Food colourants and ultra-processed foods

Consumers associate colour with quality and children are particularly attracted to highly coloured products often with poor nutritional value such as sweets and soft drinks. While some food dyes are extracted from natural sources, the most commonly employed are synthetic organic compounds, known as artificial food colourants (AFCs). Azo dyes are produced in the largest quantities, constituting over 60% used in foods and drinks, but also pharmaceuticals, cosmetics, paper and textiles. The most frequently added to foods are tartrazine (E-102), sunset yellow (E-110), carmosine (E-122), amaranth (E-123), ponceau 4R (E-124), allura red AC (E-129), brilliant blue (E-133) and HT brown (E-155). As they have no nutritional value or preservation properties, the improvement in appearance of food products must be balanced against increasing concerns about hazards to health. Given the attractiveness of highly coloured processed confectionery, soft drinks, etc, it is not surprising that in children, estimates of the average cumulative intake in the diet have exceeded WHO-acceptable daily intake levels in a number of countries.4 5

Ultra-processed foods (UPFs) and drinks commonly contain combinations of AFCs, benzoate preservatives, non-caloric sweeteners and emulsifiers. Recent publications have highlighted adverse effects of non-caloric sweeteners and emulsifiers in adults.6 7 No studies have established the long-term health consequences of cocktails of additives, but consistent evidence points to individual additives in highly processed foods being hazardous to health, and experimental models have identified potential mechanisms.8 Of greatest concern is high intake in children who have the potential to have even greater long-term adverse effects than have been observed in adult studies. A literature review revealed high intake of UPFs in children not only in high but also middle and low-income countries. In the USA, Canada and the UK, UPFs constitute over 50% of childhood daily energy intake.9

AFCs/benzoate preservatives and behaviour

Feingold published an uncontrolled study suggesting that 68% of children with learning difficulties and behaviour problems improved on a diet excluding AFCs, preservatives and supposed salicylate-containing fruits and vegetables.10 Subsequent conflicting outcomes from studies of dietary restriction in children with behaviour disorders have been compromised by poor study design and the variability of definitions of ADHD and conduct disorders. It is, therefore, tempting to dismiss the concept of an association between AFCs and behaviour disturbance; indeed, the European Food Safety Authority (EFSA), in 2008, reviewed past and more recent data on the association, and despite evidence to the contrary, rejected the relevance of a link.11 However, the repeated assertions from parents warrant further investigations.

A systematic review and meta-analysis of published double-blind placebo-controlled trials showed a significant effect of AFCs and other additives on the behaviour of children with ADHD. A secondary analysis in non-hyperactive children showed a significant adverse effect on behaviour when exposed to additives compared with placebo.12 Following the systematic review, my team conducted two large studies investigating the effects of a mixture of commonly used azo dye AFCs and sodium benzoate (E-211) in whole population samples of children.13 14 They were challenged in double-blind form with the mixture of additives or placebo in a drink administered daily for 1 week while on an otherwise additive-free diet. There were consistently higher levels of hyperactive behaviour when on artificial additives rather than placebo. There were no differences in response between children with or without pre-existing allergy, conduct disorder or ADHD. The effect size was similar to that calculated in the meta-analysis (0.28 (0.08–0.49)).12 The accumulated evidence indicates that mixtures of AFCs and benzoate preservatives have a significant effect on children’s behaviour. While the effect size is relatively small, in children with pre-existing conduct disorders, this can be ‘the last straw to break the camel’s back’, and impacts on long-term educational achievement and career attainment are unknown.

Potential mechanisms for adverse responses to AFCs and benzoates

While past concerns have focused on additives inducing allergies, the only confirmed IgE antibody-mediated responses have been to colourants from natural sources such as carmine (E-120) and annatto (E-160). They are likely to contain sufficient protein to act as conventional allergens and are the most common food colourants to cause urticaria/angio-oedema.15 In vitro and human challenge studies have indicated that tartrazine induces non-IgE-mediated histamine release.16 17 We genotyped the children from our double-blind AFC and sodium benzoate challenge study.14 There was a gene/environment interaction in that children who had an adverse behavioural response to additive challenge were significantly more likely to have a polymorphism in the histamine-N-methyl-transferase gene.18 The polymorphism is associated with impaired histamine degradation so that AFC triggering of histamine release would be more likely to have the adverse effects on behaviour. Histamine is released from cells in the third ventricle and posterior hypothalamus, and it is likely that AFC small molecules or their metabolites will cross the blood–brain barrier inducing non-IgE-mediated histamine release. Histamine 1, 2 and 3 receptors are present on central nervous system neurons and influence sleep/wake cycles and behaviour. This provides additional credibility to a causal association between AFC exposure and disturbed behaviour.19 20

The concerns expressed in publications about the EFSA dismissive response to evidence in relation to child behavioural effects21 eventually resulted in the European parliament adopting a legislative package mandating that products containing azo dye AFCs should be labelled with a health warning, ‘may have an adverse effect on activity and attention in children’. This has impacted on food manufacturers moving to natural alternatively sourced colourants, but regulators will need to be mindful of allergenicity before granting approval. We do not need one problem to be replaced by another. However, some of the natural colourants have additional nutraceutical properties such as beta-carotene (160e) from carrots, betanin (E-162) from beetroot and curcumin (E-100) from turmeric.22

Sweeteners: depression and cardiovascular disease

A recent large US cohort study followed depression-free middle-aged women with food frequency questionnaires administered every 4 years. Those with a high UPF consumption were more likely to have developed new-onset depression even allowing for confounders (cigarette smoking, less exercise and comorbidities including diabetes, hypertension and dyslipidaemia). Discriminating individual additive components highlighted that artificially sweeteners had the greatest effect on depression risk.23 Non-caloric artificial sweeteners (sucralose E-955, acesulfame E-950 and saccharin E-944) differ from natural sugars in stimulating purinergic, rather than glutamatergic neurotransmission from the gut.24 Purinergic activity has been long been known to be linked to depression.25

A very large French cohort study followed adults for a median of 9 years with regular 24-hour diet diaries. New-onset cardiovascular disease occurred more frequently in those with a high artificial sweetener intake. Aspartame (E-951) was specifically associated with cerebrovascular disease, while acesulfame potassium (E-950) or sucralose (E-955) increased coronary artery disease risk.6 Statistical modelling accounted for potential sociodemographic, anthropometric, dietary and lifestyle confounders, but the co-intake of other additives including colourants and emulsifiers could have amplified the effect. As discussed in the paper, the observations are in line with other studies and there are potential mechanisms focused on glucose homeostasis and dyslipidaemia.6

The literature review discussing UPF intake in children focused on non-caloric sweetened foods and drinks and linked their intake to childhood obesity. The intake of attractive and palatable foods and drinks potentially induces hyperphagia through ‘brain reward systems’.9 In vitro and animal model studies have clearly shown effects on behaviour likely through changes in neural circuitry.

Emulsifiers (most between E-400 and E-495) and cardiovascular disease

Recent epidemiological studies have linked high intake of highly processed foods to increased risks of obesity, cardiovascular disease, cancers and type 2 diabetes.1 Based on mechanistic considerations, attention has focused on the extensive use of emulsifiers which thicken and prolong shelf life of foods.7 Most approved additive emulsifiers do not have a legal upper quantitative limit being classed as ‘quantum satis’. The large French study followed over 95 000 adults with no prior diagnosis of cardiovascular disease for a median of 7 years with at least three 24-hour food diary records. Emulsifiers were most frequently present in processed fruits and vegetables, cakes, biscuits and dairy products. Analysis accounting for a wide range of confounders showed significantly more cardiovascular disease among those consuming high amounts of five individual and two groups of commonly used food additive emulsifiers.7 As with the sweetener study cause cannot yet be ascribed to an association as there may be additional confounding by other lifestyle differences and co-intake of many other additives. However, this is yet another clarion call for more research to investigate which if any food additives, individually or in combinations, have adverse effects on health. As with AFCs and sweeteners, there is mechanistic credibility to the indirect adverse effects of emulsifiers.

High intake of emulsifiers has been associated with altered gut microbiota (dysbiosis) which, based on predictions, would lead to differences on faecal short-chain fatty acids. Reduced butyrate and increased propionate were associated with increased markers of gut inflammation.26 A review of published data from human and animal model studies connects gut dysbiosis with metabolic syndrome through the microbial production of bioactive metabolites.27 This suggests a causal link between emulsifiers, dysbiosis and adverse metabolic consequences.

Polysorbates (E-432–E-436) and polyethylene glycol (PEG—E-1521) are included as permitted emulsifier additives and are subject to a quantitative limit. Polysorbate 80 (E-433) and PEG are used in several medicinal products including AstraZeneca and Johnson & Johnson COVID-19 vaccines. There is concern that rare cases of acute allergic reactions to drugs and these vaccines are due to either PEG or polysorbate.28 The question of sensitisation occurring through ingestion of these emulsifiers in foods must be investigated and should be possible from further analysis of the French cohort study.7