Author + information
- Received May 29, 1998
- Revision received December 1, 1998
- Accepted January 11, 1999
- Published online April 1, 1999.
- Steven P Marso, MDa,* (, )
- Stephen G Ellis, MD, FACCa,
- E.Murat Tuzcu, MD, FACCa,
- Patrick L Whitlow, MD, FACCa,
- Irving Franco, MDa,
- Russell E Raymond, DO, FACCa and
- Eric J Topol, MD, FACCa
- ↵*Reprint requests and correspondence: Dr. Steven P. Marso, Department of Cardiology/F25, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195
The aims of this study were to compare mortality and clinical events following percutaneous coronary intervention (PCI) between nondiabetics and diabetics with and without proteinuria.
Diabetics have increased rates of late myocardial infarction, repeat revascularization and mortality when compared with nondiabetics following PCI. Proteinuria is a marker for diabetic nephropathy and potentially a surrogate marker for advanced atherosclerosis. It is unknown if proteinuria is a predictor of outcome in diabetics following PCI.
We performed an observational study of 2,784 patients who underwent PCI at the Cleveland Clinic between January 1993 and December 1995. There were 2,247 nondiabetics and 537 diabetics with urinalysis and follow-up data available (proteinuria n = 217, nonproteinuria n = 320). The diabetic proteinuria group was further prospectively stratified into low concentration (n = 182) and high concentration (n = 35). The end points were all-cause mortality and the composite end point of death, nonfatal myocardial infarction (MI) and need for revascularization.
The mean follow-up time was 20.2 months. The two-year mortality rate was 7.3% and 13.5% for nondiabetics and diabetics, respectively (p < 0.001). The two-year mortality rate was 9.1% and 20.3% for the nonproteinuria and proteinuria groups, respectively (p < 0.001). There was a graded increase in mortality comparing the diabetic group. The two-year mortality rate was 9.1%, 16.2% and 43.1% for the nonproteinuria, low concentration and high concentration groups, respectively (p < 0.001). The difference in survival between the nondiabetic and nonproteinuric diabetics was not significant (p = 0.8).
The presence of proteinuria is the key determinant of risk following PCI for diabetics. Diabetics without evidence of proteinuria have similar survival compared with nondiabetics.
Diabetes affects more than 100 million patients worldwide (1). Whether insulin or noninsulin dependent, diabetes is a critical independent risk factor for atherosclerotic coronary artery disease (2,3). In fact, over 80 percent of all deaths in diabetic patients are due to complications from atherosclerosis (4). Given the high prevalence of coronary artery disease in diabetics, the need for coronary artery revascularization is frequent. Although the initial success and complication rates among diabetic and nondiabetic patients following percutaneous coronary revascularization are similar, diabetics have increased rates of late myocardial infarction, repeat revascularization and mortality when compared with nondiabetics (5–7). Diabetes is the most important risk factor for angiographic and clinical restenosis following percutaneous coronary revascularization. Overall, the odds ratio for restenosis among diabetics is 1.35 relative to their nondiabetic counterparts.
Proteinuria is an important marker for diabetic nephropathy and potentially a surrogate marker for advanced atherosclerosis. The aims of this study were to compare mortality and clinical event rates following percutaneous coronary intervention (PCI) between nondiabetics and diabetics with and without proteinuria. Therefore, we performed an analysis of all patients who underwent percutaneous coronary revascularization at the Cleveland Clinic from January 1993 through December 1995.
We identified 2,784 patients who underwent percutaneous coronary revascularization at the Cleveland Clinic Foundation and were flagged for long-term follow-up. Only the first 1,000 patients undergoing PCI in the calendar year are designated for follow-up in the interventional registry; of this group 2,247 were nondiabetic. There were originally 714 diabetic patients who had both short and long-term follow-up data available. We were able to obtain urinalysis information on 537 patients. This group of 2,784 patients (2,247 nondiabetics and 537 diabetics) was the focus of our study.
Patients undergoing percutaneous coronary revascularization at the Cleveland Clinic have been entered prospectively into an interventional registry (8). There are dedicated personnel that enter information on baseline demographic, clinical and angiographic characteristics into this registry. Initial follow-up information is obtained by telephone contact with patients at 30 days. Thereafter, patients are contacted at 12-month intervals for up to five years following the initial procedure. Follow-up data include information concerning need for repeat revascularization (either PCI or coronary artery bypass grafting [CABG]), nonfatal myocardial infarction (MI) and mortality. These data are collected and recorded by personnel dedicated solely to the interventional database.
The decision regarding whether patients underwent PCI, CABG or medical treatment was made by the attending physician at the time of angiography. Of patients referred for PCI, the majority underwent balloon angioplasty, although coronary artery stenting, directional coronary atherectomy, excimer laser, extraction and rotational atherectomy were also utilized at the discretion of the operator. Patients only underwent follow-up angiography if clinically indicated as determined by their attending physician or if required by enrollment in a randomized controlled trial.
The baseline clinical characteristics of the study groups were obtained from the Cleveland Clinic Interventional Registry. Acute MI was defined as myocardial infarction within 24 h and recent MI was within 1 to 14 days of PCI. Angina was defined according to the Canadian Cardiovascular Society. Patients with diabetes were identified from the registry if they were coded as having either insulin or noninsulin requiring diabetes. The noninsulin group could have been treated with either oral hypoglycemic agents or by diet alone. Patients were coded as having hypertension if they were diagnosed prior to PCI.
One or two observers using a caliper technique determined the severity of coronary stenosis (8). Significant stenosis was defined as greater than 50% stenosis in a major epicardial artery. The calcification score was defined by subjective visual assessment. Zero indicated that there was no visible calcification, 1 = faintly visible, 2 = moderate and 3 = severe. Chronic total occlusion was defined as a vessel known to be occluded for ≥3 months. Ostial stenosis was any lesion within 2 mm from the ostium. The lesions were also characterized by the modified (American College of Cardiology/American Heart Association) ACC/AHA classification scores. Procedure success required all treated sites to have a less than 50% residual stenosis and >TIMI (thrombolysis in myocardial infarction) grade two flow. Complete revascularization required vessels greater than 1.5 mm with greater than a 50% stenosis to be successfully treated to less than a 50% residual stenosis and a total decrease in the stenosis of greater than 20%. Left ventricular ejection fraction (LVEF) was determined by left ventriculography when available. Echocardiography was used if patients did not have a left ventriculogram.
This was an observational study. All the baseline clinical, angiographic and end point data elements were collected prospectively by personnel dedicated to the interventional registry. As urinalysis data were not recorded in the interventional registry, proteinuria was obtained retrospectively from a laboratory computer system by an investigator blinded to the outcomes of the study patients.
The presence of proteinuria was determined by a random urinalysis during the evaluation of this group of patients at the Cleveland Clinic. It was predetermined that the urinalysis needed to be within a time frame of six months prior to and three months postpercutaneous coronary intervention. Urinalyses done within 48 h after the intervention were not used in data analysis, owing to a higher false positive rate in patients given intravenous contrast. Further, the color indicator approximates a trace reading when the specific gravity is elevated. Such is the case when patients have had nothing by mouth for 12 to 24 h prior to angiography. If more than one urinalysis was available, the highest grade for proteinuria was used. Bayer’s Multistix 10 SG reagent strips (Bayer Corporation, Elkhart, Illinois) were used for the urinalysis. The test for proteinuria is based on the protein-error-of-indicators principle. The color indicator ranges from yellow (negative for protein) to yellow green to green to blue for positive reactions. Each color was semiquantified by laboratory personnel as trace, 1+, 2+, 3+ or greater than 3+, without knowledge of the clinical outcome.
Proteinuria in diabetic patients has been traditionally divided into micro- and macroalbuminuria. Microalbuminuria is defined as a 24 h protein excretion of 30 to 300 mg/24 h.
Macroalbuminuria is defined as greater than 300 mg/24 h. In this study, we used routine urinalysis chemstix that only identified those patients with macroalbuminuria. These chemstix generally only identify spot urine albumin to creatinine ratios greater than 300 μg albumin per mg creatinine which fall into the macroalbuminura range on timed urinalysis.
The predetermined end points of this study were all cause mortality and the composite end point of death, nonfatal myocardial infarction and need for repeat revascularization, either PCI or CABG. It was decided apriori to analyze the end points in two ways with respect to proteinuria. In the first analysis, the proteinuria was stratified into two groups: proteinuria and nonproteinuria. In the second analysis, proteinuria was stratified into three groups: nonproteinuria, low concentration and a high concentration. The low concentration group had a urinary protein concentration of trace, 1+ or 2+. The high concentration group had a urinary protein concentration of ≥3+. Comparison groups included nondiabetics, diabetics and diabetics stratified by proteinuria.
Data are listed as mean ±SD or median and interquartile range if nonnormally distributed. Statistical analysis was conducted using SYSTAT version 7.0.1 (SPSS Inc., Evanston, Illinois). The Student ttest was used for continuous variables and the chi-square test for categorical variables. The end points were analyzed using Cox proportional hazards estimation and presented as Kaplan-Meier freedom from event survival curves. A multivariable Cox proportional hazards model was developed for the end point of death with backward entry of the following covariates: acute myocardial infarction, insulin requiring diabetes mellitus, serum creatinine, LVEF, number of diseased vessels, treatment score and the presence of proteinuria. A treatment score was developed based on whether the following medications were given to diabetic patients: beta-adrenergic blocking agents, angiotensin-converting enzyme inhibitors (ACE), aspirin and 3-hydroxy-3-methylglutaryl-coenzyme A (HMG Co-A) reductase inhibitors. Each patient was assigned a number from 0–4 based on the number of medications they were given at the time of the procedure. Each treatment and the treatment score was analyzed by univariate Cox analysis for the end point of death. The treatment score along with the above mentioned covariates were then analyzed by backward entry in a Cox multivariate model.
The baseline clinical characteristics of the study groups are shown in Tables 1 and 2. ⇓⇓Diabetic patients were older, more often female, hypertensive, had higher NYHA classification and Canadian Classification for unstable angina. They were more likely to have had prior bypass surgery, greater number of diseased vessels and lower ejection fractions. Diabetics were less likely to be current and more likely to be past smokers. The baseline characteristics of diabetic patients based on proteinuria status are shown in Table 2. The proteinuria group was more likely to be hypertensive, insulin dependent, have a higher serum creatinine level and present with acute MI than the nonproteinuria group. There were only 91 gycosylated hemoglobins available for analysis among the study groups. There were 46 for the nonproteinuria and 45 for the proteinuria group. The mean levels were 8.3 ± 2.1 mg/dl and 8.1 ± 2.4 mg/dl, respectively (p = 0.7). The angiographic and lesion specific characteristics of the two groups are shown in Table 3. The median left ventricular ejection fraction was lower in the proteinuria compared with the nonproteinuria group (50% vs. 55%, p = 0.03), respectively. The proteinuria group was also more likely to have a greater number of diseased vessels than the nonproteinuria group, 2.3 versus 2.1, respectively (p = 0.008).
The procedural and in-hospital outcomes are listed in Table 4. The proteinuria group had somewhat higher rates of in-hospital death and MI and lower rates of procedural success and complete revascularization than the nonproteinuria and nondiabetic groups.
The probability of event-free survival for the study groups is shown in Figures 1a–d. The mean follow-up time was 20.2 months. The two year mortality rate was 7.3% for nondiabetics and 13.5% for diabetic patients (p < 0.001) (Fig. 1a). The two-year mortality rate was 9.1% and 20.3% for the nonproteinuria and proteinuria groups (p < 0.001) (Fig. 1b), respectively. There was a graded increase in mortality comparing the diabetic group stratified by proteinuria status shown in Figure 1c. The two-year mortality rate was 9.1%, 16.2% and 43.1% for the nonproteinuria, low concentration and high concentration groups, respectively (p < 0.001). The differences between each of these survival curves are significant (nonproteinuria vs. low concentration, p = 0.009; nonproteinuria vs. high concentration, p < 0.001; low concentration vs. high concentration, p = 0.001). Diabetics without proteinuria had similar two year mortality rates compared with nondiabetics (9.1% vs. 7.3%, p = 0.8) as shown in Figure 1d.
The composite event rate (death, MI or revascularization) at two years for the nonproteinuria and proteinuria groups were 25% versus 32%, respectively (p = 0.009). This difference was driven by the increased mortality in the proteinuria group. The rates of revascularization and myocardial infarction were similar among the two groups. The revascularization rate after two years was 15.8% and 11% for the nonproteinuria and proteinuria groups, respectively (p = 0.55). Similarly, the rate of MI among the proteinuria and nonproteinuria groups were nearly identical (3.3% vs. 2.5%, p = 0.7), respectively.
The two year composite event rates for the nonproteinuria, low concentration and high concentration groups were 25%, 29% and 50%, respectively (p = 0.002). The revascularization rates among the three groups were 15.8%, 12.4% and 3.4% (p = 0.3). The rate of nonfatal MI in the high concentration group was significantly higher than that of the other two groups (12% vs. 2%, p = 0.03).
The causes of death within the proteinuria and nonproteinuria groups are shown in Table 5. The majority of deaths in both groups were cardiac. There were 34 (16%) cardiac deaths in the proteinuria group and 18 (6%) cardiac deaths in the nonproteinuria group. There were 4 (2%) noncardiac deaths in the proteinuria group and 3 (1%) noncardiac deaths in the nonproteinuria group. The cause of death could not be determined for seven patients—three in the proteinuria group and four in the nonproteinuria group.
The baseline pharmacologic treatments are listed in Table 2. There was a trend for increased use of aspirin, beta-blockers and HMG CoA reductase inhibitors in the nonproteinuria group. There was a greater usage of ACE inhibitors in the proteinuria group. Given that these drugs have all been shown to be beneficial in patients with known atherosclerotic coronary disease and the trend for increased usage of three out of four agents were in the nonproteinuria group, we sought to determine if there was any significant treatment interaction with the end point of death for these agents. Therefore, each of these drugs was analyzed using univariate Cox analysis for the end point of death. This is shown in Table 6. All drugs, with the exception of ACE inhibitors, showed a nonsignificant trend for improved survival in those patients receiving therapy. For ACE inhibitors, there was a nonsignificant trend for an increased hazard for death in those patients receiving therapy. In order to determine the importance of combined pharmacologic therapy, a treatment score based on the number of drugs each patient was receiving at the time of intervention was developed. Fifteen diabetics received none of the 4 drugs, 276 received only 1 drug, 326 received 2 drugs, 118 received 3 drugs and 6 received all 4 drugs. The treatment score was significantly associated with improved survival using univariate Cox analysis. However, as shown in Table 7B, it was not independently associated with improved outcomes when forced into the multivariate model. Table 7, A–B shows the results of the univariate and multivariate Cox analysis for the end point of death for diabetic patients. Univariate predictors of outcome are depicted in Table 6. In this multivariate model, only proteinuria and LVEF were significant predictors of death shown in Table 7b.
Similar to other reports, our findings show that diabetic patients have decreased survival following percutaneous revascularization compared with nondiabetics (5,6,9). However, for the first time these data suggest that the increased mortality is limited to those diabetics with evidence of proteinuria on screening urinalysis. Furthermore, there is a graded response such that patients with a high concentration of proteinuria have a nearly four-fold increase in mortality compared with nonproteinuric diabetics. This group also had a marked increase in nonfatal MI rate when compared with other diabetics. The association of increasing mortality with increased urinary protein concentration further strengthens the association of proteinuria and increased risk of death following percutaneous coronary revascularization.
Comparison with prior studies
Proteinuria in diabetic patients indicates underlying nephropathy and increased risk of developing renal failure (10). Attention has been drawn to the association of proteinuria and cardiac mortality. Patients with non-insulin dependent diabetes mellitus (NIDDM) and proteinuria at baseline in the recent World Health Organization (WHO) Multinational Study of Vascular Disease in Diabetes had an increased risk of cardiovascular death three to four times that of diabetics without proteinuria (11). In 1984, Jarret et al. and Mogensen et al. reported an excess of all cause mortality in patients with NIDDM and microalbuminuria (12,13). Since then, there have been a number of longitudinal studies linking microalbuminuria and increased mortality in maturity onset diabetes (12,14–24). However, only six cross-sectional studies reported cardiac endpoints (16–18,20,22,23). The cardiac endpoints were determined clinically by a history of angina, MI or an abnormal electrocardiogram. Prior to our study, there has been a paucity of data concerning clinical outcomes of diabetics with proteinuria coupled with angiographic data.
A great deal of attention was drawn to the strategy of percutaneous coronary artery revascularization in diabetics with multivessel disease after the findings of the BARI study (7). This trial was a multicenter design randomizing over 1,800 patients with multivessel disease to an initial revascularization strategy of CABG or PTCA. Among treated diabetics, the five year survival for the CABG group was 80.6% and 65.5% for the PTCA group (p = 0.003). Following this trial, the National Heart Lung and Blood Institute issued a clinical alert concluding that CABG should be the preferred treatment for patients with treated diabetes and multivessel disease requiring a first time revascularization procedure (25). Given the findings of BARI, identifying clinical predictors of poor outcome in diabetics is fundamental to applying these findings to clinical practice.
Proteinuria and risk assessment
These data would suggest that screening diabetic patients for proteinuria by urinalysis prior to percutaneous coronary revascularization appears to be an effective way to risk-stratify this subset of patients. The BARI Study showed that insulin treated diabetics were at increased risk compared with noninsulin treated patients. Our data would suggest that diabetics with evidence of proteinuria, rather than insulin treated patients, are those at highest risk for future fatal cardiac events. Importantly, the risk of fatal events in diabetics without proteinuria is similar to nondiabetics.
We were unable to analyze proteinuria with a more sensitive assay. We identified only those patients with macroalbuminuria with the traditional chemstix utilized in this study. As the sensitivity of detecting proteinuria falls with decreasing urinary protein excretion, newer assays have been developed to detect lower concentrations of proteinuria-microalbuminuria. The detection of microalbuminuria with higher sensitivity may identify patients at increased risk after percutaneous coronary revascularization. Proteinuria was determined during a period of six months prior to three months post PCI. Therefore, nonproteinuria patients who developed proteinuria in our follow-up period were not reclassified. As the quantity of proteinuria varies throughout the day, timed urinalysis may be a more accurate way to identify high risk diabetics undergoing revascularization. Further, a limitation of routine urinalysis is that urine volume effects the result, i.e., a more concentrated urine sample has a higher concentration of protein compared with a dilute sample.
Objectively characterizing the severity and duration of diabetes is often difficult given the insidious onset of NIDDM (26). Thus, we are left with surrogate markers of disease severity, such as markers of glycemic control, insulin dependence and complications of diabetes. There were only 91 glycosylated hemoglobins available for analysis among the study patients. Although glycosylated hemoglobin was not a predictor of outcome by univariate Cox analysis, the lack of complete data prohibit us from making meaningful conclusions regarding glycemic control prior to percutaneous coronary revascularization in our study. In this cohort of patients, proteinuria appears to objectively identify those patients with disease severity such that proteinuric diabetics are at increased risk for adverse outcome compared with their nonproteinuric counterparts.
Only 23% of patients were receiving therapy with ACE inhibitors in our study. Therapy with these agents in this study was actually associated with increased risk of death. This is likely secondary to selection bias. Diabetics in this study receiving ACE inhibitors had significantly lower ejection fractions and a high incidence of clinical heart failure. Although ACE inhibitors have been shown to delay the progression of proteinuria in diabetics independent of the antihypertensive effects, they have not been shown to abolish proteinuria or improve the cardiovascular morbidity and mortality in diabetic patients with proteinuria (27–32). Data from Lewis et al. demonstrated captopril therapy reduced the risk of doubling the serum creatinine (12% in the captopril group and 21% in the placebo group, p = 0.007) and there was a 50% reduction in the combined endpoints of death, dialysis and renal transplantation in the ACE inhibitor group (31). These data were not published until November, 1993 which likely accounts for the low percentage of patients on ACE inhibitor therapy in our study. Currently, the DIAB-HYCAR study is testing the hypothesis that low dose ramipril will reduce the cardiovascular morbidity and mortality in NIDDM patients with microalbuminuria or proteinuria (33). Given that ACE inhibitor therapy slows the progression rather than eradicating proteinuria, determination of proteinuria, even in patients receiving ACE inhibitor therapy, is likely to remain an effective screening tool in identifying high risk diabetic patients undergoing PCI.
These data support the theory that proteinuria is a surrogate marker for advanced atherosclerosis. Diabetic patients with proteinuria had smaller arteries, a greater number of diseased vessels and longer lesion length as evaluated by angiography when compared with the nonproteinuria group. Our data suggest that the increased mortality seen in diabetic patients is limited to those with evidence of proteinuria and that the survival rate for nonproteinuric diabetics is similar to nondiabetics following PCI. Proteinuria, the determination of which is simple, effective and inexpensive, appears to be an important prognostic marker for diabetics.
- angiotensin converting enzyme
- coronary artery bypass grafting
- left ventricular ejection fraction
- myocardial infarction
- percutaneous coronary intervention
- Received May 29, 1998.
- Revision received December 1, 1998.
- Accepted January 11, 1999.
- American College of Cardiology
- Harris M.I,
- Hadden W.C,
- Knowler W.C,
- Bennet P.H
- Schwartz C.J,
- Sprague E.A,
- Kelley J.L,
- Cayatte A.J,
- Rozek M.M
- Stamler J,
- Vaccaro O,
- Neaton J.D,
- Wentworth D
- Wingard D.L,
- Barrett-Conner E
- Stein B,
- Weintraub W.S,
- Gebhart S.P,
- et al.
- Kip K.E,
- Faxon D.P,
- Detre K.M,
- Yeh W,
- Kelsey S.F,
- Currier J.W
- Abdel-Meguid A.E,
- Whitlow P.L,
- Simpfendorfer C,
- et al.
- (1997) Influence of diabetes on 5-year mortality and morbidity in a randomized trial comparing CABG and PTCA in patients with multivessel disease: the Bypass Angioplasty Revascularization Investigation (BARI). Circulation 96:1761–1769, [see comments].
- Mogensen C.E,
- Damsgaard E.M,
- Froland A,
- Nielsen S,
- de Fine Olivarius N,
- Schmitz A
- Mattock M.B,
- Morrish N.J,
- Viberti G,
- Keen H,
- Fitzgerald A.P,
- Jackson G
- Neil A,
- Hawkins M,
- Potok M,
- Thorogood M,
- Cohen D,
- Mann J
- ↵(1995) NHLI BARI clinical alert on diabetics treated with angioplasty. Circulation 92.