Discordance in diagnosis of osteoporosis using spine and hip bone densitometry.

Détails

Mis à jour le 24 juin 2012

Osteoporosis is a metabolic bone disorder characterized by low bone mass and microarchitectural deterioration with a subsequent increase in bone fragility and susceptibility to fracture. Dual X-ray absorptiometry (DXA) is recognized as the reference method to measure bone mineral density (BMD) accurately and reproducibly. The World Health Organization (WHO) has established DXA as the best densitometric technique for assessing BMD in postmenopausal women and based the definitions of osteopenia and osteoporosis on its results (table I) [12]. In clinical practice BMD measurements are widely used to diagnose osteoporosis and to assess its severity and changes in bone mass are commonly used as a surrogate for fracture risk [3]. The BMD values (in g/cm2) are not used for diagnosing osteoporosis. Instead a working group of the WHO proposed to define osteoporosis on the basis of the T-score (which is the difference between the measured BMD and the mean value of young adults expressed in standard deviations (SD) for a normative population of the same ethnicity) [45]. Despite its limitations; this definition is currently applied worldwide. Thus the WHO diagnostic criteria for osteoporosis define osteoporosis in terms of a T-score below −2.5 and osteopenia when T-score is between -2.5 and -1. These figures are usually calculated separately for two different sites of lumbar spine and total hip.

Table I: WHO Osteoporosis Classification

Although the BMD at different anatomic regions is correlated the agreement between sites is low when it comes to classifying individual subjects as osteoporotic or not [6]. Thus T-score discordance between the lumbar spine and total hip testing sites is a commonly observed phenomenon in densitometery. T-score discordance is the observation that the T-score of an individual patient varies from one key measurement site to another. This phenomenon has been divided into two groups: major and minor [7]. Minor discordance happens when the different diagnostic classes are adjacent; i.e. patient is diagnosed as osteoporotic in one site and osteopenic in the other site or osteopenic in one site and normal in the other site. If the diagnosis is osteoporosis in one site and the other site is in the normal range the discordance falls into the major class. Actually as the presence of discordance can affect the diagnosis and therapeutic plan in an individual person it is highly recommended to measure BMD in several sites.

Prevalence and risk factors of T-score discordance

Various studies have analyzed the prevalence and impact of T-score discordance on the management of osteoporosis [7-12]. Only two studies focused on risk factors of this commonly observed discordance [1213]. Table II shows the results of data analysis of three major studies of T-score discordance between spine and total hip measurement.

Table II: Distribution of diagnostic discordances using WHO criteria in three studies.

Numbers are presented as frequency (percentage in parenthesis).

Nelson et al. [14] reported data on 537 patients who were osteoporotic at one site and normal at another site. They found a 3.5% prevalence of major discordance between the total hip and the posteroanterior (PA) lumbar spine. Woodson data analysis about 5627 patients [7] showed that using the WHO diagnostic classification system simultaneously measured T-scores at the PA L1–L4 total spine and total hip are concordant in 56% of patients and discordant by at least one diagnostic class in 44%. Minor discordance present when the PA L1–L4 total spine and total hip sites differ by only one WHO diagnostic class was found to be common occurring in 39% of patients. Major discordance present when one site is osteoporotic and the other is normal was rare having a prevalence of only 5%. The first study to look for potential risk factors for T-score discordance enrolled 4188 persons [12]. Totally 518 participants were diagnosed in osteoporotic range in hip area and 1036 participants in the lumbar area. Major discordance was observed in BMD results of 115 (2.7%) participants. Minor discordance was observed in 1631 (38.9%) participants and T-score categories of two measurement sites in other 2442 (58.3%) participants were not different. T-score discordance was more prevalent in women than men (42.2% versus 36.5% P = 0.042). The mean age of participants with discordance (54.8 years) was higher than the other group (52.5 years P < 0.001). In 3848 female participants the number of post-menopausal women with diagnostic discordances (951 of 2027) was significantly higher than pre-menopausal participants with discordance (671 of 1821; P < 0.001). In multivariate analysis participants with late menopause (age at menopause > 50) were more likely to show T-score discordances. Obesity defined as BMI over 30 was recognized as a risk factor for major discordance and smoking as a protective factor against minor discordance. Hormone replacement therapy was a significant protector against both. To look for prevalence and risk factors of T-score discordance in our centre we studied recently 3015 persons [13]: there were 259 participants diagnosed in osteoporotic range at the hip site and 792 participants at the PA L1–L4 total spine. Major discordance was observed in BMD results of 129 (4.3%) participants. Minor discordance was observed in 1250 (41.5%) participants and T-score categories of two measurement sites in other 1636 (54.3%) participants were concordant. T-score discordance was equally observed in women and men (44.1% vs 46.0% p = 0. 42). The mean age of participants with discordance (56.1 years) was higher than the other group (52.6 years p < 0.0001). In 2486 female participants the number of post-menopausal women with diagnostic discordances (915 of 1739) was significantly higher than pre-menopausal participants with discordance (231 of 747; p < 0.0001). In multivariate analysis participants with menopause obesity (defined as BMI over 30) and history of fractures were more likely to show major T-score discordance.

T-score discordance aetiologies

Five different causes for occurrence of discordance between the spine and the hip sites have been described [7].

1.      Physiologic discordance is related to the skeleton's natural adaptive reaction to normal external and internal factors and forces. Mechanical strain especially related to weight bearing plays a key role in this kind of discordance. An example of this type of discordance is the difference observed between the dominant and non-dominant total hip.  The explanation is that weight bearing can cause rise in bone density especially in the hip and femur regions [15]. This mechanism could be the reason of more major T-score discordances observed by increment of BMI as confirmed by the multivariate analysis in Moayyeri et al. and El Maghraoui et al. studies. Moreover the spine and hips usually start out with different T-scores (the spine is said to reach peak at least 5 yrs before the hip). And finally bone loss observed with age in an individual may be more rapid and important in trabecular than cortical bone is another explanation. In both major and minor discordances observed in our study lower BMD for lumbar spine was more prevalent. The main explanation was that rates of bone loss differ substantially between the anatomic regions in the same individual [16]. Trabecular bones (typical of lumbar area) are known to have a more rapid rate of deprivation in early post-menopausal state in comparison to cortical bone (typical of proximal femur). Indeed major T-score discordances was associated with menopause in both Moayyeri et al. and El Maghraoui et al. studies (OR = 1.7; 95% CI: 1.01 – 2.7 and 6.04; 95% CI: 2.75 – 13.28 respectively).

2.      The second type of discordance described as pathophysiologic discordance and which can also be called secondary discordance is seen secondary to a disease or medication use. Two subtypes of secondary discordance can be observed. The first subtype is related to a disease responsible of falsely elevated lumbar spine T-score. Common examples observed in the elderly include vertebral osteophytosis vertebral end plate and facet sclerosis osteochondrosis and aortic calcification [1718]. Another important cause in younger patients is ankylosing spondylitis syndesmophytes  [19-21]. The abnormal calcium deposition within the field of the DXA region of interest (ROI) leads to the falsely elevated spine T-score. The second subtype is a secondary true discordance resulting from a more decreased BMD in the lumbar spine than the hips. Indeed most of the aetiologies of the secondary osteoporosis (such as glucocorticoid excess hyperthyroidism malabsorption liver disease rheumatoid arthritis) first affect spinal column [2021]. This will lead to higher prevalence of lumbar osteoporosis.

3.      Anatomic discordance is owing to differences in the composition of bone envelopes tested. An example is the difference in T-scores found for the posteroanterior lumbar spine and the supine lateral lumbar spine in the same patient.

4.      Artifactual discordance occurs when dense synthetic manmade substances are within the field of ROI of the test (e.g. barium sulphate metal from zipper coin clip or other metallic object).

5.      And finally technical discordance occurs because of device errors technician variability patients’ movements and variation due to other unpredictable sources. With respect to positioning error some studies showed that either excessive internal or external rotation of the femur during test acquisition resulted in a BMD difference of as much as 10% compared with correct positioning. This is equal to a 1 SD difference or a full T-score clearly enough to cause discordance with the spine. We demonstrated in a previous study that DXA in vivo reproducibility is two-fold better in the hips than the spine especially when measuring both hips [22]. Technical discordance can also result when the method used by the technician to analyze the test is incorrect [23]. In addition discordance can occur if the device used to obtain the test is incorrectly calibrated shifting the BMD up or down. Although in the usual case both the hip and spine would be shifted in the same direction the result of such a shift might be cause the apparent discordance. Finally technical discordance can occur due to the normative reference data used by the device software to analyze the test. This type of discordance occurs when the average BMD of the normative group used to calculate the T-score is significantly different from the average value found for the whole population.

Consequences of T-score discordance on osteoporosis management

The high prevalence of T-score discordance could induce some problems for the physicians in decision-making regarding these patients. In general high prevalence of discordance between lumbar spine and hip T-scores suggests some defects in the cut-off values for definition of osteoporosis and osteopenia proposed with the WHO. The international societies interested in osteoporosis management recommend using DXA to measure BMD in both the hip and spine and classifying the patient based on the lowest T-score of these measurements. The inconsistencies in the diagnostic classification of osteoporosis between skeletal sites lend credence to the notion that BMD should be used as only one of the factors in making therapeutic decisions when evaluating patients with osteoporosis. An international team convened by The WHO is trying to develop a globally applicable measure of absolute fracture risk based upon multiple risk factors including BMD. This could silence much of the controversy regarding choice of reference data for T-score calculation and usefulness of relatively arbitrary densitometric categorizations [14].

In summary the densitometrists and clinicians should be prepared to expect that at least four of every ten patients tested by DXA to demonstrate either minor or major T-score discordance between spine and total hip measurement sites. T-score discordance can occur for a variety of reasons related to physiologic and pathologic patient factors as well as the performance or analysis of DXA itself.

• References: