COMT (catechol-O-methyltransferase)

In brief

What the enzyme is, and what a catechol is

Start with the shape, because the whole subject follows from it. A catechol is a six-carbon benzene ring carrying two hydroxyl groups -- two oxygen-hydrogen pairs -- on carbons that sit next to each other. That arrangement is chemically reactive and biologically meaningful: it is the business end of dopamine, of norepinephrine and of epinephrine, the three catecholamines, and of the Parkinson's drug levodopa. COMT takes a methyl group, a single carbon with three hydrogens, from S-adenosyl-L-methionine, the molecule the body uses as its universal methyl donor, and sticks it onto one of those two hydroxyls ¹. One cap, and the catechol is no longer a catechol. For a catecholamine or for levodopa, that ends its activity ¹. Not every product is inert, though: one of them, 2-methoxyestradiol, has effects of its own, and the section on catechol estrogens below returns to it ⁴⁵.

The enzyme that does this is small and has been known in atomic detail since 1994, when the rat enzyme was crystallized and solved to 2.0 Angstrom resolution, an Angstrom being a ten-billionth of a meter. The paper's most interesting observation was not about catecholamines at all: the part of the enzyme that grips the methyl donor looked strikingly like the equivalent part of an S-adenosyl-L-methionine-dependent DNA methylase, which suggested that every enzyme using this donor might share one architecture ¹. Later work, using a fungal compound called sinefungin that mimics the moment of transfer, pinned down the mechanism further: a magnesium ion sits in the site, a lysine at position 144 takes a proton from the catechol, and the pocket tightens sideways around the methyl group as it moves, rather than the donor and acceptor sliding toward each other ². The human enzyme itself was not crystallized until 2008, when both the valine and the methionine forms were solved with the methyl donor and a substrate analog bound; that paper notes that inhibitor development had until then relied on the rat structures, that the human and rat proteins differ appreciably in which substrates they prefer, and that the bulkier methionine displaces the protein backbone by 0.7 Angstrom near the polymorphic site ⁷.

One gene makes two versions of the protein. The human gene has six exons -- the stretches of a gene kept in the finished messenger RNA -- of which the first two carry no protein code; both start codons sit in exon 3, and two separate promoters -- the stretches of DNA that switch transcription on -- drive them. The proximal promoter produces the shorter 1.3-kilobase message for the soluble enzyme and the distal promoter the longer 1.5-kilobase message for the membrane-bound one, a kilobase being a thousand letters of genetic code, and the ratio differs between tissues ⁴. The membrane-bound protein is simply the soluble one with an extra 50 amino acids tacked on the front, which anchors it. That is the entire reason the same genetic variant is called Val108Met by some authors and Val158Met by others: the residue is the same, the numbering starts 50 places earlier in the shorter protein ^4,5,6.

Where the two forms sit was worked out by expressing them in cultured cells and staining. The membrane-bound form appeared in the endoplasmic reticulum, the folded membrane system inside the cell where proteins are made, and in the nuclear membrane -- and, importantly, not at the plasma membrane, the cell's outer skin ⁸. On that evidence the authors inferred that substrates may have to be internalized before they can be methylated ⁸. Soluble COMT also stained the nucleus, and enzyme assays on rat liver cytoplasmic and nuclear fractions put roughly 5 percent of it there ⁸. That picture of the membrane-bound form facing inward was later challenged: in rat cortical neurons, membrane-bound COMT was found in cell bodies, axons and dendrites with its catalytic domain facing the extracellular space, where it could inactivate dopamine at and around the synapse, and the authors describe the orientation of the enzyme on the membrane as controversial ⁹. In the brain, messenger RNA for the enzyme turned up mainly in neurons rather than in glia, in large pyramidal and smaller neurons across every layer of the human prefrontal cortex, and at lower levels in the striatum, which receives far denser dopamine input ¹⁰. That distribution is the anatomical basis for every claim later made about COMT and the frontal lobe.

The whole history, including the reversals

The discovery is unusually clean. In August 1957 Julius Axelrod published a two-page report in Science showing that epinephrine and other catechols are O-methylated both in the test tube and in the living animal ³⁰. The following year, with LaBrosse and Kety, he established what that meant for people: O-methylation to metanephrine is the principal route by which epinephrine is metabolized in man, and the enzyme responsible is the one mainly responsible for terminating epinephrine's action ²⁹. Before this it was widely assumed that monoamine oxidase, found two decades earlier, did most of the work of ending catecholamine signaling. Axelrod's finding did not displace monoamine oxidase so much as split the job in two, and in 1970 he shared the Nobel Prize in Physiology or Medicine with Bernard Katz and Ulf von Euler for work on how nerve transmitters are stored, released and inactivated ⁴⁹. It is worth being exact about what the prize was for, because the oldest confusion in this subject, the one the last section of this page returns to, turns on it. Axelrod's share was awarded not for the methylation work but for a later discovery that partly displaced it: that noradrenaline released from a sympathetic nerve terminal is inactivated mainly by being taken back up into the terminal that released it, shown by infusing tritiated noradrenaline and recovering it in the nerve endings ⁵⁰, with the corollary that antidepressant drugs act by blocking that uptake ^51,52. So the man who established that O-methylation terminates the action of circulating epinephrine is the same man who then showed that it is not how a synapse is switched off. His 1971 Science article on the fate of noradrenaline and the control of its biosynthesis sets out the whole scheme as he then understood it ⁵³.

What followed was a slow accumulation of structure. Weinshilboum and Raymond showed in 1977 that human COMT activity measured in red cells is bimodal and heritable, with about a quarter of people in a low-activity group and family data fitting an autosomal recessive allele ¹⁴. The human complementary DNA was cloned in 1991 by two groups independently, one of which already predicted the two initiation codons and showed by cell-free translation that only the longer product associated with membranes ^54,55. In 1992 the gene was mapped to the long arm of chromosome 22, in the band 22q11.1 to q11.2, using panels of human-rodent hybrid cells ³ -- a location that mattered enormously, because it is inside the stretch deleted in 22q11.2 deletion syndrome. In 1994 the gene's exon-intron structure and its two promoters were worked out ⁴ and the rat protein was crystallized ¹. In 1996 Lachman and colleagues found the single letter responsible for the activity difference Weinshilboum had seen nineteen years earlier, and in the same year proposed that it might matter for neuropsychiatric illness ⁵.

That proposal is where the trouble starts, and it is worth being exact about the sequence. The gene sat in a deleted region associated with schizophrenia. It had a common functional variant. It metabolized dopamine, and dopamine was the leading hypothesis for schizophrenia. Three plausible facts lined up, and a very large literature was built on top of them. Then, between 2008 and 2022, three independent methods -- meta-analysis of the cognitive literature ³⁴, genome-wide association studies in tens of thousands of patients ^43,44, and formal evaluation of historical candidate genes ^35,36 -- each found that the specific claims did not hold. That is the reversal this page exists to state plainly, and it says nothing against the enzyme, which is as real and as important as it ever was.

Soluble and membrane-bound: the same enzyme doing two jobs

The distinction between the two forms is not a technicality. They differ in where they are, how tightly they grip their substrate and, as a result, what they are for. The two transcripts are present in different ratios in different tissues, which is where the division of labor begins ⁴; the membrane-bound form is the one that matters in the brain, and the soluble form is what a red-cell assay measures ¹⁴. The brain picture came from in situ hybridization, a method that labels messenger RNA where it sits in a slice of tissue, and the study was explicitly set up against an expectation: the authors wrote that contrary to expectations that COMT would be expressed predominantly in non-neuronal cells, neurons were the main cell population expressing it in prefrontal cortex and striatum ¹⁰. Combining that messenger RNA map with earlier protein work, the same authors suggested -- and were careful to call it a suggestion -- that the membrane-bound form, which has the higher affinity for dopamine, is the one sitting at neuronal dendrites in human cortex ¹⁰.

The reason this matters for the frontal lobe is an absence elsewhere. In the striatum, dopamine released into a synapse is mostly taken back up by the dopamine transporter. The prefrontal cortex has comparatively little of that transporter, so a larger share of the clearing falls to enzymatic breakdown, and that is where a three-to-four-fold difference in COMT activity could plausibly have a measurable effect while making no difference at all in the striatum. This regional argument is the load-bearing element of the whole prefrontal story ^6,60, and unlike the clinical associations built on top of it, it has not been overturned.

The second place the distinction decides something is a tumor. When pheochromocytomas were examined by Western blot -- a method that separates proteins by size and then stains for one of them, so it reports how much of a particular protein is present -- together with enzyme assay and immunohistochemistry, the predominant form was membrane-bound, not soluble, and it was located in the same chromaffin cells where catecholamines are loaded into storage vesicles ¹⁶. Free metanephrine concentrations inside the tumors were more than 10,000 times the concentrations in the same patients' plasma before surgery, and more than 93% of the elevated circulating metanephrines came from metabolism before, not after, catecholamines were released into the blood ¹⁶. That is the sentence that explains why the test works even in tumors that do not secrete: all of them appear to metabolize.

What the membrane-bound form does besides clear dopamine is a 2026 question. Unbiased proteomic and biochemical analysis of a cell line in which the membrane-bound isoform had been deleted found that its loss disrupts dopamine metabolism and also perturbs pathways governing synaptic and mitochondrial function, iron and copper handling, and the balance between oxidation and reduction; limited-proteolysis mass spectrometry, a method that reports on protein shape rather than protein amount, showed widespread structural changes of the kind usually described in neurodegeneration rather than in psychiatric illness, and the altered proteins overlapped those dysregulated in postmortem prefrontal cortex from people with schizophrenia. The authors propose the membrane-bound enzyme as a hub connecting dopamine metabolism to cellular stress responses ⁷². This is laboratory work in edited cells and postmortem tissue, not a finding about any living patient, and it does nothing to revive the Val158Met claims; it is a reason to stop treating the two isoforms as the same enzyme in two places.

Val158Met: what it is, what it does, and what it does not do

The variant is one letter. At codon 158 of the membrane-bound protein, a G becomes an A, and the amino acid valine becomes methionine ⁵. The methionine version is less thermally stable ⁵, and in human brain less of the protein is present ¹³; people carrying two methionine copies have three to four times lower total enzyme activity than those carrying two valine copies ⁵. It is common: the minor allele frequency was 0.42 in one large United States treatment study ¹¹ and 0.47 in another large cohort ¹², so roughly four to five in every ten copies of the gene carry it in those populations, and the frequency differs between ancestry groups. These facts are solid and have been confirmed by direct measurement of protein and activity in 108 postmortem human prefrontal cortex samples, where the variant changed protein abundance and enzyme activity but not messenger RNA levels, and where two other variants previously proposed as part of a schizophrenia-associated haplotype changed nothing at all ¹³. That conclusion has since been tested systematically. A review that screened 22,728 articles and included 255 eligible studies asked, for every commonly studied polymorphism in the dopamine system, whether there is real evidence that it changes the expression, abundance, activity or affinity of the gene product; for most of them the evidence was negative, inconclusive or absent, and COMT Val158Met was one of the handful that passed, with marked effects on enzyme activity, protein abundance and protein stability ⁷³. The variant is functional. That is a different question from whether it predicts anything about a person.

What happened next is a case study in how a real biochemical fact can support an unreal clinical edifice. In 2001 Egan and colleagues reported three things at once: a dose-dependent relationship between genotype and Wisconsin Card Sorting Test performance explaining 4% of the variance in perseverative errors; a more efficient prefrontal response on functional MRI in methionine carriers; and an excess transmission of the valine allele to affected offspring in 104 family trios ⁶. Each component is modest. Four percent of the variance in one score is small. The imaging was done in three subgroups of 11 to 16 people. The transmission result is a family-based association in 104 trios. Read together, with a plausible mechanism attached, the package was extraordinarily persuasive, and a companion review in the same year described it as the first plausible mechanism for a genetic effect on normal human cognition and risk for mental illness ⁶⁰.

By 2008 the cognitive half had collapsed. A meta-analysis of six cognitive phenotypes found no association with most of them, an association with IQ of d = 0.06 -- a difference so small that it is invisible in any individual -- no robust association for the n-back working memory task, evidence of publication bias, and effect sizes that shrank as publication year advanced ³⁴. The last of these is the signature of a literature correcting itself under the weight of unpublished null results.

By 2022 the schizophrenia half had collapsed too, and in a particularly complete way. The 2014 genome-wide study of up to 36,989 cases found 108 loci and did not highlight COMT ⁴³. The 2015 evaluation of 25 historical candidate genes, COMT included, concluded that the literature had not yielded clear insight, most plausibly because of inadequate statistical power ³⁵. The 2017 gene-set analysis showed the candidate genes as a group were no more associated with schizophrenia than control sets of non-candidate genes ³⁶. And the 2022 study of up to 76,755 cases and 243,649 controls mapped 287 loci and fine-mapped 120 genes concentrated in synaptic biology, without COMT among them ⁴⁴. None of this means COMT does nothing in the brain. It means the specific claim -- that this common variant is a meaningful determinant of cognition or of schizophrenia risk in the general population -- is not supported.

Warrior and worrier: where the phrase came from and what it is worth

The phrase did not come from an experiment. It is the title of a short 2006 article in CNS Spectrums by Stein, Newman, Savitz and Ramesar, which proposed that valine carriers, with higher COMT activity and lower prefrontal dopamine, may have an advantage in processing aversive stimuli -- a warrior strategy -- while methionine carriers may have an advantage in memory and attention tasks -- a worrier strategy -- with the ranking reversing under stress, when dopamine release rises ⁶³. The authors introduced the supporting evidence with the words some evidence suggests: some evidence suggests valine alleles are associated with schizophrenia, and methionine alleles with anxiety ⁶³. No earlier published use of the warrior and worrier phrasing appears in PubMed, and the article does not claim to be reporting an experiment. Related framings were in the air at the time: Goldman's 2005 review of addiction genetics places COMT among the variants affecting reward, behavioral control and the stress response ⁷⁴, and his companion pharmacogenetics paper treats Val158Met as a common functional variant with threefold relevance to executive function, stress and anxiety responses, and brain endogenous opioid function ¹¹.

Two things are worth separating. The underlying inverted-U idea -- that prefrontal function is best at an intermediate level of dopamine, so more enzyme can help or hurt depending on where you start -- is a serious pharmacological hypothesis, and it is the reason a single allele could plausibly look good on one task and bad on another. The popular claim built on it, that a person is constitutionally a warrior or a worrier according to one letter of their DNA, is not supported by the evidence, and the article that named it did not claim it was.

The empirical state as of 2026 is that the cognitive associations do not replicate at meaningful size ³⁴; that the schizophrenia association does not appear in genome-wide data ^35,36,43,44; that some sex-specific associations have held up better than others, with the Val158Met variant associated with obsessive-compulsive disorder in men and with anxiety phenotypes in women, and a larger effect on cognitive function in boys than girls, in a review that calls much of this evidence incomplete and lacking a unifying explanation ⁷⁵; and that direct-to-consumer panels continue to sell the warrior-worrier reading. A 2026 systematic review of dopaminergic gene polymorphisms and personality in athletes is instructive on how this literature behaves: fourteen studies, mostly from one research network with overlapping cohorts, no gene-by-trait association replicated in an independent sample, and certainty of evidence rated weak for every association examined ⁷⁶.

One line of evidence has held up better than the rest, and this page has to say so. A meta-analysis of the available neuroimaging studies of rs4680 found a significant association between genotype and prefrontal activation with a large effect size (d = 0.73) and, unlike the behavioral literature, no evidence of publication bias; the effects were strong and in opposite directions for executive-cognition paradigms, which favored methionine carriers, and emotional paradigms, which favored valine carriers ⁷⁷. That is the warrior and worrier pleiotropy itself, measured at the level of brain activation rather than of behavior, and the authors present it as a case for using such intermediate measures to bridge genes and behavior. It is also an intermediate measure and not a clinical outcome: it says the genotype changes how the prefrontal cortex works a task, not that it makes anyone a warrior or a worrier; the imaging studies pooled in it were individually small, and it was published in 2009, before the behavioral literature was corrected; and it does nothing to rescue the schizophrenia or the cognitive-performance claims ^34,43,44.

22q11.2 deletion syndrome: what losing the gene did and did not teach

Chromosome 22q11.2 deletion syndrome -- previously called velo-cardio-facial syndrome and DiGeorge syndrome -- is caused by the loss of a stretch of one copy of chromosome 22 containing dozens of genes. It produces a characteristic facial appearance, cleft palate, heart defects and learning difficulties ⁵⁹, and it carries an extraordinary psychiatric burden. In the largest collaborative series, 1,402 people aged 6 to 68, attention-deficit hyperactivity disorder was the commonest diagnosis in children at 37.10%, anxiety disorders outnumbered mood disorders at every age, and psychotic disorders were present in 41% of adults over 25 ³⁷. That makes the deletion one of the strongest known genetic risk factors for psychosis.

COMT sits inside the deleted region ³, so people with the deletion have one working copy instead of two, and whichever allele sits on the remaining chromosome determines their enzyme activity outright -- there is no second copy to average with. This is as close to a natural experiment as human genetics offers, and in 1996 Lachman and colleagues reported that the low-activity allele was associated with bipolar spectrum disorder, particularly a rapid-cycling form, in patients with the syndrome ⁵⁹.

The natural experiment was then run properly, and the result was negative. Murphy, Jones and Owen assessed 50 adults with the syndrome using structured clinical interviews, confirmed the high psychosis rate, and genotyped all 48 individuals deleted for the relevant probe. They found no evidence that possession of the low-activity allele was associated with schizophrenia in their sample ³⁸.

It was then run again, larger, with the same answer. Eighty-three adults with the deletion were compared with 90 people who had schizophrenia without the deletion and 316 controls, assessed with the Schedules for Clinical Assessment in Neuropsychiatry and two schizotypy questionnaires, and genotyped for several COMT polymorphisms including the Ashkenazi risk haplotype. Psychotic disorder was present in 29% and schizophrenia in 22%, confirming the rate; the symptom profile did not differ from schizophrenia without the deletion; and COMT was associated with neither schizophrenia status nor schizotypy score ³⁹.

A later longitudinal study did find a signal for the low-activity allele in a different outcome: in adolescents with the deletion, the low-activity allele was a risk factor for decline in prefrontal cortical volume and cognition and for the development of psychotic symptoms ⁴⁰. A 2024 study of 30 children and adolescents with the deletion then found the opposite of the expected direction for anxiety -- individuals carrying the higher-activity valine allele had higher anxiety levels (p = 0.021), which the authors flag as contrary to expectation -- and found higher salivary alpha-amylase, a marker of sympathetic activation, in the low-activity group (p = 0.026, uncorrected) ⁴¹.

Five studies, no consistent direction. The honest summary is that hemizygosity for COMT has not been shown to account for the psychiatric phenotype of 22q11.2 deletion syndrome, that the deletion removes many other genes with plausible neural roles, and that the samples that have tested the question directly are in the tens, not the thousands. What the syndrome did teach is something broader and more durable: that a rare, large deletion can raise psychosis risk enormously, which is now a central fact of psychiatric genetics, while the common variant in one of its genes does not ^37,43,44.

COMT and pain: the strongest soft finding in the field

The pain literature is where COMT has done best, and it still needs careful handling. In 2003 Zubieta and colleagues scanned volunteers with positron emission tomography during sustained pain induced by injecting saline into a jaw muscle. People homozygous for the methionine allele showed diminished regional mu-opioid system responses -- that is, less engagement of the brain's own opioid machinery -- along with higher sensory and affective pain ratings and a more negative internal affective state; valine homozygotes showed the opposite ⁶¹. The proposed chain is that chronically higher synaptic dopamine in methionine carriers downregulates the endogenous opioid system, leaving less in reserve when pain arrives.

Two years later Diatchenko and colleagues reported the haplotype work. Three COMT haplotypes, labeled low, average and high pain sensitivity, covered 96% of the population studied; five combinations of them were associated with variation in experimental pain sensitivity at p = 0.0004; and carrying even a single low-pain-sensitivity haplotype reduced the risk of myogenous temporomandibular joint disorder -- jaw muscle pain -- by as much as 2.3 times. The low-pain-sensitivity haplotype produced much higher enzyme activity, and inhibiting COMT in rats produced a profound increase in pain sensitivity ⁶². The animal half of that result is an experiment with a cause; the human half is an association.

The mechanism turned out to be unexpected and is a genuine contribution to molecular biology regardless of what one thinks of the clinical claims. The haplotypes differ at two synonymous positions -- changes that do not alter the protein -- and one that does. The haplotypes differing only in synonymous positions showed the largest difference in enzyme activity, because the messenger RNA folds into local stem-loop structures of different stability; the most stable structure was translated least, and mutating it away restored protein output ⁶⁴. Before this, synonymous changes were widely treated as inert.

The replication record is mixed, and should be read exactly as it stands. The OPPERA case-control study, which typed 3,295 variants across 358 pain genes in 166 chronic temporomandibular disorder cases and 1,442 controls plus a further 182 cases and 170 controls, supported previously reported associations for COMT and the serotonin receptor gene HTR2A, with the authors explicitly stating the findings need replication ⁶⁶. A systematic review and meta-analysis across three chronic pain conditions found an association only for fibromyalgia or chronic widespread pain, with the low-activity methionine allele as the risk allele, and no association with migrainous headache or chronic musculoskeletal pain ⁶⁷. Then the unbiased test: a genome-wide association study of multisite chronic pain in about 380,000 UK Biobank participants found 76 lead variants at 39 loci, enriched in neurogenesis and synaptic plasticity genes, and COMT was not among them ⁶⁸. A 2025 systematic review of 63 studies on temporomandibular disorders, seven of which entered a network meta-analysis, found its clearest COMT signal not at rs4680 but at a different COMT variant, rs165774: in pairwise meta-analysis two genotypes were protective for myalgia (odds ratios 0.33, 95% confidence interval 0.14-0.76, and 0.32, 0.14-0.74, both p < 0.01), and that variant carried the highest probability of association with myalgia (97%) and with arthralgia (93%) ⁷⁸.

Two further results belong in that record, one because it is the largest negative test and one because it is randomized. In 2,294 adult European patients taking a World Health Organization step III opioid for cancer pain, 112 single-nucleotide polymorphisms in 25 candidate genes -- COMT among them, alongside the opioid receptor, serotonin receptor and transporter genes -- were tested against opioid dose, with the patients split at random into a development and a validation sample. Not one variant was significantly associated with dose in both samples, and the authors offered the study as a demonstration of why genetic associations must be validated before being reported as findings ⁷⁹.

The randomized result is the propranolol story, and it is the one place in this literature where a COMT genotype changed the effect of a drug in a trial. A crossover pilot in 40 women with temporomandibular disorder suggested that benefit from the beta-blocker propranolol was confined to those without the high-activity COMT haplotype, absent in high-activity homozygotes and intermediate in heterozygotes ⁸⁰. The full trial then randomized 200 patients with temporomandibular disorder myalgia to extended-release propranolol 60 mg twice daily or placebo; it missed its primary endpoint, the mean change in a facial pain index at week 9 (adjusted difference -1.8, 95% confidence limits -6.2 to 2.6, p = 0.41), but more patients reached a 30% reduction on the drug than on placebo (69.0% against 52.6%, number needed to treat 6.1, meaning about six patients had to be treated for one extra responder, p = 0.03) ⁸¹. The prespecified genotype analysis of 143 of those patients then found the predicted interaction pointing the wrong way: for a 30% or greater reduction the odds ratio favoring propranolol was 10.9 (95% confidence limits 2.4 to 50.7) in G:G homozygotes and 0.8 (0.2 to 3.2) in A:A homozygotes, interaction p = 0.035, and the authors state that the antagonistic effect of the A allele was the opposite of the synergistic effect they had hypothesized in advance ⁸². A randomized genotype-by-drug interaction that comes out backwards is still a real interaction, and it is also one more reversal.

Smaller recent studies continue in both directions. In 60 women with fibromyalgia and 60 matched controls, the AA genotype was associated with higher pain severity scores (p = 0.013 and 0.019 on two scales) and higher neuropathic pain scores ⁸³. In 150 patients with acute renal colic and 100 controls, no difference in rs4680 genotype or allele frequency was found between patients and controls, and the genotype-phenotype signal that did emerge was at a drug-transporter gene rather than COMT ⁸⁴. The fair summary in 2026: COMT genotype makes a real, small, condition-specific contribution to pain sensitivity that is best supported in temporomandibular disorder and fibromyalgia, is not supported as a general determinant of chronic pain, and is not strong enough to guide the treatment of an individual patient.

• Plasma free metanephrinesThe methylated products of epinephrine and norepinephrine, which exist because COMT made them; measured in blood to find or exclude a pheochromocytoma ^15,16.One of the two first-line biochemical tests for pheochromocytoma and paraganglioma, alongside urinary fractionated metanephrines, in the Endocrine Society guideline ⁸⁵In 214 patients with the tumor and 644 without, sensitivity was 99% (95% confidence interval 96-100%) and specificity 89% (87-92%); no combination of other tests beat it ¹⁵.Posture and food wreck the specificity. Sampling seated and non-fasting raised normetanephrine and methoxytyramine by 49-51% in people without the tumor and produced a 5.7-fold increase in false positives against cutoffs built from supine fasting samples ⁸⁶.
• Red-cell COMT activityThe enzyme assayed directly in erythrocytes, the measurement that first showed human COMT activity is inherited ¹⁴.The original phenotype assay, now largely replaced by genotypingIn 373 randomly selected 16-to-18-year-olds the distribution was bimodal with a trough at about 8 units, 23% falling below it, and segregation in 48 families fitted an autosomal recessive low-activity allele ¹⁴.It measures the soluble form in a red cell, which is not the membrane-bound form in a cortical dendrite; the correspondence had to be established separately in brain tissue ^10,13.
• COMT genotyping (rs4680)A single-letter test for the valine-to-methionine change, originally read with a restriction enzyme after polymerase chain reaction ⁵.Research; sold commercially in direct-to-consumer and wellness panels, where its claims far outrun the evidenceThe genotype does predict enzyme activity: in 108 postmortem prefrontal cortex samples it significantly affected protein abundance and enzyme activity, though not messenger RNA level ¹³.Predicting a person from the genotype is where it fails. Meta-analysis found essentially no association with memory or executive function and evidence of publication bias ³⁴, and genome-wide studies did not support it as a schizophrenia gene ^35,36,44.
• Enzyme activity in tissue, with Western blotMeasuring how much protein is present and how fast it methylates a catechol, in the tissue of interest rather than in blood ^13,16. A Western blot separates proteins by size and then stains for one of them, so it reports how much of a particular protein a tissue contains.The method that settled which COMT form a pheochromocytoma uses, and what the Val158Met variant actually changesIt showed that pheochromocytomas contain predominantly the membrane-bound and not the soluble form, and that free metanephrine concentrations inside tumors were more than 10,000 times those in the same patients' plasma ¹⁶.It requires tissue, so it is a research and pathology method, not a clinical test on a living patient.

Where the catecholamines go, and why metanephrines are the pheochromocytoma test

The clinical payoff of Axelrod's chemistry is a blood test, and the logic behind it is worth following because it is counterintuitive. Most catecholamine metabolism does not happen after release. It happens inside the cells that make the catecholamines, where amine continuously leaks out of its storage vesicles into the cell fluid and is deaminated by monoamine oxidase; O-methylation outside the neuron to form metanephrines is, by volume, a minor route ⁴⁶. The same review corrects several textbook errors in passing: in sympathetic nerves the aldehyde made from norepinephrine becomes 3,4-dihydroxyphenylglycol, not 3,4-dihydroxymandelic acid, and vanillylmandelic acid is made in the liver rather than in the nerve ⁴⁶.

Although O-methylation is a minor route overall, the single largest source of metanephrines in the body is the adrenal medulla ⁴⁶, where chromaffin cells carry membrane-bound COMT alongside the vesicles they load with catecholamines ¹⁶. A pheochromocytoma is a tumor of those cells, and it behaves the same way, only more so. Measured directly, free metanephrine concentrations inside tumors were over 10,000 times the levels in the same patients' plasma before the tumors were removed, and more than 93% of the raised circulating metanephrines in these patients came from metabolism occurring before catecholamines ever reached the bloodstream ¹⁶. The practical consequence is decisive: some pheochromocytomas do not secrete catecholamines, but all of them appear to metabolize, so a metabolite test catches tumors that a hormone test misses ¹⁶.

That prediction was tested in 858 patients across four referral centers between 1994 and 2001, 214 with a confirmed pheochromocytoma and 644 without. Plasma free metanephrines were 99% sensitive (95% confidence interval 96-100%) and 89% specific (87-92%). Urinary fractionated metanephrines were 97% sensitive (92-99%) but only 69% specific (64-72%). Plasma catecholamines were 84% sensitive (78-89%), urinary catecholamines 86% (80-91%), urinary total metanephrines 77% (68-85%) and urinary vanillylmandelic acid, long the standard test, only 64% sensitive (55-71%) though 95% specific (93-97%). Combining tests did not beat plasma free metanephrines alone ¹⁵. The Endocrine Society's 2014 guideline recommends initial testing with plasma free or urinary fractionated metanephrines, and specifically warns that pre-analytical factors cause both false positives and false negatives and that all positive results require follow-up ⁸⁵.

The pre-analytical warning is not boilerplate. In 762 patients tested at six centers, two of which sampled supine after an overnight fast and four of which did not, plasma normetanephrine and methoxytyramine were 49-51% higher in people without a tumor when sampled seated and non-fasting. The upper reference limits differed accordingly: normetanephrine 1.29 against 0.79 nanomoles per liter, metanephrine 0.49 against 0.41, methoxytyramine 0.42 against 0.18. Using cutoffs derived from seated non-fasting sampling dropped sensitivity from 98% to 85%; using supine fasting cutoffs on seated samples dropped specificity from 95% to 71%, a 5.7-fold increase in false positives ⁸⁶. A test with 99% sensitivity is only 99% sensitive when the blood is drawn the way the reference range was built.

The three COMT inhibitors, and the deaths that shaped the class

Levodopa is a catechol, so COMT methylates it, and much of an oral dose is lost that way before it reaches the brain. Blocking the enzyme therefore increases how much levodopa gets through and extends how long a dose lasts, which in Parkinson's disease means fewer hours of the immobile off state. The pharmacology was reviewed in detail in 1999, when the new selective inhibitors were reaching the clinic ²⁸.

Tolcapone was first in the clinic, though not quite first in the randomized literature: a one-month double-blind crossover study of entacapone was published in January 1996 ¹⁷, and a multicenter placebo-controlled tolcapone trial in 151 completers, in which off time fell about 40% and on time rose about 25% at all three doses, appeared in January 1997 ¹⁸. Three larger placebo-controlled trials of tolcapone followed later the same year, and they are the ones usually cited. In 177 fluctuating patients, tolcapone 200 mg three times daily reduced off time by 26.2% of baseline and increased on time by 20.6% (p < 0.01 against placebo), while allowing a 122 mg reduction in the mean daily levodopa dose from a baseline of 676 mg (p < 0.01); diarrhea was the commonest non-dopaminergic adverse event and the commonest reason for withdrawal ¹⁹. In 202 patients, the same dose cut daily off time by 3.25 hours from baseline, significantly more than placebo (p < 0.01), but dyskinesia -- involuntary writhing movements -- developed or worsened in 64% of that group against 18% on placebo ²⁰. In 298 patients who had not yet developed fluctuations, both doses improved activities of daily living and motor scores at six months, maintained at twelve ²¹.

Then the liver. A 1998 Lancet letter reported fulminant hepatitis on tolcapone ³¹. In clinical trials, liver chemistry tests had exceeded three times the upper limit of normal in about 1% of patients on 100 mg and about 3% on 200 mg; post-marketing surveillance recorded three instances of acute liver failure with death after some 60,000 patients had taken the drug for about 40,000 patient-years, and the drug was withdrawn in Europe and Canada and given a boxed warning in the United States ³². A panel reporting in 2000 reviewed four cases of severe hepatic dysfunction and reported that liver dysfunction had not developed in any patient monitored according to the original prescribing information, and that virtually all enzyme abnormalities appeared within six months of starting ⁸⁷. By 2007, Olanow and Watkins counted four cases of acute hepatotoxicity with three fatalities, and reported that in the more than 40,000 patient-years accumulated since the 1998 labeling restrictions there had been only three reports of severe but reversible liver injury and no hepatic fatalities; on that basis the drug was reintroduced in several European countries, and in the United States the Food and Drug Administration relaxed its monitoring requirements: the new labeling recommends liver function tests every two to four weeks for six months and at the physician's discretion thereafter, and stopping the drug if enzymes exceed twice the upper limit of normal ³³. A 2006 review records the same reintroduction and notes that tolcapone, unlike the others, has central as well as peripheral effects, and that in a switch study it gave greater on time than remaining on entacapone ⁸⁸. The exact dates of the withdrawal, the boxed warning and the reinstatement belong to regulators' decision documents, which are not indexed in PubMed.

Entacapone was the safer sibling: no increase in liver enzyme elevation over placebo and no attributed liver failure ³². Its 1997 trial randomized 205 patients to 200 mg with every levodopa dose or placebo for 24 weeks; patients averaged about 10 hours of on time per day at baseline (60.5% of waking hours) and entacapone added 5.0 percentage points of on time, with the effect larger in those who started with less ²². Then came the reversal. STRIDE-PD tested the appealing theory that giving entacapone from the start, by smoothing levodopa delivery, would delay the dyskinesia that long-term levodopa causes. It randomized 747 patients to levodopa-carbidopa or levodopa-carbidopa-entacapone four times daily at 3.5-hour intervals for 134 weeks. The entacapone group reached dyskinesia sooner (hazard ratio 1.29, p = 0.04) and had it more often at week 134 (42% against 32%, p = 0.02), with the effect more pronounced in those already on a dopamine agonist ⁶⁵. The authors note that the regimen did not actually achieve continuous levodopa availability and that the entacapone group received higher levodopa dose equivalents, which may explain it. Either way, the prediction failed in the direction opposite to the hypothesis.

Opicapone is the third generation: once daily, more potent, and designed around the liver problem. BIPARK-I randomized 600 patients across 106 centers in 19 European countries and Russia to opicapone 5, 25 or 50 mg, placebo, or entacapone 200 mg with every levodopa dose. Mean change in off time was -56.0 minutes (standard error 13.4) on placebo, -96.3 on entacapone, -91.3 on opicapone 5 mg, -85.9 on 25 mg and -116.8 on 50 mg. Only the 50 mg dose beat placebo (difference -60.8 minutes, 95% confidence interval -97.2 to -24.4, p = 0.0015) and it was non-inferior to entacapone (-26.2 minutes, -63.8 to 11.4, p = 0.0051); 5 mg (p = 0.056) and 25 mg (p = 0.080) did not. The trial was funded by the manufacturer ²³. BIPARK-II randomized 427 patients (mean age 63.1, standard deviation 8.8) to 25 mg, 50 mg or placebo: least-squares mean change in off time was -64.5 minutes (standard error 14.4) on placebo, -101.7 (14.9) on 25 mg and -118.8 (13.8) on 50 mg; only 50 mg was significant against placebo (treatment effect -54.3 minutes, 95% confidence interval -96.2 to -12.4, p = .008), with the 25 mg dose at p = .11, and the effect was maintained through a one-year open-label phase ²⁴. The European Commission granted the marketing authorization in June 2016 ⁸⁹ and the United States Food and Drug Administration approved the drug, in combination with levodopa-carbidopa for wearing off, on 24 April 2020 ⁹⁰; both dates are taken here from peer-reviewed reviews rather than from the agencies' own decision documents, which were not opened for this page.

The recent record is mixed in the usual way. A one-year open-label extension of the EPSILON trial, in patients who had not yet developed motor fluctuations, reported a sustained motor improvement and a numerically but not statistically higher proportion free of motor complications (80.2% against 69.7%, p = 0.1) ⁷¹. A pooled post hoc analysis of 227 patients whose fluctuations had begun within the previous two years found that starting opicapone early reduced off time by 65.6 minutes against placebo (95% confidence interval -105.5 to -25.6, p = 0.0014) and increased good on time by 88.3 minutes (47.0 to 129.6, p < 0.0001), with the long-term advantage over late starters numerical rather than significant (p = 0.2 and 0.09) ⁹¹. And the OCEAN trial, which asked whether opicapone relieves the pain that comes with off periods, missed its primary endpoint outright: change in domain 3 of the King's Parkinson's Disease Pain Scale was -8.8 points (standard deviation 6.9) on opicapone against -9.3 (6.2) on placebo, p = 0.7940 ⁷⁰. A real-world retrospective study of 101 patients aged 65 and over, followed for a mean of 20.6 months, found adverse events in 48.5% and discontinuation in 28.7%, mostly for neuropsychiatric and cardiovascular or autonomic reasons; for comparison, treatment-emergent adverse events were reported in 54% of patients on opicapone 50 mg and 50% on placebo in BIPARK-I ²³, and the authors of the real-world study concluded that the drug was generally well tolerated in this elderly cohort ⁹². Drug development continues: a screen of natural flavonoids published online in December 2025 identified baicalein as a lead and optimized it into a non-nitrocatechol inhibitor with a binding constant of 89.28 nanomolar, 26-fold better cellular inhibition than the parent compound and improved oral availability, tested in rats ⁹³. That is a preclinical result in animals, not a drug.

Catechol estrogens, 2-methoxyestradiol, and the breast cancer associations at their true strength

Estrogens are not catechols, but the body makes them into catechols. Hydroxylating estradiol or estrone at the 2 or the 4 position produces a catechol estrogen, and COMT caps those exactly as it caps dopamine ²⁵. The two positions have opposite reputations. Capping the 2-hydroxy catechol gives 2-methoxyestradiol, which in 1994 was reported in Nature to be a potent inhibitor of endothelial cell proliferation and migration and of angiogenesis -- the growth of new blood vessels -- in culture, and to suppress the growth of solid tumors in mice when given by mouth, without the heparin co-administration that other steroid angiogenesis inhibitors required ⁴⁵. Failing to cap the 4-hydroxy catechol leaves it available for oxidation to a quinone that reacts with DNA to form depurinating adducts, leaving apurinic sites and hence mutations, which is the proposed chemistry of estrogen-initiated cancer ⁴⁷.

Both lines predicted that low COMT activity should raise breast cancer risk, and the prediction was tested repeatedly. The first study, in 1997, was a nested case-control analysis of 112 matched pairs. The odds ratios were 1.30 (confidence interval 0.66-2.58) for one low-activity allele and 1.45 (0.69-3.07) for two -- neither significant. In postmenopausal women with two low-activity alleles the odds ratio was 2.18 (0.93-5.11), again not significant; it became significant only in the subgroup with a body mass index above 24.47 kg/m2, where it was 3.58 (1.07-11.98) ²⁵. A subgroup of a subgroup of 112 pairs, with a confidence interval spanning an eleven-fold range, is a hypothesis, not a finding.

Thirteen years and tens of thousands of women later, the answer is no. A meta-analysis of 26 studies covering 16,693 cases and 18,261 controls found no significant association in any genetic model in the overall, Asian or European populations; after excluding studies whose control genotype frequencies violated Hardy-Weinberg equilibrium, a statistical check on whether genotype counts behave as a population should, a borderline decreased risk appeared in Europeans (recessive model odds ratio 0.90, with the 95% confidence interval printed in the abstract as 0.90-1.00, a lower bound equal to the point estimate that is almost certainly a typographical error) ⁹⁴. A larger meta-analysis of 30,199 cases and 38,922 controls found no association in the dominant model (odds ratio 0.99, 0.94-1.04), the recessive model (0.97, 0.92-1.02) or the additive model (0.98, 0.91-1.05), with substantial heterogeneity between studies; in Caucasians specifically the recessive model gave 0.96 (0.92-1.00), a decrease, which is the opposite direction to the original hypothesis ⁹⁵. A meta-analysis restricted to Chinese populations, 18 studies with 5,034 cases and 6,234 controls, did find increased overall cancer risk (odds ratios 1.34, 1.04-1.73; 1.39, 1.06-1.82; and 1.13, 1.01-1.27 in different models), with the signal in breast cancer and esophageal squamous cell carcinoma ⁹⁶ -- a population-specific result from a smaller literature, which should not be generalized. For mammographic density, a known breast cancer risk factor, a meta-analysis of eight studies found an effect of -1.41 percentage points in the recessive model that did not reach significance (confidence interval -2.86 to 0.05, p = 0.06), becoming -1.93 (-3.49 to -0.37, p = 0.02) when breast cancer patients were excluded ⁹⁷. For ovarian cancer, a 2025 systematic review and meta-analysis of 30 studies found the A allele of rs4680 associated with reduced risk (odds ratio 0.85, 0.73-0.99, p = 0.0373), and noted that the limited number of studies and population heterogeneity may affect robustness ⁹⁸.

The most interesting modern result is not an association study at all. In a pharmacogenetic analysis of two randomized trials of alpha-tocopherol -- vitamin E -- supplementation, pooling the Women's Genome Health Study (23,294 participants) and a case-control subset of the Alpha-Tocopherol Beta-Carotene trial, the supplement's effect on total cancer differed by genotype: hazard ratio 0.88 (0.80-0.97, p = .01) in methionine homozygotes, 0.99 (0.92-1.06) in heterozygotes and 1.18 (1.06-1.31, p = .002) in valine homozygotes, with a significant interaction (p < .001) ⁹⁹. That is to say, in the 22.8% of participants who were valine homozygotes, a common over-the-counter supplement was associated with more cancer, not less. The finding is from two randomized trials rather than observational cohorts, which makes it stronger than the association literature around it; it is also a subgroup analysis of trials not designed to ask this, and no third trial replicating it was found.

As for the methylated product itself: in the one human trial cited here, 2-methoxyestradiol did not work. A phase 2 study in metastatic renal cell carcinoma that had progressed on sunitinib enrolled 17 patients, of whom 12 were evaluable, and produced no objective responses; the study was terminated because the drug was not tolerable at the recommended phase 2 dose, with 35% of patients discontinuing for toxicity ⁴⁸. A clean mechanism in cell culture and mice ⁴⁵ did not become a medicine in that trial ⁴⁸; it is one small single-arm study, not the whole development program. A 2025 study of 24 women using a combined oral contraceptive and 25 controls measured the pathway directly by mass spectrometry: contraceptive users excreted less free estradiol and estrone and more 2-methoxyestrone, with higher methylated-to-parent ratios, and -- against the genotoxic prediction -- no increase in estrogen-DNA adducts; the authors also found lower serum betaine and a correlation between methyl-donor availability and methoxyestrogen levels, suggesting the methylation step is limited by how much methyl group is available rather than by enzyme alone ¹⁰⁰.

The sex difference, and estrogen's grip on the gene

COMT activity differs between men and women ⁷⁵, and the difference was settled enough by 1999 to be cited as the background that prompted the work on how the gene is regulated ¹⁰¹. Direct measurement of enzyme activity in 108 postmortem human prefrontal cortex samples found that the subject's sex had a small effect on activity ¹³. In mice, deleting the gene produced sexually dimorphic effects: region-specific dopamine changes, most notably in frontal cortex; homozygous females but not males impaired in a light-dark test of anxiety-like behavior; and heterozygous males, not females, showing increased aggression ⁴².

A plausible mechanism exists and was demonstrated in cells. Physiological concentrations of 17-beta-estradiol, from a billionth to a ten-millionth of a mole per liter, decreased the 1.3-kilobase COMT messenger RNA in MCF-7 breast cancer cells in a time- and dose-dependent way, through an estrogen-receptor-dependent route. Cloning the two promoters revealed several half-palindromic estrogen response elements and CCAAT/enhancer binding protein sites, and deletion analysis localized the effect to a 280-base-pair fragment of the proximal promoter -- a base pair being one rung of the DNA ladder -- carrying two putative response elements and a 323-base-pair fragment of the distal promoter carrying two putative binding protein sites ¹⁰¹. In cultured breast cancer cells, estrogen turns the gene down. Whether it does so enough in a woman to lower her enzyme activity and slow the clearance of her catecholamines or catechol estrogens was not tested: that study measured messenger RNA and promoter activity in cell lines and reporter constructs, not enzyme activity or catecholamines in people ¹⁰¹.

The temptation is to run that chain straight into clinical conclusions, and the most careful review of the subject declines to. Harrison and Tunbridge survey the evidence for sexually dimorphic effects of COMT and conclude that although many reported differences are unconfirmed or minor, some appear reasonably robust: the Val158Met variant is associated with obsessive-compulsive disorder in men, with anxiety phenotypes in women, and has a greater impact on cognitive function in boys than girls. They note that sex-specific effects are usually attributed to transcriptional regulation by estrogens, but state plainly that additional mechanisms are likely to be at least as important, and that the evidence is incomplete and lacks a unifying explanation ⁷⁵. That is where the subject still sits. A real sex difference in a real enzyme, a demonstrated transcriptional mechanism in cultured cells ¹⁰¹, and clinical associations that are not strong enough to carry the weight placed on them. The obsessive-compulsive half of that claim has since met the same test as the schizophrenia claim. A genome-wide association meta-analysis of 53,660 people with obsessive-compulsive disorder and 2,044,417 controls found 30 independent genome-wide significant loci and prioritized 25 genes as most likely causal, with roughly 11,500 variants accounting for 90% of the disorder's genetic heritability and risk concentrated in excitatory neurons of the hippocampus and cortex and in dopamine-receptor-bearing medium spiny neurons; COMT is not among the genes it prioritizes ¹⁰². The dopaminergic biology survives in that picture. The single common variant does not.

Green tea and COMT genotype: one positive interaction, and the larger studies that did not find it

Tea polyphenols are catechols, so COMT methylates them too, and they inhibit the enzyme while it does. The enzymology was worked out in 2003 for epigallocatechin gallate (EGCG) and epigallocatechin (EGC), the main polyphenols in green tea. With human liver cytosol, the Michaelis constant -- the substrate concentration at which the enzyme runs at half speed -- was 4.0 micromolar for EGC against 0.16 micromolar for EGCG, with maximum velocities of 1.28 against 0.16 nanomoles per milligram per minute; rat liver had higher activity than human or mouse, which matters when reading rodent studies of tea. EGCG and its two methylated products were all potent inhibitors of COMT with half-maximal inhibitory concentrations around 0.2 micromolar ²⁶.

A companion study measured inhibition of catechol estrogen methylation specifically, in human liver cytosol. For capping 2-hydroxyestradiol, the concentration producing half-maximal inhibition was 14-17 micromolar for catechin, 44-65 for epicatechin and 0.04-0.07 for EGCG; for 4-hydroxyestradiol it was 5-7, 10-18 and 0.2-0.5 micromolar respectively. Quercetin and fisetin, two common bioflavonoids, fell in between. Crude extracts of green and black tea both showed very strong inhibition, and the kinetics were mixed competitive and non-competitive ²⁷. These are in vitro experiments in human liver preparations. They show the chemistry is real; they do not show what happens in a person who drinks tea, and the authors say so, writing that more studies are warranted to determine the extent of such inhibition in human subjects.

The first study to connect this chemistry to an outcome was a case-control study of 589 incident breast cancer cases and 563 population-based controls among Chinese-, Japanese- and Filipino-American women in Los Angeles County. Risk fell with tea intake, particularly green tea -- but only in women carrying at least one low-activity COMT allele, where tea drinkers had an adjusted odds ratio of 0.48 (95% confidence interval 0.29-0.77) against non-drinkers. Among women homozygous for the high-activity allele, the odds ratio was 1.02 (0.66-1.60), that is, nothing. The authors' interpretation is that slow methylators clear tea catechins less efficiently and so are exposed to more of them ¹⁰³. It is a single case-control study with a biologically coherent interaction, which is a reason to take it seriously and not a reason to act on it.

It has not held up as a general finding. A Shanghai case-control study six times larger, 3,454 cases and 3,474 controls, found a modest inverse association with green tea (odds ratio 0.88, 95% confidence interval 0.79-0.98) that was not modified by rs4680 genotype ¹⁰⁴, and a Japanese case-control study of 369 matched pairs found no inverse association at all and no effect modification by COMT (odds ratio 1.27, 0.75-2.14, for the highest against the lowest intake) ¹⁰⁵.

And the trial has been done. The Minnesota Green Tea Trial randomized 1,075 healthy postmenopausal women with dense breast tissue to four decaffeinated green tea extract capsules a day -- 1,315 mg of total catechins, of which 843 plus or minus 44 mg was epigallocatechin gallate -- or to placebo for twelve months, with randomization deliberately stratified by COMT genotype activity, high against low or intermediate, so that the interaction could be tested directly; 937 women completed it ¹⁰⁶. Twelve months of that dose did not significantly change percent or absolute mammographic density overall, and did not change it within any COMT genotype subgroup, although percent density fell by 4.40 percentage points against placebo in women aged 50 to 55 (P = 0.05) ¹⁰⁷. Genotype did not modify the effect on plasma F2-isoprostanes, a marker of oxidative stress (P = 0.85) ¹⁰⁸, nor on C-reactive protein, interleukin-6 or tumor necrosis factor alpha in a 97-woman subset (all interaction P values above 0.40) ¹⁰⁹. The one place genotype appeared to matter was harm rather than benefit, and only between the two variant genotypes: from baseline to month 3 the mean change in the aspartate-to-alanine aminotransferase ratio was plus 1.0% in rs4680 heterozygotes against minus 4.8% in methionine homozygotes (P = 0.03), a difference the authors themselves call clinically negligible, while a variant in a different gene, UGT1A rs6755571, went with far larger alanine aminotransferase rises at months 6 and 9 (78.1% and 82.1% against 28.0% and 30.1%, P < 0.001 and P = 0.004) ¹¹⁰. So the honest statement in 2026 is that the chemistry is real in human liver preparations, that one case-control study found the interaction, and that the large randomized trial designed to find it did not.

Cannabinoid hyperemesis syndrome: what a forty-person candidate-gene study can and cannot show

Cannabinoid hyperemesis syndrome is a real and well-characterized clinical picture: cyclical nausea, vomiting and abdominal pain in a heavy, long-term cannabis user, with the peculiar feature that hot showers or baths relieve it, and resolution on stopping cannabis ^69,111. It is frequently misdiagnosed as cyclic vomiting syndrome ⁶⁹; the cannabinoid hyperemesis researchers regard the two as clearly distinct in presentation and pathophysiology ¹¹¹, while a review of cyclic vomiting research treats cannabinoid hyperemesis as a related condition ¹¹². In 2022 a group led by Ethan Russo published what remains the largest genomic investigation of it, and it is worth setting out exactly what was done and what it supports.

The design was an online questionnaire distributed through cannabis-related mailing lists, websites and direct solicitation. Of 585 respondents, 205 qualified as the patient pool and 54 as controls -- current cannabis users without symptoms. Genomic test kits were offered; 99 patients agreed to receive them, and 28 returned them, along with 12 controls. So the genetic analysis is 28 patients against 12 controls, forty people ⁶⁹. Genotyping used a commercial high-throughput array, and variants were selected for analysis by looking for common variants among patients using proprietary analysis software, then tested with Fisher's exact test ⁶⁹.

The strongest reported finding is in COMT: an intronic variant, rs4646316, heterozygous, present in 57.1% of patients against 10% of controls, odds ratio 12, p = 0.012. The confidence limits are given as 1.3 to 88.1 in the abstract and 1.3 to 98.1 in the table, a discrepancy within the paper itself ⁶⁹. Four other variants were reported as significant: a downstream TRPV1 variant (odds ratio 5.8, 1.2-28.4, p = 0.015), an intronic CYP2C9 variant (7.8, 1.1-70.1, p = 0.043), an intronic DRD2 variant (6.2, 1.1-34.7, p = 0.031) and a synonymous ABCA1 variant (8.4, 1.5-48.1, p = 0.012). A second COMT variant, rs165656, did not reach significance (4.6, 0.8-25.7, p = 0.069) ⁶⁹. Notably, the variant in the cannabinoid receptor gene CNR1, which the authors' own stated hypothesis predicted and which they report is associated with cyclic vomiting syndrome, was not significant ⁶⁹.

Now the limits, stated as plainly as the findings. With 12 controls, a control frequency of 10% does not correspond to any whole number of people -- one of twelve is 8.3%, two is 16.7% -- so the reported percentages and the odds ratios computed from them should be read as approximate. Every confidence interval is enormous: an odds ratio of 12 with a lower bound of 1.3 is compatible with a trivial effect. Multiple variants were examined and the significant ones reported, without a stated correction for multiple comparisons. Participants were self-selected from online communities and the patient and control groups differed in age (mean 33.8 against 43.8 years) and in several clinical variables by design ⁶⁹. The paper reports internally inconsistent numbers: the abstract gives 15.6% of patients with cannabis dependency diagnoses, 56.6% with withdrawal symptoms and 87.7% improving after cessation, where the results section gives 21.5%, 75% and 90.1% for what appear to be the same quantities ⁶⁹. Two authors are affiliated with a commercial cannabis-genomics company, Endocanna Health, and two with CReDO Science ⁶⁹, and the paper's conclusion proposes that the constellation of five variants may represent a valid diagnostic battery, pending corroboration from future testing ⁶⁹.

The authors are not hiding any of this. They describe the small number of returned kits as the primary limitation and attribute it partly to active discouragement within the online patient community ⁶⁹, and the 2024 review by the same group restates the five variants while continuing to call them a basis for future research ¹¹¹. What a study of this design and size can show is that a hypothesis is worth testing properly. What it cannot show is that COMT variation causes cannabinoid hyperemesis syndrome, or that a genetic test can identify who will develop it. No replication in an independent cohort was found. A review of research priorities for cyclic vomiting syndrome, published online in 2024 as the proceedings of a symposium, describes that condition's pathophysiology as multifactorial, with contributions from migraine, dysautonomia, endogenous cannabinoids, mitochondrial dysfunction, genetic abnormalities and rapid gastric emptying, notes that cannabinoid hyperemesis is hard to manage partly because sustained cannabis abstinence is difficult, and calls for controlled trials run by multicenter collaborations ¹¹². Nothing in it supports a genetic test for cannabinoid hyperemesis, and that is the right verdict: the question is open, and forty self-selected people have not closed it.

What was believed, and is not

COMT is the single gene where a physician can watch the whole machinery of a false consensus assemble and then come apart, and the useful thing is that none of the underlying biochemistry had to be withdrawn. What follows is the list, each item with the claim, the evidence it rested on, and the evidence that undid it.

That COMT Val158Met is a schizophrenia risk gene. The claim rested on a family-based association in 104 trios combined with cognitive and imaging data ⁶, a plausible dopamine mechanism, and the gene's location in a deletion region associated with psychosis ^3,37. It did not survive the genome-wide era: 108 loci in up to 36,989 cases without it ⁴³, a formal evaluation concluding the candidate-gene literature yielded no clear insight ³⁵, a gene-set analysis finding the candidate genes no more associated than random control genes ³⁶, and 287 loci and 120 fine-mapped genes in up to 76,755 cases, again without it ⁴⁴.

That the variant meaningfully shapes cognition. Meta-analysis of six cognitive phenotypes found no association with most, an effect on IQ of d = 0.06, evidence of publication bias, and effect sizes declining with publication year ³⁴.

That a person is a warrior or a worrier by genotype. The phrase comes from a short article that introduced its supporting evidence with the words some evidence suggests ⁶³, and the cognitive and psychiatric associations beneath it have not held ^34,43,44. What has held, at the level of brain activation rather than behavior, is a meta-analytic association between genotype and prefrontal activation with opposing directions for cognitive and emotional tasks ⁷⁷ -- which is a finding about how a region works a task, not about what kind of person someone is. Where sex-specific effects do appear, the most careful review of them says the evidence is incomplete and lacks a unifying explanation ⁷⁵. The same pattern -- one research network, overlapping cohorts, no independent replication, weak certainty -- is visible in the 2026 athlete-personality literature ⁷⁶.

That low COMT activity raises breast cancer risk. Predicted by the chemistry of catechol estrogens ^25,47, tested in meta-analyses of 26 studies and then of 30,199 cases, and not found; the European signal, where present, pointed the other way ^94,95.

That smoothing levodopa delivery with entacapone from the start would delay dyskinesia. A well-reasoned prediction, randomized in 747 patients, and wrong in the opposite direction: dyskinesia came sooner and was more frequent ⁶⁵.

That COMT genotype sets pain sensitivity in general. Supported for experimental pain and temporomandibular disorder ^61,62,66, narrowed by meta-analysis to fibromyalgia and chronic widespread pain only ⁶⁷, and absent from the 39 loci of an unbiased genome-wide study of multisite chronic pain in 380,000 people ⁶⁸. Where a randomized trial did find a COMT-by-drug interaction in pain, it ran opposite to the direction predicted in advance ⁸².

That synonymous DNA changes are silent. Overturned, in COMT itself, by the demonstration that haplotypes differing only in synonymous positions had the largest difference in enzyme activity, through messenger RNA stem-loop stability ⁶⁴. This one is a reversal in COMT's favor, and it generalizes far beyond this gene.

What is still unknown

Whether COMT activity matters clinically in any individual. The enzyme's activity varies three-to-four-fold by genotype ⁵ and that variation is real at the level of protein and tissue ¹³. What has never been shown is that knowing a person's genotype reliably changes what should be done for them -- not in psychiatry ^35,36, not in oncology ^94,95, and not yet in pain, where the unbiased scan does not find the gene ⁶⁸ and the clearest candidate signal sits at a different variant ⁷⁸. Randomized analyses have, though, produced genotype-dependent differences in outcome more than once: for alpha-tocopherol and total cancer ⁹⁹, for aspirin and cardiovascular disease in the same Women's Genome Health Study ¹², for tolcapone and verbal episodic memory in a 47-person crossover trial in healthy volunteers ¹¹³, and for propranolol and facial pain in a prespecified subset of 143 trial participants, where the interaction ran opposite to the prior hypothesis ⁸². Every one of them is a secondary or small analysis, none has been replicated in a trial designed for the question, and together they are a reason to keep testing the idea rather than a reason to genotype a patient.

Why the directions keep flipping within the 22q11.2 deletion syndrome literature. The low-activity allele was associated with bipolar spectrum disorder in 1996 ⁵⁹, with cortical and cognitive decline and psychotic symptoms in 2005 ⁴⁰, with nothing for schizophrenia in 1999 ³⁸, and in 2024 the high-activity allele was associated with higher anxiety in a direction the authors call contrary to expectation ⁴¹. The samples are tens of people each; whether there is a real effect being measured noisily, or no effect at all, is not settled.

How much of the methylation capacity is set by the enzyme and how much by the methyl supply. A 2025 study of combined oral contraceptive users found that methylated estrogen output correlated with methyl-donor availability and that serum betaine fell in users, suggesting the limiting step may sometimes be the donor rather than the enzyme ¹⁰⁰. If that is general, then everything written about COMT activity as a fixed personal trait is missing half the equation.

Whether tea catechins inhibit COMT enough in a living person to matter. The in vitro inhibition is powerful and well quantified in human liver preparations ^26,27, and the epidemiology is split: one case-control study found the tea and breast cancer association only in low-activity genotypes ¹⁰³, two larger case-control studies found no modification by genotype ^104,105, and the randomized test of the interaction, in 1,075 postmenopausal women stratified by COMT genotype stratum and given 843 mg of epigallocatechin gallate daily for a year, was largely negative for mammographic density, oxidative stress and inflammatory markers ^{106,107,108,109}. In 660 daily green tea drinkers, men homozygous for the low-activity allele excreted less tea polyphenol in their urine, which is at least consistent with slower methylation ¹¹⁴. What is still missing is a direct measurement of COMT activity, or of methylated catechol output, in people actually drinking tea: the trials measured downstream outcomes, not the enzyme.

Whether cannabinoid hyperemesis syndrome has a genetic basis at all. One candidate-gene study of 28 patients and 12 controls reported five variants, COMT among them ⁶⁹, with no independent replication, wide confidence intervals, a proposed diagnostic battery and two authors affiliated with a commercial cannabis-genomics company ^69,111. The question is live and unanswered ¹¹².

Whether a safer, more potent COMT inhibitor would do anything new. Opicapone is taken once daily and showed no liver signal in its trials, with treatment-emergent adverse events in 54% of patients on 50 mg against 57% on entacapone and 50% on placebo in the one head-to-head trial ²³; its benefit is measured in roughly an hour a day of off time at the licensed dose ^23,24; it did not beat placebo on off-period pain in its one trial for that indication ⁷⁰; and in patients over 65, nearly half had adverse events in routine use ⁹². A non-nitrocatechol inhibitor has been designed and tested in rats ⁹³. There is a live argument that a better inhibitor would do something new, and it is not about Parkinson's disease: a 2026 review sets out the human evidence that the brain-penetrant inhibitor tolcapone improves working memory in a way that depends on baseline prefrontal dopamine signaling -- better in those who start low, worse in those who start high, which is the inverted-U again -- and that the effects extend to impulsive and risky behavior, with rodent data pointing the same way, while stating plainly that the human studies are small; the first author's affiliation on that paper is a pharmaceutical company ¹¹⁵. Separately, a 2026 retrospective study of 237 patients treated with opicapone between 2017 and 2025 found heterozygosity at COMT rs4818 and rs6269 and the low-activity COMT haplotype associated with improvement in motor fluctuations, with no variant associated with stopping the drug; it is retrospective, single-center, scored from clinical records rather than rating scales and without pharmacokinetic data, and its authors call for prospective validation ¹¹⁶. Whether the class has a ceiling set by the pharmacology or by the trial designs is not known.

And the oldest one. Axelrod showed that O-methylation is the principal route of epinephrine metabolism in man ²⁹, but the modern accounting says most catecholamine turnover is intraneuronal deamination and that O-methylation outside the neuron is, by volume, minor ⁴⁶. Both statements are true of different things -- the fate of circulating epinephrine against the fate of all catecholamine made in the body -- and the tension between them is still the commonest source of confusion in teaching this subject. Axelrod supplied the missing piece himself, four years after the enzymology: a synapse is switched off mainly by reuptake of the transmitter into the terminal that released it, which is the discovery his share of the Nobel Prize was given for ^50,51.

Where it connects

Sources

116 sources, numbered as they are cited. Every one was checked against PubMed or its publisher before it was cited here; the note under each says what it shows and what it does not.

1. 1
   Vidgren J, Svensson LA, Liljas A. Crystal structure of catechol O-methyltransferase.Nature · 1994 · 368(6469):354-8doi:10.1038/368354a0 · PMID 8127373

   The atomic structure at 2.0 Angstrom resolution, with the observation that the coenzyme-binding domain resembles that of an S-adenosyl-L-methionine-dependent DNA methylase. Also states the enzyme's role in inactivating levodopa and the rationale for COMT inhibitors in Parkinson's disease.

2. 2
   Czarnota S, Johannissen LO, Baxter NJ, Rummel F, Wilson AL, Cliff MJ, Levy CW, Scrutton NS, Waltho JP, Hay S. Equatorial Active Site Compaction and Electrostatic Reorganization in Catechol-O-methyltransferase.ACS Catal · 2019 · 9(5):4394-4401doi:10.1021/acscatal.9b00174 · PMID 31080692

   X-ray structures and NMR of human soluble COMT with magnesium and either the methyl donor or sinefungin as a transition-state analog, with density functional theory and molecular dynamics. Supports active-site compaction with proton transfer to lysine 144. A physical-chemistry result in purified enzyme.

3. 3
   Grossman MH, Emanuel BS, Budarf ML. Chromosomal mapping of the human catechol-O-methyltransferase gene to 22q11.1----q11.2.Genomics · 1992 · 12(4):822-5doi:10.1016/0888-7543(92)90316-k · PMID 1572656

   Southern blotting against human-rodent and human-hamster hybrid cell panels localized the gene to 22q11.1-q11.2. Places COMT inside the region deleted in 22q11.2 deletion syndrome, which is the origin of the whole psychiatric literature; the mapping itself says nothing about psychiatric disease.

4. 4
   Tenhunen J, Salminen M, Lundström K, Kiviluoto T, Savolainen R, Ulmanen I. Genomic organization of the human catechol O-methyltransferase gene and its expression from two distinct promoters.Eur J Biochem · 1994 · 223(3):1049-59doi:10.1111/j.1432-1033.1994.tb19083.x · PMID 8055944

   Six exons, exons 1 and 2 non-coding, both start codons in exon 3; a proximal promoter giving the 1.3-kilobase soluble transcript and a distal promoter giving the 1.5-kilobase membrane-bound transcript, with tissue-specific differences in the ratio. The structural basis of the 108 versus 158 numbering.

5. 5
   Lachman HM, Papolos DF, Saito T, Yu YM, Szumlanski CL, Weinshilboum RM. Human catechol-O-methyltransferase pharmacogenetics: description of a functional polymorphism and its potential application to neuropsychiatric disorders.Pharmacogenetics · 1996 · 6(3):243-50doi:10.1097/00008571-199606000-00007 · PMID 8807664

   Identifies the G-to-A transition at codon 158 producing the valine-to-methionine substitution behind the three-to-four-fold activity difference, with a restriction-fragment assay. The neuropsychiatric applications are proposed as possibilities in the discussion; no clinical association is tested in this paper.

6. 6
   Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, Goldman D, Weinberger DR. Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia.Proc Natl Acad Sci U S A · 2001 · 98(12):6917-22doi:10.1073/pnas.111134598 · PMID 11381111

   175 patients, 219 siblings, 55 controls: allele-dosage relationship with Wisconsin Card Sorting Test perseverative errors explaining 4% of variance (p = 0.001); functional MRI in three subgroups of 11-16 people; excess transmission of the valine allele in 104 trios. The paper that made COMT famous. Each component is small, and the package did not survive genome-wide testing.

7. 7
   Rutherford K, Le Trong I, Stenkamp RE, Parson WW. Crystal structures of human 108V and 108M catechol O-methyltransferase.J Mol Biol · 2008 · 380(1):120-30 (online 23 Apr 2008)doi:10.1016/j.jmb.2008.04.040 · PMID 18486144

   The first structures of the human soluble enzyme, both the valine and the methionine form, with S-adenosylmethionine and a substrate analog. Residue 108 sits in a hydrophobic pocket about 16 Angstrom from the methyl-donor site, and the bulkier methionine displaces the backbone by 0.7 Angstrom. The paper states that the human and rat proteins have significantly different substrate specificities and that inhibitor development had until then relied on structures of the rat enzyme.

8. 8
   Ulmanen I, Peränen J, Tenhunen J, Tilgmann C, Karhunen T, Panula P, Bernasconi L, Aubry JP, Lundström K. Expression and intracellular localization of catechol O-methyltransferase in transfected mammalian cells.Eur J Biochem · 1997 · 243(1-2):452-9doi:10.1111/j.1432-1033.1997.0452a.x · PMID 9030772

   Membrane-bound COMT localizes to the endoplasmic reticulum and nuclear membrane with no plasma membrane signal; about 5 percent of the soluble form is nuclear. Transfected cell lines and rat primary neurons, so it is a cell-biology result, not a measurement in intact human brain.

9. 9
   Chen J, Song J, Yuan P, Tian Q, Ji Y, Ren-Patterson R, Liu G, Sei Y, Weinberger DR. Orientation and cellular distribution of membrane-bound catechol-O-methyltransferase in cortical neurons: implications for drug development.J Biol Chem · 2011 · 286(40):34752-60doi:10.1074/jbc.M111.262790 · PMID 21846718

   In rat cortical neurons, membrane-bound COMT was located in cell bodies, axons and dendrites, and computer simulation, flow cytometry and a cell-surface enzyme assay placed its C-terminal catalytic domain in the extracellular space, where it could inactivate synaptic and extrasynaptic dopamine. The authors state that the orientation of the enzyme on the membrane is controversial. Rat neurons and human lymphoblastoid cells, not intact human brain.

10. 10
    Matsumoto M, Weickert CS, Akil M, Lipska BK, Hyde TM, Herman MM, Kleinman JE, Weinberger DR. Catechol O-methyltransferase mRNA expression in human and rat brain: evidence for a role in cortical neuronal function.Neuroscience · 2003 · 116(1):127-37doi:10.1016/s0306-4522(02)00556-0 · PMID 12535946

    In situ hybridization showing COMT messenger RNA mainly in neurons, in all cortical layers of human prefrontal cortex and in striatal neurons, with lower levels in striatum than cortex. Contradicts the prior expectation of predominantly glial expression. Messenger RNA, not protein or activity.

11. 11
    Goldman D, Oroszi G, O'Malley S, Anton R. COMBINE genetics study: the pharmacogenetics of alcoholism treatment response: genes and mechanisms.J Stud Alcohol Suppl · 2005 · (15):56-64doi:10.15288/jsas.2005.s15.56 · PMID 16223057

    Describes Val158Met as a common functional polymorphism with minor allele frequency 0.42 and threefold relevance to executive function, stress and anxiety responses, and brain endogenous opioid function. Cited for the allele frequency and for the framing; it is a study-design paper, not a result.

12. 12
    Hall KT, Nelson CP, Davis RB, Buring JE, Kirsch I, Mittleman MA, Loscalzo J, Samani NJ, Ridker PM, Kaptchuk TJ, Chasman DI. Polymorphisms in catechol-O-methyltransferase modify treatment effects of aspirin on risk of cardiovascular disease.Arterioscler Thromb Vasc Biol · 2014 · 34(9):2160-7doi:10.1161/ATVBAHA.114.303845 · PMID 25035343

    Women's Genome Health Study, rs4680 minor allele frequency 0.47, with randomized allocation to aspirin or vitamin E against placebo and ten-year follow-up. Randomized aspirin abolished the genotype association seen on placebo: valine homozygotes had higher cardiovascular disease rates on aspirin than placebo (hazard ratio 1.85, 95% confidence interval 1.05-3.25, P = 0.033) and methionine homozygotes lower rates (0.60, 0.39-0.93, P = 0.023). The authors state plainly that replication is required.

13. 13
    Chen J, Lipska BK, Halim N, Ma QD, Matsumoto M, Melhem S, Kolachana BS, Hyde TM, Herman MM, Apud J, Egan MF, Kleinman JE, Weinberger DR. Functional analysis of genetic variation in catechol-O-methyltransferase (COMT): effects on mRNA, protein, and enzyme activity in postmortem human brain.Am J Hum Genet · 2004 · 75(5):807-21doi:10.1086/425589 · PMID 15457404

    108 postmortem prefrontal cortex samples: Val158Met affected protein abundance and enzyme activity but not messenger RNA, so the mechanism is protein stability. Two other variants previously reported as part of a schizophrenia haplotype had no effect on any measure. Also reports small effects of sex and of a distal promoter variant on activity.

14. 14
    Weinshilboum RM, Raymond FA. Inheritance of low erythrocyte catechol-O-methyltransferase activity in man.Am J Hum Genet · 1977 · 29(2):125-35PMID 848488

    373 randomly selected 16-to-18-year-olds, 262 adult blood donors and 201 first-degree relatives. Bimodal red-cell activity with a trough near 8 units, 23% below it; segregation in 48 families compatible with autosomal recessive inheritance of a low-activity allele. A phenotype, not a gene: the responsible DNA change was not identified until 1996.

15. 15
    Lenders JWM, Pacak K, Walther MM, Linehan WM, Mannelli M, Friberg P, Keiser HR, Goldstein DS, Eisenhofer G. Biochemical diagnosis of pheochromocytoma: which test is best?JAMA · 2002 · 287(11):1427-34doi:10.1001/jama.287.11.1427 · PMID 11903030

    214 patients with confirmed pheochromocytoma and 644 without, four referral centers, 1994-2001. Plasma free metanephrines 99% sensitive (96-100%) and 89% specific (87-92%), beating every alternative and every combination. A multicenter cohort at referral centers, so the pretest probability is higher than in general practice.

16. 16
    Eisenhofer G, Keiser H, Friberg P, et al. Plasma metanephrines are markers of pheochromocytoma produced by catechol-O-methyltransferase within tumors.J Clin Endocrinol Metab · 1998 · 83(6):2175-85doi:10.1210/jcem.83.6.4870 · PMID 9626157

    Western blot, enzyme assay and immunohistochemistry showed predominantly membrane-bound COMT in pheochromocytomas, colocalized with chromaffin cells; free metanephrines inside tumors exceeded the same patients' plasma levels more than 10,000-fold; over 93% of raised circulating metanephrines came from metabolism before release. The mechanistic basis of the test.

17. 17
    Ruottinen HM, Rinne UK. Entacapone prolongs levodopa response in a one month double blind study in parkinsonian patients with levodopa related fluctuations.J Neurol Neurosurg Psychiatry · 1996 · 60(1):36-40doi:10.1136/jnnp.60.1.36 · PMID 8558148

    A one-month double-blind crossover study, entacapone 200 mg with each levodopa dose. Entacapone prolonged the motor response to a single levodopa dose by 34 minutes (24%, P = 0.001) and dyskinesia by 39 minutes (37%, P = 0.002) without changing their magnitude, allowed a 16% reduction in mean total daily levodopa, and added 2.1 hours of mean daily on time by home diary. Small and crossover, but randomized, double-blind and published a year before the larger trials.

18. 18
    Kurth MC, Adler CH, Hilaire MS, Singer C, Waters C, LeWitt P, Chernik DA, Dorflinger EE, Yoo K. Tolcapone improves motor function and reduces levodopa requirement in patients with Parkinson's disease experiencing motor fluctuations: a multicenter, double-blind, randomized, placebo-controlled trial. Tolcapone Fluctuator Study Group I.Neurology · 1997 · 48(1):81-7doi:10.1212/wnl.48.1.81 · PMID 9008498

    Tolcapone 50, 200 and 400 mg three times daily against placebo in fluctuating patients, with two 10-hour clinical evaluations; 151 patients completed. Off time fell an average of 40% and total on time rose about 25% at all three doses against placebo, with significant reductions in levodopa-carbidopa dose and frequency. Published in the January 1997 issue, so it precedes the three larger tolcapone trials cited here.

19. 19
    Baas H, Beiske AG, Ghika J, Jackson M, Oertel WH, Poewe W, Ransmayr G. Catechol-O-methyltransferase inhibition with tolcapone reduces the wearing off phenomenon and levodopa requirements in fluctuating parkinsonian patients.J Neurol Neurosurg Psychiatry · 1997 · 63(4):421-8doi:10.1136/jnnp.63.4.421 · PMID 9343116

    58 placebo, 60 tolcapone 100 mg three times daily, 59 at 200 mg. At three months on 200 mg: off time down 26.2% of baseline, on time up 20.6% (p < 0.01), mean daily levodopa down 122 mg from 676 mg (p < 0.01), maintained to nine months. Diarrhea was the commonest non-dopaminergic event and the commonest reason for withdrawal.

20. 20
    Rajput AH, Martin W, Saint-Hilaire MH, Dorflinger E, Pedder S. Tolcapone improves motor function in parkinsonian patients with the wearing-off phenomenon: a double-blind, placebo-controlled, multicenter trial.Neurology · 1997 · 49(4):1066-71doi:10.1212/wnl.49.4.1066 · PMID 9339691

    202 patients. On 200 mg three times daily, daily off time fell 3.25 hours from baseline, significantly more than placebo (p < 0.01). Dyskinesia developed or worsened in 18% on placebo, 51% on 100 mg and 64% on 200 mg; 18% of all patients withdrew for adverse events. The efficacy and the dose-related harm in one trial.

21. 21
    Waters CH, Kurth M, Bailey P, Shulman LM, LeWitt P, Dorflinger E, Deptula D, Pedder S. Tolcapone in stable Parkinson's disease: efficacy and safety of long-term treatment. The Tolcapone Stable Study Group.Neurology · 1997 · 49(3):665-71doi:10.1212/wnl.49.3.665 · PMID 9305320

    298 patients on levodopa without motor fluctuations. Both doses significantly improved the activities of daily living and motor subscales of the Unified Parkinson's Disease Rating Scale at six months, maintained to twelve, with reductions in daily levodopa dose. Extends the drug beyond fluctuating patients; predates the liver findings.

22. 22
    Parkinson Study Group. Entacapone improves motor fluctuations in levodopa-treated Parkinson's disease patients.Ann Neurol · 1997 · 42(5):747-55doi:10.1002/ana.410420511 · PMID 9392574

    205 patients randomized to entacapone 200 mg with every levodopa dose or placebo for 24 weeks. Baseline about 10 hours of on time per day (60.5% of waking hours); entacapone added 5.0 percentage points of on time, with a larger effect in those starting below 55%. Home diaries at 30-minute intervals as the primary measure.

23. 23
    Ferreira JJ, Lees A, Rocha JF, Poewe W, Rascol O, Soares-da-Silva P. Opicapone as an adjunct to levodopa in patients with Parkinson's disease and end-of-dose motor fluctuations: a randomised, double-blind, controlled trial.Lancet Neurol · 2016 · 15(2):154-165doi:10.1016/S1474-4422(15)00336-1 · PMID 26725544

    600 randomized across 106 centers in 19 European countries and Russia. Only opicapone 50 mg beat placebo (-60.8 minutes of off time, 95% confidence interval -97.2 to -24.4, p = 0.0015) and was non-inferior to entacapone (-26.2 minutes, -63.8 to 11.4, p = 0.0051); 5 mg and 25 mg did not separate. Funded by BIAL, the manufacturer.

24. 24
    Lees AJ, Ferreira J, Rascol O, Poewe W, Rocha JF, McCrory M, Soares-da-Silva P. Opicapone as Adjunct to Levodopa Therapy in Patients With Parkinson Disease and Motor Fluctuations: A Randomized Clinical Trial.JAMA Neurol · 2017 · 74(2):197-206doi:10.1001/jamaneurol.2016.4703 · PMID 28027332

    427 patients (mean age 63.1, standard deviation 8.8). Least-squares mean change in off time -64.5 minutes (standard error 14.4) placebo, -101.7 (14.9) on 25 mg, -118.8 (13.8) on 50 mg; only 50 mg significant (treatment effect -54.3 minutes, 95% confidence interval -96.2 to -12.4, p = .008). Benefit sustained through a one-year open-label phase; 11.9% discontinued during the double-blind phase.

25. 25
    Lavigne JA, Helzlsouer KJ, Huang HY, Strickland PT, Bell DA, Selmin O, Watson MA, Hoffman S, Comstock GW, Yager JD. An association between the allele coding for a low activity variant of catechol-O-methyltransferase and the risk for breast cancer.Cancer Res · 1997 · 57(24):5493-7PMID 9407957

    112 matched nested case-control pairs. Odds ratios 1.30 (0.66-2.58) for one low-activity allele and 1.45 (0.69-3.07) for two, neither significant; 2.18 (0.93-5.11) in postmenopausal women, also not significant; significant only in the subgroup with body mass index above 24.47 kg/m2, at 3.58 (1.07-11.98). A subgroup finding in a small study, which later meta-analyses did not confirm.

26. 26
    Lu H, Meng X, Yang CS. Enzymology of methylation of tea catechins and inhibition of catechol-O-methyltransferase by (-)-epigallocatechin gallate.Drug Metab Dispos · 2003 · 31(5):572-9doi:10.1124/dmd.31.5.572 · PMID 12695345

    With human liver cytosol, Michaelis constants 4.0 micromolar for epigallocatechin versus 0.16 for epigallocatechin gallate, maximum velocities 1.28 versus 0.16 nanomoles per milligram of protein per minute; rat liver had higher activity than human or mouse. Epigallocatechin gallate (EGCG) and its methylated products inhibited COMT with half-maximal concentrations around 0.2 micromolar. In vitro liver preparations only.

27. 27
    Nagai M, Conney AH, Zhu BT. Strong inhibitory effects of common tea catechins and bioflavonoids on the O-methylation of catechol estrogens catalyzed by human liver cytosolic catechol-O-methyltransferase.Drug Metab Dispos · 2004 · 32(5):497-504doi:10.1124/dmd.32.5.497 · PMID 15100171

    Half-maximal inhibitory concentrations for 2-hydroxyestradiol methylation: catechin 14-17, epicatechin 44-65 and epigallocatechin gallate (EGCG) 0.04-0.07 micromolar; for 4-hydroxyestradiol: 5-7, 10-18 and 0.2-0.5 micromolar. Crude green and black tea extracts also inhibited strongly; mixed competitive and non-competitive kinetics. The authors state that more studies are needed to determine the extent of such inhibition in human subjects.

28. 28
    Männistö PT, Kaakkola S. Catechol-O-methyltransferase (COMT): biochemistry, molecular biology, pharmacology, and clinical efficacy of the new selective COMT inhibitors.Pharmacol Rev · 1999 · 51(4):593-628PMID 10581325

    The reference pharmacology review from the period when the selective inhibitors reached the clinic. The PubMed record carries no abstract, so it is cited for the existence, scope and date of the synthesis rather than for any specific figure.

29. 29
    LABROSSE EH, AXELROD J, KETY SS. O-Methylation, the principal route of metabolism of epinephrine in man.Science · 1958 · 128(3324):593-4doi:10.1126/science.128.3324.593 · PMID 13580222

    Establishes that O-methylation to metanephrine is the principal pathway of epinephrine metabolism in man and names COMT as the enzyme mainly involved in terminating epinephrine's action. Does not address catecholamines made and metabolized inside neurons, which later work showed dominates total turnover.

30. 30
    AXELROD J. O-methylation of epinephrine and other catechols in vitro and in vivo.Science · 1957 · 126(3270):400-1doi:10.1126/science.126.3270.400 · PMID 13467217

    The discovery paper. The PubMed record carries no abstract, so it is cited for the existence and date of the finding that catechols are O-methylated in the test tube and in the living animal, not for a number.

31. 31
    Assal F, Spahr L, Hadengue A, Rubbia-Brandt L, Burkhard PR. Tolcapone and fulminant hepatitis.Lancet · 1998 · 352(9132):958doi:10.1016/s0140-6736(05)61511-5 · PMID 9752821

    The case report letter that opened the liver question. The PubMed record carries no abstract; cited for the existence, date and nature of the first published report of fulminant hepatitis on tolcapone, not for any rate.

32. 32
    Watkins P. COMT inhibitors and liver toxicity.Neurology · 2000 · 55(11 Suppl 4):S51-2; discussion S53-6PMID 11147510

    Liver chemistry tests above three times the upper limit of normal in about 1% of patients on tolcapone 100 mg and about 3% on 200 mg; post-marketing, three instances of acute liver failure with death after 60,000 patients and about 40,000 patient-years; withdrawal in Europe and Canada and a boxed warning in the United States. Entacapone showed no excess over placebo and no attributed liver failure.

33. 33
    Olanow CW, Watkins PB. Tolcapone: an efficacy and safety review (2007).Clin Neuropharmacol · 2007 · 30(5):287-94doi:10.1097/wnf.0b013e318038d2b6 · PMID 17909307

    Counts four cases of acute hepatotoxicity with three fatalities, and reports more than 40,000 patient-years since the 1998 labeling restrictions with only three severe but reversible liver injuries and no hepatic fatality. On that basis the drug was reintroduced in several European countries, with liver monitoring every two to four weeks for six months and withdrawal above twice the upper limit of normal.

34. 34
    Barnett JH, Scoriels L, Munafò MR. Meta-analysis of the cognitive effects of the catechol-O-methyltransferase gene Val158/108Met polymorphism.Biol Psychiatry · 2008 · 64(2):137-44doi:10.1016/j.biopsych.2008.01.005 · PMID 18339359

    Six cognitive phenotypes meta-analyzed: no association with most, IQ effect d = 0.06, no robust n-back association, evidence of publication bias, effect sizes decreasing with later publication year. Concludes the polymorphism appears to have little if any association with cognitive function. The clearest single refutation of the cognition claim.

35. 35
    Farrell MS, Werge T, Sklar P, Owen MJ, Ophoff RA, O'Donovan MC, Corvin A, Cichon S, Sullivan PF. Evaluating historical candidate genes for schizophrenia.Mol Psychiatry · 2015 · 20(5):555-62doi:10.1038/mp.2015.16 · PMID 25754081

    Twenty-five historical candidate genes including COMT, assessed by meta-analysis of their own literature, against the largest genomic study then available, and by the ratings of informed researchers and 24 schizophrenia geneticists. Concludes the candidate gene literature did not yield clear insights, most likely from inadequate power.

36. 36
    Johnson EC, Border R, Melroy-Greif WE, de Leeuw CA, Ehringer MA, Keller MC. No Evidence That Schizophrenia Candidate Genes Are More Associated With Schizophrenia Than Noncandidate Genes.Biol Psychiatry · 2017 · 82(10):702-708doi:10.1016/j.biopsych.2017.06.033 · PMID 28823710

    Gene-set analysis of the historical candidate genes against control sets of non-candidate genes using the largest schizophrenia genome-wide data. As a group the candidates were no more associated than the controls. Offered by the authors as a cautionary tale for candidate-gene work on any complex trait.

37. 37
    Schneider M, Debbané M, Bassett AS, et al. Psychiatric disorders from childhood to adulthood in 22q11.2 deletion syndrome: results from the International Consortium on Brain and Behavior in 22q11.2 Deletion Syndrome.Am J Psychiatry · 2014 · 171(6):627-39doi:10.1176/appi.ajp.2013.13070864 · PMID 24577245

    1,402 participants aged 6-68 assessed with validated instruments: attention-deficit hyperactivity disorder in 37.10% of children, anxiety disorders more prevalent than mood disorders at all ages, psychotic disorders in 41% of adults over 25. Establishes the magnitude of the risk; does not attribute it to any single gene in the deleted region.

38. 38
    Murphy KC, Jones LA, Owen MJ. High rates of schizophrenia in adults with velo-cardio-facial syndrome.Arch Gen Psychiatry · 1999 · 56(10):940-5doi:10.1001/archpsyc.56.10.940 · PMID 10530637

    Fifty adults with the syndrome, structured interviews: 30% had a psychotic disorder, 24% met criteria for schizophrenia. All 48 deleted individuals were genotyped for COMT and the authors found no evidence that the low-activity allele was associated with schizophrenia. The negative result inside the syndrome that generated the hypothesis.

39. 39
    Monks S, Niarchou M, Davies AR, Walters JTR, Williams N, Owen MJ, van den Bree MBM, Murphy KC. Further evidence for high rates of schizophrenia in 22q11.2 deletion syndrome.Schizophr Res · 2014 · 153(1-3):231-6 (online 15 Feb 2014)doi:10.1016/j.schres.2014.01.020 · PMID 24534796

    Eighty-three adults with the deletion against 90 people with schizophrenia without the deletion and 316 controls, assessed with the Schedules for Clinical Assessment in Neuropsychiatry and two schizotypy questionnaires, and genotyped for several COMT polymorphisms including the Ashkenazi risk haplotype. Psychotic disorder was present in 29% and schizophrenia in 22%; the symptom profile did not differ from sporadic schizophrenia; and COMT was associated with neither schizophrenia status nor schizotypy. The largest direct test of the COMT hypothesis inside the syndrome.

40. 40
    Gothelf D, Eliez S, Thompson T, Hinard C, Penniman L, Feinstein C, Kwon H, Jin S, Jo B, Antonarakis SE, Morris MA, Reiss AL. COMT genotype predicts longitudinal cognitive decline and psychosis in 22q11.2 deletion syndrome.Nat Neurosci · 2005 · 8(11):1500-2doi:10.1038/nn1572 · PMID 16234808

    Longitudinal study of adolescents with the deletion identifying the low-activity allele as a risk factor for prefrontal volume and cognitive decline and for psychotic symptoms. A small longitudinal cohort of adolescents; the direction of effect is opposite to the high-activity allele implicated in general-population schizophrenia work.

41. 41
    Beebe JL, Martin CR, Sanders AFP, Guidry J, Faheem F, Atallah J, Beaton EA. Short communication: catechol-O-methyltransferase allelic variation in relation to psychological and hormonal indices of stress in children and adolescents with chromosome 22q11.2 deletion syndrome (22q11.2DS).J Appl Genet · 2025 · 66(2):383-388 (online 10 Aug 2024)doi:10.1007/s13353-024-00898-5 · PMID 39126435

    Thirty children and adolescents with the deletion. Contrary to the authors' stated expectation, the higher-activity valine allele went with higher anxiety (p = 0.021); salivary alpha-amylase was higher in the low-activity group (p = 0.026, described as uncorrected); cortisol did not differ. Thirty participants, so hypothesis-generating only.

42. 42
    Gogos JA, Morgan M, Luine V, Santha M, Ogawa S, Pfaff D, Karayiorgou M. Catechol-O-methyltransferase-deficient mice exhibit sexually dimorphic changes in catecholamine levels and behavior.Proc Natl Acad Sci U S A · 1998 · 95(17):9991-6doi:10.1073/pnas.95.17.9991 · PMID 9707588

    Knockout mice are viable, with region-specific and sex-specific dopamine changes, notably frontal cortex; homozygous females but not males impaired in a light-dark anxiety test; heterozygous males more aggressive. A mouse model of enzyme loss, not of any human psychiatric disorder.

43. 43
    Schizophrenia Working Group of the Psychiatric Genomics Consortium. Biological insights from 108 schizophrenia-associated genetic loci.Nature · 2014 · 511(7510):421-7doi:10.1038/nature13595 · PMID 25056061

    Up to 36,989 cases and 113,075 controls, 128 independent associations across 108 loci, 83 new, enriched in brain-expressed genes with hits at the dopamine D2 receptor and glutamatergic genes. COMT is not among the highlighted loci. Absence from a list is weaker evidence than a direct test, which is why it is cited alongside the formal evaluations.

44. 44
    Trubetskoy V, Pardiñas AF, Qi T, et al. Mapping genomic loci implicates genes and synaptic biology in schizophrenia.Nature · 2022 · 604(7906):502-508doi:10.1038/s41586-022-04434-5 · PMID 35396580

    Up to 76,755 cases and 243,649 controls, 287 distinct loci, 120 fine-mapped likely causal genes concentrated in excitatory and inhibitory neurons and converging on synaptic organization and transmission. COMT is not among the prioritized genes. The current picture of schizophrenia genetics.

45. 45
    Fotsis T, Zhang Y, Pepper MS, Adlercreutz H, Montesano R, Nawroth PP, Schweigerer L. The endogenous oestrogen metabolite 2-methoxyoestradiol inhibits angiogenesis and suppresses tumour growth.Nature · 1994 · 368(6468):237-9doi:10.1038/368237a0 · PMID 7511798

    2-methoxyestradiol, the COMT product of 2-hydroxyestradiol, inhibited endothelial proliferation, migration and angiogenesis in vitro and suppressed solid tumor growth in mice given it orally, without the heparin co-administration other angiostatic steroids needed. Cell culture and mice; no human evidence.

46. 46
    Eisenhofer G, Kopin IJ, Goldstein DS. Catecholamine metabolism: a contemporary view with implications for physiology and medicine.Pharmacol Rev · 2004 · 56(3):331-49doi:10.1124/pr.56.3.1 · PMID 15317907

    Corrects several textbook errors: most catecholamine metabolism is intraneuronal, extraneuronal O-methylation to metanephrines is a minor pathway by volume, the adrenal medulla is the single largest source of metanephrines, and vanillylmandelic acid is made in the liver rather than in the nerve.

47. 47
    Cavalieri EL, Rogan EG. Depurinating estrogen-DNA adducts, generators of cancer initiation: their minimization leads to cancer prevention.Clin Transl Med · 2016 · 5(1):12doi:10.1186/s40169-016-0088-3 · PMID 26979321

    States the genotoxic hypothesis: catechol estrogen-3,4-quinones react with DNA to form depurinating adducts leaving apurinic sites and mutations, with COMT as the protective capping enzyme in the balance. A review by the hypothesis's own authors, which also advocates specific supplements; treat the prevention claims as the authors' position, not as settled evidence.

48. 48
    Bruce JY, Eickhoff J, Pili R, Logan T, Carducci M, Arnott J, Treston A, Wilding G, Liu G. A phase II study of 2-methoxyestradiol nanocrystal colloidal dispersion alone and in combination with sunitinib malate in patients with metastatic renal cell carcinoma progressing on sunitinib malate.Invest New Drugs · 2012 · 30(2):794-802doi:10.1007/s10637-010-9618-9 · PMID 21174224

    17 patients enrolled, 12 evaluable; no objective responses; study terminated because the drug was not tolerable at the recommended phase 2 dose, with 35% discontinuing for toxicity. The human end of the 2-methoxyestradiol story, and a negative one. Small and single-arm within strata.

49. 49
    Neurophysiologists honored (Ulf von Euler, Julius Axelrod, Bernard Katz).Nature · 1970 · 228(5269):304PMID 4319740

    Contemporary news report of the 1970 Nobel Prize in Physiology or Medicine, indexed in PubMed and carrying no abstract; cited for the fact, date and three laureates. The Nobel Foundation's own announcement is the primary document but was not opened for this page, so it is not cited.

50. 50
    Hertting G, Axelrod J. Fate of tritiated noradrenaline at the sympathetic nerve-endings.Nature · 1961 · 192:172-3doi:10.1038/192172a0 · PMID 13906919

    The reuptake experiment: tritiated noradrenaline infused and recovered in the sympathetic nerve endings. The PubMed record carries no abstract, so it is cited for the existence and date of the finding, not for a number.

51. 51
    Iversen L. Julius Axelrod: 20 May 1912 - 29 December 2004.Biogr Mem Fellows R Soc · 2006 · 52:1-13doi:10.1098/rsbm.2006.0002 · PMID 18543469

    Royal Society biographical memoir. States that Axelrod shared the 1970 Nobel Prize in Physiology or Medicine for his discovery that reuptake of noradrenaline into the nerve endings that released it is a novel mechanism of neurotransmitter inactivation, that an important corollary was that antidepressant drugs inhibit that uptake, and separately that among his early achievements was establishing COMT as an important enzyme in catecholamine metabolism. A memoir, so it is cited for the historical record rather than for data.

52. 52
    Snyder SH. Turning off neurotransmitters.Cell · 2006 · 125(1):13-5doi:10.1016/j.cell.2006.03.023 · PMID 16615882

    Short historical commentary stating that the catecholamine neurotransmitters of the sympathetic nervous system are inactivated through reuptake by presynaptic nerve terminals, and that this discovery paved the way for modern antidepressants. A commentary, cited for the historical framing.

53. 53
    Axelrod J. Noradrenaline: fate and control of its biosynthesis.Science · 1971 · 173(3997):598-606doi:10.1126/science.173.3997.598 · PMID 4397955

    Axelrod's account of the whole noradrenaline scheme as understood at the time of his Nobel Prize; the PubMed record gives no publication type beyond journal article. The PubMed record carries no abstract; cited for the existence, date and scope of the synthesis.

54. 54
    Bertocci B, Miggiano V, Da Prada M, Dembic Z, Lahm HW, Malherbe P. Human catechol-O-methyltransferase: cloning and expression of the membrane-associated form.Proc Natl Acad Sci U S A · 1991 · 88(4):1416-20doi:10.1073/pnas.88.4.1416 · PMID 1847521

    Complementary DNA cloned from a human hepatoma cell line: a deduced 271-residue protein of about 30 kilodaltons with a hydrophobic 21-residue N-terminal segment that may insert it into the endoplasmic reticulum membrane. Endogenous activity in transfected human kidney cells rose from about 9.98 to 206 units per milligram of protein, and the recombinant protein was mainly membrane-associated by immunoblot. One of the two 1991 cloning papers.

55. 55
    Lundström K, Salminen M, Jalanko A, Savolainen R, Ulmanen I. Cloning and characterization of human placental catechol-O-methyltransferase cDNA.DNA Cell Biol · 1991 · 10(3):181-9doi:10.1089/dna.1991.10.181 · PMID 1707278

    Complementary DNA cloned from human placenta. All four clones contained an open reading frame for the soluble enzyme; two carried a 5' extension coding a further 50 amino acids with a hydrophobic signal-like N terminus. Cell-free translation gave both a 30-kilodalton and a 26-kilodalton product, implying two initiation codons, and only the longer product associated with microsomal membranes. The 50 extra amino acids behind the 108-versus-158 numbering were therefore described three years before the gene structure was solved.

56. 56
    ARMSTRONG MD, McMILLAN A, SHAW KN. 3-Methoxy-4-hydroxy-D-mandelic acid, a urinary metabolite of norepinephrine.Biochim Biophys Acta · 1957 · 25(2):422-3doi:10.1016/0006-3002(57)90491-2 · PMID 13471587

    The identification of vanillylmandelic acid as a urinary metabolite of norepinephrine, published in the same month as Axelrod's report of the O-methylation reaction. The PubMed record carries no abstract; cited for the existence and date of the identification.

57. 57
    ARMSTRONG MD, McMILLAN A. Studies on the formation of 3-methoxy-4-hydroxy-D-mandelic acid, a urinary metabolite of norepinephrine and epinephrine.Pharmacol Rev · 1959 · 11(2 Pt 2):394-401PMID 13667420

    The follow-up setting out the formation of vanillylmandelic acid from norepinephrine and epinephrine. The PubMed record carries no abstract; cited for the existence and date of the work.

58. 58
    AXELROD J, TOMCHICK R. Enzymatic O-methylation of epinephrine and other catechols.J Biol Chem · 1958 · 233(3):702-5PMID 13575440

    The enzymology of the reaction, a year after the two-page report of its existence. The PubMed record carries no abstract; cited for the existence and date of the characterization.

59. 59
    Lachman HM, Morrow B, Shprintzen R, Veit S, Parsia SS, Faedda G, Goldberg R, Kucherlapati R, Papolos DF. Association of codon 108/158 catechol-O-methyltransferase gene polymorphism with the psychiatric manifestations of velo-cardio-facial syndrome.Am J Med Genet · 1996 · 67(5):468-72doi:10.1002/(SICI)1096-8628(19960920)67:5<468::AID-AJMG5>3.0.CO;2-G · PMID 8886163

    Reports an apparent association between the low-activity allele and bipolar spectrum disorder, particularly rapid-cycling, in patients with velo-cardio-facial syndrome. A small association study using the word apparent; the first clean test of the same idea for schizophrenia, three years later, was negative.

60. 60
    Weinberger DR, Egan MF, Bertolino A, Callicott JH, Mattay VS, Lipska BK, Berman KF, Goldberg TE. Prefrontal neurons and the genetics of schizophrenia.Biol Psychiatry · 2001 · 50(11):825-44doi:10.1016/s0006-3223(01)01252-5 · PMID 11743939

    The companion review describing the COMT finding as the first plausible mechanism for a genetic effect on normal human cognition and risk for mental illness. Useful as the clearest statement of what the field believed in 2001, and so as the baseline against which the later reversal should be read.

61. 61
    Zubieta JK, Heitzeg MM, Smith YR, Bueller JA, Xu K, Xu Y, Koeppe RA, Stohler CS, Goldman D. COMT val158met genotype affects mu-opioid neurotransmitter responses to a pain stressor.Science · 2003 · 299(5610):1240-3doi:10.1126/science.1078546 · PMID 12595695

    Positron emission tomography during sustained muscle pain: methionine homozygotes showed diminished regional mu-opioid responses with higher pain and more negative affect; valine homozygotes the opposite. A small, mechanistically specific human imaging study. It measures brain response to experimental pain, not clinical pain outcomes.

62. 62
    Diatchenko L, Slade GD, Nackley AG, Bhalang K, Sigurdsson A, Belfer I, Goldman D, Xu K, Shabalina SA, Shagin D, Max MB, Makarov SS, Maixner W. Genetic basis for individual variations in pain perception and the development of a chronic pain condition.Hum Mol Genet · 2005 · 14(1):135-43doi:10.1093/hmg/ddi013 · PMID 15537663

    Three haplotypes covering 96% of the population studied; five combinations associated with experimental pain sensitivity (p = 0.0004); a single low-pain-sensitivity haplotype reduced myogenous temporomandibular disorder risk by as much as 2.3 times; COMT inhibition in rats increased pain sensitivity. Human association plus a rat experiment; the human arm is an association.

63. 63
    Stein DJ, Newman TK, Savitz J, Ramesar R. Warriors versus worriers: the role of COMT gene variants.CNS Spectr · 2006 · 11(10):745-8doi:10.1017/s1092852900014863 · PMID 17008817

    The review that named the framing: valine alleles proposed as advantageous in processing aversive stimuli, methionine alleles in memory and attention, reversing under stress. A summary of associations, whose supporting evidence it introduces with the words some evidence suggests. PubMed lists its publication types as case reports and journal article; it is not an experiment and does not claim to be one. The phrase unconfirmed or minor belongs to Harrison and Tunbridge [75], not to this paper. No earlier published use of the warrior and worrier phrasing appears in PubMed.

64. 64
    Nackley AG, Shabalina SA, Tchivileva IE, Satterfield K, Korchynskyi O, Makarov SS, Maixner W, Diatchenko L. Human catechol-O-methyltransferase haplotypes modulate protein expression by altering mRNA secondary structure.Science · 2006 · 314(5807):1930-3doi:10.1126/science.1131262 · PMID 17185601

    Haplotypes differing only at synonymous positions showed the largest activity difference, through messenger RNA stem-loop stability; removing the stable structure by site-directed mutagenesis restored protein output. A general result about synonymous variants that happens to have been found in COMT, and one of the few COMT findings that has strengthened rather than weakened.

65. 65
    Stocchi F, Rascol O, Kieburtz K, Poewe W, Jankovic J, Tolosa E, Barone P, Lang AE, Olanow CW. Initiating levodopa/carbidopa therapy with and without entacapone in early Parkinson disease: the STRIDE-PD study.Ann Neurol · 2010 · 68(1):18-27doi:10.1002/ana.22060 · PMID 20582993

    747 patients, 134 weeks, double-blind. Adding entacapone from the start shortened time to dyskinesia (hazard ratio 1.29, p = 0.04) and raised its frequency at week 134 (42% versus 32%, p = 0.02), the opposite of the continuous-delivery hypothesis. The authors note the regimen did not achieve continuous levodopa availability and that the entacapone group received higher levodopa dose equivalents.

66. 66
    Smith SB, Maixner DW, Greenspan JD, Dubner R, Fillingim RB, Ohrbach R, Knott C, Slade GD, Bair E, Gibson DG, Zaykin DV, Weir BS, Maixner W, Diatchenko L. Potential genetic risk factors for chronic TMD: genetic associations from the OPPERA case control study.J Pain · 2011 · 12(11 Suppl):T92-101doi:10.1016/j.jpain.2011.08.005 · PMID 22074755

    3,295 variants across 358 pain genes in 166 cases and 1,442 controls, plus 182 cases and 170 controls from a similar study. Supports prior associations for HTR2A and COMT and raises new candidates. The authors state explicitly that the findings need replication in independent cohorts.

67. 67
    Tammimäki A, Männistö PT. Catechol-O-methyltransferase gene polymorphism and chronic human pain: a systematic review and meta-analysis.Pharmacogenet Genomics · 2012 · 22(9):673-91doi:10.1097/FPC.0b013e3283560c46 · PMID 22722321

    Meta-analysis across migrainous headache, fibromyalgia or chronic widespread pain, and chronic musculoskeletal pain. Only fibromyalgia and chronic widespread pain could be associated with rs4680, with the methionine allele as the risk allele; no association with migraine or chronic musculoskeletal pain. Narrows a broad claim to a specific one.

68. 68
    Johnston KJA, Adams MJ, Nicholl BI, Ward J, Strawbridge RJ, Ferguson A, McIntosh AM, Bailey MES, Smith DJ. Genome-wide association study of multisite chronic pain in UK Biobank.PLoS Genet · 2019 · 15(6):e1008164doi:10.1371/journal.pgen.1008164 · PMID 31194737

    About 380,000 participants, single-nucleotide-polymorphism heritability 10.2%, 76 independent lead variants at 39 risk loci, gene-level enrichment in neurogenesis and synaptic plasticity. COMT is not among the loci. An unbiased scan of a pain phenotype, and so the natural comparator for the candidate-gene pain literature.

69. 69
    Russo EB, Spooner C, May L, Leslie R, Whiteley VL. Cannabinoid Hyperemesis Syndrome Survey and Genomic Investigation.Cannabis Cannabinoid Res · 2022 · 7(3):336-344doi:10.1089/can.2021.0046 · PMID 34227878

    Online survey of 585 respondents; genotyping of 28 patients and 12 controls. Strongest finding COMT rs4646316, 57.1% versus 10%, odds ratio 12, p = 0.012, with confidence limits 1.3-88.1 in the abstract and 1.3-98.1 in the table. CNR1, predicted by the authors' own hypothesis, was not significant. Forty people, very wide intervals, no stated multiple-comparison correction, self-selected recruitment, internally inconsistent percentages between abstract and results, and two authors affiliated with Endocanna Health, a commercial cannabis-genomics company.

70. 70
    Chaudhuri KR, Ferreira JJ, Antonini A, Rascol O, Kurtis MM, Storch A, Rocha JF, Costa R, Magalhães D, Castilla-Fernández G, Brigas HC, Holenz J, Odin P. Challenges of Pain in Parkinson's Disease: Results from the OCEAN Study.Mov Disord · 2026 · published online 8 June 2026doi:10.1002/mds.70390 · PMID 42259665

    122 patients randomized to opicapone 50 mg or placebo for 24 weeks for fluctuation-related pain. Primary endpoint, change in King's Parkinson's Disease Pain Scale domain 3: -8.8 (standard deviation 6.9) versus -9.3 (standard deviation 6.2), P = 0.7940. A clean negative result on a pre-specified primary endpoint. Several authors are employees of the manufacturer.

71. 71
    Ferreira JJ, Rascol O, Stocchi F, Antonini A, Moreira J, Fonseca MM, Rocha JF, Brigas HC, Holenz J, Poewe W. Long-term follow up of opicapone as add-on to levodopa in Parkinson's patients without motor fluctuations.J Parkinsons Dis · 2026 · 16(4):797-802doi:10.1177/1877718X261436957 · PMID 41995421

    One-year open-label extension of the EPSILON trial (NCT04978597). Adjusted mean change in the motor score of the Movement Disorder Society Unified Parkinson's Disease Rating Scale from double-blind baseline to week 76 was -7.4 plus or minus 0.81; placebo-to-opicapone switchers -6.1 plus or minus 0.79. Motor-complication-free at week 76: 80.2% versus 69.7%, p = 0.1, so not significant. Several authors are employees of the manufacturer.

72. 72
    Tripathi SJ, Chakraborty S, Wood NB, Hoopes D, Barker S, Vázquez-Rosa E, An J, Ma C, Hou Y, Sharma SM, Cheng F, Thomas B, Orsburn BC, Fried SD, Snyder SH, Pieper AA, Paul BD. Catechol-O-methyltransferase connects dopamine homeostasis to redox signaling, metal homeostasis, and protein folding in schizophrenia.Proc Natl Acad Sci U S A · 2026 · 123(38):e2606205123 (online 16 Sep 2026)doi:10.1073/pnas.2606205123 · PMID 42748150

    PC12 cells in which the membrane-bound isoform had been deleted by gene editing showed impaired dopamine metabolism, greater vulnerability to hydrogen peroxide and to dopamine itself, and unbiased proteomics implicating synaptic and mitochondrial function, iron and copper handling and redox balance; limited-proteolysis mass spectrometry found widespread changes in protein structure. The proteomic signature overlapped genes and proteins dysregulated in postmortem prefrontal cortex from people with schizophrenia. A cell-line and postmortem-tissue study, not a finding about any living patient.

73. 73
    Tunbridge EM, Narajos M, Harrison CH, Beresford C, Cipriani A, Harrison PJ. Which Dopamine Polymorphisms Are Functional? Systematic Review and Meta-analysis of COMT, DAT, DBH, DDC, DRD1-5, MAOA, MAOB, TH, VMAT1, and VMAT2.Biol Psychiatry · 2019 · 86(8):608-620doi:10.1016/j.biopsych.2019.05.014 · PMID 31303260

    Systematic review and meta-analysis that screened 22,728 articles and included 255 eligible studies, asking for every commonly studied polymorphism in the dopamine system whether it changes the expression, abundance, activity or affinity of the gene product. For most the evidence was negative, inconclusive or absent. COMT Val158Met was one of a handful with robust, medium-to-large effects, markedly affecting enzyme activity, protein abundance and protein stability. It addresses function, not clinical prediction.

74. 74
    Goldman D, Oroszi G, Ducci F. The genetics of addictions: uncovering the genes.Nat Rev Genet · 2005 · 6(7):521-32doi:10.1038/nrg1635 · PMID 15995696

    Review placing COMT among variants affecting reward, behavioral control and stress response in addiction. Context for how the gene was being framed in 2005; contains no new data on COMT.

75. 75
    Harrison PJ, Tunbridge EM. Catechol-O-methyltransferase (COMT): a gene contributing to sex differences in brain function, and to sexual dimorphism in the predisposition to psychiatric disorders.Neuropsychopharmacology · 2008 · 33(13):3037-45doi:10.1038/sj.npp.1301543 · PMID 17805313

    Review of sexually dimorphic COMT effects. Reports Val158Met associated with obsessive-compulsive disorder in men, anxiety phenotypes in women, and greater cognitive impact in boys than girls, while stating that many differences are unconfirmed or minor, that estrogen regulation is the usual explanation but other mechanisms are likely at least as important, and that the evidence lacks a unifying explanation.

76. 76
    Łosińska K, Hamza AS, Depta J, Recław R, Grzywacz A, Jaworski Ł, Maszczyk A. Dopaminergic Gene Polymorphisms and Personality Traits in Athletes: A Systematic Review and Exploratory Quantitative Synthesis.Int J Mol Sci · 2026 · 27(19):8484doi:10.3390/ijms27198484 · PMID 42883732

    Fourteen studies, mostly overlapping cohorts from a single research network. The COMT rs4680 methionine allele was less frequent in athletes than controls (odds ratio 0.75, 0.59-0.96), but no gene-by-trait association was replicated in an independent sample and the certainty of evidence was rated weak for every association examined. A current illustration of how this literature still behaves.

77. 77
    Mier D, Kirsch P, Meyer-Lindenberg A. Neural substrates of pleiotropic action of genetic variation in COMT: a meta-analysis.Mol Psychiatry · 2010 · 15(9):918-27 (online 5 May 2009)doi:10.1038/mp.2009.36 · PMID 19417742

    Meta-analysis of the available neuroimaging studies of rs4680: a significant association with prefrontal activation, effect size d = 0.73, without evidence of publication bias, with strong and opposing effects for executive-cognition paradigms (favoring methionine carriers) and emotional paradigms (favoring valine). The authors present it as validating intermediate phenotypes. Activation is an intermediate measure, not an outcome, and a 2009-vintage imaging meta-analysis cannot exclude the small-study problems that afflict the behavioral literature.

78. 78
    Almeida DAF, Brant CF, da Costa Siqueira L, Soares LFF, Oliveira JA, de Oliveira DSB, Pigossi SC, Soares CJ. Genetic polymorphisms on temporomandibular disorders: Network meta-analysis.Arch Oral Biol · 2025 · 174:106235doi:10.1016/j.archoralbio.2025.106235 · PMID 40107000

    Sixty-three studies reviewed, seven in the network meta-analysis, covering 120 genes and 206 polymorphisms. The clearest signal is at COMT rs165774, not rs4680: odds ratios 0.33 (0.14-0.76) and 0.32 (0.14-0.74), both p < 0.01, protective for myalgia, with that variant ranking highest for myalgia (97%) and arthralgia (93%).

79. 79
    Klepstad P, Fladvad T, Skorpen F, Bjordal K, Caraceni A, Dale O, Davies A, Kloke M, Lundström S, Maltoni M, Radbruch L, Sabatowski R, Sigurdardottir V, Strasser F, Fayers PM, Kaasa S. Influence from genetic variability on opioid use for cancer pain: a European genetic association study of 2294 cancer pain patients.Pain · 2011 · 152(5):1139-1145doi:10.1016/j.pain.2011.01.040 · PMID 21398039

    2,294 adult European patients taking a World Health Organization step III opioid for cancer pain, tested for 112 single-nucleotide polymorphisms in 25 candidate genes including COMT, with the patients split into a development and a validation sample. None of the 112 variants was significantly associated with opioid dose in both samples. The authors offer the study as a demonstration of why genetic associations must be validated before being reported as findings.

80. 80
    Tchivileva IE, Lim PF, Smith SB, Slade GD, Diatchenko L, McLean SA, Maixner W. Effect of catechol-O-methyltransferase polymorphism on response to propranolol therapy in chronic musculoskeletal pain: a randomized, double-blind, placebo-controlled, crossover pilot study.Pharmacogenet Genomics · 2010 · 20(4):239-48doi:10.1097/FPC.0b013e328337f9ab · PMID 20216107

    Forty women meeting research diagnostic criteria for temporomandibular disorder, genotyped for COMT and crossed over between propranolol and placebo. More patients reported reduced pain intensity on propranolol (P = 0.014) and a composite pain index fell (P = 0.02), while other clinical and experimental pain measures did not. Benefit was seen in patients not carrying the high-activity haplotype, was diminished in heterozygotes and was absent in homozygotes. A 40-person pilot, offered by its authors as hypothesis-generating.

81. 81
    Tchivileva IE, Hadgraft H, Lim PF, Di Giosia M, Ribeiro-Dasilva M, Campbell JH, Willis J, James R, Herman-Giddens M, Fillingim RB, Ohrbach R, Arbes SJ, Slade GD. Efficacy and safety of propranolol for treatment of temporomandibular disorder pain: a randomized, placebo-controlled clinical trial.Pain · 2020 · 161(8):1755-1767doi:10.1097/j.pain.0000000000001882 · PMID 32701836

    Phase 2b trial, 200 patients with temporomandibular disorder myalgia randomized to extended-release propranolol 60 mg twice daily or placebo, 199 analyzed. The primary endpoint, mean change in the facial pain index at week 9, did not differ (adjusted difference -1.8, 95% confidence limits -6.2 to 2.6, P = 0.41), but a 30% or greater reduction was more common on propranolol (69.0% against 52.6%, number needed to treat 6.1, P = 0.03), as was a 50% reduction (number needed to treat 6.1, P = 0.03). Fatigue, dizziness and sleep disorder were more frequent on the drug.

82. 82
    Slade GD, Fillingim RB, Ohrbach R, Hadgraft H, Willis J, Arbes SJ, Tchivileva IE. Genotype and Efficacy of Propranolol for TMD Pain: A Randomized Trial.J Dent Res · 2021 · 100(2):163-170 (online 8 Oct 2020)doi:10.1177/0022034520962733 · PMID 33030089

    Prespecified genotype analysis of 143 non-Hispanic White participants in the randomized placebo-controlled propranolol trial (ClinicalTrials.gov NCT02437383). Mean reduction in the facial pain index did not differ by genotype (P = 0.277), but for a 30% or greater reduction at week 9 the odds ratio favoring propranolol was 10.9 (95% confidence limits 2.4-50.7) in G:G homozygotes against 0.8 (0.2-3.2) in A:A homozygotes, interaction P = 0.035, confirmed on cumulative response curves (difference in area under the curve 43.7, 15.4-72.1, P = 0.003, against 6.5, -30.2 to 43.2). The authors state that the antagonistic effect of the A allele was the opposite of the synergistic effect they had hypothesized in advance.

83. 83
    Elsawy NA, Hassan M, Ghazala RA, Hamed R, Elnemr R. Neuropathic Pain in Female Patients With Fibromyalgia: Clinical, Electrodiagnostic, and Genetic Aspects.Clin J Pain · 2025 · 41(10)doi:10.1097/AJP.0000000000001311 · PMID 40662353

    Sixty women with fibromyalgia and 60 matched controls. The AA genotype went with higher pain severity (p = 0.013 and 0.019 on two scales) and both AA and AG with higher neuropathic pain scores than GG. A single-center case-control study of 120 people; consistent with the fibromyalgia signal in the meta-analysis, not independent confirmation of a mechanism.

84. 84
    Ün S, Sabırlı R, Türkçüer İ, Lengerova G, Bozhkova M, Petrov S, Köseler A. Associations of OPRM1, COMT and ABCB1 Variants with Opioid Analgesic Response in Acute Renal Colic: A Candidate-Gene Study.Pharmaceuticals (Basel) · 2026 · 19(9):1343doi:10.3390/ph19091343 · PMID 42797389

    150 patients with acute renal colic and 100 controls. No significant difference in rs4680 genotype or allele frequency between patients and controls; the only adjusted association was at the drug-transporter gene, which the authors themselves call preliminary and hypothesis-generating. A recent negative result for COMT in acute clinical pain.

85. 85
    Lenders JWM, Duh QY, Eisenhofer G, Gimenez-Roqueplo AP, Grebe SKG, Murad MH, Naruse M, Pacak K, Young WF. Pheochromocytoma and paraganglioma: an endocrine society clinical practice guideline.J Clin Endocrinol Metab · 2014 · 99(6):1915-42doi:10.1210/jc.2014-1498 · PMID 24893135

    Recommends initial biochemical testing with plasma free or urinary fractionated metanephrines, warns that pre-analytical factors cause false positives and negatives, and requires follow-up of all positive results. A guideline graded with GRADE, the standard system for rating how certain a body of evidence is; the task force reports no corporate funding.

86. 86
    Därr R, Pamporaki C, Peitzsch M, et al. Biochemical diagnosis of phaeochromocytoma using plasma-free normetanephrine, metanephrine and methoxytyramine: importance of supine sampling under fasting conditions.Clin Endocrinol (Oxf) · 2014 · 80(4):478-86 (online 17 Oct 2013)doi:10.1111/cen.12327 · PMID 24102244

    762 patients at six centers. Seated non-fasting sampling raised normetanephrine and methoxytyramine 49-51% in people without the tumor; upper reference limits 1.29 versus 0.79, 0.49 versus 0.41 and 0.42 versus 0.18 nmol/L; sensitivity fell from 98% to 85% with seated cutoffs, specificity from 95% to 71% with supine cutoffs on seated samples, a 5.7-fold rise in false positives.

87. 87
    Olanow CW. Tolcapone and hepatotoxic effects. Tasmar Advisory Panel.Arch Neurol · 2000 · 57(2):263-7doi:10.1001/archneur.57.2.263 · PMID 10681087

    Panel review of four patients with severe hepatic dysfunction. Reports that liver dysfunction did not develop in any patient monitored per the original prescribing information and that virtually all abnormalities appeared within six months, and recommends less frequent monitoring thereafter. A consensus panel, not a new study.

88. 88
    Stocchi F, De Pandis MF. Utility of tolcapone in fluctuating Parkinson's disease.Clin Interv Aging · 2006 · 1(4):317-25doi:10.2147/ciia.2006.1.4.317 · PMID 18046910

    Review recording the European reintroduction and reduced United States monitoring requirements, and noting that tolcapone has central as well as peripheral effects and that in a switch study it gave greater on time than remaining on entacapone. A narrative review, so the switch comparison is reported, not independently verified here.

89. 89
    Fabbri M, Rosa MM, Ferreira JJ. Clinical pharmacology review of opicapone for the treatment of Parkinson's disease.Neurodegener Dis Manag · 2016 · 6(5):349-62doi:10.2217/nmt-2016-0022 · PMID 27599671

    States that in June 2016 the European Commission granted a marketing authorization valid throughout the European Union for opicapone as an adjunct to levodopa decarboxylase inhibitors in adults with Parkinson's disease and end-of-dose motor fluctuations. A narrative clinical pharmacology review, cited for the authorization and its date rather than for a trial result.

90. 90
    Craft BM, Baker DE, Levien TL. Opicapone: Once-Daily COMT Inhibitor for the Treatment of Wearing Off in Parkinson's Disease.Sr Care Pharm · 2022 · 37(2):55-61doi:10.4140/TCP.n.2022.55 · PMID 35082010

    Drug review stating that the United States Food and Drug Administration approved opicapone in combination with levodopa-carbidopa for wearing off in Parkinson's disease on 24 April 2020. A narrative review, cited for the approval and its date.

91. 91
    Ferreira JJ, Stocchi F, Antonini A, Rascol O, Ebersbach G, Kulisevsky J, Fonseca MM, Ramos D, Rocha JF, Brigas HC, Holenz J, Poewe W. Early start of opicapone in Parkinson's disease: evidence from a pooled analysis of phase 3 trials for sustained benefit in patients with recent onset of motor fluctuations.Front Neurol · 2025 · 16:1715748doi:10.3389/fneur.2025.1715748 · PMID 41473225

    Post-hoc pooled subgroup analysis of 227 patients whose fluctuations began within two years: off time -65.6 minutes versus placebo (95% confidence interval -105.5 to -25.6, p = 0.0014), good on time +88.3 minutes (47.0 to 129.6, p < 0.0001). The long-term advantage of early over late start was numerical only (p = 0.2 and 0.09). Post hoc and manufacturer-authored, so hypothesis-generating.

92. 92
    De Carolis L, Bianchini E, Alborghetti M, et al. The safety and tolerability of opicapone in elderly Parkinson's disease patients: a real-world, multicenter, retrospective study.J Neural Transm (Vienna) · 2026 · published online 3 October 2026doi:10.1007/s00702-026-03284-z · PMID 42829398

    101 patients aged 65 and over followed a mean of 20.6 months. Adverse events in 48.5%, discontinuation in 28.7% within a mean of 9.4 months, most often for neuropsychiatric and cardiovascular or autonomic reasons; disease stage was the only baseline feature independently associated with adverse events. Retrospective and uncontrolled, but closer to routine practice than the trials.

93. 93
    Zhu R, Wang P, Qi SL, Zhang Y, Xie CL, Yang Y, Hu C, Zhou ZQ, Zhao DF, Yang C, Sun J, Yang XW, Ge GB, Wang P. Discovery and optimization of a novel non-nitrocatechol COMT inhibitor for modulating levodopa metabolism.J Pharm Anal · 16(6):101520 (online 11 Dec 2025)doi:10.1016/j.jpha.2025.101520 · PMID 42434335

    Flavonoid screening identified baicalein as a lead; optimization produced a competitive non-nitrocatechol inhibitor binding the catechol pocket with a constant of 89.28 nanomolar, 26-fold better cellular inhibition and about tenfold better metabolic stability, permeability and oral bioavailability, with effects on levodopa metabolism in rats. Preclinical: cells and rats, no human data.

94. 94
    Ding H, Fu Y, Chen W, Wang Z. COMT Val158Met polymorphism and breast cancer risk: evidence from 26 case-control studies.Breast Cancer Res Treat · 2010 · 123(1):265-70doi:10.1007/s10549-010-0759-5 · PMID 20130981

    26 studies, 16,693 cases, 18,261 controls. No significant association in any genetic model overall or in Asian or European populations; after excluding studies violating Hardy-Weinberg equilibrium, a borderline decreased risk in Europeans (recessive model odds ratio 0.90). The abstract prints the confidence interval as 0.90-1.00, a lower bound equal to the point estimate, which appears to be a typographical error.

95. 95
    He XF, Wei W, Li SX, Su J, Zhang Y, Ye XH, Liu Y, Wang W. Association between the COMT Val158Met polymorphism and breast cancer risk: a meta-analysis of 30,199 cases and 38,922 controls.Mol Biol Rep · 2012 · 39(6):6811-23doi:10.1007/s11033-012-1506-2 · PMID 22297695

    No significant association in the dominant (odds ratio 0.99, 0.94-1.04), recessive (0.97, 0.92-1.02) or additive (0.98, 0.91-1.05) models, with significant heterogeneity. In Caucasians the recessive model gave 0.96 (0.92-1.00), a decrease, which is the opposite direction to the catechol estrogen hypothesis. The largest breast cancer meta-analysis of this variant.

96. 96
    Tian C, Liu L, Yang X, Wu H, Ouyang Q. The Val158Met polymorphism in the COMT gene is associated with increased cancer risks in Chinese population.Tumour Biol · 2014 · 35(4):3003-8 (online 5 Dec 2013)doi:10.1007/s13277-013-1387-6 · PMID 24307619

    18 case-control studies, 5,034 cases and 6,234 controls, restricted to Chinese populations. Significant associations in three models (odds ratios 1.34, 1.04-1.73; 1.39, 1.06-1.82; 1.13, 1.01-1.27), with the signal in breast cancer and esophageal squamous cell carcinoma. A population-specific result from a smaller literature than the two large meta-analyses, and in the opposite direction.

97. 97
    Kallionpää RA, Uusitalo E, Peltonen J. Association of Catechol-O-methyltransferase polymorphism Val158Met and mammographic density: A meta-analysis.Gene · 2017 · 624:34-42doi:10.1016/j.gene.2017.04.049 · PMID 28473194

    Eight studies. Overall effect on percent mammographic density in the recessive model -1.41 (95% confidence interval -2.86 to 0.05; P = 0.06), rising to -1.93 (-3.49 to -0.37; P = 0.02) when breast cancer patients were excluded. The authors note opposite apparent effects in postmenopausal hormone users versus other women, so menopausal status and hormone therapy confound the comparison.

98. 98
    Aguiar MP, Gomes MO, Ananias LF, Silva-Martins LF, Espindula AP. Impact of single nucleotide variants in estrogen genes on ovarian cancer risk: a systematic review and meta-analysis.Endocr Oncol · 2025 · 5(1):e250007doi:10.1530/EO-25-0007 · PMID 41158532

    Thirty studies. The A allele of COMT rs4680 was associated with reduced ovarian cancer risk (odds ratio 0.85, 0.73-0.99, p = 0.0373). The authors state that the limited number of studies and population heterogeneity may affect robustness and generalizability; registered in PROSPERO.

99. 99
    Hall KT, Buring JE, Mukamal KJ, Vinayaga Moorthy M, Wayne PM, Kaptchuk TJ, Battinelli EM, Ridker PM, Sesso HD, Weinstein SJ, Albanes D, Cook NR, Chasman DI. COMT and Alpha-Tocopherol Effects in Cancer Prevention: Gene-Supplement Interactions in Two Randomized Clinical Trials.J Natl Cancer Inst · 2019 · 111(7):684-694doi:10.1093/jnci/djy204 · PMID 30624689

    Pooling the Women's Genome Health Study (23,294) and an ATBC case-control subset: alpha-tocopherol versus placebo gave hazard ratios 0.88 (0.80-0.97, p = .01) in met/met, 0.99 (0.92-1.06) in val/met and 1.18 (1.06-1.31, p = .002) in val/val, interaction p < .001. From randomized trials, which is strong; but a genetic subgroup analysis of trials not designed for the question.

100. 100
     Venter G, van der Berg CL, Jacobs T, van der Westhuizen FH, Erasmus E. Oral contraceptives containing ethinyl estradiol and drospirenone increase hydroxylation and methylation of endogenous estrogen but not genotoxic estrogen DNA-adduct formation.Sci Rep · 2025 · 15(1):31468doi:10.1038/s41598-025-16892-8 · PMID 40858744

     24 contraceptive users and 25 controls, urinary and serum metabolites by liquid chromatography-tandem mass spectrometry. Lower free estradiol and estrone, higher 2-methoxyestrone, higher methylated-to-parent ratios, lower serum betaine, and no increase in estrogen-DNA adducts. Small and cross-sectional, but it measures the pathway directly rather than inferring it from genotype.

101. 101
     Xie T, Ho SL, Ramsden D. Characterization and implications of estrogenic down-regulation of human catechol-O-methyltransferase gene transcription.Mol Pharmacol · 1999 · 56(1):31-8doi:10.1124/mol.56.1.31 · PMID 10385681

     17-beta-estradiol at a billionth to a ten-millionth of a mole per liter decreased the 1.3-kilobase COMT messenger RNA in MCF-7 cells, time- and dose-dependently and through the estrogen receptor; deletion analysis localized the effect to a 280-base-pair proximal fragment with two half-palindromic response elements and a 323-base-pair distal fragment with two binding-protein sites. Cell lines and reporter constructs, not human tissue.

102. 102
     Strom NI, Gerring ZF, Galimberti M, Yu D, Halvorsen MW, Abdellaoui A, et al. Genome-wide analyses identify 30 loci associated with obsessive-compulsive disorder.Nat Genet · 2025 · 57(6):1389-1401 (online 13 May 2025)doi:10.1038/s41588-025-02189-z · PMID 40360802

     Genome-wide association meta-analysis of 53,660 cases and 2,044,417 controls: 30 independent genome-wide significant loci, 249 potential effector genes with 25 classed as most likely causal, and an estimated 11,500 variants explaining 90% of the genetic heritability. Risk was enriched in excitatory neurons of the hippocampus and cortex and in dopamine-receptor-bearing medium spiny neurons. COMT is not among the genes it prioritizes; absence from a prioritized list is weaker evidence than a direct test.

103. 103
     Wu AH, Tseng CC, Van Den Berg D, Yu MC. Tea intake, COMT genotype, and breast cancer in Asian-American women.Cancer Res · 2003 · 63(21):7526-9PMID 14612555

     589 incident cases and 563 population-based controls among Chinese-, Japanese- and Filipino-American women in Los Angeles County. Among women with at least one low-activity allele, tea drinkers had an adjusted odds ratio of 0.48 (0.29-0.77); among high-activity homozygotes, 1.02 (0.66-1.60). A single case-control study with a biologically coherent interaction, not a trial.

104. 104
     Shrubsole MJ, Lu W, Chen Z, Shu XO, Zheng Y, Dai Q, Cai Q, Gu K, Ruan ZX, Gao YT, Zheng W. Drinking green tea modestly reduces breast cancer risk.J Nutr · 2009 · 139(2):310-6doi:10.3945/jn.108.098699 · PMID 19074205

     Population-based case-control study in Shanghai, 1996-2005: 3,454 incident breast cancer cases and 3,474 controls aged 20-74. Regular green tea drinking carried an odds ratio of 0.88 (95% confidence interval 0.79-0.98) against non-drinkers. The authors state that COMT rs4680 genotypes did not have a modifying effect on the association. Six times the size of the Los Angeles study it contradicts on that point.

105. 105
     Iwasaki M, Mizusawa J, Kasuga Y, Yokoyama S, Onuma H, Nishimura H, Kusama R, Tsugane S. Green tea consumption and breast cancer risk in Japanese women: a case-control study.Nutr Cancer · 2014 · 66(1):57-67 (online 25 Nov 2013)doi:10.1080/01635581.2014.847963 · PMID 24274352

     Hospital-based case-control study in Nagano, Japan, 369 matched pairs with food frequency questionnaires and blood samples genotyped for four variants including COMT rs4680. No inverse association with green tea (adjusted odds ratio 1.27, 95% confidence interval 0.75-2.14 for more than 600 mL per day against less than 120 mL; P for trend 0.20) and no substantial effect modification by any of the four variants.

106. 106
     Samavat H, Dostal AM, Wang R, Bedell S, Emory TH, Ursin G, Torkelson CJ, Gross MD, Le CT, Yu MC, Yang CS, Yee D, Wu AH, Yuan JM, Kurzer MS. The Minnesota Green Tea Trial (MGTT), a randomized controlled trial of the efficacy of green tea extract on biomarkers of breast cancer risk: study rationale, design, methods, and participant characteristics.Cancer Causes Control · 2015 · 26(10):1405-19doi:10.1007/s10552-015-0632-2 · PMID 26206423

     The design paper. 1,075 healthy postmenopausal women with dense breast tissue randomized to a green tea catechin extract containing 843.0 plus or minus 44.0 mg per day of epigallocatechin gallate or placebo for 12 months, with randomization stratified by COMT genotype activity, high against low or intermediate. 937 women completed; overall dropout 12.8%. This is the trial built to test the COMT-catechin interaction directly.

107. 107
     Samavat H, Ursin G, Emory TH, Lee E, Wang R, Torkelson CJ, Dostal AM, Swenson K, Le CT, Yang CS, Yu MC, Yee D, Wu AH, Yuan JM, Kurzer MS. A Randomized Controlled Trial of Green Tea Extract Supplementation and Mammographic Density in Postmenopausal Women at Increased Risk of Breast Cancer.Cancer Prev Res (Phila) · 2017 · 10(12):710-718 (online 13 Sep 2017)doi:10.1158/1940-6207.CAPR-17-0187 · PMID 28904061

     Mammographic density outcome of the Minnesota Green Tea Trial, assessed in 932 completers (462 extract, 470 placebo). Twelve months of four decaffeinated capsules a day, 1,315 mg total catechins including 843 mg epigallocatechin gallate, did not significantly change percent or absolute mammographic density in all women, and did not change it within subgroups defined by COMT genotype or body mass index. In women aged 50-55 percent density fell by 4.40% against placebo (P = 0.05).

108. 108
     Bathgate JR, Radler DR, Kurzer M, Samavat H. Green tea extract supplementation does not modify plasma concentration of F2-isoprostanes in women who are postmenopause: Findings from a randomized controlled trial.Nutr Res · 2023 · 113:29-38doi:10.1016/j.nutres.2023.03.001 · PMID 37011435

     Secondary analysis of the Minnesota Green Tea Trial. Twelve months of 843 mg per day of epigallocatechin gallate did not significantly change plasma F2-isoprostanes, a marker of oxidative stress (P for overall treatment 0.07), and COMT genotype did not modify the effect (P = 0.85).

109. 109
     Cunningham A, Gomes A, Meng L, Shapses S, Byham-Gray L, Samavat H. Effects of Green Tea Extract Supplementation on Inflammatory Cytokines Among Postmenopausal Women with Overweight or Obesity-A Secondary Analysis of a Randomized Controlled Trial.Nutrients · 2026 · 18(1):143 (online 1 Jan 2026)doi:10.3390/nu18010143 · PMID 41515260

     Secondary analysis of a random subset of 97 Minnesota Green Tea Trial participants with overweight or obesity. Twelve months of the extract did not significantly change C-reactive protein (P = 0.24), interleukin-6 (P = 0.59) or tumor necrosis factor alpha (P = 0.36), and there was no significant interaction between treatment group and COMT genotype for any of them (all P above 0.40). A small subset, so a null result with wide limits.

110. 110
     Acosta L, Byham-Gray L, Kurzer M, Samavat H. Hepatotoxicity with High-Dose Green Tea Extract: Effect of COMT and UGT1A Genotypes.J Diet Suppl · 2023 · 20(6):850-869 (online 30 Sep 2022)doi:10.1080/19390211.2022.2128501 · PMID 36178169

     Secondary analysis of 799 Minnesota Green Tea Trial participants, 400 on the extract. The only COMT signal was between the two variant genotypes: from baseline to month 3 the mean aspartate-to-alanine aminotransferase ratio rose 1.0% in rs4680 heterozygotes and fell 4.8% in methionine homozygotes (P = 0.03), with neither differing from valine homozygotes, and the authors call the difference clinically negligible. A variant in a different gene, UGT1A rs6755571, was associated with much larger alanine aminotransferase rises at months 6 and 9 (78.1% and 82.1% against 28.0% and 30.1%, P < 0.001 and P = 0.004).

111. 111
     Russo EB, Whiteley VL. Cannabinoid hyperemesis syndrome: genetic susceptibility to toxic exposure.Front Toxicol · 2024 · 6:1465728doi:10.3389/ftox.2024.1465728 · PMID 39507417

     Review by the same group restating the five variants as affecting TRPV1, two dopamine genes, CYP2C9 and an ATP-binding cassette transporter, and rejecting pesticide and neem explanations. A narrative review of the authors' own prior work; no new data and no independent replication.

112. 112
     Hasler WL, Li BUK, Levinthal DJ, Venkatesan T. Cyclic vomiting syndrome: Future clinical and research priorities for: Special supplement/proceedings of 3rd international symposium.Neurogastroenterol Motil · 2025 · 37(3):e14825doi:10.1111/nmo.14825 · PMID 38775195

     Consensus proceedings placing genetic abnormalities among several contributors to a multifactorial pathophysiology for cyclic vomiting syndrome, alongside migraine, dysautonomia, endogenous cannabinoids, mitochondrial dysfunction and rapid gastric emptying, and calling for multicenter controlled trials. Expert consensus, not primary data.

113. 113
     Apud JA, Mattay V, Chen J, Kolachana BS, Callicott JH, Rasetti R, Alce G, Iudicello JE, Akbar N, Egan MF, Goldberg TE, Weinberger DR. Tolcapone improves cognition and cortical information processing in normal human subjects.Neuropsychopharmacology · 2007 · 32(5):1011-20 (online 25 Oct 2006)doi:10.1038/sj.npp.1301227 · PMID 17063156

     Randomized, double-blind, placebo-controlled crossover trial of the brain-penetrant COMT inhibitor tolcapone in 47 healthy volunteers stratified by Val158Met genotype, with functional magnetic resonance imaging in 34. Significant drug effects on executive function and verbal episodic memory, and a significant drug-by-genotype interaction on the latter: valine homozygotes improved, methionine homozygotes worsened. A proof-of-concept study in healthy people, not a treatment trial.

114. 114
     Inoue-Choi M, Yuan JM, Yang CS, Van Den Berg DJ, Lee MJ, Gao YT, Yu MC. Genetic Association Between the COMT Genotype and Urinary Levels of Tea Polyphenols and Their Metabolites among Daily Green Tea Drinkers.Int J Mol Epidemiol Genet · 2010 · 1(2):114-123PMID 21191472

     Cross-sectional analysis within the Shanghai Cohort Study of 18,244 men: 660 self-identified daily green tea drinkers genotyped at rs4680, with five urinary tea polyphenol metabolites measured and expressed per gram of creatinine. Men homozygous for the low-activity allele excreted significantly less of each metabolite and of all five together than high-activity homozygotes or heterozygotes, whose levels were comparable. A measurement of excretion in tea drinkers, not of enzyme activity, and in men only.

115. 115
     Tunbridge EM, Carr GV, Weinberger DR. The therapeutic potential of COMT inhibition: cognition and beyond.Mol Psychiatry · published online 19 August 2026doi:10.1038/s41380-026-03824-9 · PMID 42618604

     Review of the evidence linking COMT to executive function, centered on human studies with the brain-penetrant inhibitor tolcapone. States that multiple studies, although small in scale, show working memory improving after COMT inhibition in a manner dependent on baseline prefrontal dopamine signaling, consistent with the inverted-U relationship, and that the effects extend to impulsive and risky behavior, with rodent data pointing the same way. Commercial interest: the first author's affiliation on this paper is Boehringer Ingelheim Pharma GmbH and Co. KG.

116. 116
     Ojeda-Lepe E, García-Díaz S, Muñoz-Delgado L, Martín-Bórnez M, Jesús S, Buiza-Rueda D, Bonilla-Toribio M, Díaz-Belloso R, Macías-García D, Adarmes-Gómez A, Luque-Ambrosiani A, Pineda-Sánchez R, Clavijo-Jiménez L, Carrillo-García F, Gómez-Garre P, Mir P. Effect of COMT and UGT1A Variants on Clinical Response to Opicapone in Parkinson's Disease.Mov Disord Clin Pract · published online 20 July 2026doi:10.1002/mdc3.70751 · PMID 42473901

     Retrospective single-center study of 237 patients treated with opicapone between May 2017 and June 2025, genotyped at COMT rs4818, rs4680, rs4633 and rs6269 and at UGT1A rs1105880. Heterozygosity at rs4818 (odds ratio 2.55) and rs6269 (2.63) and the low-activity COMT haplotype (2.19) were associated with improvement in motor fluctuations after correction for multiple comparisons, as was the UGT1A rs1105880 dominant model (2.41). No variant was associated with stopping the drug. Retrospective, with clinical records rather than rating scales and no pharmacokinetic data; the authors call for prospective validation. Several authors report honoraria from Bial, which makes opicapone.