Disuse Osteoporosis: Clinical and Mechanistic Insights

Skeletal integrity is maintained by the process of bone remodeling, i.e., the balanced removal of old bone matrix by bone-resorbing osteoclasts and the deposition of new bone tissue by bone-forming osteoblasts [1]. Since the early work of Julius Wolff in 1892 [2], it has been well established that the skeleton represents a dynamic tissue undergoing adaptive changes in response to loading to better withstand these loads (“Wolff’s law”). There are several clinical examples of the anabolic response of bone to loading, one of them being the observation that the bone mass is markedly higher in the dominant arm than in the nondominant arm in tennis players [3]. In recent years, a number of essential molecular load-sensation pathways to build the optimal amount of bone required for adequate stability have been uncovered. Osteocytes, which represent terminally differentiated osteoblasts embedded into the bone matrix, have gained research attention due to their ability to sense loading and translate mechanical signals into biochemical signals to orchestrate bone remodeling [4]. The discovery of osteocyte-secreted proteins such as sclerostin has already led to new osteoporosis treatments [5, 6], and more recently, other molecular osteocytic and nonosteocytic proteins and nanostructures involved in the process of mechanosensation, such as the calcium channel Piezo1, were revealed [7‐9].

Clinically, disuse represents one of the major risk factors for osteoporosis, a widespread, multifactorial disorder characterized by low bone mass, impaired bone quality and increased risk of fragility fractures [10]. It has been shown that disuse conditions are characterized by an unfavorable combination of high bone resorption and low bone formation [11], leading to immediate bone loss and ultimately to osteoporosis with increased fracture risk [12]. In contrast, skeletal loading leads to an anabolic bone response in terms of increased bone formation [13]. Physical activity represents an effective option to increase bone mineral density (BMD) [14], which is the reason that adequate exercise constitutes the most natural basic treatment for patients with osteoporosis.

The term “disuse osteoporosis” comprises various clinical situations of mechanical unloading or pathological immobilization associated with bone loss, including spinal cord injury [15], neuromuscular diseases [16], bed rest [17], spaceflight [18] and others. Relative unloading has also been discussed as an important factor contributing to the relevant bone loss observed in obese patients after bariatric surgery [19]. Bone loss during disuse or unloading occurs due to a lack of movement and muscular insertions, but it is assumed that lower mechanical stimulation in connection with non-weight-bearing, and especially microgravity, has additional negative effects on skeletal integrity. This is also the reason why bone microstructure studies often compare weight-bearing bones (such as the tibia) with non-weight-bearing bones (such as the radius). The interaction of muscle and bone (“crosstalk”) has opened up a whole new field of science, including both clinical studies on the bone status in sarcopenia in the elderly [20] and, at the basic research level, studies on the interaction of muscle and bone and the role of osteocytes in this relationship [21] (not discussed in detail in this review). This review will focus on recent clinical and molecular advances in understanding the specific effects of unloading on the skeleton. In particular, microstructural changes, mechanistic concepts and options for the treatment of disuse osteoporosis will be discussed.

Disuse-associated bone loss has been outlined in many studies, which documented the magnitude of bone loss in a variety of disuse conditions early on. Dual-energy X-ray absorptiometry (DXA), regularly performed for the proximal femur and lumbar spine, represents the most established method to quantify the areal BMD and to determine the diagnosis of osteoporosis or low bone mass based on T- and Z-scores, respectively. DXA studies have been widely performed in disuse conditions, with both cross-sectional and longitudinal study designs.

To obtain more detailed data on the bone loss associated with disuse beyond DXA-based observations, volumetric and microstructural approaches, primarily peripheral quantitative computed tomography (pQCT) and, more recently, high-resolution peripheral quantitative computed tomography (HR-pQCT), have been introduced. The bone microstructural changes following immobilization have also been discussed in review articles [57], including those specifically discussing bone loss patterns in patients with spinal cord injury [58] and other neuromuscular disorders [59].

Previous pQCT studies with cross-sectional study designs have demonstrated negative associations between the duration of disuse and bone microstructural parameters in patients with spinal cord injury, where steady states were reached after 3–8 years and the total amount of bone loss ranged from 25–35% (diaphyses) to 50–60% (epiphyses) [60]. Only a few studies with longitudinal approaches, i.e., repeated measurements of the same parameters and in the same patients at standardized time intervals after disuse onset, are available. A selection of relevant longitudinal pQCT and HR-pQCT studies is presented in Table 1. Simulated (i.e., bed rest) or real microgravity studies are typically performed in young and healthy individuals. These conditions have been shown to lead to a pattern of both trabecular and cortical bone loss [37, 57]. During bed rest, it was reported that the relative bone loss in the tibia was larger in the cortical than the trabecular compartment [61]. Furthermore, relevant site specificity was demonstrated, with all tibial regions (epiphyseal to diaphyseal) depicting cortical loss, while considerable trabecular loss could only be detected in the proximal tibia epiphysis. However, a similar team of authors also observed greater trabecular loss in the distal tibial epiphysis in another bed rest study [62]. Greater trabecular than cortical loss was also detected after acute spinal cord injury [29]. More recently, HR-pQCT bed rest studies have confirmed that bone loss in both male and female individuals occurs primarily in weight-bearing bones such as the tibia [63, 64]. These studies demonstrated combined cortical and trabecular bone loss during bed rest.

During spaceflight, cortical and trabecular bone loss was observed in the tibia [65]. Notably, trabecular bone showed both more severe and earlier bone loss than cortical bone, although a high degree of variability among the subjects was noted. There have also been QCT studies with measurements of the lumbar spine and hip, in which trabecular and cortical bone loss has been documented [37]. A recent HR-pQCT study in subjects before and after spaceflight showed that the loss of cortical bone was not only characterized by cortical thinning but also primarily by an increase in cortical porosity of 15% at the distal tibia [38]. Interestingly, a similar increase in cortical porosity (+16.1%) was also observed in the affected limbs of patients after 6 weeks of non-weight-bearing following surgery [41]. In summary, both trabecular and cortical compartments are affected by disuse, and while site and compartment differences exist among the disuse conditions, the most pronounced bone loss seems to occur in the epiphyses, with cortical porosity being the microstructural parameter that has shown the highest increases.

The recovery of bone mass in response to physical exercise programs, pharmacotherapy or both has been studied across different disuse and unloading models. Indeed, the most natural treatment for disuse osteoporosis is physical exercise or remobilization (loading) of the affected bones, although some of the disuse conditions per se do not allow remobilization. A recent systematic review found that the favorable effect of exercise on BMD in postmenopausal women is largely independent of the type of exercise [66]. Experiments on recovery from disuse conditions (mostly bed rest) often compare a study group with different exercise programs during or after immobilization with a control group (i.e., the study group without any additional exercise program). It is interesting to note that bed rest-induced bone loss continues in the first days/weeks after reambulation [61, 67]. In different control groups, the bone mass did not fully recover within the same time that the individuals were immobilized and, in most studies, even after long-term follow-up [32, 67] (Table 1). For instance, a remaining significant reduction in the tibial epiphyseal bone mineral content (BMC) was noted at approximately one year after reambulation [67]. While HR-pQCT bed rest studies have indicated a recovery of most microstructural parameters (including the cortical thickness or trabecular bone volume fraction) after mid-term follow-up, the trabecular number in the distal tibia remained significantly reduced after two years of follow-up in both male and female subjects [63, 64]. In individuals returning from space, the tibial cortical density recovered, while the cortical porosity or trabecular bone did not fully recover during the year after landing [38]. Increased cortical porosity after 6 weeks of non-weight-bearing did also not recover after 13 weeks of full weight-bearing [41].

Regarding the optimal exercise program, various studies with different protocols have been conducted, from which diverging findings regarding their success could be derived. In individuals with spinal cord injury, an overall positive and site-specific effect of functional electrical stimulation-induced cycling exercise on bone mass has been observed [68‐70]. These studies also suggested a dose-dependent effect, with the best results after high-volume training [68]. To counteract bone loss during bed rest, the addition of whole-body vibration to high-load resistive exercise has been found to be effective in male subjects [67]. HR-pQCT studies have further confirmed these findings, with the combination of vibration and resistive exercise showing overall better preservation of cortical and trabecular parameters than resistive exercise alone [71]. In women, a previous pQCT study found that a combination of resistance and aerobic training did not provide any benefit compared to the control protocol [32]. In another study, this was also confirmed by HR-pQCT, where combined resistive and endurance exercise countermeasures were ineffective in preventing skeletal deterioration during disuse in women [64]. Such findings suggest a sex-specific effect in the skeletal response to loading, and the difference in the loading response may be attributed to sex hormones (estrogens and androgens) and their receptors [72]; however, the exact mechanisms are complex and remain partly unexplored. Of note, previous studies in non-immobilized patients examining the treatment effects of whole-body vibration on BMD have shown conflicting results with both positive and no effects [73, 74].

The fact that increased bone resorption plays a primary role in the development of disuse osteoporosis is supported by the findings of several previous therapeutic studies, in which antiresorptive treatment with bisphosphonates or the receptor activator of NF-κB ligand (RANKL) antibody denosumab attenuated bone loss [44, 75]. For example, intravenous bisphosphonates were partly effective in preserving bone mass in male subjects during bed rest [76]. Furthermore, denosumab treatment led to preserved bone mass in the knee region in patients with subacute spinal cord injury [75]. Compatible with the observation that chronic unloading also leads to low bone formation, the osteoanabolic agent teriparatide led to a solid increase in BMD in individuals with chronic spinal cord injury [77]. Interestingly, this increase was not augmented by additional vibration stimulation [77]. A previous study in postmenopausal, non-immobilized, osteoporotic women demonstrated a clinically relevant treatment effect in lumbar spine BMD with twelve months of whole-body vibration exercise and teriparatide compared to teriparatide alone [78].

Taken together, the treatment of disuse osteoporosis by combining bone-specific medications with physical exercise appears to be effective in at least partially counteracting the observed bone loss. In particular, the prompt increase in bone resorption markers after the onset of disuse [24] argues for the early use of anti-resorptive drugs such as denosumab. Individual exercise programs using vibration or electrical stimulation have shown positive, but sometimes inconsistent, results. The combination of bone anabolic drugs with physical exercise protocols appears to be promising but remains largely unexplored in conditions of disuse osteoporosis.

To decipher the mechanisms of disuse-induced bone loss, loading and unloading conditions have been modeled in a variety of human and animal studies. In one of our previous experimental human cadaveric studies analyzing the micromorphology of cortical bone, we found increased cortical porosity in long-term immobilization compared to controls and postmenopausal osteoporosis [79] (Fig. 3a), which supports previous clinical findings of a primary effect on cortical porosity after disuse [38]. Cortical deterioration was associated with lower osteocyte viability and impaired osteocyte connectivity [79] (Fig. 3b).

Regarding the evaluation of recovery or therapeutic effects in disuse osteoporosis, there have been many experimental studies taking advantage of animal models. One such study reported that different types of exercises (e.g., swimming, jumping, or vibration) applied to rats with immobilization-induced bone loss were equally efficient in preventing bone loss [80], thereby confirming clinical observations. Disuse in rats attenuated the bone anabolic response to teriparatide treatment [81]. In another experimental study in dogs, the cortical bone loss observed in disuse osteoporosis could only partially be counteracted by bisphosphonates, which indicates that the strong osteoclastogenic stimulus of disuse may not fully be antagonized by antiresorptive agents [82]. Together, these results suggest that mechanical loading is of critical importance in optimizing the treatment effects of bone-specific agents. Of note, an emerging model to study bone and its adaption to physical activity is the zebrafish [83]. Modification of the physical exercise intensity in a swim tunnel experiment led to rapid bone formation and increased bone volume and mineralization [84].

There have been a number of studies in rodents focusing on mechanical loading, bone remodeling and especially osteocyte mechanobiology. In mice, spaceflight led to both trabecular and cortical deterioration along with increased numbers of cortical empty osteocyte lacunae [85]. Early studies found evidence of osteocytic changes following immobilization, including increased osteocyte apoptosis [86]. More recently, it was demonstrated that osteocyte apoptosis following weightlessness (i.e., unloading by tail suspension) in mice triggers osteocytic RANKL production along with activation of both cortical and trabecular bone resorption [87]. While osteocyte apoptosis preceded the recruitment of osteoclastic cells and bone loss [88], another study by the same group found that inhibition of osteocyte apoptosis did not prevent unloading-induced bone loss although osteocytic RANKL production was sufficiently inhibited [89]. Tatsumi et al. developed a transgenic mouse model with the ablation of osteocytes through the injection of diphtheria toxin [90]. Importantly, these mice did not develop unloading-induced bone loss, underlining the osteocyte’s function in sensing mechanical loading and in the development of disuse osteoporosis. It is interesting to note that other models, including Botox-induced immobilization, were shown to lead to significant changes in bone mechanical and structural properties without affecting cortical osteocyte lacunar properties in rats [91]. A reason for these diverging findings, and especially the lack of osteocytic changes in the latter work, may be due to the almost nonexistent intracortical remodeling in the rats studied as well as the fact that osteocyte apoptosis was not studied [91]. Overall, it is likely that certain factors influence how osteocytes respond to unloading, for example, cortical vs. trabecular bone compartments, local loading history, and remodeling rates.

In the search for molecular mechanisms involved in the pathogenesis of disuse osteoporosis, several other important signaling pathways and proteins involved in the process of mechanically driven bone remodeling have been discovered (Fig. 4). Indeed, matrix-embedded osteocytes represent the skeletal cell type that most likely plays a major role during mechanotransduction, i.e., the process of converting external mechanical forces into biochemical responses. This is achieved by sensing local mechanical signals and responding to these signals both directly and indirectly [92]. Notably, both mechanically induced bone formation and disuse-induced bone loss and skeletal fragility may be mediated by osteocytes. Osteocytes are embedded in a lacuno-canalicular system (LCS), which becomes evident in acid-etched bone with subsequent scanning electron microscopy. Inside the LCS reside the osteocytes and the dendritic processes that originate from the osteocyte cell body. The osteocyte-secreted protein sclerostin was originally discovered through the identification of loss-of-function mutations in the SOST gene in individuals with the genetic high-bone-mass disorder sclerosteosis (van Buchem disease) [93]. Further research has demonstrated that increased sclerostin expression is primarily caused by mechanical unloading in mice, which leads to low bone formation [94]. While in vitro studies have revealed that disuse promotes osteocyte apoptosis and mechanical stimulation by fluid shear stress promotes osteocyte survival [95], it has further been demonstrated that increased sclerostin expression is specifically caused by unloading in osteocytic cells [96]. Translating these findings into therapeutic concepts, it is most likely that the recently approved sclerostin antibody romosozumab [6, 97] is especially effective in disuse osteoporosis, but the efficacy of romosozumab in association with conditions of disuse osteoporosis has to be investigated in future clinical studies.

The mechanosensitive function of the skeleton was recently evidenced by the identification of mechanically activated ion channels of the Piezo family, specifically Piezo1 in osteoblasts and osteocytes [7, 8]. Interestingly, the deletion of Piezo1 reduced the expression of Wnt1 [7], which is a key regulator of bone formation [98]. By a systematic analysis of mice with inactivated Piezo proteins in different skeletal cell populations, our group confirmed that Piezo1 acts as a mechanosensor in osteocytes and demonstrated that it represents an essential osteogenic differentiation factor during endochondral bone formation [9]. The latter finding supports the assumption that various cell types are involved in mechanotransduction, which is highly relevant not only for bone remodeling but also for developmental and regenerative processes [99]. Together, these collective data suggest that Piezo1 plays a crucial role in mechanosensation and that Piezo channels could be a novel therapeutic target for osteoanabolic treatment in mechanical unloading-induced bone loss.

Another molecular example of musculoskeletal mechanotransduction, more specifically the close interaction between muscle and bone during this process, is the identification of the myokine irisin, a protein that is derived from muscle in response to exercise. Irisin injection increased cortical bone mass in mice, partly through suppression of sclerostin expression [100]. In this way, irisin prevented not only muscular atrophy but also the development of disuse osteoporosis accompanied by decreased osteocyte apoptosis [101]. In human subjects, circulating irisin levels were associated with the risk of osteoporotic fracture [102]. Beyond this associative evidence, there is currently no strong clinical evidence demonstrating that muscle or bone modulate each other’s tissue response to load changes.

Finally, the molecular mechanosensors in osteocytes are the subject of ongoing research, as recently discussed in a detailed review article [92]. Briefly summarized (and as partly outlined above), osteocytic mechanosensors include the osteocyte cytoskeleton and dendritic processes within the lacuno-canalicular system (LCS), as well as ion channels (e.g., Piezo1). However, the extracellular matrix (ECM), primary cilia, integrin-based focal adhesions and connexin-based intercellular junctions have also been identified to contribute to mechanotransduction in osteocytes. During skeletal development and mechanical stimulation, there is interactive communication between osteocytes and the ECM, which is mainly composed of proteoglycans, glycoproteins, and hyaluronic acid. One example highlighting the role of the ECM during the osteocyte mechanotransduction process is perlecan (HSPG2), a proteoglycan that is localized in the pericellular space of osteocytes within the LCS [103]. Importantly, perlecan-deficient mice showed dysfunctional bone formation in response to loading [104]. Primary cilia are solitary, rigid structures that span from the cell body into the extracellular space [105]. Interestingly, primary cilia translate extracellular mechanical stimuli into cellular responses via an ion-channel-independent mechanism [106]. In addition to these important structures responsible for mechanotransduction in osteocytes, integrin-based focal adhesions play an important role in this complex process. Integrins are transmembrane receptors with an extracellular and a cytoplasmic domain composed of an α and β subunit. Both in vitro and in vivo studies have revealed the involvement of the different subunits during mechanotransduction [107]. Finally, cellular communication via hemichannels and gap junctions, which are composed of the protein connexin, has been found to be essential during mechanotransduction. While multiple types of connexins are expressed in bone cells, Cx43 is the major candidate for mechanotransduction in osteocytes. The cell-specific deletion of Cx43 in osteoblasts and osteocytes led to an attenuated response to both loading and unloading [108, 109].

Although a number of fundamental questions concerning the clinical, molecular and therapeutic aspects of disuse osteoporosis remain unanswered, the combined effort of both clinical and basic research has revealed the microstructural and mechanistic basis of the profound negative effects of disuse on skeletal integrity. Clinical studies with real or modeled unloading conditions, including spinal cord injury, limb suspension, bed rest and microgravity (spaceflight), have demonstrated that early bone loss is characterized by a relative increase in bone resorption compared to decreased bone formation. As a consequence, combined cortical and trabecular microstructural deterioration was noted in most studies, with subtle disease-, sex-, site- and compartment-specific differences that can be accurately imaged using high-resolution CT-based techniques such as HR-pQCT. In a broader sense, disuse osteoporosis also includes clinical conditions of bone loss in response to local skeletal unloading, such as after surgery or around implants. Furthermore, full and timely recovery from bone loss may not be achieved by loading, but bone loss may be counteracted by high-frequency physical exercise programs and bone-specific agents such as bisphosphonates/denosumab and teriparatide. The use of animal models has allowed the cellular and molecular mechanisms of disuse-induced bone loss to be elucidated. Matrix-embedded osteocytes have been identified as the primary mechanoresponsive cell type translating mechanical into biochemical signals to orchestrate mechanically induced bone formation and remodeling, which is in fact a regulatory process with high complexity also involving neighboring bone cells. Osteocyte-specific proteins such as sclerostin and Piezo1 are involved in this process of mechanotransduction, and new therapeutic options have already been introduced; however, their efficacy needs to be clarified in further experiments and clinical trials acknowledging the mechanical environment.