Tau Protein Elevation During a Single Hemodialysis Session in End-Stage Renal Disease Patients and Its Associations with Cardiovascular and Inflammatory Biomarkers

Introduction

End-stage renal disease (ESRD) is a growing global health problem with increasing prevalence.^1-3 In patients with ESRD, the primary causes of death are mostly related to cardiovascular complications, which account for a significant proportion of mortality. This is largely due to the high prevalence of comorbid conditions such as diabetes and hypertension, which contribute to cardiovascular disease. Additionally, complications arising from hemodialysis (HD), including infections, fluid overload, multiorgan ischemia, and electrolyte imbalances, further exacerbate the risk of mortality.

Neurological complications and cognitive impairment have been reported in patients with ESRD undergoing maintenance HD. Proposed mechanisms include uremic toxin exposure, oxidative stress, inflammation, and intradialytic hemodynamic instability, which may contribute to cerebral hypoperfusion and neuronal stress.^1,4-7 Rapid fluid and electrolyte shifts, particularly in patients with pre-existing vascular disease, can lead to cerebral hypoperfusion and subsequent neurological damage.^1,7 Additionally, chronic inflammation and malnutrition associated with ESRD may contribute to accelerated neurodegeneration.⁶ In the context of neuronal damage, particularly in patients with ESRD undergoing HD, several biomarkers have been identified as significant indicators of neuronal injury. These include Tau protein, a microtubule-associated protein involved in axonal structure and function. Increased circulating Tau levels have been reported in several conditions associated with neuronal injury or stress; however, these findings in ESRD is complex, as renal dysfunction and dialysis-related factors may affect circulating biomarker concentrations.^8-12,S1-S5 Other markers include Glial fibrillary acidic protein (GFAP), a structural protein predominantly released upon astrocytic injury, with elevated serum levels associated with brain damage; ^8,13,14 myelin basic protein (MBP), a key component of the myelin sheath in the central nervous system, serving as a marker for demyelination and axonal injury^8,15 and S100B protein (S100B), a calcium-binding protein primarily produced by astrocytes, involved in neuroinflammation and neuronal survival.^10,14

The brain–heart axis provides a conceptual framework for potential interactions between neuronal stress and cardiovascular vulnerability in patients with ESRD. However, direct evidence linking intradialytic changes in neuronal injury biomarkers with cardiovascular biomarkers remains limited. This study aimed to assess changes in selected neuronal injury biomarkers during a single HD session and to explore their associations with cardiovascular and inflammatory markers.

Materials and Methods

The study was approved by the Bioethics Committee of the Jan Kochanowski University in Kielce (6/2022), and each participant signed a written consent. This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

All available patients with ESRD undergoing dialysis at our hospital’s dialysis center who did not meet the exclusion criteria were included in this study. The exclusion criteria were cardiovascular complications in the last 3 months (such as myocardial infarction, brain stroke), severe infection or exacerbation of chronic disease in the last 1 month, cancer or other severe health issue with a predicted survival rate <6 months, cardiac arrhythmias (such as atrial fibrillation or flutter) occurring on the day of study, or lack of patient cooperation. From the group of 48 patients enrolled in the study, 4 patients were excluded: 3 patients due to atrial fibrillation episodes during HD sessions and one patient due to his resignation (n = 1).

All included patients were treated with HD three times a week for 4 h per session. The dialysate flow rate was at least 500 mL/min, while the blood flow rate was at least 300 mL/min. During HD, anticoagulation was performed using unfractionated heparin (UFH) at a total dose of ∼3.000–7.000 IU, administered in one or two divided doses and adjusted according to individual patient requirements. All HD sessions were carried out using Fresenius HD machines and Fresenius FX10 dialyzers. The FX10 is a low-flux Helixone^® dialyzer with an effective membrane surface area of 1.8 m², according to the manufacturer’s data. Dialysis prescription was individualized for each patient in line with routine clinical practice, including adjustment of dialysate sodium and bicarbonate concentrations, dialysate composition, and target ultrafiltration. Typical dialysis settings comprised a dialysate sodium concentration of 137 mmol/L, potassium concentration of 2–4 mmol/L, and calcium concentration of 1.25–1.5 mmol/L, with dialysate temperature maintained at 37°C. Blood pressure was monitored during each session, and no intradialytic hypotensive episodes were observed.

The blood was collected from participants immediately before (immediately before connection) and after (after rinseback, before saline rinse) the HD session. To assess neuronal injury in the study participants, 4 biomarkers were selected: Tau protein, GFAP, MBP, and S100B protein. Above-mentioned biomarkers, as well as endothelin 1 (ET-1), oxidized LDL cholesterol (oxLDL), and interleukin 6 (IL-6) concentrations in serum were measured using commercial ELISA kits by Cloud-Clone Corp. (HEB983Hu, SEA068Hu, SEA539Hu, SEA567Hu), Immundiagnostik (K 7810), and Diaclone (950.030.096), respectively, and Lambda 25 UV/VIS Spectrophotometer (Waltham, MA). Laboratory tests routinely performed on patients on HD, including complete blood count, electrolytes, parameters of iron status, indices of calcium–phosphate metabolism, and other standard biochemical measures, were determined using automated analyzers and commercial kits. All assays were performed as single measurements by a highly experienced analyst, following the manufacturer’s instructions for the respective kits. No blinding was applied during the analytical procedure. According to the manufacturers’ specifications, the following analytical performance parameters apply: Tau protein (Cloud-Clone HEB983Hu/SEB983Hu), detection range 31.2–2000 pg/mL, minimum detectable dose <12.8 pg/mL, intra-assay CV <10%, inter-assay CV <12%; GFAP (Cloud-Clone SEA068Hu), detection range 0.156–10 ng/mL, minimum detectable dose <0.056 ng/mL, intra-assay CV <10%, inter-assay CV <12%; MBP (Cloud-Clone SEA539Hu), detection range 15.63–1000 pg/mL, minimum detectable dose <5.7 pg/mL, intra-assay CV <10%, inter-assay CV <12%; S100B (Cloud-Clone SEA567Hu), detection range 0.156–10 ng/mL, minimum detectable dose <0.056 ng/mL, intra-assay CV <10%, inter-assay CV <12%; oxLDL (Immundiagnostik K 7810), analytical sensitivity 4.13 ng/mL, intra-assay CV 3.9–5.7%, inter-assay CV 9.0–11.0%; IL-6 (Diaclone 950.030.096), standard curve range 6.25–200 pg/mL, minimum detectable dose 2 pg/mL, intra-assay CV 3.6%, inter-assay CV 7.7%.

Electrocardiographic parameters were obtained before and immediately after the HD session using the Cardiax device (IMED Co. Ltd., Budapest, Hungary). This involved automatic calculating of the spatial QRS-T angle (QRS-Tan) with the inverse Dower method by Cardiax.

Results

Forty-four patients’ data were analyzed in this study (22 females and 22 males). The mean age in the analyzed group was 61 ± 12.91. Vascular access for HD was provided by an arteriovenous fistula in 26 patients and by a central venous catheter in 18 patients. In the examined group, 11 patients suffered from diabetes mellitus. All patients were taking beta-blockers, while none of them was using any QT-prolonging drugs.

Mean dialysis vintage was 4.5 ± 2.6 years. Mean ultrafiltration volume (UF) was 2.25 ± 1.07, while the mean Kt/V was 1.6 ± 0.3. The mean blood pressure decreased during HD (from 147.94 ± 27.77/81.85 ± 15.50 mmHg to 137.48 ± 28.29/80.23 ± 15.45 mmHg; p = 0.0001 and p = 0.2539, respectively).

Analysis revealed changes in serum concentrations of tau protein, C-reactive protein (CRP), ET-1, and oxLDL. Remaining neuronal injury biomarkers (MBP, GFAP, S100B protein) didn’t change significantly during a single HD session. Changes in neuronal injury and other biochemical markers have been presented in Table 1. 

Additionally, the potential effect of hemoconcentration on changes in tau protein concentration was analyzed. Tau protein concentration increased by a mean Δtau of 75.42 ± 165.43. Body weight decreased from 79.20 ± 18.64 kg to 76.95 ± 18.20 kg, corresponding to a mean ultrafiltration-related weight change of −2.25 ± 1.07 kg and relative weight loss of 2.86 ± 1.19%. Δtau was not significantly correlated with absolute ultrafiltration-related weight loss (R = −0.060, p = 0.699) nor percentage weight loss (R = −0.063, p = 0.683). Hematocrit showed only a small, non-significant increase (p = 0.0712), suggesting no marked hemoconcentration effect at the group level. Urea decreased from 123.32 ± 30.45 to 31.51 ± 11.22, with an absolute urea reduction of 91.81 ± 23.69. Δtau was not significantly associated with absolute urea reduction (R = −0.177, p = 0.333). Similarly, Kt/V was 1.58 ± 0.31 and was not significantly correlated with Δtau (R = −0.281, p = 0.126).

Correlation analyses were performed to identify candidate predictors for regression models explaining Tau-postHD and Δtau. In Supplementary Table 1, correlations with the highest absolute R values for Tau-postHD are presented. The strongest associations were observed for ET-1, IL-6, ferritin, QTc, and QRS-T angle. In Supplementary Table 2, the strongest correlations for Δtau are presented; variables with the highest absolute R values included ΔET-1, ΔIL-6, oxLDL-preHD, PTH-preHD, and QRS-Tan-postHD.

A multiple regression model constructed for Δtau involving change in ET-1 concentration (ΔET-1), parathormone concentration before HD (PTH-preHD), and oxLDL concentration before HD (oxLDL-preHD) described its variance in 46.54% [Table 2]. Regression assumptions were checked.

Similarly, a multiple regression model for Tau-postHD was constructed. In the constructed models with the highest R², ranging from about 0.4 to 0.6, the leading predictors were invariably ET-1-postHD and ΔQTc, while ferritin-preHD and IL-6-postHD were mutually exclusive. Correlation analysis and striving to obtain the best possible model resulted in the inclusion of ferritin-preHD in the final model, not IL-6-postHD. However, taking into account the available data, including a stronger correlation of IL-6 with ΔQTc than ferritin-preHD with ΔQTc, as well as a stronger correlation of IL-6-postHD with Tau-postHD than ferritin-preHD with Tau-postHD, it seems most appropriate to treat ferritin-preHD as a marker of inflammation in the interpretation of this model. Selected model involving ET-1-postHD, ferritin-preHD, and ΔQTc described its variance in 60.51% [Table 3]. Given the exploratory nature of the regression analyses and the biological plausibility of IL-6, we additionally presented an alternative model including IL-6-postHD instead of ferritin-preHD [Table 4]. This model explained 42.23% of the variance in Tau-postHD, thereby allowing a more transparent comparison of the competing candidate models. Regression models’ assumptions were checked.

Discussion

In this study, we show that serum Tau increased after a single HD session, consistent with prior reports in patients with renal failure, while GFAP, MBP, and S100B did not change significantly.

The increase in serum Δtau was also observed by other researchers examining patients with kidney impairment; its causes and effects stay unclear. Rubenstein et al.¹⁶ and Kitaguchi et al.⁹ have both reported elevated Tau in renal failure, with the latter suggesting a role for impaired renal Tau degradation. The above-mentioned study suggested that the kidneys play a role in tau degradation, and the increase in tau levels during HD sessions indicates that tau may be a relevant biomarker for assessing neuronal damage in this population.⁹

In our study, the intradialytic increase in tau protein concentration also did not appear to be directly explained by hemoconcentration. Although ultrafiltration during HD may reduce plasma volume and thereby increase the measured concentration of circulating solutes, Δtau was not associated with absolute or relative weight loss. Moreover, the hematocrit increase was modest and statistically nonsignificant, which does not support substantial hemoconcentration as the dominant mechanism. We also found no evidence that tau was removed from serum proportionally to urea during HD in our patient group, as Δtau was not associated with urea reduction or Kt/V.

The intradialytic increase in Tau observed in our study may be consistent with neuronal stress in patients undergoing HD. However, because no neuroimaging, cerebral perfusion assessment, cerebrospinal fluid analysis, or cognitive testing was performed, serum Tau elevation alone cannot be interpreted as definitive evidence of structural brain injury.

The absence of significant changes in GFAP, MBP, and S100B is consistent with prior studies in ESRD;^10,13-15 notably, S100B has been linked to depression and cognitive impairment in patients on dialysis rather than acute neuronal injury.^17,18,S6

We observed a relationship between the increase in tau protein concentration and both ET-1 and oxLDL. The association of tau-related outcomes with ET-1 and oxLDL may suggest that endothelial activation and oxidative stress are part of the broader biological context of intradialytic tau changes. However, the present observational design does not allow us to determine whether these pathways directly contribute to tau release or reflect parallel dialysis-related processes.^19,S7,S8 In addition, ET-1 secreted by astrocytes worsens the prognosis in patients with brain damage, among other things, by sensitizing tissues to ischemia. Considering that one of the mechanisms of putative neuronal damage during HD is precisely the decrease in cerebral blood supply during HD, the pathophysiological role of ET-1 seems to be particularly important here.

Although UFH generally suppresses ET-1 release in experimental data, a potential anticoagulation-related bias on ET-1-postHD cannot be excluded in the absence of a heparin-free comparator.^24-28 However, because no heparin-free or alternative-anticoagulation comparator was included in our study, a potential UFH-related bias affecting ET-1-postHD levels cannot be completely excluded and should be considered when interpreting the association between ET-1-postHD and Tau-postHD.

The associations of tau-related outcomes with ferritin and IL-6 may indicate that inflammatory activity contributes to the broader pathophysiological background of intradialytic biomarker changes. This interpretation remains exploratory, particularly because inflammatory markers, endothelial dysfunction, oxidative stress, and cardiovascular risk are closely interrelated in ESRD.^4,5,21,23

QTc prolongation may represent a potential pathway through which HD-related neuronal stress could contribute to cardiovascular risk in patients with ESRD, consistent with the brain–heart axis concept.^{4,29-33,S9-S12} In our exploratory analyses, Tau-related measures were associated with QTc changes, which may suggest a possible interaction between neuronal stress biomarkers and cardiac electrical instability in patients with ESRD. However, the directionality of this association cannot be determined from the present study, and no causal inference should be made. We also observed a nonsignificant trend in correlation analyses, indicating an indirect association of Tau protein with the spatial QRS-T angle, a reliable marker of cardiovascular risk.^5,29,34,35

This study has some limitations - it is an exploratory, hypothesis-generating, single-center observational study without a control group. The study is limited by the relatively small sample size; therefore, the reported associations should be interpreted as associative rather than causal. Biomarker concentrations were assessed using single measurements, and inter-assay coefficients of variation of up to 12% may have introduced measurement imprecision. Although the performed analyses did not indicate that the Tau increase was secondary to hemoconcentration, direct plasma-volume-corrected Tau values were not available; therefore, hemoconcentration cannot be completely excluded. In addition, all patients received UFH during HD, and no heparin-free or alternative-anticoagulation comparator was included, so a potential anticoagulation-related effect on ET-1-postHD cannot be excluded. Finally, neuronal injury biomarkers were measured only in serum, and no direct neurological assessment, including cerebral perfusion imaging, structural neuroimaging, or cognitive testing, was performed.

To conclude, a single HD session was associated with an increase in serum ΔTau in patients with ESRD. The observed associations between Tau protein and cardiovascular and inflammatory markers, as well as QTc changes, might suggest a potential link between intradialytic Tau dynamics and endothelial/inflammatory activation. These findings should be interpreted as exploratory and hypothesis-generating, and further clinical and mechanistic studies in larger, independent cohorts are needed to clarify their significance.