Calcif Tissue Int (1992) 51:169-172
Calcified Tissue International 9 1992 Springer-Verlag New York Inc.
Editorial
The Bone "Quality" Problem A symposium devoted to exploring models of basic science and clinical aspects of bone quality convened on August 27, 1991 in San Diego, California during the annual meeting of the American Society for Bone and Mineral Research. Seven speakers presented a wide range of topics on this subject and 20 scientists participated in the ensuing discussions. In opening the symposium, Dr. Joel D. Feinblatt of the Sandoz Research Institute discussed the increasing need for a quantitative assessment of bone quality which could be correlated with bone mass in animal and human models. He emphasized that recent experience with therapeutic agents with the potential to influence bone quality and quantity differentially, such as sodium fluoride and the bisphosphonates, makes such assessments mandatory. Furthermore, he acknowledged that the ultimate endpoint of therapeutic benefit is resistance to fracture, but stressed that it was increasingly difficult to conduct adequate statistically valid clinical investigations using fractures as the primary endpoint for efficacy. Consequently, it is essential that appropriate surrogates that measure bone quality as well as bone mass be developed which can be applied to human studies. The first speaker was Dr. Thomas Einhorn of the Department of Orthopedics at the Mt. Sinai School of Medicine, who presented material developed during his collaboration with Dr. Moise Azria of Sandoz Pharmaceuticals Ltd., Basel, Switzerland. Dr. Einhorn discussed animal models that lend themselves to in vitro mechanical testing. The type of fractures seen in these models depends on the type of force applied. He stressed the fact that appropriate tests should include as many types of mechanical loading as possible. Compressive loads result in fractures with an oblique configuration; bending and tensile loads result in transverse fractures; and torsional loads result in spiral fractures. Measurements of the mechanical properties of bone under loading conditions produce load-deformation curves which have a linear portion initially and then become nonlinear to the point where fracture occurs. Pattern analysis of these curves yields parameters of bone quality which include elastic modulus, stiffness, strength, toughness, and deformability. Stiffness allows bone to accept mechanical loads. Strength was defined as the maximum stress applied immediately prior to fracture. The strain (i.e., deformation) at the point of failure is designated ductility (Fig. 1). The properties of stiffness and strength result from a combination of the proteinaceous matrix and its incorporation of hydroxyapatite mineral. However, the transition from the linear to the nonlinear portion of the stress-strain curve, leading to the permanent deformation of a bone, is solely a function of the matrix. Thus, biochemical alterations Offprint requests to: Stanley Wallach, M.D., Catholic Medical Center, 88-25 153 St., Jamaica, NY 11432, USA
in collagen and bone matrix proteins may weaken bone by first changing its ability to resist deformation. This is one mechanism whereby a metabolic bone disease such as osteogenesis imperfecta, which affects the collagen component of bone, can lead to an alteration in mechanical properties. Alternatively, a condition such as osteomalacia, where the amount of mineral in an otherwise normal collagenous matrix is decreased, will affect both the slope of the curve (stiffness) and the point at which it ends in bone failure (strength). Similarly, a condition such as Paget's disease, where the remodeling of bone is abnormal, results in an aberration in the way in which the composite material is formed (i.e., the relative amounts of woven bone and mature lamellar bone). This will alter the collagenous-mineral composite material and will affect both stiffness and strength. A number of animal models have shed light on the effects of altered tissue quality on mechanical properties. A major effect of prior ovariectomy in rats is a loss of the bone's ability to resist cyclic stress. Streptozotocin-induced diabetes mellitus causes bone to become stiffer but less strong (i.e., more brittle). The bone of the spontaneously diabetic BB rat, on the other hand, lacks both strength and stiffness. The genetically altered MOV-13 mouse has a 50% reduction in bone collagen and also manifests brittle, less strong bone. Interestingly, as a compensation for the reduction in bone collagen, an adaptive increase in subperiosteal bone formation occurs which to a large degree restores the mechanical properties of MOV-13 bone towards normal. Dr. Einhorn concluded his remarks by stating that increased use of such models should help elucidate further the interrelations among bone composition, experimentally determined bone quality, and clinical consequences. Following Dr. Einhorn's presentation, Dr. Harry Genant stressed the need to avoid artifacts during sample preparation, such as surface scratches and dehydrating conditions, as these can alter mechanical properties. Dr. Sol Epstein emphasized the need to study bone under a large number of mechanical loads applied cyclically as well as statically, as small but important nuances may be missed. In his experience, Dr. Ken Faulkner noted that the rate at which strain is produced also has a fundamental effect on the mechanical load data. In response to comments from Dr. Daniel Baran, Dr. Einhorn pointed out that in these models there is an excellent correlation between bone density and stiffness and strength, and offered these data as a reason why bone mineral density in humans predicts mechanical failure. The next speaker, Dr. Piet Geusens from the University of Pellenberg (Belgium), then described experiments in sheep in which he combined measurements of bone mass using both dual energy X-ray absorptiometry (DEXA) and dual energy quantitative computed tomography (DEQCT) with mechanical testing. Sheep were utilized because they possess skeletal turnover kinetics similar to humans. The 28
Editorial
170 Point of failure Ultimate . . . . . . . . . . . . . . . . . . . . . . . . . . . . strength Y,e,d
p
Elastic limit . . . . . . . . . coc~
oughness
ILl r I-cO
..........
Ductility . . . . . . . . . STRAIN
-~
Fig. l. A standard stress/strain curve of bone loaded in tension. The linear portion of the curve represents the elastic region during which bone will return to its original shape after the load is removed. The nonlinear portion of the curve represents the plastic region during which the bone will be permanently deformed by the load. The junction of these two regions is defined as the yield point. The stress at this point is known as the elastic limit. The maximum stress at the point of failure is the ultimate tensile strength. The maximum strain at this point represents the bone's ductility. The area under the curve is known as the strain energy, and the total energy stored at the point of fracture defines the toughness of the material.
animals in the study were divided into four groups: (1) control, (2) ovariectomized, (3) ovariectomized treated with 50 IU of synthetic salmon calcitonin (SCT) daily for 6 months, and (4) ovariectomized treated with 100 IU of SCT daily for 6 months. The bone measurements were made on excised femoral neck and lumbar vertebral samples. The mechanical strength measurements employed compression and torsional stress. In addition, the resonant frequency of the samples was measured because this bone "quality" has been demonstrated to correlate with buckling strength. Ovariectomy resulted in decreased bone mineral density (BMD) by both DEXA and DEQCT and decreased resistance to rapid compression but not to torsion. Treatment with SCT produced a 10% increase in BMD, but due to high variance among the groups, the observed increase in BMD was not statistically significant. No differences in torsion testing results were observed with SCT, but there was a 25% increase in resistance to rapid compression, which was significant at the p < 0.05 level. Dr. Geusen's presentation was followed by discussion relating to the general skeletal effects of calcitonin and other "anti-resorber" drugs such as estrogens and bisphosphonates. Many recommendations for follow-up studies were made. It was concluded that 6 months was too short a period and that the number of animals too small for such studies considering the large intragroup variances noted in the measured parameters. It was suggested that the studies be extended to 2-3 years in larger groups, possibly substituting dogs or monkeys for sheep. It was also observed that the correlation between BMD and mechanical properties noted in these studies was to be expected in view of the multiplicity of animal and human studies demonstrating similar correlations. Dr. Kenneth Faulkner from the Department of Radiology and the Osteoporosis Research Group at the University of California-San Francisco shifted the discussion to noninvasive studies of bone quality in humans. He noted that current methods for assessing osteoporotic fracture risk involve
measuring the content and/or density of bone at various skeletal sites and relating the measurement to that in either agematched or young normal patients measured at the same site with the same technique. It had been demonstrated that these densitometric methods possess the ability to predict fractures. Although bone mineral density is believed to be the largest contributing factor to bone strength and fracture risk, studies have shown that other age-related factors, such as increased risk of falling and changes in bone quality and structure independent of bone mineral density, also influence bone strength and individual fracture risk. The influences of these other factors is thought to explain, at least in part, the observed overlap in bone mineral measurements between patients with and without previous osteoporotic fractures, irrespective of measurement site or technique. Due to the random nature of fractures and falling, and the fact that bone mass is continuously distributed throughout the population, he concluded that densitometric measurements from patients with and without previous osteoporotic fractures will always overlap to some degree. In an attempt to improve the separation of normal and osteoporotic patients and enhance the fracture risk relation of bone mass measurements, Dr. Faulkner described work, performed with collaborators, designed to combine structural and densitometric information for a more precise estimate of bone strength. High resolution CT was used to measure bone density and visualize gross trabecular structure in vivo. Using contiguous 5-mm slices through the lumbar spine and femoral neck, three-dimensional representations of both cortical and trabecular structure were made. Subcomponent regions were identified and labeled using a coordinate system so that regional changes in bone density would be determined at follow-up measurements. Finite element analyses were then made of subcomponent regions, and calculations of elastic modulus, yield stress, Poisson ratio, and other mechanical parameters were made using both simple and complex assumptions of mechanical properties based on bone density and the distribution and orientation of bone tissue in the subcomponent regions. The goal of this finite element analysis was to deduce the overall mechanical properties of the bone itseff. The data accumulated to date for lumbar spine and femoral neck indicate that BMD remains a strong but not the sole determinant of mechanical strength. Dr. Faulkner also discussed other techniques that might permit assessment of bone quality characteristics independently of BMD. Using ultrasound (US) parameters of broad band attenuation (BUA) and speed of sound (SOS), and correlating them with single photon absorptiometry measurements of the same region of the calcaneus, Dr. Claus Glueer noted a correlation coefficient of 0.7. However, what qualities other than BMD are reflected by BUA and SOS measurements is still unclear. Other researchers in his study group are working with nuclear magnetic resonance (NMR) representations of bone. Because bone itself is a void in hydrogen-based NMR, the signal originates from inhomogeneities in the marrow signal by the presence of interspaced bony trabeculae. Using gradient rather than spin echo images, a linear relationship (r = 0.85--0.94) between BMD and NMR signal can be obtained both in vivo and in vitro. However, these data may depend solely on trabecular spacing, which is reflected by measurements of BMD using dual energy (DEQCT). Dr. Faulkner emphasized the fact that although many of the techniques he discussed are currently limited in their clinical applicability, they could provide insights into the importance of regional bone distribution as well as trabecular orientation and cortical structure to overall skeletal
Editorial strength. If significant relationships among bone strength, bone structure, and fracture risk are verified, the incentive will exist to improve and expand current techniques so that they may move from the research laboratory into clinical practice. In discussion of this presentation, Dr. Gary Brandenberger described an US study of the patella in osteoporotic patients matched to nonosteoporotic subjects who had similar BMD measurements by both single photon absorptiometry and dual photon absorptiometry. He demonstrated that SOS was significantly decreased in the osteoporotic group, giving further credence to the concept that US parameters give additional information. Dr. Glueer added that BUA and BMD measurements obtained at the same bony sites were related with a correlation coefficient of 0.7. This result indicated that although half the BUA signal related to BMD, the remainder did not. Dr. Baran further pointed out, however, that using SOS at the calcaneal site, the correlation coefficient with BMD is 0.82, indicating a major impact of BMD at this site. With regard to finite element analysis, Dr. Toby Hayes commented on the cascade of assumptions of varying quality and robustness that underpins the technique. It was generally agreed that the validity of these newer techniques will have to await verification by application to studies of osteoporotic patients undergoing therapeutic intervention. Dr. Richard Eastall of the University of Sheffield (UK), presented data on US measurements in osteoporotic patients. If it is assumed that US measurements of bone reflect its architecture, it follows that US energy passing through the perforated trabecular plates characteristic of the osteoporotic process will be less scattered, and therefore undergo less attenuation. As the velocity of US through bone is also affected by the elastic modulus of bone and by its density, US velocity will also reflect mechanical properties (i.e., bone quality as well as quantity). Studies on calcanei removed at autopsy have shown a strong correlation between BUA and the physical density of bone. These observations raise the question whether the techniques currently used measure quality and quantity or only bone density. In this regard, Dr. Eastall described the results of his recent studies. Fifty-three normal women between the ages of 20 and 75 years, and 18 women with type I osteoporosis (vertebral and/or forearm fractures) were studied. BUA and SOS were measured through the calcaneus using a contact ultrasonic bone analyzer (Ultrasonic Innovations). Bone density of the same site in the calcaneus was measured using DEXA. In the normal women, the correlation between DEXA and BUA (r = 0.48) and the correlation between DEXA and SOS (r = 0.24) were both lower than would be expected from the reproducibility and variance of these techniques. Also in the normal women there was a significant decrease in BUA with age (r = -0.56, P < 0.00l), equivalent to a 41% decrease between ages 20 and 75 years. This decrement was greater than the 13% decrease in DEXA and the 7% decrease in SOS observed over this same age range. The regression line describing the relationship between BUA and DEXA of the calcaneus was used to derive z scores for the subjects with osteoporosis. The mean z score in type I osteoporosis was -0.59, significantly different from 0, suggesting that in osteoporosis there is a relatively greater decrease in BUA than in bone density. Dr. Eastall concluded that US measurements of bone do not measure bone density alone. The relatively greater decrease in BUA than in BMD observed in his studies may reflect the trabecular perforation in osteoporosis, resulting in fragility much greater than that suggested by BMD measurements alone.
171 In discussion, Dr. Richard Mazess noted that the instrumentation must permit a precision of no greater than 0.5% as the numerical values obtained are unavoidably admixed with fixed signals due to the contributions of water and fat to the total signal. It was conceded that many systems in current use do not meet this requirement. Concerning correlations with BMD, r values between 0.24 and 0.90 have been reported, and this great variability seems to depend on the population being studied, the instrumentation, the sites selected for measurement, and whether BUA or SOS is the correlative parameter. All discussants agreed that additional research was essential before concluding that US measurements provide a better predictive system for osteoporotic fracture than BMD. Dr. Carlo Gennari of the University of Siena (Italy) then described additional collaborative clinical studies of US measurements in the patella. He noted that in comparing pre- and postmenopausal women and older women with and without established osteoporosis, US values have definite correlations with menopausal status, age, and osteoporosis status. However, in his experiments, US did not demonstrate differences with any greater precision than observed with BMD. In pilot longitudinal studies, Dr. Gennari described the results to date of measurements of the SOS and BMD in four groups of osteoporotic women: (1) control, (2) those receiving calcium supplements alone, (3) those on estrogen replacement, and (4) those on nasal salmon calcitonin (SCT) treatment. Calcium had no effect on the expected slow decrease in the measured parameters, whereas both estrogen and SCT treatment resulted in either stabilization or a small increase in both SOS and BMD. Dr. Gennari concluded that SOS measurements of the patella are as sensitive as BMD in monitoring treatment, and are also less costly without the need for radiation exposure. Drs. Avioli and Baran, in discussing these data, pointed out that walking increases broad band attenuation (BUA) in the calcaneus, and exercise or heel impact considerations should also be taken into account in analyzing US data. Dr. Eastall mentioned that he is presently attempting US measurements in the distal radius, but the spine and femoral neck pose insurmountable problems. Dr. Michael Parfitt expressed the opinion that appendicular skeletal measurements may not be relevant to treatment-induced changes in the axial skeleton in advanced cases of osteoporosis because the appendicular skeleton does not have a hemopoietic marrow space. He believed that such studies may be more fruitful during treatment trials at an earlier, osteopenic phase of the condition. Dr. Toby Hayes of the Orthopedic Biomechanics Laboratory at Harvard Medical School then reviewed material describing the contribution of falls to osteoporotic fractures. He indicated that the traditional view that an inadequate BMD is the major determinant of fractures, even if additional deficits in bone quality are taken into consideration, is inadequate because of the consistent overlap of densitometric data between those who fracture and those who do not fracture, even when matched for age and gender. Dr. Hayes stressed that the failure of various therapies to show impressive or consistent reductions in fracture incidence is due to the confounding influences of falls and their severity. From a surveillance study of almost 1000 falls in a group of nursing home residents, Dr. Hayes observed that upper femoral fractures occurred most frequently when the subjects fell to the side rather than the back and failed to extend their hands to cushion the fall. Under these conditions, the energy exerted against the upper femur was I0-16 times greater than the energy required to fracture the elderly femur, as deduced
Editorial
172 from in vitro studies of cadaveric hip joints. There was some effect of subject height and body mass index that affected the energy transmitted to the femur during the process of falling. Gender had a very small effect as in men, but not women, there was a borderline association of fractures with decreased BMD of the upper femur. Dr. Hayes expressed the opinion that the mechanics of the falls themselves, because of their magnitude relative to BMD, dominated the factors contributing to upper femoral fractures. He applied similar concepts to vertebral body fractures and noted that a normal adult vertebra should support a load of 3000-4000 newtons. Quiet standing is associated with a force of 700 newtons, coughing with 1000, sit-ups with 2000, lifting 20 kg from the ground 3500, and lifting 50 kg 5400 newtons. Here again, the energy applied may exceed vertebral ability to resist force, although there is a greater concordance of the two than in femoral fractures. Because many vertebral fractures occur with activities that produce forces below 3000 newtons, there are more complex interactions among BMD, bone quality, and loading conditions in the spine than the hip. With regard to bone quality specifically, Dr. Hayes emphasized the fact that there is as yet no objective definition that can be used operationally. If the term "quality" refers to various aspects of bone morphology such as microdamage, connectivity, trabecular dimensions and contiguity, and state of mineralization, there are few data demonstrating the relationship of these qualities to mechanical properties and structural failure. Quantitative assessments of these stereologic characteristics in cubes of trabecular bone from the vertebrae and femoral necks of the elderly populations correlate closely with BMD and do not add to predictions of mechanical properties of bone. Nevertheless, these parameters, as well as BMD, should be quantitated in therapeutic studies in animal models. He noted in conclusion that in recent experiments with rats and baboons, the second generation bisphosphonate, alendronate, was noted to have a dose-dependent protective effect against mechanical failure. Dr. Parfitt concurred for the most part with Dr. Hayes' thesis that architecture does not have much to do with hip fractures. However, he noted that architectural considerations probably play some role in vertebral fractures as the estimates of in vivo loads during various activities are inexact and do not address fracture resistance with respect to shearing forces, twisting, bending, etc. Dr. Jean Yves Reginster of the University of Liege (Belgium) reviewed recent data describing the effects of estrogens, calcitonin, and first and second generation bisphosphonates on various bone parametes. He pointed out that estrogen, when used as a preventative agent and studied epidemiologically, is the only agent that has been shown to reduce fracture rates in the spine and femur. None of the agents have been shown in prospective studies to reduce fracture rate, possibly because of the intrinsic difficulty in conducting such longitudinal clinical studies. Therefore, using surrogates for fracture risk potential such as BMD, both estrogen and SCT have been shown to prevent postmenopausal bone loss for relatively prolonged periods of time. There are relatively few data regarding preventative effects of estrogens and calcitonin on architectural features or mechanical properties of bone, except for the data previously presented at this conference regarding the ability of estrogen and SCT to influence mechanical and US parameters in animal models (Geusens) and in humans (Gennari). Dr. Reginster stressed the need to add histologic studies to BMD and bone quality measurements in assessing the effects of these
agents. This is especially highlighted by recent experiences with fluoride and also by concerns regarding potential deleterious effects of long-term etidronate treatment based on earlier observations in Paget's disease. With regard to the treatment of established osteoporosis, Dr. Reginster expressed the opinion that fracture rates must be an important component of drug evaluations. Dr. Peter Antich presented briefly some data suggesting that sustained release fluoride may have some beneficial effects on trabecular bone. His studies involved a new method of utilizing US in which the US beam undergoes reflection off a bone sample at a critical angle dependent on the velocity of the pressure wave in bone, which itself is related to the intrinsic symmetry of bone structure. The critical angle is correlated with age in both pre- and postmenopausal women and is significantly decreased in established osteoporosis. This parameter is favorably affected by sustained released fluoride. Dr. Antich and his collaborator, Dr. Charles Pak, believe the reflection US technique is a measurement of bone quality. Dr. Wallach thanked the participants for their presentations and critical discussion, stressing that mechanical testing in animal models is becoming very sophisticated and should be included in long-term studies. He also noted that in humans, new techniques such as finite element analysis of three-dimensional QCT, MRI, and new US-based techniques should be introduced into protocols which presently employ only BMD and fracture frequency as end-points in order to validate these methodologies. These newer methods of studying bone quality in vivo should also be incorporated into animal studies as correlates of mechanical parameters. Finally, far more attention should be paid to falls in the elderly, their causes and characteristics, as inroads into fracture prevention can be made here as well as by improving BMD and bone quality. Dr. Avioli expressed the hope that techniques designed to measure bone "quality" might also be used predictively in advance of fracture to better define the patient at risk, especially in hip fracture cases where femoral BMD does not differ significantly between patients who fracture and those who do not. He also expected that novel methods of measuring bone quality and strength would evolve in the future, especially as long-term longitudinal fracture rate studies prove so compromising and difficult to interpret in studies designed to determine the efficacy of pharmaceutical intervention. Stanley Wallach, M.D. Department of Internal Medicine University of South Florida College of Medicine Tampa, Florida 33612 USA Joel D. Feinblatt, Ph.D. John H. Carstens, Jr., M.D. The Sandoz Research Institute East Hanover, New Jersey 07936 USA Louis V. Avioli, M.D. Washington University School of Medicine The Jewish Hospital of St. Louis St. Louis, Missouri 63110 USA Received April 3, 1992
Calcified Tissue International 9 1992 Springer-Verlag New York Inc.
Editorial
The Bone "Quality" Problem A symposium devoted to exploring models of basic science and clinical aspects of bone quality convened on August 27, 1991 in San Diego, California during the annual meeting of the American Society for Bone and Mineral Research. Seven speakers presented a wide range of topics on this subject and 20 scientists participated in the ensuing discussions. In opening the symposium, Dr. Joel D. Feinblatt of the Sandoz Research Institute discussed the increasing need for a quantitative assessment of bone quality which could be correlated with bone mass in animal and human models. He emphasized that recent experience with therapeutic agents with the potential to influence bone quality and quantity differentially, such as sodium fluoride and the bisphosphonates, makes such assessments mandatory. Furthermore, he acknowledged that the ultimate endpoint of therapeutic benefit is resistance to fracture, but stressed that it was increasingly difficult to conduct adequate statistically valid clinical investigations using fractures as the primary endpoint for efficacy. Consequently, it is essential that appropriate surrogates that measure bone quality as well as bone mass be developed which can be applied to human studies. The first speaker was Dr. Thomas Einhorn of the Department of Orthopedics at the Mt. Sinai School of Medicine, who presented material developed during his collaboration with Dr. Moise Azria of Sandoz Pharmaceuticals Ltd., Basel, Switzerland. Dr. Einhorn discussed animal models that lend themselves to in vitro mechanical testing. The type of fractures seen in these models depends on the type of force applied. He stressed the fact that appropriate tests should include as many types of mechanical loading as possible. Compressive loads result in fractures with an oblique configuration; bending and tensile loads result in transverse fractures; and torsional loads result in spiral fractures. Measurements of the mechanical properties of bone under loading conditions produce load-deformation curves which have a linear portion initially and then become nonlinear to the point where fracture occurs. Pattern analysis of these curves yields parameters of bone quality which include elastic modulus, stiffness, strength, toughness, and deformability. Stiffness allows bone to accept mechanical loads. Strength was defined as the maximum stress applied immediately prior to fracture. The strain (i.e., deformation) at the point of failure is designated ductility (Fig. 1). The properties of stiffness and strength result from a combination of the proteinaceous matrix and its incorporation of hydroxyapatite mineral. However, the transition from the linear to the nonlinear portion of the stress-strain curve, leading to the permanent deformation of a bone, is solely a function of the matrix. Thus, biochemical alterations Offprint requests to: Stanley Wallach, M.D., Catholic Medical Center, 88-25 153 St., Jamaica, NY 11432, USA
in collagen and bone matrix proteins may weaken bone by first changing its ability to resist deformation. This is one mechanism whereby a metabolic bone disease such as osteogenesis imperfecta, which affects the collagen component of bone, can lead to an alteration in mechanical properties. Alternatively, a condition such as osteomalacia, where the amount of mineral in an otherwise normal collagenous matrix is decreased, will affect both the slope of the curve (stiffness) and the point at which it ends in bone failure (strength). Similarly, a condition such as Paget's disease, where the remodeling of bone is abnormal, results in an aberration in the way in which the composite material is formed (i.e., the relative amounts of woven bone and mature lamellar bone). This will alter the collagenous-mineral composite material and will affect both stiffness and strength. A number of animal models have shed light on the effects of altered tissue quality on mechanical properties. A major effect of prior ovariectomy in rats is a loss of the bone's ability to resist cyclic stress. Streptozotocin-induced diabetes mellitus causes bone to become stiffer but less strong (i.e., more brittle). The bone of the spontaneously diabetic BB rat, on the other hand, lacks both strength and stiffness. The genetically altered MOV-13 mouse has a 50% reduction in bone collagen and also manifests brittle, less strong bone. Interestingly, as a compensation for the reduction in bone collagen, an adaptive increase in subperiosteal bone formation occurs which to a large degree restores the mechanical properties of MOV-13 bone towards normal. Dr. Einhorn concluded his remarks by stating that increased use of such models should help elucidate further the interrelations among bone composition, experimentally determined bone quality, and clinical consequences. Following Dr. Einhorn's presentation, Dr. Harry Genant stressed the need to avoid artifacts during sample preparation, such as surface scratches and dehydrating conditions, as these can alter mechanical properties. Dr. Sol Epstein emphasized the need to study bone under a large number of mechanical loads applied cyclically as well as statically, as small but important nuances may be missed. In his experience, Dr. Ken Faulkner noted that the rate at which strain is produced also has a fundamental effect on the mechanical load data. In response to comments from Dr. Daniel Baran, Dr. Einhorn pointed out that in these models there is an excellent correlation between bone density and stiffness and strength, and offered these data as a reason why bone mineral density in humans predicts mechanical failure. The next speaker, Dr. Piet Geusens from the University of Pellenberg (Belgium), then described experiments in sheep in which he combined measurements of bone mass using both dual energy X-ray absorptiometry (DEXA) and dual energy quantitative computed tomography (DEQCT) with mechanical testing. Sheep were utilized because they possess skeletal turnover kinetics similar to humans. The 28
Editorial
170 Point of failure Ultimate . . . . . . . . . . . . . . . . . . . . . . . . . . . . strength Y,e,d
p
Elastic limit . . . . . . . . . coc~
oughness
ILl r I-cO
..........
Ductility . . . . . . . . . STRAIN
-~
Fig. l. A standard stress/strain curve of bone loaded in tension. The linear portion of the curve represents the elastic region during which bone will return to its original shape after the load is removed. The nonlinear portion of the curve represents the plastic region during which the bone will be permanently deformed by the load. The junction of these two regions is defined as the yield point. The stress at this point is known as the elastic limit. The maximum stress at the point of failure is the ultimate tensile strength. The maximum strain at this point represents the bone's ductility. The area under the curve is known as the strain energy, and the total energy stored at the point of fracture defines the toughness of the material.
animals in the study were divided into four groups: (1) control, (2) ovariectomized, (3) ovariectomized treated with 50 IU of synthetic salmon calcitonin (SCT) daily for 6 months, and (4) ovariectomized treated with 100 IU of SCT daily for 6 months. The bone measurements were made on excised femoral neck and lumbar vertebral samples. The mechanical strength measurements employed compression and torsional stress. In addition, the resonant frequency of the samples was measured because this bone "quality" has been demonstrated to correlate with buckling strength. Ovariectomy resulted in decreased bone mineral density (BMD) by both DEXA and DEQCT and decreased resistance to rapid compression but not to torsion. Treatment with SCT produced a 10% increase in BMD, but due to high variance among the groups, the observed increase in BMD was not statistically significant. No differences in torsion testing results were observed with SCT, but there was a 25% increase in resistance to rapid compression, which was significant at the p < 0.05 level. Dr. Geusen's presentation was followed by discussion relating to the general skeletal effects of calcitonin and other "anti-resorber" drugs such as estrogens and bisphosphonates. Many recommendations for follow-up studies were made. It was concluded that 6 months was too short a period and that the number of animals too small for such studies considering the large intragroup variances noted in the measured parameters. It was suggested that the studies be extended to 2-3 years in larger groups, possibly substituting dogs or monkeys for sheep. It was also observed that the correlation between BMD and mechanical properties noted in these studies was to be expected in view of the multiplicity of animal and human studies demonstrating similar correlations. Dr. Kenneth Faulkner from the Department of Radiology and the Osteoporosis Research Group at the University of California-San Francisco shifted the discussion to noninvasive studies of bone quality in humans. He noted that current methods for assessing osteoporotic fracture risk involve
measuring the content and/or density of bone at various skeletal sites and relating the measurement to that in either agematched or young normal patients measured at the same site with the same technique. It had been demonstrated that these densitometric methods possess the ability to predict fractures. Although bone mineral density is believed to be the largest contributing factor to bone strength and fracture risk, studies have shown that other age-related factors, such as increased risk of falling and changes in bone quality and structure independent of bone mineral density, also influence bone strength and individual fracture risk. The influences of these other factors is thought to explain, at least in part, the observed overlap in bone mineral measurements between patients with and without previous osteoporotic fractures, irrespective of measurement site or technique. Due to the random nature of fractures and falling, and the fact that bone mass is continuously distributed throughout the population, he concluded that densitometric measurements from patients with and without previous osteoporotic fractures will always overlap to some degree. In an attempt to improve the separation of normal and osteoporotic patients and enhance the fracture risk relation of bone mass measurements, Dr. Faulkner described work, performed with collaborators, designed to combine structural and densitometric information for a more precise estimate of bone strength. High resolution CT was used to measure bone density and visualize gross trabecular structure in vivo. Using contiguous 5-mm slices through the lumbar spine and femoral neck, three-dimensional representations of both cortical and trabecular structure were made. Subcomponent regions were identified and labeled using a coordinate system so that regional changes in bone density would be determined at follow-up measurements. Finite element analyses were then made of subcomponent regions, and calculations of elastic modulus, yield stress, Poisson ratio, and other mechanical parameters were made using both simple and complex assumptions of mechanical properties based on bone density and the distribution and orientation of bone tissue in the subcomponent regions. The goal of this finite element analysis was to deduce the overall mechanical properties of the bone itseff. The data accumulated to date for lumbar spine and femoral neck indicate that BMD remains a strong but not the sole determinant of mechanical strength. Dr. Faulkner also discussed other techniques that might permit assessment of bone quality characteristics independently of BMD. Using ultrasound (US) parameters of broad band attenuation (BUA) and speed of sound (SOS), and correlating them with single photon absorptiometry measurements of the same region of the calcaneus, Dr. Claus Glueer noted a correlation coefficient of 0.7. However, what qualities other than BMD are reflected by BUA and SOS measurements is still unclear. Other researchers in his study group are working with nuclear magnetic resonance (NMR) representations of bone. Because bone itself is a void in hydrogen-based NMR, the signal originates from inhomogeneities in the marrow signal by the presence of interspaced bony trabeculae. Using gradient rather than spin echo images, a linear relationship (r = 0.85--0.94) between BMD and NMR signal can be obtained both in vivo and in vitro. However, these data may depend solely on trabecular spacing, which is reflected by measurements of BMD using dual energy (DEQCT). Dr. Faulkner emphasized the fact that although many of the techniques he discussed are currently limited in their clinical applicability, they could provide insights into the importance of regional bone distribution as well as trabecular orientation and cortical structure to overall skeletal
Editorial strength. If significant relationships among bone strength, bone structure, and fracture risk are verified, the incentive will exist to improve and expand current techniques so that they may move from the research laboratory into clinical practice. In discussion of this presentation, Dr. Gary Brandenberger described an US study of the patella in osteoporotic patients matched to nonosteoporotic subjects who had similar BMD measurements by both single photon absorptiometry and dual photon absorptiometry. He demonstrated that SOS was significantly decreased in the osteoporotic group, giving further credence to the concept that US parameters give additional information. Dr. Glueer added that BUA and BMD measurements obtained at the same bony sites were related with a correlation coefficient of 0.7. This result indicated that although half the BUA signal related to BMD, the remainder did not. Dr. Baran further pointed out, however, that using SOS at the calcaneal site, the correlation coefficient with BMD is 0.82, indicating a major impact of BMD at this site. With regard to finite element analysis, Dr. Toby Hayes commented on the cascade of assumptions of varying quality and robustness that underpins the technique. It was generally agreed that the validity of these newer techniques will have to await verification by application to studies of osteoporotic patients undergoing therapeutic intervention. Dr. Richard Eastall of the University of Sheffield (UK), presented data on US measurements in osteoporotic patients. If it is assumed that US measurements of bone reflect its architecture, it follows that US energy passing through the perforated trabecular plates characteristic of the osteoporotic process will be less scattered, and therefore undergo less attenuation. As the velocity of US through bone is also affected by the elastic modulus of bone and by its density, US velocity will also reflect mechanical properties (i.e., bone quality as well as quantity). Studies on calcanei removed at autopsy have shown a strong correlation between BUA and the physical density of bone. These observations raise the question whether the techniques currently used measure quality and quantity or only bone density. In this regard, Dr. Eastall described the results of his recent studies. Fifty-three normal women between the ages of 20 and 75 years, and 18 women with type I osteoporosis (vertebral and/or forearm fractures) were studied. BUA and SOS were measured through the calcaneus using a contact ultrasonic bone analyzer (Ultrasonic Innovations). Bone density of the same site in the calcaneus was measured using DEXA. In the normal women, the correlation between DEXA and BUA (r = 0.48) and the correlation between DEXA and SOS (r = 0.24) were both lower than would be expected from the reproducibility and variance of these techniques. Also in the normal women there was a significant decrease in BUA with age (r = -0.56, P < 0.00l), equivalent to a 41% decrease between ages 20 and 75 years. This decrement was greater than the 13% decrease in DEXA and the 7% decrease in SOS observed over this same age range. The regression line describing the relationship between BUA and DEXA of the calcaneus was used to derive z scores for the subjects with osteoporosis. The mean z score in type I osteoporosis was -0.59, significantly different from 0, suggesting that in osteoporosis there is a relatively greater decrease in BUA than in bone density. Dr. Eastall concluded that US measurements of bone do not measure bone density alone. The relatively greater decrease in BUA than in BMD observed in his studies may reflect the trabecular perforation in osteoporosis, resulting in fragility much greater than that suggested by BMD measurements alone.
171 In discussion, Dr. Richard Mazess noted that the instrumentation must permit a precision of no greater than 0.5% as the numerical values obtained are unavoidably admixed with fixed signals due to the contributions of water and fat to the total signal. It was conceded that many systems in current use do not meet this requirement. Concerning correlations with BMD, r values between 0.24 and 0.90 have been reported, and this great variability seems to depend on the population being studied, the instrumentation, the sites selected for measurement, and whether BUA or SOS is the correlative parameter. All discussants agreed that additional research was essential before concluding that US measurements provide a better predictive system for osteoporotic fracture than BMD. Dr. Carlo Gennari of the University of Siena (Italy) then described additional collaborative clinical studies of US measurements in the patella. He noted that in comparing pre- and postmenopausal women and older women with and without established osteoporosis, US values have definite correlations with menopausal status, age, and osteoporosis status. However, in his experiments, US did not demonstrate differences with any greater precision than observed with BMD. In pilot longitudinal studies, Dr. Gennari described the results to date of measurements of the SOS and BMD in four groups of osteoporotic women: (1) control, (2) those receiving calcium supplements alone, (3) those on estrogen replacement, and (4) those on nasal salmon calcitonin (SCT) treatment. Calcium had no effect on the expected slow decrease in the measured parameters, whereas both estrogen and SCT treatment resulted in either stabilization or a small increase in both SOS and BMD. Dr. Gennari concluded that SOS measurements of the patella are as sensitive as BMD in monitoring treatment, and are also less costly without the need for radiation exposure. Drs. Avioli and Baran, in discussing these data, pointed out that walking increases broad band attenuation (BUA) in the calcaneus, and exercise or heel impact considerations should also be taken into account in analyzing US data. Dr. Eastall mentioned that he is presently attempting US measurements in the distal radius, but the spine and femoral neck pose insurmountable problems. Dr. Michael Parfitt expressed the opinion that appendicular skeletal measurements may not be relevant to treatment-induced changes in the axial skeleton in advanced cases of osteoporosis because the appendicular skeleton does not have a hemopoietic marrow space. He believed that such studies may be more fruitful during treatment trials at an earlier, osteopenic phase of the condition. Dr. Toby Hayes of the Orthopedic Biomechanics Laboratory at Harvard Medical School then reviewed material describing the contribution of falls to osteoporotic fractures. He indicated that the traditional view that an inadequate BMD is the major determinant of fractures, even if additional deficits in bone quality are taken into consideration, is inadequate because of the consistent overlap of densitometric data between those who fracture and those who do not fracture, even when matched for age and gender. Dr. Hayes stressed that the failure of various therapies to show impressive or consistent reductions in fracture incidence is due to the confounding influences of falls and their severity. From a surveillance study of almost 1000 falls in a group of nursing home residents, Dr. Hayes observed that upper femoral fractures occurred most frequently when the subjects fell to the side rather than the back and failed to extend their hands to cushion the fall. Under these conditions, the energy exerted against the upper femur was I0-16 times greater than the energy required to fracture the elderly femur, as deduced
Editorial
172 from in vitro studies of cadaveric hip joints. There was some effect of subject height and body mass index that affected the energy transmitted to the femur during the process of falling. Gender had a very small effect as in men, but not women, there was a borderline association of fractures with decreased BMD of the upper femur. Dr. Hayes expressed the opinion that the mechanics of the falls themselves, because of their magnitude relative to BMD, dominated the factors contributing to upper femoral fractures. He applied similar concepts to vertebral body fractures and noted that a normal adult vertebra should support a load of 3000-4000 newtons. Quiet standing is associated with a force of 700 newtons, coughing with 1000, sit-ups with 2000, lifting 20 kg from the ground 3500, and lifting 50 kg 5400 newtons. Here again, the energy applied may exceed vertebral ability to resist force, although there is a greater concordance of the two than in femoral fractures. Because many vertebral fractures occur with activities that produce forces below 3000 newtons, there are more complex interactions among BMD, bone quality, and loading conditions in the spine than the hip. With regard to bone quality specifically, Dr. Hayes emphasized the fact that there is as yet no objective definition that can be used operationally. If the term "quality" refers to various aspects of bone morphology such as microdamage, connectivity, trabecular dimensions and contiguity, and state of mineralization, there are few data demonstrating the relationship of these qualities to mechanical properties and structural failure. Quantitative assessments of these stereologic characteristics in cubes of trabecular bone from the vertebrae and femoral necks of the elderly populations correlate closely with BMD and do not add to predictions of mechanical properties of bone. Nevertheless, these parameters, as well as BMD, should be quantitated in therapeutic studies in animal models. He noted in conclusion that in recent experiments with rats and baboons, the second generation bisphosphonate, alendronate, was noted to have a dose-dependent protective effect against mechanical failure. Dr. Parfitt concurred for the most part with Dr. Hayes' thesis that architecture does not have much to do with hip fractures. However, he noted that architectural considerations probably play some role in vertebral fractures as the estimates of in vivo loads during various activities are inexact and do not address fracture resistance with respect to shearing forces, twisting, bending, etc. Dr. Jean Yves Reginster of the University of Liege (Belgium) reviewed recent data describing the effects of estrogens, calcitonin, and first and second generation bisphosphonates on various bone parametes. He pointed out that estrogen, when used as a preventative agent and studied epidemiologically, is the only agent that has been shown to reduce fracture rates in the spine and femur. None of the agents have been shown in prospective studies to reduce fracture rate, possibly because of the intrinsic difficulty in conducting such longitudinal clinical studies. Therefore, using surrogates for fracture risk potential such as BMD, both estrogen and SCT have been shown to prevent postmenopausal bone loss for relatively prolonged periods of time. There are relatively few data regarding preventative effects of estrogens and calcitonin on architectural features or mechanical properties of bone, except for the data previously presented at this conference regarding the ability of estrogen and SCT to influence mechanical and US parameters in animal models (Geusens) and in humans (Gennari). Dr. Reginster stressed the need to add histologic studies to BMD and bone quality measurements in assessing the effects of these
agents. This is especially highlighted by recent experiences with fluoride and also by concerns regarding potential deleterious effects of long-term etidronate treatment based on earlier observations in Paget's disease. With regard to the treatment of established osteoporosis, Dr. Reginster expressed the opinion that fracture rates must be an important component of drug evaluations. Dr. Peter Antich presented briefly some data suggesting that sustained release fluoride may have some beneficial effects on trabecular bone. His studies involved a new method of utilizing US in which the US beam undergoes reflection off a bone sample at a critical angle dependent on the velocity of the pressure wave in bone, which itself is related to the intrinsic symmetry of bone structure. The critical angle is correlated with age in both pre- and postmenopausal women and is significantly decreased in established osteoporosis. This parameter is favorably affected by sustained released fluoride. Dr. Antich and his collaborator, Dr. Charles Pak, believe the reflection US technique is a measurement of bone quality. Dr. Wallach thanked the participants for their presentations and critical discussion, stressing that mechanical testing in animal models is becoming very sophisticated and should be included in long-term studies. He also noted that in humans, new techniques such as finite element analysis of three-dimensional QCT, MRI, and new US-based techniques should be introduced into protocols which presently employ only BMD and fracture frequency as end-points in order to validate these methodologies. These newer methods of studying bone quality in vivo should also be incorporated into animal studies as correlates of mechanical parameters. Finally, far more attention should be paid to falls in the elderly, their causes and characteristics, as inroads into fracture prevention can be made here as well as by improving BMD and bone quality. Dr. Avioli expressed the hope that techniques designed to measure bone "quality" might also be used predictively in advance of fracture to better define the patient at risk, especially in hip fracture cases where femoral BMD does not differ significantly between patients who fracture and those who do not. He also expected that novel methods of measuring bone quality and strength would evolve in the future, especially as long-term longitudinal fracture rate studies prove so compromising and difficult to interpret in studies designed to determine the efficacy of pharmaceutical intervention. Stanley Wallach, M.D. Department of Internal Medicine University of South Florida College of Medicine Tampa, Florida 33612 USA Joel D. Feinblatt, Ph.D. John H. Carstens, Jr., M.D. The Sandoz Research Institute East Hanover, New Jersey 07936 USA Louis V. Avioli, M.D. Washington University School of Medicine The Jewish Hospital of St. Louis St. Louis, Missouri 63110 USA Received April 3, 1992
