Rhodiola rosea: salidroside, monoamines and mechanisms of fatigue

Rhodiola rosea is known as “golden root”, but for a professional the name refers to several different objects: a herbal substance, a multicomponent extract and individual compounds, particularly salidroside. These are not interchangeable. An experiment with a purified compound helps identify a possible biological target, whereas a clinical trial describes a particular preparation, dose, population and treatment duration. Between these levels lie absorption, metabolism, tissue access and exposure: without these conditions, even a persuasive cellular effect does not become a treatment.

Interest in rhodiola centres on fatigue, stress and physical performance. These conditions overlap in neuroendocrine regulation and energy metabolism, but they have different causes and call for different decisions. Fatigue after sleep loss, depressive slowing, low energy availability in an athlete and post-exertional worsening in myalgic encephalomyelitis/chronic fatigue syndrome do not constitute a single indication for an “adaptogen”. A detailed molecular account is therefore useful primarily for framing the right question, choosing an outcome to follow and setting limits on its interpretation.

Adaptogenesis: a historical concept and a testable hypothesis

In Soviet pharmacology, the term “adaptogen” developed around the idea of increasing resistance to different stressors. The classical definition emphasised nonspecific action, normalisation of disturbed functions and low toxicity under the conditions of use. These features constituted a research programme, rather than trial results for every plant. Claims of “harmlessness” deserve particular care: observation duration, extract composition, concurrent medicines and individual susceptibility all change the risk profile.

The molecular version of this concept proposes a role for cellular defence responses: chaperones, antioxidant systems, metabolic sensors and stress signalling pathways. This is sometimes described through hormesis, a response to a moderate challenge that increases resistance to a subsequent one. Hormesis, however, requires a demonstrated dependence on dose and experimental conditions. The appearance of phosphorylated AMPK or an increase in Hsp72 does not by itself show that an ordinary rhodiola capsule produces a beneficial hormetic exposure in a human brain or muscle.

Comparisons with caffeine and amphetamines also have limits. No common pattern of effects has been established across all rhodiola products that would guarantee an absence of tolerance, withdrawal, activation or sleep disruption. Short-term tolerability in small studies appears relatively favourable, but that is a narrower statement than “does not exhaust reserves and causes no adverse effects”. The European regulatory monograph considers certain preparations for temporary relief of stress symptoms within a traditional-use framework; this status is not equivalent to convincingly demonstrated efficacy for a particular disease.

What an extract contains: salidroside, rosavins and botanical authenticity

Rhodiola rosea L. belongs to the Crassulaceae family. Roots and rhizomes are used in preparations, and their chemical profile depends on plant origin, processing and extraction method. A species name and extract mass without composition data leave substantial uncertainty. Different solvents extract different combinations of phenolic compounds, glycosides and other secondary metabolites.

Salidroside and tyrosol

Salidroside, also called rhodioloside, is a glycoside of tyrosol. Its molecular formula is C₁₄H₂₀O₇ and its molar mass is approximately 300.30 g/mol. The β-D-glucopyranose residue is linked through the oxygen of tyrosol’s hydroxyethyl side chain; the phenolic hydroxyl in the para position remains free. This is essential when drawing the molecule correctly: glucose is not attached directly to the phenolic oxygen.

The glycosidic bond can undergo hydrolysis to yield the aglycone and glucose. In abbreviated form, the reaction is salidroside + H₂O → tyrosol + D-glucose; β-glucosidase can serve as a catalyst. This notation describes a chemical transformation and does not specify the fraction of a dose hydrolysed in a particular person. It cannot establish that all salidroside is converted to tyrosol in the intestine or that free tyrosol reaches a clinically relevant concentration in the brain.

Adding a sugar increases molecular polarity, but does not by itself establish “high metabolic stability”. Absorption, conjugation and distribution require separate pharmacokinetic studies. In particular, a compound’s ability to affect cultured nerve cells is not a measurement of blood–brain barrier passage after oral administration. For a professional, this is an important boundary between structural biochemistry and explaining a preparation’s action in a patient.

Rosavins and rosiridin: different chemical groups

The collective term rosavins usually refers to rosavin, rosin and rosarin, glycosides of cinnamyl alcohol. Rosin has a glucose residue attached to the aglycone; rosavin and rosarin contain an additional arabinose, but differ in its ring form. Rosavin contains an α-L-arabinopyranosyl residue linked to position 6 of glucose, whereas rosarin contains an arabinofuranosyl residue. Rosavin has the formula C₂₀H₂₈O₁₀; the trans configuration of the double bond in its aglycone should be preserved in a structural illustration.

Rosiridin is a separate monoterpene glycoside, not one of the rosavins. This distinction is particularly important when reading monoamine oxidase studies: a strong effect of rosiridin on an enzyme in vitro cannot be attributed to the entire phenylpropanoid group. Russian accounts sometimes use a name resembling “rosaridin” and confuse these objects; the compound discussed here is rosiridin.

Why standardisation is not evidence of efficacy

Salidroside occurs in different members of the genus, so its presence does not establish the authenticity of R. rosea. The rosavin profile is more useful for identification, but should still be considered alongside botanical expertise and a chromatographic fingerprint. An absolute claim that no other species ever contains rosavins is unnecessary: analytical identification relies on a combination of features, not a universal exclusion applying to the whole genus.

Analysis of commercial products has revealed discrepancies with declared composition and cases of botanical substitution. Quality control therefore needs to confirm the species, plant part, extraction process, several marker concentrations and contaminants. A percentage on a label does not replace documented batch analysis. Requirements from one pharmacopoeia should not automatically be extended to every preparation and market.

The ratio “3% rosavins and 1% salidroside” is widely used in descriptions of commercial standardisation, but is not a universal pharmacopoeial rule and does not guarantee equivalence to SHR-5 or WS 1375. Two extracts with identical markers may differ in other compounds. Moreover, demonstrating synergy requires a comparison of a mixture with its components at comparable exposures; the presence of several potentially active compounds does not establish clinical synergy.

Monoamines: where MAO is located and what inhibition means

MAO-A and MAO-B are flavin-containing enzymes of the outer mitochondrial membrane. They catalyse oxidative deamination and participate in intracellular amine metabolism. MAO-A is mainly discussed in relation to serotonin and noradrenaline, and MAO-B in relation to phenylethylamine; dopamine can be a substrate for both isoforms. Their contribution depends on tissue, species and physiological state, so a fixed fraction such as “70–80%” should not be presented as a general constant of dopamine metabolism in the human brain.

The general reaction is R–CH₂–NH₂ + O₂ + H₂O → R–CHO + NH₃ + H₂O₂. Here R represents the rest of the amine molecule and R–CHO the corresponding aldehyde. MAO uses FAD as a cofactor; the resulting aldehyde is subsequently metabolised by other enzymes. The reaction also explains hydrogen peroxide formation, but does not establish that blocking MAO automatically provides neuroprotection: its physiological significance depends on the entire metabolic pathway and transmitter concentrations.

Van Diermen’s study tested methanolic and aqueous extracts, rather than a clinically standardised hydroethanolic preparation. At 100 µg/mL, MAO-A inhibition was 92.5% and 84.3%, respectively, and MAO-B inhibition was 81.8% and 88.9%. Rosiridin produced more than 80% MAO-B inhibition at 10 µM, with a reported pIC₅₀ of 5.38 ± 0.05. These are enzyme-assay results; they do not establish an antidepressant effect in humans, despite the authors’ broader interpretation.

The expression pIC₅₀ = −log₁₀(IC₅₀/[1 mol/L]) allows the concentration scale to be recovered: the central estimate corresponds to approximately 4.2 µM. IC₅₀ is the concentration that halves activity under the conditions of a particular assay. It is not a binding constant Kᵢ, a brain concentration or a capsule dose. For a mixture, a value in µg/mL cannot be directly compared with the µM concentration of a purified component.

Within a synapse, transmitter release, receptor interaction, reuptake and intracellular degradation are different processes. MAO should not be depicted as an enzyme of the synaptic cleft. If its intracellular activity does change, monoamine availability may be affected, but moving from an enzyme assay to the duration of a synaptic signal requires confirmation in a living system. Rhodiola has not been shown to improve attention or fatigue in clinical trials specifically through this pathway.

Neither competitive inhibition nor reversibility across all extract components can be considered established merely from the percentage reduction in activity. For the same reason, an absence of tyramine interactions cannot be promised on the basis of assumed reversibility. Rhodiola is not a clinically characterised substitute for pharmacological MAO inhibitors; comparisons with phenelzine, tranylcypromine or moclobemide do not justify inferring the same effect or an opposite guarantee of safety.

NPY and Hsp72: a cellular response to stress

Neuropeptide Y, or NPY, contains 36 amino acids and participates in nervous, autonomic and metabolic regulation. It should not be described as a single “anti-stress brake”: its effects depend on the receptor, brain region and compartment. In particular, the properties of Y₁ and Y₂ do not justify treating every increase in NPY as anxiolytic. A result in cultured cells does not measure hippocampal neurogenesis or a patient’s emotional resilience.

Panossian’s study used T98G, a cell line derived from a human glioblastoma. The interventions included salidroside and ADAPT-232, a combination of rhodiola, eleutherococcus and schisandra extracts. Changes in NPY and Hsp72 expression and release were observed, and individual elements were suppressed to investigate how the responses were related. This design supports a cellular hypothesis, but is not an experiment on a healthy human brain and does not allow all effects of the combination to be attributed to rhodiola alone.

A subsequent study of molecular markers extended this experimental programme. It is useful for designing further research, but does not make NPY or Hsp72 clinical laboratory tests for selecting adaptogens. Likewise, gene changes in a microarray analysis identify affected pathways; they do not confirm treatment of every disease named in a software annotation of those pathways.

Hsp72 belongs to the inducible members of the Hsp70 family. Chaperones bind regions of misfolded proteins and help restore their structure, using a cycle of ATP binding and hydrolysis. The simplified energy expression ATP + H₂O → ADP + Pᵢ denotes energy expenditure within that cycle; Pᵢ is inorganic phosphate. An increase in protein abundance, however, does not demonstrate that all of its functions have increased in a particular tissue.

In experimental systems, Hsp70 can affect JNK and mitochondrial apoptosis, including interactions with Apaf-1 and recruitment of procaspase-9 to the apoptosome. These findings describe chaperone biology. Separate evidence is needed to claim that rhodiola prevents motor neuron or cardiomyocyte injury through this mechanism in humans.

A further distinction concerns chaperone-mediated autophagy. Its central chaperone is constitutive Hsc70, which recognises appropriate cytosolic proteins and delivers them to LAMP-2A on the lysosomal membrane. Inducible Hsp72 and Hsc70 belong to the same family but are not interchangeable. Thus, the statement “rhodiola increases Hsp72, therefore it activates autophagy through LAMP-2A” contains an untested step.

β-Endorphin: stress-related changes and the limits of the opioid hypothesis

β-Endorphin is formed during processing of proopiomelanocortin, or POMC. This system links production of ACTH and opioid peptides, although the set of products depends on cell type and processing enzymes. Different peptide sources also matter: pituitary release into the blood cannot automatically be equated with central opioid transmission. The plasma concentration of immunoreactive β-endorphin is not a quantitative measure of analgesia in the brain.

In an early experiment in albino rats, four hours of stress increased β-endorphin and altered hormonal measures. Prior adaptation through brief stress exposures or administration of rhodiola extract was associated with a moderate baseline increase in immunoreactive β-endorphin and a reduction of the subsequent stress-related rise. This was not a study of chronic exhaustion in humans. It did not establish an obligatory sequence of “pathological peak, depletion of stores, rescuing plateau”.

Antiarrhythmic experiments also require precise reading. In a study of adrenaline-induced arrhythmia in rats, blocking μ- and δ-receptors under the conditions used did not abolish the protective effect; a change in response at a higher naloxone dose led the authors to discuss κ-receptors. The original explanation based on obligatory stimulation of μ- and δ-receptors therefore does not match that study. Other opioid mechanisms may depend on the model, but these experiments do not establish clinical prevention of arrhythmias.

For sports practice, a particularly undesirable inference is that changing endorphins allows a person to safely overcome a “critical fatigue barrier”. Pain, an abrupt decline in performance and impaired coordination can signal injury, heat illness or another dangerous condition. Neither an animal hormonal experiment nor proposed opioid modulation justifies continuing exercise through such symptoms.

AMPK and muscle energetics: what has actually been shown

During muscular work, ATP demand rises, and the contributions of phosphocreatine, glycolysis and oxidative phosphorylation vary with exercise intensity and duration. Fatigue cannot be reduced to a single metabolite: central regulation, electrolyte changes, substrate availability and recovery of the contractile apparatus all contribute. Lactate is a transported and oxidisable metabolite, rather than exclusively a “toxin of exhaustion”; its rise should not be equated with the sole cause of H⁺ accumulation.

AMPK is a heterotrimeric complex with α-, β- and γ-subunits that helps detect cellular energy status. Its regulation involves adenine nucleotide binding and phosphorylation of the activation loop, usually designated Thr172. The upstream kinase LKB1 is important in muscle, but belonging to the same pathway does not establish a direct action of salidroside on LKB1. The AMP/ATP ratio is used to assess an energy shift, rather than a promise to “add energy” from outside.

In differentiated rat L6 cells, salidroside increased glucose uptake and phosphorylation of AMPK and ACC. Compound C reduced stimulated uptake, whereas PI3K blockade under the conditions studied did not produce the same effect; enhancement of the insulin response was also described. These results support AMPK involvement, although a pharmacological intervention does not itself establish absolute specificity. Nor do they measure a training outcome in humans.

Another study used diabetic db/db mice, primary mouse hepatocytes and isolated mitochondria. Salidroside affected AMPK/PI3K/Akt/GSK3β; in the mitochondrial system it limited complex I activity, changed coupling and membrane potential, and transiently increased AMP/ATP. This is an insulin-resistance model, not evidence of improved energetics in a healthy athlete. Later work using C2C12 cells and mice examined AMPK/SIRT1 and mitochondrial quality control, also at the preclinical level.

The ACC pathway helps explain the direction of the proposed effect. Acetyl-CoA carboxylase catalyses ATP-dependent malonyl-CoA formation: acetyl-CoA + HCO₃⁻ + ATP → malonyl-CoA + ADP + Pᵢ, in abbreviated biochemical notation. Phosphorylation of ACC by AMPK can reduce enzyme activity. Malonyl-CoA inhibits CPT1, so lowering it can relieve this restriction on entry of long-chain fatty acids into the carnitine transport pathway.

CPT1 is located on the outer mitochondrial membrane and transfers an acyl group from acyl-CoA to carnitine: acyl-CoA + carnitine ⇌ acylcarnitine + CoA-SH. The inner-membrane translocase and CPT2 are then required to regenerate acyl-CoA in the matrix. A single arrow labelled “CPT1 → β-oxidation” should therefore represent a multistep process, rather than free diffusion of a fatty acid into the mitochondrion. ACC is not a “Cys enzyme”; biotin participates in its carboxylase reaction.

GLUT4 transports glucose across the plasma membrane after the relevant vesicles relocate. AMPK-dependent regulation can intersect with insulin mechanisms, but it does not remove perfusion constraints or provide an unlimited substrate supply. It is important to distinguish a change in transporter expression, transporter translocation and directly measured glucose flux: they are different experimental outcomes.

PGC-1α, SIRT1 and mitochondrial biogenesis

PGC-1α is a transcriptional coactivator linking metabolic signals to expression of energy-metabolism genes. AMPK can phosphorylate it, and SIRT1 can participate in its deacetylation. The system includes nuclear respiratory factors and TFAM, which is needed for mitochondrial genome regulation. Nuclear respiratory factor NRF-2/GABP should not be confused with Nrf2, the antioxidant-response transcription factor.

An experiment using primary muscle cells and genetic models demonstrated a functional link between AMPK and PGC-1α, including direct phosphorylation. This explains why researchers investigate such pathways for salidroside. However, increases in mitochondrial mass, cristae number, complex IV activity, MCT1 expression and VO₂peak are different endpoints. They cannot be joined into a pathway proven for rhodiola without studies of each transition. An AMPK increase in a cell does not replace measurement of maximal oxygen uptake or establish remodelling of muscle fibres in humans.

The HPA axis, cortisol and recovery after stress

The hypothalamic–pituitary–adrenal axis links hypothalamic CRH release, pituitary ACTH secretion and cortisol synthesis in the adrenal cortex. Cortisol participates in feedback and follows a daily rhythm. Stress, shift work, sleep quality and awakening time can alter the observed profile, but “allostatic load” is not an automatic diagnosis from a single saliva sample.

The term “adrenal fatigue” does not describe an established endocrine disorder and should not replace assessment for adrenal insufficiency. Chronic stress also does not invariably produce low morning cortisol: the direction of change depends on the condition and measurement method. Depicting a single “pathological flat curve” therefore creates false diagnostic certainty.

A randomised trial of SHR-5 in 60 participants with stress-related fatigue used 576 mg of extract daily for 28 days. Differences were reported in some fatigue and attention outcomes and the salivary cortisol response after awakening. The authors described a reduction in the cortisol response, rather than restoration of a low morning peak. This study cannot establish normalisation of the complete daily rhythm, evening clearance or glucocorticoid receptor sensitivity.

Research assessment of the cortisol awakening response, or CAR, requires sampling to be linked to actual awakening and adherence to the protocol. Samples immediately after awakening and a short time later describe this early response; midday and evening samples belong to the daily profile. A self-arranged four-point panel is not obligatory monitoring for rhodiola and does not show that a patient needs the preparation or that it has “restored the axis”.

Clinical studies: the same words, different outcomes

Human evidence cannot be condensed into a single formula such as “effective for chronic fatigue”. Trials differ in populations, extracts, scales, duration and controls. The distinction between a within-group change and an advantage over placebo is particularly important. Even a low p value in a before-and-after comparison does not establish a preparation-specific effect.

Work-related fatigue and examination stress

The study of physicians on night shifts assessed memory, calculation, concentration and perception tests. Its three two-week periods included a washout between two treatment phases, rather than three courses of active preparation. Improved tests in the first period are of interest, but do not demonstrate fewer medical errors, correction of circadian misalignment or the safety of chronic sleep restriction. An extract does not replace sleep or appropriate shift organisation.

The student pilot reported improvements in some measures and wellbeing. Its 100 mg/day dose was a research choice; calling it “subtherapeutic” is inappropriate because no established therapeutic range exists for these tasks. The study cannot show that increasing the dose to 200–300 mg improves academic productivity. That would require direct dose comparisons and prespecified educational outcomes.

The open-label WS 1375 study and the mild-anxiety trial add information about tolerability and subjective symptoms. However, participant and researcher expectations, changes in routine and natural symptom fluctuation may account for part of the observed effect. Professionals should retain these findings as hypotheses for controlled investigation, rather than confirmation of treatment for an anxiety disorder or burnout.

Endurance, lactate and muscle injury

In De Bock’s study, time to exhaustion after a single dose increased from 16.8 to 17.2 minutes, approximately 24 seconds in a particular exercise test. The subsequent four-week phase did not produce the same convincing improvement. This matters more than the advertising formula “increases endurance by 3%”: the result belongs to a protocol and does not define the expected benefit in every sport. Nor does it establish a universal increase in VO₂peak by a fixed amount.

Noreen’s trial assessed a six-mile time trial, rather than exactly 10 km. Changes in completion time and perceived exertion were observed, but the sample was small and the intervention acute. In Parisi’s study, some post-exercise biomarkers fell while test duration, VO₂max and subjective outcomes did not change significantly. This is a particularly clear example of the distinction between a surrogate measure and a sporting outcome.

The review of sports studies included in the original bibliography should not be described as a meta-analysis proving accelerated lactate clearance. It combines heterogeneous studies, and lactate concentration at one time point is not a measurement of its removal rate. A reduction in CK also depends on exercise, sampling time and variability between individuals. For practice, a reproducible standardised test result, recovery and tolerability matter more than a promise to improve antioxidant “capacity” through an isolated marker.

Depression: why favourable tolerability does not establish equal efficacy

In the three-arm trial of rhodiola, sertraline and placebo, Hamilton scale changes were −5.1, −8.2 and −4.6 points, respectively. No convincing between-group difference was obtained. Adverse events were recorded in 30.0% of the rhodiola group, 63.2% of the sertraline group and 16.7% of the placebo group. These figures do not mean that rhodiola has the same risk as placebo or greater overall benefit than a standard antidepressant.

A preliminary trial does not confirm noninferiority of rhodiola or justify replacing prescribed treatment. Clinical decisions require assessment of symptom severity and duration, suicide risk, a history of affective episodes and access to methods with established efficacy. Treating “natural” origin as a reason to consider the preparation suitable for self-treatment of a depressive disorder is particularly unsafe.

How professionals can apply this knowledge

The first step when a person presents with fatigue is to identify its phenotype and temporal relationship to exertion, sleep, infection, medicines and dietary changes. Inadequate recovery, anaemia, thyroid disease, sleep apnoea, depression and other medical conditions may produce similar complaints. Investigations should depend on history and clinical findings; a universal “adaptogen panel” is unnecessary. Post-exertional worsening, syncope, unexplained weight loss or marked psychiatric symptoms make diagnostic assessment more important than consideration of an extract.

The second step is to identify the outcome that actually matters. For someone with stress-related fatigue, this may be daily functioning and fatigue severity; for an athlete, a prespecified test under consistent conditions. Cortisol, NPY, Hsp72 or changes in AMPK are not obligatory clinical targets. If a preparation is used within its authorised instructions and the professional’s scope of practice, the particular product, composition, concurrent interventions and reassessment time should be recorded, so that the entire benefit of a restored routine is not attributed to it.

The doses in the trial table are included to explain the evidence. They do not constitute a regimen of “asthenia 200–300 mg, chronic fatigue 400–600 mg, sport 300–400 mg”. Changing the form, standardisation or solvent makes such comparisons imprecise. Individual preparations have their own instructions, and use outside them requires separate clinical justification; doses of purified salidroside from animal experiments should not be converted into a home prescription.

Sleep should be discussed before administration and followed afterwards. Difficulty falling asleep, agitation or worsening anxiety calls for reassessment of the preparation. Evidence is insufficient for a universal “prohibited after 14:00” rule or established suppression of REM sleep, slow-wave sleep or the nocturnal growth hormone peak in humans by rhodiola. Preferring earlier administration when activating symptoms occur is a practical tolerability measure, not a precise chronobiological protocol.

Psychiatric safety and medicine combinations

A case of mania temporally associated with rhodiola use has been published. One case does not determine risk frequency or establish a mechanism, but warrants attention in people on the bipolar spectrum, those with previous activation episodes and those with a relevant family history. Reduced need for sleep, increasing agitation, rapid speech, impulsivity and unusual confidence require assessment, rather than a higher dose “for fatigue”. Self-directed use with such a history lacks sufficient justification.

In a crossover study of 13 healthy volunteers, two weeks of a commercial product reduced the losartan metabolite-to-parent ratio by 21%, consistent with CYP2C9 inhibition. No substantial effect was shown for the other isoenzymes studied. This is a limited signal, not proof of an interaction with every medicine listed in reference books; even such a signal deserves attention for a narrow therapeutic index.

Signs of possible serotonin toxicity, including agitation with tremor, hyperreflexia or clonus, autonomic disturbances and increased temperature, require prompt medical assessment, particularly when several serotonergic agents are involved. However, general features of this syndrome should not be presented as an inevitable effect of rhodiola with any antidepressant. A known risk from a particular medicine combination must be distinguished from a hypothesis for a herbal product that has not yet been quantified.

Evidence for self-directed use during pregnancy, breastfeeding or childhood is insufficient. An absence of identified harm in a brief adult trial does not resolve these questions. Allergic reactions, gastrointestinal complaints and individual activation should also be considered, even if most trial participants tolerated the preparation well.

Which investigations are actually useful?

Monitoring should be linked to a clinical question. CK and LDH can be relevant when muscle injury is suspected, but are not routine measures of “rhodiola efficacy”. Lactate can be useful in standardised sports testing with consistent sampling time and exercise load. MDA, oxLDL and GSH remain mainly research measures: their changes do not provide a generally accepted criterion for starting or continuing an extract.

Likewise, ALT, AST, glucose and HbA₁c should be assessed according to the patient’s condition and concurrent treatment. HbA₁c reflects a longer period of glycaemia and is poorly suited to assessing a brief course of a preparation “for stress”. For warfarin, the prescribed anticoagulation monitoring matters, rather than an abstract antioxidant panel. Linking investigations to their purpose reduces both overdiagnosis and a false sense of safety.

Rhodiola remains an interesting subject for neuropharmacology and metabolic research. The strongest descriptions concern individual preclinical interactions and some signals from small trials of particular extracts. For professionals, the value lies in precisely distinguishing composition, model, outcome and risk. This allows a discussion of potential benefit without promising to restore hormones, protect the heart or eliminate physical exhaustion through a single preparation.