Osteogenesis Imperfecta: A Review with Clinical Examples

Osteogenesis imperfecta (OI) comprises a heterogeneous group of diseases characterized by susceptibility to bone fractures with variable severity and, in most cases, with presumed or proven defects in collagen type I biosynthesis [Van Dijk et al., 2010c]. Other clinical manifestations include short stature, blue sclerae, dentinogenesis imperfecta, and hearing loss.

OI has a birth prevalence of approximately 6–7/100,000 [Steiner et al., 1993]. The prevalence and incidence of the OI types are different from each other, with OI type I and OI type IV accounting for considerably more than half of all OI cases [Steiner et al., 1993]. In 1979, Sillence et al. [1979] reported a prevalence of 3–4/100,000 and an incidence of 3.5/100,000 for OI type I in Victoria, Australia. For OI type II the incidence is about 1–2/100,000 [Steiner et al., 1993], prevalence data are not available due to early lethality [Steiner et al., 1993]. OI type III has a prevalence of 1–2/100,000 [Steiner et al., 1993]. An incidence of 1.6/100,000 has been reported in Australia [Sillence et al., 1979]. OI type IV was believed to be a rare entity [Sillence et al., 1979], but this proved not to be the case [Steiner et al., 1993].

The earliest known patient with OI probably dates from about 1000 BC and appears to be an Egyptian infant. This conclusion can be drawn after studying the remains of an Egyptian mummy [Lowenstein, 2009].

The first scientific description of OI was given by the Swedish army surgeon Olaus Jakob Ekman (1788) who, in his thesis on ‘congenital osteomalacia’, described a 3-generation family with hereditary bone fragility [Peltier, 1981; Baljet, 2002]. Since then, many names have been used to describe familial bone fragility. Willem Vrolik (1801–1863), a Dutch professor at the Athenaeum Illustre of Amsterdam, introduced the term ‘osteogenesis imperfecta’ [Baljet, 1984] and was one of the first to realize that OI might be due to insufficient intrinsic ‘generative energy’ [Baljet, 2002] as opposed to the result of an acquired disease.

In the 20th century, it became clear that OI was a disease showing remarkable clinical variability with severity ranging from death in the perinatal period to subtle increase in fracture frequency. Looser [1906] made the first classification of OI into congenital and tarda. Attempts were made to further classify OI, and in 1979 the ‘Sillence classification’ [Sillence et al., 1979] was proposed which, though in an adjusted form, is still in use today.

In 1974, bone collagen aggregation abnormalities were described by scanning electron microscopy of bone collagen in 3 patients with OI [Teitelbaum et al., 1974]. Sykes et al. [1977] reported altered relation of 2 collagen types in pepsin digests of skin of OI patients using a method of interrupted polyacrylamide-gel electrophoresis. In 1983, the presence of an internal deletion of about 0.5 kb in 1 allele for the pro-α1(I) chain in a patient with OI [Chu et al., 1983] was discovered. This finding was the beginning of the unraveling of the collagen type I biosynthesis and the pathogenic mechanisms of OI which are still not completely known today.

In 1979, Sillence et al. [1979] proposed a numerical classification of OI into 4 types based on clinical and genetic findings in OI patients ascertained in Victoria, Australia. This classification distinguished type I (mild OI, blue sclerae, autosomal dominant inheritance), type II (lethal perinatal OI, autosomal recessive inheritance, later subdivided in II-A, -B, and -C based on radiographic features [Sillence et al., 1984], type III (progressively deforming, autosomal recessive inheritance), and type IV (dominantly inherited OI with normal sclerae). It is evident that Sillence et al. [1979] assumed that OI was a heterogeneous condition.

OI was considered for some time to be an autosomal dominant disease with recurrence due to germ line mosaicism since heterozygous collagen type I mutations (COL1A1 or COL1A2 mutations) were discovered in all OI types. The recurrence risk of OI type II was observed to be less than 10% [Byers et al., 1988; Cohn et al., 1990; Cohen-Solal et al., 1991; Pepin et al., 1997]. The Sillence classification was used only for the clinical/radiological classification of OI. Thereafter, studies were reported of families, in some cases consanguineous, with OI not caused by pathogenic variants in the COL1A1 or COL1A2 genes [Wallis et al., 1993]. The original Sillence classification was extended with OI types V–VII based on OI cases with unknown genetic etiology and/or distinctive clinical manifestations [Rauch et al., 2004].

In 2006, the first genetic cause of autosomal recessive lethal OI was discovered, i.e. bi-allelic variants in the CRTAP gene causing complete loss of protein function [Barnes et al., 2006]. Partial loss of function CRTAP variants encoding cartilage-associated protein (CRTAP) were found to cause OI type VII [Morello et al., 2006]. Presently, 6 more causes of recessive OI (causative variants in LEPRE1 [Cabral et al., 2007], PPIB [Van Dijk et al., 2009b; Barnes et al., 2010], SERPINH1[Christiansen et al., 2010], FKBP10 [Alanay et al., 2010], SP7[Lapunzina et al., 2010], and SERPINF1 [Becker et al., 2011]) have been described, all but 2 (SP7 and SERPINF1) concerning genes encoding proteins involved in collagen type I biosynthesis.

There is some debate in the literature about how to involve this newly discovered heterogeneity of OI in the classification. One research group proposes to have OI caused by recessive causative variants in PPIB, SERPINH1, FKBP10, SP7, and SERPINF1 added to the current classification as OI type IX [Barnes et al., 2010] and presumably X, XI, XII, and XIII, respectively. Another viewpoint expressed by our research group is that the clinical/radiological characteristics of recessive OI do not substantially differ from OI type II-B, III or IV caused by dominant OI, so we propose to have recessive OI classified as OI type II-B/III/IV-PPIB/SERPINH1/FKBP10/ SP7/SERPINF1- related. OI type I and OI type II-A appear to be exclusively caused by causative COL1A1/2 variants [Van Dijk et al., 2009a]. We also proposed to only add a new numerical type when the phenotype of the patients differs from the numerical types described so far [Van Dijk et al., 2010c]. There is also debate as to whether Bruck syndrome type 1 [Breslau-Siderius et al., 1998] and type 2 [van der Slot et al., 2003], characterized clinically by bone fragility and congenital contractures of the large joints, should be classified as subtypes of OI.

The International Nomenclature group for Constitutional disorders of the Skeleton, among others concerning the classification of OI, proposed to retain the Sillence classification into 5 types and free the classification from direct molecular reference. Bruck syndrome 1 and 2 were not listed as subtypes of OI [Warman et al., 2011].

Collagen type I biosynthesis consists of various important steps, which are explained in this section and depicted in figure 1.

Approximately 90% of individuals affected with OI are heterozygous for a causative variant in 1 of the 2 genes, COL1A1 or COL1A2[Sykes et al., 1990; Körkkö et al., 1998], which encode the pro-α1(I) and pro-α2(I) chains of type I procollagen, respectively.

Procollagen type I is cotranslationally translocated into the lumen of the endoplasmic reticulum [Canty and Kadler, 2005].

Procollagen type I contains C- and N-terminal propeptides and a large ‘triple helix’ domain comprising predominantly Gly-X-Y triplets. In the rough endoplasmic reticulum, 2 α1(I)-collagen chains encoded by COL1A1 and 1 α2(I)-collagen chain encoded by COL1A2 align. Interactions between the C propeptides are largely stabilized by interchain disulphide bounds to ensure correct alignment [Prockop et al., 1989]. Protein disulphide isomerase has also been implicated in the formation of inter-chain disulphide bonds [Canty and Kadler, 2005]. The 2 pro-α1 chains and 1 pro-α2 chain then assemble in the C- to N-direction to form a triple helix. During folding, collagen is modified by, among others, specific enzymes that hydroxylate lysine and proline residues and glycosylate hydroxylysyl residues. This post-translational modification stops when triple helix assembly is complete [Engel et al., 1991]. The CRTAP/P3H1/CyPB complex, encoded by the CRTAP, LEPRE1 and PPIB genes, is responsible for the 3-hydroxylation of P986 (p.P1164 counting from the methionine which initiates translation), but the complex most likely also acts as a proline cis-trans isomerase and a molecular chaperone [Ishikawa et al., 2009]. FKBP65 encoded by FKBP10 also acts as a molecular chaperone for type I (pro)collagen [Alanay et al., 2010]. The protein product of PLOD2 hydroxylates telopeptide lysines in the rough endoplasmic reticulum [van der Slot et al., 2003]. HSP47 encoded by SERPINH1 is thought to maintain the stability of the triple helix [Christiansen et al., 2010].

After folding, the procollagen molecules are transported through the Golgi apparatus and plasma membrane into the extracellular matrix.

In the extracellular matrix, cleavage of the N- and C-terminal propeptides occurs, and collagen molecules aggregate to form fibrils. Covalent cross-links occur within and between triple-helical collagen molecules in fibrils. These fibrils converge into collagen type I fibers.

At present, more than 1,000 distinct variants in the COL1A1 and COL1A2 genes have been identified that give rise to OI (https://oi.gene.le.ac.uk) [Dalgleish, 1997, 1998]. Types of mutation as well as position appear to influence the phenotype.

OI type I is mostly characterized by a 50% reduction of the amount of collagen type I (quantitative or haploinsufficiency effect) usually resulting from variants in 1 COL1A1 allele (frameshift, nonsense, and splice-site alterations) that lead to mRNA instability and haploinsufficiency [Marini et al., 2007b] and sometimes from a deletion of the complete COL1A1 allele [van Dijk et al., 2010a] or from substitutions for glycine by small amino acids (cysteine, alanine and serine) near the amino terminal ends of the triple helical domains in either one COL1A1 or COL1A2 allele.

OI types II–IV are characterized by intertwining of mutated and normal collagen type I chains resulting in production of abnormal collagen type I (dominant-negative effect). This occurs usually due to causative variants in either COL1A1 or COL1A2 that result in substitutions for glycine. Less common causative variants include splice-site alterations, insertion/deletion/duplication events that lead to in-frame sequence alterations and variants in the carboxyl-terminal propeptide coding domains. Most of these variants result in synthesis of an abnormal type I procollagen molecule which has to intertwine with normal pro-α chain(s) [Marini et al., 2007b]. This assembly leads to disturbed helical folding resulting in overprocessing by the enzymes responsible for post-translational modification of (pro)collagen type I. Post-translational overmodification of the triple-helical domain results in alterations visible by SDS-polyacrylamide gel electrophoresis.

It is estimated that 10% of OI cases will be caused by recessive causative variants in 1 of the currently known genes (CRTAP, LEPRE1, PPIB, SERPINH1, FKBP10, PLOD2, SP7, SERPINF1) or in genes that have yet to be discovered.

CRTAP forms a complex with prolyl-3-hydroxylase-1 (P3H1) and Cyclophilin B (CyPB). One known function of the complex is the 3-hydroxylation of a single proline residue at position 986 (P986) in the pro-α1(I)-collagen chain [Marini et al., 2007a]. It is highly likely that the complex also acts as a cis-trans isomerase and a molecular chaperone [Ishikawa et al., 2009] initiating chain recognition and helical folding [Pyott et al., 2011b]. Most CRTAP variants are reported to cause autosomal recessive lethal/severe OI [Barnes et al., 2006] types II-B and III, but CRTAP variants do not seem to cause OI type II-A [Van Dijk et al., 2009a]. Most causative variants result in null alleles with absence or severe reduction of gene transcripts and proteins [Marini et al., 2010b]. Biochemically, decreased prolyl 3-hydroxylation of P986 is evident as well as post-translational overmodification on (pro)collagen gel electrophoresis [Barnes et al., 2006].

LEPRE1 encodes P3H1 and pathogenic variants in LEPRE1 cause autosomal recessive lethal/severe OI [Cabral et al., 2007; Baldridge et al., 2008], types II-B/ III [Van Dijk et al., 2010b]. In most patients, the causative variants result in null alleles with absence or severe reduction of gene transcripts and proteins [Cabral et al., 2007; Willaert et al., 2009; Marini et al., 2010a]. As in CRTAP-related OI, decreased prolyl 3-hydroxylation of P986 in the pro-α1(I)-collagen chains as well as post-translational overmodification are observed [Cabral et al., 2007].

PPIB encodes the protein cyclophilin B, a collagen-specific proline cis-trans isomerase. CyPB is bound in a complex with P3H1 and CRTAP. Recessive variants in PPIB were described to cause OI with decreased P986 3-hydroxylation and post-translational overmodification of type I (pro)collagen in 2 families [Van Dijk et al., 2009b]. These findings were confirmed in other patients with OI types II, III and IV due to PPIB mutations [Pyott et al., 2011a]. Interestingly, in another family, a homozygous PPIB variant in the presumed start codon of PPIB (c.2T>G in the sequence of PPIB described by Price et al. [1991], c.26T>G in the current reference sequence of PPIB (NM_000942.4))appeared to cause only a moderately deforming type of OI without decreased P986 3-hydroxylation and detected post-translational overmodification but with absence of CyPB on Western blot [Barnes et al., 2010]. At this point, it would be correct to state that recessive variants in the PPIB gene can cause OI type II-B, III and IV with, in the majority of cases, decreased P986 3-hydroxylation in the pro-α1(I)-collagen chains and post-translational overmodification, as is seen in CRTAP and LEPRE1-related OI.

First detected in Dachshund pedigrees with OI [Drögemüller et al., 2009], bi-allelic causative missense variants in the SERPINH1 gene encoding collagen chaperone protein HSP47 appear to result in severe recessive OI. HSP47 monitors the integrity of the triple helix of type I procollagen at the endoplasmic reticulum/cis-Golgi boundary. When absent, the rate of transit of procollagen type I chains from endoplasmic reticulum to Golgi appears to be increased and the triple helical structure is compromised. Since the action of HSP47 appears not to be required for procollagen type I chain modification, no post-translational overmodification is visible on (pro)-collagen gel electrophoresis [Christiansen et al., 2010].

Four individuals (2 related and 2 unrelated individuals) with OI type III, and without collagen type I overmodification on electrophoresis, were recently described to harbor bi-allelic causative variants in SERPINF1 leading to a loss of pigment-epithelium-derived factor (PEDF). The causative SERPINF1 variants in the index patient were found by next generation sequencing. PEDF is known mainly for its strong inhibition of angiogenesis. However, expression analyses in bone tissue from wild-type mice and in vitro experiments with murine cell systems support a role for PEDF in bone formation and remodeling. It is speculated that a loss of PEDF causes OI independent of collagen type I biosynthesis [Becker et al., 2011].

Recently, a homozygous causative variant in SP7 was described in a patient with a moderately severe recessive form of OI with recurrent fractures, mild bone deformities, delayed tooth eruption, normal hearing, and white sclerae. SP7 does not seem to encode a protein involved in the collagen type I biosynthesis pathway. However, it should be noted that in vivo studies to assess the effect on collagen type I production could not be performed [Lapunzina et al., 2010]. SP7 encodes Osterix, an osteoblast-specific transcription factor that has been shown to be essential for bone formation in mice [Sun et al., 2010; Zhou et al., 2010].

PLOD2 and FKBP10

In 1998, a family was reported with bone fragility, congenital contractures of the large joints and aberrant cross-linking of bone collagen [Breslau-Siderius et al., 1998]. The affected individuals were diagnosed as having Bruck syndrome which is characterized by bone fragility in combination with congenital joint contractures. A locus responsible for Bruck syndrome in this family was mapped to 17p12 [Bank et al., 1999]. In other families with similar clinical and biochemical features, recessive variants in PLOD2, encoding a bone-specific telopeptidase lysyl hydroxylase-2, appeared to be causative [van der Slot et al., 2003; Ha-Vinh et al., 2004]. All 3 causative variants identified in PLOD2 are homozygote missense mutations affecting highly conserved residues in exon 17 of PLOD2. These variants possibly alterate folding and isomerization of the protein, leading to severe reduction of the enzymatic activity [Hyry et al., 2009], which results in aberrant cross-linking of bone collagen due to underhydroxylation of lysine residues in the telopeptides.

However, in the family described by Breslau-Siderius et al. [1998], no recessive variants in PLOD2 were found, leading to a classification of Bruck syndrome in type I (with unknown genetic cause) and II (with recessive variants in PLOD2) [van der Slot et al., 2003]. Recently, however, causative variants in FKBP10 (locus 17q12) encoding FKBP65, were described to cause (a) recessive OI without post-translational overmodication of (pro)collagen type I most closely resembling OI type III [Alanay et al., 2010] or (b) Bruck syndrome [Shaheen et al., 2010; Kelley et al., 2011]. The genetic cause of Bruck syndrome type I in the originally described family has not been reported yet.

Although causative variants in many genes are now known to result in OI, the pathogenic mechanism leading from causative variant(s) to OI is still largely unknown, which emphasizes the need to study normal bone formation.

Bone consists of cells and mineralized extracellular matrix, it provides support and protection, is an insertion site for muscles and serves as a storage site for calcium and phosphate. Four cell types are active in bone tissue: (1) osteoprogenitor cells (resting cells that can transform into osteoblasts, chondrocytes and adipocytes), (2) osteoblasts, (3) osteocytes (mature osteoblasts), and (4) osteoclasts. The bone matrix (osteoid) is secreted by osteoblasts and consists of type I collagen and ground substance containing proteoglycans and non-collageneous glycoproteins. Resorption of the bone matrix is performed by macrophage-related osteoclasts and is important for adaption to growth, repair and mineral mobilization [Ross et al., 1995].

Bone consists of an outer layer of mature compact bone (cortex) largely composed of cylindrical units called osteons or Haversian systems. The inner layer of adult spongy bone is arranged as trabeculae or spicules (fig. 2A). Bone is formed by endochondral (long bones, ribs and vertebrae) (fig. 2B) or intramembranous (flat bones of skull and face, mandible, clavicle, ileum) ossification. Intramembranous ossification takes place by direct differentiation of mesenchymal cells into osteoblasts that secrete osteoid. Osteoblasts retreat or become entrapped as osteocytes in the osteoid. The osteoid calcifies to form spicules of spongy bone, the spicules unite and form trabeculae. Endochondral bone formation is characterized by the presence of a cartilaginous model in which chondrocytes differentiate [Ross et al., 1995] (fig. 2B).

Paracrine factors and transcription factors appear to be active in the transition of cartilage to bone. For example, SOX9 is important for differentiation of mesenchymal cells into chondroblast (mutations in SOX9 cause campomelic dysplasia [Unger et al., 1993]), whereas RUNX2 (mutations in RUNX2 cause cleidocranial dysplasia [Mundlos et al., 1997]) is critical for differentiation of prehypertrophic chondrocytes into preosteoblasts and activation of Osterix (mutations in SP7, encoding Osterix, cause OI [Lapunzina et al., 2010]), which activates bone-specific proteins [Gilbert, 2010].

Mineralized cartilage is replaced by bone through remodeling. During life, bone is remodeled through the action of osteoblasts and osteoclasts, replacing old bone with new bone. Normally, remodeling is coupled such that the level of resorption is equal to the level of formation and bone density remains constant. The continued growth of the long bones is dependent on the presence of epiphyseal cartilage (epiphyseal plate) between the primary and secondary ossification centers. This continues to form new cartilage which is replaced by bone, resulting in increased length of the bone. Growth continues until the cartilage in the plate is replaced by bone.

A study of iliac bone specimens of 70 children with OI type I, III and IV compared to 27 age-matched controls reported that in OI bone thickness is reduced because of delayed periosteal bone formation. Trabeculae are reduced in number and abnormally thin. The overall bone formation is increased but counteracted by increased activity of bone resorption [Rauch and Glorieux, 2004]. Figure 2C compares bone tissue of an individual affected with OI to a control.

The severity of clinical features of OI at birth ranges from no clinical features to prenatally lethal skeletal abnormalities. The clinical variability of OI led to a classification of OI originally in 4 types [Sillence et al., 1979]. In 2004, OI type V, VI and VII were added to this classification. Important for the prenatal as well as postnatal diagnosis of OI is that a continuum of severity is observed in OI with clinical overlap between OI types I and IV, II and III, III and IV. Tables 1 and 2 represent an overview of the clinical and radiological characteristics of OI type I–VI. In our opinion, the addition of OI type VII and the newly proposed OI types VII–XII are unnecessary [Van Dijk et al., 2010c]. Figures 3, 4, 5, 6, 7 are clinical pictures and radiographs of patients with OI type I, II, III, IV with case descriptions in the legend.

OI type II and III (fig. 4, 5, 6) can prenatally be diagnosed with ultrasonography since they are likely to have prenatal fractures that can be observed. Increased nuchal translucency can be the earliest (nonspecific) sonographic sign of OI type II [Viora et al., 2002]. Sonographic signs of OI type II can be detected as early as 14 weeks [Marini et al., 2007b] due to reduced echogenicity of the fetal bones, followed by multiple fractures at various stages of healing and deformity of the long bones, ribs and skull [Morgan and Marcus, 2010]. Sonographic details of OI type III are usually visible from 18 weeks [Marini et al., 2007b], whereas OI type IV may occasionally be detected after 20 weeks [Marini et al., 2007b] and may consist of bowing of the long bones with or without shortening and without evidence of fractures or osteopenia. Prenatal ultrasonographic diagnosis of OI type I is unreliable. For differential diagnosis, campomelic dysplasia and perinatal lethal hypophosphatasia can be considered [Morgan and Marcus, 2010]. Pathognomonic radiological/ultrasonographic features of campomelic dysplasia are cervical spine abnormalities, scapular hypoplasia, narrow iliac wings, bowing of the femora and the tibiae, and club feet [Unger et al., 1993]. Distinguishing features of perinatal lethal hypophosphatasia are (i) deficient ossification of the spine notably in the thoracic region with patchy ossification of ribs and vertebrae, and (ii) deep cupping of the metaphyses of the long bones [Zankl et al., 2008].

The ultrasonographic prenatal diagnosis of OI can be confirmed by laboratory investigations either by (a) performing a chorion villus biopsy with cultured chorion villi cells showing abnormal production of collagen type I, visible as post-translational overmodification on procollagen electrophoresis, or (b) by performing a chorion villus biopsy/amniocentesis to obtain fetal DNA for molecular analysis of genes involved in OI. Recently, guidelines have been established for the laboratory diagnosis of OI. Molecular analysis of the COL1A1/2 genes will be the first step, followed by DNA analysis of the other causative genes upon re-evaluation and confirmation of the diagnosis OI [van Dijk et al., 2011].

Prenatal or preimplantation genetic diagnosis with the intention of terminating a pregnancy or not selecting embryos carrying the causative variant(s) is possible in the case of identification of known disease-causing variants [van Dijk et al., 2011].

OI types I–V are clinically diagnosed peri- (types II, III and sometimes IV) and postnatally (all types). A flow diagram for the postnatal diagnosis of OI is presented in figure 8. Clinical case examples are presented in figures 3, 6 and 7. For differential diagnosis, rare genetic conditions such as Bruck syndrome (MIM 259450, 609220), Osteoporosis pseudoglioma syndrome (MIM 259770), Cole-Carpenter syndrome (MIM 112240), Hajdu-Cheney (MIM 102500), and gerodermia osteodysplastica (MIM 231070) can be considered. Furthermore, idiopathic juvenile osteoporosis (MIM 259750) and isolated dentinogenesis imperfecta (MIM 125490) can be important differential diagnostic considerations. The most common to be considered is nonaccidental injury (NAI), frequently in cases of suspected OI type I or IV [Bilo et al., 2010]. Fractures resulting from NAI occur in 24/10,000 children under 3 years of age, whereas the OI prevalence is 1/10,000–20,000 [Marlowe et al., 2002]. In the study by Marlowe et al. [2002], the reported incidence of OI among children evaluated for NAI was 2–5%. Differentiation between OI and NAI is aided by an experienced clinician familiar with OI [Ablin et al., 1990]. A positive family history, the presence of blue sclerae, osteoporosis, multiple wormian bones in the skull, and platyspondyly point to OI [Chapman and Hall, 1997]. The laboratory confirmation of OI is made preferably by DNA analysis of the genes involved in OI or by decreased or abnormal production of (pro)collagen type I by fibroblasts measured on procollagen electrophoresis.

In a large majority of patients, OI type I is caused by dominant (de novo or recurrent) causative variants in the COL1A1/2 genes.

OI types II–IV can be autosomal dominantly and autosomal recessively inherited. In case of autosomal dominant inheritance, the causative variant is either de novo or, with clinically unaffected parents, recurrent due to germ line mosaicism in a parent. The empirical recurrence risk is a mixture of recurrence risk due to gonadal mosaicism and autosomal recessive inheritance.

Pepin et al. [1997] reported a 2% empirical recurrence risk of lethal OI for families with one previous affected child. In a recent study conducted by Pyott et al. [2011b], a recurrence rate of 1.3% was observed for lethal OI (based on 1 recurrence in 76 families with 1 previous affected child). Interestingly, it was reported that approximately 16% of parents with 1 affected child due to a causative variant in COL1A1/2 was mosaic in somatic cells [Pyott et al., 2011b], resulting in a higher risk of recurrence of OI in these families.

Identifying the causative variant(s) and DNA analysis of the parents in case of a causative variant in COL1A1/2 is necessary for an accurate estimate of the recurrence risk, which is important for genetic counseling in case of early prenatal diagnosis and preimplantation genetic diagnosis [Pyott et al., 2011b].

When the diagnosis OI has been established, the affected individual should preferentially be evaluated by a multidisciplinary team [Steiner et al., 1993]. Important members of the team would be orthopedic surgeons, rehabilitation physicians, endocrinologists, physical therapists, and pediatricians. Referral to other disciplines can take place upon individual needs and for routine surveillance such as dental controls. Management consists of pharmacological treatment, orthopedic treatment, physical medicine, dental treatment, treatment for hearing loss, and prevention of primary (e.g. basilar impression) and secondary (e.g. problems due to general medical disciplines) complications [Steiner et al., 1993].

Oral and intravenous bisphosphonates are commonly prescribed for all OI types, adults and children since the first publication of the effect of bisphosphonate treatment in children with severe OI [Glorieux et al., 1998]. The main rationale for bisphosphonate therapy is based on several (not placebo-controlled) clinical trials that showed improvements of bone mineral density in individuals with OI. Nitrogenous bisphosphonates disrupt osteoclast formation, survival and cytoskeletal dynamics, and non-nitrogenous bisphosphonates initiate osteoclast apoptosis. A recently published systematic review of bisphosphonate treatment in OI [Philippi et al., 2009] concluded that in a relatively small group of patients, there is significant improvement in bone mineral density in individuals affected with OI and treated with oral or intravenous bisphosphonates. However, the most important question arising is whether increase in bone mineral density leads to fracture reduction and functional improvement; this has not been answered yet and warrants further investigations [Philippi et al., 2009]. The use of growth hormone to affect short stature in types III and IV OI [Marini et al., 2003] is still under active investigation [Marini et al., 2010a].

In case of decreased bone mineralization, high fracture frequency and/or bone deformities, intramedullary rods will be placed in the majority of patients with OI types III and IV and sometimes in OI type I [Monti et al., 2010]. These rods are inserted in the bone marrow canal in the center of the long bones and are used to align and stabilize fractures (fig. 6Ba and 7C, D). Severe scoliosis occurs most often in patients with OI type III (fig. 6Be) and sometimes IV and appears not to be related to the number of vertebral compression fractures. Since severe scoliosis can lead to pulmonary insufficiency, corrective surgery is often performed when the curvature is less than 60° [Marini et al., 2010a]. In case of anesthesia, precautions should be undertaken during intubation because of possible cervical fragility, and the patient should be carefully monitored during surgery because of the (possibly weak) association with hyperthermia during anesthesia [Monti et al., 2010; Oakley and Reece, 2010]. Nonsurgical management consists of bracing and splinting interventions [Monti et al., 2010].

An intensive rehabilitation program is necessary especially in OI types III and IV [Monti et al., 2010] with early intervention such as correct positioning of the child and proper head support, muscle strengthening (isotonic) and aerobic conditioning [Marini et al., 2010a].

In patients with dentinogenesis imperfecta, fractures and excessive wear of fragile teeth often occurs (fig. 9). This can be treated by capping teeth with hard polymers in order to prevent infections and facial deformities due to the loss of (parts of) teeth and/or malocclusion [Monti et al., 2010].

Hearing loss often occurs in adults with OI. Initially it concerns conductive hearing loss, but as the hearing loss progresses, a significant sensorineural component emerges. Surveillance for hearing loss is advised after adolescence every 3–5 years [Steiner et al., 1993]. Initially, hearing aids will be sufficient. As the hearing loss progresses, stapedectomy can be considered for which successful outcomes have been reported; however, long-term hearing restoration may be unsatisfactory due to fragility of the ossicular middle ear structures. Cochlear implantation has been reported because of the sensorineural hearing loss, but data are too limited to draw conclusions on effectiveness [Marini et al., 2010a].

Basilar invagination is a rare complication occurring in adults with OI type III when the top of the C2 vertebra migrates upward which may lead to (partial) closure of the foramen magnum with hydrocephalus, pressure on the brain stem, syrinx formation, and hindbrain herniation, requiring ventricular shunt placement or surgery. Only prolonged orthotic immobilization has been proven to stabilize symptoms and arrest progression [Monti et al., 2010].

Pregnant women with OI who have significant skeletal deformity and short stature should be monitored in high-risk prenatal care centers, not only for maternal reasons but also because of the risk that the fetus is affected with OI as well [Steiner et al., 1993]. In a large retrospective study of 167 pregnancies in which a child was affected with OI, there was an unusually high rate of breech presentation at term (37%). Cesarean delivery did not decrease fracture rates at birth in infants with OI types I, III and IV nor did it prolong survival in OI type II. Prenatal diagnosis did not influence the mode of delivery in most instances [Cubert et al., 2001].

Ninety percent of OI patients have a causative variant in the COL1A1/2 genes encoding either the (pro)-α1 chain or the (pro)-α2 chain of collagen type I. OI types II–IV result from intertwining of mutated and normal collagen type I chains resulting in production of abnormal collagen type I (dominant-negative effect), whereas OI type I is mostly due to decreased or nonexpression of 1 COL1A1 allele (haploinsufficiency effect). It was therefore thought that the future treatment of OI should consist of so-called antisense suppression therapy, directed towards decreasing or silencing of the allele containing the causative variant. This would transform a severe type of OI into a mild OI type. Various murine models (oim [Saban et al., 1996], brtl IV [Kozloff et al., 2004], crtap–/– [Morello et al., 2006], Mov-13 [Bonadio et al., 1990], OASIS–/– [Sekiya et al., 2010]) have been developed, which can be used for the purpose of investigating gene therapy in OI. However, the various antisense techniques (short oligonucleotides [Wang et al., 1996], ribozymes [Grassi et al., 1997], silencing RNA [Millington-Ward et al., 2004]) all lack true specificity against the mutant transcript. Other problems are stability of the antisense molecules apart from specific problems which each technique. As such, antisense suppression therapy is currently limited to in vitro studies [Monti et al., 2010].

Another approach is the replacement of cells harboring the causative variant with normal cells by bone marrow transplantation leading to engraftment of functional mesenchymal stem cells that differentiate into bone cells [Undale et al., 2009]. This approach may hold promises for OI treatment as is illustrated by various in vitro and in vivo studies (children with severe OI) [Horwitz et al., 1999, 2001, 2002; Panaroni et al., 2009]. Further studies are warranted to evaluate whether bone marrow transplantation could be a possible treatment for patients with OI.

Osteogenesis imperfecta is a complex hereditary disease with (i) a remarkable clinical variability warranting a logical classification system; (ii) causative recessive or dominant variants in 8 different genes with 6 of 8 genes encoding proteins involved in collagen type I biosynthesis; (iii) a need for multidisciplinary management and further investigations of therapeutic approaches such as bisphosphonates, growth hormone therapy and gene therapy.