation, grade III as blood in the lateral ventricles with ventricular dilatation, and grade IV as blood in the ventricular system with parenchymal extension. We studied 207 infants without IVH, 28 with grade I, 10 with grade II, 16 with grade III, and 10 with grade IV. Demographics of this population are described in Table 1. A second phase of the study was performed after removing confounders that could potentially interfere with inflammation/infection and coagulation genetic associations. Exclusions included infants of women with heart disease (0.37%), bleeding disorder (1.1%), autoimmune disease (0.0%), thrombocytopenia (2.2%), gestational diabetes (3.0%), type I diabetes (0.74%), type II diabetes (1.1%), chronic hypertension (8.9%), preeclampsia (28.0%), eclampsia (1.1%), gestational hypertension (5.5%), hemolysis, elevated liver enzymes and low platelet count (HELLP) syndrome (7.7%), infants with congenital anomalies (3.0%), infants who were a twin (17.7%), and infants with one or both parents of non-Caucasian descent (23.2%). Race of infant was determined through self-reported questionnaire of the mother. The remaining subset included 103 infants, 81 without IVH, 10 with grade I, 4 with grade II, 8 with grade III, and none with grade IV.

Table 1 Demographic characteristics of 271 unrelated infants with (n= 64) and without (n= 207) IVH
Full size table

DNA processing and genotyping.

DNA was extracted from blood or buccal swabs (26). Ten SNPs were chosen for analysis with IVH; seven have been shown to associate with IVH previously (IL10-1082 rs1800896, TNF-308 rs1800629, FII G20210A rs1799963, FV G1691A rs6025, FXIII Val34Leu rs5985, ITGB3 Leu33Pro rs5918, ESR1 rs2234693) and three were in strong LD with SNPs previously reported to be associated with IVH (IL1β-31 rs1143627 with IL1β-511 rs16944 r2 = 0.96; IL6 rs2069832 with IL6-174 rs1800795, r2 = 0.96; and IL4 rs2243270 with IL4-590 rs2243250, r2 = 0.94). LD was determined using the Caucasian population (CEU) from Hapmap. Genotyping was performed using TaqMan (Applied Biosystems, Foster City, CA), as previously described (24). Allele scoring was done using the Sequence Detection Systems software (version 2.2, Applied Biosystems). The genotype data were uploaded into a Progeny database (Progeny Software, LLC, South Bend, IN), also containing phenotypic data, for subsequent statistical analysis.

Statistical analysis.

Demographic characteristics were compared between infants without IVH (n = 207) and infants with IVH grades I–IV (n = 64) using the Wilcoxon rank sum test when comparing continuous traits and Fisher's exact test for dichotomous traits. The first dataset included 271 infants born <32 wk of gestation and weighing <1500 g. The second set included 103 infants after excluding the potential confounders described above. Markers were tested for deviations from Hardy-Weinberg Equilibrium (HWE) with Fisher's exact tests. Fisher's exact tests were also used to compare genotype and allele frequencies between the following groups; 1) infants without IVH and those with IVH grades I–IV; 2) infants without IVH and those with IVH grades I–II, and 3) infants without IVH and those with IVH grades III–IV. A Bonferroni significance level of p < 5 × 10−3 (0.05/10 independent tests) was used to correct for multiple testing. Logistic regression was performed on SNPs with significant (p < 0.05) genotype differences. The OR and CI using the Woolf test was calculated for SNPs with significant (p < 0.05) allele frequency differences. Statistical analysis was performed in Stata version 10.1 (Stata Corp, College Station, TX). In addition, logistic regression was performed after controlling for factors that differed by IVH status (Table 1), specifically appearance, pulse, grimace, activity, respiration (Apgar score) at 1 and 5 min, cystic PVL (c-PVL), twin status, and PDA. Power analysis was performed with PS Power (27).

RESULTS

Demographic data.

Birth weight, GA, race, gender, retinopathy of prematurity (ROP), RDS, necrotizing entercolitis, (NEC), sepsis, pneumothorax, smoking during pregnancy, diabetes, hypertension, PPROM, maternal preexisting conditions, thrombocytopenia, and congenital anomalies did not differ between infants with IVH compared to those without IVH (Table 1). As expected c-PVL was more frequent in infants with IVH (17.2%) compared to those without IVH (2.4%; p = 9.7 × 10−5). This association has been documented in previous reports (10). PDA was also more common in infants with IVH (53.1%) compared to those without IVH (35.8%; p = 0.02). In addition, as expected infants with IVH had lower Apgar scores at 1 and 5 min compared to those without IVH. There was also a higher incidence of twins among infants with IVH (31.3%) compared to those without (13.5%).

Genetic associations with IVH.

All markers were in HWE in infants with and without IVH in the full dataset (n = 271). In the dataset after exclusion of potential confounders all the markers were in HWE with the exception of ESR1 rs2234693 (p = 0.03) in infants with IVH. When comparing infants with and without IVH, we detected allele and genotype frequency differences for IL1 β rs1143627 and FV rs6025 in both the full dataset and the dataset with exclusions (Table 2). Infants with the CC genotype (OR = 3.1, CI = 1.3–7.5, p = 0.01), the CT genotype (OR = 2.2, CI = 1.1–4.7, p = 0.04), or the C allele (OR = 1.8, CI = 1.2–2.8, p = 5 × 10−3) of IL1 β rs1143627 were at increased risk for IVH compared to infants with the TT genotype or T allele. After excluding infants with potential confounders the association with IVH and the CT genotype of IL1 β rs1143627 remained (OR = 4.2, CI = 1.2–14.6, p = 0.03); however, the associations with the CC genotype (p = 0.56) and the C allele (p = 0.23) were no longer significant, possibly because of the decreased sample size and lack of power to detect the effects (Table 2). After controlling for factors that significantly differed by IVH in Table 1; i.e. Apgar scores at 1 and 5 min, c-PVL, twin status, and PDA, the CT (OR = 2.8, CI = 1.2–6.8, p = 0.02) and CC (OR = 3.7, CI = 1.3–10.2, p = 0.02) genotypes were associated with increased risk for IVH in the full data and the CT genotype only (OR = 4.0, CI = 1.1–14.8, p = 0.04) was associated with increased risk for IVH in the data after exclusion criteria.

Table 2 Genetic associations comparing premature infants (<32 wk gestation and <1500 g) with and without IVH
Full size table

Infants heterozygous (OR = 4.9, CI = 1.6–14.8, p = 5 × 10−3) or with the A allele (OR = 4.6, CI = 1.6–13.6, p = 2 × 10−3) of the FV Leiden mutation were at increased risk for IVH. After excluding infants with potential confounders, infants heterozygous for the FV Leiden mutation (OR = 14.4, CI = 1.4–147.9, p = 0.02) or with the A allele (OR = 13.3, CI = 1.3–131.6, p = 5 × 10−3) were at increased risk for IVH. Only the association between FV rs6025 and IVH in the full dataset was significant after correction for multiple testing with Bonferroni. After controlling for Apgar scores at 1 and 5 min, c-PVL, twin status, and PDA, heterozygotes for the FV Leiden mutation were still at increased risk for IVH in the full data (OR = 4.0, CI = 1.1–14.9, p = 0.04) and after exclusion criteria (OR = 17.7, 1.3–244.7, p = 0.03).

Genetic associations with grade of IVH.

IL1 β rs1143627 was significant or marginally significant for allele and genotype differences when comparing infants without IVH to those with IVH grades I–II or IVH grades III–IV (Table 3). Infants with the CT genotype (OR = 3.0, CI = 1.1–7.9, p = 0.03) or the C allele (OR = 1.7, CI = 0.99–3.0, p = 0.05) of IL1 β rs1143627 were at increased risk for IVH grades I–II compared to infants with the TT genotype or T allele. Infants with the CC genotype (OR = 3.6, CI = 1.1–11.1, p = 0.03) or the C allele (OR = 2.0, CI = 1.1–3.8, p = 0.02) of IL1 β rs1143627 were at increased risk for IVH grades III–IV compared to infants with the TT genotype or T allele. These associations were not significant after correction for multiple testing with Bonferroni nor did they remain significant after excluding infants for potential confounders (Table 3). After controlling for Apgar scores at 1 and 5 min, c-PVL, twin status, and PDA, the CT (OR = 3.1, 1.1–8.8, p = 0.03) but not CC genotype (p = 0.1) increased risk for IVH grades I–II and the CC (OR = 6.5, 1.3–33.8, p = 0.03) but not CT (p = 0.2) genotype increased risk for IVH grades III–IV in the full data. These results did not remain significant (p > 0.05) in the data after exclusions.

Table 3 Genetic associations comparing premature infants (<32 weeks gestation and <1500 g) without IVH to those with grades I–II and those with grades III–IV
Full size table

When comparing infants with grades I–II IVH to those without IVH, there were allele and genotype frequency differences for FV rs6025 in both the full dataset and the dataset with exclusions (Table 3). Associations in both the full dataset and the dataset with exclusions were significant after correction for multiple testing with Bonferroni. Infants heterozygous (OR = 8.1, CI = 2.5–26.0, p = 4 × 10−4) or with the A allele (OR = 7.3, CI = 2.4–22.5, p = 1 × 10−4) for the FV Leiden mutation were at increased risk for IVH grades I–II. After excluding infants for potential confounders, the association with IVH and both the heterozygous genotype (OR = 28.9, CI = 2.7–311.2, p = 6 × 10−3) and the A allele (OR = 24.5, CI = 2.4–247.2, p = 1 × 10−4) remained significant; however, there was no association in either dataset when comparing IVH grades III–IV to infants without IVH (p > 0.5). However, the power to detect the same effect in the A allele as observed in IVH grades I–II was 70% and therefore lack of association cannot be adequately established. After controlling for Apgar scores at 1 and 5 min, c-PVL, twin status, and PDA, heterozygotes for the FV Leiden mutation were still at increased risk for IVH grades I–II in the full data (OR = 5.5, CI = 1.5–20.0, p = 0.01) and after exclusion criteria (OR = 40.5, 2.0–840.1, p = 0.02).

DISCUSSION

Genetic studies of IVH have largely focused on genes involved in inflammation and infection, because this pathway is strongly implicated in the pathophysiology of perinatal brain injury, supported by animal models and studies in human preterm infants (10). We sought to replicate 10 candidate genes from the inflammation/infection and complement/coagulation pathways for association with IVH. We identified two genes (IL1 β (rs16944) and FV (rs6025)) that associated with the risk for IVH in our cohort; thereby replicating previous studies.

Previously, the IL1β-511 (rs16944) T allele was associated with an increased risk of IVH in 215 VLBW infants compared to the C allele (OR = 3.0, CI = 1.4–6.4, p = 0.003) (10). We validated this result by finding that the IL1 β-31 (rs1143627) C allele was associated with an increased risk of IVH (p = 0.007). This finding was significant or marginally significant for both IVH grades I–II and IVH grades III–IV. The C allele of IL1 β-31 is in strong LD (r2 = 0.96) with the T allele of IL1 β-511. The IL1 β-31 C allele is associated with an increased production of IL1β in vivo (28). There is substantive evidence that IL1β is involved in the pathophysiology of perinatal brain injury. Injection of IL1β causes brain injury in neonatal rats, and increased amniotic and/or cord blood levels of IL1β are observed in infants with PVL and IVH (29–33). The exact mechanisms by which IL1β is involved in IVH and perinatal injury is still not entirely clear; however, our research has identified that the same allele of IL1 β-31 (C allele) associated with increased levels of IL1β is also associated with an increased risk of IVH.

Coagulation factors, specifically FII, FV, and FXIII have been considered as possible candidate genes for IVH because of likely interactions between thrombophilic factors and the pathology of IVH. It is hypothesized that increased fibrinolytic activity and decreased levels of clotting factors may contribute to the severity of intracranial bleeding that can occur in preterm infants (34,35). We found no significant associations with FII or FXIII; however, the FV Leiden mutation (rs6025) and IVH were strongly associated. The heterozygous genotype of rs6025 was previously shown to associate with an increased risk for IVH and to protect against the extension and/or progression to more severe grades of IVH (14,22). Our study supports these findings as the heterozygous genotype of the FV Leiden mutation was significantly associated in infants with IVH grades I–II (p = 5 × 10−3) but not IVH grades III–IV (p = 1.0). However, the lack of association detected in the severe grades of IVH must be interpreted with caution, because of the small sample size in the more severe grades of IVH (n = 8) and therefore lack of power (70%) to detect the same effect seen in IVH grades I–II (OR = 24.5). Moreover, it is to be noted that the effect size seen in the IVH grades I–II groups after stringent exclusions is likely inflated because of the small sample size; therefore, the power to detect an effect in the severe IVH group is likely lower. The FV Leiden variant has about a 5% allele frequency in the Caucasian population and is a glutamine to arginine replacement at amino acid position 506 that results in an increased risk for thrombosis (36). The associations with the FV Leiden mutation were the only results in our study that withstood Bonferroni correction for multiple testing and lends evidence to the hypothesis that the complement coagulation pathway is involved in the risk for the development of IVH.

Although several studies have found associations with the FV Leiden mutation (rs6025) (10,14,15,21,22) there have also been other studies that failed to find association (10,37). This could be due to ethnic heterogeneity, varied study design, or lack of an adequate control group for comparison. A strength of our study was the ability to compare very preterm (<32 wk of gestation) VLBW (<1500 g) infants with IVH to very preterm VLBW infants without IVH, whereas other studies have compared infants with IVH to term infants without IVH and their associations are therefore confounded by GA and birth weight. However, one weakness of our study is a relatively small sample size and therefore associations that were not detected in our analysis may not be indicative of lack of association but rather because of the limited power to detect the effects. For example, for the genes for which no allelic association was detected (IL4, IL6, IL10, TNF, FII, FXIII, ITGB3, and ESR1), the power to detect these associations only reached 80% for effect sizes larger than an OR of 4.

However, this study does replicate two genes [IL1 β (rs16944) and FV (rs6025)] associated with the risk for IVH. IVH is a significant problem for very preterm VLBW infants. Although there is substantial research that has focused on understanding the etiology, mechanisms, and risk factors for IVH, little progress has been made in preventing this serious condition. It is important that continuing research focus on replicating previous findings and discovering new mechanisms and pathways for IVH. Identification of genetic risk factors can provide an opportunity to generate new therapeutic and preventative strategies in an era of personalized medicine.