Improved accuracy of clinical diagnosis of Lewy body Parkinson’s disease

Patients and methods.

One hundred consecutive cases with clinically diagnosed IPD at the time of the last assessment prior to death were studied. The methodology was identical to our initial study in 1992.2 The 100 cases were collected at the United Kingdom PD Society Brain Research Center (UKPDSBRC) in London between 1996 and 1998, from all over the UK. Sixty-one were preregistered brain donors and had been prospectively assessed annually by clinicians associated with the UKPDSBRC. Eighty-six cases had been followed by neurologists, seven by geriatricians, and seven by internal medicine specialists.

The pathologic diagnoses were made by one neuropathologist (SED) using standard criteria.2,4,5⇓⇓

The clinical features of each case were abstracted from the case records by one of us (AJH), and the diagnosis reevaluated according to three different sets of clinical criteria: the UKPDSBRC clinical diagnostic criteria6; those proposed by Calne et al.7 (clinically definite category); and those proposed more recently by Gelb et al.8 (clinically possible and clinically probable categories). The clinical features were also assessed according to a selected set of features (asymmetrical onset, no atypical clinical features for IPD, and no possible etiology for another parkinsonian syndrome) which have been suggested to give the highest proportion of true IPD cases (93%).9

Results.

The mean age at onset of disease was 62.2 years (range 29 to 82), mean disease duration 15 years (range 3 to 35), and mean Hoehn and Yahr stage at death was 4.6 (range 3 to 5). Fifty-six patients were men and 44 women. The last clinical assessment was mean 1.2 years before death (range 0.1 to 6). Ninety cases fulfilled the pathologic criteria for IPD. The neuropathological diagnoses in the other 10 cases were multiple system atrophy (6), progressive supranuclear palsy (2), neurofibrillary tangle pathology (1; probable postencephalitic parkinsonism), and extensive cerebrovascular disease with no other pathologic explanation for parkinsonism (1).

When the clinical features were assessed according to the different sets of clinical diagnostic criteria, 12 patients did not fulfill the UKPDSBRC clinical diagnostic criteria for IPD6 (see table). Nine had exclusion criteria (four with early severe dementia, three had neuroleptic treatment at the time of onset of parkinsonism, one had no response to an adequate trial of levodopa, and one had vertical supranuclear gaze palsy). Three other patients had inadequate supportive criteria. Nine of these 12 had pathologic IPD, whereas three had other parkinsonian syndromes. Thus the application of these diagnostic criteria improved the diagnostic accuracy only slightly, with 81 of the remaining 88 patients (92%) having pathologic IPD. Using the criteria for clinically definite IPD proposed by Calne et al.,7 11 patients were rejected, nine with IPD and two with other parkinsonian syndromes. The diagnostic accuracy was thus 91%, with 81 of the remaining 89 patients having pathologic IPD. The criteria proposed by Gelb et al.8 for clinically possible IPD rejected 16 cases, 12 with IPD and four not having IPD; whereas for clinically probable IPD, 29 cases were rejected, 25 with IPD and the same four cases without IPD. The diagnostic accuracy was thus not increased by using the more stringent criteria for clinically probable IPD (92%) compared to clinically possible IPD (93%), despite more than double the number of true IPD case being rejected. Using the three selected clinical features previously suggested to give the highest proportion of true IPD cases,9 60 of the 67 cases with all three features present had pathologic IPD. Thus the diagnostic accuracy (90%) was not increased despite 30 of the 90 confirmed IPD cases being rejected. Six cases without IPD failed to be detected by any of the criteria.

Discussion.

There are potential weaknesses inherent in all clinicopathological series which may concentrate interesting, extensively investigated, or unusual cases and select for institutionalized patients and therefore for more severe disability. Referral bias and the length of follow-up also limit the extent to which these results can be generalized to other clinical settings, particularly to early disease. Nevertheless this study shows a significant improvement (14%, 95% CI 3.8 to 24.2, p = 0.008) in the clinical diagnostic accuracy of IPD over the last decade and suggests a greater awareness among clinicians of the pitfalls in diagnosis of IPD. In the present series, multiple system atrophy was responsible for the majority of the false positive cases; but in two previous series, progressive supranuclear palsy was the disorder most likely to be clinically mistaken for IPD.2,3⇓ Similarly, there has been a reduction in cases with nigral atrophy without inclusions, Alzheimer-type pathology (tangles and plaques), neurofibrillary tangle pathology, and vascular parkinsonism being misdiagnosed as IPD.1-3⇓⇓

Although no attempt was made to establish the criteria referring clinicians used to diagnose IPD, 12% of patients failed to satisfy the clinical diagnostic criteria most familiar to those treating IPD in the UK.6 This might suggest that clinicians looked at the overall picture and did not negate the diagnosis even in the presence of accepted exclusion criteria, e.g., early severe dementia or neuroleptic treatment at the time of onset of parkinsonism. That the diagnostic accuracy was not improved by rigidly applying these criteria would support this approach. In our study from almost 10 years ago, the number of cases not satisfying the same diagnostic criteria was similar at 11%.2 However in that series, the removal of such cases increased diagnostic accuracy from 76% to 82% because, in contrast to the present series, most of the rejected cases did not have IPD.2 This suggests that clinicians are now interpreting atypical clinical features more appropriately.

The retrospective application of different diagnostic criteria resulted in a similar diagnostic accuracy to the treating clinicians. Some led to many more cases of true IPD being excluded with no gain in diagnostic accuracy. It is interesting that only four of the 10 false positive cases were rejected by the different sets of diagnostic criteria. This suggests that although rigorous application of diagnostic criteria may decrease false positive cases, a number of cases will still mimic IPD sufficiently closely to not be rejected by current diagnostic criteria. These results should not be taken as an assessment or comparison of the different diagnostic criteria in a wider setting. Comparing their ability to improve diagnostic accuracy in patients with clinically diagnosed IPD at death has little or no relationship to their applicability in other clinical settings, such as early disease or when the diagnosis is made by clinicians with different expertise to those associated with the UKPDSBRC.

It is possible that specialists in movement disorders might be more accurate at diagnosing IPD. However, in the clinicopathological study by Litvan et al.,10 the positive predictive value for the diagnosis of Lewy body disease (including both IPD and dementia with Lewy bodies) by experts in movement disorders at the last clinical visit prior to death was only 77.4%. For the primary neurologists, the positive predictive value was 70.3%. In that study, the cases were collected from several different centers, and the proportion of cases with other parkinsonian syndromes may have adversely affected the diagnostic accuracy.10 In any case, the present finding of a diagnostic accuracy of 90% compares favorably with available data.

Despite the shortcomings of available diagnostic criteria we would emphasis that in routine clinical practice it is possible for neurologists to accurately diagnose the large majority of patients with IPD. The failure to improve the diagnostic accuracy with the retrospective application of diagnostic criteria suggests that a diagnostic accuracy of 90% may be the highest that can be expected using current clinical diagnostic criteria. This needs to be taken into consideration in the interpretation and design of clinical trials.