Neonatal hypoglycaemia following maternal olanzapine therapy during pregnancy: a case report

Introduction

Information regarding the safety of olanzapine and other antipsychotics during pregnancy is limited. Due to the ethical difficulties in conducting randomized controlled trials in pregnant women, the best available data are from nonrandomized observational studies and case reports [Howard et al. 2004]. We present a case of a woman with bipolar disorder taking olanzapine during pregnancy, whose baby developed neonatal refractory hypoglycaemia due to hyperinsulinism within a few hours of birth.

Case

A 24-year-old African woman presented with an unplanned pregnancy and a 4-year history of bipolar disorder type 1, including four hospital admissions for severe relapses of rapid onset. She was also a regular cannabis and alcohol user.

Although she had been advised about the risks of conceiving whilst taking sodium valproate (1000 mg/day), she unintentionally became pregnant. She was also taking olanzapine 10 mg/day. The valproate was stopped and olanzapine increased (15 mg/day). At 16-weeks gestation, following a period of medication noncompliance, the patient developed an acute manic illness. She was irritable, with pressured speech, and grandiose and paranoid delusionals. She was admitted, prescribed promethazine 25 mg four times daily and diazepam 5 mg three times daily as needed (for 8 weeks) and olanzapine increased to 20 mg/day. At 19+5 gestation she was commenced on lithium 400 mg twice daily. Compliance was assured by supervised dosing and her mental state gradually improved. She was discharged at 36+2 weeks of gestation on olanzapine 20 mg/day and lithium 400 mg twice daily (see Figure 1). She continued to smoke cigarettes throughout the pregnancy.

Figure 1.

Timeline of medication taken by the mother throughout the pregnancy.

Investigations

A 20-week ultrasound scan demonstrated a small placenta and foetal in-utero growth restriction (IUGR). The patient often refused blood tests throughout her admission; however, lithium levels obtained were within the therapeutic range. Random blood glucose was 3.3 mmol/l at 28 weeks; urinalysis remained normal throughout pregnancy. Other investigations were normal including umbilical artery Doppler and foetal echocardiography scan at 37 weeks. Body mass index (BMI) was not recorded throughout or before the pregnancy, however, the woman was noted to be slim before and during the pregnancy.

A male infant was delivered via caesarean section at 39+4 gestation following a suboptimal cardiotocograph. Lithium was discontinued during labour (36 h). The patient did not breastfeed.

The infant was in good condition at birth with Apgar scores of 8 (1 min) and 9 (5 min). He was small for gestational age (SGA) (birth weight 2.69 kg, 0.4th centile). At 2 h, he was grunting with laboured breathing, admitted to the neonatal unit and found to have a metabolic acidosis (pH 6.9, lactate 8.9 mmol/l; normal: < 2.0) and hypoglycaemia (blood glucose < 0.6 mmol/l; normal: 2.7–5.4 mmol/l).

A hypoglycaemia screen demonstrated hyperinsulinaemia (insulin 15.5 mlU/l) despite blood glucose 0.7 mmol/l. Normal investigations included C-peptide, serum cortisol, growth hormone, serum free fatty acid, 3-hydroxybutyrate and urine organic acids. Urinary ketones were negative. The low glucose with increased lactate and virtually absent lipolytic and ketogenic response with increased glucose utilization were all consistent with hyperinsulinism. There was no evidence of genetic causes, sepsis, asphyxia or hypothermia.

The infant was treated with 10% dextrose boluses and a dextrose infusion. The highest dextrose infusion rate needed to maintain normoglycaemia was 16 mg/kg/min (day 3). Hyperinsulinaemia is considered highly likely if a neonate needs > 12 mg/kg/min dextrose infusion to maintain normoglycaemia. Initial attempts to reduce dextrose infusion by establishing milk feeds were unsuccessful, so the baby was prescribed oral diazoxide and chlorthiazide. (Diazoxide acts by inhibiting the secretion of insulin by the beta cells of the pancreas, and chlorthiazide counteracts the fluid retention caused by diazoxide.) The infant responded well, allowing the gradual reduction of dextrose and establishment of milk feeds by day 15. Magnetic resonance imaging brain scan and brain stem auditory evoked potentials at 3 weeks were normal. He was discharged on day 35.

On discharge the infant required thrice daily blood glucose measurements, fluid restriction, chlorthiazide and diazoxide treatment, with an emergency plan should blood glucose levels fall below 3.0 mmol/l, including the application of ‘hypostop’ gel on the gums and immediate referral to the emergency department. He was reviewed by a paediatrician at 3 and 5 months and was making appropriate developmental progress with no neurological sequelae of the profound hypoglycaemia. The infant’s medications were stopped at 23 weeks.

Comments

To the best of the authors’ knowledge, this is the first case report of maternal olanzapine therapy being associated with neonatal hypoglycaemia due to hyperinsulinism. Hyperinsulinism is the most common cause of persistent hypoglycaemia in neonates [Stanley and Baker, 1999]. As a consequence of the dominant effects of insulin, neonates with hyperinsulinism are especially susceptible to the adverse complications of hypoglycaemia. This is because of their inability to generate alternative fuels, such as ketones and up to 20% of infants with hyperinsulinism have associated neurological problems.

Nongenetic causes of neonatal hyperinsulinaemia are transient and include prematurity, IUGR, maternal diabetes, sepsis, asphyxia, hypothermia, erythroblastosis foetalis, exposure to beta-agonist tocolytics and Beckwith–Wiedemann syndrome [Stanley and Baker, 1999]. Although this infant was SGA (probably related to smoking and/or cannabis), the hyperinsulinism associated with this condition is usually relatively transient [Stanley and Baker, 1999]. Moreover, the association between IUGR and neonatal hypoglycaemia is usually the result of reduced glycogen and adipose stores rather than hyperinsulinaemia. Sodium valproate exposure in pregnancy has been associated with newborn hypoglycaemia, but this seems unlikely to have been the cause here as it was stopped at 7 weeks gestation [Barrett and Richens, 2003; Coban et al. 2010]. Clinically, lithium does not have a major influence on glucose homeostasis or increase insulin levels [McIntyre et al. 2011]. There have been no reports of short-term maternal benzodiazepine treatment and neonatal hypoglycaemia.

It is possible that undiagnosed gestational diabetes was present here. An oral glucose tolerance test was not carried out antenatally as close monitoring for gestational diabetes in mothers taking olanzapine during pregnancy is not recommended in the National Institute for Health and Clinical Excellence guidelines. Urinalysis was normal throughout pregnancy, but a degree of subclinical impaired glucose tolerance is possible, as olanzapine has been shown to increase basal insulin levels. Additionally, the mother was of African origin, which further increases her risk of gestational diabetes. Therefore, olanzapine seems a plausible cause of this baby’s hypoglycaemia, either through direct action on the infant’s basal insulin levels, or via undiagnosed maternal gestational diabetes.

Olanzapine has an in vivo placental passage ratio of 72.2% [Newport et al. 2007], and several case reports on its use have described uneventful pregnancies and healthy infants. The largest study to date found that olanzapine did not increase the risk of major congenital malformations, but was associated with a higher maternal BMI, maternal gestational diabetes and low birth weight [Reis and Kallen, 2008]. Another prospective study also reported a tendency towards low birth weight and neonatal intensive care admission [McKenna et al. 2005]. In contrast, others have linked maternal olanzapine with a higher incidence of large for gestational age (LGA) infants and higher mean birth weight [Babu et al. 2010; MacRitchie et al. 2006; Newham et al. 2008]. Predisposing factors for LGA infants include maternal obesity, type 1 diabetes mellitus, gestational diabetes mellitus and maternal weight gain, all conditions that have been shown to be exacerbated or precipitated by some antipsychotics, including olanzapine.

However, a large recent linkage study has found that women taking antipsychotic medication during pregnancy have an increased risk of gestational diabetes and a higher incidence of SGA infants [Boden et al. 2012]; olanzapine and clozapine therapy were not associated with a higher incidence of gestational diabetes compared with other antipsychotics. Also of note was that the high incidence of SGA infants born to women on antipsychotics was explained by confounders such as smoking, which is a possible cause in this case.

The mechanism for an olanzapine-induced metabolic syndrome is not well understood. However, the insulin resistance induced by olanzapine treatment is rapid, occurring within days, separately from weight gain [Ebenbichler et al. 2003]. Potential causative mechanisms may involve free fatty acids, leptin and tumour necrosis factor α [Kahn and Flier, 2000]. Studies have also shown that weight gain is due to the accumulation of white adipose tissue, and that low-grade adipose inflammation may play a role in this [Victoriano et al. 2010].

Conclusion

Olanzapine exposure during pregnancy was associated with neonatal hyperinsulinaemia in the absence of proven gestational diabetes. Given the mechanisms discussed above, olanzapine is a potential causative agent either acting directly on the infant’s own glucose metabolism or indirectly via the mother. It seems prudent to suggest that pregnant women treated with olanzapine should be offered an oral glucose tolerance test, particularly if they have a raised BMI: indeed, in light of recent research finding an increased risk of gestational diabetes in women prescribed antipsychotics in pregnancy [Boden et al. 2012], we would suggest all women prescribed antipsychotics with raised prepregnancy/first trimester BMI should be offered this test. In addition, it would be potentially helpful to consider the creation of an international register to monitor routinely the maternal and foetal outcomes of antipsychotic use in pregnancy [Kulkarni et al. 2008].