It Wasn’t the Ketones: What Really Cut Liver Fat in the New Keto Study | Nicola Guess

Episode Overview

• Host Simon Hill (nutritionist, physiotherapist, author, and podcast host of The Proof) and guest Dr. Nicola Guess (a clinical and academic dietitian specializing in cardiometabolic disease and sustainable nutrition whose PhD research was on fermentable carbohydrates and liver fat) examine whether diet macronutrient composition independently affects liver fat reduction, insulin sensitivity, and cardiometabolic risk when adults with obesity, prediabetes, and excess liver fat lose weight.
• Roughly 1 in 3 adults today have excess fat in their liver—many without knowing it—which impairs liver function even at a very small percentage, can progress to fibrosis and increase the risk of certain liver cancers, and is causally linked via Mendelian randomization studies to developing insulin resistance (which in turn promotes liver fat deposition in a vicious cycle).
• The discussion centers on a rigorous 6-month randomized clinical trial by Sam Klein’s group at Washington University (PMID: 42660124) that provided all food to participants, controlled for a 10% weight loss across three distinct diets (very-low-carbohydrate ketogenic, Mediterranean, and very-low-fat plant-forward), and verified adherence with biomarkers.
• While headlines praised the ketogenic diet for cutting intrahepatic triglyceride (liver fat) content by 67%, Dr. Guess and Hill unpack the underlying mechanisms—revealing that high protein intake and amino acid signaling, rather than ketones or saturated fat, drove the superior liver fat reduction alongside carbohydrate restriction and weight loss.

Key Concepts

• Speaker Disclosures and Episode Sponsors: Simon Hill notes he eats a plant-based diet and discloses he serves on the Scientific Advisory Board for IM8. Episode sponsors read by Hill include Eight Sleep (Pod 6 smart mattress cover; promo code THEPROOF for up to $200 off), Function Health (lab testing membership offering 160+ tests for $1/day; promo code SIMON25 for a $25 credit), IM8 (Daily Ultimate Essentials multinutrient supplement; promo code SIMON for 10% off plus a welcome kit, though Hill hedges that claiming a multinutrient got him his health results would be a stretch and that cognitive benefits in trials are mostly in people older than him, so he holds that loosely), and Momentous (Signature-spec Creatine; promo code THEPROOF for up to 35% off the first order). Dr. Guess also mentions a former Massachusetts company called Axcella that tested proprietary amino acid mixes, with no personal financial tie stated.
• Why Controlling for Adherence and Weight Loss Matters: In a meta-analysis of weight-loss trials aiming to test low-carbohydrate diets of <50 grams of carbs/day, actual adherence reports showed some participants were eating 180 grams of carbs/day at the end of the trial. Sam Klein’s trial overcame this by providing all food, achieving an equal 10% weight loss over a 6-month period across all three groups, and measuring adherence via urinary nitrogen for protein intake and blood ketones for carbohydrate restriction.
• The Washington University Trial Design and Results (PMID: 42660124): Per 2,000 kcal, the very-low-carbohydrate ketogenic diet contained 20 g carbohydrate (4% energy), 112 g protein (23% energy), 164 g fat (73% energy), 52 g saturated fatty acids (23% energy), and 749 mg cholesterol; the Mediterranean diet had 254 g carbohydrate (50% energy), 83 g protein (15% energy), 80 g fat (35% energy), 19 g saturated fatty acids (8% energy), and 160 mg cholesterol; and the very-low-fat plant-forward diet had 352 g carbohydrate (70% energy), 90 g protein (15% energy), 35 g fat (15% energy), 7 g saturated fatty acids (3% energy), and 15 mg cholesterol. The ketogenic group cut liver fat by 67% (-67% vs. -42% in the Mediterranean group and -45% in the very-low-fat plant-forward group, $p < 0.05$, a roughly threefold difference in liver fat reduction / de novo lipogenesis impact). The proportion of weight loss from fat did not significantly differ across groups (71% ketogenic, 74% Mediterranean, and 77% plant-forward). Dr. Guess also notes that the study authors stated they would look at vascular function (flow-mediated dilation / FMD), which was missing from the published paper.
• Weight Loss Drives Glycemic Recovery: People with prediabetes already have beta-cell dysfunction and, depending on the group studied, have lost 30% to 60% of their first-phase insulin secretory response. In the Klein trial, oral glucose tolerance test (OGTT) results showed no significant difference between the three diets after the 10% weight loss. This aligns with the DiRECT studies, where people with type 2 diabetes who lost about 10 kg normalized their glucose handling (including first-phase insulin secretion) from weight loss alone.
• Protein and Amino Acids—Not Ketones—Explain the Liver Fat Drop: Blood ketones in the trial's ketogenic group were only about 0.5 mmol/L, at the absolute bottom end compared to a proper ketogenic diet of 1.5 mmol/L or above, making ketones irrelevant to the outcome. Instead, the ketogenic diet had higher protein (23% vs. 15% of energy). Sam Klein’s paper notes liver fat reduction is driven by decreased glucose and insulin (from carb restriction) combined with an increase in glucagon (stimulated by protein/amino acids, not fat). Supporting this, the Norwegian CARBFUNC trial (PMID: 36116147; an RCT of ~200 adults over 1–2 years, with 78 and 57 participants completing 6 and 12 months) matched protein across a low-carb high-fat diet and two higher-carb cellular/acellular diets at 15–17% of calories; with equal weight loss, liver fat (measured as a secondary outcome via CT scan, not MRI) did not differ between groups. Furthermore, overfeeding studies show adding branched-chain amino acids (BCAAs) or protein protects against liver fat deposition from excess fat or fructose calories, and studies using Axcella’s proprietary amino acid mix lowered liver fat compared to an equal amount of general protein. A low-carb diet with increased protein also blunts the rise in postprandial and day-long triglycerides seen on low-carb, high-fat diets without added protein. Clinically, Dr. Guess has even observed lean vegan patients on high-processed-carb, very-low-protein diets present with severe fatty liver on MRI that resolved after increasing protein via tofu and legumes.
• Physiological Uncertainties and Hedges on Protein and Liver Fat Sources: Dr. Guess hedges that whole-body physiology is crazy complicated, notes she is not a physiologist, and emphasizes that exact molecular pathways for how protein reduces liver fat or how saturated versus unsaturated fats (and ceramides) act are not well characterized. She also states that we do not know the exact dose-response for protein on liver fat (whether it maxes out at 20% of calories or depends on muscle mass). Regarding whether liver fat comes mostly from dietary fat, adipose tissue, or de novo lipogenesis from carbs, she says it depends on the individual and degree of hepatic insulin resistance, deferring to isotope researcher Leanne Hodson because isotope studies are rare and expensive. On longevity and protein restriction, Dr. Guess admits she knows nothing about longevity research and thinks the idea of high protein affecting longevity is kind of rubbish, arguing that feeding muscles and lowering liver fat is preferable.
• Saturated Fat vs. Unsaturated Fat (PUFAs/MUFAs) and ApoB Dynamics: All three groups in the Klein trial saw reductions in ApoB and LDL cholesterol over the 6-month weight loss period with no statistically significant differences between groups (though the Mediterranean group had the highest absolute reduction in ApoB). Hill hypothesizes that ApoB did not rise on the ketogenic diet (despite 23% of calories coming from saturated fat) because massive 10% weight loss, carb restriction, and high protein reduced hepatic VLDL/LDL export enough to outweigh downregulated LDL receptor clearance. Outside of active weight loss, overfeeding trials (such as PMID: 31369090, where overeating saturated fat caused a 50% relative increase in liver fat and increased serum ceramides while polyunsaturated fat did not, despite similar weight gain of ~2.31 kg vs. 2.01 kg, as well as acute and 2-week palm oil vs. PUFA trials) and isocaloric trials consistently show PUFAs protect against liver fat deposition compared to saturated fats. PUFAs even generate higher concentrations of ketones on a high-fat ketogenic diet than saturated fats. Additionally, the PREDIMED trial showed hard-outcome cardiovascular event reductions via improved blood pressure and blood vessel function in participants eating ~25 grams/day of walnuts, almonds, and hazelnuts compared to a control low-fat/Mediterranean-light diet. While animal data suggested seed oils (like sunflower or rapeseed/canola oil) were inflammatory, human data consistently shows they are not. Dr. Guess explicitly speculates and hypothesizes that a high-protein, low-carb diet rich in MUFAs and PUFAs would outperform the high-saturated-fat version used in the Klein trial, especially during weight maintenance.
• The Role of Fiber vs. Protein: Regarding the broad recommendation of 30 grams of daily fiber, Dr. Guess notes her PhD research showed viscous/fermentable carbohydrates have a small reducing effect on liver fat (requiring a large dose of fermentable carbs) and improve glucose tolerance directly (by slowing glucose absorption) and indirectly (via short-chain fatty acids from fermentation altering fat metabolism), though high protein has a greater effect on liver fat than fiber.

Quotes

• At 4:39 - "It for the first time I think, enabled us to say, okay, if people lose 10% weight loss, and they have markedly different diets, what is the effect of those diets on certain outcomes? It was a great, really well-done study." - Dr. Nicola Guess explaining the unique value of Sam Klein's controlled feeding trial.
• At 5:31 - "And in this meta-analysis, they report what people actually did. And at the end of it, some people were having 180 grams of carb a day. So you're not testing what happens when people have less than 50 grams of carb a day, you're testing what happens when you advise people to have less than 50 grams a day." - Dr. Nicola Guess illustrating why dietary adherence biomarkers and food provision are critical in nutrition trials.
• At 14:45 - "And it's very interesting to me, there have been a number of studies done which is just low-carb and they didn't increase protein, and without weight loss, you don't really see a significant reduction in liver fat. Not to me that's convincing over the long term. You do when you combine low-carb with high-protein." - Dr. Nicola Guess highlighting that carbohydrate restriction requires elevated protein to effectively reduce liver fat without weight loss.
• At 16:07 - "There are studies which were matched for total protein, where they gave a proprietary mix of specific amino acids that lowered liver fat compared to the equal amount of general protein. So basically what that signals to me is that amino acids, some of them may actually be signaling molecules, like telling the liver not to deposit fat." - Dr. Nicola Guess describing how specific amino acids act as hepatic signaling molecules.
• At 18:52 - "The headlines were a ketogenic diet's best for liver fat. I think a better characterization is a high-protein, low-carbohydrate diet, possibly in the context of weight loss, though not necessarily, helps lower liver fat. That's my take of this study." - Dr. Nicola Guess reframing the main takeaway from the Washington University trial.
• At 23:32 - "I don't think it's the ketones because the ketones weren't even very high in this study. They were about 0.5 millimole per liter, which is probably at the absolute bottom end of ketones that you get from a proper ketogenic diet, which might be 1.5 or above." - Dr. Nicola Guess pointing out the biomarker evidence showing ketosis was barely reached in the ketogenic diet arm.
• At 32:56 - "But apart from deliciousness, I can't think of any nutritional reason why I would ever recommend someone, 'Hey, have more butter, have more cream, or coconut oil.' There's no nutritional reason to do that." - Dr. Nicola Guess contrasting saturated fat sources like butter, cream, and coconut oil against nutrient-rich or protective fats.
• At 44:41 - "I think 35 grams of protein a meal is a good starting point for most people. So with the combination of, if you're a meat eater, have chicken, fish, tofu if you're not, plant-based meat alternatives if you're not, then have some legumes, that's going to get you the protein. Have a small amount of carbohydrate if you want, like rice, potatoes, but keep it small, have loads of veggies, whatever you happen to like, and get some healthy fat on there, whether it's sunflower, canola, olive oil, that would be a very healthy main meals." - Dr. Nicola Guess outlining a practical meal template for managing fatty liver and cardiometabolic health.

Takeaways

• Prioritize a 10% weight loss to reverse prediabetes and lower liver fat: If you have obesity, prediabetes, or excess liver fat, achieving roughly a 10% reduction in body weight (~10 kg in diabetes trials) is the primary driver for normalizing glucose handling, restoring first-phase insulin secretion, lowering ApoB/LDL cholesterol, and substantially reducing liver fat across diverse diet types.
• Increase protein intake to roughly 35 grams per meal: For managing obesity, prediabetes, and fatty liver, aim to increase dietary protein by at least 30%—targeting about 35 grams of protein per meal from sources like chicken, fish, tofu, plant-based meat alternatives, and legumes—while keeping refined or starchy carbohydrate portions (like rice or potatoes) small and eating plenty of vegetables.
• Swap butter, cream, and coconut oil for unsaturated fats: Avoid adding butter, cream, or coconut oil for nutritional purposes (noting that cheese and high-fat yogurt appear neutral for cardiometabolic risk and are nutrient-rich). Instead, use polyunsaturated and monounsaturated fat sources—such as olive oil, canola/rapeseed oil, sunflower oil, and ~25 grams per day of nuts like walnuts, almonds, and hazelnuts—to protect against liver fat accumulation and reduce cardiovascular events.
• Ensure adequate protein if eating a plant-based or vegan diet: Even lean individuals eating a plant-based diet can develop severe fatty liver if their diet consists of high processed carbohydrates and very low protein; deliberately incorporate concentrated plant proteins like tofu, legumes, and plant-based meat alternatives at meals.
• Skip the "faff" of cooking and cooling starches in favor of high-protein meal pairing: Rather than relying on cooking and cooling rice or potatoes to create resistant starch, simply pair a small portion of regular white rice or potatoes with a high-protein meal (such as fish, tofu, or legumes) to naturally slow gastric absorption and blunt post-meal glucose spikes.