Statistical Challenges in the Analysis of Biomarkers of Environmental Chemical Exposures for Perinatal Epidemiology

A review of statistical strategies to integrate biomarkers of chemical exposure with biomarkers of effect applied in omic-scale environmental epidemiology

Humans are exposed to a growing list of synthetic chemicals, some of them becoming a major public health concern due to their capacity to impact multiple biological endpoints and contribute to a range of chronic diseases. The integration of endogenous (omic) biomarkers of effect in environmental health studies has been growing during the last decade, aiming to gain insight on the potential mechanisms linking the exposures and the clinical conditions. The emergence of high-throughput omic platforms has raised a list of statistical challenges posed by the large dimension and complexity of data generated. Thus, the aim of the present study was to critically review the current state-of-the-science about statistical approaches used to integrate endogenous biomarkers in environmental-health studies linking chemical exposures with health outcomes. The present review specifically focused on internal exposure to environmental chemical pollutants, involving both persistent organic pollutants (POPs), non-persistent pollutants like phthalates or bisphenols, and metals. We identified 42 eligible articles published since 2016, reporting 48 different statistical workflows, mostly focused on POPs and using metabolomic profiling in the intermediate layer. The outcomes were mainly binary and focused on metabolic disorders. A large diversity of statistical strategies were reported to integrate chemical mixtures and endogenous biomarkers to characterize their associations with health conditions. Multivariate regression models were the most predominant statistical method reported in the published workflows, however some studies applied latent based methods or multipollutant models to overcome the specific constraints of omic or exposure of data. A minority of studies used formal mediation analysis to characterize the indirect effects mediated by the endogenous biomarkers. The principles of each specific statistical method and overall workflow set-up are summarized in the light of highlighting their applicability, strengths and weaknesses or interpretability to gain insight into the causal structures underlying the triad: exposure, effect-biomarker and outcome.

Using Latent Profile Analysis to Identify Associations Between Gestational Chemical Mixtures and Child Neurodevelopment

BACKGROUND

Unsupervised machine learning techniques have become increasingly popular for studying associations between gestational exposure mixtures and human health. Latent profile analysis is one method that has not been fully explored.

METHODS

We estimated associations between gestational chemical mixtures and child neurodevelopment using latent profile analysis. Using data from the Maternal-Infant Research on Environmental Chemicals (MIREC) research platform, a longitudinal cohort of pregnant Canadian women and their children, we generated latent profiles from 27 gestational exposure biomarkers. We then examined the associations between these profiles and child Verbal IQ, Performance IQ, and Full-Scale IQ, measured with the Wechsler Preschool and Primary Scale of Intelligence, Third Edition (WPPSI-III). We validated our findings using k-means clustering.

RESULTS

Latent profile analysis detected five latent profiles of exposure: a reference profile containing 61% of the study participants, a high monoethyl phthalate (MEP) profile with moderately low persistent organic pollutants (POPs) containing 26%, a high POP profile containing 6%, a low POP profile containing 4%, and a smoking chemicals profile containing 3%. We observed negative associations between both the smoking chemicals and high MEP profiles and all IQ scores and between the high POP profile and Full-Scale and Verbal IQ scores. We also found a positive association between the low POP profile and Full-Scale and Performance IQ scores. All associations had wide 95% confidence intervals.

CONCLUSIONS

Latent profile analysis is a promising technique for identifying patterns of chemical exposure and is worthy of further study for its use in examining complicated exposure mixtures.

Log-transformation of Independent Variables: Must We?

Epidemiologic studies often quantify exposure using biomarkers, which commonly have statistically skewed distributions. Although normality assumption is not required if the biomarker is used as an independent variable in linear regression, it has become common practice to log-transform the biomarker concentrations. This transformation can be motivated by concerns for nonlinear dose-response relationship or outliers; however, such transformation may not always reduce bias. In this study, we evaluated the validity of motivations underlying the decision to log-transform an independent variable using simulations, considering eight scenarios that can give rise to skewed X and normal Y. Our simulation study demonstrates that (1) if the skewness of exposure did not arise from a biasing factor (e.g., measurement error), the analytic approach with the best overall model fit best reflected the underlying outcome generating methods and was least biased, regardless of the skewness of X and (2) all estimates were biased if the skewness of exposure was a consequence of a biasing factor. We additionally illustrate a process to determine whether the transformation of an independent variable is needed using NHANES. Our study and suggestion to divorce the shape of the exposure distribution from the decision to log-transform it may aid researchers in planning for analysis using biomarkers or other skewed independent variables.

Prenatal exposure to mixtures of persistent environmental chemicals and fetal growth outcomes in Western Australia

BACKGROUND

Environmental chemicals have been implicated in the etiology of impaired fetal growth. However, few studies have assessed the effects of chemical mixtures or considered the possibility of non-monotonic exposure-response relationships for chemicals that act through the endocrine system.

METHODS

We assessed exposure to polybrominated diphenyl ethers, organochlorine pesticides, metals, and perfluorinated alkyl substances in blood and urine samples collected approximately two weeks prior to delivery in 166 non-smoking pregnant women, and subsequent birth weight, length, and head circumference of neonates who were part of the Australian Maternal Exposures to Toxic Substances (AMETS) study. We used Bayesian structured additive regression models with spike-slab priors to estimate mixture effects, identify important exposures, and model non-linearity in exposure-response relationships.

RESULTS

Mixtures of polybrominated diphenyl ethers, organochlorine pesticides, metals, and perfluorinated alkyl substances were not associated with fetal growth outcomes. Estimated change in fetal growth outcomes for an increase in exposure from the 25th to 75th percentile suggested no meaningful associations; the strongest evidence was for a small inverse association between birth weight and cesium exposure measured in whole blood (-124 g, 90% credible interval: -240 to -3 g). We identified several chemicals that may be associated with fetal growth non-linearly; however, 90% credible intervals contained small values consistent with no meaningful association.

CONCLUSIONS

Using a Bayesian penalized regression method, we assessed the shapes of exposure-response relationships, controlled for confounding by co-exposure, and estimated the single and combined effects of a large mixture of correlated environmental chemicals on fetal growth. Our findings, based on a small sample of mother-neonate pairs, suggest that mixtures of persistent chemicals are not associated with birth weight, length, and head circumference. The potential for non-monotonic relationships between environmental chemicals and fetal growth outcomes warrants further study.

Environmental mixtures and children's health: identifying appropriate statistical approaches

PURPOSE OF REVIEW

Biomonitoring studies have shown that children are constantly exposed to complex patterns of chemical and nonchemical exposures. Here, we briefly summarize the rationale for studying multiple exposures, also called mixture, in relation to child health and key statistical approaches that can be used. We discuss advantages over traditional methods, limitations and appropriateness of the context.

RECENT FINDINGS

New approaches allow pediatric researchers to answer increasingly complex questions related to environmental mixtures. We present methods to identify the most relevant exposures among a high-multitude of variables, via shrinkage and variable selection techniques, and identify the overall mixture effect, via Weighted Quantile Sum and Bayesian Kernel Machine regressions. We then describe novel extensions that handle high-dimensional exposure data and allow identification of critical exposure windows.

SUMMARY

Recent advances in statistics and machine learning enable researchers to identify important mixture components, estimate joint mixture effects and pinpoint critical windows of exposure. Despite many advantages over single chemical approaches, measurement error and biases may be amplified in mixtures research, requiring careful study planning and design. Future research requires increased collaboration between epidemiologists, statisticians and data scientists, and further integration with causal inference methods.