Assessment of Bone Fragility in Patients With Multiple Myeloma Using QCT-Based Finite Element Modeling.

Multiple myeloma (MM) is a malignant plasma cell disease associated with severe bone destruction. Surgical intervention is often required to prevent vertebral body collapse and resulting neurological complications; however, its necessity is determined by measuring lesion size or number, without considering bone biomechanics. Finite element (FE) modeling, which simulates the physiological loading, may improve the prediction of fragility. To test this, we developed a quantitative computed tomography (QCT)-based FE model of the vertebra and applied it to a dataset of MM patients with and without prevalent fracture. FE models were generated from vertebral QCT scans of the T12 (T11 if T12 was fractured) of 104 MM patients, 45 with fracture and 59 without, using a low-dose scan protocol (1.5 mm slice thickness, 4.0 to 6.5 mSv effective dose). A calibration phantom enabled the conversion of the CT Hounsfield units to FE material properties. Compressive loading of the vertebral body was simulated and the stiffness, yield load, and work to yield determined. To compare the parameters between fracture and nonfracture groups, t tests were used, and standardized odds ratios (sOR, normalized to standard deviation) and 95% confidence intervals were calculated. FE parameters were compared to mineral and structural parameters using linear regression. Patients with fracture showed lower vertebral stiffness (-15.2%; p = 0.010; sOR = 1.73; 95% CI, 1.11 to 2.70), yield force (-21.5%; p = 0.002; sOR = 2.09; 95% CI, 1.27 to 3.43), and work to yield (-27.4%; p = 0.001; sOR = 2.28; 95% CI, 1.33 to 3.92) compared to nonfracture patients. All parameters correlated significantly with vBMD (stiffness: R2 = 0.57, yield force: R2 = 0.59, work to yield: R2 = 0.50, p < 0.001), BV/TV (stiffness: R2 = 0.56, yield force: R2 = 0.58, work to yield: R2 = 0.49, p < 0.001), and Tb.Sp (stiffness: R2 = 0.51, yield force: R2 = 0.53, work to yield: R2 = 0.45, p < 0.001). FE modeling identified MM patients with compromised mechanical integrity of the vertebra. Higher sOR values were obtained for the biomechanical compared to structural or mineral measures, suggesting that FE modeling improves fragility assessment in these patients. © 2016 American Society for Bone and Mineral Research.

Association of QCT Bone Mineral Density and Bone Structure With Vertebral Fractures in Patients With Multiple Myeloma.

Computed tomography (CT) is used for staging osteolytic lesions and detecting fractures in patients with multiple myeloma (MM). In the OsteoLysis of Metastases and Plasmacell-infiltration Computed Tomography 2 study (OLyMP-CT) study we investigated whether patients with and without vertebral fractures show differences in bone mineral density (BMD) or microstructure that could be used to identify patients at risk for fracture. We evaluated whole-body CT scans in a group of 104 MM patients without visible osteolytic lesions using an underlying lightweight calibration phantom (Image Analysis Inc., Columbia, KY, USA). QCT software (StructuralInsight) was used for the assessment of BMD and bone structure of the T11 or T12 vertebral body. Age-adjusted standardized odds ratios (sORs) per SD change were derived from logistic regression analyses, and areas under the receiver operating characteristics (ROC) curve (AUCs) analyses were calculated. Forty-six of the 104 patients had prevalent vertebral fractures (24/60 men, 22/44 women). Patients with fractures were not significantly older than patients without fractures (mean ± SD, 64 ± 9.2 versus 62 ± 12.3 years; p = 0.4). Trabecular BMD in patients with fractures versus without fractures was 169 ± 41 versus 192 ± 51 mg/cc (AUC = 0.62 ± 0.06, sOR = 1.6 [1.1 to 2.5], p = 0.02). Microstructural variables achieved optimal discriminatory power at bone thresholds of 150 mg/cc. Best fracture discrimination for single microstructural variables was observed for trabecular separation (Tb.Sp) (AUC = 0.72 ± 0.05, sOR = 2.4 (1.5 to 3.9), p < 0.0001). In multivariate models AUCs improved to 0.77 ± 0.05 for BMD and Tb.Sp, and 0.79 ± 0.05 for Tb.Sp and trabecular thickness (Tb.Th). Compared to BMD values, these improvements of AUC values were statistically significant (p < 0.0001). In MM patients, QCT-based analyses of bone structure derived from routine CT scans permit discrimination of patients with and without vertebral fractures. Rarefaction of the trabecular network due to plasma cell infiltration and osteoporosis can be measured. Deterioration of microstructural measures appear to be of value for vertebral fracture risk assessment and may indicate early stages of osteolytic processes not yet visible.