Stevia: Pharmacology and toxicology of stevioside, rebaudioside A, and steviol


Of the three compounds to be discussed in this post, stevioside and rebaudioside A are major natural glycosides found in the leaves of S. rebaudiana (henceforth in this chapter expressed as Stevia), while the aglycone, steviol is a biosynthetic precursor in the leaves and a putative mammalian metabolite of stevioside. These compounds are structurally related to ent-kaurenoic acid.

Stevia leaves contain naturally high levels of the glycosides, and selective breeding has increased these levels further. Typical concentrations range from 5 to 10% w/w of the dried leaf for stevioside, 2–4% for rebaudioside A, 1–2% for rebaudioside C, and 0.4–0.7% for dulcoside A. Newer, commercially developed strains may contain an excess of 14% diterpene glycosides.

Stevioside, in the form of the pure compound or of Stevia leaf extracts, has been widely used as a food additive, particularly in Brazil, Korea and Japan. It has been estimated, e.g. that somewhere between 85 and 170 metric tons of stevioside were consumed in Japan in 1987. This is equivalent to approximately 1,700 tons of leaf. The absence of reports of adverse reactions from these countries is primafacie evidence of lack of gross toxicity. Safety concerns, therefore, must address issues of chronic or accumulative toxicity, subtle toxic actions occurring at low frequency, and the possible existence of particularly sensitive groups (e.g. pregnant women, babies, the elderly, the sick, those taking particular drugs, or those exhibiting particular genetic or pharmacogenetic traits).

Considering the extent to which Stevia and stevioside are used, studies relevant to toxicity and pharmacology are jejune. In considering the available literature, it is necessary to interpret the findings in the light of actual or anticipated human consumption. For stevioside, per capita consumption among users has been estimated to be 300mg/day (0.37 mM). Much of the literature is in the form of incomplete reports or abstracts, or has been published in hard to obtain journals.

A number of earlier reviews on Stevia and its glycosides have appeared.

Studies on extracts

Many of the earlier toxicological studies of Stevia glycosides involved either water or ethanolic extracts of Stevia leaves containing stevioside and rebaudioside A at various concentrations. These studies are typically discussed in terms of the percentage of stevioside in the extract, but the potential contribution of other components in the leaf must always be kept in mind. A number of differences in the pharmacology of Stevia extracts and stevioside have been noted. Thus, the extract inhibits glutamate dehydrogenase; stevioside does not. The hypoglycemic effects reported for Stevia extracts are not seen with stevioside. The potent inhibition of 2, 4-dinitrophenol-stimulated A TPase of rat liver mitochondria by crude leaf extract cannot be accounted for on the basis of stevioside content. The presence in extracts of glycosides having an unesterified free carboxyl group [rebaudioside B and steviolbioside ()] may contribute to the toxicity of extracts relative to purified stevioside.


Aqueous ethanolic extracts consisting 50% of stevioside had an acute LD50 (i.p.) in rats of 3400mg/kg. In mice aqueous extracts comprising 50% stevioside had an acute LD50 of 17 g/kg. Enrichment to 40–55% stevioside decreased lethality to >42 g/kg.

Subchronic and chronic studies

Stevia extracts containing 50% stevioside were fed to rats at doses up to 1 g/day for 56 days without detectable effects on biochemical or pathological endpoints, except for a slight depression in hepatic lactate dehydrogenase activity. A Stevia extract containing 75% stevioside and 16% rebaudioside A was fed to rats for up to 24 months at doses up to 550 mg/kg/day. The only effect noted was slight growth retardation.

Pharmacological actions

Energy metabolism

Crude Stevia extracts inhibited mitochondrial glutamate dehydrogenase from beef and rat.

Carbohydrate metabolism

There have been repeated claims in hard-to-obtain sources that Stevia extracts improve glucose tolerance in alloxan-diabetic rabbits, and both diabetic and non-diabetic humans. In one study, it was reported that powdered leaves as a 10% addition to a high carbohydrate diet decreased blood glucose and hepatic glycogen levels in rats after four weeks. One study reported in an abstract claims a 35% drop in blood glucose in human volunteers eight hours after consumption of an extract. On the other hand, 0.5–1.0g/day of Stevia extract for 56 days was without effect on blood sugar.

Cardiovascular effects

Stevia extracts have been reported in humans to reduce heart rate and mean arterial blood pressure. Aqueous extracts fed to rats for 40–60 days produced hypotension, the mean arterial pressure falling from 110 mm Hg to 90 mm Hg over the 40-day-treatment period.

Renal effects

In rats, extracts of Stevia given by mouth for 40–60 days induced diuresis and natriuresis and increased renal plasma flow, indicating vasodilatation. Glomerular filtration rate remained unchanged. The extract given was equivalent to 1.33 g dry leaves, twice a day. The stevioside content was judged to be too low to be responsible for the vasodilatation. Mean arterial pressure was decreased in both normal and Goldblatt hypertensive rats given extract equivalent to 2.67 g/dry leaves per day for 30 days.


An early report that an aqueous extract of Stevia reduced fertility of female rats has led to considerable controversy. The amount fed to the rats daily was eight times the dosage per kg purportedly used by Paraguayan Indians as an oral contraceptive. Although this report has been frequently cited, its findings have not been reproduced. No effects on fertility were seen in studies on rats or rabbits. The contraceptive effect of Stevia, if any, does not appear to be due to the sweeteners in it. In one small study, a hot water extract was added to rat food to the level of 0.14% stevioside for 21 days. No reproductive effects were seen in either sex. No endocrine effects were seen in male rats), including a study in which 25–30-day-old rats were fed leaf extract for the next 60 days.

Stevia: Stevioside

Rebaudioside A

Rebaudioside A (C44H70O23; molecular weight 966) has a similar structure to stevioside. It differs only in having an extra glucopyranosyl residue attached to the sugar unit at C-13. Despite its relative abundance in S. rebaudiana, rebaudioside A has been little studied.

Absorption, distribution and metabolism

Rat intestinal microflora efficiently hydrolyze rebaudioside A to steviol. Apart from this, reports in the mainstream literature on absorption, distribution and metabolism of rebaudioside A appear non-existent.

Lethality and whole animal toxicity

‘Crude’ rebaudioside A was given to mice by gastric intubation (2 g/kg). No toxic effects or effects on organ weight were noted two weeks later.

Pharmacological actions

Chromosomal and mutagenic effects

No mutagenic effects were seen in a Salmonella typhimurium strain, either in the presence or absence of a metabolic activating system. Rebaudioside A was without effect on azoxymethane-induced colonic aberrant crypt foci in rats.


Almost nothing is known per se of the pharmacology and toxicology of rebaudioside A. It differs in structure from stevioside only in the presence of an additional glucosyl unit. If the cardiac glycosides are any guide, this type of structural variation may affect solubility and binding characteristics (and hence pharmacokinetic properties) without qualitative change in biochemical actions.

Stevia: Steviol

Overall Conclusions

Much of the literature discussed above reports studies carried out on rats. Some of the findings, such as the conversion of stevioside to steviol, may reflect the idiosyncracies of that species. Studies on a variety of species would give better insight into potential risk for human consumers. For example, such studies would include examination of human gut flora under a variety of conditions, and the metabolic abilities of biopsy or autopsied samples of liver and kidney.

Stevia extracts, although consisting largely of glycosides, are variable, complex mixtures, partially due to differences in extraction technique and partially due to variability in plant strain, cultivation conditions, and other factors. Extracts can have biological effects that are not solely attributable to their content of stevioside or rebaudioside A [e.g. the inhibition of 2,4-dinitrophenol-stimulated ATPase in rat liver mitochondria ()]. The use of purified stevioside as a food additive thus appears preferable from a public safety viewpoint.

Stevia glycosides interfere with energy metabolism at concentrations within the high µM range. It is unclear how many of these effects are produced in vivo. However, these effects may be of particular concern for children, a group with high metabolic rate, rapid physiological growth, and a seemingly insatiable demand for sweet foods.

The potential risk: benefit ratio is also a function of the denominator. The potential benefits afforded by Stevia glycosides to certain groups may be considered to outweigh the potential risks. These benefits include better dietary control in diabetes and obesity. Claims have also been made that Stevia-containing products reduce risk of dental caries and periodontal disease.

There are perhaps two ways of investigating toxic potential. One is to administer the substance to experimental animals and then humans and look for toxic endpoints. The other is to examine the biological handling and mechanisms of action of the substance and look for potential problems. By the first approach, no problems have surfaced with stevioside and rebaudioside A (albeit that the literature on the latter compound is jejune). By the second approach, a number of concerns have surfaced, particularly in the areas of energy and carbohydrate metabolism, and in the metabolic potential of the glycosides themselves. Drawbacks with the first approach include the potential for missing susceptible groups or factors that predispose towards toxicity (e.g. nutrition, disease, age, ethnicity, drugs, etc.). Furthermore, even extensive, well-designed studies may miss low frequency but serious events, subtle effects or significant endpoints that are not examined for. Drawbacks with the second approach include the potential for fishing up red herrings: potential problems may be indicated that are never realized in exposed populations.

The aglycone, steviol, has the ability to penetrate cells, and has a number of potent effects on basic cell functions. The most important of these are inhibition of monosaccharide transport, and inhibition of oxidative phosphorylation, combined with disruption of a number of mitochondrial activities. In addition, steviol is bioactivated to a mutagenic metabolite, the nature of which is still unknown. The glycosides, stevioside and rebaudioside A, are assumed not to cross the plasma membrane into cells, and — except in rat intestine — not to be deglucosylated. Neither of these assumptions has been well tested experimentally. Although Stevia products have become widely used in certain countries within a short period without indications of toxicity surfacing, until these aspects of distribution and metabolism have been appropriately studied in humans, an assumption of universal safety appears inappropriate. The conclusion of Pezzuto, although more than a decade old, is still relevant: ‘It cannot be definitively stated on the basis of any results that are currently available that stevioside is harmful to human health. However, it does appear that additional safety assessments are required to resolve this question, as steviol, a likely hydrolysis product, is metabolized to active mutagenic species by human enzymes, and appears to covalently interact with DNA.’ Phillip’s comment ‘that stevia has been in use for centuries does not necessarily make it safe’ is also relevant given that such use is often uncritically cited as evidence of safety. It is evidence of one sort, but there are examples of substances in wide use for centuries or even millennia — such as tobacco or comfrey (Symphytum spp.) — before their profound toxicity became apparent.


Selections from the book: “Stevia. The genus Stevia”. Edited by A.Douglas Kinghorn. Series: “Medicinal and Aromatic Plants — Industrial Profiles”. 2002