LONDON – A 3-month diet comprised of 70% fat improved cognition in Alzheimer’s disease patients better than any anti-amyloid drug that has ever been tested.
In a small pilot study, Alzheimer’s patients who followed the University of Kansas’s ketogenic diet program improved an average of 4 points on one of the most important cognitive assessments in dementia care, the Alzheimer’s Disease Assessment Scale–cognitive domain (ADAS-cog). Not only was this gain statistically significant, but it reached a level that clinical trialists believe to be clinically meaningful, and it was similar to the gains that won Food and Drug Administration approval for donepezil in 1996, according to, director of the University of Kansas Alzheimer’s Disease Center in Fairway.
To put the results in perspective, donepezil was approved on a 4-point spread between the active and placebo arm over 3 months, said Dr. Swerdlow, who is also the Gene and Marge Sweeney Professor of Neurology at the university. Part of this difference was driven by a 2-point decline in the placebo group. Relative to its baseline, the treatment group improved, on average, by about 2 points.
But in the Ketogenic Diet Retention and Feasibility Trail (KDRAFT), also 3 months long, patients’ ADAS-cog scores didn’t decline at all. Everyone who stayed with the diet and kept on their baseline medications improved, although to varying degrees.
KDRAFT was very small, with just 10 patients completing the intervention, and lacked a comparator group, so the results should be interpreted extremely cautiously, Dr. Swerdlow said in an interview. “We have to very careful about overinterpreting these findings. It’s a pilot study, and a small group, so we don’t know how genuine the finding is. But if it is true, it’s a big deal.”
Diet and dementia
Emerging evidence suggests that modifying diet can help prevent Alzheimer’s and may even help AD patients think and function better. But this research has largely focused on the heart-healthy diets already proven successful in preventing and treating hypertension, diabetes, and cardiovascular disease. Most notably, the Mediterranean-DASH Intervention for Neurodegenerative Delay (MIND) diet cut the risk of AD by up to 53% () and also slowed aging-related cognitive decline ( ).
MIND is a combination of the low-salt, plant-focused DASH diet, and the heart-healthy Mediterranean diet. It is a moderate-fat plan, with a ratio of 33% fat, 38% carbohydrates, and 26% protein. Ideally, only 3% of the fat should be saturated, so MIND draws on olive oil, nuts, and other foods with monounsaturated fats, largely eschewing animal fats. It’s generally considered fairly easy to follow, since it allows a wide variety of whole grains, beans, nuts, fruits, vegetables, salads, fish, and poultry. Butter, red meat, fried foods, full-fat dairy, and fast foods are strict no-nos.
A ketogenic diet, however, turns MIND on its head. With a 70% fat, 20% protein, 10% carbohydrate ratio, a typical ketogenic diet nearly eliminates most fruits, and virtually all starchy vegetables, beans, and grains. It does, however, incorporate a large amount of fat from many sources, including olive oil, butter, cream, eggs, nuts, all kinds of meat, and fish. For a ketogenic diet, Dr. Swerdlow said, the ratio of fat to protein and carbs is more critical than the source of the fat.
MIND was designed to prevent the cardiovascular and endocrine disorders than predispose to dementia over the long term. But a ketogenic diet for patients with Alzheimer’s acutely manipulates the brain’s energy metabolism system, forcing it to use ketone bodies instead of glucose for fuel.
In normal energy metabolism, carbohydrates provide a ready supply of glucose, the brain’s primary fuel. When carbs are limited or absent, serum insulin decreases and glucagon increases. This promotes lipolysis. Ketones (primarily beta-hydroxybutyrate and acetoacetate) are formed in the liver from the newly released fatty acids, and released into the circulation, including into the brain during times of decreased glucose availability – a state characteristic of Alzheimer’s disease.
Induced ketogenesis trial
Inducing ketosis through diet seems to help correct the normal, age-related decline in the brain’s ability to use glucose, said, who also presented ketogenic intervention results at AAIC. “Cognitively normal, healthy older adults experience a 10% reduction in the brain’s ability to metabolize glucose compared to healthy young people,” he said in an interview. But this decline accelerates as Alzheimer’s hits. Those with early AD have a 20% decrement in glucose utilization, compared with healthy elders.
Changes in brain glucose metabolism can develop years before any cognitive symptoms manifest and seem to increase the risk of Alzheimer’s, said Dr. Cunnane of Sherbrooke University, Que.
“We propose that this vicious cycle of presymptomatic glucose hypometabolism causes chronic brain energy deprivation, and might contribute to deteriorating neuronal function. That could cause a further decrease in the demand for glucose, leading to cognitive decline.”
“What doesn’t change, though, is the brain’s ability to take up ketone bodies,” he said. If anything, the brain appears to use ketones more efficiently as AD becomes established. “It’s almost like the brain is trying to rescue itself. If those cells were dead, they would not be able to take up ketones. Because they do, we think they are instead starving because of their inability to use glucose and that maybe we can rescue them with ketones before they die.”
At AAIC, Dr. Cunnane reported interim results of an investigation of induced ketogenesis in patients with mild cognitive impairment (MCI). The 6-month BENEFIC trial comprises 50 patients, randomized to either a daily nutritional supplement with 30 g medium chain triglycerides (MCT) in a unflavored, nondairy emulsion, or a fat-equivalent placebo drink. When consumed, the liver very quickly converts MCT fatty acids into ketone bodies, which then circulate throughout the body, including passing the blood-brain barrier.
All of the participants in the BENEFIC trial underwent brain PET scanning for both glucose and ketone uptake. Early results clearly showed that the MCI brains took up just as much acetoacetate as did the brains of cognitively normal young adults. And although the study wasn’t powered for a full cognitive assessment, Dr. Cunnane did present 6-month data on three measures in the MCI group: trail making time, verbal fluency, and the Boston Naming Test. In the active group on MCT, scores on all three measures improved “modestly” in direct correlation with brain ketone uptake. In the placebo group, scores remained unchanged.
“We don’t have enough people in the study to make any definitive statement about cognition, but it’s nice to see the trend going in the right direction, Dr. Cunnane said. “I really think of this as a dose-finding study and a chance to demonstrate the safety and tolerability of a liquid MCT supplement in people with MCI. Our next study will use a 45 g per day supplement of MCT.”
Details of the KDRAFT study
The BENEFIC study looked only at the effects of an MCT supplement, which may not deliver all the metabolic benefits of a ketogenic diet. KDRAFT, however, employed both, and assessed not only cognitive outcomes and adverse effects, but the practical matter of whether AD patients and their caregivers could implement the diet and stick to it.
Couples recruited into the trial met with a dietitian who explained the importance of sticking with the strict fat:carb:protein ratio. It’s not easy to stay in that zone, Dr. Swerdlow said, and the MCT supplement really helps there.
“Adding the MCT, which is typically done for the ketogenic diet in epilepsy, increases the fat intake so you can tolerate a bit more carbohydrate and still remain in ketosis. MCT therefore makes it easier to successfully do the diet, if we define success by time in ketosis. Ultimately, it is an iterative diet. You check your urine, and if you are in ketosis, you are doing well. If you are not in ketosis, you have to increase your fat intake, decrease your carb intake, or both.”
The study comprised 15 patients (7 with very mild AD, 4 with mild, and 4 with moderate disease). All patients were instructed to remain on their current medications for Alzheimer’s disease for the duration of the study if they were taking any. All of the patients with moderate AD and one with very mild AD dropped out of the study within the first month, citing caregiver burden. The supplement was in the form of an oil, not an emulsion like the BENEFIC supplement, and it caused diarrhea and nausea in five subjects, although none discontinued because of that.
“We found that a slow titration of the oil could deal with the GI issues. Rather, the primary deal-breaker seemed to be the stress of planning the menus and preparing the meals.”
One patient discontinued his cholinesterase inhibitor during the study, for unknown reasons. His cognitive scores declined, but was still included in the diet-compliant analysis.
The diet didn’t affect weight, blood pressure, insulin sensitivity or resistance, or glucose level, but the intervention was short-lived. Nor were there any significant changes in high-density, low-density, or total cholesterol. Liver enzymes were stable, too.
“The only thing that changed was that they really did increase their fat and decrease their carb intake,” Dr. Swerdlow said. Daily fat jumped from 91 g to 167 g, and carbs dropped from 201 g to 46 g.
Almost everyone who stuck with the diet achieved and maintained ketosis during the study, although with varying degrees of success. “Many only had a trace amount of urinary ketones,” Dr. Swerdlow said. The investigators tracked serum beta hydroxybutyrate levels every month as well, and those measures also confirmed ketosis in the group as a whole, although some patients fluctuated in and out of the state.
The cognitive changes were striking, he said. In the 10-patient analysis, ADAS-cog scores improved by an average of 4.1 points. The results were better when Dr. Swerdlow excluded the patient who stopped his cholinesterase inhibitor medication. In that nine-patient group, the ADAS-cog improved an average of 5.3 points.
While urging caution over the small sample size and lack of a control comparator, Dr. Swerdlow expressed deep satisfaction over the outcomes. A clinician as well as a researcher, he is accustomed to the slow but inexorable decline of AD patients.
“I’m going to try to relate the impression you get in the clinic with these scores,” he said. “Very rarely, but sometimes, with a cholinesterase inhibitor in patients, we’ll see something like a 7-point change. That’s a fantastic response, an improvement you can see across the room. A change of 2 points really doesn’t look that much different, although caregivers will tell you there is a subtle change, maybe a little more focus. The average we got in our 10 subjects was a 4-point improvement. That’s impressive. And a 5-point change is like rolling the clock back by a year.”
The improvements didn’t last, though. A 1-month washout period followed the intervention. By the end, both ADAS-cog and Mini-Mental State Examination scores had returned to their baseline levels. At the end of the study, a few of the patients and their partners expressed their intent to resume the diet, but the investigators do not know whether this indeed happened. Still, the results are encouraging enough that, like Dr. Cunnane, Dr. Swerdlow hopes to conduct a larger, longer study – one that would include a control group.
Future investigations of the ketogenic diet in AD might do well to also include an exercise component, both researchers mentioned. In addition to starvation, ketogenic dieting, and MCT supplementation, exercise is an effective way to induce ketogenesis.
“Exercise produces ketones, but most importantly, it increases the capacity of the brain to use ketones,” Dr. Cunnane said. The connection may help explain some of the cognitive benefits seen in exercise trials in patients with MCI and AD.
“This raises the possibility that if in fact exercise benefits the brain, ketone bodies may mediate some of that effect,” Dr. Swerdlow said. “Could exercise potentiate the ketosis from the diet? That is possible, and maybe using these interventions in conjunction would be synergistic. At this point, we are just happy to show the diet is feasible, if even for a limited period.”
Implementing KDRAFT: Research team dishes the skinny on fats
The KDRAFT study diet is surprisingly flexible despite its strict ratio of fat to protein and carbohydrate, according to the University of Kansas research team that implemented it. It only took a few counseling sessions to get most study participants enthusiastically embracing the new eating plan, even one so radically different from the way they were accustomed to eating.
“We focused mainly on the macronutrient makeup,” said Matthew Taylor, PhD, who supervised the diet study on a day-to-day basis. Instead of distributing a rigid diet plan, with prespecified meals and snacks, “We talked more in general about foods they could have and foods they couldn’t have.”
“When people think ‘ketogenic,’ they think bacon, eggs, oil, butter and cream, and may have an automatic negative connotation that this is unhealthy eating,” Dr. Taylor said in an interview. “But yes, eggs were in there and, because a lot of people really like bacon, there was bacon, too!”
The educational sessions did include teaching about healthy and unhealthy fats, and Dr. Taylor “tried to steer people toward the healthier ones, like olive oil, avocados, and nuts. But I didn’t say, ‘Eat this one and not that one.’ If it took melting butter on vegetables to get to that fat ratio, I was not as concerned about where the fat came from as about getting there and maintaining ketosis.”
KDRAFT also had a twist that’s becoming more common among ketogenic eating plans: lots of vegetables. Dr. Taylor asked participants to concentrate on nonstarchy vegetables and forgo potatoes, corn, beans, and lima beans, although some people did enjoy peas occasionally.
“We used to be think we had to restrict vegetables or people would go out of ketosis more easily. But that doesn’t seem to be true. We focused a lot on eating vegetables, and everyone increased their vegetable intake dramatically. We actually tried to use vegetables as a vehicle for fat. For example, people would roast Brussels sprouts or broccoli in olive oil and then put melted butter on it. It was pretty much, ‘Eat all the vegetables you can and put fat on them.’”
Fruits are full of sugar, so they are not liberally used in most ketogenic diets, but KDRAFT did allow one type: berries, and blueberries in particular. “We had people eating a couple of small handfuls of berries throughout the day and still being able to maintain ketosis. We did severely cut back on the amount and type of fruit people could have, but berries seemed to work well.”
Whipping cream had a place, too. “It fit really well in the diet, because it’s basically all fat,” Dr. Taylor said. “It’s used more often in pediatric ketogenic diets as a milk substitute. One thing our subjects liked to do was use it to make a sweet snack. All it takes is a packet of [stevia] sweetener and some vanilla. Then you whip and freeze it and it’s like an ice cream dessert.”
After the initial drop-outs, the remaining study pairs embraced the intervention enthusiastically.
“When the study partner took the diet on too, we had our best success. One of our last pairs had an entire family join in – children, grandchildren, everyone decided to follow the diet. That is a very helpful piece to this. It’s difficult to always say, ‘Here’s our normal food and here’s the keto food over here.’”
The dropouts occurred very early. These study pairs, all of whom included patients with moderate Alzheimer’s, never embraced the plan at all, and this is a telling point, Dr. Taylor noted.
“When you get to a level of dementia there are so many other things in the caregiving process that taking on big behavioral changes is very difficult.”
Although the study showed that the diet wasn’t practical for sicker patients at home, it still might be beneficial in other settings, said Debra Sullivan, PhD, RD. Dr. Sullivan chairs the department of dietetics and nutrition at the University of Kansas Medical Center and holds the Midwest Dairy Council Endowed Professorship in Clinical Nutrition.
“I think that we might be able to create a version of the diet that could be used in an institutional setting for our more advanced patients,” she said. “But there’s no denying that this can be challenging. It’s a big change from the way the typical American eats.”
None of the KDRAFT participants experienced any lipid changes, for either better or worse. The 3-month intervention was long enough to have picked up such changes if they were in the offing, said principal investigator Russell Swerdlow, MD. While there are mixed data on ketogenic diets’ atherogenic effects, many people respond positively, with improved cholesterol.
“Much of what it comes down to is, are you in a catabolic or anabolic states? Are you building up or tearing down? Excessive cholesterol is a sign of being overfed and laying down energy supplies. You take in carbon and turn it into cholesterol. But if you can trick your body into a catabolic state – essentially make it think it’s starving, which a ketogenic diet does – then you have consistently low insulin levels, and you don’t turn on the cholesterol synthesis pathway. You may increase your cholesterol intake through diet, but you’re not synthesizing it in your body, and that synthesis is what really drives your cholesterol level. If you’re not overeating, your body’s production goes down.”
Brain Energy and Memory (BEAM) study
Dr. Swerdlow isn’t the only clinician researcher looking at how a ketogenic diet might influence cognition., well known for her investigations of the role of insulin signaling and therapy in AD, is running a ketogenic diet trial as well.
As noted on, the 24-week Brain Energy and Memory (BEAM) study aimed to recruit 25 subjects in two cohorts: adults with mild memory complaints, and cognitively normal adults with prediabetes. A comparator group of healthy controls will contribute cognitive assessments, blood and stool sample collection, neuroimaging, and lumbar puncture at baseline.
Both active groups will be randomized to 6 weeks of either a low-fat, high-carbohydrate diet, with carbs making up 50%-60% of daily caloric intake, or a modified ketogenic-Mediterranean Diet with carbs comprising less than 10% of daily caloric intake.
BEAM’s primary outcome will be changes in the AD cerebrospinal fluid biomarkers beta-amyloid and tau. Secondary endpoints include cognitive assessments, brain ketone uptake on PET scanning, and insulin sensitivity.
Dr. Cunnane has no financial interest in the MCT emulsion, which was supplied by Abitec. He reported conference travel support from Abitec, Nisshin OilliO, and Pruvit. He also reported receiving research project funding from Nestlé and Bulletproof.
Dr. Swerdlow had no financial disclosures.