Impact of prenatal tobacco smoking on infant telomere length trajectory and ADHD symptoms at 18 months: a longitudinal cohort study

Despite the established negative effects of prenatal maternal smoking on her offspring, as of 2016, one in 14 mothers in the USA reports smoking during pregnancy [1]. Maternal prenatal smoking is associated with preterm birth [2], placental abruption [3], low birthweight and fetal growth restriction [2‐4], stillbirth [3], and sudden infant death syndrome [3]. These adverse effects extend into childhood with links to childhood asthma [5] and obesity [6, 7], and meta-analytic data associates maternal prenatal smoking with attention deficit/hyperactivity disorder (ADHD) [8, 9]. The pathophysiological mechanisms of these broad consequences of prenatal maternal smoking are poorly understood; however, changes to mitochondrial function, oxidative stress, and placental apoptosis are hypothesized contributors [10‐12]. One molecular marker tied to oxidative stress and cellular apoptosis that is also associated with smoking is telomere length (TL) [13].

Telomeres are the protective nucleoprotein cap at the end of each chromosome that, due to the end replication problem, shorten with each cell cycle [14]. Both oxidative stress and DNA damage accelerate telomere loss. In adults, shorter TL is associated with smoking [13], with some evidence of a dose dependent effect [15, 16]. A recent meta-analysis found shorter TL in smoking adults compared to non-smoking adults; however, no association was seen between smoking and the rate of TL attrition across the life span. The authors suggested that the observed cross-sectional relationship, in adults, between shorter TL and smoking was due to unmeasured effects early in life [17]. Beyond these direct effects within an individual, maternal prenatal smoking has been previously associated with shorter TL in her offspring [18‐22].

To date, seven studies have reported on the relation between maternal prenatal smoking and TL in her offspring at different ages from different sources of DNA [20, 21, 23‐27]. Of these studies, six found that exposure to prenatal smoke was associated with shorter TL in offspring [20, 21, 23‐25, 27]; however, one study reported maternal smoking associated with longer newborn TL [26]. Shorter cord blood TL has also been reported in infants born to mothers exposed to second-hand smoke prenatally [21], as well as in mothers with higher concentrations of urinary cadmium—one of the main toxic components in cigarette smoke [27]. Maternal smoking during pregnancy has been associated with shorter cord blood TL [24]; however, maternal prenatal smoking was also reported to be associated with longer TL in lymphocyte subpopulations. In this study, the authors hypothesized that longer TL was the consequence of a shift in the replicative history of specific lymphocyte sub-populations [26]. Beyond effects on newborn TL, maternal prenatal smoking has also been associated with shorter TL in two cross-sectional studies in older children [20] and adolescents [25]. Collectively, these studies indicate both an initial and persistent impact of maternal prenatal smoking on offspring TL; however, the directionality and potential differences in the relation due to the tissue type in which TL was measured require further exploration. To date, no studies have explored the trajectory of offspring TL in relation to prenatal smoking, nor have any studies tested the association between prenatal smoking, TL, and child outcomes.

Accelerated TL shortening, in both adults and pediatric populations, has been related to many aging related diseases including cardiovascular disease [28], obesity [29], and diabetes [30]. Further accelerated TL loss has been associated with environmental exposures related to increased oxidative stress [31]. Lastly, TL in preclinical and animal models has been associated with cellular differentiation, cellular senescence, and cellular aging [32]. In addition to the established link between maternal prenatal smoking and ADHD in her offspring, studies have reported associations between ADHD and TL. Costa Dde et al. demonstrated that higher levels of hyperactivity-impulsivity were associated with shorter TL in both children and their mothers [33]. In contrast, Momany et al. demonstrated that previous childhood hyperactivity and impulsivity symptoms, but not current adult symptoms or persistence of ADHD symptoms, were associated with longer TL [34]. Dyskeratosis congenita is a telomeropathy with higher incidence of ADHD, suggesting a link between telomere pathology and ADHD. Given these results, additional prospective studies exploring the link between TL and ADHD, particularly given the evidence associating maternal smoking with both TL and ADHD, are needed.

This study examined the association between maternal prenatal smoking and infant TL trajectory across the first 18 months of life and how the change in TL influenced the relation between prenatal maternal smoking and maternal report of child ADHD symptoms at 18 months of age. First, we hypothesized that fetal exposure to maternal prenatal smoking would be associated with accelerated TL loss across the first 18 months of life in the infant. Second, we hypothesized that, consistent with previous studies, maternal prenatal smoking would be associated with higher ADHD symptoms in children at 18 months of age. Lastly, we hypothesized that the change in TL would moderate the association between maternal prenatal smoking and ADHD symptoms.

Maternal report of prenatal smoking was significantly associated with infant TL attrition across the first 18 months of life. Even after accounting for SES, race, birth weight, and infant sex, maternal prenatal smoking was associated with a greater rate of TL shortening from four to 18 months of infant age. These results are consistent with cross-sectional studies linking maternal prenatal smoking to shorter offspring TL and augment previous findings linking maternal prenatal smoking, secondhand smoke exposure, and maternal cadmium levels with alterations in offspring TL both in newborns and across childhood [20, 21, 23‐27]. Further, our findings provide initial evidence that altered TL trajectories, as a function of prenatal maternal smoking, is one mechanistic pathway that contributes to the lasting negative health effects of maternal smoking for health outcomes in offspring [43].

Consistent with meta-analytic findings, in this relatively small cohort, prenatal smoking was correlated with maternal report of ADHD symptoms in her 18-month-old child. In this cohort, a significant association was found between maternal report of depressive symptoms and both prenatal smoking and ADHD symptoms in her toddler at 18 months. As our modest sample size precluded testing of mediation, larger studies with adequate power may be useful in determining this relationship.

As our primary initial hypothesis was to examine TL as a biological pathway linking maternal prenatal smoking and offspring ADHD, we subsequently report that TL change across the first 18 months of life moderated the association between prenatal smoking and early symptoms of ADHD in non-depressed mothers. For children whose mothers reported prenatal smoking, higher maternal report of ADHD symptoms at 18 months of age was associated with lower TL attrition. While initially counter-intuitive, these findings fit with existing neurodevelopmental models of ADHD that suggest delayed cortical maturation contributes to ADHD symptoms and also with previous reports of longer TL in subpopulations of cord blood lymphocytes with maternal prenatal smoking [26]. Almanzar et al. suggest that maternal prenatal smoking results in the preferential loss of older lymphocytes, resulting in a younger population of circulating cells with longer TL. Alternatively, similar to previous studies that correlated telomere biology with oligodendrocyte maturation, one proposed hypothesis is that the delayed maturation and differentiation of the neurons in the prefrontal cortex associated with ADHD may be associated with longer TL [44]. If TL is reflective of cellular replicative history [26, 44], the finding of less TL attrition may be reflective of slower overall neurodevelopmental processes reflecting decreased cellular differentiation and less mature cell populations. Notably, as this study reports on TL measured from buccal epithelial cells, which are neuroectodermally derived, while previous studies have utilized mesodermally derived lymphocytes, the relation between neurodevelopmental trajectories and buccal TL may be more tightly aligned. Most recently, Cicek et al. reported an association between telomerase, the enzyme that elongates telomeres, and both ADHD symptoms and cognitive function associated with ADHD. The authors hypothesized that neuroinflammation, one mechanism increasingly tied to ADHD, resulted in elevated telomerase, which, while not measured in this study, would theoretically predict longer telomere length [45]. While highly speculative, the moderation by longer TL may reflect the influence of smoking induced inflammation triggering elevated telomerase and ADHD, perhaps indicating that while smoking contributes to this pathway the effect on TL is through telomerase and the relation between smoking and ADHD may be driven by inflammation, as such our findings reflect joint consequences of inflammation rather than more standard considerations of elevated oxidative stress and subsequent shorter TL. Future studies targeting additional markers of cellular differentiation, including telomerase and cellular markers of DNA damage, as well as explorations of the relation with inflammation, complimented by preclinical animal models, are needed to better disentangle these complex pathways.

While these results supplement the growing body of literature relating early childhood TL and prenatal tobacco exposure, they are not without limitations. Despite the repeated TL measurements within infants, this remains a modest sample in which to examine TL as both a marker of exposure and a predictor of a health outcome. Particularly given concerns about assay repeatability with the qPCR-based TL measurement, close attention to both sample size and assay repeatability calculated with ICCs is necessary. The calculated ICCs for the TL assays in this cohort were high, and, using available sample size estimates (tulane.trn.edu) and the calculated ICC > 0.90, we detected significant associations between the change in TL and prenatal smoking [41]. Another limitation is the reliance on maternal self-report and medical record data for the determination of maternal prenatal smoking [41, 46]. Our measure of exposure is limited to maternal use and did not capture additional sources of tobacco exposure, such as paternal smoking or maternal second-hand smoke exposure, nor were additional exposures such as alcohol, marijuana, prescribed or illicit drug use, or environmental contaminants, which are often co-occurring with tobacco use, included in our study design. Our data did not permit exploration of testing a dose-response effect of smoke exposure on TL. Additionally, our small sample size prohibited consideration of additional pregnancy covariates (i.e. pre-eclampsia). As there is sparse evidence to suggest a directionality of the relationship between TL attrition and ADHD symptomatology during infancy, challenges exist in discussing potential molecular mechanisms. Furthermore, the focus of this study is on prenatal exposure and did not capture postnatal smoking exposure; however, large-scale studies demonstrate that prenatal exposure is more predictive of negative health outcomes than postnatal exposure [47, 48]. Although our primary aim was to examine the impact of prenatal smoking, we subsequently examined if maternal postnatal smoking contributed to additional variance in the model. Maternal postnatal smoking was collected via self-report when the infant was 4 and 12 months of age. Although maternal prenatal smoking was positively correlated with maternal postnatal smoking, maternal postnatal smoking was not associated with infant TL at any time nor with ADHD symptoms. Since meta-analyses evaluating telomere attrition rates in adult smokers do not demonstrate greater attrition in smokers compared to non-smokers [17], further studies are needed to evaluate TL attrition at various life stages to help direct clinicians to the most influential point of exposure as it relates to long-term outcomes. As maternal depression has been associated with associated with both higher risk of maternal prenatal smoking [49, 50] and early childhood ADHD symptoms [51], depressed mothers were excluded from our moderation analysis; future studies sufficiently powered to test the biological pathways in women with depression are needed. Given the potential role of prematurity in infant TL dynamics and ADHD symptoms as a function of maternal prenatal smoking, the analyses excluding infants with gestational ages less than 37 weeks we run and the findings were consistent. We did not consider shared genetic factors that would be related to smoking behaviors and TL [52], and although there is evidence of heritability in telomere length to date there is no evidence of shared genetic risk with smoking behaviors and telomere length [53], future studies accounting for genetic contributions are warranted. Finally, maternal report of symptoms was done at a very early age and was based on the CBCL subscale scores, which are not diagnostic of ADHD, as such caution is warranted. However, the correlation between ADHD subscale scores and later diagnosis of ADHD is high although later variability is possible. Additional studies are needed to better understand the relation between prenatal exposure to smoking, the trajectory of telomere length attrition and socioemotional and behavioral outcomes throughout childhood.

Our findings confirm previously published studies linking offspring TL and prenatal maternal smoking and support the literature base demonstrating the intergenerational consequences of prenatal tobacco exposure. Previous data reiterates that continued smoking during pregnancy or third trimester smoking is linked to worse outcomes than for those mothers who either never smoked or stopped smoking early in pregnancy [4, 54]. In the future, TL may serve as a predictive index of later offspring health risks. Our findings provide a foundation to investigate this further from both a clinical and a mechanistic standpoint. Detailed prospective studies evaluating the trajectory of TL in children of mothers who smoke prenatally, from birth through adulthood, and the correlation with long-term negative health outcomes are needed to better understand these pathways. In addition, larger studies adequate powered to explore dose and duration of exposure may define sensitive periods of heightened neurodevelopment risk and additional factors in both the prenatal and the postnatal environment (i.e. gestational diabetes, pre-eclampsia) that can moderate long-term outcomes.

Given the accumulating public health concerns related to multiple prenatal substance exposures, evaluation of the independent and interactive effect of multiple substances of abuse (e.g., marijuana, opiates) on TL and long-term infant outcomes is an important future research direction. Despite well-documented adverse consequences of prenatal smoking, it remains unfortunately prevalent. To better understand the pathophysiologic mechanisms by which tobacco smoking during pregnancy influences the next generation, multigenerational studies could shed light on critical events and outcomes during and after fetal development.