|Home | About | Journals | Submit | Contact Us | Français|
Chronic kidney disease (CKD) is associated with a high risk of death from coronary artery disease and may modify the response to standard cardiovascular therapies. Treatment of subjects with CKD should ideally be based on evidence from randomized, clinical trials, but how often subjects with CKD have been excluded from these trials is uncertain. We undertook this study in order to quantify how often subjects with moderate to advanced CKD were excluded from large cardiovascular trials. MEDLINE and the reference list of selected articles were searched in order to identify large, randomized, controlled trials of five different coronary artery disease therapies published between 1998 and 2005. Exclusion criteria and reported clinical characteristics of subjects were abstracted. Rates of exclusion and reporting of baseline characteristics of study participants were compared for CKD, diabetes, history of smoking, and hypertension. Eighty-six trials randomizing 411 653 patients were identified. More than 80% of trials excluded subjects with end-stage renal disease and 75.0% excluded patients with CKD. Subjects with diabetes, hypertension, or a history of smoking were excluded less than 4% of the time. Baseline renal function of study participant was reported in only 7% of trials. Patients with CKD are frequently excluded from coronary artery disease trials and renal function of randomized subjects is rarely reported. These findings reinforce the notion that available data on the treatment of coronary artery disease in subjects with CKD have significant limitations and should be generalized to the treatment of subjects with CKD cautiously.
Roughly 11% of the US population has chronic kidney disease (CKD),1 and by 2030 there will be more than two million people with end-stage renal disease (ESRD) and many times that number with moderate impairment of kidney function.2 Both ESRD and moderate CKD are associated with high risks of death from coronary artery disease.3,4 This risk is not fully explained by traditional cardiovascular risk factors, and the usual relationship of these risk factors to cardiovascular outcomes is significantly altered in the setting of renal failure.3–10
Given the unique features of coronary artery disease in CKD it is possible that established cardiovascular therapies, or at least those tested solely in populations with normal renal function, will prove to be less effective in patients with CKD than in those with normal renal function. Alternatively, the high baseline risk in patients with CKD could magnify the absolute benefit of treatment with standard cardiovascular therapies. Indeed, the importance of specifically testing established cardiovascular therapies in subjects with CKD was highlighted by the negative findings of two recent trials that randomized patients with advanced kidney disease to treatment with statins vs placebo.11,12 Thus, both theoretical considerations and the available randomized evidence suggest that uncertainty about the appropriate role for standard therapies in subjects with CKD is warranted when those therapies have not been broadly tested on subjects with impaired kidney function.
How often renal function has been used to exclude participation in clinical trials and whether renal function is used to exclude participation in clinical trials more frequently than other high-risk conditions have not been systematically studied or quantified. We undertook the current study in order to systematically determine whether subjects with moderate CKD or dialysis-dependent ESRD are excluded from enrolment in large, coronary artery disease trials more frequently than subjects with other high-risk conditions such as diabetes, hypertension, or smoking and to compare the extent to which those trials report on the presence of these conditions at baseline.
A total of 3076 articles were identified via electronic and hand searches. As shown in Figure 1, 86 trials randomizing 411 653 subjects met the inclusion criteria and were selected for further analysis. Twenty-one trials randomized patients to GPIIb–IIIa inhibitors, 15 to oral anti-platelet agents, 30 to statins, seven to beta blockers, and 18 to percutaneous coronary intervention or devices. Five trials randomized patients to more than one of these interventions (Table 1).
Subjects with chronic renal insufficiency or ESRD were most likely to be excluded from trials of statins or GIIb/IIIa inhibitors –90% (95% confidence interval (CI): 73.5–97.9) of statin trials and 85.7% (95% CI: 63.7–97.0) GIIb/IIIa trials excluded subjects with either condition. Trials of oral anti-platelet agents included subjects with CKD more frequently, but approximately half used CKD as an exclusion criteria – 60% (95% CI: 32.3–83.7) excluded subjects with ESRD and 46.6% (95% CI: 16.3–67.7) excluded subjects with earlier stages of CKD.
Exclusion of patients with CKD was typically based on the serum creatinine. In only five trials was the threshold based on estimated GFR or creatinine clearance. A threshold creatinine clearance of 30 cm3/min or less was used in five trials. Two trials used a threshold creatinine below 1.5 mg/dl, 35 trials used a threshold between 1.5 and 2.0 mg/dl, and 18 trials used a threshold creatinine above 2.0 mg/dl.
All five types of trials were unlikely to report the baseline renal function or frequency of renal impairment in trial participants. This information was reported most frequently in trials of beta blocking agents, but was still available only 28.6% (95% CI: 3.7–70.9) of the time in this setting. Results from the individual trials are summarized in Tables 2–6.
Out of 86 trials, 69 (80.6%, 95% CI: 70.2–88.0) excluded patients with ESRD and 64 (74.4, 95% CI: 63.9–88.2) excluded patients with moderate to severe renal insufficiency. Patients with diabetes, hypertension, or a history of smoking were rarely excluded from participation in trials. Only three studies (3.5%, 95% CI: 0.7–9.8) excluded patients with diabetes and no study excluded patients with hypertension or a history of smoking (P < 0.0001 for all comparisons, Table 1 and Figure 2).
The percentage of patients with impaired renal function or the mean baseline serum creatinine of randomized patients was reported by only six of 86 studies (7.0, 95% CI: 2.6–14.6). The percentage of patients with hypertension (or mean baseline blood pressure), diabetes, or a history of smoking was all reported significantly more frequently (P < 0.0001 for all comparisons). Baseline blood pressure and smoking history were each recorded in 70 of 86 (81.4, 95% CI: 71.6–89.0) studies. The baseline percentage of subjects with diabetes was reported in 75 out of 83 studies (90.4%, 95% CI: 81.9–95.7) that included diabetic patients, Figure 3.
We systematically reviewed the reports of large, randomized coronary artery disease trials published between 1998 and 2005 to assess whether these trials exclude patients with moderate or dialysis-dependent CKD more frequently than they exclude subjects with other cardiovascular risk-factors. We found that more than 80% of trials exclude patients with ESRD and nearly 75% exclude patients with moderate renal insufficiency whereas subjects with other common risk factors for cardiovascular disease are excluded only rarely. Further, we found that fewer than 10% of trials provide information on baseline serum creatinine. Data on the estimated glomerular filtration rate or creatinine clearance of randomized subjects, which are better markers of renal function than serum creatinine alone, are provided even less frequently. Our results thus demonstrate that among common cardiovascular risk factors, there is a unique failure of large cardiovascular trials to produce data on the treatment of cardiovascular disease in the setting of moderate or advanced CKD.
This failure to test coronary artery disease therapies in patients with advanced renal insufficiency has significant implications when considered in light of elevated cardiovascular mortality rates in subjects with CKD,4 the fact that as many as 50% of subjects admitted with a myocardial infarction have stage 3 or worse CKD13 and the common failure to administer standard cardiovascular therapies to patients with CKD even when they are diagnosed with myocardial infarction.14–16 Because there is insufficient randomized evidence on the effectiveness of typical therapies in subjects with CKD, it is difficult to know whether this low use represents appropriate concern about the use of unproven therapies, appropriate response to comorbidity in patients with CKD,17 or whether it partially explains the high rates of cardiovascular morbidity and mortality in the CKD population.5–8,18
More importantly, there is a growing body of literature suggesting that standard treatment strategies may act differently in patients with and without CKD 14,15,19 – a concept reinforced by two trials of statins that were conducted in populations with CKD.11,12 Available evidence, including the present study, thus suggests that the general standard of care for coronary artery disease should be extrapolated to subjects with CKD cautiously, except in those rare cases where subjects with CKD have been randomized in significant numbers.
This failure of large coronary artery disease trials to include patients with CKD is particularly concerning when one considers that the number of randomized controlled trail published in nephrology is low and compounded by poor quality and reporting20 and thus unlikely to provide definitive answers on cardiovascular care in this population. Our findings, highlight and quantify the limitations of current evidence on the treatment of cardiovascular disease in subjects with CKD and provide a strong rationale for including greater numbers of subjects with CKD in future trials or for specifically targeting subjects with CKD as the population for future trials of standard and emerging therapies of coronary artery disease. Finally, the reliance of serum creatinine as the measure of renal function in the majority of studies we analyzed highlights the need to educate cardiovascular trialists on the availability of better estimates of renal function for use during randomization and follow-up of patients.
Our findings should be interpreted within the context of our methodology. Our search was limited to peer-reviewed trials randomizing at least 1000 patients to five different therapies. We cannot rule out the possibility that small trials or trials of other therapies are more likely to include patients with kidney disease. However, our results were consistent across a broad array of both medical and interventional therapies. Furthermore, large trials with power to provide definitive answers to clinical questions provide important data that profoundly influences evidenced-based medical practice. The clinical importance of our findings would be only slightly diminished in the event that small or unpublished trials are more likely to enrol subjects with kidney disease than the trials we studied.
We did not directly assess whether separate, smaller trials are being conducted in patients with renal insufficiency. Nevertheless, our search strategy incidentally identified only a single, randomized trial published between 1998 and March 2005 that tested whether a drug or device prevents myocardial infarction or cardiovascular death in patients with renal insufficiency.21
Finally, the low rate of reporting on the baseline renal function of trial participants that we found does not account for the possibility of subsequent publication of subgroup analyses of the treatment effect in subjects with vs without renal insufficiency. Such publications can partially mitigate the import of an initial failure to report the percent of subjects with renal disease, but that initial failure nevertheless reflects a markedly different approach by clinical trialists towards CKD as a coronary disease risk factor than they have for other common conditions such as hypertension, diabetes, and smoking. Given the high risk of cardiovascular events in subjects with CKD and high frequency of this condition in the general population, we find this practice puzzling. Additionally, subgroup analyses, particularly if not pre-specified, are susceptible to false-positive and false-negative findings.22 Further, their publication is frequently delayed by several years compared to the initial publication of the overall results. For both reasons, having enough information to draw conclusions about a trial’s relevance to subjects with CKD at the time of initial publication would be preferable to current practices.
An important question raised by our findings is whether there are sufficiently compelling reasons to exclude patients with CKD despite the clear need for better data about this group. Inclusion of a new subgroup of patients in cardiovascular trials might skew the overall results or the power to detect a therapeutic effect by altering the rate of cardiovascular events, the case-fatality rate, or the frequency of non-cardiovascular death in the trial cohort, although stratified randomization or limiting the numbers of randomized patients with a particularly high-risk condition can be used to mitigate the influence of that condition on trial results. Moreover, the risk of cardiovascular disease among patients with moderate renal impairment is actually similar to the risk among patients with diabetes, smoking, or hypertension,5–9,18,23 and randomized trials are unlikely to randomize substantially more patients with moderate or advanced renal insufficiency than they currently randomize with diabetes or hypertension. The risk profile is significantly different in patients with ESRD where the mortality rate is significantly higher than in the general population.24 However, roughly half of deaths in patients with ESRD are due to cardiovascular disease,25 and any loss of power owing to an increased rate of non-cardiovascular deaths might well be neutralized by a concurrent increase in the rate of cardiovascular morbidity.
A reduction in the glomerular filtration rate can alter the safety profile of some drugs and devices, and might be an appropriate reason to exclude patients from participation in trials, especially phase I and phase II trials. However, none of the therapies we studied is contraindicated in patients with renal insufficiency. Indeed, a high proportion of trials of beta blockers, statins, and anti-platelet therapies excluded patients with renal insufficiency despite the widespread clinical use of these agents in studies on patients with ESRD.14
Finally, if the altered epidemiology and pathophysiology of cardiovascular disease in patients with renal disease10 confers resistance to standard therapies, then the inclusion of patients with renal insufficiency in trials will decrease the observed treatment effect and bias towards the null hypothesis. Belief in this theory may partially explain why patients with renal disease are so frequently excluded from cardiovascular trials, but it remains an unproven hypothesis that at a minimum deserves to be tested in clinical trials.
A significant and growing minority of people in this country has CKD. They suffer from an increased risk of developing and dying from coronary artery disease and whether they respond differently to cardiovascular therapies is uncertain. It is therefore crucial that we advance our knowledge of how to treat and prevent cardiovascular disease in this population. Unfortunately, large, contemporary trials have largely ignored these patients despite compelling reasons to increase our understanding of the unique features of their disease. Government and industry-funded trials must enrol greater numbers of patients with impaired kidney function in trials or should specifically target trials towards the CKD population so that we can provide the best possible cardiovascular care for this crucial population.
Our objective in this study was to understand how contemporary clinical research applies to patients with renal disease. Because small trials or trials published only in abstract form are less likely to influence clinical practice or lead to Food and Drug Administration approval of new drugs and devices than larger trials that have been published in peer-reviewed journals, we restricted our analysis to published trials that randomized at least 1000 subjects. We looked at five different interventions selected because as a group they represent the standard therapies now used for the treatment of coronary artery disease. Trials randomizing subjects to beta-blockers, GIIb/IIIa inhibitors, non-GIIb/IIIa oral anti-platelet agents, HMG-CoA reductase inhibitors, or a percutaneous coronary intervention or coronary stents were included in this study. In order to examine contemporary practices, we restricted our analysis to trials whose principal results were published in peer-reviewed journals between January 1998 and March 2005.
Trials that studied the aforementioned agents for purposes other than the treatment of coronary artery disease were excluded from further analysis. Thus, trials were included when at least one of the following conditions was met: (a) ischemic coronary artery disease was a prerequisite for randomization; (b) the primary end point assessed the occurrence of new or recurrent coronary artery disease that included myocardial infarction, coronary revascularization, and cardiovascular death; and (c) the primary end point assessed the effect of a medication on the low-density lipoprotein cholesterol concentration. The last criterion was chosen because low-density lipoprotein cholesterol lowering has been widely accepted as a valid surrogate end point for statin therapy, is widely used to guide the use of statins, and has been used to justify the approval of new statin medications.
We chose not to study trials of angiotensin-converting enzyme inhibitors) and angiotensin receptor blockers. Angiotensin-converting enzyme inhibitors and angiotensin receptor blockers have recently emerged as important agents in the treatment and prevention of cardiovascular disease. However, in contrast to the other therapies we studied, the use of angiotensin-converting enzyme inhibitors and/or angiotensin receptor blockers forms the standard of care in the therapy of CKD. These agents are routinely prescribed to reduce proteinuria and slow progression of renal disease in patients with CKD regardless of their role in the treatment of cardiovascular disease. Thus, whereas it would have been interesting to study trials of these medications, we felt that the clear nephrologic indications for their use mitigated the clinical impact of exclusion of patients with CKD from cardiovascular trials of angiotensin-converting enzyme inhibitors and angiotensin receptor blockers.
MEDLINE searches using the PubMed interface were conducted for the period between January 1998 and March 2005. Search terms included the following MeSH headings: beta blocker, platelet glycoprotein GPIIb–IIIa complex, HMG-CoA reductase inhibitors, angioplasty, stent, aspirin, and ticlopidine. Additional search terms included the generic names of individual beta blockers, GIIb/IIIa inhibitors, statins, and anti-platelet agents. The following limits were used: English-language; randomized, controlled trial; and adults greater than 19 years old. The reference lists of selected trials were also manually reviewed in order to identify additional studies. Eligibility for inclusion was determined after review of titles, abstracts or, where necessary, the full manuscript.
For each trial, the exclusion criteria were abstracted from the published reports. Exclusion of subjects with CKD was defined as the use of ‘renal insufficiency’ or a threshold of serum creatinine, blood urea nitrogen, or creatinine clearance to exclude subjects from randomization. Authors were contacted for clarification of the exclusion criteria when the published reports were unclear or when individual investigators were allowed to exclude patients based on the presence of conditions they believed would compromise life expectancy or a subject’s ability to comply with the assigned treatment. In the event that the authors could not to be reached for clarification, trials were considered to have included patients with CKD. The proportion of studies excluding patients with moderate CKD or dialysis-dependent CKD was calculated from these data.
For comparative purposes, we also assessed how frequently subjects with a clinical history of diabetes, smoking, or hypertension – other common risk factors for cardiovascular disease – were excluded from trial participation. In addition, we abstracted whether studies reported the baseline percentage of patients with diabetes, a history of smoking, CKD, or hypertension. Reporting on the mean baseline serum creatinine was considered equivalent to reporting on the number of subjects with CKD. Reporting mean baseline blood pressure was considered equivalent to reporting on the percentage of subjects with diagnosed hypertension.
Proportions were compared using an exact binomial test. P < 0.05 was considered significant. All calculations were performed in Stata version 8.0 (Stata Corporation, College Station, TX, USA).
This work was supported by NIH Grant T32 DK007527-22. Dr Kuntz is currently employed at Medtronic Inc., a maker of cardiovascular devices.