Polychlorinated Biphenyls (PCBs) in Fish Roe


Polychlorinated biphenyls (PCBs) are a suite of 209 possible chlorinated structures, or congeners, having lipophilic properties. They are persistent pollutants that bioaccumulate in the environment and have been suspected to cause a variety of health effects in humans, including cancer. Humans may be exposed to PCBs through consumption of contaminated food items, such as fish, meat and dairy products. Previous studies have shown that fish may contain high PCB levels through exposure to these contaminants in water and ingestion of contaminated prey items. Additional studies suggest that female fish may shunt PCBs from their body via roe production. The objective of this study was to quantify the presence and levels of 110 PCB congeners in six roe samples commonly sold as caviar. Using gas chromatography with an electron capture detector (GC-ECD), total PCB concentrations ranged from 13-458 ng/g. The highest concentration was observed in farmed-raised paddlefish roe from the United States. A positive correlative (r2 equal to 0.58; p less than 0.005) between PCB levels and lipid content (for all samples except paddlefish) was observed suggesting fat content is a strong factor in predicting contaminant accumulation in roe. Results from these analyses were similar to a previous study quantifying six PCB congeners in European and Russian caviar. Although PCBs were detected in all roe samples analyzed in this study, concentrations fell below the U.S. Food and Drug Administration's consumption guideline of 2 ppm (2.000 ng/g wet weight) and would therefore be deemed safe to ingest.


Polychlorinated biphenyls (PCBs) are a group of synthetic organic chemicals that contain 209 possible individual chlorinated biphenyl compounds. These chemically related compounds are called congeners and vary in their physicochemical properties (Beyer 2002) and toxicity (Safe 1984). Previously, these compounds were used in hundreds of commercial and industrial applications due to their chemical stability, high heat capacity, low flammability, and insulating properties (Erickson 1997). These properties lead to the widespread use of PCBs in a variety of applications including dielectric fluids in transformers and capacitors, printing ink, paints, de-dusting agents, pesticides, hydraulic fluids, plasticizers, adhesives, fire retardants, and lubricants (Durfee et al. 1976). Due to human health concerns, the manufacture of PCBs was banned in the United States in 1976 (ATSDR 2001). Numerous studies have indicated adverse health effects from exposure to PCBs including cancer and effects on the cardiovascular, hepatic, immune, musculoskeletal, endocrine, gastrointestinal, reproductive and dermal systems (ATSDR 2001). Though PCBs were banned in the U.S. three decades ago, these health concerns and issues about persistence in the environment have made PCBs a notorious class of anthropogenic compounds that are still garnering much research attention.

Because PCBs are no longer manufactured or widely used today, there are relatively few ways that people can be exposed to concentrated PCBs. The two main routes of PCB exposure are by ingestion of contaminated food and/or water and by inhalation of contaminated air (EPA 2005). PCBs continue to enter the environment from landfills containing PCB waste materials and products, incineration of municipal refuse, sewage sludge, and though both legal and illegal use and disposal of PCB-containing materials (EPA 2005). Given that PCBs do not readily degrade in the environment after disposal and they are lipophilic (fat-soluble), they are environmentally persistent and tend to bioaccumulate in biota (Erickson 1997). Furthermore, PCBs may biomagnify through food chains. This process results in higher trophic level organisms having significantly more contaminant within their bodies than their prey items of lower trophic position (e.g., Danuta et al. 1997).

Through urban run-off, dumping and/or air transport, PCBs may enter water bodies and once there, tend to accumulate in the sediments (Ashley and Baker 1999). Additionally, PCBs accumulate in the fatty tissues of fish and other aquatic organisms. If humans ingest these items, this may represent a major pathway of PCB exposure (Cordle 1982; Birmingham et al. 1989).

PCB body burdens in wild animal populations are often related to age (exposure time), with the frequency of high levels being greater among older individuals (Norstrom et al. 1988; Aguilar et al. 1988). However, for some organisms, PCB concentrations are also related to gender because females of some species may eliminate these hydrophobic organic pollutants via the fetus, during roe (egg) production, and/or by lactation (e.g., Norstrom et al. 1988, Aguilar et al. 1988). For example, as observed by Niimi (1983), about 10-15% of the body weight of female fish consists of germinal tissue (relating to, or having the characteristics of a germ cell or early embryo) during the winter. Because germinal tissue contains 2.7-5% fat and muscle contains only 0.5-0.7%, the amount of fat remaining in the body after spawning is about equal to the fat that is eliminated in roe. Female fish may significantly eliminate PCBs during annual spawning events, producing eggs that contain about 5% of fat (on a wet weight basis). Thus large amounts of fat and persistent pollutants (over 10 times the remaining body burden) may be released in roe (Larsson et al. 1991).

Caviar, a popular worldwide delicacy, consists of female fish eggs that are high in lipid/fat. Most caviar is commonly obtained or often ingested by humans are from three types of sturgeon fish: beluga, ossetra, and sevruga, however roe from other species such as flying fish, paddlefish and salmon are popular as well. Roe is processed by passing it through a very fine mesh screen, which separates the eggs based on size. Salt is added to prevent freezing, as the caviar must be stored between 28 to 31 degrees Fahrenheit.

The U.S. Food and Drug Administration (U.S. FDA) has set a limit of 2 ppm for PCBs (two parts of PCBs per million parts of fish tissue). Any fish and fish products that are above this limit are deemed unsafe for human consumption. We hypothesized that because fish roe that is harvested for caviar are produced by large female fish, often eating at or near the top of the aquatic food chains, the presence of PCBs may be high. This study presents the results from PCB analyses of several roe samples from various sources. The objectives of this study were to quantify the levels of PCBs in caviar samples, compare these values to those previously reported in the literature, and identify the factors responsible for accumulation of PCBs.

Materials and Methods

The caviar samples analyzed in this study were obtained from two purveyors in Philadelphia, PA and were chosen based on their popularity by those who eat caviar. Flying fish and salmon roe were obtained from a local Asian restaurant specializing in sushi and sashimi. At an upscale epicurean boutique offering a wide range of roe from different countries, four additional roe samples were purchased: salmon roe, American paddlefish, and sturgeon (Russian sevruga and Uruguayan ossetra). Since two different salmon roe samples were analyzed, the one purchased at the Asian restaurant was designated "salmon (sushi)" and the latter sample designated "salmon (caviar)".

To assess analytical precision, the sturgeon roe sample was analyzed in duplication. To assess accuracy in the determination of PCB concentrations, a standard reference material (SRM 1974b – Organic in Mussel Tissue), purchased from the National Institute of Standards and Technology (NIST) was analyzed using the same analytical procedure as the roe samples. Upon quantification and calculation of the final concentrations of PCBs in the SRM, comparison to the published values reported by NIST was made and accuracy judged.

The preparation of roe samples for PCB followed previously published methods for biotic samples (Ashley et al. 2000). One gram of each caviar sample was mixed with a 1:10 ratio of sodium sulfate, a drying agent used to eliminate water. After the mixtures were completely dried, they were pulverized using a mortar and pestle. The dried samples were then extracted for 24 hours using a Soxhlet apparatus containing 200 mL of dichloromethane. Prior to initiation of extraction, 0.100 mL of a surrogate solution, containing known concentrations (350 ng/mL) of three PCB congeners (IUPAC numbers: 14, 65, and 166), was added to each sample. These three PCB congeners were never industrially produced and therefore are not constituents of any environmental sample. Addition of these three congeners allows for the assessment of analytical loss of PCBs through the analytical preparation of samples. To reduce the solvent volume upon termination of 24-hour extraction, a roto-evaporator was used. The volume was reduced to 3-4 mL and the process of solvent evaporation was repeated two more times after addition of a small volume of hexane each time. Following the third hexane rinse, the samples were transferred to centrifuge tubes and hexane was added until the final volume was 10.0 mL in preparation for gravimetric lipid analysis.

Figure 1. As part of the quality assurance plan of this project, a standard reference material (SRM 1974B) was analyzed. Total PCB concentrations for the evaluated SRM were compared to those reported by NIST.  Error bars for the NIST values represent standard deviations of the mean for replicate analyses.

Figure 1. As part of the quality assurance plan of this project, a standard reference material (SRM 1974B) was analyzed. Total PCB concentrations for the evaluated SRM were compared to those reported by NIST. Error bars for the NIST values represent standard deviations of the mean for replicate analyses.

For lipid determination, 1.00 mL of the extract was removed and placed in a pre-tarred aluminum tray. Hexane was allowed to evaporate over 24 hours. The percent lipid was calculated using the mass of the residue divided by the original weight of sample extracted and multiplying by 100 to express lipid content on a percent basis.

Meanwhile, the remaining 9 mL of extract were reduced in volume (1-2 mL) and further concentrated by placing the tubes under a stream of N2 gas. Concentrated sulfuric acid (1 mL) was added to each sample to hydrolyze the lipids because they would interfere with the instrumental detection of PCBs. Upon full separation (24 hours), the top organic layer was removed and placed in a clean test tube. The remaining bottom layer was rinsed two more times with hexane and the washes were added to the organic layer collection. The aqueous phase was discarded and the organic phase was reduced in volume to ca. 1 mL using N2 gas. The remaining 1 mL of extracted sample was further cleaned using liquid-solid chromatography with florisil (8 g) as the stationationary phase. Approximately 35 mL of hexane was used as the mobile phase. The mobile phase was collected in a clean 250 mL round bottom flask which were then transferred to the roto-evaporation system for solvent reduction. The samples were reduced to only a few milliliters and placed clean vials. At this point, an internal standard solution containing 2,3,6-trichlorobiphenyl (congener 30) and 2,2',3,4,4',5,6,6'-octachlorobiphenyl (congener 204) was added to all of the samples.

Using Agilent 6890 (Palo Alto, CA, USA) gas chromatography equipped with a 63Ni electron capture detector and a 5% phenyl- methyl silicon capillary column, 110 PCB congeners were analyzed to determine the total amount of PCBs present in each sample. A 60 m long column having a 0.25 mm internal diameter with a 250 nm stationary phase film thickness (DB-5, J&W Scientific, Folsom, CA, USA) was used. Hydrogen and argon/methane were used as the carrier and make-up gases, respectively (flow rates = 30 ml/min), and the inlet pressure was 100 kPa. An optimized temperature program was used to elute all congeners with an approximately 80 minute run per sample. The oven was set at 100˚C for 2 min, ramped from 100 to 170˚C at 4˚C/min, 170 to 280˚C at 3˚C/min and finally held at 280˚C for the last 5 minutes. The injector and detector temperatures were set at 225 and 285˚C, respectively. A 0.001 mL sample was injected in splitless injection mode using an auto sampler and the data was acquired by a computer operating Chemstation software (Hewlett-Packard). The method outlined by Swackhamer (1987) was used to identify and quantify the PCB congeners. In this, the identities and concentrations of each congener in a mixed Aroclor standard (25:18:18 mixture of Aroclors 1232, 1348, and 1262) were determined by calibration with individual PCB congener standards. Based on chromatographic retention time, congener identities in the sample were determined relative to the internal standards added.

A matrix blank was generated to monitor possible laboratory contamination and to calculate the detection limits for PCBs. The matrix blank, consisting of approximately 30 g of clean Na2SO4, was analyzed using the same procedures as the samples.


Quality Assurance/Quality Control

Average recoveries of congeners 14, 65 and 166 were 67 +/- 7%, 68 +/- 7% and 71 +/- 7%. Due to the relatively high surrogate recoveries and the low standard deviations, all reported values for PCB concentrations, except for the SRM comparison, were not corrected for analyte loss The detection limits for total PCBs (6 ng), defined as the sum of all quantified congeners, was calculated as three times the mass determined in the matrix blank. The matrix blank chromatogram was void of significant peaks suggesting that little contamination through laboratory exposure occurred.

Figure 2. Total PCB concentrations for each roe sample are shown on a ng/g wet weight basis (top) and on a lipid normalized basis (bottom). Under both the wet weight and lipid normalize basis, American paddle fish roe contained the highest levels of PCBs.

Figure 2. Total PCB concentrations for each roe sample are shown on a ng/g wet weight basis (top) and on a lipid normalized basis (bottom). Under both the wet weight and lipid normalize basis, American paddle fish roe contained the highest levels of PCBs.

Standard reference material 1974b (Organics in Mussel Tissue) was used to determine analytical accuracy. Our values were corrected for analyte loss using the average surrogate recovery for this sample (72%). Although we quantified 110 congeners in the SRM; NIST only reports values for 20 singly or co-eluting congeners (Figure 1). Our determined values were within analytically acceptable ranges of the true (NIST) values (Figure 1).

To evaluate our precision, a duplicate of the Uruguayan ossetra sturgeon was analyzed. Total PCB values for the duplicate analyses were 81.6 ng/g and 83.0 ng/g. The relative percent difference (defined as the range of the duplicates divided by the mean, expressed as a percentage) was very low (less than 2%) denoting a very high level of analytical precision.

PCB Levels in Roe

Total PCB concentrations were determined on an ng/g wet weight basis for the six samples of roe. The concentrations ranged from 13 to 458 ng/g wet weight (Figure 2). The highest concentration was observed in farmed-raised paddlefish roe. Total PCBs were also normalized to the amount of fat in the roe samples. The concentrations ranged from 169 to 4,082 ng/g lipid (Figure 2). A linear correlation (r2 equal to 0.58; p less than 0.005) existed between the percent lipid of caviar samples and total PCBs (Figure 3). The paddlefish sample was excluded from this treatment because it was calculated to be an outlier (Q Test: Q equal to 0.849; p less than 0.005), having a high level of PCBs (458 ng/g) but relatively low lipid content (11.2%).

Figure 3. Total PCBs were compared to the lipid percent of the caviar (paddlefish value not shown). The positive linear regression line suggests that lipid content of roe is a large factor in determining the PCB concentration.

Figure 3. Total PCBs were compared to the lipid percent of the caviar (paddlefish value not shown). The positive linear regression line suggests that lipid content of roe is a large factor in determining the PCB concentration.


PCBs have "octanol-water partition coefficients" (Kow) that range from 3.76 to 8.26 (EPA 1999), indicating they are very lipophilic (fat-loving) molecules. Concentrations of PCBs and other lipophilic contaminants are often related to the lipid content of the tissue being analyzed (EPA 1995). Barring the paddlefish sample for this treatment, a relatively high degree of correlation (r2 equal to 0.5814) with lipid content was observed (Figure 3). This suggests that fat content of roe is a major determining factor in the concentrations of PCBs present. Other studies have found that PCB body burdens in fish increase with increasing lipid content (e.g., Amstrong and Sloan 1988; Kidd et al. 1998; Rasmussen 1993).

Paddlefish (Polydon spathula) roe (t-PCB concentration of 457 ng/g), did not follow the expected PCB-lipid trend. We suggest that paddlefish may be accumulating PCBs by exposure to environmental factors such as contaminated water or groundwater, contaminated food sources, and possibly from their reproductive processes.

The elevated PCB value for paddlefish roe may be due to the presence of PCBs in the farm ponds and/or from food fed to these fish during farming. Recently, Hites et al. (2004) found that farm raised salmon had higher concentrations compared to natural populations. They suggested that contaminated fish food might be the vector supplying PCBs to these fish. This may be the process explaining the elevated levels observed in the paddlefish sample, whose source is a farm in the southeastern, US. In a natural setting, American paddlefish filter feed on large zooplankton (e.g., Daphnia sp.) and insect larvae; however, when the species is farmed it is often trained to consume trout/fish crumbles (ATSDR 1997; SAC 2000). This alteration in food source could be leading to elevated PCB levels.

Paddlefish require very specific environmental conditions to ensure successful reproduction, which include: photoperiod, water temperature, and water flow (Jennings 2000). One study (SAC 2000) suggests that when these requirements are not met, female paddlefish will resorb their eggs. By resorbing their eggs, paddlefish may be greatly increasing their PCB concentrations since they only reproduce every four to seven years in a wild setting (Jennings 2000). This resorption of roe would suggest that female paddlefish do not shunt PCB concentrations annually like previous studies with other fish species have suggested (e.g., Larsson et al. 1993). This notion is further supported by a study that stated fish with high fat content that did not spawn yearly had increased concentrations of persistent pollutants (Hites et al. 2004).

Our PCB values were compared to the only known published values of PCBs in fish roe (Krüger and Pudenz 2002) using "total PCBs" defined as the sum of only the following congeners: 28, 52, 101, 138, 153, and 180 (Table 1). That study, which evaluated caviar from European and Russian sources, found values ranging from 14-388 ng/g while this study observed a range from 2-157 ng/g wet weight. There is little difference between all the geographically dispersed roe samples, which may suggest a similar "global" source of PCBs to all these fish rather than a local one.

Table 1. Comparison of the sum of PCB congeners 28, 52, 101, 138, 153, and 180 in purchased caviar purchased from U.S. suppliers (this study) to those reported by Krüger and Pudenz* (2002).

Table 1. Comparison of the sum of PCB congeners 28, 52, 101, 138, 153, and 180 in purchased caviar purchased from U.S. suppliers (this study) to those reported by Krüger and Pudenz* (2002).

While this study quantified PCB levels that all fell below the U.S. Food and Drug Administration's consumption guideline of 2,000 ng/g (2 ppm) per serving, there may be concern for roe produced by farmed paddlefish. Paddlefish roe PCB concentrations could be high enough to cause embryotic development problems (Johnston et al. 2002) yet toxicological studies would need to be completed to evaluate this.

Future studies should further investigate the high concentration of PCBs that were found in the farmed Paddlefish roe. While the levels found are considered safe, they were exceedingly high compared to the other samples PCB concentrations in this study. More replicates should be assessed to ensure accurate ranges for the concentration of PCBs in fish roe.


Aguilar, A and A. Borrell. (1988). Age and sex related changes in organochlorine compound levels in fin whales (Balaenoptera physalus) from the eastern North Atlantic. Mat. Environ. Res. 25:195-211.

Armstrong, R.W. and R. J. Sloan. (1988). PCB Patterns in Hudson River Fish: I. Resident Freshwater Species Fisheries Research in the Hudson River. State University of New York Press Albany. 304-324.

Ashley, J.T.F., J.E. Baker, E. Zlokovitz, D. Secor, and S. Wales. (2000). Linking habitat use of Hudson River striped bass to accumulation of PCB congeners. Env. Sci. Technol. 34: 1023-1029.

Ashley, J.T.F., and J. E. Baker. (1999). Hydrophobic organic contaminants in surficial sediments of Baltimore Harbor: Inventories and sources. Environ. Tox. Chem. 18: 838-849.

ATSDR. Polychlorinated Biphenyls (PCBs), CAS# 1336-36-3, September 1997.

ATSDR. Public Health Implications of Exposure to Polychlorinated Biphenyls (PCBs). June 20, 2001.

Beyer, A., F. Wania, T. Gouin, D. Mackay, and M. Matthies. (2002). Selecting internally consistent physicochemical properties of organic compounds. Environmental Toxicology and Chemistry. 21: 941–953.

Birmingham, B., A. Gilman, D. Grant, J. Salminen, M. Boddington, B. Thorpe, I. Wile, P. Tofe, and V. Armstrong (1989). PCDD/PCDF multimedia exposure analysis for the Canadian population: detailed exposure estimation. Chemosphere. 19: 637-42.

Cordle, F., R. Locke, and J. Springer. (1982). Risk assessment in a federal regulatory agency: an assessment of risk associated with the human consumption of some species of fish contaminated with polychlorinated biphenyls (PCBs). Environ Health Perspect. 45:171-82.

Danuta T. Z., R.W. Griffiths, and N.K. Kaushik. (1997). Biomagnification of polychlorinated biphenyls through a riverine food web. Environmental Toxicology and Chemistry. 16:1463–1471.

Durfee R.L., G. Contos, F.C. Whitmore, J.D. Barden, E.E. Hackman, III, and R.A. Westin RA (1976). "PCBs in the United States – Industrial Use and Environmental Distributions." USEPA, EPA 560/6-76-005, NTIS No. PB-252012.

EPA. (1995). National Study of Chemical Residues in Fish. Volume I. Office of Science and Technology, Washington, DC. EPA 823-R-92-008a.

EPA. (1999). Guidance for Assessing Chemical Contaminant Data for Use in Fish Advisories. Risk Assessment and Fish Consumption Limits. Office of Water.

EPA. (2005). Technical Factsheet on: Polychlorinated Biphenyls (PCBs). U.S Environmental Protection Agency Website. 18 Nov 2005.

Erickson, M. D. (1997). Analytical Chemistry of PCBs- Second Edition. Florida: Lewis Publishing.

Hites, R., J.A. Foran, D.O. Carpenter, M.C. Hamilton, B.A. Knuth, and S.J. Schwager. (2004). Global Assessment of Organic Contaminants in Farmed Salmon. Science 303: 226-229.

Jennings C., and S.J. Zigler (2000). Ecology and biology of paddlefish in North America: historical perspectives, management approaches, and research priorities. Rev. in Fish Biology and Fisheries. 10:167-187.

Kidd, K., D.W. Schindler, R.H. Hesslein, and D.C.G. Muir. (1998). Effects of trophic position and lipid on organochlorine concentrations in fishes from subarctic lakes in Yukon Territory. Can. J. Fish. Aquat. Sci./J. Can. Sci. Halieut. Aquat. 55:869-881.

Krüger, A., and S. Pundenz (2002). Chlorinated Hydrocarbon Pollution in Caviar Samples. Inter. Rev. of Hydrobio. 87:637-644.

Larsson, P., L. Okla and S.F. Hamrin. (1991). Factors determining the uptake of persistent pollutants in an eel population (Anguilla anguilla). Environ. Pollut. 69:39-50.

Larsson, P; L. Okla., and L. Collvin. (1993). Reproductive status and lipid content as factors in PCB, DDT and HCH contamination of a population of pike (Esox lucius L.). Environmental Toxicology and Chemistry. Vol 12. 5:855-861.

Niimi, A. J. (1983). Biological and toxicological effects of environmental contaminants in fish and their eggs. Can. J. Fish. Aquat. Sci. 40:306-312.

Norstrom, R.J., D.C.G. Muir, and M. Simon. (1988). Organochlorine contaminants in arctic marine food chains: Accumulation of specific polychlorinated biphenlys and chlordane- related compounds. Environ. Sci. Technol. 22:1071-1079.

Rasmussen, J.B., D.J. Rowan, D.R.S. Lean and J.H. Carey. (1990). Food chain structure in Ontario lake determines PCB levels in lake trout (Salvelinus namaycush) and other pelagic fish. Canadian Journal of Fisheries and Aquatic Sciences. Vol 47. 10:2030-2038.

Safe S. (1984). Polychlorinated biphenyls (PCBs) and polybrominated biphenyls (PBBs): biochemistry, toxicology, and mechanism of action. Crit Rev Toxicol. 13:319-95.

Swackhamer, D.L. (1987). Quality Assurance Plan for Green Bay Mass Balance Study - PCBs and Dieldrin. U.S. Environmental Protection Agency, Great Lakes National Program Office.

UK Marine Special Areas of Conservation (SAC) Project. "PCBs" (2000).

Do you want to receive updates about JYI's recent publications, events, and staff openings? Sign up for our email newsletter.
Follow Us
For all the latest news from JYI, join our Facebook.
For all the latest news from JYI, join our Youtube.
For all the latest news from JYI, join our twitter.
For all the latest news from JYI, join our email list.