L Hagmar 1, M Persson-Moschos 2,BA Ê kesson 2 and A SchuÈtz 1

Transcription

1 European Journal of Clinical Nutrition (1998) 52, 796±800 ß 1998 Stockton Press. All rights reserved 0954±3007/98 $ Plasma levels of selenium, selenoprotein P and glutathione peroxidase and their correlations to sh intake and serum levels of thyrotropin and thyroid hormones: A study on Latvian sh consumers L Hagmar 1, M Persson-Moschos 2,BA Ê kesson 2 and A SchuÈtz 1 1 Department of Occupational and Environmental Medicine, University Hospital, Lund; and 2 Department of Applied Nutrition and Food Chemistry, University of Lund, Lund, Sweden Objective: To study the relationships between sh intake and different markers of selenium status and thyroid hormone function. Design: Cross-sectional study. Setting and subjects: Sixty-eight men (age 24±79 years) were recruited among coastal shermen and inland subjects from Latvia. None of the subjects was on selenium medication or had any known endocrine disease. Main outcome measures: Correlations between sh intake, plasma levels of selenium, selenoprotein P, glutathione peroxidase, organic mercury in erythrocytes and TSH in serum. Results: Selenium in plasma ranged from 0.30 to 1.56 mmol=1, selenoprotein P from 0.54 to 2.21 arbitrary units relative to pooled plasma, and glutathione peroxidase from 1.20 to 5.73 mg=1. The number of sh meals per month was correlated with plasma selenium, selenoprotein P and glutathione peroxidase (r ˆ 0.63, r ˆ 0.62 and r ˆ 0.50, respectively; P < 0.001). Plasma selenium was correlated with selenoprotein P and glutathione peroxidase (r ˆ 0.88 and r ˆ 0.67, respectively; P < 0.001), and also selenoprotein P and glutathione peroxidase were correlated (r ˆ 0.63, P < 0.001). The mean plasma selenium level in those with a high sh intake (21±50 sh meals=month), was 81% higher than in those with lowest sh intake. TSH in serum was inversely correlated with plasma selenium and selenoprotein P. Thyroid hormone levels were not correlated with plasma selenium, selenoproteins or sh intake. Conclusions: In this study group, selenium from sh intake had a marked impact on all variables studied on selenium status. No impact of selenium status on T3 and T4 levels was observed. The slightly negative correlation of selenium status with TSH levels might indicate a higher TSH secretion at low selenium status. Sponsorship: The study was supported by grants from the Swedish Environmental Protection Agency, the Medical Faculty, Lund University, and the PaÊhlsson Foundation. B.AÊ. was supported by FAIR CT Descriptors: sh intake; selenium; selenoprotein P; glutathione peroxidase; thyrotropin; thyroid hormones; bioavailability Introduction The bioavailability and metabolic fate of selenium from dietary sh in humans have not been studied extensively. Svensson et al (1992) found that Swedish healthy subjects with a high consumption of sh had slightly higher concentrations of selenium in plasma than those with low sh consumption. Moreover, a change to a diet rich in sh increased plasma selenium of healthy subjects (Robinson et al, 1978; Thorngren & AÊ kesson, However, in Swedish sh consumers with slightly higher plasma selenium levels compared with nonconsumers, no increase could be observed in the plasma levels of the selenoproteins, selenoprotein P and glutathione peroxidase (Huang et al, 1995), a nding that may have several explanations. Pronounced changes in thyroid hormone metabolism occur in experimental selenium de ciency, mainly because Correspondence: Dr L Hagmar, Department of Occupational and Environmental Medicine, University Hospital, S Lund, Sweden. Received 3 May 1998; revised 23 June 1998; accepted 30 June 1998 the three iodothyronine deiodinases are selenoproteins. The functional effects of low selenium status on thyroid function in humans have so far not been fully characterized (Beech et al, 1993; Terwolbek et al, 1993; Arthur & Beckett, 1994). In addition to the relationship between selenium and deiodinases, it was recently hypothesized that extracellular glutathione peroxidase may be a regulator of thyroid hormone synthesis (Howie et al, 1995). These ndings prompted the study of thyroid function in individuals with large variations in selenium intake. The aim of the present study was rstly to assess the importance of the dietary contribution of selenium from sh for the levels of selenoproteins in plasma in subjects with a larger variation in selenium status than in previously studied cohorts. Latvian shermen at the Gulf of Riga, a part of the Baltic Sea, with a high consumption of locally caught sh, and subjects living in the inland of Latvia, with a low sh consumption, were considered as appropriate study populations. Secondly, the goal was to assess correlations between plasma levels of selenium and selenoproteins, and thyroid hormones in serum.

2 Lars Hagmar was responsible for the study design, obtained fundings and performed the statistical analysis. Andrejs SchuÈtz was responsible for the eld study in Latvia, and for the analyses of organic mercury in erythrocytes and selenium in plasma. Marie Persson-Moschos and BjoÈrn A Ê kesson performed the analyses of selenoprotein P and glutathione peroxidase in plasma. The latter also obtained fundings. All four investigators contributed to the writing of the paper. Lars Hagmar is the guarantor for the integrity of the article as a whole. Methods Study groups and assessment of sh intake Sixty-eight men were recruited among coastal shermen from ve shing villages around the Gulf of Riga, and from Riga and four small villages in the Latvian inland. Each subject was interviewed about his average consumption of different species of sh using a food frequency method. Information on present health status and intake of medical drugs was obtained as described previously (Hagmar et al, 1995). The estimated monthly intake of sh ranged from 0 to 50 meals per month (median 11.8). In terms of sh species, the median intake of salmon from the Baltic Sea was 0.8 meals per month (range 0±18), that of herring from the Baltic Sea was 4 meals per month (range 0±30), and that of pike 2 was meals per month (range 0±18). For some calculations the subjects were divided into three groups according to sh consumption (0±3, 4±20 and 21±50 meals per month). The median age in the study group was 48 years (range 24±79). Forty-eight per cent of the subjects were smokers. None of the subjects was on selenium medication or had any known endocrine disease. Blood sampling and transport Venous blood sampling was made in the morning before the subjects had performed any major physical activity. Blood was collected in tubes containing heparin, and plasma was stored frozen until analysis. The samples received by the analytical laboratories were coded with respect to sh consumption. Selenium in plasma Selenium was determined uorometrically (Lalonde et al, 1982). The detection limit was 0.04 mmol=l (3 mg=l). All samples were analyzed in duplicate, and the coef cient of variation (CV) was 6%. The average value for a serum reference sample (Seronorm Trace Elements, batch 10017, Nycomed AS, Oslo) was 1.2 (s.d. 0.08) mmol=l (n ˆ 10), and coincided with the recommended 1.2 (1.12±1.25) mmol=l. Mercury in erythrocytes Total and inorganic mercury were determined in aciddigested samples (0.5 g) by an automated cold vapour atomic absorption (CV-AA) technique, as described in detail previously (Svensson et al, 1994; Bergdahl et al, 1995). All samples were analyzed in duplicate. The precision calculated as the CV for the duplicate analysis was 5% for both methods, and was independent of the concentration range. The results for a reference sample (Seronorm Trace Elements, lyophilized blood, batch 904, Nycomed, Oslo) averaged 17 nmol=l (range 15.4±17.5; n ˆ 4), as compared to the 19 nmol=l (range 10±30) obtained by other laboratories. The difference between total mercury and inorganic mercury was used as a proxy for the levels of organic mercury in the erythrocytes. Selenoprotein P and glutathione peroxidase Selenoprotein P and extracellular glutathione peroxidase were measured by radioimmunoassays as described elsewhere (Huang & A Ê kesson, 1993; Persson-Moschos et al, 1995). The concentration of selenoprotein P was expressed in arbitrary units (au) relative to a standard of pooled plasma. For selenoprotein P analysis the imprecision expressed as CV was 5.4% for intra-assay and 6.7% for inter-assay variation. For the measurement of extracellular glutathione peroxidase the imprecision was 7.0% for intraassay and 3.5% for inter-assay variation. Assay of thyroid hormones and creatinine Serum thyrotropin (TSH) was analyzed with a uoroimmunoassay technique (Del a, LKB, Finland), with an intra-assay CV of 5% and an inter-assay CV of 2.6%. Serum free triiodothyronine (T3) and free thyroxine (T4) were assayed by a radioimmunoassay (Amersham International, Amersham, UK, with intra- and inter-assay CVs of 2.9% and 4.7%, and 3.7% and 6.9%, respectively, and total serum T3 and T4 were analyzed by radioligand assays. Creatinine was measured by an enzymatic method using creatininase. These analyses were performed at the Laboratory of Clinical Chemistry, Lund University Hospital. Statistical methods Pearson's test was used to assess correlations. The study design was approved by the Ethics committee of Lund University. Results The concentration of selenium in plasma varied from 0.30 to 1.56 mmol=1 (Table 1), and it was signi cantly correlated with the estimated number of sh meals per month (r ˆ 0.63, P < ; Figure 1). Selenoprotein P ranged from 0.54 to 2.21 au and the concentration of glutathione peroxidase varied from 1.20 to 5.73 mg=1 (Table 1); they were also signi cantly correlated with the number of sh meals (r ˆ 0.62, P < and r ˆ 0.50, P < 0.001, respectively, Figures 2 and 3). The mean plasma selenium among those 31 subjects with the highest estimated sh intake (21±50 sh meals=month) was 81% higher than for those 21 subjects with a low sh consumption (0±3 sh meals=month) (Table 1). The corresponding values for selenoprotein P and glutathione peroxidase were 68% and 45%, respectively. Organic mercury ranged from 0.4 to nmol=1, and was associated with sh consumption (Table 1). The level of inorganic mercury did not exceed 13 nmol=1 in any of the subjects. Strong correlations were also observed between plasma selenium and selenoprotein P (r ˆ 0.88, P < 0.001; Figure 4), and between selenium and glutathione peroxidase (r ˆ 0.67, P < 0.001). Consequently, the plasma levels of selenoprotein P and glutathione peroxidase were also highly correlated (r ˆ 0.63, P < 0.001). Organic mercury in erythrocytes also correlated with both selenium in plasma (r ˆ 0.76, P < ), selenoprotein P (r ˆ 0.66, P < ) and glutathione peroxidase (r ˆ 0.66, P < ). Since especially extracellular glutathione peroxidase may be related to kidney function, selenium status in relation to the serum creatinine level was examined. 797

3 798 Table 1 Selenium, selenoprotein P and glutathione peroxidase levels in plasma, and organic mercury in erythrocytes in 68 Latvian males in relation to sh consumption a Fish intake (meals per month) 0±3 (n ˆ 21) 4±20 (n ˆ 16) 21±50 (n ˆ 31) Median Median Median (Range) (Range) (Range) Selenium (mmol=l) (0.30±1.14) (0.46±1.47) (0.66±1.56) Selenoprotein P (au) (0.54±1.15) (0.63±1.83) 0.75±2.21) Glutathione peroxidase (mg=l) (1.20±4.32) (2.31±4.65) (2.69±5.73) Organic mercury (nmol=l) (0.4±34.8) (5.1±115.6) (43.5±218.1) a The subjects were divided into three groups according to sh consumption (0±3, 4±20 or 21±50 sh meals per month). Figure 1 The correlation between the estimated number of sh meals per month and selenium in plasma (r ˆ 0.63, P < ). Figure 3 The correlation between the estimated number of sh meals per month and glutathione peroxidase in plasma (r ˆ 0.50, P < ). Figure 2 The correlation between the estimated number of sh meals per month and selenoprotein P in plasma (r ˆ 0.62, P < ). The concentration of selenoprotein P was expressed in arbitrary units (au) relative to a standard of pooled plasma. Figure 4 The correlation between selenium and selenoprotein P in plasma (r ˆ 0.88, P < ). The concentration of selenoprotein P was expressed in arbitrary units (au) relative to a standard of pooled plasma.

4 Figure 5 The correlation between selenium and TSH in plasma (r ˆ , P ˆ 0.01). Plasma selenium was signi cantly correlated with serum creatinine (r ˆ 0.38, P ˆ 0.001), even more strongly with seleno-protein P (r ˆ 0.50, P < 0.001), but not signi cantly with glutathione peroxidase (r ˆ 0.19; P ˆ 0.12). The median TSH level among the 68 male subjects was 1.3 IU=l (range 0.1±3.7). The median level for free T3 was 5.1 pmol=1 (range 4.0±6.8); for total T3 it was 1.9 nmol=1 (range 1.4±2.7); for free T4 it was 13 pmol=1 (range 7.8± 18); and for total T4 it was 82 nmol=1 (range 57±124). Both plasma selenium and selenoprotein P were signi cantly negatively correlated with TSH (r ˆ , P ˆ 0.01 and r ˆ , P ˆ 0.03, respectively; Figure 5), but not at all correlated with the total or free T3 and T4 levels (data not shown). The weak correlation between glutathione peroxidase and TSH was not statistically signi cant (r ˆ , P ˆ 0.11), and was similar to that found between TSH and the erythrocyte level of organic mercury. The estimated sh intake was not correlated with TSH and thyroid hormone levels. Discussion A low selenium status has been implicated as a risk factor for several chronic diseases (Salonen et al, 1982; Fex et al, 1987; FloheÂ, 1989). Accordingly it is necessary to study the in uence of selenium intake from major food groups on selenium status in different populations. Fish is rich in selenium. Herring, salmon and pike collected in Finnish waters contained 0.18±0.26 mg Se=kg (Koivistoinen, 1980), whereas herring and salmon collected west of Norway contained 0.5 mg Se=kg (Lie et al, 1994). According to the Finnish data, 100 g of sh would thus provide 20 mg of selenium. Studies in several animal species have indicated that the bioavailability of selenium from sh is 50% of that for selenite (Miller et al, 1972; Cantor et al, 1975; Alexander et al, 1983), but in other studies values between 50% and 100% were obtained (Ringdal et al, 1985; Mutanen et al, 1986; Hassan et al, 1987). The divergent results probably re ect the use of different animals and sh feeds, and variations in the methods for assessment of bioavailability (Mutanen, 1986). It is also intriguing that sh species contain varying proportions of high-molecular-mass and low-molecular-mass selenium compounds, the bioavailability of which is not known (A Ê kesson & Srikumar, 1994). Much less is known about the bioavailability of selenium from sh in humans, although a change to a diet rich in sh increased the selenium concentrations in plasma of healthy subjects (Robinson et al, 1978; Thorngren & A Ê kesson, 1987). On the other hand, Meltzer et al (1993) found that a diet rich in sh containing 50% more selenium than a control diet did not signi cantly affect serum and platelet selenium in healthy subjects. In the present study high correlations were found between sh intake and plasma selenium and selenoproteins, indicating that sh in this study group had a considerable impact on the selenium status. Also a previous study of men living close to the Swedish Baltic coast indicated a positive correlation of sh intake with plasma selenium levels (Svensson et al, 1992) but not with the plasma levels of selenoprotein P and glutathione peroxidase, although the two selenoprotein levels were highly intercorrelated (Huang et al, 1995). There are at least two possible reasons for this discrepancy. First, the relatively small variation in plasma selenium between the Swedish high and low consumers of sh might have prevented the detection of relationships between selenoproteins and other variables. Second, the biosynthesis of plasma selenoproteins may have approached saturation at the higher plasma selenium level in the low- sh-intake group of the previous study, 0.99 mmol=1 (Huang et al, 1995), compared to that observed in the present one, 0.66 mmol=1, which would decrease the correlation (Alfthan et al, 1991; Marchaluk et al, 1995). There were several other differences between the Swedish and Latvian study groups. The mean selenoprotein P levels in the Swedish groups were 1.49±1.61 au, in close agreement with ndings in another group of healthy Swedish adults (Marchaluk et al, 1995), whereas the three Latvian groups had means ranging from 0.82 to 1.38 au Similarly, the mean level of glutathione peroxidase in the low-consumption group of the present study, 2.73 mg=1, was generally lower than that found previously, 3.51± 3.67 mg=1 (Huang et al, 1995). Glutathione peroxidase, the rst speci c selenoprotein identi ed, forms part of the defense system against oxidative stress and may also have other functions (FloheÂ, 1989). The glutathione peroxidase family has at least four member enzymes, but little is known about their physiological function. Extracellular glutathione peroxidase is the predominant form in plasma and is mainly produced in the kidney (Avissar et al, 1989; Huang & AÊ kesson, 1993; Avissar et al, 1994; Huang, 1996), and it is also found in milk (Avissar et al, 1991) and aqueous humor (Huang et al, 1997). It is also the major 75 Se-substituted protein secreted by human thyrocytes incubated with ( 75 Se)selenite, and it was hypothesized that it may be a regulator of thyroid hormone synthesis (Howie et al, 1995). Furthermore, the other major selenoprotein in human plasma, selenoprotein P (AÊ kesson et al, 1994) may also act as an extracellular antioxidant (Burk et al, 1995). Selenium status also plays a role in thyroid function because the three iodothyronine deiodinases that catalyze conversions of thyroxine to triiodo- and diiodothyronines have been identi ed as selenoproteins (Arthur et al, 1990; Behne et al, 1990; Salvatore et al, 1996). Pronounced changes in thyroid hormone metabolism occur in experimental selenium de ciency, but the functional effects of 799

5 800 low selenium status on thyroid function in humans have not been fully investigated (Beech et al, 1993; Terwolbek et al, 1993; Arthur & Beckett 1994). In the present study, no relation between T3 or T4 and the selenium status variables or sh intake was observed. The weak negative correlations between TSH and selenium status might indicate a higher TSH secretion at low selenium status. In contrast, in children with low selenium status, a high T4 but no change in T3 and TSH was observed (Terwolbek et al, 1993). Evidently, further studies are necessary on thyroid status in subjects with different selenium status. Thus, the medical importance of the differences in selenium status between the groups studied is hard to assess at present. Moreover, the present data cannot tell whether the differences in selenium status between the Swedish and Latvian groups re ect differences in selenium intake from sh, in selenium forms of the sh species consumed, or in the general food consumption pattern. Acknowledgements ÐWe are grateful to Ms Anita Nilsson, RN, Dr Marcis Leja, and Ms Arija SchuÈtz, RN, for performing blood sampling and interviews in Latvia. Ms Anna Akantis and Mr Giovanni Ferrari skilfully analyzed selenium in plasma and mercury in blood, and Ms Monica Persson assisted in the analysis of selenoproteins. References AÊ kesson B & Srikumar TS (1994): Occurrence of low-molecular-weight and high-molecular-weight selenium compounds in sh. Food Chem. 51, 45±49. AÊ kesson B, Bellew T & Burk RF (1994): Puri cation of selenoprotein P from human plasma. Biochim. Biophys. Acta 1204, 243±249. Alexander AR, Whanger PD & Miller LT (1983): Bioavailability to rats of selenium in various tuna and wheat products. J. Nutr. 113, 196±204. Alfthan G, Aro A, Arvilommi H & Huttunen JK (1991): Selenium metabolism and platelet glutathione peroxidase activity in healthy Finnish men: effects of selenium yeast, selenite, and selenate. Am. J. Clin. Nutr. 53, 120±125. Arthur JR & Beckett GJ (1994): Roles of selenium in type I iodothyronine 5 0 -deiodinase and in thyroid hormone and iodine metabolism. In Selenium in Biology and Human Health, ed. RF Burk, pp 93±116. New York: Springer-Verlag. Arthur JR, Nicol F & Beckett G (1990): Hepatic 5 0 -deiodinase: the role of selenium. Biochem. J. 272, 537±540. Avissar N, Whitin JC, Allen PZ, Palmer IS & Cohen HJ (1989): Antihuman plasma glutathione peroxidase antibodies: immunologic investigations to determine plasma glutathione peroxidase protein and selenium content in plasma. Blood 73, 318±323. Avissar N, Slemmon JR, Palmer IS & Cohen HJ (1991): Partial sequence of human plasma glutathione peroxidase and immunologic identi cation of milk glutathione peroxidase as the plasma enzyme. J. Nutr. 121, 1243±1249. Avissar N, Ornt DB, Yagil Y, Horowitz S, Watkins RH, Kerl EA, Takahashi K, Palmer IS & Cohen HJ (1994): Human kidney proximal tubules are the main source of plasma glutathione peroxidase. Am. J. Physiol. 266, C367±C375. Beech S, Walker SW, Arthur JR & Beckett GJ (1993): Selenium status and thyroidal IDI activity in rat and human thyrocytes. In Trace Elements in Man and Animals (TEMA 8). eds M Anke, D Meissner, CF Mills, pp 1062±1066. Gersdorf: Verlag Media Touristik. Behne D, Kyriakopoulos A, Meinhold H & KoÈhrle J (1990): Identi cation of type I iodothyronine 5 0 -deiodinase as a selenoenzyme. Biochem. Biophys. Res. Commun. 173, 1143±1149. Bergdahl I, SchuÈtz A & Hansson G-AÊ (1995): Automated determination of inorganic mercury in blood after sulfuric acid treatment using cold vapour atomic absorption spectrometry and an inductively heated gold trap. Analyst 120, 1205±1209. Burk RF, Hill KE, Awad JA, Morrow JD, Kato T, Cockell KA & Lyons PR (1995): Pathogenesis of diquat-induced liver necrosis in seleniumde cient rats: assessment of the roles of lipid peroxidation and selenoprotein P. Hepatology 21, 561±569. Cantor AH, Scott ML & Noguchi T (1975): Biological availability of selenium in feedstuffs and selenium compounds for prevention of exudative diathesis in chicks. J. Nutr. 105, 96±105. Fex G, Pettersson B & AÊ kesson B (1987): Low plasma selenium as a risk factor for cancer death in middle-aged men. Nutr. Cancer 10, 221±229. Flohe L (1989): The selenoprotein glutathione peroxidase. In Glutathione Ð Chemical, Biochemical, and Medical Aspects, eds D Dolphin, O Avramovic, R Poulson, pp 643±731 New York: Wiley-Interscience. Hagmar L, Hallberg T, Leja M, Nilsson A & SchuÈtz (1995): High consumption of fatty sh from the Baltic Sea is associated with changes in human lymphocyte subset levels. Toxicol. Lett. 77, 335±342. Hassan S, Hakkarainen J & Lindberg PO (1987): Bioavailability to chicks of selenium in wheat and sh meal. J. Vet. Med. A, 34, 353±363. Howie AF, Walker SW, AÊ kesson B, Arthur JR & Beckett GJ (1995): Thyroidal extracellular glutathione peroxidase: a potential regulator of thyroid hormone synthesis? Biochem. J. 308, 713±717. Huang W (1996): Extracellular glutathione peroxidase. Puri cation, immunoassay, nutritional regulation and clinical aspects. Doctoral thesis, University of Lund. Huang W & AÊ kesson B (1993): Radioimmunoassay of glutathione peroxidase in human serum. Clin. Chim. Acta 219, 139±148. Huang W, AÊ kesson B, Svensson BG, SchuÈtz A, Burk RF & Skerfving S (1995): Selenoprotein P and glutathione peroxidase (EC ) in plasma as indices of selenium status in relation to the intake of sh. Br. J. Nutr. 73, 455±461. Huang W, Koralewska-MakaÂr A, Bauer B & AÊ kesson B (1997): Extracellular glutathione peroxidase and ascorbic acid in aqueous humor and serum of patients operated on for cataract. Clin. Chim. Acta 261, 117±130. Koivistoinen P (ed.) (1980): Mineral element composition of Finnish foods: N, K, Ca, Mg, P, S, Fe, Cu, Mn, Zn, Mo, Co, Ni, Cr, F, Se, Si, Rb, Al, B, Br, Hg, As, Cd, Pb and ash. Acta. Agric. Scand. vol 22 (suppl), 1±171. LaLonde L, Jean Y, Roberts KD, Chapdelaine A & Bleau G (1982): Fluorometry of selenium in serum or urine. Clin. Chem. 28, 172±174. Lie é, Lied E, Maage A, Njaa LR & Sandnes K (1994): Nutrient content in sh and shell sh. Fisk. Dir. Skr. Ser. Ernoering 6, 83±105. Marchaluk E, Persson-Moschos M, Thorling EB & AÊ kesson B (1995): Variation in selenoprotein P concentration in serum from different European regions. Eur. J. Clin. Nutr. 49, 42±48. Meltzer HM, Bibow K, Paulsen IT, Mundal HH, Norheim G & Holm H (1993): Different bioavailability in humans of wheat and sh selenium as measured by blood platelet response to increased dietary Se. Biol. Trace Elem. Res. 36, 229±241. Miller D, Soares Jr JH, Bauersfeld Jr P & Cuppett SL (1972): Comparative selenium retention by chicks fed sodium selenite, selenomethionine, sh meal and sh solubles. Poultry Sci. 51, 1669±1673. Mutanen M (1986): Bioavailability of selenium. Ann. Clin. Res. 18, 48±54. Mutanen M, Koivistoinen P, Morris VC, Levander OA (1986): Nutritional availability to rats of selenium in four seafoods: crab (Callinectes sapidus), oyster (Crassostrea virginica), shrimp (Penaeus duorarum) and Baltic herring (Clupea harengus). Br. J. Nutr. 55(2); 219±225. Persson-Moschos M, Huang W, Srikumar TS, Lindeberg S & AÊ kesson B (1995): Selenoprotein P in serum as a biochemical marker of selenium status. Analyst 120, 833±836. Ringdal O, BjoÈrnestad EOÈ & Julshamn K (1985): Comparative utilization of sh selenium and inorganic selenite by rats of normal selenium status. Ann. Nutr. Metab. 29, 297±305. Robinson MF, Rea HM, Friend GM, Stewart RDH, Snow PC & Thomson CD (1978): On supplementing the selenium intake of New Zealanders. 2. Prolonged metabolic experiments with daily supplements of selenomethionine, selenite, and sh. Br. J. Nutr. 39, 589±600. Salonen JT, Alfthan G, Huttunen JK, Pikkarainen J & Puska P (1982): Association between cardiovascular death and myocardial infarction and serum selenium in a matched-pair longitudinal study. Lancet ii, 175±179. Salvatore D, Bartha T, Harney JW, Larsen PR (1996): Molecular biological and biochemical characterization of the human type 2 selenodeiodinase. Endocrinology 137, 3308±3315. Svensson B-G, SchuÈtz A, Nilsson A, AÊ kesson I, AÊ kesson B & Skerfving S (1992): Fish as a source of exposure to mercury and selenium. Sci. Tot. Environ. 126, 61±74. Svensson B-G, Hallberg T, Nilsson A, SchuÈtz A & Hagmar L (1994): Parameters of immunological competence in subjects with high consumption of sh contaminated with persistent organochlorine compounds. Int. Arch. Occup. Environ. Health. 65, 351±358. Terwolbeck K, Behne D, Meinhold H, Menzel H & Lombeck I (1993): Increased plasma T4-levels in children with low selenium state due to reduced type I iodothyronine 5 0 -deiodinase activity? J. Trace Elem. Electrolytes Health Dis. 7, 53±55. Thorngren M & AÊ kesson B (1987): Effect of dietary sh on plasma selenium and its relation to haemostatic changes in healthy adults. Int. J. Vitam. Nutr. Res. 57, 429±435.