Depression is a debilitating chronic illness affecting 1 in 6 adults in the United Kingdom (UK) at any one time. Antidepressants and psychological therapy are the main treatments, but some people do not respond to these. Neurons and signals in the brain are greatly disrupted in people with severe depression. A ketogenic diet, a high-fat and very low-carbohydrate diet, supplies a form of energy that appears to help brain cells communicate and may improve the treatment of depression. Our goal is to find out whether a ketogenic diet could be an additional effective treatment for patients with depression for which antidepressants do not work.

Using social media advertising, 100 patients, ages 18-65, who have previously tried at least two different antidepressant medications within the current depressive episode will be recruited. Enrolment, consent, and data collection will be collected online using self-report questionnaires. Participants will be allocated by minimisation 1:1 to the KD group or control group based on depressive severity (moderately severe vs. severe) and body mass index (<30kg/m2 vs. 30+ kg/m2). The intervention group will receive 6-weeks of prepared ketogenic diet meals (3 meals with snacks per day) and weekly ketogenic diet-focused nutrition counselling. The control group will be asked to follow a diet to reduce their saturated fat intake and increase vegetable consumption by one portion a day. The control group will receive vouchers to assist with purchases and will be provided with weekly nutritional counselling. Existing treatment for depression will remain in both groups.

The primary outcome is the change in depression symptoms at six weeks. All participants will complete assessments of depression and anxiety every two weeks, starting before treatment to post-intervention (week 6), and again at week 12. Additional outcomes include participants' ability to experience pleasure, quality of life, ability to socialise and work, cognitive processing, morning cortisol, and gut microbiome.

At all stages of the study, adults with lived experience of depression will advise the research team to take into account the needs and views of patients.

This study will provide evidence of whether following a ketogenic diet leads to a short-term improvement in depression in people whose depression cannot be relieved by antidepressants.

One in 6 people suffer from clinical depression. Antidepressants are a first-line treatment for depression, but at least 1 in 3 patients with depression do not respond to antidepressants and not everyone responds to, or wants psychological therapy. Furthermore, there is still a high relapse rate after psychological treatment. Longer depressive episodes, a higher number of hospitalisations, moderate to high suicide risk, physical and psychiatric comorbidities, and higher dosage of medication, were partly explained by the low response (16-17%) and remission rates (13%). Hence, new treatment options for treatment resistant depression (TRD) are needed.

Recent research in depression indicates that the imbalance between excitatory (e.g. glutamatergic system) and inhibitory (e.g. GABA system) systems may be a better way of understanding depression. There is strong evidence that GABAergic neurotransmission is disrupted in depression and concentrations of GABA are significantly reduced, especially in treatment-resistant depression. This suggests increasing inhibitory neurotransmitters (e.g. GABA) could be an alternative way to treat depression.

KDs (ketogenic diets), characterised by high-fat and very low-carbohydrate intake, have been recently suggested as a possible intervention for depression due to the influence on neurotransmitters. KDs increase production of the inhibitory neurotransmitter GABA, and restore microglial activation and neuronal excitability in the lateral habenula, a region involved in negative reward processing. Furthermore, the mechanisms of action of KD appear to include several additional potential mechanisms of action in depression including cellular bioenergetics, reduced oxidative stress, improved brain vascularity, and influence on the HPA axis (via lowered cortisol), and changes in gut microbiome.

Despite several plausible beneficial effects and increased public interest and rapid market growth of KD food products (estimated $12.4 billion by 2024), and evidence of biological plausibility demonstrated in epilepsy and animal models, the efficacy of KDs for treatment-resistant depression has not yet been established. To our knowledge, there are only three registered clinical trials underway testing KD for mental disorders, however, none examine efficacy of KD for treatment-resistant depression and there are no currently published results. A recent systematic review is the first to appraise the literature on KD for mental disorders in human samples and highlights the paucity of robust research. Twelve studies examining 388 participants (9 case reports, 2 cohort studies, and one observational study) met inclusion criteria and found no high-quality evidence of KD efficacy for mental disorders. The absence of controlled trials, heavy reliance on individual case studies, grouping of various clinical disorders together, lack of uniform definition of KD across studies, including five studies not reporting KD nutritional intake, and no measurement of ketosis in half the studies make interpretation of the literature difficult. However, the review data suggest that people with various mental illnesses who follow a KD can have marked improvements in mental health. Moreover, a recent review suggests that symptom scales such as the PHQ-9 may be useful but should be better combined with measures of functioning and quality of life to fulfil patients' perspectives.

The current study proposes to test whether a 6-week ketogenic diet is an effective treatment for treatment-resistant depression. Recruitment of 100 participants with treatment-resistant depression based on a sample size calculation will be conducted. A 5-point difference in PHQ-9 will be considered to represent a clinically important difference based on prior research. This study reported that the Standard Deviation (SD) of the change in PHQ-9 over 3 months as 5.8 and as 6.1 over 6 months. Assuming an SD of 6 over 6 weeks would suggest a standardised effect size of 0.83 in PHQ-9 to be clinically relevant. The sample size to test differences between groups at 90% power and at a type one error rate of 5% would be 64 (32 per group). A sample size of 100 participants will be recruited to make the study robust against up to 35% attrition/missing data. In simulations, it has been confirmed that this sample size remains robust under scenarios such as unequal sizes of the strata used in minimisation.

After all baseline assessments are complete, participants will be allocated by minimisation 1:1 to the KD group or control group. A researcher will embed a non-deterministic algorithm in the Redcap database. This algorithm aims to produce treatment groups balanced for important prognostic factors by minimising on the following variables (as below). Only after eligibility and consent is confirmed will the database reveal the allocation. Investigators delivering the intervention will not be blinded, but the outcome is collected blind (by self-report), and those involved in analysing the outcome data will be blinded to allocation.

• BMI (less than 30 kg/m2 vs. 30+ kg/m2)
• Depression severity (PHQ-9 score: 15-19 [moderately severe] vs. 20-27 [severe])

All follow-up is done online and without trial staff and will therefore not be subject to observation bias.

Following randomisation, participants will be given information and advice for their respective group. The KD group involves 6-week pre-prepared ketogenic diet with weekly KD-focused behavioural and nutritional counselling. The control group will receive weekly nutritional counselling to increase vegetable consumption and reduce saturated fat intake. Participants in the control group will receive food vouchers to help purchase these items. This aims to be a plausible placebo dietary treatment for depression. Usual treatment for depression will continue, and leaflet material is provided in both groups.

Psychometric questionnaires and biological outcomes will be assessed pre-post intervention, and at 12-week follow-up, to investigate changes in depression symptom severity and potential mechanistic pathways.

Initial descriptive statistics will present the profile of the subjects by study arm without using statistical comparisons. For the primary hypothesis, the change in PHQ-9 scores from baseline to week 6 will be compared between groups. A mixed effect model that includes treatment group, time and their interaction as fixed effects, and individual subjects as random effects will be fit using PHQ-9 at all available assessments. A linear contrast will be used to test the difference between groups in changes from baseline to week 6 in PHQ-9 scores. Two tailed tests and significance level of 0.05 will be used. Secondary outcomes will be analysed using analogous mixed effects generalised linear models. Pre-specified subgroup analyses of the primary outcome will be explored by baseline depression severity (severe versus moderate) and duration of depression at baseline split at the median. An exploratory mediation analysis using previously validated methods to examine whether adherence to the diet, changes in microbiome, and cortisol awakening response appear to mediate changes in depression. Spearman correlations will determine associations between alpha diversity metrics, individual microbes, and PHQ-9 scores. Kruskal-Wallis one-way analysis of variance tests will be performed to compare relative abundance of the top ten genera, with false discovery rate (FDR) corrections between dietary groups. Complete case analysis will be used as the primary analysis if the proportion of missing data is below 5%. A sensitivity analysis to explore reasons for missingness and determine appropriate handling of missing data will be conducted.

A 12-member advisory panel of adults with lived experience of mental and physical illness (50% from ethnic minority backgrounds) in collaboration with The McPin Foundation has been formed to inform all stages of this research trial. Members have expressed interest to know whether KDs can treat depression, and they welcome this kind of research. An additional 8 members of the public who have depression, of whom 3 had previously followed a KD diet were involved in the study design (e.g. social media and public advertising to engage participants, prepared meals preferred, motivational messages, strategies aimed at addressing relapsing and relapse following KD diets) before ethics submission. Participants in the trial will be offered the opportunity to hear the results of the study upon completion, and a lay summary and infographic will be provided.

A six-member trial steering committee (TSC) has been formed to provide independent oversight of the trial. Three independent members (50%) whose scientific, psychiatric, and statistical expertise is directly relevant to the trial, have verified they have no relationship to the investigators, the trial funders, nor employed by the same institution. The TSC will meet at least annually at appropriate time-points to ensure the trial is conducted in accordance with clinical trial standards.

There are few established and widely available treatments available for people with TRD. This is the first randomised, controlled trial assessing the effects of KDs on symptoms of depression in patients with TRD. Findings from this trial will provide evidence of efficacy of a new option for alleviating depressive symptoms in people who do not respond to antidepressants. If the treatment proves efficacious, it will create a potential new way to treat these patients.

Inclusion Criteria:

• Aged 18-65 years
• Have been diagnosed with depression
• Currently on an antidepressant medication for a period of 4-weeks or longer
• Have tried 2 or more antidepressant treatment trials of adequate dose and duration within the current depressive episode to which they did not fully respond. An adequate treatment trial is defined as at least 4 weeks of a medication at a minimum National Institute for Health and Care Excellence (NICE) approved dose
• With a Patient Health Questionnaire-9 (PHQ-9) total score greater than or equal to 15 at baseline
• Able to understand and be willing to adhere to the demands of the study
• Provision of written informed consent
• Have access to a tablet/computer for online assessments, follow-ups with the registered dietitian, and able to attend appointments for assessments and treatment and adhere to study procedures
• Have both a fridge and a freezer at home
• Complete all baseline assessments

Exclusion Criteria:

• Currently following a low carbohydrate or ketogenic diet
• Currently following a vegan or vegetarian diet as these diets are more challenging to accommodate in a Ketogenic Diet and adding vegetables in the control group is unlikely to be seen as helpful.
• Currently receiving, or have received, in-patient psychiatric treatment or electroconvulsive therapy (electric shock to the brain under brief general anaesthetic) within the past year, or scheduled to receive such treatment during the study
• Currently using St John's wort or other remedies for depression that were bought without a doctor's prescription
• Currently have suicidal ideation with intent* or attempted suicide within the past two months
• Ever had an eating disorder, bipolar disorder, schizophrenia, or psychosis
• Have substance use or alcohol dependence
• Have epilepsy
• Have serious food allergies (experiencing food hypersensitivity that leads to anaphylaxis or other severe symptoms, which may require hospitalisation or are life-threatening) or otherwise require a special diet that cannot be accommodated within a KD such as phenylketonuria or lactose intolerance
• Treated with insulin, sulfonylureas, Glucagon-like peptide-1 (GLP-1) agonists, or Sodium-glucose co-transporter-2 (SGLT2) inhibitors
• Women who are pregnant, planning pregnancy in the next three months, or breastfeeding
• Have a body mass index (BMI) of <18.5 kg/m2
• Have unstable or severe medical conditions (e.g., cancer, cardiovascular, renal, lung, psychiatric, or bleeding disorders, diabetes, etc.), currently receiving cancer treatment except hormonal treatment for breast cancer or non-melanoma skin cancer treatment
• Have gallstones, renal tubular acidosis, kidney stones, small bowel malabsorption or a history of pancreatitis
• Have scheduled a major surgery in the next 3 months
• Taking part in other studies that may compromise this study or this study may compromise the other study/ies
• Have read the trial protocol or the clinical trial registration information and therefore are unblinded
• Live in the same household as another participant in the trial
• Not able to complete the online task with a tablet/computer
• Not willing to provide saliva, urine and stool samples

We plan to share the study protocol, statistical analysis plan, and individual participant data (IPD) that underlie the results in a publication.

Data will be available after publication in a peer-reviewed journal for a period of up to 12-months.

Please contact the primary investigators listed on this trial registration for access.