The Alzheimer’s Disease Amyloid-Beta Hypothesis in Cardiovascular Aging and Disease

1Several cardiovascular risk factors have long been associated with a greater risk for future cognitive decline in nondemented individuals. Control of vascular risk factors effectively reduces the incidence of dementia in both healthy and cognitively impaired individuals. The presence of intracerebral atherosclerotic vascular disease exacerbates all types of dementia and has been independently associated with worse cognitive performance even in nondemented individuals. These observations indicate that the aging-related inflammatory nature of both atherosclerosis and dementia involves multiple common cellular and molecular mechanisms. Recent accumulating evidence points toward the existence of a possible nonexclusive shared systems biology process that may drive aging-associated diseases, atherosclerotic cardiovascular disease (CVD), and dementia (Figure 1).

2Production and accumulation of amyloid-beta (Aβ) peptides in the brain are considered the hallmark of Alzheimer’s disease (AD) amyloid hypothesis. The prototypic cerebrovascular disease associated with Αβ40 deposits is cerebral amyloid angiopathy (CAA). CAA describes a group of aging-associated brain disorders with characteristic pathological findings of amyloid deposits predominantly in the arteriolar wall. Clinical and imaging features of CAA vary from asymptomatic microbleeds to severe hemorrhage, neurological deficits, cognitive impairment, dementia, and death. Defective perivascular drainage of neuronal-derived Aβ is probably the main mechanism of Αβ deposition. Among Aβ peptides, Αβ1-40 is the main peptide involved in the pathogenesis of CAA, whereas Αβ1-42 is mainly involved in development of AD. The vascular preference of Aβ1-40 has led to the hypothesis that this molecule may exert proinflammatory properties not only in cerebral but also in peripheral vasculature, mediating arterial disease as depicted in Figure 1, suggesting a continuum of Aβ1-40 deposits in the circulatory system ranging from leptomeningeal and cortical cerebral microvasculature (CAA) to intracerebral, carotid, aortic, or coronary vascular wall or heart. Interestingly, in contrast to studies examining associations between Aβ1-40 plasma levels and cardiovascular disease, studies assessing the association of plasma Αβ1-40 with cognitive function have not yielded consistent results. The detrimental properties of Αβ1-40 species on vascular brain pathology affecting memory and cognition secondarily to microvasculature damage rather than through direct neurotoxicity, may explain this discrepancy.

3In this review, we present contemporary evidence that links Αβ peptides with vascular inflammation and a wide range of associated extracerebral atherosclerotic manifestations and myocardial dysfunction, as well as adverse CVD outcomes and mortality (Central Illustration). Based on this evidence, we discuss the potential clinical utility of Αβ1-40 as a biomarker for risk stratification for mortality and present therapeutic interventions that may alter Αβ accumulation.

4Aβ peptides are proteolytic fragments of amyloid precursor protein (APP), an integral membrane protein. The APP gene produces 3 major splice variants, APP695, APP751, and APP770, produced in neurons, endothelial cells, and platelets, respectively. The exact physiological function of this well-conserved, site-specific APP/Αβ pathway is not fully elucidated, but it is associated with natural antimicrobial defense and coagulation cascade proteolytic events. The latter is mediated by a Kunitz-type serine protease inhibitor domain contained in APP751 and APP770 molecules.

5APP can be initially cleaved by α-secretases generating nonamyloidogenic products depending on its location on plasma membrane, the site of processing (membrane or endosomes), and environmental pH, or by β-secretases, also known as beta amyloid cleaving enzymes (BACE) (Figure 2). The β-secretase–mediated cleavage of APP retains the integrity of Αβ fragments within the remaining C99 peptide, while C99 subsequent cleavage by γ-secretases releases Aβ peptides. C99 cutting site by γ-secretases depends on the location of processing (endosomes or Golgi network) and generates amino acid peptides of length 40 (Αβ1-40 mostly found in vascular lesions) and 42 (Αβ1-42, mainly found in AD-associated brain lesions), as well as the intracellular domain of APP (Figure 2). Several factors, including aging, inflammation, renal dysfunction, ischemia, polymorphisms, and drugs, increase circulating levels and subsequent tissue deposition of Αβ by augmenting APP production and processing or by decreasing Αβ clearance and degradation (Figure 2, Online Tables 1 to 3). Under normal conditions an equilibrium exists between Aβ production and removal in various compartments inside or outside of the central nervous system. Deregulation of this equilibrium may lead to accumulation of Αβ1-40 in blood, vascular wall, and heart tissues, which has been associated with CVD.

6Although APP processing in different cell types gives rise preferentially to Αβ1-40 or -42, it is not known what drives this differential final processing of the amyloidogenic pathway of APP. In cases of CAA, neuronal-derived Αβ (either Αβ1-40 or -42) fails to drain away from the leptomeningeal vessels, capillaries, and brain parenchyma. This defective depletion leads to its accumulation in brain arterioles. Αβ deposits are observed in the tunica media in close proximity as well as inside of the smooth muscle cells and in the adventitia, avoiding endothelial cells even at higher degrees of CAA. Because impairment of adventitial lymphatic capillaries in peripheral vessels also aggravates atherosclerosis, the role of lymphatic drainage in Aβ-related cardiovascular disease should be further explored. In peripheral atherosclerotic lesions, Αβ deposits consist almost exclusively from the Αβ1-40 species. Using mass spectrometry, Aβ1-40 peptide was on average 100 times more abundant than Aβ1-42 in human aortic atherosclerotic plaques. The 2-peptide-amino-acid-longer species Αβ1-42, being more hydrophobic and fibrillogenic, is the main amyloid peptide found in parenchymal lesions of AD; however, its “vascular” involvement is limited to deposits in pericapillary spaces and glia limitans, parenchymal brain vessels, and leptomeningeal vessels. Yet, overexpression of Αβ1-42 promotes Αβ1-40 vascular depositions in the brain, and factors that alter the Αβ1-40/-42 ratio, such as human apolipoprotein E4, favor amyloid deposits in the form of CAA compared with parenchymal plaques. This differential tissue preference of Aβ species may be explained by the following observations: 1) using 3D models of cerebrovascular vessels, researchers have recently demonstrated that HDL and apolipoprotein E (ApoE) synergistically promote vascular clearance of Aβ1-42 more than that of Aβ1-40; 2) Αβ1-40 is produced in significant amounts from platelets, plaque invading macrophages, endothelial cells, and vascular smooth muscle cells; and 3) different ApoE isoforms, which are proteins with an impact in cholesterol transport system, seem to differentially regulate Aβ production, aggregation, and clearance. More specifically, ApoE4 may inhibit Aβ clearance by competitively binding to the low-density lipoprotein receptor-related protein 1, and its presence has been associated with brain Αβ accumulation and increased AD risk. Interestingly, ApoE seems to affect also Αβ kinetics in blood.

7Aβ1-40 is critically involved in vascular aging. SIRT1, a class III histone deacetylase, plays a pivotal protective role in vascular aging as it up-regulates α-secretase activity shifting Aβ metabolism towards the non-amyloidogenic pathway (Figure 2). However, activation of the amyloidogenic pathway results in impairment of the vasodilating properties of small arterioles by enhancement of endothelin-1 expression, reduction of eNOS activity and endothelium-dependent vasodilation, enhancement of oxidative stress, and increased responsiveness to vasoconstrictors (Table 1, Figure 3). Further, Aβ oligomers may inhibit telomerase activity leading to telomere shortening, which actively promotes vascular aging. This experimental evidence generates the hypothesis that increased Aβ systemic concentrations may be associated with measurable, accelerated arterial aging and deteriorated vascular function and structure in humans. Arterial pulse wave velocity is a well-established, noninvasive marker of arterial stiffness and vascular aging. Interestingly, the severity of cerebral β-amyloid deposition measured by positron emission tomography scan and its change over 2-year follow-up was associated with higher pulse wave velocity in nondemented elderly adults. To assess whether Aβ1-40 is involved in early processes of arterial disease and aging, we prospectively examined changes in pulse wave velocity and plasma Aβ1-40 in 107 young to middle-aged healthy adults (mean age 46.2 years), clinically followed for 5 years. We found that the 5-year change of plasma Aβ1-40 levels was an independent determinant of the 5-year change in aortic stiffness. Because Aβ1-40 deposits have been found in carotid human atherosclerotic plaques and aortas, we examined whether plasma Aβ1-40 levels are associated with subclinical atherosclerosis in a population of 394 individuals with a wide range of CVD risk profiles. After adjustment for age, traditional CVD risk factors, and renal function, increased Αβ1-40 was independently associated with higher carotid intima-media thickness, lower ankle-brachial index, and the severity and extent of arterial damage assessed in the carotid and femoral arteries, aorta, and coronary circulation. Plasma Aβ1-40 was also associated with the severity of coronary artery calcium score in a sample of 3,266 adults from the Dallas Heart Study without clinically overt CVD.

8Overall, these findings are indicative of direct and indirect roles of Aβ1-40 in accelerated arterial aging, atherosclerosis at various stages, and vascular beds, taking place long before the establishment of clinically overt CVD.

9Circulating Aβ1-40 levels were independently associated with the presence of angiographically documented stable coronary artery disease (CAD) in 2 independent cohorts consisting of 514 and 396 patients. This association was confirmed in subsequent studies, including adults with normal cognitive function or patients with AD.

10Experimental evidence indicates that Aβ peptides may be actively involved in downstream pathways leading to plaque rupture, thrombosis, and subsequent clinical manifestations of the acute coronary syndrome (ACS) (Figure 3). Αβ1-40 stimulates platelet activation and adhesion in humans and mice (Table 1) and induces release of matrix metalloproteinases by human monocytes to increase plaque vulnerability. Interestingly, in a myocardial infarction rat model, early surges in plasma sAPP770 concentrations preceded the release of cardiac injury enzymes, while plasma sAPP was also increased in patients with ACS, suggesting that enhanced APP/Aβ processing and subsequent release of sAPP770 and Αβ1-40 may trigger plaque rupture or its sequalae in ACS. In support of this hypothesis (Figure 3), we recently reported that in 2 independent cohorts of patients with non-ST-segment elevation ACS, higher blood Aβ1-40 levels were associated with worse risk profile, including a higher GRACE (Global Registry of Acute Coronary Events) score high sensitivity cardiac troponin T and lower systolic blood pressure and estimated glomerular filtration rate, implying a concentration-dependent relation of Aβ with the severity of ACS. Overall, the results of these studies provide conceptual proof that Aβ metabolism is enhanced in CAD and Aβ1-40 levels in blood are increased and associated with its clinical presentation.

11High plasma Aβ1-40 concentrations were independently associated with increased risk of mortality in 1,254 elderly subjects after adjustment for CVD risk factors and frailty. However, significance was lost after adjustment for cystatin C, suggesting that this association may be mediated by differences in renal function and/or inflammatory status. The prognostic value of circulating Aβ1-40 in nonelderly subjects from the general population as well as its reclassification potential remain unknown.

12We have recently shown that circulating Aβ1-40 blood levels measured in 2 independent populations of patients with stable CAD were predictive of a 3-fold increased risk of cardiovascular death for highest versus lowest quartile. Importantly, adding Aβ1-40 improved risk stratification over the best predictive model by reclassifying 22% of the population to correct risk categories for cardiovascular mortality.

13In-hospital and midterm mortality in patients with ACS vary considerably from <1% to >8% according to risk score calculators. However, no indexes of vascular inflammation are currently included in risk estimation scores such as the widely recommended GRACE score assessing mortality. To this end, we have demonstrated that measuring Aβ1-40 in patients with non–ST-segment elevation ACS improves prognostic assessment and provides incremental reclassification value over the GRACE score. A single measurement of circulating Aβ1-40 at presentation was independently associated with mortality in both cohorts. Importantly, Αβ1-40 substantially improved risk stratification of patients with non–ST-segment elevation ACS into correct risk categories over the GRACE score (net reclassification index 33.4% to 47.1%).

14Collectively, these findings suggest that Αβ1-40 may be a clinically useful risk biomarker in stable CAD and particularly in non–ST-segment elevation ACS where Aβ1-40’s performance was complementary to that of the GRACE score, a commonly used risk score in clinical practice. However, clinical application of this peptide as a biomarker needs further research to set reference values and thus allow its investigation as part of novel prognostic algorithms in CAD.

15A deregulation of the BACE1/Αβ1-40 axis was identified in the hearts of nondemented individuals with ischemic heart failure, whereas histology confirmed Aβ1-40 and -42 aggregates in the heart of patients with AD, suggesting a novel form of aging-related cardiac amyloidosis that merits further investigation. Mechanistically, both peptides exert toxic effects on cardiomyocytes resulting in poorer cell viability and apoptosis. Treatment of zebrafish embryos with Aβ1-40 peptides induces impaired vascular development and angiogenesis, possibly by interfering with VEGF pathway. Because ischemia promotes both APP up-regulation and cleavage, and Aβ1-40 may induce vasoconstriction and reduced endothelium-dependent vasodilatation, the pathogenic consequences of short- or long-term myocardial ischemia on heart failure via enhanced cardiac amyloidogenesis should be explored.

16Many aspects of Aβ-related cardiac amyloidosis are supported by clinical findings. Plasma Αβ1-40 has been associated with markers of cardiac dysfunction in several clinical conditions with variable degrees of myocardial functional impairment. We have recently demonstrated that in 3,266 individuals without clinically overt CVD from the Dallas Heart Study who underwent cardiac magnetic resonance imaging, plasma Αβ1-40 was associated with increased circulating N-terminal pro–B-type natriuretic peptide and high sensitivity cardiac troponin T, indicative of involvement of this peptide in early subclinical myocardial stretch and injury. Interestingly, we also found an association of Aβ1-40 with lower left atrial emptying fraction after adjustment for CVD risk factors. In contrast, although stroke volume index was lower at higher levels of Aβ1-40 by univariate analysis, we observed no independent associations with more advanced cardiac abnormalities such as left ventricular systolic dysfunction or remodeling, possibly because the population under study was free of established heart disease and such late changes were not discernible. Indeed, increased plasma Αβ1-40 was found in patients with established CAD and lower left ventricular ejection fraction. Given that Aβ1-40 is associated with lower cardiorespiratory fitness (VO2 max) independently of daily activity and with left atrial dysfunction, further studies are needed to assess whether lower VO2 max is of cardiac origin possibly related to Αβ1-40–mediated subclinical myocardial damage. Accordingly, the presence of Aβ1-40 in the heart has been associated with echocardiographic findings of early diastolic dysfunction. Furthermore, in a prospective study of 939 patients with heart failure showing reduced or preserved ejection fraction, plasma Αβ1-40 levels were associated with symptoms of heart failure as described in New York Heart Association’s functional classification system. Because diastolic dysfunction and heart failure with preserved ejection fraction are considered prominent manifestations of myocardial aging, blood concentrations of Αβ1-40 may reflect the extent of its vascular and myocardial involvement in CVD. The clinical relevance of this concept is supported by recent findings showing that circulating Aβ1-40 predicts adverse clinical outcomes and mortality and improves risk stratification in patients with heart failure.

17A dementia-CVD continuum hypothesis is further demonstrated through the vascular involvement of dementia-prone transgenic APP mice. The Tg2576 mouse model expresses 5 times the levels of endogenous murine APP and shows progressive impairment of cognitive function together with Αβ1-40–dependent and ROS-mediated endothelial dysfunction, impaired vascular reactivity, and 30% attenuation in cerebral blood flow. B6Tg2576 mice develop more extensive aortic lesions than control mice when fed the same atherogenic or normal diet under similar lipid profiles. APP23 mice, which overexpress APP and Αβ1-40, show enhanced platelet integrin activation and degranulation as well as accelerated thrombus formation. Dementia-prone APP23 mice crossed with atherosclerosis-prone apolipoprotein E–deficient (ApoE−/−) mice develop larger and more inflammatory aortic atherosclerotic lesions compared with ApoE−/− mice. Conversely, ApoE−/− mice crossed with animals lacking APP (APP−/−) have significantly reduced atherosclerotic plaque size in thoracic and abdominal aorta (90% and 75% reduction, respectively) compared with ApoE−/− mice despite comparable cholesterol levels. More importantly, atherosclerotic plaques in APP−/−/ApoE−/− mice have reduced macrophage content, increased amount of collagen, and a thicker fibrous cap indicating a more stable plaque morphology. Mechanistically, a series of experimental studies summarized in Table 1 present Αβ as a potent proinflammatory, proapoptotic, and proatherogenic molecule affecting the function of endothelial cells, platelets, vascular smooth muscle cells, and macrophages (Figure 3).

18A healthy lifestyle, including adherence to Mediterranean diet, omega-3 fatty acids, and caloric restriction may reduce Aβ brain deposits and exert antiamyloidogenic properties (Online Table 4). We recently demonstrated that increased daily activity assessed by accelerometer recordings and lower physical fitness, as assessed by VO2 max, in 3,266 participants without CVD from the Dallas Heart Study were independently associated with plasma levels of Aβ1-40. However, changes of Aβ peptide blood levels over time in response to physical activity have not been assessed. Yet, accumulating evidence suggests that an unhealthy lifestyle such as a high-fat diet and cigarette smoking may enhance the amyloidogenic pathway (Online Table 4). These findings suggest that cardiovascular effects of lifestyle modifications may be partly mediated by altering Aβ metabolism, but further research should explore these effects in humans, particularly with regards to Aβ1-40 as a direct effector molecule in cardiovascular disease.

19Experimental evidence indicates that statins reduce brain and intracellular Aβ levels in vitro and in vivo, by down-regulating its upstream pathway, reducing cellular uptake of Aβ peptides, and enhancing its clearance through the blood brain barrier (Table 2). However, results of 2 randomized clinical studies evaluating blood Αβ1-40 peptides after statin treatment were inconsistent, possibly due to statins’ effect on equilibrium between brain and circulating Αβ (Table 2).

20Most classes of antihypertensive drugs used in clinical practice influence APP/Aβ metabolism (Table 3). Inhibition of the angiotensin-converting enzyme increases Aβ1-40 or Aβ1-42 availability due to attenuation of its breakdown or through blockade of Aβ1-42 conversion to Aβ1-40, respectively. Consequently, plasma levels of Aβ1-42 were found to increase after angiotensin-converting enzyme inhibition, but results of Aβ1-40 levels were not consistent, showing either increase or no change (Table 3). The favorable effects of angiotensin receptor antagonists on Aβ metabolism shown in the central nervous system (Table 3) have not been investigated on the cardiovascular system in humans, similar to the effect of β-blockers, calcium-channel blockers, and diuretic agents (Table 3).

21A new heart failure drug class, the angiotensin receptor-neprilysin inhibitors, involves the inhibition of neprilysin, which is an Αβ degrading enzyme and thus may increase Aβ1-40 plasma levels (Table 3). In light of new evidence showing that Aβ1-40 blood levels are associated with increased mortality in patients with heart failure not receiving angiotensin receptor-neprilysin inhibitors and that Aβ1-40 is widely expressed in the myocardium of patients with heart failure, it remains unknown whether some beneficiary effects of angiotensin receptor-neprilysin inhibitors may be partly offset due to increased systemic Aβ1-40 availability. This may be particularly important in regard to long-term outcomes, as deposition diseases need time to evolve. Peritoneal dialysis and hemodialysis reduce plasma levels of Aβ1-40 and -42 (Table 3), supporting the significance of Αβ renal clearance indicating a definite interventional target on Aβ1-40 availability.

22Although some evidence indicates that at low concentrations, anticoagulant agents may increase Aβ metabolism, most experimental studies indicate that mainly due to their glycosaminoglycan structure, heparin and enoxaparin inhibit Aβ neurotoxic effects by affecting APP function and BACE1 activity (Online Table 5). However, whether these protective effects are extended systemically to the cardiovascular system merits further investigation. In contrast, 1 experimental study showed that treatment of C57BL/6 mice with anticoagulants greatly increased plasma levels of Aβ (>20-fold) through down-regulation of the factor XII–factor VII pathway, which is involved in Aβ degradation. Clopidogrel or aspirin may interfere with APP/Aβ generation from platelets, but further studies are needed to confirm this relationship (Online Table 5).

23Finally, although most phase III trials assessing antiamyloid-specific, targeted therapies were negative regarding efficacy in AD, their impact on CVD is unknown and merits further investigation.

24Several issues merit clarification. Although patients with CAD are more likely to develop AD-like neuropathological lesions than those without CAD, whether atherogenesis occurs in parallel or independently from brain parenchyma amyloid load in humans is unknown. In B6Tg2576 mice, brain Aβ load is positively correlated with the area of aortic atherosclerotic lesions, while APP23/ApoE−/− mice developed aortic atherosclerotic lesions well before any parenchymal brain depositions. The association between Αβ1-40 and normal or premature cardiovascular aging needs to be further elucidated. Understanding the mechanisms responsible for the vascular preference of Aβ1-40 over -42 can elucidate the precise biological role of this peptide in the complex pathophysiology of vascular inflammation.

25A pathophysiological role of Aβ1-40 across the continuum of cardiovascular disease is suggested through its independent association with a broad spectrum of vascular and cardiac involvement from early functional vascular alterations and subclinical atherosclerosis to overt symptomatic CAD, ACS, and heart failure. This is robustly supported by experimental evidence that APP and Aβ1-40 are critically involved in vascular inflammation, vascular and cardiac aging, and atherothrombosis. The association of Aβ1-40 with mortality has been consistently shown in a total population of about 5,000 patients in 6 independent cohorts derived from 8 countries. Thus, Aβ1-40 fulfills several criteria for consideration as a new biomarker for risk stratification in cardiovascular disease, including proof of concept, clinical utility, prospective validation, incremental and reclassification value for risk prediction, and ease of use. The implementation of a universally accepted method of sampling, preparation, storage, and measurement of circulating Aβ1-40 in plasma and the definition of normal and reference values as well as the conduction of studies with strict protocols of measurement in well-defined populations will allow the clinical application of this peptide as a new risk biomarker in patients with established cardiovascular disease. Interestingly, the association of Aβ1-40 with subclinical functional vascular alterations in healthy individuals and its association with all-cause mortality in the general population indicate that it should be further tested as a possible biomarker of cardiovascular risk in primary prevention as well. Most importantly, multiple lines of evidence clearly indicate that manipulating APP/Aβ turnover and aggregation or blocking its inflammatory reactions is feasible, potentially improving our understanding and means to simultaneously protect the brain, heart, and vessels during physiological or premature aging.