Technology and Health Care, 1 (1993) 127-131 0928-7329/93/$06.00 © 1993 Elsevier Science Publishers B.V. All rights reserved
127
Multicentre European COMAC-BME study on the standardisation of bone densitometry procedures J. Dequeker, J. Reeve, J. Pearson, J. Bright, D. Felsenberg, W. Kalender, C. Langton, A.M. Laval-Jeantet, P. Riiegsegger, G. Van der Perre and the COMAC-BME Quantitative Assessment of Osteoporosis Study Group * (Received 27 May, 1993, Accepted 2 July 1993)
Abstract 26 European centres participated in a concerted research action Biomedical Engineering: Quantitative Assessment of Osteoporosis. With a newly designed European spine and forearm phantom, the stability, accuracy, precision of dual energy absorption (DXA) and quantitative computer tomography (QCT) densitometry machines have been evaluated. Marked and clinically significant differences were found between brands and between techniques. Cross-calibration formulae have been made and normative data evaluated for different regions (spine, femoral neck, femoral trochanteric and forearm). A general fit for all data obtained from different machines was established. The cross-calibration formulae will allow a sensitivity analysis to assist the choice of equipment for clinical management of different categories of patients with bone disease. The present results obtained with an internationally accepted European spine and forearm phantom can now serve to stimulate the manufacturers to improve the comparability of bone measurements between machines. Key words: Osteoporosis; Bone densitometry; QCT; DXA
1. Introduction
Osteoporosis is one of the major health problems of our ageing population. Treatment of a clinically proven osteoporosis (with fractures) is not satisfactory. What is needed is an early diagnosis (in the sense of fracture risk prediction) and the ability to monitor bone response to therapy on
Correspondence to: Prof. Dr. J. Dequeker, Arthritis and Metabolic Bone Disease Research Unit, K.U. Leuven, B-3212 Pellenberg, Belgium.
* CO MAC BME Quantitative Assessment of Osteoporosis Participating centres: Adams J.E., Manchester; Birkenhiiger J.c., Rotterdam; Bonjour J.P., Geneva; Braillon P., Lyon; Diaz Curiel M., Madrid; Felsenberg D., Berlin; Fischer M., Kassel; Galan F., Sevilla; Gennari C., Siena; Geusens P., Leuven; Hemmingsson A, Uppsala; Hyldstrup L., Hvidovre; Jaeger Ph., Bern; Jonson R., Goteborg; Kalef-Ezra J., Ioannina; Kotzki P.O., Montpellier; Kroger H., Kuopio; LavalJeantet A.M., Paris; Lips P., Amsterdam; Osteaux M., Brussels; Reeve J., Harrow; Reid D., Aberdeen; Reiners c., Essen; Ribot c., Toulouse; Riiegsegger P., Ziirich; Schneider P., Wiirzburg. Project Management Group: Dequeker J. (Project Leader), Felsenberg D., Kalender W., Langton c., Laval-Jeantet AM., Reeve J., Riiegsegger P., Van der Perre G.
128
an individual basis. Biomedical engineering has to provide the means so that a patient at risk can be detected in time, and a preventive therapy can be started. Prevention of osteoporotic fractures would improve life quality of 10-20% of the population older than 65 years and would have a considerable influence on the exploding health care cost. In the past, considerable disagreement on measurement and treatment of osteoporosis have done damage to the acceptance and comparability of the respective studies. One of the primary goals of the EEC supported Concerted Research Action on Quantitative Assessment of Osteoporosis is to standardize and unify measurements, to enable the intercomparability of the various measuring procedures and to provide a standardized test bed for new procedures. The development of new methods and procedures will require continuous validation. Measurements of bone mass can now be made precisely. However, measurements made using different manufacturers hardware and software are not directly comparable. The EEC supported concerted action was started in 1990 to stimulate and coordinate the research of this particular problem. The aims of this study were to standardize measurements, to establish reference ranges for normal subjects and to compare the measurement techniques for their power to discriminate between people with spinal osteoporosis, thyroid excess, hip fracture, hyperparathyroidism and long-term corticosteroid treatment. 2. Experimental procedures The measurements were standardized by the use of reproducible operating procedures and by calibration using two newly designed semi-anthropomorphic phantoms. Phantoms were designed for the spine (ESP) and the forearm (EFP) [1,3]. The ESP and the EFP do not contain real muscle and bone tissue, no bone marrow is added. Both phantoms had three measurement sites with specified densities 0.5, 1.0 and 1.5 g/cm 2 or 50, 100 and 200 mg/cm 3 . Each centre was asked to measure the phantoms once per day for the first week and every
1. Dequeker et al. / Technology and Health Care 1 (1993) 127 -131
week the rafter. Centres measured normal subjects and patients at the same time as the phantom measurements. Each subject was measured at least twice on different machines and/or measurement sites. 22 centres participated in the study, 21 centres measured the ESP and 9 the EFP on a variety machines: quantitative computer tomography (QCT), dual X-ray energy absorptiometry (DXA) and single energy photon absorptiometry (SPA). 3. Results In a standard way 977 phantom forms, 6740 subject forms, 2471 questionnaires and 562 laboratory forms were submitted for analysis. The data for the first week measurements and the subsequent weekly measurements of the European spine and forearm phantoms were combined for each machine. Six machines, one DXA spine, three QCT spine, one DXA forearm and one SPA forearm showed evidence of a significant (P < 0.01) drift ranging from -5.3% to +3.9% over a period of 3 months. Most of the machines change less than 3%. Important differences in accuracy and precision were disclosed using ESP and EFP phantoms between different machines and between different brands of machines. Precision and accuracy can be defined as follows: precision is the ability of a system to obtain the same results in repeated measurements of the same subject or object; accuracy is the ability of a system to obtain measured results, that on the average reflect the true values for the measured parameters. The results of the ESP measurements from one centre (Table 1) show that the DXA machine 11 gives lower density values at the specified 1.5 g/cm2 site, and at the other sites to a lesser extent. The DXA I machine measures accurately, but with less precision. The results from another centre, show that a QCT machine gives slightly higher density values at each specified density site, but measures each site with similar precision. The machines not only differ in accuracy in relation to specified densities, but also in precision and this was also dependent on the level of bone density measured.
1. Dequeker et at. / Technology and Health Care 1 (1993) 127-131
Procedures and equations for cross-calibration have been developed for interconversion of results from e.g. DXA on the equipment of one manufacturer in conjunction with measurements made on the machines of another manufacturer. The cross-calibrated data obtained from three different DXA machines (Hologic, Lunar and Norland) were brought together in order to establish combined reference ranges which could serve as European normative data. Figure 1 shows an example for women using the European spine phantom for cross-calibration. The combined reference ranges fit the data from each brand of machine very well. There is a linear decrease in log bone density with age (P < 0.001), with a standard deviation increasing with age (P < 0.02). Similar combined reference ranges were made for DXA spine men, QCT spine men
Table 1 Mean and standard deviation for measurements of ESP on two DXA and one QCT spine machine. The results show differences in calibration procedures used for the different machines Specified density
Cortical & Trabecular (g/ cm 2 ) DXAI
DXAII
mean
sd
mean
sd
0.5 1.0 1.5
0.543 1.091 1.513
0.013 0.016 0.023
0.463 0.898 1.236
0.005 0.008 0.015
Specified density
Trabecular (g/cm 3 )
0.050 0.100 0.200
QCT mean
sd
0.052 0.103 0.210
0.0017 0.0019 0.0027
129
DXA spine reference lines. all CHART¥PE=alL
1.B
•
A
1.7
•
SEX=female
•
1.6 1.5
'"E
1.4
~ 1.3
0
il:j
1.2
£; .,
c: 1.1
'" .,
"0 "0 CD
'E
1.0
•• •
§'" 0.9
(/)
O.B 0.7 0.6 0.5
•
•
0.4 10
20
30
40
50
60
70
BO
90
Age(years)
Fig. 1. Combined reference ranges DXA spine after cross-calibration normal data obtained three machines. a, overall data. b, cross-calibration data obtained by each brand of machine separately.
1. Dequeker et al. / Technology and Health Care 1 (1993) 127 -131
130
Overall reference lines DXA spine M'1'YPE~I!olog~c
1.6
'e1
!:,F:X
lemJJ"
•
1.7
•
1.6
•
•
•
1.5
••
•
•
1.4 N
E
~
Cl
1.3
0
•
::E 1.2
'"
~
.~
Q)
1.1
&
1.0
u u
'E
• • "'~~-!II-,-~~ .-~-~
•
'" '"
• ••
u 0.9 c
Ui
-.
0.6 0.7 0.6
0.5 O.~
10
20
30
40
50
60
70
80
90
Age (years,
Overall reference lines. DXA spine M'j'YP1:- Lunar
1.8
:.;I·:X
[en'dlc
B2
1.7
•
1.6 1.5 1.4 C'J
E
.~
en
1.3
0
::E 1.2
'".~
cQ) 1.1
"0 U
:J; 1.0
:e '"
u
§
0.9
(J)
0.8 0.7 0.6 0.5
10
20
30
40
50 Aqe(years)
Fig. 1. (continued).
•
•
0.4 60
70
60
90
131
1. Dequeker et al. / Technology and Health Care 1 (1993) 127-131 Overall reference lines. DXA spine MTYPJ·;' N o r l d n d
1.6
83
1.7 1.6
- -'
-~---
--_
SI':X~fcmale
• ••
.•
•
1.5 1.4 N
E 0 01 1.3 0
::! 1.2
127
Multicentre European COMAC-BME study on the standardisation of bone densitometry procedures J. Dequeker, J. Reeve, J. Pearson, J. Bright, D. Felsenberg, W. Kalender, C. Langton, A.M. Laval-Jeantet, P. Riiegsegger, G. Van der Perre and the COMAC-BME Quantitative Assessment of Osteoporosis Study Group * (Received 27 May, 1993, Accepted 2 July 1993)
Abstract 26 European centres participated in a concerted research action Biomedical Engineering: Quantitative Assessment of Osteoporosis. With a newly designed European spine and forearm phantom, the stability, accuracy, precision of dual energy absorption (DXA) and quantitative computer tomography (QCT) densitometry machines have been evaluated. Marked and clinically significant differences were found between brands and between techniques. Cross-calibration formulae have been made and normative data evaluated for different regions (spine, femoral neck, femoral trochanteric and forearm). A general fit for all data obtained from different machines was established. The cross-calibration formulae will allow a sensitivity analysis to assist the choice of equipment for clinical management of different categories of patients with bone disease. The present results obtained with an internationally accepted European spine and forearm phantom can now serve to stimulate the manufacturers to improve the comparability of bone measurements between machines. Key words: Osteoporosis; Bone densitometry; QCT; DXA
1. Introduction
Osteoporosis is one of the major health problems of our ageing population. Treatment of a clinically proven osteoporosis (with fractures) is not satisfactory. What is needed is an early diagnosis (in the sense of fracture risk prediction) and the ability to monitor bone response to therapy on
Correspondence to: Prof. Dr. J. Dequeker, Arthritis and Metabolic Bone Disease Research Unit, K.U. Leuven, B-3212 Pellenberg, Belgium.
* CO MAC BME Quantitative Assessment of Osteoporosis Participating centres: Adams J.E., Manchester; Birkenhiiger J.c., Rotterdam; Bonjour J.P., Geneva; Braillon P., Lyon; Diaz Curiel M., Madrid; Felsenberg D., Berlin; Fischer M., Kassel; Galan F., Sevilla; Gennari C., Siena; Geusens P., Leuven; Hemmingsson A, Uppsala; Hyldstrup L., Hvidovre; Jaeger Ph., Bern; Jonson R., Goteborg; Kalef-Ezra J., Ioannina; Kotzki P.O., Montpellier; Kroger H., Kuopio; LavalJeantet A.M., Paris; Lips P., Amsterdam; Osteaux M., Brussels; Reeve J., Harrow; Reid D., Aberdeen; Reiners c., Essen; Ribot c., Toulouse; Riiegsegger P., Ziirich; Schneider P., Wiirzburg. Project Management Group: Dequeker J. (Project Leader), Felsenberg D., Kalender W., Langton c., Laval-Jeantet AM., Reeve J., Riiegsegger P., Van der Perre G.
128
an individual basis. Biomedical engineering has to provide the means so that a patient at risk can be detected in time, and a preventive therapy can be started. Prevention of osteoporotic fractures would improve life quality of 10-20% of the population older than 65 years and would have a considerable influence on the exploding health care cost. In the past, considerable disagreement on measurement and treatment of osteoporosis have done damage to the acceptance and comparability of the respective studies. One of the primary goals of the EEC supported Concerted Research Action on Quantitative Assessment of Osteoporosis is to standardize and unify measurements, to enable the intercomparability of the various measuring procedures and to provide a standardized test bed for new procedures. The development of new methods and procedures will require continuous validation. Measurements of bone mass can now be made precisely. However, measurements made using different manufacturers hardware and software are not directly comparable. The EEC supported concerted action was started in 1990 to stimulate and coordinate the research of this particular problem. The aims of this study were to standardize measurements, to establish reference ranges for normal subjects and to compare the measurement techniques for their power to discriminate between people with spinal osteoporosis, thyroid excess, hip fracture, hyperparathyroidism and long-term corticosteroid treatment. 2. Experimental procedures The measurements were standardized by the use of reproducible operating procedures and by calibration using two newly designed semi-anthropomorphic phantoms. Phantoms were designed for the spine (ESP) and the forearm (EFP) [1,3]. The ESP and the EFP do not contain real muscle and bone tissue, no bone marrow is added. Both phantoms had three measurement sites with specified densities 0.5, 1.0 and 1.5 g/cm 2 or 50, 100 and 200 mg/cm 3 . Each centre was asked to measure the phantoms once per day for the first week and every
1. Dequeker et al. / Technology and Health Care 1 (1993) 127 -131
week the rafter. Centres measured normal subjects and patients at the same time as the phantom measurements. Each subject was measured at least twice on different machines and/or measurement sites. 22 centres participated in the study, 21 centres measured the ESP and 9 the EFP on a variety machines: quantitative computer tomography (QCT), dual X-ray energy absorptiometry (DXA) and single energy photon absorptiometry (SPA). 3. Results In a standard way 977 phantom forms, 6740 subject forms, 2471 questionnaires and 562 laboratory forms were submitted for analysis. The data for the first week measurements and the subsequent weekly measurements of the European spine and forearm phantoms were combined for each machine. Six machines, one DXA spine, three QCT spine, one DXA forearm and one SPA forearm showed evidence of a significant (P < 0.01) drift ranging from -5.3% to +3.9% over a period of 3 months. Most of the machines change less than 3%. Important differences in accuracy and precision were disclosed using ESP and EFP phantoms between different machines and between different brands of machines. Precision and accuracy can be defined as follows: precision is the ability of a system to obtain the same results in repeated measurements of the same subject or object; accuracy is the ability of a system to obtain measured results, that on the average reflect the true values for the measured parameters. The results of the ESP measurements from one centre (Table 1) show that the DXA machine 11 gives lower density values at the specified 1.5 g/cm2 site, and at the other sites to a lesser extent. The DXA I machine measures accurately, but with less precision. The results from another centre, show that a QCT machine gives slightly higher density values at each specified density site, but measures each site with similar precision. The machines not only differ in accuracy in relation to specified densities, but also in precision and this was also dependent on the level of bone density measured.
1. Dequeker et at. / Technology and Health Care 1 (1993) 127-131
Procedures and equations for cross-calibration have been developed for interconversion of results from e.g. DXA on the equipment of one manufacturer in conjunction with measurements made on the machines of another manufacturer. The cross-calibrated data obtained from three different DXA machines (Hologic, Lunar and Norland) were brought together in order to establish combined reference ranges which could serve as European normative data. Figure 1 shows an example for women using the European spine phantom for cross-calibration. The combined reference ranges fit the data from each brand of machine very well. There is a linear decrease in log bone density with age (P < 0.001), with a standard deviation increasing with age (P < 0.02). Similar combined reference ranges were made for DXA spine men, QCT spine men
Table 1 Mean and standard deviation for measurements of ESP on two DXA and one QCT spine machine. The results show differences in calibration procedures used for the different machines Specified density
Cortical & Trabecular (g/ cm 2 ) DXAI
DXAII
mean
sd
mean
sd
0.5 1.0 1.5
0.543 1.091 1.513
0.013 0.016 0.023
0.463 0.898 1.236
0.005 0.008 0.015
Specified density
Trabecular (g/cm 3 )
0.050 0.100 0.200
QCT mean
sd
0.052 0.103 0.210
0.0017 0.0019 0.0027
129
DXA spine reference lines. all CHART¥PE=alL
1.B
•
A
1.7
•
SEX=female
•
1.6 1.5
'"E
1.4
~ 1.3
0
il:j
1.2
£; .,
c: 1.1
'" .,
"0 "0 CD
'E
1.0
•• •
§'" 0.9
(/)
O.B 0.7 0.6 0.5
•
•
0.4 10
20
30
40
50
60
70
BO
90
Age(years)
Fig. 1. Combined reference ranges DXA spine after cross-calibration normal data obtained three machines. a, overall data. b, cross-calibration data obtained by each brand of machine separately.
1. Dequeker et al. / Technology and Health Care 1 (1993) 127 -131
130
Overall reference lines DXA spine M'1'YPE~I!olog~c
1.6
'e1
!:,F:X
lemJJ"
•
1.7
•
1.6
•
•
•
1.5
••
•
•
1.4 N
E
~
Cl
1.3
0
•
::E 1.2
'"
~
.~
Q)
1.1
&
1.0
u u
'E
• • "'~~-!II-,-~~ .-~-~
•
'" '"
• ••
u 0.9 c
Ui
-.
0.6 0.7 0.6
0.5 O.~
10
20
30
40
50
60
70
80
90
Age (years,
Overall reference lines. DXA spine M'j'YP1:- Lunar
1.8
:.;I·:X
[en'dlc
B2
1.7
•
1.6 1.5 1.4 C'J
E
.~
en
1.3
0
::E 1.2
'".~
cQ) 1.1
"0 U
:J; 1.0
:e '"
u
§
0.9
(J)
0.8 0.7 0.6 0.5
10
20
30
40
50 Aqe(years)
Fig. 1. (continued).
•
•
0.4 60
70
60
90
131
1. Dequeker et al. / Technology and Health Care 1 (1993) 127-131 Overall reference lines. DXA spine MTYPJ·;' N o r l d n d
1.6
83
1.7 1.6
- -'
-~---
--_
SI':X~fcmale
• ••
.•
•
1.5 1.4 N
E 0 01 1.3 0
::! 1.2
