The cardiovascular–dialysis nexus: the transition to dialysis is a treacherous time for the heart

Abstract

Chronic kidney disease (CKD) patients require dialysis to manage the progressive complications of uraemia. Yet, many physicians and patients do not recognize that dialysis initiation, although often necessary, subjects patients to substantial risk for cardiovascular (CV) death. While most recognize CV mortality risk approximately doubles with CKD the new data presented here show that this risk spikes to >20 times higher than the US population average at the initiation of chronic renal replacement therapy, and this elevated CV risk continues through the first 4 months of dialysis. Moreover, this peak reflects how dialysis itself changes the pathophysiology of CV disease and transforms its presentation, progression, and prognosis. This article reviews how dialysis initiation modifies the interpretation of circulating biomarkers, alters the accuracy of CV imaging, and worsens prognosis. We advocate a multidisciplinary approach and outline the issues practitioners should consider to optimize CV care for this unique and vulnerable population during a perilous passage.

Graphical abstract

The initiation of dialysis superimposes new CV challenges on the substrate of chronic kidney disease (CKD). These factors include increased myocardial oxygen demand, for example, due to access-associated increase in cardiac output, and hypotension due to volume shifts during dialysis. Intra-dialytic hypotension can also decrease oxygen supply to the myocardium, particularly a lower diastolic pressure head for coronary artery perfusion augmented by arteriosclerosis affecting both the intimal and medial layers of conduit arteries. Hypotension superimposed on macro- and micro-vascular disease can lead to the myocardial stunning documented during and immediately post-dialysis. In addition, dialysis involves inflammatory insults from innate immune activation due to contact with the dialysis membrane and catheters. Rapid fluid and electrolyte shifts during dialysis can predispose towards arrhythmia, contributing to the augmented sudden death that follows the initiation of dialysis. All of these new factors related to dialysis initiation superimpose upon the chronic anaemia, the uraemic cardiomyopathy, and uraemic toxins associated with CKD.

Introduction

Worldwide, more than 700 000 patients will start chronic dialysis this year, and this number is expected to double in the next 20 years.¹ Both the European Society of Cardiology and American Heart Association Guidelines recognize chronic kidney disease (CKD) as a marker of heart disease, but neither articulates that starting dialysis for end-stage kidney disease (ESKD) may be a period of high risk because the cardiovascular (CV) death rate unexpectedly surges in the first 4 months of dialysis therapy, as demonstrated by the new data presented here (Figure 1).^3–5

Figure 1.

Standardized mortality rates in the chronic kidney disease and dialysis population, Medicare 5% CKD sample (2011–2014). Our analysis of a 5% random sample of Medicare claims highlights ‘the spike’ in cardiovascular mortality that occurs when dialysis starts that extends into the first 4 months of chronic dialysis. Direct standardized rates for age, gender, and race were calculated as per methods outlined by Ahmad et al.² See Supplementary material online for details regarding the generation of this figure.

Nephrologists only recently recognized this spike in CV mortality because traditional studies followed patients after Day 90 when patients become Medicare eligible. Other practitioners who co-manage these vulnerable patients may not share this awareness of ‘the spike’ mandating its broader recognition. Starting dialysis entails a major shift in the pathophysiology of disease that changes the clinical CV manifestations of heart disease that underlies this peak in CV mortality.^{4  ,  6}

This review article summarizes how this pathophysiological shift changes the presentation, diagnosis, and management of CV disease the moment CKD patients start dialysis.

We advocate a multi-disciplinary approach to mitigate the spike in CV mortality precipitated by starting chronic dialysis, concordant with a recent Kidney Disease Improving Global Outcomes (KDIGO) Controversies Conference.⁷

Epidemiology surrounding the start of dialysis

Almost 60% of the US population will develop CKD (Table 1) with consequently enhanced CV mortality.^{8  ,  9} The lifetime risk of ESKD is 3.6%, which is higher than death from suicide, automobile accidents, firearms, or cancer of either breast, colon, or prostate.^{8  ,  10  ,  11}

Table 1.

Chronic kidney disease classification and grade

CKD grade	Classification	Glomerular filtration rate (mL/min)
Stage 1–2	Mild	60–89
Stage 3	Moderate	30–59
Stage 4	Severe	15–29
Stage 5	Kidney failure	<15
Dialysis	End-stage renal failure	Dialysis

CV mortality during the first 4 months following the start of chronic dialysis exceeds by 20-fold that of age-matched individuals in the general US population.^{3  ,  4  ,  6  ,  12  ,  13} We present new data in Figure 1 illustrating this critical spike in mortality when dialysis begins where the combined rate of CV death and sudden cardiac death (SCD) is 14 events per 100 patient-years during these first 4 months. CV/SCD mortality peaks during Month 2 of dialysis, yielding an annualized rate of 20.4 deaths per 100 patient-years, then subsequently plateaus to a steady state of 7 deaths per 100 patient-years at Month 8. The combined CV and SCD death risk in the first 4 months is remarkably 50% higher than Stage 5 CKD, an equally comorbid population who are not yet on dialysis but quickly approaching ESKD.

Dialysis initiation changes the causes of CV death. Ischaemic heart disease causes the majority of deaths in CKD. Once dialysis starts, SCD risk doubles and becomes the most common cause of death, suggesting a shift in mechanisms (Figure 1).¹⁴ The rate of SCD exceeds by 3.5-fold that of pre-dialysis CKD 4 patients. Myocardial infarction (MI) also peaks during this period (Figure 2). The incidence of MI is 0.27 events per 100 patient-years in the general population, ranges from 1 to 2 events per 100 patient-years in mild CKD, increases to >3 events per 100 patient-years on dialysis, and ‘spikes’ to 34 events per 100 patient-years in the first month of chronic renal replacement therapy, which is >11 times higher than pre-dialysis CKD 4 patients (Figure 2).^{15  ,  16} Stroke rates parallel those of MI possibly from repeated cerebral hypoperfusion, anticoagulation, and inflammation from dialysis.¹⁷ The incident stroke and transient ischaemic attack rate in the year before dialysis of 0.25 event per 100 patient-months increases 7-fold on dialysis initiation and then declines in the following month.¹⁷

Figure 2.

Standardized acute coronary syndrome rates in the chronic kidney disease and dialysis population, Medicare 5% CKD sample (2011–2014). Our analysis of a 5% random sample of Medicare claims shows the rate of myocardial infarction, while is already heightened in chronic kidney disease, and peaks right after dialysis starts. Direct standardized rates for age, gender, and race were calculated as per methods outlined by Ahmad et al.² See supplementary material for details regarding the generation of this figure.

Compared to the general population, both ST-segment elevation myocardial infarction (STEMI) and non-ST-segment elevation myocardial infarction (NSTEMI) increase in CKD. While STEMI rates remain flat around 0.2 events per 100 patient-years for all levels of CKD, NSTEMI rates increase as glomerular filtration rate (GFR) decreases, suggesting that the presentation, mechanisms, and diagnosis of acute coronary syndrome (ACS) change during the progression of renal dysfunction.¹⁸ Moreover, kidney disease impairs the clearance of biomarkers of cardiac injury e.g. troponin, perhaps heightening the detection of NSTEMI.¹⁹

Presentation: detection of cardiovascular disease in patients starting dialysis

ESKD complicates the presentation of heart disease for patients starting dialysis by impairing the performance of biomarker assays, decreasing the sensitivity of non-invasive testing, and increasing the risk of procedural complication. Dialysis patients with coronary artery disease (CAD) may not exhibit typical symptoms due to coexistent diabetes, uraemic neuropathy, and limited exercise capacity. Overall, these challenges may hinder accurate diagnoses, delay vital therapies, and complicate testing for these patients at the very time of greatest need for optimal CV care.

Imaging modalities

Over 50% of patients starting dialysis have angiographically obstructive CAD, with multi-vessel involvement in 25–40%.^{20  ,  21} Dobutamine stress ECHO (DSE) predicts severe CAD in dialysis patients (Table 2) even though ∼50% of patients starting dialysis have left ventricular (LV) hypertrophy (LVH) that blunts the detection of wall-motion abnormalities.^{22  ,  23} Myocardial perfusion imaging (MPI) has lower sensitivity in dialysis patients than in the general population due to reduced coronary vasodilatory response and balanced ischaemia because of multi-vessel involvement (Table 2).^24–26 Frequent baseline electrocardiogram (ECG) abnormalities and inadequate functional capacity limit the predictive ability of exercise electrocardiography (sensitivity = 35%, specificity = 60%) in the dialysis population.²⁷ Only two small dialysis studies evaluate the performance of coronary artery calcification (CAC) scoring for detecting CAD: Rosario et al. reported a sensitivity = 0.64, specificity = 0.65.^{24  ,  28} Winther et al.²⁹ reported a sensitivity = 0.67, specificity = 0.77 (27) Similarly, data on coronary computed tomography angiography is sparse: 93% sensitivity and 63% specificity in 138 advanced CKD patients who underwent concurrent coronary arteriography.

Table 2.

Performance characteristics of fifth-generation hs-TNI assay, dobutamine stress ECHO, and myocardial perfusion scintigraphy CAD in the dialysis and general population

	Sensitivity		specificity
	Dialysis population	General population	Dialysis population	General population
Fifth-generation hs-TNI (27)	0.41	0.93	1.00	0.94
Dobutamine stress ECHO (35)	0.79	0.85	0.89	0.77
Myocardial perfusion scintigraphy (35)	0.74	0.87	0.70	0.64

DSE, MPI, or CAC testing in patients starting dialysis can effectively risk stratify subjects for future CAD events.³⁰ However, results from ISCHEMIA-CKD indicate that advanced CKD and dialysis patients with a positive non-invasive stress test do not benefit from angiography/revascularization additional to optimal medical therapy; moreover, angiography/revascularization causes harm where the risk of stroke was 3.76 higher than the medication alone group.³¹ A surprising 31.1% of angiograms showed no obstructive lesions, suggesting that epicardial coronary artery stenoses alone do not account for excess mortality in end-stage renal disease. Thus, these invasive procedures generally should be reserved for severe cases where patients become refractory to maximal medical management, have decompensated heart failure, with recent ACS, or have an LV ejection fraction < 35% who were excluded from the ISCHEMIA-CKD trial.

For severe and refractory symptomatic patients approaching dialysis [estimated GFR (eGFR) < 30 mL/min], a positive DSE or MPI can present a clinical dilemma. In the ISCHEMIA-CKD trial, angiogram +/- revascularization resulted in 7.9% risk for contrast nephropathy and 48% risk increase in death and dialysis initiation when compared to medical treatment alone. The risk of dialysis initiation after coronary artery bypass grafting (CABG) was 11–12.5%. Thus, invasive management can provoke contrast-induced nephropathy and CABG can hasten dialysis start and premature exposure to the CV spike. Risk-scoring tools for contrast-induced acute kidney injury and dialysis can inform decisions about the risk vs. benefit of invasive procedures (Table 3).³²

Table 3.

Risk prediction of contrast-induced nephropathy after percutaneous coronary intervention³²

Risk factor	Scoring
Hypotension Systolic pressure <80 mmHg for >1 h, AND Patient requires inotropic support or an intra-aortic balloon pump within 24 h after the procedure	5 points
Use of intra-aortic balloon pump	5 points
Heart failure: New York Heart Association class III or IV, OR History of pulmonary oedema	5 points
Elderly: >75 years old	4 points
Anaemia: haematocrit <39% for men or <36% for women	3 points
Diabetes	3 points
Volume of contrast medium	1 point per 100 cc
Serum creatinine level >1.5 mg/dL (133 μmol/L)   OR   Estimated GFR <60 mL/min/1.73 m2 body surface area	4 points   2 points for GFR = 40–60 mL/min   4 points for GFR = 20–39 mL/min   6 points for GFR < 20 mL/min

Total score	Risk of contrast-induced nephropathya	Risk of dialysis
≤5	7.5%	0.04%
6–10	14%	0.12%
11–16	26.1%	1.09%
≥16	57.3%	12.6%

^a>25% or >5 mg/dL (44 umol/L) increase in serum creatinine.

Circulating biomarkers

Troponin concentrations increase immediately after haemodialysis (HD), suggesting that the procedure itself can induce haemoconcentration and/or cardiac injury from rapid fluid removal and coronary hypoperfusion in a population with predisposing obstructive or microvascular CAD^{33  ,  34}; furthermore, both cardiac troponin T and cardiac troponin I fragments depend on renal clearance, which may contribute to raised serum troponin concentrations due to myocardial injury.¹⁹ These mechanisms may also contribute to the ‘spike’ in MI, NSTEMI, and stroke following dialysis initiation (Figure 2).¹⁷

Newer fifth-generation troponin testing (hs-TNI) detects levels 100-fold lower than traditional assays, which complicates the diagnostic interpretation of a positive troponin for MI in dialysis patients.³⁵ Asymptomatic dialysis patients already have elevated hs-TNI (>99th percentile values) from increased ventricular pressure, coronary microvascular dysfunction, anaemia, hypotension, and myocardial uraemic toxicity with impaired renal hs-TNI clearance.^{34  ,  36} These mechanisms decrease the test specificity for ACS in this population.^{37  ,  38} (Table 2). As such, The 2018 Universal Definition of MI³⁹ requires a > 20% increase in serial hs-TNI values in symptomatic dialysis patients with EKG changes for ACS, although this strategy has not undergone rigorous testing in trials.^40–44

Patients starting dialysis have high proBNP and B-type natriuretic peptide (BNP) levels due to reduced renal clearance, prevalent LVH, and fluid retention from advanced CKD. ProBNP (more than BNP) are cleared by the kidney and dialysis, further increasing inter- and intra-patient variability,^{45  ,  46} such that establishing cut-off values for these biomarkers remains difficult, limiting their clinical utility in dialysis patients.⁴⁷

Pathophysiology: the mechanisms of cardiovascular disease shift when dialysis starts

Myocardial stunning

Haemodialysis is associated with intra-dialytic myocardial stunning.⁴⁸ In 70 HD patients, echocardiograms performed before, during, and after HD determined the direct effect of dialysis on cardiac function. Surprisingly, 64% of patients had new or worsening regional wall motion abnormalities (RWMA) during dialysis, indicating that the procedure can precipitate regional ischaemia. Furthermore, such RWMA associated with death or a major CV event 1 year later.⁴⁸ In a follow-up study, 12 patients underwent intradialytic cardiac magnetic resonance scanning, which showed immediate worsening of LV function (cardiac and stroke volume index), peak systolic circumferential/longitudinal strain, and decreasing myocardial perfusion during HD.⁴⁹ These studies link HD to cardiac dysfunction.

Other mechanisms also contribute to the excess CV death with dialysis initiation. HD rapidly removes fluid, potassium, calcium, bicarbonate, and magnesium from the intravascular space. Electrolyte shifts can trigger arrhythmias and drive excess SCD peri-dialysis. Excessive ultrafiltration decreases intravascular volume, reducing coronary perfusion, while the extracorporeal circuit subsumes approximately one unit of blood, reducing the already impaired oxygen-carrying capacity caused by the chronic anaemia commonly encountered in ESKD. Further, the creation of an arterio-venous shunt for dialysis access also reduces total peripheral resistance and diastolic pressure that drives coronary perfusion, and increases cardiac output, increasing myocardial oxygen demand.^{50  ,  51} The high prevalence of obstructive coronary plaques further impedes coronary blood flow.⁵² In non-CKD patients, cardiac injury doubles and CV mortality increases 30% with a diastolic blood pressure (DBP) <60 mmHg relative to 80–89 mmHg.⁵³ Twenty percent of dialysis patients have a pre-dialysis DBP < 65 and an additional 17% of patients will become hypotensive during HD.^{54  ,  55} Indeed, low DBP or intra-dialytic hypotension are associated with increased CV mortality.^54–56 Thus, aggressive blood pressure (BP) targets (e.g. 120/70 mmHg) derived from landmark non-dialysis trials such as SPRINT may not apply during dialysis.⁵⁷ In fact, a dialysis randomized controlled trial (RCT) showed the opposite: lowering pre-dialysis BP to 143/74 increased CV events by 18% when compared to a pre-dialysis BP of 156/82 mmHg.⁵⁸ These data collectively suggest that BP targets differ in dialysis, and highlight the pitfalls of extrapolating results of general outcome trials to patients starting dialysis.

Inflammation

Systemic inflammation links strongly to atherosclerosis and subsequent CV events in both the general and dialysis population.^{59  ,  60} Thirty to sixty percent of US and European dialysis patients have C-reactive protein (CRP) levels > 3 mg/L, a level designated as high risk.^{61  ,  62} Exceeding this cut-off point is associated with a doubling of CV mortality risk in the general population.⁶³ Contact of blood with the dialysis membrane can trigger innate immune activation. A single HD session recirculates 100 litres of patient whole blood against 1.2 m² of polymer fibre membrane and 200 litres of pre-mixed dialysate. Indeed, in vitro and in vivo studies suggest the large interface of human blood against an engineered membrane incites an inflammatory reaction through immune cell stimulation, complement activation, and cytokine generation.^64–67 Similarly, peritoneal dialysis patients expose their abdominal cavity to 8–16 L of dialysate daily that contains plasticizers, which incite a direct inflammatory response.⁶⁸ Dialysate from poorly processed municipal water can contain bacterial products including endotoxins that can cross the dialysis membrane into the patient’s bloodstream to activate monocytes that induce inflammation.⁶⁸ In addition, 80% of patients in the USA initiate chronic dialysis using a catheter as their vascular access.⁶⁹ Multiple studies show a statistically significant association between elevated CRP levels and catheter use in dialysis patients.⁷⁰ Both anti-inflammatory and LDL-lowering effects of statins might decrease the risk of coronary events in CKD patients, but we lack evidence of their efficacy once patients start dialysis. In the Study of Heart and Renal Protection (SHARP), simvastatin with ezetimibe decreased MIs, strokes, and arterial revascularization procedures in CKD, but not in dialysis patients.⁷¹ Two separate dialysis trials also demonstrated the lack of improved CV and ACS outcomes with LDL lowering.^72–74 The failure of statins to improve outcomes in ESKD in SHARP and other trials may reflect the distinct pathophysiology of CV events and/or the advanced state of arterial disease in this population. While MIs account for many CKD deaths, statins do not directly prevent arrythmias, which cause much mortality in dialysis patients.⁷⁵

Anti-inflammatory therapies to prevent CV complications in the dialysis population remains in its infancy. Moderate CKD patients randomized to an antibody that neutralizes the pro-inflammatory cytokine interleukin-1β, canakinumab, significantly reduced their risk of major adverse CV events by 18% compared to placebo.⁷⁶ The beneficial effect was most pronounced [hazard ratio (HR) = 0.68] in those whose hsCRP fell to <2 mg/L on canakinumab. These data support further trials of anti-inflammatory treatments in patients starting dialysis.

Vascular calcification

Coronary and other arterial calcification may contribute to CV disease and are highly prevalent in patients starting dialysis. While only 10.1% of the Multi-Ethnic Study of Atherosclerosis population has CAC, more than 70% of patients starting dialysis will have CAC.^77–79 In dialysis patients, CAC is associated with increased mortality and diastolic dysfunction by echocardiography.^{80  ,  81} Calciphylaxis occurs almost exclusively in the dialysis population when microvascular calcification of the dermo-hypodermic arterioles leads to necrotic skin lesion.⁸² The 6-month survival of patients with calciphylaxis is ∼50%, mostly due to sepsis.⁸³

The genesis of CAC may differ in dialysis patients. While autopsy studies show similar coronary plaque burden in dialysis and non-CKD patients who die of cardiac death, atheromata of dialysis patients harbour markedly more calcium.⁸⁴ The extensive arterial calcifications in ESKD extend to intra-myocardial arterioles.⁸⁵ Aortic calcification causes arterial stiffness that increases pulse pressure, reduces DBP, and thus decreases coronary perfusion.⁸⁶ Aortic stiffness also raises LV afterload, which can increase LV mass, an independent predictor of death, in > 70% of those starting dialysis.⁸⁷ These abnormalities may predispose to heart failure with preserved systolic function in patients with CKD and on dialysis.

The complex pathogenesis of uraemic vascular calcification includes systemic inflammation and disordered mineral metabolism.^88–90 Arterial smooth muscle cells can acquire osteoblastic functions that promote arterial calcification in CKD due to hyperphosphataemia, oxidative stress, and activity of the osteoblastic transcription factor RUNX2.^{91  ,  92} Furthermore, ESKD patients have low concentrations of calcification inhibitors such as fetuin-A and matrix gla-protein, favouring calcification.^93–95 FGF23, which increases markedly in patients undergoing dialysis, can induce LVH in uraemic rats.^{96  ,  97} For these reasons, normalizing calcium, phosphorus, PTH, and FGF23 merit consideration as interventions to limit CV events in those starting dialysis.

Impaired autonomic modulation

Finally, several heart rate variability studies provide evidence of impaired cardiac autonomic modulation in dialysis patients.⁹⁸ This autonomic imbalance may predispose to lethal arrythmias.⁹⁹

Prevention: decreasing cardiovascular burden as dialysis starts

Reducing CV risk in patients starting dialysis merits a combined approach. Nephrologists optimize dialysate prescription, dialysis temperature, and ultrafiltration rate to remove excess fluid and normalize electrolytes while minimizing hypotension and coronary hypoperfusion during dialysis.^{100  ,  101} Peritoneal dialysis (PD) places fewer haemodynamic demands on the heart. According to observational studies, PD may mitigate the ‘cardiovascular spike’; however, it remains inconclusive if HD vs. PD improves long-term survival.^102–104 The only attempted randomized controlled trial of HD vs. PD halted because of inadequate enrolment, as most patients clearly prefer one modality.¹⁰⁵ Frequent, shorter dialysis (6 times per week) can control BP, phosphorus, and improve LV function over conventional protocols (3 times per week)^{106  ,  107}; however, these RCTs lack adequate power for mortality and did not enrol patients just starting dialysis.

CV disease management during dialysis overlaps with outpatient care, mandating an integrated, multidisciplinary approach. Dialysis initiation should signal the care team to shift management targets strategies to align with the altered mechanisms of disease. As CV clinical trials typically exclude those with advanced CKD and dialysis, we lack conclusive data to inform diabetes, BP, cardiomyopathy, and arrhythmia management of dialysis.¹⁰⁸ We must therefore base practice on the pathophysiology of disease with extrapolation from small validation trials in dialysis patients or on trials done in the non-dialysis population. The safety of such extrapolation remains unproved, and all members of the care team should acknowledge these limitations.

Blood pressure management

During the first 4 months of dialysis, fluctuating BP presents a management challenge.¹⁰⁹ Sodium restriction and dialysis fluid removal offer first-line therapies for hypertension at this juncture.^{110  ,  111} Most patients have excess volume at dialysis initiation; however, ultrafiltration in the first months of dialysis will reduce BP and permit reduction of hypertensive medications. Nephrologists often advise omission of BP medication before a session to achieve a pre-dialysis BP = 150–159/80–99 that reduces intra-dialytic hypotension and vascular access thrombosis⁵⁸ although pre- and post-dialysis BP correlate minimally with CV mortality.¹¹² In contrast, a 20/10 mmHg increase in inter-dialytic BP taken at home or in the doctor’s office is associated with a significant 50% increase in CV mortality.¹¹³ A meta-analysis of dialysis subgroups from RCTs reports that a 4–5/2–3 mmHg lowering of BP with anti-hypertensive medication in dialysis patients reduces both CV events and mortality by 29%.¹¹⁴ No dialysis RCTs have evaluated BP targets, but extrapolation from general population studies suggests that dialysis patients could target a home or outpatient-measured BP of 120–135/60–80 mmHg in between HD treatments.¹¹⁵

Anti-hypertenisve medications

Some, albeit limited, RCTs have evaluated classes of anti-hypertensive drugs in this context. Beta-blockers might seem a logical first choice, as arrhythmia may contribute to death in dialysis initiation. One trial randomized 200 hypertensive dialysis patients to atenolol vs. lisinopril and stopped early as the incidence of MI, stroke, chronic heart failure (CHF), and CV death increased two-fold in those allocated to lisinopril.¹¹² Even though a meta-analysis of five RCTs of angiotensin-converting enzyme inhibitor (ACEI)/angiotensin II receptor blocker (ARB) vs. placebo reported significant reductions in LV mass index, these findings have not consistently translated into improved outcomes.¹¹⁶ Four small trials randomized dialysis patients to an ACEI/ARB or placebo. Two of these trials showed CV benefit with ACEI/ARB while two other trials did not, such that the efficacy of renin–angiotensin–aldosterone system (RAAS) inhibition remains unclear in ESKD.¹¹⁰ Amlodipine (vs. placebo) in dialysis patients reduces mortality, non-fatal stroke, and coronary/peripheral revascularization by 47% when compared to placebo in one small trial (n = 251).¹¹⁷ Finally, 253 non-heart failure dialysis patients randomly received 25 mg spironolactone or placebo. Those treated with spironolactone had 58% less cardiocerebrovascular death, sudden death, and cardiac arrest, but no statistically significant difference in the incidence of hyperkalaemia.¹¹⁸ Another RCT found that spironolactone (25 mg/day) significantly decreased CV/cerebral hospitalization and death by 60% and all-cause mortality by 65% in HD when compared to placebo (n = 157), while serious hyperkalaemia occurred in only 1.9% of patients on the drug.¹¹⁹ Given the small size of these trials and their inclusion of subjects without a history of hyperkalaemia, the use of spironolactone in those starting dialysis requires caution pending the results from two ongoing larger trials examining the safety and efficacy of mineralocorticoid antagonists in dialysis.^{120  ,  121}

In addition to class of drug, whether dialysis removes the prescribed drug requires consideration. A propensity-matched observational study reported 40% higher mortality in dialysis patients prescribed dialysable beta-blockers (atenolol, metoprolol) when compared to non-dialysable beta-blockers (propranolol, carvedilol).¹²² ARBs are not cleared by the kidney or removed by dialysis and thus do not require dosing adjustments, which simplifies their prescribing in dialysis. The use of an ACEI can present more challenges, as each drug has a different renal and dialysis clearance properties (see Table 4). Dialysis does not remove amlodipine, diltiazem, or doxazocin. Dialysable medications should be avoided or taken after dialysis to avoid their removal from the bloodstream.

Table 4.

Cardiovascular medications dosing and clearance for patients starting dialysis^123–125

Medication	Drug clearance		Comments
	Kidneys	Dialysis
Anticoagulant and antiplatelet
Enoxaparin	Yes	No	20–30 mg SQ for DVT prophylaxis likely acceptable; avoid otherwise
Apixaban	Yes	Negligible	PK modelled dose of 2.5 mg b.i.d.; but no data to support safety and efficacy of drug; prescribe drug at own discretion
Dabigatran	Yes	Yes	Contraindicated; drug bio-accumulation can cause bleeding
Edoxaban	Yes	Negligible	Not recommended; drug bio-accumulation can cause bleeding
Rivaroxaban	Yes	No	PK models suggest 15 mg qd; but no data to support safety and efficacy of drug; prescribe drug at own discretion
Fondaparinux	Yes	Unknown	Not indicated
Eptifibatide	Yes	Yes	Contraindicated; drug bio-accumulation can cause bleeding
Prasugrel	Yes	No	No dosage adjustment
Ticagrelor	No	No	No dosage adjustment
Dalteparin	Yes	No	5000 units prophylaxis ×7 days likely safe; monitor anti-Xa levels for treatment level dosing
Aspirin	Yes	Yes	Administer after HD on dialysis days
Clopidogrel	Yes	No
Cangrelor	Yes	Unknown	No dose adjustment
Diabetic drugs
Insulin	Yes	No	Decrease initial dose by 50%
Glipizide	No	Unknown	No dose reduction needed in dialysis
Repaglinide	No	unknown	No dose reduction needed in dialysis
Metformin	Yes	Yes	Contraindicated; causes lactic acidosis
Thiazolidinediones (rosiglitazone, pioglitazone)	No	No	Not recommended; risk of oedema and CHF in anuric dialysis patients
Diabetic drugs: GLP-1 inhibitors
Exenatide	Yes	Unknown	Not recommended: severe GI side effects, many reports of AKI126
Liraglutide	No	No127	Not recommended; no safety and efficacy data
Dulaglitide	No	Unknown	Not recommended; no safety and efficacy data, increased risk AKI128
All SGLT2 inhibitors	Yes	Unknown	Contraindicated
Diabetic drugs: DPP-4 inhibitors
Sitaglipin	Yes	Minimal	25 mg qd
Saxaglipin	Yes	Yes	2.5 mg qd, post-dialysis
Linagliptin	No	No	No dose adjustment
Aglopliptin	Yes	No	6.25 mg qd
B-blockers
Atenolol	Yes	Yes	Administer after HD, reduce starting dose ∼50%
Metoprolol	No	Yes	Administer after HD
Carvdedilol,	No	No
Labetalol	No	No
Propranolol	No	No	No dose adjustment, but use with caution as weak active metabolite cleared by kidney
Bisoprolol	yes	no	Initial: 2.5 mg daily; increase cautiously
Other anti-hypertensive drugs
ACEI	Yes/No	Yes/No	Renal and dialysis clearance vary by ACEI type; Use ARB instead as all ARB require no dose adjustment thus simpler to dose
ARB	No	No
Amlodipine	No	No
Clonidine	Yes	No	Decrease starting dose but no mcg recommendations provided
Doxazocin	No	No
Isosorbide mononitrate	Yes	Yes	Dose after dialysis
Spironolactone	Yes	Unknown	25 mg qd
Anti-arrhythmic
Procainamide (Ia)	Yes	Yes	Monitor procainamide/N-acetylprocainamide (NAPA) concentrations; supplementation may be necessary
Quinidine (Ia)	Yes	Yes	No dosage adjustment provided in manufacturer’s labelling. Use with caution. Some clinicians advocate 75% normal dose after HD
Disopyramide (Ia)	Yes	No	Reduce dose; monitor drug levels and ECG; increased risk of torsades do pointes
Lidocaine (Ib)	No	No
Mexiletine (Ib)	No	Yes
Propafenone (Ic)	No	No	Accumulation of active metabolites in ESKD; no dose recommendation on label but some recommend 50% dose reduction
Flecainide (Ic)	Yes	No	Reduce dose 50%; plasma levels to guide dosing
Sotalol (III)	Yes	Yes	Contraindicated
Dofetilide (III)	Yes	No	Contraindicated
Amiodarone (III)	No	No	Safest anti-arrhythmic to use in ESKD
Digoxin	Yes	No	Keep serum level ≤1 if not avoid; rapid shifts in potassium during haemodialysis may precipitate arrhythmias through hypokalaemia2

For dialysis patients with reduced ejection fraction (EF) heart failure, small ESKD trials validate RCT findings in the general population. Carvedilol (up to 25 mg BID) significantly improved LV EF by 27%, CV mortality by 78%, and survival by 49% when compared to placebo (n = 114) in dialysis patients with an EF < 35%.^{129  ,  130} Similarly, two small placebo-controlled spironolactone (25 mg/day) trials discussed above indicate promise for mineralocorticoid antagonism; however, we await confirmatory results from an ongoing larger trial¹²⁰ conducted in dialysis patients with an EF < 40%. While RAAS dialysis trials in the non-heart failure cohorts described above were inconclusive, ACEI/ARB therapy may have efficacy when dialysis patients have coexisting heart failure. In a double-blind trial of 332 HD patients with an EF < 40% on ACEI randomly receiving add-on telmisartan (40–80 mg/day) vs. placebo, the ARB significantly reduced CV mortality (HR = 0.42) and CHF hospitalization (HR = 0.38) without causing excess hyperkalaemia.¹³¹

Medication reconciliation

The start of dialysis should prompt a careful review of a patient’s medications. Some CV medications depend on renal clearance, necessitating dose reduction or discontinuation, as renal function is minimal at dialysis initiation. Additionally, dialysis clears some medications, which may require a supplementary dose given after dialysis to maintain therapeutic drug levels (see Table 4). The use of antiarrhythmic drugs proves challenging in this population, as these drugs are often dialysable and cleared by the kidney, which can lead to fluctuations in the drug levels. Additionally, many anti-arrhythmics have a narrow therapeutic window, and toxicity can precipitate lethal arrhythmias in a population that is already prone to SCD.¹³² Dose reduction, plasma level testing and ECG monitoring may enhance the use of these drugs in dialysis patients.¹³² Amiodarone merits consideration in these patients due to its predictable elimination/clearance unaffected by dialysis.¹³² Forty percent of dialysis patients have a prolonged QTc interval, and QTc intervals increase after HD, likely from the rapid removal of potassium, magnesium, and calcium.¹³³ Clinicians should consider omitting QT-prolonging medications before patients initiate dialysis, especially if the baseline ECG shows a long QT interval.

Hyperkalaemia

Hyperkalaemia is prevalent in patients starting dialysis and is associated with increased risk of SCD.^{134  ,  135} Hyperkalaemia refractory to dietary restriction and dialysis removal can be treated with sodium polystyrene or newer agents that remove potassium by exchanging cations for potassium in the distal colon.¹³⁶

Anticoagulation

The practice of anticoagulation in dialysis persists in a state of clinical equipoise. About 30% of HD patient have atrial fibrillation at the start of dialysis; yet, it remains unclear if the benefits of warfarin outweigh the risks of accelerating bleeding, CAC, and calciphylaxis in dialysis.¹³⁷ Direct oral anticoagulants do not aggravate arterial calcification, but these drugs depend on renal clearance to variable extents, such that drug accumulation can enhance bleeding risk.¹³⁸ The US FDA label does make pharmacokinetically-based dose recommendation for the use of rivaroxaban and apixaban in dialysis even though ESKD patients were excluded from all phase 3 clinical trials for atrial fibrillation and venous thromboembolism prophylaxis.^{139  ,  140} The risk of bleeding was 20% higher in apixaban vs. warfarin (statistically non-significant) in an RCT of dialysis patients with atrial fibrillation; however, the trial was underpowered and terminated early because of slow enrolment.¹⁴¹

Implantable cardioverter-defibrillator

Given the high risk of SCD and frequent cardiomyopathy, CV implantable cardioverter-defibrillator (ICD) use is prevalent (10%) in the dialysis population.¹⁴² ICDs may have reduced efficacy in dialysis patients compared to non-dialysis patients.^{143  ,  144} More study is needed on the risk–benefit of ICD therapy once patients start dialysis. Concurrent comorbid conditions, the dialysis access, and repeated cannulation raise the risk of device infection.¹⁴⁵ Additionally, ICD infection-related mortality in dialysis remains significantly higher than in the general population even after prompt device extraction.^146–149

Anaemia

Erythropoietin and IV iron can treat the anaemia of CKD to increase haemoglobin to 10-12 g/dL to attempt to limit cardiac complications and regress LV mass,¹⁵⁰ yet raising haemoglobin levels beyond 12 g/dL is associated with CV harm.¹⁵¹ Ongoing trials for Hypoxia Inducible Factors (HIF) inhibitors may demonstrate that these drugs can raise haemoglobin without the CV risks seen with EPO.¹⁵²

Diabetes management

Unfortunately, we lack trial evidence to inform diabetes management in patients starting dialysis, and the efficacy and safety of different anti-diabetic agents. Because of this information gap, HD patients are typically switched to insulin, although some physicians will elect to use an oral agent. Among the sulphonylurea/meglitinide drugs, glipizide, and repaglinide depend least on renal clearance.^{153  ,  154} Metformin (lactic acidosis), thiazolidinediones (oedema and CHF), GLP-1 agonists, and SGLT2 inhibitors are all not recommended in dialysis. FDA labels do support DDP-4 inhibitor prescription in dialysis. A small dialysis trial (n = 129) showed that sitagliptin is equivalent to glipizide at lowering Hba1c with less severe hypoglycaemia, but larger studies are needed.¹²³ Given the reduction in CV events recently demonstrated with SGLT2 inhibitors and GLP-1 agonists, we urgently need more information about the utility and safety of these classes of agents in the dialysis setting.

Finally, providers should ensure that CKD patients have nephrology care by the time their eGFR falls below 30 mL/min for vascular access and pre-emptive transplantation planning.¹²⁴ The best outcome and quality of life for patients with progressive CKD is pre-emptive kidney transplantation, when clinical circumstances permit.¹²⁵ Fistula or graft vascular access should be placed for HD because it associated with better cardiac outcomes than those using a catheter.⁶ There is evidence that brachial plexus block is superior to general and local anaesthesia in fistula placement because it improves haemodynamics and fistula maturation.¹²⁶ If urgent, new-generation multilaminated grafts can be placed that can be used 24–72 h after implantation.¹²⁷ Altered dialysis initiation practices to preserve residual renal function, such as peritoneal dialysis and starting with twice weekly dialysis, are under study.¹²⁸

Conclusion

Dialysis initiation encompasses the period of highest CV risk in the spectrum of CKD, entailing a 20-fold mortality excess over age-matched controls. This ‘cardiovascular spike’ coincides with the start of dialysis, which shifts the mechanism of disease to alter the manifestations, response, and outcomes of CV disease in this vulnerable population. This critical period affords an opportunity for clinicians to heighten their vigilance for CV complications while ongoing research attempts to address the many gaps in our knowledge regarding optimum management of this perilous point in the patient’s journey with CKD. A cooperative approach between the nephrology and CV care teams assumes even greater importance during this critical passage.

Search strategy

Search strategy and selection criteria: Data for this seminar were identified from PubMed, MEDLINE, and references from relevant articles using the search terms ‘incident dialysis’, ‘cardiovascular’, ‘dialysis initiation’, and ‘dialysis early mortality’. Only articles published in English after 1980 were included.

Supplementary material

Supplementary material is available at European Heart Journal online.

Data availability

Standard analytic file data provided by USRDS Data Coordinating Center.

Funding

P.L. receives funding support from the National Heart, Lung, and Blood Institute (1R01HL134892), the American Heart Association (18CSA34080399), the RRM Charitable Fund, and the Simard Fund.

Conflict of interest: P.L. is an unpaid consultant to, or involved in, clinical trials for Amgen, AstraZeneca, Baim Institute, Beren Therapeutics, Esperion, Therapeutics, Genentech, Kancera, Kowa Pharmaceuticals, Medimmune, Merck, Norvo Nordisk, Merck, Novartis, Pfizer, Sanofi-Regeneron. P.L. is a member of scientific advisory board for Amgen, Corvidia Therapeutics, DalCor Pharmaceuticals, Kowa Pharmaceuticals, Olatec Therapeutics, Medimmune, Novartis, and XBiotech, Inc. P.L. laboratory has received research funding in the last 2 years from Novartis. P.L. is on the Board of Directors of XBiotech, Inc. P.L. has a financial interest in Xbiotech, a company developing therapeutic human antibodies. P.L. interests were reviewed and are managed by Brigham and Women’s Hospital and Partners HealthCare in accordance with their conflict-of-interest policies.