Pharmacological and Biological Therapeutic Strategies for Osteogenesis Imperfecta

Abstract

Osteogenesis Imperfecta (OI) is a connective tissue disorder characterized by bone fragility, low bone mass and bone deformities. The majority of cases are caused by autosomal dominant pathogenic variants in the COL1A1 and COL1A2 genes that encode type I collagen, the major component of the bone matrix. The remaining cases are caused by autosomal recessively- or dominantly-inherited mutations in genes that are involved in the post-translational modification of type I collagen, act as type I collagen chaperones, or are members of the signaling pathways that regulate bone homeostasis. The main goals of treatment in OI are to decrease fracture incidence, relieve bone pain, and promote mobility and growth. This requires a multi-disciplinary approach, utilizing pharmacological interventions, physical therapy, orthopedic surgery, and monitoring nutrition with appropriate calcium and vitamin D supplementation. Bisphosphonate therapy, which has become the mainstay of treatment in OI, has proven beneficial in increasing bone mass, and to some extent reducing fracture risk. However, the response to treatment is not as robust as is seen in osteoporosis, and it seems less effective in certain types of OI, and in adult OI patients as compared to most pediatric cases. New pharmacological treatments are currently being developed, including anti-resorptive agents, anabolic treatment, and gene- and cell-therapy approaches. These therapies are under different stages of investigation from the bench-side, to pre-clinical and clinical trials. In this review we will summarize the recent findings regarding the pharmacological and biological strategies for the treatment of patients with OI.

Introduction

Osteogenesis imperfecta (OI) is a clinically and genetically heterogeneous disorder characterized by a prominent skeletal phenotype with brittle bones, low bone mass, growth deficiency, bone deformities and fractures [Forlino and Marini 2016; Shapiro and Sponsellor 2009; Van Dijk and Sillence 2014]. Additionally, OI patients may exhibit extra-skeletal manifestations, including dentinogenesis imperfecta (dental anomalies especially affecting primary dentition), hearing impairment (a mixture of conductive and sensorineural hearing loss), craniofacial abnormalities, and joint hypermobility. Rarer extra-skeletal symptoms include cardiovascular abnormalities (primarily aortic root dilation and valvular abnormalities), pulmonary disease, and muscle weakness [Forlino et al. 2011; Harrington et al. 2014; Van Dijk and Sillence 2014].

The prevalence of OI is 1 out of 15,000-20,000 births [Forlino and Marini 2016; Stoll et al. 1989] and the phenotypic spectrum ranges in severity from mild skeletal phenotypes with few to no fractures, to perinatal lethality. Originally, OI was classified into four types according to the clinical and radiological findings: OI type I patients have mild disease and non-deforming bone; OI type II causes perinatal lethality; OI type III is the most severe non-lethal form with progressive bone deformities and fractures; OI type IV patients have moderate disease severity [Sillence et al. 1979]. Since this original classification, at least 12 new molecularly defined types of OI have been described, and research to discover new genes is ongoing (Table 1). However, the original Sillence classification still holds most use from a clinical perspective in the context of patient care.

Table 1:

Classification of Osteogenesis Imperfecta types and main clinical features

OI type	Inheritance	Gene	Severity & unique clinical features	Biochemical abnormalities
OI types where classification is clinically useful for diagnosis and patient counseling
I	AD	COL1A1	Mild. Normal stature, little or no deformities	50% reduction in type I collagen synthesis
II	AD AR (rare)	COL1A1, COL1A2 AR genes below	Lethal. Minimal calvarial mineralization, beaded ribs, long bone deformities	Structural alterations of type I collagen chains leading to collagen overmodification
III	AD AR	COL1A1, COL1A2 AR genes below	Severe. Progressively deforming bones	Structural alterations of type I collagen chains leading to collagen overmodification
IV	AD	COL1A1, COL1A2	Moderate severity with short stature	Structural alterations of type I collagen chains leading to collagen overmodification
V	AD	IFITM5	Variable severity. Calcification of interosseous membrane of the forearm, hyperplastic callus formation	Dysregulated mineralization
VI	AR	SERPINF1	Moderate to severe. Accumulation of un-mineralized osteoid, biopsy shows fish-scale pattern of the lamellae	Defects in signaling molecules regulating bone homeostasis
OI types where classification is primarily useful for molecular distinction
VII	AR	CRTAP	Severe to lethal, rhizomelia	Collagen prolyl-3-hydroxylation complex defects
VIII	AR	P3H1	Severe to lethal, rhizomelia, coxa vara, popcorn metaphyses	Collagen prolyl-3-hydroxylation complex defects
IX	AR	PPIB	Severe. Short bowed femurs with anterior bowing of the tibiae	Collagen prolyl-3-hydroxylation complex defects
X	AR	SERPINH1	Severe	Collagen chaperone defects with abnormal collagen crosslinking
XI	AR	FKBP10	Moderate to severe. Joint contractures, biopsy shows distorted lamellar structure and a fish scale-like pattern	Collagen chaperone defects with abnormal collagen crosslinking
XII	AR	SP7	Moderate severity	Defect in osteoblast differentiation
XIII	AR	BMP1	Severe	Collagen C-propeptide cleavage defect
XIV	AR	TMEM38B	Moderate to severe	Calcium channel defect
XV	AR (AD causing osteoporosis)	WNT1	Moderate to severe. Also have brain malformations.	Defects in signaling molecules regulating bone homeostasis
XVI	AR	CREB3L1	Severe. Perinatal fractures, multiple fractured tubular bones with an accordion-like broadened appearance	Defects in ER-stress transducer regulating type I procollagen expression
Un-classified	AR	PLOD2	Moderate to severe. Joint contractures	Defects in collagen post-translational modification and cross-linking
Un-classified	X-linked	PLS3	Osteoporosis with fractures, clinical overlap with OI	Unknown mechanism
Un-classified	AR	SPARC	Moderate to severe. Early-onset scoliosis	Unknown mechanism

Abbreviations: AD: autosomal dominant; AR: autosomal recessive; OI: osteogenesis imperfecta.

Genetic causes of Osteogenesis Imperfecta

Type I collagen defects (classic autosomal dominant OI)

The majority of OI cases (~85-90%) are caused by autosomal dominant, heterozygous pathogenic variants in the genes encoding type I collagen, the most abundant protein in bone extracellular matrix [Forlino and Marini 2016; Marini et al. 2007; Van Dijk and Sillence 2014]. Type I collagen is a triple-helical molecule formed by two α1 chains (encoded by COL1A1) and one α2 chain (encoded by COL1A2). The helical domains exhibit Gly-X-Y repeats, where X and Y are typically proline or hydroxyproline [van der Rest and Garrone 1991]. These repeats are essential for proper collagen trimer assembly, with glycine, the smallest amino-acid, facing the inner part of the helix and the proline and hydroxyproline groups facing the outer part of the helix where they can interact with adjacent collagen molecules. After the triple helix is formed, type I collagen undergoes extensive post-translational modifications before secretion into the extracellular matrix as procollagen [Prockop and Kivirikko 1995]. The dominant mutations in type I collagen genes resulting in OI can be divided into two groups: mutations that result in a reduction in collagen quantity (i.e. quantitative mutations), and qualitative (dominant-negative) mutations, which result in structural abnormalities of the collagen triple helix.

Quantitative collagen defects in OI primarily result from mutations that introduce premature stop codons in COL1A1, with loss of the affected transcript by nonsense-mediated decay. Similarly, splice variants in COL1A1 may lead to the activation of cryptic splice sites, inclusion of introns or exon skipping, which can result in frameshifts that create null alleles [Forlino and Marini 2016; Marini et al. 2007]. Together, these types of mutations are typically associated with mild, non-deforming OI type I, as the unaffected allele can still produce structurally normal α1 chains, although in reduced amount [Slayton et al. 2000]. Heterozygous loss-of-function quantitative mutations in COL1A2 leading to α2 chain haploinsufficiency do not seem to cause a clinically significant phenotype, while compound heterozygosity or homozygosity that leads to complete absence of α2 chain results in joint hypermobility, skin laxity and valvular problems that are consistent with Ehlers-Danlos phenotype [Malfait et al. 2006; Schwarze et al. 2004].

In contrast, qualitative mutations in the COL1A1 and COL1A2 genes that result in structural defects of type I collagen typically lead to the more severe OI types II, III, or IV [Ben Amor et al. 2011; Forlino et al. 2011; Forlino and Marini 2016; Marini et al. 2007; Rauch et al. 2010; Sato et al. 2016]. The most severe are pathogenic variants that lead to glycine substitutions in the Gly-X-Y repeats within the helical domains, which can disrupt and delay proper triple helix assembly, and result in structurally abnormal type I collagen molecules [Forlino et al. 2011; Marini et al. 2007]. Additionally, delayed folding can expose type I collagen to modifying enzymes for a longer period of time, leading to abnormal post-translational modifications. This can further impair collagen secretion and processing, and may affect normal collagen-cell or collagen-protein interactions within the extracellular matrix [Forlino et al. 2011; Marini et al. 2007; Raghunath et al. 1994]. Hence, the type and location of type I collagen mutations in OI may predict the clinical severity [Ben Amor et al. 2011; Forlino et al. 2011; Forlino and Marini 2016; Marini et al. 2007; Rauch et al. 2010]. Importantly, the genotype-phenotype correlation is not sufficiently clear to be used in clinical setting for the purpose of genetic counseling, such as in the prenatal setting.

Other genes in Osteogenesis Imperfecta

During the past 10 years pathogenic variants in a number of additional genes have been identified that can cause autosomal recessive, autosomal dominant, and X-linked forms of OI (Table 1). These non-collagenous genes are involved in the post-translational modification of type I collagen, act as type I collagen chaperones, are members of intra- and extracellular signaling pathways that regulate bone homeostasis, or have yet unknown functions (Figure 1).

Figure 1:

The role of different proteins in the synthesis and processing of type 1 collagen. Two protein complexes in the endoplasmic reticulum work to ensure proper collagen helical assembly, chaperoning, post-translational modifications and intra-cellular trafficking: 1) CRTAP, P3H1 and PPIB form the prolyl-3-hydroxylation complex, and 2) PPIase FKBP10 forms a complex with serpin H1 and lysyl hydroxylase 2. BMP1 cleaves the C-propeptide of type I procollagen molecules. IFITM5 (localized at the osteoblast cell membrane), and PEDF and WNT1 (which reside in the extracellular matrix) are involved in signaling pathways that regulate differentiation and mineralization.

Collagen 3-hydroxylation complex defects

The first identified gene associated with autosomal recessive OI was cartilage-associated protein (CRTAP, encoded by CRTAP) [Morello et al. 2006]. CRTAP forms a complex with the prolyl 3-hydroxylase 1 (also known as LEPRE1, encoded by P3H1) and peptidyl-prolyl isomerase B (also known as CYPB, encoded by PPIB) proteins in the endoplasmic reticulum (ER) [Ishikawa et al. 2009; Vranka et al. 2004]. The roles of this complex are to ensure proper collagen helical assembly, chaperoning, and post-translational modification. This complex is required for prolyl-3 hydroxylation specifically at a single proline residue of the α1(I) chain (α1(I) Pro986) and of the α2(I) chain (α2(I) Pro707) [Eyre and Weis 2013; Kefalides 1973; Weis et al. 2010]. Loss of CRTAP is associated with reduced prolyl-3 hydroxylation, delayed collagen assembly and type I collagen post-translational over-modification [Morello et al. 2006]. These changes may disrupt collagen-protein/proteoglycan interactions in the extracellular matrix, and may lead to abnormal collagen cross-linking and mineralization [Eyre and Weis 2013; Grafe et al. 2014]. Depending on the level of residual CRTAP expression, biallelic pathogenic variants in CRTAP can lead to a severe OI phenotype with bone fragility and rhizomelia, described as OI type VII, or to a severe lethal phenotype found primarily in the First Nations population in Quebec, Canada that recapitulates the phenotype of OI type II [Morello et al. 2006; Ward et al. 2002].

Not surprisingly, OI-causing biallelic loss-of-function mutations were found subsequently in the two other members of the P3H1 complex, P3H1 and PPIB, in patients affected with severe forms of OI termed OI type VIII and IX, respectively [Baldridge et al. 2008; Pyott et al. 2011; van Dijk et al. 2009; Willaert et al. 2009]. P3H1 shares N-terminal homology with CRTAP and has the catalytic C-terminal domain for prolyl-3 hydroxylation. PPIB plays an important role in the cis-trans conversion of proline for proper folding of the collagen helix [Ishikawa et al. 2009].

Collagen chaperone defects

Collagen chaperones are proteins that reside in the ER, bind to and assist with folding of procollagen, and stabilize the folded collagen molecule to ensure proper post-translational processing [Makareeva et al. 2011]. SERPINH1 and FKBP10 encode the Serpin H1 (also known as HSP47) and Peptidyl-prolyl cis-trans isomerase FKBP10 (PPIase FKBP10, also known as FKBP65) proteins, respectively. Both have collagen chaperone activity. In 2010, an OI-causing homozygous missense mutation in SERPINH1 was identified in a patient with a severe and progressive skeletal phenotype [Christiansen et al. 2010]. Serpin H1 interacts with the triple helical domain of type I collagen and is necessary for proper collagen secretion from the ER and for collagen triple helix stability [Christiansen et al. 2010]. This suggests that loss-of-function Serpin H1 defects may cause OI primarily by affecting intracellular collagen trafficking, type I collagen secretion and stability. FKBP10 mutations were first identified as the cause of OI in a cohort of five consanguineous Turkish families, in whom the OI phenotype co-segregated with autosomal recessive epidermolysis bullosa simplex, and a Mexican family (presenting only with OI phenotype) through homozygosity mapping [Alanay et al. 2010]. PPIase FKBP10 protein has been shown to play a role in folding of collagen helices [Zeng et al. 1998], and to interact with collagen as a chaperone to prevent aggregation [Ishikawa et al. 2008; Murphy et al. 2011]. Furthermore, in primary OI patient’s fibroblasts, loss of PPIase FKBP10 led to reduced collagen deposition in the extracellular matrix and decreased collagen cross-linking [Barnes et al. 2012]. Interestingly, Bruck syndrome, a rare variant form of OI characterized by brittle bone with joint contractures, was also found to be caused by recessive loss-of-function biallelic pathogenic variants in either FKBP10 [Kelley et al. 2011; Lietman et al. 2014; Shaheen et al. 2011] or PLOD2 [Ha-Vinh et al. 2004; Puig-Hervás et al. 2012]. The PLOD2 gene encodes the lysyl hydroxylase 2 enzyme, which likely interacts with PPIase FKBP10 and plays an important role in collagen telopeptide lysine hydroxylation and collagen cross-linking [Eyre and Weis 2013; Schwarze et al. 2013].

Collagen C-propeptide cleavage defect (late collagen-processing defects)

Biallelic loss-of-function mutations in bone morphogenetic protein 1 (BMP1 encoded by BMP1), which cleaves the C-propeptide of type I procollagen molecules, were reported in a number of families with a recessively inherited severe form of OI (classified as OI type XIII) [Cho et al. 2015; Puig-Hervás et al. 2012; Syx et al. 2015]. BMP1 is also involved in the processing of other types of procollagens and in the activation of lysyl oxidase, which functions in collagen cross-linking, both of which may contribute to the severe phenotype [Asharani et al. 2012; Puig-Hervás et al. 2012].

Defects in signaling molecules, and other genes regulating bone homeostasis

Homozygous or compound heterozygous biallelic loss-of-function WNT1 mutations were detected by whole exome sequencing in patients with recessive OI with a moderate and progressively deforming phenotype [Laine et al. 2013]. Interestingly, heterozygous WNT1 mutations can cause early-onset osteoporosis [Keupp et al. 2013; Laine et al. 2013]. Recent studies have shown that WNT1 mutations can impair canonical WNT signaling and osteoblast differentiation [Baron and Kneissel 2013; Joeng et al. 2014; Laine et al. 2013].

Several groups have reported autosomal recessive loss-of-function mutations in SERPINF1 causing OI type VI, which results in a moderate to severe OI phenotype [Becker et al. 2011; Glorieux et al. 2002; Homan et al. 2011; Rauch et al. 2012; Venturi et al. 2012]. This OI type was recognized as a distinct clinical entity due to the unique mineralization defect and accumulation of unmineralized osteoid seen in histological analyses. Type I collagen in OI type VI patients has normal post-translational modifications, collagen folding and secretion, indicating that mutations in SERPINF1 do not impair type I collagen synthesis or processing [Becker et al. 2011]. SERPINF1 encodes pigment epithelium derived factor (PEDF), a ubiquitously expressed and secreted protein that inhibits osteoclast differentiation through upregulation of the tumor necrosis factor receptor superfamily member 11B (TNFRSF11B, also known as osteoprotegerin) [Akiyama et al. 2010]). Studies in Serpinf1^−/− mice showed that Pedf replacement via helper-dependent adenoviral vector-mediated expression in the liver failed to correct the skeletal phenotype, suggesting that PEDF acts in a paracrine fashion in bone [Rajagopal et al. 2016].

Other genes that have been reported to cause autosomal recessive OI include: TMEM38B, encoding TRICB (Trimeric Intracellular Cation Channel type B), which is involved in calcium release from the ER/sarcoplasmic reticulum [Shaheen et al. 2012; Volodarsky et al. 2013]; SP7, encoding transcription factor Sp7 (zinc finger protein osterix), a key transcription factor that regulates osteoblast differentiation [Lapunzina et al. 2010]; SPARC, encoding the Secreted protein acidic and rich in cysteine (also known as osteonectin) [Mendoza-Londono et al. 2015]; and CREB3L1, encoding Cyclic AMP-responsive element-binding protein 3-like protein 1, an ER stress transducer that is involved in procollagen expression [Symoens et al. 2013]. Additionally, loss-of-function mutations in PLS3, encoding plastin-3, have been described in X-linked osteoporosis with fractures that shows clinical overlap with OI [Fahiminiya et al. 2014; van Dijk et al. 2013].

OI type V is characterized by the unique clinical features of hypertrophic callus formation and calcification of the interosseous membrane [Glorieux et al. 2000]. It is caused by a recurrent, dominant mutation (c.−14C>T) in interferon inducible transmembrane protein family 5 (IFITM5, also known as BRIL) [Cho et al. 2012; Rauch et al. 2013; Semler et al. 2012] [Shapiro et al. 2013; Takagi et al. 2013]. This mutation introduces a new start codon adding 5 amino acid residues to the IFITM5 protein, and is thought to have a neomorphic effect through unknown mechanisms [Hanagata 2016; Lietman et al. 2015]. IFITM5 belongs to the highly conserved IFITM (interferon-induced-transmembrane) family of proteins that are implicated in innate immune response. Given that the localization of the mutant IFITM5 is unaltered [Patoine et al. 2014], protein interactions and downstream signaling cascades may be affected by this mutant protein, which inhibits differentiation and mineralization in bone.

Lastly, heterozygous, likely dominant-negative mutations in P4HB, encoding the collagen type I chaperone protein disulfide isomerase (PDI), and biallelic loss-of-function mutations in SEC24D, encoding a component of the coat protein II (COPII) complex that functions in protein transport from the ER, have been reported in Cole-Carpenter syndrome, a syndromic bone fragility disorder characterized by features of osteogenesis imperfecta and craniofacial abnormalities [Garbes et al. 2015; Moosa et al. 2015; Rauch et al. 2015].

More recently suggested classifications of OI propose to include these new genetic discoveries in the clinical classification of OI, currently termed OI types I-XVI. Please refer to recent reviews for a comprehensive overview of the genetics causes of OI, clinical characteristics, and classification [Forlino and Marini 2016; Marini et al. 2014; Van Dijk and Sillence 2014]. Future studies have to validate if there are clinical benefits for genotype-specific treatment strategies for OI patients, as most patients with recessive OI or IFITM5 mutations can be clinically categorized according to the original Sillence classification.

Diagnosis of OI

The diagnosis of OI is based primarily on the clinical and radiographic features. Historically, the diagnosis was confirmed by biochemical analysis of type I collagen molecules derived from cultured dermal fibroblasts [van Dijk et al. 2012]. Analysis of the electrophoretic mobility of collagen on SDS-urea-PAGE gels can detect both quantitative and qualitative variations [Wenstrup et al. 1990]. Quantitative variations can be appreciated by detecting decreased amount of normal collagen in affected individuals. Structural variations that result in collagen post-translational over-modification, due to abnormal folding and processing of the collagen molecule, delay the electrophoretic migration [Cabral et al. 2006; Wenstrup et al. 1990]. However, the biochemical analysis will not be useful in the diagnosis of OI types that are not associated with altered collagen synthesis (such as OI type V and OI type VI). Thus, molecular diagnosis by DNA sequencing is often necessary [Cabral et al. 2006; van Dijk et al. 2012]. As the technology of next-generation sequencing (NGS) became available, gene panels were designed to facilitate sequencing of multiple OI genes in a cost-effective and rapid manner [Sule et al. 2013]. Whole exome sequencing enabled the discovery of new OI genes, but also identified variants of uncertain clinical significance in known OI genes, and rare modifier alleles with a unique genotype-phenotype correlate [Lu et al. 2014]. Chromosomal microarray analysis may be utilized to search for copy number variations when sequencing results are normal. RNA and protein analysis from cultured dermal fibroblasts plays an important role in studying the pathogenicity of splicing or missense variants of uncertain clinical significance. Obtaining an accurate molecular diagnosis is required for the purpose of genetic counseling, anticipatory guidance, prenatal genetic diagnosis and identifying relatives at risk by sequencing of the known familial mutation. In the future, a genetic diagnosis may guide the utilization of specific treatments.

Bone abnormalities in OI

Bone contains three major cell types: two of mesenchymal origin (osteoblasts and osteocytes), and one of hematopoietic origin (osteoclasts) [Sheng 2015; Xiao et al. 2015]. Osteoblasts are bone-forming cells which reside on the bone surface and deposit new bone matrix. Osteoclasts are multinucleated cells that facilitate bone resorption. Old bone is constantly removed by osteoclasts and replaced with new bone by osteoblasts, a process termed bone-remodeling. Bone remodeling rate is much higher in the pediatric age group, due to active growth [Parfitt et al. 2000; Rauch et al. 2007]. In healthy bone, resorption and formation are balanced to maintain total bone mass; however, in bone disorders, a relatively higher resorption or reduced formation can lead to a net bone loss. For instance, bone from patients with more severe forms of OI typically exhibits an increased bone-remodeling rate, or high-turnover, with increased numbers of osteoclasts and osteoblasts, leading to an overall bone loss and reduced bone quantity [Doty and Mathews 1971; Falvo and Bullough 1973].

Some osteoblasts can differentiate into osteocytes, which reside in lacunae within the mineralized bone matrix, forming a 3-dimensional network connected by dendrite-like morphology [Dallas and Bonewald 2010; Dallas et al. 2013]. Osteocytes play an important role in sensing various stimuli in the bone environment such as mechanical loading [Bonewald 2006], and they actively control bone-remodeling by osteoblasts and osteoclasts via secretion of factors such as sclerostin or Receptor activator of nuclear factor kappa-B ligand (RANKL) [Nakashima et al. 2011; Wijenayaka et al. 2011].

Several studies in mouse models and utilizing bone biopsies from OI patients have shown that the bone abnormalities in OI not only include a reduction of bone quantity, but also abnormal bone quality. This includes the altered bone cellular composition, impaired biomechanical properties of the bone material, an abnormal bone matrix composition, altered mineralization, and abnormal type I collagen-collagen cross-linking [Bishop 2016; Doty and Mathews 1971; Falvo and Bullough 1973; Rauch et al. 2000; Sarathchandra et al. 2000]. Despite considerable clinical variability and genetic heterogeneity in OI, most OI types exhibit a severely increased “brittleness” of the bone material, which contributes to the dramatically increased fracture incidence that is seen in most cases. The pathogenesis of the “brittle bone” material has been under intense study and remains as an open question. Current theories on what causes this brittleness include consequences of the abnormal collagen synthesis or processing (post-translational collagen modifications, chaperoning, and assembly), altered mineralization associated with abnormal collagen cross-links, or other mechanisms [Bishop 2016; Eyre and Weis 2013; Marini et al. 2014].

Therapeutic interventions in Osteogenesis Imperfecta

The treatment goals for patients with OI are to decrease fracture incidence, relieve bone pain, promote mobility and growth, and improve patient independence. Most pharmacological approaches used to date were originally developed for the treatment of osteoporosis, and they aim to increase bone mass by inhibiting the activity of bone resorbing osteoclasts, or increasing bone formation by osteoblasts. Bone densitometry by Dual Energy X-ray Absorptiometry (DEXA) is frequently used to clinically assess the bone mineral density (BMD) in patients, for follow up on disease progression, to decide on treatment initiation and to follow up on treatment outcome.

Overall, many patients achieve a better functional outcome from combined pharmacological and physical therapy, occupational therapy and rehabilitation [Montpetit et al. 2015]. Orthopedic surgery interventions are indicated for the treatment of fractures, long bone deformities and scoliosis [Harrington et al. 2014; Thomas and DiMeglio 2016].

With respect to nutrition, an adequate intake of calcium and vitamin D is recommended [Chagas et al. 2012; Harrington et al. 2014]. Meta-analysis of randomized controlled trials studying the effect of calcium and vitamin D supplementation on fracture prevention in osteoporosis supports their combined use to reduce fracture risk [Weaver et al. 2016]. In pediatric patients with OI, serum levels of 25-hydroxyvitamin D were positively correlated with BMD [Edouard et al. 2011]. However, high-dose vitamin D (2000 IU) did not prove to be superior to low-dose vitamin D (400 IU) in increasing lumbar spine bone density [Plante et al. 2016]. As OI patients are often small-for-age, weight and height should be considered when adjusting the dose of supplements to prevent potential side effects (primarily hypercalciuria).

Pharmacologic interventions

Bisphosphonate therapy has become the mainstay of care in pediatric patients with moderate to severe OI [Glorieux et al. 1998; Hald et al. 2015; Shi et al. 2016]. A number of other pharmacologic approaches are currently being studied in OI, such as inhibition of the receptor activator of nuclear factor kappa-B ligand (RANKL), osteo-anabolic agents including human parathyroid hormone analogue, sclerostin-inhibition, and inhibition of transforming growth factor beta (TGFβ) signaling (Figure 2).

Figure 2:

The bone remodeling unit is the target of therapeutic interventions as illustrated in the figure. Bisphosphonates (BPs) reside in the bone matrix and act to inhibit osteoclast activity. Cathepsin K inhibitors such as Odanacatib (ODN) also inhibit the resorptive activity of osteoclasts. Denosumab targets the receptor activator of nuclear factor kappa-B ligand (RANKL) that stimulates the formation and activation of osteoclasts. Teriparatide (a PTH analogue) exerts an anabolic effect by promoting osteoblast differentiation and activity. Sclerostin inhibitory antibodies (Scl-Ab) increase bone formation via Wnt signaling. Studies show increased TGF-β signaling in OI, and TGF-β inhibitors are thought to have a beneficiary effect by inhibition of this pathway in bone.

Bisphosphonates

Bisphosphonates (BPs) are stable analogues of pyrophosphate that inhibit bone resorption by osteoclasts [Fleisch 1998]. They are currently considered the standard of pharmacological care in patients with moderate to severe OI, and have been shown to increase bone mass, and to some extent, reduce the fracture risk in OI patients [Glorieux et al. 1998; Lindahl et al. 2016; Shi et al. 2016].

The BPs currently in clinical use (Pamidronate, Alendronate, Risedronate and Zoledronic acid) contain nitrogen side chains that are thought to block the mevalonate pathway of sterol synthesis and interfere with post-translational modifications of GTP-binding proteins, which are essential for osteoclastic cytoskeletal function [Soares et al. 2016]. BPs can be administered either orally or intravenously. Advantages of intravenous administration include a lower risk for gastrointestinal side effects, better bioavailability, the option to adjust dosing based on body weight, and increased patient compliance compared with oral administration. However, comparative studies failed to show a significant difference in effectiveness between intravenous and oral administration [George et al. 2015; Hald et al. 2015]. Following administration, BPs are quickly concentrated at the bone matrix and may reside there for many months. They are excreted through the kidney and dose should be adjusted to renal function [Soares et al. 2016].

In adults, BPs have been used for a few decades in the treatment of metabolic bone disorders, including osteoporosis, hypercalcemia of malignancy and Paget disease of bone. However, they were found to be less effective for the treatment of adult OI patients [Adami et al. 2003; Chevrel et al. 2006; Shi et al. 2016]. In children, BPs can increase BMD [Falk et al. 2003; Glorieux et al. 1998; Gatti et al. 2005; Land et al. 2006; Letocha et al. 2005; Rauch and Glorieux 2004; Rauch et al. 2009; Rauch et al. 2003; Sakkers et al. 2004; Shi et al. 2016; Ward et al. 2011], improve vertebral height and reshape vertebral deformities [Glorieux et al. 1998; Letocha et al. 2005; Semler et al. 2011], promote growth [Gatti et al. 2005], and in some studies improve mobility [Falk et al. 2003; Glorieux et al. 1998; Land et al. 2006]. Although some clinical trials found that BPs can reduce the fracture incidence in children with OI at certain skeletal sites [Gatti et al. 2005; Glorieux et al. 1998; Letocha et al. 2005; Lindahl et al. 2016; Rauch et al. 2009; Sakkers et al. 2004], this has not been observed in all studies [Ward et al. 2011]. Additionally, BPs have weaker effects in the appendicular skeleton of the pediatric population [Rauch et al. 2009].

Several meta-analyses have investigated overall effectiveness of BPs in OI patients. A Cochrane analysis, which reviewed 14 randomized-controlled trials conducted in OI patients, concluded that treatment with BPs significantly improves BMD [Dwan et al. 2014]. However, data from this analysis were inconclusive regarding a decrease in fracture incidence and improvement in growth, mobility and bone pain. Two recent meta-analyses found only a moderate reduction in fracture incidence in OI patients treated with BPs; one study showed a 20% decrease in fractures [Shi et al. 2016], while the other showed a reduction that did not reach statistical significance [Hald et al. 2015]. The difficulty of showing improvement in fracture endpoints in small clinical studies include observational bias for fractures, stratification in cohorts with different severities of OI, and potential effect of treatment on activity and hence fracture risk. With respect to other outcomes, studies failed to demonstrate a beneficial effect of BPs on the occurrence and progression of scoliosis [Palomo et al. 2015; Sato et al. 2016].

BPs are generally well-tolerated [George et al. 2015]. The most common side effects include gastrointestinal problems when taken orally and an acute phase infusion reaction (hyperpyrexia, myalgia, and weakness) that typically occurs in BP-naïve patients during the first infusion. This is likely to be an immune-mediated reaction, although the exact mechanism is not well understood. It can be prevented by the concurrent administration of nonsteroidal anti-inflammatory drugs or acetaminophen [Glorieux et al. 1998; Kennel and Drake 2009]. Treatment may be associated with transient hypocalcemia, which can be prevented by supplementing with calcium and vitamin D. Osteonecrosis of the jaw is a well-documented but rare side effect in adults, but it has never been reported in pediatric patients [Soares et al. 2016]. With long term use, a concern exists for continuous suppression of bone turnover, which ultimately may affect linear growth in pediatric patients. However, this is not reflected in clinical studies [Zeitlin et al. 2003]. In addition, atypical femur fractures, more commonly seen in adult patients with osteoporosis during and after treatment with BPs [Saita et al. 2015; Schilcher et al. 2014], have been rarely reported in pediatric patients with OI [Etxebarria-Foronda and Carpintero 2015; van de Laarschot and Zillikens 2016].

In summary, treatment with BPs has significantly improved the condition of OI patients and has become the mainstay of pharmacological care, particularly in moderate to severe OI. In patients with milder forms of OI, treatment with bisphosphonates may be considered when there is a history of recurrent fractures, low-trauma fractures or vertebral compression fractures, or with significant bone pain. Treatment increases total bone mass, and may to some extent reduce the fracture risk, particularly in pediatric patients; however, the effects of BPs in OI are less robust than in patients with osteoporosis, and BPs may be less effective in adult OI patients [Hald et al. 2015; Shi et al. 2016].

Denosumab

Denosumab (Prolia, Amgen) is a human monoclonal antibody that targets the receptor activator of nuclear factor kappa-B ligand (RANKL). RANKL is a cytokine that is expressed and secreted by osteoblasts and osteocytes in bone, and stimulates the formation, activation and survival of osteoclasts. Similar to BPs, inhibition of RANKL decreases osteoclast activity and bone resorption. Pre-clinical study in cynomolgus monkey model of post-menopausal osteoporosis showed that treatment with denosumab resulted in continuous periosteal bone formation in addition to inhibited resorption [Ominsky et al. 2015], suggesting a potential advantage of denosumab therapy compared to BPs. In the oim/oim mouse model of OI (due to a mutation in Col1a2), RANKL inhibitors improve bone density and bone mechanical properties, and also decrease fracture incidence [Bargman et al. 2010; Bargman et al. 2012].

Clinical trials are ongoing to evaluate the long-term effect and safety of denosumab therapy in OI patients. Semler and colleagues studied the effect of denosumab in four pediatric patients with OI type VI (cause by PEDF mutations) who responded poorly to bisphosphonate therapy [Semler et al. 2012]. Treatment was well-tolerated, and resulted in a decrease in biochemical markers of bone resorption. After 2 years of treatment, BMD increased in all patients. In addition, the results indicated a decrease in fracture rate and mild improvement in spine morphology after 2 years of therapy in this cohort [Hoyer-Kuhn et al. 2014]. A phase 2 clinical trial in ten patients with OI showed a significant increase in lumbar spine BMD, with no significant change in spine morphology or in mobility [Hoyer-Kuhn et al. 2016].

Currently, there are limited data regarding the clinical use of denosumab in OI, and it is only approved for the treatment of osteoporosis and metastatic bone disease in adults [Farrier et al. 2016]. The reported side effects include rare hypocalcemia, osteonecrosis of the jaw, atypical hip fractures, and risk for rebound hypercalcemia after cessation of treatment [Hoyer-Kuhn et al. 2016].

Cathepsin K inhibition

Cathepsin K is a lysosomal protease expressed and secreted primarily by osteoclasts that participates in bone resorption by degrading type I collagen in the extracellular matrix [Duong et al. 2016; Mukherjee and Chattopadhyay 2016]. Loss-of-function variants in CTSK, encoding Cathepsin K, are the genetic cause of pycnodysostosis [Gelb et al. 1996; Johnson et al. 1996], an autosomal recessive disorder characterized by a defect in osteoclast function leading to increased bone mass, short stature and bone fragility.

Cathepsin K inhibition in bone reduces the enzymatic resorption process of the extracellular matrix by mature osteoclasts, while still allowing acid release [Duong et al. 2016], [Mukherjee and Chattopadhyay 2016]. This effectively maintains osteoclast-mediated release of “coupling” factors from the bone matrix that attract osteoblasts to induce bone formation, while preventing degradation of the matrix. Unlike BPs, which can reside for months on the bone surface, the effects of cathepsin K inhibitors are transient and reversible, which may provide an advantage in clinical management [Zhuo et al. 2014]. Pre-clinical studies with Odanacatib (ODN, Merck & Co., Inc. (Kenilworth, NJ)) showed improved bone mass and strength in ovariectomized monkeys [Pennypacker et al. 2014]. A phase 2 clinical trial in patients with osteoporosis demonstrated a continuous increase in BMD during 8 years of treatment [Rizzoli et al. 2016]. However, the phase 3 Long-Term ODN Fracture Trial (LOFT) in postmenopausal osteoporosis, which included 16,071 participants, has been recently discontinued after analysis of adverse cardiovascular events confirmed an increased risk of stroke (http://www.mercknewsroom.com/news-release/research-and-development-news/merck-provides-update-odanacatib-development-program). The effects of Cathepsin K inhibitors were not studied in patients with OI.

Growth Hormone

Osteo-anabolic therapies, which are targeted to increase bone formation by osteoblasts, are currently being investigated as an alternative, or complementary approach to anti-resorptive therapies to increase bone mass and strength in OI. Growth hormone therapy was studied in pediatric patients with OI as a potential anabolic agent and to address growth deficiency. In response to treatment with recombinant human growth hormone (rhGH) injections, patients with OI type IV exhibited increased bone formation rate and a positive effect on linear growth [Marini et al. 2003]. In contrast, patients with more severe OI type III did not respond efficiently to the therapy. Combined treatment with rhGH and BPs had a synergistic effect on growth velocity and BMD but did not reduce fracture incidence [Antoniazzi et al. 2010]. Sufficient clinical data are lacking to support the safety and effectiveness of this treatment in individuals with OI, and it is currently not in standard clinical use.

Teriparatide

Teriparatide (recombinant human PTH 1-34, Forteo, Eli Lilly & Co. (Indianapolis, IN)) is a parathyroid hormone (PTH) analogue with bone-anabolic activity that is approved for use in adult patients with osteoporosis. PTH stimulates both bone formation and resorption. However, while continuous PTH infusion leads to net bone loss, intermittent PTH administration results in net increase in bone mass [Neer et al. 2001]. Teriparatide is a recombinant protein that is identical to the biologically-active N terminal region of PTH (amino acids 1-34), and is given via daily injections. Pre-clinical studies in rats that received high-dose PTH 1-34 treatment raised concerns for a higher risk for the development of osteosarcoma [Neer et al. 2001; Vahle et al. 2002]. Due to this concern, the approved duration of treatment with teriparatide in patients with osteoporosis is limited to 24 months, and it is not approved for use in pediatric patients. However, after years of employing teriparatide as a treatment in osteoporosis patients, recent studies have shown no indications of increased occurrence of osteosarcoma [Cipriani et al. 2012; Lindsay et al. 2016]. A meta-analysis of the effects of teriparatide in osteoporosis concluded that there is a positive effect on bone formation, an increase in bone mass and strength and a decrease in fracture risk, with a relative risk reduction of up to 85% for vertebral fractures, and 40-60% for non-vertebral fractures, depending on the study [Lindsay et al. 2016]. It is of note that cessation of teriparatide therapy leads to accelerated bone loss, thus consolidation with an anti-resorptive agent (such as BPs or Denosumab) should be considered [Black et al. 2005].

Data on treatment in adult OI patients are limited. A study in a cohort of 13 post-menopausal women affected with OI type I, showed that teriparatide therapy led to increased BMD at the spine but not at the hip, accompanied by an anabolic response of biochemical bone remodeling markers [Gatti et al. 2013]. The first controlled trial of PTH therapy in OI tested the effects of teriparatide in 79 adult patients that were randomized to receive teriparatide or placebo treatment for 18 months [Orwoll et al. 2014]. Results showed a significant increase in BMD and estimated vertebral strength in patients with mild OI (type I), and an anabolic pattern of change in biochemical bone remodeling markers. However, patients with moderate to severe OI (type III and IV) showed no increase in BMD in response to teriparatide therapy, suggesting that teriparatide treatment is not effective for the treatment of patients with more severe forms of OI. The study was not powered to address the effect on fracture risk [Orwoll et al. 2014].

Sclerostin-inhibitory antibodies

Sclerostin (encoded by the SOST gene) is a glycoprotein that acts as a negative regulator of bone mass by inhibiting WNT/β-catenin signaling in osteoblasts [Sharifi et al. 2015]). It is secreted predominantly by osteocytes and exerts its effect by binding to the WNT co-receptors low-density lipoprotein receptor-related proteins 5 and 6 (LRP5 and LRP6) on the membrane of osteoblastic cells [Li et al. 2005; Poole et al. 2005; Semënov et al. 2005]. Pathogenic loss-of-function variants in the SOST gene cause sclerosteosis, a rare disorder characterized by high bone mass [Balemans et al. 2001; Brunkow et al. 2001].

Monoclonal antibodies that target sclerostin (Scl-Ab/Romosozumab, Amgen Inc. (Thousand Oaks, CA); Blosozumab, Eli Lilly; BSP804, Novartis (Basel, Switzerland) / Morphosys (Planegg, Germany)) have been developed as a novel anabolic therapy to increase bone mass. Presumably, sclerostin inhibition may have a beneficial effect in disorders wherein abnormal WNT signaling plays a major role, such as WNT1-related forms of OI [Keupp et al. 2013; Laine et al. 2013], or LRP5-mediated osteoporosis-pseudoglioma syndrome [Kedlaya et al. 2013]. Pre-clinical studies in OI animal models demonstrated beneficial effects of Scl-Ab therapy on bone formation and BMD. In the Brtl/+ mouse model of dominant OI (moderate to severe phenotype), treatment with Scl-Ab resulted in increased bone mass and improved bone mechanical strength in both adult and growing mice, but did not have major effects on the increased brittleness of the bone material [Sinder et al. 2015; Sinder et al. 2014]. However, in another mouse model of dominant OI, Col1a1^Jrt/+ with a severe OI phenotype, treatment with BPS804 was found to be less effective [Roschger et al. 2014]. In Crtap^−/− mice that model a severe form of recessive OI, Scl-Ab administration improved bone mass, bone strength and bone formation rate, but again did not improve the bone brittleness [Grafe et al. 2016].

Phase 1 and 2 clinical trials of Romosozumab in post-menopausal osteoporosis revealed a significant increase in bone formation and, interestingly, also decreased bone resorption, ultimately resulting in increased BMD [McClung et al. 2014]. The medication was administered subcutaneously at 1-3 month intervals, and the response to treatment was greater in the Scl-Ab group compared with placebo, bisphosphonate or teriparatide therapy. A similar beneficial response was observed in another Phase 2 clinical trial with Blosozumab in patients with post-menopausal osteoporosis [Recker et al. 2015]. Together, these data suggest that treatment with sclerostin-neutralizing antibodies may be beneficial for the future treatment of OI patients to increase bone mass and strength.

TGF-β inhibition

TGF-β is secreted by osteoblasts, resides in the bone matrix and is known to modulate bone remodeling by coupling the activity of bone resorbing osteoclasts with bone forming osteoblasts [Balooch et al. 2005; Tang et al. 2009]. Increased TGF-β signaling had been associated with osteopenia and bone fragility in mouse models [Borton et al. 2001; Erlebacher and Derynck 1996]. In wild-type mice, treatment with a TGF-β receptor kinase inhibitor (SD-208, Scios Inc. (Fremont, CA)) or a TGF-β neutralizing antibody (1D11, Sanofi Genzyme (Cambridge, MA)) was associated with an increase in osteoblast numbers, decreased osteoclast activity and improved bone mass and strength [Edwards et al. 2010; Mohammad et al. 2009]. A recent study in mouse models of both recessive and dominant OI has shown that increased TGF-β signaling plays an important role in the development of the OI phenotype [Grafe et al. 2014]. Interestingly, OI mice treated with the TGF-β inhibiting antibody 1D11 showed normalized bone mass and improved bone strength [Grafe et al. 2014]. However, similar to other pharmacological approaches, TGF-β inhibition did not improve the increased brittleness of the bone material. A phase I clinical trial to study the effects of anti-TGF-β therapy with Fresolimumab (GC1008, Genzyme) in OI patients is expected to launch in the near future (http://www.rarediseasesnetwork.org/cms/BBD).

Combination therapy

Combination of anabolic and anti-resorptive therapies may offer an additive or synergistic effect to improve bone mass more than each therapy alone. Preliminary findings from pre-clinical studies in the Col1a2 G610C mouse model of moderate dominant OI [Daley et al. 2010] showed that combining sclerostin-antibody and zoledronic acid therapies resulted in increased BMD and bone strength that was superior to either therapy alone [Munns et al. 2015]. A recent trial in patients with severe osteoporosis showed that adding teriparatide to ongoing alendronate treatment resulted in increased bone formation rate and mineralizing surface in iliac crest biopsies [Dempster et al. 2016]. Future studies are needed to determine the potential benefits of combinatorial therapy on bone mass and fracture risk in OI.

The pharmacological treatments described thus far aim to target the activity of bone resorbing osteoclasts and bone forming osteoblasts to increase bone mass, but do not provide an ultimate cure for this genetic disorder. These treatments do not address the defective type I collagen, which leads not only to reduced bone quantity, but also to abnormal bone matrix and bone quality. The potential of cell and gene therapy approaches to correct the genetic defects and also improve bone quality is currently being studied.

Cell therapy

Bone marrow-derived mesenchymal cells are believed to be the source of osteoprogenitors and can differentiate to osteoblasts in vitro [Jethva et al. 2009]. Bone marrow transplantation was attempted in both animal models and OI patients, with the intention of introducing bone-forming cells that produce normal type I collagen. Despite low levels of mesenchymal stem cell engraftment, most results showed an increase in BMD, accelerated growth rate and decreased fracture risk following transplantation, suggesting that the introduction of even a low amount of normal type I collagen may be beneficial to improve the impaired bone quantity and quality in OI [Götherström et al. 2014; Horwitz et al. 1999; Otsuru et al. 2012; Westgren and Gotherstrom 2015]. However, this therapeutic approach is currently still considered experimental and controversial with respect to the risks and benefits in patients with OI. Its clinical use is also limited by the toxicity associated with bone marrow transplantation and lack of suitable donors.

Gene therapy

As discussed above, most severe forms of OI result from pathogenic variants that produce structurally abnormal proα1(I) and proα2(I) collagen chains. Such variants (mostly glycine substitutions) exert a dominant-negative effect by disrupting the triple helix structure of the collagen molecule. Gene targeting of the abnormal allele may reduce the expression of mutant proα1(I) or proα2(I) chains, and transform a qualitative collagen defect to a quantitative defect with a milder OI phenotype.

In a proof-of-concept experiment, Chamberlain and colleagues infected OI patient-derived marrow stromal cells with an Adeno-Associated Virus (AAV)- vector designed to disrupt exon 1 of the COL1A1 gene [Chamberlain et al. 2004]. Random integration resulted in transduced clones that expressed only wild-type collagen polypeptides, and were able to differentiate and form bone in vivo. Even though the AAV vector can equally disrupt either the mutant or the wild-type allele, the authors postulated that genetic mosaicism with a subset of cells expressing wild-type collagen may be sufficient to induce a beneficial clinical effect. Others have demonstrated that silencing of COL1A1 or COL1A2 expression by RNA-interference causes a significant reduction of mutant collagen in cell cultures [Lindahl et al. 2013; Rousseau et al. 2014]. The issues of efficient delivery to bone tissue, safety of genetic manipulation and maintaining long-term effects in vivo remain to be addressed.

Conclusion

Anti-resorptive BPs, that inhibit the activity and survival of the bone-resorbing osteoclasts, are currently the pharmacological “standard of care” for the treatment of OI and have certainly improved the clinical course of OI patients. The clinical value of new anti-resorptive drugs for the treatment of OI, including the RANKL-inhibitory antibody, denosumab or cathepsin K inhibitors has yet to be demonstrated in OI. However, they may offer potential advantages over BPs including the “uncoupling” of bone resorption from bone formation, so that anti-resorptive effects do not also cause secondary suppression in bone formation as is the case with BPs. Treatment with sclerostin-inhibitory antibodies may provide a new osteo-anabolic option for the treatment of patients with more severe forms of OI, as treatment with teriparatide appears to be less effective in patients with moderate to severe OI. Most of the pharmacological approaches have in common that they were originally designed for the treatment of osteoporosis, and hence it is not surprising that they seem to be comparatively less effective in OI as they are not based on the specific genetic and molecular pathophysiology of OI. Ultimately, the effect size of different treatments may vary depending on molecular lesion and mechanisms so that personalized, genotyped based treatment strategies could be warranted. For example, augmentation of WNT signaling with sclerostin inhibition may be the best approach in WNT cases of OI. Another example may be inhibition of the increased TGF-β signaling activity that has been found in bone of mouse models of both dominant and recessive OI caused by collagen and/or collagen modification defects. Although several of these pharmacological treatments have been shown to increase BMD (as determined by DEXA scan) in OI patients and bone mass in mouse models, it is likely that the bone quality, including matrix composition and collagen cross-linking, is still abnormal and the bone material remains fragile. Thus, eventual targeting of bone quality may be required for maximal clinical benefit, though the field as a whole has moved dramatically forward since the discovery of type I collagen mutation in the 1980’s.

Biographies

Ronit Marom is an Assistant Professor in the Department of Molecular and Human Genetics at Baylor College of Medicine in Houston, Texas. She is a clinician and researcher with special interest in skeletal dysplasias and inborn errors of metabolism.

Yi-Chien Lee is a fifth year PhD student in the Department of Molecular and Human Genetics, and Program of Integrative Molecular Biomedical Sciences at Baylor College of Medicine. Her interests include the disease mechanisms and treatments options of Osteogenesis Imperfecta.

Ingo Grafe is an Instructor in the Department of Molecular and Human Genetics, Baylor College of Medicine, Houston, Texas, USA. He is a physician researcher interested in understanding the mechanisms leading to bone diseases such as Osteogenesis Imperfecta, in order to improve the treatment options for patients.

Dr. Brendan Lee is the Robert and Janice McNair Endowed Chair in the Department of Molecular and Human Genetics at Baylor College of Medicine. He has focused on the translational study of skeletal dysplasias and inborn errors of metabolisms throughout his career.