Interpretation of DEXA Measures

DEXA measures “areal” bone mineral density in two dimensions only. It is calculated by dividing the mineral content of the site of bone being assessed by the area of bone and is represented by the units of g/cm². Thus if a large and a small bone have the same mineral density, the larger bone will always appear to have a greater BMD owing to the fact that it has a smaller cross sectional area compared to the smaller bone when surface area is used as the denominator.¹¹ As such a more accurate assessment of BMD would need to take into account the depth of bone and thereby provide a measure of volumetric bone density, which would be less influenced by the size of the actual bone.¹¹ In the clinical setting however, this measure is not used. Hence, it is important to realise that individuals of short stature will have lower BMD measurements than their taller counterparts. A component of ethnic differences in BMD is explained by differences in stature and, as a result, bone size.

Commercially available DEXA measures primarily bone mineral density at the hip (femoral neck) and lumbar spine. The measurement of BMD at these sites is sufficient to detect osteoporosis and provides an indication of potential risk of future fracture.¹²

When performing serial measures of bone mineral density, it is important to use a radiology service with good quality control measures with regards to performing DEXA studies. It is also important to realize that there is variability in the precision of the test itself with bone mineral density measures varying between 1 – 3% at the hip and spine.^13-15 Different imaging machines will also produce different results. Thus it is important for patients to have serial measures performed on the same machine.

This has implications in terms of interpretation of serial measures and whether specific osteoporotic therapies have been successful in an individual patient. When evaluating the success of therapy, small decreases in bone mineral density measures of < 5% at the hip or < 3% at the spine may not necessarily indicate a failure of therapy. It may be that the loss of BMD may have actually been worse had the patient not been on therapy, it may reflect use of different imaging machines or precision variability differences in the test.¹¹ Decreases in BMD may also be due to non-compliance with therapy and hence it is important to evaluate this in patients before assuming that therapy has been ineffective.

There also is a phenomenon known as “regression to the mean”. This phenomenon refers to changes in the second year of therapy regressing back to the mean of the group in the first year of therapy. This has been shown for instances in studies with alendronate and raloxifene.^11,16 In a study using alendronate, BMD in the hip in some patients had decreased by 4% or more in the first year of therapy, but increased in 83% of patients by the second year of therapy. Conversely, those patients with the greatest increase in BMD in the first year were found to have a slight decrease in the second year of therapy. The implications of this finding are that even if there is a decrease in BMD when measured 12 months after initiating therapy, patients may still go on and have an increase in BMD the following year. Hence it is probably best to monitor therapy with a BMD measurement 2 years after initiating therapy. It may be necessary to perform a BMD measure after the first 12 months of therapy if there is concern about the effectiveness of therapy especially if there are concurrent factors that may affect the efficacy of therapy such as the use of corticosteroid therapy.¹¹

Frequency of Monitoring

For women who are on specific therapies for osteoporosis, a measure of BMD by DEXA every 2 years is usually sufficient to assess the effectiveness of therapy.

In postmenopausal women who do not have osteoporosis or osteopaenia (BMD T score > - 1.0) the frequency of monitoring for the onset of osteoporosis is difficult to determine. In the immediate years following menopause, there is a rapid phase of bone loss, which can last from 4 – 8 years.⁴ Usually, however, most bone is lost in the first 3 years after menopause and then the rate of bone loss slows. During the first 3 years after menopause, women may lose up to approximately 2% of bone mass per year, especially in the vertebral bones. The extent of bone loss in the hip and wrist may be similar or slightly less.^9,17,18 Once the rapid phase of bone loss subsides, women then undergo a slow phase of bone loss, which continues indefinitely. This phase of bone loss is part of normal ageing and not only affects women but also men. During this phase women can expect to lose 1% of bone mass from the hips per year and just under 1% of bone mass from the spine.^4,11 This equates to an average rate of bone loss that is less than 0.1 T score units annually. Thus if a woman has a normal bone density (T score > -1.0), it is unlikely that her T score will fall below a T score of –2.5 for at least 10 years. In this setting it is probably advisable to recommend a repeat DEXA in 5 or more years, provided that the woman is of good health, not on corticosteroids, at least 3 - 4 years beyond menopause and has stable weight.¹¹

Who should have a DEXA scan?

Patients who have an increased risk of osteoporosis should have a DEXA scan. But not all can have full Medicare rebate for the scan.

High Risk Group:

• Family history of osteoporosis
• Premature menopause
• Long-term corticosteroids (eg prednisolone)
• Malabsorption eg coeliac disease
• Rheumatoid arthritis
• Smoking
• Long-term inadequate exercise/immobilisation
• Long-term low dietary calcium
• House-bound / institutionalised elderly
• Patients who sustain a fracture after minimal trauma eg falling from standing height.

There are specific indications that allow for full Medicare Rebate:

• People aged 70 years and over (available 1 April 2007)
• ≥ 1 fractures occurring after minimal trauma
• Monitoring of low bone mineral density proven by previous bone densitometry.
• Prolonged steroid therapy
• Female hypogonadism (oestrogen deficiency) lasting more than 6 months before age of 45.
• Male hypogonadism (testosterone deficiency)
• Primary hyperparathyroidism
• Chronic liver disease
• Chronic kidney disease
• Malabsorption
• Rheumatoid arthritis
• Hyperthyroidism

Prediction of Fracture Risk

Fracture risk increases as bone mineral density decreases. There is however no specific cut off value for BMD, which equates to an abrupt increase in fracture risk. Rather, declining BMD represents a continuous risk factor for fracture.¹¹ In a meta-analysis evaluating the predictive ability of BMD measures to predict later fractures in women, it was found that for each standard deviation (SD) below the age adjusted mean for BMD measured at any site, there was approximately a 1.5 fold increase in the relative risk of any type of fracture. However, it was also found that lumbar spine BMD and hip BMD measures were far better predictors of their own site-specific fractures. A decrease of 1 SD in lumbar spine BMD results in a 2.3 (95% CI; 1.9-2.8) fold relative risk increase for vertebral fractures. Likewise, a 1 SD decrease in hip BMD increases the age adjusted risk of hip fracture by a factor of 2.6 fold (95% CI; 2.0-3.5).¹² Although femoral neck BMD best predicts hip fracture, femoral neck and spine BMD also have similar accuracy for predicting vertebral fractures and fractures at all sites.¹² The advantage of femoral neck BMD measures over spine BMD for predicting overall risk of fracture is that the femoral neck is not affected by osteoarthritis as is the spine, which can spuriously increase spinal BMD measures.

It is also possible to estimate the lifetime risk of hip fractures for white women based on age and level of femoral neck BMD. This is illustrated in the table below.

	Lifetime Risk (Percentage) of Hip Fracture for White Women*
	Femoral Neck Bone Mineral Density T Score Age, y -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 50 49 41 33 27 21 16 13 10 60 47 40 33 27 21 17 13 10 70 46 39 33 27 21 17 13 10 80 41 35 30 24 20 16 12 10 * These estimate are based on a stochastic model using a logistic model of the relationship between bone mineral density and risk of hip fracture derived from the Study of Osteoporotic Fractures (relative risk= 2.6/SD decrease). T scores were based on bone mineral density results from Hologic QDR 1000 densitometers, with T scores derived from the National Health and Nutrition Examination Survey III 1995 database. Overall mortality rates are based on US Vital Statistics. Statistical methods are published in Black et al. JAMA, Oct 16, 2002 – Vol 288, No. 15, p. 1892. “Copyrighted 2002, American Medical Association. All rights reserved”.

Pitfalls of DEXA

Although BMD measures can predict the risk of fracture, it cannot specifically identify individual women who are going to fracture. It is merely a guide to potential risk.

Fracture risk is also dependent on other factors, in addition to BMD measurements. Pre-existing vertebral fractures in postmenopausal women is associated with a 4 fold increase in the risk of a further vertebral fracture, independent of BMD.¹⁹ However, when previous vertebral fractures are present with low BMD measurements, the predictive risk of future vertebral fracture and fractures at other sites is further increased.^20,21 Hip fracture risk is increased independent of BMD, in the setting of a maternal history of hip fracture, increasing age, predisposing conditions leading to falls, previous fractures and oestrogen deficiency states.^22-24

Spinal BMD measures with DEXA are affected by the presence of osteoarthritis in the lumbar spine and calcification of the abdominal aorta.²⁵ These conditions lead to a spuriously increased BMD measurement at the spine, which may underestimate the presence of osteoporosis. In addition it may lead to discordance between hip and spine BMD measures, particularly in the setting of significantly low BMD measures at the hip and a near normal BMD at the lumbar spine. In this setting it is appropriate to perform a plain X-ray of the lumbar spine to determine if there is arthritis in the spine. Femoral neck BMD measures by DEXA are not affected by osteoarthritis.

Content updated December 22, 2006