Causes of Age-Related Bone Loss and Bone Fragility: An Alternative View

WE CONGRATULATE Lanyon and Skerry1 for attempting to address the question, “What are the mechanically derived signals that tell the cells of bone how to fashion a mineralized skeleton that can do its job?” A careful application of biomechanical principles to bone biology is welcome and long overdue. In the preceding article, the investigators present their analysis based on a conference on “Osteoporosis as a Failure of Bone's Adaptation to Functional Load Bearing” held in 1999. They found that a consensus document from this meeting was not feasible. Indeed, one of us (L.G.R.) presented an overview with the title “From the Evidence Presented at This Meeting, Mechanically Adaptive Remodeling Is an Important Influence on Bone Architecture but Is Not of Prime Importance in the Etiology of Osteoporosis.” Our position is that this statement is most consistent with available data. We will present some of the arguments for viewing osteoporosis as a multifactorial disease or group of diseases and review some of the pathogenetic mechanisms. Lanyon and Skerry1 propose that osteoporosis is the result of a maladaptation to loading. They suggest that estrogen deficiency or perhaps a reduction in estrogen receptor numbers impairs the detection of microstrain. An adaptation follows in which bone resorption exceeds formation so that “unneeded” bone is removed until microstrains increase in the remaining bone and switch off bone resorption. This thought-provoking “mechanostat” theory originally proposed by Harold Frost remains a challenging but unproved hypothesis. Lanyon and Skerry1 also state that osteoporosis has … “received considerable attention as to its epidemiology, genetics, endocrinology … ” but that this has “not led to an understanding of the condition's actual cause ….” However, in our opinion, there is no single “actual cause” of bone fragility. Osteoporosis or low bone mineral density (BMD) or “skeletal fragility” is not a single disease with a single cause like tuberculosis. No single pathogenetic mechanism can explain all the observations in patients with fractures. There are many candidates including not only maladaptation to loading, but also cellular effects of estrogen deficiency, calcium, and vitamin D deficiency; altered parathyroid hormone (PTH) secretion or sensitivity; and abnormal response to or production of local regulators or genetic defects. In fact, all of these are likely to be important. What then is the evidence to support our view that the structural basis of bone fragility is heterogenous and that this structural heterogeneity has many causes operating throughout life? The offspring and other first-degree relatives of women and men with fractures have deficits in bone mass (relative to their age-matched peers) that are about half the deficit in the patients with fractures.2, 3 This is consistent with the genetic hypothesis, the notion that the magnitude of the familial trait resemblance in first-degree relatives (r ∼ 0.5) is the result of sharing half their genes. Thus, regional-specific deficits in bone mass in patients with subsequent fractures may partly originate during growth. More rapid appendicular growth dominates until puberty and then slows while axial growth accelerates. During puberty, the accumulation of cortical mass occurs by periosteal apposition, with a large contribution from endocortical apposition in girls. Periosteal apposition accounts for most of the cortical thickness in boys. The differing temporal pattern of axial and appendicular growth and the sex-specific pattern of cortical thickening predispose to region-specific deficits in size and structure should illnesses or other factors affecting the skeleton intervene. This could explain some of the subsequent morphological changes in men and women with fractures.4-6 Estrogen deficiency in young females results in (i) increased periosteal apposition (because estrogen inhibits periosteal apposition) enlarging external bone diameter, (ii) reduced endocortical apposition producing a thinner cortex, and (iii) increased appendicular length, as estrogen contributes to epiphyseal closure. A longer and wider bone with reduced cortical thickness is predisposed to buckling.7 Reduced peripubertal axial growth results in reduced bone size and decreased bone mass because of reduced trabecular thickening.8, 9 Testosterone deficiency in males results in reduced periosteal apposition and, consequently, results in reduced cortical thickness. Thus, growth-related factors are likely to contribute to structural abnormalities observed in adulthood in cross-sectional studies such as smaller bone size, thinner cortices, and thinner trabeculae that predispose to fractures many years later. Peak bone mass and size is achieved around the age of 15-20 years in women and later in men. There is little evidence for any prolonged period of stability in whichneither bone gain or bone loss occurred. Indeed, there is evidence for bone loss between approximately 30 and 49 years of age before estrogen deficiency has occurred.10, 11 Undefined genetic factors account for a portion of the variance in modeling and remodeling during growth. These genetically determined factors presumably turn off, as reflected in the rapid decline in biochemical measures of bone remodeling after puberty. Bone remodeling then continues, but at a lower rate. It appears that “aging” begins when “growth” stops but we do not know what regulates the transition. The net amount of bone lost during aging is the result of two simultaneous but opposing processes; bone resorption within the endosteal envelope and bone formation beneath the periosteal envelope. Subperiosteal bone formation increases bone size reducing load per unit of bone area and so partly offsets endosteal resorption.12 The importance of this subperiosteal apposition is inadequately appreciated. Net age-related bone loss is less in men than in women because subperiosteal bone formation is greater in men, not because endosteal resorption is greater in women. Reduced vertebral size in men and women with spine fractures cannot be explained by increased bone resorption but might be caused by, in part, reduced subperiosteal apposition. In patients with osteoporotic fractures, reduced iliac crest cortical thickness is associated with reduced core length, the distance between periosteal surfaces of the inner and outer cortices, consistent with reduced periosteal apposition, which in turn may have occurred during growth or aging or both.13, 14 Medullary diameter, the distance between endocortical surfaces of the inner and outer cortices is not increased, as would be expected if endocortical resorption were the dominant mechanism. The smaller bone size in women than in men and in patients with spine fractures than in controls could well be the result of impaired adaptation or reduced loading earlier in life, but as yet there are no data to support this view. Bone loss on the endosteal (endocortical, trabecular, and intracortical) surfaces occurs because of the negative balance between the volumes of bone resorbed and the volumes of bone formed at each remodeling site or basic multicellular unit (BMU). This negative bone balance is caused by, in part, an age-related fall in bone formation as evidenced by a reduction in mean wall thickness, which is linear, similar in men and women, and begins before menopause.15 Hence, it is difficult to attribute this cause of negative bone balance to estrogen deficiency or an estrogen receptor defect. The rapid increase in bone mass after administration of antiresorptive drugs is the result of a reduction in activation frequency and filling of the remodeling space caused by resorption-formation coupling.16, 17 The reverse occurs with estrogen deficiency: activation frequency increases, expanding the size of the reversible remodeling space and bone mass falls rapidly. Estrogen deficiency increases the life span of osteoclasts by reducing apoptosis and reduces the life span of osteoblasts by increasing apoptosis.18, 19 Subsequently, bone loss occurs more slowly in most individuals than in the immediate postmenopausal state patients because a new steady state is restored, but this is a higher remodeling rate than before menopause. Bone loss continues because of persistent high remodeling and negative balance at the BMU. The rate of bone loss may accelerate if bone balance at the BMU becomes more negative than before or during menopause because of the change in the life span of the osteoclast and osteoblasts referred to previously. The loss of bone is important because of the structural damage it produces such as cortical thinning, intracortical porosity, trabecular thinning, and loss of connectivity. The biomechanical consequence of the structural changes produced by bone loss depends on bone morphology at peak. In a bone with thicker trabeculae or a thicker cortex, the same endosteal bone loss may have less severe biomechanical consequences than in a bone with thinner cortices and trabeculae. The adaptive response of the skeleton may be impaired after menopause as suggested by Lanyon and Skerry.1 It is feasible that as endosteal bone loss proceeds, strain increases on the periosteal surface, stimulating bone formation to occur until bone size increases and dissipates the strain. However, there is no evidence that the changes on these surfaces are different in estrogen-replete or -deficient states. The mechanism of action of estrogen on bone is controversial and complex. Lanyon's group have reported that some estrogen effects are related to the response to mechanical loading.20, 21 However, effects such as the increased production of bone-resorbing cytokines by marrow cells and inhibitory effects on resorption mediated through local factors are likely to be independent of loading. For example, in rodents, bone loss after ovariectomy can be prevented by blocking interleukin-1 (IL-1) or tumor necrosis factor (TNF) or deleting the IL-1 receptor.22-24 Estrogen may inhibit osteoclast formation and activity by increasing the production of osteoprotegerin or transforming growth factor β (TGF-β).25, 26 The interaction of marrow and bone is supported further by the occurrence of bone loss in patients with proliferative hematopoietic disorders26 and estrogen loss can increase hematopoiesis particularly of B cells.27, 28 The endosteal resorption and loss of trabecular structure characteristic of primary osteoporosis further support an interaction between marrow and bone. The ability of changes in loading to affect these interactions is not known. Cytokines produced by hematopoietic cells as well as bone cells could play a role in pathogenesis of bone fragility. Marrow cells, including macrophages, B and T lymphocytes, and mast cells can produce factors that act on skeletal tissue and changes in these cells have been found in human and animal models of osteoporosis.29-32 The conversion of hematopoietic marrow to fat also may play a role. Cells that differentiate into osteoblasts also can be made to differentiate into adipocytes.33 The age-related increase in marrow fat may be at the expense of osteoblast precursors.34 The vessels in bone also might play a role. Endothelial cells can produce factors that influence bone cells.35, 36 There also are abundant nerve endings in bone that produce neuropeptides influencing bone cell function.37, 38 PTH may even act on cytokine production by the liver.39 The focus on bone resorption as the primary pathogenetic mechanism during the last 30-40 years has been based on the fact that on average the biochemical and histological measures of bone remodeling are increased. However, these values may be normal, increased, or decreased in individual patients with fractures.40-45 This variability suggests that bone fragility has a widely varying pathogenesis. There may not only be increased resorption, but also impaired endosteal, cortical, or periosteal bone formation. It seems unlikely that these different mechanisms have a single underlying cause. A high bone-remodeling rate with negative bone balance in each BMU may not only produce structural damage, but will cause a decrease in bone mineral content out of proportion to the decrease in bone volume. This occurs because younger, less mineralized bone replaces older more mineralized bone. In older individuals, this increase is driven not only by estrogen deficiency but also by calcium and vitamin D deficiency, both contributing to secondary hyperparathyroidism.48, 49 An increase in PTH will increase bone remodeling and may increase intracortical porosity and, hence, the surface available for remodeling in the cortex. Meanwhile, trabecular remodeling may diminish in osteoporotic patients because of a loss of trabecular mass. Thus, in the elderly, bone loss may be predominantly cortical and absolute rates of bone loss may increase because of greater remodeling imbalance. It is not clear how this process could be considered a failure of adaptation to loading, beyond the obvious, that is, it leads to a fracture, which is the ultimate failure of adaptation. There are many pathways to bone fragility. Thus, a given deficit in bone mass and strength in patients with fractures could be caused by reduced bone size because of reduced periosteal apposition during growth or aging or both. Reduced cortical thickness may be the result of reduced periosteal apposition, reduced endosteal apposition, or net endosteal resorption during growth as well as during aging. Increased intracortical porosity is a function of the remodeling rate and the BMU imbalance. Reduced trabecular number may have its origins in embryonic development (as trabecular numbers are determined at the growth plate) as well as late remodeling imbalance and reduced trabecular thickness may be growth or age-related. The loss of trabecular connectivity may be a function of whether the deficit in bone balance is the result of deeper resorption cavities produced by older but still vigorous osteoclasts. Reduced bone formation favors thinning rather than perforation of trabeculae, a mechanism that may occur more in men than in women during aging. Thus, many gender-, growth-, age-, surface-, and site-specific factors contribute to the structural abnormalities that increase bone fragility. It seems unlikely that all these changes can be explained by a single cause. Evidence for a causal relationship between a change in the response to microstrain and endosteal bone loss, increased resorption or decreased formation at the BMU, increased activation frequency, or decreased subperiosteal bone formation is not available. This remains, as Lanyon and Skerry1 state, a hypothesis. Complete immobilization produces a striking remodeling imbalance with a rapid, albeit transient, increase in bone resorption followed by a sustained decrease in bone formation.49 Moreover there is abundant animal work testifying to the efficacy of local loading to stimulate bone formation. However the evidence that exercise reduces fracture incidence in adult humans is weak.50 Thus, even if aging or osteoporosis are associated with a change in the threshold for the detection of strain, there is little evidence that increasing exercise-induced loading compensates and so reduces skeletal fragility. It might be a good idea to reset the mechanostat so that watching TV is like weight training, but how is this to be achieved? Adding a vibrating platform to the TV chair is an intriguing approach, but might cause as much joint damage as skeletal benefit. The views expressed here and those presented by Lanyon and Skerry1 are not mutually exclusive. A unifying hypothesis that explains bone fragility in old age would be an elegant insight and there is little doubt that estrogen deficiency plays a key role in many aspects of skeletal growth and aging.50 However, estrogen deficiency acts through many different pathways. Whether it also is the factor frustrating the biomechanically driven changes in bone structure and preventing adaptation to altered loading is uncertain. An answer to these questions requires a broad approach to pathogenetic mechanisms regulating bone structure. In conclusion, we suggest that bone fragility in old age can have its seed planted 9 months before birth and the evolution of bone fragility occurs during growth as well as during aging. We agree with Lanyon and Skerry1 that a structural approach should be taken to understand the pathogenesis of bone fragility. However, the full spectrum of causes of bone fragility will only emerge when we dispense with the notion of osteoporosis as a disease with one cause. Lumping individuals into diagnostic categories based on BMD below −2.5 SD, one or more spine fractures, or a fracture with a fall from no greater than the standing position obscures the heterogeneity in the structural, cellular, and biomechanical basis of bone fragility. Ignoring heterogeneity will impede the development of methods for the precise detection of risk and prevention of fractures in individual patients, and that is the bottom line. We would like to thank Barbara Capella for her help in the preparation of this article. This work was supported by National Institutes of Health (NIH) grants M01-RR-06192 and AR-18063 (L.G.R.).