There continues to be uncertainty about the classification of available drugs for treating osteoporosis. We find that grouping them into anti-catabolic and anabolic classes based on the mechanisms of their action on bone remodeling and fracture reduction removes ambiguities and provides a relatively straightforward classification.
The recent introduction of teriparatide into clinical practice initiated the era of anabolic therapy for osteoporosis, but it is still unclear how to define an anabolic drug. All drugs that increase bone mass do so by affecting bone remodeling. When their mechanisms of action on bone remodeling and on fracture reduction are considered, we find that anti-osteoporotic drugs fall naturally into either anti-catabolic or anabolic classes. Anti-catabolic drugs increase bone strength and reduce fractures mainly by decreasing the number of bone multicellular units (BMUs). This reduces perforative resorption and preserves skeletal microarchitecture (by preventing further structural damage to trabecular bone and increased porosity in cortical bone induced by high bone remodeling). Reduction in bone remodeling by anti-catabolic drugs may increase bone mass moderately during the interval in which previously initiated BMUs are completing mineralization. Some anti-catabolic drugs may also enhance the formation phase of the remodeling cycle, but their major action is to reduce overall bone turnover (i.e., the number of BMUs in bone). In contrast, anabolic drugs increase bone strength and reduce fractures by substantially increasing bone mass as a result of an overall increase in the number of BMUs combined with a positive BMU balance (the magnitude of the formation phase is greater than that of the resorption phase). Some anabolic drugs also induce renewed modeling, increase periosteal apposition and repair of trabecular microstructure. We hope that this classification will serve as a starting point for continued discussion on the important issue of nomenclature.
THE PHILOSOPHER LUDWIG Wittgenstein has famously written that most philosophical problems arise from linguistic mistakes. This probably also applies to modern biomedical science. As an example from our own field, the enthusiasm in the 1980s for treating osteoporosis with sodium fluoride was largely because of the linguistic mistake of defining osteoporosis as a disease of reduced bone mass rather than as a disease of reduced bone strength. In a large randomized clinical trial, Riggs et al.(1) found that the sodium fluoride treatment group had increases over placebo of 35% for lumbar spine BMD (LS-BMD) and of 12% at femoral neck BMD (FN-BMD) after 4 years. However, because fluoride's known effect on bone crystal structure could adversely affect bone strength, fracture reduction was used as the criterion for efficacy. Despite the huge increases in BMD, fluoride therapy did not reduce the risk of either vertebral or nonvertebral fractures.
The introduction into clinical practice of teriparatide, the 1–34 fragment of parathyroid hormone, in early 2003 heralded the onset of the era of anabolic drugs that possess the potential to increase bone mass and reduce fracture risk dramatically. Currently, basic scientists and pharmaceutical companies are working to bring new and even better anabolic drugs into clinical practice. Because of the enormous importance of this class of drugs, the American Society for Bone and Mineral Research, several of the NIH, and a number of private foundations dedicated to improving care of patients with metabolic diseases jointly sponsored a conference on “Advances in Skeletal Anabolic Agents for the Treatment of Osteoporosis” that was held in Bethesda, MD, on May 24–25, 2004. In addition to lectures on established anabolic agents such a teriparatide, there were also lectures on potential anabolic effects of estrogen, selective estrogen receptor modulators, calcium, vitamin D, and strontium. Also, some authorities have suggested that the amino-bisphosphonates may have anabolic effects on the bone remodeling sequence.(2) Thus, it is clear that there is not a broad consensus on which drugs should be classified as anabolic for bone. Part of the difficulty is that criteria used for the assessment of efficacy, such as increases in BMD and fracture reduction, represent continuous variables that do not clearly distinguish true anabolic drugs from those that are only efficacious. We wish here to address the issue of nomenclature and to suggest that anti-osteoporotic drugs should be classified based on their effects on bone remodeling. When this is done, most efficacious drugs that increase bone strength can be classified operationally as either anabolic (that act by increasing bone remodeling but increase formation more than resorption) or as anti-catabolic (that act by reducing bone remodeling). Moreover, the major mechanisms of fracture reduction also differ between these two classes.
PRINCIPLES OF BONE REMODELING
Bone resorption and bone formation do not occur randomly throughout the skeleton, but, as reviewed by Parfitt,(3) are coupled together at some 106 discrete foci called basic multicellular units (BMUs). The progressive stages of the bone remodeling sequence within BMUs are shown schematically in Fig. 1. At the beginning of the sequence, lining cells on the surface of bone become activated and retract. Osteoclasts are recruited to the active site from precursors in bone marrow or from circulating precursors, and these excavate a resorption cavity, the furthermost extension of which becomes bounded by a sclerotic border called the cement line. The resorption phase requires only about 2–4 weeks to complete and is terminated when the osteoclasts disappear by apoptosis. After a brief reversal phase, the resorption cavity becomes lined by osteoblasts. The osteoblasts secrete osteoid that, after a lag period, mineralizes to form new bone. Over a period of some 4–6 months, the formation phase refills the resorption cavity with bone to form a completed osteon.
Remodeling is physiologically important because it maintains normal skeletal mass, repairs microdamage to the skeleton, and participates in regulation of systemic calcium homeostasis. The components of the bone remodeling sequence can be assessed quantitatively only by histomorphometry of bone biopsy samples. The equation that describes the remodeling sequence at each BMU is
where ΔBS is the rate at which bone is added or removed from bone surfaces, Ac.f is the activation frequency, W.Th is the wall thickness of the completed osteon, E.De is the erosion depth of the resorption cavity, and (W.Th − E.De) is the remodeling balance.(4) Ac.f is proportional to the number of new BMUs being formed, W.Th to the magnitude of the formation phase, E.De to the magnitude of the resorption phase, and (W.Th − E.De) to the net gain or loss of bone at the BMU level. Because the equation was developed for use in bone histomorphometry, its components are in 2D areal units, whereas actual remodeling occurs in 3D volumetric units.
Operationally, the equation can be separated into two major components—activation frequency and remodeling balance. The activation frequency is the statistical probability that bone remodeling will be initiated on any surface at any given time and is the summation of several processes, most notably the recruitment and differentiation of osteoclasts from precursors in bone marrow or in the circulation. Because it varies among individuals and among disease states by up to 10-fold, activation frequency is the major component of bone remodeling affected by drug therapy. The other major component, the remodeling balance, is the algebraic difference of the formation phase minus the resorption phase at the BMU level. If the resorption cavity is underfilled by the osteoblasts, the remodeling balance will be negative, whereas if it is overfilled, it will be positive. Remodeling balance varies among individuals and among disease states by only ±10%, which is 1/100 of the variability of the activation frequency. Nonetheless, it is the small change in remodeling balance that allows bone to be gained or lost, because steady-state changes in activation frequency will not cause changes in bone mass if the remodeling balance is zero. However, as will be discussed later, changes in activation frequency may have short-term effects on bone mass because of expansion or contraction of the remodeling space, but these do not continue after steady state is achieved. Thus, the steady-state changes in bone mass can be estimated by the product of Ac.f × (W.Th − E.De) (i.e., the rate that new BMUs are initiated times the mean remodeling balance of each of them).
In the adult skeleton, virtually all BMUs are committed to the remodeling process. However, during skeletal growth and development, many BMUs are committed to a different process termed bone modeling. This differs from remodeling in that the formation phase occurs without a preceding resorption phase (E.De = 0), and thus, it is capable of inducing large increases in bone mass. Modeling occurs primarily on the periosteal surfaces and thus is mainly responsible for the large radial increases in bone size that occur during growth. As will be discussed below, there is evidence that teriparatide therapy can lead to a renewal of the modeling process.
DETERMINANTS OF BONE STRENGTH
There is now a consensus that osteoporosis should be defined as a disease of decreased bone strength rather than a disease of decreased bone mass and that all efficacious drugs should show increases in bone strength as evidenced by a decrease in fracture incidence. Although bone mass is still the major determinant of bone strength, other important determinants include bone geometry, bone material properties, and bone microstructure.(5) The last three of these are often grouped together under the rubric of “bone quality.” Indeed, some studies suggest that decreases in bone mass account for only about one-half of the decreases in bone strength with aging.(6) Bone geometry is important because larger bones are stronger than smaller bones, even with equivalent volumetric BMD. Indeed, bone strength increases with the fourth power of the distribution of bone mass from the central axis.(7) The material properties of bone also contribute, as was shown by the trial of treatment of osteoporosis with sodium fluoride discussed earlier.(1) Finally, bone strength is augmented by its complex microstructure. The main microarchitectural elements that contribute to bone strength are intact trabecular plates, normal trabecular width, and normal connectivity of the trabeculae, including the cross-strutting of the vertical trabeculae with horizontal trabeculae. As with a suspension bridge or a three-legged stool, connectivity supplies enormous strength. According to Euler's theorem, connecting two narrow vertical columns with only three horizontal supports would increase the strength of the columns by 16-fold.(8) All of these three microarchitectural elements are damaged by high turnover states such as are found in estrogen-deficient women and osteoporosis.(9, 10) In addition, the presence of resorption cavities on trabecular surfaces may act as “stress risers” that concentrate biomechanical strain focally.
ACTION OF ANTI-CATABOLIC DRUGS ON BONE REMODELING
Before the recent release of teriparatide, almost all available drugs for treating osteoporosis fell into this class. Their cardinal class characteristic is to reduce overall bone turnover, and they are conveniently divided into weak and more potent subclasses. Supplementary calcium and vitamin D are generally given with other anti-catabolic agents as enhancers whose small, but significant, effects depend mainly on whether there is nutritional deficiency. The weak anti-catabolic drugs used for primary treatment include nasal spray calcitonin and the selective estrogen receptor modulator, raloxifene. Over intervals of up to 3 years, they have only small effects on LS-BMD and FN-BMD, although they reduce markers of bone resorption by about 20–30%.(11, 12) Nonetheless, they reduce vertebral fractures by 30% and 40%, respectively, but do not significantly reduce nonvertebral fractures.(13) In contrast, the more potent anti-catabolic drugs such as estrogen and the amino-bisphosphonates, alendronate and risedronate, over intervals of up to 4 years decrease markers of bone resorption by about 40–60%,(14–16) increase LS-BMD by 4.5–7.5% and FN-BMD by 2.7–4.2%, and reduce vertebral fractures by 34–48% and nonvertebral fractures by 27–44%.(13, 17)
As assessed by direct histomorphometric measurements, anti-catabolic drugs act both by reducing activation frequency and by improving BMU balance, although the former is clearly more important. Increases in remodeling balance without changes in Ac.f, if this occurs, would lead to only a slow gain in bone. As assessed by bone histomorphometry, the major effect of two years of treatment with the amino-bisphosphonate, alendronate, on bone remodeling in osteoporotic patients was to reduce Ac.f by 87%, but there also was a trend toward induction of a positive BMU balance that was associated with decreases in E.De and, possibly, with borderline increases in W.Th.(18) In another study in early postmenopausal women treated for 2 years with either estrogen/progestin compared with placebo, there was only a trend for a decrease in Ac.f, but BMU balance increased because of a decrease in the resorption phase (E.De).(10) However, in postmenopausal women treated long-term with pharmacological dosages of estradiol implants for ≥14 years, W.Th was increased compared with controls, suggesting that estrogen in high dosages can stimulate the formation phase of BMU balance.(19) Thus, the effect of estrogen treatment on bone remodeling seems to vary with the estrogen preparation, dosage, duration of treatment, and underlying clinical condition. Both amino-bisphosphonates and estrogen reduce biochemical markers of bone turnover by about 40–60%, usually with similar reductions in the resorption and formation markers.(14, 20) Bisphosphonates reduce bone turnover, largely by decreasing Ac.f and the recruitment of osteoclast precursors,(21) but they also reduce the amount of bone removed during the resorption phase of the BMU cycle by either decreasing osteoclast work capacity or by increasing apoptosis.(21–24) Bisphosphonates and pharmacological dosages of estrogen also may stimulate the formation phase, probably mainly by decreasing apoptosis.(23, 25) However, the effects of treatment with these agents on bone turnover markers suggest that a reduction of activation frequency may be their most important action.
Testosterone, tibolone, and strontium ranelate are anti-osteoporotic drugs that have anti-catabolic characteristics but do not fit neatly into the mechanism suggested above. Based on studies in elderly men who received replacement after having been made acutely sex hormone-deficient, the predominant effect of testosterone was to reduce bone resorption (largely through its aromatization to estrogen), but there also was evidence that testosterone maintained levels of serum osteocalcin, a formation marker.(26) In a 12-month study of testosterone replacement in elderly men with low normal serum levels, Kenny et al.(27) found a trend for decreases in resorption markers but no decrease in formation markers. Tibolone, a drug available in Europe, but not in the United States, has a mixture of estrogenic, androgenic, and progestogenic actions.(28) Its action on bone markers and on BMD changes suggest that it should be classified as a potent anti-catabolic drug. As of now, however, trials on its antifracture efficacy and its effects on bone remodeling assessed by bone histomorphometry have not been made. Strontium ranelate is more difficult to classify. Treatment results in large increases LS-BMD and FN-BMD,(29) but these are difficult to interpret, even with corrective algorithms, because of the attenuating effect of strontium in bone on DXA results. In a large clinical trial in osteoporotic women, strontium reduced vertebral fractures by 41%.(29) However, the changes in bone turnover markers were modest, and there was directional divergence between the formation (+8.1%) and resorption markers (−12.2%). The possibility has not been excluded that strontium in bone has physical effects on bone strength that account for much of its antifracture activity. Until there are histomorphometry data to the contrary, we believe that strontium ranelate should be either classified as an anti-catabolic drug or its classification should be deferred.
Characteristically, during treatment with the more potent anti-catabolic drugs, BMD increases moderately over the first 1–3 years and then plateaus or increases only slightly thereafter. This phenomenon represents a remodeling transient caused by contraction of the remodeling space.(30) In high remodeling states, such as postmenopausal osteoporosis, there are more BMUs. When a potent anti-catabolic drug is administered, the initiation of new BMUs is sharply reduced. However, those BMUs that were previously initiated will continue in the remodeling sequence and will be eventually refilled by the formation phase, leading to increases in BMD. Although the resorption and formation phases of BMU construction take some 4–6 months to complete, BMD may continue to increase for at least 3 years because of secondary mineralization,(18) a process whereby newly formed osteons slowly increase from 70% to full mineralization without the synthesis of additional matrix.
A recently developed concept is that the fracture reduction associated with anti-catabolic drugs is largely caused by inhibition of high bone turnover. As mentioned earlier, high turnover leads to plate perforation, loss of connectivity, and increases in surface stress risers in trabecular bone. High bone turnover also weakens cortical bone by increasing porosity. That high bone turnover can increase the risk of fractures independent of its effect on inducing bone loss was suggested over 10 years ago(31, 32) and was supported by a reanalysis of the inter-relationship of vertebral fractures, LS-BMD, and bone histomorphometry.(33, 34) However, it did not become widely accepted until large clinical trials with anti-catabolic drugs were reanalyzed using an algorithm from observational studies on the relationship of changes in BMD to fracture incidence. These showed that only 3–30% of the antifracture effect of anti-catabolic drugs could be accounted for by post-treatment increases in BMD.(35–39)
Bone remodeling characteristics in normal young adult women and in women with postmenopausal osteoporosis are shown in cartoon form in Figs. 2A and 2B, and the effects of anti-catabolic drugs are shown in Fig. 2C. The mechanisms by which anti-catabolic drugs increase bone strength and decrease fragility fractures are shown in cartoon form in Fig. 3.
ACTION OF ANABOLIC DRUGS ON BONE REMODELING
Although its effects on bone remodeling have most of the characteristics of an anabolic agent, sodium fluoride must be excluded from this class because of its adverse effects on bone material properties and bone strength. Thus, the only currently available drug that is clearly anabolic is teriparatide [PTH(1–34)] by daily injection, although a similar regimen using full-length parathyroid hormone [PTH(1–34)] may soon become clinically available. In a large world-wide trial of 1637 osteoporotic women, Neer et al.(40) found that, compared with placebo, the respective changes after 21 months of teriparatide treatment with 20 and 40 μg/day were +9% and +13% for LS-BMD, +3% and +6% for FN-BMD, −65% and −69% for vertebral fractures, and −53% and −54% for nonvertebral fractures. These results are somewhat better than those obtained with the more potent anti-catabolic drugs. Several smaller studies on the effect of teriparatide for the treatment of osteoporosis have shown similar increases in BMD.(41–43) Preliminary studies with PTH(1–84) show comparable increases in LS-BMD, although the changes in FN-BMD were not significant.(44)
The effect of teriparatide treatment on increasing bone turnover differs strikingly from the anti-catabolic drugs that decrease it.(42) With teriparatide treatment, however, there is a greater increase in formation markers over resorption markers early in treatment the ratio of which progressively narrows until there is little difference by 6 months. Also, the levels of both types of markers peak at about 6 months and then decrease progressively, so that by 36 months, there is little difference between the treatment and placebo groups. Despite the transient nature of the stimulation of bone formation over bone resorption, BMD continued to increase at all measurement sites for 30 months or more, possibly because of a remodeling transient in which PTH therapy increased the remodeling space that gradually filled in as bone turnover progressively decreased.
As would be expected from the results of bone turnover markers, bone histomorphometry studies showed increased bone remodeling in the early phase of treatment that falls to normal with continued treatment.(45, 46) During the active phase of bone remodeling, bone histomorphometry studies have shown large increases in activation frequency (there are more BMUs) and a positive BMU remodeling balance (the amount of bone laid down at each BMU is increased) because of an increase in the formation phase.(46) The increase in the formation phase probably results from the combination of increased osteoblast function(47) and an inhibition of apoptosis that extends osteoblast life span.(48) The combination of high activation frequency and a positive remodeling balance leads to large increases in bone mass during the transient interval when formation exceeds resorption. Thus, teriparatide has effects on bone remodeling that are the polar opposite to those of anti-catabolic drugs.
In addition to increasing activation frequency and inducing a positive BMU balance, teriparatide therapy has three remarkable effects that enhance bone strength further. First, it induces renewed modeling, as shown by finding active formation surfaces in the absence of an underlying scalloped cement line.(45, 49) Modeling is normally present during skeletal growth and is an efficient way to increase bone mass, because the formation phase of the BMU is not preceded by a resorption phase. However, Parfitt(50) has suggested that the same histomorphometric feature could occur if more osteoblasts are formed than are needed to fill the BMU cavity, leading to an overflow of osteoblasts onto adjacent quiescent surfaces. Nonetheless, it seems likely that the major mechanism is renewed modeling, particularly on the periosteal surfaces where bone formation normally is extremely low.(51) The origin of the osteoblasts that participate in bone modeling is not clear, although rats seem to be capable of converting inactive lining cells into osteoblasts after acute treatment with teriparatide.(52)
Second, teriparatide therapy leads to substantial increases in cortical thickness(53, 54) and to increases in cross-sectional bone area of up to 10%.(55) The teriparatide-induced increases in cross-sectional area of the femoral neck, which presumably were caused by enhanced periosteal apposition, were associated with improved biomechanical indices of bone strength.(55) Similar increases in cross-sectional bone area have been reported for the lumbar spine.(56) The increases in periosteal apposition do not seem to be unique to teriparatide therapy, however, because increases in cross-sectional bone area have also been reported after growth hormone therapy for osteoporosis.(57)
Finally, as assessed by quantitative histology and 3D μCT analysis, teriparatide therapy improves the microarchitecture of bone by increasing connectivity of trabeculae, thickening of trabeculae, and converting it from a more rod-like to a more normal plate-like appearance.(53, 54) It is unclear whether therapy stimulates the budding of new trabeculae de novo or, alternatively, leads to the thickening of trabeculae that are bifurcated by tunneling resorption.(58, 59) Either or both mechanisms would lead to the observed increase in the connectivity index.
The effects of anabolic drugs on bone remodeling are shown in the cartoons in Figs. 2D and 2E. The mechanisms by which anabolic drugs increase bone strength and decrease fragility fractures are shown in the cartoon in Fig. 4.
We propose that drugs used for treating osteoporosis be classified as either anti-catabolic or anabolic using the definitions shown in Table 1. This nomenclature is similar to that of antiresorptive or formation-stimulating classes previously proposed by Riggs and Melton in 1986(60) and that has been widely used since then. However, the new classification corrects the linguistic mistakes of the earlier classification. Thus, the term antiresorptive does not take into account that the major action of anti-catabolic drugs to decrease activation frequency, leading to decreases in both resorption and formation. The term anti-catabolic is more appropriate because it is consistent not only with a decrease in resorption but also with a decrease in the total number of BMUs, an event that will oppose catabolism of bone. Also, the term formation-stimulating is not strictly correct because if resorption is stimulated to the same degree as formation or more so, as occurs, for example, in hyperthyroidism or with pharmacological dosages of thyroxine,(61) bone mass will not increase. Finally, the terms anabolic and anti-catabolic not only describe the therapeutic effects more clearly but are semantically more appropriate because the literal definition of anabolic is to build up and that of catabolic is to break down. Both anti-catabolic and anabolic drugs must show increases in bone strength, which presently can be established clinically only by showing a reduction in fracture risk.
Table Table 1. Proposed Classification of Drugs Used to Treat Osteoporosis and Definition of Differences Between Classes
At present, it seems best to consider anti-catabolic drugs as members of a single class rather than subdividing them into pure and mixed types. However, more detailed information on the mechanism of action on bone remodeling may require that some drugs, such as strontium ranelate, be classified in this way. Our definition of an anabolic drug requires both an increase in overall bone remodeling and a formation phase that is of greater magnitude than the resorption phase at the BMU level. This helps to resolve an issue that some have found semantically troubling—how to classify a drug that decreases activation frequency but also increases the formation phase of the BMU. Bisphosphonates and high-dose estrogen may have such mixed properties, as discussed previously. However, as assessed by bone biochemical markers, treatment with these drugs decreases overall turnover at the skeletal level as assessed by biochemical markers, consistent with our classification of them as anti-catabolic. If new agents are developed that increase overall bone formation at the tissue level without reducing activation frequency, the present classification will need to be modified. According to Frost's extended mechanostat model,(62) vigorous mechanical usage may change remodeling and modeling in opposite directions. A drug that potentially might possess such biphasic properties is the steroid, estren, that has been reported to increase bone mass in mice by inhibiting osteoblast apoptosis. This was associated with absolute increases in the bone formation marker serum osteocalcin.(63) Human studies with this agent have not yet been reported.
In conclusion, now that the era of anabolic drugs is on us, it is critical that a consensus be developed on what constitutes an anabolic drug. We have proposed a nomenclature in which anti-osteoporotic drugs are divided into two classes—anti-catabolic and anabolic—that differ with respect to their effects on bone remodeling and with respect to mechanisms of fracture reduction. We hope that this classification will serve as a starting point for continued discussion on this important issue.
This study was supported in part by NIH Grants AR27065, P01 AG004875, and P01 AG13918.