OUP user menu

Chronic kidney disease and statin therapy: to treat or not to treat?

C. Wanner
DOI: http://dx.doi.org/10.1093/eurheartj/eht136 1772-1774 First published online: 19 April 2013

This editorial refers to ‘Effect of statin therapy on cardiovascular and renal outcomes in patients with chronic kidney disease: a systematic review and meta-analysis’, by W. Hou et al., on page 1807

Scientific knowledge and available good quality studies are ranked according to an accepted ‘evidence ladder’. The ladder is headed by a randomized controlled trial (RCT), if possible confirmed by a second, followed by prospective cohort studies and case–control studies. The latter are observational studies, usually subject to bias and confounding. In case the evidence is insufficient or of moderate quality, systematic reviews are undertaken, in order to integrate the available knowledge into a meta-analysis which then goes to the top of the hierarchy of evidence.

Hou and colleagues now present a meta-analysis from a therapeutic area which has been studied most extensively in clinical medicine.1 The evidence is abundant and the issue of statin therapy for elevated LDL-cholesterol in persons from the general population is considered as being solved.2 Apparently this is not the case for patients with chronic kidney disease (CKD), and the debate on who to treat and not to treat is ongoing. Thus, the analyses of Hou and colleagues contribute to solving these uncertainties. In their analysis, statin therapy produced a 23% reduction in the risk of major cardiovascular events (but not cerebrovascular disease) or an 18% per 1 mmol/L lowering of LDL-cholesterol in patients with CKD including those receiving dialysis. These results may be discussed controversially among nephrologists caring for dialysis patients. It is important to recognize that in this meta-analysis, trials from non-dialysis and dialysis patients were combined, although the trials exhibit different levels of hierarchy on the ‘evidence ladder’. In the absence of many RCTs, meta-analyses tend to include observational studies, and post-hoc analysis of large statin trials from the general population. It is certainly permissible to ask whether data from CKD subgroups with mild or moderately impaired kidney function, extracted from the general population lipid trials by the use of glomerular filtration rate (GFR) equations, should be combined with data from RCTs carried out in dialysis patients. Hou et al. decided to do so by justifying that all the available clinical trial data should be synthesized in order to better define the balance of risks and benefits of statin therapy in patients with CKD and also the effect of kidney function on statin use. They also observed that the beneficial effects demonstrated a progressively smaller relative risk reduction with lower estimated GFR (eGFR) and appeared to be smaller in people with stage 5 kidney disease and those requiring dialysis. Thus, subgroup analysis showed that the statin effect was significantly modified by kidney function.

Recently, another meta-analysis by Palmer and colleagues3 has reported similar results by looking separately at stages of CKD and renal replacement therapy (dialysis or transplantation). They found that statins decrease mortality and cardiovascular events in persons with early stages of CKD and have little or no effect in persons receiving dialysis. To compare the published evidence, Table 1 outlines the data of Hou and co-workers as well as those of two recently published meta-analyses.1,3,4 The differences in study selection, analysis criteria and outcome parameters are apparent. The selection criteria applied to studies for subsequent analysis lead to a different number between 18 and 80 studies. Thus, every meta-analysis has to be looked at separately according to its research aim and selection criteria in order to appreciate the differences.

View this table:
Table 1

Common characteristics and differences of systematic reviews and meta-analyses

Hou et al.1Palmer et al.3Upadhyay et al.4
Search period1970–November 2011February 20122000–November 2011
Source (databases)Cochrane, EMBASE, MedlineCochrane, EMBASEMedline, Cochrane
Type of studiesRCTsRTsRCTs
No. of studies318018
Type of patientsCKDCKD, TXCKD, TX
No. of patients48 42951 09936 528
Type of events6690 MCV events11 782 MCV events
6653 deaths11 363 deaths
Non-dialysisDialysisNon-dialysisDialysisND + dialysis
All-cause mortality0.86 (0.73–1.00)0.96 (0.90–1.02)0.81 (0.74–0.88)0.96 (0.88–1.04)0.91 (0.83–0.99)
CV death0.86 (0.77–0.96)0.96 (0.86–1.08)0.78 (0.68–0.89)0.94 (0.82–1.07)0.82 (0.74–0.91)
CV events0.76 (0.73–0.80)0.95 (0.87–1.03)0.78 (0.7–0.86)
MI0.55 (0.42–0.72)0.87 (0.71–1.07)0.74 (0.6–0.81)
Stroke0.61 (0.39–0.95)1.16 (0.91–1.47)0.61 (0.38–0.98)1.30 (0.79–2.11)
CE0.69 (0.58–0.83)0.91 (0.81–1.02)
MCE0.77 (0.70–0.84)
CE or death0.91 (0.84–0.99)
Kidney failure0.95 (0.90–1.01)
Adverse hepatic events0.87 (0.92–1.39)NSNS
Muscular disorders1.02 (0.95–1.09)NSNS
  • CE, coronary events; CKD, chronic kidney disease; CV, cardiovascular; MCE, major cardiovascular events; MI, myocardial infarction; ND, non-dialysis; NS, not significant; RCT, randomized controlled trial; RT, randomized trial; TX, transplantation.

Hou and co-workers bring their results into context and explain methods of data extraction, quality assessment, and outcome estimation.1 Although controversial, they are to some extent in line with many clinicians who separate patients into a pre-dialysis group (stage 3–5 CKD) and a dialysis group (stage 5D). This approach is used because the disease is heterogeneous with respect to co-morbidities and vascular pathology. The risk of dying during progression of kidney disease, before entering a dialysis programme, is high. Those who develop end-stage renal disease are considered a prevalent population of survivors, and data indicate that vascular disease is transforming from atherosclerosis to arteriosclerosis. Causes of death are of cardiac and vascular origin, including stroke. In the late stage, risk factors also differ substantially (traditional and non-traditional or uraemia related) and treatment approaches are urgently awaited that take the underlying pathophysiology into account. Thus, analyses according to stages of CKD, or to pre-dialysis vs. dialysis, would match the clinical situation and how clinicians would like to approach their patients.

The backbone of all meta-analyses outlined in Table 1 is three RCTs, the 4D study (Die Deutsche Diabetes Dialyse Studie), AURORA (A Study to Evaluate the Use of Rosuvastatin in Subjects on Regular Hemodialysis: An Assessment of Survival and Cardiovascular Events), and SHARP (Study of Heart and Renal Protection). 4D and AURORA assigned >4000 diabetic and non-diabetic dialysis patients to statin or placebo therapy, whereas SHARP included a mixed cohort of 6029 patients not on dialysis (mean GFR 27 mL/min/1.73 m2) and 3023 patients on dialysis.57

A difficulty common to all the meta-analyses in the interpretation of the randomized trials, as well as the observational trials with CKD subgroup results, is that primary endpoints are not uniform. This makes comparison of the trials and the interpretation of the meta-analysis difficult. In addition, endpoints of the three RCTs were adjudicated differently. Re-adjudication of all patients and analysis of single data of individual patients in the trials would be ideal, but most probably is not feasible.

The present analysis of Hou et al. is also of value for future guideline writing, such as an update of the ESC/EAS Guidelines for the management of dyslipidaemias8 or the KDIGO (Kidney Disease Improving Global Outcomes). The US National Kidney Foundation KDOQI (Kidney Disease Outcomes Quality Initiative) has recently released a 2012 update recommendation of the 2007 KDOQI clinical practice guidelines for diabetes and CKD. They recommend not initiating statin therapy in patients with diabetes who are treated by dialysis.9 It is interesting to note that these guidelines, based on clinical trial results, have not been followed in the past,10 and physicians may have preferred to wait for a more condensed analysis.

The finding of a smaller relative risk reduction with lower GFR may reactivate the discussion on the search for the so-called point of no return, beyond which statins may lose their benefit. However, a strategy such as treat early and stop treatment once the patient has reached dialysis is not applicable in clinical practice. A wait and see strategy in the presence of conflict and unresolved issues is also not appropriate since vascular changes are logically a continuous process. For example, the SHARP study has shown a significant 17% relative risk reduction in 9052 patients across all stages of CKD. In fact, 3023 patients were already on dialysis and 2066 started dialysis therapy during the course of the study, making the discussion of a responsive pre-dialysis group vs. a non-responsive dialysis condition less valid. In fact, discussion along these line is erroneous since treatment in earlier stages of nephropathy is effective, and most, if not all, patients will be on a statin regimen once they start dialysis. Therefore, few practical consequences emerge when following the KDOQI guideline ‘not to initiate treatment’, because stopping treatment is not recommended by any guideline.

Despite the analysis of Hou et al. and in light of the other meta-analyses, the debate ‘to treat or not to treat’ will continue among nephrologists who care for elderly dialysis patients with strong symptoms of wasting. The approach to handle the situation from a clinical point of view may be most appropriate. Parameters such as advanced age, frailty, and short life expectancy provide a good rationale to limit treatment efforts in dialysis patients. In the case of doubt, one can be reassured by the present analysis that no harm is apparent from doses of statins given in the various trials.

Conflict of interest: C.W. received lecture honoraria from MSD and Astellas.


  • The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal or of the European Society of Cardiology.

  • doi:10.1093/eurheartj/eht065.


View Abstract