The research team, led by investigators at Brigham and Women’s Hospital and Harvard Medical School including M. Elizabeth Moss and Bruce Levy, applied a technique known as lipid mediator metabololipidomics, using liquid chromatography coupled with tandem mass spectrometry, to profile hundreds of fat-derived signaling molecules in blood plasma. Their subjects came from two cohorts totaling 106 participants with recent SARS-CoV-2 infection. Sixty-four went on to meet clinical criteria for Long COVID, defined by the National Academies as an infection-associated chronic condition persisting at least three months, while 42 fully recovered. By comparing lipid profiles between these groups at different time points, the researchers could ask whether the chemistry of inflammation resolution differs in people who develop the syndrome.

The logic behind the approach rests on a well-established but underappreciated branch of immunology. Essential polyunsaturated fatty acids such as arachidonic acid, docosahexaenoic acid, and eicosapentaenoic acid are not merely fuel; they are raw materials for an entire pharmacopeia of signaling molecules. Lipoxygenase enzymes convert these fats into specialized pro-resolving mediators, including lipoxins, resolvins, protectins, and maresins, which actively coordinate the return of tissues to homeostasis after injury or infection. In a self-limited inflammatory response, the body performs a metabolic class switch, pivoting from pro-inflammatory prostaglandins and leukotrienes to these pro-resolving mediators. If that switch fails, inflammation can smolder indefinitely.

The first major finding came from the acute phase. In samples drawn within 30 days of a positive test, participants who would later develop Long COVID showed significantly elevated levels of three products of the 12-lipoxygenase enzyme: 12-hydroxyeicosatetraenoic acid from arachidonic acid, 14-hydroxydocosahexaenoic acid from DHA, and 12-hydroxyeicosapentaenoic acid from EPA. This association held even after the researchers statistically controlled for the severity of the initial infection, gender, and self-reported race. A composite score aggregating all 12-lipoxygenase products was likewise a significant predictor of later Long COVID, as were aggregate measures of arachidonic acid and EPA metabolism. Notably, the pattern was not simply a byproduct of generalized platelet activation, since thromboxane products of platelet cyclooxygenase were not elevated.

The 12-lipoxygenase signal is intriguing because the enzyme is highly expressed in platelets and sits at the crossroads of clotting and inflammation. Acute SARS-CoV-2 infection is known to trigger endothelial inflammation, angiogenesis, platelet activation, and microthrombi, and the new data suggest that selective 12-lipoxygenase activation, which can occur through thrombin receptors, collagen signaling, or immune complexes independently of the cyclooxygenase pathway, may link these processes to the later development of persistent symptoms. The researchers also observed that reduced levels of interleukin-22 and of resolvin D1, a DHA-derived pro-resolving mediator, in the acute phase predicted Long COVID risk, hinting that a deficit in resolution chemistry may accompany the excess lipoxygenase activity.

After the first month, the picture shifted. In cross-sectional samples collected more than a month post-infection, the correlation structure of the lipidome itself had changed. In recovered participants, intermediates derived from the same parent fatty acid tended to rise and fall together, indicating that metabolism was organized around the available fats. In Long COVID participants, this within-family correlation was lost for the DHA and EPA families, while a strong signature of 5-lipoxygenase activation emerged across fatty acid families. The authors interpret this as an uncoupling of PUFA metabolism in Long COVID, in which enzyme activity, rather than substrate availability, drives the lipid profile, consistent with disrupted lipid mediator class switching.

Longitudinal sampling over the first year after infection sharpened the temporal picture. Levels of 15-HETE, an arachidonic acid product, remained significantly higher in Long COVID participants, while resolvin D1 was consistently lower; indeed, resolvin D1 was undetectable in 14.3 percent of Long COVID samples but in none of the recovered samples. Several metabolites, including 17-HDoHE, 14-HDoHE, and lipoxin A4, showed significant interactions with time, diverging between the groups in the three-to-nine-month window after infection. These trajectories suggest that lipoxygenase activity remains elevated for months in Long COVID even as production of key pro-resolving mediators stays depressed, a combination that could sustain defective resolution of inflammation.

The study also intersected with another leading hypothesis about Long COVID: persistence of viral proteins. Using ultra-sensitive single molecule array assays, the researchers detected SARS-CoV-2 spike protein in a subset of both Long COVID and recovered participants long after acute infection, and unsupervised clustering of the multidimensional data separated samples largely by the presence or absence of spike. Within Long COVID samples, plasma levels of 17-HDoHE correlated significantly with spike protein abundance, a relationship absent in recovered participants. When the team clustered Long COVID samples using lipid mediators alone, four distinct biochemical subtypes emerged, one enriched in 17-HDoHE and another almost entirely lacking resolvin D1, hinting that Long COVID may encompass several metabolically distinct endotypes rather than a single disease process.

The authors are careful to note limitations. The cohorts were modest in size, particularly the recovered group in the longitudinal arm, and clinical status was determined by chart review rather than validated survey instruments. Medication use, especially aspirin, which can suppress 12-lipoxygenase activity, and dietary intake of omega-3 fatty acids were not captured, and parent fatty acid levels were not quantified. Many bioactive mediators fell below detection limits in plasma, prompting the authors to call for serum-based analyses that could better distinguish deficient substrate from dysregulated enzymatic machinery.

Even so, the findings open concrete paths forward. Elevated 12-lipoxygenase products in the first month of infection could serve as an early biomarker identifying patients at risk of Long COVID before symptoms become chronic, and the results dovetail with an ongoing randomized controlled trial, ARACOV-02, testing whether supplementation with pro-resolving mediator precursors such as 17-HDoHE, 14-HDoHE, and 18-HEPE, combined with telerehabilitation, can improve outcomes in established disease. More broadly, the work reframes Long COVID as a disorder of failed inflammatory resolution, a concept that may ultimately connect viral persistence, platelet biology, and fat metabolism into a single mechanistic story, and offers researchers a biochemical handle on one of the pandemic’s most stubborn legacies.

Subject of Research: Lipoxygenase-catalyzed metabolism of essential polyunsaturated fatty acids in Long COVID

Article Title: Long COVID is associated with lipoxygenase-catalyzed metabolism of essential polyunsaturated fatty acids

Article References: Long COVID is associated with lipoxygenase-catalyzed metabolism of essential polyunsaturated fatty acids. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: Long COVID, SARS-CoV-2, lipoxygenase, polyunsaturated fatty acids, specialized pro-resolving mediators, lipidomics, inflammation resolution, resolvin D1, platelets, spike protein, 12-HETE, iScience