Neurofibromatosis is an autosomal dominant genetic disease with familial inheritance (1), with neurofibromatosis type 1 (NF-1) being the most common form of neurofibromatosis. NF-1primarily affects the peripheral nervous system (2), and often involves bones, skin, and soft tissues with an incidence of 10–60% (3). Approximately 50% of NF-1 patients are susceptible to bone malnutrition, with scoliosis being the highly prevalent bone damage (4). NF-1 scoliosis can be classified as dystrophic and non-dystrophic, with significant differences of natural history and treatment plans. Non-malnourished scoliosis, which represents the more common spinal deformity of NF-1 patients, resembles adolescent idiopathic scoliosis in terms of imaging manifestations, treatment plans, and complications (5). If the Cobb angle of the main curvature of this type of scoliosis exceeds 40°, posterior spinal fusion is required. Dystrophic scoliosis, the more severe subtype, usually presents with early onset, rapid progression, and distinctive radiographic features, including vertebral scalloping and rib penciling. Given its aggressive nature, timely surgical intervention and long-term management are essential to prevent severe deformity and related complications.

Previous studies have shown that the main cause of NF-1 is related to mutations of NF-1 gene, leading to abnormalities in the encoding of neurofibromatosis proteins (6). The major reason for bone dysplasia in NF-1 is attributed to the reduced expression of neurofibromatosis proteins in osteoblasts, osteoclasts, chondrocytes, fibroblasts, and vascular endothelial cells (7). The NF-1 gene is located on chromosome 17 (17q11.2), containing approximately 60 exons and spanning over 300 bases (8). Mutations types include nonsense, missense, insertion or deletion (frameshift), splicing, and complete gene deletion. Neurofibromatosis proteins are the gene product of NF-1, containing functional protein of 2,818 amino acids with molecular weight of 280kD. Neurofibromatosis proteins have homology between the 25–40 kDa fragment of NF-1GRD and the 250–400aa fragment of p120GAP, and act as GTPase-activating proteins (GAPs) on Ras to perform their primary function (9). GAPs are currently the only clear functional fragments of neurofibromatosis proteins. NF-1 mRNA is widely expressed in neural crest-derived tissues, bones, and vascular endothelium, and its protein product, neurofibromin regulates cell proliferation and differentiation mainly through the Ras/MAPK and PI3K/AKT signaling pathways. Moreover, specific NF-1 variants have been correlated with distinct clinical phenotypes, suggesting important genotype–phenotype associations in disease severity and manifestations.

Unlike adolescent idiopathic scoliosis, malnourished scoliosis is often accompanied by significant spinal abnormalities, such as acute angular deformity, vertebral rotation or subluxation, shortened ribs, and slender pedicle (10). The typical site is the upper thoracic spine, which is often accompanied by obvious shoulder imbalance. Early and active surgical intervention is necessary to prevent the progression of scoliosis in patients with neurofibromatosis. While previous studies have conducted comparative analyses of the surgical treatment, imaging manifestations, and other aspects of scoliosis in neurofibromatosis (11–14), there remains limited research on the intrinsic genetic initiating mutations and genetics of scoliosis in neurofibromatosis. Identification of specific pathogenic mutations and genetic characteristics in neurofibromatosis patients is of great significance for the prevention of scoliosis.

Whole-exome sequencing (WES) allows for the sequencing of more than 4,000 genes that currently have known biological functions. Exons, which comprise about 1.5% of the genome, account for approximately 85% of the pathogenic variants (15, 16). WES exhibit several benefits over other genomic techniques, including lower cost, higher throughput, and greater diagnostic potential. This method has wide coverage and can comprehensively screen for pathogenic genes related to skeletal developmental abnormalities. Research has shown that WES technology improves the diagnostic rate of fetal skeletal system diseases (17, 18). With decrease in sequencing costs, its clinical applications have become increasingly widespread in the recent years.

Therefore, this study conducted WES of neurofibromatosis patients with scoliosis, combined with relevant imaging data, to explore specific gene mutations and provide a theoretical basis for the prevention and treatment of scoliosis.

In 1987, the National Research Institute of the United States proposed seven diagnostic criteria for NF-1, including skin milky coffee spots, neurofibromas, inguinal freckles, optic gliomas, Lisch nodules, characteristic skeletal change, and family history (25). Due to some diagnostic features changing with age, the international NF-1 diagnostic standard has some limitations, which may lead to delayed diagnosis in some patients. Spinal scoliosis, as the most common bone system lesion in NF-1, can be diagnosed on changes in patient’s physical appearance clinical examination, typically presenting during the first peak period between 5 and 10 years of age (26). According to previous studies, the incidence of NF spinal deformities is 10–64% with the faster progression rate of deformities than those of AIS (27). Therefore, scoliosis screening should be performed in all pre-adolescent children with NF-1. Due to the significant differences in treatment and prognosis between the two types of scoliosis associated with NF-1, it is vital to make timely diagnosis on the types of scoliosis.

Patients with NF-1 accompanied by scoliosis can be divided into malnourished and non-malnourished based on imaging features. Non-malnourished scoliosis typically presents with long, smooth curves. In contrast, malnourished scoliosis, compared with AIS, typically involves four to six vertebral bodies and is characterized by short and sharp lateral curvature, with or without angular kyphosis. In addition, the undernourished scoliosis, which often develops gradually in young age, is accompanied by pencil-shaped changes of ribs, vertebral rotation, vertebral wedge-shaped changes, and lateral curvature of atypical positions (28). Li et al. measured 2,652 pedicle abnormalities of 56 NF-1 patients with dystrophic scoliosis and 22 NF-1 patients with non-dystrophic scoliosis (29). The results showed that the overall incidence of pedicle abnormalities in NF-1 patients was as high as 67%, and the upper thoracic spine (87%) was the most concentrated area of abnormal pedicle abnormalities. In addition to scoliosis, patients with neurofibromatosis often have abnormalities in the skin, blood vessels, eyes, and nervous system. Approximately 20–50% of NF-1 patients have osteoporosis, which is associated with alterations in bone transformation biochemical markers, including decreased serum 25-hydroxyvitamin D and elevated serum PTH levels (30). In this study, one patients also experienced changes in multiple imaging indicators, such as intraspinal rib head(T9, T10),a hemivertebra(T9,)and abnormal pedicles(T9, T10). The accurate recognition of these radiological features is beneficial for early diagnosis of patients with malnourished scoliosis to develop timely and appropriate treatment plans.

With the development of modern diagnostic technology, NF-1 can be diagnosed early through genetic testing. In 2021, the International Consensus Group of Diagnostic Standards for Neurofibromatosis proposed revisions of the NF-1 diagnostic standard (1987) mainly incorporating genetic diagnosis (31). Genetic testing was able to confirm diagnosis before clinical diagnosis (or before the appearance of the second clinical feature), which could help screen out children with atypical features. The NF-1 gene has been recognized as the pathogenic factor for NF-1 in the medical community. However, the correlation between different mutation types and sites of NF-1 and the clinical phenotype of NF-1 patients has always been a hot research topic in the medical community. Although scoliosis is the most common type of skeletal changes in NF-1, there is limited research on the correlation between specific mutation types and mutation sites in NF-1 patients with scoliosis.

In this study, we conducted WES on 18 individuals from seven probands and their immediate relatives to explore gene variants associated with NF-1. As previous studies have shown that mutations in the NF-1 gene lead to the occurrence of NF-1, we performed family-based analysis to identify the variants of NF-1 gene associated with phenotypes and characterized their effects on the amino acid sequence and structure (32). The NF-1 gene codes for neurofibromin, a protein of 2,818 amino acids that negatively regulates the RAS pathway, a signaling cascade involved in cell growth and differentiation (33). Neurofibromin acts as a GAP for Ras, a family of small GTP-binding proteins that are key regulators of the RAS pathway. A 360 amino-acid segment of neurofibromin, called the GAP-related domain (GRD), enhances the GTP hydrolysis of Ras, thereby inactivating it (34). NF-1 mutations impair the function of neurofibromin, resulting in the accumulation of active Ras and the hyperactivation of the RAS pathway, which underlies the tumorigenesis and other phenotypes of NF-1. We identified two mutations located in the GRD region in patients 1(c.4445 T > C, p. Ile1482Thr) and 6(c.4084C > T, p. Arg1362Ter). c.4445 T > C is a single nucleotide variation located at chr17:31260383 with a minor allele frequency of 0.000024, resulting in a missense mutation of isoleucine to threonine at codon 1,482. This SNV is associated with not only NF-1 (NF-1) but also hereditary cancer predisposition syndrome. c.4084C > T is located at chr17:31249093 with a minor allele frequency of 0.000008, resulting in a premature stop codon in the mRNA transcript and leading to a truncated protein product. Eisenbarth et al. (35) identified c.4084C > T as a tumor-specific point mutation in seven neurofibromas from four different NF-1 patients and demonstrated that loss of function of NF-1 gene is a common mechanism in tumorigenesis. Hence, variants in the GRD may impair the function of neurofibromin and contribute to the genetic predisposition for neurofibromatosis.

In the family-based analysis, we identified c.1885G > A in the NF-1 gene in patient 3 and his sibling, both of whom exhibited NF-1 phenotypes, and this variant was absent in their normal parents. Furthermore, patient 3 uniquely presented with vertebral body dysplasia, indicating that c.1885G > A is a genetic factor that contributes to the development of NF-1. c.1885G > A is an SNV at chr17:31249093 with a minor allele frequency of 0.000024, resulting in a nonconservative amino acid substitution in the CSRD of neurofibromin. The CSRD of neurofibromin undergoes phosphorylation by protein kinase A (PKA) and protein kinase C (PKC). The PKC-mediated phosphorylation of the CSRD modulates the conformation and function of neurofibromin, a Ras-GAP, and promotes its interaction with actin. In our cohort study, none of the studied families had consanguineous marriage, suggesting that the identified variants were not attributable to consanguinity.

Through a one-sided analysis between the patient and control group, we detected five NF-1 related pathogenic variants with statistically significant frequency. We observed that two variants, rs2916067 and rs80221306, were present in all the individuals with NF-1 in our study. Variant rs2916067 is an SNV in the coding region of MAGEF1, a ubiquitin ligase enhancer and a member of the melanoma-associated antigen (MAGE) super family (36). MAGEA2 is overexpressed in various cancers and suppresses the RAS pathway. Correspondingly, loss-of-function mutations in the NF-1 gene inhibit the RAS pathway, impair bone development, and lead to skeletal deformities such as scoliosis and pseudarthrosis. Variant rs80221306 is located at gene filaggrin (FLG), which affects differentiation of keratinocytes (37). The dysregulation of the keratinocyte–melanocyte axis results in the formation of café-au-lait macules and malignant melanoma in NF-1, but the underlying molecular basis of this dysregulation remains elusive. The FLG gene might play a role in facilitating phenotypic expression.

There are still some limitations that need to be addressed. First, the small sample size of this study reduces the statistical power. However, the NF-1 is quite rare condition, and only a few of patients present with the dystrophic scoliosis. In addition, WES is an expensive test for patients, which limit the sample size. Second, all participants in the present study were from a single ethnic group, which limits the generalizability of the findings to the broader human population. Third, the study would have been more robust if genetically healthy humans were considered as the control group. Another limitation of this study is that we did not perform functional assays to validate the biological effects of the identified NF-1 variants, and further experimental studies are required to confirm their pathogenicity.

In our cohort study, NF-1 patients with dystrophic scoliosis predominantly exhibited upper thoracic involvement, and surgical correction achieved satisfactory outcomes, with a maximum postoperative Cobb angle correction rate of 85.3% and no loss of correction during follow-up. The c. G1885A variant in the NF-1 gene may influence the phenotypic severity of the disease, and rs112819846, rs2916067, and rs80221306 may be the disease-related pathogenic variants of NF-1 patients with dystrophic scoliosis.