Metabolic Interventions for Treatment-Resistant Depression

When depression does not lift after two or more adequate trials of antidepressant medication, it is described as treatment-resistant. For many people this is the point at which they begin to look beyond the standard menu of pharmacology and talk therapy. Increasingly, that search leads them to metabolic psychiatry, and to two interventions in particular: ketogenic metabolic therapy and intermittent fasting. The two are often discussed in the same breath. The evidence base behind them is very different.

What The Research Is Showing

The most clinically meaningful evidence for ketogenic metabolic therapy in refractory mood illness comes not from large trials but from carefully delivered pilot studies and case series, where the therapy was provided the way it is actually practised. The 2022 inpatient series from Toulouse, led by Albert Danan and Georgia Ede, followed 31 patients with treatment-refractory major depression, bipolar II disorder or schizoaffective disorder through a hospital-supervised ketogenic diet. Depression and mood ratings improved substantially, and the improvements were maintained at discharge. In 2024, a Stanford pilot led by Shebani Sethi followed 21 adults with bipolar disorder or schizophrenia, all of whom were on antipsychotic medication and all of whom had a metabolic abnormality at baseline, through a four-month ketogenic diet. Around four in five participants showed meaningful psychiatric improvement, and every participant who started the study with metabolic syndrome had it resolve. The Edinburgh group, led by Iain Campbell at the University of Edinburgh, has reported similar findings in bipolar disorder, including measurable changes in brain chemistry on MR spectroscopy that fit the proposed mechanism. A 2025 open trial in young adults with major depressive disorder reported substantial reductions in depression scores alongside a rise in BDNF, a protein that supports brain plasticity.

The mechanistic picture that emerges is consistent. The depressed brain often shows reduced capacity to use glucose efficiently for energy. Ketone bodies, produced when carbohydrate intake is low enough to shift the body into nutritional ketosis, offer an alternative fuel that bypasses this bottleneck. Ketosis also appears to reduce neuroinflammation, improve mitochondrial function, and shift the balance of the brain’s two main signalling chemicals, GABA and glutamate, in a direction that tends to calm overactive circuits.

The first attempt at a randomised controlled trial in treatment-resistant depression was published in February 2026. The Oxford DIME trial compared a prescribed ketogenic diet of pre-packaged meals against a control diet with matched weekly check-ins, over six weeks. The ketogenic group did improve more than the control group, with the difference most pronounced in participants with the most severe depression. Several aspects of the trial limit how much can be drawn from it, however. Ketosis was assessed by urine strips, which are not an accurate measure of blood ketones. The intervention was a uniform meal-delivery service rather than an education in ketogenic metabolic therapy. Medication changes during the trial were not recorded. And six weeks is shorter than the period over which mood effects in refractory illness typically consolidate. The trial confirms that the approach is safe and that a signal can be detected even under difficult conditions. It is not, on its own, the evidence on which clinical decisions should rest.

The evidence is genuinely promising. It is also early, and adherence is the central practical challenge. This is one of the reasons we provide structured, multidisciplinary support rather than handing patients a meal plan and wishing them luck.

What About Intermittent Fasting?

Intermittent fasting has captured public attention as a low-cost, low-effort metabolic intervention. The evidence in depression specifically is much thinner than for the ketogenic diet, and there are three reasons we are cautious about recommending it, particularly in refractory cases.

The first is the stress response. Fasting reliably activates the body’s stress system, raising cortisol. This matters for two reasons. In someone who is already running on chronic stress, poor sleep, or a dysregulated HPA axis, which describes a great many people with refractory depression, adding a daily physiological stressor is not obviously helpful and may worsen mood, sleep, and anxiety. It also undermines the metabolic goal. Elevated cortisol drives gluconeogenesis and pushes up insulin, and insulin is the single biggest suppressor of ketone production. In a person attempting to access the benefits of nutritional ketosis, a stress-driven cortisol spike from fasting can quietly cancel out the very mechanism the dietary intervention is meant to produce. Poor sleep does the same. The body cannot make ketones efficiently while it is in a stress-and-glucose state.

The second is the autophagy story. Autophagy is the cellular housekeeping process in which damaged components are broken down and recycled, and it is frequently invoked as the headline benefit of fasting. Most of what is claimed about autophagy in humans is extrapolated from research in mice. Mice have a basal metabolic rate roughly six times faster than ours. What happens in a fasted mouse at 24 hours does not happen in a fasted human at 24 hours. The handful of studies that have actually measured autophagy in fasted humans have found weak and inconsistent signals. The popular claim that 16 hours without food triggers meaningful cellular cleaning is not well supported.

The third concern, and the one we see most often in clinical practice, is the way intermittent fasting can disguise a problem as a solution. Many of the patients we see arrive convinced that skipping breakfast is healthy, because their blood glucose stays lower for longer and they feel no morning hunger. The trouble is that not being hungry in the morning is usually a sign of metabolic dysfunction, not metabolic flexibility. In most of the people we assess, the reason they are not hungry at breakfast is that they ate a large meal late the evening before, often with snacks afterwards, and they are still processing it. The wellness framing of 16:8 lets a maintaining factor for poor metabolic and mood health be reinterpreted as a virtuous practice. The eating window has shifted into the evening, circadian biology is working against them, insulin and cortisol rhythms are dysregulated, and the morning fast is the visible symptom rather than the cure. Until that pattern is recognised and reversed, no amount of fasting is going to address it.

How We Approach This Clinically?

For someone with treatment-resistant depression, particularly where there is also weight gain on medication, insulin resistance, metabolic syndrome, or a bipolar pattern, properly tailored ketogenic metabolic therapy is a reasonable next step alongside existing treatment. Done well, KMT is a clinical intervention that includes structured education on how and why the diet works, individualised macronutrient targets, blood ketone monitoring, ongoing review of mood, energy, sleep, and other symptoms, coordination with your prescribing doctors. The goal is for the patient to understand the therapy well enough to maintain it themselves long after the formal program ends.

Would we as a rule, recommend intermittent fasting as a metabolic intervention for refractory depression? The evidence base is not there, the physiological stress is real, and the cortisol response actively suppresses the ketosis that the dietary approach is trying to produce. If someone is already eating within a comfortable window, such as finishing dinner by eight and having breakfast at eight, that is unremarkable and we leave it alone. Actively prescribing longer fasts, particularly to women, to younger adults, or to anyone with a history of restrictive eating or evening-loaded eating patterns, is not something the current research supports.

This is the work of metabolic psychology: sorting the signal from the noise, matching the intervention to the person, and providing the structure and skills people need to do this, sustain it, and build it into their lives if they choose to.