Sensitive Windows for In Utero Exposures and Asthma Development. Layers of Complexity

There is strong scientific support for a relationship between in utero exposures such as environmental pollution or maternal smoking during pregnancy and lung growth and development (1, 2) as well as subsequent pulmonary function (3, 4). Until recently, most studies have examined in utero exposures globally (i.e., cumulative or trimester exposure during pregnancy) (5, 6). However, evidence increasingly points to a more complex interplay among the toxic exposure, timing of exposure, and individual characteristics such as sex and genetic predisposition that culminate in altered lung structure and function (7).

In this issue of the Journal, Bose and colleagues (pp. 1396–1403) report that they employed novel Bayesian distributed lag interaction models to identify sensitive prenatal windows for the influence of nitrate (NO₃⁻) exposure on childhood asthma, accounting for effect modification by fetal sex and maternal psychological stress (8). In this primarily data-driven analysis, the relationship between prospectively collected cumulative daily prenatal NO₃⁻ exposure and the overall incidence of asthma by 6 years of age was not significant (odds ratio [OR], 1.20; 95% confidence interval [CI], 0.93–1.70; per interquartile range increase in ln NO₃⁻) among the total sample of 752 mother–child dyads born later than 37 weeks of gestation, and no sensitive gestational window for exposure was identified.

However, more detailed slicing of the data revealed a significant interaction with offspring sex and maternal psychological stress (high vs. low) as measured by the Crisis in Family Systems–Revised survey (9). Two distinct sensitive windows, at 7–19 weeks and 33–40 weeks of gestation, were identified in males exposed to high prenatal maternal stress. In addition, for male offspring of mothers who reported high stress, the odds of being diagnosed with asthma by 6 years of age was significantly increased (OR, 2.64; 95% CI, 1.27–5.39; per interquartile range increase in ln NO₃⁻). No significant relationship was found between NO₃⁻ exposure and asthma diagnosis among males whose mothers reported low prenatal stress or among females with low or high maternal prenatal stress exposure. Sensitive windows were not significant for any female offspring or for males of mothers who did not report high prenatal stress.

This article boasts a number of strengths. It uniquely examines the potential three-way interactions of prenatal maternal stress and fetal sex and their modifying effects on the relationship between prenatal NO₃⁻ pollution exposure and the development of childhood asthma. Other strengths include its prospective, longitudinal design and ethnically diverse population, with 54% being Hispanic and 29% being black. The authors adjusted for a large number of potential confounders, measured daily prenatal NO₃⁻ exposures using a hybrid chemical transport land-use regression model (10), and applied novel Bayesian distribution lag interaction models that adopt sliding windows of NO₃⁻ exposures throughout the pregnancy (11). This model identifies sensitive windows that are defined by where the estimated pointwise 95% CI does not include an OR of 1.

Bose and colleagues conclude that increased prenatal NO₃⁻ exposure during distinct sensitive windows was associated with incident asthma in boys concurrently exposed to high prenatal stress. Of note, the 95% CIs for this group are quite wide, and there were only 169 boys with high prenatal stress. This raises some concerns regarding study power of the subset analysis. In addition, a relatively large portion of subjects, 176 infants (about 24%), were born after 37 weeks but prior to the last week of gestation. For these patients, postnatal NO₃⁻ estimates corresponding to time were used. When the authors performed a sensitivity analysis using the imputed NO₃⁻ values at Weeks 37–39 for those infants, the missing data points and their imputed values greatly influenced the identification of the sensitive windows. That is, their Figure E2 suggests that there was no longer a sensitive window for boys with high prenatal maternal stress when imputed data were used. The authors adopted a polynomial spline regression imputation for missing values. This model-based imputation method might have helped to reduce bias or increase precision if missingness mechanisms (e.g., missing at random) had been examined (12).

Although the study is novel and well designed, the usual limitations of observational studies apply, and caution should be taken in interpreting the results. There is potential for misclassification of the exposure and the outcome (maternal report of physician diagnosis of asthma), unmeasured confounding, and the inability to infer causation. For instance, multiple risk factors for asthma have been identified, which are difficult to disentangle in such a homogeneous population of low socioeconomic status, given that people with limited resources have historically been forced to live in areas with more exposure to pollution. For infants born between 37 and 40 weeks, the authors used postnatal NO₃⁻ estimates, which may have different biological effects than a known intrauterine exposure.

This paper also raises interesting questions of interpretation. Does the fact that female offspring, or male offspring of low stress mothers, are less sensitive mean that there is less concern for NO₃⁻ exposure in these pregnancies, or rather does the identification of specific sensitivities reinforce the toxicity of NO₃⁻ exposures in general and begin to point to mechanisms of injury? The biologic underpinning for the effect modification by fetal sex suggested by the authors is focused on NO₃⁻ and stress exposures in relation to slower lung maturation among male fetuses. This raises some questions. Primarily, regardless of more rapid lung maturation, how do the authors explain the apparent lack of effect of NO₃⁻ and high stress exposure during the first critical window on the outcome of asthma in girls? In addition, it is not clear that sensitive windows can be clearly tied to stages of lung development at the sensitive time, because the window of sensitivity may be affecting subsequent patterns of gene expression in later periods of lung development.

In summary, this article reinforces the dangers of in utero exposure to air pollution and stresses the importance of first considering and identifying critical windows for exposure and not using a one-size-fits-all model of the dangers of toxic exposures. In addition, it is highly likely that different components of air pollution will have different critical windows, requiring further analysis of the specific temporal and subgroup toxicities of pollutants.