Biology subjects

Grandjean, P.

Publications and source records attributed to Grandjean, P..

2 recordsLinked to original sources

bioRxiv2017

Application of benchmark analysis for mixed contaminant exposures: Mutual adjustment of two perfluoroalkylate substances associated with immunotoxicity

BackgroundDevelopmental exposure to perfluorinated substances is associated with deficient IgG antibody responses to childhood vaccines as an indication of depressed immune system functions. As this outcome may represent a critical effect of these substances, calculation of benchmark dose (BMD) results would be useful for standards setting to protect exposed populations against adverse effects. However, in the mixed exposure setting of most epidemiological evidence, the two major and inter-related substances associated with this adverse effect have shown similar benchmark results that raise concerns about possible confounding.\n\nMethodsWith the aim of better characterizing the immunotoxicity impact of the two major perfluorinated substances in the mixed exposures, we carried out BMD calculations on prospective data from two prospective birth cohort studies from the Faroe Islands with a total of 1,146 children. Exposure data included serum concentrations of perfluorooctane sulfonate and perfluorooctanoate at birth and at age 5 years and, as outcome parameters, the serum concentrations of specific IgG antibodies against tetanus and diphtheria at ages 5 and 7. We calculated the BMDs and their lower confidence bounds (BMDLs) and included mutual adjustment for the two compounds.\n\nResultsThe BMDLs for the two immunotoxicants were of similar magnitude before and after adjustment. Both substances showed lower results for a logarithmic dose-response model, which also provided a slightly better fit than a linear dose model for both antibodies. We also used a broken curve shape that allowed a different slope below the median exposure. Postnatal exposure as represented by the age 5 serum concentration, showed a stronger association with the antibody outcomes than the prenatal exposure. Due to the correlation between the two immunotoxicants, the mutual adjustment resulted in elevated BMD results and p values. However, the BMDL values were virtually unchanged.\n\nConclusionsAdjustment for co-exposure to another immunotoxicant increased the variance and the BMD values, but affected the BMDL values only to a negligible extent. These calculations are in accordance with an interpretation that, when two toxicants appear to affect an outcome to an almost equal degree and none of them is known to be solely responsible, the exposures should both be considered responsible and attract equal regulatory attention until further evidence shows otherwise.

epidemiology↗

bioRxiv2017

Combining Ensemble Learning Techniques and G-Computation to Investigate Chemical Mixtures in Environmental Epidemiology Studies

BackgroundAlthough biomonitoring studies demonstrate that the general population experiences exposure to multiple chemicals, most environmental epidemiology studies consider each chemical separately when assessing adverse effects of environmental exposures. Hence, the critical need for novel approaches to handle multiple correlated exposures.\n\nMethodsWe propose a novel approach using the G-formula, a maximum likelihood-based substitution estimator, combined with an ensemble learning technique (i.e. SuperLearner) to infer causal effect estimates for a multi-pollutant mixture. We simulated four continuous outcomes from real data on 5 correlated exposures under four exposure-response relationships with increasing complexity and 500 replications. The first simulated exposure-response was generated as a linear function depending on two exposures; the second was based on a univariate nonlinear exposure-response relationship; the third was generated as a linear exposure-response relationship depending on two exposures and their interaction; the fourth simulation was based on a non-linear exposure-response relationship with an effect modification by sex and a linear relationship with a second exposure. We assessed the method based on its predictive performance (Minimum Square error [MSE]), its ability to detect the true predictors and interactions (i.e. false discovery proportion, sensitivity), and its bias. We compared the method with generalized linear and additive models, elastic net, random forests, and Extreme gradient boosting. Finally, we reconstructed the exposure-response relationships and developed a toolbox for interactions visualization using individual conditional expectations.\n\nResultsThe proposed method yielded the best average MSE across all the scenarios, and was therefore able to adapt to the true underlying structure of the data. The method succeeded to detect the true predictors and interactions, and was less biased in all the scenarios. Finally, we could correctly reconstruct the exposure-response relationships in all the simulations.\n\nConclusionsThis is the first approach combining ensemble learning techniques and causal inference to unravel the effects of chemical mixtures and their interactions in epidemiological studies. Additional developments including high dimensional exposure data, and testing for detection of low to moderate associations will be carried out in future developments.

epidemiology↗