The REAL Science of Diets: Keto, Paleo, Vegan and More | Alexis Cowan

Cowan frames this as keeping fasting pathways active — if ketogenesis enzymes are never exercised because every meal is carbohydrate-dominant, the enzymatic machinery downregulates over years. Periodic high-fat mornings re-engage the pathway enough to maintain it. She suggests pushing carbohydrates toward later in the day when glycogen demands from activity have already been incurred, so dietary glucose is directed toward refilling muscle glycogen rather than flowing into storage.

Cowan describes the beta-cell concern as 'still emerging' but significant enough to warrant a pragmatic hedge. The underlying evolutionary logic: humans in any ancestral environment would have encountered some carbohydrate-containing foods, so the body was not designed to operate in strict carbohydrate exclusion indefinitely. Cyclic keto captures the fasting-mimetic metabolic benefits while mimicking the natural feast-famine cycle that the pancreas and thyroid evolved around.

Cowan notes that the original protocol used fresh apple peels, but apple peel powder supplements are now available because the powder is more practical than peeling six apples per day for an effective dose. The HMO compound 2-fucosyllactose is the most widely available form; Cowan anticipates 5–10 more HMO variants entering the supplement market in coming years. Layer Origin is the company she works with as scientific content director. She emphasizes that the gut protocol is additive, not restrictive — an important psychological distinction for people with fraught histories around food elimination.

Cowan distinguishes magnesium L-threonate from the inorganic form (magnesium oxide) and other organic chelates (glycinate, malate): all organic forms have better bioavailability than the oxide, but L-threonate has the unique CNS-penetrating property. The electrolyte issue is severe enough in early keto that it can be mistaken for cardiac symptoms; Lyon mentions using LMNT electrolyte packets (1000mg sodium, 200mg potassium, 60mg magnesium) for travel cramping, illustrating that sodium management is an equally important parallel element.

Cowan personally used a combined glucose + ketone meter while experimenting with the ketogenic diet, which let her correlate how she felt with both blood glucose and ketone readings simultaneously. She found the pairing more informative than either measure alone and recommends it as the most data-rich self-experiment a person can run before committing to an extended dietary protocol. A continuous glucose monitor adds further resolution by showing glucose trajectory in response to individual meals, sleep, and exercise.

Cowan's pivotal observation: when she first tried to reintroduce dairy after 2–3 years of elimination — before discovering the gut protocol — she experienced throat swelling and a full-body hot sensation. This was more severe than her original UC trigger response and suggests the elimination period had not healed the underlying gut barrier. After six months on the gut protocol, the same food produced no response. The mechanistic interpretation: the gut protocol restored bifidobacteria and akkermansia, which rebuilt butyrate production, which healed colonocyte integrity, which reduced LPS-mediated systemic inflammatory priming — so the immune system's response to dairy peptides was no longer amplified by a background inflammatory state.

Cowan warns explicitly against calcium supplementation as a first-line intervention because chronically elevated serum calcium is associated with arterial calcification and atherosclerosis. The indirect route — vitamin D to improve calcium absorption from dietary sources, K2 to handle calcium routing — is safer and effective without the calcification risk. On a high-fat diet, fat-soluble K2 absorption is enhanced because fat in the gut facilitates micellar uptake of fat-soluble vitamins.

The rodent ketogenic diet used in the study was a very-high-fat classic formulation (~90% fat, ~8% protein), closer to the clinical epilepsy protocol than what most self-experimenters actually follow. Muscle substrate breakdown was roughly 30–40% fatty acids, 10–20% ketones, with the remainder lactate and amino acids. Brain tissue continued burning significant lactate and glucose even after 6–8 weeks of keto — aligning with Cahill's famous 45-day human fast studies from Harvard in the 1960s, where subjects' glucose levels fell to 2 mmol/L without loss of consciousness, because ketones had substantially taken over brain energy supply. Critically, Cowan notes that rodents have a ~10x higher metabolic rate than humans, and that standard lab chow is casein-based and vegetarian, limiting the direct translation of any keto rodent data to human physiology.

Cowan's core methodological critique: (1) LDL-C is a poor proxy for cardiovascular risk compared to apoB or oxidized-LDL; the drop seen on vegan diet is well-known to be transient with any dietary change from a Western baseline and re-equilibrates; (2) most study participants were likely transitioning from a standard American diet, so the improvement reflects ultraprocessed food removal rather than a diet-specific signal; (3) the study did not measure markers Cowan would have chosen — apoB, CRP, or oxLDL to capture inflammatory status. Lyon adds that the vegan arm's lean-mass loss may not have been skeletal muscle alone — could have included liver or organ tissue — since the study did not specify compartment.

Cowan is careful to note that the research is still emerging and the mechanistic explanation is not confirmed. The evolutionary argument: in any natural setting, an organism would not sustain strict carbohydrate exclusion for a full year; the body would inevitably encounter some carbohydrate-containing foods. Beta cells are designed to be periodically activated by glucose and amino-acid-driven insulin signals; indefinite non-stimulation is not a state the system was designed to tolerate. Cowan's practical recommendation in response: a cyclic ketogenic approach — introducing one or two higher-carbohydrate meals per week — preserves the metabolic benefits of keto while maintaining beta-cell function and avoiding the thyroid stress (elevated reverse T3) also associated with sustained very-low-carb eating.

The mechanistic cascade: HMOs (e.g., 2-fucosyllactose, the currently most available form) → bifidobacteria proliferation → acetate + lactate production → butyrogenic bacteria stimulation → butyrate → colonocyte fuel → intact gut barrier → reduced LPS leakage → less systemic inflammation. Apple peel powder (specific anthocyanin pigments) and red fruit powder (grape extract) also feed akkermansia, a separately important gut microbe associated with metabolic health. Cowan personally buys meat from regenerative local farms and receives weekly seasonal produce deliveries, emphasizing that the animal-protein portion of the diet supports small intestinal enterocytes (which primarily use amino acids for energy) while plant-fiber portion supports colonic microbiome and butyrate production — arguing these are complementary rather than competing priorities.

The evolutionary frame: humans are the most metabolically adaptable mammal on earth, capable of surviving on almost any macronutrient ratio. That adaptability depended on cycling through feast-famine — eating what was available, then going without. Chronic feeding without ever engaging fasting pathways is the novel stressor, not occasional fasting. Cowan also notes the parallel to muscle glycogen management: if glycogen pools in muscle never turn over, the muscle becomes a 'full sponge' that cannot take up additional glucose or fatty acids — impairing both metabolic flexibility and contractile function.

The clinical implication: individuals on a high-protein diet effectively outsource blood glucose supply to gluconeogenesis rather than dietary carbohydrates. The rate is more regulated and produces a smaller, slower insulin response than equivalent carbohydrate grams — which Cowan and Lyon both characterize as favorable for body composition, insulin sensitivity, and satiety. Lyon adds the caveat that intense exercise still depletes glycogen faster than gluconeogenesis can replenish it, so carbohydrate timing around training remains relevant for performance athletes regardless of their baseline dietary pattern.

Akkermansia lives in the mucus layer of the gut and helps maintain mucus integrity — a first barrier before the epithelial lining itself. When akkermansia is depleted (by antibiotics, processed foods, pesticides), mucus thins, enterocytes are more exposed, and the LPS-translocation pathway opens. Stool testing can assess relative akkermansia levels, but absolute reference ranges do not exist — the relevant question is whether your own levels are improving over time in response to interventions. Cowan distinguishes akkermansia-feeding (apple peel polyphenols) from bifidobacteria-feeding (HMOs) as separate but complementary arms of the gut protocol.