Build strength

An umbrella review inherits the primary trials' limitations and cannot fix them, and moderate certainty still means the true effect may differ substantially from the estimate. No pooled effect sizes appear, only the gradings, so nothing here says how big the reduction is. Searches ran only to May 2022.

The lean-mass figure is the solid one; the fat-loss figure ran anywhere from 1.34kg down to 0.11kg, close enough to nothing at one end that the effect there is unsettled. Everyone pooled here was under 50. In a small share of the trials, how people were assigned or assessed left room for the result to tilt.

The pooled training ran about 12.6 weeks on average, so nothing here shows whether the gains hold over a year. The pool of qualifying trials is small, and the authors ask for more work in this age group before the functional gains are treated as durable.

These trials were built to test performance rather than to hunt for harm, so anything rare or slow enough to slip past them would have slipped past. The adverse-event reports are filed voluntarily, which distorts the count in both directions. Average intake across the pooled trials was about 12.5g of creatine a day.

Twenty of the 23 trials in one analysis enrolled only men, and just 40 women appear across its two female-only trials, so the weaker showing in women reflects how little has been measured rather than a settled absence. Gains were most consistent in younger adults. And these are pooled averages, not what any one lifter should expect.

Between two and five trials per outcome, so a real effect could still be hiding in an evidence base this thin. What argues against that is how closely the trials agreed with each other: they scattered almost not at all around the null. The nicotinamide riboside results could not be pooled and ran in both directions, longer six-minute walks in peripheral artery disease against worse physical performance scores in mild cognitive impairment.

In about 80% of the pooled trials people were already eating at least 1.2 g per kg a day, so this measures going from enough to more, not from too little to enough. The 1.62 figure comes with a range running from 1.03 to 2.20, wide enough to be a region rather than a line. Most trials ran six to twelve weeks. Effects shrank with age and were larger in people already training.

A null result is harder to establish than a positive one, and this one carries the weight of the finding. Strength was pooled across one-rep-max tests, isometric tests and handgrip, which do not always move together. The spline curves are smooth in a way the underlying trials are not. The search covered two databases.

Most of these trials enrolled men by age rather than by a low testosterone reading, so many participants were not deficient. A gain in lean mass is not by itself a gain in anything a man would notice. In the 16-trial pooling, whose men averaged 77 years old, no musculoskeletal difference from placebo was left by 12 months, which could mean the effect fades or only that fewer trials reported that far out.

Baseline protein intake was not consistently controlled across the trials, so some of these people may already have been eating enough for a supplement to add nothing. The handgrip analysis was highly inconsistent between studies. Appendicular muscle mass did improve, but on two trials that disagreed sharply with each other. Older adults only, so this says nothing about young trained people.

This measures what was in the blood, not whether anyone took a supplement, so a powder started after the draw would be invisible. The intervals are wide enough to include modest risk either way, leucine running from 0.57 to 1.33. Estimates below 1 are not evidence of protection.

Only five of twelve studies could be pooled, and the main analysis compares before with after inside the supplemented group rather than against a control, so it cannot separate the supplement from everything else that changed. Cirrhosis is a malnourished state. None of this transfers to a well-fed person.

A food-composition database, not a study: it says how much is in what, and nothing about what these amino acids do. Values are averages across sampled products and shift with cut, brand and cooking loss. The figures are per 100 g, not per portion.

Ten men, one four-hour window, and all of them already trained. The comparator was placebo, so this shows branched-chain amino acids beat nothing rather than that they match protein. The authors put the rise at roughly half what intact protein produces in similar designs, which this study never measured directly.

An overview of reviews inherits every weakness of the trials underneath it, and the same small primary studies are counted inside several of the eleven. The authors grade those reviews and report the recurring flaws: no justification for excluding studies, and no funding disclosure for the primary trials.

Fifteen men per group, healthy and aged around 71, at rest with no exercise. The doses are not matched: 6 g of branched-chain amino acids against 30 g of milk protein is the choice a shopper faces, not a gram-for-gram test of which does more. Nothing here was measured beyond five hours.

One real exception sits inside the largest of these studies: among American footballers alone, sweat measurements did carry a weak signal, and in the other ten sports none did. Beyond that, cramping was self-reported throughout, and serum electrolytes were sampled before and after a race rather than inside the muscle at the moment a cramp struck, which is where the theory actually locates the problem. Forty-three crampers is enough to rule out a dominant effect, not a small one.

Twenty crampers is a small group, and the finish-line enzyme gap is borderline at p = 0.034 with spreads wide enough to swallow it. The blood was drawn after the cramp, so muscle damage could as easily follow a cramp as cause one. Crampers and non-crampers only diverged in pace from 25 kilometres onward, which points back at effort. The paper's own abstract calls 20 of 88 a rate of 24 percent; it is 22.7 percent.

This is a second analysis of the same 31 participants, not an independent replication, so it inherits that trial's size and design unchanged. A null result across the markers measured narrows the candidates rather than showing that no mechanism exists. Muscle fibre size specifically was unaffected, which sits awkwardly next to the whole-muscle scan difference the earlier report described, and the authors do not reconcile the two.

The abstract never says how many men took part, which is the first thing you would want from a dose-response study. The adverse signal is a blood chemistry change, not a symptom anyone reported. Healthy young men, fed briefly, so nothing here reaches long-term intake.

Nine trials is a small pool and the searches ended in December 2013, so a decade of later work is missing. The two outcomes are measured on different clocks: synthesis takes hours, lean mass takes weeks to months, so part of the contrast is a contrast of study durations. Older adults only.

Eight men per arm, one acute measurement, nothing past four and a half hours and nothing on muscle size. Part-funded and co-authored with a commercial protein-beverage research centre, a direct interest in a result about beverage formulation. A fifth arm added isoleucine and valine, and its value was never published.

Twelve men over 21 days is a mechanistic study, not an efficacy trial. What rose was the NAD metabolome, the family of related compounds including clearance products; the paper does not report that muscle NAD itself rose in proportion. The gene-expression changes ran downward in the energy and mitochondrial pathways, the opposite direction to the one the rationale predicts. One author invented the licensed patents behind the product and holds stock in the company selling it.

Expressed per kilo of whole body mass, the widely quoted 0.40 against 0.24, the same comparison reached only p = 0.055. This pools data from studies run earlier for other purposes rather than a trial built for the question. Men only, in both age groups. It measures a building rate over hours, not muscle held onto over years.

A position stand grades existing literature and computes no pooled estimate of its own, so these ranges are considered judgements rather than measurements. The society receives industry support and several authors declare commercial ties to supplement companies, which is worth weighing against recommendations that favour supplementing. It was published in 2017, before the large-bolus work and before trials that compared timings head to head.

One acute recovery period with a tracer, not muscle measured over a training block. Thirty resistance-trained men, no women and no older adults. Only 20 g and 40 g were compared, so this shows the per-meal ceiling is above 20 g without showing where it is. The advantage of 40 g over 20 g was real but small, 0.059 against 0.049 percent per hour.

Only the immediate before-versus-after contrast has been tested this way; nobody has randomised protein eaten hours from training. The direct-comparison review had five studies to work with, and in it lower-body strength did favour eating beforehand at p = 0.048, before the subgroup test declined to confirm it. The ten-week trial had 21 men, too few to prove equivalence. The author group behind two of these is publicly associated with the position they support.

Twelve men per group is not enough to rule out a modest real difference between the two proteins; the comparisons came out at p = 0.69 and p = 0.44. One night, no training outcome, no women. The exercise was endurance rather than resistance training, so the mitochondrial finding does not carry straight over to lifting. Both proteins were measured against a non-caloric placebo, so part of what is captured is protein against nothing.

Twenty people per arm across four arms, analysed per protocol rather than by intention to treat, in healthy adults over sixty days. The first author works for an NMN company and two co-authors for NMN raw-material producers. The trial's biological-age outcome comes from a free online calculator rather than a validated clinical instrument.

Whether NMN helps people walk further is unsettled. The one trial that found it was run by companies selling NMN and tested healthy middle-aged adults; a later pooling in older adults did not repeat it.

The pooled trials were short and produced only 3.55kg of loss, so those kilograms do not scale up to someone losing a seventh of their body weight; the proportion is what carries over. Lean mass as a share of body weight did not change. Not one trial in the pool measured strength, mobility or falls, so whether the loss costs anything a person would notice is untested.

Whether losing lean mass on semaglutide costs anything a person would notice has not been measured either way. Trials rank it weak at holding lean mass; one study with no comparison group says it was preserved.

Only one of the seven handgrip trials found a gain on its own; the pooled significance comes from combining mostly null results, over an interval running from 0.17 to 3.0 kgf. The leg-press result sits exactly at p = 0.05. Physical performance, the outcome closest to something a man would notice, did not improve, and rested on only two comparable trials.

Whether testosterone makes men stronger is unsettled. The pooled gains rest on mostly null trials and other studies found none, but a knee, a hand squeeze and a leg press measure different things and can disagree without any being wrong.

Eleven people took acetaminophen in this trial, out of 36 across three arms, and one muscle group was measured. Healthy older adults on a supervised programme are not a general population and not athletes. The dose was 4 g a day, the labelled maximum, which carries liver considerations this trial was neither designed nor powered to detect. The drug did nothing to the untrained hamstrings, so whatever is happening depends on the training rather than on the drug alone.

Ten men, one four-hour window, no size or strength endpoint. A p value of .039 and an interval running from -0.002 to 0.028 do not agree, and the interval is the more cautious reading. Same laboratory and near-identical design to the same group's 2017 trial, so the two are not independent.

Genetic proxies for circulating levels are not a stand-in for swallowing a supplement, and neither study gave anyone one. Both draw on European-ancestry samples, and the UK Biobank cohort is healthier than the general population. Neither rules out a smaller effect running the other way.

Fourteen of the 35 trials carried some risk of bias, and the authors report high heterogeneity. Older adults, many of them at risk of undernutrition, so the baseline protein intake is not a well-fed lifter's. Physical performance sat exactly on the significance line, an interval starting at zero.

Weak discrimination is not a screening test, and the authors do not offer it as one. Cramp history was self-reported, and sweat was collected during ordinary training, never during a cramp. Forearm patches estimate whole-body sweat sodium indirectly and carry real measurement error. One other significant result among roughly fourteen comparisons, potassium content in women, is what chance produces at that many tests.

Thirty-one people in one trial and thirty-six in the other, one muscle group each. The comparators were not the same: the younger trial used low-dose aspirin rather than a true placebo, so it set a high dose against a low one rather than a drug against nothing. The training protocols and durations differed too, eight weeks against twelve. The older trial's authors state the direction was the opposite of what they expected.

Ibuprofen during strength training roughly halved muscle growth in young adults and added to it in older ones. Age, comparator, training and programme length all differed, so what flips the effect is unknown.

These numbers come from a programme run under the International Council on Amino Acid Science, a nonprofit funded by the amino acid industry. The review says so itself. Short-term trials in healthy adults only, so nothing about long-term intake or about people with liver or kidney disease.

A secondary outcome is exploratory unless it was named in advance and corrected for testing many things at once, and this paper reports no such correction. Three of the four authors work for Meiji Holdings, which sells NMN. The authors highlight that walking-time changes tracked changes in blood NAD, 2PY and 4PY, which is an association inside one trial rather than evidence that raising NAD caused the walking.

Thirteen people is very small even for a crossover design, which makes the two positive results the fragile half and the long list of nulls the sturdy one. The fat-free mass change was one of many outcomes tested, exactly the setup in which a chance finding appears. Six weeks, in healthy overweight and obese volunteers.

The randomised evidence applies only to people whose kidneys were healthy at the start, and early kidney disease is common and silent. Diet in the cohort came from food frequency questionnaires, which are imprecise, and its top quartile is a moderate Korean intake rather than a deliberate high-protein diet. In the cohort, a rapid decline meant the estimated filtering rate falling by more than 3 mL a minute each year, a research threshold rather than a diagnosis of kidney disease. In a sixteen-week trial at 3.2 g per kg a day, kidney and liver markers rose within the high-protein groups even though the comparison between groups came out level.

Short trials show filtration rising without declining; a long cohort saw a signal but read kidney function off creatinine, which protein itself raises. Nothing here shows harm, and nothing establishes safety over decades.

All four are acute tracer studies in young trained men, measuring a building rate over hours rather than muscle held onto over months, and the two do not always track. The original plateau rests on telling a 20 g response apart from a 40 g one in six people. The 100 g result comes from a four-tracer method used by very few laboratories. Nobody tested whether the same protein split across the day would do as well.

A 20 g per-meal ceiling shows up four hours after a one-leg session and not twelve hours after whole-body work, so it belongs to the measurement. No trial has fed large doses for months and weighed the muscle kept.

These measure a building rate across one night, not muscle retained across a training block. The forty-gram result in older men reached p = 0.02, and its supporting tracer analysis only p = 0.08. The founding study's clearly significant effects were whole-body, which includes gut and liver protein, not muscle specifically. Young men in two of the three trials, older men in the other, no women anywhere.

The trials for and against bedtime protein came from one laboratory, so dose likely explains the split: the effect appears somewhere between 30 and 40 g. Where exactly, and whether any muscle is kept, is untested.

Vitamin C and vitamin E were given together, so neither can be credited or blamed on its own. The marker changes carry enormous spread: the placebo group's rise averaged 59% with a standard deviation of 97%. Reviewers add that the blunting is inconsistent and rarely reached in ordinary training.

Both trials used people who were not training beforehand, who gain readily from the training alone and so make group differences hard to see. In the nine-month trial everyone ate 1.1 to 1.4 g per kg a day, below the level where gains flatten, so it may be measuring a shortfall corrected. Leucine-matching meant unequal grams, 26 g of soy against 19 g of whey. Whey-industry funding is common across this literature.

Whey beat soy matched on grams, tied matched on leucine, pointing at leucine. That stays open because in the grams trial soy gained less lean mass than a carbohydrate control, an unlikely result, so its gaps may be noisier than they look.

There was no control group and no randomisation, so nothing separates the drug from the specialist nutrition care these patients also received, or from people simply moving more as they got lighter. Grip strength rising during weight loss is partly expected because a lighter body is easier to move, and the study did not adjust for that. Mean body mass index was 46.3, far heavier than the trial populations. Calling 3kg preserved is the authors' reading of a real loss.

What this study can show is how lean mass fell over time, stopping by month seven. With nobody off the drug measured alongside, it cannot show that semaglutide, rather than anything else, preserved it.