Dual-Energy X-ray Absorptiometry Measurement and Accuracy of Bone Mineral After Unilateral Total Hip Arthroplasty

The Journal of Arthroplasty Vol. 21 No. 4 2006

Dual-Energy X-ray Absorptiometry Measurement and Accuracy of Bone Mineral After Unilateral Total Hip Arthroplasty Roy D. Bloebaum, PhD,*yz Derek W. Liau, MD,§ D. Kevin Lester, MD,t and Teri G. Rosenbaum, BS*yz

Abstract: The standard technique for monitoring bone mineral in hip arthroplasty has been dual-energy x-ray absorptiometry (DEXA). The accuracy of DEXA in the cortical bone adjacent to femoral components has not been established. This study evaluated bone mineral in the cortical bone adjacent to the femoral component comparing DEXA and ashing. Seven pairs of human femora from postmortem donors with unilateral hip implants were examined. Twenty-eight ashed core specimens from both the medial and lateral sides were taken. Cortical bone loss was seen to be greater in the proximal and medial regions of the implanted femora. Dualenergy x-ray absorptiometry failed to show an acceptable level of accuracy compared with ash data (r = 0.56; P = .002). It did show relative patterns of bone loss. Bone loss was consistent with implant-induced stress shielding. Key words: Dual-energy x-ray absorptiometry, total hip arthroplasty, DEXA accuracy, bone remodeling. n 2006 Elsevier Inc. All rights reserved.

Total hip arthroplasty (THA) has been a highly successful procedure in the treatment of degenerative bone disease. Despite its overwhelming success, some failures occur, causing patient morbidity,

reduced quality of life, and incurring of significant rehabilitation costs. One mechanism of clinical failure in THA continues to be loosening of the femoral component [1]. Clinical failure of the femoral component has been often caused by low bone density of the underlying trabecular bone [2]. This low bone density has been attributed to stress shielding. Stress shielding and subsequent bone mineral loss result from the mechanical unloading and principles of Wolff’s law [3]. Periprosthetic bone loss may also allow for subsidence of the prosthesis. Therefore, bone density determination in the proximal human femur, both preoperatively and in postoperative follow-up, would help surgeons in terms of prosthesis choice, fixation method, and monitoring bone remodeling in response to implant modulus and design. Dual-energy x-ray absorptiometry (DEXA) was developed for the quantitative assessment of bone mineralization [4] and has become a common method for determining cancellous bone mineral density (BMD) in the proximal femur [5,6].

From the *Bone and Joint Research Laboratory, Salt Lake City Veterans Administration Health Care System, Salt Lake City, UT; y Department of Bioengineering, University of Utah, UT; z Department of Orthopedic Surgery, School of Medicine, University of Utah, UT; § School of Medicine, University of Washington, WA; andt Department of Orthopedics, University of California San Francisco–Fresno, Fresno, CA. Submitted January 26, 2004; accepted November 16, 2005. Benefits or funds were received in partial or total support of the research material described in this article from Senator Warren G. Magnuson Scholarship, FY 2005 Pre-Doctoral Associated Health Rehabilitation Research Fellowship Program, The Department of Orthopaedics, University of Utah, and Research and Development (R&D) Medical Research Service, Department Veterans Administration Salt Lake City Health Care System. Reprint requests: Roy D. Bloebaum, PhD, Bone and Joint Research Laboratory (151F), 500 Foothill Blvd, Salt Lake City, UT 84148. n 2006 Elsevier Inc. All rights reserved. 0883-5403/06/1906-0004$32.00/0 doi:10.1016/j.arth.2005.11.010

612

DEXA Accuracy in Total Hip Arthroplasty ! Bloebaum et al 613

Because of its ability to compensate for soft tissue, DEXA has emerged as the current gold standard in determining bone density in the lumbar spine and proximal femur. Numerous studies have also shown DEXA to have the necessary precision [7] and accuracy [8,9] for determining BMD in cancellous bone, but questions still remain concerning the accuracy of DEXA bone mineral measurements of the cortical bone [10]. It is known that an implant affects DEXA scans by creating a bshadowing Q of the bone. Recently developed software allow DEXA to overcome effects of the implant on BMD measurements. Several studies on periprosthetic bone mineral have shown an acceptable level of precision error (1.4%-3.8%) and reliability using DEXA scanning on THA patients with variations in femoral rotation and repositioning [4,11-15]. Although some studies have shown its accuracy compared with ash studies on the long bones of rats [16-18], a clear understanding of the accuracy of DEXA measurements of the cortical bone adjacent to a femoral component has not been established in the literature. In addition to rat bones, several attempts have been made to assess the accuracy of DEXA measurements using ashed bones of spines [9], hydroxyapatite phantoms, and hydroxyapatite phantoms with overlying hip implants [11,19]. A study by Kilgus et al [14] showed 1.1% accuracy using a calcium hydroxyapatite powder in lucite plates. The investigators also attempted to analyze DEXA scans with ash data from an implant placed in one femur postmortem. They reported that it was not possible to control for important variables that rendered their cadaveric accuracy portion invalid. Kilgus et al [14] also noted that they could not resect precise segments of cadaver bone that had been scanned as regions of interest (ROIs) by DEXA. Second, they could not obtain reliable ash data because of the limited number of samples: one implanted femur. Lochmuller et al [20] performed studies on cadaveric femora without components comparing DEXA and ash data. Although a correlation was found between the 2, the results suggested that accuracy errors of DEXA in the femur limit the prediction of actual mechanical failure loads. This study coincides with studies by Currey [21-23] and Currey et al [24] which indicated that differences in mineralization have a profound effect on a bone’s mechanical properties. Engh et al [25] and Sychterz and Engh [26] conducted cadaveric studies examining periprosthetic bone remodeling using DEXA. They examined human cadaveric femora implanted unilaterally with a femoral component. A postmortem femoral

component was implanted in the contralateral control femur. No accuracy study was performed; neither were comparisons made with ash data. In this study, the objective was to first evaluate bone mineral content (BMC) using DEXA and ash studies measuring the cortical bone next to femoral components using donor femora implanted in situ for various periods. The purpose was to examine regional differences in BMD and ash data by comparing the unimplanted and implanted femora from the same donor to determine if, and to what extent, stress shielding may have occurred. Donors in this study had similar implant in situ times as did the donors in the studies by Engh et al [25] and Sychterz and Engh [26]. The second objective was to evaluate DEXA accuracy in measuring the cortical bone relative to ash data. The null hypothesis tested was that DEXA software could predict mineralization results similar to ash data. The contralateral unimplanted side was used as the control; the unimplanted mineral content obtained through ashing and DEXA scans was compared with the implanted femur.

Materials and Methods Specimen Preparation Fourteen human femora were obtained from 7 donors at autopsy; each donor (all were white: 2 males and 5 females; age range = 52 to 94 years; mean F SD = 82 F 14.8 years) had a unilateral THA. The average length of time in vivo for the implants ranged from 6 to 130 months (mean F SD = 55.4 F 46.2 months). Implants were all press fit, nonporous coated, and implanted with the Alloclassic5 (1 donor), Alloclassic6 (3 donors), Allopro4 (2 donors), and Vector5 (1 donor). Allopro was renamed as Alloclassic, and the Vector femoral component was the generic version of the Alloclassic. Therefore, all the femoral components in this study had the same design, geometry, and material composition (titanium alloy). The donors’ weights ranged from 82 to 222 lb (mean F SD = 138 F 44 lb). All donors died of cardiac arrest. Three donors had a diagnosis of degenerative joint disease, and 4 had fractured at the time of hip arthroplasty. None of the donors took therapeutic drugs that could have influenced bone density. All implants were pressfit types and were clinically successful at the time of the donors’ death, without radiologic evidence of loosening. The femora were manually cleaned of soft tissue and stored in 70% alcohol at the time of death.

614 The Journal of Arthroplasty Vol. 21 No. 4 June 2006 Dual-Energy X-ray Scans Bone mineral density and BMC were determined with a dual-energy x-ray scanner (Lunar ExpertXL, Lunar Corporation, Madison, Wis) by a certified technician with settings consistent with those of a clinical diagnostic setting. The software, Orthopedic Lunar Expert Version 1.91 (Lunar Corporation), used specialized automated algorithms to detect the soft tissue-bone and boneimplant surfaces that automatically excluded the shadowing effects produced by the metallic implants. To simulate soft tissue in the clinical setting, we placed the femora in an acrylic tank filled with 15 cm of water. Water submersion of the proximal femur has been demonstrated to accurately simulate surrounding soft tissue found in the clinical setting [8,13]. The neck and head of the femur or implant were oriented perpendicularly each time to the DEXA beams. This limited rotational effects that can alter DEXA measurements [11]. Analysis of Scans To allow us to define the same areas on differently sized femora and implants and obtain the same scan area on the contralateral femur, we measured the implant on the x-rays and the bones were marked with pins at 25%, 45%, 65%, and 85% of the length of the implants. These percentages marked ROIs and compensated for the varying lengths of different sizes of hip implants and their effects on the surrounding bone (Fig. 1). Each of the 4 ROIs scanned was 1 cm in height.

Fig. 2. Radiographs of a femur pair with a unilateral femoral component. The pins mark the ROIs or levels normalized to the percentage of implant length from which dual-energy x-ray and ashing measurements were taken from medial and lateral regions.

To analyze both the medial and lateral aspects of the diaphysis, we shot the DEXA scans anteroposteriorly. To maintain consistency between the specimens, we analyzed the 1-cm regions proximal to the 4 marker pins that demarcated our ROIs (Fig. 2). Level 1 was the most proximal ROI and level 4 the most distal ROI. The DEXA machine scanned each level in an anteroposterior orientation in 1-cm-high sections that included the medial and lateral aspects of the femur. Both the implanted femora and the contralateral femora were scanned at the same 4 levels. The DEXA software generated BMD (g/cm2) and BMC (g) for each region. Bone mineral density was influenced by the amount of porosity, collagen, and degree of mineralization, whereas BMC was influenced by amount of porosity, collagen, cortical bone thickness, and degree of mineralization. Ash Density

Fig. 1. Schematic drawing of the ROIs or levels normalized to the percentage of implant length from which dual-energy x-ray and ashing measurements were taken from medial and lateral regions.

After the DEXA scan and x-ray findings were obtained, the bones were cored medially and laterally at the same levels of the DEXA analysis (proximal to the pins at 25%, 45%, 65%, and 85% of the length of the implants). The same strict methods used to determine ash density have been used and cited in peer-reviewed journals [27-29]. Cores were taken using a standard drill press and custom-made machine core drill bits with a 9-mm inner diameter. These cortical bone samples represented 90% of the cortical bone scanned by DEXA. This far exceeds most statistical standards for

DEXA Accuracy in Total Hip Arthroplasty ! Bloebaum et al 615

Fig. 3. Percentage of loss attributable to the femoral component at each level as measured by dual-energy x-ray and ash studies (n = 7) showing a general trend of greater proximal than distal bone mineral losses when comparing unimplanted with implanted femora. Ash percentage showed a much lower percentage of mineral loss when compared with other measurements.

between unimplanted and implanted bones. These ratios provided 2 assessments of the comparative reliability between DEXA BMC and ash weight. If the ratios differed greatly, then this would indicate that the DEXA approach did not provide a reliable bone density measurement as it would not agree with ash data. Comparison of ratios between the DEXA and ash approaches was done using paired-sample t tests. The P values of the paired t tests were adjusted for multiple comparisons using Hommell’s multiplecomparison procedure [31]. Finally, the percentage of bone mineral loss attributable to the implant was calculated as the difference in a bone measurement (unimplanted minus implanted bone in same donor) divided

Table 1. Percentage of Loss Attributable to the Femoral Component at Each Level and Medial/Lateral Region as Measured by Ash Weight and Ash Percentage (n = 7) n

sample regions of tissue processing [27-29]. Cores were prepared for ashing, which determines the percentage of the bone mass that is composed of mineral. First, the bones were placed in chloroform for 21 days to remove fats and oils. They were then dried in an oven at 808C for 5 days, returned to room temperature in a desiccator, and weighed. The latter step measured dried defatted weight. The bones were then placed in an oven at 5808C for 24 hours to burn off the organic matrix. After returning to room temperature in a desiccator, the bones were weighed again. This step measured ash weight, which reflected the thickness of the 9-mm core, degree of mineralization, and porosity. Ash percentage was calculated as Ash Weight/Dried Defatted Weight * 100 and only reflected the degree of mineralization. Statistical Methods Four cross-sections were taken from 7 pairs of human femora, providing 28 bone sampling sites. Twenty-eight ashed core specimens were taken for comparison with the DEXA data. Statistical analysis was conducted using Pearson’s correlation coefficient computed for the pairs of measurements (DEXA and ash data) made on each bone specimen [30]. Ratios of bone densities of implanted bone and unimplanted bone in the same donor were computed. Differences in bone density were examined

Percentage of loss* (mean)

95% confidence interval

Ash weight (medial) Level 1 6 69.4 42.7 to 96.0 Level 2 7 71.2 58.9 to 83.5 Level 3 6 64.7 52.4 to 77.0 Level 4 7 34.5 15.7 to 53.3 Ash weight (lateral) Level 1 4 58.9 23.1 to 94.6 Level 2 5 45.3 25.6 to 64.9 Level 3 7 54.8 24.7 to 85.0 Level 4 7 37.7 4.1 to 71.2 Ash weight (average of medial and lateral) Level 1 7 70.3 47.0 to 93.6 Level 2 7 60.6 48.5 to 72.7 Level 3 7 59.7 41.9 to 77.5 Level 4 7 36.1 12.4 to 59.8 Ash percentage (medial) Level 1 6 6.7 0.7 to 12.8 Level 2 7 5.6 0.9 to 10.2 Level 3 6 3.3 1.4 to 5.1 Level 4 7 1.2 0.3 to 2.1 Ash percentage (lateral) Level 1 4 8.3 0.0 to 16.6 Level 2 5 3.2 0.5 to 6.9 Level 3 7 2.3 0.8 to 5.4 Level 4 7 1.2 1.3 to 3.6 Ash percentage (average of medial and lateral) Level 1 7 9.4 3.5 to 15.3 Level 2 7 4.7 1.2 to 8.2 Level 3 7 2.6 0.3 to 4.8 Level 4 7 1.2 0.5 to 2.8

Adjusted Py

.002 b.001 b.001 .004 .027 .009 .013 .034 .001 b.001 b.001 .010 .035 .035 .024 .035 .152 .191 .254 .300 .032 .049 .067 .129

The data show a general trend of greater proximal than distal and greater medial than lateral bone mineral losses when comparing unimplanted with implanted femora. *For example, Ash Weight Percentage of Loss = (Unimplanted Ash Weight  Implanted Ash Weight)/(Unimplanted Ash Weight)  100%. yOne-sample t test comparing mean percentage of loss with 0, with P values adjusted for multiple comparisons using Hommel’s procedure.

616 The Journal of Arthroplasty Vol. 21 No. 4 June 2006 by the unimplanted bone measurement times 100. To illustrate: Ash Weight Percentage of Loss ¼ ðUnimplanted Ash Weight  Implanted Ash WeightÞ= ðUnimplanted Ash WeightÞ  100%:

Ninety-five percent confidence intervals for the percentage of loss were reported. The same method was used to find the percentage of bone mineral loss from the medial/lateral sides within the same femur.

Results Results of Bone Mineral Loss Attributable to the Implant Bone mineral losses attributable to implant-induced stress shielding as measured by DEXA and ash studies by each level are shown in Fig. 3. The graphs in Fig. 3 show greater bone mineral losses in the proximal regions compared with the fourth distal region. In addition, Fig. 3 shows the discrepancy between ash percentage–estimated bone mineral losses and other measurements. These descriptive statistics are shown in Tables 1 and 2. The bone mineral measurements compared the unimplanted with the implanted femur pairs and the mineral losses attributed to implant-induced stress shielding. Fig. 4 shows the percentage of bone mineral loss attributable to the femoral component in the medial/ Table 2. Percentage of Loss Attributable to the Femoral Component at Each Level and Medial/Lateral Region as Measured by Dual-Energy X-ray BMD and BMC (n = 7)

DEXA BMD Level 1 Level 2 Level 3 Level 4 DEXA BMC Level 1 Level 2 Level 3 Level 4

n

Percentage of loss* (mean)

95% confidence interval

Adjusted P y

7 7 7 7

48.9 55.3 47.0 33.3

37.8-60.0 31.1-79.4 26.4-67.5 14.8-51.9

b.001 .003 .003 .005

7 7 7 7

71.4 57.3 59.9 42.8

58.6-84.2 36.3-78.3 48.0-71.8 32.4-53.3

b.001 .001 b.001 b.001

The data show a general trend of greater proximal than distal and greater medial than lateral bone mineral losses when comparing unimplanted with implanted femora. *For example, DEXA BMD Percentage of Loss = (Unimplanted BMD  Implanted BMD)/(Unimplanted BMD)  100%. yOne-sample t test comparing mean percentage of loss with 0, with P values adjusted for multiple comparisons using Hommel’s procedure.

Fig. 4. Percentage of bone mineral loss attributable to the femoral component at each level and medial/lateral region as measured by ash percentage (n = 7). The data showed a general trend of greater proximal than distal and greater medial than lateral bone mineral losses except in levels 1 and 4 compared with the contralateral unimplanted femora.

lateral regions at each level as measured by ash percentage. The results showed the general trend of greater proximal than distal and greater medial than lateral bone mineral losses except in level 1. Level 1 was the only region where lateral bone loss exceeded medial bone loss according to ash percentage measurements. Level 4 showed symmetric bone loss in the medial and lateral regions. Fig. 5 shows ash weight measurements of these same medial/lateral bone loss discrepancies at all 4 levels, with the medial region having higher bone loss in levels 1 to 3. Level 4 showed the least bone

Fig. 5. Percentage of loss attributable to the femoral component at each level and medial/lateral region as measured by ash weight (n = 7) showing a general trend of greater proximal than distal and greater medial than lateral bone mineral losses except in level 4 compared with the contralateral unimplanted femora.

DEXA Accuracy in Total Hip Arthroplasty ! Bloebaum et al 617 Table 3. Percentage of Loss of Medial/Lateral Regions Within the Same Femora in Both Unimplanted and Implanted Femora Measured by Ash Studies Percentage n of loss* (mean)

95% confidence interval

Ash weight (unimplanted) Level 1 6 76.7 64.3 to 89.2 Level 2 6 10.7 13.9 to 35.3 Level 3 7 36.2 104.8 to 32.4 Level 4 7 6.9 10.0 to 23.8 Ash weight (implanted) Level 1 4 39.7 61.7 to 141.0 Level 2 6 94.2 188.3 to 0.2 Level 3 6 53.0 178.4 to 72.3 Level 4 7 16.7 27.7 to 61.2 Ash percentage (unimplanted) Level 1 6 7.7 2.3 to 17.6 Level 2 6 3.6 12.0 to 4.7 Level 3 7 0.2 0.6 to 1.0 Level 4 7 0.6 0.05 to 1.2 Ash percentage (implanted) Level 1 4 7.6 2.8 to 12.4 Level 2 6 5.9 10.1 to 1.6 Level 3 6 0.4 3.8 to 3.0 Level 4 7 0.6 1.2 to 2.4

Adjusted P y

b.001 .357 .357 .357 .393 .199 .393 .393 .312 .547 .547 .154 .045 .052 .761 .761

The P values show that in all but level 1, the difference was not statistically significant. *For example, Ash Weight Percentage of Loss = (Medial Ash Weight  Lateral Ash Weight)/(Medial Ash Weight)  100%. yOne-sample t test comparing mean percentage of loss with 0, with P values adjusted for multiple comparisons using Hommel’s procedure.

loss, with the lateral region showing a more significant ( P = .034) bone loss than the medial region. Both the ash weight and ash percentage

Fig. 7. Dual-energy x-ray BMC and ash weight measurement pairs in implanted femora shown with a linear regression line and Pearson’s correlation coefficient. Dual-energy BMC scans in implanted femora showed a high correlation with ash weight and, therefore, acceptable accuracy of the mineral content of the femora.

showed the same general trend of greater proximal than distal and greater medial than lateral bone mineral losses except in levels 1 and 4. Medial vs lateral bone mineral losses within the same femur were also studied using ash weight and percentage data (Table 3). The descriptive statistics showed that according to ash studies, there was no statistically significant bone mineral difference between the medial and lateral regions ( P N .05) within the same femur except at level 1. Level 1 showed a significantly higher mineral level in the medial region (Table 3). Results of Dual-Energy X-ray Accuracy Studies on Implanted Femora Comparisons between DEXA-derived BMD and ash percentage showed that BMD scans moderately predicted the degree of mineralization of the implanted femora (Fig. 6). A low-to-moderate Table 4. Descriptive Statistics of the Ash and DualEnergy X-ray Measurements for the Implanted Femora

Fig. 6. Dual-energy x-ray BMD and ash percentage measurement pairs in the implanted femora shown with a linear regression line and Pearson’s correlation coefficient. Dual-energy BMD scans in implanted femora showed a low/moderate correlation with ash percentage and, therefore, unacceptable predictability of the degree of mineralization of the femora.

Mean Median SD Minimum Maximum n

Ash percentage

Ash weight

DEXA BMD

DEXA BMC

65.3 66.0 3.4 57.9 69.6 28

0.12 0.09 0.09 0.01 0.28 28

0.8 0.6 0.5 0.1 2.2 28

1.0 0.8 0.7 0.1 2.7 28

618 The Journal of Arthroplasty Vol. 21 No. 4 June 2006 80

Table 5. Descriptive Statistics of the Ash and DualEnergy X-ray Measurements for the Unimplanted Femora

Ash Percentage 70

75

r = 0.01 , P = .955

Ash percentage

Ash weight

DEXA BMD

DEXA BMC

68.4 68.5 2.5 61.9 77.4 28

0.27 0.23 0.12 0.11 0.55 28

1.5 1.5 0.5 0.7 2.7 28

2.2 2.1 0.9 1.2 4.6 28

60

65

Mean Median SD Minimum Maximum n

.5

1

1.5 2 DEXA AP BMD (g/cmˆ2)

2.5

3

Fig. 8. Dual-energy x-ray BMD and ash percentage measurement pairs in the unimplanted femora shown with a linear regression line and Pearson’s correlation coefficient. Dual-energy BMD scans in unimplanted femora showed no correlation and, therefore, no predictability of the degree of mineralization of the femora.

correlation was observed between DEXA BMD and ash percentage (r = 0.62; P b .001) of the implanted femora. Although BMD lacked a high predictive value for ash percentage, DEXA BMC and ash weight in the implanted femora showed a statistically significant strong correlation (r = 0.90; P b .001) and high

levels of accuracy between these measurements (Fig. 7). Descriptive statistics for DEXA and ash measurement comparisons for implanted femora are shown in Table 4. Results of Dual-Energy X-ray Accuracy Studies on Unimplanted Femora Comparisons between DEXA-derived BMD and ash percentage showed that the BMD scans did not predict the degree of mineralization of the unimplanted femora (Fig. 8). No correlation was observed when comparing DEXA BMD and ash percentage measures for unimplanted femora (r = 0.01; P = .955). In addition, comparisons between DEXA-derived BMC and ash weight showed that the BMC scans only vaguely predicted the degree of mineralization in unimplanted femora. A low correlation (r = 0.56; P = .002) was observed

.3

.4

Ratio Unimplanted Implanted percentage* [mean (SD)] [mean (SD)] [mean (SD)]

.2

Ash Weight

.5

.6

Table 6. Comparison of Ratios of Implanted and Unimplanted Femoral Cortical Bone Mineral in the Same Donor (n = 7)

.1

r = 0.56 , P = .002 1

2

3 DEXA AP BMC (g)

4

5

Fig. 9. Dual-energy x-ray BMC and ash weight measurement pairs in unimplanted femora shown with a linear regression line and Pearson’s correlation coefficient. Dual-energy BMC scans in unimplanted femora showed a low correlation and, therefore, unacceptable accuracy of the mineral content of the femora.

Ash weight Level 1 Level 2 Level 3 Level 4 All levels DEXA BMC Level 1 Level 2 Level 3 Level 4 All levels

Py

0.21 0.23 0.31 0.34 0.27

(0.13) (0.09) (0.13) (0.12) (0.12)

0.05 0.09 0.13 0.20 0.12

(0.06) (0.06) (0.09) (0.08) (0.09)

29.7 39.4 40.3 63.9 43.3

(25.2) (13.1) (19.2) (25.6) (23.9)

Refy Ref Ref Ref Ref

2.2 2.0 2.2 2.5 2.2

(0.9) (0.7) (0.9) (1.1) (0.9)

0.7 0.9 0.9 1.5 1.0

(0.6) (0.7) (0.6) (0.7) (0.7)

28.6 42.7 40.1 57.2 42.2

(13.9) (22.7) (12.9) (11.3) (18.2)

0.882 0.561 0.967 0.373 0.692

The data show similar ratios and, therefore, high reliability of dual-energy x-ray BMC scans when compared with ash weight. *Ratio of implanted bone to unimplanted bone  100. yPaired t test comparing DEXA ratio with ash weight ratio (the referent group) at the same level, with P values not adjusted for multiple comparisons (adjusted P values would be even less significant).

DEXA Accuracy in Total Hip Arthroplasty ! Bloebaum et al 619 Table 7. Comparison of Ratios of Implanted and Unimplanted Femoral Cortical Bone Mineral in the Same Donor (n = 7) Ratio Unimplanted Implanted percentage* Adjusted Py [mean (SD)] [mean (SD)] [mean (SD)] Ash percentage Level 1 68.2 Level 2 67.6 Level 3 68.8 Level 4 69.0 All levels 68.4 DEXA BMD Level 1 0.9 Level 2 1.5 Level 3 1.7 Level 4 1.8 All levels 1.5

(4.4) (2.5) (0.9) (0.8) (2.5)

61.7 64.3 67.0 68.2 65.3

(3.4) (2.6) (1.6) (1.2) (3.4)

90.6 95.3 97.4 98.8 95.6

(6.4) (3.8) (2.5) (1.8) (4.9)

Refy Ref Ref Ref Ref

(0.3) (0.4) (0.4) (0.5) (0.5)

0.5 0.7 0.9 1.2 0.8

(0.3) (0.5) (0.6) (0.6) (0.5)

51.1 44.7 53.0 66.7 53.9

(12.0) (26.1) (22.2) (20.1) (21.2)

b.001 .003 .003 .004 b.001

The data show large ratio differences and, therefore, a low reliability of dual-energy x-ray BMD scans when compared with ash percentage. *Ratio of implanted bone to unimplanted bone  100. yPaired t test comparing DEXA ratio with ash percentage ratio (the referent group) at the same level, with P values adjusted for multiple comparisons.

between DEXA BMC and ash weight measures for unimplanted femora (Fig. 9). Descriptive statistics for DEXA and ash measurement comparisons for unimplanted femora are shown in Table 5. Results of Unimplanted vs Implanted Reliability Studies The ratios using DEXA BMC compared with ash weight were not significantly different (Table 6). The ratios using dual-energy BMC (~ 42%) were not significantly different from the ash weight (~ 43%). Therefore, using DEXA BMC data showed high reliability when comparing implanted and unimplanted femur pairs using DEXA vs ash weight. Ratios using the DEXA BMD approach (~ 42%) and the ash percentage approach (~ 95%) were significantly different. Therefore, using DEXA BMD data (Table 7) was unreliable when comparing implanted and unimplanted femur pairs for the purpose of estimating ash percentage ratios.

Discussion Both DEXA and ash data from this study showed significant bone loss going from proximal to distal in the cortical bone surrounding the femoral component when comparing implanted with unimplanted femora in the same donor. The stress shielding that was observed in the proximal 3 regions showed a more significant bone loss

medially than laterally in all but the most distal fourth level of implantation where radiographic analysis supported that implant impingement occurred. Bone mineral losses in the medial and lateral regions of the implanted femur compared with the unimplanted indicated that there was only significant bone loss medially and not laterally. In comparing bone mineral losses in this study with those in the study by Sychterz and Engh [26], our ash and DEXA BMC data showed more bone loss at all levels compared with their donor population (n = 11). Our bone losses according to DEXA BMC were 71.4% proximal, 58% midsectional, and 42.8% distal. Our ash weight data showed similar losses (70.3%, 60%, and 36.1%, respectively). Their losses according to BMC were 42.1% proximal, 23% midsectional, and 5.5% distally [26]. Although similarities exist between this study and those by Engh et al [25] and Sychterz and Engh [26], such as implant times, it is difficult to suggest direct comparisons with Engh et al’s [25] bone mineral loss results because of variables such as sex population differences and implant types. The donors in this study consisted of 71% females vs 40% in the study by Engh et al [25] and 55% in that by Sychterz and Engh [26]. The larger percentage of males in their population may account for the lesser amount of BMC loss because of differences in female bone composition and remodeling response to femoral prostheses [32]. Sychterz and Engh [26] found that females exhibited a significantly higher (12.3%) bone loss than males. The population in this study also had reduced body weights compared with their population (62.6 vs 68.8 kg). There were also differences in implant types between the implants used this study and the 11 implants used by Engh et al [25] and Sychterz and Engh [26]. Sychterz and Engh [26] used an uncemented extensively porous-coated Anatomic Medullary Locking prosthesis with a circular crosssection made of cobalt chromium [25,26], whereas the femoral component used in this study was an uncemented grit-blasted prosthesis with a rectangular cross-section made of titanium alloy. Despite these differences, both implant types led to greater proximal bone loss compared with distal bone loss. In the studies by Engh et al [26,33], a femoral prosthesis was inserted postmortem to decrease variability in DEXA BMC readings owing to electron scatter caused by metal implants. The insertion of the femoral implant ex vivo may have caused significant bone loss from the implantation process by reaming out excessive periprosthetic cortical bone. This reaming may have caused artifact-reducing bone mineral loss measurements

620 The Journal of Arthroplasty Vol. 21 No. 4 June 2006 by decreasing the amount of bone originally present in the unimplanted femur. Because Engh et al [25] and Sychterz and Engh [26] did not have ash studies validating DEXA BMC reliability, the authors needed to place an implant in the control femur to eliminate any variability in DEXA scans between the 2 study groups. The results of this study on DEXA BMC reliability compared with ash weight indicated that this was unnecessary. High reliability ratios were found (Table 6) when comparing the unimplanted and implanted femora. This indicated that the new DEXA absorptiometry software was able to compensate for prosthetic effects on BMC scans when compared with ash weight as the standard, thus negating the need for femoral implant insertion ex vivo into the unimplanted control femur. In the second hypothesis, testing DEXA accuracy, the data indicated that BMC accuracy predicted ash weight in the implanted femora (r = 0.90) and to a lesser extent in the unimplanted femora (r = 0.56). Bone mineral density had a low predictive value for ash percentage in the implanted (r = 0.62) and no predictive value in the unimplanted (r = 0.01) femora. The lower correlations between BMD and ash percentage as compared with BMC and ash weight were likely caused by porosity and collagen contributing to BMD but not to ash percentage. Porosity has been shown to contribute to 71.6% of the variance in BMD in unimplanted femora [34]. Because BMD and ash percentage correlated in the implanted femora but not in the unimplanted femora, perhaps the stress shielding caused by implantation resulted in the amount of porosity and collagen having a smaller impact on BMD, thereby resulting in a correlation between BMD and ash percentage in the implanted femora. When measuring differences between the implanted and unimplanted femora, both BMD and BMC demonstrated high reliability for determining relative bone changes after THA that agreed with the ash data. Therefore, BMD and BMC can be used to predict mineralization differences between implanted and unimplanted femora. In assessing DEXA accuracy, this study is unique. Past accuracy studies have shown high correlations between DEXA compared with hydroxyapatite phantoms, rat bones, and lumbar spines [11,16-19]. In contrast, this study used human cadaveric femora with implanted femoral components compared with ash data to assess accuracy in mineralization. Although Kilgus et al [14] attempted to assess accuracy using ashing data from an implant placed in one cadaveric femur, they could

not complete their study because of imprecise bone resection and the small sample size (n = 1). Studies by Currey [22,35] and Currey et al [24] showed that even small differences in BMC have shown large changes in the mechanical properties of bone as measured by modulus of elasticity, bending strength, impact energy, and work of fracture. A recent study by Lochmuller et al [20] showed that the ash weight was a better predictor of femoral failure loads (r = 0.78; P b .01) than DXA (r = 0.67; P b .01). If an accurate correlation could be found between ash data, DEXA readings, and mechanical properties in the periprosthetic cortical bone in the proximal femur, then bone mineral data from scans may possibly be used in the future to measure mineral changes in cortical bone surrounding femoral components and accurately estimate biomechanical properties of bone postoperatively. This study is unique in its testing of DEXA vs ash data in the periprosthetic cortical bone. Ash percentage data in this study showed similar patterns of bone mineral losses but a much lower percentage of bone losses at each level as compared with DEXA BMC, BMD, and ash weight (see Tables 1 and 2; Fig. 3). This experimental discrepancy may be caused by porosity, collagen, or cortical bone thickness contributing to the DEXA BMC, BMD, and ash weight measurements. Future mechanical studies need to be conducted to gain an understanding of the meaning of the mineralization, porosity, collagen, and cross-sectional geometry changes in relation to mechanical properties of bone in the proximal femur. In summary, DEXA is most reliable in measuring relative mineral changes in unimplanted vs implanted femora but not actual bone mineral changes that occurred in the cortical bone surrounding the implant. In addition, Lochmuller et al [20] showed that accuracy errors of femoral measurements by DEXA limit the actual prediction of mechanical failure loads and that, hence, the measurements have less correlation with biomechanical performance. Future DEXA studies on THA should be aware of the limits raised in this study and in others such as that by Lochmuller et al [20]. Further investigations should separately quantify changes in bone such as its mineralization, porosity, collagen, and bone geometry to better understand the effect of femoral components on proximal stress shielding in the human femur.

Acknowledgments Funding for this study was provided by a Senator Warren G. Magnuson Scholarship grant and the

DEXA Accuracy in Total Hip Arthroplasty ! Bloebaum et al 621

Veterans Affairs Pre-Doctoral Associated Health Rehabilitation Research Fellowship Program. We thank the Research and Development Medical Research Service of the Department of Veterans Administration Salt Lake City Health Care System, the Department of Orthopedics of the University of Utah, and the University of Washington School of Medicine for their support. We also thank Greg Stoddard for assisting with this study’s statistics and Erin Whitaker for assisting in data collection.

References 1. Malchau H, Herberts P, Ahnfelt L. Prognosis of total hip replacement in Sweden. Follow-up of 92,675 operations performed 1978-1990. Acta Orthop Scand 1993;64:497. 2. Bobyn JD, Mortimer ES, Glassman AH, et al. Producing and avoiding stress shielding. Laboratory and clinical observations of noncemented total hip arthroplasty. Clin Orthop 1992;274:79. 3. Wolff J. The law of bone remodelling (Das Gesetz der Transformation der Knochen, Hirschwald). Berlin: Springer-Verlag; 1892. 4. Sartoris DJ, Resnick D. Current and innovative methods for noninvasive bone densitometry. Radiol Clin North Am 1990;28:257. 5. Mirsky EC, Einhorn TA. Current concepts review. Bone densitometry in orthopaedic practice. J Bone Joint Surg [Am] 1998;80-A:1687. 6. Compston JE, Cooper C, Kanis JA. Bone densitometry in clinical practice. BMJ 1995;310:1507. 7. Duboeuf F, Braillon P, Chapuy MC, et al. Bone mineral density of the hip measured with dualenergy x-ray absorptiometry in normal elderly women and in patients with hip fracture. Osteoporos Int 1991;1:242. 8. Wahner HW, Dunn WL, Brown ML, et al. Comparison of dual-energy x-ray absorptiometry and dual photon absorptiometry for bone mineral measurements of the lumbar spine. Mayo Clin Proc 1988; 63:1075. 9. Ho CP, Kim RW, Schaffler MB, et al. Accuracy of dual-energy radiographic absorptiometry of the lumbar spine: cadaver study. Radiology 1990;176:171. 10. Lundeen GA, Knecht SL, Vajda EG, et al. The contribution of cortical and cancellous bone on dual energy x-ray absorptiometry measurements in the female proximal femur. Osteoporos Int 2001; 12:192. 11. Cohen B, Rushton N. Accuracy of DEXA measurement of bone mineral density after total hip arthroplasty. J Bone Joint Surg [Br] 1995;77-B:479. 12. Trevisan C, Bigoni M, Denti M, et al. Bone assessment after total knee arthroplasty by dual-energy x-ray absorptiometry: analysis protocol and reproducibility. Calcif Tissue Int 1998;62:359.

13. Mazess RB, Collick B, Trempe J, et al. Performance evaluation of a dual energy x-ray bone densitometer. Calcif Tissue Int 1989;44:228. 14. Kilgus DJ, Shimaoka EE, Tipton JS, et al. Dualenergy x-ray absorptiometry measurement of bone mineral density around porous-coated cementless femoral implants. J Bone Joint Surg [Br] 1993; 75-B:279. 15. McCarthy CK, Steinberg GG, Agren M, et al. Quantifying bone loss from the proximal femur after total hip arthroplasty. J Bone Joint Surg [Br] 1991; 73-B:774. 16. Kastl S, Sommer T, Klein P, et al. Accuracy and precision of bone mineral density and bone mineral content in excised rat humeri using fan beam dualenergy X-ray absorptiometry. Bone 2002;30:243. 17. Nagy TR, Prince CW, Li J. Validation of peripheral dual-energy X-ray absorptiometry for the measurement of bone mineral in intact and excised long bones of rats. J Bone Miner Res 2001;16:1682. 18. Paniagua JG, Diaz-Curiel M, Gordo CDO, et al. Bone mass assessment in rats by dual energy x-ray absorptiometry. Br J Radiol 1998;71:754. 19. Kiratli BJ, Heiner JP, McBeath AA, et al. Determination of bone mineral density by dual x-ray absorptiometry in patients with uncemented total hip arthroplasty. J Orthop Res 1992;10:836. 20. Lochmuller EM, Miller P, Burklein D, et al. In situ femoral dual-energy X-ray absorptiometry related to ash weight, bone size and density, and its relationship with mechanical failure loads of the proximal femur. Osteoporos Int 2000;11:361. 21. Currey JD. Mechanical properties of bone tissues with greatly differing functions. J Biomech 1979;12:313. 22. Currey JD. Effects of differences in mineralization on the mechanical properties of bone. Philos Trans R Soc Lond (Biol) B 1984;304:509. 23. Currey JD. The mechanical consequences of variation in the mineral content of bone. J Biomech 1969;2:1. 24. Currey JD, Brear K, Zioupos P. The effects of ageing and changes in mineral content in degrading the toughness of human femora. J Biomech 1996; 29:257. 25. Engh CA, McGovern TF, Bobyn JD, et al. A quantitative evaluation of periprosthetic boneremodeling after cementless total hip arthroplasty. J Bone Joint Surg [Am] 1992;74-A:1009. 26. Sychterz CJ, Engh CA. The influence of clinical factors on periprosthetic bone remodeling. Clin Orthop 1996;322:285. 27. Skedros JG, Mason MW, Nelson MC. Evidence of structural and material adaptation to specific strain features in cortical bone. Anat Rec 1996;246:47. 28. Skedros JG, Su SC, Bloebaum RD. Biomechanical implications of mineral content and microstructural variations in cortical bone of horse, elk and sheep calcanei. Anat Rec 1997;249:297. 29. Bloebaum RD, Skedros JG, Vajda EG, et al. Determining mineral content variations in bone

622 The Journal of Arthroplasty Vol. 21 No. 4 June 2006 using backscattered electron imaging. Bone 1997; 20:485. 30. Streiner DL, Norman GR. Health measurement scales: a practical guide to their development and use. 2nd ed. New York: Oxford University Press; 1995. 31. Wright S. Adjusted p-values for simultaneous inference. Biometrics 1992;48:1005. 32. Sabo D, Reiter A, Simank HG, et al. Periprosthetic mineralization around cementless total hip endoprosthesis: longitudinal study and cross-sectional study on titanium threaded acetabular cup and

cementless spotorno stem with DEXA. Calcif Tissue Int 1998;62:177. 33. Engh CA, Bobyn JD, Glassman AH. Porous-coated hip replacement. The factors governing bone ingrowth, stress shielding, and clinical results. J Bone Joint Surg [Br] 1987;69-B:45. 34. Bousson V, Bergot C, Meunier A, et al. CT of the middiaphyseal femur: cortical bone mineral density and relation to porosity. Radiology 2000;217:179. 35. Currey JD. The effect of porosity and mineral content on the Young’s modulus of elasticity of compact bone. J Biomech 1988;21:131.