Molecular and clinical findings in Osteogenesis Imperfecta: A cohort study from a single tertiary center

Abstract

OBJECTIVE

Osteogenesis imperfecta (OI) is a clinically and genetically heterogeneous skeletal fragility disorder. Although COL1A1/COL1A2 variants account for most cases, non-COL1A genes and overlapping monogenic bone fragility conditions complicate diagnosis and genotype–phenotype interpretation in routine practice. We aimed to describe the clinical spectrum and molecular findings of patients evaluated for suspected or confirmed OI at a tertiary referral center, and to assess phenotype patterns across genotype-informed groups.

METHODS

We conducted a single-center retrospective cohort study at Ankara City Hospital. Pediatric and adult patients evaluated for suspected/confirmed OI were included. Clinical, radiologic, and treatment data were extracted from medical records. All patients underwent targeted next-generation sequencing panel testing, and variants were interpreted according to ACMG criteria. Findings were summarized descriptively and analyzed across COL1A1/COL1A2, non-COL1A OI-associated genes, and non-OI bone-related genes.

RESULTS

Forty-two individuals were included (age range at genetic evaluation: 0.1–47 years). The most common referral indication was recurrent long-bone fractures (81.0%), while 31.0% presented with or additional vertebral compression fractures and/or osteoporosis. Variants in OI-associated genes were identified in 30/42 patients (71.4%); 23/42 (54.7%) had molecularly confirmatory pathogenic/likely pathogenic variants in established OI-associated genes. Among genetically confirmed OI cases, 17/23 (73.9%) involved COL1A1/COL1A2. In the non-COL1A subgroup, biallelic pathogenic/likely pathogenic variants were found in CRTAP, P3H1, and FKBP10 (n=5), and one patient had a heterozygous WNT1 variant. Seven patients (16.7%) had negative panel results. At last follow-up, 83.3% were independently ambulatory.

CONCLUSION

In this tertiary-center cohort, integrating phenotype with molecular testing improved diagnostic stratification in suspected OI and highlighted the predominance of COL1A-related disease. However, the continued clinical relevance of VUS and panel-negative results underscores the need for broader second-tier genetic testing and periodic reanalysis in cases where clinical suspicion persists.

Highlight key points

• In this tertiary-care OI cohort, targeted sequencing provided a molecular diagnosis in more than half of patients, supporting its use as a practical first-line test in routine clinical evaluation.

• The predominance of COL1A1/COL1A2 variants indicates that COL1A-focused interpretation remains central for rapid diagnostic decision-making in most suspected OI cases.

• Detection of pathogenic variants in non-COL1A genes (e.g., CRTAP, P3H1, FKBP10, WNT1) shows that relying only on classic COL1A phenotypes may miss clinically relevant subgroups and delay accurate classification.

• The presence of VUS and panel-negative cases confirms that a negative targeted panel does not exclude inherited bone fragility; these patients require phenotype-guided follow-up, periodic reanalysis, and, when appropriate, broader genomic testing.

• Overall, a clinicogenetic approach (clinical phenotype + molecular data) improves diagnostic stratification and directly supports treatment planning, prognosis assessment, and family counseling.

Osteogenesis imperfecta (OI) comprises a genetically and phenotypically heterogeneous group of skeletal disorders primarily characterized by bone fragility. The reported incidence ranges from approximately 1 in 15,000 to 1 in 20,000 live births [1, 2]. Clinical severity spans a wide spectrum, from mild disease with recurrent low-trauma fractures and little or no deformity to severe forms marked by progressive long-bone deformities, vertebral compression fractures, growth impairment, scoliosis, chronic pain, and substantial functional limitation. In the most extreme cases, OI is perinatally lethal. Extra-skeletal features such as blue sclerae, dentinogenesis imperfecta, hearing loss, joint hypermobility, and other connective-tissue features may accompany the skeletal phenotype and vary widely across individuals [3, 4]. At the molecular level, OI is best understood as a disorder of type I collagen and related pathways involved in extracellular matrix formation and bone mineralization. Type I collagen is a heterotrimer composed of two pro-α1 chains and one pro-α2 chain, encoded by COL1A1 and COL1A2, respectively. Pathogenic variants in COL1A1 and COL1A2 account for approximately 85% of OI cases. Variants other than COL1A1 and COL1A2 genes contribute a substantial minority of cases and may affect multiple steps of collagen biosynthesis and matrix homeostasis, including triple-helix formation, collagen folding (notably via the prolyl-3-hydroxylation complex), post-translational processing and cross-linking, and osteoblast differentiation and function. The expanding molecular data has also highlighted collagen-independent mechanisms, some associated with distinct abnormalities of matrix organization and mineralization [4–7]. Although traditional clinical classifications (e.g., Sillence phenotypes) remain useful in practice, they do not fully capture biological heterogeneity: similar clinical presentations may arise from different molecular mechanisms, and variants within the same pathway may produce markedly variable phenotypes. Accordingly, contemporary approaches increasingly emphasize genotype and pathomechanism based grouping, which may better inform prognosis, multidisciplinary surveillance, and genetic counseling. In children, recurrent long-bone fractures with minimal trauma, vertebral compression fractures, family history, and supportive extra-skeletal features provide key diagnostic clues; however, the breadth of the clinical spectrum means that diagnosis particularly in milder forms often relies on a high index of suspicion [5, 8].

In this single center retrospective cohort study, we aimed to describe the clinical spectrum and molecular etiology of patients with OI followed at our center and to examine whether genotype-informed groupings are associated with clinically meaningful phenotypic patterns.

MATERIALS AND METHODS

Study Design and Setting

This study is a single-center, retrospective cohort study conducted at Ankara City Hospital, a tertiary referral center in Turkiye. Medical records of all patients followed for suspected or confirmed OI at the pediatric endocrinology, pediatric genetics, and medical genetics, clinics were systematically reviewed between May and November 2022.

Participant Selection

Pediatric and adult patients of any age were eligible for inclusion if they were evaluated or followed at our center for suspected or confirmed OI and had sufficient clinical documentation to support diagnostic confirmation and phenotype characterization. Patients were included if they met at least one of the following criteria: (a) identification of a pathogenic or likely pathogenic variant in an OI-associated gene; (b) a clinical phenotype strongly suggestive of OI (e.g., recurrent low-trauma fractures and/or typical extra-skeletal features such as blue sclerae or dentinogenesis imperfecta); or (c) radiologic findings consistent with OI (e.g., vertebral compression fractures and/or long-bone deformities) documented in imaging reports. Exclusion criteria were the presence of an alternative diagnosis explaining bone fragility (e.g., untreated rickets/osteomalacia, osteopetrosis, primary hyperparathyroidism, or prolonged high-dose glucocorticoid exposure without supportive evidence of OI) and insufficient records for core outcomes, predefined as missing documentation of fracture history and key phenotype features.

Data Collection

Data were retrospectively extracted from electronic medical records, outpatient clinic notes, hospital discharge summaries, radiology reports, laboratory results, and genetic testing reports. A standardized case report form was used to ensure systematic and uniform data collection. The following variables were recorded:

• Demographics: Sex, date of birth, age at first presentation to our center, age at OI diagnosis, family history of OI or unexplained bone fragility, and parental consanguinity.

• Clinical phenotype: Anthropometric measurements (height and weight at last visit; height standard deviation score (SDS) according to age- and sex-specific references when available). Fracture history: Total number of fractures, number of long-bone fractures, timing of fractures (prenatal, perinatal, or postnatal periods when documented), and presence of vertebral compression fractures. Skeletal deformities: Long-bone bowing or deformities, scoliosis (measured Cobb angle when available), kyphosis, basilar invagination, and platybasia. Functional status: Mobility level (independent ambulation, use of assistive devices, or wheelchair dependence) and overall functional impairment.

• Extra-skeletal features: Presence and severity of blue or gray sclerae, dentinogenesis imperfecta (confirmed by dental examination when available), hearing loss (sensorineural, conductive, or mixed; age at onset), joint hypermobility, easy bruising, and other connective tissue manifestations (e.g., skin hyperextensibility, aortic root dilatation).

• Imaging and bone densitometry: Key radiographic findings (e.g., wormian bones, thin cortices, multiple vertebral compression fractures, or characteristic long-bone deformities). DXA results were recorded when available, including lumbar spine and/or total body less head bone mineral density (BMD) expressed as age- and sex-specific Z-scores in children and adolescents, or T-scores and Z-scores in adults.

• Treatment history: Exposure to OI-specific therapies, including type of bisphosphonate (pamidronate, zoledronic acid, or oral agents), route of administration, age at initiation, and total cumulative dose or duration. Use of calcium and/or vitamin D supplementation, orthopedic interventions (e.g., intramedullary rodding, spinal fusion, corrective osteotomies), physiotherapy, and any investigational or off-label therapies were also documented.

Data extraction was performed by two investigators independently, with discrepancies resolved by consensus or consultation with a senior clinician.

Genetic Testing and Variant Interpretation

Targeted next-generation sequencing panel encompassing genes most frequently associated with OI was applied to all patients. The genes included in the targeted panel were categorized into three subgroups based on their molecular and clinical relevance to OI. COL1A1/COL1A2 subgroup comprised COL1A1 and COL1A2, while the non-COL1A osteogenesis imperfecta-associated genes subgroup included IFITM5, P3H1, CRTAP, PPIB, SERPINH1, SERPINF1, FKBP10, TMEM38B, PLS3, BMP1, SP7, ALPL, CREB3L1 and WNT1. Finally, the non-OI bone-related genetic findings subgroup consisted of LRP5 and PLOD2. Sequence data were analyzed using a standardized bioinformatics pipeline and detected variants were annotated and intrepreted in accordance with the American College of Medical Genetics and Genomics (ACMG) [9]. Targeted next-generation sequencing (NGS) panel encompassing genes most frequently associated with OI was applied to all patients. Target enrichment was performed using the IDT xGen Custom Hybridization Capture Panel (Integrated DNA Technologies, Coralville, IA, USA), designed to capture all coding exons and exon–intron boundaries of the selected genes. Genomic DNA libraries were prepared using the QIAseq Fx DNA Library Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions. Following enrichment, paired-end sequencing was performed on the Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, USA).

Statistical Analysis

Continuous variables were summarized as mean (SD) or median (IQR) depending on distribution; categorical variables were summarized as counts and percentages. Statistical analyses were performed using the IBM Statistical Package for the Social Sciences, version 25.0 (SPSS Inc., Armonk, NY, IBM Corp., USA). The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and applicable international ethical standards. Ethical approval was obtained from the Ethics Committee of the Ankara City Hospital (approval no. E2-22-1814; approval date: May 11, 2022). Given the retrospective design and the use of routinely collected clinical data, the requirement for informed consent was waived in accordance with local regulations. All patient data were deidentified prior to analysis.

RESULTS

Patient Characteristics

A total of 42 individuals evaluated for suspected or confirmed OI were included in the cohort. Genetic testing (targeted panel) was performed in all patients. Variants in OI-associated genes were identified in 30 patients (71.42%), of whom 23 carried pathogenic or likely pathogenic variants according to ACMG guidelines in established OI-associated genes. Among these, 17 variants were in COL1A1 or COL1A2. Seven patients (16.66%) had negative genetic testing results. The age at first presentation to our center for genetic evaluation was 11.8±13.3 years (range, 0.1–47). Females comprised 38.09% of the cohort (n=16). Consanguinity was reported in five families. The primary reason for referral was recurrent long-bone fractures in 80.95% (n=34), while 30.95% (n=13) presented primarily or concomittley with vertebral compression fractures and/or osteoporosis without a clear history of low-trauma long-bone fractures. At the last follow-up, 83.33% of patients (n=35) were independently ambulatory, whereas 16.66% (n=7) required assistive devices or wheelchair support.

COL1A1/COL1A2 Subgroup

Among the 17 patients with molecularly confirmed OI, pathogenic or likely pathogenic were identified in the COL1A1 gene in 10 patients, in the COL1A2 gene in 7 patients. The median age at the time of genetic testing in this subgroup was 8.4±11.7 years (range: 0.1–43 years), with females comprising 35.29% (n=6). The most common presenting feature was recurrent long-bone fractures in all patients, while three patient (17.64%) presented with additional vertebral compression fractures and/or osteoporosis without prominent long-bone fractures. At last follow-up, 16 patients (94.1%) in the COL1A subgroup were independently ambulatory, whereas one (5.9%) required assistive devices or wheelchair support. Variant classification was performed by the reporting clinical laboratory according to and Association for Molecular Pathology (AMP) guidelines, with additional review supported by public databases (e.g., ClinVar, gnomAD). In the COL1A1 group, 8 variants were classified as pathogenic and 2 variants as likely pathogenic. In the COL1A2 group, 7 variants were pathogenic. Detailed variant information, including nucleotide and protein changes is provided in Table 1.

Table 1.

Variant-level information for the COL1A subgroup

Study ID	Gene	Transcript	cDNA (HGVS)	Protein (HGVS)	Zyg.	ACMG	Type (missense-frameshift, splice…)
COL1A-01	COL1A1	NM_000088.4	c.1056+1G>A	—	Het	Pathogenic	Splice site
COL1A-02	COL1A1	NM_000088.4	c.658C>T	p.Arg220*	Het	Pathogenic	Nonsense
COL1A-03	COL1A1	NM_000088.4	c.2037_2038delAG	p.Gly680fs*29	Het	Pathogenic	Frameshift
COL1A-04	COL1A1	NM_000088.4	c.3494delC	p.Pro1165fs*74	Het	Pathogenic	Frameshift
COL1A-05	COL1A1	NM_000088.4	c.1128delT	p.Gly377fs*164	Het	Pathogenic	Frameshift
COL1A-06	COL1A1	NM_000088.4	c.1792C>T	p.Arg598*	Het	Pathogenic	Nonsense
COL1A-07	COL1A1	NM_000088.4	c.2775delT	p.Gly926fs*182	Het	Pathogenic	Frameshift
COL1A-08	COL1A1	NM_000088.4	c.3505G>A	p.Gly1196Ser	Het	Pathogenic	Missense
COL1A-09	COL1A1	NM_000088.4	c.4292C>T	p.Thr1431Ile	Het	Likely Pathogenic	Missense
COL1A-10	COL1A1	NM_000088.4	c.802delA	p.Arg268fs*273	Het	Likely Pathogenic	Frameshift
COL1A-11	COL1A2	NM_000089.4	c.2521G>A	p.Gly841Ser	Het	Pathogenic	Missense
COL1A-12	COL1A2	NM_000089.4	c.1667G>A	p.Gly556Asp	Het	Pathogenic	Missense
COL1A-13	COL1A2	NM_000089.4	c.3116G>T	p.Gly1039Val	Het	Pathogenic	Missense
COL1A-14	COL1A2	NM_000089.4	c.2305G>T	p.Gly769Cys	Het	Pathogenic	Missense
COL1A-15	COL1A2	NM_000089.4	c.983G>A	p.Gly328Asp	Het	Pathogenic	Missense
COL1A-16	COL1A2	NM_000089.4	c.964G>A	p.Gly322Ser	Het	Pathogenic	Missense
COL1A-17	COL1A2	NM_000089.4	c.1378G>A	p.Gly460Ser	Het	Pathogenic	Missense

ID: Identifier; HGVS: Human Genome Variation Society variant nomenclature; ACMG: American College of Medical Genetics and Genomics (variant classification); Zyg: Zygosity; Het: Heterozy-gous; fs: Frameshift. NM_… indicates the NCBI RefSeq transcript accession (version shown after the dot). c. denotes coding DNA (cDNA) change; p. denotes predicted protein change.

Non-COL1A OI Associated Genes Subgroup

Variants associated previously with OI other than COL1A genes were identified in 13 individuals. Because most non-COL1A OI genes classically follow an autosomal recessive inheritance pattern, interpretation in this subgroup prioritized biallelic pathogenic/likely pathogenic variation as the key criterion for molecular confirmation.

Using gene-associated OI type labels (historical/OMIM-based), biallelic involvement was documented in 5 patients: CRTAP (OI type VII; n=2), P3H1 (OI type VIII; n=2), and FKBP10 (OI type XI; n=1). In addition, one patient carried a heterozygous WNT1 variant (OI type XV), consistent with the reported autosomal dominant bone fragility spectrum of this gene. The remaining patients had monoallelic variants, including SERPINH1 (OI type X; n=2), SERPINF1 (OI type VI; n=1), FKBP10 (OI type XI; n=2), and BMP1 (OI type XIII; n=2). Variant-level details are presented in Table 2.

Table 2.

Variant-level details for non-COL1A OI-associated genes

Study ID	Gene / OI subtype	Transcript	HGVS cDNA	HGVS protein	Zygosity	ACMG class	Type (missense-frameshift, splice…)
NCOI-01	SERPINF1 / Type-6	NM_002615.7	c.893T>C	p.Ile298Thr	Het	VUS	Missense
NCOI-02	SERPINH1	NM_001207014.3	c.1222C>T	p.Arg408Trp	Het	VUS	Missense
NCOI-03	SERPINH1	NM_001207014.3	c.623C>T	p.Pro208Leu	Het	VUS	Missense
NCOI-04	CRTAP / Type-7	NM_006371.5	c.535G>T	p.Glu179*	Hom	P	Nonsense
NCOI-05	CRTAP	NM_006371.5	c.471+3A>G	—	Hom	VUS	Intronic
NCOI-06	P3H1 / Type-8	NM_001243246.2	c.1345+1G>T	—	Hom	LP	Splice site
NCOI-07	P3H1	NM_022356.4	c.2154dupC	p.Glu719fs*11	Hom	P	Frameshift
NCOI-08	FKBP10 / Type-11	NM_021939.4	c.890_897dupTGATGGAC	p.Gly300*	Het	P	Nonsense
NCOI-09	FKBP10	NM_021939.4	c.616G>A	p.Gly206Ser	Het	VUS	Missense
NCOI-10	FKBP10	NM_021939.4	c.603T>A	p.Tyr201*	Hom	LP	Nonsense
NCOI-11	BMP1 / Type-12	NM_006129.5	c.17G>C	p.Arg6Pro	Het	VUS	Missense
NCOI-12	BMP1	NM_006129.5	c.2870A>T	p.Lys957Met	Het	VUS	Missense
NCOI-13	WNT1 / Type-15	NM_005430.4	c.691_695delACGTG	p.Thr231fs*84	Het	LP	Frameshift

ID:Identifier; HGVS: Human Genome Variation Society variant nomenclature; ACMG: American College of Medical Genetics and Genomics (variant classification); Zyg: zygosity; Het: Heterozy-gous; Hom: Homozygous. NM_… indicates the NCBI RefSeq transcript accession (version shown after the dot). c. denotes coding DNA (cDNA) change; p. denotes predicted protein change.

Non-OI Bone-related Genetic findings (LRP5/PLOD2)

In five patients, variants were identified in genes which primarily involve in bone mass regulation or bone matrix biology rather than classical OI. Specifically, three individuals carried heterozygous VUS in LRP5, a gene in which heterozygous variants may be consistent with an autosomal dominant bone fragility/low bone mass spectrum, and two individuals carried heterozygous VUS in PLOD2, a gene classically associated with autosomal recessive disease when biallelic pathogenic variants are present. Given their VUS classification and zygosity, these variants were considered non-confirmatory and were reported descriptively as alternative/uncertain findings. Variant-level details are presented in Table 3.

Table 3.

Variant-level details for non-OI bone-related genes

Study ID	Gene	Transcript	cDNA (HGVS)	Protein (HGVS)	Zygosity	ACMG class	Count	Confirmatory for OI?
BONE-01	PLOD2	NM_182943.3	c.1313A>G	p.Tyr438Cys	Het	VUS	1	No
BONE-02	PLOD2	NM_182943.3	c.464C>A	p.Pro155His	Het	VUS	1	No
BONE-03	LRP5	NM_002335.4	c.1544G>T	p.Gly515Val	Het	VUS	2	No
BONE-04	LRP5	NM_002335.4	c.4414C>T	p.Leu1472Phe	Het	VUS	1	No
BONE-05	LRP5	NM_002335.4	c.2774G>A	p.Arg925His	Het	VUS	1	No

ID: Identifier; HGVS: Human Genome Variation Society variant nomenclature; ACMG: American College of Medical Genetics and Genomics (variant classification); Zyg: Zygosity; Het: Hete-rozygous; Hom: Homozygous. NM_… indicates the NCBI RefSeq transcript accession (version shown after the dot). c. denotes coding DNA (cDNA) change; p. denotes predicted protein change.

Genetically Negative Cases

Among the seven patients with negative (“normal”) genetic testing, the age at genetic testing was 14.0±10.8 years, and three were females. At initial presentation, four patients were referred due to long-bone fractures, whereas three presented with vertebral compression fractures and/or osteoporosis. All patients in this subgroup were ambulatory at last follow-up.

DISCUSSION

In this single-center retrospective cohort of 42 individuals evaluated for suspected or confirmed OI at a tertiary referral center in Turkiye, variants in OI-associated genes were identified in 30 patients (71.4%). Molecularly confirmatory OI variants defined as pathogenic/likely pathogenic variants were present in 23 patients (54.7%). In our cohort, variants in COL1A1 and COL1A2 together accounted for approximately three fourths of all genetically confirmed OI cases, consistent with the dominance of type I collagen defects across populations. The recent large-scale Chinese cohort by Zhou et al. [10] (2025) similarly reported COL1A1 and COL1A2 variants in ~76% of 298 families, with nearly equal distribution between the two genes. This parallel underscores the central pathogenic role of type I collagen structural or quantitative alterations in OI worldwide. However, the relative proportions can vary by region: European and Japanese cohorts often report higher COL1A1 frequencies and fewer COL1A2 defects [11, 12]. COL1A1 variants were slightly more frequent and were largely predicted loss-of-function (canonical splice-site or truncating), with 8/10 variants being truncating or canonical splice-site (5 frameshift, 2 nonsense, 1 splice-site) supporting haploinsufficiency (often via nonsense-mediated decay) and a “quantitative” collagen defect with a typically milder OI phenotype in our cohort.

Conversely, COL1A2 variants were predominantly glycine substitutions within the collagen triple-helical domain, and all COL1A2 variants in our cohort were glycine-substituting missense changes (7/7), supporting a “qualitative/dominant-negative” mechanism consistent with dominant-negative effects and a wider, and more severe, clinical spectrum [13]. Consistent with this mechanism, the COL1A1 subgroup in our series was enriched for truncating and canonical splice-site variants, whereas COL1A2 findings were dominated by missense glycine substitutions within the triple-helical region, supporting a dominant-negative effect profile. The predominance of COL1A-related disease highlights the need for targeted diagnostic testing strategies prioritizing these genes, especially in resource-limited settings. Taken together, the high proportion of COL1A1/COL1A2-positive cases across diverse populations confirms that COL1A-related defects remain the major etiology of OI pathogenesis, despite increasing recognition of recessive and non-collagenous forms.

In addition to type I collagen genes, we identified a meaningful subset of individuals with variants in non-COL1A genes previously associated with OI. The proportion of “non–type I collagen” diagnoses varies across cohorts and is strongly influenced by ascertainment and referral pathways. For example, in England’s national Highly Specialised Service for severe/complex atypical paediatric OI, most atypical referrals had genetic causes outside COL1A1/COL1A2.

Non-collagen pathogenic variants were identified in about 20% of children with OI, most commonly in IFITM5, P3H1-complex genes, SERPINF1, and BMP1 [14]. Because many non-COL1A OI genes follow an autosomal recessive inheritance pattern, interpretation in this subgroup appropriately prioritized biallelic pathogenic/likely pathogenic variation as the principal criterion for molecular confirmation [15]. Biallelic involvement was documented in five patients CRTAP (type VII; n=2), P3H1/LEPRE1 (type VIII; n=2), and FKBP10 (type XI; n=1). These genes act through endoplasmic reticulum–based collagen biosynthesis and post-translational processing. As a group, non–type I collagen OI genes primarily disrupt procollagen modification, helix folding, and bone matrix mineralization, rather than directly changing collagen-I chain structure. Consistent with this mechanism, the biallelic variants in our cohort were mostly truncating and/or splice-altering (nonsense, frameshift, or canonical splice-site), supporting loss-of-function effects [16, 17]. We also observed one patient carrying a heterozygous WNT1 variant (historical OI type XV). While biallelic WNT1 variants are a recognized cause of recessive OI, heterozygous variants have been reported across an autosomal-dominant bone fragility/early-onset osteoporosis continuum. Accordingly, this result was interpreted within the broader monogenic bone fragility spectrum rather than treating it as a definitive recessive OI diagnosis [18, 19].

The remaining individuals in this subgroup carried monoallelic variants in genes that are typically recessive (including SERPINH1, FKBP10, and BMP1). These single-allele findings (often VUS) were considered insufficient to establish disease and instead framed as (i) possible carrier status, (ii) an incomplete molecular diagnosis with an undetected second allele (e.g., deep intronic/regulatory variants, CNVs/structural variants), or (iii) coincidental findings in a clinically selected skeletal fragility population. This underscores a practical diagnostic point: when a recessive OI gene is implicated by a single variant in a clinically convincing case, escalation to segregation testing, CNV analysis, and/or exome/genome-based reanalysis may be necessary for the diagnosis. Building on this point, a fully negative (“normal”) targeted OI result likewise should not be interpreted as excluding a genetic etiology. Among the seven genetically negative patients (mean age at testing 14.0±10.8 years), referral indications were mixed-four presented after long-bone fractures, while three were evaluated primarily for vertebral compression fractures and/or osteoporosis and all remained ambulatory at last follow-up, suggesting an overall milder fragility trajectory. Similar to monoallelic findings in recessive genes, negative panel results may reflect technical and interpretive blind spots. When clinical suspicion remains high particularly in patients with vertebral compression fractures or recurrent low-trauma fractures escalation to exome/genome sequencing with CNV calling and periodic reanalysis, alongside a standardized bone fragility work-up, may improve diagnostic resolution and counseling.

Skeletal fragility referrals often extend beyond “classical” OI, and our cohort illustrates this diagnostic overlap. In a small subset, we detected heterozygous variants in genes that primarily influence bone mass regulation or bone matrix biology rather than canonical OI mechanisms. Importantly, while these findings were not considered confirmatory given their VUS classification. In particular, heterozygous (autosomal-dominant) LRP5 variation can be clinically meaningful within an early-onset osteoporosis/low–bone-mass spectrum, and thus may provide a plausible genetic framework for fracture susceptibility in selected patients, especially when targeted OI testing is negative or yields only inconclusive results [20]. By contrast, PLOD2 is classically implicated in a contracture–fragility (Bruck syndrome) phenotype when biallelic pathogenic variants are present; therefore, isolated heterozygous VUS in PLOD2 are best interpreted cautiously as non-confirmatory [21]. Overall, these observations reinforce that fracture-based genetic evaluation can uncover overlapping monogenic fragility conditions beyond “classical” OI and highlight the need for phenotype-anchored interpretation and structured follow-up findings.

This study provides real-world tertiary-center data on the clinical spectrum and molecular findings among individuals evaluated for suspected OI, reflecting routine referral patterns dominated by fracture-related presentations. A key strength is that genetic testing results were available for all patients, allowing a transparent estimation of diagnostic yield and the proportion of molecularly confirmed OI, genetically negative cases, and alternative bone-related genetic findings. Several limitations should be acknowledged. First, the study is retrospective and single-center, which may limit generalizability. Second, the cohort represents a referral population, potentially enriching for more clinically apparent cases and influencing the observed genetic distribution. Third, genetic evaluation relied primarily on a targeted gene panel, which may not capture all disease-causing variation (e.g., copy-number variants, deep intronic/splice-altering variants, structural variants, or low-level mosaicism), and therefore a negative result does not exclude a genetic etiology. Finally, several findings particularly in non-COL1A genes and non-OI bone-related genes were classified as VUS, and functional validation or segregation analyses were not systematically available; thus, genotype–phenotype inference remains limited.

Conclusion

This tertiary-center cohort highlights the clinical value of combining phenotypic assessment with molecular testing in patients evaluated for suspected OI. Genotype-based classification can help confirm the diagnosis, guide counseling, and plan follow-up. However, VUS and panel-negative results emphasize that genetic results are not always definitive and may need review as new evidence emerges. Overall, these findings support a practical, genetics-guided approach to skeletal fragility referrals, while acknowledging that some patients may require broader evaluation beyond classical OI.

Ethics Committee Approval

The Ankara City Hospital Chair of the No. 2 Clinical Research Ethics Committee granted approval for this study (No: E-22-1814, Date: 11.05.2022).

Informed Consent

Informed consent was not required due to the retrospective nature of this study.

Conflict of Interest

The authors declare that there is no conflict of interest.

Financial Disclosure

The authors declared that this study has received no financial support.

Use of AI for Writing Assistance

The authors declared that artificial intelligence was not used in the study.

Authorship Contributions

Concept – EO, EK, HMT, MA, VT, BT, AB, PK, MB, FG; Design – EO, EK, HMT, MA, VT, BT, AB, PK, MB, FG; Supervision – EO, EK, MA, VT,PK, MB, FG; Fundings; EO, EK, MB, FG; Materials – EO, EK, HMT, MA, VT, BT, AB, PK, MB, FG; Data collection and/or processing – EO, EK, HMT, MA, VT, BT, AB, PK, MB, FG; Analysis and/or interpretation – EO, EK, HMT, MA, MB, FG; Literature search – EO, EK, HMT, MA, VT, PK, MB, FG; Writing – EO, EK, HMT, MA, VT, BT, AB, PK, MB, FG; Critical review – EO, EK, MA, VTPK, MB, FG.

Peer-review

Externally peer-reviewed.