Circulating angiopoietin-2 and angiogenic microRNAs associate with cerebral small vessel disease and cognitive decline in older patients reaching end-stage renal disease

Abstract

Background

The prevalence of end-stage renal disease (ESRD) is increasing worldwide, with the majority of new ESRD cases diagnosed in patients >60 years of age. These older patients are at increased risk for impaired cognitive functioning, potentially through cerebral small vessel disease (SVD). Novel markers of vascular integrity may be of clinical value for identifying patients at high risk for cognitive impairment.

Methods

We aimed to associate the levels of angiopoietin-2 (Ang-2), asymmetric dimethylarginine and a selection of eight circulating angiogenic microRNAs (miRNAs) with SVD and cognitive impairment in older patients reaching ESRD that did not yet initiate renal replacement therapy (n = 129; mean age 75.3 years, mean eGFR 16.4 mL/min). We assessed brain magnetic resonance imaging changes of SVD (white matter hyperintensity volume, microbleeds and the presence of lacunes) and measures of cognition in domains of memory, psychomotor speed and executive function in a neuropsychological test battery.

Results

Older patients reaching ESRD showed an unfavourable angiogenic profile, as indicated by aberrant levels of Ang-2 and five angiogenic miRNAs (miR-27a, miR-126, miR-132, miR-223 and miR-326), compared with healthy persons and patients with diabetic nephropathy. Moreover, Ang-2 was associated with SVD and with the domains of psychomotor speed and executive function, while miR-223 and miR-29a were associated with memory function.

Conclusions

Taken together, these novel angiogenic markers might serve to identify older patients with ESRD at risk of cognitive decline, as well as provide insights into the underlying (vascular) pathophysiology.

INTRODUCTION

The number of patients reaching end-stage renal disease (ESRD) is increasing worldwide, while half of the new ESRD cases are patients ≥60 years of age [1, 2]. These older patients are at increased risk for adverse health outcomes [3], including impaired cognitive functioning [4], which decreases quality of life and independently associates with increased mortality and other adverse health outcomes [5, 6]. However, the underlying pathophysiological mechanisms of cognitive impairment in this group of patients are not fully understood.

Diseases affecting small arteries and arterioles in the brain, also known as cerebral small vessel disease (SVD), play an important role in the development of cerebrovascular disease and cognitive impairment [7, 8]. A history of vascular disease is a determinant of cognitive impairment in patients reaching ESRD (before receiving renal replacement therapy) [9, 10] and SVD has been associated with impaired cognition [11]. Being exposed to high-volume blood flow, the brain and kidney share a high vulnerability for endothelial dysfunction and vascular damage [12]. Within the brain this can subsequently result in cognitive impairment [13]. Furthermore, accumulation of uraemic toxins [such as asymmetric dimethylarginine (ADMA)] may cause cerebral vascular injury and lead to neurodegenerative damage in the brain affecting the cognitive domains of attention and speed [14]. Microvascular destabilization factors such as angiopoietin-2 (Ang-2), together with (circulating) angiogenic microRNAs (miRNAs), play a major role in mediating vascular injury or maintaining microvascular integrity and have also been linked to endothelial dysfunction in cerebral SVD [15]. As such, they may provide insights into specific deregulated molecular mechanisms [16, 17], but also represent promising markers of microvascular disease [17–20]. Assessing and understanding these determinants of cognitive functioning in older patients reaching ESRD may prove beneficial in treatment decisions and may have implications for prevention of further cognitive decline.

However, so far no studies have been performed on circulating levels of ADMA, Ang-2 and angiogenic miRNAs in the older ESRD population or their relationship with SVD and cognitive impairment in this population. The aim of this study was to determine the levels of ADMA, Ang-2 and angiogenic miRNAs and their association with brain magnetic resonance imaging (MRI) markers of SVD and cognitive impairment in older patients reaching ESRD.

MATERIALS AND METHODS

Patient cohort

The full design, methods and baseline characteristics of the Cognitive decline in Older Patients with End-stage renal disease (COPE) study have been published previously [10, 21]. The patients selected for this study were those patients for which a cardiac MRI was available. In brief, the COPE study is a prospective, multicentre cohort study based in the Netherlands comprising patients ≥65 years of age reaching ESRD [estimated glomerular filtration rate (eGFR)  ≤20 mL/min/1.73 m²] before receiving renal replacement therapy. Patients who were illiterate were excluded. As part of the pre-dialysis nephrogeriatric work-up, a physical examination, comprehensive geriatric assessment, brain MRI and neuropsychological testing were performed. Written informed consent was obtained from all patients. The study protocol was approved by the medical ethics committee (approval number NL46389.058.13) of all centres that participated [Leiden University Medical Centre, Haga Hospital (The Hague), Dialysis Centre Zoetermeer (Zoetermeer; part of the Haga Hospital) and Reinier de Graaf Groep (Delft)].

Renal care

eGFR was estimated using the Modification of Diet in Renal Disease or Chronic Kidney Disease Epidemiology Collaboration equation depending on the time of inclusion or protocol used in the different hospitals. Vascular or non-vascular cause of kidney disease was defined based on the European Renal Association–European Dialysis and Transplant Association primary renal diagnosis code, which was determined by the treating nephrologist. A history of vascular disease was defined as the presence of myocardial infarction and/or cerebral vascular incident (CVA) and/or peripheral vascular disease.

Neuropsychological testing

A standardized neuropsychological test battery was administered by trained geriatric or dialysis nurses. The tests were designed to assess different domains of cognitive functioning, including global cognition, visuoconstruction, memory, psychomotor speed and executive function, which have been successfully used in several study cohorts [22–24]. Global cognition was tested by the Mini-Mental State Examination (MMSE), which ranges from 0 to 30 points, with higher scores representing better cognitive performance [25]. Clock drawing was used to determine visuoconstructive abilities and executive function, with scores ranging from 0 to 14 points, where higher scores indicate better performance [26]. Memory was tested using the 15-word verbal learning test (15WVLT), where both immediate (total score after five trials) and delayed recall were used, with higher scores representing better function [27]. To test for long-term memory reproduction, we used the Visual Attention Test, with better function indicated by higher scores [28]. To test for psychomotor speed, the Letter Digit Substitution Test (LDST) was used. The number of correct substitutions in 60 s was used, with higher scores representing better function [29]. Executive function was assessed with the Trail Making Test A and B (TMT-A and -B), with lower scores representing better function [30]. To distinguish between cognitive flexibility and processing speed as the explanation for the test result, the score on the TMT-B was corrected for the score on the TMT-A. In addition, the Stroop Colour Word Test (SCWT) was used to assess executive function, with lower scores representing better function [31]. To distinguish between cognitive inhibition and processing speed as an explanation of this test result, the SCWT III score (interference card) was corrected for the SCWT II score (colour naming card). The 15WVLT, TMT-B and LDST were used for the association analyses as representative tests for the different neuropsychological domains of memory, executive function and psychomotor speed, respectively.

MRI

In 70 patients, a brain MRI scan was performed. Patients were excluded if metal was present in their body, if they suffered from claustrophobia, if no consent was given for MRI or MRI was not available at the particular medical centre. As previously published, the MRI protocol consisted of three-dimensional (3D) fluid-attenuated inversion recovery, 3D T1-weighted, T2-weighted turbo spin echo images, diffusion-weighted imaging and susceptibility-weighted imaging sequences [21]. White matter hyperintensity (WMH) volume and intracranial volume (in mL) were determined by an automated method (Quantib Brain, Quantib, Rotterdam, The Netherlands) after manual segmentation of infarcts and other pathologies. WMH volume served as a marker of cerebral SVD. The presence of lacunes and microbleeds was scored by a radiologist according to the Standards for Reporting Vascular Changes on Neuroimaging criteria [32]. In addition, we applied the Scheltens scale to evaluate WMH [33]. Of note, this method was chosen over the Fazekas scale because the Scheltens scale provides a more detailed visual scale.

Reference cohorts

To compare our miRNA measurements with healthy controls and patients with diabetic nephropathy (DN) as a ‘young’ CKD group, we used archival data on circulating miRNAs from a patient study that was set up at the Leiden University Medical Centre, as previously described [20].

Laboratory assessments

For miRNA analysis, plasma was harvested by centrifugation of ethylenediaminetetraacetic acid–anticoagulated blood for 10 min at 3000 rpm and subsequently stored at −80°C. At the same time, blood was collected for analysis of serum Ang-2 and ADMA. Both Ang-2 and ADMA were measured by enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN, USA) according to the instructions of the manufacturer.

RNA isolation and measurements

Total RNA was isolated from 200 µL of plasma followed by miRNA profiling for miR-27a, miR-29a, miR-126, miR-132, miR-137, miR-192, miR-223 and miR-326 using reverse transcription quantitative polymerase chain reaction (miRCURY LNA Universal RT microRNA polymerase chain reaction) performed by Exiqon (Vaedbek, Denmark). For the RNA extraction of plasma, two pre-mixed RNA spike-ins (UniSp2 and UniSp4), each at different concentrations in 100-fold increments, were added. This set of spike-ins was intended as an RNA isolation control and to ensure that the quality of the input RNA was sufficiently high for effective amplification. For the reverse transcription step, one spike-in (UniSp6) was added. This control was used to confirm that the reverse transcription and amplification occurs with equal efficiency in all samples, indicating similar and sufficient quality of the RNA. For normalization of the data we applied the median of the assays detected in all samples (n = 129 samples), as this was found to be the most stable normalizer. For comparison with a previous cohort [20] (Figure 1), data were normalized to miR-16. miR-16 was used since we observed a strong correlation between the median of all samples (the most preferable normalization method, but not applicable in our comparison to other cohorts) and miR-16 (r = 0.92, P < 0.0001), suggesting that miR-16 is suitable as a normalizer in this study. Moreover, in the original study that described the DN group and healthy controls (where we compared our results), we used miR-16 as a normalizer after careful evaluation of their levels in different samples and different groups and found it the most stable normalizer [20].

Selection of miRNAs

miRNAs to be assessed (miR-27a, miR-29a, miR-126, miR-132, miR-137, miR-192, miR-223 and miR-326) were selected based on their previously determined relation to vascular injury and cognitive dysfunction (Supplementary data, Table S1).

Statistical analyses

Patient characteristics are presented as mean with standard deviation (SD) when normally distributed. In the case of a skewed distribution, data are presented as median with interquartile range (IQR) or as number (n) with percentage (%). Function in the different cognitive domains (memory, executive function and psychomotor speed) is indicated as percentiles (mean with IQR). miRNA, ADMA and Ang-2 levels are stratified in tertiles and mean scores are presented as mean [standard error of the mean (SEM)]. Associations between miRNA, ADMA and Ang-2 levels as continuous variables with measures from the brain MRI measurements and different cognitive tests were assessed via linear regression analyses adjusted for age and gender. All analyses were carried out using SPSS version 23 (IBM, Armonk, NY, USA).

RESULTS

Patient cohort description

The clinical characteristics of the study population are presented in Table 1. The study population consisted of 129 participants with a median age of 75.3 years and 85 (65.9%) participants were male. At study enrolment, the mean eGFR was 16.4 mL/min (SD 4.3) and the mean decline in eGFR in the past 3 years was 9.2 mL/min (SD 7.3). In 80 (62.0%) patients the primary kidney disease had a vascular cause, mainly hypertension or diabetes mellitus. Fifty-nine participants (45.7%) had a history of vascular disease, defined by the presence of myocardial infarction and/or CVA and/or peripheral vascular disease.

In Table 1 we also report the performance on global cognitive function and different cognitive domains for this study population. The population had a median MMSE score of 28 out of 30 points (IQR 27–29). Nine patients had an MMSE score ˂25 (7.1%), a previously identified significant cut-off value to diagnose cognitive dysfunction [34]. The mean score on the immediate memory test (15WVLT) was 30.7 words remembered (SD 9.8). The mean score on the executive function test (TMT-B) was 178.1 s (SD 76.0). The mean score on psychomotor speed (LDST) was 21.8 correct substitutions (SD 6.8).

miRNA, Ang-2 and ADMA measurements in older patients reaching ESRD

To identify miRNAs that may associate with cognitive decline we selected miRNAs that were previously described to strongly associate with vascular injury and cognition (Supplementary data, Table S1; detailed description can be found in the Materials and methods section). As such, we selected miR-27a, miR-29a, miR-126, miR-132, miR-137, miR-192, miR-223 and miR-326 as well as vascular injury marker Ang-2 and ADMA. Table 2 summarizes these measurements in our cohort of 129 patients.

Comparison with younger healthy controls and CKD patients

The miRNA levels in our cohort were first compared with the miRNA levels measured in a healthy (young) control group (n = 19, mean age 44 years) as well as in a group of patients with DN (n = 21, mean age 44 years, mean eGFR 17 mL/min) from an earlier study [20] (patient characteristics are described in Table 3). For this comparison, data were available for miR-223, miR-126, miR-132, miR-27a and miR-326, as these were previously assessed (levels normalized to stable miR-16), and we found that the patients from this study have strongly altered miRNA levels as compared with both healthy young individuals and patients with DN (Figure 1; these graphs are presented with individual data points in Supplementary data, Figure S1). Specifically, we found lower levels of miR-223, miR-126 and miR-132, while miR-27a and miR-326 displayed higher levels in older patients, confirming previous reports that renal failure associated with vascular injury was linked to this change in circulating miRNA levels [20]. In addition, mean Ang-2 levels were 1901 and 5871 pg/mL in the healthy controls and patients with DN, respectively. Here, mean Ang-2 levels were 3608 pg/mL, indicating that Ang-2 levels in these older patients reaching ESRD are higher than normal, but not as high as in a group of (younger) DN patients. Next we compared the observed ADMA levels to reference values, indicating that 0.74 µmol/L is higher than observed in healthy individuals but comparable to other ESRD patient groups [35, 36].

Specific miRNAs and Ang-2 associate with cognitive function

Next we aimed to determine whether circulating miRNA, Ang-2 and ADMA levels associated with cognitive function in different cognitive domains. To that end we grouped the ESRD patients based on their ADMA, Ang-2 and miRNA levels in tertiles (Tertile 1, lowest levels; Tertile 3, highest levels) and plotted the cognitive test scores against the tertiles of these vascular markers (Figure 2; these graphs are presented with individual data points in Supplementary data, Figure S2). We found lower levels of miR-223 (P = 0.006) and higher levels of miR-29a (P = 0.011) significantly associated with worse memory function (15WVLT), while higher Ang-2 (P = 0.004) associated with worse psychomotor speed (LDST). In addition, we found several close to statistically significant associations; lower miR-27a associated with worse memory (15WVLT; P = 0.067) and executive function (TMT-B; P = 0.067), while higher Ang-2 (P = 0.057) and lower miR-326 (P = 0.08) associated with worse executive function (TMT-B).

Ang-2 associates with imaging parameters of cerebral SVD

Given that it was previously demonstrated that imaging parameters of SVD are associated with measures of cognitive dysfunction, we assessed the relation of circulating vascular markers with MRI parameters of SVD. Figure 3A indicates both the qualitative and quantitative data from 93 and 70 patients, respectively, for which MRI data were available. Microbleeds were present in 56 patients (60.2%); the presence of lacunes of presumed vascular origin was observed in 44 participants (48%). The Scheltens score of the white matter hyperintensities (WMHs) was on average 15.8. The median of the quantitative measurement of the WMH was 3.16 mL. We observed no relation of ADMA, Ang-2 or miRNA levels with microbleeds or lacunes (data not shown). In contrast, Ang-2 (P = 0.08, r = 0.21) and miR-29 (P = 0.02, r = 0.28) were both (close to) significantly positively correlated with the Scheltens WMH score. Moreover, Figure 3B (individual data points in Supplementary data, Figure S3) illustrates that when we analysed quantitative measures of WMH, while no relationship was found for miRNAs, Ang-2 levels significantly correlated with quantitative WMH scores on MRI (P = 0.016).

DISCUSSION

Our study in older patients with ESRD shows that five selected circulating angiogenic miRNAs (miR-223, miR-126, miR-132, miR-27a and miR-326) and Ang-2 display altered (unfavourable) levels as compared with younger CKD patients and healthy controls. In addition, higher levels of Ang-2 and miR-29a and lower levels of miR-223, miR-27a and miR-326 are associated with worse cognitive function (either psychomotor speed, executive function or memory). Furthermore, while no relation was found with microbleeds or the presence of lacunes, Ang-2 and miR-29a positively correlated with the Scheltens WMH score, while Ang-2 levels also related to quantitative WMH volume on MRI as a marker of cerebral SVD.

We observed that higher Ang-2 levels associated with worse cognitive function in the domain of psychomotor speed, but we also found trends towards an association with worse memory and executive function scores. Interestingly, Ang-2 was also shown previously to associate with cognition in maintenance haemodialysis patients [37]. Ang-2 is known to destabilize vessels via negative interference with Ang-1-mediated Tie-2 signalling, which disrupts pericyte–endothelial cell interaction [38], resulting in abnormal microvascular remodelling and blood flow [39]. Moreover, circulating Ang-2 levels and a disturbed Ang-2/Ang-1 balance have been previously demonstrated to strongly associate with vascular injury in the settings of CKD, diabetes mellitus, acute coronary syndrome and other diseases characterized by endothelial dysfunction and microvascular inflammation [38, 40, 41]. Given the observed associations of Ang-2 with WMH (as an imaging parameter of SVD) and cognitive function in our study, this suggests that vascular injury may indeed be an important determinant of poor cognitive function, while Ang-2 may also serve as a biomarker for cognitive impairment in this population. This is also emphasized by our finding that Ang-2 associated with quantitative measures of WMH of presumed vascular origin, which are a manifestation of SVD on MRI [32]. Indeed, the association between WMH and cognitive dysfunction (among others with executive function) has been well-established [11, 42, 43]. However, we also find that Ang-2 levels are higher in (younger) DN patients, indicating that the suitability of Ang-2 as biomarker depends on patient characteristics and needs to be carefully evaluated.

Regarding the miRNAs, it is increasingly recognized that circulating miRNAs are involved in cell–cell communication and can be transferred to (vascular) target cells, thereby mediating vascular (dys)function [16]. Circulating miRNAs are carried in exosomes, lipoprotein complexes [mainly high-density lipoprotein (HDL)] or associated to RNA-binding proteins (Argonaute2). These carrier–miRNA complexes have been described as being transferred to and functional in target cells in various tissues [44, 45]. Furthermore, ‘brain miRNAs’ may be detected in the circulation, as the blood–brain barrier does not block the passage of miRNAs between cerebrospinal fluid and blood (although in lower concentrations) [46], while in pathological states, miRNAs can pass from the brain tissue into the blood through the injured blood–brain barrier [47]. As such, our hypothesis is that the circulating miRNAs we identified may play a role in the pathophysiology of cognitive decline and associated vascular injury in (older) ESRD patients. One such potentially relevant miRNA that we identified is miR-223, where we observed that lower levels of miR-223 were accompanied by impaired memory function. Interestingly, miR-223 has been described to be neuroprotective by targeting glutamate receptors while the absence of miR-223 leads to contextual memory deficits [48]. Other studies have implicated miR-223 in inflammation and vascular injury, suggesting that miR-223 may play a role in the link between vascular injury and inflammation in cognitive deficits [49, 50]. For example, the fact that miR-223 promotes microglia repair [51] could link its lower levels to ongoing injury (or lack of a reparative response), as microglia are intimately linked to memory [52]. We also observed an association of higher miR-29a with poorer memory function and higher WMH scores (Scheltens scale). This appears to be in contrast with previous studies that described lower miR-29a levels in Alzheimer’s disease [53] and loss of miR-29a causing neuronal dysfunction [54]. Further studies are necessary to clarify the relations between differential miRNA levels and cognitive function.

We also found several non–statistically significant trends for an association of Ang-2 or miRNAs with cognitive function in the different domains of memory, executive function and psychomotor speed. These might reflect real associations, but the lack of statistical significance might be due to the relatively small group size or more likely due to the fact that our cohort comprises a rather homogeneous group of patients who already have severe kidney failure and comparable cognitive (dys)function. Differences in levels of Ang-2 and miRNAs might be greater if we were to compare these measurements to similarly aged healthy people or patients spanning a larger range of cognitive function. In fact, our comparison of the levels of five angiogenic miRNAs (miR-27a, miR-126, miR-132, miR-223 and miR-326) and Ang-2 in our population of older patients with ESRD with those in young healthy people, as well as with those in patients with DN, already indicate quite large differences in the levels of circulating miRNAs and Ang-2, although the group sizes of healthy controls and DN patients are limited. We previously also showed that cognitive function in the COPE cohort is impaired compared with people without kidney failure of the same age [10], suggesting the combination of old age and kidney failure contribute to changes in the levels of Ang-2 and circulating miRNAs. Furthermore, in that comparison, altered miR-223, miR-126, miR-326 and miR-27a levels correspond to their previously demonstrated relation with vascular injury [20]. Interestingly, circulating miR-126 levels have already been shown to decrease in older patients with advanced CKD stage [55], while plasma miR-126 levels have also been shown to increase with age [56], suggesting a CKD-dependent decrease. Moreover, miR-126 has been demonstrated to be deregulated in small brain vessels during CKD [57]. miR-132 levels, however, were lower, while miR-132 was expected to be higher when more vascular injury is present. On the other hand, lower miR-132 levels have been extensively associated with decreased cognition [58].

Taken together, although further studies are warranted to understand the exact relation between circulating Ang-2 and angiogenic miRNA levels, our data suggest that Ang-2 and miRNAs might serve as biomarkers of cognitive function in older patients reaching ESRD and may provide insights into the underlying (vascular) pathophysiology.

SUPPLEMENTARY DATA

Supplementary data are available at ndt online.

FUNDING

R.B. is supported by a grant from the Dutch Kidney Foundation (16OKG16) and R.B. and A.J.V.Z. are supported by a grant from the European Foundation for the Study of Diabetes. Additional support was provided to A.J.V.Z by the Dutch Heart Foundation in the context of the CVON consortium RECONNECT and to M.J.P.V.O., M.A.V.B., E.E.B. and H.C.A. in the context of the CVON consortium Heart Brain Connection (CVON 2018-28 and 2012-06).

DATA AVAILABILITY STATEMENT

The data underlying this article are available in the article and its online supplementary material.

AUTHORS’ CONTRIBUTIONS

R.B. conducted experiments, acquired and analysed data and wrote the manuscript. M.H.K., L.E.Z, B.M.V.B., J.D.B., S.H., E.E.B., N.C.B.B. and H.A. acquired and analysed data. R.B., M.A.V.B., W.J.W.B., D.V.H., T.J.R., M.V.B. and S.M. conceptualized the study. R.B., A.J.V.Z., M.V.B. and S.M. contributed to the discussion and reviewed and edited the manuscript. R.B., M.V.B. and S.M. are guarantors of this work and, as such, had full access to all the data in the study and responsibility for the integrity of the data and the accuracy of the data analyses.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to disclose.

REFERENCES

Author notes