Linda Valeri

Mediation analysis allowing for exposure-mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros.

Mediation analysis is a useful and widely employed approach to studies in the field of psychology and in the social and biomedical sciences. The contributions of this article are several-fold. First we seek to bring the developments in mediation analysis for nonlinear models within the counterfactual framework to the psychology audience in an accessible format and compare the sorts of inferences about mediation that are possible in the presence of exposure-mediator interaction when using a counterfactual versus the standard statistical approach. Second, the work by VanderWeele and Vansteelandt (2009, 2010) is extended here to allow for dichotomous mediators and count outcomes. Third, we provide SAS and SPSS macros to implement all of these mediation analysis techniques automatically, and we compare the types of inferences about mediation that are allowed by a variety of software macros.

DNA Hypomethylation, Ambient Particulate Matter, and Increased Blood Pressure: Findings From Controlled Human Exposure Experiments

Background Short‐term exposures to fine (<2.5 μm aerodynamic diameter) ambient particulate‐matter (PM) have been related with increased blood pressure (BP) in controlled‐human exposure and community‐based studies. However, whether coarse (2.5 to 10 μm) PM exposure increases BP is uncertain. Recent observational studies have linked PM exposures with blood DNA hypomethylation, an epigenetic alteration that activates inflammatory and vascular responses. No experimental evidence is available to confirm those observational data and demonstrate the relations between PM, hypomethylation, and BP. Methods and Results We conducted a cross‐over trial of controlled‐human exposure to concentrated ambient particles (CAPs). Fifteen healthy adult participants were exposed for 130 minutes to fine CAPs, coarse CAPs, or HEPA‐filtered medical air (control) in randomized order with ≥2‐week washout. Repetitive‐element (Alu, long interspersed nuclear element‐1 [LINE‐1]) and candidate‐gene (TLR4, IL‐12, IL‐6, iNOS) blood methylation, systolic and diastolic BP were measured pre‐ and postexposure. After adjustment for multiple comparisons, fine CAPs exposure lowered Alu methylation (β‐standardized=−0.74, adjusted‐P=0.03); coarse CAPs exposure lowered TLR4 methylation (β‐standardized=−0.27, adjusted‐P=0.04). Both fine and coarse CAPs determined significantly increased systolic BP (β=2.53 mm Hg, P=0.001; β=1.56 mm Hg, P=0.03, respectively) and nonsignificantly increased diastolic BP (β=0.98 mm Hg, P=0.12; β=0.82 mm Hg, P=0.11, respectively). Decreased Alu and TLR4 methylation was associated with higher postexposure DBP (β‐standardized=0.41, P=0.04; and β‐standardized=0.84, P=0.02; respectively). Decreased TLR4 methylation was associated with higher postexposure SBP (β‐standardized=1.45, P=0.01). Conclusions Our findings provide novel evidence of effects of coarse PM on BP and confirm effects of fine PM. Our results provide the first experimental evidence of PM‐induced DNA hypomethylation and its correlation to BP.

The role of measurement error and misclassification in mediation analysis: mediation and measurement error.

Methods to estimate direct and indirect effects have been rapidly expanding.1–14 It is now well documented that such mediation analyses are subject to strong no-confounding assumptions and that an unmeasured confounder of the mediator-outcome relationship can lead to substantial bias in direct and indirect effect estimates.1,2,6,7,14,15 Much less attention has been given to the question of how measurement error may bias estimates of direct and indirect effects. le Cessie and colleagues.16 have done a service to investigators interested in direct effects by providing simple correction formulas for direct effects estimates in a variety of mediator measurement-error scenarios. Here we will consider the implications of these and other results as they relate to making inferences, not just about direct effects, but also about mediation and indirect effects.

Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures

Because humans are invariably exposed to complex chemical mixtures, estimating the health effects of multi-pollutant exposures is of critical concern in environmental epidemiology, and to regulatory agencies such as the U.S. Environmental Protection Agency. However, most health effects studies focus on single agents or consider simple two-way interaction models, in part because we lack the statistical methodology to more realistically capture the complexity of mixed exposures. We introduce Bayesian kernel machine regression (BKMR) as a new approach to study mixtures, in which the health outcome is regressed on a flexible function of the mixture (e.g. air pollution or toxic waste) components that is specified using a kernel function. In high-dimensional settings, a novel hierarchical variable selection approach is incorporated to identify important mixture components and account for the correlated structure of the mixture. Simulation studies demonstrate the success of BKMR in estimating the exposure-response function and in identifying the individual components of the mixture responsible for health effects. We demonstrate the features of the method through epidemiology and toxicology applications.

SAS macro for causal mediation analysis with survival data.

We let Ta and Ma denote respectively the values of the time-to-event outcome and mediator that would have been observed had the exposure A been set to level a. We let Tam denote the value of the time-to-event outcome that would have been observed had the exposure, A, and mediator, M , been set to levels a and m, respectively. The average controlled direct effect comparing exposure level a to a∗ and fixing the mediator to level m on the mean survival time ratio scale is defined by CDEa,a∗(m) = E[Tam]/E[Ta∗m]. The average natural direct effect is then defined by NDEa,a∗(a ∗) = E[TaMa∗ ]/E[Ta∗Ma∗ ]. The average natural indirect effect can be defined as NIEa,a∗(a) = E[TaMa ]/E[TaMa∗ ], which compares the effect of the mediator at levels Ma and Ma∗ on the survival outcome when exposure A is set to a. Controlled direct effects and natural direct and indirect effects within strata of C = c are then defined by: CDEa,a∗|c(m) = E[Tam|c]/E[Ta∗m|c], NDEa,a∗|c(a ∗) = E[TaMa∗ |c]/E[Ta∗Ma∗ |c] and NIEa,a∗|c(a) = E[TaMa |c]/E[TaMa∗ |c] respectively. For an arbitrary time-to-event variable V , let λV (t) and λV (t|c) denote the hazard or hazard conditional on covariates c at time t, that is the instantaneous rate of the event conditional on V ≥ t. The causal effects can also be defined on the hazard ratio scale, replacing E[·] with λ(·).

Assessing Global Adoption of One Health Approaches

Transdisciplinary One Health (OH) approaches have been rediscovered as a promising tactic for addressing complex health risks at the human-animal-ecosystem interface. However, there is little evidence of widespread adoption of OH approaches as the new operating normal for addressing these complex health issues. We have used a transformational change model as an evaluation tool and part of an overall assessment of the global adoption of OH approaches. This assessment establishes a point of reference for measuring progress toward OH approaches being the new operating normal. Global adoption of OH approaches will require more strategic efforts to build the case (value proposition), recruiting a broader pool of One Health champions, solidifying partnerships and unifying OH efforts.

The Joint Effect of Prenatal Exposure to Metal Mixtures on Neurodevelopmental Outcomes at 20–40 Months of Age: Evidence from Rural Bangladesh

Background: Exposure to chemical mixtures is recognized as the real-life scenario in all populations, needing new statistical methods that can assess their complex effects. Objectives: We aimed to assess the joint effect of in utero exposure to arsenic, manganese, and lead on children’s neurodevelopment. Methods: We employed a novel statistical approach, Bayesian kernel machine regression (BKMR), to study the joint effect of coexposure to arsenic, manganese, and lead on neurodevelopment using an adapted Bayley Scale of Infant and Toddler Development™. Third Edition, in 825 mother–child pairs recruited into a prospective birth cohort from two clinics in the Pabna and Sirajdikhan districts of Bangladesh. Metals were measured in cord blood using inductively coupled plasma-mass spectrometry. Results: Analyses were stratified by clinic due to differences in exposure profiles. In the Pabna district, which displayed high manganese levels [interquartile range (IQR): 4.8, 18μg/dl], we found a statistically significant negative effect of the mixture of arsenic, lead, and manganese on cognitive score when cord blood metals concentrations were all above the 60th percentile (As≥0.7μg/dl, Mn≥6.6μg/dl, Pb≥4.2μg/dl) compared to the median (As=0.5μg/dl, Mn=5.8μg/dl, Pb=3.1μg/dl). Evidence of a nonlinear effect of manganese was found. A change in log manganese from the 25th to the 75th percentile when arsenic and manganese were at the median was associated with a decrease in cognitive score of −0.3 (−0.5, −0.1) standard deviations. Our study suggests that arsenic might be a potentiator of manganese toxicity. Conclusions: Employing a novel statistical method for the study of the health effects of chemical mixtures, we found evidence of neurotoxicity of the mixture, as well as potential synergism between arsenic and manganese. https://doi.org/10.1289/EHP614

Molecular responses in the telomere-mitochondrial axis of ageing in the elderly: a candidate gene approach.

Experimental evidence shows that telomere shortening induces mitochondrial damage but so far studies in humans are scarce. Here, we investigated the association between leukocyte telomere length (LTL) and mitochondrial DNA (mtDNA) content in elderly and explored possible intermediate mechanisms by determining the gene expression profile of candidate genes in the telomere-mitochondrial axis of ageing. Among 166 non-smoking elderly, LTL, leukocyte mtDNA content and expression of candidate genes: sirtuin1 (SIRT1), tumor protein p53 (TP53), peroxisome proliferator-activated receptor γ-coactivator1α (PGC-1α), peroxisome proliferator-activated receptor γ-coactivator1β (PGC-1β), nuclear respiratory factor 1 (NRF1) and nuclear factor, erythroid 2 like 2 (NRF2), using a quantitave real time polymerase chain assay (qPCR). Statistical mediation analysis was used to study intermediate mechanisms of the telomere-mitochondrial axis of ageing. LTL correlated with leukocyte mtDNA content in our studied elderly (r = 0.23, p = 0.0047). SIRT1 gene expression correlated positively with LTL (r = 0.26, p = 0.0094) and leukocyte mtDNA content (r = 0.43, p < 0.0001). The other studied candidates showed significant correlations in the telomere-mitochondrial interactome but not independent from SIRT1. SIRT1 gene expression was estimated to mediate 40% of the positive association between LTL and leukocyte mtDNA content. The key finding of our study was that SIRT1 expression plays a pivotal role in the telomere-mitochondrial interactome.

The estimation of direct and indirect causal effects in the presence of misclassified binary mediator

Mediation analysis serves to quantify the effect of an exposure on an outcome mediated by a certain intermediate and to quantify the extent to which the effect is direct. When the mediator is misclassified, the validity of mediation analysis can be severely undermined. The contribution of the present work is to study the effects of non-differential misclassification of a binary mediator in the estimation of direct and indirect causal effects when the outcome is either continuous or binary and exposure-mediator interaction can be present, and to allow the correction of misclassification. A hybrid of likelihood-based and predictive value weighting method for misclassification correction coupled with sensitivity analysis is proposed and a second approach using the expectation-maximization algorithm is developed. The correction strategy requires knowledge of a plausible range of sensitivity and specificity parameters. The approaches are applied to a perinatal epidemiological study of the determinants of pre-term birth.

Polymorphisms in maternal folate pathway genes interact with arsenic in drinking water to influence risk of myelomeningocele

BACKGROUND Arsenic induces neural tube defects in many animal models. Additionally, studies have shown that mice with specific genetic defects in folate metabolism and transport are more susceptible to arsenic-induced neural tube defects. We sought to determine whether 14 single-nucleotide polymorphisms in genes involved in folate metabolism modified the effect of exposure to drinking water contaminated with inorganic arsenic and posterior neural tube defect (myelomeningocele) risk. METHODS Fifty-four mothers of children with myelomeningocele and 55 controls were enrolled through clinical sites in rural Bangladesh in a case-control study of the association between environmental arsenic exposure and risk of myelomeningocele. We assessed participants for level of myelomeningocele, administered questionnaires, conducted biological and environmental sample collection, and performed genotyping. Inductively coupled plasma mass spectrometry was used to measure inorganic arsenic concentration in drinking water. Candidate single-nucleotide polymorphisms were identified through review of the literature. RESULTS Drinking water inorganic arsenic concentration was associated with increased risk of myelomeningocele for participants with 4 of the 14 studied single-nucleotide polymorphisms in genes involved in folate metabolism: the AA/AG genotype of rs2236225 (MTHFD1), the GG genotype of rs1051266 (SLC19A1), the TT genotype of rs7560488 (DNMT3A), and the GG genotype of rs3740393 (AS3MT) with adjusted odds ratio of 1.13, 1.31, 1.20, and 1.25 for rs2236225, rs1051266, rs7560488, and rs3740393, respectively. CONCLUSION Our results support the hypothesis that environmental arsenic exposure increases the risk of myelomeningocele by means of interaction with folate metabolic pathways.