Wide ranging cognitive deficits are major drivers of functional decline and poor outcomes in people with schizophrenia (SZ) and bipolar disorder (BD). Medications do not target pathophysiological mechanisms thought to underlie these deficits. In the search for interventions targeting underlying cognitive impairment in SZ and BD, we look comprehensively beyond just the brain and to the potential role of dysfunctional systemic metabolism.

Disrupted insulin and glucose metabolism are seen in medication-naïve first-episode SZ, suggesting that SZ itself, and not just the medications used to treat it, is associated with risk of Type 2 diabetes, cardiovascular morbidity and mortality, and more generally, accelerated aging. Even young people with SZ have increased risk of metabolic disease and cognitive deficits. Sadly, their life span is shortened by 15-20 years. BD is associated with similar but less severe disruptions in glucose and insulin metabolism and life expectancy.

Although the human brain is 2% of the body's volume, it consumes over 20% of its energy, and accordingly, the brain is particularly vulnerable to the dysregulation of glucose metabolism seen in SZ and BD. While glucose is considered to be the brain's default fuel, ketones provide 27% more free energy and are a major source of energy for the brain. Ketones prevent or improve various age-associated diseases, and a ketogenic diet (70% fat, 20% protein, 10% carbohydrates) has been posited as an anti-aging and dementia antidote.

The premise of the work is based on recent evidence that ketogenic diets improve dynamic neural network instability, related to cognitive deficits, aging, and Type 2 diabetes (Mujica-Parodi et al., Proc Natl Acad Sci U S A. 2020;117(11):6170-7.). The rigor of the work rests on findings of (1) poor cerebral glucose homeostasis in SZ and BD, (2) neural network instability in SZ and BD, and (3) direct effects of ketosis on network instability. Unknown is whether ketogenic diets can improve network instability in people with SZ and BD.

The diagnosis of schizophrenia (SZ) and bipolar disorder (BPD) as traditionally been based on "positive symptoms," such as delusions and hallucinations, and "negative symptoms," such as anhedonia and amotivation. Although not part of the diagnostic criteria, wide ranging cognitive deficits are common, and they are major drivers of functional decline, as well as poor social and occupational outcomes experienced as illness chronicity sets in. While antipsychotic medications treat positive symptoms, they do not improve cognitive deficits, nor do they target pathophysiological mechanisms thought to underlie these deficits. Accordingly, in the search for interventions targeting brain dysfunction underlying cognitive impairment in SZ, the investigators will look comprehensively beyond the brain to the potential role of dysfunctional systemic metabolism, given that obesity, insulin resistance, and associated systemic inflammation are co-morbidities. Modern anti-psychotic medications disrupt metabolic homeostasis, which may contribute to the brain dysconnectivity thought to underlie cognitive deficits. However, both SZ and BPD have been associated with disrupted insulin and glucose metabolism, reported appearing well before the advent of antipsychotic treatment, and consistent with a recent meta-analysis indicating these metabolic disturbances. In fact, insulin resistance and both SZ and BPD have been genetically linked. Thus, SZ and BPD themselves are associated with metabolic disease, while the anti-psychotic medications acutely induce insulin resistance, independent of food intake and weight gain, compounding the associated metabolic susceptibilities. The cause-and-consequence relationship of these disorders and insulin resistance is unknown, and whether re-establishing metabolic homeostasis improves the underlying neural substrates of cognition is also unknown.

The brain is an obligate "glucovore" and is particularly vulnerable to changes in glucose metabolism. Robust energy demands of the brain cannot be met by lipid transformation, and during times of glucose deprivation, they must be satisfied by ketone bodies. Disrupted central glucose metabolism, as observed in SZ and BPD patients, modulates peripheral metabolism by re-allocation of nutrients towards a brain-centric focus to maintain critical central functions. Low-carbohydrate high fat, or ketogenic, diets are an emerging therapy for insulin resistance, Type 2 diabetes, and associated co-morbidities. Increased ketones prevent or improve the symptoms of various age-associated diseases, reduce inflammation and the production of reactive oxygen species, and upregulate mitochondria in the brain. In addition, ketogenic diets have shown promise, but without the needed controls.

The premise of this proposal is based on a recent paper showing a ketogenic diet reduced 7T resting state fMRI neural network dynamic instability, a measure of how long a network of independent nodes maintains a stable connection. Instability is related to cognitive deficits, aging, and Type 2 diabetes in neurotypical adults. The investigator's fMRI data show similar network dynamic instability in SZ and BPD, adding to a larger literature showing static brain network dysconnectivity underlying neurocognitive deficits. Unknown is whether network instability can be rescued with a ketogenic diet, and whether improvements are mediated by ketogenic diet-induced increases in available ketone bodies as brain fuel, and/or with reductions in systemic inflammation and indices of metabolic syndrome.

The rigor of the proposed work rests on findings of (a) poor glucose homeostasis in SZ and BPD, (b) neural network instability in SZ and BPD, and (c) direct effects of ketosis on network instability in neurotypical adults. Unknown is how ketogenic diets might improve network instability in overweight/obese SZ and BPD with risk of insulin resistance. The investigators propose a mechanistic, prospective, pilot clinical study comparing 4-weeks of ketogenic diet (KETO) vs. diet as usual (DAU) on neural network instability in SZ and BPD. They will randomize 70 SZ and BPD (40-65 years old, balanced for sex) to KETO (n=35) or DAU (n=35). KETO meals will be delivered to participants by Metabolic Meals. Metabolic, inflammatory, and 7T MRI data will be acquired before and after the 4-week diet.

Aim 1: Assess changes in network instability with KETO and DAU in SZ and BD over the 4-week period. Hypothesis 1: KETO, relative to DAU, will improve network stability.

Aim 2: Establish metabolic and inflammatory indices as correlates of change in network instability with the KETO diet. Hypothesis 2: Improvements in network stability will be correlated with increased circulating ketone levels, and improved insulin sensitivity, reduced visceral fat, weight loss, and reduced systemic inflammation.

Aim 3: Assess neuropsychological function at baseline to determine whether it is correlated with baseline network instability in SZ and BD, similar to what has been reported in neurotypical adults. Hypothesis 3: Cognitive deficits will be related to network instability in SZ and BD at baseline. The over-arching hypothesis is: Disrupted metabolic homeostasis contributes to neural network instability in SZ and BD and that induction of ketosis restores it.

Inclusion Criteria:

• Male or female ages 18-65 years old
• Able to attend 2 in-person visits at the San Francisco VA Medical Center
• SCID-5 Schizophrenia or schizoaffective disorder stable on 2nd generation anti-psychotic (SZ) or bipolar disorder (BD)
• Not currently on KETO diet
• Willing to adhere to 4 wk. KETO diet or 4 wk. non-KETO diet per random assignment
• Speak, read, comprehend English
• Access to internet
• Willing and able to heat up KETO meals

Exclusion Criteria:

• Current Cancer diagnosis
• Other SCID-5 Axis 1 disorder
• Pregnancy, breastfeeding, or planned pregnancy
• Current diagnosis of type 1 Diabetes Mellitus
• Glucose-lowering drugs (other than metformin) or weight loss pills
• History of gastric bypass surgery or any weight loss surgery
• >10% weight fluctuation in past 2 years