Metabolic Endotoxemia: A Potential Underlying Mechanism of the Relationship between Dietary Fat Intake and Risk for Cognitive Impairments in Humans?

(1) Background: Nutrition is a major lifestyle factor that can prevent the risk of cognitive impairment and dementia. Diet-induced metabolic endotoxemia has been proposed as a major root cause of inflammation and these pathways emerge as detrimental factors of healthy ageing. The aim of this paper was to update research focusing on the relationship between a fat-rich diet and endotoxemia, and to discuss the potential role of endotoxemia in cognitive performances. (2) Methods: We conducted a non-systematic literature review based on the PubMed database related to fat-rich meals, metabolic endotoxemia and cognitive disorders including dementia in humans. A total of 40 articles out of 942 in the first screening met the inclusion criteria. (3) Results: Evidence suggested that a fat-rich diet, depending on its quality, quantity and concomitant healthy food components, could influence metabolic endotoxemia. Since only heterogeneous cross-sectional studies are available, it remains unclear to what extent endotoxemia could be associated or not with cognitive disorders and dementia. (4) Conclusions: A fat-rich diet has the capability to provide significant increases in circulating endotoxins, which highlights nutritional strategies as a promising area for future research on inflammatory-associated diseases. The role of endotoxemia in cognitive disorders and dementia remains unclear and deserves further investigation.


Introduction
As a consequence of the ageing population, the prevalence of dementia, characterized by a progressive deterioration of cognitive performances in multiple domains (i.e., memory, reasoning, judgement and ability to perform daily activities) that evolves into a pathological diagnosis, is increasing [1]. The aetiology of dementia is multi-factorial and consists of a dynamic interaction between genetic susceptibility, non-modifiable factors (i.e., age and sex), pathological processes and environmental factors, some of them being potentially preventable [2,3].
In this context, nutrition-a major lifelong environmental factor-is of growing interest and offers an interesting strategy for the prevention of cognitive decline and subsequent dementia [4]. Numerous studies have suggested a benefit of higher adherence to healthy dietary patterns, with a beneficial balance in favour of unsaturated fatty acids to the detriment of saturated fatty acids such as

Associations between Long-Term Dietary Interventions and Circulating LPS
In an intervention study, the increase in energy intake (+70 g fat for eight weeks) was associated with an acute rise in endotoxin levels in the postprandial state, but not in the fasting state [46] (Table 2). More recently, three weeks of an intervention based on a low-fat high-carbohydrate diet enriched in n-3 PUFA has been shown to increase the postprandial levels of LPS, but decrease the fasting levels of LPS, compared with a Mediterranean diet enriched in MUFA or a SFA-rich diet, among healthy older subjects [47]. Among 75 metabolically impaired subjects, adherence to a high-fat high-saturated-fatty acid diet for 12 weeks led to an increase in the postprandial levels of LPS, but not the fasting levels [48].
In a crossover study among eight healthy older adults (55 y and over), Pendyala et al. [49] demonstrated that following a Western-type diet for four weeks significantly increased fasting plasma levels of LPS by 71%. In contrast, a 38% decrease in fasting plasma LPS levels was observed after four weeks of adherence to a prudent-type diet with equivalent energy intake to the Western-type diet. Regardless of the source of energy, Breusing et al. [50] observed that the 31% increase in fasting endotoxemia after one week of overfeeding (+50% of the energy requirement) was reversed by three weeks of caloric restriction (−50% of the energy requirement) among 15 healthy adults. (Table 2) Although of major interest, only very few studies have investigated the relationship between usual diet and circulating levels of LPS, all using a cross-sectional design analysis and exhibiting inconsistent results. In a subsample of 201 healthy men, fat and energy, but not carbohydrate or protein, intake was positively associated with fasting levels of LPS [51]. More recently, Kallio et al. [52] reported an association between energy intake and the levels of LPS observed only among lean subjects. Surprisingly, no association between fat intake and the fasting levels of LPS was observed in this study. On the other hand, two studies did not find an association between nutrient intake and fasting LPS levels among overweight and obese pregnant women [53] or type 1 diabetes patients [54]. In this latter study [54], higher consumption of fish and healthy snacks (including fruits and berries, fresh vegetables, soft drinks, yoghurt and low-fat cheese) and higher adherence to a modern diet (composed of fresh vegetables, pasta and rice, poultry, meat dishes and fried or grilled foods) were all significantly associated with lower fasting LPS levels in serum [54]. Finally, among elderly patients with nonvalvular atrial fibrillation, higher adherence to a Mediterranean-type diet was inversely correlated with fasting circulating LPS levels [55]. Interestingly, among Mediterranean diet food components, higher intake of fruits and legumes showed a major association with lower levels of LPS.

Associations between Lifestyle Dietary Patterns and Circulating LPS
Overall, these results underlined that a transient rise in circulating levels of LPS can be induced by a large variety of high-fat diets, especially those devoid of healthy food components. Moreover, fasting or postprandial assessment of LPS could also partly explain the discrepancies observed in several studies. Considering that exposure to an overload of endotoxins could contribute to the development of adverse health outcomes such as elevated systemic inflammation, neuroinflammation, neurodegeneration and neural death in experimental studies [15,56], we hypothesized that lifelong exposure to endotoxins inherent to adherence to unhealthy diets and to ageing could worsen the detrimental outcomes of endotoxemia. The second part of this state-of-the-art paper describes the few studies focusing on the relationship between endotoxemia and cognitive disorders and dementia in humans.

LPS Injection and Short-Term Cognitive Function Assessment in Interventional Studies
In the field of LPS-related inflammation in association with dementia or cognitive performances, few studies have been identified following our research strategy.
First, all studies were interested in the impact of LPS injection on the inflammatory response of the host. Transient increases in pro-inflammatory cytokines (for instance IL-6 and TNFα, on average between 1 and 6 h post-infection) followed by the release of anti-inflammatory cytokines (for instance IL1-Ra and IL-10, on average between 3 and 8 h post-infection), were observed after LPS injection in all intervention studies. This result was consistent regardless of the LPS dose injected. Second, several lines of evidence suggested that inflammatory cytokines such as IL-6 or TNFα could be involved in cognitive disturbances [26,27]. However, it remains unknown whether and to what extent cognitive functions could be affected during transient immune activation induced by a single injection of LPS, as provided by the selected intervention studies described below (Table 3).
In a crossover study, Reichenberg et al. [57] reported a significant impairment in declarative memory until 10 h after the injection of LPS. With this intervention, decreased performances in declarative memory have also been observed in a subsample of subjects, as well as improvements in working memory compared to the injection of placebo [58]. These last controversial results are, however, not generalizable to the whole study sample [57], which raises questions. Finally, in both studies [57,58], there were no statistical or clinical differences regarding attention or executive functions following the injection of LPS compared to placebo.
In another study, using an injection of a lower dose of LPS (0.2 ng/kg body weight) a negative correlation was found between the increased IL-6 levels and memory and learning performance after 4.5-6 h [59]. Among all results, the injection of LPS did not alter working or executive functions or attention among young healthy volunteers. With higher doses of LPS (2 ng/kg body weight), the results are also controversial. This treatment did not induce any alteration of working memory, psychomotor speed capacity and information processing ability, fine control motor and attention performances until 10 h post-injection compared with the placebo group [60]. Surprisingly, the authors observed an increase in attention performance in the treated group compared to the placebo group.
In 2010, Grigoleit et al. [61] found that the injection of LPS did not affect the subscales of the Wechsler Memory Scale, analysing performances in verbal, visual or delayed memory, as well as attention and executive control processes. Using a double-blind crossover study, the same authors [62] observed that LPS injection did not affect accuracy in working memory performance, but improved reaction time in the high-dose (0.8 ng/kg body weight) 2 h post-injection compared to placebo; a result that was not observed with the lowest LPS dose (0.4 ng/kg body weight).
Alteration in emotional/social processing was observed following the injection of LPS at 0.8 ng/kg body weight [63], but not with a lower dose of LPS (0.4 ng/kg body weight) [64].
Altogether, these results underlined that the LPS dose, the delay and the targeted samples have different responses on the cognitive performances assessed (themselves being heterogeneous between studies), which deserves further research with emphasis on both experimental conditions and outcomes.

Endotoxemia and HIV-Associated Neurocognitive Disorders in Observational Studies
In the field of LPS and cognition, we identified 2 cross-sectional observational studies on non-healthy individuals, focusing on HIV-infected participants with or without HIV-associated neurocognitive disorders (HAND) [65,66] (Table 3). The median fasting levels of LPS were higher among the HAND group than among the no-HAND participants (116.1 ρg/mL vs. 98.2 ρg/mL) [65]. Interestingly, circulating levels of LPS were not associated with the severity of HAND. In the other study, plasma LPS levels did not differ according to the score of a neurocognitive test battery designed to assess several domains of cognitive function (i.e., motor skills, speed of information processing, attention, learning, memory, language fluency and executive function) [66]. More recently, Jespersen et al. [67] reported a lack of association between fasting levels of LPS and markers of axonal damage or monocyte activation in the central nervous system among HIV-infected adults without evidence of impaired cognitive function. In this specific sample, LPS was not detectable in the cerebrospinal fluid.
Although not tested in a human model of endotoxemia, repeated lifelong exposure to endotoxin may lead to a long-term alteration in all cognitive domains. Dementia diagnosis is the result of a long insidious process where cognitive disorders evolve to pathology. However, the relationship between endotoxemia and dementia has been explored in very few studies, which are described below (Table 4).

Endotoxemia and Alzheimer's Disease or Dementia Diagnosis in Observational Studies
First, we identified two studies interested in the anatomy of the brain in post-mortem patients. Recent investigations reported higher LPS abundance in grey matter (superior temporal gyrus lobe) and white matter (frontal lobe) in brains from patients with Alzheimer's disease (AD) than in those from participants free of dementia [68]. Of greater interest, LPS colocalized with Aβ1-40/42 in amyloid plaques and with Aβ1-40/42 around blood in AD brains [68]. Another study reported an average three-fold higher abundance of in the hippocampus-an anatomical region of the AD brain that develops the earliest and most profound neuropathology-from four AD brains compared to two age-matched control brains, as well as a two-fold higher abundance in neocortical extracts from six AD brains and six age-matched control brains [69]. In some advanced AD patients (criteria not defined in the study), hippocampal brain lysate exhibited up to a 26-fold increase in LPS [69].
In 2008, Ancuta et al. [70] observed that plasma LPS levels were significantly higher among dementia-associated HIV participants than among participants without neurocognitive impairment, independent of plasma viral load and CD4 counts. Surprisingly, LPS levels in participants with minor cognitive and motor disorders, asymptomatic neurocognitive impairments or neuropsychiatric impairments did not differ from those in participants without neurocognitive impairment. Finally, the last study we identified reported that plasma LPS levels were three-fold higher in 18 AD participants (mean 61 ρg/mL) than in 18 healthy controls (mean 21 ρg/mL) [71].
Altogether, the scarce available literature in this field suggests that higher LPS levels are observed (i) among the brains of AD patients, in several regions of interest and (ii) among living AD participants, and that these results could be partially explained by higher immunosenescence also among HIV patients. These results allow us to speculate that not only transient immune activation induced by LPS, as described earlier, but also increased fasting levels of LPS (from different sources) could be partly involved in the pathogenesis of dementia. Longitudinal studies are still required to test this hypothesis.

Discussion
The hypothesis tested in this "state-of-the-art" paper was that metabolic endotoxemia could be an underlying mechanism of the relationship between nutrition (and mainly fats intake) and age-related cognitive impairments.
First, to the best of our knowledge, there was no study focusing on this whole association (i.e., nutritional habits that could modulate endotoxemia, which itself could be part of the pathological processes leading to dementia) in a single sample setting. Therefore, we conducted two approaches, based on the available literature: we synthesized studies focusing (i) on the role of nutritional habits and interventions on the modulation of metabolic endotoxemia, and (ii) on the association between endotoxemia and cognitive impairments.
To some extent, overall metabolic endotoxemia may be attributed to a balance between the number of LPS-containing Gram-negative bacteria in the gut microbiota and the subsequent translocation of the LPS across the gastro-intestinal barrier into the bloodstream; the fat content of the diet being a putative key actor of this translocation.
Our synthesis provided convincing evidence that fat-rich meals have an undoubted capacity to transiently modulate postprandial metabolic endotoxemia in humans. Indeed, recent studies have demonstrated that fat absorption and digestion is a step where dissociated LPS can be incorporated into chylomicrons (i.e., lipoproteins responsible for the transport of lipids through the gut barrier) thereby enabling their translocation into the bloodstream [21,22]. We highlighted that not only the quantity but also the quality of dietary fats may influence metabolic endotoxemia. Precisely, imbalanced diets in favour of saturated fatty acids have been associated with higher postprandial levels of LPS while a combination of these diets with other healthy nutritional components, such as fibre [41], is able to limit this increase.
On the other hand, long-term consumption of a high-energy-density diet, especially those derived from fat, has been associated with gut microbiota dysbiosis [72]. By shifting the balance in favour of, or at the expense of, LPS-containing Gram-negative bacteria, diet could also contribute to the amount of LPS in the gut and their translocation into the bloodstream. Metabolic diseases such as obesity and T2DM are often associated with gut microbiota dysbiosis [73,74], most likely worsened by lifelong consumption of an unhealthy fat-rich diet. This latter result could explain, at least in part, the results reporting increased fasting levels of LPS and an exacerbated postprandial increase of LPS following a high-fat-saturated meal in metabolically impaired individuals compared to healthy individuals.
Moreover, gut dysbiosis, and the subsequent release of LPS can be managed by several dietary factors, that are not discussed in the present paper, including, for instance, the intake of refined sugars, alcohol or nutritional supplements with pre-and probiotics for details see, [19]. In particular, high consumption of glucose or fructose, which is part of the Western diet, could induce an increase in circulating levels of LPS in mice [75]. Pronounced intestinal permeability and increased plasma levels of LPS were found in patients with chronic alcohol abuse [76]. Pre-and probiotics have demonstrated the ability to manipulate gut microbiota and to influence the circulating levels of LPS [19]. However, we did not intend to review all potential and nutritional factors that might increase levels of LPS and prefer to limit our hypothesis to lipids, which is a well-known risk factor for AD [77].
As a limitation of the selected studies, the discrepancies observed between postprandial and fasting levels of circulating LPS after dietary interventions are questionable. These observations suggest that metabolic endotoxemia has a fluctuating nature in humans, and that fasting levels of LPS may therefore not be an accurate marker of chronic exposure to endotoxins over time. Some authors have also suggested the measurement of LBP, which is considered a longer-term marker of endotoxin-related exposure than LPS [78,79].
The main consequence of chronic exposure to endotoxemia is the onset and maintenance of a low-grade inflammation state, with associated deleterious outcomes in elderly individuals [26,27]. However, it remains unclear to what extent acute exposure to LPS could induce age-related cognitive disorders. Indeed, our literature research showed no convincing evidence that exposure to an intravenous overload of endotoxins was associated with major cognitive alterations in healthy individuals; a result that is in line with a previous review [80]. As already mentioned, the large heterogeneity of interventional studies reported in this paper (i.e., in terms of the injected doses of LPS, the neurocognitive tests used to assess short-term cognitive performances or the various delays between injection of LPS and cognitive assessment, for instance) limits us in drawing definitive conclusions regarding the association between endotoxemia and cognitive performances. Additionally, speculative, controversial results of LPS injection on some selected cognitive tests also ask for questions on possible compensatory mechanisms, which deserves further research.
Regarding the association between endotoxemia and dementia, the results seem to be more consistent, while they are based on a very small number of studies. Dementia results in a long insidious process accompanied by molecular and physiological changes, including oxidative stress, impairment in neuronal function and the death of neuronal cells, which may be caused or worsened by neuroinflammation [26,81]. We thought that accumulated lifelong exposure to endotoxin may therefore be associated with more severe stages of cognitive disorders and dementia. Due to the lack of prospective studies to support this hypothesis, we cannot exclude a possible reverse causation. For instance, individuals with advanced stages of dementia are more likely to develop bacterial infections (i.e., in part due to a lack of hygiene inherent to the deterioration of cognitive performances and disability status) and therefore to be exposed to higher endotoxemia. In addition, and inherent to ageing, increased permeability in physiological barriers (i.e., intestinal and blood-brain barriers) is observed in AD individuals [82,83], which could also promote the translocation of higher amounts of neurotoxic molecules such as LPS.
Finally, the LPS doses used in most interventional studies may also be discussed, in addition to those of natural exposure pathways which may gradually and intermittently deliver smaller amounts of endotoxin over time.
Among the limitations of this state-of-the-art paper, we acknowledge the lack of adopting a strict systematic literature review methodology. However, only partial studies have been identified to respond to our whole hypothesis which confirmed our innovative approach. As a limitation, we cannot exclude a publication bias; studies that are not statistically significant have been available more often than those with significant results. The strengths of the present analysis are therefore to update, by a holistic approach, in a single article the experimental and observational literature in the field of fat-rich nutrition, endotoxemia and cognition in humans and to identify some gaps to be completed in the near future.

Conclusions
Nutrition has been proposed as a promising non-medical strategy to prevent cognitive decline and subsequent dementia. Affected by the quantity and the quality of ingested fats, metabolic endotoxemia, involving a potent pro-inflammatory response of the host, could be one of the underlying mechanisms. As the postprandial state represents a stressful condition in which our current society spends most of its time, the identification of an individual-adapted dietary pattern associated with lower metabolic endotoxemia and subsequent inflammation is a promising area for future research focusing on inflammatory-associated diseases. However, there is an important need for research to understand to what extent transient but also chronic low-exposure to LPS, through repeated measurements of postprandial and fasting levels of LPS over time, could be associated with long-term cognitive changes until the diagnosis of dementia.
Funding: This research received no external funding.