Lithium-induced nephrotoxicity is a form of chronic tubulointerstitial nephropathy known since lithium was introduced in the treatment of mood disorders and reported from the mid-1970s (Lindop and Padfield 1975). Renal complications associated with lithium treatment include (a) impairment of tubular function and urinary concentrating ability with polyuria and polydipsia and development of nephrogenic diabetes insipidus that may become irreversible in 15 % of patients after long-term lithium exposure and renal tubular acidosis, (b) chronic kidney disease secondary to tubulointerstitial nephritis and (c) infrequent and relatively mild renal insufficiency (Vestergaard and Schou 1981; Boton et al. 1987).

Early lithium reports established that lithium-induced nephropathy was usually characterised by tubulointerstitial nephritis with minimal glomerular involvement (Hansen 1981) and rare significant changes in glomerular filtration rate (GFR), even in presence of a long-term lithium usage (Jensen and Rickers 1984; DePaulo et al. 1986; Schou and Vestergaard 1988). More recently, several reports have clarified that a very high percentage of patients treated chronically with lithium have low GFR and that GFR monitoring is frequently neglected, with a risk of progression to end-stage renal disease (ESRD) (Bassilios et al. 2008; Janowsky et al. 2009). ESRD was considered an unlikely event in patients taking lithium, and only three case reports of lithium-induced ESRD were known in the world literature up to the early nineties (Von Knorring et al. 1990; Gitlin 1993). Since then, pharmacovigilance reports have uncovered adverse events more accurately, and from 2004 to present, data recorded by FDA, despite small differences related to age and sex, detect rates around 0.1 % for the most severe adverse events such as renal impairment (N = 34), acute renal failure (N = 132), increased blood creatinine (N = 91), renal failure (N = 60) and haemodialysis (N = 68).

The presence of renal damage induced by long-term lithium treatment in chronically treated patients was confirmed by Markowitz et al. (2000). Lithium-induced chronic renal disease is slowly progressive, and its rate of progression is related to the duration of lithium administration. Regular monitoring of estimated creatinine clearance is mandatory in long-term lithium-treated patients. A survey of lithium-induced ESRD conducted in France adds further information on this topic, demonstrating that lithium-related ESRD represents only 0.22 % of all causes of ESRD in France and that the rate of progression is related to the duration of lithium administration (Presne et al. 2003). Lepkifker et al. (2004) reported similar findings in a retrospective study showing that in long-term lithium therapy, dose reduction or discontinuation of lithium resulted in stabilisation of plasma creatinine levels and that about 20 % of long-term lithium developed renal insufficiency.

These results are slightly different from a large nationally representative sample, such as the Third National Health and Nutrition Examination Survey, estimating the prevalence and distribution of chronic kidney disease in the United States. The results of this survey show that the risk of ESRD, in lithium-treated patients, might be increased compared with healthy controls, but the absolute risk seems to be relatively low (0.53 % compared to 0.2 % of the general population) (Coresh et al. 2003). In contrast with these results, Bendz et al. (2010), reviewing the data of The Swedish Registry for Active Treatment of Uremia in two Swedish regions (2.7 million inhabitants), observed a substantially higher prevalence of lithium-induced ESRD in patients on renal replacement therapy.

McKnight et al. (2012), in a recent meta-analysis, clarified that GFR impairment secondary to lithium treatment is not clinically significant in most patients with a reduction ranging from 0 to 5 mL/min that represents only 5 % of the minimum normal GFR.

At the moment, two elements, strictly related, seem to be very important to prevent irreversible renal damage and glomerular failure in patients treated with lithium: the duration of lithium treatment and the age of treated patients (Bocchetta et al. 2013).

Pharmacovigilance reports emphasise that renal function must be carefully assessed in every patient starting lithium treatment. To minimise the risk of adverse events during treatment, assessment of renal function at least twice a year is mandatory. This assessment must provide (1) complete 24-h urine collection; (2) glomerular filtration rate either by 24-h creatinine clearance or estimated glomerular filtration rate; (3) in case of a chronic kidney disease, a nephrologist should be consulted before lithium treatment as long as the creatinine clearance is >40 mL/min and (4) the decision should be taken also on the basis of the patient’s age and the duration of lithium treatment.

Long-term treatment with lithium results in its accumulation in the thyroid with many effects on the physiology of the gland. The pathogenetic mechanism of lithium-induced hypothyroidism is manifold; lithium acts through five different mechanisms: (1) inhibition of thyroidal iodine uptake, (2) inhibition of iodotyrosine coupling, (3) changes of the thyroglobulin structure, (4) inhibition of thyroid hormone (thyroxine) secretion and (5) increase of TSH levels as a result of reduced availability of thyroxine (Berens et al. 1970; Burrow et al. 1971). Inhibition of thyroid hormone secretion associated with high rates of hypothyroidism and thyrotoxicosis is the result of these effects on the gland function (Bocchetta et al. 2001; Barclay et al. 1994).

Adverse events submitted to the FDA from 2004 to present have revealed 54 cases of hypothyroidism in adult patients treated with lithium and, although minor differences related to age and sex, rates are around 0.29 % of the total FDA reports.

Hypothyroidism is the most common thyroid disorder caused by lithium treatment. It is very difficult to evaluate its prevalence because records range from 3.3 to 35.4 %. Eight case-control studies reveal a prevalence of clinical and subclinical hypothyroidism of 9.2 % in patients treated with lithium compared with a prevalence in the general population comprised between 0.5 and 1 %. The risk of hypothyroidism increases about six times in patients taking lithium (OR = 5.78) (McKnight et al. 2012). The main risk factors for the onset of hypothyroidism in lithium-treated patients are female gender, age between 40 and 60 years, a personal or family history of thyroid disorders and the presence of autoantibodies (Malhi et al. 2012). The risk amongst women is greater than in men and in the general population and is related to the age of patients and the duration of treatment (Grandjean and Aubry 2009). However, it is important to stress that, irrespective of lithium treatment, patients with mood disorders show higher rates of thyroid abnormalities (hypothyroidism and hyperthyroidism) than the general population (Chakrabarti 2011).

Goitre is a clinical finding associated with lithium therapy, probably linked to the inhibition of thyroid hormone synthesis and release, determining an increase of TSH and a final thyroid enlargement. The prevalence of goitre is highly variable, ranging from 3.6 to 51 %. This variability is probably due to the presence of various geographic risk factors, specifically the reduced availability of iodine, the different duration of the exposure to lithium and different diagnostic methods (Kibirige et al. 2013). The presence of hypothyroidism or goitre is not a contraindication to lithium treatment; therefore, patients successfully treated should continue treatment, even in the presence of hypothyroidism, and compensate with a hormone replacement treatment.

Case reports of hyperthyroidism and thyrotoxicosis have been described in the literature since the 1970s (Rosser 1976). More recent studies clarified that it is a rare condition whose incidence is comparable to that of the general population (Vanderpump et al. 1995). Thyrotoxicosis occurs in the early stages of treatment and at a young age, especially in women.

Hyperparathormonemia and hypercalcaemia are unrecognised and underappreciated adverse effects of lithium treatment despite the evidence that in healthy volunteers a single dose of lithium (600 mg) is enough to induce a transient, and statistically significant, rise in the serum PTH (Seely et al. 1989) and a prevalence ranging from 6.3 to 50 % (Livingstone and Rampes 2006).

Health Canada has reviewed the available evidence and scientific literature concerning the association between lithium treatment and hypercalcaemia associated with hyperparathyroidism. Following this evaluation, Health Canada advised health professionals about the risk of hypercalcaemia/hyperparathyroidism associated with lithium treatment and the need of considering calcium and parathormone blood levels before starting lithium treatment and of repeating this evaluation every 6 months in order to reduce the risk of hypercalcaemia (www.​healthycanadians​.​gc.​ca).

Adverse events submitted to the FDA from 2004 to present have revealed 29 cases of hypothyroidism in adult patients treated with lithium, and, despite minor differences related to age and sex, rates are around 0.24 % of the total FDA reports (Table 11.1).

Saunders et al. (2009) reviewed the effect of lithium exposure on parathyroid cell function confirming that lithium causes hypercalcaemia and isolated hyperparathormonemia and the need of a screening of patients on chronic lithium therapy for hypercalcaemia. More recently, a meta-analysis identified 60 studies (14 case-control studies, 36 case reports, 6 cross-sectional studies) and a 10 % increase of calcium and parathormone (PTH) levels in patients treated with lithium. This adverse effect remains unrecognised by most psychiatrists with an underestimation of the long-term effects of hypercalcaemia and hyperparathyroidism (McKnight et al. 2012).

This evidence indicates that the assessment of serum calcium before and during lithium treatment is mandatory. Currently, only the clinical guideline of The International Society for Bipolar Disorders (ISBD) suggests this assessment before starting lithium treatment and, in the absence of clinical elements suggestive of impaired parathyroid function, a re-evaluation after 6 months and then annually (Pacchiarotti et al. 2013).

Potential skin changes associated with lithium treatment are generally limited, and the evidence supporting this association is restricted to descriptions of several case reports, retrospective studies and few case-control studies. There are only two recent randomised controlled trials comparing lithium with lamotrigine and placebo for 18 months. The combined analysis of the results of these two studies showed no significant difference in the prevalence of cutaneous adverse effects between patients given lithium and those given placebo (Goodwin et al. 2004). It is therefore not clear whether lithium exposes them to a greater risk of developing cutaneous adverse effects in general. This statement is confirmed by the difficulty in assessing the real prevalence rates of this adverse effect. The few controlled studies on prevalence seem to suggest an increase of these adverse effects in patients treated with lithium, compared to the general population, ranging from 13.6 to 34 % in the study by Sarantidis et al. (1983) and from 25 to 45 % in the study by Chan et al. (2000).

The situation is slightly different for psoriasis since the presence of this skin disorder in lithium-treated patients is reported in several cases as de novo onset of psoriasis or as a marked worsening of a previously diagnosed disease (actually, reports are greater for acute exacerbation of a known disease).

There are only three case-control studies evaluating the prevalence of psoriasis in patients treated with lithium. The above-mentioned studies found a prevalence of 2.2 % in patients compared to 0 % in controls (Sarantidis and Waters 1983) and 6 % in patients compared to 0 % in controls (Chan et al. 2000). The third one is a large-scale epidemiological case-control study that found a small but significant increase in the risk of psoriasis in lithium-treated patients (Brauchli et al. 2009).

The greater part of pharmacovigilance reports on lithium treatment-associated weight gain has been published in the 1970s and 1980s (Rockwell et al. 1983), and present reports are often the continuation of data well established in the scientific literature of recent years. Reports on adverse events submitted to the FDA from 2004 to present have revealed 93 cases of weight gain representing, despite minor differences related to age and sex, the 0.66 % of the total FDA reports.

Bipolar patients display a persistent cognitive impairment in the different stages of the disease including euthymia. This impairment, evident in a range of neuropsychological tests including memory and executive functioning (Frangou et al. 2005), is present in unaffected first-degree relatives and is thought to be related to illness severity, specific neurodevelopmental features, medical co-morbidity and patient’s lifestyle (Bourne et al. 2013).

Data in this area of research are highly controversial mainly due to the presence of numerous methodological problems which reduce their reliability. Lithium treatment is associated with mild impairment in psychomotor speed, verbal memory and psychomotor speed functioning without a clear positive effect on cognition (Pachet and Wisniewski 2003). In a recent study, Wingo et al. (2009) reviewed the results of 12 studies involving 276 lithium-treated bipolar patients. Long-term lithium treatment was associated with impairment in immediate verbal learning and memory and creative psychomotor performance with no evidence of cognitive improvements.

Concerns about lithium adverse effects in older patients have led to both declining rates of lithium use and questions regarding the most useful approach to the use of lithium in this population.

Despite the lack of randomised placebo-controlled trials, it is assumed that lithium is as effective in the elderly as in the younger population for prophylaxis of affective disorders and for resistant unipolar depression (Bech 2006), but several concerns about neurotoxicity have led to questions about the effectiveness and safety of lithium in older bipolar patients.

The decrease in total body water and the decline of glomerular filtration rate are common amongst older persons, and in bipolar patients treated with lithium, this can result in a decrease in lithium clearance and increased serum level (Slater et al. 1984; Sproule et al. 2000). In a 2-year study of unipolar and bipolar out-patients (21–78 years) on long-term lithium treatment, Murray et al. (1983) found polydipsia/polyuria in 44 % of patients and hand tremor in 29 % of them. The prevalence and severity of the tremor tended to increase with age, but polydipsia/polyuria didn’t. More recently, van Melick et al. (2013) evaluated 759 patients aged 40 or older and treated with lithium with at least 2 years follow-up in a retrospective study and assumed that age was not a determinant of serum lithium concentration instability and, above all, was not a reason not to initiate or to discontinue lithium therapy.

Lithium is a drug that continues to have a critical role in the treatment of bipolar disorder in the elderly. It becomes clear that, in the elderly more than in the young bipolar patients, lithium requires a careful evaluation in order to prevent adverse effects or toxicity. Lithium treatment in the elderly may be appropriate only after a clear evaluation of benefits and risks in each individual patient, and if the patient is monitored correctly, it is possible to avoid commonly reported adverse effects that can occur even at therapeutic dosages.

Guidelines for lithium concentrations in geriatric bipolar population are based on limited evidence, and a recent study recommends a low concentration range (0.5–0.6 mmol/L) for patients of 50 years and over (Wijeratne and Draper 2011).

A report by Tica et al. (2013) highlighted the possibility that phenobarbital (PH)/carbamazepine (CBZ) therapy during foetal organogenesis could induce sirenomelia by a synergistic teratogenic effect and supported the recommendation to use only one drug in pregnant epileptic or bipolar women. At birth, the newborn, delivered by an epileptic woman after 37 weeks of gestation, weighed 2.2 kg and presented with sirenomelia type II, with some of its “classic” features: oligohydramnios and absence of kidneys, bladder, rectum, uterus and a single umbilical artery. Some other “particularities” included the absence of Potter’s face and no significant cardio-pulmonary abnormalities. The authors postulated that combined therapy with PH and CBZ (both strong enzyme inductors, especially PH) had potentiated their teratogenicity, by producing supplementary quantities of epoxides and/or other oxides, which accumulated in the foetal tissues who received in the first 4 months of pregnancy PH (0.1 g/day) and CBZ (0.4 g/day), followed only by PH 0.1 g/day, until delivery.

Another report presented an infant born with renal and cardiac malformations who developed a withdrawal syndrome and hyponatraemia following in utero exposure to oxcarbazepine. The infant was born at 35 weeks’ gestation by urgent caesarean section to a mother in status epilepticus who had been treated with oxcarbazepine throughout her pregnancy. Evaluation for congenital anomalies identified mild aortic stenosis, a bicuspid aortic valve, patent foramen ovale, patent ductus arteriosus and severe left hydronephrosis due to left ureteropelvic junction stenosis. On the third day of life, the infant developed clinical signs of a withdrawal syndrome, which peaked on day 7 and resolved by day 12. Transient hyponatraemia resolved by day 8 of life. Follow-up showed normal development at 15 months (Rolnitsky et al. 2013).

The prevalence of neurodevelopmental disorders in children prenatally exposed to antiepileptic drugs was studied in a prospective cohort of women with epilepsy and a control group of women without epilepsy. The children of this cohort were followed longitudinally until 6 years of age (N = 415), and the analysis revealed an increase in risk for children exposed to monotherapy sodium valproate and in those exposed to polytherapy with sodium valproate compared to control children (4/214; 1.87 %). Autistic spectrum disorder was the most frequent diagnosis. No significant increase was found amongst children exposed to carbamazepine (1/50) or lamotrigine (2/30) (Bromley et al. 2013).

Valproate is associated with polycystic ovary syndrome as well as congenital malformations and developmental delays of infants who were prenatally exposed. In a study by Wisner et al. (2011), using New York State Medicaid Claims for Persons with Psychiatric Disorders, the authors concluded that over 20 % of childbearing-aged women receiving mood stabilisers were treated with valproate.

In women, according to FDA-reported side effects, the use of valproic acid during pregnancy led to spontaneous abortion in a different percentage, depending on the age of patients (0.1 %, 0.02 % and 0.04 %, respectively, for women’s age of 10–19, 20–29 and 30–39 years). Intrauterine deaths were reported, instead, only in few cases (0.03 %). Carbamazepine and oxcarbazepine use during pregnancy was associated to spontaneous abortion only in 0.12 % and 0.16 %, respectively, of all treated patients. Lamotrigine, on the other hand, was reported to FDA to induce spontaneous abortion in 0.5 % of cases, with a peak of 0.12 % in women ageing from 20 to 29 years.

Cutaneous adverse drug reactions (CADRs) are the most prevalent ADRs in hospitalised children, with an estimated rate of 2–3 % (Ross et al. 2007). An analysis of reports from the Canadian Pharmacogenomics Network for Drug Safety (CPNDS) included 326 CADR cases of which 214 (65.6 %) were severe and 112 (34.4 %) non-severe. Overall, carbamazepine (N = 17, 4.9 %) and lamotrigine (N = 13, 3.7 %) accounted for almost 9 % of all suspected medications (Castro-Pastrana et al. 2011).

A case-control study of severe cutaneous drug reactions (SCDR) with carbamazepine was reported by Chong et al. (2014). In recruited patients, HLA-B*1502 positivity increased the odds of carbamazepine-induced SCDR in children of Chinese and Malay ethnicity and they occurred within 2 weeks and at low doses. Stevens-Johnson Syndrome (SJS), on the other hand, was induced by oxcarbazepine, and HLA genotyping showed a HLA-B15 variant in this patient (HLA-B*1518/B*4001) (Lin et al. 2009). A 4-year-old girl with a life-threatening clinical course of drug rash with eosinophilia and systemic symptoms syndrome (DRESS) with massive pulmonary involvement was also reported by Irga et al. (2013).

The use of combined antiepileptic drugs can cause toxicity by affecting the clearance of the drugs, especially in children. A case with SJS triggered by the combination of clobazam, lamotrigine and valproic acid treatment was reported in a 4-year-old boy admitted to the hospital with a 3-day history of fever, oral mucosa ulcerations and skin lesions. The patient had been under the treatment of valproic acid (900 mg/day) for 3 years with the diagnosis of epilepsy. Because of the poor control of the seizures, lamotrigine (75 mg/day) had been added to the treatment 1 month before and clobazam (20 mg/day) 10 days before. Weintraub et al. (2005) reported that valproic acid decreased lamotrigine clearance by approximately 60 % in a study with 570 patients. It has been reported that valproic acid may interfere with the metabolism of lamotrigine by inhibiting glucuronides, leading to increased blood levels of the drug, or resulting in accumulation of toxic metabolites of the drug. New AEDs (clobazam, etc.) are reported not to affect the clearance of lamotrigine significantly, but the skin lesions of SJS in the above patient appeared when clobazam was added to valproic acid and lamotrigine treatment. Therefore, it could be concluded that clobazam can affect the clearance of combined drugs. Adverse reactions have been reported in children co-treated with lamotrigine and valproate (Kocak et al. 2007; Levi et al. 2009). Sixty-three cases of interaction between lamotrigine and other drugs have been reported to FDA from 2004 to present.

In 2011, the US FDA informed the public that children born to mothers who take the anti-seizure medication sodium valproate or related products (valproic acid and divalproex sodium) during pregnancy have an increased risk of lower cognitive test scores than children exposed to other anti-seizure medications during pregnancy. This conclusion was based on the results of epidemiologic studies. The largest of these studies is a prospective cohort study conducted in the United States and United Kingdom (Meador et al. 2009). They found that children with prenatal exposure to valproate throughout pregnancy had lower Differential Ability Scale (DAS) scores at age 3 than children with prenatal exposure to the other evaluated antiepileptic drug monotherapy treatments: lamotrigine, carbamazepine and phenytoin. Although all of the available studies have methodological limitations, the weight of the evidence supports the conclusion that valproate exposure in utero causes subsequent adverse effects on cognitive development in offspring (Gaily et al. 2004; Adab et al. 2004).

The FDA reported that for AEDs, the most common side effects in children and adolescents (0–19 years of age) were represented by the following: