Multiplicity-adjusted Inferences in Risk Assessment: Benchmark Analysis with Continuous Response Data

We develop and study multiplicity adjustments for low-dose inferences in environmental risk assessment. Application is intended for risk analysis studies where human, animal, or ecological data are used to set safe levels of a hazardous environmental agent. A modern method for making inferences in this setting is known as benchmark analysis, where attention centers on the dose at which a fixed benchmark level of risk is achieved. Both upper confidence limits on the risk and lower confidence limits on the “benchmark dose” are of interest. In practice, a number of possible benchmark risks may be under study; if so, corrections must be applied to adjust the limits for multiplicity. In this note, we discuss approaches for doing so with continuous, nonquantal response data.     

Comparing model averaging with other model selection strategies for benchmark dose estimation

Model averaging (MA) has been proposed as a method of accommodating model uncertainty when estimating risk. Although the use of MA is inherently appealing, little is known about its performance using general modeling conditions. We investigate the use of MA for estimating excess risk using a Monte Carlo simulation. Dichotomous response data are simulated under various assumed underlying dose–response curves, and nine dose–response models (from the USEPA Benchmark dose model suite) are fit to obtain both model specific and MA risk estimates. The benchmark dose estimates (BMDs) from the MA method, as well as estimates from other commonly selected models, e.g., best fitting model or the model resulting in the smallest BMD, are compared to the true benchmark dose value to better understand both bias and coverage behavior in the estimation procedure. The MA method has a small bias when estimating the BMD that is similar to the bias of BMD estimates derived from the assumed model. Further, when a broader range of models are included in the family of models considered in the MA process, the lower bound estimate provided coverage close to the nominal level, which is superior to the other strategies considered. This approach provides an alternative method for risk managers to estimate risk while incorporating model uncertainty.

An isobole-based statistical model and test for synergism/antagonism in binary mixture toxicity experiments

Synergism and antagonism are often defined in relation to the model of Concentration Addition (CA). Hence, it is vital for the conclusion of mixture toxicity studies to be able to test whether an observed deviation from CA reflects a true deviation or whether it is simply due to random variation. In this paper we consider a non-linear regression model for the classical ray designs for binary mixture experiments. The model combines dose–response curves for each mixture in the experiment with an isobole model, describing possible deviations from CA. The method allows us to test whether the chosen isobole model is reasonable for the data and to test the hypothesis of CA. Furthermore, it provides us with a measure of the degree of synergism/antagonism. The method is flexible since both the dose–response relationships and the isobole model can be chosen arbitrarily. We demonstrate the use of the method on datasets where combinations of pesticides are tested on a floating plant, Lemna minor, and an algae, Pseudokirchneriella subcapitata. Furthermore, we conduct a simulation study in order to explore the power with which a specific deviation from CA can be distinguished in different test-systems.     

Beta-normal distribution in dose–response modeling and risk assessment for quantitative responses

To establish allowable daily intakes for humans from animal bioassay experiments, benchmark doses corresponding to low levels of risk have been proposed to replace the no-observed-adverse-effect level for non-cancer endpoints. When the experimental outcomes are quantal, each animal can be classified with or without the disease. The proportion of affected animals is observed as a function of dose and calculation of the benchmark dose is relatively simple. For quantitative responses, on the other hand, one method is to convert the continuous data to quantal data and proceed with benchmark dose estimation. Another method which has found more popularity (Crump, Risk Anal 15:79–89; 1995) is to fit an appropriate dose–response model to the continuous data, and directly estimate the risk and benchmark doses. The normal distribution has often been used in the past as a dose–response model. However, for non-symmetric data, the normal distribution can lead to erroneous results. Here, we propose the use of the class of beta-normal distribution and demonstrate its application in risk assessment for quantitative responses. The most important feature of this class of distributions is its generality, encompassing a wide range of distributional shapes including the normal distribution as a special case. The properties of the model are briefly discussed and risk estimates are derived based on the asymptotic properties of the maximum likelihood estimates. An example is used for illustration.

Evaluation of the validity of the US EPA's cancer risk assessment of arsenic for low-level exposures: a likelihood ratio approach

Skin cancers associated with ingesting of arsenic have been documented since the 19th century. A study in the southwestern coastal area of Taiwan where people drank well water containing arsenic is generally recognised as providing the best data available for quantifying the risk, and the US Environmental Protection Agency (EPA) used these data to conduct a risk assessment of arsenic ingestion. However, the lowest exposure category in the Taiwan study included arsenic levels up to 290 µg L^–1, which is nearly six times higher than the current EPA maximum contaminant level (MCL), 50 µg L^–1. Therefore, the EPA risk assessment model extrapolated data on high-level exposures to generate risk estimates for low-level exposures. To evaluate the validity of this model, we conducted a quantitative review of epidemiological studies observing arsenic exposures below 290 µg L^–1. A ratio of the likelihood of the EPA model being inappropriate to that of it being appropriate was calculated for each study population as a measurement of the validity of the EPA model. Although existing human data on low-level exposures are limited, the review suggested that the EPA model is unlikely to be able to predict the risk of skin cancer accurately when the arsenic exposure level is between 170 and 270 µg L^–1.     

Immunity in children with exposure to environmental lead: II. Effects on humoral immunity

Lead has been found to depress the immune system in animal studies at levels far below those responsible for overt toxicity. Literature studies in animal systems most clearly showed an effect of lead on response to a specific immunogenic stimulus. Data are sparse concerning the effects of lead on the immune system in the human population at greatest risk for exposure-children up to six years of age. This portion of the Phase I study reports concentrations of IgG, IgM, IgA, and IgE, as well as antibody titers to the specific antigenic stimuli provided by the vaccines against diphtheria, tetanus, and Rubella. The study population consisted of a group of 193 children, ages 9 months to 6 years, who participate in the WIC (Women, Infants and Children) and Lead Poisoning Prevention Programs in the urban area of Springfield-Greene County Missouri. Blood lead levels ranged from 1 to 50 g dL^–1. Total Ig levels were determined and the data were analysed. No consistent significant differences were observed among the risk categories in the five age groups examined. A single Ig class in each of three age groups showed apparent significant differences among the various risk categories, but these differences were not correlated with blood lead. An analysis of specific antibody titers to diphtheria, tetanus, and Rubella was performed. Regression analyses of current data in Phase I of this study suggest a detrimental effect of lead on the antibody titres to diphtheria and Rubella.     

From data and theory to environmental models and indices formation

A prerequisite for environmental indices is that they represent environmental pressure, and the state of, and impact on environmental conditions. In other words, they should capture as much as possible of the cause-effect chains they represent and relate pressure and effect to criteria of environmental quality. The approach proposed in the article attempts to link the pressure–state–impact–response framework of indicators to the integrated environmental model, based on the method of response function (MRF). The MRF allows to construct purposeful, credible models from data and prior knowledge or information. The data are usually time series observations of system inputs and outputs, and sometimes of internal states. The output of such models is presented with highly aggregated environmental indices, reflecting the main pressure–state–impact–response cause-effect chains. The proposed approach is illustrated with the example of soil erosion indices.     

Statistical models of health risk due to microbial contamination of foods

Between 6 million and 33 million cases of food-related illness are estimated to occur in the United States each year, with about 5000 episodes resulting in death. Growing concerns about the safety of food prompted the National Food Safety Initiative of 1997, the goal of which is to reduce the incidence of illness caused by food-borne pathogens. A key component of the food safety initiative is the improvement of farm-to-table risk assessment capabilities, including the development of improved dose-response models for estimating risk. When sufficient data are available, allowable contamination levels of specific micro-organisms in food are established using dose-response models to predict risk at very low doses based on experimental data at much higher doses. This necessitates having reliable models for setting allowable exposures to food-borne pathogens. While only limited data on relatively few micro-organisms that occur in food are available at present for dose-response modeling and risk estimation, still none of the two-parameter models proposed so far, including the popular Beta-Poisson (BP) model, appears to be completely satisfactory for describing and fitting all of the present data (Holcomb et al., 1999). The Weibull–Gamma (WG) model is the only three-parameter model that has been proposed to date. In this paper, new competitive three-parameter models are derived, using a formulation that can be parameterized to represent statistical variation with respect to the dose of micro-organism received by the host and the hosts susceptibility to infection. Parameters of the models are estimated using the maximum likelihood method. Experimental data on several common microbial contaminants in food are used to illustrate the methodology.     

Comparing discriminant analysis, neural networks and logistic regression for predicting species distributions: a case study with a Himalayan river bird

We assessed the occurrence of a common river bird, the Plumbeous Redstart Rhyacornis fuliginosus, along 180 independent streams in the Indian and Nepali Himalaya. We then compared the performance of multiple discrimant analysis (MDA), logistic regression (LR) and artificial neural networks (ANN) in predicting this species’ presence or absence from 32 variables describing stream altitude, slope, habitat structure, chemistry and invertebrate abundance. Using the entire data (=training set) and a threshold for accepting presence in ANN and LR set to P≥0.5, ANN correctly classified marginally more cases (88%) than either LR (83%) or MDA (84%). Model performance was assessed from two methods of data partitioning. In a ‘leave-one-out’ approach, LR correctly predicted more cases (82%) than MDA (73%) or ANN (69%). However, in a holdout procedure, all the methods performed similarly (73–75%). All methods predicted true absence (i.e. specificity in holdout: 81–85%) better than true presence (i.e. sensitivity: 57–60%). These effects reflect species’ prevalence (=frequency of occurrence), but are seldom considered in distribution modelling. Despite occurring at only 36% of the sites, Plumbeous Redstarts are one of the most common Himalayan river birds, and problems will be greater with less common species. Both LR and ANN require an arbitrary threshold probability (often P=0.5) at which to accept species presence from model prediction. Simulations involving varied prevalence revealed that LR was particularly sensitive to threshold effects. ROC plots (received operating characteristic) were therefore used to compare model performance on test data at a range of thresholds; LR always outperformed ANN. This case study supports the need to test species’ distribution models with independent data, and to use a range of criteria in assessing model performance. ANN do not yet have major advantages over conventional multivariate methods for assessing bird distributions. LR and MDA were both more efficient in the use of computer time than ANN, and also more straightforward in providing testable hypotheses about environmental effects on occurrence. However, LR was apparently subject to chance significant effects from explanatory variables, emphasising the well-known risks of models based purely on correlative data.     

Current uses of the EPA lead model to assess health risk and action levels for soil

The EPA lead model predicts mean blood lead levels and risk of elevated blood lead levels in children based on lead uptake from multiple sources. In the latest model versions, environmental data from individual homes within a community can be used to predict the overall blood lead distribution and percent risk of exceeding a specific blood lead level (i.e. 10 g dl^–1). Recent criteria used by the EPA to evaluate this information include no more than 5% of houses with a greater than 5% lead risk, and a community weighted-average risk below 5%. Environmental (primarily soil) and blood lead data from a residential community near a smelter were used to illustrate recent uses of the model. Scheduled remediation in the community will remove soil for approximately 60% of the houses (i.e. those with lead levels > 1000 mg kg^–1). After remediation, the model results indicate a relatively low community risk (0.5–1.9%), although the percentage of houses with lead risks above 5% ranged from 3 to as high as 13%, depending on the variation in blood lead and assuming the model's 7 g dl^–1 increase in blood lead with each 1000 mg kg^–1 increase in soil lead level. A comparison of the limited blood lead data with soil lead levels below 1000 mg kg^–1, however, indicated no apparent relationship. Given these uncertainties, less invasive actions than additional soil removal (e.g. exposure intervention, monitoring conditions, and follow-up as necessary) may be appropriate under the new EPA guidance for lead in soil.     

Immunity in children with exposure to environmental lead: I. Effects on cell numbers and cell-mediated immunity

Studies conducted in animal systems have shown that lead is an immunosuppressive agent at levels far below those causing overt toxicity. Children less than six years of age are the population at highest risk for exposure to environmental lead; however little data were available to assess effects on the developing immune system in this age group. Reported here is the completed Phase I study on 193 children, ages 9 months to 6 years, with blood lead levels from 1 to 50 (g dL^–1, recruited from the urban population of Springfield-Greene County, Missouri, through their participation in the WIC (Women, Infants, and Children) and Lead Poisoning Prevention Programs. This portion of the study dealt with enumeration of cells involved with the immune response andin vitro mitogenic stimulation of lymphocytes. The percent lymphocytes, monocytes, granulocytes, T cells (total), B cells, CD4+T's and CD8+T's and CD4+/CD8+ ratios were determined and the data were analysed. No consistent significant differences were seen among the various risk categories currently identified by the CDC. Though two age groups showed some possible effect of lead, none of the various cellular parameters within these age groups showed significant correlation with blood lead. The lymphocyte response toin vitro mitogenic stimulus was studied on 42 children (including 17 in risk classifications IIA and higher) using the mitogens phytohemagglutinin (PHA), Concanavalin A (Con A), and Pokeweed mitogen (PWM). No consistent statistically significant differences were seen among the various risk categories; the effects of lead, if present, are most likely subtle and obscured by the interindividual and time-dependent variation inherent in this type of study.     

Environmental risk assessment of metals contaminated soils at silvermines abandoned mine site,Co Tipperary,Ireland

A centuries long history of mining and mineral processing has resulted in elevated Cd, Pb and Zn soil concentrations in the vicinity of the Silvermines abandoned mine site (AMS), Co. Tipperary, Ireland. A process for preliminary evaluation of environmental risk was developed and implemented. Potential pathways of metal compound transport and deposition were mapped and used to optimise the subsequent site investigation. Elevated soil metals are shown to be predominantly in areas where metal deposition in soil is associated with water related pathways (surface runoff, seasonal groundwater seepage and floodplains). Extensive areas of soil in the surrounding district are classified as contaminated on the basis of Cd, Pb and Zn concentrations, both total and potential bioavailable (EDTA-extractable). The most affected areas, with metal concentrations in soil comparable with that within the AMS, were floodplains located 2–3 km downstream from the site. Assessment of the sequential effects on grass and grazing animals indicates that Pb poses the greatest risk due to its high toxicity and high concentrations in soil (more than 10 000 mg kg^–1). Within floodplain areas grazing cattle may intake a lethal dose of Pb. On the basis of the investigation an approach to risk assessment was developed which allowed quantified assessment of the risks related to individual metals, areas of contamination and contamination targets.     

Water Quality Criteria for European Freshwater Fish

1. For water pollution control purposes, the concentration-addition model for describing the joint effects of mixtures of toxicants on aquatic organisms is appropriate; in this model the contribution of each component in the mixture is expressed as a proportion of the aqueous concentration producing a given response in a given time (e.g. p 96-h LC50).

2. Examination of available data using this model shows that for mixtures of toxicants found in sewage and industrial effluents, the joint acutely-lethal toxicity to fish and other aquatic organisms is close to that predicted, assuming simple addition of the proportional contribution from each toxicant. The observed median value for the joint effect of these toxicants on fish is 0.95 of that predicted, and the corresponding collective value for sewage effluents, river waters, and a few industrial wastes, based on the toxicity of their constituents, is 0.85, while that for pesticides is 1.3.

3. The less-than-predicted effect of commonly-occurring toxicants in some mixtures may be partly attributable to small fractions of their respective LC50 values having a less-than-additional effect. However, recent research has shown that for some organic chemicals which have a common quantitative structure-activity relationship (QSAR), their joint action as determined by acute toxicity is additive at all concentrations.

4. The few (unpublished) data available for the long-term lethal joint effect on fish of toxicants in mixtures suggest that they may be markedly more than additive, a phenomenon that needs confirmation and further investigation.

5. In the few studies on the sub-lethal effects on fish (eg growth), the joint effect of toxicants has been consistently less-than-additive which suggests that as concentrations of toxicants are reduced towards the levels of no effect, their potential for addition is also reduced. There appear to be no marked and consistent differences between the response of different species to mixtures of toxicants.

6. Field studies have shown that reasonably accurate toxicity predictions based on chemical analysis can be made if the waters which are polluted are acutely lethal to fish, and that a fish population of some kind can exist where the median 2 p t LCSOs (rainbow trout) is < 0.2. It is not known whether this condition is equivalent to a C p NOEC of 4.0 (ie the sum of the individual fractions of the NOEC for the species present), or to a NOEC of < 1.0 for each individual toxicant (i.e. fractions of the NOEC are not summed).

7. In general, the joint effect of the common toxicants on lethal and sub-lethal responses of fish is not explained by variations in the uptake of the individual toxicants concerned; this may not apply for those chemicals with a common QSAR, although there is little experimental evidence in this field.

8. There is an immediate need for more empirical studies on the joint effect of mixtures of toxic units of individual components, and the relation between long- and short-term lethal and non-lethal joint effects. This applies to mixtures of commonly occurring toxicants as well as to mixtures of organic chemicals with a common QSAR. The data obtained should be reinforced by studies on the mechanisms of interaction of toxicants. More field studies which relate water quality to the structure and productivity of fish populations are also required, involving direct measurements of fractional toxicity of the river water wherever possible.

9. The concentration-addition model appears to be adequate to describe the joint effect of commonly-occurring constituents of sewage and industrial wastes, and for tentative predictions of the joint effect on fish populations of toxicants present at concentrations higher than the EIFAC recommended values. However, concentrations lower than the EIFAC recommended values may make an increasingly lesser contribution to the toxicity of mixtures of toxicants and there may be a need to adjust the tentative water quality criteria downwards where two or more toxicants are present at concentrations close to these values. For toxicants with a common QSAR, their additive joint action may necessitate the setting of water quality criteriafor this group as a whole and not on the basis of individual compounds. However, too little is known of their precise joint action where the combined concentration produces a sub-lethal response.     

以REACH为数据源进行生命周期影响评估

由于潜在生态毒理数据的限制,生态周期影响评估中的毒性模型仅仅描述了市场上的一小部分物质。改进现有的LCIA数据情况可以通过发现新的数据来源,比如欧盟化学品注册、评估、授权和限制(REACH)数据库。本研究通过对比记录在REACH数据库和UNEP/SETAC的科学统一模型USEtox中相同化学物质的生态毒理数据来探究REACH是否具有作为数据来源的潜力。数据根据数据点的个数,报道可靠性和测试时间评估,并将每种化学物质的50%涵盖物种的危险浓度与USEtox中的数据做对比。结果强调了REACH和USEtox之间不同的数据可用性。REACH和USEtox的生态毒理学数据对比表明REACH是一个LCIA毒性鉴别的潜在生态毒理学数据来源,也显示出REACH标准的数据存在着一致性的问题,以及REACH中监管风险评估的假设可能与LCIA所需数据有出入。因此,在考虑REACH标准下的数据在LCIA的运用之前,数据质量,预处理和可运用性需要进一步的研究。探究其他可用的数据来源,发表的研究与报告也需要更深入的调查。
精选自Nienke Müller, Dick de Zwart, Michael Hauschild, Ga?l Kijko, Peter Fantke. Exploring REACH as potential data source for characterizing ecotoxicity in life cycle assessment. Environmental Toxicology and Chemistry: Volume 36, Issue 2, pages 492–500, July 2017. DOI: 10.1002/etc.3542
详情请见http://onlinelibrary.wiley.com/wol1/doi/10.1002/etc.3542/full
     

Diversity–stability relationships revisited: scaling rules for biological communities near equilibrium

The debate on diversity–stability relationships has a long history of theoretical interest and plays a central role in development of modern ecology. But such debate has recently re-opened under critical scrutiny both in theoretical and empirical studies. In this paper we use allometric (or energetic) scaling and statistical physics approaches to this problem. On the basis of local Damuth symmetry, a Markov model of transfer of energy between different species, and the fluctuation–dissipation theorem, scaling rules of species number and population variability of biological communities near equilibrium are derived. These scaling rules indicate that the diversity–stability relationship may be an energetic and thermodynamic consequence of ecological systems near equilibrium, not a simple statistical consequence as derived by other recent theoretical work.     

Modeling effects of toxin exposure in fish on long-term population size, with an application to selenium toxicity in bluegill (Lepomis macrochirus)

A primary goal in ecotoxicology is the prediction of population-level effects of contaminant exposure based on individual-level response. Assessment of toxicity at the population level has predominately focused on the population growth rate (PGR), but the PGR may not be a relevant toxicological endpoint for populations at equilibrium. Equilibrium population size may be a more meaningful endpoint than the PGR because a population with smaller equilibrium size is more susceptible to the negative effects of environmental variability. We address the individual-to-population extrapolation problem with modeling utilizing classical mathematical theory. We developed and analyzed a general model applicable to many freshwater fish species, that includes density-dependent juvenile survival and additional juvenile mortality due to toxicity exposure, and we quantified effect on equilibrium population size as a means of assessing toxicity. Individual-level effects are typically greater than population-level effects until the individual effect is large, due to compensatory density-dependent relationships. These effects are sensitive to the recruitment potential of a population, in particular the low-density first-year survival rate S_b. Assuming high S_b could result in underestimating effects of population-level toxicity. The equilibrium size depends directly on S_b, the reproductive potential, the toxin concentration at which mean mortality is 50% (LC₅₀), and the rate at which individual mortality increases with increasing toxin concentration. More experimental data are needed to decrease the uncertainty in estimating these parameters. We then used existing data for selenium toxicity in bluegill sunfish to parameterize a simulation version of the model as an example to assess the effects of environmental stochasticity on toxicity response. Effects of environmental variability resulted in simulated extinctions at much lower toxin concentrations than predicted deterministically.     

Unintended biological invasions: Does risk vary by trading partner?

International trade is the primary conduit for unintentional and damaging species introductions. But biogeographic heterogeneity, and differences in historical trade exposure across trade partners suggest that not all imports are equally risky. We develop an analytical model linking exotic species introductions and discoveries to trade volumes. The model is estimated using a novel historical data set on global trade and species introductions by region. Our estimates support theoretical predictions that trade from different regions poses different risks and that the cumulative number of introductions from a region is a concave function of imports. For each trade region we then calculate the marginal and cumulative invasion risk from additional trade. Simple volume restrictions on imports to reduce NIS introductions are not advisable based on coarse cost–benefit calculations.     

A Bayesian hierarchical model for monitoring harbor seal changes in Prince William Sound, Alaska

Bayesian hierarchical models were used to assess trends of harbor seals, Phoca vitulina richardsi, in Prince William Sound, Alaska, following the 1989 Exxon Valdez oil spill. Data consisted of 4–10 replicate observations per year at 25 sites over 10 years. We had multiple objectives, including estimating the effects of covariates on seal counts, and estimating trend and abundance, both per site and overall. We considered a Bayesian hierarchical model to meet our objectives. The model consists of a Poisson regression model for each site. For each observation the logarithm of the mean of the Poisson distribution was a linear model with the following factors: (1) intercept for each site and year, (2) time of year, (3) time of day, (4) time relative to low tide, and (5) tide height. The intercept for each site was then given a linear trend model for year. As part of the hierarchical model, parameters for each site were given a prior distribution to summarize overall effects. Results showed that at most sites, (1) trend is down; counts decreased yearly, (2) counts decrease throughout August, (3) counts decrease throughout the day, (4) counts are at a maximum very near to low tide, and (5) counts decrease as the height of the low tide increases; however, there was considerable variation among sites. To get overall trend we used a weighted average of the trend at each site, where the weights depended on the overall abundance of a site. Results indicate a 3.3% decrease per year over the time period.     

Altruism and environmental risks to health of parents and their children

This paper tests an equilibrium condition from a model that incorporates: (1) altruism of parents toward their young children, and (2) household production of latent health risks. The model demonstrates that an altruistic parent's marginal rate of substitution between an environmental health risk to herself and to her child is equal to the ratio of marginal risk reduction costs. Econometric estimates support this prediction based on data from a stated preference survey involving 488 parents of children aged 3–12 years. This outcome implies that parents reallocate family resources to at least partly offset the effectiveness of public programs that aim to reduce their children's environmental risks.     

生理毒代动力学模型及其在有机污染物水生生态毒理研究中的应用

人类生产和生活使用各种人工合成的化学品,种类和数量急剧增长,对生态系统和人体健康造成了极大威胁。因此,亟需采用高效的方法对数量巨大的化合物进行毒性评价。对生理毒代动力学(PBTK)模型的建立过程及其在污染物生态毒理研究中的应用进行了综述。PBTK模型,又称生理药代动力学(PBPK)模型,是利用生理学和解剖学等原理,将生物体简化为用血流连接的肝、肾和脂肪等各组织器官房室,模拟化合物在生物体内的吸收、分布、代谢和排泄过程。模型参数包括生理参数和生化参数2个部分,可用MATLAB等软件进行模拟。模型已应用于数百余种有机污染物在鱼体等水生生物体的毒代动力学模拟。已有模拟结果能够预测化合物在生物体内的有效剂量,对化合物毒性进行评估,并可用于不同物种、不同剂量和不同暴露途径间的外推,有力推进了污染物生态毒理研究工作的开展。