Julius Wolff’s “Law of Bone Transformation” in 1892 and Harold Frost’s “Mechanostat Theory” nearly a century later, laid the foundational framework in which bone—once thought of as an inert, unchanging tissue—functionally adapts to its environment simultaneously meeting the contradictory needs of stiffness and flexibility during impact loading, muscle contraction and joint movement (Wolff, 1986; Frost, 1987; Seeman, 2002; Frost, 2003; Jepsen, 2009). Mechanical loading as a consequence of exercise training causes varying degrees of tissue-level deformation of bone’s extracellular matrix dependent on whole bone stiffness. This deformation results in a strain stimulus which if greater than customary, and heavily influenced by osteocyte viability, may elicit one of two adaptive processes: formation modeling or targeted remodeling (Hughes and Bouxsein, 2020). Targeted remodeling, whereby osteoblast and osteoclast activity are temporally coupled with the objective of resorbing and replacing small packets of damaged tissue is important for modulating microdamage—that is, tissue deterioration at the micro- and nanostructure level (Morgan, Unnikrisnan, and Hussein 2018)—with the resultant porous space creating a transient negative bone balance that can reduce the stiffness, strength, and fatigue resistance of bone until the site is fully replenished by secondary mineralization, an extensive process that can take upwards of 30 months (Boivin et al., 2009; Plotkin, 2014; Hughes et al., 2017; Mancuso et al., 2022). Conversely, adaptive bone formation, a process of formation modeling, is the independent action of osteoblasts uncoupled from prior osteoclastic bone resorption remodeling that ultimately deposits bone at mechanically advantageous places intended to improve the morphology of the structure (Hughes et al., 2017; Hughes and Bouxsein, 2020). Further, adaptive bone formation, bypassing the resorption period with a median duration of 30–40 days (Eriksen, 2010), is temporally more expedient than is remodeling, foregoing the transient porosity and subsequent increased risk of bone stress injury.

Reports from in vivo loading models dating back to the 1960s have established the ability to capture these complex physiologic pathways in animal species (Frost, 2003). Such studies have shown improvement in bone mineral density, trabecular microstructure, and bone geometry in as little as 4 weeks (Mustafy et al., 2019). In another, after 8 weeks of 10, 45 cm freefall drops, bone volume fraction, trabecular and cortical thickness improved in rat ulnas, with greater adaptation resulting from impact than from calcium supplementation (Welch et al., 2008). A study performed over 16 weeks subjected rats to 360 load cycles per day, resulting in significant increases in biomechanical resilience, specifically ultimate force and energy to failure, occurring despite only modest gains in areal bone mineral density (aBMD) and bone mineral content (BMC) (Robling and Turner, 2009). These results have provided the scope in which exercise interventions in human research commonly range in duration from just 4–12 weeks (Linke, Gallo, and Norman, 2011; Hecksteden et al., 2018; Ashton et al., 2020). However, due to rodent cortical bone lacking well-developed haversian remodeling activity (Bentolila et al., 1998)—and the majority of animal models designed for compact bone—resorption data from animal models is limited (Robling and Turner, 2009). Capturing modeling and remodeling in response to exercise in humans is even more challenging, with recent improvements as higher resolution technologies become available.

High-Resolution Peripheral Quantitative Computed Tomography (HR-pQCT) has the potential to capture small and subtle changes in human bone (Mancuso and Troy, 2020). HR-pQCT allows for the assessment of bone compartments by assessing three-dimensional volumetric bone mineral density (vBMD) and structural measurements of total, trabecular and cortical bone due to its isotropic voxel size of 61 μm (XtremeCT II) or 82 μm (XtremeCT) (Whittier et al., 2020). In addition to differentiating between skeletal compartments, advanced HR-pQCT software allows for microfinite element analysis (FEA), capable of estimating the mechanical competence of bone (Boutroy et al., 2007). The ability to discern small changes in bone microarchitecture over a short time has the unequivocal potential to determine intervention efficacy within healthy populations and preemptively identify those at unsuspecting risk of injury or fracture. Additionally, early detection of even modest changes in bone microstructure and morphology contribute to large changes in bone strength, even in lieu of changes to bone density or mass (Turner and Robling, 2003; Bass, Eser, and Daly, 2005). Therefore, the purpose of this review is to synthesize existing training-induced HR-pQCT data in three distinct, healthy, adult populations, identify similarities and contrasts with prior modalities and cumulatively interpret results within the scope of bone functional adaptation.

Eligibility was confirmed by full-text screening of papers written in English that utilized the HR-pQCT within a healthy, non-diseased, adult (>18 years) population. Reviews, case-reports, conference proceedings and editorials were excluded. Only studies that included an exercise or training intervention (i.e., non-observational) without the use of pharmacological drugs or nutritional supplementation were included. Consequently, 10 studies were included and were further stratified into 3 groups: 1) Military interventions without supplementation (4 studies), 2) Non-osteoporotic, healthy older adults interventions (3 studies), and 2) General population interventions (3 studies). All studies featured here are also presented in Tables 1, 2.

Perhaps the best evidence we have to support the ability of the HR-pQCT to detect short-term yet meaningful changes in skeletal microarchitecture lies in studies of basic military training (BCT), as these 8–44 weeks programs represents a brief period of often unaccustomed physical activity. Prior to the advent of the HR-pQCT, Peripheral Quantitative Computed Tomography (pQCT) studies demonstrated modest anabolic changes in bone density, geometry and strength in response to short-term (<10 weeks) exercise interventions in male cadets (Izard et al., 2016) and in density, content and cortical microarchitecture in female cadets (Gaffney-Stomberg et al., 2014), though others, demonstrated no changes (Baker et al., 2022). However, pQCT studies have also consistently reported that certain geometric phenotypes (smaller/narrower bone, lower cross-sectional area) predispose military cadets to bone stress injury (Jepsen et al., 2011; Wentz et al., 2011; Davey et al., 2015; O’Leary et al., 2021), though some results have been inconsistent in women(Moran et al., 2008; Cosman et al., 2013). pQCT holds some advantages over HR-pQCT, specifically capturing midshaft and proximal sites and the ability to measure muscle cross sectional area. However, pQCT suffers from partial volume artifacts and lacks the necessary resolution (170 μm) to examine bone microarchitecture (Nishiyama and Shane, 2013). While limited, current HR-pQCT evidence from field training environments demonstrates that short periods of military training (<13 weeks) can elicit favorable adaptations in the tibia, including increased density, size and strength in both women (Hughes et al., 2018) and men (O’Leary et al., 2019a).

In just 8 weeks of Army BCT, 91 female cadets (21.5 ± 3.3 years) experienced a significant increase of approximately 2% in total and trabecular vBMD and trabecular BV/TV at the tibial metaphysis, as well as 1.13%–1.21% increases in trabecular number and thickness (Hughes et al., 2018). Changes however were less robust at the cortical compartment as cortical thickness increased by almost 1% while cortical vBMD and TMD decreased in the tibial metaphysis and diaphysis (Hughes et al., 2018). Stiffness and failure load increased at both sites, conferring much larger magnitude changes at the metaphysis (2.41%–2.45% versus 0.21%–0.77% at diaphysis), though both changes were significant. In 13 weeks of British Army BCT, 43 young men (21 ± 3 years) experienced a significant increase in total (1.7%–2.02%), trabecular (1.3%–1.9%) and cortical density (0.56%–0.9%) as well as Tb BV/TV (1.05%–2.1%), cortical area (2.7%) and cortical thickness (3.1%–3.8%) at the distal tibia. Trabecular area (−0.4%) significantly declined while trabecular number, thickness, spacing, cortical perimeter and porosity did not significantly change (O’Leary et al., 2019a). Longer training (44 weeks) in female Officer cadets demonstrated greater changes in trabecular and cortical microarchitecture at the metaphysis than did shorter training, demonstrating a magnitude of change 1.5–3 fold higher than any previous military study, with the largest changes occurring at the metaphysis in trabecular vBMD (3%), trabecular number (3.5%), cortical thickness (4%) and cortical area (4.8%) (O’Leary et al., 2021).

BCT environments elicit mild energy deficits (Adler et al., 2015; O’Leary et al., 2021). Alternatively, longitudinal assessment of bone microarchitecture under conditions of severe physiological stress are limited. One such example included six healthy British servicewomen (32 ± 3 years) traversing the Antarctic, scanned via HR-pQCT approximately 1 month before the expedition and 2 weeks afterwards. As a result, there were no changes at the tibial metaphysis or diaphysis. The authors reported significant declines in axial aBMD ranging from −2.6% to −5% (O’Leary et al., 2019b) with concurrent preservation of the appendicular skeleton and bone biomarkers indicative of remodeling (O’Leary et al., 2019). Bone resorption, specifically, has been shown to increase with military training (Evans et al., 2008; Lutz et al., 2012; Hughes et al., 2014). HR-pQCT, by neither capturing discernable gains in vBMD nor improvements in microarchitecture, indicated that intense physical activity conducted under multiple stressors (i.e., prolonged severe exercise stress and reduced energy availability) does not favor anabolic formation modeling in women. Aforementioned changes in vBMD and microarchitecture over a short period of time are highly indicative of bone formation modeling, or de novo bone formation (Hughes et al., 2017; Hughes and Bouxsein, 2020) on existing surfaces, without prior resorption. Training-induced decreases in cortical vBMD at both sites also occurred, suggestive of potentially concurrent intracortical targeted remodeling. This suggests that bones can adapt in structure and strength to even brief periods of heightened physical activity in healthy, young adult populations. As many of the military trainees are beyond the age of peak height velocity and residual bone deposition due to growth, these HR-pQCT in military trainees offer evidence that even the mature young adult skeleton retains mechanosensitivity.

Following short-term training (8–13 weeks) robust changes are observed during military training with sex-specific differences in bone adaptation. Military initial level training involves a drastic and sudden increase in both volume and frequency of weight bearing activities such as unaccustomed load carriage and military drill, during which the lower extremities experience intense loading (O’Leary et al., 2021). Greater sensitivity to heightened mechanical loading consistently occurred distally, despite animal studies demonstrating that long bone adaptations in response to mechanical loading tends to occur at the midshaft and more proximally (Warden et al., 2004; Wallace et al., 2007). This discrepancy could be the result of limited scan breadth on HR-pQCT lending to best capturing only the distal appendicular skeleton (Lieberman et al., 2003; Pearson and Lieberman, 2004). However, the second generation HR-pQCT allows for expansion of the diaphyseal region to include up to 66% of total tibial length. Although to date, no training-specific studies have utilized this methodology, future longitudinal capture at this site will allow for further elucidation of cortical bone, muscle and fat in response to exercise or training (Whittier et al., 2020). Longer training periods in female Officer cadets demonstrated greater changes in trabecular and cortical microarchitecture at the metaphysis than did shorter training, a magnitude of change 1.5-3 fold higher than any prior military study (O’Leary et al., 2021). Maintenance of whole body aBMD in conjunction with these high magnitude adaptations implies that unaccustomed mechanical loading was protective against periodic low energy availability(Gifford et al., 2021) despite significant changes in total body, fat and lean mass(O’Leary et al., 2021). Further research is needed to further elucidate factors promoting versus preventing adaptation, from which we’ll discover new risk factors (i.e., energy availability, psychological stress, sleep, NSAIDs) (Hughes et al., 2022) for bone stress injury, stress fracture and bone fragility.

Within the general population, healthy adult women undergoing 12 months of compressive loading of the forearm showed significant increases in BMC associated with strain ranges experienced during activities of daily living (Troy et al., 2020). This finding conflicts with the previously held belief that extremely high strain magnitudes were necessary to elicit an adaptive response, specifically in women (Martyn-St James and Carroll, 2010). In the absence of loading, men demonstrated a pattern of bone recovery after disuse most pronounced at the distal tibia with recovery of cortical bone (namely, Ct.Ar and Ct.Th) beginning as early as 3 days after bedrest (Belavý, 2011). Women, however, demonstrated the greatest loss in Ct.Th at the distal tibia, despite additional protein supplementation (Armbrecht et al., 2011). These findings demonstrate sex-specific differences in skeletal recovery following disuse, such that men require substantially less time than do women for recovery of bone structure and density (Armbrecht et al., 2011).

Drawing sex-specific conclusions in aging populations remains difficult, as few HR-pQCT studies exist with healthy, aging populations and those that do, include exclusively women. However, existing research demonstrates that bone mechanosensitivity remains viable in women well into their 6th decade of life with changes favoring the distal trabecular compartment. Cumulatively, these results demonstrate that age may not be the decisive factor in whether or not an adaptive response occurs, but instead the length of the intervention (>6 months), type of stimulus (multidirectional, high impact) and granularity of the imaging technique (regional analysis). There remains work to be done to establish the effectiveness of exercise not just in preventing bone loss but encouraging anabolism into old age. The exercise-specific characteristics and environmental considerations that will promote these processes have yet to be fully elucidated and captured with HR-pQCT technology.

Introduction of the HR-pQCT has provided researchers the ability to investigate bone microarchitecture in vivo and quantify densitometric, morphological and geometric adaptation in response to exercise and training exposure. While pharmaceutical interventions have shown efficacy, they are often contraindicated until late in life. Non-pharmacological, non-invasive training intervention studies provide insight into the optimal exercise program to enhance bone strength and reduce susceptibility to bone injury, stress and osteoporotic fracture. Ultimately, the utilization of highly sensitive bone imaging modalities like the HR-pQCT maximizes the ability to detect even modest change that may, in turn, manifest into substantial effects over longer periods of time. Promising horizons are ahead for the role of exercise to promote anabolism and bone strength across the lifespan.