Adaptogens & the stress response

The term was coined in 1947 by Nikolai Lazarev, a Soviet toxicologist, and developed through the 1950s and 1960s by Israel Brekhman and Igor Dardymov in Vladivostok ^[4]. Their 1969 review in Annual Review of Pharmacology is the single most cited source in the field and the origin of the three-part definition still quoted today ^[1].

The programme's aims were applied rather than theoretical, and the plants it worked on were the ones growing in Soviet territory: eleuthero, Eleutherococcus senticosus, plus Rhodiola rosea, Schisandra chinensis and Rhaponticum carthamoides ^[4]. Eleuthero is still widely sold as Siberian ginseng, which is a marketing coinage rather than a botanical fact, since it is not a Panax species at all.

That is the inheritance, and it needs stating plainly. A large share of the primary literature behind this category consists of studies published inside a system with no external peer review, in a language most later citers could not read, using designs that predate the modern trial. Reviews sympathetic to the field concede the shape of the problem directly: the plants have a rich history of use, and clinical trials in humans remain limited ^[4].

None of that makes the plants inert. Eleuthero, rhodiola and schisandra were investigated by competent scientists testing real hypotheses, and several of them have since acquired modern evidence. The failure mode to guard against is different and more specific: a claim can be repeated for sixty years without ever having been tested. When a supplement page says a plant has been used as an adaptogen since the 1960s, that sentence is true and carries no information about whether it works.

A related caution about the modern literature. The most influential contemporary defence of the category comes from Alexander Panossian, whose earlier papers carry a Swedish Herbal Institute affiliation and whose later ones a supplement-company affiliation ^[2][5]. That is common in botanical pharmacology and it is not disqualifying, but the theory and the products come from overlapping places, and a reader should know that.

The original definition set three conditions. An adaptogen should raise nonspecific resistance to a broad range of physical, chemical and biological stressors; it should have a normalising effect, correcting in whichever direction the disturbance runs; and it should be innocuous, disturbing normal physiological function as little as possible ^[1].

Read as science rather than as a slogan, each condition is hard to falsify. Nonspecific resistance has no assay. There is no single measurement whose failure would show a substance does not have it, which means the claim can survive any particular negative result. Normalising predicts both directions at once, so a study finding an increase and a study finding a decrease are both compatible with the hypothesis; a theory that accommodates opposite outcomes is not making a risky prediction. Innocuous is a long-term safety claim, and almost none of these plants has the surveillance data that would support it. The ashwagandha liver-injury series discussed below is the direct counterexample.

The definition has also moved. The modern formulation broadens adaptogens to substances that increase an organism's ability to adapt and survive, and maps their action onto molecular chaperones such as HSP70, the stress-activated kinase JNK1, the FOXO transcription factor DAF-16, nitric oxide and cortisol ^[2]. That is a real list of mediators with real experiments behind it. It is also broad enough to be compatible with a great many observations and to forbid very few.

The most candid statement of the problem comes from inside the field. Panossian has argued explicitly that the classical reductionist model of pharmacology is insufficiently complex to describe adaptogenic activity, and that network pharmacology is the only suitable frame ^[5]. Take that argument seriously and it has a cost worth naming: a mechanism that cannot be reduced to a target is a mechanism no binding assay can refute. It moves the category further from the kind of claim that a single decisive experiment could settle, which is precisely the kind of claim the rest of this site tries to hold compounds to.

So the working position taken here is this. Adaptogen is a useful shorthand for a family of slow-acting plant extracts with plausible effects on stress physiology, and it is not a mechanism. Judge each plant on its own trials.

The framework these compounds were built on is Hans Selye's general adaptation syndrome, described in its mature form in 1950: a stereotyped three-phase response to any sufficiently intense demand, running alarm, then resistance, then exhaustion, mediated through the pituitary and adrenal cortex ^[3]. Brekhman's programme was an attempt to find substances that extended the resistance phase.

The machinery is the HPA axis. Under a stressor the hypothalamus releases corticotropin-releasing hormone, the pituitary releases adrenocorticotropic hormone, and the adrenal cortex releases cortisol, which then feeds back to shut the loop down. Short bursts are healthy and necessary. Chronic elevation is corrosive, wearing on mood, sleep, memory and metabolism. See the HPA axis & cortisol for the full picture.

The evidence that adaptogens touch this axis in humans is real but narrow, and it comes almost entirely from ashwagandha. In sixty-four adults with chronic stress, sixty days of a full-spectrum root extract reduced serum cortisol substantially against placebo ^[6]. In a separate sixty-person trial, a standardised extract produced greater reductions in morning cortisol and in DHEA-S than placebo, and the authors proposed HPA moderation as the mechanism ^[7]. Rhodiola moved a different marker: in people with stress-related fatigue, the cortisol response to awakening differed significantly between the extract and control groups after 28 days ^[12].

Here is where to be careful. A cortisol change is a biomarker, not an outcome. Lower is not automatically better; cortisol is a regulator with a necessary circadian shape, and the same trial that found morning cortisol falling also found DHEA-S falling, which is not obviously a desirable direction ^[7]. The reason those trials count at all is that the clinical scales moved in the same direction as the hormones, not that the hormones moved.

The broader molecular story, HSP70 and JNK1 and DAF-16 and nitric oxide ^[2], is best read as a set of candidate mechanisms rather than an established pathway. It is drawn largely from animal work and isolated cells, and no part of it has been shown to be the step through which a human benefit is produced.

The classical list is short: Panax ginseng, eleuthero, Rhodiola rosea, schisandra, and in the Ayurvedic rather than Soviet tradition, ashwagandha. Everything else now sold under the label was added later and by commerce.

Ashwagandha is the one with a modern evidence base, and it is genuinely the best-supported botanical in this whole area for stress and sleep. Its active constituents are the withanolides, a family of steroidal lactones, and the standardised extracts used in trials (KSM-66, Sensoril, Shoden) differ from each other in withanolide content and are not interchangeable.

Rhodiola is the stimulating end of the family. Its marker compounds are rosavins and salidroside, it is the one most associated with anti-fatigue rather than anti-anxiety effects, and its trial literature is larger than ashwagandha's in count and much weaker in quality, which the next section quantifies.

Panax ginseng is the most studied and the least convincing for cognition. The Cochrane review found no convincing evidence of a cognitive-enhancing effect in healthy participants and no high-quality evidence in dementia ^[15]. American ginseng is a different species with a different ginsenoside profile and a smaller literature again.

Eleuthero and schisandra are the ones running on inheritance. Both sit in every classical roster, both were central to the Soviet programme, and the modern controlled literature on either is thin ^[4]. That is the clearest example of the gap between how often a plant is named and how often it has been tested.

Bacopa monnieri is regularly filed here and does not belong, because it is an Ayurvedic nootropic rather than an anti-stress agent by the 1969 criteria. It is included in the table because it has a better meta-analysis than most things that do belong ^[16]. Cordyceps, reishi, astragalus, maca and tongkat ali are modern additions to the category by marketing; each may have its own merits and none of them was an adaptogen under the original definition.

Adaptogens sit in an awkward middle ground: a real and growing body of randomised human trials, wrapped in a great deal of marketing and a genuinely fuzzy theory. The way through is to read the trials individually and notice what is small about them.

The ashwagandha trials are positive and they are small. Sixty-four people, sixty people, sixty people, all single-centre, all using one proprietary extract. Some clear significance comfortably ^[6][8] and one clears it narrowly on its primary anxiety scale while missing on its stress scale altogether ^[7]. Consistent, modest, and never large.

The rhodiola literature is the cautionary one. Its best individual trial found both arms improving substantially, with the extract separating only on a burnout scale and three attention indices ^[12]. Two independent systematic reviews then examined the whole set. One concluded rhodiola may have beneficial effects and named the decisive problem, a lack of independent replication of any individual study ^[13]. The other was harsher: every included trial exhibited either a high risk of bias or reporting flaws that prevented assessing its true validity ^[14].

The ginseng answer is simply negative for cognition. Only five of nine qualifying trials had extractable data, pooling was impossible because of heterogeneity, and the conclusion was a lack of convincing evidence ^[15].

Three patterns are worth carrying away. First, positive results cluster in stressed or symptomatic populations, which is the same pattern seen across the rest of this site: a compound that corrects a deficit looks better than a compound asked to improve a healthy baseline. Second, effect sizes are modest and slow, which means an honest trial of one takes weeks and a two-day impression is worthless. Third, the evidence attaches to extracts, not to plants. Every positive result above used a named, standardised preparation with a specified withanolide or rosavin content, and a generic capsule labelled with the plant's name is not the substance that was tested.

Read that way, treated as gentle long-game stress support rather than a quick fix, adaptogens remain one of the more sensible places in this field to experiment. The claim being made for them is small, and the evidence is a reasonable match for a small claim.

Ashwagandha can injure the liver. Five cases of hepatotoxicity attributed to ashwagandha-containing supplements were reported from Iceland and the US Drug-Induced Liver Injury Network. All five developed jaundice after a latency of two to twelve weeks, with cholestatic or mixed injury, and with prolonged pruritus and hyperbilirubinaemia lasting five to twenty weeks. One biopsy showed acute cholestatic hepatitis. Nobody developed hepatic failure and liver tests normalised in one to five months in the four who were followed. Chemical analysis confirmed ashwagandha in the products and identified no other toxic compound ^[10].

Five cases is not an incidence rate, and this is a very widely used supplement, so the absolute risk is presumably low. The point is narrower and it matters: a plant marketed under a definition whose third clause is innocuous has a documented, characterised pattern of drug-induced liver injury. That clause was never a finding; it was an assumption written into the definition in 1969, and it can be wrong.

It also moves thyroid hormones. In fifty patients with subclinical hypothyroidism, eight weeks of 600 mg daily of ashwagandha root extract significantly improved serum TSH, T3 and T4 against placebo ^[11]. In that population that is presented as a benefit. In anyone who is euthyroid, taking levothyroxine, or hyperthyroid, the same effect is an unmanaged endocrine change from a supplement, and it is a good reason to tell a doctor it is being taken.

Sedation stacks. Ashwagandha measurably shortens sleep onset latency and raises sleep efficiency ^[8], which is the desired effect and also means the combination with alcohol, antihistamines or prescription sedatives is an additive-effects question rather than a neutral one. See nootropic stacking.

Product quality is the unmanaged variable. The most careful modern review of ashwagandha states the problem for the whole category: the significant variability between the extracts examined prevents any consensus on the optimum preparation or dose, and potential herb-drug interactions remain the important open question for anyone combining it with prescription medicines ^[9]. That is a direct instruction to buy the extract a trial used, at the dose the trial used, and to distrust everything else on the label.

And the one contraindication with no data behind it in either direction is pregnancy, where ashwagandha is traditionally avoided and where no controlled safety evidence exists. Absence of evidence is the whole answer there. None of this is medical advice.