Bone Demineralization and Osteoporosis
There is a small increased potential risk for osteoporosis during treatment with carbamazepine or SSRIs. Older adults or other patients who have an increased risk for osteoporosis may warrant periodic monitoring of bone mineral density during treatment with anticonvulsants or SSRIs, and consideration may be given to alternative pharmacotherapies when feasible.
Certain anticonvulsant drugs, particularly those that induce liver enzymes (e.g., carbamazepine), are known to decrease bone mineral density. In nongeriatric adults with epilepsy, long-term use of divalproex has been linked with an increased risk for osteoporosis or osteopenia in some studies but not others. Some investigators have suggested that drugs that inhibit histone deacetylase (e.g., divalproex) may have value in promoting osteoblast maturation, thereby mitigating a risk for bone demineralization. To date, no published reports have linked osteoporosis or osteopenia with gabapentin, lamotrigine, oxcarbazepine, topiramate, levetiracetam, or tiagabine.
Separately, SSRIs have been shown to reduce bone mineral density in hip and lumbar spinal joints by approximately 4%–6% in older adult men and women, presumably due to inhibition of serotonin sites in osteoblasts and osteoclasts. One prospective population-based study of 7,983 individuals found a 2.35-fold increased risk for nonvertebral fractures among SSRI recipients among individuals ≥age 55 (Ziere et al. 2008). Whether these observations pose a broad, clinically meaningful risk for fractures in older adults remains the subject of controversy, and at present no formal recommendation has been made either to perform bone densitometry studies on patients before and during treatment with SSRIs or to refrain from using SSRIs in osteoporotic patients for whom no other safety concerns exist regarding their use. Furthermore, depression itself has been suggested to induce bone loss via hypothalamic-pituitary-adrenocortical axis hyperactivity (Schweiger et al. 2000).
Of note, the use of hypermetabolic thyroid hormone as a psychotropic intervention (e.g., as may occur in rapid-cycling bipolar disorder) poses a potential risk for hastening bone demineralization. Some authorities advise periodic bone mineral densitometry testing in patients who continue high-dose thyroid hormone as a long-term therapy.
Hyperprolactinemia, Galactorrhea, and Gynecomastia
Symptomatic hyperprolactinemia warrants either changing from a prolactin-elevating to a within-class prolactin-sparing drug or augmentation with a dopamine agonist such as bromocriptine or amantadine. Adjunctive aripiprazole also has preliminarily been shown to counteract hyperprolactinemia caused by other antipsychotics. Clinically asymptomatic hyperprolactinemia poses a long-term risk for osteoporosis and infertility and merits either periodic laboratory monitoring (if current benefits outweigh potential risks) or intervening by a medication change or augmentation with a dopamine agonist.
Antipsychotic-induced blockade of dopamine D2 receptors on mammotropic cells of the anterior pituitary can cause release of prolactin. Hyperprolactinemia is a well-known consequence of virtually all FGAs and several SGAs (notably, risperidone and paliperidone), as described in Table 11–1. The clinician must bear in mind that hyperprolactinemia can result from other medications, including calcium channel blockers, TCAs, the sleep aid ramelteon, opiates, and histamine H2 antagonists. Psychotropic agents other than antipsychotics have also been reported, although more rarely, to cause hyperprolactinemia and consequent gynecomastia with galactorrhea; these include venlafaxine, fluoxetine, paroxetine, diazepam (in the case of diazepam, possibly secondary to estrogen elevation), and methamphetamine. Hyperprolactinemia from some serotonergic antidepressants may arise via indirect GABAergic modulation of tuberoinfundibular dopaminergic projections (Emiliano and Fudge 2004). Among the SSRIs, the risk for hyperprolactinemia may be especially low with sertraline (Sagud et al. 2002). Rarely, carbamazepine and divalproex also have been implicated as causes of hyperprolactinemia.
Hyperprolactinemia also can result from a wide range of primary medical conditions, including pituitary tumors, macroprolactinomas, polycystic ovary syndrome, pregnancy, sarcoidosis, adrenal insufficiency, and hypothyroidism, as well as from decreased elimination due to renal or hepatic failure. Serum prolactin elevations due to antipsychotics or other psychotropic medications typically rise no higher than 100 mg/dL, although serum levels exceeding 200 mg/dL (usually otherwise suggestive of prolactinomas) have rarely been reported with use of risperidone (Melmed et al. 2011). A clinical practice guideline of The Endocrine Society advises against treating asymptomatic drug-induced hyperprolactinemia (Melmed et al. 2011). Limited patient-specific risk factors for the development of hyperprolactinemia during antipsychotic treatment have been identified and include female sex, possibly premenopausal > postmenopausal reproductive status, and antipsychotic dosage.
Reported rates of hyperprolactinemia with SGAs in controlled trials are summarized in Table 11–1.
Some SGAs, including aripiprazole and asenapine, have been reported in clinical trials to significantly lower serum prolactin levels from baseline. However, it is often a matter of speculation as to whether reductions in prolactin occur via an active normalizing effect of a newly introduced agent, or rather, the diminution of hyperprolactinemia after simply discontinuing a previous prolactin-elevating drug. SGAs appear less prone to elevate serum prolactin if they minimally interfere with tuberoinfundibular dopamine transmission and have “loose” D2 binding affinities (i.e., high dissociation constants), as seen with quetiapine and clozapine. On the other hand, lurasidone demonstrates minimal prolactin elevation despite its low D2 dissociation constant. It has been suggested that differences in prolactin elevation among SGAs may reflect their differential blood-brain disposition, with greater likelihood for hyperprolactinemia among agents with higher pituitary than striatal D2 occupancy (Kapur et al. 2002).
Risk for hyperprolactinemia with a given agent may also vary on the basis of patients’ clinical characteristics. For example, in studies that controlled for risperidone dosage, episode number, or illness duration, serum prolactin levels were significantly higher among risperidone-treated patients with paranoid versus disorganized schizophrenia or schizoaffective disorder—possibly reflecting differences in basal dopaminergic tone among subtypes of psychotic disorders. The clinician does not routinely measure serum prolactin levels in patients taking antipsychotics without a clinical reason to do so. For example, because hyperprolactinemia is expectable with FGAs, its documentation may help to affirm treatment adherence. In patients who develop gynecological disturbances (e.g., amenorrhea, oligomenorrhea, gynecomastia, galactorrhea), measurement of serum prolactin can help clarify etiology.
Changes from baseline in serum prolactin levels
SMD (95% CI)a
Schizophrenia: No appreciable changes (Hamner 2002).
Schizophrenia: No appreciable changes (Hamner 2002).
Schizophrenia: –13.4 to –14.1 mg/dL.
Schizophrenia: –18.3 mg/dL (48-week open-label flexible-dose studies).
Schizophrenia: –56.5% (Marder et al. 2003).
Schizophrenia: –34.2 mg/dL over 26 weeks in a multisite randomized industry-supported trial in schizophrenia (Hanssens et al. 2008).
–0.22 (–0.46, 0.03)
Bipolar disorder: 84-day placebo-controlled monotherapy trial, –15.3±40.9 mg/dL (no different from placebo) (Adler et al. 2007).
Schizophrenia: –10.6 mg/dL over 9.2-month median exposure in CATIE (Lieberman et al. 2005).
–0.05 (–0.23, 0.13)
Bipolar disorder: +4.9 mg/dL over 3 weeks.
Schizophrenia: –6.5 mg/dL over 6 weeks (manufacturer’s package insert, Merck Pharmaceuticals).
Schizophrenia: –26.9 mg/dL over 52 weeks.
0.12 (–0.12, 0.37)
Pooled FDA registration trial data across adult indications: +30% increase from baseline over 12 weeks (cf. 10.5% of placebo-treated patients).
Adolescents with schizophrenia or bipolar disorder: Serum prolactin elevations observed in 47% (cf. 7% with placebo) (manufacturer’s package insert, Eli Lilly and Company).
–8.1 mg/dL over 9.2-month median exposure in CATIE (Lieberman et al. 2005).
0.14 (0.00, 0.28)
Schizophrenia: FDA registration trials reported a mean serum prolactin change of +2.6 mg/dL over 4 weeks, with elevated prolactin levels seen in 26% of iloperidone recipients. A pooled analysis of 6-week acute phase trials demonstrated significant reductions from baseline in serum prolactin levels (ranging from –23.1 to –38.0 mg/dL) (Weiden et al. 2008).
0.21 (–0.09, 0.51)
Schizophrenia: Pooled findings from three acute trials identified mean serum prolactin changes ranging from –1.31 to 3.42 mg/dL (females) and –2.16 to –0.47 mg/dL (males) (Correll et al. 2016).
Major depressive disorder: 2 mg/day adjunctive therapy over 6 weeks: +8.3 mg/dL (females), +2.2 mg/dL (males) (Thase et al. 2015a); 1 or 3 mg/day adjunctive therapy over 6 weeks: 0 patients taking 1 mg/day and 0.4% of those taking 3 mg/day had serum prolactin levels >3 times the upper limit of normal (Thase et al. 2015b).
Transient elevation from baseline that normalizes in healthy volunteers or across studies in psychotic disorders (Hamner 2002).
–5.6 mg/dL over 9.2 month median exposure in CATIE (Lieberman et al. 2005).
0.25 (0.01, 0.49)
Schizophrenia: –1.1 mg/dL (dose dependent; +0.3 ng/mL with 40 mg/day; +1.1 mg/dL with 80 mg/day; +3.3 mg/dL with 120 mg/day; manufacturer’s package insert, Sunovion Pharmaceuticals).
–1.9 mg/dL at 24 weeks; –5.4 mg/dL at 36 weeks; –3.3 mg/dL at 52 weeks (from schizophrenia open-label extension phase studies by manufacturer).
0.34 (0.11, 0.57)
Schizophrenia: Up to 66% of women and 45% of men demonstrated significant elevations in serum prolactin from baseline (Kinon et al. 2003).
Across diagnoses in children and adolescents: In FDA registration trials, up to 87% had dose-dependent serum prolactin elevations (manufacturer’s package insert, Janssen Pharmaceutica).
+13.8 mg/dL over 9.2-month median exposure in CATIE (Lieberman et al. 2005).
1.23 (1.06, 1.40)
Schizophrenia: 6-week placebo-controlled monotherapy (comparison with quetiapine)b trial +38.4±42.8 mg/dL (Canuso et al. 2009).
1.30 (1.08, 1.51)
Note. CATIE=Clinical Antipsychotic Trials of Intervention Effectiveness; FDA=U.S. Food and Drug Administration; SD=standard deviation; SMD=standardized mean difference.
bIndustry trial sponsored by Ortho-McNeil Janssen Scientific Affairs, Johnson & Johnson Pharmaceutical Research and Development, and Janssen-Cilag.
No authoritative recommendation indicates whether or when iatrogenic hyperprolactinemia requires intervention. Persistent hyperprolactinemia can lead to osteoporosis, infertility, and hypogonadism in men. Some authors advise favoring prolactin-sparing SGAs in patients with existing osteoporosis, or in women with breast cancer or a history of breast cancer, inasmuch as prolactin may be trophic to some breast tumors (such that aripiprazole may be the preferred agent) (Citrome 2008).
In symptomatic patients with antipsychotic-induced hyperprolactinemia, most authorities favor changing from a prolactin-elevating antipsychotic (e.g., risperidone, paliperidone) to a prolactin-sparing antipsychotic (e.g., cariprazine, aripiprazole, quetiapine; see Table 11–1) as the first-line intervention. Alternatively, the cautious use of adjunctive dopamine agonists (e.g., bromocriptine 2.5–10 mg/day, amantadine 100–300 mg/day, pramipexole ≤1 mg/day, pergolide 0.05–0.1 mg/day, or ropinirole 0.75–3 mg/day) may effectively suppress antipsychotic-induced hyperprolactinemia, although the clinician must observe for rare but possible exacerbations of psychosis or mania. Adjunctive low-dose aripiprazole (3 mg/day) has also been described in open trials as an effective strategy to normalize hyperprolactinemia induced by risperidone or paliperidone and in a randomized placebo-controlled study (dosed at 15–30 mg/day) to counteract haloperidol-induced prolactin elevation (Shim et al. 2007), with no adverse effect on psychopathology.
Gynecomastia occurs from a hypertrophic effect of prolactin on mammary tissue. Although gynecomastia is not medically significant, it can be a distressing side effect for men or women. Long-standing high prolactin exposure to mammary tissue may cause a pharmacologically irreversible hypertrophy that may be remediable only by surgical procedures (e.g., liposuction, breast reduction). Treatment with dopamine agonists has not shown reductions in hypertrophic mammary tissue caused by elevated prolactin levels.
Several apparently safe herbal remedies have been reported to counteract antipsychotic-induced hyperprolactinemia, although their mechanisms of action are not well defined. The herb Peony-glycyrrhiza decoction dosed at 45 g/day significantly reduced risperidone-induced hyperprolactinemia over 4 weeks, with magnitude comparable to that seen with bromocriptine 5 mg/day; other herbal remedies with open or preliminary randomized controlled data to reduce antipsychotic-induced hyperprolactinemia without exacerbating psychiatric symptoms include Shakuyaku-kanzo-to, Zhuang Yang capsule, and Tongdatang serial recipe (reviewed by Hasani-Ranjbar et al. ).
Menstrual Disturbances and Polycystic Ovary Syndrome
New-onset menstrual disturbances should be evaluated for changes related to medications that may elevate serum prolactin or otherwise interfere with the menstrual cycle. In women of childbearing potential, menstrual cycles should be monitored during treatment with divalproex due to a potential increased risk for polycystic ovary syndrome (PCOS).
A number of psychotropic agents can cause menstrual irregularities through a variety of mechanisms. Irregular menstruation caused by hyperprolactinemia (e.g., secondary to antipsychotic medications) is readily diagnosed by measurement of serum prolactin levels.
PCOS was defined by the National Institutes of Health as the presence of hyperandrogenism with oligomenorrhea (Zawadzki and Dunaif 1992). Documented anatomical evidence of subcapsular ovarian cysts (known as polycystic ovaries) is not considered necessary for the diagnosis of PCOS, although it is often present.
Concerns that divalproex might contribute to PCOS arose following a large observational study of 238 women with epilepsy, in whom the use of divalproex was associated with a higher prevalence of irregular menses (45%) or anatomical evidence of polycystic ovaries (43%) than occurred during treatment with carbamazepine or other anticonvulsants (Isojärvi et al. 1993). Extensive subsequent debate ensued about parsing the effects of divalproex on the menstrual cycle relative to the potential unique contributions of epilepsy, obesity, and other factors. A later study specifically in women with bipolar disorder found a 7.5-fold increased relative risk for developing PCOS during divalproex treatment than during therapy with other anticonvulsants (Joffe et al. 2006). Collectively, these reproductive findings and the drug’s teratogenic risks have prompted some authorities to advise against the use of divalproex among women of childbearing potential. Less extreme perspectives would favor monitoring the menstrual cycle of any woman of reproductive potential who receives divalproex, and further evaluation (i.e., measurement of serum androgen levels) would likely be warranted in the presence of clinical signs of a change in menstrual patterns. The treatment of PCOS involves the discontinuation of agents that may be causing it or the administration of oral contraceptives.
There is no absolute contraindication to prescribing divalproex in women with preexisting PCOS. In other words, no evidence indicates clinical worsening of existing PCOS after superimposition of divalproex, although most practitioners would likely be reluctant to risk further disruption of the hypothalamic-pituitary-gonadal steroid axis by prescribing divalproex.
Table 11–2 provides recommendations for monitoring reproductive safety in women who take divalproex during reproductive years.
Noniatrogenic causes of hypercalcemia in conjunction with hyperparathyroidism should be investigated medically. Iatrogenic hyperparathyroidism may result from lithium and thiazide diuretics. Lithium discontinuation may not necessarily reverse secondary hyperparathyroidism. Imaging studies to discern hyperplasia from multiglandular pathology should occur in conjunction with endocrinological consultation. Parathyroidectomy or use of calcimimetics may be necessary to restore calcium homeostasis.
Hypercalcemia in conjunction with both hyperplastic and multiglandular hyperparathyroidism has been reported to occur in about 10%–25% of patients taking lithium, more commonly (4:1 ratio) in women than in men (Albert et al. 2013; Meehan et al. 2015; Shapiro and Davis 2015). Lithium can cause a shift in the inhibitory set point for parathyroid hormone secretion to a higher serum calcium concentration, while antagonizing the calcium sensing receptor (CaSR) located on the surface of parathyroid cells, which results in an increase in the threshold levels of extracellular calcium required to suppress PTH release from the parathyroid gland, leading to increased serum PTH. Lithium also can inhibit renal excretion of calcium, causing hypercalcemia with inappropriately low-normal urinary calcium excretion, measurable by 24-hour urine collection. (By contrast, primary hyperparathyroidism typically causes both hypercalcemia and hypercalciuria.) Duration of lithium exposure may be a predisposing risk factor. We recommend periodic monitoring of serum calcium and PTH in patients taking lithium.
1. Discuss possibility of reproductive and endocrinological side effects with female patients before beginning divalproex treatment.
2. Measure baseline body weight and body mass index, and follow both at each visit; evaluate weight gain.
3. Obtain baseline information about menstrual cycle and assess for menstrual irregularities at each visit.
4. Consider obtaining baseline and subsequent information about ovarian structure (by ultrasound) and serum sex hormone concentrations, including the following:
•Lutenizing hormone (LH)/follicle stimulating hormone (FSH) ratio (if >2–3 may be consistent with polycystic ovary syndrome [PCOS])
•Morning, fasting 17-hydroxyprogesterone during follicular phases of menstrual cycle (values >200 ng/dL may suggest 21-hydroxylase deficiency due to late-onset congenital adrenal hyperplasia)
•Dehydroepiandrosterone (DHEA; usually normal or slightly elevated in PCOS; values >800 μg/dL may suggest possible androgen-secreting adrenal tumor)
•Prolactin (usually normal or only mildly elevated in PCOS; marked elevation in the absence of dopamine-blocking drugs may suggest pituitary hyperprolactinoma)
•Total testosterone (may be normal [<150 ng/dL] in PCOS; higher levels may reflect an androgen-secreting ovarian or adrenal tumor)
5. Obtain baseline and annual serum lipid profiles.
6. Evaluate at baseline signs of androgen excess (e.g., hirsutism, alopecia, acne) and follow up at each visit.
7. Refer to a reproductive endocrinologist if abnormalities in the laboratory values described above are accompanied by two or more of the following symptoms: hirsutism, menstrual disturbances, obesity, alopecia, or infertility; consider discontinuing divalproex if a thorough risk-benefit analysis demonstrates unfavorable risk.
8. Counsel patient on nutrition and weight management strategies.
9. For women <age 20 years:
•Divalproex is not contraindicated, but use with caution and follow the recommendations listed above.
•Consider pretreatment workup consisting of serum testosterone level and pelvic ultrasound, and consider repeating annually.
•Consider changing to a different mood stabilizer if clinical symptoms of hyperandrogenism or PCOS appear.
Source. Adapted from Ernst CL, Goldberg JF: “The Reproductive Safety Profile of Mood Stabilizers, Atypical Antipsychotics, and Broad-Spectrum Psychotropics.” Journal of Clinical Psychiatry 63 (suppl 4):42–55, 2002.
Exacerbations of preexisting hyperparathyroidism, as well as multiglandular disease, have both been implicated in the pathogenesis of lithium-induced hyperparathyroidism (Szalat et al. 2009). Cessation of lithium does not necessarily lead to normalization of parathyroid function. The calcimimetic drug cinacalcet is sometimes used to treat secondary hyperparathyroidism (via activating the CaSR, in turn downregulating PTH release), usually for instances in which parathyroidectomy is either not indicated or unsuccessful.
Lithium, carbamazepine, and quetiapine all may infrequently cause secondary hypothyroidism. Baseline measurement of thyroid function tests should include measurement of antithyroid antibodies when serum thyroid-stimulating hormone (TSH) is elevated. Repletion typically involves supplemental L-thyroxine (T4), usually begun at 0.025 mg/day, followed by reassessment of thyroid function tests after 6 weeks, with increases by 0.025 mg every 3–6 weeks until TSH levels normalize.
Subclinical hypothyroidism may manifest as a consequence of lithium use in 5%–35% of individuals receiving lithium, particularly women (Kraszewska et al. 2016), and usually within the first 6–18 months of treatment. The duration of exposure appears not to contribute to risk (Kraszewska et al. 2016). The condition is thought to result from several effects of lithium, including antagonism of TSH, reduced deiodination of peripheral T4 to triiodothyronine (T3), and interference with cyclic adenosine monophosphate–mediated production of thyroid hormone within the thyroid gland. More rarely, lithium may also cause hyperthyroidism due to thyroiditis or Graves’ disease (Lazarus 2009). Some studies suggest that lithium-associated hypothyroidism may be more common among individuals who have circulating thyroid antibodies. Typically, in patients with an elevated serum TSH level, the clinician measures antiperoxidase and antithyroglobulin to determine whether autoimmune thyroiditis is present. Note that in contrast to goiter, thyroid nodules are generally not thought to result from treatment with lithium.
No firm consensus exists on when supplemental thyroid hormone should be added to the regimen of patients with an elevated serum TSH level who are taking lithium. Some authorities advocate more frequent monitoring of serum TSH levels (e.g., every 3 months) in the setting of biochemical hypothyroidism, without adding exogenous thyroid hormone unless serum TSH levels exceed 10 mU/L or clinical manifestations emerge. Others advise a lower threshold for adding supplemental thyroid hormone whenever serum TSH levels rise above the upper limit of a laboratory’s reference range, particularly in the setting of affective symptoms. T4 is usually preferred over T3 to produce steadier hormone levels, even though T3 may exert more potent psychotropic (e.g., antidepressant) effects.
Supplementation of thyroid hormone for lithium-induced hypothyroidism usually begins with the addition of 0.025 mg/day of T4, followed by a reassessment of thyroid function tests after 6 weeks, and iterative increases of T4 by 0.025 mg/day until TSH levels have normalized. High-dose exogenous thyroid hormone—as is sometimes used to achieve suprametabolic levels of free T4 in rapid-cycling bipolar disorder—requires clinicians to recognize the potential for developing arrhythmias (notably, atrial fibrillation) and promoting bone demineralization and osteoporosis (sometimes prompting a need for obtaining periodic bone mineral densitometry assessments).
Carbamazepine also hastens the metabolism of T4 and T3 and may produce secondary hypothyroidism. In addition, a handful of case reports have described hypothyroidism induced by quetiapine, usually in patients with a history of past thyroid abnormalities (Kelly and Conley 2005), and potentially subject to spontaneous resolution (Kontaxakis et al. 2009). Proposed mechanisms include competitive metabolism of thyroid hormones and quetiapine by uridine diphosphate–glucuronosyltransferase (Kelly and Conley 2005) or possibly an autoimmune-mediated process, although the rarity of this phenomenon does not appear to warrant routine monitoring of thyroid function during quetiapine treatment.