Figures
Abstract
Adolescent idiopathic scoliosis (AIS) is a clinically significant disorder with high heritability that affects 2–4% of the population. Genome-wide association studies have identified LBX1 as a strong susceptibility locus for AIS in Asian and Caucasian populations. Here we further dissect the genetic association with AIS in a Caucasian population. To identify genetic markers associated with AIS we employed a genome-wide association study (GWAS) design comparing 620 female Caucasian patients who developed idiopathic scoliosis during adolescence with 1,287 ethnically matched females who had normal spinal curves by skeletal maturity. The genomic region around LBX1 was imputed and haplotypes investigated for genetic signals under different inheritance models. The strongest signal was identified upstream of LBX1 (rs11190878, Ptrend = 4.18×10-9, OR = 0.63[0.54–0.74]). None of the remaining SNPs pass the genome-wide significance threshold. We found rs11190870, downstream of LBX1 and previously associated with AIS in Asian populations, to be in modest linkage disequilibrium (LD) with rs11190878 (r2 = 0.40, D' = 0.81). Haplotype analysis shows that rs11190870 and rs11190878 track a single risk factor that resides on the ancestral haplotype and is shared across ethnic groups. We identify six haplotypes at the LBX1 locus including two strongly associated haplotypes; a recessive risk haplotype (TTA, Controlfreq = 0.52, P = 1.25×10-9, OR = 1.56), and a co-dominant protective haplotype (CCG, Controlfreq = 0.28, P = 2.75×10-7, OR = 0.65). Together the association signals from LBX1 explain 1.4% of phenotypic variance. Our results identify two clinically relevant haplotypes in the LBX1-region with opposite effects on AIS risk. The study demonstrates the utility of haplotypes over un-phased SNPs for individualized risk assessment by more strongly delineating individuals at risk for AIS without compromising the effect size.
Citation: Chettier R, Nelson L, Ogilvie JW, Albertsen HM, Ward K (2015) Haplotypes at LBX1 Have Distinct Inheritance Patterns with Opposite Effects in Adolescent Idiopathic Scoliosis. PLoS ONE 10(2): e0117708. https://doi.org/10.1371/journal.pone.0117708
Academic Editor: Shenying Fang, University of Texas, UNITED STATES
Received: September 25, 2014; Accepted: December 30, 2014; Published: February 12, 2015
Copyright: © 2015 Chettier et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This study was funded by Affiliated Genetics, Inc. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Juneau Biosciences, LLC provided support in the form of salaries for authors HMA and KW, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section.
Competing interests: This study was funded by Affiliated Genetics, Inc., the employer of Rakesh Chettier, Lesa Nelson and Kenneth Ward. Rakesh Chettier, Lesa Nelson and Kenneth Ward have direct financial interest in Affiliated Genetics. A US provisional patent application has been filed by Affiliated Genetics, Inc. that include the results and inventions reported in the manuscript. Title: Method of Treating Adolescent Idiopathic Scoliosis Using Genetic Markers. Serial No.: 62/015/695 Filing date: June 23, 2014. Hans M. Albertsen and Kenneth Ward are employed by Juneau Biosciences, LLC. There are no further patents, products in development or marketed products to declare. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.
Introduction
Adolescent idiopathic scoliosis (AIS) is a structural, lateral curvature of the spine greater than 10 degrees, without a known cause, diagnosed between age 10 and skeletal maturity. It affects 2–4% of the teenage population. Although scoliosis can be a primary or secondary feature in several dozen syndromes and can arise secondary to a variety of neuromuscular conditions, idiopathic scoliosis arising during puberty accounts for 80% of pediatric scoliosis. About 10–15% of youth diagnosed with AIS have progressive disease, and 2–4% progress to severe scoliosis [1]. Girls are much more likely than boys to have their spines progress to a severe degree [2]. AIS spinal curve is most likely to worsen during the adolescent growth spurt, sometimes leading to cosmetic deformity, back pain, psychological and social impairments, and respiratory and pulmonary limitations [3–5]. Usually an adolescent is first suspected of having AIS either during pediatric physical examination or during school screening [6].
The genetic component of AIS is both clear and significant as previously described [7–9], with heritability reported between 0.38 [10] and 0.87 [11]. Several genome-wide association studies (GWAS) have established a strong link to LBX1 (Ladybird homeobox 1) [12–15], and a recent meta-study has confirmed LBX1 as a major susceptibility locus for AIS in both Asian and Caucasian populations [16]. LBX1 is a homeobox transcription factor. Its involvement in spinal cord differentiation and patterning, and somatosensory signal transduction make it a strong biological candidate for AIS [17, 18]. In addition to LBX1 three other loci; GPR126, SOX9 and CHL1, are putatively associated with AIS [14, 19]. The primary objective of the present study is to use GWAS to further resolve the genetic components of LBX1 in AIS and to enhance individualized risk assessment in a Caucasian population.
Subjects and Methods
Study Subjects
Quorum Review IRB (Seattle, WA 98101) reviewed and approved the study (Quorum file # 23147/1). Adult subjects and controls provided written informed consent in accordance with study protocols. For minors, both an assent form (signed by the child) and a parental consent form were provided. Patients diagnosed with scoliosis were recruited with assistance from collaborating spine surgeons across the United States and control subjects were recruited from the general population. All data was de-identified.
Inclusion Criteria
Female subjects were included if they had been diagnosed with idiopathic scoliosis between the ages of 10 and 13 and had reached skeletal maturity. Skeletal maturity was defined by a Risser score of 5 or by chronological age (≥16 years or 2 years post menarche). Risser scores quantify the ossification of the iliac apophysis on an anterior-posterior radiograph of the pelvis) [20]. Idiopathic scoliosis has been historically classified as “juvenile” if diagnosed between the ages of 3 and 10 and as “adolescent” if diagnosed from age 10 up until skeletal maturity [21]. Patients with congenital, juvenile, or adult-onset scoliosis, and patients diagnosed with syndromic scoliosis, spine trauma, myopathy, cerebral palsy, or other neurologic condition, were excluded.
Phenotyping and Medical Record Review
All clinical data were collected and analyzed prior to genotyping. Medical records and spinal radiographs of each patient were blinded and reviewed by an orthopedic surgeon (JWO) who assessed Cobb angle measurements, Risser scores, and Lenke classification to provide a diagnosis [22]. Progression to a severe phenotype (curve) was defined according to usual clinical criteria; progression to a >40° curve in an adolescent, or progression to a >50° curve in an adult [23]. Mild and moderate AIS phenotypes were defined by Cobb angles of 10–24° and 25–40° respectively.
DNA Extraction
DNA was obtained from saliva samples, collected using the Oragene DNA Self-Collection Kit (OG-300 Tube Format) (DNA Genotek; Ottawa, Ontario, Canada). DNA was extracted using an automated instrument, AutoPure LS (Qiagen; Valencia, CA), and manufacturer's reagents and protocols. DNA quality was evaluated by calculating absorbance ratio OD260/OD280, and DNA quantification was measured using Picogreen reagent (Invitrogen; Carlsbad, CA).
Genotyping
Genotyping was performed on AIS cases and controls using the Affymetrix HuSNP 6.0 Microarray, with 906,600 SNPs (Affymetrix; Santa Clara, CA). Microarray processing was completed per manufacturer’s protocols. Genotypes were extracted from the scanned images using the Birdseed version 2 algorithm of the Affymetrix Power Tools package (v 1.10.2).
Sample Quality Control
Samples were excluded from the analysis if they failed any of the following quality thresholds: (a) an individual was a relative of another subject in the study ( ≥ 0.20) using the Identity By State (IBS) estimation implemented in PLINK; (b) Affymetrix Quality Control (QC) call rate of < 95%; (c) gender ambiguity or error, determined using the observed genotypes of SNPs on chromosomes X and Y.
SNP Quality Control
SNPs that failed any of the following quality metrics were excluded from the study: (a) p ≤10–3 in a test for deviation from Hardy-Weinberg Equilibrium (HWE) in cases and controls; (b) Minor allele frequency ≤3%; (c) SNP call rate ≤98%; and (d) visual inspection of genotype clusters for the 500 most strongly associated SNPs. SNPs from copy number variant regions and SNPs known to have frequent Mendelian or genotyping errors in other studies [24] were also removed.
SNP Association Testing
After all quality control measures were performed, genetic association was determined using the whole-genome association analysis toolset, PLINK (v 1.07) [25]. Differences in allele frequencies between AIS case and control subjects were tested for each SNP using Cochran-Armitage Trend test. The allelic odds ratios were calculated along with 95% confidence interval. SNPs that passed the quality control parameters were used to generate Quantile-Quantile (QQ) plots ranking the observed χ2 values and plotting them against their expected values. Principal component analysis (PCA) was performed using EIGENSTRAT [26]. Haplotype-based association tests were calculated by 1-degree of freedom χ2 test, along with their respective odds ratios using R [27]. Different inheritance models were analyzed using logistic regression, which was also used to test for independence of SNP/haplotype effects and to calculate the phenotypic variance. The variance explained by a logistic regression model was calculated using the Cox Snell and Nagelkerke pseudo R2 method which is similar to the R2 concept of linear regression [28].
Principal Component Analysis (PCA)
Validation of self-reported ethnicity was performed by PCA with the EIGENSTRAT program. To account for linkage disequilibrium (LD) effects, we performed LD pruning using a window size of 50 SNPs, window increment of 5 SNPs and LD threshold of r2>0.2. We performed PCA on the set of LD pruned markers and analyzed the top 10 eigenvectors. We then eliminated self-reported Caucasian samples that fell outside of the core European cluster. Subsequently we performed a secondary PCA on the core European cluster to determine any remaining population substructure.
Imputation Analysis
IMPUTE2 (ver.2.1.2) [29] was used for imputing SNPs against the 1000-genomes (phase 1 data, v3). We performed the pre-phasing step on the genotyped samples prior to imputing haplotypes using a set of reference panel haplotypes from 1000 genomes. Imputation of the targeted region was carried out within ±2.5 Mb of the main marker of interest. Only SNPs that passed the confidence score of ≥0.9 and call rate of 0.98 with MAF>0.03 from imputation were retained for further analysis. The resulting haplotypes were used for haplotype based association tests.
Results
An initial cohort of 906 AIS cases and 1480 controls who self-reported being of European descent was genotyped using Affymetrix HuSNP 6.0 Microarray chip. After applying the sample quality filters and subsequently eliminating samples of admixture using PCA (Fig. A in S2 File) a final population of 620 AIS case and 1,287 control samples remained for further analysis. Since the case-population was recruited from surgical practices, the subjects tended to have more severe scoliosis; the 620 cases included 495 with severe AIS, 63 with moderate and 62 with mild AIS. After applying the SNP quality filters 457,206 SNPs remained for analysis. The genomic inflation factor (λ) was estimated to be λ = 1.08 as shown by QQ plot in Fig. B in S2 File.
After visual inspection of the genotype clusters of the top 500 associated SNPs, we identified rs11190878 (Ptrend = 4.18×10–9, OR = 0.63) in the LBX1 locus as associated with AIS (Fig. C in S2 File). None of the other genotyped markers reached the genome-wide significance threshold (P<5×10–8). Results for the 30 most strongly AIS-associated SNPs from the GWAS are shown in Table A in S2 File.
Through imputation of the LBX1 region we identified 10 additional markers that passed the threshold of Ptrend = 5×10–7 as shown in Table 1 (excluding markers in perfect LD). The table include rs11190870 (Pimputed = 5.43×10–9, OR = 0.66), previously reported to be associated with AIS is Asian populations, with association strength similar to rs11190878 identified here in a Caucasian population. The association plot of genotyped and imputed markers in the LBX1 region (Fig. 1) indicates that rs11190878 is the most strongly associated marker in this region. The marker rs11190870 originally reported in Asian population, is in modest LD with rs11190878 (r2 = 0.40, D' = 0.81). To investigate if the association signals observed with rs11190870 and rs11190878 are independent, we selected the 2 strongly associated genotyped markers from this study (rs7893223 and rs11190879) together with the imputed marker rs11190870 for haplotype analysis. We found six distinct haplotypes (frequencies > 0.01) at the LBX1 locus as shown in Table 2. The table shows two significantly associated haplotypes from the LBX1 locus (hap-2 and hap-6) with opposite effects. To investigate the inheritance pattern for the risk (hap-6) and protective (hap-2) haplotype we evaluated the effect size of the haplotype-pairs on AIS (Table 3). It is evident that the risk haplotype (TTA) only is associated with AIS under a recessive inheritance model, whereas the protective haplotype (CCG) appears to have an additive effect. To confirm which genetic inheritance model best explains the protective haplotype hap-2 (CCG) we applied logistic regression using the most common haplotype hap-6 (TTA) as reference. These results (Table 4) show the co-dominant model as having the lowest AIC indicating that this model best explains the inheritance pattern of the protective haplotype (P = 2.09×10–15). We also examined the inheritance models for the remaining 4 haplotypes together (hap-1, hap-3, hap-4 and hap-5) again using hap-6 (TTA) as reference. Contrary to hap-2, this group only has minimal effect (Table B in S2 File), but similar to hap-2 this analysis also supports a recessive inheritance model for hap-6. The variance explained by LBX1 was estimated using a multivariate logistic regression that included the haplotypes with the inheritance models established above. This analysis, summarized in Table C in S2 File, indicates that the risk factors have independent effects and together explain 1.4% of the variance for AIS.
The panel shows-log10 of p-values for genotyped SNPs (●) and imputed SNPs (×) plotted against Chromosomal positions (hg19) in MB along the X-axis and-log10(Ptrend) values along the Y-axis. The European genetic recombination rate estimated from 1000G data is shown in blue line. Key SNPs are identified by arrows. The SNP rs11190870 that originally was reported to be associated with AIS was imputed in the present dataset. The gene identified as FLJ41350 immediately upstream of LBX1, has recently been characterized as LBX1-AS1 (LBX1 antisense RNA 1).
A comparison of results from the present study to previously reported GWAS associations is shown in Table 5. The table shows strong replication of rs11190870 in LBX1 and nominal support for rs6570507 in GPR126 [30]. The present data does not support association to rs10510181 in CHL1 [14] or rs12946942 in SOX9 [19] in our Caucasian population. A lack of association between CHL1 and AIS has also recently been reported in a Han Chinese population [31].
Discussion
This study validates the association of LBX1 with AIS in subjects of European descendence and presents evidence of two independent, opposing association signals at the LBX1 locus. Our results show rs11190870 and rs11190878 with equally strong association to AIS, yet the two markers are only in moderate LD with one another (r2 = 0.40, D' = 0.81). This raises the question whether their respective association signals are independent or share an ancestral origin. Based on our haplotype analysis, it is evident that the ancestral (common) alleles of both rs11190870 (T, freq = 0.551) and rs11190878 (A, freq = 0.671) reside on the ancestral haplotype hap-6 (TTA) (Table 2). It is therefore reasonable to conclude that the Asian and European populations share the ancestral haplotype associated with a sharply increased risk for AIS.
Hap-6 (TTA, P = 1.25×10–9, OR = 1.56), the most common of the six haplotypes, is associated with an increased risk for AIS (Table 2). Our results also reveal a second haplotype, hap-2 (CCG, P = 2.75×10–7, OR = 0.65), with a protective effect. Further investigation of the inheritance patterns of the risk (hap-6) and protective (hap-2) haplotypes demonstrated that hap-6 (TTA) is inherited in a recessive fashion, whereas hap-2 (CCG) is inherited in a co-dominant fashion in the Caucasian population (Tables 3 and 4). Individuals carrying both hap-2 (CCG) and hap-6 (TTA) have neither increased nor decreased risk, suggesting that these two haplotypes neutralize each other. It remains to be determined if a protective haplotype also exists in Asian populations.
The power of haplotypes as clinical predictors compared to un-phased SNPs become clear in Tables 1–3, which in the case of rs11190878, show that the case frequency of the two homozygous genotypes (nnA/nnA + nnG/nnG) is 0.616, whereas the corresponding frequency of the two associated haplotypes (TTA/TTA + CCG/CCG) is 0.394. This comparison demonstrates the utility of haplotypes over un-phased SNPs by more strongly delineating the group of individuals at risk for AIS without compromising the effect size. We also estimate the phenotypic variance explained by hap-2 and hap-6 at LBX1 for AIS to be 1.4% (Table C in S2 File), which is large for a single locus in a complex disease.
In Fig. D in S2 File we show the frequency of patients grouped by disease severity and by haplotypes. The graph shows a clear trend whereby the TTA-TTA group has an increased frequency of moderate and severe patients and reduced frequency of mild patients, while patients with CCG-other and CCG-CCG has a reduced frequency of severe patients and increased frequency of mild patients. The trend fails to reach statistically significance in our study and it will be interesting to see if the correlation between haplotypes and severity can be independently confirmed.
The genomic architecture of the LBX1 locus (Fig. E in S2 File) reveals a highly conserved gene with extensive regulatory mechanisms. Of particular interest are the head-to-head orientation of LBX1 with its antisense counterpart LBX1-AS1, because antisense transcription increasingly is being recognized as an important regulator of gene expression [32]. The figure also shows some very extensive CpG islands (shown in Green) that flank LBX1and are a hallmark for epigenetic transcriptional regulation. Indeed, it has been shown that antisense expression can affect methylation patterns and lead to long-term repression [33]. Finally, Fig. E shows a pattern of remarkably extensive nucleotide conservation across vertebrate-species that not only include the protein coding regions but extends about 10 kb down-stream of LBX1. Together, the presence of LBX1-AS1, the CpG islands and the conserved regions suggests that LBX1 expression is very tightly regulated. A detailed functional description of the regulatory mechanisms that govern the LBX1 locus is not yet available, but we expect that sequencing of the LBX1 locus will identify rare and private mutations with large effects to segregate in some AIS families.
Supporting Information
S1 File. Includes all genotypes and case status information necessary to reproduce the imputation and haplotype analysis reported here.
https://doi.org/10.1371/journal.pone.0117708.s001
(TXT)
S2 File. Includes the following files: Fig. A.
PCA Plots. PCA plots of the 2 first eigenvectors at different geographical resolution are shown. Panel A show the Case and Control samples that self-reported as being of European ancestry (blue and green circles) together with previously characterized hapmap samples of African, Asian and European origin (black, purple and red circles). Samples determined by PCA to be admixed (blue circles) were eliminated from the study. A second PCA was performed on samples of PCA-verified European ancestry (blue rectangle) and shown in Panel B. In Panel B the cases are represented in green and controls in red. As suggested in panel B the genetic variance between case and control samples is minimal and the genomic inflation measure lambda λ = 1.08. Fig. B. Quantile-Quantile (Q-Q) plot. QQ plot of the allelic association analysis of expected versus observed χ2. The genomic inflation factor, λ = 1.08, indicate minimal stratification that does not require further PCA-adjustment. The QQ-plot was generated using 77,366 SNPs in linkage equilibrium (r2<0.2). Fig. C. Genotype Cluster Plots. Genotype cluster plots for the two genotyped SNPs in Table 1 are shown here using the A and B allelic intensity values. The distinct colors for each of the homozygous and heterozygous clusters represent cases (n = 853) and controls (n = 1,368). Fig. D. Haplotype vs. Severity Bar Plot. The bar-plot shows the case percentages for each of the three AIS classes (mild, moderate and severe) per haplotype groups. The graph shows a clear trend whereby the TTA-TTA group has an increased frequency of severe patients and reduced frequency of mild patients, while patients with CCG-other and CCG-CCG has a reduced frequency of severe patients and increased frequency of mild patients. Severity among individuals with TTA-CCG show equal frequency across the three groups, which supports the notion that the effects of TTA and CCG neutralize each other. Fig. E. Genomic architecture of the LBX1 locus. The genomic architecture around LBX1 on chromosome 10q24.31 reveals a highly conserved gene with extensive regulatory mechanisms. Of particular interest are the head-to-head orientation of LBX1 with is antisense counterpart LBX1-AS1 together with the very extensive CpG islands (shown in Green) that include and flank LBX1. In addition to the highly conserved coding regions of LBX1 a series of conserved regions are shown in blue mostly downstream of LBX1. Table A. Top 30 SNPs from GWAS. Table B. Logistic Regression Analysis of the Four Minor Haplotypes Combined. Table C. Multivariate Conditional Logistic Regression.
https://doi.org/10.1371/journal.pone.0117708.s002
(DOCX)
Acknowledgments
The authors gratefully acknowledge the scoliosis patients and spine surgeons who contributed to this research.
Author Contributions
Conceived and designed the experiments: KW LN RC. Performed the experiments: LN. Analyzed the data: RC HMA. Contributed reagents/materials/analysis tools: KW JWO. Wrote the paper: RC HMA KW. Clinical review: JWO.
References
- 1. Miller NH (1999) Cause and natural history of adolescent idiopathic scoliosis. Orthop Clin North Am 30: 343–352, vii. pmid:10393760
- 2. Lenke LG (2005) Lenke classification system of adolescent idiopathic scoliosis: treatment recommendations. Instructional course lectures 54: 537–542. pmid:15948478
- 3. Barois A (1999) [Respiratory problems in severe scoliosis]. Bulletin de l'Academie nationale de medecine 183: 721–730. pmid:10437294
- 4. Bengtsson G, Fallstrom K, Jansson B, Nachemson A (1974) A psychological and psychiatric investigation of the adjustment of female scoliosis patients. Acta psychiatrica Scandinavica 50: 50–59. pmid:4275025
- 5. Pehrsson K, Bake B, Larsson S, Nachemson A (1991) Lung function in adult idiopathic scoliosis: a 20 year follow up. Thorax 46: 474–478. pmid:1877034
- 6.
Renshaw TS (1988) Screening school children for scoliosis. Clinical orthopaedics and related research: 26–33.
- 7. Cheng JC, Tang NL, Yeung HY, Miller N (2007) Genetic association of complex traits: using idiopathic scoliosis as an example. Clinical orthopaedics and related research 462: 38–44. pmid:17534191
- 8. Miller NH (2007) Genetics of familial idiopathic scoliosis. Clinical orthopaedics and related research 462: 6–10. pmid:17603389
- 9. Ward K, Ogilvie J, Argyle V, Nelson L, Meade M, et al. (2010) Polygenic inheritance of adolescent idiopathic scoliosis: a study of extended families in Utah. American journal of medical genetics Part A 152A: 1178–1188. pmid:20425822
- 10. Grauers A, Rahman I, Gerdhem P (2012) Heritability of scoliosis. European spine journal: official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society 21: 1069–1074.
- 11. Tang NL, Yeung HY, Hung VW, Di Liao C, Lam TP, et al. (2012) Genetic epidemiology and heritability of AIS: A study of 415 Chinese female patients. Journal of orthopaedic research: official publication of the Orthopaedic Research Society 30: 1464–1469. pmid:22362628
- 12. Fan YH, Song YQ, Chan D, Takahashi Y, Ikegawa S, et al. (2012) SNP rs11190870 near LBX1 is associated with adolescent idiopathic scoliosis in southern Chinese. Journal of human genetics 57: 244–246. pmid:22301463
- 13. Gao W, Peng Y, Liang G, Liang A, Ye W, et al. (2013) Association between common variants near LBX1 and adolescent idiopathic scoliosis replicated in the Chinese Han population. PloS one 8: e53234. pmid:23308168
- 14. Sharma S, Gao X, Londono D, Devroy SE, Mauldin KN, et al. (2011) Genome-wide association studies of adolescent idiopathic scoliosis suggest candidate susceptibility genes. Human molecular genetics 20: 1456–1466. pmid:21216876
- 15. Takahashi Y, Kou I, Takahashi A, Johnson TA, Kono K, et al. (2011) A genome-wide association study identifies common variants near LBX1 associated with adolescent idiopathic scoliosis. Nature genetics 43: 1237–1240. pmid:22019779
- 16. Londono D, Kou I, Johnson TA, Sharma S, Ogura Y, et al. (2014) A meta-analysis identifies adolescent idiopathic scoliosis association with LBX1 locus in multiple ethnic groups. Journal of medical genetics 51: 401–406. pmid:24721834
- 17. Gross MK, Dottori M, Goulding M (2002) Lbx1 specifies somatosensory association interneurons in the dorsal spinal cord. Neuron 34: 535–549. pmid:12062038
- 18. Jagla K, Dolle P, Mattei MG, Jagla T, Schuhbaur B, et al. (1995) Mouse Lbx1 and human LBX1 define a novel mammalian homeobox gene family related to the Drosophila lady bird genes. Mechanisms of development 53: 345–356. pmid:8645601
- 19. Miyake A, Kou I, Takahashi Y, Johnson TA, Ogura Y, et al. (2013) Identification of a susceptibility locus for severe adolescent idiopathic scoliosis on chromosome 17q24.3. PloS one 8: e72802. pmid:24023777
- 20. Kotwicki T (2008) Risser sign: the value of the lateral spinal radiograph to assess the excursion of the iliac apophysis. Studies in health technology and informatics 140: 44–47. pmid:18809997
- 21. Dell DD, Regan R (1987) Juvenile idiopathic scoliosis. Orthopaedic nursing / National Association of Orthopaedic Nurses 6: 23–29, 50.
- 22. Lenke LG, Edwards CC 2nd, Bridwell KH (2003) The Lenke classification of adolescent idiopathic scoliosis: how it organizes curve patterns as a template to perform selective fusions of the spine. Spine 28: S199–207. pmid:14560193
- 23. Lonstein JE, Carlson JM (1984) The prediction of curve progression in untreated idiopathic scoliosis during growth. The Journal of bone and joint surgery American volume 66: 1061–1071. pmid:6480635
- 24. Hosking L, Lumsden S, Lewis K, Yeo A, McCarthy L, et al. (2004) Detection of genotyping errors by Hardy-Weinberg equilibrium testing. European journal of human genetics: EJHG 12: 395–399. pmid:14872201
- 25. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, et al. (2007) PLINK: a tool set for whole-genome association and population-based linkage analyses. American journal of human genetics 81: 559–575. pmid:17701901
- 26. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, et al. (2006) Principal components analysis corrects for stratification in genome-wide association studies. Nature genetics 38: 904–909. pmid:16862161
- 27.
R Core Team (2013) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://doi.org/10.1097/ACC.0b013e3182973a4f pmid:25611960
- 28. Nagelkerke NJD (1991) A note on a general definition of the coefficient of determination. Biometrika 78: 691–692.
- 29. Howie BN, Donnelly P, Marchini J (2009) A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS genetics 5: e1000529. pmid:19543373
- 30. Kou I, Takahashi Y, Johnson TA, Takahashi A, Guo L, et al. (2013) Genetic variants in GPR126 are associated with adolescent idiopathic scoliosis. Nature genetics 45: 676–679. pmid:23666238
- 31. Qiu XS, Lv F, Zhu ZZ, Qian BP, Wang B, et al. (2014) Lack of association between the CHL1 gene and adolescent idiopathic scoliosis susceptibility in Han Chinese: a case-control study. BMC musculoskeletal disorders 15: 38. pmid:24512353
- 32. Pelechano V, Steinmetz LM (2013) Gene regulation by antisense transcription. Nature reviews Genetics 14: 880–893. pmid:24217315
- 33. Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, et al. (2009) Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462: 315–322. pmid:19829295