CKD appears to increase the risk of cardioembolic stroke predominantly through atrial fibrillation (AF). Population-based studies have confirmed a higher prevalence of AF among patients with CKD. For example, the prevalence of AF was up to 3 times higher in The Chronic Renal Insufficiency Cohort than in the general population (19). CKD triggers several mechanisms which may result in an increased risk for AF. Renin-angiotensin-aldosterone dysfunction, chronic inflammation, vascular calcification, and left ventricular hypertrophy all increase the risk of cardioembolic stroke through AF as its predisposing risk factor (20). Beyond AF, CKD also increases the risk of thromboembolic events. A meta-analysis reviewing 25 studies of patients with AF and end stage renal disease (ESRD) showed that the presence of severe kidney impairment doubles the risk of stroke (21). Another study showed that kidney dysfunction (defined as either reduced eGFR or proteinuria) is associated with a higher incidence of thromboembolism independent of other stroke risk factors (HR = 1.39; 95% CI = 1.31–1.71 for eGFR <45 and HR = 1.54; 95% CI = 1.29–1.85 for proteinuria) (22). Taken together, this evidence suggests that CKD increases the risk for developing AF as well as the risk of thromboembolic events in the context of AF, both of which increase the risk for cardioembolic stroke.

CKD is an established risk factor for atherosclerosis, and as such may increase the risk of large artery stroke (LAS) via promoting extra- and intracranial atherosclerosis. In the Japanese Suita Study of urban residents, CKD was associated with carotid artery stenosis [adjusted odds ratio (OR) = 3.16; 95% CI = 2.05–4.88] independent of hypertension (23). Similarly, in the Intervention Project on Cerebrovascular Diseases and Dementia, a community-based cohort study with 3,364 participants, individuals in the lower quartile of eGFR had the greatest increase in carotid intima-media thickness (2.4%; 95% CI = 2.0–2.7%), an ultrasound marker of carotid atherosclerosis (24).

Arterial stiffness represents another clinical measure proven to be independently predictive of fatal and non-fatal cardiovascular events and as such is a useful surrogate end point for cardiovascular disease outcomes (25). It reflects the functional and structural changes in the vascular wall and correlates with atherosclerosis of the large arteries often involved in LAS (26). Even mildly impaired renal function is associated with increased arterial stiffness and subsequent independent increase in stroke risk (27, 28). This association between renal impairment and large artery atherosclerosis results in an increased risk of atherosclerotic ischemic stroke.

In the CHOICE study, which evaluated stroke in dialysis patients, the overall stroke incidence was almost ten times the incidence in the general population, with large-vessel atherosclerosis found in 11% (29). In a Japanese study of 639 subjects, severe kidney dysfunction almost doubled the risk of the atherosclerotic stroke (OR = 1.81; 95% CI = 1.23-2.68) (30). A recent study which followed subjects after a transient ischemic attack (TIA) or minor stroke found that microalbuminuria was associated with recurrent events and significant internal carotid artery stenosis (OR = 3.4; 95% CI = 2.2–5.2) independent of other vascular risks factors (31).

There are few epidemiological studies which have specifically assessed the contribution of CKD to the SVD subtype of ischemic stroke, commonly referred to as lacunar infarction. The Cardiovascular Health Study reported a linear association between decreasing kidney function and prevalence of lacunar infarction (OR = 1.20; 95% CI = 1.09–1.32 for each SD of decreased cystatin C clearance) after multivariable adjustments (32). Similar ORs were also reported in the Rotterdam Scan Study for decreased eGFR (33). These findings were confirmed in a meta-analysis which showed a nearly three-fold increased risk of silent cerebral infarctions in patients with low eGFR (OR = 1.77; 95% CI = 1.36–2.11) (34). In contrast, in a retrospective observational study of 639 patients with stroke and ESRD, severe kidney dysfunction was associated with atherothrombotic stroke (OR = 1.81; 95% CI = 1.23–2.68) and cardioembolic stroke (OR = 2.25; 95% CI = 1.32–3.83) and showed an OR of similar magnitude for lacunar stroke, which however did not reach the level of statistical significance (OR = 1.67; 95% CI = 0.98–2.84) (30). While some studies may not have found an excess risk of lacunar stroke in CKD patients, associations have been reported for SVD neuroimaging features beyond lacunar stroke. The United Kingdom Young Lacunar Stroke DNA Study recruited 1,023 patients with lacunar infarction. In this study, decreased eGFR did not represent a risk factor for multiple lacunar infarcts vs. isolated lacunar infarcts, but increased the risk of moderate/severe white matter hyperintensity (WMH), a marker of chronic cerebral SVD (35). Similarly, a meta-analysis including 37 publications and 20,379 subjects calculated the risk of having renal impairment for patients with lacunar stroke compared to patients with non-lacunar stroke. Although no specific association between renal impairment and lacunar stroke was found (OR = 0.88; 95% CI = 0.6–1.30), the presence of SVD features on imaging was associated with worse renal function (36).

Available data of kidney function in patients with ESUS are limited and few studies assess the relationship between CKD and ESUS. In the largest ESUS dataset, patients with intact or impaired renal function had the same risk of suffering from ESUS (37). As such, whether kidney dysfunction could lead to ESUS or other forms of cryptogenic stroke remains uncertain.

CKD increases the risk of ICH. Up to 46% of patients presenting with ICH are also affected by CKD (38, 39). In a retrospective cohort study of 516,197 adults, individuals with CKD had almost double the RR of hemorrhage (including ICH) (RR = 1.9; 95% CI = 1.5–2.4) when compared to subjects with normal kidney function (21). A prospective population-based cohort study in China found that subjects with proteinuria had almost double the risk of ICH (HR = 1.90; 95% CI = 1.35–2.67) compared to those without. This association was confirmed after adjustment for established cardiovascular risk factors for proteinuria, but not for low eGFR (40). A cohort study conducted with more than 10,000 people with ESRD found a similar increase in risk of ICH when compared to the general population (41). Concordant findings were also reported in a population-based study from UK primary care (42). In contrast, a more recent retrospective cohort study in South Korea which analyzed almost 200,000 subjects did not find any association between eGFR and ICH (43).

The high comorbidity between ICH and CKD is in part attributable to the burden of hypertension. Around 30% of patients affected by CKD have elevated ambulatory BP with normal office BP (masked hypertension), and up to 40% suffer from resistant hypertension (44). Therefore, the impact on risk of ICH is likely substantial considering that patients with a higher than normal systolic BP had a 5.5 (95 % CI = 3.0–10.0) fold increased rate of ICH compared with normotensive individuals (45).

Other hypothesized mechanisms include metabolic derangements common in advanced renal disease, such as high serum phosphate and low serum calcium levels. Calcium and phosphorus are pivotal in the maintenance of cell function in the endothelium and vascular smooth muscle, hence their abnormality can result in endothelial dysfunction and impairment of cerebral autoregulation (46). In a prospective study of patients on dialysis, when comparing patients with different levels of phosphate, every 1 mmol/L increase in serum phosphate corresponded to a two-fold increased risk of developing ICH (HR = 2.07; 95% CI = 1.10–3.81) (47). Lastly, patients suffering from CKD may exhibit uremia, which affects the balance between platelet adhesion activators and inhibitors, resulting in a net activation defect. Uremia also reduces platelet receptor glycoprotein, leading to impaired platelet adhesion to the sub-endothelium (48). As a consequence, patients with advanced CKD are at increased risk of hemorrhage (49, 50).

Further, CKD can increase the risk of ICH via the use of AF-related anticoagulants. As discussed earlier, the incidence and prevalence of AF in patients with CKD are higher than in the general population. A systematic meta-analysis showed that AF in patients with ESRD increased the risk of stroke, ICH and mortality, although with a high degree of variability (21).

The “strain vessel hypothesis,” based on hypertensive vascular damage, has been suggested as a possible mechanism explaining the association between CKD and stroke (52). Kidney and brain share commonalities in terms of vascularization, as previously discussed. Both organs have vessels therefore referred to as “strain vessels,” as they are systems with low vascular resistance allowing continuous high-volume perfusion (53). These features make the kidney and the brain extremely vulnerable to hypertension. In fact, high BP not only impacts the histopathology of arterioles leading to a replacement of smooth muscle cells by lipohyalinosis, but also impacts the hemodynamics of large arteries mostly via arterial stiffness which in turn exacerbates the deleterious effects on strain vessels (54). As a result, these cerebral subcortical perforating arteries and renal juxtamedullary afferent arterioles lose their autoregulation and thus impair regional blood flow. Furthermore, damage to renal juxtamedullary afferent arterioles leads to glomerular hypertension and sclerosis as well as impaired downstream circulation in the vasa recta and medulla (55). The first causes a progressive loss of renal function, while the second affects the sodium balance with an overall summative effect on worsening systemic hypertension (55).

Albuminuria occurs when podocyte density decreases below a certain threshold. It represents a marker of glomerular barrier impairment, specifically glomerular endothelium damage. Given the similarities in the anatomy of the kidney and brain, it has been hypothesized that the endothelial dysfunction which occurs at the glomerular level and causes albuminuria happens simultaneously at a systemic scale (56, 57). In this sense the presence of albuminuria reflects a more generalized impairment of vascular endothelial function, and the vascular leakage which occurs as a consequence will ultimately impair tissue perfusion with resultant ischemia. In the Framingham Heart Study, any level of urinary albumin excretion was associated with progressively increased cardiovascular risk and mortality, independent of diabetes or hypertension (58).

Additional evidence is provided by research on derivatives of the aminoacid L-arginine, such as asymmetric dimethylarginine (ADMA) and symmetric dimethylarginine (SDMA). ADMA and SDMA are elevated in patients with kidney dysfunction as well as in patients with other SVD manifestations, independent of classical vascular risk factors (59). ADMA levels were found to correlate with the extent of WMH, risk of small vessel disease, and were found to be elevated in subjects affected by heritable forms of SVD such as Cerebral Autosomal Dominant Arteriopathy with Sub-cortical Infarcts and Leukoencephalopathy (CADASIL) (59). L-arginine derivatives, similar to albuminuria, may represent biomarkers of the same systemic endothelial pathology shared by the kidney and the brain.

Accelerated atherosclerosis has been described in patients with kidney dysfunction. Dyslipidemia is common in patients affected by CKD. As a result of proteinuria and lipoprotein transport impairment, CKD patients often have an increase in serum triglycerides, very-low-density lipoprotein, and low-density lipoprotein (LDL) cholesterol, all of which have significant atherogenic potential (60, 61). Hypercholesterolemia may also be worsened by the upregulation of 3-hydroxy-3-methylglutaryl CoA reductase in patients with CKD (62).

Modified lipoproteins from elevated LDL have been hypothesized to trigger innate immune reactions, which result in inflammation and accelerated diastasis of lipids in the intima of the vessel. Additionally, CKD has been described as a pro-inflammatory state, and enhanced levels of fibrinogen and matrix metalloproteinases have been variably described in this patient population. Inflammation is known to affect atherosclerotic lesions' stability and risk of rupture.

Instability and rupture of atherosclerotic carotid plaques have been found to be significantly higher in patients with CKD compared to patients with normal renal function in a retrospective study of patients undergoing carotid endarterectomy (83 vs. 52%, p = 0.001, and 59 vs. 36%, p = 0.039, respectively) (63). However, data showing that patients with CKD have higher percentages of calcification and lower collagenous content of carotid plaques when compared to subjects with normal renal function represent an important counterfactual observation, as heavily calcified plaque is typically considered less prone to rupture and cause ischemia (64).

Taking this evidence together, increased atherosclerosis in CKD patients is likely to arise in the setting of combined insults of dyslipidemia and accentuated inflammation (65, 66). Concomitantly, higher levels of factor VIII and von Willebrand factor, which have been described in CKD patients as compared to the normal population, may further increase the risk of thrombotic and atheroembolic events.

Closely related to atherosclerotic changes, even moderate stages of CKD have been associated with changes in the biomaterial and biological characteristics of the vessels, resulting in arterial stiffness (67). Most likely via impaired renal excretion of vascular toxins and metabolic derangements of the calcium homeostasis, kidney disease promotes vascular calcification and eventually arterial stiffness. A prospective study evaluating patients with mild to moderate CKD found that progression to ESRD was an independent determinant of carotid stiffness (HR = 2.48; 95% CI = 1.63–3.78) (68), with the above mentioned impact on stroke risk and cardiovascular events.

Cerebral vasculature is able to maintain stable cerebral blood flow (CBF) despite changes in BP. Impaired kidney function can lead to impaired cerebral autoregulation and subsequently make brain perfusion more strictly dependent on systemic BP (69, 70). In this scenario, patients with CKD are at risk for both cerebral hypoperfusion and hyperperfusion depending on systemic pressure. This disconnection between tissue perfusion and metabolic demand can be measured by deviation from the expected CBF. Demonstrating this autoregulatory failure, studies have reported that patients with CKD have both higher and lower CBF compared to patients with normal kidney function. In the Systolic Blood Pressure Intervention Trial, reduced kidney function was independently associated with higher global and white matter CBF (71). In the Rotterdam Study, lower eGFR was independently associated with lower CBF (0.42 ml/ min/100 ml decrease in CBF for each standard deviation of eGFR reduction) (72).

Impairment of CBF may be a repercussion of an effect of kidney disease on nitric oxide, which is crucial in vascular responsivity. As discussed above, kidney impairment increases ADMA and SDMA, which are inhibitors of nitric oxide synthase, and therefore have been implicated in vascular disease (73). Although limited by small sample size, studies have shown that doses of ADMA increase vascular stiffness and decrease cerebral perfusion in healthy subjects, and that increased concentrations of ADMA are associated with LAS and cardioembolic stroke (59, 74).

Lastly, CKD can also lead to CBF dysregulation and hence stroke through anemia. Anemia in fact is a well-known complicating feature of CKD, present in up to two thirds of subjects with severe kidney impairment (75). Concomitantly, anemia has been shown to increase the risk of stroke by almost 50%, possibly through impairment of oxygen delivery to tissue, blood supply and CBF (76). This relationship between anemia and CBF may explain the further increase in stroke risk observed in CKD patients with anemia, compared to CKD patients without anemia (11).

Chronic low-grade inflammation can be seen in patients at all stages of CKD. Levels of pro-inflammatory molecules Interleukin (IL)-6, IL-1 and Tumor Necrosis Factor (TNF)-α are higher even in patients in the early stages of CKD, and this inflammation appears to influence CKD progression. As CKD worsens, uremic toxins, indoxyl sulfate, and guanidino compounds enter the central nervous system and promote neuroinflammation via macrophage and microglia polarization toward a pro-inflammatory phenotype (77).

Patients with CKD suffer from impaired production of many antioxidant sources, such as glutathione peroxidases and mitochondrial superoxide dismutase (78), with commensurate increases in uremic toxins such as indoxyl sulfate, p-cresyl sulfate, advanced glycation end products, oxidized LDL, and activated Nicotinamide Adenine Dinucleotide Phosphateoxidase (79). In this context, it is perhaps unsurprising that levels of biomarkers for oxidative stress (such as malondialdehyde) have been found to be increased in patients suffering from CKD and cardiovascular disease (80). Finally, iron therapy, frequently required in advanced stages of kidney dysfunction as a result of chronic disease anemia, may also contribute to oxidative stress (81).

This oxidative stress and consequent inflammation have profound vascular effects and influence on cerebral blood flow. Free radicals increase endothelial permeability, platelet aggregation, and alter reactivity to vasodilators (82). Oxidative stress eventually contributes to endothelial dysfunction, arterial stiffness and development of atherogenesis as discussed above in patients with CKD (83).

The blood–brain barrier (BBB) is made of vascular endothelium, a specialized basement membrane, astrocyte foot processes, and pericytes. Due to this specialized barrier, access to the brain from the blood is highly regulated. Animal experiments have demonstrated impairment of BBB integrity in the setting of both acute kidney injury and CKD (84). This BBB disruption can lead to leakage of toxic compounds and proteins into perivascular tissues. The consequent edema, arteriolar stiffening and impaired vasoregulation can contribute to hypo-oxygenation, hypoperfusion, and ultimately ischemic changes. Each of these features has been identified in SVD and chronic cerebrovascular diseases (85, 86).

The aforementioned observational studies and many others have recognized the role of CKD in cerebrovascular disease and quantified the associations between the two. However, conventional analytical methods in observational research are limited in providing evidence for causal effects, and therefore insights about the mechanisms underlying the observed associations are constrained. In particular, it remains unclear whether both the kidney and the brain suffer concomitantly from similar injuries and ultimately perfusion failure or whether damage to one can lead to injury in the other. Human genetics can provide valuable information about the causal networks underlying CKD and cerebrovascular pathologies. Genetic variation is determined at conception and is therefore not influenced by confounding processes occurring later in life. Thus, studying genetic data can provide useful additional and orthogonal anchors to causality.

Conventional genetic research, genome wide association studies (GWAS) and post-GWAS research instruments, such as Mendelian randomization (MR), may clarify the complex relationship between the two diseases. In parallel with the axes inferred by epidemiological studies, genetic data may be used to test the following hypotheses: kidney dysfunction and cerebrovascular disease are the result of a single shared biological defect, as is the case in rare monogenic diseases (axis 1); kidney dysfunction and cerebrovascular disease share common pathologic pathways that contribute to both diseases (axis 2); kidney disease causally contributes to the risk of specific cerebrovascular pathologies that increase the risk of stroke (axis 3).