Authors: Joseph F. Goldberg, Michael Q. Steinman, Stephen M. Stahl
Categories: Review, Systematic review, Antidepressants, Trazodone once-a-day (OAD), Adverse events, Activating effects, Sexual dysfunction
Source: Neuroscience Applied
Authors: Joseph F. Goldberg, Michael Q. Steinman, Stephen M. Stahl
Psychotropic drug properties can vary across different formulations of a compound. Trazodone is a multifunctional drug that exemplifies this phenomenon. Although it is a serotonin reuptake inhibitor, trazodone is a significantly more potent blocker of serotonin 5-HT2A receptors. It also acts as an antagonist at several other 5-HT receptors and acts as a partial agonist at 5-HT1A receptors. Additional targets of trazodone blockade include histamine H1 receptors and alpha-1 adrenergic receptors. Trazodone's newer once-a-day (OAD) formulation is thought to have improved safety and tolerability over earlier formulations due to release kinetics that avoid the multiple higher peaks in plasma concentrations observed with earlier formulations. Here, we systematically reviewed the clinical data from randomized controlled trials (RCTs). We performed a PubMed and Cochrane Library database search for RCTs on trazodone OAD that either included placebo or active comparator controls or used a crossover design with at least one additional treatment. Three studies met inclusion criteria. Trazodone OAD was compared to placebo in one study, to venlafaxine extended-release (XR) in another, and to trazodone immediate-release in an additional study. Trazodone OAD demonstrated a favorable safety and tolerability profile across studies and was associated with low levels of sexual dysfunction and weight gain. It improved sleep disturbance scores and was associated with time point-dependent improvements in anxiety/somatization scores on the 17-item Hamilton Rating Scale for Depression (HAM-D-17) as compared to placebo or venlafaxine XR, suggesting that it does not tend to promote anxiety or insomnia. Trazodone OAD represents a well-tolerated treatment for MDD patients who are concerned by anxiety symptoms and are seeking to minimize adverse drug effects involving sexual dysfunction, weight gain, and psychomotor activation.
Balancing efficacy and tolerability poses a significant clinical challenge in the pharmacotherapy of major depressive disorder (MDD). Adverse drug effects occur in nearly two-thirds of antidepressant-treated patients with MDD (Gartlehner et al., 2011) and are a significant contributor to nonadherence (Ashton et al., 2005; Bull et al., 2002; Demyttenaere et al., 2001). Studies suggest that between 23% and 43% of MDD patients discontinue antidepressant treatment due to adverse events (AEs) such as fatigue, anxiety, headache, insomnia, sexual dysfunction, and weight gain (Ashton et al., 2005; Bull et al., 2002; Demyttenaere et al., 2001). Antidepressants with a more tolerable safety profile or low levels of specific AEs that are intolerable to a particular patient may reduce nonadherence and represent an unmet need in the treatment of MDD (Rakesh et al., 2017; Gartlehner et al., 2012).
One approach to addressing safety and tolerability concerns with antidepressants and other psychotropic drugs may be to control how the drug is delivered by leveraging different available formulations. This may permit a more individualized approach to treatment (Fuller et al., 2013). Drug delivery can impact safety and tolerability, as formulations with greater fluctuating or peak plasma concentrations are associated with increased rates of AEs (Ereshefsky and Dugan, 2000; Stahl, 2009a). The multifunctional drug trazodone is thought to possess more than one therapeutic mechanism and comes in several formulations with different release kinetics (Stahl, 2009a, 2009b; Fagiolini et al., 2012). In this review, we will focus on the safety and tolerability of the once-a-day (OAD) formulation of trazodone, which generates smooth increases in plasma concentrations that reach lower peaks than earlier formulations and remain at antidepressant concentrations over a 24-h period (Stahl, 2009a; Hidalgo and Sheehan, 2010; Albert et al., 2021).
The most potent binding property of trazodone is serotonin 5-HT2A receptor antagonism (Stahl, 2009a; Kasper et al., 2005; Fagiolini et al., 2020). Trazodone displays moderate serotonin transporter (SERT) blockade that is 100-fold less potent than its 5-HT2A receptor antagonism, as well as partial serotonin 5-HT1A receptor agonism, serotonin 5-HT2C receptor antagonism, and histamine H1 receptor and alpha-1 adrenergic receptor blockade (Stahl, 2009a). At low doses, trazodone exhibits sedative-hypnotic effects due to 5-HT2A receptor, H1 receptor, and alpha-1 adrenergic receptor blockade. At higher doses ranging from 150 to 300 mg, trazodone exhibits antidepressant effects as a result of SERT inhibition and 5-HT1A partial agonism (Stahl, 2009a; Fagiolini et al., 2020; Sheehan et al., 2009a).
The antidepressant efficacy of trazodone has been shown to be significantly correlated with its steady-state plasma levels (Monteleone and Gnocchi, 1990). The minimum concentration of trazodone required to achieve antidepressant activity has been reported to be 0.5 mg/L (Albert et al., 2021). When administered at doses that induce sedative-hypnotic effects, trazodone concentrations only briefly, if ever, reach these levels (Stahl, 2009a). Three different formulations of trazodone are immediate-release (IR), prolonged-release (PR/AC), and the aforementioned OAD (Stahl, 2009a; Fagiolini et al., 2012; Monteleone and Delrio, 1993, Cešková et al., 2018). The IR formulation, which is short acting, is associated with a rapid peak and decline in plasma levels and has a recommended dosing schedule of three times daily (Stahl, 2009a). This can generate a sawtooth pattern of plasma trazodone concentrations that greatly exceed the minimum antidepressant concentration by approximately 2- to 3-fold for the majority of the day yet drop below the therapeutic threshold overnight (Stahl, 2009a). Such elevated trazodone concentrations can lead to daytime sedation in a subset of patients. The PR/AC formulation displays delayed and lower peak plasma levels than the IR formulation and may be administered twice daily (Fagiolini et al., 2012; Monteleone and Delrio, 1993, Cešková et al., 2018). In contrast, the OAD formulation contains specific excipients that generate a slow rise and fall in plasma levels, which remain above the therapeutic antidepressant activity threshold across a day following a single administration (Stahl, 2009a; Hidalgo and Sheehan, 2010). This trazodone OAD formulation has been reported to have a low incidence of sedation and may provide the best option for sustaining plasma concentrations that exceed the therapeutic antidepressant activity threshold while retaining adequate tolerability (Stahl, 2009a). We propose that when release kinetics of different formulations are considered, then trazodone may be an example of “different delivery–different drug.” This could allow for a specific trazodone formulation to be selected according to an individual patient's needs based on its pharmacokinetic profile (Stahl, 2009a).
The present study systematically reviews the results from randomized controlled trials (RCTs) examining the efficacy, safety, and tolerability of the OAD formulation of trazodone. In addition, we provided context by examining in a narrative review format how the AE profile of trazodone OAD compares to those of other antidepressants, with emphasis on selective serotonin reuptake inhibitors (SSRIs), serotonin and norepinephrine reuptake inhibitors (SNRIs), and the serotonin modulators and stimulators (SMSs) vortioxetine and vilazodone.
Comprehensive searches of PubMed and the Cochrane Library were performed using a priori search terms to conduct a systematic review of studies reporting the safety and tolerability of trazodone OAD. The search was limited to RCTs. Due to the manageable number of studies, including search cutoff dates proved to be unnecessary. The OAD formulation of trazodone has been referred to as prolonged-release or extended-release depending on the article or study (Stahl, 2009a; Fagiolini et al., 2012; Goracci et al., 2016), so we included variations on these terms in addition to variations on the phrase “once a day.” Specifically, the articles included in this review were identified using the following search In PubMed, ((trazodone [Title/Abstract]) AND (once a day [Title/Abstract])) OR ((trazodone[Title/Abstract]) AND (once daily [Title/Abstract])) OR ((trazodone[Title/Abstract]) AND (once-a-day [Title/Abstract])) OR ((trazodone [Title/Abstract]) AND (prolonged release [Title/Abstract])) OR ((trazodone[Title/Abstract]) AND (prolonged-release [Title/Abstract])) OR ((trazodone [Title/Abstract]) AND (extended-release [Title/Abstract])) OR ((trazodone [Title/Abstract]) AND (extended release [Title/Abstract])), and in the “title abstract keyword” search bar of the Cochrane Library, (trazodone once a day) OR (trazodone once-a-day) OR (trazodone prolonged release) OR (trazodone prolonged-release) OR (trazodone extended-release) OR (trazodone extended release). Studies were eligible for inclusion in the systematic review if they were RCTs that compared trazodone OAD to an active comparator or placebo or if trazodone OAD was compared to at least one different drug treatment (not a different dose of trazodone OAD) in a crossover design.
The specified search term criteria yielded 58 articles in PubMed and 41 articles in Cochrane Library, as shown in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart (Fig. 1). Of these, only three articles reported the RCTs that met the eligibility criteria for study inclusion of a placebo, active comparator, or crossover design that included at least one additional treatment and evaluated either the safety and tolerability profile of trazodone OAD in participants with MDD or the pharmacokinetic properties, safety, and tolerability of trazodone OAD in healthy participants (Fig. 1). One article reported the results of a randomized, double-blind, placebo-controlled study evaluating the safety and efficacy of trazodone OAD vs placebo for the treatment of MDD (Sheehan et al., 2009a). Another article reported the results of a randomized, double-blind, active-controlled, parallel-group study evaluating the efficacy and safety of trazodone OAD vs venlafaxine extended-release (XR) for the treatment of MDD (Fagiolini et al., 2020). The third article reported the results of three substudies from a comprehensive Phase I program that evaluated the pharmacokinetics, safety, and tolerability of trazodone OAD in healthy participants (Karhu et al., 2011). One of these studies was a single-dose comparative study consisting of a randomized, open-label, two-way crossover design comparing the bioavailability of trazodone OAD and trazodone IR (Karhu et al., 2011). Another was a multiple-dose comparative study that employed a randomized, open-label, two-way crossover design comparing the steady-state pharmacokinetic profiles of trazodone OAD and trazodone IR (Karhu et al., 2011). The third study was a food effect study that did not use a comparator against trazodone OAD and therefore was not included in the systematic review (Karhu et al., 2011). Here, we present an in-depth review of the three articles and discuss the results as they relate to the safety and tolerability of trazodone OAD (Fagiolini et al., 2020; Sheehan et al., 2009a; Karhu et al., 2011).Fig. 1Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chartThe initial literature search for randomized controlled trials (RCTs) on once-a-day (OAD) trazodone formulation identified 58 PubMed and 41 Cochrane Library articles, 13 of which were duplicate articles. Of the 86 articles screened, only three met the eligibility RCTs that included a placebo, an active comparator, or crossover design to evaluate the safety and tolerability of trazodone OAD in either patients with major depressive disorder or healthy participants. Studies that only examined different doses of trazodone OAD were also excluded.Fig. 1
It is worth briefly summarizing the efficacy data from the two RCTs that reported them to put the safety and tolerability data in better context. Sheehan et al. (2009a) compared a trazodone OAD (n = 202 intent-to-treat [ITT]) population to patients on placebo (n = 204 ITT). The authors observed a statistically significant difference in the change from baseline in the 17-item Hamilton Rating Scale for Depression (HAM-D-17) total score for individuals with MDD treated with trazodone OAD (dose range during the first 150–225 mg/day) as compared to placebo-treated individuals with MDD (5.6 vs 3.9, respectively; P = 0.005) within the first week of the double-blind treatment phase (Sheehan et al., 2009a). Significant differences in the change from baseline in the HAM-D-17 total score were maintained at all assessment points throughout the duration of the study (Sheehan et al., 2009a).
Fagiolini et al. compared trazodone OAD (n = 162 ITT) to venlafaxine XR (n = 152 ITT). This study also demonstrated a statistically significant difference in the change from baseline in the HAM-D-17 total score for individuals with MDD treated with trazodone OAD (dose during the first 150 mg/day) as compared to individuals with MDD treated with venlafaxine XR (dose during the first 75 mg/day) within the first week of the double-blind treatment phase (4.3 vs 3.5, respectively; P = 0.05) (Fagiolini et al., 2020). The antidepressant efficacy of trazodone OAD and venlafaxine XR did not differ significantly at any subsequent assessment points through the study endpoint (Fagiolini et al., 2020).
The HAM-D-17 total score includes measurement of anxiety/somatization and sleep disturbances/insomnia (Fagiolini et al., 2020; Sheehan et al., 2009a; Hamilton, 1960). These clinical features can be present as residual symptoms of MDD (McClintock et al., 2011) as well as treatment-emergent AEs from antidepressants (Sinclair et al., 2009). In the Fagiolini et al. study, trazodone OAD showed superiority over venlafaxine XR in improving the HAM-D-17 sleep disturbance score from baseline at almost all time points in both the ITT and per-protocol (PP; n = 122 trazodone, n = 127 venlafaxine XR) populations, including the final assessment at Day 56 for the PP population (P = 0.022) (Fagiolini et al., 2020). Although the anxiety/somatization score improved numerically for both treatments, the effect was significantly greater for venlafaxine XR vs trazodone OAD at Day 56 in both the ITT (−3.5 vs −4.5; P = 0.002) and PP populations (−4.4 vs −5.0; P = 0.038) (Fagiolini et al., 2020). Sheehan et al. did not report subanalyses for anxiety/somatization or sleep disturbances, although these as well as additional analyses of data from Fagiolini et al. were reported in a recent systematic review, which is discussed later (Albert et al., 2021).
The single-dose comparative bioavailability study in healthy participants compared the pharmacokinetic profile of single-dose administration of 300-mg trazodone OAD caplets with that of 100-mg trazodone HCl IR tablets administered as three doses, 8 h apart. Karhu et al. observed that the mean plasma concentration vs time profile following the administration of a single 300-mg oral dose of trazodone OAD demonstrated controlled release of the drug over a 24-h period (Karhu et al., 2011). The mean half-life of trazodone OAD was 11.8 h vs 8.3 h for 100-mg trazodone IR three times per day, every 8 h (Karhu et al., 2011). The area under the curve parameters for trazodone OAD were approximately 20% lower than those for trazodone IR, and the maximum serum concentration (Cmax) was 60% lower for the OAD formulation than for the IR formulation (Karhu et al., 2011).
The multiple-dose comparative study in healthy participants compared the steady-state pharmacokinetic profiles of 300-mg trazodone OAD caplets and 100-mg trazodone HCl IR tablets (Karhu et al., 2011). After a 2-day titration, participants received a 300-mg daily dose of the study medication for a period of 7 days, followed by a 2-day down-titration (Karhu et al., 2011). Karhu et al. observed that steady state was attained within 48 h of completion of the titration phase (Karhu et al., 2011). The pharmacokinetic profile was evaluated over a 24-h interval on Day 9, and the mean maximum and minimum steady-state concentrations were observed to be 43% and 27% lower, respectively, for the trazodone OAD formulation than for the trazodone IR formulation (Karhu et al., 2011). Variations in maximum and minimum trazodone concentrations were 42% and 31% lower, respectively, for the trazodone OAD formulation than for the trazodone IR formulation (Karhu et al., 2011).
Trazodone OAD was well tolerated in the efficacy and safety studies in individuals with MDD. In both the Sheehan et al. and Fagiolini et al. studies, the majority of AEs were mild to moderate. The most commonly observed AEs (≥5% in any treatment group) included headache, somnolence, dry mouth, dizziness, and nausea (Table 1) (Albert et al., 2021; Fagiolini et al., 2020; Sheehan et al., 2009a).Table 1Summary of the most frequent adverse events (AEs) reported with trazodone once-a-day (OAD) and its comparators in 3 randomized controlled trials.Table 1Most frequent AEs (≥5% in any group)Sheehan et al. (Sheehan et al., 2009a)Fagiolini et al. (Fagiolini et al., 2020)Karhu et al. (Karhu et al., 2011)Single-dose comparisonMultiple-dose comparisonTrazodone OAD (n = 202)Placebo(n = 204)Trazodone OAD (n = 165)Venlafaxine XR (n = 156)TrazodoneTrazodoneOAD (n = 24)IR (n = 25)OAD (n = 28)IR (n = 29)Headache33.2%27%6.83%11.8%4%12%18%10%Somnolence31.2%15.7%8.7%–––––Dry mouth25.2%12.7%6.83%1.86%––11%3%Dizziness24.8%12.3%11.18%3.73%21%16%7%14%Electrocardiogram QT––5.59%3.73%––––Nausea20.8%12.7%6.21%14.29%17%8%7%7%Palpitations––2.48%5.59%––––Sedation16.8%3.4%––––––Fatigue14.9%8.3%––––––Diarrhea9.4%11.3%––––––Constipation7.9%2%––––7%3%Back pain5.4%3.4%––––––Vision blurred5.4%–––––––Nasal congestion––––4%4%14%3%Insomnia–––––––2%Orthostatic hypotension––––––4%–Abdominal pain (upper)––––––––Dyspnea–––––––7%Trazodone OAD was given at 150–375 mg/day (Sheehan et al., 2009a), 300–450 mg/day (Fagiolini et al., 2020), or 300 mg/day (Karhu et al., 2011). Trazodone immediate-release (IR) was given at 100 mg 3x/day (Karhu et al., 2011). Venlafaxine extended-release (XR) was given at 75–225 mg 1x/day (Fagiolini et al., 2020).
In the Sheehan et al. study (Sheehan et al., 2009a), one participant (0.5%) in the trazodone OAD-treated group reported experiencing anxiety during the study period vs five participants (2.5%) in the placebo-treated group (Sheehan et al., 2009a). The incidence of sexual dysfunction was 4.9% in the trazodone OAD-treated group and 2.5% in the placebo-treated group (Sheehan et al., 2009a). In the trazodone OAD-treated group, decreased libido was the most commonly reported sexual dysfunction (1.48%), followed by delayed ejaculation (1%) and erectile dysfunction (1%). No incidences of priapism were reported. No clinically significant changes in vital signs or body weight were observed in either treatment group during the study (Sheehan et al., 2009a).
Quality of sleep was also evaluated in the Sheehan et al. study (Sheehan et al., 2009a). At the end of the study, significant improvements in the quality of sleep were reported in the trazodone OAD-treated group (Sheehan et al., 2009a). Based on Likert scale responses, 60.2% of trazodone OAD-treated participants reported having “excellent” or “good” overall sleep quality vs 45.1% of placebo-treated participants (Sheehan et al., 2009a). In addition, 74.6% of trazodone OAD-treated participants reported “never” or “rarely” having trouble falling asleep vs 9.8% of placebo-treated participants, and 69.7% of trazodone OAD-treated participants reported “never” or “rarely” awakening during the night vs 54.4% of placebo-treated participants (Sheehan et al., 2009a). Significant improvements in overall quality of sleep and awakening during the night in the trazodone OAD-treated group occurred within the first 7 days of the titration phase and were maintained throughout most days of the treatment period (Sheehan et al., 2009a).
No electrocardiogram (ECG) abnormalities were reported in the Sheehan et al. study (Sheehan et al., 2009a). In the Fagiolini et al. study, statistically significant differences in ECG parameters were observed between the screening visit and the post-randomization visits (Fagiolini et al., 2020). At the final study visit, corrected QT (QTcF), QT, RR, and QRS significantly differed from screening in both the trazodone OAD- and venlafaxine XR-treated groups (Fagiolini et al., 2020). In the trazodone OAD-treated group, 5.59% of participants experienced ECG QT prolongation, which was defined as QTcF values exceeding 450 ms or showing prolongation higher than 60 ms at any visit, while 3.73% of participants in the venlafaxine XR-treated group experienced such ECG QT prolongation. However, the difference between the two treatment groups was not statistically significant (Fagiolini et al., 2020).
In both studies, the incidence of serious adverse events (SAEs) was low. In the Sheehan et al. study, 1.5% of trazodone OAD-treated participants experienced SAEs, while 1.8% of trazodone OAD-treated participants experienced SAEs in the Fagiolini et al. study (Fagiolini et al., 2020; Sheehan et al., 2009a). None of the SAEs in the Sheehan et al. study were judged to be related to the study medication (Sheehan et al., 2009a). In the Fagiolini et al. study, one patient exhibited mental impairment and dizziness, which were considered to be related to study treatment, but all other SAEs were considered to be unrelated to treatment (Fagiolini et al., 2020; Sheehan et al., 2009a).
A greater percentage of trazodone OAD-treated participants discontinued treatment due to AEs in both studies (Sheehan et al.: 12.4% in the trazodone OAD-treated group vs 2.9% in the placebo-treated group; Fagiolini et al.: 15.2% in the trazodone OAD-treated group vs 9.61% in the venlafaxine XR-treated group) (Fagiolini et al., 2020; Sheehan et al., 2009a). The most commonly cited reasons for discontinuation due to AEs were dizziness, sedation, somnolence, and ECG abnormalities/ECG QT prolongation (Fagiolini et al., 2020; Sheehan et al., 2009a).
Trazodone OAD was well tolerated in healthy participants in the short-term single-dose comparative and multiple-dose comparative studies by Karhu et al. Sample sizes in these studies ranged from 24 to 29. The most commonly reported AEs were headache, dizziness, and nausea (Karhu et al., 2011).
In both of the substudies, all AEs were rated as either mild or moderate in intensity. No SAEs were reported, and no clinically significant medication-related changes in vital signs, physical findings, laboratory values, or ECG evaluation were observed (Karhu et al., 2011).
As discussed, the favorable safety and tolerability profile of trazodone OAD was demonstrated in several RCTs. This section will examine the prevalence and types of AEs associated with trazodone OAD as compared with other second-generation antidepressants. We will focus the comparison on the SSRIs escitalopram, fluoxetine, paroxetine, and sertraline and the SNRIs venlafaxine and duloxetine, as well as other formulations of trazodone. We will also discuss the SMSs vortioxetine and vilazodone, which are distinct from the other classes and have some structural pharmacodynamic overlap with trazodone (Stahl, 2009a; Connolly and Thase, 2016; Deardorff and Grossberg, 2014; Basgiouraki et al., 2016; Rosenblat et al., 2020).
Sexual dysfunction has been reported to be the most bothersome AE among patients taking antidepressants (Ashton et al., 2005; Hu et al., 2004). Concerns with sexual function are frequently reported to be a major obstacle to antidepressant adherence and to the treatment of depression (Ashton et al., 2005; Clayton et al., 2014). Trazodone's low rates of treatment-emergent sexual dysfunction have been recognized for decades (Boyarsky and Hirschfeld, 2000), and even as far back as the 1980s it was reported that trazodone may increase sexual interest above premorbid levels in some female patients treated for depression (Gartrell, 1986). Although quite rare, cases of priapism have also been attributed to trazodone in both males and females (Battaglia and Venturoli, 2009; Bardin and Krieger, 1990; Jayaram and Rao, 2005; Goldberg and Ernst, 2019).
As described earlier, Sheehan et al. (2009a) recapitulated the finding that trazodone is associated with relatively low levels of sexual dysfunction, showing that for the OAD formulation it was present in just under 5% of patients (Sheehan et al., 2009a). This is considerably lower than the 33.9% of patients who, in a survey-based study, reported experiencing sexual dysfunction within the first 3 months of SSRI treatment (Hu et al., 2004). However, the overall prevalence of sexual dysfunction from SSRIs may be even higher, with one estimate putting it at 50%–70% (Carvalho et al., 2016). As shown in Table 2, antidepressant drugs that have rates of sexual dysfunction reported in the literature that exceed those reported for trazodone OAD in Sheehan et al. (2009a) by at least 2-fold include escitalopram, fluoxetine, paroxetine, sertraline, venlafaxine, duloxetine, and vortioxetine (Serretti and Chiesa, 2009; Montejo et al., 2001; Jacobsen et al., 2019; Nelson et al., 2006; Citrome, 2012). Rates of sexual dysfunction were similar for vilazodone vs trazodone OAD, occurring in under 5% of patients (Table 2) (Croft et al., 2014).Table 2Likelihood of adverse events (AEs) with antidepressants relative to trazodone once-a-day (OAD) (Albert et al., 2021; Fagiolini et al., 2020; Sheehan et al., 2009a).Table 2DrugSexual dysfunctionaAnxietyaSomnolencea^,^bTrazodone prolonged-releaseFrequency not foundFrequency not found= (Kasper et al., 2005; Zhang et al., 2014; Munizza et al., 2006)Trazodone immediate-releaseFrequency not found> (Beasley et al., 1991a)> (Beasley et al., 1991b)Escitalopram> (Serretti and Chiesa, 2009)>c (Bandelow et al., 2007; Urade et al., 2015; Burke et al., 2002)< (Baldwin et al., 2007; Jeong et al., 2015)Fluoxetine> (Montejo et al., 2001)>d (Beasley et al., 1991a; Davidson et al., 2002; Schatzberg and Roose, 2006; Rudolph and Feiger, 1999)= (Rudolph and Feiger, 1999; Beasley et al., 1991b)Paroxetine> (Montejo et al., 2001; Jacobsen et al., 2019; Nelson et al., 2006)> (Jacobsen et al., 2019)< (Kasper et al., 2005; Nelson et al., 2006) or = (Jacobsen et al., 2019)Sertraline> (Montejo et al., 2001)>d (Stahl, 2000; Nemeroff et al., 1995)= (Croft et al., 1999)Venlafaxine> (Montejo et al., 2001)Greater reductione (Albert et al., 2021)< (Baldwin et al., 2007; Rudolph and Feiger, 1999)Duloxetine> (Serretti and Chiesa, 2009; Nelson et al., 2006)>c (Detke et al., 2004)< (Nelson et al., 2006; Baldwin et al., 2016)Vortioxetine> (Jacobsen et al., 2019)>d (Jacobsen et al., 2019)< (Jacobsen et al., 2019; Baldwin et al., 2016)Vilazodone= (Citrome, 2012)Frequency not foundd (Mathews et al., 2015)< (Mathews et al., 2015)a>, greater likelihood, 2-fold increase of AE frequency compared to trazodone OAD; <, 2-fold decrease of average AE frequency compared to trazodone OAD; = , similar likelihood, average AE frequency is less than a 2-fold difference from trazodone OAD.bData from Fagiolini et al. (2020) and Sheehan et al. (2009a) were combined for analysis using weighted sample sizes.cSignificantly improved Hamilton Anxiety Scale (HAM-A) score or decreased frequency of anxiety compared to placebo at any one or more measured time points within individual studies.dNo difference in HAM-A score and/or frequency of anxiety compared to placebo in individual studies.eFrequency data not available, greater reduction in Hamilton Rating Scale for Depression anxiety/somatization score from baseline than trazodone OAD in a head-to-head trial.
Weight gain is another AE that is considered particularly bothersome among patients taking antidepressants (Ashton et al., 2005; Hu et al., 2004). Among SSRIs, paroxetine appears to be particularly robust in its tendency to promote weight gain over a period of at least 4 months (Serretti and Mandelli, 2010; Goldberg and Stahl, 2021). Similarly, mirtazapine, a noradrenergic and specific serotonergic antidepressant drug that, like trazodone, antagonizes 5-HT2 and alpha-2 adrenergic receptors and is associated with low levels of sexual dysfunction, has been reported to cause significant weight gain in patients (Serretti and Mandelli, 2010; Nutt, 2002). Fluoxetine, escitalopram, sertraline, duloxetine, venlafaxine, and vortioxetine may not be associated with significant long-term weight gain (Serretti and Mandelli, 2010; Robinson et al., 2011; Baldwin et al., 2016; Dent et al., 2012). Likewise, trazodone OAD was demonstrated to be weight neutral over 56 days in the placebo-controlled trial by Sheehan et al. (2009a), and neither the trazodone group nor the placebo group increased notably in weight (Sheehan et al., 2009a). Trazodone PR/AC was also found to be weight neutral across multiple trials; however, paroxetine also did not lead to marked changes in weight when used as an active comparator over 42 days (Kasper et al., 2005; Zhang et al., 2014).
As with the previously discussed AEs, anxiety and insomnia are additional AEs that are considered difficult to tolerate (Ashton et al., 2005; Hu et al., 2004). They stand out, however, in that they may also be among the AEs most likely to clinically impair occupational functioning (Lam et al., 2012). Another commonality is that both anxiety and insomnia are components of activating effects, sometimes also called jitteriness/anxiety syndrome, which may be precipitated by certain antidepressants (Sinclair et al., 2009). This syndrome may manifest as a constellation of these and other related symptoms such as irritability, restlessness, and increased energy, although the precise cluster of symptoms remains to be validated (Sinclair et al., 2009). Fluoxetine and paroxetine, in particular, have been associated with increased activating effects, which may develop in the first month of treatment and then wane in the second (Sinclair et al., 2009). One study found that 28% of patients who were rapidly titrated to 60–80 mg/day fluoxetine experienced activating effects (Beasley et al., 1991a).
In a prospective study where patients took SSRIs or SNRIs, the 7% of patients who experienced activating effects were primarily taking fluoxetine and paroxetine, which together were prescribed to 80.6% of study participants (Harada et al., 2014). Of these patients, 57.1% experienced one symptom, 28.6% experienced two symptoms, and 14.3% experienced three symptoms that would be classified as activating effects (Harada et al., 2014). Furthermore, among these patients, symptom prevalence was 42.9% for insomnia, 33.3% for irritability, 23.8% for anxiety and agitation, 14.3% for panic attacks, 9.5% for impulsivity, and 4.8% for akathisia and hypomania (Harada et al., 2014). There was a trend for the patients who showed activating effects to be younger, with a mean age of 38.2 vs 45.9 years old (P = 0.057) (Harada et al., 2014).
A study from 1991 specifically compared rates of activating effects in patients taking fluoxetine to those taking trazodone IR (Beasley et al., 1991a). Here, 95.3% of patients on fluoxetine took 20 mg/day, and the remaining took 40 mg/day. For patients on trazodone, 66.1% took 200–300 mg/day, 20.3% took 50–150 mg/day, and 13.6% took 350–400 mg/day (Beasley et al., 1991a). There were no statistically significant differences in rates of insomnia or anxiety between treatments; however, the incidence of having experienced at least one activating effect was elevated in the fluoxetine group vs the trazodone IR group, occurring in 15.4% and 3.3% of patients, respectively (Beasley et al., 1991a). This was measured by accounting for whether patients had experienced at least one event on one occasion involving agitation, anxiety, nervousness, or insomnia, in the absence of any sedation-related events. Activating effects were earlier and more transient in patients taking trazodone (Beasley et al., 1991a). As might be expected, the presence of a sedative event consisting of somnolence or asthenia was greater in the trazodone IR group, occurring in 42.5% of patients, vs the fluoxetine group, where it occurred in 21.5% of patients (Beasley et al., 1991a).
In an RCT where the PR/AC formulation of trazodone was compared with paroxetine, the latter significantly reduced the Day 21 mean HAM-D-17 anxiety/somatization score compared with trazodone PR/AC (Kasper et al., 2005). By completion of the study, trazodone PR/AC was associated with significantly greater improvements in the HAM-D-17 mean sleep scores than paroxetine (Kasper et al., 2005).
As discussed earlier, somnolence was among the most frequently reported AEs for trazodone OAD. In the Sheehan et al. study (Sheehan et al., 2009a), there were corresponding significant improvements in sleep quality compared with placebo (Sheehan et al., 2009a). Moreover, Fagiolini et al. (2020) reported greater improvements in the HAM-D-17 sleep disturbance score from baseline compared with venlafaxine XR at nearly all assessments in both the ITT and PP populations (Fagiolini et al., 2020). This was also demonstrated in a recent systematic review by Albert et al., which used data collected by both Sheehan et al. (2009a) and Fagiolini et al. (2020) to demonstrate that trazodone OAD improved the HAM-D-17 sleep disturbance score from baseline compared to placebo at all assessment points in both the ITT and PP populations (Albert et al., 2021). Importantly, effects on sleep disturbances were rapid, emerging by 7 days from the start of treatment (Albert et al., 2021). These findings for trazodone contrast with those of many other antidepressants that have been shown to promote insomnia. A 2015 meta-analysis demonstrated that SSRIs, including escitalopram, paroxetine, fluoxetine, and sertraline, as well as SNRIs such as duloxetine and venlafaxine, can produce elevated insomnia relative to placebo (Alberti et al., 2015). For vilazodone, insomnia has been reported to be among the AEs found in at least 5% of patients at twice the rate of placebo (Mathews et al., 2015). Vortioxetine, in contrast, has been reported to have low incidence of insomnia, ranging from 2.0% to 5.1% vs 4.0% for placebo (Baldwin et al., 2016).
Beyond its favorable profile in terms of insomnia, the literature suggests that trazodone OAD may be a suitable choice for patients who present clinically with anxious depression or who develop anxiety when taking other medications. Trazodone OAD has been associated with a numerically lower frequency of anxiety (1/202, 0.5%) than placebo (5/204, 2.5%), as reported in the Sheehan et al. study (Sheehan et al., 2009a). A recent analysis of data from that study demonstrated that the mean HAM-D-17 anxiety/somatization score was significantly lower at Day 21 in patients on trazodone vs placebo in both the ITT and PP populations (Albert et al., 2021). Importantly, anxiety/somatization was reduced from baseline at all time points, indicating that trazodone OAD was not increasing anxiety/somatization. This pattern was likewise observed in the data from the Fagiolini et al. study, as reported in the Albert et al. systematic review, although venlafaxine XR significantly improved these scores compared with trazodone at Days 35 and 56 in the ITT population and Day 56 in the PP population (Albert et al., 2021; Fagiolini et al., 2020). Altogether, trazodone demonstrates a favorable profile in terms of activating effects, especially concerning anxiety and insomnia. Table 2 summarizes the prevalence of these anxiety and somnolence AEs for the previously discussed second-generation antidepressants relative to trazodone OAD (Albert et al., 2021; Kasper et al., 2005; Sheehan et al., 2009a; Serretti and Chiesa, 2009; Montejo et al., 2001; Jacobsen et al., 2015, 2019; Nelson et al., 2006; Citrome, 2012, 2014; Serretti and Mandelli, 2010; Baldwin et al., 2007, 2016; Zhang et al., 2014; Beasley et al., 1991a, 1991b; Harada et al., 2014; Mathews et al., 2015; Bandelow et al., 2007; Davidson et al., 2002; Detke et al., 2004; Urade et al., 2015; Burke et al., 2002; Schatzberg and Roose, 2006; Stahl, 2000; Munizza et al., 2006; Rudolph and Feiger, 1999; Croft et al., 1999; Jeong et al., 2015; Nemeroff et al., 1995).
Some additional AEs that were observed in at least 5% of patients in the Sheehan et al. placebo-controlled trazodone OAD study include headache, dry mouth, and dizziness (Sheehan et al., 2009a). Nausea, headache, and dizziness were reported to be among the most frequent treatment-related AEs by Fagiolini et al. (2020). In both studies, the frequent AEs were mild to moderate, and these AEs were consistent with those reported for the PR/AC formulation (Fagiolini et al., 2020; Sheehan et al., 2009a; Zhang et al., 2014). Venlafaxine appears to have a particularly strong association with dizziness (Hansen et al., 2005), although the XR formulation had lower incidence than trazodone OAD (3.73% vs 11.18%) (Fagiolini et al., 2020). SSRIs may tend to be associated with comparable levels of nausea to trazodone and duloxetine (Kasper et al., 2005; Nelson et al., 2006; Beasley et al., 1991a; Munizza et al., 2006; Papakostas, 2007; Perahia et al., 2006) but lower levels of nausea than venlafaxine (Rudolph and Feiger, 1999; Papakostas, 2007; Montgomery et al., 2004). This is consistent with Fagiolini et al., where venlafaxine XR had a prevalence of 14.29% vs 6.21% for trazodone OAD (Fagiolini et al., 2020). In a comprehensive and retrospective US claims study that included a cohort of 40,017 patients, headaches were the most common AE among those examined in adults and adolescents on either SSRIs or SNRIs (Anderson et al., 2016). Headache was also the most common AE in the Sheehan et al. study; however, it was the highest for placebo as well (33.2% vs 27.0%) (Fagiolini et al., 2020). Indeed, among nocebo effects, patients appear to be particularly susceptible to headaches, and studies are likely to significantly overestimate actual headache prevalence in patients on active treatments (Goldberg and Ernst, 2019).
Some antidepressants are known to be associated with QT prolongation, especially the SSRI citalopram (Aronow and Shamliyan, 2020; Jasiak and Bostwick, 2014). Although relatively low in specificity, the presence of prolonged QT interval is considered a risk factor for sudden cardiac death from the prodysrhythmia torsades de pointes (Fabritz and Kirchhof, 2010). The US Food and Drug Administration considers a QTc interval increase of 20 ms to be clinically meaningful, and a QT interval prolongation in excess of 500 ms or QTc prolongation greater than 60 ms is generally considered to be indicative of elevated prodysrhythmic risk (Goldberg and Stahl, 2021; Fabritz and Kirchhof, 2010). In the Fagiolini et al. study, 11 patients taking trazodone OAD demonstrated QTc values above 450 ms or prolongation higher than 60 ms at any visit vs six patients taking venlafaxine XR (Fagiolini et al., 2020). There was no statistical difference between the groups on this outcome measure (Fagiolini et al., 2020). While a recent study concluded that trazodone is unlikely to increase risk for ventricular proarrhythmia, concomitant use of trazodone with drugs that are associated with prolonged QT or increased trazodone exposure should be avoided (Fagiolini et al., 2012; Aronow and Shamliyan, 2020; Tellone et al., 2020). Overall, the risk of QT prolongation from antidepressants appears to be low, especially in the absence of overdose, and the risk from trazodone is within the same order of magnitude as other such drugs (Aronow and Shamliyan, 2020; Jasiak and Bostwick, 2014; Tellone et al., 2020; Beach et al., 2018).
Another consideration concerning the safe use of antidepressants is discontinuation symptoms upon abrupt cessation. Discontinuation symptoms may be observed in as many as two-thirds of patients who suddenly cease use of SSRIs (Goldberg and Stahl, 2021). Such symptoms may be detrimental, not only in terms of impairing patients’ function and quality of life but also with respect to leading to misdiagnosed conditions such as MDD relapse or emergent disorders sometimes nebulously categorized as “medically unexplained symptoms” or “functional neurological disorder” (Guy et al., 2020). The types of potential discontinuation symptoms from SSRIs and SNRIs are quite broad. The most common symptoms include dizziness, nausea, headache, sleep disturbances, and irritability and emotional lability (Henssler et al., 2019). It is also possible to have flu-like symptoms, weakness and fatigue, pain, malaise, tachycardia, hypotension, sexual dysfunction, vomiting, diarrhea, hypomania, depersonalization, visual and auditory hallucinations, and nightmares, among numerous other symptoms (Henssler et al., 2019; Cosci and Chouinard, 2020). Discontinuation symptoms typically emerge within 1 week of discontinuation and then resolve within 2–6 weeks in a half-life-dependent manner (Henssler et al., 2019).
The level of prevalence and severity of antidepressant discontinuation symptoms is controversial, and there has been much debate among groups who have discussed the topic (Davies and Read, 2019; Jauhar and Hayes, 2019; Horowitz and Taylor, 2019; Selvaraj et al., 2019). This disagreement may stem from differences in symptoms among antidepressants as well as the types of studies used to assess discontinuation symptoms (Selvaraj et al., 2019). For example, placebo-controlled clinical trials may report fewer discontinuation symptoms, while survey-based studies may report greater severity due to a proneness to attributional bias (Selvaraj et al., 2019). Higher estimates put rates of withdrawal following SSRI discontinuation at 56% (Fava et al., 2007). On the lower end of the estimate, a study compared SSRIs with shorter half-lives (fluvoxamine or paroxetine) to those with longer half-lives (sertraline or fluoxetine) and observed a symptom prevalence of 17.2% and 1.5%, respectively (Coupland et al., 1996). Henssler et al. classified antidepressants according to their risk of discontinuation symptoms. Fluoxetine was classified as low-risk; escitalopram, sertraline, duloxetine, and vortioxetine as moderate-risk; and paroxetine and venlafaxine as high-risk (Henssler et al., 2019). Vilazodone does not appear to be associated with acute discontinuation symptoms, but there is very little literature about potential longer-term effects (Cosci and Chouinard, 2020; Reinhold et al., 2012). SSRIs and SNRIs with short half-lives may necessitate a gradual tapering period of many days to limit withdrawal symptoms, which can be extended if symptoms are troublesome (Goldberg and Ernst, 2019).
There do not appear to be very many cases or studies of trazodone withdrawal in the literature. To our knowledge, the most cited report was by Otani et al., which covered three case studies where 27- and 28-year-old male patients and a 63-year-old female patient experienced withdrawal symptoms despite protracted tapering from therapeutic doses (Otani et al., 1994). Symptoms experienced by the patients included insomnia, nausea and vomiting, moderate lassitude, anorexia, diarrhea or constipation, mild tremor, headache, myalgia with cold-like symptoms, and akathisia (Otani et al., 1994). Reintroduction or dose increases of trazodone tended to alleviate symptoms (Otani et al., 1994). The authors attributed the withdrawal symptoms to a noradrenergic rebound driven by serotonergic responses to reduced trazodone exposure (Otani et al., 1994). They concluded that trazodone should be tapered very slowly when discontinuing (Otani et al., 1994). Other reported withdrawal symptoms of psychiatric relevance include hypomania, nightmares, depersonalization, and formication (Cosci and Chouinard, 2020). These reports all predate trazodone OAD, and, to our knowledge, not much is known about withdrawal symptoms associated with this formulation (Cosci and Chouinard, 2020; Otani et al., 1994). The slower decline in plasma concentrations seen in trazodone OAD vs PR/AC and IR could reduce frequency of withdrawal symptoms and reduce the need to taper (Stahl, 2009a).
The data from RCTs have demonstrated that trazodone OAD has a favorable safety and tolerability profile, especially in terms of many of the AEs that patients consider especially troublesome, such as sexual dysfunction, weight gain, anxiety, and insomnia (Fagiolini et al., 2012; Sheehan et al., 2009a; Karhu et al., 2011; Hu et al., 2004; Clayton et al., 2014). Moreover, anxiety and insomnia, which can be activating effects, are known to cause significant impairment in occupational functioning (Sinclair et al., 2009; Lam et al., 2012).
The slow and sustained release kinetics of trazodone OAD are thought to underlie its safety and tolerability. The various formulations of trazodone may have different pharmacodynamic and psychopharmacological properties according to the proposed “different dose–different delivery–different drug” framework (Stahl, 2009a, 2013). The original IR formulation of trazodone is short acting, and peak plasma concentrations are reached rapidly, followed by a precipitous decline (Stahl, 2009a). Furthermore, this formulation may be dosed three times per day, which generates a sawtooth pattern (Stahl, 2009a). These characteristics are associated with hypnotic effects that are resistant to the development of tolerance (Stahl, 2009a). At the same time, saw-toothed patterns of antidepressant delivery are known to produce greater risk of AEs at the peaks and reduced efficacy at the troughs (Kilts, 2003). It is important to note that due to trazodone's relatively low affinity for SERT, the 50- to 100-mg doses used to achieve this hypnotic effect are generally too low to cross the antidepressant threshold, and doses of 150–300 mg are typically necessary to adequately block serotonin reuptake (Stahl, 2009a). This translates to even higher peaks at antidepressant doses (Stahl, 2009a; Nilsen et al., 1993). In contrast to the IR formulation, trazodone PR/AC produces two peaks in a 24-h period (Albert et al., 2021). When trazodone PR/AC is administered at 150 mg per dose, both of these peaks exceed the peak observed with trazodone OAD dosed at 300 mg (Albert et al., 2021). In contrast, trazodone OAD avoids the pulsatility and excessive peaks in drug exposure observed with other formulations, especially the IR version (Stahl, 2009a). This would be expected to cause fewer peak-dose AEs while facilitating development of tolerance to AEs over time, without compromising efficacy (Stahl, 2009a; Khouzam, 2017). Indeed, many of the AEs associated with trazodone OAD have been shown to be transient and resolve within days to weeks (Sheehan et al., 2009a, Cešková et al., 2018). The phenomenon of peak drug concentrations being associated with AEs is also recognized for other classes of drugs, such as analgesics, including tramadol, which notably blocks SERT (Ogawa et al., 2014; Gudin et al., 2019; Mongin, 2007).
Trazodone's pharmacodynamics further help to explain the pattern and rates of AEs associated with its medicinal use. Trazodone demonstrates medium to high binding affinity for numerous receptor subtypes (Albert et al., 2021). This multitarget drug has the highest binding affinity for 5-HT2A receptors on which it acts as an antagonist (Albert et al., 2021). In addition to its 5-HT2A binding, trazodone is recognized for its partial agonism at 5-HT1A receptors and antagonism at 5-HT2C, alpha-1A adrenergic, and H1 receptors, in addition to acting as a SERT inhibitor (Stahl, 2009a, 2013; Marek et al., 2003; Montalbano et al., 2019). However, trazodone can also antagonize alpha-1B and alpha-2C adrenergic receptors as well as serotonin 5-HT1D and 5-HT7 receptors (Albert et al., 2021). Recent evidence even suggests that trazodone's affinity for 5-HT7 is higher than previously thought and within the same order of magnitude as SERT inhibition (Albert et al., 2021). The binding affinities of trazodone for monoamine receptors relative to SERT are shown in Fig. 2 and derive from previously reported proprietary in vitro data (Albert et al., 2021; Mangano et al., 2020).Fig. 2Relative binding affinity of trazodone to monoamine receptors compared to serotonin transporter (SERT)Bars represent the log10 of the fold difference between trazodone's Kis for serotonin (5-HT), alpha adrenergic (alpha), and histamine H1 (H1) receptors vs the Ki for SERT derived from previously reported proprietary in vitro data (Albert et al., 2021; Mangano et al., 2020). The Ki of trazodone for each target is listed in nanomolar units. Trazodone has a greater than 10-fold higher affinity for 5-HT2A, alpha-1B, and 5-HT1D as compared with SERT. The other receptors are within an order of magnitude of SERT, with trazodone having a lower affinity for 5-HT2C and 5-HT7 than SERT.Fig. 2
When SSRIs and SNRIs block SERT, they promote an accumulation of extracellular serotonin throughout the brain that can then bind any available 5-HT receptors (Stahl, 2009a). While serotonergic agonism of 5-HT1A receptors is thought to exert antidepressant effects, the activation of 5-HT2A and 5-HT2C receptors is believed to be associated with AEs, including insomnia, sexual dysfunction, and anxiety (Fig. 3) (Stahl, 2009a). Blockade of 5-HT2A/C may help to diminish these AEs (Stahl, 2009a; Goldberg, 2000). Electrophysiological measurements conducted in rats have demonstrated that agonism of 5-HT2C can substantially decrease the firing of dopamine neurons in the ventral tegmental area (VTA) and that administration of trazodone can reverse this effect (Ghanbari et al., 2012). Likewise, agonism of 5-HT2A inhibits firing of locus coeruleus norepinephrine neurons, which can be ameliorated by trazodone administration, and when trazodone is already on board it can prevent this 5-HT2A-mediated inhibition of this firing (Ghanbari et al., 2012). These findings may have relevance for the low rate of anxiety and sexual dysfunction associated with trazodone. Withdrawal during abstinence from addictive substances, which represents an anxiogenic state, is marked by decreases in dopamine neuron firing in the VTA (Koob, 2013; Liu and Jin, 2004). Moreover, dopamine promotes appetitive and consummatory behaviors, including those involved in sexual behaviors (Clayton et al., 2014; Fiorino et al., 1997). Norepinephrine is also involved in sexual behavior and promotes sexual arousal and vasocongestion (Fiorino et al., 1997), so trazodone could potentially avoid emergence of related AEs by preventing serotonin-mediated inhibition of norepinephrine neurons. It is also possible that trazodone's partial agonism at 5-HT1A accounts for its low rates of sexual dysfunction as compared with SSRIs. Indeed, both fluoxetine (Li et al., 1996) and paroxetine (Li et al., 1997) have been demonstrated to reduce hypothalamic expression of G proteins of the Gi/o family through which 5-HT1A signals (Raymond et al., 1999). This is accompanied by a reduction in the level of plasma oxytocin released in response to 5-HT1A agonism (Li et al., 1996, 1997). Oxytocin plays established roles in male and female sexual response; therefore, attenuated peripheral oxytocin secretion may underlie sexual dysfunction associated with SSRIs (de Jong et al., 2007; Blaicher et al., 1999). It is possible that trazodone's partial agonism at 5-HT1A could have a sparing effect on sexual function by not modulating second messenger activity as much as full serotonin agonism, resulting in weaker negative feedback on G protein function.Fig. 3Model of hypothesized effects of the modulation of distinct serotonin receptors(A) Agonism of the serotonin (5-HT) receptor 5-HT1A is thought to be responsible for antidepressant activity, whereas agonism of 5-HT2A/2C/7 is believed to contribute to sleep disturbances, anxiety, and/or sexual dysfunction. These effects may be anticipated with general serotonin transporter (SERT) blockade. (B) Partial agonism of 5-HT1A may reduce sexual dysfunction. Antagonism of 5-HT2A/2C/7 is believed to reduce sleep disturbances, anxiety, and/or sexual dysfunction. These are pharmacodynamic features of trazodone that may help explain its safety and tolerability profile.Fig. 3
Blockade of 5-HT2A is a feature that trazodone shares with atypical antipsychotics and certain other antidepressants (Pedzich et al., 2022). It therefore seems almost counterintuitive that there is growing interest in pursuing the clinical use of psychedelics in patients with treatment-resistant depression and other psychiatric disorders given that these substances act as 5-HT2A agonists or partial agonists (Pedzich et al., 2022; Sellers et al., 2018; Dodd et al., 2022). Examples of such substances include psilocybin, N,N-dimethyltryptamine, lysergic acid diethylamide, and mescaline (Pedzich et al., 2022). Evidence suggests that agonizing 5-HT2A with psychedelics could potentially downregulate or desensitize the receptor, which has been speculated to mediate the antidepressant effects (Pedzich et al., 2022; Raval et al., 2021; Karaki et al., 2014; de la Fuente Revenga et al., 2022; McCorvy et al., 2016). Therefore, following the initial transient activation of the receptor, agonism of 5-HT2A by psychedelics could lead to a net reduction in 5-HT2A activity much like an antagonist. While this hypothesis is intriguing, there has been some contradictory evidence. In a preclinical study, mice that were preexposed to swim stress and later administered the serotoninergic psychedelic 2,5-dimethoxy-4-iodoamphetamine 1 day before a forced swim test exhibited a dose-dependent reduction in 5-HT2A protein expression in the medial prefrontal cortex, but without any accompanying alleviation of depressive-like behavior during testing (Pedzich et al., 2022). Furthermore, a recent case study involving a 45-year-old male with an extended history of treatment-resistant depression reported that psilocybin generated rapid, pronounced, and sustained antidepressant effects despite the patient having premedicated with trazodone against the direction of the investigators (Rosenblat et al., 2023). This anecdotal evidence supports the counterhypothesis that the antidepressant effects of psilocybin occur independently from 5-HT2A agonism (Rosenblat et al., 2023). Indeed, psychedelics have numerous other targets beyond 5-HT2A that could potentially contribute to antidepressant effects (Sellers et al., 2018; Dodd et al., 2022; McCorvy et al., 2016).
Insomnia occurs in about 90% of patients with MDD (Stahl, 2013). While it is a common cause of concern in patients taking antidepressants (Hu et al., 2004), insomnia has been shown to be improved in patients taking trazodone OAD vs placebo (Sheehan et al., 2009a, 2009b) or venlafaxine XR (Fagiolini et al., 2020). When present as an AE, insomnia is thought to be caused by serotonergic activity in the brainstem sleep centers, so it is possible that blockade of various 5-HT receptor subtypes, especially 5-HT2, contributes to trazodone's low rate of this AE (Stahl, 1998). Insomnia is also one of the most common residual symptoms in patients treated with SSRIs, and drugs with hypnotic properties may help alleviate this symptom when it persists (Stahl, 2013). Trazodone's antagonism of alpha-1A adrenergic and H1 receptors, as well as 5-HT2A, is believed to contribute to the somnolence that may counteract insomnia (Stahl, 2009a, 2013).
It is worth exploring how trazodone shares a few pharmacodynamic features with the SMSs vortioxetine and vilazodone, which could potentially contribute to some of the overlap in their AE profiles. This is likely in part because they all possess a piperazine group that facilitates binding to multiple types of serotonin receptors (Connolly and Thase, 2016). Like trazodone, vortioxetine is a SERT inhibitor as well as an antagonist at 5-HT1D and 5-HT7 receptors (Albert et al., 2021; Deardorff and Grossberg, 2014; Basgiouraki et al., 2016). It is also an agonist at 5-HT1A receptors (Deardorff and Grossberg, 2014; Basgiouraki et al., 2016). Preclinical research has demonstrated a role of the 5-HT7 receptor in regulation of sexual behavior (Siddiqui et al., 2007; Becnel et al., 2011), and it has been suggested that it may be a target for treating sexual dysfunction (Sutar et al., 2020). Therefore, it is possible that antagonism of this receptor contributes to the low frequency of sexual dysfunction attributed to these drugs, which share Category B designation, indicating lack of a significant impact on sexual function (Fig. 3) (Chokka and Hankey, 2018). In fact, drugs that agonize both 5-HT1A and 5-HT7 were shown to block sexual receptive behavior in female rats, which could be rescued through administration of a selective 5-HT7 antagonist (Siddiqui et al., 2007). Trazodone (Sheehan et al., 2009a, 2009b) and vortioxetine (Baldwin et al., 2016) are also both relatively unlikely to promote insomnia. 5-HT7 has been implicated in the regulation of sleep-wake cycles, which has provoked speculation that antagonization of 5-HT7 could help to normalize disrupted sleep patterns in patients with depression (Fig. 3) (Thomas et al., 2003; Lee and Choo, 2019).
Vilazodone is also a SERT inhibitor and a partial agonist at 5-HT1A receptors (Deardorff and Grossberg, 2014). Its relatively weak association with sexual dysfunction (Clayton et al., 2013; Jacobsen et al., 2020) supports the hypothesis that partial agonism of 5-HT1A receptors could be associated with lower rates of this class of AE. Indeed, this has been proposed for vilazodone (Montejo et al., 2018). While antipsychotic medications are often associated with sexual dysfunction, aripiprazole is an antipsychotic that acts as a partial agonist at 5-HT1A receptors (Stahl, 2013), for which there are case reports of hypersexuality in males (Priya and Moorthy, 2021) and females (Cheon et al., 2013) as well as spontaneous ejaculation (Eğilmez et al., 2016). Interestingly, as its name suggests, aripiprazole is a piperazine-based drug like trazodone, vortioxetine, and vilazodone (Kumar et al., 2021).
AEs represent a principle trade-off when second-generation antidepressants are used to treat MDD (Gartlehner et al., 2012). Over the course of a clinical study, as many as 90% of patients treated with this group of medications may experience at least one AE, although the majority are mild and tolerable (Gartlehner et al., 2005). Additionally, the discontinuation rate due to intolerable AEs in these trials tends to be about 15% (Gartlehner et al., 2005). In one prospective analysis, 46.9% of patients were found to be nonadherent to their antidepressant medication over a 2-year follow-up period (ten Doesschate et al., 2009). In a separate survey-based study where 22% of patients reported nonadherence, the most common reported reason for not taking medication was difficulty remembering to take it (43%) (Ashton et al., 2005). Significant weight gain was the primary motivation for nonadherence in 27% of patients, followed by anorgasmia in 20% of patients and loss of interest in sex in another 20% of patients (Ashton et al., 2005). These findings are significant because remaining adherent has been reported to be associated with a 70% lower likelihood of relapse of depression (Geddes et al., 2003). It is also notable that, anecdotally, in two cases where increased sex drive was attributed to trazodone, the female patients expressed concerns about eventual discontinuation of the medication (Gartrell, 1986). This highlights how relief of a condition can promote adherence, while emergence or exacerbation can diminish it.
Patients may be more adherent if the selection of a medication accounts for their personal history or level of concern with specific AEs (Gartlehner et al., 2012). Therefore, past treatment experiences may be informative (Gartlehner et al., 2012). An additional factor in medication choice may be the baseline presentation of clinical features of depression that can also be treatment-emergent AEs. Prescribers often consider these features when making a decision about choice of antidepressants (Zimmerman et al., 2004). High baseline anxiety has been reported to be the most frequent consideration among prescribers, followed by insomnia, fatigue, anger or irritability, and increased appetite (Zimmerman et al., 2004).
AEs may pose yet another challenge in that they can potentially confound measurements of medication efficacy, especially earlier on in the treatment before any tolerance is established (Hieronymus et al., 2021). Conversely, symptoms of depression, or even non-iatrogenic worsening of depression, may be mistaken for AEs (Hieronymus et al., 2021; Goldberg and Ernst, 2016). A post hoc analysis of covariance performed on data from the Sheehan et al. study suggested baseline insomnia did not significantly impact efficacy of trazodone OAD (Sheehan et al., 2009b). It is also interesting to point out that a recent study challenges the oft-cited view that AEs in clinical trials may improve treatment efficacy by unblinding patients and raters to treatment (Hieronymus et al., 2021).
A primary limitation of this systematic review is that safety and tolerability of trazodone OAD was assessed in a limited number of small-size RCTs. In this case, we found three articles that met our eligibility criteria for inclusion in the analysis. Additional large, randomized, controlled, double-blind studies would provide more insight into trazodone's safety and tolerability profile. Additionally, long-term head-to-head studies between trazodone OAD and other trazodone formulations would help to clarify more quantitatively what advantages trazodone OAD may derive from its slower and more stable release kinetics.
Although there is a limited number of RCTs performed using trazodone OAD, numerous other double-blinded RCTs have been conducted using the original IR formulation of trazodone, and these suggest that the molecule is an efficacious antidepressant with a favorable safety and tolerability profile. In head-to-head comparisons, trazodone IR demonstrated reductions in the 21-item Hamilton Rating Scale for Depression (HAM-D-21) total score that were overall similar to those observed with fluoxetine (Beasley et al., 1991b) and the norepinephrine-dopamine reuptake inhibitor bupropion (Cunningham et al., 1994). Notably, at Day 7 this effect was greater for bupropion (Weisler et al., 1994), which is consistent with the rapid efficacy seen in trazodone OAD trials (Albert et al., 2021; Fagiolini et al., 2020; Sheehan et al., 2009a). In a study assessing trazodone IR and venlafaxine against placebo, both active treatments significantly reduced the HAM-D-21 total score vs placebo at Week 4 and the final evaluation in which patients either were still taking their medication or were within 3 days of their most recent dose (Cunningham et al., 1994). Here, trazodone IR was associated with an outcome that is consistent with trazodone OAD RCTs (Albert et al., 2021). Venlafaxine was more effective on cognitive disturbance and retardation factors, which for the latter measure is consistent with data from Fagiolini et al. (Fagiolini et al., 2020; Cunningham et al., 1994). In terms of safety and tolerability, the most common AEs observed in patients who were treated with trazodone IR were similar to what was observed in the OAD trials and included headache, somnolence, dry mouth, dizziness, and nausea (Fagiolini et al., 2020; Sheehan et al., 2009a; Beasley et al., 1991b; Weisler et al., 1994; Cunningham et al., 1994). Anxiety was significantly lower in patients taking trazodone IR vs those on bupropion, at 1.67% vs 14.52% (Weisler et al., 1994), and agitation was significantly lower with trazodone IR than venlafaxine at 0% vs 10% (Cunningham et al., 1994). Somnolence rates reported for trazodone IR in these studies were generally markedly higher than what was reported in trazodone OAD RCTs and ranged from 45% to 61% (Beasley et al., 1991b; Weisler et al., 1994; Cunningham et al., 1994) vs 8.7% (Fagiolini et al., 2020) and 31.2% (Sheehan et al., 2009a) for trazodone OAD. Weight gain was significantly higher for trazodone IR than venlafaxine and placebo; however, nausea rates were significantly higher in patients treated with venlafaxine than patients treated with trazodone at 44% vs 19%, and patients on venlafaxine lost weight (Cunningham et al., 1994). Patients on trazodone IR gained an average of 1–2 pounds, whereas venlafaxine-treated patients lost 2 to 3 pounds (Cunningham et al., 1994). In a 1988 review of double-blinded trazodone IR trials, the authors conclude that, overall, 25%–30% had some difficulty with using trazodone, while the remaining majority tolerated it well (Feighner and Boyer, 1988). These studies did not examine sexual AE rates (Beasley et al., 1991b; Weisler et al., 1994; Cunningham et al., 1994), but, as previously discussed, trazodone is recognized to be associated with low rates of sexual dysfunction (Boyarsky and Hirschfeld, 2000). Ultimately, data from RCTs on trazodone IR support the view that trazodone is a safe and effective antidepressant molecule.
Taken together, the present systematic review of the safety and tolerability of trazodone OAD in RCTs followed by a narrative approach to comparing this drug with other second-generation antidepressants demonstrates that trazodone OAD has a favorable safety and tolerability profile. This is especially the case in terms of certain AEs that may interfere with or be considered particularly bothersome to daily life, such as sexual dysfunction, weight gain, or anxiety and insomnia. Trazodone OAD may be an appropriate choice of treatment for MDD in patients who have difficulty tolerating these or related AEs.
This work was funded by 10.13039/501100006546Angelini Pharma S.p.A.
The authors declare the following financial interests/personal relationships which may be considered as potential competing
Stephen M. Stahl, MD, PhD, DSc (Hon), is a Clinical Professor of Psychiatry and Neuroscience at the University of California, Riverside; an Adjunct Professor of Psychiatry at the University of California, San Diego; Honorary Visiting Senior Fellow at the University of Cambridge, UK; Director of Psychopharmacology for the California Department of State Hospitals; and Editor-in-Chief for CNS Spectrums. Over the past 36 months, Dr. Stahl has served as a consultant to Acadia, Alkermes, Allergan, AbbVie, AstraZeneca, Avanir, Axsome, Biogen, Biomarin, Biopharma, ClearView, Done, EMD Serono, Eisai Pharmaceuticals, Eurolink, Genomind, Impel NeuroPharma, Innovative Science Solutions, Intra-Cellular Therapies, Ironshore Pharmaceuticals, Janssen, Jazz, Karuna Therapeutics, Lilly, Lundbeck, Neurocrine Biosciences, Noveida, Otsuka, Pfizer, Pierre Fabre, Proxymm, Relmada Therapeutics, Sage Therapeutics, Servier, Sprout, Sunovion, Takeda, Taliaz, Teva, TMS NeuroHealth, Tonix, Tris Pharma, Trius, Vanda, Vertex, and Vifor Pharma. He holds options in Delix, Genomind, Lipidio, and NeuraWell; he has served on speakers bureaus for 10.13039/100009427Acadia, 10.13039/100004328Genentech, Janssen, 10.13039/501100013327Lundbeck, Neurocrine, Otsuka, 10.13039/501100011725Servier, 10.13039/100009655Sunovion, 10.13039/100007723Takeda, and Teva; and he has received research and/or grant support from 10.13039/100009427Acadia, Alkermes, 10.13039/100007819Allergan/10.13039/100006483AbbVie, Arbor Pharmaceuticals, AssureX, 10.13039/100004325AstraZeneca, Avanir, Axovant, 10.13039/100005614Biogen, 10.13039/100001003Boehringer Ingelheim, Braeburn Pharmaceuticals, 10.13039/100002491Bristol Myers Squibb, 10.13039/100006436Celgene, 10.13039/100006437CeNeRx, 10.13039/100004388Cephalon, Daiichi Sankyo Brazil, Dey, 10.13039/501100003769Eisai, Eli Lilly, Forest, Genomind, 10.13039/100004330GlaxoSmithKline, Harmony Biosciences, 10.13039/100011404Indivior, Intra-Cellular Therapies, Ironshore, ISSWSH, Janssen, JayMac, Jazz, 10.13039/501100013327Lundbeck, 10.13039/100004334Merck, Neurocrine, Neuronetics, 10.13039/100004336Novartis, Otsuka, Pear Therapeutics, 10.13039/100004319Pfizer, Reviva, 10.13039/100004337Roche, Sage, 10.13039/501100011725Servier, Shire, Sprout, 10.13039/100009655Sunovion, Supernus, 10.13039/100007723Takeda, Teva, TMS NeuroHealth Centers, Tonix, Torrent, and 10.13039/100010902Vanda.
Joseph F. Goldberg, MD, is a Clinical Professor of Psychiatry at the Icahn School of Medicine at Mount Sinai. During or since 2022/2023, he has served as a consultant to BioXcel, Neumora, Neurelis, Otsuka, Sunovion, and Supernus. He has served on speakers bureaus for AbbVie, Alkermes, Axsome, and Intra-Cellular Therapies; and he has received royalties from American Psychiatric Publishing and Cambridge University Press.
Michael Q. Steinman, PhD, has nothing to disclose.