What 71 Studies Actually Found — A 2026 Research Digest
We read 71 peer-reviewed studies on food additives, cosmetic ingredients, and household chemicals. Here's what the science actually says — grouped by what matters.
Building an ingredient safety database means reading a lot of research. Not press releases. Not headlines. The actual papers — peer-reviewed, published in journals like The Lancet, Food and Chemical Toxicology, the EFSA Journal, and IARC Monographs.
GradeMyLabel’s database currently tracks 71 studies spanning 13 categories: food additives, cosmetic preservatives, cleaning agents, sweeteners, colorants, emulsifiers, flavor enhancers, antimicrobials, sunscreen actives, pet food additives, supplements, vape liquids, and oral care ingredients. Each study is AI-summarized and linked to the specific ingredients it covers, so when you scan a product, you see the science behind every flag.
This article distills what those 71 studies actually found. Not every study, but the findings that matter most — organized by health concern rather than ingredient category.
Cancer and Genotoxicity
This is the category that gets the most attention, and for good reason. Several widely used additives have raised red flags in long-term toxicology studies.
Titanium dioxide and DNA damage. The landmark 2021 EFSA opinion on titanium dioxide (E171) concluded that the agency “could not rule out genotoxicity concerns” for titanium dioxide nanoparticles. The panel reviewed over 100 studies and found that TiO2 nanoparticles can induce DNA strand breaks, chromosomal damage, and oxidative stress in mammalian cells. Published in the EFSA Journal (2021;19(5):6585), this assessment led directly to the EU’s food ban in 2022. What made the EFSA opinion unusual was its focus on particle size — the same compound in larger particles posed less concern, but the nanoparticle fraction found in food-grade TiO2 could not be deemed safe.
BHA and tumor promotion. Butylated hydroxyanisole (BHA), a common antioxidant preservative in cereals, snack foods, and chewing gum, was classified as “possibly carcinogenic to humans” (Group 2B) by the International Agency for Research on Cancer. The classification, documented in IARC Monographs Volume 40, was based on sufficient evidence of carcinogenicity in animal studies — specifically, forestomach tumors in rats and hamsters at high doses. The human evidence remains limited, which is why it sits at Group 2B rather than a higher classification. BHA is still approved in most countries, typically at concentrations below 200 ppm.
Red Dye 3 and thyroid tumors. Erythrosine (Red Dye 3, E127) has one of the longest regulatory histories of any food colorant. Studies published in Food and Chemical Toxicology demonstrated that high-dose erythrosine exposure caused thyroid follicular cell tumors in male rats. The mechanism involves disruption of thyroid hormone metabolism — erythrosine inhibits thyroid peroxidase, leading to elevated TSH levels, which in turn drives follicular cell proliferation. The FDA announced a ban on Red Dye 3 in food in January 2025, decades after it was removed from cosmetics. The delay between the animal findings and regulatory action is itself a case study in how slowly the system moves.
Azo dyes and genotoxicity signals. Several synthetic azo dyes — including Allura Red (E129), Sunset Yellow (E110), and Tartrazine (E102) — have been evaluated for genotoxicity with mixed results. A 2012 review in Food and Chemical Toxicology (Vol 50, Issues 3–4) examined the Ames test results and in vivo comet assay data across multiple azo colorants. The findings were not uniform: some dyes showed no genotoxic activity under standard conditions, while others produced positive results in specific assay systems. The scientific consensus is that azo dyes as a class do not present a clear genotoxic risk at approved levels, but the data is inconsistent enough that several regulatory bodies continue to require re-evaluation.
Pattern: Genotoxicity findings tend to be dose-dependent and particle-size-dependent. The gap between laboratory doses and real-world exposure is real — but so are the biological mechanisms these studies identify.
Gut Health and the Microbiome
The gut microbiome has become one of the most active areas of food additive research in the last decade. Several common emulsifiers and sweeteners have been shown to alter gut bacteria composition in ways that may promote inflammation.
Carrageenan and intestinal inflammation. Carrageenan, a seaweed-derived thickener found in dairy alternatives, deli meats, and infant formula, has been studied extensively for its inflammatory effects. Research published in Environmental Health Perspectives demonstrated that degraded carrageenan (poligeenan) induces intestinal inflammation and colitis-like symptoms in animal models. The more relevant question for consumers is whether food-grade carrageenan (undegraded) causes similar effects. Studies in the Journal of Mammary Gland Biology and Neoplasia and elsewhere suggest that even food-grade carrageenan can trigger inflammatory pathways, including NF-kB activation, in intestinal epithelial cells. The Joint FAO/WHO Expert Committee has maintained its safety assessment, but the National Organic Standards Board removed carrageenan from the approved list for organic foods in 2018.
Polysorbate 80 and gut barrier disruption. A widely cited 2015 study in Nature (Vol 519, pp. 92–96) by Chassaing et al. demonstrated that the emulsifiers polysorbate 80 (E433) and carboxymethylcellulose (E466), at concentrations relevant to processed food consumption, altered gut microbiota composition in mice. The changes included reduced microbial diversity and increased penetration of bacteria into the intestinal mucus layer — effectively thinning the gut barrier. The treated mice developed low-grade inflammation, metabolic syndrome features, and in genetically susceptible animals, colitis. This study is frequently cited because it used doses comparable to what humans actually consume, not the extreme doses typical of toxicology studies.
Artificial sweeteners and microbiome shifts. A 2022 study in Cell (Vol 185, Issue 18) by Suez et al. conducted a randomized controlled trial in 120 healthy adults, testing sucralose, aspartame, saccharin, and stevia against glucose and no-supplement controls. All four non-nutritive sweeteners altered the gut microbiome — but saccharin and sucralose had the most pronounced effects, significantly changing microbial composition and function within two weeks. Sucralose exposure was associated with altered glucose tolerance in a subset of participants. Critically, the researchers performed fecal transplants from human participants into germ-free mice and reproduced the glycemic effects, establishing a causal link through the microbiome rather than direct metabolic action.
Pattern: Emulsifiers and sweeteners do not need to be toxic in the traditional sense to matter. Microbiome disruption at realistic doses is a relatively new finding, and regulatory frameworks have been slow to incorporate it.
Endocrine Disruption
Endocrine disruptors interfere with hormone signaling, often at very low concentrations. Several ingredients tracked in our database have demonstrated endocrine activity in controlled studies.
Parabens in breast tissue. Parabens (methylparaben, ethylparaben, propylparaben, butylparaben) are antimicrobial preservatives found in cosmetics, personal care products, and some foods. A study by Darbre et al., published in the Journal of Applied Toxicology (2004;24:5–13), detected intact parabens in human breast tumor tissue samples. The study did not prove that parabens caused the tumors, but it demonstrated that parabens are absorbed through skin and accumulate in tissue without being fully metabolized. Subsequent research confirmed that parabens exhibit weak estrogenic activity — roughly 10,000 to 100,000 times weaker than estradiol — but the concern is chronic low-dose exposure from multiple products simultaneously. The EU’s Scientific Committee on Consumer Safety set concentration limits of 0.4% for individual parabens and 0.8% for mixtures.
BPA and hormonal mimicry. Bisphenol A, used in can linings, thermal receipt paper, and some food packaging, is one of the most studied endocrine disruptors. Research published in Reproductive Toxicology and reviewed by EFSA in 2023 (the agency reduced its tolerable daily intake by a factor of 20,000 — from 4 micrograms to 0.2 nanograms per kilogram of body weight per day) documented estrogenic effects, altered reproductive development, and immune system changes at low doses. The dramatic TDI reduction reflected the panel’s assessment that BPA affects the immune system at far lower levels than previously assumed. BPA is not a food additive per se, but it migrates into food from contact materials, making it relevant to anyone scanning canned goods.
Triclosan and thyroid interference. Triclosan, an antimicrobial once ubiquitous in antibacterial soaps and toothpaste, was shown to interfere with thyroid hormone metabolism in studies published in Aquatic Toxicology and Environmental Health Perspectives. In rat models, triclosan exposure reduced circulating T4 levels by accelerating thyroid hormone clearance through liver enzyme induction. The FDA banned triclosan from consumer hand soaps in 2016, though it remains approved in certain toothpaste formulations (where it reduces gingivitis) and in some industrial and hospital applications.
Pattern: Endocrine disruptors challenge traditional toxicology because effects can occur at very low concentrations — sometimes exhibiting non-monotonic dose-response curves where lower doses produce stronger effects than higher ones.
Neurological and Developmental Effects
A smaller but significant subset of studies in our database addresses effects on brain function and child development.
Artificial colors and child behavior. The 2007 Southampton study, published in The Lancet (Vol 370, Issue 9598), remains the most influential trial on food colorings and behavior. McCann et al. conducted a randomized, double-blind, placebo-controlled study of 153 three-year-olds and 144 eight-to-nine-year-olds. Children given drinks containing a mix of artificial colors (Sunset Yellow, Carmoisine, Tartrazine, Ponceau 4R, Allura Red, Quinoline Yellow) plus sodium benzoate showed increased hyperactive behavior compared to placebo, as measured by both parental observation and objective classroom assessments. The effect sizes were modest but statistically significant. This study directly prompted the EU’s requirement for warning labels (“may have an adverse effect on activity and attention in children”) on foods containing these six dyes.
MSG and symptom sensitivity. Monosodium glutamate has been the subject of scientific debate for decades. A double-blind, placebo-controlled study published in the Journal of Allergy and Clinical Immunology tested self-described MSG-sensitive individuals and found that when MSG was administered without food (in capsules), a subset of participants reported headache, numbness, and flushing at doses of 2.5 grams or higher. However, when MSG was consumed with food — as it would be in normal eating — the effects were not reproducible. The scientific consensus, supported by reviews from FASEB and JECFA, is that MSG is safe for the general population, but a small percentage of people may experience transient symptoms at high doses on an empty stomach. Our database classifies MSG as low risk with a note on individual sensitivity.
Aluminum compounds in antiperspirants. Aluminum chlorohydrate and aluminum zirconium, the active ingredients in most antiperspirants, have been studied for potential neurotoxic effects due to aluminum’s known ability to cross the blood-brain barrier. Research published in the Journal of Inorganic Biochemistry examined dermal absorption rates and found that approximately 0.012% of applied aluminum is absorbed through intact skin — a very small fraction. The hypothesized link between antiperspirant aluminum and Alzheimer’s disease has not been supported by epidemiological studies, though the Alzheimer’s Society notes that the question is not fully resolved. EFSA’s 2008 opinion set a tolerable weekly intake for dietary aluminum, but dermal exposure from cosmetics falls under a separate regulatory framework.
What “Under Review” Actually Means
In GradeMyLabel’s scoring system, some ingredients carry an “under review” designation. This is not a risk level — it is a regulatory status that signals active scientific reassessment.
When a major regulatory body announces a formal review of an ingredient, it means new data has raised questions that existing safety assessments did not address. The ingredient has not been declared unsafe, but the agency has determined that the current evidence base is insufficient to maintain confidence in prior approvals.
For consumers, “under review” is arguably the most actionable status. A “banned” ingredient is already off the shelves in that jurisdiction. An “approved” ingredient has been assessed and cleared. But an “under review” ingredient is in limbo — still legal, still in products, but with enough scientific uncertainty that regulators are actively re-examining it. Our scoring system treats these ingredients with a moderate base score (40 out of 100), reflecting the genuine uncertainty.
Three ingredients in our database that currently carry this status illustrate the pattern: they were approved years or decades ago based on the science available at the time, new studies (often involving microbiome effects, endocrine activity, or nanoparticle behavior) have raised concerns, and regulatory bodies are now conducting formal reassessments that may take years to complete.
The Research Is Not Settled — But the Patterns Are
Seventy-one studies is not exhaustive. The food additive literature alone runs into thousands of papers. But patterns emerge clearly across the research we track.
First, dose matters, but so does duration. Many additives are safe at acute exposure levels but raise questions about chronic, low-dose, daily exposure over years or decades. Traditional toxicology was designed to catch the former. The latter is harder to study and harder to regulate.
Second, the microbiome has changed the conversation. Ten years ago, emulsifier safety was evaluated primarily through cancer and organ toxicity endpoints. The discovery that common emulsifiers alter gut bacteria at dietary-relevant doses has introduced an entirely new category of concern that existing regulatory frameworks were not built to assess.
Third, regulatory action consistently lags behind science. The gap between Red Dye 3’s animal tumor data and the FDA’s 2025 ban was measured in decades. EFSA’s titanium dioxide ban came years after genotoxicity signals appeared in the literature. This lag is not necessarily negligent — regulatory certainty requires a high evidence bar — but it means consumers who want to act on emerging science cannot rely solely on “approved” status.
GradeMyLabel does not tell you what to eat or what to avoid. It gives you access to the same research that regulators and scientists use — scored, summarized, and linked to the products in your hand. Every ingredient score is backed by the studies in our database, and every study is traceable to its source journal.
Scan a label. Read the research. Make your own call.