# Lemon Balm and Rosmarinic Acid: The GABA Shunt, Stress, and the Limits of Clinical Evidence

- Canonical: https://nutrifit.health/knowledge/lemon-balm-rosmarinic-acid-gaba-t-inhibition
- Language: en
- Published: 2026-10-08T04:39:14.886Z
- Updated: 2026-10-08T04:39:14.886Z
- Author: NutriFit Editorial

> How the composition of Melissa officinalis relates to GABA-T, cholinergic activity, and autonomic symptoms. Chemistry, biosynthesis, pharmacokinetics, clinical studies, thyroid experiments, and how to read dosing regimens safely.

The lemony scent of lemon balm is easy to associate with evening tea and a sense of calm. Yet that scent comes mainly from volatile terpenes, while one of the most discussed neuropharmacological mechanisms concerns another part of the plant: its non-volatile polyphenols. Rosmarinic acid and some accompanying compounds can reduce the activity of an enzyme involved in breaking down gamma-aminobutyric acid. This raises a substantive question: can inhibitory neurotransmission be changed by acting on neurotransmitter metabolism rather than directly on its receptor?

The answer currently has several layers. In a laboratory system, inhibition of GABA transaminase has indeed been observed; in small human studies, certain lemon balm preparations changed self-rated calmness, cognitive-task results, and symptoms of disturbed sleep. But these observations do not complete the causal chain from swallowing a capsule to the GABA concentration in a particular brain region. Between those steps lie extract composition, absorption, metabolism, and access of the compound to an intracellular enzyme.

Lemon balm is therefore interesting both as a subject of biochemistry and as an example of how to read herbal pharmacology carefully. It cannot be reduced to a “natural benzodiazepine,” and its origin does not justify promises of calm without drowsiness, interactions, or limitations. The sections below cover its chemical composition, the GABA shunt, cholinergic and autonomic hypotheses, clinical studies, and the special question of the thyroid. Each mechanism is discussed with a clear boundary between established physiology and a proposed effect of the plant.

## What is actually in lemon balm leaves

### Two fractions that should not be treated as interchangeable

Lemon balm, *Melissa officinalis* L., belongs to the mint family, *Lamiaceae*. Its leaves contain polar phenolic compounds, essential oil, and other secondary metabolites. The botanical name matters: the everyday label “lemon mint” does not guarantee that a product contains this species, and a blend of several herbs is not equivalent to a single-herb lemon balm preparation.

An aqueous infusion, a hydroalcoholic extract, and an essential oil extract different parts of the plant’s chemical mixture. The polar fraction includes rosmarinic and other hydroxycinnamic acids, as well as flavonoids. The volatile fraction contains citral, citronellal, geraniol, nerol, linalool, and sesquiterpenes. Relative amounts depend on the source material, growing conditions, harvest time, drying, and extraction method; no single fixed percentage composition can be assigned to all leaves or oils.

| Compound group | Examples | Why this matters when reading studies |
| --- | --- | --- |
| Hydroxycinnamic acids and derivatives | Rosmarinic, caffeic, and chlorogenic acids; more complex phenolic derivatives | They differ in structure, oxidation, and metabolism; marker content does not describe all extract activity |
| Flavones and glycosides | Luteolin, cynaroside; apigenin, cosmosiin | A free aglycone and its sugar conjugate have different properties and should not be counted as the same amount of an active compound |
| Pentacyclic triterpenoid acids | Ursolic and oleanolic acids | Identified among compounds active in a laboratory GABA-T assay; they are not volatile essential-oil components |
| Monoterpenes | Geranial and neral, together called citral; citronellal, geraniol, nerol, linalool | They explain the aroma and some experimental effects of the oil, but do not replace studies of a polyphenol extract |
| Sesquiterpenes | β-Caryophyllene, germacrene D, caryophyllene oxide | Finding a compound with a known molecular target does not determine the effect of the entire ingested mixture |

This distinction helps prevent a common error: transferring an essential-oil effect to tea, or results from a standardized extract to any capsule containing leaf powder. β-Caryophyllene, for example, is a plant compound, not an endogenous human mediator. Its experimental activity at CB₂ receptors does not establish an anti-inflammatory or anxiolytic effect of lemon balm at a clinical dose.

### Rosmarinic acid: the chemistry of two aromatic fragments

Rosmarinic acid (RA) is an ester of caffeic acid and 3,4-dihydroxyphenyllactic acid. Its molecular formula is C₁₈H₁₆O₈, and its molar mass is about 360.31 g/mol; the natural molecule has a defined spatial configuration at its chiral center, usually designated R. For biochemistry, the key features are its two catechol aromatic rings, ester bond, and free carboxyl group—not the impressive length of its systematic name.

Each catechol ring has two adjacent phenolic OH groups. They participate in electron transfer and hydrogen bonding, which is why the compound reacts in radical-scavenging assays and can interact with proteins. But the chemical ability to scavenge radicals in solution is not equivalent to proven “antioxidant protection of the brain” after oral intake. Oxidation of the phenolic groups also changes the molecule: the parent acid, its conjugates, and its oxidation products may have different activities.

Hydrolysis provides a useful way to describe the structure without replacing an exact structural formula with a schematic drawing of rings:

$$
\text{Rosmarinic acid} + H_2O
\longrightarrow
\text{Caffeic acid} + \text{3,4-dihydroxyphenyllactic acid}.
$$

In a neutral overall equation, this is C₁₈H₁₆O₈ + H₂O → C₉H₈O₄ + C₉H₁₀O₅. This shows where the two fragments come from, but does not by itself specify where, how quickly, or to what extent hydrolysis occurs in the body. Those questions require separate studies of digestion and metabolism.

### How the plant assembles rosmarinic acid

Two aromatic-amino-acid pathways converge in plant cells. The phenylalanine branch produces an activated hydroxycinnamoyl fragment, while the tyrosine branch produces a hydroxyphenyllactate fragment. Their coupling, followed by hydroxylation, forms a molecule with two catechol rings. This is plant biosynthesis, not a pathway for synthesizing rosmarinic acid in the human brain.

$$
\text{L-Phenylalanine}
\xrightarrow{\mathrm{PAL}}
\text{Cinnamic acid}
\xrightarrow{\mathrm{C4H}}
\text{4-Coumaric acid}
\xrightarrow{\mathrm{4CL}}
\text{4-Coumaroyl-CoA}.
$$

$$
\text{L-Tyrosine}
\xrightarrow{\mathrm{TAT}}
\text{4-Hydroxyphenylpyruvate}
\xrightarrow{\mathrm{HPPR}}
\text{4-Hydroxyphenyllactate}.
$$

PAL stands for phenylalanine ammonia-lyase, C4H for cinnamic acid 4-hydroxylase, and 4CL for 4-coumarate-CoA ligase. In the other branch, TAT transfers tyrosine’s amino group, while HPPR reduces hydroxyphenylpyruvate. Rosmarinic acid synthase (RAS) transfers the hydroxycinnamoyl residue from CoA to hydroxyphenyllactate; CYP98-family enzymes carry out the required hydroxylations of the aromatic fragments. The detailed order and substrate specificity depend on the plant enzyme system, so this is a working map of biosynthesis rather than a universal sequence for every species.

Related phenolic acids also need precise names. Chlorogenic acid is an ester of caffeic and quinic acids; lithospermic and salvianolic acids are more complex derivatives. Calling lithospermic acid simply a “caffeic acid tetramer” is inaccurate, and the presence of individual complex acids cannot be treated as a fixed quantitative property of every lemon balm extract. Similarly, cynaroside is luteolin-7-O-glucoside, while cosmosiin is apigenin-7-O-glucoside; aglycone and glycoside names describe different molecules. Hesperetin and quercitrin, mentioned in phytochemical descriptions, are not varieties of these flavones: hesperetin is a flavanone, while quercitrin is a glycoside of the flavonol quercetin. Their presence and amounts also require analysis of the specific sample.

## The GABA shunt: the target is inside the mitochondrion

### From neurotransmitter synthesis to its breakdown

Gamma-aminobutyric acid (GABA) is the main inhibitory neurotransmitter in the mature central nervous system. It is formed from glutamate by decarboxylation: glutamate decarboxylase (GAD), including the GAD65 and GAD67 isoforms, uses pyridoxal phosphate (PLP), a derivative of vitamin B6. This cellular synthesis of the neurotransmitter should be distinguished from taking exogenous GABA and from acting on its receptors.

$$
\text{L-Glutamate}
\xrightarrow{\mathrm{GAD},\,PLP}
\text{GABA} + CO_2.
$$

At the nerve terminal, VGAT, also called VIAAT, helps package GABA into synaptic vesicles. After release, the neurotransmitter interacts with receptors, while GAT transporters remove it from the extracellular space; GAT-1 and GAT-3 are particularly relevant to neuronal and glial uptake. Their distribution depends on the region and cell type, so the simplified claim that one transporter is found only in neurons and another only in astrocytes does not describe the whole system.
Further catabolism is linked to mitochondrial enzymes. GABA transaminase (GABA-T), encoded by the *ABAT* gene, is a PLP-dependent aminotransferase. It transfers GABA’s amino group to α-ketoglutarate, producing glutamate and succinic semialdehyde (SSA). SSA is then oxidized by succinic semialdehyde dehydrogenase (SSADH) to succinate, which enters the tricarboxylic acid cycle.

$$
\text{GABA} + \alpha\text{-Ketoglutarate}
\xrightarrow{\mathrm{GABA\!-\!T}\,/\,ABAT,\,PLP}
\text{Succinic semialdehyde} + \text{L-Glutamate}.
$$

$$
\text{Succinic semialdehyde} + NAD^+ + H_2O
\xrightarrow{\mathrm{SSADH}}
\text{Succinate} + NADH + H^+.
$$

GABA-T is a dimeric enzyme; PLP is linked to a catalytic lysine through an internal aldimine, or Schiff base. On contact with the substrate, an external aldimine forms, and amino-group transfer proceeds through the pyridoxamine phosphate form of the cofactor. The second half-reaction with α-ketoglutarate restores PLP and produces glutamate. The catalytic residue number depends on the sequence used, so a specific protein numbering should not be transferred to every inhibition model.

The GABA shunt thus links neurotransmitter metabolism with energy metabolism. At physiological pH, compounds occur in their corresponding ionic forms; the equations show the biochemical direction of conversion, not every acid–base equilibrium. PLP acts as a regenerated cofactor, not a molecule consumed and needing to be added to every reaction product.

![The GABA shunt: synthesis in the cytosol and catabolism via GABA-T and SSADH in mitochondria; inhibition by lemon balm components is shown in a laboratory system]([![GABA is made from glutamate in the cytosol; after signaling and reuptake, its mitochondrial catabolism via GABA-T and SSADH yields succinate. Extract inhibition is shown in vitro; an effect in the human brain has not been established.](https://api.nutrifit.health/api/v2/public/knowledge/files/images/nutrifit-scientific-20261002-lemon-balm-gaba-shunt.en.png)](https://api.nutrifit.health/api/v2/public/knowledge/files/images/nutrifit-scientific-20261002-lemon-balm-gaba-shunt.en.png))

### What extract fractionation found

In a 2009 study, Awad and colleagues assessed extract activity using a GABA-T assay from rat brain. Repeatedly separating the mixture and reassessing activity identified rosmarinic, ursolic, and oleanolic acids. For rosmarinic acid, the authors described about 40% inhibition at 100 μg/mL; a possible contribution from interactions among components was discussed, but this alone did not establish clinical synergy.

This was an in-vitro result: the compound directly contacted the experimental enzyme system. The study did not measure brain GABA in people after a capsule or establish selectivity among all enzymes in the body. “Inactivation” should also not imply a proven irreversible blockade. Without full kinetic data, a competitive or mixed inhibition type cannot be confidently selected, and without structural analysis, specific amino-acid contacts should not be listed as fact.

The numbers need the same care. A concentration in mg/mL refers to the mass of a particular compound or mixture; conversion to molarity is possible only for a compound with a known molar mass. If 0.35 mg/mL refers specifically to pure rosmarinic acid, the calculation is:

$$
\frac{0.35\,\mathrm{g/L}}{360.31\,\mathrm{g/mol}}
\approx 0.000971\,\mathrm{mol/L}
= 971\,\mathrm{μmol/L}.
$$

It is therefore not equivalent to 40–65 μmol/L. The conversion itself does not confirm that 0.35 mg/mL is a reliable IC₅₀ for the pure compound; it demonstrates the unit error in that claim. IC₅₀—the concentration producing half-maximal inhibition in a specified assay—depends on assay conditions and is not a required plasma concentration or a therapeutic dose for people.

Comparison with vigabatrin is useful only for understanding the target. Vigabatrin is a mechanism-based, irreversible GABA-T inhibitor with distinct indications and serious limitations, including risk of visual damage. Sharing an enzyme name does not make lemon balm its mild equivalent, a substitute for antiseizure treatment, or a product with proven long-term ophthalmic safety.

### Phasic and tonic inhibition: a possible consequence, not a measured effect

Phasic inhibition is associated with brief synaptic events after GABA release. Tonic inhibition arises from more sustained activation of extrasynaptic receptors by ambient neurotransmitter; its magnitude depends on release, uptake, spatial diffusion, and receptor properties. Some extrasynaptic GABA_A complexes contain a δ subunit, but receptor composition is not identical across brain regions.

GABA_A is a ligand-gated anion channel. The net inhibitory effect depends on electrochemical gradients and membrane conductance: it may involve hyperpolarization and shunting of excitatory currents. GABA_B is a metabotropic receptor; its effects are mediated by G proteins, changes in potassium- and calcium-channel activity, and regulation of neurotransmitter release. The two receptors should not be collapsed into a single “inward chloride current.”

In theory, inhibiting catabolism could change the available GABA pool. It does not automatically follow that all vesicles become more loaded, every quantal release grows in amplitude, or a particular receptor-subunit combination is activated. For lemon balm, these sequences remain hypotheses until measured in the relevant tissue at an achievable exposure. It is also not established that botanical inhibition rules out tolerance, desensitization, or symptoms after stopping long-term use.

Benzodiazepines and Z-drugs affect receptor transmission differently: they are positive allosteric modulators of certain GABA_A complexes, not orthosteric substitutes for GABA. Their effects and risks vary by drug and regimen. Comparing them helps distinguish receptor and metabolic targets, but does not justify promising that lemon balm will preserve alertness in every user.

## Why a cup of tea is not equivalent to a concentration in an enzyme assay

### Absorption, conjugation, and intestinal transformations

For an intracellular target, it is not enough for a compound to be present in the plant. After oral intake it must be released from the preparation, absorbed, survive metabolic transformations, reach the relevant tissue, and enter the relevant compartment. Each step can change not only the amount of the parent compound but also which molecule actually interacts with a protein.

Rosmarinic acid undergoes hydrolysis and metabolism; possible transformations of phenolic compounds include methylation, sulfation, and glucuronidation. Body tissues and the gut microbiota participate, but not every reaction should be attributed to microbial esterases. An esterase cleaves an ester bond; subsequent conjugation is a different biochemical step. The parent acid, its conjugates, and derivatives of its caffeic-acid fragment should therefore be distinguished in both analytical methods and discussions of activity.

$$
\text{Parent phenolic compound}
\longrightarrow
\begin{cases}
\text{hydrolyzed derivatives},\\
\text{methylated derivatives},\\
\text{sulfate and glucuronide conjugates}.
\end{cases}
$$

This scheme lists classes of transformations, not one linear route. A total concentration measured after analytical deconjugation is not the same as the free concentration of the parent compound before sample processing. Free exposure is especially important when comparing it with an enzyme-inhibition assay.

A focused example comes from a small pharmacokinetic study by Noguchi-Shinohara and colleagues in 2015. It included 11 healthy volunteers and examined single doses of an extract with known rosmarinic-acid content and the effect of food. For the fasted regimen containing 500 mg of rosmarinic acid itself, the authors reported a peak total serum concentration of about 162 nmol/L after one hour. That is about 0.162 μmol/L, not hundreds of micromoles; 500 mg of the marker compound must also not be confused with 500 mg of extract.

This observation cannot directly establish clinical efficacy, but it shows why enzyme-assay and serum concentrations are not interchangeable. The study reported that food affected exposure and time to peak, but did not measure inhibition of brain GABA-T. Tolerability after a single dose in a small group does not establish the safety of long-term use, use during pregnancy, or complex drug combinations.

### The blood–brain barrier is a separate question

The blood–brain barrier (BBB) regulates exchange between blood and nervous tissue. For a polar phenolic compound, its presence in plasma proves neither adequate entry into the brain nor access to a mitochondrial target. Even finding a compound in animal tissue requires distinguishing parent molecule from metabolites, free from bound fractions, and tissue concentration from concentration at the site of action.

The materials provided do not establish human transport of rosmarinic acid through monocarboxylate transporters (MCTs) or accumulation of therapeutic micromolar concentrations in the hippocampus. This mechanism should therefore not be drawn as a solid, proven arrow. Similarly, one half-life range does not justify mandatory twice-daily dosing: a regimen should be based on the specific product, clinical data, and its directions, not solely on the presumed kinetics of one component.

Improvement in well-being does not necessarily require the entire central GABA-T hypothesis to be correct. An extract may have several targets, and a clinical change may have several explanations. The research task is to distinguish them; when reading the evidence in practice, the observed effect of a product and the proof of its proposed mechanism should be assessed separately.

## Cholinergic activity and “calm alertness”: an appealing explanation with mixed results

### Receptor binding does not mean receptor activation

Acetylcholine participates in regulating attention, memory, and autonomic functions. Nicotinic receptors (nAChRs) are ion channels, whereas muscarinic receptors (mAChRs) signal through G proteins. Acetylcholinesterase (AChE) breaks down the neurotransmitter; changing its activity and directly interacting with a receptor are distinct mechanisms that cannot be combined based on one positive assay.

Studies by Wake and colleagues and the Kennedy research series included receptor-binding assessments of plant preparations. Displacement of a labeled ligand—for example, a nicotinic or muscarinic marker—shows interaction with the system under study. By itself, it does not establish whether a component is an agonist, antagonist, or allosteric modulator, and does not prove a central effect after oral intake. Extracts with different binding results therefore cannot be treated as pharmacologically identical.

It is useful to retain the normal signaling pathways when considering the hypothesis. Some nicotinic receptors, including α7- and α4β2-containing complexes, participate in regulating transmission and attention. M1 muscarinic receptors can activate the Gq/11–phospholipase cascade:

$$
M_1 \longrightarrow G_{q/11}
\longrightarrow PLC
\longrightarrow IP_3 + DAG
\longrightarrow \text{Ca}^{2+}\text{ signal and PKC}.
$$

PLC cleaves a membrane phosphoinositide substrate to form the second messengers IP₃ and DAG. IP₃ is associated with mobilizing intracellular calcium, while DAG participates in protein kinase C activation. These mechanisms describe receptor physiology; they do not confirm that lemon balm selectively activates M1, induces long-term potentiation, or prevents cortisol-related memory impairment in people.

Individual findings on AChE inhibition should likewise be read at the level of the specific extract and experimental method. Without confirmed exposure, one cannot claim that an IC₅₀ is the same in erythrocyte and cortical systems or promise a physiological increase in acetylcholine. Lemon balm does not become equivalent to a studied medicinal AChE inhibitor merely because such assays exist.

### What remains of the cognitive paradox

“Calm alertness” describes the desired combination of reduced tension and preserved performance. In studies, mood and task performance could indeed change in different ways depending on dose, sample, and measurement time. It is therefore more accurate to describe individual findings than a universal profile in which cholinergic activity necessarily offsets sedation.

In Kennedy’s 2004 experiment, 18 healthy participants received 300 mg, 600 mg, and placebo in a crossover design. At 600 mg during laboratory stress, self-rated calmness increased but self-rated alertness decreased; faster mathematical processing without reduced accuracy was observed at 300 mg. These were different dose-related findings. They do not show that the higher dose improved all cognitive measures at once and ruled out drowsiness.

The earlier studies by Kennedy in 2002 and 2003 are important as investigations of acute changes in mood and task performance. Small crossover experiments allow comparison of each participant across conditions, but are sensitive to task selection, measurement timing, and multiple comparisons. They cannot be turned into promises of improved memory with long-term use, dementia treatment, or equal efficacy for every extract labeled with a percentage of rosmarinic acid.

The flavonoid fraction adds another hypothesis but does not remove these limitations. Experiments on individual flavones, including apigenin, do not justify calling all lemon balm flavones “weak partial agonists at the benzodiazepine site.” The benzodiazepine site is allosteric, and the direction of an effect depends on the compound and receptor system. A separate article on [apigenin, GABA receptors, and sleep](/en/knowledge/apigenin-gaba-cd38-nad-sleep) discusses this distinction; its conclusions also should not automatically be transferred to a lemon balm mixture.

## Palpitations and gut symptoms: how stress becomes physical

### Autonomic pathways and the limits of attributing them to lemon balm

Anxiety can accompany a rapid heartbeat, a sense of skipped beats, muscle tension, and abdominal discomfort. Limbic structures, the hypothalamus, and brainstem centers participate in stress signaling. The amygdala helps assess threat salience, the hypothalamic paraventricular nucleus is linked to the neuroendocrine response, the locus coeruleus to noradrenergic regulation, and the rostral ventrolateral medulla (RVLM) to control of sympathetic outflow.

This network is not a direct sequence of equivalent inhibitory switches. GABA can inhibit different cells with different downstream effects; for example, inhibiting an inhibitory neuron disinhibits the next element. The nucleus ambiguus and dorsal motor nucleus of the vagus participate in parasympathetic control, but vagal activity cannot be increased simply by adding GABA to all of these structures. The outcome depends on cell type, inputs, and local circuitry.

For lemon balm, a hypothesis has been proposed in which reduced stress reactivity changes bodily sensations downstream. It does not prove measured blockade of GABA-T in the RVLM, direct activation of vagal nuclei, or selective reduction of effects at cardiac β₁ receptors. A symptomatic reduction in palpitations also does not establish antiarrhythmic efficacy or absence of an effect on cardiac conduction. Terms such as “cardioneurosis” and “vegetative dystonia” should not replace a description of symptoms and diagnostic assessment of their causes.

The neuroendocrine branch of stress is described by the hypothalamic–pituitary–adrenal axis: CRH → ACTH → cortisol. A change in salivary cortisol in an individual experiment may be an interesting biomarker, but does not prove “normalization” of the entire axis. Interpretation requires the time of day, stressor, measurement method, and clinical outcomes; the concentration of one hormone does not automatically explain every change in sleep, attention, or palpitations.

### What heart-rate variability measures

Heart-rate variability (HRV) reflects fluctuations in the intervals between successive heartbeats. Analyses generally use normal-to-normal (NN) intervals after artifact processing and exclusion of non-sinus events. HRV is a useful measure of sinus-node regulation, but it depends on breathing, body position, recording duration, age, fitness, and medication. It is not a diagnosis by itself or a direct measure of “calmness.”

One time-domain measure is RMSSD, the square root of the mean of squared successive differences between NN intervals. It is usually expressed in milliseconds and, under comparable short-recording conditions, is associated with vagal modulation. pNN50 is the percentage of adjacent interval pairs that differ by more than 50 ms. Defining the measure is different from making a clinical claim that lemon balm necessarily increases it.

$$
RMSSD = \sqrt{\frac{1}{N-1}\sum_{i=1}^{N-1}(NN_{i+1}-NN_i)^2}.
$$

$$
pNN50 = \frac{\#\{|NN_{i+1}-NN_i|>50\,\mathrm{ms}\}}{N-1}\times100\%.
$$

Spectral analysis often uses the LF range of 0.04–0.15 Hz and HF range of 0.15–0.40 Hz. Under certain conditions, HF is associated with respiratory modulation and vagal influences, but its magnitude is not independent of breathing rate. LF reflects mixed influences and baroreflex regulation; the LF/HF ratio should not uncritically be called a precise sympathovagal balance. This is especially problematic when interpreting short wearable-device recordings without respiratory and artifact controls.

The clinical evidence provided does not justify promising a specific increase in RMSSD, pNN50, or HF after lemon balm. Observations of fewer palpitations do not replace a dedicated HRV study. If these measures are used to track a person’s state, they require a comparable protocol and independent clinical interpretation—not herbal dose selection based on a daily number in an app.

### The gut–brain axis and a local antispasmodic hypothesis

Stress affects motility, perception of visceral signals, and eating behavior. The central nervous system, enteric networks, and autonomic signaling all participate. Reduced tension may therefore accompany reduced subjective discomfort, but this link does not prove that an extract treats irritable bowel syndrome or removes the cause of pain.

Local effects on smooth muscle are also described for plant compounds. Terpenes and flavonoids may alter contraction in experimental systems where calcium entry, including through voltage-gated channels, matters. However, isolated tissue, essential oil, and an oral polyphenol extract are different interventions. Specific blockade of intestinal Caᵥ1.2 after an ordinary lemon balm dose cannot be presented as an established human mechanism.

The practical boundary is broader than the laboratory hypothesis. Traditional use of the leaf for mild bloating and flatulence does not justify treating persistent pain, bleeding, weight loss, or severe diarrhea with it. Likewise, new palpitations accompanied by fainting, chest pain, or marked shortness of breath require medical assessment; attributing symptoms to stress is appropriate only after other causes have been considered.

## What has actually been studied in people

### Replace the general word “effective” with a specific research question

Clinical research on lemon balm includes acute laboratory-stress studies, small trials involving sleep disturbance, and work in specific patient groups. These studies differ in product, control, duration, and outcome. A lower self-report score, a change in calculation speed, and fewer episodes of palpitations are not the same endpoint and do not establish one general treatment indication.

It is especially important to distinguish an open observation from a randomized, placebo-controlled trial. In the former, improvement after starting treatment may include natural symptom fluctuation, expectations, and regression to the mean. In the latter, a control helps separate these factors, but a small sample and short duration still leave uncertainty. Statistical significance on one scale does not automatically tell us how noticeably or durably a person’s daily functioning changed.

| Study or group of studies | What was studied | Appropriate interpretation |
| --- | --- | --- |
| Kennedy, 2002 and 2003 | Acute changes in mood and cognitive tasks after specific lemon balm samples | Individual measures may be modulated; a single effect on memory and alertness across all products is not established |
| Kennedy, 2004 | Crossover study in 18 healthy volunteers, 300 and 600 mg, laboratory multitasking load | Changes depended on dose; calmness and subjective alertness did not improve together at 600 mg |
| Scholey, 2014 | Foods containing lemon balm and an acute stress protocol | Food matrix and intervention composition matter; results from a drink or yogurt cannot simply be transferred to a capsule |
| Cases, 2011 | Open pilot in 20 volunteers with anxiety and sleep problems; 600 mg extract daily, 15 days | Symptoms improved, but without placebo the causal treatment effect cannot be determined |
| Alijaniha, 2015 | Short randomized trial for benign palpitations; 500 mg extract twice daily, 14 days | Symptom results for a specific preparation, not proof of arrhythmia treatment |
| Haybar, 2018 | 80 patients with stable angina; 3 g plant material daily, 8 weeks | Positive status and sleep-scale results in a defined group; not proof of improved coronary blood flow |
| Soltanpour, 2019 | Anxiety and sleep quality in the context of coronary bypass surgery | A narrow clinical setting; findings do not replace perioperative care or automatically apply to healthy people |
| Ghazizadeh, 2021 | Systematic review and meta-analysis of anxiety and depression trials | Pooled signals of benefit remain limited by heterogeneity of the original interventions and studies |

### Laboratory stress and products containing the extract

Defined Intensity Stressor Simulation (DISS) creates simultaneous demands on several cognitive functions. The protocol makes it possible to assess responses to a controlled task rather than waiting for an unpredictable stressful event. However, healthy volunteers taking part in a brief experiment are not equivalent to patients being treated for generalized anxiety disorder or chronic insomnia.

The Kennedy findings already show why a universal promise of calm concentration is too strong. For practical decisions, a person’s own drowsiness, reaction speed, and specific task matter, not only group averages. A new preparation should not be tried for the first time before an exam, a long drive, or work requiring fast reactions just because one small study found improvement on a mathematics task.

In the study by Scholey and colleagues, the extract was included in a food matrix. This is a distinct research approach: the drink or yogurt composition, conditions of intake, and combination of ingredients can affect tolerability and exposure. Even if a particular regimen changes self-rated anxiety or a stress biomarker, that does not establish universal prevention of a cortisol response with any cup of tea. The experimental conditions are part of the result, not a secondary detail.

### A pilot study of sleep and anxiety

The Cases study examined Cyracos, a standardized extract, in a short open-label regimen. The source describes 600 mg per day divided into two doses, 20 participants, and 15 days of observation. It reported reductions of about 18% in anxiety manifestations and about 42% in insomnia symptoms; these percentages refer to changes in scores within a small observational study, not differences from placebo.

They therefore should not be presented as a probability of recovery or an expected effect for every patient. Even a high proportion of participants who respond does not remove the absence of a control group or the limited duration. The useful conclusion is that the preparation merited further controlled study, not that it produced remission of a diagnosed disorder. Persistent insomnia also has causes and behavioral factors that a brief herbal pilot does not address.

### Palpitations and stable angina

The Alijaniha trial concerned complaints of benign palpitations, not treatment of every form of tachycardia. It used dried leaf extract at 500 mg twice daily for two weeks; the supplied evidence describes a reduction in episode frequency and some associated symptoms. The word “extract” matters: it should not be turned into 500 mg of unprocessed dried leaf in the same table.

A short-term improvement in subjective episodes does not prove prevention of atrial fibrillation, ventricular arrhythmias, or complications. It also does not confirm “100% safety”: a small trial may fail to detect rare adverse reactions. In someone with palpitations, the first step is to determine what the person is experiencing and whether there is a structural, electrical, endocrine, or medication-related cause.

The Haybar study in stable angina used DASS-21 and the Pittsburgh Sleep Quality Index (PSQI), not the HADS scale cited in the source. Of 80 enrolled patients, 73 completed the trial; the regimen was 3 g daily for eight weeks. Improvements were reported in depression, anxiety, stress, and sleep scores. These results should not be assigned an effect on HRV, ischemia, or prognosis; a reduction in hs-CRP should likewise not be attributed to this publication on psychological scales without a separate source.

### Menopause and review findings

Sleep disturbance and anxiety occur during peri- and postmenopause, but their origin cannot be reduced to a universal decline in GABA. Hot flashes, thermoregulation changes, mood, pain, and coexisting conditions can affect sleep at night in different ways. A change in a subjective sleep scale cannot therefore be equated with restoration of slow-wave sleep stages without an instrumental sleep study.

The reference provided for an alleged 2021 Shirazi trial does not match the stated title: the cited article identifier refers to a review of prenatal music therapy. It does not support claims that lemon balm reduces hot flashes or restores sleep architecture. Until a correctly identified primary study is available, these specific clinical promises are excluded, and the use of lemon balm for menopausal symptoms remains a separate question not adequately supported by this evidence base.

A meta-analysis of anxiety and depression helps summarize the overall findings, but does not make different extracts identical. Its interpretation must account for the design of the original studies, concomitant medicines, populations, scales, and duration. Taiwo’s mouse behavioral experiments are preclinical: anxiety-like or depression-like test results are not a human diagnosis and do not confirm efficacy for treating clinical depression.

## The thyroid: older experiments and treatment claims that go too far

### Antithyrotropic activity is not the same as treatment for thyrotoxicosis

Thyroid-stimulating hormone (TSH) binds the TSH receptor on the surface of a thyroid cell and initiates intracellular signaling. In Graves’ disease, stimulating antibodies can activate the same receptor. A substance’s effect on TSH or stimulating-immunoglobulin binding is therefore of interest in experimental endocrinology, but must be distinguished from the efficacy and safety of oral treatment.

In studies by Auf’mkolk and colleagues in the mid-1980s, plant extracts and oxidized phenolic components were examined. The studies described formation of products that interact with TSH and attenuation of receptor binding or biological activity of stimulating immunoglobulins in experimental systems. Sample composition and oxidation mattered: the effects cannot be attributed without qualification to any unoxidized rosmarinic acid in a fresh infusion.

Different results were also obtained for different plants and preparations. For example, a study of Graves’ immunoglobulins directly examined the binding of labeled antibodies to human thyroid membranes for an extract of *Lithospermum officinale*. The presence of *Melissa officinalis* among other tested extracts does not allow every experiment in that paper to be attributed automatically to lemon balm. Human membranes or immunoglobulins in a laboratory experiment also do not constitute a clinical treatment trial in patients.

The signaling sequence is useful to write separately from the proposed intervention point:

$$
\text{TSH or stimulating antibody}
\longrightarrow TSHR
\longrightarrow G_s
\longrightarrow \text{adenylyl cyclase}
\longrightarrow cAMP
\longrightarrow PKA.
$$

Adenylyl cyclase forms cAMP from ATP, and cAMP activates protein kinase A; cAMP itself is not a protein kinase. An effect on ligand binding occurs upstream of these steps. It does not necessarily mean direct blockade of the receptor protein, and a list of “covalent and pseudocovalent bonds” should not replace experimentally established evidence about the nature of the interaction.

![Thyroid signaling: TSH and stimulating antibodies activate the TSH receptor and cAMP cascade; effects of oxidized plant components are experimental findings]([![TSH or stimulating antibodies activate the thyrocyte TSH receptor and the Gs–adenylyl cyclase–cAMP–PKA pathway. Experimental effects of oxidized plant components do not prove that lemon balm treats hyperthyroidism.](https://api.nutrifit.health/api/v2/public/knowledge/files/images/nutrifit-scientific-20261002-lemon-balm-thyroid-boundary.en.png)](https://api.nutrifit.health/api/v2/public/knowledge/files/images/nutrifit-scientific-20261002-lemon-balm-thyroid-boundary.en.png))

### Deiodination and clinical caution

Another process is the conversion of thyroxine (T₄) to triiodothyronine (T₃). An iodine atom is removed from the outer ring of the iodothyronine; the selenoprotein deiodinases D1 and D2, which have different tissue distributions, participate in producing active T₃. This is a distinct target and is not the same as TSH receptor activity.

$$
T_4 \xrightarrow{\mathrm{D1/D2},\,\text{outer-ring deiodination}} T_3.
$$

The evidence supplied does not establish clinically meaningful inhibition of D1 after ordinary lemon balm use. Still less can data from other plants or direct contact between a compound and an enzyme be transferred to capsule treatment of hyperthyroidism. Deiodinases therefore remain a possible level of intervention to discuss, not a proven second pathway for antithyroid treatment.

People with hypothyroidism, autoimmune thyroiditis, or levothyroxine treatment should reasonably discuss regular use of a concentrated extract with their clinician. However, laboratory findings alone do not prove that every cup of lemon balm will worsen thyroid function or that the plant is an absolute contraindication for every thyroid diagnosis. Decisions about laboratory monitoring depend on the person’s condition, treatment changes, and clinical question; no universal testing schedule after lemon balm is established here.

In particular, an obligatory four-hour interval between lemon balm and levothyroxine cannot be inferred from an antithyrotropic experiment. Separating doses can help with certain proven absorption interactions involving other substances, but does not eliminate a possible systemic pharmacodynamic interaction. A specific interval and the nature of an interaction with lemon balm are not confirmed in this evidence base. Follow the usual instructions and individualized plan for prescribed levothyroxine.

These experiments also do not justify recommending lemon balm instead of treatment for Graves’ disease, subclinical hyperthyroidism, goiter, or “thyrotoxic autonomic dysfunction.” When anxiety, tremor, rapid pulse, and weight change occur together, it is especially important to assess a thyroid cause rather than mask symptoms with a calming product. Core questions about hormone synthesis and testing are covered separately in the article on [iodine and thyroid function](/en/knowledge/iodine-thyroid-hormones-tests-safe-intake).

## Practice starts with the identity of the product and the reason for using it

### A standardization marker controls composition; it does not guarantee an effect

Rosmarinic-acid content is useful as one analytical marker. But the marker percentage does not automatically tell us the amount of all active compounds, their bioavailability, or the degree to which they affect GABA-T. A range of 2–7% should not be declared a universal therapeutic standard, nor an extraction ratio of 4:1–6:1 a necessary condition for clinical efficacy. These are characteristics of particular products and require their own justification.

To compare a product with a study, relevant details include plant species, plant part, formulation, extraction solvent, drug-to-extract ratio, analytical method, and dose. The drug-extract ratio (DER) describes the ratio of the starting plant material’s mass to the extract mass, but does not replace quantitative analysis. HPLC can assess marker compounds; additional checks of identity, contaminants, and stability concern quality and do not prove treatment efficacy.

The marker amount is easy to calculate if the label really gives its mass fraction in the extract:

$$
\text{Marker mass}
= \text{Extract mass}\times\frac{\text{Marker content, \%}}{100}.
$$

For example, 300 mg of an extract containing 7% rosmarinic acid mathematically contains 21 mg of the marker. This is an illustrative calculation, not a dose recommendation. It does not prove that two extracts with the same amount of rosmarinic acid have the same effect, and does not allow the extract to be replaced with a corresponding amount of essential oil.

The Cyracos product discussed in the Cases pilot used a specific extract standardized for phenolic components. This fact should be retained when reading the result, but neither the brand nor its percentage turns an open-label study into a proven protocol. For another batch or formulation, comparability must be established separately; the study dose should not be treated as a general market-wide rule for supplements.

### What can be carried into a discussion about use

EMA/HMPC considers lemon balm leaf under traditional use for mild symptoms of stress, to aid sleep, and for mild digestive complaints including bloating and flatulence. “Traditional” here means that long-standing use supports the plausibility of use, while clinical data are insufficient for a firm conclusion about efficacy. It is not regulatory confirmation of treatment for an anxiety disorder, depression, arrhythmia, or thyroid disease.

For the relevant medicinal preparations, the regulatory description concerns adults and adolescents over 12 years of age; medical advice is recommended if symptoms persist after two weeks. The specific regimen should be taken from the product instructions. These conditions do not establish one general course for every food, essential oil, or concentrated supplement.

The practical goal should be defined before choosing a dose. Mild, short-lived tension, persistent insomnia, panic episodes, and palpitations have different diagnostic and treatment pathways. The table of study regimens above helps explain the literature but is not a prescribing plan: one should not simply select the “anxiety” row and increase a dose to a figure from another study. Universal instructions to take the product 45–60 minutes before stress, use it for 4–8 weeks, or treat tachycardia for 3–6 weeks are also unsupported.

If a product is used according to its directions, it is useful to observe a specific outcome and tolerability: is it easier to fall asleep, has tension decreased, has daytime drowsiness appeared? Continually adding new sedative ingredients makes this assessment harder. If there is no benefit, reconsider the cause of the symptoms and the treatment approach rather than treating lack of effect as proof of “insufficient GABA-T blockade” and automatically increasing the dose.

## Safety: uncertainty must not be presented as absence of risk

### Drowsiness and drug combinations

Reduced subjective alertness in a clinical experiment already shows why absence of sedation cannot be promised. Combining lemon balm with alcohol, sleeping medication, or other central nervous system depressants may add drowsiness and impair reaction. Until an individual response is known, consider driving and operating machinery; “plant-based” origin does not remove this practical limitation.

Combinations with benzodiazepines, Z-drugs, barbiturates, and other sedatives require consideration of the specific combination. Do not reduce a prescribed medicine’s dose on your own to “make room” for lemon balm. The evidence provided does not confirm a protocol for using lemon balm to prevent benzodiazepine withdrawal. Discontinuation or dose reduction should follow a separate medical plan, especially after long-term use.

Nor can general compatibility with all SSRIs and tricyclic antidepressants be guaranteed, or a reduction in treatment-emergent agitation promised. Possible effects depend on the medicine, other ingredients in the mixture, the person’s condition, and treatment goals. Experimental receptor activity of lemon balm does not prove clinical antagonism with any anticholinergic medicine; assessing that requires real exposure and outcomes from the combination.

### Metabolic interactions and vulnerable groups

The cytochrome P450 system metabolizes many medicines. An in-vitro inhibition result for one isoenzyme is not equivalent to an interaction in a patient, but the absence of strong induction in one system does not guarantee the absence of all interactions. Comparison with St John’s wort does not confirm lemon balm’s safety with every substrate of CYP1A2, CYP2D6, CYP3A4, CYP2C9, or CYP2E1. For medicines with a narrow therapeutic range in particular, consider the specific combination rather than relying on a general claim of “insignificant CYP effects.”

Short studies and traditional use provide information about tolerability of certain forms, but do not cover every circumstance. Irritation, an allergic reaction, or gastrointestinal discomfort can occur with a plant product too. Leaf powder, extract, tincture, and essential oil have different limitations; concentrated oil should not be treated as a dose-equivalent substitute for an infusion or taken orally on the basis of leaf studies.

There are insufficient clinical data on concentrated preparations during pregnancy and breastfeeding. The absence of a particular toxic signal in an animal experiment does not establish safety for a fetus or child. For the same reason, this article excludes unreferenced claims that teratogenicity and mutagenicity studies were uniformly negative and that one LD₅₀ applies to all extracts: an acute toxicity measure does not describe chronic use or replace human data.

People with thyroid disease, complex psychotropic treatment, a cardiovascular diagnosis, or planned surgery should tell their care team about herbal products. This helps clinicians interpret symptoms, possible drowsiness, and other medicines. Such a discussion is especially useful when several supplements are used at once and the ingredients or dose of each product are unclear.

## What lemon balm can teach us about herbal neuropharmacology

The strength of this topic is a specific biochemical question. There is a plant molecule with a defined structure, an enzyme in the GABA shunt, and a laboratory finding that can be reproduced and refined. Other levels sit alongside it: receptor binding, stress symptoms, and small human trials. They add to the picture but cannot substitute for one another.

For a specialist, the most useful sequence is “which preparation — what exposure — which system — what outcome.” It preserves interest in GABA-T, cholinergic activity, and thyroid experiments without turning them into nonexistent clinical protocols. For a general reader, the conclusion is more practical: product quality and formulation matter, promises of calm without drowsiness warrant caution, and persistent or new symptoms need their causes assessed. This precision makes an article about lemon balm more useful than the marketing image of a universal calming herb.

## FAQ

### Does lemon balm really increase GABA in the brain?

Components of the extract have been shown to inhibit GABA-T in a laboratory system. Adequate exposure and reduced activity of this enzyme in the human brain after ordinary use have not been established; feeling better does not by itself confirm the mechanism.

### Can I expect to feel calm without becoming drowsy?

That cannot be guaranteed. In a laboratory-stress experiment, 600 mg increased calmness but reduced subjective alertness. The result depends on the product, dose, and person; consider driving and combining it with other sedatives.

### Are tea, essential oil, and standardized extract interchangeable?

No. They extract different leaf fractions and produce different exposures. The percentage of rosmarinic acid describes one marker but does not make all preparations equivalent; essential oil is not a dose-for-dose substitute for an infusion.

### What doses were used in studies?

Regimens varied: a specific extract was studied at 300 and 600 mg in an acute-stress experiment, at 600 mg daily in an open pilot, and at 500 mg extract twice daily for benign palpitations. These are study descriptions, not a universal dosing plan; follow the directions for the specific product.

### Can lemon balm treat arrhythmia?

A trial in people with benign palpitations assessed symptoms in a particular group. It does not confirm treatment of arrhythmias or prevention of complications. New palpitations, especially with fainting, chest pain, or shortness of breath, need medical assessment.

### Can lemon balm be used for hyperthyroidism or Graves’ disease?

Laboratory antithyrotropic effects do not confirm clinical efficacy of oral treatment. Lemon balm does not replace diagnosis, antithyroid therapy, or monitoring; people with thyroid disease should discuss regular use of a concentrated extract with their clinician.

### Should lemon balm and levothyroxine be separated by four hours?

The evidence provided does not confirm that a mandatory interval is needed specifically for lemon balm. Separating doses does not eliminate a possible systemic interaction. Take levothyroxine according to its instructions and individualized plan, and discuss adding a regular extract separately.

### Can lemon balm replace sleeping medication or ease benzodiazepine withdrawal?

Do not replace prescribed treatment or reduce its dose on your own. The evidence base does not establish a lemon balm protocol for preventing benzodiazepine withdrawal; reducing medication requires a separate medical plan.

### What does traditional herbal use mean?

It reflects long-standing use and plausible use for mild complaints, not proven treatment of an anxiety disorder or insomnia. EMA/HMPC notes that clinical data are insufficient and advises seeking medical advice if symptoms persist after two weeks of appropriate use.

### Is lemon balm proven to reduce menopausal hot flashes?

The reference given in the source does not support that claim; it concerns a different topic. Reduction of hot flashes and restoration of sleep stages are not presented as established effects of lemon balm; a correctly identified primary clinical source would be needed.

## Sources

- [Awad R et al. Bioassay-guided fractionation of lemon balm (Melissa officinalis L.) using an in vitro measure of GABA transaminase activity](https://doi.org/10.1002/ptr.2712)
- [Kennedy D.O et al. Modulation of mood and cognitive performance following acute administration of Melissa officinalis (lemon balm)](https://doi.org/10.1016/s0091-3057(02)00777-3)
- [Kennedy D.O et al. Modulation of mood and cognitive performance following acute administration of single doses of Melissa officinalis (Lemon balm) with human CNS nicotinic and muscarinic receptor-binding properties](https://doi.org/10.1038/sj.npp.1300230)
- [Kennedy D.O et al. Attenuation of laboratory-induced stress in humans after acute administration of Melissa officinalis (Lemon Balm)](https://doi.org/10.1097/01.psy.0000132877.72833.71)
- [Scholey A et al. Anti-stress effects of lemon balm-containing foods](https://doi.org/10.3390/nu6114805)
- [Cases J et al. Pilot trial of Melissa officinalis L. leaf extract in the treatment of volunteers suffering from mild-to-moderate anxiety disorders and sleep disturbances](https://doi.org/10.1007/s12349-010-0045-4)
- [Alijaniha F et al. Heart palpitation relief with Melissa officinalis leaf extract: double blind, randomized, placebo controlled trial of efficacy and safety](https://doi.org/10.1016/j.jep.2015.02.007)
- [Haybar H et al. The effects of Melissa officinalis supplementation on depression, anxiety, stress, and sleep disorder in patients with chronic stable angina](https://doi.org/10.1016/j.clnesp.2018.04.015)
- [Auf'mkolk M et al. The active principles of plant extracts with antithyrotropic activity: oxidation products of derivatives of 3,4-dihydroxycinnamic acid](https://doi.org/10.1210/endo-116-5-1677)
- [Auf'mkolk M et al. Extracts and auto-oxidized constituents of certain plants inhibit the receptor-binding and the biological activity of Graves' immunoglobulins](https://doi.org/10.1210/endo-116-5-1687)
- [Sourgens H et al. Antihormonal effects of plant extracts. TSH- and prolactin-suppressing properties of Lithospermum officinale and other plants](https://doi.org/10.1055/s-2007-971249)
- [Wake G et al. CNS acetylcholine receptor activity in European medicinal plants traditionally used to improve failing memory](https://doi.org/10.1016/s0378-8741(99)00113-0)
- [Perry E.K et al. Medicinal plants and Alzheimer's disease: from ethnobotany to phytotherapy](https://doi.org/10.1211/0022357991772808)
- [Taiwo A.E et al. Anxiolytic and antidepressant-like effects of Melissa officinalis dilute alcohol extract in mice](https://doi.org/10.1007/s00213-011-2607-0)
- [Soltanpour A et al. Effects of Melissa officinalis on anxiety and sleep quality in patients undergoing coronary artery bypass graft surgery: A double-blind randomized clinical trial](https://doi.org/10.1186/s40816-019-0134-8)
- [Ghazizadeh J et al. The effects of Melissa officinalis on depression and anxiety in clinical trials: A systematic review and meta-analysis](https://doi.org/10.1002/ptr.7252)
- [EMA/HMPC. European Union herbal monograph on Melissa officinalis L., folium; assessment and traditional-use summary](https://www.ema.europa.eu/en/medicines/herbal/melissae-folium)
- [Noguchi-Shinohara M et al. Pharmacokinetics, Safety and Tolerability of Melissa officinalis Extract which Contained Rosmarinic Acid in Healthy Individuals: A Randomized Controlled Trial](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0126422)