The widespread use of the selective serotonin reuptake inhibitors (SSRIs) has been accompanied by numerous reports describing a potential association with hyperprolactinemia. Antipsychotics are commonly known to elevate serum prolactin (PRL) through blockade of dopamine receptors in the pituitary. However, there is little awareness of the mechanisms by which SSRIs stimulate PRL release. Hyperprolactinemia may result in overt symptoms such as galactorrhea, which may be accompanied by impaired fertility. Long-term clinical sequelae include decreased bone density and the possibility of an increased risk of breast cancer. Through literature review, we explore the possible pathways involved in serotonin-induced PRL release. While the classic mechanism of antipsychotic-induced hyperprolactinemia directly involves dopamine cells in the tuberoinfundibular pathway, SSRIs may act on this system indirectly through GABAergic neurons. Alternate pathways involve serotonin stimulation of vasoactive intestinal peptide (VIP) and oxytocin (OT) release. We conclude with a comprehensive review of clinical sequelae associated with hyperprolactinemia, and the potential role of SSRIs in this phenomenon.
Hyperprolactinemia is an undesirable effect of several classes of psychotropic medications. While it is well recognized in relation to antipsychotic use (Conner and Fried, 1998; Dickson and Glazer, 1999; Kleinberg et al, 1971; Meltzer et al, 1979), there is limited awareness of this adverse effect of selective serotonin reuptake inhibitors (SSRIs). A growing number of case reports and small studies have described PRL abnormalities and/or manifestations such as galactorrhea, amenorrhea, and breast tenderness in association with the use of SSRIs in women (Peterson, 2001; Amsterdam et al, 1997; Bronzo and Stahl, 1993; Morrison et al, 2001; Arya and Taylor, 1995; Iancu et al, 1992; Attenburrow et al, 2001; Bonin et al, 1997; Jeffries et al, 1992; Spigset and Mjorndal, 1997; Cowen and Sargent, 1997; Dulchin et al, 2001; Laine et al, 1997; Urban and Veldhuis, 1991). Possible long-term clinical consequences of hyperprolactinemia, including decreased bone density and a potential increased risk of breast cancer, are only beginning to be investigated (Klibanski et al, 1980; Wang et al, 2002). Subtle impacts on fertility mediated by PRL due to its effects on circulating gonadotropins have also not been studied in patients exposed to SSRIs.
In the first part of this review, we examine the related neuroanatomy and neurobiological mechanisms by which SSRIs may cause hyperprolactinemia. Next, we review the clinical manifestations of hyperprolactinemia. Finally, we outline questions for further research to determine the incidence of, risk factors for, and significance of SSRI-induced hyperprolactinemia.
THE SSRIS AND PROLACTIN
The stimulus for the development of the SSRIs originated in the serotonin hypothesis of depression. This hypothesis was proposed by Carlsson, Van Praag, Asberg, and others (Agurell, 1983; Asberg et al, 1976), based on studies showing a low CSF 5-HIAA response to probenecid in depressed individuals (Van Praag, 1977), decreased central 5-HT in the brains of suicide victims (Pare et al, 1969), and reports of antidepressant effects of tryptophan, a serotonin precursor (Berger, 1975). Additional support for the serotonin hypothesis of depression came from the demonstration of reversal of monoamine antidepressant action in patients pretreated with a 5-HT synthesis inhibitor (Shopsin et al, 1976).
Neuroendocrine challenge tests later helped to substantiate the serotonin hypothesis of depression (Coccaro et al, 1989). These tests are employed to provide an index of central serotonergic function, based on the premise that 5-HT stimulation leads to the release of pituitary hormones such as ACTH and PRL(Yatham and Steiner, 1993). Neuroendocrine probes of serotonergic function include 5-HT precursors (L-tryptophan, 5-hydroxytryptophan), releasing agents (fenfluramine), reuptake inhibitors (clomipramine), and receptor agonists (m-CPP) among others. In humans, the observation that oral administration of fenfluramine to healthy volunteers induced a dose-related increase in PRL secretion provided an index of central serotonergic function (Quattrone et al, 1983), a phenomenon that has been reliably reproduced many times over. Fenfluramine, a serotonin-releasing agent, which also inhibits the reuptake of synaptic 5-HT and stimulates postsynaptic 5-HT2A and 2C receptors (Newman et al, 1998) induces a rapid surge of PRL in laboratory animals and humans (Lu and Meites, 1973; Pinder et al, 1975; Slater et al, 1976). In contrast to healthy controls, a blunted PRL response occurs in some depressed patients when challenged with fenfluramine (Siever et al, 1984; Cleare et al, 1995; Coccaro et al, 1989; Mann et al, 1995; O'Keane and Dinan, 1991; Lopez-Ibor et al, 1988; Mitchell and Smythe, 1990), suggesting abnormal central serotonergic transmission in this condition. However, many studies have failed to show similar results (Asnis et al, 1988; Abel et al, 1997; Maes et al, 1991; Weizman et al, 1988; Kavoussi et al, 1998). The main criticism of studies documenting a positive correlation between a blunted PRL response and depression is the use of heterogeneous patient populations with other psychiatric comorbidities (Newman et al, 1998). More recent studies indicate that blunted PRL response to fenfluramine is associated with impulsive aggression and suicidal behavior, traits associated with depression (Coccaro et al, 1989, 1997a, 1997b; Fava et al, 2000; New and Siever, 2002; Placidi et al, 2001). These findings suggest that serotonin dysfunction may be a marker of traits associated with some forms of depression and other psychiatric conditions, rather than a causal factor of major depression. In patients chronically treated with SSRIs, a normalization of the blunted PRL response would be expected. However, results have been conflicting with some studies showing enhanced PRL response (Kasper et al, 1990; O'Keane et al, 1992), and others finding no change or even decrements in the response (Dulchin et al, 2001; Kavoussi et al, 1999). These studies underscore how individual differences in serotonin responsivity are reflected in variable PRL responsivity across patient samples.
The exact mechanisms through which SSRIs achieve their therapeutic and neuroendocrine effects are not known. SSRIs elevate serotonin levels at the synapse through reuptake blockade of the serotonin transporter (ser-T) (Hyttel, 1984; Tatsumi et al, 1997) which interrupts the normal negative feedback control of presynaptic serotonin release and increases serotonin at the synapse (Dubovsky, 1994). With the exception of fluoxetine, SSRIs generally have little interaction with postsynaptic 5-HT receptors and are mainly thought to work through this presynaptic mechanism. While several SSRIs differentially interact with other neurotransmitter systems (Goodnick and Goldstein, 1998; Hyttel, 1984; Tatsumi et al, 1997), including dopamine (sertraline) and norepinephrine (paroxetine) stimulation of PRL likely involves serotonergic mechanisms, since all SSRIs have been implicated in hyperprolactinemia, regardless of their effects on these other transmitter systems (Attenburrow et al, 2001; Bronzo and Stahl, 1993; Spigset and Mjorndal, 1997; Cowen and Sargent, 1997; Peterson, 2001; Morrison et al, 2001).
Prolactin's Physiologic Role
PRL is a polypeptide hormone, which is secreted in a pulsatile fashion and is structurally related to GH and human placental lactogen (hPL) (Niall et al, 1971). PRL release follows a circadian rhythm, with increased secretion in the early evening hours (Sassin et al, 1972; Waldstreicher et al, 1996). Peak PRL levels occur in the early morning secondary to its increased secretion with prolonged sleep. PRL levels also fluctuate with the menstrual cycle, peaking with ovulation (Seppala, 1978). Physiological increases in plasma PRL levels are seen with stress, pregnancy, sleep, exercise, meals, sexual intercourse, and breastfeeding (see Review by Yazigi (Yazigi et al, 1997)). PRL's actions are mediated at the PRL receptor, which is most densely concentrated in the choroid plexus where active uptake of serum PRL occurs. To a lesser degree, the PRL receptor is also distributed in the paraventricular nucleus (PVN) and other hypothalamic sites (Bakowska and Morrell, 1997; Chiu and Wise, 1994; Crumeyrolle-Arias et al, 1993; Walsh et al, 1987) with generally higher densities found in females (Chiu and Wise, 1994; Muccioli et al, 1991; Pi and Grattan, 1998).
In addition to mammotropic and lactogenic functions (Frantz, 1978), PRL plays a role in developing neuroendocrine and behavioral adaptations in the maternal brain (Grattan, 2001), and influences reproductive function in both males and females (Doherty et al, 1981; Harlan et al, 1983; Witcher and Freeman, 1985). Elevated levels of PRL interfere with the normal pulsatile secretion of LH and FSH, leading to the inhibition of gonadal function (Seppala, 1978; Bohnet et al, 1976; Greenspan, 2001). In humans, elevated PRL levels are associated with hypogonadism and infertility (see below) (Gomez et al, 1977; Katz and Adashi, 1990), while subtle perturbations in PRL dynamics have been linked to infertility of unknown etiology (see below) (Subramanian et al, 1997).
PRL also has a broad role in complex behaviors outside reproduction and lactation, including grooming behaviors (Drago et al, 1983), food intake (Noel and Woodside, 1993; Li et al, 1995) and the response to acute physiological stress (Drago et al, 1990; Holsboer and Barden, 1996; Neill, 1970; Torner et al, 2001). Various forms of stress are associated with increased PRL release, including ether stress (Johnston and Negro-Vilar, 1986), restraint stress (Torner et al, 2001; Gala, 1990; Neill, 1970; Rossier et al, 1980; Seggie and Brown, 1975) thermal stress (Vaha-Eskeli et al, 1991), social conflict in mice (Huhman et al, 1995), and academic stress in humans (Malarkey et al, 1991). PRL circadian rhythm disturbances occur in patients with premenstrual dysphoric disorder, which may be linked to chronobiologic abnormalities in this condition (Parry et al, 1996). Evidence from animal models indicates that PRL interfaces with the HPA axis, blocking stress-induced increases in corticosterone. Consistent with these neuroendocrine effects, PRL exerts anxiolytic effects in these animals (Torner et al, 2001). In humans, low scores on the Hamilton anxiety scale correlated with high PRL levels in healthy lactating women (Asher et al, 1995).
Related to its role in the stress response, PRL participates in the regulation of the immune system (Gala, 1990). PRL enhances thymic function in PRL-deficient mice (Chen et al, 1972) and stimulates T lymphocyte mitogenesis (Viselli et al, 1991; Shiu et al, 1983). PRL involvement in the immune response is further supported by increased lymphocytic PRL gene expression in the setting of graft rejection (Shen et al, 1992). Taken together, these data indicate that PRL is an important part of the physiological response to stress.
ANATOMY OF PRL RESPONSE: HYPOTHALAMIC–PITUITARY CONNECTIONS
The following overview aims to orient the reader to specific hypothalamic–pituitary pathways implicated in PRL release, which can be influenced by serotonin. It is not intended as a comprehensive review of the vast network of hypothalamic–pituitary connections.
PRL is secreted by the anterior pituitary, which is regulated by the hypothalamus. The hypothalamus orchestrates the control of the neuroendocrine system, including PRL release, through a complex series of connections. Different hypothalamic nuclei are composed of cells with unique characteristics and functions. Differential outputs of the various hypothalamic nuclei control autonomic, endocrine, and behavioral responses. Furthermore, many hypothalamic cells also project outside the pituitary, influencing diverse regions in the brain and brainstem (Buijs and Van Heerikhuize, 1982; Sofroniew et al, 1981).
The hypothalamus has three main subdivisions: the periventricular zone, the medial hypothalamus, and the lateral hypothalamus, which are arranged medial-to-lateral from the midline. The periventricular and medial hypothalamic subdivisions are the main regions associated with PRL homeostasis (Figure 1). The periventricular zone is apposed to the ventricular wall, and includes the periventricular nucleus and the arcuate nucleus. Both of these nuclei contain dopaminergic cells, which tonically inhibit PRL release via projections to the anterior pituitary (AP) (Macleod, 1976). This system is known as the ‘tuberoinfundibular dopamine pathway’ (TIDA) (Schofield and Everitt, 1981). PRL release associated with antipsychotic drugs is mediated by this pathway, through blockade of D2 receptors located on the lactotrophs (Bunzow et al, 1988).
The medial hypothalamus contains several prominent nuclei which are implicated in PRL regulation. The paraventricular nucleus (PVN) and supraoptic nucleus (SON) are associated with a variety of functions, which include regulation of metabolism and body temperature, control of water and food intake, and cardiovascular and gastrointestinal functioning (Lechan and Jackson, 1982; Crawley and Kiss, 1985; Leibowitz, 1978; Sawchenko et al, 1981; Porter and Brody, 1985). The PVN contains neurosecretory magnocellular cells, which produce oxytocin (OT) and vasopressin (AVP), and parvocellular cells, which synthesize corticotropin releasing factor (CRF), vasoactive intestinal peptide (VIP), thyroid releasing hormone (TRH), and other neuropeptides (Bloom et al, 1982; Kiss, 1988; Lechan and Jackson, 1982; Merchenthaler et al, 1982; Mezey and Kiss, 1985). The magnocellular OT and AVP cells project to the posterior lobe of the pituitary, while the parvocellular division projects to the AP via the median eminence. OT, which is found only in mammals, is associated with reproductive and maternal behaviors, and is also implicated in a variety of social behaviors (Hughes et al, 1987; Insel, 1992; Winslow and Insel, 1991). It is specifically involved in the initiation of parturition and milk ejection, and is itself a stimulator of PRL secretion (see below) (Samson et al, 1986). VIP, also a major mediator of PRL secretion, is found in cells of the parvocellular PVN (see below) (Ceccatelli et al, 1989). The SON, which has developmental and cellular similarities to the PVN, contains OT-positive cells (Saper, 1990; Van de Kar et al, 1995).
The pituitary is composed of anterior, posterior, and intermediate lobes. The AP arises from the ectoderm of the primitive pharynx known as Rathke's pouch, and is considered a glandular structure, composed of secretory cells (Sheng and Westphal, 1999). PRL is secreted by lactotrophs, and gonadotrophs, somatotrophs, thyrotrophs, and corticotrophs synthesize LH and FSH, GH, TSH, and ACTH, respectively (Laycock, 1983; Perez et al, 1995; Horvath and Kovacs, 1994). PRL and GH are also secreted by mammosomatotrophs which are found in very small numbers in the normal pituitary (Horvath and Kovacs, 1994). Releasing factors produced in the hypothalamus are secreted into the blood supply of the AP, diffusing locally to stimulate release of their respective hormones. Paracrine and autocrine mechanisms are also thought to play an important role in pituitary regulation (Denef, 1986).
The neurohypophysis, also known as posterior pituitary (PP), derives from neural tissue of the primitive hypothalamus, and is devoid of secretory cells (Stopa et al, 1993). OT and AVP produced in the PVN and SON are transported anterogradely and stored in the axon terminals found in the PP. These neurohormones are subsequently released in the systemic circulation in response to a variety of peripheral stimuli (Strand, 1999). Milk ejection in response to suckling is one example in which peripheral stimulation, in this case represented by afferents in the breast tissue, leads to the release of OT (Crowley and Armstrong, 1992).
Although the AP and PP have different embryonic origins, by week 16 of development they are fused together. These two pituitary lobes communicate via short portal vessels, which guarantee that hormones and releasing factors will diffuse between them via their shared blood supply (Liu and Ben-Jonathan, 1994). Thus, hormones released into the PP can influence the AP directly via communicating vessels as well as through the systemic circulation.
NEUROHORMONES AND PEPTIDES INVOLVED IN PRL SECRETION
A number of different agents are involved in the fine balance of stimulation vs inhibition of PRL release (McCann et al, 1984). These are organized according to documented ‘direct’ and ‘indirect’ effects on PRL release, based on in vitro and in vivo studies. Prolactin-inhibiting factors (PIFs) directly decrease PRL secretion, while prolactin releasing factors (PRFs) directly stimulate PRL secretion from lactotrophs. PRFs may also participate in indirect pathways. Neurotransmitters and neuropeptides that rely on intermediaries to stimulate PRL release are considered ‘modulators’ or ‘indirect regulators’ of PRL secretion (Ben-Jonathan, 1994). Indirect regulators can also have important direct effects on PRL gene expression, despite their inability to directly stimulate PRL secretion.
It is important to note that there are a myriad of substances that affect PRL release, although their physiological significance in PRL homeostasis is unclear. These substances include angiotensin II, somatostatin, substance P, neurotensin, galanin, endothelin, calcitonin, and atrial natriuretic peptide, which are not included in this review (for discussion, see Freeman (Freeman et al, 2000)).
PIFs and Other Inhibiting Agents
Dopamine is considered the primary physiologic PIF, and exerts tonic inhibition via two main pathways (Everitt et al, 1984; Martinez de la Escalera and Weiner, 1992). The main pathway is the TIDA, described above. Dopamine cell bodies in the arcuate nucleus send projections to the median eminence to release DA into the portal vessels, through which it reaches the lactotrophs in the anterior pituitary. PRL release is inhibited by the binding of dopamine to D2 receptors on the lactotrophs (Caron et al, 1978; Meller et al, 1991). The other inhibitory pathway, known as the tuberohypophyseal tract (Bjorklund et al, 1973), also originates in the arcuate nucleus but involves DA release into the blood supply of the posterior pituitary. Through that pathway, dopamine can reach the lactotrophs via short portal vessels. When tonic inhibition exerted by dopamine is overcome, PRL release ensues.
In addition to its role in thyroid function, triiodothyronine (T3) is an inhibitor of human PRL gene transcription (Pernasetti et al, 1997), and is likely to participate in the physiological regulation of PRL secretion. Interestingly, TRH, the releasing factor for T3, exerts the opposite effect, acting as a PRF (see below). PRL exerts a short-loop negative feedback on its own release by stimulating TIDA cells through binding to PRL receptors on these neurons (Grattan, 2001).
VIP: VIP stimulates PRL release in vivo and in vitro (Ben-Jonathan, 1994). At the cellular level, VIP directly binds to its receptors on the lactotroph cell membrane, stimulating adenyl cyclase activity, and PRL gene transcription and release (Wanke and Rorstad, 1990). Systemic administration of VIP to rhesus monkeys and man causes a significant elevation of plasma PRL (Frawley and Neill, 1981; Lightman and Young, 1987). Conversely, passive immunoneutralization with VIP antisera reduces or abolishes the normal PRL surge associated with ether stress, suckling, and direct serotonergic stimulation (Shimatsu et al, 1984; Abe et al, 1985) and (Kaji et al, 1985). Afferent sources of VIP include the suprachiasmatic nucleus (SCN) and the PVN (Lam, 1991). However, VIP immunoreactive granules have also been localized in the lactotrophs (Segerson et al, 1989; Morel et al, 1982). VIP can thus regulate PRL release via hypothalamic afferents, and also through direct paracrine and autocrine mechanisms in the AP.
Oxytocin: OT, which is associated with parturition and maternal behaviors, also directly causes PRL release in vitro and in vivo (Mogg and Samson, 1990). Consistent with this, OT receptor mRNA expression is highly specific and restricted to lactotrophs in the AP (Breton et al, 1995). In lactating rats, elevation of OT plasma levels occurs just before the accompanying PRL surge associated with suckling. Administration of antioxytocin serum significantly reduces this PRL surge (Samson et al, 1986). A link between OT and VIP's action on PRL release is suggested by experiments in which animals injected with anti-OT serum prior to intraventricular administration of VIP fail to show elevation of plasma PRL. These results indicate that that OT participates in VIP-induced PRL release (Samson et al, 1989).
TRH: Thyrotropin-releasing hormone stimulates the release of PRL from the AP (Aizawa and Hinkle, 1985). Besides stimulating the secretion of PRL in vivo and in vitro, TRH has also been shown to stimulate PRL gene expression (Benker et al, 1990). TRH binds to membrane receptors on the lactotrophs, activating the phospholipase C signaling pathway (Bjoro et al, 1990). This pathway culminates in the activation of protein kinase C, with consequent increased PRL gene expression (Benker et al, 1990).
Estrogen: Estrogen is a well-known modulator of PRL (Ben-Jonathan, 1994). Daily estrogen use in females increases basal and stimulated PRL secretion within 2–3 days (Greenspan, 2001). Estrogen's influence on PRL occurs through regulation of PRL gene expression (Schaufele, 1999) and indirect stimulation of PRL release (Chen and Meites, 1970; Grosvenor, 1960). The lactotroph expresses both nuclear estrogen receptor α and β subtypes (ERα and ERβ), as well as a truncated form of ERα (Mitchner et al, 1999). Based on information to date, ERβ is most significantly involved in estrogen-induced PRL gene expression (Mitchner et al, 1999). Estrogen binding to its β subtype receptor stimulates the PRL gene promoter, resulting in transcription of the PRL gene (Schaufele, 1999).
In addition to its direct actions on PRL gene expression, estrogen also interacts with PRFs at the hypothalamic level (Benker et al, 1990). Estrogen treatment stimulates PRL release by influencing hypothalamic VIP (Lam et al, 1990), and also OT gene expression (Shughrue et al, 2002).
Opioids: Endogenous opioid peptides such as enkephalins and dynorphins are involved in the regulation of PRL secretion during pregnancy, lactation, and stress (Ferland et al, 1978; Rossier et al, 1980; Sagrillo and Voogt, 1991). The suckling-induced PRL surge is blocked by treatment with naloxone, a nonselective opioid antagonist (Selmanoff and Gregerson, 1986). Naloxone also decreases stress- and estrogen-induced PRL secretion (Dupont et al, 1980; Petraglia et al, 1987).
Opioids act on a hypothalamic level, although a direct effect on the pituitary has not been completely excluded (Enjalbert et al, 1979). Endogenous opioid peptides and synthetic opiates suppress TH activity and m-RNA levels in the TIDA neurons (Arbogast and Voogt, 1998; Deyo et al, 1979; Reymond et al, 1983; Andrews and Grattan, 2003). Therefore, one established mechanism of opioid-induced PRL secretion is through inhibition of TIDA neuronal activity, probably modulated by the opioid receptor subtypes μ and κ (see review, Ben-Jonathan, 1994; Soaje and Deis, 1994). Contacts between opioid-containing terminals and TIDA neurons include pro-enkephalin, pro-dynorphin and pro-opiomelanocortin terminals (Fitzsimmons et al, 1992).
Serotonin: It has long been recognized, based on classic experiments mentioned above, that acute injections of 5-HT or its precursor, 5-HTP, stimulate PRL release (Kamberi et al, 1971). Conversely, blocking either serotonin synthesis or transmission results in a blunted PRL release (Kordon et al, 1973; Gil-ad et al, 1976; Kamberi et al, 1971). These effects depend on one or more PRFs, making serotonin an indirect, albeit potent, modulator of PRL release (Ben-Jonathan, 1994; Freeman et al, 2000) (Figure 1).
VIP and OT are among the best-studied candidates for serotonin's downstream actions on PRL release (McCann et al, 1984; Shimatsu et al, 1982, 1984; Van de Kar et al, 1995, 2001). VIP and OT are both found in the PVN, implicating this structure as an important anatomic substrate of PRL release (Mezey and Kiss, 1985; Sofroniew et al, 1981; Swanson and Sawchenko, 1983). Consequently, the PVN is among the best-studied sites of serotonin-induced PRL release, and is known to contain the postsynaptic 5-HT1A, 5-HT2A, and 5-HT2C receptor subtypes (Van de Kar et al, 2001; Jorgensen et al, 1992, 2003; Appel et al, 1990; Rittenhouse et al, 1993; Van de Kar et al, 1989). 5-HT2A/2C agonists increase Fos immunoreactivity (an index of synaptic activity) in the PVN (Van de Kar et al, 2001) ,while ablation of the PVN results in a markedly decreased PRL response to serotonin agonists (Bagdy, Rittenhouse #7066, Van de Kar #7052).
Recruitment of VIP neurons in the serotonin-induced PRL response is implied by the fact that VIP antiserum significantly blunts the usual PRL surge associated with serotonergic stimulation in vivo (Kaji et al, 1985; Shimatsu et al, 1982,). While it is hypothesized that VIP neurons of the PVN are directly stimulated by serotonergic fibers (Kiss et al, 1984; Lam, 1991; Reichlin, 1988), to date, serotonergic terminals have not been demonstrated on these VIP-expressing cells (Kiss et al, 1984).
Oxytocin, another PRF found in the PVN (magnocellular subdivision), is also stimulated by serotonin agonists in male and female animals (Moos, F l983; Saydoff, J l991; Van de Kar et al, 1995). Recent studies show that OT is also stimulated by fenfluramine in healthy men (Lee et al, 2003). These data, combined with the observation that bilateral PVN lesions inhibit the effect of the 5-HT-receptor agonists on PRL release, suggest that serotonin may stimulate PRL secretion through the OT system (Arey and Freeman, 1989; Van de Kar et al, 1995). In support of this idea, D-fenfluramine administration induces Fos- immunoreactivity in oxytocin-containing cells in the PVN (Javed et al, 1999).
Taken together, pharmacologic and anatomic data indicate that the PVN represents a major regulatory site of serotonin-induced PRL release, although the relative role of VIP and OT pathways in this response is not established. Furthermore, while ablation of the PVN blunts the PRL response to 5-HT agonists, it does not entirely abolish it (Bagdy, 1996; Rittenhouse et al, 1993), demonstrating that other pathways contribute to serotonin-induced PRL release.
An alternate path for serotonin-induced PRL release is inhibition of the tuberoinfundibular dopamine cells (TIDA). However, there is little synaptic contact between serotonergic fibers and the dopamine cells, indicating that if direct inhibition of dopamine cells occurs, it is through volume transmission of serotonin in the region (Kiss and Halasz, 1986). There is more direct evidence for serotonergic stimulation of GABAergic neurons in the vicinity of the TIDA dopamine cells, based on the presence of 5-HT1A receptors on these cells (Mirkes and Bethea, 2001). Assuming that these GABAergic cells are interneurons, their stimulation by 5HT would result in inhibition of TIDA cells, releasing the tonic inhibition of PRL (see Figure 1). Consistent with this idea, several pharmacologic studies show that GABA agonists injected in the hypothalamus promote PRL secretion (Ondo and Dom, 1986; Fuchs et al, 1984; Wagner et al, 1994), as does alprazolam administration (Bondolfi et al, 1997; Shioiri et al, 1996; Zemishlany et al, 1990). However, intraventricular injection of GABA has yielded differential effects depending on the dose utilized (with a reduction in plasma PRL seen with low (physiologic) doses, and increases with higher doses) (McCann et al, 1984; Vijayan and McCann, 1978). Several benzodiazepines have also been reported to decrease PRL secretion (Jarvinen et al, 1992). These collective results suggest that PRL levels are differentially influenced by GABAergic drugs depending on variables such as drug potency (affinity for the GABA receptor) and dose.
Hyperprolactinemia: Established pharmacologic causes of hyperprolactinemia include D2 receptor antagonists, tricyclic antidepressants, methyldopa, reserpine, and verapamil. Diseases related to hyperprolactinemia include prolactinomas and tumors compressing the pituitary stalk, hypothyroidism, and decreased clearance from renal insufficiency (Conner and Fried, 1998; Katz and Adashi, 1990; Klein et al, 1964; Quigley et al, 1979; Wieck and Haddad, 2003; Yazigi et al, 1997). The upper normal value for serum PRL in most laboratories is 20 ng/ml (nanograms per milliliter) for men and 25 ng/ml for women. Serum PRL levels between 20 and 200 ng/ml are generally associated with any cause of hyperprolactinemia, and values above 200 ng/ml are usually seen with prolactinomas (Gomez et al, 1977; Melmed, 2001; Schlechte et al, 1989).
The classic manifestations of hyperprolactinemia are galactorrhea, amenorrhea, infertility, and decreased libido in women, and erectile dysfunction, hypogonadism, and infertility in males (Gomez et al, 1977; Carter et al, 1978; Segal et al, 1979; Seppala, 1978). Long-term clinical consequences of hyperprolactinemia are less obvious, and include osteopenia in men and women (Greenspan et al, 1986; Jackson et al, 1986; Klibanski et al, 1988) and the possibility of an increased risk of breast cancer in women (Welsch and Nagasawa, 1977; Kwa et al, 1981). PRL levels associated with impaired fertility, decreased bone density and breast cancer have not been established. However, these conditions have a major public health impact and are therefore of special interest.
Menstrual disturbances associated with hyperprolactinemia range from luteal phase dysfunction with normal menses to frank amenorrhea (Bohnet et al, 1976; Katz and Adashi, 1990; Yazigi et al, 1997). Subtle abnormalities in PRL secretion may impact fertility. In a study of women with infertility of unknown etiology, the normal midcycle increase of PRL was absent in the patients, but maintained in healthy controls. All subjects had normal basal PRL levels, implicating loss of normal pulsatility in fertility problems (Subramanian et al, 1997). Consistent with this, altered fertility associated with only modestly elevated PRL levels is well recognized with antipsychotic drug use (Dickson and Glazer, 1999; Kinon et al, 2003; Smith, 2003), but has not been studied with respect to SSRIs. Aberrant PRL levels can interfere with female fertility through both CNS and peripheral mechanisms. At the hypothalamic level, high PRL interferes with LH pulsatility due to abnormalities in GnRH secretion (Sauder et al, 1984). Another mechanism is a direct inhibitory effect of PRL on ovarian follicular development and steroidogenesis (McNatty, 1979). Corpus luteum deficiency is also a mechanism of infertility due to PRL's luteolytic effect (Kredentser et al, 1981).
Decreased Bone Density
Bone loss and the long-term development of osteoporosis are deleterious effects of hyperprolactinemia in both men and women. There is little evidence supporting a direct role for PRL in bone homeostasis (Schlechte et al, 1983; Ciccarelli et al, 1988). Rather, elevated PRL levels disrupt the normal pulsatile secretion of LH and FSH, leading to hypogonadism (Greenspan, 2001; Bohnet et al, 1976; Seppala, 1978; Greenspan et al, 1986; Quigley et al, 1979; Sauder et al, 1984), and subsequent osteopenia (Klibanski et al, 1988, 1980; Schlechte et al, 1987, 1992; Koppelman et al, 1984; Di Somma et al, 1998; Carter et al, 1978). Hypoestrogenemia as a requisite for osteopenia in women is supported by studies showing that hyperprolactinemic women with normal menstrual cycles do not show decreased bone density (Ciccarelli et al, 1988; Klibanski et al, 1988; Schlechte et al, 1992; Wardlaw and Bilezikian, 1992). One study reports that hyperprolactinemic women who were amenorrheic for less than a year did not have significant bone loss (Cann et al, 1984), and several studies indicate an absence of continued bone loss even when exposure to elevated PRL levels occurs over several years (Biller et al, 1992; Koppelman et al, 1984; Schlechte et al, 1992). However, there are no definitive data describing what duration or degree of elevated PRL levels are necessary to result in decreased bone density.
Bone loss, once established, may persist after resolution of hyperprolactinemia, indicating that even transient elevations in PRL may be an important risk factor for osteoporosis (Biller et al, 1992; Di Somma et al, 1998; Schlechte et al, 1987, 1983). Hyperprolactinemia in adolescents is associated with persistent bone loss (Coelho et al, 1999). Since the achievement of peak bone mass represents an important protection against osteoporosis (Bonjour, 1996), adolescents are an important population in which to assess drug-induced PRL abnormalities. In a small uncontrolled study of subjects on antipsychotics for at least 6 months, bone mineral density and PRL levels were inversely correlated (Abraham et al, 2003).
A subject of controversy is the role of PRL in the genesis of breast cancer. Although it has been demonstrated that PRL increases the growth of malignant breast cells in vitro (Welsch and Nagasawa, 1977), the clinical significance of such findings is an area of much debate.
The best epidemiologic study evaluating the association between elevated PRL and breast cancer was a nested case–control study conducted within the prospective Nurses' Health Study cohort of 121 700 women (Hankinson et al, 1999). In this study, 306 women diagnosed with breast cancer were matched with 448 controls. All participants had blood samples collected between the period of 1989–1990, at which point they were 43–69 years of age and all postmenopausal. Case patients were women diagnosed with breast cancer after blood collection. A statistically significant higher risk of breast cancer was found in women who had PRL levels above 9.7 ng/ml (relative risk of 2.03 with a 95% CI (1.24–3.31)), even when several established breast cancer risk factors were controlled for. In another prospective study with 40 postmenopausal breast cancer patients, Wang et al found a positive but nonsignificant association between elevated PRL levels and the risk of developing breast cancer (Wang et al, 1992).
There is also an association between elevated PRL levels and poor survival in breast cancer patients. PRL levels above 12.6 ng/ml were linked to a poor survival prognosis in postmenopausal subjects who had undergone mastectomy for breast cancer (Wang et al, 1995). Plasma PRL levels above 32 ng/ml were seen in 8% of 149 women with metastatic breast cancer, in contrast to none of the 221 control subjects (p<0.001) (Holtkamp et al, 1984). In the same study, mean survival time after mastectomy in the group with normal PRL was 154 months, compared with 89 months in the hyperprolactinemic group. Patients on PRL elevating or lowering drugs were excluded from the study.
Drug-induced risks from antidepressant medications are only beginning to be investigated. A recently published case–control study investigated the association between prior antidepressant use and risk of breast cancer (Moorman et al, 2003). Overall, there was no increased risk of breast cancer with ever-use when all antidepressant categories were taken into account (tricyclics, monoamine oxidase inhibitors, SSRIs). However, there was a trend towards increased risk with use of SSRIs for 36 months or longer. In another case–control study, Kelly et al also found a borderline statistically significant increased risk for breast cancer associated with SSRI use, noting that these results were ‘not reassuring’(Kelly et al, 1999). It is not entirely clear whether a possible association between antidepressants and increased risk of breast cancer is related to elevation in PRL levels or direct tumor growth-promoting properties of certain antidepressants (Brandes et al, 1992). However, the results from these studies warrant further investigation of the potential association between serotonin-enhancing drugs, PRL, and breast cancer.
CLINICAL RELEVANCE OF SSRI-INDUCED HYPERPROLACTINEMIA: EMERGING DATA
While the link between acute serotonin stimulation and PRL release has been long established, the clinical effects of chronic serotonin stimulation on PRL has only been investigated recently. A review of the literature retrieved 13 single case reports describing nonpuerperal lactation associated with the use of SSRIs in women (Bronzo and Stahl, 1993; Morrison et al, 2001; Peterson, 2001; Arya and Taylor, 1995; Iancu et al, 1992; Bonin et al, 1997; Jeffries et al, 1992; Lesaca, 1996; Bondolfi et al, 1997; Davenport and Velamoor, 2002; Gonzalez et al, 2000; Otero et al, 2002; Pablos et al, 2001). In these case reports, the patients were all females, most of them premenopausal. Plasma PRL levels ranged between 28 and 60 ng/ml. A common theme of the single case reports is the onset of galactorrhea, usually associated with amenorrhea, shortly after initiation of treatment with an SSRI. In all reports, symptoms promptly subsided with discontinuation of the drug. All SSRIs are implicated in these reports.
There have also been several uncontrolled studies assessing changes in PRL level with therapy with SSRIs (Amsterdam et al, 1997; Attenburrow et al, 2001; Bronzo and Stahl, 1993; Morrison et al, 2001; Peterson, 2001; Spigset and Mjorndal, 1997; Cowen and Sargent, 1997; Dulchin et al, 2001; Laine et al, 1997; Urban and Veldhuis, 1991). All reports showed varied degrees of basal PRL elevation with SSRI treatment. Amsterdam et al (1997) examined the incidence of mammoplasia (breast enlargement) in 59 women taking SSRIs or venlafaxine, finding the highest rate with paroxetine compared to the other antidepressants. Paroxetine-treated patients exhibited statistically significant elevations in PRL levels, although all subjects on fluoxetine, sertraline or venlafaxine showed nonsignificant elevation of their basal PRL. Although significant effects were only seen for paroxetine, this may be related to the relatively greater number of patients on paroxetine (n=28), compared to the other drugs (fluoxetine (n=8), sertraline (n=4), and venlafaxine (n=19)), resulting in inadequate power to detect changes in the latter groups.
The incidence and prevalence of hyperprolactinemia in patients taking SSRIs will be important to pursue in future controlled studies. Based on cumulative case reports, all SSRIs have the potential to cause elevation of basal PRL. This observation was recently confirmed by the French Pharmacovigilance Database Study, an epidemiological study that investigated the rates of hyperprolactinemia induced by multiple prescription medications from 1985 to 2000 (Petit et al, 2003). Of the total of 159 cases of drug-induced hyperprolactinemia, 17% had been induced by SSRIs, which included sertraline (OR=odds ratio 15.74), fluoxetine (OR 49), paroxetine (OR 8.10), fluvoxamine (OR 5.96), and citalopram (OR 3.62). Citalopram was the only SSRI not to reach statistical significance. The available data indicate that SSRI-induced hyperprolactinemia is a class-related effect.
The small number of case reports may be an indication that SSRI-induced hyperprolactinemia is not a common phenomenon. An alternative explanation is that symptoms may go unrecognized by treating clinicians either because they are subtle, or patients tend not to report them. Based on case reports, women seem to be at higher risk for SSRI-induced hyperprolactinemia. Because recognition of galactorrhea and amenorrhea induced by SSRIs leads to prompt discontinuation of the drug without further deleterious consequences, other than the interruption of treatment, the issue of SSRI-induced hyperprolactinemia has been generally seen as a benign one. However, classic definitions of clinical hyperprolactinemia, based on primary pituitary pathology, may not be entirely applicable for drug-induced PRL hypersecretion. It remains to be answered if minimal elevations of PRL, which do not qualify as hyperprolactinemia by current laboratory criteria, impact gonadal function, with consequent effects on fertility and bone density. As mentioned above, subtle alterations in PRL have been demonstrated in women with unexplained infertility (Subramanian et al, 1997). It is thus important that future studies measuring changes in PRL take into consideration the individual baseline levels of the study subjects. It is possible that a specific PRL level that would not cause any problems in one individual may induce clinically important pathology in another. Given the large number of patients taking SSRIs (Frank et al, 2001), the incidence of PRL-associated effects may be significant, even if a small percentage of patients are affected. Moreover, SSRIs are commonly prescribed to postmenopausal women, a population in which the diagnosis of hyperprolactinemia is difficult to recognize, given their hypoestrogenic state, and lack of obvious clinical manifestations. It is unknown whether even mild PRL elevations would add to the risk of osteoporosis and breast cancer in that population.
This review highlights possible mechanisms and manifestations of serotonergic stimulation of PRL, considering them in the context of SSRI use. There are a number of different pathways through which SSRIs may cause PRL abnormalities, including release of a PRF such as VIP or OT, and the inhibition of the TIDA system. The wide-ranging effects of PRL on reproduction, lactation, stress, and immune responses, suggest a fundamental role in maintaining the homeostasis of the organism. This view is supported by the anatomy of this system, which involves complex, and perhaps redundant, pathways mediating PRL regulation.
Future studies should characterize changes in PRL associated with SSRI use, and determine the magnitude and duration of PRL elevations and their association with chronic sequelae such as blunted fertility, decreased bone density, and elevated risk of breast cancer. Measuring pre- and post-treatment PRL levels in well-characterized patient groups will begin to address these issues, advancing our understanding of serotonin–prolactin interactions.
We wish to thank Dr Barbara Gracious and Dr Lawrence Guttmacher for their critical reading and comments.