At the National Institutes of Health (NIH) Consensus Conference in 2000, osteoporosis was defined as a skeletal disorder characterized by compromised bone strength that predisposes to an increased risk of fracture (1). Bone strength reflects the integration of two features: BMD and bone quality. BMD is one of the strongest risk factor for fractures and its measurement has long been used to define osteoporosis. Clinical risk factors have also been integrated with BMD measurements in an attempt to help clinicians better identify patients requiring osteoporosis therapy (2). The FRAX calculator is a user-friendly web-based tool that provides immediate quantification of risks and treatment decision making based on a very simple algorithm. However, it should always be interpreted within the clinical context as it does not take into account a number of important clinical variables.

The World Health Organization defines osteoporosis as two and a half SD below the peak bone mass [i.e., the maximum amount acquired post bone maturation around the age of 18 in women (3) and 20 in men (4), but bone growth can continue up to the age of 30]. It is expressed as grams of mineral per area or volume. On the other hand, bone quality reflects a combination of bone microarchitecture, bone turnover, and mineralization. Peak bone mass is therefore a critical parameter that will impact bone strength as the skeleton is aging. Genetic factors appear to account for over 50% of the variation in peak bone mass acquisition (5). As bone is progressively lost overtime, the higher the peak bone mass, the longer the skeleton could theoretically withstand damage. This progressive bone loss from peak bone mass occurs predominantly as a result of reduced bone formation from osteoblast (6, 7) and resultant protein composition (8) and persists for decades thereafter. Additionally, accelerated bone resorption predominates in women as a result of estrogen deficiency at the menopause but also to a lesser extent in men after the fifth decade (6). However, postmenopausal women have the ability to produce estrogens from the peripheral conversion in fat tissues of testosterone to estradiol. Adipocytes indeed express the cytochrome P450 enzyme, aromatase, which can produce estradiol from testosterone. This peripheral production of estradiol has been proposed as a protective mechanism against bone loss in overweight women (9–11).

Bone strength is highly dependent on its structural and material properties. The balance between bone formation and resorption, also called bone turnover, greatly influences the material properties of bone such as tissue mineral density and collagen cross-linking. Enhanced bone turnover, as seen with a lack of estrogen in postmenopausal women, influences the structural and material properties that lead to bone microdamage. With aging, the reduction in bone strength is further compounded by progressive muscle weakness and the increased risk for falls due to lack of balance and coordination. Maintenance of bone mineralization within a relatively narrow range is also critical to the maintenance of bone strength (12). Poorly mineralized bone loses its stiffness, whereas excessive mineralization makes bone more brittle. Bisphosphonates, the most widely used drugs to treat osteoporosis by excessive bone turnover, also lead to increased mineralization and stiffness overtime. Impairment of microdamage repair is another potential side effect of bisphosphonates since normal bone turnover replaces old bone with new bone and protects against microdamage. Long-term use of bisphosphonates has been linked to atypical fractures, and one could hypothesize that the combination of abnormal mineralization and reduced turnover may play a role in its development.

The bone microenvironment is comprised of several compartments, including hematopoietic cells, bone cells, and stromal cells (13). Bone cells, also referred to as the bone remodeling unit (BMU), are composed of bone-forming osteoblasts, bone-resorbing osteoclasts, and osteocytes embedded within the bone matrix. The BMU is also in close contact to stromal elements of the marrow and the blood vessels supply (14). Osteoclasts are of hematopoietic origin, whereas osteoblasts originate from bone marrow mesenchymal stem cells (MSCs) (11, 14, 15). One of the most interesting occurrences in this environment is the accumulation of fat cells during aging and in some pathological conditions. The functional significance of this “marrow fat (MF)” accumulation correlates strongly and inversely with bone strength (16). However, its causal relationship to bone degradation as well as its potential for therapeutic targeting in osteoporosis remains to be determined. There is indeed a significant gap of knowledge in our understanding of the mechanistic relationships between fat and bone especially during the aging process.

In humans, white adipose tissue (WAT) is principally located beneath the skin (subcutaneous fat) and around internal organs (visceral fat or abdominal fat). The main cellular component of WAT is the adipocyte but other cell types are also present, including fibroblasts, macrophages, and blood vessels. Its main function is energy storage. Adipose tissue also accounts for a significant proportion of the breast tissue and is found around other organs (such as pericardial and gonadal fat) providing protective padding. Adipocytes are also found in small amounts outside adipose tissues, including muscle, liver, pancreas, and heart, which are also referred as ectopic fat. Fat cells are also found in the bone marrow, “MF,” and have been the subject of enormous research interest to explore their relationship with the bone microenvironment.

Another form of adipose tissue is known as brown fat or brown adipose tissue (BAT) located mainly around the neck and large blood vessels of the thorax of neonates whose main function is to generate heat and protect neonates against cold (17). Recent studies indicate that BAT is also found in the neck and trunk of adults albeit in lesser amounts (18). Although this review focuses mainly on white fat, the relationship between BAT and bone will be briefly discussed.

Many studies have clearly demonstrated the positive impact of muscle strength on bone strength, but we will not cover this important area of research here. However, an interesting, but much less explored, area is the relationship between muscle fat and muscle strength and by extension its impact on bone strength. Fat accumulation in muscle may also have indirect effects on bone. Intermuscular adipose tissue accumulation occurs during aging or in pathological conditions such as Duchenne muscular dystrophy, which has been linked to decreased muscle strength, a known risk factor for osteoporosis and fractures (19). Increased intermuscular adipose tissue is associated with poor mobility (20) and increased risk of hip fractures (21). However, it is not yet known whether intermuscular adipose tissue accumulation is simply a marker of muscle dysfunction or has a direct causal effect on muscle function. The relationship between vitamin D and intermuscular adipose tissue is discussed later in this review.

As we age, the cortex of the bones become thinner encircling concomitantly larger marrow cavities filled with fat, but whether this is a result of a passive accumulation of fat as bone is lost and marrow space increases or an age-related shift in MSC differentiation with predominant adipogenesis against osteoblastogenesis is difficult to elucidate.

Meunier et al. studied 81 iliac crest biopsies from elderly women and found that bone marrow samples from women with osteoporosis had a pronounced accumulation of adipocytes, relative to levels in healthy young subjects (22). Subsequent studies showed increased bone marrow adiposity in postmenopausal women with osteoporosis and a negative association between bone–MF and rate of bone formation (23–25). Investigations using magnetic resonance imaging (MRI) have shown that the accumulation of bone–MF in the vertebral bodies of older women with low bone mass confers an additional risk for compression fracture beyond that associated with low BMD (26). Further support for this notion are data showing an association between exogenous glucocorticoid use and endogenous over production of cortisol and marked bone marrow infiltration by adipocytes with a significant increase in fracture risk (27–29).

Marrow stromal cells isolated from postmenopausal osteoporotic patients express more adipocytic differentiation markers than those with normal bone mass and are more likely to enter an adipocyte than an osteoblast differentiation program (30, 31). Fat in bone marrow may also promote bone resorption since marrow adipocytes, much like fat cells elsewhere, secrete inflammatory cytokines capable of recruiting osteoclasts (32).

Pluripotent bone marrow MSCs have the ability to become osteoblasts, chondrocytes, myocytes, or adipocytes under the influence of specific cell-derived differentiation factors (33). This process has been well demonstrated in vitro to control the fate of MSC into osteoblasts or adipocytes. This process is bidirectional and considerable plasticity has been observed both in vitro and in vivo in the ability of bone cells to become adipocytes and vice versa. Mechanical stimuli on the skeleton can also modify the differentiation of MSC into the cell lineages responsible for bone and fat formation (34–39) such that increases in bone strain add to increased osteogenic activity, whereas decreases favor the adipogenic differentiation. Lastly, the lack of estrogen in rats following oophorectomy has been reported to lead to profound fatty bone marrow infiltration, suggesting that estrogen must play an important role in regulating adipocyte recruitment (40).

Data from pathological specimens and imaging studies have consistently observed a reciprocal relationship between bone mass and increased marrow adiposity in elderly humans (22, 41–43). A recent study found that the accumulation of MF during aging is linked to increased expression of RANKL, a finding that could explain at least in part age-related bone loss (44). However, whether the relation between these two tissues in the elderly represents the clinical translation of preferential differentiation by MSC into the adipose cell lineage or is merely the unintended consequence of a passive accumulation of adipose tissue as bone is lost and marrow space increases has been a matter of considerable debate. To avoid this confounding effect, young subjects were examined and found that bone acquisition is tightly linked with decreases in marrow adiposity (16). The inverse relation between the amount of bone and MF is observed at all sites along the shaft of the bone in the young and the old regardless of age, gender, or anthropometric measures (45, 46) (Figure 1). Moreover, prospective longitudinal studies have found that bone acquisition in the appendicular skeleton of healthy young females is inversely related to changes in marrow adiposity (16). Consequently, one could make a strong argument that during the aging process, differentiation of MSCs into adipocytes is favored at the expense of osteoblasts, resulting in MF accumulation and decreased bone mass. However, this causal relationship has not yet been demonstrated.

In summary, mounting evidence supports a mechanistic relationship between MF accumulation and bone loss, pointing out the potential to target this pathway to prevent or even reverse the process of bone aging.

Postmenopausal women have the ability to produce estrogens from the peripheral conversion of testosterone to estradiol in fat tissues. Adipocytes express the cytochrome P450 enzyme, aromatase, which can produce estradiol from testosterone. This peripheral production of estradiol has been proposed as protective mechanism against bone loss in overweight women (9–11). There are also reports showing an inverse relationship between BMI and osteoclast activity in normal postmenopausal women (47) and an increase in bone resorption following weight loss (48). As discussed further in the next section, fat accumulation leads to hyperinsulinemia, which is anabolic to bone, and adipocytes produce estrogen and adiponectin, which have a positive effect on bone strength and could therefore explain this positive association observed clinically (49).

Bone mineral density measured by dual-energy X-ray absorptiometry (DXA) is positively related to body weight and BMI (49, 50), possibly because higher body weight may increase mechanical loading on the skeleton, a mechanism known to stimulate bone formation. However, DXA measurements are falsely elevated by increased body fat and therefore DXA may overestimate BMD in obese individuals (51–54). Indeed, other studies have found a strong positive association between lean mass and BMD in young women and a much weaker association between BMD and fat mass (50).

A meta-analysis indicates that a high BMI appears to protect against fractures at any site in both men and women (55). Similarly, a European study found that a higher BMI protects against vertebral fractures (56). In the study of osteoporotic fractures, body weight in the lowest quartile was found to double the risk of hip fracture (57). In contrast, other studies found that the risk of hip fractures is positively correlated with fat mass in a cohort of French women (58) and Chinese men (59). Interestingly, visceral adiposity has been linked to deterioration of bone structure and skeletal fragility (60, 61), suggesting that fat compartments may have different effects on bone strength. In summary, clinical observations linking body fat and bone strength are inconsistent, and more mechanistic studies are needed to support the purported beneficial effect of obesity on osteoporosis.

PPARγ2 is the most important regulator of adipogenesis. In vitro PPARγ2 directs the commitment of MSCs into adipocytes and inhibits their differentiation into osteoblasts (9). Ablation of the PPARγ gene leads to enhanced osteoblastogenesis of embryonic stem cells in vitro and results in enhanced bone mass in vivo and reduced bone marrow adiposity (93). At the cellular level, ex vivo examination of MSCs shows commitment toward osteoblastogenesis and reduced adipogenesis (94). Administration of rosiglitazone, a specific activator of PPARγ, in mice decreases osteoblastogenesis and enhances adipogenesis in the bone marrow (95).

The canonical Wnt/beta-catenin pathway and non-canonical Wnt signaling have been implicated in this reciprocal regulation via PPAR-gamma 2. In the canonical pathway following ligand activation, Wnt binds to a transmembrane coreceptor complex consisting of Frizzled receptors and LRP5 to stimulate bone formation (96). Although Wnt10b, Wnt 3a, and Wnt 7 can stimulate the differentiation of MSC into osteoblast while inhibiting adipogenesis (97–101), only Wnt 7 has been shown to block PPAR-gamma 2 (99). Similarly, the non-canonical ligand Wnt5a was found to induce Runx2-mediated osteoblastogenesis while simultaneously suppressing adipogenesis in bone marrow MSC through the formation of a corepressor that blocks PPAR-gamma 2 gene transcription (102). In addition, PPAR-gamma 2 acts downstream of the Wnt receptor complex to enhance the proteosomic degradation of beta catenin, thereby acting as a direct regulator of osteoblastogenesis (103).

Duque et al. provided in vitro and in vivo evidence that interferon (IFN)-gamma is a potent inducer of MSC differentiation into mature osteoblasts and a key regulator of bone formation in mice and has therefore the potential to become an efficient drug target in osteoporosis (104, 105). It was further demonstrated that IFN-gamma inhibits adipogenesis in vitro and prevents MF infiltration in oophorectomized mice in vivo (106). In addition, IFN-gamma suppresses osteoclast differentiation by interfering with RANKL signaling (107), thus acting synergistically on bone cells to enhance bone strength. As discussed earlier and independently of IFN-gamma, MF accumulation during aging is linked to increased expression of RANKL, highlighting another mechanism linking bone loss to MF (44). It remains to be established whether other proinflammatory and anti-inflammatory cytokines could affect the balance between MF and osteoblastogenesis.

Excessive production or supre-physiological administration of GC excess results in inhibition of osteoblastogenesis and accelerated adipogenesis (108) through suppression of Wnt signaling (109) and induction of PPAR-gamma 2 expression (110).

Parathyroid hormone is a major regulator of calcium and bone homeostasis, but studies on its effect of fat have been so far limited. Two epidemiological studies suggest a possible positive association between circulating levels of PTH and fat mass. The first showed that circulating PTH concentrations are directly correlated with fat mass (169), and the other showed that body weight is increased in women with primary hyperparathyroidism as compared to controls (170).