Patterns of Misdiagnosis and Discordance in Detecting Osteoporosis: A Comparison of Dual-energy X-ray Absorptiometry and Lumbar Computed Tomography Hounsfield Units

STUDY DESIGN

This was a retrospective chart review.

OBJECTIVE

This study aims to identify the prevalence of osteoporosis (OP) by lumbar computed tomography (CT) Hounsfield units (HUs) in patients who have normal or osteopenic bone determined by dual-energy x-ray absorptiometry (DEXA).

SUMMARY OF BACKGROUND DATA

OP is a critical issue in the postmenopausal and aging population. Bone mineral density assessment by DEXA has been described as insensitive for diagnosing OP in the lumbar spine. Improving the detection of OP can bring more patients to treatment and reduce the risks associated with low bone mineral density.

PATIENTS AND METHODS

We retrospectively reviewed all patients with DEXA scans and noncontrast CTs of the lumbar spine over a 15-year period. Patients were diagnosed as non-OP if they had a normal DEXA T -score (≥ -1) or osteopenic DEXA T -score (between -1.1 and -2.4). Patients in this cohort were considered osteoporotic by CT if L1-HU ≤110. Demographics and lumbar HUs were compared between these stratified groups.

RESULTS

A total of 74 patients were included for analysis. All patients were demographically, similar, and the average patient age was 70 years. The prevalence of OP determined by CT L1-HU ≤110 was 46% (normal DEXA: 9%, osteopenic DEXA: 63%). A significant number of males in our study were considered osteoporotic by L1-HU ≤110 (74%, P = 0.03). All individual axial and sagittal lumbar HU measurements including L1-L5 average lumbar HUs were statistically significant among non-OP and OP groups except for the lower lumbar levels ( P > 0.05 for L4 axial HUs, and L4-L5 sagittal HUs).

CONCLUSIONS

The prevalence of OP in patients with normal or osteopenic T -scores is high. Among those with osteopenia by DEXA, more than 50% may lack appropriate medical treatment. The DEXA scan may be particularly insensitive to male bone quality making the CT HU the diagnostic method of choice for detecting OP.

LEVEL OF EVIDENCE

Level III.

Patient physiology influences the MRI-based vertebral bone quality score

BACKGROUND CONTEXT

Osteoporosis is a critical issue affecting postmenopausal women and the aging population. A novel magnetic resonance imaging (MRI)-based vertebral bone quality (VBQ) score has been proposed as a method to identify poor bone quality and predict fragility fractures. The diagnostic accuracy of this tool is not well understood.

PURPOSE

To examine the ability of VBQ to predict osteoporosis and osteopenia, its correlation with dual-energy x-ray absorptiometry (DEXA), and the influence of patient-specific factors upon the score.

STUDY DESIGN

Retrospective cohort study.

PATIENT SAMPLE

Patients over the age of 18 with a DEXA scan and noncontrast, T1-weighted MRI of the lumbar spine completed within a 2-year period.

OUTCOME MEASURES

Area-under-curve (AUC) values of the VBQ score predicting osteopenia and osteoporosis when controlling for patient characteristics.

METHODS

Patients with noncontrast, T1-weighted MRIs of the lumbar spine and DEXA scans completed within a 2-year time frame were retrospectively reviewed. Patient demographics and medical risk factors for osteoporosis were identified and compared. VBQ scores were measured by two trained researchers and interrater reliability was calculated. Patients were separated into three groups defined by lowest DEXA T-score: Healthy Bone, Osteopenia, and Osteoporosis. analysis of variance, Kruskal-Wallis test, chi-square, t tests, Mann-Whitney U tests, and multivariate linear regression were performed to examine the relationship between patient characteristics, DEXA t-scores, and VBQ scores. Receiver operating characteristic analysis and AUC values were generated for the prediction of osteopenia and osteoporosis.

RESULTS

A total of 156 patients were included for analysis. Sufficient inter-rater reliability was determined for VBQ measures (intraclass correlation coefficient: 0.81). Most patients were female (83%), postmenopausal (81%), and had hyperlipidemia (64%). Patients with hyperlipidemia and healthy bone density by DEXA had elevated baseline VBQ scores (p<.001) reflective of values seen in osteopenia and osteoporosis. The AUC of the VBQ score predicting osteopenia and osteoporosis changed to be more concordant with DEXA results after controlling for hyperlipidemia (AUC=0.72, 0.70 vs. AUC=0.88, 0.89; p<.001). Sub-analysis of hyperlipidemia subtypes revealed that elevated high-density lipoprotein is associated with elevated VBQ scores.

CONCLUSIONS

Hyperlipidemia increased the MRI-based VBQ score in our healthy bone population. The high signal intensities resembled values seen in osteopenia and osteoporosis, suggesting that physiologic variables which impact bone composition may influence the VBQ score. Specifically, elevated high-density lipoprotein may contribute to this. The microarchitectural changes and the clinical implications of these factors need further exploration.