Review of data on the food additive aspartame
The present document has been produced and adopted by the bodies identified above as author(s). This task has been carried out exclusively by the author(s) in the context of a contract between the European Food Safety Authority and the author(s), awarded following a tender procedure. The present document is published complying with the transparency principle to which the European Food Safety Authority is subject. It may not be considered as an output adopted by EFSA. EFSA reserves its rights, view and position as regards the issues addressed and the conclusions reached in the present document, without prejudice to the rights of the authors.
AbstractNo abstract available
Project developed on the procurement project CT/EFSA/ANS/2011/03
This draft preparatory document contains a review of data on aspartame, undertaken to support the EFSA re-evaluation of the safety of this food additive. The material evaluated during this review comprised information submitted in response to the EFSA call for data on aspartame and publications identified in the course of a comprehensive search of the published literature on aspartame. This document summarises the findings of the review, provides an overview of current knowledge regarding the metabolism and toxicity of aspartame and presents an in depth quality assessment of 15 toxicity studies.
Results of literature search
A tiered strategy was adopted for the literature search. In the first tier, a wide-ranging, high level overview of the literature was undertaken in order to identify all publications in which aspartame and/or other intense artificial sweeteners is/are mentioned. This generated a database containing 5023 references.
In the second tier the primary database was filtered, focusing on the published literature that could contribute directly or indirectly to the safety assessment of aspartame, and unpublished documents from the EFSA call for data were added manually. The result of this process was a filtered database listing 1326 documents of direct or indirect relevance to the risk assessment of aspartame. Copies of references identified during the literature search which are not included in the EFSA database were obtained and will be provided to EFSA upon completion of the project.
The database was subsequently amended twice; 37 references on the epidemiology of bladder cancer which also mention artificial sweeteners were reinstated in response to a request from EFSA and three references were added to support statements made in reports. The version of the database submitted upon completion of the project therefore contains 1366 references.
In order to ensure evaluation of all these documents in a consistent, well-structured and timely manner, a proforma-based approach was adopted in the next stage of the project. The proforma used for this review was designed to facilitate the review of each document in the database by:
capturing the publication details, source, topic, type and quality of each document;
recording an expert judgment as to which studies meet the reliability, relevance and adequacy criteria for further in depth analysis;
identifying documents which require input from the quality assurance (QA) and/or pathology experts on the team.
The results of the proforma-based review were used to facilitate the selection of documents for further evaluation. The tiered selection strategy involved initial selection by document category, then selection by topic. Following application of these criteria, 358 documents were taken forward for further evaluation. Information from these documents was used to create the summary of data which will inform the risk assessment of aspartame.
Summary of the data
The approved use of aspartame is as a dietary sweetener; as such, the only route by which consumers are exposed to this compound is via ingestion (although other routes may become relevant in the occupational setting).
Following oral exposure, aspartame undergoes hydrolysis catalysed by esterases and dipeptidases leading to release of its individual components (aspartate, phenylalanine and methanol). This may occur either in the lumen of the gastrointestinal tract or within intestinal mucosal cells; either way, it is the individual components which undergo absorption. Intact aspartame is not detected in the systemic circulation, and aspartame therefore has an effective oral bioavailability of zero.
Two of the three components of aspartame (aspartate and methanol) are cleared rapidly from the circulation. Under some circumstances (e.g. when aspartame is ingested in the absence of carbohydrates) aspartate is briefly measurable in the plasma following ingestion, but under normal conditions of use aspartate is rapidly metabolised and excreted predominantly in expired CO2, although a small proportion may be incorporated into proteins. Even after ingestion of large doses of aspartame, blood methanol levels usually remain below or very close to the limits of detection.
The effective pharmacokinetics of aspartame are therefore those of its phenylalanine component. Plasma phenylalanine usually peaks about 60 min after ingestion and is then cleared from the circulation over a period of about 2-3 hours depending on the dosing regimen used. If repeated high doses are administered (e.g. at 2 hour intervals), phenylalanine may accumulate and reach a plateau after about 6 hour; however, levels invariably return to background following overnight abstinence. The ultimate fate of the phenylalanine component of aspartame is incorporation into proteins.
The potential physiological consequences of exposure to aspartame-derived phenylalanine are a function of its effects on the phenylalanine/large neutral amino acid ratio. It has been suggested that changes in this ratio could affect the uptake of other large neutral amino acids via the blood-brain barrier leading to changes in neurotransmitter levels in key brain regions; however, no consistent changes in neurotransmitter levels either in whole brain or specific brain regions have been identified following exposure of rodents to aspartame, even at high doses. In this context, a species difference in the metabolism of phenylalanine should be noted: rodents have much higher hepatic phenylalanine hydroxylase activity than do humans. Rodents therefore metabolise aspartame-derived phenylalanine rapidly to form tyrosine, so that dosing with aspartame leads to an increase in plasma tyrosine and the tyrosine/large neutral amino acid ratio, whereas in humans the main effect observed is on plasma phenylalanine itself and on the phenylalanine/large neutral amino acid ratio.
No significant acute or subchronic toxicity has been observed in animal models even at the highest doses of aspartame which could reasonably be administered, and humans given high doses over periods of up to 27 weeks have reported no significant adverse effects. Early concerns that aspartame might cause neurotoxicity in neonates and infants were not substantiated by subsequent experimental data.
There is no evidence to indicate that aspartame is genotoxic, either in vitro (with or without metabolic activation) or in vivo. Occasional marginal positive results have been reported, but such results only occur sporadically and do not indicate any particular cause for concern.
The results of chronic toxicity studies carried out in the 1970s did not indicate any significant neoplastic or non-neoplastic pathological effects in rodents dosed with aspartame via the diet. These studies were unsatisfactory in a number of ways, including the presence of infections requiring antibiotic treatment in one study, and they were subjected to detailed investigation by the US Bureau of Foods in the 1970s. The outcome of the investigation was that the studies were essentially reliable but that they were limited by a number of shortcomings in terms of diet preparation, protocol compliance and QA. Furthermore, when the studies were issued, concern was expressed regarding a possible excess of brain tumours in rats dosed with aspartame via the diet; however, further detailed examination of tissue from these studies did not indicate any treatment related effects on brain tumours.
Three additional chronic toxicity studies (two in rats and one in mice) were recently reported by the European Ramazzini Foundation of Oncology and Environmental Sciences (ERF). In these studies, aspartame was administered from early life (in utero in one study) until the natural death of the animals. Increased incidences of a variety of tumour types (including, in particular, lymphomas and leukaemias) were reported in these studies, which have been the subject of intense controversy since the first one was published in 2005. The studies are difficult to interpret because the increased tumour incidences are close to those seen in control animals in other studies and there is evidence that the animals may have been infected with the organism Mycoplasma pulmonis. Furthermore, a number of methodological issues call these studies into question. These include allowing the animals to die naturally, storing the carcasses for up to 19 hr before performing necropsies, and fixing the tissues in ethanol rather than a conventional fixative such as 10% neutral buffered formalin which would give better morphology.
Overall, the available chronic toxicity studies do not indicate any overt carcinogenic effect in experimental animals due to aspartame, but all the conventional studies are limited in various ways. These studies were subjected to more detailed review during the quality assessment of key studies received by EFSA. Reassurance has recently been provided by a Good Laboratory Practice (GLP)-compliant study conducted by the National Toxicology Program (NTP) using transgenic mouse models. This indicated that aspartame had no carcinogenic effect in Tg.AC, p53+/- or Cdkn2a-/- mice dosed for 9 months via the diet.
Concern regarding possible carcinogenic effects of aspartame has mainly focussed upon tumours of the bladder and brain. The issue of bladder carcinogenesis arose mainly by analogy with saccharin, which is known to induce bladder tumours via a male rat specific mechanism involving α2-microglobulin. A number of epidemiological studies were conducted in the between the 1970s and the 1990s in an attempt to determine the relevance of these findings to humans; their results were variable but the majority did not indicate any increase in risk of bladder cancer in humans due to artificial sweetener consumption. None of these studies addressed aspartame directly and many of them were carried out in populations diagnosed before aspartame entered the market.
The issue of brain tumour induction by aspartame arose partly because of the suggestion of increased brain tumour incidence in chronic rat studies and partly following the publication of a paper stating that an increase in brain tumour incidence in the US had coincided with the release of aspartame. Subsequent epidemiological studies have not identified any association between brain tumour incidence and use of aspartame; indeed, studies of artificial sweetener use in which risks associated with saccharin and other sweeteners (presumed to represent mainly aspartame) are considered separately provide no evidence for an increased risk of any tumour type in humans.
The reproductive toxicity studies carried out prior to release did not indicate any adverse effects due to aspartame in mice, rats or rabbits. However, these studies were not conducted to modern standards (they predate the introduction of GLP regulations) and several of them are compromised by problems with the use of an inappropriate control diet. Dietary studies in rabbits also proved problematic because the rabbits failed to consume diets containing high doses of aspartame, so that the actual doses ingested did not correspond to those planned. There are no modern, GLP-compliant studies addressing the reproductive toxicity of aspartame.
Only one epidemiological study has addressed the possible reproductive effects of aspartame in humans. This study addressed preterm delivery and examined the effects of soft drink consumption thereon. It found an increased risk of preterm delivery in women who frequently ingested either carbonated or (to a lesser extent) non-carbonated diet drinks. This study did not address aspartame directly and it was subject to a number of confounding factors, but as a large, well-designed study its results merit careful consideration and further investigation.
Ever since its release, anecdotal reports have attributed numerous adverse effects in humans to consumption of aspartame. The effects most commonly reported include hypersensitivity reactions, headaches, seizure induction and behavioural changes. However, when these are addressed by properly designed clinical trials it usually proves impossible to detect any effect due to aspartame. Only sporadic effects in individual subjects are observed, and these occur in response to placebo at least as often as in response to aspartame. Similarly, when effects such as seizure induction and behavioural change are addressed in animal models the number of positive findings is well within the range expected by chance, although it should be noted that grand mal seizures were observed in monkeys treated with high doses of aspartame during the only chronic toxicity in primates. These seizures were attributed to high plasma phenylalanine concentrations (similar effects were observed in response to equimolar doses of phenylalanine) and to stress caused by handling.
With regard to potentially vulnerable groups, it is important to note that phenylketonurics (who have a genetic deficiency in phenylalanine hydroxylase activity) represent a known and well-characterised susceptible subgroup within the population. These individuals need to restrict their intake of aspartame, and aspartame-containing products are required to carry warning labels to this effect. However, controlled administration of single doses of aspartame to phenylketonurics has provided reassurance that the inadvertent ingestion of a small dose by these individuals is not hazardous. Furthermore, studies on phenylketonuria heterozygotes (“carriers”) have demonstrated that, while they exhibit higher plasma phenylalanine peaks and slower clearance than controls following ingestion of aspartame, no accumulation or long term effects are observed.
Other population subgroups whose potential vulnerability to adverse effects following ingestion of aspartame has been investigated include monosodium glutamate-sensitive individuals, the obese, diabetics, individuals with alcoholic liver disease, Parkinson’s disease sufferers and individuals with mood disorders. No untoward effects have been reported in any of these groups.
There is no consistent evidence that aspartame has adverse effects, either in healthy individuals or in potentially susceptible groups, under normal conditions of use although phenylketonurics do need to regulate their intake of aspartame for health reasons. They are supported in doing this by clear labelling of aspartame-containing products.
Quality assessment of key studies
For the detailed review of 15 key studies, a consistent approach was ensured by adopting a proforma-based methodology. Every study was reviewed by at least two team members; carcinogenicity studies were considered by all three reviewers.
The studies conducted by or on behalf of Searle in the early 1970s were performed according to standards which were considered appropriate at the time; indeed, they were probably “state of the art” for the period. In order to provide a modern day perspective, the QA assessment conducted in this review evaluated these studies against current (i.e. 2012) GLP regulations and Organisation for Economic Co-operation and Development (OECD) guidelines.
In general, no major deficiencies were identified but every study was found to lack certain key items that would exist in a GLP study and minor inconsistencies in data presentation were identified. It was not possible to confirm protocol objectives and any changes, as protocols for these studies were not provided. The main criticism made, which applies to all the Searle studies, was that they contain insufficient information to support adequate characterisation of the test item(s), including the batch(es) used, and for the duration of the test its actual concentration, homogeneity and stability in the diet mix, and the actual concentrations consumed.
Despite these caveats, the view of the review team is that these studies are adequate for use in risk assessment as long as their limitations are taken into account.
The study reported by Soffriti (Soffritti and Belpoggi, 2005) has been the subject of intense controversy. The most important concerns of the review team were that it claims compliance with GLP but lacks supporting documentation and that there are significant doubts about the pathology findings. The opinion of the review team is that, before the results of this study are used for risk assessment, the pathology should be subjected to a comprehensive independent peer review and the whole study should be audited.
The study reported by Ajinomoto in 2006 (Iwata, 2006) was difficult to evaluate because it only comprises histopathological re-evaluation of a previous study which could not be identified unequivocally. The review team had significant concerns regarding this study and took the view that it should not be used for risk assessment unless/until these issues are resolved.
Overall, the opinion of the review team is that the Searle studies can be used in risk assessment if their limitations are taken into account, but issues relating to the Ajinomoto and ERF studies require resolution before these studies can be relied upon. The team notes that the only two generation study, while without major quality issues, has serious limitations. Reassurance would be provided by conducting another two generation study according to current OECD guidelines.
A huge amount of evidence already exists regarding the potential adverse effects of aspartame, both in animal models and in humans, and most of this is reassuring. Overall, the opinion of the review team is that the studies conducted by or on behalf of Searle in the 1970s can be used in risk assessment if their limitations are taken into account, but issues relating to the ERF and Ajinomoto studies require resolution before they are relied upon. Specifically:
The ERF study should be subjected to independent audit; its pathology findings should undergo a comprehensive independent peer review.
The Ajinomoto study should be authenticated with respect to the identity of the original study to which it relates; in particular, the genetic background of the rats used in this study should be confirmed.
The team notes that the only two generation reproductive toxicity study, while without major quality issues, is limited with respect to procedures carried out, parameters measured and outcomes reported. Reassurance would be provided by conducting another two generation study according to current OECD guidelines.
There are, therefore, two areas in which the existing data, while still essentially reassuring, are compromised by issues of methodology and QA:
The original chronic toxicity studies conducted by Searle and the more recent studies carried out by the European Ramazzini Foundation of Oncology and Environmental Sciences have created vigorous debate regarding possible carcinogenic effects due to aspartame. Recent National Toxicology Program studies in transgenic mice are reassuring; however, a GLP-compliant 2-year bioassay conducted by an organisation such as the National Toxicology Program would help to resolve the issue of carcinogenicity once and for all.
The reproductive toxicity studies conducted by Searle in the 1970s, while essentially reassuring, were conducted to the standards current at the time and beset by technical difficulties. A modern, GLP-compliant reproductive toxicity study conducted to the appropriate guidelines would, it is to be hoped, confirm the lack of any embryotoxic, teratogenic or reproductive effect due to aspartame. Such further data would provide further reassurance, especially in the light of recent epidemiological findings.Published: 1 March 2013