Definition of chronic kidney disease

Chronic kidney disease (CKD) is defined by the Kidney Disease Improving Global Outcomes (KDIGO) guidelines as an abnormality in kidney function or structure that is present for greater than 3 months. It is classically defined by an estimated glomerular filtration rate (eGFR) less than 60 mL/min/1.73 m² and/or by the presence of persistent kidney damage (albumin to creatinine ratio (ACR) of >30 mg/g, urine sediment abnormalities, tubular dysfunction, history of kidney transplant).1 According to the United States Renal Data System (USRDS), CKD prevalence in the USA was approximately 15% from 2011 to 2014, which translates to approximately 36 million American adults.2

CKD is divided into five stages based on severity of kidney disease (table 1). More advanced stages of CKD (defined by lower eGFR or higher urine ACR) are associated with worse prognosis, including progression to end-stage renal disease (ESRD), cardiovascular disease (CVD) and death.3

Epidemiology of heart failure and other cardiovascular disease in patients with kidney disease

Individuals with CKD have mortality rates that are more than double the rate in the general population. Greater than 50% of deaths in patients with CKD are from CVD. In fact, patients with CKD are more likely to die from CVD than to progress to end-stage-renal-disease (ESRD). CKD is now considered a coronary artery disease risk equivalent akin to diabetes.4 All forms of CVD including acute myocardial infarction (MI), heart failure (HF), valvular disease, cerebrovascular accidents (CVA), peripheral artery disease (PAD), thromboembolic disease and sudden cardiac death are more common in those with CKD.5

HF is a leading cause of CVD among patients with CKD and ESRD.5 6 Almost 30% of CKD Medicare patients have HF, compared with just 6% of Medicare patients without CKD.2 The large, biracial Atherosclerosis Risk in Communities (ARIC) study of almost 15 000 middle-aged participants found that the risk of incident HF is threefold greater in individuals with eGFR <60 mL/min/1.73 m² compared with those with eGFR >90 mL/min/1.73 m².6 HF prevalence and incidence also increases with the severity of CKD. Studies estimate that HF is 12–36 times more common in dialysis patients compared with the general population, and among dialysis patients, the incidence of HF is 7% per year.7 Patients with CKD can have either HF with reduced ejection fraction (HFrEF) or HF with preserved ejection fraction (HFpEF), with HFpEF being most common.

There are important racial and ethnic differences in HF incidence among patients with CKD as well. A recent publication of three diverse community-based cohorts reported that the risk of HF in patients with CKD was particularly high among African–Americans and Hispanics, where the adjusted risk of developing HF with CKD was 7.8 per 1000 person-years compared with 3.5 per 1000 person-years without CKD.5 Similarly, in the diverse Coronary Artery Risk Development in Young Adults (CARDIA) study, black patients developed HF before age 50 at a rate that was 20 times the incidence in white patients, with kidney disease being one of the strongest risk factors (HR of 19.8).8

HF contributes greatly to morbidity and mortality in patients with CKD and ESRD. It remains one of the leading causes of emergency department visits and hospitalisations in this population,9 which translates to poor health-related quality of life.10 Currently, 2-year survival for Medicare patients with CKD and HF is only 76% compared with 93% for those without either condition.2

Diagnosis of heart failure in patients with kidney disease

Despite its prevalence, there remain significant challenges to HF diagnosis in the CKD population, leading to underdiagnosis (and thus, undertreatment). HF is a clinical diagnosis based on a characteristic history and physical exam indicative of inadequate cardiac output or cardiac filling to meet systemic demands. Accurate diagnosis relies on signs or symptoms of volume overload (shortness of breath, orthopnea, paroxysmal nocturnal dyspnoea, oedema) and signs of elevated cardiac filling pressures (jugular venous distention, displaced apical impulse). These symptoms/signs can also be seen in patients with CKD, particularly as they near ESRD when it can be difficult to distinguish HF from systemic volume overload.

Early HF symptoms in particular are often missed or underappreciated in patients with CKD. In a well-defined CKD population of nearly 3000 patients without a clinical diagnosis of HF, 25% reported significant HF symptoms as determined by the validated 23-item Kansas City Cardiomyopathy Questionnaire (KCCQ).11 Furthermore, this study showed an inverse association between eGFR level and the burden of HF symptoms. In a follow-up study, greater KCCQ symptom burden was associated with a higher short-term risk of incident HF (OR of 3.30 for those with the highest symptom burden compared with those with the lowest symptom burden).11 These data suggest that the KCCQ may detect subclinical HF in symptoms and shows promise for future HF screening in CKD.

Ancillary imaging and lab tests such as ECG, echocardiograms, chest radiographs, N-terminal pro-BNP (NT pro-BNP) and high-sensitivity troponin T (hs-TnT) are often difficult to interpret in CKD and ESRD. Park et al recently reviewed echocardiograms from patients with CKD in the multicenter Chronic Renal Insufficiency Cohort (CRIC) study and found that there was a 50% prevalence of left ventricular hypertrophy (LVH) among CKD individuals without clinical signs of HF. Furthermore, among patients with ESRD, nearly 75% have incidental LVH. Thus, structural abnormalities in left ventricular mass and geometry may not necessarily help discriminate HF.12

Similarly, traditional biomarkers like NT pro-BNP and hs-TnT may also be difficult to interpret in patients with CKD and ESRD as circulating concentrations of these biomarkers are affected by lower kidney excretion. Despite this concern, a number of studies have shown that elevations in NT pro-BNP and hs-TnT are in fact strongly associated with incident HF in patients with CKD, even after accounting for eGFR level and urine ACR. The 2008 African American Study of Kidney Disease and Hypertension (AASK) evaluated over 900 hypertensive adults with CKD and showed that there was a 70% increase in HF risk per doubling of NT pro-BNP.13 In the European Prevention of Renal and Vascular Endstage Disease (PREVEND) study of over 1000 patients with CKD, elevations in NT pro-BNP and hs-TnT were significantly associated with cardiovascular events (defined as MI, ischaemic heart disease and coronary artery bypass grafting or percutaneous coronary angioplasty).14 The 2015 CRIC study specifically evaluated for incident HF and found that patients with CKD in the highest quartiles of NT pro-BNP and hs-TnT had a nearly 10-fold and fivefold higher risk of incident HF, respectively.15 However, the use of these biomarkers in the clinical setting still remains controversial since there are no accepted cut-offsin the CKD population.

There are also current investigations into novel cardiac biomarkers that are known to be widely expressed in tissue inflammation, fibrosis, hypoxia and oxidative stress as predictors of clinical and subclinical HF in kidney disease that may help with diagnosis of HF in the future.16–18 Until then, HF in CKD remains a diagnosis based on characteristic symptoms and correlated structural/functional abnormalities on cardiac imaging.

Risk factors for heart failure in patients with kidney disease

There are a myriad of different mechanisms that contribute to the development of HF in patients with kidney disease. The multifactorial inflammatory, neurohormonal, metabolic, nutritional and haemodynamic changes that occur in kidney disease also affect the cardiovascular system. The heart and kidneys interact in a complex, bidirectional manner to influence each other such that a declining eGFR accelerates HF progression just as HF contributes to declining eGFR (figure 1).19 It is unclear whether one organ is the primary driver of dysfunction in the other organ, or if both the heart and kidneys are affected by similar parallel biological insults.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Heart failure and chronic kidney disease. Proposed paradigm for the bidirectional interactions between the heart and the kidneys in patients with heart failure and chronic kidney disease, with emphasis on both the shared and novel kidney disease risk factors that contribute to the pathophysiology of these conditions. RAAS, renin–angiotensin aldosterone system.

As outlined in table 2, traditional risk factors for the development of HF have been confirmed in patients with CKD and ESRD. Interestingly, these HF risk factors are also risk factors for the development of CKD itself. Older age is one of the strongest risk factors for both kidney disease and HF. The pathological changes in the heart and kidney that accompany ageing are complex, multifactorial and not completely understood, but likely involve accelerated endothelial damage leading to macrovascular and microvascular disease, cell senescence and fibrosis. It is plausible that similar pathways lead to age-related changes in both organs. Men are also more likely to develop CKD and HF; however, previous studies from our group and others have not found that sex significantly modifies the association between kidney disease and HF5 20; the association is strong in both male and females.

Modifiable risk factors include hypertension, smoking, obesity and glycaemic control in diabetes mellitus.21–23 We recently analysed blood pressure control among CRIC study participants and found that wider pulse pressure (>68 mm Hg vs <51 mm Hg) was associated with greater risk of HF in patients with stage 4 and 5 CKD.22 Foley et al investigated smoking status among haemodialysis patients in the USRDS Wave 2 study and found that smoking was associated with a 60% increase in new-onset HF compared with non-smokers.21 A recent German study evaluated glycemic control in 1255 haemodialysis patients and found that a glycated haemoglobin (A1c) level of >8% had a twofold higher risk of sudden death compared with those with an A1c of <6%. Furthermore, each 1% increase in A1c increased cardiovascular events and mortality by 8% with a specific trend towards higher risk of CVA and increased death from HF.24 Focus on reversing these modifiable risk factors may help to decrease mortality in patients with kidney disease.

The unique biological changes that occur with the progression of kidney disease also promote HF (Table 2). In patients with stages 4 and 5 CKD, metabolic abnormalities, anaemia, uraemic toxins, sympathetic overactivity and volume dysregulation all likely play a role in cardiac disease. Several studies evaluating anaemia in patients with HF have shown that anaemia is both common and independently associated with increased mortality.25 However, there are no consensus guidelines on transfusion thresholds for patients with HF. In patients with CKD, there is an expected gradual decline in haemoglobin with worsening renal function. KDIGO currently recommends consideration of erythropoietin stimulating agent (ESA) therapy in patients with CKD with haemoglobin <10 g/dL after appropriate treatment for modifiable causes of anaemia.1 This threshold is based off the landmark Normal Hematocrit Trial where ESA supplementation to maintain a haematocrit (HCT) of 42 versus 30 in dialysis patients resulted in a 7% higher mortality rate in the HCT 42 group.26 Subsequent trials have expanded these findings to patients with CKD stage 3–4 where normalisation of haemoglobin did not change CVD risk. In fact, use of ESA agents may lead to increased hospitalisations and thrombosis; thus, the utility of anaemia correction remains unclear.27

Additionally, the metabolic derangements seen in CKD can increase CV risk. CRIC investigators found that a bicarbonate >26 mmol/L was independently associated with a 66% increase in relative risk of HF hospitalisations compared with those participants with bicarbonate 22–26 mmol/L. In the same cohort, bone mineral metabolism was evaluated with higher serum levels of phosphate and of fibroblast growth factor-23 (FGF-23), a key regulator of phosphate metabolism, being associated with higher rates of HF hospitalisations and LVH. CRIC investigators found a 45% increased risk of HF hospitalisation per doubling of FGF-23 level after adjustments for confounders.28 Likewise, in the Cardiovascular Health Study, the subset of patients with CKD with the highest quartile of FGF-23 had a 94% increased risk of incident HF compared with patients with CKD with the lowest quartile of FGF-23.29 Further investigations into the role of malnutrition and sympathetic activity in promoting cardiac disease are also needed and may identify novel therapeutic targets for HF in patients with CKD.

Once patients reach ESRD, changes in volume status and electrolyte shifts during dialysis may also contribute to cardiac remodelling and the development of HF. A 2008 study showed that there is a significant reduction of myocardial blood flow during haemodialysis treatments (as assessed by cardiac positron emission tomography scans).30 These ‘myocardial stunning’ events were associated with a longitudinal decline in left ventricular ejection fraction (LVEF) of almost 10% per year, higher risk of subsequent CVD and death.31

Heart failure as a risk factor for CKD progression

HF itself impacts the development and progression of CKD confirming the bidirectional interaction between the heart and the kidneys. The prevalence of CKD among those with chronic HF is estimated to be 40%.32 Studies have shown that worsening HF is associated with progressive loss of kidney function. In the CRIC study, a history of HF in patients with CKD was independently associated with a 29% higher risk of progression to ESRD or a 50% decline in eGFR compared with those without HF.33 There are several postulated mechanisms to explain this bidirectional association between kidney disease and HF. For example, haemodynamic changes are common in patients with HF and can lead to increased renal venous congestion and poor forward flow, which may compromise kidney function. Additionally, adverse effects from HF pharmacotherapies, like aggressive diuresis, may cause acute renal injury and subsequent progression of CKD. Similarly, kidney disease leads to increased sodium retention and increased blood pressure and afterload, which can contribute to cardiac damage. Several other hormonal and inflammatory pathways may be activated, leading to both kidney and cardiac stress and damage. Further study of other mechanisms involved in this complex bidirectional relationship is warranted.

In addition to clinical HF, biomarkers and imaging suggestive of subclinical HF are also associated with a progressive decline in kidney function. In the Cardiovascular Health Study of older participants, those in the highest quartile of NT pro-BNP had a 38% higher risk of incident CKD and a 67% higher risk of rapid eGFR decline compared with those in the lowest quartile.34 Another study of diverse participants with no CVD reported that those in the highest quartile of cardiac concentricity as assessed by cardiac MRI had an additional 21% decline in kidney function compared with those in the lowest quartile.35 Similarly, LVH measured on echocardiograms among African–Americans in the Jackson Heart Study was associated with kidney function decline.36 Thus, developing more effective HF therapies may also help preserve kidney function, which in turn, would lead to better prognosis.

Progression of subclinical HF in kidney disease

Several recent investigators have explored the natural history of subclinical HF as CKD progresses with emphasis on the physiologic transition from CKD to the initiation of dialysis. A substudy of the New Zealand IDEAL trial obtained echocardiograms of patients with late stage CKD at baseline and 12 months after randomisation to either early (eGFR 10–14 mL/min/1.73 m²) or late (eGFR 5–7 mL/min/1.73 m²) initiation of haemodialysis. They found no significant change in echocardiogram metrics after the dialysis transition, with similar increased diastolic dysfunction (DD) and increased left ventricular mass index (LVMI) present in both groups at baseline and follow-up.37 In contrast, other studies have shown structural changes to the heart after the dialysis initiation. In the USA, the multicentre CRIC study collected predialysis and postdialysis initiation echocardiographic data for a CKD cohort with a mean time between studies of 2.9 years. They found that mean LVEF worsened (52.5%–48.6%%), while mean LVMI (60.4 g/m^2.7–58.4 g/m^2.7) and diastolic relaxation (11.11%–4.94%% had moderately/severely abnormal relaxation) improved. As demonstrated in the CRIC study and others, echocardiographic changes can be seen in the absence of clinical HF in patients with CKD. These subclinical echocardiographic changes are likely predictive of future HF development, but to date it is still unknown whether screening for subclinical HF improves outcomes in patients with CKD.

Treatment of HF in kidney disease

Once diagnosed, appropriate HF treatment in CKD patients can be challenging. As most CKD-HF patients have HFpEF, there are no current evidence-based recommendations for therapies that improve outcomes.38 Evidence-based HFrEF therapies, like β-blockers and renin–angiotensin aldosterone system (RAAS) inhibitors, could be beneficial in this group, and more trials need to be conducted in this patient population. In one observational study of dialysis patients, β-blocker use was significantly associated with a 31% reduction in risk of HF.39 There are also ongoing trials regarding spironolactone efficacy and safety in dialysis patients (NCT02285920).

Novel HF regimens may also play a key role in improving outcomes in CKD patients with HF. Emerging antihypertensives like neprilysin inhibitors show promise in augmenting RAAS inhibition and the endogenous natriuretic peptide system.40 Thus far, there have been no studies of neprilysin inhibitors specifically in patients with CKD or dialysis patients. Similarly, the new sodium-glucose transporter-2 (SGLT-2) inhibitor empagliflozin also shows promise. A trial recently demonstrated that use of empagliflozin led to a 38% reduction in the risk of cardiovascular mortality in patients with diabetes and, despite modest effects on long-term glycemic control, highly significant reductions in HF admissions and ESRD.41 The SGLT-2 inhibitors block sodium/glucose uptake in the proximal tubule, inducing plasma volume contraction and decreasing glomerular hyperfiltration which leads to better long-term kidney preservation and improves diuretic and natriuretic responses to other diuretic agents. Moreover, SGLT-2 inhibitors might improve the efficiency of myocardial energetics. Thus, this new class of drugs has promise to improve HF and kidney disease in diabetics (and potentially non-diabetics).

Once patients reach ESRD, the mode and frequency of dialysis may also be important to mitigate HF risk. Studies have shown that peritoneal dialysis and more frequent dialysis are likely beneficial. A 2003 analysis of data from the USRDS Dialysis Morbidity and Mortality Wave 2 database found that those on haemodialysis were 56% more likely to experience de novo HF at 3-year follow-up compared with those on peritoneal dialysis.42 Several other studies have shown that peritoneal dialysis in HF significantly improves New York Heart Association Class and reduces the number of days hospitalised for HF.43 44 Hypothesised reasons for these improvements include better volume control with continuous gentle ultrafiltration that minimises haemodynamic changes, improves serum lipid profile, improves control of potassium and sodium haemostasis and removes proinflammatory mediators.

The impact of haemodialysis frequency on HF risk has also been studied. A 2010 randomised clinical trial compared haemodialysis six times per week versus three times per week and found that more frequent dialysis showed less increase in LVMI and less death (HR 0.61).45 The subsequent 2013 Frequent Hemodialysis Network trial also enrolled participants in six times per week haemodialysis versus three times per week and found significant decreases in left and right ventricular end diastolic volumes with more frequent dialysis, though no changes in left ventricular remodelling at 12 months.46

A limited number of studies have explored the impact of correcting novel HF risk factors in dialysis like anaemia, acid-base disturbances and bone metabolism abnormalities. One randomised controlled trial evaluating cinacalcet in reducing parathyroid hormone abnormalities in ESRD failed to reduce the risk of HF.47

While novel therapies are explored further, we can do better to ensure that all patients with CKD with HFrEF are on the few treatments that are evidence based. β-blockers and RAAS inhibitors that are well validated for HFrEF are underprescribed in patients with CKD/ESRD. A 2006 study found that almost 75% of dialysis patients with known HFrEF were not on the recommended β-blocker and ACE inhibitor (ACEI) combination. According to the study, the most common reason cited by nephrologists for this discrepancy is ‘concern about adverse reactions’, such as intradialytic hypotension and hyperkalemia.48 The actual frequency of these events attributed to the antihypertensives is unknown. Newer potassium-lowering agents, such as patiromer,49 may potentially expand the use and safety of RAAS inhibitors in patients with CKD. Thus, further research is needed to augment implementation of known HF therapies as well as to develop novel kidney-specific therapies for HF in CKD.