Abstract
Background
According to several published studies, a hypercaloric diet (HD) could be considered a risk factor for neurodegenerative diseases, including Alzheimer's disease (AD). To fully understand the molecular pathways involved, HD has been investigated in several animal models.
Objective
The aim of this systematic review (SR) is to provide an overview of recent published data on the effects of HD on animal models of AD to gain insight into the molecular mechanisms potentially involved and to highlight current knowledge gaps for future studies.
Methods
Structured bibliographic searches were carried out on PubMed, the Cochrane Library, and ISI Web of Knowledge. The SR was conducted following the Cochrane Handbook and the PRISMA statement. Studies enrolling only wild-type models or not using standard diet (SD) as control were excluded, as were non-original publications. Included studies were assessed for risk of bias using SYRCLE's tool.
Results
A total of 77 studies met inclusion criteria. Most reported significant behavioral differences in HD-exposed mice (Morris Water Maze, Open Field, Y-maze), though with considerable variability due to protocol heterogeneity. A significant increase in tau and amyloid deposition was observed after HD exposure, and most studies reported negatively affected learning and memory. However, nearly half found no significant differences between HD and SD groups, likely reflecting heterogeneity in diet duration and type, animal age, and strain susceptibility. Methodological quality varied widely, with many studies omitting randomization, blinding, or sex-stratified analyses.
Conclusions
Despite variability, evidence suggests HD worsens behavioral performance and increases tau and amyloid expression in mouse brain, representing a risk factor for dementia. More rigorous, standardized, and sex-balanced preclinical studies are needed, and findings support dietary interventions as early non-pharmacological strategies in AD prevention.
This is a visual representation of the abstract.
Introduction
Over the years, the frequency of neurodegenerative diseases has widely increased. 1 Almost 50 million people worldwide have a diagnosis of dementia, with Alzheimer's disease (AD) being one of the most prevalent, involving approximately 17% of the population aged 75 and older.1,2 In its early stages, AD is characterized by the accumulation of neurofibrillary tangles, amyloid plaques, and dystrophic neurites. Neurofibrillary tangles and amyloid plaques can lead to neuronal degeneration and cell death, resulting in memory loss, cognitive deterioration, and language impairments. 3 Several risk factors have been reported to contribute to the development of AD, including age, genetic factors, head trauma, vascular diseases, infections, and environmental factors.4,5 The complexity of these interactions led to consider AD as a multifactorial condition. Interestigly, in people aged 45 and older additional potentially modifiable risk factors become increasingly significant. These include lack of cognitive activity, depression, stress, diabetes, hypertension, obesity, physical inactivity, and cardiovascular diseases.6,7 Nutrition has been reported as a potential non-pharmacological strategy for the prevention of AD. Adherence to a “Mediterranean diet” has been reported as associated to a lower incidence of AD,8,9 while a high intake of refined carbohydrates, particularly sugars, has been linked to a higher risk of obesity, diabetes and AD.10–12 This can induce metabolic alterations that could impair neuronal function and ultimately trigger neuronal apoptosis, leading to cognitive deterioration. 13 Conversely, other studies reported that HD could improve cognitive function by improving the functioning of the blood-brain barrier and reducing brain atrophy in AD models. However, these diets did not seem to be able to reduce amyloid-β (Aβ) levels.14,15 Therefore, the impact of a HD on the progression of AD remains a topic of debate within the scientific community. Several reviews focusing on diet and neurodegenerative disease have been conducted in recent years. Many of these have concentrated on the supplementation of common dietary components such as ketogenic supplements to evaluate their effects on animal models or non-transgenic mice.16,17 To the best of our knowledge, there are currently no systematic reviews investigating the effects of a hypercaloric diet on murine models of AD. Therefore, we sought to address this gap by conducting a systematic review and, where feasible, a meta-analysis of the available preclinical evidence. The aim was to evaluate the effects of HD on cognitive decline, amyloid pathology, and neurodegeneration in animal models of dementia to allow clarify the molecular mechanisms underlying the relationship between diet and dementia, thus highlighting its potential role in the prevention of AD. At the same time, we aimed to assess whether the conditions required to perform a meta-analysis were met and, if not, to critically examine the main methodological and reporting limitations hindering quantitative synthesis.
Methods
Search strategy and selection criteria
The present SR was carried out following the methodology published in the Cochrane Handbook for systematic reviews 18 (Available from www.training.cochrane.org/handbook and reported following the PRISMA statement for reporting systematic reviews and meta-analyses 19 . Structured literature searches were performed in the PubMed, Embase, and Web of Science databases using the following search terms ((diet* OR HFSSD) AND (dementia* OR Alzheimer*) AND (mouse OR mice OR rat OR rats OR transgenic*) AND (“cognitive impairment” OR “behavioural impairment” OR “behaviour impairment” OR “behavior impairment” OR “behavioral impairment” OR “oxidative stress” OR neuroinflammat* OR “mitochondrial dysfunction” OR amyloid* OR “tau hyperphosphorylation” OR gliosis OR “neuronal loss” OR “neuron loss” OR lipid peroxidation)). All literature available up to February 2026 was included, with no restrictions applied regarding publication date, study design, or language. A comprehensive list of all relevant published studies with available results was obtained from all databases, and duplicates were removed. Studies were initially selected based on their relevance and pertinence to the topic of the review.
Study selection
After removing duplicates, the records obtained from the bibliographic searches were selected based on titles and abstracts by using online tool Rayyan (https://www.rayyan.ai/). The full texts of selected studies were retrieved and applied the predefined eligibility criteria. Bibliographic references of all included studies were also analyzed to identify further potentially relevant publications. The following eligibility criteria were applied: all studies reporting data on the effect of HD on transgenic mouse models of AD, and studies comparing HD to a control diet were included. Inclusion criteria for HD were strictly defined to ensure consistency across the selected studies. To be included, a dietary intervention had to include a fat content or a combination of high fat and cholesterol like a high fat diet (HFD) or western diet (WD) or a high cholesterol diet (HCD), supplementation with high concentrations of purified carbohydrates, such as sucrose or fructose like high sugar diet (HSD). Diets focused solely on single micronutrient supplementation without an increase in total caloric intake were excluded. Studies enrolling only wild type models or adopting other control conditions (e.g., low-fat diets used as control instead of standard diet, etc.) were excluded. Letters, commentaries, preprint and grey literature, editorials, conference proceedings, case reports or case series, and non-systematic or narrative reviews were also excluded. Preprints were excluded because they have not yet undergone reviewer assessment and comments, which are essential to define the validity and accuracy required for publication. Exclusion of grey literature was applied due to a potential bias related to the possible lack of replicability. Disagreements in the decision to include studies, when present, were resolved by consensus or by involving an independent reviewer. Systematic reviews, when available, were considered separately to check for references and consistency of results. Literature selection, data extraction, and qualitative assessment were performed by two independent reviewers.
Data extraction and quality assessment
Results were summarized in a narrative form. The qualitative assessment of included studies was carried out using the Syrcle's risk of bias tool specific for animal research. 20 Based on this tool, the risk of bias for each study was classified as “high”, “moderate” or “unclear”. 20 This tool contains 10 items addressing 6 domains of potential bias: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other sources of biases.
The qualitative data were reported as summary of scores for each domain for each included study, as described in previous articles.21,22
Results
Summary of included studies
Bibliographic searches yielded a total of 5552 records, including 1955 records from PubMed, 2005 from Embase, and 1592 from Web of Science. To facilitate the screening and classification of articles after duplicate removal, the lists of records were imported on Rayyan (https://www.rayyan.ai/), a web-based tool that had been used by the reviewers for a previously published review.23,24 Based on this preliminary screening, 4972 articles were excluded, while 580 articles were retrieved in full-text and were assessed for eligibility based on the predefined eligibility criteria. Of the studies analyzed, 503 were further excluded as they were conference posters, abstracts or reviews. A total of 77 studies met the inclusion criteria. The study selection process is reported in Figure 1. The characteristics of included studies are reported in Table 1. Data was extracted using a standardized form. Collected data included year of publication, characteristics of the animal model, type of diet, and results for all considered outcomes. Supplementary materials were reviewed and included when available. In Table 2, we provide a sum up of the main characteristics and results of the outcomes of the studies with available data. All studies used mice as animal models, except for four that used rats.25–28 36 studies used transgenic animal models carrying the APP/PS1 double mutation12,15,26,27,29–60 and the most represented model was APPswe/ PS1ΔE9, though four studies did not describe the type of mutations on the APP and PS1 genes without specific mutation.33,34,46,61 In 16 studies, animal model had only one mutation in APP gene, and the most used was APP Swedish mutation.43,62–76 Moreover, a total of 13 studies used the 3xTg all expressing mutant PS1M146 V, APPswe, and TauP301L.68,77–88 Finally, three studies used two experimental models with Aβ injection.28,78,89 While one 25 used a TG Fischer rat carrying a human Aβ. Overall, 27 studies included both male and female mice,12,25–2729,34,40,43,50,51,57–59,61,62,67,68,77–80,82,83 10 studies included only female mice,14,44,52,54,71,73,74,85,87,88 while the remaining 38 studies included only male mice. The age at the start of diet ranged from 1 to 6 months in most studies except for four studies25,62,66,90 that initiated diet in mice aged 8–10 months and three studies that initiated diet in mice younger than one month.31,45 The duration of the dietary intervention ranged from 2 to10 months, except in one study 88 in which lasted less than one month and in 5 studies42,49,51,76,82 in which the diet lasted one month and other 7 studies in which it lasted 10–14 months.37–3947,69,86,91
Modified PRISMA flow diagram for clinical trial selection.
Characteristics of included studies.
SD: standard diet; HCHF: High carbohydrates and high fat diet; HFD: high fat diet; HCD: High-cholesterol diet; WD: western diet; SFA: saturated fatty acid; MUFAs: multiple unsaturated fatty acids; PUFAs: polyunsaturated fatty acids; LFD: low fat diet; CHOL: Cholesterol-enriched diet; APPswe: APP Swedish; KM: Kunming; NR: not reported.
Summary of the main characteristics and results of the outcome of included studies with available data.
SD: standard diet, HCHF: high carbohydrates and high fat diet, HFD: high fat diet, HCD: high-cholesterol diet, WD: western diet, SFA: saturated fatty acid; MUFAs: multiple unsaturated fatty acids; PUFAs: polyunsaturated fatty acids, LFD: low fat diet, NOR: Novel object recognition; MWM: Morris Water Maze; NOL: novel object localization; OF: open field test, BM: Barnes maze test; TSPO: translocator protein; NR: not reported; CA1so: cornu ammonis 1; DGpo: polymorphic layer of the dentate gyrus.
HD is defined as any dietary regimen with a caloric density exceeding that of standard laboratory chow. Specifically, we categorized HD into three main types based on their macronutrient composition: HFD, characterized by a fat content ranging from 45% to 60% of total kilocalories; HSD, primarily enriched with sucrose or fructose; and WD, which combine high fat content (approx. 40–45%) with high cholesterol and purified carbohydrates, mimicking typical human WD patterns; HCD, specifically supplemented with 1–2% cholesterol to induce hypercholesterolemia; and high-carbohydrate high-fat diets, designed to mimic a combination of caloric excesses. This classification allows for a better understanding of how different metabolic triggers influence AD pathology.
Qualitative assessment
Results of the qualitative assessment for each domain in all included studies are reported in Figure 2. Overall, the methodological quality of the included articles was widely variable. Only a minority of studies had a relatively low risk of bias across domains, with more than 60% of the items rated as “Yes”.26,27,98 These studies reported an adequate baseline comparability between groups, had an appropriate management of missing data, and clearly defined outcome measures. Most studies had insufficient methodological transparency. The most common methodological limitations included the absence of reported methods for random allocation, lack of blinding of caregivers and outcome assessors, and unclear or inappropriate management of missing outcome data. Moreover, most of the included studies had a limited sample size, and no information was provided whether animals who dropped out of the initial experimental samples or control group were replaced with other animals to maintain the sample size. Specifically, 13 studies were rated as having an overall low risk of bias,27,36–3841,44,47,49,54,57,70,83,98 36 studies were rated as having a moderate risk of bias since most domains were scored as unclear,14,15,25,26,28,30,33,35,40,42,45,46,48,50,55,60–63,69,71,73,75,77,79,81,82,84,85,88–90,94,96,100 and 28 studies were rated as at high risk of bias, since most domains were scored as high risk.29,31,32,34,39,51–53,56,58,59,64,66,67,72,74,76,78,80,86,87,93,95,97,100,101 Among the objectives of this systematic review, we initially planned to perform a meta-analysis on the identified outcomes to estimate the effect of a hypercaloric diet on murine models of AD. However, this was not feasible, primarily because the majority of the included studies reported their results only in graphical form, without providing numerical data such as means or medians with corresponding standard deviations or standard errors and also because during the evaluation of the studies, it was not possible to assess the effect measures, as in many studies these were not reported or had not been evaluated.
Risk of bias included studies (n = 72). Summary of scores for each domain for each included study. Green indicates a low risk of bias, yellow indicates an unclear risk of bias, red indicates a high risk of bias, while white indicates that the item was not applicable.
Effects of hypercaloric diet
Behavioral and cognitive impairment
To evaluate the impact of HD on cognitive functions in AD animal models, most included studies employed standardized behavioral tests such as the Morris Water Maze (MWM), Y-maze, Barnes Maze, and Novel Object Recognition (NOR), which assess spatial learning, memory, and exploratory behavior. Regarding cognitive functions, our semi-quantitative synthesis shows that approximately 50% of the studies employing the MWM reported significant learning and memory impairments in HD-fed animals compared to those on SD. Conversely, the remaining 50% of the studies included failed to detect significant differences between groups, highlighting a high degree of heterogeneity in behavioral responses. Approximately half of the studies using the MWM reported significant impairments in HD-fed animals compared to those on a SD, including longer escape latencies, reduced time spent in the target quadrant, and greater distances traveled, indicating deficits in spatial learning and memory.14,28,32,35,36,47,62,70,71,76,89,98,101 Similar trends were observed in Y-maze tests, where HD-fed animals exhibited lower spontaneous alternation rates and less time in the novel arm, suggesting compromised working memory and only one study 83 did not perform this behavioral test. The NOR test was performed in about 18% of the studies. Five studies reported poorer object recognition in HD-fed animals,15,30,34,45,84 while four32,56,82,96 found no significant differences. Barnes Maze results were consistent with these findings: animals on HD generally required more time to locate the escape zone, confirming spatial memory deficits except for 87 and, 96 found no differences between groups in locomotor activity while 96 showed a worsening in anxiety-like behavior. Despite these observations, nearly 50% of the studies did not detect significant differences between HD and SD groups. This variability may reflect differences in diet composition, duration of exposure, age at intervention onset, strain susceptibility, and the frequent lack of sex-stratified analyses. Notably, a few studies even reported improvements in specific cognitive domains, suggesting that the effects of HD are context dependent. Animal models demonstrated varying degrees of vulnerability to dietary triggers. APP/PS1 and 3xTg-AD models proved most sensitive, with 61.1% and 69.2% of studies respectively reporting significant pathological exacerbation. In contrast, models with a single mutation (e.g., APP Swedish) showed lower reactivity (37.5%), often requiring more extreme dietary protocols or longer durations to manifest comparable AD-like hallmarks. Indeed, this differential sensitivity is further modulated by dietary composition; specifically, our findings indicate that WD, which combines high fat with purified carbohydrates, act as more potent drivers of cognitive and pathological decline compared to pure HFD, triggering deficits in 65% versus 48% of relevant studies, respectively.
Amyloid and tau burden
The impact of hypercaloric diets on Aβ or tau pathology was a primary focus in 75.3% of the included studies. Within this group, an increase in the insoluble fraction of the Aβ1−40 fragment was documented in 59.1% of the studies reporting this specific outcome. In contrast, the effects on tau pathology were less consistent, with only 33.3% of the relevant studies reporting a significant increase in tau phosphorylation. Overall, 58 studies investigated Aβ or tau pathology in mouse models14,15,25–27,29–43,45,47–49,51–58,60–65,68–71,73,75,76,78,80,81,85,87,88,90,91,93,95,97,98,100,101 and 3 studies investigated them in rat models.25,26 Out of these, 23 studies provided results on possible variations of Aβ1−40 peptide levels in its soluble fraction29,31,34,35,41,48,51,53–55,62,64,76,80,81,87,91,95,97,98,100,101 and 22 studies provided results on its insoluble fraction29,31,34,41,47–49,51,53,55,57,62,64,65,71,73,76,81,87,91,97 Overall, 12 studies reported a significant increase in soluble Aβ1−40,29,31,34,35,41,48,54,55,62,76,80,87 while 11 studies reported no variations51,53,61,64,67,81,91,95,97,98,100 and 1 study reported a significant reduction in soluble Aβ1−40. 30 Thirteen studies reported a significant increase in the concentration of the insoluble form of the Aβ1−40 fragment,29,31,34,49,55,62,64,65,67,71,81,91,97 one study reported a reduction in its concentration, 41 and 8 studies reported no variations.47,48,51,53,57,73,76,87 A comprehensive analysis of brain Aβ1−42 peptide expression in relation to HD exposure was performed in 34 studies.15,27,29–31,33,35,36,41,45,47–49,51,53–55,57,58,62,64,65,67,70,71,73,76,81,87,91,95,97,98,100 Of these, 25 reported data on the presence of the peptide in its soluble fraction15,27,29,30,32,33,35,41,48,51,53–55,62,64,67,70,76,81,87,91,95,97,98,100 and 25 reported data on its insoluble fraction 15,27,29–31,33,35,41,48,51,53,55,62,64,67,70,76,81,87,91,95,97,98,100 Twelve studies observed an increase in soluble Aβ1−42,30,35,41,48,54,55,62,76,87,91,95,97 while 13 observed no significant changes in time.15,27,29,31,33,51,53,64,67,70,81,98,100 When considering the insoluble fraction, 12 studies reported an increase in Aβ1−42 levels,31,45,47,49,55,58,62,64,65,71,81,97 one study reported a decrease in Aβ1−42 levels, 41 and 13 studies reported no variation in time.29,36,48,51,53,57,67,70,73,76,87,91 Only one study observed an increase in the Aβ42/40 ratio because after exposure to HD, 30 while four studies reported no significant changes over time.57,87,91,98 One study that investigated the effects of WD reported a significant increase in Aβ1−34 and Aβ1−38 protein levels over time. 49 However, previous findings reported no changes in Aβ1–38 concentrations over time. 64 Conversely, one study reported that mice exposed to a high-fat diet showed significantly lower concentration of total Aβ. 56 Overall, 41 studies reported results from morphometric analyses of amyloid accumulation in brain tissue,15,25,26,29–32,34–43,45,48,53,54,56,58,60,62–65,67,69,71,73,76,78,87,88,90,93,97,98,101 specifically in the cortex15,26,29–32,34,35,37,39–43,45,48,53,56,58,62,64,71,87,97,98, hippocampus,15,29,34–37,40–42,45,48,54,56,58,62–65,67,71,78,97,98,101 striatum, 26 or more generally in either composite areas (cortex plus hippocampus),60,90 or in the whole brain.25,37,73,88,95 The overall presence of amyloid plaques in the cortex of animals fed a HD was measured in terms of amyloid burden, number of plaques, plaque density, percentage of cortex area covered, percentage of cortex volume. Sixteen studies reported a statistically significant increase in the area of immunoreactivity and plaque size,26,29–3234,39,40,42,43,48,53,62,64,71,97 8 studies reported no change in the same outcomes,15,35,37,41,45,58,87,98 and one study reported a statistically significant decrease in the same outcomes. 56 Dietary duration emerges as a critical factor for proteinopathy development: long-term exposure, exceeding 6 months, was required to observe a significant increase in insoluble Aβ1−40 in 59.1% of the studies. This confirms that advanced amyloid accumulation is a chronic process strictly dependent on the length of exposure. Thirteen studies reported an increase in amyloid burden or plaques within the hippocampus in animals fed a HD,29,34,37,54,58,62,64,66,67,71,78,97,101 while 11 studies reported no significant changes in time.15,35,36,40–42,45,48,56,63,98 However, one study among the latter group reported results from two different transgenic models that showed an increase in hippocampal amyloid load after a HD. 98 One study reported a gender-specific response, with an higher presence of both diffuse and fibrillar amyloid plaques in the cortex and hippocampus exclusively in female mice. 90 Another study analyzed the striatum in rats and found no significant diet-induced changes in amyloid deposition. 26 A similar result was observed in three studies that analyzed the whole brain.25,38,95 In contrast, one study showed an increase in amyloid deposition in whole brain slices. 49 Similarly, another study reported an 5-months earlier onset of Aβ plaques in animas fed a HD diet compared to controls. 69 Another study showed that intraneuronal Aβ oligomers accumulation started about two months earlier in both the hippocampus and cortex of two different transgenic models. 76 An increase in Aβ among the total number of immunoreactive neurons was observed in the cortex 93 and in the hippocampus 73 of animals fed with HFD. Since amyloid peptides are derived from the secretory processing of the APP holoprotein, several authors have analyzed APP expression profiles in brain tissue following high-fat diet administration. Specifically, four studies showed an increase in APP expression,31,35,69,71 whereas six studies found no significant differences.49,57,75,81,87,90 The analysis of APP transcripts largely confirmed previous observations, with one study reporting an increase in one article 31 and another reporting no changes. 32 Some studies further investigated into the molecular mechanism underlying APP processing, with four reporting an accumulation of APP CTFβ49,65,69,81,87,101 whereas no significant differences in expression levels were observed in others.81,87 The expression of APP CTFα was reported as increased in, 67 while in 49 was decreased or unchanged in. 81 One study analyzed the expression of soluble α-APP and found no variations associated with HD administration. 81 The relationship between HD and tau pathology has been extensively investigated by several researchers. Results from five studies showed no alterations in tau protein expression levels.31,42,57,80,88 However, one study observed an increase in both the soluble and insoluble fractions of this protein. Included studies predominantly focused on tau phosphorylation. 81 Five studies reported an increase in tau phosphorylation14,40,42,76,100 while ten studies reported no significant variations.31,34,45,48,57,78,80,81,85,91 The most investigated phospho-epitopes were Ser202, Thr205 (detected using antibodies AT8, CP13, 44738G), with one study reporting an increase in phosphorylation 42 and six studies reporting no changes in animals fed HD.31,48,57,80,81,94 Similarly, for Ser396 and Ser404 (detected using antibodies PHF1,44752G, 44748G), three studies reported an increase in their expression levels,14,40,76 and four studies reporter no changes.31,45,80,81 A single study that quantified the tau isoforms associated with the number of tubulin-binding repeats reported a significant increase in mRNA and protein levels of 3R-tau, 4R-tau and 3R-tau/4R-tau ratio. 52
Neuroinflammation and synaptic plasticity state
Neuroinflammation is well-known to play a critical role in the pathogenesis of AD, a process that might be exacerbated by HD. Neuroinflammation is typically mediated by the activation of glial cells, such as microglia and astrocytes, with subsequent production of pro-inflammatory cytokines, chemokines, ultimately contributing to synaptic dysfunction and neuronal loss.102–104 Neuroinflammation was explicitly investigated in 50.6% of the systematic review's corpus. Among these, 51.3% provided clear evidence of microgliosis and/or astrogliosis in regions such as the cortex and hippocampus, while 15.4% reported no significant neuroinflammatory changes associated with the diet. Among the 76 studies included in this systematic review, 39 specifically investigated the effects of HD on neuroinflammation in AD mouse models, while the remaining 33 primarily focused on amyloid/tau expression and behavioral testing, as previously described.14,35,37,38,44,45,47,49,51,53,56,64–67,70,71,74,75,79,80,82,84–87,90,93–95,97,100,101 Specifically, 20 studies reported evidence of microgliosis (Iba1+) and/or astrogliosis (GFAP+, OX-6+) in several brain regions, including the cortex and hippocampus.25,27,32,36,39,42,48,50,55,58–60,69,73,74,77,81,91,94,100 Interestingly, two studies69,94 observed higher levels of neuroinflammation, but lower levels of CD68 levels, a marker of microglia phagocytic activity, in the hippocampus of mice fed a HD compared to mice on a standard diet. These findings suggest that HD might impair the ability of microglia to phagocyte Aβ deposits. 69 Furthermore, one study 77 observed gender differences in hypothalamic gliosis, with higher astrogliosis observed only in female mice fed an HD. 77 Conversely, four studies reported a lower gliosis in AD mice after HD administration, suggesting that this type of diet might reduce inflammation in some AD mouse models.26,40,90,98 Metabolic and inflammatory shifts appear to occur early in the disease progression. Indeed, short-term interventions, less than 8 weeks, primarily induced neuroinflammation in 51.3% of relevant studies, often preceding over amyloid deposition. Several studies also assessed the release of one or more pro-inflammatory cytokines, such as TNFα and interleukins in brain tissue. Eight studies reported higher levels of these cytokines in the cortex and hippocampus of mice fed HD compared to those on a standard diet39,46,54,60,62,88,91,96 while five studies reported lower levels of the same cytokines in animals on HD compared to controls.26,33,41,52,80 Notably, one study observed an increase in IL-6 specifically in male mice and a decrease in IL-6 and TNFα mRNA levels in females mice on HD compared to those on a standard diet. 72 One study found a significant decrease in cortical TNF-alpha, COX2, and IL-6 mRNA expression but an increase in iNOS (a pro-inflammatory protein) mRNA expression in mice on HD compared to standard diet. 41 Six studies reported no significant changes in neuroinflammation between the two groups, both in terms of gliosis and in the levels of pro-inflammatory cytokines.29–31,46,57,63 However, one study reported an isolated increase in TNFα, with no corresponding changes in IL-6, IL-1β, and TGFα levels. 46 To explore potential associations between this selective increase and other neuroinflammatory factors, the study also assessed transcript levels of astrocyte, macrophages, and adhesion cell markers, but found no significant differences between groups. 46 To further investigate the molecular mechanism underlying the worsening of cognitive decline observed in AD mice on HD, six studies investigated synaptic plasticity, by focusing on synapse-associated proteins and synaptic density. Five of these studies reported a significant decrease in presynaptic proteins (Synapsin-1, Synaptophysin, etc.) in mice fed HD,15,34,54,81 while one study reported no significant alterations in synaptic density in both the cortex and hippocampus of AD mice, suggesting that the effects of HD on synaptic integrity may be model- or context-dependent. 83
Mitochondrial dysfunction. Mitochondria are known to be the powerhouse of the cell, as they can produce large quantities of ATP, which has the role of transporting and providing cells with the energy necessary for their functioning. 104 Mitochondrial dysfunction and bioenergetic crisis have been suggested to play a crucial role in the pathophysiology of neurodegenerative disease, including AD. 105 Overall, only 5 studies explicitly focused on the potential interrelation between mitochondria status, HD diet, and dementia. One study investigated the potential relationship between neuropathological Aβ plaque formation in the hippocampus and obesity at an early presymptomatic disease stage in APP/PS1 mice. 31 The authors reported an almost 50% decrease in the levels of peroxisome proliferator-activated receptor-gamma coactivator-1alpha (PGC-1α), a member of a family of transcription coactivators that plays a central role in the regulation of cellular energy metabolism, in the hippocampus of APP/PS1 mice fed with HD, suggesting a possible involvement of this protein in both mitochondrial biogenesis alterations and metabolic function. However, the study observed no alterations in downstream effectors of PGC-1α, such as nuclear respiratory factors 1 (NRF1) and 2 (NRF2). Another study investigated the potential effects of HD on the OXPHOS enzymatic system, mRNAs, and protein levels of molecules associated with mitochondrial energy status and biogenesis in APP/PS1 transgenic mice. 45 The authors reported a significant decrease in mitochondrial OXPHOS metabolism in the hippocampus of animals on HD. Moreover, when considering the APP/PS1 mice group, small but statistically significant differences were observed in the mRNA expression of NRF1 and NRF2 between mice on HD compared to mice on standard diet. Two studies investigated the potential effects of HD in 3xTgAD mice. One study assessed the potential effects of HD on mitochondrial respiration in intact hippocampal slices from 3xTgAD mice. 82 The study reported a small impact of the AD genotype on overall oxygen consumption rates (OCR), but a 25% increase in basal OCR in the 3xTgAD mice on standard diet. However, a significant decrease in basal and maximal OCR levels was observed to be 3xTgAD mice on HD, showing a significant impact of HD on tissue OCR. These results suggested that an 8-week HD could significantly compromise mitochondrial function in the hippocampus. One study investigated endothelial and neuronal mitochondria number and morphology along with synaptic density using transmission electron microscopy in the subiculum area of the hippocampal formation and in the cortex of 8-months-old mice, thus at a stage before amyloid plaque pathology is observable. 82 Results showed no changes in the number of mitochondria in the hippocampus and cortex in 3xTgAD mice regardless of the type of diet. However, 3xTgAD mice on HD showed a higher percentage of elongated mitochondria (∼38%, p<0.05 versus SD) in the hippocampus but not in the cortex. Moreover, data also showed severe morphological alterations in the mitochondria in the hippocampus of 3xTgAD mice on HD, including enlarged, swollen, and/or absent cristae, and less dense or empty matrix. Another study investigated the effects of a 3-month HD diet on APP23 transgenic mice in term of amyloid deposition and cognitive functions. 90 Since the phospholipid composition of cellular membranes is extremely important for cell signaling, including apoptosis and phagocytosis the study carried out a specific analysis of the brain lipidomic profile. In this context, the most significant changes observed in response to HD were associated with cardiolipin (CL) a major phospholipid of the inner mitochondrial membrane. A wide imbalance between CL and other phospholipids, such as phosphatidylinositol biphosphate and phosphatidyl triphosphate, was observed in the brain lipidome of APP23 mice on HD. This imbalance was closely related to mitochondrial dysfunction, 106 characterized by a different and significant modulation of several genes associated with mitochondria components.
Oxidative stress
Oxidative stress (OS) is also reported as having a relevant role in the pathogenesis of neurodegenerative diseases of the central nervous system (CNS), AD, and Parkinson's disease. 92 An increasing number of studies suggested that diets rich in saturated fats and/or refined sugars might activate pro-inflammatory pathways that can result in the production of reactive oxygen species (ROS), and initiate a process of neurodegeneration.92,107 However, the potential role of HD in modulating OS in neurodegenerative disorders remains unclear. To attempt to clarify whether a direct correlation exists between HD and OS, we included in this systematic review studies that specifically investigated the effects of HD on this specific outcome. Only 11.7% of the studies analyzed OS. Only nine of the included studies analyzed the impact of different HD on OS. Of these, eight studies used APP/PS1 mice models of AD, while one study used 3xTgAD mice. 85 Results from all studies consistently reported an association, regardless of the type of mouse model used, between HD and OS and antioxidant defense mechanisms. One study reported higher levels of ROS and endostatin expression in the brain of mice fed with HD, along with a significant decrease in glutathione (GSH). 100 Similarly, two other studies reported a decreased expression of several proteins involved in antioxidant response mechanisms (Nrf2, MnSOD, HO-1) and a burst of ROS.46,85 One study investigated the mRNA hepatic expression of antioxidant defense enzymes in APP/PS1 mice fed with high cholesterol diet. 46 The study reported a decrease in SOD1, SOD2 and NOX4 mRNA levels along with an overexpression of the HO-1 and NOX2 genes. There results were not consistent with results from another study, likely due to differences in the dietary composition and the specific animal model selected. 85 Only one study reported that HD induced prediabetes in AD mice, along with an increase in OS, higher metalloproteinase activity, and amyloid plaques deposition. 53 In addition, the study evaluated the deposition of Aβ plaques around arterioles, a condition known as cerebral amyloid angiopathy. 53 Results showed a significantly higher frequency of cerebral amyloid angiopathy in mice on HD, suggesting a disruption of antioxidant defense pathways in the AD brain. Two studies investigated the impact of HD on lipid peroxidation and protein oxidation in APP mouse models. While one of the studies observed no significant differences between HD and standard diet, 31 the other study reported higher levels of lipid hydroperoxides and protein nitrosylation in animals on HD compared to standard diet. 51 These findings are consistent with other studies on mice models suggesting that HD might induce the accumulation of potentially toxic products that may promote neuronal stress. 108 The complex mechanistic pathways linking hypercaloric dietary triggers to metabolic stress, neuroinflammation, and the subsequent exacerbation of AD pathological hallmarks are visually summarized in the Infographic.
Discussion
Results from the studies included suggested that the exposure to HDs in transgenic animal models of AD is frequently associated with a decline in cognitive functions and an increase in the misfolding of Aβ and tau proteins, along with neuroinflammation, mitochondrial dysfunction, and OS. Our findings align with human epidemiological data linking WD to AD risk, although structural differences in rodent metabolism and the use of familial mutations limit direct translation. Nevertheless, the early onset of diet-induced neuroinflammation suggests that managing metabolic health represents a key non-pharmacological “window of opportunity” for human AD prevention. However, these findings should be interpreted taking into due consideration the substantial heterogeneity across studies in terms of experimental design, dietary composition, type of animal model, and methodological quality. A quantitative analysis of data through meta-analyses should be useful but was not feasible mainly due to this heterogeneity and lack of quantitative data of the outcome considered. Most studies investigating behavioral outcomes reported that mice on HDs had worse performances in learning and memory tasks, especially in spatial navigation tasks such as MWM and Y-maze. However, nearly half of the studies included found no significant differences between animals on HD and SD. This discrepancy may be due to the heterogeneity observed in the duration of dietary intervention, the age at the beginning of the intervention, and the strain susceptibility. Notably, studies using diets with higher saturated fat content and longer duration of the intervention were more likely to report a negative effect of the diet on cognitive functions, indicating that both the dose and the length of exposure may play a relevant role. The observed heterogeneity in AD-like hallmarks is strictly modulated by experimental variables: specifically, APP/PS1 (61.1%) and 3xTg-AD (69.2%) models demonstrate significantly higher sensitivity to dietary manipulation than single transgenic strains (37.5%). Furthermore, while WD induce cognitive decline more consistently (65%) than pure HFD (48%), the progression of proteinopathy remains duration-dependent, with chronic exposure (>6 months) being a fundamental requirement for the accumulation of insoluble Aβ1−40 in 59.1% of studies. However, some studies present conflicting evidence, reporting even an improvement at the behavioral level. The causes for these discrepancies are likely diverse, including the duration of the diet, the specific animal model employed, the sex of the animals, and other factors such as genetic background and the precise nutritional composition of the diet as indicated in a recent study. 109 A key question emerging from our synthesis is whether diet-induced obesity and the resulting metabolic phenotype are mandatory prerequisites for the exacerbation of AD. Our analysis suggests that while severe cognitive impairment and amyloid deposition are often strongly correlated with significant weight gain and insulin resistance, obesity is not an absolute requirement for neurodegeneration. In several included studies, even short-term exposure to HD triggered neuroinflammatory markers and mitochondrial dysfunction prior to the onset of a full metabolic syndrome or overt obesity. This indicates that the nutritional composition of the diet may exert direct neurotoxic effects, likely through increased blood-brain barrier permeability and OS, independent of the systemic accumulation of adipose tissue. Therefore, while a metabolic phenotype characterized by obesity and insulin resistance acts as a potent accelerator of AD, the diet itself can initiate early pathological shifts regardless of the animal's final body weight. The divergence in cognitive outcomes (approximately 50% of studies reporting no significant impairment) is closely linked to diet composition. Our subgroup analysis reveals that WD, which combines fats and sugars, are more effective in triggering deficits (65% of cases) compared to pure HFD (48%). This suggests that the synergistic effect of saturated fats and purified carbohydrates is a more potent driver of synaptic dysfunction than lipid intake alone. When considering neuropathological markers, a significant number of studies reported higher levels of both soluble and insoluble Aβ1−40 and of Aβ1−42 in animals on HD, particularly in the cortex and hippocampus, which are involved in memory and cognitive processing. Similarly, higher tau phosphorylation at multiple phospho-epitopes (Ser202, Thr205, Ser396, Ser404) was observed in a subset of studies. However, several studies reported conflicting results, including no change in amyloid burden or even a lower amyloid burden, emphasizing how complex and context-dependent relationship can be between diet and neuropathology. Gender, genetic background, and gut microbiota composition could also be involved in modulating the response to dietary interventions. Neuroinflammation was also reported as a relevant mediator of brain pathology after HD. Several studies reported increased gliosis (microgliosis and astrogliosis) and higher levels of pro-inflammatory cytokines such as TNF-α and IL-6 in mice on HD. Interestingly, some studies showed the opposite effect or gender-specific patterns, with females showing higher vulnerability in some instances. These findings suggest that the inflammatory response to HD may vary not only in relation to gender, but also in relation to brain region and may also change over time. Moreover, reductions in microglial phagocytic activity suggest a detrimental shift toward a dysfunctional immune phenotype that may impair Aβ clearance. Mitochondrial dysfunction was also frequently reported in animals after being fed with HD. HD was reported as being associated with impaired oxidative phosphorylation, lower expression of key regulatory factors such as PGC-1α and TFAM, and the onset of structural abnormalities in hippocampal mitochondria. These alterations, particularly the loss of mitochondrial cristae integrity and the disruption of lipid composition such as cardiolipin imbalance, can indicate energy failure and a higher vulnerability to neurodegeneration. Some studies also reported a reduction in presynaptic markers such as synaptophysin, thus supporting the hypothesis that HD might contribute to synaptopathy even before the evident deposition of plaques. 110 OS is consistently reported as having a relevant role in the neurodegenerative processes. Results from included studies confirmed that HD can increase ROS production, while impairing antioxidant defenses, including lower levels of Nrf2, MnSOD, and HO-1. Considering the central role of redox imbalance in triggering inflammatory and apoptotic cascades, future studies should systematically incorporate OS markers. A significant limitation of included studies was the underrepresentation of female animals, which may prevent the identification of potential gender-specific responses to diet. 111 Despite the consistently reported higher prevalence of AD in women, most studies relied exclusively on male mice or did not stratify data according to gender. This is a significant limitation considering the well-documented modulatory effects of estrogens on amyloid metabolism, tau phosphorylation, and mitochondrial function. 108 While a small number of studies highlighted gender-specific effects of HD, these were the exception rather than the norm. Future research should specifically consider gender as a biological variable and explore its interactions with hormonal status and APOE genotype. One critical issue raised by the data of this systematic review concerns the role of sex as a contributor to the observed heterogeneity and as a potential source of bias. The pronounced underrepresentation of female animals within the analyzed body of literature represents a significant methodological gap, hindering the identification of sex-specific responses to hypercaloric challenges. Given that sex acts as a pivotal biological modifier influencing amyloid metabolism, tau phosphorylation, and mitochondrial integrity, the predominance of male-only models (49.4%) and the lack of sex-stratified reporting limit the generalizability of the current evidence; thus, future research must systematically incorporate both sexes to accurately reflect the differential vulnerability observed in human AD pathology. The marked predominance of male-only murine models, together with the frequent lack of sex-stratified analyses, limits the generalizability of the findings. As a result, the overall worsening of behavioral and pathological outcomes associated with hypercaloric diet exposure should be interpreted with caution, as it predominantly reflects evidence derived from male models. Importantly, the limited number of studies that included female animals and reported sex-specific outcomes suggests that the effects of a hypercaloric diet are not uniform across sexes. Both exacerbating and divergent effects were observed in female models, indicating that sex may act as a biological modifier rather than a simple confounder. These findings highlight the need for systematic inclusion of both sexes and for sex-disaggregated data reporting in future preclinical studies.
A significant result from our review was the low methodological quality of many of the included studies. Key aspects such as randomization, blinding, and transparency in the reporting of exclusions were inconsistently applied. Only 4.28% of the included studies explicitly reported having implemented a form of randomization,65,93,100 and even fewer studies provided details on the randomization procedures used. Moreover, blinding of outcome assessors was reported by only 20% of the studies, implying a substantial risk of bias that may affect the reliability of the reported results.15,25,26,29,30,36,44,47,70,77,80,83,94,100 A very small number of studies included clearly stated the number of animals excluded from the analyses, thus limiting the transparency and reproducibility. This lack of transparency affects the internal validity and the reproducibility of results, which is a critical issue often reported in preclinical research. However, some critical issues should be observed in relation to the checklist used in this systematic review to assess the quality of included studies, as some criteria cannot be uniformly applied or equally weighted across all experimental contexts. While the SYRCLE's RoB Tool is widely adopted for assessing internal validity in animal studies, some of its items may not be fully appropriate for all experimental designs, particularly in studies involving AD models fed specific types of diets. 112 Several domains, such as random housing and blinding of caregivers or investigators, are difficult to be consistently applied across preclinical studies where group allocation may not be feasible or clearly reported. Similarly, the domain assessing selective outcome reporting often requires accessing pre-registered protocols, which are uncommon in animal research. This limitation has been reported in literature, with recent studies 112 underlining that some SYRCLE items lack empirical validation within the context of animal research and may result in a high number of “Unclear” ratings due to poor reporting rather than actual methodological flaws. For this reason, our assessment adopted a conservative approach: we maintained all SYRCLE items to ensure comparability with previous reviews, but we explicitly marked domains such as random housing and caregiver blinding as “Not applicable” when they were impossible to be applied due to the specific study design. This adjustment was aimed at preserving the interpretability of quality scores, while acknowledging the structural limitations of applying a clinical-oriented RoB framework to studies on animal models. 113 In conclusion, while most studies focused on more common endpoints such as the accumulation of Aβ and tau, emerging areas such as the modulation of the gut-brain axis, neurovascular integrity, and brain connectivity remain largely unexplored in studies exploring the potential relation between HDs and AD. Moreover, few studies adopted a longitudinal design capable of distinguishing between the short- and long-term effects of the exposure to HD. The integration of multi-omics approaches, such as transcriptomic, lipidomic, metabolomic could provide a more comprehensive understanding of how HDs might influence the pathophysiology of AD. To increase their translational relevance and scientific robustness, future preclinical studies should standardize dietary protocols, such as fat type, cholesterol content, and duration of exposure, and ensure the inclusion of female mouse models. Additionally, research should explore novel endpoints, including neurovascular integrity. Despite many studies suggesting detrimental effects of HD on cognitive function and AD-related pathology in animal models, wide heterogeneity and several methodological flaws significantly limit the strength of these conclusions. A more rigorous and comprehensive approach is essential to clarify the complex relationship between nutrition and neurodegeneration, considering also gender-specific characteristics, that could provide an appropriate basis to guide future preventive and therapeutic strategies.
Footnotes
Acknowledgements
The authors have no acknowledgments to report.
Ethical considerations
Not applicable
Consent to participate
Not applicable
Author contribution(s)
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the European Union—Next Generation EU –PNRR M6C2—Investment 2.1 Enhancement and strengthening of biomedical research in the NHS, under the project identification number PNRR-MAD-2022-12375822, CUP I55E22000560006; Ministero della Salute.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement
Further inquiries can be directed to the corresponding author.