Effect of cerebral microhemorrhages on neurocognitive functions in patients with end-stage renal disease

Introduction

Chronic renal failure (CRF) is defined as failure of adjustment of fluid-solute balance due to a decrease in glomerular filtration rate (GFR) and chronic progressive deterioration in metabolic-endocrine functions. End-stage renal disease (ESRD) represents the last stage of CRF characterized by irreversible kidney function loss in which patients require regular dialysis or renal transplantation for sustaining their lives. In patients with ESRD, hemodialysis (HD), peritoneal dialysis, and renal transplantation are the treatment options (1).

Susceptibility weighted imaging (SWI) is an advanced magnetic resonance imaging (MRI) technique that exploits the magnetic susceptibility differences of various substances, such as blood, iron, and calcium, as a new source of contrast enhancement. SWI has become widely used in addition to routine MRI protocols for various brain pathologies. SWI has emerged, especially in conditions such as hypertensive encephalopathy, stroke, and diffuse axonal injury where presence of microbleeding affects the success of treatment (2).

Patients with ESRD are at risk of various systemic complications, some with fatal prognosis, caused by significant reduction in renal functions and HD, which is often the preferred treatment choice. Among the neurological complications, intracranial hemorrhage is a common complication seen in HD patients and patients with hemorrhagic stroke in particular have a poor prognosis in terms of morbidity and mortality (3).

The aim of the present study was to detect intracerebral macro- or microhemorrhage using SWI in patients with ESRD and to evaluate the effect of presence of intracerebral hemorrhage on neurocognitive function impairment.

Material and Methods

A total of 49 HD cases (25 men, 24 women; mean age = 57.2 ± 15.4 years; age range = 22–87 years) with ESRD, previously referred to our Nephrology Department, were enrolled in the study. Patients underwent routine cranial MRI and SWI. Axial T2-weighted (T2W) and fluid-attenuated inversion recovery (FLAIR), three-dimensional (3D) T1-weighted (T1W), diffusion-weighted imaging (DWI), and SWI sequences were obtained. A section thickness of 5 mm was used (matrix = 128 × 128, field of view =200 × 230 mm). MRI was performed on a 1.5-T system (Siemens, Avanto, Erlangen, Germany) using a head coil. Intravenous contrast material and drugs for sedation were not used in any of the patients. First, routine brain imaging with turbo spin echo 3D T1W images in the axial plane (TR/TE = 500/12 ms), turbo spin echo T2W images in the axial plane (TR/TE = 4280/91 ms), FLAIR images in the axial plane (TR/TE = 8000/118 ms), DWI in the axial plane (b values = 0 and 1000 s/mm²), and apparent diffusion coefficient (ADC) maps were obtained. Then, the SWI sequence was applied. The SWI protocol consisted of a magnitude and phase scan. The minimum intensity projection algorithm was obtained. The presence of macro- or microbleeds was investigated with SWI (Figs. 1 and 2).

OPEN IN VIEWER

Fig. 1.

In two different hemodialysis patients, SWI sequence phase (a, c) and final SWI (b, d) images show intracerebral microbleeds (arrows). SWI, susceptibility weighted imaging.

OPEN IN VIEWER

Fig. 2.

In two different hemodialysis patients, SWI sequence phase (a, c) and final SWI (b, d) images demonstrate intracerebral macrobleeds (arrows). SWI, susceptibility weighted imaging.

Patients with co-morbidities predisposing to hemorrhage such as cirrhosis, malignity, and coagulation disorders were excluded from the study. The demographic characteristics of cases, the total time of HD period, co-morbid situations such as diabetes mellitus (DM) and hypertension were queried and their relations with microbleeds were investigated.

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee (Bezmialem Vakif University, reference number: 71306642/050-01-04/174) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. All individuals were fully informed and gave their written informed consent.

The localization of microbleeds were classified as basal ganglion + thalamus, cortical + subcortical area, and cerebellum; distribution of bleeding was examined. Globus pallidus in which signal intensity changes secondary to physiological mineralization due to aging may be seen was not included into classification (Fig. 3).

OPEN IN VIEWER

Fig. 3.

SWI sequence phase (a) and final SWI (b) images shows signal changes because of physiological mineralization-related aging (arrows). SWI, susceptibility weighted imaging.

Among the 49 cases, the standardized mini-mental test (SMT) was performed by a neurologist in 43 cases to determine whether neurocognitive functions were affected. We thought the test results would not reliable because of the history of ischemic cerebrovascular attack in six cases, so in these patients the SMT was not performed. The maximum test score was 30 points: ≤9 points was defined as severe cognitive impairment; 10–18 points was defined as moderate cognitive impairment; and 19–24 points were defined as mild cognitive impairment. Cases in which neurocognitive tests were performed were separated into two groups: cases with intracerebral microbleeding (group 1, n = 26) and cases without intracerebral microbleeding (group 2, n = 17). We investigated whether there was any difference between the groups in terms of impairment of neurocognitive function.

Results

Mini-mental test findings

Patients in whom the SMT was performed were separated into two groups: group 1 (n = 26, mean age = 59.1 ± 15 years) cases with microbleeding and group 2 (n = 17, mean age = 55.5 ± 16.9 years) cases without microbleeding. Cognitive impairment was detected in 10 (38.4%) cases in group 1. Among these cases, six had mild cognitive dysfunction and four had moderate cognitive dysfunction. Cognitive dysfunction was detected in two cases in group 2, both of which were mild.

A negative correlation was found between the cognitive test score and the presence of microbleeding (P = 0.031). In group 1, the test score was 24.3 ± 5.1; in group 2, it was 27 ± 2.7. There was no statistically significant difference between attention disorder, which was a major criterion queried during testing, and microbleeding. However, there was a statistically significant difference between short-term memory impairment and microbleeding. In cases with microbleeding, short-term memory impairment was found to be more frequent (P = 0.027). The ratio of cases with short-term memory impairment to cases without short-term memory impairment was 15/11 (57.6%) in group 1 and 4/13 (23.5%) in group 2.

In patient in whom the SMT was performed, microbleeds were classified according to locations. Microbleeds in the cortical + subcortical region were detected in 24 (58.8%) cases, in the cerebellum in 10 (23.2%) cases, in the basal ganglion + thalamus in 8 (18.6%) cases, and in the brainstem in 6 (13.9%) cases. When the microbleeding locations were evaluated, no correlation was found between the location and SMT score.

Receiver operating characteristic (ROC) analysis shows, if a test score is accepted as ≤ 23, the presence of microbleeding can be considered with a sensitivity of 38.5% and specificity of 88.2% (area under the curve [AUC] = 0.637, P = 0.1026; Fig. 4).

OPEN IN VIEWER

Fig. 4.

ROC analysis shows, if test score accepted as ≤23, presence of microbleeding can be considered with a sensitivity of 38.5% and specificity of 88.2% (AUC = 0.637, P = 0.1026). AUC, area under the ROC curve; ROC, receiver operating characteristic.

Discussion

CRF is characterized by a progressive and usually irreversible decrease in GFR. The most common causes of CRF are DM, hypertension, and glomerulonephritis. When the GFR declines to < 5 mL/min, it is called ESRD (4). At this stage, many complications may develop due to both significant reduction in renal functions and HD is one of the most preferred methods of treatment for ESRD. Frequent complications induced by HD include hypotension, headache, nausea, vomiting, chest and back pain, fever, chills, muscle cramps, itching, and restless leg syndrome. Rare but serious complications are cardiac tamponade, air embolism, hemolysis, hypoxemia, anaphylactic reactions, disequilibrium syndrome, arrhythmias, seizures, and intracranial hemorrhages (5).

In dialysis patients with CRF, neurological complications are frequently seen. Peripheral neuropathy is the most common complication. Sensory, motor, and cranial nerves may be involved. Involvement may take the form of carpal tunnel syndrome, ischemic monomelic neuropathy, anterior ischemic optic neuropathy, or polyneuropathy. As a central nervous system complication, dialysis dementia, disequilibrium syndrome, Wernicke encephalopathy, central pontine myelinolysis, atherosclerosis related with cerebrovascular disease, intracranial hemorrhages, and posterior leukoencephalopathy are the most frequently seen situations (6).

Intracranial hemorrhage is a complication with serious morbidity and mortality rates, and is approximately 10 times more common in HD patients compared to a normal healthy population. The main factors that play a role in hemorrhage include a marked increased risk of atherosclerosis, anemia, thrombocytopenia, anticoagulant drugs such as heparin used during dialysis, inadequate hypertension control, thrombocyte aggregation due to hyperuricemia, defect in thrombocyte–vessel wall adhesion, and abnormal production of nitric oxide-prostaglandin I2.

In the literature, in HD patients with ESRD, some studies have been performed regarding intracranial hemorrhages in the last 10 years. In HD patients, intracerebral macrohemorrhage foci are frequently observed in basal ganglion + thalamic and subcortical regions (7).

In the present study, microhemorrhage foci, in order of frequency, were found in cortical + subcortical region, cerebellum, basal ganglion + thalamus and brainstem. Unlike macrohemorrhages, the presence of microhemorrhages, frequently in the cortical + subcortical regions, suggested that the possibility of microangiopathy is more prominent at these locations. There was also no significant association between the location of hemorrhage and neurocognitive dysfunction.

Pai et al. (3) reported high mortality rates in dialysis patients with hemorrhagic stroke. In these cases, prognosis was poor and the main cause of mortality was related to the deterioration of neurological function. In addition, they reported that co-morbidities such as hypertension and DM frequently accompanied these cases.

In the study by Watanebe et al. (8), frequency of microhemorrhages and their relationship with chronic intracerebral hematomas were studied with T2W MRI in chronic dialysis patients. They found microhemorrhages in 35% of cases. They also reported that frequency of microhemorrhages in cases with macrohemorrhages and the frequency of macrohemorrhages in cases with microhemorrhages were significantly higher. They suggested that microhemorrhages may be a predictor of macrohemorrhages. In the present study, the presence of microbleeding was evaluated with SWI, which is a more accurate MRI method for diagnosis of hemorrhage. Microbleeding was detected in 59% of our cases and macrobleeding was detected in 14%. Microbleeding was also present in all cases with macrobleeding. Byun et al. (7) found an association between the presence of macrohemorrhage and age, duration of dialysis and hypertension, whereas in the present study, no correlation was found between these factors and microhemorrhages. This discordance between the two studies may indicate that these factors may play a role in the conversion of a microhemorrhage (precursor of macrohemorrhage) to a macrohemorrhage rather than the development of a microhemorrhage.

In the present study, we found a significant positive correlation between DM and the presence of microbleeding. Murakami et al. (9) reported that DM was a poor prognostic factor in HD patients with intracranial macrohemorrhages. We thought that the cause of poor prognosis in patients with DM may be related to the fact that DM is a triggering factor for both the development of microhemorrhages and the conversion of microhemorrhages to macrohemorrhages.

In the present study, in cases with microhemorrhages, the SMT was performed to determine the effect on microhemorrhages on neurocognitive function. Goos et al. (10) investigated the effect of the presence of microbleeds on neurocognitive function in patients with Alzheimer’s disease. In this study, the presence of microbleeding was evaluated with T2W MRI; neurocognitive functions were evaluated with the SMT, similar to the present study. They found that neurocognitive functions were worse in patients with Alzheimer’s disease with multiple microbleeds. Similar to their study, in our study there was a negative correlation between test score and the presence of microbleeding. In patients with deteriorated neurocognitive functions, mild cognitive impairment was present in 60% of cases and moderate cognitive impairment was present in 40% of cases.

In the present study, ROC analysis, drawn up in terms of relationship between the presence of microbleeds and SMT score, showed a sensitivity of 38.5% and specificity of 88.2% in cases with SMT score ≤23 (AUC = 0.637, P = 0.1026).

With regard to attention disorder and close memory disorder, which were the two basic criteria during the scoring of the SMT, no correlation was found between the microbleeding and attention disorder, whereas a positive correlation was found between microbleeding and close memory disorder. This finding can be attributed to presence of microbleeds at the cortical + subcortical regions, in which microbleeding was commonly seen compared with other brain regions.

The present study had some limitations. The lack of homogenity in terms of age and duration of dialysis created a limitation in the evaluation of the relationship between these factors and microbleeding. The patient population of the subgroups was relatively small. To display the relation between the microbleeding and neurocognitive functions with higher sensitivity and specificity values, new studies with a wider patient population, including more homogenous groups, can be useful.

In conclusion, to detect microbleeding, which may be predictor of macrobleeding and may cause decrease in cognitive functions, SWI should be added to routine MRI sequences in dialysis patients with ESRD. Particularly in patients with DM with declining neurocognitive functions, SWI has the potential to increase the specificity of cranial MRI in the detection of microhemorrhages, which may provide guidance in terms of treatment protocols in an early period. However, more extensive studies with larger populations are needed to clearly document the effectiveness of the SWI.