Linoleic Acid & Metabolic Disease

The rest of this paper uses a handful of terms constantly. They're worth ten minutes up front, because once the basic chemistry clicks, most of the argument follows from it — and the argument really does rest on a single physical fact about how these molecules are built.

P.1All fats are chains — what differs is how many reactive sites they have

A fatty acid is a chain of carbon atoms. Two things make one fat different from another: how long the chain is (the number of carbons), and how many double bonds it contains. Chain length matters — it separates short-, medium-, and long-chain fats — but for this paper the double bonds are what count, because a double bond is a kink — and, importantly, it creates a reactive site where oxidation begins. (Strictly, the vulnerable spot is the carbon next to the double bond, whose hydrogen is easily pulled off — but “the double bond is where oxidation starts” is close enough to carry the argument.) That single fact drives nearly everything below.

Fats fall into three groups by how many double bonds they carry, and the count is what determines both their texture and their stability. Saturated fats (butter, tallow, coconut) have none — straight, tightly-packing, stable, solid at room temperature. Monounsaturated fats or MUFA (olive, avocado; chiefly oleic acid) have one. Polyunsaturated fats or PUFA (soybean, corn, sunflower) have two or more, which makes them liquid even when chilled and chemically fragile — they oxidize far faster, especially with heat, light, and air. The diagram makes the pattern visible:

P.2Omega-6 and omega-3: two families of PUFA that oppose each other

Polyunsaturated fats come in two families, named for where the first double bond sits when you count from the tail end of the chain. Both are essential — the body can't make either, so both must come from food.

The name is literally a location. Count carbon atoms starting from the methyl end of the chain — the end chemists call “omega,” the last letter of the Greek alphabet. The number tells you how many carbons in from that end the first double bond sits — position 6 for omega-6, position 3 for omega-3. (The number is the position of the first double bond, not the count of them: the parent omega-6, linoleic acid, has 2 double bonds; the parent omega-3, alpha-linolenic acid, has 3. Their longer derivatives have more — arachidonic acid, 4; DHA, 6.)

Before making the case against excess, the case for adequacy has to be stated plainly, because it's real and the rest of the paper depends on getting it right. Linoleic acid is an essential fatty acid — the body cannot synthesize it, and it is required. The argument here is not that LA is a toxin; it's that the dose-response is biphasic: beneficial and necessary at low-to-adequate intake, with the benefits saturating and the harms taking over as intake climbs into the modern range.

½.1The genuine benefits — at adequate intake

• Cell membranes. LA is a structural component of membrane phospholipids and is the most abundant PUFA in skin. It is built into the ω-hydroxy-ceramides of the epidermal barrier — the waterproofing lipids that keep the skin intact. Severe LA deficiency causes scaly dermatitis, water loss, poor growth, and impaired immunity. This is not in dispute.
• It arrives with its own antioxidant. In whole foods, PUFA does not come alone: nuts, seeds, and unrefined oils are also among the richest natural sources of vitamin E (tocopherols), the body's main fat-soluble antioxidant, which sits inside membranes and terminates the peroxidation chain reaction on PUFA before it propagates. Nature packages the fragile fat together with the molecule that protects it. This is part of why whole-food PUFA and refined, oxidized seed-oil PUFA are not the same exposure — industrial processing and the frying that follows strip and consume that protective vitamin E, delivering the substrate without its built-in defence (a theme returned to in Part VI).
• The anti-inflammatory metabolite branch. LA is the parent of γ-linolenic acid (GLA) → dihomo-γ-linolenic acid (DGLA) → prostaglandin E1 (PGE1) and 15-HETrE, which are anti-inflammatory and vasodilatory. LA's own direct oxidation also yields 13-HODE, a PPAR-γ ligand (II.2). So a real fraction of LA's metabolism runs toward resolution, not away from it.
• Inflammation is not the enemy — unresolved inflammation is. The acute inflammatory response is protective and necessary: it clears pathogens, initiates wound healing, and then resolves. The paper's target throughout is chronic, non-resolving inflammation. A functional inflammatory capacity, partly supplied by ω-6 mediators, is a feature. The pathology is when it never switches off (see Part V).

StrongLA essentiality, the skin-barrier role, and the GLA→DGLA→PGE1 branch are established, uncontroversial nutritional biochemistry.

½.2What happens to dietary LA — and the two pathways this paper will use

Before the disease arguments, it helps to lay out what actually happens to a molecule of linoleic acid once it is eaten, because the rest of the paper refers back to these names. LA has several possible fates, and it is worth being clear that the two this paper concentrates on are not the quantitatively dominant ones.

The full set of fates. Most dietary LA is simply β-oxidised for energy in mitochondria — burned like any other fatty acid. Much of the remainder is stored, esterified into adipose triglyceride (which is why adipose LA content, with its slow turnover, is the long-term intake biomarker used later), or incorporated structurally into membrane phospholipids — including mitochondrial cardiolipin — and into circulating lipoproteins such as LDL. Only a small fraction is desaturated and elongated down the D6D/D5D chain toward arachidonic acid. And peroxidation — the route to the OXLAMs — is not really a parallel destination at all: it is something that happens to LA already sitting in those membrane, lipoprotein, and storage pools.

That the bulk of LA is burned does not defuse the argument; it sets its scale. The fates are not independent: raising intake raises LA in every pool at once — more in membranes and LDL particles (more oxidisable substrate exactly where Part I needs it), more in slow-turnover adipose (keeping tissue LA elevated for years), and more substrate presented to the desaturases. The two pathways below are the ones the disease chapters use, and both are fractions of a much larger and rising total.

The desaturase chain. The first step, LA → GLA, is run by delta-6-desaturase (D6D), and it is the rate-limiting, saturable step of the pathway — a fixed enzyme meeting a variable input, so the throughput to GLA/DGLA plateaus rather than rising indefinitely with intake. (This is why GLA is sold as a direct supplement — evening primrose, borage, hemp — to enter the chain past D6D.) The chain continues through an elongase to DGLA (a favourable intermediate, precursor to the anti-inflammatory series-1 eicosanoids) and then, via D5D, to arachidonic acid — the pivot. AA is the substrate for the 2-series eicosanoids and, through CYP1B1 and the lipoxygenases, for the mid-chain HETEs (5-, 12-HETE) that recur throughout Part II. The point to carry forward is just the vocabulary and the shape: LA → GLA → DGLA → AA → its metabolites.

The oxidation chain. LA does not have to be enzymatically built up at all; its two fragile double bonds also make it a peroxidation substrate. Attacked by lipoxygenases, COX, or free radicals, LA becomes the hydroperoxides (9-/13-HpODE), reduced to the HODEs (9-/13-HODE) — the OXLAMs, oxidised linoleic-acid metabolites — and on to the oxo-ODEs. The same branch also yields the paper's other key product. Peroxidation is a chain reaction: once a radical abstracts a hydrogen from a bis-allylic carbon, the resulting lipid radical propagates to neighbouring molecules, and the hydroperoxides formed along the way are unstable. When they fragment — by β-scission of the carbon chain — they break into shorter reactive aldehydes, the most consequential of which is 4-hydroxynonenal (4-HNE)²², a nine-carbon aldehyde derived specifically from omega-6 fatty acids. So the HODEs are what LA becomes when peroxidation stops at the hydroperoxide stage; 4-HNE is what appears when the chain runs on and the molecule breaks apart. Both are referenced throughout Part II. This branch is entirely separate from the desaturase chain: it does not need D6D, and it is where the cardiovascular and much of the insulin-resistance story (Parts I–II) actually operate.

The popular version of this story reduces to one number: LA rising from ~2% to ~8–9% of calories across the century. The fuller picture, from USDA food-availability reconstructions, involves several simultaneous but distinct trends worth separating — because only one of them moves in a single direction the whole way through.

Total and saturated fat rose steadily from 1909, peaked around 2000, and have since modestly declined. If either alone explained metabolic disease, incidence should have climbed through the 1800s and fallen after 2000. It didn't. Trans fat spiked mid-century then fell sharply after regulation — yet metabolic dysfunction kept climbing afterward, so trans fat alone is not sufficient either.

One oil dominates that shift. Soybean oil alone supplies roughly 40–45% of all the LA Americans eat, and about 7% of total calories — because it is the default “vegetable oil” in processed food, restaurant fryers, salad dressings, and packaged baked goods. Soybean oil is roughly half LA by weight, so its rise is, in practical terms, the rise of LA in the food supply.

The one trend without reversal is the ω-6 : ω-3 ratio. Omega-6 (chiefly LA) rose from the early 1900s while omega-3 declined, producing a ratio now roughly an order of magnitude above the ancestral baseline. This shift was structurally driven by agricultural economics, not consumer preference for PUFA: post-1940s grain-yield gains drove grain prices down, which (a) made grain-derived vegetable oils cheap enough for mass-market processed food, and (b) shifted livestock to grain feed, altering the fatty-acid profile of meat itself — grain-fed animals carry more omega-6 and less omega-3 than grass-fed.

Critically, “vegetable oil” is not a single fatty-acid profile. Soybean, corn, and canola oil dominate the US supply, but the specific oils that came to dominate were disproportionately LA-heavy — largely for reasons of cost, not nutritional design. Treating the category as a monolith obscures that the food supply didn't just get “more vegetable oil,” it got specifically more LA.

Strongfor the food-availability trends themselves (USDA ERS series). Moderatefor the claim that LA specifically — rather than another correlated post-1940s change — is the operative variable.

I.1The oxidized-LDL mechanism — from oxidized fat to foam cell to plaque

Native LDL is not inherently atherogenic; oxidation is the event that starts the disease. The chain runs in a few steps, and dietary LA enters at a specific one. An LDL particle carries linoleic acid as part of its cargo; when the particle gets trapped in an artery wall, that LA is what peroxidizes; and only the oxidized particle triggers the runaway immune uptake that builds a plaque. So the argument has three links to establish — why the particle carries LA, why it oxidizes, and how the oxidized particle becomes plaque — each explained below and traced in the diagram.

Why the particle carries LA in the first place. An LDL particle's core is mostly cholesteryl esters — each a cholesterol molecule joined to a single fatty acid — along with triglycerides, wrapped in a shell of phospholipids and the apoB protein. Every one of those components carries a fatty acid, and which fatty acid broadly reflects what is available in the body's pool, so a diet high in LA builds particles whose lipids are correspondingly high in LA (this is why serum cholesteryl-ester composition is used as a biomarker of LA intake — it tracks the diet closely). The coupling is in fact stronger than passive availability: the plasma enzyme that attaches fatty acids to cholesterol to form those esters — LCAT — has a substrate preference for LA, actively favouring it over saturated and monounsaturated fatty acids. So raising dietary LA does not merely make more LA available; it feeds the enzyme already biased toward putting LA into the particle. More dietary LA means a more linoleate-rich, and hence more oxidisable, LDL particle.

Why the particle oxidizes in the first place is a question of location and time. LDL particles routinely cross into the artery wall and back out again — this is normal traffic, and a particle that makes the round trip is fine: it spends little time in the wall, stays surrounded by the antioxidants it carries and that circulate in blood, and leaves before it can be damaged. The problem begins only when a particle is retained. Its apoB protein is electrically attracted to the proteoglycans of the wall's connective-tissue matrix, and if it binds there it stops moving. Now the clock runs against it: stuck in the wall, it is progressively cut off from circulating antioxidants and left sitting in tissue where the vessel lining and resident immune cells continuously generate reactive oxygen. The longer it stays, the more its LA peroxidizes. So retention is the pivotal event — it converts routine, harmless LDL traffic into a stationary target, and a particle carrying more LA has more to lose once it is trapped.

What happens if it doesn't oxidize. This is the step that isolates the role of the fat itself. A retained particle carrying mostly saturated fat — no fragile double bonds to peroxidize — largely does not oxidize, so the scavenger receptors never recognize it and the runaway uptake never starts; it is cleared through regulated housekeeping, or un-sticks and diffuses back out. No foam cell forms.

From foam cell to fatal plaque. The oxidized particle is then taken up by macrophages — but through a different door than native LDL uses. Native LDL enters via the normal LDL receptor, which shuts off when the cell is cholesterol-full (the Brown & Goldstein receptor work, 1980s), so uptake is self-limiting and no overload occurs. Oxidized LDL instead enters via scavenger receptors (SR-A, CD36), which have no feedback brake — so the macrophage gorges without limit into a lipid-bloated foam cell. That absent brake is the hinge: it lets a handful of oxidized particles become a lesion. Foam cells then build plaque — accumulating into a fatty streak, recruiting more monocytes, and dying into a necrotic core (the spilled, oxidized contents of dead cells) that smooth muscle walls off with a fibrous cap. A thin cap over a large core is the vulnerable kind: plaque size doesn't kill; cap instability does.

Consistent with this: the blood of cardiovascular-disease patients shows elevated LA, with the most atherosclerotic individuals showing the highest serum LA.

StrongThe foam-cell/oxLDL/scavenger-receptor pathway is textbook atherogenesis, independent of any advocacy source. The specific weight assigned to dietary LA as the rate-limiting substrate is the paper's emphasis, not a settled quantity.

I.2Cholesterol quantity vs. cholesterol oxidation — and a natural experiment

The mainstream case for LA rests on one hard endpoint: it reliably lowers LDL cholesterol, and LDL-C lowering is assumed to lower risk. That assumption smuggles in a category error — it treats LDL as a quantity problem when the mechanism in I.1 says the problem is oxidation. LA can do both at once: lower the LDL number while making each particle more oxidizable.

Two lines of evidence that oxidation, not concentration, is the operative variable:

• Oxidized LDL outpredicts LDL. Circulating oxidized LDL is a stronger marker of coronary artery disease than LDL-C itself, and MDA-modified LDL specifically flags plaque instability — the rupture-prone state that causes events (Holvoet et al.³²). The oxidized fraction carries the prognostic signal.
• LA is the substrate that oxidizes. Linoleic acid is the principal fatty acid oxidized within the LDL particle, so a diet that raises LDL's LA content raises its oxidizability — and dietary MUFA (olive oil) reduces LDL's susceptibility to oxidation relative to LA-rich oils. Same particle count, different vulnerability.

A natural experiment. If high LDL alone caused heart disease, people with lifelong genetically elevated LDL should have had raised cardiac mortality in every era. They did not. Sijbrands et al.³⁶ traced a Dutch familial-hypercholesterolaemia pedigree back eight generations to a single 19th-century ancestral pair — a design that avoids the usual bias of recruiting FH patients who already have heart disease. Mortality was not raised in mutation carriers during the 19th and early 20th century; it rose after 1915, peaked in 1935–1964, then fell — and differed more than threefold between two branches carrying the same mutation. The authors conclude this “points to a strong interaction with environmental factors”: same genotype, same lifelong high LDL, radically different outcomes by era and branch.

The timing overlaps the industrialisation of the fat supply, but the study does not name the factor, and competing explanations aren't excluded — shorter 19th-century life expectancy with infectious disease dominating, a possible protective effect of cholesterol against infection, and smoking rising and falling on a similar curve. It is a single pedigree measuring all-cause mortality. It establishes that FH risk is strongly environment-dependent — not that LA is the environment in question.

Strongfor oxLDL outpredicting LDL-C, LA as the oxidation substrate, the MCE recovered-data finding, and the FH pedigree's era-and-branch variation implying powerful environmental modulation of a fixed lipid phenotype. Contestedfor the strong “LDL-C doesn't matter” framing, and for attributing the FH modulation to seed oils specifically.

I.3A second cardiovascular route: HETEs and cardiac hypertrophy

Atherosclerosis is not the only way LA's oxidation products reach the heart. Arachidonic acid drawn from the LA-fed oxylipin pool is metabolised by CYP1B1 into mid-chain HETEs (5-, 12-, 15-HETE), and these are directly cardiotoxic: in human ventricular cardiomyocytes they activate NF-κB and MAPK and induce cellular hypertrophy — an increase in cardiomyocyte surface area and total protein content, the cell physically growing and adding sarcomeres without dividing, alongside the standard hypertrophy gene markers (Maayah & El-Kadi¹⁶). CYP1B1 and its mid-chain HETEs are elevated across pressure-overload, isoproterenol, and angiotensin models of hypertrophy, and CYP1B1 inhibition ameliorates it. This is a distinct harm from plaque — myocardial remodelling rather than arterial occlusion — and it broadens the cardiovascular case.

Animal / in-vitrothe mechanism is largely one lab's, in a single cardiomyocyte line plus rodent models, and the studies dose the HETEs directly — none shows that dietary LA causes hypertrophy. The honest link is upstream and partial: LA feeds the AA pool the HETEs are made from. A real second route, graded in-vitro/rodent.

Part I leaned on oxidation — LA oxidizing inside LDL to drive atherosclerosis — though it also carried a desaturase-pathway route (the HETE/hypertrophy argument of I.3). Insulin resistance runs on the same two mechanisms, with the weighting reversed: oxidation contributes (II.2–II.3), but the better-evidenced lead here is desaturase flux: when linoleic acid is drawn down the delta-6-desaturase pathway (D6D, then D5D, then the CYP4A ω-hydroxylases), the throughput of that pathway — not the size of the leftover LA pool — is what tracks, and by Mendelian randomization causes, type 2 diabetes. The pathway's products (specific oxylipins such as 20-HETE, and the endocannabinoids of Part III) impair insulin signalling, and the flux itself is entangled with the cell's redox state.

This is a deliberate change of emphasis — from an oxidation-of-LA story to a flux story — made because that is where the human evidence for insulin resistance points. Excess dietary LA remains the modifiable input, the substrate the flux runs on, but the causal core is the enzyme activity rather than the peroxidation of a stagnant pool. The direct-oxidation route (II.2–II.3) is kept because it is real cell biology and because it is the mechanism shared with the cardiovascular story; for insulin resistance it is a contributor, not the spine.

One implication follows and is worth naming so it is not mistaken for a contradiction. Because the causal quantity is flux, and flux is shaped substantially by genotype (the FADS haplotypes) and metabolic state, the dietary lever for this endpoint is more targeted than a blanket “stop eating seed oils” message implies: diet supplies the substrate, while enzyme activity is set by genetics and metabolic state (the flux is not fixed by genotype alone, as later sections show). That does not weaken the mechanism — both inputs are simply true at once. Reducing LA reduces the raw material the pathway consumes; it does not rewrite anyone's genes. The dietary claim for insulin resistance is therefore real but proportionate, and the paper states it that way rather than overselling it.

The section runs in that order: the flux mechanism and its causal evidence (II.1), then the direct-oxidation routes flux operates alongside — the OXLAMs and JNK/IRS-1 signalling that carry LA oxidation to the insulin switch (II.2–II.3). The objection that blood LA and dietary LA point in opposite directions is taken up in the counterargument (Part V).

II.1The desaturase mechanism: how flux through D6D drives insulin resistance

This is the lead mechanism for insulin resistance (per the section opening). When linoleic acid is drawn down the delta-6-desaturase pathway (D6D → D5D → the CYP4A ω-hydroxylases), the throughput of that pathway — not the size of the leftover LA pool — is what tracks, and by Mendelian randomization causes, type 2 diabetes (Jäger et al.⁹). The mechanism is not bulk arachidonic-acid production — dietary LA barely raises tissue AA (Rett & Whelan⁴) — but a shift in the balance of pathway products and the redox state that accompanies the flux. What the enzyme's activity does is convert LA into specific bioactive oxylipins; the question is which ones reach insulin signalling, and how.

The short version: beyond diet setting D6D activity, there is evidence of a self-amplifying loop in which LA-driven oxidation itself feeds back to raise desaturase activity — oxidized LDL → inflammation → the AhR receptor → more desaturase and oxylipin-producing enzymes → more oxidation. Each link is documented; the whole loop is not proven end-to-end. The enzyme-by-enzyme mechanism follows — skip it if you don't need the details.

Strongfor the causal signal: MR links desaturase flux to diabetes, and 20-HETE supplies a concrete insulin-signalling endpoint. Contestedfor the dietary lever — flux is largely genetic and metabolic-state-driven, so diet moves it only partly.

II.2LA oxidation products that drive insulin resistance

Two families of oxidation product carry the insulin-resistance signal, and they differ in evidentiary weight. Both act largely through the IRS-1 switch formalised in II.3. OXLAMs are the diet-responsive, well-correlated but mechanistically lighter case; 4-HNE is the single best-evidenced product. They are treated in that order.

This is a contributor route, not the lead mechanism for insulin resistance (the lead is desaturase flux, II.1). It is kept because it is the mechanism shared with the cardiovascular story of Part I — the same chemistry, now in metabolic tissue — and because parts of it are well-replicated cell biology.

LA can oxidize directly, without becoming arachidonic acid — enzymatically via 12/15-lipoxygenase, COX, or CYP450, and non-enzymatically by free radicals — into oxidized linoleic-acid metabolites (OXLAMs): chiefly 9-HODE, 13-HODE, and the 9- and 13-oxoODEs. This is the same OXLAM family that drives the arterial oxidation story in Part I. Its role in insulin resistance is real but, on the evidence, secondary to the flux mechanism.

Strongfor the human diet→OXLAM link and the disease correlations; the 9-HODE signalling mechanism is well-characterised but largely rodent/in-vitro, and its quantitative contribution to human insulin resistance is unestablished. This is a plausible contributor, not a demonstrated master switch.

Of the two products in this section, 4-hydroxynonenal (4-HNE) has the better human evidence — so the oxidation story was not wrong for insulin resistance, it was pointing at the wrong molecule. It partly revives the oxidation claim that the flux mechanism (II.1) otherwise leads.

Strongfor the 4-HNE→insulin-resistance link: human tissue correlation, rodent-causal, and an LA-dependent source in cardiolipin. Inferredfor attributing it to dietary LA alone — LA is the substrate, but the driving ROS has other sources.

II.3How LA oxidation switches off insulin signalling (the IRS-1 hub)

This is the second of the two oxidation routes, and it converges on the same molecular switch as 20-HETE did (II.1) — a single site on the insulin-signaling machinery that, flipped the wrong way, turns insulin resistance on. The point of this section is to show how LA-derived oxidation flips it. First the switch itself.

Insulin signaling proceeds via tyrosine phosphorylation of insulin-receptor substrate-1 (IRS-1), recruiting PI3K toward glucose uptake. IRS-1 also has inhibitory serine sites (notably Ser307) that, when phosphorylated instead, block the IRS-1–receptor interaction and shut signaling down.

The responsible kinases — JNK and IKKβ — are activated directly by LA: in adipocytes, LA activates both, and inhibiting either prevents the LA-induced loss of functional IRS-1, with PKCθ upstream. A parallel, inflammation-independent route runs through pure oxidative stress: H₂O₂ exposure alone phosphorylates IRS-1 at Ser307 while activating IKKβ/JNK — so LA's own peroxidation products (including the OXLAMs of II.2a) can drive the endpoint without an inflammatory intermediate.

Strongas cell biology — the JNK/IKKβ→Ser307 mechanism is well replicated. But the node is shared: saturated fat activates it too, so it is a route to insulin resistance, not an LA-specific one.

This is the endpoint most associated with seed-oil arguments, and it is where the animal work is cleanest in this paper — while the corresponding human experiment has not been run. Both the rodent evidence and the state of the human literature are presented. The section closes by stating plainly what can and cannot be claimed.

III.1The endocannabinoid route — LA drives hunger itself

The last row of Simopoulos's table points at a mechanism that closes a loop the others don't: LA doesn't just alter how the body handles energy, it alters how much you seek. Both major endocannabinoids — 2-AG and anandamide (AEA) — are built downstream of LA, and they activate CB1 receptors that stimulate appetite and food intake. The relevant question is whether dietary LA moves that pool, and it does.

Alvheim et al.⁴⁷ isolated this cleanly: raising dietary LA from 1% to 8% in mice — mirroring the exact 20th-century US shift — roughly tripled liver 2-AG and AEA, and raised food intake, feed efficiency, and adiposity. Adding just 1% EPA/DHA shrank the endocannabinoid precursor pool and blunted the effect. The striking part: this happened even on a low-fat diet, meaning LA made a low-fat diet more obesogenic without adding calories from fat.

And the effect is not only appetite — which matters for how this argument is tested. The obvious reading of an endocannabinoid mechanism is “CB1 makes you hungry, you eat more, you gain weight” — and if that were the whole story, isocaloric trials would be entitled to dismiss it. But the knockout data say otherwise. Mice lacking CB1 are lean and resistant to diet-induced obesity, steatosis, and insulin and leptin resistance — and they resist those changes even though their total caloric intake is no different from wild-type mice on the same diet. Same calories in, different body composition: so endocannabinoid tone is altering energy partitioning and hepatic metabolism directly, not merely driving intake. Tissue-specific work locates a large part of this in the liver — hepatic CB1 is required for diet-induced steatosis, dyslipidaemia, and insulin and leptin resistance. The appetite arm is real (CB1 blockade was pursued as an obesity drug on exactly that basis), but it is not load-bearing on its own. That matters for III.3: the mechanism predicts effects a matched-calorie design could detect, so “isocaloric trials show nothing” is a weaker rebuttal than it first appears.

Strongin the animal model (LA isolated as the single variable, dose-response, reversed by omega-3). Human confirmation is inferential — the CB1-antagonist drug rimonabant did cause real weight loss but was pulled for psychiatric side effects, which supports the pathway's relevance while cautioning against pharmacologic mimicry.

III.2The controlled feeding evidence: soybean oil vs. coconut oil, lard, and fructose

The cleanest test of all this comes from the Sladek lab (UC Riverside), because it isolates fat type at matched calories and includes the control that pins the effect on LA specifically.

A 2025 follow-up extended the mechanism to LA-derived oxylipins and a genetically variable liver protein (HNF4α), and found that only liver-tissue oxylipins, not blood oxylipins, correlated with body weight — independent support for the compartment point that tissue damage need not show up on a blood test.

Strongas a controlled isolation of LA in mice — the Plenish arm is a rare design that changes only LA content. Moderatefor human translation: murine, and oxylipin/genotype-dependent rather than LA-universal.

III.3The human evidence is missing, not contrary

Everything above is rodent. The honest position on humans is not that the evidence contradicts this section — it is that the relevant experiment has not been run. What human trials exist were designed for other questions: they are isocaloric (so structurally blind to an appetite mechanism, III.1), short, and use endpoints such as liver fat that PUFA manipulate directly through a known pathway (SREBP-1c, CC.2) — measuring quantity rather than the composition and oxidative accumulation this argument is about. Those trials, and the case built on them, are taken up properly in the counterargument (Part V), where they can be weighed against the whole thesis rather than answered piecemeal here.

Strongin rodents, with LA isolated as the variable (III.2), and the human exposure biomarker has risen 136%. Animal / in-vitroThe honest claim: LA promotes fat accumulation in rodents; in humans the relevant experiment — ad libitum intake, long duration, adiposity and oxidative endpoints — has not been run. Existing trials are isocaloric, short, and use endpoints PUFA act on directly; they are addressed in Part V.

The claim in this section is deliberately narrower than “LA causes cancer,” and it should be read that way throughout. Cancer has many initiating causes and the human body is layered with defenses; no single dietary fatty acid pulls the trigger. The claim is probabilistic and environmental: a high-LA diet helps create the metabolic and inflammatory terrain in which cancer is more likely to take hold and progress. Direction, not magnitude — more likely, not inevitable.

IV.1The terrain is the one the earlier chapters already built

This section does not need new mechanisms. Everything it requires has already been established elsewhere in the paper for other endpoints — and the argument is simply that the state those mechanisms produce is, item for item, the state in which tumours are favoured. Cancer biology has a standard name for these conditions: tumour-promoting inflammation and deregulated cellular energetics are two of the named hallmarks (Hanahan & Weinberg), so the terrain being described here is mainstream oncology, not a fringe construct. What follows is the mapping.

What a human tumour actually looks like on this account. The prediction is specific and checkable: tumour tissue should show an over-supply of arachidonic-acid-derived inflammatory mediators together with a deficit of the pro-resolving ones. That is what is found. Soundararajan et al.⁶² ran lipidomics and spatial transcriptomics on paired human colorectal tumour and normal tissue and reported exactly this asymmetry — over-expression of AA-derived leukotrienes and the machinery that makes and senses them, alongside near-absent lipoxins (LXA4, LXB4). They call it defective lipid class switching: normal wound healing runs inflammation → resolution, and here the resolution phase never fires — Virchow's old description of a tumour as a wound that does not heal, now with the lipid mediators named.

The honest limit of that study is worth stating once: it measured no diet. It documents the terrain; it does not show what produced it. What it establishes is that the state this paper's mechanisms produce is the state actually present in human tumour tissue — which is the premise the dietary argument needs, not the conclusion it wants.

Strongfor each mechanism individually — all five are established literature. Inferredfor the assembly: that dietary LA moves someone along this terrain far enough to change cancer risk. Magnitude unquantified.

IV.2A worked example of context-dependence: LA → FABP5 → mTORC1

One 2025 Science paper (Koundouros & Blenis⁶¹) shows how LA can also feed growth signaling directly, in the right context: LA is sensed by fatty-acid-binding protein 5 (FABP5), which binds Raptor to enhance mTORC1 assembly — a central growth switch. In triple-negative breast cancer models (high FABP5), a high-LA diet raised FABP5, boosted mTORC1, and grew larger tumors; omega-3 had no such effect.

ModerateStrong, specific mechanism in one susceptible context (TNBC, animal/cell). Not generalizable to “LA causes cancer” — and under the terrain framing it isn't meant to be; it exemplifies context-dependent risk modification.

IV.3The epidemiology is consistent with a modest probabilistic effect

Under a direct-cause claim, the mixed human data would be damning. Under the terrain claim, it's what you'd expect. A 2025 dose-response meta-analysis (Atashi et al.⁶³) found LA associated with colorectal cancer — a modest elevation, the right size for an environmental risk modifier, not a potent carcinogen. Total omega-6 and arachidonic acid were not associated, and other cohorts (e.g. Cancer Prevention Study-II) find null or even inverse relationships.

ModerateA small LA–CRC association in the largest meta-analysis, nulls elsewhere — consistent with a probabilistic contributory factor. Direction defensible; magnitude unquantified. Not evidence of direct causation, and not claimed as such.

The threads below are not specific to any one disease — they cut across the cardiovascular, metabolic, obesity, and cancer arguments made in the four Parts above, and several answer standard objections to the LA thesis. They are collected here, after the disease-specific case, rather than repeated in each section. This is a cross-cutting interlude, not a numbered Part.

CC.1The bridge to humans: does eating LA actually cause oxidative damage?

This is the load-bearing question for the entire paper. Mechanism in cells and mice is abundant; what matters is whether changing dietary LA changes oxidative damage in living people. Three controlled human studies address it directly, and they are the most important citations in this document.

Strongfor the diet→oxidation coupling itself: two controlled human trials moving in opposite directions with consistent results, one using a genuine low-LA run-in and a MUFA comparator. Inferredfor the step from marker change to disease outcome.

CC.2Self-enrichment: PUFA suppresses the synthesis of the fats that would dilute it

A mechanism worth stating separately, because it explains an otherwise awkward result and because it is easy to get backwards. Polyunsaturated fatty acids are potent suppressors of hepatic de novo lipogenesis (DNL) — the manufacture of new fatty acids from carbohydrate.

But invert that last observation and a real consequence appears.

Strongfor the mechanism: PUFA-specific LXR antagonism suppressing SREBP-1c and de novo lipogenesis, with SFA and MUFA largely inactive. Speculativefor the enrichment consequence and its application to the liver-fat trials — coherent and directly testable, but not yet measured.

CC.3The ω-6 : ω-3 axis — opposing effects on the same targets

The routes above are one half of a balance. Simopoulos's 2016 review⁷⁴ lays out how omega-6 and omega-3 act in opposite directions on the same metabolic targets: adipogenesis, inflammation, insulin/leptin signaling, and the endocannabinoid system. Because both families compete for the same desaturase enzymes (FADS1/FADS2, which actually prefer ALA but are outcompeted by the sheer mass of dietary LA), the ratio — not just absolute LA — sets the downstream mediator balance. This is also the resolution side of the cancer argument in Part IV: the mediators that end inflammation are built from omega-3s.

StrongMainstream review; the opposing-effects framework is well-supported. The ratio-as-primary-lever emphasis is Simopoulos's interpretive stance, shared but not universal in the field.

CC.4Membranes: the same oxidation, one level down

The same chemistry applies to LA built into cell membranes, and the consequence is worth stating plainly: a membrane's oxidizability is set by its fatty-acid composition. Membranes rich in polyunsaturated phospholipids carry more of the fragile double bonds that peroxidation chains propagate through; monounsaturated or saturated phospholipids carry fewer. This is directly diet-responsive — exogenous monounsaturated fat (oleic acid) displaces PUFA from membrane phospholipids in an ACSL3-dependent manner, measurably reducing peroxidation susceptibility (Magtanong et al.⁴⁶). That substrate-competition effect — not antioxidant activity — is the mechanistic rationale for the MUFA substitution recommended in Part VI.

Strongfor the core principle — membrane PUFA content sets peroxidation susceptibility, and MUFA displaces PUFA via ACSL3. Animal / in-vitrofor cardiometabolic tissue specifically; membrane oxidizability is a mechanism here, not evidence.

CC.5The adipocyte paradox — and why it is not actually a contradiction

An apparent inconsistency has to be addressed head-on, because a careful reader will find it. Several lines of evidence suggest LA increases adipocyte insulin sensitivity and promotes fat storage — which appears to run directly opposite to II.2, where LA's oxidation products drive insulin resistance. It can be resolved, and the resolution turns out to be mainstream physiology rather than special pleading.

The dissociation is also visible in the feeding data already cited: in mouse studies, insulin-tolerance-test performance keeps worsening with LA intake after weight gain has plateaued — glucose handling degrades independently of further fat accumulation, which is what a ceiling-then-spillover model predicts and a simple “more fat equals more resistance” model does not.

Strongfor each component: HODEs and LA are documented PPARγ ligands; PPARγ is the master adipogenic regulator; TZDs improve insulin sensitivity by expanding adipose despite weight gain; the expandability/spillover framework is mainstream. Inferredfor the specific claim that dietary LA moves a person along this trajectory at realistic intakes — that step is inferred from the components, not directly demonstrated.

CC.6Synthesis: why the same double bonds are a liability in two different places — and what that implies about saturated fat

The mechanisms above can be read as pointing in two directions, and rather than hide the tension this section resolves it explicitly. The resolution also produces the paper's one substantive statement about saturated fat, so it is placed here, reasoned rather than asserted. What follows mixes well-supported chemistry with a frankly speculative mechanism; the two are labelled throughout, and the speculative half is not load-bearing for anything else in the paper.

The practical upshot — that saturated fat compares favourably to excess LA — is developed as dietary guidance in Part VI.3b, where the epidemiology (which does not convict saturated fat as a class) is set alongside this mechanistic reasoning.

Strongfor arm 1 (uncontrolled LA peroxidation → damaging products) — the paper's spine. Speculativefor arm 2 (LA under-generates a controlled Complex-I satiety signal) — mechanistically grounded but unproven at human dietary intakes; presented as an integrative hypothesis, not a demonstrated result.

A paper that only presents its own case is advocacy. This section states the best version of the argument against the thesis — not a caricature, but the case a well-informed lipid researcher would actually make — and then answers it. Where the reply is weak, that is said plainly.

The reply, point by point.

Translating the mechanisms into food. Three framing points first, because they determine whether any of this is worth the effort:

• One change outweighs all the rest. Most of an American's LA doesn't come from the bottle at home — it comes from restaurant, fried, and packaged food, where soybean oil is the default. Cooking at home with almost anything beats optimizing which fat you cook with.
• Don't chase zero. LA is essential and present in trace amounts across nearly all whole foods. The target is the concentrated industrial sources.
• This is a food-quality argument, not a calorie argument. Swapping fats while still eating mostly ultra-processed food misses the point — those foods carry the LA and the refined starch and sugar that the animal data shows are worst in combination.

VI.1The levers

Before the specific foods, the interventions themselves — ordered by how directly each addresses the mechanism. The first matters far more than the rest.

1. Eat less LA. Not zero — down from the modern ~8–10% toward the ~1–2% the body needs. Cut the concentrated sources (soybean and corn oil, fried and ultra-processed food), not the trace LA in whole foods. Strong
2. Get vitamin E from food, not a bottle. Vitamin E is the chain-breaking antioxidant that stops peroxidation propagating, and the amount you need scales with how much PUFA you eat. Whole high-LA foods — nuts, seeds, unrefined oils — arrive with their own vitamin E; refined oil and fried food have had much of it stripped or spent. Eat LA in its original packaging. Strong
3. Selenium-dependent GPX4. The enzyme that detoxifies lipid hydroperoxides before they decompose into HNE/MDA/OXLAMs — the same enzyme that limits lipid-peroxide damage generally. Selenium is rate-limiting. Strong
4. Omega-3 to dilute LA. ALA/EPA/DHA compete with LA for D6D and for membrane incorporation. Human tracer data (Emken⁵) supports real competition in mammals; a rainbow-trout study found none, cautioning against over-generalizing across taxa. Moderate
5. MUFA substitution (oleic acid). Actively displaces PUFA from membrane phospholipids via ACSL3, measurably lowering peroxidation susceptibility (CC.4) — a substrate-competition mechanism, distinct from both simple LA-avoidance and from antioxidant protection. This is the mechanistic case for olive and avocado oil beyond “heart-healthy.” Strong
6. Odd-chain SFAs (C15:0, C17:0). Zero double bonds, so chemically incapable of chain propagation — the same substrate logic as MUFA substitution above. Contested — sound chemistry, but the disease-outcome literature is dominated by one financially-interested researcher with little independent replication.
7. Avoid heated / reused frying oil. Distinct exposure route (VI.2) delivering pre-formed OXLAMs/aldehydes. Strong

VI.2Fried food: the worst exposure

Frying is a distinct exposure, additive to everything above. Every other mechanism in this paper requires LA to oxidise inside your body; frying delivers the oxidation products already made. Heat plus oxygen applied to PUFA is non-enzymatic combustion, yielding the same reactive aldehydes found in cigarette smoke — 4-HNE, acrolein — plus trans fats, which soak into the food and are eaten with it. Saturated fats, lacking the vulnerable double bonds, do not generate the same cocktail.

Commercial frying is the worst case, and the reason is time. A restaurant fryer sits hot and aerated for the entire service, and the oil is typically not changed for days or weeks. Peroxidation is a chain reaction and its products accumulate: what is being cooked in by the end of that period is not fresh oil but a concentrated aldehyde load. Home frying in fresh oil, discarded after, is a different exposure from the same nominal food.

Strongthat heating PUFA generates aldehydes and OXLAMs, and that reused frying oil accumulates them. Inferredfor how much of fried food's association with disease is attributable to those products specifically, as against the calories, salt, and food matrix it arrives in.

VI.3Think in fatty-acid classes, not “seed oils”

Stop sorting fats into “vegetable” versus “animal” — those categories get abused precisely because they're imprecise. Sort by the dominant fatty acid.

On olive oil: adulteration with cheaper high-LA oils is common. Buy single-origin, harvest-dated, dark-bottled where possible.

Strongas a classification principle — fatty-acid composition, not plant-vs-animal origin, is what determines oxidative behaviour, and the compositions themselves are straightforward analytical data.

VI.3bOn saturated fat: not the villain — and plausibly the better choice

A paper that tells people to cut linoleic acid owes them an answer to the obvious question: if not seed oils, then what — and isn't saturated fat supposed to be worse? The honest answer has two parts, one solid and one reasoned.

Strongthat saturated fat has not been shown to raise cardiovascular mortality in the major meta-analyses (Siri-Tarino⁶⁶, de Souza, Chowdhury⁶⁸) — the negative claim this section actually needs. Inferredfor the positive claim that SFA is preferable; that rests on the oxidative-stability argument of this paper rather than on trials showing benefit from adding it.

VI.4Animal foods — and what livestock feeding reveals

There's a biological asymmetry here that determines where sourcing is worth paying for — and an uncomfortable observation hiding inside standard agricultural practice.

Strongfor the underlying biology — ruminant biohydrogenation buffers dietary PUFA, monogastric tissue fat mirrors feed, and livestock feed composition has shifted with the oil supply. Moderatefor the health consequence of sourcing choices in humans; the compositional difference is measurable, the outcome difference is not directly demonstrated.

VI.5Nuts and seeds — where the intuition needs correcting

The realistic risk isn't the nuts — it's snack mixes and nut butters roasted in seed oils, plus the ease of eating 400 calories of them without noticing.

Strongfor the compositional rankings themselves, including the walnut correction (high ALA but higher LA). Moderatefor translating composition into a recommendation — whole nuts arrive with fibre, tocopherols, and an intact matrix, and cohort data on nut intake is broadly favourable despite the LA content.

VI.6The avoidance hierarchy

Inferredas a ranking. The ordering follows from the mechanisms argued above — pre-formed oxidation products first, bulk exposure next — but no trial has compared these avoidance tiers head-to-head, so the hierarchy is reasoned, not measured.

Part V answers the strongest opposing case. What follows is the residue — the weaknesses in this paper's own construction. In one sentence: the mechanisms are strong and the claim they support is a terrain-shifting one — that high LA raises disease risk by degrading metabolic conditions, not that it single-handedly causes any endpoint.

• The sourcing is not neutral. Several threads trace to one network with commercial or ideological stakes. The primary citations are mostly sound and mainstream reviews are cited alongside, but the framing did not arise from neutral ground.
• Oxidation is the strong claim for cardiovascular disease, not for insulin resistance. Part I is the firmest mechanism. For insulin resistance the case is deliberately multifactorial: strong cell biology, thin human quantification, an unsettled desaturase strand, and a JNK/Ser307 node that saturated fat activates too. This paper does not claim LA oxidation is the master cause of insulin resistance.
• This is about linoleic acid, not a claim that saturated fat is clean. The lipotoxicity literature implicates palmitate as a driver of mitochondrial ROS and insulin resistance through partly distinct mechanisms. The two are complementary: saturated fat supplies much of the oxidative spark, LA the oxidisable substrate.
• The loops are inferred, not demonstrated. The AhR feed-forward loop (II.1) and the self-enrichment mechanism (CC.2) are each assembled from independently documented links; no single experiment shows either chain end-to-end, and the work is largely mouse and cell.
• The human trials are short, and there is a mechanistic reason short trials mislead. The controlled trials carrying the diet→oxidation bridge run 4–12 weeks with 38–80 participants. Tissue PUFA turns over slowly — adipose linoleate takes one to several years to re-equilibrate — so a weeks-long trial measures a body still carrying its old load. No trial has taken people to low LA and measured hard events.
• The obesity claim is rodent-only. The Deol comparisons (III.2) are the cleanest isolation of LA available, and they are in mice. The human replies in Part V explain why trials might miss an effect; they are not evidence the effect is there.
• The cancer claim is the weakest. It is a terrain argument built on mechanisms established for other endpoints; the human epidemiology is modest and contested, and several biomarker datasets point the other way (IV.3).
• The recommendations are untested as a package. These are mechanism-derived and low-risk, but the honest claim is “plausibly better,” not “proven.” None of it is medical advice, and it matters far less than not smoking, sleeping enough, and staying active.

All previously outstanding citation-audit items have been resolved: the EPIC-Potsdam desaturase paper is confirmed as Kröger et al. 2011 (Am J Clin Nutr 93:127); the HODE review as Vangaveti et al. 2010 (Ther Adv Endocrinol Metab); the UK Biobank adiposity paper as Lai et al. 2025 (Br J Nutr, and its pro-omega-6 provenance is noted); the colon-cancer paper as Soundararajan et al. 2025 (Gut 74:586); the skin-barrier fact to the Linus Pauling Institute review and the PNPLA1 mechanism; and the β-cell ferroptosis paper as Krümmel et al. 2022 (J Nutr Biochem 106:109013), cited as a saturated-fat liability. The OMEMI trial (Circulation 2021) previously bundled with the desaturase material was removed as mis-bundled (it is an omega-3 secondary-prevention trial, not a test of the LA/D6D point).