The seed-oil debate bundles several questions together. Does linoleic acid, an omega-6 fat, promote inflammation? Can extraction, refining or repeated heating introduce harmful substances? And how do foods containing these oils affect long-term health? Each question needs a different kind of evidence.
A small soybean-oil trial helps address the first. It does not settle the others. Understanding that distinction explains much of the apparent gap between alarming mechanistic arguments and generally reassuring findings on common inflammatory markers.
What was actually tested
Published in January 2026, this secondary analysis involved 10 adults with overweight or obesity. In a randomized, double-masked crossover trial, each consumed foods supplying 30 grams of soybean oil daily for four weeks and equivalent foods containing palm oil for another four weeks, in randomized order, separated by a two-week washout. [1]
The comparison was soybean versus palm oil, not oil versus no oil. Participants incorporated the study foods into their usual diets with dietitian support; this was not a fully controlled feeding study in which researchers supplied every meal. [1]
No significant changes were detected in the measured inflammatory markers or oxidized LDL. The downward trend in IL-6 was not statistically significant (p = 0.09). Red-blood-cell arachidonic acid fell, but so did DHA; neither change alone establishes a health benefit or harm. [1]
A non-significant result does not prove exactly zero effect: a small pilot can miss meaningful differences. Likewise, a downward trend is not a demonstrated anti-inflammatory benefit.
The broader evidence matters more
A 2017 meta-analysis combined 30 randomized trials involving 1,377 participants. Increasing dietary linoleic acid, the omega-6 fat abundant in many seed oils, did not significantly change the main blood inflammatory markers overall. The authors did report a signal suggesting that larger increases in intake might affect CRP, and evidence for some markers was limited. That qualification belongs alongside the overall result. [2]
For example, a 10-week randomized trial in 67 adults with abdominal obesity compared diets richer in vegetable omega-6 polyunsaturated fats with diets richer in saturated fat, mainly from butter. Liver fat was lower with the polyunsaturated-fat diet, without an increase in inflammation. This was a different experiment, with a different comparator and outcome, rather than a replication of the soybean pilot. [3]
Why the mechanism does not predict the whole outcome
The familiar hypothesis runs from linoleic acid to arachidonic acid and then to inflammatory signalling molecules. The pathway exists, but its presence does not show that eating more linoleic acid increases its output. A systematic review of adult intervention studies found no consistent increase in arachidonic acid in the measured blood lipid compartments when dietary linoleic acid increased within Western-type diets. That finding has a defined population and measurement scope; it is not a claim about every tissue or dietary setting. [6]
Arachidonic acid also supplies molecules involved in resolving inflammation. Describing it solely as an inflammatory ingredient leaves out part of its biology. The soybean paper discusses both roles. [1] The net effect depends on how pathways operate together, not simply on how much precursor a food contains.
Experimental conditions matter too. Exposing isolated cells to a substance can identify a possible mechanism. It cannot reproduce digestion, distribution and regulation throughout a person. Translating such a result requires checking whether the experimental concentration, exposure and outcome resemble real consumption.
This does not mean that human trials always detect every harm. Common blood markers capture only part of biology, and short studies cannot answer questions about disease developing over decades. The mismatch can therefore reflect an oversimplified prediction, an outcome that was not measured, or inadequate precision. Those explanations need testing rather than choosing whichever suits a preferred conclusion.
Processing deserves its own assessment
Some oils are solvent-extracted and refined; others are mechanically pressed. Refining can remove unwanted substances, but processing conditions also matter. FAO describes bleaching with adsorbent materials and deodorisation using steam under reduced pressure. High-temperature treatment can produce some trans isomers; it does not turn all the oil into trans fat. Oxidation and trans-fat formation are distinct chemical changes. [7]
Hexane residues are an exposure question. EFSA called for a fresh safety assessment of technical hexane used in food extraction and subsequently requested additional data. This is a real evidence gap, not evidence that every solvent-extracted oil causes illness. Assessment requires reliable residue measurements, intake estimates and toxicology relevant to those exposures. [8]
There are also established processing contaminants. EFSA has assessed glycidyl esters and 3-MCPD esters, which can form during high-temperature refining. Glycidol, released from glycidyl esters, is genotoxic and carcinogenic. Its 2016 assessment found particularly high contaminant concentrations in palm oils and fats. This historical finding should not be treated as a measurement of every product sold today. It does show why refining conditions and finished-product composition deserve attention. [9]
The soybean pilot did not characterize these contaminants or test repeatedly heated frying oil. Measuring oxidized LDL in blood is not equivalent to measuring oxidation products in the oil itself. A comparison with palm oil also cannot isolate a contaminant that might occur in both products. Its reassuring inflammatory-marker findings therefore cannot certify all processing methods, storage conditions or cooking practices as safe.
Inflammatory markers are not a longevity verdict
A blood marker, a clinical event and lifespan are separate outcomes. Evidence about one should not be silently promoted into evidence about the others.
A Cochrane review of longer-term omega-6 trials found uncertainty around several cardiovascular outcomes. It reported that increased intake may have little or no effect on mortality or overall cardiovascular events, while a possible reduction in heart attacks rested on low-quality evidence. These findings do not justify a claim that seed oils extend life. [4]
Dietary guidance also asks a comparison question. WHO recommends that dietary fats be primarily unsaturated and identifies polyunsaturated fats, plant-derived monounsaturated fats and fibre-rich carbohydrate foods as replacements for saturated fats. The relevant question includes what replaces what in the diet. [5]
Food context adds another layer. In a randomized inpatient study, an ultra-processed diet led participants to eat more and gain weight compared with an unprocessed diet. That experiment tested whole dietary patterns; it did not identify seed oils as the cause. [10] Neither an ingredient list nor a blood-marker result can stand in for assessing the complete food and its effects on intake.
What would help resolve the remaining questions?
The next studies should separate the questions currently grouped under “seed oils.” The following are research priorities, rather than findings established by the pilot.
First, characterize the oils actually consumed: fatty-acid profile, extraction and refining method, storage and heating history, relevant solvent residues, oxidation products and processing contaminants. Comparisons should use measured composition rather than assuming that every bottle within a category is equivalent.
Second, run adequately powered, preregistered feeding trials that specify the replacement nutrient and keep energy intake and the rest of the diet comparable. Where feasible, compare oils with similar fatty-acid profiles but different processing histories, and compare fatty-acid profiles under matched processing conditions. That would help distinguish the effect of the fat from the effect of its manufacture. Studies should stay within ethical food-safety limits.
Third, connect mechanism to outcomes: prespecified inflammatory measures, lipid and metabolic responses, and relevant exposure biomarkers can test short-term predictions. Longer follow-up is needed for clinical outcomes; contaminant safety also requires toxicology and exposure assessment. Confidence intervals should show which effect sizes a study can reasonably exclude, rather than letting a non-significant p-value carry the whole conclusion.
Independent replication, transparent funding and accessible methods and data would strengthen confidence. The pilot was supported by the United Soybean Board and Soy Nutrition Institute Global alongside public and academic funding; Cargill donated the oil. The authors report no sponsor role in design, conduct, analysis or publication, and one disclosed soybean-board travel reimbursement. [1]
The present evidence gives us a useful, bounded conclusion: human trials do not support a blanket claim that dietary linoleic acid raises common inflammatory markers. Questions about processing contaminants, particular exposures and long-term health still require their own evidence. Progress depends on measuring those distinctions directly.