Authors: Bryan Shapiro, Daniel Cohrs
Categories: Commentary
Source: Journal of Psychiatry & Neuroscience : JPN
Doi: 10.1503/jpn.250054
Authors: Bryan Shapiro, Daniel Cohrs
Stopping treatment with serotonin reuptake inhibitors (SRIs) often leads to withdrawal symptoms, which can be mitigated by a slow, hyperbolic taper using subtherapeutic dosage strengths. Unfortunately, conventional drug formularies lack the necessary breadth of dosing options to gradually wean SRIs, and alternative methods (e.g., bead counting, homemade dilutions, compounding) are difficult to implement for many patients. It has been suggested that fluoxetine, a widely available antidepressant with an unusually long elimination half-life, can help patients successfully discontinue SRIs, but the technique is poorly characterized. We propose a standardized fluoxetine substitution protocol that facilitates the discontinuation of compatible SRIs while minimizing adverse events.
Patients stopping serotonin reuptake inhibitor (SRI) antidepressants are susceptible to SRI withdrawal, a constellation of somatic and cognitive-emotional symptoms usually occurring within 24–48 hours of stopping or reducing dosages. The syndrome may follow a protracted course and is often associated with substantial morbidity and functional impairment.1,2 Systematic reviews have reported the incidence of antidepressant withdrawal to be 33%–56%, based on short-term studies.3,4
In recent years, the high rates of SRI withdrawal despite ostensibly gradual taper protocols have been explained by positron emission technology (PET) radioligand neuroimaging data. These studies show that about 80% occupancy of the serotonin transporter (SERT) occurs at minimum therapeutic doses of SRI antidepressants, with considerable occupancy even at subtherapeutic doses, as represented by a hyperbolic curve (Table 1).5–7 To avoid precipitous changes in SERT occupancy, which are thought to induce neurobiological instability, dosage reductions should respect this hyperbolic relationship. Progressively lower subtherapeutic doses must be incorporated into an SRI taper to maximally attenuate the risk of withdrawal symptoms.5–9
Currently, available dosage strengths in standard drug formularies are too limited to support gradual SRI tapers (Table 2). For instance, the lowest marketed dose of duloxetine (20 mg) averages at least 70% SERT occupancy with short-term treatment, 9 and 10 mg of citalopram averages at least 60% SERT occupancy. 9 Discontinuation at these dosages is a precipitous and high-risk discontinuation strategy. Several selective SRIs (SSRIs) are available in prepackaged prescription liquid form, but newer SRIs — including the serotonin–norepinephrine reuptake inhibitors (SNRIs), vilazodone, and vortioxetine — are not available as liquids (Table 2). To obtain dosages at lower SERT occupancies, patients often split small unscored tablets, open capsules to count beads or weigh powder, self-manufacture liquid dilutions, or rely on a compounding pharmacy.
Although a number of leading guidelines clearly recommend hyperbolic tapering of SRIs (which can involve transitioning to liquid formulations to obtain the needed doses), for some, these methods can be prone to error, time consuming, and expensive.6,12 Certain patients may lack the physical or cognitive capacity to consistently carry out extemporaneous tapering methods. Psychiatric providers often receive limited training in deprescribing psychotropics, including hyperbolic tapering as a means of safely and successfully deprescribing SRIs.8
In light of current deprescribing challenges at the level of the pharmaceutical formulary, patient, and clinician, there is a clear need for novel antidepressant tapering approaches that are safe and effective but also practical and easy to implement for patients.
Fluoxetine is a racemate of R(−) and S(+) enantiomers that demonstrates a high binding affinity for SERT (1.1 nM) in human receptor assays.13 It is highly selective for SERT, with relatively modest affinities for other neuronal binding sites (5-HT^2C^ 72 nM; norepinephrine transporter [NET]: 599 nM; M1 702 nM; dopamine transporter [DAT]: 3764 nM).13 Its sole active metabolite, norfluoxetine, achieves similar serum levels in humans and is roughly equipotent in terms of SERT inhibition.13,14
Compared with other SRIs, fluoxetine has an unusually long elimination half-life, with a slow decline in plasma concentration and SERT occupancy after reducing or discontinuing treatment (Table 2). Fluoxetine’s half-life is 1–3 days after acute administration and extends to 4–6 days with chronic dosing to inhibit its own metabolism to norfluoxetine via cytochrome CYP2D6. The half-life of norfluoxetine is 4–16 days.15 Thus, norfluoxetine accumulates more slowly than fluoxetine, and comparable levels of these 2 active moieties are reached after 30 days.15 Fluoxetine’s long combined half-life was the basis for Fluoxetine Weekly, a 90-mg once-weekly capsule marketed for the treatment of mood disorders. 15 Thus, fluoxetine can be administered at less-than-daily frequencies, which essentially mimics once-daily dosing with subtherapeutic dosage strengths. As a strong inhibitor of CYP2D6, fluoxetine inhibits the metabolism of several SRIs,15,16 effectively increasing their elimination half-life and theoretically reducing the risk of withdrawal reactions when the SRI is discontinued. Notably, in vitro studies show that S-fluoxetine and S-norfluoxetine’s binding affinity for 2D6 are both less than 100 nmol, suggesting inhibition even at subtherapeutic dosages based on typical serum concentrations. 16 Strong CYP2D6 inhibition may be particularly helpful in tapering high-risk SRIs such as paroxetine and venlafaxine, which have short half-lives but are predominantly metabolized by CYP2D6.17,18 To date, this feature of fluoxetine has not been specifically recognized. Fluoxetine and norfluoxetine also appear to cause mild-to-moderate inhibition of CYP2C19 and CYP3A4,19 although the clinical significance of this association is unclear.
In terms of its dosage formulary, fluoxetine is a generic drug widely manufactured as a scored tablet that, when split, achieves a 5 mg dose that is one-quarter of its minimum therapeutic dose of 20 mg/d (Table 2). Fluoxetine is also available as a liquid formulation that can be precisely titrated to various subtherapeutic dosage strengths.15 If we posit that successfully tapering an SRI antidepressant requires a gradual reduction to very low SERT occupancies to mitigate withdrawal symptoms, fluoxetine is an optimal choice. Splitting scored 10-mg tablets of fluoxetine to 5-mg halves, given weekly, equates to a daily dose of 0.7 mg, which has an estimated SERT occupancy of about 23%.20 By contrast, the lowest commercially available solid dosages of other SRIs typically occupy 60%–70% of SERT9 and incur substantial risk of withdrawal when discontinued.
Several randomized, placebo-controlled studies have reported a lower incidence of discontinuation-emergent withdrawal symptoms with fluoxetine compared with other SRIs.21–23 In a study involving 242 patients with remitted depression, Rosenbaum and colleagues evaluated maintenance fluoxetine, sertraline, or paroxetine (treatment duration 4–24 mo) and assessed withdrawal symptom scores occurring 5–8 days after placebo substitution.21 Withdrawal symptom score changes were significantly higher in the sertraline and paroxetine groups than the fluoxetine group (p < 0.001), as assessed by the 43-item Discontinuation Emergent Signs and Symptoms checklist (DESS), and no changes in withdrawal symptom scores were observed in the fluoxetine group (p = 0.578). Such comparative studies have been criticized for discontinuation periods that are likely too short (≤ 1 wk) to detect fluoxetine withdrawal, considering its unusually long elimination half-life. To date, only Zajecka and colleagues have examined longer-term fluoxetine withdrawal symptoms.24 Among 395 patients with depression on a 12-week course of fluoxetine (20 mg/d), those who were abruptly switched to placebo for 6 weeks had equivalent rates of new or worsening adverse events at week 1, week 2, week 4, and week 6, compared with those maintained on fluoxetine. Rates of dizziness were significantly higher in the placebo-substituted group at week 4 (7% v. 1%, p = 0.023), which persisted through week 6 (5% v. 1%, p < 0.048), suggesting the possibility of delayed-onset withdrawal in fluoxetine-treated patients. However, SRI withdrawal symptoms were not systematically evaluated (e.g., via the DESS scale) in this study, and the 12-week treatment period in this study was relatively short.
Two retrospective studies have compared fluoxetine’s withdrawal risk to other SRIs. In a retrospective chart review assessing the incidence of withdrawal symptoms among 171 patients who discontinued SRI treatment, no incident cases of fluoxetine withdrawal were noted, although the fluoxetine sample was quite small (20 patients) and the withdrawal monitoring methodology was poorly defined.25 A pharmacovigilance analysis of 31 688 reports from the World Health Organization’s Spontaneous Reporting Database, using naltrexone as a control, found lower odds of withdrawal with fluoxetine (odds ratio [OR] 3.18) relative to paroxetine (OR 45.07) and sertraline (OR 7.22), a risk that was similar to tricyclic antidepressants.26
Clinicians have previously considered fluoxetine’s potential role in helping patients discontinue SRI antidepressants, including both SSRIs and SNRIs, at least partially on the basis of its long elimination half-life and potentially lower risk or severity of withdrawal symptoms.
Giakas and Davis documented 3 patients experiencing intractable withdrawal symptoms after stopping chronic treatment with venlafaxine.27 Venlafaxine was restarted at dosages that relieved withdrawal symptoms (18.75 mg/d, 37.5 mg/d, and 150 mg/d, respectively), and fluoxetine was started at 10–20 mg/d and maintained for 4 weeks. For all patients, the combination was well tolerated, and venlafaxine was subsequently discontinued without recurrence of withdrawal symptoms.27 One patient attempted to discontinue fluoxetine at 10 mg/d and had no signs of withdrawal.
A separate case series recounted 2 patients experiencing intolerable withdrawal symptoms after discontinuing subtherapeutic doses of venlafaxine and sertraline. The first patient, stable on venlafaxine 75 mg/d for several months, was tapered to 37.5 mg/d for 7 days, then 18.75 mg/d for 10 days before discontinuation. After experiencing signs of venlafaxine withdrawal — including dizziness, anxiety, agitation, and nausea — venlafaxine was reinstated at 75 mg/d, resulting in complete resolution of her symptoms within 1 day. The patient was subsequently started on fluoxetine 20 mg/d and re-tapered successfully, experiencing mild paresthesia and anxiety lasting 1 week. The fluoxetine was reduced 1 week later to 10 mg/d for 5 days, then discontinued without signs of withdrawal over subsequent weeks. In the second case, a patient stable on sertraline 50 mg/d for several months was tapered at 25 mg/d for 10 days, then 12.5 mg/d for 10 days before discontinuation, but experienced intolerable withdrawal symptoms. After reinstating sertraline 25 mg/d for 1 month, the patient was tapered again to 12.5 mg/d for 1 week before discontinuation, taking fluoxetine 5 mg/d from the start. Sertraline was successfully tapered without onset of withdrawal symptoms, and fluoxetine was discontinued without subsequent observations of withdrawal.28
In another case report by the same author, a patient was treated for at least 1 year with clomipramine (150 mg/d), then tapered weekly until discontinuing a 37.5 mg/d dose. Because of withdrawal symptoms, clomipramine 75 mg/d was reinstated, resulting in resolution of symptoms. Fluoxetine 20 mg/d was introduced for 1 week and clomipramine was tapered with no recurrence of withdrawal. Fluoxetine was then successfully tapered over the course of a week.29
Bet and colleagues documented a 48-year-old woman who had been taking venlafaxine for 13 years, initially prescribed for an adjustment disorder with depressed mood.30 Because of symptomatic improvement and possible iatrogenic adverse effects of excessive anxiety, tension, and sweating, the patient began to taper her venlafaxine (75 mg/d). She experienced severe headaches when venlafaxine was discontinued at 37.5 mg/d, which she also experienced when missing a maintenance dose of venlafaxine. Venlafaxine was reinstated at 37 mg/d with resolution of her headaches. She was unable to afford lower compounded dosages of venlafaxine, and a consulting pharmacist recommended a direct switch to fluoxetine 10 mg/d for 2 weeks. During this switch, the patient experienced a reduction in anxiety, tension, and sweating, but with onset of diarrhea that disappeared after stopping fluoxetine. The patient subsequently stayed off antidepressants for several months without reported withdrawal symptoms.
Several methodological limitations of these case studies bear mention. Patients were initially tapered from their SRI rapidly (dose reductions every 7–10 d), so it is unclear whether these were truly taper-resistant cases necessitating fluoxetine substitution. With the exception of the patient described by Bet and colleagues,30 patients were also treated for up to around 1 year with their SRI, a duration that is often shorter than seen in clinical practice and hence may incur a lower risk of severe withdrawal reactions. It is also unclear how carefully patients were monitored for symptoms or adverse effects after starting adjunctive fluoxetine; further, all but 1 case report did not specify a follow-up period after ultimate discontinuation of fluoxetine, so it is unclear whether the monitoring period was sufficient to observe any potential withdrawals that may have emerged. To date, no controlled studies have evaluated the efficacy or tolerability of fluoxetine substitution, including comparisons to conventional hyperbolic tapers of SRI antidepressants.
Several factors may indicate that a patient is a candidate for fluoxetine substitution. Ideal patients include those who have been prescribed a potent SRI that is relatively selective for SERT (Table 3), those with at least 1 previous unsuccessful taper of a prescribed SRI using its conventional dosage formulary, those taking an SRI with an extremely limited dosage formulary, those unable to taper using alternative dosing techniques (e.g., liquids, beads, powders), those unable to access or afford compounding services, those with no history of serious adverse effects or allergies associated with fluoxetine, those not taking medications affected by CYP2D6 inhibition (e.g., tamoxifen), and those with excellent adherence to medication and capacity for close follow-up and monitoring.
Before trialling fluoxetine substitution, we recommend at least 1 attempt to gradually taper the patient’s SRI, if reasonably practicable. In many cases, a sufficiently gradual taper is not possible because of patient- and pharmacy-related challenges. It is essential that candidates for fluoxetine substitution have a history of excellent adherence to their medication regimens and be highly motivated to taper their antidepressant. Patients must be willing to follow up closely with their providers (often weekly or every other week) to monitor for decompensation, including suicidal ideation.6
Fluoxetine substitution is not appropriate for patients who have previously had adverse effects with fluoxetine or who are taking other medications (e.g., tamoxifen, metoprolol, certain opioids) whose activity may be affected by fluoxetine’s potent enzymatic inhibition of CYP2D6. This method is also contraindicated for tapering monoamine oxidase inhibitors owing to the risk of serotonin syndrome (Table 4).
Based purely on fluoxetine’s high selectivity for SERT, fluoxetine substitution is most appropriate for discontinuing potent and selective SRIs (Table 3). Most SRIs approved for the treatment of mood disorders are highly selective for SERT, with affinities for other receptors that are several orders of magnitude lower.13,18,31 For instance, the SNRI venlafaxine has a binding affinity for human NET (2480 nM) that is roughly 30-fold less than for SERT (82 nM),33 although NET inhibition has been demonstrated within its therapeutic dosage range.34,35 Fluoxetine substitution can be considered for SNRIs that are stronger reuptake inhibitors of norepinephrine (e.g., desvenlafaxine, duloxetine, and levomilnacipran), but patients may be potentially susceptible to norepinephrine withdrawal not addressed by fluoxetine substitution. Although many tricyclic antidepressants are classified as serotonin and norepinephrine reuptake inhibitors, only clomipramine and imipramine are considered potent SRIs.36
Theoretically, fluoxetine substitution would not address withdrawal from non-SERT receptor targets that may be independently associated with withdrawal reactions. Examples include NET, DAT, and postsynaptic targets (e.g., histamine H1 receptor, cholinergic M1 receptor). Several newer SRIs interact with these targets, although corresponding binding affinities are often modest, and the clinical importance of these effects is controversial (Table 3).36 Tricyclic antidepressants also generally have high affinities for other receptor targets, including NET, histamine H1 receptor, and cholinergic M1 receptor, often substantially higher than for SERT itself.36 Thus, fluoxetine substitution is not recommended for tricyclic antidepressants other than clomipramine and imipramine (Table 4).
Overall, citalopram, escitalopram, fluvoxamine, paroxetine, sertraline, venlafaxine, vilazodone, and vortioxetine appear to be most suited for fluoxetine substitution, given their potency and relative selectivity for SERT, although citalopram, escitalopram, and sertraline are generally available in liquid formulations and paroxetine and venlafaxine do have off-target effects of variable potency. Duloxetine, paroxetine, and venlafaxine are also considered particularly high risk in terms of withdrawal, and bind potently to SERT, although they have considerable non-SERT activity (M1 antagonism in the case of paroxetine;31 NET inhibition in the case of venlafaxine and duloxetine).
Our suggested approach to fluoxetine substitution includes 4 steps.
Before introducing fluoxetine, the original SRI should be tapered as gradually as practicable to the lowest tolerated dosage in the standard drug formulary (e.g., 5 mg/d for escitalopram or vortioxetine or 12.5 mg/d for sertraline by splitting the scored 25 mg tablet). Tapering the SRI as much as possible before introducing fluoxetine may reduce the risk of serotonergic and other adverse effects associated with SRI cotreatment. The initial phases are typically well tolerated as changes in SERT occupancy are modest within the therapeutic dosage range.5,6,37 Withdrawal from SERT and other lower-affinity targets (e.g., NET) may occur even at relatively high SRI dosages and requires close monitoring. Of note, some patients may experience notable serotonin withdrawal symptoms within the typical therapeutic dose range, suggesting that typical dose-occupancy frameworks may not fully predict withdrawal outcome. The patient should be monitored for serotonin withdrawal (Figure 1) and withdrawal from other receptor targets. Norepinephrine withdrawal symptoms (e.g., related to duloxetine or clomipramine) may include fatigue, increased appetite, low mood, and diarrhea. Anticholinergic (e.g., paroxetine, tricyclic antidepressants) withdrawal symptoms often include nausea, diarrhea, sweating, and urinary urgency.
The taper process should be sufficiently gradual (e.g., initially, 50% dosage reduction every 4 weeks) to mitigate acute withdrawal symptoms (Table 5). If intolerable withdrawal symptoms occur, the previous dose should be reinstated, followed by a more gradual dosage reduction (e.g., 25% dosage reduction every 4 weeks or the smallest dosage reductions allowed by the standard dosage formulary), if possible, before moving to the next step.
For example, for a patient on sertraline 100 mg/d planning a fluoxetine substitution, a gradual sertraline taper should be attempted (e.g., 50% dose reduction, carried forward, every 4 weeks) before introducing fluoxetine. Should withdrawal symptoms occur after discontinuation of the 12.5 mg/d dose, the patient would resume the 12.5 mg/d dose to relieve withdrawal symptoms. After stabilizing at this dose, fluoxetine can be started at 5 mg/d and maintained for 4 weeks while sertraline 12.5 mg/d is continued (Table 5).
Some authors have proposed direct switches from an SRI to fluoxetine either before or during the emergence of withdrawal symptoms, rather than implementing an initial cotreatment phase.32,38 This approach incurs unnecessary risk of serotonin withdrawal because blood levels of fluoxetine and norfluoxetine take several weeks to reach steady state and sufficient SERT occupancy is slow to accumulate. The level of SERT occupancy associated with a single dose of fluoxetine has not been evaluated in humans9 and is estimated to be markedly lower than that of chronically dosed fluoxetine (28 d of treatment), given fluoxetine’s long half-life. In a direct switch from SRI to fluoxetine, the resulting serotonin withdrawal symptoms may be difficult to distinguish from fluoxetine-associated adverse effects (e.g., nausea, anxiety, agitation, insomnia) 6 and may compromise the success of the taper. This risk is mitigated by the approach explained in Step 2.
Based on estimates modelled from PET radioligand data, 4 weeks of fluoxetine treatment at 5 mg/d results in an average 60% SERT occupancy (Table 1).7,9 At 60% occupancy, fluoxetine should mitigate withdrawal effects when the original SRI is tapered and discontinued. Significantly higher doses of fluoxetine (e.g., 15–20 mg) may unnecessarily increase the risk of adverse effects without commensurate benefit. At 4 weeks, blood fluoxetine concentrations are likely to be at steady state, whereas norfluoxetine levels may modestly increase for an additional 6 weeks in some individuals.
Co-treatment with fluoxetine and an SRI, even if temporary, may somewhat increase the risk of adverse events. In general, time-limited SRI co-treatment often occurs in clinical practice when drugs are being cross-titrated (e.g., SSRI to SSRI or SNRI). Fluoxetine is a strong inhibitor of CYP2D6 and may increase serum concentrations of several SRIs metabolized by CYP2D6 (e.g., paroxetine, vortioxetine), increasing the risk of adverse effects when combined.16 At the same time, this inhibition may mitigate the severity of withdrawal from these antidepressants by increasing their half-lives. Inhibition of CYP2C19 and CYP3A4 by fluoxetine and norfluoxetine may also be clinically relevant, influencing serum levels of antidepressants such as citalopram, escitalopram, and vilazodone.
Adding fluoxetine to an SRI may theoretically increase the risk of serotonin syndrome. However, severe, life-threatening forms of serotonin toxicity appear to occur only when SRIs are combined with monoamine oxidase inhibitors. 39,40 Moreover, overdoses of SRIs alone do not cause severe forms of serotonin syndrome, and moderate serotonin toxicity occurs in only 1 in 7 overdoses.39,41 The combination of fluoxetine and SRI we have described is also unlikely to be high risk considering that the dose of fluoxetine is low; fluoxetine and other SRIs, other than vilazodone, 42 compete for SERT occupancy, such that dual-SRI treatment resembles the risk of higher-dose monotherapy with a single SRI; and fluoxetine substitution is implemented when the dose of the original SRI is low (after Step 1) and SERT occupancy is well below thresholds concerning for serotonin toxicity.5
After 4 weeks of treatment with fluoxetine, circulating fluoxetine and norfluoxetine form a SERT occupancy buffer for tapering of the original SRI, which can now be implemented. This taper should proceed at a reasonable pace and to the lowest available conventional dosage (e.g., splitting scored tablets or capsules) before the drug is completely stopped. Ideally, because of the fluoxetine–norfluoxetine buffer, the patient would not need to resort to alternative dosing strategies (e.g., liquid formulations, bead counting, powder weighing, compounding) to successfully discontinue the original SRI, even if previously necessary. We suggest an initial taper schedule of 50% dose reductions every 4 weeks to the lowest available conventional dosage before discontinuing the SRI. If intolerable withdrawal symptoms occur at this pace, the lowest tolerable dose should be reinstated, followed by a more gradual dosage reduction (e.g., 25% reduction every 4 weeks), if commercially available.
If intolerable serotonin withdrawal symptoms occur at any point during the taper of the original SRI, a higher fluoxetine dose may be required. Because of potential individual variability in blood concentrations and SERT occupancies at a given SRI dose,14,43 5 mg/d of fluoxetine, despite occupying 60% of SERT on average,7,9 may not adequately mitigate withdrawal in a subset of individuals. In these cases, the previous dose of the original SRI should be reinstated, followed by a stabilization period, wherein fluoxetine is increased to 10 mg/d. After 4 weeks of co-treatment, discontinuation can be reattempted.
After the patient has fully discontinued their original SRI, we recommend monitoring for at least 4 weeks to assess for withdrawal symptoms longitudinally and to ensure stability on fluoxetine before tapering. After this period, fluoxetine should be switched from 5 mg/d to a a once-weekly dosing phase, with reductions occurring every 4 weeks, starting at 20 mg/week, followed by 10 mg/week, then 5 mg/week, then full discontinuation (Figure 2). This schedule yields equivalent daily fluoxetine dosages of 2.9 mg/d, 1.4 mg/d, and 0.7 mg/d, respectively, which corresponds to an estimated SERT occupancy as low as 23%.20 Fluoxetine’s active metabolite, norfluoxetine, sees only small reductions (10%–15%) in plasma concentration after 1 week of treatment cessation, and so weekly dosing with fluoxetine is feasible.44
Because of fluoxetine’s long elimination half-life, clinicians should monitor for withdrawal symptoms for several months after its discontinuation. If withdrawal symptoms occur and are sufficiently severe, fluoxetine 5 mg/d can be reinstated, followed by a more gradual taper using a lower dosage (e.g., 2.5 mg or 1.25 mg weekly via a liquid formulation) or with less frequent dosing.
Fluoxetine substitution may support the discontinuation of difficult-to-taper SRI antidepressants and has been implemented successfully, as evidenced by case reports. The standardized approach presented in this commentary strikes a balance between mitigating risk of withdrawal symptoms and risk of adverse effects associated with SRI co-treatment. Limitations of this approach should be considered, including fluoxetine’s selective binding profile, the lack of clinical trials examining fluoxetine substitution, and the potential of a patient having difficulty tapering the original SRI who may not be able to tolerate the drug changes entailed by this method. Overall, given avoidably high rates of antidepressant withdrawal, pervasive limitations in drug formularies, and patient-centred challenges in implementing alternative tapering strategies, this approach warrants empirical validation for consideration in clinical practice.