Alzheimer’s disease (AD) is the most common cause of neurodegenerative dementia and one of the most burdening diseases worldwide (1). The main pathophysiological hypothesis of AD is the amyloid-beta (Aβ) cascade (2), with abnormal phosphorylation of tau protein, resulting in the formation of neurofibrillary tangles, representing another hallmark of AD pathophysiology. Both Aβ-induced plaques formation and tau accumulation result in neuronal degeneration, and indeed these mechanisms constitute the bases of a recent biological classification of AD, relying on biomarkers evidence of amyloid, tau and neurodegeneration (3). However, other pathophysiological mechanisms are known to exert a role in AD development, with cerebrovascular disease (CVD) being a crucial one.

CVD comprises a series of different lesions, such as white matter hyperintensities, cerebral microbleeds, enlarged perivascular spaces and lacunar infarcts, visible on pathological examination and detectable in vivo with magnetic resonance imaging (MRI) (4). Vascular abnormalities have been reported in at least half of AD cases and have an additive effect on AD pathology (5). In particular, older subjects with silent brain infarcts on MRI have an increased risk of dementia and a steeper decline in cognitive function (6) and both large and small brain vessels diseases increase the likelihood of AD diagnosis (7). Clinicopathological studies have shown that silent brain infarcts lower the threshold for AD pathology to become clinically significant (8, 9). However, the mechanisms behind this increase in risk remain unclear (10).

Importantly, traditional cardiovascular risk factors such as obesity, hypertension, and diabetes can all increase the risk of cognitive decline and dementia (11). In particular, high midlife Body Mass Index (BMI), as well as central adiposity, results in a doubled risk of developing dementia in late life (12). The mechanisms underlying this increased risk might lie in the endocrine actions of the white adipose tissue, secreting hundreds of cell-signaling molecules, known as adipokines. Of those, leptin and adiponectin have been investigated as key mediators of the communication between periphery and the central nervous system (CNS) in healthy and pathological condition (13).

Leptin and adiponectin are adipocyte-derived hormones shown to have neuroprotective actions by modulating synaptic plasticity, improving learning and memory, and reducing Aβ levels and tau hyperphosphorylation in preclinical models (14, 15). Interestingly, leptin treatment has shown a neuroprotective effect in preclinical models of ischemic brain damage (16), while leptin deficiency resulted in larger infarct volume after focal cerebral ischemia in animal experiments (17). However, some epidemiological studies have found that high circulating leptin levels were associated with ischemic cerebrovascular disease (18, 19), while others have found no association (20). Low circulating adiponectin levels have been associated with ischemic cerebrovascular disease and with increased mortality after ischemic stroke (21, 22), although others have reported opposite findings (23).

Overall, evidence linking adipokines to CVD is controversial and it is not yet known what its actual contribution to AD pathophysiology could be. Here, we evaluated whether baseline serum leptin and adiponectin levels could predict the development of silent brain infarcts in subjects with MCI and AD from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) cohort.

Data used in the preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). The ADNI was launched in 2003 as a public-private partnership, led by Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test whether serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early Alzheimer’s disease (AD). All ADNI studies are conducted according to the Good Clinical Practice guidelines, the Declaration of Helsinki, and U.S. 21 CFR Part 50 (Protection of Human Subjects), and Part 56 (Institutional Review Boards). Written informed consent was obtained from all participants. The ADNI protocol was approved by the Institutional Review Boards of all of the participating institutions. The ethics committees/institutional review board that approved the ADNI study are listed within Supplementary Information File .

The cohort of subjects extracted from ADNI database for this study consisted of 566 individuals, of which 58 HC, 396 patients diagnosed with MCI and 112 AD patients. The demographic features and baseline biomarkers levels of the entire population are summarized in Table 1 . Overall, the study population mean age was 74.8 ± 7.4 years and comprised 38% of female. The three groups were homogenous in terms of age, gender, education levels and BMI. The proportion of ApoE4 carriers was significantly higher in AD group compared to both HC and MCI, as well as in MCI compared to HC, as expected. The prevalence of the most common cardiovascular risk factors (smoke, diabetes, hypertension, dyslipidaemia, atrial fibrillation) was not different among the three groups.

As expected, the neuropsychometric tests showed worse cognitive performance and behavioral symptoms in AD compared to MCI and HC, and in MCI compared to HC.

As for the CSF biomarkers, levels of Aβ were significantly lower in in AD compared to both HC and MCI and in MCI compared to HC. An opposite trend (higher levels) was observed for tau and phospho-tau. Brain metabolism evaluated by FDG-PET in the main regions of interest (temporal lobes and angular right and left gyri and at the posterior bilateral cingulum), was significantly reduced in AD and MCI compared to HC, except for the temporal right lobe which did not show significantly different uptake in MCI vs HC.

On average, 7.4% of the whole population had at least one detectable brain infarct on MRI. The mean number of infarcts was not significantly different among the three groups.

We evaluated whether the levels of serum adipokines leptin and adiponectin differed among the three groups. Only serum leptin levels were significantly lower in MCI compared to HC. Adiponectin levels were similar among the three groups ( Table 1 ). In the whole study population, an inverse correlation was observed between adiponectin and leptin levels (Pearson’s r = −0.18 [− 0.25–0.09] p < 0.00).

Since it has been demonstrated that leptin and adiponectin levels tend to be generally higher in women than in men (31), we wanted to test whether this could be the case also in our study population. Therefore, we stratified our three groups by gender and compared serum leptin and adiponectin levels between men and women within each diagnostic group. As expected, both adipokines levels were significantly higher in women in each of the three groups, as shown in Supplementary Figure 1.

To establish which are the variables that could be associated with baseline serum leptin levels in our population, we performed a multivariate regression analysis for leptin levels in our disease group (MCI and AD), building a model that included age, gender, diagnosis, BMI, ADAS-Cog, MMSE, dyslipidaemia, hypertension and brain infarcts. The variables that showed a significant association with serum leptin levels in our population were age, gender, BMI, MMSE and hypertension as shown in Supplementary Table 1 . A good fraction of the variability in plasma leptin levels was explained by this model, as indicated by the global R² of 0.58. A large fraction of this contribution is shared by both gender and BMI as indicated by the relative contribution to the global R² (45.5% and 48.5% respectively), while a marginal impact on the association is evidenced by the other 3 significant factors (age, MMSE and hypertension) as shown by their almost irrelevant, albeit statistically significant, contribution to R² .

Similarly, we performed a multivariate regression analysis with the same predictors for adiponectin levels, showing that age, gender, BMI and dyslipidaemia were associated with baseline adiponectin levels with a global R² of 0.16. As for leptin, a large fraction of the R² is related to the gender and BMI partial contribution ( Supplementary Table 2 ).

To evaluate the influence of leptin and adiponectin on the presence of brain infarcts in AD and MCI subjects, we built two multivariable fractional polynomial logistic regression models for the presence or absence of silent brain infarct at baseline. The model for leptin included as independent predictors age, gender, leptin levels, diagnosis, BMI, diabetes, dyslipidaemia, hypertension, smoke and atrial fibrillation. The model proved significant, with and overall modest pR^{2 =} 0.07, with age and leptin being the only significant predictors of brain infarcts at baseline, as shown in Table 2 . Of note, as indicated by the percent of contribution to the global pR2, the greatest fraction is attributable to age (81.6%), leaving to leptin a modest contribution (18.4%).

Similarly to leptin we built a logistic model for the presence of baseline brain infarcts, using age, gender, adiponectin levels, diagnosis, BMI, diabetes, dyslipidaemia, hypertension, smoke and atrial fibrillation as independent predictors. The overall model proved significant (χ2 (1) =12.96, p<.01), with age (odds ratio of 2.45), being the only significant predictor of brain infarcts at baseline ( Table 3 ).

Finally, to establish whether circulating adipokines might influence the development of brain infarcts in this cohort of 508 MCI and AD subjects, we conducted a multivariable fractional polynomial Cox regression analysis with age, gender, baseline leptin and adiponectin levels, BMI, hypertension and ApoE4 carrier status as predictors ( Table 4 ). 83 new brain infarcts were observed for a period of 15 years. The Cox analysis showed that only age was a significant predictor for the development of brain infarcts in this population (hazard ratio 1.63). At a maximum follow up time of 15 years, 50% of the population had developed at least one brain infarct ( Figure 1A ), and the effect of age on brain infarct incidence at follow up is shown if Figure 1B . After adjusting for all the variables included in the Cox regression model, an increase in age by 10 years would increase the risk of developing a brain infarct by 63%. In the representative Figure 1B , while the risk of developing a brain infarct at 50 years of age is 20%, at 90 years of age it almost reaches 80% in this study population.

The focus of this work has been on the role of two adipokines, leptin and adiponectin, as potential CVD biomarkers in the AD continuum. We showed that leptin levels at baseline were associated with the presence of brain infarcts and, although we failed to demonstrate a predictive value of serum leptin and adiponectin on brain infarcts development in this population, our findings have shed further light on this topic, which is still much debated.

Leptin is implicated in body weight homeostasis by regulating feeding behavior, energy expenditure, thermogenesis, autonomic function, and systemic metabolism (13). Leptin has been shown to have a neuroprotective effect in experimental models (32). In clinical AD studies though, evidence is controversial, as some studies have shown decreased leptin levels in CSF and plasma in AD patients, while others have shown increased or unaffected levels (33–35). Leptin treatment ameliorates ischemic damage in models of oxygen-glucose deprivation in vitro and in animal models of cerebral ischemia (36). However, several epidemiological studies have found that high circulating leptin levels are associated with ischemic cerebrovascular disease (20).

Adiponectin modulates inflammatory responses, energy expenditure, central food intake, and several metabolic processes (12). Low circulating adiponectin levels have been associated with heart failure, coronary artery disease, dyslipidemia, and type 2 diabetes (11). Low circulating levels of this adipokine have been associated with ischemic cerebrovascular disease (37), and with increased mortality after ischemic stroke, predicting neurological sequelae severity and functional outcome (23). Adiponectin has important anti-inflammatory and anti-atherogenic effects by suppressing the production of pro-inflammatory cytokines and modulating the expression of anti-inflammatory cytokines on different cell types. Moreover, adiponectin is considered as an immune mediator because it is significantly involved in innate and adaptative immune response (38).

In our population, adiponectin levels did not significantly differ among the three diagnostic groups. However, leptin levels were lower in MCI compared to controls. This is consistent with previous findings reported by Holden et al. who suggested higher leptin concentration were correlated with a less expressed cognitive decline in the elderly in a prospective, longitudinal cohort study (39). The data from Khemka also confirmed that lower levels of leptin were found in subjects with both AD and MCI compared with controls (33). Our data suggest a trend towards higher levels of adiponectin with increasing cognitive decline, however larger populations studies might highlight any significant differences. Some studies have shown a significant positive association between higher serum adiponectin levels and better cognitive function in postmenopausal women (40), while other reports suggest that in AD patients increased adiponectin serum levels could suggest a compensatory mechanism against neurodegeneration (38).

Consistent with previous findings, leptin and adiponectin levels in our study population were significantly higher in women than men (31). Indeed, the sexual dimorphism of leptin concentration is not uncommon in animal and human studies. It has been shown that circulating leptin concentration was significantly higher in women with heart failure or diabetes than men (41). Although the detailed mechanism is not yet fully understood, the significant difference in leptin concentrations in diabetic women might be explained by a higher proportion of body fat in women and the counter-regulatory effect between leptin and testosterone (42), high level of adiponectin in female, as well as high oestrogen that functions as negative regulator of leptin (43).

Serum leptin levels in our study population were significantly associated with gender, BMI and hypertension, in accordance with previous evidence on other populations (44, 45). Similarly, baseline adiponectin levels were associated with age, gender, BMI and dyslipidaemia.

In our study, we focused on the association of circulating adipokines with cerebrovascular pathology in MCI and AD. As Alzheimer’s clinical syndrome is multifactorial and might be due to mixed pathologies (46), other factors than amyloid and tau will likely contribute to and/or modify onset and progression of symptoms. Among these factors, CVD plays a critical role (47). Neuropathological studies have shown that cerebrovascular pathology is a major risk factor for clinically diagnosed AD-type dementia (48). A large autopsy-based neuropathological study importantly revealed that 80% of patients diagnosed with AD and no evidence of mixed (vascular) dementia had vascular pathology including cortical infarcts, lacunes, cerebral microbleeds, and multiple microinfarcts indicative of small vessel disease (SVD), intracranial atherosclerosis, arteriolosclerosis, perivascular spacing, and cerebral amyloid angiopathy, supporting the concept that cerebrovascular dysfunction is prominent in AD and lowers the threshold for dementia for a given AD pathology burden (49). In our population, 7.4% of participants had at least one brain infarct detected on brain MRI. At baseline, leptin levels were significantly associated with the presence of brain infarcts. The prevalence of brain infarcts in the ADNI population is generally lower compared to other cohorts; this might limit the power to find significant associations with adipokine levels and it might be because the prevalence of cardiovascular risk factors is generally lower in the ADNI cohort compared to the American population. However, our analysis did not find significant association between cardiovascular risk factors and brain infarcts, both at baseline and at the longitudinal analysis since age was the only predictor of brain infarcts in this population.

This work has some limitations: body fat and other parameters (e.g., insulin, testosterone, oestrogen levels), which may also play an important role on leptin, adiponectin and cognitive decline, were not collected. Moreover, leptin and adiponectin levels were only measured at baseline. We evaluated serum leptin and adiponectin levels; however, these might not reflect central adipokines levels, therefore CSF leptin and adiponectin should also be evaluated. Longitudinal cohort studies with larger sample size, different ages, and different vascular risk factors are required to comprehensively expand the understanding of the role of adipokines in AD. A strength of this work is represented by the fact that the ADNI cohort is well characterized in terms of biomarkers, and with a long follow up time.

Data used in preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at: http://adni.loni.usc.edu/wp-content/uploads/how to apply/ADNI Acknowledgement List.pdf.

Publicly available datasets were analyzed in this study. This data can be found here: http://adni.loni.usc.edu.

The studies involving humans were approved by Alzheimer’s Disease Neuroimaging Initiative (ADNI) Ethics approval. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

GC: Writing – original draft. LB: Writing – original draft. MS: Writing – original draft. ND: Writing – review & editing. MP: Writing – review & editing. EE: Writing – review & editing. FS: Writing – review & editing. PE: Writing – review & editing. NF: Writing – review & editing. DV: Writing – review & editing. GR: Writing – review & editing. GF: Conceptualization, Writing – review & editing.