SAMe (S-adenosylmethionine)

S-adenosylmethionine is the molecule the body makes from methionine to carry methyl groups, and it sits one step downstream of betaine in the pathway. That is how it reaches this guide at all — but the case for SAMe and cartilage is a separate one from the methylation story, and it rests on something much smaller than the pathway does.

The mechanism, and how far it has been taken

The case starts in cell culture and it has been carried further than it once was. Human chondrocyte-like cells exposed to SAMe at 10 micrograms per millilitre accumulated 10.6 percent more aggrecan by day seven and 31.3 percent more by day fourteen. What makes that more than a staining result is the control the same experiment ran in the other direction: cell proliferation did not rise at all, and fell at the concentration that worked best, so the extra matrix is each cell producing more rather than there being more cells. Gene expression of aggrecan, type II collagen, SOX9, CCN2 and the chondroitin sulfate synthesis enzymes rose alongside it.

The route was then identified rather than assumed. Blocking the enzyme that makes SAMe inside the cell, first with an inhibitor and again by knocking the gene down, lowered aggrecan and type II collagen expression, and adding SAMe from outside reversed it. Polyamine levels rose in treated cells, and inhibiting polyamine synthesis removed most of the effect on aggrecan — which is what identifies polyamine production as the mechanism rather than a correlate.

Two limits belong beside it. These are cultured cell lines, not diseased human cartilage, and a culture concentration has no established relationship to what a swallowed dose delivers to a knee. No animal model of osteoarthritis has been run on SAMe at all.

What the placebo-controlled trials found

Four randomised placebo-controlled trials of SAMe in knee or hip osteoarthritis exist, and Cochrane pooled them. Not every outcome draws on all four, which is worth knowing before reading the numbers.

Pain pooled two trials and 533 patients at a standardised mean difference of -0.17, on a 95% confidence interval from -0.35 to 0.01 — which the review translates as 0.4 cm on a 10 cm visual analogue scale. There was no disagreement between the two. Function pooled three trials and 542 patients at 0.02, on an interval running from -0.68 to 0.71.

The flagship trial supplies 96 percent of the weight in that pain estimate, and on its own it comes to -0.17 against placebo, on an interval from -0.35 to 0.02. It ran 489 patients on SAMe or placebo across 33 Italian centres for thirty days, with no other analgesic allowed. Its naproxen arm was excluded from the review, so nothing Cochrane reports here bears on the comparison with a drug. Its patients were not only knees and hips: 272 knees, 115 hips, 69 spines and 33 hands.

Cochrane rated the methodological quality and the quality of reporting poor, called its own finding inconclusive, and advised against routine use.

The trials that make up most of the literature

Nearly every other SAMe trial in this disease uses an anti-inflammatory drug as its comparator rather than a placebo. That is not an impression; it is what Cochrane found when it went looking. Twelve studies were excluded from its review for having only an active control — six of them against ibuprofen, and one each against naproxen, aspirin, indomethacin, piroxicam, celecoxib and sulindac. Add the naproxen arm it dropped from the trial it did include, and the shape of this literature is clear: it is much larger than four trials, and much less able to answer whether SAMe beats doing nothing.

Only one of those active-comparator trials has been checked against its paper, and it is the most recent. Sixty-one adults took SAMe or celecoxib in a sixteen-week crossover: celecoxib was ahead in the first month and the two were level by the second. That is the pattern the older trials are usually summarised as showing — SAMe arriving where an anti-inflammatory arrives, more slowly — and this is the one trial of them whose paper says so directly.

One trial does have both a placebo and an unusual route, and Cochrane describes it in detail. Patients received intravenous SAMe at 400 mg once daily for five days and then 200 mg orally three times a day, against a matching placebo regimen, over four weeks. It ran at two sites, and the randomisation — concealed, using a coded pharmacy and a random-number table, the best allocation procedure in the whole set — still produced markedly different patients at the two sites, so the investigators reported each site separately. Other analgesics were allowed, and their use differed between arms in opposite directions at the two sites. Cochrane's extraction of its pain outcome comes to -0.22, on an interval from -1.16 to 0.73.

The one time anyone imaged a knee

A 2023 pilot randomised 120 people with knee osteoarthritis to placebo or one of two strengths of a fixed combination — glucosamine sulfate, non-animal chondroitin sulfate and SAMe — for six months, with ultrasound cartilage thickness and pain as co-primary endpoints. It is the only trial in this entry to measure a joint structurally, and three things about it need saying together.

It was null between groups. The abstract reports a minor rise in cartilage thickness in the higher-dose arm; the results section states that at baseline, three months and six months there was no significant difference between groups in cartilage thickness at any measured site in either knee. Pain, WOMAC, the Tegner Lysholm score and SF-36 did not differ between groups either, and pain fell in all three arms — furthest in the placebo arm at three months.

Nothing in it can be attributed to SAMe. It is a three-way combination, and its SAMe dose was 100 or 200 mg a day against the 600 to 1,200 mg the osteoarthritis trials used. The study and the product were sponsored by the manufacturer.

What it does establish is that the measurement is feasible: the ultrasonographer's own repeat measurements agreed closely. A trial of SAMe alone, at a trial dose, with this endpoint, is a study somebody could run.

What most people take it for

The larger and better SAMe literature is about mood, and this guide covers it because that is what a person swallowing SAMe is usually relying on.

A systematic review of eight double-blind randomised trials, 1,011 subjects and eleven comparisons counts them rather than pooling them, which suits a literature this size better than an average would. Of five comparisons against placebo, SAMe was significantly better in three. Of four comparisons against imipramine or escitalopram, none separated in either direction. Of the two that added SAMe to an antidepressant already being taken, one found it accelerated the response to imipramine between days four and twelve with the scores level again by day fourteen, and the other found the combination better than a serotonin reuptake inhibitor alone. Side effects across all eight were mild, transient or not clinically relevant, and the reviewers call the picture encouraging while saying larger trials are needed before anyone concludes from it.

Those are mood endpoints rather than joint endpoints, so they inform this page without bearing on its cartilage tier. The scope point is worth making plainly: SAMe has better evidence for one thing and equivocal evidence for another, and this guide grades the second.

Practical notes

The placebo-controlled trials did not all use the same dose, which is worth knowing before reading a supplement label. Oral doses across the four were 200 mg three times a day in one, 400 mg three times a day in two, and 200 mg six times a day in the largest — 600 to 1,200 mg a day, always split. The review reporting them contradicts itself on the largest trial's dose, listing it as 400 mg six times daily in its narrative and 200 mg six times daily in its table of that trial; the table is the more detailed record and is what this page follows. Onset is slow — the celecoxib crossover found SAMe behind at one month and level at two — so a four-week judgement is likely to be an unfair one.

That crossover is also the one place this literature looked at what was actually in the capsule. Three quarters of the way through, a routine check found the study's SAMe had lost about half its potency, and the trial was paused until a fresh batch arrived. The trial could not show that the loss changed its result, and no other SAMe trial reports having checked at all.

The trials ran between four and sixteen weeks, and none ran longer. That is worth knowing about a supplement people take for years.

Tolerability is where this entry has most often been over-read, including here. Pooled across the four placebo-controlled trials, the risk of any adverse event on SAMe came to 1.27 times placebo — 79 events on SAMe against 59 on placebo, on an interval from 0.94 to 1.71 — and the risk of withdrawing because of one to 0.94 times. In the largest trial the split was 57 of 236 on SAMe against 42 of 220 on placebo. So the statement available is that SAMe is not clearly worse tolerated than placebo, not that it is equal to it, and certainly not that it is better.

That is also a finding about minor side effects in short trials. It is not a finding about serious ones, and the reason is worth stating plainly: none of the four placebo-controlled trials described serious adverse events at all. Cochrane's own review names that as a concern.

One caution this guide cannot source and therefore does not state as a finding: the mood literature is the reason SAMe gets discussed alongside antidepressants, and the interaction questions that follow — serotonergic drugs, and use in bipolar disorder — are not addressed by any record here. Anyone on psychiatric medication needs a clinician rather than a page.

What would move the tier

A modern, adequately powered, placebo-controlled trial. The whole placebo-controlled evidence base predates 1995 and was rated poor by the people who pooled it, so the question of whether SAMe beats nothing for a knee has not so much been answered as never properly asked — and a compound that holds its own against celecoxib is worth asking properly about.

The more interesting study is smaller and cheaper. The proteoglycan work has been carried as far as a named mechanism in cell culture and no further: no animal model of osteoarthritis has been run on SAMe, and the one human imaging trial gave it in a three-way combination at a sixth of the trial dose. An animal model with a cartilage endpoint, or a human trial of SAMe alone with a biomarker like CTX-II, would say whether there is anything structural here at all — and either result would be worth having.