Is early chronic kidney disease an important risk factor for cardiovascular disease?: A Background Paper prepared for the UK Consensus Conference on Early Chronic Kidney Disease

Patients with severe chronic kidney disease (CKD) who require renal replacement therapy (RRT), either dialysis or renal transplantation, have a markedly increased risk of cardiovascular disease [ 1–4 ]. This increased incidence and prevalence of vascular disease accounts for much of the reduced life expectancy of these patients. The average life expectancy of a 40- to 44-year-old white male in the general population in the US is more than 35 years. However, if he is on long-term dialysis he can expect, on average, to live for only eight more years [ 4 ]. In Scotland, a patient aged between 20 and 44 on starting renal replacement can expect to survive, on average, a further 16.9 years [ 5 ]. Once patients have reached RRT, modification of traditional risk factors such as smoking, obesity, hyperlipidaemia and hypertension may have a disappointing impact upon overall mortality. One of the reasons adduced for this is that by the time RRT is reached it is too late. The burden of years of cardiovascular risk factors appears to overwhelm any intervention. In 1995, in an attempt to improve the quality of life of dialysis patients, the Dialysis Outcomes Quality Initiative (DOQI) was launched by the National Kidney Foundation in the US. In 1997, clinical practice guidelines (DOQI guidelines) were published [ 6–9 ]. However, during the development of these guidelines, it became clear that to further improve the quality of life outcomes of patients on dialysis and perhaps impact on the early, persistent, marked and seemingly resistant separation of the survival curves of these patients compared with the general population [ 10–12 ] intervention at an earlier stage of progressive CKD was likely to be essential. Therefore, in 1999 the Dialysis Outcomes Quality Initiative became the Kidney Disease Outcomes Quality Initiative (K/DOQI). This resulted in the publication in 2002 of clinical practice guidelines for CKD [ 3 ]. During the course of this initiative two factors further shifted the emphasis towards earlier CKD. Firstly, patients with early CKD appeared to have a significantly higher overall mortality and cardiovascular disease burden compared with the general population. Secondly, the majority of patients with CKD died before progressing to renal replacement [ 13 ]. A recent large observational study of patients with early CKD followed for 5.5 years found that though only 3.1% progressed to requiring RRT, 24.9% died before reaching dialysis. Cardiovascular Disease (CVD) may have accounted for many of these deaths [ 14 ]. This early CKD population deserved attention to their cardiovascular risks in their own right not just to improve the dialysis survival curves of the relative few who progressed to RRT. This excess cardiovascular risk was estimated by the K/DOQI guidelines to be high enough to suggest that ‘All patients with chronic kidney disease should be considered in the “highest risk” group for cardiovascular disease, irrespective of levels of traditional CVD risk factors’ [ 3 ]. This, together with the possibility that the optimal management of cardiovascular risk factors in early CKD may in addition delay or reduce progression of CKD, led to increasing interest in identifying CKD earlier and considering modifying these risk factors earlier [ 3 , 15 ]. This aritcle will review the evidence as to whether early CKD is an important risk factor for the presence, severity and progression of CVD. An internationally accepted definition and classification of severity of CKD had been surprisingly absent until the K/DOQI proposal in 2002 (Figure 1) [ 3 ]. This classification has subsequently been largely accepted and adopted or adapted by many national renal organizations including the Renal Association in the UK [ 15 ], the Canadian Society of Nephrology and the Australian and New Zealand Society of Nephrology [ 16 ]. The glomerular filtration rate (GFR) in millilitres per minute is the primary measure of kidney function. Mathematical equations (prediction equations) to estimate GFR (eGFR) have been devised. These equations are populated by relatively easily obtainable data such as age, gender, ethnicity, serum creatinine, weight and serum albumin [ 17 , 18 ]. The K/DOQI guidelines elected to use the most recently developed prediction equation [ 18 ] for estimating GFR in their definition and classification of CKD. This had been derived as part of the Modification of Diet in Renal Disease study [ 19–21 ]. Even a modified simplified version of this equation including four variables (gender, ethnicity, serum creatinine and age) has been shown to more accurately predict actual GFR in adults compared with other commonly used methods [ 22 ]. However, it is not without criticism. At higher levels of kidney function it is increasingly less precise and at lower levels of kidney function it may overestimate renal function because of its dependence on serum creatinine which may fail to rise relative to a fall in actual GFR because of a reduction in creatinine generation and increased renal tubular creatinine secretion. In addition, it was both generated and validated in a population known to have CKD [ 23 , 24 ]. Patients with ESRD have a markedly increased burden of cardiovascular disease including excess peripheral vascular disease, coronary artery disease, cerebrovascular disease and left ventricular dysfunction [ 3 , 4 ]. Cardiovascular deaths have been reported as the commonest cause of death in the United States Renal Data System reports from 1994 to 2005 with cardiovascular death rates 15–20 times that of the general population [ 1 , 4 , 10 , 25 ]. The excess burden of CVD is already evident at the time of initiation of dialysis [ 4 , 10 , 11 , 26 ]. It is biologically plausible that patients with earlier CKD may also have an excess burden of CVD. In addition to traditional risk factors ( Table 2 ) some of which they have suffered more frequently, for longer and with greater severity than the general population they may also have non-traditional risk factors ( Table 2 ) [ 27 ] which may make them more likely to have CVD, for that disease to progress differently and perhaps be less responsive to traditional risk factor modification. Biological plausibility, however, is not enough to assume an association let alone causation. The need to consider non-traditional risk factors has been driven to an extent by the failure to find an adequate explanation for the cardiovascular burden from traditional factors. Systolic hypertension and high serum cholesterol appear to lose their association with excess mortality at least in patients who have progressed to dialysis. This is probably explained by falling blood pressure as left ventricular function fails and falling cholesterol as nutrition becomes inadequate. However, even in patients with earlier CKD, traditional cardiovascular risk factors do not adequately explain the excess cardiovascular morbidity and mortality [ 27–29 ]. Stages of chronic kidney disease [ 3 ] Chronic kidney disease is defined as either kidney damage or GFR <60 ml/min/1.73 m 2 for >3 months. Kidney damage is defined as pathological abnormalities or markers of damage, including abnormalities in blood or urine tests or imaging studies. Stages of chronic kidney disease [ 3 ] Chronic kidney disease is defined as either kidney damage or GFR <60 ml/min/1.73 m 2 for >3 months. Kidney damage is defined as pathological abnormalities or markers of damage, including abnormalities in blood or urine tests or imaging studies. Cardiovascular disease risk factors [ 27 ] Cardiovascular disease risk factors [ 27 ] The K/DOQI guidelines 2002 reviewed the evidence for an association between CKD and cardiovascular disease [ 3 ]. Its work was based on data from the National Kidney Foundation Task Force on cardiovascular disease in chronic renal disease report [ 10 ]. Eight studies were identified which considered reduced GFR as a predictor of cardiovascular disease [ 30–37 ] ( Table 3 ). Some reported a number of different cardiovascular outcomes. All except Culleton et al. [ 37 ] reported a statistically significant association between reduced GFR and CVD. In addition, 11 studies were identified which considered reduced GFR as a predictor of either total mortality or cardiovascular death [ 30 , 31 , 33 , 38–45 ]. (some studies reported both) ( Table 4 ). Again only one failed to report a statistically significant association. These studies, however, were of variable quality, not all adjusted their findings to take into account covariates, and the measure of GFR (or surrogate markers for GFR) varied. Decreased GFR as a predictor of cardiovascular disease [ 3 ] Se creat, Se creatinine units of measurement as per study reports. RR, relative risk; HR, hazard ratio; CI, confidence interval. *Adjusted comparison if reported. Decreased GFR as a predictor of cardiovascular disease [ 3 ] Se creat, Se creatinine units of measurement as per study reports. RR, relative risk; HR, hazard ratio; CI, confidence interval. *Adjusted comparison if reported. Decreased GFR as a predictor of mortality [ 3 ] Se creat, Se creatinine units of measurement as per study reports. RR, relative risk; OR, odds ratio; HR, hazard ratio. *Adjusted comparison if reported. Decreased GFR as a predictor of mortality [ 3 ] Se creat, Se creatinine units of measurement as per study reports. RR, relative risk; OR, odds ratio; HR, hazard ratio. *Adjusted comparison if reported. In addition, they identified five studies which considered proteinuria as a predictor of cardiovascular disease [ 46–50 ]. Only one of these studies, Culleton et al. [ 48 ] failed to report a statistically significant association with at least one cardiovascular outcome for at least one gender. Six studies were identified which considered proteinuria as a predictor of cardiovascular mortality [ 46 , 48 , 50–53 ]. Five reported a statistically significant association for at least one gender. Nearly all of these studies were considered to be of good or of moderate quality in terms of avoidance of bias. The measures of proteinuria however varied between studies. Though not completely consistent, of varying methodological quality and using different assessments of GFR and proteinuria and examining different populations the final K/DOQUI guidelines found the scientific evidence at the time persuasive that their was an association between early CKD and CVD, cardiovascular death and total mortality. However, causation was certainly not demonstrated [ 3 ]. It is possible that reduced GFR and proteinuria are just markers of severity or duration of traditional cardiovascular risk factors [ 3 ]. Even though it accepted the limitations of available epidemiology studies describing an association between early CKD and cardiovascular disease, agreed that causation was far from determined and highlighted the lack of interventional studies of risk factor modification, the K/DOQI guidelines ultimately concluded that it was probably acceptable to extrapolate risk factor modification data from the general population to the CKD population. It supported this position by indicating that traditional cardiovascular risk factors could be modified in this population (hypertension and hyperlipidaemia respond to treatment), there was little evidence that therapy to modify these factors held greater risk in this group and finally, that most patients with CKD lived long enough for the benefits of risk modification to be realistically realised [ 3 ]. Most of these and subsequent studies which supported CKD as an independent risk factor for CVD were undertaken in populations with known cardiovascular disease and using recurrent events as the outcome or in populations with a known high risk profile for cardiovascular events and recording incident events. Studies of a more general population with incident cardiovascular events as the outcome were less able to demonstrate that CKD was an independent risk factor. Culleton et al. [ 37 ] analysed data from 6233 participants in the Framingham Heart Study. Though CKD appeared to be an independent risk factor after multivariable-adjustment for all-cause mortality in men [hazard ratio (HR) 1.31, 95%CI 1.02–1.67] it was not an independent risk factor for CVD events in men (HR 1.17, 95% CI 0.88–1.57); for CVD events in women (HR 1.04, 95%CI 0.79–1.37) nor all-cause mortality in women (HR 1.08, 95%CI 0.87–1.34). After adjusting data for traditional cardiovascular risk factors from 2352 adult participants without baseline cardiovascular disease from the US First National Health and Nutrition Examination Survey (NHANES I, 1974–1975) and NHANES I Epidemiologic Follow-up Study (NHEFS, 1992) Garg et al. [ 54 ] found that moderate CKD (GFR 30–60 mls/min) was not an independent risk factor for total mortality (HR 1.0, 95% CI 0.8–1.4) nor cardiovascular mortality (HR 1.2, 95% CI 0.8–1.8). Garg et al. [ 54 ] also reviewed previous prospective studies (some included in K/DOQI guidelines) with more than 1000 participants and concluded that an independent association between CKD and cardiovascular disease was most clear cut in study populations with traditional cardiovascular risk factors such as pre-existing cardiovascular disease [ 33 , 42 , 55 ] hypertension [ 30 , 31 , 38 ] or older age [ 40 , 56 ]. In the general population they suggested that the association is either weak [ 39 ] or not there [ 37 ]. In 2004, Weiner et al. [ 57 ] pooled data, excluding participants with known CVD, from four community-based longitudinal studies: Athersclerotic Risk in Communities Study [ 58 ], Cardiovascular Health Study [ 59 ], Framingham Heart Study [ 60 ] and the Framingham Offspring Study [ 61 ]. Of the total pooled population of 22 634, 7.4% had CKD (defined as GFR between 15 and 60 mls/min, stage 3/4 CKD). The study concluded that after adjusting for study, demographic variables, educational status and other cardiovascular risk factors CKD was not an independent risk factor in white individuals for the cardiovascular outcomes (myocardial infarction, fatal coronary heart disease and stroke) though for the primary composite study end point which included these cardiovascular outcomes plus death it was an independent risk factor (HR 1.13, 95% CI 1.02–1.26). CKD appeared to be a more important risk factor for the primary composite end point for black individuals (HR 1.76, 95% CI 1.35–2.31). More recently, Weiner and colleagues [ 62 ] again analysed data from the same pooled studies but on this occasion reported that both baseline CKD (HR 1.26%, 95% CI 1.16–1.35, P < 0.0001) and baseline CVD (HR 1.83, 95%CI 1.72–1.95, P < 0.0001) were independent risk factors for the primary composite outcome (stroke, cardiac events and death) in this general population. However, the effect of having both CKD and CVD at baseline on the primary outcome appeared to be additive and not synergistic. Early CKD appears to be an independent risk factor for cardiovascular events in individuals who already have CVD or have traditional risk factors for the development of CVD such as hypertension or diabetes. It remains uncertain whether this is a causal relationship by means of one or more of the non-traditional CVD risk factors linked with CKD such as hyperparathyroidism or hyperhomocysteinaemia. It is possible that reduced GFR or proteinuria are just markers for the severity of the underlying CVD or for the severity, number and duration of traditional CVD risk factors. Early CKD has not yet been clearly demonstrated to be an independent risk factor for CVD events in the general population. Future identification, classification in terms of CKD severity and registration should facilitate long-term, prospective observational studies to clarify whether CKD is an independent risk factor for CVD in this population. Though many of the large primary and secondary CVD risk factor modification studies excluded patients with CKD it is reasonable to extrapolate and use interventions proven to be of benefit in patients with CVD without CKD to patients with CVD or traditional CVD risk factors and CKD. There is no evidence that patients with early CKD do not respond to these interventions or suffer excess adverse effects. Studies of new agents or interventions aimed at traditional risk factor modification should specifically include participants with early CKD. Interventional studies aimed at non-traditional risk factors may help elucidate any causal relationship between risk factor and CVD. Perhaps in future years non-traditional risk factor identification and modification in the CKD population may both elucidate pathogenesis of CVD and suggest novel interventions in the general population. If CVD risk factor modification decreases CVD events in the CKD population it would be naïve to assume that it would necessarily be accompanied by an equal reduction in the progression from early CKD to CKD 5 and the need for RRT. A burgeoning dialysis population may be the ultimate marker of success of more active management of early CKD. Conflict of interest statement . None declared.