Quantifying postural tremor in workers exposed to low levels of manganese

https://doi.org/10.1016/j.jneumeth.2004.05.002Get rights and content

Abstract

The aim of this study was: (1) To determine the minimum number of characteristics necessary to discriminate between postural tremor recorded in control subjects (CO), in subjects exposed to manganese (MN), and in patients with Parkinson’s disease (PD), and (2) to examine the continuum of changes between the three groups examined.

Workers previously exposed to Mn (n = 10), patients with PD (n = 10), and control subjects (CO) (n = 11) underwent a clinical examination. Blood Mn was measured at the end of exposure time for the MN group and 12 months later at the beginning of the experiment for all groups. Postural tremor with visual feedback was recorded in the index finger with a laser system. Statistical criteria were used to reduce computed tremor characteristics to a minimal set of reliable discriminating variables.

Two variables were retained namely corrected wobble (CW), describing the morphology of the tremor oscillations, and variability ratio (VR), describing proportional power of tremor. Both variables had an overall correct classification rate of 77.4%. Blood Mn levels at the time of the experiment were similar for all groups and had insignificant correlation with tremor variables. However, blood Mn levels in workers which were also measured at the end of exposure time (i.e., 12 months before) showed significant correlation (Spearman’s rank coefficient) with both harmonic index (ρ = 0.70, P = 0.03) and first maximum of the autocorrelation function (ρ = 0.89, P = 0.001).

We conclude that (1) the tremor of workers exposed to Mn could be adequately described with only two variables; (2) a continuum of changes between tremor recorded in control subjects, in subjects exposed to Mn and in patients with PD was observed, with the MN group always found in between the control (CO) and the PD groups; (3) while blood Mn levels in workers were back at control levels at the time of the experiment, the effect of Mn on postural tremor was still detected. Thus our method has the potential to detect the effect of Mn on tremor with only two variables even after Mn level in the blood is back to normal values.

Introduction

During early progression of neurodegenerative disorders, such as Parkinson’s disease (PD), subtle changes in motor functions may develop which are consistently observed across movement disorders. For example, when tremor is affected, its amplitude and within-subject variability increase, its median frequency shifts toward lower values and its power is reorganised over new intervals in the power spectrum. A number of characteristics developed by Beuter and Edwards, 1998, Beuter and Edwards, 1999 and Edwards and Beuter, 1999, Edwards and Beuter, 2000 can be used to detect these changes validly, reliably even if they sometimes are not detected during the clinical examination. Indeed, it is known that around 25–30% of PD patients, for example, do not develop a perceptible tremor (Beuter et al., 1999). However, while amplitude remains within normal values in these patients the morphology of their tremor has been shown to be different from that of control subjects (Edwards and Beuter, 1999).

One question that has been left relatively unexplored is the number of variables necessary to discriminate between normal and slightly abnormal tremors. For example, according to a recent study (Beuter and Edwards, 1999) the proportion of the power spectrum contained between two frequency ranges (7–12 Hz and 4–6 Hz) also called proportional power is a reliable characteristic for discriminating between normal and abnormal (PD) tremors. Finding out the smallest number of discriminating variables could be helpful in field studies in which many subjects must be tested in a short time and the changes, often subtle and fluctuating, must be detected reliably.

To explore this question, two conditions (i.e., Parkinson’s disease and Mn exposure) which are known to affect tremor, were selected. Recently long-term exposure to manganese was associated with a continuum (i.e., a sequence of ordered changes forming a consistent and systematic progression) of neurological dysfunction including early subtle changes in tremor at lower manganese exposure levels and increased tremor amplitude at higher levels of exposure (Beuter et al., 1999, Mergler et al., 1994, Mergler et al., 1999). In a community based study (Beuter et al., 1999), hand postural tremor measured with an accelerometer appeared to be more regular with higher Mn blood levels as indicated by a lower standard deviation of the median frequency and a higher harmonic index (Table 1). This last measure quantifies how the spectrum concentrates around peaks. According to Mergler et al. (1994), Mn intoxication follows three stages and advanced manganism is clinically similar to advanced PD. Note that “manganism” is the medical term used in the literature to describe the effects of a chronic exposure to Mn (see also Mergler and Baldwin, 1997). Manganism is associated with a degeneration of the globus pallidus (Clarkson et al., 1988) and PD with a degeneration of the nigrostriatal system (Mergler et al., 1999). In the continuum of neurological dysfunction, scores of Mn exposed subjects should be systematically found in between those of controls (CO) and those of patients with PD.

Thus, the first objective of the present study was to identify variables discriminating postural tremor in workers previously exposed to manganese (MN) from those in control subjects (CO) and patients with Parkinson’s disease (PD). Postural tremor was used because it has been shown to produce more discriminating results than rest tremor between CO and PD subjects in the frequency (Beuter and Edwards, 1999) and time (Edwards and Beuter, 2000) domains. In the present study it is defined as an involuntary oscillatory behaviour of limb extremities while maintaining a posture using visual feedback. By providing visual feedback we created a compensatory tracking task. As a consequence drifting of the finger was controlled by the subject using visual feedback and this avoided the occurrence of saturations occasionally caused by drift out of the range of the recording system. Ten identified characteristics (see Table 1) found valid, reliable and sensitive in previous tremor studies (Beuter and Edwards, 1998, Beuter and Edwards, 1999) and blood Mn levels were evaluated for this purpose. A second objective was to examine the continuum of changes between the three experimental groups. It was hypothesised that the MN group would be placed on the continuum line between the CO and the PD groups. Therefore, using tremor measures, we examined if the scores of the three subgroups lay on a continuum going from normal values (CO) to abnormal values (PD) with intermediate values for the MN group.

Section snippets

Subjects

Ten male workers previously exposed to Mn in a ferromanganese alloy production plant were tested. The Mn produced is used for the manufacturing of steel. Other uses include the production of nonferrous alloys, dry cell batteries, fertilizers, dyes, catalysts, and glass, etc. (Clarkson et al., 1988). In addition 10 patients with idiopathic PD, and 11 control subjects (CO) were recruited for the study. Mean age was 49.8 years (S.D. = 3.7) for the workers, 59.5 years (S.D. = 2.4) for the patients

Using the within-subject variability ratio (VR) for classification

We used the estimated standard deviations of proportional power between the two trials in the frequency range 4–6 Hz (SDP4–6) and in the frequency range 7–12 Hz (SDP7–12) to discriminate MN tremor from control and Parkinsonian tremors (Beuter and Edwards, 1999) (see also Section 1 and Fig. 2, Fig. 3). The Spearman rank correlation between SDP7–12 and SDP4–6, was not significant (ρ = 0.06, P = 0.72). The box and whiskers plots (Campbell, 1987) for SDP4–6 and SDP7–12 are presented in Fig. 2, Fig. 3

Discussion

We have shown that it was possible to identify a small number of principal variables adequately discriminating the postural tremor of workers formerly exposed to MN. This was done using a larger set of reliable and sensitive tremor characteristics (Beuter and Edwards, 1999, Edwards and Beuter, 2000) and statistical methods based on stepwise discriminant analyses. The two variables identified (i.e., corrected wobble—CW and variability ratio—VR) correctly classified 70.0 and 77.4% of the

Acknowledgements

The authors wish to thank all the participants who gave their time so willingly, Prof. Donna Mergler who made this study possible, and FCAR of Québec and NSERC of Canada for financial support.

References (21)

There are more references available in the full text version of this article.

Cited by (0)

View full text