FormalPara Key Points

Promising results indicate a positive effect of selective serotonin reuptake inhibitors (SSRIs) on recovery after stroke.

The results from several ongoing large trials are awaited before routine use after stroke may be considered.

1 Introduction

The main indication for treatment with selective serotonin reuptake inhibitors (SSRIs) is moderate-to-severe depression. Depression after stroke occurs in approximately one in three patients [1], so SSRI treatment is common in stroke [2]. SSRIs inhibit the presynaptic serotonin transporter in neurons, thereby increasing the level of serotonin in the synaptic cleft [3]. Serotonin modulates practically all human behavioral processes as well as other brain functions such as mood, memory, aggression, motor control and sleep [4]. When released from platelets, serotonin facilitates platelet aggregation. The serotonin transporter in platelets is also inhibited by SSRIs, and treatment leads to a reduction of serotonin content in platelets [5], which in theory could lead to an antithrombotic effect as well as an increased risk of bleeding events.

SSRIs have been of interest in the search for potential pharmacological treatments to facilitate neural repair and rehabilitation after stroke. In earlier reports, the main focus was on the antidepressant effect; however, several more recent studies have focused on a potential neurorestorative effect of SSRIs. This effect has an expected therapeutic time window of days to weeks or even longer, corresponding to the recovery stage and the chronic state after stroke [6]. A potential antithrombotic effect caused by the inhibition of platelet aggregation may further play a role in both the acute and the more chronic phase after stroke, an aspect that has also gained attention more recently. This article reviews the evidence behind SSRIs as potential enhancers of stroke recovery.

A search of all published articles in English until July 2018 was conducted in PubMed using the search terms “serotonin uptake inhibitors,” “SSRI” OR “selective serotonin reuptake inhibitors” AND “stroke”. Titles, then relevant abstracts, were scrutinized, and the relevant articles were subsequently retrieved. Reference lists from relevant articles were also searched for other potentially relevant references.

2 Selective Serotonin Reuptake Inhibitors (SSRIs) and Functional Outcome after Stroke

Table 1 shows the identified randomized trials of SSRI treatment after stroke that had functional/motor outcome as the primary outcome measure.

Table 1 Chronological list of randomized trials of SSRI treatment after stroke with functional/motor outcome as primary outcome measure
Full size table

Our search identified four randomized placebo-controlled trials [7,8,9,10] of SSRI treatment in nondepressed stroke patients with a treatment duration of 1–3 months and a primary outcome measure of motor function (Table 1). The trials were all relatively small but have shown promising results. Dam et al. [7] studied the effect of 3 months of treatment with the SSRI fluoxetine, the tetracyclic antidepressant maprotiline or placebo in 52 stroke patients. They found improvement in motor and functional outcome in all three groups, with the greatest improvement in the fluoxetine group [7]. Acler et al. [8] randomized 20 patients to daily citalopram or placebo treatment for 1 month and found a significant improvement in neurological status and a decrease in motor excitability, measured by transcranial magnetic stimulation, over the unaffected hemisphere in patients receiving citalopram. The FLAME (fluoxetine for motor recovery after acute ischaemic stroke) study [9] randomized 118 ischemic stroke patients with moderate-to-severe motor deficits to fluoxetine or placebo treatment for 3 months within 5–10 days after stroke onset. At day 90, the fluoxetine-treated patients had significantly greater improvement on the Fugl-Meyer motor score (FMMS) compared with the placebo group, and the proportion of independent patients [modified Rankin Score (mRS) of 0–2] was significantly higher in this group [9]. Finally, Savadi Oskouie et al. [10] conducted a randomized, placebo-controlled trial in which 144 ischemic stroke patients were randomized to citalopram or placebo for 3 months within 7 days after stroke onset. Significantly more patients evidenced a reduction in National Institute of Health Stroke Scale (NIHSS) score of at least 50% and a favorable outcome on the mRS (score of 0–2) at 3 months in the citalopram group [10].

Two randomized single-blinded trials were identified where the control group was given standard treatment but not placebo [11, 12] (Table 1). The trials were larger and included a 6-month follow-up after 3 months of treatment with fluoxetine. He et al. [11] included 374 patients and observed improved functional outcome in the treated group at both 3 and 6 months and that neurological function improved at 6 months. Guo et al. [12] included 300 patients within 1 week after stroke and initiated treatment upon, or 1 week after, randomization. They found improved functional outcome and neurological function in the early-treated group compared with the standard-treated group at 3 months; at 6 months, improvement was observed compared with both the standard treated and the later-treated group. It appears that no comparison was made between the group treated upon randomization and those treated within 1 week. The results from these trials may indicate SSRI treatment exerts an effect beyond the treatment period and that treatment initiated in the early phase may be beneficial.

Three small (eight to ten patients) randomized placebo-controlled trials studying the effect of a single dose of SSRI were identified [13,14,15] (Table 1). Motor function was tested 2–5 h after drug administration, corresponding to presumed maximal plasma concentrations. Pariente et al. [13] and Zittel et al. [14] observed improved motor function. Further, using functional magnetic resonance imaging (fMRI), Pariente et al. [13] found hyperactivation in the ipsilesional primary motor cortex. On the other hand, Gourab et al. [15] found that treatment was associated with increased lower limb spasticity and no increased strength or lower limb function. In a crossover placebo-controlled trial in ten patients, Berends et al. [16] used electromyography to test muscle activation 5 h after a single dose of fluoxetine, and motor function was assessed using grip strength as a secondary outcome. The authors found that fluoxetine increased activation in both agonist and antagonist muscles in the arm, but grip strength was not affected. They argued that the increased rigidity may even decrease motor function. Optimal timing for testing after a single dose of SSRI is unclear, and although Gourab et al. [15] and Berends et al. [16] found no positive effect on motor function but rather increased spasticity and possibly rigidity, these studies indicate that SSRI to some extent influenced motor output and cortical activation. Accordingly, modulation of cerebral motor activity after a single dose of SSRI has also been shown in healthy subjects [17, 18], whereas a more recent study found that a single dose did not enhance corticomotorneuronal excitability, motor performance or practice-dependent plasticity in healthy subjects [19].

Finally, three systematic reviews and meta-analyses on SSRI treatment and functional outcome after stroke were identified [20,21,22]. A Cochrane review from 2012 included 52 randomized controlled trials comprising 4060 participants in a meta-analysis [20, 23]. Some trials included patients with depression, whereas other trials included only patients without depression. The mean time since stroke to study inclusion was 0–3 months for the majority of trials (31 of 52), but ten trials did not report the time from onset to trial inclusion. Duration of treatment varied between trials, ranging from weeks to months, and only eight trials followed patients after the end of the treatment period. As described and also concluded by the authors, there was heterogeneity between the trials, and several trials had substantial methodological limitations. In this comprehensive review and meta-analysis, SSRI treatment appeared to improve dependence, disability, neurological impairment, anxiety and depression after stroke, with an effect also seen in patients without depression. On the other hand, treatment also appeared to be associated with an increased risk of adverse events. The authors concluded evidence was insufficient to recommend routine use of SSRIs in stroke recovery, and questions remained unanswered regarding what class of SSRI to use, when to initiate treatment, length of treatment, and dosage [20]. A more recent meta-analysis of trials studying the effects of different central nervous system drugs on stroke recovery included 17 trials (1575 patients) studying the effects of SSRIs [21]. The studied outcomes were gross motor function, cognition, disability, dependency and quality of life (QOL). They found that SSRI treatment improved gross motor function, disability and QOL. They found insufficient evidence that treatment enhanced global cognition and dependency. Ten of the studies did not consider time since stroke as an inclusion criterion. Five studies included only ischemic stroke patients, and 12 studies included both ischemic and hemorrhagic stroke patients. Most studies included strokes with moderate severity, but four studies did not report severity. Importantly, seven of the trials used no control drug [21]. In another recent meta-analysis, Gu et al. [22] focused on early administration (≤ 30 days) of SSRI. The primary outcome was decrease in NIHSS, and secondary outcomes were improvement in Barthel Index (BI) score and functional independence, defined as a score of 0–2 on the mRS. They included eight trials comprising 1549 nondepressed patients and found that treatment with SSRIs compared with placebo was associated with a greater decrease in NIHSS, an improved BI score and a significantly higher rate of functional independence. They found that the primary outcome was significantly better among trials with higher NIHSS scores at baseline (≥ 10), which may partly have been caused by a ceiling effect. They found no significant difference between SSRI and placebo on incidence of depression or the risk of adverse events [22].

The randomized trials in Table 1 were all conducted among nondepressed stroke patients, and primary outcomes were functional and motor outcomes associated with SSRI treatment [7,8,9,10,11,12,13,14,15, 20,21,22]. The SSRI escitalopram has also been associated with improvement in cognitive function after stroke [24]. In a recent randomized, double-blind, placebo-controlled study including 478 ischemic or hemorrhagic stroke patients within 21 days after stroke onset with a mean baseline NIHSS of five, patients were treated with either escitalopram or placebo for 3 months and followed for 6 months [25]. The primary outcome was frequency of depressive symptoms, but secondary outcomes were motor function on the Hemispheric Stroke Scale, NIHSS and BI scores, mRS, and cognitive functioning as measured by the Montreal Cognitive Assessment (MoCA) score at 3 and 6 months. The authors found no difference in outcomes between the two groups [25].

Although the results from most of the reviewed trials are encouraging and indicate a positive effect from SSRI treatment in stroke recovery, the reviewed trials were all relatively small, and the largest trials were not placebo controlled [11, 12]. The results from the meta-analyses also indicate a positive effect of SSRI treatment, but they are based on small heterogeneous trials, and several trials had substantial methodological limitations [20,21,22]. The single-dose studies by Gourab et al. [15] and Berends et al. [16] and the study by Kim et al. [25] found no positive effect of treatment on functional outcome. The patients in the study by Kim et al. [25] had a relatively low mean NIHSS score, which may have made a potential improvement insignificant. Compared with the FLAME study [9] and that by Savadi Oskouie et al. [10], Kim et al. [25] included patients with more severe depression, and patients were included in a later time window of up to 21 days after stroke onset.

3 Safety of SSRI Treatment after Stroke

The results from the FLAME study [9] boosted discussion around the use of SSRIs as part of routine treatment after stroke [26,27,28,29,30,31]. Although the results gave rise to optimism, several concerns were also raised. One concern included potential drug–drug interactions, specifically a potential interaction with clopidogrel, which is often used in secondary prophylaxis after stroke. Fluoxetine inhibits cytochrome P450 2C19, which is involved in the activation of clopidogrel into its active metabolite, and thus may inhibit the activation and thereby the efficacy of the drug [27]. Other concerns include potential side effects and a lack of knowledge about potential mechanisms, about the optimal timing and duration of treatment, and about the effect on milder strokes and on deficits other than motor impairment [30]. Besides the more well-known SSRI-related side effects such as gastrointestinal symptoms, hyponatremia, and sexual dysfunction, SSRIs have also been associated with an increased risk of hemorrhagic stroke, ranging from one expected additional intracerebral hemorrhage per 33,333 individuals in a meta-analysis by Shin et al. [33] to one in 10,000 treated individuals in a meta-analysis by Hackam and Mrkobrada [32]. The increased bleeding risk could be explained by the inhibition of serotonin uptake in platelets by SSRIs and corresponds with the well-described risk of gastrointestinal bleedings [34,35,36]. Importantly, the studies included in the meta-analyses by Shin et al. [33] and Hackam and Mrkobrada [32] were all conducted in nonstroke populations. The risk of stroke recurrence associated with SSRI treatment was studied in a follow-up study by Mortensen et al. [37] including 5833 ischemic stroke patients treated with SSRIs and 5833 propensity score-matched controls not treated with SSRIs. SSRI treatment was associated with a significantly increased risk of bleeding complications, but the absolute risk of intracranial bleedings was low and not statistically significant [37]. SSRI treatment was associated with a reduced risk of recurrent ischemic stroke, which is possibly due to an antithrombotic effect [37]. On the other hand, vasospastic effects and serotonin syndrome have been proposed as possible mechanisms behind the increased risk of ischemic stroke [38, 39] found in a cohort study by Juang et al. [40], conducted among 16,770 stroke patients. SSRI treatment was not associated with an increased risk of hemorrhagic stroke in this study [40]. Wang et al. [41] conducted a case–control study that included 3536 cases and 6679 controls and found no association between the use of SSRIs and recurrent stroke risk. Importantly, in the study by Juang et al. [40], a distinction between ischemic and hemorrhagic stroke was not made for incident strokes, and in the case–control study by Wang et al. [41], neither incident nor recurrent events were distinguished between ischemic and hemorrhagic events. Further, in observational studies, the risk of potential residual confounding, despite controlling for potential confounding variables in the statistical analysis, needs to be considered. In particular, confounding by indication should be considered, as depression is associated with an increased risk of cardiovascular disease [42]. He et al. [43] studied the risk of stroke recurrence in a randomized controlled trial. A reduced risk of 3-year recurrence of ischemic stroke was found in a randomized controlled trial of 404 ischemic stroke patients treated with fluoxetine or placebo for 90 days. The risk of hemorrhagic stroke was not studied.

In line with the possible increased risk of bleeding complications associated with SSRI treatment, the risk of bleeding complications in acute ischemic stroke patients treated with thrombolysis and pre-stroke SSRI treatment has been studied [44,45,46]. The cohorts in these studies included relatively small groups of pre-stroke SSRI-treated patients (135, 266 and 22, respectively). Overall, however, treatment with thrombolysis did not appear to increase the risk of bleeding complications.

4 Mechanisms

The exact mechanism(s) behind a potential beneficial effect of SSRIs in stroke recovery is not clear. A beneficial effect on remission of depression and reduction of depressive symptoms with antidepressant treatment could contribute to an increased functional recovery, as post-stroke depression itself increases the risk of suboptimal recovery, recurrent vascular events, poorer QOL and mortality [47, 48]. Treating depression could increase patients’ motivation and participation in rehabilitation efforts and thus the functional outcome. On the other hand, a positive effect on functional recovery was found in studies including clinically nondepressed patients, such as the FLAME study [9], and an effect of pharmacological therapy on the prevention of depression after stroke has not been established [49]. It is worth mentioning that 7% of the patients in the FLAME study developed depression during follow-up, and these patients were not excluded from the final analyses. In this particular study, a positive effect on outcome, at least partly due to an antidepressant effect, cannot be ruled out. Several previously mentioned studies found an effect of a single dose of SSRI on both cortical excitability and functional performance, and the antidepressant effect is not likely to be the sole mechanism [13, 14, 17, 18]. Further, in their randomized controlled trial, Dam et al. [7] found fluoxetine to be superior to the tetracyclic antidepressant maprotiline in enhancing recovery, whereas both drugs improved depressive symptoms significantly. The modulation of cerebral motor activity has been proposed as one of the more specific mechanisms behind an enhancement of motor function. Stroke leads to a disruption of the excitation–inhibition balance, and it has been proposed that SSRIs may reestablish this balance, possibly through an acute effect on enhancement of excitatory activity followed by an increase in inhibitory activity [50]. Several other possible mechanisms have been identified in animal models, including a neurotrophic effect promoting brain plasticity, which can be defined as the capacity of cerebral neurons and neural circuits to structurally and functionally change [51]. SSRIs have thus been found to promote both neurogenesis [52] and angiogenesis [53], possibly mediated through the regulation of brain-derived neurotrophic factor and vascular endothelial growth factor [54, 55]. Whereas neurotrophic agents exert their effect days to weeks or even longer after stroke, agents with a neuroprotective effect, i.e., agents with the ability to preserve neurons, exert their effect within the initial hours, or at most days, after stroke [6]. It has been proposed that at least part of the positive effect of SSRIs on rehabilitation after stroke is caused by neuroprotection possibly mediated through an anti-inflammatory effect [56] or through the enhancement of specific protein expression [57].

Different mechanisms may be behind the different effects on outcome. Hippocampal neurogenesis may improve post-stroke cognitive function, whereas the effect on cortical excitability may affect motor function. There may be a more immediate effect, including the inhibition of serotonin reuptake and longer-term effects such as the regulation of neurotrophic factors. As previously stated, the exact mechanisms behind a beneficial effect of SSRIs after stroke are not clear, but a multimodal effect is likely, as SSRIs affect different brain structures and the balance between different neurotransmitters. Some of the effects, particularly excitation and inhibition, may even be opposing, and, as indicated by the single-dose studies by Gourab et al. [15] and Berends et al. [16], SSRIs may increase motor tone, possibly through an effect on the spinal canal, and thereby affect motor function negatively.

5 Future Trials and Implications for the Management of Recovering Stroke Patients

The possible increased risk of hemorrhagic stroke, and the less unequivocal association with ischemic stroke in SSRI treatment may be acceptable when treating patients with post-stroke depression or emotional lability, as these are potentially devastating consequences of stroke. When using SSRIs to potentiate rehabilitation, it is equally important that the potential gain outweighs the potential adverse effects. An antithrombotic effect that could potentially reduce the rate of recurrent ischemic events after stroke has been studied in the TALOS trial. In this trial, 600 ischemic stroke patients were randomized to citalopram or placebo treatment for 6 months with two co-primary outcomes: mRS score at 6 months and a composite vascular endpoint of transient ischemic attack/stroke, myocardial infarction or vascular mortality during the first 6 months [58]. The results from the TALOS trial are awaiting publication. The randomized controlled trials, FOCUS (UK), AFFINITY (Australia, New Zealand and Vietnam) and EFFECTS (Sweden) are studying the effects of 6 months of fluoxetine treatment in acute (2–15 days after onset) ischemic and hemorrhagic stroke patients with persisting neurological deficits. The primary outcome is mRS at 6 months. A core protocol for the three trials has been published, and a total of 6000 patients will be included [59]. Plans are for inclusion and follow-up for FOCUS to be completed by the end of 2018 and for AFFINITY and EFFECTS by 2020. As FOCUS, AFFINITY and EFFECTS include both ischemic and hemorrhagic stroke patients, the results from these trials may help resolve the question of safety also in hemorrhagic stroke. The results will hopefully also help answer the question of whether SSRI treatment after stroke facilitates rehabilitation and whether these drugs should be considered as part of the routine treatment after stroke. Regardless, a decision must always be based on considerations of indication and potential gain versus drug–drug interactions and potential side effects in individual patients. A potential aspect to consider in future trials when studying personalized medicine, is genetics. As an example, serotonin transporter (SERT) gene polymorphisms have been associated with the risk of stroke [60], the risk of post-stroke depression [61] and, finally, post-stroke neurological recovery after SSRI use [62]. This is also an important aspect to consider when comparing trials with different ethnic profiles, as the distribution of SERT gene polymorphisms may differ according to ethnicity [63].

6 Conclusions

SSRIs may have neurorestorative effects with the potential to facilitate rehabilitation in the sub-acute and chronic phase after stroke and improve outcome in a greater proportion of stroke patients. A neuroprotective effect may also be present if treatment is given in the acute phase. Several possible mechanisms may be behind these potential effects, although they remain speculative. The effect is most likely multimodal, and excitatory and inhibitory effects may affect motor outcome in opposing directions. Results from several large ongoing trials are awaited before a potential beneficial effect may be clarified and before routine treatment may be considered. Further, studies exploring the underlying mechanisms, optimal timing, treatment duration, interactions and effects on bleeding complications and other potential adverse effects as well as genetic implications are warranted. A potential beneficial effect of treatment must always be weighed against potential adverse effects in each individual patient.