واکنش پذیری استرس های بیولوژیکی مغز و اعصاب و رفتاری در کودکان در معرض دخانیات قبل از تولد

Summary This study examined neurobiological and behavioral stress reactivity in children who had been prenatally exposed to tobacco. Neurobiological stress reactivity was measured using salivary cortisol and alpha-amylase levels at five different time points throughout a stressful neuropsychological test session, which involved a competition against a videotaped opponent. Participants (mean age: 10.6 years, SD 1.3) were 14 prenatally exposed (PE) children, 9 children with disruptive behavior problems (DBD), and 15 normal controls (NC). For cortisol responses, no significant differences between the three groups were observed. Normal controls, however, had significantly higher alpha-amylase levels than PE-children throughout the test session, and their alpha-amylase levels also increased throughout the session, whereas these remained low and stable for PE-children. Alpha-amylase levels and trajectory of PE-children were similar to those observed for DBD-children. PE-children also showed significantly increased behavioral stress reactivity compared to NC-children, and neurobiological and behavioral stress reactivity were inversely related in PE-children, again similar to what was observed for DBD-children. These results support the hypothesis that prenatal smoking may lead to long-lasting neurobiological and behavioral changes in exposed offspring.

1. Introduction Prenatal tobacco exposure has been related, in both and animal and human studies, to structural and neurobiological changes in the offspring brain. These changes include cortical thinning (e.g., of the orbitofrontal cortex) (Toro et al., 2008), smaller volume of specific brain structures (e.g., the corpus callosum (Paus et al., 2008), disruptions of white matter microstructure (Jacobsen et al., 2007), reduction of binding sites for serotonin transporter (Xu et al., 2001), and both increases and decreases (in specific brain regions) in serotonin and acetylcholine receptor density (Falk et al., 2005 and Slotkin et al., 2006). Stimulation of nicotinic acetylcholine receptors (nAChRs) triggers catecholamine (epinephrine and norepinephrine) secretion, so prenatal tobacco exposure is likely to have further (indirect) effects on neurotransmitter availability (Oncken et al., 2003). Both structural damage and in- and decreases in receptor density are likely to have functional consequences. Whereas structural damage and reductions in receptor density may be relatively easy to associate with systematic or functional down-regulation, an important hypothesis in the context of prenatal tobacco exposure is that increases in numbers of receptors, specifically nAChRs, will also lead to systematic or functional down-regulation (of both cholinergic and catecholaminergic activity), as the nAChRs will be under-stimulated after conception because the direct nicotine exposure experienced prenatally has stopped (Oncken et al., 2003). Acetylcholine and norepinephrine are the two most important neurotransmitters for functioning of the hypothalamic–pituitary–adrenal (HPA-) axis and the Sympathetic Nervous System (SNS), which are both involved in stress regulation. The SNS is part of the autonomic nervous system and is involved in quick (fight-or-flight type) reactions to stress; at the hormonal level its activity is represented by the enzyme alpha-amylase, for which increases can be observed immediately after a stressor. Compared to the SNS, the HPA-axis has more reciprocal connections with cortical structures involved in cognitive(-emotional) control and memory, such as the orbitofrontal cortex and the hippocampus: in humans, its activity is generally represented by (changes in) the hormone cortisol, which reaches its peak level approximately 20–30 min after a stressor. Based on the evidence for PE-related changes to brain structures and neurotransmitter systems important for stress regulation, it may be hypothesized that children of mothers who smoked during pregnancy will have altered stress reactivity. Studies to date have indeed provided indications that this is the case. Behaviorally, it was shown that PE-children had less frustration or stress tolerance than non-exposed controls during delays in a (simple) cognitive task (Huijbregts et al., 2008a). Findings regarding neurobiological stress reactivity are mixed, with reports of increased cortisol reactions in PE-children (Schuetze et al., 2008) and reports of the absence of any difference in cortisol and alpha-amylase reactivity with non-exposed children (Granger et al., 2007). There are also reports showing reduced physiological stress reactivity in exposed children, although these generally used non-hormonal measurements, such as several different heart rate indices (Fifer et al., 2009). An important consideration here might be that these studies almost exclusively focused on infants, whose stress systems work differently from those of older children and whose (neurobiological) stress reactivity may also change following (continuously) high exposure to stress hormones early in life (from hyper-reactive to hypo-reactive: Miller et al., 2007). This is the first study to investigate hormonal stress reactivity (in combination with mood changes and behavioral stress reactivity) in exposed children from an older age group. For this group we hypothesize lower neurobiological stress reactivity compared to controls, in part based on the evidence suggesting reduced functional activity of cholinergic and catecholaminergic systems. We further expect to replicate our earlier finding of behavioral hyper-reactivity in the exposed group (Huijbregts et al., 2008a), more negative moods and more extreme mood changes. A similar discrepancy between neurobiological and behavioral stress reactivity has been reported for children with disruptive behavior disorders (DBDs) (Van Goozen et al., 2000, Van Goozen et al., 2007 and Ortiz and Raine, 2004) Prenatal tobacco exposure has repeatedly been related to externalizing behavior problems in children, even when controlling for other established risk factors, but the mechanisms underlying this association are still unclear (Wakschlag et al., 2006 and Huijbregts et al., 2008b). Altered patterns of behavioral and neurobiological stress reactivity in exposed children might, in part, explain their elevated levels of externalizing behavior. Thus, our two main research questions were (1) do children prenatally exposed to tobacco have altered stress reactivity, both neurobiologically and behaviorally? and (2) does stress reactivity of children prenatally exposed to tobacco resemble that of children with DBDs?

3. Results 3.1. Externalizing behavior Mean CBCL externalizing scores were 8.5 (SD = 7.0) for the PE-group, 21.3 (SD = 9.8) for the DBD-group, and 4.6 (SD = 3.8) for NC. The three groups differed significantly from each other [F(2,35) = 11.2, p < .001, View the MathML sourceηp2=.39], with the DBD-group displaying significantly more externalizing behavior than both the NC [(Fisher's LSD: p < .001] and the PE-group [(Fisher's LSD: p = .001], but no significant difference between PE and NC-children. 3.2. Mood Repeated measures ANOVAs showed significant overall group differences for reported anger [F(2,35) = 7.0, p = .003, View the MathML sourceηp2=.29], unhappiness [F(2,35) = 6.4, p = .004, View the MathML sourceηp2=.27], and lack of cheerfulness [F(2,35) = 4.3, p = .022, View the MathML sourceηp2=20]. DBD-children generally indicated feeling more angry and unhappy, and less cheerful than both PE- and NC-children. There were no significant differences between PE- and NC-children. There was one trend suggesting a group by rating time interaction. This was observed for reports of feeling stressed [F(4,70) = 2.4, p = .061, View the MathML sourceηp2=.12]. Whereas generally the stress (and recovery) manipulations appeared to work [F(2,70) = 21.5, p < .001, View the MathML sourceηp2=.38], with significantly higher reported stress at rating point 2 compared to rating point 1, and significantly lower reported stress at rating point 3 compared to rating point 2, DBD-children's reports of feeling stressed remained stable between rating points 1 and 2, whereas PE and NC-children showed an increase in reported stress between rating points 1 and 2. Note, however, that there were significant group differences in reported stress at rating point 1 [F(2,35) = 4.3, p = .022, View the MathML sourceηp2=.20], with DBD-children feeling more stressed than both PE-children (Fisher's LSD: p = .036) and NC-children (Fisher's LSD: p = .007). 3.3. Behavioral stress reactivity Univariate analysis of variance showed significant group differences regarding behavioral stress reactivity (i.e., the square root transformed number of response button presses during 20s-delays in the delay frustration task): F(2,35) = 19.8, p < .001, View the MathML sourceηp2=.53. Both the PE-group (M = 23.8, SD = 4.3) and the DBD-group (M = 22.2, SD = 4.8) pressed the response button significantly more often than controls (M = 13.1, SD = 5.4) (Fisher's LSD: p < .001 in both instances). PE- and DBD-children did not differ from each other in behavioral stress reactivity. 3.4. Neurobiological stress reactivity Repeated measures ANOVA showed a decline in cortisol levels throughout the test session [F(4,136) = 14.8, p < 0.001, View the MathML sourceηp2=.30]. There were no significant differences in cortisol levels between groups, either at particular measurement points or with respect to the pattern of responses ( Fig. 1). Repeated measures ANOVA further showed a significant increase in sAa-levels during the test session [F(4,140) = 8.2, p < 0.001, View the MathML sourceηp2=.19]. A significant interaction between group and sampling time [F(8,140) = 2.2, p = .032, View the MathML sourceηp2=.11] indicated that the sAa-reactivity pattern differed between groups: whereas sAa-levels in both the DBD-group and the PE-group did not change, sAa levels of controls increased over the session ( Fig. 2). Contrast analyses showed that the interaction was best captured by a cubic contrast [F(2,35) = 4.5, p = .018, View the MathML sourceηp2=.21], indicating that, whereas sAa-levels remained stable for PE and DBD, they increased at the beginning of the session for NC (between measurements 1 and 2), then became more stable, and then increased again (starting between measurements 3 and 4, with a further increase between measurements 4 and 5). The ANOVA further showed an almost significant overall group difference for mean sAa-level [F(2,35) = 3.2, p = .055, View the MathML sourceηp2=.15], which was determined by significantly higher sAa-levels for NC- compared to PE-children (p = .017). This overall difference, in turn, was determined by PE–NC differences at assessments 2–5. Salivary cortisol levels for children prenatally exposed to tobacco (PE), ... Figure 1. Salivary cortisol levels for children prenatally exposed to tobacco (PE), children characterized by high levels of disruptive behavior (DBD), and normal controls (NC) during a stress-inducing test session. Figure options Salivary alpha-amylase levels for children prenatally exposed to tobacco (PE), ... Figure 2. Salivary alpha-amylase levels for children prenatally exposed to tobacco (PE), children characterized by high levels of disruptive behavior (DBD), and normal controls (NC) during a stress-inducing test session. Figure options Further non-parametric analyses with PE children divided into those exposed to <10 cigarettes/day and those exposed to ≥10 cigarettes/day, showed that differences in both sAa-levels and sAa-slope were particularly evident when NC-children were compared to PE-children exposed to ≥10 cigarettes/day. Mann–Whitney tests comparing the sAa-slopes showed a significant difference between NC and PE ≥ 10 (U = 19, z = −2.6, p = .007) and a non-significant trend for the difference between NC and PE < 10 (U = 23, z = −1.7, p = .087). NC-children had significantly higher sAa-levels than PE ≥ 10-children at assessments 2–5 (#2: p = .014; #3: p = .029; #4:, p = .025; #5: p = .012), whereas mean sAa-levels were only higher for NC than for PE < 10 at assessment 3 (p = .042), with a trend observed at assessment 5 (p = .079). 3.5. Associations between mood, behavioral stress reactivity and neurobiological stress reactivity There were only few significant correlations between indices of neurobiological and behavioral stress reactivity on the one hand and (changes in) mood on the other. Overall, reported anger and stress at baseline were positively related to behavioral stress reactivity (i.e., number of button presses during delays in the DF-task), although these correlations just failed to reach significance (r = .23, p = .087 for baseline stress, and r = .24, p = .072 for baseline anger). These correlations (both in direction and magnitude) were similar for all groups. Furthermore, there were differential correlations between sAa and reported stress for PE-children and NC-children at rating point 2 (i.e., after a series of stress-inducing tasks and film clips). There was a negative association for PE-children (r = −.48, p = .042) and a positive association for NC-children (r = .46, p = .043). These results indicate that, for PE-children, lower sAa-levels were associated with more stress feelings, whereas for NC-children higher sAa-levels were associated with more stress feelings. Something similar was observed for associations between sAa and reported anger at rating point 2: PE: r = −.66, p = .005; NC: r = .39, p = .077. Most importantly, there was an overall inverse association between sAa-reactivity and DF-reactivity (i.e., between neurobiological and behavioral stress reactivity): r = −.49, p = .001, indicating that less sAa-reactivity was associated with more behavioral reactivity. This association was accounted for by the PE-children (r = −.73, p = .001) and a trend for the DBD-children (r = −.53, p = .070), while there was no significant association for NC.