Thoracic bone mineral density measured by quantitative computed tomography in patients undergoing spine surgery

The thoracic spine is a common location for vertebral fractures as well as instrumentation failure after long spinal fusion procedures. The association between those complications and bone mineral density (BMD) are well recognized. Due to the overlying sternum and ribs in the thoracic spine, projectional BMD assessment tools such as dual energy x-ray absorptiometry (DXA) are limited to the lumbar spine. Quantitative computed tomography circumvents several shortcomings of DXA and allows for level-specific BMD measurements. Studies comprehensively quantifying BMD of the entire thoracic spine in patients undergoing spine surgery are limited.

The objective of this study was: (1) to assess the reliability of thoracic QCT measurements, (2) to determine possible level-specific BMD variation throughout the thoracic spine and (3) to assess the correlation between BMDs of the T1-T12 spinal levels.

Cross-sectional observation study.

Patients undergoing spine surgery from 2016-2020 at a single, academic institution with available preoperative CT imaging of the thoracic spine were included in this study.

The outcome measure was BMD measured by QCT.

Patients undergoing spine surgery from 2016-2020 at a single, academic institution with available preoperative CT imaging of the thoracic spine were included in this study. Subjects with previous instrumentation at any thoracic level, concurrent vertebral fractures, a Cobb angle of more than 20 degrees, or incomplete thoracic spine CT imaging were excluded. Asynchronous quantitative computed tomography (QCT) measurements of T1-T12 were performed. To assess inter- and intra-observer reliability, a validation study was performed on 120 vertebrae in 10 randomly selected patients. The interclass correlation coefficient (ICC) was calculated. A pairwise comparison of BMD was conducted and correlations between each thoracic level were evaluated. The statistical significance level was set at p<.05.

60 patients (men, 51.7%) met inclusion criteria. The study population was 90% Caucasian with a mean age of 62.2 years and a mean BMI of 30.2 kg/m². The inter- and intra-observer reliability of the thoracic QCT measurements was excellent (ICC of 0.97 and 0.97, respectively). The trabecular BMD was highest in the upper thoracic spine and decreased in the caudal direction (T1 = 182.3 mg/cm³, T2 = 168.1 mg/cm³, T3 = 163.5 mg/cm³, T4 = 164.7 mg/cm³, T5 = 161.4 mg/cm³, T6 = 152.5 mg/cm³, T7 = 143.5 mg/cm³, T8 = 141.3 mg/cm³, T9 = 143.5 mg/cm³, T10 = 145.1 mg/cm³, T11 = 145.3 mg/cm³, T12 = 133.6 mg/cm³). The BMD of all thoracic levels cranial to T6 was statistically higher than the BMD of all levels caudal to T6 (p < .001). Nonetheless, significant correlations in BMD among all measured thoracic levels were observed, with a Pearson's correlation coefficient ranging from 0.74 to 0.97.

There is significant regional BMD variation in the thoracic spine depending on spinal level. This BMD variation might contribute to several clinically relevant phenomena. First, vertebral fractures occur most commonly at the thoracolumbar junction including T12. In addition to mechanical reasons, these fractures might be partially attributed to thoracic BMD that is lowest at T12. Second, the optimal upper instrumented vertebra (UIV) for stopping long fusions to the sacrum and pelvis is controversial. The BMD of surgically relevant upper thoracic stopping points (T2-T4) was significantly higher than the BMD of lower thoracic stopping points (T10-T12). Besides stress concentration at the relatively mobile lower thoracic segments, the low BMD at these levels might contribute to previously suggested higher rates of junctional failures with short fusions.