Author + information
- Received August 7, 1998
- Revision received March 2, 1999
- Accepted April 14, 1999
- Published online August 1, 1999.
- Lynn Palacol Roppolo, MDa,
- Robert Fitzgerald, PhDa,
- Jennifer Dillow, BSa,
- Thomas Ziegler, MDa,
- Mitchell Rice, RNa and
- Alan Maisel, MD, FACCa,* ()
- ↵*Reprint requests and correspondence: Dr. Alan Maisel, VAMC, Cardiology 111-A, 3350 La Jolla Village Drive, San Diego, California 92161
The purpose of this study was to prospectively evaluate the usefulness of the cardiac troponins as predictors of subsequent cardiac events in patients with chronic renal failure undergoing dialysis.
Cardiac troponin T (cTnT) and I (cTnI) are cardiac markers that are specific for cardiac muscle. They are also excellent prognostic indicators for patients presenting with chest pain. Although cardiac disease is the leading cause of death in dialysis patients, standard methods to diagnose acute coronary syndromes in patients with renal failure are often misleading.
A six-month prospective study was done in a university-affiliated Veterans Hospital’s dialysis clinic. Forty-nine patients undergoing chronic dialysis with no complaints of chest pain were followed for cardiac events occurring in the six months after cardiac troponin measurements. These included unstable angina, acute myocardial infarction and cardiac death. An additional 83 patients with renal failure but who were not undergoing dialysis were also examined.
Within six months all three dialysis patients with elevated cTnI at entry into the study suffered an adverse complication (specificity and positive predictive value = 100%). Twenty-five patients had cTnT elevated at >0.10 ng/ml (53%). Patients with diabetes were more likely to have elevated troponin T levels (64% vs. 25%, p = 0.01). All six patients developing cardiac events within three months had elevations of cTnT >0.1 ng/ml (sensitivity = 100%). Whereas the specificity of cTnT was only 56% for a near-term cardiac event, the negative predictive value of cTnT using a cutoff of ≤0.1 ng/ml was 100%. On restratifying patients using a cutoff value of cTnT of >0.2 ng/ml, only nine of 49 dialysis patients (18%) had elevated levels. In patients with renal failure not undergoing dialysis, only three of 83 (4%) had elevated troponin I or T. None of these patients suffered a cardiac event in the next six months.
This prospective pilot study clearly delineates the troponins as important prognosticators in asymptomatic otherwise “stable” patients on chronic dialysis, especially those with concomitant diabetes mellitus. It also appears that troponins are more likely to be elevated in dialysis patients than other patients with renal failure not on dialysis. The above suggests that the combination of cTnI and cTnT might be very effective in elucidating cardiac risks of patients with renal failure undergoing chronic dialysis.
Cardiac troponin T (cTnT) and troponin I (cTnI) are structural proteins that act to regulate muscle contraction (1,2). They are released into the bloodstream from injured muscle cells during cardiac ischemia with no overlap with skeletal muscle troponins under normal conditions (3,4). Multiple studies have demonstrated that both cTnI and cTnT are important prognostic indicators in patients presenting with chest pain, even when creatine kinase (CK), MB fraction is not elevated (5–7). As such, the troponins are gaining wider acceptance as a tool to stratify patients with chest pain. Thus, they might also be useful prognosticators in high risk patients in whom standard evaluation of myocardial ischemia is not always accurate.
Patients with chronic renal failure are at risk for the accelerated development of coronary artery atherosclerosis and its consequences such as acute myocardial infarction (8). Cardiac disease is the leading cause of death in dialysis patients (9). Diabetes, hypertension and left ventricular hypertrophy are recognized cardiac risk factors commonly found in this patient population (10). The conventional tools used to diagnose acute coronary syndromes in patients with renal failure, which include a 12-lead electrocardiogram (ECG) and serial measurements of CK isoenzyme (CK-MB), are often misleading (11,12).
The purpose of the present pilot study was to prospectively evaluate the usefulness of the cardiac troponins as predictors of subsequent cardiac events in our patients with chronic renal failure undergoing dialysis. We also included a group of patients with chronic renal failure not on dialysis. Because cTnI has been found to have greater specificity for myocardial injury in chronic renal failure patients than cTnT (13–19), we hypothesized that cTnI might also be of greater prognostic value than cTnT.
The study was approved by the Human Subjects Committee at the University of California, San Diego. All patients receiving dialysis at the VA Medical Center in La Jolla were asked to participate in this study (N = 51). Informed consent was obtained on 48 hemodialysis patients and one patient receiving peritoneal dialysis (94% of dialysis patients). All blood samples were drawn before starting hemodialysis on their routinely scheduled day within a one-week time frame. All patients denied having chest pain at the time their blood was drawn. Demographic data, medical history including cardiac risk factors and laboratory data, were obtained by chart review, hospital information systems and patient interviews. Hemodialysis was performed with a Hemoflow F-Series (High Flux) Capillary dialyzer (Fresenius, Walnut Creek, California) in all patients.
In addition, 83 sequential patients with chronic renal failure not on dialysis but followed in the renal failure clinic were also examined over a three-month period.
Follow-up period to assess cardiac events was designated at six months. Cardiac events included: 1) admission for unstable angina, classified as type IIIB in the Braunwald classification (20); 2) acute myocardial infarction, diagnosed by ECG changes, wall motion abnormality by multigated angiogram; echocardiography, angiography or autopsy; and 3) sudden death, presumed cardiac in origin.
Venous blood was collected immediately before dialysis and taken immediately to the blood gas lab and centrifuged for 120 s at 11,000 rpm. An aliquot of plasma was analyzed in the gas lab for total CK, CK-MB, and cTnI. The remaining plasma was frozen at −20°C and stored for three to five weeks. The samples were then analyzed for cTnT by batch method in the main laboratory. Cardiac cTnI, total CK and CK-MB were measured with a fluorogenic enzyme-linked immunoassay designed by Behring Diagnostics on the OPUS analyzer. The OPUS cTnI assay uses purified polyclonal antibodies that are able to recognize polypeptide segments unique to the cardiac isoform of troponin I. We selected <0.5 ng/ml as the cutoff for normal cTnI based on our experience with over 600 normal specimens. The OPUS CK-MB assay utilizes anti–CK-MB monoclonal antibodies. The OPUS total CK assay couples its enzymatic activity to hexokinase and glucose-6-phosphate dehydrogenase. The nicotinamide adenine dinucleotide phosphate (reduced form) production measured by the OPUS analyzer is proportional to the amount of total CK in the sample. The normal expected value for CK-MB is <8.99 ng/ml. The normal range for total CK in human serum is 38 to 174 U/liter for men and 26 to 140 U/liter for women. The CK index was calculated by dividing the CK-MB by the total CK and multiplying by 100. Values >4.0% suggest cardiac damage.
The Elecsys Troponin T STAT CARDIAC T test was the immunoassay used for the quantitative determination of cTnT. This assay was designed by Boehringer Mannheim and uses monoclonal antibodies specifically directed against human cTnT and uses the cardiac-specific M11-7 antibody. The recommended clinical threshold value of cTnT using this assay is 0.1 ng/ml. This has a specificity of greater than 99.5%.
InStat and Excel statistical software was used for data analysis. Unpaired ttests were performed on all numeric data comparing patient variables as they related to levels of cardiac troponin T. Fisher exact test was performed for frequency comparison (i.e., cardiac risk factors) as well as event outcomes using the cardiac markers. Significant p values were <0.05.
Table 1shows the baseline characteristics of the 49 dialysis patients. Within a three-month follow-up period, six patients (12%) suffered significant cardiac events (Table 2). These included four documented myocardial infarctions, two of which were fatal, one episode of bradycardia and cardiac arrest culminating in death and presumed to be myocardial infarction and one episode of unstable angina necessitating urgent coronary angioplasty. Only one of these patients had prior evidence of left ventricular dysfunction on echocardiogram. All six patients suffered their events within the first three months of follow-up.
Cardiac troponin I
All three patients with elevated cTnI at entry into the study suffered an adverse complication within three months (specificity and positive predictive value = 100%, p < 0.005) (Table 3). The three patients with elevated cTnI also had elevated cTnT levels (ranging from 0.25 ng/ml to 0.7 ng/ml). In addition, there were three other patients who suffered a complication in the two-month period in whom cTnI was negative (sensitivity = 50%, negative predictive value = 93%).
Cardiac troponin T >0.1 ng/ml
Twenty-five patients (53%) had cTnT elevated at >0.10 ng/ml (Tables 3 and 4). ⇓Patients with elevated cTnT tended to be slightly older and have lesser elevations of serum creatinine. There were no significant differences in levels of total CK, CK-MB and relative index in patients with normal versus elevated cTnT.
Table 5compares cardiac risk factors based on troponin T levels. Patients with a history of tobacco use, hyperlipidemia, hypertension, a positive family history for heart disease or a documented cardiac history showed no propensity toward elevated troponin T levels. Of note, patients with diabetes were more likely to have elevated troponin T levels (64% vs. 25%, p = 0.01).
All six patients developing cardiac events within two months had elevations of troponin T >0.1 ng/ml (sensitivity = 100%, p < 0.05) (Table 3). Two of the patients had previously normal ECGs within the previous year and four had wall motion abnormalities consistent with previous myocardial infarction. The only patient (Patient #2) in the study group receiving peritoneal dialysis (he had received hemodialysis intermittently over 13 years) died two weeks after his blood sample was obtained. He had a history of coronary artery disease and multiple cardiac risk factors including diabetes. His cTnT level was 0.766 ng/ml. His cTnI level was normal and his CK-MB was 2.3 ng/ml. His autopsy revealed an acute anterior wall myocardial infarction. Using a cutoff of 0.1 ng/ml cTnT was only 56% specific for a near-term cardiac event. However the negative predictive value of cTnT using the 0.1 ng/ml cutoff was 100%.
Cardiac troponin T >0.2 ng/ml
Because of the low specificity of cTnT at values >0.1 ng/ml, we restratified patients using a cutoff value of cTnT of >0.2 ng/ml (Tables 3 and 4). Nine of 49 dialysis patients (18%) had cTnT levels >0.2 ng/ml. Of the six patients suffering cardiac events, cTnT was greater than 0.2 ng/ml in five of them, improving the specificity to 91% and the positive predictive value to 56% (p < 0.005). There was only a slight effect on sensitivity (83%) and little effect on the negative predictive value (98%).
Creatine kinase, MB fraction
There were two patients in this study with elevated CK-MB. One of these patients had elevated cTnT levels >0.2 ng/ml. There were no untoward cardiac events in either patient.
Renal failure patients not undergoing dialysis
Table 6delineates the baseline characteristics of patients with chronic renal failure not yet undergoing dialysis. Only three patients in the entire group had elevation of cardiac troponin I (n = 2) or cardiac troponin T (n = 1) (4% vs. 51% of dialysis patients, p < 0.0001). In the six-month follow-up period, two patients suffered myocardial infarction, none of whom had elevated troponins.
Multiple studies have demonstrated that an elevated level of either cTnI or cTnT in an ischemic setting is associated with subsequent myocardial infarction and death, even when CK-MB is not elevated (5–7,21,22). Such studies suggest that troponins are useful in stratifying patients who come to the emergency room with chest pain. In fact, adverse outcomes in those with negative levels proved so rare that they suggested a patient may be discharged from an emergency room mainly on the basis of these markers (5). The reason why detection of small rises in troponins has such important prognostic implications in patients with chest pain probably involves its ability to detect minor amounts of focal myocardial necrosis and distal embolization seen during disruption of coronary plaques (23).
The importance of this pilot study is that cardiac troponins proved to be extremely effective predictors of cardiac events in a group of asymptomatic, but high risk patients undergoing chronic dialysis at our medical center. All six patients suffering near-term cardiac events had elevations of one or both of the cardiac troponins. Although other studies have detailed poor prognostic outcomes in renal failure patients who have elevated cardiac markers (15–18), this was the first prospective evaluation of a group of dialysis patients utilizing both cardiac troponin T and I. Conversely, very few patients with chronic renal failure not undergoing dialysis had elevation of troponins.
Studies have clearly demonstrated the clinical accuracy of cTnI in diagnosing an acute myocardial infarction (1,2). The reason cTnI has such high specificity for cardiac injury is because it is not found in skeletal muscle during neonatal development or adulthood (4,24–27). Furthermore, it is not expressed in regenerating human skeletal muscle tissue. Accordingly, elevations do not occur, even in patients with acute or chronic skeletal muscle damage, unless myocardial injury is present. The concentration of cTnI is generally not increased in renal failure in the absence of myocardial injury (1,15). Hence, it is likely that the three patients in our study who had elevated cTnI had subclinical myocardial necrosis eventually leading to cardiac events, two of them fatal. Whereas the positive predictive value of elevated cTnI in these patients was 100%, the sensitivity for detection of future events was only 50%.
Cardiac troponin T, on the other hand, identified every patient with a subsequent cardiac event that occurred in the dialysis patients (sensitivity of 100%). Clearly the negative predictive value of cTnT was this marker’s strong point. The drawback with cTnT in the present study was the large number of positive patients (levels greater than 0.1 ng/ml) who did not have any documented cardiac damage (specificity of 56%). Because higher concentrations of cTnT were more strongly associated with adverse cardiac events, we restratified patients to a cutoff of >0.2 ng/ml. This improved specificity to 91% without compromising sensitivity to a large degree (5/6 patients with adverse cardiac events at the 3-month follow-up point had elevations of cTnT above 0.2 ng/ml). However, the large number of cTnT positive patients precluded a major change in treatment strategy for those patients, although they did provide some additional biochemical data that could be considered in the context of the patients’ histories and physical findings.
We do not know if the apparent “false positives” observed with cTnT represented minor myocardial damage that could not be detected with other means or if they were caused by an analytic artifact. There have been several reports suggesting that cTnT is reexpressed as its cardiac isoenzyme (27–29); however, the antibodies used to detect cTnT in these studies were different than those used in the commercially available assay we tested. Another hypothesis is that the uremic state causes myocardial damage that is detected by cTnT. Haller et al. found no correlation between residual diuresis, serum creatinine, blood urea nitrogen and cTnT (16). Another study reported no significant differences in cTnT levels pre- and postdialysis (14).
Is it possible that minor myocardial injury undetected clinically might have been responsible for positive cTnT values?
Fifty-three percent of our dialysis patients had troponin T levels greater than 0.1 ng/ml. Although we doubt that every one of these patients would go on to have a cardiac event, it is nevertheless possible that cTnT might be more sensitive than cTnI in detecting small areas of necrosis or disruption of the coronary plaque (28,29)and thus be a better prognostic indicator of patients with acute coronary syndromes. Although the exact nature of release of troponins from the contractile apparatus of striated muscle has not been fully elucidated, both are probably released from the cytosolic pool into circulation after necrosis. The proteins differ in the proportion contained in the cytosolic pool, however, representing 6% to 8% of total cTnT but only about 2.5% of total cTnI (13). Thus, it might be possible that cTnI is less sensitive than cTnT or that cTnI may retain its cardiac-specific expression during uremia (16).
In our series, there was a preponderance of diabetes in patients whose troponin T was >0.1 ng/ml. Similar findings have been reported by Li et al., who also found no cTnT elevations in a separate group of diabetic patients with normal creatinine levels (17). Diabetes is a common cause of renal failure. Thirty-five to 45% of patients with insulin-dependent diabetes develop nephropathy (30). In this study, all six of the patients with significant cardiac outcomes had diabetes. It is well documented that patients with diabetes are at risk for coronary artery disease (31). The increased sensitivity of cTnT may detect microvascular changes that are not clinically evident, and in some cases, perhaps not detected with cTnI.
Conclusions and clinical implications
Although this prospective study included small numbers of patients and is therefore considered a pilot study, it did encompass nearly 95% of all patients undergoing chronic dialysis at a Veterans Hospital as well as over 50% of the hospital’s renal failure clinic. Examination of larger populations of dialysis patients is needed to substantiate our conclusion that cardiac troponins are important prognosticators in this cohort of asymptomatic otherwise “stable” patients. This may be especially true in patients with concomitant diabetes mellitus. A positive troponin I was virtually 100% specific and 100% predictive for a future cardiac event in our study. Yet its negative predictive value was only 50%. Cardiac troponin T, on the other hand, was 100% sensitive in detecting dialysis patients who would have future cardiac events, with a negative predictive value of 100%. Our data suggested that the combination of cTnI and cTnT might be effective in elucidating cardiac risks of patients with renal failure undergoing chronic dialysis. If the results of our pilot study were confirmed by other studies, all patients with elevations of cTnI should urgently be worked up with a functional cardiac study or coronary angiography, especially if an elective procedure is being planned. If only cTnT is being used, it is possible that its cutoff value should be increased to >0.2 ng/ml. This would greatly improve the specificity with only minimal compromise in sensitivity.
On the other hand, if the low incidence of cardiac marker positivity and their subsequent lack of predictive value in patients with renal failure not undergoing dialysis is substantiated, this might ease physicians’ minds about their value in this group of patients, as long as they are not presenting with ischemic symptoms. We believe it is possible that the higher frequency of marker elevation in renal failure patients once they begin dialysis was a consequence of the ongoing “inflammatory condition” of dialysis. This might possibly lead to subsequent plaque rupture and eventual cardiac complications. Further studies in this area are suggested.
We gratefully acknowledge the Dialysis nurses, the Blood Gas Lab staff and the Laboratory staff at the VA Medical Center in La Jolla, without which this work would not have been completed.
- creatine kinase
- cardiac troponin I
- cardiac troponin T
- Received August 7, 1998.
- Revision received March 2, 1999.
- Accepted April 14, 1999.
- American College of Cardiology
- Adams J.E,
- Bodor G.S,
- Davila-Roman V.G,
- et al.
- Adams J.E,
- Schechtman K.B,
- Landt Y,
- Ladenson J.H,
- Jaffe A.S
- Anderson P.A,
- Malouf N.N,
- Oakeley A.E,
- Pagani E.D,
- Allen P.D
- Galvani M,
- Ottani F,
- Ferrini D,
- et al.
- Rutsky E.A,
- Rostand S.G
- ↵U.S. Renal Data System, USRDS. Annual Data Report. Bethesda, MD: The National Institute of Health, National Institute of Diabetes and Digestive and Kidney Diseases, August 1990.
- Bhayana V,
- Gougoulias T,
- Cohoe S,
- Henderson A.R
- Frankel W.L,
- Herold D.A,
- Ziegler T.W,
- Fitzgerald R.L
- Hafner G,
- Thome-Kromer B,
- Schaube J,
- et al.
- Haller C,
- Stevanovich A,
- Katus H.A
- Wu A.H.B,
- Feng Y.J,
- Contois J.H,
- Pervaiz S
- Braunwald E
- Lindahl B,
- Venge P,
- Wallentin L
- Saggin L,
- Gorza L,
- Ausoni S,
- Schiaffino S
- Toyota N,
- Shimada Y
- Lindahl B,
- Toss H,
- Venge P,
- Wallentin L
- Christenson R.H,
- Duh S.H,
- Newby L.K