Using the Spearman rank correlation coefficient procedure, ve found a sig¬ 
nificant (p=0.003) positive correlation (rg=0.48) between the prevalence 
of hepatic neoplasms in English sole and sediment concentrations of AHs 
(1st grcup) (Malins et al. 1984). Moreover, in a recently completed 
summary analysis of the data from the original study of 40 stations 
combined with data from subsequent studies of 30 stations (e.g., stations 
in Eagle Harbor and near Mikilteo), a significant (p=0.0001) positive 
correlation between concentrations of sediment-associated AHs and the 
prevalence of hepatic neoplasms was found using Spearman rank 
correlation (Table 1). 
We also performed a detailed analysis of the results of six field studies 
using logistic regression (Malins et al. 1987). This method permits the 
construction of a series of multivariate statistical nodels relating the 
prevalences of particular categories of lesions to the combined levels of 
several different categories of sediment-associated contaminants. Speci¬ 
fically, logistic regression was used to assess the relationships between 
neoplasm prevalence, fish length, and sediment concentrations of three 
classes of chemical compounds (AHs, PCBs, and CBDs). The best logistic 
regression model for neoplasnus prevalence accounted for 40.4% of the 
variation in neoplasm prevalence among the 59 collections of fish at the 
46 stations sampled (Malins et al. 1987) (Figure 12). Neoplasm preva¬ 
lences v^re positively correlated with sediment concentrations of both 
AHs (p£0.00001) and PCBs (p^O.OOOl), but negatively correlated with 
sediment concentrations of CBDs (p£0.05). ^feoplasm prevalences were also 
n^atively correlated with the interaction term for AHs and PCBs (p£0.0001). 
Fish size range exerted an important effect (p^O.OOOOl), with observed 
neoplasm prevalence being greater when only large fish were collected 
than when the entire size range was employed. 
Bile metabolite concentrations measured (at the BaP wavelength pair) in 
English sole from 11 sites in Puget Sound were also compared statistically 
(Spearman rank correlation) to the prevalences of certain types of hepatic 
lesions, including neoplasms in these fish (Table 2) (Krahn et al, 1986). 
A significant (p<0.002) positive correlation (rg=0.85) was found. These 
results provide supportive evidence for the putative relationship betveen 
the arcmatic compounds found in the environment and serious liver diseases 
in bottom-dv^lling fish such as the hepatic neoplasm. 
Studies conducted by our laboaratory yielded results suggesting that the 
presence of certain liver lesions in English sole is associated with im¬ 
paired organ function. Casillas et al. (1985) found an association be¬ 
tween certain liver lesions and abnormal values of several serum chemistry 
parameters characteristic of liver dysfunction and/or damage (Table 3). 
For example, sole with neoplasms had significantly (p£0.05) higher serum 
concentrations of bilirubin, and significantly lower concentrations of 
albumin and calcium. 
99 
