Selective serotonin reuptake inhibitors (SSRIs) are widely used in the treatment of depressive disorders (Petersen et al., 2002; Chaudhry et al., 2011) and anxiety disorders (van der Linden et al., 2000; Hedges et al., 2007). Total SSRI prescription volume increased threefold between 1995/1996 and 2006/2007 (Lockhart and Guthrie, 2011). Despite this, the use of SSRIs is not without flaws. Apart from potential adverse side effects arising from long-term antidepressant treatment, as well as issues of relapse, and recurrence following remission of a depressive episode (Rucci et al., 2011; Rush et al., 2012), the existence of a SSRI discontinuation syndrome (also known as SSRI withdrawal syndrome) has been suggested in some cases. A decade ago, Harvey et al. (2003) highlighted that although significant progress had been made to uncover the pathophysiology of depression and the mechanisms of actions of SSRIs, the neurobiology of SSRI discontinuation syndrome had not been comprehensively addressed. Following an overview of the clinical evidence, we will discuss the possible molecular mechanisms implicated in the pathology of SSRI discontinuation syndrome, and comment on the use of animal models to better understand this condition.

About 15 years ago, Zajecka et al. (1997) proposed a definition of the SSRI discontinuation syndrome as the onset of a cluster of symptoms following the discontinuation of a SSRI, not attributable to other causes (i.e., concomitant medication, illness). Along with sensory and gastrointestinal symptoms, the SSRI discontinuation syndrome includes somatic symptoms such as dizziness, lethargy, and sleep disturbances, as well as psychological symptoms such as anxiety/agitation, irritability, and poor concentration (Warner et al., 2006; Haddad and Anderson, 2007). The literature on these symptoms initially consisted mainly of case reports (Coupland et al., 1996; Price et al., 1996; Schatzberg et al., 1997; Bryois et al., 1998; Goldstein et al., 1999). Haddad (1997) reviewed 47 separate reports of SSRI discontinuation syndrome, 30 of which involved paroxetine compared to only seven involving fluoxetine, despite the latter being prescribed more frequently. Black et al. (2000) identified 53 different symptoms within the condition (with dizziness being the most common) and proposed a diagnostic criteria for SSRI discontinuation syndrome that requires two or more of the described symptoms developing within 1–7 days of discontinuation (or reduction in dosage) of an SSRI after at least 1 month of treatment.

These case reports have since been followed up by a number of controlled studies (i.e., prospective studies, with a randomized, double-blind interruption period, which included a systematic method for discontinuation symptoms data collection). For example, using the Discontinuation Emergent Signs and Symptoms (DESS) checklist, Rosenbaum et al. (1998) assessed the effects of a 1-week placebo substitution period in patients diagnosed with unipolar depressive disorder who had been maintained on fluoxetine, sertraline, or paroxetine for similar periods of time (∼11 months). Following treatment interruption, there was a significant increase in the number of DESS items observed in the sertraline- and paroxetine-treated patients. In contrast, this was not detected in fluoxetine-treated patients. Another double-blind trial found that placebo substitution for paroxetine was associated with increased frequency and severity of specific physical and psychological symptoms, which arose as early as after the second missed dose (Michelson et al., 2000). In that trial, patients with a history of depression had been undergoing SSRI treatment for similar periods of time (∼12–15 months) and were equivalent on several parameters (e.g., similar baseline symptom severity). The discontinuation symptoms seemed to be specific to paroxetine in this study, since patients treated with sertraline or fluoxetine did not exhibit discontinuation syndrome within the 5-day placebo substitution period. However, since sertraline and fluoxetine (but not paroxetine) have active metabolites with half-lives around 2–3 and 7–15 days respectively (Table 1), the effects of longer withdrawal periods are worth assessing in future studies. In the case of fluoxetine, it is possible that discontinuation symptoms may emerge when the levels of the active metabolite decrease after 15 days and beyond. Consistent with that possibility, Zajecka et al. (1998) found a small increase in reports of dizziness among patients who discontinued fluoxetine 4 and 6 weeks after placebo substitution. Bogetto et al. (2002) investigated 97 outpatients diagnosed with dysthymic disorder who were instructed to stop their medication (paroxetine or fluoxetine) after a successful treatment period of at least 8 weeks. With a mean time of onset of symptoms of 2 days after drug discontinuation, discontinuation syndrome was reported by 27% of patients (of which 85% had been treated with paroxetine compared to only 15% treated with fluoxetine). Analyzing randomized controlled studies of escitalopram and paroxetine for the treatment of anxiety disorders (in which treatments were followed by a prospectively defined discontinuation period), Baldwin et al. (2007) reported that individuals taking either SSRI showed more discontinuation symptoms compared to placebo. Following a 12–24 week treatment period, patients were abruptly switched to placebo and were assessed using the DESS checklist 1 and 2 weeks after drug cessation. Overall, paroxetine withdrawal induced significantly more discontinuation symptoms than escitalopram.

There are a variety of symptoms associated with paroxetine withdrawal as Murata et al. (2010) recently provided based on observations of a group of Japanese outpatients (n = 56). These symptoms included dizziness (50% of patients), increased dreaming/vivid dreams (35%), fatigue (30%), nausea/vomiting (30%), headache (25%), anxiety/agitation (25%), paresthesia (15%), insomnia (10%), diarrhea (10%), visual disturbances (10%), fever (10%), tremor (5%), irritability (5%), and chills (5%). In addition to these symptoms, in an assessment of 87 patients who had their treatment interrupted for 4–7 days in a double-blind placebo study, Hindmarch et al. (2000) also reported greater cognitive deficits, poorer quality of sleep and increased depressive symptoms associated with paroxetine discontinuation, symptoms which were not evident in patients ceasing fluoxetine, sertraline, or citalopram treatment.

To date, there is no evidence to suggest that the length of SSRI treatment is associated with the development of more symptoms or with the severity of those symptoms (Rosenbaum et al., 1998; Michelson et al., 2000; Baldwin et al., 2007). However, several studies suggest that an abrupt interruption of treatment results in more symptoms of the discontinuation syndrome compared to a gradual tapering of the drug (van Geffen et al., 2005; Himei and Okamura, 2006; Murata et al., 2010). This suggests that the incidence, timing, and severity of SSRI discontinuation symptoms may be related to plasma elimination characteristics of each drug. In view of this, Michelson et al. (2000) found a statistically significant relationship between the percentage reduction in plasma concentration of the drug and the appearance of discontinuation symptoms resulting from placebo substitution across all drug groups (fluoxetine, sertraline, and paroxetine). Comparing drug concentrations before and after placebo substitution, Henry et al. (2000) found that lower steady-state brain levels of paroxetine (but not fluoxetine) after placebo substitution were associated with greater risk for discontinuation-related adverse events. Overall, discontinuation symptoms are more frequent after the abrupt cessation of drugs with shorter half-lives (Rosenbaum et al., 1998; Bogetto et al., 2002; Judge et al., 2002; Baldwin et al., 2007). However, that relationship may not be an absolute predictor. Indeed, despite paroxetine having a similar half-life to that of fluvoxamine, the rate of withdrawal reactions for the latter drug is 10 times lower (Price et al., 1996). For example, in the 46 patients with discontinuation symptoms reviewed by Black et al. (2000), paroxetine was stopped in 65% of these cases compared to only 7% of cases with fluvoxamine. To date, there are no double-blind trials with placebo substitution comparing the effects of discontinuation from paroxetine and fluvoxamine, and the higher number of reports of withdrawal reactions with paroxetine is likely to reflect the greater prescription rate of paroxetine over fluvoxamine (Lockhart and Guthrie, 2011). Other factors (i.e., pharmacological) could also contribute to the high frequency of discontinuation syndrome with paroxetine.

Notably, paroxetine has significant off-target effects, including an affinity for cholinergic receptors similar to the tricyclic antidepressants (Owens et al., 1997, 2001), raising the possibility for a cholinergic rebound during discontinuation. Fujishiro et al. (2002) further demonstrated that the ability of paroxetine to induce an anticholinergic effect in mice was similar to that of tricyclic antidepressants, whereas fluvoxamine was much less potent in that regard. Paroxetine also exhibits moderately high affinity for the norepinephrine transporter, unlike other SSRIs (Tatsumi et al., 1997). In comparison, the longer half-lives of the active compounds of fluoxetine and sertraline may be the reason underlying fewer symptoms of the discontinuation syndrome. However, as raised previously, the major limitation of most of the clinical studies examining the discontinuation syndrome is the restriction of treatment interruption to only few days, when longer periods (i.e., weeks) might be more suitable for the study of fluoxetine withdrawal. Finally, whether or not the pharmacodynamic characteristics of a drug could also help to explain the clinical observations remains to be addressed. For instance, fluoxetine is also a 5-HT_2C receptor antagonist (Sanchez and Hyttel, 1999) with affinity for 5-HT_2A/2C receptors (see Table 1), whereas sertraline is the sole SSRI which also inhibits the dopamine transporter (Tatsumi et al., 1997).

The discontinuation syndrome has been reported in relation to nearly every SSRI, although there are increased reports of occurrence in patients ceasing treatment of paroxetine. However, the full extent of discontinuation syndrome might only emerge if future studies are designed to take into account the pharmacokinetics of the drug and availability of its active metabolite. Double-blind trials with placebo substitution have yet to be undertaken to directly compare the effects of discontinuation from SSRIs with shorter half-lives (i.e., paroxetine and fluvoxamine). Overall, the crucial question of whether the discontinuation syndrome is equally related to the lack of specificity and/or the neurobiology related to the half-life of the SSRI remains unclear. Potential genetic-drug interactions should also be systematically assessed in future studies. Indeed, a recent clinical study suggested a possible involvement of the C(-1019)G polymorphism of the serotonin 5-HT_1A receptor gene in the occurrence of paroxetine discontinuation syndrome, as patients with the −1019C allele experienced paroxetine discontinuation syndrome more frequently than patients who were −1019G homozygous (Murata et al., 2010).

Although initial observations relied only on case reports, there is now a substantial body of clinical evidence suggesting that, in some cases, SSRIs may be associated with a withdrawal response when halted after a period of regular use. The existence of these symptoms, known as the SSRI discontinuation syndrome, has now been confirmed by a number of well-controlled studies (i.e., double-blind randomized placebo-controlled design, in which treatment is followed by a prospectively defined discontinuation period). These studies suggest that the best route of action for cessation of SSRI treatment is to taper down the dose of the medication rather than abrupt termination, as tapering is likely to decrease the possibility of the occurrence of discontinuation symptoms. Due to the broad range of symptoms which can develop as part of the discontinuation syndrome, prior to drug initiation, patients and caregivers need to be provided with adequate education and realistic, objective appraisals of possible outcomes which can develop during and following antidepressant treatment.

The current evidence suggests that the discontinuation syndrome is dependent on the SSRI half-life, with more reports of symptoms occurring in patients treated with paroxetine compared to other SSRIs. However, studies designed to assess the onset of discontinuation syndrome, and how the syndrome coincides temporally with pharmacokinetic withdrawal, are still lacking. Such studies are critical in order to draw conclusions when comparing SSRIs with short versus long half-lives. It is currently unclear as to whether the discontinuation syndrome is equally related to the pharmacological properties of a given SSRI, and/or its half-life. Future studies could also examine possible associations of the syndrome with clinical characteristics, as one study reported that the discontinuation syndrome was more common in patients with earlier onset of dysthymic disorder, and was also more common in females (Bogetto et al., 2002). On the other hand, Baldwin et al. (2007) reported no difference in discontinuation symptoms in patients with depression compared to patients with anxiety disorders. Future clinical investigations need to have enough statistical power to enable examination of within-group comparisons. Current understanding of the pathophysiology associated with the SSRI discontinuation syndrome remains largely speculative (Blier and Tremblay, 2006; Delgado, 2006). In fact, the sole clinical investigation looking at the possible chemical and molecular mechanisms underlying the SSRI discontinuation syndrome was based on a single subject (Kaufman et al., 2003).

Notably, not all patients treated with SSRIs (which represents a very heterogeneous population) experience discontinuation symptoms. In that regard, a recent clinical study indicated a possible involvement of the C(-1019)G polymorphism of the serotonin 5-HT_1A receptor gene in the occurrence of paroxetine discontinuation syndrome (Murata et al., 2010). Whether the development of discontinuation symptoms has a genetic component requires additional studies to provide more conclusive results. The use of genetic animal models might be able to shed light in that regard. Interestingly, using mice with higher (1A-High) or lower (1A-Low) autoreceptor levels, Richardson-Jones et al. (2010) found a negative relationship between 5-HT_1A autoreceptor level and response to antidepressants. Further studies looking at the effects of 5-HT_1A receptor function on vulnerability to the discontinuation syndrome are of interest. Most of the animal studies performed thus far have naturally focused on the serotonergic system, and have mainly used fluoxetine, despite the fact that this specific SSRI has been associated with a lower incidence of discontinuation syndrome in the clinic. However, based on the numerous targets known to be changed adaptively during chronic treatment with SSRIs in animal models, further studies assessing non-serotonergic pathways would be worthwhile. Finally, preclinical studies using appropriate animal models of anxiety/depression are still lacking when it comes to the study of SSRI discontinuation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.