Osteogenesis Imperfecta

Literature and regulatory information updated: 6 October 2026; integration of the French National Diagnosis and Care Protocol (PNDS, HAS, 2017): 6 November 2026.

Brittle bone disease — all genetic types and subtypes (OI 1 to 23): comprehensive lecture for medical geneticists and specialist healthcare professionals

---

0. Initial Reference Table

Table adapted and synthesised from Sillence et al. (1979) [1], Van Dijk & Sillence (2014) [2], GeneReviews (May 2025 update) [3], Orphanet [4] and table 14.1.1 of Spranger et al. [5]. Types 1 to 5 correspond, with minor nomenclature variations, to the original Sillence phenotypic classification; types 6 to 23 were progressively incorporated as new causative genes were identified by next-generation sequencing [6]. In current OMIM nomenclature, type 18 corresponds to TENT5A and type 19 to the X-linked form due to MBTPS2*. International Classification of Diseases (ICD-10): Q78.0 [7]. Abbreviations: AD, autosomal dominant; AR, autosomal recessive; XL, X-linked; OMIM, Online Mendelian Inheritance in Man.*

---

1. Introduction

1.1 Definition

Osteogenesis imperfecta (OI), popularly known in Portuguese as “glass bone disease” — from the English brittle bone disease, meaning “fragile bone disease” — is a clinically, radiographically and genetically heterogeneous genetic connective tissue disorder caused by quantitative or structural deficiencies of type 1 collagen — the structural protein of bone tissue — or by abnormalities in proteins that interact with this collagen post-translationally, that is, after protein synthesis, resulting in increased bone fragility [5]. The disease is fundamentally characterised by excessive bone fragility, resulting in fractures following minimal trauma or, in some forms, even in the complete absence of trauma. In addition to skeletal fragility, the condition affects a wide range of structures derived from type 1 collagen, the principal structural protein of bone, skin, tendons, ligaments, teeth and the sclera, the white protective layer of the eye. For this reason, osteogenesis imperfecta is now understood as a systemic connective tissue disease rather than merely an isolated skeletal disorder. For all subtypes, OI is defined by the presence of bone fractures and osteopenia [21].

The number of fractures, secondary deformities and variable soft tissue abnormalities produce a spectrum of clinical and radiographic phenotypes — that is, distinct presentations on physical examination and imaging investigations — ranging from perinatally lethal forms to forms with a mild predisposition to fractures, normal stature and normal life expectancy. This phenotypic system is useful for prognosis, but each phenotypic group is aetiologically heterogeneous [5].

1.2 Historical overview and founding article

Osteogenesis imperfecta has been recognised clinically for centuries, with descriptions attributed to Ekman in 1788 and subsequently to Lobstein, Vrolik and Van der Hoeve, whose names still echo in the historical synonyms “Lobstein disease” or “Ekman–Lobstein disease” (non-deforming form, osteogenesis imperfecta tarda), “Vrolik disease” (perinatally lethal form) and “Van der Hoeve syndrome” (late form with deafness) [21–23]. One of the first modern obstetric series, by Freda and colleagues (1961), described 16 cases of osteogenesis imperfecta congenita [21]. However, the founding milestone of modern disease nosology is the work of Australian researcher David Sillence with Alison Senn and David Danks, published in the Journal of Medical Genetics in 1979 under the title “Genetic heterogeneity in osteogenesis imperfecta” [1]. In this article, the authors analysed a cohort of Australian families and proposed, for the first time, a systematic classification into four types — I to IV — based on clinical and radiographic criteria, the inheritance pattern observed in families and natural history, rather than solely on the old dichotomy between “congenital OI” and “OI tarda”, which proved inconsistent because infants with severe forms could be born without apparent fractures and children with mild forms could already fracture at birth. The Sillence classification remained the worldwide reference for more than two decades and is still used in clinical settings; Sillence himself, in collaboration with Barlow, Cole, Garber and others, published detailed delineations of phenotypes II (1984) and III (1986) during the 1980s, consolidating the system [21, 24, 25].

Having understood the historical origin of the classification, it is important to place the reader within the subsequent trajectory of the disease as a molecular entity. During the 1980s and 1990s, the prevailing understanding was that OI resulted almost exclusively from mutations in COL1A1 and COL1A2, which encode type I collagen chains. Peter Byers and colleagues played a central role during this period, biochemically characterising structural collagen mutations in patients with lethal and progressively deforming forms; Willing and colleagues demonstrated in 1992 that type I OI usually results from a COL1A1 null allele [21]. With the advent of molecular biology and genetic sequencing techniques from the 1990s, and particularly in the 2000s and 2010s, it became possible to identify the causative genes of each disease form, revealing much broader molecular heterogeneity than previously assumed. Rauch and Glorieux expanded the classification in 2004 to include types 5 to 7 (type 8 was added in subsequent adaptations of this table) [9, 26]. From 2006, this monogenic dominant paradigm was definitively overturned: Barnes and colleagues demonstrated in the New England Journal of Medicine that mutations in CRTAP caused a lethal recessive form of OI clinically indistinguishable from classic type II, inaugurating the era of recessive disease forms [5]. In subsequent years, a succession of discoveries — LEPRE1 in 2007, PPIB in 2009–2010, SERPINH1 and FKBP10 in 2010, SP7 in 2010 and SERPINF1 in 2011, BMP1 and TMEM38B in 2012, WNT1 and CREB3L1 in 2013, SPARC in 2015, among others — progressively expanded the genetic spectrum of the disease, such that osteogenesis imperfecta is now formally recognised in at least 22 to 23 genetically distinct types, caused by pathogenic variants in more than twenty different genes encoding proteins ranging from the type 1 collagen chains themselves to proteins involved in collagen modification, processing, intracellular transport and stability, as well as proteins regulating differentiation and function of the osteoblast, the cell responsible for bone formation — all converging, directly or indirectly, on the type I collagen synthesis, processing, secretion or mineralisation pathway [3, 5, 6, 27, 28].

This growth in the genetic spectrum led to a formal nomenclature revision. Van Dijk and Sillence, in 2014, proposed in the American Journal of Medical Genetics Part A an update to the original classification, retaining Sillence's four primary phenotypic groups — non-deforming with blue sclerae, perinatally lethal, progressively deforming and moderately deforming with normal sclerae — as the central axis of clinical communication, while naming additional genetic types (5 onwards) descriptively, linked to the gene or molecular mechanism involved [2]. A further revision appeared in 2024: the dyadic nosology of OI and bone fragility syndromes proposed by Sillence (Calcified Tissue International, 2024; doi:10.1007/s00223-024-01248-7), which develops for OI the 11th revision of the International Skeletal Dysplasia Society's Nosology of Genetic Skeletal Disorders (Unger et al., American Journal of Medical Genetics Part A, 2023; doi:10.1002/ajmg.a.63132) and combines the Sillence phenotypic group (types 1 to 5) with the HGNC symbol of the causative gene (for example, “OI1-COL1A1”); an alternative classification based on pathogenic mechanisms, proposed by Yu and colleagues in 2023, was criticised in 2024 for conflicting with the numbering in use (Dalgleish et al., Orphanet Journal of Rare Diseases, 2024; doi:10.1186/s13023-024-03269-9). To date, this is the nomenclature recommended by GeneReviews in its most recent revision, updated in May 2025: the terms “OI type I”, “OI type II”, “OI type III” and “OI type IV” have formally been replaced in reference texts by “classic non-deforming OI with blue sclerae”, “perinatally lethal OI”, “progressively deforming OI” and “common variable OI with normal sclerae”, respectively — although Roman numbering remains widely used in everyday clinical practice and will be retained throughout this lecture for teaching clarity, with the descriptive nomenclature in parentheses [3]. In Smith's Recognizable Patterns of Human Malformation, type I collagenopathy phenotypes are described as a continuum of short stature, fractures with minimal or absent trauma, skeletal deformities, dentinogenesis imperfecta, hearing loss and skin and joint manifestations, frequently described as a continuous spectrum between mild and severe forms [21].

This text aims to present, in an integrated and in-depth manner, the molecular genetics, pathophysiology, clinical spectrum, diagnostic investigation, treatment — including the most recent therapeutic advances, with results published up to 2026 — and genetic counselling of osteogenesis imperfecta, for an audience of clinical geneticists and specialists in bone disease.

---

2. Molecular Genetics and Pathophysiology

2.1 Structure and biosynthesis of type I collagen

Type I collagen is the most abundant structural protein in bone, accounting for approximately 40% of total bone volume and constituting the fibrillar framework on which mineral hydroxyapatite is deposited; it is a fibrillar protein forming the basis of the bone extracellular matrix, providing resistance to mechanical tension. It is a heterotrimer, that is, it consists of three intertwined polypeptide chains: two pro-α1(I) chains, encoded by COL1A1 on the long arm of chromosome 17 (17q21.33), and one pro-α2(I) chain, encoded by COL1A2 on the long arm of chromosome 7 (7q21.3). Together, these three chains wind into a triple helical structure conferring mechanical strength on bone. Each chain contains a helical domain with uninterrupted repeats of the glycine-X-Y tripeptide, in which X and Y are frequently proline and hydroxyproline, respectively. Glycine, being the smallest amino acid, is the only one capable of occupying the steric centre of the triple helix without distorting it — an absolute structural requirement explaining why glycine substitutions are so deleterious.

Procollagen synthesis occurs in the rough endoplasmic reticulum of the osteoblast and involves extensive post-translational modifications: hydroxylation of proline and lysine residues, glycosylation of selected hydroxylysines and formation of disulphide bridges in the C-terminal propeptide, which nucleates triple helix folding towards the amino terminus in a “zipper-like folding” process. Procollagen is then transported through the Golgi complex and secreted into the extracellular environment, where specific proteases — including the protein encoded by BMP1 — cleave the N- and C-terminal propeptides, generating mature collagen, which organises into fibrils and subsequently into strong fibres that react with adjacent molecules outside the cell to form covalent cross-links responsible for fibre tension [7]. When type 1 collagen is defective, bones lose much of their strength and become fragile and consequently easily fractured [7].

2.2 The two central molecular mechanisms

Two molecular mechanisms explain the vast majority of cases caused by COL1A1/COL1A2 [9, 21].

The first is haploinsufficiency due to a functional null allele, in which one of the two gene copies ceases to function, that is, does not generate functional protein. Nonsense mutations, which introduce a premature termination codon, frameshift mutations, which alter the entire subsequent reading frame, or splice-site mutations, which impair messenger ribonucleic acid (messenger RNA) processing, generate transcripts degraded by nonsense-mediated decay. The affected allele ceases to contribute pro-α1(I) chains, and the cell produces only around 50% of the normal amount of type I collagen — but structurally normal collagen. The result is bone with a reduced quantity of collagen but preserved quality. This mechanism is responsible for the overwhelming majority of cases of classic non-deforming OI with blue sclerae (type I), explaining why this is the mildest disease form: there is a quantitative but not qualitative collagen deficiency. More rarely, type I OI may also be caused by mutations leading to structural defects in procollagen I molecules [22].

The second mechanism is the dominant negative effect of glycine substitution in the triple helix, typical of perinatally lethal, progressively deforming and common variable forms (types II, III and IV). When glycine is replaced by another, larger amino acid, the helical structure is disrupted — an effect termed dominant negative because the defective chain also impairs normal chains when attempting to associate with them. As each type I collagen molecule contains two pro-α1(I) chains and only one pro-α2(I) chain, a single dominant negative mutation in a COL1A1 allele can statistically compromise up to three-quarters of all collagen molecules produced by the cell (by the theorem of combinatorial probability of random chain association), resulting in a much more severe presentation than the simple quantitative reduction observed in haploinsufficiency. The result is structurally abnormal collagen compromising bone integrity and producing more severe phenotypes.

The phenotypic severity resulting from glycine substitution depends on multiple factors: the substituting amino acid (bulky, charged side chains, such as arginine, glutamic acid, aspartic acid, valine and tryptophan, tend to be more deleterious than substitutions with serine, alanine or cysteine, which tend to produce more variable phenotypes); the substitution's position along the helical domain, with bulky amino acid substitutions in the most carboxy-terminal 70% of the α1(I) chain being almost invariably lethal, possibly because they interfere with the initiation of C→N propagation of helical folding; and the affected chain, with evidence that α2(I) chain mutations generally show greater phenotypic variability and, on average, milder phenotypes than equivalent α1(I) chain mutations, even when involving amino acids with large side chains. Splice-site mutations resulting in exon skipping beyond exon 14 of COL1A1 or beyond exon 25 of COL1A2 tend to be lethal, whereas variants closer to the amino terminus produce more variable phenotypes, sometimes with prominent joint hypermobility as the dominant manifestation [27, 28]. OI/Ehlers–Danlos overlap syndrome is caused in most cases by helical mutations near the amino-proteinase cleavage site of type I procollagen [21], and C-propeptide cleavage-site mutations give rise to so-called high-bone-mass OI, which manifests after the second year of life with fractures and normal bone density or even increased mineralisation (Lindahl et al., 2011) [5, 22].

2.3 Functional categories of recessive and non-collagen genes

In addition to mutations in collagen genes, defects in proteins participating in collagen processing, folding or stabilisation — as well as abnormalities in factors regulating differentiation or activity of osteoblasts, the cells responsible for bone formation — also cause OI [9]. As new causative genes were identified, it became possible to group them into at least four broad functional categories reflecting different stages of collagen and osteoblast biology [27–29].

The first category comprises genes involved in post-translational collagen modification, notably the prolyl 3-hydroxylation complex formed by cartilage-associated protein (CRTAP), prolyl 3-hydroxylase 1 (encoded by LEPRE1, also called P3H1), which catalyses the reaction, and cyclophilin B (encoded by PPIB), which also acts as a collagen chaperone, that is, an accessory protein guiding collagen towards correct folding. This complex catalyses hydroxylation of the proline residue at position 986 of the α1(I) chain, a chemical modification essential for triple helix thermal stability; its absence results in typically severe to lethal recessive forms (types 7, 8 and 9), with phenotypic overlap with type II or III. This category also includes the endoplasmic reticulum chaperones HSP47 (SERPINH1), a collagen-binding chaperone, and FKBP65 (FKBP10), which assist correct procollagen folding and stabilisation and participate in its maturation; FKBP10 mutations cause both recessive type 11 OI and, with certain variants, Bruck syndrome type 1, characterised by the association of OI with congenital joint contractures.

The second category concerns genes involved in bone mineralisation, examples being IFITM5, encoding interferon-induced transmembrane protein 5, also called BRIL, involved in the osteoblast mineralisation phase and causing type 5 through a recurrent gain-of-function point mutation, and SERPINF1, encoding pigment epithelium-derived factor (PEDF), an endogenous angiogenesis-inhibiting protein that binds collagen, whose loss of function causes type 6, characterised by accumulation of unmineralised osteoid with a “fish-scale” histological pattern.

The third category brings together genes related to osteoblast differentiation and function, including SP7 (osterix transcription factor, essential for osteoblast differentiation), WNT1 (canonical Wnt/β-catenin pathway ligand, essential for bone formation and bone homeostasis), CREB3L1 (OASIS transcription factor, an endoplasmic reticulum stress sensor and transducer), SPARC (osteonectin, a calcium- and collagen-binding matricellular glycoprotein of the extracellular matrix), TMEM38B (TRIC-B cation channel, regulating calcium flux in the endoplasmic reticulum), MBTPS2 (S2P protease of the regulated intramembrane proteolysis pathway, the only gene causing an X-linked recessive form), BMP1 (a metalloproteinase cleaving the C-terminal propeptide of procollagen; the Spranger source designates the protein as bone morphogenetic protein 1/mTLD) and KDELR2 (a receptor for retrograde protein transport from the Golgi to the endoplasmic reticulum).

A fourth, more recently discovered category brings together genes with diverse regulatory functions: TENT5A (also called FAM46A), a non-canonical polyadenylase increasing the stability and translation of COL1A1, COL1A2 and SERPINF1 messenger RNAs; MESD, an endoplasmic reticulum chaperone for the Wnt pathway LRP5/LRP6 receptors, whose biallelic deficiency causes type 20, notable for reduced bone mineralisation — unlike the hypermineralisation pattern seen in most other types; and CCDC134, a secreted protein inhibiting phosphorylation of the ERK and JNK kinases of the MAPK (mitogen-activated protein kinase) pathway, whose impairment causes type 22 with cortical thickening despite normal mineralisation — a finding bringing this form of OI closer to syndromes caused by RAS-MAPK pathway dysregulation, such as Noonan syndrome and neurofibromatosis type 1. More recently, biallelic PHLDB1 variants were associated with an autosomal recessive type 23 (OMIM 620639), initially described in five children from two Turkish families with mild OI and regressive spondylometaphyseal abnormalities (Tuysuz et al., Journal of Medical Genetics, 2023; doi:10.1136/jmg-2022-108763), raising the number of described types to at least 23 [6]. Outside the numbering system, mutations in PLOD2, causing Bruck syndrome, may also cause isolated OI with a type III phenotype (Puig-Hervas et al., 2012), and PLS3 mutations were identified as a cause of X-linked OI/osteoporosis (Van Dijk et al., 2013), an inheritance pattern not previously described in this disease [5]. In a Dutch cohort of 28 patients from 11 families with PLS3 variants (18 men and 10 women), men had a lumbar Z-score of −2.8 ± 1.7 and a median of 12 fractures, whereas most women had normal bone mineral density (median of 2 fractures); mildly blue sclerae (around one-third in both sexes), joint hypermobility and skin hyperlaxity bring this entity closer to mild OI, without a relationship between variant type or location and severity; all men and 40% of women received bone therapy, predominantly oral bisphosphonates, with a mean 16.6% increase in lumbar bone mineral density in adult men [30]. Spranger also mentions SEC24D among genes described in OI [21].

2.4 Ultrastructural histological findings

From a histomorphometric perspective, most OI types paradoxically show bone hypermineralisation — an increase in the mean degree of collagen fibril mineralisation which, rather than conferring greater strength, makes bone denser but more brittle because it reduces bone's ability to deform and absorb energy before fracturing, contributing directly to fragility. Types 5 and 6 are exceptions to the general bone lamination pattern: type 5 displays an irregular mesh-like lamellar pattern under polarised light [31], whereas type 6 shows the characteristic “fish-scale” pattern, with unmineralised osteoid accumulation and elevated alkaline phosphatase in childhood. Types 14 (TMEM38B) and 20 (MESD) are additional exceptions to the hypermineralisation rule, notable for low bone turnover and reduced mineralisation, respectively. In type 11 OI (FKBP10), Spranger describes a distorted lamellar structure and fish-scale-like pattern, with elevated serum alkaline phosphatase [32].

2.5 Germline mosaicism

A pathophysiological concept with direct implications for genetic counselling is germline mosaicism (or gonadal mosaicism): a condition in which only a fraction of the reproductive cells of one clinically unaffected parent carries the pathogenic mutation, without its being present in that parent's somatic tissues in a sufficient proportion to cause clinical manifestations. This phenomenon explains the occurrence of more than one child affected by severe OI forms (types II and III) in couples without a previous family history of the disease [22]. In Pyott and colleagues' study (2011), the parental mosaicism rate in families in which a dominant variant was identified in a first child with lethal OI was 16%, but the recurrence rate actually measured after the first affected pregnancy was 1.3%; after two or more affected children, recurrence rose to 32% (27% in families with parental mosaicism, 31% in recessive forms) [10]. The historically estimated value was around 7% in Cole and Dalgleish's work (1995) [11], suggesting a high gonadal mosaicism rate, and Spranger reports, for the IIB/III phenotype, an empirical recurrence risk of 7% to 8%, explained by germline mosaicism, homozygosity for COL1A2 mutations and recessive inheritance [25]. This finding alone justifies recommending molecular prenatal diagnosis in subsequent pregnancies of couples with a child previously affected by a severe OI form, even without a clinical diagnosis in the parents.

2.6 Distribution between inherited cases and de novo mutations

The distribution of cases between inherited forms and forms resulting from de novo mutations, that is, mutations arising for the first time in the patient without either parent being a carrier, reveals that approximately 60 per cent of mild cases are due to de novo mutations, a proportion rising to nearly 100 per cent in lethal and severe cases [3, 7]. This pattern is explained by natural selection: individuals with lethal or severely disabling forms rarely reproduce, such that most severe cases are new mutational events in each generation. Many individuals diagnosed with milder forms (type I and some with type IV) have an affected parent [7].

---

3. Clinical Spectrum and Manifestations by System

Regarding clinical presentation, in addition to fractures, patients may have mild short stature, muscle weakness, joint laxity, scoliosis, dentinogenesis imperfecta and hearing loss; skin is thin, with prominent vessels and easy bruising; more severe complications due to bone softening, such as platybasia and basilar invagination, rarely occur [5]. The Brazilian source lists possible manifestations: excessive, atypical or spontaneous fractures; bone deformities; short stature; scoliosis; bluish or greyish sclerae; basilar skull deformities, which may cause nerve compression and neurological symptoms; progressive postpubertal hearing loss; dentinogenesis imperfecta (opalescent teeth that wear rapidly); ligamentous laxity and other signs of connective tissue abnormality; Wormian bones; chest deformity; and clinodactyly (abnormal curvature of the fingers) [7].

3.1 Skeletal system

Bone fragility with recurrent fractures following disproportionately mild trauma or even in the absence of identifiable trauma is the cardinal and universal sign of osteogenesis imperfecta. The long bones of the limbs — femur, tibia, radius and ulna — are most frequently affected, although any bone may fracture. A clinical finding of high differential diagnostic value is the predominantly diaphyseal location of OI fractures: metaphyseal fractures are rare in osteogenesis imperfecta, contrasting with the metaphyseal predominance observed in childhood non-accidental injury. Fractures are concentrated in childhood, particularly during the prepubertal period of greater physical activity while the skeleton is still growing, become less frequent after puberty, especially in types 1 and 4 — a reduction attributed to increased sex steroids promoting more efficient bone remodelling — and increase again in later adulthood, particularly in postmenopausal women and men over 70 years, when bone remodelling becomes unfavourable again [22]. Inactivity, pregnancy, lactation and menopause increase fracture probability, and scoliosis progresses during puberty [21]. In the French national protocol, the frequency of spinal deformities ranges from 20% to 80% depending on the series and age, with loss of height and restriction of lung volumes, and pregnancy may reduce height by a few centimetres [8]. In severe forms, fractures may already be present in utero and are visible on the second-trimester foetal anomaly ultrasound scan. Fractures heal at a normal rate in most patients, although pseudarthroses may occur [25].

The phenotypic pattern varies widely according to type. In the classic non-deforming form (type I), the typical patient has normal or slightly reduced stature, accumulating a few dozen to as many as a hundred fractures throughout life, all due to minimal trauma, without significant residual long bone deformity. In the progressively deforming form (type III), the most severe among forms compatible with prolonged survival, a single patient may accumulate around 200 fractures throughout life, with progressive deformity even in the absence of evident fractures — the result of abnormal bone remodelling combined with biomechanical forces exerted on already fragile bone. Adult stature in this form has a median of approximately 97.94 cm in women and 118.06 cm in men, values reduced relative to the general population [3].

In the perinatally lethal form (type II), manifestations already occur in utero: limbs are short, bowed and extensively fractured, frequently in abduction (“frog-leg position”, with abducted and externally rotated thighs), the chest is narrow and the rib cage has multiple rib fractures at different stages of healing, conferring the pathognomonic radiographic appearance of beaded ribs. The skull is large relative to the body, with wide fontanelles and a poorly mineralised calvarium (caput membranaceum). More than 60% of affected newborns die on the first day of life and around 80% in the first week, mainly from respiratory failure due to rib cage collapse; data from specific type 2A series place median survival at only around two hours, with 90% of deaths occurring within four weeks [24]. With modern neonatal intensive care resources, a minority of patients survive the immediate neonatal period, generally with intensive respiratory and nutritional support, occasionally evolving towards a phenotype compatible with type III.

The radiographic subdivision of type II into three variants — 2A, with broad crumpled long bones and continuously beaded ribs; 2B, with similarly broad crumpled bones but discontinuous or absent rib beading; and 2C, with thin fractured long bones and thin beaded ribs — has prognostic and differential diagnostic value, although the distinction between types IIB and III is now recognised as largely a matter of degree: neonatal lethality remains the most reliable defining criterion between these two categories, with considerable morphological overlap [9, 25].

The common variable form with normal sclerae (type IV) occupies an intermediate position, with mildly to moderately reduced stature (adult median approximately 136.95 cm in women and 148.04 cm in men [3]; Spranger reports adult heights between 145 and 160 cm in some patients [23]) and generally less functional impairment than type III, although there is considerable phenotypic overlap between these forms, requiring molecular confirmation for precise classification in borderline cases [25].

Additional skeletal deformities include bowing of long bones — particularly femora and tibiae — present in moderate to severe forms; scoliosis, the lateral curvature of the spine resulting both from ligamentous laxity and weakening of spinal ligaments and from vertebral compression fractures, with curves above 30 degrees observed in more than one-third of type III patients, contributing directly to adult respiratory restriction; platybasia, flattening of the skull base associated with basilar invagination; and craniofacial abnormalities characteristic of severe forms, notably a prominent forehead, triangular face and relative midfacial retrognathism (backward projection of the midface) with dental malocclusion, resulting from relative microcephaly and mandibular weakening. Relative macrocephaly may be present in all types to varying degrees; shallow orbits, a small nose and low nasal base are seen in lethal forms [21].

3.2 Sclerae, cornea and globe

Bluish or greyish scleral colour — a feature frequently regarded as the disease's most emblematic sign — results from scleral thinning, allowing partial visualisation of the naturally pigmented underlying choroid; in type II the colour is dark blue [21]. Two clinical caveats with diagnostic importance should be emphasised. First, slightly bluish sclerae are a normal physiological finding in infants, when the sclera is naturally thinner — a circumstance complicating early diagnosis based solely on this sign; the age limit varies between sources, from 6 months [9] to 18 months of life. Second, blue sclerae are classically absent in certain OI types, notably types 5, 6, 7, 8, 12 and 19 and most rarer autosomal recessive forms, in which sclerae are typically white; therefore, absence of blue sclerae does not exclude OI. Conversely, blue sclerae persisting beyond 18 months in a child with recurrent fractures strongly reinforce diagnostic suspicion. In Marlowe and colleagues' series, four children clinically considered to have OI had normal biochemical studies, a false positive largely attributed to the use of scleral colour, an age-dependent feature, as a major diagnostic criterion [33]. Early arcus senilis is frequent in adults with type I and III OI [22, 25].

Reduced scleral and corneal thickness, a universal finding in the disease, also increases the risk of additional ophthalmological complications: ocular rupture or laceration following minor trauma, glaucoma, cataract, refractive errors and retinal disease. Additional occasional ocular findings described in the classic literature include posterior embryotoxon (peripheral corneal opacity), megalocornea and keratoconus [21].

3.3 Dentition

Dentinogenesis imperfecta (DI), when present, is highly suggestive of the diagnosis, representing a primary dysplasia of dentine — the deeper tooth layer immediately beneath enamel — with dentine and pulp hypoplasia [21]. Teeth have a grey, brown, yellowish or blue-grey colour, or a translucent appearance (amber, opalescent), wear and fracture easily, with enamel abrasion and fragmentation, show increased susceptibility to caries, irregular positioning and delayed eruption, and may display radiographic obliteration or calcification of the pulp chamber and spontaneous crown fractures [21, 22]. Both primary and permanent dentition may be affected, with a historical predilection for primary teeth. DI is frequent in types III and IV, uncommon in type I and classically absent in types 5 and 6 — information with important differential diagnostic value between subtypes. Dentinogenesis imperfecta shows broad intrafamilial and interfamilial variability in distribution and severity and is not useful for type I OI subcategorisation [22]. Teeth with deficient roots may also be visible on skull radiographs [9]. The French PNDS places the frequency of DI between 36% and 65% (up to 100% in some series), with preferential involvement of the incisors and first molars, discolouration from grey to brown, wear, attrition and dental fractures, morphological abnormalities of the crown and root and early canal obliteration; the severity of dental involvement does not correlate with that of bone involvement, and the craniofacial morphology favours dentomaxillary dysmorphoses with occlusal disturbances [8].

3.4 Hearing

Hearing loss affects on average 28% of individuals with COL1A1/COL1A2-related OI, with cumulative lifetime prevalence above 50% among adults with the classic non-deforming form — typically manifesting only from the second or third decade, potentially progressing to profound deafness, tinnitus and vertigo in the fourth to fifth decade of life [3, 21, 22]. Hearing loss occurs through two principal mechanisms. Among patients with hearing loss, approximately 43% have a conductive pattern — due to fixation of the stapes footplate, one of the middle ear ossicles, a mechanism resembling otosclerosis, or fractures of the middle ear ossicles, in which sound is not transmitted to the inner ear — 32% a sensorineural pattern, with damage to the auditory nerve or inner ear structures, and 24% a mixed pattern. Conductive loss predominates in individuals under 20 years, whereas sensorineural and mixed patterns become more common after this age. In terms of severity, 66.6 per cent of losses are mild, 22.7 per cent moderate and 10.6 per cent severe or profound; loss is bilateral in 55 per cent and unilateral in 45 per cent of cases [3]. Audiological assessment should be part of these patients' routine follow-up from childhood. The French PNDS describes deafness in 22% to 58% of patients, progressive, usually bilateral and late, becoming apparent between 15 and 40 years of age, conductive (stapes involvement), sensorineural or mixed, of all degrees of severity; it is exceptional in children and affects about 40% of adults, and hearing loss due to otitis media with effusion appears more frequent in young children with OI than in the general population [8].

3.5 Central nervous system

The most relevant neurological manifestations are basilar invagination (basilar impression) and platybasia. Basilar invagination is upward migration of the foramen magnum margins, with displacement of the odontoid process of the second cervical vertebra towards the posterior fossa — the upper cervical vertebrae sliding into the skull base — progressively compressing the brainstem and potentially causing obstructive hydrocephalus and syringomyelia [21]. It occurs in around 25% to 37% of type III patients; regarding type IV, sources differ — it is described as rare in GeneReviews, whereas Spranger reports posterior fossa compression syndromes due to basilar impression in 30% of cases [3, 23]. Symptoms include occipital and cervical headache, sensory deficits in the C2 dermatome, paraesthesiae (tingling) in the fourth and fifth fingers, numbness on the medial forearm, and Lhermitte's sign — an electrical sensation descending the spine on neck flexion. Severe cases may involve sleep apnoea and death due to medullary compression. Intelligence is normal in the great majority of patients, in both deforming and non-deforming forms [21]; cognitive impairment is rarely associated with intracranial haemorrhage, an uncommon event in the disease except in the specific context of perinatally lethal type II, in which severe macrocephaly and primary cortical development and neuronal migration abnormalities — agyria, hippocampal malrotation, abnormal neuronal lamination, perivascular gliosis, diffuse haemorrhage and periventricular leukomalacia — may occur, described in histopathological series of affected foetuses and neonates (Emery et al., 1999) [21]. Hypopituitarism has been described in type III, probably due to traumatic pituitary stalk injury at birth [25], and heterozygotes for WNT1 variants may combine central nervous system abnormalities with increased fracture susceptibility [21].

3.6 Respiratory system

Pulmonary function is impaired by extrinsic factors — short stature, immobility, scoliosis, rib fractures, muscle weakness and rib cage deformity — and intrinsic factors related to the collagen defect itself in the parenchyma and chest wall, resulting in a restrictive pattern. Forced vital capacity (FVC) and forced expiratory volume in one second (FEV1) are markedly reduced in progressively deforming forms. Complications include neonatal pulmonary hypoplasia (a central determinant of type II mortality, in which thoracic cavity size predicts lethality [21]), pulmonary arterial hypertension [25] and chronic restrictive lung disease, an important cause of adult morbidity and mortality in severe forms.

3.7 Cardiovascular system

Emerging data indicate increased risk of structural cardiovascular disease, including aortic root dilatation — generally non-progressive, sometimes segregating in type I OI families, but with documented aortic dissection cases requiring surgical correction [21, 22] — arterial dissection, aneurysms and mitral and aortic valve regurgitation. Mitral valve prolapse is the cardiac finding most frequently cited in the literature, with higher prevalence in adults [9], although most valve abnormalities are trivial or mild; a minority subgroup of patients has clinically relevant valve dysfunction. The Brazilian source notes that type 3 patients may have cardiovascular abnormalities [7]. In the absence of a previous consensus on assessment and management of cardiovascular disease in OI, a multidisciplinary group published in 2025, based on a systematic review and modified Delphi technique, 14 clinical recommendations for adults with OI, also identifying knowledge gaps and research priorities [34]. The French PNDS stresses that cardiovascular abnormalities, described in all types, are generally mild and slowly progressive, usually discovered incidentally in adults, mainly affect the valves of the left heart chambers, the aortic root and the descending aorta, and do not correlate with bone severity; pulmonary hypertension should be sought on echocardiography in patients with respiratory failure on nocturnal non-invasive ventilation [8].

3.8 Skin, ligaments and vessels

Skin is frequently thin, smooth, soft, slightly hyperelastic and translucent, with widened scars (broad hypertrophic scars in type III), inguinal or umbilical hernias, and a greater tendency to bruising after the slightest trauma due to increased capillary fragility, microvascular fragility and poor peripheral connective tissue support [21, 22, 25]. Joint hypermobility (ligamentous laxity), present to some degree in virtually all patients (100% according to Smith's [21]), contributes to joint instability, dislocations and partial dislocations — notably radial head dislocation, particularly characteristic of type 5 — and early development of adult degenerative osteoarthritis. Muscle hypotonia is described particularly in severe forms [21, 25]. Standardised hypermobility assessment uses the Beighton Score, a standardised nine-point scale whose criteria include the ability to bend the fifth finger backwards more than ninety degrees, bend the thumb to touch the forearm, hyperextend the elbows and knees more than ten degrees and place the palms on the floor with the knees straight; scores above 5 suggest generalised hypermobility, but, as around 10% of the general population scores above five on this scale, isolated ligamentous laxity does not confirm OI and should be interpreted alongside other findings [9].

Bleeding tendency. Patients with OI appear to have an increased bleeding risk, particularly during surgery and after moderate trauma, including intracranial haemorrhage; the mechanism remains poorly explained and the bleeding time is prolonged. In the presence of a clear bleeding tendency, the French PNDS recommends haemostasis testing to look for an associated abnormality, in particular von Willebrand factor deficiency, which is common in the general population; perioperative management is described in section 7.11 [8].

3.9 Chronic pain and gastrointestinal and urinary manifestations

Chronic pain affects more than 40% of OI patients, with prevalence above 70% among elderly people with the disease, and is multifactorial in origin — related to fractures, deformities and a myofascial component — predominantly affecting the lumbar region, followed by pain in multiple joints. Limitations in walking and participation in physical and social activities, and increased school and work absenteeism, are direct consequences of chronic pain. Gastrointestinal manifestations include constipation, diarrhoea, non-specific abdominal pain and gastro-oesophageal reflux, more frequent than in the general population, particularly in adults; urinary incontinence and bladder dysfunction have been reported in children, possibly related to collagen abnormalities in the bladder wall [3]. In severe forms, infant feeding difficulties, including pain during sucking and swallowing, reflux and muscle weakness, may compromise caloric intake. Additional occasional findings include syndactyly, hydrops fetalis and hydrocephalus in type II [21, 24].

3.10 Comparative table of classic types (expanded Sillence classification)

Data from GeneReviews [3], Spranger et al. [22–25, 31] and Oxford Desk Reference (table 2.11, adapted from Rauch & Glorieux, 2004) [9]. The range of clinical severity in OI is a continuum; categorisation is useful for determining prognosis and assessing treatment [9].

3.11 Description by type

OI classification guides prognosis and therapeutic planning. Each type is described in detail below.

Type 1 — Classic non-deforming form with blue sclerae (OMIM 166200). Type 1, listed in the Online Mendelian Inheritance in Man database — abbreviated OMIM — under number 166200, also termed the classic non-deforming form with blue sclerae, is the most common of all types (2 to 5 cases per 100,000 people) and follows autosomal dominant inheritance, with 100 per cent penetrance although variable expressivity [9]. It is caused almost invariably by loss-of-function mutations, including nonsense, frameshift or splice-site mutations, in COL1A1 (17q21.33), resulting in type 1 collagen haploinsufficiency; rarely by COL1A2 (7q21.3). The affected protein is type I alpha-1 collagen or, rarely, type I alpha-2 collagen. Historical synonyms include osteogenesis imperfecta tarda and Van der Hoeve syndrome [22]. Clinically, patients have normal stature or a slight reduction, with a few to around a hundred lifetime fractures, all resulting from minimal trauma, without long bone deformity. Sclerae are blue or greyish due to their reduced thickness allowing visualisation of the underlying choroid. Dentinogenesis imperfecta, characterised by grey or brown teeth with a translucent appearance and increased susceptibility to fractures and wear, is uncommon in type 1. Hearing loss may occur in childhood but is more characteristic of adulthood, affecting more than 50 per cent of adults with this form. Radiographically, tubular bones are straight, with mild bowing correcting spontaneously with growth, and prognosis is good [5]. Wormian bones, additional sutural skull bones with more than 10 elements and a diameter greater than 6 millimetres by 4 millimetres, are present in approximately 35 per cent of type 1 patients, in a mosaic pattern, usually in the lambdoid and posterior sagittal sutures [22]. Generalised osteopenia, thin tubular bone and rib diaphyses, thinning of bone cortices and sparsely trabeculated cancellous bone may also be observed [22]. Codfish vertebrae are seen in adults. Bone mineral density measured by bone densitometry, DXA, an abbreviation for dual-energy X-ray absorptiometry, shows average lumbar spine Z-scores between −1.52 and −1.4 in young people aged 9 to 18 years. At birth, bone morphology is usually normal, but there may be Wormian bones, mild osteopenia, thin ribs and mild femoral bowing; fractures are rare in the first year and become more frequent in childhood; bowing deformities tend to improve with age; and cardiovascular problems due to aortic root dilatation and valve incompetence may occur but are not frequent [22]. There are usually no prenatal morphological manifestations detectable by ultrasound [22]. Bone abnormalities in types I and IV are frequently identical; differentiation between the two phenotypes may be desirable given type IV's slightly less favourable prognosis [22].

Note on Wormian bones. Worm's or Wormian bones, also known as sutural bones, are small supernumerary bone fragments developing within the sutures (fibrous joints) of the human skull. These bones are named after Danish physician Ole Worm, who described them in detail in the seventeenth century. They occur due to isolated ossification centres appearing in addition to normal growth centres of the cranial bones. They are most frequently found in the lambdoid suture (separating the occipital bone from the parietal bones at the back of the head). They are technically considered anatomical variations, meaning that not everyone has them. Some theories suggest that mechanical stress or intracranial pressure during foetal development may stimulate the appearance of these extra ossification centres to “fill” spaces in the sutures.

Type 2 — Perinatally lethal form (OMIM 166210). Type 2, recorded in OMIM under number 166210 and termed the perinatally lethal form, is the most severe disease manifestation and results almost always from dominant negative mutations, particularly glycine substitutions, in COL1A1 or COL1A2, arising predominantly as de novo mutations; inheritance is autosomal dominant. Manifestations appear in utero. The newborn is typically small for gestational age, has intensely blue sclerae, prominent eyes, a small nose with a depressed nasal base, a skull large relative to the body with softened palpable fontanelles due to deficient mineralisation, usually normal-looking hands and feet, inguinal hernias and, in some cases, hydrops fetalis [24]. Limbs are short, curved and extensively fractured, frequently adopting a frog-leg position with lower-limb abduction. Radiologically, the pathognomonic finding is beaded or necklace-like ribs, resulting from multiple fractures and bone calluses along the rib axis. Long bones are thick, compressed and angulated — rectangular femora with an undulating “accordion-like” outline and anteriorly angulated tibiae and fibulae — vertebral bodies show generalised height loss due to collapse, and the skull has severely reduced mineralisation [24]. The condition is lethal in the perinatal period, with rare survival transitioning to a type three phenotype; many patients are stillborn [24]. Type 2 is subdivided into three radiological variants: type 2A (synonyms: neonatal lethal thick-bone OI; Vrolik type), with broad crumpled long bones and continuously beaded ribs due to healing fractures; type 2B, with broad crumpled long bones but ribs with discontinuous or absent beading, or intact ribs; and type 2C, with thin fractured long bones and thin beaded ribs. Type IIC (dense bone variant) is further characterised by a small triangular face with protruding eyes and hypertelorism, relatively long twisted extremities, gross calvarial and facial demineralisation, distorted scapulae, long angulated ischia, increased density of metaphyseal margins and fracture sites with a stippled appearance, and an almost normal spine; heterozygous mutations in the COL1A1 C-propeptide region were detected (Takagi et al., 2011), and autosomal recessive inheritance was proposed based on a family with three affected siblings of unaffected consanguineous parents (Sillence et al., 1984); it is rare and probably a heterogeneous phenotype rather than a genetic entity [35]. In all cases, death is caused by respiratory failure resulting from rib fractures. The responsible mutations are glycine substitutions in type one collagen chains, producing the dominant negative effect described in the pathophysiology section.

Type 3 — Progressively deforming form (OMIM 259420). Type 3, recorded in OMIM under number 259420, the progressively deforming form, is the most severe non-lethal form compatible with prolonged survival and causes severe lifelong functional impairment. It is most commonly caused by de novo dominant negative mutations in COL1A1 or COL1A2, although autosomal recessive forms associated with other genetic causes, such as mutations in CRTAP, LEPRE1, PPIB, FKBP10, SERPINH1, SERPINF1 and WNT1, may produce similar presentations; a recessive form has been described in southern African populations and reported in Irish and other consanguineous families [9]. Fractures are already present at birth or occur with minimal handling in the first weeks of life. A single patient may accumulate around 200 lifetime fractures, occurring extremely frequently during childhood phases of greater physical activity. Clinically, type three has markedly reduced stature, with progressive limb and spinal deformity resulting from lifelong fractures, even in the absence of evident fractures, due to abnormal bone remodelling and biomechanical forces on weak bones. Severe scoliosis may progress to respiratory difficulty, the principal cause of mortality in these patients [9]. The facial phenotype is characteristic, with frontal and temporal bossing, a prominent forehead, triangular face, backward projection of the midface, wide infant fontanelles and dental malocclusion; the triangular facies results from relative microcephaly and weakening of mandibular bones [25]. Dentinogenesis imperfecta is frequently present, and sclerae are greyish (bluish in infancy and white later, according to Spranger [25]). Hearing loss usually begins in adolescence. Basilar impression, a condition in which the skull base invaginates the foramen magnum, pressing on the brainstem, occurs in 25 to 37 per cent of type 3 patients and represents one of the most severe neurological complications. Most patients need a wheelchair or other assisted mobility devices but can lead an active life [25]. Radiologically, fractures may occur before birth, and tubular bones, sometimes thick at birth with a beaded appearance, become progressively thin and curved; Wormian bones are present in 96 per cent of cases, platybasia in 70 per cent, codfish vertebrae are common, the pelvis may assume a triradiate shape with acetabular protrusion, and long bone metaphyses show the so-called popcorn or honeycomb appearance resulting from irregular growth cartilage calcification [25]. Bone densitometry reveals lumbar spine Z-scores between −2.79 and −4.33 in young people aged 9 to 18 years. In the IIB/III phenotype, many infants die at birth or in the first year from respiratory failure, and prognosis may be partially predicted by analysing neonatal bone abnormalities (crumpled long bones); bone fragility improves at puberty, and survival to adulthood is expected with bisphosphonate therapy [25].

Type 4 — Common variable form with normal sclerae (OMIM 166220). Type 4, recorded in OMIM under number 166220 and termed moderately deforming or the moderate form with normal sclerae, is the most variable of all forms in clinical expression (Spranger synonyms: OI tarda, Lobstein disease, common variable OI) [23]. It follows autosomal dominant inheritance and is caused by glycine substitutions or other structural mutations in COL1A1 or COL1A2; according to Spranger, with rare exceptions, COL1A2 mutations produce a type IV phenotype [23]. Stature is mildly to moderately reduced, with an adult median of 136.95 centimetres in women and 148.04 centimetres in men, with mild to moderate scoliosis. Dentinogenesis imperfecta is common and may be mild. Sclerae are normal or slightly greyish at birth, progressively lightening with age (greyish to white). Conductive or mixed hearing loss may occur in adolescence or adulthood. Basilar impression is described as rare in GeneReviews [3], but Spranger reports posterior fossa compression syndromes in 30% [23]. Radiologically, Wormian bones are present in 78 per cent of cases and platybasia in 20 per cent. Bone densitometry Z-scores range from −1.84 to −2.29 in young people aged 9 to 18 years. Fractures may occur as early as the third trimester of pregnancy and in the perinatal period but much less frequently than in type 3, and tend to occur later than in type three; incidence declines at puberty [23]. There is phenotypic overlap with type three, but generally less functional impairment. At birth, generalised osteopenia without fractures or Wormian bones may be the only manifestation; congenital femoral angulation occasionally leads to an erroneous “kyphomelic dysplasia” diagnosis, with spontaneous improvement of deformities, and some cases progress in adulthood towards a type III radiological appearance [23]. Spranger proposes a genotype–phenotype correlation: cases with α1(I) chain abnormalities and blue sclerae usually have phenotype I and a good prognosis; cases with α1(I) abnormalities and white or greyish sclerae more frequently have phenotype IIB/III and a poor prognosis; cases with α2(I) abnormalities and white sclerae should be assigned to type IV, with a relatively good prognosis [23].

Type 5 — Form with a calcified interosseous membrane and hyperplastic callus (OMIM 610967). Type 5, recorded in OMIM under number 610967, has a phenotype ranging from mild to severe. It follows autosomal dominant inheritance and is caused uniquely by a gain-of-function mutation, the c.-14C>T variant in the 5' untranslated region of IFITM5 (11p15.5), encoding interferon-induced transmembrane protein five (BRIL). This mutation creates an additional initiation codon adding the pentapeptide MALEP (methionine–alanine–leucine–glutamic acid–proline) to the protein's N terminus, altering its function during the osteoblast mineralisation phase. It is the only OI form not caused by collagen defects that follows dominant inheritance [7]. Type 5 is clinically distinguished by hyperplastic callus, excessive uncontrolled formation of reparative bone tissue at healing fracture sites, particularly during periods of rapid growth, with limb swelling and pain, potentially mimicking bone tumours. Another unique characteristic is progressive calcification of the forearm (and leg) interosseous membrane, limiting wrist and hand pronation and supination and potentially leading to radial head dislocation [31]. Spranger also describes ligamentous laxity, particularly in the elbows and fingers, scoliosis, dysmorphic features (a short upturned nose, small mouth, prominent chin), neonatal rickets-like metaphyseal abnormalities followed by sclerotic widened metaphyses with dense bands, and a poorly mineralised calvarium; Spranger's text describes Wormian bones as absent, although figures in the same chapter show patients with multiple Wormian bones [31]. Sclerae are white and there is no dentinogenesis imperfecta. Histologically, bone has an irregular mesh-like lamellar pattern with hypermineralisation. Around 5% of non-lethal OI may correspond to type V, with highly variable interfamilial and intrafamilial expressivity [31]. In a cohort of 17 individuals from 12 families, all carrying the same c.-14C>T variant, 13 had interosseous membrane calcification, 14 radial head dislocation, only 10 hyperplastic callus and 9 long bone bowing, 11 walked without assistance and 1 had mild unilateral mixed hearing loss, and bone mineral density varied greatly even within families — so absence of some typical signs does not exclude the diagnosis [36]. As the variant is in the 5' untranslated region [36], an assay covering only coding exons may not detect it; molecular testing should therefore specifically include position c.-14.

Type 6 — Form with a mineralisation defect (OMIM 613982). Type 6 has moderate to severe severity. It follows autosomal recessive inheritance and is caused by biallelic loss-of-function mutations in SERPINF1 (17p13.3), encoding pigment epithelium-derived factor (PEDF), an endogenous angiogenesis inhibitor. Patients have a normal appearance at birth, with progressive onset of fractures in childhood. Alkaline phosphatase is elevated in childhood, a distinguishing laboratory feature of type 6 relative to other forms. Long bones are broad and curved and, in adulthood, “popcorn” epiphyses may develop [5]. Sclerae are white and there is no dentinogenesis imperfecta. Bone histology reveals a so-called fish-scale pattern of bone lamination, with accumulation of unmineralised osteoid tissue and paradoxical hypermineralisation of the remaining bone. Serum PEDF is undetectable in these patients.

Type 7 — Form due to CRTAP deficiency (OMIM 610682). Type 7 has autosomal recessive inheritance, a variable and generally severe phenotype with some lethal forms, and is caused by mutations in CRTAP (3p22.3), encoding cartilage-associated protein, an essential component of the collagen prolyl 3-hydroxylation enzyme complex. The phenotype includes slightly reduced stature, proximal limb shortening, also called rhizomelia — shortening of the segment between shoulder and elbow and between hip and knee — and coxa vara, a hip deformity with a reduced femoral neck angle (varus of the neck–shaft angle). Sclerae are white and there is no dentinogenesis imperfecta. The most severe forms overlap with lethal type 2.

Type 8 — Form due to prolyl 3-hydroxylase deficiency (OMIM 610915). Type 8 also has autosomal recessive inheritance and is caused by mutations in LEPRE1 (1p34.2), encoding prolyl 3-hydroxylase 1, the catalytic component of the same prolyl 3-hydroxylation complex. The phenotype ranges from severe to lethal, overlapping with type 2 and severe type 3, with white sclerae. Bones appear severely hypomineralised on radiographs. A West African LEPRE1 founder variant causes lethal recessive OI: it is carried by around 0.4% of African Americans in the mid-Atlantic region of the United States (predicted incidence 1 in 260,000 births) and 1.48% of individuals in Nigeria and Ghana (predicted incidence 1 in 18,260 births, equal to that of dominant de novo OI), and was not detected in Africans from neighbouring countries; all carriers share a common haplotype, and the variant's origin was estimated between AD 1100 and 1350 [37].

Types 9 to 23 — Additional recessive and X-linked forms. Type 9 (OMIM 259440) is autosomal recessive, caused by mutations in PPIB, also called CYPB (15q22.31), encoding cyclophilin B, the third component of the prolyl 3-hydroxylation complex and a chaperone acting in the same enzyme complex as CRTAP and LEPRE1, with a phenotype ranging from mild or moderate to perinatally lethal forms. Type 10 (OMIM 613848) is autosomal recessive, caused by mutations in SERPINH1 (11q13.5), encoding the collagen-specific endoplasmic reticulum chaperone HSP47; the form is moderate to severe or lethal, with broad curved long bones, and patients have blue sclerae [5]. Type 11 (OMIM 610968) is autosomal recessive and progressively deforming, caused by mutations in FKBP10 (17q21.2), encoding the binding protein FKBP65, an endoplasmic reticulum chaperone; it has a moderate to severe phenotype with white sclerae and is also associated with Bruck syndrome, characterised by osteogenesis imperfecta with congenital joint contractures, and Kuskokwim syndrome, with specific contractures in Indigenous Alaskan populations [32]. Type 12 (OMIM 613849) is caused by mutations in SP7 (12q13.13), encoding the transcription factor osterix, a regulator of osteoblast differentiation; inheritance is autosomal recessive in the reference table, although autosomal dominant inheritance has also been described (see below); patients have moderate bone bowing, white sclerae, delayed tooth eruption and craniofacial abnormalities. In a second sibship, homozygous for SP7 NM_001173467.3:c.946C>T p.(Arg316Cys), with osteoporosis, low-impact fractures and short stature, two siblings developed progressive moderate to severe and severe to profound hearing loss due to otospongiosis, with deficient mineralisation of the ossicles and petrous temporal bone, and bone biopsy showed marked cortical porosity and high trabecular turnover, leading the authors to recommend early audiological screening in this subtype [38]. Dominant inheritance was demonstrated in two young adult siblings heterozygous for SP7 NM_001173467.3:c.1019A>C p.(Glu340Ala), in an osterix zinc finger domain, with diaphyseal fractures, poor healing, scoliosis, dental malocclusion, very low cortical volumetric bone mineral density and low bone turnover with reduced osteoblast function [39]. Type 13 (OMIM 614856) is autosomal recessive and progressively deforming, caused by mutations in BMP1 (8p21.3; BMP1/mTLD protein), encoding a metalloproteinase required for collagen C-terminal propeptide cleavage; the phenotype ranges from mild to severe with bone hypermineralisation, bluish sclerae and severe kyphoscoliosis [5]. In most of the fewer than 20 patients described up to 2016, bone fragility was associated with increased bone mineral density; in a British series of three children (two siblings homozygous for c.2188delC, affecting only the short BMP1 isoform, with normal areal density, and one compound heterozygote with a high-bone-mass phenotype), all hypermobile and without dentinogenesis imperfecta, intravenous bisphosphonate was stopped in two and reduced in the third because of concern that increased bone stiffness could promote “chalk-stick” fractures, with the indication for antiresorptive therapy in this form remaining uncertain [40]. Type 14 (OMIM 615066) is autosomal recessive, caused by mutations in TMEM38B (9q31.2), encoding the endoplasmic reticulum TRIC-B cation channel regulating calcium influx, initially described in Saudi Arabian families (Shaheen et al., 2012) and Israeli Bedouin families (Volodarsky et al., 2013) [5]; the phenotype is variable, from asymptomatic to severe, with white sclerae, and is distinguished by low bone turnover, a unique characteristic among osteogenesis imperfecta types. Type 15 (OMIM 615220), progressively deforming, may have autosomal recessive inheritance when biallelic, or autosomal dominant inheritance when heterozygous, and is caused by mutations in WNT1 (12q13.12), a canonical Wnt/β-catenin pathway ligand required for regulation of bone mass and osteoblast differentiation; heterozygotes have early-onset osteoporosis, whereas homozygotes have fractures, short stature and, in some cases, variable neurological impairment, developmental delay and learning difficulties [7]. Type 16 (OMIM 616229) is autosomal recessive and progressively deforming, caused by mutations in CREB3L1 (11p11.2), encoding the transcription factor and endoplasmic reticulum stress transducer called OASIS; the phenotype ranges from mild to perinatally lethal, with blue sclerae. Type 17 (OMIM 616507) is autosomal recessive, caused by mutations in SPARC (5q33.1), encoding osteonectin, a matricellular glycoprotein of the extracellular matrix binding collagen in a calcium-dependent manner; patients have a progressively severe form with muscle hypotonia and bone hypermineralisation resulting in wheelchair dependence. Type 18 (OMIM 617952) is autosomal recessive, caused by mutations in TENT5A, also known as FAM46A, a non-canonical polyadenylase increasing the stability and translation of COL1A1, COL1A2 and SERPINF1 messenger RNAs; the phenotype is severe with congenital limb bowing and progressive scoliosis. Type 19 (OMIM 301014) is the only type with X-linked recessive inheritance, caused by mutations in MBTPS2, encoding the S2P protease of the regulated intramembrane proteolysis pathway; male patients have short stature, fractures, bone bowing, a barrel chest and scoliosis, with normal sclerae. Type 20 (OMIM 618644) is autosomal recessive, caused by biallelic mutations in MESD, encoding an endoplasmic reticulum chaperone for the LRP5 and LRP6 receptors; it has a lethal to severe phenotype with blue sclerae, Wormian bones, intellectual disability and dental abnormalities, notable for reduced bone mineralisation rather than the usual hypermineralisation pattern. Type 21 (OMIM 619131) is autosomal recessive, caused by mutations in KDELR2, a receptor responsible for retrograde protein transport from the Golgi apparatus to the endoplasmic reticulum; the phenotype is progressively severe with thin disorganised collagen fibrils. Type 22 (OMIM 619795) is autosomal recessive, caused by mutations in CCDC134, a secreted protein inhibiting ERK and JNK kinase phosphorylation; the phenotype is severe, with thickened bone cortices despite normal mineralisation, overlapping with syndromes resulting from MAPK-ERK pathway dysregulation, such as Noonan syndrome and neurofibromatosis type 1. Finally, type 23 (OMIM 620639), associated with biallelic PHLDB1 variants, has been described in few families and expands the nosology to at least 23 types [6]. Types 6 to 17 were numbered in Spranger's table according to their molecular defects, have variable phenotypes and are inherited as autosomal recessive traits — requiring inheritance of two altered gene copies, one from each parent — representing around ten per cent of cases [5].

---

4. Clinical Recognition and Warning Signs

Early recognition of OI depends on systematic integration of history, physical examination and radiographic findings, with particular attention to differentiation from childhood non-accidental injury — the most critical differential diagnosis clinically, ethically and medicolegally. Between one-third and half of all children sustain a fracture during childhood, and twenty per cent sustain two or more; fractures following trivial trauma are the most relevant clinical marker [9]. Severe OI forms have marked clinical and radiological abnormalities manifesting prenatally on ultrasound or neonatally; the structured clinical approach is primarily intended to diagnose milder forms, and when the possibility of non-accidental injury has been raised, questioning should be conducted sensitively [9].

4.1 History

When assessing fracture history, the clinician should investigate the following points. First: a three-generation pedigree should specifically investigate fractures, blue sclerae, dental problems, hearing loss, short stature and osteoporosis in relatives (fractures and loss of height with age, kyphosis), as well as age at onset of walking [9]. Second: the age at each fracture and whether fractures were present at birth should be determined — remembering that infants with arthrogryposis, a condition of multiple congenital contractures, sustain fractures during labour itself in 8% to 10% of cases, an important differential diagnosis of neonatal fractures. Third: the total number of fractures and which bones were affected should be recorded. Fourth: whether the traumatic event was witnessed and whether the reported history is consistent with the observed injury mechanism — the possibility of non-accidental injury, or NAI, should be systematically considered in every child with multiple fractures without this excluding a concurrent OI diagnosis a priori, since the two conditions may coexist. Fifth: the quality of healing of previous fractures should also be asked about — exuberant bone callus formation, characteristic of type 5, may be confused radiologically with a bone tumour. Sixth: whether there is limb deformity. Seventh: whether there are recurrent joint dislocations, reflecting OI ligamentous laxity, or joint contractures. Hearing loss and teeth that chip or break should also be investigated [9]. If NAI is suspected, the detailed perinatal history should include prematurity, because former preterm infants born below 1,000 g probably have an increased fracture risk up to 6 months, and nutritional deficiencies (scurvy, rickets) should be sought [41].

4.2 Physical examination

Physical examination should focus on the following findings. Stature is reduced in moderate to severe forms. Limb deformities are present in more severe forms and absent in mild forms. Blue sclerae result from reduced scleral thickness making the underlying choroid visible and are frequently described as more evident when fractures occur [9]. Infants up to six months have blue sclerae as a normal finding, complicating early diagnosis. Scoliosis is present in moderate to severe forms, resulting from ligamentous laxity and vertebral fractures. Ligamentous laxity should be assessed using the Beighton Score. Skin may be smooth, soft and bruise after the slightest trauma. Atypical scars or deficient wound healing raise suspicion of Ehlers–Danlos syndrome — abbreviated EDS — a differential diagnosis. Teeth should be examined for dentinogenesis imperfecta, manifesting as discolouration, deficient growth and defective dentine. Adults have a higher prevalence of mitral valve prolapse. Neurological assessment is indicated in the presence of spinal abnormalities or basilar invagination [9]. If abuse is suspected, growth parameters (height, weight and head circumference), anterior fontanelle size, photographic documentation of bruises and fractures and investigation for sparse hair (Menkes syndrome) should be recorded [41].

4.3 Warning signs

Warning signs that should strongly raise clinical suspicion of OI include: multiple fractures at different radiographic healing stages, associated with a disproportionately mild traumatic mechanism; bluish sclerae persisting beyond 18 months of life; Wormian bones on skull radiographs — small accessory sutural bones, more than ten in number with minimum dimensions of 6 by 4 millimetres each being highly suggestive of the diagnosis, although their absence does not exclude it, particularly in types 5, 12 and 19; dentinogenesis imperfecta; a positive family history of recurrent fractures, early-onset deafness or disproportionate short stature; and generalised joint hypermobility associated with the preceding findings. Any one of these signs alone has limited value, but their combination substantially increases the disease's pretest probability and justifies targeted radiological and molecular investigation. As a chronological feature, child abuse frequently presents in infants without independent mobility, whereas fractures in mild OI usually begin after mobility is acquired [22]; the Brazilian source states that, in mild forms, there are usually no fractures before the child begins crawling or walking [7]. Increased use of densitometry in young women has identified individuals with density so reduced that they may have a mild OI form; endocrine causes of premature osteoporosis should also be sought [9].

Pathway to suspicion in the French PNDS. The diagnosis is usually raised by the orthopaedic surgeon, paediatrician, radiologist or family doctor, through family screening or, prenatally, by the obstetrician or sonographer, and is confirmed by a physician specialised in constitutional bone diseases at a reference or competence centre, who organises the multidisciplinary team (including a psychologist, physiotherapist, occupational therapist and social worker). The PNDS lists as minor and inconstant signs, which strengthen the suspicion in the presence of repeated fractures with moderate trauma: blue-grey sclerae (about 80% of patients), ligamentous hyperlaxity, deformities (anterior sternal protrusion or, more rarely, pectus carinatum, kyphosis due to vertebral fractures, scoliosis, bowing of the lower limbs and, in severe forms, «sabre-blade» deformity of the upper limbs), relatively short stature, deafness (exceptional in children, about 40% in adults), transparent skin with easy bruising, DI and, rarely, valvulopathies; the same signs in a first-degree relative strengthen the hypothesis. When OI is suspected, the child and family should be referred to a reference or competence centre to confirm the diagnosis and initiate care [8].

---

5. Diagnosis: Criteria, Tests and Algorithms

Osteogenesis imperfecta is diagnosed through a combination of clinical, radiological and laboratory findings, confirmed by identification of a pathogenic variant through molecular analysis [3, 7].

5.1 Radiological investigation

A complete skeletal survey is a first-line investigation, documenting old and recent fractures at different stages of healing, deformities and the characteristic mineralisation pattern, and is important for distinguishing genetic conditions from child abuse; if NAI is suspected, it should be reviewed by a paediatric radiologist according to British Society of Paediatric Radiology criteria [9, 41]. Anteroposterior and lateral skull radiographs are essential for investigating Wormian bones — small irregular bones interposed in skull sutures — frequently the principal radiological clue in type I, in which the rest of the skeletal survey may be entirely normal [9]. Lateral spinal radiographs have a specific role in infants under 12 months with suspected mild disease, allowing identification of vertebral body biconcavity and wedge fractures before the therapeutic decision to start bisphosphonates [9].

Bone densitometry by dual-energy X-ray absorptiometry (DXA) — an investigation measuring bone mineral density — should begin around 5 years and be repeated according to clinical severity and therapeutic response; it may be reduced in moderate to severe forms but is frequently normal in type one, and should be interpreted together with a physician with a special interest in metabolic bone disease [9]. Normal lumbar spine Z-score reference values lie between −2 and +2, with values below −2 considered low for chronological age. In the classic non-deforming form, mean Z-scores range between −1.52 and −1.4; in the progressively deforming form, between −2.79 and −4.33; and in the moderate form, between −1.84 and −2.29 — thus with considerable overlap between groups, so densitometry assists longitudinal follow-up but does not replace clinical and molecular characterisation for classification purposes [3].

Basilar invagination is the skull base abnormality causing morbidity in severe forms [9]. If suspected, cranial computed tomography or magnetic resonance imaging with sections through the skull base is indicated in patients with platybasia, moderate to severe disease, or suggestive neurological signs and symptoms [7]. Flexion and extension cervical spine radiographs are recommended in cooperative children before sporting activities or elective anaesthetic/surgical procedures. Cranial computed tomography is recommended if abuse is suspected [41].

The French PNDS defines a minimum radiological work-up, guided by the clinical picture and subject to the ALARA principle (justification of the request and optimisation of the technique): anteroposterior and lateral skull radiographs and a lateral radiograph of the thoracolumbar spine, with anteroposterior radiographs of the long bones and rib cage when necessary, particularly in infants. The features sought are fracture sequelae, periosteal appositions on the long bones, thin cortices and excessive bone lucency, a short skull widened transversely, excessive numbers of Wormian bones, bowing of the long bones, platybasia, sternal protrusion, progressive gracility of the skeleton, «onion-bulb» metaphyseal widening, acetabular protrusion and basilar impression, defined by the position of the tip of the odontoid and the anterior arch of the atlas above Chamberlain's line. No isolated clinical or paraclinical element allows the diagnosis to be established [8].

For densitometry, the PNDS uses the Z-score in children and the T-score in adults, notes that device-specific reference standards exist from 3 or 5 years of age, that the examination is difficult in very young children and that a normal density does not exclude the diagnosis, and reserves it for doubtful forms and for the work-up before bisphosphonates. The 2006 HAS assessment cited in the PNDS supports two-site DXA (four studies, 70 children and adults, with lumbar ± femoral density lower than in unaffected individuals) and stresses that, when abuse is suspected, a normal density for age and sex does not rule out type I OI (one study, n = 63) and that a low density is not sufficient to establish OI or to exclude abuse; whole-body DXA is preferred in the presence of multiple vertebral or peripheral fractures, and paediatric interpretation should be left to specialised centres [8].

Initial assessment of severity (French PNDS). The initial work-up is adapted to age, the circumstances of diagnosis and severity: cervical spine radiographs in severe forms and limb radiographs in the presence of long bone fractures or bowing; general condition; height, weight and head circumference charts, body mass index and, in the presence of scoliosis or inability to stand, supine length and sitting height; functional status (walking distance, joint limitations, pain, hyperlaxity); dental status; respiratory function, with pulmonary function tests in forms with thoracic deformity or short stature and before bisphosphonates; hearing acuity; cardiac auscultation; and lifestyle, home accessibility and quality of life [8].

5.2 Laboratory investigation

A laboratory finding of central diagnostic importance — frequently undervalued — is that basic serum biochemical parameters, including calcium, phosphorus, vitamin D and alkaline phosphatase, are typically normal in classic osteogenesis imperfecta; alkaline phosphatase — an enzyme released by bone-forming cells — may rise transiently immediately after a recent fracture as part of normal bone repair, and the Oxford Desk Reference describes it as normal or elevated in OI [9]. This laboratory normality helps to distinguish OI from hypophosphatasia, a condition which clinically mimics OI but has markedly reduced or absent serum alkaline phosphatase, reflecting deficiency of the tissue-nonspecific isoenzyme (TNSALP), encoded by ALPL. Exceptionally, in type 6 OI (SERPINF1 deficiency), alkaline phosphatase is elevated from childhood and serum pigment epithelium-derived factor (PEDF) is undetectable — a specific biochemical finding assisting differentiation of this subtype.

Measurement of collagen degradation and bone resorption markers, such as urinary type I collagen N-telopeptide (NTx) — the most useful marker, reflecting the rate of osteoclastic activity — is useful for monitoring the therapeutic response to bisphosphonates [9]. Urinary amino acid testing aims to exclude homocystinuria, a relevant differential diagnosis of fractures and skeletal deformities; if abuse with a subdural haematoma is suspected, organic acid testing (glutaric aciduria), a full blood count with platelets and coagulation studies are added and, in males, copper and caeruloplasmin (Menkes syndrome) [41]. Hearing assessment should also be performed [9].

5.3 Molecular diagnosis

Blood should be collected for DNA storage, and molecular confirmation is particularly useful in prenatal situations [9]. Molecular diagnosis is the gold-standard confirmatory method and should be performed through next-generation sequencing, analysing all known OI genes simultaneously in a bone fragility gene panel — including at least COL1A1 and COL1A2, but ideally all known causative genes — through whole-exome sequencing or, in clinical settings of greater diagnostic complexity or phenotypic overlap with other skeletal dysplasias or bone fragility disorders, whole-genome sequencing [3, 9]. Sequencing detects more than 95% of known pathogenic variants; when the initial result is negative, exon or whole-gene deletion and duplication analysis should be performed using gene dosage techniques (MLPA, multiplex ligation-dependent probe amplification, or next-generation sequencing read-depth analysis), identifying an additional approximately 1% to 2% of causative variants — including complete COL1A1 deletions, complex structural rearrangements with simultaneous deletion and duplication, and contiguous deletions involving neighbouring genes, as already documented for COL1A1 and DLX3 in families with an overlapping phenotype of OI and tricho-dento-osseous syndrome [3]. In a series of 598 individuals from 487 families with a typical OI phenotype, sequencing of known genes identified the causative variant in 98% (97% in type I OI and 99% in moderate to severe forms); in type I OI all variants were dominant and restricted to COL1A1/COL1A2, whereas in moderate to severe forms dominant variants were found in COL1A1/COL1A2 (77%), IFITM5 (9%) and P4HB (0.6%), and recessive gene variants in 12%, particularly SERPINF1 (4.0%) and CRTAP (2.9%) [42].

The French PNDS, written in 2016, takes a more restrictive position: the diagnosis rests on a combination of clinical and paraclinical arguments, and molecular testing, performed with next-generation sequencing panels in accredited laboratories (list of the OSCAR network) and supplemented if necessary by MLPA, was not then indicated for diagnostic purposes, because it was time-consuming and of incomplete sensitivity (variant interpretation, false negatives). It is, however, increasingly offered to families to identify the gene and inform genetic counselling, and is recommended in particular for people with a severe form who are considering prenatal diagnosis and for parents of a child or foetus with a severe form who wish to have it in a future pregnancy; it should only be requested after a dedicated genetics clinic consultation, and the result only changes management in the forms less responsive to bisphosphonates (published data for SERPINF1 and IFITM5) [8]. The divergence from sources treating sequencing as routine confirmation is analysed in section 7.9.

The Brazilian source proposes a testing algorithm derived from Basel & Steiner (2009): when there is clinical suspicion (history, physical examination and radiological assessment indicative of OI), an NGS panel of OI genes is requested as the most cost-effective approach; if negative, testing is expanded to include other disorders with reduced bone density through exome sequencing; if this is also negative, clinical follow-up continues [7].

flowchart TD
A[Clinical suspicion: fractures with minimal trauma, blue sclerae, Wormian bones, DI, family history] --> B[Skeletal survey, AP/lateral skull X-ray, biochemistry: Ca, P, 25-OH-vit D, alkaline phosphatase]
B -->|Low ALP| H[Hypophosphatasia: ALPL testing]
B -->|Normal/elevated ALP| C[NGS bone fragility gene panel]
C -->|P/LP variant| D[Confirmed diagnosis: type, inheritance, counselling]
C -->|Negative| E[MLPA / CNV: COL1A1/COL1A2 deletions-duplications]
E -->|Negative| F[Exome / genome; consider collagen studies in fibroblasts]
F -->|Negative| G[Clinical follow-up and periodic reassessment]

Variant classification should follow American College of Medical Genetics and Genomics (ACMG) criteria, with the five standardised categories — pathogenic, likely pathogenic, variant of uncertain significance (VUS), likely benign and benign — considering population, computational, functional and family segregation criteria. Specific COL1A1 and COL1A2 variant databases, compiled by systematic literature review, currently catalogue more than 1,500 unique COL1A1 variants (approximately 55% classified as pathogenic, 26% as likely pathogenic and 14% as VUS) and more than 770 unique COL1A2 variants, with a similar distribution between classification categories, reflecting the disease's extensive allelic heterogeneity [43]. The ClinVar portal lists documented pathogenic variants in each gene.

Skin biopsy for fibroblast culture followed by biochemical analysis of synthesis and structure of type I collagen produced in vitro remains useful in selected cases — to confirm diagnosis or assist prenatal diagnosis, particularly when access to molecular genetics laboratories is difficult or sequencing results are inconclusive — although availability of collagen protein analysis is now restricted in most centres [9]. In a fibroblast biochemical analysis study in 262 children investigated for suspected non-accidental injury (Marlowe et al., 2002), 11 samples (around 4%) showed abnormalities in type I collagen quantity or structure consistent with OI, and in another 11 OI could not be excluded; referring physicians had clinically suspected OI in 6 of the 11 confirmed cases — illustrating both the utility and limitations of this approach in medicolegal settings [33, 41].

5.4 Prenatal diagnosis

In prenatal diagnosis, obstetric ultrasound scanning (USS) in an experienced reference centre is the principal tool for detecting severe forms: type II may be suspected before 20 weeks of gestation through visualisation of short curved fractured long bones, flattened beaded ribs and a malformed skull, with increased nuchal translucency [24]. In lethal OI, bone abnormalities may be observed in the first trimester on ultrasound at 13–14 weeks of gestation; at 16 weeks, femoral length is usually two or more weeks behind, calvarial mineralisation is virtually absent and ribs generally show fractures; in type III OI, limb length usually begins to fall below the growth curve between 17 and 18 weeks, requiring serial examinations to confirm the trend; milder forms may be detected later in pregnancy when fractures or deformities occur [7]. Smith's states that undermineralisation and deformities allow diagnosis as early as 12 weeks of pregnancy [21]. In type IIC, ultrasound shows short deformed extremities and a large unossified skull [35]. In ultrasound assessment of short limbs, bowing (OI with fractures, campomelic dysplasia, hypophosphatasia), bone density and mineralisation (OI, achondrogenesis, hypophosphatasia), fractures and skull mineralisation (reduced in hypophosphatasia, OI IIA/IIC and achondrogenesis) should be documented; in a series of 31 prenatally detected skeletal dysplasias (Parilla et al., 2003), six were OI [44]. In type I, there are usually no prenatal morphological manifestations detectable by ultrasound, and in at-risk pregnancies prenatal diagnosis may be obtained by molecular analysis in amniocytes or type I collagen protein studies in chorionic villi [22]. Chorionic villus sampling allows direct molecular diagnosis from 11 weeks of gestation when the familial variant is already known; amniocentesis, in turn, is not suitable for collagen protein studies, although it allows foetal DNA extraction for targeted molecular analysis when the variant is already known in the family [9]. If this is the first case in the family, more comprehensive molecular testing, such as a genetic panel or exome sequencing, may be performed [7].

The French PNDS organises prenatal diagnosis within a multidisciplinary prenatal diagnosis centre (CPDPN), with a consultation with a genetic counsellor and a psychologist offered to all couples and psychological support throughout the process. If the parental variant is known, molecular diagnosis is performed on chorionic villus sampling at 11–12 weeks of gestation or on amniotic fluid from 16 weeks; if it is not known, the PNDS does not accept molecular prenatal diagnosis, because of the time required for identification and the lack of prognostic value of the variant, and follow-up is imaging only, with ultrasound from 14–16 weeks and foetal bone CT, with adapted acquisition parameters, from 26–28 weeks. In women with a severe form, the diagnosis should be made as early as possible, because termination of pregnancy for medical reasons may be technically difficult, and preimplantation genetic diagnosis may be offered in selected cases [8].

5.5 Limitations of genetic testing and report content

The high sequencing detection rate in OI with a typical phenotype (98% in the series of 598 individuals) refers to patients with a firm clinical diagnosis studied in a reference centre; it is not the sensitivity of any panel applied to any child with fractures, and a negative result does not exclude OI in the presence of a convincing phenotype [42]. There are four predictable technical gaps. First, exon or whole-gene deletions (1–2% of causative variants) require MLPA, read-depth analysis or genome sequencing: in the Japanese family described in section 10, the exome without point variants suggested a deletion of around 57 kb through CNV analysis, and only the genome defined a 101 kb deletion including all of COL1A1 [3, 45]. Second, the recurrent type 5 variant, IFITM5 NM_001025295.3:c.-14C>T, lies in the 5' untranslated region and is not detected by a design restricted to coding exons [17, 36]. Third, intronic variants creating cryptic splicing sites lie outside most panels and may require RNA or collagen studies in fibroblasts [3, 9]. Fourth, parental mosaicism, present in 16% of parents in families with dominant lethal OI, may be below the detection threshold of blood Sanger sequencing, and mosaicism confined to the germline is undetectable in blood, so a negative parental study does not reduce recurrence risk to zero [10].

The report should state the gene, transcript with version, HGVS nomenclature, zygosity, ACMG/AMP classification and limitations (CNV, non-coding regions, mosaicism). MANE v1.5 transcripts are COL1A1 NM_000088.4, COL1A2 NM_000089.4, IFITM5 NM_001025295.3, SERPINF1 NM_002615.7, CRTAP NM_006371.5, P3H1 NM_022356.4, PPIB NM_000942.5, SP7 NM_001173467.3, WNT1 NM_005430.4, TMEM38B NM_018112.3 and PLS3 NM_005032.7; BMP1 has a MANE Select transcript (NM_006129.5) and a MANE Plus Clinical transcript (NM_001199.4), which matters because some variants affect only one isoform, such as c.2188delC described in the British series [17, 40]. Classic collagen literature numbers residues from the first amino acid of the triple helical domain: in COL1A1, this domain begins at residue 179 of the pre-pro-α1(I) chain, so HGVS numbering corresponds to helical numbering plus 178 [46]. Thus, the Caffey disease variant described as R836C corresponds to NM_000088.4:c.3040C>T, p.(Arg1014Cys), and the arginine-to-cysteine variant associated with Ehlers–Danlos forms, R134C in helical numbering, corresponds to c.934C>T, p.(Arg312Cys); COL1A2 has its own offset, which the laboratory should specify when citing older literature [46–48]. A variant of uncertain significance does not become pathogenic through response to bisphosphonates or partial segregation in a small family; extended segregation, confirmed de novo occurrence and biochemical collagen studies in fibroblasts are additional investigations that may support reclassification according to ACMG/AMP criteria [9, 43].

---

6. Differential Diagnosis

The most critical differential diagnosis, clinically and ethically/legally, is non-accidental injury (child abuse), also termed non-accidental violence. Non-accidental injury is suspected when an infant has several fractures of different ages without an explanatory history, a subdural haematoma or recurrent bruising; for an infant or child with fractures, the most important differential diagnosis is OI, and establishing the genetic diagnosis promptly and rigorously is essential to avoid inappropriate suspicion and investigation of child abuse [41]. Differentiation rests on three pillars: clinical, radiological and molecular. Osteogenesis imperfecta should always be considered when there are multiple fractures at different healing stages, fractures following disproportionately mild trauma, and inconsistency between the caregiver's account and clinical findings. Persistent blue sclerae, Wormian bones, thin ribs, clearly reduced bone density, dentinogenesis imperfecta, a positive family history of fractures and bone fragility, and molecular confirmation of a pathogenic variant in a causative gene favour OI [22]. Signs of inflicted trauma — suggestive skin marks, retinal haemorrhage, metaphyseal avulsion fractures — or a fracture pattern incompatible with caregivers' accounts, particularly in an infant without independent mobility, favour suspicion of abuse. Without a positive family history, differentiation is frequently difficult and sometimes impossible, and none of these distinguishing signs may be present, and biochemical analysis of type I collagen synthesis in cultured fibroblasts may be indicated [22]. Spranger documents a 1-month-old infant with distal submetaphyseal radius and ulna fractures, in whom abuse was considered but skull abnormalities were typical of OI — emphasising that the two conditions may coexist [22]. This possibility of coexistence makes multidisciplinary assessment involving clinical genetics, paediatric radiology and social work mandatory in cases of diagnostic doubt, according to the paediatric team's protocol, coordinating with the child protection team, social services and, in some circumstances, the police and forensic physician [41].

Hypophosphatasia is an inherited disorder, autosomal recessive in most cases (although certain variants have a dominant negative effect and are transmitted dominantly, according to Lia-Baldini et al., 2001), caused by mutations in ALPL (TNSALP) and characterised by deficient bone mineralisation associated with markedly reduced serum alkaline phosphatase — clearly distinguishing it from OI, in which this parameter is typically normal or slightly elevated after a recent fracture. Variability of clinical expression in hypophosphatasia is extreme, from stillbirths with a poorly mineralised skeleton to pathological fractures only in later adulthood and mild dental forms, and parents of severely affected children may have absent or minimal symptoms; molecular analysis and parental alkaline phosphatase measurement may be useful [9, 41]. It is also distinguished by absence of Wormian bones and metaphyseal irregularities; the infantile form is distinguished by deficient metaphyseal ossification and absence of fractures, and sometimes presents with bowed femora, with biochemical assessment being definitive [22–24]. Severe cases require early referral for recombinant enzyme replacement therapy with asfotase alfa.

Idiopathic juvenile osteoporosis is characterised by reduced bone density beginning around school age (mid- to late childhood), with axial rather than appendicular skeletal involvement predominating, a tendency towards spontaneous resolution around puberty, rarely familial occurrence, and absence of Wormian bones or ocular and dental abnormalities — findings which, when present, point towards OI [22]. It may occasionally be caused by heterozygous LRP5 and WNT1 mutations [9]. Premature osteoporosis in young adults should prompt investigation of endocrine causes, and secondary osteoporosis (nutritional, endocrine or metabolic) is differentiated by history and associated signs [9, 22].

Osteoporosis–pseudoglioma syndrome, a rare autosomal recessive disorder caused by biallelic LRP5 mutations (11q13.2; the Oxford Desk Reference gives 11q11–12), combines severe juvenile-onset osteoporosis with congenital or early-onset blindness due to hyperplasia of the persistent primary vitreous, corneal opacity and secondary glaucoma; muscle hypotonia, ligamentous laxity, mild intellectual disability and seizures may be present [9, 22]. Canonical Wnt pathway abnormalities through LRP5 also cause dominant osteoporosis [9].

Bruck syndrome combines classic OI features — the Oxford Desk Reference mentions blue sclerae and Wormian bones — with congenital large-joint contractures, bilateral talipes equinovarus (congenital clubfoot) and sometimes skin webs (pterygia) at the elbows and knees [9]. Its molecular basis is deficiency of bone telopeptide lysyl hydroxylase, responsible for mature bone collagen cross-linking (Bank et al., 1999); it is caused by mutations in FKBP10 (17q21.2; Bruck syndrome type 1) or PLOD2 (Bruck syndrome type 2, chromosome 3q23-q24). Spranger adds a type 3, allelic with type XI OI and also caused by FKBP10, in which contractures may be subtle and restricted to peripheral joints, making differentiation from OI difficult; unlike the Oxford Desk Reference, Spranger states that Bruck syndrome is not associated with blue sclerae, hearing loss or dentinogenesis imperfecta, although it has multiple Wormian bones, platybasia and basilar invagination, and that secondary OI contractures (due to deformity and prolonged immobility) differ clinically from congenital contractures [32]. Kuskokwim syndrome, also due to FKBP10, is a recessive congenital contracture disorder described in Alaskan Yupik populations [32].

The French PNDS adds to the list of differential diagnoses, to be selected according to the clinical and paraclinical picture, cleidocranial dysplasia, progeroid-type Ehlers–Danlos syndrome (B3GALT6, B3GAT3, autosomal recessive), Cole–Carpenter syndrome (P4HB, dominant; SEC24D, recessive), Stüve–Wiedemann syndrome, mucolipidoses II and III and spondylo-ocular dysplasia (XYLT2, recessive), and, among overlapping phenotypes, X-linked osteoporosis (PLS3) and primary ± familial osteoporosis (LRP5 or WNT1, dominant), together with Bruck syndrome (FKBP10 or PLOD2), osteoporosis–pseudoglioma syndrome (biallelic LRP5), idiopathic juvenile osteoporosis (sporadic), hypophosphatasia and geroderma osteodysplastica (GORAB) [8]. On abuse, the PNDS points out that the investigation may conclude that constitutional bone fragility coexists with inappropriate handling or abusive acts, and that both failing to recognise a child in danger and wrongly accusing a family have serious and lasting consequences [8].

Hajdu–Cheney syndrome is distinguished by the association of characteristic facial appearance, acro-osteolysis and distal phalangeal shortening, and a disproportionately long serpentine fibula — findings absent in typical OI [22].

Geroderma osteodysplasticum presents with wrinkled skin, progeroid features and a peculiar radiographic finding termed a knee “metaphyseal peg” [22].

Arthrogryposis multiplex congenita, a form of intrauterine hypomobility or confinement, results in thin poorly ossified tubular bones that may fracture during labour itself (in 8% to 10% of cases) or iatrogenically during attempts to mobilise contracted limbs; unlike OI, fractures occur without bowing, Wormian bones are absent, sclerae are white and the underlying bone structure is normal [9, 22].

Short stature, optic atrophy and Pelger–Huët anomaly syndrome, an autosomal recessive disorder due to biallelic NBAS mutation (2p24.3), involves bone fractures, recurrent infections, neutropenia — a reduction in neutrophil white blood cells — and elevated alanine aminotransferase, a liver enzyme (Segarra et al., 2015) [9].

Skeletal dysplasias with Wormian bones should be distinguished from OI based on their other characteristics [9]. Additional differential diagnoses include thanatophoric dysplasia, caused by FGFR3 mutations, lethal, with extreme limb shortening; campomelic dysplasia, caused by SOX9 mutations; achondrogenesis, which in type IA differs through crescent-shaped ilia and shorter tubular bones [24]; chondrodysplasias with long-limb bowing; Ehlers–Danlos syndrome, a heterogeneous group of connective tissue diseases, particularly forms caused by type I collagen mutations which may directly overlap phenotypically with OI — arthrochalasia EDS (aEDS, COL1A1/COL1A2, with mild osteopenia; autosomal dominant, with severe generalised joint hypermobility and recurrent multiple subluxations, congenital hip dislocation, hyperextensible skin, tissue fragility, easy bruising, muscle hypotonia and kyphoscoliosis [49]; the clinical diagnosis rests on congenital hip dislocation associated with severe generalised hypermobility with recurrent dislocations and/or hyperextensible skin, and is confirmed by sequencing of COL1A1 and COL1A2 with analysis of exon 6 [50]), cardiac-valvular EDS (cvEDS, biallelic COL1A2 variants with absence of pro-α2(I) chains), rare classical and vascular forms due to arginine-to-cysteine COL1A1 variants (p.(Arg312Cys), among others), and, in severe EDS forms, extensive mucocutaneous lesions after minor trauma potentially mimicking abuse [41, 51]; brittle cornea syndrome (ZNF469, PRDM5), with blue sclerae, extreme corneal thinning with rupture, high myopia and deafness, whose corneal rupture may be wrongly attributed to NAI [41]; congenital insensitivity to pain (SCN9A, among others) and congenital insensitivity to pain with anhidrosis (NTRK1), with frequent painless fractures [41]; Menkes syndrome (ATP7A, X-linked), with sparse hair and Wormian bones [41]; polyostotic fibrous dysplasia, with fractures, bone deformity and a ground-glass radiographic appearance [7]; Caffey disease (infantile cortical hyperostosis), allelic with OI, in which the heterozygous COL1A1 p.Arg836Cys variant (helical numbering; NM_000088.4:c.3040C>T, p.(Arg1014Cys) in HGVS nomenclature) causes spontaneous episodes of subperiosteal new bone formation beginning in the first 5 months of life, associated in carriers with joint hyperlaxity, hyperextensible skin and inguinal hernias resembling a mild EDS form [47]. Bone Dysplasias describes onset in the first 6 months, with irritability, fever, anorexia and painful, firm soft-tissue swellings (face, mandible, scapular region and limbs), sometimes with pseudoparalysis of a limb, and laboratory inflammatory signs in the acute phase in about 50% of cases; the cortical hyperostosis spares the metaphyses and epiphyses. In most cases, clinical and radiographic recovery is complete within weeks to months; in about 25%, the course is protracted, with remissions and exacerbations that may last into adolescence, bowing and elongation of the long bones and motor delay; severe prenatal forms, with short and angulated bones, polyhydramnios or hydrops, may result in stillbirth or death at birth from respiratory failure. Adults with a history of Caffey disease may have joint laxity, hyperextensible skin, hernias and an increased risk of fractures and/or bone deformities, but, unlike OI, the presentation does not result from recurrent fractures due to osteopenia; corticosteroids halt the process in the acute phase and non-steroidal anti-inflammatory drugs provide symptomatic relief [52]; and rickets and osteomalacia, due to vitamin D or phosphorus deficiency, always to be considered in the differential diagnosis of children with fractures and osteopenia — in rickets, metaphyseal radiographic findings are characteristic, without scleral abnormalities or hearing loss [7]. Hyperplastic callus may also occur in neuropathic diseases and child abuse due to inadequate fracture immobilisation [31].

---

7. Treatments and Risk Management Strategies

As a structural genetic condition, osteogenesis imperfecta has no established cure to date. Therapeutic goals are to reduce fracture rate, increase bone mineral density, prevent and correct deformities, minimise pain, preserve functional mobility and improve quality of life through a coordinated multidisciplinary approach individualised according to severity, degree of impairment and age [7], involving a clinical geneticist and genetic counselling, paediatric or adult bone disease endocrinologist, orthopaedic surgeon, rehabilitation medicine, physiotherapist, occupational therapist, dentist, otorhinolaryngologist/audiologist, ophthalmologist, neurosurgeon, respiratory physician, nephrologist, gastroenterologist, dietitian, social support, chronic pain specialist and mental healthcare [3, 7].

7.1 Non-pharmacological management

Safe patient handling should be taught in detail to caregivers: in infants, broad body support with the hands, slow gentle movements, avoidance of limb twisting, lifting the child while supporting the trunk, neck and pelvic girdle, avoiding concentrated pressure on any single point of a long bone, avoiding tossing the child in the air and taking special care during nappy changes not to pull the legs are recommended. Positioning should alternate between supine, prone and lateral positions to minimise skull and limb deformities. Adoption of an upright posture should begin only after adequate head and trunk control. Physiotherapy and occupational therapy have a central role, with goals of assessing and minimising motor development and mobility delays, strengthening trunk and limb muscles, preventing contractures and head and spinal deformities, increasing aerobic fitness — aquatic physical activity is particularly well tolerated because buoyancy reduces skeletal loading — controlling pain, promoting functional independence and providing orthoses, crutches, walking frames or wheelchairs according to functional needs [7]. Physical activity prevents additional bone loss [25]. Contact and high-impact sports are contraindicated [7]. During general anaesthesia, appropriate positioning on the operating table and cushioning, such as “egg-crate” foam, may help prevent fractures [7]. An international consensus on physical rehabilitation in children and adolescents with OI systematises assessment and physiotherapy interventions by developmental stage and severity, and is the reference cited by clinical sources for this treatment component [53].

Therapeutic education and rehabilitation according to the French PNDS. The measures to be taught to parents include handling the fragile infant while keeping the spinal axis aligned and avoiding abrupt manoeuvres, dressing the child in loose clothing opening at the front, adapting the home with an occupational therapist in severe forms, avoiding trauma and activities such as trampolining, slides and violent team sports, and taking seriously any unusual pain or a chronic pain syndrome. Entry into nursery and schooling are not contraindicated in most cases and are prepared with an individualised care plan (PAI, projet d'accueil individualisé), completed by the teaching team, the school doctor, the parents and the reference physician; management of a fracture (analgesia and immobilisation, section 7.10) should be taught very early, and the family receives the care and emergency card and useful contacts. Physical and rehabilitation medicine is indicated in all forms and is the only treatment in mild forms; it aims to prevent bone loss due to immobility, strengthen the musculature and optimise functional independence, autonomy, socialisation and quality of life. The rehabilitation annex of the PNDS provides for early physiotherapy in severe forms (educating parents in care and play, protecting the environment with a padded cot and bath chair, stimulating motor acquisitions according to neuromotor age: rolling, sitting, crawling, creeping on all fours), then progression to standing and walking with assistive devices, global and play-based muscle strengthening without pulley-based mechanotherapy (Klein ball, balneotherapy), progressive proprioceptive work, regaining range of motion after surgery involving joint opening, respiratory work and exercise training, adaptation to periods of surgery (non-weight-bearing periods, exercises during cast immobilisation) and training in wheelchair transfers. Readaptation includes adapting the home with support from the MDPH (departmental centre for disabled people), the PAI (immobilisation equipment and analgesics on site, travel and positioning in the classroom, playground, school outings, adapted physical education), the use of computers in severe forms and adapted sports or artistic activities [8].

7.2 Bisphosphonates

Bisphosphonates are the first-line pharmacological class and the most widely used in OI treatment, although no bisphosphonate has formal regulatory approval specifically for this indication in most countries — all use remains off-label, since there is no licensed treatment for children with OI to date. They do not constitute a cure but are prescribed to improve bone density, with particular indications in patients with vertebral fractures, recurrent long bone fractures or moderate to severe disease. Smith's states that treatment is recommended in children born with multiple fractures, long bone deformity and radiographic demineralisation, and suggests treating children with three fractures in total or more than two fractures in one year, including vertebral fractures, with decreased bone mineral content (Z-score ≤ −2) [21]. Mechanistically, they are inorganic pyrophosphate analogues binding to bone matrix hydroxyapatite, inhibiting dissolution of mineral crystals; they prevent osteoclast attachment to the bone matrix and reduce recruitment and viability of these cells, thereby reducing the bone resorption rate and shifting the remodelling balance towards net bone formation. Cyclical bisphosphonate therapy reduces bone turnover, increases bone density and reduces fracture frequency [22, 25]. Paediatric reviews systematise these indications and regimens [54]. The French PNDS sets clinical and radiological, not densitometric, indications, discussed at a medical–surgical multidisciplinary meeting (orthopaedic surgeons, radiologists, bone disease specialists, physical and rehabilitation medicine, endocrinology): more than three fractures, vertebrae included, from minor trauma in the previous two years; vertebral compression fractures, with or without postural disturbance (scoliosis, kyphosis); threatening basilar impression or marked acetabular protrusion, usually associated with other criteria; and neonatal forms of severe appearance, particularly with prenatal and perinatal fractures. The aims are to reduce fractures and new vertebral compressions, reduce bone pain and increase mobility. No study or recommendation defines when to stop: the maximum gain in density and cortical thickness occurs in the first two years, some propose 2 to 4 years of treatment guided by fracture incidence, vertebral height, mobility, remodelling markers and DXA, and, in severe OI, treatment continues throughout growth with decreasing annual cumulative doses (dose reduction or longer intervals between infusions); a dynamic remodelling marker was under study to define the minimum effective dose [8]. Each bisphosphonate is described below with its formulation, dose, dosage regimen, administration route and treatment duration.

Disodium pamidronate (trade name Aredia®, among others) is the reference intravenous bisphosphonate and has the largest accumulated evidence base in children, still considered the historical reference standard since Glorieux and colleagues' pioneering study published in the New England Journal of Medicine in 1998: in 30 children aged 3 to 16 years with severe OI, intravenous pamidronate (mean dose 6.8 mg/kg/year, at 4- to 6-month intervals, for 1.3 to 5 years) produced a mean annual 41.9% increase in bone mineral density, Z-score improvement from −5.3 to −3.4, a 27% annual increase in metacarpal cortical width, a reduction in fracture incidence of 1.7 per year and relief of chronic pain and fatigue in all children, without altering fracture healing, growth or growth plates [57]. In infants under 2 years with severe OI, Plotkin and colleagues (2000) obtained, with 3-day cycles and mean cumulative doses of 12.4 mg/kg over 12 months, bone mineral density increases of 86% to 227% and a lower fracture rate (2.6 vs 6.3 fractures/year) than untreated controls [58]. It is marketed in 30 mg/10 mL or 90 mg/10 mL vials, diluted in 0.9% saline or 5% glucose (paediatric reviews specify 100 to 250 mL normal saline over 3 to 4 hours) [55]. The classic protocol consists of administering 1 milligram per kilogram body weight per dose for 3 consecutive days, with cycles repeated every 4 months (9 mg/kg/year). In infants under 2 years, the dose is 0.5 mg/kg/day for 3 consecutive days, with cycles every 2 months; in children aged 2 to 3 years, the Montreal protocol uses 0.75 mg/kg/day for 3 days every 3 months; and above 3 years, 1 mg/kg/day for 3 days every 4 months, with a maximum dose of 60 mg per dose and 11.5 mg/kg/year [54, 55]. Modified protocols of a single infusion per cycle have been tested, with comparable results and shorter hospital stays, with dose individualised according to centre protocol. All this paediatric use is off-label, and cited doses come from published protocols [54, 55, 57, 58]. In adults, use is also off-label and the dose is individualised according to specialist centre protocol, adapted to clinical response. Infusion should last 2 to 4 hours per day over the three days of the cycle. Treatment is continuous and generally maintained until skeletal maturity in children. In adults, duration is periodically reassessed, with a drug holiday considered after three to five years of use depending on bone density and fracture rate.

French PNDS protocol. The PNDS considers intravenous bisphosphonates apparently more effective in children and uses the same pamidronate regimen (9 mg/kg/year: 0.5 mg/kg/day × 3 days every 2 months under 2 years; 0.75 mg/kg/day × 3 days every 3 months from 2 to 3 years; 1 mg/kg/day × 3 days every 4 months over 3 years; paediatric reviews set maxima of 60 mg per dose and 11.5 mg/kg/year [54, 55]), with half the dose on the first day of the first cycle, slow infusion over 3 to 4 hours in normal saline and flushing of the vein before and after the infusion; zoledronic acid is given at a dose of 0.025 to 0.050 mg/kg twice a year (maximum 4 mg per infusion in paediatric reviews [54, 55]). During the cycle, calcium 1 g/m²/day is supplemented, with body surface area calculated, below 40 kg, as (4 × weight in kg + 7)/(weight in kg + 90); between cycles, calcium covers age requirements taking the diet into account (600–800 mg/day before puberty, 1,000 mg/day in adolescence), and vitamin D is given as 100,000 IU every 3 months or 1,200 IU/day (fixed doses, not scaled by weight). Temperature is monitored during the first cycles, with paracetamol at the usual doses. Precautions are an oral and dental assessment with hygiene advice (risk of osteonecrosis of the jaw, according to the 2007 Afssaps recommendation), a pregnancy test in adolescent girls before each cycle (risk of foetal osteosclerosis), perioperative suspension (section 7.7) and respiratory monitoring in infants with precarious respiratory function. At each cycle, pain, fractures, deformities, mobility, height, weight and the oral and dental examination are assessed, and calcium, phosphorus, 25-hydroxyvitamin D, PTH, osteocalcin (formation), CTX (resorption) and urinary calcium and creatinine on a spot sample are measured; once a year, lumbar DXA is performed, to measure efficacy and avoid iatrogenic osteopetrosis, together with a lateral radiograph of the thoracolumbar spine, with low-dose techniques (EOS type), to look for vertebral fractures and assess vertebral height [8]. In adults, the PNDS concludes that there is no consensus on bisphosphonates, that the data are discordant and that no benefit has been formally demonstrated, whatever the route [8].

Zoledronic acid (Zometa® for oncology use; Aclasta®/Reclast® for metabolic bone diseases), also termed zoledronate, is the most potent intravenous bisphosphonate and the most convenient to administer, with one or two infusions per year, an infusion of only 15 to 45 minutes and a longer dosing interval. It is marketed in vials containing four milligrams in one hundred millilitres of ready-to-use solution or concentrate vials containing four milligrams in five millilitres for dilution. In children (off-label use), the most studied dose is 0.05 mg/kg every 6 months (maximum dose 4 mg per infusion), with a total annual dose of approximately 0.1 mg/kg, infused over 30 to 45 minutes; paediatric reviews describe a range of 0.0125 to 0.05 mg/kg every 6 to 12 months [54, 55]. In adults, the label dosage for osteoporosis is 5 mg intravenously once a year, infused over at least 15 minutes, with adequate hydration and renal function monitoring before each administration; in OI this use is off-label. It is contraindicated in patients with creatinine clearance below 35 mL/min (Reclast®/Aclasta® label). Zoledronic acid has been increasingly used in children with OI because of the shorter infusion time and longer interval between doses compared with pamidronate. Observational studies demonstrated similar effects to pamidronate, with increased bone mineral density, reduced fracture rate and improved pain. Direct comparison of zoledronic acid and pamidronate in two controlled trials found no differences in primary outcomes, although studies differed regarding the relative benefit of one over the other for lumbar bone mineral density at twelve months. A network meta-analysis published in 2025 (9 randomised trials, 595 children; Wang et al., Archives of Osteoporosis, 2025; doi:10.1007/s11657-025-01515-6) showed that, at 2 years, pamidronate produced the greatest improvement in lumbar spine areal bone mineral density, whereas zoledronic acid produced the greatest lumbar spine Z-score gains at both 1 and 2 years (low-certainty evidence), at the cost of a higher adverse event rate; in the limited fracture analysis, only olpadronate reduced total fracture number relative to placebo, without a demonstrated difference between pamidronate and zoledronic acid.

Risedronate (Actonel®, among others) is the oral bisphosphonate most studied in children with OI, with a slightly better gastrointestinal adverse effect profile than alendronate, available in 2.5, 5, 35 and 75 mg tablets. In children, the pivotal trial used daily doses of 2.5 or 5 mg, in children aged 4 to 15 years, with no maximum dose defined in the source [56]; in adults, the label dosage for osteoporosis is 5 mg/day or 35 mg/week (off-label use in OI). A multicentre randomised double-blind placebo-controlled trial published in the Lancet in 2013 by Bishop and colleagues, with 147 children aged 4 to 15 years, demonstrated increased lumbar bone mineral density at 1 year (16.3% vs 7.6%; difference 8.7%) and reduced clinical fracture rate (31% vs 49%; p = 0.0446), with an adverse effect profile similar to placebo [56]. In an observational study of 32 adults with type 1 osteogenesis imperfecta (27 completed), risedronate at 35 milligrams weekly for 24 months modestly increased lumbar bone mineral density (+3.9%), without a significant total hip change, and reduced bone turnover (P1NP −37%), but fracture incidence remained high — without demonstration of reduced fracture risk or rate (Bradbury et al., Osteoporosis International, 2012; doi:10.1007/s00198-011-1658-2).

Sodium alendronate (Fosamax®, among others), available in 10 mg (daily) and 70 mg (weekly) tablets, is an alternative for patients in whom intravenous infusion is impractical or for mild disease forms; it was assessed in a randomised placebo-controlled trial in paediatric OI (Ward et al., 2011) [22]. In children (off-label use), Ward's trial used 5 mg/day below 40 kg and 10 mg/day from 40 kg; paediatric reviews also describe 1 to 2 mg/kg/week (maximum 70 mg/week) or, by weight, 35 mg/week (<40 kg) and 70 mg/week (≥40 kg) [55]; in adults, the label dosage for osteoporosis is 10 mg/day or 70 mg/week. The medicine should be taken fasting immediately on waking, with a full glass of water, at least 30 minutes before any food, drink other than water or other medication, and the patient should remain sitting or standing during this period to avoid oesophagitis. Principal adverse effects are gastrointestinal: oesophagitis, gastric or oesophageal ulcers, reflux and gastritis.

Neridronate, widely used in Italy, where it is the only bisphosphonate with regulatory approval specifically for OI, is administered intravenously at a paediatric dose of 1 to 2 mg/kg every 3 to 4 months [55] and, in adults, 100 mg every 3 months. Ibandronate (Bonviva®/Boniva®) may be administered orally (150 mg monthly) or intravenously (3 mg every 3 months), according to label dosage for postmenopausal osteoporosis (off-label use in OI). Oral olpadronate also has favourable evidence from European clinical trials, with data showing reduced fracture rate and increased bone mineral density. Neridronate and olpadronate are not commercially available in Brazil at the time of this review.

All oral bisphosphonates should be administered fasting with a full glass of water, at least 30 minutes before the day's first food, drink or medicine (60 minutes for ibandronate), with the patient upright during this period, according to the respective labels, avoiding calcium, antacids and medicines containing iron or magnesium in the two hours before or after, to prevent oesophagitis and ensure adequate absorption.

Acute-phase reaction and adverse effects. The acute-phase reaction — a transient inflammatory response with fever up to 38.5°C, myalgia, arthralgia, diffuse musculoskeletal pain, bone pain, headache, nausea, rash and flu-like malaise in the 12 to 24 hours after the first intravenous infusion, resolving within one to three days — is mediated by γδ T lymphocyte activation and tends not to recur, or to attenuate substantially, with subsequent infusions. Premedication with a non-steroidal anti-inflammatory drug (NSAID) is recommended, particularly in the first cycle: in paediatric clinical practice, ibuprofen is usually used at 10 mg/kg orally, administered 30 to 60 minutes before infusion begins, repeated every 6 to 8 hours over the following 24 to 48 hours as needed (paediatric ibuprofen label analgesic dose: 10 mg/kg every 6 to 8 hours, maximum 40 mg/kg/day, from 6 months); in adults, ibuprofen may be used at 400 mg orally at the same frequency, or paracetamol at 500 to 1,000 mg (maximum 4 g/day) (or metamizole) as an alternative for patients with contraindications to non-steroidal anti-inflammatory drugs, such as active peptic disease or renal impairment. In infusions after the first cycle, the acute-phase reaction tends to become progressively less intense and frequently does not require systematic premedication, although many services maintain the preventive protocol for at least the first two or three cycles. Adequate hydration, with generous oral fluid intake or concurrent intravenous hydration according to patient tolerance, is also recommended before and during infusion to reduce the risk of transient hypocalcaemia and, with zoledronic acid, nephrotoxicity. Before each intravenous bisphosphonate cycle, serum calcium, phosphorus, magnesium and renal function should be checked, correcting any vitamin D deficiency before infusion to minimise symptomatic hypocalcaemia risk. Shared class adverse effects also include oesophagitis and upper gastrointestinal irritation with oral formulations if administration rules are not followed. Osteonecrosis of the jaw and atypical femoral fractures associated with prolonged adult use, particularly in association with invasive dental procedures, are rare and not consistently documented in children with OI (there are no reports of jaw osteonecrosis in children receiving pamidronate), but justify dental assessment before treatment and maintenance of adequate oral hygiene throughout its duration. The possibility of delayed bone healing after fractures or osteotomies with prolonged bisphosphonate use is under investigation in ongoing studies. Bisphosphonates cross the placenta and should be stopped before planned pregnancy. Radiographically, cyclical treatment produces transverse sclerotic metaphyseal bands [32].

Expected results. A Cochrane systematic review, updated in 2016 and analysing 14 clinical trials with 819 participants, concluded that oral and intravenous bisphosphonates consistently increase bone mineral density in children and adults with OI; however, fracture rate reduction was not consistently demonstrated in all studies, although no study reported an increased fracture rate with treatment; there was no conclusive improvement in pain, growth or functional mobility; comparison of oral and intravenous bisphosphonates showed no primary outcome differences; and optimal dose, treatment duration and long-term safety remain under investigation [59]. The Oxford Desk Reference, in turn, considers bisphosphonates safe and effective therapy integrated within multidisciplinary management, and Smith's states that they reduce bone pain and help maintain good mobility [9, 21].

What the labels say. US labels for three bisphosphonates describe paediatric OI trials only in the paediatric use section, without granting an indication. The zoledronic acid label (Reclast®) reports a one-year active-controlled trial in 152 patients aged 1 to 17 years with severe OI, with increased bone mineral density but no individual correlation with fracture risk or pain; fever (61%), arthralgia (26%), hypocalcaemia (22%) and headache (22%) were more frequent than in adults, particularly in the three days after the first infusion, and the label allows use in children only if potential benefit outweighs risk, given prolonged bone retention [12]. The risedronate label (Actonel®) describes the one-year trial in 143 children aged 4 to 16 years, 85% with type I OI, and concludes that treatment increased lumbar density but did not reduce fracture risk, with more vomiting (15% versus 6%) [13]; publication of the same trial reports fewer children with clinical fractures (31% versus 49%; p = 0.0446), a secondary outcome, and this discrepancy shows the difference between regulatory analysis and the authors' interpretation [56]. The alendronate label describes a two-year trial in 139 children with severe OI: lumbar Z-score increased by 1.3 versus 0.1 with placebo, fracture risk did not decrease, and 16% of treated children with a confirmed fracture in the first year had delayed healing or pseudarthrosis at 24 months, versus 9% with placebo [14]. In the European Union, the Aclasta® SmPC states that it should not be used below 18 years [60].

Response by genetic type. Response is not uniform. In type 6 OI (SERPINF1), ten children treated for three years with cyclical pamidronate improved lumbar density, vertebral shape, grip strength and gross motor function, and fracture incidence fell from 3.1 to 1.4 per year, but incidence remained higher and motor function lower than in matched patients with types I, III and IV OI, and the mineralisation defect did not change [61]. In type 13 OI (BMP1) with high bone mass, the indication for antiresorptives is uncertain (section 3.11) [40].

7.3 Denosumab

Denosumab is an IgG2 monoclonal antibody directed against RANKL (receptor activator of nuclear factor kappa-B ligand), a signalling molecule essential for osteoclast differentiation, activity and survival; by blocking RANKL, it prevents formation of new osteoclasts and reduces bone resorption. It is marketed as Prolia® (adult osteoporosis) and Xgeva® (oncological indications), without formal approval for paediatric OI. Prospective studies in children with COL1A1 or COL1A2 mutations and types 1, 3 and 4 used 1 mg/kg subcutaneously every 12 weeks (3 months; trial dose, with no maximum dose defined in the source), demonstrating significant improvement in lumbar spine bone mineral density Z-score without a significant change in mobility [62]. A randomised comparative trial of 84 children and adolescents, published in 2024 in the Journal of Clinical Endocrinology & Metabolism by Liu and colleagues, compared subcutaneous denosumab every 6 months with a single zoledronic acid infusion: after 12 months, bone mineral density increased by 29.3%, 27.8% and 30.2% at the lumbar spine, femoral neck and total hip with denosumab, and by 32.2%, 47.1% and 41.1% with zoledronic acid; rebound hypercalcaemia was a common, serious adverse effect of denosumab, reaching hypercalcaemic crisis in 14.3% of cases, and was attenuated by transition to zoledronic acid [63]. Denosumab has been presented as an alternative for patients unresponsive to bisphosphonates.

One pharmacological advantage of denosumab over bisphosphonates is its short degradation time (3 to 4 months), avoiding the prolonged skeletal accumulation characteristic of bisphosphonates, whose residual effect persists for years after withdrawal. However, this same pharmacokinetic feature is its principal safety limitation: the rebound effect after denosumab discontinuation can cause a marked, abrupt increase in bone resorption, with loss of bone density, increased fractures and risk of severe, sometimes potentially fatal hypercalcaemia if the drug is stopped without replacement by another antiresorptive agent. This risk was specifically documented in an Amgen-sponsored paediatric study (protocol 20130173, NCT02352753, 153 participants, and its extension 20170534, NCT03638128, both terminated early in March 2022 because of hypercalcaemia): from 30 September 2021, following review by the independent data monitoring committee, further denosumab administration in the study was permanently suspended because of life-threatening hypercalcaemia events reported in paediatric participants — reinforcing the recommendation that denosumab be used in children with OI exclusively under a research protocol or rigorous specialist follow-up, with strict serum calcium monitoring and careful planning of a possible transition to replacement antiresorptive therapy (typically a bisphosphonate) before denosumab withdrawal, never stopping it abruptly and without supervision. Consequently, the UK MHRA (Drug Safety Update, May 2022) and other agencies began recommending that denosumab 60 mg (Prolia®) should not be used below 18 years. The European Prolia® SmPC now incorporates this restriction — it should not be used below 18 years because of severe hypercalcaemia risk and possible inhibition of bone growth and tooth eruption — and summarises the single-arm phase 3 study in 153 children aged 2 to 17 years with OI: with 1 mg/kg (maximum 60 mg) every 6 months and, in 60 patients, every 3 months, lumbar Z-score increased by a mean of 1.01 at 12 months of quarterly administration, but hypercalcaemia occurred in 19% (six-monthly) and 36.7% (quarterly), severe hypercalcaemia in 13.3% with quarterly administration and 18.5% in the extension (n = 75), and the studies were terminated early [64]. The US label adds that reduction in fracture risk, the study objective, was not demonstrated [65]. Paediatric OI use is therefore off-label in both jurisdictions and contrary to the European SmPC.

7.4 Recombinant human growth hormone

It may be used in children with moderate to severe OI to increase growth velocity and, in responders, improve bone mineral density. This is off-label use; studied doses range from 0.1 to 0.2 IU/kg daily, administered six days weekly for at least one year [66], or 0.2 mg/kg weekly (0.6 IU/kg weekly) in six subcutaneous injections [67]; these are trial doses, with no maximum dose defined in the sources. Responders show a 5% to 7% increase in lumbar bone mineral density Z-score, lower fracture frequency and greater growth velocity; however, long-term follow-up data remain incomplete and the impact on final height is not definitively established. The French PNDS considers the effect of growth hormone in OI modest and inconsistent and does not recommend it routinely [8].

7.5 Teriparatide and the TOPaZ trial

Teriparatide (Forteo®, among others), a synthetic amino-terminal fragment of human parathyroid hormone (amino acids 1-34), has an anabolic action on bone by directly stimulating osteoblast differentiation and activity. Studies in adults with OI type I demonstrated increased bone mineral density and improved remodelling: in a double-blind placebo-controlled trial of 79 adults treated for 18 months with 20 μg/day, areal density increased by 6.1% at the lumbar spine (versus 2.8% with placebo) and 2.6% at the total hip (versus −2.4%), with improved volumetric density and estimated vertebral strength; response was robust in OI type I but no benefit was observed in types III/IV, and there was no difference in self-reported fractures [68]. Its use is restricted to adults because of concern about osteosarcoma risk observed in animal models with open growth plates, and limited to 24 months under the European SmPC (the FDA label permits use beyond 2 years only if fracture risk remains high), followed by antiresorptive therapy (bisphosphonate or denosumab) to consolidate and maintain bone gains. The large randomised controlled TOPaZ (Treatment of Osteogenesis Imperfecta with Parathyroid Hormone and Zoledronic Acid) trial, multicentre, open-label and parallel-group, conducted at 27 referral centres in the United Kingdom and Europe, recruited 350 adults with clinically diagnosed OI between 2017 and 2022 (76.2% type I, 15.8% type IV, 5.4% type III), with baseline characteristics published in 2025 [69]. Results were published in JAMA in July 2026: the active group received 20 μg teriparatide subcutaneously daily for 2 years, followed by a 5 mg zoledronic acid infusion, while the usual-care group could use bisphosphonates and other antiresorptives, but not anabolic agents. Among 349 evaluable participants (mean age 43.7 years; 53.9% women), imaging-confirmed incident fractures occurred in 36.9% of the teriparatide–zoledronate group and 36.4% of the usual-care group (risk ratio 0.97; 95% CI 0.68–1.38), despite significantly greater lumbar and hip bone mineral density increases and several quality-of-life indicators favouring the active group, with similar adverse events; the authors conclude that reduced bone quality, rather than low density, determines fractures in OI [70].

7.6 Biological therapies in advanced investigation

The OI therapeutic landscape underwent an important shift at the end of 2025. Setrusumab (UX143, also designated BPS-804), a human monoclonal antibody inhibiting sclerostin — a protein that physiologically suppresses the bone-forming Wnt pathway —, had shown promising phase 2b results (ASTEROID study, 112 adults with OI types 1, 3 or 4): although it did not meet its primary endpoint of increased radial trabecular bone mineral density, it produced areal bone mineral density increases of 8.81% at the spine for type 1 and 10.38% for types 3/4 over 12 months, and 2.48% to 3.37% at the total hip and femoral neck [71]. In phase 2 of ORBIT, in 24 participants aged 5 to 25 years, the highest dose produced a 21.6% increase in spinal areal density at 6 months. However, the pivotal phase 3 ORBIT (NCT05125809, 159 patients aged 5 to 25 years in the phase 3 part, setrusumab versus placebo) and COSMIC studies (NCT05768854, 69 children aged 2 to under 7 years, setrusumab versus intravenous bisphosphonate), whose topline results were announced by Ultragenyx Pharmaceutical on 29 December 2025, did not achieve statistical significance for the primary endpoint of reduced annualised clinical fracture rate compared with their respective comparators, although both studies met the key secondary endpoint of statistically significant bone mineral density improvement, without a change in the observed safety profile [15]. In 2026, Mereo BioPharma and Ultragenyx continued additional analyses of fracture data and patient-reported outcomes, particularly in the 2- to 18-year subgroup, to determine whether a regulatory pathway for setrusumab is possible, and an open-label extension remains available to trial participants [72]. Sponsor figures presented in January 2026, still without a peer-reviewed publication, show the scale of the negative result: in ORBIT (159 patients aged 5 to 25 years, 2:1 randomisation, double-blind, 18- to 24-month follow-up), estimated annualised confirmed clinical fracture rates were 0.71 with setrusumab and 0.55 with placebo (rate ratio 1.28; 95% CI 0.80–2.06; p = 0.305), and, counting all fractures, 1.16 versus 1.12 (ratio 1.03; 95% CI 0.71–1.52); 31 patients (19.5%) exited under rescue criteria at 12 months, mainly because of fractures; among OI type III or IV patients, 13 of 31 (42%) exited in the placebo group and 15 of 64 (23%) in the setrusumab group, an imbalance which, according to the sponsor, may have reduced the observed fracture rate in controls [73]. In COSMIC (69 children aged 2 to 7 years, open-label, comparison with intravenous bisphosphonate), the rate ratio was 0.79 (95% CI 0.48–1.28; p = 0.338); a post hoc analysis of all vertebral fractures showed 19 versus 46 fractures (ratio 0.44; 95% CI 0.18–1.11; p = 0.081), an exploratory, non-significant result [73]. In August 2026, Mereo reported that the FDA was open to alternative fracture-analysis approaches, that further discussions would be required to define supplementary clinical data potentially supporting an authorisation application, that the UK MHRA encouraged further dialogue, and that it expected to provide an update by the end of 2026; no authorisation application has been submitted [16]. Setrusumab retains orphan drug designation in the United States and European Union, which does not imply demonstrated efficacy [16].

Romosozumab (Evenity®), another sclerostin inhibitor already approved for adult postmenopausal osteoporosis, is being assessed in an Amgen-sponsored phase 3 trial in children and adolescents with OI, comparing it with bisphosphonates (NCT05972551, 111 participants, recruitment closed, estimated primary completion May 2027), following a phase 1b study completed in 2023 (NCT04545554, 25 participants); no efficacy results have yet been published. In March 2026, a paediatric safety follow-up study began for patients previously treated with romosozumab (NCT07366086, phase 3, 71 planned participants) [74]. In adults, published experience is limited to small series: in a retrospective Hamburg analysis, five postmenopausal women with classical OI treated for 12 months had significant increases in lumbar areal density and osteocalcin, but no significant HR-pQCT microarchitectural changes and smaller hip gains than women with severe osteoporosis; romosozumab is authorised in the USA and EU only for severe postmenopausal osteoporosis, making this use off-label [75, 76].

A more recently developed agent with an innovative mechanism is AGA2115 (Angitia Biopharmaceuticals), a humanised bispecific antibody simultaneously neutralising sclerostin and Dickkopf-1 (DKK1), two distinct endogenous inhibitors of the bone-forming Wnt pathway. The first-in-human study (NCT06086613, 72 healthy volunteers, completed in October 2024), with topline results announced on 12 June 2025 and presented at the ASBMR meeting in September 2025, demonstrated a favourable safety profile, with rapidly increased bone formation, simultaneously reduced bone resorption and bone mineral density gains in both single- and multiple-dose cohorts of healthy volunteers. The molecule received orphan drug and rare paediatric disease designations from the FDA (Food and Drug Administration), alongside orphan drug designation from the EMA (European Medicines Agency), and the phase 2 IDUN study (NCT07062588), assessing safety and efficacy in adults with OI types I, III or IV, dosed its first participant in January 2026, with estimated primary completion in November 2027 [77].

Anti-transforming growth factor beta (TGF-β) antibodies remain in development, based on the observation that excessive TGF-β signalling in bone is a central pathogenic mechanism in OI models. Fresolimumab was first used in adults with OI in a phase 1 study (NCT03064074) of 8 adults: it was well tolerated and associated with increased lumbar areal density in OI type IV participants, whereas density remained unchanged or decreased in types III and VIII; in parallel, histology and RNA sequencing of bone obtained from children with OI outside the trial showed TGF-β pathway activation, identifying it as a therapeutic target [78]. Sanofi's SAR439459 antibody was assessed in a phase 1 single ascending-dose study in adults with OI (NCT05231668), terminated early in November 2024 with 16 participants, without a reason specified in the registry [74]. Losartan, an antihypertensive indirectly modulating the TGF-β pathway, is being assessed in adolescents and adults with OI within the European REMEDi4ALL consortium.

In cellular therapies, mesenchymal stem cell therapy has been investigated in animal models and preliminary clinical studies. BoostB4 (NCT03706482), a phase 1/2 trial, assesses postnatal-only or combined prenatal and postnatal administration of expanded allogeneic foetal mesenchymal stem cells (BOOST cells), derived from first-trimester foetal liver, to treat severe OI (type III or severe type IV) in infants and foetuses, comparing outcomes with untreated historical and prospective controls; the protocol remains underway across centres in Sweden, the United Kingdom and the Netherlands, with a follow-up study called BOOST2B, assessing repeated intravenous and intraosseous doses in children, also underway in India. Definitive efficacy results have not yet been published [79]. In the current registry, BOOSTB4 is sponsored by Karolinska Institutet, with 18 enrolled participants, recruitment closed and estimated primary completion in 2030, while the BOOST2B protocol, published in 2025, plans four intravenous and intraosseous doses in 15 children aged 1 to 5 years in India [74, 80]. Foetal mesenchymal stem cells, losartan and the bispecific antibody AGA2115 have European orphan designation for OI (EU/3/21/2534, EU/3/22/2622 and EU/3/25/3069), as does cisatracurium since July 2026 (EU/3/26/3272); orphan designation is based on plausibility and rarity and does not demonstrate clinical benefit [81].

4-Phenylbutyric acid, already approved for a urea cycle disorder, experimentally acts as a chemical chaperone, assisting correct protein folding in the endoplasmic reticulum, with promising results in OI cell and animal models, still in advanced preclinical development. Gene therapies seeking to correct or replace defective genes — including CRISPR-Cas9 genome editing — and mesenchymal and haematopoietic stem cell therapies represent the research frontier for a durable solution, still experimental or in early clinical trials [82].

7.7 Orthopaedic surgical treatment

Orthopaedic surgery includes fracture management with shorter immobilisation than in the general population, lightweight supports and physiotherapy immediately after cast removal; prolonged post-fracture immobilisation should be avoided because it accelerates disuse bone loss, worsening underlying osteoporosis. OI fractures heal normally in approximately eighty-five per cent of patients. Intramedullary rodding — metal pins inserted into the bone's medullary canal to correct deformities and prevent fractures — is indicated in moderate to severe OI, or mild forms with disability from recurrent fractures, pseudarthrosis or progressive deformity, and improves weight-bearing capacity and allows earlier walking [22, 23]. Extensible telescopic rods (Fassier–Duval system, among others) are preferred in growing children because they lengthen with bone growth, reducing repeated surgical revisions. In children already receiving bisphosphonates, percutaneous multiple osteotomy with intramedullary fixation is safe and effective, allowing experienced teams to correct up to four lower-limb long-bone segments in a single operation with shorter postoperative immobilisation.

The French PNDS specifies the principles of palliative osteosynthesis, which protects a fragile, deformed diaphysis that is the site of repeated fractures. Segmental osteosyntheses should be avoided except in rare indications, such as the proximal femur combined with intramedullary fixation; epiphysis-to-epiphysis intramedullary rodding is the reference method, with hardware (wires or rods) that is not too bulky and a telescopic effect through epiphyseal anchorage, and usually remains in place for life; the technique, the approach (open, minimally invasive or percutaneous under image intensifier) and the hardware depend on the site, the type of fracture and the team; cast immobilisation after osteosynthesis is not systematic and mobilisation is resumed progressively. Coxa vara, which is frequent, is prevented by fixing the femoral neck in valgus; a fracture over a rod usually heals with simple immobilisation and only rarely requires the hardware to be replaced; infection may require its removal; and, in forms with hypertrophic callus (type 5), an anti-inflammatory drug is recommended post-operatively for 3 to 4 weeks, without a specified drug or dose (using ibuprofen at the PNDS's own paediatric dose, 10 mg/kg every 8 hours, with the maximum of 30 mg/kg/day of the French SmPC [83], is an extrapolation) [8]. Bisphosphonates should be suspended at least 48 hours before surgery and not prescribed in the following three months, to avoid delayed healing; the dosage annex of the same PNDS is more cautious for planned long bone surgery (realignment osteotomy and intramedullary rodding): no cycle in the preceding week and none until the healing callus is visible [8]. In fractures treated conservatively, lightweight materials (resins), shorter immobilisation than outside OI, transition to a posterior plaster splint on the lower limb after early cast removal and maintenance of muscle trophism (contractions in the cast, standing, physiotherapy after removal) limit immobilisation osteopenia [8].

In severe OI-associated scoliosis, technically challenging because of early onset and curve magnitude, short stature, bone fragility and associated comorbidities, modern segmental pedicle-screw fixation and multimodal surgical approaches including bone morphogenetic protein type two have enabled relevant corrections — a study of 30 patients with different OI types demonstrated mean improvement of the main curve from 76 to 36 degrees, with minimal loss of correction during follow-up. Historically, bone marrow transplantation from HLA-matched sibling donors was associated with improved growth, significantly increased bone mineral content and reduced fractures in severe type III patients (Horwitz and colleagues, 2001) [25], although this approach has not become established in current practice given the procedure's risk profile.

According to the French PNDS, spinal deformities cause disability, loss of height and restriction of lung volumes; bracing is reserved for moderate forms and posterior spinal fusion, increasingly used, stabilises the deformity. The surgical decision weighs three criteria — worsening of the curve, respiratory deterioration and end of growth — and preoperative halo traction allows progressive correction with improvement of respiratory capacity through rehabilitation. Complications of the craniocervical junction (basilar impression, odontoid instability, cervical dislocation), possible in severe or progressive forms, may be silent or cause headaches, vertigo, neuralgia, torticollis or reflex changes and, in severe cases, swallowing disturbances, paralysis and even sudden death. The PNDS recommends MRI of the craniocervical junction in the first year of life in all children with a severe form or with marked platybasia, repeated in the presence of symptoms; treatment combines immobilisation in a Minerva brace and surgery to reduce the basilar impression with bone grafting [8].

Surgical treatment of basilar invagination is reserved for symptomatic cases with neurological deficits, particularly from brainstem and upper cervical cord compression, and should ideally be performed in referral centres with specific skull-base neurosurgical experience because of the technical complexity of surgery in structurally poor bone. In type V, indomethacin helps prevent hyperplastic callus progression [31]; in Bruck syndrome, bisphosphonates may be beneficial, and orthopaedic treatment of congenital contractures is necessary [32].

Transition and adulthood. The French PNDS recommends supporting the move from paediatric to adult care with joint transition consultations of the paediatric and adult teams. Fractures decrease after the end of puberty, with relative stabilisation in young adults, and reappear at menopause; they are treated as outside OI, but with light immobilisation and early mobilisation of the non-immobilised segments and of the fractured limb. In severe adult forms, progressive kyphoscolioses affect functional, respiratory and vital prognosis and may require posterior spinal fusion [8].

7.8 Nutritional, dental, hearing and pain support

Adequate calcium, vitamin D and phosphate intake is part of treatment. Calcium and vitamin D intake should be ensured in all patients, preferably through diet and supplementation when dietary intake is insufficient, before and during any antiresorptive therapy. In children, reference daily calcium requirements are 200 mg (0–6 months), 260 mg (6–12 months), 700 mg (1–3 years), 1,000 mg (4–8 years) and 1,300 mg (9–18 years), and vitamin D 400 IU in the first year of life and 600 IU thereafter [55]; adult reference intakes are 1,000 to 1,200 mg/day calcium and 600 to 800 IU/day vitamin D [84]. Vitamin D dose should be adjusted to serum 25-hydroxyvitamin D levels, maintained above 20 ng/mL. Calorie control is indicated in non-ambulant patients with severe OI because of their lower energy expenditure. Nutritional assessment should occur annually or as needed, maintaining healthy weight for height because both underweight and excess weight adversely affect bone health and mobility.

Dental care should begin early, with six-monthly appointments for patients with or at risk of dentinogenesis imperfecta; for those without dentinogenesis imperfecta, annual reviews suffice. Prosthetic rehabilitation, tooth coverage or veneering, preventive treatment and orthodontic follow-up are often necessary [21]. OI oral health specialists recommend full-coverage restorations (stainless steel or zirconia crowns) for damaged teeth, avoiding intracoronal restorations that promote dental structure loss; simple extractions should not be performed immediately before or after intravenous bisphosphonate infusions, necessary dental surgery should be scheduled before treatment starts or after it ends whenever possible (treatment should resume only after healing), orthognathic surgery is discouraged in severe forms and clear aligners are a promising orthodontic option [3]. For conductive hearing loss, middle-ear surgery may restore or improve hearing in selected cases [21]; sensorineural loss is treated with hearing aids and, when indicated, cochlear implantation. Hearing defects should be recognised and treated, with repeated hearing testing [22, 25]. Periodic otological assessment should begin around 5 years, every 3 years until hearing loss is identified and thereafter as needed. Chronic pain management should be multidisciplinary, combining non-steroidal anti-inflammatory drugs — the most commonly used agents for acute and chronic pain —, the documented analgesic effect of bisphosphonates themselves, physiotherapy, cognitive behavioural therapy and, in adults, participation in peer support groups.

The French PNDS recommends an audiogram before 5 years of age in all children, repeated on average every 3 years or earlier in the presence of hearing loss, and that otological surgery, when indicated, be performed by a team experienced with OI, because of the greater risk of complications, particularly haemorrhagic ones [8]. In oral health, prevention includes learning oral and dental hygiene, age-appropriate dietary advice, oral and/or topical fluoride therapy (varnishes or gels in trays) according to the 2008 Afssaps national recommendation, and at least annual check-ups with the dentist, more frequent under bisphosphonates. Restorative treatment is adapted to the tissue loss and the number of missing teeth (glass ionomers, composites, inlays and onlays, crowns, bridges, conventional or implant-supported prostheses, implants); endodontic treatment must take account of early canal obliteration; extractions are difficult because of the tendency of the hypomineralised tooth to fracture; and orthodontics is not contraindicated. Care is provided in liaison with reference or competence centres for the oral and dental manifestations of rare diseases [8].

7.9 Regulatory status, pivotal trials and limitations of evidence

Therapeutic information in this section represents distinct evidence levels that should not be confused: peer-reviewed published trials (bisphosphonates, teriparatide, TOPaZ, denosumab compared with zoledronic acid); labels and SmPCs, reflecting regulatory analysis and sometimes differing from authors' interpretations; sponsor press releases and presentations (phase 3 setrusumab results, not yet journal-published); trial registries; and orphan designations, which do not require demonstrated benefit. In October 2026, no medicine has an FDA- or EMA-approved OI indication, as the setrusumab sponsor itself acknowledges [16]. Authorisation for another indication also does not guarantee reimbursement for OI use in Portugal or Brazil; the consulted sources identified no INFARMED, ANVISA or CONITEC decision specifically concerning OI, and absence of this information does not prove that none exists.

In France, the PNDS points out that prescription outside the marketing authorisation (AMM) is possible, in the absence of an appropriate medicinal alternative, if there is a temporary recommendation for use (RTU) from the ANSM or if the prescriber considers it indispensable according to the scientific data; the patient is informed of the off-label nature, the lack of an alternative, the risks, benefits and reimbursement conditions, the prescription bears the mention of off-label prescription and the decision is justified in the clinical record; the ANSM may draw on the PNDS to draw up an RTU, and exceptional temporary reimbursement requires a prior RTU and an opinion from the HAS. No bisphosphonate has a marketing authorisation for children with OI. OI is covered by the «off-list» long-term condition scheme for exemption from patient co-payment, with a care protocol established by the attending physician with the health insurance medical adviser [8]. The PNDS literature search covered 1992–2016, with monitoring until November 2016, so it does not include TOPaZ, the phase 3 trials of setrusumab or the paediatric experience with denosumab described above [8].

The trials do not allow ranking of drugs by indirect comparison. Endpoints differ: most measure bone mineral density, a surrogate marker weakly related to fractures in OI, as shown by TOPaZ and ORBIT, in which density improved without fracture reduction [15, 70]. Populations differ in age, OI type and previous treatment (COSMIC compares with active bisphosphonate; ORBIT with placebo; TOPaZ allowed bisphosphonates in controls), and fracture counts are highly skewed distributions: in ORBIT, the individual median fracture rate was 0.58 with setrusumab and 0.00 with placebo, while means were 0.92 and 0.80, so a median of zero does not mean the group had no fractures [73]. Relative reductions should be read alongside the absolute residual rate (for example, in the risedronate trial, 31% of treated children had clinical fractures within a year), and differences in secondary or post hoc endpoints do not replace a negative primary endpoint [56, 73]. Negative trials — ASTEROID for its primary endpoint, ORBIT, COSMIC, TOPaZ and the paediatric denosumab study — are part of the evidence and must not be omitted [15, 65, 70, 71].

Distinguishing evidence extrapolation from off-label use is essential here. Almost all pharmacological treatment of paediatric OI is off-label: US zoledronic acid, risedronate and alendronate labels state that they are not indicated in children, the European Aclasta® SmPC states it should not be used below 18 years, and doses used derive from published protocols and trials rather than labels [12–14, 54, 55, 60]. In adults, bisphosphonate, teriparatide or romosozumab use is within the label only when the patient meets the osteoporosis indication in the relevant jurisdiction (for romosozumab, only in postmenopausal women, and in the EU only for severe osteoporosis); application to OI extrapolates evidence obtained in common osteoporosis [76, 85]. Teriparatide is excluded below 18 years and in young adults with open epiphyses, and denosumab is contrary to the European SmPC below 18 years [64, 86]. In pregnancy, bisphosphonates cross the placenta and should be withdrawn before a planned pregnancy (section 7.2), and none of these medicines has specific safety data in pregnant women with OI.

Divergences between the French PNDS and the other sources. The rule followed is that the guideline is the reference and the label the comparison; when two recommendations diverge, both are presented, with the date and the population to which they apply.

7.10 Acute fracture: analgesia and immobilisation

The French PNDS refers to the Orphanet Urgences (emergency) sheet for OI and organises management in two stages: oral analgesia at home, with appropriate analgesics kept in a safe place, before transfer to hospital, and immobilisation, which the protocol considers the first of the analgesics and which facilitates transport and radiology [8].

The PNDS tramadol and morphine doses, written in 2016, differ from the current French SmPCs, which space tramadol doses 6 to 8 hours apart, limit each dose to 100 mg and warn about ultrarapid metabolisers and accidental overdose (bottle kept out of the reach of children), and from the US label, which contraindicates tramadol under 12 years. The protocol dose is the reference within its French framework; the labels serve as comparison and as a safety limit, and, where tramadol is not authorised at the child's age, the protocol's own alternative is titrated oral morphine [8, 87, 88, 91].

«Home-made» immobilisation. The equipment should always be ready in all the places where the child lives, prepared with a professional from the team (physiotherapist, physiatrist, orthopaedic surgeon): a small bag with 2 or 3 Velpeau bandages and weight-adjusted analgesics. On the lower limb, a slightly curved cardboard splint is used, to gain rigidity, padded with cotton wool and extending from the axilla to the foot; in the presence of a femoral or tibial fracture, the limb is held in gentle traction for a few minutes until the muscles relax, the splint is placed along the body, from under the arm to the foot, and fixed with circular bandages. On the upper limb, a small splint the length of the arm and forearm, with the elbow at 90°, immobilises the limb against the body with bandages passing over the shoulder on the fractured side and not the contralateral one. The child is not undressed on site (the splint is placed over the clothing); the most distal hand exerts gentle traction along the axis of the limb to align it; it is preferable not to be alone; and, on arrival at the emergency department, a comfortable immobilisation should not be changed: the first radiographs are taken with the splint, which is compatible with the vacuum mattress of the emergency services. For small, non-serious fractures, these immobilisations allow a return home or to the usual hospital in good conditions [8].

7.11 Respiratory, cardiovascular, haemorrhagic and anaesthetic care

In moderate to severe forms with thoracic deformity, kyphoscoliosis or a short thorax, the French PNDS recommends pulmonary function tests in all children with a moderate or severe form (from 5 years in forms with vertebral fractures, repeated every 2 to 5 years according to severity) and before spinal surgery; polysomnography in severe forms, in complications of the cervical spine and when obstructive sleep apnoea is suspected; prevention of respiratory infections (influenza vaccination, avoidance of contagion and of smoking); a low threshold for antibiotic therapy in the presence of an infection of bacterial appearance; bacteriological examination of sputum in the presence of repeated or chronic bronchorrhoea; and daily hyperinflation techniques and/or nocturnal non-invasive ventilation, according to the pulmonary function tests and the sleep study, initiated and monitored by a qualified pulmonologist, with minimally traumatic interfaces (mouthpiece for hyperinflation, nasal cannulae or pillows for ventilation) [8].

On the cardiovascular side, the PNDS proposes echocardiography at 10 years in all patients, repeated every 2 or 3 years in forms with marked hyperlaxity or in the presence of symptoms, with screening for pulmonary hypertension in patients with respiratory failure; cardiac surgery, delicate because of tissue fragility and the bleeding risk, is performed in a specialised setting [8].

Prevention of the bleeding risk is based on careful surgical haemostasis and prolonged postoperative monitoring. In the presence of a clear bleeding tendency, haemostasis testing is performed (in particular, screening for von Willebrand factor deficiency) and, if no abnormality is identified, desmopressin 0.3 µg/kg in 30 mL of normal saline over 30 minutes or tranexamic acid 20 mg/kg/day in 3 doses is discussed before surgery [8]. The French SmPC for injectable desmopressin (Minirin®) provides for the same dose of 0.3 µg/kg (0.2 µg/kg in the elderly or with cardiovascular disease), as an infusion over 15 to 30 minutes diluted in 50 to 100 mL of normal saline, immediately before the procedure, repeatable 1 to 2 times every 12 or 24 hours with progressive loss of effect, with no absolute maximum dose stated (the US label sets a maximum of 20 µg [92]), and requires fluid restriction and monitoring of weight and serum sodium, particularly in children, because of the risk of hyponatraemia [93]; the SmPC for tranexamic acid (Exacyl®) indicates about 20 mg/kg/day in children from 1 year, with limited efficacy and safety data, and contraindicates it in severe renal impairment, acute venous or arterial thrombosis and with a history of seizures [94]. The desmopressin dose is concordant between the PNDS and the SmPC; only the dilution volume differs (30 mL in the PNDS, 50–100 mL in the SmPC).

In anaesthesia, in severe forms, the PNDS prefers regional to general anaesthesia because of possible intubation difficulty, mentions some published cases of malignant hyperthermia and accepts epidural or spinal anaesthesia if there is no spinal fusion hardware in the puncture area [8].

---

8. Reproductive Options and Family Counselling

Genetic counselling should be systematically offered to all parents of a child with OI, necessarily addressing the possibility of germline mosaicism and its associated recurrence risk [9].

Disclosure of the diagnosis and psychological support (French PNDS). The diagnosis is disclosed in a dedicated consultation, by a physician experienced in OI, to both parents, and covers the manifestations and their variability, the genetic origin and the indications for molecular testing, the broad outlines of medical and surgical care, the organisation of follow-up and the reference physicians, the lifestyle rules that reduce fracture risk, attendance at nursery and school (possible in most cases), adjustments in the workplace and administrative procedures; in severe forms it is often done progressively, in several stages. Psychological support, offered to the child, the parents and the siblings, by a local psychologist (medico-psycho-pedagogical centre or private practice), or at the competence or reference centre, is particularly important when the disease arises de novo and abuse was suspected, sometimes with judicial measures and temporary removal of the child: wrongly suspected parents may be left feeling guilty and have difficulty parenting. Throughout childhood, it is recommended to provide schooling with the necessary adjustments, not to restrict the child more than necessary, to help the child find other interests when sport is not possible, to avoid a depressive atmosphere around the child and to support entry into employment (in France, through the commission for the rights and autonomy of disabled people of the MDPH); the family should receive the contact details of the patient association (Association de l'Ostéogenèse Imparfaite, AOI) [8].

For types with autosomal dominant inheritance and complete penetrance — including types 1 to 5 and the heterozygous form of type 15 —, each child of a clinically affected parent has a 50% probability of inheriting the pathogenic variant and therefore developing the disease, regardless of the child's sex [7].

When the index case results from a de novo mutation — a circumstance characterising most lethal and severe forms, reflecting natural selection against reproduction by individuals with disabling forms; virtually 100% of probands with OI types II and III are sporadic cases [7] —, recurrence risk in subsequent pregnancies of the same parents is low but not zero, because of parental germline mosaicism (somatic and/or germline). If the proband's variant is not detected in either parent, possible explanations include a de novo variant or very low-level somatic mosaicism involving the germline in a parent [7]. Parental mosaicism identified in families with lethal OI and a dominant variant was 16%, corresponding to a measured recurrence risk of 1.3% after one affected pregnancy and 32% after two or more affected children [10]; the historical reference value from Cole and Dalgleish (1995) was 7% [11], and some of these cases may have been recessive forms with phenotypic overlap [9]. For type three, the most common form is autosomal dominant, with a germline mosaicism risk similar to type two; Spranger reports an empirical risk of 7% to 8% for the IIB/III phenotype [9, 25]. For this group of families, molecular prenatal diagnosis in future pregnancies is always recommended when possible, even when neither parent has any clinical manifestation of the disease. For unaffected parents with a first affected child, the French PNDS uses a recurrence risk of the order of 5%, attributed to germline mosaicism, a value lying between the measured risk of 1.3% and the historical estimate of 7% (section 7.9) [8].

For types with autosomal recessive inheritance — including types 6 to 14, 16, 17, 18, 20, 21, 22 and 23 —, parents are typically asymptomatic heterozygous carriers of a single copy of the pathogenic variant, and recurrence risk is 25% in each pregnancy regardless of the child's sex, with an additional 50% probability that the child will be an asymptomatic carrier and 25% that the child will be genetically a non-carrier. Genetic testing to detect heterozygous carrier status can be performed [7]. Children of an individual with a recessive form will be obligate carriers and will only develop the disease if they receive another pathogenic variant from the other parent; risk to other relatives depends on the proband's parents' status, and in recessive forms the greatest risk is concentrated in the proband's siblings (25%) [7].

For type 19, with X-linked recessive inheritance (MBTPS2 gene), carrier mothers have a 50% probability of transmitting the variant to each child; sons inheriting the affected maternal X chromosome will be clinically affected, whereas daughters inheriting the variant will generally be asymptomatic carriers or have attenuated manifestations, depending on the X-inactivation pattern. The French PNDS describes the X-linked forms as symptomatic mainly in boys, sometimes with minor signs in the carrier mother, and counselling as reassuring for the children of an affected man, whose daughters will be healthy or paucisymptomatic carriers [8].

Carrier detection is possible in relatives when the causative variant is already known in the family, through targeted molecular analysis (single-variant testing). Careful clinical assessment of parents is always recommended before assuming a mutation is de novo, including clinical examination and molecular genetic testing if the proband's variant has been identified, with targeted assessment of blue sclerae, dentinogenesis imperfecta, joint hypermobility and previous fractures — a parent may have a mild form with subtle phenotypic expression that was previously undiagnosed [7].

Prenatal diagnosis is requested particularly by couples with a child with OI type two or type three [9]. Available options include obstetric ultrasound at a referral centre, capable of detecting recurrence of type II and most type III cases from the second trimester — safe and effective, but providing relatively late diagnosis [9]; chorionic villus sampling (CVS), possible from 11 weeks of gestation for direct molecular analysis when the familial variant is already known, or when diagnosis has been confirmed by molecular methods or collagen studies; and preimplantation genetic diagnosis (PGD), involving genetic analysis of embryos obtained through in vitro fertilisation before uterine transfer, technically possible when the familial mutation is known in couples choosing in vitro fertilisation, eliminating the need to consider termination of potentially affected pregnancies. Amniocentesis is unsuitable for collagen studies [9].

Regarding mode of delivery, a review of 167 affected pregnancies by Cubert and colleagues in 2001 found a 37% incidence of breech presentation at term — substantially above that in the general population — and new fracture rates at birth of 40% among caesarean births and 32% among vaginal births in foetuses with non-lethal forms [89]. Caesarean section did not reduce fracture rates at birth in non-lethal forms or prolong survival in lethal forms; Smith's likewise reports that delivery mode does not appear to alter fracture probability [21]. Thus, when a foetus is diagnosed with OI before birth, vaginal delivery is generally advised in the absence of specific obstetric indications for caesarean section, reserving caesarean section for inadequate progress in the second stage of labour or foetal distress, and avoiding instrumental delivery whenever possible [9].

Pregnancy, delivery and breastfeeding in women with OI (French PNDS). In moderate or severe forms, follow-up and delivery take place at a reference or competence centre, or in liaison with one, in a level 2b or 3 maternity unit with facilities adapted for the mother; in mild forms, pregnancy is followed under the usual conditions. Preconception consultation is indispensable in moderate or severe forms, ideally with the reference centre physician and the obstetrician who will follow the pregnancy, and includes assessment of the disease (radiological if necessary), standing and supine pulmonary function tests to predict tolerance of late pregnancy, and adaptation of treatments (suspension of bisphosphonates, section 7.2). Tolerance depends on the type and severity, height and respiratory function, and pregnancy may reduce height by a few centimetres. Thoracolumbar back pain worsens in more than 50% of pregnancies and should prompt radiographs in the third trimester to look for vertebral fractures, being treated with rest and analgesics authorised in pregnancy; the uterus becomes intra-abdominal from the first trimester, which is uncomfortable particularly for women of short stature, and tends to anteversion because of weakness of the abdominal wall; the abdominal volume may require a wheelchair and hospitalisation at the end of pregnancy, with thromboprophylaxis; and, in the most severe cases, with pelvic abnormalities, the cervix may be long, which matters for the management of miscarriages and terminations of pregnancy for medical reasons. Foetal ultrasound follow-up is the usual one, with an additional scan at 18 weeks of gestation when prenatal diagnosis has not been performed; psychological support is offered, because pregnancy may be experienced negatively owing to limitations and physical suffering, and postpartum fatigue justifies home help. Calcium (plentiful dairy products) and vitamin D should start at the beginning of pregnancy, with daily vitamin D or one ampoule in the third trimester [8]; the reference intake in pregnancy is 600 IU/day of vitamin D and 1,000 mg/day of calcium from 19 to 50 years [84].

The mode of delivery depends on the type and severity: caesarean section is not systematic in moderate and intermediate forms, all women undergo CT pelvimetry (the pelvis is frequently narrow), and in severe forms, in which caesarean section is mandatory, regional anaesthesia is preferred because of possible intubation difficulty, epidural or spinal anaesthesia being possible without spinal fusion hardware in the puncture area (otherwise, the decision is discussed with a spinal surgeon). The bleeding risk, theoretically greater even with normal haemostasis, weighs on the choice of route, and the uterus, being more fragile, follows the protocols for the scarred uterus. Breastfeeding is discussed with the woman and prolonged breastfeeding should be avoided, because it may cause asthenia and back pain and worsen osteopenia, although it is possible with well-supervised maternal calcium intake [8]. These recommendations concern the affected mother; when the foetus is affected and the mother is not, the series by Cubert and colleagues supports vaginal delivery in the absence of an obstetric indication [89].

---

9. Special Considerations: Population Diversity and Epidemiology

Osteogenesis imperfecta is considered a rare disease. Estimated global prevalence across all types is between 6 and 7 cases per 100,000 individuals — in a country with two hundred million inhabitants, this corresponds to between twelve thousand and fourteen thousand people with the disease —, with birth incidence, representing new cases identified in the neonatal period, between 1 in 15,000 and 1 in 20,000 live births [3, 4, 27]; the Brazilian source cites a global incidence of 1:10,000 births [7]. Type 1, the mildest, non-deforming form, has the highest individual prevalence, estimated at between 2 and 5 cases per hundred thousand inhabitants [9], and approximately 5% of non-lethal OI is type V [31]. The French PNDS estimates the birth prevalence at about 1 per 10,000, with about 90% autosomal dominant forms due to monoallelic variants in COL1A1, COL1A2 or IFITM5 and about 10% recessive forms, mainly autosomal (14 genes known in September 2016) and exceptionally X-linked (PLS3, MBTPS2), usually of intermediate to severe severity; the protocol does not give numbers of patients followed in France [8]. The PNDS classifies PLS3 forms as X-linked recessive; in the largest published cohort (28 patients from 11 Dutch families), men had low bone mineral density and early fractures, but 3 of the 10 heterozygous women had osteoporosis or osteopenia, and the risk in carriers warrants its own assessment [30].

The aetiological distribution of osteogenesis imperfecta shows relevant geographical variation, directly related to the frequency of consanguineous marriages in each population. Approximately ninety per cent of cases are caused by COL1A1 and COL1A2 mutations (85–90% according to the Oxford Desk Reference) [5, 9]; in Western countries, these mutations account for 80% to 85% of cases; in countries or regions with a high prevalence of consanguinity — described particularly in parts of the Middle East, southern Asia and southern Africa —, this proportion falls to approximately 60%, reflecting proportionally greater representation of autosomal recessive forms, whose manifestation depends on inheritance of two copies of the altered allele, one from each parent, an event statistically favoured by unions between relatives. In populations with a high carrier frequency for recessive forms, the proportion of recurrences attributable to parental mosaicism versus recessive inheritance changes, reinforcing the value of variant identification for counselling [10].

Several specific population founder effects merit emphasis for teaching. The autosomal recessive form of type III was originally described in southern African populations (predominantly of Black South African ancestry) and independently reported in Irish families and other consanguineous groups [9], illustrating how the same severe phenotypic presentation can arise from distinct, geographically dispersed mutational events, each maintained within its population of origin through genetic drift and possibly an unproven heterozygous advantage. In West Africa, a LEPRE1 founder variant (type 8), approximately 650 to 900 years old and carried by 1.48% of individuals in Nigeria and Ghana and 0.4% of African Americans in the US mid-Atlantic region, predicts an incidence of lethal recessive OI of 1 in 18,260 births in those countries — equal to that of de novo dominant OI — and 1 in 260,000 among African Americans, consistent with introduction of the variant into the Americas through the transatlantic slave trade [37]. Type 14 (TMEM38B) was defined through study of autosomal recessive OI in Saudi Arabia, and a TMEM38B deletion was described in consanguineous Israeli families [5]. Type 5, caused by a single recurrent point variant in IFITM5 (NM_001025295.3:c.-14C>T), represents a rare example of a truly recurrent hotspot mutation in the OI literature, independently identified in multiple unrelated populations (including Korean and European populations in the foundational studies by Cho and Semler, 2012), without evidence of a single founder effect — suggesting an intrinsically hypermutable site rather than shared ancestry [31].

Kuskokwim syndrome, associated with specific FKBP10 variants, is characterised by congenital joint contractures described in indigenous Yupik populations in Alaska (Kuskokwim region), illustrating a geographically isolated founder effect [32].

There is no evidence of a significant predilection for a particular sex or ethnic group in any autosomal type; in X-linked type 19, full clinical expression is, as expected, predominantly restricted to males. The sources consulted for this article did not yield robust population epidemiological data for Latin America, including Brazil, concerning the relative distribution of genetic OI subtypes, and the Brazilian molecular series cited in these sources are scarce; this observation reflects the scope of the sources and does not demonstrate the absence of published studies. The same applies to Portugal, sub-Saharan Africa (apart from the South African recessive form and the LEPRE1 founder variant carrier frequency in Nigeria and Ghana) and most of Asia, where subtype prevalence data are scarce. Incidence shows no demonstrated ethnic variation beyond the specific founder effects already described [3, 4, 27].

---

10. Illustrative Clinical Case

10.1 Composite case (illustrative): infant with recurrent fractures and maternal transmission

The following case is composed for educational purposes from typical features described in this article's sources; it does not correspond to a published patient.

Identification. Female infant referred to clinical genetics at 4 months for a right femoral fracture sustained during nappy changing, without a significant traumatic mechanism reported by caregivers.

History. Born at term by vaginal delivery, without reported complications at birth; birth weight and length appropriate for gestational age. At 2 months, she had a left humeral fracture, attributed at the time to routine handling during bathing, without further investigation. Family history: the mother reports “many fractures” in childhood and adolescence, informally attributed to “weak bones”; the maternal grandmother had hearing loss beginning in her fourth decade, of unexplained cause; there is no reported parental consanguinity.

Physical examination. Active infant with age-appropriate neuropsychomotor development. Clearly blue sclerae, also present more subtly in the mother on targeted reassessment. Height and head circumference within the normal range for age. No evident limb deformity. Mild to moderate joint hyperextensibility.

Further investigations. Skeletal survey showed a healing right femoral fracture and a healed left humeral fracture, without other occult bone lesions identified; skull radiography showed multiple Wormian bones in the lambdoid suture. Bone densitometry, performed later at 5 years, showed a lumbar spine Z-score of −1.6. Serum biochemical tests, including calcium, phosphate, vitamin D and alkaline phosphatase, were normal, a finding which, together with the other features, made hypophosphatasia unlikely and reinforced suspicion of osteogenesis imperfecta.

Differential diagnosis assessment. In an infant with multiple fractures at an early age, non-accidental trauma was formally considered and discussed with the service's child protection team, following institutional protocol. Bilateral blue sclerae (in the child and mother), radiographic Wormian bones, a consistent family history of bone fragility across three generations and absence of any signs suggesting inflicted trauma (no skin lesions, no metaphyseal avulsion fractures, no discrepancy between the reported mechanism and findings) strongly directed investigation towards a genetic cause, without excluding, at the time, the need for continuing surveillance.

Molecular diagnosis. The bone fragility gene panel identified the heterozygous nonsense variant NM_000088.4(COL1A1):c.2869C>T, p.(Gln957Ter), classified as pathogenic in ClinVar by several laboratories for OI type I; targeted testing showed the same variant in the mother, consistent with autosomal dominant inheritance and the classical non-deforming form with blue sclerae (OI type I) [48]. Teaching point: a premature stop codon in the helical domain leads to mRNA degradation through nonsense-mediated decay and type I collagen haploinsufficiency, the mechanism underlying most type I cases (section 2.2); a glycine substitution, conversely, would require more cautious interpretation because it is frequently associated with more severe phenotypes, and any variant requires formal application of ACMG/AMP criteria and consultation of locus-specific databases [43].

Management instituted. The child was referred to the multidisciplinary paediatric bone disease team, with detailed safe-handling advice for caregivers, baseline physiotherapy assessment and plans for serial bone mineral density and hearing reassessments (from 5 years). Given two fractures within a short interval, cyclical intravenous pamidronate was indicated under the paediatric protocol (0.5 mg/kg/day for 3 days, every 2 months, below 2 years), with paracetamol premedication and prior vitamin D correction. Parents were offered genetic counselling, explaining the 50% recurrence risk in future pregnancies given confirmed autosomal dominant inheritance with maternal transmission, and discussing available prenatal diagnostic options for future pregnancies.

Teaching comment. This case illustrates several principles discussed throughout this lesson: the importance of systematically assessing subtle signs (blue sclerae, Wormian bones, family history) when distinguishing mild OI from non-accidental trauma; the expected normal laboratory findings in classical OI; possible mild, undiagnosed manifestations in an apparently minimally symptomatic parent (the mother, whose history of “many childhood fractures” had been informally attributed to non-specific causes); and direct application of genetic counselling based on molecularly confirmed autosomal dominant inheritance.

10.2 Cases documented in the reference literature (Spranger et al.)

Complementing the illustrative case, the reference atlases document real cases highlighting specific teaching points. In type I, a 1-month-old infant (patient 4) had osteopenia and fractures in the distal submetaphyseal segments of the radius and ulna, with a lateral incomplete fracture of the distal humerus; the metaphyseal fractures raised suspicion of child abuse, but skull changes were typical of OI, illustrating the possibility of coexistence of both conditions; at 5 months, the same patient showed rib fractures and callus at the costochondral junctions [22]. In type IV, a newborn (patient 4) with mild femoral bowing and a distal-third left femoral fracture healed without deformity by 2 years, but by 25 years disease had progressed with rarefaction, thinning and bowing of the tibiae and fibulae, resembling OI type III — an example of overlap between phenotypes IV and III [23]. In type V, a 7-year-old girl (patient 7) with hyperplastic callus of the left femur and an intramedullary rod showed almost complete callus resolution at 8 years, with thin osteopenic long bones but sclerotic metaphyses [31]. In Bruck syndrome, a 2-year-old boy homozygous for PLOD2 p.Arg598His variant (published designation, NM_000935.3:c.1793G>A; in the MANE transcript NM_182943.3:c.1856G>A p.(Arg619His)) had elbow and knee flexion contractures with antecubital and popliteal pterygia and residual clubfoot, and a 12-year-old boy compound heterozygous for FKBP10 variants had finger and toe contractures, platybasia and multiple Wormian bones [32].

10.3 Published case: whole-COL1A1 deletion identified by genome sequencing

In 2026, Yamamoto and colleagues described a Japanese boy assessed at five years after three fractures at four years [45]. He was born at term without fractures or deformities, had normal height and weight (−0.80 and −0.60 standard deviations), blue sclerae, no dentinogenesis imperfecta or deformities; calcium, phosphate, alkaline phosphatase and parathyroid hormone were normal, but 25-hydroxyvitamin D was 6 ng/mL. Radiographs showed only slightly thin cortices, without Wormian bones, and lumbar density was 0.39 g/cm² (height-adjusted Z-score −1.8). His father, clinically diagnosed with OI, had had more than ten fractures, blue sclerae and osteoporosis treated with a bisphosphonate, and a younger brother had blue sclerae and one fracture. OI type I was clinically diagnosed and a bisphosphonate started. Exome sequencing found no point variants in the approximately 40 bone fragility genes analysed, but CNV analysis suggested an approximately 57 kb COL1A1 deletion; genome and Sanger sequencing defined a heterozygous 101 kb deletion (chr17:50,143,793–50,245,237, GRCh38) encompassing all of COL1A1, SGCA and part of PPP1R9B, also present in the father and brother, with a breakpoint in an Alu element and 5 base pairs of microhomology [45]. Figures in the original article show the patient's radiographs (figure 1) and the deletion in read coverage, PCR confirmation and breakpoint sequences (figure 2), available open access at https://pmc.ncbi.nlm.nih.gov/articles/PMC12878673/.

The case illustrates three points of this article. Absence of Wormian bones and only slightly low density do not exclude OI type I when fractures, blue sclerae and a dominant family history are present (sections 3.11 and 4.3). An exome “negative” for point variants does not exclude the diagnosis; whole-gene deletions, representing approximately 1% of pathogenic COL1A1 variants, require CNV analysis and genome sequencing to define breakpoints (section 5.5). A contiguous deletion may also reveal carrier states unrelated to OI: the authors stress that the patients are carriers of recessive SGCA-associated limb-girdle muscular dystrophy, warranting specific reproductive counselling, although they had neither weakness nor elevated creatine kinase [45]. Documented vitamin D deficiency should be corrected before any intravenous bisphosphonate (section 7.2); the case report does not describe the dose used, so no regimen is extrapolated from this report.

11. Prognosis and Long-Term Follow-Up

Prognosis in osteogenesis imperfecta varies widely according to genetic type and individual phenotypic severity, and recessive types vary according to the specific mutation. Patients with the classical non-deforming form (type I) have essentially normal life expectancy, with preserved function and full participation in educational, social and occupational activities in most cases with appropriate support; fracture frequency tends to decrease after puberty and may rise again in adulthood, particularly in postmenopausal women, who have a greater likelihood of osteoporosis [7]. The Brazilian source describes mild forms as having few fractures, no long-bone deformities, minimal deformity, usually normal height and possible premature hearing loss; moderate to severe forms as having increased fracture frequency, kyphoscoliosis, mild to moderate bone deformities, earlier osteoporosis and hearing loss, pain due to hypermobility and higher mortality risk than the general population; and the perinatal lethal form as having frequent death in utero or before the end of early childhood [7].

The perinatal lethal form (type II) is fatal in most cases, typically in the immediate neonatal period due to respiratory failure secondary to pulmonary hypoplasia and multiple rib fractures; affected children are usually stillborn or die in early infancy [21]; survival beyond the neonatal period is exceptional and, when it occurs, is associated with severe functional impairment and need for continuous intensive support. However, in a cohort of 18 infants with prenatally diagnosed OI and a predicted poor prognosis, followed at a specialist centre — 12 of whom had prenatal predictions of perinatal lethal OI based on genetic testing and/or ultrasound —, most survived the neonatal period and were discharged home, many with respiratory and/or feeding support, without correlation between prenatal assessment and postnatal survival; GeneReviews therefore stresses the need for caution in prenatal counselling [3].

In the progressively deforming form (type III), functional prognosis is the most guarded among non-lethal forms, with substantial functional impairment and a high risk of death from respiratory failure due to severe scoliosis or rib fractures, and most patients depend on a wheelchair or other assisted mobility devices; progressive trunk deformity may lead to restrictive lung disease [21], and survival is influenced mainly by the extent of skeletal deformities, cardiorespiratory impairment from severe scoliosis and neurological complications, particularly basilar invagination, and spinal cord compression from kyphoscoliosis [25]. Moderate forms (types IV and V) occupy an intermediate position, with near-normal life expectancy and variable functional limitations according to the degree of deformity and effectiveness of treatment instituted.

A Danish population-based cohort study (Folkestad et al., 2016), with 687 patients followed from 1977 to 2013, demonstrated an all-cause mortality hazard ratio of 2.90 relative to the reference population and median survival for men with OI (all types combined) of 72.4 years, compared with 81.9 years in the matched reference population, and 77.4 years for women with OI, compared with 84.5 years in the reference population — an average reduction of approximately 7 to 9 years in life expectancy, predominantly attributable to respiratory disease, gastrointestinal complications and traumatic events [95].

Recommended longitudinal follow-up includes orthopaedic assessment every 3 months up to 1 year of age, every 6 months between 1 and 3 years, and annually (or with new fractures) thereafter; serial bone densitometry from 5 years, with frequency according to severity and treatment response; audiological assessment from 5 years, repeated every 3 years until any hearing loss is identified; ophthalmological assessment every 2 to 3 years in adults; pulmonary function assessment by spirometry every 1 to 2 years in adults, with polysomnography (sleep study) for symptoms suggesting sleep apnoea; six-monthly or annual dental assessment; and cardiology assessment according to symptoms, given the possibility — although infrequent — of clinically relevant valvular or aortic involvement. Annual nutritional assessment, attending to the balance between underweight and overweight — both detrimental to bone health and mobility —, completes the recommended multidisciplinary follow-up programme [3].

The French PNDS provides for specialised hospital consultations once a year in children and every 3 years in adults, adapted to severity and complications, in follow-up shared with the attending physician (paediatrician or family doctor), who monitors height and weight growth with body mass index, adherence to and understanding of treatment and changes in course (pain, growth deceleration, increased fracture frequency), which are reported to the centre; imaging is preferably performed in paediatric or osteoarticular imaging facilities, with low-dose techniques (EOS) in children who undergo many radiographs. The follow-up annex distinguishes the moderate form from the complicated or severe form and adds, by age, a genetics clinic consultation every 5 years in adolescence, rodding of the long bones from 12–18 months in severe forms, bracing and spinal fusion, the long-term condition scheme and the MDPH, weight control, physical activity, genetic counselling and psychological support [8].

The interval divergences between GeneReviews and the PNDS (audiometry, echocardiography, orthopaedics in the first year, oral health) are discussed in section 7.9.

---

12. Summary and Final Considerations

Osteogenesis imperfecta is now a paradigm of genetic heterogeneity unified by a common pathophysiological pathway: direct or indirect impairment of type I collagen synthesis, processing or mineralisation. Since Sillence's phenotypic classification in 1979, the field has evolved from a strictly dominant, monogenic model centred on COL1A1 and COL1A2 to current recognition of at least 22 to 23 genetically distinct types, encompassing haploinsufficiency, dominant-negative effects, enzymatic deficiencies of post-translational modification, molecular chaperone dysfunction, altered osteoblast differentiation and primary mineralisation defects — a spectrum continuing to expand as next-generation sequencing is applied to previously unexplained cases, illuminating formerly unknown pathophysiological mechanisms and opening new therapeutic windows.

Therapeutically, intravenous bisphosphonates, particularly pamidronate and zoledronic acid, remain, more than two decades after Glorieux's study, the central pharmacological pillar of treatment, with established paediatric and adult protocols and demonstrated efficacy on bone mineral density, although without consistent evidence of fracture-rate reduction in systematic reviews. The 2025–2026 period marked a point of reassessment for the field: failure of the pivotal phase 3 setrusumab studies to meet their primary fracture-reduction endpoints — after phase 2 results considered promising — and the neutral TOPaZ trial result with teriparatide followed by zoledronic acid in adults, both with significant bone mineral density gains, illustrate the difficulty in translating bone mineral density gains into statistically demonstrable fracture reduction in this heterogeneous population, point to bone quality as the principal determinant of fragility, and reinforce the need for refined study designs, sensitive composite endpoints and genotype stratification in future clinical trials. Denosumab requires caution because of rebound hypercalcaemia in children. New molecules with dual mechanisms, such as the bispecific antibody AGA2115, anti-TGF-β antibodies, and experimental cellular approaches, such as foetal mesenchymal stem cells in the BoostB4 programme, are in clinical investigation, while genome editing-based gene therapies lie further on the horizon, with clinical implementation estimated a decade or more away; no therapy with OI-specific regulatory approval is currently available except neridronate in Italy. In 2026, setrusumab remained without an authorisation application, with dialogue with the FDA and MHRA on additional data, and follow-up of romosozumab and AGA2115 trials extends to 2027–2028; recent negative trials should be read alongside positive trials when deciding treatment [16, 74].

Genetic counselling remains an essential component of care, requiring specialists simultaneously to understand the different inheritance patterns involved, germline mosaicism and its practical implications for recurrence risk, and to approach these conversations sensitively in a setting often marked by family suffering associated with pregnancy or neonatal loss in the most severe forms.

Although incurable with current therapeutic resources, the course of osteogenesis imperfecta can be modified by early, systematic multidisciplinary intervention. Coordinated multidisciplinary management — involving clinical genetics, orthopaedics, bone endocrinology, physiotherapy, dentistry, otorhinolaryngology, ophthalmology, respiratory medicine and specialist pain management — remains the central pillar of care for the patient with osteogenesis imperfecta, regardless of pharmacological advances that may become established in coming years.

12.1 Online resources for professionals and families

For healthcare professionals, information sources include GeneReviews on osteogenesis imperfecta caused by COL1A1 and COL1A2, available through NCBI Bookshelf at www.ncbi.nlm.nih.gov/books/NBK1295, providing a regularly updated clinical and molecular review of the disease [3]; OMIM (www.omim.org), the online Mendelian inheritance database, with entries 166200 for type 1, 166210 for type 2, 259420 for type 3 and 166220 for type 4, alongside individual entries for all other types with updated references; Orphanet (ORPHA:666) [4]; ClinVar, listing documented pathogenic variants in each gene; Medscape Genetics at emedicine.medscape.com/article/947588, with clinical reviews of diagnosis and treatment; the European Skeletal Dysplasia Network at esdn.org and UK Skeletal Dysplasia Group at skeletaldysplasia-group.org.uk, additional resources for skeletal dysplasia specialists. For families, international support groups include the United States Osteogenesis Imperfecta Foundation at oif.org [96] and the United Kingdom Brittle Bone Society at brittlebone.org [97], offering educational materials, links to peer support groups and clinical research updates for patients, caregivers and healthcare professionals. In Brazil, Associação Brasileira de Osteogénese Imperfeita (ABOI) provides Portuguese-language support for patients and families [98]. In Europe, the relevant European Reference Network is ERN BOND (European Reference Network on Rare Bone Diseases), linking rare bone disease specialist centres, and Osteogenesis Imperfecta Federation Europe (OIFE) brings together national patient associations; in Portugal, APOI — Associação Portuguesa de Osteogénese Imperfeita — supports patients and families [18–20].

In France, care is organised at the reference centre for constitutional bone diseases (CR MOC; paediatric coordination at Necker-Enfants Malades Hospital, with Armand-Trousseau Hospital, and adult sites at the Cochin and Lariboisière hospitals), with 13 regional competence centres in the 2016 list (Caen, Clermont-Ferrand, La Réunion, Lille, Lyon, Marseille, Montpellier, Nancy, Nantes, Nice, Rennes, Strasbourg and Toulouse), linked with the reference centres for calcium and phosphate metabolism and for rare endocrine growth disorders and, for oral health, with the reference centre for malformations of the face and oral cavity (MAFACE), the reference centre for the oral and dental manifestations of rare diseases (Strasbourg) and their competence centres. The OSCAR health network (filière; rare diseases of bone, calcium and cartilage), based at Necker, brought together in 2016 three coordinating reference centres, 10 constituent centres, 39 competence centres, 19 patient associations, 14 laboratories, 18 research units and 9 scientific societies, and maintains the list of laboratories accredited for molecular testing and educational materials, such as the videos on «home-made» immobilisation; the patient association is the Association de l'Ostéogenèse Imparfaite (AOI, www.aoi.asso.fr), and the Tous à l'école platform has a bone fragility information sheet for schools. For fracture management, the PNDS refers to the Orphanet Urgences (emergency) sheet for OI [8]. These French structures belong to the European network ERN BOND.

---

13. References

1. Sillence DO, Senn A, Danks DM. Genetic heterogeneity in osteogenesis imperfecta. J Med Genet. 1979;16(2):101-116. doi:10.1136/jmg.16.2.101
2. Van Dijk FS, Sillence DO. Osteogenesis imperfecta: clinical diagnosis, nomenclature and severity assessment. Am J Med Genet A. 2014;164A(6):1470-1481. doi:10.1002/ajmg.a.36545
3. Rodriguez Celin M, Steiner RD, Basel D. COL1A1- and COL1A2-Related Osteogenesis Imperfecta. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 2005 Jan 28 [Updated 2025 May 29]. Available at: https://www.ncbi.nlm.nih.gov/books/NBK1295/
4. Orphanet. Osteogenesis imperfecta (ORPHA:666) and subtypes ORPHA:216796, 216804, 216812. Orphanet: The portal for rare diseases and orphan drugs. Available at: https://www.orpha.net/en/disease/detail/666
5. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 14.1, Osteogenesis imperfecta: overview and Table 14.1.1. p. 497–500.
6. Wang XQ, Zhang C, Li X, et al. Pathogenesis and precision management of osteogenesis imperfecta: from genetic mechanisms to novel therapies. World J Pediatr. 2026 Sep 30. doi:10.1007/s12519-026-01086-7
7. Pereira DL, Carvalho EDF, Ribeiro EM. Osteogênese imperfeita. In: Ribeiro EM, coord. Síndromes Genéticas: Reconhecer para Tratar. Fortaleza: EdUnichristus; 2021. p. 269–272.
8. Haute Autorité de Santé. Protocole National de Diagnostic et de Soins (PNDS): Ostéogenèse imparfaite. Saint-Denis La Plaine: HAS; 2017 Jan 4 (text by the Centre de Référence Maladies Osseuses Constitutionnelles, filière OSCAR, February 2017) [accessed 2026 Nov 6]. Available at: https://www.has-sante.fr/jcms/c_2734083
9. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Fractures. p. 158–161.
10. Pyott SM, Pepin MG, Schwarze U, Yang K, Smith G, Byers PH. Recurrence of perinatal lethal osteogenesis imperfecta in sibships: parsing the risk between parental mosaicism for dominant mutations and autosomal recessive inheritance. Genet Med. 2011;13(2):125-130. doi:10.1097/GIM.0b013e318202e0f6
11. Cole WG, Dalgleish R. Perinatal lethal osteogenesis imperfecta. J Med Genet. 1995;32(4):284-289. doi:10.1136/jmg.32.4.284
12. U.S. Food and Drug Administration. RECLAST (zoledronic acid) injection. Prescribing information, section 8.4 Pediatric Use. NDA 021817; version dated 03/02/2026 [accessed via openFDA on 2026 Oct 6]. Available at: https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22RECLAST%22
13. U.S. Food and Drug Administration. ACTONEL (risedronate sodium) tablets. Prescribing information, section 8.4 Pediatric Use. NDA 020835; version dated 04/02/2026 [accessed via openFDA on 2026 Oct 6]. Available at: https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22ACTONEL%22
14. U.S. Food and Drug Administration. FOSAMAX PLUS D (alendronate sodium/cholecalciferol) tablets. Prescribing information, section 8.4 Pediatric Use (paediatric alendronate trial in OI). NDA 021762; version dated 03/02/2026 [accessed via openFDA on 2026 Oct 6]. Available at: https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22FOSAMAX%22
15. Ultragenyx Pharmaceutical Inc. Ultragenyx Announces Phase 3 Orbit and Cosmic Results for Setrusumab (UX143) in Osteogenesis Imperfecta [press release]. 29 Dec 2025. Available at: https://ir.ultragenyx.com/news-releases/news-release-details/ultragenyx-announces-phase-3-orbit-and-cosmic-results-setrusumab
16. Mereo BioPharma Group plc. Mereo BioPharma reports second quarter 2026 financial results and provides corporate update [press release; exhibit 99.1 to Form 8-K]. London; 2026 Aug 11. SEC EDGAR, accession 0001193125-26-343519. Available at: https://www.sec.gov/Archives/edgar/data/1719714/000119312526343519/mreo-ex99_1.htm
17. National Center for Biotechnology Information; EMBL-EBI. Matched Annotation from NCBI and EMBL-EBI (MANE), version 1.5, file MANE.GRCh38.v1.5.summary [accessed 2026 Oct 6]. Available at: https://ftp.ncbi.nlm.nih.gov/refseq/MANE/MANE_human/release_1.5/
18. ERN BOND — European Reference Network on Rare Bone Diseases [Internet; accessed 2026 Oct 6]. Available at: https://ernbond.eu
19. Osteogenesis Imperfecta Federation Europe (OIFE) [Internet; accessed 2026 Oct 6]. Available at: https://oife.org
20. APOI — Associação Portuguesa de Osteogénese Imperfeita [Internet; accessed 2026 Oct 6]. Available at: https://www.apoi.pt
21. Jones KL, Jones MC, del Campo M. Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. Osteogenesis imperfecta. p. 682–687.
22. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 14.2, Osteogenesis imperfecta, type I. p. 501–505.
23. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 14.6, Osteogenesis imperfecta, type IV. p. 517–520.
24. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 14.3, Osteogenesis imperfecta, type IIA. p. 506–508.
25. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 14.5, Osteogenesis imperfecta, type III/IIB. p. 511–516.
26. Rauch F, Glorieux FH. Osteogenesis imperfecta. Lancet. 2004;363(9418):1377-1385. doi:10.1016/S0140-6736(04)16051-0
27. Forlino A, Marini JC. Osteogenesis imperfecta. Lancet. 2016;387(10028):1657-1671. doi:10.1016/S0140-6736(15)00728-X
28. Marini JC, Blissett AR. New genes in bone development: what's new in osteogenesis imperfecta. J Clin Endocrinol Metab. 2013;98(8):3095-3103. doi:10.1210/jc.2013-1505
29. Byers PH, Pyott SM. Recessively inherited forms of osteogenesis imperfecta. Annu Rev Genet. 2012;46:475-497. doi:10.1146/annurev-genet-110711-155608
30. Zervou Z, Bruggenwirth HT, Demirdas S, Zillikens MC. Phenotypic and genetic characteristics of a Dutch cohort of patients with X-linked osteoporosis due to PLS3 genetic variants. JBMR Plus. 2025;9(6):ziaf046. doi:10.1093/jbmrpl/ziaf046. PMID: 40353206
31. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 14.7, Osteogenesis imperfecta, type V. p. 521–525.
32. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 14.9, Bruck syndrome. p. 529–531.
33. Marlowe A, Pepin MG, Byers PH. Testing for osteogenesis imperfecta in cases of suspected non-accidental injury. J Med Genet. 2002;39(6):382-386. doi:10.1136/jmg.39.6.382
34. Folkestad L, Prakash SK, Nagamani SCS, Andersen NH, Carter E, Hald JD, et al. Cardiovascular disease in adults with osteogenesis imperfecta: clinical characteristics, care recommendations, and research priorities identified using a modified Delphi technique. J Bone Miner Res. 2025;40(2):211-21. doi:10.1093/jbmr/zjae197. PMID: 39665364
35. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 14.4, Osteogenesis imperfecta, type IIC. p. 509–510.
36. Shapiro JR, Lietman C, Grover M, Lu JT, Nagamani SC, Dawson BC, et al. Phenotypic variability of osteogenesis imperfecta type V caused by an IFITM5 mutation. J Bone Miner Res. 2013;28(7):1523-30. doi:10.1002/jbmr.1891. PMID: 23408678
37. Cabral WA, Barnes AM, Adeyemo A, Cushing K, Chitayat D, Porter FD, et al. A founder mutation in LEPRE1 carried by 1.5% of West Africans and 0.4% of African Americans causes lethal recessive osteogenesis imperfecta. Genet Med. 2012;14(5):543-51. doi:10.1038/gim.2011.44. PMID: 22281939
38. Fiscaletti M, Biggin A, Bennetts B, Wong K, Briody J, Pacey V, et al. Novel variant in Sp7/Osx associated with recessive osteogenesis imperfecta with bone fragility and hearing impairment. Bone. 2018;110:66-75. doi:10.1016/j.bone.2018.01.031. PMID: 29382611
39. Ludwig K, Ward LM, Khan N, Robinson ME, Miranda V, Bardai G, et al. Dominant osteogenesis imperfecta with low bone turnover caused by a heterozygous SP7 variant. Bone. 2022;160:116400. doi:10.1016/j.bone.2022.116400. PMID: 35367406
40. Pollitt RC, Saraff V, Dalton A, Webb EA, Shaw NJ, Sobey GJ, et al. Phenotypic variability in patients with osteogenesis imperfecta caused by BMP1 mutations. Am J Med Genet A. 2016;170(12):3150-6. doi:10.1002/ajmg.a.37958. PMID: 27576954
41. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Suspected non-accidental injury. p. 326–327.
42. Bardai G, Moffatt P, Glorieux FH, Rauch F. DNA sequence analysis in 598 individuals with a clinical diagnosis of osteogenesis imperfecta: diagnostic yield and mutation spectrum. Osteoporos Int. 2016;27(12):3607-13. doi:10.1007/s00198-016-3709-1. PMID: 27509835
43. P122: A locus-specific database of COL1A1 and COL1A2 gene variants associated with osteogenesis imperfecta. Genetics in Medicine Open. 2026. Available at: https://www.sciencedirect.com/science/article/pii/S2949774426001263
44. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Short limbs. p. 784–787.
45. Yamamoto K, Nakayama H, Ito Y, Hattori M, Shimada T, Ueda I, et al. Osteogenesis imperfecta with a gross deletion including the COL1A1 gene, induced by Alu-driven microhomology-mediated end joining. Bone Rep. 2026;28:101901. doi:10.1016/j.bonr.2026.101901. PMID: 41660581; PMCID: PMC12878673
46. UniProt Consortium. UniProtKB P02452 (CO1A1_HUMAN): signal peptide 1–22, N-terminal propeptide 23–161, N-terminal non-helical region 162–178, triple-helical region 179–1192 [accessed 2026 Oct 6]. Available at: https://rest.uniprot.org/uniprotkb/P02452
47. Gensure RC, Mäkitie O, Barclay C, Chan C, Depalma SR, Bastepe M, et al. A novel COL1A1 mutation in infantile cortical hyperostosis (Caffey disease) expands the spectrum of collagen-related disorders. J Clin Invest. 2005;115(5):1250-7. doi:10.1172/JCI22760. PMID: 15864348
48. National Center for Biotechnology Information. ClinVar: records NM_000088.4(COL1A1):c.2869C>T (p.Gln957Ter), c.3040C>T (p.Arg1014Cys; VCV000017347) and c.934C>T (p.Arg312Cys; VCV000017343); NM_001025295.3(IFITM5):c.-14C>T [accessed 2026 Oct 6]. Available at: https://www.ncbi.nlm.nih.gov/clinvar/
49. Jones KL, Jones MC, del Campo M. Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. Ehlers-Danlos syndromes. p. 674–677.
50. Aragão JVA, Ribeiro EM. Síndrome de Ehlers-Danlos. In: Ribeiro EM, coord. Síndromes Genéticas: Reconhecer para Tratar. Fortaleza: EdUnichristus; 2021. p. 256–260.
51. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Ehlers–Danlos syndrome (EDS). p. 406–407.
52. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. Chapter 17.2, Infantile cortical hyperostosis. p. 623–626.
53. Mueller B, Engelbert R, Baratta-Ziska F, Bartels B, Blanc N, Brizola E, et al. Consensus statement on physical rehabilitation in children and adolescents with osteogenesis imperfecta. Orphanet J Rare Dis. 2018;13(1):158. doi:10.1186/s13023-018-0905-4. PMID: 30201006
54. Trejo P, Rauch F. Osteogenesis imperfecta in children and adolescents-new developments in diagnosis and treatment. Osteoporos Int. 2016;27(12):3427-3437. doi:10.1007/s00198-016-3723-3
55. Cannalire G, Biasucci G, Bertolini L, Patianna V, Petraroli M, Pilloni S, et al. Osteoporosis and bone fragility in children: diagnostic and treatment strategies. J Clin Med. 2024;13(16):4951. doi:10.3390/jcm13164951
56. Bishop N, Adami S, Ahmed SF, et al. Risedronate in children with osteogenesis imperfecta: a randomised, double-blind, placebo-controlled trial. Lancet. 2013;382(9902):1424-1432. doi:10.1016/S0140-6736(13)61091-0
57. Glorieux FH, Bishop NJ, Plotkin H, Chabot G, Lanoue G, Travers R. Cyclic administration of pamidronate in children with severe osteogenesis imperfecta. N Engl J Med. 1998;339(14):947-952. doi:10.1056/NEJM199810013391402
58. Plotkin H, Rauch F, Bishop NJ, et al. Pamidronate treatment of severe osteogenesis imperfecta in children under 3 years of age. J Clin Endocrinol Metab. 2000;85(5):1846-1850. doi:10.1210/jcem.85.5.6584
59. Dwan K, Phillipi CA, Steiner RD, Basel D. Bisphosphonate therapy for osteogenesis imperfecta. Cochrane Database Syst Rev. 2016;10(10):CD005088. doi:10.1002/14651858.CD005088.pub4
60. European Medicines Agency. Aclasta (zoledronic acid): summary of product characteristics, section 4.2 [Internet]. Amsterdam: EMA [accessed 2026 Oct 6]. Available at: https://www.ema.europa.eu/en/documents/product-information/aclasta-epar-product-information_en.pdf
61. Land C, Rauch F, Travers R, Glorieux FH. Osteogenesis imperfecta type VI in childhood and adolescence: effects of cyclical intravenous pamidronate treatment. Bone. 2007;40(3):638-44. doi:10.1016/j.bone.2006.10.010. PMID: 17127117
62. Hoyer-Kuhn H, Franklin J, Allo G, Kron M, Netzer C, Eysel P, et al. Safety and efficacy of denosumab in children with osteogenesis imperfect--a first prospective trial. J Musculoskelet Neuronal Interact. 2016;16(1):24-32. PMID: 26944820
63. Liu J, Lin X, Sun L, et al. Safety and efficacy of denosumab in children with osteogenesis imperfecta—the first prospective comparative study. J Clin Endocrinol Metab. 2024;109(7):1827-1836. doi:10.1210/clinem/dgad732
64. European Medicines Agency. Prolia (denosumab): summary of product characteristics, sections 4.2 and 5.1 [Internet]. Amsterdam: EMA [accessed 2026 Oct 6]. Available at: https://www.ema.europa.eu/en/documents/product-information/prolia-epar-product-information_en.pdf
65. U.S. Food and Drug Administration. PROLIA (denosumab) injection. Prescribing information, section 8.4 Pediatric Use. BLA 125320; version dated 26/06/2026 [accessed via openFDA on 2026 Oct 6]. Available at: https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22PROLIA%22
66. Marini JC, Hopkins E, Glorieux FH, Chrousos GP, Reynolds JC, Gundberg CM, et al. Positive linear growth and bone responses to growth hormone treatment in children with types III and IV osteogenesis imperfecta: high predictive value of the carboxyterminal propeptide of type I procollagen. J Bone Miner Res. 2003;18(2):237-43. doi:10.1359/jbmr.2003.18.2.237
67. Antoniazzi F, Bertoldo F, Mottes M, Valli M, Sirpresi S, Zamboni G, et al. Growth hormone treatment in osteogenesis imperfecta with quantitative defect of type I collagen synthesis. J Pediatr. 1996;129(3):432-9. doi:10.1016/s0022-3476(96)70077-x
68. Orwoll ES, Shapiro J, Veith S, Wang Y, Lapidus J, Vanek C, et al. Evaluation of teriparatide treatment in adults with osteogenesis imperfecta. J Clin Invest. 2014;124(2):491-8. doi:10.1172/JCI71101. PMID: 24463451
69. Hald JD, Weir C, Keerie C, et al. Baseline Characteristics of the TOPaZ Study: Randomised Trial of Teriparatide and Zoledronic Acid Compared with Standard Care in Adults with Osteogenesis Imperfecta. Calcif Tissue Int. 2025;116(1):136. doi:10.1007/s00223-025-01440-3
70. Hald JD, Weir CJ, Keerie C, et al.; TOPaZ Investigators. Teriparatide plus zoledronic acid for osteogenesis imperfecta: a randomized clinical trial. JAMA. 2026;336(2):116-124. doi:10.1001/jama.2026.6889
71. Glorieux FH, Langdahl B, Chapurlat R, et al. Setrusumab for the treatment of osteogenesis imperfecta: 12-month results from the phase 2b ASTEROID study. J Bone Miner Res. 2024;39(9):1215-1228. doi:10.1093/jbmr/zjae112
72. Mereo BioPharma Group plc. Mereo BioPharma reports full year 2025 financial results and provides corporate highlights [press release]. 2026. Available at: https://www.biospace.com/press-releases/mereo-biopharma-reports-full-year-2025-financial-results-and-provides-corporate-highlights
73. Mereo BioPharma Group plc. Corporate presentation, J.P. Morgan Healthcare Conference 2026: setrusumab Phase 3 Orbit and Cosmic results [sponsor presentation; exhibit 99.2 to Form 8-K]. 2026 Jan 12. SEC EDGAR, accession 0001193125-26-009829. Available at: https://www.sec.gov/Archives/edgar/data/1719714/000119312526009829/mreo-ex99_2.htm
74. ClinicalTrials.gov. NCT05125809 (ORBIT, active not recruiting, 183 enrolled in phases 2 and 3); NCT05768854 (COSMIC, 69); NCT05972551 (romosozumab, phase 3, 111, estimated primary completion May 2027); NCT07366086 (romosozumab safety follow-up, start 18/03/2026); NCT07062588 (AGA2115, IDUN, recruiting, 80 planned); NCT03706482 (BOOSTB4, 18 enrolled); NCT05231668 (SAR439459, terminated, 16 enrolled); NCT03064074 (fresolimumab, phase 1, completed) [Internet]. Bethesda (MD): National Library of Medicine [accessed via API v2 on 2026 Oct 6]. Available at: https://clinicaltrials.gov/
75. Bartosik M, Windels O, von Brackel FN, Amling M, Oheim R. Romosozumab in postmenopausal women with classical osteogenesis imperfecta. Arch Osteoporos. 2026;21(1):100. doi:10.1007/s11657-026-01744-3. PMID: 42470570
76. European Medicines Agency. Medicines output — medicines report: Zometa (MA 20/03/2001), Aclasta (15/04/2005), Forsteo (10/06/2003), Prolia (26/05/2010) and Evenity (09/12/2019), with current indications [Internet]. Amsterdam: EMA [accessed 2026 Oct 6]. Available at: https://www.ema.europa.eu/en/documents/report/medicines-output-medicines_json-report_en.json
77. Angitia Biopharmaceuticals. Angitia Biopharmaceuticals Announces Dosing of First Participant in Phase 2 IDUN Trial of AGA2115 for the Treatment of Osteogenesis Imperfecta [press release]. 12 Jan 2026. Available at: https://www.globenewswire.com/news-release/2026/01/12/3216764/0/en/Angitia-Biopharmaceuticals-Announces-Dosing-of-First-Participant-in-Phase-2-IDUN-Trial-of-AGA2115-for-the-Treatment-of-Osteogenesis-Imperfecta.html
78. Song IW, Nagamani SC, Nguyen D, Grafe I, Sutton VR, Gannon FH, et al. Targeting TGF-β for treatment of osteogenesis imperfecta. J Clin Invest. 2022;132(7):e152571. doi:10.1172/JCI152571. PMID: 35113812; PMCID: PMC8970679
79. Sagar RL, Åström E, Chitty LS, et al. An exploratory open-label multicentre phase I/II trial evaluating the safety and efficacy of postnatal or prenatal and postnatal administration of allogeneic expanded fetal mesenchymal stem cells for the treatment of severe osteogenesis imperfecta in infants and fetuses: the BOOSTB4 trial protocol. BMJ Open. 2024;14:e079767. Available at: https://pmc.ncbi.nlm.nih.gov/articles/PMC11163617/
80. Madhuri V, Ramesh S, Goos A, Paul TV, Nidugala Kesava S, Mathews V, et al. Evaluation of safety and efficacy of multiple intravenous and intraosseous doses of foetal liver-derived mesenchymal stem cells in children with severe osteogenesis imperfecta: the BOOST2B clinical trial protocol. Bone Jt Open. 2025;6(3):361-72. doi:10.1302/2633-1462.63.BJO-2024-0115.R1
81. European Medicines Agency. Orphan designations report: EU/3/21/2534 (allogeneic fetal mesenchymal stem cells, 10/12/2021), EU/3/22/2622 (losartan, 21/06/2022), EU/3/25/3069 (bispecific anti-sclerostin and anti-DKK1 antibody, 20/06/2025) and EU/3/26/3272 (cisatracurium, 28/07/2026), all for the treatment of osteogenesis imperfecta [Internet]. Amsterdam: EMA [accessed 2026 Oct 6]. Available at: https://www.ema.europa.eu/en/documents/report/medicines-output-orphan_designations-json-report_en.json
82. Chaugule S, Constantinou CK, John AA, Micha D, Eekhoff M, Gravallese E, Gao G, Shim JH. Comprehensive Review of Osteogenesis Imperfecta: Current Treatments and Future Innovations. Hum Gene Ther. 2025. doi:10.1089/hum.2024.191
83. Agence nationale de sécurité du médicament et des produits de santé. ADVILMED ENFANTS ET NOURRISSONS 20 mg/1 ml, suspension buvable (ibuprofène): résumé des caractéristiques du produit, rubrique 4.2 [updated 2025 Jun 19; accessed 2026 Nov 6]. Base de données publique des médicaments. Available at: https://base-donnees-publique.medicaments.gouv.fr/medicament/61902232/extrait
84. Institute of Medicine (US) Committee to Review Dietary Reference Intakes for Vitamin D and Calcium. Dietary Reference Intakes for Calcium and Vitamin D. Washington, DC: National Academies Press; 2011. doi:10.17226/13050
85. U.S. Food and Drug Administration. Drugs@FDA (openFDA drugsfda): original approval dates of Aredia (NDA 020036, 31/10/1991), Zometa (NDA 021223, 20/08/2001), Reclast (NDA 021817, 16/04/2007), Actonel (NDA 020835, 27/03/1998), Fosamax (NDA 020560, 29/09/1995), Prolia (BLA 125320, 01/06/2010), Forteo (NDA 021318, 26/11/2002) and Evenity (BLA 761062, 09/04/2019) [accessed 2026 Oct 6]. Available at: https://api.fda.gov/drug/drugsfda.json
86. European Medicines Agency. Forsteo (teriparatide): summary of product characteristics, section 4.2 [Internet]. Amsterdam: EMA [accessed 2026 Oct 6]. Available at: https://www.ema.europa.eu/en/documents/product-information/forsteo-epar-product-information_en.pdf
87. Agence nationale de sécurité du médicament et des produits de santé. CONTRAMAL 100 mg/ml, solution buvable (chlorhydrate de tramadol): résumé des caractéristiques du produit, rubriques 4.2–4.4 [updated 2025 Oct 30; accessed 2026 Nov 6]. Base de données publique des médicaments. Available at: https://base-donnees-publique.medicaments.gouv.fr/medicament/60477650/extrait
88. U.S. Food and Drug Administration. Tramadol hydrochloride tablets: prescribing information, section 4 (Contraindications) [label dated 2026 Apr 10; set id 007bf37f-0e46-426a-ac8c-be63d4b7414c; accessed 2026 Nov 6]. DailyMed. Available at: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=007bf37f-0e46-426a-ac8c-be63d4b7414c
89. Cubert R, Cheng EY, Mack S, Pepin MG, Byers PH. Osteogenesis imperfecta: mode of delivery and neonatal outcome. Obstet Gynecol. 2001;97(1):66-69. doi:10.1016/s0029-7844(00)01100-5. PMID: 11152910
90. Agence nationale de sécurité du médicament et des produits de santé. DOLIPRANE 2,4 POUR CENT, suspension buvable (paracétamol): résumé des caractéristiques du produit, rubriques 4.2 and 4.4 [updated 2026 Oct 5; accessed 2026 Nov 6]. Base de données publique des médicaments. Available at: https://base-donnees-publique.medicaments.gouv.fr/medicament/66477582/extrait
91. Agence nationale de sécurité du médicament et des produits de santé. ORAMORPH 20 mg/1 ml, solution buvable en gouttes (sulfate de morphine): résumé des caractéristiques du produit, rubrique 4.2 [updated 2025 Oct 2; accessed 2026 Nov 6]. Base de données publique des médicaments. Available at: https://base-donnees-publique.medicaments.gouv.fr/medicament/66089989/extrait
92. Ferring Pharmaceuticals Inc. DDAVP (desmopressin acetate) injection. Prescribing information. NDA018938. Revision 2022-09-28. Available at: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=651f6fee-a2c7-431b-8d5d-58b156c72244
93. Agence nationale de sécurité du médicament et des produits de santé. MINIRIN 4 microgrammes/mL, solution injectable (desmopressine): résumé des caractéristiques du produit, rubriques 4.2 and 4.4 [updated 2026 May 21; accessed 2026 Nov 6]. Base de données publique des médicaments. Available at: https://base-donnees-publique.medicaments.gouv.fr/medicament/67576220/extrait
94. Agence nationale de sécurité du médicament et des produits de santé. EXACYL 0,5 g/5 ml I.V., solution injectable, and EXACYL 1 g/10 ml, solution buvable (acide tranexamique): résumés des caractéristiques du produit, rubriques 4.2–4.3 [updated 2026 May 22 and 2026 Apr 10; accessed 2026 Nov 6]. Base de données publique des médicaments. Available at: https://base-donnees-publique.medicaments.gouv.fr/medicament/62330898/extrait
95. Folkestad L, Hald JD, Canudas-Romo V, et al. Mortality and causes of death in patients with osteogenesis imperfecta: a register-based nationwide cohort study. J Bone Miner Res. 2016;31(12):2159-2166. doi:10.1002/jbmr.2895
96. Osteogenesis Imperfecta Foundation (USA). Available at: https://oif.org
97. Brittle Bone Society (United Kingdom). Available at: https://brittlebone.org
98. Associação Brasileira de Osteogênese Imperfeita (ABOI). Portuguese-language support resource for patients and families.