Jump to main content.

Assessment and Remediation of Contaminated Sediments (ARCS) Program

Monitoring Links

exit EPA [About PDF]

Table of Contents

Back To Index 

Assessment Guidance Document
Chapter 8

US Environmental Protection Agency. 1994. ARCS Assessment Guidance Document. EPA 905-B94-002. Chicago, Ill.: Great Lakes National Program Office.

Table Of Contents 


8. Fish Tumors and Abnormalities


The purpose of this chapter is to provide guidelines for performing a survey of resident fish for the presence of liver tumors (using the Ashtabula River AOC as a demonstration site) and to provide information useful for interpreting the data obtained in such surveys. Additional detailed information on the use of histopathological surveys for environmental assessments is provided by USEPA (1987b).

Role of Fish Tumor Surveys in Assessing Sediment Contamination

Laboratory toxicity tests (as discussed in Chapter 6) are effective methods for assessing the toxicity of contaminated bottom sediments. These tests measure changes in survival, growth, reproduction, or other endpoints and can be used in concert with chemical evaluations to evaluate the biological effects of sediment contamination.

Some biological effects that result from exposure to environmental contaminants take a long time to develop and cannot be evaluated using short-term toxicity tests. Carcinogenesis is a prime example. A long time is usually needed to develop pathologic lesions in tissue that can be identified as cancer. The methods available to determine this type of effect include laboratory toxicity tests with long exposure and grow-out times; however, these tests are costly and have been used with only a few species. Currently, the most effective method for assessing the potential carcinogenicity of contaminated sediments is to survey resident organisms, particularly species that are known to be sensitive to the development of cancer. However, because fish move, it is not possible to determine whether specific locations of contaminated sediments are responsible for the observed tumors or abnormalities. Although mutagenicity assays of sediment extracts also provide information on potential carcinogenicity, these assays cannot address questions of availability and do not recognize nongenotoxic carcinogens (Mac and Johnson 1989).

Use of Fish Tumor Surveys to Infer Cause-and-Effect Linkages

The role of contaminated sediments in inducing liver cancer in wild fishes has become better understood in the last decade. In a survey of hepatic neoplasms (i.e., liver tumors) in fishes from North America, 14 species from 41 geographic regions were found to have tumors that were related to environmental contamination (Harshbarger and Clark 1990). Most of these fishes were benthic-dwelling bottom feeders. In a study on the etiology of hepatic neoplasms in wild English sole (Parophrys vetulus) from Puget Sound, Washington, Myers et al. (1990) reported evidence for a link between exposure to sediment-associated contaminants (mainly PAHs) and the development of liver lesions, including neoplasms. Hepatic lesions that are suspected of being induced by contaminants have also been found in fishes from Great Lakes tributaries such as the Niagara River (New York) (Hickey et al. 1990); the Buffalo River (New York), Cuyahoga River (Ohio), and Black River (Ohio) (Baumann et al. 1991; Couch and Harshbarger 1985); the Detroit River (Michigan) (Kreis et al. 1989); and the Fox River (Illinois) (Brown et al. 1973, 1977); and from areas such as Torch Lake (Michigan) and Boston Harbor (Massachusetts) (Couch and Harshbarger 1985).

Liver tumors have been induced in fishes by exposure to contaminated sediments in the laboratory (Black 1983; Myers et al. 1990). Laboratory exposure to contaminants from sediments has also caused skin cancers and other hyperplastic abnormalities (Black 1982). Many of the parent chemical compounds found in the sediments are not necessarily the carcinogens found in the fish themselves because of metabolic transformations within the fish. Often, parent compounds (such as PAHs) are metabolized into carcinogenic metabolites that can form DNA-aromatic adducts. Adduct formation is indicative of the initiation phase of carcinogenesis (Dunn et al. 1987; Varanasi et al. 1987). Findings such as these help to substantiate a causal relationship between cancer and sediment contaminants.

Bottom-dwelling fishes are particularly susceptible to sediment-associated carcinogens by virtue of direct contact with the sediments and direct absorption through the skin or gills, or by exposure via dietary routes through ingestion of contaminated sediments and detritus or benthic invertebrates that have body burdens of carcinogenic compounds. In a study evaluating the neoplasms in bottom-dwelling flatfishes and highly migratory salmon from the same study area, a much higher prevalence of neoplasms was found in the flatfishes, presumably as a result of their direct exposure to contaminated sediments (Couch and Harshbarger 1985). Surveys of neoplasms in bottom-dwelling fishes are particularly effective in providing tangible evidence of damage to resident organisms from exposure to contaminated sediments.

Histopathology as a Sensitive Assessment Tool

While tumor surveys may be an effective means of determining damage to aquatic organisms from exposure to contaminated sediments, gross examination of fish for tumors during autopsies is not sufficient for accurately determining the prevalence of tumors. Although large nodules that are easily detectable to the naked eye are often tumors, histopathological examination is critical to determine the origin of the lesion and the type of neoplasm. Furthermore, some lesions and nodules have other etiologies such as parasitic infestations, which often resemble neoplastic nodules on gross examination. In addition, precancerous lesions or small neoplasms such as those seen in the liver are only detectable by microscopic examination of the liver. Also, the amount of liver tissue examined influences the chances of detecting lesions. Studies need to be conducted to determine the amount of tissue that should be examined from a liver to achieve an acceptable level of confidence. This information would enhance the sensitivity and interpretation of the surveys. In a study of brown bullheads (Ameirus nebulosus) taken from the Black River, Ohio, fish less than or equal to 2 years old had a prevalence of grossly observable neoplasms of 33 percent. Subsequent sampling of brown bullheads from the same area during the same year revealed an 80-percent prevalence of neoplasms when livers were examined histologically (Couch and Harshbarger 1985). Histopathological examination is the most sensitive tool for evaluation of tissue damage resulting from exposure to contaminated sediments.

Methods and Materials

This section describes recommended methods for fish collection, fish processing, and evaluation of tissue samples, as well as recommended QA/QC procedures for fish tumor surveys.

Fish Collection

Although numerous methods can be used to collect fishes for a tumor survey, it is critical that fish are collected alive with the least amount of physical damage. Electroshocking or trap netting meet these criteria. Electroshocking is preferable to trap netting because fish often sustain physical damage in trap nets. This is especially true with bullheads, because fighting between captured individuals leads to external wounds. However, in some areas electroshocking will be ineffective due to physical limitations (e.g., depth, water hardness). If destructive sampling gear such as gill nets must be used, frequent tending of the nets is necessary to minimize physical damage to the fish. Once fish are caught, they should be held in a manner that will keep them alive until processing. It is preferable to use a live well or other kind of tank for holding individuals prior to processing.

Fish Processing

Before beginning the external examination and general autopsy of captured fish, fish must be sacrificed using a humane method that minimizes trauma to the tissues. The preferred method is an overdose of an anesthetic. Fish are sacrificed individually to minimize any post-mortem tissue changes that may confound or interfere with histopathological analyses. Immediately following death, fish length and weight should be measured and a careful examination of the external body surface should be made. Abnormalities often associated with contaminated sediments, such as lip papillomas and stubbed barbels of bullheads, should be recorded. The skin should also be examined for any changes in thickness or coloration. Melanomas or skin tumors are common in fish from contaminated areas. Spines (from scaleless fish), scales, or otoliths should be taken to determine the age of the fish (Blouin and Hall 1990).

Upon completion of the external examination, a ventral incision on the midline running from the anus to just above the pectoral fins should be made, keeping the scalpel just under the skin so internal organs are not damaged. The incision should then be opened to expose the internal organs for examination, and to allow the excision of appropriate tissue samples for later histopathological examination. The liver should first be removed and weighed, and then grossly examined for the presence of any abnormalities such as swelling or nodules. If these abnormalities are present, tissue slices should be taken from the nodule, taking care to include "normal appearing" tissue in the same slice. If the liver appears normal, a 1-cm slice should be taken diagonally from top to bottom of the entire liver. Additional tissue samples can be taken at this time. Slices should not exceed 1 cm in thickness to allow for maximum penetration of the tissue fixative. Tissues should be placed in individual labeled jars with a volume of fixative at least 3 times the volume of the tissue. Labels should include a sample identification number reflecting the nature of the project, date, and individual fish number.

Various tissue fixation procedures can be used depending on the type of microscopic procedures used for histopathological analysis (Yevich and Barszcz 1981). For tumor surveys, the most common analytical procedure is fixation with neutral buffered formalin (10-percent solution) or Bouin's fluid (a picric acid-formalin mixture) and examination by light microscopy. It is critical that tissues be placed in the fixation medium as soon as possible after death of the fish to minimize post-mortem changes. Caution should be exercised when using picric acid (explosion potential) and formalin (carcinogen). After a fixation period of 24 hours, samples should be transferred to a 70-percent ethanol solution for storage until processing. Tissues should be processed using routine paraffin embedment procedures (Humason 1979), cut at 6 um, and stained with hematoxylin and eosin.

Evaluation of Tissue Samples

Histopathological evaluation of tissues is relatively subjective and agreement among pathologists is sometimes variable because of different interpretations about the fate of specific kinds of lesions. Because most of our knowledge about tumor progression is based on mammalian models, there is much speculation about the significance of similar lesions in fish. For tumor surveys, the primary goal is to determine the presence or absence of tumors, and to evaluate whether any observed tumors are cancerous or non-cancerous. The determination of the exact origin and morphologic description of the various lesions is less critical. Some researchers use morphologic classification schemes for diagnosis (Myers et al. 1987), while others may rely on two main categories of lesions: preneoplastic and neoplastic (Baumann et al. 1991). Preneoplastic lesions infer a potential for neoplasia, and neoplastic lesions indicate a cancerous condition. There are many kinds of lesions that fall under each category. Establishing specific criteria for lesion classification is critical because it assists other pathologists in understanding the exact nature of the diagnosis and facilitates validation.

When evaluating livers for tumors, it is generally too costly to microscopically examine entire livers. It is recommended that a sample be composed of at least four different tissue subsamples (four separate slides) from the same fish for histological examination. Each slide should be examined and all abnormalities recorded on a data sheet. Often, one section of a liver may exhibit more than one type of lesion or abnormality. All lesions should be recorded, even though the most serious lesion is considered to be the definitive diagnosis and is usually the one used in data analysis.

Quality Assurance and Quality Control

USEPA quality assurance policy stipulates that every monitoring and measurement project must have a written and approved QAPP. Guidance on preparation of a QAPP is presented in Chapter 2. The six primary topics covered in a QAPP include detection limit, bias, precision, representativeness, comparability, and completeness. The following sections discuss the latter five topics with regard to fish tumor surveys.

The bias associated with a tumor survey is controlled by sample size (number of fish examined histologically). Sample sizes should be large enough to be able to detect one tumor-bearing fish (at a 95-percent level of confidence) if the tumor prevalence in the population is at least 2 percent. Tumor prevalences <=2 percent have been estimated as the expected values for bullhead populations in unimpacted areas (Hartig and Mikol 1992). The minimum number of fish needed to achieve this criterion is 85 individuals from each individual sampling site. If fewer than 85 fish are examined, the level of confidence in sampling at least one tumor-bearing fish must be reduced accordingly.

Precision is a function of the histopathological diagnosis, which is a relatively subjective determination. Because there are no manuals for diagnosing fish tumors, acceptable criteria should be established. Precision can be evaluated by having the diagnoses for at least 10 percent of the slides validated by another pathologist. When different diagnoses occur, both pathologists should confer until they arrive at a common diagnosis. If disagreements occur for more than 25 percent of the samples checked, the original pathologist should reevaluate all samples after consultation with the other pathologist.

Criteria should also be established for determining the number of liver sections that should be evaluated to provide a representative assessment of the entire organ. Although there are currently no commonly accepted guidelines on this issue, four sections were used in the Ashtabula River survey and are recommended for similar surveys elsewhere. Although selection of the appropriate number of sections is somewhat arbitrary, four represents a balance between the need to assess a representative portion of the organ and the cost of evaluating individual sections.

Because fish move, it is not possible to determine how long they have resided in any one location. Representativeness is therefore difficult to demonstrate. Despite this uncertainty, fish should be collected from areas within, or areas believed to be representative of, the study area.

In studies involving multiple sample sites, tumor survey data should be obtained in the same manner at all sites. In addition, because of the subjectiveness of the histopathological diagnoses, it is preferable if one pathologist evaluates samples from all sites.

If the target number of fish is 85, a minimum of 50 fish (59 percent) should be considered adequate for estimating tumor prevalence, although this estimate will have a reduced confidence level. If 50 fish are not available, use of an alternative species should be evaluated.

The following section describes a case study of a tumor survey conducted in the Ashtabula River AOC under the ARCS Program.

The Ashtabula River AOC Tumor Survey

A total of 98 brown bullheads were collected by electroshocking in three areas of the Ashtabula River AOC--the harbor, breakwater, and river. The sample consisted of 40 males and 57 females (1 unknown sex) ranging in age from 3 to 7 years old. Attempts to collect sufficient numbers of fish for similar surveys in the Indiana Harbor and Saginaw River AOCs were unsuccessful.

External Abnormalities

External abnormalities such as skin discolorations, stubbed barbels, and lip papillomas were found in brown bullheads from all three areas. Skin discolorations, confirmed through histological examination to be pigmented nevi, were observed in 41 percent of the captured fish. Stubbed barbels, an abnormality thought to result from intimate contact with contaminated sediments, were found in 35 percent of the fish. Lip papillomas, common in fish from contaminated areas, were present in 16 percent of the fish.

Histological Findings

Liver lesions were generally classified as either "preneoplastic," inferring a potential for neoplasia; "neoplastic," indicating a cancerous condition; or non-neoplastic. In counting the number of fish with each class of lesions, the more severe class was the determinant. Hence, a fish with both neoplastic and preneoplastic lesions was counted in the neoplastic category to avoid double counting. Complete morphological descriptions and classifications were made for each liver sample. "Preneoplastic" lesions included areas of hepatocellular alteration, hepatocellular adenoma, and cholangioma, whereas "neoplastic" lesions included hepatocellular carcinomas and cholangiocarcinomas.

Approximately 20 percent (20 of 97 fish for which the sex could be determined) of the total sample of bullheads from the Ashtabula River AOC (all three areas combined) had preneoplastic liver lesions; most of these lesions were areas of hepatocellular alteration. Of the three areas sampled in the Ashtabula River system, the prevalence of preneoplastic lesions (as a percentage of the total catch from each area) was highest in the river (64 percent, 9 of 14 fish), followed by the breakwater (14 percent, 6 of 44 fish) and harbor (13 percent, 5 of 39 fish). Given the relatively small numbers of fish examined from each of these areas, such differences in prevalences may be indicative of trends but should not be considered statistically rigorous. Larger numbers of fish would need to be collected from each of the three areas (river, harbor, and breakwater) if statistical comparisons were to be made, especially if the prevalences had been much lower.

Neoplastic lesions were observed in four fish (4 percent of the total sample of bullheads from the Ashtabula River AOC). Three of these fish were from the river and one fish was from the breakwater. Three fish had hepatocellular carcinomas while the third individual had a mixed hepatocellular and cholangiohepatocellular carcinoma. All of these neoplastic lesions were considerably advanced in development.

Biological Correlations

All biological comparisons in this section were made using chi-square analysis of contingency tables.

Hepatosomatic Index
A hepatosomatic index (HSI) was derived for 82 of the fish by dividing the liver wet weight by the total body wet weight and multiplying by 100. (Livers of the remaining 14 fish were chemically analyzed and not weighed.) This index is often related to the presence of liver lesions because the density or mass of affected livers is often increased in response to preneoplastic or neoplastic conditions. In addition, enlarged livers can be caused by increased activity of mixed-function oxidase enzymes as a result of exposure to certain contaminants. In the Ashtabula River sample, the mean HSI (2.31) was significantly higher (P<=0.05) for fish with preneoplastic and neoplastic lesions relative to fish without these lesions (mean HSI of 1.85), and was significantly higher (P<=0.05) in females compared to males.

The bullheads collected ranged from 3 to 7 years old. Analysis of the prevalence of preneoplastic lesions for each age group showed a general trend of increasing prevalence with increasing age. In addition, neoplastic lesions were only found in older fish (i.e., ages 5-6 years old). This pattern is consistent with the hypothesis of chemical causation, in which a latent period between initiation and tumor development is expected (Baumann et al. 1990). Older fish have a longer period for exposure and development of lesions than do younger fish.

There was no correlation between most external abnormalities or liver lesions and sex. Approximately the same numbers of fish were collected for both sexes (i.e., 40 males and 57 females). There were no significant differences (P>0.05) in the prevalence of liver lesions, stubbed barbels, or skin discolorations between the sexes. However, there was a significantly higher (P<=0.05) prevalence of lip papillomas in males than in females.

Internal and External Abnormalities
There were no significant differences (P<=0.05) in the prevalence of external abnormalities between fish with and without liver lesions (either neoplastic or preneoplastic).

Contaminant Correlations

The suspected relationship between liver lesions in fish and contaminated sediments is based on four main lines of evidence:

Evidence for Chemical Etiology
Much evidence exists in the literature to support the hypothesis of chemical causation of liver lesions. Studies of wild English sole from Puget Sound have shown a relationship between PAHs in the sediments and elevated prevalences of hepatic lesions (Myers et al. 1990). These findings were further validated by laboratory studies in which PAH-enriched extracts from contaminated sediments were injected into English sole and induced lesions identical to those found in fish from the environment. Similar relationships have been reported by Baumann et al. (1991) for brown bullheads taken from the Cuyahoga River and the Black River (Baumann et al. 1990) and for flatfish from other areas of Puget Sound (Malins et al. 1984).

Presence of Contaminants in the Sediments
Sediment samples collected from the Ashtabula River AOC showed the presence of PAHs, PCBs, metals, and other chemicals (Ohio EPA 1991). It should be noted that the sediment samples were collected at a different time than the fish samples. When compared to USEPA guidelines for determining the extent of contamination, results of the sediment analyses indicate that four stations were characterized by low to moderate concentrations of metals, while PCB and PAH concentrations were lower than the concentrations in other areas (the Black and Cuyahoga Rivers) where epizootics of liver tumors in fish have been found (Ohio EPA 1991) (Table 8-1).

Bioaccumulation of Contaminants by Brown Bullhead
There is little information on bioaccumulation of contaminants in brown bullheads from the Ashtabula River because only one composite sample of 10 fish was analyzed, and few chemicals were analyzed. The only relevant concentration available is for total PCBs (0.7 ug/g) (Ohio EPA 1991). Most parent PAH compounds would not be expected to be found in fish tissue because they would be metabolized soon after uptake.

Tumor Prevalence
The prevalence of preneoplastic and neoplastic liver lesions in brown bullheads from the Ashtabula River AOC was 23 percent. This prevalence is much higher than would be expected from an unpolluted site such as the Huron River, Ohio, in which the prevalence was only 4 percent (Table 8-1). Compared to other areas in which there were unusually high incidences of tumors in bullheads, the prevalence observed for the Ashtabula River is not quite as high (Table 8-1). Within the Ashtabula River system, however, there was a significantly higher (P<=0.05) prevalence of liver tumors in samples taken from the river, compared to samples from the harbor or breakwater.


Limitations of Results

Using the fish tumor survey of the Ashtabula River AOC study as an example, it is evident that the brown bullhead population is suffering adverse effects as a result of some environmental factor(s). A fish tumor survey is one method of establishing a link between these demonstrable effects and possible causes. While conclusions have been drawn based on the available data, there are limitations of the results that must be considered. Although it is clear that the fish in this area have liver abnormalities, histological diagnoses are somewhat subjective and open to a variety of interpretations relative to the biological significance to the fish.

The two most relevant areas of uncertainty are in assigning a diagnosis and predicting an outcome based on the kind of lesion. Although pathologists may differ in the detailed morphologic description and identification of each lesion, there is relatively good agreement in assigning lesions to the more general categories of either cancerous or precancerous. These major categories of lesions, neoplastic and preneoplastic, imply irreparable damage with possibly fatal consequences. Therefore, for tumor surveys concerned with the presence or absence of injury, discrepancies in diagnoses are not likely to affect the overall evaluation of impact to fish populations. However, these surveys do not address the significance of the tumors to the fish (e.g., whether the presence of liver lesions affects the ability of a fish to reproduce and survive). Although the survey found fish with liver tumors, there appeared to be a stable resident population of reproducing fish. Very few older fish were found, suggesting that exposure to environmental factors may ultimately be lethal. To lend credence to this suggestion, it would be necessary to compare the age structure of the population with that in a reference area, but this was not done. Nevertheless, the presence of environmentally induced tumors in wild fish populations may serve as a warning system for the environmental health of all animals, including humans.

Another important factor to consider is the limits inherent in the sampling methods. While the condition of bottom-dwelling fish may be indicative of exposure to sediment contaminants, the extent of exposure is difficult to determine. Movement of the fish makes correlations with sediment contamination difficult. In many studies, sediment samples have been collected simultaneously with fish samples, and, in these cases, contaminant concentrations may be representative of actual exposure. However, many fish collections are made after sediment samples have been collected and an area is determined to have a sediment contamination problem. In the Ashtabula River survey, the observed prevalence of tumors is indicative of a highly contaminated area; however, supporting data on sediment chemistry are lacking. This may be a result of the fish visiting more contaminated areas or the collection of sediment samples from areas that were not representative of the entire AOC. Correlations of tumor prevalence and chemical contamination can be strengthened by conducting supporting laboratory studies that determine the direct effect of sediment extracts on fish by producing lesions identical to those observed in field-collected specimens (Myers et al. 1990). Injection of sediment extracts may be preferable to exposing fish directly to contaminated sediments because of the difficulties in holding fish in laboratory exposures for sufficiently long periods for contaminant uptake to occur. Additional evidence can be provided by conducting surveys several years after remediation of sediment contamination and determining whether tumor prevalence declined in response to remediation.

For the Ashtabula River survey, the apparent relationship between liver lesions in fish and contaminated sediments supports a hypothesis of chemical causation of the lesions. However, these correlations are not enough to prove chemical etiology. Instead, they provide a body of evidence that is consistent with, but not proof of, the hypothesis of chemical causation.


Improvements in Study Design
Statistical studies should be conducted to determine the amount of liver tissue that should be examined to ensure a high probability of detecting lesions. In the Ashtabula River survey, liver tissue that was removed for pathology included both routine tissue subsamples and any tissue that appeared abnormal. For histological examination, four slides were prepared for each liver. This number was selected based on the recommendations of other pathologists and cost constraints; however, no information is available on the number needed to ensure adequate confidence that the sections are representative of the entire liver.

Establishing Links Between Chemical Contamination and Tumors
Significant progress has been made during the last decade in determining strong correlations between some chemicals and certain tumors (particularly liver tumors). However, in assessment and remediation studies, the information available for a particular site is often inadequate to make definitive associations between tumor prevalences and specific chemicals. Laboratory studies in which wild fish are exposed to sediment extracts to induce lesions similar to those observed in the field are effective in providing the necessary evidence to make strong correlations between sediment contamination and tumors. Studies to determine the effect of exposure to chemical fractions of sediment extracts are also useful. Other types of studies that would provide similar information include cages positioned along a suspected sediment contamination gradient, laboratory testing using surrogate species that have a relatively short latency period for tumor development (e.g., Japanese medaka), and in vitro testing of sediment extracts for genotoxic and non-genotoxic potential.

Species Sensitivity
Studies have shown that certain species of bottom-dwelling fish are more sensitive to the effects of contaminated sediments than are other species. The nature of this selective sensitivity should be evaluated to determine if it is a function of the unique physiology of the sensitive species (e.g., immune system, nutrition, metabolic pathways, lifespan, behavior). This information would be critical for elucidating mechanisms of toxicity and for determining the appropriate remedial measures for reestablishing healthy, self-sustaining fish populations.

Species Movement
Inherent to tumor surveys that are coupled with evaluations of sediment contamination is the assumption that the fish observed with abnormalities are exposed to the contaminants at the site from which the fish were collected. Little information is available describing the movement patterns of many of these bottom-dwelling fish. Tagging studies to address the potential migratory behavior of these fish would improve the confidence in determining location and duration of contaminant exposure.

Post-Remediation Monitoring
Long-term monitoring is essential to any sediment remediation project. Tumor surveys are useful methods for assessing the effectiveness of remediation. Comparisons can be made between tumor prevalences in specific year classes of fish evaluated before and after sediment remediation. Because tumor prevalence should decline in the absence of sediment contamination, surveys conducted 2-3 years after remediation should reflect improved conditions or identify a recurring problem.

Manual on Liver Tumors
An atlas describing the various liver lesions would greatly expand the application of tumor surveys for assessment and remediation purposes. A manual standardizing the classification and interpretation of liver tumors would reduce the subjectivity of diagnoses and allow for preliminary screening of liver samples by less specialized personnel.


Local Navigation

Jump to main content.