you can’t tell but under this pile of chopped veggies (maybe 50 cal), peanuts (100 cal), and freshly grated ginger is half a block of silken tofu (110 cals). Peaking through: peanut butter-chili crisp sauce (150 cals). Sometimes I swap chili crisp for sriracha to knock off 50 cals. About 20g protein, plus fiber and healthy fats. Super satisfying! Sometimes I add cucumber, carrot, daikon, celery, or use a sesame oil dressing 😋

Free gift link to recipe

I'm a physician. I see metabolic syndrome and prediabetes every week, and I'm tired of patients arriving with shopping bags of supplements they bought because an influencer with a ring light told them to (I see one of them at least once a month now). So here is what the actual literature supports for improving HOMA-IR, fasting glucose, HbA1c, and triglycerides, graded by RCT and meta-analysis evidence, not mechanism handwaving.

This is not personalized medical advice. It's an evidence review. Some links below are affiliate (DoNotAge, where their lineup happens to align with the literature); other recommendations are brand-agnostic and I have no financial relationship with them. The pharmacology is independent of point of purchase.

A note before any of this: pharmacotherapy comes first when indicated

If you meet criteria for prediabetes (HbA1c 5.7–6.4%, fasting glucose 100–125 mg/dL) or T2D, the first-line interventions with the largest effect sizes are not on this list. They are:

• Metformin: decades of safety data, prevents progression to T2D in the Diabetes Prevention Program by ~31%, costs about $4/month
• GLP-1 receptor agonists (semaglutide, tirzepatide): the largest single intervention effect on metabolic disease in the modern pharmacopeia, both for glycemia and weight
• SGLT-2 inhibitors in the right candidate (especially with comorbid HFrEF or CKD)

If you are a candidate for these, take them. Supplements are an adjunct for patients who cannot or will not start pharmacotherapy, who are still pre-prediabetic, or who want to optimize metabolic markers alongside drug therapy. Do not let supplement enthusiasm delay an indicated metformin script. Patients who progress from prediabetes to T2D over years of "I'll try diet and supplements first" are common enough that I now address it explicitly at the first visit.

With that established:

The stack

1. Berberine: first-line, full stop

The most evidence-backed nutraceutical for insulin resistance currently available. The 2025 Frontiers in Pharmacology meta-analysis of placebo-controlled RCTs in metabolic syndrome demonstrated reductions in fasting plasma glucose (WMD −0.52 mmol/L), triglycerides (−0.37 mmol/L), LDL-C (−0.50 mmol/L), total cholesterol (−0.45 mmol/L), and waist circumference (−3.27 cm) (source). A separate meta-analysis of 46 RCTs in T2D reported reductions in HbA1c (−0.73%), fasting glucose (−0.86 mmol/L), and HOMA-IR (−0.71) (source).

Mechanism: AMPK activation, the same pathway as metformin. Effect size is comparable to a low-dose oral hypoglycemic in some trials.

Clinical considerations:

• Bioavailability ~1%. Single daily dosing is ineffective. Split TID with meals.
• CYP3A4 and P-glycoprotein inhibitor. Significant interactions with statins (especially simvastatin, atorvastatin), calcineurin inhibitors, DOACs, macrolides, and others. Review the full medication list before recommending.
• GI tolerability is rate-limiting. Titrate from 500 mg daily upward over 1–2 weeks.
• Do not co-administer with metformin without close glucose monitoring.
• Pregnancy: contraindicated (placental crossing, bilirubin displacement from albumin).

Dosing: 500 mg three times daily with meals.
Source: DoNotAge Pure Berberine (affiliate).
Brand-agnostic alternative: any berberine HCl product third-party tested for purity. Dihydroberberine is a more bioavailable derivative if cost is not a concern.

2. Psyllium husk — the most underused intervention in this entire space

I am putting this near the top deliberately. The evidence is among the strongest of anything discussed here, and it costs roughly $0.20 per dose at any pharmacy.

The 2015 meta-analysis of 35 RCTs (Gibb et al., Am J Clin Nutr) demonstrated that psyllium dosed before meals in patients with T2D reduced fasting blood glucose by 37 mg/dL (p<0.001) and HbA1c by 0.97% (source). A 2023 dose-response meta-analysis on viscous soluble fiber including psyllium confirmed HbA1c reduction of −0.47% along with LDL-C reduction of −0.24 mmol/L (source). A 2021 overview of medicinal-plant meta-analyses for T2D placed psyllium in the top three botanical interventions for HbA1c reduction (−0.97%), comparable to aloe vera and fenugreek (source).

A 0.97% HbA1c reduction is pharmaceutical-tier. Most oral hypoglycemics deliver 0.5–1.0%. The fact that this is sold as a constipation supplement next to the prune juice is a market failure.

Mechanism: viscous gel formation slows gastric emptying and carbohydrate absorption, blunts postprandial glucose excursion, sequesters bile acids (driving the LDL reduction), and lowers caloric absorption.

Clinical considerations:

• Must be taken before meals, not after, with adequate water (~250 mL minimum). Post-meal dosing loses most of the effect.
• Begin at 3.4 g (one teaspoon) before one meal daily, titrate to 5–10 g/day divided. Higher doses produce more effect but more bloating in week 1.
• Separate from oral medications by 2 hours — psyllium can reduce absorption of levothyroxine, lithium, carbamazepine, and others.
• Contraindicated in known bowel obstruction or significant dysmotility.

Dosing: 5–10 g/day total, divided before meals.
Source: Generic psyllium husk powder (Metamucil sugar-free, NOW Foods, Yerba Prima, etc.). Avoid sugar-sweetened versions. Buy whatever is cheap and pure.

3. Sulforaphane (glucoraphanin + active myrosinase) — targets hepatic gluconeogenesis

Axelsson et al.'s 2017 RCT in Science Translational Medicine (n=97, obese T2D) demonstrated that broccoli sprout extract reduced fasting glucose and HbA1c with effect sizes comparable to metformin specifically for hepatic glucose output, mediated through NRF2 translocation and downregulation of gluconeogenic enzymes including PEPCK (source). The 2025 Nature Microbiology RCT in prediabetics confirmed a smaller but statistically significant fasting glucose reduction (−0.2 mmol/L, 95% CI −0.44 to −0.01, p=0.04), with the critical finding that response is gut microbiome–dependent. Non-responders lacked the Bacteroides-encoded transcriptional regulator required to convert glucoraphanin to bioactive sulforaphane (source).

Critical formulation point: Sulforaphane itself is unstable. Supplements containing only glucoraphanin without active myrosinase rely entirely on intestinal microbial conversion, which fails in roughly 25% of patients. A formulation combining glucoraphanin with exogenous myrosinase (typically from radish) bypasses this. Without myrosinase, the supplement is pharmacologically inert in a substantial subset of patients. Most retail sulforaphane products fail this test. Read the label.

Mechanism is complementary to berberine, not redundant — berberine acts peripherally (skeletal muscle AMPK), sulforaphane acts hepatically (NRF2-mediated suppression of gluconeogenesis).

Clinical considerations:

• NRF2 activation has theoretical interactions with cytotoxic chemotherapy. Avoid in active oncology patients without consultation.
• GI side effects in clinical trials were mild and self-limiting.

Dosing: 200–460 mg daily (of standardized extract), divided, with food.
Source: DoNotAge SulforaBoost (affiliate).
The non-affiliate alternative I will name is Avmacol or Prostaphane; both have been used in clinical trials. Avoid any product that lists "broccoli extract" without specifying glucoraphanin content and myrosinase.

4. Omega-3 (EPA/DHA) — for the triglyceride and inflammatory components

The 2025 Nutrients meta-analysis (21 RCTs in MetS) is unambiguous: marine omega-3 produces substantial triglyceride reduction at doses >2000 mg/day for ≥8 weeks (source). The 2025 Food Science & Nutrition dose-response meta-analysis confirms this (SMD −0.25 for triglycerides) (source).

Clinical considerations:

• The same meta-analysis flagged a small but real LDL-C increase, particularly at lower doses. In statin-treated patients with controlled LDL, monitor on follow-up panels.
• Effects on fasting glucose are essentially null. This is a triglyceride and inflammation intervention, not a glycemic one.
• Subtherapeutic dosing is the most common error. 1 g "fish oil" softgels containing 300 mg EPA+DHA will not produce trial-level outcomes. Total combined EPA+DHA must be ≥2 g/day.
• Mild antiplatelet effect; relevant in patients on anticoagulation or pre-operatively.
• For severe hypertriglyceridemia (>500 mg/dL), prescription icosapent ethyl (Vascepa) has the strongest CV-outcome data (REDUCE-IT). OTC fish oil is not equivalent.

Dosing: 2–3 g combined EPA+DHA daily with meals. Source:
DoNotAge Pure Omega 3 (affiliate).
Brand-agnostic alternatives: Nordic Naturals ProOmega, Carlson, or any IFOS-certified product. Check the label for actual EPA+DHA mg, not total fish oil mg.

5. Magnesium — usually deficient, almost always worth supplementing

Magnesium is a cofactor in over 300 enzymatic reactions including those of insulin signaling. The Simental-Mendía et al. meta-analysis of 18 RCTs found magnesium supplementation for ≥4 months significantly improved both fasting glucose and HOMA-IR (HOMA-IR WMD −0.67, 95% CI −1.20 to −0.14) (source). A 2026 prediabetes-specific meta-analysis showed improvements in 2-hour OGTT glucose (MD −0.99 mmol/L, p<0.00001), HOMA-IR (MD −1.10, p=0.03), and triglycerides (MD −14.57 mg/dL, p=0.04) (source). The 2022 Frontiers in Nutrition pooled analysis of 24 RCTs in T2D confirmed HbA1c reduction with magnesium supplementation (source).

Effect is most pronounced in patients with hypomagnesemia or low dietary intake, which describes most of the modern Western population eating refined grains and minimal leafy greens.

Clinical considerations:

• Form matters. Magnesium oxide is poorly absorbed (~4%). Use magnesium glycinate, citrate, or malate. Threonate is reasonable for sleep/cognitive applications but more expensive.
• The DoNotAge D3/K2/Mg combo product may not contain enough elemental magnesium for therapeutic effect — check the label. Most adults need 200–400 mg of elemental magnesium daily.
• Caution in advanced CKD (eGFR <30); risk of hypermagnesemia.
• Mild GI laxative effect, especially with citrate. Glycinate is better tolerated.

Dosing: 200–400 mg elemental magnesium daily, evening (often improves sleep onset). Source: Brand-agnostic. Pure Encapsulations, Doctor's Best, Klaire Labs, Thorne, any third-party tested glycinate or citrate. The DoNotAge D3/K2/Mg combo provides some, but most patients will need a separate magnesium product.

6. Vitamin D3 / K2, repletion, not supplementation

The 2021 Nutrients meta-analysis (29 RCTs, n=3,792) demonstrated improvements in fasting glucose (SMD −0.38), HbA1c (SMD −0.14), and fasting insulin in prediabetics on vitamin D (source). The 2018 meta-analysis (28 RCTs, n=3,848) reported HOMA-IR reduction of −0.39 (source). However, the 2025 umbrella review correctly identified that benefit is concentrated in deficient and insufficient individuals, with high heterogeneity in replete cohorts (source).

This is a deficiency-correction intervention, not a tonic. Order a 25(OH)D before recommending. If the patient is >40 ng/mL (>100 nmol/L), additional supplementation will not improve insulin sensitivity. Supraphysiologic dosing has no metabolic benefit and creates risk of hypercalcemia in chronic use.

K2 (MK-7) directs calcium deposition away from vascular tissue and pairs logically with D3.

Dosing: 2,000–4,000 IU D3 daily for repletion if 25(OH)D <30 ng/mL. Recheck at 8–12 weeks.
Source: DoNotAge Pure D3, K2 & Magnesium (affiliate) covers all three but verify magnesium adequacy (see above).
Brand-agnostic: Thorne D/K, Pure Encapsulations.

7. Creatine monohydrate, adjunct in patients who train

The 2025 Nutrients review on creatine and T2D prevention is clear: in combination with resistance training, creatine increases GLUT4 translocation, augments muscle glycogen storage, and supports glycemic control while attenuating sarcopenia (source). Skeletal muscle is the dominant site of postprandial glucose disposal. Preserving and increasing muscle mass is one of the most underappreciated interventions in metabolic medicine.

Creatine without exercise produces minimal glycemic effect. Creatine with structured resistance training is one of the most cost-effective interventions in this entire stack.

Clinical considerations:

• Renal function: long-standing concerns are unsupported in patients with normal baseline creatinine clearance, but recheck in patients with CKD before initiating.
• Serum creatinine will rise modestly on supplementation; this reflects metabolic load, not renal injury. Use cystatin C if you need an unbiased GFR estimate.
• No loading phase is required. 5 g daily produces saturation within 3–4 weeks.

Dosing: 5 g daily.
Source: DoNotAge Creatine Monohydrate (affiliate).
Honestly any Creapure-certified monohydrate is identical. Bulk Supplements, Thorne, Optimum Nutrition. Do not pay for "advanced" forms (HCl, ethyl ester, buffered) — monohydrate is the only form with the trial data.

Indication-specific additions

Inositol (myo-inositol + D-chiro-inositol, 40:1 ratio), for PCOS-driven insulin resistance

If insulin resistance is occurring in the context of PCOS, inositol becomes a targeted intervention with evidence approaching first-line status. The 2023 systematic review informing the international PCOS guidelines reviewed 30 trials (n=2,230) and found benefit on metabolic and ovulatory outcomes, with myo-inositol producing fewer GI adverse events than metformin (source). The Benelli et al. RCT specifically demonstrated significant reductions in LH, free testosterone, fasting insulin, and HOMA index with combined MI+DCI 40:1 versus placebo over 6 months (source).

The 40:1 MI:DCI ratio recapitulates physiologic plasma ratios. Single-isomer DCI at high doses paradoxically worsens ovulatory outcomes and should be avoided.

Dosing: 2 g myo-inositol + 50 mg D-chiro-inositol twice daily.
Brands Ovasitol (Theralogix) is the most-studied product. Wholesome Story is a cheaper alternative with the same ratio.

Alpha-lipoic acid — defensible for diabetic neuropathy, mixed for HOMA-IR

The 2020 dose-response meta-analysis of 28 RCTs found ALA reduced fasting insulin and HOMA-IR (WMD −0.48, p=0.002) but did not consistently move HbA1c (source). A 2019 meta-analysis of 41 RCTs found benefit on HbA1c, FBG, and inflammatory markers but not HOMA-IR (source). Heterogeneity is high.

The clearest indication for ALA is established diabetic peripheral neuropathy, where multiple trials and a Cochrane-level signal support symptomatic improvement. For uncomplicated insulin resistance, the evidence is weaker than the prior items in this list. Reserve for patients with neuropathic symptoms.

Dosing: 600 mg daily (R-ALA preferred over racemic). Source: Brand-agnostic. Doctor's Best R-Lipoic Acid, Jarrow.

What I am explicitly not including, and why

The longevity supplement market has bled into the metabolic supplement market and the resulting noise has been actively unhelpful in clinic. The following have no place in an insulin resistance stack on current evidence:

• NMN and NR. Mouse data is impressive. Human RCTs for insulin sensitivity are small, short, and inconsistent. There is no meta-analytic signal for HOMA-IR or HbA1c. If patients want to take NAD+ precursors for other reasons, that is their decision; it does not belong in a metabolic stack on the basis of current evidence. (NMN / NR for those proceeding anyway.)
• Resveratrol. Bioavailability is poor and three decades of human trials have not delivered a consistent metabolic signal. The discontinuation of the Sirtris program at GlaxoSmithKline tells you what large-scale due diligence concluded.
• Probiotics. Strain- and indication-specific. Without an RCT for the exact formulation in metabolic syndrome, generic probiotic recommendations fail on evidence grounds. Dietary fiber diversity will outperform any capsule.
• Cinnamon, chromium, gymnema. Trialed extensively, signal is weak to absent. Sometimes mentioned for completeness; I do not recommend them.

Administration schedule

• Pre-breakfast (15 min before): Psyllium 3.4 g in water
• Breakfast: Berberine 500 mg; Omega-3; Vitamin D3/K2; SulforaBoost ×1
• Pre-lunch: Psyllium 3.4 g in water
• Lunch: Berberine 500 mg
• Any time: Creatine 5 g
• Pre-dinner: Psyllium 3.4 g (optional third dose)
• Dinner: Berberine 500 mg; SulforaBoost ×1
• Evening: Magnesium glycinate 200–400 mg

Separate psyllium from medications by 2 hours.

Monitoring

Do not run an open-loop intervention. Order labs.

Baseline and at 12 weeks:

• Fasting glucose, fasting insulin (calculate HOMA-IR: glucose [mg/dL] × insulin [µIU/mL] ÷ 405)
• HbA1c
• Full lipid panel including triglycerides
• 25(OH)D
• hs-CRP
• Liver enzymes (berberine has rare hepatotoxicity reports)
• Serum magnesium and RBC magnesium if available (RBC is the better marker)

Adjunct: A two-week continuous glucose monitor period yields more actionable behavioral data than any single supplement in this stack. Recommend it routinely.

If HOMA-IR, HbA1c, and triglycerides have not moved meaningfully at 12 weeks with adherence to the stack and reasonable lifestyle measures, the underlying driver is something else — sleep-disordered breathing, PCOS, subclinical hypothyroidism, alcohol intake the patient is underreporting, hypercortisolism — and supplements will not fix it. Investigate, don't escalate the supplement count.

TL;DR

Evidence-graded stack for insulin resistance and metabolic syndrome:

Core (everyone):

1. Psyllium husk 5–10 g/day pre-meals — pharmaceutical-tier HbA1c reduction (−0.97%) and LDL-C reduction. Cheapest and most underused intervention here.
2. Berberine 1500 mg/day split TID — strongest HOMA-IR data outside of pharmaceuticals.
3. Sulforaphane with myrosinase 200–460 mg/day — hepatic gluconeogenesis suppression via NRF2.
4. Magnesium glycinate 200–400 mg/day — broadly deficient, HOMA-IR and FPG benefit at ≥4 months.
5. Omega-3 2–3 g EPA+DHA/day — triglycerides and inflammation.
6. Vitamin D3/K2 if 25(OH)D <30 ng/mL — repletion only.

Conditional:

• Creatine 5 g/day if the patient lifts.
• Inositol 40:1 MI:DCI if PCOS-driven IR.
• ALA 600 mg/day if diabetic peripheral neuropathy is present.

Skip: NMN, generic probiotics, cinnamon, chromium.

Above all: if the patient is a candidate for metformin or a GLP-1, that is the conversation. Supplements are an adjunct, not a substitute. Track HOMA-IR, HbA1c, and triglycerides at baseline and 12 weeks. Screen medications for berberine interactions before initiating. Confirm vitamin D and magnesium status before supplementing.

Affiliate links disclosed inline; non-affiliate alternatives provided where relevant. Comments and disagreement welcomed. Bring data.

Just a simple meal that cannot be beat!
Veggies tossed in olive oil, S&P. Chicken seasoned with paprika, chili powder, S&P. I actually seasoned the chicken leg with salt overnight so it has time to penetrate and make it juicy! Roasted at 380 for around 40-45 mins, or whenever the chicken skin looks golden. I don’t eat a lot of meat but have been craving chicken for a few days. This leg is from a pasture-raised, slow-growth heirloom breed raised locally. I’ll prob have half this tonight and half for lunch tomorrow 😋