Management of Depression in Parkinson's Disease: A Systematic Review

Abstract

Background

Depression is a frequent psychiatric condition in Parkinson's disease (PD). The treatment of depression has been examined in several randomized controlled trials and meta‐analyses, but no clear guidelines are available.

Methods

We carried out a systematic review of pharmacological and non‐pharmacological treatments for depression in patients with PD using the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidelines. We searched main medical databases up to December 12, 2016, and included randomized controlled trials, patient‐control studies, and case series with data on treatment modality, outcome measures, and side effects.

Results

Selective serotonergic reuptake inhibitors and tricyclic antidepressants may have efficacy for the treatment of depression in patients with PD, although the evidence is not strong. The antidepressant efficacy of dopamine agonists is still controversial, and initial results were positive for pramipexole but not for rotigotine. Cognitive‐behavioral therapy showed promising results in two recent randomized controlled trials, but further evidence is required. Studies using repetitive transcranial magnetic stimulation produced conflicting results, and the efficacy results for this treatment have been inconsistent. On the other hand, electroconclusive therapy produced strong positive results in patients with severe depression, but no randomized controlled trials are available.

Conclusion

Selective serotonergic reuptake inhibitors and cognitive‐behavioral therapy are currently first‐line treatments for depression in patients with PD, although the evidence is still weak. The heterogeneity among contributory factors for depression in PD should be considered for the most effective treatment of depression in this condition.

Introduction

Depression is among the most common psychiatric disorders in Parkinson's disease (PD). Prevalence rates for major depression range from 8% in general population studies and 24% in outpatient studies1 to 40% for patients attending specialized movement disorders clinics.2 The 1‐year incidence of minor depression was estimated at 18%.3 Although depression is associated with worse motor function and increased disease severity,4 it is already present in 29% of patients with de novo PD and in 40% of patients with Hoehn and Yahr stage I PD.5

In this systematic review of the current evidence for the treatment of depression in PD, we begin with a discussion of the methodology being used to diagnose depression in patients with PD, including the most useful instruments used in busy PD clinics. This is followed by a discussion of the management options, as provided by pharmacological and nonpharmacological treatment studies (Table S1), and we conclude with final recommendations based on a narrative review of the literature and clinical experience.

Materials and Methods

Search Strategy and Literature Sources

We conducted our systematic review according to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) Protocols Statement.6 We searched for research‐based articles catalogued in PubMed, the Cochrane Library, Medline, Web of Science, Science Direct, Cumulative Index of Nursing and Allied Health Literature (CINAHL) Plus, ProQuest science journals, Scopus, and Wiley from database inception to December 12, 2016, without restriction on year, age, or language. The keywords “depression,” “Parkinson,” and “treatment OR management” were combined with “AND” to search within the title and abstract fields for each database. We also included relevant keyword derivatives, such as Parkinson's, Parkinson's disease, and PD. Bibliographies for all included articles were also searched.

Selection and Eligibility

All citations were independently reviewed for eligibility by both authors. Citations were initially filtered by analysis of the title (S.B.), and those that reported on the efficacy of any treatment of depression in patients with PD, regardless of publication type, were included for initial abstract review. Any disagreement regarding eligibility resulted in including the full article for review. Eligibility at the full text stage required the following: the article had to include patients with PD ages 18 years and older (both sexes), to report on the efficacy of treating depression (as diagnosed by depression scales and/or diagnostic criteria) as the main outcome measure, to be a randomized controlled trial (RTC) with a sample size of 10 or more patients, and to be written in English or Spanish (S.E.S. is a native Spanish speaker). During the review, if it was clear that publications stemmed from the same database, then only the original publication was included.

Data Extraction and Analyses

The following data were extracted from the selected articles: authors, type of study, participant characteristics, intervention type, diagnostic criteria, primary and secondary outcomes measures, efficacy and safety of the intervention, and summary measures, including risk ratio and difference in means.

Bias Risk Assessment

We analyzed eligibility criteria for participants of the sample, the random allocation of participants, the presence and type of a control group, the presentation of results, and the intergroup variability of the results.

Results

Our electronic database search generated a total of 1405 studies (Fig. 1). After duplicates were removed (n = 1107), an additional 551 articles were excluded because they did not satisfy keyword criteria, 10 were excluded for being in a language other than Spanish or English, 4 were excluded because they could not be sourced, and 153 were excluded because of the publication date and/or because they had insufficient data. The remaining 80 articles underwent a full text review, and an additional 18 articles were excluded because: depression was not the main outcome measure (n = 6), they were not an RCT (n = 6), they were based on previously reported databases (n = 5), and they included less than 10 participants (n = 1). Before discussing the findings, it is important to provide an overview on the complexities of diagnosing depression in the context of PD.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISM) flow diagram. RCT indicates randomized controlled trial.

The Diagnosis of Depression in PD

In most recent research studies, depression in PD has been diagnosed based on standardized criteria for major depression, such criteria from the Diagnostic and Statistical Manual of Mental Disorders (DSM), fifth edition (DSM‐5).7 The main problem of using these criteria is that some symptoms of depression, such as psychomotor retardation, insomnia, and loss of energy, are also frequently found in PD. A committee convened by the National Institute of Neurological Disorders and Stroke raised the important question of whether DSM criteria for major depression, which were designed for individuals without PD, are valid in the context of a disease with a variety of motor and nonmotor manifestations.8 The committee recommended that depressed mood must be present (i.e., decreased interest alone is not sufficient to satisfy core features, because apathy in PD may present as a discrete entity) and that evaluations should be done in the “on” state for patients who experience fluctuations. Furthermore, to avoid “false‐negatives,” the National Institute of Neurological Disorders and Stroke committee suggested using the “inclusive approach” to diagnosis, which consist of rating the presence of all symptoms of depression, independent of whether they may be related to the motor disorder.

One important nosological problem beyond the field of depression in PD is the validity of the DSM‐5 criteria for major depression, which are most often used as the diagnostic “gold standard.” It has been the objective of recent editions of the DSM to have high reliability, but the validity of many diagnoses, including depression, remains uncertain. Fried and colleagues9, 10 reported marked variation in symptoms among individuals without PD who met DSM‐5 criteria for major depression, suggesting that major depression is not a homogeneous syndrome. This suggests that the DSM‐5 syndrome of major depression may include different syndromes with a variety of underlying causes, different clinical manifestations over time, and different responses to treatment. This heterogeneity in the clinical expression and causative factors of major depression may also be present in patients with PD, adding complexity to a diagnostic process that is already complicated by the overlap between symptoms of depression and the motor symptoms of PD. Nevertheless, using latent class analysis, our group has demonstrated that all 9 DSM‐5 symptoms of major depression are valid for use in PD.11 This was confirmed by a drug trial by Dobkin et al.,12 who reported improvement in both motor and nonmotor symptoms of depression after a trial with nortriptyline.

Another limitation is that depression has to be screened for in the context of busy clinics, where the main focus is on the motor symptoms of PD. Ideally, every patient should be screened for the presence of depression, raising the issue of choosing the best instrument. Williams et al.13 examined the psychometric attributes of 9 scales to detect depression in patients with PD and concluded that the self‐reported Geriatric Depression Scale (GDS‐30) had the best efficient profile, given its relative brevity, good psychometric properties, and no copyright protection. Another option is using the self‐rated Patient Health Questionnaire‐9 (PHQ‐9),14 an instrument that has been used extensively in both research studies and clinical practice in many clinical conditions. In the study by Williams and colleagues, the PHQ‐9 had a psychometric profile similar to that of the GDS‐30; in addition, it had a shorter completion time and the advantage of assessing all of the symptoms necessary for a DSM‐5 diagnosis of major depression. Chagas et al.15 reported that a cutoff score of 9 on the PHQ‐9 had 100% sensitivity and 83% specificity for discriminating major depression from no depression. To conclude, we believe that the most effective method to screen for depression in PD is using the PHQ‐9 or the GDS‐30, whereas major depression should be diagnosed based on DSM‐5 criteria until specific diagnostic criteria are validated for use in this condition.

The Treatment of Depression in PD

Our PRISMA review included all RCTs for depression in PD, patient‐control studies, and case series collected as described above. We focused on efficacy and side effects, and the results of pharmacological and nonpharmacological treatments are summarized below.

Pharmacological treatment

Efficacy of antidepressant medication

Several RCTs examined the efficacy of Selective serotonergic reuptake inhibitors (SSRIs) and tricyclic antidepressants (TCAs) in PD, but the findings are difficult to consolidate into concrete recommendations. The study with the largest sample size (N = 115) examined the efficacy of the SSRI paroxetine and the selective serotonin‐norepinephrine reuptake inhibitor (SNRI) venlafaxine compared with placebo.16 After 12 weeks of treatment, both drugs showed significantly greater efficacy compared with placebo. An earlier RCT compared paroxetine with the TCA nortriptyline and placebo and found that nortriptyline was more effective than either paroxetine or placebo.17 Moreover, and different from the findings of Richard and colleagues, paroxetine was no more effective than placebo. Low power in the study by Menza et al. may account for the reported lack of efficacy of paroxetine, but the response rate to placebo was almost double the response to paroxetine (24% vs. 11%, respectively), suggesting that power may not account for the difference between studies. Another important difference was that the study by Richard et al. demonstrated a much higher response rate for both paroxetine (68%) and placebo (44%). Richard and colleagues’ study had a 12‐week duration compared with 8 weeks in the study by Menza et al., although it is unlikely that this difference can account for the large discrepancies in paroxetine and placebo responses. Furthermore, the dose of paroxetine was similar in both studies (range, 24–28 mg/day). An important methodological difference is that Menza et al. included only patients who met DSM‐IV criteria for either major depression or dysthymia, whereas Richard et al. also included patients who had minor and subsyndromal depression. A second study comparing an SSRI (citalopram) and a TCA (desipramine) against placebo was a 4‐week RCT that included patients with PD who met DSM‐IV criteria for major depression.18 That study demonstrated significant improvements for both active drugs compared with placebo by the end of treatment. Finally, most patients who have PD with depression are currently treated with SSRIs (an analysis of the 2002 US Veterans Affairs database demonstrated that 63% of patients with PD who were on antidepressants were taking SSRIs, and only 7% were taking TCAs19). In recent RCTs, TCAs, citalopram and venlafaxine proved to be effective, whereas the evidence for paroxetine was conflicting. The main limitations of RCTs include the short duration of most of them, the relatively small number of patients included, and the heterogeneity in the assessment and diagnosis of depression. Most studies have excluded patients with dementia or severe motor fluctuations, which are commonly seen in the clinic, and patients with PD who are recruited for the trials may not represent the most common presentations in clinical practice.

We also reviewed published meta‐analyses of antidepressants in PD that included several older RCTs. A meta‐analysis by Skapinakis and colleagues20 focused on the efficacy of SSRIs only and included 10 studies that used citalopram, sertraline, fluoxetine, paroxetine, or fluvoxamine. There was no efficacy for SSRIs in patients with PD, with a crude response rate of 34% for SSRIs and 36% for placebo. This response rate is similar to that reported for placebo in meta‐analysis for antidepressants in non‐PD depression,21 suggesting that depression in PD is responsive to treatment but that the efficacy is rendered nonsignificant by a high placebo effect. STAR*D (Sequenced Treatment Alternatives to Relieve Depression),22 a drug trial for “primary” depression that included patients who had clinical profiles similar to patients who attended general practices, showed a remission rate for citalopram of 27% as measured by the Hamilton Rating Scale for Depression. Skapinakis et al. also concluded that the small number of RCTs that used TCAs does not allow conclusive recommendations regarding this type of antidepressant.

A later meta‐analysis by Liu et al.23 that included the report by Richards et al. (which was not included in Skapinakis and colleagues’ meta‐analysis) replicated the lack of efficacy for both SSRIs and for SNRIs and suggested higher efficacy for TCAs over SSRIs. Nevertheless, it is important to note that their meta‐analysis included only 1 study that used venlafaxine, which reported positive results. A more recent meta‐analysis by Troeung et al.24 calculated a pool effect size for their selection of 9 RCTs and observed a moderate but statistically nonsignificant effect for antidepressants compared with placebo. Those authors stressed that the lack of statistical significance could be explained by a Type II error due to small sample sizes. They also confirmed the higher efficacy of TCAs over SSRIs for depression. The most recent meta‐analysis by Bomasang‐Layno and colleagues25 did not include any new RCTs compared with the meta‐analysis by Troeung et al. but reported significant efficacy for SSRIs over TCAs.

Finally, meta‐analyses of reports on antidepressants in PD have produced unclear results, and several factors may explain the lack of consistency. First, meta‐analyses used different statistical techniques and different inclusion criteria. RCTs for depression in PD mostly include small patient samples and are biased toward individuals who have milder parkinsonism and no cognitive deficits. Furthermore, the inclusion/exclusion of a few studies may have a large impact on the final results. The second problem is the heterogeneity in terms of diagnostic instruments, trial duration, and primary outcome measures between the rather limited numbers of RCTs. The third problem is the large placebo effect and the lack of a difference between the active and placebo arms, which may be addressed in the future by better trial designs. Therefore, we conclude that meta‐analyses may provide some guidance about which antidepressant to choose but without firm suggestions.

Before concluding this section, it is important to at least briefly address the main side effects and pharmacological interactions of the antidepressants discussed for use in PD. TCAs should be avoided in patients who have cardiac problems, such as bifascicular block, left bundle branch block, a prolonged QTc interval, or a recent myocardial infarction, and in patients with narrow angle‐closure glaucoma. TCAs have a high lethality index and should be used cautiously in patients with suicidal ideation. Several side effects of TCAs are of special concern in PD, such as orthostatic hypotension, with the concomitant risk of falls and injuries, constipation, and urinary retention, because these are also common features of PD. Confusion and delirium may mostly occur among patients who have PD with comorbid dementia. On the other hand, the anticholinergic effects of TCAs may be useful in reducing tremor. TCAs are also sedating, which may increase the risk of falls. On the other hand, sedation may help to improve insomnia. Nortriptyline is the only TCA with a therapeutic window (between 50 and 150 ng/mL); it can be started at 10 mg once daily at bedtime and may be increased by 25 mg every 7 days, up to a maximum of 100 mg daily. SSRIs have a milder profile of side effects and a lower lethality index than TCAs. Most SSRIs may produce nausea, sedation, a fast action tremor, and sexual dysfunction. Sodium levels should be regularly checked in the elderly due to the risk of producing hyponatremia. Citalopram is a well‐tolerated SSRI; it is usually started at 10 mg daily and may be increased by 10 mg daily every 7 days up to a maximum of 40 mg daily. The US Food and Drug Administration (FDA) recommends limiting the citalopram dose to 20 mg daily in the elderly because of potential for an increased QTc interval.26 Selegiline is a compound with weak antidepressant efficacy and is contraindicated in patients who are taking other antidepressants, whereas rasagiline is contraindicated in patients who are taking the monoamine oxidase inhibitors fluoxetine or fluvoxamine. It is important to note that no study using any other monoamine oxidase inhibitors met the inclusion criteria for this systematic review. Rasagiline should be used cautiously together with other SSRIs because of the risk of producing a serotonergic syndrome.27 It is believed that this risk is relatively low because, at the dosages typically used for PD (5 mg or 10 mg daily), there is dopamine selectivity, and it becomes nonselective only at higher dosages.

It is also important to consider RCTs of antidepressants for “primary” depression, which may help to extrapolate some results to the context of depression in PD. A recent review and meta‐analysis28 suggested that individuals ages 65 years and older may not benefit from antidepressants (see also Tedeschini et al.29). Most elderly patients do not achieve remission during acute treatment, and RCTs include samples that are biased toward individuals without complex clinical or psychiatric problems, suicidal ideation, or dementia. Patients in RCTs are considered to consist of relatively uncomplicated individuals uncommonly seen in clinical practice.22 Recent studies30, 31 have also demonstrated an increased risk of falls among elderly patients with “primary” depression who are taking SSRIs.

The question arises regarding what should be tried when the patient fails the first antidepressant trial. In patients with primary depression, the STAR*D study22 demonstrated that 27% of nonresponders achieved remission after a second antidepressant trial, and 31% achieved remission with cognitive‐behavioral therapy (CBT), which was better tolerated than antidepressants. Nevertheless, the efficacy of switching antidepressants has been challenged in another recent meta‐analysis.32

Efficacy of antiparkinsonian medication

The antidepressant efficacy of dopaminergic agonists has been the focus of recent RCTs. A meta‐analysis by Leentjens et al.33 suggesting antidepressant efficacy for pramipexole, a potent D2 agonist, was confirmed in a more recent RCT by Barone et al.34 Nevertheless, some limitations of that study should be considered. First, the RCT included patients with very mild to moderate depression. Moreover, the difference between the active and placebo groups at the end of the study was of only 2 points on the Beck Depression Inventory, which is of questionable clinical significance, and only 27% responded to pramipexole compared with 18% in the placebo arm. The fact that the pramipexole group had significant motor improvement could have influenced mood improvement, although this was statistically controlled. Rotigotine, another dopaminergic agonist, was studied as a potential antidepressant in PD in a recent RCT.35 At the end of the 8‐week trial, there was no significant difference between the active group and the placebo group, with response rates of 41% and 43%, respectively. Rasagiline is an irreversible monoamine oxidase‐B inhibitor with proved efficacy to improve motor problems in PD. A recent 12‐week RCT36 demonstrated no efficacy for this compound to improve depression in patients with PD.

Psychotherapy

Psychotherapy presents as a good option for patients with PD who may not want to use antidepressant medication or have contraindications to their use. Only 2 RCTs of CBT have been published to date,12, 37 and both showed significant benefits of psychotherapy over placebo, with large effect sizes and response rates. Cognitive‐based mindfulness therapy is a new type of psychotherapy based on meditation techniques, breathing exercises, and yoga postures that may prove beneficial for patients with mild to moderate depression and anxiety.38

Repetitive transcranial magnetic stimulation

Repetitive transcranial magnetic stimulation (rTMS) works by producing excitation or inhibition of brain cortical regions and has demonstrated efficacy in treating primary depression. The most frequent rTMS technique in primary depression consists of delivering a high‐frequency stimulus to the left dorsolateral‐prefrontal cortex (DLPFC) and a low‐frequency stimulus to the right DLPFC. A recent meta‐analysis demonstrated that bilateral stimulation had higher efficacy for primary depression than newer techniques.39 That single meta‐analysis of rTMS for depression in patients with PD identified 7 studies comprising 290 patients.40 Five studies compared rTMS against an SSRI (fluoxetine in 3 studies, paroxetine and sertraline in 1 study each) and found similar efficacy for both treatments. The unclear efficacy of SSRIs for depression in PD and the lack of a placebo arm make the potential efficacy for rTMS in PD difficult to ascertain. rTMS may improve motor function in PD,40 and this important confounder should be considered in any rTMS RCT. The remaining 2 studies compared active versus sham rTMS, and the results did not suggest a significant benefit for the active treatment.

Three new RCTs using rTMS for depression in PD have been published since the meta‐analysis by Xie et al.40 Pal and colleagues41 stimulated the left DLPFC using low frequency and observed a significant benefit for active treatment over placebo. One limitation of the study was that “response” was defined as a reduction on the Montgomery‐Asberg Depression Rating Scale of at least 2 points, which had dubious clinical significance, and the effect size of the treatment was not provided. Brys et al.42 delivered high‐frequency rTMS to either the left DLPFC, the primary motor cortex bilaterally, to both sites, or to neither site for patients randomly allocated in a 1:1:1:1 manner. Those investigators found no significant differences between the active and sham treatments after 1, 3, and 6 months of treatment. Finally, Shin and colleagues43 carried out a 6‐week study of high‐frequency rTMS to the left DLPFC on a small sample (N = 18) and reported significant benefit from the active treatment, as evidenced by significant reductions in scores on the Hamilton Rating Scale for Depression and the Montgomery‐Asberg Depression Rating Scale (48% and 45% change, respectively), but not on the Beck Depression Inventory (37% change), compared with sham treatment (the percentage change for this group was not reported).

Electroconvulsive therapy

Electroconvulsive therapy (ECT) may be considered for patients who have PD with treatment‐resistant depression. Borisovskaya et al.44 carried out a systematic review of ECT in patients with PD and included 43 publications consisting of small case series or case reports, but no RCTs. Their analysis included 116 patients, most of whom had major depression. The authors reported improvement in depression in 93% of patients, with 83% also reporting motor improvements. Side effects were not systematically analyzed, but 27% or patients had delirium, 9% had transient confusion, and 1 died with an acute myocardial infarction. The impact on cognition was not formally examined, and there were inconsistencies in the number of ECT sessions per patient, bilateral versus unilateral electrode placement, seizure duration, and stimulation parameters. The authors suggested that the results may have been biased by the potential lack of reporting of negative studies. Even in the absence of RCTs, and based on published reports and personal experience, ECT should be considered an effective treatment for patients with PD who do not improve after 2 full trials of different types of antidepressants and psychotherapy. Before undergoing ECT, it is recommended that patients undergo a comprehensive medical evaluation encompassing a full medical history, a review of current medications, and initiation of the consent process. Family members should be encouraged to attend so they can corroborate the patient's history, assist the patient's understanding of the treatment process, and receive education on how to care for the patient for the duration of the treatment program.45 The systematic review of ECT included several patients who had successful short‐term responses and were successfully maintained on ECT for long periods; however, there are a few reports of patients who developed cognitive deficits or delusions during maintenance ECT.

Conclusions

Currently, no definite guidelines for the treatment of depression in patients with PD can be provided. We reviewed findings from RCTs and meta‐analyses for psychoactive medication, psychotherapy, rTMS, and ECT, and identified several problems. Limitations common to all RCTs are small samples, short‐term duration, differences in baseline characteristics, variability for inclusion/exclusion criteria, variability in the assessment and diagnosis of depression, the inclusion of different severities of depression, differences in the severity of motor problems, different primary outcome measures, and different statistical techniques to analyze findings. There is no “best” treatment for depression in PD, and much will depend on the patient's medical status, severity of depression, and preference for a given treatment along with the neurologist's expertise in using antidepressants, psychiatric expertise, and availability of social supports.

It is highly likely that most patients who have PD with mild to moderate depression will be started on citalopram, venlafaxine, or paroxetine. If there is no response with 1 of these 3 antidepressants, then patients likely will be switched to another medication from this group. TCAs will be used most commonly if there is no response to the second SSRI or SNRI and provided there are no contraindications. Nortriptyline may be the TCA of choice given its lower profile of anticholinergic side effects compared with other TCAs and the possibility of checking whether the therapeutic window is reached. For patients who do not want to be on antidepressants and who have the necessary motivation and financial resources, CBT may be the treatment of choice. Patients in the early stages of PD with mild to moderate depression who are considered to need a dopaminergic agonist may be started on pramipexole, carefully watching for side effects like impulse‐control disorders. For patients with severe major depression or refractory depression, ECT may be the treatment of choice. The evidence for rTMS is still relatively weak, but this treatment may be considered before starting ECT.

In conclusion, it is important to stress that effective treatments depend on a proper knowledge of contributory factors. Depression in the context of PD is multifactorial, and the patients’ personality and character, their resilience and stress tolerance, the presence of fears and anxiety, contextual factors like financial status and levels of support, and yet to be clarified biologic factors have an important role in causing depression and also in the rate of successful treatment. Finally, the role of the physician is of great relevance, not only in choosing the most adequate treatment but also in approaching the patient with the necessary time, care, and understanding.

Author Roles

1. Research Project: A. Conception, B. Organization, C. Execution; 2. Statistical Analysis: A. Design, B. Execution, C. Review and Critique; 3. Manuscript Preparation: A. Writing the First Draft, B. Review and Critique.

S.E.S.: 1A, 3B

S.B.: 1B, 1C, 3A, 3B

Disclosures

Ethical Compliance Statement: We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

Funding Sources and Conflict of Interest: This research was funded by Parkinson's WA and the Zrinski Family through the award of a Zrinski Research Grant.

Financial Disclosures for the previous 12 months: The authors report no sources of funding and no conflicts of interest.

Supporting information

Table S1. Summary of studies meeting inclusion criteria for systematic review