Bone Densitometry: The Best Way to Detect Osteoporosis and to Monitor Therapy

The development of objective, noninvasive, and highly sensitive techniques to quantitate bone mineral density (BMD) has provided the clinician with a powerful diagnostic tool. Bone mass measurement is the best way to make the diagnosis of osteoporosis, one of the major diseases of our time. Using this technology, osteoporosis can now be detected well before it is obvious by conventional x-rays or when fractures eventuate. The definition of osteoporosis by the World Health Organization (WHO) is a BMD that is 2.5 sd or more below the mean of a young, normal reference population (1). This definition offers the practitioner an objective standard by which to make a diagnosis and to make subsequent management decisions. It has also profound therapeutic implications. Fortuitously, the advent of this technology has been accompanied by an exciting array of new, effective therapies to prevent and treat osteoporosis. The therapeutic optimism resulting from these pharmacological breakthroughs has fueled discussions about the potential widespread applicability of bone mass measurement to the entire population at risk. It has sparked controversy over how widely bone mass measurement modalities should be applied to detect the population at risk and what devices should be used to this end. In this article we provide a perspective on these issues. The aging population is inevitably going to become more osteoporotic unless we intervene first with diagnostic tools and then with preventive therapy. The National Osteoporosis Foundation has estimated that the number of postmenopausal women in this country will double from 40 to 80 million over the next 20 yr (2). Projections also call for a tripling of the number of osteoporotic fractures by 2040 from current annual estimates of 1.3 million. Staggering costs are surely going to exceed the 1995 figure of 13.8 billion dollars by countless billions more. The need is clear both to discover the population with osteoporosis and, more importantly, to identify those at risk for the disease. Measurement of bone mass is the single best predictor of fracture risk (3–5). The advent of this technology into the clinical arena over the past 15 years is similar in importance to the development of the sphygmomanometer and assays for serum cholesterol. As blood pressure and cholesterol determinations are predictors of stroke and cardiovascular disease, so bone density predicts fracture risk (2, 3–5). In fact, the measurement of bone mass is a much more powerful indicator of fracture risk than cholesterol determination is a predictor of myocardial infarction. Epidemiological studies indicate that the totality of risk for fracture, as represented by bone mass measurement, is 60–70% (6). Because of its predictive power, it is, thus, the most important information to gain about fracture risk. Other independent predictors of fractures are readily known, such as age, history of previous fracture, and low body weight. Together, the risk profile of an individual can be rather accurately determined as long as the BMD is known. The most widely used technique is dual energy x-ray absorptiometry (DXA). DXA can measure central sites, such as lumbar spine and hip, as well as peripheral sites, such as distal forearm, heel, and phalanges. DXA is noninvasive, rapid, accurate, and safe. The high precision of the technique usually allows DXA testing of central sites (not peripheral sites) to be used for monitoring as well as diagnosis. The effectiveness of therapy with bisphosphonates is well correlated with changes in bone mass (7). Most of the recent literature with respect to treatment response is based upon central measurements (8–10). It is the gold standard with which all other technologies are compared. Other approaches to bone mass measurement have been developed (11). Quantitative computed tomography (QCT) measures lumbar spine (12) and more recently has been adapted to measure peripheral sites as well (13). The potential advantage of QCT over DXA is in its ability to measure true volumetric density (grams per cm3) compared with DXA, which gives an areal density (grams per cm2). Moreover, QCT measures cancellous bone of the lumbar spine exclusively, devoid of the cortical envelope. It is less likely, therefore, to detect artifacts of aging, such as osteophytes and aortic calcifications, than DXA. On the other hand, changes in the bone marrow space with aging can confound the lumbar spine density measurement by QCT. Other disadvantages of QCT are the cost of the machines, poor availability, cost of the test, and, to a certain extent, radiation exposure, the latter of which becomes a consideration when patients are monitored. Fracture prediction by QCT is as good as but no better than that by DXA (14). A number of less costly and more portable devices that measure peripheral sites have been developed and approved by the FDA (15). The rationale to develop these densitometers rests with the fact that central machines are relatively expensive and, in some settings, not readily accessible. Some believe that central densitometers are unlikely to proliferate sufficiently or to be distributed proportionately throughout the United States so as to be accessible to all who are in need of bone mass measurement. There are currently about 5000 central DXA machines in the United States. As noted, DXA and QCT have been adapted to peripheral sites. In addition, the radius can be measured by single energy absorptiometry (16), and the phalanges can be measured by radiogrammetric absorptiometry (17) and DXA. In addition to machines that use ionizing radiation, the technique of ultrasound has been developed to measure bone mass (18). Ultrasound can measure the speed of sound as well as broadband ultrasonic attenuation of the site in question. These indexes give a measure of bone mass, although it is still not clear whether ultrasound may also detect certain qualitative aspects of bone. The machines available in the United States at this time detect bone mass in heel and proximal tibia. Ultrasound machines are attractive because they are small, portable, relatively inexpensive, and do not use ionizing radiation. The heel is of particular interest, because its composition, primarily cancellous bone, is similar to the composition of the spine. An example of how useful ultrasound can be in the quest to detect the population at risk is in Japan, where the introduction of ultrasound was associated with a 30% increase in diagnosis (19). The technologies described above give highly accurate and reproducible information. In general, bone mass at peripheral sites correlates well with measurements at more central (and more important) sites, such as hip and spine. Correlation coefficients between peripheral and central sites in general will be between 0.6–0.70 (Fig. 1). These reasonably good correlations seem to be the case regardless of the sites and the devices compared. Such correlations have led to arguments for and against the use of peripheral measurements to discover the population at risk. General optimism (15, 20) has led to views by some that peripheral sites can be used in selected subsets and that some therapeutic decisions can be based on them (21). More tempered opinions have questioned the potential utility of peripheral measurements (22, 23). The facts that the machines that measure peripheral sites can be portable and are less expensive are an important part of the argument to use peripheral techniques. If we are going to discover the millions of Americans who are at risk, we must use screening approaches that are the easiest and the most affordable. In these respects, peripheral devices are most attractive. If one can identify a large number of individuals who would otherwise not be discovered and do it in a cost-effective manner, a major point against screening the at-risk population, namely cost, is minimized. Within limits, this argument has merit. The relationship between bone mineral measurements at a peripheral site (distal radius) and a central site (spine). The study population consisted of 172 postmenopausal and osteoporotic women. Solid lines show the WHO thresholds for the definition of osteoporosis (t score, <2.5). Data are rom Ref. 36. The relationship between bone mineral measurements at a peripheral site (distal radius) and a central site (spine). The study population consisted of 172 postmenopausal and osteoporotic women. Solid lines show the WHO thresholds for the definition of osteoporosis (t score, <2.5). Data are rom Ref. 36. The good correlation between peripheral and central measurements is precisely the genesis of the counterargument, namely that peripheral sites will miss a substantial number of individuals with osteopenia or osteoporosis. Significant concordance between sites with r values from 0.6–0.70 is not good enough when it comes to predicting bone mass from one site to another in a particular individual (24–26). This discordance has the potential to be a real problem. An individual with normal bone mass at a peripheral site could have substantial reductions at a central site and, therefore, need additional testing for these to be discovered (20). In general, measurement of the site in question gives the best predictive value of the risk of fracture at that site (3). For example, measurement of hip bone density (at any of the five sites: femoral neck, trochanter, intertrochanter, Wards area, and total hip) gives much better predictive information about hip fracture than does measurement of the spine, distal radius, or heel (Fig. 2). This is generally true for the spine also except for older individuals in whom osteoarthritic changes may give artifactually elevated values in the anterio-posterior projection. For these individuals, the lateral spine or hip measurement is a better predictive index than the anterio-posterior spine bone density. Risk ratio values for hip fracture as determined by bone densitometry at the hip and other sites. Measurements of bone density were obtained at five different hip regions, two forearms regions, the heel, and the spine. Although relative risk (RR) for hip fracture was related to bone mass measurement at any site (>1), at every hip site, the RR showed much higher values (>2.5). Data are from Ref. 36. Risk ratio values for hip fracture as determined by bone densitometry at the hip and other sites. Measurements of bone density were obtained at five different hip regions, two forearms regions, the heel, and the spine. Although relative risk (RR) for hip fracture was related to bone mass measurement at any site (>1), at every hip site, the RR showed much higher values (>2.5). Data are from Ref. 36. One point on which most experts agree at this time is that peripheral measurements have not yet been shown to be useful to monitor the course of therapy (20, 27). For this important feature of bone mass measurement technology, central sites need to be used. This point raises an important problem. Certainly, peripheral measurement can be used, at least in some individuals, to discover osteoporosis. Bone density at osteoporotic levels in the radius, for example, does define a problem that needs attention (14, 3), particularly in older individuals, in whom bone mass is likely to be low at all skeletal sites. However, the best way to monitor that patient when therapy is established is to use central sites, where remodeling occurs at a more substantial pace. Thus, changes in bone mass can be detected more readily in the course of therapy. A peripheral site, such as the heel, comprised of similar cancellous bone as the lumbar spine, has a greater potential to be used for monitoring purposes, but to date the data are not conclusive, nor has the FDA yet approved this site for monitoring the course of therapy. The FDA has recognized that the precision error of the heel is low enough to be capable of monitoring, but the responsiveness of this site to therapy has not yet been established. At the present time, therefore, it would appear that central sites will have to be measured both at baseline and thereafter if densitometry is going to be used to monitor patients. Discordance among skeletal sites is not surprising, as the composition and metabolism of bone are not uniform from site to site. Different hormonal and mechanical influences lead to differential changes in bone mass as a specific function of site. For example, early postmenopausal bone loss is going to affect the cancellous skeleton first, a feature of estrogen deficiency. In this setting, therefore, a central site will show bone loss first. As osteoporosis is primarily a disorder of postmenopausal women, these central sites have reasonably been emphasized more. With age, the concordance between peripheral and central sites tends to improve, but in women in their early postmenopausal years, a discordance between peripheral and central sites is a source of concern. From the forgoing discussion, there will be a number of individuals whose peripheral measurement is normal, but whose hip or spine is osteopenic or osteoporotic. Miller et al. (20) recently suggested a group of individuals for whom one is encouraged to test beyond the peripheral site simply because they are at much greater risk. For example, any postmenopausal individual with significant risk factors for osteoporosis or a history of fragility fractures should have central measurements (Table 1), even if a peripheral site is normal. An early postmenopausal woman not taking estrogen replacement therapy who is concerned about low bone mass and would consider preventive therapy should also have a central measurement if the peripheral site measured is normal. Another important consideration is the rate of bone metabolism. If bone turnover is elevated, as determined by measurements of biochemical bone markers, these subjects, too, should have the benefit of central measurements even if the peripheral site(s) is normal. Normal peripheral screening and indication for central bone mass measurement Adapted from Miller et al. (20). Normal peripheral screening and indication for central bone mass measurement Adapted from Miller et al. (20). Ideally, one would like to test all individuals at risk for osteoporosis. As all postmenopausal individuals are at risk for developing osteoporosis if they live long enough, one would like to develop an approach that allows the entire population to be screened. Such an approach has important precedents with respect to other widely pervasive diseases, such as hypertension, diabetes, hyperlipidemia, and breast and cervical cancer. In contrast to these conditions, for which inexpensive screening tests have been developed, are readily available, or have simply been accepted, there is no such equivalent with respect to surveying the population for osteoporosis. Nevertheless, osteoporosis is a disorder that meets requirements for which a screening approach is justifiable. It is a common disease associated with high cost, it causes major morbidity and mortality, accurate and safe diagnostic tests are available, and therapy is efficacious. As more densitometers become available at a lower cost, the concept of screening the population will gain more support. At the moment, the more acceptable approach to this issue is to consider the case for selective screening. The carriers who provide reimbursement for bone densitometry have understandably been slow in developing guidelines for reimbursement, but do agree with the concept of selective screening. The Bone Mass Measurement Act of 1998 detailed, for the first time, a set of uniform indications for bone measurement for which reimbursement is justified (Table 2). They include an estrogen-deficient women at risk for osteoporosis. Although the regulation is somewhat ambiguous on this point, it is reasonable to expect that estrogen-deficient subjects at risk include those with a family history of osteoporosis, low body weight, history of anorexia, amenorrhea for at least 1 yr during the reproductive years, associated diseases associated with bone loss and certain medications. Other indications contained in the Bone Mass Measurement Act include any individual with a vertebral abnormality, receiving long term glucocorticoid therapy, or with primary hyperparathyroidism and for the purposes of monitoring a patient’s response to an FDA-approved therapy for osteoporosis. Bone mass measurement act:a effective July 1, 1998 (HCFA) Ref. 35. Bone mass measurement act:a effective July 1, 1998 (HCFA) Ref. 35. The National Osteoporosis Foundation (NOF) has just issued its document, “Osteoporosis: review of the evidence for prevention, diagnosis, and treatment and cost-effective analysis” (2). In that report, the NOF proceeds one step beyond the Health Care Finance Administration guidelines. It defines three categories of individuals who should have a bone mass measurement: all women over age 60–65 yr, all women with vertebral fracture, and all woman, 50–60 yr old, with at least one important risk factor. As we further refine and define our criteria for bone mass measurement and as technologies become more user-friendly, available, and affordable, it is clear that bone mineral densitometry is going to be even more widely application to the population. One looks forward to the day when bone mass measurement will be as much a standard of preventive care as is measurement of blood pressure, blood sugar, and cholesterol and mam-mography. Bone mass measurement is indispensable to the early diagnosis of osteoporosis. Early recognition of this disease, before the first or next fracture occurs, is leading to a therapeutic imperative, namely to intervene. With new approaches to prevention and therapy, it is important to be able to monitor the effectiveness of the therapy that has been instituted. It is not enough to state simply that the therapy has been shown to “work” and that all one has to do is to follow the therapeutic plan. The end point of therapy is a reduction in fracture incidence. Certainly, this end point can be monitored in large clinical trials, but it is relatively useless in the individual patient. It is logical to expect that as bone mass measurement defines risk of fracture, it should be useful as an index of therapeutic effectiveness. Indeed, the large clinical trial experience with the bisphosphonate, alendronate, has shown clearly that impressive increases in bone mass in the lumbar spine are associated with a substantial reduction in fracture incidence (9, 10). Recent analysis of these data also suggests that there is a significant relationship between the magnitude of the change in bone mass and the magnitude of the reduction in fractures (7). These data have led to acceptance of bone mass measurement as a surrogate marker for the true end point, fracture reduction. The bone mass measurement act of 1998 acknowledges this point by providing reimbursement for monitoring the therapeutic course of an osteoporotic subject. The use of bone densitometry to monitor the population receiving agents to prevent or treat osteoporosis is clearly important. However, recent evidence suggests that changes in bone mass do not account for the entire risk reduction associated with a specific therapeutic intervention. In the EPIDOS (28) and SOF (29) studies, large epidemiological studies in Europe and the United States, respectively, changes in bone mass have been shown to account only in part for the reduction in fracture incidence. An additional, important predictive index is the change in bone markers. A reduction in indexes of bone turnover contributes importantly and independently to fracture risk reduction. In a carefully conducted, double blind, placebo-controlled study of calcium and vitamin D, Dawson-Hughes et al. (30) have shown that fracture incidence is significantly over a This reduction in fracture incidence is associated with changes in bone Bone turnover is by calcium and vitamin The of the large study of show that fracture incidence is but bone density nor bone change data from a large clinical trial with a selective estrogen have shown that changes in bone mass and bone are much less than one would expect the major reduction in fracture incidence These data do not the value of monitoring changes in bone mass with therapy, but that there are other factors that will be in the response to therapy. In some the of bone will provide not only evidence that a is and that bone mass will increase over time, but also will provide independent data to the risk reduction by the change in bone et al. have for example, that the of reduction of the of predicts the change in lumbar spine and hip density of the potential value of bone has led the Health Care Finance Administration to a set of for their use in the of FDA-approved therapy They include two baseline a and measurements The in the of osteoporosis has been and by the advent of bone mass measurement As they become more widely and more affordable, it is that we have the potential to discover the millions of individuals at risk for or with the disease. With effective therapies at hand, it is now to prevent and treat osteoporosis. There is every therefore, to bone mass measurements as widely as to discover those subjects at risk for osteoporosis in a that is effective and affordable.