Patterns of Asymmetry Do Not Change Over the Course of Idiopathic Parkinsonism

Traditional neurology has emphasized the value of studying the temporal and spatial distribution of symptoms and signs. Careful documentation of how (in terms of time course) and where (in terms of distribution) clinical features emerge and evolve has often allowed conclusions to be drawn concerning the site and nature of a neuropathologic lesion. For example, the sudden onset of weakness on one side of the body followed by rapid but incomplete improvement leads clinicians to infer a vascular disturbance damaging the contralateral corticospinal pathway. The clinical phenomenology of several other neurologic problems, such as multiple sclerosis and intracranial neoplasia, has been equally rewarding in providing some understanding of the underlying pathogenesis. The neurodegenerative disorders are also potentially accessible to this type of analysis. The pathogenesis of neurodegeneration is more obscure than that of any other major category of neurologic disorder. However, there is no shortage of hypotheses that invoke putative mechanisms that might be responsible for idiopathic loss of functionally related groups of neurons.

We have recently found that in idiopathic parkinsonism (IP), the bradykinesia scores derived from the Unified Parkinson's Disease Rating Scale (UPDRS) correlate inversely but linearly with the fluorodopa uptake constant measured by PET ^[1,2]. The fluorodopa uptake constant also correlates linearly with the number of nigral dopaminergic neurons ^[3,4]. We can therefore estimate the degree of integrity of the nigrostriatal pathology by assessing bradykinesia ^[1]. We have also presented a justification for temporary withdrawal of antiparkinsonian treatment for 12 hours to obtain an index of the severity of the nigral lesion ^[1]. From our analysis of 238 patients, we concluded that (1) the time course of evolution of the nigral lesion in IP is in keeping with a model of pathogenesis deriving from a transient event that kills some neurons and damages others so that they may survive in a functional state for several years but will ultimately undergo premature death. (2) Alternatively, our observations could be explained in terms of an event that initiates a process that engages and kills healthy neurons at a constant rate over a protracted time course.

We now seek to explore which of these two types of mechanisms is likely to be occurring, using evidence relating to the time course of progression of the asymmetry and the focality of bradykinesia in IP. The rationale for this study is that a transient causal event may be expected to establish localized lesions, each of which would evolve with similar rates to a level of severity that would be determined by the extent of the initial insult. In contrast, a process that continued to invade normal neurons over a prolonged period would lead to the convergence of all lesions toward the maximum level attainable within the patient's life span. In this way, our observations should help to choose between an "event" or "process" hypothesis. Our observations involved 198 of the 238 patients that we previously studied.

Methods. Subjects. Two hundred thirty-eight patients with IP were seen between August 1984 and October 1992. Most were recruited from our movement disorder clinic and from our inpatient neurology service; we also recruited patients from chronic care facilities to ensure a sufficient number of patients with advanced disease. We included patients with a diagnosis of definite IP according to the criteria of Calne et al ^[5]. Exclusion criteria comprised solely unilateral deficits as well as bradykinesia scores attaining the maximal value on one or both sides; these exclusions guarded against the introduction of bias toward decreased asymmetry at the very early and very late durations and reduced distortion due to a ceiling effect. In addition, exclusions included a history of neurosurgical procedures or the presence of other neurologic diseases. The number of subjects who met the criteria was 198 (134 men, 64 women). The mean age was 64.59 +-\11.53 years (mean +-\SD; range, 33 to 90), and the mean duration of symptoms was 9.1 +-\5.82 years (range, 1 to 26). The mean bradykinesia score was 6.72 +-\2.17 (range, 1 to 11) for the more affected side and 4.84 +-\2.35 (range, 1 to 11) for the less affected side. Most patients were taking carbidopa/levodopa, bromocriptine, or both. Some patients were taking selegiline. Patients taking controlled-release carbidopa/levodopa were not included in the study. Duration of disease was recorded from the onset of symptoms.

Measurements. After 12 hours off all antiparkinsonian drugs, patients were assessed in the morning at least 1 hour after arising. Clinical severity was quantified by bradykinesia scores of the limbs, derived from the UPDRS. The maximum possible total score for bradykinesia was 24 (8 for each arm, 4 for each leg).

Statistical methods. We explored the asymmetry of focal deficits by examining the distribution of deficits in the four limbs. By standardizing the scores and weighting tied values, we were able to identify patterns of asymmetry, comparing the two more advanced limbs with the two limbs with lower scores. We recorded the relative frequencies of the following various focal asymmetries: lateralization, upper versus lower limbs, and diagonal (crossed) asymmetries.

The relationship between the total scores for bradykinesia, the patients' ages, and the durations of symptoms was analyzed by multiple linear and nonlinear regression.

To compare rates of progression between the more affected and the less affected limbs or sides, analysis was carried out for the higher scores (more affected) and for the lower scores (less affected) of bradykinesia in each patient. Since the higher and lower scores within the same patient were correlated, the analysis was carried out as a bivariate multiple regression. The two scores to be compared were jointly regressed on age and on duration of symptoms, and an adequate fit was obtained. The differences between the higher and lower scores (ie, the asymmetry) were compared statistically for different durations, adjusting for associations between differences. We undertook this multivariate analysis to take into account correlations of assigned scores within each subject--clearly, a patient with early IP is more likely to have lower scores in all limbs than a patient with advanced pathology. Independent univariate analysis would have failed to address this association of values within each patient.

Results. Lateral asymmetries were far more common than asymmetries between arms and legs; the latter type in turn dominated over diagonal asymmetries (right arm/left leg versus left arm/right leg). The corresponding frequencies were 103 for lateral asymmetries, 73 for asymmetries between arms and legs, and 22 for diagonal asymmetries.

The scores for lateral asymmetry are summarized in table 1. The linearly age-adjusted bradykinesia scores were well fitted by quadratic curves on duration (table 2). The age-adjusted quadratic curves of bradykinesia versus duration for the more affected and the less affected sides are shown in the figure 1. The corresponding curves for the upper limbs and lower limbs also are shown in the figure.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Comparison between sides: quadratic regression curves of age-adjusted bradykinesia scores versus duration for more affected side (curve A) and less affected side (curve B) (curve A: 4.225 + 0.0257 x age + 0.243 x duration - 0.0091 x (duration) ^[2]; curve B: 1.263 + 0.0392 x age + 0.284 x duration - 0.0097 x (duration) ^[2]; N = 198). Comparison between arms and legs: quadratic curves of the age-adjusted standardized bradykinesia scores versus duration for the more affected (curve C) and the less affected (curve D) of the arms versus the legs (curve C: 2.332 + 0.070 x age + 0.434 x duration - 0.0158 x (duration) ^[2]; curve D: 0.690 + 0.0722 x age + 0.275 x duration - 0.00887 x (duration) ^[2]; N = 198)

The bradykinesia scores of the more affected and the less affected sides were significantly different over the entire range of duration (mean score difference +-\SE, 2.046 +-\0.475 at 1 year of duration, p < 0.001; 1.593 +-\0.303 at 15 years of duration, p < 0.001). There was no significant difference between the age-adjusted differences in bradykinesia scores at 1 year of duration and at 15 years of duration (p = 0.75). Analogous findings were obtained when comparing one arm with the other, and both arms with both legs.

Discussion. We have focused our attention on measurements of bradykinesia because UPDRS scores for this particular deficit have been shown to have a direct linear correlation with the number of surviving dopaminergic nigral neurons ^[1-4]. Measures of asymmetry allow "within-patient" comparisons that have several advantages. Each measure was derived within the same genetic constitution and duration of disease. Furthermore, each patient was examined by the same evaluator at the same time in the same circumstances (including therapeutic state). We have outlined elsewhere ^[1] that 12 hours off antiparkinsonian medication provides a satisfactory background for assessment, as recommended by the Core Assessment Program for Intracerebral Transplantation (CAPIT) ^[6].

We have recently reported ^[1] that the progression of bradykinesia follows a quadratic time course and have inferred that this reflects the temporal profile of death of nigral dopaminergic neurons. The initial, faster phase is followed by a slower stage that approaches the normal age-related decline. We have been interested in the natural history of the relative deficit in the individual limbs of patients with IP. Characteristically, IP produces bradykinesia of differing severity in the four limbs. Elucidating the time course of the relationship between these focal deficits should help us to understand the nature of the pathogenesis. Does the focal asymmetry increase, decrease, or stay the same as pathology advances? The answer to this question will have implications on whether a continuing pathologic process is engaging healthy nerve cells or a one-time pathologic event has afflicted the nigra unevenly, with damage confined to the areas involved.

Here we report that with increasing duration of symptoms, the focal asymmetry of bradykinesia is sustained without any significant changes, in spite of the progressively accumulating deficits that appear with advancing pathology. This finding is unexpected, because some current hypotheses that explain the pathogenesis of IP envision a process that engages healthy neurons ^[7,8]. Such a mechanism would be anticipated to proceed toward the same extent of neuronal loss in all regions; in other words, the neuronal counts for the less afflicted regions of the nigra should converge toward those of the worse areas as the more involved regions reach a plateau. We have excluded the possibility that the plateau is an artifact due to a ceiling effect by limiting the analysis to scores below the upper limit of the protocol.

We undertook a cross-sectional design in our study because a longitudinal series of observations over the substantial period necessary (15 years) was not feasible. We examined our findings to exclude any systematic bias in the selection of patients, so our population represents as random a sample as can be obtained.

Our study of the distribution of deficits indicates that one side tends to be affected predominantly and persistently throughout the course of illness. The arms are involved to a greater extent than the legs. The major tendency is for greater deficits to occur on one side. We can infer from these findings that the lesion is most frequently asymmetric from side to side but is also asymmetric between the upper and lower limbs. The pathology is clearly focal throughout its natural history rather than one that conforms to a diffuse pattern of spread.

How can we resolve the paradox between our observations and certain present theoretical constructs that explain underlying pathogenesis? Our findings can be construed as evidence in favor of an event hypothesis rather than a process hypothesis ^[9,10]. In biologic terms, the event hypothesis is analogous to the concept of "intrinsic cell death," whereas the process hypothesis would correspond to "interactive cell death" ^[11].

As models for an event, we can consider poliomyelitis and von Economo's encephalitis. Acute infection induces damage that is distributed haphazardly and unequally to motor neurons on each side of the CNS. There is, however, an important difference between the natural history of IP and that of poliomyelitis or von Economo's encephalitis; IP advances inexorably, whereas poliomyelitis and von Economo's encephalitis are relatively stable for several decades after the causal event. What category of biological mechanism could account for this difference? Evidence from several sources ^[12-15] suggests that this spatiotemporal pattern of pathogenesis might result from DNA damage induced by a toxin or virus. The concept of damage to DNA leading to chronic disease is not new ^[12-15]. DNA displays limited stability; spontaneous decay has been cited as a cause of neoplasm and aging ^[16].

A hypothesis of a genomic disturbance in IP is in accord with following considerations:

1. The natural history of IP is rather similar to that of genetic disorders. Both have a long latency, slow progression over decades, and marked interindividual variation.

2. There exist familial forms of IP that are indistinguishable from nonfamilial IP clinically and pathologically ^[17,18].

3. Damaged neurons cannot be replaced with new cells because neurons are postmitotic ^[13]. The accumulation of damage to DNA is more pronounced in nondividing cells ^[13,19]. Mullaart et al ^[20] claim that increased breakdown of DNA occurs in the brains of patients with Alzheimer's disease, ^[21] supporting the hypothesis of the accumulation of nonrepaired or misrepaired DNA damage as a component of underlying pathogenesis. Alzheimer's disease shares many features with IP ^[22].

4. Epidemiologic studies of spouses of patients with IP do not support the notion that protracted exposure to toxins or repeated (or chronic) infection causes disease progression (personal communication, L.I. Golbe).

Our findings are not in keeping with the suggestion that IP is the result of an accelerating pathogenesis, such as inflammation. In their study on the HLA-DR stained autopsy material, McGeer et al ^[23] suggested an increasing velocity of neuronal loss because their estimates of the rate of neuronal destruction were not compatible with the known natural history of IP. Attempts to estimate the duration of pathogenesis from postmortem findings in a neurodegenerative disorder are tenuous, and other evidence ^[1,10,24,25] militates against the accelerating neuronal death suggested by McGeer et al ^[23].

We have argued that aging provides a background of linearly declining nigral neurons that contributes to the progression of neurologic deficits in IP. This position is predicated upon the most recent reports of dopaminergic nigral cell loss ^[25] and striatal depletion of dopamine ^[26] in normal aging. We recognize that the pathology of IP is focused in the ventral tier within the nigra, ^[27] whereas the effect of aging is more diffuse. However, the attrition of neurons with senescence includes the ventral tier, so aging must be regarded as contributing to the total impact of deleterious forces contributing to the destruction of the nigrostriatal pathway in IP.

In conclusion, our findings, taken in the context of the relevant literature, suggest that neurons already damaged by an event ultimately undergo a premature death, leaving unaffected neurons with a normal expectation of survival. If a process were causing neuronal death, all the nigral dopaminergic neurons that are vulnerable should ultimately become equally affected, eventually leading to uniform pathology. However, we found no significant convergence toward such a pattern. The evidence suggests that the fate of the damaged cells is being sealed by an initial causal event through a change in the genome.

Acknowledgments.

We would like to thank the Medical Research Council of Canada, the National Parkinson Foundation (Miami), the Dystonia Medical Research Foundation (Chicago), the Parkinson Foundation of Canada, and the Movement Disorder Institute (Vancouver) for their continued support.