- Email: [email protected]
Cortical and total bone mineral content of the radius: accuracy of peripheral computed tomography
Bone Vol. 18, No. 5 May 1996:467472 ELSEVIER
Cortical and Total Bone Mineral Content of the Radius: Accuracy of Peripheral Computed Tomography O. L O U I S , 1 S. S O Y K E N S , e J. W I L L N E C K E R , 3 P. V A N D E N W I N K E L , 2 a n d M. O S T E A U X j
i Radiology Department and 2Cyclotron Department, Akademisch Ziekenhuis, Vrije Universiteit Brussel, Brussels, Belgium 3 Stratec Medizintechnik, Birkenfeld, Germany
its distal end. Subsequently, specimens matched with these two sites were embedded and submitted to two reference methods, namely neutron activation analysis and flame atomic absorption spectrometry. Our goal was to evaluate the accuracy of the Stratec pQCT device, at both a site of pure cortical bone and at the distal end used in clinical practice, where some amount of trabecular bone is still present.
The aim of the study was to evaluate the accuracy of an XCT 960 Stratec peripheral quantitative computed tomography (pQCT) device in assessing bone mineral content of the radius. We scanned 27 left forearm specimens excised from cadavers and focused on cortical bone mineral content (BMCc) at the junction of the middle and distal third and on total bone mineral content (UMCtot) at the distal end of the radius. Cylindrical specimens matched with those two sites were cut using a diamond circular saw, embedded in a polyester resin and subsequently submitted to two reference methods, nondestructive neutron activation analysis and flame atomic absorption spectrometry. Mineral contents measured by pQCT were closely correlated with those assessed by using the two reference methods, with correlation coefficients ranging from 0.862 to 0.960. The standard error of the estimate amounted 7-10% for the BMC c (junction of the middle and distal third), and 17-18% for the BMCto t (distal end). We conclude that pQCT is able to measure either cortical or total mineral content of the radius with a high degree of accuracy. (Bone 18:467-472; 1996)
Materials and Methods Materials This study was performed on 27 left forearm specimens excised from 27 cadavers obtained from the Anatomy Department of our School of Medicine. All cadavers had been preserved using an injection into the vessels of formalin solution. All cadavers were women (mean age 72 years, range 61-85). The cause of death was cerebrovascular insult (11 cases), heart failure (10 cases), and acute infection (6 cases). Malignancy, chronic infection, diabetes, or treatment known to interfere with bone metabolism were not present in any of the cases.
Key Words: Bone mineral content; Calcium content; Peripheral quantitative computed tomography.
Methods Peripheral quantitative computed tomography (pQCT). All scans were performed using a newly developed, specially built scanner (single energy pQCT; XCT 960, Stratec, Germany) with an X-ray tube (38.5 kV) as the source of radiation. Computed tomographic single slice measurements (2.5 mm thick) were made at two distinct sites. The first site was the junction of the middle third and the distal third as in previous studies by our group. 12'13 The second site was the distal end as currently evaluated in clinical practice.18 After completion of the scanning, two metal pins were screwed into the forearm specimen at the exact site of the two scanned slices. Trabecular (BMDt) , cortical (BMDc), and total (BMDtot) bone mineral density was assessed separately and expressed as milligrams hydroxyapatite of calcium per milliliter. The calibration attributed the attenuation coefficient of fat (0.21 cm -1) to a density of 0.0 mg/cm 3. The threshold used to define cortical bone was fixed at a linear attenuation coefficient of 0.93 cm -~ to minimize partial volume effects. The software used in the current study also calculated trabecular (At), cortical (At), and total (Atot) areas. Results were reported as cortical bone mineral content (BMCc) at the junction of the middle and the distal third and as total bone mineral content (BMCtot) at the distal site. BMCc was calculated as A c ×
Introduction Peripheral quantitative computed tomography (pQCT) has been available for many years to assess the bone mineral density of the radius. Various devices have been described. 6'15"16'2° In vitro and in vivo reproducibility has been evaluated 5 and reference values in a healthy population have been reported. 2"15 Several studies have shown that the method is able to assess the rate of bone loss with age and/or disease. 15.18.19 Furthermore, the mineral content of radius specimens assessed with pQCT has been found to correlate with compressive strength on biomechanical testing. 12 However, to the best of our knowledge, the accuracy of the method, i.e., its ability to assess the true calcium content, has never been validated. In the present investigation, we have studied 27 forearm specimens with an XCT 960 Stratec pQCT device. We scanned both the junction of the middle and distal third of the radius and
Address for correspondence and reprints: Dr. Olivia Louis, Department of Radiology, Akademisch Ziekenhuis, Vrije Universiteit Brussel, Laarbeeklaan 101, B-1090 Brussels, Belgium. © 1996 by Elsevier Science Inc. All rights reserved.
467
8756-3282/96/$15.00 Pll $8756-3282(96)00040-3
468
O. Louis et al. Accuracy of peripheral computed tomography
Bone Vol. 18, No. 5 May 1996:467~-72
BMD c x 2.5 mm (slice thickness); and BMCto ~as Atot x BMD~ot x 2.5 mm, both expressed as milligrams hydroxyapatite.
Preparation of Specimens The forearms were dissected after pQCT examination to obtain radius specimens. At each site (junction of middle and distal third, and distal end), 5-ram-thick slices were cut with a diamond circular saw, each slice including in its central part the 2.5-ramthick region studied on pQCT, located using the metal pin. Every effort was made to cut perpendicular to the long axis of the radius. In the specimen obtained at the junction of the middle and the distal third, the endosteal trabecular bone was removed by grinding with carborundum paper. 23 All specimens were embedded in a polyester resin. Subsequently, the thickness was reduced to 2.5 mm.
Flame Atomic Absorption Spectrometry The embedded samples were placed in a Mikrodismembrator iI (Van der Heyden, Brussels, Belgium) with steel pearls and liquid nitrogen, to be reduced to a fine powder. We added 5 mL of a mixture of oxidizing acids (HNO 3 1 vol. for 3 vol. HCIO4) to 100-300 mg of bone powder and subsequently the preparation was heated for 20 min at 100°C. The flame atomic spectrometer we used has been fully described in a previous study. ~ Briefly, the solution obtained after heating was transformed by the atomizer-burner into a fine mist. In the flame, the atoms present in the mist absorbed selected wavelengths of radiation from a suitable light source. Subsequently the detector transformed the transmitted light beam intensity into an electric current. The accuracy error of atomic absorption is considered to be 1-2%. 2~
Statistical Analysis Nondestructive Neutron Activation Analysis A Thetis reactor (School of Nuclear Sciences, Rijksuniversiteit Gent, Ghent, Belgium) was used. Each specimen was brought into a polyethylene vial that was then transferred with a pneumatical rabbit. The flux of the radiation was 2.6 x 1012 n cm-= sec -~. On elapse of 300 sec irradiation time, the capsule was returned automatically to the load port. After a cooling period of 120 sec, the samples were placed between two 5 in. x 5 in. Nal (TI) scintillation detections coupled to a 4.000 channel analyzer ($40 Camberra, USA). The measured 49Ca activity obtained after the 300 sec counting time was corrected for any 24Na spectral interference by dual counting and subsequent correction for decay. Calcium hydroxyapatite was used as a standard. The accuracy error of neutron activation analysis is estimated to amount to 50/0.24 550
'
500
,
Y:
Linear regression analysis was used to evaluate the relation between BMC~. (junction of the middle and distal third) and BMCt,,t (distal end) assessed with pQCT, on the one hand, and the corresponding contents obtained by neutron activation analysis (BMC~ NAA, BMCto t NAA) and by flame atomic absorption spectrometry (BMC c AAS, BMCto t AAS) on the other. The reported results include the correlation coefficient (r), the regression parameters (slope and intercept) and the standard error of the estimate (SEE).
Results The reported results involve 27 specimens for both the junction of the middle and distal third and for the distal end. All BMC
1.106"X + 5.90 r= 0.960
450,
400,
350,
300,
250
200
,
150,
100,
SO,
0, 0
50
100
150
200
250
300
350
400
450
BMCe NAA
Figure 1. Cortical bone mineral content assessed by pQCT at the junction of the middle and the distal third of the radius evaluated against that measured by nondestructive neutron activation analysis, both expressed in milligrams of calcium hydroxyapatite. The regression line is drawn with its 95% confidence interval.
Bone Vol. 18, No. 5 May 1996:467-472
O. Louis et al. Accuracy of peripheral computed tomography
469
500 '
Y= 0.973"X + 29.991 r= 0.916
450 -
'i
400
3SO.! 300 '
A A
Aj
-
•
200 .i
.~"~
150 --
1 O0
j.
•A
.J//"
50. 0
0
50
100
150
200 BMCe AAS
250
300
3SO
400
4S0
Figure 2. Cortical bone mineral content assessed by pQCT at the junction of the middle and distal third of the radius evaluated against that measured by flame atomic absorption spectrometry. Otherwise, same legend data as for Figure 1.
6O0
SS0,
Y= | . 0 3 1 ' X - 4.59| r= 0.865
•
S00 •
480
400
350
300,
250,
200 •
lS0
100
50,
0 0
I
I
I
I
50
100
I S0
200
I
I
250 300 B M C TOt N A A
I
I
I
I
I
350
400
450
500
EB0
Figure 3. Total bone mineral content assessed by pQCT at the distal end of the radius evaluated against that measured by nondestructive neutron activation analysis. Otherwise, same legend data as for Figure h
470
O. Louis et al. Accuracy of peripheral computed tomography
Bone Vol. 18, No. 5 May 1996:467-472
600,
5SO'
&
Y - 0.974"X + 9.85
500'
r-
0.862
~
450.
400 • •
&
350 •
300 •
2SO
2OO
1SO
~
too
SO 0
I S0
•"& I 1 O0
I 1 SO
I 200
I I 260 300 BMCTo T AAS
I 350
I 400
I 450
I SO0
I S50
Figure 4. Total bone mineral content assessed by pQCT at the distal end of the radius evaluated against that measured by flame atomic absorption spectrometry. Otherwise, same legend data as for Figure 1.
measurements a r e e x p r e s s e d in m i l l i g r a m s o f c a l c i u m hydroxyapatite. T h e two reference m e t h o d s s h o w e d close relation: B M C c A A S = 1.022 B M C c N A A + 3.41 m g , (p < 0.001); B M C t o t A A S = 1.025 B M C t o t N A A - 4.98 m g , (p < 0.001).
r = 0.942 r = 0.971
Discussion
T h e linear r e g r e s s i o n s o f p Q C T m e a s u r e m e n t s against the corr e s p o n d i n g contents obtained by the reference m e t h o d s are BMC c (p < 0.001), BMC c (p < 0.001), BMCt,,t (p < 0.001), BMCto t (p < 0.001).
= 1.106 B M C c N A A + 5 . 9 0 m g ,
T h e relation b e t w e e n the content a s s e s s e d u s i n g p Q C T , either at the j u n c t i o n o f the middle a n d distal third or at the distal end, a n d the c o r r e s p o n d i n g content obtained by N A A or A A S is illustrated in F i g u r e s 1, 2, 3, a n d 4. T h e standard errors o f the estimate are s h o w n in T a b l e 1. A s expected from r values, SEEs are h i g h e r for the distal end.
r = 0.960
= 0.973 B M C , , A A S + 2 9 . 9 9 rag, r = 0.916 = 1.031 B M C t o t N A A - 4.59 m g , r = 0.865 = 0.974 B M C t o t A A S + 9.85 m g , r = 0.862
N o n e o f the slopes o f these r e g r e s s i o n s w a s significantly different f r o m unity.
In the p r e s e n t s t u d y i n v o l v i n g 27 f o r e a r m s p e c i m e n s we f o u n d that the mineral content o f the radius can be accurately a s s e s s e d u s i n g p Q C T . Indeed, u s i n g n o n d e s t r u c t i v e neutron activation analysis and atomic absorption s p e c t r o m e t r y as the reference m e t h o d s , we d e m o n s t r a t e d close a g r e e m e n t b e t w e e n the mineral contents obtained by p Q C T and those a s s e s s e d by the reference m e t h o d s , with g o o d proportionality (slopes o f the r e g r e s s i o n lines 0 . 9 7 3 - 1 . 1 0 6 ) . Previous studies involving p Q C T h a v e foc u s e d on either reproducibility 15"2° or sensitivity for clinical use15,18,19 of mineral density m e a s u r e m e n t s . In contrast, we were unable to find any data in the literature dealing with the accuracy o f p Q C T , w h e r e a s s u c h data are available for other bone densit o m e t r y t e c h n i q u e s s u c h as d u a l - e n e r g y X - r a y a b s o r p t i o m -
Table 1. Accuracy of pQCT: standard errors of the estimate (SEEs) Junction of the middle and distal end BMC,JBMC c NAA BMC,JBMC,. AAS BMCc NAA/BMC,: AAS
18.1 (7.2%) 25.8 (10.3%) 20.2 (8.8%)
Distal end BMCtot/BMCtot NAA BMCtot/BMCto, AAS BMCtot/BMCtot AAS
50.6 (17.4%) 51.1 (17.6%) 21.5 (7.4%)
SEEs are expressed in milligrams of calcium hydroxyapatite, as well as percentages of the mean.
B o n e Vol. 18, No, 5 May 1996:467-472
O. Louis et al. A c c u r a c y o f peripheral c o m p u t e d t o m o g r a p h y
471
Table 2. A c c u r a c y error o f p Q C T : c o m p a r i s o n with other methods Spine DEXA B u r g e s s et al. 1 Gltier et al. 4 H o et al. 5 Laval-Jeantet et al. s L o u i s et al. l° L o u i s et a l ) I Present study
Spine Q C T (single energy) 10.7% 10.5%
Spine Q C T (dual energy)
pQCT
6.7% 6.4%
9% 14%
5.5%
13% 19% 7.2-17.6%
All accuracy errors are expressed as SEEs (%).
etry 5'1°'22 or spinal quantitative computed tomography) '4'8'it A comparison between the accuracy error reported in the present study about pQCT, expressed as a SEE, and that reported in previous cadaver studies involving spine DEXA or spine QCT, is shown in Table 2. In this study, we evaluated both the junction of the middle and distal third of the radius and its distal end. The former site gave us the opportunity to study pure cortical bone, the remaining trabecular bone being removed by grinding with carborundum paper. Furthermore, this site has increased vulnerability to fractures, as judged by recent data about area moment of inertia 7 and was used by us in previous studies dealing with compressive strength 12 and evaluation of cortical thickness) 3 The latter site is that usually evaluated in clinical practice when using pQCT 2'~5 and includes trabecular as well as cortical bone. Our results suggest that the Stratec pQCT device may be less accurate in studying the ultradistal site than it is for the junction of the middle and distal third, as judged from somewhat weaker correlation coefficients and higher SEEs. This lower accuracy at the ultradistal site may be due to the trabecular component of B M f t o t assessed at this site. Indeed, the marrow contains a variable proportion of fat influencing measurement of the mineral content of trabecular bone when using quantitative computed tomography. 9'~4 As an alternative explanation, the difference in accuracy between the two sites may be related to our experimental procedure. Indeed, the distal end is a thin anatomic site, containing friable trabecular bone, and the loss of some bone material during the cutting procedure may have induced a given degree of inaccuracy. 3 In most previous studies involved with the accuracy of bone densitometry methods, the gold standard was ashing, the result being expressed as an ash weight or ash density. 4'5'8 In the present study we first used neutron activation analysis. This method has the advantage of being nondestructive and its validity has been established by others 17 and ourselves, t° We used a Thetis reactor with a high irradiation flux and a short irradiation time, particularly well suited for small specimens such as our radius cylindrical samplesY Our second reference method was flame atomic absorption spectrometry, which is a calcium specific analysis considered as the best available gold standard. 3 Where the junction of the middle and distal third is concerned, the SEE reflecting the inaccuracy of pQCT in assessing the cortical content was as low as <--10%, and hence, compared very well with that reflecting the inaccuracy of either reference method.
Acknowledgments: The
authors t h a n k P. C o v e n s for d r a w i n g the figures a n d W. Rijsselaere for secretarial assistance.
References 1. Burgess, A., Colborne, B., and Zoffmann, E. Vertebral trabecular bone: Comparison of single and dual energy CT measurements with chemical analysis. J Comput Assist Tomogr 11:506-515; 1987. 2. Butz, S., W0ster, C., Scheidt-Nave, C., G6tz, M., and Liegler, R. Forearm BMD as measured by peripheral quantitative computed tomography (pQCT) in a German reference population. Osteopor Int 4:179-184; 1994. 3. Cann, C.E. Quantitative CT for determination of bone mineral density: A review. Radiology 166:509-512; 1988. 4. G10er, C., Reiser, U., Davis, C., Rutt, B., and Genant, H. K. Vertebral mineral determination of quantitative computed tomography (QCT): Accuracy of single and dual energy measurements. J Comput Assist Tomogr 12:242-258; 1988. 5. Ho, C., Kim, R., Schaffer, M., and Sartoris, D. Accuracy of dual energy radiographic absorptiometry of the lumbar spine: Cadaver study. Radiology 176:171-173; 1990. 6. Hosie, C. J. Measurement of trabecular bone mineral density in the distal radius by two gamma-ray computed tomography scanners. Physiol Meas 14:269-276; 1993. 7. Hsu, E., Patwardhan, A.G., Meade, K., Light, T., and Martin, W. Crosssectional geometrical properties and bone mineral contents of the human radius and ulna. J Biomech 26:1307-1318; 1993. 8. Laval-Jeantet, A. M., Cann, C., Roger, B., and Dallant, P. A post-processing dual energy technique for vertebral CT densitometry. J Comput Assist Tomogr 8:1164-1167; 1984. 9. Laval-.leantet, A.M., Roger, B., Boysse, S., Bergot, C., and Mazess, R.B. Influence of vertebral fat content on quantitative CT density. Radiology 159:463-466; 1986. 10. Louis, O., Van Den Winkel, P., Covens, P., Schoutens, A., and Osteaux, M. Dual energy X-ray absorptiometry~of lumbar vertebrae: Relative contribution of body and posterior elements and accuracy in relation with neutron activation analysis. Bone 13:317-320; 1992. 11. Louis, O., Van Den Winkel, P., Covens, P., Schoutens, A., and Osteaux, M. Mineral content of vertebral trabecular bone: Accuracy of dual energy quantitative computed tomography evaluated against neutron activation analysis and flame atomic absorption spectrometry. Bone 15:35-39; 1994. 12. Louis, O., Boulpaep, F., Willnecker, J., Van Den Winkel, P., and Osteaux, M. Cortical mineral content of the radius assessed by peripheral QCT predicts compressive strength on biomechanical testing. Bone 16:375-379; 1995. 13. Louis, O., Willnecker, J., Soykens, S., Van Den Winkel, P., and Osteaux, M. Cortical thickness assessed by peripheral quantitative computed tomography: Accuracy evaluated in radius specimens. Osteopor Int 5:466-449; 1995. 14. Mazess, R. B. Errors in measuring trabecular bone by computed tomography due to marrow and bone composition. Calcif Tissue lnt 35:148-152; 1983. 15. MOiler, A., RUegsegger, E., and Riiegsegger, P. Peripheral QCT: A low risk procedure to identify women predisposed to osteoporosis. Phys Med Biol 34:741-749; 1989. 16 ,erton, T. and Wheeler, G. D. Bone mass measurements in the distal forearm using dual energy X-ray absorptiometry and "/-ray computed tomography: A longitudinal, in vivo comparative study. J Bone Miner Res 7:371-375; 1992. 17. Parks, N. L, Jee, W. S., Dell, R. B., and Miller, G. E. Assessment of cortical and trabecular bone distribution in the beagle skeleton by neutron activation analysis. Anat Rec 215:230-250; 1986. 18. Rico, H., Diaz, H., Duran, C., Villa, L., and Penela, S. Quantitative peripheral computed tomodensitometry study of cortical and trabecular bone mass in relation with menopause. Maturitas 18:183-189; 1994.
472
O. Louis et al. Accuracy of peripheral computed tomography
19. Rtiegsegger, P., Durand, P., and Dambacher, M. A. Differential effects of age ing and disease in trabecular and compact bone density of the radius. Bone 12:99-105; 1991. 20. Schneider, P. and B6rner, W. Periphere quantitative computertmnographie zur Knochenmineralmessung mit einem neuen speziallen QCT-Scanner. Nuklearmediciner 11:145-152; 1988. 21. Skoog, D. Atomic spectroscopy based upon flame and electrothermal atomization. Principles of instrumental analysis. New York: 1985. 22. Svendsen, O., Hassager, C., Skodt, V., and Christiansen, C. Impact of sotl tissue on in vivo accuracy of bone numeral measurements in the spine, hip and li~rearm: A human cadaver study. J Bone Miner Res 10:860-873; 1995. 23. Thompson, D. Age changes in bone mineralization, cortical thickness and aversian canal area. Calcif Tissue lnt 31:5 11; 1980.
Bone Vol. 18, No. 5 May 1996:467~-72
24. Tothill, P. Methods of bone mineral measurement. Phys Med Biol 34:543 572; 1989. 25. Van Den Winkel, P. De bepaling van sporenelementen in biologisch materiaal door neutron activeringsanalyse. Verhandelingen van de Koninklijke Vlaamse Academie your Wetenschappen, Letteren en Schone Kunsten van BelgiE, Klasse der Wetenschappen, Jaargang XXXII, No. 114, Brussels; 1970.
Date Received: October 31, 1995 Date Revised: January 25, 1996 Date Accepted: January 25~ 1996
Cortical and Total Bone Mineral Content of the Radius: Accuracy of Peripheral Computed Tomography O. L O U I S , 1 S. S O Y K E N S , e J. W I L L N E C K E R , 3 P. V A N D E N W I N K E L , 2 a n d M. O S T E A U X j
i Radiology Department and 2Cyclotron Department, Akademisch Ziekenhuis, Vrije Universiteit Brussel, Brussels, Belgium 3 Stratec Medizintechnik, Birkenfeld, Germany
its distal end. Subsequently, specimens matched with these two sites were embedded and submitted to two reference methods, namely neutron activation analysis and flame atomic absorption spectrometry. Our goal was to evaluate the accuracy of the Stratec pQCT device, at both a site of pure cortical bone and at the distal end used in clinical practice, where some amount of trabecular bone is still present.
The aim of the study was to evaluate the accuracy of an XCT 960 Stratec peripheral quantitative computed tomography (pQCT) device in assessing bone mineral content of the radius. We scanned 27 left forearm specimens excised from cadavers and focused on cortical bone mineral content (BMCc) at the junction of the middle and distal third and on total bone mineral content (UMCtot) at the distal end of the radius. Cylindrical specimens matched with those two sites were cut using a diamond circular saw, embedded in a polyester resin and subsequently submitted to two reference methods, nondestructive neutron activation analysis and flame atomic absorption spectrometry. Mineral contents measured by pQCT were closely correlated with those assessed by using the two reference methods, with correlation coefficients ranging from 0.862 to 0.960. The standard error of the estimate amounted 7-10% for the BMC c (junction of the middle and distal third), and 17-18% for the BMCto t (distal end). We conclude that pQCT is able to measure either cortical or total mineral content of the radius with a high degree of accuracy. (Bone 18:467-472; 1996)
Materials and Methods Materials This study was performed on 27 left forearm specimens excised from 27 cadavers obtained from the Anatomy Department of our School of Medicine. All cadavers had been preserved using an injection into the vessels of formalin solution. All cadavers were women (mean age 72 years, range 61-85). The cause of death was cerebrovascular insult (11 cases), heart failure (10 cases), and acute infection (6 cases). Malignancy, chronic infection, diabetes, or treatment known to interfere with bone metabolism were not present in any of the cases.
Key Words: Bone mineral content; Calcium content; Peripheral quantitative computed tomography.
Methods Peripheral quantitative computed tomography (pQCT). All scans were performed using a newly developed, specially built scanner (single energy pQCT; XCT 960, Stratec, Germany) with an X-ray tube (38.5 kV) as the source of radiation. Computed tomographic single slice measurements (2.5 mm thick) were made at two distinct sites. The first site was the junction of the middle third and the distal third as in previous studies by our group. 12'13 The second site was the distal end as currently evaluated in clinical practice.18 After completion of the scanning, two metal pins were screwed into the forearm specimen at the exact site of the two scanned slices. Trabecular (BMDt) , cortical (BMDc), and total (BMDtot) bone mineral density was assessed separately and expressed as milligrams hydroxyapatite of calcium per milliliter. The calibration attributed the attenuation coefficient of fat (0.21 cm -1) to a density of 0.0 mg/cm 3. The threshold used to define cortical bone was fixed at a linear attenuation coefficient of 0.93 cm -~ to minimize partial volume effects. The software used in the current study also calculated trabecular (At), cortical (At), and total (Atot) areas. Results were reported as cortical bone mineral content (BMCc) at the junction of the middle and the distal third and as total bone mineral content (BMCtot) at the distal site. BMCc was calculated as A c ×
Introduction Peripheral quantitative computed tomography (pQCT) has been available for many years to assess the bone mineral density of the radius. Various devices have been described. 6'15"16'2° In vitro and in vivo reproducibility has been evaluated 5 and reference values in a healthy population have been reported. 2"15 Several studies have shown that the method is able to assess the rate of bone loss with age and/or disease. 15.18.19 Furthermore, the mineral content of radius specimens assessed with pQCT has been found to correlate with compressive strength on biomechanical testing. 12 However, to the best of our knowledge, the accuracy of the method, i.e., its ability to assess the true calcium content, has never been validated. In the present investigation, we have studied 27 forearm specimens with an XCT 960 Stratec pQCT device. We scanned both the junction of the middle and distal third of the radius and
Address for correspondence and reprints: Dr. Olivia Louis, Department of Radiology, Akademisch Ziekenhuis, Vrije Universiteit Brussel, Laarbeeklaan 101, B-1090 Brussels, Belgium. © 1996 by Elsevier Science Inc. All rights reserved.
467
8756-3282/96/$15.00 Pll $8756-3282(96)00040-3
468
O. Louis et al. Accuracy of peripheral computed tomography
Bone Vol. 18, No. 5 May 1996:467~-72
BMD c x 2.5 mm (slice thickness); and BMCto ~as Atot x BMD~ot x 2.5 mm, both expressed as milligrams hydroxyapatite.
Preparation of Specimens The forearms were dissected after pQCT examination to obtain radius specimens. At each site (junction of middle and distal third, and distal end), 5-ram-thick slices were cut with a diamond circular saw, each slice including in its central part the 2.5-ramthick region studied on pQCT, located using the metal pin. Every effort was made to cut perpendicular to the long axis of the radius. In the specimen obtained at the junction of the middle and the distal third, the endosteal trabecular bone was removed by grinding with carborundum paper. 23 All specimens were embedded in a polyester resin. Subsequently, the thickness was reduced to 2.5 mm.
Flame Atomic Absorption Spectrometry The embedded samples were placed in a Mikrodismembrator iI (Van der Heyden, Brussels, Belgium) with steel pearls and liquid nitrogen, to be reduced to a fine powder. We added 5 mL of a mixture of oxidizing acids (HNO 3 1 vol. for 3 vol. HCIO4) to 100-300 mg of bone powder and subsequently the preparation was heated for 20 min at 100°C. The flame atomic spectrometer we used has been fully described in a previous study. ~ Briefly, the solution obtained after heating was transformed by the atomizer-burner into a fine mist. In the flame, the atoms present in the mist absorbed selected wavelengths of radiation from a suitable light source. Subsequently the detector transformed the transmitted light beam intensity into an electric current. The accuracy error of atomic absorption is considered to be 1-2%. 2~
Statistical Analysis Nondestructive Neutron Activation Analysis A Thetis reactor (School of Nuclear Sciences, Rijksuniversiteit Gent, Ghent, Belgium) was used. Each specimen was brought into a polyethylene vial that was then transferred with a pneumatical rabbit. The flux of the radiation was 2.6 x 1012 n cm-= sec -~. On elapse of 300 sec irradiation time, the capsule was returned automatically to the load port. After a cooling period of 120 sec, the samples were placed between two 5 in. x 5 in. Nal (TI) scintillation detections coupled to a 4.000 channel analyzer ($40 Camberra, USA). The measured 49Ca activity obtained after the 300 sec counting time was corrected for any 24Na spectral interference by dual counting and subsequent correction for decay. Calcium hydroxyapatite was used as a standard. The accuracy error of neutron activation analysis is estimated to amount to 50/0.24 550
'
500
,
Y:
Linear regression analysis was used to evaluate the relation between BMC~. (junction of the middle and distal third) and BMCt,,t (distal end) assessed with pQCT, on the one hand, and the corresponding contents obtained by neutron activation analysis (BMC~ NAA, BMCto t NAA) and by flame atomic absorption spectrometry (BMC c AAS, BMCto t AAS) on the other. The reported results include the correlation coefficient (r), the regression parameters (slope and intercept) and the standard error of the estimate (SEE).
Results The reported results involve 27 specimens for both the junction of the middle and distal third and for the distal end. All BMC
1.106"X + 5.90 r= 0.960
450,
400,
350,
300,
250
200
,
150,
100,
SO,
0, 0
50
100
150
200
250
300
350
400
450
BMCe NAA
Figure 1. Cortical bone mineral content assessed by pQCT at the junction of the middle and the distal third of the radius evaluated against that measured by nondestructive neutron activation analysis, both expressed in milligrams of calcium hydroxyapatite. The regression line is drawn with its 95% confidence interval.
Bone Vol. 18, No. 5 May 1996:467-472
O. Louis et al. Accuracy of peripheral computed tomography
469
500 '
Y= 0.973"X + 29.991 r= 0.916
450 -
'i
400
3SO.! 300 '
A A
Aj
-
•
200 .i
.~"~
150 --
1 O0
j.
•A
.J//"
50. 0
0
50
100
150
200 BMCe AAS
250
300
3SO
400
4S0
Figure 2. Cortical bone mineral content assessed by pQCT at the junction of the middle and distal third of the radius evaluated against that measured by flame atomic absorption spectrometry. Otherwise, same legend data as for Figure 1.
6O0
SS0,
Y= | . 0 3 1 ' X - 4.59| r= 0.865
•
S00 •
480
400
350
300,
250,
200 •
lS0
100
50,
0 0
I
I
I
I
50
100
I S0
200
I
I
250 300 B M C TOt N A A
I
I
I
I
I
350
400
450
500
EB0
Figure 3. Total bone mineral content assessed by pQCT at the distal end of the radius evaluated against that measured by nondestructive neutron activation analysis. Otherwise, same legend data as for Figure h
470
O. Louis et al. Accuracy of peripheral computed tomography
Bone Vol. 18, No. 5 May 1996:467-472
600,
5SO'
&
Y - 0.974"X + 9.85
500'
r-
0.862
~
450.
400 • •
&
350 •
300 •
2SO
2OO
1SO
~
too
SO 0
I S0
•"& I 1 O0
I 1 SO
I 200
I I 260 300 BMCTo T AAS
I 350
I 400
I 450
I SO0
I S50
Figure 4. Total bone mineral content assessed by pQCT at the distal end of the radius evaluated against that measured by flame atomic absorption spectrometry. Otherwise, same legend data as for Figure 1.
measurements a r e e x p r e s s e d in m i l l i g r a m s o f c a l c i u m hydroxyapatite. T h e two reference m e t h o d s s h o w e d close relation: B M C c A A S = 1.022 B M C c N A A + 3.41 m g , (p < 0.001); B M C t o t A A S = 1.025 B M C t o t N A A - 4.98 m g , (p < 0.001).
r = 0.942 r = 0.971
Discussion
T h e linear r e g r e s s i o n s o f p Q C T m e a s u r e m e n t s against the corr e s p o n d i n g contents obtained by the reference m e t h o d s are BMC c (p < 0.001), BMC c (p < 0.001), BMCt,,t (p < 0.001), BMCto t (p < 0.001).
= 1.106 B M C c N A A + 5 . 9 0 m g ,
T h e relation b e t w e e n the content a s s e s s e d u s i n g p Q C T , either at the j u n c t i o n o f the middle a n d distal third or at the distal end, a n d the c o r r e s p o n d i n g content obtained by N A A or A A S is illustrated in F i g u r e s 1, 2, 3, a n d 4. T h e standard errors o f the estimate are s h o w n in T a b l e 1. A s expected from r values, SEEs are h i g h e r for the distal end.
r = 0.960
= 0.973 B M C , , A A S + 2 9 . 9 9 rag, r = 0.916 = 1.031 B M C t o t N A A - 4.59 m g , r = 0.865 = 0.974 B M C t o t A A S + 9.85 m g , r = 0.862
N o n e o f the slopes o f these r e g r e s s i o n s w a s significantly different f r o m unity.
In the p r e s e n t s t u d y i n v o l v i n g 27 f o r e a r m s p e c i m e n s we f o u n d that the mineral content o f the radius can be accurately a s s e s s e d u s i n g p Q C T . Indeed, u s i n g n o n d e s t r u c t i v e neutron activation analysis and atomic absorption s p e c t r o m e t r y as the reference m e t h o d s , we d e m o n s t r a t e d close a g r e e m e n t b e t w e e n the mineral contents obtained by p Q C T and those a s s e s s e d by the reference m e t h o d s , with g o o d proportionality (slopes o f the r e g r e s s i o n lines 0 . 9 7 3 - 1 . 1 0 6 ) . Previous studies involving p Q C T h a v e foc u s e d on either reproducibility 15"2° or sensitivity for clinical use15,18,19 of mineral density m e a s u r e m e n t s . In contrast, we were unable to find any data in the literature dealing with the accuracy o f p Q C T , w h e r e a s s u c h data are available for other bone densit o m e t r y t e c h n i q u e s s u c h as d u a l - e n e r g y X - r a y a b s o r p t i o m -
Table 1. Accuracy of pQCT: standard errors of the estimate (SEEs) Junction of the middle and distal end BMC,JBMC c NAA BMC,JBMC,. AAS BMCc NAA/BMC,: AAS
18.1 (7.2%) 25.8 (10.3%) 20.2 (8.8%)
Distal end BMCtot/BMCtot NAA BMCtot/BMCto, AAS BMCtot/BMCtot AAS
50.6 (17.4%) 51.1 (17.6%) 21.5 (7.4%)
SEEs are expressed in milligrams of calcium hydroxyapatite, as well as percentages of the mean.
B o n e Vol. 18, No, 5 May 1996:467-472
O. Louis et al. A c c u r a c y o f peripheral c o m p u t e d t o m o g r a p h y
471
Table 2. A c c u r a c y error o f p Q C T : c o m p a r i s o n with other methods Spine DEXA B u r g e s s et al. 1 Gltier et al. 4 H o et al. 5 Laval-Jeantet et al. s L o u i s et al. l° L o u i s et a l ) I Present study
Spine Q C T (single energy) 10.7% 10.5%
Spine Q C T (dual energy)
pQCT
6.7% 6.4%
9% 14%
5.5%
13% 19% 7.2-17.6%
All accuracy errors are expressed as SEEs (%).
etry 5'1°'22 or spinal quantitative computed tomography) '4'8'it A comparison between the accuracy error reported in the present study about pQCT, expressed as a SEE, and that reported in previous cadaver studies involving spine DEXA or spine QCT, is shown in Table 2. In this study, we evaluated both the junction of the middle and distal third of the radius and its distal end. The former site gave us the opportunity to study pure cortical bone, the remaining trabecular bone being removed by grinding with carborundum paper. Furthermore, this site has increased vulnerability to fractures, as judged by recent data about area moment of inertia 7 and was used by us in previous studies dealing with compressive strength 12 and evaluation of cortical thickness) 3 The latter site is that usually evaluated in clinical practice when using pQCT 2'~5 and includes trabecular as well as cortical bone. Our results suggest that the Stratec pQCT device may be less accurate in studying the ultradistal site than it is for the junction of the middle and distal third, as judged from somewhat weaker correlation coefficients and higher SEEs. This lower accuracy at the ultradistal site may be due to the trabecular component of B M f t o t assessed at this site. Indeed, the marrow contains a variable proportion of fat influencing measurement of the mineral content of trabecular bone when using quantitative computed tomography. 9'~4 As an alternative explanation, the difference in accuracy between the two sites may be related to our experimental procedure. Indeed, the distal end is a thin anatomic site, containing friable trabecular bone, and the loss of some bone material during the cutting procedure may have induced a given degree of inaccuracy. 3 In most previous studies involved with the accuracy of bone densitometry methods, the gold standard was ashing, the result being expressed as an ash weight or ash density. 4'5'8 In the present study we first used neutron activation analysis. This method has the advantage of being nondestructive and its validity has been established by others 17 and ourselves, t° We used a Thetis reactor with a high irradiation flux and a short irradiation time, particularly well suited for small specimens such as our radius cylindrical samplesY Our second reference method was flame atomic absorption spectrometry, which is a calcium specific analysis considered as the best available gold standard. 3 Where the junction of the middle and distal third is concerned, the SEE reflecting the inaccuracy of pQCT in assessing the cortical content was as low as <--10%, and hence, compared very well with that reflecting the inaccuracy of either reference method.
Acknowledgments: The
authors t h a n k P. C o v e n s for d r a w i n g the figures a n d W. Rijsselaere for secretarial assistance.
References 1. Burgess, A., Colborne, B., and Zoffmann, E. Vertebral trabecular bone: Comparison of single and dual energy CT measurements with chemical analysis. J Comput Assist Tomogr 11:506-515; 1987. 2. Butz, S., W0ster, C., Scheidt-Nave, C., G6tz, M., and Liegler, R. Forearm BMD as measured by peripheral quantitative computed tomography (pQCT) in a German reference population. Osteopor Int 4:179-184; 1994. 3. Cann, C.E. Quantitative CT for determination of bone mineral density: A review. Radiology 166:509-512; 1988. 4. G10er, C., Reiser, U., Davis, C., Rutt, B., and Genant, H. K. Vertebral mineral determination of quantitative computed tomography (QCT): Accuracy of single and dual energy measurements. J Comput Assist Tomogr 12:242-258; 1988. 5. Ho, C., Kim, R., Schaffer, M., and Sartoris, D. Accuracy of dual energy radiographic absorptiometry of the lumbar spine: Cadaver study. Radiology 176:171-173; 1990. 6. Hosie, C. J. Measurement of trabecular bone mineral density in the distal radius by two gamma-ray computed tomography scanners. Physiol Meas 14:269-276; 1993. 7. Hsu, E., Patwardhan, A.G., Meade, K., Light, T., and Martin, W. Crosssectional geometrical properties and bone mineral contents of the human radius and ulna. J Biomech 26:1307-1318; 1993. 8. Laval-Jeantet, A. M., Cann, C., Roger, B., and Dallant, P. A post-processing dual energy technique for vertebral CT densitometry. J Comput Assist Tomogr 8:1164-1167; 1984. 9. Laval-.leantet, A.M., Roger, B., Boysse, S., Bergot, C., and Mazess, R.B. Influence of vertebral fat content on quantitative CT density. Radiology 159:463-466; 1986. 10. Louis, O., Van Den Winkel, P., Covens, P., Schoutens, A., and Osteaux, M. Dual energy X-ray absorptiometry~of lumbar vertebrae: Relative contribution of body and posterior elements and accuracy in relation with neutron activation analysis. Bone 13:317-320; 1992. 11. Louis, O., Van Den Winkel, P., Covens, P., Schoutens, A., and Osteaux, M. Mineral content of vertebral trabecular bone: Accuracy of dual energy quantitative computed tomography evaluated against neutron activation analysis and flame atomic absorption spectrometry. Bone 15:35-39; 1994. 12. Louis, O., Boulpaep, F., Willnecker, J., Van Den Winkel, P., and Osteaux, M. Cortical mineral content of the radius assessed by peripheral QCT predicts compressive strength on biomechanical testing. Bone 16:375-379; 1995. 13. Louis, O., Willnecker, J., Soykens, S., Van Den Winkel, P., and Osteaux, M. Cortical thickness assessed by peripheral quantitative computed tomography: Accuracy evaluated in radius specimens. Osteopor Int 5:466-449; 1995. 14. Mazess, R. B. Errors in measuring trabecular bone by computed tomography due to marrow and bone composition. Calcif Tissue lnt 35:148-152; 1983. 15. MOiler, A., RUegsegger, E., and Riiegsegger, P. Peripheral QCT: A low risk procedure to identify women predisposed to osteoporosis. Phys Med Biol 34:741-749; 1989. 16 ,erton, T. and Wheeler, G. D. Bone mass measurements in the distal forearm using dual energy X-ray absorptiometry and "/-ray computed tomography: A longitudinal, in vivo comparative study. J Bone Miner Res 7:371-375; 1992. 17. Parks, N. L, Jee, W. S., Dell, R. B., and Miller, G. E. Assessment of cortical and trabecular bone distribution in the beagle skeleton by neutron activation analysis. Anat Rec 215:230-250; 1986. 18. Rico, H., Diaz, H., Duran, C., Villa, L., and Penela, S. Quantitative peripheral computed tomodensitometry study of cortical and trabecular bone mass in relation with menopause. Maturitas 18:183-189; 1994.
472
O. Louis et al. Accuracy of peripheral computed tomography
19. Rtiegsegger, P., Durand, P., and Dambacher, M. A. Differential effects of age ing and disease in trabecular and compact bone density of the radius. Bone 12:99-105; 1991. 20. Schneider, P. and B6rner, W. Periphere quantitative computertmnographie zur Knochenmineralmessung mit einem neuen speziallen QCT-Scanner. Nuklearmediciner 11:145-152; 1988. 21. Skoog, D. Atomic spectroscopy based upon flame and electrothermal atomization. Principles of instrumental analysis. New York: 1985. 22. Svendsen, O., Hassager, C., Skodt, V., and Christiansen, C. Impact of sotl tissue on in vivo accuracy of bone numeral measurements in the spine, hip and li~rearm: A human cadaver study. J Bone Miner Res 10:860-873; 1995. 23. Thompson, D. Age changes in bone mineralization, cortical thickness and aversian canal area. Calcif Tissue lnt 31:5 11; 1980.
Bone Vol. 18, No. 5 May 1996:467~-72
24. Tothill, P. Methods of bone mineral measurement. Phys Med Biol 34:543 572; 1989. 25. Van Den Winkel, P. De bepaling van sporenelementen in biologisch materiaal door neutron activeringsanalyse. Verhandelingen van de Koninklijke Vlaamse Academie your Wetenschappen, Letteren en Schone Kunsten van BelgiE, Klasse der Wetenschappen, Jaargang XXXII, No. 114, Brussels; 1970.
Date Received: October 31, 1995 Date Revised: January 25, 1996 Date Accepted: January 25~ 1996