Genome-Wide Association Study of Plasma Polyunsaturated Fatty Acids in the InCHIANTI Study

If two people eat identical diets — the same fish, the same oils, and the same portion sizes — you might expect their blood levels of omega-3 fatty acids to look roughly the same. They don't. Some individuals convert dietary fats into long-chain fatty acids with striking efficiency. Others, eating the exact same food, end up with dramatically lower circulating levels. A single spot in the genome, near a gene called FADS1 on chromosome 11, explains nearly 19 percent of that variation. Before we get to that number, it helps to understand what these fatty acids actually do and why one genetic variant can matter so much. Polyunsaturated fatty acids, or PUFAs, are fats with two or more double bonds in their carbon chains. The two major families are omega-6 and omega-3, classified by where the first double bond sits relative to the methyl end of the molecule. Your body can't make the essential precursors from scratch — linoleic acid for omega-6s and alpha-linolenic acid for omega-3s have to come from food. But from those two starting points, the body can build longer, more biologically active molecules: arachidonic acid, eicosapentaenoic acid, and docosahexaenoic acid. Tanaka and colleagues describe these long-chain PUFAs as central to energy production, the modulation of inflammation, and the maintenance of cell membrane integrity. They also compete with each other for the same enzymatic machinery, so the balance between omega-3 and omega-6 products influences inflammatory signaling, eicosanoid production, and cardiovascular risk. Epidemiological studies have linked higher plasma PUFA concentrations, particularly omega-3s, to reduced risk of cardiovascular disease, diabetes, and cognitive decline. However, dietary questionnaires are unreliable. People misremember what they ate. Plasma measurements are better because they capture both what you consumed and what your body actually converted, which turns out to be a crucial distinction. That conversion process is what makes the genetics interesting. The pathway from dietary precursors to long-chain PUFAs runs through a series of enzymatic steps, alternating between desaturation and elongation. The first and rate-limiting step is delta-6 desaturation, carried out by an enzyme encoded by the FADS2 gene. After that, a series of elongation enzymes add carbon pairs to the chain, and further desaturation steps shape the final products. ELOVL2, an elongase, handles the two critical elongation steps that push eicosapentaenoic acid toward the longer omega-3 products, including docosahexaenoic acid. These enzymes are bottlenecks. Variation in the genes that encode them should show up directly in blood fatty acid levels, and the InCHIANTI data confirm exactly that. The InCHIANTI study is a population-based aging cohort in the Chianti region of Tuscany, Italy. Tanaka and colleagues measured plasma levels of six fatty acids in one thousand seventy-five participants, ranging from 21 to 102 years of age, with a mean age of around 68, using gas chromatography after overnight fasting. They genotyped participants on the Illumina HumanHap550 chip, covering roughly 495,000 genetic variants, and ran a genome-wide association study, a systematic scan asking which of those variants correlate with each fatty acid level. Before any scanning, the heritability estimates justified the effort. Arachidonic acid showed narrow-sense heritability — the proportion of variation explained by additive genetic factors — of 37.7 percent. Linoleic acid came in at 35.9 percent, eicosadienoic acid at 33.3 percent, and eicosapentaenoic acid at 24.4 percent. Docosahexaenoic acid was lowest at 12 percent. Substantial heritability across the board shows that genes are clearly in play. The biggest signal in the genome came from a single nucleotide polymorphism, a one-letter DNA variant, called rs174537, located near FADS1 on chromosome 11. The association with arachidonic acid had a p-value of five point nine five times ten to the negative forty-six. That's not a typo. And rs174537 alone explained 18.6 percent of the additive variance in plasma arachidonic acid concentrations. To put that in perspective, complex traits are typically influenced by dozens or hundreds of variants, each explaining a fraction of a percent. A single common variant explaining nearly a fifth of a blood trait is remarkable. The allele effect is visible in the raw numbers. Participants homozygous for the major G allele had mean arachidonic acid levels of 8.72 percent of total fatty acids. Heterozygotes came in at 7.39 percent. And minor-allele T/T homozygotes averaged 6.35 percent. That's a meaningful biological gradient across three genotype groups. The minor T allele was associated with lower levels of the long-chain products, arachidonic acid, eicosapentaenoic acid, and eicosadienoic acid, and higher levels of the upstream precursors, linoleic acid and alpha-linolenic acid. That pattern is exactly what you'd expect if the T allele impairs the desaturation step, leaving precursors to accumulate while downstream products fall. The same variant also associated with eicosapentaenoic acid at a p-value of one point zero four times ten to the negative fourteen, and with eicosadienoic acid at a p-value of six point seven eight times ten to the negative nine. Then comes the cardiovascular connection. Tanaka and colleagues found that rs174537 associated with both total cholesterol and low-density lipoprotein cholesterol. Minor-allele homozygotes had roughly 8 milligrams per deciliter lower total cholesterol and 9 milligrams per deciliter lower low-density lipoprotein compared to major-allele homozygotes. High-density lipoprotein and triglycerides were unaffected. The allele that depresses arachidonic acid and eicosapentaenoic acid also lowers low-density lipoprotein, a connection that threads fatty acid metabolism directly into cardiovascular risk pathways. Beyond the FADS cluster, the next strongest signal came from chromosome 6, in the region encoding the elongase ELOVL2. The variant rs953413 associated with plasma eicosapentaenoic acid in InCHIANTI at a p-value of one point one times ten to the negative six. In that sample, the A allele linked to higher eicosapentaenoic acid and lower docosahexaenoic acid, consistent with ELOVL2's role. If elongation from eicosapentaenoic acid onward is less efficient, eicosapentaenoic acid accumulates and docosahexaenoic acid falls. Replication is where the story gets its texture. Tanaka and colleagues tested both hits in an independent sample of one thousand seventy-six subjects from the GOLDN study. The FADS1 signal replicated powerfully — rs174537 showed significant associations with arachidonic acid, eicosapentaenoic acid, docosahexaenoic acid, linoleic acid, and alpha-linolenic acid in GOLDN, all in the same allelic direction, with p-values below 0.001 across the board. Total and low-density lipoprotein cholesterol associations replicated as well. Minor-allele homozygotes in GOLDN showed lower total cholesterol and lower low-density lipoprotein, confirming the lipid connection in an independent dataset. The ELOVL2 result was more nuanced. In GOLDN, rs953413 showed no significant association with eicosapentaenoic acid, the substrate that drove the InCHIANTI signal. Instead, it associated with docosapentaenoic acid and docosahexaenoic acid, the downstream elongation products. That's not a failed replication — it's a refinement. InCHIANTI measured plasma fatty acids; GOLDN measured erythrocyte membranes, which reflect a different time window. The mechanistic interpretation holds: ELOVL2 governs elongation steps further along the omega-3 chain, so depending on where in the pathway you sample and when, you catch the effect at different points. The association moved from substrate to product, which is what a bottleneck enzyme should do. Taken together, these findings shift how we should think about plasma fatty acid measurements. They are not a clean readout of diet. They reflect diet filtered through the lens of individual metabolic capacity, and that capacity is substantially genetic. Two people eating identical omega-3-rich diets can end up with meaningfully different plasma profiles depending on their FADS genotype. That helps explain why observational studies linking dietary fatty acids to cardiovascular outcomes have been inconsistent. Some of what looks like dietary variation is actually genetic variation in conversion efficiency. Tanaka and colleagues also point to other loci, including a docosahexaenoic acid-associated variant on chromosome 12, suggesting that the genetic architecture of PUFA metabolism extends beyond FADS and ELOVL2. What remains unresolved is how these genotype effects play out in specific tissues. InCHIANTI measured plasma; GOLDN measured erythrocytes; neither tells us what's happening in the liver, brain, or vascular wall, where PUFA biology may matter most. What this study establishes is a framework: genetic variants in the enzymes that build long-chain fatty acids are major determinants of what circulates in your blood, and those variants associate with the same lipid traits that cardiovascular medicine has spent decades trying to modify. Any serious effort to connect diet to disease, or to personalize nutritional advice, will need to account for the fact that the same meal means different things to different genomes. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.