Ingestion of the non-nutritive sweetener erythritol, but not glucose, enhances platelet reactivity and thrombosis potential in healthy volunteers

Abstract

Background:

Although artificial and non-nutritive sweeteners are widely used and “generally recognized as safe” (GRAS) by US and EU regulatory agencies, there have been no clinical trials to assess either long-term cardiovascular disease (CVD) risks or short-term CVD-relevant phenotypes. Recent studies report fasting plasma levels of erythritol, a commonly used sweetener, are clinically associated with heightened incident CVD risks and enhance thrombosis potential in vitro and in animal models. Effects of dietary erythritol on thrombosis phenotypes in humans have not been examined.

Methods:

Using a prospective interventional study design (clinicaltrials.gov:NCT04731363), we tested the impact of erythritol or glucose consumption on multiple indices of stimulus-dependent platelet responsiveness in healthy volunteers (n=10 per group). Erythritol plasma levels were quantified with LC-MS/MS. Platelet function at baseline and following erythritol or glucose ingestion was assessed via both aggregometry and analysis of granule markers released.

Results:

Dietary erythritol (30g), but not glucose (30g), lead to a >1000-fold increase in erythritol plasma concentration (6480[5930–7300]μM versus 3.75[3.35–3.87]μM, P<0.0001) and exhibited acute enhancement of stimulus-dependent aggregation responses in all subjects, agonists and doses examined. Erythritol ingestion also enhanced stimulus-dependent release of the platelet dense granule marker serotonin (P<0.0001 for TRAP6; P=0.004 for ADP) and the platelet alpha granule marker CXCL4 (C-X-C motif ligand-4) (P<0.0001 for TRAP6; P=0.06 for ADP). In contrast, glucose ingestion triggered no significant increases in stimulus-dependent release of either serotonin or CXCL4.

Conclusion:

Ingestion of a typical quantity of the non-nutritive sweetener erythritol, but not glucose, enhances platelet reactivity in healthy volunteers, raising concerns that erythritol consumption may enhance thrombosis potential. Combined with recent large-scale clinical observational studies and mechanistic cell-based and animal model studies, the present findings suggest that discussion of whether erythritol should be re-evaluated as a food additive with GRAS designation is warranted.

Graphical Abstract

Introduction

Artificial sweeteners (including non-nutritive sweeteners) are widely used and generally recognized as safe (GRAS); however, a growing number of epidemiological studies have associated their use with cardiovascular disease (CVD) risk.¹ Epidemiological studies are limited by potential unmodeled confounding, including reverse causation. Further, due to current limitations in food labelling requirements, epidemiology studies generally do not quantify individual sweeteners, but instead use food questionnaires to estimate consumption in broad categories (e.g. “natural sweeteners”, or “sugar alcohols”). While numerous health bodies (e.g. American Diabetes Association (ADA), American Heart Association (AHA), and European Food Safety Authority (EFSA)) recommend use of sugar substitutes over sugar to patients most at risk for thrombotic events (e.g. those with diabetes, obesity, and metabolic syndrome),² human intervention studies to directly assess adverse effects of sweeteners are limited.

Using untargeted metabolomics as a discovery platform, we recently reported fasting plasma levels of erythritol were associated with incident major adverse cardiovascular event (MACE=myocardial infarction, stroke or death) risks independent of traditional cardiovascular disease (CVD) risk factors. In subsequent validation studies using a quantitative and specific mass spectrometry assay, we confirmed that circulating fasting plasma erythritol levels are clinically associated with incident (3 year) MACE risks in both US and EU cohorts.³ We also showed elevated circulating erythritol levels enhanced thrombosis potential in animal models, and elicited pro-thrombotic phenotypic changes to human platelets in multiple in vitro studies.³ Moreover, consumption of erythritol led to prolonged (days) heightened plasma levels in healthy volunteers, further raising safety concerns.³ While our previous studies suggest an association of erythritol blood levels with a pro-thrombotic state, proof-of-concept interventional studies to confirm direct effects of erythritol consumption on thrombosis phenotypes in humans have not yet been performed. Here, using a prospective interventional study design with a relevant dietary exposure in healthy volunteers, we assessed the potential pro-thrombotic effects of dietary erythritol (E 968), one of the fastest growing Food and Drug Administration (FDA) and EFSA approved non-nutritive sweetener,⁴ versus an equivalent amount of glucose.

Methods

The human study (clinicaltrials.gov:NCT04731363) abided by the Declaration of Helsinki and all subjects provided written informed consent. The Institutional Review Board of the Cleveland Clinic approved the erythritol intervention study (IRB 21–005). Plasma levels of erythritol were assessed using stable isotope dilution liquid-chromatography tandem mass spectrometry (LC-MS/MS) using methods specifically designed to distinguish erythritol from its structural isomers, as previously described.³ Aggregometry studies in platelet-rich plasma (PRP) were performed as described.⁵ PRP was isolated before and immediately (30 minutes) after ingestion of erythritol or glucose from blood anticoagulated with sodium citrate (0.109 M). Platelet aggregation was induced by adding the agonists adenosine diphosphate (ADP; up to 5 μM, Catalogue # 384, Chronolog, Havertown, PA, US) or thrombin activator peptide 6 (TRAP6) (TFLLR-NH2, up to 10 μM, Catalogue # 464, Tocris, Bristol, UK) as indicated. For all platelet aggregometry experiments, blood was always processed within approximately 30 minutes of collection, and isolated platelets were used within 120 minutes of isolation. Serotonin levels were quantified by LC-MS/MS, and CXCL4 concentrations were analyzed via ELISA (R&D Systems, MN, USA).³

Results

This single center trial was approved by the Cleveland Clinic Institutional Review Board (clinicaltrials.gov identifier NCT04731363). Following subject consent (nonsmokers without CVD, hypertension, or diabetes, with normal renal function, no history of recent (1 month) anti-platelet medication, and no clinical history of bleeding, bruising or documented bleeding disorder, Table S1), blood was drawn after overnight fast and 30 min following consumption of water mixed with 30g of either glucose (n=10 subjects, 30.1±11 years of age, 40% male), or erythritol (n=10 subjects, 30.5±8 years of age, 50% male), a quantity commonly found in erythritol-sweetened foods and the daily intake of some subjects based on 2013–2014 National Health and Nutrition Examination Survey (NHANES) data and FDA filings.^4,6

Following erythritol consumption, postprandial circulating erythritol levels were increased >1000-fold compared to baseline levels (median [inter quartile range], 6480 [5930–7300] μM versus 3.8 [3.4–3.9] μM, P<0.0001). In contrast, circulating erythritol levels remained similar before versus after glucose consumption (3.0 [2.6–4.0] μM versus 2.9 [2.7–3.8] μM, P=0.87), while glucose levels were modestly increased (87 [82–93] mg/dL versus 127 [122–132] mg/dL, P=0.002). A striking increase in platelet aggregation responses to multiple sub-maximal levels of both ADP and TRAP6 was observed following erythritol ingestion (Figure 1). Every subject showed the same effect (increased responsiveness with erythritol consumption, Supplemental Figure 1). In contrast, glucose consumption had no effect on platelet aggregation (Figure 1, Supplemental Figure 2), in line with previous in vitro studies showing that much higher glucose concentrations (>400 mg/dL) are needed to enhance agonist-induced platelet activation.⁷ The erythritol-dependent increase in platelet responsiveness showed significant correlation among all subjects (erythritol levels versus agonist-induced aggregation; Spearman rho 0.65 and 0.68, P<0.0001 each, for ADP and TRAP6, respectively). Erythritol ingestion also markedly enhanced stimulus-dependent release of both the dense granule marker serotonin (P<0.0001 for TRAP6; P=0.004 for ADP), and the alpha granule marker CXCL4 (P<0.0001 for TRAP6; P=0.064 for ADP; Figure 2). In contrast, no significant increases were observed in stimulus (ADP, TRAP6) dependent release of either serotonin or CXCL4 following glucose ingestion (Figure 2).

Figure 1. Ingestion of erythritol, but not glucose, enhances platelet responsiveness to multiple agonists in healthy volunteers.

(A, B) Platelet aggregation in response to the indicated concentrations of ADP (top) or TRAP6 (bottom) at baseline (blue) and 30 min post glucose (orange) or erythritol ingestion (red). Multiple replicates of paired samples (connected by lines) from each subject are shown. Boxes represent interquartile range (IQR) with median (thicker line within box). Lower whiskers represent smallest observations (≥25% quantile-1.5×IQR), and upper whiskers represent largest observations (≤75% quantile+1.5×IQR). The total number of paired (baseline/postprandial) replicates from subjects for challenges (glucose or erythritol) are shown. To evaluate difference across groups (erythritol versus glucose, and before versus after consumption), a 2-factor non-parametric p value was calculated using Friedman test. When it was significant, p values for pairwise comparisons were performed with Wilcoxon signed rank test.

Figure 2. Ingestion of erythritol, but not glucose, increases agonist-induced platelet alpha and dense granule release.

(A, B) Erythritol ingestion enhances platelet stimulus-dependent release of both alpha granule (CXCL4) and dense granule (serotonin) products in response to sub-maximal levels of agonists (ADP(2 μM), panel A; and TRAP6 (7.5 μM), panel B) in healthy volunteers. Platelet release of CXCL4 (ELISA) and serotonin (LC-MS/MS) in response to the indicated concentration of ADP or TRAP6 at baseline (blue) and 30 min post ingestion of 30 g of either erythritol (red) or glucose (orange). The total number of paired (baseline/postprandial) replicates from the indicated number of subjects for both challenges are shown. Bars represent median levels in each group. To evaluate difference across groups (erythritol versus glucose, and before versus after consumption), a 2-factor non-parametric p value was calculated using Friedman test. When it was significant, p values for pairwise comparisons were performed with Wilcoxon signed rank test.

Discussion

Upon re-evaluation of erythritol (E968) as a food additive, the EFSA recently noted that further studies are needed to test the direct effects of erythritol exposure in humans.⁸ The present studies show that a standard serving of erythritol enhances platelet responsiveness in healthy subjects with normal renal function. Moreover, erythritol consumption led to statistically significant enhancement in multiple indices of stimulus-dependent platelet responsiveness, including ADP- and TRAP6-induced aggregation (at multiple submaximal doses examined), as well as alpha and dense granule release, in every subject examined. When coupled with recently reported large-scale clinical studies and mechanistic animal model studies linking erythritol to CVD risks,³ the present findings raise concerns that erythritol consumption in humans may provoke a direct pro-thrombotic effect. They also suggest re-evaluation of the safety of erythritol as a food additive with GRAS designation should be considered.

Our present studies have several limitations. First, while the effects of erythritol ingestion were uniformly observed in every subject enrolled and were statistically significant, the study size was small and replication of results with different cohorts is needed. Further, we did not test long-term changes in platelet function following erythritol consumption. This was because in previous pharmacokinetics data it was observed that erythritol is rapidly absorbed, with circulating erythritol levels being increased within 15 minutes of ingestion. Further, both in vitro and ex vivo (whole blood) studies showed enhanced platelet responsiveness elicited with erythritol within 15 minutes of exposure³. Given that the life span of a platelet is only 5–7 days, we focused our studies on examining the impact of acute dietary exposure. Our protocol had two advantages – it both minimized inconvenience to volunteers by enabling collection of both (baseline and post-prandial) samples at the same visit/day, and it reduced variation in intake for each participant to be able to compare changes in platelet responsiveness within a given subject before vs after exposure, and also across comparable exposures between subjects. The chronic effects of erythritol intake warrant further investigation. As with any study, another limitation is the possibility for residual confounding by unknown factors. Use of a glucose control arm with comparable study design (baseline versus 30 min post prandial blood collection) that showed no enhancement of platelet responsiveness in subjects with any functional measure indicated that the findings are specific to erythritol.

While intervention studies with sweeteners monitoring thrombosis relevant phenotypes have not yet been reported, several studies have tested whether alternative dietary interventions modulate platelet function (i.e. adhesion, activation and aggregation), mainly with the goal of improving primary and secondary preventive efforts. Among a large number of studies, most evidence for platelet-inhibitory effects comes from studies involving supplementation with polyunsaturated fatty acids⁹ and plant polyphenols¹⁰. Other studies tested food constituents, such as olive oil ^11,12, garlic ¹³, and tomato extract ¹⁴, with the latter found to induce approximately 30% platelet inhibition compared to aspirin. Despite their ever-growing use, the limited studies on sugar substitutes have primarily focused on glucose control and weight reduction. The present studies showing heightened platelet responsiveness following erythritol ingestion are consistent with recent clinical associations observed between erythritol levels and incident major adverse cardiac event risks in observational cohort studies and in mechanistic cell and animal model studies.³ They also are consistent with similar clinical and mechanistic findings recently reported with the sugar alcohol xylitol ¹⁵. Further studies exploring the cardiovascular safety of sugar alcohols like erythritol as a sugar substitute, particularly among subjects at heightened thrombosis risk, seem prudent.

Highlights.

• Human intervention studies have not yet tested for pro-thrombotic effects of ingesting the artificial sweetener erythritol.

• A typical dietary exposure to erythritol enhanced multiple indices of stimulus-dependent platelet activation and aggregation in healthy subjects.

• Comparable glucose intake did not increase platelet responsiveness in study participants.

• Our studies show ingestion of the non-nutritive sweetener erythritol, but not glucose, enhances platelet reactivity and thrombosis potential in subjects.

Data and code availability

All data and materials have been made publicly available at the public data sharing repository Zenodo and can be accessed at [http://doi.org/10.5281/zenodo.10594509]