This prospective observational study included patient populations from our “Dementias database” National Institute of Neurology IRCCS Mondino Foundation, Pavia Italy. The selection criteria were Alzheimer’s disease (AD) diagnosis, frontotemporal neurodegeneration (FTD) and cerebrovascular disease (VaD). Moreover, subjects with a diagnosis of mild cognitive impairment (MCI) were selected. Among these populations, only demented patients and MCI subjects with available plasma and cerebrospinal fluid (CSF) levels of amino acids (AAs), lactate and pyruvate, and malondialdehyde (MDA) were considered.

Summarized below are the main characteristics of the demented patients and MCI subjects and the procedures they underwent.

The various dementia subtypes were cognitively impaired at the Mini Mental State Examination (MMSE scores): AD 16.24 ± 6.43; FTD 22.25 ± 6.44; VaD 20.73 ± 6.34. In MCI subjects, MMSE scores were 24.90 ± 3.00.

Both demented patients and MCI subjects underwent diagnostic testing in the Department of General Neurology at the National Institute of Neurology IRCCS Mondino Foundation, Pavia, Italy between 2014 and 2018.

Patients with dementia had a clinical dementia rating (CDR) ≥1 (Morris, 1993) and received an etiological diagnosis of typical AD (McKhann et al., 2011), behavioral variant of frontotemporal dementia (bvFTD) (Rascovsky et al., 2011), or vascular dementia (VaD) (Román et al., 1993). All patients with non-vascular dementia had a score <4 on the Modified Hachinski Ischemic Scale (Hachinski et al., 1975). The diagnosis of MCI was made according to the following criteria: (1) cognitive concern reflecting a change in cognition reported by patient, informant or clinician (i.e., historical or observed evidence of decline over time); (2) objective evidence of cognitive decline with cognitive test scores that are typically 1 to 1.5 standard deviations below the mean for their age and education matched peers on culturally appropriate normative data, CDR 0.5; and (3) preservation of independence in functional abilities (Albert et al., 2011).

Control subjects (CTRL) were selected from patients who were hospitalized in the same Department for reasons that were unrelated to cognitive disorders or inflammatory diseases. They were 11 males and 4 females, aged 73.6 ± 6.3 years with normal body weight (body mass index, BMI-kg/m² −25.4 ± 1.7).

All procedures complied with the ethical standards of human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. The study was approved by the Local Ethical Committee of Ospedale San Raffaele-Istituto di Ricovero e Cura a Carattere Scientifico-Milano, Italy at the time of the previous study (Aquilani et al., 2020) (Project identification code: MAIR2016; ethical approval: 130/INT/2016; date: 8 September 2016).

Participants or their legal representatives provided written informed consent to participate in the study. No participant received financial compensation.

The first objective of the study was to document possible CSF reductions in AAPs between demented patients, MCI subjects, and CTRL.

The second objective of the study was to document possible reductions in CSF phenylalanine levels and to understand whether the reductions were similarly shared between demented patients and MCI subjects.

Descriptive statistics are reported as mean ± SD for continuous variables, and as numbers (N) and percentages (%) for discrete variables. Between-group comparisons for continuous variables were carried out by one-way analysis of variance (ANOVA) and by the Chi-square test for categorical variables. A significant result from ANOVA was followed up by post hoc analysis for pairwise comparisons (Dunn-Sidak method). The association between couples of variables was assessed by the Spearman’s correlation coefficient. All tests were two-tailed. A p-value < 0.05 was considered statistically significant. All statistical analyses were carried out using the SAS/STAT statistical package, release 9.4 (SAS Institute Inc., Cary, NC, USA).

Table 1 shows that experimental groups (demented and MCI) and control subjects were similar in terms of age and anthropometric variables. Biohumoral parameters and biomarker concentrations in CSF were similar across the dementia subtypes and MCI, except for p-tau, which was higher in MCI than in FTD. Both demented groups and MCI subjects had a state of insulin resistance.

The results showed that experimental groups, compared to CTRL, had altered levels of metabolic substrates, increased oxidative stress and similar AA levels. Indeed, in Table 2 lactate levels were similar between demented groups, MCI, and CTRL. In contrast, pyruvate in each dementia subtype and MCI was higher than in CTRL. Consequently, lactate/pyruvate ratios were lower in experimental groups than in CTRL.

Tyr and Trp were similar in each dementia subtype, MCI, and CTRL. For technical reasons, Met concentrations were not available. There was no difference in Phenyl- levels between experimental populations and CTRL.

Significantly higher levels of the oxidative stress marker (MDA) were found in demented patients and MCI subjects, compared to CTRL.

In the CSF compartment of demented patients and MCI subjects, compared to CTRL, altered energy substrate levels, increased marker of oxidative stress and lower AA levels were found. Indeed, in Table 3 demented patients (with the exception of AD), MCI subjects, and CTRL had similar levels of lactate in the CSF compartment. In AD, lactate was higher than in CTRL (p = 0.003). Pyruvate levels were elevated in each dementia subtype and in MCI, in comparison to CTRL. Consequently, lactate/pyruvate ratios were significantly reduced in demented groups and MCI, in comparison to CTRL.

The oxidative stress marker MDA was elevated in each dementia subtype and in MCI, in comparison to CTRL. The levels of MDA were not significantly different among the dementia subtypes and between these subtypes and MCI.

Tyrosine levels in AD (p = 0.03), FTD (p = 0.004), VaD (p = 0.001), and MCI (p = 0.002) were lower than in CTRL.

Tryptophan was also lower in demented patients and MCI subjects (from p = 0.006 to p < 0.0001), in comparison to CTRL.

No differences in Tyr and Trp were found in AD, FTD, VaD, and MCI groups.

Compared to CTRL, Met was lower in AD (p = 0.002) and FTD (p = 0.005) but not in VaD or in MCI.

Phenyl- concentrations in the experimental populations were significantly lower than in CTRL and were not different among demented subtypes and between demented subtypes and MCI.

The results showed lower CSF/P ratios of AAPs in experimental groups than in CTRL. Indeed, in Table 4, compared to CTRL, CSF/P Tyr, and CSF/P Trp were lower in both demented patients and MCI subjects. CSF/P Phenyl- was lower in demented patients and in MCI subjects (from p = 0.009 to p = 0.002).

The study found several correlations between plasma, CSF energy substrates and MDA. Indeed, plasma and CSF lactate levels were significantly associated in AD (r = 0.65, p < 0.0001) and in FTD (r = 0.57, p = 0.047), but not in CTRL, VaD or MCI. The levels of circulating pyruvate showed a direct correlation with CSF pyruvate in AD (r = 0.56, p = 0.0002) but not in CTRL, FTD, VaD, or MCI.

In summary (Table 5), in CSF, the three dementia subtypes and MCI shared similar reductions in Phenyl- and AAPs, except for Met, which was only lower in AD and FTD. Furthermore, dementia subtypes and MCI shared increased MDA and pyruvate. In contrast, increased lactate was only observed in AD. In the plasma compartment, dementia subtypes and MCI shared similar reductions in Phenyl- and similar increases in pyruvate and MDA concentrations. The lactate levels were the same in patients as they were in CTRL.

Several factors shared by AD, FTD, VaD, and MCI may explain the overutilization of amino acids in the brain, leading to lower AAPs in the CSF compartment. These factors include insulin resistance, the biosynthesis of neuropathological markers and the mutual influence of these two elements on each other (Aquilani et al., 2022).

Insulin resistance (Apelt et al., 1999; Bigl et al., 2003; Mazzola and Sirover, 2003; Freude et al., 2009; Ahmed et al., 2014, 2016; Cooper et al., 2015; Hong et al., 2021; Aquilani et al., 2022) reduces glucose utilization in mitochondria by blocking the mitochondrial pyruvate dehydrogenase enzyme complex (Lazar, 2002) thus leading to impaired energy generation. Given the brain’s inability to process fatty acids for energy production (Griffin and Bradshaw, 2017), an excess of AAs is used in the tricarboxylic acid cycle (TCA) as an energy substrate (Burns et al., 2010; Griffin and Bradshaw, 2017). The alterations of anaerobic glucose breakdown in AD patients were likely more pronounced than in the other dementia subtypes and in MCI, thus accounting for high CSF lactate. Translocated intestinal bacteria, bacterial amyloid, and toxins may increase brain inflammation, favoring the development of insulin resistance (Bhattacharjee and Lukiw, 2013; Friedland, 2015; Zhao et al., 2015; Aquilani et al., 2022).

Another mechanism increasing AAP utilization is promoted by the Aβ that can activate mTOR signaling (Li et al., 2005; Caccamo et al., 2011), which stimulates the synthesis of tau protein, the main component of the neurofibrillary tangles (Jahrling and Laberge, 2015), and mitochondrial activity (Schieke et al., 2006).

The interrelationship between insulin resistance and neuropathological markers may further explain AAP overutilization. Hyperinsulinemia increases the phosphorylation of tau proteins (Clodfelder-Miller et al., 2006; Planel et al., 2007), the formation of senile plaques (De la Monte and Wands, 2005), and inhibits the degradation of extraneuronal Aβ (Qiu et al., 1998). In turn, both Aβ and tau further impair brain insulin signaling (Biundo et al., 2018) (vicious cycle). Brain amino acid overutilization in FTD may follow increased synthesis of protein aggregates (Raz et al., 2016), including cellular p-tau (Lee et al., 2001) and FTD-transactive response DNA-binding protein-43 (TDP-43) (Davidson et al., 2007; Hu and Grossman, 2009).

In VaD brain tissue, AA overutilization may be due to increased amyloid synthesis and deposition (Bell et al., 2010). Amyloid formation is due to prolonged vasculotoxic and neurotoxic effects of vascular damage-induced hypoxia and blood-brain barrier permeability (Wardlaw et al., 2013; Iadecola, 2017; Sweeney et al., 2018; Nation et al., 2019). On the other hand, cerebral vessel disease contributes to approximately half of all dementias (Iadecola, 2017). In turn, Aβ42 deposition (Iadecola, 2017; Kisler et al., 2017; Montagne et al., 2017) and tau protein (Bennett et al., 2018) cause vascular damage. Vessel damage-induced ischemia leads to reduced oxygen supply and further reduction of monoamine neurotransmitters. Indeed, the syntheses of biogenic amines are very sensitive to a reduced oxygen supply, as tyrosine dehydrolase and tryptophan hydroxylase activities need molecular oxygen for catalyzing the syntheses of monoamine neurotransmitters (Frazer and Hensler, 1999; Kuhar et al., 1999). Previous studies have documented that vascular damage contributes to cognitive decline (Zlokovic, 2011; Snyder et al., 2015; Gottesman et al., 2017; Iadecola, 2017; Sweeney et al., 2018).

It is of clinical interest to note that individuals with early cognitive impairment may have brain capillary damage and BBB disruption in the hippocampus, independently of neuropathologic marker deposition (Nation et al., 2019).

Brain amino acid overutilization from increased protein synthesis may also occur in subjects with MCI following the upregulation of brain genes associated with anabolic processes and mitochondrial energy generation in synaptic genes (Berchtold et al., 2014).

In the current study, the reduction of CSF Met, a precursor of choline, which is the essential substrate for ACh synthesis, was only found in AD and FTD cohorts. This suggests that VaD patients and MCI subjects have a better conservation of the cholinergic system, although MCI subjects had a reduced forebrain, the pacemaker of cholinergic transmission (Peter et al., 2021). We speculate that the choline formed from Met could ensure the formation of ACh, thus limiting the negative impact of basal forebrain atrophy. Maintaining ACh generation as far as possible is important because ACh released under the activity of cholinesterase supplies choline to neurons (Taylor and Brown, 1999).

It is important to consider that another source of brain choline is the phosphatidylcholine catabolism (Taylor and Brown, 1999).

In the current study, the high levels of pyruvate both in plasma and CSF, together with the finding of normal plasma lactate concentrations, low lactate/pyruvate ratios in FTD, VaD, and MCI suggest that mitochondrial dysfunction is more prevalent than dysfunction in the glycolytic pathway in the brain and extracerebral districts. The current investigation confirms a previous study reporting high CSF levels of pyruvate in AD and VaD (Parnetti et al., 2000). The derangements of both cell mitochondrial structure and enzymes leading to impaired energy production (ATP) have been reviewed in several studies (Mosconi et al., 2008; Ferreira et al., 2010; Verri et al., 2012). The high CSF and plasma pyruvate levels and low lactate/pyruvate ratios in experimental groups indicate that (1) the deficit of energy production (hypometabolism) (Liguori et al., 2015, 2016) is due to combined defects of anaerobic and aerobic metabolism energy-producing in brain and extra-brain organs/tissues; and (2) the alterations of the aerobic pathway of energy generation are more pronounced than the alterations of the anaerobic metabolism.

The high lactate in AD patients is in line with previous investigations reporting elevated CSF lactate (Liguori et al., 2015, 2016).

The high MDA levels found in all the experimental groups analyzed in the current study suggest that abnormal mitochondrial activity is responsible for the increased generation of free radicals and the development of a state of oxidative stress leading to cell membrane lipolysis. The lipid hyper-oxidation can have a devastating impact on brain and extra brain cell integrity due to an excess of free radicals that results in cell lysis (Sharp et al., 2004; Huerta-Alardín et al., 2005), contributing to brain and extra brain tissue atrophy.

High plasma and CSF pyruvate indicates an inefficient aerobic pathway of energy generation. Reduced pyruvate utilization in TCA is likely to be an effect of cell insulin resistance. Another important negative effect of low pyruvate utilization in the TCA cycle is the reduction of choline formation and consequently ACh synthesis (Taylor and Brown, 1999). Both energy deficit and oxidative stress lead to brain (and extra brain) cell destruction.

Given that the levels of lactate in the plasma and CSF compartments correlate in AD and FTD while those of pyruvate in plasma and CSF only correlate in AD, we postulate that plasma supplies these substrates to AD and FTD brains. The supply of lactate is not only extremely useful during exercise (Ide et al., 1999, 2000), in which the brain increases lactate uptake, but also maintains cerebral blood flow regulation as far as possible (Hein et al., 2006; Yamanishi et al., 2006; Gordon et al., 2008).

The results of this study suggest that metabolic alterations may potentially lead to the worsening of brain structural and functional changes and to the progression of cerebral damage (Figure 1).

Low AAPs indicate that the lack of neurotransmitter formation, the first step of neurochemical transmission processes, may contribute to deficits in synaptic transmission. The reductions of the essential Trp, Phenyl-, and Met can impair the structure and function of the neuronal network (Shepherd and Huganir, 2007; Bramham, 2008) by limiting the rate of protein synthesis and activation, in particular for the remodeling of synapse plasticity.

Altered synapse proteins could be responsible for alterations of the synaptic response to subsequent neurotransmitter release (Todorova and Blokland, 2017), reduced dendritic and spine growth, and synaptogenesis (Wong and Ghosh, 2002; Mattson et al., 2008).

Brain mitochondria dysfunction and reduced energy production (ATP), as inferred by pyruvate accumulation, reduces synaptic function and plasticity that normally depend on mitochondria (Cheng et al., 2010; Amaral and Pozzo-Miller, 2012; Ivannikov et al., 2013; Manczak et al., 2013; Brot et al., 2014; Su et al., 2014).

Moreover, deficit in ATP generation does not support (or only to a very small degree) synaptic vesicle recruitment and release, protein phosphorylation processes (Stefani et al., 1997; Attwell and Laughlin, 2001; Verstreken et al., 2005; Valenti et al., 2014), neurotransmitter release via vesicle exocytosis and mobilization, or dendritic remodeling (Cheng et al., 2010; Valenti et al., 2014).

Inadequate mitochondrial ATP production impairs long-term-potentiation (LTP) (Cunha et al., 1996), a well-documented experimental model of learning and memory (Morris, 2013).

Brain ATP deficit reduces the development of neuronal pre- and post-synaptic compartments, the generation of membrane potential, synaptic vesicle recruitment and release, and protein phosphorylation activities (Stefani et al., 1997; Attwell and Laughlin, 2001; Verstreken et al., 2005; Valenti et al., 2014).

As mitochondrial and synaptic plasticity influence each other (Figure 2; Amaral and Pozzo-Miller, 2012), deficits in synaptic plasticity lead to reduced mitochondrial biogenesis. Indeed, some signaling pathways regulating neuroplasticity, such as estrogen, growth factors, and nitric oxide (Foy et al., 2008; Brinton, 2009; Harooni et al., 2009; Minichiello, 2009), change mitochondrial formation (Williams et al., 1998; Gutsaeva et al., 2008; Klinge, 2008; Renton et al., 2010). Impaired synaptic activation and LTP formation (Mattson and Liu, 2003) can reduce mitochondrial energy production, Ca²⁺ pump activity and gene expression.

Mitochondria damage-increased ROS and high oxidative stress expose both demented patients and MCI subjects to DNA mutations (Mattson and Liu, 2003) and mitochondrial damage (vicious circle). Oxidative stress is associated with synaptic impairments, in particular in aging (Brawek et al., 2010). This contributes to explaining the cognitive impairment in the MCI subjects in the current study.

There is a need for future investigations to overcome the limitations of this study.

Other common neurodegenerative dementias such as Parkinson’s disease dementia and dementia with Lewy bodies were not considered in this study.

The results of the study should be confirmed in a larger population of FTD, VaD patients, and MCI subjects. We believe this is especially important for CSF lactate levels. Indeed, in the current investigation, we also found high CSF lactate levels in FTD, VaD patients and MCI subjects although they were not statistically significant, probably because of the small sample sizes.

Another limit of the study is that the brain magnetic resonance imaging was performed, but due to differences in protocols could not be used to draw any correlation and analysis.

In the current study we did not analyze and discuss other signaling molecules including glutamate, which plays a critical role in synaptic plasticity and energy cellular metabolism (Mattson et al., 1995; Trenkner et al., 1996) by increasing mitochondrial oxygen consumption and thereby ATP production (Schuchmann et al., 2005; Mattson, 2007). In our opinion, possible differences in CSF glutamate among the dementia subtypes and between the dementia subtypes and MCI may be particularly important.

The results of the current and previous studies (Aquilani et al., 2020, 2022) may raise the question of whether oral supplementation with a specific cluster of AAs including adequate amounts of Tyr, Trp, Phenyl-, and Met could improve patients’ levels in the CSF compartment while reducing CSF pyruvate accumulation in patients with AD, FTD, and VaD, and subjects with MCI. Moreover, it may be important to document whether the addition of antioxidant agents to the AA cluster could reduce brain oxidative stress.