Scoliosis is one of the most commonly occurring abnormalities of the vertebral column, defined as a side-to-side deviation of the spine from the normal axis of the body.1 Lateral curvature of the spine of at least 10° is seen on upright spinal radiographs, accompanied by rotation of the spine.1 2 Adolescent idiopathic scoliosis (AIS) is a disorder of unknown aetiology, and occurs in otherwise healthy children with no underlying muscular or neurological abnormalities.3 Appearing as early as 7 years of age and progressing rapidly in some cases, AIS, if untreated, can result in loss of mobility, back pain, respiratory and cardiac disease. Most curves observed in adolescence probably begin during the juvenile years; the curvature is then enhanced due to rapid growth occurring at puberty, which has a destabilising effect on the curved spine.4 However, not all curves progress. It is this variability that poses problems to the physicians caring for a patient with AIS.

It is widely recognised that there is a genetic predisposition to AIS. The mode of inheritance is still unclear in some families, but evidence for autosomal dominant inheritance with partial penetrance,5^–7 X-linked inheritance,8 9 sex-influenced autosomal dominant inheritance with a female:male ratio of 8:1,10 and a complex trait or multifactorial mode of inheritance11 12 has been described.

Genetic analyses have identified several candidate loci predisposing AIS. Chromosomal regions on 6q, 10q and 18q,13 17p11.2,14 19p13.3,15 8q1116 and Xq23-26.19 have been reported. More recently, significant evidence of linkage to various regions on chromosomes 6, 9, 16 and 1717 and the previously described locus on chromosome 19p1315 has been independently confirmed.18 These studies indicate substantial genetic heterogeneity in the aetiology of AIS.

To date, no disease genes causing AIS have been identified; however, polymorphisms associated with susceptibility loci for idiopathic scoliosis have been described in the γ-1 syntrophin gene (SNTG1; OMIM*608714) on 8q11.22,16 the matrillin-1 gene (MATNl; OMIM*115437) on 1p35,19 and the chromodomain helicase DNA-binding gene (CHED7; OMIM*608892) on 8q12.1.20

In this study, we report our findings from a genome-wide scan of 25 British families affected with autosomal dominant AIS and the identification of two significant genetic loci in 10 of these families. To our knowledge, this is the first report of linkage analysis for AIS in the British population, describing refinement of a major locus for AIS on chromosome 9q and the identification of a novel locus on chromosome17q.

METHODS

Family ascertainment and clinical assessment

We studied 25 unrelated multiplex AIS families, referred to St. George’s Hospital Medical School by their orthopaedic surgeons. Of these, 24 families were Caucasian of British descent, and one was of Afro-Caribbean origin. Each family was identified through a proband presenting with adolescent scoliosis requiring orthopaedic care. All subjects, male and female, were ascertained using the same criteria and were examined by the same clinical geneticist (AHC) and/or trained genetic nurse (MA, GB). Their diagnosis was confirmed by the Adams forward bending test21 and by radiography of the spine while standing upright. The Cobb angle was measured, and all those affected had the internationally accepted criteria of at least 10° lateral spine curvature with rotation.

Syndromes featuring scoliosis as a phenotypic characteristic were excluded, including Beals syndrome, MASS (mitral valve, aorta, skeleton, and skin) syndrome and Marfan syndrome (MFS). Some of the families examined had mildly marfanoid body build that travelled with the scoliosis gene, but MFS was excluded clinically using Ghent criteria22 and by the fact that none of the 25 families linked to the FBN-1 locus on chromosome 15q21. In each case, the family phenotype did not include ocular (dislocated lens, detached retina), or cardiac (history of aortic aneurysm, cardiac valve insufficiency, sudden cardiac death) features required for a clinical diagnosis of MFS. Ophthalmological and cardiac ultrasound examinations were not performed; however, each family member was thoroughly examined for signs of connective tissue deficiency, using a standardised examination sheet (available on request). Signs of ligamentous laxity such as joint hypermobility were especially noted, together with skeletal proportions and signs of skin involvement such as lumbar striae, and thickness and elasticity of skin over the elbow. As is standard practice, all unaffected individuals were classified as “phenotype unknown” for the purpose of linkage analysis for a partially penetrant autosomal dominant model,13 including individuals >18 years of age with normal results from the Adams test and radiography investigation. Phenotypes were assigned prospectively before genotyping on the basis of clinical and radiological data. Aside from the spinal deformity, affected individuals were generally in good health.

In the largest family (SC32), all living affected and unaffected members, including spouses, were examined at home visit by one of the authors (AC). Examination included Adams forward bending test, and the scoliometer was used for all members who had not had spinal radiography or corrective surgery. All affected family members had required bracing or surgery, including all the deceased affected in generation I (see supplemental files, available online). Photographs of deceased members were studied for signs of scoliosis or kyphosis. Verbal reports of family medical history were cross-checked with all relatives in generation III. The one living unaffected member of generation II (II:11) was thoroughly examined and declared unaffected (see supplementary fig 1), as were her two adult children and five adult grandchildren, confirming her unaffected status.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Standing posterior radiographs of proband V:1 of family SC32, showing right thoracolumbar spinal curvature in the absence of congenital vertebral abnormality. (A) Cobb angle measurement of 40° at time of diagnosis, (B) Cobb angle measurement of 56° before corrective surgery, and (C) spinal fusion.

Blood samples were obtained for DNA analysis, with informed consent, from family members aged ⩾18 years. In the case of minors aged <18 years, informed consent was obtained from the responsible adult as approved by the ethics committee of St George’s Hospital NHS Trust.

Genotyping and data management

Genomic DNA was extracted using a standard procedure from all 25 families (208 people, including 116 affected). Preliminary microsatellite analysis excluded previously described autosomal dominant AIS loci on chromosomes 6p, 10q, 18q,13 17q11.214 and 19p13.315 in all families (data not shown). A genome-wide screen was subsequently performed at the Centre of Medical Genetics, Marshfield, using 410 fluorescein-labelled microsatellite markers distributed at about 10-cM intervals throughout the genome (Centre for Medical Genetics, screening set 13) as described previously.23 Genetic saturation analysis was performed using an additional 17 microsatellite markers for the 9q31.2-q34.2 locus and 9 microsatellite markers for the 17q25.3-qtel locus, and genotyped using a genetic analyser (ABI 3100; Applied Biosystems, Foster City, California, USA). Allele sizing and scoring were performed using Genescan V.3.7 software (Applied Biosystems). Allelic data were ascertained and analysed for mendelian inconsistencies using PedCheck.24

Statistical analysis

Two-point linkage analyses with MLINK V.5.1 software25 were performed to calculate LOD scores, under the assumption of an autosomal dominant mode of inheritance with disease allele frequency of 0.01 and an estimated penetrance of 80%. Cyrillic V.2.1 (Cherwell Scientific, Oxford, UK) software was used for pedigree drawing and assignment of alleles. Multipoint linkage analyses were performed using GENEHUNTER V.2.126 with an estimated disease penetrance of 80% and a gene frequency of 0.01. Non-parametric linkage (NPL) analyses were performed using MERLIN27 and the multipoint location score (log 10) from the NPL score that MERLIN calculates was extracted to generate GENEHUNTER graphical plots. Genetic heterogeneity and homogeneity were tested using the HOMOG program.28 Genetic and physical distances, and marker orders used for two-point and multipoint linkage analysis, were estimated according to sex-averaged genetic maps obtained from the Marshfield genetic linkage map,29 deCODE map30 and UCSC Genome Browser physical map database (June 2006 assembly, http://genome.cse.ucsc.edu/).

RESULTS

Clinical findings

We studied 25 multigeneration families with 116 members affected by an autosomal dominant form of AIS. All families were examined clinically by the same clinical geneticist (AHC) or a trained genetic nurse. Affected subjects had at least 10° lateral spine curvature with rotation. Each family presented with at least three living affected members in at least two generations, with onset of scoliosis during adolescence (8–18 years). There was no history of consanguinity, and the disorder in all families was transmitted to subsequent generations without skipping. Both sexes were affected, and male to male transmission occurred eight times, in seven families. Transmission of the disorder is therefore consistent with autosomal dominant inheritance. A clinical synopsis for the probands of the 10 families described in this article is shown in table 1.

Linkage of AIS to 9q31.2-q34.2 and refinement of the locus

Linkage analysis revealed several potential chromosomal regions in the 25 AIS families analysed. The highest two-point LOD score of Z_max = 3.64 (θ = 0.0) was obtained for microsatellite marker D9S2157 in a five-generation AIS kindred, family SC32 (see supplementary fig 1). The severity of curvature in this family ranged from 15° to 65° in 8 affected people from 21 sampled, with the proband (V:1) demonstrating a Cobb angle measurement of 40° at time of diagnosis, which progressed to 56° before surgery (fig 1).

Following the identification of potential loci from the genome-wide screen, 16 additional microsatellite markers were analysed, with subsequent linkage analyses applying a partially penetrant autosomal dominant inheritance pattern in an affected-only analysis, in which unaffected family members (including children <18 years), were analysed as “phenotype unknown”. The critical region cosegregating with disease in all affected members of family SC32 was localised within a region of ∼21 Mb, between markers D9S930 and D9S1818 on chromosome 9q31.2-q34.2 (see supplementary fig 1). The proximal boundary of the disease region is defined by a recombination observed in affected individual III:9, between markers D9S930 and D9S934; the distal boundary was observed in affected individuals IV:1 and III:11 between markers D9S1793 and D9S1818. Two-point LOD scores were calculated for these additional markers; a maximum score of Z_max = 2.76 (θ = 0.0) was generated with D9S1795 in an affected-only analysis, and a multipoint score of 3.08 was observed between markers D9S1847 and D9S1793 at 9q34.13-q34.2 (fig 2). We further extended our analysis to include seven additional AIS families each with at least three affected individuals in two or three generations (SC22, SC33, SC35, SC37, SC42, SC61 and SC109). Because of their small family size, significant LOD scores were not obtained from the genome-wide screen; however, haplotype analysis showed disease segregation with D9S2157. Additional markers on chromosome 9q were genotyped in these families (see supplementary fig 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Multipoint linkage analysis showing LOD scores for markers mapping across the AIS 9q region in family SC32. A peak LOD score of 3.08 was observed between markers D9S1847 and D9S1793 at the 9q34.13-q34.2 AIS locus.

Chromosome 9q is a major AIS locus

In order to evaluate genetic heterogeneity versus homogeneity in the eight AIS families apparently segregating with chromosome 9q, we performed a homogeneity test with the HOMOG program (A test)28 using the most informative marker, D9S2157. Maximum likelihood for linkage to D9S2157 was estimated under various values of θ and α, with α referring to the proportion of linked families. The following three maximum log₁₀ likelihoods were obtained: (1) 2.1736 with θ = 0 and α = −1 under the hypothesis of homogeneity and linkage, (2) 7.2723 for θ = 0 and α = 0.2 under the hypothesis of heterogeneity and linkage, (3) 0 for θ = 0 and α = 0 under the hypothesis of exclusion of linkage between D9S2157 haplotype and AIS. A χ2 value of 10.197 was obtained under the hypothesis “linkage and heterogeneity v linkage homogeneity,” 4.347 under the hypothesis “linkage and homogeneity v no linkage” and 14.545 under the hypothesis “linkage and heterogeneity v no linkage.” HOMOG analysis supports haplotype data and is suggestive of homogeneity for these eight families, which are assumed to be linked to the D9S2157 locus with probability values ranging from 0.20 to 0.99 (table 2). Although haplotype analysis of the additional 9q families indicates potential further refinement of the locus (fig 3), the critical interval is still considered to be defined by family SC32 due to lack of significant LOD scores in these smaller families.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Combined disease intervals for eight AIS families segregating with the chromosome 9q AIS locus. Key recombination events observed only in affected individuals define the disease interval between markers D9S930 and D9S1818 in the linked family SC32 on chromosome 9q31.2-q34.2. Eight other potentially linked families are depicted with their corresponding affecteds-only disease intervals. Grey bars denote uninformative markers.

Identification of a novel locus for AIS on 17q25.3-qtel

Two families, SC36 and SC23 (see supplementary fig 3), showed evidence of linkage to chromosome 17q from the genome-wide screen with a maximum two-point LOD score of Z_max = 2.64 (θ = 0.0) for marker TTCA006M in family SC36. Marker AAT095 on chromosome 17q25.3 generated a two-point LOD score of Z_max = 2.27 (θ = 0.0) in family SC36, and Z_max  = 1.81 (θ = 0.0) in family SC23. A cumulative two-point LOD score of Z_max = 4.08 (θ = 0.0) was obtained for this locus. SC36 is a four-generation family with 10 affected members, in which severity of disease ranged from 11° to 40° curvature, and SC23 is a two-generation AIS family with 6 affected members, in which severity of disease ranged from 14° to 55° curvature.

Subsequent fine mapping at this locus with additional microsatellite markers resulted in refinement of the locus on chromosome 17q25.3. A maximum affected-only cumulative two-point LOD score of Z_max  = 3.79 (θ = 0.0) for marker D17928, and multipoint score of 3.78 was obtained between marker D9S914 and 17qtel (fig 4). The proximal boundary of the critical interval is defined by a recombination event observed in affected individual IV:5 of family SC36. The AIS locus therefore lies distal to marker D17S1806. HOMOG analysis for marker AAT095 supports haplotype data with a conditional probability of linked type score of 0.986 for SC36 and 0.9552 for SC23. In addition, HOMOG analysis of the 9q marker D9S2157 resulted in a probability of 0 for both families (table 2). Taken together, haplotype and HOMOG analysis in families SC36 and SC23 indicates that the AIS locus on chromosome 17q25 maps within a 9.6-cM (approx. 3. Mb) interval, positioned between marker D17S1806 and the telomere at 17q25.3-qtel (fig 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 Multipoint LOD scores of microsatellite markers on chromosome 17 spanning the AIS 17q25.3 locus in families SC23 and SC36. A peak score of 3.78 was observed between D17S914 and 17q telomere.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5 Combined minimal disease intervals for two AIS families mapping to the Chromosome 17q25 AIS locus. Solid box, segregation of markers with disease in affected members of families SC23 and SC36. Minimal critical interval is between markers D17S1806 and 17q telomere, within a <3.2 Mb region on 17q25.3. Grey bar, uninformative markers.

DISCUSSION

The aim of this study was to establish the genetic loci implicated in AIS in the British population. We describe a major locus for autosomal dominant AIS segregating with markers on 9q31.2-q34.2, which is refined in this study, and a novel locus on 17q25.3-qtel.

The independent mapping of AIS in eight of our families to the chromosomal locus 9q31.2-q34.2 is in agreement with the recently published candidate AIS loci predisposing to scoliosis, with chromosome 9q being significantly implicated in the development of severe curves.17 The identification of significant linkage at this locus for AIS in British families confirms the presence of a causative gene on 9q31.2-q34.2 and suggests that this may be a major locus for AIS. Our region lies within the initially identified region, but has been refined from a 34.6 Mb region between markers D9S938 and D9S1838, to a genetic interval of ∼21 Mb between D9S930 and D9S1818 in the linked family SC32 (fig 3). Interestingly, haplotype analysis at the current resolution of markers did not reveal shared alleles between the eight linked families, suggesting that AIS may not be due to a common founder mutation. Current data indicate a novel locus for AIS, adding to the level of heterogeneity.

AIS is a complex disorder in which a major gene or genes may be responsible for susceptibility to the condition. These genes may be strongly affected by modifying factors such as age and sex, to cause phenotypic variation between affected individuals within the same family.1 This is possibly one of the major reasons why the affected female:male ratio is high, with females more severely affected than males in our families.10 It could be hypothesised that sex-influenced inheritance or female hormone dependency of the causative gene may be responsible for the female preponderance in some families. Additionally, it is possible that modification of the expression of AIS is influenced by environmental factors, although none has as yet been identified.

For ethical reasons, spinal radiographs of young siblings of affected probands could not be obtained, necessitating the use of a scoliometer to estimate spinal curvature. Genetic subgroups may be revealed once causative genes are identified, enabling genotype–phenotype correlation. At present, no identifying features can differentiate between families segregating with the 9q31.2-q34.2 and 17q25.3-qtel.

The regions on chromosomes 9q31.2-q34.2 and 17q25.3-qtel linked with AIS in our pedigrees contain many candidate genes. The 9q locus and 17q locus contain 160 and 56 genes, respectively. Strong candidate genes in these regions include transcription factors DDX31 on 9q34.2 and DDX48 on 17q25.3, which are expressed in connective tissue, bone, ligament, muscle and skin, and are believed to be involved in embryogenesis, spermatogenesis, and cellular growth and division.31

The described AIS loci on 9q31.2-q34.2 and 17q25.3-qtel do not segregate in the remaining 15 AIS families included in the genome scan, therefore further genetic analysis of these families is required to identify other AIS loci in the British population. The families studied are an excellent national resource; however, multiplex families continue to be recruited. Further linkage studies of the identified regions reported in the present study should facilitate further refinement of these regions.

This common disorder is of wide interest to various specialties including orthopaedic surgery, rheumatology, paediatric and adolescent medicine, respiratory medicine and cardiology, as well as medical genetics. It is hoped that identification of the causative genes within the critical disease regions for AIS described in this study on 9q31.2-q34.2 and 17q25.3-qtel will aid in the identification of other loci and AIS disease genes, which may act in a common pathway. The intervals contain many genes that are potential candidates for disease, such as developmental genes, growth factors and transcription factors. Each of these genes warrants consideration for future analysis, which should provide insight into the role of genetic influences that affect skeletal development of the adolescent spine. Elucidation of the causative genes and disease pathophysiology will lead to risk stratification, and a rational basis for family screening and development of preventive therapy.