Page 10: Guidelines for Canadian Drinking Water Quality: Guideline Technical Document – Selenium

Part II. Science and Technical Considerations (continued)

9.0 Health effects

9.1 Effects in humans

9.1.1 Essentiality

Selenium is considered to be an essential element by Health Canada, the Institute of Medicine (Otten et al., 2006) and other international agencies (Expert Group on Vitamins and Minerals, 2002; National Institutes of Health, 2011; WHO, 2011). It plays an important role in antioxidant defences, the immune response and the regulation of thyroid hormones by being an integral part of various selenoproteins (Zeng et al., 2009; Rayman, 2012). These selenoproteins participate directly in deoxyribonucleic acid (DNA) transcription, protein synthesis and maturation, calcium flux and oxidant scavenging (Forceville, 2006). Plasma selenium concentrations of 70-90 µg/L are considered adequate for enzyme function (Institute of Medicine, 2000).

The Institute of Medicine established a Recommended Dietary Allowance (RDA) for selenium for adolescents and adults (14-70 years of age) in Canada and the United States of 55 µg/day (Otten et al., 2006). This amount is based on two human intervention studies (Chinese male population deficient in selenium supplemented with 10-90 µg selenomethionine for 8 months, and a New Zealand population with low selenium intake given selenomethionine for 20 weeks), which demonstrated the amount needed to maximize glutathione peroxidase activity (Institute of Medicine, 2000). The average dose of selenium required to reach a plateau in glutathione peroxidase activity in both studies corresponded to 45 µg/day (including a weight adjustment for North American adults). This value is converted to the RDA of 55 µg selenium per dayFootnote 1 for people 14 years of age and over. The enzyme activity was chosen as the endpoint because a deficient activity was observed in a Keshan disease-endemic area in relation to low selenium intake.

For young people (1-18 years old), the RDA values increase progressively from 17 to 55 µg/day. For infants (0-12 months old), the RDA value is 15-20 µg/day, and for pregnant and lactating woman, it is 59-70 µg/day (Institute of Medicine, 2000).

Selenium deficiency is not expected in Canada, as food is the main source of intake, and the Canadian TDS (Health Canada, 2011) shows that the population is meeting the RDAs established by the Institute of Medicine. Natural health products (i.e., via supplementation with selenium) can also contribute significantly to the daily intake.

9.1.2 Deficiency

Selenium deficiency leads to decreased glutathione peroxidase activity, resulting in increased risk of inflammation and atherosclerotic diseases and possibly leading to an increase in the occurrence of chronic diseases (Turner and Finch, 1991; Blankenberg et al., 2003; Kohrle and Gartner, 2009).

At an average intake level of 10 µg/day, selenium is suspected to cause Keshan disease, which is characterized by cardiomyopathy (Yang, 1984), and Kashin-Beck disease, which is characterized by rheumatism. Selenium is also associated with a form of cretinism related to hypothyroidism (Spallholz, 2001; WHO and FAO, 2004; Xia et al., 2005). Keshan disease incidence was lowered and prevented by the administration of selenium in the diet in areas with very low intake of selenium in China (Yang, 1984; Cheng and Qian, 1990; Patterson and Levander, 1997). However, viruses and other nutrient deficiencies are thought to act as contributing factors together with selenium deficiency in causing these diseases (Burk, 2002; Rayman, 2008).

Selenium intake lower than 10 µg/day can lead to muscle weakness, myalgia and heart failure in some infants (WHO and FAO, 2004).

9.1.3 Acute toxicity

The most common manifestations of acute selenium toxicity are vomiting, muscle cramps, fatigue and diarrhoea. Acute selenium toxicity can also be life threatening when the following symptoms occur: pulmonary oedema, comatose state, digestive tract disturbances, heart irregularities, and renal and liver dysfunction.

Some human acute intoxications related to selenium exposure involve accidental or suicidal ingestion of products containing selenium, such as metal gun bluing agent (metal polisher). The main toxic agent in metal polishers, which consist of 2-9% selenic acid (or selenium acid) and 2-4% copper in an acid solution, is believed to be selenium. However, the product is corrosive and has a pH of 1; hence, gastric burns, among other effects, are expected.

Classical case studies are reported in Nuttall (2006), and four of these are described below.

Another complete recovery was observed after a 56-year-old man (weight not mentioned in the report) ingested 1.7 g of selenite. He vomited 1 hour after the ingestion and was admitted to hospital 4 hours later. He complained of abdominal pain, his oropharyngeal mucosa was erythematous, and electrocardiography showed a T-wave flattening. Most symptoms resolved after 2 weeks (Gasmi et al., 1997).

A 15-year-old girl (weighing 52.5 kg) ingested 400 mL of a concentrated sheep drench (vermifuge) containing selenate at a concentration of 5 mg/mL, corresponding to a dose of 22.3 mg/kg bw. She vomited 20 minutes after the ingestion and was sent to the hospital, where she was treated and recovered. Abnormal electrocardiogram (T-wave flattening) peaked at day 3 and gradually disappeared after 2 weeks. Levels of the liver enzymes aspartate aminotransferase, bilirubin and alkaline phosphatase were disturbed, but were found to be mostly normal after 1 week. The serum selenium concentration was 3100 µg/L on admission, 2100 µg/L on day 3 and 480 µg/L on day 4. The girl recovered without sequelae (Civil and McDonald, 1978).

A 40-year-old woman (weight not mentioned in the report) ingested more than 90 mL of gun bluing agent (4% selenic acid, 2.5% cupric sulphate in hydrochloric acid). She vomited and exhibited stomach haemorrhage, pulmonary oedema and kidney congestion. Death by heart failure occurred 8 hours after the incident, and her postmortem blood selenium concentration was 2600 µg/L. Tissue levels of selenium were 9-90 times those of a normal patient, whereas tissue levels of copper were about 2 times those of a normal patient. The highest levels of selenium were found in the lung (12.7 µg/g wet weight, compared with 0.15 µg/g in a normal patient) and the kidney (14.2 µg/g wet weight, compared with 1.09 µg/g in a normal patient) (Matoba et al., 1986).

The ingestion of 15 mL of a gun bluing agent by a 2-year-old boy (weight not mentioned in the report) caused multiple vomiting and diarrhoeal episodes. Initial toxic manifestations lasted for a few days and were associated with a plasma selenium concentration of 285 µg/L. Symptoms included a comatose state, digestive tract burns, heart irregularities, renal and liver dysfunction and metabolic acidosis. The boy eventually developed respiratory complications and died (Nantel et al., 1985).

Moderate intoxication has been associated with blood selenium concentrations of 1500-3100 µg/L (Civil and McDonald, 1978; Lombeck et al., 1987), whereas transient gastrointestinal disturbances occur at blood selenium concentrations of 410-930 µg/L. Ingestion of high doses of selenium contained in marketed products can be fatal; however, recovery has also been seen. Fatalities are related to the quantity ingested and blood selenium levels (Civil and McDonald, 1978; Lombeck et al., 1987; Gasmi et al., 1997; Kise et al., 2004). Postmortem blood concentrations in fatal cases have been shown to be higher than 2600 µg/L (Koppel et al., 1986; Schellmann et al., 1986).

9.1.4 Case studies: intoxication with supplements

Symptoms similar to those of long-term selenosis (hair and fingernail loss, garlic odour of the breath, gastrointestinal disturbances, irritability and fatigue) can occur following the self-administration of selenium tablets as supplements.

Inorganic selenite at a median dose of 41 mg/day from a natural health product was consumed by 210 Americans for about 2-4 weeks. One subject was hospitalized, and the majority had typical selenosis symptoms (e.g., nail abnormalities, hair loss, garlic breath, gastrointestinal disturbances) (MacFarquhar et al., 2010).

A study was performed with an intention-to-treat analysis in severe septic shock patients with an infection. Patients were injected with selenite (4000 µg on the first day, 1000 µg on each of 9 subsequent days) using continuous intravenous infusion for 10 days. No adverse events related to selenite were recorded (Forceville et al., 2007). The rationale behind the treatment relates to the fact that a drop in serum selenium level has been reported in critically ill patients and those with severe septic shock (Geoghegan et al., 2006). This could be associated with a decrease in antioxidant defences, which would be counteracted by the injection. In contrast, the pro-oxidative effect of selenium could also inhibit the binding of NF-κB to the DNA, reducing the inflammatory response.

A 36-year-old man (weight not reported) took two tablets hourly for a few days, then 10 tablets per day, each containing selenium at concentrations between 2500 and 5000 µg/g (R.F. Clark et al., 1996). He had diarrhoea, fatigue and paraesthesia (tingling sensations in extremities) and lost some hair during the first week and became bald after 2 weeks. His nails became discoloured thereafter. After he stopped the supplements for 2 weeks, he experienced hair regrowth, his neurological symptoms disappeared and his blood and clinical chemistry tests became normal. The serum selenium level was 8.26 µmol/L (normal 0.70-1.65 µmol/L), which is equivalent to 650 µg/L using a conversion factor of 78.74 µg/µmol.

A 55-year-old woman took approximately 24 mg of selenium (species and total intake period not specified) daily. She had diarrhoea for 6 weeks, followed by hair loss for 2 weeks. She also had muscle cramps, joint pain, fatigue and difficulty concentrating (Sutter et al., 2008).

A 57-year-old woman took tablets containing 31 mg of elemental and/or organic selenium and other vitamins and minerals daily for 70 days (FDA, 1984). After 2 months, she had brittle nails, hair and nail loss, swelling and purulent discharge of the fingertips, nausea and vomiting, a sour milk breath odour and increased fatigue. No information was provided on her recovery.

Patterson and Levander (1997) reported a case study (original study sources not mentioned) that described an attempt to prevent Kashin-Beck disease by administering 250 µg of selenite to Chinese children for 60 days, followed by 500 µg/day for another 60 days. No adverse effects were observed.

A 62-year-old male took 913 µg selenite tablets daily for 2 years and developed fragile nails and garlic breath odour (Yang et al., 1983). The symptoms disappeared after he stopped taking the tablets.

In a randomized, placebo-controlled interventional study, 88 healthy American adults ingested selenite, selenomethionine or selenium-enriched yeast at 200, 400 or 600 µg/day for 16 weeks. No adverse effects were observed (Burk et al., 2006).

Overall, symptoms similar to those of chronic exposure to selenium have been observed after high selenium intakes from supplements.

9.1.5 Chronic exposure

Chronic exposure to high selenium doses causes toxicity characterized as selenosis. Selenium has also been suggested to affect other body functions, such as glucose metabolism. However, these studies present limitations, and definitive conclusions cannot be drawn.

9.1.5.1 Selenosis

Long-term exposure to high levels of selenium in food results in selenosis, characterized by hair loss, nail anomalies or loss, skin anomalies, garlic odour of the breath, tooth decay and, more severely, disturbances of the nervous system.

The most complete database on selenosis comes from a fairly large population living in a seleniferous area located in Enshi County in China.

In 1961-1964, an outbreak of selenosis affected 49.2% of the 248 inhabitants of five highly exposed villages of Enshi County, with a daily selenium intake level averaging 4.99 mg (3.20-6.69 mg), or 94.4 µg/kg bw per day (Yang et al., 1983; Patterson and Levander, 1997). The levels of selenium intake were calculated after collecting information on the dietary habits of the subjects observed and analysis of typical cereals and vegetable samples. The selenium concentrations in food ranged between 4.0 and 11.9 mg/kg. This high selenium intake period occurred concurrently with a famine. The affected individuals showed symptoms of selenosis, characterized by hair and fingernail loss, garlic odour of the breath and, more severely, disturbances of the nervous system (fatigue, irritability, peripheral anaesthesia, numbness in the extremities, convulsions), skin changes and gastrointestinal upset. The symptoms disappeared after the diet was modified. The levels of selenium intake were estimated based on an analysis of selenium concentrations in cereals and vegetables (organic forms of selenium), urine and whole blood. In general, the concentration of selenium in drinking water for the study population represented 2-3% of the total selenium intake. The mean concentration of selenium in 11 drinking water samples was 54 µg/L (Yang et al., 1983). Among these, four samples were from the surface water of a village with a high prevalence of selenosis and had a mean selenium concentration of 139 µg/L (117-159 µg/L). The remaining seven samples were from different sources and had an average selenium concentration of 5 µg/L. The soil in the area derived from coal had a mean selenium concentration of 300 mg/kg and is responsible for the high selenium content of the plants. Samples were collected at three periods. The selenium concentrations in 10 coal samples from the selenosis-endemic area were 291 mg/kg in 1966, 367 mg/kg in 1967 and 84 123-92 800 mg/kg in 1978. The famine that occurred at that period resulted from a drought, forcing villagers to consume more vegetables and maize and less rice and other proteinous food, which is hypothesised to have contributed to the potency of the selenium toxicity (Yang et al., 1983).

In a subsequent study in Enshi County during 1985-1986, the average daily intake of selenium, based on lifetime exposure, was categorized as low (70 µg), moderate (200 µg) or high (1300 µg) in 50-75 families (n = 349 adults) (Yang et al., 1989a). To estimate total selenium intake, a questionnaire on food habits and history of toxic manifestations was administered to the families, and selenium was measured in a wide variety of food items (cooked and raw). A regression equation was then derived based on the correlation between intake estimates and selenium concentrations in tissues (whole blood, urine, hair, fingernails and toenails). This equation allowed the estimation of selenium intakes in individuals with known blood selenium concentrations. Selenosis symptoms were classified as + or ++ based on severity (++ being more severe than +) of fingernail and skin changes and hair loss (Yang et al., 1989b). The prevalence and severity of symptoms were not related to blood selenium concentration in a dose-response fashion, showing interindividual variability. These symptoms were not present in individuals with a blood selenium concentration of 1000 µg/L or below. Subjects with blood selenium concentrations in the range of 1000-2000 µg/L displayed selenosis symptoms with a severity of ++ (3-7%) and + (10-35%). Forty-five percent of individuals with blood selenium concentrations of 2000-3300 µg/L or higher displayed selenosis symptoms with a severity of + only (none were reported to have selenosis symptoms with a severity of ++). Symptoms were mostly (97% of the time) present in individuals older than 18 years, and no symptoms were present in individuals under 12 years of age. Prolonged prothrombin time was observed in 45% of individuals with a blood selenium concentration above 1000 µg/L. Persistent selenosis symptoms were observed in five Chinese individuals with blood selenium concentrations ranging between 1054 and 1854 µg/L. The authors calculated that a blood selenium concentration of 1054 µg/L corresponded to an intake of 910 µg/day and identified these as the minimum blood selenium concentration and minimum selenium intake causing toxicity, respectively.

Those same five patients were chosen for a follow-up assessment in 1992 because they had the lowest intake level showing clear signs of selenosis and their symptoms were persistent (Yang and Zhou, 1994). Between 1986 and 1992, part of their diet and corn consumption were replaced by imported grains (rice and market cereals) with a lower selenium content. By 1992, their symptoms had disappeared, and their blood selenium concentrations had dropped from an average of 1346 µg/L to 968 µg/L, the latter representing a mean intake of 819 µg/day. The authors set the no-observed-adverse-effect level (NOAEL) at 819 µg/day. This value was rounded to 800 µg/day by the Institute of Medicine to derive the tolerable upper intake level (UL). The details provided on the selenium intake in relation to selenosis symptoms in the Yang et al. (1989a,b) study and the follow-up study of five recovered sensitive individuals (Yang and Zhou, 1994) are sufficient to allow a risk assessment for selenium to be performed.

Some possible limitations to the findings of studies by the Yang and co-workers (Yang et al., 1989a,b; Yang and Zhou, 1994) were identified. For example, the health measurements of these observational studies were not exhaustive. No information was provided on the impact of the famine and the low protein content diet on the health effects of selenium in the population. Moreover, although the cohort was exposed to relatively high levels of selenium in drinking water, the main source of exposure was through the diet, in which selenium in the organic form selenomethionine is predominant. The burning of coal for cooking could have contributed to the selenium intake via inhalation, but no quantitative information was provided.

Although these potential biases could have influenced the quantification of the exposure, the selenium toxicity data from these studies on a relatively high number of subjects (n = approximately 400) are useful in the analysis of the dose-response. The use of this study is also supported by the relevance and the causality of selenium intake with the adverse health outcomes. The correlation between selenium content in food measured in multiple items consumed by the subjects with levels of selenium in various tissues allows a quantitative estimation of the intake and identification and use of a NOAEL in the risk assessment of selenium.

Other studies, supporting the findings of the studies by Yang and co-workers (Yang et al., 1989a,b; Yang and Zhou, 1994), are described below.

Over a 2-year period, 142 subjects of both sexes from South Dakota and eastern Wyoming, another area with high levels of selenium, were randomly recruited to participate in a clinical examination survey (Longnecker et al., 1991). This population consumed grains with high selenium concentrations (average 239 µg/day, highest intake 724 µg/day). Concentrations of selenium in blood, serum, urine and nails was correlated with the intake levels. A duplicate-plate food and beverage collection for selenium analysis was also performed. Higher selenium concentrations in whole blood and nails were associated with a decrease in symptoms such as paraesthesia and an increase in lethargy. Selenium intake also correlated with aminotransferase concentrations in the serum. As physical injury in ranchers could influence physiological parameters, the authors included an indicator variable (rancher or not) to assure that individuals were randomly selected. After correction for rancher status, the evidence for selenosis was no longer significant. Despite a lack of details on statistical analysis and study protocol provided by the authors, the doses reported to be associated with no selenosis symptoms are supportive of the results of the studies by Yang and co-workers (Yang et al., 1989a,b; Yang and Zhou, 1994).

No evidence of selenosis was observed in a cross-sectional study involving 448 individuals aged 15-87 years living in the Amazon (Tapajos River basin, Brazil) with a diet high in selenium (Lemire et al., 2012). Blood selenium concentrations ranged between 103 and 1500 µg/L. Because the population was also exposed to mercury, the authors suggested that this co-exposure might be protective of selenosis.

In a survey that compared the health status of 111 children from a rural, seleniferous zone in Venezuela with that of 50 urban children from the city of Caracas (Jaffe et al., 1972), the rural children showed a higher concentration of selenium in blood (813 µg/L versus 355 µg/L) and urine (636 µg/L versus 224 µg/L). Although some selenosis symptoms were present in children from the rural zone (dermatitis and nail anomalies), the authors concluded that there was no clear sign of toxicity that could be attributable to selenium because of the presence of multiple differences between the two subpopulations.

No evidence of selenosis was observed in Inuits of Greenland (n = 222 subjects of both sexes) (Hansen et al., 2004). Mean whole blood selenium concentrations were in the range of 80-1890 µg/L, and whale skin was found to be the main source of intake, with a selenium concentration of 47.9 µg/g.

In conclusion, selenosis effects were observed and characterized in a Chinese population having high intakes of selenium from food (Yang et al., 1989a,b; Yang and Zhou, 1994). No selenosis symptoms were observed in subpopulations of children in the studies from China, Venezuela, Greenland or the Amazon.

9.1.5.2 Effects on cancer

Selenium is not considered to be a carcinogen, and multiple studies have focused on its anticarcinogenic potential (Hurst et al., 2012; Rayman, 2012).

A Cochrane Systematic Review of several observational studies and randomized controlled trials suggested a protective effect of selenium against cancer (Dennert et al., 2011). The meta-analysis of 13 prospective observational studies showed a reduction in cancer incidence (odds ratio [OR] = 0.69, 95% CI = 0.53-0.91) and in cancer mortality (OR = 0.55, 95% CI = 0.36-0.83) in both sexes. The organs with the greatest reduction in cancer risk were the bladder (OR = 0.67, 95% CI = 0.46-0.97), lung (OR = 0.75, 95% CI = 0.54-1.03) and prostate (OR = 0.78, 95% CI = 0.66-0.92). The pooled results from the observational studies were suggestive of a slight protective effect for cancer incidence in individuals with a higher selenium status (blood, nail and hair biomarkers of exposure) compared with those with a lower status (OR = 0.69, 95% CI = 0.53-0.91). The authors stated that it is difficult to conclude on the validity of this tendency because of the heterogeneity of the data. The review also included an analysis of six randomized clinical trials focusing on prostate cancer, non-melanoma skin cancer and liver cancer, among them the Nutritional Prevention of Cancer (NPC) and the Selenium and Vitamin E Cancer Prevention Trial (SELECT), the two largest trials using selenium-only supplementation. Of the six randomized clinical trials, two were conducted with selenium-enriched yeast for the prevention of non-melanoma skin cancer, one with selenomethionine for the prevention of prostate cancer and three with selenium-enriched yeast or selenite for the prevention of liver cancer. A protection against all-cancer, colorectal, lung and prostate cancer mortality and prostate cancer incidence was observed in the NPC and in the observational study, the Third National Health and Nutrition Examination Survey (NHANES III), respectively. After a mean follow-up of 7.4 years, the subjects receiving selenium supplements in the NPC study had an increase in non-melanoma skin cancer (relative risk [RR] = 1.17, 95% CI = 1.02-1.34) and squamous cell carcinoma (RR = 1.25, 95% CI = 1.03-1.51). The increase in squamous cell carcinoma incidence was also observed in the subjects in the highest tertiles (105.6-122.0 ng/mL and > 122.4 ng/mL) (RR = 1.49, 95% CI = 1.05-2.12; and RR = 1.59, 95% CI = 1.11-2.30). However, these results should be interpreted cautiously, as described below. The increase in non-melanoma skin cancer incidence was mainly observed in a subgroup, located in one cancer centre (Macon, Georgia), receiving 200 µg of selenomethionine per day of the trial (Duffield-Lillico et al., 2003; Reid et al., 2008). The subjects consisted of patients with skin cancer history and skin that had sustained heavy sun damage (Duffield-Lillico et al., 2003; Reid et al., 2008). No dose-response relationship for non-melanoma skin cancer or squamous cell carcinoma risk was observed in relation to the administered dose: the group receiving 400 µg selenomethionine per day had no significant difference in cancer risk, with hazard ratios (HR) of 0.91 (95% CI = 0.69-1.20) and 1.05 (95% CI = 0.72-1.53), respectively. After correction for cases that occurred at the beginning of the trial, no significant increase in squamous cell carcinoma risk was seen (HR not given by the author) (Dennert et al., 2011). The reasons for the different effects observed between doses and trial locations remain unclear and were suspected to be related to chance, as distribution of factors was similar between sites (Dennert et al., 2011).

A meta-analysis of 12 studies (9 on plasma/serum selenium, 3 on toenail selenium; total = 13 254 individuals) looked at the relationship between prostate cancer risk and selenium biomarkers (Hurst et al., 2012). Randomized controlled trials, case-control studies and prospective cohort studies were included in the analysis. Increasing plasma/serum selenium concentration (up to 170 ng/mL) was related to a decrease in prostate cancer risk. Similarly, toenail selenium concentrations between 0.85 and 0.94 µg/g indicated a reduction in prostate cancer risk (estimated RR = 0.29; 95% CI = 0.14-0.61). However, toenail concentrations higher than 0.94 µg/g seemed to increase cancer risk, but the authors gave no details or comments on the tendency.

The NPC trial is a randomized, double-blind clinical trial that was designed to address skin cancer incidence following supplementation with selenium-enriched yeast (consisting mostly of selenomethionine) at 200 µg selenomethionine per day for a mean of 4.5 years between 1983 and 1991 (L.C. Clark et al., 1996, 1998). The population consisted of 1312 American male patients with skin cancer history (Duffield-Lillico et al., 2003). No skin cancer protection was observed, but significant reductions in total (HR = 0.75, 95% CI = 0.58-0.97) and prostate cancer incidence (HR = 0.48, 95% CI = 0.28-0.80) were observed compared with placebo. In a 2002 follow-up study, results were presented from 1991 through 1996 (Duffield-Lillico et al., 2002). They confirmed the cancer protection trend for total and prostate cancer incidence. However, lung (HR = 0.74, 95% CI = 0.44-1.24) and colorectal (HR = 0.46, 95% CI = 0.21-1.02) cancer incidences were no longer significantly reduced. The decrease in prostate cancer incidence was statistically significant in men with initially low baseline serum levels of selenium (< 121.2 ng/mL) and prostate-specific antigen levels (< 4 ng/mL) (Clark et al., 1998). The cancer protection observed in the NPC trial is restricted to men with plasma selenium concentrations lower than 121.6 µg/L (Duffield-Lillico et al., 2002).

SELECT is a phase III randomized, double-blind trial including 35 533 men older than 50 years without clinical evidence of prostate cancer from the United States (including Puerto Rico) and Canada (Klein et al., 2003). Selenium (200 µg selenomethionine per day) was administered as a supplement. The men, who had prostate-specific antigen levels below 4 ng/mL at baseline, were followed for 7-12 years with normal digital rectal exams. No lower prostate cancer (HR = 1.04, 95% CI 0.90-1.18) incidences were observed with selenium supplementation compared with the placebo group (Klein, 2009; Lippman et al., 2009). Also, there were no increases in secondary outcome cardiovascular events or mortality risk measurements. However, the trial was stopped early because of statistically non-significant increased risks of prostate cancer in the vitamin E group and of type II diabetes mellitus in the selenium group (RR = 1.07, 99% CI = 0.94-1.22).

NHANES III (1988-1994) was a large longitudinal epidemiological study using a multistage probability cluster designed to compile physical conditions involving a representative sample of the U.S. population (16 573 adults). The mean serum selenium concentration was 125.6 µg/L in 13 887 individuals (Bleys et al., 2008). Concentrations up to 130 µg/L were inversely correlated to all-cause, all-cancer and colorectal, lung and prostate cancer mortality HR, when comparing the tertiles (< 117 µg/L, 117-130 µg/L and > 130 µg/L) of serum selenium concentration by Cox proportional hazards regression. When serum selenium concentration exceeded 130 µg/L, all-cancer and all-cause mortality had a non-significant increase. However, no causality relationship can be drawn based on observational studies.

One study on 2065 residents of Reggio Emilia, Italy, focused on mortality during a 12-year period following a relatively high intake of selenate in the drinking water (7-9 µg/L) (Vinceti et al., 2000). The authors found a slight increase in mortality from neoplasms in people exposed to high levels of selenate (standardized mortality ratio [SMR] = 1.17, 95% CI = 0.96-1.42) compared with the rest of the municipality, who were exposed to an average concentration of selenate in drinking water of 1 µg/L. However, this study had several limitations, such as the lack of statistical analysis, physiological selenium status and adjustments based on lifestyle, smoking and alcohol consumption. In addition, the level of exposure and the characteristics of the control population were not defined. Consequently, no valid conclusions can be drawn from this study.

Other epidemiological studies found an inverse association of selenium with cancer. A lung cancer study in Maryland found an inverse, but non-significant (P = 0.08), association (r = −0.58) between serum selenium concentrations from 25 802 people and risk of lung cancer (OR not provided by the authors) (Comstock et al., 1997). Among another cohort of 10 940 people from the United States, 111 cancer patients were matched to 210 cancer-free controls based on age, sex, race, smoking history and general health status and followed for 5 years (Willett et al., 1983). Selenium concentrations in the serum (loge transformed) were found to be lower in cancer patients (0.129 ± 0.002 µg/mL [standard error of the mean (SEM)]) than in the controls (0.136 ± 0.002 µg/mL [SEM]) (paired t-tests, P = 0.02 for total cancer association with selenium levels). The gastrointestinal cancers had the strongest inverse association with selenium concentrations (cases: 0.114 µg/mL; controls: 0.134 µg/mL; P = 0.01).

The European Prospective Investigation into Cancer and Nutrition (EPIC) investigated the relationship between environmental factors and health outcome in 520 000 people in 10 European countries: Denmark, France, Germany, Greece, Italy, the Netherlands, Norway, Spain, Sweden and the United Kingdom (Allen et al., 2008). Analysis of the data from 959 patients that had prostate cancer at the time of blood sampling and 1059 matched controls with a median age of 60 years was conducted. The geometric mean plasma selenium concentrations were 71.9 µg/L and 70.6 µg/L for the cancer cases and controls, respectively. When comparing the highest quintile (> 84 µg/L) with the lowest quintile (< 62 µg/L) of plasma selenium concentration, no correlation with prostate cancer risk was found (RR = 0.96, 95% CI = 0.70-1.31), with or without adjustment for smoking, alcohol and other factors, independently of the stage of illness.

In conclusion, the human epidemiological studies on selenium and cancer are inconsistent. There is no clear indication that high selenium exposure causes a decrease or an increase in the risk of cancer, and thus no clear indication that there is any effect at all, as concluded by the World Health Organization and the Food and Agriculture Organization of the United Nations in 2004 (WHO and FAO, 2004). Inverse associations between selenium status and cancer risk have been reported in observational studies (Fairweather-Tait et al., 2011; Rayman, 2012); however, trial findings have been mixed. Beneficial effects on cancer are observed in trials in which selenium blood concentrations were relatively low before the selenium supplementation had started (Boosalis, 2008; Rayman, 2012).

9.1.5.3 Cardiovascular effects

Epidemiological evidence for the relationship between selenium and cardiovascular disease is contradictory. Observational studies have generally reported an inverse relationship between selenium status and cardiovascular disease, in particular in populations with low levels of intake (Flores-Mateo et al., 2006; Rayman et al., 2011). In contrast, more recent studies have suggested a U-shaped dose-response relationship, with potential harm occurring at selenium levels both below and above a physiologically adequate range (Stranges et al., 2010; Rees et al., 2012).

In a meta-analysis performed by Flores-Mateo et al. (2006), results from 25 observational (14 prospective cohort and 11 case-control) and 6 randomized studies published between 1982 and 2005 were pooled separately to assess the relationship between selenium biomarkers and cardiovascular health and the efficacy of supplementation in preventing coronary heart disease, respectively. In the randomized clinical studies, a non-significant reduction (RR = 0.89, 95% CI = 0.68-1.17) in risk of coronary events was observed with selenium supplementation; however, the authors concluded that the trials were small, and other nutrient supplements were taken in addition to selenium in most studies, resulting in inconclusive results. The NPC trial was one of them. No increase in risk of all cardiovascular disease was observed after supplementation with 200 µg selenomethionine per day in this study after the entire 7.6 years of follow-up (HR = 1.03, 95% CI = 0.78-1.37) (Stranges et al., 2006). In contrast, a moderate inverse association was seen in the observational studies. Based on the blood or serum selenium status, a 50% increase in selenium was associated with a significant decrease (24%) in the risk of cardiovascular disease (RR = 0.76, 95% CI = 0.62-0.93); however, the authors questioned the validity of this result, because previous observational studies with other antioxidants and vitamins were unreliable and not supported by clinical trials on their efficacy in a relationship with cardiovascular disease prevention.

In more recent cross-sectional studies and longitudinal observational studies, mixed results have been observed, suggesting a non-linear dose-response relationship between selenium and cardiovascular disease (Bleys et al., 2008). Increasing selenium levels in the body were associated with potential protection against cardiovascular disease up to a concentration of 130-150 µg/L in the mortality follow-up study of NHANES III (Bleys et al., 2008), whereas adverse cardiovascular-related effects were observed in cross-sectional analyses of populations with high selenium status, such as those in the United States (Laclaustra et al., 2009, 2010). These data come from serum selenium measurements from 13 887 U.S. adults of both sexes who participated in NHANES III.

However, results from cross-sectional studies should be interpreted with caution, because of the possibility of reverse causation and potential confounding effects by unmeasured or unknown factors. In addition, it should be noted that recent findings from a clinical trial in the United Kingdom (i.e., the UK-PRECISE trial) suggest potential beneficial effects of selenium supplements on blood lipids in a group of elderly volunteers with relatively low selenium status (Rayman et al., 2011). The applicability of these findings to other populations is uncertain.

Furthermore, in post hoc analyses from the NPC trial, selenium supplementation (200 μg/day as high-selenium yeast) was not significantly associated with any of the cardiovascular disease endpoints after 7.6 years of follow-up (HR= 1.03, 95% CI = 0.78-1.37) (Stranges et al., 2006).

In a prospective study among 636 individuals suspected to have coronary heart disease, the patients with the highest level of activity of red blood cell glutathione peroxidase had an HR of 0.27 (95% CI = 0.15-0.58; P < 0.001), compared with those with the lowest activity, suggesting a beneficial effect of selenium (Blankenberg et al., 2003).

There is no clear mode of action defining the effects of selenium on the cardiovascular system, and additional studies will need to be conducted to examine the relationship between selenium and cardiovascular disease across a wider range of selenium concentration (Stranges et al., 2010; Rayman et al., 2011).

In conclusion, the results of epidemiological studies and clinical trials on the effects of selenium on cardiovascular disease are mixed and do not suggest a protective effect of selenium on cardiovascular disease at this time (Stranges et al., 2010; Rees et al., 2012). Findings from NPC, SELECT and UK-PRECISE were based on post hoc analyses of the main trial, raising concerns about the robustness of results from secondary endpoints or subgroup analyses of clinical trials (Freemantle, 2001; Brookes et al., 2004).

9.1.5.4 Relationship with diabetes

Several studies have investigated the relationship between selenium and diabetes and have shown mixed results.

Many observational studies have found protective effects in terms of diabetes or dysglycaemia incidence in individuals with relatively high selenium status (Navarro-Alarcón et al., 1999; Stapleton, 2000; Kljai and Runje, 2001; Rajpathak et al., 2005; Bleys et al., 2007; Kornhauser et al., 2008; Akbaraly et al., 2010). For example, a recent pooled longitudinal analysis from two U.S. cohorts showed inverse associations between toenail selenium levels and incident type II diabetes, with a reduced diabetes risk across quintiles of toenail selenium (P for trend = 0.01) (Park et al., 2012). However, other observational and interventional studies, such as the secondary analysis of the Supplementation with Antioxidant Vitamins and Minerals (SU.VI.MAX) trial, found an increased risk of type II diabetes incidence or prevalence among subjects with higher selenium status (Czernichow et al., 2006; Bleys et al., 2007; Laclaustra et al., 2009).

Laclaustra et al. (2009) used NHANES III data from 917 adults aged 40 or older to determine the association between serum selenium levels and diabetes. Diabetes was defined as a self-report of current use of medication or a fasting plasma glucose greater than 126 mg/dL. The mean serum selenium level was 137.1 µg/L. The study found that the odds of having diabetes was significantly higher for the participants in the highest quartile of serum selenium (>147 µg/L) compared with the ones in lowest quartile (<124 µg/L) (OR = 7.64, 95% CI = 3.3-17.5). However, no definitive conclusions on causality can be drawn since the study has several limitations, as follows: (1) the cross-sectional design of the NHANES does not allow temporality in the association (i.e., if the high selenium levels are the cause or a consequence of the association with diabetes); (2) the single measurement of serum selenium levels as a biomarker of exposure reflects short-term intake and may be subject to within-person variability; and (3) the trend for diabetes and cardiovascular diseases risk are non-linear (no dose-response).

Two secondary analyses of randomized clinical trials, which have kept the original randomization design, have observed some positive associations between selenium levels and diabetes incidence. However, it is difficult to draw conclusions from these results, because the trials were not designed specifically to evaluate the effects of selenium on diabetes. The SELECT study (described in section 9.1.5.2) observed a non-significant increased risk of type II diabetes (RR = 1.07, 95% CI = 0.94-1.22, = 0.16) after selenomethionine was administered at 200 µg/day to 35 533 men followed for 7-12 years (Klein, 2009; Lippman et al., 2009). In the NPC trial (described in section 9.1.5.2), however, the observed increase in type II diabetes cases was significant (HR = 1.55, 95% CI = 1.03-2.33, P = 0.03). Moreover, based on plasma selenium levels, a significant dose-related increase in risk (P = 0.038) was observed across the tertiles of plasma selenium baseline concentration (HR = 2.70, 95% CI = 1.30-5.61, in the highest tertile, > 121.6 µg/L) (Stranges et al., 2007).

In both NPC and SELECT, diagnosis of type II diabetes was based on self-report or use of diabetes medication rather than on biomarker data. This may have led to some misclassification (under-diagnosis) of diabetes at baseline or during the trials. However, given the randomized design and blinding, differential misclassification according to treatment assignment is unlikely. The results cannot readily be generalized to the general public because of the selective nature of the participants randomized in these trials. The NPC trial sample consisted of elderly individuals (mean age 63.2 years) with a history of skin cancer from the eastern United States. Finally, the UK-PRECISE trial sample was based on a group of relatively healthy elderly, mostly white, volunteers, aged 60-74 years, recruited from four general practices in different parts of the United Kingdom.

The conclusions drawn from the results of randomized clinical trials on secondary endpoints such as diabetes are not appropriate for the risk assessment of selenium in drinking water. Their weaknesses include the fact that they were designed to assess the anticarcinogenicity effects of one dose of selenium supplements, which does not allow the establishment of a safe level of exposure to an environmental contaminant. Also, the post-analysis association with secondary endpoints is liable to result from confounders and improper design. Overall, the results are mixed, and no definitive conclusion can currently be drawn on the relationship between selenium intake and the onset of diabetes (Stranges et al., 2010; Rees et al., 2012). However, the consistency in the results of these studies does suggest the need for additional research on this topic.

9.1.5.5 Relationship with other diseases

A number of other diseases, such as amyotrophic lateral sclerosis (ALS) (Vinceti et al., 2010) and glaucoma (Bruhn et al., 2009), have shown a potential association with selenium, but the evidence is insufficient to draw conclusions due to methodological issues.

In a population-based case-control study in Reggio Emilia, Italy, a relationship between selenium and ALS was observed (Vinceti et al., 2010). After adjusting for cofactors and other exposure factors, consumption of "drinking water with ≥ 1 µg/L Se" was associated with an increased risk of ALS (RR = 4.2, 95% CI = 1.1-16). Although the study suggests that selenium-induced ALS may occur at low levels of exposure in drinking water, the semi-qualitative and retrospective exposure projections are not appropriate to use for risk assessment. Moreover, a selection bias may have occurred, in that the subjects were sampled from a population where a cluster of ALS and selenium contamination had already been perceived by the investigators.

In a small case-control study (Bruhn et al., 2009), a significantly higher glaucoma prevalence was observed in the middle and highest tertiles (183.5-215.9 ng/mL; 218.5-398.8 ng/mL) of serum selenium concentration in comparison with the lowest tertile (127.3-182.6 ng/mL). However, no dose-response relationship was observed across tertiles of aqueous humour selenium level of the eye. The small sample size and control-case ratio (~1:1) may have hampered adequate adjustment of confounders. The authors also mentioned an internal publication of a trial at the University of Arizona in which a relationship between glaucoma prevalence and blood selenium levels was observed, but gave no further details.

Selenium is recognized as essential to the functioning of the thyroid gland (Johnson et al., 2010). The effects of selenium on thyroid metabolism were evaluated in a review of two cross-sectional studies and three interventional studies in New Zealand (Thomson et al., 2005). Correlations between plasma selenium levels and thyroid status were not significant in the cross-sectional studies. Only one interventional study demonstrated a significant decrease (P = 0.0045) in thyroxine in response to the administration of varying levels of selenomethionine in 52 adults divided into five groups (0, 10, 20, 30 or 40 mg). The objective of the study was to evaluate the effect of selenium supplementation on glutathione peroxidase activity, and the subjects had low levels of plasma selenium (less than 100 µg/L) (Duffield et al., 1999).

9.1.6 Developmental and reproductive toxicity

Research on the developmental and reproductive toxicity of selenium in humans is sparse, and there is no conclusive evidence demonstrating toxicity to these systems. No studies demonstrated the teratogenicity of selenium in humans (OEHHA, 2010). The Institute of Medicine (2000) indicated that there are no reports of symptoms of teratogenicity in infants born to mothers with high levels of intake of selenium (without toxicity).

In Rivalta, Italy, Vinceti et al. (2000) reported the rate of spontaneous abortions and congenital malformations among women exposed to drinking water with concentrations of selenate of 7-9 µg/L for the entire gestational period, compared with women exposed to drinking water with selenate concentrations below 1 µg/L, between 1972 and 1988. Computerized birth records (n = 1974) from the General Registry Office for all Rivalta residents were reviewed, and no adverse effects on human reproduction were observed, except for a non-significant excess rate of abortion (rate ratio = 1.73, 95% CI = 0.62-4.8).

The Western Human Nutrition Research Center of the Agricultural Research Service in California evaluated the effects of selenium on about 30 healthy men aged 18-45 years for 1 year (Hawkes and Turek, 2001). The participants were divided between a high (300 µg/day) and a low (10 µg/day) food-based selenium diet for 99 days, food selenium being replaced by selenite on days 111-117 at their respective levels. In the high-selenium group, sperm motility was decreased by 32% on week 13 and in a lower fashion (17%) on week 17 compared with baseline, or week 0. Sperm concentration and number decreased significantly by more than 50% in both groups. However, environmental and dietary factors were suggested as confounders by the authors. In a recent experiment by the same investigators, with more participants (n = 42) for a longer period (48 weeks), no difference in sperm quality was observed (Hawkes et al., 2009). No effects on thyroid hormone metabolism (blood triiodothyronine and thyroxine hormone levels), body composition (fat free and fat mass) or vascular responsiveness (arterial diameter and blood flow rate) were observed in a cohort of healthy men administered 300 µg selenium-enriched yeast per day for 3-6 weeks (Hawkes et al., 2008).

OEHHA (2010) also reported case studies on sperm quality and quantity. Generally, no correlation or a positive correlation was observed between these parameters and blood selenium concentrations. For example, in a cross-sectional study in the United States, no significant relationship was observed between selenium and the rate of birth defects in 1986 in Nebraska (42 out of 453 communities had drinking water levels higher than 0.01 mg/L). Moreover, no teratogenicity effect of high selenium exposure (subjects with mean urinary levels of 0.38 mg/L) was observed in Venezuela.

In conclusion, there is no clear evidence supporting the developmental or reproductive toxicity of selenium.

9.2 Effects on experimental animals

The effects of exposure of experimental animals to selenium are variable and have been shown to depend on many factors, such as the animal species and type (rodents versus livestock), selenium species, dose, duration and route of exposure, diet, physiological status, presence of other contaminants or nutrients and stress (Valdiglesias et al., 2009). Selenium is also considered essential to animal health (Davis et al., 1999; Nogueira and Rocha, 2011).

9.2.1 Acute toxicity

Acute exposure of laboratory animals to very high levels of selenium results in respiratory failure, liver and kidney necrosis and congestion, and a decrease in locomotor activity (Civil and McDonald, 1978; Griffiths et al., 2006). For selenite, the oral median lethal doses (LD50 values) reported in the literature for mice and rats ranged between 3.2 and 50 mg/kg bw (Morss and Olcott, 1967; Pletnikova, 1970; Cummins and Kimura, 1971; Vinson and Bose, 1981; Plasterer et al., 1985; Griffiths et al., 2006). The oral LD50 in female guinea pigs was 2.3 mg/kg bw, and in female rabbits, 1 mg/kg bw (Pletnikova, 1970).

The form of selenium administered has an impact on its acute toxicity (Nuttall, 2006). Cummins and Kimura (1971) observed large differences in oral LD50 values for Sprague-Dawley rats when different forms of selenium (including oxidation states and solubility) were administered. The oral LD50 values were 7 mg/kg bw for selenite, 138 mg/kg bw for selenium sulphide and 6700 mg/kg bw for elemental selenium. Symptoms occurring within 18-72 hours following administration and prior to death included pilomotor activity, decreased activity, dyspnoea, diarrhoea, anorexia and cachexia (body weight loss). Blood selenium concentrations correlated well with the toxicity observed with the different forms of selenium administered.

Oral LD50 values have been reported in the literature for other forms of selenium: selenocysteine, 35.9 mg/kg bw (male mice); and selenium-enriched yeast, 4.07-37.3 mg/kg bw (rats and mice) (Vinson and Bose, 1981; Sayato et al., 1997). No reliable data were available for selenate administered orally.

In conclusion, acute exposure to selenium has been shown to cause respiratory failure, liver and kidney necrosis, decreased activity, dyspnoea, diarrhoea and death.

9.2.2 Short-term exposure

Short-term oral exposure of laboratory animals to selenium resulted in decreased water consumption, which was considered to be responsible for the subsequent decrease in body weight and renal papillary degeneration. Moreover, hunched posture, decreased weights of heart, spleen and thymus, centrilobular hepatocyte enlargement, hypertrophy of the zona glomerulosa cells of the adrenal glands, growth retardation and disturbances of the cardiovascular, respiratory, renal, endocrine, neurological and tegumentary (skin and hair) systems have been reported.

A study by the U.S. National Toxicology Program (NTP) (Abdo, 1994) exposed F344/N rats and B6C3F1 mice (10 of each sex per concentration) to selenate in drinking water at concentrations of 0, 3.75, 7.5, 15, 30 or 60 mg/L daily for 90 days. The concentrations were estimated to be equivalent to selenium doses of 0, 0.1, 0.2, 0.4, 0.6 and 1.1 mg/kg bw per day (males) or 0, 0.1, 0.2, 0.4, 0.6 and 0.8 mg/kg bw per day (females) for rats and 0, 0.3, 0.5, 0.8, 1.5 or 2.6 mg/kg bw per day for mice. All rats died in the 60 mg/L group (by week 11 for males and week 6 for females). Mean body weights in rats dosed with 30 mg/L and in mice dosed with 30 and 60 mg/L were reduced (13-29%) compared with control animals. Decreased water consumption was observed in rats and mice exposed to 15 mg/L and above. Administration of selenate at concentrations of 7.5 mg/L or higher was associated with increased incidences of renal papillary degeneration in rats, in which dehydration may have played a role. Dehydration was considered to be responsible for the observed decrease in urine volume and increases in erythrocyte counts, haematocrit, haemoglobin concentrations, alanine aminotransferase activities, urea nitrogen and urinary specific gravity in rats. No renal lesions were seen in mice. Blood chemistry values for treated mice were similar to those of the controls.

In the same NTP drinking water study (Abdo, 1994), F344/N rats and B6C3F1 mice (10 animals of each sex per dose group) were also exposed to selenite at concentrations of 0, 2, 4, 8, 16 or 32 mg/L daily for 13 weeks. These concentrations were estimated to be equivalent to selenium doses of 0, 0.08, 0.13, 0.2, 0.4 and 0.8 mg/kg bw per day (males) or 0, 0.08, 0.13, 0.2, 0.4 and 0.9 mg/kg bw per day (females) for rats and 0, 0.14, 0.3, 0.5, 0.9 or 1.6 mg/kg bw per day for mice. Two female rats exposed to 32 mg/L died. Mean body weights were reduced (17-54%) in rats and mice exposed to 32 mg/L compared with controls. Water consumption was reduced in rats and mice with increasing selenite concentration. Selenite induced similar changes in haematology, clinical chemistry and urinalysis parameters in rats as did selenate. These effects may also have been induced by dehydration. Selenite also increased the incidence of renal papillary degeneration in rats at 0.8 mg/L and above. Dehydration may have contributed to the effects observed in rats and mice. Selenite did not cause lesions in mice.

Although some effects on decreased water consumption and renal papillary lesions were observed at a lower dose (7.5 mg/L), the author estimated the NOAEL in rats to be 0.4 mg/kg bw per day for selenate and selenite expressed as selenium, based on mortality, body weight depression, decreased water consumption and renal papillary lesions (Abdo, 1994). The estimated NOAEL in mice was 0.8 mg/kg bw per day for selenate expressed as selenium and 0.9 mg/kg bw per day for selenite expressed as selenium, based on body weight depression and decreased water consumption.

Sprague-Dawley rats (10-13 rats of each sex per dose group) were exposed to selenate in drinking water at 0, 7.5, 15 and 30 mg/L (estimated to be equivalent to doses of 0, 0.5, 0.8 and 1.1 mg/kg bw) for 30 days (NTP, 1996). Female rats were divided among three groups: peri-conception, gestational and vaginal cytology. Maternal and male body weights were reduced in all selenate dose groups. The groups exposed to 0.8 and 1.1 mg/kg bw demonstrated symptoms of toxicity, such as pale and small adrenals, thickened stomach walls, stomach adhesion involving abdominal organs, enlarged and small kidneys, enlarged spleen and implantation sites with nodular material. No NOAEL was determined, as reduced body weight occurred in all treated groups.

Swiss mice were exposed to selenite daily in drinking water at concentrations of 1-64 mg/L for 46 days (Jacobs and Forst, 1981b). Reduced survival and decreased body weight were observed at 64 mg/L, with females being more tolerant than males, whereas concentrations of 1, 4 and 8 mg/L enhanced survival and growth The authors did not identify a NOAEL.

The effect of a selenium-enriched yeast preparation was evaluated in Sprague-Dawley rats (both sexes) and beagle dogs (both sexes) for 28 or 90 days (Griffiths et al., 2006). Selenium in yeast is in the form of selenomethionine (98%). In the 28-day study, rats were dosed with selenium at 0, 0.1, 0.51, 2.0 or 5.1 mg/kg bw per day (n = 6 per group), and dogs were dosed with selenium at 0, 0.045, 0.225 or 1.125 mg/kg bw per day (n = 4 per group). In the 90-day study, rats were dosed with selenium at 0, 0.23, 0.36 or 0.61 mg/kg bw per day (n = 20 per group), and dogs were dosed with selenium at 0, 0.06, 0.2 or 0.6 mg/kg bw per day (n = 8 per group). Another group was also administered selenite at 0.35 mg/kg bw per day (rats) and 0.6 mg/kg bw per day (dogs) in the 90-day study. No mortality occurred as a result of treatment in the groups. In the 28-day rat study, hunched posture was observed in the 0.51 mg/kg bw per day group and higher. The two highest dose groups were terminated because hunched posture occurred on day 4 and onward. Excessive salivation was observed in dogs in the high-dose group, and decreased erythrocyte count, haemoglobin concentration and packed cell volume were observed in the 0.225 mg/kg bw per day dose groups. A reduced dietary intake for all groups of rats was observed in the 90-day study. Moreover, fur loss, hunched posture, a decrease in male rat heart, spleen and thymus organ weights, centrilobular hepatocyte enlargement and hypertrophy of the zona glomerulosa cells of the adrenal glands were observed at selenium doses above 0.23 mg/kg bw per day and in the selenite group. In the 90-day treated dog groups, an emaciated appearance and decreases in erythrocyte count, haemoglobin concentration, packed erythrocyte volume and mean cell hemoglobin concentration were observed at 0.6 mg/kg bw per day. Higher cholesterol levels were observed in the 0.2 mg/kg bw per day group. The NOAELs were 0.1 mg/kg bw per day for rats and 0.045 mg/kg bw per day for dogs in the 28-day study. The NOAELs were 0.23 mg/kg bw per day for rats and 0.06 mg/kg bw per day for dogs in the 90-day study.

Signs of selenosis (loss of hair, lesions of hooves) were observed in livestock (calves, steer and pigs) administered selenite, selenomethionine or selenium-enriched yeast (organic) either in the diet or by gavage (selenite or selenium-enriched yeast at concentrations above 5-20 mg/kg of food for pigs and selenite and selenomethionine at concentrations of 0.28-0.8 mg/kg bw for steers and calves) in short-term studies (O'Toole and Raisbeck, 1995; Kim and Mahan, 2001; Kaur et al., 2003).

OEHHA (2010) reported on various studies relating to the health effects seen in mice and rats as a result of short-term oral exposure (14 days to 3 months) to selenite, selenate and organic compounds such as selenomethionine. Rats were the most sensitive species, and adverse effects observed at concentrations above the NOAELs of 0.026-0.50 mg/kg bw include growth retardation and disturbances of the cardiovascular, hepatic, respiratory, renal and endocrine systems. The highest NOAEL observed for these health effects was 1.67 mg/kg bw for selenite following a 13-week exposure. Organic selenium, such as selenomethionine, induced the same adverse effects at concentrations above the NOAELs of 0.125-0.320 mg/kg bw in rats. These endpoints were also observed in mice, with NOAELs in the range of 0.20-7.17 mg/kg bw. Gastrointestinal, ocular and muscular disturbances were observed at doses above 7.17 mg/kg bw in mice. Selenomethionine induced the same effects, in addition to neurological disturbances, at doses above the NOAELs in the range of 1.36-1.96 mg/kg bw in mice. Pigs, calves and monkeys exhibited the same adverse effects as rats, as well as neurological and dermal disturbances, after oral exposures to doses above the NOAELs of 0.014-1.25 mg/kg bw. Selenomethionine induced these effects at doses above 0.08-1.25 mg/kg bw.

9.2.3 Long-term exposure and carcinogenicity

Most carcinogenicity studies with selenate, selenite and organic selenium compounds have shown negative results in laboratory animals, and exposure to selenium may delay the onset of chemically induced tumours. The characteristic signs of chronic selenium exposure are abnormal hooves, horns and hair.

Alkali disease is the most characteristic manifestation from chronic exposure to selenium (via feed or selenium-accumulating plants) in large animals, such as cattle, steer and pigs (Zhang and Spallholz, 2011). It is characterized by emaciation, stiffness, lameness, loss of hair and cracking of hooves (O'Toole and Raisbeck, 1995). Heart and liver atrophy, anaemia and erosion of bones were also identified as symptoms of alkali disease (Moxon, 1937). Laboratory animals exposed chronically to doses of 0.1-0.57 mg/kg bw exhibited toxic effects characterized by weight loss, renal toxicity and liver toxicity (organ congestion, fatty degeneration of parenchymal cells, hyperplastic lesions, amyloidosis, nephritis, necrosis) (Nelson et al., 1943; O'Toole and Raisbeck, 1995; O'Toole et al., 1996; OEHHA, 2010).

Several chronic studies were undertaken to determine whether selenate or selenite induced tumours in laboratory animals. Schroeder and Mitchener (1971a) administered selenite or selenate in the drinking water at a concentration of 2 mg/L (0.14 mg/kg bw per day)Footnote 2 to Long-Evans rats at weaning. The dose was increased to 3 mg/L (0.21 mg/kg bw per day) at 1 year of age, and dosing continued until natural death occurred. Selenite's innate toxicity generated a high mortality rate among male rats, but a lower rate in females. Selenate did not induce any particular signs of toxicity during the first year. Toxic effects found at autopsy were increases in aortic plaques and serum cholesterol levels in both males and females. Rats fed selenate developed more tumours than control rats (30 versus 20, P ~ 0.001) and more malignant tumours (20 versus 11, P < 0.01) than the controls, and at a younger age.

In contrast, there was no significant difference in tumour incidence when selenite or selenate was administered at 3 mg/L in drinking water to CD Swiss mice for life. However, reduced body weights in females, increased body weights in males and an obvious decline in general health were observed in the mice (Schroeder and Mitchener, 1972).

In a U.S. National Cancer Institute study conducted by Tinsley et al. (1967), 1437 rats of both sexes were divided among 34 different diet groups (food given ad libitum) supplemented with selenite or selenate at concentrations ranging between 0.5 and 16 mg/kg diet with varying protein concentrations for eight durations between 28 and 1150 days. Decreased feed consumption was seen at 4 mg/kg diet and above in males. Decreased growth rate was observed with increasing selenium dose. Selenate decreased body weight in a more pronounced way than did selenite in female rats. Toxic effects of selenium on almost all tissues autopsied, but mainly on the liver (focal lesions, abnormal cells), were observed from 0.5 to 16 mg/kg diet. Other affected organs included the adrenals (vesiculation of the cortex), pancreas (interstitial oedema, hyperplasia and congestion), myocardium (macrophage and lymphocyte infiltration and adventitial tissue proliferation), spleen (congestion, hyperplasia and depletion, reticulosis, sclerosis) and kidney (interstitial nephritis, presence of cysts). The only non-affected parameters were erythrocyte concentration and spleen weight. Higher protein content of the diet reduced the intensity of the lesions. Hyperplasic liver lesions did not disappear after selenium supplementation stopped. Tolerance to selenium (based on life length) was not induced by gradually increasing the selenite or selenate dose to 4 mg/kg diet in comparison with a direct administration of 4 mg/kg diet. A follow-up study analysing the tissues of these rats indicated no induction of neoplasms (Harr et al., 1967).

Another chronic study evaluated the effect of administering selenite in drinking water at a concentration of 1, 4 or 8 mg/L (10.3, 32.3 and 49.8 µg/day) to Swiss mice of both sexes for 47 weeks (Jacobs and Forst, 1981b). No significant effect was seen on survival at any dose. All animals gained weight; however, the 8 mg/L group gained only half of the weight gained by controls. No significant changes in serum chemistry (alanine aminotransferase, aspartate aminotransferase, alkaline phosphatase) were observed. No sign of neoplasia was observed in the groups.

9.2.4 Genotoxicity

The genotoxicity of selenium has been evaluated in a large variety of systems, and positive and negative results as well as protective effects have been noted (Ferguson et al., 2012). There is a large body of evidence suggesting that selenium prevents DNA damage in vitro and in vivo (Davis et al., 1999; Letavayová et al., 2006). Moreover, the element was shown to reduce the toxic effects of several carcinogens, including cadmium and arsenic, in vitro in animal and human cells (Davis et al., 1999; Zhou et al., 2009; Zwolak and Zaporowska, 2012).

Nevertheless, there is evidence that high doses of selenite and selenate can cause genotoxicity in vitro and in vivo based on various tests, such as the Bacillus subtilis rec-assay measuring DNA damage, the Ames Salmonella assay, chromosomal aberration measurements, the comet assay, the terminal deoxynucleotidyl transferase deoxyuridine triphosphate nick end labelling (TUNEL) assay and recombination at meiosis based on crossover frequency in Drosophila melanogaster (Noda et al., 1979; Norppa et al., 1980a,b,c; Whiting et al., 1980; Ray, 1984; Khalil, 1989; WHO and FAO, 2004; Valdiglesias et al., 2009).

9.2.4.1 In vitro findings

The in vitro genotoxicity results are inconclusive and vary as a function of the concentration and form of selenium. Generally, higher concentrations and the species selenite generate positive results (Letavayová et al., 2006).

Different assays demonstrated the genotoxic potential of selenium. For example, selenate and, to a greater extent, selenite induced DNA damage in the Bacillus subtilis rec-assay (Nakamuro et al., 1976) and base-pair substitution in Salmonella typhimurium strains TA100 and TA104 (Noda et al., 1979; Kramer and Ames, 1988). Selenite and selenate have also shown DNA damage with Kada's rec-assay on Bacillus subtilis (Noda et al., 1979). In contrast, an absence of genotoxicity has been reported with the Bacillus and Ames Salmonella/microsome assays for selenite (Lofroth and Ames, 1978; U.S. EPA, 1991a; Valdiglesias et al., 2009).

Selenite was genotoxic and mutagenic by inducing gene conversion, back mutation, mitotic crossing-over, DNA double-strand breakage, frame-shift mutations and aberrant colony formation in the yeast Saccharomyces cerevisiae (Anjaria and Madhvanath, 1988; Letavayová et al., 2006). The genotoxic effects of selenite are hypothesized to be caused by the generation of reactive oxygen species (Letavayová et al., 2008). Also, the genotoxicity of selenite in yeast was enhanced by the addition of glutathione (Anjaria and Madhvanath, 1988). The reaction of selenite with glutathione generates reactive oxygen species, as detailed in Section 9.3 below (Mézes and Balogh, 2009).

Selenite, selenate and selenomethionine were genotoxic to primary human fibroblasts and purified peripheral blood lymphocytes, as measured by DNA fragmentation, chromosomal aberrations, DNA strand breaks, sister chromatid exchange and micronucleus induction (Lo et al., 1978; Ray and Altenburg, 1982; Khalil and Maslat, 1990). Selenite induced DNA damage and chromosomal aberrations (breaks and fragments) in cultured human fibroblasts, whereas selenate induced a weak increase in DNA repair (Lo et al., 1978). Also, selenite and selenomethionine induced chromosomal aberrations (fragments and breaks) and reduction of cell division in primary human lymphocytes (Khalil, 1989; Biswas et al., 2000). Selenite was more clastogenic than selenate. Selenite, selenate and selenide induced chromosomal aberrations and unscheduled DNA synthesis, as a measure of active DNA repair, in Chinese hamster ovary cells (Whiting et al., 1980). The effects were increased by the presence of S9 fractions and the addition of glutathione. The effects were attenuated by antioxidants such as superoxide dismutase, supporting the pro-oxidative effect of selenium, as explained in Drake (2006). The genotoxic effects in Chinese hamster ovary cells were not observed with organic selenium (selenocysteine, selenocystamine and selenomethionine).

In conclusion, the inorganic and selenomethionine forms of selenium have been demonstrated to be genotoxic in vitro, which is hypotesized to result from the increase in reactive oxygen species.

9.2.4.2 In vivo findings

The results from in vivo studies vary and do not demonstrate a clear genotoxicity pattern, although positive results have been observed at high concentrations (in the range of 0.15 mg/kg of food to near the LD50 of laboratory animals).

High selenite doses (5, 10 and 20 µmol/kg bw) injected intraperitoneally induced dose-dependent DNA strand breaks (comet assay) in hepatocytes of Sprague-Dawley rats (n = 5 male rats per dose) (Yu et al., 2006) and sister chromatid exchange and chromosomal aberrations in Chinese hamster (n = 20 of each sex) bone marrow (Norppa et al., 1980c). Norppa et al. (1980c) stated that at these high doses, three hamsters died a few hours after they had been injected with selenium. Another study in male NMRI mice using the same protocol also found chromosomal aberrations in bone marrow cells, but not in primary spermatocytes (Norppa et al., 1980a).

Administration of selenite (0, 0.15 or 2.0 mg/kg of food) for 10 weeks in the diet of Sprague-Dawley rats induced a dose-dependent increase in 8-hydroxy-2′-deoxyguanosine (8-OHdG), an indicator of oxidative DNA damage (Wycherly et al., 2004). This supports the in vitro studies showing the involvement of reactive oxygen species in the induction of DNA damage.

Also, 49 elderly (corresponding to 62- to 69-year-old men) male beagle dogs were fed for 7 months with a normal diet or diet supplemented with selenomethionine or selenium-enriched yeast at an adequate (3 µg/kg bw) or supranutritional (6 µg/kg bw) level (Waters et al., 2003, 2005). DNA strand breaks in cells from the brain and the prostate (> 90% epithelial cells), evaluated by the comet assay, revealed a non-linear DNA damage response--i.e., a U-shape in relation to the dose (Waters et al., 2005). In an earlier publication by the same group using the same protocol, a decrease in DNA damage (comet assay) and an increase in apoptosis (TUNEL assay) were measured (Waters et al., 2003).

In humans, selenite injected or given in oral tablets (0.004-0.050 mg/kg bw per day for 1-13.5 months) did not induce chromosomal aberrations in lymphocytes (Norppa et al., 1980c).

In conclusion, genotoxicity and oxidative effects, or protection against them in vitro and in animals, caused by organic and inorganic selenium vary depending on the dose, the test system and the test method used. Although genotoxicity is observed in vitro and in vivo, the evidence does not suggest that selenium is directly genotoxic. The positive results observed at high doses are hypothesized to be caused by the generation of reactive oxygen species, and not by the direct action of selenium on the DNA, as demonstrated by the increase in 8-OHdG (EFSA Panel on Dietetic Products, Nutrition and Allergies, 2000; Drake, 2006; Letavayová et al., 2006, 2008).

9.2.5 Reproductive and developmental toxicity

Limited studies on the reproductive and developmental toxicity of selenium were found in the literature. The few cases of reproductive and developmental toxicity observed in laboratory animals are associated with maternal toxicity.

9.2.5.1 Reproductive effects

Rosenfeld and Beath (1954) exposed rats for two generations to inorganic selenium as selenate in their drinking water at a concentration of 1.5, 2.5 or 7.5 mg/L (dose not specified by the authors). No adverse effects on reproduction were observed in rats exposed to 1.5 mg/L in drinking water. In the 2.5 mg/L group, however, maternal toxicity (loss of weight, increased mortality) was observed as well as a 50% reduction in the fecundity and fertility of the females. At the high dose (7.5 mg/L), growth and survival of pups were adversely affected. No histological examination of the reproductive organs was done.

Sprague-Dawley rats were exposed to selenate in drinking water at 0, 7.5, 15 and 30 mg/L (estimated to be equivalent to doses of 0, 0.5, 0.8 and 1.1 mg/kg bw per day) for 30 days (n = 10-13 rats of each sex per dose group) (NTP, 1996). Female rats were divided among three groups: peri-conception, gestational and vaginal cytology. Maternal and male body weights were reduced in all selenate dose groups. Reproductive functions (decreases in number and weight of live pups, number of implants and pup survival) were altered only at 1.1 mg/kg bw per day. Only minor effects on male reproductive functions were noted (changes in testicular and epididymis weights at 0.8 and 1.1 mg/kg bw per day). Selenate was not considered a reproductive or developmental toxicant, as it decreased body weight at doses below that at which it affected reproduction.

Water containing selenate at a concentration of 3 mg/L was administered ad libitum to three generations of CD mice. Selenium increased the number of runts and failures to breed in the three generations (Schroeder and Mitchener, 1971b).

Male BALB/c mice were subjected to a Baker's yeast diet supplemented with varying concentrations of selenite (0-1.02 mg/kg) for 4 or 8 weeks (Shalini and Bansal, 2008). Number of spermatozoa and sperm motility were decreased, whereas lipid peroxidation in testis, DNA strand breaks, flagellar defects and fusion of the mid-piece of the sperm were increased at selenite levels considered by the authors to be deficient (0.02 mg/kg) or in excess/supranutritional (1.02 mg/kg), compared with the level considered to be adequate (0.22 mg/kg). Degenerating mitochondria, improper chromatin condensation and DNA strand breaks of the sperm (at 4 weeks) were significantly increased at 0.02 mg/kg (deficient level) compared with excess and adequate levels. The effects were generally more pronounced after 8 weeks. No other health effects were reported by the authors.

A study was conducted on the reproductive ability of male BALB/c mice subjected to a Baker's yeast diet supplemented with varying concentrations of selenite (0-1.02 mg/kg) for 8 weeks (Kaur and Bansal, 2005). Only the mice fed a selenium-deficient diet (0.02 mg/kg) showed a significant decrease in the number of pachytene spermatocytes and young and mature spermatids, in sperm number and in fertility status in comparison with the control mice fed a diet with selenium levels considered adequate (0.2 mg/kg). The mice fed a diet with excess/supranutritional selenium (1.02 mg/kg) had no significant differences in these parameters compared with the control mice.

The increase in lipid peroxidation was observed in another study by the same author using the same protocol of mice exposure (Shalini and Bansal, 2007). Decreases in fertility (P < 0.001) and litter size (P < 0.05) were observed in mice fed a diet with excess/supranutritional selenium (1.02 mg/kg). Increases in NF-κB and nitric oxide (inflammatory indicator) were observed in the selenium-deficient mice only.

High levels of selenite ingested in the diet (2 or 4 mg/kg) by rats for 5 weeks reduced body weight, testicular and caudal epididymis weights and spermatozoid viability, whereas exposure to 4 mg/kg caused a decrease in sperm motility and an increase in abnormalities of the mid-piece region of spermatozoids (Kaur and Parshad, 1994). High variability in sperm abnormality was observed between individual rats.
Selenate in drinking water is not considered to be a reproductive/developmental toxicant, as it decreased rat body weight at doses below that at which it affected reproduction in the NTP study.

9.2.5.2 Developmental effects

Maternal exposure to selenium in drinking water at a concentration of 3 or 6 mg/L did not produce teratogenicity. Retardation of fetal growth was reported when female IVCS mice were given high doses of selenite (6 mg/L) in drinking water 30 days before and 15 days during pregnancy. The study also showed no difference in litter size between the selenium groups. No maternal effects were reported by the authors (Nobunaga et al., 1979).

One study found malformations of offspring after oral selenite (4-19 mg/kg bw) or selenate (17-21 mg/kg bw) administration to pregnant hamsters (Ferm et al., 1990). However, this was associated with severe maternal toxicity (weight loss and exhaustion), leading to a 50% mortality of the animals at the oral selenite dose of 19 mg/kg bw. The increase in encephaloceles (common in this strain) of the fetus and decreased fetal crown-rump length are associated with a decrease in the mother's body weight and/or feed consumption at either single or repeated treatment. As maternal general malnutrition and weight loss are risk factors for developing fetal abnormalities, no conclusion on selenium teratogenicity can be drawn from this study.

Four monkeys were fed a basal semipurified diet supplemented with selenite (200 µg/kg) in their last 2-4 months of pregnancy. The juveniles showed no sign of abnormalities. The female monkeys were rebred and gave birth to another set of four healthy infants (Butler et al., 1988).

Female rats fed supranutritional selenium levels in the diet (selenite at 3 or 4.5 mg/kg) for 8 weeks prior to mating had normal fetal development (Bergman et al., 1990).

In summary, selenium has been shown to cause some reproductive and developmental toxicity in laboratory animals (rats and mice); however, the results are inconsistent across studies, and no conclusions can be drawn.

9.3 Mode of action

The research on the mode of action of selenium is focused mainly on the beneficial and anticarcinogenicity potentials of the element. Its toxicity mechanisms remain unclear and probably result from multiple metabolic pathways. The perturbation of the cell oxido-reduction equilibrium by the decrease in the glutathione/glutathione disulphide ratio, a rise in oxidative events and the replacement of sulphur atoms in proteins lead to toxic effects such as nail and hair abnormalities (EFSA Panel on Dietetic Products, Nutrition and Allergies, 2000; Zhong and Oberley, 2001; Nogueira and Rocha, 2011; Zhang and Spallholz, 2011).

9.3.1 Toxic effects

The molecular mechanisms of the toxicity of selenium remain unclear, and it is suggested that multiple events operate (EFSA Panel on Dietetic Products, Nutrition and Allergies, 2006). Selenium toxicity is largely mediated by its pro-oxidant effect, leading to the production of hydrogen peroxide and reactive oxygen species involved in tissue damage (Spallholz, 1997; Letavayová et al., 2008; Brozmanová et al., 2010). This would result from a depletion in glutathione, S-adenosylmethionine (a precursor of aminopropyl groups and glutathione; it also regulates the activities of multiple enzymes) and levels of vitamin E in the liver (Anundi et al., 1984; Scarlato and Higa, 1990; Expert Group on Vitamins and Minerals, 2003; EFSA Panel on Dietetic Products, Nutrition and Allergies, 2006; Tiwary et al., 2006). In fact, incubation of yeast with 0.25 mmol/L selenomethionine decreased thiol compounds and cell growth. The adverse effects were reversed by the addition of cysteine (Kitajima et al., 2012). Moreover, incubation of hepatocytes with low levels of selenite (30-100 µmol/L) induced an increase in oxygen consumption and oxidized glutathione and a depletion of the electron donor nicotinamide adenine dinucleotide phosphate (NADPH) (Anundi et al., 1984; Kim et al., 2004). In another study, mice were injected with selenite at a dose of 5 nmol/g bw. A marked decrease in adenosylmethionine in the liver was measured, and the inhibition of the methionine adenosyltransferase was suggested as the mechanism of action (Hoffman, 1977). This also leads to a depletion of the sulphydryl group in the liver. The duality of selenium potentially acting as an antioxidative and an oxidative agent was demonstrated in a human hepatoma cell line exposed to selenite in the presence or absence of glutathione inducing/inhibiting compounds (Shen et al., 2000).

Also, both organic and inorganic selenium can interact with sulphur in critical sulphydryl groups within proteins and other molecules, such as glutathione (Spallholz, 1997). This can also lead to the formation of reactive intermediary compounds, such as selenotrisulphides and methylselenide, which react with other thiols, leading to a decrease in glutathione both in vitro and in animals, followed by the generation of the superoxide anion and hydrogen peroxide (Spallholz, 1997; Nogueira and Rocha, 2011; Zhang and Spallholz, 2011). Accordingly, selenium inhibits thiol-containing enzymes, such as methionine adenosyltransferase, succinate dehydrogenase, lactate dehydrogenase and NADP+-isocitrate dehydrogenase (Mézes and Balogh, 2009). The decrease in thiol-containing antioxidant proteins can also result in an indirect generation of reactive oxygen species (Mézes and Balogh, 2009). The increase in reactive oxygen species can lead to a cascade of events, including lipid peroxidation, DNA damage and loss of membrane integrity and permeability (e.g. organelle membrane), leading to lysosomal enzyme release and tissue necrosis (Mézes and Balogh, 2009).

Another theory on the mechanism of toxicity suggests that high levels of selenium result in the replacement of sulphur by selenium (Letavayová et al., 2006). For example, selenium can cause inadequate incorporation of sulphur into amino acids, such as in critical sulphydryl groups of glutathione involved in antioxidant defences (Pickrell and Oehme, 2002). This can inhibit protein synthesis and the function of DNA repair proteins.

Tests done in vitro on tumour cells support the involvement of oxidative events, which may not be specific to cancer cells, in selenium-induced toxicity. The increased production of reactive oxygen species and induction of apoptosis have been observed in vitro in different cancer cell lines (Yoon et al., 2001; Drake, 2006; Guan et al., 2009; Kandas et al., 2009).

The potentially increased risk of diabetes in response to high intakes of selenium is thought to be caused by an excessively high level of antioxidative activity (Steinbrenner, 2011). This would remove hydrogen peroxide, normally acting as an intermediate (second messenger) in the mechanism of pancreatic insulin secretion and in the signal transduction in response to insulin binding to its receptor. Hence, high levels of antioxidant induced by selenium could potentially impair insulin sensitivity.

9.3.2 Beneficial effects

Paradoxically, the beneficial effects of selenium are due, on one hand, to its involvement in antioxidant defences through its incorporation into selenium-dependent enzymes, such as glutathione peroxidase, and, on the other hand, to its pro-oxidant effects, leading to the apoptosis of cancer cells (Valdiglesias et al., 2009). The chemical form of selenium influences its biological activity; organic forms are generally less toxic, being incorporated more easily into selenoproteins (Chen et al., 2000; Spallholz, 2001; Yan and DeMars, 2012). However, both organic and inorganic selenium compounds increase the activity of glutathione peroxidase and the cellular antioxidative potential in vitro and in vivo (Dalla Puppa et al., 2007; Erkekoğlu et al., 2011; de Rosa et al., 2012; Moon et al., 2012).

There is consensus that selenium is an essential element (Foster and Sumar, 1997; Expert Group on Vitamins and Minerals, 2002; CCME, 2009; EFSA Panel on Dietetic Products, Nutrition and Allergies, 2010; WHO, 2011). It is involved in antioxidant defences, the regulation of immune and endothelial cell functions and the regulation of thyroid hormones by being an integral part of various selenoproteins (Zeng et al., 2009). These directly participate in DNA transcription, protein synthesis and maturation, calcium flux and oxidant scavenging (Forceville, 2006). Selenium has also been shown to antagonize the oxidative effects of other metals, such as mercury. Selenomethionine counteracted the decrease in superoxide dismutase and glutathione and the increase in malondialdehyde observed in rats exposed to mercury (Su et al., 2008). Preincubation of C6 glioma cells with 50 µmol/L selenomethionine reduced the levels of reactive oxygen species caused by mercury (Kaur et al., 2009). Likewise, selenite (10 µg/L) protected against the decrease of activity of the thioredoxin reductase in the liver of zebra seabream fish (= 78) (Branco et al., 2012). Selenoprotein activity is dependent on the presence of the amino acid selenocysteine at the catalytic site. Plasma selenium concentrations of 70-90 µg/L are considered adequate for enzyme function (Institute of Medicine, 2000). Selenium deficiency leads to malfunctioning of many systems, resulting in inflammation, atherosclerotic diseases and perhaps an increased prevalence of chronic diseases (Turner and Finch, 1991; Kohrle and Gartner, 2009).

Preincubation of human endothelial cells for 24 hours with sodium selenite at 5-40 nmol/L provided significant protection against the oxidative effects of tert-butylhydroperoxide (Miller et al., 2001). The activities of cytoplasmic glutathione peroxidase (GPX-1), phospholipid hydroperoxide glutathione peroxidase (GPX-4) and thioredoxin reductase were also each induced by selenite (Miller et al., 2001). Preincubation with low concentrations of selenite (30 nmol/L) or selenomethionine (10 nmol/L) protected LNCaP prostate cancer cells from oxidative DNA damage (comet assay) induced by ultraviolet-A or hydrogen peroxide (de Rosa et al., 2012).

Prevention mechanisms of cancer are incompletely understood. Selenium's anticarcinogenic effects are hypothesized to be mediated through multiple mechanisms (Stewart et al., 1997; Redman et al., 1998; Drake, 2006; Jackson and Combs, 2008; Valdiglesias et al., 2009). For example, selenium can alter the expression of phase I and II detoxifying enzymes, inhibit adduct formation, induce apoptosis of cancer cell lines and act as an antiproliferative and antioxidant agent (Lawson and Birt, 1983; McCarty, 1998; Davis et al., 1999; Keck and Finley, 2004; Letavayová et al., 2006; Guan et al., 2009; Jariwalla et al., 2009; Wang et al., 2009). Selenoproteins are thought to be an important element in cancer protection (Diwadkar-Navsariwala et al., 2006; Zeng et al., 2009).

Various tests were done on tumour cells demonstrating that oxidative stress is involved in apoptosis induction. For example, selenite induced a dose-dependent depletion of glutathione and an increase in apoptosis (cell detachment and DNA fragmentation, measured based on the TUNEL assay) at concentrations from 1 to 100 µmol/L (Stewart et al., 1997). Oxidative stress-induced apoptosis was also observed in human hormone-dependent prostate adenocarcinoma cells (LNCaP) and in the Cheng liver cell line exposed to high doses of selenite (Zhong and Oberley, 2001; Kim et al., 2004; Kandas et al., 2009). Also, apoptosis, through an increased production of reactive oxygen species and phosphorylation of p53 (induction of apoptosis in response to DNA damage), mitochondrial depolarization and caspase cleavage were observed in human leukaemia NB4 cells exposed to selenite (Guan et al., 2009). In fact, selenite and methylselenol are pro-oxidants involved either directly or indirectly in the oxidation of enzyme cysteine clusters, such as protein kinase C and glutathione. The direct oxidation of the catalytic centre of protein kinase C by selenite has been shown to induce apoptosis (Drake, 2006). Moreover, the reaction of selenite with glutathione produces the selenide anion (CH3Se−), which may react with oxygen to produce the free radical O−. The increase in production of reactive oxygen species causes a diminution of the glutathione pool, an increase in DNA damage and apoptosis (Drake, 2006; Letavayová et al., 2006).

Adding antioxidants to the cell media in vitro has been shown to inhibit cell mammary tumour cell death induced by selenite (Zhong and Oberley, 2001). Although these are thought to be beneficial effects, in that they inhibit cancer cell growth, they are supportive of the toxic oxidative potential of selenium compounds in normal cells (Zhang and Spallholz, 2011).

Randomized trials across a wider range of selenium status would help determine the optimal levels of selenium intake in the general population to maximize health benefits while avoiding potential chronic toxic effects. Also, optimal intake for any individual is likely to depend on polymorphisms in selenoprotein genes, which may also affect the risk of disease, including coronary heart disease and ischaemic stroke (Alanne et al., 2007; Rayman, 2012). Future work in the field examining the effect of selenium supplements on chronic disease should give attention to the potential interaction between genetic make-up and selenium intake or status.

9.3.2.1 Selenium protection against tumour formation in animals

In several species, selenium has been shown to inhibit the number and size of tumours induced chemically and to delay their age of onset (Jacobs, 1980; Jacobs and Forst, 1981a; Ankerst and Sjogren, 1982; Lane and Medina, 1985; Ip et al., 1991, 2000; U.S. EPA, 1991a; Woutersen et al., 1999; ATSDR, 2003; WHO, 2011).

In order to demonstrate that selenium protects against cancer, inbred female C3H/St mice, a strain that develops mammary adenocarcinomas at high frequency, were dosed with various concentrations of selenite in water over a lifetime, accompanied by selenium-rich or selenium-poor diets. Groups of mice that received selenite at 0.0, 0.1, 0.5 or 1.0 mg/L in water, with a protein-enriched Wayne diet (selenium concentration of 0.45 mg/kg), had a reduced tumour incidence and an increase in the age of the mice at tumour appearance (i.e., delayed tumour onset) in a dose-dependent fashion (Schrauzer et al., 1978). In another study, a group of mice receiving a low-protein Concord diet (selenium concentration of 0.15 mg/kg) had a reduction in tumour size, growth rate and malignancy when selenium was added to drinking water at 0.1, 0.5 or 2.0 mg/L compared with controls (Schrauzer et al., 1978).

Regarding chemically induced tumours, 55-day-old Sprague-Dawley rats were injected intraperitoneally with the mammary carcinogen methylnitrosourea at 50 mg/kg bw (Ip et al., 2000). A diet with either a basal level of selenium in the form of selenite (0.1 mg/kg) or a supplemented diet including methylselenocysteine (3 mg/kg) or the synthetic water-soluble triphenylselenonium chloride (30 mg/kg) was then administered for 6 weeks. Compared with the controls receiving the basal diet (0.1 mg/kg as selenite), methylselenocysteine reduced the total number of mammary premalignant intraductal lesions by 60% (23 lesions versus 57 in the controls). However, methylselenocysteine had no effect on the proliferation potential of those premalignant cells, revealed by cell cycle biomarkers (proliferating nuclear antigen, cyclin D1, replication of DNA). In the methylselenocysteine group, an increase in p27/Kip 1 protein was observed, which may play a role in tumour prevention, considering the protein's inhibitory action on cell cycle transition and its promotion of differentiation. Conversely, triphenylselenonium chloride (30 mg/kg) reduced the number of larger size lesions and cell proliferation.

An increase in activity of the antioxidant enzyme glutathione peroxidase can be involved in preventing cancer and other chronic diseases (Nogueira and Rocha, 2011). Both organic and inorganic selenium (selenite) have increased glutathione peroxidase activities and blood selenium concentrations in six female rhesus monkeys exposed to 0.25-0.5 µg/mL for 11 months (Butler et al., 1990).

In conclusion, selenium has not been shown to be carcinogenic in animal studies. Alkali disease is the most common manifestation from chronic exposure to selenium in large animals, such as cattle, steer and pigs (Zhang and Spallholz, 2011).

9.3.3 Comparison between the effects of organic and inorganic selenium species

Most of the available epidemiological literature is on exposure to organic species of selenium naturally present in food, rather than to the inorganic species. Detailed studies on the mechanistic differences between inorganic and organic selenium species are scarce. Although detailed studies in which animals are exposed to inorganic species exist, more confidence is placed in the human data on exposure to organic selenium, as explained below.

The current body of evidence suggests that organic and inorganic selenium species generally have similar health effects and share some metabolic pathways in humans. As such, both organic and inorganic selenium compounds are metabolized to selenide and incorporated into selenoproteins (Gromadzińska et al., 2008; Rayman, 2012). Preincubation with low concentrations of selenite (30-500 nmol/L) or selenomethionine (10 nmol/L) protected microglial or LNCaP prostate cancer cells from oxidative DNA damage (comet assay) induced by ultraviolet-A, phthalates or hydrogen peroxide (Dalla Puppa et al., 2007; Erkekoğlu et al., 2011; de Rosa et al., 2012). Selenite and selenomethionine did not have a significant effect in terms of change in body weight, tumour weight, apoptosis induction or tumour angiogenesis when administered orally at 3 mg/kg bw in athymic mice that had been injected with a xenograft of epithelial cancer cells, whereas methylseleninic acid decreased cancer growth (Li et al., 2008).

The speciation and metabolism differences between inorganic and organic forms of selenium have recently been highlighted in a review (Weekley and Harris, 2013). Although both forms can be metabolized to a common metabolite (selenide), they differ in their reduction pathways. Studies on cancer cell apoptosis have shown that selenite reduction is associated with an increase in reactive oxygen species (ROS) production at high level of exposure, while organic selenium follows more complex schemes of trans-sulfuration and methylation involving many lyases. All dietary forms of selenium have been shown to generate a variety of ROS and to increase radical scavenging activity (antioxidant selenoproteins). The authors suggest that more information is needed to understand the biological relevance of the differences in metabolism pathways, although their focus is on disease prevention and/or treatment.

Organic and inorganic species increase glutathione peroxidase activity and reduce levels of reactive oxygen species in vitro and in vivo when administered orally to experimental animals or humans (Thomson et al., 1982; Pehrson et al., 1999; Dorea, 2002; Kaur et al., 2005; Xia et al., 2005; Gromadzińska et al., 2008; Abedelahi et al., 2010). Moreover, Keshan disease was prevented in Chinese individuals supplemented with selenite (Yang, 1984; Cheng and Qian, 1990).

In addition, both organic and inorganic selenium species deplete thiol compounds such as glutathione and induce oxidative stress and selenosis symptoms in vitro and in vivo in humans and experimental animals at high doses (O'Toole and Raisbeck, 1995; Reid et al., 2004; Griffiths et al., 2006; Forceville, 2007; MacFarquhar et al., 2010; Kitajima et al., 2012; Misra et al., 2012). Both selenite and selenomethionine at 1 µmol/L interfered with the reactive oxygen species-induced cascade of phosphorylation in muscle cells in response to insulin (Pinto et al., 2011).

However, as described in Section 9.2, some differences in effects have also been observed between inorganic and organic selenium species. These could differently affect genetic expression, as demonstrated by the variations in gene expression levels (measured in the gastrocnemius muscle, cerebral cortex and liver with the microarray assay) induced in mice by adding 1 µg/kg bw of selenite, selenomethionine or selenium-enriched yeast to food for 100 days (Barger et al., 2012). However, mechanistic studies documenting the differences between the effects of selenium species are lacking, and no definitive conclusions can be drawn at the present time.

Page details

Date modified: