Variability in Cardiac Biomarkers during Hemodialysis: A Prospective Cohort Study

Abstract

Background

The effect of hemodialysis on cardiac biomarkers is unclear. We sought to evaluate the degree and causes of intradialytic variability of high sensitivity troponin I (hs-TnI), galectin-3 (gal-3), and heart-type fatty acid binding protein (hFABP).

Methods

hs-TnI, gal-3, and hFABP were prospectively measured pre-dialysis and post-dialysis for 1 week every month for 6 months in 178 prevalent adult hemodialysis patients at a single center in Hamilton, Canada. The degree of change from pre-dialysis to post-dialysis for each cardiac biomarker was estimated with multilevel linear regression models.

Results

The median change in the concentration of hs-TnI during hemodialysis was −1 ng/L (interquartile range [IQR] −1 to 2 ng/L) while gal-3 and hFABP changed by −36.3 ng/mL (IQR −27.7 to −46.8 ng/mL) and −19.41 ng/mL (IQR −13.61 to −26.87 ng/mL), respectively. The median (IQR) percentage intradialytic changes for hs-TnI, gal-3, and hFABP were 2.6% (−4.4% to 12.5%), −59.8% (−54.7% to −64.8%) and −35.3% (−28.4% to −42.1%), respectively. Ultrafiltration was associated with an increase in concentration of hs-TnI, gal-3, and hFABP (mean 0.99 ng/L, 1.05 ng/mL, and 1.9 ng/mL per L ultrafiltration, respectively, P < 0.001). Both gal-3 and hFABP concentrations decreased in association with the volume of blood processed (P < 0.001) and with hemodialysis treatment time (P  = 0.02 and P  = 0.04) while hs-TnI concentration decreased only in association with hemodialysis treatment time (P  < 0.001).

Conclusions

Ultrafiltration volume and hemodialysis treatment time influenced hs-TnI, gal-3, and hFABP concentrations during hemodialysis and should be considered when interpreting their measurement.

Introduction

Cardiovascular disease is the leading cause of death in patients receiving hemodialysis (1). Cardiac biomarkers are used to diagnose acute coronary syndromes and risk stratify patients with heart failure (2). However, their interpretation in the setting of hemodialysis is challenging because blood volume is changing, the procedure may induce myocardial stunning and injury, and cardiac biomarkers are commonly increased in patients with kidney disease even in the absence of clear symptoms (3–5).

High sensitivity troponin I (hs-TnI), galectin-3 (gal-3), and human fatty acid binding protein (hFABP) are cardiac biomarkers commonly used in clinical settings and research to understand myocardial injury and remodeling. TnI is a subunit of the troponin complex that controls the calcium-mediated interaction between actin and myosin. gal-3 is a soluble B-galactosidase-binding lectin that is released from activated monocytes and macrophages as a result of oxidative stress. It is a marker of fibrosis and is associated with adverse outcomes, including heart failure and mortality (6−9). hFABP is found in myocytes and is (generated or released) in response to ischemia and appears quickly in the blood and has a short half-life relative to hs-TnI, which may improve acute coronary syndromes diagnosis (10) and the prognostication of cardiovascular disease (11, 12).hs-TnI (5, 13), gal-3 (14–16), and hFABP (17, 18) are increased in patients on hemodialysis and are all associated with adverse outcomes, but their intradialytic variability and the factors, which predict change during hemodialysis are not clear. We performed a single center prospective cohort study of prevalent hemodialysis patients to evaluate the intradialytic variability of hs-TnI, gal-3 and hFABP over 6 months.

Materials and Methods

Study Cohort

The study cohort consisted of 178 adult hemodialysis patients with at least 2 hemodialysis treatments per week from single tertiary hemodialysis unit at St. Joseph’s Hospital in Hamilton, Canada.

Study Procedures

Demographics, comorbidities, medications, and laboratory tests were collected at baseline. hs-TnI, gal-3, and hFABP was measured pre- and post-dialysis every hemodialysis treatment for 1 week every month for 6 months (24 or more measurements per patient) in all study participants.

Cardiac biomarkers were collected in a 4.0 mL serum separator tube, allowed to clot upright at room temperature for 30 min and then were centrifuged at 3000 g for 6 min. The serum was transferred to 2.0 mL cryovials and stored at −80°C in liquid nitrogen vapor. Pre-dialysis, intradialytic, and post-dialysis systolic blood pressure, diastolic blood pressure, intradialytic symptoms (chest pain, other), and hemodialysis prescription (vascular access, duration, total blood processed, dialysate, membrane, ultrafiltration volume, weights) were recorded at every hemodialysis treatment in which cardiac biomarkers was measured. Participants that experienced acute events during hemodialysis requiring hospitalization did not have post-dialysis cardiac markers routinely collected, given the immediate transfer to an acute care setting. Approval for the study was obtained by the Hamilton Integrated Research Ethics Board prior to initiating any study procedures. Informed consent was obtained from all participants.

Assays

The hs-TnI assay used was the Abbott Architect assay with a lower limit of detection of 1 ng/L (19) with a coefficient of variation (CV) of <20% at approximately 4 ng/L (20). The current necrosis threshold based on the 99th percentile for the normal population is considered 30 ng/L (21) with sex-specific 99th percentiles of 16 ng/L for women and 34 ng/L for men (22). The gal-3 assay used was the Abbott Architect assay (23) with upper 95th and 99th percentiles of 25.2 ng/mL and 28.4 ng/mL, respectively, in the general population. The gal-3 assay limit of detection is 1.0 ng/mL and its CV is ≤10% at 4.0 to 114.0 ng/mL. The hFABP assay was performed with the Randox assay on a Roche Modular instrument with the following precision estimates: quality control (QC) level 1 = 4.47 µg/L, CV 7.1% from a n = 46 and a serum QC material = 2.44 µg/L, CV 12% from a n = 82. The hFABP assay upper limit of normal is 5.8 µg/L (11).

Statistical Analysis

Only hemodialysis treatments in which cardiac biomarkers were collected were included in the analysis without any imputation of missing data. Descriptive statistics are presented with means [standard deviation (SD)] and medians [(interquartile ranges (IQR)] as appropriate. Multilevel linear regression with individual hemodialysis treatments (level 1) clustered within study participants (level 2) was performed to determine independent predictors of post-dialysis hs-TnI, gal-3, and hFABP. We limited the number of predictors to 1 per 10 participants (24) without stepwise model selection to reduce bias (25). The following variables were forced into the multivariate model: age, sex, ethnicity (Caucasian vs. non-Caucasian), diabetes mellitus, heart failure, coronary artery disease (history of myocardial infarction, coronary artery bypass grafting, percutaneous transcoronary angioplasty), cerebrovascular disease (history of transient ischemic attack or stroke), peripheral vascular disease (history of amputation or stenting/bypass), residual renal function (self-reported urine output ≥250 mL/day), hemodialysis frequency (≤3x or >3x weekly), hemodialysis length (minutes), dialyzer (Revaclear 300/Revaclear 400 vs. Exeltra 190 vs. Toray), ultrafiltration volume (liters), total blood processed (liters), intradialytic hypotension (systolic blood pressure nadir <90 mm Hg) (26), pre-dialysis cardiac biomarker (hs-TnI, gal-3, hFABP) with no interaction terms. The assumptions of linear regression including linearity, homoscedasticity, and normality were assessed. Outliers were assessed using standardized level 1 and 2 residuals and those greater than +3 or less than -3 removed from the analysis. Sensitivity analyses using ultrafiltration rate instead of ultrafiltration volume and hemodialysis time were performed in addition to limiting the cohort to stable hemodialysis patients with censoring for any hospitalization. We did not adjust post-dialysis cardiac biomarkers for hemoconcentration.

All analyses were performed using Stata (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP).

Results

The study cohort is shown in Table 1. The QC results for hs-TnI, gal-3, and hFABP are shown in Supplemental Table 1. Pre-dialysis, post-dialysis, and intradialytic changes (absolute and relative) in hs-TnI, gal-3, and hFABP are shown in Table 2. A Bland Altman plot of pre-dialysis and post-dialysis cardiac biomarkers (in stable hemodialysis patients) is shown in Fig. 1. The median intradialytic changes in hs-TnI, gal-3, and hFABP and their interquartile ranges (IQR) were -1ng/L (IQR -1ng/L to 2 ng/L), 36.3 ng/mL (IQR −27.7 to −46.8 ng/mL), and −19.41 ng/mL (IQR −13.61 to −26.87 ng/mL), respectively. This corresponded to median intradialytic changes of 2.6% (−4.4% to 12.5%), −59.8% (−54.7% to −64.8%), and −35.3% (−28.4% to −42.1%) for each cardiac biomarker.

The percentage of hemodialysis treatments that yielded differences in hs-TnI that exceeded change criteria (e.g., greater than a percentage threshold) is shown in Supplemental Table 2. The percentage of hemodialysis treatments that yielded differences in hs-TnI analytical criteria for myocardial injury over 3 hours (typically the minimum duration of a hemodialysis treatment) including ≥ 4 ng/L (21), ≥12 ng/L (27), and ≥28 ng/L (28) was 17.88%, 7.85%, and 6.54%, respectively.

The multivariable linear regression analyses for post-dialysis hs-TnI, gal-3, and hFABP are shown in Table 3. Patient characteristics (sex, ethnicity, comorbidities, residual renal function) were associated with small changes that were within the analytical error of all 3 cardiac biomarker assays. Ultrafiltration volume was associated with an increase in concentration of hs-TnI, gal-3, and hFABP (mean 0.99 ng/L, 1.05 ng/mL, and 1.9 ng/mL per L ultrafiltration volume respectively, P < 0.001 for all). The total blood processed during hemodialysis was associated with decreases in gal-3 and hFABP concentrations (mean −0.139 ng/mL and 0.072 ng/mL per L of blood, respectively, P < 0.001 for both) but not hs-TnI. Hemodialysis treatment time was associated with decreases in all 3 cardiac biomarkers (mean −0.025 ng/L per min for hs-TnI, −0.009 ng/mL per min for gal-3, −0.011 ng/mL for hFABP per min, P < 0.05 for all). The use of an Exeltra 190 membrane was associated with a decrease in gal-3 and hFABP (mean −2.512 ng/mL and −4.905 ng/mL,respectively, P < 0.001 for both). The presence of intradialytic hypotension was associated with a small decrease in hFABP (mean −0.522, P = 0.031). The pre-dialysis values of all 3 cardiac biomarkers were associated with increases in all post-dialyses (P < 0.001 for all).

There were no differences in results when accounting for influential observations or the inclusion/exclusion of different constructs (e.g., ultrafiltration rate, the definition of intradialytic hypotension) as predictors. When censoring observations for hospitalizations (and thus defining a stable hemodialysis population), sex for both hs-TnI and gal-3, as well as diabetes and coronary artery disease, were no longer statistically significant predictors (Supplemental Tables 3 and 4).

Discussion

In this prospective cohort study of over 3000 hemodialysis treatments in 178 hemodialysis patients, hs-TnI did not change during hemodialysis, while gal-3 and hFABP decreased by a mean of 57.9% and 34.6%, respectively. The linear proportional bias for gal-3 and hFABP shown in Fig. 1 (higher mean pre-dialysis and post-dialysis values were associated with greater absolute intradialytic reductions) is perhaps due to a fixed relative dialyzability (i.e., percentage reduction in biomarkers) across all ranges that is unsaturated without compartmentalization but with variability due to specific dialysis factors such as treatment time, ultrafiltration, total blood processed, and membrane. Pre-dialysis biomarker coefficients of 0.235 ng/mL for gal-3 and 0.637 ng/mL for hFABG that were statistically significant (and both less than 1 ng/mL) support the absolute intradialytic reductions in gal-3 and hFABG being dependent on pre-dialysis values.

Previous studies inconsistently identified intradialytic changes in hs-TnI (29–31). This inconsistency in hs-TnI intradialytic variability may be due to heterogeneity in the timing of measurements, the specific assays used, and hemodialysis prescriptions (32, 33). “Hemodialysis” provides clearance by dialysis (diffusion of small solutes across a concentration gradient), “hemofiltration” provides clearance by filtration (convection of solutes due to solvent drag) and “hemodiafiltration” provides clearance by both mechanisms. The degree of dialysis and filtration during a treatment may influence the clearance of a solute with effect modification by treatment duration, blood flow, dialysate flow and membrane characteristics including flux (permeability).

There was no significant change in pre-dialysis and post-dialysis TnI measured by the Vitros ES assay in a study of 48 asymptomatic patients dialyzed 3x weekly, but no information was provided regarding hemodialysis and dialyzer characteristics (29). In another study (31) the mean reduction in hs-TnI (Architect i1000) and TnI-ultra (ADVIA Centaur) during dialysis was 30% (95% CI 22–37%) and 29% (95% CI 25–44%), respectively, (adjusted for hemoconcentration). This study (31) included 36 clinically stable patients treated by hemodiafiltration 3x weekly of which 91.5% were treated with polysulfone membranes for a mean (SD) duration of 220 (25) minutes with a mean (SD) Kt/V of 1.63 (0.25). Tarapan et al. (34) hs-TnI measured by the ARCHITECT STAT assay decreased by a median of 36% (IQR 8.3–75%) during hemodialysis (also adjusted for hemoconcentration). This cohort included 100 patients treated with HD or HDF 3x weekly with high flux membranes. Those individuals with post-dialysis reductions in hs-TnI paradoxically had lower blood flow and dialysate flow, but there was no difference in fluid removal between groups (dialysis vs. filtration and dialyzer membranes were not compared). No individual patient level meta-analysis or meta-regression using study level factors has evaluated which patient, dialysis, and assay factors are responsible for changes in hs-TnI during dialysis but the results of our study in the context of the literature suggest that hs-TnI is primarily removed by convection.

The lack of hs-TnI variability during hemodialysis in our study supports its use for the diagnosis of acute coronary syndromes peri-dialytically since hs-TnI concentrations changed minimally with a mean of <10% and <15% of hemodialysis patients exceeding the percent change values of 20% and 30%, respectively. Similarly, the estimates for the effect of each predictor of post-dialysis hs-TnI are relatively close to the analytical error of the assay (i.e., most statistically significant beta coefficients are less than 2 ng/L and the CV for hs-TnI was 3%–4% in the study with a mean pre-dialysis hs-TnI of 33.84 ng/L). Ultrafiltration and hemodialysis treatment time are the only exceptions given beta coefficients of 0.991 ng/L/L and -0.025 ng/L/minute and the mean ultrafiltration volume (2.46 L) and treatment time (201 min) in the cohort. Although no patients in this study had evidence of an acute coronary syndrome during hemodialysis, hs-TnI did exceed a percent change value of 30% in almost 10% of the cohort. This may be due to subclinical myocardial injury due to myocardial stunning during hemodialysis that is a well-described phenomenon (3, 35), which was not measured in this study.

Gal-3 was increased in the hemodialysis population in prior studies (14, 16, 36). Consistent with our results, only one study of 23 patients previously reported a decrease in gal-3 concentration (40.7% from 55 to 32.6 ng/mL) in hemodialysis (15). Similarly, hFABP was also increased in patients on hemodialysis but there are no data on intradialytic change (17, 37, 38). Importantly, in contrast to hs-TnI, the degree of intradialytic change in gal-3 and hFABP compared to their assays’ precisions was large. These data suggest the interpretation of both gal-3 and hFABP must consider when the samples were obtained relative to the hemodialysis treatment in which they were collected.

Decreases in gal-3 and hFABP, but not hs-TnI, were associated with the total blood processed and hemodialysis treatment time, suggesting clearance by diffusion and/or absorption by membranes. Given beta coefficients of −0.139 ng/mL/L and −0.072 ng/mL/L and a mean cohort total blood processed of 64.7 (SD 14.2) liters, clearance is potentially a clinically relevant contributor for gal-3 declines during hemodialysis but less so for hFABP (e.g., changes in gal-3 and hFABP of 9.00 ng/mL and 4.66 ng/mL are clinically meaningful given their assays’ precisions). hs-TnI, gal-3, and hFABP are all middle molecules similar to β2-microglobulin (11.8 kDa) with molecular weights of 24 kDa, 30 kDa, and 15 kDa, respectively, and thus are removed predominantly by convection. The finding that total blood processed is as an independent predictor of gal-3 and hFABP declines during hemodialysis is likely due to the use of high flux dialyzer membranes in this cohort, with sieving coefficients >0 for β2-microglobulin and other middle molecules. The fact that hs-TnI does not decrease during dialysis although it is smaller than gal-3 but larger than hFABP suggests that it is not significantly removed by convection and that flux and/or absorption is limited due to possibly charge or that it may exist as larger dimers or trimers in vivo. The use of the Exeltra 190 dialyzer membrane was also associated with declines in gal-3 and hFABP concentrations but not hs-TnI and is presumably related to intrinsic dialyzer properties such as flux or absorption (Supplemental Table 6). However, latent patient characteristics might also be responsible for this finding since at our center individuals with intradialytic hypotension or dialyzer reactions are more likely to be prescribed Exeltra 190 than our standard membranes.

Increases in hs-TnI, gal-3, and hFABP were associated with increasing ultrafiltration volumes indicating hemoconcentration as a potential mechanism. The median time to post-dialysis cardiac biomarker collection in our study was only 7 min (IQR 3–11), which may be too soon for plasma refilling to fully equilibrate intravascular volume. Thus, in clinical settings in which blood collection might be delayed (e.g., if hemodialysis is interrupted and the patient is sent to an acute care setting), the impact of ultrafiltration volume on post-dialysis hs-TnI, gal-3, and hFABP may be less significant. Regardless, given that the mean ultrafiltration was 2.46 L (SD 1.06) in this cohort, the contribution of ultrafiltration volume to any intradialytic cardiac biomarker increase is likely not that clinically significant in most patients.

To the best of our knowledge, this is the largest study of its kind in the hemodialysis population and its strengths include its size and the use of repeated measurements over 6 months with extensive adjustment for patient level and hemodialysis level covariates. However, it involved only a single center, but given its broadly inclusive eligibility criteria, its population is representative of a typical North American hemodialysis unit and should be generalizable to other centers where the specific cardiac biomarkers assays are used and only hemodialysis is performed (with similar membranes). Our results may not apply to settings where hemodiafiltration is predominantly performed or where other types of dialyzers are utilized where the clearance of middle molecules may be enhanced (39, 40). Lastly, our study does not include troponin T, which may exist in a small immunoreactive free form that may be better cleared by both diffusion or convection and whether our results are generalizable to other TnI assays that target different epitopes is unclear.

In summary, gal-3 and hFABP decrease during hemodialysis while hs-TnI does not. All of these cardiac biomarkers appear to increase with increasing ultrafiltration volume, presumably secondary to hemoconcentration, while gal-3 and hFABP appear to decrease with increasing clearance related to the total blood processed and hemodialysis treatment time. hs-TnI is a suitable cardiac biomarker for the diagnosis of acute coronary syndromes peridialytically given its lack of intradialytic variability. However, the degree to which hs-TnI increases during hemodialysis are clinically significant and how they influence short- and long-term patient symptoms and outcomes is uncertain. The roles of gal-3 and hFABP in the hemodialysis population remain undefined.

Supplemental Material

Supplemental material is available at Clinical Chemistry online.

Author Contributions

All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved.

D. Collister, statistical analysis; M. Walsh, statistical analysis, provision of study material or patients.

Authors’ Disclosures or Potential Conflicts of Interest

Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:

Employment or Leadership

M. Walsh, Ontario Renal Network.

Consultant or Advisory Role

P. Kavsak, Beckman, Roche, Siemens; P.J. Devereaux, Roche Diagnostics, Bayer.

Stock Ownership

None declared.

Honoraria

P. Kavsak, Beckman, Roche, Siemens.

Research Funding

This project was supported by funding from the Hamilton Academic Health Sciences Organization. Abbott Diagnostics and Randox Laboratories donated the hs-TnI, gal-3, and hFABP assays used in this study. D. Collister is supported by a Kidney Research Scientist Core Education and National Training Program Post-doctoral Fellowship award. M. Walsh is supported by a Mid Career Investigator Award from McMaster University.

Expert Testimony

None declared.

Patents

McMaster has filed patents with P. Kavsak listed as an inventor in the acute cardiac biomarker field.

Role of Sponsor

The funding organizations played a direct role in the final approval of manuscript. The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation of manuscript.

References

Nonstandard Abbreviations

 
• CV

  coefficient of variation

 
• gal-3

  galectin 3

 
• hFABP

  heart-type fatty acid binding protein

 
• hs-TnI

  high sensitivity troponin I

 
• IQR

  interquartile range

 
• QC

  quality control

 
• SD

  standard deviation