INTERNATIONAL PROGRAMME ON CHEMICAL SAFETY WORLD HEALTH ORGANIZATION SAFETY EVALUATION OF CERTAIN FOOD ADDITIVES WHO FOOD ADDITIVES SERIES: 42 Prepared by the Fifty-first meeting of the Joint FAO/WHO Expert Committee on Food Additives (JECFA) World Health Organization, Geneva, 1999 IPCS - International Programme on Chemical Safety STEVIOSIDE First draft prepared by Dr Josef Schlatter Swiss Federal Office of Public Health, Switzerland Explanation Biological data Biochemical aspects Absorption, distribution, and excretion Biotransformation Effects on enzymes and other biochemical parameters Toxicological studies Acute toxicity Short-term studies of toxicity Long-term studies of toxicity and carcinogenicity Genotoxicity Reproductive toxicity Developmental toxicity Studies on metabolites: Steviol Absorption, distribution, and excretion Effects on enzymes and other biochemical parameters Acute toxicity Genotoxicity Developmental toxicity Special studies Cariogenicity Renal function and vasodilatation Observations in humans Comments Evaluation References 1. EXPLANATION Stevioside is a glycoside of the diterpene derivative steviol (ent-13-hydroxykaur-16-en-19-oic acid). Steviol glycosides are natural constituents of the plant Stevia rebaudiana Bertoni, belonging to the Compositae family. The leaves of S. rebaudiana Bertoni contain eight different steviol glycosides, the major constituent being stevioside (triglucosylated steviol), constituting about 5-10% in dry leaves. Other main constituents are rebaudioside A (tetraglucosylated steviol), rebaudioside C, and dulcoside A. S. rebaudiana is native to South America and has been used to sweeten beverages and food for several centuries. The plant has also been distributed to Southeast Asia. Stevioside has a sweetening potency 250-300 times that of sucrose and is stable to heat. In a 62-year-old sample from a herbarium, the intense sweetness of S. rebaudiana was conserved, indicating the stability of stevioside to drying, preservation, and storage (Soejarto et al., 1982; Hanson & De Oliveira, 1993). Stevioside and its aglycone steviol may act in plants as a feeding deterrent, e.g. against the aphid Schizaphis graminum. The EC50 of stevioside was 650 mg/kg; steviol was more active, with an EC50 of 150 mg/kg. Steviol lost its deterrent activity after acetylation or glycosylation of the C-13 tertiary hydroxy group or methylation of the C-19 carboxylic acid substituent, but the activity of steviol was not greatly affected by modification of either the C-16 exomethylene group or its stereochemistry (Nanayakkara et al., 1987). The biochemical pathway for the formation of steviol in S. rebaudiana is partly known (Kim et al., 1996), and a simple, efficient method for the extraction of steviol glycosides has been described (Liu et al., 1997). The chemical structure of stevioside (Nanayakkara et al., 1987; Suttajit et al., 1993) is shown in Figure 1.2. BIOLOGICAL DATA 2.1 Biochemical aspects 2.1.1 Absorption, distribution, and excretion Rats 3H-Stevioside (specific activity, 13 or 46 µCi/mg), administered by gavage to groups of three to seven Wistar rats at a dose of 125 mg/kg bw (10-120 µCi/kg bw), was absorbed slowly, a maximal blood concentration of 4.8 µg/ml being reached by 8 h. At 4 h, the highest concentration was found in the caecum (280 µg/g stevioside equivalent). At 24 h, the concentrations of radiolabel were low in most organs, including blood, corresponding to about 2 µg/g or ml stevioside equivalent, except in liver (5.7 µg/g), adrenal gland (12 µg/g), small intestine (8.8 µg/g), caecum (40 µg/g), large intestine (12 µg/g), and fat (12 µg/g). The elimination half-life was 24 h. At 48 h, 31% of the radiolabel remained in the body. By five days, 68% had been excreted into the faeces, 24% in expired air, and 2.3% in the urine. In bile-duct cannulated rats, biliary excretion was low up to 24 h, increasing rapidly thereafter to reach 41% of the dose after three days. The authors concluded that stevioside is absorbed from the gut very slowly, that enterohepatic circulation occurs in rats, and that faecal excretion is the major route of elimination (Nakayama et al., 1986). 131I-Stevioside (specific activity, 3.7 MBq/mg, equivalent to 100 µCi/mg; 1.1 MBq, 30 µCi, equivalent to 1 mg/kg bw) was injected intravenously to male Wistar rats. The radiolabel in plasma decreased rapidly, showing rapid distribution in the body. The highest concentrations of radiolabel 10 and 120 min after injection were observed in the liver (45 and 5% of the injected dose, respectively) and the small intestine (18 and 66%). At 120 min after injection, the radiolabel eliminated in the bile represented 52% of the original dose; that excreted in the faeces and urine 24 h after injection represented 35 and 35%, respectively, of the original dose (Cardoso et al., 1996). The Committee considered that this study was of limited value since introduction of a large 131I atom into stevioside might significantly affect its absorption, distribution, metabolism, and excretion in bile or urine. The renal excretion of stevioside and its effect on the renal excretion of several other substances was studied in groups of 10 male Wistar rats, which received intravenous infusions of stevioside at doses of 4, 8, 12, or 16 mg/kg bw per hour for 30 min, after a control period of 30 min. No significant change in inulin clearance was observed, but there was a significant increase in para-aminohippuric acid clearance, fractional sodium excretion (FeNa+), urinary flow as percent of glomerular filtration rate, and glucose clearance when compared with controls at doses greater than 4 mg/kg bw per hour. Stevioside clearance was greater than inulin clearance and smaller than para-amino-hippuric acid clearance at all doses tested. The authors concluded that stevioside is secreted by the renal tubular epithelium and induces diuresis and natriuresis and a fall in renal tubular reabsorption of glucose (Melis, 1992a). 2.1.2 Biotransformation Thin-layer chromatography of the intestinal contents, faeces, and bile of groups of three to seven Wistar rats given 3H-stevioside (specific activity, 13 or 46 µCi/mg) by gavage at a dose of 125 mg/kg bw (10-120 µCi/kg bw) revealed that stevioside is decomposed by rat caecal flora to steviol and sugars. Stevioside was detected as the major component in the stomach 1 h after administration. After 4 h in the small intestine, stevioside, steviolbioside (produced by cleavage of the glucose moiety at the C-19 position), and steviol accounted for 7.6, 8, and 7.5% of the radiolabel; in the caecum, these compounds accounted for 39, 17, and 5.1% of the radiolabel, respectively. At 24 h, stevioside was not detectable in the caecum, but steviol and an unidentified metabolite accounted for 16 and 68% of the radiolabel, respectively. Steviol was found to be the major metabolite in faeces, whereas stevioside and the unidentified metabolite were not quantifiable. In bile, most of the radiolabel found up to 24 h was on the unidentified metabolite, which was probably a steviol conjugate. The authors concluded that orally administered stevioside is not readily absorbed from the upper part of the small intestine, but metabolites, formed primarily by the bacterial flora in the caecum, are absorbed from the lower part of the intestine; they also concluded that most of the stevioside is excreted as steviol in the faeces (Nakayama et al., 1986). The Committee concluded further that the faecal steviol could also have arisen from deconjugation of biliary conjugates by the gut flora. When 2.5 mg/ml stevioside were incubated under anaerobic conditions with whole-cell suspensions of bacteria from rat caecum, stevioside was completely degraded to steviol within two days. The authors concluded that similar degradation of stevioside occurs in humans (Wingard et al., 1980). Mass spectral analysis of steviol and some analogues revealed characteristic patterns reflecting differential stereochemistry and variations in the nature of the substituents present. This information was used to identify several metabolites of steviol (by gas chromatography-mass spectrometry) which are known to produce a mutagenic response in Salmonella typhimurium strain TM677 with metabolic activation. After incubation with a 9000 × g fraction derived from the livers of Aroclor 1254-pretreated rats, unchanged steviol was the predominant compound, and nine metabolites were found. The major pathways of mammalian metabolism of steviol proved to be allylic oxidation and epoxidation. 15alpha-Hydroxysteviol represented 67% of the metabolites of steviol in vitro (Compadre et al., 1988). 131I-Stevioside (specific activity, 3.7 MBq/mg, equivalent to 100 µCi/mg; 1.1 MBq, 30 µCi, equivalent to 1 mg/kg bw) was injected intravenously to male Wistar rats. The results of reverse-phase high-performance liquid chromato-graphy (RP-HPLC) of the bile showed that stevioside was degraded in vivo and that steviol was the major metabolite (47% of radiolabel); 37% of the radiolabel was on stevioside, and the remaining 15% was on an unidentified metabolite. RP-HPLC analysis of urine 90 min after injection showed the presence of stevioside and the same unidentified metabolite found in bile, but no steviol. The authors concluded that stevioside is metabolized in rat liver to steviol, which is excreted through the bile, and that similar degradation occurs in humans (Cardoso et al., 1996). The Committee concluded that there is an alternative explanation, namely that stevioside is secreted into the bile, is degraded to steviol by the gut flora, and is resorbed in the lower gut. The Committee considered that this study was of limited value since introduction of a large 131I atom into stevioside might significantly affect its absorption, distribution, metabolism, and excretion in bile or urine. Stevioside was perfused at a concentration of 0.2 or 0.5 mmol/L (equivalent to 0.16 and 0.4 mg/ml) into rat livers and was recirculated for 2 h. The concentration of stevioside remained constant throughout the perfusion. The formation of hydrolysis products, especially steviol, was investigated chromatographically, with negative results. The authors concluded that the reported metabolic transformation of intravenously injected 131I-stevioside is either a specific characteristic of this derivative or depends on factors that are absent in the isolated perfused rat liver (Ishii-Iwamoto & Bracht, 1995). The Committee concluded that the most likely explanation for the apparent discrepancy is the fact that introduction of the large 131I atom into stevioside altered its pharmacokinetic behaviour and that stevioside is secreted into the bile in vivo and is degraded to steviol by the gut flora. 2.1.3 Effects on enzymes and other biochemical parameters Stevioside given to female RCR/Ha mice did not induce glutathione S-transferase activity in liver or intestinal mucosa (Pezzuto et al., 1986). Stevioside (1 mmol/L, equivalent to 0.8 mg/ml) inhibited oxidative phosphorylation and the activity of ATPase (50% inhibition), succinate oxidase (8% inhibition), and succinate dehydrogenase (10% inhibition). No inhibition of NADH-oxidase or L-glutamate dehydrogenase activity was seen. The ADP:O ratio was slightly decreased. Substrate respiration was increased at low concentrations (up to 0.5 mmol/L, equivalent to 0.4 mg/ml) and inhibited at higher concentrations (1 mmol/L, equivalent to > 0.8 mg/ml). The authors concluded that stevioside acts as a weak uncoupler of oxidative phosphorylation (Kelmer-Bracht et al., 1985). Stevioside inhibited oxidative phosphorylation in isolated rat liver mitochondria. The concentration required for 50% inhibition of ATP synthesis was 1.2 mmol/L, equivalent to 0.97 mg/ml (Vignais et al., 1966). The effect of stevioside, an inhibitor of long-chain fatty acid transport, on ketogenesis and on 14C-carbon dioxide production from [1-14C]-palmitate (100-300 µmol/L) was investigated in isolated and haemoglobin-free perfused rat liver. Stevioside (2.5 mmol/L, equivalent to 2 mg/ml) inhibited both parameters but had a smaller effect on 14C-carbon dioxide production. At 300 µmol/L palmitate and 150 µmol/L albumin, ketogenesis was inhibited by 66%, whereas no significant inhibition of 14C-carbon dioxide was seen. The authors concluded that these results reflect different degrees of saturation of the citric acid cycle and the ketogenic pathway and that changes in the redox state of the mitochondrial NAD(+)-NADH complex occur after infusion of stevioside (Constantin et al., 1991). The Committee noted that the concentrations used in the studies conducted in vitro were very high relative to those achieved in blood after oral administration, when the major intestinal metabolite that enters the circulation is steviol. These studies may therefore be of limited significance. When single doses of 200 µmol/L stevioside, equivalent to 650 mg/kg bw, were given orally to 24-h-fasted male Wistar rats, either alone or simultaneously with fructose, stevioside increased the initial glycogen deposition in the liver. When it was given to the rats in the drinking-water at 1 or 2 mmol/L, equivalent to 81 and 160 mg/kg bw, at the beginning of a fasting period of 24 or 48 h, increased hepatic glycogen concentrations were found at 48 h (1 mmol/L) and at 24 h (2 mmol/L). The authors concluded that stevioside stimulates hepatic glycogen synthesis under gluconeogenic conditions (Hübler et al., 1994). The effect of stevioside on the transport and metabolism of D-glucose and D-fructose was investigated in isolated perfused rat liver. The maximal exchange rate of D-glucose was 700 µmol/L per min/ml, and the Km was 38 mmol/L. Stevioside inhibited D-glucose and D-fructose transport across the cell membrane. The half-maximal effect at 1 mmol/L D-glucose occurred at 0.8 mmol/L stevioside, equivalent to 0.65 mg/ml. Stevioside had no effect on D-glucose metabolism, except to cause transient changes in D-glucose release, which reflected changes in the intracellular concentration. D-Fructose consumption, however, was specifically affected (half-maximal effect at 2.8 mmol/L, equivalent to 2.3 mg/ml), as were all parameters that depend on D-fructose transformation: D-glucose production, L-lactate and pyruvate production, and extra oxygen uptake. In livers that released D-glucose from endogenous glycogen, strong inhibition of transport increased the intracellular:extracellular ratio of D-glucose concentration. The control values for this ratio, representing an average over the total intracellular water space, were all below unity (Ishii et al., 1987). Stevioside had no effect on gluconeogenesis or oxygen uptake in isolated Wistar rat renal cortical tubules at concentrations up to 3 mmol/L, equivalent to 2.4 mg/ml. The authors concluded that the lack of activity was due to the inability of stevioside to penetrate the cell membrane (Yamamoto et al., 1985). Intravenous infusion of stevioside at 150 mg/ml to male Wistar rats at a dose of 100, 150, or 200 mg/kg bw per hour raised the plasma glucose concentrations to 110, 140, and 130% of the control value during and after infusion. The glucose turnover rate was not altered, but glucose clearance was reduced by infusion of 200 mg/kg bw per hour stevioside, from 6.5 to 5 ml/min per kg bw. The plasma insulin concentration was unchanged. Pretreatment with angiotensin II and arginine vasopressin had no effect, while prazosin, an alpha-adrenergic blocker, attenuated the hyperglycaemic effect of stevioside and infused insulin inhibited it. Oral administration of stevioside at 2000 mg/kg bw had no effect on the plasma glucose concentration. The authors concluded that the hyperglycaemic effect of stevioside was due to an effect on glucose transport across the cell (Suanarunsawat & Chaiyabutr, 1997). The effects of stevioside at 1 and 5 mmol/L, equivalent to 0.8 and 4 mg/ml, on intestinal glucose absorption were examined in hamster jejunum by the everted sac technique. Glucose absorption was not inhibited (Toskulkao et al., 1995a,b). Infusion of stevioside at 15 µmol/L, equivalent to 12 µg/ml, for 20 min did not significantly alter the arginine-induced secretion of insulin or glucagon in the pancreas of male Wistar rats (Usami et al., 1980). Stevioside inhibited the action of atractyloside, a known inhibitor of the adenine nucleotide carrier of mitochondria and in consequence an inhibitor of energy metabolism, in isolated perfused rat liver. It decreased the effects of atractyloside on glycolysis, glycogenolysis, gluconeogenesis, and oxygen uptake. The concentration for half-maximal action of stevioside was 0.5 mmol/L, equivalent to 0.4 mg/ml. The authors concluded that it acts on the outside of the cell, as labelled stevioside did not penetrate the cell membranes (Ishii & Bracht, 1986). Concomitant treatment of Raji cells (human lymphoblastoid cells carrying the Epstein-Barr viral genome) with 12- O-tetradecanoylphorbol 13-acetate (TPA) and stevioside did not inhibit the induction of Epstein-Barr virus early antigen by TPA at the highest concentration tested: 50 µg/ml (18% inhibition) (Okamoto et al., 1983a). 2.2 Toxicological studies 2.2.1 Acute toxicity Studies of the the toxicity of stevioside given as single doses to rodents are summarized in Table 1. No lethality was seen within 14 days after administration, and no clinical signs of toxicity or morphological or histopathological changes were found. After intravenous administration of stevioside to pentobarbital-anaesthetized dogs at a dose of 32.5 µmol/L per kg bw (equivalent to 26 mg/kg bw), no significant changes were seen in any parameters of whole blood, plasma, or renal function, and there was no significant alteration of the renal ultrastructure. The authors concluded that stevioside is totally devoid of acute extrarenal effects (such as hypoxaemia, which could contribute to nephrotoxicity) and direct renal effects during the 6-h period following intravenous administration (Krejci & Koechel, 1992). Table 1. Acute toxicity of stevioside (purity, 96%) given orally to rodents Species Sex LD50 (g/kg bw) Reference Mouse Male and female > 15 Toskulkao et al. (1997) Mouse Male > 2 Medon et al. (1982) Rat Male and female > 15 Toskulkao et al. (1997) Hamster Male and female > 15 Toskulkao et al. (1997) 2.2.2 Short-term studies of toxicity Rats A 13-week study of toxicity was carried out in Fischer 344 rats given doses of 0, 0.31, 0.62, 1.25, 2.5, or 5% in the diet (equivalent to 160, 310, 630, 1300, and 2500 mg/kg bw per day) to determine the appropriate doses for a two-year study of carcinogenicity. The rats were randomly allocated to six groups, each consisting of 10 males and 10 females. None of the animals died during the administration period, and there was no difference in body-weight gain between the control and treated groups during administration or in food consumption in the later part of the study. The activity of lactic dehydrogenase and the incidence of single-cell necrosis in the liver were increased in all groups of treated males. The authors considered these effects to be nonspecific because of the lack of a clear dose-response relationship, the relatively low severity, and their limitation to males. Other statistically significant differences in haematological and biochemical parameters were also considered to be of minor toxicological significance. The authors concluded that a concentration of 5% in diet was a suitable maximum tolerable dose of stevioside for a two-year study in rats (Aze et al., 1991). 2.2.3 Long-term studies of toxicity and carcinogenicity Rats Groups of 45 male and 45 female inbred Wistar rats were given diets containing stevioside (purity, 85%) at 0, 0.2, 0.6, or 1.2% (equivalent to 100, 300, and 600 mg/kg bw per day) for two years. After 6, 12, and 24 months, blood was obtained from the tail vein of five male and five female rats in each dose group for haematological and clinical biochemical tests. One week later, these rats were housed in metabolism cages for urine collection and were then killed for further biochemical, pathological, and histopathological examination. All surviving animals were killed at two years. Growth, food use and consumption, general appearance, and mortality were similar in treated and control groups. The mean life span of rats given stevioside was not significantly different from that of the controls. No treatment-related changes were observed in haematological, urinary, or clinical biochemical values at any stage of the study. The incidence and severity of non-neoplastic and neoplastic changes were unrelated to the concentration of stevioside in the diet. The NOEL was 1.2%, equivalent to 600 mg/kg bw per day. The authors suggested that the acceptable daily intake of stevioside for humans was 7.9 mg/kg bw per day, on the basis of the stevioside consumption of the rats during the first three months (the average for males and females being 790 mg/kg bw per day) and a safety factor of 100 (Xili et al., 1992). Stevioside (purity, 95.6%) was added to powdered diet at concentrations of 0, 2.5, or 5% (equal to 0, 970, and 2000 mg/kg bw per day for males and 0, 1100, and 2400 for females) and pelleted every three months. The doses were selected on the basis of the results of the 13-week study and administered to groups of 50 male and 50 female Fischer 344/DuCrj rats ad libitum for 104 weeks. Thereafter, all of the groups were maintained on basal diet for four weeks. All surviving rats were killed at week 108. The body-weight gain of treated animals was slightly depressed, and a relationship was seen with the dose of stevioside: 2.3 and 4.4% in males at the low and high dose and 2.4 and 9.2% in females at the low and high dose. Food consumption did not differ between the groups. The final survival rate of males at 5% was significantly decreased, with a rate of 60% versus 78% in controls. The absolute kidney weights were decreased in male and female animals at the high dose; however, there was no significant histopathological evidence of neoplastic or non-neoplastic lesions attributable to treatment in any organ or tissue, except for a decreased incidence of mammary adenomas in females and a reduced severity of chronic nephropathy in males. The authors concluded that stevioside is not carcinogenic in Fischer 344 rats under these experimental conditions (Toyoda et al., 1995, 1997). The Committee noted that the report of Toyoda et al. (1995) gives data only for individual animals, with no summary tables or figures. The effects of stevioside on urinary bladder carcinogenesis initiated by N-nitrosobutyl- N-(4-hydroxybutyl)amine was evaluated in male Fischer 344 rats given 0.01% of the nitrosamine in their drinking-water for four weeks and then 5% stevioside in their diet, equivalent to 5000 mg/kg bw per day, for 32 weeks. All surviving rats were sacrificed after 36 weeks and examined histologically. Treatment with 5% stevioside did not affect the incidence or extent of papillary or nodular hyperplasia in nitrosamine-treated rats. No preneoplastic or neoplastic lesions of the urinary bladder were observed in rats treated with stevioside only. The authors concluded that stevioside does not promote bladder carcinogenesis (Hagiwara et al., 1984; Ito et al., 1984). 2.2.4 Genotoxicity Studies of the genotoxicity of stevioside are summarized in Table 2. Table 2. Results of assays for the genotoxicity of stevioside End-point Test object Concentration Results Reference Reverse mutation S. typhimurium TA98, TA100 50 mg/platea Negativeb Suttajit et al. (1993) (purity, 99%) Reverse mutation S. typhimurium TA97, TA98, TA100, 5 mg/platec Negative Matsui et al. (1996a)d TA102, TA104, TA1535, TA1537 1 mg/platee Negative (purity, 83%) Forward mutation S. typhimurium TM677 10 mg/platea Negative Matsui et al. (1996a) Forward mutation S. typhimurium TM677 Not specifieda Negative Medon et al. (1982) Forward mutation S. typhimurium TM677 10 mg/platea Negative Pezzuto et al. (1985a) umu Gene mutation S. typhimurium TA1535/pSK1002 5 mg/platea Negative Matsui et al. (1996a) Gene mutation B. subtilis H17 rec+, M45 rec- 10 mg/disca Negative Matsui et al. (1996a) Chromosomal aberration Chinese hamster lung fibroblasts 8 mg/mlc Negative Matsui et al. (1996a) 12 mg/mle Chromosomal aberration Human lymphocytes 10 mg/ml Negative Suttajit et al. (1993) Chromosomal aberration Chinese hamster lung fibroblasts 12 mg/mlc Negative Ishidate et al. (1984) (purity, 85%) a With and without metabolic activation b A positive response towards TA98 was seen without metabolic activation at 50 mg/ml but not at lower concentrations up to 20 mg/ml c Without metabolic activation d The same results were cited in an earlier abstract (Matsui et al., 1987). e With metabolic activation 2.2.5 Reproductive toxicity Hamsters Groups of 10 male and 10 female one-month-old golden hamsters (Mesocricetus auratus,) were force-fed with stevioside (purity, 90%) at 0, 500, 1000, or 2500 mg/kg bw per day daily. Each female was mated and allowed to bear three litters during the experiment. Females in late gestation and while lactating (one month) received stevioside in the drinking-water. Two weeks after the offspring had been weaned, the females were mated again. No abnormalities were found in the growth or fertility of animals of either sex. All of the males mated females efficiently and successfully; the females showed normal four-day oestrus cycles and became pregnant after mating. The duration of gestation, number of fetuses, and number of offspring were not significantly different from those of controls. Forty hamsters of each sex from the first and second litters were divided into four groups after weaning and force-fed stevioside at the same doses as their parents. These animals also showed normal growth and fertility. Histological examination of reproductive tissues from animals of all three generations revealed no abnormality that could be linked to treatment. The authors concluded that stevioside at doses up to 2500 mg/kg bw per day affected neither growth nor reproduction in hamsters (Yodyingyuad & Bunyawong, 1991). Rats Groups of 11 male Wistar rats were given stevioside (purity, 96%) in the diet at 0, 0.15, 0.75, or 3%, equivalent to 0, 150, 750, and 3000 mg/kg bw per day, for 60 days before and during mating, and groups of 11 female Wistar rats received the same diet for 14 days before mating and for seven days during gestation. Rats of each sex at the highest dose had slightly decreased body-weight gain. There was no treatment-related effect on mating performance or fertility, and no malformations were seen in the fetuses. The authors concluded that stevioside had no adverse effect on fertility or on the development of fetuses (Mori et al., 1981). The Committee noted a slight but not statistically significant increase in the number of dead or resorbed fetuses at the highest dose. A decoction of 5 g dry S. rebaudiana in 100 ml water was given orally to inbred, adult female rats for 18 days, resulting in an intake of approximately 40 ml/kg bw. They were mated with untreated rats during the last six days. Fertility was reduced to 21% of that of control rats and remained reduced (47%) after a 50-60-day recovery period (Mazzei-Planas & Kuc, 1968). The effects of aqueous S. rebaudiana extracts corresponding to 0.67 g dried leaves per ml, given at a dose of 2 ml/rat twice a day for 60 days, were studied in prepubertal (25-30 days old) rats. The end-points were glycaemia; serum concentrations of thyroxine and tri-iodothyronine; available binding sites in thyroid hormone-binding proteins; binding of 3H-methyltrienolone (a specific ligand of androgen receptors) to prostate cytosol; zinc content of the prostate, testis, submandibular salivary gland, and pancreas; water content of testis and prostate; body-weight gain; and the final weights of the testis, prostate, seminal vesicle, submandibular salivary gland, and adrenal. None of these parameters was significantly different from those in the control group, with the exception of the seminal vesicle weight, which fell by about 60%. The authors concluded that if the Stevia extract can decrease fertility in rats, the effect is almost certainly not exerted on males (Oliveira-Filho et al., 1989). 2.2.6 Developmental toxicity Rats Stevioside (purity, 95.6%) dissolved in distilled water was given to four groups of 25 or 26 pregnant Wistar rats by gavage once a day on days 6-15 of gestation at doses of 0, 250, 500, or 1000 mg/kg bw. The rats were sacrificed on day 20 of gestation, and their fetuses were examined for malformations. The end-points examined were maternal and fetal body weight, number of live fetuses, sex distribution, number of resorptions or dead fetuses, and incidence of malformations. No treatment-related effect on general or reproductive toxicity was observed up to the highest dose. The authors concluded that orally administered stevioside is not teratogenic in rats (Takanaka et al., 1991; Usami et al., 1995). 2.2.7 Studies on metabolites: Steviol 2.2.7.1 Absorption, distribution, and excretion: Steviol Intact or bile-duct ligated rats were given [17-14C]-steviol (specific activity, 2.9 µCi/mg, 1.7 µCi/rat, corresponding to approximately 3 mg/kg bw) either orally or by intracaecal injection. After oral administration, 1.5% of the radiolabel was excreted in the urine of intact rats and 96% in that of bile-duct-ligated animals; the corresponding amounts in faeces were 96 and 3.3%. After intracaecal administration of 14C-steviol to bile-duct ligated rats, 94 and 6% of the radiolabel was excreted in urine and faeces, respectively. When bile was collected over 72 h, all of the intracaecally injected radiolabel was recovered. Very little (0.02% of dose) was exhaled as 14C-carbon dioxide. The authors concluded that steviol is completely absorbed from the rat lower bowel (Wingard et al., 1980). 2.2.7.2 Effects on enzymes and other biochemical parameters: Steviol Steviol administered to female RCR/Ha mice did not induce glutathione S-transferase activity in liver or intestinal mucosa (Pezzuto et al., 1986). Steviol at 0.5 mmol/L, equivalent to 0.16 mg/ml, inhibited oxidative phosphorylation and the activity of ATPase (92%), NADH oxidase (45%), succinate oxidase (42%), succinate dehydrogenase (46%), and L-glutamate dehydrogenase (46%). The ADP:O ratio was decreased. Substrate respiration was increased at concentrations up to 0.5 mmol/L, equivalent to 0.16 mg/ml, and inhibited at > 1 mmol/L, equivalent to 0.32 mg/ml. Inhibition of substrate respiration was the only effect observed in uncoupled mitochondria. Net proton ejection induced by succinate and swelling induced by several substrates were inhibited. The authors concluded that steviol acts as a uncoupler of oxidative phosphorylation (Kelmer-Bracht et al., 1985). Steviol decreased glucose production and inhibited oxygen uptake in isolated Wistar rat renal cortical tubules (IC50, 0.3 mmol/L, equivalent to 96 µg/ml). The authors concluded that this effect is consistent with an inhibitory action on oxidative phosphorylation and electron transport in mitochondria (Yamamoto et al., 1985). Steviol inhibited oxidative phosphorylation in isolated rat liver mitochondria. The concentration required for 50% inhibition of ATP synthesis was 40 µmol/L, equivalent to 13 µg/ml. Steviol also inhibited the 2,4-dinitro-phenol-stimulated ATPase, phosphorylation of exogenous ADP, and exchange between exogenous 14C-ADP and endogenous adenine nucleotides. The authors concluded that steviol does not act at the level of the coupling mechanism but at the level of mitochondrial translocation of adenine nucleotides (Vignais et al., 1966). The effects of steviol (purity, 90%) on intestinal glucose absorption were examined in hamster jejunum by the everted sac technique. Thus, 1 mmol/L steviol (equivalent to 318.5 µg/ml) inhibited glucose absorption by 29-43%, and the inhibition was related to the steviol concentration and incubation time. Reductions in the intestinal mucosal ATP content and absorptive surface area were responsible for the inhibition of glucose absorption. The decrease in intestinal mucosal ATP content was accompanied by a decrease in the activities of mitochondrial NADH cytochrome c reductase and cytochrome oxidase. Steviol did not inhibit the activity of intestinal Na+/K+-ATPase or glucose uptake in the intestinal brush-border membrane vesicles. Steviol altered the morphology of the intestinal absorptive cells. The authors concluded that inhibition of intestinal glucose absorption by steviol in hamsters is due to a reduction in mucosal ATP content and alteration of the morphology of the intestinal absorptive cells (Toskulkao et al., 1995a,b). Single doses of 200 µmol/L steviol, equivalent to 255 mg/kg bw, were given orally to 24-h-fasted male Wistar rats, either alone or simultaneously with fructose. Under these conditions, steviol increased the initial glycogen deposition in the liver. When steviol was given to the rats in drinking-water at 1 or 2 mmol/L, equivalent to 32 and 64 mg/kg bw, at the beginning of a fasting period of 24 or 48 h, it had no effect on hepatic glycogen concentrations (Hübler et al., 1994). Concomitant treatment of Raji cells (human lymphoblastoid cells carrying the Epstein-Barr viral genome) with 12- O-tetradecanoylphorbol 13-acetate (TPA) and steviol strongly inhibited the induction of Epstein-Barr virus early antigen by TPA, with 50% inhibition at 25 µg/ml (Okamoto et al., 1983a). In a study of the effects of steviol at 0.2 µmol/L (equivalent to 64 ng/ml) on the induction of ornithine decarboxylase in mouse skin by TPA, the activity in the epidermis had increased by about 300-fold 4-5 h after application of 17 nmol/L TPA. TPA-induced ornithine decarboxylase activity was strongly decreased (63%) when steviol was applied to mouse skin 1 h before TPA treatment, concurrently with TPA (75%), or 1 h after TPA (71%). Steviol alone did not induce epidermal ornithine decarboxylase activity. The authors concluded that steviol interferes with the process of induction of this enzyme by TPA in mouse skin (Okamoto et al., 1983b). 2.2.7.3 Acute toxicity: Steviol In male and female mice and rats given steviol (purity, 90%) orally, the LD50 was > 15 g/kg bw, and 1/15 animals died within 14 days of administration. The LD50 values in hamsters given steviol orally were 5.2 g/kg bw in males and 6.1 g/kg bw in females. Histopathological examination of the kidneys revealed severe degeneration of the proximal tubular cells, and these structural alterations were correlated with increased serum blood urea nitrogen and creatinine. The authors concluded that the cause of death was acute renal failure (Toskulkao et al., 1997). 2.2.7.4 Genotoxicity: Steviol Studies of the genotoxicity of steviol are summarized in Table 3. The major metabolite of steviol in vitro, 15alpha-hydroxysteviol, was inactive at doses up to 7.5 mg/ml in the forward mutation assay in S. typhimurium strain TM677 with metabolic activation. 15-Oxosteviol, a product of the oxidation of 15alpha-hydroxysteviol, was a directly acting mutagen at 25-200 µg/ml and was highly toxic to bacteria. Moreover, the expression of mutagenicity required the presence of the 13-hydroxy group and the C-16 exomethylene group (Compadre et al., 1988). 15-Oxosteviol was not mutagenic in various test systems. Repetition of the experiment with S. typhimurium TM677 failed to show significant induction of 8-azaguanine-resistant mutants, even when the number of bacteria tested was greatly increased. The authors concluded that the earlier positive result reported was due to a common mishandling of data obtained in the TM677 system and that 15-oxosteviol is unlikely to be the active metabolite responsible for the mutagenicity of steviol (Procinska et al., 1991). Table 3. Results of assays for the genotoxicity of steviol End-point Test object Concentration Results Reference Reverse mutation S. typhimurium TA98 and TA100 20 mg/platea Negative Suttajit et al. (1993) Reverse mutation S. typhimurium TA97, TA98, TA100, 5 mg/platea Negative Matsui et al. (1996a)b TA102, TA104, TA1535, and TA1537 (purity, 99%) Forward mutation S. typhimurium TM677 10 mg/platec Negative Matsui et al. (1996a) 0.5-10 mg/plated Positive Forward mutation S. typhimurium TM677 10 mg/platec Negative Pezzuto et al. (1985a) 10 mg/platee Positive umu Gene mutation S. typhimurium TA1535/pSK1002 625-1250 µg/platec Positive Matsui et al. (1996a) 1259-2500 µg/plated Positive Gene mutation B. subtilis H17 rec+, M45 rec- 10 mg/disca Negative Matsui et al. (1996a) Gene mutation Chinese hamster lung fibroblasts 400 µg/mld Positivef Matsui et al. (1996a) Chromosomal aberration Chinese hamster lung fibroblasts 0.5 g/mlc Negative Matsui et al. (1996a) 1-1.5 mg/mld Positive Chromosomal aberration Human lymphocytes 0.2 mg/ml Negative Suttajit et al. (1993) Micronucleus formation MS/Ae mice 1000 mg/kg bwg Negative Matsui et al. (1996a) a With and without metabolic activation b The same results are cited in an earlier abstract (Matsui et al., 1987). c Without metabolic activation d With metabolic activation e With metabolic activation derived from phenobarbital- or Aroclor 1254-pretreated rats; fractions from control or 3-methylcholanthrene-pretreated rats were ineffective. f Diphtheria toxin-resistant colonies g Toxic: 4/6 mice at highest dose given intraperitoneally died The expression of the mutagenic activity of steviol in S. typhimurium TM677 was dependent on both metabolic activation (9000 × g fraction derived from phenobarbital- or Aroclor 1254-pretreated rats) and addition of NADPH. The authors concluded that a cytochrome P450 mediates the metabolic activation of steviol to a mutagenic species. As partially purified rat liver epoxide hydrolase did not inhibit steviol-induced mutagenicity, the authors concluded that the active metabolite is not an epoxide. A species structurally related to steviol, isosteviol, was not active in S. typhimurium TM677, regardless of whether metabolic activation was provided. Similarly, chemical reduction of the unsaturated bond linking the carbon-16 and -17 positions of steviol resulted in the generation of two isomeric products, dihydrosteviol A and B, which were not mutagenic. Ent-kaurenoic acid was also inactive. A potential metabolite of steviol, steviol-16alpha,17-epoxide, was synthesized chemically and found to be ineffective as a directly acting mutagen. The authors concluded that it is a metabolite of an integral component of stevioside that is mutagenic. The structural features necessary for the expression of mutagenic activity include a hydroxy group at position 13 and an unsaturated bond joining the carbon atoms at positions 16 and 17 (Pezzuto et al., 1985a, 1986). Steviol was mutagenic after metabolic activation in the forward mutation assay with S. typhimurium TM677. The authors confirmed first that the 8-aza-guanine resistance of the TM677 mutants resides in the chromosomal guanine phosphoribosyltransferase (gpt) gene, since it could be complemented by the gpt gene of Escherichia coli. The chromosomal DNA of TM677 and TM677 mutants was digested by several restriction enzymes (BamHI, Sau3AI, AluI, TaqI, HaeIII, HpaII, and RsaI) and analysed by Southern blot hybridization with a probe for the gpt gene in DNA of E. coli. No significant difference in DNA fragment length was found between the wild type and spontaneous or steviol-induced mutants (Matsui et al., 1988, 1989a). pSV2-gpt plasmids were treated with metabolically activated steviol (concentration not given), and the DNA was subsequently analysed by polyacrylamide gel electrophoresis after digestion with restriction endonucleases (Sau3AI, HhaI, HpaII). Steviol induced a fivefold increase in mutation frequency, and seven mutants were obtained, all showing deletions ranging from 20 bp to 2 kb (Matsui et al., 1989b). Steviol strongly induced mutations at the gpt gene of S. typhimurium TM677 when a metabolic activation system was present, but it had no activity in reverse mutation assays with E. coli WP2uvrA/pKM101 or S. typhimurium TA strains. In order to characterize the mutations induced by metabolically activated steviol, the chromosomal gpt alleles of 24 induced (ST clones) and 16 spontaneous mutants (SP clones) of S. typhimurium TM677 were sequenced, and the mutation spectra were compared. Nine out of 24 of the mutations of ST clones were localized in the region between nucleotides 280 and 330 from the starting codon ATG, whereas no mutations of SP clones were found in that region. The mutations identified included transitions (three clones), transversions (four clones), a duplication, and a deletion. There were no other marked differences between the ST and SP clones: base-change mutations predominated over frameshifts and deletions (ST clones, 20 versus three; SP clones, 16 versus two), and base-change mutations occurred more frequently at G:C pairs than at A:T pairs (ST clones, 15 versus five; SP clones, 12 versus four). The authors suggested that metabolically activated steviol interrupts DNA synthesis around nucleotide 280, thereby stimulating duplication, deletion, and untargeted mutagenesis in the defined region of the gpt gene downstream from the site of interruption (Matsui et al., 1996b). 19- O-ß-D-Glucopyranosyl steviol, a potential hydrolysis product of stevioside, was mutagenic to S. typhimurium TM677 and bactericidal in the presence of a metabolic activating system (Pezzuto et al., 1986). Microsomes derived from human liver mediated a mutagenic response of steviol (Pezzuto et al., 1985b). 2.2.7.5 Developmental toxicity: Steviol Groups of 20 pregnant golden hamsters were given steviol (purity, 90%) at doses of 0, 250, 500, 750, or 1000 mg/kg bw per day (only 12 animals at the highest dose) by gavage in corn oil on days 6-10 of gestation. A significant decrease in body-weight gain and increased mortality (1/20, 7/20, and 5/12) were observed at the three highest doses, and the number of live fetuses per litter and mean fetal weight decreased in parallel. Histopathological examination of the maternal kidneys showed a dose-dependent increase in the severity of effects on the convoluted tubules (dilatation, hyaline droplets). No dose-dependent teratogenic effects were seen. The NOEL was 250 mg/kg bw per day for both maternal and developmental toxicity (Wasuntarawat et al., 1998). 2.2.8 Special studies 2.2.8.1 Cariogenicity Groups of 15 albino Sprague-Dawley rat pups colonized with Streptococcus sobrinus received 0.5% stevioside or 30% sucrose in the basal diet or basal diet alone and were sacrificed after five weeks, when S. sobrinus was counted and caries were evaluated. There was no difference in food or water intake or in weight gain among the groups, but significant differences in sulcal caries scores and S. sobrinus counts were found between the group receiving sucrose and the other groups. There was no significant difference between the group receiving stevioside and the controls. The authors concluded that stevioside was not cariogenic under the conditions of this study (Das et al., 1992). 2.2.8.2 Renal function and vasodilatation The effect of stevioside on renal function was evaluated by clearance techniques in groups of seven pentobarbital-anesthetized male Wistar rats simultaneously with the effect of indomethacin on the renal action of stevioside, given at a priming dose of 4, 8, 12, or 16 mg/kg bw followed by an infusion rate of 4, 8, 12, or 16 mg/kg bw per hour. Mean arterial pressure and renal function were measured. Administration of stevioside resulted in a statistically significant, dose-related decrease in mean arterial pressure (120 ± 2.3 with 4 mg/kg bw to 72 ± 4.8 mm Hg with 16 mg/kg bw) and an increase in renal plasma flow (10 ± 1.2 with 4 mg/kg bw to 26 ± 2.9 ml/min per kg bw with 16 mg/kg bw), with no change in glomerular filtration rate. Stevioside also increased fractional sodium (FeNa+) and potassium (FeK+) excretion and urine flow (volume/glomerular filtration rate). The decrease in mean arterial pressure (control, 120 ± 0.93; stevioside, 91 ± 2.5 mm Hg) and increase in renal plasma flow (control, 14 ± 1.4; stevioside, 33 ± 2.8 ml/min per kg bw) induced by stevioside at 16 mg/kg bw were inhibited by simultaneous administration of indomethacin at 2 mg/kg bw, but the glomerular filtration rate was not affected. The diuretic, natriuretic, and kaliuretic effects of stevioside were also abolished by indomethacin. The authors concluded that stevioside behaves like a typical vasodilator, causing changes in mean arterial pressure, diuresis, natriuresis, and kaliuresis per millilitre of glomerular filtration, and that these effects probably depend on prostaglandins (Melis & Sainati, 1991a). The effects of intravenous administration of verapamil (0.015 mg/min) and calcium chloride (800 mEq/L, 0.025 ml/kg bw per min) on renal function and mean arterial pressure were evaluated in groups of 10 pentobarbital-anaesthetized male Wistar rats weighing 280-320 g during intravenous treatment with stevioside (16 mg/kg bw per min). Verapamil significantly increased the hypotensive effect of stevioside on mean arterial pressure (control, 120 ± 0.77; stevioside, 96 ± 1.5; stevioside plus verapamil, 67 ± 0.70 mm Hg) and on fractional sodium excretion (control, 0.76 ± 0.05; stevioside, 1.6 ± 0.10; stevioside plus verapamil, 2.7 ± 0.25%). Urinary flow, reported as percent glomerular filtration rate, and renal plasma flow were increased slightly but not significantly during administration of stevioside plus verapamil. In contrast, infusion of calcium chloride into rats pretreated with stevioside resulted in a marked attenuation of mean arterial pressure (control, 120 ± 1.8; stevioside, 70 ± 1.1; stevioside plus calcium chloride, 110 ± 1.6 mm Hg) and renal plasma flow (control, 17 ± 3.8; stevioside, 34 ± 2.6; stevioside plus calcium chloride, 17 ± 2.9 ml/min per kg bw). The diuresis and natriuresis induced by stevioside were also inhibited by simultaneous administration of calcium chloride. The authors concluded that stevioside acts on arterial pressure and renal function as a calcium antagonist, as does verapamil (Melis, 1992b). Classical clearance techniques and arterial pressure measurements in pentobarbital-anaesthetized male Wistar rats showed that stevioside at a priming dose of 8 or 16 mg/kg bw followed by an infusion rate of 8 or 16 mg/kg bw per h caused a fall in systemic blood pressure and in diuresis and natriuresis per millilitre of glomerular filtration rate. Verapamil tended to increase the renal and systemic effects of stevioside. In contrast, an infusion of calcium chloride into rats pretreated with stevioside induced marked attenuation of the vasodilatatory responses to stevioside. The authors concluded that stevioside, like verapamil, acts as a calcium antagonist (Melis & Sainati, 1991b). The effect of stevioside (purity, > 90%) at a priming dose of 16 mg/kg bw followed by an infusion rate of 16 mg/kg bw per h on renal function in normal Wistar rats and rats with experimental renal hypertension was evaluated by clearance techniques. Stevioside provoked hypotension, diuresis, and natriuresis in both groups of rats. The normal rats had increased renal plasma flow and a constant glomerular filtration rate after stevioside administration, whereas the hypertensive rats had increased renal plasma flow and glomerular filtration rate. The authors concluded that stevioside impairs a renal autoregulation mechanism in this model (Melis, 1992c). The effects of administration of aqueous S. rebaudiana extracts corresponding to 0.67 g/ml dried leaves given at 2 ml/rat twice a day for 20, 40, or 60 days on renal function and mean arterial pressure were studied in normal Wistar rats weighing 80-100 g. Rats treated for 20 days showed no significant difference from the controls, but administration of the crude extract for 40 or 60 days induced hypotension, diuresis, and natriuresis, with a constant glomerular filtration rate. Increased renal plasma flow was seen only in the group treated for 60 days. The authors concluded that oral administration of an aqueous extract of dried leaves of Stevia to rats induces systemic and renal vasodilation, causing hypotension, diuresis, and natriuresis (Melis, 1995). Normotensive and experimentally hypertensive male Wistar rats (Goldblatt GII experimental hypertension induced by clipping the left renal artery, leaving the contralateral kidney intact) received an S. rebaudiana extract corresponding to 0.67 g/ml dried leaves given at 2 ml/rat by gavage twice a day (2.7 g dry leaves per day) for 30 days. Administration of Stevia 10-12 weeks after clipping resulted in a significant decrease in mean arterial pressure in both the normotensive and hypertensive rats: normotensive, 110 ± 3.0 mm Hg in controls versus 70 ± 4.0 mm Hg in those given Stevia; hypertensive, 160 ± 3.0 mm Hg in controls versus 110 ± 4.0 mm Hg with Stevia. The glomerular filtration rate was constant in the normotensive rats but increased significantly in the hypertensive rats after Stevia treatment (16 ± 1.3 versus 14 ± 1.3 ml/min per kg bw in the controls and Stevia groups, respectively). Both normotensive and hypertensive rats had increased renal plasma flow after administration of Stevia: normotensive, 16 ± 3.1 ml/min per kg bw in controls versus 33 ± 3.2 ml/min per kg bw in the Stevia group; hypertensive, 19 ± 2.5 ml/min per kg bw in controls versus 37 ± 3.9 ml/min per kg bw in the Stevia group. Stevia increased urinary flow in both normotensive (1.4 ± 0.08% versus 2.3 ± 0.11%) and hypertensive animals (1.5 ± 0.07% versus 3.0 ± 0.13%) and also increased sodium excretion i(normotensive, 0.61 ± 0.07% in controls versus 1.6 ± 0.2% in the Stevia group; hypertensive, 0.70 ± 0.1% in controls versus 2.2 ± 0.45% in the Stevia group). The authors concluded that Stevia impairs renal autoregulation in this model (Melis, 1996). 2.3 Observations in humans S. rebaudiana has been used by Indians in Paraguay as an oral contraceptive (Mazzei-Planas & Kuc, 1968; Schvartzman et al., 1977). Aqueous extracts of 5 g of S. rebaudiana leaves were administered to 16 volunteers at 6-h intervals for three days, and glucose tolerance tests were performed before and after administration. Another six volunteers were given an aqueous solution of arabinose in order to eliminate possible effects of stress. The extract increased glucose tolerance and significantly decreased plasma glucose concentrations during the test and after overnight fasting in all volunteers (Curi et al., 1986). 3. COMMENTS After oral administration to rats, stevioside is not readily absorbed from the upper small intestine but is hydrolysed to the aglycone, steviol, before absorption from the gut. Steviol per se is completely absorbed and is excreted in the bile as conjugates; only a very small fraction is detectable in urine. After biliary excretion, the conjugates are hydrolysed, and steviol undergoes enterohepatic circulation; its elimination half-life is 24 h. Steviol is the only faecal metabolite of stevioside that has been identified, and excretion in the faeces is the major route. After intravenous administration, stevioside is rapidly distributed throughout the body, partially secreted by the renal tubular epithelium, and excreted in urine. At high concentrations, stevioside affected a variety of biochemical parameters in rat tissues in vitro. It weakly inhibited oxidative phosphorylation, and steviol was about 30 times more potent in this respect. The most likely mechanism is inhibition of the mitochondrial translocation of adenine nucleotides. Steviol also inhibited glucose absorption from rat gut by reducing the mucosal ATP content. Stevioside may also act as a calcium antagonist, showing a hypotensive effect and inducing diuresis, natriuresis, and a fall in renal tubular reabsorption of glucose. Stevioside may not, however, be able to penetrate cell membranes. Although most of these studies were performed after intravenous injection of stevioside, orally administered extracts of S. rebaudiana to rats had similar effects (hypotension and diuresis). Stevioside has very low acute oral toxicity. Oral administration of stevioside at a dietary concentration of 2.5% to rats for two years, equal to 970 and 1100 mg/kg bw per day in males and females, respectively, had no significant effect. Reduced body-weight gain and survival rate were observed at a dietary concentration of 5% stevioside. There was no indication of carcinogenic potential in a long-term study and no evidence of urinary bladder tumour promoting potential in a separate bioassay. In studies of reproductive toxicity, administration of stevioside at doses up to 2500 mg/kg bw per day to hamsters and 3000 mg/kg bw per day to rats had no effect. Although an aqueous infusion of S. rebaudiana administered orally to female rats was reported to cause a severe, long-lasting reduction in fertility, the contraceptive effect of Stevia is probably not due to stevioside. Stevioside had neither teratogenic nor embryotoxic effects in rats given up to 1000 mg/kg bw per day by gavage. The results of tests for genotoxicity with stevioside in various systems were uniformly negative. The aglycone, steviol, was more acutely toxic than stevioside to hamsters but not to rats. Steviol was clearly genotoxic after metabolic activation, inducing forward mutations in bacteria and gene mutations and chromosomal aberrations in Chinese hamster lung fibroblasts. Several mechanistic studies indicated that the structural features necessary for the expression of mutagenic activity include a hydroxyl group at position 13 and an unsaturated bond joining the carbon atoms at positions 16 and 17 of steviol. The fact that stevioside is glycosylated at position 13 could explain the absence of mutagenicity. The active metabolite of steviol responsible for its mutagenic activity is not known. While some data suggest that epoxidation may be involved in the metabolic activation of steviol, other data indicate that the active metabolite is not an epoxide. Preliminary data indicate that human liver microsomes may activate steviol to a mutagenic metabolite. 4. EVALUATION The Committee noted several shortcomings in the information available on stevioside. In some studies, the material tested (stevioside or steviol) was poorly specified or of variable quality, and no information was available on other constituents or contaminants. Furthermore, no studies of human metabolism of stevioside and steviol were available. In addition, data on long-term toxicity and carcinogenicity were available for stevioside in only one species. The mutagenic potential of steviol has been tested sufficiently only in vitro. In view of the fact that no information was provided for elaboration of specifications for stevioside and that the evaluation of the available toxicological data revealed several limitations, the Committee was unable to relate the results of the toxicological investigations to the article of commerce and could not allocate an ADI to stevioside. Before the substance is reviewed again, specifications must be developed to ensure that the material tested is representative of the material of commerce, and further information should be made available on the nature of the substance that was tested, on the human metabolism of stevioside, and on the activity of steviol in suitable studies of genotoxicity in vivo . 5. REFERENCES Aze, Y., Toyoda, K., Imaida, K., Hayashi, S., Imazawa, T., Hayashi, Y. & Takahashi, M. (1991) Subchronic oral toxicity study of stevioside in F344 rats. Bull. Natl Inst. Hyg. Sci., 48-54 (in Japanese). Cardoso, V.N., Barbosa, M.F., Muramoto, E., Mesquita, C.H. & Almeida, M.A. (1996) Pharmacokinetic studies of 131I-stevioside and its metabolites. Nucl. Med. Biol., 23, 97-100. Compadre, C.M., Hussain, R.A., Nanayakkara, N.P., Pezzuto, J.M. & Kinghorn, A.D. (1988) Mass spectral analysis of some derivatives and in vitro metabolites of steviol, the aglycone of the natural sweeteners, stevioside, rebaudioside A, and rubusoside. Biomed. Environ. Mass Spectrom., 15, 211-222. Constantin, J., Ishii-Iwamoto, E.L., Ferraresi-Filho, O., Kelmer-Bracht, A.M. & Bracht, A. (1991) Sensitivity of ketogenesis and citric acid cycle to stevioside inhibition of palmitate transport across the cell membrane. Braz. J. Med. Biol. Res., 24, 767-771. Curi, R., Alvarez, M., Bazotte, R.B., Botion, L.M., Godoy, J.L. & Bracht, A. (1986) Effect of Stevia rebaudiana on glucose tolerance in normal adult humans. Braz.J. Med. Biol. Res., 19, 771-774 (English abstract only). Das, S., Das, A.K., Murphy, R.A., Punwani, I.C., Nasution, M.P. & Kinghorn, A.D. (1992) Evaluation of the cariogenic potential of the intense natural sweeteners stevioside and rebaudioside A. Caries Res., 26, 363-366. Hagiwara, A., Fukushima, S., Kitaori, M., Shibata, M. & Ito, N. (1984) Effects of three sweeteners on rat urinary bladder carcinogenesis initiated by N-butyl-N-(4-hydroxybutyl)nitrosamine. Gann, 75, 763-768 (English abstract only). Hanson, J.R. & De Oliveira, B.H. (1993) Stevioside and related sweet diterpenoid glycosides. Nat. Prod. Rep., 10, 301-309. Hübler, M.O., Bracht, A. & Kelmer-Bracht, A.M. (1994) Influence of stevioside on hepatic glycogen levels in fasted rats. Res. Commun. Chem. Pathol. Pharmacol., 84, 111-118. Ishidate, M., Sofuni, T., Yoshikawa, K., Hayashi, M., Nohmi, T., Sawada, M. & Matsuoka, A. (1984) Primary mutagenicity screening of food additives currently used in Japan. Food Chem. Toxicol., 22, 623-636. Ishii, E.L. & Bracht, A. (1986) Stevioside, the sweet glycoside of Stevia rebaudiana, inhibits the action of atractyloside in the isolated perfused rat liver. Res. Commun. Chem. Pathol. Pharmacol., 53, 79-91. Ishii, E.L., Schwab, A.J. & Bracht, A. (1987) Inhibition of monosaccharide transport in the intact rat liver by stevioside. Biochem. Pharmacol., 36, 1417-1433. Ishii-Iwamoto, E.L. & Bracht, A. (1995) Stevioside is not metabolized in the isolated perfused rat liver. Res. Commun. Mol. Pathol.Pharmacol., 87, 167-175. Ito, N., Fukushima, S., Shirai, T., Hagiwara, A. & Imaida, K. (1984) Drugs, food additives and natural products as promoters in rat urinary bladder carcinogenesis. In: Börzsönyi, M., Day, N.E., Lapis, K. & Yamasaki, H., eds, Models, Mechanisms and Etiology of Tumour Promotion (IARC Scientific Publications No. 56), Lyon, International Agency for Research on Cancer, pp. 399-407. Kelmer-Bracht, A.M., Alvarez, M. & Bracht, A. (1985) Effects of Stevia rebaudiana natural products on rat liver mitochondria. Biochem. Pharmacol., 34, 873-882. Kim, K.K., Sawa, Y. & Shibata, H. (1996) Hydroxylation of ent-kaurenoic acid to steviol in Stevia rebaudiana Bertoni -- Purification and partial characterization of the enzyme. Arch. Biochem. Biophys., 332, 223-230. Krejci, M.E. & Koechel, D.A. (1992) Acute effects of carboxyatractyloside and stevioside, inhibitors of mitochondrial ADP/ATP translocation, on renal function and ultrastructure in pentobarbital-anesthetized dogs. Toxicology, 72, 299-313. Liu, J., Ong, C.P. & Li, S.F.Y. (1997) Subcritical fluid extraction of Stevia sweeteners from Stevia rebaudiana. J. Chromatogr.Sci., 35, 446-450. Matsui, M., Matsui, K., Kawasaki, Y., Sawada, M., Nohmi, T., Sofuni, T., Yoshihira, K., Ishidate, M., Oda, Y., Noguchi, T., Komatsubara, S. & Kitagawa, Y. (1987) Mutagenicity test on stevioside and its aglycone, steviol. Mutat. Res., 182, 366. Matsui, M., Matsui, K., Nohmi, T., Mizusawa, H. & Ishidate, M. (1988) Mutagenicity of steviol: An analytical approach using a Southern blotting system. Mutat. Res., 203, 377. Matsui, M., Matsui, K., Nohmi, T., Mizusawa, H. & Ishidate, M. (1989a) Mutagenicity of steviol: An analytical approach using the Southern blotting system. Bull. Natl Inst. Hyg.Sci., 83-87 (in Japanese). Matsui, M., Matsui, K., Nohmi, T., Mizusawa, H. & Ishidate, M. (1989b) Detection of deletion mutations in pSV2-gpt plasmids induced by metabolically activated steviol. Mutat. Res., 216, 367-368. Matsui, M., Matsui, K., Kawasaki, Y., Oda, Y., Noguchi, T., Kitagawa, Y., Sawada, M., Hayashi, M., Nohmi, T., Yoshihira, K., Ishidate, M. & Sofuni, T. (1996a) Evaluation of the genotoxicity of stevioside and steviol using six in vitro and one in vivo mutagenicity assays. Mutagenesis, 11, 573-579. Matsui, M., Sofuni, T. & Nohmi, T. (1996b) Regionally-targeted mutagenesis by metabolically-activated steviol: DNA sequence analysis of steviol-induced mutants of guanine phosphoribosyltransferase (gpt) gene of Salmonella typhimurium TM677. Mutagenesis, 11, 565-572. Mazzei-Planas, G. & Kuc, J. (1968) Contraceptive properties of Stevia rebaudiana. Science, 162, 1007. Medon, P.J., Pezzuto, J.M., Hovanec-Brown, J.M., Nanayakkara, N.P., Soejarto, D.D., Kamath, S.K. & Kinghorn, A.D. (1982) Safety assessment of some Stevia rebaudiana sweet principles. Fed. Proc., 41, 1568. Melis, M.S. (1992a) Renal excretion of stevioside in rats. J. Nat. Prod., 55, 688-690. Melis, M.S. (1992b) Influence of calcium on the blood pressure and renal effects of stevioside. Braz. J. Med.Biol. Res., 25, 943-949. Melis, M.S. (1992c) Stevioside effect on renal function of normal and hypertensive rats. J. Ethnopharmacol., 36, 213-217. Melis, M.S. (1995) Chronic administration of aqueous extract of Stevia rebaudiana in rats: Renal effects. J. Ethnopharmacol., 47, 129-134. Melis, M.S. (1996) A crude extract of Stevia rebaudiana increases the renal plasma flow of normal and hypertensive rats. Braz. J. Med.Biol. Res., 29, 669-675. Melis, M.S. & Sainati, A.R. (1991a) Participation of prostaglandins in the effect of stevioside on rat renal function and arterial pressure. Braz. J. Med. Biol. Res., 24, 1269-1276. Melis, M.S. & Sainati, A.R. (1991b) Effect of calcium and verapamil on renal function of rats during treatment with stevioside. J. Ethnopharmacol., 33, 257-262. Mori, N., Sakanoue, M., Takeuchi, M., Shimpo, K. & Tanabe, T. (1981) Effect of stevioside on fertility in rats. J. Food Hyg. Soc. Jpn., 22, 409-414 (in Japanese). Nakayama, K., Kasahara, D. & Yamamoto, F. (1986) Absorption, distribution, metabolism and excretion of stevioside in rats. J. Food Hyg. Soc. Jpn., 27, 1-8. Nanayakkara, N.P., Klocke, J.A., Compadre, C.M., Hussain, R.A., Pezzuto, J.M. & Kinghorn, A.D. (1987) Characterization and feeding deterrent effects on the aphid, Schizaphis graminum, of some derivatives of the sweet compounds, stevioside and rebaudioside A. J. Nat. Prod., 50, 434-441. Okamoto, H., Yoshida, D. & Mizusaki, S. (1983a) Inhibition of 12-O-tetradecanoylphorbol-13-acetate-induced induction in Epstein-Barr virus early antigen in Raji cells. Cancer Lett., 19, 47-53. Okamoto, H., Yoshida, D., Saito, Y. & Mizusaki, S. (1983b) Inhibition of 12-O-tetradecanoylphorbol-13-acetate-induced ornithine decarboxylase activity in mouse epidermis by sweetening agents and related compounds. Cancer Lett., 21, 29-35. Oliveira-Filho, R.M., Uehara, O.A., Minetti, C.A. & Valle, L.B. (1989) Chronic administration of aqueous extract of Stevia rebaudiana (Bert.) Bertoni in rats: Endocrine effects. Gen. Pharmacol., 20, 187-191. Pezzuto, J.M., Compadre, C.M., Swanson, S.M., Nanayakkara, D. & Kinghorn, A.D. (1985a) Metabolically activated steviol, the aglycone of stevioside, is mutagenic. Proc. Natl Acad Sci. USA, 82, 2478-2482. Pezzuto, J.M., Nanayakkara, N.P., Compadre, C.M., Swanson, S.M., Kinghorn, A.D., Guenthner, T.M., Lam, L.K., Sparnins, V.L. & Wattenberg, L.W. (1985b) Evaluation of the potential of steviol (13-hydroxy-ent-kaurenoic acid) and related diterpenes to mediate bacterial mutagenesis and induce glutathione-S-transferase in mice. Proc. Am. Assoc. Cancer Res., 26, 94. Pezzuto, J.M., Nanayakkara, N.P., Compadre, C.M., Swanson, S.M., Kinghorn, A.D., Guenthner, T.M., Sparnins, V.L. & Lam, L.K. (1986) Characterization of bacterial mutagenicity mediated by 13-hydroxy-ent-kaurenoic acid (steviol) and several structurally-related derivatives and evaluation of potential to induce glutathione S-transferase in mice. Mutat. Res., 169, 93-103. Procinska, E., Bridges, B.A. & Hanson, J.R. (1991) Interpretation of results with the 8-azaguanine resistance system in Salmonella typhimurium: No evidence for direct acting mutagenesis by 15-oxosteviol, a possible metabolite of steviol. Mutagenesis, 6, 165-167. Schvartzman, J.B., Krimer, D.B. & Moreno-Azorero, R. (1977) Cytological effects of some medicinal plants used in the control of fertility. Experientia, 33, 663-665. Soejarto, D.D., Kinghorn, A.D. & Farnsworth, N.R. (1982) Potential sweetening agents of plant origin. III. Organoleptic evaluation of Stevia leaf herbarium samples for sweetness. J. Nat. Prod., 45, 590-599. Suanarunsawat, T. & Chaiyabutr, N. (1997) The effect of stevioside on glucose metabolism in the rat. Can. J. Physiol. Pharmacol., 75, 976-982. Suttajit, M., Vinitketkaumnuen, U., Meevatee U. & Buddhasukh, D. (1993) Mutagenicity and human chromosomal effect of stevioside, a sweetener from Stevia rebaudiana Bertoni. Environ. Health Perspectives, 101, 53-56. Takanaka, T., Kawashima, K., Usami, M. & Sakami, K. (1991) A teratological study of stevioside administered orally to rats. Unpublished report from Department of Pharmacology, Biological Safety Research Center, National Institute of Hygienic Sciences, Japan. Submitted to WHO by Ministry of Health and Welfare, Food Chemistry Division, Japan. Toskulkao, C., Sutheerawatananon, M., Wanichanon, C., Saitongdee, P. & Suttajit, M. (1995a) Effects of stevioside and steviol on intestinal glucose absorption in hamsters. J. Nutr. Sci. Vitaminol., 41, 105-113. Toskulkao, C., Sutheerawattananon, M. & Piyachaturawat, P. (1995b) Inhibitory effect of steviol, a metabolite of stevioside, on glucose absorption in everted hamster intestine in vitro. Toxicol. Lett., 80, 153-159. Toskulkao, C., Chaturat, L., Temcharoen, P. & Glinsukon, T. (1997) Acute toxicity of stevioside, a natural sweetener, and its metabolite, steviol, in several animal species. Drug Chem.Toxicol., 20, 31-44. Toyoda, K., Kawanishi, T., Uneyama, C. & Takahashi, M. (1995) Re-evaluation of the safety of a food additive (reported in fiscal 1994). A chronic toxicity/carcinogenicity study of stevioside (a substance extracted from Stevia): Final report. Unpublished report from Division of Pathology, Biological Safety Research Center, National Institute of Health Sciences, Japan. Submitted to WHO by Ministry of Health and Welfare, Food Chemistry Division, Japan. Toyoda, K., Matsui, H., Shoda,T., Uneyama, C., Takada, K. & Takahashi, M. (1997) Assessment of the carcinogenicity of stevioside in F344 rats. Food Chem. Toxicol., 35, 597-603. Usami, M., Seino, Y., Takai, J., Nakahara, H., Seino, S., Ikeda, M. & Imura, H. (1980) Effect of cyclamate sodium, saccharin sodium and stevioside on arginine-induced insulin and glucagon secretion in the isolated perfused rat pancreas. Horm. Metab. Res., 12, 705-706. Usami, M., Sakemi, K., Kawashima, K., Tsuda, M. & Ohno, Y. (1995) Teratogenicity study of stevioside in rats. Bull. Natl Inst. Hyg. Sci., 113, 31-35 (in Japanese). Vignais, P.V., Duee, E.D., Vignais, P.M. & Huet, J. (1966) Effects of atractyligenin and its structural analogues on oxidative phosphorylation and on the translocation of adenine nucleotides in mitochondria. Biochim. Biophys. Acta, 118, 465-483. Wasuntarawat, C., Temcharoen, P., Toskulkao, C., Mungkornkarn, P., Suttajit, M. & Glinsukon, T. (1998) Developmental toxicity of steviol, a metabilite of stevioside, in the hamster. Drug Chem.Toxicol., 21, 207-222. Wingard, R.E., Brown, J.P., Enderlin, F.E., Dale, J.A., Hale, R.L. & Seitz, C.T. (1980) Intestinal degradation and absorption of the glycosidic sweeteners stevioside and rebaudioside A. Experientia, 36, 519-520. Xili, L., Chengjiany, B., Eryi, X., Reiming, S., Yuengming, W., Haodong, S. & Zhiyian, H. (1992) Chronic oral toxicity and carcinogenicity study of stevioside in rats. Food Chem. Toxicol., 30, 957-965. Yamamoto, N.S., Kelmer-Bracht, A.M., Ishii, E.L., Kemmelmeier, F.S., Alvarez, M. & Bracht, A. (1985) Effect of steviol and its structural analogues on glucose production and oxygen uptake in rat renal tubules. Experientia, 41, 55-57. Yodyingyuad, V. & Bunyawong, S. (1991) Effect of stevioside on growth and reproduction. Hum. Reprod., 6, 158-165.
See Also: Toxicological Abbreviations