Use of sucralose in foods heated during manufacturing does not pose a risk to human health

Experimental design

EFSA requested that realistic uses of sucralose in a range of foods subjected to heating under realistic manufacturing conditions be analyzed for the presence of specific non-sucralose chlorinated substances. The foods specified were:

Each product was to be evaluated for the presence of:

In order to determine the effects of typical baking heat processing treatments on the formation of compounds of interest, four model food systems were formulated and processed under controlled conditions. Model food systems comprised a wafer, cake, biscuit (cookie) and pizza sauce (placed on a typical pizza crust to recreate typical heat treatment conditions and then collected after baking) (Tables 1 -4).

Each model food system incorporated the use of either sugar (control), sucralose or a retail sucralose-based sweetener with maltodextrin (SPLENDA® Granulated), which provides sugar-like bulk with relatively minimal calories per serving, or maltodextrin alone. The use levels chosen for sucralose, or retail sucralose-based sweetener with maltodextrin, represented amounts typically found in commercial applications and levels commonly recommended for home product preparation or greater. In order to avoid complicating experimental variables, when sucralose ingredients were applied in the experimental recipes, no additional ingredients were added to rebalance the recipes for the removal of sugar.

The presence of the toxic compounds of interest were analyzed using validated and accepted analytical methods - high resolution gas chromatography (HRGC)/high resolution mass spectrometry (HRMS) for PCDDs and PCDFs ²⁸ and GC-MS/MS for free and bound 3-MCPD. ²⁹

Ingredients

Sucralose (SPLENDA® Sucralose, micronized powder) and maltodextrin (STARDRI® 100) were obtained from Tate & Lyle Ingredients Americas LLC (Hoffman Estates, IL, USA). SPLENDA® Granulated (1% sucralose and 99% maltodextrin by weight), a dry, granulated, low-calorie sweetener designed to replace sugar spoon-for-spoon, on a sweetness basis, and all other recipe ingredients were purchased from a local retail grocery store. Ingredients were representative of those typically used in home and/or industrial baking applications.

Formulations and preparation

Food products were prepared as described below (Tables 1 -4). Stainless steel cookware and utensils were used for preparation of pizza sauce samples. The pizza sauce samples were cooked using a Vollrath Induction burner. Dough or batter mixing was accomplished using a Hobart® planetary mixer (Model N50A) with paddle attachment. All traditional baking was accomplished using a Picard Revolution reel gas heated oven and stainless steel baking sheets or muffin tins. A stainless steel non-stick cast electric waffle cone baker (Cone Baker Model XC-XG-1) was used for preparation of the wafer samples.

Each food type was produced in triplicate. Samples (minimum 250 g) were collected from each production run, placed in a stainless steel bowl, homogenized with an immersion blender (Braun MQ505) and ground until visually homogenous. Samples of this ground material were placed in 100 mL glass sample containers sealed with plastic lids, placed in frozen storage and held at -23°C until samples were shipped to the contract laboratory for analysis.

Sample preparation and analysis

For each of the three individual baking run replicates (for each of the experimental variables), representative samples were shipped in 100 ml glass sample containers, sealed with plastic lids in insulated shipping containers held at approximately 0-5°C. Samples were shipped to Eurofins Lancaster Laboratories Environmental, Lancaster, PA, USA, for the analysis of PCDD and PCDF, and to Eurofins Food Integrity & Innovation, Madison, WI, USA, for the MCPD analyses. Upon receipt of samples, they were kept at 0-6°C at both locations until sample preparation and analysis.

PCDD and PCDF

An additional 8 samples (2 samples for each of the 4 products) were subjected to spike recovery analysis of PCDD/PCDF. Spike recovery was conducted as follows:

Spike recovery results for PCDD and PCDF are presented in Table 5.

OPEN IN VIEWER

Table 5.

Recovery of PCDDs and PCDFs.

Compound	Wafer		Cake		Cookie/biscuit		Pizza sauce
	Lower	Upper	Lower	Upper	Lower	Upper	Lower	Upper
2378-TCDD	90%	109%	94%	94%	95%	98%	97%	103%
2378-TCDF	97%	100%	92%	93%	95%	98%	92%	104%
12378-PeCDD	95%	106%	92%	98%	92%	95%	95%	98%
12378-PeCDF	102%	109%	97%	101%	99%	103%	99%	109%
23478-PeCDF	110%	113%	104%	108%	107%	109%	110%	115%
123478-HxCDD	102%	108%	98%	101%	99%	100%	100%	110%
123678-HxCDD	102%	107%	96%	98%	100%	103%	102%	104%
123789-HxCDD	91%	101%	92%	95%	92%	95%	95%	102%
123478-HxCDF	99%	106%	98%	100%	97%	98%	101%	103%
123678-HxCDF	98%	106%	98%	98%	97%	100%	102%	106%
123789-HxCDF	97%	106%	95%	97%	96%	101%	101%	102%
234678-HxCDF	98%	106%	97%	102%	96%	100%	100%	101%
1234678-HpCDD	99%	104%	96%	97%	98%	101%	100%	100%
1234678-HpCDF	98%	108%	96%	97%	97%	99%	101%	103%
1234789-HpCDF	99%	105%	94%	97%	94%	98%	100%	105%
OCDD	95%	104%	94%	95%	93%	97%	97%	101%
OCDF	96%	104%	93%	98%	94%	99%	99%	104%

3-MCPD

Four samples (two solid samples and two liquid samples) were used for spike recovery analysis of free and bound 3-MCPD. An appropriate amount of spiking reagent was added to the sample tube as follows: for solid samples, 100 µL of a mixed solution of 3-MCPD ester, 2-MCPD ester, and glycidol to obtain sample spiked at 0.05 µg/g of ester forms, and 200 µL of a mixed solution of free 3-MCPD and free 2-MCPD to obtain sample spiked at 0.01 µg/g of free forms; for liquid samples, 100 µL of a mixed solution of 3-MCPD ester, 2-MCPD ester, and glycidol to obtain sample spiked at 0.01 µg/g of ester forms, and 50 µL of a mixed solution of free 3-MCPD and free 2-MCPD to obtain sample spiked at 0.005 µg/g of free forms. The following standards were used: 1,2-bis-palmitoyl-3-chloropropanediol (PP-3-MCPD; Toronto Research Chemicals, B515200), 1,2-bis-palmitoyl-3-chloropropanediol-d5 (PP-3-MCPD-d5; Toronto Research Chemicals, B515202), 3-chloropropane-1,2-diol (3-MCPD; Sigma-Aldrich, 32406), and 3-chloropropane-1,2-diol-d5 (3-MCPD-d5; C/D/N Isotopes, D-1842). Samples were vortexed to mix and then analyzed using GC-MS/MS [AOCS Official Method Cd 29b-13 (2013), modified]. Spike recovery results for free and bound 3-MCPD are presented in Table 6.

OPEN IN VIEWER

Table 6.

Recovery of free and bound 3-MCPD.

Compound	Solid Samples	Liquid Samples
Free 3-MCPD	134%	104%
Bound 3-MCPD ester	100%	114%

Overall, the spike recovery data for PCDDs and PCDFs demonstrated recoveries of 90–115% for the 17 congeners of PCDD and PCDF in the four product types (Table 5). The spike recovery data for free and bound 3-MCPD demonstrated recoveries of 100–134% for free and bound 3-MCPD in solid and liquid samples (Table 6). This demonstrates that the analytical methods used were working well.

Statistics

Mean and standard deviation were calculated for each group (control, sucralose, and maltodextrin) of three samples each. For results reported as less than the method limit, the value was assumed to be the reporting limit for calculation of means. Groups within a product category were compared to each other using one-way ANOVA. P<0.05 was considered statistically significant.

Wafer results

Bound 3-MCPD levels for controls, sucralose, and maltodextrin-containing wafer samples were below the LOD of 0.005 µg/g and LOQ of 0.01 µg/g (Table 8). Free 3-MCPD levels in most of the wafer samples were below the LOQ of 0.005 µg/g, excepting one control sample and one of the sucralose-containing samples. There were no statistically significant differences observed for bound or free MCPD among the wafer samples.

OPEN IN VIEWER

Table 8.

Analytical results for free and bound 3-MCPD from the baking study.

Product type2	Baking sample no.	Bound 3-MCPD esters (expressed as 3-MCPD) (µg/g); LOD 0.005 µg/g1; LOQ 0.01 µg/g	Means (ppm)	SD	P values for one-way ANOVA (statistical significance p<0.05) 2	Free 3-MCPD (µg/g); LOQ 0.005 µg/g1	Means (ppm)	SD	P values for one-way ANOVA (statistical significance p<0.05)
Wafer
Wafer Control	1	<0.005				<0.0050
	2	<0.005	0.005	0		<0.0050	0.0051	0.0002
	3	<0.005				0.0054
Wafer Sucralose	1	<0.005				0.0150
	2	<0.005	0.005	0	1.000	<0.0050	0.0083	0.0058	0.435
	3	<0.005				<0.0050
Wafer maltodextrin	1	<0.005				<0.0050
	2	<0.005	0.005	0		<0.0050	0.0050	0
	3	<0.005				<0.0050
Cake
Cake control	1	<0.005				<0.0050
	2	<0.005	0.005	0		<0.0050	0.0050	0
	3	<0.005				<0.0050
Cake sucralose	1	<0.005				<0.0050
	2	<0.005	0.005	0	1.000	<0.0050	0.0050	0	1.000
	3	<0.005				<0.0050
Cake maltodextrin	1	<0.005				<0.0050
	2	<0.005	0.005	0		<0.0050	0.0050	0
	3	<0.005				<0.0050
Biscuit (Cookie)
Biscuit control	1	>0.005 (LOD) and <0.01 (LOQ)				0.0100
	2	>0.005 (LOD) and <0.01 (LOQ)	0.010	0		0.0055	0.0071	0.0025
	3	>0.005 (LOD) and <0.01 (LOQ)				0.0057
Biscuit sucralose	1	0.011				0.0160
	2	0.011	0.012	0.002	0.216	0.0066	0.0096	0.0056	0.584
	3	0.014				0.0061
Biscuit maltodextrin	1	0.011				0.0062
	2	0.011	0.012	0.002		0.0070	0.0067	0.0004
	3	0.014				0.0068
Pizza sauce3
Sauce control	1	0.021				0.0036
	2	0.016	0.023	0.009		0.0027	0.0040	0.0015
	3	0.033				0.0056
Sauce sucralose	1	0.033				0.0046
	2	0.031	0.029	0.005	0.355	0.0046	0.0048	0.0003	0.411
	3	0.024				0.0051

¹Specific parameters as requested by EFSA.

²Groups within a product category were compared to each other using one-way ANOVA.

³Maltodextrin controls were run only in the case where the retail sucralose-containing low-calorie sweetener product, which contains maltodextrin as a bulking agent, was used as the source of sucralose.

Cake results

Bound 3-MCPD levels for controls, sucralose, and maltodextrin-containing cake samples were below the LOD of 0.005 µg/g and LOQ of 0.01 µg/g. Free 3-MCPD levels in all of the cake samples were below the LOQ of 0.005 µg/g. Hence, there were no statistically significant differences observed among the cake samples.

Biscuit (Cookie) results

Bound 3-MCPD levels for sugar-containing control samples were above the LOD of 0.005 µg/g but below the LOQ of 0.01 µg/g. For the sucralose and maltodextrin-containing samples bound MCPD levels ranged between 0.011 and 0.014 µg/g. Free 3-MCPD levels in all of the cookie samples were above the LOQ of 0.005 µg/g. For sugar-containing controls free MCPD levels ranged between 0.0055 and 0.01 µg/g, between 0.0061 and 0.016 µg/g for sucralose samples, and between 0.0062 and 0.007 µg/g for maltodextrin controls. There were no statistically significant differences observed for bound or free MCPD among the cookie samples.

Pizza sauce results

Bound and free 3-MCPD were detected in all control and sucralose-containing pizza sauce samples. There were no statistically significant differences observed for bound or free 3-MCPD among these samples.

Background

In addition to conducting the analyses requested by EFSA to investigate the potential for sucralose to lead to the formation of non-sucralose chlorinated compounds that could be harmful to human health, it is important to assess critically the basis for the BfR conclusion that sucralose use in foods subjected to heat during their manufacture may result in the formation of non-sucralose chlorinated compounds that could be harmful to health.

This BfR conclusion was first reported in a review ¹⁰ of publications that BfR found through a literature search and assigned a “Klimisch” Category rating. This review was followed by a publication in Food Chemistry, ¹¹ providing similar information.

The Klimisch Category rating system is a published scoring system designed to help with assessing data reliability. ³⁰ The highest Klimisch category rating is 1, to be used when data are considered “reliable with no restrictions.” Next is 2 denoting data considered as “reliable with restrictions.” A 3 or 4 rating is for, respectively, data considered as “not reliable” and “not assignable.”

In forming its conclusions, the BfR determined to consider only publications to which it had assigned a Klimisch Category level of 1 or 2, i.e. publications BfR considered to be “reliable,” with or without “restriction.” The BfR rated no publications found as Category 1 and nine as Category 2, so these nine publications were evaluated in the BfR review. Six publications were found that that BfR assigned a Klimisch Category rating of either 3 or 4, and the BfR did not rely upon these in forming its opinions on the safety of sucralose when used in foods exposed to heat. ⁷ The 3 and 4 Klimisch Category ratings were assigned generally on the basis that BfR found the publication to present insufficient details on methodology and/or findings and/or the publication represented “secondary literature” (i.e., a book chapter).

The below critically analyzes both the studies that were the basis for the BfR opinion on sucralose (BfR assigned Klimisch Category level of 2), and the merits of reaching conclusions regarding sucralose safety without consideration of other information, including publications dismissed by BfR in evaluating available information.

Critical analysis of studies that were the basis for the concern raised by BfR

A critical analysis of these studies shows that none are reliable for assessing the safety of sucralose use in foods that are subjected to heat during manufacturing. They are studies that imposed conditions that would not be found in expected food manufacturing processes and, conversely, could be expected to affect reaction kinetics. It is well known that solute concentrations and environmental conditions can affect reaction kinetics. In many of the studies relied upon by BfR in its assessment of the potential fate of sucralose in food, the concentration of sucralose employed was extremely high, either in the absence of any other ingredient, or at high molar concentration and massively in excess of concentrations that could be expected to be found in food matrices. In addition, the testing environment commonly excluded the presence of water. Previous studies have reported already that neat (dry) sucralose can, with sufficient exposure to heat, eventually lead to chloride ion release by reaction with ambient hydrogen, to form hydrogen chloride, ^31,32 which can further catalyze chloride release. In general, most of the studies cited by the BfR as evidence of concern for sucralose stability in food found this type of chloride release. What is not taken into account by the BfR review, however, is that sucralose is a highly water soluble molecule ³³ that will preferentially partition to the water phase in food matrices, ⁶ which changes the dynamics for chemical reactions. Importantly, an aqueous matrix limits the heat level to which sucralose could be exposed during food manufacturing conditions. Indeed, loss of chloride from sucralose is not seen when sucralose is present at low levels in food. ⁵ The majority of primary reaction sites on the sucralose molecule are also occupied, ⁵ reducing its ability to interact with other substances. This helps to explain its general resistance to chemical interactions and/or breakdown in foods. Scientific publications are available that support this conclusion: Sucralose has been shown to be remarkably stable to chemical interactions with a wide range of types of substances that could be expected to be found in foods, (e.g., bases, oxidizing and reducing agents, aldehydes, ketones and metallic salt) and to pH levels normally found in food, including high acidity foods, such as carbonated soft drinks. ⁵ In a baking study, that investigated the stability of sucralose in a range of baked goods (cookies, cake, graham crackers), no loss of sucralose was detected. ⁷ The latter study used radiolabeled sucralose and extracts of radiolabeled material were analyzed by thin-layer chromatography, using two solvent systems independently as eluents. This showed only material eluting at a time consistent with sucralose standard. Sucralose has also been found to be 100% stable in a range of foods subjected to heat-pasteurization, e.g. ketchup, canned fruits, dairy products and still beverages. ³

What the studies ranked by BfR as Category 2 provide evidence of, is that, consistent with other research, release of free chloride in the presence of fats or fatty acids will result in the formation of such compounds as PCDD/Fs and/or 3-MCP and its congeners. Indeed, it is the availability of chloride ion that is key to the formation of chloropropanols. In a review of 3-MCPD formation from soy sauce production, for example, it was noted that removal of chloride ions in acid hydrolysis would eliminate the formation of 3-MCPD. ³⁴ Also important to note is that the formation of chloropropanols in foods is not unique to foods containing sucralose. They are found in a wide range of foods, including, for example, baked goods (e.g., bread, cookies, crackers), cooked meats (e.g., hamburgers, sausages, bacon), battered fish filets, cheeses, malted beverages (beer and others), roasted coffee, soups, cereals, oils and infant formula, and more ^{14 –18} Their formation occurs naturally in “fat- and salt-containing foods processed at high temperatures,” ¹⁶ via the interaction of free chlorine (chloride), which is readily available from salt, and glycerol (glycerine), which is readily available from fatty acids.

Further, even if sucralose could be a source of free chloride under normal food manufacturing conditions, the thermodynamics of which are exceedingly unsupportive, to protest the use of sucralose based on a potential for it to add chloride to food, is akin to protesting that salt cannot be used in foods. This is particularly true when sucralose use levels in foods are exceedingly low due to its high sweetening potency. On average, sucralose is about 600× sweeter than sugar. ³⁵

Finally, it must be taken into consideration that the extreme and rapid heating applied to sucralose in many of the studies discussed below, if applied in the manufacture of a typical food containing sucralose, would also be likely to induce pyrolysis (burning) of other ingredients, such as the carbohydrate in flour, which thus would seriously impact the likelihood of human exposure due to unacceptable product quality.

The below summarizes the conduct and findings of research studies that were the basis for the BfR questions regarding sucralose use in food, which illustrate the non-applicability of these studies to food safety assessment.

Hutchinson ³⁶ investigated the thermal stability of aqueous sucralose, and other sweeteners, at varying pH and temperatures, which was reported in a doctoral thesis. ³⁶ In this non-peer-reviewed series of studies, the stability of a 0.1 M (3.97%) solution of sucralose (sucralose purity unspecified; 3.97 g in 100 mL water), or of a 0.1 M solution of sucralose and 0.1 M glycine, was evaluated after 1 h at temperatures of 100°C, 140°C, and 180°C at different pH levels (3, 7, and 11). The solutions were placed in stainless steel Whitey reaction cylinders, sealed, and were heated in an oil bath. Heating in such a sealed system would lead to higher temperatures that would not be expected with usual heating of sucralose in an aqueous environment. Such investigative methodology, while useful for thermal stability research, does not represent standard cooking procedures, from either a household or industrial scale perspective. The concentrations of sucralose used by Hutchinson represents about a 2% sucralose concentration, whereas sucralose concentrations in food matrices is far (at least 100×) less. Hutchinson also noted that the inclusion of certain ions in the buffered solutions used in the testing may have interacted with sucralose hydrolysis, thus complicating the interpretation of the effect of temperature and pH on sucralose stability, including findings of chloride loss.

Bannach et al. ³⁷ investigated the effect of rapid heating of sucralose by simultaneous thermogravimetry and differential thermal analysis (TG-DTA). Neat, dry sucralose (two suppliers; one with 98% purity, the other with purity undefined) was applied to the metal sample plate of the thermal analysis system and introduced to a nitrogen-purged furnace sample tube, that was enveloped by a micro-furnace, which increased the heat of the sample tube at a rate of ∼10°C/min up to 500°C. Thermal decomposition of sucralose was noted to occur in three separate phases, the first decomposition change being detected at about 120°C and consistent with loss of chlorine, which was empirically substantiated by detection of hydrogen chloride.

This study is then, effectively, a study of the isolated burning of neat sucralose.

de Oliveira et al., ³⁸ using thermogravimetric apparatus, similarly assessed rapid heating (10°C/min) of neat, dry technical-grade sucralose (two suppliers; purity not specified). The authors reported the fate of sucralose pyrolysis based on differential scanning calorimetry and thermogravimetric analysis coupled with infrared spectroscopy (DSC/TGA/IR), hot-stage microscopy (HSM), and high resolution mass spectrometry (HRMS).

It is worth noting that, in these studies, as with the Bannach et al. ³⁷ studies, mass loss, consistent with chloride loss and HCl formation, was begun to be detected at about 125°C, and images of sucralose evaluated by HSM showed that the dry sucralose crystals underwent a fusion process followed by decomposition. If sucralose in food matrices were to so decompose with food manufacturing processes, including heat, it would be unlikely that it would be suitable for achieving sweetness.

Finally, HRMS analyses of the gaseous phase following heating of sucralose (in a closed vial) to 98°C in a water bath, was reported to reveal the presence of specific polychlorinated aromatic hydrocarbons. The finding of polychlorinated aromatics in the heated headspace at low ppm concentrations seems inconsistent with the relatively mild conditions involved (98°C for 15 min). Formation of these substances is not known to occur, generally, at such a low temperature, ³⁹ and this temperature is lower than that reported to be associated with the thermal decomposition of sucralose. ³¹ This could be the result of contamination or some other source of chlorine other than sucralose. The use of a closed vial when heating the samples may also have caused pressure changes that may have affected internal temperature and/or chemical reaction potential, and as such are not clear as having relevance to sucralose stability in food preparation.

Rahn and Yaylayan ³¹ investigated sucralose thermal decomposition and the potential of this to lead to formation of other chlorinated substances, in particular, chloropropanols. In these investigations, sucralose alone or in a 1:2 M ratio with glycerine was inserted into a pyrolysis (250°C-heated) chamber for 20 s. Rahn and Yaylayan also report pyrolysis of sucrose (common table sugar) under similar conditions. With pyrolysis, there can be oxidation of other elements, like sulfur and chlorine. ^{40 –42} The investigators showed that the pyrolysis of pure sucralose resulted in a degradation pathway with levoglucosenone as a major product, which is also a product found with the pyrolysis of sucrose and other saccharide molecules such as β-d-glucopyranose, cellobiose and cellulose. ^43,44 The formation of levoglucosenone is not surprising as sucralose is made from sucrose with selective chlorination at three of the primary hydroxyl groups, and release of chloride through oxidation would enable this type of degradation product. Levoglucosenone formation is preceded by, in the case of both sucrose and sucralose, cleavage at the glycosidic linkage that connects the inherent monosaccharides or, in the case of sucralose, chlorinated monosaccharide moieties. Indeed, Rahn and Yaylayan noted that “being [a] sucrose analog, one anticipates [sucralose] degrading in a similar fashion.” The investigators also showed that subsequent degradation of the two monosaccharide derivatives from sucralose is related to chlorine loss. The finding of chlorine loss under these particular experimental conditions is again consistent with the previously discussed thermal degradation studies.

The pyrolysis studies of a 1:2 molar mixture of sucralose and glycerine, whether in the presence of air or moisture, also showed some formation (in addition to formation of the main degradation product, levoglucosenone) of 3-monochloropropanediol (3-MPCD) and 1,2- and 1,3-dichloropropanols (1,2- and 1,3-DCP, respectively), based on chromatographic analyses. The findings are consistent with those of Rahn and Yaylayan, ³¹ which exposed a 1:2 molar mixture of sucralose and glycerol to extreme and rapidly-applied heat, would yield chloropropanols. While chloropropanol formation resulted, as discussed above, this can be expected when conditions force free chloride release in the presence of glycerol, but such molar concentrations and experimental conditions are not present in the production of food products.

Wu et al. ⁴⁵ investigated the potential for formation of various polychlorinated dibenzo-p-dioxin and dibenzofuran congeners (PCDD/Fs) by heating sucralose or 1,3-dichloropropan-2-ol (1,3-DCP) combined with oil (soybean oil) or oil + minced beef (∼1/3 ratio, w/w). The sucralose and 1,3-DCP concentration in the heated oil was 22% and 19%, respectively. The sucralose and 1,3-DCP w/w concentrations in the beef/oil mixtures ranged from ∼0.005 to 5% and ∼0.00005 to 5%, respectively. The oil mixtures were heated to ∼250°C and maintained at this temperature for another 10 min. The beef/oil +/- sucralose or 1,3-DCP mixtures were heated using the same apparatus used for heating the oil, and after 15 min the mixture temperatures were ∼145°C. All control and text mixtures were then allowed to cool for 10 min. The temperature of all the beef/oil mixtures rose slightly, to about 165°C, by the end of this period. The air above the heating pans (head space) was also trapped during the entire heating and cooling phases for subsequent analysis.

Analysis of the fluid phases (oil, oil mixtures or residual fluids after heating beef/oil +/- sucralose or 1,3-DCP) after heating showed that the concentrations of PCDD/Fs of the oils with or without sucralose or 1,3-DCP, as well as the beef/oil mixtures, were all “rather low, and most of the target PCDD/F congeners were not detected.” Analysis of the head space samples and solid phases from the beef-containing mixtures showed the presence of various PCDD/F congeners, however, the contribution was predominantly from the gas, vs. the solid phase, i.e. predominantly not in the food matrix. Thus, this study does not represent conditions of food manufacture, where sucralose would be dissolved in the water phase of a food matrix. Further, as noted by the investigators, heating the oil used in their study could produce hydrocarbons. The PCDD/F levels found with inclusion of either sucralose or 1,3-DCP, both of which have been shown to be able to be dechlorinated with high heat and in low/zero-moisture conditions, ^31,46 likely stems from free chloride interacting with the generated hydrocarbons. As noted previously, however, any possible chloride release would not be likely to meaningfully increase exposure to PCDD/Fs beyond what would already be possible with high heat application to foods containing salt and fat.

Dong et al. ⁴⁷ conducted studies that were very similar to those conducted by Wu ⁴⁵ and, indeed, were done in the same lab. These studies investigated the potential for formation of particular polychlorinated biphenyls (PCBs) when sucralose or 1,3-DCP was added to either minced beef or minced beef in soybean oil (3:1 w/w ratio). The w/w concentration of sucralose or DCP in the beef/oil mixtures was ∼0.5 – 5% and 0.005 – 5%, respectively. The mixtures were added to a stainless steel bowl kept at 400°C and heated for 15 min, and then allowed to cool for 10 min. The temperatures of the raw minced beef or minced beef mixtures was ∼160°C at the end of the heating time.

The research found that the concentrations of PCBs in control beef declined by more than 50% after cooking with or without the presence of oil. This also occurred when either sucralose or 1,3-DCP was added to the beef/oil mixtures, except at the highest 1,3-DCP concentration, which was still a PCB concentration not greater than the level found in uncooked beef. As in the Wu ⁴⁵ study, PCB congener levels were increased in the oil fumes produced after cooking. The authors said that results of their study suggest that sucralose or 1,3-DCP could act as chlorine sources, and so promote formation of PCBs in oil fumes. While it is unknown whether the formation of PCBs by reaction of chlorine with hydrocarbons occurs within the beef, the decline in PCBs in the cooked beef is evidence that any such putatively formed PCBs are subsequently volatilized and, thus, do not add to the innate meat PCB content. Nonetheless, it is more probable that the heating of the oil, while either sucralose or 1,3-DCP was present, led to free chloride, by dechlorination, and free chloride then combined with hydrocarbons from the oil, to result in PCB congeners that were then volatilized to the head space. It is also noteworthy that, compared with the total PCB content of raw beef (∼75 pg/g), the content of PCBs in oil fumes captured after cooking the beef/oil/sucralose mixtures was comparatively low (<3 pg/g). In addition, while the investigators report an increasing concentration of PCBs with increasing sucralose concentration, the maximum increase was within one order of magnitude, while the range of sucralose concentrations represented a 1000-fold increase.

Dong et al. ⁴⁸ later investigated the effect of heating oil (peanut or olive) with or without either sucralose or sodium chloride, using the same type of cooking apparatus and general methodology, and analyzing for the presence of polychlorinated napthalenes (PCNs). The authors noted that due to the fact that sucralose is hydrophilic, the sucralose/oil mixtures were mixed with a glass rod to affect a sucralose suspension as homogeneous as possible. The w/w concentration of sucralose or sodium chloride in the oils was 10%. As noted by the investigators, this is an unrealistic use of sucralose in food preparation. The mixtures were heated to ∼ 200 or 245°C. The investigators also reported heating neat sucralose under the same conditions as the control or test oil mixtures and similarly analyzed for the presence of PCNs.

No PCNs were detected, in heated neat sucralose, or any oil mixture or collected head space (off-gassed fumes following heating) when heated to ∼200°C. At 245°C, no PCNs were found in neat sucralose or the oil mixtures containing either salt or sucralose. PCNs were detected in the off-gassed oil fumes when the sucralose/oil mixture was heated to ∼245°C, but not with heating either neat sucralose or oil mixtures with sodium chloride. The investigators note that this suggests that the PCNs detected are a consequence of free chloride, released from sucralose as a result of the cooking in oil, interacting with carbon sources from the cooked oil, at this temperature. They further note that the PCNs detected could be the result of [higher] temperatures achieved in the oil mixtures in the field close to the thermal heating plate, leading to PCNs being evaporated into the oil fumes. As with the previous studies from this lab, the experimental conditions utilized do not represent conditions that would be encountered in normal food production.

Dong et al. ³⁹ reported a further investigation generally using the methodology described by Wu, ⁴⁵ where neat sucralose (5 g) was heated to 350°C or 400°C in utensils made of different materials and in the presence or absence of particular “rusts,” including Al₂O₃, Fe₂O₃, and CuO, and analyzed for the presence of PCDD/Fs. Heating sucralose up to 400°C in “cooking utensils” made of quartz, aluminium, or copper, and polished to be free of rust, failed to generate detectable amounts of PCDD/Fs. Whereas, when sucralose was heated to 350 or 400°C in a stainless steel vessel, PCDD/Fs were detected primarily in the resulting smoke. It should be noted that other research has shown that certain metal ions, and particularly Fe³⁺ and Fe²⁺, enhance the formation of 3-MCPD when vegetable oil is heated. ⁴⁹ With heating to 400°C and addition of 1 g of either Al₂O₃, Fe₂O₃, or CuO, there were increases in the levels of PCDD/Fs detected in the smoke. No significant amounts of PCDD/Fs were detected when sucralose was heated in the presence of the metal oxides at 300°C. In general, these studies are effectively pyrolysis studies, which do not represent conditions during normal food manufacture. The addition of pure rusts at high w/w concentration, which could act as catalysts for other reactions in the thermal degradation of sucralose, is also unrealistic for food manufacturing conditions. Further, the amount of dry sucralose that was heated has a sweetness equivalent to ∼3000 g sugar – hence, a concentration that has no relevance to normal use of sucralose in foods.

Dong et al. ⁵⁰ conducted one additional study that had a design and methodology very similar to the previously described study, ³⁹ this time looking for the presence of PCNs. Dry, neat sucralose (5 g) was again heated by itself to high temperatures (350 and 400°C) and PCNs were found primarily in the resulting smoke. The patterns of PCNs detected were found to be “consistent with the patterns of PCNs unintentionally produced from waste incineration and metal smelting.” The addition of rusts in the heating process demonstrated, effectively, that heated metal oxides can catalyze PCN formation processes when dry sucralose is heated at such high temperatures. The authors also noted that CuO, the added oxide resulting in the largest PCN increase, can react with HCl to produce Cl₂, and “Cu (II) would be expected to promote formation of biphenyl structures, while Cl₂ can, of course, promote chlorination of aromatic compounds.” The investigators also detected no polycyclic aromatic hydrocarbons (PAHs) in sucralose prior to heating, although they were found in the smoke after heating sucralose at the high temperatures utilized. This indicates that the heating induced effective burning of sucralose, such that aromatic hydrocarbons were made available from the breaking of the disaccharide basis of sucralose, and, along with release of free chloride. The released free chloride and available hydrocarbons can then lead to generation of PCNs.

Consideration of the scope of research, and application of Klimisch Category ratings, in the BfR reviews ^10,11

The Klimisch category ratings were originally created for evaluating the quality of experimental toxicological and ecotoxicological data and in comparison to already established standards for the conduct of such studies ³⁰ relating to preferably standardized methodology. ⁵¹ While its use has recently been extended to other types of studies, the application of this rating system to exploratory chemistry studies on ingredient stability under variable conditions is questionable. Importantly, use of the ratings infers that all studies conducted should be available, in full detail, in the published literature, and those that don’t have full details should not have weight in scientific consideration of stability evidence. There is, however, no regulatory requirement for publication of stability studies of food ingredients and there is also no historical precedent set by food ingredient research publications that should lead to an expectation for such detailed reports. In general, very detailed stability studies of any food ingredient are commonly not found in the published literature. In addition, Klimisch Category ratings of 3 and 4 do not have to be excluded in a weight of evidence approach, as they may still convey important information. ⁵² Despite the BfR ratings applied to publications found, appropriate stability studies are required by regulatory agencies in the assessment of any proposed new food ingredient ^2,53 and, in the case of sucralose, key studies have been described in various publications, ^{7,22,54 –57} including, but not limited to, ones found by BfR with their literature search. ^5,58 A summary of these studies is shown in Table 9.

OPEN IN VIEWER

Table 9.

Types of studies conducted to evaluate sucralose stability, in addition to those reviewed by BfR, ¹⁰ which are discussed in published scientific literature.* ^5,54,58

Type of study	Conditions
Stability in model aqueous systems and prototype beverages (studies relate also to foods, as sucralose will partition to the water phase in food formulations)	–Variable sucralose concentrations –Variable pH –Ambient temp –Elevated heat –Variable storage times –Some analyses include use of either 36Cl- or 14C radiolabeled sucralose
Stability to food manufacturing processes – General food processing – Pasteurization – Sterilization – Ultra-High Temperature (UHT) Processing	– Sucralose concentrations to achieve normal sweetness – Variable storage times – Range of prepared foods as applicable, e.g.:  – Beverage  – Condiment  – Canned Fruit  – Beans in sauce  – Dairy desert  – Vanilla milk
Stability in baked goods	–Sucralose concentrations to achieve normal sweetness – Range of baked goods:  – Cake  – Cookie  – Graham crackers – Use of radiolabeled sucralose
Photochemical stability in model soft drink system	Constant darkness or fluorescent daylight at ambient temp and with 9 months storage

* This is not intended to be a comprehensive list of studies that were reviewed by regulatory agencies, which can include studies beyond those reported in the published literature.

While the publications that provide this information were largely publications that summarize sucralose stability research (the primary reason BfR assigned a Klimisch Category rating of 3 or 4 to studies it found in its literature search), they provide a good picture of the types of research that was submitted to regulatory agencies for their full and complete assessment.

These publications also provide information on other research, some of which would have been outside the scope of the BfR literature search, that helps food scientists to better comprehend how sucralose might behave in food systems. This includes, for example, effects of concentration, stability in dry or low-moisture conditions, food-chemical interaction studies, and physicochemical characteristics of sucralose. ^5,6,54,58

Notably, this research makes clear that sucralose is freely soluble in water (up to 28.2 g/100 ml at 20°C) and has poor lipid solubility (insoluble in corn oil <0.1g/100 g at 20°C), ³³ and will always partition with the aqueous portion of a food system comprising both lipid and aqueous phases. ⁶ Because the presence of surrounding water limits the level of heat in the food matrix, this information must be taken into account when considering the meaningfulness of certain types of research studies, e.g. the studies cited by BfR in raising their concern about the safety of foods made with sucralose.