Dioxins, polychlorinated biphenyls, methyl mercury and omega-3 polyunsaturated fatty acids as biomarkers of fish consumption

Article metrics

Abstract

Background/Objectives:

To assess biomarkers and frequency questions as measures of fish consumption.

Subjects/Methods:

Participants in the Fishermen substudy numbered 125 men and 139 women (aged 22–74), and in the Health 2000 substudy, 577 men and 712 women (aged 45–74) participated. The aim of the Fishermen study was to examine the overall health effect of fish consumption in a high-consumption population, whereas the aim of the Health 2000 substudy was to obtain in-depth information on cardiovascular diseases and diabetes. Fish consumption was measured by the same validated food frequency questionnaire (FFQ) in both the studies, with a further two separate frequency questions used in the Fishermen substudy. Dioxins, polychlorinated biphenyls (PCBs) and methyl mercury (MeHg) (in the Fishermen substudy alone), and omega-3 polyunsaturated fatty acids (omega-3 PUFAs) (in both studies) were analyzed from fasting serum/blood samples.

Results:

The Spearman's correlation coefficients between FFQ fish consumption and dioxins, PCBs, MeHg and omega-3 PUFAs were respectively 0.46, 0.48, 0.43 and 0.38 among the Fishermen substudy men, and 0.28, 0.36, 0.45 and 0.31 among women. Similar correlation coefficients were observed between FFQ fish consumption and serum omega-3 PUFAs in the Health 2000 substudy, and also between FFQ fish consumption and the frequency questions on fish consumption in the Fishermen substudy. According to multiple regression modeling and LMG metrics, the most important fish consumption biomarkers were dioxins and PCBs among the men and MeHg among the women.

Conclusions:

Environmental contaminants seemed to be slightly better fish consumption biomarkers than omega-3 PUFAs in the Baltic Sea area. The separate frequency questions measured fish consumption equally well when compared with the FFQ.

Introduction

Fish consumption is beneficial especially to cardiovascular health (Mozaffarian, 2008; Calder and Yaqoob, 2009). Conversely, fish may also be an important source of various toxic environmental contaminants, such as polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs, called dioxins in this work) (Kiviranta et al., 2004; Isosaari et al., 2006), polychlorinated biphenyls (PCBs) (Kiviranta et al., 2004; Isosaari et al., 2006) and methyl mercury (MeHg) (EFSA, 2004).

Omega-3 polyunsaturated fatty acids (omega-3 PUFAs) are traditional fish consumption biomarkers (Hunter, 1998). The relationship between habitual fish consumption measured by a food frequency questionnaire (FFQ) and blood concentration of omega-3 PUFAs has been assessed in several studies, for instance among Norwegians (Andersen et al., 1996), Englishmen (Welch et al., 2006), Americans (Sun et al., 2007), Canadians (Philibert et al., 2006) and Australians (Mina et al., 2007). In these studies, the correlation coefficients have ranged from 0.17 to 0.50 when total fish consumption was used. The correlation coefficients have been slightly higher for fatty fish, ranging from 0.19 to 0.50 (Philibert et al., 2006; Welch et al., 2006; Mina et al., 2007), and lower for lean fish, ranging from 0.01 to 0.12 (Welch et al., 2006; Mina et al., 2007).

Many environmental contaminants are fat soluble and therefore originate mainly from fatty fish like omega-3 PUFAs. Serum concentrations of omega-3 PUFAs are affected by several dietary and nondietary factors such as metabolism, genetics and lifestyle (Hunter, 1998), and they reflect intake only for the last few days (Arab, 2003). The concentrations of environmental contaminants are known to vary according to region, fish species, and the age and size of the fish (Kiviranta et al., 2003; Isosaari et al., 2006; Domingo and Bocio, 2007) but contaminants have very slow elimination in the human body and they accumulate even at low exposures (Tuomisto et al., 1999). Owing to accumulation, environmental contaminants may reflect long-term fish intake at least in those areas where fish is an important source of exposure. To the best of our knowledge, there are no studies in which the relationship between habitual fish consumption measured by an FFQ on whole diet and tissue concentrations of environmental contaminants has been assessed. Overall, studies on the relationship between fish consumption and environmental contaminants are scarce (Svensson et al., 1991, 1995; Asplund et al., 1994; Bergdahl et al., 1998; Arisawa et al., 2003; Björnberg et al., 2005).

The aim of this study was to compare the ability of environmental contaminants and omega-3 PUFAs to reflect fish consumption and primarily to investigate the usefulness of environmental contaminants as biomarkers of fish consumption. Another aim was to assess whether separate frequency questions measure fish consumption equally well when compared with an FFQ on whole diet. The associations were studied in a population with high fish consumption, and when possible, the analyses were repeated in a larger general population subsample.

Materials and methods

Study populations

In the Nutrition, Environment and Health study, 1427 professional fishermen, their wives and other family members answered a self-administered health questionnaire (Turunen et al., 2008). This study looked at the overall health effect of fish consumption in a high fish consumption population, that is, professional fishermen and their families. A total of 309 volunteers, aged 22–74 years, and living near the Helsinki and Turku study centers participated in a health examination study (the Fishermen substudy). Of those, 125 men and 139 women reported fish consumption through the FFQ and through separate frequency questions, and had data on blood concentrations of environmental contaminants and fatty acids.

Analyses of FFQ fish consumption and serum fatty acids were repeated using data from the population-based Health 2000 health examination survey, which looked at major public health problems and their determinants in a nationally representative population sample in Finland (n=7977) (Aromaa and Koskinen, 2004). A total of 1526 volunteers, aged 45–74 years, and living near the study locations in the five university hospital districts of Finland (Helsinki, Turku, Tampere, Kuopio and Oulu) participated in an in-depth health examination study on cardiovascular disease and diabetes (the Health 2000 substudy). Of these, 577 men and 712 women reported fish consumption through the FFQ and had data on blood concentrations of fatty acids.

Both the Fishermen substudy and the Health 2000 substudy were coordinated by the National Institute for Health and Welfare in Finland (THL—which includes the former National Public Health Institute, KTL). The studies were independent of each other, and had different study populations and time frames, although they had similar study protocols, which enables comparisons. The key features of the Fishermen substudy were high fish consumption and analyzed serum concentrations of environmental contaminants, whereas the key feature of the Health 2000 substudy was a relatively large sample with fish consumption close to that of the general population.

Data collection

Fish consumption and other dietary variables

In both substudies, diet was assessed by the same validated self-administered FFQ designed to cover the whole diet during the preceding year (Männistö et al., 1996; Paalanen et al., 2006). The FFQ consisted of 128 food items and mixed dishes with specified serving sizes, including 10 fish dishes. The nine response options ranged from ‘never or seldom’ to ‘six or more times per day’ (see Appendix). Dietary data were processed in the Fineli Finnish Food Composition Database (National Public Health Institute), and daily fish consumption (g per day), energy (MJ per day), alcohol (ethanol, g per day) and fatty acid (g per day) intakes, and the prevalence of users of fish oil supplements was calculated. In the previous validation study, the same participants completed the FFQ twice, and reproducibility between the first and the second FFQ measurement was 0.63 for fish consumption. Validity between the first FFQ measurement and the 14-day food record measurement was 0.46 for fish consumption (Männistö et al., 1996).

In addition, in the Fishermen study, the health questionnaire contained two separate frequency questions to obtain fish consumption information from a larger population (n=1427). The participants were asked about the frequency of use of 12 fish dishes and 12 fish species (see Appendix). For both frequency questions, the six response options ranged from ‘never’ to ‘almost every day’ and the frequencies were summed to four variables: fish dishes, fish species, fatty fish species and lean fish species (times per month). Species containing more than 3.5% fat according to the Fineli Finnish Food Composition Database were included in fatty fish (see Appendix).

Serum/blood concentrations of environmental contaminants and fatty acids

In both studies, the blood samples were collected after 10–12 h of fasting and analyzed using the same method. Blood samples were not available for environmental contaminant analyses in the Health 2000 substudy.

Serum concentrations (pg/g or ng/g fat) of 17 dioxin and 37 PCB congeners were analyzed using a high-resolution mass spectrometer equipped with a gas chromatograph (Kiviranta et al., 2002). Toxic equivalents (TEQs) for dioxins (WHOPCDD/F-TEQ) and PCBs (WHOPCB-TEQ) were calculated with the set of toxic equivalency factors recommended by the WHO in 1998 (Van den Berg et al., 1998). In addition, four individual congeners were included in the analyses. Based on our previous studies, pentachlorodibenzofuran (2,3,4,7,8-PeCDF), PCB 126 and PCB 153 are likely to be correlated with fish consumption, whereas octachlorodibenzo-p-dioxin (OCDD) is likely to be uncorrelated with fish consumption (Kiviranta et al., 2002, 2003). Blood MeHg concentration (ng/ml) was analyzed from whole blood using an isotope dilution-gas chromatograph/mass spectrometer (Yang et al., 2003; Baxter et al., 2007). The interassay coefficients of variation were 5.3% for WHOPCDD/F-TEQ, 9.6% for WHOPCB-TEQ, 6.7% for pentachlorodibenzofuran, 6.8% for OCDD, 12% for PCB 126, 7.3% for PCB 153 and 4.2% for MeHg.

Serum fatty acids (proportion from all serum fatty acids, % FAs) were analyzed using a gas chromatograph and flame ionization detector (Jula et al., 2005). Omega-3 PUFAs were defined here as the sum of eicosapentaenoic acid (EPA), docosapentaenoic acid (DPA) and docosahexaenoic acid (DHA). α-Linolenic acid (ALA) and palmitic and stearic acids are known to be uncorrelated with fish consumption. Interassay coefficients of variation were 12% for EPA, 12% for DPA, 19% for DHA, 4.8% for ALA, 4.9% for palmitic acid and 7.5% for stearic acid.

Other variables

Weight (kg, using a typical scales), height (cm, using a wall-mounted stadiometer), and waist and hip girth (cm, using a flexible measuring tape) were measured during the health examination, and body mass index (kg/m2), waist–hip ratio and age at the time of the examination (years) were calculated. Data on smoking were obtained from a self-administered health questionnaire in the Fishermen substudy, and from a structured interview in the Health 2000 substudy. The following questions were asked in both the questionnaire and the interview: ‘Have you ever smoked?’, ‘Have you smoked at least 100 times?’, ‘Have you ever smoked regularly (that is, daily for at least one year)?’ and ‘When did you last smoke?’. The final smoking variable had five classes: ‘daily smoker’, ‘occasional smoker’, ‘ex-smoker, cessation 1–12 months ago’, ‘ex-smoker, cessation over a year ago’ and ‘never-smoker’. In this study, the prevalence of daily or occasional smoking (current smoking) was reported. Daily or occasional smoker was defined as a participant who reported having smoked most recently on the current date, the previous day or from 2 to 30 days previously.

Statistical analyses

Cross-classification was used to assess agreement between the FFQ fish consumption and the fish consumption measured by the two separate frequency questions, that is, the ability of these dietary methods to classify individuals into the same fish consumption category. The participants were categorized into quartiles by FFQ fish, fish dishes and fish species. The percentages of participants in the same quartile, in the same or adjacent quartile and in the extreme quartile were calculated.

Nonparametric Spearman's correlation coefficients were calculated to assess the relationships between FFQ fish, fish dishes, fish species, and serum/blood concentrations of environmental contaminants and fatty acids.

For multiple linear regression analyses, all dietary and biochemical variables were log-transformed according to log(x+1). FFQ fish and fish dishes were considered as dependent variables and the biomarkers were considered as regressors. Dioxins and PCBs were not simultaneously included in the models due to high correlation (r=0.9). Age and total energy intake were included in all models. Additional adjusting for body mass index and the use of fish oil supplements did not change the results of the linear regression analyses and therefore age- and energy-adjusted models are shown. LMG metrics was used to assess the regressors’ relative contributions to the model's total explanatory value, that is, the relative importance of the biomarkers. This method was chosen because regressors are typically correlated and model R2 cannot be broken down into shares from the individual regressors. In the procedure, the sequential sums of squares were averaged over all orderings of the regressors (Lindeman et al., 1980; Kruskal, 1987). The partial R2 by LMG metrics and their 95% bootstrap CIs were calculated by using the relaimpo package in the R statistical software (R Development Core Team, 2009; Grömping, 2006; Table 5).

Results

The Fishermen substudy participants were, on average, 5 years younger, and had slightly higher alcohol intake and prevalence of fish oil supplement use, and a smaller waist–hip ratio than the Health 2000 substudy participants. On the basis of the FFQ on whole diet, all the participants of the Fishermen substudy reported eating fish, whereas three men and seven women of the Health 2000 substudy reported not eating fish (Table 1). FFQ fish consumption and serum concentrations of omega-3 PUFAs were approximately 1.5-fold among the Fishermen substudy participants compared with the Health 2000 substudy participants. The Fishermen substudy participants were professional fishermen, fishermen's wives and other family members who are known to eat a lot of fish possibly due to easy availability (Table 2).

Table 1 Characteristics of the Fishermen substudy and Health 2000 substudy participants
Table 2 Age-adjusted means for the variables from the FFQ on whole diet, the frequency questions on fish consumption, and blood sample analyses among the Fishermen substudy and Health 2000 substudy participants

FFQ fish and fish dishes were able to classify 47% of the Fishermen sub study men into the same quartile and 87% into the same or adjacent quartile (data not shown). The corresponding proportions for fish species were 51 and 85%. Among the women, FFQ fish and fish dishes were able to classify 51% of the women into the same quartile and 88% into the same or adjacent quartile. The corresponding proportions for fish species were 37 and 78%. Thus, the proportion of grossly misclassified (classified into extreme quartiles) participants was approximately 2% among the men and 6% among the women, the same for fish dishes and fish species. The age-adjusted correlation coefficients between FFQ fish and separate frequency questions on fish consumption were 0.62 for fish dishes, 0.64 for fish species, 0.55 for fatty fish species and 0.32 for lean fish species among the men (data not shown). The corresponding coefficients were 0.61, 0.44, 0.39 and 0.21 among the women.

The age-adjusted correlation coefficients between FFQ fish and serum/blood environmental contaminants were 0.46 for dioxins, 0.48 for PCBs and 0.43 for MeHg among the Fishermen substudy men (Table 3). Among the women, the corresponding correlation coefficients were 0.28 for dioxins, 0.36 for PCBs and 0.45 for MeHg. When compared with FFQ fish, fish dishes yielded approximately 10–15% lower correlation coefficients with environmental contaminants among the men, and approximately 10–35% higher correlation coefficients among the women. Fish species yielded less than 10% lower correlation coefficients with environmental contaminants when compared with FFQ fish in both sexes.

Table 3 Age-adjusted Spearman's correlation coefficients between serum/blood environmental contaminants and fish consumption variables from the FFQ on whole diet and the frequency questions on fish consumption among the Fishermen substudy participants

When compared with WHOPCDD/F-TEQ, dioxin congener pentachlorodibenzofuran yielded equal correlation coefficients with all fish consumption variables among the men, and approximately 20% higher correlation coefficients among the women. PCB congener 126 yielded equal correlation coefficients with all fish consumption variables when compared with WHOPCB-TEQ in both sexes. As expected, dioxin congener OCDD had a weak correlation with fish consumption.

The age-adjusted correlation coefficients between FFQ fish and serum omega-3 PUFAs were 0.38 among the men and 0.31 among the women participants of the Fishermen substudy. The corresponding coefficients were approximately 10% lower among the Health 2000 substudy participants (Table 4). When compared with FFQ fish, fish dishes yielded less than 10% lower correlation coefficient and fish species a 16% lower correlation coefficient with serum omega-3 PUFAs among the Fishermen substudy men. Among the women, fish dishes yielded 10% higher correlation coefficient and fish species 35% lower correlation coefficient with serum omega-3 PUFAs when compared with FFQ fish. As expected, the correlation coefficients between FFQ fish and serum concentrations of ALA and the sum of palmitic and stearic acid were close to zero.

Table 4 Age-adjusted Spearman's correlation coefficients between serum fatty acids and fish consumption variables from the FFQ on whole diet and the frequency questions on fish consumption among the Fishermen substudy and Health 2000 substudy participants

In the multiple regression modeling, all four fish consumption biomarkers were statistically significantly associated with FFQ fish, when considered in separate models (Table 5). In model 5 (dioxins, MeHg and omega-3 PUFAs as regressors) and model 6 (PCBs, MeHg and omega-3 PUFAs as regressors), all three environmental contaminants, though not omega-3 PUFAs, were statistically significantly associated with FFQ fish among the Fishermen substudy men. Among the women, only MeHg was statistically significantly associated with FFQ fish. Using the LMG metrics to assess the relative importance of the regressors, we found that dioxins (partial R2 13%, 95% bootstrap CI 6.9–21) and PCBs (15%, 7.8–23) had the largest relative contribution to the model's total explanatory value among the men, although the contribution of MeHg was not notably lower (8.6%, 3.6–16 in model 5; 8.1%, 3.2–15 in model 6). In contrast, MeHg (16%, 8.9–25 in model 5; 16%, 8.5–24 in model 6) was clearly the most important biomarker among the women. When fish dishes was considered as the dependent variable, the differences between the relative contributions of the biomarkers almost disappeared among the men, whereas MeHg remained the most important biomarker among the women (Table 5).

Table 5 Results of multiple linear regression analyses between each of the two fish consumption variables (FFQ fish and fish dishes) and fish consumption biomarkers among the Fishermen substudy participants

Discussion

In this study, serum/blood environmental contaminants seemed to be slightly better fish consumption biomarkers than serum omega-3 PUFAs. Dioxins and PCBs were the most important biomarkers among the men, and MeHg was the most important among the women. There was a satisfactory agreement between fish consumption data from the FFQ on whole diet and the separate frequency questions.

This is one of the rare studies using data on both a validated FFQ on whole diet (Männistö et al., 1996; Paalanen et al., 2006) and serum/blood concentrations of environmental contaminants. To the best of our knowledge, this is also the first study to assess the relative importance of different fish consumption biomarkers. LMG metrics has only been used rarely but remains, to our knowledge, the best method to quantify the relative contributions of the regressors to the model's total explanatory value (Grömping, 2006).

As to the limitations of this study, FFQ measures the usual long-term diet, for example, over the past year, whereas serum fatty acid concentration reflects intake over the past few days. Adipose tissue would have been the most preferable media for fatty acid analyses as it reflects long-term dietary intake under homeostatic conditions (Arab, 2003). In addition, FFQ is designed to rank individuals according to their dietary intake and not to measure absolute intake. Thus, there can be some measurement error in FFQ estimates due to under- and overreporting (Willett, 1998). Our FFQ has been reported to somewhat overestimate fish consumption (Männistö et al., 1996). In addition, the use of a total fish consumption variable including both fatty (oily) and lean (white) fish may also have attenuated the studied associations as lean fish typically has lower correlations with serum omega-3 PUFAs than fatty fish. In general, concentration biomarkers are not always affordable or even eligible for validation purposes or as a surrogate source of dietary data (Jenab et al., 2009). They are affected by many nondietary factors such as metabolism, genetics and life style (for example, smoking, obesity and physical activity) (Hunter, 1998).

The correlation coefficients between FFQ fish consumption and serum omega-3 PUFAs were almost of the same magnitude in the Fishermen substudy and in the Health 2000 substudy. This indicates that the results of the Fishermen substudy are probably generalizable at least to the Finnish general population.

The validated FFQ on whole diet (Männistö et al., 1996; Paalanen et al., 2006) and separate frequency questions on fish consumption were independent and unrelated sources of fish consumption data. However, they seemed to classify the majority of the participants into the same or adjacent fish consumption quartile, and their agreement can be considered good as less than 10% of the participants were grossly misclassified (Masson et al., 2003). Thus, the nonvalidated frequency questions may be used as measures of fish consumption in further epidemiological studies.

In the Fishermen substudy, fish consumption yielded higher correlation coefficients with environmental contaminants than with omega-3 PUFAs. This is probably due to the fact that dioxins and PCBs accumulate, and their serum concentrations are fairly stable and slowly rising (Tuomisto et al., 1999).

The men in the study had higher correlation coefficients between fish consumption variables and serum/blood biomarkers compared with the women, and the importance of the biomarkers differed by sex. This may be explained partly by the higher variation in fish consumption and lower volume of distribution (lower proportion of body fat) for fat-soluble compounds among men. Furthermore, lower concentrations of dioxins and PCBs among the women may have increased the relative importance of MeHg as a fish consumption biomarker.

Among the women, fish dishes yielded approximately 10–35% higher correlation coefficients with biomarkers than FFQ fish and approximately 15–40% higher correlation coefficients than fish species. Fish dishes had more indications of food preparation methods (for example, cooked, baked, fried or smoked fish) than the FFQ (for example, frozen fish, salmon dishes, Baltic herring dishes or whitefish/perch/vendace/pike), and therefore that question may have suited the women better. Conversely, the men may be better in reporting fish species than the women, possibly due to practical experience in fishing.

Regarding those few studies using environmental contaminants as fish consumption biomarkers, a Japanese study reported correlation coefficients from 0.09 to 0.32 (depending on the investigated fish type) between fish consumption frequency (from a frequency questionnaire) and serum dioxins or PCBs in both sexes combined (Arisawa et al., 2003). Lower correlation coefficients compared with this study may be due to lower exposure to dioxins and PCBs as well as the lack of a measure for total fish consumption. In Sweden, the correlation coefficient between total fish consumption (from a dietary interview) and plasma 2,3,7,8-TCDD was 0.84 (Svensson et al., 1991), and the correlation coefficient between total fish consumption (from a frequency questionnaire) and different serum PCB congeners ranged from 0.63 to 0.87 (Asplund et al., 1994). The higher correlation coefficients compared with those given in this study are probably due to higher exposure to dioxins and PCBs and a small study group (n=34) consisting of men from extreme fish consumption groups.

In this study, MeHg was analyzed from whole blood as it was available for all the study participants. Although concentrations in hair are most commonly used in epidemiological studies, whole-blood concentrations correlate well with hair concentrations (Björnberg et al., 2005). Two Swedish studies reported correlation coefficients around 0.50 between total fish consumption (from dietary interviews) and blood MeHg (Svensson et al., 1995; Bergdahl et al., 1998), which is only marginally higher than in this study.

In this study, fatty acids were analyzed from total serum including all three lipid fractions (cholesterol esters, phospholipids and triglycerides). It reflects intake only over the last few days (Arab, 2003) but has been shown to be a feasible biomarker (Hodson et al., 2008) and more affordable than subfraction analysis (Baylin et al., 2005). Only three of the previous studies (Andersen et al., 1999; Philibert et al., 2006; Sun et al., 2007) using an FFQ have used total serum or plasma to analyze fatty acids. Two of them reported correlation coefficients around 0.50 between FFQ fish consumption (total fish)/FFQ omega-3 PUFA intake and serum omega-3 PUFAs (Andersen et al., 1999; Philibert et al., 2006). Reasons for the slightly higher correlation coefficients compared with this study may be the use of a selected occupational group as a study population (Andersen et al., 1999), and using an FFQ with a special emphasis on fish consumption (Philibert et al., 2006). In the previous studies, phospholipids have been the most common choice for fatty acid analyses. The correlation coefficients between FFQ fish consumption (total fish)/FFQ omega-3 PUFA intake and serum omega-3 PUFAs have been slightly lower than in this study, ranging from 0.09 to 0.36 and being typically around 0.20 and 0.30 (Ma et al., 1995; Andersen et al., 1996; Hjårtaker et al., 1997; Li et al., 2001; Woods et al., 2002; Kobayashi et al., 2003; Welch et al., 2006). One study using erythrocytes reported similar correlation coefficients (Mina et al., 2007) to this study. In the previous studies, the correlation coefficients between FFQ fish and serum omega-3 PUFAs were somewhat higher when fatty fish was used. In our study, only total fish consumption was available from the FFQ.

To conclude, self-reported fish consumption was reflected reasonably well in serum/blood concentrations of dioxins, PCBs, MeHg and omega-3 PUFAs. The associations were approximately at the same level as those reported in earlier studies. The results of our study indicate that serum/blood concentrations of dioxins, PCBs and MeHg may be better fish consumption biomarkers than serum concentrations of omega-3 PUFAs. However, this may be generalizable only to those populations where fish is an important source of these environmental contaminants like in the Baltic Sea area. The relative importance of the biomarkers seemed to differ between the sexes. Dioxins and PCBs were the most important biomarkers among the men, whereas MeHg was the most important biomarker among the women. The separate frequency questions appeared to yield equally good estimates of habitual fish consumption as the whole diet FFQ, and they may be used in further epidemiological studies.

Conflict of interest

The authors declare no conflict of interest.

References

  1. Andersen LF, Solvoll K, Drevon CA (1996). Very-long-chain n-3 fatty acids as biomarkers for intake of fish and n-3 fatty acid concentrates. Am J Clin Nutr 64, 305–311.

  2. Andersen LF, Solvoll K, Johansson LR, Salminen I, Aro A, Drevon CA (1999). Evaluation of a food frequency questionnaire with weighed records, fatty acids, and alpha-tocopherol in adipose tissue and serum. Am J Epidemiol 150, 75–87.

  3. Arab L (2003). Biomarkers of fat and fatty acid intake. J Nutr 133 (Suppl 3), 925S–932S.

  4. Arisawa K, Matsumura T, Tohyama C, Saito H, Satoh H, Nagai M et al. (2003). Fish intake, plasma omega-3 polyunsaturated fatty acids, and polychlorinated dibenzo-p-dioxins/polychlorinated dibenzo-furans and co-planar polychlorinated biphenyls in the blood of the Japanese population. Int Arch Occup Environ Health 76, 205–215.

  5. Aromaa A, Koskinen S (2004). Health and Functional Capacity in Finland. Baseline Results of the Health 2000 Health Examination Survey. National Public Health Institute: Helsinki.

  6. Asplund L, Svensson BG, Nilsson A, Eriksson U, Jansson B, Jensen S et al. (1994). Polychlorinated biphenyls, 1,1,1-trichloro-2,2-bis(p-chlorophenyl)ethane (p,p′-DDT) and 1,1-dichloro-2,2-bis(p-chlorophenyl)-ethylene (p,p′-DDE) in human plasma related to fish consumption. Arch Environ Health 49, 477–486.

  7. Baxter DC, Rodushkin I, Engstrom E, Klockare D, Waara H (2007). Methylmercury measurement in whole blood by isotope-dilution GC-ICPMS with 2 sample preparation methods. Clin Chem 53, 111–116.

  8. Baylin A, Kim MK, Donovan-Palmer A, Siles X, Dougherty L, Tocco P et al. (2005). Fasting whole blood as a biomarker of essential fatty acid intake in epidemiologic studies: comparison with adipose tissue and plasma. Am J Epidemiol 162, 373–381.

  9. Bergdahl IA, Schutz A, Ahlqwist M, Bengtsson C, Lapidus L, Lissner L et al. (1998). Methylmercury and inorganic mercury in serum—correlation to fish consumption and dental amalgam in a cohort of women born in 1922. Environ Res 77, 20–24.

  10. Björnberg KA, Vahter M, Grawe KP, Berglund M (2005). Methyl mercury exposure in Swedish women with high fish consumption. Sci Total Environ 341, 45–52.

  11. Calder PC, Yaqoob P (2009). Omega-3 polyunsaturated fatty acids and human health outcomes. Biofactors 35, 266–272.

  12. Domingo JL, Bocio A (2007). Levels of PCDD/PCDFs and PCBs in edible marine species and human intake: a literature review. Environ Int 33, 397–405.

  13. EFSA (2004). Opinion of the Scientific Panel on Contaminants in the Food Chain on a request from the commission related to mercury and methylmercury in food. EFSA J 34, 1–14.

  14. Grömping U (2006). Relative importance for linear regression in R: the package relaimpo. J Stat Soft 17.

  15. Hjårtaker A, Lund E, Bjerve KS (1997). Serum phospholipid fatty acid composition and habitual intake of marine foods registered by a semi-quantitative food frequency questionnaire. Eur J Clin Nutr 51, 736–742.

  16. Hodson L, Skeaff CM, Fielding BA (2008). Fatty acid composition of adipose tissue and blood in humans and its use as a biomarker of dietary intake. Prog Lipid Res 47, 348–380.

  17. Hunter DJ (1998). Biochemical indicators of dietary intake. In: Willett W (ed). Nutritional Epidemiology. Oxford University Press: New York. pp 174–243.

  18. Isosaari P, Hallikainen A, Kiviranta H, Vuorinen PJ, Parmanne R, Koistinen J et al. (2006). Polychlorinated dibenzo-p-dioxins, dibenzofurans, biphenyls, naphthalenes and polybrominated diphenyl ethers in the edible fish caught from the Baltic Sea and lakes in Finland. Environ Pollut 141, 213–225.

  19. Jenab M, Slimani N, Bictash M, Ferrari P, Bingham SA (2009). Biomarkers in nutritional epidemiology: applications, needs and new horizons. Hum Genet 125, 507–525.

  20. Jula A, Marniemi J, Ronnemaa T, Virtanen A, Huupponen R (2005). Effects of diet and simvastatin on fatty acid composition in hypercholesterolemic men: a randomized controlled trial. Arterioscler Thromb Vasc Biol 25, 1952–1959.

  21. Kiviranta H, Ovaskainen ML, Vartiainen T (2004). Market basket study on dietary intake of PCDD/Fs, PCBs, and PBDEs in Finland. Environ Int 30, 923–932.

  22. Kiviranta H, Vartiainen T, Parmanne R, Hallikainen A, Koistinen J (2003). PCDD/Fs and PCBs in Baltic herring during the 1990s. Chemosphere 50, 1201–1216.

  23. Kiviranta H, Vartiainen T, Tuomisto J (2002). Polychlorinated dibenzo-p-dioxins, dibenzofurans, and biphenyls in fishermen in Finland. Environ Health Perspect 110, 355–361.

  24. Kobayashi M, Sasaki S, Kawabata T, Hasegawa K, Tsugane S (2003). Validity of a self-administered food frequency questionnaire used in the 5-year follow-up survey of the JPHC Study Cohort I to assess fatty acid intake: comparison with dietary records and serum phospholipid level. J Epidemiol 13, S64–S81.

  25. Kruskal W (1987). Relative importance by averaging over orderings. Am Stat 41, 6–10.

  26. Li D, Zhang H, Hsu-Hage BH, Wahlqvist ML, Sinclair AJ (2001). The influence of fish, meat and polyunsaturated fat intakes on platelet phospholipid polyunsaturated fatty acids in male Melbourne Chinese and Caucasian. Eur J Clin Nutr 55, 1036–1042.

  27. Lindeman RH, Merenda PF, Gold RZ (1980). Introduction to Bivariate and Multivariate Analysis. Longman Higher Education: Glenview, IL.

  28. Ma J, Folsom AR, Shahar E, Eckfeldt JH (1995). Plasma fatty acid composition as an indicator of habitual dietary fat intake in middle-aged adults. The Atherosclerosis Risk in Communities (ARIC) Study Investigators. Am J Clin Nutr 62, 564–571.

  29. Masson LF, McNeill G, Tomany JO, Simpson JA, Peace HS, Wei L et al. (2003). Statistical approaches for assessing the relative validity of a food-frequency questionnaire: use of correlation coefficients and the kappa statistic. Public Health Nutr 6, 313–321.

  30. Mina K, Fritschi L, Knuiman M (2007). A valid semiquantitative food frequency questionnaire to measure fish consumption. Eur J Clin Nutr 61, 1023–1031.

  31. Mozaffarian D (2008). Fish and n-3 fatty acids for the prevention of fatal coronary heart disease and sudden cardiac death. Am J Clin Nutr 87, 1991S–1996S.

  32. Männistö S, Virtanen M, Mikkonen T, Pietinen P (1996). Reproducibility and validity of a food frequency questionnaire in a case–control study on breast cancer. J Clin Epidemiol 49, 401–409.

  33. National Public Health Institute (2008). Fineli Finnish Food Composition Database. http://www.fineli.fi/3.9.2008.

  34. Paalanen L, Männistö S, Virtanen MJ, Knekt P, Räsänen L, Montonen J et al. (2006). Validity of a food frequency questionnaire varied by age and body mass index. J Clin Epidemiol 59, 994–1001.

  35. Philibert A, Vanier C, Abdelouahab N, Chan HM, Mergler D (2006). Fish intake and serum fatty acid profiles from freshwater fish. Am J Clin Nutr 84, 1299–1307.

  36. R Development Core Team (2009). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org.

  37. Sun Q, Ma J, Campos H, Hankinson SE, Hu FB (2007). Comparison between plasma and erythrocyte fatty acid content as biomarkers of fatty acid intake in US women. Am J Clin Nutr 86, 74–81.

  38. Svensson BG, Nilsson A, Hansson M, Rappe C, Akesson B, Skerfving S (1991). Exposure to dioxins and dibenzofurans through the consumption of fish. N Engl J Med 324, 8–12.

  39. Svensson BG, Nilsson A, Jonsson E, Schutz A, Akesson B, Hagmar L (1995). Fish consumption and exposure to persistent organochlorine compounds, mercury, selenium and methylamines among Swedish fishermen. Scand J Work Environ Health 21, 96–105.

  40. Tuomisto J, Vartiainen T, Tuomisto J (1999). Synopsis on Dioxins and PCBs. National Public Health Institute: Kuopio.

  41. Turunen AW, Verkasalo PK, Kiviranta H, Pukkala E, Jula A, Mannisto S et al. (2008). Mortality in a cohort with high fish consumption. Int J Epidemiol 37, 1008–1017.

  42. Van den Berg M, Birnbaum L, Bosveld AT, Brunstrom B, Cook P, Feeley M et al. (1998). Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environ Health Perspect 106, 775–792.

  43. Welch AA, Bingham SA, Ive J, Friesen MD, Wareham NJ, Riboli E et al. (2006). Dietary fish intake and plasma phospholipid n-3 polyunsaturated fatty acid concentrations in men and women in the European Prospective Investigation into Cancer-Norfolk United Kingdom cohort. Am J Clin Nutr 84, 1330–1339.

  44. Willett WC (1998). Food-frequency methods. In: Willett W (ed). Nutritional Epidemiology. Oxford University Press: New York. pp 74–100.

  45. Woods RK, Stoney RM, Ireland PD, Bailey MJ, Raven JM, Thien FC et al. (2002). A valid food frequency questionnaire for measuring dietary fish intake. Asia Pac J Clin Nutr 11, 56–61.

  46. Yang L, Colombini V, Maxwell P, Mester Z, Sturgeon RE (2003). Application of isotope dilution to the determination of methylmercury in fish tissue by solid-phase microextraction gas chromatography–mass spectrometry. J Chromatogr A 1011, 135–142.

Download references

Acknowledgements

We thank the volunteers and the research staff of the Fishermen study and the Health 2000 health examination survey. This work was supported by the Academy of Finland (project numbers 77008, 205324, 206950, 124286); the Finnish Cancer Organisations; the Yrjö Jahnsson Foundation and the Juho Vainio Foundation.

Author information

Correspondence to A W Turunen.

Appendix

Appendix

Details of the original fish consumption questions

Table A1

Table a1 FFQ on whole diet (available for the Fishermen substudy and the Health 2000 substudy)

Table A2

Table a2 Frequency questions on fish consumption in a health questionnaire (available only for the Fishermen substudy)

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • fish
  • biomarkers
  • dioxins
  • polychlorinated biphenyls
  • methyl mercury
  • omega-3 fatty acids

Further reading