Tier 2 — strong

Broccoli for the brain: a review of the neuroprotective mechanisms of sulforaphane

Bessetti RN, Litwa KA

Frontiers in Cellular Neuroscience 2025 Volume 19, article 1601366

Bibliography

DOI: 10.3389/fncel.2025.1601366
PubMed: PMID 40686501
PubMed Central: PMC12271217
Funding: Supported by the National Institute of Environmental Health Sciences of the National Institutes of Health under Award Number P30ES025128 (pilot project funding to K.A. Litwa); the National Center for Complementary and Integrative Health and Office of the Director of the National Institutes of Health under award number 1R21AT011371-01A1 (to K.A. Litwa); and the National Science Foundation CAREER award number 2144912 (to K.A. Litwa).
Competing interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The authors also disclose that generative AI was used to sort cited clinical trials by theme in one figure of the manuscript.

Study snapshot

Design	Narrative review of preclinical (in vitro and in vivo) and clinical trial evidence.
Model	Review scope: cellular and animal models across epilepsy, autism spectrum disorder, schizophrenia, Alzheimer's disease, and Parkinson's disease; plus every published human clinical trial identified via PubMed search for "sulforaphane" and "clinical trial" (77 trials over 2000–2025), with detailed focus on 6 ASD trials, 2 schizophrenia trials, 1 depression trial, and 1 cognitive-ageing trial.
Sample	Cell culture, rodent, and human clinical trial evidence catalogued; largest single trial cited is a 108-participant ASD trial (Ou 2024).
Intervention	Sulforaphane and its precursor glucoraphanin (usually with active myrosinase, e.g. Avmacol, BroccoPhane); cell culture doses 0.01–50 μM; rodent doses 1–50 mg/kg; human doses 50–150 μmol/day in most ASD trials, up to 1700 mg/day in a schizophrenia trial.
Duration	Preclinical acute (single dose) through chronic (up to 8 weeks); clinical trials 10–36 weeks.
Endpoints	NRF2 nuclear translocation and target gene expression (NQO1, HO-1, GST, SOD, GPx, GSH); NFκB signalling and pro/anti-inflammatory cytokines (IL-6, IL-1β, TNFα, IL-10, TGF-β); oxidative stress markers (8-oxo-dG, 3-NT, 4-HNE); disease-specific behavioural and pathological outcomes (Aberrant Behavior Checklist, Social Responsiveness Scale, cognitive assays, dopaminergic neuron counts, Aβ and tau accumulation, PPI, hyperlocomotion)

What the study showed, in plain terms

This is a comprehensive review, not a new experiment. Bessetti and Litwa catalogue how sulphoraphane — the isothiocyanate produced when the precursor glucoraphanin in cruciferous vegetables is broken down by the enzyme myrosinase — protects nerve cells, animal brains, and, in early trials, human patients. The central mechanism is activation of a transcription factor called NRF2. Under normal conditions NRF2 is tagged for degradation by a protein called KEAP1. Sulphoraphane covalently modifies KEAP1's cysteine residues, freeing NRF2 to translocate to the nucleus and switch on antioxidant and detoxification genes.

The review then works through what this means for specific neurological conditions. In seizure models, sulphoraphane blunts the surge in reactive oxygen species that accompanies epileptic activity and preserves glutathione levels, reducing neuronal death. In autism spectrum disorder models, sulphoraphane reduces oxidative stress and inflammatory cytokines in the cerebellum, improves social behaviours, and, when given to pregnant mice as glucoraphanin, prevents autism-like behaviours in the pups. In schizophrenia models, it rescues cognitive deficits and dendritic spine loss caused by phencyclidine. In Alzheimer's models it reduces amyloid-β burden and tau hyperphosphorylation; in Parkinson's models it preserves dopaminergic neurons.

The clinical picture is more mixed. Since 2020, 8 of 20 published sulphoraphane trials have focused on CNS conditions — a sharp rise from the previous two decades. The most-studied condition is ASD, with six trials. Some trials show meaningful improvements in behavioural rating scales; others show no significant effect. Study design, patient age, dosing paradigm, and the preparation used (freeze-dried broccoli sprouts with active myrosinase versus other forms) all appear to matter. One 2025 trial in chronic schizophrenia used a very high dose (1700 mg/day) and reported improvement in negative symptoms after 24 weeks. Trials in Parkinson's, Alzheimer's, and ADHD are registered but not yet reporting.

The review closes on a hormesis argument: sulphoraphane's effects are biphasic. Neurons benefit at low doses (often 0.1 μM in culture) and can be harmed at concentrations that cancer cells tolerate. Any clinical use has to reckon with this narrow window, with the low oral bioavailability of sulphoraphane itself compared to glucoraphanin plus active myrosinase, and with the fact that most preclinical evidence favours prevention over treatment of established disease.

Key findings

NRF2 activation is the central mechanism. Sulphoraphane covalently modifies KEAP1 cysteine residues, freeing NRF2 to translocate to the nucleus and drive transcription of antioxidant and phase II detoxification genes including NAD(P)H quinone oxidoreductase 1, heme oxygenase-1, glutathione-S-transferase, and superoxide dismutase.
The oxidative stress and neuroinflammation pathways converge. NF-κB drives pro-inflammatory cytokine production (TNFα, IL-1β, IL-6) and reinforces the M1 microglial phenotype; NRF2 activation counteracts this and, via a reciprocal loop with PPARγ, promotes the anti-inflammatory M2 phenotype.
Preclinical evidence favours prevention over treatment. Across cell culture, animal, and clinical work catalogued, sulphoraphane's strongest effects appear when administered before or alongside the neuronal insult rather than after established pathology.
Dose response is biphasic and cell-type-specific. Neurons in culture benefit from sulphoraphane as low as 0.01 μM, with 0.1 μM often optimal; higher doses that cancer cells tolerate can be cytotoxic to neurons. Microglia and astrocytes tolerate slightly higher doses but still benefit from the low end of the range.
Six ASD clinical trials have been published, with mixed results. Singh 2014 (young adult men, 18 weeks, 50–150 μmol/day broccoli sprout extract) showed significant improvements in Aberrant Behavior Checklist and Social Responsiveness Scale. Bent 2018 (open-label, ages 7–21, Avmacol) showed improvements in ABC social withdrawal and SRS communication and motivation subscales. Ou 2024 (108 participants, ages 3–15) showed no caregiver-rated improvement but significant clinician-rated improvement on the OSU-DSM-IV rating scale. Magner 2023 and Zimmerman 2021 showed no significant total-score improvement, though Zimmerman reported reductions in circulating IL-6 and TNFα.
Bioavailability varies dramatically by preparation. Freeze-dried broccoli sprouts with active myrosinase, or juice with pre-hydrolysed glucoraphanin, deliver the highest bioactive sulphoraphane. Heat-processed products lacking myrosinase require higher doses.
Sulphoraphane is generally well tolerated at doses used in most trials. Reported side effects at higher doses are mild and gastrointestinal; lethargy, hypothermia, and gastrointestinal distress can occur at very high doses.
Sulphoraphane crosses the blood-brain barrier and is detected in neuronal tissue after intraperitoneal injection in mice, confirmed in NRF2 knockout studies that also confirm NRF2 is required for most of its neuroprotective effects.

What this study can and cannot tell us

This is a narrative review, not a systematic review or meta-analysis. There is no pre-registered search strategy, no formal risk-of-bias assessment, and no quantitative synthesis. Readers should treat the conclusions as informed synthesis rather than statistical summary.

The clinical evidence base for neurological indications is small and heterogeneous. As of publication, only ten trials had reported results across all CNS conditions catalogued (six ASD, two schizophrenia, one depression, one cognitive ageing in older adults). Meaningful comparison is limited by differences in patient age, dosing paradigm, delivery form, and outcome measures. No published trials yet exist for epilepsy, Parkinson's disease, or Alzheimer's disease in humans, though registered trials for PD (NCT05084365), AD (NCT04213391), and ADHD (NCT06594536) are underway or recruiting.

Many preclinical studies administer sulphoraphane simultaneously with the induced neuronal insult — a design that inflates apparent efficacy relative to the clinical scenario, where patients present with established disease. The review acknowledges this and argues sulphoraphane's strongest case is preventive rather than therapeutic, but that argument is difficult to test in adult clinical trials.

Hormesis complicates dose selection. The dose that benefits neurons in one model may be inert or harmful in another, and translation from cell culture concentrations to whole-organism doses is confounded by sulphoraphane's rapid metabolism and short half-life. This is why glucoraphanin plus active myrosinase is preferred in most clinical trials — but even then, the pharmacokinetic profile is inconsistent across preparations.

The authors disclose the use of generative AI to sort cited clinical trials by theme in one figure of the manuscript. This is a transparency note rather than a methodological flaw, but readers should be aware of it.