BoneKEy-Osteovision | Meeting Reports

Meeting report from the 24th annual meeting of the American society for bone and mineral research



DOI:10.1138/2002065

Introduction

BoneKEy-Osteovision covered the 24th Annual Meeting of the American Society for Bone and Mineral Research (ASBMR) with a team of thirteen reporters. In the meeting report we have tried to capture recurring themes in oral and poster presentations. To find topics of interest, use the Table of Contents below (just click on topics) to navigate through the text of the meeting report. Through the courtesy of ASBMR, each abstract referenced in the report contains a full citation with a link to the Abstracts online in the reference list. Abstracts are also published in the 2002 ASBMR Annual Meeting Program and Abstracts (Journal of Bone and Mineral Research, 2002;17[Supplement 1]). We could not cover every important theme that was presented at the meeting and we apologize if we have slighted an area that is of interest to you. We also extend our thanks to the staff of ASMBR and program chairs of the meeting for their generous help in making this report possible. This is the most ambitious of a series of meeting reports BoneKEy has undertaken during the past year. Please let us know what you think of the report by sending us your feedback. GJS.

Table of Content

When Bone Morphogenetic Protein Meets LRP5 - Patricia Ducy Steroid Hormones and Bone: Of Angels and Lucifers - Roger Bouillon and Dirk Vanderschueren Osteoclast Biology: M-CSF, RANKL, Src, and New Friends - Patrick Ross and Anna Teti Puzzles of Cartilage Biology - Ernestina Schipani Calciotropic Hormones - Gordon J. Strewler Phosphate Homeostasis: The Search for Phosphatonin Continues - Rajesh V. Thakker Epidemiology of Osteoporosis: It's Not About White Women - Dennis M. Black Genetics of Osteoporosis: QTLs Meet Candidates - Stuart H. Ralston Bone Quality is Job One - Ego Seeman Anything New in the Use of Bisphosphonates in the Management of Osteoporosis? - Socrates E. Papapoulos New Developments in Bone Densitometry - Paul D. Miller

When bone morphogenetic protein meets LRP5 Patricia Ducy, Department of Molecular and Human Genetics, Baylor College of Medicine, Houston, TX, USA

At last year's American Society for Bone and Mineral Research (ASBMR) meeting, low-density lipoprotein receptor-related protein 5 (LRP5) was reported as one of the most potent regulators in osteoblast biology because activating and inactivating mutations in the gene encoding this factor were found responsible for high bone mass (HBM) syndrome and osteoporosis pseudoglioma (OPPG) syndrome in humans and mice, respectively (). This year, several studies have focused on this finding, mostly analyzing mouse models and signaling pathways associated with LRP5.

Two major studies (one of which will soon be published) that analyzed transgenic mice that overexpress the G171V HBM allele or wild-type LRP5 allele were discussed (). These results were also discussed by Mark Johnson at the Plenary Symposium (LRP5: The Highs and Lows of Bone Mass). Both mouse models used the a1(I) collagen 3.6-kb promoter to drive expression of the transgene, a promoter fragment that confers expression to a transgene early in development and during osteoblastic differentiation, as well as in fibroblasts of the tendons and skin. Consistent with the broad expression of LRP5 but the bone-restricted phenotype associated with its mutant alleles, the transgenic mice do not exhibit any overt phenotype outside of the skeleton, even though the transgene is expressed elsewhere. As early as 5 weeks of age, HBM transgenic mice presented an increase in trabecular bone mineral density (BMD) that can reach 250% of the normal BMD by 52 weeks of age. Accordingly, the mineral apposition rate was increased by 11%. Cortical thickness was also increased by 42% as the result of an increase in periosteal diameter, which led to an increase in the width of the bones that mimicked the one that was observed in HBM patients.

Similar results were obtained when a wild-type LRP5 allele was overexpressed, but the amplitude of the HBM phenotype was markedly milder. Biomechanical analysis of the HBM transgenic mice also revealed that their bones were more sensitive to loading, suggesting that HBM osteoblasts respond to lower strain than normal cells and thus indicating a role for LRP5 in response to mechanical loading. The presence of stress response elements in the LRP5 promoter, as reported by Johnson in his lecture, would lend support to this hypothesis. Another role for LRP5 resides in the control of apoptosis, as less TUNEL-positive osteoblasts and osteocytes were detected in HBM transgenic mice.

These results need to be compared to those that analyze the phenotype of mice that are deficient in secreted frizzled-related protein 1 (sFRP-1), a known antagonist of wnt signaling in other systems. Females with sFRP-1 deficiency show a significant decrease in the number of TUNEL-positive osteoblasts and osteocytes in calvaria, which is associated with an increase of trabecular bone and mineralized area in femurs at 35 weeks of age (). However, in contrast to the HMB mice, the phenotype observed in sFRP-1-deficient mice was milder and transient. Indeed, it did not appear before 20 weeks of age and was not observed after 40 weeks of age. These differences suggest that sFRP-1 is not the only inhibitor of the LRP5 pathway and that other molecules not yet identified either can compensate for its absence or play an additional and earlier role in controlling LRP5 function. To add complexity, addition of sFRP-3, another decoy frizzled receptor, to MC3T3 osteoblastic cells in culture stimulates their production of alkaline phosphatase while decreasing their proliferation ().

Apart from the analysis of LRP5-deficient mice, which was published earlier this year (), the novel insights presented at this year's meeting regarding OPPG and LRP5 inactivation essentially reside in the identification of various alleles that cause this human disease (): more than 10 different mutations have now been identified. This multiplicity is contrary to the still unique G171V HBM mutation. Indeed, a study comparing LRP5 polymorphism in HBM patients (control and sporadic individuals with high BMD) confirmed that G171V was only found in HBM patients. The study further revealed only three candidate-silent mutations associated with high BMD ().

The major novelty, however, came from two independent studies, both of which established a link between bone morphogenetic protein (BMP) induction of osteoblast differentiation and the wnt/LRP5/b-catenin signaling pathway. One group () overexpressed a constitutively active form of b-catenin in C3H10T1/2 undifferentiated mesenchymal cells (or C2C12 myoblastic cells). Upon BMP2 treatment, these cells exhibited an increased response, compared to the response of control cells (osteocalcin expression increased 16- vs 2-fold, respectively; alkaline phosphatase activity increased 12- vs 6-fold, respectively). These infected cultures were also able to form mineralized nodules, while control cultures did not. In contrast, infection of a dominant negative form of b-catenin in C3H10T1/2 undifferentiated mesenchymal cells, or in MC3T3 osteoblastic cells, blocked the ability of BMP2 to induce alkaline phosphatase activity in these cells. These data indicate that a full osteoblastic response to BMP requires functional b-catenin-mediated activation. Of interest, this study also showed that the BMP/b-catenin link occurs independently of Cbfa1, confirming the observation previously made in LRP5-deficient mice—that LRP5 acts independently of the Cbfa1 regulatory pathway.

The second study () also used tissue culture experiments to demonstrate three points. First, induction of alkaline phosphatase in osteoblastic cells by wnt treatment is not blocked upon addition of Noggin, an inhibitor of BMP signaling, whereas inhibition of LRP5 signaling by the wnt signaling inhibitor Dickkopf-1 (Dkk-1) treatment blocks BMP osteo-inductive ability. Second, wnt activation does not require de novo protein synthesis, whereas BMP-mediated induction does. Third, BMP treatment of osteoblastic cells induces wnt. Together, these two lines of evidence suggest that the osteoblastic response obtained upon BMP treatment could involve BMP induction of wnt expression, which, in turn, will stimulate the LRP5/b-catenin pathway that will then directly activate osteoblastic gene expression

Putting Together More Pieces of the Transcriptional Control of the Osteoblast Differentiation Puzzle

Cbfa1 (runx2) has been known for several years to be the earliest, and a major, regulator of osteoblast differentiation. This year's meeting has seen a large number of reports that characterize physical interactions between Cbfa1 and various factors that control its transcriptional function. Among them, a spliced isoform of the adipogenic C/EBPb transcription factor lacking its transactivation domain (LIP or p20C/EBPb), which is known to inhibit adipogenesis in vitro, was expressed in primary mouse calvaria osteoblasts. In mesenchymal C3H10T1/2, cotransfection of LIP with Cbfa1 enhanced their alkaline phosphatase synthesis as well as osteocalcin promoter activity (). LIP was also found to potentiate the osteogenic effect of BMP2 in a Cbfa1-dependent manner (). These results suggest that LIP is involved in defining an osteoblastic versus adipogenic fate of mesenchymal cells via inhibition of C/EBPb-induced adipogenesis and a physical interaction with Cbfa1 that enhances its osteogenic ability. However, transgenic mice overexpressing p20C/EBPb, under the control of the a1(I) collagen 3.6-kb promoter, were osteopenic and showed a decrease in type I collagen and osteocalcin expression (). Accordingly, in tissue culture experiments, transgenic bone marrow stromal cells showed reduced ability to form mineralized nodules and to express osteopontin, bone sialoprotein, and osteocalcin (). The mechanism underlying the difference between these two studies will be an interesting point to address.

Resulting from a yeast two-hybrid screen, a positive interaction was reported between Cbfa1 and DFosB (). Cbfa1 was shown to activate its own promoter with a higher affinity when in the presence of DFosB. This effect was found to be dependent on an AP1 site present upstream of the proximal OSE2 sites previously described in the Cbfa1 promoter. Such enhancing interaction could be, at least in part, responsible for the osteosclerotic phenotype previously reported in transgenic mice that overexpress DFosB (). Negative regulators of Cbfa1 activity were also reported. For instance, Msx2 was shown to decrease Cbfa1 binding to the osteocalcin promoter in vitro () and the E3 ubiquitin ligase, Smurf-1, was shown to interact with Cbfa1 and mediate its proteasome-dependent degradation ().

Another interesting report was the presentation of OSEBF1, a novel transcription factor required for osteogenesis (). OSEBF1 is a leucine zipper molecule that binds to the OSE1 site previously described in the osteocalcin promoter. Its absence in mice causes severe dwarfism and osteopenia postnatally, but no other apparent phenotype. Surprisingly, OSEBF1 osteoblast specificity resides not in its restricted gene expression, but in its absence of degradation, in osteoblastic cells only.

Table of Contents

Steroid hormones and bone: Of angels and lucifers Roger Bouillon and Dirk Vanderschueren, Gasthuisberg, Catholic University, Leuven, Belgium

The sex steroid hormones, estradiol and testosterone, are known to be bone protective, whereas glucocorticoids (GCs) are, conversely, the devils for bone. The 2002 San Antonio American Society for Bone and Mineral Research (ASBMR) meeting provided interesting new data on these angels and lucifer.

Glucocorticoids and bone

Ample in vivo evidence leaves no doubt that excess GCs have negative effects on bone and that they ultimately result in increased fracture rates. Their mode of action is complex and involves presumed effects on gut, muscle, and bone cells. Bone formation is especially impaired, and GC excess was recently associated with increased apoptosis rate of osteoblasts and osteocytes (). In vitro data, however, have also consistently demonstrated a beneficial effect of GCs on early osteoblast differentiation, and GCs are, in fact, standard reagents used to direct pluripotent mesenchymal cells into the osteoblastic lineage. GCs impair only the proliferation of osteoblasts after confluence. Osteoblast-selective (in)sensitivity to GCs might help to elucidate this dilemma, and instead of selective GC receptor knockout, two groups have tried to evaluate the effects of selective 11b-hydroxysteroid dehydrogenase 2 (11b-HSD2) overexpression in osteoblasts to create selective insensitivity to GC effects. This enzyme is capable of transforming active cortisol or prednisol into inactive cortisone/prednisone, in contrast to the naturally more abundant 11b-HSD type 1, which catalyzes the activation of GCs.

Weinstein et al. () demonstrated that 11b-HSD2-transgenic mice still lose bone during prolonged (35 days) prednisone administration. The bone of such animals maintained its normal compression strength, however, and did not show the expected GC-induced osteoblast and osteocyte apoptosis. Thus, osteoblastic GC inactivation is partly (but far from completely) protective against bone loss. However, as prereceptor inactivation only occurs at the time of osteocalcin expression (the osteocalcin promoter being used as enzyme inducer), this result could also mean that GC inactivation comes too late in the osteoblast life cycle to allow the expected complete protection. Much earlier transgenic expression of the same 11b-HSD type 2, by using the collagen promoter, however, revealed () that selective GC inactivation in osteoblasts impaired spontaneous bone growth in female mice, especially in trabecular bone (mild decrease in trabecular bone volume and number). This suggests that normal endogenous GCs are rather beneficial for (trabecular) bone development. This interpretation is further supported by in vitro data using osteoblasts derived from GC receptor-dim mice (). In such cells, the dimerization of GC receptors is impossible, and therefore GC receptor-mediated gene regulation is excluded; however, ligand-activated GC receptor-protein interaction with AP1 and NF-k-B remains functional. GCs, in fact, enhance osteoblast differentiation and nodule formation of such GC receptor-deficient osteoblasts. It seems, therefore, that a little GC is good or necessary for bone, as suggested by previous in vitro data; however, excess, of course, is devilish. This interpretation would imply that the design or development of a (bone) selective GC receptor modulator would be even more complex than expected.

Normal osteoblasts express predominantly the activating 11bOHSD type 1, and human volunteers with greater overall activity of this enzyme (as revealed by urinary excretion of GC metabolites) are apparently more sensitive to osteoblast inhibition by pharmacological doses of GC (). This would suggest that such individuals (re)activate endogenous or exogenous GCs selectively in their osteoblasts and are therefore more prone to the deleterious bone effects of GCs.

Several other abstracts also deal with GC effects on bone and they demonstrate that these effects are certainly multifactoral with regard both to the cells/tissues involved and molecular/humoral effects. GCs indeed suppress two bone beneficial agents: osteoprotegerin (OPG) () and interleukin 11 (IL-11) ().

Estrogens as angels for bone

In his Louis Avioli Award lecture, Riggs et al. () highlighted the important role of estrogen in skeletal homeostasis in both sexes. According to recent data from his group, which were also presented at this meeting, receptor activator of NF-k-B ligand (RANKL) expression is significantly higher in bone marrow cells isolated from early untreated premenopausal women than in pre- or postmenopausal women who receive hormone replacement therapy (HRT) (). Moreover, RANKL expression (per bone marrow cell) in these women was correlated significantly with biochemical bone resorption markers. Therefore, in accordance with earlier rodent data, RANKL seems to be the principal regulator of increased osteoclastogenesis during early estrogen deficiency in women.

Estrogen deficiency not only increases bone resorption, but it may simultaneously impair osteoblast proliferation in response to mechanical loading. Indeed, as reported at this meeting, estrogen receptor a knockout (ERKO) mice show significantly lower bone formation at periosteal and endosteal sites following mechanical loading than wild-type mice (), thereby confirming earlier in vitro data with respect to the possibility of a common signaling pathway for both mechanical loading and estrogen in osteoblasts (). In contrast, however, maximal bone-breaking strength increases less (not more) in response to mechanical loading in intact female rats than in either ovariectomized female or male littermates (), suggesting that estrogen decreases (rather than stimulates) bone sensitivity to mechanical strain. Therefore, the interaction of estrogen with the “mechanostat,” although attractive in theory, still needs further clarification.

The ratio of estrogen a/b receptor seems important not only to bone health, but also for general survival, at least in female mice (). Estrogen receptor b knockout (BERKO) mice show less age-related bone loss and survive longer than wild-type mice. BERKO mice, however, remain fully responsive to effects of ovariectomy. Therefore, although BERKO mice respond normally to ovariectomy and estrogen, a low estrogen receptor (ER) b/a ratio may still be beneficial for aging female mice. Also, in postmenopausal women, relatively low estradiol concentrations may still inhibit bone resorption, as indicated by further bone loss, compared to control, following administration of the aromatase inhibitor, anastrazole (), in the context of adjuvant therapy for early breast cancer.

Results of the Women's Health Initiative Observational Study (WHI-OS), presented during the late-breaking abstract session, show the need for new “designer sex steroids” (LB1). These designer sex steroids ideally should have bone anabolic action without stimulating reproductive tissues in either sex. Along these lines was further elucidation of the activator of nongenotropic estrogen-like signal (ANGEL), estren (). Estren has an equal (or even superior) bone anabolic action compared to either dihydrotestosterone (DHT) in male orchiectomized mice or E2 in female ovariectomized mice, despite a 300-fold lower affinity for the ER, and this without stimulation of reproductive organs in either sex (). It was hypothesized that ANGELs, like sex steroids, shorten and lengthen the lifespan of osteoclasts and osteoblasts, respectively, via a sex-neutral nongenomic signaling pathway (). Furthermore, the authors also assumed that androgens and estrogens alike, without discrimination, may regulate bone cell apoptosis via nongenomic activation of any of the three sex steroid receptors. In line with this theory, therefore, androgens could function without androgen receptor (AR), and estrogens could function without ER, in both sexes. Indeed, Chen et al. () showed that estrogens regulate bone cell apoptosis, even in cells derived from estrogen receptor a/b double knockout (DERKO) mice, suggesting that estrogen stimulates AR in these cells, in the absence of any ER (). However, in contrast to these in vitro data, earlier in vivo studies clearly demonstrated that gonadectomized ERKO and DERKO mice are resistant to estrogen in vivo, despite the presence of ARs. Moreover, data at this meeting showed that bone of mice without functional ARs is also resistant in vivo to androgens (), whereas male ER a knockouts were earlier shown to be fully responsive to androgens (). Therefore, sex steroid replacement studies in gonadectomized sex steroid receptor knockout mice do not support the concept that either AR or ER is sufficient for full expression of estrogen or androgen action, respectively. Androgens may, however, inhibit bone loss in the gonadectomized female, and estrogens may inhibit bone loss in gonadectomized male mice (). This is in line with the presence of both types of sex steroid receptors in bone cells of both sexes of mice ().

Several possibilities, however, should be further explored. First, the existence of another type of sex hormone receptor that mediates common "nongenomic" action of these steroids or analogs should be considered. Such an hypothesis, however, cannot fully explain the in vivo data reported (). A second testable hypothesis that could explain most of the data assumes that Manolagas’ ANGELs might directly activate postreceptor signaling (eg, MEK/ERK pathway) induced by either androgens or estrogens, rather than by direct interaction with the full transcriptionally active steroid receptor. This would then, of course, explain the promiscuous behavior of such "angels."

Can even androgens become pure angels?

Androgens undoubtedly increase bone mass and do so probably by a combination of AR and ER activation; however, as estrogens in women, they may also increase the risk of androgen-dependent cancer cell growth. Can chemists transform the natural hormone into a bone-selective androgen and create a pure angel for male, and possibly even female, bone?

Indeed, the well known antiandrogen, cyproterone acetate, widely used in Europe to treat relative androgen excess in women, showed clearly tissue-selective bone-sparing effects in mice without stimulation of seminal vesicles (). Other selective androgen receptor modulators (SARMs) were introduced in abstract M393 ().

Table of Contents

Osteoclast biology: M-CSF, RANKL, SRC, and new friends Patrick Ross, Department of Immunology and Pathology, Washington University School of Medicine, St. Louis, MO, USA, and Anna Teti, Department of Experimental Medicine, University of L'Aquila, L'Aquila, Italy M-CSF, RANKL, and OPG in osteoclast formation

Although the indispensable role of macrophage colony-stimulating factor (M-CSF) and receptor activator of NF-k-B ligand (RANKL) remains firmly established, new findings were presented that add important information relating to these and other agonists of osteoclast formation. Thus, osteoblast-specific targeting of only the membrane-bound form of M-CSF is sufficient to rescue the phenotype of the op/op mouse (), establishing for the first time in vivo that nonsoluble M-CSF is capable of inducing significant signaling in myeloid precursors. The role of individual tyrosine residues in the cytoplasmic tail of the M-CSF receptor, c-Fms, was dissected by using a retroviral transduction system to insert a chimeric receptor comprising the external domain of the erythropoietin receptor coupled to the transmembrane and cytoplasmic regions. Surprisingly, it appears that two positions, Y559, a binding site for members of the c-src family and Y807, of unknown function, are particularly important for initiating signal transduction following cytokine ligation (). Analogous to the study that reported targeted expression of M-CSF, a mouse, in which the collagen Ia1 promoter was used to direct expression of the RANKL decoy receptor osteoprotegerin to osteoblasts, had osteopetrosis more profound than is seen when the RANKL inhibitor exists as a serum protein (). The suggested explanation for this disparity is that serum-derived osteoprotegerin (OPG) is unable to blunt RANKL action at sites of periosteal osteoclast formation and function, while the osteoblast-specific form can perform this function. A potential method for blocking RANKL-receptor activator of NF-k-B (RANK) signaling was developed using data obtained from the recent publication of a crystal structure between the RANK cytoplasmic tail and a peptide derived from TRAF6 (), a known adaptor required for RANKL-RANK signaling. The same TRAF6 peptide, coupled to an amino acid sequence rendering it membrane permeable, inhibited osteoclastogenesis in a dose-dependent manner (). The previous uncertainty concerning the capacity of tumor necrosis factor a (TNFa) to stimulate osteoclastogenesis in a RANKL-independent manner was resolved in studies using a TNFa-transgenic mouse crossed onto a RANK-deficient background (). The animals had severe osteopetrosis and no osteoclasts, while wild-type transgenic animals expressing TNFa had accelerated bone resorption (and severe arthritis, which was mechanistically unrelated to osteoclast function). Of interest, the high in vivo levels of TNFa increased both the number and osteoclastogenic capacity of CD11b+precursors. In sum, these studies support the earlier report that TNFa is a powerful synergist of RANKL signaling, but that it cannot itself stimulate osteoclast formation. Finally, RANKL levels are increased two- to three-fold on T, B, and marrow stromal cells isolated from untreated postmenopausal women, as opposed to premenopausal or estrogen-treated postmenopausal women. Furthermore, the increased RANKL values are positively correlated with serum carboxy-terminal telopeptide, a marker of bone resorption. ()

Novel regulators of osteoclastogenesis

Several interesting studies uncovered new regulators of one or more aspects of bone cell biology. A possible molecular explanation for the increase in osteoclast number and concomitant osteoporosis seen in cases of excess thyroid function is provided by the observation that both T4 and T3 (the latter hypothesized to be generated by stromal cells that express the key T4 deiodinase) can stimulate RANKL expression by osteoblasts (). In a related study, mice lacking the receptor for thyroid-stimulating hormone (TSH) and those heterozygous for the gene (with the latter maintaining their euthyroid status) manifest a high turnover osteoporosis and focal sclerosis (). Both mesenchymal and myeloid precursors express the TSH receptor, a finding that explains the in vitro observation that absence of TSH receptor leads to enhanced formation of both osteoblasts and osteoclasts. Furthermore, recombinant TSH suppresses both osteoclast and osteoblast differentiation in vitro. Data from mice lacking pax5, a transcription factor required in B-cell formation, confirm earlier studies indicating that B cells and osteoclasts probably share a common early precursor of the hematopoietic lineage. Thus, the pax5 null mice are markedly osteopenic, with increased numbers of osteoclasts, but unchanged osteoblasts. Precursors from pax5 (-/-) mice generated many more functional osteoclasts than those from wild-type littermates and, interestingly, could survive in culture for up to 6 weeks without exogenous M-CSF (). Earlier in vivo studies had revealed that lipopolysaccharide- (LPS) induced bone resorption required signaling through the p55 TNFa receptor, presumably consequent on stimulation of TNFa release by the bacterial product. LPS is also an agonist for the toll-like receptor 4 (TLR4), and several studies analyzed the role of this receptor in osteoclast function. Studies in osteoblasts showed that LPS stimulates expression of RANKL, while suppressing that of OPG (). Using specific pharmacological inhibitors, the authors were able to show that suppression of OPG requires COX2 -mediated up-regulation of PGE2, while RANKL expression involves Cai/protein kinase C- (PKC) dependent activation of the ERK pathway. Separately, and in agreement with a recent publication (), the bacterial product CpG-DNA, a ligand for TLR9, enhances RANKL production by osteoblasts. Furthermore, double-stranded viral DNA (ds-DNA), a TLR3 ligand, also accelerates both osteoclast formation and RANKL expression in mesenchymal cells. Confirming the importance of increased RANKL production, exogenous OPG completely blunts the effects of either form of DNA. Finally, secreted frizzled-related protein 1 (sFRP-1), a molecule that modulates wnt/frizzled signaling, blocks in vitro osteoclastogenesis, apparently by binding to both TNFa and RANKL ().

Mechanisms of parathyroid hormone-induced osteoclastogenesis

It is now clear that parathyroid hormone (PTH) circulates in plasma in a number of forms, including one that comprises residues 7-84, which fails to bind to the PTH/PTHrP receptor (PTH1R) type 1, but still is biologically active. Furthermore, the intact hormone acts by stimulating the protein kinase A (PKA) and/or PKC pathways. In back-to-back reports, forms of PTH that activate the PKA pathway and, hence, cyclic adenosine monophosphate (cAMP) production, stimulate osteoclast formation in cocultures of stromal cells and myeloid precursors. The mechanism of this effect involves stimulating RANKL while suppressing OPG expression by the stromal cells. In contrast, either a specific isoform of PTH or the phorbol ester, phorbol myristate acetate (PMA), both of which exclusively activate the PKC pathway, suppress in vitro osteoclastogenesis and raise levels of OPG, while not altering those of RANKL (). These results were confirmed using PTH receptor knockin mutations, which selectively activate the PKA or PKC pathways. The second study () first showed that 7-84 PTH inhibits osteoclast differentiation in both whole marrow cultures and when purified marrow macrophages are stimulated with M-CSF and RANKL. Both osteoblasts and myeloid precursors isolated from mice lacking the PTH1R still exhibit high levels of specific binding for 7-84 PTH, indicating the presence of a novel C-terminal receptor on both cell types and suggesting that, unlike the osteoclastogenic capacity of PTH containing an intact N terminus, C-terminal fragment(s) can inhibit formation of bone-resorbing polykaryons. In sum, these studies have uncovered a previously unknown level of complexity with regard to the bone resorptive actions of PTH.

GP130 receptors and osteoclast formation

The membrane receptor gp130 is the transducing component of a number of heterodimeric cytokine receptors, the most relevant of which in bone biology are those that contain the specific subunits that bind interleukin 6 (IL-6) or IL-11 (IL-6Ra and IL-11Ra). Data from mice lacking IL-6Ra, IL-11Ra, or both receptors, confirm that absence of IL-6Ra has little impact on bone phenotype, while lack of IL-11Ra results in low bone turnover and high trabecular bone volume in both male and female mice, with the double knockout reflecting that of mice lacking IL-11Ra (). Importantly, osteoclast surface was decreased significantly. In a follow-up study (), mice carrying point mutations in the cytoplasmic tail of gp130, resulting in inactivation of either the STAT or ERK pathways, were generated by knockin technology. Mice incapable of activating the ERK pathway have decreased bone volume and high bone turnover, showing that gp130-mediated ERK signals are important for maintenance of normal bone volume. Mice that cannot transduce gp130 signals to the STAT pathway have unaltered histomorphometric indices, indicating that STATs are not required for regulation of trabecular bone volume. In contrast, longitudinal bone growth is substantially decreased in the absence of gp130-initiated STAT signals. Together these reports reveal nonoverlapping and important roles for signals transduced by the gp130 family of heterodimeric cytokine receptors.

Integrins and the cytoskeleton

Many studies have tried to identify new signaling pathways to explain established results. For instance, while we already knew that occupancy of the aVb3 integrin receptor prevents osteoclast apoptosis, it was reported that b3-null osteoclasts survive better than their b3-expressing counterparts when kept in suspension, suggesting that the unoccupied integrin triggers apoptotic signals otherwise silent in this cell (). Other downstream aVb3 integrin receptor signals are now better understood. The kinase domain of PYK2 plays an important role in the dynamics of osteoclast polarization. Thus, the autophosphorylation site at Y402 is central to the organization of the ruffled border, but has little effect on generation of the sealing zone (). We also learned that structural and functional differences distinguish the roles of c-Cbl and Cbl-b (), members of a family of proteins recruited to integrin-containing complexes. While c-Cbl is an important regulator of podosome disassembly and osteoclast motility, deletion of this gene in mice does not lead to significant phenotypic alterations in bone. In contrast, a loss of function mutation in Cbl-b results in decreased trabecular volume, without changes in bone formation parameters. Consistent with this latter observation, Cbl-b-deficient osteoclasts resorb more bone than wild-type cells. Furthermore, while the bone-resorbing capacity of c-Cbl (-/-) osteoclasts can be stimulated by IL-1, that of cells lacking Cbl-b (-/-) is maximal in the absence of exogenous stimulation. Podosomes, found uniquely in osteoclasts, are key structures in osteoclast adhesion, polarization, and motility. Wiskott-Aldrich syndrome protein (WASP), a molecule that mediates actin reorganization, has been identified in osteoclast podosomes (). Attesting to the function of this protein, osteoclasts derived from WASP-deficient mice cannot assemble podosomes and they form incomplete actin rings and fail to migrate on bone slices. Introduction of eGFP-WASP, which localized to the podosome core, led to reversal of the abnormal phenotype.

c-Src and friends

The role of c-src in osteoclast function is still extensively investigated, with the development of promising new inhibitors in progress. Inhibitors of the c-src adenosine triphosphate (ATP) binding site () restrain osteoclast function, cancer cell activity, and bone metastases, both in vitro and in vivo. Given their enhanced specificity toward nonreceptor tyrosine kinases, including most members of the c-src family itself, these novel inhibitors could presage innovative therapies to suppress osteoclastic bone resorption action.

Mechanisms of osteoclast survival

New observations suggest that the capacity of M-CSF, TNFa, or RANKL to increase osteoclast survival is mediated, at least in part, via the mTOR/S6 kinase (). Colony-stimulating factor 1 (CSF-1) treatment of osteoclasts triggers the PI3K/PDK1/PKB effector complex, but fails to activate IkB. In contrast, the antiapoptotic signals induced by TNFa and RANKL only marginally affect PKB, but they enhance IkB phosphorylation. Nevertheless, all three pro-survival cytokines activate mTOR and, hence, phosphorylation of the ribosomal protein S6, an event inhibited by the mTOR inhibitor, rapamycin. When osteoclasts are exposed to rapamycin, they undergo apoptosis, exhibit increased caspase activity, and have reduced capacity to resorb bone. Thus, distinct upstream survival signals converge downstream on a common mechanism shared by the PI3K and IkB modulators. Further pro-survival signals are transmitted in osteoclasts by the small GTP-binding protein, rac1 (). Overexpression of a dominant negative mutant of this ras family member blunts the antiapoptotic effect of CSF-1, whereas a constitutively active rac1 mutant significantly enhances osteoclast survival in a PI3K- and mTOR-dependent manner. The role of rac1 and related small GTPases was discussed in an excellent state-of-the-art lecture given by Mike Rogers, who reviewed the mechanism of action of bisphosphonates. While the antiresorptive abilities of these compounds have been known for several decades, much has been learned recently regarding their mechanisms of action. The two classes of bisphosphonates, distinguished by the absence or presence of nitrogen, affect osteoclastic bone resorption by distinct pathways; however, they share a common mechanism of action, initiation of apoptosis. Bisphosphonates lacking nitrogen, like clodronate, form metabolites that block the mitochondrial adenosine diphosphate (ADP)/ATP transporter, alter mitochondrial potential, and lead to cytochrome c release and, thus, caspase activation. Alternatively, the nitrogen-containing bisphosphonates, such as alendronate, exert their principal effect via inhibition of the mevalonate pathway, resulting in blocked prenylation of small GTP-binding proteins. As a result, these proteins are prevented from targeting appropriate membranes in osteoclasts. Since the many classes of small GTP-binding proteins regulate important processes, including cytoskeletal organization, membrane trafficking, and apoptosis, the net effect is loss of cellular function.

What else is new?

We learned that it is "RAINing" on osteoclasts (). RAIN is a "RANK-associated inhibitor," a novel protein with no identifiable domains or motifs, which interacts with RANK, TRAF2, TRAF5, and TRAF6. Transfection of RAIN in Raw 264.7 cells reduced RANKL-induced NF-k-B activity by 50% and blocked osteoclast differentiation. It is probably too early to say that RAIN represents a new important RANK-induced negative regulator of osteoclastogenesis, since no in vivo data were presented, nor was RAIN expression and function investigated in authentic cells. Should this pathway be confirmed, it would represent a second RANK-dependent negative autoregulator in osteoclasts, paralleling the recently reported interferon b-dependent suppression of autoregulation of RANK signaling that is induced by RANKL (). Finally, the existence of a protein secreted by osteoclasts that contributes to osteoblast recruitment at the site of bone resorption was also reported. The chemokine Mim-1 enhances migration and differentiation of osteoblasts, regulates ERK phosphorylation in these cells, stimulates matrix mineralization, and increases Cbfa1 binding to the osteocalcin promoter. Of interest, Mim-1 also increases binding of VDR to its response element in the osteocalcin promoter in a vitamin D3-independent manner. Again, while these data need to be confirmed and information must be provided on the Mim-1 receptor, the structure of which is, at present, still unknown, the findings are potentially exciting.

Promising new areas

The results presented this year suggest that the major foci of research are still the RANKL-RANK pathway and regulation of the podosome/actin ring assembly. The next level of discovery will require extending our studies to encompass new mechanisms. As always, we can learn lessons by examining the various forms of osteopetrosis, which represent a generic failure of osteoclast function. For example, patients with one of a variety of mutations in the cDNA coding for the a3 subunit of the osteoclast vacuolar ATPase (ATP6i-deficiency) have osteoclasts with normal morphology, intense TRAP staining, and a decreased, but not abolished, ability to resorb bone (). These findings suggest that either the function of the a3 subunit of the V-ATPase is not completely abrogated by the mutations or that a compensatory mechanism may partially complement its inactivity. One approach to overcome this failure of osteoclastic acid secretion would be to identify a pharmacological reagent that stimulates proton release in these patients. New mutations of the osteoclast-specific chloride channel gene ClCN7, representing human type II autosomal dominant benign osteopetrosis, were also reported (). Since it is not possible to correlate genotype and phenotype in these patients, we cannot yet explain why this form of the disease has such a heterogeneous phenotype. Further aberrant osteoclast phenotypes are presently under investigation, an approach that undoubtedly will yield important new information in the future. In the past, one of the major obstacles to rapid advances in osteoclast cell and molecular biology has been the lack of readily available cellular models. Substantial advances have been made in recent years, and osteoclast precursors from different species, including human, are now available. Of particular interest, abstract M270 () described commercially available human osteoclast precursors, derived from hematopoietic CD34+ progenitors, which turn into mature osteoclasts upon treatment with CSF-1 and RANKL. The ready availability of these cells could be a further tool in improving our understanding of the biology and regulation of this intriguing and ever fascinating cell type.

Conclusions

There is something new in osteoclasts, but, currently, progress is less exciting than that seen in the osteogenic field, where several new genes have been reported as relevant to bone formation (). This disparity arises, at least in part, from the relative maturity of osteoclast research, coupled with the fact that there are few true anabolic compounds, while several antiresorptive drugs are available. Does all this mean we should change fields? To this, we say NO WAY! Despite our huge store of knowledge, much still remains to be uncovered. This thought should give comfort to the many researchers out there still hoping to get that great score on their grant and/or write that killer paper. In summary, while the amount of important information relating to osteoclast function at the meeting was relatively sparse, it is important that we continue to develop new insights in this area, in order to assist in the development of novel strategies for the treatment of bone diseases.

Table of Contents

Puzzles of cartilage biology Ernestina Schipani, Endocrine Unit, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA

The essential role of the PTH/PTHrP receptor (PTH1R) in delaying chondrocyte differentiation is an established finding, and it is well documented that cAMP is the major downstream effector of this receptor (). However, it is also known that PTH1R is able to activate the phospholipase C (PLC) signaling pathway. It has now been reported () that the PLC pathway may indeed contribute to mediating PTH1R action in the growth plate. A knockin mutant mouse, in which the wild-type PTH1R had been substituted with a mutant receptor (DSEL), was impaired in its ability to activate PLC and it displayed a significant delay in chondrocyte differentiation. These two downstream effectors of PTH1R, cAMP and PLC, seem thus to have opposite or distinct effects on growth plate development. Of interest, lack of inositol tris-phosphate (IP3) signaling increases PTH1R levels in chondrocytes. It would be interesting to determine whether the cartilage phenotype of the DSEL mutant mouse is mainly the result of increased level of PTH1R expression, or whether other mechanisms, yet to be identified, have a role as well (more likely!) ().

A mystery solved?

Pseudohypoparathyroidism Ia (PHP-1a), defined as resistance to parathyroid hormone (PTH) and other hormones associated with Albright's hereditary osteodystrophy, and pseudopseudohypoparathyroidism (ie, Albright's osteodystrophy only) can be caused by the same mutation in the Gsa gene (GNAS1) and generally occurs in the same family. In order to solve this apparent paradox, one mutation/two diseases, a combination of both tissue-specific imprinting and haplo-insufficiency mechanisms has been invoked as a plausible explanation (). However, analysis of bone specimens isolated from mice carrying only one functional GNSA1 allele has been, thus far, inconclusive. In an elegant chimeric growth plate model, chondrocytes carrying only one functional GNSA1 allele of either paternal or maternal origin underwent hypertrophic differentiation a bit more quickly than surrounding wild-type cells (). The result was consistent with the finding that stimulation of cAMP through the PTH1R delays chondrocyte differentiation and strongly supports the hypothesis that haplo-insufficiency of GNSA1 may be a critical pathogenetic event in Albright's hereditary osteodystrophy.

The importance of being mineralized

In the growth plate, hypertrophic chondrocytes mineralize their surrounding matrix before being replaced by bone. The molecular mechanisms that regulate this process are an active area of investigation (), but virtually unknown is the biological impact of mineralization on chondrocyte differentiation and death. Mice lacking DMP-1 (), which is a noncollagenous matrix protein present in dentin that is also expressed by osteoblasts, osteocytes, and hypertrophic chondrocytes, are dwarf, and they display a striking growth plate and bone phenotype. Bones are undermineralized, and osteoblast differentiation appears to be impaired; the metaphyseal region is widened, the chondrocyte hypertrophic region is highly disorganized, and the secondary ossification center is delayed, all features somehow reminiscent of rickets. In mice lacking the vitamin D receptor (), the expansion of the poorly mineralized hypertrophic zone, which is a typical feature of the rachitic phenotype, appears to be caused, at least in part, by decreased apoptosis of late hypertrophic chondrocytes. It would be interesting to investigate whether DMP-1 is indeed involved in this process.

TNFa and osteopontin "join" efforts

The pathogenesis of rheumatoid arthritis (RA) is a complex one and still largely unclear (). Integrins have been known to be involved; it is thus not surprising that osteopontin, a protein with a myriad of actions, such as being a ligand for integrins, has been found in the synovial fluid of patients with RA. It is also well documented that tumor necrosis factor a (TNFa) plays a central role in many inflammatory processes, including RA. An interesting and potentially very important link between TNFa and osteopontin in controlling the fate of articular chondrocytes has now been reported (). The findings are simply striking: osteopontin deficiency suppresses TNFa-mediated chondrocyte cytotoxicity and protects joints against destruction in anti-type II collagen antibody-induced arthritis. In other terms, TNFa action requires osteopontin. This finding may be related to the widespread abnormalities of macrophage function in osteopontin knockout mice. The possibility of osteopontin being a therapeutic target for RA treatment becomes more appealing.

More about the RANK/RANKL couple

Receptor activator NF-k-B ligand (RANKL) and receptor activator NF-k-B (RANK) are very well known as the critical molecules that allow the physical contact between osteoblasts and osteoclasts and, consequently, usher in a drama, (ie, osteoclast recruitment/activation and bone resorption). The activity of this couple in chondrocytes appears to be more constructive. RANK (-/-) mice have a significant decrease in growth plate chondrocyte proliferation; conversely, the hypertrophic zone is expanded, very likely as result of both decreased chondroclast function and reduced chondrocyte apoptosis (). Is there any link between the latter two phenomena? More importantly, RANKL synergizes the activity of bone morphogenetic protein 2 (BMP-2) as a positive modulator of chondrocyte nodule formation in vitro and upregulates expression of specific chondrocyte genes, such as Sox9 and COL2. Finally, RANK (-/-) mice heal their bone fractures more slowly than normal animals or op/op mice (an independent osteopetrotic control) do, with no cartilage formation at the fracture site. We look forward to learning what we still do not know about the RANKL-RANK couple!

An unusual "rescue" mechanism

It is often stated that the size of an organ depends on two variables: cell number and cell size. This is only partially true for bone and cartilage, tissues in which matrix is as important as cells (at least from a quantitative point of view!). With this in mind, it is not surprising that it is possible to increase the size of a growth plate by increasing chondrocyte matrix production. But, let's proceed in order. Achondroplasia, the most common form of human dwarfism, is caused by constitutively active FGFR3 receptors, and transgenic mice carrying these mutant receptors are indeed dwarf and they display a severe decrease in proliferation of growth plate chondrocytes, in addition to a delay in the appearance of the secondary ossification center. The achondroplasia phenotype is somehow phenocopied by mice that lack C-natriuretic peptide (CNP), and numerous lines of evidence strongly suggest that CNP is a local positive modulator of growth plate development (). Can the achondroplasia phenotype be rescued by CNP? The answer is a qualified "yes," but with an unusual mechanism (). Chondrocyte overexpression of CNP corrects the shortening of long bones in mice that express a constitutively active FGFR3 in the growth plate, and this is achieved not by modulating the decreased proliferation rate or the delay of the secondary ossification center formation, but by increasing matrix production. This is a novel mechanism, and there is probably some hope for the treatment of a devastating human disease.

A novel story is unfolding

The mammalian fetal growth plate is hypoxic, and the hypoxia-inducible transcription factor HIF-1a is critical for survival of hypoxic chondrocytes. HIF-1a protein is highly unstable and its levels are regulated by the von Hippel Lindau (VHL) protein, a novel E3-ubiquitin ligase. This ligase targets HIF-1a for proteosomal destruction. The lack of VHL in growth plate chondrocytes causes a unique phenotype ()—that is, a dramatic decrease of chondrocyte proliferation, an abundant matrix, and the appearance of unusually larger cells in the round proliferative layer near the top of the growth plate. These "hypertrophic" chondrocytes do not express collagen type X, and thus they differ from classical chondrocytes. The lack of VHL somehow uncouples cell size from cell proliferation. HIF-1a is the main target of VHL action; however, having said that, it needs to be determined whether the striking phenotype of VHL null growth plate is, at least in part, HIF-1a dependent. VHL is a tumor suppressor gene, and heterozygous loss of function mutations have been identified in the VHL gene of patients with von Hippel Lindau disease. We are thus left with the paradox that a tumor suppressor gene appears to be a positive modulator of chondrocyte proliferation. This is probably a novel story unfolding, but many pieces of the puzzle are still missing.

Table of Contents

Calciotropic hormones Gordon J. Strewler, Beth Israel-Deaconess Medical Center and Harvard Medical School, Boston, MA, USA PTH and PTHrP

Parathyroid hormone (PTH) is the classic calcium-regulating hormone, acting on bone in an endocrine fashion, but PTHrP is expressed prominently in osteoblasts, raising the possibility that it too regulates bone turnover. Conditional disruption of PTHrP expression in the osteoblast now indicates that, as anticipated, PTHrP is a local regulator of the osteoblast lineage (). PTHrP expression was disrupted in osteoblastic cells that express the cre recombinase driven by a 2.3-kb fragment of the murine a1(I) Col promoter. The resulting mice have normal levels of calcium and phosphorus, but they have osteoporosis with defective osteoblast function and reduced numbers of osteoclasts. Apoptosis of osteoblasts is increased and the formation of mineralized nodules is reduced; thus, both the origin and the fate of the osteoblast lineage may be regulated locally by PTHrP. Further work will be required to disclose the organization of the local pathway thus identified. It is possible that a local action of PTHrP was the evolutionary prologue to the eventual takeover of calcium homeostasis by PTH; if so, understanding it may well be essential to unraveling the anabolic effects of intermittent administration of PTH.

In the meantime, we have a deeper understanding of how PTH and PTHrP accomplish their work. The mutation of four amino acids in the second intracellular loop of the PTH1 receptor largely eliminates signaling through phospholipase C (PLC) and protein kinase C (PKC), leaving signaling through cAMP and protein kinase A (PKA) intact. This receptor has been introduced into mice by a knockin strategy, replacing the normal PTH receptor (). Cartilage rudiments prepared from mice that bear two copies of the mutant receptor respond normally to PTH(1-84) or to a PKA-selective analog of PTH with accelerated chondrocyte proliferation and slowed maturation of the hypertrophic zone; phorbol ester had opposite effects (). Thus, cAMP is the predominant second messenger for the effects of PTHrP and PTH on cartilage, but the receptor can exert either a stimulatory or an inhibitory effect on chondrocyte differentiation, depending on the specific downstream signal. Using the same mutant mice and selective agonists, PKA was shown to be the second messenger for effects of PTH on bone resorption in mouse bone marrow; again, PKC activation by phorbol ester had an opposite effect, increasing expression of osteoprotegerin (OPG) (). Finally, introduction of a constitutively active form of the G protein, Gaq, produces fracturing osteoporosis with very low bone formation rates and apparently normal osteoclasts (). Osteoblasts from these mice differentiate poorly in vitro, and it was suggested that PLC/PKC signaling may inhibit osteoblast differentiation.

PTH blocks the apoptosis of osteoblasts (), and the work done by osteoblasts, thus spared from cell death, is thought to be important to the anabolic effect of PTH. Transfection of dominant negative transcription factors was used to show that the antiapoptotic effect of PTH requires the transcription factors CREB and runx2/Cbfa1. Microarray analysis identified Bcl-2 as a gene whose transcription is induced by PTH, and overexpression of Bcl-2 rescued osteoblasts from apoptosis induced by etoposide (). PTH, however, also promotes the proteosome-mediated degradation of runx2/Cbfa1 (). Both expression of wild-type runx2 and an inhibitor of proteosomal proteolysis extend the duration of PTH-mediated inhibition of apoptosis, raising the possibility that the duration of apoptosis inhibition is limited by effects of PTH on the turnover of runx2/Cbfa1 ().

PTHrP is produced in vascular smooth muscle in response to vascular injury and has complex effects on the proliferation of vascular smooth muscle cells (). PTHrP could thus have a role in neointimal proliferation, a central feature of restenosis after endothelial injury during angioplasty. Rat carotid arteries were injured with a balloon; to block restenosis, a PTHrP analog deficient in the nuclear localization signal (and thus designed to have an antiproliferative effect) was then delivered using an adenoviral vector (). The PTHrP analog completely prevented formation of a neointima in the rat model. This striking result justifies further trials in standard large animal models of vascular injury associated with angioplasty.

Calcitonin

Disruption of calcitonin receptor expression by homologous recombination is, unexpectedly, an early embryonic lethal mutation (). Mice bearing two copies of a calcitonin receptor null allele do not survive beyond embryonic day 9.5, about the time of implantation. The result is surprising because both known ligands of the receptor, calcitonin and amylin, have been deleted in the mouse, and both phenotypes are mild. Thus, either nonredundant actions of calcitonin and amylin are simultaneously necessary at the time of implantation or there is an additional, as yet unidentified, ligand for the receptor. The former possibility can readily be tested by crossing calcitonin and amylin knockout mice.

The calcium-sensing receptor

Gain-of-function mutations in the parathyroid calcium sensing-receptor (CaSR) are associated with hypocalcemia and hypercalciuria (). Some cases of idiopathic hypercalciuria with renal stones might result from CaSR mutations; children from families with early-onset urolithiasis were therefore screened (). One proband out of 12 had a His994Tyr mutation in the carboxyl-terminal tail of the receptor, which segregated with hypercalciuria in the proband's family, and which, when expressed in HEK293 cells, increased the sensitivity to calcium. Why would this point mutation improve the affinity of the receptor for calcium? The authors postulated that the tyrosine residue introduced by the mutation might be a substrate for phosphorylation. Upon transient transfection, there was indeed a small increase in tyrosine phosphorylation compared to the wild-type receptor. Affected family members had normal levels of calcium and PTH. It remains to be determined whether the phenotype is truly kidney-specific (and why) or whether other physiological compensations for hypercalciuria obscure an effect of the mutation on calcium-sensing by the parathyroid glands.

Mice with null mutations of the CaSR have severe hyperparathyroidism and disorders of cartilage and bone (), as do infants with neonatal severe hyperparathyroidism (). Are these disorders the consequences of high PTH levels or does the CaSR, which signals in cartilage and bone cells, have an essential role in the physiology of either tissue? To answer this question, CaSR (-/-) mice were crossed with GCM2 (-/-) mice, which lack parathyroid glands (). The double knockout animals did not have hypercalcemia and their cartilage and bones appeared normal (). There is thus currently no evidence for a role of the CaSR in bone or cartilage.

Table of Contents

Phosphate homeostasis: The search for phosphatonin continues Rajesh V. Thakker, Nuffield Department of Clinical Medicine, University of Oxford, Oxford, UK

Phosphate homeostasis involves several organ systems, which include the gastrointestinal tract, kidneys, bone, and parathyroids. Some of the mechanisms (eg, the phosphaturic action of parathyroid hormone [PTH]) are well known, but the story is far from complete, and it is likely that other hormone(s), referred to as phosphatonin(s) (), will be involved. Much has been learned recently about phosphate homeostasis from molecular genetic studies of the hereditary forms of hypophosphatemic rickets (eg, X-linked hypophosphatemic rickets [XLH], autosomal dominant hypophosphatemic rickets [ADHR], and tumor-induced hypophosphatemic osteomalacia, also referred to as oncogenous osteomalacia [OOM]) (). Furthermore, clinical and biochemical studies of XLH, ADHR, and OOM identified similarities that indicated that there was likely to be a common pathway involved in the etiology of these diseases. Thus, patients with XLH, ADHR, and OOM all have hypophosphatemia, an inappropriately low or undetectable serum 1,25-dihydroxy vitamin D concentration, a renal tubular phosphate leak, and defective skeletal calcification, which are usually associated with normal serum calcium and PTH concentrations. XLH was caused by inactivating mutations of a gene encoding a cell surface metalloprotease, referred to as the phosphate-regulating gene with homologies to endopeptidases on the X chromosome (PHEX) (); ADHR was identified as a result of missense mutations of the gene encoding fibroblast growth factor 23 (FGF23) (); and OOM tumors have been shown to express high levels of FGF23 and also PHEX (). Moreover, injection of recombinant FGF23 into normal mice led to urinary phosphate wasting and hypophosphatemia, and overexpression of FGF23 in transgenic mice resulted in hypophosphatemia and low serum 1,25-dihydroxy vitamin D. These dual actions of FGF23 have been shown, by studies in mice that have been injected with FGF23, to be the result of decreased expression of both the sodium-phosphate cotransporter type 2a (NPT2a) and the renal 25-hydroxy vitamin D 1a-hydroxylase (). In addition, recombinant PHEX seems to degrade FGF23 in vitro, and all of this is consistent with key roles for PHEX and FGF3 in the pathway of phosphate homeostasis.

Indeed, one could postulate a unifying hypothesis to explain the etiology of the hypophosphatemia in XLH, ADHR, and OOM, based upon a single enzyme- (PHEX) substrate (FGF23, phosphatonin) interaction, which results in excess phosphatonin either because of impaired degradation (in XLH and ADHR) or increased production in OOM (). For example, in XLH, the inactivating PHEX mutations fail to degrade FGF23 (or phosphatonin), whereas, in ADHR, the mutant FGF23 protein is not cleaved by PHEX, with the net result being an excess of FGF23. In contrast, in OOM, there is overproduction of FGF23 (ie, phosphatonin) such that it exceeds the capacity of normal PHEX activity. These increased levels of FGF23 would then inhibit the renal activity of the 25-hydroxy vitamin D 1a-hydroxylase and NPT2a, thereby leading to a decrease in 1,25-dihydroxy vitamin D and hypophosphatemia (). NPT2a is a key transporter of phosphate from the urine into the proximal tubular epithelial cells, and indeed the phosphaturic action of PTH is mediated by diminishing the availability of NPT2a on the apical surface of the renal proximal tubular cells (). This important role for NPT2a is further supported by data from knockout mice (-/-) that lack the NPT2a gene, and they develop hypophosphatemia, renal phosphate wasting, raised serum 1,25-dihydroxy vitamin D concentrations, and hypercalciuria (). NPT2a also has a similarly important role in phosphate homeostasis in man, as DNA sequence analysis of the NPT2a gene in 20 patients with hypophosphatemia, which is due to renal phosphate wasting and nephrolithiasis or a low bone mass, has revealed missense mutations that reduced NPT2a function in two patients (). All of these studies indicate that the pathway regulating phosphate homeostasis involves an enzyme (PHEX), a substrate (FGF23) that acts as hormone (phosphatonin), and a target transporter, NPT2a, in the kidney.

This hypothesis seems attractive, but proof is needed to show that FGF23 is phosphatonin by showing its elevated presence in the circulation of XLH, ADHR, and OOM patients. Furthermore, FGF23 should ideally be shown in normal individuals to have a feedback mechanism in response to alterations in serum phosphate. In order to undertake such studies, one requires an assay for FGF23, and several groups () have raised antibodies against FGF23 and developed enzyme-linked immunosorbent assays (ELISAs) to undertake such studies. The results show that FGF23 is present in the circulation of normal healthy individuals (), that it is elevated in XLH () and OOM patients (), and that successful removal of the tumor in OOM patients leads to a decrease in FGF23 () that precedes the increase in serum 1,25-dihydroxy vitamin D and phosphate. The detection of FGF23 in the serum of normal individuals is intriguing and suggests that it may also have a physiological role. This was investigated by altering the phosphate intake in 12 normal male volunteers (). Increased phosphate intake led to a rise in renal phosphate excretion, which was associated with a decrease in urinary calcium. These urinary changes were not accompanied by changes in serum calcium, phosphate, or PTH, but fasting serum FGF23 was significantly increased in response to phosphate loading (). All of these findings are consistent with a phosphate-regulating role for FGF23, and the findings in mice that have homozygous (-/-) deletions of FGF23 further support this. Thus, these FGF23 null (-/-) mice survived and developed elevated serum concentrations of phosphate, calcium, and 1,25-dihydroxy vitamin D (). These findings in man and mouse provide support for the theory that FGF23 is the phosphate-regulating hormone, phosphatonin.

However, some issues remain unresolved. First, if FGF23 is phosphatonin, then it should be cleaved by PHEX. Yet, cotransfection of expression plasmids containing PHEX and epitope-tagged human FGF23 in COS7 cells failed to detect PHEX-dependent cleavage of the FGF23 by Western blot analysis of conditioned media (). Furthermore, a survey of mouse tissues by quantitative polymerase chain reaction (PCR) revealed that the highest levels of FGF23 and PHEX transcripts were in bone. These data indicate that the elevated FGF23 levels in XLH may be the result of an indirect consequence of PHEX inactivation, whereby FGF23 accumulates in bone. So, if FGF23 is not phosphatonin, then are there other candidates? Indeed, there are, and these include, among others, matrix extracellular phosphoglycoprotein (MEPE) and secreted frizzled-related protein 4 (sFRP-4), which are both highly expressed in the tumors of patients with OOM. One study has further shown that infusion of biosynthetic sFRP-4 protein in mice with intact parathyroids significantly decreased renal phosphate reabsorption (). These data are consistent with sFRP-4 being a "phosphatonin." Much still remains to be elucidated in the pathways that regulate phosphate homeostasis, and it should perhaps come as no surprise that there may be more than phosphate-regulating hormone (phosphatonin). After all, in the control of plasma sodium (Na+), there are several known regulating hormones, including aldosterone, cortisol, atrial natriuretic peptide, and antidiuretic hormone, each with its own receptor and enzymatic pathways. Perhaps the situation with phosphate regulation will be equally complex and intriguing.

Table of Contents

Epidemiology of osteoporosis: It's not about white women Dennis M. Black, Department of Epidemiology and Biostatistics, University of California, San Francisco, CA, USA

Over the past 15 years, much has been learned about the epidemiology of osteoporosis in white women, particularly in women who are older than about age 65 years. The Study of Osteoporotic Fractures (SOF) in the United States and the EPIDOS study in France have been particularly useful in defining risk factors for fractures and for characterizing the relationship between bone mineral density (BMD) and fracture risk. Other studies among white women have included the Rotterdam study and a long-term study in Malmo, Sweden.

Populations other than white women have been less well studied. For example, our knowledge about the epidemiology of osteoporosis and risk factors in men and in non-white women is much more limited. Furthermore, our knowledge about short and long-term risk factors for fracture among younger postmenopausal women (ie, women aged 50 to 65 years) is very limited.

Osteoporosis in asian and black populations

In recent years, there has been increasing interest in osteoporosis among Asian women. It is estimated that within 50 years, more than 75% of hip fractures in the world will occur in Asia (). However, there is an interesting quandary that has not been resolved among Asian women: although they generally have lower BMDs than whites, they have similar vertebral fracture rates. Furthermore, despite their lower BMDs, hip fracture rates are lower among Asians than among whites. How much lower depends on the specific population: the lowest incidence has been reported by Xu et al. () in Beijing (about 12% that of whites in the United States), while the highest incidence has been reported in Hong Kong (about 75% that of whites in the United States).

Bone structure or geometry might possibly explain this discrepancy. Japanese-American women were compared to white women with respect to structural measures of femoral neck strength (). Three measures were assessed: bending strength, compressive strength, and impact strength, which were calculated from dual-energy x-ray absorptiometry (DXA) scans using DXA BMD, body weight, and femoral neck width. Although BMD was significantly lower in the Japanese-American women (as expected), the estimated strength of the hip was greater. It was concluded that the stronger femoral neck in the Japanese women could explain the lower hip fracture risk observed in this population. Other explanations, such as reduced risk of falling or better ability for self protection in the event of a fall, may also explain some of the difference.

Calcium intake is generally very low among Asians. Lau et al. () reported a randomized study of calcium supplementation to increase bone accretion in about 300 Chinese children, aged 9 and 10 years. The children were randomized into three groups: supplementation containing 1300 mg/d of calcium, supplementation containing 650 mg/d of calcium, and a control. The study found that, after 18 months, there was 1% to 2% additional bone gain in the supplemented groups, although the differences were not statistically significant. The authors concluded that larger and longer term studies are needed to test whether calcium supplementation can help increase bone mass (and possibly later bone strength) in geographic regions with low dietary calcium intake.

Osteoporosis in India has received relatively little attention. Reddy et al. () conducted a survey of women of Indian ancestry living in the United States. The authors concluded that the bone density of these women was lower than that of the bone density of white women. They also concluded that specific normative data were warranted for this population.

Rates of hip and vertebral fracture are lower among blacks (both women and men) than among whites. These differences persist, even after adjusting for such factors as age, body weight, estrogen use, and so forth. Unlike Asians, blacks also have higher BMDs and probably larger bones than whites, which may explain some of the difference. Bone loss was explored in whites versus blacks by comparing about 6000 women from the SOF to 482 women who were newly-recruited for a parallel study (). Black women not only had higher hip BMDs, but they also had less mean bone loss than white women, even after adjustment. However, among both white and black women, there was an increase in bone loss with increasing age. For example, in black women younger than age 75 years, the mean annual bone loss at the femoral neck was about 0.1%, compared to about 0.4% for women older than age 75 years. These increases in bone loss with age in all women were evident in the total hip, as well as in the femoral neck.

Epidemiology and diagnosis of osteoporosis in men

The incidence of hip fractures in men is about one-half that of women at the same age. Given the shorter life expectancy for men, the lifetime risk of hip and other clinical fractures is only about one-third that of women. Studies in Europe suggest that vertebral fracture prevalence among men and women in their fifties is similar, but that new vertebral fractures are much more common in women after that age. However, the absolute levels of risk in men, particularly in some groups of high-risk men, are still quite high and remain an important clinical problem.

Diagnostic criteria for osteoporosis (based on T scores) are well established (although exact values remain controversial) in women. However, there is an ongoing controversy with regard to men and diagnostic criteria for osteoporosis. The relationship between BMD and fracture risk is well established for men, as is the fact that, at the same age, men have higher BMDs than women. What is not yet clear, however, is whether at the same absolute (areal) BMD, risk is lower, higher, or the same for men. If it is the same, this implies that the same absolute cutpoints (in g/cm2) should be used, which therefore implies that the female reference data and T scores should be used in males. This controversy is further confused by the question of bone size: men have larger (and therefore stronger) bones than women. Therefore, it is not clear whether T scores should be calculated from male reference data bases or whether the female reference data should be used.

Data from the Dubbo epidemiologic study () were used to calculate volumetric BMD from DXA scans. Use of calculated volumetric BMD attempts to correct for differences in bone size between sex or race groups by estimating the volume of bone and then calculating true volumetric density. This is done by making assumptions from the two-dimensional (mineral per unit area) DXA scans. While areal BMD was lower in men than in women, volumetric BMD was similar. In a case-control analysis, women and men with hip fracture had similar volumetric, but not areal, BMD. These results suggest that use of volumetric BMD could lead to a more unified diagnostic criteria for osteoporosis. However, while suggestive, previous studies have not supported the additional value of volumetric BMD calculated from DXA scans, and further study is needed.

Other epidemiology studies of osteoporosis in men

Testosterone levels were studied in 1066 men (aged 19 to 89 years) in the MINOS cohort in France. As has been previously shown, both free and bioavailable testosterone levels decline strongly with age. "Hypogonadal" testosterone levels were defined as more than 2 SDs below the mean for young normals. Using this definition, these authors found that, after adjustment for age and body mass index, hypogonadal status was associated with lower BMD and higher levels of bone markers. In men older than 49 years, hypogonadal status was associated with impaired balance and neuromuscular status. In men older than 69 years, hypogonadal status increased risk of falls. These authors concluded that hypogonadal men could have increased risk of fractures as the result of a combination of decreased bone strength and increased risk of falls. This study has some limitations, particularly the large age range. Also, previous studies have shown inconsistent relationships between testosterone levels and osteoporosis. However, the results suggest that further study may be warranted and that intervention studies of testosterone supplementation in hypogonadal men should include evaluation of osteoporosis endpoints.

Geometric parameters of the hip were calculated from DXA scans that were obtained as part of the National Health and Nutritional Examination Survery (NHANES) study in 221 men and 331 women (). Family history of osteoporosis (defined as a mother's history of hip fracture or diagnosis of osteoporosis) was specifically considered, and it was found that those men (and women) with a positive history had thinner cortical areas in the hip and somewhat lower BMDs, which may be indicative of a higher risk of fracture. This study suggested that family history of hip fracture in men may prove to be a risk factor for hip fracture, as has been shown in women.

Forsmo et al. () examined smoking habits and their relationship to BMD in a large cohort (Nord-Trøndelag Health Study [HUNT]) in Norway. The authors reported that there was a strong association between smoking history (in pack years) and low bone density. This association seemed much stronger in men than in women, thus supporting previous studies that showed that smoking, although only a weak risk factor for osteoporosis in women, may play a much more substantial role in men. Similarly, the study of Chang et al. (), from the Dubbo epidemiologic study, showed that ankle, hand, and foot fractures were particularly strong predictors of future fractures in men.

There were several other noteworthy studies presented as oral presentations in the two epidemiology sections.

Fractures that occur due to major trauma (eg, motor vehicle accidents) are routinely excluded as endpoints in epidemiologic studies and clinical trials in osteoporosis. The relationship of traumatic fractures to BMD was examined in the SOF (). The relationship of BMD to fracture risk was similar for nontraumatic and traumatic fractures. Women with a traumatic fracture were at increased risk for future nontraumatic fracture. It was concluded that fractures that occurred as a result of increased trauma should be included as endpoints in future studies. Kanis et al. () examined how many deaths could be prevented if hip fracture incidence could be reduced by 50%. The authors presented a mathematical model designed to estimate the proportion of deaths following hip fracture that could be attributed to the hip fracture. They suggested that 0.5% of all deaths could be avoided if hip fractures could be reduced by 50%. However, this number depended critically on their assumptions concerning avoidable deaths, and to test this assumption would require a trial of over 2 million  patients. Some recent studies have suggested that higher magnesium intake may be associated with higher levels of bone mass. Data from the Women's Health Initiative Observational Study (WHI-OS) were used to examine this question (). The study included over 89,000 women who had magnesium intake assessed by dietary questionnaire at baseline. The women were followed for fractures. The authors found that higher magnesium intake did not decrease fracture risk and, in fact, that wrist/lower arm fractures may be more common in those with higher magnesium intake. Because osteoporosis has been previously associated with pernicious anemia, Tucker et al. () examined the relationship between plasma vitamin B-12 levels and BMD in the Framingham Offspring study. The authors found that very low vitamin B-12 levels were associated with lower BMD in both men and women. They concluded that more work is needed to examine the effects of vitamin B-12 on bone biology, as well as the relationship between vitamin B-12, homocysteine levels, and osteoporosis. Using data from the SOF, the relationship between anticonvulsant use and rates of bone loss at the hip was examined (). The authors found that continuous users had slightly higher rates of hip and bone loss than nonusers and they concluded that anticonvulsant use should be part of the risk evaluation for osteoporosis in older women.

A meta-analysis of the use of oral corticosteroids and BMD and fracture risk was also reported (). Cumulative dose was related to bone loss, as well as fracture risk. However, the increased rate of bone loss with corticosteroid use was not sufficiently high to account for the large increase in fracture risk. This implied that there might be a direct effect of corticosteroids on bone quality, independent of their effect on bone density.

Table of Contents

Genetics of osteoporosis: QTLS meet candidates Stuart H. Ralston, University of Aberdeen Medical School, Aberdeen, Scotland, UK

With sequencing of the human genome now virtually complete, tremendous progress is being made in identifying the genes responsible for monogenic bone diseases. The candidate gene approach was successfully used to identify mutations in sodium-phosphate cotransporter type 2a(NPT2a) as the cause of a syndrome characterized by low bone density, hypophosphatemia, hypercalciuria, phosphaturia, and urolithiasis. Mutation screening of NPT2a resulted in the identification of missense mutations in 2 of 20 patients studied. Evidence for a cause-effect relationship was gained by functional studies showing that expression of messenger RNA (mRNA) for the mutant proteins in Xenopus oocytes impaired phosphate-induced transmembrane currents. Since NPT2a mutations were absent from most cases, the authors concluded that NPT2a was likely only one of the many genes that contributes to the pathogenesis of this syndrome. The results of this study have just been published ().

Following hot on the heels of the recent report of a homozygous deletion of the osteoprotegerin (OPG) gene as a cause of idiopathic hyperphosphatasia by Whyte et al. (), another group reported further missense and deletion mutations of OPG as the cause of this rare disease in a series of eight families with hypophosphatasia (). All mutations were clustered in exons 2 and 3 and were predicted to disrupt bonding between the cysteine residues in this region, which are important for normal conformation of the receptor activator NF-k-B ligand (RANKL) binding domain. Spondyloepimetaphyseal dysplasias (SEMDs) are a group of rare, genetically heterogeneous conditions characterized by defective growth and modeling of the long bones. Mutations in the collagen type II genes and ATP sulfurylase have been previously identified as causes of some SEMD variants. Linkage analysis was used to map the gene responsible for the so-called Missouri variant of SEMD to chromosome 11q22.3 (), and a positional candidate approach was then used to show that the disease was caused by a missense mutation in the MMP13 gene, an important enzyme for the regulation of cartilage degradation.

Over the past few years, various groups have successfully used linkage analysis in mice to identify several quantitative trait loci (QTL) for the regulation of bone mineral density (BMD). These regions are now being narrowed down and further characterized by the creation of congenic mouse strains, which involves backcrossing the QTL from one mouse strain to another by a selective breeding program. Elegant studies () used congenics carrying QTL from mouse chromosomes 1, 4, 13, and 18 to confirm the presence of genes that regulate BMD in these regions. These authors were also able to show, using micro computed tomography (CT), that the genes in these different loci exerted distinct effects on cortical and trabecular bone architecture at different skeletal sites. An important message to emerge from this work is that the DXA techniques that are being used for phenotypic characterization of the skeleton in human studies may be too blunt a tool with which to effectively detect the genes that regulate bone mass and bone structure in man.

One of the drawbacks of QTL mapping by linkage analysis in both mice and men is that the regions of interest generally extend over large stretches of DNA (typically 10 to 20 cM). This makes gene identification difficult and time-consuming. Microarray-based techniques are now being used to study transcript profiles of genes in linkage regions, in order to prioritize likely candidates (). Klein et al. () used the approach to good effect when they identified a four-fold upregulation of secreted frizzled-related protein 3 (sFRP-3) mRNA (Frzb1) in chondrocytes from congenic mice carrying a chromosome 2 QTL that predisposes to low BMD. Secreted frizzled proteins are strong candidates for BMD regulation because they inhibit bone formation by antagonizing wnt-induced low-density lipoprotein receptor-related protein 5 (LRP5) signaling. This was clearly demonstrated by the studies of Bodine et al. () who showed that mice with a knockout of sFRP1 had increased bone mass, apparently because of decreased osteocyte and osteoblast apoptosis. While Klein et al. () do not yet have direct evidence that allelic variation of Frzb1 is responsible for the effect they observed, one would imagine that this gene would be a fertile place to look for polymorphisms that are associated with BMD.

Less progress has been made in the search for BMD QTLs in humans, and the results that have been obtained are not as clear cut or consistent as many of us in the field would have hoped. A notable exception is the BMD QTL on chromosome 1p36, initially identified by Loretta Spotila, which has now been confirmed () as an important region for regulation of hip BMD in man. There is already very good evidence to suggest that TNFR2 (TNFRSF11A) is one of the genes responsible, although others may also exist. Evidence of a QTL was also found on 3p21, which was previously identified as being linked to BMD in a candidate locus linkage study by Duncan et al. (). The updated linkage results from the Indiana group () confirmed the previously reported spine BMD QTL over a broad region of chromosome 1q21, but identified two new hip BMD QTLs on chromosomes 14 and 15 that were not observed previously. There was disappointment for members of the Omaha group () when expansion of their population resulted in a decrease in LOD scores to values that were not statistically significant anywhere in the genome. These results most likely reflect the fact that regulation of BMD is genetically heterogeneous and that the statistical packages currently available for analyzing quantitative traits are not particularly good at dealing with this problem.

There was the usual mix of candidate gene studies. As well as the old favorites, such as VDR, COLIA1, and interleukin 6 (IL-6), new candidates, including OPG (), Fra-1 (), and SOST (), have emerged as potential regulators of BMD. Some of the best data on candidate genes continues to come from the Rotterdam study. In abstract 1221 (), the Rotterdam group presented convincing evidence for an allelic association between polymorphisms in the promoter of the IGF-1 gene and fragility fracture. The study included 7012 of the 7983 participants and is one of the largest association studies ever performed in the bone field. An interesting report from the same group () reported an association between BMD and polymorphisms in noncoding regions of DNA surrounding the SOST gene. Of special interest was the fact that the allelic association increased with age, which is consistent with the fact that the phenotype of sclerosteosis (caused by homozygous inactivating mutations of SOST) worsens with increasing age. Although one of the polymorphisms described lay in a downstream enhancer element that was known to be important for SOST regulation, the mechanisms by which SOST alleles regulate BMD remain to be determined.

Table of Contents

Bone quality is job one Ego Seeman, Austin and Repatriation Medical Centre, University of Melbourne, Melbourne, Australia Material and structural properties of bone

Bone is a lever that must be strong and stiff for weight bearing, yet light to allow movement, and flexible to allow absorption of impact energy. Mineralization of the bone tissue is responsible for the combination of stiffness, the resistance to bending and flexibility, and the ability to deform when energy is absorbed. The greater the tissue mineral content, the greater the stiffness and peak stress that the bone will tolerate. But 100% mineralized bone is brittle and will not “give” during impact loading; its “toughness” declines as the tissue mineral content exceeds about 65% (percent volume of bone that is mineral ash) ().

Strength and lightness are also achieved by structural design. Long bones are weight bearing, and stiffness is favored over flexibility by fashioning long bones with a medullary canal and placing the cortex of mineralized tissue distant from the central long axis, which confers resistance to bending (). The vertebral bodies are open, porous, mineralized interconnecting honeycomb structures of plates that function like springs that are able to store energy by deforming to one-half of their original length (). Lightness is achieved by the porous network of the vertebral body, while the strength is in this structure's ability to tolerate deformation without failing by fracture and to facilitate flexion, extension, and rotation of the whole vertebral skeleton of the upper body, should we decide to move back up into the trees.

Growth builds a bigger, wider, and longer long bone in males than in females, but the thickness of the cortex is similar or only slightly greater in males. The main sex difference in bending strength is achieved not by the thickness of the cortex, but by the further placement of the cortex from the neutral axis in males than in females (). This greater radial displacement of the cortex also produces a larger cross-sectional area (CSA) upon which compressive loads can be distributed. Growth builds a bigger vertebral body in males, not a more dense vertebral body, so that the greater peak load tolerated in males is because of the larger size, not higher density, and the load per unit area in young males and females is no different.

A mutation in the LRP5 gene causes high bone mass. This mutation was expressed in osteoblasts and osteocytes in mice (), resulting in increased volumetric bone mineral density (BMD) of the distal femur metaphysis with increased trabecular number, thickness, and connectivity. Diaphyseal volumetric BMD proximally was only slightly increased, but cortical thickness was increased as the result of increased periosteal circumference. Osteoclasts were not reduced, but bone formation indices were increased, with reduced apoptosis and increased bone formation rate. In transgenic mice with the LRP5 gene mutation, both the femoral neck and shaft had greater width, cortical thickness and strength, and stiffness (). For a given load, the stiffness of the bones was greater in the mutant than in the control. Bone formation (per unit surface per unit strain) was also greater in the mutant, suggesting that a given deformation produces more bone formation. The authors suggest that the genetically determined response to strain is altered and more sensitive, so that a smaller strain will stimulate more bone formation in this mutant.

Chinese are smaller in stature and have a shorter and narrower femoral neck than whites, with a narrower cortical thickness (before and after adjustment for femur length) (). The shorter femoral neck is associated not with differences in cortical thickness, but with a narrower femoral neck diameter, which suggests that the absolute movement of the periosteal and endosteal surfaces is race- and sex-specific. Cortical thickness is similar across race and sex. For any given femoral neck length, as the diameter of the femoral neck increases, the wider bone has a correspondingly thinner cortex, at least in females of both races, favoring minimal mass; the wider bone can maintain the same bending strength and cortical area by placing the cortical shell further from the neutral axis of the long bone. The clinical relevance of this observation is that persons with a wider femoral neck enter menopause with a narrower cortical shell, which may predispose to buckling as endocortical bone resorption proceeds during aging and after menopause.

Kalkwarf et al. () compared 165 black and 159 white women (aged around 22 years), measuring the radius at 4% and 66% of the radial endplate. After adjustment for differences in forearm length (greater in blacks), total bone area was 6.9% less, while cortical thickness was 7.4% higher as the result of a narrower endocortical diameter. Therefore, the smaller bone had a relatively thicker cortex. This is similar to the femur length-adjusted comparisons described above. Thus, the risk of fracture by buckling in Chinese and blacks, relative to whites, may in part reflect this structural difference.

Studies in 267 Japanese and 199 white women () suggest that the Japanese have greater mean values for measures of bending, impact, and compressive strength than do whites. However, the analysis was based on derived loads calculated from height and weight of the individuals. Whites are heavier, and leg length is greater than in Japanese. Asians have shorter legs than whites, but they have a similar trunk length, so the higher load may fall a shorter distance. Blacks have longer legs and a shorter trunk, so the lower load has further to fall.

The purpose of remodeling

Remodeling is aimed at replacing older damaged bone with younger bone. At some time before menopause, bone balance starts to become negative because of a reduction in the amount of bone formed in the bone morphogenetic unit (BMU), producing bone loss and structural damage. The same loads on bone are imposed on a structure diminished in CSA, so that the load per unit area increases. Trabecular thinning predominates in men, whereas loss of connectivity predominates in women. Residual strength of the vertebra decreases more greatly with loss of connectivity than with thinning. If the remodeling rate increases, older (more mineralized) bone is replaced by younger (less mineralized) bone, thereby reducing the stiffness of bone. If remodeling is slowed, more time is available for secondary mineralization and bone stiffness increases, while the tissue becomes more homogeneous and so loses its toughness or resistance to progression of microdamage. The roughness of the material is partly the result of variable collagen fiber orientation. Collagen cross-link maturity and crystallinity account for differences in bone strength in mouse strains of similar BMD that differ in bone strength (). Microdamage progression is facilitated by high mineral density and perhaps changes in collagen cross-linking. Reduced remodeling also reduces the removal of microdamage, so that microdamage burden increases both by increased production of microdamage and by decreased removal of microdamage.

Targeted remodeling is believed to be initiated by microdamage, which induces large strains that produce fluid flow in canaliculi and signals the initiation of local remodeling. However, in normal bone tissue, microstrains are about 2000, insufficient to initiate remodeling. Nicolella et al. () measured microstrain in regions of damage in matrix. Levels around 30,000 (3%) were recorded locally, values up to 15 times greater than global strains observed in vivo. These microstrains are hypothesized to initiate bone remodeling by osteocytes that are responsive to fluid flow-induced sheer stresses. Qiu et al. () reported that osteocyte number (per unit of bone area) has a higher sensitivity and specificity in identifying fracture and nonfracture patients than does trabecular bone volume.

Periosteal bone formation

Age-related periosteal bone formation offsets endosteal bone loss. Greater periosteal bone formation in men offsets endosteal bone loss more in men than in women. The lesser net decrease in BMD at the spine in men than in women is the result of deposition of more periosteal bone rather than the removal of less endosteal bone (). Structural failure emerges during aging because periosteal bone formation on the outside of the bone fails to completely offset the fragility produced by bone loss and architectural destruction inside the bone. A lower proportion of elderly men than elderly women have bone size and architectural and material properties, such as microdamage, tissue mineral density, loss of connectivity, porosity, and trabecular and cortical thinning below a critical level (or fracture threshold), where the stresses are greater than the bones ability to resist them.

Bone strength loss that is the result of bone loss was compensated for by increased bone size in a 19-year prospective study of 108 women (). Endocortical width increased by 1% per year, while periosteal width increased by 0.6% per year. The cross-sectional moment of inertia increased by 1.8% per year, while the strength index decreased by 0.7% per year. Computed tomography (CT) was used to assess geometric features (). CSA was increased at the vertebrae and proximal femur in elderly men. The increase in CSA was associated with maintenance of bone strength at the proximal femur, but not at the spine. Orwoll et al. () studied five monkeys (aged 2 to 12 years) and reported the presence of bone resorption on the periosteal surface rather than bone formation. These findings are consistent with the notion that periosteal expansion may be remodeling-based, rather than modeling-based. This is unlikely to be substantial, as the remodeling rate is so slow on this surface that loss of bone based on a tiny imbalance at each BMU is likely to be minimal.

Progress in understanding the pathogenesis of bone fragility depends on the definition of the material and structural properties of bone that determine its resistance to structural failure: bone size, cortical thickness, trabecular number, thickness, connectivity, tissue mineral content, microdamage burden, osteocyte density, and porosity. Age- and menopause-related abnormalities in remodeling rate and balance produce loss of the material and structural properties of bone. The challenge we face is to measure each of these material and structural determinants of bone strength and so better define the structural diversity responsible for fractures in the community.

Table of Contents

Anything new in the use of bisphosphonates in the management of osteoporosis? Socrates E. Papapoulos, Department of Endocrinology and Metabolic Diseases, Leiden University Medical Center, Leiden, The Netherlands

During the American Society for Bone and Mineral Research (ASBMR) meeting in San Antonio, three important issues regarding the use of bisphosphonates in the management of osteoporosis were discussed: the duration of bisphosphonate treatment, the concept of the total dose of bisphosphonate as a determinant for antifracture efficacy, and the use of combinations of bisphosphonates with other antiosteoporotic treatments.

Duration of bisphosphonate treatment

Practicing physicians need to know how long they can treat their patients with a bisphosphonate. This question is related to the specific pharmacological properties of the bisphosphonates that are embedded in bone and that have a very long skeletal half-life. Despite the fact that the biologically active bisphosphonate is that which is present on the bone surface, there have been concerns over the years about the safety of long-term use. Although up until now such concerns have not been justified, longer term data are clearly needed. Two studies with oral alendronate that was given for prevention or treatment of osteoporosis addressed this issue.

The Early Postmenopausal Interventional Cohort (EPIC) study was designed to investigate the effect of alendronate in early postmenopausal women (). The study had the following aims: to assess the long-term efficacy and tolerability of continuous alendronate therapy compared to placebo in women at risk; to compare the skeletal effects induced by continuous alendronate treatment with those induced by an estrogen/progestin (E/P) regimen; and to resolve the effect of alendronate on bone mineral density (BMD) and bone turnover following withdrawal of alendronate and E/P treatment. The EPIC study was a randomized triple-blind placebo-controlled trial that included 1609 women aged 45 to 59 years (mean, 53 years) with a mean BMD T score of -0.80; the women were at least 1 year menopausal. The following regimens were used: placebo, alendronate (2.5 mg/d), alendronate (5 mg/d), or open-label E/P for 4 years. This was a 6-year study in which there was a preplanned discontinuation of alendronate therapy in respective groups after 2 and 4 years, followed by placebo. E/P was stopped after 4 years. The preparation of E/P in the United States was Premarin (0.625 mg/d) and Provera (5 mg/d), whereas, in Europe, Kliogest was used (17 b-estradiol [2 mg/d] and norethindrone acetate [10 mg cyclically]). Over 6 years of treatment, BMD decreased steadily in placebo-treated women, while alendronate preserved bone mass; 5 mg/d was more efficacious than 2.5 mg/d. To answer the important question of how long treatment should be given, the changes in BMD and biochemical markers of bone turnover were assessed, following withdrawal of alendronate treatment after 2 and 4 years. Discontinuation of alendronate was followed by increases in bone resorption to levels similar to those induced by placebo treatment. These changes were associated with a progressive decrease in BMD. The slope of this decline was similar to that observed in placebo-treated patients. There was, therefore, no catch-up loss. In contrast, after stopping E/P, there was a rapid rebound of bone resorption, and BMD returned to pretreatment levels within 2 years. The conclusions of this study have obvious clinical consequences when prevention of postmenopausal bone loss is considered. Continuous treatment with alendronate for 6 years normalizes bone turnover and preserves BMD. Upon withdrawal of the bisphosphonate, bone turnover approaches baseline, and bone loss resumes. There is, however, no catch-up bone loss. Thus, even 2 or 4 years of alendronate therapy provide residual benefit, in contrast to E/P, the effect of which is lost within 2 years of treatment discontinuation.

Results of the third extension of a study of alendronate in the treatment of postmenopausal osteoporosis were also reported (). The 10-year data presented here are the longest follow-up of any antiosteoporotic treatment under controlled conditions. (The initial findings had been reported by Liberman et al. [].) Patients received alendronate either continuously for 10 years (5 mg/d or 10 mg/d) or for 5 years (followed by 5 years of placebo). The dose in the latter group was 20 mg/d for 2 years, followed by 5 mg/d for 3 years, providing a total alendronate dose similar to that of the 10 mg/d dose. The primary endpoint of the study was the change in spine BMD by intention-to-treat analysis. Secondary endpoints included BMD at other skeletal sites, biochemical markers of bone turnover, changes in stature, and clinical and laboratory safety parameters. Of the 597 patients originally allocated to these three treatment groups, 247 patients (51%) agreed to participate in this third extension (years 8 to 10). The mean age of the patients at the start of the study was 63 years, and there were no differences in demographics among the three groups. Lumbar spine (LS) BMD increased progressively by a mean of 13.8% and 9.8% after 10 years of treatment with 10 mg/d and 5 mg/d, respectively. During years 8 to 10, there was a significant increase in LS BMD, by 2.25% with alendronate (10 mg/d). Of interest, in patients who discontinued treatment after 5 years, LS BMD remained stable for the following 5 years. There was no accelerated bone loss, but also no increase in bone. Similar BMD results were obtained at hip sites; there was, however, a trend for total hip BMD to decrease with time. Alendronate treatment suppressed biochemical parameters of bone turnover to premenopausal levels that were maintained during the 10 years of follow-up. In patients who discontinued treatment after 5 years, there was an increase in the excretion of urinary cross-linked N-telopeptide of type I collagen (NTx), which, however, remained well below baseline during the 5-year follow-up period. It may be that the alendronate that had been embedded in bone was released in small quantities that were locally active, but had not been sufficient to keep bone resorption suppressed. This effect prevented bone loss, but it was not sufficient to increase BMD. Finally, and most importantly for long-term safety, the incidence of nonvertebral fractures, which was reported as adverse events during years 8 to 10 of the study, was similar to the incidence during years 1 to 3, although patients were 10 years older. This study makes possible the design of a therapeutic strategy with alendronate. For example, in patients at low risk, treatment can be stopped after 5 years, and follow-up is indicated. In patients at high risk, however, treatment can be given for longer periods. The validity of this conclusion awaits confirmation from the ongoing 10-year extension of the Fracture Intervention Study (FIT), in which bone biopsies will also be taken. The question as to whether these data are also applicable to all bisphosphonates cannot be answered with certainty because of possible differences in bisphosphonate binding to bone that may lead to a different resolution of the effect. Results of our earlier studies with oral pamidronate are consistent with those observed with alendronate, however ().

The total dose concept

Randomized clinical trials and systematic reviews with meta-analyses have demonstrated the antifracture efficacy of daily bisphosphonate administration. To improve patient convenience and long-term adherence to treatment, once-weekly regimens that provide the sum of seven daily doses were developed for alendronate (70 mg) and risedronate (35 mg). These weekly regimens are pharmacologically equivalent to the daily ones. The question is whether intermittent regimens with longer drug-free intervals, which may provide greater convenience, may also be successful. Previous attempts have generally failed to show antifracture efficacy with such regimens. Recker et al. () presented evidence that such an approach can be effective. The authors used the nitrogen-containing bisphosphonate, ibandronate, which is more potent than alendronate and risedronate; its optimal daily oral dose for the treatment of osteoporosis is 2.5 mg. In this study, 2946 postmenopausal women (with one to four prevalent vertebral fractures and LS BMD T scores < -2.0) were randomized to daily oral ibandronate (2.5 mg/d), intermittent oral ibandronate (20 mg every other day, for 12 doses at the start of a 3-month cycle), or placebo. Both ibandronate regimens significantly reduced the incidence of vertebral fractures by 62% and 50%, respectively, and this effect was sustained for the 3-year period of observation. This is the first time that a bisphosphonate demonstrated antifracture efficacy with a between-dose interval longer than 2 months in a properly designed clinical trial. The study also raises the issue of the effectiveness of the total dose given at intervals longer than 2 weeks, the period during which a total dose equal to the sum of the daily dose is expected to work, according to the principles of bone cell biology. It would be interesting to see whether such responses can be induced by other bisphosphonates.

Combination therapies

There have been several studies on the effects of combinations of antiresorptive therapies with different mechanisms of action, such as bisphosphonates with hormone replacement therapy (HRT) or raloxifene, in the management of patients with osteoporosis. The general conclusion of these studies is that combination treatments induce greater increases in BMD than in monotherapies. This conclusion was confirmed in a study of the effect of alendronate, HRT, or their combination, on BMD in elderly women (). This was a randomized, double-blind, placebo-controlled study that included 484 women older than age 65 years (mean age, 72 years) with a mean femoral neck BMD T score of -1.7. One-third of these women had osteoporosis. After 3 years, alendronate, HRT, and the combination of alendronate and HRT increased LS BMD by 7.4%, 6.8%, and 9.2%, respectively. As mentioned above, the increases induced by the combined therapy were significantly higher than those induced by either monotherapy. I must confess that I did not understand the rationale and the significance of this study, which, as stated in its title, was sponsored by the National Institutes of Health (NIH). Physicians have good reasons to consider combination therapies, the most important being the achievement of a better response in severely affected patients. This study examined a population of patients, the majority of whom will never be considered for combination therapy in clinical practice. Taking into account that the combination of therapies will combine both the price and the side effects of the two interventions, such approaches will probably reduce adherence to treatment and should perhaps be reserved for highly selected cases. In addition, the lack of any fracture data from the studies reported so far with combination therapies should make clinicians skeptical about using two antiresorptive agents in daily practice.

In contrast to this study, the studies presented by the Mass General Hospital group were particularly interesting, as they examined the effects of a more rational combination, namely that of a bone-forming and a bone-resorbing agent. The group tested the hypothesis that combining parathyroid hormone (PTH) with an antiresorptive agent will enhance the proven effectiveness of PTH. This hypothesis was tested in two studies (one study in men and the other study in women) of similar design. Both studies included three treatment groups. Group 1 received 6 months alendronate (10 mg/d), and they continued on alendronate thereafter. Group 2 received no treatment for 6 months and was then started on PTH (40 mg/d). Group 3 received 6 months alendronate (10 mg/d), and they continued on a combined alendronate and PTH therapy. The studies were planned for 3 years, and an18-month interim analysis was reported.

The study of male osteoporosis included men (aged 46 to 85 years) with BMD T scores < -2.0 (). The results of 63 of the 83 patients included in the study were shown. PTH monotherapy induced the greatest increase in LS BMD (13.7%), followed by combination therapy (10.7%) and alendronate monotherapy (6.1%). At the femoral neck, the greatest increases were observed in the two groups that used alendronate (4% to 5%), while the increase induced by PTH alone was lower (1.5%). Similar results were obtained for total body BMD, with increases in the alendronate groups by 2.9% and 4.0%, while there was a small decrease in the PTH group (-0.4%). In the corresponding study of women, results of 54 of the 91 women, with similar entry criteria to those of the male study, were discussed. Although there were some small differences in responses compared to those in men, the overall conclusions were similar. These results, as stated by the investigators, are difficult to interpret and they do not allow any conclusion of their significance for fracture prevention. An interesting and potentially clinically relevant conclusion is the possibility that previous antiresorptive therapy may block some of the actions of PTH, with yet unknown consequences. In addition, it may be that induction of resorption may be necessary for the full action of PTH. I wonder why the authors did not include the obvious group that clinicians will most probably treat in daily practice when PTH becomes available. This is PTH treatment followed by alendronate. Theoretically, this approach will help to consolidate the effect of PTH over a longer period. This, however, still needs to be proven. The availability of PTH as the first effective bone-forming agent opens the way for examining therapeutic regimens that combine antiosteoporotic treatments.

Table of Contents

New developments in bone densitometry Paul D. Miller, University of Colorado Health Sciences Center, Colorado Center for Bone Research, Lakewood, CO, USA

One of the hot topics in the management of patients with osteoporosis is how to improve compliance with osteoporosis-specific pharmaceutical agents. Long-term compliance with all treatments for asymptomatic chronic diseases is poor, and osteoporosis therapies are no exception. The impact of monitoring biochemical markers (BCMs) of bone turnover, with and without feedback to the patient on the BCM results, has been examined with regard to compliance. To do so, the impact of different rates of compliance on the therapeutic efficacy of raloxifene, as defined by different changes in bone mineral density (BMD), was examined in 75 postmenopausal women with World Health Organization (WHO) criteria for osteopenia (). The authors found that adherence was improved by 36% over 1 year by BCM monitoring with feedback to the patients, and adherence at 1 year was strongly correlated with the percentage change in hip BMD. Other ongoing studies are also examining the impact of monitoring either BCM and/or BMD, with or without feedback to the patient, on adherence and subsequent influence on therapeutic response. Although it is clear from clinical trials of highly preselected patients that adherence and therapeutic response to antiresorptive agents is high, the response rate is not clearly defined in clinical practice. This study from Sheffield, England provides evidence that both adherence and response in clinical practice are compromised, but that monitoring response, and patient interaction with regard to that response, influences adherence and BMD effects. Previous independent observations have also suggested that monitoring osteoporosis treatments with BCM improves adherence (). To the extent that part of the antifracture efficacy of antiresorptive agents is the result of their influence on BMD, improving compliance should also be translated into a higher proportion of patients more likely to benefit from osteoporosis-specific treatments ().

Siminoski et al. () reported on a prospective study that assessed the degree of height loss that best predicts incident vertebral compression fractures (VCFs) in the placebo arm of the risedronate North American and multinational studies. Using calibrated stadiometer measurement techniques, the odds ratio (OR) for detecting one or more incident VCFs increased with the amount of height loss. For example, height loss in the range of 0 to 1.0 cm produced an OR of 2.0, whereas height loss > 4.0 cm produced an OR of 25.1. In addition, greater height loss increased the likelihood of multiple VCFs. On balance, height loss > 2cm during prospective monitoring substantially increases the probability that one or more VCFs have occurred. It is known that even asymptomatic prevalent vertebral fractures increase the risk for incident VCFs and hip fractures, independent of baseline BMD (). In addition, prevalent VCFs can be seen in 35% of the postmenopausal population who do not have osteoporosis by BMD WHO criteria (). Current osteoporosis-specific therapies reduce incident vertebral fractures by about 50%, but they do not abolish them. The longitudinal assessment of patients being treated with osteoporosis-specific therapies, with the intent of reducing incident VCFs, requires annual lateral spine x-ray (or the use of newer vertebral assessment technologies that are now available) at a point-of-care while the patient is having dual-energy x-ray absorptiometry (DXA) BMD assessment. The lower radiation techniques, instant vertebral assessment (IVA) and lateral vertebral assessment (LVA), are as sensitive as lateral x-ray spine films for detecting incident VCFs. Nevertheless, a methodology of preselecting patients who are more likely to have sustained an incident VCF would reduce the number of monitored patients requiring lateral spine x-rays. A welcome suggestion in the Siminoski report () to use carefully measured height change (as a possible indicator that an incident VCF might have occurred) allows such patients to be selected for subsequent morphometric vertebral assessment.

Areal BMD by DXA and volumetric BMD by quantitative computerized tomography (QCT) are two distinctly separate technologies that can quantitate bone mass. The relationship between low BMD by areal measurements and the subsequent risk for vertebral, nonvertebral, and hip fracture is well established for white postmenopausal women. However, the relationship between QCT and subsequent fracture risk or the relationship between low BMD (done by either technique) and subsequent risk for fragility fracture, in women of non-white ethnicities and in white men, has not been well studied. Cummings et al. () performed BMD, by both DXA and QCT, in a large sample of both white and black women and men, and incident fracture data were captured over an average of 3.7 years of follow-up. Adjusting for both sex and race, each SD reduction in BMD by total hip DXA increased the risk for all clinical fractures 1.8-fold, and a similar reduction in BMD by spine QCT increased the risk 2.1-fold. It seems that both hip DXA and spine QCT predict an increased risk for clinical fractures in men and women of both races. In a subset of the etidronate clinical trials, it was previously shown that spine QCT and spine DXA have comparable risk prediction per SD reduction in BMD (). In addition, data from the National Osteoporosis Risk Assessment (NORA) () suggest that the relative risk (RR) for fracture per SD reduction in BMD, as assessed by peripheral BMD techniques, is similar among postmenopausal white, black, Asian American, and American Indian postmenopausal women (). Because the NORA () and Cummings et al. () data are consistent, it may be that fracture risk (per SD reduction) in the mean BMD of the study cohort in the United States is similar across ethnicities. In addition, the data of Cummings et al. suggest that, at least for white men, the risk may be similar to that of women, an observation made previously by others (). Studies by both Oden et al. () and the NORA () showed similar fracture risk between sexes and ethnicities when T scores were derived using data from a white female reference population database. An intriguing question therefore is whether, at least for risk prediction, white female databases could be universally used for risk assessment, at least in the United States and Europe.

Another study assessed the influence of bone size and height on areal and volumetric BMD in 150 healthy children, aged 6 to 21 years (). It has been suggested that the decrease in whole body (WB) DXA bone mineral content (BMC) with aging is the result of short stature (decreased height for age), narrow bones (decreased bone area adjusted for height), or decreased bone density (decreased BMC adjusted for bone area). These parameters are important considerations because BMD is a derived equation (BMC/area), and areal BMD does not capture the influence of bone size on BMC. This study compared WB DXA (predominantly cortical bone) with peripheral distal tibial QCT (pQCT), which measures cortical bone geometry and volumetric density. Correlations were made between three pQCT parameters (cross-sectional area [CSA] for tibial length, bone strength for tibial length, and volumetric density), as well as for WB DXA bone area for height and WB DXA BMC for bone area. The data demonstrate that decreased WB DXA bone area for height is significantly associated with decreased bone CSA for bone length, suggesting that WB DXA bone area adjusted for height provides a valid measure of cortical bone geometry. Decreased width of bone is associated with decreased strength because periosteal dimensions are a critical determinant of bone strength. However, WB DXA BMC that is adjusted for bone area is not significantly associated with cortical bone volumetric density, size, or strength, which suggests that it is not a valid measure of bone mineralization (the latter associated with strength as well). Therefore, decreased WB DXA BMC for bone area may give the false impression of decreased cortical bone density and strength. This disconnect between areal and volumetric parameters and bone strength has been seen clinically in postmenopausal women treated with low-dose parathyroid hormone (PTH) (). In PTH-treated women, areal BMD of the forearm (measured by DXA) declines, yet wrist fractures also decrease. BMC (measured by pQCT) in these postmenopausal women treated with low-dose PTH increases, as does periosteal bone accretion and CSA (). Because areal BMD is a derived equation and bone area increases in PTH-treated women (as it does in growing children and aging men), it is probable that areal BMD underestimates the changes in BMD, if the bone area increases more than BMC (). In future epidemiological studies that address the relationship between BMD and fracture risk, and in pharmacological clinical trials, it will be important to consider both areal and volumetric BMD in the assessment of bone strength.

Table of Contents


Creative Commons License This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.