Significant associations of PAI-1 genetic polymorphisms with osteonecrosis of the femoral head

Osteonecrosis of the femoral head (ONFH) is a devastating bone disease in which patients experience progressive collapse of the femoral head caused by a disturbance in the supply of blood and anomalies in the fibrinolytic system [1, 2]. An increased tendency for intravascular coagulation is proposed as the pathogenetic mechanism responsible for interruption of the osseous blood supply and ONFH, and a significantly higher prevalence of coagulation abnormalities is reported in patients with ONFH [1, 3]. Recent studies have suggested that genetic polymorphisms in factor V, prothrombin, methylenetetrahydrofolate reductase (MTHFR), and plasminogen activator inhibitor-1 (PAI-1) genes leading to intravascular coagulation disorders may be related to ONFH [4‐6].

PAI-1 is a critical factor that regulates coagulation and fibrinolytic systems. Reduced plasma fibrinolytic activity, mainly attributable to increased levels of PAI-1, is associated with ONFH development [7, 8]. Previously, PAI-1 levels were reported to be regulated by a common transcription-altering insertion/deletion single nucleotide polymorphism (SNP; rs1799889) of four or five guanine (4G/5G) nucleotides that is 675 bp upstream of the transcription start site. Homozygous or heterozygous carriage of the 4G allele is associated with higher PAI-1 levels [9]. In myocardiac infarction, subjects who are homozygous for the 4G allele (4G/4G genotype) have plasma PAI-1 concentrations that are approximately 25% higher than those of subjects who are homozygous for the 5G allele (5G/5G genotype) [10]. Glueck et al. first reported that the genotype frequency of 4G/4G was 41% in ONFH patients and 20% in healthy control subjects [8]. Moreover, Ferrari et al. reported a significant increase in the frequency of the 4G/4G genotype in renal transplant patients with ONFH compared to that of controls (60.3% vs. 20.6%) [7]. However, Asano et al. suggested that plasma PAI-1 levels are highest in ONFH patients with the 4G/4G genotype, but that the incidence of ONFH is not related to genotype [11]. To better understand the genetic influences of PAI-1 on ONFH, we selected SNPs in the promoter and 3'UTR that are known to be involved in the regulation of gene expression. To determine whether PAI-1 SNPs, including 4G/5G, are associated with susceptibility to ONFH, genotype and allele frequencies were analyzed. We also investigated whether pathological etiology (idiopathic, alcohol- or steroid-induced) is involved in the association of SNPs and ONFH.

To identify if PAI-1 SNPs were involved in susceptibility to ONFH, three SNPs in the promoter and 3'UTR were selected based on a minor allele frequency (MAF) > 0.05 and Hardy-Weinberg equilibrium (HWE) > 0.05 using a public database http://​www.​ncbi.​nlm.​nih.​gov/​SNP/​. Two SNPs (rs2227631, -844 G/A; rs1799889, -675 4G/5G) in the promoter region and one (rs11178, +10700 C/T) in the 3'UTR of the PAI-1 gene were genotyped in 206 ONFH and 251 control subjects (Table 2).

The p value of each polymorphism was determined for comparison between ONFH patients and controls. The rs2227631, rs1799889, and rs11178 genotypes and/or allele frequencies of the PAI-1 gene were significantly associated with ONFH risk (p = 0.002 - 0.03). These results suggest that the minor alleles of rs2227631 (A), rs1799889 (4G), and rs11178 (C) in the PAI-1 gene contribute to an increase in ONFH risk (Table 3).

Further analysis based on pathological etiology (idiopathic, alcohol- or steroid-induced) showed that the allele frequencies of PAI-1 (rs2227631 and rs11178) between idiopathic ONFH patients and controls were significantly associated (p = 0.007 and 0.021). In addition, the allele frequencies of rs1799889 and rs11178 were significantly associated with the alcohol-induced ONFH subgroup (p = 0.042 and 0.015, respectively). These results suggest that the PAI-1 SNPs were important risk factors in idiopathic (rs2227631, rs11178) and alcohol-induced (rs1799889, rs11178) ONFH (Table 4). In addition, the association of PAI-1 polymorphisms was further analyzed in the ONFH subgroup stratified according to gender. Interestingly, the genotype frequencies of rs2227631, rs1799889, and rs11178 were significantly associated with ONFH in men but not in women (Table 5). These results suggest that the PAI-1 SNPs might be important genetic factors in ONFH susceptibility in men.

We analyzed linkage LDs among the PAI-1 SNPs and did not observe complete LD for the PAI-1 gene. LD between SNPs was estimated by calculating the D' values (Table 6). We also calculated the haplotype frequencies of rs2227631, rs1799889, and rs11178 in ONFH patients and controls. The distribution of haplotype A-4G-C was significantly different between the ONFH patients and controls (p < 0.001) (Table 7). These results suggest that these SNPs should be further analyzed in future studies regarding PAI-1 functions and ONFH.

We determined the contributions of PAI-1 gene SNPs to ONFH. We identified for the first time a significant association between SNPs of the PAI-1 gene (rs2227631; -844 G/A, rs1799889; -675 4G/5G, rs11178; +10700 C/T) and ONFH in Koreans. The A, 4G, and C alleles considerably increased disease risk (Table 3).

PAI-1 is a fast-acting inhibitor of fibrinolysis, and increased plasma levels are associated with increased incidence of thrombophilia [14] and osteonecrosis [8, 11, 15, 16]. High levels of PAI-1, induced by -675 4G/5G SNP in the PAI-1 promoter, lead to suppression of fibrinolysis through inhibition of plasminogen activator and promotion of thrombosis. The resulting increase in intraosseous venous pressure which restricts flow to the femoral head may culminate in osteonecrosis [8, 16]. The PAI-1 gene is reported to be polymorphic, especially in rs1799889 (-675 4G/5G) of the promoter region. The possible association between rs1799889 and osteonecrosis risk has been studied, leading to controversial results. Glueck et al. and Ferrari et al. reported that the 4G allele is a major predisposing factor in ONFH patients [7, 8], but the findings of Asano et al. were contradictory to that conclusion [11]. Glueck et al. showed that, in Americans, twice as many patients as healthy control subjects (41% vs. 20%) were homozygous for 4G/4G, and 19% of patients and 36% of the control subjects had the 5G/5G genotype (p = 0.001) [7]. Furthermore, Ferrari P. et al. observed that in 228 glucocorticoid-treated renal transplant patient in Switzerland, the prevalence of ONFH according to genotype was 1.8% for 5G/5G, 7.7% for 4G/5G, and 30.3% for 4G/4G (p < 0.001 vs. 4G/5G and 5G/5G); the prevalence of ONFH according to genotype in subjects with persistent hyperparathyroidism was 4.2% for 5G/5G, 15.2% for 4G/5G, and 55.5% for 4G/4G (p < 0.003 vs. 4G/5G and p < 0.001 vs. 5G/5G) [8]. Asano et al. studied 31 Japanese patients with postrenal transplant ONFH and found four patients with 5G/5G, 11 with 4G/5G, and 16 with 4G/4G. However, analysis revealed no significant differences in the incidence of ONFH among these patients (p = 0.49) [11].

In addition to rs1799889, another SNP in the promoter region of PAI-gene, rs2227631 (-844 G/A), is potentially implicated in PAI-1 gene regulation. To date, a significant association has been reported between this SNP and coronary heart disease (CHD) in nonsmokers, with patients having a higher frequency of rs2227631 A allele. In addition to rs2227631 and rs1799889, rs11178 (+10700 C/T) also significantly associates with increased CHD risk in nonsmokers. The correlation between rs11178 and CHD in nonsmokers might be attributable to strong linkage LD of this SNP with rs2227631 and rs1799889 [17]. For ONFH, the results of our study confirmed an association between rs1799889 and disease risk in a Korean population. Moreover, for the first time, we showed that rs2227631 and rs11178 are essential SNPs involved in the regulation of PAI-1 gene expression in ONFH (Table 3).

Several transcription factor binding sites for PAI-1 are found in the 5' and 3' UTR regions, and the transcriptional regulation of the gene is extremely complex [18]. Several studies have shown that SNPs within the 5'UTR lead to differences in PAI-1 expression between individuals, and this could influence the etiology of a variety of pathological conditions with which PAI-1 is associated such as cancer, rheumatoid arthritis and stroke [19‐21]. We classified ONFH into two or three major subgroups based on etiology and gender: idiopathic, alcohol-induced, steroid-induced groups, and men and women. We found that the risk effects of rs2227631 and rs11178 in the idiopathic subgroup, rs1799889 and rs11178 in the alcohol-induced subgroup, and all three SNPs in men were significantly associated with ONFH. However, no association was seen in the steroid-induced group or in women. Glucocorticoid has been reported to increase PAI-1 activity and is a potential risk factor for ONFH development [7, 22]. However, we had limited data on SNPs of PAI-1 in steroid-induced ONFH. The incidence of ONFH is relatively low and gender-biased. We examined 206 cases, of which 159 (77%) were men. Moreover, epidemiologic analysis showed that the incidence of steroid-induced ONFH in overall ONFH was low (17.5%). In Korean studies, the proportion of steroid-induced ONFH was previously shown to be small (range, 12.6% to 15.4%) [23, 24]. We found that, in Koreans, steroid-induced ONFH is more rare than idiopathic or alcohol-induced ONFH. Thus, we found no significant association between PAI-1 SNPs and steroid-induced ONFH because of the small sample size. In addition, differences according to subgroup could not be clearly distinguished. Therefore, the association analysis strategy of subgrouping according to gender or etiology has limitations. However, the results of the steroid-induced group showed a similar tendency as those of the other groups, demonstrating that patients more often have risk alleles (A, 4G, and C) than do control subjects, although this finding was not significant (Table 4 and 5). Thus, we suggest that PAI-1 SNPs are involved in ONFH risk in Korean patients. To firmly establish the relationship between the PAI-1 SNPs and steroid-induced ONFH, further study with larger sample sizes is required.

The pathophysiology of ONFH is not well known, although a number of polymorphisms in candidate genes (HIF-1, VEGF, eNOS, IL23R, SREBP-2, ANXA6) were recently identified in an attempt to determine the genetic factors involved in ONFH pathogenesis in a Korean population [24‐29]. Some studies have suggested that genetic polymorphisms leading to thrombosis (factor V, prothrombin, MTHFR) may be related to ONFH [4, 6, 16, 30, 31]. Intravascular coagulopathy including thrombotic and fibrinolytic abnormalities may play an etiologic role in the disease, and studies have investigated the association between ONFH and genes involved in the coagulation and fibrinolytic system [7, 11, 32].

We found that SNPs of the PAI-1 gene, which is involved in coagulation, were significantly correlated with ONFH. These data suggest that PAI-1, along with already reported candidate genes, may be useful genetic markers to identify high-risk individuals in Korea. The results of this association study suggest that PAI-1 gene polymorphisms may be important genetic factors in ONFH susceptibility in a Korean population. To further substantiate this hypothesis, functional studies of PAI-1 regulation are required. The polymorphisms analyzed in this study may contribute to further studies on the function of PAI-1 and the development of ONFH.