Page 922 - The Toxicology of Fishes

902 The Toxicology of Fishes

TABLE 22.1
Associations between PAH Concentrations in Sediments, Stomach Contents, and Bile with
Site-Specific Prevalences of Selected Categories of Idiopathic Liver Lesions in English Sole,
as Determined by Logistic Regression (p < 0.05) a
Chemical Class/ p-Value/% Total Variance
Compartment n Neoplasms FCA SDN Prolif Necrosis

ΣLAHs
Sediment 22 0.020/54 0.001/52 0.001/45 0.001/68 NS
Stomach contents 9 NS 0.001/65 0.001/75 0.001/73 NS
Bile (FACs-L) 20 0.032/15 0.001/26 0.001/35 0.002/21 NS
ΣHAHs
Sediment 22 0.004/48 0.001/40 0.001/38 0.003/35 NS
Stomach contents 9 NS 0.001/65 0.001/75 0.001/73 NS
Bile (FACs-H) 20 NS 0.01/13 0.001/22 0.001/22 NS
a Analyses were performed while adjusting for mean age and sex ratio (female:male). Table indicates p-value,
and percent of total variance in lesions prevalence explained by risk factor (reduction in scaled deviance).
Abbreviations: FACs-H, aromatic compounds fluorescing at benzo(a)pyrene wavelengths; FACs-L, aromatic
compounds fluorescing at naphthalene wavelengths; FCA, foci of cellular alteration; HAHs, high-molecular-
weight polycyclic aromatic hydrocarbons; LAHs, low-molecular-weight polycyclic aromatic hydrocarbons;
n, number of sites; NS, not significant; Prolif, proliferative lesions; SDN, specific degenerative/necrotic lesions.
Source: Adapted from Myers, M.S. et al., Environ. Health Perspect., 102, 200–215, 1994.

Risk Characterization

Ecological risk assessment and characterization are critical components of our research program on the
effects of PAHs on marine and estuarine fish. Suter (1993) defined risk assessment as “the process of
assigning magnitudes and probabilities to adverse effects of human activities or natural catastrophes.”
This process involves identifying hazards, such as releases of toxic chemicals to surface waters that
support fisheries, and uses measurement, testing, and mathematical or statistical models to quantify the
relationship between perceived hazards and subsequent adverse health effects. The standard paradigm
for ecological risk assessment (NRC, 1983) includes four major steps: (1) hazard definition, which
involves the choice of endpoints for the assessment (e.g., mortality, cancer risk, reductions in fecundity)
and the target species of interest, environmental description, and source terms; (2) exposure assessment
and effects assessment; (3) risk characterization; and (4) risk management.
For the case study illustrated in this chapter, PAHs in the marine environment are the hazard of interest,
the primary target species is English sole, and exposure and effects assessment data have already been
presented for several endpoints, including hepatic lesions, reproductive dysfunction, and growth. This
section on risk characterization describes our efforts to use these data to estimate the types of adverse
health effects that would be expected to result from a given environmental exposure to PAHs. Uncer-
tainties in the analysis are discussed, but risk management options are not examined in detail.
The analytical methods we used in the effects assessment and risk characterization steps are primarily
logistic regression techniques. These methods calculate the risk of an adverse health effect (e.g., the risk
of cancer) associated with a given PAH exposure in an individual fish or fish subpopulation (e.g., fish
collected from a particular sampling site). We have used this approach in several studies with English
sole and other Pacific Coast bottomfish to statistically relate lesion prevalences to biological risk factors
and measures of contaminant exposure (Johnson et al., 1993; Myers et al., 1994; Stehr et al., 1997,
1998, 2004). Through stepwise logistic regression techniques, it is possible to determine the influence
of contaminant exposure-related risk factors on hepatic disease risk while simultaneously accounting
for the influence of biological risk factors such as age and sex. This technique is commonly used on
binomial or proportional data in epidemiological and epizootiological studies (Breslow and Day, 1980).