Scoreboard
Nothing has resolved yet. The first readout we are waiting on is due Q4 2026. A prediction only moves off Open when the trial reports the endpoint we named and we publish a verdict, win or lose.
Jump to prediction
Predictions on record
L-citrulline for erectile dysfunction
The standalone L-citrulline arm of NCT07642505 will not reach a clinically meaningful IIEF gain, roughly 4 points over placebo, even if the result comes back statistically positive.
- Readout expected
- Q4 2026
- Scored on
- IIEF change versus placebo, measured against the roughly 4-point clinical threshold set by the Menafra 2022 arginine benchmark.
Why we expect this
This is the first powered double-blind citrulline trial in erectile dysfunction, but the older arginine data already sets the bar for a meaningful change. A positive result under that bar is a partial win at best, not the retail breakthrough the category will announce.
What would prove us wrong
The standalone citrulline arm reports a mean IIEF improvement of about 4 points or more versus placebo.
Collagen peptides and “Ozempic face”
Posted results for NCT06787924 will not validate consumer claims about collagen for “Ozempic face” or skin laxity. The registered endpoints are scar and wound healing plus serum hydroxyproline, not facial volume.
- Readout expected
- 2026
- Scored on
- Registered primary endpoints (scar and wound healing, serum hydroxyproline) versus the consumer outcome the trial is being sold as.
Why we expect this
The trial completed without posted results as of August 2026, and the endpoint mismatch is already documented in our guide. We lock the call before results land so the record cannot be written after the fact.
What would prove us wrong
Posted results include a facial volume, laxity, or skin elasticity endpoint that was registered before enrolment closed, and it improves versus control.
Urolithin A (Mitopure) and sleep claims
NCT07060898 will not produce proof that Mitopure improves sleep. Sleep sits in this trial as a secondary self-report questionnaire inside a brain-health app study, not as measured sleep.
- Readout expected
- Q2 2027
- Scored on
- Primary endpoints are brain-health app measures. Sleep is a secondary questionnaire, never polysomnography.
Why we expect this
Our guide documents the endpoint hierarchy for this study. Marketing already cites “650 participants proved sleep” before any readout exists. We expect the primary analysis to stay on brain-health measures.
What would prove us wrong
Published results report objective sleep (polysomnography or validated actigraphy) as a pre-registered primary or co-primary endpoint, and it improves versus placebo.
Retatrutide lean-mass share at 30% weight loss
When the TRIUMPH-1 primary results publish in a peer-reviewed journal, the proportion of total weight lost as lean mass will be approximately 25 percent (the same fat-to-lean split reported for tirzepatide in SURMOUNT-1), meaning retatrutide's glucagon-receptor component does not spare lean mass relative to other incretins despite the mechanism's thermogenic rationale.
- Readout expected
- H2 2027
- Scored on
- DXA-measured fat mass and lean mass change at 80 weeks, reported as lean-mass share of total weight lost, in the TRIUMPH-1 primary publication.
Why we expect this
The Coskun phase 2 DXA substudy (n=103 paired scans, PMID 40609566) states that retatrutide's lean-mass proportion is "similar to other obesity treatments" without publishing the number. The tirzepatide SURMOUNT-1 DXA substudy (PMID 39996356) found a stable 75-to-25 fat-to-lean split across all doses and placebo. If retatrutide followed the same pattern at its larger absolute weight loss, 30% body-weight reduction would produce roughly 7.5% lean-mass loss, a clinically meaningful amount for sarcopenia-risk patients. The glucagon-receptor mechanism (energy expenditure, lipolysis) argues for possible lean sparing; the existing human data argue against it. We are placing our marker on the null: no meaningful advantage.
What would prove us wrong
The TRIUMPH-1 publication reports a DXA-verified lean-mass share below 20 percent or above 30 percent of total weight lost, or demonstrates a statistically significant lean-sparing advantage versus tirzepatide's published ratio.
Berberine-stack product trial on semaglutide
The first registered trial combining a berberine-containing product with semaglutide (NCT07195994, AMPK Charge+ by QuickSilver Scientific, estimated n=90, 84 days) will not demonstrate an incremental glycemic benefit attributable to adding the product on top of semaglutide, because the design has no semaglutide-only arm: it compares the product alone against the product plus semaglutide.
- Readout expected
- 2027
- Scored on
- Between-arm difference in change of fasting glucose, fasting insulin, and HbA1c at day 84, plus whether any publication frames the result as evidence for berberine stacking on GLP-1 therapy.
Why we expect this
Industry-sponsored (QuickSilver Scientific as lead, KGK Science as CRO), single-blind, NA-phase product trial. Arms are product versus product plus semaglutide with no semaglutide-only control, so any between-arm difference mixes the medication effect with the stack effect. Registry composition: phospholipids, diindolylmethane, quercetin dihydrate, milk thistle, resveratrol, and berberine in a liposomal matrix. Whatever publishes, marketing claims that berberine works with Ozempic will outrun the design; this entry exists to score that gap.
What would prove us wrong
Published results isolate a clinically meaningful incremental effect of the stacked product on fasting glucose, fasting insulin, or HbA1c beyond what semaglutide monotherapy would explain, and the paper attributes it to the combination.
Fiber tolerance on weight-loss meds
In the completed crossover trial of inulin, a fiber blend, and resistant starch type 4 in adults using weight-loss medications (General Mills collaborator), added fibers will not worsen the composite GI symptom score versus the low-fiber control, differences between fiber types will be small, and no arm will show a large symptomatic improvement.
- Readout expected
- 2027
- Scored on
- Composite GI symptom score across the measurement windows, by fiber type versus the low-fiber control arm.
Why we expect this
Our companion guide recommends soluble fiber for GLP-1 constipation while warning against stacking gut-slowing products. This trial is the first controlled GI-symptom readout in people on weight-loss medications; registration completed in November 2025 with results still unpublished, so the prediction scores against the eventual publication, not the registry entry.
What would prove us wrong
Any fiber arm shows a large composite-score improvement (roughly 30% or more versus control) or a statistically significant worsening.
Creatine for GLP-1 lean-mass preservation
In the University of Saskatchewan pilot of creatine 10 g per day with resistance training during GLP-1 receptor agonist therapy, the creatine arm will show a modest functional gain on the primary sit-to-stand endpoint, but the pilot will be underpowered for a statistically significant lean-tissue mass difference versus placebo.
- Readout expected
- 2028
- Scored on
- Sit-to-stand repetitions at 12 weeks (primary endpoint) and any reported lean tissue mass change with between-arm statistics.
Why we expect this
Our companion guide's line that creatine has not been proven in a direct randomized trial to prevent GLP-1-related lean-mass loss meets its first direct test here. Mechanism plus training argues for the direction; pilot size and a 12-week window argue against a powered lean-mass endpoint. The academic sponsor lowers the odds of outcome-switching inflation.
What would prove us wrong
The trial reports a statistically significant between-arm difference in lean tissue mass favoring creatine.
Protein intake during GLP-1 therapy
Middle-aged women on GLP-1 medication randomized to a structured higher-protein diet (largely pork products; National Pork Board collaborator) will preserve more muscle mass at 12 weeks than a GLP-1-only group, with a small-to-moderate effect size; the funder's framing will present the upper bound of that range.
- Readout expected
- 2028
- Scored on
- Between-group change in muscle mass (primary endpoints: body composition, muscle mass) at 12 weeks, as reported against the GLP-1-only arm.
Why we expect this
Protein is the companion framework's highest-leverage recommendation, and this is its first head-to-head test in GLP-1 users. The direction matches generic protein evidence; the magnitude over 12 weeks should stay modest, and pork-checkoff funding will shape how generously the abstract reads the same numbers.
What would prove us wrong
The trial finds no between-group muscle mass difference at 12 weeks.
How scoring works
A prediction is only worth reading if it can fail. Each entry carries a threshold and a registry ID before the readout, and one of five statuses after it.
- Open
- Published and waiting. The trial has not reported the endpoint we named.
- Hit
- Results landed and the claim held, including the threshold we set in advance.
- Miss
- Results landed and the claim failed. The entry stays up with what we got wrong.
- Partial
- Direction was right, but a number or an endpoint we named did not match.
- Void
- Scoring became impossible: the trial stopped, changed endpoints, or never posted.
Questions about the ledger
What is the Supplement Brief predictions ledger?
A public list of dated, falsifiable calls about clinical trials that have not reported yet. Each entry names the trial, the endpoint that decides it, and the result that would prove us wrong, so the call cannot be rewritten after the readout.
How is a prediction scored?
When the trial reports the endpoint we named, we compare the result against the threshold written in the entry. The status becomes Hit, Miss, Partial, or Void, and the scoreboard at the top of the page updates with it.
What happens when a prediction turns out to be wrong?
The entry stays online and is marked Miss with a resolution note explaining what we got wrong. Nothing is deleted or quietly edited, and the miss counts against our hit rate.
Is this investment or medical advice?
No. Predictions here are about what a trial will show, not about what anyone should take. Dosing and safety guidance lives in our guides and always belongs with a clinician.
Where this sits in our editorial record
The corrections log records mistakes we made in published guides and how we fixed them. This ledger runs in the other direction: forward-looking trial calls we are willing to be scored on in public. Our methodology explains how evidence gets into a guide in the first place.
Spotted a readout we should be scoring, or think a call here is already wrong? Email [email protected] with the registry ID.