Yes. AA is basically the paper’s central biochemical scaffold, especially for the oxidized phospholipid story. The authors explicitly introduce arachidonic acid (AA, 20:4) as a major substrate for lipid-oxidizing enzymes, then choose PAPC = PC(16:0/20:4) as their prototypical PUFA-containing membrane phospholipid. Oxidizing its arachidonoyl chain gives the whole oxPAPC menagerie: PEIPC, PECPC, PGPC, POVPC, KOdiAPC, etc. (PubMed)
But the review is almost an argument against collapsing AA into “pro-inflammatory fat.” AA can enter several very different biochemical universes:
- COX: prostaglandins/thromboxanes, some inflammatory, some homeostatic.
- LOX: leukotrienes, HETEs, but also lipoxins, which help terminate inflammation.
- CYP: EETs and related products, often vasoregulatory/cytoprotective.
- enzymatic oxidation while esterified in phospholipids: 12-HETE-PE, 15-HETE-PE, etc.
- uncontrolled radical oxidation of AA-containing membranes: oxPAPC species, whose effects themselves range from NRF2 activation/barrier protection to inflammasome activation, ferroptosis and tissue injury.
The review explicitly says oxidized lipids can have divergent effects depending on molecular structure and cellular target, and summarizes oxPLs as protective when transient but pathological when they accumulate. (PubMed)
So a better conceptual picture is:
AA = a substrate reservoir, not an inflammatory instruction.
What happens depends on which enzyme or radical gets to it, where that occurs, and what the receiving cell expresses.
For example, AA released by PLA₂ into one cellular context might go:
AA → 5-LOX → LTB4 → neutrophil recruitment
while another enzymatic configuration can push toward:
AA → lipoxin pathway → LXA4 → resolution/barrier protection
and an oxidatively stressed membrane can instead go:
PAPC → radical oxidation → POVPC/PGPC/etc. → DAMP-like signaling / ferroptotic stress.
Those aren’t minor variations. They’re biologically almost opposite programs built from the same 20-carbon precursor.
And your nutrition point is important
I would modify it slightly: this context-dependence is one reason that “more PUFA → more inflammation” fails, but there are several additional reasons.
Most dietary omega-6 PUFA is linoleic acid (LA, 18:2), not AA. Human metabolism does not simply pour dietary LA downstream into AA in proportion to intake. A systematic review found that even quite large changes in dietary LA generally did not materially change plasma/erythrocyte phospholipid AA in adults eating typical Western diets. The LA→AA pathway is tightly constrained enzymatically. (PubMed Central (PMC))
And empirically, the simple inflammatory prediction does badly. A meta-analysis of 30 randomized trials / 1,377 participants found increasing dietary LA had no significant overall effect on CRP, TNF-α, IL-6, MCP-1, fibrinogen and several other inflammatory markers. (PubMed)
That’s a lovely example of why biochemical pathway diagrams can mislead:
LA → AA → PGE₂
is chemically true.
Therefore,
more LA → more AA → more PGE₂ → chronic inflammation
does not automatically follow physiologically.
There are rate limits, membrane remodeling, competing pathways, feedback regulation, receptor expression, substrate competition, redox state and active resolution pathways in between. Biology inserts about twelve bureaucratic departments between the arrows.
Also, “PUFA” is an enormous category. LA, AA, ALA, EPA and DHA don’t behave interchangeably. And health outcomes depend strongly on what the PUFA replaces. Replacing saturated fat with unsaturated fats, including PUFA, is still recommended for cardiovascular risk reduction. (JACC)
This paper actually makes the AA story even stranger
The nonenzymatic branch means that you can’t even assign a sign to oxidized AA.
Consider just PAPC:
PAPC (AA-containing)
↓ oxidationPEIPC → NRF2 ↑, endothelial barrier ↑, LPS signaling ↓ but also chemokines/inflammasome effects
PGPC → can enhance DC migration/CD8 priming but also ferroptosis/barrier damage
POVPC → generally much more damaging, ferroptotic/reactive
KOdiAPC → low-dose TLR4 antagonism but higher-dose inflammatory/barrier effects
So the relevant variable isn’t really:
“How much arachidonic acid?”
It’s closer to:
AA × lipid compartment × oxidation route × enzyme expression × redox environment × dose × receptor repertoire × cell type × time.
And I think that is actually the intellectually interesting takeaway from this review. AA isn’t intrinsically pro- or anti-inflammatory. It gives the immune system an unusually large vocabulary. Chronic oxidative stress can make that vocabulary ugly, while controlled enzymatic metabolism can use essentially the same chemical feedstock for inflammation, homeostasis or resolution. (PubMed)
One further implication: this makes the question you originally asked about oxidized fish oil substantially different from “are PUFAs bad?” Oxidation state and which oxidation products formed may matter more than the mere fact that a fatty acid is highly unsaturated.
Exactly. The confusion comes from the fact that the AA, EPA, and DHA pathways use almost the same enzymatic machinery and produce nearly parallel “hydroxy-PUFA” intermediates.
But chemically:
- AA (20:4) → HETE: hydroxy-eicosatetraenoic acid
- EPA (20:5) → HEPE: hydroxy-eicosapentaenoic acid
- DHA (22:6) → HDHA: hydroxy-docosahexaenoic acid
So your intuition is basically right, with one letter changed. Fish oil produces resolvins through HEPE/HDHA intermediates analogous to the HETEs produced from AA. (PubMed Central (PMC))
The three pathways line up beautifully
Starting PUFA Important first oxygenated intermediates Major downstream programs AA, 20:4 ω-6 5-HpETE/HETE, 12-HETE, 15-HpETE/15-HETE, etc. leukotrienes, lipoxins, HETE signals EPA, 20:5 ω-3 18-HpEPE → 18-HEPE, 15-HpEPE/15-HEPE E-series resolvins DHA, 22:6 ω-3 17-HpDHA → 17-HDHA, 14-HpDHA/14-HDHA D-series resolvins, protectins, maresins This is why I wouldn’t describe fish oil’s action simply as “replacing inflammatory AA.” It’s more interesting: you alter what substrates the same lipid-oxidation network has available.
AA itself has a pro-resolving HETE branch
The poster child is 15-HETE.
A simplified route is:
AA
→ 15-LOX
→ 15S-HpETE
→ 15S-HETE
→ another cell’s 5-LOX
→ epoxide intermediates
→ LXA₄ / LXB₄Those are lipoxins, among the original specialized pro-resolving mediators. Epithelial cells can make/donate 15-HETE, for example, and leukocytes subsequently transform it into lipoxins. (PubMed Central (PMC))
So even within pure omega-6 AA metabolism:
AA → inflammation is wrong.
AA can produce leukotrienes on one branch and actively inflammation-resolving lipoxins on another.
That’s a wonderfully biological arrangement. The substrate doesn’t say “attack” or “stop.” The enzymatic sequence writes the sentence.
Some major HETEs
5-HETE sits near the classic 5-LOX inflammatory pathway:
AA → 5-HpETE → LTA₄ → LTB₄/LTC₄/etc.
5-HETE itself can also become 5-oxo-ETE, a potent leukocyte chemoattractant. This side of AA metabolism is generally much more inflammation-amplifying.
12-HETE is especially associated with platelet 12-LOX and has roles in platelet activation, vascular biology, leukocyte behavior and other signaling. It is not inherently “bad,” but frequently appears on the inflammatory/vascular signaling side.
15-HETE is particularly interesting because it can do several things. It has direct signaling activity, can appear esterified into phospholipids as 15-HETE-PE, and can serve as substrate for the lipoxin/resolution pathway.
And then there’s 20-HETE, which isn’t primarily a LOX product at all. CYP450 ω-hydroxylases convert AA into 20-HETE. It is a potent vascular and renal signal, often producing vasoconstrictive/pro-inflammatory effects, but again its phenotype is tissue-dependent.
There are also 8-, 9-, 11-HETE and various stereoisomers, some generated enzymatically and some through nonenzymatic oxidation. That’s another reason “HETE concentration” without specifying position + stereochemistry isn’t terribly informative.
Fish oil essentially gives this network new Lego bricks
Here’s the EPA pathway you were remembering:
EPA
↓ COX-2/CYP and related oxygenation
18-HpEPE
↓ reduction18-HEPE
↓ 5-LOX / 12-LOX / 15-LOX
RvE1, RvE2, RvE318-HEPE is the major recognized biosynthetic hub for E-series resolvins. RvE1 formation is especially associated with conversion of 18R-HEPE through leukocyte 5-LOX. (PubMed Central (PMC))
There’s another EPA route:
EPA → 15S-HpEPE → 15S-HEPE → RvE4
so even multiple HEPE positional isomers can be entrances into resolution biology. (PubMed Central (PMC))
And DHA does the same thing with different carbon numbering:
DHA
↓ 15-LOX17S-HpDHA
↓
17S-HDHA
↓ 5-LOX + epoxide chemistryRvD1, RvD2, RvD3, RvD4, RvD5, RvD6
The exact downstream branching is more complicated than that schematic, but 17-oxygenated DHA is the central gateway. (PubMed Central (PMC))
And:
DHA → 14-HpDHA / 14-HDHA → maresin pathway
so 14-HDHA plays a roughly analogous precursor-marker role for maresins. (PubMed Central (PMC))
There’s even an aspirin twist
This chemistry gets delightfully baroque.
Aspirin-acetylated COX-2 changes the stereochemistry of oxygen insertion:
AA → 15R-HETE → aspirin-triggered lipoxins
EPA → 18R-HEPE → aspirin-triggered/E-series resolvin pathways
DHA → 17R-HDHA → aspirin-triggered D-series resolvins
Same general oxygenation machinery, three substrates, three corresponding families. (PubMed Central (PMC))
That parallel is probably the connection you were remembering.
And this changes how I would characterize that Science Immunology review
The oxPAPC story I described earlier is really just one branch of AA oxidation:
AA esterified in phospholipid
→ oxidation
→ oxidized PC species such as PEIPC/PGPC/POVPC…But physiologically there’s an enormous parallel branch:
AA released / enzymatically accessible
→ regio- and stereospecific LOX/COX/CYP oxidation
→ HETEs, prostaglandins, leukotrienes, lipoxins…Those are very different kinds of oxidation.
And even esterified AA isn’t merely oxidative wreckage. The review discusses 12-HETE-PE and 15-HETE-PE, which are deliberately generated enzymatic oxidized phospholipids. So there’s almost a spectrum:
controlled enzymatic oxygenation
→ exquisitely specified molecular signalsversus
uncontrolled radical lipid peroxidation
→ heterogeneous oxPL mixtures + fragmentation + electrophilesrather than a clean unoxidized = good / oxidized = bad binary.
That distinction also explains a lot about the fish-oil literature. EPA/DHA being more oxidation-prone chemically does not imply that consuming them must be harmful. Their susceptibility to controlled oxygenation is precisely what permits formation of 18-HEPE, 17-HDHA and then resolvins/protectins/maresins. Random autoxidation and enzyme-directed oxygenation are chemically neighboring processes with profoundly different biological meanings. (PubMed)
There is one important scientific wrinkle: how quantitatively important endogenous resolvin production is in humans remains debated, especially because some SPM measurements are near analytical detection limits and different lipidomics methods have produced conflicting estimates. One critical review specifically questions how confidently some proposed resolvin pathways and concentrations have been established in vivo. (PubMed Central (PMC))
So the strongest claim is not “fish oil works because it becomes resolvins.” It’s:
EPA/DHA provide substrates for a parallel oxygenated-lipid network that includes HEPE/HDHA intermediates and potent pro-resolving mediators; simultaneously they alter competition with AA for membrane incorporation and COX/LOX/CYP metabolism.
That is much closer to what the biochemistry actually looks like.
Yes. I went through the review’s named species and Table 2 rather than treating “oxidized lipids” as one blob. The striking thing is that “oxidized lipid = bad” is very much not the paper’s message. The same oxidation family can be protective at one concentration/cell type and inflammatory, ferroptotic, or immunosuppressive in another. The authors explicitly frame oxPL biology as double-edged. (PubMed)
One scope note: below I count every explicitly named oxidized lipid/species or defined oxidized-lipid preparation in the review text/Table 2, rather than every molecule buried in the titles of its 200+ references.
The main inventory
Oxidized lipid Immediate precursor Potentially beneficial / protective effects reported Potentially harmful / pathogenic effects reported oxPAPC PAPC, 1-palmitoyl-2-arachidonoyl-PC = PC(16:0/20:4), therefore arachidonic acid (AA) is the oxidizable chain At lower/context-dependent exposure: blocks LPS/TLR responses; suppresses E-selectin and neutrophil adhesion; activates NRF2 antioxidant program; can enhance pulmonary endothelial barrier; prevents/reverses TNF barrier disruption; induces protective lipoxin A4; can reduce Treg suppressive phenotype in vitro Probably the most Janus-faced molecule in the review: induces CXCL8/CCL2/IL-6/CXCL1; promotes monocyte adhesion; NLRP3/caspase-11 hyperactivation and IL-1β; suppresses IL-10 through AKT/EZH2; impairs macrophage phagocytosis; disrupts barrier at high dose; inhibits DC/T-cell activation; induces T-cell anergy/exhaustion; promotes angiogenesis/nucleotide release; associated with MDSC recruitment and tumor immune suppression (Scribd) PEIPC PAPC / arachidonate-containing PC Strong endothelial barrier enhancement via Rac; NRF2 activation; suppresses LPS-induced E-selectin and some TLR inflammatory output; can increase IL-10 in monocytes Can induce CXCL8/CCL2 and UPR signaling; inflammasome/IL-1β hyperactivation; inhibits IL-12 and vitamin-D antimicrobial response to M. tuberculosis ligand; interferes with antigen presentation/T-cell activation (Scribd) PECPC PAPC / arachidonate Activates NRF2; reduces TLR-induced IL-6/IL-12 in DCs Skews DC-driven T-cell responses away from IFN-γ toward IL-4, potentially suppressing productive cellular immunity (Scribd) PGPC Oxidative fragmentation of PAPC / arachidonate Suppresses LPS-induced E-selectin; in DCs, can cause nonpyroptotic inflammasome activation that increases CCR7/migration and can improve antigen-specific CD8 responses, which the review highlights as potentially useful for cancer vaccination ROS and ferroptosis in endothelium; pulmonary barrier disruption and inflammatory gene induction; NLRP3/IL-1β hyperactivation; T-cell proliferation inhibition; LRP6 internalization and impaired osteoblast differentiation. Can promote monocyte/endothelial interactions (Scribd) POVPC Oxidative fragmentation of PAPC / arachidonate Suppresses LPS-induced E-selectin/neutrophil adhesion in some endothelial experiments; can antagonize inflammatory signaling under some conditions One of the nastier truncated products: barrier failure, inflammatory cytokine expression, mtROS/NLRP3, pulmonary epithelial ferroptosis, endothelial/VSMC ferroptosis, neuronal/oligodendrocyte toxicity, vascular calcification, EMT/autophagy and increased tumor-cell migration (Scribd) KOdiAPC PAPC / arachidonate oxidation-fragmentation At low concentrations blocks TLR4 signaling and reduces LPS-induced TNF, IFN-β and PGE2; suppresses E-selectin At higher/context-dependent exposure disrupts lung endothelial barrier and induces TNF/IL-6/CXCL8/IL-1β-family inflammatory programs (Scribd) PONPC Primarily PLPC, 1-palmitoyl-2-linoleoyl-PC = PC(16:0/18:2), hence linoleic acid rather than AA No major positive species-specific effect emphasized beyond oxPL antagonism of E-selectin/TLR responses Endothelial barrier disruption and induction of inflammatory endothelial genes (Scribd) PAzePC / PAzPC PLPC / linoleic acid, downstream oxidation of PONPC Inhibits LPS-induced E-selectin Endothelial inflammatory gene expression and barrier disruption; POVPC/PAzePC exposure is directly toxic to neurons and oligodendrocytes. Microglia can scavenge it and protect neurons (Scribd) PAPC-OOH PAPC hydroperoxidation Inhibits LPS-induced endothelial E-selectin in the cited experiments The review does not assign a separate major pathological phenotype to this isolated species (Scribd) PAPC-OH Monohydroxylated PAPC Likewise suppresses LPS-induced E-selectin No separate major adverse phenotype assigned in this review (Scribd) oxPAPE 1-palmitoyl-2-arachidonoyl-PE, therefore AA-containing PE Suppresses LPS-induced endothelial E-selectin No distinct harmful effect separated from the general oxPL pool in Table 2 (Scribd) oxPAPG 1-palmitoyl-2-arachidonoyl-PG / AA Suppresses LPS-induced E-selectin; activates NRF2; inhibits neutrophil oxidative burst Promotes endothelial migration/angiogenesis; inhibits activated human T-cell proliferation (Scribd) oxPAPS 1-palmitoyl-2-arachidonoyl-PS / AA Suppresses E-selectin; NRF2 activation; suppresses neutrophil oxidative burst Angiogenesis/migration; T-cell proliferation suppression (Scribd) oxPAPA 1-palmitoyl-2-arachidonoyl-phosphatidic acid / AA Suppresses E-selectin; NRF2 activation Angiogenesis/migration; T-cell proliferation suppression (Scribd) oxPLPC PLPC = PC(16:0/18:2), linoleic acid Suppresses stimulated neutrophil oxidative burst No distinct pathological effect separately developed in the review (Scribd) 15-HETE-PE AA-containing PE, oxygenated by LOX Physiological membrane-tethered signaling; regulation of platelet activation/coagulation and innate immune responses Depending on setting can alter/promote inflammatory neutrophil responses. The review treats this as a regulated eoxPL rather than simply “oxidative damage” (Scribd) 12-HETE-PE AA-containing PE via LOX Regulated platelet/immune-cell signaling; local control of platelet/coagulation responses Can participate in innate inflammatory/neutrophil activation (Scribd) 12-HETE-PC AA-containing PC via LOX Same broad regulated eoxPL role Same context-dependent innate immune effects; not extensively separated from 12-HETE-PE (Scribd) 15-HpETE-SAPE SAPE = PE(18:0/20:4), i.e. stearate + AA, oxidized to 15-hydroperoxy-AA while still esterified The review does not identify a beneficial role in its cancer discussion Strongly relevant to tumor immunology: reduces activated T-cell proliferation; associated with ferroptotic/MDSC lipid-mediated immunosuppression (Scribd) 15-HpETE-SAPC SAPC = PC(18:0/20:4), AA-containing PC None emphasized Reduces antigen-specific T-cell proliferation (Scribd) oxidized SAPE more generally SAPE, PE(18:0/20:4) / AA Species-dependent, not resolved Suppresses activated human/mouse T-cell proliferation (Scribd) oxidized SAPC more generally SAPC, PC(18:0/20:4) / AA Species-dependent, not resolved Suppresses activated human/mouse T-cell proliferation (Scribd) That yields 20-ish chemically distinguishable named oxPL entities, depending on whether you count the broad oxidized parent preparations separately from their individual constituents.
A caveat that matters a lot: oxPAPC is not a molecule. It is a chemically heterogeneous oxidation mixture. PEIPC, PECPC, PGPC, POVPC, KOdiAPC and numerous unnamed products can coexist in it. Consequently, apparently contradictory oxPAPC findings are not necessarily contradictory biology. They can reflect different oxidation batches, dose, oxidation time, composition, cell type and receptor usage. The authors actually flag heterogeneous preparations and supraphysiological concentrations as a major weakness of the literature. (Scribd)
The other oxidized lipids named in the review
There are a few important molecules outside the oxPL table.
PGE2 (prostaglandin E2) is an enzymatic AA-derived oxylipin rather than a spontaneous oxPL. In the tumor setting the review mostly paints it as harmful: EP2/EP4 signaling suppresses CD8 expansion and survival, reduces metabolic fitness, promotes T-cell ferroptosis, makes cDC1s dysfunctional, and drives cDC2s toward IL-10/IL-23 production and suppressive Treg induction. COX-1/2 blockade or PGE2 deficiency improves immune tumor control in the cited work. (Scribd)
Lipoxin A4 (LXA4) is also AA-derived, but is on the other side of the ledger. The review specifically describes oxPAPC inducing LXA4, which then acts autocrinely/paracrinely as an anti-inflammatory, barrier-protective mediator, preventing TNF/LPS-induced endothelial permeability. (Scribd)
The introduction additionally names the broad AA-derived oxylipin families prostaglandins, thromboxanes, leukotrienes, HETEs and epoxyeicosatrienoic acids (EETs). It does not enumerate individual members of those families, aside from the examples above, but notes that collectively they regulate both initiation and resolution of inflammation, hemostasis, and vascular tone. (Scribd)
Oxysterols are also discussed only as a class, not molecule-by-molecule. Their precursor is cholesterol/cholesterol-pathway sterols; the review says they regulate cell viability and immune responses mainly through liver X receptors. Later it notes oxysterols together with oxPLs as components of the tumor lipid environment affecting T-cell biology. No individual 25-HC, 27-HC, 7-ketocholesterol, etc. is actually developed in the text, so I would not pretend those are part of the review’s species inventory. (Scribd)
There are also oxidation products that aren’t really “oxidized lipids”
The paper mentions these because they help explain what PUFA peroxidation eventually does:
DHA → carboxyethylpyrrole (CEP) adducts. DHA oxidation yields reactive fragments that generate CEP-modified proteins/lipids. CEP itself is therefore better thought of as an oxidation-derived adduct/epitope, not an intact oxidized DHA molecule.
ω-6 PUFAs → malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE). Again, these are reactive lipid-peroxidation products rather than phospholipids. They covalently modify proteins and create oxidation-specific epitopes. (Scribd)
This distinction is biologically important. The intact oxidized phospholipid can act as a receptor ligand; its chopped-off aldehyde descendants can instead behave almost like molecular graffiti, covalently modifying proteins and changing what the immune system sees.
So, oxidized fish oil?
Not really.
There are three nearby concepts in the paper, but they shouldn’t be conflated:
DHA is mentioned. DHA is abundant in fish oil, but the review only notes DHA oxidation as an example of generating CEP adducts. It does not discuss rancid fish oil, fish-oil capsules, peroxide values, anisidine values, EPA/DHA supplement stability, or health effects of consuming oxidized fish oil. EPA is not discussed by name. (Scribd)
The paper cites a dietary study where oxidized PUFA aggravated colitis and colitis-associated cancer in mice. But that experiment used oxidized corn oil, compared with purified unoxidized corn oil and lard. It is therefore mostly an oxidized linoleic-acid-rich vegetable-oil model, not oxidized fish oil. (Scribd)
Mechanistically, fish oil is certainly eligible for this chemistry. EPA and especially DHA are extremely unsaturated and can generate hydroperoxides, aldehydes, fragmented phospholipids and other oxidation products. But adv9397 does not establish that eating normally stored fish-oil supplements produces the oxPAPC/PGPC/POVPC biology it reviews. Those are predominantly AA/linoleate-containing phospholipid stories.
And there is a neat asymmetry here: the paper’s main protagonist is arachidonic-acid-containing membrane phospholipid, not free dietary oil. PAPC is PC(16:0/20:4), embedded in membranes. Most of the named oxPL cast therefore descends from AA, while PONPC/PAzePC/oxPLPC provide the major linoleate branch. DHA is barely a cameo.
If you’re ultimately interested in “which oxidation products in actual commercial fish oil are biologically concerning versus potentially beneficial?”, that is a somewhat different and very interesting inventory. It would involve EPA/DHA hydroperoxides, F3-isoprostanes, neuroprostanes, neuroketals, 4-HHE, MDA and various epoxy/hydroxy-DHA/EPA products, and importantly distinguish enzymatic pro-resolving mediators from random autoxidation products. That comparison would make the “oxidized omega-3 = bad?” question much less muddy.