To the Editor,
The diagnosis of acute coronary syndromes (ACS) is challenging in hemodialysis (HD) patients. Chronic HD patients with acute myocardial infarction (AMI) are less likely to present with chest pain (44.4% versus 68.3%) or show ST elevation (19.1% versus 35.9%) [1] but often show unspecific symptoms, such as hypotension (28.7%) or nausea/vomiting (11.7%), especially during dialysis treatments [2].
Thus, diagnosis of AMI may frequently be delayed or missed due to atypical presentation and relies heavily on biomarkers [i.e. cardiac troponins (cTn)]. Reports on cardiac troponin T (cTnT) kinetics during HD are heterogeneous [3]. Likewise, reports on cardiac troponin I (cTnI) are inconsistent [4, 5] and baseline levels in HD patients are commonly above established recommended thresholds. Importantly, dialysis membranes and modalities differ significantly in permeability for larger molecules: “low-flux” membranes with permeability for molecules with a molecular mass of up to 5 kDa, “high-flux” up to 20 kDa and the newest, “medium cut-off” (MCO) membranes, up to 45 kDa. With molecular masses of approximately 39 kDa and 26 kDa for cTnT and cTnI, respectively, clearances differ depending on the type of membrane and amount of filtration used in dialysis.
Currently, neither the European Society of Cardiology (ESC) [6] nor American Heart Association (AHA) [7] guidelines specify the diagnostic algorithm and use of biomarkers for patients during HD. A consensus report from the SONG-HD MI Expert Working group concluded there to be insufficient evidence for standard dialysis to impact the diagnosis of AMI [8]. In this study, we aimed to explore intradialytic cTn changes with routinely used HD membranes and treatment modalities to aid in the interpretation of cTn levels in dialysis patients.
Detailed methods are provided in the Supplementary data. In brief, in this randomized, controlled crossover study, asymptomatic, clinically stable patients at least 18 years of age on chronic HD were randomized to a sequence of one treatment session with low-flux HD, high-flux HD, hemodiafiltration (HDF) and MCO HD. Cardiac troponins were measured before dialysis, after 1 h and immediately after HD to examine changes of cardiac troponins.
Twenty patients were randomized, but one patient was excluded from analysis due to non-ST elevation ACS during dialysis with substantial distortion on the absolute values of the results, thus a per protocol analysis including 19 patients (47.4% female) with a mean age of 65.5 ± 13.4 years and a median of 19 months (min. 3, max. 165) on dialysis was conducted. Sixty-eight percent had a history of coronary artery disease (CAD) and 36.8% had previously suffered an AMI. Baseline characteristics are provided in Supplementary data, Table S1.
In mixed model analysis for the relative difference in cTnT, no significant sequence or period effects were observed for relative changes from baseline to 1 h (sequence: P = .45, period: P = .97).
Kinetics of cTn are shown in Fig. 1 and Table 1. The effect of membrane on the relative change on cTnT differed significantly (P < .001). Significantly different relative changes after 1 h were observed for MCO [least square mean (LSM) −21.9; 95% confidence interval (CI) −27.3 to −16.6%] compared with low-flux (LSM +2.2; 95% CI −3.2 to 7.5%, P < .001) and MCO to high-flux (LSM −6.8; 95% CI −12.2 to −1.5%, P < .001). No difference was observed for MCO versus HDF treatment with high-flux membrane (LSM −21.2; 95% CI −26.6 to −15.7%, P = .81). Similar results were observed post-HD.
Figure 1:
(A) Relative changes (presented as LSM ± SE) of cTnT according to treatment modality. (B) Absolute changes in pg/mL of cTnT according to treatment modality after 1 h of treatment and post-HD. (C) Relative changes of cTnI (presented as LSM ± SE) according to different treatment modalities. BL, baseline; SE, standard error. ***P < .001 refer to differences to MCO cTn levels according to treatment modalities.
Table 1:
Troponin kinetics according to membrane characteristics and treatment modality.
| High-flux HD | Low-flux HD | MCO | HDF | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Troponin T | Baseline (pg/mL), mean (± SD) | 0 h | 101.3 (± 74.4) | 107.7 (± 82.1) | 105.4 (± 71.7) | 103.2 (± 67.1) | ||||
| Absolute differences (pg/mL), LSM (95% CI) | 1 h | −6.4 (−12.8 to 0.0) | P = .049 | 2.3 (−4.1 to 8.6) | P = .48 | −21.2 (−27.6 to −14.8) | P < .001 | −20.2 (−26.8 to −13.7)a | P < .001 | |
| Post-HD | −7.7 (−17.2 to 1.8)a | P = .11 | 0.3 (−9.0 to 9.6) | P = .96 | −31.4 (−40.9 to −21.9) | P < .001 | −27.8 (−37.1 to −18.5)a | P < .001 | ||
| Relative difference (%), LSM (95% CI) | 1 h | −6.8 (−12.2 to −1.5) | P = .013 | 2.2 (−3.2 to 7.5) | P = .42 | −21.9 (−27.3 to −16.6) | P < .001 | −21.2 (−26.6 to −15.7) | P < .001 | |
| Post-HD | −8.3 (−15.0 to −1.7)a | P = .015 | 3.5 (−3.0 to 9.9) | P = .28 | −33.0 (−39.6 to −26.4) | P < .001 | −28.9 (−35.3 to −22.4) | P < .001 | ||
| Troponin I | Baseline (pg/mL), mean (± SD) | 0 h | 112.4 (± 159.3) | 78.1 (± 87.2)a | 73.8 (± 89.9) | 89.3 (±145.5)a | ||||
| Absolute differences (pg/mL), LSM (95% CI) | 1 h | −10.3 (−81.5 to 60.8) | P = .77 | 40.5 (−32.2 to 113.2)a | P = .27 | 53.6 (−17.5 to 124.7) | P = .14 | 48.8 (−26.9 to 124.6)a | P = .20 | |
| Post-HD | 20.9 (−66.4 to 108.0)a | P = .63 | 102.1 (15.4 to 188.8)a | P = .022 | 47.7 (−37.1 to 132.4) | P = .26 | −1.0 (−88.8 to 86.8)a | P = .98 | ||
| Relative difference (%), LSM (95% CI) | 1 h | 201.8 (−215.4 to 618.9) | P = .34 | 223.6 (−202.9 to 650.1)a | P = .30 | 295.7 (−121.3 to 712.6) | P = .16 | 405.2 (−38.9 to 849.3)a | P = .07 | |
| Post-HD | 165.3 (−164.6 to 495.2)a | P = .32 | 462.0 (133.8 to 790.1)a | P = .007 | 149.1 (−171.7 to 469.9) | P = .35 | 112.7 (−219.6 to 445.0) | P = .50 | ||
| Troponin T corrected for hemo-concentration | Absolute differences (pg/mL), LSM (95% CI) | 1 h | −16.8 (−26.7 to −6.9) | P = .001 | −3.4 (−13.4 to 6.6) | P = .40 | −26.4 (−36.1 to −16.7) | P < .0001 | −25.2 (−35.8 to −14.6)a | P < .0001 |
| Post-HD | −23.1 (−37.0 to −9.2)a | P = .002 | −10.1 (−24.0 to 3.9) | P = .15 | −40.9 (−54.6 to −27.4) | P < .0001 | −35.7 (−49.5 to −21.9)a | P < .0001 | ||
| Relative difference (%), LSM (95% CI) | 1 h | −13.9 (−19.7 to −8.2) | P < .0001 | −2.9 (−8.7 to 2.9) | P = .32 | −27.1 (−32.7 to −21.5) | P < .0001 | −26.3 (−32.6 to −20.0) | P < .0001 | |
| Post-HD | −21.4 (−28.5 to −14.2)a | P < .0001 | −8.4 (−15.5 to −1.2) | P = .02 | −41.9 (−48.8 to −35.0) | P < .0001 | −36.9 (−43.9 to −29.8) | P < .0001 | ||
| Troponin I corrected for hemoconcentration | Absolute differences (pg/mL), LSM (95% CI) | 1 h | −25.5 (−88.6 to 37.6) | P = .42 | 35.5 (−29.9 to 100.8)a | P = .28 | 45.6 (−16.1 to 107.3) | P = .14 | 18.4 (−53.8 to 90.5)a | P = .61 |
| Post-HD | 6.1 (−74.5 to 86.8)a | P = .88 | 83.1 (0.0 to 166.2)a | P = .05 | 28.1 (−47.6 to 103.8) | P = .46 | 8.5 (−74.0 to 91.0)a | P = .84 | ||
| Relative difference (%), LSM (95% CI) | 1 h | 208.1 (−86.8 to 503.0) | P = .16 | 219.4 (−86.0 to 524.8)a | P = .15 | 268.3 (−20 to 556.6) | P = .06 | 69.6 (−267.7 to 406.9)a | P = .67 | |
| Post-HD | 146.80 (−170.8 to 464.4)a | P = .36 | 450.2 (123.2 to 777.2)a | P = .008 | 98.32 (−199.6 to 396.3) | P = .51 | 110.44 (−214.2 to 435.1) | P = .49 | ||
LSMs derived by linear mixed model analysis.
a n = 17–18 due to missing data.
Comparisons and P-values refer to differences relative to baselines.
For absolute changes, LSM for MCO were −21.2 (95% CI −27.6 to −14.8 pg/mL), −6.4 (95% CI −12.8 to −0.0 pg/mL) for high-flux, −20.2 (95% CI −26.8 to −13.7 pg/mL) for HDF treatment and +2.3 (95% CI −4.1 to 8.6 pg/mL) for low-flux hemodialysis after 1 h. There were no significant effects of sequence, period or membrane effects observed for relative changes of cTnI from baseline to 1 h (sequence: P = .14, period: P = .68, membrane: P = .91) and from baseline to post-HD (sequence: P = .58, period = 0.23, membrane = 0.41).
The results were concordant but more pronounced when accounting for ultrafiltration-induced hemoconcentration, with even higher observed relative and absolute changes of cTnT for high-flux HD, HDF and MCO HD, while the increase of cTnT seen in low-flux HD vanished. After correction for hemoconcentration, there was a statistically significant increase of cTnI seen in low-flux HD (Table 1).
In this randomized crossover trial we found significant decreases of cTnT with MCO and HDF, and smaller decreases with high-flux HD but not with low-flux HD in clinically stable patients. These data support that cTnT is cleared significantly by high-flux and MCO HD, as well as by HDF. The increase seen in low-flux is primarily due to hemoconcentration in the setting of absent cTnT clearance, and disappeared when correcting for hemoconcentration using pre/post-hematocrit [9]. If hemoglobin concentrations, hematocrit or the relative blood volumes are available at measuring times, the increase can be predicted. The effect of hemoconcentration is also present in other treatment modes and one can discuss that true intradialytic reductions are probably even larger than derived from uncorrected concentrations (Table 1). However, while hemoconcentration artificially raises cTnT levels in low-flux HD, in the clinical setting of suspected ongoing myocardial ischemia, adjusting for hemoconcentration will not be feasible and there are no data on whether this will alter test characteristics. A measurement 1 h after dialysis would show the equilibrated concentrations and reveal effects of hemoconcentration as well. Increases of cTn concentrations post-HD, indicating equilibration, have been described previously [10].
A possible reason for the reduction of cTnT being more pronounced with MCO compared with HDF is that, first, at low blood flow rates, as observed in our study, MCO seems to be more effective than HDF in terms of larger middle molecule clearances [11], and second, because total convective volume was rather low in our study.
There was no clear trend in cTnI kinetics and values varied substantially, which does not support clinical use of cTnI in this setting. This is surprising and requires further clarification, as cTnI is smaller than cTnT and a larger drop would therefore be expected with cTnT in high-flux treatments. Insignificant clearance of cTnI could also be explained by a high net negative charge, possibly caused by high phosphorylation [12], by protein binding and ultrafiltration-induced hemoconcentration, or membrane adsorption. After a cardiac insult (non-ST elevation myocardial infarction), cTnI increases faster than cTnT in a regular population [13]. Since myocardial stunning may lead to a rise of cTn, and we did not assess the occurrence of stunning, one might assume that there might be a distinct difference in release of cTnI and cTnT, with extrarenal clearance [14], which may contribute to the large intraindividual changes during each session. Given the multiple reasons for cTnT elevation in HD patients, a more specific biomarker for myocardial ischemia is highly desirable in this patient population.
Considering the algorithms of the 2023 ESC Guidelines [6], a 1 h delta of cTnT >5 pg/mL would be missed in most patients. Furthermore, a delta of >20% of cTn, as recommended by the standardised outcomes in nephrology group–haemodialysis (SONG-HD) Expert Working group [8], occurs without evidence of ACS when MCO membranes or HDF are applied. Any uncorrected increase in cTnT during dialysis (except low-flux HD) would be sensitive but not specific for AMI. While higher baseline cTnT levels in renal patients are well-described, the influence of HD on cTnT may frequently not be considered and should be included when consulting cardiology for a patient evaluation for ACS.
The main limitation is the patient population, a sizable proportion of whom had preexisting CAD, and which deliberately included exclusively stable HD patients, without suspected ongoing myocardial ischemia, to study the effect of dialysis without having to account for differences in the myocardial release of troponins. Thus, the generalizability for other patient populations is limited, specifically for patients with ACS.
Taken together, current diagnostic algorithms cannot be uncritically applied to assess AMI during hemodialysis sessions and troponin kinetics need to be interpreted cautiously in concert with clinical, electrocardiogram and imaging parameters in patients undergoing HD.
Supplementary Material
Patients’ baseline characteristics and methods are provided as part of the Supplementary data.
ACKNOWLEDGEMENTS
We thank the dialysis staff for their important help in obtaining data.
Contributor Information
Michael Kolland, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
Jascha Amenitsch, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
Nikolaus Schreiber, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
Noemi Ginthör, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
Max Schuller, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
Regina Riedl, Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz, Graz, Austria.
Peter P Rainer, Department of Internal Medicine, Division of Cardiology, Medical University of Graz, BioTechMed Graz, Graz, Austria.
Daniel Schneditz, Division of Physiology, Otto Loewi Research Center, Medical University of Graz, Graz, Austria.
Tobias Niedrist, Clinical Institute of Medical and Chemical Laboratory Diagnostics, Medical University of Graz, Graz, Austria.
Kathrin Eller, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
Benedikt Krietemeyer, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
Alexander R Rosenkranz, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
Alexander H Kirsch, Department of Internal Medicine, Division of Nephrology, Medical University of Graz, Graz, Austria.
FUNDING
The authors received no external funding.
AUTHORS’ CONTRIBUTIONS
M.K., A.H.K., A.R.R., D.S. and R.R. designed and conceived the study. J.A., N.G.O.S., M.K., A.H.K., B.K., K.E. and M.S. acquired patients’ data. M.K. wrote the first draft of the manuscript. R.R. and N.G.O.S. performed statistical analysis and designed figures. T.N. performed laboratory analysis. All authors contributed to the acquisition and interpretation of the data and provided critical revision of the manuscript for important intellectual content.
DATA AVAILABILITY STATEMENT
Data available upon reasonable request to the corresponding author.
CONFLICT OF INTEREST STATEMENT
M.K. and A.R.R. received speaker fees from Baxter (Deerfield, IL, USA). A.H.K. has received speaker fees, travel support and research support and has served on advisory boards for Baxter. D.S. is co-inventor of patents in the field of blood volume and bioimpedance applications in hemodialysis and member of the American Renal Associates research board. J.A., N.G.O.S., N.G., M.S., R.R., P.P.R., T.N., K.E. and B.K. declare no conflict of interest.
REFERENCES
- 1. Herzog CA, Littrell K, Arko Cet al. Clinical characteristics of dialysis patients with acute myocardial infarction in the United States: a collaborative project of the United States Renal Data System and the National Registry of Myocardial Infarction. Circulation 2007;116:1465–72. 10.1161/CIRCULATIONAHA.107.696765 [DOI] [PubMed] [Google Scholar]
- 2. Ali M, Ejaz A, Iram Het al. Frequency of intradialytic complications in patients of end-stage renal disease on maintenance hemodialysis. Cureus 2021;13:e12641. 10.7759/cureus.12641 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Hunderup MM, Kampmann JD, Kristensen FBet al. The short-term effect of hemodialysis on the level of high-sensitive cardiac troponin T – a systematic review. Semin Dial 2023; 1–7. 10.1111/sdi.13178 [DOI] [PubMed] [Google Scholar]
- 4. Deleaval P, Descombes E, Magnin JLet al. Differences in cardiac troponin I and T levels measured in asymptomatic hemodialysis patients with last generation immunoassays. Nephrol Ther 2006;2:75–81. 10.1016/j.nephro.2005.11.003 [DOI] [PubMed] [Google Scholar]
- 5. Tarapan T, Musikatavorn K, Phairatwet Pet al. High sensitivity troponin-I levels in asymptomatic hemodialysis patients. Ren Fail 2019;41:393–400. 10.1080/0886022X.2019.1603110 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Byrne RA, Rossello X, Coughlan JJet al. 2023 ESC guidelines for the management of acute coronary syndromes. Eur Heart J 2023; 44:3720–826. 10.1093/eurheartj/ehad191 [DOI] [PubMed] [Google Scholar]
- 7. Gulati M, Levy PD, Mukherjee Det al. 2021 AHA/ACC/ASE/CHEST/SAEM/SCCT/SCMR guideline for the evaluation and diagnosis of chest pain: a report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation 2021;144:e368–454. 10.1161/CIR.0000000000001029 [DOI] [PubMed] [Google Scholar]
- 8. O'Lone E, Apple FS, Burton JOet al. Defining myocardial infarction in trials of people receiving hemodialysis: consensus report from the SONG-HD MI Expert Working group. Kidney Int 2023;103:1028–37. 10.1016/j.kint.2023.02.033 [DOI] [PubMed] [Google Scholar]
- 9. Schneditz D, Putz-Bankuti C, Ribitsch Wet al. Correction of plasma concentrations for effects of hemoconcentration or hemodilution. ASAIO J 2012;58:160–2. 10.1097/MAT.0b013e318243660f [DOI] [PubMed] [Google Scholar]
- 10. Chen M, Gerson H, Eintracht Set al. Effect of hemodialysis on levels of high-sensitivity cardiac troponin T. Am J Cardiol 2017;120:2061–4. 10.1016/j.amjcard.2017.08.026 [DOI] [PubMed] [Google Scholar]
- 11. Kirsch AH, Lyko R, Nilsson LGet al. Performance of hemodialysis with novel medium cut-off dialyzers. Nephrol Dial Transplant 2017;32:165–72. 10.1093/ndt/gfw310 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Labugger R, Organ L, Collier Cet al. Extensive troponin I and T modification detected in serum from patients with acute myocardial infarction. Circulation 2000;102:1221–6. 10.1161/01.cir.102.11.1221 [DOI] [PubMed] [Google Scholar]
- 13. Pickering JW, Young JM, George PMet al. Early kinetic profiles of troponin I and T measured by high-sensitivity assays in patients with myocardial infarction. Clin Chim Acta 2020;505:15–25. 10.1016/j.cca.2020.02.009 [DOI] [PubMed] [Google Scholar]
- 14. Muslimovic A, Friden V, Tenstad Oet al. The liver and kidneys mediate clearance of cardiac troponin in the rat. Sci Rep 2020;10:6791. 10.1038/s41598-020-63744-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Patients’ baseline characteristics and methods are provided as part of the Supplementary data.
Data Availability Statement
Data available upon reasonable request to the corresponding author.