Main

Depression in Parkinson disease (PD) is common, affecting up to one-third of patients1, and is associated with greater disability2 and increased mortality3 and has a greater negative impact on health-related quality of life than motor symptoms4. As depression in PD often occurs early in the condition and predicts increased caregiver burden5, greater impairment in activities of daily living6 and higher costs of care7, effective treatment of depression in PD has the potential to achieve important health and economic benefits. However, current treatment guidelines for depression in PD advise the same approach as for depressed patients with other long-term conditions, despite evidence suggesting that standard antidepressant drugs are ineffective for these patients8,9.

Mood changes in PD are frequently associated with motor fluctuations, or ‘on/off’ states, that begin as end-of-dose deterioration of the effect of dopaminergic medications and later progress to unpredictable fluctuations10. One study of nonmotor fluctuations in PD found that two-thirds of patients experience depressed mood exclusively during off states11. This raises the possibility that depression in PD may be related to dopaminergic deficit and have a specific etiology, explaining why treatment recommendations for depression may not generalize to PD10.

Depression is a heterogeneous and etiologically complex syndrome. There are at least 256 possible unique symptom profiles that meet the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, (DSM-V) criteria for a diagnosis of major depressive disorder12. This degree of clinical heterogeneity has stimulated efforts to define subtypes based on symptom profiles12. In PD, motivational symptoms of depression such as apathy (diminished initiation of and engagement in activities) and anhedonia (inability to experience pleasure and loss of motivation to act in order to seek pleasure) are particularly prevalent, occurring in 40% and 46% of patients, respectively13. Motivational symptoms are also particularly challenging to treat. The ‘interest–activity’ symptom dimension in depression that includes loss of interest, diminished activity, fatigue and difficulty making decisions has been associated with poor outcome of antidepressant treatment in large prospective clinical studies14. People with PD report worse apathy and anhedonia when off dopaminergic medication10,15 and loss of dopaminergic innervation of the striatum, as measured by single photon emission computed tomography (SPECT) dopamine transporter (DAT) imaging, is associated with emergent motivational symptoms in PD16.

Dopamine activity plays a crucial role in goal-directed behavior, signaling how much better or worse an event is than predicted (a ‘reward prediction error’), as well as the valuation of reward and effort costs of actions17,18,19. The effect of dopamine signaling depends on the dynamics of its release and clearance from the synapse20. Short-latency phasic firing of dopaminergic neurons in the striatum is thought to encode reward prediction errors, crucial for reinforcement learning, while tonic levels of activity are thought to signal average reward valuation21. Understanding how dopamine signaling regulates adaptive behavior and motivation at different timescales and in different brain regions has important implications for the mechanisms underlying motivation and different depressive symptoms, and their diverse responses to different dopaminergic agents17,20,21.

Prescription of different dopaminergic medications in PD is currently largely driven by consideration of motor symptoms and side-effect profile, including impulse control disorders that most commonly occur with dopamine agonists. However, dopaminergic medications are also known to have antidepressant effects. Double-blind randomized controlled trials have shown that dopamine agonists can improve depression and apathy in patients with PD, and pramipexole, a relatively selective D3 receptor agonist, has shown promise as a treatment in chronic and severe depression in patients without PD22,23. Lab-based studies in both animals and humans utilizing dopamine depletion have consistently implicated mesolimbic dopamine activity as a key modulator of motivation24. However, it remains unclear whether the antidepressant effects of dopamine agonists are primarily caused by improvement in motivational deficits rather than mood. Therefore, we aimed to perform the first comprehensive analysis of the dimensional symptom predictors of antidepressant effects of dopamine agonists.

Other treatments, such as monoamine oxidase inhibitors (MAO-Is), are also clinically used for both their antidepressant and antiparkinsonian effects25. Selective type-A monoamine oxidase (MAO-A) inhibitors, such as moclobemide, are primarily used in depression to prevent the metabolism of serotonin and noradrenaline25. In contrast type-B monoamine oxidase (MAO-B) inhibitors, such as rasagiline and selegiline, were developed for treatment in PD owing to their ability to inhibit dopamine metabolism, thereby increasing striatal dopamine levels25. However, MAO-B inhibitors such as selegiline have also been shown to improve depression in PD26,27 and, at high doses, they are nonselective, additionally affecting serotonergic transmission, which may represent a complementary pathway to improving mood25,28. There remains limited understanding of the effects of MAO-Is on different symptom dimensions of depression in PD, and no study has examined how these effects change over time or their relationship with progression of neurodegeneration within dopaminergic pathways.

In this Article, we hypothesized that there would be dissociable effects of dopaminergic medications on different depressive symptom dimensions. Considering lab-based findings of the behavioral effects of dopamine agonists, we predicted that dopamine agonist treatment would improve motivational symptoms but not other depression symptom dimensions, whereas MAO-B inhibitors may improve mood due to their concurrent serotonergic action at higher doses29. Our secondary hypothesis was that dopaminergic medications with mechanisms of action reliant on presynaptic dopamine neuron integrity, such as MAO-B inhibitors, would have attenuated antidepressant effects as presynaptic dopaminergic neurodegeneration progressed.

Results

Participant characteristics

In total, 412 participants with PD were included at baseline with loss to follow-up of one-quarter of participants by year 5 (Table 1). Over half of patients had commenced PD medication by the end of year 1. Levodopa, dopamine agonists and MAO-B inhibitors were comparable in the prevalence of their use in year 1, but, by year 5, levodopa was the most common class of PD medication, with over 80% of patients prescribed this drug.

Table 1 Characteristics of PPMI participants with PD
Full size table

Total depression symptom scores and dimension scores both increased over time. However, only 13% of patients reported a 15-item geriatric depression scale (GDS-15) score >5 at baseline (suggestive of clinical depression) and less than 10% of participants with PD were taking antidepressant medication across all years of follow-up, though the reason for antidepressant use treatment was not available.

In almost all (98.8%) patients, DAT imaging was available at baseline and in years 1, 2 and 4, with only 17 participants imaged in years 3 and 5. As reported previously in this sample, DAT specific binding ratio (SBR) in participants with PD was, on average, around half that of healthy controls, and as expected, there was evidence of a marked decline over time (mean ± standard deviation (s.d.) percentage reduction from baseline: year 1, −9.7 ± 17.4%; year 2, −16.6 ± 17.7%; year 4, −26.6 ± 18.4%)16.

Depression symptom dimensions and medication class

We constructed longitudinal linear mixed-effects models to examine the relationship between antiparkinsonian medications and two different depression symptom dimensions within the GDS-15: a three-item ‘motivation’ factor30 and a nine-item ‘depression’ factor1). No other significant interactions with time were observed for other treatments using this cutoff and other treatments (Supplementary Table 1).

Depression symptom dimensions, medication and striatal DAT binding

Mixed-effects model analysis of the relationship between striatal DAT SBR and depression symptoms across all time points revealed that the effect of MAO-B inhibitor treatment on motivation was moderated by striatal DAT SBR (β = −0.23, 95% CI −0.39 to −0.07, P = 0.006) (Table 3 and Fig. 3). Patients with PD with higher striatal DAT SBR appeared to experience a greater antidepressant effect of MAO-B inhibitor treatment. A logistic mixed modeling analysis also found that striatal DAT SBR moderated the effect of MAO-B inhibitor treatment in reducing the risk of develo** clinically significant apathy/anhedonia (motivation factor score of ≥2). MAO-B inhibitor treatment significantly reduced the odds of develo** high apathy/anhedonia specifically in patients with higher striatal DAT SBR (odds ratio 0.28, 95% CI 0.11 to 0.73, P = 0.009) (Supplementary Table 2).

Table 3 Adjusted mixed-effects model results investigating the relationship between PD medication and depression symptom dimension by striatal DAT SBR
Full size table
Fig. 3: Relationship between ‘motivation’ factor score, MAO-B inhibitor treatment and striatal DAT binding.
figure 3

Adjusted linear mixed-effects model simulation showing ‘motivation’ factor score by striatal DAT SBR and MAO-B inhibitor treatment (n = 408, total of 1,406 scans across all six time points).

Full size image

Striatal DAT SBR did not significantly moderate the effects of dopamine agonist on individual depression symptom dimensions.

Discussion

We performed the first longitudinal analysis of how different types of dopaminergic medication for PD are associated with different symptom dimensions of depression and how these relationships are influenced by neurodegeneration of dopaminergic pathways. We found dissociable associations of PD dopaminergic medications with different dimensions of depression. Dopamine agonist treatment was associated with relatively lower motivation symptoms beyond the second year following diagnosis, but not depressed mood. In contrast, MAO-B inhibitor treatment was associated with fewer depression, but not motivation, symptoms. However, MAO-B inhibitor use did appear to improve motivation symptoms in patients with relatively preserved striatal presynaptic dopamine projections, indexed by higher striatal DAT binding. A potential explanation for this pattern is that the MAO-B inhibitor mechanism of action is dependent on presynaptic striatal dopamine neuron integrity, whereas other medication classes, such as dopamine agonists, act postsynaptically. Our results support existing evidence that striatal dopamine dysfunction plays a crucial role in motivation and highlight the need for further understanding of the effect of different dopaminergic therapies on depressive symptoms as PD pathology progresses over time16.

Optimization of dopaminergic therapies is often the first strategy in the treatment of depression in PD32. However, at present, the choice of dopaminergic therapy is largely guided by motor symptoms and potential side effects33. Our findings suggest that depressive symptom profile may also have a role to play in guiding dopaminergic treatment therapy selection. For example, dopamine agonist treatment may be an effective strategy to treat motivational symptoms beyond 2 years following diagnosis, while alternative therapies such as MAO-B inhibitors may be indicated in patients with prominent depressed mood.

Though our study was conducted in patients with PD, these results also reveal mechanistic insights and potential treatment strategies for depression in patients without PD, specifically for those with prominent motivational symptoms that are resistant to standard antidepressant medications14. The dopamine agonist pramipexole is currently used as a treatment for refractory depression, and our findings support the potential for further repurposing of PD dopaminergic therapies for treatment of motivational symptoms34.

Dopamine agonists have been identified as an effective treatment of disorders of motivation in PD previously35. However, the role of dopaminergic therapies on motivation and mood depends on dopaminergic receptor profile, pharmacodynamics, brain region and progression of underlying PD pathology21. For example, an animal study comparing D1-, D2- and D3-specific agonists in rescuing motivational deficits induced by lesions to the substantia nigra found that only D3 agonists were effective36. D3 receptors have a narrow distribution and are primarily located within the ventral striatum, a region implicated in reward processing and the cognitive mechanisms underlying motivation. Agonism of D3 receptors has been identified as a potential treatment of apathy in PD, and further research is needed to investigate whether dopamine agonists with high affinity for D3 receptors, such as pramipexole, are more effective options for the treatment of the motivational symptoms of depression15,24.

Improved understanding of the dynamics of dopamine neurotransmission underlying mood and motivation may also guide novel treatment strategies for depression in PD. Optogenetic studies have confirmed that phasic midbrain dopamine firing encodes reward prediction errors, which are crucial for learning, whereas tonic dopamine signals encode reward valuation and effort costs that drive motivation18,37. Dopaminergic therapies modulate the complex dynamics of dopamine signaling via inhibition of synaptic clearance mechanisms, post-synaptic agonism or potentiating synaptic release. Owing to the variation in metabolism, signaling and receptor distribution, manipulating dopamine can have paradoxical consequences for cognitive processing, often depending on basal levels of dopamine in different brain regions21. For example, dopamine synaptic clearance varies substantially across different brain regions. In the ventral striatum, rapid recycling via DAT predominates20. In contrast, in the prefrontal cortex, DAT recycling is minimal, and enzymatic degradation by catechol-O-methyltransferase (COMT) is the primary mechanism for clearance, modulating evoked dopamine release measured over minutes17,38,39. Though in our sample too few patients were taking COMT inhibitors to draw definitive conclusions, functional polymorphisms in COMT have been associated with motivational and mood disorders17,40.

Investigating the effects of different dopaminergic therapies on depression at the symptom dimension level provides insight into the potential utility of specific dopaminergic medications for the treatment of depression. However, large-scale randomized controlled trials of dopaminergic therapies for depression in PD are needed before any firm clinical recommendations can be made. Effective development of future treatments for depression in PD will depend on refining phenotypes and integrating our understanding of the evolution of depressive symptoms over time with the progression in PD pathology41.

Limitations

The associations we observed between different classes of treatment and depression symptom dimension could be a consequence of clinician treatment selection bias. However, this is unlikely as depression symptom dimension severity did not predict treatment selection at year 1.

The PPMI cohort only includes patients recently diagnosed with PD and may not be applicable to individuals in the later stages of the condition where the spread of neurodegeneration and systems involved are probably more complex.

The associations we observed between dopaminergic medication, striatal DAT binding and the depression symptom dimension could potentially be a consequence of other PD symptoms or functional disability. However, this is unlikely as all models were adjusted for cognition, motor symptoms, disease duration and functional disability.

Though all medication classes were incorporated into a single model, with the sample size available in PPMI it was not feasible to account for all the interactions of the various combination therapies patients were on.

Finally, depressive symptoms were analyzed as a continuous trait measure due to limited numbers of patients meeting a cut-off for clinical depression based on GDS-15 score. Future replication of our findings in a large cohort studies of PD patients with an established clinical diagnosis of depression is needed.

Conclusion

We identified dissociable effects of PD dopaminergic medications on different dimensions of depression. Dopamine agonists are a potentially effective treatment for motivational symptoms of depression in PD, especially beyond the second year following diagnosis. In contrast, MAO-B inhibitors may improve both depression and motivation symptoms, though the latter effect may be dependent on the severity of striatal dopaminergic neurodegeneration, potentially as a consequence of their mechanism of action depending on presynaptic dopaminergic neuron integrity. Further clinical trials of different dopaminergic treatment regimes are needed to identify clinically efficacious treatments for depression in PD.

Methods

We used data from the Parkinson’s Progression Markers Initiative (PPMI) cohort, an international multicenter cohort study42. Launched in 2010, PPMI enrolled untreated, patients with newly diagnosed PD and age- and sex-matched healthy controls. All participants underwent a standard battery of assessments including the MDS-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS), the 15-item GDS (GDS-15) and SPECT DAT imaging with Ioflupane [123-I] yearly, over 5 years.

To avoid capturing chronic symptoms in the context of depression with onset before the development of PD, all participants who had received a diagnosis of major depressive disorder more than 5 years before diagnosis with PD were excluded (n = 11).

We investigated two different depression symptom dimensions within the GDS-15 previously validated in independent cohorts using factor analysis: a three-item ‘motivation’ factor30 and a nine-item ‘depression’ factor31:

‘Motivation’ factor

GDS item 2—Have you dropped many of your activities or interests? Scale yes (1)/no (0)

GDS item 9—Do you prefer to stay at home, rather than going out and doing new things? Scale yes (1)/no (0)

GDS item 13—Do you feel full of energy? Scale yes (0)/no (1)

‘Depression’ factor

GDS item 1—Are you basically satisfied with your life? Scale yes (0)/no (1)

GDS item 3—Do you feel that your life is empty? Scale yes (1)/no (0)

GDS item 5—Are you in good spirits most of the time? Scale yes (0)/no (1)

GDS item 7—Do you feel happy most of the time? Scale yes (0)/no (1)

GDS item 8—Do you often feel helpless? Scale yes (1)/no (0)

GDS item 11—Do you think it is wonderful to be alive? Scale yes (0)/no (1)

GDS item 12—Do you feel pretty worthless the way you are now? Scale yes (1)/no (0)

GDS item 14—Do you feel that your situation is hopeless? Scale yes (1)/no (0)

GDS item 15—Do you think that most people are better off than you are? Scale yes (1)/no (0)

A variable recording whether patients were taking a specific medication class at each time point was created for each of the following PD medication classes: dopamine agonists, MAO-B inhibitors, levodopa, amantadine and COMT inhibitors.

Striatal DAT imaging was used as a measure of presynaptic dopaminergic neurodegeneration, indexing the established decrease in striatal DAT SBR in PD as the disease progresses, owing to a loss of presynaptic dopaminergic projections from the substantia nigra and ventral tegmental area.

Statistical analysis

We used linear mixed-effects modeling to examine the relationship between each GDS-15 factor score (dependent variable) and medication state for each class of PD medication (independent variables, all included in the same multiple regression), which were acquired contemporaneously, and how this relationship changed over the progression of illness. This involved fitting both the main effects of medication class state (across all follow-up time points) and their interactions with time. This allowed interindividual heterogeneity and unequal follow-up intervals to be accommodated by incorporating random effects. Random intercept terms at the participant level were tested.

The interaction term between medication class state and time allowed us to assess how the relationship between each GDS-15 factor and specific medication changed over time, using all available follow up GDS-15 data as the outcome.

Two sets of regressions for each GDS-15 factor were conducted: (1) unadjusted and (2) adjusted for age, sex, years of education, duration of PD, and both GDS-15 factor scores (all at baseline); plus cognition (Montreal Cognitive Assessment), impulsivity (Questionnaire for Impulsive-Compulsive Disorders in Parkinson’s Disease), severity of motor symptoms (MDS-UPDRS part III score, ‘off’ medication), stage of disease/functional disability (Hoehn and Yahr scale), levodopa equivalent dose, antidepressant medication status and the other GDS-15 factor to ensure specificity (all at each contemporaneous time point).

Model fit was tested using the Akaike information criterion. Quantile–quantile plots were obtained to assess model residual distributions (Supplementary Q–Q Plots 1–3). Where model residuals did not meet a normal distribution assumption, a logarithmic transformation was applied and normality reassessed.

Secondary analysis was conducted to assess moderation of pharmacological effects by dopaminergic neurodegeneration, replacing the time interaction with an interaction with DAT SBR. The interaction term between medication class state and DAT SBR enabled analysis of how the relationship between each GDS-15 factor and specific medication changed with striatal presynaptic dopaminergic neurodegeneration.

All statistical analyses were performed in R version 4.1.2. The R package ‘lme4’ was used for mixed-effects modeling.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.