Teratology: More than malformations

Myles et al., 2013 is the eighth meta-analysis to examine the question as to whether first trimester exposure to selective serotonin reuptake inhibitor (SSRI) antidepressants is associated with malformations. To date four have found no association (Addis and Koren, 2000; Einarson and Einarson, 2005; O’Brien et al., 2008; Rahimi et al., 2006) and two have found an association between paroxetine and cardiac malformations (Bar-Oz et al., 2007; Wurst et al., 2010). The two most recent, Myles et al., 2013 and Grigoriadis et al., 2013 have also found a significant association between paroxetine and cardiac malformations and in addition found an association between fluoxetine and malformations. This lack of consistent findings is puzzling for clinicians to interpret and anxiety-provoking for women who are prescribed antidepressants in pregnancy.

While rates of antidepressant use in pregnancy are relatively low in Australia, with a recent study showing 2.1% (Lewis et al., 2012), in North America the trend is towards increasing antidepressant usage in pregnancy, with rates documented between 5% and 13%. These rising rates of exposure to antidepressants in pregnancy have mirrored the substantial increase in publications and studies on the risks of exposure. However, many of these newer publications are based either on retrospective reviews of medical records or on retrospective studies of population registers, which frequently were not designed for research purposes. This has led to a number of studies having major limitations with crucial concerns such as accuracy of data on exposure, including timing and lack of adequate measures of maternal mental illness, and key confounding variables.

While Myles et al., 2013 and Grigoriadis et al., 2013 have both included all methodologies in their meta-analyses, earlier reviews have selected only prospective cohort studies. Interestingly, those that have selected only prospective studies have not found an association between malformations and antidepressants (Einarson and Einarson, 2005). This is one of the reasons there has been a consistent call for more prospective studies which include careful measures of exposure, maternal mental illness and confounding variables.

Given the general risk for malformations is between 2% and 3% and the principles of teratology require a specific malformation to be associated with a specific exposure, with commonly a dosage effect, untangling what may be caused by antidepressant exposure, what may be associated with other associated confounding variables and what may be explained by baseline population rates is challenging. For instance, several studies have now found the use of antidepressants in pregnancy is also associated with higher smoking and alcohol intake. In one study of malformations and antidepressant use the rate of fetal alcohol syndrome in SSRI-exposed neonates was 10 times higher than the control group (Malm et al., 2011). In studies of pregnant women taking antidepressant treatment there is yet to be an examination of other key variables associated with elevated malformation rates, such as obesity and nutritional status including prenatal folate use. In addition, maternal illnesses, such as diabetes, are defined also as teratogens, and it warrants further studies to clarify if maternal mental illnesses, particularly anxiety, are also potentially implicated. Given paroxetine is frequently prescribed for anxiety disorders and the evidence for adverse pregnancy outcomes and mental illness has been focused on anxiety in pregnancy, this clearly requires further exploration. With so many questions still to be answered there is an urgent need for well-designed, prospective studies that are adequately powered to untangle the complexity of exposures and outcomes for women and infants.

However, beyond the concern of structural teratology is the relatively unexplored territory of neurodevelopmental teratology. A systematic review in 2010 identified only 12 studies that have examined neurodevelopmental outcomes for children following antidepressant exposure; only half of these studies examined children aged 12 months or older and none at school age or beyond (Gentile and Galbally, 2010). The identified studies were methodologically heterogeneous with few using neuropsychological measures with predictive validity for effects on later child development. This contrasts with the anti-epileptic drugs where there are a number of high-quality prospective studies using comprehensive neuropsychological tests (Galbally et al., 2010). However, despite the limited spectrum of studies for antidepressant exposure and neurodevelopmental outcomes, there have now been four studies that have identified poorer motor development in children exposed to antidepressants in utero, including our own (Galbally et al., 2011). This clearly warrants further investigation with larger, prospective longitudinal studies with comprehensive neuropsychological assessment of children at an age and with measures that can accurately determine if this is a ‘real’ finding.

Systematic reviews and meta-analysis are useful in identifying potential risks, particularly for rare conditions or where studies available are mostly small and inconclusive. However, the current state of research into antidepressant exposure in pregnancy often results in meta-analysis leading to more questions than answers. There now needs to be a concerted effort to fill the knowledge gaps with purpose-designed, longitudinal, prospective studies which include measures of maternal illness, accurate records of exposure and a range of key confounding variables.

See Review by Myles et al., 2013, 47(11): 1002–1012.