We Know Quite a Bit About How a Ketogenic Diet Affects Actual Humans
I understand, obviously, that most people aren't being "algorithmed" with the same stories that I am, which means that most people haven't been hit by two waves (let alone one) of headlines on a quasi-recent study out of the University of Utah that purportedly raises concerns regarding the negative health outcomes associated with a ketogenic diet. But because I have now, for the second time, been hit with a number of articles on the topic, I'm going to use it as an opportunity to express the frustration that I have with mice. Or, more specifically, the problems I have with mouse studies and the ignorant conclusions these studies often generate.
Alarming Headlines
Mouse Study Flags a Serious Downside to Popular Weight-Loss Diet1
Ketogenic Diet's Hidden Risks Named in Mouse Study2The above are but two examples of the type of alarming headlines recently making the rounds. They are, to put it lightly, unambiguous in their proclamations that a high-fat, low-carbohydrate diet leads to health consequences that perhaps have never before been considered - fatty liver disease and elevated triglycerides highlight the results. Quotes from the researches involved serve only to emphasize the apparent dangers of the diet. Lead author Molly Gallop claims that human studies have only ever assessed weight loss, but never "other facets of metabolic health." Thus, the idea goes, these negative findings in mice may be highly suggestive of similar negative health outcomes in humans. Her conclusion, in fact, is that when a person eats a high-fat diet, the lipids consumed unavoidably "end up in the blood and liver."
The articles describing this paper and the opinions of the authors are quite direct in amplifying this position - bluntly highlighting the rodents' "impaired liver function" and "low levels of blood glucose and insulin" indicative of "metabolic disease."
The conclusions from both the paper's authors and the articles this study has spawned are clear - ketogenic diets come with dangerous side effects that we've only now, finally, been able to understand, and nobody should undertake such a diet without understanding the risks to their health. It seems so clear, but...what if its all bullshit? What if the claims about human health are completely unfounded? What if it doesn't matter what happens when you pump a rat full of soybean oil?
What This Study Found
So I want to be clear - I'm not contesting the results of this study, I don't think anyone fabricated data, anything like that. The case I'm ultimately making is that these findings don't actually matter to people. The important findings of concern were as follows:3- Increased levels of triglycerides (the common form of fat packaged and moved around your bloodstream)
- Increased levels of the liver enzyme Alanine Transaminase, or ALT, typically indicative of some degree of liver damage
- Greater degree of hepatic steatosis (ie. "fatty liver disease")
Again, to be clear - these are generally bad things. If these occur in you, an actual human being, it should be cause for concern.
Applying the Results
My issue then is not with the actual results, but instead in trying to apply these results to people. The reason for this is twofold. The first is conceptual - mice are not people. That is not to say that rodent studies have no place! I am not making that claim. Rodent studies can be a great starting place, or a potentially suitable stand-in for studies that can't be carried out in humans (gene manipulation, etc.). But! Such studies always need to be interpreted with caution because, again, mice are not actually humans.
I think that fact really matters in nutrition studies, because mice and humans have not evolved to eat the same diets. Despite being consumed heavily by many, grains have been absent from the human diet for well over 99% of human history. Not so for mice. Mice have historically eaten a more carbohydrate-heavy diet than humans, and that matters for today's version of each species.
More concretely, our physiology is not in all ways the same. Mice, for example, do not express CETP, a protein heavily involved in human lipid mechanics and the return of triglycerides to the liver. I'm not going to pretend I'm an expert in rodent lipid mechanics but I can say with certainty, based on that alone, that there does exist at least some difference between human and mouse triglyceride and fatty acid behaviors.
What Happens in Human
Now if that was the end of it, the most I'd really be able to say is that some degree of caution should probably be exercised in completely applying these results to people. These results would perhaps be concerning, but only really serve as a jumping off point for future investigation in human subjects. However, as you can probably guess (and contrary to the author's claims), similar studies in fact have been carried out many times in humans. The results are... not ambiguous.
1. Increased blood triglycerides:
Molly Gallop claims that eating fewer carbs and more fat means that more fat will inevitably end up in your blood. This may be true for rats! This study suggests that it is. But, again, rodents aren’t people. Not only is this not true, but the opposite - that carbohydrates rather than fat lead to an increase in triglyceride levels - is arguably the most thorough and consistent finding in all of nutrition science.4–40 I’ll attach a few dozen trials below, but the only reason I’m not including a couple dozen more is the tedium of it.
2. Increased liver enzymes:
I’m just going to quote from a meta-analysis (effectively a “study of studies”) that sought to assess the effects of a ketogenic diet on the levels of several liver enzymes. These studies - 21 of them - not only assessed the ALT that was elevated in these mice, but other liver enzymes that you’d also prefer to see at lower levels such as AST and GGT. What it says, in plain English, is that every liver enzyme moved in a favorable direction when eating a ketogenic diet.
“A quantitative meta-analysis showed that a KD has a significant lowering effect on levels of aspartate aminotransferase (AST) [weighted mean difference (WMD): −3.56 U/L; 95% CI: −6.61, −0.51], alanine aminotransferase (ALT) (WMD: −3.03 U/L; 95% CI: −5.26, −0.81),
gamma-glutamyl transferase (GGT) (WMD: −12.25 U/L; 95% CI: −22.08, −2.42),
and alkaline phosphatase (ALP) (WMD: −5.29 U/L; 95% CI: −9.85 to −0.74)”41
3. Increased Hepatic Steatosis/Fatty Liver Disease
I will again quote from another meta-analysis of 20 studies that assessed the effects of high-fat ketogenic diets (HFKD) and very-low carb diets (VLCKD/LCKD) on fatty liver disease, reporting consistent positive results.
“Virtually all reviewed studies assessing liver fat content report positive
results after VLCDs, VLCKDs, and HFKDs”42
And here’s another literature review on the same topic:
“Studies showed that short- to medium- term ketogenic diets, with or without
calorie restriction, are able to lower plasma triglycerides
and ameliorate hepatic steatosis”43
Implications and Conclusions
It would be a conservative estimate to say that over 100 studies have now been carried out on ketogenic/low-carb diets and their effects on triglycerides, liver enzymes, and fatty liver disease. With virtually universal consistency, these studies demonstrate improved chronic health markers in those undertaking the diets. These 100-odd studies are not, however, consistent with the one mouse study currently making headlines. I’d like to make several points on the topic:1. I want to emphasize the fact that you don’t really need any of these studies to "know" the findings and conclusions - an understanding of human physiology will work just fine. The basic reason that these results are inescapable is that triglyceride levels, the rate of triglyceride return to the liver, and the fat content of the liver itself are effectively all just a reflection of a person’s capacity to properly utilize and store fat. It is, broadly speaking, excess carbohydrates, not fat, that chronically raise insulin levels and impair this capacity. When triglycerides can’t be properly taken up by the body, levels in the blood and liver inevitably rise. My entire series on lipid mechanics explains this in much greater detail but part 6 and part 11 are particularly relevant to this discussion.
- The author(s) of this paper is/are jaw-droppingly ignorant on the very topic they are researching, and implying expertise
- The author(s) of this paper is/are blatantly lying to you
When a person is so overwhelmingly and unconditionally wrong, there really can be no other options. The question is why this is happening and allowed to happen, and what can be done about the extreme misinformation being broadcast to a public who has no reason to doubt the people spreading fabricated information. I don’t, unfortunately, have the answers to those questions.
3. The original point of this post - mice aren’t humans! Again, I am not trying to speak ill of rodent studies as a concept. They have value. But they need to be used and (more importantly) applied properly. This is an extremely clear failure of that necessity. There are so many controlled trials on human diet and health markers - there is no need to revert to a rodent study! Its really a twofold issue here - there should be questions about the funding of this study, about the researchers and the claims they make, and about the publications that promote the results. Who decides to carry out and promote the results of a rodent study when, for all intents and purposes, the same or similar study has already been carried out in humans dozens of times over?
But the second issue is that of the reader, for you the next time you see a mouse study that says X or Y about human health, especially if the study involves diet. I am not saying you can dismiss every rodent study outright. I am not saying that every rodent study can actually be interpreted to have the opposite implication in humans, even if thats how it worked out here. But what I am absolutely saying is that you shouldn’t put your full faith in such a result, or such a headline. It should raise your interest, but not necessarily more than that. It should be an opportunity for curiosity - Its an easy thing to head to Google Scholar and type “keto trial fatty liver” and see what comes up.
There you have it - rodent studies can have value, but they are also often redundant, meaningless, or a vehicle for misinformation. When a rodent study is at odds with human trials, feel free to dismiss the mice! We have different physiology, and we eat different diets. When dozens of trials make clear a consistent finding in humans, you need not care what happens in a mouse. Now if we could only get science to catch up to that same notion…
