Borage

What is Borage?

Borage is Borago officinalis L., an annual in the Boraginaceae family with bristly leaves, cucumber-flavored foliage and blue star-shaped flowers. Two commercially distinct products carry the name, and confusing them is the most consequential mistake a buyer can make. Borage seed oil is the richest common plant source of gamma-linolenic acid, or GLA, at 20 to 26 percent of total fatty acids. Borage aerial parts, the leaf and flower, are sold as tea and as a garden vegetable in Spain and Italy, and they carry hepatotoxic pyrrolizidine alkaloids that a properly refined seed oil does not.

Common Names

• Starflower and starflower oil, the standard UK retail names for the seed oil

• Bee bread, bee plant, cool tankard, tailwort

• Borraja (Spanish), bourrache (French), Borretsch (German), borragine (Italian)

Primary Active Compounds

• Gamma-linolenic acid (GLA, 18:3 n-6), 20 to 26 percent of the seed oil. Health Canada permits standardization from 18 to 27 percent

• Linoleic acid 35 to 42 percent, oleic 15 to 20, palmitic 9 to 12, stearic 3 to 4, and erucic acid 2 to 3 percent, high enough that the EU wrote borage oil a specific exemption from its erucic acid ceiling

• Pyrrolizidine alkaloids in leaf and flower: amabiline, supinine, lycopsamine, intermedine, and the acetylated forms acetyllycopsamine and acetylintermedine. Thesinine, also present, is saturated and not hepatotoxic

Key Note

GLA is not an essential nutrient. Your body makes it from dietary linoleic acid using delta-6-desaturase (FADS2), and most Americans eat 12 to 17 grams of linoleic acid daily. The commercial case rests entirely on preformed GLA doing something endogenous synthesis does not.

What follows separates two products sharing a plant name: one with a large and mostly negative evidence base, one with a liver toxin and no evidence.

Primary Functions & Benefits

Delivers preformed GLA, bypassing delta-6-desaturase

Delta-6-desaturase is the rate-limiting step converting dietary linoleic acid into the anti-inflammatory branch of omega-6 metabolism. Swallowing GLA skips it. GLA is elongated to dihomo-gamma-linolenic acid (DGLA, 20:3 n-6), and tissue DGLA does rise: Barham and colleagues, Journal of Nutrition 2000, volume 130, pages 1925 to 1931, showed neutrophil DGLA rising significantly across 3 weeks, and Henz's 1999 borage trial found erythrocyte DGLA increases in treated patients only.

Shifts eicosanoid output, in principle

DGLA feeds cyclooxygenase to make prostaglandin E1 and thromboxane A1, and 15-lipoxygenase to make 15-HETrE, which inhibits 5-lipoxygenase and therefore leukotriene B4. That mechanistic case has produced two positive rheumatoid arthritis trials, 27 negative eczema trials, and one unreplicated diabetic neuropathy result. Borage aerial parts have no controlled human efficacy evidence for any indication.

Forms & Standardization

The difference between two bottles labeled "borage oil" can be a factor of 30,000 in liver toxin content.

Refined seed oil

Degumming, alkali neutralization, bleaching and deodorization are what remove pyrrolizidine alkaloids. Wretensjö and Karlberg, Journal of the American Oil Chemists' Society 2003, volume 80, pages 963 to 970, spiked crude oil with crotaline and tracked it through each stage: pyrrolizidine content fell roughly 30,000-fold, with none detectable above a 20 ppb limit. Refining is the safety step.

Cold-pressed and expeller-pressed seed oil

Marketed as the premium option, and for PAs it is the riskier one: no alkali wash, no deodorizer. The USP-NF monograph for Borage Seed Oil Capsules defines the article as oil derived by "cold pressing or supercritical fluid extraction" and requires not less than 95.0 percent of labeled gamma-linolenic, linoleic and oleic acids. That is a fatty acid specification, not a pyrrolizidine alkaloid one. A capsule can be compendially compliant and never have been tested for the compound that can kill you.

GLA percentages actually sold

• 22 to 24 percent is the standard retail softgel grade. A 1,000 mg softgel labeled 240 mg GLA is 24 percent

• Bulk commodity oil runs 20 percent, and 27 percent is the top of Health Canada's permitted range

Other forms

Zurier's 1996 RA trial used GLA as the free fatty acid, which no retail bottle contains. Blackcurrant seed oil runs 15 to 20 percent GLA, evening primrose 8 to 10 percent, hemp seed 2 to 4 percent, which is why 500 mg of borage and 500 mg of evening primrose are not interchangeable and why so much of the eczema literature is evening primrose. Aerial part preparations are not refined at all, so everything refining removes from seed oil is present in them.

Food Sources

GLA is effectively absent from the ordinary food supply, which makes this one of the few sections in this archive where "eat the food instead" is not available. Human milk carries GLA at a few tenths of one percent of total fatty acids, and that exposure ends at weaning. Oats and barley carry only traces.

Borage as an actual vegetable

Borage leaves are eaten. In Aragón and Navarre, borraja is boiled and served with potatoes; in Liguria it goes into preboggion, the greens mixture filling pansoti. These are aerial parts, and therefore the PA-containing parts.

The nitrate problem in the culinary plant

Martinez and colleagues, Journal of Pediatric Gastroenterology and Nutrition 2013, volume 56, pages 573 to 577, reported 78 cases of diet-induced methemoglobinemia in infants in Pamplona between 1987 and 2010. Borage was the strongest vegetable association, odds ratio 5.2, 95 percent CI 1.1 to 24.6, and nitrate averaged 3,968 mg/kg across 15 borage samples, the highest of any vegetable tested, ahead of chard at 2,811 mg/kg. Storage mattered as much: a purée held 24 to 48 hours carried an odds ratio of 17.4, beyond 48 hours 24.9.

You cannot get a supplemental GLA dose from food, and the one food form of borage carries its own hazard. This is a pill-or-nothing compound, which is why the trial evidence matters.

Who Should Take Borage

For most people the honest answer is no one. The narrow exceptions:

• People with active rheumatoid arthritis, as an adjunct to disease-modifying therapy, at trial dose, with a rheumatologist. This is the only indication with two positive randomized trials behind it, at a dose far above what any bottle suggests

• People already committed to a GLA oil who want the most concentrated one. Borage at 24 percent delivers 2.7 times the GLA per gram that evening primrose at 9 percent does, an advantage in capsule burden that says nothing about whether GLA works

• People buying it as a topical emollient, which raises none of the systemic questions here

Nobody should take it for eczema; 27 trials and 1,596 participants say so. Nobody should take the aerial parts internally, where there is no efficacy evidence and a liver toxin.

Who Should AVOID or Use Caution

Contraindications

• Pregnancy and breastfeeding. PAs cross the placenta and are excreted in milk

• Liver disease, cirrhosis, hepatitis, or any history of drug-induced liver injury

• Infants and young children, for the aerial parts and for borage purée specifically

• Seizure disorder. Al-Khamees and colleagues, Journal of Medical Toxicology 2011, volume 7, pages 154 to 157, reported status epilepticus in a patient who had taken borage oil for one week. One case report is not proof of causation, but with an anticonvulsant regimen at stake it is not a risk worth taking for an oil that failed its main indication

Use Caution

• On warfarin or any anticoagulant. A study of evening primrose GLA found 9 of 12 patients had significantly increased bleeding time, and borage oil is 2 to 3 times more concentrated in GLA

• On antiplatelet agents including aspirin and clopidogrel. Stop at least 2 weeks before surgery

• On phenothiazines, where the concern is a lowered seizure threshold, and on hepatotoxic drugs including ketoconazole, methotrexate and anabolic steroids. That matters most for aerial parts, but a cold-pressed oil of unverified PA status belongs in the same conversation

• On regular NSAIDs, which inhibit the cyclooxygenase step converting DGLA to prostaglandin E1

Critical Safety Point

Pyrrolizidine alkaloid hepatic veno-occlusive disease, now usually called hepatic sinusoidal obstruction syndrome, is irreversible. Chen and Huo, Netherlands Journal of Medicine 2010, volume 68, pages 252 to 260, state that no uniformly effective therapy has been established and treatment is largely supportive. The small hepatic venules are obliterated, portal hypertension follows, and severe cases end at transplant. Damage accumulates with cumulative exposure, which is why a small daily amount over years is the shape of the problem rather than a single large dose.

Recommended Dosages

The gap between what the trials used and what a bottle tells you is the most useful number in this issue.

Rheumatoid arthritis, trial doses

• Leventhal, Annals of Internal Medicine 1993, volume 119, pages 867 to 873: 1.4 g/day of GLA in borage seed oil, 24 weeks

• Zurier, Arthritis & Rheumatism 1996, volume 39, pages 1808 to 1817: 2.8 g/day of GLA as free fatty acid, 6 months double-blind

What that means in softgels

Take the standard retail product, a 1,000 mg borage oil softgel delivering 240 mg of GLA, 24 percent by weight. Leventhal's 1.4 g of GLA is 1,400 divided by 240, which is 5.8 softgels a day, about 5.8 grams of oil. Zurier's 2.8 g is 11.7 softgels, about 11.7 grams of oil. At 22 percent GLA those requirements rise to 6.4 and 12.7 grams of oil.

A 60-count bottle dosed as the label suggests lasts 60 days. Dosed at Leventhal's level it lasts 10 days, at Zurier's 5. One softgel daily delivers 17 percent of the lower trial dose and 8.6 percent of the higher. Nothing in the RA literature supports 240 mg of GLA a day, and that is what most bottles put in front of you.

Eczema, trial doses

Takwale 2003 gave 920 mg GLA/day for 12 weeks. Henz 1999 gave 500 mg of borage oil daily at at least 23 percent GLA, roughly 115 mg GLA, for 24 weeks. Takwale ran nearly 8 times the GLA of Henz and still found nothing, so this is not a dose-response failure. It is a null effect.

Regulatory reference dose

Health Canada's Natural Health Product monograph permits 3.7 to 5 grams of oil per day for adults 18 and over. At 18 to 27 percent GLA that is 666 to 1,350 mg of GLA, so even the top of Canada's range falls short of Leventhal's 1.4 g and is under half of Zurier's 2.8 g. The permitted claims are "source of essential fatty acids" and "source of omega-6 fatty acids for the maintenance of good health." No eczema claim, no arthritis claim.

Aerial parts and duration

No defensible aerial-part dose exists. Traditional texts describe 2 g of dried herb infused three times daily, a PA exposure with no efficacy evidence attached. For the oil, RA trials ran 24 weeks and 12 months, eczema trials 12 to 24 weeks.

Timing & Administration

With food, always. Bile salt release and pancreatic lipase activity track meal fat, so a fat-containing meal improves absorption of this triglyceride and sharply reduces belching and reflux.

Split the dose if you are taking a real amount. At 6 to 12 softgels daily, taking them at once reliably produces nausea and loose stools. Both RA trials split dosing.

Refrigerate after opening. Borage oil has an iodine value around 135 to 155 g I2/100 g, a measure of how many double bonds are available to oxidize. Specification-grade oil shows a peroxide value at or below 10 meq O2/kg and an acid value at or below 3.0 mg KOH/g, and neither stays put in a warm cupboard.

Separate from NSAID doses by a few hours, since the mechanism runs through cyclooxygenase. Stop 2 weeks before surgery.

Timeline of Effects

• Days 1 to 7. No clinical change, measurable biochemistry: plasma and neutrophil DGLA begin rising within days

• Weeks 2 to 4. Adverse effects arrive before benefits. Belching, soft stools and mild nausea appear here if at all, and in the Cochrane eczema review they occurred at similar rates in oil and placebo arms

• Weeks 6 to 12. The earliest point any efficacy signal appeared in any trial. Zurier's continuous-GLA patients were still improving at 12 months, 16 of 21 better than at entry

• Week 12. A fair trial for eczema, and the answer is known. Takwale's primary endpoint came in at 1.4 points in favor of placebo, 95 percent CI -2.2 to 5.0, P = 0.45

• Weeks 24 to 52. A fair trial for joint symptoms at trial dose. Running 240 mg a day for 4 weeks and concluding it does not work is not a test of anything

Benefits of Taking Borage

Rheumatoid arthritis symptom relief, at high dose, moderate evidence

Leventhal's 37 patients on 1.4 g/day of GLA in borage seed oil against a cotton seed oil placebo showed a 36 percent reduction in tender joint count, 45 percent in tender joint score, 28 percent in swollen joint count and 41 percent in swollen joint score, all at P < 0.05, while placebo showed no change or worsening. Zurier's 56 patients on 2.8 g/day produced meaningful responses, at least 25 percent improvement in 4 measures, in 14 of 22 versus 4 of 19 on placebo, P = 0.015.

Diabetic neuropathy, one unreplicated trial. Keen and colleagues, Diabetes Care 1993, volume 16, pages 8 to 15: 111 patients, 480 mg GLA daily for a year, 13 of 16 nerve conduction and sensory parameters better than placebo. Evening primrose source, unconfirmed in 33 years.

What is not a benefit. Eczema and atopic dermatitis, per 27 trials. Asthma, where the only supporting study, Lindemann 2009, tested a GLA-plus-EPA medical food in 21 subjects for 28 days and did not improve the Asthma Control Questionnaire. Premenstrual complaints, menopausal symptoms, hair, nails and cardiovascular endpoints, all without borage-specific trials.

Potential Negatives & Side Effects

Common and dose-related

• Belching, reflux and an oily aftertaste, the most frequent complaint at multi-gram doses

• Loose stools and diarrhea. In the OMEGA trial of enteral omega-3 plus GLA in 272 adults with acute lung injury, the supplemented arm reported diarrhea on 29 percent of days versus 21 percent, P = 0.001

• Nausea, abdominal discomfort, bloating, headache

Uncommon but documented

• Increased bleeding time, particularly on warfarin

• Status epilepticus in one published case after a week of use, and contact dermatitis from topical preparations

The mechanism-level negative most people never hear

Supplementing GLA raises serum arachidonic acid. Barham and colleagues state it directly: GLA supplementation causes "a marked increase in serum arachidonate levels, a potentially harmful side effect." Arachidonic acid is the substrate for the pro-inflammatory series-2 prostaglandins and series-4 leukotrienes, the opposite of the intended effect. Their fix was adding EPA, which blocks delta-5-desaturase: 3.0 g/day of GLA plus EPA raised neutrophil DGLA, cut leukotriene synthesis, and left serum arachidonic acid unchanged. Borage oil alone does none of this.

Deficiency Symptoms

Borage is not an essential nutrient and there is no borage deficiency. GLA is not essential either. The essential omega-6 is linoleic acid, and GLA sits one step downstream of it. A person on an ordinary Western diet takes in 12 to 17 grams of linoleic acid a day, roughly a hundred times more precursor than any borage capsule supplies as product.

What borage addresses

• Not a deficiency. There is no deficiency state, no blood test, no reference range and no clinical syndrome of low GLA

• Delta-6-desaturase activity declines with age and is reduced in diabetes, heavy alcohol use and zinc or magnesium insufficiency. That is the physiological argument for supplemental GLA, plausible rather than demonstrated, since no trial has selected patients by measured desaturase activity

• Genuine essential fatty acid deficiency is real, seen on fat-free parenteral nutrition as scaly dermatitis, alopecia and impaired wound healing. It is corrected with linoleic acid, not GLA

Bottom line

Nobody needs borage. The question is not whether you are short of GLA but whether extra GLA does anything, and for the largest indication tested the answer is no.

Toxicity Symptoms

At high intake, seed oil

Acute overdose is not the failure mode. In the 2011 Cochrane pooling adverse events ran 20 percent for GLA versus 3 percent for placebo, relative risk 4.24, 95 percent CI 0.78 to 22.99, not statistically significant. The meaningful concerns are bleeding on anticoagulants and the arachidonic acid rise.

At high intake, aerial parts or unrefined oil

This is the real toxicity. The 1,2-unsaturated PAs are oxidized by hepatic CYP3A4 into pyrrolic esters that alkylate proteins and DNA in the sinusoidal endothelium, obliterating the small venules. Clinically: right upper quadrant pain, hepatomegaly, ascites, jaundice, and rising transaminases and bilirubin, presenting weeks to months after exposure begins.

Signs to reduce or stop

• Right upper quadrant abdominal pain or tenderness

• Unexplained abdominal swelling or rapid weight gain

• Yellowing of skin or eyes, dark urine, pale stools

• Unusual fatigue with loss of appetite and nausea

• New or disproportionate bruising or bleeding, particularly on warfarin, or any new seizure activity

General note

Hepatic sinusoidal obstruction syndrome is irreversible. No antidote, no chelator, no wash-out, and injury accumulates with cumulative dose, which is why regulators set PA limits in micrograms per day rather than milligrams.

How Borage Works

The pathway

• Dietary linoleic acid (18:2 n-6) is desaturated by delta-6-desaturase, the FADS2 gene product, to GLA (18:3 n-6), the rate-limiting step and the one borage oil bypasses. GLA is then elongated by ELOVL5 to DGLA (20:3 n-6), a fast conversion, which is why GLA itself barely accumulates while DGLA rises measurably

• DGLA sits at a fork. Cyclooxygenase converts it to prostaglandin E1 and thromboxane A1, and 15-lipoxygenase to 15-HETrE, which inhibits 5-lipoxygenase and suppresses leukotriene B4. The competing branch is delta-5-desaturase, the FADS1 gene product, converting DGLA to arachidonic acid (20:4 n-6), substrate for series-2 prostaglandins and series-4 leukotrienes

Why the mechanism does not settle the clinical question

Every borage marketing page describes the first branch and none the second, but the balance between them depends on desaturase activity nobody measures. More to the point, every step above is real and documented: tissue DGLA goes up, leukotriene B4 goes down, and eczema does not improve. That gap between a working mechanism and a null clinical result is the most instructive thing here.

Synergistic Supplements

EPA, the one combination with actual mechanistic evidence behind it. Eicosapentaenoic acid inhibits delta-5-desaturase, stopping GLA-derived DGLA converting onward to arachidonic acid. Barham 2000 showed 3.0 g/day of combined GLA and EPA raised neutrophil DGLA and EPA, reduced leukotriene synthesis, and left serum arachidonic acid flat. Taking GLA at a real dose without omega-3 is the version the evidence argues against.

Magnesium, zinc and vitamin E. The first two are desaturase cofactors, largely irrelevant if you are swallowing preformed GLA. Vitamin E is practical: an oil with an iodine value of 135 to 155 needs an antioxidant, so check that tocopherols appear on the label.

Not synergistic: stacking borage with evening primrose and blackcurrant seed oil. Three sources of one molecule at three concentrations means buying GLA three times and paying a premium for the two weaker sources.

Interactions & What NOT to Take

• Warfarin. The clearest interaction. In a small study of evening primrose GLA, 9 of 12 patients showed significantly increased bleeding time, and borage oil carries 2 to 3 times the GLA per gram. If you start it your INR needs monitoring, and the sensible answer is not to start

• NSAIDs: ibuprofen, naproxen, celecoxib, diclofenac. Direct pharmacodynamic antagonism, since cyclooxygenase is the enzyme turning DGLA into prostaglandin E1. Regular NSAID use plausibly blocks the mechanism you are paying for, and this is the interaction most often left off product pages

• Phenothiazines: chlorpromazine, prochlorperazine, fluphenazine. Lowered seizure threshold, which with the published status epilepticus case makes this a hard avoid

• Hepatotoxic drugs: ketoconazole, methotrexate, isoniazid, anabolic steroids, amiodarone. Applies to aerial parts and to any seed oil whose PA status you cannot verify. Strong CYP3A4 inducers such as rifampicin and carbamazepine would in principle increase formation of the reactive PA metabolite, which is inference rather than demonstrated fact

Quality, Testing & Adulteration

"PA-free" is a manufacturer claim, not a certified standard

There is no accredited PA-free designation, no USP verification program for pyrrolizidine alkaloids in borage oil, no NSF certification, and no legally binding US limit. The USP-NF capsule monograph specifies fatty acid identity and content and defines the article as cold pressed or supercritical CO2 extracted, and neither method includes the alkali refining step that cuts PA content roughly 30,000-fold.

What the numbers actually are

• Wretensjö and Karlberg found no PAs above a 20 ppb detection limit in processed borage oil, so 5 grams of oil delivers 0.1 micrograms. The frequently cited alternative estimate, that 1 to 2 g of borage seed oil daily can deliver close to 10 micrograms of unsaturated PAs, implies an oil at roughly 5,000 ppb, 250 times higher. Both figures are in the literature, and the difference between them is refining

• Memorial Sloan Kettering states certified products should contain no more than 0.5 to 1 microgram per gram, 500 to 1,000 ppb, the threshold quoted on certificates of analysis and 25 to 50 times more permissive than the refined-oil detection limit

The regulatory limits that do exist

• European Medicines Agency, HMPC public statement EMA/HMPC/893108/2011 Rev. 1: an acceptable intake of 1.0 microgram of PA per day for adults in herbal medicinal products, an earlier version having set 0.35 micrograms per day in 2014. Germany's federal health authority set 1 microgram per day for limited internal use and 0.1 micrograms per day for unrestricted use, anchoring European practice since the early 1990s

• Commission Regulation (EU) 2020/2040 of 11 December 2020, applicable from 1 July 2022 and now consolidated in Regulation (EU) 2023/915, set food maximum levels as the lower-bound sum of 21 PAs plus 14 co-eluting compounds, 35 in total: borage leaves fresh or frozen 750 µg/kg, dried borage 1,000 µg/kg, dried herbal infusions 200 µg/kg, food supplements with herbal ingredients 400 µg/kg, and infant infusions 75 µg/kg dried or 1.0 µg/kg liquid

• The United States has none of these. FDA has set no PA limit for dietary supplements

The gap inside the EU limit

Sattler and colleagues at the Max Rubner-Institut, Talanta 2023, volume 258, article 124425, found the acetylated forms, acetyllycopsamine and acetylintermedine, among the quantitatively most relevant alkaloids in B. officinalis leaves. Those forms are not on the EU's regulated list, so a borage tea can pass the 200 µg/kg infusion limit while the dominant alkaloids in it were never counted.

What a certificate of analysis should show

• Species confirmed as Borago officinalis L., plant part stated as seed rather than aerial parts, and the fatty acid profile by GC-FID: GLA 18 to 27 percent, linoleic 35 to 42, oleic 15 to 20, palmitic 9 to 12

• Extraction and refining method stated explicitly, plus peroxide value at or below 10 meq O2/kg and acid value at or below 3.0 mg KOH/g. "Cold-pressed" and "expeller-pressed, hexane-free" are marketing positives that are PA negatives

• Pyrrolizidine alkaloids by LC-MS/MS with the reporting limit in ppb and the alkaloid list named. A panel screening only the EU's 21 will miss the acetylated forms

• Erucic acid quantified. Borage runs 2 to 3 percent, and Commission Regulation (EU) 2019/1870 cut the general vegetable oil ceiling to 20 g/kg while leaving camelina, mustard and borage oils at 50 g/kg because reduction by good manufacturing practice is not possible

Adulteration and substitution

• Dilution with cheaper high-linoleic oils, usually sunflower or safflower, detectable by the GC fatty acid ratio. A product claiming 240 mg GLA from 1,000 mg of oil cannot be evening primrose oil at 8 to 10 percent

• Boraginaceae cross-contamination in aerial part material. Comfrey, viper's bugloss and heliotrope are in the same family and carry higher PA loads

• Confusion with foxglove in fresh herb gathering. Memorial Sloan Kettering documents a near-fatal poisoning in a 58-year-old woman who mistook Digitalis for borage

What a buyer can actually verify

Divide labeled GLA milligrams by labeled oil milligrams. If the answer is not between 0.18 and 0.27, ask why. Then ask the manufacturer two questions in writing: was the oil alkali-refined and deodorized, and what is the PA reporting limit in ppb and which alkaloids were counted. A company that cannot answer both has not tested for the only thing that can seriously hurt you.

Special Considerations

Pregnancy, lactation, infants. Avoid both forms. This is one of the few supplement categories where fetal exposure has a documented mechanism of harm rather than a theoretical one, and the EU's PA limit for infant infusions is 75 µg/kg dried, four times stricter than the 200 for adult infusions and more than five times stricter than the 400 for supplements.

Older adults are disproportionately on warfarin and NSAIDs, the two clearest interactions. Topical use is a different risk profile, and the one place a small positive trial exists: Kanehara and colleagues, Journal of Dermatology 2007, volume 34, pages 811 to 815, tested borage-oil-coated undershirts in 32 children with mild atopic dermatitis over 2 weeks and reported improvement in erythema and itch.

People with rheumatoid arthritis. The evidence is real and the dose is the problem. Treat 1.4 to 2.8 g of GLA daily as the operative range, budget 6 to 12 grams of oil a day, expect a 24-week trial, and tell your rheumatologist. The 2022 American College of Rheumatology guideline on integrative interventions, in Arthritis Care & Research 2023, conditionally recommends following established dietary recommendations without supplements over adding them, on very low to moderate quality evidence. That is the current professional position.

Research Status & Evidence Quality

Strong Evidence For

• Nothing, in the sense of a replicated, adequately powered, positive result at a retail-realistic dose. Strong evidence does exist in the negative direction: the 2013 Cochrane review establishes that borage oil does not treat eczema

Moderate Evidence For

• Symptomatic relief in rheumatoid arthritis at 1.4 to 2.8 g/day of GLA. Cameron, Gagnier and Chrubasik, Cochrane Database of Systematic Reviews 2011, issue 2, CD002948, 22 studies overall and 7 pooled for GLA: pain reduced by a mean difference of 32.83 points on a 100-point scale, 95 percent CI -56.25 to -9.42, disability improved 15.75 percent, 95 percent CI -27.06 to -4.44, with the stated conclusion that there is moderate evidence GLA-containing oils afford some benefit in relieving RA symptoms. The pooled studies mixed borage, evening primrose and blackcurrant seed oil, and the review notes many herbal trials had design flaws

Emerging / Preliminary Evidence For

• Diabetic neuropathy, Keen 1993, never replicated. Topical borage oil in pediatric atopic dermatitis, Kanehara 2007, 32 children. Asthma quality of life with a GLA plus EPA medical food, Lindemann 2009, 21 subjects

Research Limitations

• The eczema question is closed. Bamford, Ray, Musekiwa, van Gool, Humphreys and Ernst, Cochrane Database of Systematic Reviews, 30 April 2013, issue 4, CD004416, DOI 10.1002/14651858.CD004416.pub2: 27 studies, 1,596 participants, 19 evening primrose and 8 borage. Participant-reported global symptoms on a 0 to 100 visual analogue scale pooled to a mean difference of -2.22, 95 percent CI -10.48 to 6.04, 176 participants, 7 trials; physician-reported -3.26, 95 percent CI -6.96 to 0.45, 289 participants, 8 trials. The authors' conclusion: "Oral borage oil and evening primrose oil lack effect on eczema; improvement was similar to respective placebos used in trials. Oral BO and EPO are not effective treatments for eczema." Their implications for research: "Noting that the confidence intervals between active and placebo treatment are narrow, to exclude the possibility of any clinically useful difference, we concluded that further studies on EPO or BO for eczema would be hard to justify." That is not a call for more research. It is a call to stop

• The individual trials agree with the pooling. Takwale, BMJ 2003, volume 327, issue 7428, page 1385: 151 randomized, 140 evaluable including 69 children, 920 mg GLA/day for 12 weeks, SASSAD 30 to 27 on borage and 28 to 23 on placebo, difference 1.4 points favoring placebo, 95 percent CI -2.2 to 5.0, P = 0.45, concluding "Gamma linolenic acid is not beneficial in atopic dermatitis." Henz, British Journal of Dermatology 1999, volume 140, pages 685 to 688: 160 randomized to 500 mg/day borage oil or miglyol placebo over 24 weeks, no overall efficacy, favorable only in a post hoc subgroup. Anheyer, Dermatitis 2025, volume 36, pages 207 to 215, 51 RCTs and 3,763 participants, found systemic evening primrose oil no different from placebo across 13 RCTs, SMD 0.14, 95 percent CI -0.45 to 0.73

• A regulator did act, and only on the drug side. The UK Medicines Control Agency announced on 13 September 2002 the withdrawal of marketing authorizations for Epogam and Efamast, licensed gamolenic acid products for atopic eczema and mastalgia, on the grounds there was not enough evidence they were effective. That is what happens when a GLA product is regulated as a medicine. The identical molecule sold as a supplement was untouched, and no authority has since acted against borage oil eczema marketing

• The RA evidence is genuinely positive and genuinely limited. Leventhal's 37 patients and Zurier's 56 are small, and Zurier used free fatty acid GLA rather than whole borage oil. Both papers carry the same warning from their own authors: that GLA "is usually taken in far lower doses than used in this trial," that it "is not approved in the United States for the treatment of any condition," and that it "should not be viewed as therapy for any disease." Thirty-three years on, the confirmatory trials both called for have not been run, and no trial has tested 240 mg of GLA a day against placebo for anything

• Funding. Most included studies in the 2013 review received funding or product from companies making these oils, and the lead author's own research was funded by a Miller-Dwan Foundation grant repaid by Efamol, a manufacturer. That the pooled result came back negative anyway strengthens it

• The aerial parts have no efficacy research at all. Borage tea is sold on tradition, and the only well-characterized thing about it is its alkaloid content

Summary & Key Takeaways

Borage occupies an unusual position in this archive. Almost every other ingredient here fails for want of evidence. Borage seed oil was studied properly, at adequate scale, with placebo controls, across 12 countries and more than 1,500 participants, and it failed with evidence. The 2013 Cochrane review did not report an inconclusive result. It reported narrow confidence intervals around no effect and said further study would be hard to justify. Someone paid for those 27 trials, and what they bought was an answer that should have removed a claim from the market. The claim is still there.

The honest response is either to stop selling the claim the evidence refuted or to say plainly which claim survives. For borage oil the surviving claim is narrow: moderate evidence for symptomatic relief in rheumatoid arthritis at 1.4 to 2.8 grams of GLA a day, which is 6 to 12 standard softgels. That claim is defensible, and it is almost never the claim being made, because the dose it requires would empty a 60-count bottle in five days.

Bottom Line

Borage seed oil is the most concentrated common source of GLA at 20 to 26 percent, about 2.7 times evening primrose oil. For eczema it does not work, and that is established rather than uncertain. For rheumatoid arthritis it has moderate evidence at 1.4 to 2.8 g/day of GLA, a dose no retail label directs you to take. The aerial parts are a different product with hepatotoxic pyrrolizidine alkaloids and no efficacy evidence.

Key Safety Points

• PAs are a leaf and flower problem, not a properly refined seed oil problem. Alkali refining cuts PA content roughly 30,000-fold to below a 20 ppb detection limit. Cold pressing does not

• "PA-free" is a manufacturer claim. No USP, NSF or ConsumerLab program certifies it and the US sets no legal PA limit. The real limits are European: 1.0 microgram per day for herbal medicines under EMA/HMPC/893108/2011 Rev. 1, and 400 µg/kg for herbal supplements and 200 µg/kg for dried infusions under Regulation (EU) 2020/2040

• PA hepatic veno-occlusive disease is irreversible. No antidote, supportive treatment only, and injury accumulates with cumulative dose

• Do not combine with warfarin without INR monitoring, and stop 2 weeks before surgery

• Do not give borage purée to infants. Mean nitrate 3,968 mg/kg and an odds ratio of 5.2 for methemoglobinemia in the Pamplona series

• Avoid in pregnancy, breastfeeding, liver disease and epilepsy

Special Note

There is a pattern here worth carrying to other shelves. When a supplement's clinical question has actually been answered in the negative, watch what the market does next. It does not withdraw the product. It changes the wording. Borage oil left the eczema aisle and reappeared as "skin health," "omega-6 balance" and "essential fatty acid support," phrases no trial can refute because no trial can test them. When you see a claim vague enough to be untestable, check whether a testable version of it was tested first, and how it came out.