Worldwide, more than 50 million people suffer from dementia, in 60–70% of the cases caused by Alzheimer’s disease (AD) (World Health Organization (WHO), 2018). The social burden posed by AD places it amid the top priorities for medical research: studies on the subject receive billions of dollars each year in the United States alone (Alzheimer’s association, 2020). Even with the extensive scientific efforts taking place, AD’s pathophysiology remains far from elucidated. Currently, our comprehension of the disease’s mechanisms may be about to face a turning point. So far, most attempts to explain AD’s onset and progression have focused on the brain deposits of beta-amyloid (Aβ) protein. Lately, though, amyloid-targeting drugs have failed to show clinical benefits in successive trials. Such mounting high-quality evidence fuel an active debate around the “amyloid hypothesis” of AD and the limits of its explanatory power (Makin, 2018). Ever more attention is shifting to the disease’s vascular and inflammatory features, encouraging new models to come forward. For instance, one theory suggests that concurrent cerebrovascular dysfunction could prompt AD onset, or synergistically contribute to its progression (Solis et al., 2020). Another hypothesis points to neuroinflammation as a major component of AD’s cognitive decline (Heneka et al., 2015). In this context, the renin-angiotensin system (RAS) emerges as a credible player in AD’s pathogenesis, particularly the RAS’ components involved in cerebrovascular regulation and brain inflammation (Kehoe, 2018).

Primarily remembered as a blood pressure controller, the RAS is in fact a multifaceted system for homeostasis, carrying out diverse and intricate functions. In the past decades, important discoveries transformed the way we think about the RAS. First, active peptides were described and added to the angiotensins’ family, leading to the RAS’ division in two main components: the classical axis, led by its main effector molecule, angiotensin II (Ang II), and the alternative axis, driven by angiotensin (1–7) [Ang-(1–7)]. Second, the concept of “local RAS” (in opposition to systemic RAS) was coined after RAS compounds were found in different organs and tissues, including the central nervous system (CNS) (Mascolo et al., 2017). Following these developments, the RAS has been implicated in medical conditions outside the heart and the kidneys, including neuropsychiatric disorders, AD among them (Rocha et al., 2018). Brain RAS is present to some extent in the hippocampus and other areas affected by AD pathology, where an interaction could take place. More significant, though, are the postulated relations between systemic RAS and AD. At the neurovascular unit level, circulating Ang II deregulates the cerebral blood flow, weakens the blood-brain barrier, and promotes neuroinflammation—all actions that might contribute to AD onset and progression. Plasma Ang-(1–7), on the contrary, might protect against AD-related damages, once it increases cerebral blood flow, reduces blood-brain barrier permeability, and inhibits inflammation. These theoretical perspectives are extensively discussed in our recent review (Ribeiro et al., 2020).

Literature on RAS-AD interaction is profuse and goes beyond theoretical speculation. Experiments with animal models help to build the case for a distinct role for RAS in AD pathogenesis, many even testing RAS as drug target in AD (Saavedra, 2016). Evidence from humans is contrastingly scarce. A recent review found 20 reports of RAS’ molecules measured in AD subjects (Ribeiro et al., 2020). Most of these studies were investigating RAS’ components in the CNS and thus examined brain tissue or cerebrospinal fluid (CSF), with conflicting results. The state of systemic RAS in AD is still largely unexplored. Less than ten studies dosed RAS molecules in AD patients’ blood samples, none of which assessed both RAS axes. More often the focus has been the classical axis, especially the angiotensin-converting enzyme (ACE), and even its role remains unclear. One exception is worth noting: in a case-control study with 228 participants, plasma Ang-(1–7) was significantly lower in the AD group (Jiang et al., 2016a). However, such interesting finding only describes the alternative axis, as other molecules have not been analyzed in the same sample.

Here, we aimed to help shed more light on the complex relationship between AD and systemic RAS. Hence, we have conducted a cross-sectional exploratory study, comparing plasma levels of Ang II and Ang-(1–7) in AD patients vs. cognitively healthy age-matched subjects. Within each group, we searched for correlations between RAS biomarkers and relevant neuroimaging variables, particularly the cortical areas most hit by AD, and markers of cerebrovascular lesions. With our results, we expect to generate hypotheses about the state of both systemic RAS axes in AD, which may help to build inferences about potential mechanisms of interaction.

To find whether RAS systemic axes were unbalanced in AD, we compared Ang II and Ang-(1–7) levels between AD patients and cognitively healthy controls. Our results showed that Ang-(1–7) was reduced in AD patients, whereas no difference was found in Ang II levels. Ang-(1–7)/Ang II ratio was lower in AD patients simply reflecting the difference in Ang-(1–7). Then, to investigate if systemic RAS is linked to brain pathology, we looked for correlations between plasma angiotensins and MRI variables. In AD patients, but not in controls, plasma levels of Ang-(1–7) correlated with WMHs. No association was found between angiotensins and selected cortical volumes.

It is noteworthy that all our significant findings regarded the RAS alternative axis’ main peptide, Ang-(1–7). Before our study, Ang-(1–7) was already found to be reduced in AD patients (Jiang et al., 2016a), and mice models (Jiang et al., 2016b). The notion that AD patients may lack Ang-(1–7) is consistent with the peptide’s alleged neuroprotective properties (Farag et al., 2017). To our knowledge, no prior study has evaluated Ang-(1–7) in relation to MRI measurements, neither in AD patients nor in controls. As for Ang II, however, our results contrast with a recent report by Yasar et al. (2020), who found an association between Ang II levels and hippocampal atrophy in cognitively healthy elders. The size of our sample and the dissimilar demographics may account for this difference. Another study (Zhuang et al., 2016) described a higher ACE activity in AD patients compared to controls. This would presumably result in higher Ang II levels, which were not verified in our sample. Differences in the target population may help explain the disparities, as Zhuang and colleagues selected subjects with moderate-to-severe AD.

In our results, it was unsurprising that WMHs’ volume was higher in AD patients. In fact, MRI and post-mortem pathological studies reveal that cerebrovascular lesions are more frequent in AD (Schneider et al., 2004, 2007; Jellinger and Attems, 2005; Attems and Jellinger, 2014; Suemoto et al., 2017; Hase et al., 2018). Classically, AD and cerebrovascular disease are considered independent entities, which often happen together only because the prevalence of both increase with age. Assuming such independence, cerebrovascular lesions would contribute to cognitive decline in AD patients only by reducing the “brain reserve,” thus allowing symptoms to manifest earlier (Kapasi and Schneider, 2016; Raz et al., 2016). This notion has been challenged by mounting evidence of interaction between cerebrovascular disease and AD amyloid and tau pathologies (van Norden et al., 2012; Nucera and Hachinski, 2018; Solis et al., 2020). For instance, cerebral blood flow is dysregulated in AD (Jagust et al., 1997; Roher et al., 2012). Traditionally, ensuing brain hypoperfusion is interpreted as a late consequence of AD pathology, when neurodegeneration diminishes cerebral metabolism, and thus reduces the brain’s need for blood. But there are clinical, radiological and pathological findings suggesting that the mechanisms are likely more complex: control of cerebral perfusion can be disrupted early in AD, and the reduction in blood supply may exceed the decline in metabolic demand (Ruitenberg et al., 2005; Binnewijzend et al., 2014; Hays et al., 2016; Love and Miners, 2016a). Some hypotheses go as far as to state that hypoperfusion can precede (or even induce) other key pathological events in AD (Niedermeyer, 2006; Hays et al., 2016; de la Torre, 2018). Moreover, cerebral blood flow dysregulation is possibly caused by functional changes, rather than a result of structural vascular abnormalities (e.g., atherosclerosis, cerebral amyloid angiopathy) (Kelleher and Soiza, 2013; Love and Miners, 2016b). These changes take place at the level of the neurovascular unit, which adjusts the vascular tone so as that blood supply matches energy demand in the brain (a process named neurovascular coupling). In AD, a malfunctioning neurovascular unit fails to adequately regulate cerebral blood flow and weakens the blood-brain barrier (Benarroch, 2007; Zlokovic, 2011; Iadecola, 2017; Kisler et al., 2017).

If AD pathophysiology is actually influenced by vascular pathology and neurovascular unit dysfunction, then systemic RAS is likely an important player, including its alternative axis (Kangussu et al., 2020). Acting upon the Mas receptor, Ang-(1–7) mediates the alternative axis’ anti-inflammatory, anti-oxidative and vasodilatory properties (Santos et al., 2018). Evidence from preclinical studies suggest that Ang-(1–7) is especially important in brain response to ischemia-hypoxia, increasing cerebral blood flow and preventing blood-brain barrier breakdown (Lu et al., 2008; Zhang et al., 2008; Wu et al., 2015). In animal models of chronic cerebral hypoperfusion, Ang-(1–7) induces tolerance to ischemia and improves cognitive function (Jiang et al., 2014; Xie et al., 2014). Ang-(1–7) has also been studied in mice models of AD. In SAMP8 mice, Ang-(1–7) was reduced and inversely correlated with Tau hyperphosphorylation (Jiang et al., 2016b). When constantly given to SAMP8 mice, Ang-(1–7) counteracted Ang II and prevented cognitive decline (Cao et al., 2019). In Tg2576 mice, upregulating RAS alternative axis reduced amyloid pathology and restored cognition (Evans et al., 2020). It is worth mentioning that no mice model credibly reproduces all the key features of sporadic AD (LaFerla and Green, 2012; Neha et al., 2014). Hence, pathological inferences from animal studies should be interpreted with caution. With such caveat in mind, we can state that preclinical data support the hypothesis of an Ang-(1–7) downregulation contributing to AD.

Against this background, we risk extrapolating our findings to hypothesize that AD patients produce less Ang-(1–7), which may contribute to their disease by diminishing the magnitude of Ang-(1–7) neuroprotective effects. We also speculate that, if confirmed, the positive correlation between plasma levels of Ang-(1–7) and cerebrovascular lesions in AD might result from some sort of response mechanism: for instance, Ang-(1–7) could be upregulated in an effort to counteract the underlying cerebrovascular disease and increase tolerance to ischemia. We assume that such attempt, however, would not raise Ang-(1–7) concentrations to the same levels seen in healthy individuals.

We are aware of this study’s many limitations, starting with the sample size. Besides reducing power and generalizability, having a small sample limited our capacity to adjust results for possible confounders (e.g., hypertension, ACEi or ARB use). To minimize the chance of spurious results, we tried to keep groups balanced concerning variables that might interfere. We still recognize that when few subjects are analyzed, statistical positives can arise only by chance. Regarding the reduced Ang-(1–7) in AD, however, this possibility seems less likely in light of existing data (Jiang et al., 2016a). We also acknowledge that the study would benefit from having a second control group, ideally of patients with another dementia (e.g., vascular dementia). If such third group had been added, it would help to determine whether our findings are specific of AD or common to different dementias. Moreover, one of the inherent disadvantages of the cross-sectional design, not being able to establish causality constrained our attempts to explain mechanistically our main results. Likewise, the lack of histopathological data restricts the consistency of pathophysiological inferences we made. It should also be noticed that in this study, even the biochemical assessment of the RAS pathways was far from complete. To understand why AD patients have lower Ang-(1–7), we would have to look at the protein that produces it, angiotensin-converting enzyme 2 (ACE2). If ACE2 activity was found to be reduced in AD patients, it would explain the immediate mechanism behind their lack of Ang-(1–7). It would also be useful to measure ACE2 concentration together with its activity. Especially if the correlation between ACE2 levels and activity was strong, dosing both would guide future studies about which assays to perform (Chappell, 2015). Despite these limitations, we believe that, for an exploratory study, our methods were appropriate and achieved the goal of generating hypotheses about RAS-AD interaction.

In conclusion, our study strengthens the hypothesis that RAS alternative axis is downregulated in AD. It also points to a possible interaction between Ang-(1–7) and cerebrovascular lesions in AD patients. We hope these hypotheses will be addressed in the future by larger studies, with longitudinal follow-up and a more comprehensive assessment of the RAS molecules. We believe that, as AD pathogenesis remains largely unelucidated, it is important to follow every lead that may help to explain the disease. If confirmed, our findings corroborate the view that the RAS is a possible player in Alzheimer’s disease pathophysiology.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Comitê de Ética em Pesquisa da Universidade Federal de Minas Gerais. The patients/participants provided their written informed consent to participate in this study.

VR, AS, and LS designed the study and wrote the protocol. TC proposed the neuroimaging protocol and conducted the MRI analysis. LP and RF planned and conducted the biomarker assays. LS, PC, and AT enrolled participants and performed neurological evaluation. VR undertook the statistical analysis, reviewed by AS and LS. All authors contributed to and have approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.