Incidence and Prognostic Importance of Acute Renal Failure After Percutaneous Coronary Intervention
Background— In patients undergoing percutaneous coronary intervention (PCI) in the modern era, the incidence and prognostic implications of acute renal failure (ARF) are unknown.
Methods and Results— With a retrospective analysis of the Mayo Clinic PCI registry, we determined the incidence of, risk factors for, and prognostic implications of ARF (defined as an increase in serum creatinine [Cr] >0.5 mg/dL from baseline) after PCI. Of 7586 patients, 254 (3.3%) experienced ARF. Among patients with baseline Cr <2.0, the risk of ARF was higher among diabetic than nondiabetic patients, whereas among those with a baseline Cr >2.0, all had a significant risk of ARF. In multivariate analysis, ARF was associated with baseline serum Cr, acute myocardial infarction, shock, and volume of contrast medium administered. Twenty-two percent of patients with ARF died during the index hospitalization compared with only 1.4% of patients without ARF (P<0.0001). After adjustment, ARF remained strongly associated with death. Among hospital survivors with ARF, 1- and 5-year estimated mortality rates were 12.1% and 44.6%, respectively, much greater than the 3.7% and 14.5% mortality rates in patients without ARF (P<0.0001).
Conclusions— The overall incidence of ARF after PCI is low. Diabetic patients with baseline Cr values <2.0 mg/dL are at higher risk than nondiabetic patients, whereas all patients with a serum Cr >2.0 are at high risk for ARF. ARF was highly correlated with death during the index hospitalization and after dismissal.
Received August 14, 2001; revision received March 6, 2002; accepted March 6, 2002.
Renal insufficiency is often encountered in patients who have coronary artery disease. Previous studies have documented the adverse prognostic impact of chronic renal insufficiency after percutaneous coronary intervention (PCI).1,2⇓ Acute, or acute on chronic, renal failure may occur after PCI for multiple reasons, including hemodynamic instability, radiocontrast administration, atheroembolism, and drug toxicity.3–11⇓⇓⇓⇓⇓⇓⇓⇓ In current practice, the incidence and prognostic importance of acute renal failure (ARF) after PCI are unknown. The purposes of this study were to (1) determine the incidence of ARF after PCI in a high-volume cardiovascular referral practice, (2) characterize independent predictors of ARF after PCI, and (3) determine the prognostic implications of ARF, both short and long term, when associated with PCI.
Mayo Clinic Angioplasty Registry
A prospective interventional database has been kept at the Mayo Clinic under the auspices of an institutional review board–approved protocol since coronary angioplasty was first performed in 1979. It includes demographic, clinical, angiographic, and procedural data. Immediate and in-hospital events are recorded, and all patients are subsequently contacted by telephone at 6 and 12 months and yearly thereafter. The Mayo Clinic Institutional Review Board separately approved the present study.
Subjects and Technique
We identified all patients who had coronary interventional procedures from January 1996 through May 2000. No subgroups of patients (for example, those with acute myocardial infarction or shock) were excluded. Only patients who denied research access to their medical records were excluded (n=153), as required by Minnesota law.
In most cases, the femoral approach was used, although arm approaches were used as necessary. Procedures performed included balloon angioplasty, stent deployment, and rotational, directional, and extractional atherectomy. An excimer laser was used in a small number of patients. A single nonionic contrast agent, iopamidol (Isovue, Bracco Diagnostics), has been used almost exclusively (>99% of procedures) in our laboratory over the past 5 years. Patients with preexisting chronic renal insufficiency were admitted for intravenous hydration at the discretion of their physicians.
ARF was defined as an increase in serum creatinine (Cr) concentration of ≥0.5 mg/dL from preprocedure values. Postprocedure Cr values were measured within 48 hours or before dismissal. Successful lesion dilatation after PCI was defined as achievement of <50% residual diameter stenosis, including at least a 20% improvement, by visual estimate. Procedural success was defined as successful treatment of ≥1 lesion without in-hospital death, Q-wave myocardial infarction, or emergency bypass surgery. Myocardial infarction was considered to have occurred if 2 of the following 3 criteria were met: chest pain of ≥20 minutes’ duration, enzyme elevation ≥2 times normal, and development of ST-T-wave changes or new Q waves on electrocardiography.
Results are presented as mean±SD or a percentage of the total. Continuous data were compared by means of Student’s t test; proportions were compared with Pearson’s χ2 test. A number of patients had >1 procedure during the study period. In such cases, only the first procedure was included. In 186 patients, the preprocedure Cr value was missing, but they had normal postprocedure values; in 455 patients, baseline Cr was normal, but the postprocedure value was missing. In these cases, missing values were imputed from the median of the normal Cr values from the remainder of the patients (median 1.1 mg/dL for both preprocedure and postprocedure Cr values). A total of 223 patients (2.9%) were excluded either because of 2 missing values or because of 1 missing value with 1 abnormal value. In this manner, a final sample size of 7586 (97.1%) of 7809 patients undergoing PCI during this time period were identified for this analysis.
Multivariate logistic regression was used to identify correlates of ARF and in-hospital mortality after PCI. Results are presented as odds ratios (ORs) with 95% CIs. Models were developed with stepwise techniques and by consideration of variables that were clinically relevant. Variables included are total volume of contrast medium, age, sex, body mass index, Canadian Heart Association class, history of congestive heart failure, diabetes, hypertension, metastatic cancer, preprocedure shock, tumor, peptic ulcer disease, peripheral vascular disease, and myocardial infarction in the 24 hours before the procedure. Long-term follow-up of patients surviving the hospital phase is presented based on the Kaplan-Meier product-limit method. Survival curves based on the occurrence of ARF were adjusted for other significant predictors of late survival.
Serum Cr was measured a median of 1 day after PCI. Of the 7586 patients in the study population, 254 (3.3%) experienced ARF after PCI, 20 of whom required hemodialysis. As seen in Table 1, these patients were older; more often had a history of congestive heart failure, hypertension, or diabetes; and more often presented with acute myocardial infarction or shock.
Incidence of ARF Stratified by Baseline Cr and Diabetic Status
The observed incidence of ARF after PCI was related to the baseline serum Cr concentration and the presence of diabetes (Table 2). Among patients with a baseline serum Cr <2.0 mg/dL, diabetic patients had a significantly higher risk of ARF than nondiabetic patients (risk 3.7% versus 2.0% for Cr <1.1, P=0.05 and 4.5% versus 1.9% for Cr 1.2 to 1.9, P<0.001). When baseline Cr was >2.0 mg/dL, however, a high proportion of both diabetic and nondiabetic patients experienced ARF (risk 22.4% for Cr 2.0 to 2.9 mg/dL and 30.6% for Cr ≥3.0 mg/dL).
Variables independently associated with ARF after PCI are shown in Table 3. ARF was most strongly associated with increased baseline serum Cr and acute myocardial infarction in the 24-hour interval before the index PCI (OR 1.85, 95% CI 1.31 to 2.63, P=0.0006). Patients who had a successful PCI procedure were much less likely to experience ARF (OR 0.27, 95% CI 0.19 to 0.38, P<0.0001), as were patients who underwent PCI of the right coronary artery. Other significant correlates included age, a history of congestive heart failure, diabetes mellitus, and the presence of peripheral vascular disease. Volume of contrast medium administered at the time of PCI was weakly correlated with ARF (OR 1.12 for each 100-mL increase, 95% CI 1.02 to 1.23, P=0.02).
Procedural Complications by Change in Serum Cr
As shown in Table 4, patients who experienced ARF had a much higher incidence of coronary artery, peripheral vascular, and systemic complications after PCI. The overall procedural success rate was markedly diminished among patients with ARF (72.8% versus 94.0%, P<0.0001), and the risk of Q-wave myocardial infarction was significantly higher in ARF patients (3.9% versus 0.9%, P<0.0001). Twenty-two percent (56) of 254 patients who experienced ARF died in the hospital, compared with 1.4% of patients without ARF (P<0.0001). The vast majority of deaths in both groups were of cardiac causes (Table 5). A greater proportion of ARF patients who died in the hospital had recent myocardial infarction, shock, and 3-vessel coronary artery disease (Table 6).
Patients who experienced ARF had higher rates of femoral access-site bleeding, hematoma formation, and pseudoaneurysms. Other noncardiac complications, such as stroke, coma, adult respiratory distress syndrome, pulmonary embolus, and gastrointestinal hemorrhage, were also significantly more frequent among patients who experienced ARF. A larger proportion of patients who died underwent emergent or urgent procedures (89% versus 67%, P=0.001), whereas a smaller proportion were taking β-blockers (55% versus 64%, P=0.23; Table 6). Among 254 patients who experienced ARF, 6 (10.7%) of 56 patients who died required hemodialysis compared with 14 (7.1%) of 198 survivors (P=0.39). To ascertain whether ARF was independently associated with in-hospital death, we constructed a multivariate logistic model (Table 7). Preprocedural shock (OR 12.12, 95% CI 8.11 to 18.13, P≤0.0001) and ARF (OR 10.83, 95% CI 6.91 to 16.98, P≤0.0001) were by far the strongest predictors of in-hospital death in multivariate analysis. After adjustment for ARF, baseline serum Cr was not significantly associated with in-hospital death.
Long-Term Prognostic Implications of ARF After PCI
The Figure illustrates long-term survival data for patients who survived to hospital dismissal. Among the 6890 patients who did not experience ARF, the Kaplan-Meier estimated risk of death at 6 months, 1 year, and 5 years was 2.3%, 3.7%, and 14.5%, respectively. In contrast, the 185 hospital survivors who experienced ARF had a much higher risk of dying during follow-up, with Kaplan-Meier estimated mortality rates of 9.8%, 12.1%, and 44.6% at 6 months, 1 year, and 5 years, respectively (P<0.0001). Rates of myocardial infarction, both Q wave and non-Q wave, were also significantly higher among patients who had experienced ARF during the index hospitalization. Myocardial infarction rates at 6 months, 1 year, and 5 years were 4.3%, 7.0%, and 18.5%, respectively, among ARF patients compared with 2.7%, 3.8%, and 10.5% for patients without ARF (P=0.003).
The major findings of this study are that although the overall incidence of ARF after PCI is low (3.3%), patients with chronic renal insufficiency are at high risk of ARF, and ARF portends a very high risk of death, both in the hospital and in the long term. These data establish the prognostic importance of ARF after PCI.
Comparison With Previous Work
Although much smaller than the present study, a recently published report emphasized the adverse prognostic importance of ARF after PCI in patients with baseline Cr elevation.12 A remarkable 46% of 161 patients who had an increase in serum Cr of ≥25% died by 1 year compared with 19.4% of those who did not. Independent correlates of mortality included Cr elevation, age, and PCI of a saphenous vein graft lesion. Our data are consistent with previous outcome studies documenting high rates of cardiac death and myocardial infarction among patients with chronic renal failure on hemodialysis.13 Cardiac causes are by far the major causes of mortality among such patients. In addition, patients with chronic renal failure who undergo PCI have relatively poor outcomes, as do those who have CABG surgery.2
Mechanisms of ARF could not be established by the present study design, but these likely include the administration of iodinated contrast imaging agents. The pathobiology of contrast-induced renal insufficiency is complex and may be related to adenosine metabolism, perturbations of glomerular flow, endothelin and prostaglandin metabolism, and oxidative stress.8–10,14–18⇓⇓⇓⇓⇓⇓⇓ The incidence of ARF reported in previously published studies has varied widely among patients with normal renal function at baseline, diabetes mellitus, and chronic renal insufficiency but in general was lower than in the present study.3–11⇓⇓⇓⇓⇓⇓⇓⇓ Most of these studies were performed among stable patients who had diagnostic imaging procedures, whereas the present study addressed the incidence of ARF among patients who required high dye loads, including those who were critically ill or in shock.
Our data cannot establish whether the development of ARF is a marker of multisystem failure in critically ill patients or directly contributes to mortality, or both. The observed associations of ARF with recent myocardial infarction, shock, and a history of congestive heart failure suggest that hemodynamic deterioration plays a central role in the pathogenesis of ARF and in the prognosis of patients in whom the condition develops. To favorably influence prognosis, therefore, therapeutic maneuvers that are directed both at achieving hemodynamic stability and at preventing multiorgan failure are likely needed.
Although no patient subsets were excluded and a 4-year consecutive series was analyzed, ours is a post hoc analysis of an existing database. Because postprocedure serum Cr checks were ordered only when clinically indicated, it is possible that some patients were missed who had ARF but did not manifest an increase in Cr concentration until day 2 or 3 after the procedure, although this is unusual.19 Such misclassification, if important, would decrease the power to detect a difference between groups. The fact that clinical outcomes were so disparately different between groups suggests that any such misclassification was minor. The type of ARF (ischemic, nephrotoxic, or atheroembolic) could not be determined precisely. Renal biopsies were not performed. Because of the nature of the patients and the procedures being performed, ARF was assumed to be multifactorial.
Because of methodological limitations inherent in retrospective registry analyses, our data cannot definitely establish an etiologic link between worsening renal function after PCI and the observed increased risk of death and myocardial infarction. Our data cannot address whether prevention of such seemingly modest increases in Cr concentration can influence outcome after PCI. Only a prospective randomized study with large numbers of patients that demonstrates an effective reduction in the risk of renal failure, along with reduction in long-term risk of death, can answer that question.
We conclude that although the overall incidence of ARF after PCI is low, (1) patients with baseline elevation of serum Cr concentration are at high risk of ARF after PCI, (2) patients who experience ARF after PCI are at very high risk of in-hospital death, and (3) patients with worsening renal function are at very high risk of death or myocardial infarction in follow-up.
- ↵Bakris GL, Lass N, Gaber AO, et al. Radiocontrast medium-induced declines in renal function: a role for oxygen free radicals. Am J Physiol. 1990; 258: F115–F120.
- ↵Gross P, Bussemaker E. Endothelin: what role in acute contrast nephropathy? Nephrol Dial Transplant. 1996; 11: 1716–1718.
- ↵Heyman SN, Clark BA, Kaiser N, et al. Radiocontrast agents induce endothelin release in vivo and in vitro. J Am Soc Nephrol. 1992; 3: 58–65.