Biological gases, oxidative stress, artificial intelligence, and machine learning for neurodegeneration and metabolic disorders
Kenneth Maiese
Abstract
The close association between neurodegeneration and metabolic disease: Neurodegenerative diseases impact a large portion of the global population, are estimated to affect more than one billion individuals and result in the death of more than seven million individuals annually.1,2 Disorders of the nervous system include more than 600 disorders. However, cognitive disorders may be one of the more debilitating types of nervous system disorders, and dementia is expected to require more than two trillion United States dollars per year in the United States alone for daily care and maintenance. Additional finances will be necessary for companion care, social services, and senior home care. Interestingly, cognitive loss and dementia are currently the seventh leading cause of death, and less than 30 years from now more than 160 million individuals in the world will suffer from cognitive loss, such as Alzheimer’s disease (AD). One reason for the continued increase in the prevalence of neurodegenerative disorders is the concurrent increase in life expectancy. The lifespan approaches 80 years of age on average, and the number of individuals over the age of 65 has doubled during the prior 50 years. Life expectancy is increasing not only in developed nations but also in developing nations, with the expectation that the aging population will increase from 5% to 10%. Multiple factors may be responsible for improved lifespans, which involve greater access to healthcare, improved infectious disease and sanitation measures, and the ability to focus more effectively on populations at the highest risk. Some of the highest-risk populations involve individuals who suffer from metabolic disorders, such as diabetes mellitus (DM). Similar to that of individuals with neurodegenerative disorders,3 the number of individuals with metabolic disorders is on the rise, and within another 20 years, it is expected that more than 700 million individuals will have DM. Unfortunately, these considerations do not account for individuals who remain undiagnosed, since more than 400 million individuals throughout the world are believed to suffer from underlying metabolic disease and have a significant risk for the eventual onset of DM. It is estimated that more than 800 billion United States dollars per year are necessary to care for individuals with DM for glucose control, nutritional management, wound care, disability occurrence, and functional loss. Metabolic disorders that involve DM are chronic and progressive in nature and affect all organs of the body, leading to renal disease, liver impairment, cardiovascular disorders, and neurodegenerative disease. With respect to neurodegeneration, DM affects the nervous system through multiple pathways that result in endothelial dysfunction, mitochondrial impairment, immune system failure, loss of stem cells, and decreased resistance to infections. These pathways can ultimately lead to cognitive loss, demyelinating disorders, and peripheral nervous system impairment, such as peripheral neuropathies. A common link for degenerative disorders – oxidative stress: A prominent pathway that can lead to the onset and progression of neurodegenerative disorders and metabolic disease involves biologic gas entities that lead to the generation of reactive oxygen species (ROS) and oxidative stress.4,5 During metabolic disease and diseases of the nervous system, ROS are produced during oxidative stress. ROS are formed from gaseous entities that include singlet oxygen, nitric oxide, peroxynitrite, superoxide free radicals, and hydrogen peroxide. Oxidative stress results in programmed cell death via autophagy, apoptosis, ferroptosis, mitophagy, and pyroptosis. Oxidative stress impairs mitochondrial organelle function and leads to cytochrome c release and caspase activation, which involve programmed cell death. Mitochondria that become injured can be removed from cellular structures during mitoptosis to protect against cell death. Peroxisomes, also known as microbodies, are membrane-bound cellular organelles involved in oxidative stress that produce hydrogen peroxide and catalase to decompose hydrogen peroxide and lead to the catabolism of fatty acids in neurodegenerative disorders and metabolic disease.6,7 In addition, iron-dependent accumulation of ferrous ions can result in oxidative stress-induced phospholipid damage through ferroptosis, a programmed cell death pathway that negatively affects glutathione homeostasis.2,8 Through these processes, oxidative stress can impair the development of stem cells, lead to cardiovascular disease, destroy hepatocyte function, and result in cellular senescence by blocking the function of telomeres. This ultimately leads to the onset of age-related diseases, loss of energy homeostasis, metabolic dysfunction, nerve cell injury, loss of immune system function, and the inability of cells and tissues to respond to toxic environments. Artificial intelligence (AI) and machine learning (ML) for oxidative stress pathways that impact neurodegeneration and metabolic disorders: Given the importance of oxidative stress pathways in the onset and progression of neurodegenerative disorders and metabolic diseases such as DM, identifying new avenues to assess the generation of ROS is imperative for the development of new clinical treatments. Currently, treatments for neurodegenerative disorders are limited and usually cannot address disease progression. For example, disease-modifying therapies for multiple sclerosis can limit the number of relapses in relapsing–remitting multiple sclerosis, but these treatments do not prevent the progression of the disease. In individuals with cognitive neurodegenerative disorders such as AD, cholinesterase inhibitors reduce memory loss, and immunotherapy treatments that limit amyloid deposition in the brain may improve cognition in subgroups of patients; however, patients are at risk for microhemorrhages in the brain, and ultimately, memory loss continues to progress. Similarly, a number of strategies can focus on the treatment of metabolic disorders and DM, including nutritional guidance, obesity reduction, increased activity, hypoglycemic pharmaceutical treatments, and attempts to monitor and maintain serum glucose homeostasis. However, whether employed individually or in combination, these treatments for disorders such as DM may slow disease progression to limit multiorgan damage, but disease progression may still ensue, and off-target clinical treatments may also result in vascular and neuronal disease with organ dysfunction. New strategies that employ AI and ML are driving new directions to assess and potentially treat the underlying cellular pathways tied to oxidative stress, neurodegenerative disorders, and metabolic disease. AI with ML offers the advantage of assessing the input of vast arrays of data from multiple parameters that are nonhomogeneous, such as genetic inputs, cellular assessments, liquid biopsies, imaging data, and pathological tissue, that are formulated to offer predictive disease signature pathways that cannot be obtained from statistical methods alone. AI is applicable for multiple disease entities that can involve pulmonary fibrosis, obstructive pulmonary disease, cancer onset and progression, cognitive loss, and metabolic disease. With respect to cognitive loss and metabolic dysfunction, studies have evaluated plasma proteins in individuals with dementia and mild cognitive impairment via ML algorithms to identify potential diagnostic biomarkers. Dysregulation of several plasma proteins was identified in individuals with dementia, which could differentiate patients with severe cognitive loss from individuals with mild cognitive impairment. Furthermore, several underlying pathways, including programmed cell death mechanisms and Wnt proteins known to be associated with oxidative stress, DM, and aging, have been identified.9-11 Using ML in key gene modules in patients with AD, nucleosome assembly protein 1-like 1, which is known to be involved in nucleosome assembly and processes associated with longevity, cellular metabolism, and resistance to oxidative stress, was also identified as a potential predictive biomarker for AD.12 These studies underscore the intimate relationship of neurodegenerative disorders with cellular metabolism since current work also points to the oversight of inflammation and lipid metabolism, which can control microglial cell activity in individuals with AD.13 In addition, investigations employing ML to study aging processes have revealed that circadian rhythm genes and pathways related to longevity with silent mating type information regulation 2 homolog 1 (Saccharomyces cerevisiae) (SIRT1) are involved in the aging of cells by signaling mechanisms linked to immune cells and metabolism with insulin resistance.2 Notably, circadian rhythm clock genes play critical roles in cognitive loss, metabolic disease, aging, and mitochondrial energy maintenance. Loss of circadian rhythm function can promote dementia and cognitive loss by leading to sleep fragmentation, poor nutritional status, inability to remove toxins in the brain, such as amyloid, and cell injury through oxidative stress.14,15 Through the use of ML techniques, circadian rhythm clock genes have been associated with the infiltration of immune cells during the aging process. Furthermore, circadian rhythm dysfunction has the potential to increase the risk for aging as well as metabolic syndrome onset in the presence of gene defects.16,17 Through the use of ML applications, specific circadian rhythm clock genes have also been shown to aid in clinical prognosis and allow clinicians to place cancer patients into high-risk and low-risk groups, which impacts overall survival.18 The SIRT1 pathway oversees metabolic cellular mechanisms that control mitochondrial function, insulin sensitivity, lipid processing in the liver, and cell survival. SIRT1 also affects the circadian clock gene rhythm to prevent neurodegenerative processes. SIRT1 maintains sufficient cellular energy pools that can become depleted and lead to mitochondrial dysfunction and cognitive loss, which fluctuate with circadian rhythmicity during aging. In addition, SIRT1 can prevent immune dysregulation during oxidative stress, increase stem cell and neuronal cell survival, preserve memory function, and increase cell longevity, and studies employing ML have shown that it can regulate the aging of tissues.16 Conclusions and future directions: Neurodegenerative disorders are intimately linked to metabolic diseases, including DM, and impact a significant proportion of individuals worldwide with chronic disability and death. Metabolic disease affects all body systems, and the nervous system leads to cognitive loss, demyelinating disorders, risk for infection, and neuropathies in the peripheral nervous system.19,20 A shared underlying pathway for these disorders involves biologic gas entities that can lead to the production of ROS and oxidative stress. Oxidative stress can affect all organs of the body and lead to cellular metabolic dysfunction, stem cell impairment, programmed cell death, and mitochondrial dysfunction, which can involve mitoptosis, immune function dysregulation, and neurovascular cell injury. Given that current diagnostics and therapies for neurodegenerative disorders and metabolic diseases are limited and do not ultimately prevent the onset of disease progression, novel strategies that involve AI and ML may offer new directives to assess the fundamental pathways linked to oxidative stress, neurodegeneration, and metabolic disease. The techniques of AI and ML can provide predictive disease signature pathways and converge on common mechanisms of crosstalk leading to oxidative stress that are derived from multiple nonhomogenous data sources and cannot be obtained from statistical methods alone. Through the employment of AI and ML techniques, multiple pathways associated with oxidative stress have been identified, including Wnt proteins, nucleosome assembly protein 1-like 1, circadian clock genes, and SIRT1, which offer diagnostic biomarkers, potential treatment options, and prognostic information for disorders of the nervous system and metabolic pathways that can influence cognitive loss, metabolic disease, including DM, aging-related disorders, cancer onset, and immune system dysregulation (Figure 1). Both AI and ML offer great promise for elucidating the underlying pathways of oxidative stress that oversee neurodegeneration and metabolic disorders, but AI and ML are currently considered early in the development stages, and further investigations are needed to validate and translate these techniques into broader and clinically efficacious formats for patient care.Figure 1: Implementation of artificial intelligence and machine learning to assess disease signature pathways of neurodegenerative and metabolic disorders through oxidative stress pathways.Artificial intelligence and machine learning can offer vital assessments of disease signature pathways for neurodegenerative disorders and metabolic diseases, such as diabetes mellitus, which focus on the underlying mechanisms of oxidative stress and common pathways that can include Wnt proteins, nucleosome assembly protein 1-like 1 (NAP1L1), circadian clock genes, and silent mating type information regulation 2 (Saccharomyces cerevisiae) (SIRT1). Through the use of artificial intelligence and machine learning, new directions may unfold to assess and impact underlying disease cellular pathways that affect aging, cognition, inflammation, energy homeostasis, lipid metabolism, and cellular survival.This work was supported by the following grants to Kenneth Maiese: American Diabetes Association, American Heart Association, NIH NIEHS, NIH NIA, NIH NINDS, NS053956, and NIH ARRA.