Endpoint: surrogate — fasting plasma & urinary TMAO, chronic (~1 month) feeding, 3-arm source-comparison RCT, restricted to participants with normal renal function.
Healthy volunteers with normal renal function were randomized across ~1 month to 4 eggs/day, phosphatidylcholine capsules (matched choline dose), or choline-bitartrate tablets (also a fourth combined arm per registration NCT03039023). Only the free-choline-bitartrate arms significantly raised fasting plasma/urinary TMAO; both whole eggs and phosphatidylcholine capsules failed to raise TMAO significantly.
relevance_note: strongest single “eggs specifically don’t raise TMAO” primary in the pool because it comes from the same lab (Hazen/Tang) that built the foundational harm hypothesis — shows food-matrix/choline-form, not choline content per se, determines the TMAO response. Restricting to normal renal function also foregrounds the renal-confounding angle (see Gruppen 2017, this pool) as a deliberate design choice.
Methodology
Registered (NCT03039023) 5-arm parallel RCT, 4-week interventions, 82 of 89 completing (92%); healthy men and women with normal renal function (eGFR > 60, normal creatinine, no microalbuminuria); median age 28, 42% male. Arms: (1) 4 whole hardboiled eggs/day, 467 mg choline, n=18; (2) 2x500 mg choline-bitartrate tablets, 411 mg free choline, n=20; (3) 4 eggs + choline tablets, n=16; (4) 4 egg whites + choline tablets, n=18; (5) 6x420 mg phosphatidylcholine capsules, 410 mg choline, n=10. Outcomes: fasting plasma and 24-h/spot urinary TMAO, plasma choline, platelet responsiveness, lipid panel. Positive control = free choline; egg white arm tests egg-white antimicrobial peptides.
Results — TMAO by arm
O-62 - Four eggs-day for 4 weeks did not raise fasting TMAO or platelet reactivity in adults with normal renal function
The core egg-null: a substantial habitual dose (4 large eggs/day) did not move any TMAO readout or the downstream platelet-reactivity readout, in participants pre-selected for normal renal function.
Link to originalMethodology
Registered (NCT03039023) 5-arm parallel RCT, 4-week interventions, 82 of 89 completing (92%); healthy men and women with normal renal function (eGFR > 60, normal creatinine, no microalbuminuria); median age 28, 42% male. Arms: (1) 4 whole hardboiled eggs/day, 467 mg choline, n=18; (2) 2x500 mg choline-bitartrate tablets, 411 mg free choline, n=20; (3) 4 eggs + choline tablets, n=16; (4) 4 egg whites + choline tablets, n=18; (5) 6x420 mg phosphatidylcholine capsules, 410 mg choline, n=10. Outcomes: fasting plasma and 24-h/spot urinary TMAO, plasma choline, platelet responsiveness, lipid panel. Positive control = free choline; egg white arm tests egg-white antimicrobial peptides.
Link to original
O-63 - Free choline (choline bitartrate) raised fasting TMAO ~6-fold and increased platelet reactivity
The validated far-from-threshold positive control that anchors the trial: free choline at a dose comparable to the eggs’ choline content produced a large TMAO and platelet-reactivity response.
Link to original
O-64 - Phosphatidylcholine capsules at a matched choline dose did not raise TMAO or platelet reactivity
The phosphatidylcholine capsule arm delivers choline in the same phospholipid form as egg yolk but without the egg food matrix, and reproduces the egg null - isolating choline form as the relevant variable rather than the whole-egg matrix per se.
Link to original
O-67 - Adding egg whites to choline-bitartrate did not suppress the choline-driven TMAO rise
A designed mechanism-dissection arm: it rules out egg-white antimicrobial peptides (e.g., ovodefensins) as the explanation for the whole-egg TMAO null.
Link to original
Results — compliance and platelet correlation
O-65 - Plasma choline rose significantly in every arm including the egg arm, confirming choline was absorbed
Key control finding: the egg TMAO null is not due to non-compliance or failed choline absorption - circulating choline rose in the egg arm just as in the TMAO-raising choline-tablet arms.
Link to original
O-66 - Platelet responsiveness correlated with TMAO levels across participants
Links the TMAO differences to a functional prothrombotic readout: participants with higher TMAO had higher platelet reactivity, regardless of arm.
Link to original
Results — lipids
O-68 - No lipid-panel component changed in any arm, including 4 eggs-day for 4 weeks
Secondary lipid finding relevant to the broader eggs question: at 4 eggs/day for a month in healthy young adults, no measurable lipid-panel change.
Link to original
Discussion
H-28 - The form-food-matrix of choline, not total choline content, determines TMAO generation
The authors’ central mechanistic claim, discriminating between two rival TMAO models. Uncertain (the microbiome step is inferred, not directly measured here).
Link to original
H-29 - Habitual whole-egg consumption does not meaningfully raise TMAO in people with normal renal function
A candidate answer to the harm-mechanism sub-question, notable for coming from the lab (Hazen/Tang) that built the TMAO-harm hypothesis.
Link to original
H-30 - Impaired renal function may modify the TMAO response to egg-choline intake
A for-whom candidate answer flagged by the authors as a limitation; the trial deliberately restricted to normal renal function, so this remains untested here.
Link to original
A-22 - Choline was absorbed in the egg arm yet TMAO did not rise, isolating choline form (not dose or compliance) as the TMAO determinant
The inference chains four arm-level facts into a mechanism attribution. (1) The egg arm delivered ~467 mg choline and the choline-bitartrate arm ~411 mg — comparable choline doses. (2) Free choline (bitartrate) raised fasting TMAO ~6-fold (O-63), establishing that the assay and the gut-microbial TMA pathway are responsive at this dose. (3) Eggs did not raise TMAO (O-62); crucially, this is not because the choline was not consumed/absorbed — plasma choline rose comparably in the egg arm (O-65), ruling out non-compliance and mal-absorption as confounds. (4) Phosphatidylcholine capsules, delivering choline in the same phospholipid form as egg yolk but without the whole-egg matrix, also failed to raise TMAO (O-64), which points past “the whole-egg food matrix” to the phospholipid-bound form of choline itself as the operative variable. The only remaining variable that separates the TMAO-raising arm from the two null arms is the chemical form of choline (free vs phosphatidylcholine-bound), so the data force the form/bioavailability conclusion (supporting H-28) and, applied to whole eggs specifically, the egg-null harm-pathway conclusion (H-29). The controlled contrast (same dose, absorption confirmed) is what makes this a valid deduction rather than a bare association.
Validity (step 6): approved / checked. Reconstructed step: three candidate explanations for the egg TMAO-null are under-dosing, non-compliance/malabsorption, and choline form. The matched doses (~467 vs ~411 mg) close off under-dosing; the confirmed plasma-choline rise in the egg arm closes off non-compliance/malabsorption; the free-choline arm’s ~6-fold TMAO rise proves the assay and the gut TMA pathway are responsive at this dose, so a null is informative rather than a floor effect. With those alternatives eliminated, the sole variable separating the one TMAO-raising arm (free choline) from the two null arms (egg, phosphatidylcholine capsule) is chemical form/bioavailability — a clean four-arm eliminative contrast, not a bare association. I probed one undercutting defeater that does not deny the premises: TMAO generation depends on choline reaching colonic microbiota, and a systemic plasma-choline rise confirms absorption but not equal colonic delivery — however the statement’s conclusion is explicitly “form/bioavailability,” which is precisely the form-dependent colonic-availability channel, so this refines rather than breaks the step. The PC-capsule arm (same phospholipid form as yolk, no whole-egg matrix, also null) correctly pushes the conclusion past “food matrix” to the phospholipid-bound form itself. Conditional on the four observations, the isolation of form as the determinant follows; approved as stated. (Whether egg truly does not raise TMAO — small-trial, high-variance TMAO — is a premise-truth question priced downstream, not a validity flaw.)
Link to original
A-23 - Because TMAO is cleared almost exclusively by the kidney, the normal-renal egg-null cannot be extrapolated to renal impairment
Steady-state TMAO concentration is set by production minus clearance. TMAO is cleared almost exclusively via renal excretion (the study measures urinary TMAO precisely because the kidney is the dominant elimination route). The egg-null (O-62) was measured only in participants with eGFR > 60, normal creatinine, and no microalbuminuria — i.e., with intact clearance. Even if whole-egg intake produced a small amount of TMA/TMAO, intact kidneys would clear it, keeping fasting concentrations flat. In renal impairment the clearance term falls, so the same (or any) production could yield a higher steady-state concentration. Therefore the null is conditional on normal renal function and does not license a claim about renal-impaired or cardiovascular-disease populations - exactly the caution the authors state. This bears directly on H-30 (renal function as an effect modifier / for-whom qualifier) and bounds the scope of O-62.
Validity (step 6)
status: approved — reason_if_not_false: checked. Traced the step: the conclusion is a scope-limitation (a negative claim that the null does not license extrapolation), not a positive claim that TMAO will rise in renal impairment. Given the premises (steady-state = production/clearance; kidney is the near-exclusive clearance route; O-62 measured only in eGFR>60/no-microalbuminuria participants), the selection criterion coincides exactly with the hypothesized modifier, so a null observed in the intact-clearance stratum cannot rule out an effect in the reduced-clearance stratum. Probed for an undercutting defeater: “whole eggs produce ~zero TMA, so reduced clearance of zero stays zero” — this does NOT break the step, because it denies a premise (it asserts production is exactly zero, whereas the argument grants only “small amount,” and the conclusion is merely that the null is non-generalizable, which a near-zero-but-nonzero production still supports). No surviving defeater; the modest negative conclusion follows cleanly.
Link to original