Since the 1980s, the prevalence of obesity has more than doubled globally( 1 ) with significant implications in terms of the development of chronic conditions such as CVD, type 2 diabetes mellitus and hypertension( Reference Butland, Jebb and Kopelman 2 ). Given that the overall pattern of weight gain has been attributed to a culmination of a host of factors( Reference Butland, Jebb and Kopelman 2 ), it is of no surprise that a multi-faceted approach has been proposed to address the issue. One factor implicated in the development of weight gain, as well as a range of adverse health outcomes, is the over-consumption of non-milk extrinsic sugars from foodstuffs, particularly from sugar-sweetened beverages( Reference Malik, Schulze and Hu 3 – Reference Xi, Huang and Reilly 8 ). A logical strategy to address these issues would be to reduce the intakes of these beverages. However in practice, this approach may be difficult given the apparent innate preference for sweet taste among human subjects( Reference Drewnowski, Mennella and Johnson 9 ). To satisfy the desire for sweet-tasting products without exacerbating the problem of overconsumption of non-milk extrinsic sugars, low-calorie sweeteners (LCS) have become more commonly used as substitutes in a wide range of products( Reference Zygler, Wasik and Namiesnik 10 ).
This review will provide an overview of current understanding in relation to intakes of LCS and will consider existing methods for monitoring exposure to these compounds. It will also explore the possibility of implementing a novel biomarker approach for investigating LCS exposure, which could be used to gain a better understanding of LCS consumption and health.
Low-calorie sweeteners
LCS can be broadly divided into two groups; bulk sweeteners (e.g. polyols), which have a similar sweetness to sucrose and are also commonly used for other functional purposes in products( Reference Mortensen 11 , Reference Ghosh and Sudha 12 ), and intense sweeteners, which are many times sweeter than sucrose and are mainly used only for their sweetening properties( Reference Mortensen 11 ). The focus of the remainder of this review will be on intense sweeteners. Intense sweeteners which are currently approved for use in the European Union (EU), together with their general characteristics, are listed in Table 1. The use of LCS has increased considerably over recent years and they can now be found in a wide variety of food and non-food products( Reference Zygler, Wasik and Namiesnik 10 ). The level of sweetness that they impart, relative to sucrose, ranges from thirty times sweeter (cyclamates) to 37 000 times sweeter (advantame) and despite possessing the common characteristic of being intensely sweet, they represent a chemically diverse group of compounds. The overall contribution of LCS to energy intake is negligible and it is upon this basis, along with the fact that they are also non-cariogenic, that they are commonly used. Prior to approval, the safety of LCS consumption in human subjects is established through extensive evaluation of existing safety and toxicological data, usually culminating in the assignment of acceptable daily intake (ADI) and maximum permitted use levels( Reference Renwick 13 – Reference Logue, Peters and Gallagher 15 ).
Table 1. Intense sweeteners approved for use in Europe
NHDC, neohesperidine dihydrochalcone; ADI, acceptable daily intake; BW, body weight.
* Relative to sucrose.
† Expressed as steviol equivalents.
Acceptable daily intake
An ADI, which is expressed as mg/kg body weight, has been defined as the amount of a chemical that can be consumed daily over the period of a lifetime with no appreciable risk to health( 16 ). To assign an ADI, long-term, multiple-dose animal studies are typically used to initially establish the no observed adverse effect level by identifying the highest level of exposure which causes no adverse effects in the most sensitive species of animals( Reference Renwick 13 – Reference Logue, Peters and Gallagher 15 ). To account for variation between species and within human subjects, safety factors are then applied (Fig. 1). In the absence of serious adverse effects (e.g. teratogenicity) in animal studies, an overall safety factor of 100 is usually applied to the no observed adverse effect level( Reference Herrman and Younes 14 ), although higher or lower safety factors can also be applied.
Fig. 1. (Colour online) Safety factors applied to the no observed adverse effect level to establish the acceptable daily intake (Source: Logue et al. ( Reference Logue, Peters and Gallagher 15 )). ADI, acceptable daily intake; NOAEL, no observed adverse effect level.
Health effects of low-calorie sweeteners intake
Despite extensive evaluation of the safety of LCS, the potential health effects of consumption has remained topical within the area of nutrition research( Reference Mortensen 11 , Reference Kroger, Meister and Kava 17 , Reference Marinovich, Galli and Bosetti 18 ). Moreover, the long-term efficacy of using LCS in place of non-milk extrinsic sugars as a weight management tool has yet to be conclusively established( Reference de la Hunty, Gibson and Ashwell 19 – Reference Swithers 25 ). A recent meta-analysis of randomised controlled trials and prospective cohort studies into the effects of LCS consumption on weight status, reported that in prospective cohort studies, a small positive association was observed between LCS consumption and increased BMI (0·03 kg/m2) but not body weight or fat mass. However in randomised controlled trials, LCS consumption was associated with modest, albeit significant reductions in body weight (−0·80 kg), BMI (−0·24 kg/m2), fat mass (−1·10 kg) and waist circumference (−0·83 cm)( Reference Miller and Perez 26 ). The potential mechanisms by which LCS consumption might influence appetite and food intake were reviewed by Mattes and Popkin( Reference Mattes and Popkin 27 ) and they concluded that, although the evidence was lacking for many putative mechanisms (such as cephalic phase stimulation, gut peptide response and increased palatability of products), further research in the free living population via long-term randomised controlled trials was warranted. In addition to body weight status, LCS consumption in relation to a range of adverse health outcomes including cancer( Reference Weihrauch and Diehl 28 , Reference Schernhammer, Bertrand and Birmann 29 ), CVD( Reference Duffey, Steffen and Van Horn 30 , Reference Pereira and Odegaard 31 ), diabetes mellitus( Reference Wiebe, Padwal and Field 32 , Reference Imamura, O'Connor and Ye 33 ) and preterm deliveries( Reference La Vecchia 34 ) has also been investigated. No convincing evidence of a risk in the development of any adverse effects as a result of LCS consumption has been presented to date. The French Agency for Food, Environmental and Occupational Health and Safety recently undertook a review of the evidence with regard to many of these outcomes and concluded that, although the data do not demonstrate a risk, further research is required to establish the long-term beneficial effects of LCS consumption on health( 35 ). Furthermore, the French Agency for Food, Environmental and Occupational Health and Safety also recommended that future cohort studies should aim to distinguish the intakes of individual LCS, so that the effects of single and multiple LCS use can be investigated more effectively( 35 ). The ongoing debate around the safety of LCS consumption has served to fuel a somewhat negative perception within the lay media and the population in general. A recent study by Harricharan et al.( Reference Harricharan, Wills and Metzger 36 ) investigated the attitudes of dietitians from several European countries towards LCS and highlighted a diversity of opinions ranging from negative, ambivalent to positive; they suggested the provision of guidance similar to that which has been undertaken in the USA( 37 ).
Assessment of exposure to low-calorie sweeteners
In accordance with EU Regulation 1333/2008, EU Member States are required to monitor levels of LCS intake within the population to ensure that the ADI is not being exceeded( 38 ). Assessment of exposure to LCS requires the consideration of food intake data along with LCS concentrations within products and can be expressed, according to the International Programme on Chemical Safety( 39 ), as:
$$\displaystyle{{\Sigma ({\rm Food\; LCS\; concentration} \times {\rm Food\; consumption})} \over {{\rm Body\; weight}}}$$
Given that currently there are almost 400 food additives approved for use in the EU, the potential costs associated with collecting accurate intake data at the level of the individual are considerable. However, as the primary aim of monitoring is to ensure that the ADI is not being exceeded( 40 ), a tiered approach is usually adopted, beginning with a conservative screening step and progressing to more refined, and thus costly assessments, if indicated( 39 , 40 ).
Tiered approaches for assessment of low-calorie sweeteners exposure
An example of a tiered approach for assessing exposure to LCS is illustrated in Fig. 2. Tier 1, which is carried out at European wide level, will usually consist of an initial screening method designed to generate a highly conservative estimate so that potentially at risk groups can be easily identified. Examples of such screening methods include the Sweetener Substitution Method( Reference Renwick 41 ) and the Danish Budget Method( Reference Hansen 42 ). In the Danish Budget Method, conservative assumptions are made about the occurrence of LCS in food and beverages and an individual's energy and fluid requirements; this will result in an overestimate of exposure. If, following the initial screening step, the ADI is deemed to be exceeded, further more refined assessments would then be indicated. Tier 2 involves the assessment of actual consumption of foods and beverages known to contain LCS along with the maximum permitted use levels. A Tier 3 assessment would be indicated if the ADI is estimated to be exceeded following the Tier 2 assessment and this will consist of a more refined calculation using actual levels of consumption and actual concentrations in products in order to further elucidate the risk.
Fig. 2. Tiered approach for food additive exposure estimates (adapted from EC( 40 )). MPLs, maximum permitted levels; ADI, acceptable daily intake.
Methods of assessing exposure to low-calorie sweeteners
Dietary intake is traditionally assessed using tools such as food diaries, 24-h recalls and FFQ( Reference Thompson, Subar, Coulston, Boushey and Ferruzzi 43 ). These assessment tools rely on self-reported data and a number of inherent limitations exist which often results in inaccuracies( Reference Livingstone, Prentice and Strain 44 – Reference Kipnis, Midthune and Freedman 47 ). The optimal methodology for assessing habitual LCS exposure is a 1–2-week prospective, brand level diary, including information on portion sizes along with brand-specific information on LCS concentrations in products( Reference Renwick 48 ). Given that obtaining such refined individual level data can be time and resource intensive for investigators, as well as labour intensive for participants, a variety of dietary assessment tools have been utilised in the past including 24-h recalls( Reference Devitt, Daveman and Buccino 49 – Reference Ha, Ha and Choi 53 ), FFQ covering various durations( Reference Toledo and Ioshi 54 – Reference Lino, Costa and Pena 57 ) and food diaries lasting 2 d( Reference van Rooij-van den Bos, Konings and Heida 58 ), 5 d( Reference Garnier-Sagne, Leblanc and Verger 59 ), 7 d and 14 d( Reference Leclercq, Berardi and Sorbillo 60 ). Some studies used a combination of retrospective and prospective intake data in order to first identify potential high consumers and then to further investigate these high consumers( Reference Arcella, Le Donne and Piccinelli 61 , 62 ). It is apparent therefore that many of the published studies do not satisfy these criteria, potentially making it difficult to make direct comparisons.
Data on LCS concentrations in products can be obtained from a number of sources; the least resource intensive, yet least accurate, is to use maximum permitted use levels. This source of information is commonly used in initial, conservative estimates of exposure, or when it is not possible to obtain more refined data( Reference Renwick 48 ). However, the actual levels used in products are unlikely to meet these values, particularly when a blend of two or more LCS is used within a particular product, as is frequently done. Therefore using maximum permitted use levels in the calculation will result in an overestimate of exposure. Concentrations in specific products have been obtained from manufacturers( Reference Devitt, Daveman and Buccino 49 , Reference Serra-Majem, Bassas and Garcia-Glosas 56 , Reference Leclercq, Berardi and Sorbillo 60 – 64 ), providing a more accurate measure; however, difficulties in obtaining such information in the past have been highlighted( 65 , 66 ). Furthermore, with product innovations and changing tastes among consumers, the concentrations of LCS within products are likely to evolve over time, potentially introducing error into subsequent estimates of exposure unless ongoing data are received from manufacturers. Another method for determining LCS concentrations in products is to directly measure them analytically and to this end, numerous methodologies for the determination of LCS in foods and beverages have been published( Reference Zygler, Wasik and Namiesnik 10 ). Recent studies investigating LCS exposure have adopted this approach( Reference Chung, Suh and Yoo 50 – Reference Ha, Ha and Choi 53 , Reference Lino, Costa and Pena 57 , Reference van Rooij-van den Bos, Konings and Heida 58 ) and, although it allows for a more accurate and objective measure of LCS concentrations in products, it is also likely to be costly. Furthermore, with the ubiquity of LCS in today's market and the trend towards more widespread use, adopting such an approach as part of future assessments may prove unfeasible.
Recent developments in low-calorie sweeteners exposure assessment
A desire to harmonise food additive exposure assessment across the EU led to guidelines on how Member States should collect intake data for exposure assessments( 40 ) and this was further enhanced through the recent implementation of the Flavourings, additives and food contact exposure tool (FACET)( 66 ). As a result of the FACET project, a publicly available exposure software package has been released (available at: http://expofacts.jrc.ec.europa.eu/facet/login.php). One of the strengths of the FACET project is that cooperation was obtained from FoodDrinkEurope, an alliance of national food and drink industries in the EU, for the provision of information on the concentrations of targeted food additives in products, including aspartame and acesulfame-K and this will result in a more sustainable system of monitoring exposure over time.
Alternatively, the Monte Carlo risk assessment software tool, developed as part of the EU-wide ACROPOLIS (Aggregate and cumulative risk of pesticides: an on-line integrated strategy) project, may also be used for the assessment of exposure to chemicals in foods (available at: https://mcra.rivm.nl/Account/Login). This tool is designed to allow a cumulative assessment of exposure to multiple chemicals from multiple sources( Reference van Klaveren, Kennedy and Moretto 67 ) and has been successfully applied for the assessment of acute and chronic exposure to a group of pesticides in a number of European countries( Reference Boon, van Donkersgoed and Christodolou 68 ). Like the FACET tool, the ACROPOLIS tool is freely accessible.
Actual low-calorie sweeteners exposure
European based studies carried out over the last 20 years have largely reported that the intakes of LCS fall well within the ADI (Table 2), with only the intake of cyclamate potentially exceeding the ADI in some population sub-groups. Similar results were reported in studies conducted in Korea, Australia and New Zealand and Brazil( Reference Chung, Suh and Yoo 50 – Reference Ha, Ha and Choi 52 , Reference Leclercq, Berardi and Sorbillo 60 ). A review by Renwick( Reference Renwick 48 ) reported that the overall intakes of LCS had not increased significantly during the preceding decade, although it has been reported elsewhere that the numbers of people consuming LCS were increasing( Reference Sylvetsky, Welsh and Brown 69 ).
Table 2. Exposure estimates of low-calorie sweeteners in Europe over the past 20 years
No data reported for other LCS. ‘–’ indicates no data reported.
Ace-k, acesulfame-K; ADI, acceptable daily intake; Asp, aspartame; Cyc, cyclamate; Sac, saccharin; Suc, sucralose; FSA, Food Standards Agency.
* Mean % ADI presented with high consumers (% ADI) where available.
Nutritional biomarkers: concepts and considerations
The focus of this review will be on nutritional biomarkers of exposure and within this specific context, biomarkers will be defined as cellular, biochemical, analytical, or molecular measures that are obtained from biological media such as tissues, cells or fluids and are indicative of exposure to an agent( 39 ). As such, nutritional biomarkers of exposure, by their very nature, are independent of the sources of bias associated with self-reported dietary intake data and can therefore provide a more objective measure of intake( Reference Kaaks, Riboli and Sinha 70 , Reference Bingham 71 ). Such biomarkers can be used as measures of intake, for the assessment of nutritional status or in order to validate more traditional dietary assessment tools( Reference Potischman and Freudenheim 72 ). Although the application of a biomarker approach is not new in nutrition research, it has been suggested that many existing nutritional biomarkers have not been properly validated and the field of biomarkers is yet to be fully exploited( Reference Jenab, Slimani and Bictash 73 ). A number of considerations are important for properly implementing such an approach; the target biomarker must be specific to the food or food component, be reproducible and be sensitive to changes in intakes over time( Reference Hedrick, Dietrich and Estabrooks 74 ). Furthermore, the relevant biomarker should be obtained in a minimally invasive way( Reference Crews, Alink and Anderson 75 ) and the biological sample should be collected, processed and stored in an appropriate manner, so that a true reflection of intake can be obtained( Reference Wild, Andersson and O'Brien 76 ). Factors that may affect the validity of a biomarker include genetic variability, physiologic factors, dietary factors, the biological sample of choice and the analytical method used to measure it( Reference Jenab, Slimani and Bictash 73 ).
Four classes of nutritional biomarkers, namely recovery, concentration, replacement and predictive biomarkers, have been described according to the relationship between the biomarker and intake of the component of interest( Reference Jenab, Slimani and Bictash 73 ). Recovery biomarkers exhibit a strong time defined relationship between intake and excretion and can therefore be used to estimate absolute intakes. Examples of such biomarkers are doubly labelled water for energy intake( Reference Livingstone and Black 77 ) and urinary nitrogen( Reference Day, McKeown and Wong 78 , Reference Bingham 79 ) and potassium( Reference Day, McKeown and Wong 78 ) for protein and potassium intakes, respectively. Concentration and replacement biomarkers differ from recovery biomarkers in that they exhibit a lower correlation with absolute intake( Reference Jenab, Slimani and Bictash 73 ); however, they are useful for the purposes of ranking individuals according to intake and therefore can be used in the investigation of the relationship between the food or food component and disease( Reference Potischman and Freudenheim 72 ). Examples of such biomarkers are carotenoids and aflatoxins( Reference Jenab, Slimani and Bictash 73 ). The class of predictive biomarkers was first proposed by Tasevska et al.( Reference Tasevska, Runswick and McTaggart 80 ) when describing the use of urinary fructose and sucrose as markers for sugar intake. Although relatively small amounts of a dose were recovered in the form of urinary sucrose and fructose, it was demonstrated that a higher level of correlation (R > 0·6) with dietary intake existed than with concentration or replacement biomarkers and therefore this class of biomarker would fall between recovery and concentration biomarkers. As part of the validation process, it is important to characterise the relationship between the target biomarker and intake of the food or food component of interest( Reference Kuhnle 81 ), as such information will inform the application of the biomarker (e.g. to estimate absolute intakes, rank individuals or monitor compliance).
Two broad strategies for the development of a biomarker approach have been described; discovery- and hypothesis-driven( Reference Kuhnle 81 ). Discovery-driven approaches have become more prominent recently with the use of metabolomics to identify previously unknown biomarkers or panels of biomarkers that are associated with dietary patterns or the consumption of specific foods or food components( Reference O'Sullivan, Gibney and Brennan 82 ). Hypothesis-driven approaches differ in that prior knowledge of the component of interest and its metabolic fate are required and this information subsequently informs a more targeted approach to biomarker development( Reference Kuhnle 81 ).
Potential application of a biomarker approach for investigating low-calorie sweeteners intakes
The metabolic fates of LCS are well known (Table 3) and therefore a hypothesis-driven approach would appear to be the most appropriate for the implementation of a biomarker approach for investigating exposure. Following ingestion, aspartame is hydrolysed to aspartic acid, phenylalanine and methanol, each of which commonly occur in a normal diet( Reference Butchko, Stargel and Comer 83 , Reference Magnuson, Burdock and Doull 84 ), while thaumatin undergoes normal protein digestion( 85 ). Neohesperidine dihydrochalcone, although not known to exist in nature, is structurally similar to naturally occurring flavonoid glycosides and undergoes a similar metabolic fate to these analogues with the same or similar metabolites( Reference Borrego, Montijano and O'Brien Nabors 86 ). This finding would indicate that no obvious specific candidate biomarkers exist for these compounds. Acesulfame-K( Reference Christ and Rupp 87 ) and saccharin are almost completely absorbed and excreted unchanged via the urine,( Reference Byard, McChesney and Golberg 88 – Reference Sweatman, Renwick and Burgess 90 ) while cyclamate (30–50 %)( Reference Renwick and Williams 91 ) and sucralose (10–15 %)( Reference Grice and Goldsmith 92 , Reference Roberts, Renwick and Sims 93 ) undergo partial absorption and the absorbed proportion is excreted unchanged via the urine with the unabsorbed proportions excreted via the faeces. In about 20 % of the population, cyclamate can be converted to cyclohexylamine via bacterial hydrolysis in the gut, which is absorbed and also excreted via the urine( Reference Bopp, Sonders and Kesterson 94 ). Furthermore, the extent of cyclamate conversion to cyclohexylamine can be variable during chronic exposure( Reference Renwick, Thompson and O'Shaughnessy 95 ). Steviol glycosides also undergo bacterial hydrolysis to steviol which is then absorbed and excreted via the urine as steviol glucuronide( Reference Geuns, Buyse and Vankeirsbilck 96 – Reference Wheeler, Boileau and Winkler 98 ). Advantame is converted to advantame acid and a small proportion is absorbed (about 6 %) and excreted via the urine while about 90 % of a dose is excreted via the faeces( Reference Ubukata, Nakayama and Mihara 99 ).
Table 3. Metabolic fates and routes of excretion of low-calorie sweeteners approved in Europe
CAS, Chemical Abstract Service; NHDC, neohesperidine dihydrochalcone; N/A, not applicable as broken down to normal dietary components; JECFA, Joint FAO/WHO Expert Committee on Food Additives.
* No CAS Number.
† Principal route of excretion listed first.
Saccharin, acesulfame-K, cyclamate and sucralose undergo no or limited metabolism following absorption into the body; therefore candidate biomarkers, in the form of the parent compounds, exist for these LCS. For steviol glycosides and advantame, the excretory products may act as suitable biomarkers for intakes. A high level of specificity of the candidate biomarkers exists as they are not found elsewhere in the diet or formed endogenously and given that at least a proportion of each of these compounds is excreted via the urine, a urinary biomarker approach may be feasible. Indeed, such an approach for assessing exposure to acesulfame-K and saccharin was previously described by Wilson et al.( Reference Wilson, Wilkinson and Crews 63 ) who measured levels of excretion in 24-h urine samples and found excellent levels of correlation in an intake/excretion study (R 2 > 0·99 for both compounds), demonstrating a clear dose–response relationship for both compounds. Slightly lower correlations were observed when validated against an FFQ which is probably more indicative of a limitation with the FFQ rather than the biomarker as the FFQ did not account for non-dietary sources of the LCS. The dose–response relationships for cyclamate, sucralose, steviol glycosides and advantame, however, are less clear so future work will be required to elucidate these relationships before the usefulness of measuring urinary concentrations for investigating intakes is established.
To fully characterise the relationship between urinary excretion and intakes of these LCS, a suitable and validated analytical method is first required( Reference Wild, Andersson and O'Brien 76 ). To this end, a liquid chromatography, tandem-MS/MS method of simultaneously determining urinary levels of acesulfame-K, cyclamate, saccharin, steviol glucuronide and sucralose has recently been developed and validated( Reference Logue, Dowey and Strain 100 ) and allows for investigations of the feasibility of implementing a biomarker approach for assessing intakes of these compounds.
Potential limitations of a urinary biomarker approach for assessing low-calorie sweetener intakes
Although a biomarker approach would potentially offer many advantages over more traditional methods of investigating exposure to LCS, potential limitations should also be acknowledged. As urinary biomarkers, only short-term exposure (previous 24–48 h) can be investigated( Reference Kuhnle 81 ). The mode of sampling is also an important factor to consider for the implementation of a biomarker approach and if 24-h urine samples are required, a lack of compliance could also represent a limitation. To mitigate this limitation, however, methods of assessing compliance have been developed such as the paraminobenzoic acid method( Reference Bingham and Cummings 101 ). It must also be acknowledged that a biomarker approach, utilised on its own, will not provide information on the source of exposure. Therefore in the event that the ADI of a particular LCS is being exceeded, more traditional methods would be required to identify the main sources of exposure( 39 ). As such, for the purposes of monitoring exposure in the population, a biomarker approach would likely be most useful when used as an adjunct to more traditional methods.
In summary, knowledge of the intakes of LCS is a legislative requirement in the EU and although the recent development and implementation of the FACET project will help harmonise methods among EU Member States and improve knowledge with regard to exposure to food additives, a successfully implemented biomarker approach for investigating LCS intake would undoubtedly be a useful adjunct to such monitoring activities. Furthermore, with ongoing interest and debate around the efficacy, as well as safety, of long-term LCS use, a biomarker approach would help elucidate the intakes of specific and combinations of LCS, and thereby address a limitation in the evidence to date, as highlighted by the French Agency for Food, Environmental and Occupational Health and Safety.
Financial Support
This work was principally funded by the National Institute for Public Health and the Environment (RIVM), The Netherlands as part of a PhD studentship for C. L. Part of this work was also funded by the JPI-HDHL project Foodball (Project no. 50-52905-98-12).
Conflicts of Interest
None.
Authorship
C. L. conducted a literature search and drafted the manuscript. A. M. G., L. C. D., J. J. S. and H. V. reviewed and approved the final version.
