Prenatal nicotine exposure and child behavioural problems

Nicotine metabolism and its transfer from mother to the foetus

Nicotine is a chemical compound, an alkaloid, found in tobacco smoke, smokeless tobacco products and nicotine replacement products [5]. It can be absorbed through the mouth by chewing, the lungs as smoke or the skin via an NRT patch [5]. Once nicotine has entered the bloodstream, it is distributed throughout the body to various tissues such as the brain, lung and liver [5]. In the liver, it is metabolised and finally excreted in urine [5]. Nicotine’s major metabolite is cotinine whose main metabolic product in turn is trans-3′-hydroxycotinine [5]. By crossing the placental barrier, nicotine can be transferred from the maternal circulation to the foetus [6]. Cotinine measurements taken in the first trimester indicate an accumulation in foetal fluids: cotinine concentrations in amniotic fluid and foetal serum were higher than in maternal serum regardless of whether the mother is an active or passive smoker [7]. The main pathway of elimination of nicotine from the foetal circulation occurs by re-diffusion across the placenta into the maternal circulation [8].

Assessment of prenatal exposure to nicotine

Prenatal exposure to nicotine can be assessed via questionnaires asking, for example, about maternal smoking or ETS exposure during pregnancy. This method is often chosen for epidemiological studies as past, long-term and current exposures can be inexpensively assessed in large samples [3, 9]. However, this method has several drawbacks. First, this method is prone to exposure misclassification such as recall bias, under-reporting or concealment of smoking, which might be due to awareness of numerous negative health effects for children exposed to tobacco smoke in utero [3, 9]. Second, a single question on whether the mother smoked during pregnancy does not adequately address the amount of exposure and smoking patterns: a “yes” answer might include total abstinence after becoming aware of the pregnancy. Therefore, a detailed set of questions on numbers of cigarettes smoked per day during pregnancy or during certain trimesters and on smoking cessation should preferably be used.

A more objective method of ascertaining prenatal exposure is the measurement of exposure-specific biomarkers in biological matrices of mother and/or child. Candidate biomarkers with high specificity for active or passive tobacco smoking or usage of medication containing nicotine are nicotine itself and its metabolites, such as cotinine or trans-3′-hydroxycotinine [10, 11]. While nicotine has a short half-life of 2 h, the half-life of cotinine is on average 16 h [10]. Thus, measurements of nicotine or cotinine in blood, urine and saliva reflect recent exposures within the last few hours or last few days, respectively, and are susceptible to variations in exposure. This limits the usage for the assessment of long-term exposures. In new-borns, however, a longer half-life of nicotine and similar half-life of cotinine compared to adults have been reported which have been suggested to be due to differences in the sensitivity of the clearance rates of these two chemicals to alterations of hepatic blood flow [5]. Maternal matrices such as blood, urine, saliva and hair can be used for detection of the above-mentioned biomarkers to assess ETS exposure or active smoking [10, 12]. Llaquet et al. [13] provide an overview on possible biological matrices in the child for detection of prenatal ETS exposure. Measurements of nicotine or its metabolites in cord serum and neonatal urine reflect the exposure to ETS shortly before delivery [13]. Other matrices such as hair, nails, amniotic fluid and meconium, in which nicotine or its metabolites accumulate or are incorporated during formation, have longer detection periods ranging from first to last trimester [13]. It is often not feasible to differentiate clearly between children of mothers exposed to ETS during pregnancy and those who were not. Two studies [14, 15] reported no detectable levels of nicotine or its metabolites in amniotic fluid or meconium of new-borns whose mother was exposed to ETS during pregnancy. In another study, the measurement of trans-3′-hydroxycotinine (adjusted for creatinine) in urine of new-borns showed slight but significant differences between children of mothers exposed to ETS and those unexposed [16]. For detection of foetal tobacco smoke exposure in late pregnancy, the measurement of cotinine in cord serum seems to be the most promising biomarker due to its ability to distinguish between active maternal smoking, maternal ETS exposure and no exposure [13].

Lastly, indoor nicotine concentrations can be measured in the homes of pregnant women using active or passive air samplers [17, 18]. This approach is mainly relevant for the assessment of ETS exposure.

ADHD and symptoms of hyperactivity or inattention

An association between gestational exposure to nicotine or tobacco smoke and ADHD in children has been reported from many studies; results were summarised in several reviews (e.g. [31, 40, 44, 45, 46]). Linnet et al. [45] reviewed 24 studies published between 1975 and 2002 investigating the relationship between prenatal maternal smoking and ADHD or ADHD symptoms in the children. The authors concluded that most studies reported an increased risk for the development of such problems in children of smoking mothers, some even showing a dose–response effect in the association. However, as there were several serious shortcomings such as methodological issues related to retrospectively collected data, rough estimation of exposure by a dichotomous smoke exposure variable and statistical issues related to power, no final statement on causality was possible [45].

Latimer et al. [44] reviewed prenatal or early post-natal environmental risk factors associated with disruptive behaviour disorders. Eleven studies investigated the role of maternal smoking during pregnancy: eight of them, including population-based and case–control studies of good quality, supported the presence of a link to an increased risk for ADHD in the offspring.

Furthermore, results from about 1,600 children of the German birth cohort study LISAplus also support an association between maternal smoking during pregnancy and hyperactivity or inattention problems in 10-year-olds [22].

In a population-based record linkage case–control study of young non-Aboriginal Australians (about 1,700 cases and 3,850 controls), Silva et al. [47] recently observed that maternal smoking is a risk factor for clinically defined ADHD with additional prescription of stimulant medication. The association remained significant for both sexes even after adjustment for several characteristics related to pregnancy and birth (boys: odds ratio (OR) = 1.86, 95 % confidence interval (CI): 1.53–2.27; girls: OR = 1.67, 95 %CI: 1.07–2.61).

Thus, both the consistency of results across many studies and different study designs and the presence of dose–response relationships between exposure and outcome in some studies support the hypothesis of a causal association. A further aspect is related to the ETS exposure of women who are non-smokers during pregnancy. This exposure can be due to ETS exposure at home by the partner or other household members or it can be an exposure at the workplace. Several studies have compared the effects of active maternal smoking during pregnancy with those resulting from ETS exposure (e.g. [20, 48, 49, 50]).

Gatzke-Kopp et al. [48] observed a higher risk for ADHD symptoms not only in children exposed to maternal smoking during pregnancy but also in those whose mother did not smoke but was exposed to ETS during gestation.

A very recent study by Keyes et al. [49] compared the influence of maternal and/or paternal smoking on offspring hyperactivity at the age of 10 years. In unadjusted analyses, maternal as well as paternal smoking during pregnancy was related to increased offspring hyperactivity, respectively. After adjustment for partner’s smoking behaviour and accounting for several covariates, the association between maternal smoking and hyperactivity in the children remained stable, but the association with paternal smoking was attenuated to non-significance. Furthermore, no increased risk for hyperactivity could be observed in children whose father smoked during pregnancy and whose mother did not smoke. Nomura et al. [50] conducted a similar study in about 200 preschool children (3–4 years old). They observed an increased risk of ADHD symptoms only for children exposed to maternal smoking but not for those exposed to paternal smoking, even after adjustment for a series of confounders including ADHD symptoms of the parents, thereby decreasing the chance for confounding by genetic factors.

Langley et al. [20] used data from over 8,000 children of the ALSPAC prospective birth cohort study. They compared the risks of ADHD symptoms in children aged 7.5 years whose mother smoked during pregnancy, with those whose mother did not smoke but was exposed to the smoking behaviour of the father. Furthermore, they assessed the effect of passive smoking in families in which neither parent smoked but where the mother reported ETS exposure at work or living with household members who smoked. Maternal smoking and paternal smoking (even in the absence of maternal smoking) were both observed to be associated with increased ADHD symptoms in the offspring, while passive smoking was not. The authors concluded that the associations between maternal smoking and ADHD in the children may be confounded by genetic factors or factors on the household level and are to a lesser extent attributable to causal effects of an exposure in utero [20].

Further studies that were able to control for genetic factors suggest that the association between maternal smoking during pregnancy and ADHD might not be causal (e.g. [51, 52]).

Thapar et al. [51] tested the association with maternal smoking in children conceived with assisted reproductive technologies, comparing the ADHD risk of children genetically related and unrelated to the gestational carrier. In genetically related mother–child pairs, maternal smoking during pregnancy was related to an increased risk for ADHD symptoms in the offspring, while no association was observed for genetically unrelated pairs. This observation suggests that the effect might be rather attributed to inherited characteristics than to the exposure to prenatal smoking [51].

D’Onofrio et al. [52] compared the ADHD traits of siblings with and without prenatal maternal smoke exposure in order to account for familial and genetic effects. When children whose mother smoked during pregnancy were compared to unrelated children without prenatal smoke exposure, they showed a significantly increased risk for ADHD symptoms. However, in siblings who differed in their exposure to maternal smoking during pregnancy, the association between smoking and subsequent ADHD symptoms was small and not significant.

Abbott et al. [31] provided an overview on findings from animal studies exploring whether prenatal exposure to tobacco smoke is associated with ADHD-like symptoms. Most studies used nicotine instead of tobacco smoke and hyperactivity measured by increased locomotor activity was usually chosen as indicator for ADHD symptoms. Studies in mice mostly reported increased locomotor activity after prenatal nicotine exposure, but studies in rats were less consistent [31]. Furthermore, the authors stated that increased activity might not be representative of ADHD-like behaviour in rodents, as other symptoms such as inattention are not considered [31]. Moreover, several methodological issues complicate the transfer of results from rodent studies to humans. Dwyer et al. [53] mention three caveats: first, human foetuses are born at a more mature stage of brain development than are rodents. The first two trimesters of human development correspond approximately to the full gestational development of rodents, and the early post-natal period of rodents is used as model for the third trimester development of human foetuses. Second, the effects of a continuous exposure to nicotine as effected in rodent models might be different from an intermittent exposure related to variations in nicotine levels such as it is the case for human smoking. Third, animal models with a nicotine exposure do not reflect the exposure to tobacco smoke in humans as tobacco smoke contains numerous other chemicals besides nicotine [53].

In summary, the majority of studies, and especially several recent epidemiological studies, observed a higher likelihood for ADHD or ADHD symptoms in subjects prenatally exposed to nicotine. However, both human and animal studies have failed to provide clear evidence on causality.

Conduct or externalising behaviours and antisocial behaviour

Several review articles (e.g. [54, 55]) summarised the association between prenatal exposure to nicotine and conduct or externalising problems. The authors concluded that the existing literature strongly indicates an increased risk, but no causal association could be established due to methodological limitations.

Further support for an increased risk comes from the study of Gatzke-Kopp et al. [48] who observed that not only active maternal smoking, but also non-smoking mother’s ETS exposure, during pregnancy is related to higher symptom scores for conduct disorder in the offspring.

Two other studies investigated the association between maternal smoking during pregnancy and increased risk for externalising problems in relatively young children at an age of 18 months [56] and 4 years [19]. Using data from the population-based Norwegian Mother and Child Cohort Study with a large study sample (N > 22,500), Stene-Larsen et al. [56] reported a significantly increased risk for externalising problems for 18-month-old children whose mother smoked more than 10 cigarettes per day (OR = 1.32, 95 % CI: 1.03–1.70) but not for those who smoked less. The authors additionally reported no sex difference in the association [56]. Brion et al. [19] analysed the association of maternal smoking during pregnancy with conduct or externalising problems in 4-year-olds from two birth cohorts, one from a middle-income country (Brazilian Pelotas study) and one from a high-income country (British ALSPAC study). A significant effect was (1) present in both studies, (2) persisted even after adjustment for confounders such as socioeconomic status and parental psychopathology and (3) was also robust to adjustment for paternal smoking during pregnancy.

Recently, O’Brien et al. [57] reported a gene x environment interaction for a dopamine transporter gene (DAT1) variant that modifies the risk for externalising problems in male but not in female adolescents after prenatal exposure to maternal smoking which was assessed via repeated cotinine-corrected reports.

However, studies using specific designs to control for genetic confounding came to inconsistent results (e.g. [52, 58, 59]).

D’Onofrio et al. [58] recently assessed the relationship between maternal smoking during pregnancy and antisocial behaviour in adolescents aged 14–17 years. In unrelated individuals, the results show a significantly increased risk for antisocial behaviour symptoms and for a criminal conviction. However, the associations became smaller and lost statistical significance when comparing siblings who differed in their exposure to prenatal maternal smoking. This supports the influence of familial factors on the association between prenatal exposure to smoking and later development of antisocial behaviour. A similar result was observed in an earlier study from D’Onofrio et al. [52] who also did not observe any elevated risk for conduct problems or oppositional defiant problems in children exposed to prenatal maternal smoking compared to unexposed siblings.

However, a very new study by Gaysina et al. [59] shows a different picture. Using data from three studies, it supports a direct causal effect of prenatal maternal smoking on later conduct problems in the offspring. Three different genetic constellations for mother–child pairs were present: either genetically related (1) or genetically unrelated with an adoption of the child at birth (2) or at conception (3). Thus, the authors were able to assess the association in mother–child pairs who differed not only with respect to tobacco smoke exposure during pregnancy but also with respect to their genetic relationship. Children who were exposed to maternal smoking in utero were observed to have a higher risk of conduct problems, regardless of whether the mother was genetically related or unrelated to the child. This supports an adverse effect of the exposure and not of genetic factors. This association was also observed after adjustment for several potential confounders among which were maternal education, parenting practices and socioeconomic characteristics of the family. Furthermore, results of a meta-analysis across pairs in the three studies supported this finding [59].

A possible explanation for differences between the findings from the study of Gaysina et al. [59] and those from D’Onofrio et al. [52, 58] is proposed by Gaysina et al.: the inability of the latter two studies to account for the influence of passive gene–environment correlations. Contrary to a gene–environment interaction which refers to a different susceptibility to an environmental factor due to a certain genotype, gene–environment correlations are described by a probability of exposure to an environmental factor that differs with the genotype [60]. A passive gene–environment correlation refers to the situation that children with a certain genotype (that is inherited from the parents) are more likely to experience a certain environmental exposure occurring during childhood [60]. Gaysina et al. [59] mentioned that they were able to control for post-natal passive genotype–environment correlations by testing the association in a subgroup of children adopted at birth who share the post-natal environment, but no prenatal environmental nor genetic factors with the mother. The association between prenatal smoke exposure and children’s conduct problems was also present in this group.

In conclusion, existing literature suggests a causal effect of prenatal exposure to nicotine and conduct or externalising problems in the offspring.

Depression, anxiety or internalising disorders

Results are generally mixed on the association between maternal smoking during pregnancy and internalising symptoms, such as depression or anxiety, in children. Findings are therefore less consistent than the findings for externalising symptoms such as conduct problems or ADHD [61]. While some studies are supportive of a relationship (e.g. [61, 62, 63, 64]), others are not (e.g. [19, 65]).

One longitudinal study by Ashford et al. [61] investigated the association between maternal smoking during pregnancy and symptoms of internalising behaviour in nearly 400 children, assessed at ages 5, 10–11 and 18 years. The authors observed significant relationships with both externalising and internalising behaviours that were also robust to adjustment for potential confounders and also for co-occurring internalising and externalising behaviours, respectively. The authors state that their study has the advantage of controlling for comorbid externalising problems, as the association between maternal smoking during pregnancy and externalising problems is well established and internalising and externalising problems are often comorbid.

A similar result was observed from a small study of 84 children conducted by Indredavik et al. [62]. Maternally reported internalising scores at the age of 14 years were significantly higher for children whose mother smoked during pregnancy, and this association remained also after adjustment for confounders including socioeconomic status and maternal mental health.

Menezes et al. [63] recently reported results from 18-year-olds from the Pelotas cohort, showing a higher risk for lower levels of happiness and increased rates of depression among those prenatally exposed to maternal smoking (<20 cigarettes/day: OR = 1.38, 95 % CI: 1.03–1.84; ≥20 cigarettes/day: OR = 2.11, 95 % CI: 1.31–3.40). Smoking by the mother’s partner during pregnancy was associated with decreased adolescent happiness after adjustment for confounders, but did not show an association with offspring depression.

Ekblad et al. [64] studied the relationship between maternal smoking during pregnancy and psychiatric morbidity in young Finnish adults in a large (N > 175,000) population-based sample using registry-based data. The authors observed an increased risk for any psychiatric diagnosis. Dose–response relationships were observed for the risks of mood disorders, behavioural and emotional disorders occurring in childhood and adolescence, as well as disorders of conduct and emotion. However, while the study was able to adjust for a potential influence of maternal psychiatric morbidity on the relationship, information on other important factors such as socioeconomic factors, maternal alcohol consumption during pregnancy and post-natal exposure to tobacco smoke was not available. Therefore, the reported association should be interpreted with caution.

Other studies, however, do not observe a relationship between prenatal maternal smoking and internalising problems. Höök et al. [65] observed no such association in preschool children at 3 or at 5.5 years of age. A similar null finding was observed by Brion et al. [19] who studied the association in the British ALSPAC and the Brazilian Pelotas study. Another finding of this study was that paternal smoking during pregnancy was also unrelated to offspring internalising problems [19].

A small number of studies in rats on internalising behaviours after prenatal nicotine exposure reported consistent results of an increased anxiety-like behaviour in adolescent and also in adult rats [31].

In summary, the establishment of a final conclusion concerning the relationship between prenatal nicotine exposure and internalising problems in the offspring is complicated by insufficient data and mixed results in epidemiological studies.