A Retrospective Study of the Presentation, Diagnosis, Management, and Outcomes of 27 Patients with Osteogenesis Imperfecta at a Single Center in Türkiye

Introduction

Osteogenesis imperfecta, colloquially known as glass bone disease, is as a genetic connective tissue disorder affecting approximately 1 in every 15 000–20 000 births [1]. Coined by Vrolik in 1849, the term “Osteogenesis Imperfecta” was first used to describe a child born with multiple fractures [2]. Marked by distinctive clinical features, including blue sclera, dental impairment, hearing loss, ligament laxity, fragile bones, and bone deformities, osteogenesis imperfecta is the most common genetic cause of osteoporosis. Its clinical spectrum varies widely, ranging from a fatal form during childbirth to milder manifestations that can go unnoticed until adulthood [3,4].

Despite the inherent fragility of bones in osteogenesis imperfecta patients and the potential for vertebral compression or long-bone fractures, even with low-energy trauma, higher mean mineralization density values have been observed in age-matched bone measurements. Moreover, osteogenesis imperfecta extends its impact beyond the skeletal system, affecting various organs such as the skin, ligaments, tendons, sclera, nose, and ears [5,6].

Osteogenesis imperfecta is an uncommon hereditary disorder, usually due to mutations in the COL1A1 and COL1A2 genes [7]. Most individuals with osteogenesis imperfecta have a genetic mutation that causes problems in the quantity and quality of type I collagen molecules, affecting over 90% of osteogenesis imperfecta patients. A decade ago, a wide range of mutations associated with osteogenesis imperfecta was uncovered. Individuals diagnosed with osteogenesis imperfecta harbor diverse autosomal genetic mutations, exceeding 1500 combinations discovered so far, resulting in a deficiency of type I collagen [8]. The mode of inheritance for this condition can be either spontaneous mutation, autosomal recessive, or autosomal dominant [9,10]. Autosomal recessive forms are caused by non-collagenous proteins that participate in triple helix formation or post-translational modifications, whereas autosomal dominant forms are attributable to the direct abnormalities in type 1 collagen [1].

In 1979, Sillence and Danks first classified 4 forms of osteogenesis imperfecta according to their clinical and genetic characteristics [11]. Initially, they classified types I and IV as exhibiting autosomal dominant heredity, whereas types II and III were categorized as displaying autosomal recessive inheritance [1]. Cole expanded the Sillence classification by adding types V to XI based on a study focusing on genetic abnormalities; type V is characterized by autosomal dominant transmission, whereas types VI to XI are associated with autosomal recessive transmission [9].

The osteogenesis imperfecta classification, as determined by the International Society of Skeletal Dysplasias, is based on the mechanism of inheritance and the specific genes involved. Type I, the least affected, is characterized by non-deforming features with persistently blue sclera; type II is the lethal form, observed in the perinatal period; type III patients experience severe effects; type IV is considered moderate; and type V presents with calcification of the interosseous membranes and/or hypertrophic callus [8,12] (Table 1).

The management of fractures in osteogenesis imperfecta patients, particularly in milder instances, can be challenging, as essential indicators such as dental impairment, blue sclera, and family history that may raise suspicion of osteogenesis imperfecta are not always present. Distinguishing osteogenesis imperfecta from unaffected patients becomes complex because results provided by clinical examination and conventional radiology are often insufficient [9]. However, recent studies have emerged, shedding light on the relationship between radiographic fracture characteristics and osteogenesis imperfecta [10].

Therefore, this retrospective study aimed to evaluate the presentation, diagnosis, management, and outcomes of 27 patients diagnosed with osteogenesis imperfecta at a single center in Turkey between January 2011 and January 2020.

Material and Methods

ETHICS APPROVAL AND INFORMED CONSENT:

The study received approval from the Ethics Board for Non-Invasive Clinical Researches of the University of Çukurova, School of Medicine, with meeting and decision number 92/7. Written informed consent was obtained from all parents and legal guardians of the 27 osteogenesis imperfecta patients before participation in the study, both for the conduct of the clinical study and for evaluation of the results of genetic analysis.

STUDY DESIGN AND DATA COLLECTION:

This retrospective study involved the analysis of medical records of patients with osteogenesis imperfecta admitted to the University of Çukurova, School of Medicine, Orthopedics Clinic between January 2011 and January 2020. The study encompassed osteogenesis imperfecta patients undergoing conservative treatment or surgery due to fractures. Initially, 33 osteogenesis imperfecta patients were considered, of which 6 were excluded – 4 lacked genetic analyses, and 2 were missing recent clinical examination data due to patient death before the study analysis. Subsequently, data on the remaining 27 patients were analyzed retrospectively.

Patients were classified according to phenotypic/genotypic findings by pediatricians and medical genetic department counseling, and their information was available in patient files. The data included the clinical examination notes of the cases classified according to the Sillence and Shapiro systems, encompassed age, sex, parental consanguinity, genetic analysis (DNA isolation) results, the number and localization of past fractures, treatment methods, complications, hypermobility, and ambulation scoring.

CLINICAL PRESENTATION AND DIAGNOSIS:

Two classification systems, developed by Sillence et al and Shapiro et al, were employed for clinical classification. Sillence et al’s system divided patients into 4 main types based on clinical characteristics and inheritance patterns, further expanded to include a broader range of patient subgroups [13]. Shapiro’s classification, based on when fractures occurred, improved the prognosis of survival and ambulation [14].

Hypermobility was assessed using the Beighton score, a well-known screening technique for measuring joint flexibility. The Beighton score utilizes a 9-point system, with higher scores indicating greater joint flexibility [15].

The patients’ ambulation condition was assessed using the Hoffer scale, which classifies patients into 4 levels: community ambulator, household ambulator, therapeutic ambulator, and wheelchair-dependent [16].

GENETIC ANALYSIS:

Genetic Analysis was conducted at the genetic diagnosis center AGENTEM (Adana Genetic Diseases Diagnosis and Treatment Center) of Çukurova University.

The records show that 2 cc of peripheral blood samples obtained from the patients were used in the DNA analysis conducted by DNAmidi kit, Qiagen, Germany, on an automated DNA isolation system (QiaSymphony, Qiagen, Germany). DNA concentration was measured using Qubit (ThermoFisher, USA) and saved for the library preparation procedure. The samples were fragmented, and initial barcodes were attached in the ligation phase. Target regions for the multigene panel, designed within the center, were amplified by the polymerase chain reaction (PCR) method. Then, second barcodes and sequencing adapters were attached to the prepared samples. The quality and quantity of the libraries were then measured by qPCR, and a library pool was created accordingly. The library pool was loaded into the Miseq (Illumina, USA) next-generation sequencing system and sequenced. Bioinformatics analysis of the sequence data obtained from patient samples was performed in 3 stages. First, quality control of the sequencing data was performed to determine the suitability of the obtained data for analysis. Second, variant analyses of the received data and evaluation of the clinical meaning of the detected variants were performed. Each detected variant was initially checked using the databases [Human Genome Mutation Database (HGMD), NCBI dbSNP database, PubMed, and disease-specific data sources]. Any possible effects due to previously undetected variants or unknown clinical significance (ranging from pathological to benign) were assessed using at least 3 different in-silico PCR tools (eg, PolyPhen2, SIFT, MutationTaster). Third, clinical reporting of variants with disease etiology or prognostic significance was performed by physicians specializing in medical genetics per international standards. The primary mutation types were COL1A1, COL1A2, and SERPINF1. Other rare genetically based types of osteogenesis imperfecta whose primary mutation type could not be identified despite genetic analysis were determined as non-COL1A1-COL1A2, and sub-mutation types were evaluated as ‘undetermined’.

ORTHOPEDIC TREATMENTS, CLINICAL FOLLOW-UP, AND COMPLICATIONS:

Treatments were categorized as surgical or conservative. Surgical procedures included osteotomy, titanium elastic nailing (TEN), tension band wiring, and Ilizarov external fixation, while conservative treatments comprised splints, pelvic pedal casting, slings, velpeau, and Pavlik bandaging procedures.

Patients who were treated conservatively were invited to weekly clinical controls in the first 3 weeks after the fracture, and fracture healing was checked with direct radiographs. Patients who underwent surgical treatment were invited to clinical controls in the second week postoperatively; fracture healing was checked with direct radiographs, and sutures were removed. In both groups, after the fracture healing was completed at 6–8 weeks, patients were called for clinical controls in 3-month periods in the first 1-year period.

Complications (eg, angulation, shortening, and fixation failure following conservative or surgical treatment) were documented.

STATISTICAL ANALYSIS:

Statistical analysis was performed using the SPSS 25.0 package program. Categorical measurements are presented as numbers and percentages, while continuous measurements are summarized as mean and standard deviation (median and minimum-maximum where necessary). The comparison of continuous measurements utilized the Mann-Whitney U test or Kruskal-Wallis Test. The correlation between variables was determined using Spearman’s test, with coefficient “r” scores indicating different levels of correlation. A significance level of P<0.05 was considered statistically significant.

Results

DEMOGRAPHIC CHARACTERISTICS AND CLINICAL PRESENTATION OF PATIENTS:

A total of 27 patients were included in the analysis, comprising 13 (48.1%) males and 14 (51.9%) females. The mean age of the patients was 10.4±7.4 years (range, 3–39). Almost half (n=15, 55.6%) had consanguineous parents.

The Sillence classification identified 18 type I (66.67%), 3 type III (11.11%), and 6 type IV (22.22%) cases, while the Shapiro classification revealed no Congenita A, 8 Congenita B (29.7%), 4 Tarda A (14.8%), and 15 Tarda B (55.55%) patients.

Blue sclera was present in 17 (62.9%) patients, and no statistically significant differences were found between genetic mutation subtypes and the presence of blue sclera (P=0.925). Dentinogenesis imperfecta (DGI) was present in 12 (44.4%) patients, with no statistically significant differences between genetic mutation subtypes and DGI (P=0.667).

All patients scored more than 4 points on the Beighton joint hypermobility scale, classifying them as hypermobile, with a mean score of 6.9±1.5.

The Hoffer ambulation scale results indicated that 17 patients were community ambulatory, including 3 using crutches. Six were household ambulatory, and 4 were wheelchair-dependent.

Genetic mutation types and subtypes of all patients according to the Sillence and Shapiro classifications are presented in Table 2.

FRACTURE LOCATIONS, TREATMENT STRATEGIES, AND COMPLICATIONS:

All patients admitted to the orthopedics outpatient clinic experienced 1 or more fractures during the 9 years of follow-up. Among them, 13 (48.15%) received both surgical and conservative treatments, while the remaining 14 underwent only conservative treatments. The analysis revealed no association between the location of fractures and the Sillence or the Shapiro classes (P=0.779, P=0.252, respectively) (Table 3).

The patients had 131 fractures during the 9 years between January 2011 and January 2020. The number of fractures based on the localization was as follows: 61 in the femur (46.5%), 42 in the tibia (32.1%), 13 in the humerus (9.9%), 13 in the forearm (9.9%), and 2 in the clavicula (1.6%). Most of the fractures occurred in the lower extremities (n=103, 78.6%).

The femur was the most commonly fractured bone, which was observed on 61 occasions (29 fractures in the right [47.5%] and 32 in the left [52.5%] femur) (n=20 cases, 74%). The fractures in the femur were mainly localized in the diaphysis (n=59, 96.7%). The conservative and surgical methods used in the fixation of the femur fractures were splint, TEN, PPA, osteotomy with TEN, Pavlik bandage, and plating. The most frequent method for fixation was splinting, which was used in 26 (42.6%) fractures. The TEN, PPA, osteotomy with TEN, Pavlik bandage, and plating were used in 13, 10, 10, 1, and 1 patients, respectively. Following fixation, angulation developed in 7 patients treated with splints, shortening in 3, shortening with angulation in 1, angulation with implant failure in 1 patient treated with TEN, and angulation in 3 patients treated with PPA. There were 20 complications observed in 15 fractures, with a rate of 45% (n=9). There were no correlations of the complications between the fractures that occurred in other locations and femur fractures (P=0.320).

There were 42 fractures in the tibia, predominantly in the left (n=31), occurring in 18 (66.7%) patients. One metaphysis fracture was reported, and the remaining fractures were localized in the diaphysis. Almost all fractures (39, 92.9%) were fixated with splints, and the remaining 3 were each fixated with TEN, osteotomy with TEN, and Ilizarov external fixation methods. The reported incidents were as follows: angulation in 3 treatments performed by splints and a case with angulation with implant failure treated with TEN. There were 2 patients with 4 complication incidents, and the complication rate of tibia fracture fixation was statistically significant among all fracture cases (P=0.0001).

Regarding the fractures located in the humerus, 10 patients had 13 fractures (6 located in the right and 7 in the left humerus). There were 11 diaphysis and 2 metaphysis fractures. Five of the fractures were treated with arm slings, 4 with splints, 2 with TEN, 1 with velpeau bandaging, and 1 with cannulated screw. There was only 1 angulation complication, which occurred following a splint treatment. There was a significant (P=0.0001) difference in complication rates of the treated humerus fractures compared to fractures in the other locations.

A similar rate was observed in the fractures located in the forearm, occurring in 11 patients, with a total number of 13 (6 located in the right and 7 in the left forearm). Diaphyseal fractures (n=8) were the most common. Ten fractures were fixated by splints, 2 with the tension banding method and 1 with TEN. There was only 1 case reported with malunion with angulation following splint treatment. Complications were significantly associated with forearm fractures (P=0.0001).

There were only 2 fractures (1 located in the right and 1 in the left forearm), located exclusively in the clavicula in 1 case. Both fractures were diaphyseal and were treated with arm slings without any complications.

Regarding the most frequently fractured bone (femur, n=20, 74.07%), no significant differences in complication rates were found (P=0.320). However, patients with tibia and humerus fractures had significantly lower complication rates (P=0.0001).

No statistically significant differences in complication rates were observed based on genetic mutations, including subtypes (P=0.140). In comparing the treatment types, surgical treatment was significantly more common in cases with COL1A1 mutations (P=0.008) (Table 4).

CORRELATION ANALYSIS:

The analysis demonstrated no correlation between age and the number of fractures (r=−0.09; P=0.669). However, a strong association was found between the number of fractures and genetic mutation types (P=0.004) (Table 5).

Although there was no significant association between the number of fractures and the presence of subtype G>A (r=−0.42; P=0.202), a negative correlation was observed among mutations other than subtype G>A, showing a decrease in the number of fractures with age (r=−0.71; P=0.015).

Discussion

The examination of demographic factors, including age and sex, revealed no significant differences, ageeing with distribution patterns reported in the literature [17,18]. A remarkably high rate of consanguinity among parents (55.56%) was observed, and in another study with a similar population, this rate was 25% [19]. While various studies on osteogenesis imperfecta cases report consanguinity rates ranging from 24% to 100%, the exceptionally high rates in our study emphasize the need for further investigations to facilitate a more nuanced approach to osteogenesis imperfecta patients [19–21].

The association between hypermobility and osteogenesis imperfecta has been previously suggested, with the prevalence appearing higher in type I cases [22]. Our study population paralleled these findings, exhibiting hypermobility and an increased number of type I cases with higher Beighton hypermobility scores.

Current classification systems, such as Sillence, are preferred in managing osteogenesis imperfecta cases. In our study, the distribution of types I, III, and IV based on Sillence (66.67%, 11.11%, and 22.22%, respectively) closely mirrors findings from a large study, indicating stability in osteogenesis imperfecta classification patterns over time [18]. Blue sclera did not show a statistically significant association with genetic mutation types, consistent with previous research [23]. However, our study revealed a lower prevalence of dentinogenesis imperfecta (DGI) (44.4%) compared the literature, with 11.1% of cases exhibiting DGI in non-collagenous osteogenesis imperfecta types [18,23,24].

Contrary to prior studies, our analysis did not find significant correlations between the presence of blue sclera, DGI, and genetic mutation subtypes., which suggests the need to further explore the relationship between clinical features and genetic mutations in osteogenesis imperfecta cases. In addition, we did not find a significant genotype-phenotype correlation in patients with osteogenesis imperfecta diagnosed with COL1A1 or COL1A2 variants. The study by Erbaş et al on Turkish osteogenesis imperfecta patients also supports our finding [25].

Fracture distribution based on the Sillence and Shapiro classifications showed no statistically significant differences, aligning with findings from Wekre et al, highlighting the complexity of predicting fracture locations using current classification systems [26]. Our study reported 131 fractures over a 9-year follow-up period, predominantly in the lower extremities, with the femur being the most frequently fractured bone. Similar findings were reported in studies by Moorefield and Miller, emphasizing the significance of lower extremity issues in osteogenesis imperfecta cases [27]. Intramedullary titanium elastic nails were employed in our surgical treatments, demonstrating low complication rates, consistent with the literature [28,29].

Notably, our study revealed significantly lower complication rates in patients with tibia and humerus fractures, suggesting that these cases may be more easily managed using standard outpatient methods. Despite the lack of a direct link between genetic mutations and complication rates, the type of treatment showed an association with mutation types, emphasizing the potential role of genetic typing in osteogenesis imperfecta management. Specifically, a higher number of surgical interventions were observed in cases with COL1A1 mutations, providing valuable guidance for treatment decisions.

Our analysis found a significant association between non-collagenous osteogenesis imperfecta types (eg, SERPINF1) and a higher number of fractures. While the exact mechanisms underlying these mutations and bone fragility are not fully understood, evidence suggests a negative impact on bone matrix mineralization, contributing to fragility [30]. Patients with osteogenesis imperfecta were diagnosed at an early age, and severe deformities and fractures were observed.

The most common complication observed in both conservative and surgical treatments was angulation, highlighting the importance of continuous outpatient follow-up for early detection of complications. The results underscore the need for vigilance in managing complications, which can significantly impact the quality of life for osteogenesis imperfecta patients.

Two primary challenges individuals face with osteogenesis imperfecta are reduced bone mass and diminished bone quality. Existing therapeutic approaches enhance bone quantity but do not alter bone quality. Treatment options for primary osteoporosis differ according to the severity of the illness and the patient’s age. Due to increased fragility and osteopenia, fractures occur more frequently, often in unusual locations, with an incidence peaking in children and declining with age [31]. Studies have demonstrated that bisphosphonates enhance bone densitometry in individuals with osteogenesis imperfecta, leading to biochemical improvement [32].

Several limitations should be considered when interpreting the present results, including the retrospective design, limited sample size, geographic specificity, absence of a control group, and medication therapy management information. Future research should aim to address these limitations for a more comprehensive understanding of the relationships explored in this study.

Conclusions

Our findings indicate that the type of genetic mutation does not significantly correlate with the risk of treatment complications in osteogenesis imperfecta cases. Nevertheless, we found a significant association between the type of mutation and the number of surgeries required. Specifically, cases with the COL1A1 mutation exhibited a higher demand for surgeries than those with other mutations, suggesting the potential utility of genetic typing in guiding treatment decisions. Furthermore, our analysis underscores a distinctive pattern in non-collagenous osteogenesis imperfecta cases, such as SERPINF1, which exhibited a higher frequency of fractures compared to other mutation types. This highlights the need for a nuanced approach to understanding the diverse genetic factors influencing fracture risk in osteogenesis imperfecta patients. Our study shows the need for integrating gene typing as a valuable tool in managing osteogenesis imperfecta cases, particularly in predicting the surgical needs associated with specific genetic mutations.