Dose-response relationship of levodopa with mood and anxiety in fluctuating Parkinson's disease

Motor fluctuations in Parkinson's disease (PD) commonly appear several years after initiation of levodopa therapy, and typically increase over the course of the disease. In some cases, these fluctuations appear to be related to levodopa dose schedules and seem to parallel plasma levodopa levels--the ``wearing-off'' pattern. In other cases, fluctuations have no apparent temporal relationship with levodopa dosing--the ``on-off'' pattern. By the fifth anniversary of their diagnosis, perhaps 50% of parkinsonian patients will develop motor fluctuations. ^[1]

Symptoms of depression are common in PD, ^[2] but these may not always constitute major depression. ^[3] Anxiety disorders ^[4] or mixed depressive and anxiety disorders ^[5] also contribute to psychiatric morbidity in PD. Less well appreciated is the fact that, like motor symptoms, depression and anxiety fluctuate in severity, ^[1,6,7] presumably as a consequence of levodopa dosing and thereby brain dopamine concentrations.

In an earlier, open-label study, we described the pattern and time course of mood and anxiety changes associated with a 1-mg/kg/hr infusion of levodopa in 15 patients with motor fluctuations and the temporal relationship of mood and motor fluctuations, but we could not exclude a placebo response. ^[8] This current study investigated the dose responsiveness of levodopa-induced mood and anxiety effects under double-blind conditions.

Methods.

Results.

The Figure 1 illustrates the time-effect relationships for mood, anxiety, tapping speed, and serum levodopa concentrations during the infusions. All responses were proportional to dose. For mood, the 1-mg/kg/hr infusion rate produced a greater mean effect than the 0.5-mg/kg/hr infusion rate and placebo infusion (eg, at 11 AM [high equals 56 plus minus 23, low equals 45 plus minus 21, placebo equals 34 plus minus 23]). There was also a shorter latency, a greater peak effect, and a longer duration with the 1-mg/kg/hr infusion. In turn, the low-dose (0.5 mg/kg/hr) infusion produced a greater mean effect than the placebo infusion. The effect size here was moderate (approximately 0.5 SD) for the low-dose comparison with both high dose and placebo; it was large (approximately 1.0 SD) for the high-dose comparison with placebo. Mood elevation with the high-dose infusion began promptly after the 9 AM rating. Response to the low-dose infusion was delayed a full hour from initiation of the infusion. The duration of the high-dose effect after stopping the infusion was approximately 2 hours longer than that of the low-dose effect. Of note, the last mood rating on the high-dose day dropped noticeably below the pre-infusion baseline levels--a possible indication of a withdrawal or rebound effect with high-dose but not low-dose infusion.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Relationship of mood, anxiety, tapping speed, and levodopa concentration before, during, and after three randomly assigned levodopa infusions (0.0, 0.5, and 1.0 mg/kg/hr) on each of 3 consecutive days. Each data point is the mean of eight individuals.

Six of eight patients had a mood response (an increase in mood score of greater than 20%) during the 1-mg/kg/hr infusion, four of eight had a mood response during the 0.5-mg/kg/hr infusion, and two of eight had a mood response during the placebo infusion. We compared the mean values of the five half-hourly ratings between 9:30 AM and 11:30 AM for each infusion condition using a repeated measures nonparametric analysis, the Friedman two-way ANOVA. The mean rank for high dose was 2.75, for low dose was 2.13, and for placebo was 1.13 (chi² equals 10.75, df equals 2, p equals 0.0046). Wilcoxon's matched-pairs signed rank test was used for pair-wise comparisons using the Bonferroni correction. High dose differed significantly from placebo (Z equals 2.52, p equals 0.016); the other pair-wise comparisons did not reach significance.

The Figure 1 also shows the relationship of anxiety with infusions. A reduction of anxiety began shortly after onset of the high-dose infusion but was delayed at least 1 hour with the low-dose infusion. The duration and magnitude of this apparent antianxiety effect was more developed with the high-dose infusion than the low-dose. Uniquely, the peak effect of anxiety reduction occurred 30 minutes after the infusion had been stopped and was followed by a precipitous increase in anxiety, analogous to the marked but transient decrease in mood after the drug effects dissipated.

In general then, although the anxiety reduction was very similar to the mood elevation, the dose-response curve appeared distinctly phase delayed for anxiety reduction as compared with mood elevation at both high and low doses. Despite the visible trends in the Figure 1, mean differences for anxiety between the high dose, low dose, and placebo state were not statistically significant.

Ratings by nurse observers of patient mood and anxiety paralleled those of the patients, as shown in the Figure 1 insert. These curves demonstrate similarities and dissimilarities with their patient-rated analogs. As with the patient ratings, both showed little evidence of response with placebo infusions, particularly during the 4 hours bracketing the infusion period. Both suggested a partial dose-response relationship, ie, the 1.0-mg/kg/hr infusion condition produced a more pronounced effect than the 0.5 mg/kg/hr infusion. However, the nurse-rated mood data points for the two nonplacebo infusions between 10 AM and 12 PM merit additional comment. Mood elevation with the 1.0-mg/kg/hr dose terminated abruptly after the 10 AM rating, dropped to a low at 10:30, then resumed a climb to the 12 PM zenith. The same temporary reversal of effect was seen in the 0.5-mg/kg/hr curve but with a narrower band. From the nurses' anxiety ratings, this same pattern was evident in the 1.0-mg/kg/hr infusion, but not the 0.5-mg/kg/hr infusion.

One possible explanation for this pattern is that the nurses were subjectively factoring into their ratings the ``negative'' contributions of increasing dyskinesia and agitation that they observed at the peak of the infusions, ie, that they viewed these as counteracting or discounting the beneficial (mood-elevating and anxiety-reducing) effects of the infusion. Although this may be an over-interpretation of the data, there was anecdotal clinical support for this explanation. Several patients, who appeared calm, rational, and emotionally stable when neither maximally ``off'' nor ``on,'' were clearly disorganized, hyperkinetic, and labile during the last hour of their infusions, when they would be perhaps better classified as ``on with dyskinesia.'' ^[6] Remarkably, the patients themselves seemed to have little insight into the discrepancy between how they reported they felt subjectively and how they appeared to observers in this dyskinetic and somewhat agitated, but relatively euphoric, state.

The Figure 1 also illustrates tapping rate and plasma levodopa concentration during the three infusion conditions. As expected, peak levodopa concentrations appeared at the end of the 2-hour infusion: 6.7 nmol/ml plus minus 3.9 with the 0.5-mg/kg/hr infusion and 12.5 nmol/ml plus minus 5.3 with the 1.0-mg/kg/hr infusion.

Tapping speed generally paralleled mood changes. Of note in the Figure 1, tapping speed continued to increase to a peak that occurred 90 minutes after the 1.0-mg/kg/hr infusion was discontinued.

Discussion.

The two main findings in this study are (1) levodopa affects mood and anxiety significantly more than a placebo infusion and (2) this effect is dose responsive. These emotional effects of levodopa infusion are prompt, relatively brief, and readily identified by patients and observers. As the effects decay after infusions, there may be rebound markedly below baseline levels. Mood changes and tapping speed were somewhat discordant, which argues that mood changes are not simply a consequence of improved motor function. The results of this study match the patterns in an earlier sample of 15 patients similarly examined under open-label conditions ^[8] comparing fluctuating patients when ``on'' and when ``off,'' in that depression and anxiety were greater when ``off.''

Our results are in agreement with Nissenbaum et al ^[7] who, in a study of ``on'' and ``off'' conditions, concluded that the worse the motor state the greater the depression and anxiety. We are in partial agreement with Menza et al ^[6] who reported that mood is better when ``on'' without dyskinesia, but we did not find a deterioration in mood when dyskinesia appeared. We suspect that mood is depressed when ``on'' with dyskinesia only when the dyskinesia is relatively severe and bothersome to the patient.

Mood and motor fluctuation research in PD has been hampered by problematic methodology. Our study has several advantages over previous studies. We measured mood and motor responses in fluctuators as they were experiencing them. As a result, it does not share the limitations of post hoc questionnaire data that depend upon the subject's memory of mood during past fluctuations. ^[7] We gathered data at multiple time points and display the time course of the effect. No previous study has employed placebo or looked for dose effect. Another strength of our design is the use of independent raters to corroborate the mood and anxiety ratings of the subjects.

This study confirms anecdotal patient and clinician observations that emotional status often changes predictably with levodopa administration. Presumably, the infused levodopa exerts the effect we show after conversion to intracerebral dopamine. This study suggests that at least momentary changes in depressive mood and anxiety may be a consequence of intracerebral dopamine depletion and repletion. Future studies are needed to better characterize the relationship between the emotional response of levodopa in PD patients and risk for subsequent psychiatric morbidity. The ultimate clinical goal of such studies is the amelioration of emotional suffering and reduction of disability.

Acknowledgments

We thank the research nurses of the Clinical Research Center for careful attention to protocols and the subjects for participating in the studies.