Adolescent idiopathic scoliosis (AIS) is characterized by a structural lateral curvature of the spine with a rotary component deviation in otherwise healthy individuals. When defined as a Cobb's angle of at least 10°, AIS is the most common pediatric spinal deformity, with reports of incidences ranging from 0.5% to 10% worldwide.1 A report published in Korea revealed a scoliosis prevalence of 3.26% among 1134890 school children, higher than those of other countries.2 According to a notification released by the Korea Health Insurance Review and Assessment Service, the number of scoliosis patients and expenditures related to its treatment increased by 12.2% and 40.3% over the past five years, respectively.3

AIS is usually noticed while bathing or during physical activity by a child's parents. Sometimes, early detection of AIS occurs during periodic school screening upon discovery of a waist or rib hump.4 The Adams forward bending test is helpful in detecting a thoracic rib hump or lateralized lifting of the buttock, but exact diagnosis is difficult. A hump and/or the Adams forward bending test are sensitive but not specific. Upon suspicion of AIS, radiographs can be used to confirm the diagnosis.

Clinical features and radiographic findings suggestive of minimal or mild idiopathic scoliosis do not adequately predict the future progression of the deformity. Even mild idiopathic scoliosis has been found to progress, so that regular medical assessment over a period of years has become necessary to detect and assess progression of the deformity.1,5,6 The current standards of care suggest that children and teens with AIS should be monitored until they cease growing, usually with serial radiographs, for curve progression. If serial radiographs confirm progression of Cobb's angle (20<Cobb's angle<45), a brace should be administered as the next step in the primary treatment thereof. Approximately 8% to 10% of all patients with adolescent idiopathic scoliosis are prescribed a brace at some point during their treatment. If the spinal curve continues to progress (Cobb >45), spinal fusion is often recommended.

Brace treatment comprising wearing of a brace for a minimum of 15 hours per day is considered effective. However, practically, it is difficult to ensure that adolescents wear a brace 15 hours a day.7 In order to overcome non compliance, electronic monitoring of these individuals is now followed in certain places.8 Due to difficulties faced during brace treatment, a prognostic method is direly needed to assess whether the deformity will progress or not. Such methods could help clinicians plan the treatment schedule of AIS on an individual and more effective manner.

Factors known to predict curve progression include maturity, curve size, and position of the curve apex. Progression equations have also been developed to quantify the risk of progression. Peterson and Nachemson9 used a method combining Risser sign, level of apex, presence of trunk imbalance, and chronological age to the aforementioned list of factors to predict curve progression. Although this method might be of some help to clinicians and patients, some patients do not follow this equation, and a short period brace or exercise only often improve Cobb's angle. In some cases, progression of deformity occurs rapidly, and physicians and their patients may miss the proper timing of surgery (Fig. 1). To solve this problem, a more accurate diagnostic tool is needed to determine whether or not deterioration will occur.

A prognostic test with a sufficient negative predictive value (NPV) can alter the monitoring schedule of a patient. Moreover, such test can reveal patients whose Cobb's angle will not undergo serious progression. Accordingly, low risk patients with a definite NPV might be spared the long-term negative effects of repeated exposure to radiation, unnecessary treatments, emotional stress, cost of direct care (physician's fee, hospital charge) and indirect cost of care, such as time away from a patient's studies and hospital visit expenses.10,11 In those who might undergo serious progression, however, it is important that AIS does not get worse. An effective prognostic test would allow physicians to develop new clinical protocols for managing high risk AIS patients.

Genetic associations in AIS have been described for decades.11 Recent etiologic studies revealed that AIS is a complex genetic disorder that results from the interaction of multiple gene loci and the environment.12,13 Linkage studies of these families revealed multiple potential genetic loci, chromosomes 6, 9, 16 and 17, that may predispose individuals to the condition.14 Several genes such as bone morphogenetic protein 4, interleukin-6, leptin, matrix metalloproteinase-3, melatonin 1B receptor (MTNR1B), etc., have been shown to be associated with a combinatorial effect.15 Recently, Ward insisted that by using 53 gene marker SNPs known to be associated with AIS, a NPV greater than 99% could be obtained.5

Ward, et al.5 also reported that these 53 SNPs were closely associated with AIS severity. The odds ratios for 5 of these SNPs (rs2449539, rs1437480, rs448013, rs10493083, rs16945692) with severe scoliosis ranged from 0.35 to 1.28, as shown in Table 1. Based on their study, a diagnostic kit was developed that could predict the progression of Cobb's angle and was sold commercially. Subsequently, several studies were carried out in Oriental countries concerning AIS candidate genes and SNP markers thereof.16 Sharma, et al.6 genotyped the SNPs of an additional chromosome, 3p26.3, and tested replication in two follow-up case-control cohorts, obtaining stronger results when all three cohorts were combined (rs10510181 odds ratio of 1.49). In addition, other significant associations in their genome-wide association study (GWAS) were discovered for SNPs (rs2222973) in the Down syndrome cell adhesion molecule (DSCAM) gene, which encodes for an axon guidance protein in the same structural class.

Ethnically, Koreans are much more similar to the Japanese than Caucasians. Therefore, from a database of Japanese SNPs, we selected SNPs for study with a minor allele frequency of more than 5%. Other controversial SNPs in Asian populations were also included for confirmation. Common SNPs in the genes of matrilin 1 (MATN1) and insulin-like growth factor 1 (IGF1) are reported to be associated with AIS in Chinese, but studies to uncover candidate genes or marker SNPs for AIS in Koreans have yet to be performed. Furthermore, only one paper has reported on a relationship between AIS and osteoporosis in Korea,17 and the lack of data has limited the ability of physicians in Korea to personalize treatment of AIS, unlike Caucasian populations. The objective of this study was to reveal AIS-related genes and their relevance in Koreans, exploring the potential use of gene analysis as a diagnostic tool for personalized treatment of AIS therein.

Prior to conducting our study, we received Institutional Review Board and Clinical Research Ethics Committee approval of our study protocols. A total of 68 Korean adolescents (8-18 years of age), newly diagnosed with AIS at the authors' institution, in addition to 35 age- and sex-matched, healthy adolescents were enrolled in our study after applying the exclusion criteria detailed below. Some of the healthy controls comprised students from the school where the authors conducted the study. The following candidates were excluded: those with a history of congenital anomalies, neuromuscular diseases, endocrine disorders, skeletal dysplasia or connective tissue disorders, or mental retardation or mental illness, as well as psychiatric patients on medication affecting bone metabolism. From the 68 patients in the scoliosis group, 33 patients with a Cobb's angle between 10-40 degrees allowing for non-surgical treatment were subdivided into a lower risk group, while 35 patients whose Cobb's angle was more than 40 degrees were divided into the high risk group as they required surgery.

In this study, the following three groups were compared: 35 subjects in the control group, 33 patients in the lower risk group and 35 patients in the high risk group (Table 3). The mean Cobb's angle for lower risk group patients was 25.8° (SD=7.6°) and 58.8° (SD=21.0°) for high risk group patients. Age, sex, height and weight had no statistical significant difference among the three groups. BMI was 20.1 (SD=3.5) in the control group, 19.3 (SD=2.0) in the low risk group and 19.1 (SD=2.8) in the high risk group, and this mathematical variation was not statistically significant bone formation marker alkaline phosphatase (ALP) levels were 174.7 IU (SD=98.1) in the control group, 77.0 IU (SD=24.5) in the low risk group and 108.5 (SD=16.4) the in high risk group. Again, despite variation, its statistical value was questionable. Also, differences in CTX, representative of the degree of bone destruction, and osteocalcin, a bone formation marker, were found to be statistically inconclusive. Mean lumbar spine and femur neck bone mineral density (FNBMD) in AIS patients tended to be lower than that in controls, yet the difference failed to reach statistical significance. Vitamin 25(OH)D3 levels were 19.0 ng/mL (SD=9.0) in the high risk group, 23.1 ng/mL (SD=8.2) in the lower risk group and 26.2 ng/mL (SD=8.2) in the control group, and vitamin D levels were shown to be significantly lower in the high risk group than those in the lower risk and control groups (p=0.019).

Genotypes were shown to be in Hardy-Weinberg equilibrium. The distribution of the genotypes and alleles of the cases and controls for the ten promoter SNPs are presented in Table 4. No variations of statistical importance were observed in genotype frequencies and allele frequencies of the eight promoter SNPs among the lower risk group, high risk group and controls in the Korean population, when the SNPs were studied independently (Table 4).

The lysosomal-associated transmembrane protein 4 beta (LAPTM4B) polymorphism (rs2449539) was found to be significantly different among the controls and lower risk and high risk groups. In comparison of the three groups individually, the frequency of TT genotype was most common in the high risk group, followed by the low risk group and control group, respectively (p=0.014). In contrast, the frequency of the TC genotype was most common in the control group, followed by the low risk group and high risk group, in that order (p=0.003). In comparison between two groups (controls versus AIS patient group), the frequency of TT genotype was most common in the AIS patients group (p=0.008). Similarly, the frequency of TC genotype was most common in the control group as compared to the AIS patients group (p=0.004). These results are summarized in Fig. 2.

Additionally, IGF1 polymorphism (rs5742612) was also found to be significantly different among the controls and the lower risk and high risk groups. In comparison of the three groups individually, the frequency of GG genotype was most common in the control group, followed by the low risk group and high risk group, in that order (p=0.010). Between the two groups (controls versus AIS patient group), the frequency of GG genotype was most common in the control group (p=0.006). However, there was no significant difference in allele frequencies for one SNP (rs5742612) (p=0.059).

As shown in Table 4, two SNPs at rs2449539 (p=0.003) and rs5742612 (p=0.006) were found to exhibit significant associations with AIS in Koreans (p<0.01). The other 8 SNPs demonstrated no such significant associations. There was a significant difference in allele frequencies for one SNP (rs2449539) (p=0.002).

In Korea, AIS patients make about 100000 visits to physicians annually, thereby incurring an expenditure of millions of dollars or even more, and this expenditure has increased steeply over the past five years (Fig. 1). Clinicians and patients need to be aware of the risk of curve progression as they make treatment decisions. Peterson and Nachemson's9 progression equation might be of some help to clinicians and patients who want to base their treatment on the risk of progression. But some patients do not follow this equation, and to solve this problem, a more accurate diagnosis and diagnostic tool that can predict whether or not deterioration will occur is needed.

Though AIS is thought to be multifactorial disorder, differing degrees of interaction between multiple factors depend on linear and summation causality. Although AIS has been linked to multifactorial causes, genetic factors are considered the most important cause.

There are many reports about the supportive role of genetic inheritance in the development of AIS. A meta-analysis of different twin studies has been performed and found a concordance rate for AIS of 73% in monozygous twins versus 36% in dizygous twins.20 With recent advances of genetic analysis techniques, many different gene loci linked to AIS is reported by family-linkage studies. Miller, et al.21 reported that gene loci in the primary regions on chromosomes 6, 9, 16, and 17 and in the secondary regions on chromosomes 1, 3, 5, 7, 8, 11, 12, and 19, after scanning 202 families and 1198 individuals, were linked to AIS in 2005. Yet, other studies failed to achieve the same results, and instead, found other candidate genes. Therefore, it is hard to establish whether the etiology of AIS is due to single gene or not.22,23 Recently, many genetic association studies with the candidate gene approach have been performed with the help of information gathered from the international HapMap project. In previous studies, the selection of candidate gene selection followed to prior knowledge of the expression of observed phenotypes in AIS and different SNPs studied with the case-only and case-control analysis. More recently, a more comprehensive explanation on the possible genes involved in the etiopathogenesis of AIS has been presented by GWASs of all SNPs in the genome. In the recent report of Ward, et al.,5 the comparison of a GWAS of 1.8 million genetic markers was performed between 1200 AIS patients and 1500 controls. Their study found 202 markers associated to curve progression, and after further refinement, 30 markers were reported as the most useful prognostic markers for curve progression. After verifying the results of this study, they launched the development of commercially available kit with a NPV greater than 99% for AIS-PT. However, there was an important limitation to their work; genotype frequencies for several of the SNP markers used in the AIS-PT kit vary by race. Accordingly, their AIS-PT kit has been validated for whites only. Thus, Korean specific SNP tests should be developed to personalize the treatment of AIS therein.

In this study, the authors selected a control group that was age- and sex-matched, so that age and sex did not differ among the three groups compared. BMI was 20.1 (SD=3.5) in the control group, 19.3 (SD=2.0) in the lower risk group, 19.1 (SD=2.8) in the high risk group and arithmetically low in the AIS patient group (lower+high risk group), but this difference was not statistically significant. Several authors reported that scoliotic girls had lower mean weight and a lower bone mineral status compared to those in the control population.24,25 Leptin along with soluble leptin receptor was shown to play an important role in the regulation of bone and energy metabolism in children. Liu, et al.25 insisted that leptin and soluble leptin receptor were abnormal and associated with deranged growth and anthropometric phenotypes in AIS girls. According to Liu's study, abnormality of leptin receptor leads to lower BMI, and osteoporosis is accompanied by the possibility of scoliosis. Although our study did not show any variation of significance, increasing the sample size of the study population may help in establishing any correlation that may exist between decreased BMI and AIS.

In the present study, vitamin 25(OH)D3 level was 26.2 (SD=8.2) in the control group, 23.1 (SD=8.2) in the lower risk group and 19.0 (SD=9.0) in the high risk group, the difference of which was statistically significant in the surgically treated, high risk group (p=0.019). The BMDs of the spine were 0.81 g/cm² (SD=0.17) in the control group, 0.78 (SD=0.14) in the lower risk group and 0.76 (SD=0.09) in the high risk group, a difference which tended to be lower in the high risk group, but was not statistically significant (p=0.093). Generalized low bone mass and osteopenia in the axial and peripheral skeleton have been described in patients with AIS.26,27 The precise mechanism of bone loss in these patients is unclear. Recently, many studies have reported that gene polymorphism is related to osteoporosis, but such results vary and are difficult to interpret.27,28 Once again by increasing the size of the study population, the relationship between osteoporosis and AIS may be established.

In a previous GWAS, Sharma, et al.6 insisted that chromosome 3p26.3 SNPs in the proximity of the cell adhesion molecule with homology to the L1CAM (CHL1) gene (rs10510181) as well as SNPs in the DSCAM gene (rs222973) that encodes for an axon guidance protein were strongly associated with AIS. The prevalence of the rs10510181 genotype was nearly the same as that in the control, lower risk and high risk groups of the present study. The prevalences of the rs2222973 genotype were 42.4% (TT), 42.4% (TC) and 15.22% (CC). In particular, lower risk AIS patients more frequently exhibited the TC genotype, but this was not significant (p=0.057, 0.078). By studying SNPs, in particular rs2222973, in a larger population in Korea, we may be able to develop SNPs as race-specific markers of AIS in Koreans.

Among the 53 SNPs reported by Ward, et al. we studied the following five SNPs: rs2449539, rs1437480, rs448013, rs10493083 and rs16945692. Except for the SNP rs2449539 that encodes the LAPTM4B gene, genotype and allele frequency between the three groups did not differ (p-values from 0.198 to 0.914). LAPTM4B is a protein that in humans is encoded by the LAPTM4B gene encoded on chromosome 8q22, which has been extensively studied in cancer research.29 We found that allele T of rs2449539 involved a significant predisposition to AIS (p=0.002), and individuals with the TT genotype were at a higher risk for AIS, in comparison to TC and CC (p=0.08, 0.04). In this study, we identified that rs2449539 polymorphism in the LAPTM4B gene is associated with both susceptibility and disease progression in AIS. Accordingly, we concluded that the LAPTM4B gene may be important to disease predisposition and progression. Additionally, we also reconfirmed previous reports of ethnic differences in the association between AIS and rs2449539 polymorphism of the LAPTM4B gene. No statistically significant differences were found in the genotype frequencies of the remaining 4 SNPs between AIS patients and controls in the Korean population when the SNPs were studied independently. Similarly, the allele frequencies were comparable between cases and controls (Table 4). Thus, rs2449539 polymorphism of the LAPTM4B gene may be important to the complex genetic etiopathogenesis of AIS.

Comparing Ward's study and our study, there are lot of genetic differences between Caucasians and Asians.5 Two Chinese papers reported the associations of rs1149048 near MATN1, rs10488682 near MTNR1B and rs4753426 near insulin-IGF1 with AIS in Chinese.30,31 However, these associations were not shown to exist in a Japanese population.19 With exception of rs5742612 SNP upstream IGF1, genotype and allele frequency of two SNPs between the three groups did not differ (p from 0.145 to 0.999). We found that allele A of rs5742612 was not significantly related with a predisposition to AIS (p=0.059), but individuals with genotype GG were shown to be at a lower risk of AIS in comparison to AA and AG (p=0.006). In our study, the SNPs of rs1149048 and rs4753426 exhibited no association with AIS predisposition in Koreans, nor did we find any association for these two SNPs with curve severity. However, the rs5742612 SNP in the upstream IGF1 gene was shown to be closely related with AIS in Koreans. Our result fell somewhere between those for Chinese and Japanese populations.

There were some limitations to this study. The number of samples tested was relatively small, which diminishes the statistical power of the study and the possibility of detecting correlations. Further studies with a larger patient population are recommended. The association with other factors, such as the markers of bone metabolism and other candidate genes, should be tested.

In conclusion, this study provides evidence that the SNP rs2449539 in LAPTM4B and rs5742612 in upstream IGF1 genes were associated with both susceptibility to and curve severity in AIS. Our results suggest that LAPTM4B and IGF1 may both be important to AIS predisposition and progression. Nevertheless, expanding our database by increasing the study size so that more SNPs or candidate genes can be studied in the Korean population would prove immensely beneficial in developing a SNP marker for AIS specifically designed for the Korean population.