Background
Alzheimer’s disease (AD) is characterised by progressive synaptic and axonal dysfunction, neuronal loss and cognitive decline [
1]. There is growing evidence that AD is closely associated with metabolic abnormalities and oxidative stress linked to critical elements of neurodegeneration, such as mitochondrial dysfunction and bioenergetic impairment [
2,
3]. Indeed, increasing data indicate that systemic metabolic disorders, such as insulin resistance, are strongly associated with bioenergetic failure of nerve cells [
4,
5]. This can manifest as cognitive impairment and brain-specific neuropathology, and share common pathogenic mechanisms with AD, such as impaired glucose metabolism, oxidative stress, insulin resistance, and amyloidogenesis [
4,
6,
7]. Recent evidence suggests that patients with type 2 diabetes mellitus are at an increased risk of developing AD [
6].
Although AD is defined by accumulation of abnormal amyloid and tau proteins [
8], the mechanistic assumption of linear causality between the amyloid cascade and cognitive dysfunction in AD is still lacking, since amyloid-lowering approaches have failed to provide cognitive benefits in human clinical trials [
9]. A growing body of evidence suggests that impaired brain energy metabolism and mitochondrial dysfunction in AD may contribute to cognitive decline. At the same time, different drugs for metabolic diseases are prescribed to improve metabolic status, slow cognitive decline or prevent dementia progression [
10]. Positron emission tomography studies have revealed baseline cerebral glucose metabolism abnormalities before the onset of cognitive symptoms in AD patients [
11]. In addition, recent preclinical data indicated that ageing and AD are associated with the reorganisation of brain energy metabolism and mitochondrial dysfunction, including an overall increase in lactate secretion and downregulation of bioenergetic enzymes [
12,
13].
Several studies have suggested that impaired brain energy metabolism and oxidative stress are associated with mitochondrial degeneration and abnormal protein accumulation during the progress of AD [
14‐
18]. In this context, emerging evidence suggests that the autophagy/lysosome pathways play a critical role in removing damaged mitochondria (mitophagy), and dysfunction of autophagy results in the accumulation of dysfunctional mitochondria in neurons [
19]. This opens up a new window of beneficial effects of interventions that maintain mitochondrial health and/or stimulate mitophagy in the neurodegenerative process in AD [
20]. In line with this, the beneficial effects of nicotinamide on mitochondrial integrity, autophagy and bioenergetics-related signaling in brain cells are associated with reduced accumulation of abnormal Aβ and tau in the hippocampus, and lead to the improved cognitive performance in transgenic mice [
21,
22]. Also, recent human studies suggested that the impaired hippocampal mitophagy in AD patients responds well to mitophagy enhancement strategies and such treatment finally improves AD-related tau pathologies in human neuronal cells and memory deficits in transgenic models [
23].
Combining multiple compounds to both reduce oxidative injury and improve bioenergetics, in other words, to target multiple pathways simultaneously, has been proposed as a therapeutic strategy associated more likely with successful translational outcomes [
24]. Previous research identified limited serine availability, reduced de novo glutathione (GSH) synthesis, and altered NAD+ metabolism in a transgenic mouse model of AD based on multi-omics profiling [
25]. While NAD+ is reduced in the AD animal models, NAD+ augmentation mitigates Aβ, tau, and metabolic pathologies in laboratory models of AD [
23,
26]. These findings have been confirmed by human metabolomic data showing significantly altered cerebrospinal fluid levels of acylcarnitine in patients with AD, which are correlated with the decline of cognitive function and structural abnormalities of the brain [
27,
28].
In addition to the above-mentioned metabolic underpinnings of AD, several neuroimaging studies have revealed alterations of critical cognitive regions, including hippocampus, cortex, and inferior parietal, middle frontal and occipital regions [
29]. For instance, Nagata et al. showed that the vulnerability of the hippocampus plays a potential role in memory and executive dysfunction in AD patients [
30,
31]. Similarly, several neuroimaging studies showed that cortical thickness plays a critical role in AD pathophysiology [
32,
33]. Despite these promising studies, no research has evaluated the common pathophysiological mechanisms shared by systemic metabolic alterations and specific brain areas involved in cognitive deterioration in AD patients.
Based on the integrative network analysis of multi-omics data of non-alcoholic fatty liver disease, we have developed the combined metabolic activators (CMA) consisting of
L-serine,
N-acetyl cysteine (NAC), nicotinamide riboside (NR), and
L-carnitine tartrate (LCAT, the salt form of
L-carnitine) and showed that administration of CMA activates mitochondria, and improves inflammatory markers in animals and humans [
34‐
38]. We have found that the CMA administration promotes mitochondrial fatty acid uptake from the cytosol, facilitates fatty acid oxidation in the mitochondria, and alleviates oxidative stress [
39]. Global metabolomic and proteomic profiling revealed that CMA administration effectively increases fatty acid oxidation and de novo GSH synthesis [
34]. Moreover, plasma levels of metabolites associated with antioxidant metabolism and inflammation are improved in COVID-19 patients treated with CMA compared to the placebo [
38]. Recently, we tested the individual metabolic activators and CMA in the streptozotocin-induced AD-like rats and showed that CMA administration significantly improved behavioural scores in parallel with neurohistological outcomes in this model [
40].
Based on these studies, we hypothesized that CMA administration may be a promising treatment for improving the metabolic parameters and brain functions in AD patients. Here, we designed a randomised, double-blinded, placebo-controlled human phase 2 clinical study to investigate the effect of CMA administration on the global metabolism of AD patients through comprehensive phenomics, metabolomics, proteomics and imaging analyses.
Discussion
Our results suggested that oral administration of CMA for 84 days has a considerable effect on cognitive function in AD patients based on ADAS-Cog scores. The cognitive functions in patients with high ADAS-Cog scores were improved in the CMA group while there were no significant differences in the placebo group. We also showed beneficial effects in both severe and mild patients. Cognitive functions of AD patients were improved by 29% in the CMA group, whereas they were improved only by 14% in the placebo group after 84 days, consistent with a placebo effect that is seen in other AD clinical trials [
50,
65,
66]. The improvement of cognitive function was supported by positive alterations in cortical thickness and maintenance of the hippocampal subfield volumes in the CMA group, while no cortical thickness difference but significant volume decline was found in the placebo group. Our finding of a possible beneficial effect in severe AD is of particular value since severe AD patients lack current therapeutic regimes, except for palliative support. Apart from clinical severity, we observed that various clinical variables were also related to the treatment response. For example, patients with low ALT and increased metabolic load (i.e., increased HbA1c and insulin levels) or impaired CBC values responded better to CMA treatment.
The effect of oral administration of CMA was substantiated with a comprehensive analysis of proteins and metabolites in the plasma of patients using a multi-omics analytical platform. The clinical results were consistent with the genome-scale metabolic modelling of more than 600 AD patients showing clear evidence of mitochondrial dysfunction [
40]. It is also consistent with the results from an animal model demonstrating improvement in AD-associated histological parameters in animals treated with oral administration of CMA [
67]. Thus, the present study suggests a promising therapeutic regime that might help to improve metabolic alterations in AD patients.
Considering the role of hippocampal and frontoparietal degeneration in AD pathogenesis, our neuroimaging observation of improved hippocampal subfield volumes and cortical thickness after metabolic stimulation was not surprising. Herein, we observed significantly improved hippocampal volumes as well as frontal and parietal cortical thickness in the CMA group, whereas no positive effects were observed in the placebo group. Our results suggested a treatment effect in the CMA group in major cognitive brain regions in AD.
Numerous studies in healthy participants have suggested that metabolism and cognitive brain network organization are strongly linked [
68,
69]. This is especially true for higher-order cognitive tasks, which are the most metabolically demanding for the brain, and "burn out" would make the brain prone to neurodegenerative and age-related alterations marked by neuronal metabolic dysfunction [
70]. Metabolic cofactors improve mitochondrial metabolism by a three-step strategy: (1)
L-carnitine to enhance the transport of fatty acids across the mitochondrial membrane, (2) the NAD+ precursor nicotinamide riboside to enhance the β-oxidation of fatty acids in mitochondria, and (3) GSH precursors including serine and NAC to form GSH that is required to protect liver against oxidative stress mediated by free radicals, which are generated through increased β-oxidation of fatty acids in the mitochondria. Thus, our results of the pro-cognitive benefits of CMA may be connected to the metabolic load of cognitive networks and their responsiveness to such metabolic interventions.
Growing data indicate that significant mitochondrial dysfunction occurs in AD pathogenesis [
71]. In our previous studies, we reported that CMA administration effectively facilitates fatty acid oxidation in the mitochondria and alleviates oxidative stress by de novo GSH synthesis [
34‐
39]. However, how these beneficial effects contribute to both structural and cognitive abnormalities is unknown and needs further evaluation. Based on the current literature, enhancing NAD+ levels may help restore brain energy metabolism and oxidative stress, which are implicated in cognitive decline. Studies in animal models have shown that NAD+ precursors such as NR and NMN can normalize neuroinflammation, improve learning and motor functions, induce mitophagy, and protect against neurological damage [
72‐
77]. In parallel, preserving the redox homeostasis in elderly animals has been shown to protect against cognitive decline, potentially through preserving NMDA receptor activation [
78]. A recent study reported that astrocytic glycolysis controls cognitive functions through synaptic NMDA receptors and suggested oral
L-serine as an accessible therapy for AD patients [
79]. Similarly, NAC therapy has been shown to restore NMDA receptor activation and reduce oxidative injury in the hippocampus of PD animal models [
80]. While removal of oxidative stress with CMA administration suggests a reasonable biological explanation for the brain hypofunction that accompanies AD-related pathologies, further mechanistic studies should be performed.
The proteomic analysis in this study revealed significant alterations of levels of several critical proteins that play an essential role in the pathogenesis of AD. For instance, levels of MertK [
81,
82], EGFR [
83,
84], oncostatin [
85‐
89], PAD4 [
90,
91], LTGF [
92‐
96], and TPO [
97], known as a potent inducer of neuro-inflammation, amyloid production and apoptosis, were significantly decreased. In contrast, proteins with neuroprotective and pro-cognitive properties, such as Klotho [
98,
99] and ST3GAL1 [
100], werer increased after CMA treatment. More interestingly, most of the analysed proteins were also shown to be significantly altered in recent human AD studies [
92‐
96,
101‐
107]. KlothoB levels were also significantly altered after CMA treatment, consistent with its neuroprotective role as a cofactor and neurotrophic factor. Recent studies have shown that KlothoB indirectly regulates glucose and energy metabolism through F2F1, which is expressed in some regions of the brain and involved in learning and memory [
106]. Moreover, gamma-aminobutyric acid (GABA) signalling has also been shown to play a critical role in mediating the detrimental effects of increased dihydroxybutyrate in the progression of mild cognitive impairment (MCI) [
108]. Interestingly, we found that the dihydroxybutyrate levels were decreased after treatment based on metabolomic analysis. Although the exact pathways involved in the metabolic generation of DHBA are still far from clear, it has been hypothesized that dihydroxybutyrate levels may be a compensatory response to increased cellular stress secondary to compromise of the Krebs cycle function, creating an alternative energy production pathway in AD [
108]. This represents indirect evidence for the energetic regulatory effect of CMA. Our findings are also in agreement with a recent study by Johnson et al., which evaluated > 2000 brains and nearly 400 cerebrospinal fluid samples by quantitative proteomics and identified mitochondrial metabolism as one of the most affected six metabolic pathways showing a strong correlation with the overall cognitive and pathological changes in AD [
109]. Hence, we observed that the significantly improved proteins related to mitochondrial bioenergetic mechanisms and the neuroinflammatory process were associated with increased cognitive scores and preserved hippocampal volumes after CMA treatment.
The metabolomics data were consistent with the expected biological outcomes of CMA treatment. Levels of plasma nicotinamide and related metabolites were increased, suggesting that NR provides sufficient substrate for mitochondrial fatty acid oxidation. In addition to its role as a cellular metabolite, NAD+ functions as an essential cofactor for the DNA repair protein poly (ADP ribose) polymerase 1 (PARP1) [
110]. Hyperactivation of PARP1 and decreased NAD+ have already been identified in the brains of patients with AD [
111,
112]. Plasma levels of serine were also increased, suggesting that CMA treatment improves the serine deficiency associated with AD. For instance, a recent study showed that the adenosine triphosphate (ATP)-reducing effect of glucose hypometabolism is restored with oral serine supplementation, suggesting the potential use of oral serine as a ready-to-use therapy for AD [
79]. The exact mechanism of action also applies to cysteine. As a GSH precursor, cysteine acts as an antioxidant and anti-inflammatory agent, maintaining the mitochondrial homeostasis and key neurotransmitter systems, such as glutamate, involved in learning and memory [
113,
114]. Accordingly, NAC has been tested for treatment of AD and shown potentials for use as an alternative medication [
115]. More importantly, fatty acid oxidation and carnitine metabolism were significantly facilitated, as shown by the robust increase in plasma levels of carnitine. These findings fit well with recent human data showing that severe disturbances in carnitine metabolism frequently occur in individuals with AD, in association with severe mitochondrial dysfunction [
116,
117]. Cristofano et al. showed a progressive decrease in carnitine serum levels in individuals shifting from normal status to AD, suggesting that decreased serum concentrations of carnitine may predispose them to AD [
118]. In support of this hypothesis, human clinical studies have demonstrated pro-cognitive effects of carnitine in MCI and AD [
119‐
121].
In addition, the levels of tryptophan metabolites, including kynurenate, kynurenine, and tryptophan betaine, decreased significantly after CMA treatment. Increased levels of these metabolites are associated with increased severity of neurodegeneration and clinical cognitive impairment through a high oxidative load, and with the formation of neurofibrillary tangles (NFTs) [
122,
123]. For instance, recent data showed a synergistic relationship between β-amyloid 1–42 and enzymatic activations of the tryptophan kynurenine pathway, resulting in increased oxidative stress, which may be associated with the formation of NFTs and senile plaque development [
124]. Also, a recent study revealed that tryptophan-2,3-dioxygenase is highly expressed in the brains of AD patients and co-localised with quinolinic acid, NFTs, and amyloid deposits in the hippocampus of post-mortem brains of AD patients [
125].
We also observed significantly increased levels of NAA, sarcosine, methionine, cysteine, and
S-adenosylmethionine and decreased levels of histidine, tryptophan quinolate, and urea cycle metabolites, which play a critical role in cognitive functions. For instance, increased NAA may provide an additional energy source for intercellular metabolite trafficking during the neurodegenerative process, especially when glucose metabolism is downregulated [
52]. Similarly, increased sarcosine levels may boost cognition, as previously shown in patients with schizophrenia, a disease in which oxidative damage and impaired glucose metabolism play key roles [
126]. In addition, decreased histidine metabolism and other decreased markers, such as homocysteine and
S-adenosylhomocysteine found in our treatment group, have been shown to slow the cognitive ageing process when appropriately downregulated [
127]. For instance, increased plasma homocysteine levels are a known risk factor for AD, whereas a low-leucine and low-arginine diet yields beneficial effects on cognition [
128].
Interestingly, CMA rapidly lowered levels of uric acid and associated metabolites. Uric acid stimulates inflammation either directly or by activating NLRP3 inflammasomes [
129]. Although the extent to which uric acid reduction contributed to the regression of cognitive impairment was unclear, the effect may be linked to improved metabolic homeostasis. For instance, a recent clinical study showed increased urea metabolism in AD patients [
130]. Accordingly, decreased levels of taurine and urea metabolites are associated with a diminished risk of dementia [
131].
To date, a few studies have identified the global changes in metabolites and metabolic pathways in AD [
25,
132,
133]. Among these, some studies highlight that alterations in lipid meabolism also play an essential role in the pathophysiology of AD [
134]. In terms of lipid metabolism, significant differences in the levels of some compounds have been observed in AD patients. Despite some discrepant trends in cross-sectional studies examining lipids in AD patients [
135,
136], the plasma levels of sphingolipids, sphingomyelins [
137,
138], acylcarnitines [
139] and phosphatidylcholines [
140‐
142] exhibited statistically lower concentrations in patients with AD, even in the preclinical stages of the disease [
27]. In addition, a significant correlation among different lipid metabolites, tau and amyloid pathology, brain atrophy and cognitive decline has been observed in an AD patient study [
27]. An autopsy study of frontal cortex metabolites showed that impaired glycerophospholipid metabolism is involved in six central metabolic pathways that are altered in AD [
143]. In brief, we observed significantly increased levels of lipid metabolites after CMA treatment, including sphingomyelin, carnitine and carnitine-related by-products, which were previously reported to decrease in patients with AD.
Despite insufficient clinical AD data concerning cholesterol metabolites and dicarboxylic acids (DCAs), we observed significantly lower levels of these metabolites after CMA treatment [
144]. Levels of pregnanediol, a metabolite of pregnenolone, and DCAs, end-products of β- or omega oxidation, which were observed as decreased in the present study, were previously reported to be lower in the urine of patients with AD [
145,
146]. Considering the neurotoxic role of bile acids, along with the oxidative properties of DCAs, the detection of decreased levels of bile acid metabolites and DCA products in the present study is therefore not surprising. Similarly, allopregnanolone has already been reported to have harmful effects on cognitive functions through GABA signalling [
147]. Also, increased bile acid levels have been reported in MCI and AD. In contrast, bile acids strongly inhibit the cysteine catabolic pathway in the preclinical period, resulting in depletion of the free cysteine pool and reduced GSH concentrations [
148].
The study has limitations. One limitation of the study was the small sample size after classifying the patients into low- and high-ADAS-Cog score groups. Therefore, a clinical trial with a larger sample size is necessary to elucidate the effects of CMA on functional and structural brain alterations. Moreover, in future studies, mitochondrial functions and changes of beta-amyloid 1–42, total-tau, and phosphorylated-tau concentrations after CMA administration should be analysed. Another limitation of the study was the lack of ApoE genotyping. Such evaluation would be informative for the detection of risk variants for AD patients.