Sulforaphane & Blood Sugar: Diabetes, Prediabetes and Human RCTs

Sulforaphane is not a proven glucose-lowering treatment. Human randomized evidence is mixed: a 2017 type 2 diabetes trial reported its strongest benefit in a metabolically dysregulated subgroup, a 2025 prediabetes trial missed its prespecified primary glucose-reduction target despite exploratory responder signals, and a 2026 trial in people with diabetes and advanced chronic kidney disease found no significant glycemic benefit.

The useful question is therefore not “does sulforaphane lower blood sugar?” as a universal yes/no claim. It is which population, which formulation, which endpoint and which prespecified analysis?

The three trials that define the current glucose story

The 2017 type 2 diabetes trial: where the excitement started

The 2017 translational study first used molecular and animal work to identify sulforaphane as a candidate suppressor of excessive hepatic glucose production, then tested a concentrated broccoli sprout extract in 97 adults with type 2 diabetes.^[1]

The intervention delivered 150 µmol/day of sulforaphane for 12 weeks. The strongest clinical signal occurred in participants with obesity and dysregulated type 2 diabetes, where fasting glucose and HbA1c-related measures improved.

What the study supports: sulforaphane may influence hepatic glucose regulation in some metabolic phenotypes.

What it does not support: that every person with diabetes will respond, that lower-dose retail supplements reproduce the intervention, or that sulforaphane replaces standard treatment.

The 2025 prediabetes RCT: promising subgroup, failed prespecified target

A 2025 randomized double-blind placebo-controlled trial studied 74 drug-naive adults with prediabetes for 12 weeks.^[2]

The prespecified primary goal was a 0.3 mmol/L fasting-glucose reduction versus placebo. The trial did not meet that target in the overall population. The average between-group effect was smaller.

Exploratory analyses then identified a subgroup with particular metabolic characteristics and a distinct baseline gut microbiota that appeared to respond more strongly.

This is scientifically interesting because it raises a precision-nutrition hypothesis: baseline metabolic phenotype and microbiome may influence response. But exploratory subgroup findings are not yet validated treatment-selection rules.

The 2026 diabetes + CKD trial: a clean null result

A 2026 multicenter randomized double-blind trial enrolled 99 adults with type 2 diabetes and chronic kidney disease stages 3b–4.^[3]

After the intervention, fasting glucose, insulin and HbA1c did not significantly improve relative to placebo. The null result persisted in both intention-to-treat and per-protocol analyses and remained null at follow-up.

This does not erase the 2017 subgroup signal because the populations were clinically different. It does show why “sulforaphane lowers blood sugar” is too broad a claim.

Why might sulforaphane affect glucose metabolism?

Mechanistic research suggests several possible routes:

• reduced hepatic gluconeogenic gene expression;
• NRF2-related redox and stress-response signaling;
• effects on insulin-signaling pathways;
• possible interactions between metabolic phenotype, inflammation and the gut microbiome.

These mechanisms provide biological plausibility. Clinical efficacy still depends on reproducible human outcomes.

What does the microbiome finding mean?

The 2025 prediabetes trial is important because it suggests the microbiome may be one determinant of who converts or responds to broccoli-derived compounds. That fits the broader sulforaphane literature: glucoraphanin can be converted by gut bacteria when active plant myrosinase is absent, and conversion varies substantially between people.

But two separate questions should not be merged:

1. Bioavailability: how much sulforaphane is generated and absorbed?
2. Clinical response: does that exposure improve glucose regulation in this person?

See our sulforaphane bioavailability guide for the first question.

Does formulation matter for metabolic trials?

Yes. Human sulforaphane studies use concentrated broccoli sprout extracts, glucoraphanin systems and different reported dose units. The 150 µmol/day intervention in the 2017 diabetes study is not equivalent to “150 mg broccoli extract” on a retail label.

Our sulforaphane dosage guide explains µmol-to-mg conversion and why precursor weight is different from delivered sulforaphane.

Can sulforaphane replace metformin, insulin or GLP-1 therapy?

No. The current evidence does not establish sulforaphane as a replacement for metformin, insulin, GLP-1 receptor agonists, SGLT2 inhibitors or other evidence-based glucose-lowering therapy.

Stopping or reducing medication based on a supplement study can create real clinical risk. Medication changes should be made with the clinician managing the diabetes.

What if someone has kidney disease?

The 2026 CKD trial is directly relevant here because advanced kidney disease changes metabolism, medication burden and clinical risk. Its glycemic result was null.^[3]

People with CKD should not infer safety or benefit from trials conducted in otherwise healthier populations. Concentrated supplements should be reviewed with the nephrology or diabetes team.

Does eating broccoli help blood sugar?

Cruciferous vegetables are nutrient-dense foods and can fit into a diabetes-friendly dietary pattern. But the clinical trials discussed here used defined, concentrated interventions and should not be translated into claims that a serving of broccoli reproduces the same pharmacologic exposure.

For food sources and preparation, see Foods High in Sulforaphane.

Bottom line

The human glucose evidence for sulforaphane is mixed and phenotype-dependent. The 2017 type 2 diabetes study produced a meaningful subgroup signal. The 2025 prediabetes RCT missed its prespecified primary target and generated an exploratory responder hypothesis. The 2026 diabetes/CKD RCT was null for fasting glucose, insulin and HbA1c.

That makes sulforaphane an interesting metabolic research compound—not an established diabetes treatment.