Evaluation of the Ketogenic Diet to Improve Post Operative Cognitive Decline in Cardiac Surgery

Although there have been significant improvements in the mortality seen in cardiac surgery, post-operative cognitive decline (POCD) remains a major problem, occurring in up to 50% of elderly patients. The classification of this problem has been difficult, given the distinct aspects of post-operative delirium and post-operative cognitive dementia; however, with their many overlapping potential pathologies, there is increasing recognition that these represent a continuum of post-operative cerebral dysfunction. Overall, the mean duration of POCD is four days, but the cognitive decline can persist for months or even years after surgery. As reviewed by Greaves in coronary artery bypass graft (CABG) patients, POCD decreases to \~25% 6-12 months following surgery; however, \~40% of patients who developed acute POCD will also experience long-term POCD lasting 1-5 years. With POCD being a debilitating and costly problem for older patients, there is a need for more effective preventative and interventional measures, as well as a better understanding of underlying risk factors contributing to POCD development.

The pathophysiology of POCD is clearly complex and for a given patient, can be difficult to specifically identify causes. Overall, the literature suggests at least two major contributing and potentially overlapping hypotheses. First, multiple studies have suggested that anesthetics cause a range of molecular changes e.g., affecting apoptosis, neuronal growth as well as learning. Pre-clinical studies have found that volatile anesthetics increase amyloid-β (Aβ) peptide production and accumulation, causing morphological changes to mitochondria, disruption of calcium homeostasis, increased permeability and ultimately apoptosis through activation of several caspases, and human studies have linked such abnormalities with clinical measures of delirium and dementia. A second major hypothesis is based in inflammatory changes because of the surgery or anesthetic. Available literature has shown that volatile anesthetics can affect immune cell function and cardiac surgery itself, which includes cardio-pulmonary bypass and organ reperfusion injury, is a major activator of systemic inflammation and oxidative stress, resulting in several proinflammatory cascades and ultimately, increased neuroinflammation. From there, neuroinflammation has been linked to neurodegenerative diseases such as Alzheimer's Disease, Parkinson's Disease, and HIV-dementia, as well as cognitive dysfunction. Consistent with both hypotheses, patients with pre-surgical cerebral dysfunction have been found to be more vulnerable to POCD. Preoperative neuroimaging studies have shown that whole-brain gray matter atrophy, impaired white matter integrity, decreased functional connectivity in the orbitofrontal cortex, and cortical dysfunction are important predictors for POCD, presumably because the challenges coronary artery bypass grafting (CABG) pose include lengthy anesthesia, including volatile anesthetics, that "inactivates" the brain, are more detrimental in an already compromised brain.

With the evident complexity of POCD, it is hypothesized that a therapeutic approach that can act in multiple ways to modulate these pathophysiological effects may be constructive. With its myriad effects on cell signaling, changes in amyloid-beta or tau deposition, anti-oxidant effects and effects on immunological function, it is proposed that the ketogenic diet (KD) may be reasonable avenue for this. Indeed, the KD has now been studied in several small clinical trials for mild cognitive impairment. Taylor 2018 used a 1:1 gram ratio of lipid:non-lipid foods (calorically, 70% fats, 20% protein, 10% carbohydrates) to report on N=10 participants who were able to complete the dietary plan over a course of 3 months. All but one patient exhibited an improvement in the Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-cog) scores with the mean ADAS-cog score starting at 25.5 which decreased to 21.4, p\<0.02. The KD has long been used for epilepsy, and as reported by Poorshiri 2023, with a more aggressive diet (calorically, 90% calorie intake were fats), N=34 participants completed the diet for three months, with 21 of the 34 (62%) of the participants responding positively with >50% seizure reduction.

The KD has been considered in heart failure as well. Von Bibra 2014 compared a mild low carbohydrate (LC) diet (25% carbohydrates, 30% protein, and 45% fat) with low fat diets in patients with cardiac failure and type 2 diabetes for a period of three weeks. The participants on the LC diet (n=16) had significant improvement in both their diastolic cardiac function and fasting glucose. Thus, the KD has been studied both in the laboratory and clinic, finding improvements in both spheres. With the limited timeframe available for our study, a 14-day implementation of the KD vs. control diet (CD) comparison will be implemented. It might be argued that 14 days is insufficiently long to achieve the needed effect. However, the conversion of the brain from glucose to ketone oxidation can be performed efficiently. As discussed for clinical management, ketosis can be achieved by 2 to 4 days after starting the KD. As well, the original work from Freeman and Vining showed that a 3-day intervention in pediatric epilepsy could effectively reduce seizures by 50%. With the direct provision of food per the study, problems associated with meal preparation will be eliminated which the investigators believe will expedite the performance of the study. A diet plan that calorically provides 70:20:10 fat/protein/CHO, matching the profile of Taylor 2018, will be targeted.

While this study can be entirely performed on a clinical basis, the investigators believe that a better understanding of the success (or not) of the KD to reduce the occurrence of POCD may be had with metabolic imaging with MR spectroscopy. In vivo human brain magnetic resonance spectroscopic (MRS) imaging have been used for more than 30 years to evaluate in vivo brain metabolism and function, allowing measurements of N-acetyl aspartate (NAA), glutamate GLU, creatine Cr and others. NAA is strongly correlated with neuronal mitochondrial function, with early work from Bates and Heales finding that NAA synthesis localized to neuronal mitochondria and correlated with adenosine-triphosphate (ATP) synthesis rates. The MRS measurements can be quantified to tissue water or as a ratio taken to total Creatine (tCr = creatine + phosphocreatine) e.g., NAA/tCr; with tCr present in both neurons and glia, this ratio provides a convenient index of neuronal mitochondrial function that is corrected for cerebral spinal fluid content (which contains negligible quantities of these metabolites when compared to the tissue concentrations).

In pathology, the NAA/tCr ratio has been related to many aspects of clinical and scientific interest. For example, the team previously performed hippocampal MR spectroscopic studies in epilepsy patients who underwent intracranial monitoring and micro dialysis sampling for seizure evaluation. A significant negative relationship between NAA/tCr to the micro dialysis (extracellular) measurements of the major inhibitory neurotransmitter gamma-aminobutyric acid (GABA) was found in patients with medial temporal lobe epilepsy (MTLE), while it was positive with neocortical (non-MTLE) patients. The selectivity of these relationships was consistent with the view that in "healthier" brain, the extracellular GABA increases (appropriately) with better mitochondrial function while in diseased tissue (region of seizure onset), GABA appears to fail to suppress the abnormal seizure activity i.e., GABA increases as mitochondrial function falls.

MRS has also been used to measure the major cerebral ketone β-hydroxybutyrate (BHB). Under non-ketotic conditions, the concentration of brain BHB is \~zero; however, with fasting (or with use of the KD), brain BHB rises to near millimolar levels (the other major ketone is acetoacetate (ACAC); however, as the redox couple to BHB, the ratio of ACAC/BHB has been reported at 1:3 in fasted subjects and thus at much lower concentration). In cerebral ketosis, there is adaptation of BHB transport across the blood brain barrier, giving a linear relationship between plasma and cerebral BHB. The adaptation may be individual-dependent and thus measurements of brain BHB can provide an objective measure of the extent of cerebral ketosis. In patients, the measurement of BHB has been largely performed at 3 Tesla; however, because of its low and variable concentration, it is a measurement that requires high signal-to-noise (SNR). With increasing therapeutic use of ketosis and the KD for neurological disorders, a robust measurement of cerebral BHB can be very informative, which should benefit from the higher SNR at 7T.