1 Introduction
A recent global analysis estimates that approximately 880 million adults and 159 million children worldwide are currently obese [
1], and over the past three decades the global number of people living with diabetes has more than quadrupled to 828 million [
2]. This has led to a more aggressive effort to reduce sugar in the diet and to look for alternatives that preserve sweetness and palatability while reducing calories. Against this background, artificial sweeteners have been widely incorporated into foods and beverages because they provide intense sweetness with little or no caloric contribution and generally exert limited immediate effects on blood glucose levels [
3]. Their expanding use has therefore been framed not only as a technological solution for food formulation, but also as a public-health strategy for weight management and glycaemic control.
Despite their widespread use, artificial sweeteners are not metabolically neutral substitutes for sugar. The debate over their potential health risks has persisted since the discovery of saccharin in 1879 [
4]. This debate intensified after the 2023 International Agency for Research on Cancer (IARC) classification of aspartame as possibly carcinogenic to humans [
5], which renewed attention to the broader toxicological and metabolic consequences of artificial sweetener use. Increasing evidence now implicates these compounds in altered insulin sensitivity, dysregulated energy balance, microbiota perturbation, and disease-related processes relevant to diabetes, inflammatory bowel disease (IBD), cancer, and cardiovascular pathology [
6,
7].
The central challenge is therefore no longer whether artificial sweeteners can replace sugar in a narrow caloric sense, but under what conditions such replacement confers benefit, neutrality, or unintended risk. This review addresses that question by integrating historical development, compound-specific properties, regulatory context, mechanistic evidence, and human observational data. Rather than listing individual studies in isolation, we evaluate where findings converge, where they diverge, and why differences in dose, exposure duration, developmental timing, model system, and host susceptibility matter for interpretation.
Two related but separate questions are often combined in this literature and should be kept apart. The first is the exposure effect, meaning the biological and metabolic consequences of consuming artificial sweeteners compared with consuming none. The second is the substitution effect, meaning whether replacing sugar with artificial sweeteners improves or worsens health compared with continued sugar intake. These two questions rely on different comparison groups and therefore support different causal interpretations. Most mechanistic studies, animal experiments, and observational cohorts that compare consumers with non-consumers mainly inform the exposure question rather than the substitution question. Formal substitution analyses and randomized trials that use sugar as the comparator are still comparatively scarce. For this reason, the present review focuses on the health effects and mechanisms associated with artificial sweetener exposure, and it does not attempt a formal benefit and risk assessment of sugar substitution. Where the evidence allows, we indicate whether a given finding reflects a comparison with non-consumption or a comparison with sugar intake, so that correlation and substitution are not treated as the same thing.
2 Classification of Artificial Sweeteners, Regulatory Frameworks, and the Origins of Safety Controversy
Artificial sweeteners are a chemically heterogeneous group of high-intensity sweetening agents whose shared sensory function masks substantial differences in molecular structure, metabolism, and biological fate. Since the introduction of saccharin in 1879, these compounds have been adopted across foods, beverages, and pharmaceutical formulations because they deliver sweetness at very low use levels [
4,
8]. However, assessing the safety of artificial sweeteners is not straightforward. Although they share the common feature of providing sweetness with little or no caloric contribution, individual compounds differ markedly in potency, absorption and metabolism, gastrointestinal exposure, and acceptable intake thresholds. These differences complicate attempts to evaluate artificial sweeteners as a single class and underscore the need for compound-specific interpretation. It should be noted that this review focuses on fully synthetic artificial sweeteners and therefore excludes sugar alcohols such as erythritol, xylitol, and sorbitol. These compounds are primarily derived from natural or fermentation processes and exhibit distinct absorption and metabolic profiles compared with classical artificial sweeteners. For example, erythritol is largely absorbed and excreted unchanged, with minimal systemic metabolism. To ensure conceptual consistency and comparability of mechanistic analyses, sugar alcohols were not included in this review, despite emerging evidence linking erythritol to cardiometabolic outcomes. The basic information about artificial sweeteners is presented in Table 1.
2.1 Saccharin and saccharin sodium
Saccharin and sodium saccharin exemplify how regulatory controversy can persist long after an initial toxicological signal. Both compounds are approximately 300–500 times sweeter than sucrose [
9,
10] (Table 1), but their historical significance lies less in their physicochemical properties than in the way rodent carcinogenicity data shaped public and regulatory perceptions. Early high-dose studies linked saccharin exposure to bladder tumors in rats, prompting restrictions in several jurisdictions [
11]. Subsequent mechanistic work, however, indicated that the effect depended on rat-specific urinary conditions and was not readily generalizable to humans [
12]. The resulting divergence between early hazard signals and later mechanistic reinterpretation remains instructive for the entire sweetener field. It also helps explain why acceptable daily intake (ADI) values for saccharin differ across agencies, reflecting not only toxicological evaluation but also different regulatory approaches to uncertainty (Table 1).
2.2 Aspartame, neotame, advantame, and alitame
Aspartame occupies a distinctive position within the artificial sweetener literature because, unlike many other compounds in this category, it is extensively metabolized after ingestion. Approximately 200 times sweeter than sucrose [
13] (Table 1), aspartame is a methyl ester of L-phenylalanine and L-aspartic acid [
14]. Its long-standing controversy stems from this metabolic profile: hydrolysis yields phenylalanine, aspartic acid, and methanol, each of which has been discussed in relation to neurological or carcinogenic risk when exposure is excessive or biologically vulnerable populations are considered. For individuals with phenylketonuria, this concern is clinically established because phenylalanine metabolism is impaired [
15,
16]. More broadly, aspartame illustrates how toxicological evaluation must account not only for the parent compound but also for downstream metabolites, exposure level, and susceptible subgroups.
The development of neotame, advantame, and alitame reflects an industrial effort to retain the sensory advantages of aspartame-like chemistry while improving potency, stability, and in some cases metabolic constraints. Neotame and advantame are substantially sweeter than sucrose and are used at extremely low concentrations, which may reduce mass exposure despite their high sweetening intensity [
17,
18] (Table 1). In addition, neotame and alitame do not pose the same phenylalanine-related concern as aspartame, which is relevant for people with phenylketonuria [
19,
20]. However, lower historical controversy should not be conflated with established long-term safety. For these second-generation compounds, the evidence base remains comparatively sparse, and emerging studies already suggest possible effects on gut ecology and epithelial homeostasis, indicating that limited scrutiny should not be interpreted as evidence of safety [
21-
23].
2.3 Sucralose, acesulfame potassium, and cyclamate
Sucralose, acesulfame potassium (acesulfame K), and cyclamate further illustrate why class-level generalizations are inadequate. Sucralose is a chlorinated sucrose derivative that is largely resistant to digestion and absorption, thereby increasing the likelihood of distal intestinal exposure [
24-
27] (Table 1). Acesulfame K is also highly potent and widely used, but differs structurally and toxicokinetically from both sucralose and dipeptide sweeteners [
28]. Cyclamate, by contrast, is less intensely sweet yet historically important because its regulatory status remains geographically inconsistent: it continues to be used in many countries, whereas its use remains banned in the United States following earlier rodent carcinogenicity [
29-
31] (Table 1). These contrasts matter because disease associations may arise not from sweetness per se, but from differences in absorption, metabolism, impurities, gut exposure, and context of use.
The health risks and pathogenic mechanisms of common artificial sweeteners are shown in Fig. 1 and Table 2.
3 Counterintuitive: Long-Term Intake of Artificial Sweeteners May Increase the Risk of Insulin Resistance
Artificial sweeteners are frequently promoted as tools for glycaemic management because they preserve sweet taste without delivering the caloric load of sucrose. Yet the central question is not whether they reduce acute glucose exposure relative to sugar, but whether repeated consumption alters the regulatory systems that govern appetite, insulin secretion, and metabolic homeostasis. Evidence from epidemiology, animal models, and mechanistic studies increasingly suggests that chronic exposure may not be metabolically neutral [
32-
34].
Experimental studies provide biologically plausible mechanisms for impaired glycaemic control, but their interpretation depends heavily on exposure design. In mice, sucralose combined with a high-fat diet aggravated insulin resistance and glucose intolerance and was linked to altered hepatic insulin signaling through a taste receptor type 1 member 3 (T1R3)–extracellular signal-regulated kinase 1/2 (ERK1/2)-related pathway [
35]. This model is informative because it captures an exposure context in which non-caloric sweeteners are consumed within energy-dense dietary patterns rather than in isolation. At the same time, such studies do not demonstrate that sucralose alone is sufficient to induce the same phenotype under all nutritional conditions. Instead, they suggest that artificial sweeteners may interact with obesogenic diets to amplify metabolic dysfunction, a distinction that is critical for both mechanistic interpretation and risk communication.
Observational research extends these concerns to population settings, although causal inference remains more limited. Long-term cohort studies have linked frequent intake of artificially sweetened beverages (ASBs) to greater weight gain and to a higher incidence of type 2 diabetes relative to non-consumption [
36,
37]. These findings suggest that although ASBs are generally considered preferable to sugar-sweetened beverages, they are not free of health concerns, and long-term consumption may still contribute to metabolic disturbances (Fig. 2A). One reason why evidence from animal studies remains difficult to translate directly to humans is that relatively few studies administer artificial sweeteners in the form of sweetener-containing foods or beverages. Instead, most experimental designs rely on purified sweetener standards, whereas real-world human exposure usually occurs within complex food matrices, where multiple sweeteners and other dietary components may coexist and potentially generate additive or interactive effects.
More recent cohort data reinforce the need for compound-specific evaluation. In the prospective NutriNet-Santé study, higher intake of aspartame, acesulfame K, and sucralose was associated with a greater risk of type 2 diabetes, even at habitual exposure levels below current ADI thresholds [
38]. Developmental animal studies similarly suggest that exposure timing may modify metabolic vulnerability: prenatal and early-life aspartame exposure has been associated with impaired insulin sensitivity and increased offspring adiposity in mice [
39,
40]. Additionally, compelling molecular evidence suggests that aspartame directly interacts with insulin, promoting cross-seeding and co-aggregation, which results in the formation of protease-resistant and cytotoxic amyloid-like fibrillar assemblies, thereby potentially contributing to the progression of diabetes-associated complications [
41]. Nevertheless, these studies vary substantially in dose selection, route of administration, and the transparency of human-equivalent dose assumptions. Their value therefore lies less in providing direct quantitative risk estimates than in identifying sensitive windows of exposure and candidate mechanisms that should be examined in better standardized human research.
A complementary mechanistic hypothesis concerns the uncoupling of sweet taste from caloric delivery. Experimental work in
Drosophila and mice suggests that intense sweetness without corresponding energy input may alter feeding behavior, promote compensatory intake, and engage fasting-related neural pathways involving neuropeptide Y and reward-learning processes [
42,
43]. These studies are conceptually important because they frame artificial sweeteners not only as metabolic additives but also as sensory signals capable of reshaping anticipatory regulation. Even so, cross-species extrapolation remains uncertain, particularly for insect models. Overall, the available evidence suggests that sweet taste–calorie mismatch represents a plausible mechanism contributing to insulin dysregulation, but one that still requires rigorous validation under human-relevant exposure conditions. Collectively, studies in a variety of models, from insects to mammals, suggest that artificial sweeteners may have important effects on insulin secretion and its regulation (Fig. 2B). In weighing this evidence, it is important to keep the two categories distinct. The cohort findings establish associations between artificial sweetener intake and metabolic outcomes such as weight gain and type 2 diabetes, but they remain vulnerable to confounding and reverse causation and therefore cannot on their own demonstrate causation. The animal and molecular studies provide biologically plausible mechanisms and identify sensitive exposure windows, but they often rely on doses, routes, and model systems that do not correspond directly to habitual human intake. These two lines of evidence are best read as complementary rather than interchangeable, with the population data indicating where associations exist and the mechanistic data suggesting how such effects might arise. Accordingly, the balance of current evidence supports concern and further investigation, but it does not yet establish a confirmed causal effect of artificial sweeteners on insulin resistance in humans.
4 Potential Associations Between Artificial Sweeteners and the Risk of Specific Cancers
Carcinogenicity has remained the most persistent and publicly visible concern in artificial sweetener research. The importance of this topic lies not only in the severity of the endpoint, but also in the way it exposes recurring tensions between toxicological hazard identification, dose relevance, mechanistic plausibility, and human epidemiological inference. Aspartame sits at the center of this debate because it is extensively metabolized, including to methanol and subsequently to formaldehyde and formic acid [
44,
45]. Yet the presence of a potentially hazardous metabolite does not by itself establish meaningful cancer risk at customary dietary exposure levels. For this reason, the carcinogenicity literature must be interpreted through a framework that distinguishes biological possibility from quantitatively supported risk.
That distinction is particularly important in interpreting the 2023 IARC classification of aspartame as possibly carcinogenic to humans (Group 2B) [
46]. IARC evaluates hazard potential rather than real-world risk under current intake patterns, which explains why this classification can coexist with regulatory decisions that retain existing ADI values. The apparent inconsistency is therefore methodological rather than merely rhetorical. It reflects the fact that hazard identification, regulatory risk assessment, and public communication are not interchangeable exercises. For high-level academic discussion, the crucial issue is to specify which question each evidentiary framework is answering and to avoid conflating hazard labels with demonstrated population-level cancer burden.
Long-term rodent studies remain influential because they address cumulative exposure, lifespan effects, and in some cases prenatal exposure windows that are difficult to examine experimentally in humans. Studies by Soffritti and colleagues reported increased incidences of hematologic and other tumors in rats exposed to aspartame over the life course and from prenatal stages onward [
47,
48]. Comparable concerns were later raised for sucralose in lifespan mouse experiments initiated during fetal development [
49]. These findings warrant attention because developmental timing may materially alter susceptibility. At the same time, the studies remain controversial owing to debate over pathology interpretation, dose framing, and external validity for human dietary exposure. Their strongest contribution is therefore to identify conditions under which carcinogenic hazard signals emerge, rather than to settle the quantitative risk question for consumers.
The carcinogenicity discussion has also expanded beyond parent compounds to include impurities and transformation products. In the case of sucralose, particular attention has focused on sucralose-6-acetate, which has been described as an impurity and as a potential
in vivo transformation product [
23,
50,
51]. This shift is analytically important because safety evaluation of food additives increasingly requires consideration of manufacturing by-products and gastrointestinal conversion rather than assessment of the nominal sweetener alone. However, current evidence on the toxicological significance of sucralose-6-acetate remains limited.
Human observational data provide a different but complementary line of evidence. In the NutriNet-Santé cohort, higher artificial sweetener intake was associated with increased overall cancer risk and with selected site-specific outcomes, including breast and obesity-related cancers. These associations were driven mainly by aspartame and acesulfame K [
52]. Mechanistic literature has attempted to contextualize these associations by focusing on formaldehyde-related adduct formation after aspartame metabolism [
53,
54] and, more recently, by proposing molecular targets through network toxicology and docking analyses [
55]. These approaches are useful for hypothesis generation, but they occupy different evidentiary tiers. Radioisotope-binding studies, epidemiological associations, and molecular docking cannot be interpreted as mutually equivalent proof. A robust synthesis therefore requires weighing them according to exposure relevance, biological plausibility, and methodological limitations.
The historical trajectory of sodium saccharin is especially valuable because it demonstrates how apparent carcinogenicity can be reinterpreted once a species-specific mode of action is established. Early findings linked high-dose saccharin exposure to bladder cancer in male rats, but subsequent work showed that the effect depended on urinary precipitates, α2u-globulin-related processes, and bladder conditions not present in humans [
56-
59]. This case illustrates a broader principle that is highly relevant for sweetener safety assessments: positive animal signals are most informative when their mode of action is identified and then evaluated for human plausibility. Accordingly, an increasing number of population-based studies have emerged in recent years. These studies provide important complementary evidence to the potential risks suggested by animal experiments [
60-
62]. Although observational studies cannot fully eliminate confounding factors, their findings nonetheless suggest that long-term consumption of certain artificial sweeteners may be associated with an increased risk of cancer. Without this step, both overestimation and underestimation of risk become likely.
Because this section draws on several types of evidence, it is useful to state how they are weighted. Long-term rodent carcinogenicity studies and human epidemiological studies carry the most weight for judging cancer relevance, because they address disease endpoints under cumulative exposure, although each has characteristic limitations, including species specific modes of action in rodents and residual confounding in observational cohorts. Mechanistic findings from adduct formation, network toxicology, and molecular docking carry less independent weight, since they establish biological plausibility and generate hypotheses rather than demonstrate cancer risk at dietary exposure levels. When these categories are ordered by their relevance to human cancer risk, converging evidence across long-term animal studies and population cohorts is treated as more informative than any single mechanistic signal, and mechanistic data are used to interpret and contextualize those findings rather than to establish risk on their own. On this basis, the assessment in this review does not rest primarily on one evidence type, but on the degree of agreement among long-term population studies, lifetime animal studies, and mechanistic work, with hazard identification and dietary risk assessment kept as separate questions throughout.
Taken together, current evidence does not support a binary conclusion that artificial sweeteners are either carcinogenic as a class or uniformly safe. Instead, cancer-related signals appear to be compound-specific and shaped by exposure timing, dose range, impurities or metabolites, species differences, and the evidentiary standard applied. Future progress will depend on integrating long-term toxicology, mechanism-focused studies, and epidemiology with improved exposure assessment and confounder control. Equally important, scholarly and regulatory discussions should communicate clearly that hazard classification and dietary risk assessment answer related but distinct questions. Fig. 3 summarizes the principal cancer-associated pathways and evidence domains discussed in this section.
5 Potential Effects of Artificial Sweeteners on the Progression of Inflammatory Bowel Disease
Because many artificial sweeteners reach the intestinal lumen either intact or only partially metabolized, their potential effects on mucosal homeostasis are especially relevant to IBD [
4]. IBD is a multifactorial disorder shaped by genetic susceptibility, immune dysregulation, environmental exposures, microbial ecology, and diet [
63,
64]. Against this background, artificial sweeteners are not merely passive food additives; they are plausible modulators of epithelial barrier integrity, luminal signaling, and inflammatory tone. The scientific question is therefore whether chronic exposure contributes to disease initiation, disease exacerbation, or both, and whether such effects differ between healthy and already inflamed intestines.
Current evidence is strongest for sucralose, although it is still derived predominantly from animal models. Patient survey data indicate that individuals with IBD often favor artificially sweetened products over table sugar, which raises the practical relevance of this question [
65]. In rodent colitis models, sucralose aggravated disease severity, increased inflammatory mediators, impaired barrier function, and altered luminal enzyme activity and permeability markers. In both studies, sucralose was administered by adding it to the drinking water at a concentration of 1.5 mg/mL for continuous intake over 6 weeks. After dose conversion, the exposure level remained far below the human ADI [
66,
67]. These studies converge on a coherent pathogenic pattern characterized by barrier disruption, cytokine amplification, and innate immune activation, but they do not yet establish that the same magnitude of effect occurs in human IBD under habitual dietary exposure. The main value of the current evidence is therefore mechanistic plausibility rather than definitive clinical translation.
Evidence for other sweeteners is emerging but remains more difficult to interpret. In murine colitis models, aspartame has been associated with worsened histological injury, reduced tight-junction protein expression, and enhanced inflammatory infiltration [
68], while acesulfame K has been linked to IBD-like phenotypes and pathway alterations involving focal adhesion and phosphoinositide 3-kinase (PI3K)–protein kinase B (Akt) signaling [
69]. Yet these studies also illustrate a recurrent limitation in the literature: uncertainty regarding dose alignment, route of exposure, and the validity of cross-species extrapolation. Accordingly, a high-level synthesis should conclude that several artificial sweeteners can plausibly exacerbate intestinal inflammation under experimental conditions, but that the exposure thresholds, susceptible populations, and clinical relevance remain insufficiently resolved. Fig. 4 provides an overview of the principal IBD-related mechanisms proposed to date.
6 Multiple Pathways Linking Artificial Sweeteners to Neurotoxicity and Brain Injury
Neurobiological evaluation of artificial sweeteners must begin with the physiology of sweet sensing itself. Sweet compounds activate the T1R2/T1R3 receptor complex on taste cells and initiate downstream signaling cascades involving G protein-coupled receptor (GPCR)-mediated pathways such as cyclic adenosine monophosphate (cAMP) and inositol 1,4,5-trisphosphate (IP3), with subsequent relay to central gustatory circuits [
70-
72]. Artificial sweeteners share access to this sensory system with nutritive sugars, but they differ in post-ingestive consequences because they deliver little or no energy and may display distinct receptor-binding profiles or signaling biases. The resulting discrepancy between oral sweetness and metabolic outcome provides a mechanistic basis for considering effects not only on taste perception, but also on reward processing, appetite regulation, and potentially neural function.
Several studies suggest that the neural representation of artificial sweetness differs from that of sucrose, even when conscious sweetness perception appears similar. Human imaging data indicate that sucralose activates gustatory pathways but produces weaker responses than sucrose in reward-related regions, implying that sensory equivalence does not necessarily translate into equivalent central valuation [
73]. Animal studies extend this observation by linking high-dose exposure to altered peripheral signaling and behavioral phenotypes, including increased diuresis through sweet receptor pathways and anxiety-like or learning-related changes after saccharin exposure [
74]. In the zebrafish model, long-term exposure to sublethal concentrations of saccharin has been reported to cause anxiety-like behaviors, impair learning and memory abilities, and disrupt the homeostasis of neurotransmitters [
75]. Therefore, these studies support the view that “long-term exposure to sweeteners may affect the nervous system” at the mechanism level, but the exposure doses and patterns they used are often difficult to directly correspond to the actual human intake level, so extrapolation to population health effects still needs to be cautious.
Aspartame has received the most sustained attention in this domain because its metabolites intersect with neurotransmitter biology. Phenylalanine can influence catecholaminergic pathways, aspartic acid relates to excitatory amino acid metabolism, and methanol can be converted to potentially toxic intermediates [
53]. Although these metabolites are generally regarded as negligible within the acceptable daily intake range, concerns persist regarding cumulative or long-term neurological consequences. One study showed that chronic intake of aspartame equivalent to 7% to 28% of the U.S. Food and Drug Administration (FDA)-regulated ADI caused mice to exhibit memory impairment by inhibition of the brain-derived neurotrophic factor (BDNF)/tyrosine kinase B (TrkB) pathway [
76]. In addition, a study on neonatal Wistar rats showed that consuming 75 mg/kg bw/d or 125 mg/kg bw/d of aspartame for 21 days resulted in anxiety-like behaviors and impaired cortical spreading depression (CSD) transmission [
77]. However, the administered doses substantially exceed human-relevant exposure levels when referenced to the ADI. In another study, chronic aspartame consumption has been shown to delay the time required for rats to find a reward in the T-maze, suggesting potential impairments in long-term memory [
78]. Additionally, significant increases in muscarinic receptor densities were observed in various brain regions, including the frontal cortex, hippocampus, and cerebellum. Notably, the midbrain showed a marked increase in the activity of Na
+, K
+-ATPase, an enzyme linked to memory function. A similar study using adult zebrafish found that exposure to aspartame within the ADI range for two months led to altered behavioral traits, such as increased phototropism and impaired exploratory and color-preference behaviors [
79]. Lebda et al. reported that subchronic exposure to aspartame 240 mg/kg daily for two months has been shown to severely disrupt energy production in the brain [
80]. However, the aspartame dose used in that study was higher than the ADI for humans established by the European Food Safety Authority (EFSA). The authors suggested that higher doses may be required in rodents because of their faster metabolic rate to achieve comparable internal exposure. In contrast, other studies, based on body-surface-area conversion, have suggested that the equivalent rodent dose may be lower than the corresponding human dose. These differing assumptions indicate that a standardized and widely accepted approach for translating animal doses to human exposure has not yet been established, which limits the interpretation and cross-species extrapolation of such findings.
Sucralose is often considered relatively inert from a neurotoxicity standpoint, yet emerging evidence indicates that its effects on neural function and reward processing merit closer scrutiny. One study found that chronic sucralose intake (0.02%) led to the overexpression of Delta FosB in regions associated with food reward, such as the inferior limbic cortex and amygdala [
81]. This indicates that chronic administration of foods containing sucralose causes long-term changes in the reward system. Importantly, artificial sweeteners are rarely consumed in isolation. In real diets, they often co-occur with carbohydrate-rich foods, making combined exposure scenarios especially relevant. For example, research by Dalenberg et al. showed that healthy individuals who consumed carbohydrate beverages containing the daily normal exposure dose of sucralose for 10 days experienced a decrease in insulin sensitivity [
82]. This effect is linked to diminished brain response to sweetness in areas such as the midbrain, insula, and cingulate gyrus. The presence of carbohydrates with sucralose seems to disrupt glucose metabolism and decrease brain sensitivity to sweetness, likely due to dysregulation in the gut-brain axis. These findings suggest that the consumption of low-calorie or calorie-free sweeteners may interfere with the body’s normal response to sugar and increase the risk of sugar overconsumption and neurological dysfunction [
83]. Additionally, artificial sweeteners such as acesulfame K may exert neurotoxic effects by inducing oxidative stress and DNA damage, promoting apoptosis, and disrupting neurotransmitter homeostasis, representing a canonical pathway of neurotoxicity [
84]. In summary, although artificial sweeteners bind to sweet taste receptors in a manner similar to caloric sugars, their distinct metabolic and absorption profiles may lead to prolonged or altered receptor stimulation, potentially contributing to neurobiological effects. These observations also raise a question about how acceptable daily intakes are derived. Current ADI values are established mainly from animal studies that administer a single purified sweetener, whereas real diets combine sweeteners with carbohydrates and other food components, and the study by Dalenberg et al. indicates that such combined exposure can alter the brain response to sweetness even at habitual doses. Because single sweetener testing does not capture these matrix and carbohydrate interactions, an ADI derived only from isolated sweetener data may not fully reflect the exposure conditions under which neurological effects occur. This gap, together with the absence of a standardized method for converting animal doses to human exposure, means that ADI thresholds based solely on single sweetener studies could underestimate the long term risk of central nervous system effects in populations. Incorporating food matrix and carbohydrate co exposure into the assessment framework would therefore help align ADI standards more closely with real world consumption, although the magnitude of any such underestimation remains to be quantified. Fig. 5 summarizes the principal neurobiological pathways discussed here.
7 Potential Associations Between Artificial Sweeteners and Cardiovascular Disease
Cardiovascular risk has emerged more recently as a major theme in artificial sweetener research, in part because the metabolic pathways implicated elsewhere in this review, including insulin resistance, dyslipidaemia, inflammation, oxidative stress and microbiota perturbation, are also central to atherosclerotic disease. Prospective human data have associated higher intakes of aspartame, acesulfame K, and sucralose with increased cardiovascular risk [
85]. Specifically, aspartame intake was associated with a heightened risk of cerebrovascular events, while acesulfame K and sucralose were linked to a greater risk of coronary heart disease. The specific mechanism of cardiovascular disease needs to be elucidated through animal models. In a mouse model study, researchers found that chronic consumption of aspartame at doses of 80 and 160 mg/kg over 90 days resulted in higher serum malondialdehyde (MDA) levels and lower nitric oxide (NO), creatine kinase (CK), and CK-myocardial band (MB) levels. Additionally, the expression of pro-apoptotic genes such as tumor protein p53 (P53), B-cell lymphoma 2 (Bcl-2), Bcl-2-associated X gene (Bax), and cysteine-aspartic acid protease 3 (Caspase-3/CASP3) was altered at the higher dose of 160 mg/kg [
86]. Although the doses in this study far exceed the human ADI, the oxidative stress and apoptosis observed in myocardial tissue after long-term high-dose aspartame intake remain mechanistically informative and warrant further investigation under human-relevant exposure levels. Separate supplementation with saccharin has also been shown to exacerbate cardiovascular disease risks associated with high-fat diets. A study using apolipoprotein E knockout (ApoE
−/−) mice revealed that saccharin exposure at ADI doses intensifies lipid metabolism disorders. This occurs through Niemann-Pick C1-Like Protein 1 (Npc1l1), which accelerates intestinal cholesterol absorption while inhibiting cholesterol excretion, thereby further promoting the progression of atherosclerosis [
87]. Furthermore, another recent study demonstrated that chronic lower-dose aspartame intake in ApoE
−/− mice exacerbates atherosclerotic plaque formation and growth via insulin- and CX3CL1-CX3CR1-dependent mechanisms [
88]. The study also validated the findings from the mouse model using a monkey model, further enhancing its relevance to humans. Importantly, these findings do not suggest a single cardiovascular mechanism shared uniformly by all compounds. Rather, they point to multiple converging pathways through which sweetener-specific exposures may interact with host metabolic vulnerability to accelerate vascular pathology.
At the population level, interpretation remains complex but increasingly important. UK Biobank analyses reported positive associations between artificial sweetener intake and incident cardiovascular outcomes, including coronary disease and peripheral arterial disease [
89], while studies in people living with human immunodeficiency virus (HIV) linked aspartame intake to coronary plaque burden and inflammatory markers such as monocyte chemoattractant protein-1 (MCP-1) and lipoprotein-associated phospholipase A2 (Lp-PLA2) [
90]. Consistent with these findings, a cross-sectional analysis based on the European Health Interview Survey in Hungary also reported that artificial sweetener use was independently associated with a higher prevalence of cardiovascular disease after adjustment for conventional risk factors [
91]. These observations strengthen the argument that cardiovascular effects deserve independent consideration rather than being treated as secondary consequences of obesity alone. Nevertheless, observational designs remain vulnerable to residual confounding, dietary clustering, and reverse causation. The current state of evidence therefore supports concern and further investigation, but not yet a simple causal statement applicable across all sweeteners, populations, and exposure ranges. Fig. 6 summarizes the principal cardiovascular pathways implicated to date.
8 Artificial Sweeteners Disrupt Host Metabolism by Affecting the Balance of Gut Microbiota
Gut microbiota disruption is one of the most consistent mechanistic themes across the artificial sweetener literature and serves as a conceptual bridge linking metabolic, inflammatory, carcinogenic, and cardiovascular outcomes. With the exception of compounds that are substantially metabolized before reaching the distal gut, several artificial sweeteners can arrive in the intestine in forms capable of direct microbial interaction [
6,
18,
92,
93]. This feature is biologically important because microbiota composition and metabolite production influence epithelial integrity, bile acid signaling, insulin sensitivity, lipid homeostasis, and immune activation. Accordingly, the gut microbiota should be viewed not as an isolated endpoint, but as an integrating mechanistic layer through which multiple disease associations may be mediated.
Aspartame-related evidence illustrates both the promise and the complexity of this framework. In diet-induced obesity models, low-dose aspartame has been associated with impaired glucose regulation alongside shifts in taxa such as
Enterobacteriaceae,
Clostridium leptum, and
Roseburia and with altered short-chain fatty acid profiles [
6]. This effect was associated with an increase in the relative abundance of
Enterobacteriaceae and
Clostridium leptum. Furthermore, aspartame reduced the typical HFD-induced increase in the
Firmicutes/
Bacteroidetes ratio and was associated with an increase in the short-chain fatty acid propionate [
6]. Aspartame not only has an impact on the individual’s own gut microbiota, but may also indirectly affect the gut microbiota balance of the offspring mice through the mother’s intake of aspartame. Relevant reports indicate that this intergenerational inheritance can lead to an increase in the relative abundance of certain bacterial genera such as
Negativibacillus,
Enterocloster, and
Parabacteroides_B_862066 in newborns, thereby disrupting purine metabolism and causing oxidative stress and activation of the inflammasome NOD-like receptor family pyrin domain-containing 3 (NLRP3), resulting in oxidative-reductive imbalance in the lungs [
94]. These findings broaden the biological scope of sweetener research beyond direct host toxicity, but they also intensify the need for careful interpretation because microbiome datasets are highly sensitive to host background, diet, analytical pipeline, and species differences.
The effects of sucralose on health have been extensively studied, particularly in relation to gut microbiota and metabolic homeostasis. In mouse models, sucralose consumption at a dose approximately equivalent to the human ADI was shown to significantly increase the relative abundance of intestinal genera
Bacteroides and
Clostridium, leading to the production and accumulation of deoxycholic acid (DCA) in feces, serum, and liver. This accumulation was linked to the development of nonalcoholic fatty liver disease (NAFLD) in mice [
93]. In addition, Li et al. found that a low dose sucralose exacerbates colitis and alters the gut microbiota in a mouse model of colorectal cancer [
50]. Sucralose has also been reported to increase the relative abundance of
Firmicutes,
Actinomycetes,
Peptostreptococcus stomatis,
Clostridium symbiosum, and
Peptostreptococcus anaerobius, while reducing the relative abundance of
Proteobacteria. These changes disrupted the intestinal barrier, impaired the deconjugated bilirubin-mediated inactivation of digestive proteases, and potentially elevated the risk of colorectal cancer [
50]. Evidence also suggests that sucralose disrupts gut microbiota homeostasis and promotes inflammation. In a study where C57BL/6 male mice consumed sucralose in drinking water for six months at the human ADI [
95], researchers observed elevated expression of pro-inflammatory genes in the liver. Sucralose affected the composition and developmental dynamics of the gut microbiota, implying that long-term consumption could increase the risk of tissue inflammation. Further studies showed that long-term intake of sucralose at doses equivalent to ADI in humans for 6 months disrupted bile acid metabolism and hepatic lipid regulation, resulting in a decline in gut bacteria involved in bile acid metabolism, such as
Lactobacillus and
Ruminococcus [
7], and resulted in a reduction in the relative abundance of secondary bile acid biosynthesis pathways and bacterial bile salt hydrolase genes. In addition, sucralose-treated mice showed increased expression of hepatic lipogenic genes and disturbances in lipid metabolism, including a decreased phosphatidylcholine/phosphatidylethanolamine (PC/PE) ratio, reduced ceramide levels, and elevated cholesterol levels. Overall, these findings suggest that sucralose disrupts hepatic lipid and cholesterol homeostasis by altering bile acid metabolism mediated by the gut microbiota [
7].
Interestingly, clinical studies involving human populations revealed results that differ significantly from those observed in animal models. Chen et al. found that in humans, sucralose consumption was associated with a notably higher relative abundance of
Bacteroidetes compared to controls, whereas the opposite effect was observed in mice and rats [
96]. Additionally, sucralose ingestion in humans led to a significant increase in the relative abundance of
Actinobacteria in the gut, and a marked decrease in the relative abundance of
Verrucomicrobia seen in animal models. These findings suggest divergent effects of sucralose on gut microbiota and related phenotypes between species. In humans, sucralose may reduce obesity by decreasing the ratio of
Firmicutes/
Bacteroidetes and increasing
Actinobacteria [
96]. Such divergence has major interpretive implications. It suggests that rodent microbiome findings are useful for mechanism generation but insufficient for direct inference about human benefit or harm without clinical corroboration. More broadly, it highlights a central challenge in this field: artificial sweeteners may act through the microbiota, but the microbiota itself is species-specific, environmentally sensitive, and methodologically variable.
Evidence for saccharin and other less extensively studied sweeteners supports the broader principle that microbiota effects are not unique to sucralose. In mice, prolonged saccharin exposure has been associated with altered microbial composition and increased pro-inflammatory markers, implying that hepatic or systemic inflammatory outcomes may be microbiota-mediated rather than purely direct toxic effects [
97]. In C57BL/6J male mice, the addition of saccharin to drinking water for 3–6 months at a human ADI dose resulted in increased expression of pro-inflammatory markers, including inducible nitric oxide synthase (iNOS) and TNF-α in the liver. For gut microbiota, they observed that after three months of saccharin consumption, the relative abundance of genera such as
Sporosarcina,
Jeotgalicoccus,
Akkermansia, and
Oscillospira significantly increased, while
Anaerostipes and
Ruminococcus declined. By six months, the relative abundance of
Corynebacterium,
Roseburia, and
Turicibacter increased further, while
Ruminococcus,
Adlercreutzia, and
Dorea decreased further. Several of these bacterial genera, such as
Corynebacterium,
Turicibacter,
Anaerostipes,
Dorea,
Roseburia, and
Ruminococcus, are associated with inflammation, suggesting that the pro-inflammatory effects of saccharin may be mediated through changes in gut microbial composition [
97]. This observation is important because it shifts the discussion from individual taxa to functional consequences. The key analytical task is therefore to identify which microbial changes are reproducible, which are incidental, and which are linked to measurable host endpoints such as inflammation, permeability, bile acid metabolism, or carcinogenesis.
Acesulfame K and neotame further extend this mechanistic landscape. Acesulfame K has been linked to inflammatory metabolite profiles and depletion of beneficial taxa in mice [
92]. Specifically, acesulfame K downregulated the relative abundance of beneficial bacteria such as
Bacillus,
Eggerthella, and
Bifidobacter, as well as anti-inflammatory species like
Faecalibacterium and
Akkermansia. At the same time, it increased the relative abundance of
Collinsella, a genus positively associated with inflammation. These microbial changes coincided with disturbances in fatty acid metabolism, including elevated concentrations of several long-chain fatty acids such as C18:3n3, C18:2n6c, and C18:3n6, together with increased lipopolysaccharides (LPS), and these changes accompanied the aggravation of systemic and hepatic inflammation. Similarly, neotame consumption has been shown to disrupt gut microbiota and fecal metabolite profiles. Another study on neotame revealed that administering of 2.5 times the human ADI dose of neotame to male CD-1 mice for four consecutive weeks reduced α-diversity in the gut microbiota and altered β-diversity, characterized by a significant decrease in the relative abundance of
Firmicutes and an increase in
Bacteroidetes [
98]. Another
in vitro study by Shil et al. based on co-culturing intestinal epithelial cells (Caco-2) with
Escherichia coli and
Enterococcus faecalis showed that neotame has an adverse effect on intestinal epithelium through signaling via T1R3 [
21]. Specifically, neotame exposure led to intestinal epithelial cell apoptosis and cell death, while knockdown of T1R3 expression via small interfering RNA (siRNA) significantly mitigated these effects. Neotame also disrupted the intestinal barrier, as evidenced by increased monolayer leakage, reduced claudin-3 expression on the cell surface, and enhanced adhesion of
Escherichia coli and
Enterococcus faecalis to Caco-2 cells. These studies are valuable because they suggest that both microbial and epithelial compartments may be targets of sweetener exposure. At the same time, the available evidence remains uneven in depth and quality across compounds, which is why next-generation sweeteners should not be assumed safer simply because fewer studies currently exist.
The species divergence seen for sucralose also has direct implications for how acceptable daily intakes should be interpreted. Current ADI values rest largely on animal studies, yet the microbiota responds to sweeteners in a species specific way, and the human and rodent findings for sucralose point in opposite directions. Most of these studies also test a single sweetener rather than the composite diets in which sweeteners are actually consumed, so they capture neither the interactions among dietary components nor the between species differences in microbial ecology. For these reasons, an ADI threshold derived mainly from animal gut microbiota data cannot be assumed to reflect long term exposure risk in humans with confidence. Microbiota related endpoints in particular call for confirmation in human studies before they can inform intake thresholds, and animal data are better used to generate mechanistic hypotheses than to set human limits on their own.
Future research should move beyond cataloguing taxonomic shifts and toward integrated microbiota–host models that capture exposure dose, dietary matrix, metabolite production, epithelial responses, and disease-relevant phenotypes. Standardized multi-omics pipelines, longitudinal human cohorts, and intervention studies stratified by dietary pattern and baseline microbiome composition will be especially important. In parallel, researchers should account for regional dietary habits, developmental stage, and inter-individual variability in microbial ecology. Fig. 7 summarizes the gut microbiota-centered mechanisms that may connect artificial sweetener exposure with chronic disease outcomes.
9 Conclusions and Future Perspectives
Artificial sweeteners were introduced and widely adopted as sugar substitutes because they offer a practical means of reducing caloric intake while preserving sweetness. The evidence reviewed here, however, indicates that their health effects cannot be evaluated solely through this narrow substitution logic. Across metabolic, oncologic, gastrointestinal, neurobiological, cardiovascular, and microbiota-centered domains, the literature points to a more complex picture in which risk is compound-specific and modified by dose, exposure window, background diet, developmental stage, and host susceptibility. The strength of evidence also differs by compound and by disease area, and stating this helps readers judge how firm each conclusion is. For aspartame, acesulfame K, and sucralose, associations with type 2 diabetes, cancer, and cardiovascular outcomes are supported by large prospective cohorts as well as by animal and mechanistic work, so the concern rests on more than one line of evidence, although the cohort associations still cannot establish causation on their own. For saccharin, the cancer signal comes mainly from rodent studies whose mode of action is now known to be species specific and of limited relevance to humans, while its cardiovascular and microbiota effects rest largely on animal data. For the second generation sweeteners such as neotame, advantame, and alitame, the evidence is sparse and derives almost entirely from animal or in vitro studies, so their apparent safety reflects limited study rather than demonstrated absence of risk. Readers should therefore weight conclusions drawn from human population studies more heavily than those based only on animal or mechanistic findings, and treat compound specific certainty rather than a single class level judgement as the appropriate basis for dietary and regulatory decisions. Taken together, the evidence does not support a simple conclusion that artificial sweeteners are uniformly harmful, but that they should not be treated as a biologically homogeneous or metabolically inert class. Advancing the field will require standardized long-term human studies, mechanistic experiments designed around realistic exposure scenarios, clearer frameworks for cross-species dose translation, and tighter integration of microbiota, metabolism, and disease phenotyping. Such work will be essential for refining regulatory assessment, improving dietary guidance, and determining whether specific sweeteners can be used safely in targeted populations.
The Author(s) 2026. This article is published by Higher Education Press.
PDF
(15028KB)
