Magnetic resonance imaging for the diagnosis of multiple sclerosis
The quality of health care has improved immeasurably in recent years
because of the combined efforts of doctors and researchers to use the best
available evidence to inform decisions about clinical practice for
individual patients. In the context of medical interventions, a large body
of research clearly shows that it is wise to execute caution when
estimating intervention effects from personal clinical experience, and
that undertaking systematic reviews may act more towards patients’ best
interests.(1) We hope that our paper will promote collaborations between
doctors and researchers in the context of the diagnosis of multiple
sclerosis (MS), rather than dividing different groups into opposing camps.
We agree with Miller et al., and emphasised in our paper, that there
are multiple reasons that magnetic resonance imaging (MRI) may be required
in the clinical workup of patients with suspected MS, in particular its
contribution to differential diagnoses. Our paper focussed on the narrower
question of the diagnostic accuracy of MRI for MS. We also agree with
Polman et al. that pooling 20 years of research on this topic into a
single analysis would be misleading. For this reason, our review did not
“combine studies of different design and quality”, as Polman et al.
assert. In a paper that presented eight figures displaying different
aspects of the diagnostic accuracy of MRI for MS, we quoted only two
summary statistics for diagnostic accuracy.(2) These demonstrated
conclusively (1) that MRI is diagnostic for MS (diagnostic odds ratio for
cohort studies 9, 95% CI 5 to 16) and (2) that its diagnostic accuracy has
been seriously exaggerated in case-control type evaluations that compare
results of MRI images in a group of patients with MS with a separate group
known not to have MS (diagnostic odds ratio 213, 95% CI 85 to 535).
Given that the most widely used diagnostic algorithm (3) and a number
of rules developed for interpreting MRI (4-6) were based on evidence from
case-control studies as well as cohort studies, we proceeded to evaluate
their diagnostic accuracy using available evidence from cohort studies
alone. We were able to present results of studies evaluating both the
McDonald 2001 criteria and commonly used MRI criteria (Barkhof, Paty,
Fazekas) in cohort studies with up to 3 years’ follow up (McDonald,
Barkhof, Fazekas) or 6 years’ follow up (Paty). To our knowledge, we
evaluated all available evidence. Therefore, we do not know why Polman et
al. believe in the existence of an algorithm that “was not properly tested
in this study”. Even if a decision rule that includes MRI has good
diagnostic accuracy this does not tell us the value of including MRI in
the rule – to assess this requires a comparison of the performance of
rules with and without inclusion of MRI. We need research that examines
the additional diagnostic accuracy that results from adding MRI to
clinical history and examination.
As the diagnosis of MS depends on long-term clinical follow up, the
diagnostic accuracy of newer MRI technologies will by definition not yet
have been evaluated in cohort studies. Further, we found only limited
evidence as to the diagnostic accuracy of gadolinium enhancement, spinal
cord imaging and repeat scanning and therefore were unable to evaluate
these separately, although they form part of the Barkhof and McDonald
criteria. There was also insufficient evidence to evaluate the accuracy
of MRI in patients presenting with different clinical symptoms. We
question the assertion of Miller et al. that “a cord syndrome is the
presenting feature of ~50% of MS patients”: the frequency of spinal cord
lesions ranged from 9% to 45% in the highly selected series included in
the review cited. Therefore, in clinical practice the accuracy of brain
MRI in patients with a first clinical event suggestive of MS remains an
important issue. We hope that our paper will stimulate further research
that will address this and other relevant questions including the
diagnostic value of newer MRI technologies, and hence clarify the best use
of MRI in the diagnosis of MS.
1. Antman EM, Lau J, Kupelnick B, Mosteller F, Chalmers TC. A
comparison of results of meta-analyses of randomized control trials and
recommendations of clinical experts. Treatments for myocardial infarction.
JAMA. 1992;268:240-248.
2. Whiting P, Harbord R, Main C, Deeks JJ, Filippini G, Egger M,
Sterne JAC.
Accuracy of magnetic resonance imaging for the diagnosis of multiple
sclerosis: systematic review. BMJ online, 2006
3. McDonald WI, Compston A, Edan G et al. Recommended diagnostic
criteria for multiple sclerosis: guidelines from the International Panel
on the diagnosis of multiple sclerosis. Ann Neurol. 2001;50:121-127.
4. Barkhof F, Filippi M, Miller DH et al. Comparison of MRI criteria
at first presentation to predict conversion to clinically definite
multiple sclerosis. Brain. 1997;120:2059-2069
5. Fazekas F, Offenbacher H, Fuchs S et al. Criteria for an increased
specificity of MRI interpretation in elderly subjects with suspected
multiple sclerosis. Neurology. 1988;38:1822-1825.
6. Paty DW, Oger JJ, Kastrukoff LF et al. MRI in the diagnosis of MS:
a prospective study with comparison of clinical evaluation, evoked
potentials, oligoclonal banding, and CT. Neurology. 1988;38:180-185.
Competing interests:
None declared
Competing interests:
No competing interests
22 April 2006
Jonathan A.C. Sterne
Reader
Jonathan J. Deeks, Penny Whiting, Roger Harbord, and Graziella Filippini.
MRC Health Services Research Collaboration, Department of Social Medicine, University of Bristol
Rapid Response:
Magnetic resonance imaging for the diagnosis of multiple sclerosis
The quality of health care has improved immeasurably in recent years
because of the combined efforts of doctors and researchers to use the best
available evidence to inform decisions about clinical practice for
individual patients. In the context of medical interventions, a large body
of research clearly shows that it is wise to execute caution when
estimating intervention effects from personal clinical experience, and
that undertaking systematic reviews may act more towards patients’ best
interests.(1) We hope that our paper will promote collaborations between
doctors and researchers in the context of the diagnosis of multiple
sclerosis (MS), rather than dividing different groups into opposing camps.
We agree with Miller et al., and emphasised in our paper, that there
are multiple reasons that magnetic resonance imaging (MRI) may be required
in the clinical workup of patients with suspected MS, in particular its
contribution to differential diagnoses. Our paper focussed on the narrower
question of the diagnostic accuracy of MRI for MS. We also agree with
Polman et al. that pooling 20 years of research on this topic into a
single analysis would be misleading. For this reason, our review did not
“combine studies of different design and quality”, as Polman et al.
assert. In a paper that presented eight figures displaying different
aspects of the diagnostic accuracy of MRI for MS, we quoted only two
summary statistics for diagnostic accuracy.(2) These demonstrated
conclusively (1) that MRI is diagnostic for MS (diagnostic odds ratio for
cohort studies 9, 95% CI 5 to 16) and (2) that its diagnostic accuracy has
been seriously exaggerated in case-control type evaluations that compare
results of MRI images in a group of patients with MS with a separate group
known not to have MS (diagnostic odds ratio 213, 95% CI 85 to 535).
Given that the most widely used diagnostic algorithm (3) and a number
of rules developed for interpreting MRI (4-6) were based on evidence from
case-control studies as well as cohort studies, we proceeded to evaluate
their diagnostic accuracy using available evidence from cohort studies
alone. We were able to present results of studies evaluating both the
McDonald 2001 criteria and commonly used MRI criteria (Barkhof, Paty,
Fazekas) in cohort studies with up to 3 years’ follow up (McDonald,
Barkhof, Fazekas) or 6 years’ follow up (Paty). To our knowledge, we
evaluated all available evidence. Therefore, we do not know why Polman et
al. believe in the existence of an algorithm that “was not properly tested
in this study”. Even if a decision rule that includes MRI has good
diagnostic accuracy this does not tell us the value of including MRI in
the rule – to assess this requires a comparison of the performance of
rules with and without inclusion of MRI. We need research that examines
the additional diagnostic accuracy that results from adding MRI to
clinical history and examination.
As the diagnosis of MS depends on long-term clinical follow up, the
diagnostic accuracy of newer MRI technologies will by definition not yet
have been evaluated in cohort studies. Further, we found only limited
evidence as to the diagnostic accuracy of gadolinium enhancement, spinal
cord imaging and repeat scanning and therefore were unable to evaluate
these separately, although they form part of the Barkhof and McDonald
criteria. There was also insufficient evidence to evaluate the accuracy
of MRI in patients presenting with different clinical symptoms. We
question the assertion of Miller et al. that “a cord syndrome is the
presenting feature of ~50% of MS patients”: the frequency of spinal cord
lesions ranged from 9% to 45% in the highly selected series included in
the review cited. Therefore, in clinical practice the accuracy of brain
MRI in patients with a first clinical event suggestive of MS remains an
important issue. We hope that our paper will stimulate further research
that will address this and other relevant questions including the
diagnostic value of newer MRI technologies, and hence clarify the best use
of MRI in the diagnosis of MS.
1. Antman EM, Lau J, Kupelnick B, Mosteller F, Chalmers TC. A
comparison of results of meta-analyses of randomized control trials and
recommendations of clinical experts. Treatments for myocardial infarction.
JAMA. 1992;268:240-248.
2. Whiting P, Harbord R, Main C, Deeks JJ, Filippini G, Egger M,
Sterne JAC.
Accuracy of magnetic resonance imaging for the diagnosis of multiple
sclerosis: systematic review. BMJ online, 2006
3. McDonald WI, Compston A, Edan G et al. Recommended diagnostic
criteria for multiple sclerosis: guidelines from the International Panel
on the diagnosis of multiple sclerosis. Ann Neurol. 2001;50:121-127.
4. Barkhof F, Filippi M, Miller DH et al. Comparison of MRI criteria
at first presentation to predict conversion to clinically definite
multiple sclerosis. Brain. 1997;120:2059-2069
5. Fazekas F, Offenbacher H, Fuchs S et al. Criteria for an increased
specificity of MRI interpretation in elderly subjects with suspected
multiple sclerosis. Neurology. 1988;38:1822-1825.
6. Paty DW, Oger JJ, Kastrukoff LF et al. MRI in the diagnosis of MS:
a prospective study with comparison of clinical evaluation, evoked
potentials, oligoclonal banding, and CT. Neurology. 1988;38:180-185.
Competing interests:
None declared
Competing interests: No competing interests