Heterotopic ossification and the elucidation of pathologic differentiation

Abstract

Tissue regeneration following acute or persistent inflammation can manifest a spectrum of phenotypes ranging from the adaptive to the pathologic. Heterotopic Ossification (HO), the endochondral formation of bone within soft-tissue structures following severe injury serves as a prominent example of pathologic differentiation; and remains a persistent clinical issue incurring significant patient morbidity and expense to adequately diagnose and treat. The pathogenesis of HO provides an intriguing opportunity to better characterize the cellular and cell-signaling contributors to aberrant differentiation. Indeed, recent work has continued to resolve the unique cellular lineages, and causative pathways responsible for ectopic bone development yielding promising avenues for the development of novel therapeutic strategies shown to be successful in analogous animal models of HO development. This review details advances in the understanding of HO in the context of inciting inflammation, and explains how these advances inform the current standards of diagnosis and treatment.

1. Introduction

1.1. Heterotopic ossification an intriguing and enigmatic pathology

The consequence of tissue healing and regeneration in response to trauma or persistent inflammation can run the physiologic spectrum from protective to overtly pathologic - the balance between these can often be precarious. This paradigm is particularly evident in the pathogenesis of heterotopic ossification (HO), the formation of endochondral bone at extraskeletal sites within soft and connective tissue. HO is a debilitating sequela of local and systemic inflammatory insults, and commonly occurs following orthopedic procedures, severe burns, brain and spinal cord damage, and blast injuries suffered in combat [1]. HO also forms spontaneously in patients who suffer from Fibrodysplasia Ossificans Progressiva (FOP), Progressive Osseous Heteroplasia (POH), and GNAS-based spectrum of HO disorders in which inborn genetic error results in hyperactivity of pro-osteogenic signaling pathways [2,3]. Lessons gleaned from embryologic and developmental research demonstrate bone development to be an exquisitely regulated process, in which the fate of mesenchymal progenitors is directed by cellular, paracrine, and mechanical stimuli. As such, HO development remains a unique pathophysiologic process through which we can better understand the processes of wound healing and tissue regeneration on the level of cellular population dynamics and specific signaling pathways.

A body of research has emerged over recent years to elucidate the processes underlying the genesis and propagation of ectopic bone. The paradigm of cellular differentiation, previously thought to be terminally unidirectional and hierarchical, we now know is more complex; cell and tissue fate is fluid and can be shaped by the homeostatic environment, including biochemical and biomechanical stimuli, to profound effect. Despite this complexity, our understanding of contributory cellular lineages to bony foci, and why these foci form in characteristic distributions continues to resolve. The pathological differentiation driving the formation of ectopic bone is attributable to a confluence of signaling pathways. Not least among them, the type I bone morphogenetic proteins (BMP) receptor family and down-stream mediators stand out as promising therapeutic targets. Discovery of the highly conserved R206H mutation found in ACVR1/ALK2 culpable for FOP, and subsequent study of BMP signaling has elucidated many of the mechanisms through which ectopic osteogenesis can occur [4]. No genetic mutation has been associated with patients who develop traumatic HO as it appears to be a process that can occur in any patient subjected to extensive trauma. The complex interplay between the BMP-SMAD axis and other signaling pathways involved in growth regulation, hypoxia sensing, and retinoic acid signaling provides valuable insights through which novel, specifically targeted therapies can be designed. Informed by these advances, the assessment, diagnosis, and treatment of patients at risk of developing HO continue to evolve.

1.2. HO as a consistent clinical issue

The development and propagation of ectopic bone in response to trauma remains a persistent clinical issue, incurring significant cost and patient morbidity. Cases of HO are broadly classified into three etiological categories: neurogenic including traumatic brain and spinal cord injury; orthopedic resulting from arthroscopy, fracture fixation, and joint replacement; and finally traumatic (severe burns and high-velocity blast injury fall within this category) (Fig. 1). Multiple animal models exist to recreate and study spontaneous HO development, and each varies in the physiologic accuracy in their approximation of the inciting traumatic etiologies [5]. The burn-tenotomy in small rodents utilizes concomitant burn and connective tissue injury to yield consistent ectopic bone with close to 100% penetrance [6]. Recently, a rat hind limb extremity/amputation injury model was developed to recreate the multifactorial contributions of polytrauma using a pneumatically driven shock tube. This model incorporates all components of a blast injury; specifically, exposure to blast over pressure (BOP), femur fracture, crush injury, amputation through the zone of injury and infection at the site of injury [7]. Other models rely on supraphysiologic concentrations of recombinant cytokines, most often BMP-2, and BMP-4, integrated into a biocompatible scaffold and surgically embedded within musculo-skeletal fascial planes of experimental animals [8]. Until recently, no animal model had existed to successfully recreate HO formation in response to neural injury; Genèt et al. were able to induce HO using spinal cord transection in conjunction with intramuscular administration of the snake venom, cardiotoxin [9]. The distribution of bone formation along fascial planes in close proximity to the hip joint approximated the analogous formation characteristic of human patients [9].

Fig. 1.

HO Development: (A) Traumatic etiologies for incitation of HO development. (B) Cellular Populations present at sites of characteristic HO Development. (C) Stages of HO development: cellular infiltration, chondrogenic differentiation, and finally mineralization and ossification.

While genetic causes of HO remain exceedingly rare in the general population, (1 in a million), the incidence of traumatic HO remains significant. HO forms in up to 44% of patients undergoing hip arthroscopy or total hip replacement, in 10–20% of CNS injury, as a sequela in > 60% of blast injuries affecting the limbs and appendicular skeleton, and in up to 4% of patients suffering from > 30% TBSA burns [1,10–15]. Patients with HO are prone to severe pain, limited joint mobility and nerve entrapment, all of which can lead to a profound decrement in quality of life [15]. Multiple site-specific classification schemes exist to characterize the severity of HO formation including the Brooker, and Hastings and Graham scales, each referencing the hip and elbow joints respectively [16,17]. The former classification system groups bony lesions into four classes ascending in severity, (I-IV), inversely proportional to the distance between HO foci and bones of the hip joint; stage IV is classified as ankylosis of the hip [16]. The Hastings and Graham scale is a functional scale of three classes (I-III) oriented by functional limitation. The most severe, III, is complete ankylosis of the elbow joints, and class II is subdivided by three grades (A-C) of range of motion limitation [17].

Due to the unique endochondral pathogenesis, (ossification proceeds through the mineralization of a cartilaginous intermediate), and subsequent diagnostic limitations, treatment options remain insufficient. Current prophylactic regimens include use of Indomethacin, radiation, and bisphosphonates, each with problematic risk and side-effect profiles (Table 1). As such, surgical resection remains the therapeutic mainstay once patients become symptomatic. However, recent studies report the efficacious use of novel, specifically targeted pharmacologic agents in animal models, which present exciting solutions to many of the drawbacks of the current interventional techniques [18–20].

Table 1.

Current HO Prophylaxis.

Current treatments for Ho prophylaxis
Treatment	Population	Intervention Dosage	Frequency	Adverse Effects	Study
Indomethacin	SCI	75 mg	Once daily for 3 weeks	Gastrointestinal side effects, excessive bleeding,mental confusion	Banovac, 2001
	Orthopedic procedure	50 mg	Twice daily for 3 weeks		Romano, 2004
Celecoxib	Orthopedic procedure	200 mg	Twice daily for 3 weeks	Cardiovascular events	Yeung, 2016
Etidronate	SCI	300 mg (IV) initial and 20 mg/kg (PO) maintenance	IV for 3 days followed by PO for 6 months	Gastrointestinal symptoms, hyperphosphatemia	Banovac, 1997
	Orthopedic procedure	20 mg/kg (PO) prophylaxis	PO for 3 months		Vasileiadis, 2010
Radiation	SCI	700 cGy	Once postoperatively within 72 h of surgery	Delayed wound healing	Museler, 2017
	Orthopedic procedure	700 cGy	Once postoperatively within 72 h of surgery		Milakovic, 2015

2. Pathogenesis: the role of the inflammatory response

2.1. HO is an inflammatory process

A litany of illnesses attributable to aberrant tissue healing and regeneration in response to inflammation is well documented in the medical literature. Barret’s esophagus, formation of surgical adhesions, the fibrotic replacement of essential organs, and even atherosclerosis are ubiquitous and attributable, in large part, to severe or chronic inflammation [21,22]. One notable case study documents the cartilaginous differentiation of the serosal mesentery following abdominal surgery; clearly, tissue repair yields mutability in regards to cell and tissue fate [23]. Similarly, HO lesions are invariably preceded by inflammatory insult. In genetic cases, bone formation can present in response to unidentified trauma as patients often report prodromal symptoms classically consistent with local inflammation including pain, swelling, warmth and redness prior to overt bone formation [24]. In cases of traumatic etiologies, the extent and progression of bone formation are proportional with the severity of injury incurred, and correlate directly with factors including total body surface area (TBSA) of severe burn trauma, ventilation assistance requirement, re-operation, and sepsis [25]. In murine models of ectopic bone formation, concomitant trauma and presence of bacterial wound colonization increase the extent and incidence of bone formation [6,7]. Intriguingly, risk factors that confer a susceptibility to chronic inflammatory disorders, such as human leukocyte antigen (HLA)-B27 positivity, also confer a higher incidence for HO development in response to neurogenic traumas [26,27].

Appropriately, the most efficacious prophylactic treatments used in common practice are nonsteroidal anti-inflammatory drugs (NSAIDS) [28], and many drugs currently in development demonstrate significant anti-inflammatory properties. The local use of apyrase has been demonstrated to reduce the amount of HO development in a murine trauma model of HO development [29]. As an extracellular messenger, purinergic phosphates induce a wide-range of pro-inflammatory responses including platelet degranulation, neutrophil activation, and the promotion of chronic inflammation via the P2X and P2Y family of receptors [30]. Despite evidence suggesting purinergic surface receptors can augment osteogenic differentiation, the cleavage of ATP to dephosphorylated purines has been demonstrated to resolve local inflammation [31,32]. Treatment with apyrase likely complements the coordinated activity of cell surface enzymes CD39 and CD73, which progressively hydrolyze ATP to adenosine, thereby depleting the pro-inflammatory signal.

Multiple studies have focused on the role of BMP type I receptor inhibitors to mitigate HO formation. The BMP-receptor and downstream pathways, as will be discussed more in depth, are upregulated and causative in the formation of traumatic and genetic HO [4,18]. A structural analogue of dorsomorphin, LDN 193189, and the more recent compound LDN 212854 are specifically targeted small molecule inhibitors of the serine/threonine kinase BMP-receptors [18,33]. The LDNs have been demonstrated to not only reduce bone formation in models of sporadic and trauma induced HO, but have also been demonstrated to reduce the morbidity of chronic disease through reduction of reactive oxygen species (ROS) locally, as well as systemically reduce production of inflammatory cytokines including IL-6 [34]. The activation of BMP signaling not only drives chondrogenic and osteogenic programming, but also modulates the inflammatory setting more broadly. Taken together, the efficacy of these agents suggests that an inflammatory milieu is permissive of ectopic bone formation, and the progression of HO can be mitigated by parallel efforts to reduce inflammation at the site of injury.

Unsurprisingly, the incidence of HO formation in traumatic injury can be correlated with the concentration of cytokines and biomarkers classically associated with inflammation. Assays of wound effluent and serum for cytokines such as IL-3, and IL-12 are independently associated with HO development following severe trauma [35]. It is interesting to note that the cytokine profile is distinct between developing HO foci of traumatic and genetic etiologies [36]. The inflammatory milieu is affected by the release of local “danger signals”, (including damage and pathogen associated molecular patterns-DAMP, PAMPs) and inflammatory mediators from damaged and injured tissues that alert the host to cell death. Moreover, the introduction of microorganisms stimulates potent inflammation [37].

The quality, not just the quantity, of the inflammatory setting may also dictate the responsive tissue differentiation subsequent to severe injury. Neurogenic inflammatory peptides, including substance-P and calcitonin gene related peptide were shown to augment the recruitment and differentiation of osteoblast progenitors, and the inactivation of sensory nerve signaling was able to abrogate bone formation [38]. Coupled with the identification of contributory endoneural stem cells to stem-cell populations, the unique cytokine milieu induced by neural inflammation proves a unique contributor to HO formation [38,39].

Inhibition of substance P receptor signaling, however, was not able to mitigate HO formation in a study modeling neurogenic HO formation in wild-type mice [9]. The same study, however, revealed that depletion of macrophage populations using the bisphosphonate analogue, clodronate, was able to succeed in inhibiting HO, where specific receptor inhibition had failed. Numerous studies document the relationship between macrophages, osteoinductive cytokine production, and subsequent endochondral bone development; impairment of macrophage proliferation and function reliably reduces HO formation in animal models [9,40–42]. Given this, macrophages, as cellular effectors and orchestrators of both innate and adaptive immune systems, prove a robust target of inquiry in the connection between inflammation and pathologic differentiation.

3. Fluidity of cell fate: pathological differentiation

3.1. Questioning terminal differentiation

The determination of cell fate had long been regarded as unidirectional, definitive, and irreversible. Recent advances of studies performed in vitro, and observations made from multi-potent lineages in vivo have begun to challenge these assumptions. The discovery of reprogrammable stemness through induction of the four Yamanaka genes, Oct4, Sox2, c-Myc, and Klf4, heralded a shift in the paradigm of cellular specification and differentiation [43] [44]. In recent years the induction of pluripotency in multiple cell types including in cells of mesenchymal lineages, has become routine. The existence of novel multipotent stem cell populations isolated from bone marrow and adipose tissue revolutionized the stem cell field [45,46]. Sun et al. showed that induced pluripotent stem (iPS) cells can be generated from (adipose-derived stem cells) ASCs freshly isolated from patients [47]. Other studies have found that the surrounding niche was capable of not only supporting cellular survival, but also guided iPS cells toward an osteogenic fate, thereby allowing for functional engraftment into regenerating tissue [48]. The induction of mutability in regards to cellular differentiation underlies the multipotency of mesenchymal stem populations in the setting of trauma. Even cell types thought to be terminally differentiated have been demonstrated to contribute to HO formation in vivo [49]. In the setting of ectopic bone formation, multipotent cellular contributors have been identified from a host of tissues including marrow, muscle, adipose, and sensory nerves. While there exists no evidence that these undifferentiated cellular types are pluripotent (capable of differentiation into all terminally differentiated cells of an embryonic lineage), purported progenitor cells have been implicated in a host of fibrotic phenotypes following injury, suggesting multipotency dependent on the surrounding milieu [50,51]. Identification of individual factors that have overwhelming influence on cell fate via the promotion or prevention of mesenchymal cell differentiation toward specific lineages remains an important endeavor in today’s efforts to develop therapeutic strategies, particularly for HO where current strategies are plagued by limitations and complications.

3.2. Cell lineages contributing to HO formation

Heterotopic ossification (HO) is thought to result from inappropriate differentiation of progenitor cells that is induced by a pathological imbalance of local or systemic factors (Table 2.). In a previous study, we found that HO which formed within the soft tissue exhibited >90% Prx-Cre⁺ cells (marker of mesenchymal lineage) with Prx-Cre⁻ cells being present only in the marrow space of the mature HO or penetrating vessels of the anlagen [19]. However, contributory cellular lineages are not limited to exclusively progenitor cells of the mesenchymal lineage. While local resident mesenchymal stem cells (MSCs) are a logical candidate as HO progenitors, the contributions of endothelial, neuronal, and epithelial lineage cells to developing HO lesions have all been reported in the literature [49,52–55]. In the setting of mutable cell fate, it is likely no single cell type is predominately contributory, and likely numerous cellular lineages contribute directly to de novo bone formation.

Table 2.

Contributory Cellular Lineages in HO Development.

HO contributory cellular lineages
	Cell lineage/phenotype	Source	Study
Local	Prx-cre +	Local Mesenchymal tissue derived from lateral platemesoderm.	Agarwal, 2016
	Scx-cre +	Tendon, periosteum, fascia/connective tissue.	Dey, 2016 Agarwal, 2016
	Wntl-cre/	Endoneurium	Olmsted-Davis, 2017
	Tie-2 +/Ve-Cadherin-cre +	Vascular endothelium and endothelial progenitors	Lounev, 2009 Medici, 2012
	Glast-creERT/+	Possibly pericyte populations of the CNS.	Kan, 2013
	Mxl-cre/+	Muscle interstitium, bone marrow derived.	Dey, 2016
	PDGFR+/Sca-1 +/CD45−/CD31-	Extra-laminin skeletal muscle interstitium.	Wocszyna, 2012
	Nfatcl-cre/+	Mesodermal derived tissue	Agarwal, 2015
Circulatory	Circulating osteogenic precursors (Coll +/CD45 +/CD13 +/CD34 +/Fibronectin/CXCR4 +)	Bone marrow derived osteoprogenitors.	Suda, 2009
	Bone cartilage stromal progenitors (AlphaV+/CD105 +/Tie2−/CD45 −/Thyl−/6C3−)	Non-hematopoietic bone marrow derived.	Agarwal, 2016
	Circulating Osteogenic Precursors (CD45+/OCN+/COL1 +)	Bone marrow derived osteoprogenitors.	Egan, 2011

Theoretically, any progenitor cell population that has the potential to differentiate into the osteogenic lineage could be responsible for HO formation. Development of ectopic bone is an endochondral process, a phenomenon that occurs via multiple steps of coordinated differentiation, including the formation of a cartilaginous anlagen and subsequent vascularization, culminating in mineralization. Vascular endothelial cells initially emerged as the leading candidate for the cellular origin of heterotopic cartilage and bone in HO, with approximately 50% of the cells in HO identified to have endothelial origination based on Tie2-Cre lineage tracing and expression of various endothelial markers in these cells (Tie1, Tie2, vWF, VE-cadherin) [49,56,57]. However, in addition to endothelial cells, Tie2-Cre also labels hematopoietic lineages, thereby making it difficult to differentiate the contribution from endothelial lineage cells versus circulating populations. What is more, trauma and changes to the inflammatory milieu are sufficient to change transcriptional and immunophenotypic surface markers, further complicating discrimination of disparate cellular lineages. Indeed, Wosczyna et al. showed that the endothelial fraction of cells (Tie2⁺CD31⁺) does not participate in HO formation whereas the non-endothelial fraction of cells (Tie2⁺CD31⁻) does [50]. Furthermore, they found that a skeletal muscle resident subpopulation of Tie2⁺CD31⁻ cells (Tie2⁺ PDGFRα⁺Sca-1⁺) contributes to HO, pointing to a mesenchymal rather than an endothelial origin of these progenitors. These markers were consistent with mesenchymal progenitors implicated in fibrosis formation following skeletal muscle damage [58]. This cellular population, commonly referred to as fibro-adipogenic progenitors (FAPs), demonstrate osteogenic differentiation potential when cultured in the appropriate medium in vitro, suggesting undifferentiated mesenchymal progenitors can mediate a spectrum of fibrotic phenotypes [51].

Based on this, other groups tested additional Cre lines to cover a broader lineage range and further narrow down the potential target populations. In a study by Kan et al., FoxD1-Cre⁺ cells contributed to normal bone formation, whereas Glast-CreERT⁺ cells contributed significantly to HO at all stages [52]. The characteristic anatomic distribution and labeling pattern of Glast-CreERT⁺ cells was similar to those of pericytes contributing to scar-forming stromal cells in response to spinal cord injuries. In a more recent study, Downey et al. demonstrated that CD73⁺/CD105⁺/CD90⁻ human skeletal muscle MSCs were clonally multipotent and could give rise to all mesenchymal lineages present within HO [59]. Importantly, the models used in these previous studies were not representative of HO caused by trauma, but rather required introduction of BMP, either through exogenous administration or genetic upregulation. Therefore, our group evaluated the contribution of known bone-chondro-stromal progenitor (BCSP) cell and pericyte populations to the development of HO in the setting of trauma or constitutive BMP receptor activity in the absence of exogenous ligand [53]. We demonstrated that BCSPs (AlphaV⁺/CD105⁺/Tie2⁻/CD45⁻/CD90⁻/BP1⁻) isolated from neonatal mice or developing HO incorporated into new HO lesions upon transplantation but did not constitute a large percentage of the final HO lesion. These findings support previous studies highlighting the contribution of circulating cellular populations to HO. These cells are able to mobilize and respond to inciting trauma [60,61]. While the identification of a single HO precursor cell is enticing, the initiation and progression of HO is likely mediated by a heterogeneous composition of cells, derived from both autochthonous and circulatory reservoirs.

Clinically, HO often appears in connective tissue within myofascial planes and joints, even when inciting trauma occurs anatomically distant from bony foci. Scleraxis, a basic helix-loop-helix transcription factor, is expressed in connective tissues including tendon, ligaments, and perimysium between muscle fibers [62]. Using Scx-Cre and ScxCreERT2 for lineage tracing, we recently demonstrated that scleraxislineage restricted cells have the capacity to form HO in the settings of trauma and with hyperactive BMP receptor activity [63]. Another study identified two cell lineages that were sufficient to initiate HO, resulting from the cell-autonomous effects of ACVR1 mutations [64]. They found that an Mx1⁺ interstitial lineage in muscle gave rise to injury-dependent intramuscular HO and a Scx⁺ lineage that gave rise to spontaneous HO of tendons and ligaments [64]. These studies suggest that the phenotypic heterogeneity of HO can be attributed, in part, to the cellular lineage(s) in close proximity to the injury site.

The ability of cells to change their differentiation state in response to injury and stress may explain recent discoveries that the phenotype of cells may not always provide an accurate account of its site of origin. The phenomena of endothelial to mesenchymal transitions, and epithelial to mesenchymal transitions underlie the fluidity of cell fate, particularly in an inflammatory setting. In a recent study of ours, we demonstrated that musculoskeletal injury induces expression of CD31, VE-cadherin, and Tie2 in mesenchymal cells. Using cell transplantation assays, we also confirmed that endothelial cells isolated from uninjured muscle tissue undergoes in vivo endothelial mesenchymal transition when transplanted directly into the wound [65]. More recently, Olmsted-Davis et al. found that the endoneurium played a key functional role in HO formation using Wnt1-CreERT for lineage tracing [39]. Wnt1-Cre⁺ cells were observed in SP7⁺ osteoblasts, Sox9⁺ prechondrocytes, and UCP1⁺ brown adipocytes, suggesting the potential stem cell nature of these endoneurial cells and capability of these cells to undergo an epithelial to mesenchymal transition [39].

4. Cell signaling: pathways of differentiation, growth, and proliferation

4.1. The tumor growth factor (TGF)-β superfamily and downstream signaling

Since the discovery of the highly conserved R206H mutation in the type -I BMP receptor, ACVR1/ALK2, as the driver of FOP, the BMP signaling pathway, as well as other members of the TGF-β superfamily have been of primary interest in the study of HO (Fig. 2). The association of hyperactive BMP signaling and ectopic bone growth has prompted a great deal of research into the pathogenesis driving both genetic and traumatic HO development. The BMP family of cytokines, as its name would suggest has long been implicated in osteoinduction and in driving osteogenic differentiation [66]. The BMPs are evolutionarily conserved members of the TGF-Beta superfamily, and mediate a wide range of physiologic and developmental functions [67]. Of these proteins, BMP-2 and BMP-7 are regarded as particularly osteoinductive - the recombinant synthesis of which are increasingly employed to augment bone growth in spinal fusions and other orthopedic procedures [68]. When dimerized and bound to a cell-surface, multimeric receptor complex consisting of both type-1 and type-2 serine/threonine kinases, pro-osteogenic BMPs activate the phosphorylation of SMAD1/5 signaling proteins, and subsequent transcription of pro-chondrogenic and osteogenic genes [69,70].

Fig. 2.

Cell Signaling in tHO: (A) Cell signaling pathways implicated in cellular proliferation and differentiation responsible for HO development. (B) Diagramatic representation of cellular progression in developing HO lesions.

Recently, the work of Hatsell et al. and Hino et al. signaled a shift in the understanding of the ALK2 hyperactivity responsible for FOP [71,72]. The R206H mutation does not drive constitutive activation of the receptor, but rather confers promiscuous binding of the mutant receptor to Activin-A, another member of the TGF-β superfamily that is antagonistic to the downstream BMP pathways under physiologic conditions. In vitro experiments using iPSCs transduced with the ACVR1/ALK2 mutation confirmed the sufficiency of Activin-A to initiate osteogenic and chondrogenic programming. Interestingly, this effect was not universally applicable to all induced cellular lineages, suggesting cell type and unique epigenetic landscapes can lead to differential effects, and can help to explain the distribution of HO foci [73]. Indeed, the epigenetic landscape unique to specific cellular lineages is able to confer significant differences in transcriptional response to BMP-receptor and SMAD signaling, and underlies the characteristic distribution of HO foci in close proximity to connective tissue structures and fascial planes [74,75]. Despite this variance in cell-specific response to BMP signaling, these findings underscore a salient driver of HO pathology: an unbalanced activation of the BMP-SMAD1/5 axis resultant in ectopic and pathologic differentiation of mesenchymal tissue.

4.2. Anabolic cell signaling: role of AKT, mTOR pathways

The phosphatidylinositol-3-kinase (PI3K) and downstream mTOR/AKT signaling axis integrates growth and proliferation signals from the surrounding micro-environment to mediate appropriate cellular responses to homeostatic perturbations [76]. The downstream effects of mTOR/AKT pathway are cell-specific, but control a vast array of cellular functions including autophagy, inflammatory polarization of T-cell populations, and proliferation of fibroblasts and chondrocytes [77]. In traumatic and genetic HO models, mTOR antagonism has been documented in a number of studies to mitigate HO formation using rapamycin [19,78]. Notably, rapamycin was able to reduce bone formation despite a preservation of BMP signaling as evidenced by levels of SMAD1/5 activity [78]. Whether the effect of rapamycin can be completely attributed to the decrement in mTOR/AKT signaling, or whether the drug affects downstream pathways associated with hypoxia and subsequent chondrogenic signaling remain to be answered. Though the multifactorial effects of mTOR inhibition may occlude our understanding of how exactly the PI3K-mTOR-AKT axis mediates bone formation, rapamycin remains an efficacious and well-tolerated therapeutic in preliminary studies.

4.3. Retinoic acid receptor and differentiation

Vitamin A and its active form, all-trans retinoic acid (ATRA), bind to members of the Retinoic Acid Receptor family (RAR-α, RAR-β, RAR-γ) localized in the nucleus. After hetero-dimerization with the closely related RXR family of nuclear receptors, retinoids mediate a wide range of transcriptional activation. An excess of retinoids have been reported to mediate cortical bone loss through activation of osteoclasts, as well as increased RANKL mRNA and protein expression. For these reasons, retinoid agonists, specifically the RAR-γ agonist, palovarotene, has emerged as a viable prophylactic therapy for both genetic and traumatic HO [20,79]. A recent study has also demonstrated the synergistic use of palovarotene in combination with glucocorticoids, and has shown that the osteo-inhibitory effect of retinoids is independent of its anti-inflammatory effects [80]. In an extremity trauma model of HO involving a MRSA wound infection, palovarotene administered orally every other day for 2-weeks starting at postoperative day-1 completely inhibited ectopic bone formation, however, this treatment led to a >50% wound complication rate [79].

4.4. Endochondral bone development is intricately linked with hypoxic cell signaling

In an environment of low oxygen tension the transcription factor hypoxia inducible factor-1alpha (HIF-1α), avoids being tagged by prolyl hydroxylases and subsequent degradation, upon which the protein translocates into the nucleus, dimerizes with HIF-1β, and activates transcription of a host of downstream targets associated with angiogenesis and chondrogenic programming [81]. Multiple studies have documented the dependence of both cartilage formation and subsequent mineralization on HIF-1α driven signaling [82]. Predictably, the endochondral formation of traumatic HO is also dependent upon HIF-1α activity, and recent studies have leveraged this insight to use tissue hypoxia as a potential therapeutic target [19,83]. The recent work of Wang et al. demonstrates that hypoxia induces retained expression of the mutated ACVR1 receptor in FOP patients through increased activity of Rabaptin-5 (RABEP1) [84]. In FOP lesions, predominating hypoxia ensures sustained expression and sensitivity of resident cells to hyperactive BMP signaling.

The interactions of these complex cell-signaling pathways are not desultory; an aberration in one signaling axis invariably affects the activity of another. Fortunately, we are steadily gaining a greater understanding of this intricate network, and each pathway provides a locus at which specifically targeted and efficacious medical therapy can be developed to mitigate HO development.

5. The evolving practice of diagnosis and treatment

5.1. Development and utilization of novel diagnostic technologies

Unfortunately, in the clinical setting, definitive diagnosis of HO lesions can only be made after lesions have reached maturity, and mineralization can be recognized by radiographic imaging, usually no earlier than six weeks out from the inciting injury [85]. This delay presents multiple challenges; chief among them is the closure of a prophylactic therapeutic window before diagnosis can be confirmed. The identification of at-risk populations can vary in concordance with the inciting injury. For example, patients who undergo a total hip replacement may develop HO more often than patients who may suffer from a large burn [86]. Risk factors can be etiologically specific: the degree of spasticity in spinal cord injury (SCI) patients, need for prolonged mechanical ventilation in patients with severe burns, and surgical approach for orthopedic procedures all prove to be prominent predictive factors for HO development [11,87–89]. A number of these factors have been incorporated into scoring systems used to predict HO development with successful results in burn patients [90]. In most cases, however, these predictive factors are insufficient to dictate treatment algorithms, highlighting the need for early and accurate diagnosis before lesions are formed. The persistent inability to positively predict HO formation in at-risk patients continues to complicate the administration and efficacy of prophylactic treatment.

Currently, the gold standard of diagnosis remains computerized tomography (CT) or x-ray imaging, although other modalities, including magnetic resonance imaging (MRI) and ultrasound have been used with demonstrated efficacy, and are particularly useful in operative planning. Considering the chondrogenic pathogenesis of HO lesions, multiple novel modalities are in development for the purposes of identifying developing lesions at different stages of maturity. Recently, Raman spectroscopy has come into prominence as a viable diagnostic tool due to the ability to accurately register unique chemical signatures specific to biological macromolecules characteristic of tissue types [91,92]. Other technologies include near infrared optical imaging, single- photon emission CT, and vibrational spectroscopy [93–95]. Despite the specificity and resolution provided by Raman spectroscopy, the technology has yet to be widely adopted by the clinical community. Elsewhere in this issue, Xue et al. demonstrate the efficacy of spectroscopic ultra- sound (SUSI) in the identification and visualization of developing HO lesions [96]. SUSI is a novel technology that utilizes supra-spectral ultrasound frequency to penetrate dense tissue structures otherwise impermeable to conventional ultrasound readings. This modality offers objective readouts of tissue properties including acoustic diameter and acoustic scattering, uncorrupted by user dependent variation and biases.

5.2. Current prophylactic treatment options

Compounded by diagnostic limitations and difficult risk stratification, there is little consensus on optimal treatment strategies for HO prophylaxis. Selection of the appropriate treatment regimen is further complicated by the variability of inciting injury predisposing to ectopic bone formation. Each of etiology for traumatic HO brings a unique prevalence, risk profile, and area of formation that should be taken into consideration by treating physicians. For example, orthopedic procedures where HO can be a complication in up to 80% of patients, the balance of physiologic bone repair and prevention of excessive bone formation can be precarious. The most commonly prescribed treatment regimens feature anti-inflammatory agents (e.g. NSAIDS), localized low-dose radiation or some combination of both [28,97]. Recent studies have continued to demonstrate the efficacy of such treatments, but also highlight the risk profile inherent to each [98].

Prostaglandins, particularly PGE₂, are necessary for osteogenic differentiation of mesenchymal progenitors, and are necessary for bone growth and healing [99,100]. Logically, the suppression of prostaglandin signaling via inhibition of the cyclooxygenase (COX), family of enzymes can have ambivalent effects. Extended administration of indomethacin, a non-selective COX inhibitor, for six weeks significantly increased the rates of acetabular nonunion following fixation, despite no added reduction in HO prevention in a comparable patient group that received the drug for only a week [101]. Further studies have been dedicated to parsing the relative efficacy of specific COX-2 inhibitors over the non-specific COX inhibitors – despite theoretical benefits of specificity there are contradictory reports in the literature regarding which is superior [102–104].

Principles gleaned from studies assessing the impact of radiation on fracture healing have long been applied to the prophylactic use of radiotherapy to prevent HO. Likely due to the disproportionate effect of radiation damage on proliferative mesenchymal progenitors, radiation treatment remains a mainstay of HO prophylaxis, and is most effective when administered in close temporal proximity to the inciting injury or operation [105,106]. Nonunion following radiation administration remains a persistent concern following fracture repair [107]. Multiple prospective studies and systematic reviews demonstrate comparative efficacy of radiotherapy to NSAIDS, although differences in each case did not reach significance [108,109]. Proven efficacy of both radiation and treatment encourages clinical discretion and a consideration of case and patient specific factors. Indeed, some clinicians recommend combined treatment for optimal risk reduction [110].

Bisphosphonates are pharmacologic agents structurally similar to pyrophosphates, and are classically used as anti-resorptive agents. Despite this seemingly paradoxical method of action, first generation bisphosphonates non-selectively affect osteoblasts as well as osteoclasts. While some studies document efficacy in HO prophylaxis, others have shown an increased risk of HO formation with the use of bisphosphonates [111,112]. As such the use of bisphosphonates are discouraged in HO prophylaxis in neurogenic, burn, or orthopedic injuries.

A greater understanding of the underlying signaling pathways in the pathogenesis of HO has yielded a generation of novel therapeutic options. The LDN class of drugs (chemically optimized dorsomorphin derivatives), which target the type BMP receptors and downstream pathways, have emerged as an efficacious and specifically targeted therapy to mitigate ectopic bone formation [18]. While specific BMP-Receptor inhibitors may remain ineffective in cases of genetic HO due to unfortunate consequences of chronic use, their use may prove preferable to standard treatments when used appropriately. As discussed previously, drugs targeting signaling pathways associated with growth/proliferation, retinoid signaling, and hypoxia have all been efficacious in preventing ectopic bone growth [19,78,80].

5.3. Surgical management

Consequential to the inadequacy of pharmacologic prophylaxis, surgical resection remains the only treatment to extirpate existing HO. General indications include symptomatic disability and radiographic evidence that bony lesions have ceased to progress in serial imaging. Surgeons increasingly rely on the use of established technology, SPECT-CT prominently among them, to clarify diagnosis and refine indications for surgical resection [113]. The use of SPEC-CT allows visualization of HO lesions to demarcate proximity to, and entrapment of, vital structures, allowing for better informed decisions on when to take patients to the OR [114]. When employed with careful planning and sound surgical technique, surgery proves an efficacious solution. Yet, even meticulous surgical technique cannot abrogate the risks of wound infection, recurrent contracture, and nerve injury. Timing of surgery remains dependent on the maturation of the HO lesion, which can occur as late as 18 months in cases of neurogenic etiologies [11]. Earlier resection is favored in cases secondary to orthopedic intervention for the purposes of maintaining discrimination between ectopic bone and healing callus. As no surgery is atraumatic, resection can give rise to many of the same conditions responsible for the initial incitation of ectopic bone formation and thus can be complicated by high rates of recurrence [115]. For these reasons, adjuvant radiotherapy is often used to complement surgical resection [116].

6. Conclusions

The confluence of severe inflammation and proximity to connective and soft tissue structures manifests as a unique pathologic recapitultion of endochondral bone development with devastating consequences. HO formation in response to trauma presents a persistent clinical concern. The unique response of local and circulating cellular populations to inciting inflammation presents an opportunity to interrogate the determinants of pathologic differentiation. Fortunately, the targeted study into the nuanced mechanisms of ectopic bone formation following trauma has already begun to bear fruit in clinical practice. Our understanding of the many pathways and complex interaction that govern pathologic osteogenic differentiation has provided specifically targetable foci for HO prevention, and potentially for other illnesses defined by pathologic differentiation. With continued research into the nuances of tissue regeneration, physicians can become better equipped to tip the balances of wound healing toward the recreation of pre-injury structures and preservation of function.