SUN-380 Opportunistic Screening with Abdominal CT in Patients with Diabetes Can Identify Those at High Risk of Osteoporosis and Osteopenia

Abstract

Background: Diabetes mellitus (DM) increases the risk of fracture at any given bone mineral density (BMD). However, the optimal strategy for osteoporosis screening with DXA is unknown in those with DM. A previously described strategy in the general population known as “Opportunistic Osteoporosis Screening” uses computer tomography (CT) images done for other reasons to assess the attenuation (density) of L1 in Hounsfield units (HU)—this was found to correlate with DXA-derived T-score. However, neither the methodology nor the cut-points have been specifically validated in those with DM. Thus, the goal of this study was to examine the performance of this methodology and define thresholds corresponding with low BMD in those with DM. Methods: This was retrospective study using electronic medical record data. Patients with DM were identified by ICD code. Those with both abdominal CT and DXA within a 6-month period were included, excluding patients with CKD stage 5, solid organ transplantation, bariatric surgery, or L1 hardware. L1 attenuation, measured by 2 readers on sagittal view, were averaged. A different reader assessed for vertebral fractures. Fractures of the hip, forearm, humerus, and pelvis were identified by ICD code. The lowest T-score of lumbar spine, femoral neck, total hip, or forearm (available in 11 subjects) was used to compare to L1 attenuation. ROC curves were derived from univariate logistic regression. Results: 320 subjects met study criteria; 10 (3.2%) had vertebral fractures, 8 (2.6%) had prior major non-vertebral fracture, and 33 (10.3%) had osteoporosis by BMD.The 18 subjects with major fractures had lower T-scores (-2.3 ± 1.4 vs. -0.8 ± 1.4, p<0.001) and lower L1 attenuation (104 ± 46 vs. 149 ± 47 HU, p<0.001). T-score and L1 attenuation had similar discrimination for prior fracture by area under the ROC curve (0.77 vs. 0.76, p=NS). Moderate osteopenia (T-score -1.5 or less) and L1 attenuation of 130 HU or less had identical sensitivities (72.2% for both) and similar specificities (69.2% vs. 62.5%, respectively) for prior fracture.In regards to L1 attenuation corresponding to DXA diagnosis of osteoporosis, 160 HU was 94% sensitive, while 110 HU was 80% specific. This is similar to the 90% sensitivity for 160 HU and 90% specificity for 110 HU previously reported in the general population. Given higher fracture risk in DM, moderate osteopenia (n=106) was also examined as an outcome: 130 HU was 61% sensitive and 71% specific. This threshold had similar or improved sensitivity and specificity among subgroups of insulin users, men, and women under age 65. Conclusion: Our results validate the use of opportunistic osteoporosis screening in patients with DM, which could help clinicians decide on the need for screening DXA. Patients with diabetes and L1 attenuation below 130 HU on CT scan should be considered for DXA screening to formally assess the risk of fracture.