MODIFIED KETO DIET
Brain energy metabolism is impaired in Alzheimer’s disease (AD), which may be mitigated by a ketogenic diet. We conducted a randomized crossover trial to determine whether a 12-week modified ketogenic diet improved cognition, daily function, or quality of life in a hospital clinic of AD patients.
We randomly assigned patients with clinically confirmed diagnoses of AD to a modified ketogenic diet or usual diet supplemented with low-fat healthy-eating guidelines and enrolled them in a single-phase, assessor-blinded, two-period crossover trial (two 12-week treatment periods, separated by a 10-week washout period). Primary outcomes were mean within-individual changes in the Addenbrookes Cognitive Examination – III (ACE-III) scale, AD Cooperative Study – Activities of Daily Living (ADCS-ADL) inventory, and Quality of Life in AD (QOL-AD) questionnaire over 12 weeks. Secondary outcomes considered changes in cardiovascular risk factors and adverse effects.
We randomized 26 patients, of whom 21 (81%) completed the ketogenic diet; only one withdrawal was attributed to the ketogenic diet. While on the ketogenic diet, patients achieved sustained physiological ketosis (12-week mean beta-hydroxybutyrate level: 0.95 ± 0.34 mmol/L). Compared with usual diet, patients on the ketogenic diet increased their mean within-individual ADCS-ADL (+ 3.13 ± 5.01 points, P = 0.0067) and QOL-AD (+ 3.37 ± 6.86 points, P = 0.023) scores; the ACE-III also increased, but not significantly (+ 2.12 ± 8.70 points, P = 0.24). Changes in cardiovascular risk factors were mostly favourable, and adverse effects were mild.
This is the first randomized trial to investigate the impact of a ketogenic diet in patients with uniform diagnoses of AD. High rates of retention, adherence, and safety appear to be achievable in applying a 12-week modified ketogenic diet to AD patients. Compared with a usual diet supplemented with low-fat healthy-eating guidelines, patients on the ketogenic diet improved in daily function and quality of life, two factors of great importance to people living with dementia.
This trial is registered on the Australia New Zealand Clinical Trials Registry, number ACTRN12618001450202. The trial was registered on August 28, 2018.
Despite extensive efforts towards prevention and remediation, dementia remains an urgent public health priority, affecting over 50 million people worldwide . The majority of people with dementia have Alzheimer’s disease (AD), a disorder that characteristically results in progressive cognitive and functional decline . Supportive care remains the mainstay of treatment, and new strategies are needed.
Brain energy metabolism is impaired in AD. Compared with healthy controls, people with AD display lower levels of brain insulin signaling and fewer brain insulin receptors, culminating in brain insulin resistance [3, 4]. PET studies demonstrate a 20–25% deficiency in cerebral glucose metabolism . AD neurons also exhibit diminished numbers of mitochondria, many of which show reduced citric acid cycle and respiratory chain activity, culminating in decreased energy production .
Ketogenic diets can theoretically mitigate impaired brain energy metabolism in AD, leading to improved cognition, daily function, or quality of life. Ketogenic diets are high-fat, low-carbohydrate diets that shift the body towards fat metabolism. Neurons cannot metabolize fats directly, but the liver converts fats into ketones, which can serve as a major neuron energy source . During a typical western diet, the concentration of the primary blood ketone, beta-hydroxybutyrate, supplies less than 5% of brain energy requirements and its blood concentration rarely exceeds 0.5 mmol/L. By contrast, a ketogenic diet induces a state of “physiological ketosis” in which beta-hydroxybutyrate provides a greater contribution to brain energy metabolism and its blood concentration exceeds 0.5–0.6 mmol/L. [8, 9] Compared with glucose, ketones produce more energy per unit oxygen . Cerebral ketone metabolism remains normal in AD and can potentially compensate for brain insulin resistance and deficient glucose metabolism . Ketogenic diets also typically upregulate mitochondria biogenesis and induce expression of genes associated with the citric acid cycle and respiratory chain, thus increasing neuron energy production .
To date, two clinical trials have examined the symptomatic effects of a ketogenic diet in AD. A single-arm study examined the impact of a 12-week ketogenic diet in 15 AD patients . The 11 completers improved their cognitive scores, but the lack of a control group meant that additional contributing factors could not be ruled out. A second ongoing randomized controlled trial has provided preliminary data on the impact of a 12-week Modified Atkins diet versus a recommended diet in 14 patients with mild cognitive impairment or AD . The most adherent patients improved their memory scores, but overall adherence was only fair, and function did not improve.
On this background, we conducted a randomized crossover trial to determine whether a 12-week modified ketogenic diet was well-tolerated and improved cognition, daily function, or quality of life in a hospital clinic of AD patients.
Materials and methods
This was a single-phase, assessor-blinded, two-period randomized crossover trial conducted at Waikato Hospital, a tertiary hospital in Hamilton, New Zealand. The trial was approved by the Waikato Maori Consultation Research Review Committee and Health and Disability Ethics Committee of New Zealand.
Patients and trial partners attended a screening visit in July 2019 and a diet instruction visit in August 2019. In September 2019, patients were randomized (1:1 allocation) to a modified ketogenic diet (intervention diet) or their usual diet supplemented with low-fat healthy-eating guidelines and optional recipes (control diet). The crossover design specified two 12-week treatment periods separated by a 10-week washout period during which patients resumed their usual diet (cognitive changes induced by a ketogenic diet in AD return to baseline after 1 month) . For each treatment period, assessments were made at baseline, week 6, and week 12.
The trial was advertised in newspapers and regional dementia organizations. Volunteers attended a 2-h screening visit that included a description of the trial, medical history, evaluation of current diagnostic criteria for probable AD , dementia severity rating scale , AD informed consent questionnaire (with the word “medication” replaced with “diet”) , geriatric depression scale (short form) , Hachinski ischemia scale , body mass index calculations, and (assuming consent capacity) written informed consent from both patient and trial partner.
Eligible patients were 50 to 90 years of age, satisfied the revised NINCDS-ADRDA criteria for probable AD (confirmed by a neurologist or geriatrician), had a dementia severity rating scale score < 19, a body mass index > 18.5, and a cohabiting trial partner willing to (at least partly) partake in a ketogenic diet. Exclusion criteria included moderate or severe depression (geriatric depression scale score > 8), substantial cerebrovascular disease (Hachinski ischemia scale score > 4), a change in acetylcholinesterase inhibitor dose within the past 6 weeks, and a concurrent medical or psychiatric disorder judged likely to create difficulty in completing the trial. Patients also had recent (within 1 year) blood investigations (cell count, electrolytes, creatinine, liver function, thyroid-stimulating hormone, B12, and folate) in the normal range, as well as recent (within 2 years) CT or MRI brain imaging showing no ischemic changes beyond age-appropriate leukoariosis (missing investigations for otherwise eligible patients were performed after screening).
Following screening, eligible patients and trial partners attended a 1-h diet instruction visit and were shown how to use a complimentary blood glucose and ketone (beta-hydroxybutyrate) monitor (FreeStyle Neo, Abbott Diabetes Care, Whitney, UK), complete a 3-day (two weekdays, one weekend day) food record, and follow the diet plans. Blood was taken for apolipoprotein E genotyping.
Both diet plans contained guidelines, space to record daily (bedtime) blood glucose and ketone levels, and recipes (for full plans, see Supplementary Material). Patients on the ketogenic diet were instructed to eat all meals from the plan (unless they attended a social event, in which case meal advice was given), with numerous recipe options providing an average macronutrient ratio of 58% fat (26% saturated, 32% non-saturated), 29% protein, 7% fibre, and 6% net carbohydrate by weight. Main dietary constituents were green vegetables, meats, eggs, nuts, seeds, creams, and natural oils. The “usual” diet plan contained optional low-fat recipes in accordance with New Zealand healthy-eating guidelines, providing an average ratio of 11% fat (3% saturated, 8% non-saturated), 19% protein, 8% fibre, and 62% net carbohydrate by weight. Constituents were mainly green and root vegetables, meats, legumes, whole grains, and fruits. Both diets were supplemented by a daily multivitamin (Multivitamin and Mineral Boost, Clinicians Ltd., Auckland, New Zealand).
Randomization and blinding
Following stratification by dementia severity rating scale score and body mass index (below baseline mean, above baseline mean), the trial statistician randomized patients (1:1 allocation, block size of four) to the intervention or control diet using SAS statistical software (SAS Institute, Cary, USA).
Diet-related discussion between assessors and patients (or trial partners) was prohibited throughout the trial. To prevent detection of acetone breath, a fragrance-diffusing scent (Naturals Diffuser, The Aromatherapy Co., Auckland, New Zealand) was placed between assessors and patients (or trial partners) at every assessment.
Patients and trial partners attended three 1-h assessments over each 12-week treatment period. A baseline assessment was made during the week prior to commencing the treatment period, followed by assessments during weeks 6 and 12. Assessments evaluated patient cognition, daily function, and quality of life; patients and trial partners were evaluated by the same assessor at baseline and week 12, on the same weekday and hour of the day. Cognition was assessed using the Addenbrookes Cognitive Examination – III (ACE-III) scale, administered by an ACE-III-trained neurologist, neuropsychologist, psychiatrist, or geriatrician (New Zealand version A at baseline, version B in week 6, and version C in week 12). The ACE-III assesses 19 activities pertaining to five cognitive domains: attention, memory, fluency, language, and visuospatial ability (scores range from 0 to 100, with higher numbers indicating better cognition) . It has been objectively validated, with high levels of correlation shown between domain scores and performance on standard neuropsychological measures , and is the primary cognitive assessment battery used in our hospital. The AD Cooperative Study – Activities of Daily Living (ADCS-ADL) inventory was administered to the trial partner. The ADCS-ADL assesses 23 items (scores range from 0 to 78, with higher numbers indicating better daily function) and has good test-retest reliability . The Quality of Life in AD (QOL-AD) questionnaire was also administered to the trial partner. The QOL-AD assesses 13 items (scores range from 13 to 52, with higher numbers indicating better quality of life) and has good test-retest reliability . Assessments included body weight measurements and blood tests for glycosylated haemoglobin (HbA1C), triglycerides, high-density lipoprotein (HDL), low-density lipoprotein (LDL), and total cholesterol. An adverse effects questionnaire was given to both patient and trial partner by a nutrition specialist at weeks 6 and 12 within each treatment period. The week 12 assessment for the second treatment period included a questionnaire asking whether the patient or trial partner would continue the ketogenic diet after completing the trial.
During the first treatment period, patients and trial partners were prohibited from making copies of the diet plans. At the week 12 assessment of the first treatment period, all plans were returned and a 10-week washout period ensued. During this interval, patients and trial partners were repeatedly reminded to return to their usual diet (all trial partners vouched that the patients complied).
The lead investigator and nutrition specialists delivered an educational programme, consisting of two global e-mails per week and a 10-min video posted on the trial website every weekend. The programme provided information about facts and misconceptions relating to ketogenic and low-fat diets. Both diet approaches were consistently presented as potentially healthy and all patients were encouraged to eat until satiation.
Primary and secondary outcomes
Primary outcomes were mean within-individual changes in cognition (ACE-III), daily function (ADCS-ADL), and quality of life (QOL-AD) from baseline to week 12. Secondary outcomes included mean within-individual changes in cardiovascular risk factors (weight, body mass index, HbA1C, triglycerides, HDL, LDL, and total cholesterol) from baseline to week 12.
Sample size calculations considered that a 5-point change on the ACE-III was clinically meaningful (generally agreed by physicians at our hospital), while a 2-point change on the ADCS-ADL was considered clinically meaningful as this degree of change represents a gain or loss of independence in one domain of daily function . A 3-point change on the QOL-AD was considered clinically meaningful as this represents a change from very poor to excellent, or vice versa, in one domain of quality of life . For each outcome, we conservatively assumed a within-individual standard deviation equal to 50% of the between-individual standard deviation, resulting in a calculated conversion factor of 10 for the crossover design . To obtain 90% power at a significance level of 0.05, 18 patients were needed to complete both diets to detect an ACE-III change of 5 ± 10 points, an ADCS-ADL change of 2 ± 4 points, or a QOL-AD change of 3 ± 6 points. Since the only previous study involving a ketogenic diet in patients with uniform diagnoses of AD showed a 27% dropout rate over 12 weeks , we sought to recruit 25 to 30 patients.
Given the pre-trial uncertainty regarding data distribution, all outcomes were analysed using (nonparametric) Wilcoxon signed-rank tests. To check for a period effect, we performed Mann-Whitney U tests on the baseline means for all comparison groups (by treatment period, by treatment sequence, and for all patients). Statistical tests were two-tailed and considered an alpha of 5% as statistically significant. Data are presented as mean ± standard deviation unless stated otherwise.
We analysed primary and secondary outcomes using data from all randomized patients, with missing data imputed using regression imputation. We also performed an efficacy analysis on primary outcomes using data solely from “completers” who remained on protocol for both treatment periods and achieved sustained physiological ketosis (12-week mean beta-hydroxybutyrate level ≥ 0.6 mmol/L) during the ketogenic diet intervention.