Congress Resources: Papers, posters and presentations

< Return to abstract list

Risk Assessment Of Mixture Toxicities In Natural Source Waters

Congress: 2015
Author(s): Nur H. Orak, Mitchell Small
Carnegie Mellon University1

Keyword(s): Sub-theme 2: Surface water and groundwater,
AbstractPoster Submission

Introduction

Scientists have been trying to define chemicals and their possible risks to humans and the environment for a long time. Today there are approximately 30,000 unique chemicals (>1 ton/year) used commercially in the U.S., and only a small part of this group is well characterized, that is, their toxicities are defined for regulatory actions (Judson et al. 2008). Currently, the laboratory methods used to characterize environmental risks for the majority of these chemicals can only measure toxicities above a certain concentration; additionally these methods are expensive and time consuming. The growing field of computational toxicology is successful in estimating the unknown toxicities of single chemicals. However, this method is often insufficient and variances for mixture toxicity prediction. Prior to characterizing interaction effects on model identification and parameter estimation, it is necessary to understand their implications in a predictive model of mixture toxicity for a range of potential exposure conditions and assumptions that could occur for human and ecological receptors: that is the purpose of this paper. This study reviews the potential risks of missing chemical data and concentration variability on mixture toxicity by developing 27 occurrence scenarios based on literature data. Random concentrations were simulated by the @RISK software assuming multivariate lognormal distributions for the mixture components. Systematic variations of the lognormal parameters were simulated to represent low to high variations in concentration medians.

Methods

This statistical model was built on the experimental work of Backhaus, Scholze and Grimme (1999) from which, ten different compounds were selected: conixacin, enoxacin, flumequine, lomefloxacin, nalidixic acid, norflozacin, ofloxacin, oxolinic acid, pipedimic acid, and piromidic acid. All of these chemicals belong to an important group of synthetic antibiotics called quinolones and these chemicals have been tested on marine bacterium Vibrio fischeri (Backhaus, Scholze and Grimme 1999). As a first step, concentration-response relationships were predicted for individual chemicals and the mixture of these chemicals. There was assumed to be no variability in concentrations: mean values were chosen for chemicals' concentrations. Twenty-seven (27) scenarios were generated in order to analyze the effect of concentration variability on toxicity. The effect of chemical omission was determined in order to understand the importance of individual toxicities on the total mixture toxicity. Using a similar process, the effect of concentration correlation and chemical omission on toxicity was analyzed. The multivariate lognormal distribution was chosen to describe the joint distribution of chemical concentrations at a site.

The mixture toxicities were calculated by independent joint action theory using single toxicities of compounds (E(ci) (Backhaus, Scholze and Grimme 1999).

Results and Discussion

The effect of variability in all components (E[Ci]; i=1,10) was tested by the generation of different sets of simulated concentrations using two alternative coefficients of variation (ν=1 and ν=3). Three of the ten compounds (representative of high, middle and low toxicity) are shown for graphical comparison: Ofloxacin, Oxolinic Acid and Pipemidic Acid. There is a rapid increase of toxicity with median concentration and with the amount of variability in the system. The biggest enhancement of toxicity due to variability occurs at lower concentrations, since the presence of variability allows for occasional high concentrations and very high associated toxicity. However if the coefficient of variation is very high, toxicity is slightly lower for high median concentration values because occasionally high and low concentration values occur in the system. In this high variability case (ν=3) estimates of total mixture toxicity based on median concentrations will significantly underestimate the expected toxicity.

Contrary to our original hypothesis, our initial findings show that correlation among component concentrations did not increase mixture toxicity.

The omission factor (OF) results suggest that the increasing individual component concentration (ν=0) does not change the behavior of individual toxicities' impact on the mixture toxicity. Omitting the most toxic compound (Ofloxacin) causes a notable difference in the mixture toxicity. The ν=1 scenario increases the estimated toxicity for the compounds more than the ν=3 scenario; this is because higher variation brings both lower and higher concentrations at the same time.

Conclusion

We found that higher variability in individual chemical concentrations can increase mixture toxicity by a significant amount, especially when this variability extends over the range of low to high toxicity for the chemicals considered. Variability in the system has a relatively small effect on scenarios in which compounds actually present are assumed absent and omitted from the toxicity calculation. As expected, underestimation of the mixture toxicity was most pronounced when the more-toxic compounds were omitted from the toxicity calculation. Contrary to our original hypothesis, our initial findings show that correlation among component concentrations did not increase mixture toxicity. These results suggest the need for rigorous characterization of ambient concentrations and exposures to mixtures when designing and interpreting toxicity studies. These findings have implications for regulatory actions and computational mixture toxicity assessment. There are several chemical types that we have a little information, but could have severe mixture toxic effects.

Key Words: Mixture toxicology, risk assessment, statistical model, chemical mixtures, computational modeling, synthetic antibiotics 1. Judson R, et al. (2008) ACToR--Aggregated Computational Toxicology Resource. Toxicology and applied pharmacology 233: 7-13

2. European Commission (2001) White Paper Strategy for a Future Chemical Policy. In: Communities CotE (ed). Brussels

3. Borgert CJ, et al. (2001) Evaluating Chemical Interaction Studies for Mixture Risk Assessment. Human and Ecological Risk Assessment 7: 259-306

4. Bliss CI (1939) The Toxicity of Poisons Applied Jointly Annals of Applied Biology 26: 585-615

5. Backhaus T, et al. (1999) The single substance and mixture toxicity of quinolones to the bioluminescent bacterium Vibrio fischer. Aquatic Toxicology.

6. USEPA (1997) Guiding Principles for Monte Carlo Analysis. In: Agency USEP (ed).

2011 IWRA - International Water Resources Association office@iwra.org - http://www.iwra.org - Admin