Subjects and study design
Our analytical sample included 147 healthy subjects, with the goal of studying sex-related variations in cortical bone of humeral head region in the dominant upper extremity. Data were acquired as part of the aging and osteoporotic proximal humerus fracture Study, a single-center prospective ongoing population study of Chinese men and women. The early part of study has been reported [22]. The subjects consisted of 71 men and 76 women, with range from 20 to 88 years (mean, 49.79 years). Arm dominance was determined as the arm with which subjects would throw a ball. Subjects with a history of or evidence of metabolic bone disease, as well as those receiving chronic treatment that may affect bone metabolism were excluded from the study. Three groups of volunteers were identified according to age: group A (aged 20-39 years), group B (aged 40-59 years), and group C (aged> 60 years). There were 24 men and 22 women in group A, with average age of 29.35 years (range, 20-39 years). Group B consisted of 27 men and 22 women, with average age of 49.67 years (range, 40-59 years). Group C consisted of 20 men and 32 women, with average age of 67.98 years (range, 60-88 years). For this study, no DXA screening was performed prior to enrollment; therefore, no bone mineral density (BMD) inclusion/exclusion criteria were used. Written informed consent was obtained from all participants and the study was approved by the Institutional Review Board of Tianjin Hospital.
Cortical bone mapping
CT scans of the proximal humerus were performed with a GE Light Speed VCT (GE, Milwaukee, WI, USA). Scan parameters for CT scanners were 120 kVp, mA Auto, 5mm slice thickness, 50 cm field of view, and 512×512 matrix in spiral and standard reconstructions. CT values of pixels were recorded in Hounsfield units (HU).Subjects were positioned in supine with their arms in neutral position and centered within the gantry of the machine. Each image was analyzed from the slice that included the top of acromion to the slice that included the inferior angle of scapula. The entire CT scanning was performed by Dr. Jian Li. HU values were automatically corrected for HU values below the water equivalent of the European Forearm Phantom (EFP) by adjusting the histogram.
The cortical parameter measurement and mapping technique has been previously described [14, 15]. Cortical thickness measurement was performed using CBM, implemented by a freely available in-house program called Stradwin (http://mi.eng.cam.ac.uk/~rwp/stradwin/). First, an approximate segmentation of each proximal humerus from the CT data was performed using Stradwin, and results in a triangulated surface mesh with ~104 vertices distributed uniformly over the proximal humerus surface. Secondly, the CT data was sampled at each vertex of the mesh using 18 mm lines perpendicular to and passing through the humeral cortex and trabeculae. Finally, a model that accounts for the imaging blur was fitted to the data samples. This validated model-based deconvolution process allows the measurement of much smaller features than would normally be visible in the CT data. This process was repeated at all vertices. As a result, color maps on the proximal humerus were created for accurately estimating the cortical thickness (CTh, mm) and cortical mass surface density (CM, the cortical mass per unit cortical surface area, HUmm), as well as endocortical trabecular density (ECTD, HU). ECTD is the average trabecular density immediately adjacent to the cortex, to a depth of about 8 mm, excluding any effects due to blurring of the endocortical edge.
Definition of the regions of interest for cortical bone distribution assessment
For the evaluation of the bone morphometric analysis, specific regions of interest (ROIs) were defined within the proximal end of the humerus. The specific methodology has been described in detail previously and will be briefly outlined here [22].The cortical bone in humeral head region was defined as anterior, lateral and posterior wall. In anatomical perspective, the anterior wall is equivalent to the less tuberosity. The lateral and posterior part of greater tuberosity corresponds to lateral and posterior wall. After creation of a single 3D thickness map, the humeral head height was determined by measuring the distance between the highest point of the humeral head and the most distal margin of the articular surface (Fig. 1). The height of humeral head was then quartered by parallel plane 1-3. In each slice, to obtain more details of cortical bone tissue, the longest line (Line 1) between the joint surface and greater tuberosity was drawn; this line was divided into a medial and a lateral segment by line 2 (Fig.2). The intersection of the above two lines with cortical shell in each plane was the measurement point (ROI 1-9) (Fig.3).
Statistical Analysis
The age group difference within the same gender was compared using one-way ANOVA for normally distributed values and Kruskal-Wallis Test for non-normally distributed values. Gender and age differences were evaluated using generalized linear models (GLMs). Correlation between cortical parameters and age in ROI 1-9 was studied by linear regression analysis. All statistical analysis was performed using IBM SPSS Statistics for Windows version 20.0 (IBM SPSS Inc., Chicago, IL). Significance level was set at P<0.05 for all statistical tests.