Following a request from the European Commission, the Panel on Food Additives and Nutrient Sources added to Food (ANS) of the European Food Safety Authority (EFSA) was asked to deliver a scientific opinion on the re-evaluation of aspartame (E 951) as a food additive.
Aspartame (E 951) is a sweetener authorised as a food additive in the EU that was previously evaluated by the Joint FAO/WHO Expert Committee on Food Additives (JECFA), the EU Scientific Committee for Food (SCF) and the European Food Safety Authority (EFSA). Both JECFA and SCF established an Acceptable Daily Intake (ADI) of 40 mg/kg body weight (bw)/day.
The Panel based its evaluation on original study reports and information submitted following public calls for data, previous evaluations, and additional literature that has become available until the end of the public consultation on the draft Scientific Opinion on the re-evaluation of aspartame (E 951) as a food additive (15th February 2013). The Panel also evaluated literature published after the end of the public consultation, until 15th November 2013 (EFSA ANS Panel, 2013). The Panel noted that although many of the studies were old and were not performed according to current standards (e.g. Good Laboratory Practice (GLP) and Organisation for Economic Co-operation and Development (OECD) guidelines), they should be considered in the re-evaluation of the sweetener as long as the design of such studies and the reporting of the data were considered appropriate. In its re-evaluation of aspartame, the Panel also considered the safety of its gut hydrolysis metabolites methanol, phenylalanine and aspartic acid and of its degradation products 5-benzyl-3,6-dioxo-2-piperazine acetic acid (DKP) and β-aspartame, which may be present in the sweetener as impurities.
Aspartame (E 951) is a dipeptide of L-phenylalanine methyl ester and L-aspartic acid bearing an amino group at the α-position from the carbon of the peptide bond (α-aspartame). The major hydrolysis and degradation products of aspartame are L-phenylalanine, aspartic acid, methanol and DKP. DKP is formed through the intramolecular reaction of the primary amine with the methyl ester group of aspartame. β-Aspartame is a non-sweet isomer of α-aspartame.
Specifications have been defined in the European Commission Regulation (EU) No 231/2012 and by JECFA.
Studies in experimental animals and humans have shown that after oral ingestion, aspartame is fully hydrolysed within the gastro-intestinal tract. The products resulting from these reactions are methanol and the amino acids aspartic acid and phenylalanine. Hydrolysis of aspartame releases a corresponding 10 % by weight of methanol. Due to the very efficient hydrolysis in the gastro-intestinal tract the amount of intact aspartame that enters the bloodstream has been reported as undetectable in several studies conducted in various species, including rats, dogs, monkeys and humans. Further studies conducted in monkeys and pigs have also shown that the potential intermediate metabolite, phenylalanine methyl ester, is rapidly broken down to phenylalanine and methanol in the gastro-intestinal tract. Therefore, the Panel considered that phenylalanine, aspartic acid and methanol are absorbed and enter normal endogenous metabolic pathways.
The acute toxicity of aspartame was tested in mice, rats, rabbits and dogs and was found to be very low. Similarly, sub-acute and sub-chronic studies did not indicate any significant toxic effects in rats, mice or dogs.
Aspartame has been tested for genotoxicity in a number of in vitro and in vivo studies. The Panel concluded that in mammalian systems, apart from a valid UDS study, which was negative, no conclusion could be drawn at the gene and chromosomal level, as no studies dealing with these endpoints were available. However, the Panel considered that the weight-of-evidence was sufficient to conclude that aspartame was not mutagenic in bacterial systems. In vivo, the majority of investigations on genotoxicity reported negative findings. Equivocal findings were only described in one NTP (US National Toxicology Program) study, positive in female but not in male p53 haploinsufficient mice. In two other transgenic mouse strains the genotoxicity results were negative. The available in vitro data did not indicatea direct genotoxic activity of aspartame that might predispose to a site of first contact effect in vivo. Overall, the Panel concluded that the available data do not indicatea genotoxic concern for aspartame.
The results from three chronic toxicity and carcinogenicity studies in rats and one in mice revealed no aspartame-related increase in any type of neoplasms at all doses tested. The incidences of intracranial neoplasms observed in some studies were within the range of spontaneous brain tumours observed in the strain of rats used. In the rat studies, the highest doses tested (4000 or 8000 mg aspartame/kg bw/day) produced minor renal changes, considered by the Panel to be of minimal toxicological significance. A dose-dependent depression of body weight gain at 2000 and 4000 mg/kg bw/day correlating with decreased feed consumption was reported in one study. Overall, the Panel derived a no observable adverse effect level (NOAEL) of 4000 mg/kg bw/day from the four studies.
Furthermore, the NTP carried out several 9-month carcinogenicity studies with aspartame in genetically modified Tg.AC hemizygous, p53 haploinsufficient and Cdkn2a deficient mice. The Panel agreed that there was no evidence of treatment-related neoplastic or non-neoplastic lesions in any of these studies.
Since the last evaluation of aspartame by the SCF in 2002, two new long term carcinogenicity studies on aspartame in rats and one in mice were published by the European Ramazzini Foundation. The two rat studies have already been evaluated by the former Scientific Panel on Food Additives, Flavourings, Processing Aids and Materials in Contact with Food (AFC) and the ANS Panel and were considered to have methodological flaws. In addition to a high background incidence of chronic inflammatory changes in the lungs and other vital organs and tissues there is uncertainty about the diagnoses of some tumour types, which rendered the validity of the findings questionable. Moreover, EPA has recently concluded that many of the malignant neoplasms and the lymphoid dysplasias diagnosed in the studies from the European Ramazzini Foundation were hyperplasias related to unknown chronic infection in the animals and not related to aspartame intake. Furthermore, in the mouse study, the ANS Panel noted that the hepatic and pulmonary tumour incidences reported fell within the institute’s own historical control ranges for spontaneous tumours.
The available reproductive and developmental toxicity studies on aspartame comprised nine studies: an embryotoxicity and teratogenicity study performed in the mouse, a two-generation reproduction toxicity study in the rat, five peri- and postnatal developmental studies in the rat, a reproductive performance and developmental study in the rat and an embryotoxicity and teratogenicity study in the rat. In addition, eight embryotoxicity and teratogenicity studies were performed in the rabbit, four with administration of aspartame by diet and four by gavage.
From the rodent studies, the Panel identified a NOAEL of 5700 mg/kg bw/day, the highest dose level tested, in a developmental toxicity study in the mouse. The results of the reproductive and developmental toxicity studies in rats indicated NOAELs that ranged from 2000 to 4000 mg aspartame/kg bw/day. The Panel noted that developmental changes in pup body weight were observed at birth in studies at the dose of 4000 mg aspartame/kg bw/day and considered, these could be attributed to a combination of malnutrition and nutritional imbalance due to excessive exposure to phenylalanine derived from aspartame. In support of this hypothesis, the Panel noted that administration of a dose of L-phenylalanine equimolar to aspartame led to a similar decrease in maternal and pup body weight, as observed in a concurrent aspartame group.
Several reproductive and developmental toxicity studies performed in rabbits, where aspartame was administered via the diet or by gavage, were available to the Panel. Overall, the Panel considered that the data from these studies were confounded both by the decrease in feed intake (when aspartame was administered via the diet or by gavage), or the poor health of the animals, and, in many cases, by the number of deaths of pregnant rabbits in the treated groups possibly related to misdosing during gavage treatment. In one particular study with aspartame, pregnant rabbits were also dosed by gavage with L-phenylalanine and L-aspartic acid at dose levels equimolar to the top dose of 2000 mg aspartame/kg bw tested. A decrease in feed consumption was observed in the high aspartame dose and the L-phenylalanine-treated animals in this study and a significant body weight loss in the high aspartame dose was reported. Maternal toxicity and growth reduction were observed in the high dose aspartame group and to a lesser extent in the L-phenylalanine group compared to the controls. Mean fetal body weight and length were significantly reduced in both the high aspartame group and the L-phenylalanine group animals and a significantly higher rate of total (major and minor) malformations in the 2000 mg aspartame/kg bw/day group animals as compared to the concurrent control group was reported. The Panel considered the possibility that, in addition to a reduced feed intake by the mothers and gastrointestinal disturbances, exposure to high levels of aspartame-derived phenylalanine may be in part responsible for these effects in the high dose aspartame group because similar effects, though less severe, were seen in the phenylalanine group. Based on the above considerations the Panel identified a NOAEL of 1000 mg aspartame/kg bw/day for maternal (weight loss) and developmental toxicity (weight loss and malformations).
The Panel noted there was no epidemiological evidence for possible associations of aspartame with various cancers in the human population.
A large prospective cohort study in Denmark found no consistent association between the consumption of artificially sweetened beverages (but not with aspartame specifically) during pregnancy and the diagnosis of asthma or allergic rhinitis in children.
Another analysis of the same cohort showed a small but significantly elevated risk of medically induced pre-term delivery in women with higher reported consumption of artificially sweetened drinks.owever, another prospective study in Norway found that the association of pre-term delivery with artificially sweetened soft drinks was much weaker and barely discernible, and applied more to spontaneous than medically induced deliveries and was exceeded by an association with consumption of sugar-sweetened soft drinks.
Methanol is a metabolite of aspartame and is subject to significant first pass metabolism. The main route of metabolism of methanol proceeds by stepwise oxidation via formaldehyde to formate and then to carbon dioxide. Formate can also enter the one-carbon metabolic pool. Since some authors have suggested that methanol is responsible for the potential carcinogenicity and toxicity of aspartame, the Panel evaluated the available toxicological information on methanol.
The Panel considered the database on the genotoxicity of methanol and concluded that the data set was limited but that the available reliable in vitro and in vivo data did not indicate a genotoxic potential for methanol.
The oral studies on chronic toxicity and carcinogenicity of methanol are limited to a mouse study and a rat study. Overall, the Panel concluded that the mouse study was inadequate for the assessment of the carcinogenic potential of methanol and that the rat study was not suitable for the cancer risk assessment of methanol.
The reproductive and developmental toxicity database of methanol is limited. As oral studies available on methanol performed at high dose levels (4000 or 5000 mg/kg bw/day) did not allow the Panel to identify a NOAEL for reproductive and developmental toxicity for methanol by oral exposure, the Panel calculated the oral NOAEL using available data from animals exposed to methanol via inhalation. The Panel identified NOAECs of 1300 mg methanol/m3 in mice and 6500 mg methanol/m3 in rats that were exposed to methanol via the inhalation route. Based on these NOAECs, the Panel calculated oral NOAELs for mice and rats of approximately 560 and 2070 mg/kg bw/day, respectively.
The Panel considered the NOAEL of 560 mg/kg bw/day as the most conservative taking into account that the developmental effects observed in the mouse study tend to disappear as the pups grow. The Panel noted that the calculated NOAELs for methanol by oral exposure are 140 and 515-fold higher than the maximum amount of methanol that could be released when aspartame is consumed at the ADI.
The Panel concluded that the data on reproductive and developmental toxicity did not suggest that there was a risk from methanol derived from aspartame at the current exposure estimates or at the ADI of 40 mg/kg bw/day.
In addition, the Panel concluded that, based on recent measurements of basal levels of formaldehyde in blood and on the modelling of its biological turnover and steady state concentration in cells, formaldehyde formed from aspartame-derived methanol was not of safety concern at the current exposure estimates or at the ADI of 40 mg/kg bw/day.
Another aspartame metabolite, aspartic acid is itself a neurotransmitter and can be converted to the more potent excitatory neurotransmitter glutamate. The Panel noted that there was no evidence in vivo for neurotoxicity associated with aspartame exposure. The Panel concluded, however, that aspartic acid generated from aspartame was not of safety concern at the current exposure estimates or at the ADI of 40 mg/kg bw/day.
Concerning the third aspartame metabolite, phenylalanine, the Panel concluded that it is the main metabolite of concern in terms of potential developmental effects in humans. The Panel considered that it was plausible that phenylalanine could be responsible for some or all of the adverse effects reported for aspartame in rat and rabbit developmental toxicity studies.
Humans heterozygous for phenylalanine hydroxylase mutations, show a slightly reduced capacity to metabolise phenylalanine compared to normal individuals. Individuals homozygous for phenylalanine hydroxylase mutations (phenylketonuria (PKU) patients) have a markedly reduced capacity for phenylalanine metabolism. After birth, homozygous PKU babies show severe impairment in development and cognition if the phenylalanine intake via the diet is not strictly controlled.
PKU mothers with poorly controlled phenylalanine intake in their diet during pregnancy may give birth to babies with congenital heart diseases, microcephalus and impaired neurological function.
The Panel considered that adverse effects reported for aspartame in animal studies could be attributed to the metabolite phenylalanine, which was particularly the case for the rat and rabbit developmental toxicity studies. The Panel noted that adverse developmental effects were seen in children born to PKU patients and that these effects appeared to be related to maternal phenylalanine levels. The Panel was aware that the knowledge on effects of phenylalanine in PKU mothers and their children both before and after birth had developed considerably since the initial evaluation of aspartame.
The Panel undertook a formal Mode of Action (MoA) analysis of the putative role of phenylalanine in the developmental toxicity seen in animal studies. This MoA analysis is described in Section 11.
The Panel considered that the proposed MoA is plausible and relevant based on the weight-of-evidence of the available data, summarised in the opinion. There are uncertainties from the limited kinetic data in animals and in the human aspartame dose-phenylalanine concentration response data. The Panel decided to base the risk characterisation on comparison of plasma phenylalanine levels following aspartame administration with plasma phenylalanine levels associated with developmental effects in children born from mothers with PKU. The Panel decided these human data were more appropriate than the results of animal studies of reproductive and developmental toxicity for the risk characterization of aspartame.
Having established that the MoA was plausible and relevant, the Panel reviewed the information on plasma levels of phenylalanine associated with adverse effects on the fetuses of mothers with PKU. The Panel noted that current clinical guidelines recommend that plasma levels of phenylalanine should be maintained below 360 μM. In calculating a safe level of aspartame exposure (based on plasma phenylalanine concentrations), the Panel assumed the worst-case scenario that intake of aspartame occurs in combination with a meal which leads to circulating plasma phenylalanine concentrations of 120 µM (the maximum plasma concentration based on conservative assumptions of dietary exposure to phenylalanine). The concentration of plasma phenylalanine derived from aspartame was therefore set to 240 µM (i.e. 360 µM minus 120 µM) by the Panel. Based on the modelling, a plasma phenylalanine concentration of 240 µM would result from the administration of a bolus dose of 103 mg aspartame/kg bw (lower bound distributions: 88 mg aspartame/kg bw; 95th percentile, CI 59-125) to a normal subject. For a PKU heterozygous individual the concentration of 240 µM would be reached by the administration of a bolus dose of 59 mg aspartame/kg bw (lower bound distributions: 50 mg aspartame/kg bw; 95th percentile, CI 28-69). The Panel considered that given the conservative assumptions and the confidence intervals provided by the modelling, for realistic (i.e. non-bolus) dietary intake of aspartame, the peak plasma phenylalanine levels would not exceed 240 µM.
The Panel noted that in the normal population the 95th percentile confidence interval of the lower bound estimate of the dose resulting in a peak plasma level of 240 µM following a bolus administration of aspartame was greater than 40 mg/kg bw (which is equivalent to the current ADI). In the PKU heterozygous population the 95th percentile confidence interval for the lower bound of the dose resulting in a peak plasma level of 240 µM following a bolus administration of aspartame was greater than 40 mg/kg bw (which is equivalent to the current ADI) in 82 % of the simulations. The Panel considered that following bolus administration of aspartame of 40 mg/kg bw (which is equivalent to the current ADI) the PKU heterozygous population would not exceed the current clinical guideline of 360 µM.
The Panel also noted that in order to exceed the phenylalanine plasma concentration of 240 μM following repeated administration of aspartame in normal individuals, a bolus administration at 40 mg/kg bw (which is equivalent to the current ADI) would need to be given every hour.
The Panel considered the following:
- the conservative assumptions used in the modelling, which all overestimate peak plasma concentrations
- the available information on adverse effects on development in humans with PKU
- the allocation of 2/3 of the current clinical guideline level of 360 µM phenylalanine in plasma to phenylalanine from ingested aspartame, in order to account for simultaneous ingestion of phenylalanine from other components of the diet results of the modelling
- kinetic data from repeated oral administration of aspartame in humans
- bolus intakes based on consumption of one litre of soft drink containing aspartame at the maximum permitted level (MPL) of 600 mg/L by a child of 20-30 kg will not exceed 30 mg aspartame/kg bw.
Based on these considerations and evaluations, the Panel concluded that under realistic conditions of aspartame intake, phenylalanine plasma levels would not exceed 240 µM in normal or PKU heterozygous individuals. The Panel noted that this was well below the concentrations at which adverse effects in the fetus are reported and is also below the current clinical guideline (360 µM) for prevention of effects in the fetuses of pregnant PKU patients. The Panel noted that in young children who did not suffer from PKU, plasma levels of phenylalanine resulting from aspartame ingestion at or below the ADI (as either a bolus or other aspartame consumption patterns) were likely to remain below 240 μM. For pregnant women, the Panel noted that there was no risk to the fetus from phenylalanine derived from aspartame at the current ADI (40 mg/kg bw/day) in normal or PKU heterozygous individuals.
The Panel noted that it was currently not possible to include chronic endpoints in the postulated MoA. The Panel noted that the ADI previously derived by JECFA and SCF of 40 mg/kg bw/day was established based on the long-term animal studies using the default uncertainty factor of 100. The Panel considered that this remained appropriate for the evaluation of long-term effects of aspartame.
The current evaluation was based on analysis of human reproductive and developmental effects of phenylalanine in PKU patients, who are more susceptible than the general and PKU heterozygous population. Therefore, no additional allowance for toxicodynamic variability was required. The modelling of the aspartame dose-phenylalanine concentration response was based on data from PKU heterozygous individuals who at any dose would have a higher plasma phenylalanine concentration than the normal population; therefore, no additional allowance for toxicokinetic variability was required for the general population. The Panel concluded that exposures at or below the current ADI were not of safety concern for reproductive and developmental toxicity in humans excluding PKU homozygous individuals.
The Panel also conducted an uncertainty analysis of the assumptions used in the postulated MoA. While the Panel was not able to place a specific numerical value on the uncertainties related to these assumptions, the Panel considered that these assumptions would be more likely to overestimate than underestimate any potential developmental risk. Therefore, the Panel considered that the results of the uncertainty analysis further support its conclusion, that there is no safety concern for aspartame at the current ADI in normal and heterozygous subjects.
Overall, the Panel concluded from the present assessment of aspartame that there were no safety concerns at the current ADI of 40 mg/kg bw/day. Therefore, there was no reason to revise the ADI for aspartame.
The Panel emphasised that its evaluation of phenylalanine plasma levels from a dose of aspartame at the ensuing ADI is not applicable to PKU patients. These individuals require total control of dietary phenylalanine intake to manage the risk from elevated phenylalanine plasma levels. The Panel noted it is a requirement of EU legislation that products containing aspartame indicate through labelling that they contain a source of phenylalanine.
Conservative estimates of exposure to aspartame made by the Panel for the general population were up to 36 mg/kg bw/day at the 95th percentile. These were below the ADI. The current ADI for DKP is 7.5 mg/kg bw/day. Estimates of DKP exposure at the ADI for aspartame are below this ADI based on the current specification for DKP in aspartame (1.5 %) and up to the highest value (24 % for soft drinks) used from the whole database; this latter percentage was taken into account for all food categories where no degradation percentage of DKP was available. The Panel noted that high-level exposure estimates for the general population are up to 5.5 mg/kg bw/day at the 95th percentile, which is below the ADI. Finally, conservative estimates of exposure to methanol showed that aspartame-derived methanol contributed to less than 10 % of the total mean anticipated exposure to methanol from all sources.