Supplements · Metabolic & Cardiometabolic
mulberry leaf (1-deoxynojirimycin) decreases postprandial glucose
In plain terms: Does mulberry-leaf extract lower blood-sugar spikes?
Strong support Supplements 💰 Industry COI noted
RefutedContestedStrong support
consensus score 1.00
Small RCTs (often industry-linked) show acute postprandial glucose reduction, strongest in impaired-glucose subjects; evidence is positive but low-to-moderate quality.
📅 Last reviewed: 2026-07-14 ⓘ
Evidence ladder
How far up the ladder this claim has climbed. A high consensus on a low rung means "consistent so far," not "proven in people."
Top evidence so far: Human trials (RCT / n-of-1)
MechanismIn-vitroAnimalObservationalRCTMeta-analysis
How the studies fall
11 support 0 contradict 0 tested null 0 mixed · 11 sources, 11 independent groups · 1 superseded — pooled inside a review, counted once
What the evidence shows
Mulberry-leaf DNJ blunts postprandial glucose/insulin via intestinal alpha-glucosidase inhibition (acarbose-like) at low dose — mechanistically ideal for a starchy bread, but thermal survival through baking is the critical unproven gate.
The evidence (16)
| Source | Grade | Stance | Quality | Finding |
|---|---|---|---|---|
| Shinkawa Y, Yasuda M, Nishida Y, Tokiya M, Takagi Y, Matsumoto A, Kawaguchi A, Hara M 2025 · Nutrients | RCT | supports | moderate | Full text (Nutrients 2025;17:2308). Randomized double-blind placebo-controlled 2-period crossover, n=31 healthy non-diabetic Japanese adults (median age 56, HbA1c 5.5%), UMIN000055353, Saga University. EXPOSURE IS NOT MULBERRY ALONE: each 3 g sachet = 2.85 g mulberry leaf + 0.15 g water chestnut husk, delivering 5.10 mg DNJ + 87.3 mg polyphenols, before each meal for 2 weeks; no mulberry-only arm, and the authors credit 'the combined use of mulberry leaves and water chestnut husks' with a synergistic effect. The headline p=0.0006 is the PRIMARY outcome - the coefficient of variation of CGM glucose over the central 10 days (LSMean difference -0.0203, 95% CI -0.031 to -0.010, n=31) - i.e. glycemic VARIABILITY, not this claim's object. The on-object endpoint is SECONDARY and ACUTE, from a single standardized meal on day 1 of each period only: incremental 1 h AUC 1704 (SD 1145) vs 2263 (SD 1289), p=0.020, with 2 h (p=0.267) and 3 h (p=0.484) both null and no correction for multiple comparisons, which the authors state outright. 1,5-AG rose (p=0.005, n=30); fasting glucose, HbA1c, insulin and glycoalbumin unchanged. Funded by Kanzaki City, which developed the product. Tea not bread - bake-stability unproven. |
| Cui W, Luo K, Xiao Q, Sun Z, Wang Y, Cui C, Chen F, Xu B, Shen W, Wan F, Cheng A 2023 · Food Funct 2023;14(3):1277-1289 | meta-analysis | supports | moderate | OFF-SCOPE for postprandial glucose. Abstract-grade (NCBI efetch; Food Funct 2023;14(3):1277-1289, doi 10.1039/d2fo02645g - subscription-only, Europe PMC confirms pmcid null / isOpenAccess N, so no full text). Systematic review + random-effects meta-analysis of RCTs whose outcome set is stated exhaustively and is entirely FASTING: the aim was to 'examine the effect of ML/MLE supplementation on glycemic traits in adults, including fasting blood glucose (FBG), glycosylated hemoglobin (HbA1c), and fasting plasma insulin (FPI)'. THREE separate pools, one per fasting outcome - 'a significant reduction in FBG by -0.47 mmol L-1, HbA1c by -2.92 mmol mol-1, and FPI by -0.58' uIU/mL - and NO postprandial, post-load or iAUC glucose pool appears anywhere. The prior extract's '12 RCTs (615 subjects)' is the ELIGIBILITY total spanning all three outcomes, not the denominator of any one of them; per-outcome pool sizes, CIs and heterogeneity statistics are not in the abstract (cannot-tell, not absent), and the prior extract dropped FPI entirely. SECOND, INDEPENDENT OFF-SCOPE GROUND - POOLED EXPOSURE: the exposure is mulberry leaf OR mulberry leaf extract with no DNJ-isolated arm, and the authors attribute the effect to a mixture, 'phytochemicals they contain, which are mainly 1-deoxynojirimycin, flavonoids, phenolics, and polysaccharides'; this identical row would vote the same way on a mulberry-flavonoid or mulberry-polysaccharide claim, so it measured none of them in isolation. What it DID find points away from this claim's acute alpha-glucosidase mechanism: 'long-term supplementation of ML/MLE (>=8 weeks) was more effective for regulation of the glycemic traits in the non-healthy and baseline FPG >6.1 mmol L-1 subgroups' - a chronic, elevated-baseline effect, not a post-meal blunting. Quality held at moderate: sensitivity analysis and formal GRADE certainty were performed and random effects were chosen for stated heterogeneity, but the GRADE ratings and any publication-bias test are not reported in the abstract. Consistent with this claim's existing precedent - s20838571 was off-scoped on exactly this ground for measuring only fasting endpoints. Capsules/powders, not bread: says nothing about bake stability. |
| Thaipitakwong T, Supasyndh O, Rasmi Y, Aramwit P 2020 · Complement Ther Med | RCT | supports | moderate | Abstract-grade (efetch; not in PMC OA, publisher paywalled). Complement Ther Med 2020;49:102292, doi 10.1016/j.ctim.2019.102292. TWO randomized studies in one paper, and only the FIRST bears on this claim's object. Dose-finding: 'healthy adults were recruited into the dose-finding study and randomly allocated to ingest sucrose solution concurrently with mulberry leaf powder at weights equivalent to 0 (control), 6, 12, and 18 mg of DNJ. Postprandial glucose and undesirable effects were evaluated', giving the on-object result 'Twelve mg of mulberry DNJ was the minimum effective dose attenuating postprandial hyperglycemia' - acute co-ingestion with a sucrose solution in healthy adults, with no n, no effect size and no p-value in the abstract (cannot-tell, not absent). The SECOND study - 12 weeks, obese persons with borderline diabetes, 12 mg DNJ three times daily vs a control group - reported FASTING endpoints and reported them against baseline rather than against its own randomized control: 'Mulberry leaves decreased fasting plasma glucose (FPG) by 3.86 +/- 5.99 mg/dL (p = 0.002) and glycated hemoglobin (HbA1c) by 0.11 +/- 0.22 % (p = 0.011) when compared with the baseline levels', with insulin resistance only 'tended to ameliorate' (p = 0.057). Its one postprandial-adjacent reading was NULL: 'Improvement in glucose tolerance was not observed' - kept out of the stance because it tests durable tolerance after chronic dosing, not the acute alpha-glucosidase blunting this claim asserts (rubric v1.1.1 axis 1), and logged as a candidate. The prior extract credited the postprandial finding to the borderline-diabetic 12-week arm; the finding is the healthy-adult sucrose-load arm, and fasting glucose and HbA1c are not this claim's object. Adverse events were gastrointestinal: bloating, flatulence, loose stools. 'No conflict of interest to declare'; funding and blinding not stated. Says nothing about bake stability - vehicle was a sucrose solution. |
| Lown, Fuller, Lightowler, Fraser, Gallagher, Stuart, Byrne, Lewith 2017 · PLoS One | RCT | supports | moderate | MULBERRY trial (ISRCTN14597438), double-blind placebo-controlled 4-way crossover, n=37 healthy normoglycaemic adults (BMI 20-30, none >25; only 2 aged >50), Reducose mulberry leaf extract standardised to 5% DNJ with 50 g maltodextrin: glucose pIAUC over 120 min fell 6.1% at 125 mg (NS, p=0.316), 14.0% at the normal 250 mg dose (-26.0 to -2.0; p=0.022) and 22.0% at 500 mg (-33.9 to -10.0; p<0.001), a dose gradient with no carry-over (F=1.04, p=0.377). Insulin pIAUC fell 23.8% and 24.7% at the two higher doses. Funded by an Innovate UK grant, not by the manufacturer, but Phynova's COO is a co-author and set the dose regimen from unpublished company phase-1 data; acute single dose, pure maltodextrin in water, normoglycaemic only, and the authors caution against extrapolating to dysglycaemia. |
| Kwon HJ, Chung JY, Kim JY, Kwon O 2011 · J Agric Food Chem | animal | supports | moderate | Abstract-grade (efetch). J Agric Food Chem 2011;59(7):3014-9, doi 10.1021/jf103463f. Preclinical head-to-head designed to separate whole aqueous mulberry leaf extract (MLE) from isolated DNJ, i.e. to test the very conflation this claim's subject entity makes. THE TWO AGREE ON THIS CLAIM'S OWN STATED MECHANISM: 'Both MLE and DNJ were active in inhibiting alpha-glucosidase.' They diverge on a second, transporter route: 'in Caco-2 cells, only MLE showed significant inhibition of 2-deoxyglucose uptake, whereas DNJ was ineffective', the authors concluding 'the unabsorbed phytochemicals in MLE compete with glucose for intestinal glucose transporters, but DNJ itself does not.' In vivo the comparator is DNJ, NOT vehicle: 'For glucose loading, co-administration of MLE resulted in potent inhibitions of glucose responses compared to those by DNJ in Sprague Dawley (SD) rats, but this was not found for maltose loading' - so what is absent on maltose is MLE's superiority over DNJ, which is what the shared alpha-glucosidase route predicts. Separately, timing mattered: MLE given 30 min before the glucose load significantly lowered IAUC versus simultaneous dosing, mirrored by Caco-2 pretreatment. No rat group sizes, no doses of either agent, no effect sizes, no p-values and no funding statement appear in the abstract - cannot-tell, not absent. Rats and cells, not humans, and not bread: says nothing about bake stability. ADJUDICATION NOTE: the Caco-2 2-deoxyglucose-uptake null is a transport-route result outside this claim's alpha-glucosidase mechanism scope and does not count against the object; the in-claim evidence is one-directional (calibration #594, both-reader texts agree on the facts). |
| Dong YS, Yu N, Li X, Zhang B, Xing Y, Zhuang C, Xiu ZL 2020 · J Agric Food Chem | in-vitro | supports | moderate | Abstract-grade (NCBI efetch; ACS title, not in the PMC OA subset - BioC returned no result). J Agric Food Chem 2020;68(33):8774-8787, doi 10.1021/acs.jafc.0c01668. ON THIS CLAIM'S OWN MECHANISM, BUT THE EXPOSURE IS A COMBINATION, NOT DNJ ALONE: 'a combination of dietary 5,6,7-trihydroxy-flavonoid aglycones with 1-DNJ showed synergistic inhibitory activity against maltase of mice alpha-glucosidase and recombinant C- and N-termini of maltase-glucoamylase (MGAM) and baicalein with 1-DNJ exhibited the strongest synergistic effect', and the single in-vivo line is 'The synergistic effect of the combination was also confirmed by the maltose tolerance test in vivo.' The prior extract read 'DNJ inhibited maltase-glucoamylase and lowered postprandial glucose in mice' - a SOLO-DNJ in-vivo glucose result the abstract does not report; what was confirmed in vivo is the combination's synergy, and DNJ's own alpha-glucosidase activity appears as established BACKGROUND ('exhibiting inhibitory activity against alpha-glucosidase'), not as a measurement made here. No mouse group sizes, no doses, no glucose values, no p-values, no funding or COI statement (cannot-tell, not absent). The mechanism is the solid part, from four orthogonal methods - enzyme kinetics, molecular docking, fluorescence spectrum and circular dichroism: 'baicalein was a positive allosteric inhibitor and bound to the noncompetitive site of MGAM, causing an increase of the binding affinity of 1-DNJ to MGAM.' The authors' own premise is a direct caveat to this claim as it applies to real products: 'due to the low content of 1-DNJ in mulberry products, its level cannot meet the lowest dose to exhibit its activity' - under-dosing is the problem the synergy is proposed to solve. Recombinant enzyme and mice, not humans; a maltose solution, not bread - says nothing about bake stability. GRADE FLAGGED, NOT CHANGED: the only endpoint bearing on this claim's object is the ANIMAL maltose tolerance test while the row is graded in-vitro; an upgrade is cultivar's call (rubric axis 4). |
| Thondre PS, Butler I, Tammam J, Achebe I, Young E, Lane M, Gallagher A 2024 · Nutrients | RCT | supports | moderate | Full text (Nutrients 2024;16(11):1670, PMC11174565, doi 10.3390/nu16111670; ISRCTN18212231, RETROSPECTIVELY registered). Double-blind, placebo-controlled, four-way crossover with Latin-square sequences and concealed allocation, run to ISO 26642; 43 recruited, n=37 completers who are the denominator for EVERY number below (22F/15M), healthy lean normoglycaemic adults 18-56 y, BMI mean 23.2, fasting glucose <=6.1 mmol/L; powered a priori at 31 for 80%. Exposure is Reducose mulberry leaf extract standardised to 5% (w/w) DNJ at 200/225/250 mg (5% = 10/11.25/12.5 mg DNJ, arithmetic not stated by the authors), one capsule 10 min before a MIXED MEAL - 150 g white bread + 50 g egg mayo filler, 68.7 g carbohydrate, 18.7 g protein, 10.2 g fat - which is this paper's real novelty, since every earlier Reducose trial used maltodextrin or sucrose in water. ON-OBJECT RESULT, the pre-specified primary endpoint: glucose iAUC 120 fell 30% (p=0.003), 33% (p=0.001) and 32% (p=0.002); iAUC was significantly lower at every window from 60 to 180 min; mean peak glucose 6.0/5.9/6.0 vs 6.6 mmol/L placebo (p=0.001, <0.001, 0.002) with time-to-peak unchanged (p=0.135). PRIOR EXTRACT CORRECTED - THE EFFECT IS NOT DOSE-DEPENDENT: 'There were no significant differences between the glucose iAUCs for the three doses of Reducose' and 'There were no significant differences in the glycaemic responses between the three doses' - a flat 30-33% plateau from 200 mg upward, the opposite of what the previous one-line extract asserted (the dose gradient belongs to s28225835, which spanned 125-500 mg). The separation is EARLY and gone by 90 min: 'Between the placebo and three doses of Reducose groups, there was no significant change in actual blood glucose measurements at time points 90 min, 120 min, 150 min and 180 min'. Insulin iAUC 120 fell 31/34/38%, and all three arms sat ABOVE placebo at 180 min, which the authors attribute to late-phase GLP-1 - a hypothesis, no incretin was assayed. QUALITY HELD AT MODERATE, both directions weighed: blinded own-control crossover, matched placebo, concealed allocation, daily-calibrated HemoCue, GCP and a real power calculation, against retrospective registration, no formal carry-over test, and total sponsor ownership - 'This research was funded by Phynova Group Limited', two authors are Phynova employees who 'drafted the original draft manuscript', the sponsor designed the study and generated the randomisation, the corresponding author writes from phynova.com, and 'The data are not publicly available due to intellectual property rights'. Capsule taken BEFORE bread, never baked into it - says nothing about DNJ thermal survival. INDEPENDENCE FLAG (not acted on): Reducose is Phynova's proprietary product and appears in the vault under three group_ids - s28225835 (Lown-Reducose, co-authored by this paper's corresponding author A. Gallagher), s36855010 (mulb-36855010) and this note (thondre). |
| Kojima Y et al 2010 · J Clin Biochem Nutr | observational | tested-null | low | OFF-SCOPE for postprandial glucose. Full text (J Clin Biochem Nutr 2010;47:155-161, PMC2935155). Open-label, single-arm, uncontrolled 12-week before-after study in 10 Japanese men with TG >=200 mg/dl (9 analysed; 1 excluded for chylous samples), 12 mg DNJ three times daily before meals. Stated objective: 'to evaluate the effects of DNJ-rich mulberry leaf extract on plasma lipid profiles in humans.' All blood draws were FASTING - 'no foods or beverages were allowed from the end of dinner on the day before tests to the completion of tests', samples taken 'after a 12 h fast' - so no postprandial glucose was measured; the only glucose data are fasting glucose (89 +/- 5 to 86 +/- 6 mg/dl, ns) and HbA1c (5.2 to 5.1%, ns). The paper's one postprandial-glucose statement, 'We previously showed that DNJ-rich mulberry leaf extract suppressed elevation of postprandial blood glucose in humans', is a citation to ref 11 (Kimura 2007, PMID 17555327), already counted here as mulberry-2007. On its own lipid endpoints (n=9) the result was largely null: TG 312 +/- 90 to 252 +/- 78 mg/dl, p=0.058, and 'no significant change from baseline at week 6 or 12 in total cholesterol, LDL-cholesterol, or HDL-cholesterol'; only subfractions moved (CM-TG 18.6 to 9.0 mg/dl, p=0.031; VLDL-TG rose at week 6, p=0.007) with 'No multiplicity was taken into consideration'. Corresponding author is employed by Minato Pharmaceutical, which supplied the extract. |
| Kim JY et al 2015 · J Med Food | RCT | supports | moderate | Abstract-grade. 4-week randomized double-blind placebo-controlled trial, n=36 total with impaired fasting glucose; 5 g/day is the mulberry leaf AQUEOUS EXTRACT dose, not a DNJ dose (DNJ content not stated in the abstract). Glucose was attenuated at 30 and 60 min post-load (P=.003 and .0325), phrased as attenuation within the MLAE arm after treatment, so the denominator is that arm, not the headline 36 (per-arm n not stated). The p=.02 previously headlined here is the INSULIN iAUC (P=.0207) - the only explicitly MLAE-vs-placebo statistic in the abstract, and a different outcome from this claim's object; no glucose iAUC is reported in the abstract. |
| Yu JT, Li CP, Hsiao Y, Cheng KP, Tsai RY 2025 · Int J Mol Sci 2025;26(17):8380 | meta-analysis | supports | moderate | Off-scope on the claim's object. Random-effects meta-analysis of 13 RCTs (1294 randomized, 1202 analyzed; PROSPERO CRD42024603080) of mulberry LEAF, FRUIT and TWIG extracts in adults with metabolic disorders - the abstract's 'Fifteen trials' contradicts the body's own count ('Of the thirteen studies included...'). It pools NO postprandial glucose outcome: 'The primary outcomes assessed in this study included fasting blood glucose, BW, and liver function markers such as AST and ALT... Secondary outcomes consisted of metabolic indicators such as fasting plasma insulin, HbA1c, and the HOMA-IR.' What it did pool, each with its own denominator, not the headline 1202: fasting glucose SMD -0.303 (95% CI -0.435 to -0.171) from 11 trials, of which Qu 2021 alone holds 59% of the weight; HbA1c -0.289 (-0.427 to -0.152) from 8 trials, Qu 2021 holding 64%; HOMA-IR -0.422 (-0.764 to -0.081) from just 3 trials (the running text prints -0.442 for that same CI); TC -0.334, TG -0.499, LDL -0.283. The nulls the previous extract omitted: fasting insulin -0.257 (-0.541 to 0.027) NOT significant, plus HDL, ALT, body weight, systolic and diastolic BP all null. The 'shorter duration and <500 mg/day' subgroup clause rests on strata the authors caveat themselves - 'Because several subgroups include only a single randomized controlled trial, these estimates should be interpreted with caution' - and is not even directionally consistent (the large TC effect came from <4 weeks at >3000 mg/day, the opposite band). |
| Mohamed M 2023 · Diabetes Ther | RCT | supports | moderate | Full text (Diabetes Ther 2023;14(4):749-766, PMC10064401, doi 10.1007/s13300-023-01379-4; NCT04877366). Randomized double-blind two-center exploratory crossover, 1-week washout; 30 T2D adults of Asian origin enrolled (12 women, mean age 58.9 y, HbA1c 7.1%, BMI 26.5, 73.3% on metformin), 29 completing all sequences - so the per-comparison denominator is 29 completers, not the headline 30. THE PRODUCT IS A BLEND BUT THE RANDOMISED CONTRAST ISOLATES MLE: 250 mg Reducose MLE (12.5 mg DNJ) + 1.75 g resistant maltodextrin + 0.75 ug vitamin D3 + 75 ug chromium, against a comparator that was the same blend 'with the exception of the MLE, which was substituted by fiber (250 mg Fibersol-2) to achieve weight equivalence' - fiber, vitamin D and chromium sat in BOTH arms, and the MLE arm in fact carried slightly LESS total fiber (1.75 g vs 2.0 g), which biases against MLE rather than for it. Sprinkled on a fixed 350.2-kcal breakfast (55.4 g carbohydrate: white bread, gouda, apple juice). ON-OBJECT RESULT: glucose iAUC -19.5% at 1 h (delta -0.6 [-0.8, -0.3] mmol/L x h, p<0.0001), -16.6% at 2 h (-1.2 [-1.8, -0.6], p=0.0002), -14.9% at 3 h (-1.4 [-2.4, -0.5], p=0.0032), Cmax -0.8 (-1.2, -0.3) mmol/L, p=0.0006, Tmax unchanged; no sequence or period effect, so washout was complete. WHICH NUMBER IS PRE-SPECIFIED MATTERS: 'we defined the 3-h incremental area under the curve (iAUC) as the primary analysis', i.e. the -14.9% - the headline -20% is the 1 h window, one of three, and 'P values for the primary, secondary, and additional post hoc exploratory endpoints were not adjusted for multiple comparisons'. The 2-h glucose LEVEL (as distinct from iAUC) was null: -0.4 (-1.0, 0.1), p=0.1220. Insulin fell only EARLY: iAUC 0-60 min -23.7% (p=0.0236) with 2 h (p=0.1708) and 3 h (p=0.3849) null and Cmax/Tmax unchanged. Mechanistically the paper points at alpha-glucosidase inhibition and away from incretin or gastric routes: GLP-1, IL-6, Matsuda index and gastric emptying (acetaminophen) all null, GIP markedly reduced (iAUC 0-3 h -24.5 [-37.2, -11.8] pmol/L x h, p=0.0003). QUALITY MOVED high -> moderate: self-described 'exploratory' with unadjusted p-values, a single acute breakfast, Asian-origin participants only, and total sponsor ownership - 'funded by Nestle Health Science (NHSc)', five of eleven authors NHSc employees, the original draft written by one of them, statistics by an NHSc-funded CRO, and every non-employee author except one declaring NHSc fees; the authors themselves flag that 'the combinatory approach might also be considered a limitation of the current study, as it gives rise to difficulties in comparing the results with studies involving only MLE versus a metabolically inert comparator'. Sprinkled ON bread, not BAKED into it - says nothing about thermal survival of DNJ. INDEPENDENCE FLAG (not acted on): the extract is Phynova's Reducose, the same proprietary product as s28225835 and s38892603. An open-label acarbose period in these same participants is 'reported separately' - a future duplicate-population risk. |
| Kimura T et al 2007 · J Agric Food Chem | RCT | supports | moderate | Acute human dose-finding, healthy volunteers (n not stated), 50 g sucrose challenge: the top 2 of 4 arms - 0.8 g and 1.2 g DNJ-enriched powder, i.e. 12 and 18 mg DNJ (NOT 0.8-1.2 g of DNJ) - significantly suppressed postprandial glucose and insulin; the 0.4 g / 6 mg arm is not among the doses reported effective. Randomization/blinding not stated in the abstract. |
| Asai A, Nakagawa K, Higuchi O, Kimura T, Kojima Y, Kariya J, Miyazawa T, Oikawa S 2011 · J Diabetes Investig | RCT | supports | moderate | Full text (J Diabetes Investig 2011;2(4):318-323, PMC4014974). TWO trials in one paper, and only the SMALL one measured this claim's object. STUDY 1 (on-object): randomized double-blind four-period crossover, 12 enrolled but n=10 analysed, FPG 100-140 mg/dL, single dose of 0/3/6/9 mg DNJ 15 min before 200 g boiled white rice (70.0 g carbohydrate). Glucose group x time interaction P=0.006, but the authors phrase it as 'tended to attenuate postprandial acute glycemia in a dose-dependent manner' and the ONLY separated timepoint is 30 min: placebo 145+/-28 vs 121+/-19 (6 mg) and 114+/-12 (9 mg) mg/dL, with 3 mg (131+/-21) not separated from placebo. At 60, 90 and 120 min the arms are indistinguishable and 6 mg is numerically HIGHER than placebo at 90 min (178+/-34 vs 162+/-37) - so the finding is a blunted early peak in n=10, not a lowered 2 h curve. Insulin at 30 min was lower in all three extract arms but 'the overall change in the insulin concentration over 120 min did not differ statistically among the four groups' (P=0.395). STUDY 2 (12 weeks, 6 mg DNJ t.i.d., 76 randomized / 65 analysed = 33 extract vs 32 placebo) MEASURED NO POSTPRANDIAL GLUCOSE AT ALL - every draw was fasting, and FPG (P=0.326), insulin (P=0.740), HbA1c (P=0.687) and glycated albumin (P=0.608) were all null between groups. Its sole positive is the SURROGATE 1,5-anhydroglucitol (17.5 to 20.5 ug/mL vs placebo 17.0 to 17.4, P<0.001, reversing to baseline 4 weeks after withdrawal), from which the authors INFER that 'the increased 1,5AG concentration could reflect continuous reduction in postprandial hyperglycemic spikes'. The abstract's headline conclusion about long-term postprandial control is therefore an inference from a marker, not a measurement. Registered UMIN000003154 / UMIN000003155; compliance 99%. Three of eight authors are employees or the CEO of Project M and Minato Pharmaceutical, 'the manufacturers of the mulberry leaf extract used in this study'. Authors' own limitation: 'The heterogeneity might limit the strength of our conclusions'. Capsules/tablets, not bread - bake stability untested. ↩ SUPERSEDED — pooled in the review above, counted once |
| Kume D, Nishiwaki M, Ito M 2025 · J Nutr Sci Vitaminol | RCT | supports | low | Abstract-grade (NCBI efetch; J Nutr Sci Vitaminol (Tokyo) 2025;71(1):63-69, doi 10.3177/jnsv.71.63 - Europe PMC returns pmcid null / isOpenAccess N / inEPMC N with the sole full-text link marked 'Subscription required', so no full text exists on either route). GLUCOSE IS THE SECONDARY MEASURE, NOT THIS PAPER'S AIM: 'This study aimed to investigate the effect of powdered M. australis leaf intake on the arterial stiffness response after sucrose ingestion', and the stated conclusion is vascular, that the powder 'can prevent the transient arterial stiffening after sucrose ingestion in healthy young men' - a DIFFERENT object, logged as a candidate and kept out of this extract. DESIGN AND DENOMINATOR: 'In a randomized crossover design, 12 healthy young men completed two trials: with (mulberry [M] trial) and without (control [C] trial) intake of powdered M. australis leaves before ingestion of 75 g of sucrose' - n=12 is the denominator for every number, it is a with-vs-WITHOUT contrast with no placebo powder and no blinding stated, and the authors title it 'A Pilot Study'. ON-OBJECT RESULT IS ONE TIMEPOINT AND THE SPIKE STILL HAPPENED: glucose was sampled 'at baseline and 30, 60, and 120 min after ingestion', and 'Both trials caused a significant increase in blood glucose levels at 30 and 60 min after ingestion, with significantly lower values at 30 min after ingestion in the M trial than in the C trial' - so the finding is a blunted early peak in n=12, not an abolished excursion, with 60 and 120 min not reported as separated. No iAUC, no glucose concentrations, no effect size and no p-value appear anywhere in the abstract (cannot-tell, not absent). SUBJECT CAVEAT: the exposure is WHOLE POWDERED LEAF of Morus australis, a Ryukyu species, not the Morus alba standardised extracts the rest of this claim rests on, and NO DNJ CONTENT OR DOSE IS STATED - the DNJ attribution is inferential, resting on the authors' own prior characterisation, 'We previously developed an alpha-glucosidase-inhibiting tea powder from the leaves of Morus australis'. Quality held at low: n=12 pilot, no matched placebo, blinding unstated, glucose secondary to a vascular primary, no numbers on the on-object result, and the investigators developed the tested product; no funding or COI statement in the abstract (cannot-tell). Powder before a sucrose solution, not baked into bread - says nothing about DNJ thermal survival. INDEPENDENCE FLAG (not acted on): group 'kume' cites its OWN earlier demonstration of this same glucose effect with this same powder as background; that prior paper is not in the vault, so this claim holds one voice from the group today, but the predecessor must share group_id 'kume' if it is ever ingested. |
| Wang R, Li Y, Mu W, Li Z, Sun J, Wang B, Zhong Z, Luo X, Xie C, Huang Y 2018 · Medicine (Baltimore) | observational | supports | low | Full text (Medicine (Baltimore) 2018;97(34):e11996, PMC6113008, doi 10.1097/MD.0000000000011996; ChiCTR-IPR-15006484, registered 2015-05-28 for a trial run Jun-Dec 2014). DIRECTION SUPPORTS, BUT THE CLAIM-BEARING COMPARISON IS NOT RANDOMISED - regraded from RCT to observational on the exposure-randomised test. Single-center, randomised, OPEN-LABEL, 7-cycle self-controlled crossover; n=15 healthy normoglycaemic volunteers (9M/6F, 18-40 y, BMI 18-25, FPG 3.9-5.5 mmol/L, mainly TCM-university students), all 15 completing; '50 g of carbohydrate powder with or without 750 mg of MLE (equivalent to 7.5 mg of DNJ) was dissolved in 150 mL of drinking water'; capillary finger-stick glucose (OneTouch Ultra meter) at -15 to 120 min. THE HEADLINE IS A HISTORICAL CONTROL: 'Compared with previously reported GI values for the various carbohydrates in the absence of MLE (Table S5), the addition of MLE reduced the GIs for maltose, sucrose, maltodextrin, and glucose by 53.11%, 33.51%, 31.00%, and 8.12%, respectively' - and the authors' own second limitation is that 'measurements of the GIs of maltose, sucrose, and maltodextrin in the absence of MLE were not made, so direct comparisons of these GI values with those in the presence of MLE could not be made.' Sucrose, maltose and maltodextrin were NEVER given without MLE, so for three of the four carbohydrates the trial holds no randomised MLE-free comparator at all. The ONLY within-trial MLE-vs-no-MLE contrast is glucose+MLE (GI 91.88 +/- 27.24; IAUC 175.86 +/- 59.71 against the 3-visit reference mean 181.28, SD 10.73, CV 5.9%) - an 8.12% reduction reported with NO p-value, which the authors predict as a null: 'MLE had only a small effect on the absorption of glucose from the gastrointestinal tract, as expected since glucose is directly absorbed without the requirement of alpha-glucosidase.' The significant within-study result - 'At 30 minutes, the change in blood glucose level was significantly lower for sucrose + MLE (P < .01), maltose + MLE (P < .01), and maltodextrin + MLE (P < .05) than for glucose or glucose + MLE' - compares DIFFERENT SUGARS against glucose, not MLE against its absence, and sucrose's own published GI already sits far below glucose's. Open-label ('Neither the participants nor the investigators were blinded'), no insulin measured ('the effects of MLE on insulin secretion and sensitivity were not explored'), self-described as 'exploratory in nature and ... the results are tentative'. Sponsored by Botanic Century (Beijing) Pharmaceuticals, which prepared the extract and employs a co-author, with the corresponding author writing from a botaniccentury.com address, yet 'The authors have no conflicts of interest to disclose.' Numeric inconsistency in the paper: sucrose+MLE IAUC printed as 43.22 +/- 31.66, identical to its GI of 43.22, though 43.22/181.28 x 100 = 23.8. Powder in water, not bread - says nothing about bake stability. |
| Mela DJ, Alssema M, Hiemstra H, Hoogenraad AR, Kadam T 2024 · Nutrients | RCT | supports | moderate | OFF-SCOPE ON SUBJECT - this trial tested mulberry FRUIT extract, and never gave mulberry leaf in any arm. Full text (Nutrients 2024;16(14):2177, PMC11280168, doi 10.3390/nu16142177; NCT02256332, prospectively registered; clinical phase run 16 Feb - 16 Jul 2015 at three sites in India, published 2024). Randomized, double-blind, multi-center, 3-period balanced-order crossover. 'Of the 24 individuals entering the trial, 22 completed all treatments', and after blind-review exclusion of non-fasted baselines the PER-ARM analysis denominators in Table 2 are control 21, 0.37 g 22, 0.75 g 20 - not the headline 24. Population: unmedicated T2D adults (HbA1c 7.2 +/- 0.5%, controlled by diet and exercise only, mean age 51, BMI 27.5). Exposure: 0.37 or 0.75 g of a commercial mulberry FRUIT extract - 'The MFE (batch No. MF-DC-KQ-111207 Draco Natural Products Inc., San Jose, CA, USA) contained, by weight, 0.5% DNJ' (~1.85 and 3.75 mg DNJ) - stirred into boiled rice supplying ~50 g available carbohydrate. THE SUBJECT MISMATCH IS THE PAPER'S OWN FRAMING: 'In contrast, most research testing mulberry products for glycemic control have used mulberry leaf extract (MLE) with 2-10 times higher extract doses and DNJ contents'; the authors attribute the effect mainly to DNJ but concede 'It is, however, also possible that other iminosugars or minor components of MFE could also contribute to these effects', and warn that 'Comparisons between the MFE used here and studies using other mulberry extracts must be made with caution, however, due to differences in (often unreported) DNJ levels and overall product compositions'. ON ITS OWN (FRUIT) SUBJECT THE DIRECTION SUPPORTS, BUT ONLY AT THE HIGH DOSE: primary outcome PPG +iAUC 2hr fell 22.4% (95% CI -38.6 to -1.9) at 0.75 g, while the low dose was NOT separated from control - 8.2% (-20.8 to 6.6), CI spanning zero - as was its insulin figure, 9.6% (-20.7 to 3.0); only 0.75 g insulin cleared zero at -17.5% (-27.9 to -5.7). The prior extract read both doses as reductions and called the result 'dose-de[pendent]' without noting that the lower dose is a non-separation; the authors themselves say only that 'The effects were more consistent and robust for a dose of 0.75 g of MFE than a 0.37 g dose'. Insulin is tAUC (total, not incremental) and was a SECONDARY outcome. The effect is early and does not persist: 'There were limited effects of MFE on the PPG responses when summed over 3 and 4 hr'. NO P-VALUES ANYWHERE - 'This was a pilot study that was not planned or powered for formal statistical hypothesis testing (determination of p-values)'; inference rests on Dunnett-corrected 95% CIs alone. Tolerance was clean: 'There were no indications of adverse events or gastrointestinal discomfort'. 'This research was funded entirely by Unilever'; all five authors are or were Unilever staff, who 'contributed to the analysis and interpretation of data and the writing of this report' though blind to treatment codes until database unlock, and the datasets 'are not publicly available'. Vehicle was rice with extract stirred in, not bread - says nothing about DNJ thermal survival in baking. |
Disagree, or know a study we missed?
We grade by evidence, not opinions. The way to weigh in is to point us to a study we haven't cited (check the evidence table above first), or to flag a problem with one we have. Every submission is reviewed; if it holds up, the grade updates and shows in Science Changes Its Mind.
Opens a short form. You'll sign in with Google so submissions are tied to a real account — we don't display your identity, and we only accept a link we can verify (PubMed, DOI, ClinicalTrials.gov).
Educational only, not medical advice. Grades and scores reflect published evidence weighted by study design and quality; see the methodology.