MINERAL QUANTIFICATION IN OSTEOPOROTIC AND NORMAL WOMEN

Participants: B.R. McCreadie, J. Champion, S.A. Goldstein

Keywords: bone, mineral, osteoporosis

Introduction

Although low bone mineral density is a strong predictor of osteoporotic fractures, there is a significant overlap between those who fracture and those who do not. Recently, differences in the micro-architecture of trabecular bone have been found between these two groups. In a continuing effort to elucidate differences between osteoporotic and normal individuals, the purpose of this study was to test the hypothesis that the profile of mineral content in trabecular tissue is different between fractured and non-fractured subjects.

Materials and Methods

Bone volume fraction-matched sets of trabecular bone specimens were obtained from individuals undergoing hip arthroplasty due to hip fracture (fractured group) or from those who died without fracture (control group). Hip fractures due to extreme trauma such as a motor vehicle accident or falling from a ladder were excluded. Eighteen fracture and ten control specimens were obtained. Eight mm trabecular bone cubes were obtained from a region near the superior surface of the femoral neck, generally aligned with the principal orientation of the trabeculae, and near the fracture site in most of the fracture cases. The bone specimens were fixed, embedded in PMMA, and sectioned at 600 microns. One randomly selected section from each specimen was mounted to a plastic slide and micro-milled flat using a Reichert-Jung Polycut fit with an ultra-milling attachment. Machining artifacts were removed with a light polish of 1 micron diamond dust. The specimens were lightly coated with carbon.

Specimens were scanned with an Amray 1810 scanning electron microscope (SEM) equipped with a Robinson retractable backscattered electron detector (ETPSEMRA). The current was maintained at 165 +-1.6 (mean +- standard deviation) pA and was measured by means of a custom-fabricated Faraday cup and a picoammeter (Keithley Instruments). Four locations in each specimen were scanned. The specimens were imaged and analyzed in a random order, and the operator was blinded to their origin until the statistical analysis.

The SEM gray scale was calibrated and standardized using polished 6061 aluminum and carbon rod (grade 1, Ted Pella, Inc.) as reference materials. The reference materials were scanned before and after each specimen. The correction of each bone pixel resulted in a value that is a fractional distance between the two standard pixel values at the time of the scan, assuming a linear change in the references with time.

The remaining analysis was conducted using a procedure written in PV-Wave (Visual Numerics). First, all images were discriminated with a threshold determined by Otsu’s adaptive technique. Then, a Sobel operator was employed to find the edges between the bone matrix and background pixels (PMMA, osteocyte lacunae, etc.). These edges were removed from the identified bone pixels to insure that all the analyzed pixels were completely within the bone regions. User-identified bone regions were then grown from seed pixels to the edges determined by the thresholding and Sobel operator. The region-growing procedure effectively removed any regions with similar coloring as bone that were not bone matrix. The bone matrix pixels from the four images of each specimen were grouped together and a mean and standard deviation were calculated for each individual.

Results

No significant differences were found in the means of pixel values between fractured and control individuals (p=0.32) (fig. 1), although the average tissue mineral content for the fractured individuals was slightly lower. There was a significantly greater standard deviation in the fractured individuals (p=0.0004) (fig. 2). When the standard deviation of the pixel values is plotted against the mean for each individual, it becomes clear that there is a difference between the two groups (fig. 3). This difference can be quantified by the slope of the best-fit lines for each group, both of which are significantly different from zero (p=0.003 for fracture group, p=0.01 for control).

The results of this study suggest that the average bone tissue mineral content is not altered from normal in osteoporotic individuals, although there is more variability in those who sustained a fracture. These differences in mineralization profiles may have mechanical consequences due to localized regions of stiffer tissue alongside regions of relatively compliant tissue. Future research may investigate the spatial distribution of the high- and low-mineral content tissue.

Status

This paper has been submitted for publication.

Figure 1. Comparison of the means of the Figure 2. Comparison of the standard

pixel values, corresponding to a comparison deviations of the pixel values, corresponding