Marcel D. Waldinger, MD, PhD, Department of Psychiatry and Neurosexology, HagaHospital Leyenburg, Leyweg 275, 2545 CH The Hague, the Netherlands. Tel: (+31) 70 3592086; Fax: (+31) 70 3594902; E-mail: email@example.com
Introduction. Selective serotonin reuptake inhibitors (SSRIs) differ in the severity of induced ejaculation delay. Various studies indicate that oxytocin is involved in sexual behavior.
Aim. To review and evaluate the involvement of oxytocin in SSRI-induced ejaculation delay.
Main Outcome Measures. Oxytocine release, 5-hydroxytryptamine (5-HT) neurotransmission, and desensitization of 5-HT1A receptors.
Methods. A review and critical analysis of animal studies investigating the interaction of serotonergic and oxytocinergic neurotransmission in relation to the ejaculation process.
Results. Although acute treatment with the SSRIs fluoxetine and paroxetine immediately causes increased serotonin levels, delayed ejaculation does not occur. The increased serotonin levels induce oxytocin release via activation of 5-HT1A receptors, and this might compensate for the inhibitory actions of serotonin on sexual behavior. Chronic treatment with fluoxetine and paroxetine desensitizes 5-HT1A receptors on oxytocin neurons, and that might in part determine the onset of delayed ejaculation. Desensitization of 5-HT1A receptors is less strong following chronic treatment with the SSRIs fluvoxamine or citalopram, which may attenuate the degree of delayed ejaculation.
Conclusions. Preliminary data suggest that the severity of chronic SSRI treatment-induced delayed ejaculation and the differences between the various SSRIs in inducing ejaculation delay is related to gradual desensitization of 5-HT1A receptors on oxytocin neurons. de Jong TR, Veening JG, Olivier B, and Waldinger MD. Oxytocin involvement in SSRI-induced delayed ejaculation: A review of animal studies. J Sex Med 2007;4:14–28.
The introduction of the selective serotonin reuptake inhibitors (SSRIs) meant a revolutionary change in the understanding and treatment of premature ejaculation (PE). In the past decade, an increasing number of placebo-controlled drug treatment trials have demonstrated the clinically relevant ejaculation-delaying effects of daily treatment with the SSRIs paroxetine, sertraline, fluoxetine, and the tricyclic antidepressant clomipramine . Interestingly, despite the similar mechanism of action of the various SSRIs, differences in the extent of ejaculation delay have been found both in clinical studies in men with PE and in animal sexual behavioral experiments [2–4]. More knowledge of the neurobiological background of SSRI-induced delayed ejaculation might shed light on differences in mechanism or mode of action of serotonergic antidepressants, and is useful in the search for new drugs to treat ejaculatory dysfunctions [5,6]. The present review explores one theory concerning these questions, namely the possibility that the neuropeptide oxytocin plays an important role in both SSRI-induced delayed ejaculation and functional differences between SSRIs.
Effects of SSRIs on Depression and Ejaculation
When serotonin (5-hydroxytryptamine; 5-HT) is released from axon terminals, it binds to nearby serotonin receptors and induces a wide variety of effects. Under normal circumstances, serotonin is then quickly removed from the extracellular space back into the serotonergic cell via 5-HT transporters (5-HTTs). However, SSRIs block these 5-HTTs and thus prolong the elevation of extracellular 5-HT levels in response to activation of serotonergic neurons . As a consequence, acute systemic injection of the SSRIs fluoxetine, paroxetine, fluvoxamine, citalopram, or sertraline leads in 1 hour to a 2- to 4-fold increase in 5-HT levels in several brain areas, as measured with microdialysis in rats [8–14].
The effects of SSRIs on mood are generally attributed to this ability to increase extracellular 5-HT levels. For example, 5-HT is thought to mediate the fluoxetine-induced decrease in immobility time in the forced swimming test, a well-known animal paradigm for the detection of antidepressant effects of drugs . Abnormalities in serotonergic neurotransmission are thought to play a role in the pathophysiology of depression [16,17].
According to these findings, it would be expected that acute treatment with SSRIs, which immediately increases 5-HT levels, is able to improve the mood of depressed patients. However, one of the major disadvantages of SSRI treatment is the latency time of about 4–6 weeks until the drugs affect the depressive mood [18,19]. This delayed onset of antidepressant effects during SSRI treatment has been attributed to the stimulation of negative feedback systems that control serotonergic neurotransmission. For example, 5-HT activates autoreceptors located on serotonergic neurons, among which somatodendritic 5-HT1A autoreceptors are the most well known, that inhibit 5-HT release and attenuate the acute effects of SSRIs. However, during continuous (daily) administration of SSRIs, the increased activation by 5-HT of 5-HT1A autoreceptors induces desensitization of these receptors, which is probably one of the starting signals for the antidepressant effects to emerge [20–26].
Since the early 1980s, it is known that increased 5-HT levels inhibit ejaculation in laboratory animals. Central injection of 5-HT and systemic injection of the 5-HT precursor 5-hydroxytryptophan (5-HTP) cause delayed ejaculation in rats [27–33]. Increased activation of postsynaptic 5-HT1B and/or 5-HT2C probably mediates this effect [29,33–37].
Based on microdialysis studies of 5-HT neurotransmission, it would be expected that the immediately increased 5-HT levels upon acute SSRI administration would lead to inhibition of ejaculation. Indeed, some studies did report acute inhibition of ejaculation by paroxetine and fluoxetine in rats [3,38]. However, acute systemic injection of paroxetine and fluoxetine more often failed to delay ejaculation in rats [39–41], in contrast to chronic treatment [3,41,42]. Similarly, acute (on-demand) paroxetine treatment in humans has been shown to have hardly any ejaculation-delaying effect , whereas chronic treatment with paroxetine or fluoxetine causes a strong delay [2,44]. It has to be noted that, in a phase III clinical trial, the newly developed SSRI dapoxetine at a dose of 60 mg has been found to cause a significant, but mild, increase in intravaginal ejaculation latency time (3.6-fold vs. 1.9-fold with placebo) . It is not yet known why dapoxetine behaves differently from other SSRIs in this respect, and this question deserves full attention in further research.
The delayed onset in clinically effective ejaculation delay resembles the effects of SSRIs on mood, and suggests that desensitization of 5-HT1A autoreceptors might underlie SSRI-induced delayed ejaculation as well. Indeed, acute coadministration of the 5-HT1A receptor antagonist WAY-100635 with an SSRI, which strongly potentiates the SSRI-induced elevation of 5-HT levels [14,46], causes a more rapid onset of both the antidepressant effect [47,48] and the inhibition of ejaculation [49,50]. Reversely, activation of 5-HT1A receptors with selective agonists strongly facilitates ejaculation [51–53], which can be reversed by selective 5-HT1A receptor antagonists [29,34], supporting the idea that desensitization of these receptors might affect ejaculatory behavior.
Interestingly, a difference seems to exist between individual SSRIs in their ability to desensitize 5-HT1A autoreceptors and to induce delayed ejaculation. Paroxetine and fluoxetine seem to desensitize 5-HT1A autoreceptors [21, 24,54–57] more strongly than citalopram and fluvoxamine [10,58,59]. Likewise, paroxetine and fluoxetine are much more prone to delay ejaculation than citalopram and fluvoxamine in humans and rats, despite their comparable ability to elevate 5-HT levels [1–3,41,44,49]. In addition, chronic pretreatment with paroxetine strongly attenuated the facilitatory effect of 8-hydroxy-2-(di-n-propylamino)-tetralin (8-OH-DPAT) on ejaculation, in contrast to chronic treatment with fluvoxamine . These interesting findings raise two important questions. First, if chronic treatment with citalopram and fluvoxamine fails to desensitize 5-HT1A autoreceptors as strongly as paroxetine and fluoxetine, then how do they exert their antidepressant effects? Second, if increased serotonergic neurotransmission is responsible for SSRI-induced delayed ejaculation, then why does acute administration of SSRIs in rats not affect sexual behavior 1 hour after injection, when 5-HT levels are significantly elevated? The answer to both questions might involve another group of 5-HT1A receptors, located postsynaptically on a variety of non-serotonergic neurons in several brain areas, during SSRI treatment. One group of postsynaptic 5-HT1A receptors, located on oxytocinergic neurons in the hypothalamus, is interesting in this respect.
Interaction of Serotonergic and Oxytocinergic Neurotransmission
Two oxytocinergic neuronal systems in the central nervous system can be distinguished: the magnocellular and the parvocellular neurons. Magnocellular oxytocin-containing cells are located in the hypothalamic supraoptic nucleus (SON) and in the anterior and posterior paraventricular hypothalamic nucleus (PVH). These magnocellular neurons release oxytocin into the bloodstream via axons terminating in the posterior pituitary. In addition, magnocellular neurons are able to release large amounts of oxytocin within the PVN and SON via their dendrites [60,61].
In addition, the dorsal and lateral PVH contain smaller, parvocellular neurons that project to several brain and spinal cord areas [61–65].
5-HT is one out of a range of neurotransmitters that modulate oxytocin release from oxytocinergic neurons. All oxytocinergic cell groups receive some serotonergic innervation, mainly originating from the dorsal and median raphe nuclei in the brainstem [66–68]. In rats it has been demonstrated that intracerebroventricular injection of 5-HT stimulates oxytocin secretion , and systemic injection of the 5-HT releasers fenfluramine and p-chloroamphetamine (p-CA) significantly increases both plasma oxytocin levels [70,71] and Fos-expression in oxytocinergic neurons [72,73]. Several subtypes of 5-HT receptors have been found to mediate the 5-HT-induced release of oxytocin into the blood plasma [69,70,74–76]. Of those, postsynaptic 5-HT1A receptors located on the cell body of oxytocinergic neurons have been studied most extensively . Systemic injections of selective 5-HT1A receptor agonists, such as 8-OH-DPAT, cause a strong increase in plasma oxytocin levels, which can be blocked by 5-HT1A receptor antagonists injected systemically or locally into the PVH [78–81]. In addition, 8-OH-DPAT strongly induces Fos-expression in oxytocinergic neurons in the PVH .
SSRIs and Oxytocin Release
Interestingly, chronic systemic treatment with paroxetine and fluoxetine do not change basal oxytocin levels, but attenuate the oxytocin-releasing response to 5-HT1A receptor agonists . This attenuation starts after 3 days of treatment and lasts for at least 60 days after termination of the treatment [83–87]. Conversely, chronic treatment with citalopram increased basal plasma oxytocin levels and did not attenuate the oxytocin-releasing effect of a challenge dose of the SSRI zimelidine . As both citalopram and fluvoxamine exert less ejaculation-delaying effects than paroxetine and fluoxetine, it is questioned whether the increased oxytocin levels are associated with the rather mild ejaculation-delaying effects of citalopram. In this regard, it is unfortunate that there have been no reports on the effects of acute or chronic treatment with fluvoxamine on 5-HT-mediated oxytocin release.
Similarly, it would be interesting as well to know whether the increased plasma levels of oxytocin during citalopram treatment are related to its antidepressant effect. Indeed, some reports suggest that it may be possible that citalopram, and perhaps fluvoxamine, exert part of their antidepressant effects through an increased release of oxytocin [88,89]. This is supported by studies showing that acute or chronic systemic administration of oxytocin, like SSRIs, reduce several anxiety- and stress-related parameters [90–97].
If chronic treatment with citalopram or fluvoxamine increases oxytocinergic neurotransmission whereas chronic treatment with fluoxetine or paroxetine inhibits it, this could underlie the differences between these SSRIs on ejaculation latency. Indeed, systemic injection of oxytocin reversed the fluoxetine-induced inhibition of ejaculation (Figure 1) . Moreover, chronic pretreatment with paroxetine attenuated not only the facilitatory effects of 8-OH-DPAT on ejaculation, but also the 8-OH-DPAT-induced increased Fos-expression in oxytocinergic neurons in the PVH in the same rats (Figure 2) . It would be interesting to find out whether oxytocin release prevents delayed ejaculation in response to an acute injection with any SSRI, and in response to chronic treatment with citalopram or fluvoxamine. To further explore this hypothesis, it needs to be established how oxytocin facilitates ejaculation, which at this moment is not yet fully understood.
Oxytocin and Male Sexual Behavior
Oxytocin is known to facilitate reproduction in mammals . The role of oxytocin in female reproductive behavior is well known: oxytocin is released during sexual behavior, parturition, and suckling, and is involved in uterine smooth muscle contraction and milk ejection [100–104]. In addition, many studies suggest that oxytocin is important for male sexual behavior as well.
In men, plasma oxytocin levels are found to be elevated during sexual arousal, erection, and at the time of orgasm, although the degree of the elevation varies between different studies [105–108]. After ejaculation, oxytocin levels are raised in the blood plasma in rabbits  and in the cerebrospinal fluid in rats . Plasma oxytocin levels are elevated in sexually naïve, but not in experienced male rats following sexual behavior, and correlate highly with the number of intromissions . In sexually experienced male rats, exposure to odors of an estrous female without sexual contact is enough to increase Fos-expression in oxytocinergic neurons in the PVH, which is thought to reflect oxytocin release from these neurons . Furthermore, Fos-expression is increased in both parvocellular and magnocellular oxytocinergic neurons of the PVH in response to sexual behavior in rats [112–114]. Copulation also increases the number of Fos-positive oxytocinergic neurons in the supraoptic nucleus , and this increase is significantly stronger in “rapid-ejaculating” rats compared with “normal-ejaculating” rats . Consistently, electrical stimulation of the dorsal penile nerve, which relays sensory information from the genitals to the spinal cord, as well as tactile stimulation of the glans penis by a paintbrush, produces excitation in about half of the oxytocin cells in the PVH and SON of rats [116,117].
Despite these indications that different components of male sexual behavior activate the oxytocin system, depletion of oxytocin does not seem to block copulation or ejaculation but to cause rather mild effects on male sexual behavior. Selective lesions of the parvocellular neurons in the PVH, which reduced the number of oxytocin-immunoreactive fibers in the lumbosacral spinal cord and abolished the ejaculation-induced increase of oxytocin levels in the cerebrospinal fluid, cause a longer latency to, and a lower frequency of, noncontact erections and decrease seminal emission while leaving copulatory parameters intact [118,119], or prolong mount and intromission latencies and reduce the postejaculatory interval . Lesions of both the magnocellular and parvocellular neurons in the PVH decrease the number of noncontact erections and increases their latency, as well as the mount frequency and latency to ejaculation during copulation . On the other hand, intracerebroventricular administration of an oxytocin antagonist increases the intromission latency, decreases the number of mounts and intromissions, and abolishes ejaculation in rats, which is indicative of both erectile and ejaculatory dysfunctions [120,121]. In addition, impotence is associated with a reduced expression of oxytocin mRNA in the magnocellular PVH .
Taken together, oxytocin appears to play at least a modulating role in erection and ejaculation. Both peripheral and central oxytocin release seem to be involved.
Role of Peripheral Oxytocin in Erection and Ejaculation
Because oxytocin is a peptide and cannot readily cross the blood—brain barrier, the many effects of systemically injected oxytocin on ejaculation and erection suggest that peripheral release of oxytocin is important for male sexual behavior.
Intravenous injection of oxytocin increased the concentration of sperm in the ejaculate of rams  and oligozoospermic men , although other studies did not find this effect of oxytocin administered intravenously  or intranasally . Systemic administration of oxytocin reduced the number of intromissions required for ejaculation in young adult rats , whereas it shortened the mount, intromission, and ejaculation latencies and postejaculation intervals in old rats, especially in sexually sluggish individuals . In bulls, systemic oxytocin treatment decreased the time to semen emission and tended to decrease the number of electro-ejaculation stimuli to semen emission . These effects could have been exerted via oxytocin receptors that mediate the contractility of smooth muscle cells and have been found in the testis, epididymis, ductus deferens, prostate and penis of rats, rams and humans might be involved [123,124,129,130].
Interestingly, some evidence indicates that peripherally injected oxytocin might have an inhibiting rather than stimulating effect on erection. Systemic oxytocin treatment increased the time to penile protrusion in bulls  and inhibited the increase in intracavernous pressure elicited by electrical stimulation of the cavernous nerve in rats, which could be prevented by an oxytocin antagonist . The oxytocin receptor antagonist alone facilitated the nerve stimulation-induced increase in intracavernous pressure . Apparently, peripheral oxytocin receptors in the corpus cavernosum are involved in penile detumescence.
It has to be noted that at least some of the effects of systemically injected oxytocin might have been mediated by central oxytocin receptors, because a small proportion of subcutaneously injected oxytocin does cross the blood—brain barrier . This idea is supported by the finding that the ejaculatory motor pattern, which is exerted by motor neurons in the spinal cord innervating the bulbospongiosus muscle, could be induced by systemic administration of oxytocin in rats .
Role of Central Oxytocin in Erection and Ejaculation
The release of oxytocin from centrally projecting parvocellular neurons is well known to influence erection and, less clearly, ejaculation.
The pro-erectile effects of oxytocin are well defined within the central nervous system of the rat [133,134]. Increased oxytocinergic neurotransmission in the PVH or hippocampus induces either an increase in the number of penile erections or an increase in intracavernous pressure, which is an indication of erection [135,136]. In addition, intrathecal injection of oxytocin at the lumbosacral, but not thoracolumbar, level increased intracavernous pressure, which was impaired by pretreatment with an oxytocin receptor antagonist . These results indicate that the pro-erectile effects of oxytocin are mediated by neurons in the lumbosacral spinal cord. Indeed, oxytocinergic neurons in the dorsal parvocellular part of the PVH innervate the spinal cord , and neurons in the same area have been found to project to the erectile tissue in the penis, to the striated ischiocavernosus and bulbospongiosus muscles in the pelvic floor, and to the epididymis and prostate gland that are involved in erection and/or ejaculation [138–141]. Varicosities containing oxytocin or neurophysin (the coproduct of oxytocin and vasopressin) have been found in close apposition to neurons in the sacral parasympathetic nucleus, ventral horn, and dorsal gray commissure that directly or indirectly control the pelvic nerve, the corpus cavernosum, and the bulbospongiosus muscle [142–144]. However, in another study, the oxytocinergic neurons in PVH remained unlabeled following retrograde transneuronal tracing with rabies virus from the bulbospongiosus muscle, suggesting that oxytocinergic spinal projections are more likely to influence sacral parasympathetic than somatic outflow .
The second component of central oxytocin release, from magnocellular dendrites, has been proposed to play a role in male sexual behavior . Dendritic oxytocin release within the PVH might facilitate erection via the activation of parvocellular oxytocinergic neurons . Furthermore, dendritic oxytocin release is required for the typical burst-firing pattern and pulsatile secretion of oxytocin neurons at parturition and milk ejection that might be analogous to the pulsatile oxytocin secretion that occurs at the time of ejaculation [146,147].
Serotonin, Oxytocin, and Lifelong Premature Ejaculation
Besides a role in normal male sexual behavior, Waldinger et al. postulated that both serotonin and oxytocin might be involved in the etiology of lifelong premature ejaculation (ejaculatio praecox) and that lifelong premature ejaculation is a genetically determined sexual dysfunction [148–151]. It was further postulated that this genetic vulnerability will lead to a biological variability of the intravaginal ejaculation latency time (IELT) in men [148–151]. Indeed, a biological continuum of the IELT in men was found in 2005 with the outcome of a stopwatch study in a large random sample of 491 men in the general male population of five countries (the Netherlands, Spain, Turkey, the United Kingdom, and the United States) . It demonstrated a positively skewed IELT distribution, with a median IELT of 5.4 minutes (range 0.55–44.1 minutes). According to the researchers, lifelong premature ejaculation is the clinical manifestation of the extreme left part of the IELT curve [150,153]. Because the 0.5 percentile equated to an IELT of 0.9 minutes and the 2.5 percentile to an IELT of 1.3 minutes, IELTs of less than 1 minute may be considered as a dysfunction [150,152,153]. By this dynamic approach of endophenotypes, lifelong premature ejaculation ought to be defined in terms of the duration of the IELT [150,154,155].
Waldinger et al. proposed that the genetic component of premature ejaculation could involve dysfunction of the serotonergic receptor system, for example hyposensitivity of the 5-HT2C receptor and/or hypersensitivity of the 5-HT1A receptor . In addition, premature ejaculation might be associated with an increased oxytocin release during intercourse . This hypothesis was based on a rather unknown, and in the literature highly neglected, but clinically very relevant phenomenon. Originally described by Shapiro  in 1943, Waldinger emphasized that many men with lifelong premature ejaculation report having rapidly occurring erections (erectio praecox) as well . Because oxytocin strongly facilitates erection, erectio praecox in the context of lifelong ejaculatio praecox may be associated with increased oxytocin release during coitus [151,155].
In addition, we now hypothesize that the ejaculatory threshold in men with lifelong premature ejaculation might be decreased by a higher oxytocin release, possibly induced by sociosexual stimuli. Although highly speculative, the involvement of oxytocin in social behavior and pair bonding [157,158] might perhaps explain the remarkable clinical phenomenon that many men with complaints of lifelong premature ejaculation report rapid ejaculation only during intravaginal contact and less so during masturbation (unpublished observations by Waldinger et al.).
Theoretical Serotonin–Oxytocin Model
Serotonin is currently believed to exert a tonic inhibition on sexual behavior and to control the ejaculatory threshold [159–161]. Tonic inhibition of sexual behavior is useful, because the initiation of copulation is not always appropriate and would waste valuable energy. The ejaculatory threshold is favorable as well, because this will ensure a high number of intromissions, which in turn increases the chance that a female rat will become pregnant . The presence of sufficient cues, for example pheromones of an estrous female, signaling that copulation might lead to ejaculation and ultimately reproduction, probably causes a gradual disinhibition of sexual reflexes by reducing 5-HT levels [163–166].
Acute injection with SSRIs elevates 5-HT levels throughout the central nervous system. This probably means that the presence of a receptive female is not able to cause the normal reduction in 5-HT levels in male rats treated with an SSRI. The fact that this does not immediately cause inhibition of copulation might be explained by the fact that SSRIs not only elevate 5-HT levels in brain or lumbosacral spinal cord areas that inhibit sexual behavior, but increase serotonergic neurotransmission in many other areas, leading to a variety of effects. One of those effects is the increase of oxytocin release, which might be able to activate sexual reflexes such as erection and ejaculation by itself. Over time, this “compensatory” pathway becomes less efficient because 5-HT1A receptors on oxytocin neurons start to desensitize, ultimately leading to delayed ejaculation. Injection of oxytocin or oxytocin agonists in this situation would reinstall normal sexual behavior . A schematic representation of the theoretic model is given in Figure 3. Hopefully, this highly simplified model will evolve with the addition of future scientific discoveries.
Interestingly, this hypothesis is able to explain not only the delayed onset of ejaculatory dysfunctions during SSRI treatment, but also the differences between the SSRIs: while citalopram and fluvoxamine might continue to compensate the effects on ejaculation through an increased release of oxytocin, paroxetine and fluoxetine could lose this ability via desensitization of hypothalamic 5-HT1A receptors. In addition, this hypothesis gives a possible explanation for the higher incidence of erectile dysfunction in patients treated with paroxetine and fluoxetine compared with fluvoxamine .
Several components of the theory described above have not yet been investigated. First of all, the exact mechanisms of oxytocin release and the relative roles of peripheral and central oxytocin release during normal male sexual behavior need to be studied extensively. In addition, a thorough comparison of citalopram and fluvoxamine on one hand, and fluoxetine and paroxetine on the other, on oxytocinergic neurotransmission should be made. Furthermore, the effects of central and/or peripheral acting oxytocin antagonists in combination with 8-OH-DPAT or SSRIs on male sexual behavior need to be investigated. If the results fit the theory, then oxytocin-related compounds would be very promising drugs to implement in the treatment of various spontaneous or SSRI-induced ejaculatory dysfunctions.