Article Text

Measuring the rate of progression and estimating the preclinical period of Parkinson’s disease with [18F] dopa PET
  1. St George’s Healthcare, Wolfson Neuro-rehabilitation Centre, Regional Neurosciences Unit, Atkinson Morley’s Hospital, Copse Hill, Wimbledon, London SW20 0NE, UK.
    1. Department of Neurology, University Hospital of Wales, Heath Park, Cardiff CF4 4XN, UK

      Statistics from

      Morrish et al 1 report in great detail the PET data on 32 patients with Parkinson’s disease, from which they conclude that the mean pre-clinical period “is unlikely to be longer than 7 years”. This conclusion is based on calculations using the [18F] dopa influx constant (Ki) of the putamen, although they acknowledge that other methods of analysis and extrapolation yielded estimates of anything between 2.8 and 37.2 years. The authors justify using putamen Ki because it was “more sensitive to increasing disability” than either total striatal assessment or using the alternative ratio approach, but fail to justify a much more fundamental and unwise assumption on which their arguments rest—that is, the intercorrelation between the PET index, clinical progression, and the UPDRS.

      The paper gives little detail about how the UPDRS was administered, presumably only once, before each of the two PET scans an average of 18 months apart. A linear regression was then applied to the mean of each patient’s two UPDRS and PET assessments, the gradient of which was expressed as a percentage change in the PET index for a change of 10 points “in the total UPDRS”.

      Some questions can be raised:

      (1) Did the same observer administer the UPDRS blinded to the clinical diagnosis, on their 16 normal controls as well as to each patient on both occasions and, if not, was interobserver reliability studied?

      (2) Presumably the “total UPDRS”, judging by the scale shown on their figure A, was actually the total score from the 14 items in the motor subset of the UPDRS, which measures impairment rather than disability.

      (3) The UPDRS is neither a perfect nor a linear scale. Indeed two coauthors of this paper have pointed out elsewhere the low interrater reliability in some items and redundancy in others.2 It is a composite multi-item index of severity of disease, each item being an ordinal rather than an interval 0–4 scale of one clinical feature. The key distinction is that an ordinal scale permits the recording of data in rank order (for example, mild, moderate, severe) but without uniform intervals. Thus tremor score 4 is not twice as bad as 2, still less a total motor UPDRS score of 40/70 v 20/70. For these reasons, the use of simple arithmetic means as well as other parametric statistical methods is inappropriate, however tempting.

      One illustration of the non-linearity of the UPDRS manifest to anyone who has used it regularly in clinical trials is the bias towards intermediate scores. Those with advanced disease and high scores are seldom if ever recruited, and some of the items scored as 1, indicating slight or mild impairment, “could be normal for some” according to the definition. In one study of Alzheimer’s disease, 56% of 78 cases and 12 of the 20 age matched controls were found to have isolated extrapyramidal signs with motor UPDRS scores of 4.5 (± 4.8) and 2.8 (± 1.8) respectively using observers not blinded to the diagnosis.3 It would be interesting to know whether a UPDRS score>0 is sensitive to or predictive of preclinical parkinsonism and/or abnormal PET.

      Furthermore, as it is acknowledged that Parkinson’s disease may progress at varying rates between patients and possibly within the same patient at different ages and stages, it is perhaps not surprising that the authors found no significant correlation between change in UPDRS and change in any PET index over 18 months. Thus it seems unwise to draw such firm conclusions based on the assumption that both measures are linear and directly correlated.


      Morrish replies:

      We thank Hardie for his comments but are surprised that he finds difficulty in our assumption of a relation between PET index, clinical progression, and the UPDRS. Whether clinical severity is measured by UPDRS, bradykinesia scores, rigidity scores or Purdue pegboard scores1-1 1-2 such a relation has been a consistent finding in [18F]dopa PET imaging studies of Parkinson’s disease. The UPDRS was administered on 57 of 64 occasions by one observer (PKM) and on seven occasions by a second observer (JSR). UPDRS scoring was not carried out on the normal volunteers. Gonera et al have identified some non-specific symptoms that may predate the development of Parkinson’s disease1-3 but we know of no population study of the predictive value of UPDRS score in normal subjects. By total UPDRS score we mean the combined scores of sections I, II, III, and IV. Similar results were found when motor scores alone were examined.1-4 The UPDRS scale is the most widely used index of global disease severity in Parkinson’s disease. We accept that a linear correlation between UPDRS and PET index may have been inappropriate. The PET index represents a figure of mean [18F] dopa metabolism throughout the putamen, caudate, or total striatum whereas the clinical presentation and severity of parkinsonism is likely to depend on the distribution and severity of loss of dopaminergic function (and that of other neurotransmitters) within and outside the basal ganglia. It is unlikely that the relation is so simple yet this approach has allowed the demonstration of an aspect of the measurement of progression by PET that has not previously been considered, that of sensitivity to clinical severity. It should be noted that this discussion is not relevant to the major findings of the study (that measurement of progression is dependent on the PET method and that the average preclinical period is likely to be short), only to our explanation of these findings. However, it does suggest an important debate; should clinical indices or functional imaging indices be used independently in studies of progression in Parkinson’s disease? When the reproducibility of both measurements is taken into account it is, as Hardie comments, not surprising that we found no significant correlation between change in UPDRS and change in PET index.


      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.