Steel syndrome (OMIM# 615155) is a rare autosomal recessive skeletal dysplasia, caused by biallelic pathogenic variants in COL27A1 gene. The main clinical and radiological features include congenital hip and radial head dislocation, carpal coalition, acetabular dysplasia, short stature, and craniofacial dysmorphism. It was first delineated by Steel et al. in 1993, who reported a cohort of 23 Puerto Rican patients exhibiting similar clinical manifestations (1). Later, Flynn et al. described an additional 14 patients from the same population, and re-examined 18 subjects from Steel's cohort (1, 2). The presence of hip and radial head dislocations combined with joint laxity allowed for the inclusion of Larsen and Ehlers-Danlos syndromes in the differential diagnosis, while short stature suggested achondroplasia and pseudoachondroplasia. However, using targeted approaches to identify causative variants in disease-related genes, such as COL5A1, COL1A1, COMP, and FGFR3, did not yield any results (2). In 2015 Gonzaga-Jauregui et al. identified potential founder homozygous pathogenic variant c.2089G > C p. (Gly697Arg) in COL27A1 causing Steel syndrome in a Puerto Rican family through exome sequencing (3). To date, approximately 66 patients with STLS have been reported in the literature, 51 of whom are of Puerto Rican origin (1, 2, 4). Additionally, two fetuses from one Caucasian family with suspected STLS were described by Frigola et al. (5). Molecular genetic testing was performed on a total of thirty-six individuals, including two fetuses. Among them eighteen were of Puerto Rican origin and one Hispanic, with all cases homozygous for the p.Gly697Arg variant, consistent with a founder effect (3, 6–8). In non-Puerto Rican patients twenty nucleotide variants were found, six of which resulted in Gly amino acid substitutions located in the triple-helical domain. Currently, two acceptor splice-site variants: c.3556 – 2A > G, c.4261 – 1G > A and one donor splice-site variant c.3249 + 1G > T have been reported in three unrelated families of Yemeni, Emirati, and Caucasian origin (5, 9, 10). Gradually accumulating data continue to expand the spectrum of not only radiological but also clinical manifestations of Steel syndrome, including various extraskeletal features such as sensorineural hearing loss, coloboma, genitourinary anomalies, and developmental delay (11–13).

Here, we report two pediatric cases of Steel syndrome: one associated with two novel compound heterozygous splice variants, and the other with a novel homozygous missense variant in COL27A1 gene. We also explore potential genotype–phenotype correlations based on a review of clinical and radiological features in all reported cases with pathogenic variants in COL27A1.

Blood samples were collected from the probands and their unaffected parents. Genomic DNA was extracted using standard methods. Whole-exome sequencing (WES) was used to perform DNA diagnosis in the probands. RT-PCR followed by deep sequencing was performed to analyze the mRNA structure of the two novel variants, using RNA extracted from the skin fibroblasts of proband 1 and his mother. Statistical analysis was performed to evaluate the prevalence and significance of clinical and radiological features reported in patients with Steel syndrome. Detailed description of the clinical, genetic, functional, and statistical methods used to analyze reported patients with Steel syndrome is provided in Supplementary Material 1.

A detailed clinical and radiological data were obtained from two unrelated patients. All participants gave informed consent to the clinical and radiological examination and the publication of their anonymized data.

We report two novel cases of Steel syndrome in pediatric patients of Russian and Uzbek origin. The clinical manifestation of STLS in our patients included hallmark features such as acetabular dysplasia or congenital hip dislocation, radial head dislocation, carpal coalition and facial dysmorphism. In P1, the disease was associated with two novel splice variants in the COL27A1 – c.2619 + 1G > A and c.2673 + 4A > G, causing skipping of exons 17 and 18, confirmed by RT-PCR followed by deep sequencing. In P2 a homozygous missense variant c.3988G > C p. (Gly1330Arg) located in the triple-helical domain of the protein was identified.

COL27A1 gene is located on human chromosome 9q32–33 and consists of 61 exons. The proα1(XXVII) protein, encoded by this gene, is 1860 amino acids long and like other proteins belonging to this large family contains a triple-helical domain composed of repeating Gly-Xaa-Yaa motifs (16). It is significantly expressed in cartilage, eye, inner ear structures, bronchial and lung epithelium, ameloblasts in developing teeth, as well as major heart arteries during early embryonic development, which may explain the presence of extraskeletal features such as sensorineural hearing loss, crowded teeth, septal defects, and partial syndactyly in patients with STLS (6, 16). However, in later developmental stages and throughout adulthood, its expression remains predominantly restricted to cartilage, especially within the growth plate, which undergoes constant renewal during endochondral bone growth (16–18).

The majority of patients with STLS reported to date are of Puerto Rican origin. Among them, 19 individuals, including one patient of Hispanic origin, were found to have the homozygous variant p.Gly697Arg in the COL27A1 (3, 6–8). This Gly amino acid substitution is located in the triple-helical domain of the protein (6, 16). Interestingly, most diseases associated with Gly amino acid substitutions within this domain of other collagens follow an autosomal dominant mode of inheritance, demonstrating a dominant-negative effect by impairing proper protein folding and leading to endoplasmic reticulum stress (19, 20). This results from the fact that only glycine is small enough to fit into the center of the assembled triple-helical structure without disrupting its stability. Therefore, any deviation from the conserved glycine residues in vertebrate fibrillar collagens often leads to structural instability and disease (16). However, COL27A1 as a member of type C clade of vertebrate fibrillar collagens, along with COL24A1, may be more resistant to disruptions of the Gly-Xaa-Yaa motif particularly regarding folding and secretion, which suggests that this mechanism is unlikely to play a significant role in the development of COL27A1– related skeletal dysplasia (6, 17, 21).

To date, twenty one pathogenic nucleotide variants distributed throughout the COL27A1 gene have been reported (Figure 3; Supplementary Material 2 Table S4) (3, 4, 6, 9–13, 22–24).

Approximately half of the reported variants are frameshift or nonsense, likely leading to protein truncation, consequent degradation and loss of function. The variant c.3556 – 2A > G identified in the Emirati patient with STLS was shown to disrupt splicing by activating alternative cryptic splice sites. This resulted in the production of three distinct transcripts, each leading to a different deletion. Based on these findings, the authors suggested that the truncated protein may lead to aberrant folding and could be targeted for subsequent degradation, that may be more damaging than previously reported missense variants (9). Indeed, a 9-year-old Syrian patient with compound heterozygous variants p.Phe32LeufsTer71 and p.Lys1026ArgfsTer33 exhibited profound short stature (85.5 cm, – 7.9 SD) along with extraskeletal features including early-onset SNHL, iris coloboma, and developmental delay (13). While, among all individuals with normal (less than −2 SD) or slightly decreased height, missense variants, typically homozygous Gly substitutions in the triple-helical domain were detected, except for two sibs from Satoh et al. study, where the compound heterozygous variants p.Gly676Arg and p.Met788ValfsTer34 were identified (Supplementary Material 2 Tables S1–S2) (8, 11, 22). However, the presence of a frameshift variant on one of their alleles may have contributed to their clinical presentation. STLS phenotype observed in P1 was associated with in-frame deletions p.Gly874_Leu891del and p.Gly856_Lys873del, located within the triple helical domain. Disease presentation in P1 was characterized by the absence of short stature (−1.75 SD) and hip dislocation combined with extraskeletal features (Table 1). This suggests that in-frame deletions likely cause misfolding, leading to a less profound functional defect than truncating variants such as frameshift or nonsense, possibly explaining the milder skeletal phenotype and supporting the idea that COL27A1 may be more resistant to disruptions of the Gly-Xaa-Yaa motif (6, 17, 21). Since precise genotype-phenotype correlations remain undefined in Steel syndrome, we conducted a comprehensive review of clinical and radiological manifestations in all reported cases (Table 1).

Phenotypic analysis of STLS cases revealed features that appear to be characteristic of particular patient subgroups. Interestingly, pes cavus deformity and carpal coalition are most common in individuals of Puerto Rican origin, while equinovarus or planovalgus deformities and hearing loss are more often noted in other populations (Table 1) (4, 6, 11, 23). Furthermore, certain features, such as cervical spine abnormalities and associated complications such as cervical stenosis and cervical cord compression, likely manifest only during adulthood (2, 6, 7).

Recently, it was suggested that hearing loss in individuals with STLS associated with the homozygous founder allele p.Gly697Arg may occur later, after the age of 30. In contrast, individuals with more ‘severe' alleles experience the early-onset hearing loss (6). Indeed, three Puerto Rican individuals homozygous for the p.Gly697Arg variant were diagnosed with hearing loss in adulthood (6). According to reported data, 7 out of 16 patients with early-onset hearing loss carried variants predicted to result in premature translation termination and possible loss of function. Two individuals, including P2, were diagnosed with conductive hearing loss unrelated to STLS (6). Patients with the homozygous p.Gly802Glu, p.Gly904Arg, and p.Gly841Arg variants, as well as P1 with in-frame deletions within the triple helical domain, were diagnosed with early-onset hearing loss (Supplementary Material 2 Tables S1–S3). These variants are predicted to cause protein misfolding. Early-onset hearing loss was also diagnosed in three additional patients: two with compound heterozygous variants p.Ala99Thr and p.Pro1019His and one with a homozygous splice variant c.3556 – 2A > G (4, 9). The analysis indicates that the age of hearing loss onset is not correlated with the location or type of genetic variant. However, this hypothesis requires further validation through the accumulation of additional clinical patient data.

Notably, P1 presented with extraskeletal manifestations, including SNHL, mild developmental delay, and cryptorchidism possibly associated with prematurity, but did not exhibit the classic diagnostic criteria of Steel syndrome—congenital hip dislocation and short stature (Figure 1). Despite initial data on obligatory presence of hip dislocation, there is more evidence of less specific importance of this condition as a part of the syndrome. Thus, in both the original Steel et al. study and the subsequent Flynn et al. study hip dislocations were a universal finding (Table 1) (1, 2). Surprisingly, the first European patient reported by Kritioti et al. had no hip dislocations but only minor stable hip changes, radiographically similar to our P1 case (22). Kim et al. reported a Korean patient presented with bilateral hip dysplasia and pointed out that hip abnormalities in STLS may vary from acetabular dysplasia to complete dislocation (12). These observations suggest that the ‘classic' phenotype may be predominantly linked to Gly substitutions in the triple helical domain. This is supported by our findings, particularly in P2, who exhibited the complete spectrum of STLS criteria as defined by Steel et al., while lacking extraskeletal manifestations typically observed in non-Puerto Rican individuals (Table 1) (1). Hand radiographs demonstrated abnormal ossification patterns characterized by carpal bone coalescence in both patients. While the whole extent of fusion was challenging to evaluate due to the early ossification stage, typical capitate-hamate coalition was easily recognizable in both patients. One more interesting finding in P2 was pseudoepiphyses of the second metacarpals (Figure 1). Multiple pseudoepiphyses of the metacarpals and metatarsals may be seen in otherwise healthy children but can also occur in patients with skeletal dysplasias (25). Radiographic analysis demonstrated that our patients exhibited the same ossification pattern as the Indian monozygotic twins with STLS reported by Girisha et al. in 2022, but this phenomenon was not mentioned (4).

A comprehensive review of symptoms prevalence across reported patients including P1 and P2 from present study highlighted core features occurring more frequent among all STLS patients such as short stature, congenital hip and radial dislocation, carpal coalition and facial dysmorphism (Supplementary Material 2 Figure S1). These symptoms represent a core cluster of STLS skeletal features that may facilitate early diagnostic suspicion and help to avoid unnecessary surgical procedures for hip dislocation correction (2). However, comparison across Puerto Rican patients and individuals from other populations revealed phenotypic variability, including missing hallmark features like short stature or hip dislocation and increased incidence of extraskeletal manifestations, which should also be considered (Table 1).

In conclusion, we report two pediatric cases of STLS associated with novel biallelic variants in the COL27A1. This study provides the first characterization of two splice donor site variants confirming their effect on splicing. Review of reported clinical data enabled identification of a phenotypic cluster characteristic of STLS. Furthermore, our findings indicate that the presence of protein-truncating variants may be associated with more severe disease course characterized by extraskeletal manifestations, developmental delay and pronounced short stature. In contrast, missense variants, particularly glycine substitutions within the triple-helical domain, are more often associated with ‘classic' phenotype initially reported by Steel et al. (1). To date, most reported cases have involved Puerto Rican patients with a homozygous founder allele. Therefore, expanding the genetic and phenotypic spectrum of STLS in diverse populations and age groups will be essential for improving diagnostic accuracy and clinical management.