Antioxidants & oxidative stress

Most reactive oxygen species are a byproduct of energy production. Electrons moving down the mitochondrial electron transport chain occasionally react with oxygen prematurely, producing superoxide, from which hydrogen peroxide, hydroxyl radical and the rest follow. Complexes I and III are the main sites ^[4]. Immune cells add more on purpose: the NADPH oxidase in a neutrophil generates a deliberate oxidative burst to kill what it has engulfed. Inflammation, ultraviolet light, pollution, smoking and heavy alcohol all raise the load.

The damage is real where it occurs. Radicals abstract hydrogen from polyunsaturated fatty acids in cell membranes, starting a self-propagating chain called lipid peroxidation; they oxidise protein side chains, which can misfold or aggregate the protein; and they modify DNA bases. The brain is unusually exposed on every count: it takes a large share of the body's oxygen, is rich in exactly the fats that peroxidise most readily, and carries modest antioxidant enzyme capacity, which is much of why oxidative damage features in every major neurodegenerative disease ^[2].

Two cautions belong here rather than at the end. Association is not direction: increased oxidative damage in a diseased brain can be a consequence of dying cells rather than the cause of their death, and disentangling the two is what the field has struggled with for thirty years ^[2]. And the quantitative claims are shakier than they read. Rates measured in isolated mitochondria depend on variables that are hard to measure in a living animal, and extrapolating from bench to body is explicitly called misleading in the primary literature ^[4].

The single most useful correction to the popular account: the major role in antioxidant defence is filled by antioxidant enzymes, not by small-molecule antioxidant compounds ^[1]. That sentence comes from the researcher who coined the term oxidative stress, and it reframes the whole supplement category.

The reason is arithmetic. An enzyme is catalytic; one molecule of superoxide dismutase processes superoxide over and over, at rates approaching the diffusion limit. A small-molecule antioxidant is stoichiometric; one molecule of vitamin E neutralises roughly one radical and is then itself a radical needing to be recycled. To match enzymatic capacity by swallowing scavengers, the concentrations required are far beyond what oral dosing achieves in tissue.

The layered system works like this. Superoxide dismutase converts superoxide to hydrogen peroxide. Catalase and the glutathione peroxidases convert that to water. Glutathione is the cell's main redox buffer at millimolar concentrations, regenerated by glutathione reductase using NADPH; the parallel thioredoxin system does related work on protein thiols ^[3]. Dietary antioxidants such as vitamin C and vitamin E sit at the edges of this and are themselves recycled by the enzymatic machinery.

Above all of it sits a control system. Keap1 senses electrophilic and oxidative stress through reactive cysteines and releases the transcription factor Nrf2, which switches on a battery of genes: glutathione synthesis, thioredoxin, detoxifying enzymes ^[3]. That is why compounds like sulforaphane count as antioxidants despite scavenging almost nothing directly. They are mild irritants that trip the switch, and the cell mounts its own response, a categorically different and more promising strategy than adding scavengers from outside ^[14].

This is also the honest frame for NAC. It is not a powerful antioxidant in its own right; it is a cysteine donor whose benefit depends on being converted to glutathione, so it works where glutathione is genuinely depleted and should not be expected to do much in cells already replete ^[15]. That is precisely why it is a first-line antidote in paracetamol overdose, where hepatic glutathione has been consumed, and why its record in general antioxidant indications is inconsistent.

Redox biology now distinguishes oxidative eustress, a physiological level of oxidant challenge required for signalling, from oxidative distress, the damaging excess ^[3]. Hydrogen peroxide is the main messenger: it is not a free radical, it is relatively stable, it crosses membranes through specific channels, and it acts by reversibly oxidising particular cysteine residues on target proteins, which is a switch as specific as phosphorylation. Cells make it deliberately through regulated NADPH oxidases.

Once that is on the table, indiscriminate scavenging stops looking obviously good. Suppressing a messenger is not the same as preventing damage, and a compound cannot tell the difference between a peroxide molecule doing signalling work and one about to oxidise a membrane lipid.

The same logic produces hormesis: a mild stress that provokes an adaptive response leaving the organism better off than baseline. Interventions that extend lifespan across model organisms, including calorie restriction and exercise, converge on increased mitochondrial respiration and a transient rise in ROS, which then induce endogenous defences; antioxidant supplements that block that signal also block the benefit, an effect named mitohormesis ^[11]. That framework is a direct challenge to the simple free radical theory of ageing rather than a footnote to it.

The theory was tested properly, and this is the part usually omitted from supplement copy.

ATBC randomised 29,133 Finnish male smokers to alpha-tocopherol, beta carotene, both, or placebo for five to eight years. Vitamin E produced no reduction in lung cancer. Beta carotene produced an 18% higher lung cancer incidence and 8% higher total mortality than placebo ^[5]. CARET independently randomised 18,314 smokers, former smokers and asbestos-exposed workers to beta carotene plus retinol or placebo, and was stopped 21 months early: 28% higher lung cancer incidence and 17% higher all-cause mortality in the active arm ^[6]. Two large, well conducted trials in the highest risk populations found the opposite of the hypothesis.

SELECT randomised 35,533 relatively healthy men to selenium, vitamin E, both or placebo. On extended follow-up, vitamin E significantly increased prostate cancer incidence, hazard ratio 1.17 ^[7]. The pooled picture is the Cochrane review of 78 randomised trials in 296,707 participants: in the 56 trials at low risk of bias, antioxidant supplements significantly increased mortality, with beta carotene and vitamin E each significantly increasing it and higher doses of vitamin A trending the same way. Vitamin C and selenium were neutral. The reviewers concluded there is no evidence supporting antioxidant supplements for primary or secondary prevention ^[8].

Mechanistic work has since offered a plausible reason rather than treating the results as noise. In mouse models of B-RAF and K-RAS driven lung cancer, dietary N-acetylcysteine and vitamin E markedly accelerated tumour progression and shortened survival, by reducing ROS and consequently reducing DNA damage and p53 activation; inactivating p53 reproduced the effect and abolished the antioxidants' contribution ^[12]. A companion study found NAC and a soluble vitamin E analogue increased lymph node metastasis in a mouse melanoma model without changing primary tumour size, acting through glutathione and RHOA signalling ^[13]. Precancerous cells appear to be constrained by oxidative stress, and relieving that constraint helps them.

None of this means dietary antioxidants are poison. It means high-dose isolated supplementation is an intervention with real risks, tested at scale, that mostly failed. The postmortem is straightforward: total antioxidant capacity is homeostatically defended and largely unresponsive to dietary loading, so oxidative damage to key biomolecules rarely changes, and manipulating endogenous defences, including by supplying weak pro-oxidants, may be more useful than consuming large doses ^[14].

The exercise literature is the cleanest demonstration of ROS as signal, because it has a measurable adaptation to blunt.

In a controlled study, 39 healthy young men trained for four weeks with or without 1,000 mg vitamin C and 400 IU vitamin E daily. Insulin sensitivity by hyperinsulinaemic euglycaemic clamp improved with training only in the group not taking the vitamins, in both previously untrained and pretrained participants. Expression of PPAR-gamma, PGC-1alpha and PGC-1beta rose only without antioxidants, as did the endogenous defence enzymes superoxide dismutase 1 and 2 and glutathione peroxidase ^[9]. Supplementation blocked the body's own antioxidant upregulation, which is precisely backwards from the intent.

A larger and longer trial randomised 54 adults to the same vitamins or placebo across 11 weeks of running. Performance outcomes did not separate: VO2max rose about 8% in both arms and shuttle-run performance improved in both. But in muscle biopsies the mitochondrial marker COX4 rose 59% in placebo and not at all in the supplemented group, and cytosolic PGC-1alpha rose 19% in placebo and fell with supplementation ^[10]. The authors' framing is the right one: cellular adaptations were hampered, this did not reach the performance tests used, and caution is warranted about combining antioxidants with endurance training. A companion analysis found long-term antioxidant and heat shock adaptations largely unaffected but acute stress responses amplified in the supplemented group, suggesting the arms reached similar endpoints by different routes ^[17].

The practical reading is narrow and worth keeping narrow. This concerns chronic high-dose isolated vitamin C and E around a training block, not eating fruit. Nobody has shown dietary antioxidant intake blunts training, and the performance effects were small to absent; what moved was the molecular signature of adaptation.

A defensible position on current evidence looks like this.

Eat the food. Diets rich in fruit and vegetables are consistently associated with better outcomes, and no trial of an isolated extracted compound has reproduced that. Whatever is doing the work in a whole diet is not the thing that was put in the capsules.

Treat targeted use as pharmacology, not nutrition. NAC in paracetamol overdose works because it repletes a specifically depleted glutathione pool ^[15]. Edaravone is a free radical scavenger approved in amyotrophic lateral sclerosis on a trial that worked only in a narrowly defined early-stage subgroup, whose authors wrote that there is no indication of effectiveness in a wider ALS population ^[16]. Both are scavengers doing real work in a specific damaged context, and neither generalises to healthy people.

Be sceptical of high-dose isolated supplementation, especially around training and especially of vitamin E and beta carotene. The mortality signal is from low risk-of-bias trials in nearly 300,000 people ^[8], and the trend across the animal work is that reducing oxidative stress can relieve a constraint on precancerous cells ^[12][13].

What is genuinely not known is substantial. It is not established whether oxidative damage causes the neurodegenerative diseases it accompanies or merely accumulates alongside them ^[2]. There is no validated way to measure a healthy person's oxidative status, and the commercial panels are not fit for it. Whether Nrf2 inducers translate into human outcomes is open, and Nrf2 activation is itself exploited by some tumours. And the failure of the trials does not falsify redox biology; it falsifies one intervention drawn from it. Balance beats maximum here, and the balance is maintained by enzymes rather than purchased. Not medical advice.