The Insulin Sensitivity Protocol
A layered supplement protocol for insulin resistance, built in the order the evidence supports and graded to its weakest load-bearing layer. Layer 0 is measurement (fasting glucose and insulin for HOMA-IR, HbA1c, optional CGM) plus magnesium repletion. Layer 1 is berberine, the supplement analogue of metformin. Layer 2 is a fork — myo-inositol for women with PCOS, alpha-lipoic acid for people with neuropathy symptoms or no PCOS. Layer 3 is optional chromium, honestly labelled as useful only in established diabetes. Includes the statement that diet, resistance training, sleep and weight loss outperform every layer, the hypoglycaemia rules for anyone on medication, and the 12-week HbA1c judgement point.
Daily Schedule
Timing and dosage for each step
Layer 0 — evening with food, from week 0
2000 mg
~2 g threonate (≈144 mg elemental). Any well-absorbed form works for the metabolic role — citrate or glycinate at 300–400 mg elemental is the more usual trial dose. Meta-analysis of 10 treatment arms: HOMA-IR −0.67 versus placebo, with the effect confined to trials of ≥ 4 months. This is repletion, not pharmacology, and it is the only layer that continues regardless of what the 12-week bloods show.
Layer 1 — with each main meal, from week 2
1500 mg
500 mg three times daily with meals (or 500 mg liposomal once daily). Meta-analysis of 37 RCTs in type 2 diabetes: HbA1c −0.63%, fasting glucose −0.82 mmol/L, with no increase in hypoglycaemia. Start at 500 mg once daily for a week — a third of patients get transient GI effects. ⚠️ CYP3A4 / P-gp inhibitor: raises cyclosporine, tacrolimus and some statins. ⚠️ Additive glucose lowering with insulin and sulfonylureas.
Layer 2A — morning and evening, from week 4 (women with PCOS)
4000 mg
2 g twice daily as powder, ideally at a 40:1 myo:D-chiro ratio. Nine-trial meta-analysis in PCOS: fasting insulin SMD −1.02, HOMA index −0.59; 26-trial meta-analysis: 1.79× the odds of a regular cycle versus placebo, non-inferior to metformin. Choose THIS fork if you have PCOS. Outside PCOS the outcome data are thin, and the 2025 JAMA pregnancy trial in PCOS was null.
Layer 2B — 30 minutes before breakfast, from week 4 (neuropathy symptoms, or anyone not a candidate for 2A)
600 mg
600 mg racemic, empty stomach — food reduces absorption. The dose that improved neuropathy symptoms by ~50% versus 32% on placebo in SYDNEY 2; higher doses added nausea and vertigo without benefit. For glucose the best pooled data show HOMA-IR −0.48 and NO change in HbA1c. ⚠️ Stop immediately at any unexplained hypoglycaemia — insulin autoimmune syndrome is a documented ALA trigger in HLA-DRB1*04:03 / *04:06 carriers. ⚠️ Additive glucose lowering with berberine, insulin and sulfonylureas.
Layer 3 — optional, with breakfast, from week 4 (established type 2 diabetes only)
400 mcg
200–400 mcg picolinate. Meta-analyses in type 2 diabetes: HbA1c −0.55% to −0.71%, clearer above 200 mcg and in poor baseline control. In people WITHOUT diabetes: no effect on clamp-measured insulin sensitivity in a 16-week RCT, and the 2002 meta-analysis found nothing. Grade D outside diabetes. Skip it without regret unless your HbA1c is already diabetic.
Read This Before the Protocol
Diet, resistance training, sleep and weight loss outperform every supplement on this page, and the trial that proves it is old enough to have stopped being controversial. In the Diabetes Prevention Program, 3,234 people with impaired glucose tolerance were randomised to placebo, metformin 850 mg twice daily, or a lifestyle programme targeting 7% weight loss and 150 minutes of weekly activity. Over 2.8 years, lifestyle cut diabetes incidence by 58%; metformin by 31%. Lifestyle beat the drug. No supplement has ever been put in that trial, and none of them would have beaten metformin.
That sets the tiers. Metformin is the drug tier. Berberine is its supplement analogue — same AMPK pathway, comparable HbA1c effect in the head-to-head data, which is covered in full in Berberine vs. Metformin. Everything below berberine in this protocol is weaker than berberine.
Two populations need to read the safety section first. Anyone on insulin or a sulfonylurea is stacking hypoglycaemic agents when they add berberine or ALA, and that is a real risk rather than a label warning — it needs a prescriber's involvement. And women with PCOS are the group with the best supplement data in this whole space, because myo-inositol has been tested against placebo and against metformin in that population specifically, and they should take the inositol fork rather than the ALA one.
Finally: an HbA1c of 6.5% or above is diabetes. That is a diagnosis under the American Diabetes Association standards, not a stack. If that is where your baseline lands, this protocol is an adjunct to a clinician's plan and not a substitute for it.
The Layer Logic
| Layer | Purpose | Evidence grade | Optional? |
|---|---|---|---|
| 0 — Measure + magnesium | Baseline HOMA-IR and HbA1c; replete the cofactor | B (magnesium) | No |
| 1 — Berberine | The metformin analogue | B | No |
| 2A — Myo-inositol | PCOS insulin sensitiser | B in PCOS, C elsewhere | Fork |
| 2B — Alpha-lipoic acid | Neuropathy symptoms; weak glucose effect | C | Fork |
| 3 — Chromium | Trace layer | C in diabetes, D without | Yes, freely skippable |
The protocol is graded C — to its weakest load-bearing layer. Berberine and magnesium on their own are a solid B. The long-form reasoning, the 12-week schedule and the HOMA-IR worked example are in The Insulin Sensitivity Stack.
Layer 0 — Measure First, Then Magnesium
You cannot judge an insulin-sensitivity protocol without a baseline that includes insulin. Fasting glucose alone is the last number to move; by the time it rises, insulin has been elevated for years.
HOMA-IR is fasting glucose multiplied by fasting insulin, divided by a constant (405 in mg/dL units, 22.5 in mmol/L). It correlated with the euglycaemic clamp at Rs = 0.88 in the original 1985 validation. It also has a coefficient of variation around 31%, which is why a single reading is a screening number and not a verdict — and why this protocol is judged at 12 weeks on HbA1c, with HOMA-IR as the supporting trend.
A continuous glucose monitor is optional and useful for people who want to see post-meal excursions rather than averages; the measurement-led approach is set out in the CGM and berberine article, which is the companion to this protocol rather than a duplicate of it.
Magnesium goes in at week 0 because it is repletion, not pharmacology. Meta-analysis of randomised trials found supplementation lowered HOMA-IR by 0.67 versus placebo (95% CI −1.20 to −0.14), with the effect confined to trials lasting at least four months — so it is the one layer that runs the whole 12 weeks and beyond, regardless of what else is judged.
Layer 1 — Berberine
Berberine is the layer with the trial base. A meta-analysis of 37 randomised trials in 3,048 people with type 2 diabetes found it lowered HbA1c by 0.63% (95% CI −0.72 to −0.53), fasting glucose by 0.82 mmol/L, and two-hour post-load glucose by 1.16 mmol/L, without increasing hypoglycaemia (RR 0.48, 95% CI 0.21–1.08). In the original 2008 pilot that compared it directly with metformin 500 mg three times daily, both arms took HbA1c from 9.5% to 7.5% over three months.
Dose is 500 mg three times daily with meals, or 500 mg of a liposomal preparation once daily. Start at one dose a day for the first week: in the 2008 trial, 34.5% of patients had transient gastrointestinal effects.
The two interactions that matter: berberine inhibits CYP3A4 and P-glycoprotein, raising levels of cyclosporine, tacrolimus and several statins; and its glucose lowering is additive with insulin, sulfonylureas and — inside this protocol — alpha-lipoic acid.
Layer 2 — The Fork
This is where the protocol splits, and the split is by population rather than preference.
Take 2A — myo-inositol — if you have PCOS. The evidence is population-specific: nine randomised trials pooled in 2017 found fasting insulin fell (SMD −1.02, P = 0.009) and HOMA index fell (SMD −0.59, P = 0.041); 26 trials pooled in 2023 found 1.79 times the odds of a regular cycle versus placebo and non-inferiority to metformin. Dose is 2 g twice daily, ideally at the 40:1 myo:D-chiro ratio that restored ovulation best in a seven-arm comparison. Full profile in Myo-Inositol: The Insulin Sensitiser for PCOS and Beyond.
Take 2B — alpha-lipoic acid — if you have neuropathy symptoms, or if you are not a candidate for 2A. Be clear about what you are buying. For neuropathy, 600 mg/day cut symptom scores by 51% versus 32% on placebo in five weeks, with a 62% responder rate versus 26%. For glucose, the most rigorous meta-analysis found HOMA-IR fell by 0.48 and HbA1c did not change. That is a C, and it is why 2B is the weaker fork. The trial series and the autoimmune hypoglycaemia risk are in Alpha-Lipoic Acid: Glucose Disposal, Neuropathy and the Trials That Matter.
Do not run both. There is no trial of the combination, and stacking three glucose-lowering agents in someone who is medication-free is unnecessary; in someone on medication it is unwise.
Layer 3 — Chromium, Labelled Honestly
Chromium picolinate is optional and should be understood as useful only in established type 2 diabetes. Meta-analyses in that population put the HbA1c effect at −0.55% to −0.71%, clearer above 200 mcg/day and in people with poor baseline control. In people without diabetes, a 16-week clamp study found no effect on insulin sensitivity (P = 0.83), and the 2002 meta-analysis found nothing in 425 non-diabetic participants.
200–400 mcg with breakfast if your HbA1c is already in the diabetic range and you are working with a clinician. Otherwise skip it. It is graded D outside diabetes for a reason.
Interactions Inside the Stack
- Berberine + ALA — both lower glucose by different mechanisms. Additive hypoglycaemia is the concern in anyone also on insulin or a sulfonylurea. In a medication-free person the combination is reasonable with fasting-glucose monitoring.
- Berberine + any CYP3A4 / P-gp substrate — cyclosporine, tacrolimus, simvastatin, some calcium-channel blockers. Check the interaction checker before adding it to a prescription list.
- ALA + insulin autoimmune syndrome — not a drug interaction but a stop rule: any unexplained hypoglycaemia on ALA means stop ALA and measure insulin and insulin antibodies.
- Chromium + levothyroxine — reduced thyroxine absorption when co-ingested; separate by several hours.
When to Stop and See a Clinician
- Baseline HbA1c ≥ 6.5%. Diabetes is a diagnosis. Run this as an adjunct under a clinician, not alone.
- Any hypoglycaemic episode — shakiness, sweating, confusion, a reading below 70 mg/dL (3.9 mmol/L). Stop berberine and ALA the same day.
- You are on insulin, a sulfonylurea or a glinide. The stack needs a prescriber's sign-off before week 2.
- Neuropathy symptoms — burning, numbness, stabbing pain in the feet. ALA treats the symptom; the cause needs an HbA1c and a foot examination.
- HbA1c has not moved at 12 weeks despite adherence. That is the signal to escalate to the drug tier, not to add a fourth supplement.
Monitoring
| Marker | When | Why |
|---|---|---|
| Fasting glucose + fasting insulin (HOMA-IR) | Baseline, 12 weeks | The insulin-sensitivity signal; CV ~31%, so read the trend |
| HbA1c | Baseline, 12 weeks | The judgement number; ≥ 6.5% is diabetes |
| Serum magnesium (or RBC magnesium) | Baseline | Confirms Layer 0 is repletion rather than decoration |
| Lipids incl. triglycerides | Baseline, 12 weeks | Berberine moves these too; a second signal it is working |
| Home fasting glucose | Weekly from week 4 | Catches additive hypoglycaemia on the berberine + ALA fork |
| ALT / AST, creatinine | Baseline, 12 weeks | Berberine and chromium both carry rare hepatic / renal reports |
| Optional CGM | Weeks 0–2 and 10–12 | Post-meal excursions and variability, per the CGM article |
Safety and Scope
This protocol contains no prescription agents. It is not a treatment for diabetes, it does not replace metformin where metformin is indicated, and it does not replace the weight-loss and activity targets that outperformed metformin in the only trial that tested them head to head. Pregnant women should not use berberine or ALA; myo-inositol in pregnancy is a clinician conversation given the 2025 null trial in PCOS. Anyone with kidney disease should skip chromium.