The Association of Kidney Function with Plasma Amyloid-β Levels and Brain Amyloid Deposition

INTRODUCTION

Patients with kidney impairment exhibit multiple features of accelerated brain aging [1]. Chronic kidney disease (CKD) is strongly related to brain atrophy [2] and an elevated risk of all-cause dementia and Alzheimer’s disease [3]. Up to 50% of patients with mild to moderate chronic kidney disease have cognitive impairment, and some studies have estimated this proportion to be as high as 70% in patients with end stage kidney disease [1, 4]. This association is not limited to advanced kidney impairment, as even mild to moderate degrees of kidney impairment are related to subclinical brain pathologies and decline in cognitive function [5, 6]. Despite the compelling evidence that individuals with kidney dysfunction are at a higher risk of cognitive impairment, the exact mechanisms behind this association remain unknown. A role for shared cardiovascular risk factors, such as hypertension, diabetes, and hyperlipidemia, has been implicated [1, 7]. However, multiple studies have demonstrated that kidney dysfunction is associated with cognitive impairment independent of these factors [6, 8–10], suggesting that other pathologic pathways such as neurodegenerative changes may play a role.

The kidney controls levels of many blood metabolites via its filtering and excretory functions [11, 12]. Several studies have suggested a role for the kidneys in clearance of plasma amyloid-β [11, 13–17]. Serum amyloid-β levels and brain amyloid-β deposition were found to be higher in patients with advanced kidney impairment compared to healthy controls, which might suggest that lowered kidney-mediated amyloid-β clearance can contribute to Alzheimer’s disease pathology in the brain [11, 14–16]. Liu et al. showed higher levels of serum amyloid-β levels in CKD patients compared to controls. Another study including CKD patients stages 3-5 and a control group found a stage-dependent plasma amyloid-β increase across CKD stages [18]. However, a secondary analysis of a randomized clinical trial including persons with memory complaints but free of dementia did not find an association between mild to moderate decline in the estimated glomerular filtration rate (eGFR) and cortical amyloid-β deposition [19]. There is lack of evidence to determine to what extent mild to moderate degrees of kidney impairment in a population-based setting are associated with brain amyloid deposition and levels of plasma amyloid-β, and whether the strength of this association varies between two measures of kidney function: the estimated glomerular filtration rate (eGFR) and urine albumin-to-creatinine ratio (ACR). eGFR is generally considered a measure of kidney filtration and decreased eGFR can be due to both vascular and nonvascular etiologies, whereas albuminuria, or higher ACR, is believed to be reflective of kidney damage or glomerular angiopathy secondary to microvascular endothelial inflammation, and frequently precedes a decline in eGFR [20]. We aimed to investigate the association of kidney function measures with both brain amyloid deposition, measured by florbetapir positron emission tomography (PET), and plasma amyloid-β levels in the community-based Atherosclerosis Risk in Communities (ARIC) Study cohort.

METHODS

RESULTS

Table 1 presents overall characteristics of the participants in both the amyloid PET study and plasma amyloid-β components. Participants with PET data had mean age of 76±5, 43% were male, 42% Black adults and had average eGFRcrcys of 66.5±17.6. Participants with plasma amyloid-β data had mean age of 77±5, 43% were male, 23% Black adults and had average eGFRcrcys of 63.0±18.4 (Table 1). Participants’ characteristics according to their kidney function status and ACR and eGFRcrcys strata are presented in Table 2 and Supplementary Table 1.

In the crude model, doubling in ACR was associated with a 1.17 (95% CI: 1.09, 1.27) higher odds of brain amyloid positivity (Table 2). Adjustment for potential confounders attenuated the estimates (1.10 (95% CI: 1.01, 1.20)). In the crude model, 15 ml/min/1.73 m² lower eGFRcys was associated with 1.17 (95% CI: 1.07, 1.29) higher odds of brain amyloid positivity, but the association did not reach statistical significance after adjusting for potential confounders. eGFRcr and eGFRcrcys were not statistically significantly associated with brain amyloid positivity (all p-value>0.05).

After adjusting for potential confounders, doubling in ACR was associated with higher plasma amyloid-β_1 - 40 (standardized difference: 0.12; 95% CI: 0.09, 0.14), higher amyloid-β_1 - 42 (standardized difference: 0.08; 95% CI: 0.05, 0.10), and lower amyloid-β _{1 - 42/1 -40} ratio (standardized difference: 0.04; 95% CI: -0.06, -0.01). Lower levels of eGFRcr, eGFRcys and eGFRcrcys were associated with a higher plasma amyloid-β_1 - 40 and amyloid-β_1 - 42 and lower levels of amyloid-β _{1 - 42/1 -40} ratio (Table 2).

When ACR and eGFR were analyzed together using clinical cut points, there was evidence of dose-dependent relation between kidney impairment and plasma amyloid-β_1 - 40 and amyloid-β_1 - 42 (p for trend < 0.001), indicating higher plasma amyloid-β_1 - 40 and amyloid-β_1 - 42 levels in persons with worse kidney function (Fig. 1). For example, participants in the kidney damage-low function category (eGFR < 60 ml/min per 1.73 m² and albumin-to-creatinine ratio > 30 mg/g) and those in no damage-low function category (eGFR < 60 ml/min per 1.73 m² and albumin-to-creatinine ratio < 30 mg/g) had higher plasma amyloid-β_1 - 40 and amyloid-β_1 - 40 compared to participants in normal category (eGFR (>60 ml/min per 1.73 m² and albumin-to-creatinine ratio < 30 mg/g). In an extra analysis, we compared the effect estimates of chronological age with ACR and eGFRcrcys in relation to plasma and PET measurements of amyloid-β. Comparing the effect estimates of age with ACR in relation to amyloid positivity, we showed that a doubling in ACR may be equivalent to a 0.6-year increase in age. A doubling in ACR corresponds to higher amyloid-β_1 - 40, amyloid-β_1 - 42, and lower amyloid-β_{1 - 42/1 -40} ratio by amounts equal to 6.1, 6.5, and 3.6 years increase in age, respectively. Similarly, 15 ml/min/1.73 m² lower eGFRcrcys had an estimated magnitude of association equal to 1.2, 18.5, 27.8, and 6 years increase in age for brain amyloid positivity, amyloid-β_1 - 40, amyloid-β_1 - 42, and decrease in amyloid-β _{1 - 42/1 -40} ratio, respectively (Supplementary Table 2). Using eGFRcrcys calculated based on the equation without race yielded in similar findings (Supplementary Table 3). Adjusting the analysis additionally for HbA1c did not change the findings (Supplementary Table 4).

OPEN IN VIEWER

Fig. 1

Participants with worse kidney function have higher plasma Aβ_1 - 40, Aβ_1 - 42, and lower Aβ_{1 - 42/1 -40} ratio: The ARIC Study 2011-2013. Marginal means and SEMs of plasma Aβ_1 - 40, Aβ_1 - 42, and Aβ_{1 - 42/1 -40} ratio levels in different categories of kidney function. Normal category indicates eGFR (based on creatinine and cystatin C)>60 ml/min per 1.73 m² and albumin-to-creatinine ratio < 30 mg/g (n = 1,092). Kidney damage-intact function category indicates eGFR > 60 ml/min per 1.73 m² and albumin-to-creatinine ratio > 30 mg/g (n = 203). No damage-low function category indicates eGFR < 60 ml/min per 1.73 m² and albumin-to-creatinine ratio < 30 mg/g (n = 709). Kidney damage-low function category indicates eGFR < 60 ml/min per 1.73 m² and albumin-to-creatinine ratio > 30 mg/g (n = 300). Means were adjusted for age, sex, race-field center. *Indicates statistically significant different mean values compared to normal category.

DISCUSSION

In this community–based study, higher ACR was associated with brain amyloid positivity independent of cardiovascular risk factors. However, lower eGFR was not independently statistically significantly associated with brain amyloid positivity. Higher ACR and lower eGFR were both associated independently with higher levels of plasma amyloid-β_1 - 40 and amyloid-β_1 - 42 and lower plasma amyloid-β_{1 - 42/1 -40} ratio. When ACR and eGFR were analyzed together using clinical cut points, there was evidence of dose-dependent association between kidney function impairment and plasma amyloid levels. Our result is in agreement with multiple previous studies evaluating clearance of amyloid-β levels in kidney impairment [11, 13–16]. However, a secondary analysis of the Multidomain Alzheimer Preventive Trial (MAPT) randomized clinical trial including persons with memory complaints but free of dementia did not find an association between mild to moderate decline in the eGFR and cortical amyloid-β deposition [19]. Inclusion of relatively healthier participants with higher kidney function (eGFR: 72 compared to 66 in the current study) and less cardiovascular burden could be the reason for negative finding.

Plasma and PET measurements of amyloid-β have been increasingly used to improve the accuracy of the clinical diagnosis of Alzheimer’s disease and related dementias and to better understand the disease [17, 32]. Impaired clearance of amyloid-β from the central nervous system (CNS) is posited to play a central role in the pathogenesis of Alzheimer’s disease. Peripheral amyloid-β clearance is suggested to contribute to reducing amyloid-β accumulation in the brain [13, 17] which results in lower ratio of amyloid β_{1 - 42/1 -40} in both the CSF and plasma [33, 34]. Several studies have shown that kidneys may be involved in clearance of circulating amyloid-β [11, 14]. Different mechanisms such as interstitial injury in CKD, decreased renal tissue integrity, tubular dysfunction, and inflammation have been suggested as factors that can result in amyloid-β accumulation [18]. As such, low clearance of plasma amyloid-β in patients with kidney impairment can potentially result in further deposition of amyloid-β in the brain and ultimately lead to cognitive decline. In a radioisotopic experiment, injected intracranial or intravenous I₁₂₅-labelled amyloid-β was detected in the kidneys and urine suggesting the role of kidneys in clearance of amyloid-β [15]. In a chronic unilateral nephrectomy model in Alzheimer’s disease mice brain amyloid-β deposition and aggravated Alzheimer’s disease pathologies was observed [35]. Another study detected soluble amyloid-β in human urine [11]. Other studies investigating renal clearance of circulating amyloid-β found significantly higher circulating amyloid-β levels in CKD patients with low glomerular filtration rate compared to controls [11, 16]. Here, we extend the literature and show that levels of plasma and cerebral amyloid-β are higher even in mild-moderate kidney impairment. Dialysis is suggested to decrease plasma amyloid-β levels and consequently decrease the deposition in the brain [14]. Wang et al. measured plasma amyloid-β levels before and immediately after peritoneal dialysis in patients with chronic kidney disease and in Alzheimer’s disease mice models [36]. In both cases, plasma amyloid-β_1 - 40 and amyloid-β_1 - 42 levels were significantly reduced after dialysis which could be considered as an understudied benefit of dialysis.

In this study, we observed that lower kidney function is associated with both higher plasma amyloid-β levels and brain amyloid positivity. The temporality of this association should be tested in future longitudinal studies with multiple measurements of amyloid-β in both plasma and brain. Although lower eGFR was associated with higher levels of plasma amyloid-β_1 - 40 and amyloid-β_1 - 42, but not brain amyloid positivity, our sample size was small and larger studies with a wider range of severity of kidney function and brain amyloid burden are needed. It is possible that only more severe kidney impairment, e.g., eGFR < 60 or 30 would be associated with brain amyloid positivity as this level of kidney dysfunction is also associated with more severe cognitive impairment [37]. Our analysis using splines with knot at 60 (data not shown) suggested a direct association between lower eGFRcr less than 60 and higher amyloid positivity; albeit not statistically significant. This work also underscores the necessity to account for kidney function in plasma biomarker development and interpretation in future clinical trial enrollment or clinical diagnostic models using amyloid-β measurements [38]. A limitation of our study is that urinary albumin and creatinine were based on a single spot urine sample rather than 24-h urine sampling, which may have led to random misclassification of the exposure and potential underestimation of the true association between kidney dysfunction and amyloid-β levels.