This article is part of the IBD Index. Last updated on April 27, 2022.
As the name would suggest, a ketogenic diet is a diet that promotes the metabolic state of ketosis. This is generally accomplished by consuming very few carbohydrates, a moderate amount of protein, and getting most calories from fat.
Note that I discuss the carnivore diet separately, because while a carnivore diet is necessarily ketogenic, its defining feature is the elimination of all plant foods, which is not necessary on a non-carnivore ketogenic diet.
For information about exogenous ketones (including brands), MCT oil, and some comments about breath vs. urine vs. blood testing for ketosis, check out my article on ketone supplementation.
Table of Contents
How did the ketogenic diet originate?
What is ketosis?
What is a ketogenic diet?
Controversy in the literature: “the science says…”
How might a ketogenic diet be beneficial for IBD?
Improving colonic mitochondrial resilience against stress
Reduction of inflammatory Th17 cells
Overcoming intestinal energy starvation
Is there any clinical evidence to support a ketogenic diet for IBD?
Mouse studies: how does a keto diet both alleviate and aggravate colitis?
Differences between the study designs
Length of time on the diet
Type of fat
Are there any risks to a ketogenic diet?
Increased absorption of lipopolysaccharide (LPS, or endotoxin)
Effects on the gut microbiome
Hydrogen sulfide toxicity
Muscle growth, bone density, and cognitive performance
Thyroid function and metabolism
Is a long-term ketogenic diet suitable for humans?
The Inuit: a case study in long-term ketosis?
The bottom line
Review of the literature
How Did the Ketogenic Diet Originate?
This might come as a surprise given how the “keto” diet has taken off as a fad offshoot of the Paleo and low-carb movements, but the ketogenic diet actually traces back to 1921 when the term was coined by Russel Wilder at the Mayo Clinic, where he became the first to propose that such a diet might be a viable alternative to fasting for treating epilepsy. (Source) Fasting had been used as an epilepsy treatment since the time of Hippocrates, but obviously a therapy that allowed the patient to eat food would be easier to maintain long-term. And thus, the ketogenic diet was born.
At its roots, the goal of a ketogenic diet is to get as close to a metabolic state of fasting as possible while still eating food. The question wasn’t “what bad foods should be removed from the diet to make it better,” the question was “what foods can be added to a state of not eating that will maintain the benefits of not eating?”.
A true ketogenic diet is defined not by the foods you do or don’t eat, but by the metabolic state you achieve, making it somewhat unique amongst therapeutic diets.
What is Ketosis?
In the process of oxidizing fats for energy, fatty acids are broken down to form a molecule called acetyl-CoA. Acetyl-CoA combines with another molecule called oxaloacetate to enter the Krebs cycle, ultimately producing ATP. When there’s more acetyl-CoA than there is oxaloacetate, excess acetyl-CoA is converted into the ketone bodies beta-hydroxybutyrate and acetoacetate. This process is called ketogenesis.
(For a fun glimpse into our understanding of ketosis circa 1966, check out this paper by Hans Krebs: The regulation of the release of ketone bodies by the liver.)
Ketogenesis occurs mainly in the liver, but can also occur in the intestinal epithelium, astrocytes (a type of brain cell), and perhaps some other cell types too (we aren’t sure yet). In any case, the liver is by far the primary exporter of ketones, because it doesn’t have the enzymes necessary to use ketones for energy itself.
It’s likely that ketones produced by other tissues are simply burned for energy locally. Most tissues can and will oxidize ketones, but they’re especially important for neurons, which – unlike most other tissues – can’t oxidize fatty acids directly. (Source 1, 2, 3, 4)
Several factors play into how much ketogenesis is occurring at any given time, but generally, anything that a) increases the amount of acetyl-CoA being produced from fatty acid oxidation in the liver and b) limits the amount of oxaloacetate available to bring the acetyl-CoA into the Krebs cycle will increase ketogenesis, and vice versa.
The main thing that increases fatty acid oxidation in the liver is low blood glucose and decreased insulin signaling from not eating carbohydrate, because this prompts the body to release fatty acids from fat stores, raising their concentration in the plasma. And the liver appears to take up fatty acids in a relatively unregulated manner, directly in proportion to their concentration in the blood. (Source)
Of course, eating fat also raises plasma fatty acid concentrations, and one specific category of fat – medium-chain triglycerides – is uniquely pro-ketogenic. This is because MCTs travel directly to the liver without circulating around the body first, and they can enter the mitochondria to be oxidized without activation by carnitine. (Source)
As for the second half of the equation – oxaloacetate provision – this is where protein comes in. In addition to prompting some insulin release, protein can also supply oxaloacetate. This is why protein consumption is often kept in check on a ketogenic diet.
The regulatory mechanisms involved in ketogenesis are far more complicated than I’ve laid out above, but this at least gives you the gist. And it’s important to note that like most biological processes, ketosis isn’t an on/off switch – it’s a spectrum. Even if you eat plenty of carbohydrates, you’re always going to have some ketone bodies floating around in your circulation.
For that reason, a strict definition of “ketosis” can be tricky, but I’ve seen it functionally defined in a couple different sources as a beta-hydroxybutyrate concentration of at least 0.5 mM, which is indicative of a state where ketones serve as a “prominent metabolite.” (Source 1, 2)
The below table provides ranges of circulating beta-hydroxybutyrate concentrations under various conditions. As you can see, a ketogenic diet promotes levels of ketosis comparable to extended fasting, and significantly higher than the levels achieved during normal circadian variation and overnight fasting. However, it’s also worth noting that prolonged exercise (however they’ve defined “prolonged”) can also get you well into ketosis, albeit temporarily.
What is a Ketogenic Diet?
I’ll admit that I’m not particularly plugged in to the keto community, but even still, I’m aware that there’s controversy surrounding what constitutes a ketogenic diet.
How many carbs can you have without being “kicked out” of ketosis? Forget carbs, how much protein can you have without disrupting ketosis?
The original ketogenic diet, as devised by Wilder to treat epilepsy, comprised 1 g of protein per kilogram of body weight in children, 10–15 g of carbohydrates per day, and the remainder of the calories in fat. Another source describes the original ketogenic diet as about 90% fat, 6% protein, and 4% carbohydrate. (Source)
For reference, a slice of bread has about 15 g carbohydrate, and an average banana has around 27 g.
More relaxed iterations of the ketogenic diet have cropped up since those early days of epilepsy research, including the Modified Atkins Diet, the MCT (medium-chain triglyceride) ketogenic diet, and the low-glycemic index treatment. The formulations of these additional types are discussed here, and all still show efficacy for epilepsy treatment while avoiding some of the downsides of the classic ketogenic diet.
But what about the average person who isn’t being clinically treated for epilepsy? Most people who undertake a ketogenic diet outside of a clinical setting stick to macronutrient ratios along the lines of 65-75% calories from fat, 15‑25% from protein, and 5-10% from carbohydrates. Some people opt to track just their carbohydrate consumption; common targets are under 20 g or 50 g.
For a description of what this type of ketogenic diet looks like, I’ll refer you to Mark Sisson’s What to Eat guide. You can also check out his Definitive Guide to Keto for a pretty solid primer. (He of course has his biases – he wrote a keto diet book, after all – but I generally find his takes to be fairly balanced and well-referenced.)
This looks to be another fairly comprehensive source, but I am not familiar with the website or the author.
As I said earlier, ketogenesis isn’t an on/off switch, and you can see from the table in the previous section that several weeks on a ketogenic diet could result in beta-hydroxybutyrate concentrations anywhere from 0.5 mM to 5 mM. That is an enormous range, and the level of ketogenesis achieved on a given diet will depend on a huge variety of factors, from relative and absolute amounts of macronutrients to fatty acid composition to calorie content, as well as demographic factors and health status.
To try to quantify the macronutrient effects, some of the early ketogenic diet researchers devised “Withrow’s equation,” which expresses a “ketogenic ratio” as the sum of the relative ketogenic and antiketogenic effect of each macronutrient (Source):
Ketogenic Ratio = (0.9F + 0.46P) : (C + 0.58P + 0.1F)
where F = fat (grams), P = protein (grams), and C = carbohydrate (grams)
As you can tell from the equation, amino acids (protein) can have both ketogenic and antiketogenic effects depending on their ability to supply oxaloacetate.
A ratio above 2 will reliably put someone solidly in a state of ketosis, with a somewhat gray area between 1.5 and 2 that could be considered ketogenic depending on context.
Unfortunately, this ketogenic ratio is almost never used to describe experimental diets in metabolic research, as the authors of this short 2018 review lament. They found that the ketogenic ratio of “ketogenic” diets among a selection of studies ranged from 0.36 (ie, decidedly not ketogenic) to above 6. Notably, the range of ketogenic ratios for “ketogenic” diets overlapped with the range for “high-fat” diets (ie, the diet commonly used to promote obesity and metabolic derangement).
This is crucial to keep in mind when interpreting research using “ketogenic” and “high-fat” diets, because the difference in clinical outcomes is dramatic, with “high-fat” diets being notoriously detrimental while ketogenic diets appear to be generally therapeutic.
All that to say, nobody really agrees on where the line is between a “ketogenic” and a “not ketogenic” diet – probably because there is no line. Thus, implementation of a ketogenic diet should be highly dependent on a person’s existing state of health or disease, as well as their goals. (We’ll get to what this means for people with IBD at the end!)
Controversy in the Literature: “The Science Says…”
Because the ketogenic diet originated as a medical treatment and has been widely discussed concerning non-IBD disorders (mainly epilepsy, neurodegenerative diseases like Alzheimer’s disease, metabolic disorders such as obesity and diabetes, and cancer), I thought it would be good to provide a bit of context before diving into the IBD-specific information, which is largely mechanistic and speculative in the absence of any clinical research at this time.
When trying to get a sense of a field of scientific literature, particularly if I don’t plan to familiarize myself deeply with all the primary studies, my usual approach is to look for a few recently-published review papers. In this case, I found one published in January 2022 entitled Ketogenic diet for human diseases: the underlying mechanisms and potential for clinical implementations, and another published in July 2021 entitled Ketogenic Diets and Chronic Disease: Weighing the Benefits Against the Risks.
Despite being published only six months apart and covering largely the same topics (down to very similar subsection headings), they leave the reader with wildly different impressions of a ketogenic diet.
My best attempt to summarize the true state of the science is that aside from epilepsy, the ketogenic diet has not been adequately vetted as an intervention for disease in humans; in other words, not enough high-quality clinical trials have been conducted. At this point, it shows most promise for cancer and Alzheimer’s disease. Some positive results have also been reported in metabolic disorders, but it’s unclear whether these results a) could be maintained long-term without adverse effects, and b) are unique to a ketogenic diet (as opposed to being achievable through other strategies). Some negative results have also been observed, but as discussed above, this could be due to sloppy terminology.
So, we don’t have much solid clinical research in humans to work with. The first review I mentioned above fills in the gaps with animal and in vitro research and mechanistic hypotheses, making a ketogenic diet sound like practically a cure-all in certain sections.
On the other hand, the second review fills in the gaps with epidemiological research and tired, vague warnings about saturated fat and animal foods, making a ketogenic diet sound like the worst possible choice for health.
Unfortunately, both reviews succumb to the curse of the sloppy terminology, and neither does a good job specifying whether the study in question confirmed that the “ketogenic diet” being tested was actually ketogenic.
All that to say, this demonstrates why, in many cases, “evidence-based” health recommendations can vary so much, or simply don’t exist. Because “the science says” a lot of things – the voices of science rarely speak in unison.
How Might a Ketogenic Diet Be Beneficial for IBD?
With that stage-setting out of the way, there are several potential mechanisms by which a ketogenic diet could be beneficial for those with IBD (with the last one being most interesting by far, in my opinion).
This is really more a basket of factors rather than a specific mechanism, but there’s certainly evidence that carbohydrates, more than proteins or fats, can contribute to symptoms in those with gut disorders.
In fact, this line of thinking is what led to the development of the Specific Carbohydrate Diet, which was created in part on the basis of clinical observations in the early 1900s that children with gut disorders tended to improve when carbohydrates were removed from their diets.
It’s certainly possible for someone to have broad intolerance to carbohydrate, but it’s just as likely (if not more) that similar symptom reduction could be achieved through selective limitation of specific carbohydrates, such as in a low-FODMAP diet. The Autoimmune Protocol also includes careful exclusion of certain carbohydrates while not limiting overall quantity.
Improving colonic mitochondrial resilience against stress
In one rat study modeling stress-induced IBS, researchers found that a ketogenic diet reduced the harmful effects of stress on mitochondria in the gut. They observed reduced inflammation and oxidative stress, as well as increased mitochondrial biogenesis, in the gut epithelium of rats given a ketogenic diet compared to controls. These effects are in line with the hormetic view of ketosis, discussed briefly later on.
Reduction of inflammatory Th17 cells
In her excellent article Is a high-fat or ketogenic diet bad for your gut?, Dr. Lucy Mailing describes preliminary research suggesting that a ketogenic diet may reduce levels of pro-inflammatory Th17 cells in the small intestine, and also improve post-injury regeneration of the gut epithelium by supporting stem cell function. (Source) The reduction of Th17 cells in the small intestine was observed in the mouse portion of this particular study, but a reduction of circulating Th17 cells has been observed in humans in at least one study of a ketogenic diet in epileptic children. (Source)
A couple of interesting things to note about this paper. One is that the most notable effect of the ketogenic diet on the microbiome was a consistent decrease in Bifidobacterium. The other is that the reduction of Th17 cells observed in mice after being given the human ketogenic diet microbiome was reversed by the transfer of Bifidobacterium into those mice. (This was not surprising – it’s well established that Bifidobacterium strongly induces Th17 cells in the intestine.)
For a healthy person, these effects are most likely neither here nor there, but merely the body’s adaptive reactions to its changing environment. But in the context of IBD, where the immune system is responding inappropriately to commensal microbiota and levels of Th17 cells are predictably elevated compared to healthy controls (Source), it seems plausible that this mechanism could have therapeutic relevance.
Overcoming intestinal energy starvation
We know that butyrate is the “preferred” energy substrate of intestinal epithelial cells. We also know that one of the hallmarks of IBD, particularly UC, is energy deficiency and impaired butyrate metabolism in the enterocytes. What if a ketogenic diet could circumvent the dysfunctional pathways of butyrate metabolism and rescue starving enterocytes?
This hypothesis is broached in the same article from Dr. Mailing that I mentioned above, and expanded upon in this review article published in mBio. Evidence suggests that in the absence of butyrate, several other molecules can fill in as energy substrates and key signaling molecules in the gut. These include isobutyrate, which is produced by bacterial metabolism of proteins; acylcarnitines from bile; and ketones (namely acetoacetate and beta-hydroxybutyrate) from circulation, either produced endogenously by the liver or supplied exogenously through supplementation.
Dr. Mailing draws the following conclusions:
So what does this mean? If you have a healthy microbiome and gut mucosa, butyrate is probably well equipped to deal with all your gut’s needs, no ketones needed. However, if you:
–have ulcerative colitis or other mucosal damage, where butyrate uptake is impaired,
–have gut dysbiosis characterized by a lack of butyrate-producers, or
–are on restrictive diet such as low-FODMAP or SCD, resulting in reduced butyrate production,
it may be wise to try therapeutic nutritional ketosis to support gut epithelial cell metabolism, at least until treating the underlying gut pathologies and healing the gut mucosa.Lucy Mailing, “Is a high-fat or ketogenic diet bad for your gut?” [links above added by me]
Note that a ketogenic diet may not be necessary to achieve ketosis; check out my article on ketone supplementation for more on this. And while this is the most compelling proposed mechanism I’m aware of that would recommend a ketogenic diet for IBD, it has yet to be clinically vetted. Which brings us to…
Is There Any Clinical Evidence to Support a Ketogenic Diet for IBD?
Despite those promising mechanisms, we don’t yet have any clinical studies on a ketogenic diet for IBD. However, we do have one case report indicating that a ketone supplement in the context of a low-carb (but not ketogenic) diet can be therapeutic for Crohn’s disease: The effects of exogenous ketones on biomarkers of Crohn’s disease: A case report.
The authors report the case of a 51-year-old woman with fairly severe untreated Crohn’s disease (diagnosed for 25 years, unmedicated for 15) who felt “markedly better” after only one week of beta-hydroxybutyrate supplementation, and achieved full clinical remission after 3 months, along with significant improvements in body composition (fat loss with maintenance of lean mass).
She took 4 g sodium beta-hydroxybutyrate in water each morning for 2 weeks, then 4 g of a mixture of sodium, calcium, and magnesium beta-hydroxybutyrate twice per day thereafter. She experienced flatulence the first week, but reported no other side effects.
Her diet was described as a “flexible, non-ketogenic low-carb diet” with at least one higher-carb day per week. She couldn’t eat large amounts of oils, and ate salad frequently.
As with any case report, the results are compelling, and add support to the idea that the unique benefits of a ketogenic diet for IBD are mediated by ketone bodies themselves rather than the state of metabolic ketosis.
This provides some very preliminary evidence in support of ketone supplements as a treatment option for IBD, but doesn’t give us much insight into the potential benefits (or detriments) of a ketogenic diet itself for this patient population.
We do have one additional case report titled Crohn’s disease successfully treated with the paleolithic ketogenic diet, which I’ve included in the review of the literature section at the bottom of this article. This patient’s diet was definitely ketogenic: he maintained a 2:1 ratio of fat to protein, and urinary monitoring ensured that he remained in ketosis.
However, the patient reportedly did not do well with the introduction of small amounts of fruit or vegetables, and had a flare-up in response to a “paleo” cake (which could have been keto-friendly, but they didn’t specify). This indicates to me that while being in ketosis was very obviously not detrimental to his healing (and could’ve been a key factor in achieving remission), his body was reacting to something in plant foods. As such, this case report is really more supportive of a carnivore diet specifically.
Mouse Studies: How Does a Ketogenic Diet Both Alleviate and Aggravate Colitis?
Along with a dearth of human research on a ketogenic diet for IBD, we also have a dearth of animal research: I only found two animal studies (both published in 2021) testing a ketogenic diet in the treatment of colitis.
Much to my surprise, their results were polar opposite: one found a ketogenic diet to be protective against colitis (Kong et al), while the other found a ketogenic diet to markedly exacerbate colitis (Li et al). What could explain this discrepancy?
Differences between the study designs
Both groups used the same exact mice: SPF C57BL/6J mice purchased from Shanghai Laboratory Animal Co. The experimental diets had almost identical macronutrient composition: 10% vs 9% kcal from protein, 89% vs 91% kcal from fat, and ≤1% kcal from carbs (Kong and Li, respectively).
One difference was length of dietary intervention prior to DSS administration. Kong et al had their mice on the experimental diets for 16 weeks prior to inducing colitis, while Li et al only had their mice on the experimental diets for 3 weeks before inducing colitis.
A second difference was length and concentration of DSS administration. Kong et al induced colitis with 1.5 weeks of 3% DSS, while Li et al induced colitis with one week of 2% DSS. This seems unlikely to explain the difference in results, since the mice that got more DSS for longer were the ones whose colitis was ameliorated by the ketogenic diet.
A third difference was the fat source in the ketogenic diet. When comparing the composition of the two diets (found in the supplemental materials of both studies), the only notable difference was that while Kong et al used a diet with the bulk of fat coming from lard, and the rest from soybean oil, Li et al used a diet with the bulk of fat coming from hydrogenated vegetable shortening, and the rest from corn oil.
Length of time on the diet
It’s well established that changes in the microbiota occur rapidly in response to dietary shifts; the seminal study by David et al published in Nature in 2014 reported dramatic shifts within a matter of days for people on plant vs animal-based diets.
But even so, microbial adaptation doesn’t just stop after the initial rapid shift, and a five-fold increase in duration is significant, especially given the lifespan of a mouse.
One study on MS patients found a biphasic response in the microbiome, with an initial decline in bacterial concentrations and diversity followed by an improvement (even over baseline values) after several months on the diet. The authors even say that “the long-term effects of diet were opposite to the immediate response to the intervention.”
So although I sadly don’t have the expertise necessary to give a viable analysis, this data makes it seem plausible that a change in duration of the exact same intervention could make the difference between severely aggravating colitis and markedly ameliorating it.
Type of fat
The diet used in the Li et al study was from Envigo, and after some digging, I confirmed that the vegetable shortening used is indeed fully (rather than partially) hydrogenated, therefore containing minimal to no trans fats, making this fat source essentially 100% saturated.
For some reason I wasn’t able to find the manufacturer of the diet used in the Kong et al study (maybe it’s in Chinese), but lard is generally going to be mostly monounsaturated fat (40-50%), almost as much saturated fat (35-45%), and the remainder polyunsaturated fat (5-15%). This means that the mice in these two experiments were getting wildly different fatty acid profiles, even though their macronutrient ratios were almost identical.
I think inferring anything from these results would be premature (especially given the limitations of animal research in general), but I found the analysis interesting. In mice, at least, it appears that outcomes are better when they’ve had more time to adapt to their new diet prior to being poisoned, and also when the diet contains natural fats with a balance of fatty acids.
Are There Any Risks to a Ketogenic Diet?
Because the ketogenic diet is a medical diet that has been studied somewhat extensively for epilepsy, it has the unique advantage (or disadvantage, depending on how you look at it) of having its associated “adverse events” thoroughly documented. And there are quite a few. And perhaps a bit disconcertingly for those with IBD, many common side effects are gastrointestinal symptoms. (Source)
Aside from the technical and often unhelpful discussion of “adverse events” in the medical literature, there are a few overarching concerns regarding a ketogenic diet that are generally raised, including: increased absorption of lipopolysaccharides (also known as endotoxin) from the higher fat content in the diet; adverse effects on the gut microbiome from reduced fermentable carbohydrate; increased hydrogen sulfide production; and adverse effects on muscle growth, bone density, cognitive performance, thyroid health, and overall metabolism.
Increased absorption of lipopolysaccharide (LPS, or endotoxin)
Lucy tackles this claim in her article that I mentioned previously. Essentially, she makes the case that absorption of LPS via chylomicrons (the mechanism that would be increased by a high-fat diet) simply results in the relatively safe transport of LPS to the liver for detoxification.
She cites evidence indicating that LPS carried by chylomicrons is no longer toxic or immune-stimulating. This is opposed to LPS that gets into circulation as a result of increased intestinal permeability, which can wreak havoc through widespread immune activation. She also cites evidence that ketogenic diets typically reduce systemic inflammation – a rather unexpected finding, if high-fat diets did indeed promote endotoxemia.
To add to her argument, I’ll also offer up a 2020 study entitled Endotoxin May Not Be the Major Cause of Postprandial Inflammation in Adults Who Consume a Single High-Fat or Moderately High-Fat Meal, which concludes on the basis of their results that “the prevailing concept of HFD-induced metabolic endotoxemia requires careful re‑evaluation.”
My personal take is that there’s quite a lot more to explore here before categorizing this as a non-issue; for instance, what is involved in liver detoxification of LPS, and what are the consequences if that pathway isn’t running smoothly?
That same paper above admits that even still, “the fate of LPS absorbed from the gut is not fully understood. The profile of the LPS metabolites in the blood and their biological effects have not been determined in our study. Such a task would be required for comprehensive understanding of the role of the diet-induced absorption of LPS in enhancing inflammation” (Mo et al, 2020).
I think we can safely say that endotoxemia is not always a concern with a high-fat ketogenic diet; perhaps even that is usually isn’t. However, I don’t think we can yet conclude that it’s never a concern, particularly in the context of IBD where normal pathways might not be functioning correctly.
That said, I’d be very interested to see more research in this area, because it seems plausible that a ketogenic diet could potentially reduce LPS by reducing overall bacterial load.
Effects on the gut microbiome
One of the most commonly-cited risks of a ketogenic diet is potential adverse effects on the gut microbiome. A popular theme in animal research is that high-fat diets promote a disease-associated microbiota, although one report finally pointed out in 2020 that in most of these studies, the fiber content is probably the more important factor.
In any case, I talk about this issue a bit in my article about the AIP, but unlike the AIP where fruits and starchy carbs can be included, a ketogenic diet is very low in carbohydrates by definition, and is therefore mostly devoid of certain types of soluble prebiotic fibers.
Interestingly, although the mechanisms responsible for the therapeutic effects of the ketogenic diet in a number of disease states are still not fully understood, it is becoming increasingly apparent that the intestinal microbiota is a key mediator. So whatever the effects of a ketogenic diet on the gut microbiota are, they are often therapeutic.
Lucy’s article Is a high-fat or ketogenic diet bad for your gut? that I’ve referenced throughout makes a compelling case that a ketogenic diet is not detrimental to the health of the microbiome. One especially insightful point she makes is that we don’t even know what a “healthy” microbiome looks like. As she puts it, “a “healthy” microbiome is simply the microbiome that you have when you’re healthy.”
As far as specific changes in response to a ketogenic diet, I mentioned earlier that ketogenic diets appear to reduce Bifidobacterium levels. A number of studies have also observed increases in Akkermansia species in response to a ketogenic diet. (You can read my linked article Akkermansia Muciniphila: Bane or Boon for IBD? for more than you ever wanted to know on that particular bacterium.)
As mentioned earlier, a small study in MS patients found that a 6-month ketogenic diet notably improved the patients’ microbiome, despite the initial effects on the microbiota being interpreted as negative.
I expect that the plague of sloppy terminology discussed at the beginning of this article is at least partially responsible for the disparate conclusions regarding ketogenic diets and the microbiome. We also have the variable of duration, previously discussed.
However, individual factors including demographics, state of health or disease, and existing microbiome contribute additional variability, which leads me to…
Hydrogen sulfide toxicity
This is another topic Lucy mentions in her article, albeit briefly. Hydrogen sulfide is a gas that is produced endogenously by many tissues in the body, as well as by certain gut bacteria (known as “sulfate-reducing” bacteria). At normal levels, hydrogen sulfide has antioxidant effects and acts as an important signaling molecule, but at high levels, it becomes toxic.
Hydrogen sulfide is of particular interest to people with ulcerative colitis, because high levels impair oxidation of butyrate in colonocytes, among other cytotoxic effects. In fact, the “H2S toxin hypothesis” of UC has received a fair bit of attention in the literature, citing several lines of evidence in support: fecal samples from patients with UC often have higher levels of hydrogen sulfide than controls; the sulfate-reducing bacteria from such samples are more metabolically active; the distribution of gut inflammation seen in UC, typically increasing from proximal to distal, follows the typical concentration gradient of hydrogen sulfide; among others. (Source 1, 2)
Looking at UC through the hydrogen sulfide toxicity lens, a ketogenic diet is probably one of the worst dietary choices you could make, aside from a full carnivore diet.
One substrate for sulfate-reducing bacteria is taurine-conjugated bile acids, which will necessarily increase on a ketogenic diet to help digest the extra fat. Two other substrates are the sulfur-containing amino acids and intestinal mucin. And although a “true” ketogenic diet is low or moderate in protein, and also shouldn’t substantially change the amount of mucin available to bacteria, evidence suggests that bacterial metabolism of both of these substrates will increase in the absence of fermentable carbohydrates and fibers.
It appears that availability of fermentable carbohydrate may be a more important factor influencing hydrogen sulfide production than protein/sulfur availability or even levels of sulfate-reducing bacteria. This study found that antibiotics knocked back levels of sulfate-reducing bacteria but didn’t reduce hydrogen sulfide levels in the feces, while prebiotic supplementation had the opposite effect: levels of sulfate-reducing bacteria were unchanged, but hydrogen sulfide concentration decreased.
Preliminary evidence in mice supports the notion that a high-fat diet will lead to greater production of hydrogen sulfide, even if protein is kept low. (Source) And in the paper I mentioned earlier, David et al demonstrated that a ketogenic diet increased microbial sulfite reductase gene expression in healthy people in the short-term.
Unfortunately, testing for hydrogen sulfide production in the gut is tricky and inconsistent, and all of this evidence is quite preliminary. That said, the potential adverse effects of a ketogenic diet on hydrogen sulfide production should not be ignored, especially in the context of IBD.
Muscle growth, bone density, and cognitive performance
Mechanistically, it’s well accepted that ketosis reduces anabolic activity in the body, largely due to reduced insulin signaling. (Source) The limited clinical evidence we have appears to bear this out, with a 2021 review concluding that ketogenic diets may impair gains in muscle mass and performance from resistance training. (Source)
Sarah Ballentyne covers concerns over all three of these topics in her article The Case for More Carbs: Insulin’s Nonmetabolic Roles in the Human Body, so you can check that out if you’re interested.
Importantly, a ketogenic diet demonstrably improves cognitive performance in people with neurodegenerative diseases such as Alzheimer’s or Parkinson’s disease, but as Sarah discusses in her article, the opposite effect is observed in healthy people.
Thyroid function and metabolism
It’s no secret that the ketogenic diet mimics the metabolism seen in fasting or starvation. In fact, this is a feature, not a bug. But through an evolutionary lens, one would expect certain body adaptations to signals of starvation, prime among them being a lower metabolic rate.
Anecdotally, many who have followed a ketogenic diet long term report symptoms consistent with reduced thyroid function, such as hair loss, insomnia, impaired temperature regulation, and even amenorrhea.
For a generally pro-keto take on the topic of thyroid health, you can read Mark Sisson’s article Is Keto Bad For Your Thyroid?.
The high points: many animal studies on low-carb, high-fat diets use oils high in omega-6 fatty acids, which notoriously tank thyroid function. One 56-day study found that the body temperature regulation of keto dieters was as good or better than that of high-carb, low-fat dieters. Other factors could contribute to lower thyroid function on a ketogenic diet, including calorie restriction and weight loss.
And most importantly (from his perspective), having lower T3 levels on keto isn’t necessarily a problem, and could potentially even improve longevity.
For a generally anti-keto take, you can read the aforementioned Sarah Ballentyne article. Her main bit of evidence is one study of epileptic children, in which 16.7% developed hypothyroidism requiring medication within 6 months of beginning a ketogenic diet (with females at higher risk). Of note, the main source of fat in the diet was extra virgin olive oil (sorry Mark – no corn oil here!).
However, she fails to mention the subsequent studies in epileptic children that failed to replicate those results, finding instead the type of subclinical hypothyroidism that Mark argues isn’t a concern (these study authors agree). (Source 1, 2)
As far as amenorrhea and overall fertility, despite ample anecdotal reports to the contrary, the only clinical evidence I’ve found actually demonstrated the benefits of a ketogenic diet for fertility, albeit in a very specific population: women with PCOS.
One case series describes four women with PCOS who all regained their period after 6 months of a ketogenic diet (two also became pregnant!), and a recent clinical trial reported that 50 out of 56 (!!) women with PCOS regained their periods after 6 months of alternating ketogenic and very low-carb diets, both relatively high in protein and quite low in calories.
As far as hard clinical evidence goes, I think the jury is still out on ketogenic diets, thyroid health, and metabolism. But the thing is, basically all the clinical evidence we have is short term – on the order of months to a year. And from an evolutionary perspective, it would make sense that the real detrimental effects would come after longer periods of “starvation” metabolism. Which leads us to the question…
Is a Long-Term Ketogenic Diet Suitable for Humans?
To my knowledge, the longest study we have followed three patients with GLUT-1 transporter deficiency on a strict ketogenic diet for 5 years. (Source) No adverse effects on bone health, menstrual cycles, or otherwise, were documented.
Which is great and all, but not exactly generalizable.
This review provides an interesting discussion of ketones as messengers signaling the danger of starvation to the body, and the various adaptive mechanisms that engenders. Metabolically, the oxidation of ketones and fatty acids increases oxidative stress and pro-inflammatory mediators (this paper describes some of the mechanisms involved, such as an altered FADH2 to NADH ratio placing a greater demand on complex II of the electron transport chain). The body adapts by upregulating various anti-oxidant, anti-inflammatory, and neuroprotective mechanisms.
In this way, ketosis can be considered metabolic hormesis, of sorts – a stressor that induces positive adaptation. This helps explain some of the therapeutic benefits seen in various diseases.
But the thing about hormetic stress is that it can only lead to positive adaptation if it’s temporary. For other hormetic stressors like exercise, cold water immersion, and even phytochemicals in fruits and vegetables, the dose makes the medicine – or the poison. But does this apply to metabolic ketosis?
In the absence of long-term studies, and not wanting to rely too heavily on mechanistic theories (which don’t always bear out in practical reality) or anecdotal evidence (which is never conclusive), we can always turn to evolutionary theory and anthropological observations. Enter: the Inuit. (Yeah, you knew that was coming.)
The Inuit: A Case Study in Long-Term Ketosis?
Everyone knows that the only two populations to have consistently consumed a very low-carbohydrate diet are the Maasai and the Inuit (and related Arctic populations), with the Inuit being the more extensively studied of the two.
It’s also probably common knowledge by this point that the Inuit aren’t generally in ketosis, even when fasting. And that their diet is arguably too high in protein to even be “ketogenic.” Yada yada.
These things are true, and are notable to the extent that it deprives us of a historical model of a human population thriving long-term while maintaining a constant state of ketosis.
However, I’ve seen people use the Inuit and their low ketone levels as evidence that long-term ketosis is dangerous or detrimental, arguing that they’ve adapted as a population to avoid being in ketosis.
A 2020 review article titled Inuit metabolism revisited: what drove the selective sweep of CPT1a L479? argues against that idea. The whole thing is worth a read if you’re even slightly interested in the topic; I found it fascinating and well-written.
First of all, it’s noteworthy that the selective sweep analyzed in this paper is possibly the strongest known selective sweep in human history. The variant reaches 93% prevalence in the Canadian Inuit, with lower prevalence in other northern populations (down to a low of 68% in northeast Siberians), and is completely absent in all other populations globally.
In any case, the paper author builds a strong case that this selection was likely not against ketosis, but rather for glucose conservation:
Though future scholars may find other evidence to question the health consequences of a very low carbohydrate diet, the Inuit case study does not provide strong evidence of a selective sweep “against ketosis” in the context of such a diet. Considering all available data, rather, the most plausible current hypothesis suggests the necessity of glucose conservation, possibly in combination with the requirement to mitigate effects of cold exposure and/or very high protein intake.Nicola Hale, Inuit metabolism revisited: what drove the selective sweep of CPT1a L479?; 2020
The other fascinating tidbit I took away from this paper is that the extremely high omega-3 content of the traditional Inuit diet may have been instrumental in the selective sweep of the L479 variant.
In people with that variant of the CPT1a gene, the pancreas secretes less glucagon than normal in response to falling blood glucose levels, resulting in less hepatic homeostasis-maintaining behavior: glycogenolysis, gluconeogenesis, fatty acid oxidation, and ketone production. This results in a state known as “hypoketotic hypoglycemia,” which is exactly what it sounds like, and to this day is a leading cause of SIDS in babies in Inuit populations.
Here’s the nifty part: it appears that omega-3 fatty acids upregulate fatty acid oxidation in both the liver and the pancreas, improving the response of both of these organs to low-energy states. The paper suggests that the high rate of SIDS in the modern Inuit may derive from a shortage of omega-3 fatty acids in the modern diet – a classic example of evolutionary mismatch.
All told, the example of the Inuit is probably not evidence of the detrimental effects of long-term ketosis, as some have claimed. However, it’s also decidedly not evidence in support of a long-term very low-carbohydrate diet. It merely provides an example of how one specific population has adapted over generations to an extremely specific diet in a harsh climate.
The Bottom Line
Ultimately, in a healthy person, the most realistic way to view any effect of a ketogenic diet – whether metabolic, microbiome, or otherwise – is through the lens of adaptation. Does a ketogenic diet upregulate protective anti-oxidant and anti-inflammatory mechanisms? Absolutely – but that isn’t because ketosis is “better” – it’s because the metabolic state of ketosis results in more oxidative stress. The body is simply doing what is necessary to adapt.
Same with the gut microbiome changes. A ketogenic diet doesn’t down-regulate Th17 expression because it’s “healthier” – it’s an adaptive response to the lower levels of certain types of gut bacteria, which change in response to the food present in their environment.
From this zoomed-out perspective, the role of ketosis in a healthy human seems fairly obvious. Insofar as a ketogenic diet mimics the state of starvation and/or low carbohydrate availability that humans undoubtedly encountered periodically throughout our evolutionary history, we have well-developed adaptive mechanisms at every level of our physiology enabling us to survive those periods.
There’s also plenty of evidence to support the idea that not only can we survive such periods, but that they’re good for us, at least in certain contexts. The well-documented health benefits of fasting, caloric restriction, and – yes – ketogenic diets, attest to that.
Through this same lens, it seems equally obvious that maintaining such a state chronically is not ideal. And if you want to point to the Inuit or the Maasai, go ahead. But unless you are descended from those people and are eating all the seafood and cow blood that they ate traditionally, then their ability to maintain good health and reproductive fitness long term on their highly specific diet has no bearing on your ability to do so.
However, the equation changes in the context of a chronic illness like ulcerative colitis or Crohn’s disease, because the body is no longer responding in adaptive ways to its environment. So we must ask whether a ketogenic diet could have therapeutic potential in the specific context of IBD, and if so, what method of implementation will maximize benefits while minimizing risks.
Here’s a brief summary of what we know:
- Ketogenic diets promote increased circulating levels of ketones, with higher levels achievable by keeping carbohydrates and protein to a minimum and consuming sources of medium-chain triglycerides.
- The two potential therapeutic mechanisms that appear unique to a ketogenic diet – namely reduction in immune system activation in response to lower Bifidobacterium levels, and provision of an alternate fuel source to colonocytes – have not yet been explored by clinical studies.
- The only clinical data we have is one case report where exogenous ketone supplementation in the context of a low-carb but non-ketogenic diet induced remission in Crohn’s disease.
- There are several potential risks to a ketogenic diet, with increased hydrogen sulfide production being the most specifically relevant to those with IBD.
- Evidence (both anthropological and clinical) supporting the long-term suitability of a ketogenic diet is lacking, and anecdotally, many long-term adherents have reported undesirable effects, mainly on fertility and metabolic health.
On the basis of what I’ve covered here, these are what I’d consider the most reasonable practical takeaways for someone with IBD:
- Any benefits in symptom reduction from elimination of carbohydrates could likely be realized on less restrictive (and less risky) diets, such as low-FODMAP or the Autoimmune Protocol.
- For those who have tried other approaches without success, raising circulating levels of ketones could be worth a shot, particularly if there’s reason to believe that butyrate availability to colonocytes is significantly impaired.
- This could be achieved through exogenous ketones (check out my article on ketone supplementation for more on this approach)
- If a ketogenic diet is undertaken, keeping protein intake moderate could help both increase the generation of ketones and reduce the risk of hydrogen sulfide overproduction.
- Strategic inclusion of low-carb veggies with higher prebiotic content may help reduce the risk of hydrogen sulfide overproduction as well and maintain a microbiome with some level of butyrate production, but may also contribute to GI symptoms.
- On the other hand, allowing bacterial numbers to naturally drop in response to fewer fermentable carbohydrates could reduce immune activation and plausibly improve inflammation. This could be one mechanism at play for people who experience significant benefit from a carnivore diet.
To sum up, I think a short-term therapeutic ketogenic diet is absolutely a viable strategy to try for someone with ulcerative colitis or Crohn’s disease, but the specific context must be considered, and it won’t be a good idea for everyone. And as with any restrictive diet, even if benefit is achieved, the goal should be to resolve the body’s maladaptive responses to its environment, ultimately enabling a return to a more flexible and liberalized healthy diet.
Review of the Literature
I’ve included a couple animal studies here (discussed at length in the post above), since the literature on this topic is sparse! Remember that while animal studies can be extremely useful in IBD research, their translational relevance to human disease is still quite limited.
*asterisks denote papers that I thought were particularly interesting, relevant, well-written, or otherwise worth reading
*Lowery et al. The effects of exogenous ketones on biomarkers of Crohn’s disease: A case report. 2017. J Gast Dig Gis.
- “In this paper, we report on a case study in which exogenous ketone supplementation lowered markers of CRP significantly and improved quality of life in an individual suffering from Crohn’s disease.”
- 51-year-old woman with fairly severe untreated CD (for 25 years, unmedicated for 15), took 4g sodium BHB in water each morning for 2 weeks, then 4g of a mixture of sodium, calcium, and magnesium BHB twice per day; experienced flatulence the first week but not after that. Followed a flexible, non-ketogenic LCD with at least one higher-carb day per week; couldn’t eat large amounts of oils, at salad frequently
- Felt “markedly better” after only one week, and achieved full clinical remission after 3 months, along with significant improvements in body composition (fat loss with maintenance of lean mass)
Tóth et al. Crohn’s disease successfully treated with the paleolithic ketogenic diet. (Case report). 2016. Int J Case Rep.
- This case report describes a 14 year old boy with treatment-refractory Crohn’s disease who achieved complete remission after 10 months on a strict ketogenic diet
- Diet consisted of “animal fat, meat, offal and eggs with an approximate 2:1 fat:protein ratio,” with an emphasis on red and organ meats over poultry, and very small amounts of honey
- Urine was monitored to ensure ketosis was maintained
- The patient reportedly did not do well with the introduction of small amounts of fruit or vegetables, and had a flare-up in response to a “paleo” cake (which could have been keto-friendly, but they didn’t specify), so this case report is really more supportive of a carnivore diet than a ketogenic diet in general
- From what I can tell, this case report is not relevant to a true well-designed ketogenic diet; however, I’ve included it here because the title and the description of the diet in the text (“extreme low-carbohydrate and extreme high-fat diet”) could lead someone to conflate this diet with a keto diet, so I figured I’d get ahead of any confusion and dispel that notion!
- From Table 1 in the paper (which rated his “low-carb” diet based on the plant-based diet score), it looks like an overall unhealthy diet, with regular consumption of meat, rice, bread, alcohol, and even sweets and soft drinks/juice, with limited consumption of fruits and vegetables. Frankly, I question how his diet could look like that and still be considered low-carb at all, much less “extreme low-carbohydrate and extreme high-fat,” but unfortunately the authors do not elaborate
- I discuss the other aspects of this case study in my IBD Index article on the semivegetarian / plant-based diet, but in short, this man’s UC was quickly put into remission by the hospital-provided plant-based diet I describe in that article
- Study in mice using a DSS-induced model of colitis testing a keto diet (KD) against a low-carb (non-keto) diet (LCD) and a control diet
- After colitis induction, the KD significantly reduced inflammatory responses, protected intestinal barrier function, and reduced ILC3 production and the expression of related inflammatory cytokines, whereas the opposite effects were observed for the LCD
- KD mice showed increased abundance of Akkermansia species, which are generally recognized as health-promoting, while LCD mice showed increased abundance of Escherichia and Shigella species that have often been associated with IBD
- The researchers confirmed that these results were due to modulation of the gut microbiota by performing a fecal microbiota transplant from the KD mice into germ-free mice
- These researchers observed opposite results as the study above: the mice on a KD had significantly worse colitis than control mice, with an up-regulated inflammatory response, increased intestinal barrier permeability, and microbial and metabolite profiles associated with disease (including many of the changes observed in the LCD mice in the above study, such as increased prevalence of Escherichia, Shigella, and Proteobacteria
- See above for discussion of the discrepancy between these two studies
Any statements made in this article that don’t include a specific source were derived from the following papers:
Attaye et al. The Role of the Gut Microbiota on the Beneficial Effects of Ketogenic Diets. 2022. Nutrients.
Kolb et al. Ketone bodies: from enemy to friend and guardian angel. 2021. BMC Med.
Murakami, Mari and Tognini, Paola. Molecular Mechanisms Underlying the Bioactive Properties of a Ketogenic Diet. 2022. Nutrients.
Paoli et al. Ketogenic Diet and Microbiota: Friends or Enemies? 2019. Genes (Basel).
Puchalska, Patrycja and Crawford, Peter A. Metabolic and Signaling Roles of Ketone Bodies in Health and Disease. 2021. Annu Rev Nutr.
Puchalska, Patrycja and Crawford, Peter A. Multi-dimensional roles of ketone bodies in fuel metabolism, signaling, and therapeutics. 2017. Cell Metab.
Zhu et al. Ketogenic diet for human diseases: the underlying mechanisms and potential for clinical implementations. 2022. Signal Transduct Target Ther.