Osteoporosis is emerging as a serious health problem worldwide^24,33). An accurate examination of the bone mineral density (BMD) is important for treating the elderly of osteoporosis; it affects the prediction of the clinical prognosis, such as the fracture risk calculation, and the treatment choice¹⁸⁾. As the elderly population increases, the proportion of patients with osteoporosis is expected to increase. In particular, in spinal fusion surgery, osteoporosis is closely related to the clinical outcome because the fusion rate and degree of subsidence differ according to the patient’s BMD. The prevalence of osteoporosis in patients requiring spinal surgery has also increased.

Dual energy X-ray absorptiometry (DXA) is the gold standard method of measuring the areal BMD of the lumbar spine and hip¹⁰⁾. This technique determines the two-dimensional BMD, including both the trabecular and cortical bone. During DXA measurements, two X-ray beams are directed at the bone. The bone density can be calculated from the absorption of each beam by the bone after subtracting the beam quantity absorbed by soft tissue²⁵⁾. In contrast, quantitative computed tomography (QCT) is a representative technology to complement DXA. With the progression of computed tomography (CT) diagnosis technology, QCT was developed to measure bone density. QCT is performed on a CT scanner using a calibration phantom and dedicated computer software for analysis. Calibration phantoms are devices composed of material similar to the bone mineral structure. they are needed to convert the attenuation measured in Hounsfield units into BMD values^3,15). In an estimation of the BMD, the trabecular bone is more important than the cortical bone because most of the bone mineral metabolism occurs in the trabecular bone. QCT can differentiate the cancellous bone so that it is more sensitive to the detection of osteoporosis and for monitoring the changes in the bony structures than DXA^1,23). Owing to the difference between DXA and QCT, clinicians sometimes experience conflicting results from the two studies. For example, the BMD measured by DXA may be overestimated due to increased bone surface with degenerative changes^7,31,39). Another study reported that 55% to 90% of osteoporotic fractures appeared in individuals without a diagnosis of osteoporosis by DXA²⁸⁾.

Despite the advanced volumetric measurements by QCT, DXA remains the most popular diagnostic measurement with several advantages, including low radiation dose, rapid data analysis, medium price, and easy handling^2,18). In most cases, clinicians make decisions on patients with potential osteoporosis using only DXA. Therefore, it is important to recognize the risk factors that cause an overestimation of the DXA values. Many studies have reported that patients with obesity, bone degenerative changes, and aortic calcification may have an overestimate BMD value when examined with DXA³⁶⁾. On the other hand, most studies lacked advanced spinal imaging, such as CT and magnetic resonance imaging (MRI). This study examined the factors causing the discrepancy between DXA and QCT using imaging data, such as spinal MRI and CT of the lumbar spine.

The discrepancy between the DXA and QCT was attributed to technical differences during BMD acquisition. QCT can evaluate trabecular and cortical bone independently. QCT measured BMD volumetrically and interpreted in three dimensions, whereas DXA evaluates areal bone density and reflects the amount of cortical and trabecular tissue present within a vertebrae structure. It has been reported that intra- and an interobserver compromise was outstanding in L1 density measurement by QCT. According to Gerety et al.¹¹⁾, there were intra-class correlation coefficients for each observer’s measurements at two separate time points with a coefficient for observer 1 of 0.94 (95% CI, 0.88-0.97) and observer 2 and 3 of 0.99 (95% CI, 0.98-1). They also presented interobserver measurements had an intra-class correlation coefficient of 0.98 (95% CI, 0.96-0.99)¹¹⁾. For these reasons, QCT has less impact on degenerative changes on BMD compared to DXA¹⁵⁾.

The interruption maximizes at the lumbar spine compared to the femur. Previous studies reported that several risk factors, including spondylosis and aortic calcification, affect the areal BMD measured by DXA. On the other hand, their results were limited because prior studies relied on only plain radiographs. In contrast, the present study was conducted with additional spinal imaging results, such as CT and MRI. Therefore, an evaluation of a more detailed structural classification was possible with this study, which had never been recognized before. As a result, two novel risk factors showed a stronger relationship with the discrepancy between DXA and QCT.

Measurements at a fracture site are a form of technical error regarding the DXA outputs. After obtaining a spine bone density image, the interpreting physician should mark the ROI. According to the guideline, the region should include the inferior portion of T12 and the superior portion of L5⁴⁾. Setting the ROI, the physician easily excludes the area with artificial devices, such as augmenting cement and spinal instrumentation. Detecting and excluding fracture sites are inconvenient using only an unclear anteroposterior bone density scan. As a result, fracture sites are often included in ROI at DXA examination in clinical practice. The present study showed that the inclusion of the fracture sites in the ROI resulted in a powerful misinterpretation of DXA. This would have resulted from the DXA calculation using the areal BMD. The condensed bony structure by a compression fracture may lead to an overestimation of the BMD because a wedging deformity cannot be detected in two-dimensional images. Therefore, the interpreting physician needs to confirm the presence of a wedging deformity with a lateral plain radiograph if possible.

Degenerative changes of the lumbar spine are a confounding factor of DXA measurement^12,13,16).

According to Yu et al.³⁹⁾, lumbar degenerative changes were found in all 114 patients in a group whose BMD was overestimated at the DXA examination, but they focused only on the anterior spinal column. One study reported a significant increase in BMD in subjects with vertebral osteophytes. In contrast, Reid et al.³²⁾ reported that-spinal osteophytes are unlikely to interfere with the BMD estimation, which is similar to the results of the present study. In this study, the discordance group showed a higher KL grade of spondylosis, which evaluates the degenerative changes in the vertebral osteophytes, but the difference failed to reach statistical significance. On the other hand, facet degeneration grade was a significant factor affecting the measurements from DXA. In previous papers, it has not been reported intensively whether facet joint hypertrophy has a greater effect on overestimated BMD at the DXA examination. The facet degeneration grade by Weishaupt et al.³⁸⁾ also concentrat ed on the narrowing of joint space and sclerosis. Nevertheless, the most significant change by a higher grade would be hypertrophy of the facet joint. Facet joint hypertrophy appears to be in proportion to the volumetric increase in the total bony structures⁹⁾.

The positional relationship during DXA scanning was evaluated by the presence of scoliosis and lumbar lordosis. None of these two factors showed a significant difference between the concordance and discordance groups. On the other hand, these results do not mean that the position is irrelevant in an examination. The technician is required to confirm whether the patient has a straightforward position. The result cannot be compared if the patient had a different position on repetitive scans from the baseline study.

AAC also influences the BMD result because DXA scanning cannot distinguish between a pure vertebral structure and calcification within an adjacent blood vessel²¹⁾. In this study, AAC showed a significant difference between the two groups but failed to improve the significance in multivariate analysis. Li et al.²³⁾ reported that 41/140 patients (29.3%) were diagnosed with osteoporosis only by QCT, not by DXA. A relatively large proportion of patients (25 patients, 61.0%) were found to have AAC. AAC also acts as an important independent determinant of cardiovascular disease³⁵⁾. A previous study reported that the severity of aortic calcifications was negatively correlated with the BMD results at the lumbar spine²⁷⁾.

Two patients were diagnosed with osteoporosis on DXA but osteopenia on QCT. They were included in the concordance group in this study. They consisted of one male and one female. They had relatively short stature (male, 164 cm; BMI, 24.72; female, 149 cm; BMI, 19.82). There has been a report that a short stature can be associated with an inaccurate measurement of bone mass³⁴⁾. The female patient was on the treatment of osteoporosis with denosumab. It is reported that desunomab reduces cortical porosity and increases the volume ratio of trabecular bone to total bone volume⁴⁰⁾. Although underestimation of DXA could be caused by changes in the proportion of cortical bone depending on the drug and medical condition, the exact cause of underestimation of DXA was unclear in these two patients.

This study had several limitations. This research had a retrospective design with a relatively small sample size. Furthermore, because this study was conducted in a single institution, the results might have been affected by the distinct characteristics of the area. A multi-center study with a larger number of patients from diverse regions will be needed to decrease the possibility of bias.

QCT showed a higher detection rate of osteoporosis than DXA. DXA might miss a diagnosis of osteoporosis. Facet joint hypertrophy due to degeneration revealed a stronger association with the overestimation of the BMD by DXA than the formation of osteophytes at the vertebral body. The inclusion of a region-of-interest at the fracture site, which is a common technical error, also resulted in the incorrect measurement of osteoporosis by DXA. Therefore, the results of DXA should be interpreted carefully.