Robin McDougall

Associations of Perfluoroalkyl Substances (PFAS) with Lower Birth Weight: An Evaluation of Potential Confounding by Glomerular Filtration Rate Using a Physiologically Based Pharmacokinetic Model (PBPK)

Background Prenatal exposure to perfluoroalkyl substances (PFAS) has been associated with lower birth weight in epidemiologic studies. This association could be attributable to glomerular filtration rate (GFR), which is related to PFAS concentration and birth weight. Objectives We used a physiologically based pharmacokinetic (PBPK) model of pregnancy to assess how much of the PFAS–birth weight association observed in epidemiologic studies might be attributable to GFR. Methods We modified a PBPK model to reflect the association of GFR with birth weight (estimated from three studies of GFR and birth weight) and used it to simulate PFAS concentrations in maternal and cord plasma. The model was run 250,000 times, with variation in parameters, to simulate a population. Simulated data were analyzed to evaluate the association between PFAS levels and birth weight due to GFR. We compared simulated estimates with those from a meta-analysis of epidemiologic data. Results The reduction in birth weight for each 1-ng/mL increase in simulated cord plasma for perfluorooctane sulfonate (PFOS) was 2.72 g (95% CI: –3.40, –2.04), and for perfluorooctanoic acid (PFOA) was 7.13 g (95% CI: –8.46, –5.80); results based on maternal plasma at term were similar. Results were sensitive to variations in PFAS level distributions and the strength of the GFR–birth weight association. In comparison, our meta-analysis of epidemiologic studies suggested that each 1-ng/mL increase in prenatal PFOS and PFOA levels was associated with 5.00 g (95% CI: –21.66, –7.78) and 14.72 g (95% CI: –8.92, –1.09) reductions in birth weight, respectively. Conclusion Results of our simulations suggest that a substantial proportion of the association between prenatal PFAS and birth weight may be attributable to confounding by GFR and that confounding by GFR may be more important in studies with sample collection later in pregnancy. Citation Verner MA, Loccisano AE, Morken NH, Yoon M, Wu H, McDougall R, Maisonet M, Marcus M, Kishi R, Miyashita C, Chen MH, Hsieh WS, Andersen ME, Clewell HJ III, Longnecker MP. 2015. Associations of perfluoroalkyl substances (PFAS) with lower birth weight: an evaluation of potential confounding by glomerular filtration rate using a physiologically based pharmacokinetic model (PBPK). Environ Health Perspect 123:1317–1324; http://dx.doi.org/10.1289/ehp.1408837

Is the Relationship between Prenatal Exposure to PCB-153 and Decreased Birth Weight Attributable to Pharmacokinetics?

Background: A recent meta-analysis based on data from > 7,000 pregnancies reported an association between prenatal polychlorinated biphenyl (PCB)–153 exposure and reduced birth weight. Gestational weight gain, which is associated negatively with PCB levels in maternal and cord blood, and positively with birth weight, could substantially confound this association. Objective: We sought to estimate the influence of gestational weight gain on the association between PCB-153 exposure and birth weight using a pharmacokinetic model. Methods: We modified a recently published pharmacokinetic model and ran Monte Carlo simulations accounting for variability in physiologic parameters and their correlations. We evaluated the pharmacokinetic model by comparing simulated plasma PCB-153 levels during pregnancy to serial measurements in 10 pregnant women from another study population. We estimated the association between simulated plasma PCB-153 levels and birth weight using linear regression models. Results: The plasma PCB-153 level profiles generated with the pharmacokinetic model were comparable to measured levels in 10 pregnant women. We estimated a 118-g decrease in birth weight (95% CI: –129, –106 g) for each 1-μg/L increase in simulated cord plasma PCB-153, compared with the 150-g decrease estimated based on the previous meta-analysis. The estimated decrease in birth weight was reduced to –6 g (95% CI: –18, 6 g) when adjusted for simulated gestational weight gain. Conclusion: Our findings suggest that associations previously noted between PCB levels and birth weight may be attributable to confounding by maternal weight gain during pregnancy. Citation: Verner MA, McDougall R, Glynn A, Andersen ME, Clewell HJ III, Longnecker MP. 2013. Is the relationship between prenatal exposure to PCB-153 and decreased birth weight attributable to pharmacokinetics? Environ Health Perspect 121:1219–1224; http://dx.doi.org/10.1289/ehp.1206457

A Case Study Addressing the Reliability of Polychlorinated Biphenyl Levels Measured at the Time of Breast Cancer Diagnosis in Representing Early-Life Exposure

Background: To date, breast cancer epidemiologic studies have relied on blood or tissue specimens sampled at the time of diagnosis or a few years prior to assess lifetime exposure to polychlorinated biphenyls (PCB). In this study, we evaluated whether such PCB measurements are indicative of early-life levels by reconstructing lifetime toxicokinetic profiles for women included in the CECILE case–control study, using a physiologically based pharmacokinetic (PBPK) model. Methods: We simulated lifetime toxicokinetic profiles of PCB-153 for 2,134 French women by incorporating information on body weight history, height, pregnancies, and breast-feeding in the PBPK model. Oral dose was calculated by considering measured blood PCB-153 and the temporal trend of environmental contamination. Area under the concentration versus time curve (AUC) for each decade of life and maximum blood concentration (Cmax) were compiled and compared with measured levels, using Pearson partial correlation analyses adjusting for age at diagnosis. Results: When considering all individuals, simulated AUCs correlated with measured PCBs, with coefficients ranging from 0.735 to 0.981. The weakest correlations were obtained with AUCs for the first decades of life. Stratified analyses suggested that breast-feeding reduces the reliability of late-life blood levels in representing lifetime exposure. Conclusion: Results of this study suggest that PCB levels measured at the time of diagnosis do not fully represent early-life exposures. Impact: PBPK-derived estimates of early-life levels circumvent the limitations of current approaches in assessing PCB lifetime exposure and may be used to address hypothesized windows of breast vulnerability (e.g., puberty) in this population. Cancer Epidemiol Biomarkers Prev; 20(2); 281–6. ©2010 AACR.

SYNTHESIS OF GRASHOF FOUR-BAR MECHANISMS USING PARTICLE SWARM OPTIMIZATION

A Particle Swarm Optimization algorithm (PSO) and Grashof criteria for a four-bar mechanism are used to develop a general Grashof four-bar mechanism synthesis routine. The use of the stochastic PSO algorithm to minimize the non-linear objective function removes the requirement for sequential parameter transforms and design constraints previously employed in tandem with traditional deterministic quasi-Newtonian optimization routines to achieve convergence on meaningful global minima which often artificially constrict the feasible solution set space under exploration. The performance of this implementation compares favourably with other mechanism synthesis algorithms, particularly as the number of specified precision points is increased.© 2008 ASME

Using population physiologically based pharmacokinetic modeling to determine optimal sampling times and to interpret biological exposure markers: The example of occupational exposure to styrene

BACKGROUND Biomonitoring of chemicals in the workplace provides an integrated characterization of exposure that accounts for uptake through multiple pathways and physiological parameters influencing the toxicokinetics. OBJECTIVES We used the case of styrene to (i) determine the best times to sample venous blood and end-exhaled air, (ii) characterize the inter-individual variability in biological levels following occupational exposure and (iii) propose biological limit values using a population physiologically based pharmacokinetic (PBPK) model. METHODS We performed Monte Carlo simulations with various physiological, exposure and workload scenarios. Optimal sampling times were identified through regression analyses between levels in biological samples and 24-h area under the arterial blood concentration vs. time curve. We characterized the variability in levels of styrene in biological samples for exposures to a time weighted average (TWA) of 20ppm. RESULTS Simulations suggest that the best times to sample venous blood are at the end of shift in poorly ventilated workplaces and 15min after the shift in highly ventilated workplaces. Exhaled air samples are most informative 15min after the shift. For a light workload, simulated styrene levels have a median (5th-95th percentiles) of 0.4mg/l (0.2-0.6) in venous blood at the end of shift and 0.5ppm (0.3-0.8) in exhaled air 15min after the end of shift. CONCLUSION This study supports the current BEI(®) of the ACGIH of 0.2mg/l of styrene in venous blood at the end of shift and indicates a biological limit value of 0.3ppm in end-exhaled air 15min after the end of shift.

A mobile robotic platform for generating radiation maps

The use of mobile robots to collect the sensor readings required to generate radiation maps has the significant advantage of eliminating the risk of exposure that humans would otherwise face by collecting the readings by hand. In this work, a mobile robotic platform designed specifically to collect this information to synthesize radiation maps is presented. Details of the design are discussed, focusing in particular on the physical map generating capabilities of this new platform that are necessary to enable the generation of the radiation maps. The physical maps are generated using a laser range finder based implementation of Simultaneous Localization and Mapping (SLAM).

On the Application of Multi-Objective Parallel Asynchronous Particle Swarm Optimization to Engineering Design Problems

A distributed variant of multi-objective particle swarm optimization (MOPSO) called multi-objective parallel asynchronous particle swarm optimization (MOPAPSO) is presented and the effects of distribution of objective function calculations to slave processors on the results and performance are investigated. Two benchmark examples were used to verify the capability of this implementation of MOPAPSO to match previously published results of MOPSO. The computationally intensive task of multi-objective Optimization Based Mechanism Synthesis (OBMS) was used to verify that significant performance improvements were realized through parallelization. The results show that MOPAPSO is able to match the results of MOPSO in significantly less time. The fact that MOPAPSO is distributed results in an effective optimization tool for complex multi-objective design problems.Copyright