Currently the workhorse technique for detecting mineral-deficient bones is dual X-ray absorptiometry (DXA). By targeting a beam containing X-rays of two distinct energies at an area of tissue, clinicians can use the differences between the energy dependence of X-ray absorption in bone and soft tissue to measure the mass of mineral present. The outcome of a DXA test is the mass of mineral in a bone divided by the projected area occupied by the bone – a value known as the bone mineral density (BMD).
World Health Organization guidelines define osteoporosis in terms of the difference between the BMD measured in a patient and the expected BMD for a young, healthy adult of the same sex. The significance of the BMD parameter was established via clinical trials that recruited thousands of subjects (mostly post-menopausal women). These studies showed that a deficiency in baseline BMD was related to the likelihood that a subject would suffer a fracture during the follow-up period.
As a result of these trials, physicians can determine the probability of future fracture and, if necessary, prescribe a treatment that will decrease this fracture risk. The efficacy of any prescribed treatment is then also evaluated using DXA.
The whole story
Unfortunately, many patients with apparently acceptable BMD values still suffer fractures. Equally, many patients with BMD values in the fracture zone never break a bone. This apparent contradiction arises due to the existence of many other variables that influence whether a bone can or cannot withstand an applied load. These additional factors include age, gender, family history of osteoporosis, previous fracture, rate of bone turnover, inactivity, propensity for falling, bone size and shape, and the extent of mineralization within bone.
A DXA prediction concerning the likelihood of fracture is not reliable in an individual unless these other contributing factors are also evaluated. However, when examining large groups of subjects, some of these variables may cancel each other out, thus revealing the relationship between BMD and fracture.
The influence of these other variables can also be seen in the BMD response to effective treatments for osteoporosis. Drugs such as bisphosphonates and synthetic parathyroid hormone reduce the risk of fracture by roughly a factor of two, yet they only produce small increases in BMD. The improvement in fracture risk is a consequence of the rearrangement of mineral within bone, rather than a dramatic increase in the actual amount of mineral within bone.
Similarly, it was once thought that fluoride was an effective treatment for osteoporosis because it produced linear, sustained increases in spine BMD. However, fluoride treatment did not reduce the number of fractures, again implying that area density is not necessarily related to fracture likelihood.
Risk factors
To develop an accurate means of fracture risk assessment for individuals, it's therefore essential to include the other factors that contribute to risk. Three of the most important variables are bone size, mineral distribution and the extent of mineralization of hydroxyapatite crystals within bone. Each of these can be evaluated non-invasively at a clinically relevant site, such as the lower arm, using CT.
Cross-sectional CT images provide the external dimensions of the bone and the thickness of the hard outer cortical shell. By recording multiple CT images, changes in shape along the long axis of the bone can be extracted. Multiple images also provide the pattern of distribution of the trabecular bone that occupies the distal end of the medullary cavity. Meanwhile, the CT number (the mean X-ray attenuation associated with each element of a CT image) of the solid cortical bone supplies information about the mineralization of the bone.
Unfortunately, there are two objections to the use of CT. The first problem is that it entails the delivery of radiation dose to the patient. The second, and more significant, issue is the difficulty in gaining timely access to clinical machines. Not only are CT scanners expensive to purchase and operate but within hospitals these scanners are usually tied up throughout the day performing multiple general radiology exams.
This access problem can be solved through the use of small CT scanners dedicated to the task of evaluating bones of the arms and legs. Such instruments, called peripheral quantitative CT (pQCT) scanners, have in-plane spatial resolutions of a few hundred microns. Importantly, they also have much lower capital and operating costs when compared with whole-body CT scanners.
Recently, versions of these peripheral scanners that offer an in vivo spatial resolution of tens of microns have become available. But the problem of the radiation dose remains. While the dose is actually of an acceptable level (approximately 1 microSievert effective dose per CT slice), the perceived hazard has inhibited progress in the past and will continue to do so in the future.
One way to address this issue is to perform the routine measurements of bone size and shape using a small-bore MRI device designed specifically for arm and leg examinations. Figure 1 shows a 3D representation of the internal structure of a bone, measured in vivo using a 1 T, small-bore MRI machine called the OrthOne (ONI Medical Systems, Wilmington, MA).
This detailed information regarding a bone's size, shape and internal structure can be used to construct a computer model of it. The in vivo composition of the simulated bone is derived from a single-slice image taken with a pQCT scanner. Computationally, the modelled bone is subjected to a compressive load and the resultant stress determined. Such modelling can establish the load and the location at which the modelled bone will fail.
The validity of this analysis has been established in vitro using a compression test. Figure 2 shows such a comparison for dried human bone. The computed failure load is about 3200 N, while the observed load at failure is close to 3700 N. Further model refinements should reduce this discrepancy.
A prediction of the ultimate load that can be sustained by a bone is of little help to a physician, however, unless it's accompanied by an estimate of the maximum load that might be placed on the bone. It is possible to predict the load placed on the radius when somebody falls onto an outstretched hand using such factors as their age, height, weight and gender.
The ratio of expected load (in the case of a fall) to sustainable load can be termed the "bone fracture susceptibility" (BFS). If the BFS approaches or exceeds unity, fracture risk is high and intervention is required to lower that risk. It is anticipated that such BFS estimates will predict fracture likelihood with greater accuracy than BMD measurements.
At present, BFS predictions are not yet available at rheumatology clinics. The next step is to perform extensive clinical trials on at-risk populations. It is hoped that such tests will prove that the measurement of BFS bears a closer relationship to fracture outcome than a measurement of BMD.