KETO-MOOD: Ketogenic Diet for Microbiome Optimization and Overcoming Depression – A Protocol for a Randomized Controlled Trial

Major depressive disorder (MDD) is a serious disorder marked by at least one discrete depressive episode lasting at least 2 weeks and involving marked changes in mood, interests, and pleasure, changes in cognition, and vegetative symptoms [1]. MDD occurs about twice as often in women than men and affects about 6% of the adult population worldwide every year [2, 3]. MDD is the second leading contributor to the global chronic disease burden (measured by years lived with disability). In addition, MDD is associated with an increased risk of developing conditions such as diabetes mellitus, heart disease, cancer, and stroke, further increasing the disease burden. MDD strongly increases the risk of suicide. Up to 50% of the approximately 800,000 worldwide suicides annually occur within a depressive episode. Patients with MDD are almost 20-fold more likely to die by suicide than the general population [1]. Remission rates following antidepressant treatment appear lower than previously estimated. A recent reanalysis of the STAR*D trial indicates a cumulative remission rate of 35.0% after up to four antidepressant treatment trials [4]. Treatment-resistant depression is associated with more somatic comorbidity, substance abuse, and excess mortality, including death by suicide [5, 6]. Given the high individual and societal disease burden, there is a strong incentive to identify alternative or augmentative treatments. Moreover, pharmacological antidepressant treatment has significant side effects [7], which alternative approaches might potentially circumvent. In the realm of psychiatry, we find ourselves at a critical juncture. While the existing model primarily centered around psychopharmacology has yielded some modest success in tackling the global challenges of mental health, it falls short of addressing the complexity of the ongoing mental health crisis [8]. Recent evidence underscores the significance of nutrition in the prevalence and onset of mental disorders, highlighting its equivalence in importance to other medical disciplines [9]. Typical Western diets are rich in simple carbohydrates [10]. Recently, there has been increasing concern regarding the potential adverse effects of elevated carbohydrate consumption on the brain and mental health [11]. Prospective observational and interventional studies show that diets with a high glycemic load increase depressive symptoms in healthy participants [12, 13]. High dietary glycemic load and the resultant compensatory responses lower plasma glucose to concentrations that trigger the secretion of autonomic counter-regulatory hormones. These hormones are mainly cortisol, adrenaline, growth hormone, and glucagon, which exhibit a negative impact on mood [11].

The ketogenic diet is a form of nutrition that significantly reduces carbohydrate intake compared to standard dietary recommendations, typically allowing only 20–50 g of net carbohydrates per day, which accounts for about 5% of daily caloric intake. At the same time, fat intake is increased to around 75–80% of calories. By limiting carbohydrate intake and inducing lipolysis, the KD prompts the production of circulating ketone bodies. These ketones can be measured through standardized, reliable, and widely available self-administered tests, making ketogenic diets the only nutritional therapy with a straightforward objective biomarker to assess both physiological target achievement and adherence [14]. These ketone bodies serve as an additional source of fuel for the brain, thereby reducing its reliance on glucose [15]. Furthermore, ketone bodies offer a multitude of effects that can be therapeutically leveraged, including the amelioration of metabolic dysfunction such as improvement of lipid profile and stabilization of insulin levels, inhibition of the mTOR signaling pathway, enhancement of mitochondrial function and energy production, reduction of oxidative stress and inflammation, and rebalancing of the inhibitory-excitatory balance in the brain [16‒19]. Therefore, we propose reserving the term “ketogenic diet” for an intriguing but as yet insufficiently scientifically substantiated approach aimed at illness prevention and wellness preservation in otherwise healthy individuals. Meanwhile, its therapeutic application in serious disorders, grounded in pathophysiological reasoning, should be termed “ketogenic metabolic therapy” (KMT). Ketosis has a long track record in human history. Mild ketosis is frequently observed in both mothers and infants during the third trimester of pregnancy and at the time of birth [20]. Prolonged fasting and ketosis were the norm during human evolution, especially in the Paleolithic, when social structure was based on small clusters of hunter-gatherers [21, 22]. The use of ketogenic diets for therapeutic aims in modern medicine dates back to the 1920s [23]. A century ago, the ketogenic diet was a standard of care in diabetes, used to prolong the life of children with type 1 diabetes and to control the symptoms of type 2 diabetes in adults. However, the discovery of insulin in the 1920s enabled people with diabetes to countervail hyperglycemia despite high-carbohydrate diets [21]. Nowadays, the ketogenic diet is an established, effective non-pharmacologic treatment for drug-resistant epilepsy in both children and adults, and, thus, has a firm place in contemporary epilepsy treatment guidelines [24]. Randomized controlled trials have established that short-term ketogenic interventions reliably reduce body weight and visceral adipose tissue, improve metabolic health markers – such as increased HDL cholesterol, lowered triglycerides (TGs), HbA1c, and diastolic blood pressure – while maintaining muscle mass [19]. Multiple lines of evidence point toward favorable impacts of the ketogenic diet on the brain and psyche [18]. However, the research on ketogenic diet’s efficacy in psychiatric illness is still in its infancy. In a study involving 28 treatment-resistant patients with MDD, bipolar disorder, and schizophrenia, the ketogenic diet resulted in a notable mitigation of psychiatric symptoms in 100% of patients. Clinical remission was attained by 43% of patients, while 64% were discharged from the hospital with reduced psychiatric medication. Metabolic health showed improvement, and nearly all patients, with the exception of one, experienced a significant weight loss [25]. In a 4-month single-arm pilot trial involving 23 patients with bipolar disorder and schizophrenia, ketogenic diet therapy led to the reversal of metabolic syndrome. Participants with schizophrenia demonstrated an average improvement of 32% according to the Brief Psychiatric Rating Scale. Additionally, 69% of participants with bipolar disorder exhibited more than a one-point improvement in the Clinical Global Impression score. Overall, this pilot trial indicates dual metabolic-psychiatric benefits from ketogenic therapy [26]. In another pilot trial lasting 6–8 weeks, applying KMT in bipolar disorder, involving 27 recruited participants, demonstrated the feasibility and safety of this approach in a psychiatric population [27]. Before the contemporary era of new trials on KMT in mental illness, a series of studies in the pediatric domain have demonstrated potential to enhance core symptoms and improve the core features of autism spectrum disorder [28‒31]. Despite evident limitations such as small sample sizes and the absence of a control group, existing studies thus far have demonstrated the feasibility, tolerability, and significant improvements in psychiatric symptoms associated with the application of KMT in various serious, chronic, and refractory mental disorders. Moreover, it has been shown that the highly favorable safety profile known from a century of research in epilepsy and decades of research in obesity medicine is reproducible in these vulnerable cohorts. However, adequate real-world patient data are lacking. Further studies encompassing a broader range of disorders, particularly controlled trials, are urgently needed to establish conclusive evidence regarding the efficacy of KMT for various psychiatric conditions.

Depressive disorders are multifactorial, with no single cause. Risk factors include sex, gender, socioeconomic status, social support, stress levels, substance use, genetic and epigenetic influences, inflammation, comorbid medical conditions, endothelial dysfunction, and dietary patterns [32, 33]. These factors interact and influence one another, forming a complex web of risk factors. The idea that immune dysfunction, particularly chronic low-level inflammation, could contribute to depression and serve as a treatment target was first suggested in the early 20th century [33]. Patients with depression consistently exhibit elevated levels of inflammatory cytokines, including interleukin-6 (IL-6), tumor necrosis factor-α (TNFα), and C-reactive protein (CRP), compared to nondepressed controls [34]. Several mechanisms may link inflammation to depression. Early life stress and environmental stressors have been shown to elevate pro-inflammatory cytokine levels, particularly in males [35]. These cytokines can influence second messengers like prostaglandins, activate microglia (brain immune cells), and affect pathways including the glutamatergic system and the tryptophan-kynurenine pathway [36‒38]. These pathways are crucial for neurotransmitter regulation and neuronal function, and their dysregulation may contribute to the onset and progression of depression [39, 40]. Additionally, cerebral inflammation has been causally linked to glutamate-mediated excitotoxicity [41]. Meta-analytic evidence suggests that anti-inflammatory treatments, including celecoxib, other NSAIDs, and omega-3 fatty acids, show some benefit in reducing depressive symptoms [42]. However, their effects lack sufficient validation and reproducibility to be recommended as standard care.

KMT’s anti-inflammatory efficacy is mainly supported by animal studies. In a lipopolysaccharide (LPS)-induced fever model in rats, a ketogenic diet reduced pro-inflammatory cytokine levels compared to controls, likely due to increased anti-inflammatory n-3 PUFA derivatives [43]. Mice on an eucaloric ketogenic diet challenged with systemic LPS showed reduced levels of circulating IL-6 and TNF-α [44]. In a murine model of MPTP-induced neurotoxicity, a ketogenic diet resulted in lower levels of pro-inflammatory cytokines IL-1β, IL-6, and TNF-α in the brain [45]. Additionally, β-hydroxybutyrate (β-OHB) was shown to alleviate depressive symptoms in rodents and reduce neuroinflammatory markers [46]. A single dose of β-OHB reduced hippocampal IL-1β and TNF-α levels, possibly via NLRP3 inflammasome inhibition [47]. Microglia, brain-specific immune cells, regulate neuronal (or brain) development, synaptic plasticity, and mood. Alterations in microglia are implicated in the pathophysiology of depression [48]. In mice with LPS-induced depressive behavior, β-OHB exerted an antidepressant effect by promoting microglial ramification, a morphological change affecting immune function [46]. Together, these findings support the antidepressant effects of ketone bodies through their immunomodulatory actions.

The connection between gut microbiota and mental health is a key research area, with evidence suggesting it may affect brain function and behavior in MDD and other mental health disorders [49]. The mechanisms of this connection are complex [50]. A recent review summarized five mechanisms by which the microbiota may influence brain health across the lifespan: (1) modulation of hypothalamic-pituitary-adrenal axis function, (2) alteration of inflammatory marker release and gut permeability, (3) stimulation of the vagus nerve affecting permeability, motility, and epithelial secretion, (4) production of neurotransmitters and hormonal changes, and (5) modification of myelination, myelin plasticity, and mRNA expression in the prefrontal cortex [51]. Preliminary evidence suggests ketogenic nutrition alters the gut microbiome. Well-designed human trials suggest potential neuroprotective microbiome modulation through well-formulated ketogenic nutrition. In a randomized crossover trial in subjects with mild cognitive impairment, a Mediterranean-style ketogenic diet, compared to the American Heart Association (AHA) diet, increased the beneficial genus Akkermansia, reduced fecal lactate and acetate, and increased propionate and butyrate [52]. In a weight-loss trial, a very low-calorie ketogenic diet increased gut microbiota diversity compared to a standard low-calorie diet [53]. In an RCT with obese individuals, calorie restriction (CR), intermittent fasting, and KD significantly increased microbiota diversity compared to a habitual ad libitum diet; only the KD group showed an increase in the beneficial Faecalibacterium prausnitzii [54]. Secondary analysis in this trial suggested that LPS induced inflammatory processes may have mediated the favorable impact of gut microbiota modulation on mitochondrial function. Changes with ketogenic nutrition depend on the specific formulation and control condition. In a recent RCT comparing therapeutic carbohydrate restriction (<8% of daily calories) with a moderate-sugar as well as a low-sugar diet meeting public health guidelines, no differences were observed between groups after 12 weeks in taxonomic composition, circulating short-chain fatty acids, or LPS-binding protein (a proxy for intestinal barrier integrity) [55]. Therefore, further high-resolution studies of microbiota changes in controlled human trials are needed to understand the impact of ketogenic regimes on human gut microbiota, particularly in the psychiatric context, where such studies are currently lacking.

Our study aims to investigate the efficacy of the KMT in alleviating MDD symptoms, filling a critical gap in psychiatric treatment options and offering a novel dietary approach with potential to mitigate disease burden and enhance mental and physical well-being. This study is a prospective, assessor-blinded, controlled trial with a randomized, parallel-arm design, classified as phase 2. The focus of the study is on a nutritional intervention as the independent variable. The research will take place at the University Psychiatric Clinics (UPK) located in Basel, Switzerland. Patients have the option to enroll in the study either as outpatients, inpatients, or initially as inpatients with subsequent transition to outpatient care. The study duration for each participant in both treatment arms is 10 weeks. The study intervention involves a modified Atkins ketogenic diet over 10 weeks, including up to 2 weeks of gradual adaptation followed by 8 weeks of full ketogenic implementation. The control group will receive counseling based on a standard balanced mixed diet in accordance with public health recommendations. We will track contact time with all participants, including both direct (e.g., meetings, calls) and indirect (e.g., emails, texts) interactions, using a participant log. Both dietary interventions are designed to provide comparable engagement (Fig. 1). This study includes both inpatients and outpatients. Inpatients in both arms will receive meals provided by the hospital kitchen, with ketogenic meals prepared by a team trained in specialized diets for medical needs. Outpatients, or those who transition to outpatient status, will be instructed on how to purchase suitable groceries and prepare meals at home. Regarding compensation, after careful consideration, it has been decided not to offer compensation to avoid unintentionally incentivizing or exploiting vulnerable populations, as compensation in clinical trials presents complex ethical concerns [56].

The 10-week duration is considered appropriate for depressive episodes. Determining the optimal duration for ketogenic therapy in depression is challenging due to limited data. In antidepressant trials, therapeutic plasma levels are typically reached by week 2 [57], similar to the adaptation phase in this study. Most younger patients who do not show significant improvement by weeks 4–5 are unlikely to respond to extended treatment [58‒61]. A pooled analysis of 12-week antidepressant trials in older adults indicates that those not achieving a 30% reduction in depression scores by week 4 have a low likelihood of achieving remission by week 12 [62]. Additionally, a meta-analysis suggests a plateau in depression score reduction between 8 and 12 weeks [63], supporting the adequacy of a 10-week period. This duration is seen as an optimal balance, minimizing the drawbacks of longer trials, such as lower retention, participant fatigue, higher resource costs, and prolonged exposure to experimental treatments. Based on preliminary retrospective results, the initial response to KMT is expected to occur after 3 weeks in a fat-adapted state [25]. In a recent case series on KMT for depression, two patients experienced remission of major depression (PHQ-9 scores ≤4) and generalized anxiety (GAD-7 scores ≤4) within 7 weeks of starting therapeutic nutritional ketosis, while one patient took 12 weeks to achieve similar results [64].

Follow-up cessation support will be provided at the end of the trial or to participants who wish to discontinue. For those interested in continuing the ketogenic diet long-term, individualized support will be offered based on their specific needs.

Individuals who meet the ICD-10/11 criteria for a diagnosis of unipolar MDD or are experiencing a current depressive episode within the context of bipolar affective disorder will be enrolled. For recruitment, patients at the UPK Basel, who are entering or undergoing either inpatient or outpatient treatment for depression, will be asked whether they are willing to participate in our study. There will be a consecutive ongoing recruitment in the daily clinical practice. If participants declare interest, they will be contacted by the study investigator and screened for inclusion and exclusion criteria. The investigators will explain to each participant the nature of the study, its purpose, the procedures involved, the expected duration, the potential risks and benefits and any discomfort it may entail. Each participant will be informed that the participation in the study is voluntary and that he or she may withdraw from the study at any time and that withdrawal of consent will not affect his or her subsequent medical treatment. Patients will also have the opportunity to ask any questions regarding their study participation. All prospective participants will receive a participant information sheet and consent form detailing the study, allowing them to make an informed decision. Participants will have adequate time to decide whether to participate. If eligibility is confirmed, they will be given 48 h to decide. Formal consent, using the approved form, will be obtained prior to any study procedures.

Inclusion criteria:

• Unequivocal diagnosis of MDD or bipolar depression according to ICD-10 and ICD-11 (as soon as approved in Switzerland) criteria

• Patient Health Questionnaire nine-item depression (PHQ-9) score of ≥10 (indicating at least moderate severity) [65]

• Age ≥18 years

• The patient can give informed consent as documented by signature

• Interest in trying a dietary intervention.

Exclusion criteria:

• Inability to follow the study procedures, e.g., because of a language barrier, neurological and interfering mental disorders, dementia

• Anorexia nervosa

• BMI <18.5 kg/m²

• Adherence to an exclusively vegan diet

• Pregnancy or breast feeding

• Current electroconvulsive therapy

• Concurrent ketamine therapy

• Porphyria

• Substance use disorder according to ICD10 F10-F19

• Type 1 diabetes

• Insulin-dependent type 2 diabetes

• Contraindicated medical conditions; besides rare hereditary metabolic disorders (typically diagnosed in childhood), contraindications comprise acute pancreatitis, nephrolithiasis, advanced renal failure, advanced liver failure, advanced congestive heart failure, advanced pulmonary disease with respiratory failure, and concurrent use of SGLT2 inhibitors [55, 66].

Several variations of the classical ketogenic diet are currently in common use: typical/classical ketogenic diet, the medium chain TG diet, the modified Atkins diet (MAD), and low glycemic index treatment [67]. In our study, we will utilize the MAD ketogenic diet approach. Previous observations have indicated that individuals following the MAD show improved adherence to the diet, as evidenced by either self-reported positive urinary ketosis or accurate diet recall (56%) compared to classic ketogenic diet (38%) [68], while the anti-seizure efficacy is comparable [69]. The MAD also produces fewer side effects and is overall better tolerable than the classic ketogenic diet [70]. On the MAD, the daily net carbohydrate intake is limited to 20 g for adolescents and adults. Carbohydrates are mainly derived from foods with a high fiber content, such as non-starchy vegetables, nuts, and seeds. Protein intake will be personalized based on individual needs. The usual recommendation for the MAD is 1.2–1.6 g/kg of reference body weight per day, with the possibility of increasing up to 2 g/kg per day depending on factors such as age, activity level, and health status. Physically highly active as well as and sicker individuals may require higher protein intake. Fat intake is strongly encouraged on the MAD diet, ideally composing 60%–70% of total calories. Vitamins, minerals, and trace minerals will be supplemented as needed, guided by the initial laboratory measurements. We assume a 2-week fat-adaptation period. Food photo journaling will be conducted using the YouAte app, Cronometer or MealLogger and evaluated during regular dietitian meetings. During the initial 4 weeks, dietary counseling will be provided weekly, and thereafter biweekly. We are offering a hybrid counseling model where the initial consultation is typically conducted in person, and subsequent sessions can be held either in person or via secure video call, according to the participant’s preference. Additionally, participants will engage in regular phone or online communication via secure email or the secure messenger Element with members of the research team. Patients can also send meal photos via a messenger app; no special app is required. The study spans 10 weeks, starting with a 2-week adaptation phase. During this phase, one meal out of the three daily meals will be replaced with a ketogenic meal every 3–4 days. By the end of the first week, all patients will be on a fully ketogenic diet, with another week allocated for the body to adapt. Following this, participants will maintain a steady state of ketosis for the remaining 8 weeks. Participants in the experimental group will thus undergo an 8-week ketogenic diet intervention, characterized by reduced carbohydrate intake and increased fat consumption, following the initial 2-week adaptation phase. Successful adoption of the ketogenic diet is defined by maintaining capillary β-OHB levels ≥0.8 mmol/L and a glucose ketone index <6.

The control group will be provided with a standard balanced mixed diet following the recommendations for healthy nutrition by the Schweizerische Gesellschaft für Ernährung (Société Suisse de Nutrition). This diet will consist of an average daily caloric supply from carbohydrates of approximately 45–60%, as per the guidelines [71]. These recommendations focus on complex carbohydrate sources, such as whole grain products. This group will be managed with a flexible carbohydrate approach, allowing patients the option to reduce their intake to a low-carb level. Our non-ketogenic nutrition program provides a carbohydrate intake slightly above what is typically considered low-carb non-ketogenic, with 100–115 g of carbohydrates from sources such as grains, starchy vegetables, fruits, and any added sugars. Additionally, there are extra carbohydrates from nuts, seeds, a substantial amount of non-starchy vegetables, and dairy products (while a low daily carbohydrate intake [non-ketogenic] ranges from 50 to 125 g per day, which constitutes 10–25% of total calories, a moderate daily carbohydrate intake is between 130 and 220 g per day, or 26–44% of total calories). This approach is expected to lower and stabilize glucose and insulin levels, providing metabolic benefits. The control group will receive the same nutritional counseling and support as the ketogenic therapy group. Besides the diet protocol, which is the cornerstone of the treatment plan, participants will also receive the usual care available to all patients in the respective unit, including psychotherapy, pharmacotherapy and complementary treatments such as occupational therapy or mindfulness training, as needed.

We employed simple randomization through the “randomizeR” R package, generating a random sequence and utilizing a random allocation rule. During the trial, any conversation regarding diets between assessors and patients (or trial partners) is strictly forbidden. Allocation concealment will be ensured using RedCap software, preventing the assessor from being aware of the group allocation.

The current treatments available for major depression often fall short in providing sufficient relief. In light of this, our main hypothesis is that the inclusion of an add-on KMT will exhibit superior efficacy in alleviating symptoms of MDD compared to standard of care treatments. The primary objective of this study was to investigate the effectiveness and safety of a 10-week course of KMT.

Furthermore, a secondary objective of this study was to assess whether KMT has a positive impact on relevant biological markers associated with depression, including circulating inflammatory parameters and the gut microbiome. We hypothesize that KMT, compared to the control diet, will lead to greater reductions in inflammatory parameters, increased gut microbiota diversity, and an expansion of beneficial, short-chain fatty acid-producing taxa like Faecalibacterium prausnitzii, along with corresponding microbial metabolic pathways.

Primary outcome measure:

• 1.

  Depressive symptoms severity:

  • -

    Assessment tool: Patient Health Questionnaire nine-item depression scale (PHQ-9)

  • -

    Time points: baseline (week 0), week 6, week 10.

Secondary outcome measures:

• 1.

  Mindfulness (present-focused attention, awareness, and nonjudgmental acceptance of thoughts and emotions):

  • -

    Assessment tool: Cognitive and Affective Mindfulness Scale-Revised (CAMS-R)

  • -

    Time points: baseline (week 0), week 6, week 10.

• 2.

  Self-efficacy:

  • -

    Assessment tool: General Self-Efficacy Scale (GSE)

  • -

    Time points: baseline (week 0), week 6, week 10.

• 3.

  Sleep disturbance:

  • -

    Neuro-QOL SF v1.0 – sleep disturbance

  • -

    Time points: baseline (week 0), week 6, week 10.

• 4.

  Cognitive function:

  • -

    Neuro-QOL SF v2.0 – cognitive function

  • -

    Time points: baseline (week 0), week 6, week 10.

• 5.

  Functional Outcome:

  • -

    Assessment tool: Work and Social Adjustment Scale (WSAS)

  • -

    Time points: baseline (week 0), week 6, week 10.

• 6.

  Serum beta-hydroxybutyrate (β-OHB) monitoring:

  • -

    Assessment Tool: Keto Mojo GK+ Blood Glucose & Ketone Meter

  • -

    Time Points: Participants in both arms will be required to collect daily finger-prick blood samples (1.0 µL) 2 h before dinner, typically between 15:00 and 16:00.

• 7.

  Glucose ketone index (molar ratio of circulating glucose over β-OHB):

  • -

    Assessment tool: keto mojo GK+ blood glucose and ketone meter

  • -

    Time points: daily from baseline onward.

• 8.

  Serum level of highly sensitive CRP:

  • -

    Assessment tool: ultrasensitive ELISA

  • -

    Time points: baseline (week 0), week 6, week 10.

• 9.

  Homeostatic Model Assessment for Insulin Resistance (HOMA-IR; includes fasting glucose and insulin measurements):

  • -

    Time points: baseline (week 0), week 6, week 10.

• 10.

  Body weight (kg):

  • -

    Time points: baseline (week 0), week 6, week 10.

• 11.

  Waist-hip ratio:

  • -

    Time points: baseline (week 0), week 6, week 10.

• 12.

  Visceral adiposity index:

  • -

    Time points: baseline (week 0), week 6, week 10.

• 13.

  Fecal metagenomics shotgun sequencing:

  • -

    Assessment tool: fecal metagenomics shotgun sequencing

  • -

    Time points: baseline (week 0), week 10

  • -

    Analyzing fecal samples via shotgun sequencing to determine gut microbiota composition and diversity as well as metabolic functional potential.

Additionally, we employ a series of nutritional adequacy and safety measures detailed in online supplementary File 1 (for all online suppl. material, see https://doi.org/10.1159/000542979).

The planned statistical method for testing the hypothesis regarding the primary endpoint is a two-sample t test conducted at the 10-week assessment. The sample size calculation for the primary endpoint, the PHQ-9 score, was performed based on a two-sample t test, assuming normally distributed data. This superiority pilot trial seeks to demonstrate the efficacy of KMT in reducing PHQ-9 scores after 10 weeks, compared to the control intervention, in patients with depression. A 5-point difference in the PHQ-9 total score (scale 0–27) is regarded as the minimal clinically important difference, as proposed by Löwe et al. [65]. The study reported standard deviations (SDs) for the change in PHQ-9 scores of 5.8 over 3 months and 6.1 over 6 months. Assuming an SD of 6 over 10 weeks, a standardized effect size (Cohen’s d) of 0.83 on the PHQ-9 would be considered clinically relevant. To detect differences between groups with 90% power and a 5% type I error rate, a sample size of 64 participants (32 per group) is required. Considering dropout rates, typically around 20% for behavioral interventions [72, 73], 30% for drug treatments [74‒76], and up to 35% for short-term (≤3 months) ketogenic diet interventions [77], a final sample size estimation of 43 participants per treatment arm, or a total of n = 86 participants, was determined to ensure adequate statistical power after study attrition.

Demographics and clinical data will be recorded for all patients. Categorical data will be analyzed as frequencies and percentages. For continuous variables, the mean and SD will be reported. We will present the distribution of relevant baseline characteristics in a table. Despite efforts to control for covariates through randomization, mismatches may occur. Relevant covariates, whether statistically significant or not [78], will be adjusted for in statistical analyses, particularly those related to socioeconomic status, body composition, metabolic syndrome, and illness severity. Discrepancies in contact time will also be addressed as potential confounders in the models. For the primary hypothesis, changes in PHQ-9 scores from baseline to week 10 will be compared between groups. A mixed-effects model will be employed, incorporating treatment group, time, and their interaction as fixed effects, with individual subjects as random effects. Mixed models are reliable in psychiatric clinical trials, effectively managing missing data and accounting for multiple comparisons [79‒81]. In the primary analysis, mixed models will use available case analysis, assuming data are missing at random. Sensitivity analyses will assess the robustness of conclusions to the missing at random assumption. A linear contrast will be applied to test for differences between groups in changes from baseline to week 10. Two-tailed tests will be conducted with a significance level set at 0.05. Secondary outcomes will be analyzed using similar mixed-effects generalized linear models. Pre-specified subgroup analyses of the primary outcome will examine baseline depression severity (severe vs. moderate) and depression duration, categorized by the median. Differences in adverse events between study arms will be reported as proportions, with significance evaluated using Fisher’s exact test. Considering an estimated treatment adherence rate of approximately 65% for antidepressants [82], both intention-to-treat and per-protocol analyses will be conducted, with patients compliant to the ketogenic diet on at least 65% of ketone measurements being included in the per-protocol analysis.

Fecal metagenome samples will be processed with the Illumina Nextera DNA Library Preparation kit and sequenced on the HiSeq Platform with 2 × 125 bp paired-end reads. We will use the vegan package for numerical ecology analyses, calculating Shannon entropy, Bray-Curtis dissimilarity, and gene family richness. Multidimensional scaling and permutational ANOVA will assess between-sample diversity, while differential abundance testing will utilize edgeR. The metabolic functional potential will be analyzed using the HUMAnN pipeline.

We will conduct exploratory mediation analyses [83] to assess whether changes in waist-hip ratio, gut microbiome parameters, high-sensitivity CRP, area under the curve (AUC) for β-OHB as measure of euketonemia, and cognitive and affective mindfulness mediate changes in depressive symptoms as measured by PHQ-9. Recently, the visceral adiposity index – an empirical index derived from simple anthropometric measures (BMI and waist circumference) and lipid parameters (TGs and HDL cholesterol) – has demonstrated a strong correlation with MRI-measured visceral adiposity. Consequently, VAI is regarded as a reliable, sex-specific marker for assessing visceral adipose distribution and function [84, 85]. We will calculate VAI and include its change over time in the mediation analysis. We will quantify alpha diversity, abundance of key taxa (e.g., F. prausnitzii and Akkermansia muciniphila), and functional pathways, using changes in these metrics in our mediation analysis.

Depressive disorders are widespread and often severe, with current treatments proving inadequate. There is a critical need for novel therapies, particularly for patients resistant to pharmacological options. Ketogenic nutritional therapy, a safe metabolic intervention with significant effects on brain function, may offer a novel approach by directly targeting disease mechanisms and alleviating comorbidities like obesity and type 2 diabetes. This study aims to address the need for alternative treatments, potentially reducing suffering and improving quality of life for patients. The ketogenic diet offers a range of benefits due to its nonpharmacological nature, making it relatively safe. With a well-established track record of over 100 years, it has demonstrated efficacy in epilepsy management. We are advancing gut-brain axis research by evaluating ketogenic therapy as a dietary strategy to restore healthy microbiota in depression. Several reviews have explored ketogenic therapy’s potential to alleviate depressive symptoms, primarily based on theoretical biological reasoning [18, 86‒88]. Animal studies provide mechanistic support for its antidepressant efficacy [89‒93]; however, no prospective human trials exist, and current evidence is limited to retrospective case reports and small case series [25, 64, 94‒96]. This study introduces a rigorously designed randomized controlled trial with an active comparator, incorporating high-resolution, in-depth metagenomic profiling. This novel approach will yield valuable insights into the clinical efficacy and gut-brain axis signaling effects of ketogenic therapy in humans.

The trial has 80% power to detect at least medium effects, but may miss smaller, potentially valuable effects. An active comparator certain intervention, expected to improve insulin levels, metabolism, and neuroinflammation to a limited extent, is included to balance expectations and reduce Hawthorne effects. If effective, the treatment could offer a new management approach, with further research needed to evaluate ketogenic therapies in routine clinical practice. Depressive disorders are inherently heterogeneous, with patients often presenting nonoverlapping symptom clusters, yet still meeting the ICD or DSM criteria for MDD [97]. Furthermore, comorbidities are common in real-world clinical settings, and we anticipate that a proportion of our participants will have conditions such as obesity, metabolic syndrome, or type 2 diabetes [98]. Consequently, we aim to adopt a more inclusive study design that mirrors real-life patient populations. It is well recognized that RCT populations often fail to reflect the diversity seen in clinical practice, especially with regard to vulnerable characteristics [99], and we seek to avoid this limitation. We are confident that the substantial effect size of our intervention will accommodate this inherent variability and complexity.

We express our gratitude to Professor Shebani Sethi, MD, for her valuable conceptual advice. We also thank the Department of Clinical Research at the University of Basel for their support, particularly Klaus Ehrlich, Head of Monitoring, and the regulatory team, especially Stephanie Spannl, PhD, and the Team Coordination & Project Management, with special thanks to Sina Hansen, PhD. We are grateful to Andrea Wiencierz PhD for her statistical advice. Our thanks also go to the entire hospital kitchen team for their invaluable assistance in planning the ketogenic meal support.

The study was approved by the Local Ethics Committee (Ethikkommission Nordwest-und Zentralschweiz, EKNZ), under Project ID 2023-01945, and is registered at https://clinicaltrials.gov/with the trial registration number NCT06105762. At the time of submitting this protocol, no participants had been recruited for the study yet. All participants planning to participate in the study will be provided with a participant information sheet and a consent form describing the study and providing sufficient information for participants to make an informed decision about their participation in the study. The formal consent of a participant, using the approved consent form, will be obtained before the participant is submitted to any study procedure.

K.H. is the founder and CEO of flexiMENTE Health and Resilience Coaching. A.L. is the founder of MyKetoBrain. E.M., U.L., and O.S. are partners at Mason & Lanz Metabolic Solutions, while Y.R. serves as a consultant. T.L. is a medical advisor to Mason & Lanz Metabolic Solutions. The authors recognize the importance of adhering to general nutritional guidelines and advocate for dietary modifications only when there is a clear health rationale, supported by professional guidance from experts in dietary counseling and therapy. Authors who practice dietary counseling provide a range of tailored approaches, from standard public health recommendations to therapeutic dietary restrictions – including carbohydrate restriction – when supported by evidence (e.g., for conditions such as obesity or metabolic syndrome). Individual dietary practices vary among the authors and have also changed over time; for instance, some have personally explored both carbohydrate-based vegan diets and animal-based ketogenic diets.

The study is supported by the nonprofit organization Gertrud Thalmann Fonds.

Katarzyna Hongler: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, software, validation, visualization, writing – original draft, and writing – review and editing; Astrid Lounici: conceptualization, data curation, investigation, methodology, project administration, resources, software, and writing – review and editing; Erin Maurer, Orsolya Szathmari, Yvonne Reuter, Sandra Nussbaum, Ines Steinborn, Annika Haedrich, Melina Anastasia Mölling, Ulf Wein, Iona Bocek, Luca Hersberger, and Ueli Lanz: project administration and writing – review and editing; Annette B. Brühl and Undine E. Lang: resources and writing – review and editing; Timur Liwinski: conceptualization, funding acquisition, supervision, writing – review and editing, data curation, formal analysis, investigation, methodology, project administration, resources, software, validation, and visualization.

Katarzyna Hongler and Astrid Lounici share first authorship.Undine E. Lang and Timur Liwinski share last authorship.

Next-generation sequencing data and analysis code will be made publicly available. The clinical data supporting the findings of this study will not be made publicly available, as they contain information that could potentially compromise the privacy of research participants. However, these data can be made available upon reasonable request from the corresponding author (T.L.).