Evidence Trace, complete trial record
In plain terms
In plain terms: SAM-e (S-adenosyl-L-methionine) shows early signals in human research but the evidence is limited -- small trials, surrogate endpoints, or not yet independently replicated. It is an emerging area, not yet a proven one.
SAM-e (S-adenosyl-L-methionine)
C
Limited: small, mixed or unreplicated trials
The endpoint this grade is based on: Validated scale scored by a clinician blinded to treatment, Methyl donor marketed for depression and for osteoarthritis joint pain. Use-specific: as a stand-alone treatment for depression the pooled randomized trials disagree (Grade C); added to an antidepressant, and for osteoarthritis, the pooled trials did not separate from placebo (Grade D).
The evidence as the record states it
Limveeraprajak et al. (Prog Neuropsychopharmacol Biol Psychiatry, 2024; PMID 38423354): 23 randomized trials, 2,183 patients; SAM-e alone beat placebo (SMD -0.58) while SAM-e added to an antidepressant did not. Galizia et al. (Cochrane Database of Systematic Reviews, 2016; PMID 27727432): 8 randomized trials in depression, 934 adults, no strong evidence of a difference from placebo as monotherapy. Peng et al. (General Hospital Psychiatry, 2024; PMID 38199136): 14 trials in depression, 1,522 subjects, no significant difference from placebo as monotherapy or as adjuvant therapy, and no difference from imipramine or escitalopram either; equivalence to an active drug is not proof of efficacy when neither beat placebo in the modern trials. Cheng et al. (Psychol Med, 2025; PMID 40314175): network meta-analysis of 192 nutraceutical trials, SAM-e added to an antidepressant ranked effective, a pooled ranking that the direct add-on trials below do not support. Mischoulon et al. (Journal of Clinical Psychiatry, 2014; PMID 24500245): SAM-e versus escitalopram versus placebo in depression, no arm separated from placebo. Sarris et al. (Psychopharmacology, 2020; PMID 31712971): 49-person pilot, 800 mg/day for 8 weeks, a 3.76-point MADRS difference favoring SAM-e (effect size 0.72) that was not significant (p=0.13), with a 53 percent placebo response. Papakostas et al. (Am J Psychiatry, 2010; PMID 20595412), Targum et al. (J Psychiatr Res, 2018; PMID 30368163) and Sarris et al. (Eur Neuropsychopharmacol, 2018; PMID 30115553): the add-on trials, one early positive and two null; Alkema et al. (Epigenomics, 2026; PMID 42027147): 31 adults with trauma-related depression, 1,200 mg added to trauma-focused therapy, no added clinical benefit, a third null. Rutjes et al. (Cochrane Database of Systematic Reviews, 2009; PMID 19821403): 4 trials, 656 patients with knee or hip osteoarthritis, small non-significant pain difference on self-reported pain scales; Soeken et al. (J Fam Pract, 2002; PMID 12019049): no pain benefit over placebo in 2 trials, a functional limitation signal from the same 2 trials, and comparability with NSAIDs. Grade C for monotherapy in depression, mixed pooled evidence; D on Route 2 for add-on use and for osteoarthritis.
PMID 27727432 PMID 38199136 PMID 19821403 PMID 24500245 PMID 31712971 PMID 38423354 PMID 20595412 PMID 30368163 PMID 30115553 PMID 40314175 PMID 42027147 PMID 12019049
Evidence by specific use
Major depressive disorder, SAM-e as monotherapy
Grade C
Pooled analyses disagree. Limveeraprajak et al. (Prog Neuropsychopharmacol Biol Psychiatry, 2024; PMID 38423354): 23 trials, 2,183 patients; SAM-e alone beat placebo (SMD -0.58, 95% CI -0.93 to -0.23, 11 placebo trials, high heterogeneity), rated moderate certainty by the authors. Galizia et al. (Cochrane, 2016; PMID 27727432): 8 trials, 934 adults, no strong evidence of a difference from placebo. Peng et al. (Gen Hosp Psychiatry, 2024; PMID 38199136): 14 trials, 1,522 patients, no significant difference from placebo as monotherapy. The best modern trials are null: Mischoulon et al. (J Clin Psychiatry, 2014; PMID 24500245), SAM-e against escitalopram against placebo, no arm separated; Sarris et al. (Psychopharmacology, 2020; PMID 31712971), non-significant. Positive and null pooled analyses side by side on a clinician-rated scale is mixed randomized evidence: C, not D on Route 2. The two null pooled analyses are carried on this use as tested-no-benefit payloads beside the positive one; the letter is C because they disagree, not because either is discounted. Sarris 2020 was a 49-person pilot with a 3.76-point MADRS difference favoring SAM-e that did not reach significance against a 53 percent placebo response, and Peng 2024 also found no difference from imipramine or escitalopram.
PMID 38423354 PMID 27727432 PMID 38199136 PMID 24500245 PMID 31712971
Major depressive disorder, SAM-e added to an antidepressant
Grade D
Tested as an add-on and found no benefit. Papakostas et al. (Am J Psychiatry, 2010; PMID 20595412): 73 non-responders, 6 weeks, higher response and remission on SAM-e; the authors called the result preliminary. Three later trials are null: Targum et al. (J Psychiatr Res, 2018; PMID 30368163), 234 patients, no difference on any of three rating scales; Sarris et al. (Eur Neuropsychopharmacol, 2018; PMID 30115553), 107 non-remitters, no between-group difference; and the pooled add-on comparison in Limveeraprajak et al. (2024; PMID 38423354; SMD -0.22, 95% CI -0.63 to 0.19) and Peng et al. (2024; PMID 38199136) is not significant. One early positive trial and three null trials, pooled null: D on Route 2. A 2026 trial of 1,200 mg added to trauma-focused therapy in 31 adults found no added clinical benefit (Alkema et al., Epigenomics, 2026; PMID 42027147). A 2025 network meta-analysis ranked the add-on effective by indirect comparison (Cheng et al., PMID 40314175); the direct trials, including the largest, are null, and direct evidence governs the letter.
PMID 20595412 PMID 30368163 PMID 30115553 PMID 38423354 PMID 38199136 PMID 42027147 PMID 40314175
Knee or hip osteoarthritis, pain and function
Grade D
Tested without benefit on pain. The Cochrane review pooled 4 placebo-controlled trials, 656 patients: pain SMD -0.17 with a confidence interval crossing zero, no effect on function, poor trial quality (Rutjes et al., 2009; PMID 19821403). Soeken et al. (2002; PMID 12019049) likewise found no pain benefit over placebo in 2 trials, with a functional limitation signal from those same 2 trials. Pain on a visual analog scale is self-reported. D on Route 2.
PMID 19821403 PMID 12019049
Why Grade C Route by_use and not Grade B
Grade C reflects early or limited human evidence -- small trials, surrogate endpoints, or preliminary findings not yet independently replicated. The evidence points in a promising direction but does not yet meet the standard for Grade B. A single well-powered, independently replicated trial with a meaningful endpoint would move this to Grade B or higher.
What this grade covers and what it does not
Grade C, the record's best use: SAM-e as monotherapy for major depressive disorder, where pooled analyses of randomized trials disagree. Limveeraprajak et al. (Prog Neuropsychopharmacol Biol Psychiatry, 2024; PMID 38423354) pooled 23 trials, 2,183 patients, and found SAM-e alone beat placebo (SMD -0.58, high heterogeneity, moderate certainty by the authors); the Cochrane review (Galizia et al., 2016; PMID 27727432; 8 trials, 934 adults) and Peng et al. (Gen Hosp Psychiatry, 2024; PMID 38199136; 14 trials, 1,522 patients) found no significant difference from placebo. The best modern trials are null: Mischoulon et al. (J Clin Psychiatry, 2014; PMID 24500245), SAM-e against escitalopram against placebo, all three arms improved 5 to 6 points on the HDRS-17 and neither drug beat placebo, and Sarris et al. (Psychopharmacology, 2020; PMID 31712971), non-significant. Positive and null pooled analyses side by side on a clinician-rated scale is mixed randomized evidence, which the written rule calls C. WHY NOT D: Route 2 requires that the trials did not show benefit, and one moderate-certainty pooling says they did. WHY NOT B: the pooled analyses contradict each other and the largest modern trials are null. TWO USES ARE D ON ROUTE 2 AND STAY THERE: SAM-e added to an antidepressant, where Targum et al. (J Psychiatr Res, 2018; PMID 30368163; 234 patients) and Sarris et al. (Eur Neuropsychopharmacol, 2018; PMID 30115553; 107 patients) were null and the pooled add-on comparison is not significant, against one early 73-person trial (Papakostas et al., Am J Psychiatry, 2010; PMID 20595412); and knee or hip osteoarthritis, where the Cochrane review (PMID 19821403; 4 trials, 656 patients) found a pain difference of about 0.4 cm on a 10 cm scale with a confidence interval crossing zero. WHAT THIS GRADE DOES NOT COVER: intravenous or parenteral SAM-e, and its use in liver disease, where the endpoints studied are biochemical rather than clinical. Reading rewritten October 2026 to match the letter the record has held since October.
What would move this grade
A well-powered randomized trial with an objective, clinically meaningful primary endpoint and at least one independent replication would support Grade B or higher.
How to track whether this is working
The trials measured depression with a repeated validated rating scale (HDRS-17 or MADRS) and osteoarthritis with a pain and function score. There is no blood marker that reports whether supplemental SAM-e is working.