Adolescent idiopathic scoliosis (AIS) is a complex 3-dimensional deformity of the spine that occurs most commonly in girls during the peripubertal period [1]. Despite decades of research, its specific cause has not been determined [2]. To date, many factors have been proposed to be involved in its etiology, such as genetic factor, nervous system, skeletal growth, hormone, and metabolic dysfunction [3]. It has been strongly suggested that genetic factor plays an important role in the development and progression of AIS [4, 5]. Many AIS susceptibility loci have been identified through genome-wide linkage analyses and genome-wide association study [6–8]. Several promising candidate genes for AIS susceptibility have been reported, including the genes for estrogen receptor [9], melatonin 1B receptor [10], matrilin-1 [11] and ladybird homeobox 1 [6]. However, to our knowledge, only few genes can be replicated for their association in different ethnic populations [12–14].

Recently, Sharma et al. [15] performed a genome-wide association study (GWAS) based on transmission-disequilibrium tests in 419 AIS families, which identified associated single nucleotide polymorphisms (SNPs) in the proximity of the cell adhesion molecule with homology to the L1CAM (CHL1) gene. They genotyped additional SNPs in chromosome 3p26.3 and tested the association in two follow-up case-control cohorts. They obtained the strongest association in rs10510181 with all three cohorts combined (P = 2.58 × 10^-8). CHL1 encodes an axon guidance protein related to Robo3 [16], mutations of which could lead to horizontal gaze palsy with progressive scoliosis (HGPPS), a rare disease marked by severe scoliosis [17]. To confirm the association between rs10510181 and AIS, and to explore whether the CHL1 gene is associated with the occurrence of AIS, we conducted the present genetic association study in a Han Chinese population.

In stage I, the genotype frequencies for rs2055314 and rs2272522 were significantly different between case and control groups (P <0.05), and it was marginally different for rs10510181 (P = 0.08). Similarly, the allelic frequencies for rs2055314, rs2272522, and rs10510181 showed significant differences between case and control groups (P <0.05) (Table 3). Otherwise, the genotype and allelic frequencies for the other two SNPs (rs331894 and rs2272524) were comparable between case and control groups (P >0.05) (Table 3).

In stage II, the sample size was enlarged for rs2055314, rs2272522, and rs10510181. The pooled data were shown in Table 4. For rs10510181, the genotype frequencies were GG 28.8%, GA 46.2%, and AA 25.0% in AIS patients and GG 29.8%, GA 48.8%, and AA 21.4% in controls. No significant difference was found in genotype distribution between cases and controls (P = 0.39). In addition, there was no obvious difference in allelic distribution between the two groups (G: 51.9% in cases vs. 54.2% in controls, A: 48.1% in cases vs. 45.8% in controls, P = 0.30). Similarly, the genotype and allelic distributions for rs2055314 and rs2272522 were comparable between case and control groups (Table 4).

It is generally recognized that abnormal growth is associated with the development and progression of the scoliotic curves. After the onset of puberty, significantly longer corrected height, arm span, and various body segments were found. Furthermore, BMI of AIS was significantly lower than that of controls [19]. In the present study, we also found that the body weight was significantly lower in AlS patients than in normal controls.

GWAS is a powerful method for the detection of genetic contributions to polygenic diseases, and it has been used to identify genetic predisposition in AIS, which has been recently regarded as one of the most common complex genetic disorders of the musculoskeletal system [6, 15]. However, this method may produce spurious associations, as the findings could be confounded by systematic biases in comparison groups, such as population stratification and effect heterogeneity [20, 21]. Therefore, replicating the associations in different ethnic groups, as well as studies with large sample sizes, is important to confirm the results of GWAS [22]. Recently, Sharma et al. [15] reported the association between SNP rs10510181, in the proximity of the CHL1 gene, and AIS by performing GWAS in 419 Caucasian families with AIS. The present study was conducted to validate the association of SNP rs10510181 with AIS. The results showed that rs10510181 was not associated with AIS in a Han Chinese population. Furthermore, 4 SNPs covering the entire region of the CHL1 gene were not associated with the occurrence of AIS either, indicating that the CHL1 gene might not be associated with AIS.

The previously reported association of rs10510181 with AIS predisposition in Caucasians was not replicated in our study. Several factors could lead to this lack of replication, which is not rare in genetic association studies of common diseases. First, the allele and genotype frequencies of rs10510181 could vary with ethnicity. Because of differences in variant frequency and underlying linkage disequilibrium (LD) structure between ancestral populations, the relationship between rs10510181 and a pathogenic variant may be influenced. Second, the inclusion criteria were different between the two studies. The subjects included in Sharma's study [15] had a curvature larger than 10 degrees, as well as axial rotation toward the side of the deviation. In the present study, patients with a Cobb angle greater than 20 degrees were included.

The CHL1 gene encodes an axon guidance protein that plays an important role in the guidance of thalamocortical axons [23], as well as the proliferation and differentiation of neural progenitor cells [24]. Mutations in the CHL1 gene may alter the axonal guidance and brain anatomy in mice [25]. Moreover, mutations in the Robo3 gene, which encodes an axon guidance protein related to CHL1, could cause HGPPS [17]. Among the hypotheses of AIS etiology, abnormalities in the central nervous system have long been thought to play a key role [26, 27]. Disturbance of the central nervous system may impair somatosensory function and motor adaptation, which lead to the asymmetry of neuromuscular condition causing AIS [27, 28]. Thus, it seems quite plausible that CHL1 could be involved in the etiology of scoliosis through disturbance of the central nervous system. Although the present study did not find any association between the four SNPs spanning the whole region of the CHL1 gene and AIS, we cannot entirely exclude the role of this gene in the development of AIS because the CHL1 gene spans more than 100 kb and contains thousands of SNPs.

The current study did not find an association between the SNPs in and around the CHL1 gene and AIS predisposition. Considering the power calculation and sample size of the present study, we conclude that there is no association between the CHL1 gene and AIS predisposition in a Han Chinese population. However, the role of the CHL1 gene in other populations cannot be excluded, and replication studies in other ethnic groups are needed to understand the overall significance of the CHL1 gene in AIS.