Article Text


Clinical usefulness of MRI in multisystem atrophy
  1. Department of Neurology
  2. Austin and Repatriation Medical Centre
  3. Banksia Street, West Heidelberg 3081, Australia
  1. Dr Carl Counsell, Department of Neurology, Austin and Repatriation Medical Centre, Banksia Street, West Heidelberg 3081, Australia. Telephone 00613 9496 2845; fax 00613 9496 4065; email counsells{at}
  1. A SCHRAG,
  2. N P QUINN
  1. Department of Clinical Neurology, Institute of Neurology
  2. Queen Square, London

    Statistics from

    Schrag et al suggest that certain putaminal and infratentorial changes on MRI are useful in distinguishing between patients with multisystem atrophy (MSA) and patients with idiopathic Parkinson’s disease.1 The specificity and positive predictive value of these changes were both about 90%. However, whether these changes will be useful in clinical practice or epidemiological research is unclear for several reasons.

    The number of patients included was small and so the confidence intervals were wide. For example, the specificity of the MRI changes for MSA could be as low as 80%. Moreover, only patients with clinically probable MSA were included. In this group of patients the clinical diagnosis alone had a positive predictive value as high as that of MRI and so there would seem to be little added value of MRI (14/15 (93%) patients with probable MSA had the diagnosis confirmed on postmortem). A more relevant question is whether the MRI changes are equally specific in those with possible MSA in whom the clinical diagnosis is much less certain. Indeed it is also unclear from this study whether the MRI changes are specific to MSA as patients with other conditions that enter into the differential diagnosis were not included. It may therefore be more correct to state that the MRI changes are helpful in excluding Parkinson’s disease rather than in confirming MSA.

    Finally, the positive predictive value of MRI quoted in this study is likely to be an overestimate compared with its routine use in most movement disorder clinics. Schrag et alincluded a very high proportion of patients with MSA (nearly 50%) compared with Parkinson’s disease. As the positive predictive value is directly related to the prevalence of the disease in a given population,2 this resulted in a high positive predictive value. In a typical movement disorder clinic, fewer than 10% of patients will have MSA, in which case, even if the specificity of MRI is 90%, the positive predictive value would only be about 50%—that is, only half of those with the MRI changes would turn out to have the disease.

    It is, therefore, too early to include specific MRI changes as part of the diagnostic criteria for MSA.


    Schrag and Quinn reply:

    Counsell and Hughes raise several potential drawbacks to our study1-1 that we willingly acknowledge. Firstly, that the number of patients included was small (about 45 in each group); secondly, that the 1:1 proportion of patients with Parkinson’s disease and those with multiple system atrophy (MSA) atrophy is unrepresentative of the real life situation in which the ratio is >10:1. A more ideal study might include 100 patients with MSA (to help counter the criticism of small numbers) and more than 1000 patients with Parkinson’s disease, but would be impractical, particularly for the Parkinson’s disease group who, unlike patients with atypical disease, are not usually subjected to MRI. We agree that, as clearly stated in the abstract, our study was restricted to a comparison between multisystem atrophy, Parkinson’s disease, and controls, and are currently conducting a further study additionally including patients with other degenerative syndromes. Since the completion of the report, patients with Machado-Joseph disease have been reported with similar infratentorial abnormalities.1-2 We also agree that in the diagnosis of clinically probable MSA, there is little added value of MRI; involvement of the cerebellum and its pathways is usually already clinically evident before its demonstration by MRI or CT,1-3 and even in cases without cerebellar involvement the diagnosis is still a clinical one. Moreover, as we emphasised, a minority of patients with probable multisystem atrophy have a normal MRI. Therefore, unlike others,1-4 we have never proposed MRI changes as part of the diagnostic criteria for multisystem atrophy. As discussed, the sensitivity of the method may be lower early, and higher late in the disease. However, for the purpose of validation of a proposed diagnostic aid imaging findings need to be related to a clinically probable diagnosis rather than a possible diagnosis. The “gold standard” is definite, pathologically confirmed disease, but this was achieved in only one patient in our series. In conclusion, we nevertheless think that our, admittedly imperfect, blinded MRI study (the first conducted in MSA) has helped to determine the prevalence of certain MRI abnormalities in patients with clinically probable MSA in comparison to patients with Parkinson’s disease and controls . It has also, perhaps more importantly, revealed the limitations of MRI in this context. However, expert clinical evaluation remains the cornerstone of the diagnosis in life, and it is also more cost effective than resorting to expensive imaging techniques.


    1. 1-1.
    2. 1-2.
    3. 1-3.
    4. 1-4.
    View Abstract

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.