Separating Measurement Error and Signal in Environmental Data: Use of Replicates to Address Uncertainty.

Measurement uncertainty has long been a concern in the characterizing and interpreting environmental and toxicological measurements. We compared statistical analysis approaches when there are replicates: a Naïve approach that omits replicates, a Hybrid approach that inappropriately treats replicates as independent samples, and a Measurement Error Model (MEM) approach in a random effects analysis of variance (ANOVA) model that appropriately incorporates replicates. A simulation study assessed the effects of sample size and levels of replication, signal variance, and measurement error on estimates from the three statistical approaches. MEM results were superior overall with confidence intervals for the observed mean narrower on average than those from the Naïve approach, giving improved characterization. The MEM approach also featured an unparalleled advantage in estimating signal and measurement error variance separately, directly addressing measurement uncertainty. These MEM estimates were approximately unbiased on average with more replication and larger sample sizes. Case studies illustrated analyzing normally distributed arsenic and log-normally distributed chromium concentrations in tap water and calculating MEM confidence intervals for the true, latent signal mean and latent signal geometric mean (i.e., with measurement error removed). MEM estimates are valuable for study planning; we used simulation to compare various sample sizes and levels of replication.

Genomic comparisons between hepatocarcinogenic and non-hepatocarcinogenic organophosphate insecticides in the mouse liver.

Short-term biomarkers of toxicity have an increasingly important role in the screening and prioritization of new chemicals. In this study, we examined early indicators of liver toxicity for three reference organophosphate (OP) chemicals, which are among the most widely used insecticides in the world. The OP methidathion was previously shown to increase the incidence of liver toxicity, including hepatocellular tumors, in male mice. To provide insights into the adverse outcome pathway (AOP) that underlies these tumors, effects of methidathion in the male mouse liver were examined after 7 and 28 day exposures and compared to those of two other OPs that either do not increase (fenthion) or possibly suppress liver cancer (parathion) in mice. None of the chemicals caused increases in liver weight/body weight or histopathological changes in the liver. Parathion decreased liver cell proliferation after 7 and 28 days while the other chemicals had no effects. There was no evidence for hepatotoxicity in any of the treatment groups. Full-genome microarray analysis of the livers from the 7 and 28 day treatments demonstrated that methidathion and fenthion regulated a large number of overlapping genes, while parathion regulated a unique set of genes. Examination of cytochrome P450 enzyme activities and use of predictive gene expression biomarkers found no consistent evidence for activation of AhR, CAR, PXR, or PPARα. Parathion suppressed the male-specific gene expression pattern through STAT5b, similar to genetic and dietary conditions that decrease liver tumor incidence in mice. Overall, these findings indicate that methidathion causes liver cancer by a mechanism that does not involve common mechanisms of liver cancer induction.

Censoring Trace-Level Environmental Data: Statistical Analysis Considerations to Limit Bias.

Trace-level environmental data typically include values near or below detection and quantitation thresholds where health effects may result from low-concentration exposures to one chemical over time or to multiple chemicals. In a cook stove case study, bias in dibenzo[a,h]anthracene concentration means and standard deviations (SDs) was assessed following censoring at thresholds for selected analysis approaches: substituting threshold/2, maximum likelihood estimation, robust regression on order statistics, Kaplan-Meier, and omitting censored observations. Means and SDs for gas chromatography-mass spectrometry-determined concentrations were calculated after censoring at detection and calibration thresholds, 17% and 55% of the data, respectively. Threshold/2 substitution was the least biased. Measurement values were subsequently simulated from two log-normal distributions at two sample sizes. Means and SDs were calculated for 30%, 50%, and 80% censoring levels and compared to known distribution counterparts. Simulation results illustrated (1) threshold/2 substitution to be inferior to modern after-censoring statistical approaches and (2) all after-censoring approaches to be inferior to including all measurement data in analysis. Additionally, differences in stove-specific group means were tested for uncensored samples and after censoring. Group differences of means tests varied depending on censoring and distributional decisions. Investigators should guard against censoring-related bias from (explicit or implicit) distributional and analysis approach decisions.

Particle emissions from fused deposition modeling 3D printers: Evaluation and meta-analysis.

Fused deposition modeling (FDM) 3D printers, the most popular choice among home hobbyists, have been shown to release volatile organic chemicals (VOCs) and billions of airborne particles per minute, indicating the potential for consumer inhalation exposure and consequent health risks. Publications on FDM 3D printer emissions however, contain large heterogeneity of testing methods and analytical procedures making it difficult to reach overall conclusions for particle characteristics or particle number emission rates across the field. In this publication, data were collected over the printing time from 3D printer emission studies including particle count diameters (PCDs) (nanometers), particle number concentrations (PNCs) (particles/cm3), and particle number emission rates (PNERs) (particles min-1). Despite heterogeneity in methods, the majority of particles released were reported as ultrafine in size (i.e., <100 nm) indicating that using both acrylonitrile butadiene styrene (ABS) and poly-lactic acid (PLA) may present a risk of exposure to respirable particles. Mean PNC emitted in 3D printing tests ranged over several orders of magnitude across publications with overall means of 300,980 particles/cm3 for ABS and 65,482 particles/cm3 for PLA. Although mean PNC data were available from only 7 of the 16 papers reviewed, ABS resulted in greater particle numbers than PLA suggesting increased exposure to ultrafine particles. A linear mixed model was fitted for mean PNCs to further explore the impact of nozzle temperature and filament material. Finally, the PNER calculation method especially regarding losses, varied widely across studies, and directly impacted the PNERs reported. To strengthen direct comparability of results going forward, it is recommended that standard emissions testing protocols be developed for FDM 3D printers and particle influxes and losses be more uniformly calculated.

Moderate perinatal thyroid hormone insufficiency alters visual system function in adult rats.

Thyroid hormone (TH) is critical for many aspects of neurodevelopment and can be disrupted by a variety of environmental contaminants. Sensory systems, including audition and vision are vulnerable to TH insufficiencies, but little data are available on visual system development at less than severe levels of TH deprivation. The goal of the current experiments was to explore dose-response relations between graded levels of TH insufficiency during development and the visual function of adult offspring. Pregnant Long Evans rats received 0 or 3 ppm (Experiment 1), or 0, 1, 2, or 3 ppm (Experiment 2) of propylthiouracil (PTU), an inhibitor of thyroid hormone synthesis, in drinking water from gestation day (GD) 6 to postnatal day (PN) 21. Treatment with PTU caused dose-related reductions of serum T4, with recovery on termination of exposure, and euthyroidism by the time of visual function testing. Tests of retinal (electroretinograms; ERGs) and visual cortex (visual evoked potentials; VEPs) function were assessed in adult offspring. Dark-adapted ERG a-waves, reflecting rod photoreceptors, were increased in amplitude by PTU. Light-adapted green flicker ERGs, reflecting M-cone photoreceptors, were reduced by PTU exposure. UV-flicker ERGs, reflecting S-cones, were not altered. Pattern-elicited VEPs were significantly reduced by 2 and 3 ppm PTU across a range of stimulus contrast values. The slope of VEP amplitude-log contrast functions was reduced by PTU, suggesting impaired visual contrast gain. Visual contrast gain primarily reflects function of visual cortex, and is responsible for adjusting sensitivity of perceptual mechanisms in response to changing visual scenes. The results indicate that moderate levels of pre-and post-natal TH insufficiency led to alterations in visual function of adult rats, including both retinal and visual cortex sites of dysfunction.

From the Cover: Genomic Effects of Androstenedione and Sex-Specific Liver Cancer Susceptibility in Mice.

Current strategies for predicting carcinogenic mode of action for nongenotoxic chemicals are based on identification of early key events in toxicity pathways. The goal of this study was to evaluate short-term key event indicators resulting from exposure to androstenedione (A4), an androgen receptor agonist and known liver carcinogen in mice. Liver cancer is more prevalent in men compared with women, but androgen-related pathways underlying this sex difference have not been clearly identified. Short-term hepatic effects of A4 were compared with reference agonists of the estrogen receptor (ethinyl estradiol, EE) and glucocorticoid receptor (prednisone, PRED). Male B6C3F1 mice were exposed for 7 or 28 days to A4, EE, or PRED. EE increased and PRED suppressed hepatocyte proliferation, while A4 had no detectable effects. In a microarray analysis, EE and PRED altered >3000 and >670 genes, respectively, in a dose-dependent manner, whereas A4 did not significantly alter any genes. Gene expression was subsequently examined in archival liver samples from male and female B6C3F1 mice exposed to A4 for 90 days. A4 altered more genes in females than males and did not alter expression of genes linked to activation of the mitogenic xenobiotic receptors AhR, CAR, and PPARα in either sex. A gene expression biomarker was used to show that in female mice, the high dose of A4 activated the growth hormone-regulated transcription factor STAT5b, which controls sexually dimorphic gene expression in the liver. These findings suggest that A4 induces subtle age-related effects on STAT5b signaling that may contribute to the higher risk of liver cancer in males compared with females.

Chemical-agnostic hazard prediction: statistical inference of in vitro toxicity pathways from proteomics responses to chemical mixtures.

Toxicity pathways have been defined as normal cellular pathways that, when sufficiently perturbed as a consequence of chemical exposure, lead to an adverse outcome. If an exposure alters one or more normal biological pathways to an extent that leads to an adverse toxicity outcome, a significant correlation must exist between the exposure, the extent of pathway alteration, and the degree of adverse outcome. Biological pathways are regulated at multiple levels, including transcriptional, post-transcriptional, post-translational, and targeted degradation, each of which can affect the levels and extents of modification of proteins involved in the pathways. Significant alterations of toxicity pathways resulting from changes in regulation at any of these levels therefore are likely to be detectable as alterations in the proteome. We hypothesize that significant correlations between exposures, adverse outcomes, and changes in the proteome have the potential to identify putative toxicity pathways, facilitating selection of candidate targets for high throughput screening, even in the absence of a priori knowledge of either the specific pathways involved or the specific agents inducing the pathway alterations. We explored this hypothesis in vitro in BEAS-2B human airway epithelial cells exposed to different concentrations of Ni2+, Cd2+, and Cr6+, alone and in defined mixtures. Levels and phosphorylation status of a variety of signaling pathway proteins and cytokines were measured after 48 hours exposure, together with cytotoxicity. Least Absolute Shrinkage and Selection Operator (LASSO) multiple regression was used to identify a subset of these proteins that constitute a putative toxicity pathway capable of predicting cytotoxicity. The putative toxicity pathway for cytotoxicity of these metals and metal mixtures identified by LASSO is composed of phospho-RPS6KB1, phospho-p53, cleaved CASP3, phospho-MAPK8, IL-10, and Hif-1α. As this approach does not depend on knowledge of the chemical composition of the mixtures, it may be generally useful for identifying sets of proteins predictive of adverse effects for a variety of mixtures, including complex environmental mixtures of unknown composition.

Dose and Effect Thresholds for Early Key Events in a PPARα-Mediated Mode of Action.

Current strategies for predicting adverse health outcomes of environmental chemicals are centered on early key events in toxicity pathways. However, quantitative relationships between early molecular changes in a given pathway and later health effects are often poorly defined. The goal of this study was to evaluate short-term key event indicators using qualitative and quantitative methods in an established pathway of mouse liver tumorigenesis mediated by peroxisome proliferator-activated receptor alpha (PPARα). Male B6C3F1 mice were exposed for 7 days to di (2-ethylhexyl) phthalate (DEHP), di-n-octyl phthalate (DNOP), and n-butyl benzyl phthalate (BBP), which vary in PPARα activity and liver tumorigenicity. Each phthalate increased expression of select PPARα target genes at 7 days, while only DEHP significantly increased liver cell proliferation labeling index (LI). Transcriptional benchmark dose (BMDT) estimates for dose-related genomic markers stratified phthalates according to hypothetical tumorigenic potencies, unlike BMDs for non-genomic endpoints (relative liver weights or proliferation). The 7-day BMDT values for Acot1 as a surrogate measure for PPARα activation were 29, 370, and 676 mg/kg/day for DEHP, DNOP, and BBP, respectively, distinguishing DEHP (liver tumor BMD of 35 mg/kg/day) from non-tumorigenic DNOP and BBP. Effect thresholds were generated using linear regression of DEHP effects at 7 days and 2-year tumor incidence values to anchor early response molecular indicators and a later phenotypic outcome. Thresholds varied widely by marker, from 2-fold (Pdk4 and proliferation LI) to 30-fold (Acot1) induction to reach hypothetical tumorigenic expression levels. These findings highlight key issues in defining thresholds for biological adversity based on molecular changes.

Effects of Cold Temperature and Ethanol Content on VOC Emissions from Light-Duty Gasoline Vehicles.

Emissions of speciated volatile organic compounds (VOCs), including mobile source air toxics (MSATs), were measured in vehicle exhaust from three light-duty spark ignition vehicles operating on summer and winter grade gasoline (E0) and ethanol blended (E10 and E85) fuels. Vehicle testing was conducted using a three-phase LA92 driving cycle in a temperature-controlled chassis dynamometer at two ambient temperatures (-7 and 24 °C). The cold start driving phase and cold ambient temperature increased VOC and MSAT emissions up to several orders of magnitude compared to emissions during other vehicle operation phases and warm ambient temperature testing, respectively. As a result, calculated ozone formation potentials (OFPs) were 7 to 21 times greater for the cold starts during cold temperature tests than comparable warm temperature tests. The use of E85 fuel generally led to substantial reductions in hydrocarbons and increases in oxygenates such as ethanol and acetaldehyde compared to E0 and E10 fuels. However, at the same ambient temperature, the VOC emissions from the E0 and E10 fuels and OFPs from all fuels were not significantly different. Cold temperature effects on cold start MSAT emissions varied by individual MSAT compound, but were consistent over a range of modern spark ignition vehicles.

Uses of NHANES Biomarker Data for Chemical Risk Assessment: Trends, Challenges, and Opportunities.

BACKGROUND: Each year, the U.S. NHANES measures hundreds of chemical biomarkers in samples from thousands of study participants. These biomarker measurements are used to establish population reference ranges, track exposure trends, identify population subsets with elevated exposures, and prioritize research needs. There is now interest in further utilizing the NHANES data to inform chemical risk assessments. OBJECTIVES: This article highlights a) the extent to which U.S. NHANES chemical biomarker data have been evaluated, b) groups of chemicals that have been studied, c) data analysis approaches and challenges, and d) opportunities for using these data to inform risk assessments. METHODS: A literature search (1999-2013) was performed to identify publications in which U.S. NHANES data were reported. Manual curation identified only the subset of publications that clearly utilized chemical biomarker data. This subset was evaluated for chemical groupings, data analysis approaches, and overall trends. RESULTS: A small percentage of the sampled NHANES-related publications reported on chemical biomarkers (8% yearly average). Of 11 chemical groups, metals/metalloids were most frequently evaluated (49%), followed by pesticides (9%) and environmental phenols (7%). Studies of multiple chemical groups were also common (8%). Publications linking chemical biomarkers to health metrics have increased dramatically in recent years. New studies are addressing challenges related to NHANES data interpretation in health risk contexts. CONCLUSIONS: This article demonstrates growing use of NHANES chemical biomarker data in studies that can impact risk assessments. Best practices for analysis and interpretation must be defined and adopted to allow the full potential of NHANES to be realized.

Raising the bar for reproducible science at the U.S. Environmental Protection Agency Office of Research and Development.

Considerable concern has been raised regarding research reproducibility both within and outside the scientific community. Several factors possibly contribute to a lack of reproducibility, including a failure to adequately employ statistical considerations during study design, bias in sample selection or subject recruitment, errors in developing data inclusion/exclusion criteria, and flawed statistical analysis. To address some of these issues, several publishers have developed checklists that authors must complete. Others have either enhanced statistical expertise on existing editorial boards, or formed distinct statistics editorial boards. Although the U.S. Environmental Protection Agency, Office of Research and Development, already has a strong Quality Assurance Program, an initiative was undertaken to further strengthen statistics consideration and other factors in study design and also to ensure these same factors are evaluated during the review and approval of study protocols. To raise awareness of the importance of statistical issues and provide a forum for robust discussion, a Community of Practice for Statistics was formed in January 2014. In addition, three working groups were established to develop a series of questions or criteria that should be considered when designing or reviewing experimental, observational, or modeling focused research. This article describes the process used to develop these study design guidance documents, their contents, how they are being employed by the Agency's research enterprise, and expected benefits to Agency science. The process and guidance documents presented here may be of utility for any research enterprise interested in enhancing the reproducibility of its science.

Data-driven asthma endotypes defined from blood biomarker and gene expression data.

The diagnosis and treatment of childhood asthma is complicated by its mechanistically distinct subtypes (endotypes) driven by genetic susceptibility and modulating environmental factors. Clinical biomarkers and blood gene expression were collected from a stratified, cross-sectional study of asthmatic and non-asthmatic children from Detroit, MI. This study describes four distinct asthma endotypes identified via a purely data-driven method. Our method was specifically designed to integrate blood gene expression and clinical biomarkers in a way that provides new mechanistic insights regarding the different asthma endotypes. For example, we describe metabolic syndrome-induced systemic inflammation as an associated factor in three of the four asthma endotypes. Context provided by the clinical biomarker data was essential in interpreting gene expression patterns and identifying putative endotypes, which emphasizes the importance of integrated approaches when studying complex disease etiologies. These synthesized patterns of gene expression and clinical markers from our research may lead to development of novel serum-based biomarker panels.

Investigating the American Time Use Survey from an exposure modeling perspective.

This paper describes an evaluation of the US Bureau of Labor Statistics' American Time Use Survey (ATUS) for potential use in modeling human exposures to environmental pollutants. The ATUS is a large, on-going, cross-sectional survey of where Americans spend time and what activities they undertake in those locations. The data are reported as a series of sequential activities over a 24-h time period--a "diary day"--starting at 0400 hours. Between 12,000 and 13,000 surveys are obtained each year and the Bureau has plans to continue ATUS for the foreseeable future. The ATUS already has about 73,000 diary days of data, more than twice as many as that which currently exists in the US Environmental Protection Agency's (EPA) "Consolidated Human Activity Database" (CHAD) that the Agency uses for exposure modeling purposes. There are limitations for using ATUS in modeling human exposures to environmental pollutants. The ATUS does not report the location for a number of activities regarded as "personal." For 2006, personal activities with missing location information totaled 572 min/day, on average, for survey participants: about 40% of their day. Another limitation is that ATUS does not distinguish between indoor and outdoor activities at home, two of the traditional locational demarcations used in human exposure modeling. This lack of information affects exposure estimates to both indoor and outdoor air pollutants and potentially affects non-dietary ingestion estimates for children, which can vary widely depending on whether or not a child is indoors. Finally, a detailed analysis of the work travel activity in a subsample from ATUS 2006 indicates that the coding scheme is not fully consistent with a CHAD-based exposure modeling approach. For ATUS respondents in this subsample who reported work as an activity, roughly 48% of their days were missing work travel at one or both ends of the work shift or reported within work-shift travel inconsistently. An extensive effort would be needed to recode work travel data from ATUS for EPA's exposure modeling purposes.

Relationship between PM2.5 collected at residential outdoor locations and a central site.

Regression models are developed to describe the relationship between ambient PM2.5 (particulate matter [PM] < or = 2.5 microm in aerodynamic diameter) mass concentrations measured at a central-site monitor with those at residential outdoor monitors. Understanding the determinants and magnitude of variability and uncertainty in this relationship is critical for understanding personal exposures in the evaluation of epidemiological data. The repeated measures regression models presented here address temporal and spatial characteristics of data measured in the 2004-2007 Detroit Exposure and Aerosol Research Study, and they take into account missing data and other data features. The models incorporate turbulence kinetic energy and planetary boundary layer height, meteorological data that are not routinely considered in models that relate central-site concentrations to exposure to health effects. It was found that turbulence kinetic energy was highly statistically significant in explaining the relationship of PM2.5 measured at a particular stationary outdoor air monitoring site with PM2.5 measured outside nearby residences for the temporal coverage of the data.

Intra- and inter-individual variability in location data for two U.S. health-compromised elderly cohorts.

This study provides descriptive statistical data on daily time spent in three locations of exposure assessment interest for two panel studies of health-compromised elderly individuals >65-year-old having multiple days of human activity data. The panel studies include individuals living in Los Angeles (CA) and Baltimore (MD) in various housing types. Three general locations are evaluated: outdoors, in vehicles, and total indoors. Of particular interest is providing information regarding the within- and between-individual variability in the time use data for the three locations. The data are analyzed using non-parametric statistics and alternative statistical models. Within and between variability are evaluated using intraclass correlation coefficients (ICCs); daily "lag-one" autocorrelation coefficients are also provided for the two samples. There were significant gender differences for selected seasonal and/or day-of-the-week metrics for: (1) outdoor time in Los Angeles, but not in Baltimore, and (2) in-vehicle time in both areas. Elderly women spent more time in these locations than similarly aged men. The ICC statistic indicates that most of the variability in the time spent in the three locations is due to intraindividual variability rather than to inter-individual variability. The results indicate that US Environmental Protection Agency should consider gender, day-of-the-week, and time-of-day data in its exposure modeling of daily activities undertaken by the health-compromised elderly population.