Reading the L-Theanine Evidence: Part Three: Stress, and the Combination Problem

Abstract

Every review in this series leans on the same background assumption: that L-theanine's effect on stress is the well-established part, and that its effects on attention and sleep may follow from it. The stress literature is where the compound is supposed to be strong.

This part reads the four most recent trials to test that assumption. Two were funded by companies selling the ingredient, one by a pharmaceutical company, one by nobody. Two were double-blind, one single-blind, one open-label.

A pattern emerges that is worth naming in advance: the more tightly a trial isolates L-theanine and the better it controls expectation, the smaller the result becomes.

1.The Cleanest Design, and Its Primary Outcome

Marc Moulin, Erin Lewis and colleagues at KGK Science in Ontario ran a randomised, double-blind, placebo-controlled trial of 400 mg a day of a branded L-theanine for 28 days in 30 healthy adults with moderate stress, defined by a Perceived Stress Scale score between 14 and 26. It is registered on ClinicalTrials.gov and reported to CONSORT standards.

The primary outcome was pre-specified: change in salivary cortisol at days 14 and 28, theanine against placebo. The result was null. There were no significant differences between groups for change in salivary cortisol.

The abstract nonetheless reports that the supplemented group had decreases of 12.92 per cent and 17.98 per cent in Perceived Stress Scale scores at days 14 and 28. Both figures are within-group changes from baseline. Placebo, in the same sentence, decreased 9.74 per cent and 17.88 per cent — and the placebo result at day 28 has the stronger p-value of the two.

There were no significant differences between groups on that scale either. The paper says so.

On the second stress instrument the direction reverses. Total score on the Depression, Anxiety and Stress Scale improved in the placebo group and worsened slightly in the theanine group, a difference that was significant at day 28 (−9.20 against +1.07, p = 0.017), with the stress subscale following the same pattern (−4.67 against +0.53, p = 0.012). On the Profile of Mood States, the theanine group showed no significant change on anything; the placebo group improved significantly on total mood disturbance, fatigue, tension–anxiety and depression.

The authors handle this honestly in the discussion, attributing the pattern to a placebo effect they explicitly acknowledge, and calling for longer trials to see whether it wanes. Nine adverse events were reported, eight of them in the theanine arm. The study was funded by Ethical Naturals, Inc., which makes the ingredient, and one author is its employee.

Read as a whole: a well-run 28-day trial in which the pre-specified marker did not move, both arms improved on the questionnaire that did, and the second questionnaire favoured placebo. That is a negative trial, and the conclusion — that supplementation significantly decreased perceived stress — describes a within-group change that placebo matched.

Readers of our article on real-world dementia data will recognise the shape. Null on the outcome the study was designed around, positive on something else, and the second sentence is the one that travels.

2.The Trial That Worked, and What Was In It

Lionel Noah, Gisèle Pickering and colleagues, working between Sanofi and the University Hospital of Clermont-Ferrand, randomised 106 chronically stressed but otherwise healthy adults to a supplement or placebo for 28 days, with follow-up at day 56. One hundred completed.

This one met its primary endpoint. Stress scores on the 42-item Depression Anxiety Stress Scale fell further on the supplement than on placebo — effect size −0.29, confidence interval −0.57 to −0.01, p = 0.04 — with the same three-point difference at day 14 (p = 0.006) and still present at day 56, a month after supplementation stopped.

Then read the intervention. Each tablet contained 150 mg of magnesium, 0.7 mg of vitamin B6, 0.1 mg of vitamin B9, 1.25 micrograms of vitamin B12, 222 mg of rhodiola extract, and 125 mg of green tea extract — of which 50 mg was L-theanine.

Fifty milligrams is the lowest theanine dose anywhere in this series, below the level at which the cognition meta-analysis in Part One found anything, and roughly two cups of tea. Magnesium has an authorised European health claim for the reduction of tiredness and fatigue and for normal psychological function; B6 has one for normal psychological function. Rhodiola has its own trial literature on stress.

The authors are careful about this. They attribute the effect to synergism or an additive effect of the ingredients and do not claim it for any single one. Their own discussion notes that both groups improved substantially — 33 per cent and 25 per cent — pointing toward a placebo effect, with the between-group difference reaching the three-point threshold their clinicians had defined as clinically relevant.

One design detail limits the reading further: the trial was single-blinded, because the active and placebo tablets lacked structural similarity. Investigators were blinded; participants only partly.

So the strongest positive trial in this part tested five active substances at once, at least two of which have independent evidence and authorised claims, with theanine present at a dose no isolated trial has found active, in participants who could potentially tell which tablet they had. It is a good trial of that product. It is not evidence about L-theanine.

3.Acute Stress, With Nobody Paying

Matthew McAllister, Hunter Martaindale and colleagues at Texas State University did something the rest of this literature mostly does not: they applied a real stressor and measured the physiology, with no commercial funding and no conflict of interest to declare.

Eighty participants were randomised to 200 mg of L-theanine, 2,000 mg of L-tyrosine, or placebo, 40 minutes before a mental stress challenge combining a Stroop task, mental arithmetic and a virtual-reality active shooter drill in which the participant was the responding officer. The team cite prior work finding that the physiological response to this simulation matches that of a live scenario with actors.

The challenge worked. Salivary alpha-amylase, secretory immunoglobulin A, heart rate and state anxiety all rose significantly.

The supplements did not change any of it. There was no treatment-by-time interaction on any of the four stress measures, and the authors state the conclusion without qualification: ingestion of L-theanine and L-tyrosine did not impact markers of stress.

On the cognitive side, the tyrosine group made significantly fewer missed Stroop responses than placebo. Theanine did not differ from placebo on anything. The authors also flag, to their credit, that no baseline cognitive assessment was taken, so between-group differences cannot confidently be attributed to the supplements at all.

This is the trial in the set that most closely resembles the situation the compound is marketed for — an acute demand, in a healthy adult, an hour after taking it — and it found nothing for theanine.

4.Children, Tics and an Open Label

The fourth trial belongs to a different world and is included because it is the only recent work testing this compound in a paediatric clinical population.

Renata Rizzo, Adriana Prato and colleagues at Catania University enrolled 34 children aged 4 to 17 with Tourette syndrome or chronic tic disorder plus anxiety symptoms, and randomised them to two months of L-theanine 200 mg with vitamin B6 2.8 mg, or to eight sessions of psychoeducation.

Tic severity fell further in the supplement group: a mean reduction of 8.85 points on the Yale Global Tic Severity Scale against 3.4, p = 0.046, with 70.6 per cent achieving at least a 30 per cent reduction against 17.6 per cent.

Anxiety, which was the stated rationale for using theanine at all, did not differ between groups. The mean scores at two months were 30.9 and 31.5, p = 0.85.

The design constrains what any of this can mean. It was open-label, so families knew whether their child was receiving drops. The comparator was psychoeducation rather than placebo — the authors list the absence of a control group among their limitations. Tic severity fluctuates markedly on its own. And the intervention again contained two substances.

The authors' own framing is appropriately narrow: preliminary rather than conclusive, and further placebo-controlled trials are needed. Tourette syndrome is a medical diagnosis and nothing here should be read as guidance for treating it; that belongs with a paediatric neurologist.

5.What the Four Trials Show Together

Arrange them by how much they isolate the compound and how well they control expectation, and the results line up almost monotonically.

Double-blind, standalone, objective marker, acute stressor: nothing. Double-blind, standalone, 28 days, objective primary outcome: nothing on the primary, and the questionnaires split. Single-blind, five ingredients: a small significant effect on a self-reported scale. Open-label, two ingredients, clinician-rated: an effect on tics, nothing on the anxiety it was meant to address.

Not one of the four found L-theanine alone moving an objective marker of stress. The two that measured biology directly — salivary cortisol, salivary alpha-amylase, immunoglobulin A, heart rate — found nothing.

Set against the earlier systematic review that proposed 200 to 400 mg as the effective range for stress and anxiety, and which every subsequent review cites as settled background, the recent trials do not support it as confidently as the citation chain implies. The claim has become load-bearing for the sleep and cognition literatures without being re-tested successfully in the period since.

This is the combination problem in its clearest form, and it is not confined to theanine. When a product contains five plausible ingredients and improves an outcome, the trial has tested the product. Attributing the result to whichever ingredient the marketing leads with is an inference nobody has earned, and it is the single most common error in this category.

6.The Positive Trial We Missed

This section was added after publication. A fifth randomised trial belongs in this part, we did not cover it, and it points the other way from the four above.

Shinsuke Hidese, Hiroshi Kunugi and colleagues at Japan's National Center of Neurology and Psychiatry ran a randomised, double-blind, placebo-controlled crossover trial of 200 mg of L-theanine daily for four weeks in thirty healthy adults — nine men and twenty-one women, mean age 48.

Stress-related symptom scores fell after L-theanine relative to placebo, measured with the Self-rating Depression Scale, the State-Trait Anxiety Inventory and the Pittsburgh Sleep Quality Index. On the sleep index, the sleep latency and daytime dysfunction subscales improved specifically.

Cognition moved as well: verbal fluency, and letter fluency in particular, along with executive function scores.

That is a four-week crossover with a real placebo arm on a single, well-specified compound at a dose we could recognise. On design, it is stronger than two of the four trials above.

One thing has to be said alongside it. Two of the seven authors are employed by the Nutrition Division of Taiyo Kagaku, the company that manufactures the branded L-theanine used in this literature. That does not make the result wrong and we are not treating it as though it did. It is the kind of fact a reader is entitled to have when weighing a positive finding, and we would want it disclosed about a trial of ours.

The outcomes are also self-reported. Depression, anxiety and sleep-quality scales measure what a participant says about their state, which is the same limitation we raised against the trials in Sections 2 and 3 — and it is why the objective-marker studies covered above still matter.

So the summary in Section 5 needs one amendment rather than a rewrite. On objective physiological markers of stress, the evidence for L-theanine alone remains unpersuasive. On self-reported stress symptoms over four weeks, there is a reasonably designed positive trial, with a manufacturer at the table, that we should have put in front of readers the first time.

We found it by re-checking sources we had recorded as unavailable without testing each one. It was available throughout.

Editorial Comment

L-theanine has no authorised health claim in the European Union. We do not claim that it reduces stress, and after reading these four trials we would regard such a claim as unsupported by the recent evidence regardless of what the regulations permitted.

One disclosure is owed here in particular. The Sanofi formulation described in section 2 — magnesium, B vitamins, an adaptogen extract and L-theanine — resembles our Clarity closely enough that quoting its positive result would be commercially convenient for us. We are not going to, and we would ask readers to notice when others do. Our formulation pages state the magnesium and B-vitamin claims in the exact wording the European register authorises, and say nothing about theanine, because nothing is authorised to be said.

Our own use of the compound rests on the attention evidence in Part One and the caffeine pairing, not on stress. Nothing in this part changes that, and nothing in it would let us widen the story.

If stress has stopped being an occasional state and become a constant one, that is a matter for a doctor rather than a supplement aisle. The trials above enrolled people scoring in the moderate range on validated scales and found, at best, small differences against placebo — and in the best-controlled of them, the placebo did just as well.

Part Four leaves the brain and looks at what else this molecule has been shown to do, in immunity and in the gut — where the mechanistic work is far more interesting than the human evidence and the distance between the two is the whole story.