Benfotiamine has one of the cleanest stories in the supplement aisle. People with diabetes are measurably short of thiamine; benfotiamine is a form of thiamine that actually gets absorbed; and the enzyme it feeds, transketolase, sits exactly where the biochemistry of diabetic complications goes wrong. Every link in that chain is real and published. The question this page exists to answer is the one the story skips: does topping up that pathway change anything a patient can feel or measure years later? As of 2026 the best trial ever run on that question came back negative, and that changes how the rest of the evidence should be read.
What benfotiamine actually is (and the “fat-soluble B1” myth)
Almost every label calls benfotiamine a fat-soluble or lipid-soluble form of vitamin B1. That is not correct, and the correction is genuinely useful rather than pedantic.
Benfotiamine is S-benzoylthiamine O-monophosphate, a synthetic S-acyl thiamine derivative. When it was characterized directly, it turned out to be practically insoluble in water, organic solvents and oil — the experimenters had to solubilize it in cyclodextrin to dose it at all. It is amphiphilic, not lipophilic, and it should be distinguished from the genuinely lipid-soluble thiamine disulfide derivatives such as allithiamine, sulbutiamine and fursultiamine, which are absorbed by a different route and behave differently.[1]
The actual absorption sequence is: benfotiamine is dephosphorylated by intestinal alkaline phosphatase at the brush border, enters the bloodstream as S-benzoylthiamine, and is then converted to thiamine in erythrocytes and the liver — where it is phosphorylated to thiamine monophosphate and the active cofactor thiamine diphosphate (TDP).[1] So benfotiamine is a prodrug that solves an uptake problem, not a molecule with new pharmacology of its own.
The mechanism: one enzyme, three damage pathways
Three of the biochemical pathways blamed for hyperglycemia-induced vascular damage — the hexosamine pathway, advanced glycation end-product (AGE) formation, and the diacylglycerol–protein kinase C (PKC) pathway — are all driven by an oversupply of the same two glycolytic metabolites: glyceraldehyde-3-phosphate and fructose-6-phosphate.
Transketolase, a thiamine-diphosphate-dependent enzyme of the pentose phosphate pathway, consumes exactly those two metabolites and converts them into pentose-5-phosphates. The foundational finding, published in Nature Medicine in 2003, is that benfotiamine activates transketolase and thereby inhibits all three pathways at once, along with hyperglycemia-associated NF-κB activation — and that in diabetic animals this prevented experimental diabetic retinopathy.[2] A parallel line of work showed high-dose thiamine and benfotiamine countering triosephosphate accumulation in diabetic rats and strongly inhibiting the development of microalbuminuria, without changing blood glucose or HbA1c at all.[3] That last detail is the elegant part of the pitch: a way to blunt glucose damage without touching glucose.
There is also a plausible deficit to correct. Measured properly, plasma thiamine was about 75–76% lower in people with type 1 and type 2 diabetes than in healthy controls (roughly 15–16 nmol/L versus 64 nmol/L), driven by a 16- to 24-fold increase in renal thiamine clearance — and conventional assessment missed it, because erythrocyte thiamine transporter content was up-regulated in compensation.[7] So this is not a manufactured deficiency. It is a real, renal-loss-driven thiamine deficit that standard testing hides.
Bioavailability: the part that is genuinely proven
If you take one evidence-backed claim away from this page, take this one. In a randomized human pharmacokinetic study comparing oral benfotiamine against oral thiamine hydrochloride, the bioavailability of thiamine in plasma after benfotiamine was 1147 ± 490% of thiamine HCl, and of thiamine diphosphate in erythrocytes 196 ± 34% — roughly an eleven-fold advantage in circulating thiamine and a doubling of the active intracellular cofactor.[4] An earlier pharmacokinetic review put peak plasma thiamine about five times higher after benfotiamine, with overall bioavailability up to 3.6× that of thiamine salts, and noted that the functional marker — erythrocyte transketolase activation — improved only after benfotiamine.[5]
The effect holds in the patients most likely to need it. In 20 end-stage renal disease patients randomized to 100 mg benfotiamine or 100 mg thiamine nitrate, benfotiamine produced a 4.3× higher TDP exposure, raised erythrocyte TDP from 159 to 326 ng/mL over 24 hours (versus 166 to 201 ng/mL on thiamine nitrate), and significantly improved erythrocyte transketolase activity where thiamine nitrate did not.[6] And the modern trials confirm the pharmacology end-to-end: BOND measured six thiamine analytes in blood and benfotiamine raised every one of them (p ≤ 0.003 versus placebo).[11]
This is the rare supplement where the delivery claim is not marketing. It is the outcome claim that has to carry the weight — and the same trial that proved the delivery is the one that failed to find the outcome.
Diabetic polyneuropathy: the most-studied use, and the trial that changed it
This is where benfotiamine earned its reputation, and the older evidence is real but narrow: small, short, symptom-score trials, mostly conducted in German-speaking Europe, several of them run or funded by the companies that market the product.
- BEDIP (2005) — 40 inpatients randomized to 400 mg/day benfotiamine or placebo for three weeks. The neuropathy score improved significantly versus placebo (p = 0.0287), with the clearest effect on pain (p = 0.0414). The tuning-fork vibration test did not change, and HbA1c and glucose were unaffected — so the effect was not indirect glycemic control.[8]
- Winkler (1999) — 36 patients, six weeks, comparing 320 mg/day of a benfotiamine–B-vitamin combination against 150 mg/day benfotiamine monotherapy. All groups improved on pain, vibration and current perception threshold; the high-dose group improved most, which is where the “more is better” dosing folklore comes from. Note the design: there was no placebo arm.[9]
- BENDIP (2008) — the largest of the classical trials: 165 patients randomized to 600 mg/day, 300 mg/day or placebo for six weeks. The primary endpoint, the Neuropathy Symptom Score, differed significantly between groups in the per-protocol population (p = 0.033) but missed significance in the intention-to-treat analysis (p = 0.055). The Total Symptom Score showed no significant difference at six weeks. Benefit was larger at the higher dose and grew with treatment duration; tolerability was good.[10]
Read honestly, that is a consistent but fragile signal: symptom questionnaires move, objective nerve measures mostly do not, follow-up is three to six weeks, and the flagship trial’s primary endpoint survived only in the per-protocol analysis. Which is exactly why the next trial matters so much.
Broader pooling is equally sober. A 2026 meta-analysis of 13 randomized trials (834 participants) of B vitamins in diabetic peripheral neuropathy found improvements in the Michigan Neuropathy Screening Instrument questionnaire (MD −1.44) and examination (MD −0.39) — but the examination effect fell short of the 0.5-point minimal clinically important difference, pain on numeric and visual analogue scales showed no significant effect (MD −0.44, I² = 84.5%), and results across motor nerves were inconsistent, with tibial conduction velocity actually favouring control.[12] Sural nerve conduction did improve. For comparison, alpha-lipoic acid — the other pathogenetically-aimed agent in this space — has a larger and more replicated symptom-trial base, and it too has never demonstrated structural nerve regeneration or a disease-modifying effect.[13]
Kidney disease and endothelial function: where the negative trials live
The rodent nephropathy data were striking,[3] which makes the human result instructive. In a double-blind, randomized, placebo-controlled trial, 82 patients with type 2 diabetes and urinary albumin excretion of 15–300 mg/24h despite ACE inhibition or ARB therapy received benfotiamine 900 mg/day or placebo for 12 weeks. Thiamine status improved highly significantly (p < 0.001) — and urinary albumin excretion did not fall, nor did excretion of the tubular injury marker KIM-1.[14]
The same cohort was analysed for the intermediate biology, and that failed too: benfotiamine produced no significant reduction in plasma or urinary AGEs (CML, CEL, MG-H1), in markers of endothelial dysfunction (sVCAM-1, sICAM-1, sE-selectin), or in low-grade inflammation (hs-CRP, serum amyloid-A, myeloperoxidase) over 12 weeks.[15] This is the single most important negative result in the benfotiamine literature after BOND, because it tested the mechanism itself — and the mechanism did not move in humans at 900 mg/day.
A Cochrane review of vitamin B and its derivatives in diabetic kidney disease (nine trials, 1,354 participants) reached the corresponding verdict: an absence of evidence to recommend vitamin B therapy for delaying progression of diabetic kidney disease, with no improvement in kidney function or blood pressure, and a note that four of the included studies were pharmaceutical-company funded and judged at high risk of bias.[16]
AGEs and oxidative stress: a real acute effect that never became an outcome
The most vivid positive human finding is also the shortest. Thirteen people with type 2 diabetes ate a heat-processed, AGE-rich test meal (15,100 AGE kU) before and after three days of benfotiamine at 1,050 mg/day. The meal alone crushed vascular function — microvascular reactive hyperemia fell 60% at two hours and flow-mediated dilatation 35% at four hours. Those effects were completely prevented by benfotiamine, and markers of endothelial dysfunction, oxidative stress and circulating AGEs were significantly reduced.[17] It is a striking result, and it is where most of the “anti-glycation” marketing traces back to.
The follow-up is the part that rarely gets quoted. The same group ran a larger, double-blind, placebo-controlled crossover trial in 31 patients: 900 mg/day for six weeks, testing postprandial macrovascular, microvascular and autonomic function. Baseline flow-mediated dilatation was already severely impaired (2.63%), the test meal did not worsen it further in the placebo condition, and six weeks of benfotiamine did not alter the pattern either fasting or postprandially.[18] A three-day acute-challenge effect in 13 people did not survive translation to six weeks in 31.
Nor does benfotiamine lower blood sugar, which is worth saying plainly because it is often shelved beside compounds that do. A systematic review and meta-analysis of six trials (364 participants) of thiamine or benfotiamine at 100–900 mg/day for up to three months found no significant effect on HbA1c (MD −0.02%), fasting glucose or postprandial glucose; HDL rose slightly and benfotiamine at 120 mg/day reduced triglycerides, an effect not reproduced at higher doses.[19] If glycemic control is the goal, this is the wrong molecule — see berberine or metformin for compounds that actually target that.
| Indication or claim | Best human evidence | Verdict |
|---|---|---|
| Raising blood and tissue thiamine / TDP | ~11× plasma thiamine and ~2× erythrocyte TDP vs thiamine HCl; all six thiamine analytes raised in BOND | Proven |
| Diabetic polyneuropathy — symptom scores, 3–6 weeks | BEDIP (n=40) and BENDIP (n=165) positive on symptom scores; BENDIP primary endpoint significant per-protocol only (p=0.033), not ITT (p=0.055) | Weak positive |
| Diabetic polyneuropathy — nerve structure and function, 12 months | BOND (n=57): no change in corneal nerve fibre length or in any morphometric, neurophysiological, clinical or quality-of-life outcome | Negative |
| Diabetic nephropathy — albuminuria | 900 mg/day × 12 weeks (n=82): no reduction in urinary albumin excretion or KIM-1 despite improved thiamine status | Negative |
| AGE and endothelial/inflammatory markers | Acute 3-day study (n=13) prevented post-meal endothelial dysfunction; 12-week (n=82) and 6-week (n=31) trials found nothing | Contradictory |
| Glycemic control (HbA1c, fasting glucose) | Meta-analysis of 6 trials, 364 participants: no significant effect | Negative |
| Alzheimer's disease / cognitive decline | One 12-month phase IIa pilot (n=70): primary ADAS-Cog endpoint not significant (p=0.125); CDR secondary favourable (p=0.034) | Pilot only |
| Diabetic retinopathy prevention in humans | Rodent data only — never tested as a clinical endpoint in a human trial | Untested |
| Anti-aging / anti-glycation in healthy people | No trial. Human dosing studies in healthy subjects measured pharmacokinetics and safety, not outcomes | Untested |
Alzheimer’s disease: a genuine pilot, and a genuine paradox
There is one randomized human trial, and it should be graded as exactly what it is: a phase IIa pilot. Seventy participants with amnestic mild cognitive impairment or mild dementia due to Alzheimer’s disease were randomized to benfotiamine (34) or placebo (36) for twelve months. On the primary clinical outcome, ADAS-Cog, the increase was 43% smaller in the benfotiamine group — but this was not statistically significant (p = 0.125). Secondary outcomes were friendlier: worsening on the Clinical Dementia Rating was 77% lower (p = 0.034), circulating AGEs rose less (p = 0.044), and both effects were stronger in APOEε4 non-carriers. Treatment was safe.[20]
A pilot with a null primary endpoint and positive secondaries is a reason to run a real trial, not a reason to take a supplement. That trial is running: BenfoTeam, a seamless phase 2A–2B, randomized, double-blind, placebo-controlled study of 406 participants with early Alzheimer’s disease over 72 weeks, with an adaptive choice between 600 mg and 1,200 mg daily and co-primary endpoints of CDR-Sum of Boxes and ADAS-Cog13.[21] Until it reports, anyone selling benfotiamine as a cognitive or Alzheimer’s intervention is selling a 70-person pilot.
And the paradox is worth stating rather than smoothing over: the mouse pharmacology says benfotiamine does not raise brain thiamine.[1] Either the central effect is indirect — via peripheral AGE burden, glucose metabolism or vascular health — or the rodent result does not generalize. Nobody has resolved this, and the honest position is that the mechanism for a central benefit is currently unexplained.
Dose, regulatory status and safety
Dose. Supplements typically supply 150–600 mg/day. The trial doses cluster higher than the shelf: 300 mg twice daily (600 mg/day) in BOND and the higher BENDIP arm,[10][11] 400 mg/day in BEDIP,[8] and 900 mg/day in the nephropathy and vascular trials.[14][18] There is no established dose–response for benefit in humans, because no dose has yet produced a durable clinical benefit to titrate against. Winkler’s finding that higher doses worked better was in a six-week, placebo-free comparison of 36 patients.[9]
Regulatory status. These are genuinely different legal products on either side of the Atlantic. In Germany, benfotiamine is a licensed medicinal product — for example milgamma® protekt, 300 mg benfotiamine per film-coated tablet, non-prescription but apothekenpflichtig (pharmacy-only), with an approved indication of neuropathies and cardiovascular disorders caused by vitamin B1 deficiency.[25] Read that indication carefully: even the European licence is framed around deficiency, not diabetic neuropathy in general. In the United States, benfotiamine is a synthetic thiamine derivative used in dietary supplements, with no FDA approval for any indication and no premarket review of identity, potency or purity.[24]
Safety. Do not hand-wave this by saying “thiamine is water-soluble.” Benfotiamine is a distinct synthetic entity that is not water-soluble,[1] and it needs its own data — which, fortunately, exists. Two randomized, double-blind, placebo-controlled phase I trials tested single doses of 150–1,200 mg and multiple ascending doses of 150, 300 or 600 mg in healthy subjects. Incidence and severity of adverse events were similar to placebo; the drug-related events most commonly reported were increased ALT and urinary white cells. Thiamine and TDP accumulated about two-fold on repeated dosing.[22] Twelve months at 600 mg/day in BOND was likewise well tolerated with no relevant between-group difference in adverse-event rates.[11] Separately, the Food and Nutrition Board has set no tolerable upper intake level for thiamine, because no adverse effects from high intakes have been reported.[24]
One interaction is worth knowing about, stated at the strength the evidence supports. A screen of 1,360 compounds for inhibition of the intestinal thiamine transporter ThTR-2 (SLC19A3) found 146 hits, and metformin was among the oral drugs predicted to reach intestinal concentrations high enough to cause a transporter-mediated drug–nutrient interaction; electronic health record analysis showed lower measured thiamine values in people taking drugs that significantly inhibit ThTR-2, particularly in vulnerable groups.[23] That is a mechanistic and pharmaco-epidemiological signal, not a demonstration that metformin causes clinical thiamine deficiency, and not on its own a reason to supplement. It is a reason to have thiamine status measured rather than guessed if you have long-standing diabetes on metformin — and the Thornalley data say conventional erythrocyte-based testing can miss it.[7]
Who this plausibly makes sense for
- Someone with diabetes and a documented low thiamine status. This is the defensible case, and it is narrow. Benfotiamine corrects the deficit better than thiamine salts do,[4][6] the deficit is real and renally driven,[7] and correcting a measured deficiency is a different act from taking a supplement on a hypothesis.
- Someone with symptomatic diabetic polyneuropathy who has exhausted better options. A three-to-six-month trial at 300 mg twice daily is cheap and well tolerated, and the symptom-score data are not nothing. Go in knowing the twelve-month structural trial was negative,[11] and set a stopping rule before you start: if your symptoms have not changed by three months, they are not going to.
- A healthy person buying it to fight glycation or aging. There is no trial. Human studies in healthy subjects have measured pharmacokinetics and tolerability, not outcomes,[22] and even the acute AGE-meal result was in people with diabetes and did not survive to six weeks in the same research group’s larger study.[17][18] This is the claim with the widest gap between marketing and evidence.
- Someone with a family history of Alzheimer’s. Wait for BenfoTeam.[21] A 70-person pilot that missed its primary endpoint is a hypothesis, and a 406-person trial designed to test it is already underway.[20]
The honest bottom line
Benfotiamine is the best-argued molecule in the longevity aisle that has not yet been shown to work. The chemistry is correctly described (once you drop the “fat-soluble” label), the bioavailability advantage over thiamine hydrochloride is large and repeatedly measured, the deficiency it addresses in diabetes is real, and the transketolase-shunt mechanism is one of the most satisfying stories in diabetes biochemistry.[2][4][7]
What has not happened is a clinical payoff. Twelve weeks at 900 mg/day did not touch albuminuria or a single AGE, endothelial or inflammatory marker in patients with nephropathy.[14][15] Twelve months at 600 mg/day did not change corneal nerve fibre length, nerve conduction, sensory testing, autonomic function or quality of life in patients with polyneuropathy — in a trial that proved the drug was being absorbed and acting.[11] The remaining positive signal is a set of three-to-six-week symptom questionnaires, one of which missed significance in its own intention-to-treat analysis.[10]
So the fair summary splits cleanly. “Benfotiamine corrects a thiamine-pathway deficit in people with diabetes better than ordinary vitamin B1 does” is well supported. “Benfotiamine is an anti-glycation, anti-aging supplement” is not demonstrated in anyone. It is inexpensive, well tolerated and worth a defined trial in the right patient. It is not a reason to stop doing the things that actually change diabetic complication rates. Weigh it against the rest in our longevity evidence matrix.
This article is research information, not medical advice. In the United States benfotiamine is sold as a dietary supplement and is not approved by the FDA to treat, prevent or manage diabetes, diabetic neuropathy, kidney disease or Alzheimer’s disease; the European licence it holds in some markets is written around vitamin B1 deficiency, not diabetic complications in general. Diabetic neuropathy and nephropathy require medical diagnosis and management, and benfotiamine does not lower blood glucose — it is not a substitute for glycemic control, blood-pressure control, ACE inhibitor or ARB therapy, or any prescribed treatment. Talk to a licensed clinician before starting it, particularly if you have diabetes, kidney disease, liver disease, are taking metformin or other prescription medication, or are pregnant or breastfeeding.