Effect of Hemodialysis on Cognitive Function in ESRD Patients

Abstract

Background. Uremia is associated with impairment of different cognitive functions. However the pathogenesis of this cognitive dysfunction is unknown. Objective. In this study, long-latency event related potentials (ERPs) were used to assess changes in cortical function due to hemodialysis treatment. Methods. In this cross-sectional study, we measured event related potentials in 15 end stage renal disease (ESRD) patients maintained on hemodialysis, two hours before and two hours after they underwent hemodialysis and compared their data with a strictly age and sex matched healthy control group. The P3 was elicited by using standard auditory “odd-ball” paradigm and the data obtained was statistically analyzed (Wilcoxon signed ranks, Mann Whitney). Results. Before hemodialysis, the patients' P3 latency (347.73 ± 39.47 ms) was significantly increased as compared with that of healthy control group (308.4 ± 13.73 ms) (p = 0.001). After hemodialysis, P3 latency of the patients showed a significant decrease (347.73 ± 39.47 ms to 325.20 ± 37.15 ms, p = 0.001). P3 latency after dialysis was not significantly different from controls. No significant correlation was noted between various biochemical parameters (hemoglobin, blood urea, creatinine, uric acid and calcium) and P3 latency or amplitude. Conclusions. Removal of uremic toxins by hemodialysis leads to an improvement in cognitive processing.

INTRODUCTION

Cognitive impairment is highly prevalent in end-stage renal disease (ESRD) patients when compared to general population.[1],[2] Some studies comparing dialysis patients with controls have shown greater cognitive impairment in dialysis patients.[3–5] Very few studies have been performed in the current era of increased dialysis adequacy. More recently, Pliskin and Umans, by using neuropsychological test batteries, have demonstrated an absence of significant neurocognitive defects in attention or mental processing speed in well-dialyzed, well-nourished, and medically stable dialysis patients.[6],[7]

Two broad classes of tests can be used to assess cognitive function. Neuropsychological tests use validated questions and screening tests to evaluate cognition. On the other hand, neurophysiological tests use electrophysiological methods (e.g., electroencephalogram and event-related potentials [ERPs]) to assess cognitive dysfunction.[8] Performance on cognitive function is provided by assessment of cognitive ERPs. In the prototypical ERP trace, the most prominent and studied is the P3 (or P300, the third positive wave, or the wave with a 300 ms latency; see Figure 1). The P3 wave is evoked by a task known as the odd-ball paradigm. The latency of the P3 corresponds to the speed of cognitive processing and memory[9],[10] and the amplitude of P3 varies with the attention, on the relevancy of the task or with stimulus novelty.[11] Prolongation of P3 latency has been shown to be the earliest sign of cognitive dysfunction in metabolic encephalopathies.[12] Thus, these ERPs may be a more sensitive and useful quantitative method than neuropsychological tests to assess cognitive function.

Figure 1. A prototypical ERP trace.

ERPs have been used to study the effects of hemodialysis on ESRD patients[13–16]; however, many of these previous studies have failed to take into account various co-morbidities (such as coronary artery disease, diabetes mellitus, cerebrovascular disease, anemia, malnutrition etc.) that may contribute to cognitive dysfunction in these patients.[17] Therefore, we sought to assess the effect of hemodialysis on ESRD patients on cognitive event related potentials after carefully excluding patients with medical co-morbidities.

MATERIAL AND METHODS

Subjects

Fifteen stable patients of ESRD in the age group 21–50 years on maintenance hemodialysis for at least one month duration were recruited from the Renal Clinic and Medical Wards of our hospital for the study. Age- and sex-matched healthy controls were also studied. Age-matching was done within every subgroup of 5 years. Informed consent was taken from all study subjects.

The various causes of ESRD in the patients were chronic glomerulonephritis (n = 10), hypertensive nephrosclerosis (n = 1), chronic interstitial nephritis (n = 3), and chronic rejection of transplanted kidney (n = 1). It may be mentioned at this stage that patients with diabetic nephropathy as a cause of ESRD were excluded from the present study. In addition, patients were excluded from the study if they had primary neurological or psychiatric disorder, hepatic disease, dyselectrolytemia, overt cognitive dysfunction (MMSE ≤ 24), accelerated hypertension, severe anemia (Hb <7gm/dL), known myocardial infarction/unstable angina, hearing impairment, active collagen vascular disorders or vasculitis requiring use of cytotoxic drugs or steroids (at doses greater than 10 mg prednisone/d), or evidence of protein malnutrition (serum albumin <3.5g/dL), ingested alcohol during the preceding one month, consumed centrally acting drugs, or were on treatment with recombinant human erythropoietin (to exclude its effects on P3ERPs).

A Mini Mental State Examination (MMSE), allowing an estimate of cognitive function, was performed to exclude patients with significant impairment (scoring ≤ 24); however, it was not used to monitor mental changes in patients due to its low sensitivity.[18]

Dialysis Delivery

The mean duration of hemodialysis in our patients was 21.13 ± 21.17 weeks (range 6–67 weeks). The patients were dialyzed twice weekly for four hours at every session with cellulose acetate hollow fiber dialyzer using hemodialysis fluid containing bicarbonate on volumetric hemodialysis machine (Althin Altratouch). The adequacy of dialysis was ensured by measuring urea reduction ratio (URR) and keeping the URR ≥70%.[19] The sample for estimating blood BUN/urea was collected as per the prescribed guidelines.[20]

Testing Procedure

The P3ERPs were measured in all subjects. The hemodialysis patients were tested two hours before (66 hours after the last dialysis) and after a standard hemodialysis session. ERPs were recorded using the MEB-9100 Version 0.3-0.6 Neuropack μ (Nihon Kohden, Japan). The P300 was recorded with Ag/AgCl disk electrodes placed at standard scalp locations of 10–20 international system. The active electrodes were placed at F_z and P_z, with reference electrodes at A₁ and A₂ positions. The ground electrode was placed at Fpz site. The skin-electrode contact impedance was kept below 5Ω. The P300 wave was elicited by delivering the standard auditory odd-ball paradigm binaurally. The frequent tones (80%) and the rare tones (20%) were set at 1 kHz and 2 kHz, respectively (intensity, 70 dB; rise fall time, 10 ms; plateau,100 ms). Stimulus sequence was random, and the signals were in phase at two ears. The patient responded to the rare tones by pressing a button. The recorder settings were properly selected, and evoked responses to the frequent and rare stimuli were filtered with a band pass 0.1–50 Hz and averaged simultaneously for 30 responses. Data obtained was stored, analyzed, and averaged by the computer. The latencies of N1, N2, P2, and P3 were recorded. In addition, amplitudes of waves N1–P2, P2–N2, and N2–P3 were also recorded.

Data Analysis

The data was analyzed by using SPSS 13 statistical software using Wilcoxon signed ranks and Mann Whitney U tests. We also performed correlation analysis using Pearson correlation coefficient between hemoglobin, serum creatinine, uric acid, and calcium and blood urea levels with an amplitude and latency of P3 wave. p values < 0.05 were considered statistically significant.

RESULTS

Demographic characteristics and mean laboratory values are presented in Table 1. By design, there were no significant differences in age and sex distribution between the controls and the study group. Each group had 10 males and 5 females.

Table 1 Demographic and laboratory parameters of controls and ESRD patients

	Controls (n = 15)	ESRD (n = 15)
Age	31.8 (±9.28)	32.13 (±9.37)
Hemoglobin (g/dL)	13.2 (±1.02)	9.19 (±0.78)
Blood urea (mg/dL)	22.93 (±5.47)	154.67 (±26.15)
BUN	10.72 (±1.62)	72.27 (±12.22)
S. creatinine (mg/dL)	0.81 (±0.11)	10.1 (±1.62)
S. uric acid (mg/dL)	3.34 (±0.55)	5.29 (±1.10)
S. calcium (mg/dL)	9.37 (±0.40)	8.26 (±0.66)
MMSE	29.00 (±0.85)	28.4 (±1.24)

All values expressed as mean (±1SD).

Auditory Event-Related Potentials

Table 2 shows findings of auditory ERPs in the 30 subjects. There were no significant differences between all the groups as far as latencies of N1, P2, and N2 and amplitudes of N1–P2, P2–N2, and N2–P3 were concerned. The only significant differences were noted for P3 latencies.

Table 2 Electrophysiological studies in patients of chronic kidney disease

		ESRD (on HD) (n = 15)
Wave	Controls (n = 15)	Predialysis	Postdialysis	p
Latencies (ms)
N1	91.33 (±7.19)	105 (±16.47)	99.27 (±12.96)	0.675
P2	165.2 (±16.99)	178.27 (±27.11)	175 (±30.76)	0.094
N2	216.13 (±18.58)	238.67 (±43.62)	234.47 (±46.11)	0.820
P3	308.4 (±13.73)	347.73 (±39.47)	325.2 (±37.15)	0.001
Amplitudes (μV)
N1–P2	9.58 (±4.69)	9.725 (±5.608)	10.61 (±3.686)	0.427
P2–N2	5.79 (±3.56)	5.953 (±5.051)	7.076 (±4.501)	0.609
N2–P3	16.70 (±9.10)	13.272 (±10.03)	14.29 (±9.836)	0.82

Mean (±1SD) values of ERP measures recorded at Pz electrode in the subjects examined.

*results of Wilcoxon signed ranks (predialysis vs postdialysis values), significant at p < 0.05.

Latency of P3 Component

In the present study, it was found that P3 latencies were significantly increased prior to hemodialysis (347.73 ± 39.47ms) when compared with controls (308.4 ± 13.73ms, p = 0.001). However, when these patients were subjected to electrophysiological testing two hours after dialysis, their P3 latencies (325.2 ± 37.15ms) were not statistically different from those of controls (p = 0.11). No significant difference was noted between latencies of N1, P2, and N2 and the amplitudes of N1–P2, P2–N2, and N2–P3 between the controls and the ESRD patients. The effect of HD on latencies of other waveforms (i.e., N1, P2, and N2) was not significant. Although the amplitudes of N1–P2, P2–N2, and N2–P3 increased after dialysis, the increment was not found to be statistically significant

Correlations of P3ERPs with Some Laboratory Parameters

Intercorrelations between P3 latencies and amplitudes and the values of hemoglobin, blood urea, serum creatinine, calcium, and uric acid were assessed (see Table 3). There was no correlation of these laboratory parameters with either the latency or the amplitude of P3 wave.

Table 3 Correlation of P3 latency and amplitude with laboratory parameters

Laboratory parameter	P3 Wave	Coefficient of correlation (r)	p	Significance
S. creatinine	Latency	−0.349	0.202	NS
	Amplitude	−0.232	0.405	NS
Blood urea	Latency	−0.225	0.42	NS
	Amplitude	−0.414	0.125	NS
Hemoglobin	Latency	0.219	0.433	NS
	Amplitude	−0.201	0.473	NS
S. calcium	Latency	−0.408	0.132	NS
	Amplitude	−0.159	0.570	NS
S. uric acid	Latency	−0.197	0.481	NS
	Amplitude	0.185	0.508	NS

All values are Pearson's coefficients of correlation.

*p values significant at <0.05.

NS = non-significant.

DISCUSSION

The subjects recruited in the present study had no overt cognitive dysfunction (by design those with MMSE score <24 had been excluded). However, on neurophysiological testing, P300 latency was significantly prolonged in ESRD patients prior to dialysis when compared with controls. These data indicate that there is a deficit in cerebral cognitive function in uremic patients even in clinically asymptomatic stages, thus highlighting the sensitivity of ERPs in assessing cognitive dysfunction even when it is not clinically apparent.

ERPs have been used in previous studies also to examine the immediate effect of hemodialysis.[13–16] However, we have now recently recognized that many comorbid conditions that are frequent in ESRD patients (e.g. coronary artery disease, cerebrovascular disease, anemia, malnutrition) can also contribute to neuropsychological dysfunction. These conditions were not considered in the older studies. The present study demonstrated a significant reduction in P3 latency after hemodialysis and the P3 latency values observed post-dialysis were not statistically different from age- and sex-matched normal controls. Similar findings were also reported by other workers in spite of some methodological variations as noted above.[14],[15] Some investigators who have measured ERPs after 24 hours of hemodialysis session have demonstrated prolonged P3 latency when compared with controls.[21],[22] This suggests that improvements in P3 latency that occur after 1–2 hours after dialysis are only temporary and may not be sustainable at 24 hours after dialysis. There is some evidence to suggest that patients perform best on neuropsychological test batteries 24 hours after the last dialysis session.[6],[23] Whether the improvement in electrophysiological tests precedes the actual improvement on neuropsychological tests is a question that can only be answered by studies assessing cognitive function by performing both types of tests simultaneously at different time intervals after dialysis. Even though the reduction in P300 latency is only temporary, it still demonstrates the effectiveness of the therapeutic treatment in reversing the alterations in cognitive cerebral activity.

There were no significant correlation between any of the measured laboratory measures and the electrophysiological values in the present study, but this lack of correlation between blood urea, creatinine, and other biochemical indices and the electrophysiological measures is not totally consistent with earlier studies. (For example, in the previous study by Evers et al.[15], a high level of BUN was significantly correlated with decreased P300 latencies in dialyzed ESRD patients.) In contrast, in a study by Marsh et al.,[21] there were no significant correlations between serum creatinine and BUN levels and any of the ERP measures in dialyzed patients. Studies have indicated that a relationship between BUN and creatinine levels and electrophysiological function is complex and may be related to direction and rate of change in these values rather than the absolute levels.[24] Additionally, it has been seen that variations in BUN and creatinine resulting from systematic alterations in the hemodialysis regimen are not always associated with electrophysiological changes.[25] In the present study, there was no significant correlation of hemoglobin with P3ERPs. Thus, the finding that P300 latencies are increased by low hemoglobin in dialyzed patients levels as has been suggested by other investigators[22],[26] was not seen in our study. Our dialyzed patients showed lower levels of hemoglobin but no severe anemia (mean Hb 9.2 gm %), as these patients had been excluded (Hb < 7gm %). This may explain why we did not find a significant correlation of P300 latency with hemoglobin, unlike recent studies with mean levels of 7.5g/dL hemoglobin[26].

Our patients received dialysis delivery that is usually considered adequate for a developing country like ours, although it may not meet the current levels of dialysis adequacy in the developed world. Even then, the P3 latency after dialysis was comparable to controls. It seems that neuropsychological dysfunction resolves at some threshold level of hemodialysis adequacy, even though it is not sustained. Evidently studies designed to estimate this precise level of adequacy may not be ethically feasible.

Our data indicate that cognitive functions as measured by ERPs improve in the immediate period after hemodialysis. There is no correlation of this improvement with any of the measured laboratory parameters, suggesting that the etiopathogenesis of this cognitive dysfunction is complex. Therefore, a single session of hemodialysis improves cognitive function in ESRD patients probably by removal of uremic toxins that may require specialized biochemical investigation techniques. Additional studies may be required to elucidate the various factors causing cognitive dysfunction in these patients. Various strategies need to be devised and evaluated for the amelioration of cognitive dysfunction in CKD.