INTRODUCTION

Mood disturbance, and especially major depressive disorder, is a common condition in Parkinson's disease (PD), with an average prevalence of 25–40% in outpatient settings (Leentjens, 2004). Depression is associated with a reduced quality of life, increased functional disability, more severe cognitive symptoms, and enhanced caregiver's stress (Troster et al, 1995; Liu et al, 1997; Aarsland, 1999; Hobson et al, 1999). Although there is some knowledge about risk factors associated with depression in PD, little is known about its pathophysiology. The serotonergic hypothesis is one of the few hypotheses that have tried to link the pathophysiology of PD with an increased risk of depression (Mayeux, 1990). This hypothesis is based on three observations. The first is that serotonergic activity is reduced in PD. Indeed, degeneration of serotonergic nerve cells, decreased brain serotonin content, and alterations in the activities of various types of serotonin receptors, have all been demonstrated in post-mortem studies using neurochemical and autoradiographic techniques (Jellinger, 1991) (Scatton et al, 1983; Chen et al, 1998). Moreover, in vivo studies have consistently demonstrated reduced levels of 5-hydroxyindoleacetic acid (5-HIAA), a breakdown product of serotonin, in the cerebrospinal fluid (CSF) of PD patients (Johanson and Roos, 1967; Kuhn et al, 1996), with some studies reporting an additional reduction of 5-HIAA in depressed PD patients (Mayeux et al, 1984; Kostic et al, 1987). These findings show the involvement of serotonin in PD. The second observation is the finding in animal studies that serotonin has the ability to inhibit striatal dopamine release (Gerson and Baldessarini, 1980; Jenner et al, 1983; Jacobs and Fornal, 1993). This implies that reduction of serotonergic activity leads to less inhibition and a greater dopamine availability. The third is that a reduced serotonergic tone is a known a risk factor for depression (Van Praag and De Haen, 1979). Based on these observations Mayeux et al have formulated the serotonergic hypothesis of depression in PD. This hypothesis considers the reduced serotonergic tone a physiological adaptation to the reduced dopamine activity, while at the same constituting a risk factor for depression (Mayeux et al, 1984; Mayeux, 1990). Although there are attractive alternative hypotheses about the role of serotonin in PD, this hypothesis is appealing because it provides an explanation for some common clinical observations. The presence of this biological risk factor for depression may explain the high prevalence of this condition in patients with PD (Leentjens, 2004). It may also explain the increased incidence of depression preceding the diagnosis of PD, because of the fact that pathophysiological compensatory mechanisms are already in action long before clinical symptoms become apparent (Leentjens et al, 2003b). Finally, it may provide an explanation for the exacerbation of extrapyramidal symptoms that occur in some PD patients treated with selective serotonin reuptake inhibitors (SSRIs) (Leo, 1996; Gerber and Lynd, 1998). In spite of the fact that this hypothesis was launched in 1984, to date it has not been experimentally verified.

The aim of this study is to test the serotonergic hypothesis of depression in PD in an experimental approach using the acute tryptophan depletion (ATD) paradigm.

PATIENTS AND METHODS

Subjects

In all, 15 consecutively referred eligible patients with PD, as defined by the United Kingdom Parkinson's Disease Society Brain Bank (UK-PDS-BB) criteria, were included in the study (De Rijk et al, 1997). Criteria for exclusion were the presence of concomitant neurological disorders other than PD, and the presence of concurrent psychiatric disorders, notably major depressive disorder and dementia, as defined by the criteria of the DSM IV (American Psychiatric Association, 2000a). The presence of these psychiatric disorders was assessed in a clinical interview. A prior personal or family history of major depressive disorder, as defined by DSM IV criteria, were also considered exclusion criteria. This was carried out with the intention of excluding patients with pre-existing risk factors for depression that are not associated with PD-related neurotransmitter changes. A prior personal or family history of depression are known independent risk factors for depression in the general population, that also play a role in patients with PD, and inclusion of patients with these risk factors would confound the recognition of a specific vulnerability related to the pathophysiology of PD (Leentjens et al, 2002). For the same reason, as well as to ensure optimal cooperation with and reliability of the procedure, patients with dementia, as well as those with a score lower than 23 on the Mini Mental State Examination (MMSE), were excluded (Folstein et al, 1975). Patients currently using psychopharmacological medications could not participate. Although ideally only drug-naïve, de novo patients should be included, this was not considered feasible. For pragmatic reasons the use of antiparkinsonian medication as such was not considered a ground for exclusion, except for preparations with a known strong interference with the serotonergic neurotransmitter system, such as levodopa preparations, lisuride, and selegeline. Patients on stable doses of dopamine agonists or anticholinergics were included. No patients were taken off any medication for the sake of the study. The PD patients were individually matched for sex, age, and educational level with healthy control persons from an existing bank of volunteer subjects of the Institute of Brain and Behaviour of Maastricht University, to which the same exclusion criteria were applied.

Prior to participation, the general health of the subjects was ascertained by physical examination, screening blood tests, and an electrocardiogram. Moreover, an MMSE and a Hamilton Depression Rating Scale (HAMD) were administered, to assess the patients’ cognitive and affective status (Hamilton 1960). If these investigations revealed additional grounds for exclusion, the patient could not participate. Finally, the patient was classified according to the Hoehn and Yahr (H&Y) staging system, in order to describe the global severity of PD (Hoehn and Yahr, 1967).

Patients were given a verbal explanation and written information of the study and the procedure. All participants gave their written informed consent. Our hospitals’ Medical Ethics Committee approved the study.

Design

The experiment was designed as a double-blind, placebo-controlled, randomized order, crossover study. Every subject underwent the ATD procedure twice: once with an amino-acid mixture without tryptophan (the active condition), and once with an amino-acid mixture containing a balanced amount of tryptophan (the placebo condition). These interventions were performed in a randomized order and spaced at least 1 week apart in order to exclude carry-over effects. Thus, both within-subject comparison between the active and placebo condition, as well as a comparison between PD patients and controls became possible.

Amino-Acid Mixtures

The preparation and composition of the amino-acid mixture was the same as described by Klaassen et al (1999b) and Riedel et al (1999) In the active condition, 3 g/100 g tryptophan was left out. Apart from the amino acids, the mixture contained 63 g carbohydrates and 33 g fat in order to dissolve the amino acids and provide caloric value. On both test occasions, the subjects ingested 75 g of amino-acid mixture, dissolved in 250 ml of water.

ATD reduces the availability of tryptophan, the precursor of serotonin, in two ways. Firstly, protein synthesis is stimulated, which uses circulating tryptophan and reduces serum tryptophan levels. Secondly, tryptophan competes with the large neutral amino acids (LNAAs: valine, leucine, isoleucine, phenylalanine, and tyrosine) for active transport over the blood–brain barrier, which results in less tryptophan entering the brain. Both plasma concentrations of tryptophan start falling 2 h, and CSF levels 2.5 h after ingestion of the amino-acid mixture, and reach a minimum after 5–7 h (Carpenter et al, 1998; Williams et al, 1999). As continuous supply and synthesis of serotonin in the brain is necessary to maintain adequate levels of serotonergic transmission, ATD creates a temporary deficiency of serotonin. The clinical and physiological consequences of ATD can be followed over a period of several hours. After the intervention, tryptophan levels quickly return to normal upon return to a normal diet (Klaassen et al, 1999b).

Procedure and Measures

At the days of the intervention, the subjects fasted from midnight. Upon arrival at our department at 0900 a baseline amino-acid spectrum was obtained, and the 32 item abbreviated Dutch version of the Profile of Mood States (POMS) questionnaire was administered (McNair et al, 1971). This version of the POMS is a measure of mood states that assesses five different qualities of mood: sadness, tension, anger, vigor, and fatigue on a 100 mm visual analogue scale. Lower scores indicate higher symptom levels. As a result of its sensitivity for transient mood changes, this questionnaire is more appropriate, and often used, in acute interventional studies than depression scales such as the HAMD or Beck Depression Inventory (BDI), because these last two are designed to measure mood over a longer period of time (typically 1 or 2 weeks) (Hamilton, 1960; Beck et al, 1961). Patients were allowed 30 min for ingestion of the amino-acid drink. At 3 and 6 h after the start of ingestion, the POMS was taken again. Subjects were provided a light protein-free lunch at noon.

Power Calculation and Statistical Procedures

As this is the first study using the ATD paradigm in PD patients a formal power calculation based on previously reported findings of effect sizes and standard deviations (SD) could not be performed. Hence, sample size calculation was based on the effect sizes and SD results of two earlier tryptophan depletion studies involving healthy volunteers, conducted in our department (Klaassen et al, 1999a, 1999b). A sample size of 15 persons per group would be adequate for detecting a difference of 1 SD with a power of 80% in a between-group analysis. As a result of two reasons we expected that with this calculation the actual power in our study would be on the safe side. First it was expected that the effect sizes (but possibly also the SD) in PD patients would be greater than in healthy control subjects because of the hypothesized greater vulnerability of PD patients. Second, sample size calculation was based on a between-group comparison only, and addition of a within-group comparison within the same multivariate analysis of variance (MANOVA) would further increase the power.

A planned interim analysis was performed after inclusion of five patients and five control subjects in order to evaluate whether ATD would be a feasible intervention in PD, and to ascertain adequate levels of tryptophan depletion (Leentjens, 2002).

Demographic variables were compared by Student's t-tests. Testing was always carried out two-tailed with the level of significance set at 0.05. The score on the POMS sadness subscale was considered the primary outcome measure. As serotonin is also associated with other disturbances in mood, the other POMS subscales were considered secondary outcome measures. Using a multiple repeated measures MANOVA, the effects of disease, disease by intervention, and disease by intervention by time were analyzed for all POMS subscales. All calculations were performed with the Statistical Package for the Social Sciences (SPSS), version 10.0 (SPSS Inc., Chicago, 1998).

RESULTS

Participants

Nine male and six female patients with an average age of 61.9 years (SD 7.51) were included in the study. Their average MMSE score was 28.9 (SD 1.4), and their average HAMD score was 2.1 (SD 1.4). The median H&Y stage was II (average stage 2.4, SD 0.6, range I–III). Two patients were medication-free de novo patients; the other patients were using an NMDA antagonist (n=1), dopamine agonists (n=4) or a combination of both (n=8). Control subjects were individually matched for sex, age, and educational level. Thus, there were no significant differences in age and MMSE score. There was however a significant, albeit not clinically relevant, difference in HAMD score between PD patients and controls that was due to higher scores of PD patients on some of the somatic symptoms of this scale. Neither patients nor control subjects had a prior personal or family history of depression. The demographic characteristics are summarized in Table 1.

Table 1 Demographic Characteristics of the Participating Subjects
Full size table

Procedure

All subjects tolerated the procedure well. At the time of the interim analysis, there were no gross clinical changes in motor, affective, and cognitive symptomatology that would make continuation of this study unethical. Apart from the nausea, that occurred in four patients and two control subjects, there were no adverse events. In both PD patients and controls a significant reduction of serum tryptophan was achieved during the active procedure, but not during the placebo procedure. After 3 h, the ratio of tryptophan to other LNAA (trp/LNAA) in patients had fallen with 85%, from 7.2 to 1.1%. In control subjects the trp/LNAA ratio had fallen with 65% from 7.4 to 2.6% in the active condition. These rates of depletion were maintained until after the second set of measurements at 6 h. In the placebo condition there was a slight increase in trp/LNAA ratio. The effect of ATD is visualized in Figure 1.

Figure 1
figure 1

Tryptophan/large neutral amino acids (LNAA) ratio during the ATD procedure and the placebo condition in 15 PD patients and 15 matched control subjects, at baseline, t=3 and 6 h.

Full size image

Mood

The scores on all POMS subscales for patients and control subjects during the procedure are tabulated in Table 2. PD patients scored significantly lower (worse) on three of the subscales of the POMS: sadness, fatigue, and vigor, indicating a significant ‘disease effect’ (for depression F=6.49, df=1,28, p=0.017; for fatigue F=4.88, df=1,28, p=0.035; for vigor F=5.82, df=1,28, p=0.020). No significant ‘disease effect’ could be found for hostility and tension (for hostility F=3.09, df=1,28, p=0.089; for tension F=0.92, df=1,28, F=0.346). There was no significant ‘disease by intervention’ effect, nor a ‘disease by intervention by time’ effect for any of the POMS subscales. This means that for both patients and control subjects there was no within-group effect of ATD on POMS scores, nor was there any difference in response on the POMS between the two groups.

Table 2 Mean Scores and Standard Deviations (SD, between Brackets) Given on 0–100 mm Visual Analogue Scales for all Subscales of the POMS, of 15 PD Patients and 15 Control Subjects During the Placebo Condition and after ATD at t=0 and 6 h
Full size table

DISCUSSION

ATD Procedure

ATD has been extensively used in psychiatry to study the role of serotonin in aspects of mood regulation and cognition. Most studies report mood lowering effects of ATD in patients at risk of depression, such as patients with a mood disorder in remission, or a family history of depression or bipolar disorder (Delgado et al, 1990; Åberg-Wistedt et al, 1998; Klaassen et al, 1999a; Sobczak et al, 2002). In vulnerable individuals, ATD also exacerbates anxiety, panic, and aggression (Kent et al, 1996; Klaassen et al, 1998; LeMarquand et al, 1998).

In PD patients ATD has so far not been used to study the role of serotonin in mood or other symptom areas. Perhaps this is due to the negative experience of ATD in the only case history of a PD patient described so far (McCance-Katz et al, 1992). In this case history, the patient experienced a serious exacerbation of motor symptoms as well as emergence of depressive symptoms and a significant bradyphrenia occurred during the ATD procedure. All of these symptoms resolved within 2 h of discontinuing the testing and ingesting a meal containing a tryptophan supplement. In our study, none of these adverse events occurred and adequate levels of tryptophan depletion were achieved. Our experience is that ATD is a feasible paradigm to assess serotonergic function in PD patients.

Mood

In spite of lower baseline scores on the ‘sadness,’ ‘fatigue,’ and ‘vigor’ subscales of the POMS, there was no differential effect of ATD and the placebo condition within each group, nor was there a difference in response to the interventions between the two groups. With respect to mood, this is in contrast with earlier studies that have used the POMS during ATD to assess mood changes in non-PD subjects at risk of depression because of a positive personal or family history of depression. In our study, subjects with these known risk factors were excluded in order to study a potential risk of depression that would be specifically attributable to PD. We could find no differential responses between PD patients and control subjects that would support such a specific serotonergic vulnerability for depression.

Several potential explanations for these negative findings should be considered. A first possibility would be a possible underpowerment of the study. The difficulties in performing an adequate sample size calculation were discussed in Patients and methods. However, estimation of requested sample size was performed on the basis of two earlier studies. These studies showed that the POMS is sensitive enough to detect differential responses between study groups in vulnerable individuals even with a lower number of included subjects (Klaassen et al, 1999a, 1999b). Another possible explanation may be the existence of a floor effect. It may not be possible to further lower serotonergic activity, and thus elicit mood symptoms, in persons with an already diminished serotonergic function, such as is the case in patients with PD. A similar explanation was given by Delgado et al who reported no additional mood changes during ATD in untreated depressed patients (Delgado et al, 1994). However, our PD patients were not depressed and thus lowering of mood would be possible as a reaction to ATD. Yet another explanation is the fact that ATD is a method that is especially suitable to demonstrate presynaptic serotonergic dysfunction, while being less sensitive to demonstrate postsynaptic dysfunction. In the case of postsynaptic serotonergic dysfunction the postsynaptic cells may already be less responsive to serotonin anyway (Delgado et al, 1994). There is some evidence for postsynaptic serotonergic dysfunction in PD. Three interventions assessed serotonergic function in PD patients with different serotonin agonists. Blunted cortisol, ACTH, and prolactine responses to a fenfluramine challenge, and a blunted growth hormone response to a 5HT1A receptor challenge with sumatriptan have been reported (Kostic et al, 1996; Volpi et al, 1997a, 1997b). These studies are all indicative of a defective serotonergic control of the hypothalamic–pituitary–adrenal (HPA) axis in PD patients. A limitation of these interventions is that in a design where agonist substances are used, which enhance serotonin availability, it is not possible to elicit mood symptoms. Hence these designs are not suitable to study the serotonergic hypothesis of depression in PD. With the exception one pilot study that assessed postsynaptic 5HT2a receptor binding in PD, only presynaptic parameters, such as the 5HT transporter (5HTT) and the 5HT1a receptor have been studied. In the pilot study 5HT2a receptors were differentially increased and decreased in different brain regions that could not be linked to depression (Van Kroonenburgh et al, 2001).

If we accept that these explanations are unlikely, the only feasible explanation is that, contrary to other vulnerable groups, the known vulnerability for depression in PD patients is not directly related to the reduced serotonergic activity.

Other limitations of this study that ought to be mentioned are the fact that the allowed medication may still have played a confounding role, and the fact that there are no data on validity or reliability of the POMS in patients with PD.

Implications

In our study, the vulnerability of PD patients for depression cannot be directly linked to a reduction in serotonin activity. It can also be hypothesized that serotonin plays a more indirect role as a regulator of other neurotransmitters involved in the pathophysiology of depression, such as dopamine and noradrenalin. Although in men most studies point at an antagonistic interaction between serotonin and dopamine, animal studies also support an agonistic interaction (Scholtissen et al (accepted)). Some authors suggest that in PD serotonergic degeneration may be primary, and the degeneration of the dopamine system secondary (Steinbusch and De Vente, 1997; Braak et al, 2003). Moreover, other neurotransmitters may be more directly related to mood symptoms in PD, such as dopamine. The ‘dopaminergic hypothesis for depression in PD,’ was formulated by Fibiger in the same year as the serotonergic hypothesis was formulated (Fibiger, 1984). He considers the reduced responsible for the high incidence of depression. Deficiency in this system would lead to malfunction of self-reward systems that would constitute a risk for depression. This hypothesis would also provide an explanation for the high prevalence of depression in PD and the fact that depression may precede PD. It also provides an explanation for the beneficial effects on mood of some of the dopamine agonists, and for the problem of dopamine dependence that exists in some patients(Corrigan et al, 2000; DeBattista et al, 2000; Lawrence et al, 2003; Goldberg et al, 2004). Although formulated in 1984, this hypothesis too is still in need of experimental verification.

The pathophysiological basis of depression in PD may also influence the clinical approach to treatment. In clinical practice the treatment of depression in PD patients, is largely focused around selective serotonin reuptake inhibitors (SSRIs) or atypical agents such as venlafaxine and mirtazepine. In the ‘Practice guideline for the treatment of major depression in adults’ of the American Psychiatric Association (APA) SSRIs are mentioned, alongside bupropion, as a first choice treatment for depression in PD (American Psychiatric Association, 2000b). This advice is based on side-effect profiles, but not on efficacy or pathophysiological arguments. The two placebo-controlled trials with an SSRI in PD are characterized by a high placebo response without superior efficacy of citalopram or sertraline, respectively (Wermuth et al, 1998; Leentjens et al, 2003a). In a Cochrane review no evidence was found for superior efficacy of any antidepressant over placebo in depressed PD patients (Ghazi-Noori et al, 2004). If serotonergic deficiency is not the main pathophysiological mechanism in depression in PD, it may be worthwhile to look at other potential treatment options, including agents that more specifically address the noradrenergic and the dopaminergic system.

Conclusion

ATD is a feasible research method to assess central serotonergic function in PD. Adequate levels of tryptophan depletion were achieved without clinical adverse effects. No differential response on the POMS subscales were observed between the active and placebo condition within each group, nor was there any difference during the active condition between PD patients and controls. Although serotonin is clearly involved in the pathophysiology of PD, our study did not find evidence for a direct relation between serotonergic activity and mood symptoms. Thus, this study provides no support for the serotonergic hypothesis of depression PD. In this light, alternative hypotheses, such as the ‘dopaminergic hypothesis,’ would merit experimental investigation. In the clinical practice of treating depressed PD patients, nonserotonergic antidepressants may be interesting treatment options that should be further explored.