The role of genetics and molecular mechanisms in early onset scoliosis

Abstract

Early onset scoliosis (EOS), a spinal deformity that occurs before 10 years of age, imposes significant morbidity due to its rapid Cobb angle progression, three-dimensional spinal curvature, and potential respiratory compromise from thoracic cage distortion. This condition, classified into idiopathic, congenital, neuromuscular, and syndromic subtypes, exhibits high phenotypic heterogeneity and multisystem involvement. Current treatments like bracing and surgery focus on modulating curve progression and preserving growth, but the genetic and molecular overview is not clear. This review delineates both the genetic basis and the pathogenic mechanisms of EOS. We summarize 79 genes implicated in EOS subtypes and discuss the underlying pathogenetic mechanisms, including somitogenesis defects, abnormal vertebral development, and neuromuscular disorders. We also highlight emerging therapeutic strategies and discuss future directions. By integrating genetic discoveries with molecular pathophysiology, this review provides a foundation for advancing precision medicine in EOS and highlights critical future research directions in the field.

1. Introduction

Early onset scoliosis (EOS), defined as a spinal deformity with a Cobb angle exceeding 10° presenting before 10 years of age, represents a significant clinical challenge in paediatric orthopaedics.¹ Unlike adolescent scoliosis, EOS occurs during critical phases of spinal, thoracic, and pulmonary maturation, often leading to severe three-dimensional deformities, rapid curve progression, and life-threatening restrictive lung disease secondary to thoracic insufficiency syndrome.² The management of EOS extends beyond spinal alignment to include cardiopulmonary function, mobility, and quality of life.

Clinically, EOS is classified into four primary etiological subtypes³: congenital EOS involves vertebral malformations; neuromuscular EOS is linked to muscle imbalance due to neurological or muscular disorders^3,4; syndromic EOS is associated with genetic syndromes except neuromuscular disorders³; and idiopathic EOS has an unknown origin and is not associated with known genetic syndromes.^5,6 This phenotypic heterogeneity complicates diagnosis and treatment.³

Current therapeutic strategies - including serial casting, bracing, and growth-friendly surgical interventions like Magnetically Controlled Growing Rods⁷ aim to modulate curve progression while preserving spinal and thoracic growth. However, these approaches are reactive and do not address underlying pathogenic mechanisms.

Despite decades of clinical research, the fundamental molecular drivers of EOS initiation and progression remain poorly understood. This knowledge gap impedes the development of targeted therapies and preventive strategies. Recent advances in genomic technologies, including whole-genome-sequencing (WGS), whole-exome sequencing (WES), genome-wide association studies (GWAS), and epigenetic analyses—unravel the complex genetic architecture of EOS.

This review summaries the current understanding of the complex genetic and molecular landscape of EOS across its subtypes. We critically examine key genes and signalling pathways implicated in critical developmental processes including somitogenesis, vertebrae development, and neuromuscular functions. We explore subtype-specific pathogenic drivers revealed through emerging insights from genetic alteration and functional studies. Finally, we evaluate the translational potential of this genetic knowledge for improving molecular diagnostics and guiding the development of future precision medicine therapeutics.

2. Genetic basis of EOS

We performed a comprehensive literature search in PubMed and Web of Science databases from December 2000 to December 2025: (“Early Onset Scoliosis” OR “EOS”) AND (“genetic variant” OR “mutation” OR “copy number variation” OR “epigenetics”). Studies examined genetic associations in human EOS, with a defined age of onset before 10 years, were included. A total of 45 EOS genetics studies, including 29 cohort studies, were reviewed, identifying 79 genes associated with EOS. We summarized the genetic variants associated with different subtypes of EOS (Table 1). Furthermore, we identified the most frequently implicated genes in EOS, estimated the prevalence of EOS subtypes, and discuss other genomic alterations contributed to EOS.

Table 1.

Genes and phenotypes associated with early onset scoliosis subtypes.

Subtypes	Classification		Gene Symbol	Disease/Clinical Phenotype
Congenital Scoliosis	Vertebral Malformation Phenotype	Segmentation Defects	TBX610,21,72 DLL373,77 VANGL113 VANGL213 PTK715,20 FGFR116 ROBO28,17 IGHM17 KAT6B9	Spinal curvature abnormalities Vertebral Segmentation Defects8,9,17 Congenital scoliosis with such as vertebral segmentation defects, hemivertebrae10,13,16,21,72,73,77
		Formation Defects	FBN118,20 SOX919,20 TBXT11,20 NOTCH220,21 DSCAM21,23 MYSM121,22 COL5A117,20 COL11A285,86	with cardiovascular anomalies11,19,20 or associated with Epilepsy/Limb deformities15,20, 21, 22, 23
		Mixed Defects	DHX4021 NBPF2021 SNTG121 GRID116 GSE116 IGHG117,94 IGHG316 RGS317,20 RNF21317,20	Unclear vertebral malformation mechanisms17,20,21,94,95
Neuromuscular Scoliosis	Pathogenic System	Myopathy	MSTO139 MYMX36	Hypomyelinating leukodystrophy39 Congenital myopathy36
		Neuropathy	ROBO332,33 SCAF426 EGR227,28 KCNQ229 SCN2A30 ALS231 NF-134,96 CCDC4735 SCAF426 EGR227,28 KCNQ229 MSTO139 ALS231 SCN2A30	Autosomal recessive demyelinating neuropathy26 Charcot–Marie–Tooth disease27,28 Early-onset epileptic encephalopathy29,30 Neurofibromatosis type 134,96 Infantile-onset ascending hereditary spastic paraplegia31 Trichohepatoneurodevelopmental syndrome35 Hypomyelinating leukodystrophy39
		Mixed Phenotype	TUBB4A37 FKRP40 SELENON40 SOX641 TRPV438 RASA220,21	Leukodystrophy37 Horizontal gaze palsy with progressive scoliosis32,33 Congenital muscular dystrophy with neurodevelopmental delay40 Neurodevelopmental disorder with dystonia41 Scoliosis with cardiovascular defects13,15,20,21,23 Scoliosis with facial dysmorphism20,21
Syndromic Scoliosis	Single or multi-system	Isolated bone Phenotype	COL1A142,43 COL1A242,43 FKBP1044 NSD145 FBN297 MEOX147,98 CDKL599 SLC35B248 RIPPLY249 NOTCH2100 PIK3CA51 AKT151 FBN146	Osteogenesis imperfecta42, 43, 44 Congenital contractural arachnodactyly45 Isolated macrodactyly97 Klippel-Feil anomaly47,98 CDD syndrome99 Spondylocostal dysostosis48 Ehlers-Danlos syndrome49 Marfan syndrome100 Steel syndrome51 Chiari malformation46,51
		Multi-System Phenotype	PTPN11101 LZTR154 SOS155 TRPV438 MAGEL256 ALPPK3102 UGGT157 MIA358 TANGO158 WWOX103 ECEL1104 DDX3X105 TRPM3106 MEGF10107 TTN60 COL5A2108 MED13L109 PLOD1110	Noonan syndrome54,55,101 Skeletal dysplasias38 Prader–Willi syndrome56 hypertrophic cardiomyopathy102 Congenital disorder of glycosylation57 Odontochondrodysplasia58 Infantile epileptic encephalopathy103 Distal arthrogryposis104 Dandy-Walker malformation105 Congenital Hypogonadotropic hypogonadism106 Sotos syndrome107 Hajdu Cheney syndrome60 MED13L-related intellectual disability108,109 Neurodevelopmental disorder with dystonia110
Idiopathic Scoliosis	NA	NA	COMT25 MTHFR25 LBX161 MMP-362 CHD763 ESR264 VDR65 TTLL1165 AKAP265 CALM165	Prader–Willi syndrome (risk gene association)25

2.1. Congenital scoliosis

Congenital scoliosis arises from embryonic vertebral malformations, affecting 0.5–1 per 1000 live births.^8,9 Congenital scoliosis can be categorized into three phenotypes based on vertebral malformation: segmentation defects, formation defects, and mixed defects.

Vertebral segmentation defects are characterized by failures in vertebral separation, involving genes like TBX6,^10,11 DLL3,¹² VANGL1/VANGL2,¹³ PTK7,^14,15 FGFR1,¹⁶ ROBO2,^8,17 IGHM¹⁷ and KAT6B.⁹ These gene variants drive spinal curvature and fusion anomalies.^8,9,17

Vertebral formation defects, marked by incomplete vertebral development such as hemivertebrae, are associated with mutations in FBN1,¹⁸ SOX9,¹⁹ TBXT^11,20, NOTCH2,^20,21 MYSM1^21,22 and COL5A1.^17,20 Vertebral formation defects may associate with other syndromes. Mutations in DSCAM^21,23 are linked to congenital scoliosis with epilepsy or limb deformities.

2.2. Neuromuscular scoliosis

Neuromuscular scoliosis primarily resulted from dysfunction of the neuromuscular system^24,25 is classified into three phenotypic categories: neuropathy, myopathy, and mixed phenotype.

The neuropathic category is associated with neural disorders. SCAF4²⁶ and EGR2^27,28 mutations are implicated in peripheral neuropathies, such as Charcot-Marie-Tooth disease, where motor nerve dysfunction leads to muscle denervation and subsequent scoliosis. KCNQ2²⁹ and SCN2A³⁰ affect neuronal function, often presenting with seizures and encephalopathy before the onset of spinal deformities. ALS2³¹ variants cause infantile onset ascending hereditary spastic paraplegia, often presenting with spastic spinal deformities. ROBO3^32,33 variants cause oculomotor abnormalities due to neural dysfunction, which in turn leads to scoliosis. In some neuropathic syndromes, the connection between neural dysregulation and scoliosis remains unclear. For example, NF-1 mutations underlie neurofibromatosis type 1, which is characterized by cutaneous neurofibromas, optic gliomas, and dystrophic scoliosis.³⁴ Moreover, CCDC47-related trichohepatoneurodevelopmental syndrome manifests with hair, nail, and neurological deficits in addition to scoliosis.³⁵

The myopathy category includes disorders characterized by muscle dysfunction. Variants in MYMX are associated with congenital myopathy and facial palsy.³⁶

The mixed phenotype category includes disorders with both neural and muscular involvement. Mutations in TUBB4A³⁷ causes scoliosis associated with myopathy and epilepsy, reflecting underlying neuropathic features. TRPV4-related ion channelopathies involved in both skeletal and neurological manifestations.³⁸ MSTO1 mutations cause myopathy and ataxia.³⁹ FKRP and SELENON mutations result in congenital muscular dystrophy accompanied by neurodevelopmental delay.⁴⁰ SOX6 variants are linked to neurodevelopmental disorders and dystonia, connecting motor control deficits to scoliosis.⁴¹ RASA2 mutations present with facial dysmorphism alongside neuromuscular involvement.²¹

2.3. Syndromic scoliosis

Syndromic scoliosis refers to spinal deformity that occurs in conjunction with underlying systemic disorders. It is broadly classified into single-system and multi-system phenotypes, based on the range of symptoms present.

The single-system disorder refers to skeletal abnormalities alone. For example, osteogenesis imperfecta, caused by COL1A1 or COL1A2 mutations, results in defective collagen synthesis, bone fragility, and scoliosis.^42,43 FKBP10 mutations also lead to severe osteogenesis imperfecta with features such as dentinogenesis imperfecta.⁴⁴ Other examples include NSD1-associated congenital contractual arachnodactyly, presenting with limb contractures and spinal deformities⁴⁵; FBN2-related isolated macrodactyly, which involves soft tissue overgrowth and scoliosis⁴⁶; and MEOX1 mutations causing Klippel-Feil anomaly, characterized by cervical vertebral fusion and scoliosis.⁴⁷ Additionally, mutations in genes such as SLC35B2⁴⁸ and RIPPLY2⁴⁹ are linked to spondylocostal dysostosis and Ehlers-Danlos syndrome, respectively, both associated with vertebral segmentation defects and scoliosis. Mutations in NOTCH2,⁵⁰ PIK3CA,⁵¹ and AKT1⁵¹ are implicated in Marfan syndrome, Steel syndrome, and Chiari malformation, respectively, each presenting with varying skeletal and connective tissue abnormalities.⁵²

The multi-system phenotype includes complex genetic syndromes in which scoliosis is one of several clinical features. Variants in PTPN11,⁵³ LZTR1,⁵⁴ and SOS1⁵⁵ cause Noonan syndrome, which is characterized by cardiac abnormalities such as pulmonic stenosis, short stature, and scoliosis. Mutations in MAGEL2 are associated with Prader–Willi syndrome, leading to hypotonia, obesity, and scoliosis as a result of hypothalamic dysfunction.⁵⁶ Other examples include UGGT1-related congenital disorders of glycosylation, which present with multisystem glycosylation defects.⁵⁷ Mutations in MIA3 and TANGO1 cause odontochondrodysplasia marked by dental and spinal abnormalities.⁵⁸ Rare conditions such as MEGF10-related EMARDD syndrome,⁵⁹ featuring early-onset respiratory distress, and TTN-associated Hajdu–Cheney syndrome, characterized by craniofacial abnormalities, further illustrate the diversity of multi-system involvement in syndromic scoliosis.⁶⁰

2.4. Idiopathic scoliosis

Idiopathic scoliosis is a structural spinal deformity of unknown cause, characterized by a three-dimensional curvature of the spine. The genetic susceptibility is predominantly driven by low-penetrance gene polymorphisms with significant racial and regional specificity: LBX1 is closely linked to susceptibility in East Asian and Turkish populations,⁶¹ MMP-3 shows stronger associations in Caucasians,⁶² while CHD7 and ESR2 correlate with curve severity in Polish Caucasian and Caucasian females,^63,64 respectively, reflecting sex-specific genetic regulation. Additional polymorphisms in VDR, TTLL11, AKAP2, and CALM1 further contribute to the genetic basis of sporadic cases,⁶⁵ with their effect sizes varying across ethnic groups and underscoring the disease's complex aetiology. COMT and MTHFR have also been linked to idiopathic scoliosis with unknow aetiology²⁵

2.5. Prevalence of EOS subtypes and frequently implicated genes

No systematic studies exist on the frequency of individual EOS subtypes. We estimated the frequency of individual EOS subtypes, based on the 45 EOS cases reported. Congenital scoliosis was the most common subtype, representing 53.8 % of cases, followed by neuromuscular (19.6 %), syndromic (17 %), and idiopathic (9.6 %; Fig. 1A).

Fig. 1.

Genetic Landscape of Early Onset Scoliosis.

A. Distribution of EOS subtypes based on their reported frequency.

B. Overlap among genes associated with different EOS subtypes.

C. The most extensively studied genes associated with EOS.

D. Mutations frequency in EOS. Sample sizes are given in parentheses.

To characterize the genetic relationships underlying EOS clinical heterogeneity, we analysed the mutations associated with the four major subtypes. Congenital and Syndromic share two genes: FBN1, NOTCH2. Syndromic and Neuromuscular share one gene: TRPV4 (Fig. 1B). These overlapping genes suggest that the pathogenesis of different EOS subtypes may converge on similar signalling pathways during spine deformation. Environmental influences or additional genetic variants contribute to the manifestation of different EOS subtypes.

The most frequently studied genes in EOS included TBX6, DLL3, COL1A1, COL1A2, FBN1 ROBO3, NOTCH2, SCAF4 and PTK7. TBX6 was reported in five studies, DLL3 in 4 studies, COL1A1/2 in 4 studies, FBN1 in 3 studies, and ROBO3 in 3 studies, collectively dominating current research focus in EOS aetiology (Fig. 1C). Furthermore, we performed a meta-analysis of the cohort studies reviewed to identify the most frequent mutations among EOS patients. Mutations in COL1A1 and COL1A2, representing 26.2 % and 23.8 % of EOS cases, respectively, are the most prevalent, followed by TBX6 mutations identified in 8.8 % EOS cases (Fig. 1D). These prevalent mutations underscore key molecular drivers of EOS susceptibility, highlighting targets for potential diagnostic and therapeutic strategies.

2.6. Other genomic alteration associated with EOS

Apart from genetic mutations, the copy number variations (CNVs) and epigenetics are also crucial in EOS research. Congenital scoliosis involves significant genetic contributions, with CNV emerging as key etiological factors. The 16p11.2 microdeletion and TBX6 haploinsufficiency account for 5–10 % of congenital scoliosis cases,²¹ while broader CNV analyses reveal additional pathogenic variants in NOTCH2, DSCAM, SNTG1 linked to vertebral development. Beyond CNVs, epigenetic dysregulation plays a critical role: genome-wide methylation studies identify hyper/hypo-methylated genes (IGHG1, SORCS2, COL5A1) enriched in MAPK signalling and axon guidance pathways,¹⁷ with KAT6B hypermethylation impairing chondrocyte proliferation via RUNX2/Wnt/β-catenin suppression.⁹ Collectively, these findings establish CNV-epigenetic interactions as central to congenital scoliosis pathogenesis.

3. Pathogenetic mechanisms of EOS

The spine consists of vertebrae, discs, ligaments, and muscles. EOS results from abnormal spinal development during embryogenesis and childhood, primarily involving vertebrae formation and muscle functions.²⁴ Vertebrae development begins with somitogenesis, followed by chondrification and ossification during the embryonic period. Postnatal remodelling continues throughout life66, 67, 68. On the other hand, neuromuscular disorders can affect the balance of muscle and the symmetry of vertebrae. More than 79 genes have been linked to EOS, shedding light on its genetic basis and underlying EOS pathogenesis. We broadly classified the pathogenesis of EOS into three categories: somitogenesis defects, abnormal vertebral development, and neuromuscular disorders.

3.1. Somitogenesis defects

Somite formation occurs during gestational weeks 3–5 and is a critical process for early spinal development. The segmentation clock, which orchestrates somite formation, operates through the Notch, Wnt, and FGF signalling pathways during embryogenesis.⁶⁹ Mutations in genes involved in these signalling pathways may impair vertebral segmentation and patterning, resulting in characteristic asymmetric spinal malformations, representing a primary aetiology of EOS (Fig. 2A).

Fig. 2.

Pathogenic insights of Early Onset Scoliosis

A. Spine Deformation Phenotypes of Genetic Variants in Different Signalling Pathways. The Notch, FGF, and Wnt signalling pathways synergistically regulate the somitogenesis process with periodic and rhythmic oscillations manner during embryogenesis around to week 3–5, which called segmentation clock.

Mutations in these pathways will lead to asymmetric spinal deformities include vertebrae formation failure like wedge-shaped vertebrae, hemivertebrae, butterfly vertebrae, and vertebral aplasia; fusion of vertebrae; and fusion and absence of ribs.

B. Structural skeletal defects of signalling variants in chondro-osseous development. The endochondral ossification process, following somitogenesis, can be further divided into chondrification and ossification. Several pathways regulating this process including Hedgehog, Wnt/β-catenin, and TGF-β. Moreover, extracellular proteins like FBN1 and collagens are essential for maintaining the vertebrae structure.

The mutations in these genes will disrupt the development of vertebrae, leading to the structural failure.

C. Neuromuscular dysfunction of genetic variants in neural cytoskeletal pathways. TUBB4A encodes a key microtubule component; mutations impair neuron axons, Schwann cell myelination, and muscle function. Ion channels, such as TRPV4 (active in both neurons and muscle) and SCN2A (mainly in neurons), facilitate action potentials and muscle contraction. MSTO1 regulates mitochondrial dynamics vital for nerve and muscle cells. MYMX promotes muscle maturation by mediating myoblast fusion into mature fibres.

TBX6, a crucial T-box transcription factor involved in Notch signalling, controls paraxial mesoderm specification and maintains segmental boundaries by regulating downstream genes like MESP2 during somitogenesis.^70,71 TBX6 malfunction will cause vertebral malformations like hemivertebrae or butterfly vertebrae.^72,73 Consistent with human findings, Tbx6 variants induce vertebral and rib defects in mice^74,75 and in zebrafish models.⁷⁶ DLL3 is another key regulator in Notch Pathway during somitogenesis, which controls the segmentation clock. Patients with DLL3 mutations often exhibit the “pebble beach sign” vertebrae and spondylocostal dysostosis.^12,73,77

Mutations in PTK7, VANGL1, and VANGL2 are also commonly identified in EOS. These mutations disrupt the Wnt/planar cell polarity (Wnt/PCP) signalling pathway, which plays a crucial role in somite development. PTK7 functions as a receptor involved in both canonical and non-canonical Wnt signalling; mutations in PTK7 can result in segmentation failures, fusion defects, and absence of ribs.¹⁵ Different VANGL mutations can lead to varying degrees of scoliosis, ranging from isolated vertebral malformations, such as single-segmented hemivertebrae, to extensive vertebral fusion.¹³

FGFR1, a key receptor in FGF signalling, is essential for proper segmentation of the paraxial mesoderm and somitogenesis.⁷⁸ Mutations in FGFR1 have been associated with EOS cases characterized by hemivertebrae, butterfly vertebrae, and rib absence.⁷⁹

3.2. Abnormal vertebrae development

Following Somitogenesis, the somite undergoes chondrification and ossification, ultimately forming the vertebrae. This continuous process is precisely coordinated spatiotemporally by several key signalling pathways, including TGF-β, BMP, Wnt/β-catenin, Hedgehog, and Notch.^68,80 Mutations in these pathways can disrupt the structural integrity of vertebrae, contributing to EOS (Fig. 2B).

Mutations in SOX9 and PAX1 have been linked to congenital vertebral malformations.^19,81 These two transcription factors regulate key signalling pathways, including Hedgehog, Wnt/β-catenin, and TGF-β. The TGF-β pathway regulates chondrogenesis throughout the process of chondrocyte differentiation, maturation, and ossification. Dysregulation of TGF-β and its receptors has been reported in scoliosis patients.⁸² Mutations affecting the TGF-β pathway can lead to abnormal vertebral development.

BMPs, a member of the TGF-beta superfamily, play a crucial role in skeletal development, including the formation of vertebrae.⁸³ Mutations in BMP7 and BMPER have been associated with severe scoliosis and other skeletal abnormalities.⁸⁴

Extracellular matrix (ECM) proteins such as collagens (type I to XII) are essential for the proper chondrification and ossification of vertebrae. Mutations in collagens are the most common risk factors of scoliosis. The COL11A2 variant has been functionally validated as associated with congenital scoliosis.^85,86 Col11a2 defect in a zebra fish model presents abnormal recruitment of osteoblasts during spinal development.⁸⁵

3.3. Neuromuscular disorders

Somitogenesis and vertebral development directly influence spine formation, while neuromuscular disorders indirectly affect spinal symmetry through muscle imbalance. Neuromuscular disorders contribute to EOS primarily through neuropathies and myopathies, disrupting spinal symmetry via muscle imbalance or direct muscle dysfunction (Fig. 2C).

Neurodevelopment depends on proper neuron formation and the integrity of the myelin sheath surrounding axons. Mutations in neurodevelopment related genes like ROBO3, EGR2, and SCAF4 are linked to neuromuscular scoliosis. ROBO3 guides neurons and their axons to their proper locations within the nervous system.⁸⁷ EGR2 mutations can cause demyelinating polyneuropathy, characterized by severe sensory loss, progressive thoracolumbar scoliosis, and trigeminal neuralgia.³⁸ SCAF4 functions as an anti-termination protein; SCAF4 mutations can lead to neurodevelopmental disorders and epilepsy.²⁶

Neural functions critically depend on ion channels, and mutations in key components of these channels can disrupt neural activity and lead to EOS. For example, mutations in the sodium channel protein type 2 subunit alpha (SCN2A) are associated with EOS and a range of neurological conditions, including epilepsy and movement disorders.³⁰

While neural dysregulation may lead to EOS through muscle imbalance, muscle dysfunction can directly contribute to EOS development. MYMX mediate myoblasts fusion into myofibers; MYMX mutations links congenital myopathy with scoliosis.³⁶

In some cases, both neuropathies and myopathies contribute to EOS development. The conduction of nerve impulses and muscle fibre contraction depend on the coordinated activity of ion channels. Mutations in the calcium-permeable ion channel protein TRPV4 lead to EOS likely caused by combined neuronal and muscle dysfunctions.³⁸ TUBB4A encoding β-tubulin is essential for axon integrity, axonal transport, myelination, cellular morphology and neuronal migration.^37,88 TUBB4A also regulates muscle function likely through microtubules.⁸⁹ TUBB4A defect have been linked to neuromuscular scoliosis.³⁷ MSTO1 is vital for maintaining the structure and function of mitochondria. MSTO1 mutations can cause myopathy and ataxia, contributing to EOS.^39,90

3.4. Emerging pathogenic mechanisms underlying EOS

The genetic heterogeneity of EOS presents significant challenges to understanding its molecular basis. To elucidate underlying mechanisms, we conducted functional enrichment analyses on curated gene sets associated with various EOS subtypes, uncovering key pathways involved in spinal development and disease pathogenesis.

We utilized the Human Phenotype Ontology (HPO) to validate gene sets linked to scoliosis-related phenotypes (Fig. 3). The congenital EOS gene set was enriched for spinal and vertebral deformities; the neuromuscular EOS gene set showed associations with kyphoscoliosis; the syndromic EOS gene set was linked to abnormalities of the limbs and knee joints; and the idiopathic EOS gene set was over-representative for Abnormal thymus morphology.

Fig. 3.

Human phenotype ontology enrichment analysis of individual early onset scoliosis subtypes.

Biological process enrichment revealed that congenital EOS genes are involved in multiple embryonic organ development pathways, including skeletal and nervous system formation. Notably, pattern specification processes - crucial for spine development - were also enriched (Fig. 4). These findings emphasize the prominent role of neuromuscular factors in congenital EOS. Genes associated with neuromuscular EOS primarily participate in muscle development as well as glial cell differentiation, suggesting that muscle dysfunction significantly impacts this subtype (Fig. 4). The syndromic EOS gene set was enriched for skeletal system development and key signalling pathways such as TGF-beta, EGFR, and ERBB. Additionally, metabolic pathways - including insulin signalling, glucose, carbohydrate, and amino acid metabolism - were notably involved, indicating metabolic dysregulation as a contributing factor in syndromic EOS (Fig. 4). Idiopathic EOS associated genes were enriched for tissues morphogenesis (Fig. 4).

Fig. 4.

Biological processes enrichment analysis of individual early onset scoliosis subtypes.

Overall, distinct gene variants across EOS subtypes reflect unique pathogenic mechanisms underlying each form. These insights also provide valuable directions for future research, highlighting neuromuscular disorders as central to both neuromuscular and congenital EOS, and identifying metabolic disturbances as emerging contributors to syndromic EOS.

4. Emerging clinical translation strategies for EOS

The clinical translation of EOS needs to integrate genetic screening, risk stratification and targeted therapy. Among them, genetic screening is the premise of precise intervention, and therapy is the goal. The two together constitute the core link of precision medicine for EOS.

4.1. Genetic screening and risk stratification

Advanced genetic screening represents a cornerstone of emerging preventive strategies. Preimplantation genetic testing enables identification of embryos carrying pathogenic variants in high-penetrance EOS-associated genes, allowing selection of unaffected embryos in high-risk families. Post-conception genetic testing facilitates early diagnosis of EOS, enabling targeted prenatal monitoring and immediate postnatal intervention planning. Cascade screening of first-degree relatives further identifies at-risk individuals for proactive surveillance, shifting from reactive to predictive management.

4.2. Genome editing technologies

CRISPR-Cas9 systems show transformative potential for correcting EOS-associated mutations in preclinical models.³⁸ Key challenges remain in delivery efficiency, temporal precision, and ethical frameworks for germline applications. Ongoing advances in base editing and prime editing offer promising alternatives with reduced indel risks.

4.3. Interventions in the perinatal period

Nutraceutical modulation of key developmental pathways represents a promising frontier in perinatal EOS prevention. Periconceptional folic acid supplementation reduces the risk of neural tube defects, potentially mitigating pathways that predispose to spinal deformities.⁹¹ Randomized interventional trials suggest that combined supplementation with melatonin, calcium, and vitamin D may aid in the correction of spinal curvature.⁹² Consequently, perinatal preventive interventions hold significant potential for reducing the incidence and severity of EOS.

5. Future perspectives

5.1. Comprehensive genomic profiling in EOS

WGS and WES offer superior capacity for exploring genetic contributions compared with GWAS. Current successes with WES - such as identifying pathogenic variants in genes like SHISA3 and HDAC4 in congenital scoliosis - demonstrate the value of deep sequencing and highlight the even greater potential of WGS.⁹³ The decreasing costs of high-throughput sequencing have made WGS more accessible, providing comprehensive coverage of coding regions, non-coding regulatory elements, structural variants, and epigenetic modifications - for instance, whole-genome bisulfite sequencing (WGBS) has been used to identify hyper/hypo-differentially methylated genes (e.g., IGHG1, SORCS2, COL5A1) linked to CS pathogenesis.¹⁷ Implementing WGS or WES in routine care requires standardized pipelines for variant interpretation and cost-effective strategies for diverse healthcare settings.

5.2. Large-scale multicentre cohort studies for EOS

Given the rarity and heterogeneity of EOS, large-scale multicentre cohort studies with diverse populations are essential. Longitudinal phenotyping throughout curve development and treatment is vital for correlating genotypes with clinical outcomes. Establishing shared biorepositories and standardized ontologies for genomic, imaging, and phenotypic data will facilitate data harmonization, accelerating the discovery of novel susceptibility loci and refining genotype-phenotype correlations across different EOS subtypes.

5.3. Family-based pedigree analysis and genetic studies

Family-based studies, particularly those involving large multigenerational pedigrees, offer unparalleled opportunities to identify high-penetrance variants. Such studies can pinpoint causal mutations and clarify how modifier effects from secondary genetic or environmental factors influence disease expressivity. They also help elucidate inheritance patterns, whether autosomal dominant, recessive, or oligogenic, providing deeper insights into heritable mechanisms often obscured in sporadic cases.

5.4. Functional genomics using somite organoids and CRISPR screening

Functional genomics, especially CRISPR-based screening in human somite organoids, are invaluable for modelling early vertebral development. These organoids enable systematic interrogation of gene function in somitogenesis. High-throughput drug testing in this human-relevant system can evaluate candidate therapeutics, while mechanistic studies can reveal how EOS-associated variants disrupt segmentation clock dynamics or cell polarity. Such integrative efforts will bridge genetic discoveries with disease pathophysiology, accelerating the identification of novel therapeutic targets.

6. Conclusions

EOS presents a significant clinical challenge due to its onset during critical developmental periods, leading to complex three-dimensional deformities and potential cardiopulmonary compromise. This condition exhibits substantial phenotypic heterogeneity, classified into four primary etiological subtypes: congenital, syndromic, neuromuscular, and idiopathic.

Advances in genetic research have illuminated the molecular underpinnings of EOS. Key mutations disrupt critical pathways governing somitogenesis, vertebral development, and neuromuscular regulation. Congenital scoliosis is strongly linked to somitogenesis defects. Neuromuscular scoliosis arises from primary dysfunction of nerves or muscle, often leading to imbalance and spinal asymmetry. Syndromic cases involve systemic disorders. In contrast, the pathogenesis of idiopathic scoliosis remains elusive, though genetic associations suggest potential susceptibility factors.

Current management relies heavily on mechanical interventions to modulate curve progression and preserve growth. However, emerging genetic insights offer transformative potential. Preimplantation genetic testing enables preventive embryo selection in high-risk families, while prenatal genetic screening facilitates early diagnosis, risk stratification, and proactive intervention planning. The integration of genomic technologies, large-scale cohort studies, and functional genomics holds promise for identifying novel druggable targets and advancing precision therapeutics.

Ultimately, a deeper understanding of the genetic architecture and molecular pathophysiology of EOS is paramount. This knowledge is crucial for developing improved diagnostic tools, enabling genetic counselling, stratifying risk, and guiding the future development of targeted therapies aimed at addressing the root causes of the disease, thereby improving long-term patient outcomes and quality of life.

Guardian/patient's consent

This research did not require guardian/patient's consent as this was a review article only.

Credit author statement

SF: Methodology, investigation, data curation, manuscript-original draft and editing.

YCD: Methodology, investigation, data curation, manuscript-original draft and editing.

SWC: Project administration, manuscript-review and editing.

KSCC: Project administration, manuscript-review and editing.

JPYC: Conceptualization, study design, methodology, project administration, resources, supervision, manuscript-review and editing.

ZZS: Study design, methodology, project administration, supervision, manuscript-original draft and editing.

Ethical statement

Not applicable for this review article.

Funding statement

This work was supported by the SRS-Cotrel Foundation Basic Science Research Grant [AR240008, 2024] and the Li Shu Fan Professorship in Orthopaedic Surgery.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.