As Your Serene Highness well knows, I discovered in the heavens many things that had not been seen before our own age. The novelty of these things, as well as some consequences which followed from them in contradiction to the physical notions commonly held among academic philosophers, stirred up against me no small number of professors—as if I had placed these things in the sky with my own hands in order to upset Nature and overturn the sciences. They seemed to forget that the increase of known truths stimulates the investigation, establishment, and growth of the arts; not their diminution or destruction. (Galileo's Daughter, Dava Sobel, Walker & Company, New York, NY, U.S.A., 1999.)

For over five decades, osteoporotic fractures have been recognized as one of the most devastating complications of chronic treatment with corticosteroids. However, until now, the relationship between the dose and duration of oral glucocorticoid use and the risk of fractures or the effect of discontinuing treatment on subsequent risk were unknown. In this issue, van Staa and coworkers report the results of a retrospective cohort study involving an astounding number of a quarter of a million oral steroid users and an equal number of age and sex-matched control patients from a computerized medical record database of general practitioners in the United Kingdom.(1) They show that the risk of fractures, especially for the vertebrae and the proximal femur, increases in direct proportion with the daily dose of the steroid. This observation could be reasonably anticipated. However, what is completely unexpected is the finding that the fracture risk increases rapidly (as early as 3 months) after the initiation of steroid therapy. And, even more strikingly, it reverses sharply after the discontinuation of the medication toward baseline, although it does not reach it. The sheer size of the population studied by van Staa et al.(1) makes it unlikely that a similar source of information will be available any time soon. Therefore, I fully expect that the findings of this study and their theoretical and practical implications will reverberate in the field for some time. I welcome the opportunity to give my perspective and get the debate going.

The retrospective design of the study has, of course, the usual limitations and potential pitfalls, which van Staa et al.(1) appropriately recognize and discuss, but these hardly detract from the validity of the observations. Bone mineral density (BMD) values apparently are not available for this large epidemiological analysis. Albeit, I doubt that these rapid changes in fracture incidence could be explained away by corresponding changes in BMD or even true bone mass because they seem to occur long before bone mass could have been lost or restored to such an extent as to alter bone strength as drastically as the results indicate. The contention that changes in bone mass alone cannot explain the rapid onset and offset of fractures in this study is supported by observations that the incidence of fractures is higher in patients with corticosteroid-induced osteoporosis than postmenopausal or involutional osteoporosis, even though BMD is relatively higher in the former.(2–4) Furthermore, there is abundant evidence for an incongruity between BMD and fractures from numerous studies showing that the antifracture efficacy of traditional therapies for osteoporosis, for example, estrogen, bisphosphonates, and calcitonin, is far greater than the BMD changes produced by these agents.(5) This incongruity holds even when one takes into consideration the fact that the relationship of BMD and fractures is a power function. Taken together with the results of van Staa et al.(1) all these lines of evidence challenge traditional ideas about the mechanisms leading to increased fragility in osteoporosis, the relationship between bone mass and fractures, or our ability to predict fracture risk by conventional means.

van Staa et al.(1) offer three explanations for their findings: the apoptosis of osteoblasts and osteocytes induced by corticosteroids; a marked alteration (I suspect they mean an increase) in bone turnover, through induction of microarchitectural changes in bone quality; and nonskeletal mechanisms like increased risk of falling. I will address the last one first. Unquestionably, falls play a very important role in osteoporotic fractures in general and may indeed contribute to the rapid onset and offset of fracture risk in the analysis of this report. However, I seriously doubt that falls alone can account for these findings, especially because vertebral as well as nonvertebral fractures exhibit practically identical time kinetics and forearm is the site affected least. Van Staa et al.(1) suggest that marked alteration in bone turnover is in agreement with the evidence of a protective effect of bisphosphonates in corticosteroid-induced osteoporosis. This is in my opinion unconvincing, if not outright erroneous. Turnover in corticosteroid-induced osteoporosis is suppressed. Therefore, it is very unlikely that the protective effect of bisphosphonates, agents that unquestionably slow turnover, is a result of this mechanism. Based on this and other even more compelling reasoning that will be discussed below, I much favor osteocyte apoptosis induction by corticosteroids as the most logical explanation of these data.

It is widely believed that as in other types of the disease, steroid-induced osteoporosis results from loss of bone mass and microarchitectural deterioration predisposing to fragility and fractures. Nonetheless, the histological features and pathogenetic mechanisms responsible for the adverse affects of corticosteroids on the skeleton are quite distinct from the adverse affects of other causative factors, for example, sex-steroid deficiency. Indeed, unlike sex-steroid deficiency in which both bone resorption and formation increase, the key histological feature of glucocorticoid-induced osteoporosis is a decrease in the rate of bone formation, manifested as a reduction in cancellous bone area and trabecular width; as well as so-called “necrosis” of sections of the skeleton.(6) All these pathognomonic features of chronic glucocorticoid therapy can be explained by the evidence that steroid excess promotes the apoptosis of osteoblasts and osteocytes in mice and rats, as well as humans, and also suppresses the production of new osteoblasts and osteoclasts, at least in mice.(7–9) Consistent with the latter, biochemical markers of bone resorption in humans do not change, even at the early stages of steroid therapy.(10) Albeit, some of the bone loss during the early phases of the disease may be caused by a “relative” increase in bone resorption, brought about by the profound and rapid reduction of bone formation. This possibility is supported by the finding that early on (7 days), after glucocorticoid administration to mice, the number of osteoclasts in bone sections doubles even though osteoclastogenesis in ex vivo bone marrow cultures is significantly decreased (Weinstein et al., 2000, unpublished observations). This evidence, together with in vitro findings that steroids inhibit osteoprotegerin and concurrently stimulate the expression of the receptor activator of NF-kb (RANK)-ligand by osteoblastic cells,(11) raise the possibility that the initial rapid phase of bone loss with glucocorticoid treatment could be caused by in part an extension of the life span of preexisting osteoclasts.(12)

In agreement with the idea that osteocyte apoptosis is a critical pathogenetic mechanism for corticosteroid-induced osteoporosis and an explanation for the rapid onset and offset of fracture risk in the study of van Staa et al.,(1) whole femoral heads obtained from patients with glucocorticoid-induced osteoporosis, but not appropriate controls, exhibit abundant apoptotic osteocytes adjacent to the subchondral fracture crescent.(8) Moreover, bisphosphonates and intermittent parathyroid hormone (PTH) administration—another therapeutic regimen that like bisphosphonates recently has been shown to be effective in steroid-induced osteoporosis(13)—both have potent antiapoptotic effects on osteoblasts and osteocytes in vitro and in vivo.(14, 15) Notably and in regard to the suggestion by van Staa et al.(1) for a connection between bone turnover and the effectiveness of bisphosphonates, intermittent PTH has either no effect or the opposite effect of bisphosphonates on turnover.

How can the rapid increase in fracture risk, as early as the first 3 months of treatment, result from osteoblast and osteocyte apoptosis induced by corticosteroids? Sudden death of osteoblasts by apoptosis in existing basic multicellular units (BMUs) accompanied by a “relative” increase in resorption caused by prolongation of the life span of existing osteoclasts, would certainly lead to bone loss. However, on an average, the life span of a BMU (and therefore the duration of a cycle of remodeling) is 6–9 months and the rate of turnover of the whole skeleton is approximately 10% per year (based on an average 4% turnover per year in cortical bone, which represents roughly 75% of the entire skeleton, and an average 28% turnover per year in trabecular bone, which represents roughly 25% of the skeleton). Considering this and the fact that corticosteroid excess decreases dramatically activation frequency, that is, the rate of generation of new BMUs—a phenomenon consistent with the findings of suppressive effects of steroid excess on osteoblastogenesis and osteoclastogenesis—one cannot see how an imbalance in the work performed by existing BMUs could lead to that much bone loss that can account for increased fractures within 3 months. Even if one accepts as a possibility that the “sudden” imbalance in the life span of osteoclasts and osteoblasts leads to perforation of complete trabeculae, still one cannot see how this effect can be then reversed quickly on cessation of the treatment, so as to decrease bone fracture risk within a time period that corresponds to one-third or one-half of a single remodeling cycle. In fact, if discontinuation of steroids after several years of treatment would have led to a rapid restoration of bone mass—via an increase in remodeling brought about by the generation of new BMUs—bone mass, according to conventional wisdom, should decrease temporarily because of an expansion of the remodeling space.

Increased osteoblast and osteocyte apoptosis is not a unique feature of corticosteroid-induced osteoporosis, as similar phenomena also have been documented in estrogen- or androgen-deficient rodents and humans.(16–22) Why then is the incidence of fractures in glucocorticoid-treated patients higher compared with postmenopausal women, even though BMD in the former is relatively higher? It is very likely that this apparent difference relates to the fact that glucocorticoid excess suppresses osteoblastogenesis and osteoclastogenesis (thereby the formation of new BMUs and the ability of the tissue to repair itself) whereas estrogen deficiency has exactly the opposite effect on the birth of these two cell types. Collapse of the femoral head, which as I mentioned above can be explained by the demonstration of increased osteocyte apoptosis, is far more common in corticosteroid-induced osteoporosis as compared with postmenopausal osteoporosis. This leads me to suggest that there might be a quantitative difference in the prevalence of apoptosis, being greater in the former.

Although bone remodeling is the function of osteoclasts and osteoblasts, sensing of the need of the skeleton for mechanical adaptation is thought to be the function of osteocytes—osteoblasts that have been entombed in the mineralized matrix.(23) However, unlike the function of osteoclasts and osteoblasts that has been studied extensively and largely elucidated, very little is known about the biology and function of osteocytes, let alone their contribution to bone remodeling or strength. Osteocytes are by far the most abundant cell type in bone—there are 10 times more osteocytes than osteoblasts. They are spaced regularly throughout the mineralized matrix and communicate with one another, as well as osteoblasts and lining cells on the surface of the bone, through dendritic processes that run along the canaliculi.(24, 25) It is believed that osteocytes sense the need for remodeling at a specific time and place and communicate it to lining cells, which in turn provide the “homing” signal targeting osteoclast precursors to the specific location.(26) In support of the idea that osteocyte apoptosis provides a key part of the activation or signaling mechanisms by which osteoclasts target bone for removal, a strong association between microdamage, osteocyte apoptosis, and osteoclastic resorption has been observed in rats.(27) Two other hypotheses state that osteocytes slow production of osteoid by osteoblasts so as to enable them to be buried in the matrix as the next osteocyte (28, 29); and simultaneously they restrain remodeling by providing inhibitory signals to lining cells.(30) According to the latter hypothesis, apoptosis of osteocytes resulting from microdamage or from a direct effect of changing hormonal milieu would interrupt the transmission of these inhibitory signals, resulting in increased remodeling. Although attractive as a mechanism of perceiving and repairing focal microdamage, it is very unlikely that such a mechanism plays a substantial role in the pathogenesis of systemic diseases such as osteoporosis, perhaps with the exception of the osteoporosis that results from immobilization. Indeed, in corticosteroid-induced osteoporosis and aging, two of the conditions in which osteocyte apoptosis is increased, there is no resulting increased remodeling. In fact, the opposite is true. Likewise, in sex-steroid deficiency, another hormonal change associated with increased osteoblast and osteocyte apoptosis, increased remodeling clearly results from removal of inhibitory effects of these hormones on the production of osteoblasts and osteoclasts, as evidenced by the prevention of these phenomena with specific cytokine antibodies, antagonists, or in cytokine gene knockout models.(31, 32) Be that as it may, sex steroids and mechanical loading may share similar signaling pathways for preventing osteocyte apoptosis. However, although apoptosis induced by changes in the levels of hormones is systemic, changes in loading or microfractures are obviously localized.

It seems counterintuitive that osteocytes have no function other than waiting idly to die and serve as sacrificial lambs and beacons to the purpose of repairing local microdamage by remodeling. In this context, it is important to note that osteocytes are present in bones that are never remodeled, like the os ossicle of mammals.(33) In addition, osteocytes are found in teleosts, in which the skeleton is not remodeled.(34, 35) In distinction to, but not necessarily exclusive of, the idea that osteocyte apoptosis leads to localized tissue repair, there is evidence that disruption of the osteocyte network by apoptosis increases bone fragility.(7) Conversely, preservation of the osteocyte network seems to be an important mechanism by which bisphosphonates, calcitonin and estrogen replacement therapy decrease bone fragility.(15, 21, 22) Moreover, inhibition of osteoblast and osteocyte apoptosis is the most likely mechanism of the anabolic effect of intermittent PTH therapy; and, in fact, prevention of osteoblast (and osteocyte) apoptosis predictably leads to more numerous and closely spaced osteocytes.(14) The inexorable conclusion from all this evidence is that osteocytes play an important role in the maintenance of bone quality. Exactly how preservation of the osteocyte network contributes to bone quality and strength is only a matter of conjecture. Prevention of complete mineralization, so that bone (I refer to the substance) density does not exceed critical limits,(36) or perhaps control of collagen fibril orientation and degradation(37) or control of fluid traffic may be only few of many mechanisms by which this most abundant and least studied “third cell kind” contributes to the microarchitectural quality and strength of the skeleton (Fig. 1).

thumbnail image

Figure FIG. 1.. Hypothetical mechanisms offered as explanations for the rapid onset and offset of fracture risk with glucocorticoid treatment in the study of van Staa et al.(1)

Download figure to PowerPoint

Last, the data of van Staa et al.(1) bear on the growing appreciation of the limitations of BMD as a means of predicting fracture risks. Minus the results of van Staa et al.,(1) it is inconceivable that even if BMD was included as an end point in a prospective future study, anyone involved in the care of the patients or the design of the study could have prophetic powers, let alone the resources, to record them in 3-month intervals. More seriously, for the reasons I discussed above, I do not think that BMD measurement would have revealed the imminent risk of fractures in the first few months of treatment with corticosteroids in those patients that succumb to them. I also suspect that BMD could not predict the rapid offset of fracture risk so that some patients no longer on steroids could have avoided unnecessary treatment for osteoporosis. Unlike true bone density (which is the density of mass per unit volume of bone as a substance or material) or apparent bone density (which is the density of a whole bone as an organ, representing mass divided by external volume), the BMD used routinely for the diagnosis and management of patients with osteoporosis is a measurement of bone mineral mass partly normalized by bone area. I suspect that in patients with corticosteroid-induced osteoporosis (untreated or treated with bisphosphonates), like in other complex and unpredictable bone diseases, changes in numerical value for BMD may no longer have the same structural significance as in a normal population.(38)

As in every encounter with the unknown, the landmark observations of van Staa et al.(1) are likely to cause some stirring and apprehension because they do not fit into traditional thinking of bone biology. However, they also are likely to open avenues leading to exciting new vistas of concepts and ideas of direct relevance to clinical medicine. For this, the field should be grateful to these authors for a truly remarkable contribution. In the interim, it seems safe to recommend that prevention of osteoblast and osteocyte apoptosis by initiating treatment with antiapoptotic regimens (e.g., bisphosphonates) shortly before or simultaneously with corticosteroid therapy is the most rationale approach for decreasing fractures and their debilitating consequences.


  1. Top of page
  • 1
    van Staa T, Leufkens H, Abenhaim L, Zhang B, Cooper C 2000 Use of oral corticosteroids and risk of fractures. J Bone Miner Res. J Bone Miner Res 15:9931000.
  • 2
    Peel NF, Moore DJ, Barrington NA, Bax DE, Eastell R 1995 Risk of vertebral fracture and relationship to bone mineral density in steroid treated rheumatoid arthritis. Ann Rheum Dis 54:801806.
  • 3
    Luengo M, Picado C, Del Rio L, Guanabens N, Montserrat JM, Setoain J 1991 Vertebral fractures in steroid dependent asthma and involutional osteoporosis: A comparative study. Thorax 46:803806.
  • 4
    Dennison E 1999 Epidemioogy of glucocorticoid-induced osteoporosis. Osteoporos Int 9:S16. (abstract)
  • 5
    Cummings S, Black D, Vogt TM 1996 Changes in BMD substantially underestimate the anti-fracture effects of alendronate and other antiresorptive drugs. J Bone Miner Res 11:S102. (abstract)
  • 6
    Manolagas SC, Weinstein RS 1999 New developments in the pathogenesis and treatment of steroid-induced osteoporosis. J Bone Miner Res 14:10611066.
  • 7
    Weinstein RS, Jilka RL, Parfitt AM, Manolagas SC 1998 Inhibition of osteoblastogenesis and promotion of apoptosis of osteoblasts and osteocytes by glucocorticoids: Potential mechanisms of their deleterious effects on bone. J Clin Invest 102:274282.
  • 8
    Weinstein RS, Nicholas RW, Manolagas SC 2000 Apoptotsis of osteocytes in glucocorticoid-induced osteonecrosis of the hip. J Clin Endo Metab in press.
  • 9
    Silvestrini G, Ballanti P, Patacchioli FR, Mocetti P, Di Grezia R, Wedard BM, Angelucci L, Bonucci E 2000 Evaluation of apoptosis and the glucocorticoid receptor in the cartilage growth plate and metaphyseal bone cells of rats after high-dose treatment with corticosterone. Bone 26:3342.
  • 10
    Pearce G, Tabensky DA, Delmas PD, Baker HW, Seeman E 1998 Corticosteroid-induced bone loss in men. J Clin Endocrinol Metab 83:801806.
  • 11
    Hofbauer LC, Gori F, Riggs BL, Lacey DL, Dunstan CR, Spelsberg TC, Khosla S 1999 Stimulation of osteoprotegerin ligand and inhibition of osteoprotegerin production by glucocorticoids in human osteoblastic lineage cells: Potential paracrine mechanisms of glucocorticoid-induced osteoporosis. Endocrinology 140:43824389.
  • 12
    Manolagas SC 1999 Cell number versus cell vigor—what really matters to a regenerating skeleton? Endocrinology 140:43774381.
  • 13
    Lane NE, Sanchez S, Modin GW, Genant HK, Pierini E, Arnaud C 1998 Parathyroid hormone treatment can reverse corticosteroid-induced osteoporosis. J Clin Invest 102:16271633.
  • 14
    Jilka RL, Weinstein RS, Bellido T, Roberson P, Parfitt AM, Manolagas SC 1999 Increased bone formation by prevention of osteoblast apoptosis with PTH. J Clin Invest 104:439446.
  • 15
    Plotkin L, Weinstein RS, Parfitt AM, Roberson PK, Manolagas SC, Bellido T 1999 Prevention of osteocyte and osteoblasts apoptosis by bisphosphonates and calcitonin. J Clin Invest 104:13631374.
  • 16
    Tomkinson A, Gevers EF, Wit JM, Reeve J, Noble BS 1998 The role of estrogen in the control of rat osteocyte apoptosis. J Bone Miner Res 13:12431250.
  • 17
    Tomkinson A, Reeve J, Shaw RW, Noble BS 1997 The death of osteocytes via apoptosis accompanies estrogen withdrawal in human bone. J Clin Endocrinol Metab 82:31283135.
  • 18
    Noble BS, Stevens H, Loveridge N, Reeve J 1997 Identification of apoptotic changes in osteocytes in normal and pathological human bone. Bone 20:273282.
  • 19
    Bellido T, Plotkin LI, Han L, Manolagas SC 1999 Estrogen inhibit apoptosis of osteoblasts and osteocytes through rapid (non-genomic) activation of extracellular signal-regulated kinases (ERKs). J Bone Miner Res 14:S342. (abstract)
  • 20
    Bodenner DL, Yamamoto M, Kozlowski M, Manolagas SC 1999 Essential requirement of the estrogen receptor alpha or beta for (non-genomic) anti-apoptotic effects of estrogen. J Bone Miner Res 14:S227. (abstract)
  • 21
    Manolagas SC, Weinstein RS, Bellido T, Bodenner DL, Jilka RL, Parfitt AM 1999 Activators of non-genomic estrogen-like signalling (ANGELS): A novel class of small molecules with bone anabolic properties. J Bone Miner Res 14:S180. (abstract)
  • 22
    Weinstein RS, Bellido T, Chambers TM, Crawford JA, Swain FL, Han L, Manolagas SC 1999 Like estrogen, androgen exert potent and direct anti-apoptotic effects on osteoblasts and osteocytes in vivo and in vitro. J Bone Miner Res 14:S451. (abstract)
  • 23
    Aarden EM, Burger EH, Nijweide PJ 1994 Function of osteocytes in bone. J Cell Biochem 55:287299.
  • 24
    Marotti G, Cane V, Palazzini S, Palumbo C 1990 Structure-function relationships in the osteocyte. Ital J Miner Electro Metab 4:93106.
  • 25
    Nijweide PJ, Burger EH, Klein-Nulend J, van der Pluijm G 1996 The osteocyte. In: BilezikianJP, RaiszLG, RodanGA (eds.) Principles of Bone Biology. Academic Press, San Diego, CA, U.S.A., pp. 115126.
  • 26
    Parfitt AM, Mundy GR, Roodman GD, Hughes DE, Boyce BF 1996 A new model for the regulation of bone resorption, with particular reference to the effects of bisphosphonates. J Bone Miner Res 11:150159.
  • 27
    Verborgt O, Gibson GJ, Schaffler MB 2000 Loss of osteocyte integrity in association with microdamage and bone remodeling after fatigue in vivo. J Bone Miner Res 15:6067.
  • 28
    Marotti G 1996 The structure of bone tissues and the cellular control of their deposition. Ital J Anat Embryol 101:2579.
  • 29
    Marotti G, Ferretti M, Muglia MA, Palumbo C, Palazzini S 1992 A quantitative evaluation of osteoblast-osteocyte relationships on growing endosteal surface of rabbit tibiae. Bone 13:363368.
  • 30
    Martin RB 2000 Toward a unifying theory of bone remodeling. Bone 26:16.
  • 31
    Manolagas SC 1998 Cellular and molecular mechanisms of osteoporosis. Aging Clin Exp Res 10:182190.
  • 32
    Pacifici R 1998 Cytokines, estrogen, and postmenopausal osteoporosis—the second decade. Endocrinology 139:26592661.
  • 33
    Friedmann I 1974 Pathology of the Ear. Blackwell Scientific Publications, Oxford, U.K.
  • 34
    Hughes DR, Bassett JR, Moffat LA 1994 Histological identification of osteocytes in the allegedly acellular bone of the sea breams Acanthopagrus australis, Pagrus auratus, and Rhabdosargus sarba (Sparidae, Perciformes, Teleostei). Anat Embryol (Berl) 190:163179.
  • 35
    Hughes DR, Bassett JR, Moffat LA 1994 Structure and origin of the tooth pedicel (the so-called bone of attachment) and dental-ridge bone in the mandibles of the sea breams Acanthopagrus australis, Pagrus auratus, and Rhabdosargus sarba (Sparidae, Perciformes, Teleostei). Anat Embryol (Berl) 189:5169.
  • 36
    Frost HM 1960 Micropetrosis. J Bone Joint Surg Am 42A:144150.
  • 37
    Zhao W, Byrne M, Boyce B, Krane S 1999 Bone resorption induced by parathyroid hormone is strikingly diminished in collagenase-resistant mutant mice. J Clin Invest 103:517524.
  • 38
    Parfitt AM 1998 A structural approach to renal bone disease. J Bone Miner Res 13:12131220.