Reverse epidemiology is not necessarily reverse aetiology, argue Dr Craig Nelson and Professor Gavin J Becker of the department of nephrology, Royal Melbourne Hospital, Victoria, Australia.

Death from cardiovascular disease (CVD) in end-stage renal failure (ESRF) is 20 to 40 times higher than in the general population. This may be more exaggerated in those dialysis patients at a younger age where the incidence may be greater than 100 times. Figures from Australia and New Zealand Dialysis and Transplant Registry (ANZDATA) report 2004, suggest 40 per cent of deaths of patients on haemodialysis are of cardiac origin with 21 per cent of these relating to acute myocardial infarction (AMI). The inverse association of cardiovascular events and survival with classic vascular risk factors in this population, termed ‘reverse epidemiology’, has generated much interest. In this review we discuss issues relating to blood pressure, body mass index (BMI) and cholesterol, and the biases and hypotheses that may explain this association. Definitive answers will only be found in properly designed prospective randomised intervention trials.

Blood Pressure

High blood pressure is an accepted major cardiovascular risk in the general population and randomised trials have proved drug therapy decreases risk of stroke, coronary heart disease and mortality, however in the ESRF population, low blood pressure appears more dangerous. Up to 80 per cent of individuals with pre-dialysis renal disease have hypertension. There is a greater prevalence of hypertension as glomerular filtration rate (GFR) declines and proteinuria increases. Observational studies have demonstrated that higher systolic blood pressure is an independent risk factor for left ventricular hypertrophy (LVH) and for CVD events in non-diabetic and diabetic chronic kidney disease (CKD). Intervention with antihypertensives in CKD is not only associated with reduced renal decline but also reduces cardiovascular complications in those with mild and moderate renal impairment.

However, the relationship between blood pressure and CVD in dialysis patients is complex and may be U-shaped. Higher blood pressure has been shown to be an independent risk factor for left ventricular hypertrophy (LVH), ischaemic heart disease (IHD), and congestive cardiac failure (CCF) in a prospective study. Low blood pressure, on the other hand, has prospectively been associated with a higher risk of all-cause mortality. This increased risk for all-cause mortality in individuals with lower blood pressure forms part of the concept of reverse epidemiology of classic vascular risk in end-stage renal failure (ESRF). This finding may be due to confounding from comorbidities associated with lower blood pressure, such as structural cardiac disease, leading to a higher risk of CVD events.

Body Mass Index

Obesity is an established independent risk factor for cardiovascular disease in the general population, yet obesity seems protective, at least in the haemodialysis population. It is also known to accelerate renal decline in CKD, and more recently has been implicated in the onset of CKD itself. Conversely, a one-year prospective study has suggested a survival advantage for those on haemodialysis with BMI in the highest quartile, compared with middle and lower quartiles of BMI. Similar results have been obtained in retrospective analysis of a prospective cohort, the dialysis outcomes and practice patterns study (DOPPS), suggesting a ‘reverse epidemiology’ with respect to BMI and CVD and survival in dialysis patients.

Cholesterol

Cholesterol is a known risk factor for vascular disease in the general population and lowering cholesterol for primary and secondary prevention in this population has been shown to be associated with improved vascular outcomes and survival, yet in ESRF a low cholesterol is an all-cause mortality risk factor.

In this setting, renal disease dyslipidaemia is complex. The prevalence of dyslipidaemia in CKD
varies according to GFR and level of proteinuria. As GFR declines, triglyceride levels increase and
HDL cholesterol levels decrease. As proteinuria increases, total cholesterol, LDL cholesterol, and triglyceride levels all increase, while HDL cholesterol levels decrease. There is very little data linking dyslipidaemia with CVD events in renal disease although one study has shown an association of low HDL with CVD outcomes.

Selection Bias

The impact of renal disease on CVD and its associated mortality may be grossly underestimated by addressing the ESRF group alone. Similarly, given the fluctuating nature of classic vascular risk profiles at different stages of disease and indeed with different treatment or renal replacement modalities, it is impossible to assume that measurement at a single time point reflects the true impact of these risk factors on the overall clinical course.

Incidence and prevalence data suggests people are dying from CKDrelated cardiovascular disease before reaching renal replacement therapy (dialysis or transplant). The ANZDATA registry report 2004 suggests that of those new to the commencement of renal replacement therapy (dialysis or transplant), 30.5 per cent have known coronary artery disease, 19 per cent known peripheral vascular disease and 11 per cent known cerebrovascular disease, suggesting the problem is already present at the time of commencement of dialysis. Hence CVD would seem to have its aetiology in CKD.

Classic Risk Factors

A study of classic risk factors in CKD has shown 89 per cent of patients had hypertension, 68 per cent had dyslipidaemia, 32 per cent were diabetic and 38 per cent were previously smokers. In the AUSDIAB study, age, diabetes mellitus, and hypertension were independently associated with proteinuria; age, male gender and hypertension with haematuria; and age, male gender, and hypertension with reduced GFR.

Studies applying the Framingham risk scores from general population risk profiles to people with CKD and those on haemodialysis have found similar or slightly greater coronary risk scores to those of the reference populations. This did not mirror the increased incidence of CVD seen in these population groups. It is possible to hypothesise from this that there may be qualitative changes in these risk factors or other non-traditional risk factors in CKD and ESRF compared with those bserved in the general population. The most likely explanation is that this scoring system is inaccurate in these patient populations, as the data was not derived from similar patient groups.

Survival Bias

An alternative explanation for reverse epidemiology is that the population reaching renal replacement therapy may represent a subgroup with vascular disease that does not have grounding in the classic vascular risks. This,however, does not explain why those subjects with improved classic vascular risk profiles do worse than those with higher classic vascular risk factors at this stage of disease.

Inflammation and Cachexia

Reverse epidemiology and cardiovascular mortality have been associated not only with the ESRF population, but also other end-stage illnesses, including chronic heart failure, malignancies, AIDS and advancing age. Cachexia is a wasting syndrome that frequently develops in the setting of chronic diseases including cancer, congestive heart failure, chronic obstructive pulmonary disease, AIDS, renal failure and liver failure.

Cachexia is a potential confounder in reverse epidemiology and is linked to the malnutrition-inflammation complex syndrome (MICS).

Beware Jumping to Conclusions

The assumption of ‘reverse epidemiology’ as the reverse aetiology of vascular risks in ESRF is difficult to justify in terms of the numerous potential biases involved as outlined above. Classic vascular risks are associated with the onset of CKD, the progression of CKD and many patients die of vascular disease before reaching renal replacement therapy.

Pending further research, we have no choice but to assume that the traditional risk factors have a bearing on cardiovascular mortality in ESRF patients, However, the caveats regarding the lack of an evidence base must be taken into consideration when prescribing intervention in the individual patient.