Three-Dimensional Characterization of Trabecular Bone Mineral Density of the Distal Radius Utilizing Quantitative Computed Tomography

Abstract

Background: Distal radius (DR) fractures demonstrate patterns of predictable fragments. Bone mineral density (BMD) measurements of these regions of interest (ROIs) may guide more precise treatment. Methods: Computed tomography (CT) scans of the DR of 42 healthy volunteers (23 female) were analyzed using quantitative CT software, measuring BMD within trabecular bone. Seven ROIs were described by alignment with the distal (volar ulnar distal [VUD], dorsal ulnar distal [DUD], volar radial distal [VRD], and dorsal radial distal [DRD]) or proximal (middle ulnar proximal [MUP], middle proximal [MP], and middle radial proximal [MRP]) sigmoid notch. Additional ROIs were the radial styloid (RS) and metadiaphysis (MD). A general estimation equation assessed subject’s BMDs with predictive factors of gender, ROI, and age. The interaction between gender, ROI, and age was included in the model to allow for differences in ROI to vary with gender and/or age. Results: Comparing ROIs within the same gender and, separately, within the same age group revealed significantly higher BMD adjacent to the radioulnar and radiocarpal joints. Male and female individuals aged ≥50 years (mean: 172.7 mg/cm³ ± 6.1) had significantly lower BMD than those aged <50 years (mean: 202.7 mg/cm³ ± 5.8) when all ROIs were considered. Males had higher mean BMD at each ROI compared with females; these differences were significant in 5 of the 9 ROIs: VUD, DUD, DRD, RS, MUP. Conclusions: Trabecular BMD of the DR is highest adjacent to the radioulnar and radiocarpal joints. Female patients and those ≥50 years have lower trabecular BMD.

Introduction

Fractures of the distal radius (DR) are among the most common of all fractures, and their incidence is increasing both in the United States and abroad.^5,9,20,21 Nellans et al have demonstrated that this increase is occurring across pediatric, adult, and elderly populations.²¹ Multiple treatment options exist for DR fractures; however, open reduction and internal fixation (ORIF) with volar locked plates is quickly becoming the treatment of choice by a significant number of hand surgeons.³²

The incidence of osteoporosis in the United States is rising with an increasing elderly population.²⁷ The reduction of bone mineral density (BMD) in persons with osteoporosis makes them particularly susceptible to fragility fractures, including those of the DR. While advanced age remains the best predictor of osteoporosis, postmenopausal women are disproportionately affected.^26,27 Current estimates suggest that 50% of women and 20% of men will suffer from a fragility fracture in their lifetime.²⁴

Numerous studies of DR fractures have demonstrated predictable fracture patterns including volar ulnar, dorsal ulnar, volar radial, dorsal radial, and radial styloid (RS) fragments.^{2,16,18,19,30,35} This has encouraged some authors to advocate for fragment-specific fixation of intra-articular fractures.^7,28,32 BMD measurements of these regions of interest (ROIs) would improve our understanding of this commonly fractured bone’s anatomy and guide more precise treatment of these injuries. Furthermore, given that low BMD has been linked to fixation failure, knowledge of areas of maximum BMD in the DR may improve fixation techniques and implant designs by targeting these areas.^23,25

Quantitative computed tomography (QCT) is a validated technology that allows for measurement of trabecular and cortical BMD using computed tomography (CT) scans.^22,24 It was developed as an alternative to dual-energy x-ray absorptiometry (DEXA) in the diagnosis of osteoporosis.¹⁷ Unlike DEXA, QCT is able to look at trabecular bone separately from cortical bone, which can be advantageous in the diagnosing osteoporosis due to the higher turnover rate of trabecular bone making QCT more sensitive.²⁴ Furthermore, QCT has the advantages over DEXA of better avoiding confounders such as joint-space narrowing, osteophyte formation, and extraosseous calcification that can all artificially elevate DEXA results.^8,13,33,38

QCT’s potential for use in clinical research has been demonstrated. Diederichs et al determined that QCT could be used to map the trabecular and cortical BMD of the distal humerus in cadavers.⁶ Furthermore, MacNeil and Boyd illustrated that QCT could be used to predict bone strength in the DR.¹⁵ Recently, modern QCT techniques have been shown to allow for validated BMD analysis to be performed on CT scans that were obtained for routine purposes (eg, trauma evaluations, screening colonography, etc), expanding research potential to virtually any CT scan.^1,24,25

Previous authors have used QCT to study DR bone morphology.^11,15 However, to our knowledge, never before has the trabecular bone of the adult DR been mapped in a 3 dimensions with QCT. Furthermore, we are not aware of any previous study has used QCT to describe the bone morphology of the DR in the context of commonly described fracture patterns.

The aim of this study was to describe the three-dimensional (3D) distribution of trabecular BMD of the DR using quantitative computerized tomography. We hypothesized that trabecular bone adjacent to joints such as the radiocarpal joint and distal radioulnar joint (DRUJ) would have the highest BMD of the DR when compared with other areas. Our supposition follows Wolff law, predicting higher trabecular BMD near areas of bone loading.⁴ Secondarily, we hypothesized that in all areas of the DR, trabecular BMD would be higher in males and patients <50 years of age. This postulate follows well-described patterns of lower BMD in females and older persons.^26,27

Materials and Methods

CT scans of the DR and hand of 46 healthy volunteers were obtained as a retrospective analysis of data acquired from a study on musculoskeletal health in the upper extremity after institutional review board approval (No. 403511) and appropriate informed consent was obtained from all individual participants included in the study. Inclusion criteria were age ≥18 years with skeletal maturity, a CT scan including the wrist from the study institution, and an unexceptional appearing DR on review of the CT scan without evidence of osteoporosis or prior fracture. As such, each scan was reviewed by the first author to rule out any evidence of abnormality of the DR including previous trauma, bony tumors, osteoarthritis, deformity, or hardware. Other exclusion criteria included history of metabolic bone disease and poor-quality CT scans preventing satisfactory QCT analysis. Four patients were excluded from the study: Two patients had poor-quality CT scans that could not be reliably analyzed with the QCT software, 1 patient had a bone cyst in the metadiaphysis (MD), and 1 had a bone island in the volar portion of the DR. Thus, 42 patients (23 females; 19 males) with a mean age of 42 years (range: 18-75 years) were included in the final analysis. Each scan was obtained on one of 2 machines with thin (0.625 mm: 23 patients; 0.600 mm: 19 patients) cuts to ensure optimization of data quality.

Each patient’s CT scan was converted from standard DICOM files to QCT files that could be analyzed using quantitative CT software (QCT Pro Version 6.1, Model 4 CT Calibration Phantom, MindWays Software Inc, Austin, Texas). The QCT software is normalized to the CT scanner to optimize accuracy of data collection using a phantom device of known density and is able to calculate BMD in specific areas within a CT scan defined by 3D location and volume. These “regions of interest” were designated for the purposes of this study to be spheres 5 mm in diameter positioned in critical points throughout the DR based on predictable fracture patterns described previously (Figure 1).^{2,16,18,19,30,35} ROI sphere diameter of 5 mm was found to be optimal during study piloting as this size was easily reproducible in the QCT software ensuring precision between samples and consistently allowed for positioning on the scans such that maximum measurement of trabecular bone was achieved without catching cortical bone and causing spurious increases in BMD readings.

Figure 1.

Regions of interest.

Note. The study regions of interest (ROIs) are pictured. The top row of computed tomography cuts displays and names each ROI in the axial plane, and the bottom row shows each ROI’s corresponding location in the coronal plane.

Nine ROIs focusing on trabecular BMD were examined in each patient. Trabecular BMD, which is selectively affected more so than cortical BMD in osteoporosis, correlates with both risk of fragility fracture and risk of failure of fixation.²³ Furthermore, modern volar locked plating techniques for the treatment of DR fractures call for unicortical screws to avoid extensor tendon irritation, relying more on trabecular bone for stability of the screw/plate/bone construct than bicortical screws.³⁷ Therefore, our study focused on trabecular BMD of the DR to maximize clinical applicability.

Four ROIs were aligned in the coronal plane with the distal sigmoid notch with 2 volar and 2 dorsal ROIs in the axial plane at this level: volar ulnar distal (VUD), dorsal ulnar distal (DUD), volar radial distal (VRD), and dorsal radial distal (DRD). One ROI was placed in the center of the RS. These 5 ROIs in particular were chosen based on previously described fracture patterns of the DR (Figure 1).^{2,16,18,19,30,35}

Three ROIs were aligned in the coronal plane with the proximal sigmoid notch. At this level, the points were positioned across the center of the trabecular bone in the axial plane: middle ulnar proximal (MUP), middle proximal (MP), and middle radial proximal (MRP). One final ROI was placed at the center of the radial canal in the axial plane at the level of the MD in the coronal plane. These 4 ROIs were chosen to complete the BMD map of the DR (Figure 1).

The locations of all ROIs were standardized, and each scan was individually processed by the first author to ensure measurement precision. Similar to previous upper extremity QCT analysis protocols, easily recognizable bony landmarks, such as Lister’s tubercle and the sigmoid notch, were used as guides to determine optimal placement of each ROI sphere on each scan.⁶ To further maximize reliability of ROI placement, CT slice images with spheres in place were saved during initial analysis. These were reviewed after all scans had been analyzed to ensure precision between patients.

A general estimation equation (normal distribution) was used to assess subject BMD with the predictive factors of gender, region of interest, and age. Age was categorized into 2 discrete groups (age <50 years: 24 patients, age ≥50 years: 18 patients) to be included in the model. The interaction between gender, region of interest, and age was included in the model to allow for differences in region of interest to vary with gender and/or age. A Tukey adjustment was used to maintain family wise alpha at 0.05 across multiple comparisons. Sandwich estimation was used to adjust for any model misspecification. In all cases, statistical significance was set to P < .05 a priori. Post hoc power analysis for each comparison was performed. SAS was used for all statistical analyses (version 9.4, Cary, North Carolina).

Results

Gender and Region of Interest

A significant relationship was found between gender and region of interest (P = .001). When comparing discreet ROIs within the same gender across all age categories, there were numerous significant differences (Figure 2). DR trabecular BMD was significantly higher in more ulnar and distal locations adjacent to the DRUJ and radiocarpal joint, with the highest BMD being in the ROIs at the level of the distal sigmoid notch (VUD, DUD, VRD, DRD). Interestingly, while males had significantly higher BMD ulnarly compared with radially at this level, neither males nor females showed any difference when comparing volar with dorsal (Figure 2).

Figure 2.

Distal radius trabecular BMD by gender.

Note. The statistical analysis of the quantitative computed tomography data comparing all regions of interest (ROIs) to each other among the study females (a) and the study males (b) is demonstrated. At each box, the mean bone mineral density (BMD) of the ROI on the x-axis is subtracted from the mean BMD of the ROI on the y-axis giving a difference. If the difference is positive (y-axis > x-axis), the box is red; if the difference is negative (y-axis < x-axis), the box is blue. If there is no difference, the box is white. Significant differences are labeled with an “*.” The key on the far right can be used to determine the amount of BMD difference for each comparison.

Region of Interest and Age Category

No interaction was found between region of interest and age category (P = .222). When comparing discreet ROIs in patients either <50 or ≥50 years across gender, there were numerous significant differences (Figure 3). Again, DR trabecular BMD was significantly higher in more ulnar and distal locations adjacent to the DRUJ and radiocarpal joint, with the highest BMD being in the ROIs at the level of the distal sigmoid notch (VUD, DUD, VRD, DRD). Interestingly, while patients ≥50 years had significantly higher BMD ulnarly compared with radially at this level, neither age group showed any difference when comparing volar with dorsal (Figure 3).

Figure 3.

Distal radius trabecular BMD by age.

Note. The statistical analysis of the quantitative computed tomography data comparing all regions of interest (ROIs) to each other among study patients <50 years (a) and study patients ≥50 years (b) is demonstrated. At each box, the mean bone mineral density (BMD) of the ROI on the x-axis is subtracted from the mean BMD of the ROI on the y-axis giving a difference. If the difference is positive (y-axis > x-axis), the box is red; if the difference is negative (y-axis < x-axis), the box is blue. If there is no difference, the box is white. Significant differences are labeled with an “*.” The key on the far right can be used to determine the amount of BMD difference for each comparison.

Gender and Age Category

No interaction was found between gender and age category (P = .711). When all ROIs were combined across sex, male and female individuals aged ≥50 years (mean: 172.7 mg/cm³ ± 6.1) had significantly lower mean BMD than those aged <50 years (mean: 202.7 mg/cm³ ± 5.8) (P = .001; 95% confidence interval [CI], 12.902-46.970). When each ROI was compared separately, individuals aged ≥50 years had lower mean BMD at each ROI; however, none of these differences reached significance (Figure 4a).

Figure 4.

Distal radius trabecular BMD heat maps—age comparisons.

Note. Heat maps comparing regions of interest in all patients <50 years (ulnar semicircles) with all patients ≥50 years (radial semicircles), all male patients <50 years (ulnar semicircles) with all male patients ≥50 years (radial semicircles), and all female patients <50 years (ulnar semicircles) with all female patients ≥50 years (radial semicircles) are displayed (a, b, c, respectively). Significant differences are labeled with an “*.” On the far right is a heat map key that can be used to determine the mean bone mineral density (BMD) for each semicircle.

Similarly, when males ≥50 years were compared with males <50 years, males ≥50 years had lower mean BMD at each ROI; nevertheless, none of these differences were significant (Figure 4b). When females ≥50 years were compared with females <50 years, females ≥50 years had lower mean BMD at each ROI; however, the only region with a significant difference was the RS (Figure 4c).

Gender Comparisons

When all ages were combined, males had higher mean BMD at each ROI compared with females. In this case, the difference was significant in 5 ROIs: VUD, DUD, DRD, RS, MUP (Figure 5a). When males ≥50 years were compared with females ≥50 years, females had lower mean BMD at each ROI; though, the difference was only significant in 2 ROIs: DUD and RS (Figure 5b). When males <50 years were compared with females <50 years, females had lower mean BMD at each ROI; however, none of these differences were significant (Figure 5c).

Figure 5.

Distal radius trabecular BMD heat maps—gender comparisons.

Note. Heat maps comparing regions of interest in all male patients (ulnar semicircles) with all female patients (radial semicircles), all male patients ≥50 years (ulnar semicircles) with all female patients ≥50 years (radial semicircles), all male patients <50 years (ulnar semicircles) with all female patients <50 years (radial semicircles) are displayed (a, b, c, respectively). Significant differences are labeled with an “*.” On the far right is a heat map key that can be used to determine the mean bone mineral density (BMD) for each semicircle.

Post Hoc Power Analysis

A post hoc power analysis was performed for each statistical comparison assuming alpha error of 0.05 and beta error of 0.80. For within gender comparisons (Figure 2, Figure 4b and 4c), with a sample size of 23 females (SD: 52.7) and 19 males (SD: 61.7), the study was powered to find a difference of 64.6 mg/cm³ and 84.1 mg/cm³, respectively. For within age category comparisons (Figure 3, Figure 5b and 5c), with a sample size of 18 for individuals ≥50 years (SD: 59.0) and 24 for those <50 years (SD: 58.6), the study was powered to find a difference of 83.1 mg/cm³ and 70.1 mg/cm³, respectively.

As seen in Figures 2 and 3, when different ROI locations throughout the DR where compared with other ROI locations (eg, comparing DUD with RS) in the context of gender or age, they frequently differed from each other by 60 to 100 mg/cm³, suggesting that the study was appropriately powered for this analysis. However, as seen in Figures 4b, 4c, 5b, and 5c, when the same ROI was compared between subgroups (eg, comparing DUD between males ≥50 years and <50 years), the differences were generally much smaller. Thus, the study was underpowered for subgroup analysis of the same ROIs.

Nevertheless, when comparing the same ROI between major categories including all individuals ≥50 years versus all individuals <50 years patients (Figure 4a) and all males versus all females (Figure 5a), the study was powered to detect a difference of 53.7 mg/cm³. This increase in power reflects the increased sample size to 42 with these kinds of comparisons in contrast to the subgroups above. For gender comparisons (Figure 5a), many observed differences ranged from 40 mg/cm³ to 60 mg/cm³, suggesting appropriate power. However, for age comparisons (Figure 4a), most observed differences ranged from 20 mg/cm³ to 40 mg/cm³, suggesting the study was underpowered.

Discussion

Joints represent areas of bone loading; therefore, according to Wolff law, it would follow that trabecular bone adjacent to joints would tend to have higher BMD.⁴ For example, in a QCT study looking specifically at the subchondral plate of the DR, Rhee and Baek found that BMD was higher in the scaphoid and lunate fossae compared with the interfossa ridge or the outer edge of the subchondral plate.²⁹ Our study further exhibits Wolff law in the DR by demonstrating patterns of higher trabecular BMD in more ulnar locations (adjacent to the sigmoid notch of the DRUJ) and more distal locations (adjacent to the radiocarpal joint). The only exception was that the RS, while generally of higher BMD than ROIs at or proximal to the proximal sigmoid notch (MUP, MP, MRP, MD), had mostly lower BMD than the distal sigmoid notch ROIs (VUD, DUD, VRD, DRD). However, it would make sense that these ROIs which are adjacent to the scaphoid and lunate fossae would have higher BMD than the RS considering these areas, unlike the styloid, are central in the principal axis of load from the carpus to the DR.¹²

Given these findings, surgeons treating DR fractures with volar locked plates can optimize screw purchase by positioning plates such that distal ulnar-sided screws neighbor, but do not penetrate, the sigmoid notch. Furthermore, volar plates should be placed as distally as possible without crossing the watershed line (to avoid flexor tendon irritation) such that distal screws find maximum purchase just proximal to the radiocarpal joint.³⁴ In fact, these strategies can guide the order of fixation especially in cases with more comminution. To obtain optimal reduction and rigid fixation, it may be helpful to start reduction and fixation maneuvers on the volar-ulnar side of the DR.

Interestingly, no difference within either gender or age group was found between dorsal ROIs and volar ROIs at the level of the distal sigmoid notch. Thus, our study provides no evidence that the dorsal trabecular bone of the DR is comparatively higher in BMD than volar. This may be reassuring to surgeons who use shorter screws to avoid extensor tendon irritation when performing volar locked plating and could be one explanation as to why biomechanical studies have demonstrated that constructs with shorter screws have no loss in stability.^3,36,37

When all ages were considered, our study confirmed the hypothesis that male patients would have higher DR trabecular BMD than female patients. Interestingly, this was true for all ROIs; however, the difference was only significant in 5 of the 9 studied: VUD, DUD, DRD, RS, MUP. Based on these data, surgeons treating DR fractures with ORIF should be aware that female patients may not have as strong fixation distally as male patients and that locking screws in distal locations may be even more important in females than in males to avoid failure of fixation.

Our data also confirmed the hypothesis that patients <50 years of age have higher DR trabecular BMD than patients ≥50 years of age when both genders were combined. Interestingly, this significance was only present when all ROIs were considered together. Thus, while we can make a general claim that BMD is lower in the DR of patients ≥50 years of age, we cannot direct surgeons treating these fractures with ORIF how to best alternatively position screws in older patients.

Previous studies have demonstrated a stronger negative correlation between advancing age and BMD in females compared with males.¹⁴ In our subgroup analyses, our data demonstrated significant differences when comparing females ≥50 years with females <50 years at the RS and when comparing males ≥50 years with females ≥50 years at the RS and DUD region. Otherwise, while males and females <50 years had higher BMD than males and females ≥50 years, respectively, at all ROIs, these differences did not achieve further significance. Similarly, while males ≥50 and <50 years had higher BMD than females ≥50 and <50 years, respectively, at all ROIs, no further significance in these differences was achieved. Given these findings, we are unable to further comment on fixation goals in these subgroups.

The threshold for clinical relevance concerning differences in trabecular BMD among several ROIs in the same anatomical area as measured by QCT is difficult to define. In one study using QCT to look at age-related changes of BMD in the DR in women, it was predicted that overall DR trabecular BMD would decrease by 76.1 mg/cm³ from age 20 to 89 years old.¹⁰ Although this example is extreme, several of our statistically significant differences are at or above this number. In another study of premenopausal women, those who suffered DR fractures had lower overall mean trabecular BMD (146 ± 32 mg/cm³) by QCT than controls (172 ± 33 mg/cm³).³¹ This represents a 15% difference or about 26 mg/cm³. A vast majority of our statistically significant findings are well over this threshold, suggesting our findings have clinical relevance.

Not surprisingly, the most at-risk group for poor screw purchase during volar plating of DR fractures appears to be females ≥50 years. Our data indicate that the RS may be a particularly problematic area in this subgroup. Thus in addition to optimizing plate position to achieve locking screw purchase in subchondral bone distally and ulnarly, it may be worthwhile in this subgroup to enhance RS fixation by maximizing the length of this screw without penetrating the joint, using a plating system that allows for 2 RS screws, or using a separate radial column plate. This would be particularly important in intra-articular fractures with a free RS fragment at risk for escape.

Our study has several limitations. First, as the CT scan data were recorded anonymously to protect patient confidentiality, we had limited access to demographics which may have been helpful in further characterizing DR trabecular BMD. Second, although our study did look at a healthy number of CT scans, it was underpowered particularly for the subgroup analyses which may have caused false negative findings. Finally, our study only looked at healthy subjects and did not look at outcomes of DR fractures.