NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Michos ED, Berger Z, Yeh HC, et al. Cardiac Troponins Used as Diagnostic and Prognostic Tests in Patients With Kidney Disease [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2014 Aug. (Comparative Effectiveness Review, No. 135.)
Cardiac Troponins Used as Diagnostic and Prognostic Tests in Patients With Kidney Disease [Internet].
Show detailsSearch Results
After removing duplicate citations from our searches, we retrieved 6,809 unique citations (Figure 3). After reviewing titles, abstracts, and full articles, we included 126 studies (in 130 publications). We included 14 studies that evaluated the diagnostic accuracy of a troponin elevation in the diagnosis of acute coronary syndrome (ACS) in patients with chronic kidney disease (CKD) (Key Question [KQ] 1).37-50 We did not find any studies that directly assessed how troponin levels affect management strategies of ACS in patients with CKD (KQ 2). However, we discuss one study that reported troponin levels by management strategies in patients with CKD and symptoms of ACS.51 We found 12 studies in 14 publications that addressed short- and long-term prognosis in patients with CKD after presentation with ACS by troponin levels (KQ 3).40, 52-64 We included 98 studies (in 105 publications) that evaluated use of troponin levels for risk stratification among patients with CKD without ACS symptoms (KQ 4).9, 11, 25, 26, 47, 65-163 One study reported on both KQ 1 and KQ 3.40 One study reported on both KQ 3 and KQ 4.47
Figure 3
Summary of the literature search. * Total may exceed number in corresponding box, as articles could be excluded for more than one reason at this level. ** One study indirectly addressed this Key Question
KQ 1. Use of Troponin for Diagnosis of Acute Coronary Syndrome Among Chronic Kidney Disease Patients
Study Design Characteristics
We included 14 studies for this KQ. Of these, six used a prospective cohort design, five used a retrospective design, two used a cross-sectional design, and one used a prospective case-control design. All studies took place in the acute care setting, and all but two took place in the hospital setting. Of these two, one took place in a mixed setting, including the emergency department, intensive care unit, and internal medicine wards;40 and the setting for the second was unknown.42 Five studies took place in the United States,41, 43, 45-47 six in Europe,37, 40, 42, 48-50 two in Asia,38, 44 and one in the Middle East.39
Seven studies did not explicitly give dates of enrollment. For those seven studies which did report enrollment, start dates ranged from 1999 to 2009 and end dates ranged from 1999 to 2010.37-40, 43, 49, 50 Seven studies did not report mean length of followup. For those studies that did report length of followup, it ranged from 30 days to 2 years.37, 41-43, 47, 49
Of the 14 studies included for this KQ, different numbers of studies addressed various operating characteristics; some studies addressed more than one type of operating characteristic. Table 4, below, presents the number of unique studies addressing each type of operating characteristic, and the relevant KQ to which they apply.
Table 4
Number of unique studies addressing each type of operating characteristic.
Study Population Characteristics
The total number of patients enrolled ranged from 31 to 1,601. Five studies reported explicit adjudication of an acute coronary syndrome (ACS) diagnosis, all with panels; two included cardiologists;39, 43 and three did not include cardiologists.46, 48, 49 Table 5 summarizes the adjudication criteria the studies used.
Table 5
Adjudication criteria researchers used to define acute coronary syndromes in studies that evaluated the use of troponin to diagnosis acute coronary syndromes among patients with chronic kidney disease.
The studies included patients with various stages of CKD. Four studies included patients on dialysis.38, 42-44 One study included patients in stages 1-4 of CKD.40 Another study included patients in stages 1-5 of chronic kidney disease (CKD) but did not include patients on dialysis.48 No studies exclusively included patients in stage 1 or 2. Two studies included only patients in stage 3 or 4.37, 50
The mean age of those enrolled ranged from 48 to 80 years. Three studies did not provide this information.44-46 The percentage of men among those enrolled ranged from 35 to 76; three studies did not report gender distribution.44-46 Five of the studies reported distribution of race or ethnicity. The percentage of African American patients ranged from 48 to 86, and the percentage of White patients ranged from 12 to 65.41, 43, 47
Study Quality
The quality of the included studies varied. Three studies were of good quality.37-39 One study was of poor quality.40 The remainder of the studies were of fair quality.
KQ 1.1. Operating Characteristics of a Troponin Elevation (Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value)
Key Points
- In six studies, the sensitivity of the troponin T assay for ACS in patients with CKD ranged from 71 to 100 percent, and its specificity ranged from 31 to 86 percent. Three studies reported a positive predictive value (PPV) and negative predictive value (NPV) for troponin T for the diagnosis of ACS. The PPV for troponin T ranged from 66 to 77; the NPV ranged from 71 to 98. In one study, the assay was associated with a greater PPV and NPV for the subgroup of patients with age less than 65 years (strength of evidence: low).
- In eight studies, the sensitivity of the troponin I assay for ACS ranged from 43 to 94 percent, and its specificity ranged from 48 to 100 percent. In five studies, that reported PPV and NPV, PPV ranged from 74 to 100; the NPV ranged from 93 to 98 percent. The broad range of these findings can be attributed to heterogeneity in regard to study populations, definitions of ACS, assays used, and assay cutoffs used (strength of evidence: low).
- One study found that the magnitude of change in the troponin T assay did not differ between patients with ACS and a control group, during 24 hours after admission. The rate of change did differ but this rate displayed marked variability during the 24 hours. This was a single study with a small sample size and imprecise results, and thus not conclusive (strength of evidence: insufficient).
- One study, which included details of ACS adjudication, reported sensitivity and specificity for troponin I elevation which appeared roughly comparable to that of other studies, though direct comparison is impossible.
Results
Ten unique studies reported on the sensitivity or specificity of a troponin assay to diagnose ACS.37, 38, 40, 42, 44-47, 49, 50 Three studies reported explicit adjudication of an ACS diagnosis, all included panels, two including cardiologists,39, 43 and three did not include cardiologists.46, 48, 49 Two studies reported other diagnostic criteria of ACS; two used criteria from the European Society of Cardiology,37, 38 and two electrocardiogram and clinical criteria.41, 50 We were unable to conduct a meta-analysis because the number of studies was too small, and thus we do not have an aggregate estimate of the sensitivity and specificity. We presented the results for troponin T and troponin I separately below.
Troponin T
Six studies examined the operating characteristics of the troponin T assay in their entire study population (Table 6).38, 42, 44, 48-50 Two studies used a cutoff of 0.1 mcg/L, both used the Roche Elecsys assay.38, 42 One study used a cutoff of 0.16 mcg/L and did not specify the manufacture or assay.44 Another study used multiple cutoffs, including 0.014 mcg/L and 0.0358 mcg/L and the high-sensitivity Roche Elecsys assay.48 One study reported a cutoff of 0.009 mcg/L for the Roche Elecsys assay, and a cutoff of 0.0194 mcg/L for the high-sensitivity Roche Elecsys assay.49 One study used a cutoff of 0.014 mcg/L for the Roche Modular E170 assay.50 The sample size of those studies using the troponin T assay ranged from 31 to 382. The sensitivity in these studies ranged from 71 to 100 percent, and the specificity ranged from 31 to 86 percent (Figure 4). The heterogeneity of these results using the same cutoff and assay can potentially be understood in the light of the different geographic settings of the studies; moreover, while one study adjudicated ACS according to the standards of the European Society of Cardiology;38 and two others did so with cardiologist adjudication, according to the criteria of the Global Consensus on MI;48, 49 two other studies did not explicitly report adjudication standards.42, 44
Table 6
Operating characteristics of elevated troponin in the diagnosis of acute coronary syndrome among patients with chronic kidney disease.
Troponin I
Eight studies examined the operating characteristics of the troponin I assay in their entire study population (Table 6).37, 40, 42, 44-47, 49 The cutoff values they used for the diagnosis of ACS differed (with some studies evaluating multiple different cutoffs). One study used a cutoff of 0.11 mcg/L,37 one study used 0.4 mcg/L,45 two studies used 0.5 mcg/L,37, 40 one study used 0.6 mcg/L,46one study used 1.0 mcg/L42 and two studies used 0.8 mcg/L.44, 47 One study used two cutoffs, 0.0063 mcg/L and 0.0099 mcg/L.49 The sample size of these studies ranged from 31 to 1,601.
The troponin I assays in these studies were of a variety of types from a range of manufacturers. Three studies used an assay from the same manufacturer, Beckman.37, 40, 49 Other studies used the manufacturers Vidas, Biosite, Baxter, Dade, DPC, and Siemens. One study did not report a manufacturer.44
One study46 that reported details of ACS adjudication, showed values of sensitivity and specificity, which did not appear to differ markedly from those of the other studies using troponin I; however, we can make no conclusions due to the heterogeneity of cutpoints.
The sensitivity in these studies ranged from 43 to 94 percent, and the specificity ranged from 48 to 100 percent (Figure 5).
KQ 1.1.a. Positive and Negative Predictive Values
Results
Five studies estimated the positive and negative predictive values for troponin I in the assessment of ACS in their entire study population.37, 40, 46, 47, 49 They used multiple cutoffs. One used 0.11 mcg/L,37 two used 0.5 mcg/L,37, 40 one used 0.6 mcg/L,46 one used 0.8 mcg/L,47 and one used two cutoffs, 0.0063 mcg/L and 0.0099 mcg/L. For troponin I in the diagnosis of ACS, the PPV ranged from 7 to 100 percent; the NPV ranged from 93 to 98 percent. Given the heterogeneity of the cutoffs and manufacturers used in these studies, it was not possible to identify a trend relating the cutoff value to NPV or PPV. We were unable to conduct a meta-analysis because the studies were insufficient in number, and thus cannot provide an aggregate estimate of PPV or NPV.
One study estimated the NPV or PPV of troponin T for the diagnosis of ACS for two subgroups41 (Table 7); two studies did so for the entire study population.48, 49 The PPV for troponin T ranged from 6 to 77 percent; the NPV ranged from 71 to 98 percent. In one study, the assay was associated with a greater PPV and NPV for the subgroup of patients with age less than 65 years.
Table 7
Operating characteristics of elevated troponin in the diagnosis of acute coronary syndrome among patients with chronic kidney disease.
KQ 1.1.b. Change in Troponin Values Versus Single Troponin Elevation
Results
One study addressed this KQ, with a total sample size of 46.38 This study was performed in CKD patients in stages 3, 4, and 5, including nine patients on hemodialysis. The authors found that the magnitude of change in the troponin T assay in the first 24 hours after admission did not significantly differ between the control group and the group with ACS; neither did the rate of change from 0 to 6, or 6 to 12 hours after admission. While the rate of change from 0 to 24 hours after admission was greater in the group with ACS, there was great variability in this rate of change.
Strength of Evidence
The strength of evidence for the body of literature addressing KQ1.1, 1.1a, and 1.1b is explained in Tables 8 and 9.
Table 8
Elevated troponin T or I versus nonelevated troponin T or I in terms of diagnostic accuracy among patients with chronic kidney disease: Strength of evidence domains for KQ 1.1, 1.1a, and 1.1b.
Table 9
Elevated troponin T or I versus nonelevated troponin T or I in terms of diagnostic accuracy among patients with chronic kidney disease: Details regarding strength of evidence domains.
KQ 1.2. Operating Characteristics of a Troponin Elevation by Subgroups
Key Points
- Although a few studies have looked at how age and CKD stage affect the operating characteristics of troponin, they are small, of poor quality, and use different cutoffs for different categories. Therefore we were unable to draw any conclusions.
- There were no studies of troponin operating characteristics for ACS diagnosis in CKD patients with regard to history of coronary artery disease, electrocardiogram abnormalities, other comorbidities, or race and ethnicity.
Results
Two studies reported the operating characteristics of elevated troponin in diagnosing ACS among subgroups of patients with CKD. These studies reported one or more of sensitivity, specificity, PPV, or NPV by subgroups of age or CKD.39, 41
While these studies both examined the operating characteristics of the troponin T assay, they did so using different values of age and creatinine in their subgroups; thus their results cannot be directly compared except to say that the operating characteristics of troponin T appeared to vary by age and creatinine level (Table 10). Another study reported values of the area under the curve for subgroups (Table 11).43
Table 10
Operating characteristics of elevated troponin in the diagnosis of acute coronary syndrome among subgroups of patients with chronic kidney disease.
Table 11
Area under the curve for elevated troponin in the diagnosis of acute coronary syndrome among subgroups of patients with chronic kidney disease.
Two of the studies reporting results for subgroups39, 43 reported details of ACS adjudication, in contrast to other studies in this KQ. However, we can draw no conclusions about the operating characteristics of troponin assays in these studies compared with others, owing to heterogeneity in the type of operating characteristics reported.
This literature did not report many other subgroup characteristics that might be relevant to understanding the operating characteristics of a troponin assay in diagnosing ACS, including history of coronary artery disease, presence or absence of ischemic or other electrocardiogram changes, diabetes or other comorbidities, or race or ethnicity.
Strength of Evidence
We described the strength of the evidence addressing KQ1.2 in Tables 12 and 13.
Table 12
Numbers of studies and subjects, strength of evidence domains, magnitude of effect, and strength of evidence for the operating characteristics of elevated troponin among subgroups of patients with chronic kidney disease.
Table 13
Elevated troponin T versus nonelevated troponin T in terms of diagnostic accuracy in subgroups of age and chronic kidney disease stage among patients with chronic kidney disease: Details regarding strength of evidence domains.
KQ 1.3. Harms Associated with a False-Positive Diagnosis
Results
We found no studies addressing this KQ.
KQ 1.4. Direct Comparisons Between Troponin Assays
Results
Troponin T Versus Troponin I
One study addressed this question.42 The troponin T, Roche Elecsys assay using a cutoff of 0.1 mcg/L, was associated with a 100 percent sensitivity for ACS and a 42 percent specificity. By contrast, the Troponin I, DPC Immulite assay, using a cutoff of 1.0 mcg/L, had a sensitivity of 45 percent and a specificity of 100 percent. Both troponin assays predicted an increased risk of ACS, with area under the curve ranging from 0.7 to 0.8. We found no studies performing direct comparisons between troponin assays from the same manufacturer or using the same cutoff for the assay to diagnose ACS.
Troponin T Versus High-Sensitivity Troponin T
We found no studies addressing this comparison.
Troponin I Versus High-Sensitivity Troponin I
We found no studies addressing this comparison.
Strength of Evidence
The strength of evidence for KQ1.4 is insufficient given that it is based on one study of poor quality that is indirect, imprecise, and lacks consistency (since it is a single study).
KQ 1.5. Direct Comparisons of Troponin Testing in Patients with Chronic Kidney Disease Versus Patients with Normal Renal Function
Results
Although the studies reviewed in the previous section did include patients with normal renal function, we were not able to draw conclusions because of the size and quality of the studies. We found no studies that carried out direct a priori comparisons of troponin testing in patients with CKD versus patients with normal renal function.
KQ 2. Management of Acute Coronary Syndrome by Troponin Levels
We did not find any study that directly addressed the question of whether troponin levels can affect management strategies in chronic kidney disease (CKD) patients with acute coronary syndrome (ACS) symptoms. We identified one study by Barthelemy et al. that did not directly address this question since the study did not treat patients according to troponin levels, but they reported on troponin levels.51 This study did not answer KQ 2 as we defined it, but we discussed it here since it is the only study found that addressed troponin levels and management options in CKD patients with ACS symptoms.
Barthelemy et al. included patients with non-ST elevation ACS (diagnosis based on symptoms, ECG changes, and elevated troponin) scheduled for percutaneous coronary intervention and divided them according to those with and without renal failure. The study randomized ACS patients presenting to the emergency department to receive immediate or next working-day invasive management. In patients with a creatinine clearance less than 60 mL/min (n = 75), the peak cardiac troponin I level during hospitalization was not significantly different between those receiving immediate or next-day ACS management (P = 0.36). The study did not present a composite outcome of death, acute myocardial infarction (MI), urgent revascularization, or recurrent ischemia at 1 month separately based on elevated cardiac troponin I in the reported results; however, the authors stated in the discussion that “there was no increase in MI as evaluated by troponin I release.”51
We did not identify any additional studies meeting the criteria for KQ2.
KQ 2.1. Modification of a Troponin Elevation on Comparative Effectiveness of Interventions or Management Strategies for Acute Coronary Syndrome
Key Points
- The one study evaluating management of ACS in CKD patients did not find a significant difference in peak cardiac troponin I between the management groups (immediate vs. delayed invasive strategy) (strength of evidence: insufficient).
KQ 2.2. Modification of a Troponin Elevation on Comparative Effectiveness of Interventions or Management Strategies for Acute Coronary Syndrome by Subgroups
Barthelemy et al. did not do any subgroup analysis.
KQ 3. Short- and Long-Term Prognosis After Presentation with Acute Coronary Syndrome by Troponin Levels
Study Design Characteristics
We found 12 unique studies in 14 publications assessing the value of troponin in establishing prognosis for patients with CKD who presented with signs/symptoms of suspected ACS.40, 52-64
These studies included seven prospective studies,54, 56, 58, 59, 62-64 four retrospective studies,40, 53, 57, 60 and three post hoc analyses52, 55, 61 of previously published large randomized controlled trials (RCTs). The studies were published between 1999 and 2012 and enrolled patients from 1994 to 2008 with followups ranging from 1 month to 2 years. Three of the studies did not report the dates of enrollment52, 56, 63 and four of the studies did not specify the length of followup.40, 53, 55, 59 Studies did not report relevant details of study design uniformly.
The studies originated from the United States (nine studies),52, 54-56, 58-60, 62, 64 Europe (two studies, one from Germany63 and one from Spain40), one from Canada,57 one from Asia (Singapore)53), and one was a multinational study that recruited patients from 24 countries.61 Six studies enrolled the patients from the hospital,52-55, 61, 62 four from the emergency department,56, 57, 60, 64 two from the coronary care unit,58, 59 one from the dialysis unit,63 and one from two outpatient clinics as well as patients from the emergency department and the intensive care unit40 (Tables 14 through 17).
Table 14
Study design characteristics of studies evaluating the short- or long-term prognosis of patients with chronic kidney disease after presentation of acute coronary syndrome by troponin T levels.
Table 15
Study design characteristics of studies evaluating the short- or long-term prognosis of patients with chronic kidney disease after presentation of acute coronary syndrome by troponin I levels.
Table 16
Study design characteristics of studies evaluating the short- or long-term prognosis of patients with chronic kidney disease after presentation of acute coronary syndrome by troponin T and I levels.
Table 17
Study design characteristics of studies evaluating the short- or long-term prognosis of patients with chronic kidney disease after presentation of acute coronary syndrome by unspecified troponin levels.
Study Population Characteristics
These 12 studies included 46,988 subjects and varied widely in size. Two studies included less than 100 patients,60, 63 six studies included between 100 and 1,000 patients,40, 53, 56, 57, 62, 64 five studies included between 1,000 and 10,000 patients,52, 54, 58, 59, 61 and one study included 31,586 patients.55
Three studies by Kontos et al.54, 58, 59 recruited patients during the same time period, in the same institution, and under the same protocol, but aimed to predict mortality in patients admitted for exclusion of myocardial ischemia in different ways; Cockcroft-Gault equation versus Modification of Diet in Renal Disease equation,54 specific short-term and long-term prognostic value of troponin I for patients with and without CKD,58 and short-term and long-term outcomes and prognostic value of multiple variables (troponin, ejection fraction, and renal function).59 Even if the total population for these studies is not the same, some of the patients may recur from study to study.
All the studies included patients older than 40 years, with means ranging between 56 and 71 and medians ranging between 63 and 80. All studies included similar proportions of men and women. One study included many more men (72 percent) than women62 and one study did not report gender of participants.63 Only five studies reported race.53-56, 60 Han et al.60 recruited 83 percent African Americans, Melloni et al.55 recruited 82 percent Whites, Apple et al.56 and Kontos et al.54 recruited a more balanced population, and Chew et al.53 recruited a prevalently Chinese population (Singapore).
We included studies with very heterogeneous baseline diagnosis, comparators, and aims. All studies had the presentation of suspected ACS at enrollment, but the definition of ACS varied among them. Apple et al., defined its patients only by the presence of clinical symptoms.56 While other studies required the presence of symptoms and ECG and enzymatic changes,53, 55, 59, 61, 64 two studies categorized the patients as low, moderate, or high risk ACS,52, 54 one based it on medical records,60 and five studies did not specify any criteria for diagnosis.40, 57, 58, 62, 63 Only three studies reported how the diagnosis was adjudicated52, 53, 64 and whether there was a cardiologist involved.53 Only 50 percent of studies reported presence of CAD, which ranged from 14 to 68 percent in those studies that did report this variable.40, 53, 54, 57, 60-62
All studies included patients with renal failure but again, the definition of renal failure varied amongst them. Seven studies defined renal failure as a creatinine clearance less than 60 mL/min,40, 52, 54-56, 58, 59 three studies used serum creatinine to set the cutoff,60, 62, 64 one study classified patients per quartiles of creatinine clearance,61 and three studies did not specify definition or cutoffs.53, 57, 63 Four studies used the Cockcroft-Gault equation to calculate glomerular filtration rate,52, 58, 59, 62 three studies used the Modification of Diet in Renal Disease equation,40, 55, 56 one used both since its purpose was to compare them,54 and six studies did not specify the equation used.53, 57, 60, 61, 63, 64 Three studies included patients in all renal failure stages including end-stage patients requiring dialysis.53, 55, 64 Two studies included patients in all renal failure stages but excluded patients on dialysis54, 62 and four studies included patients in all CKD stages and did not specify if dialysis patients were included or not.56, 58, 59, 166 Two studies included only dialysis patients,57, 63 one study included only patients with severe stage patients (including patients both in medical treatment and dialysis),60 and one study included only patients with moderate renal failure.52
Seven studies evaluated troponin I,40, 54, 55, 57-59, 62 three studies evaluated troponin T,53, 60, 61 and three studies evaluated both types of troponin assay.56, 63, 64 One study did not specify which troponin it measured (Table 18 and Table 19).52
Table 18
Study population characteristics of studies evaluating the short- or long-term prognosis of patients with chronic kidney disease after presentation of acute coronary syndrome by troponin levels.
Table 19
Definitions used to define cardiac and renal populations in studies evaluating the short- or long-term prognosis of patients with chronic kidney disease after presentation of acute coronary syndrome by troponin levels.
Study Quality
The overall quality in the 14 studies evaluating the value of troponin in establishing prognosis for patients with renal failure who presented with signs/symptoms of suspected ACS was generally fair (three were good,52, 55, 57 eight were of fair quality,54, 56, 58-62, 64 and three were poor).40, 53, 63 All studies appropriately described their objective, interventions, outcomes and findings. Only one study did not describe the characteristics of the patients included.64 We felt that the included populations were representative of the general population in nine studies52, 55-58, 60-62, 64 and the setting (staff and facilities) was representative of a normal setting in eight studies.55, 57-59, 61-64 All the studies recruited their intervention groups from the same population and at the same time.
All the studies described the statistical methods used; none of the studies reported calculation of power (we found the power calculation for one study in the original RCT but not in the study we included52), seven studies reported on withdrawals,40, 57, 60-64 but all the studies took into account the losses to followup for the analyses. The authors described an adequate adjustment for confounding in the analyses in six studies,52, 55, 57, 58, 60, 64 only 21 percent of the studies (n=3)61, 63, 64 reported blinding the personnel who measured outcomes—43 percent (n=6)54-57, 59, 62 did not blind, and in 43 percent (n=6)40, 52, 53, 58, 60 blinding was not feasible due to the study design. Only one study did not do data dredging.62 All the studies reported accurate outcomes measures. Three studies did not report random variability estimates,40, 53, 55 and four studies did not report actual probability values.40, 54, 55, 60
In regard to funding, industry sponsored four studies52, 55, 61, 64 and government sponsored one.56 One study reported having no sponsorship57 and in eight studies this information was unclear.40, 53, 54, 58-60, 62, 63
KQ 3.1. Troponin Associations With Long-Term and Short-Term Outcomes
Key Points
- Elevated troponin I or T were associated with higher risk of short-term mortality (<1 year) and cardiac outcomes (strength of evidence: low).
- A similar trend was observed for long-term mortality (≥1 year) with troponin I (strength of evidence: low), but less evidence was found for long-term cardiac events for troponin I and long-term outcomes for troponin T (strength of evidence: insufficient).
- Patients with advanced stages of CKD tend to have worse prognosis with elevated troponin I than those without elevation (strength of evidence: moderate).
All-Cause Mortality
Troponin T
Four studies evaluated all-cause mortality, following a presentation for suspected ACS, in the context of troponin T levels: one with a long-term followup period (greater than 1 year),63 one with an unreported followup period,53 and two with short-term followup periods (Table 20).52, 55 The long-term study and one short-term study used a troponin T cutoff of 0.1 mcg/L, while the others did not specify the upper limits of normal.
Table 20
Association of an elevated troponin T level with all-cause mortality among patients with chronic kidney disease presenting with symptoms of acute coronary syndrome.
Wayand et al. conducted a small prospective cohort study that followed dialysis patients for 2 years and included 28 patients with myocardial discomfort or evidence of myocardial injury. The study analyzed both cardiac troponin T and I. Three patients with elevated cardiac troponin T values (>0.1 mcg/L) (n = 9) and one patient with a nonelevated cardiac troponin T died during followup (odds ratio [OR], 6.3; 95% confidence interval [CI], 0.6 to 69.7; P = 0.13). The study did not report the timing of these deaths.63
In the second study, Chew et al. found no significant difference in all-cause mortality between those with elevated (≥ 0.1 mcg/L) and nonelevated troponin T levels (P = 0.614). This was a retrospective study of 227 CKD patients with unstable angina pectoris, although the study did not report the number of patients in each group. Additionally, it did not give the duration of followup.53
The largest study of troponin T with an all-cause mortality outcome used data from an observational registry of patients admitted with ACS. A total of 13,843 patients had an estimated glomerular filtration rate less than 60 mL/min/1.73 m2 based on creatinine clearance. The study analyzed patients with mild CKD and normal kidney function jointly, so we did not consider data for stages 1 and 2 CKD for this review. Melloni et al. found an association between increases in troponin levels and death during initial hospitalization, though the study did not report the durations of hospital stays. The study grouped cardiac troponin T measurements by multiples of the assay's upper limits of normal. The study saw a trend toward death in those with higher troponin values assays. In those with an estimated GFR of 30-60 mL/min/1.73m2, the study saw mortality in 3.7, 5.3, and 7.3 percent of those with a troponin T value less than 1, 1 to 3, and greater than 3 times the upper limit of normal, respectively. For those with more severe CKD, these percentages were 7, 5.7, and 14 percent, respectively. However, after adjustment, troponin T elevation did not remain a significant predictor of mortality.55
Acharji et al. evaluated both cardiac troponin T and I, but did not distinguish between the two in the results or analysis, and therefore we did not include it in the strength of evidence analysis. This was a post hoc analysis of a large RCT reporting all-cause mortality in patients that had troponin measured prior to undergoing cardiac catheterization and revascularization after presenting with ACS. They analyzed data from the subjects in the RCT who had both CKD and baseline troponin T or I levels. The study did not list cutoff values for an elevated versus nonelevated test. They evaluated all-cause mortality at both 30 days and 1 year after presentation with ACS. Death within 30 days occurred in 4.7 percent (n = 60) of those with an elevated troponin versus only 1.0 percent (n = 9) with a non-elevated troponin (P < 0.0001). Similarly, 10.7 percent (n = 127) of those with an elevated troponin were dead at 1 year compared with 6.8 percent (n = 51) of those with non-elevated troponins (P = 0.0005). The study did not perform adjustment for this individual outcome.52
Troponin I
Seven studies investigated cardiac troponin I with an outcome of all-cause mortality.52, 54, 55, 58, 59, 62, 63 Because of overlap in patient cohorts and populations that were not exclusively ACS patients, we could not perform a pooled analysis (Table 21). The troponin I cutoff values ranged from 0.15 mcg/L to 1 mcg/L; two studies did not report a threshold.
Table 21
Association of an elevated troponin I level with all-cause mortality among patients with chronic kidney disease presenting with symptoms of acute coronary syndrome.
The only study we identified that reported on troponin I with a long-term outcome was the same study identified for troponin T that we described above. Out of a total of 28 patients, 14 had elevated cardiac troponin I values (≥ 0.4 mcg/L), and four of these patients died, whereas no patients with non-elevated cardiac troponin I died (OR, 9.0; 95% CI, 0.44 to 182.8; P = 0.15).63
A large study by Melloni et al., that used both troponin T and I (described above), grouped troponin values by multiples of the upper limit of normal, but do not specify the number of patients studied for each marker. After adjusting for patient characteristics and clinical factors, the only remaining significant association they found was between in-hospital mortality and elevated troponin I greater than 3-times the upper limit of normal in patients with an estimated GFR of 30-60 mL/min/1.73m2 (OR, 1.8; 95% CI, 1.3 to 2.5).55
Kontos et al. evaluated all-cause mortality in patients admitted to a large hospital after presenting to the emergency department with chest pain. This included 1,084 patients with creatinine clearance less than 60 mL/min; however, those with mild kidney dysfunction (creatinine clearance greater than 60 mL/min) and patients with normal kidney function were analyzed as a single group and therefore not appropriate for evaluation in this review. A significantly larger number of patients with creatinine clearance less than 60 mL/min who presented with elevated troponin levels died within 1 year, (12.6 percent) compared with those with non-elevated troponin I levels (6.8 percent; OR, 1.9; 95% CI, 1.4 to 2.5; P = 0.0001). Notably, this population excluded patients with ST elevation acute MI and was not exclusively ACS, as it may have included those with stable angina.58
Two additional studies by the same author meeting inclusion criteria for this review also included all-cause mortality as an outcome in ACS patients with CKD.54, 59
Acharji et al. evaluated both cardiac troponin T and I, but did not distinguish between the two in the results or analysis. We described the results above.52
Major Adverse Cardiovascular Events
Troponin T
In addition to the outcome of all-cause mortality, we also considered composite cardiac mortality, acute MI, cardiac ischemia, revascularization, dysrhythmia, and congestive heart failure exacerbation, as well as various composites of these endpoints. We did not identify any studies of cardiac troponin T that met inclusion criteria and evaluated MACE with a followup period of greater than 1 year.
We identified four studies of troponin T using short-term MACE outcomes following a presentation of suspected ACS (Table 22).52, 56, 60, 61 Troponin T cutoff values ranged from 0.01 mcg/L to 0.1 mcg/L. One report justified using a 0.1 mcg/L threshold by noting that the 99th percentile in the reference population was below the lower limit of detection of 0.01 mcg/L.61
Table 22
Association of elevated troponin T with major adverse cardiac events among patients with chronic kidney disease presenting with symptoms of acute coronary syndrome.
A post hoc analysis of an RCT with a composite outcome of 30-day acute MI or death found significant differences between patients with elevated and nonelevated troponin T. This study included patients with and without kidney dysfunction and presented results by quartile of creatinine clearance. There was a higher percentage of events in those with an elevated versus nonelevated troponin T when using a cutoff value of either 0.1 mcg/L (12.4 percent vs. 6.9 percent, respectively) or 0.03 mcg/L (12.2 percent vs. 5.3 percent). We presented results of the higher cutoff in Table 22. The results of the first two quartiles were significant after adjusting for sex, older age, ST-segment depression, and a history of angina, acute MI, stroke, diabetes, bypass surgery, and angioplasty. We provide an analysis of the quartiles separately below.61
Apple et al. reported a 6-month composite outcome of acute MI or death in 135 CKD patients with estimated glomerular filtration rates of less than 60 mL/min/1.73m2. The difference in event rate in those with elevated versus nonelevated troponin T was not statistically significant. (OR, 2.5; 95% CI, 1.0 to 6.3; P = 0.06).56
The study by Acharji et al. (described above) presented several outcomes for patients with measured troponin T or I, although the analysis did not distinguished the type of troponin. These outcomes included rate of cardiac death, which was significantly higher in the elevated troponin group than in the nonelevated troponin group at 30 days (P < 0.001) and 1 year (P = 0.0001). At both 30 days and 1 year, rates of ischemia and acute MI were higher in those with elevated troponin values than non-elevated troponin values (P < 0.05 for both). Differences in rates of unplanned revascularization were not significant. The only outcome presented as adjusted data was composite death or acute MI. Death or MI remained statistically significant after adjusting for baseline clinical characteristics and ECG and laboratory findings. This was true at 30 days (HR, 2.1; 95% CI, 1.5 to 2.8; P < 0.0001) and 1 year (HR, 1.7; 95% CI, 1.4 to 2.2; P < 0.0001).52
A study of 90 CKD patients presenting to the emergency department with symptoms of ACS by Han et al. used a composite endpoint of acute MI, unstable angina, revascularization, cardiac dysrhythmias, all-cause mortality, or congestive heart failure exacerbation. Using receiver operating curve analysis, the authors found that an increase in troponin T of 0.11 mcg/L compared with a prior non-ACS measure had a sensitivity of 27 percent and a specificity of 96 percent for the composite outcome at 6 months (positive likelihood ratio 7.2). The study did not provide the rate of events in groups with and without an elevated troponin T.60
Troponin I
Three of the studies reporting on short-term MACE outcomes for troponin I by the same author included substantial overlap in patient populations;54, 58, 59 therefore, we presented the most relevant results here. We identified five additional studies of troponin I.40, 52, 56, 57, 62 These included a wide range of troponin I cutoff values, from 0.0001 mcg/L to 1 mcg/L, although one study did not specify the threshold used (Table 23).
Table 23
Association of an elevated troponin I level with major adverse cardiac events among patients with chronic kidney disease presenting with symptoms of acute coronary syndrome.
Apple et al. reported a 6-month composite outcome of acute MI or death in CKD patients with estimated glomerular filtration rates less than 60 mL/min/1.73m2 for three troponin I assays. All assays resulted in a statistically significant higher event rate in those with elevated troponin levels (Dade: OR, 3.0; 95% CI, 1.3 to 6.8, P = 0.01; Beckman: OR, 3.0; 95% CI, 1.2 to 7.1, P = 0.01; Tosoh: OR, 3.6; 95% CI, 1.1 to 11.4; P = 0.03); however, there was some variation between assays. In the Tosoh and Beckman studies, respectively, event rates ranged from 9.6 to 15.6 percent in those with non-elevated troponin levels, and from 34.4 to 42.6 percent in those with elevated troponin values.56
Kontos et al. recruited patients who presented to an emergency department with chest pain, although the study excluded those with ST-segment elevation. The study defined cardiac death as death caused by acute MI, CAD, or arrhythmia. In 1,084 patients with creatinine clearance less than 60 mL/min, there were significantly fewer cardiac deaths in those with non-elevated troponin I levels (3.2%) than in those with elevated troponin I levels (9.3%).58
Flores et al. presented results of a retrospective study of 467 patients with creatinine clearance less than 60 mL/min and with suspected myocardial injury. They found an increased incidence of acute MI as primary diagnosis on discharge in those with troponin I between 0.05 and 0.5 mcg/L (8.3 percent, n = 14) and over 0.5 mcg/L (50.8 percent, n = 33) compared with those with a non-elevated troponin I (n = 0).40
A study of 149 chronic dialysis patients used a composite endpoint that included cardiac death, acute MI, revascularization, or de novo congestive heart failure within 30 days of presentation. Bueti et al. found that a troponin I greater than 0.0001 mcg/L had a strong association with the outcome (OR, 15.2; 95% CI, 5.3 to 43.6; P = 0.0000004). This remained strongly significant when adjusting separately for sex, blood pressure, and prior cardiovascular disease. This study included patients presenting to the emergency department for any reason who had a troponin I value recorded: 29 percent presented with chest pain and 20 percent presented with symptoms that were noted to be clearly non-cardiac. The association between clinical presentation and troponin I was not significant (P = 0.7), suggesting that the ability of troponin I to predict the outcome was similar in those presenting with cardiac and non-cardiac complaints.57
Although Gruberg et al. found 1-year all-cause mortality different between those with elevated versus non-elevated troponin I (as described above), there were no significant differences between troponin I groups for 1-year acute MI (P = 0.06), revascularization (P = 0.88), or composite MACE (death, acute MI, or revascularization) (P = 0.16).62
We presented results above of a study by Acharji et al., which did not distinguish between troponin T and I values.52
Strength of Evidence
We list the strength of evidence for the body of literature addressing KQ3.1 in Tables 24 and 25.
Table 24
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of prognosis after acute coronary syndrome among patients with chronic kidney disease: Strength of evidence domains.
Table 25
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of prognosis after acute coronary syndrome among patients with chronic kidney disease: Details regarding strength of evidence domains.
KQ 3.2. Troponin Associations with Long-Term and Short-Term Outcomes by Subgroups
Key Points
- Patients with more advanced stages of CKD and elevated troponin I seem to be at higher risk of adverse outcomes than those with nonelevated troponin I (strength of evidence: moderate).
- Elevated troponin was associated with a higher risk of adverse cardiac outcome in dialysis patients with ACS compared with normal troponin levels, although the quality and heterogeneity of study designs limits the strength of this finding (strength of evidence: low).
- We did not find any studies that reported on the ability of elevated troponin to estimate prognosis after ACS in subgroups of CKD patients based on sex, age, status after renal transplant, presence of previously elevated troponin, ECG changes, comorbidities, smoking status, 10-year CAD risk, or history of CAD (strength of evidence: insufficient).
Results
The only subgroups presented in the studies meeting criteria for KQ3 were extent of kidney disease and utilization of dialysis.
Stage of Chronic Kidney Disease or Creatinine Clearance
Troponin T
Aviles et al. presented their study results by quartile of creatinine clearance, rather than standard stage of CKD. The authors found a significantly higher rate of death or MI in those with a troponin T greater than 0.1 mcg/L in creatinine clearance groups less than 58.4 mL/min and 58.4 to 76.9 mL/min (P < 0.001 for both). The difference for creatinine clearance 77.0 to 98.6 mL/min was insignificant (P = 0.16); however, this result became significant when they used a lower troponin T cutoff value of 0.03 mcg/L for analysis (P < 0.001).61
Melloni et al. did not find a significant difference in in-hospital mortality between those with elevated and nonelevated troponin T based on the hospital's upper limit of normal value when they considered stages of CKD separately.55
In a post hoc analysis of an RCT, Acharji et al. considered patients with creatinine clearance less than 30 mL/min separately from those with creatinine clearance 30 to 60 mL/min. Types of troponin included both T and I (threshold not specified) but they did not distinguish between the two in the analysis. The only statistically significant difference in outcomes between troponin groups that the study saw were in the creatinine clearance 30 to 60 mL/min subgroup. These included all-cause mortality, cardiac death, acute MI, and composite death or acute MI (P ≤ 0.001 for all) at both 30 days and 1 year.52
Troponin I
In their large analysis of registry data, Melloni et al. grouped patients by estimated glomerular filtration calculated via the Modification of Diet in Renal Failure method. After adjusting for patient characteristics and other factors known to be associated with in-hospital mortality, the only association that remained statistically significant was death in stage 3 CKD patients with elevated troponin I at more than 3-times the hospital-specified upper limit of normal (OR, 1.8; 95% CI, 1.3 to 2.5; P < 0.0012). They did not report ORs for insignificant adjusted analyses.55
One multivariate analysis, Kontos et al., that adjusted for age, sex, hypertension, prior revascularization or acute MI, left ventricular hypertrophy, and ischemic ECG changes, reported that an elevated troponin I (>1 mcg/L for Opus assay and >0.3 mcg/L for Bayer assay) was a predictor of 1-year all-cause mortality in patients with creatinine clearance 30 to 60 mL/min (HR, 1.7; 95% CI, 1.1 to 2.6) and creatinine clearance less than 30 mL/min (HR, 3.0; 95% CI, 1.8 to 5.0). Additionally, elevated troponin I was a predictor of 1-year cardiac mortality in patients with creatinine clearance 30 to 60 mL/min (HR, 2.2; 95% CI, 1.3 to 3.8) and with creatinine clearance less than 30 mL/min (HR, 3.3; 95% CI, 1.8 to 6.1). Thirty-day all-cause mortality was higher in those with an elevated versus nonelevated troponin I by CKD subgroup (10 vs. 3.8 percent in those with creatinine clearance 30 to 60 mL/min and 26 vs. 9.7 percent in those with creatinine clearance less than 30 mL/min).58
We presented the results of a CKD subgroup analysis for a study considering troponins T and I jointly above.52
Dialysis Status
Melloni et al. analyzed a nondialysis subgroup from a large cohort of CKD patients and did not demonstrate a significant difference from the results for the entire population of CKD patients. The study saw a trend toward death in those with higher troponin values for both troponin T and I in those with CKD not undergoing dialysis.55
Two studies only included those undergoing chronic dialysis (described above), and these have limitations.57, 63 The former was a small cohort of 28 patients and, although it reported long-term mortality as an outcome, it did not report timing of patient deaths. The latter study found an elevated troponin I had a strong association with a composite 30-day outcome including cardiac death, acute MI, revascularization, or de novo congestive heart failure (OR, 15.2; 95 percent CI, 5.3 to 43.6; P = 0.0000004). Limitations of this study included a low cutoff value for elevated troponin I (0.0001 mcg/L) and that it included all dialysis patients presenting to the emergency department (i.e., not strictly an ACS population).
Strength of Evidence
We listed the strength of evidence for the body of literature addressing KQ3.2 in Tables 26 and 27.
Table 26
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of prognosis after acute coronary syndrome by subgroups of patients with chronic kidney disease: Strength of evidence domains.
Table 27
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of prognosis after acute coronary syndrome by subgroups of patients with chronic kidney disease: Details regarding strength of evidence domains.
KQ 3.3. Direct Comparisons Between Troponin Assays to Estimate Prognosis After Acute Coronary Syndrome
Key Points
- We are unable to determine if there is a difference in the performance of troponin assays to estimate prognosis after ACS in patients with kidney disease based on three very heterogeneous studies with indirect and imprecise estimates (strength of evidence: insufficient).
- We did not identify any studies that included high sensitivity troponin I or T.
Results
Troponin T Versus Troponin I
Two studies directly compared troponin T and I by measuring performance in the prediction of composite cardiac ischemic endpoints; however they used different cutoff values and there were differences in the cardiac events comprising the outcome (Table 28).63, 64 From these results, it is difficult to determine the extent to which differences in predicting prognosis are due to the type of troponin or to the cutoff the studies used. One of these studies also compared receiver operating curve characteristics and found the difference between the area under the curve for troponin T and I to be insignificant (P = 0.213).63
Table 28
Results from studies directly comparing troponin T with troponin I to estimate prognosis after acute coronary syndrome.
A study by Apple et al. compared four troponin assays in ACS patients with an estimated glomerular filtration rate less than 60 mL/min/1.73m2 for a composite outcome of acute MI or death. These included troponin I by Beckman, Dade, and Tosoh, and troponin T by Roche. Six-month event rates were significantly different in elevated versus nonelevated troponin groups for all assays (P < 0.05 for all). Although there were differences in exact event rates between the assays, the study reported no measures of significance for these differences.56
Troponin T Versus High-Sensitivity Troponin T
We did not identify any studies that met inclusion criteria and evaluated troponin T versus high-sensitivity troponin T.
Troponin I Versus High-Sensitivity Troponin I
We did not identify any studies that met inclusion criteria and evaluated troponin I versus high-sensitivity troponin I.
Strength of Evidence
We listed the strength of evidence for the body of literature addressing KQ3.3 in Tables 29 and 30. The strength of evidence is insufficient to compare the performance of troponin subclasses because the effects were not consistent; the precision could not be determined; the magnitude of effect was weak; and the rating is limited by the heterogeneity of the overall risk of bias of the assays the studies used, and the populations the studies included.
Table 29
Comparisons between troponin assays to estimate prognosis after acute coronary syndrome among patients with chronic kidney disease: Strength of evidence domains.
Table 30
Comparisons between troponin assays to estimate prognosis after acute coronary syndrome among patients with chronic kidney disease: Details regarding strength of evidence domains.
KQ 4. Use of Troponin for Risk Stratification Among Chronic Kidney Disease Patients Without Acute Coronary Syndrome
Study Design Characteristics
We included 98 studies (in 105 publications) that evaluated the use of troponin levels for risk stratification among patients with chronic kidney disease (CKD) without acute coronary syndrome (ACS) symptoms, KQ 4.9, 11, 25, 26, 47, 65-163
The studies took place in diverse countries, including 18 in the United States, 8 in Canada, 60 in Europe, 10 in Asia, 3 in Middle-East, 1 in Mexico, 6 in Australia, and 1 in multiple countries.
Studies varied in their sources of support. Twenty-four received industry funding, 26 reported no industry support, and the remainder did not report on support.
All studies were observational cohort studies. Enrollment into 21 studies started and ended before or in 2000,26, 89, 94, 98, 100, 104, 114, 118, 121, 122, 124, 127, 133, 138, 140, 142-144, 151, 152, 167 while 45 studies did not report the dates of enrollment period.11, 47, 69, 70, 75-77, 79, 81, 82, 87, 91, 93, 102, 103, 106, 110-113, 116, 117, 120, 123, 125, 126, 128-132, 134-137, 139, 141, 145-150, 153, 155
The median study followup time ranged from 30 days to 5 years.
Forty-six studies recruited patients in the outpatient setting, 50 took place in hospital setting, and 36 in dialysis centers.
Study Population Characteristics
The characteristics of studies included in KQ4 are outlined in Table 31. The study sample size ranged from 16151 to 8,121.69 Five studies did not report the age distributions.65, 70, 76, 121, 140 Among others, the mean/median age of study populations ranged from 32115 to 77 years.25 Six studies did not report gender distribution.65, 76, 116, 126, 129, 146 Two studies included only men.93, 148 Among other studies, the percentage of men ranged from 14 percent154 to about 80 percent.121
Table 31
Population characteristics of studies evaluating the use of troponin levels in risk stratification among patients with chronic kidney disease without symptoms of acute coronary syndrome.
Sixty-five studies specifically excluded ACS patients, while 38 studies did not report ACS inclusion/exclusions. Seven studies included patients with CKD stage 1 to 4; one included patients with CKD stage 3 to 4; eight included patients with CKD stage 5; 75 included dialysis patients; and six studies included kidney transplant patients. Eight used the Modification of Diet in Renal Disease equation; two used the CKD-Epi equation; and five used the Cockcroft-Gault formula.
Study Quality
Table 32 describes the quality of studies for KQ4. We rated the overall study quality fair to good as described in methods section. Although the adjustment of confounders was one of the factors considered in study quality assessment, it was not the only factor (i.e., a study could still have fair or good quality even without confounder adjustment if it was otherwise a well-done study with clear cutpoints, clear reporting of outcome ascertainment, and appropriate statistical methods, etc). Industry funding was not factored into the overall quality assessment, but we listed it here for reference.
Table 32
Select quality scores for studies evaluating the risk associated with elevated troponin among patients with chronic kidney disease.
Results: Inclusion of Studies in Meta-Analysis for KQ 4
Appendix E Tables 1-7 outline the studies used in meta-analysis for KQ4, and whether they were included in meta-analyses for hazard ratios (HR), odds ratios (OR), or excluded from both meta-analyses. We excluded studies from meta-analyses if there was insufficient information to derive any HR or OR, if the study presented troponin as a continuous variable rather than a cutpoint, or if the cutpoint for troponin elevation was unclear. Also, if there are multiple papers of results derived from the same cohort, we presented results from each unique cohort once for each outcome. We also noted the reasons for exclusion in Appendix E Tables 1-7. The meta-anlayses for HRs are stratified by the level of adjustment. The list of covariates for each study is presented in Appendix F.
After performing the literature search, it became clear that the majority of studies reported results in a cohort of patients receiving dialysis. The other studies were a mix of CKD stages 1-5 including or excluding dialysis patients. To avoid further heterogeneity, we presented outcome results for dialysis and nondialysis patients separately in regard to KQs 4.1 and 4.2.
Results for Patients on Dialysis
Key Points
- Among dialysis patients without suspected ACS, a baseline elevated value of cardiac troponin was associated with a higher risk (∼2-4 fold) for subsequent short- and long-term outcomes including all-cause mortality, cardiovascular-specific mortality, and MACE (i.e., “composite” outcome of MI, cardiovascular death, and/or revascularization) in models adjusted at least for age and CAD or risk equivalent (Figure 6).
- More of the studies we included in the pooled meta-analyses reported outcomes for all-cause mortality than for other outcomes. Thus, the evidence from the pooled meta-analysis is strongest for the association of elevated cardiac troponin with all-cause mortality.
- We found approximately a 3-fold increased risk for the association of cardiac troponin T and I with all-cause mortality, which was highly significant (strength of evidence: moderate). The evidence from meta-analyses for an association of elevated cardiac troponin with cardiovascular-specific mortality and MACE showed similar effect sizes but with wider confidence intervals (CIs) from fewer studies (strength of evidence: moderate for CVD mortality for Troponin T and I; moderate for MACE for Troponin T with insufficient evidence for MACE for Troponin I).
- The association of elevated troponin with adverse outcomes among dialysis patients was generally similar for troponin T versus I. Few studies reported results for high-sensitivity troponin T and high-sensitivity troponin I assays; thus, less is known about how well these assays predict risk (strength of evidence: low).
- A sensitive assay identifies more patients as being elevated.
- While almost all studies supported a positive association for elevated cardiac troponin with adverse cardiovascular outcomes, particularly mortality, there was substantial heterogeneity among the studies, even though troponin T and I were analyzed separately.
- Much of the heterogeneity across results stemmed from differences across the literature between the various types of troponin assays used (different manufacturers, different assay platforms). Troponin assays have been changing, and newer generations of assays can detect lower concentrations of cardiac troponin. Many of the articles did not report which generation of assay they used.
- The studies varied markedly regarding which cutpoints they selected to define as elevated. Many studies did not report what the manufacturer-reported 99th percentile threshold was for that assay. The 99th percentile threshold was also a changing target depending on what reference population or assay generation was used. The reference populations for the 99th percentiles were largely unclear, and were most likely not taken from a dialysis cohort. Therefore, we were not able to perform meta-analyses using the 99th percentile cutpoint, but instead compared the highest cutpoint reported in each study with the lowest cutpoint for consistency.
- The meta-analyses we performed for the pooled ORs were unadjusted results using number of events in each arm. For the meta-analyses for HRs, we selected the most-adjusted regression model. Many studies only reported an unadjusted HR. While many studies did adjust for age, few studies adjusted for a history of CAD, CAD risk equivalent (such as diabetes mellitus), or for other causes of elevated troponin, such as heart failure. Even fewer studies adjusted more comprehensively for other cardiovascular risk factors, such as systolic blood pressure, dyslipidemia, and smoking. However, associations generally did remain robust in adjusted models (when available) and thus we considered them reliable.
- Studies almost exclusively looked at the association of troponin with outcomes in regression models, but did not examine the utility of troponin as a useful marker for prediction by metrics of reclassification or discrimination (i.e. show whether troponin improves re-classification of individuals into lower or higher clinical risk groups).
- Only one study104 rigorously examined whether troponin testing among dialysis patients added incremental prognosis over routine clinical and laboratory factors, but troponin testing did not change the area under curve in their survival model. Thus, it is unknown whether measuring cardiac troponin facilitates risk prediction in dialysis patients better than a traditional risk prediction model using only clinical variables.
- All of the studies related to this question were observational cohort studies. We did not find any intervention studies that compared management strategies of dialysis patients (without suspected ACS) on the basis of elevated troponin. Thus, while elevated cardiac troponin is clearly a marker of increased risk for subsequent cardiac events, it is unknown whether changing patient management (such as more intensified preventive efforts) on the basis of elevated troponin can reduce cardiovascular morbidity and mortality.
KQ 4.1.a. Distribution of Troponin Values Among Patients on Dialysis
The number (percent) of the study populations with elevated troponin values is noted in Table 33. This was only available for studies that listed the number of patients with elevated values out of the total sample. Some studies did not provide this information. As outlined in our methods section, we only abstracted data from studies that also reported outcomes. We did not abstract studies that reported on the prevalence of elevated troponin in their cohort but had no outcome data and thus we did not include them in this list.
Table 33
Prevalence of elevated baseline troponin T and I levels at maximum cut-point among patients on dialysis.
Prevalences depend on the clinical characteristics (i.e., pre-test probability) of each study group as well as the heterogeneity between assays and cutpoints. As such, we found prevalences widely varied across studies. For troponin T, the prevalence of dialysis patients with troponin levels above cutpoints ranged from 12 to 82 percent. Some of the heterogeneity was due to different cutpoints the studies used to define “elevated,” but heterogeneity across studies remained even when studies used similar cutpoints. In general, lower cutpoints (i.e., 0.03 mcg/L) identified a higher prevalence of patients defined as elevated, as would be anticipated by a more sensitive cutpoint. For example, the prevalence of elevated troponin for cutpoints 0.01 to 0.03 mcg/L ranged from 45 to 82 percent. A more conservative cutpoint (such as 0.1 mcg/L) had a lower prevalence of patients defined as elevated. Still, even for a cutpoint of 0.1 mcg/L, the prevalence ranged from 12 to 50 percent across cohorts, averaging around 25 percent.
For troponin I, the prevalence of patients defined as elevated ranged from 6 to 60 percent. There was no clear pattern of prevalence by cutpoints across studies. For low cutpoints (0.01 to 0.03 mcg/L), the prevalence ranged from 19 to 60 percent. For higher cutpoints (0.3 mcg/L), the prevalence ranged from 6 to 30 percent. Of note, one study used a high cutpoint of 2.3 mcg/L,135 and the prevalence was still high at 21 percent.
Some studies evaluated prevalences of elevated troponin in the same population as noted in Table 33 below. For example, in a study by Apple et al 2002, the prevalence of elevated troponin T (>0.1 mcg/L) was 20 percent but the prevalence of elevated troponin I (>0.1 mcg/L) was 6 percent when tested in the same cohort of patients.
KQ 4.2a. Troponin Associations With Short- and Long-term Outcomes Among Patients on Dialysis
The Association of Cardiac Troponin T With All-Cause Mortality Among Patients on Dialysis
Overview
Forty-three unique patient cohorts (among 49 publications) presented results regarding the association of baseline troponin T levels with all-cause mortality among dialysis (only) patients without symptoms of ACS.26, 66, 67, 72, 74, 75, 77, 81, 83, 85-87, 89, 92, 95, 96, 98, 104, 106, 110, 112-114, 116, 118, 122, 124, 126, 127, 130, 133, 134, 136-142, 144, 145, 147-149, 153, 155, 157, 160, 163
We excluded ten studies from the meta-analyses of both HRs and ORs due to insufficient data, or because the studies did not present results separately for dialysis patients only. We included the remaining studies in HR meta-analysis, OR meta-analysis, or both. We included a summary of the inclusion and exclusion reasons in Appendix E, Table 1.
Followup Time
All studies except one had a followup time for mortality events equal or greater to 1 year with time ranging from 1 to 5 years. Choy126 reported a followup time of only 6 months.
Assays and Cut-Points
The cardiac troponin T assay was generally measured by one manufacturer (Roche) or by Boehringer Mannheim, which was acquired by Roche Diagnostics in 1997. The most common cut-point studies used to define elevated troponin was a troponin T greater than 0.1 mcg/L, with a cut-point of more than 0.03 mcg/L being the second most common. These do not clearly reflect the 99th percentile (as compared with Appendix G, which outlines the 99th percentile by assay as described by the manufacturer). However, the 99th percentile is a changing target based on the assay generation and reference population. Many of the articles did not clearly state which generation of assay they used, or whether the cut-point was the 99th percentile value or some other threshold. Some studies chose a value selected by a Receiver Operator Curve analysis. Therefore it was difficult to compare studies across the 99th percentile.
Hazard Ratio for All-Cause Mortality Associated With Elevated Cardiac Troponin T
We listed the results from the meta-analysis (n=21 studies) stratified by the level of adjustment that presented HRs for the association of elevated troponin T with all-cause mortality among dialysis patients in Figure 7. All studies we included in this meta-analysis have reported a HR with CIs or we were able to derive the CIs using the spreadsheet provided by Tierney et al.32
Of these studies, six were unadjusted, 15 adjusted at least for age, and 11 adjusted at least for age and history of CAD (or CAD risk equivalents such as cardiovascular disease, congestive heart failure, ejection fraction, or diabetes mellitus) in their models. In two studies, the authors performed a more thorough regression model by additionally adjusting for numerous cardiovascular risk factors including blood pressure, lipids, and diabetes.
In all studies, there was a positive association between elevated cardiac troponin T and all-cause mortality (HR >1.0), although the HRs varied widely—from as low as 1.07 up to 15.5. Most studies were statistically significant, but in three of the 21 studies the CIs crossed 1.0, although the effect estimate was similar to the other studies which were statistically significant. The pooled meta-analysis for the HR among studies that adjusted for at least age and CAD or risk equivalent was statistically significant and provided evidence for a 3-fold increased risk of all-cause mortality associated with elevated troponin T (HR, 3.0; 95% CI, 2.4 to 4.3); heterogeneity not significant (I-squared, 42 percent).
Figure 7 includes studies stratified by level of adjustment, but we also viewed data sorted by year of publication. We found no temporal trends (in terms of strength of association or statistical significance).
Odds Ratio for All-Cause Mortality Associated With Elevated Cardiac Troponin T
Twenty-four studies provided the number of events among elevated and nonelevated troponin T groups making it possible to determine an unadjusted OR. Figure 8 presents the results from the pooled meta-analysis for the unadjusted OR for all-cause mortality by elevated troponin T level among dialysis patients.
All studies showed a positive association of elevated cardiac troponin T with all-cause mortality (OR >1.0). Most of the studies were statistically significant, but three of the 24 studies reported insignificant associations (CIs crossed 1.0), although the effect estimation was similar to the other studies. The overall pooled OR showed a five-fold increased risk (OR, 4.7; 3.6 to 6.5) with significant heterogeneity (I-squared, 53 percent.)
The Association of Cardiac Troponin I With All-Cause Mortality Among Patients on Dialysis
Overview
We identified 31 publications representing 30 unique patient cohorts that presented results regarding the association of baseline cardiac troponin I levels with all-cause mortality among dialysis patients without symptoms of ACS.11, 65-67, 70, 75, 77, 81, 88, 92, 93, 96, 101, 112, 113, 117-119, 126, 131, 133, 134, 138, 139, 147-150, 157, 162, 168
We excluded seven studies from the meta-analysis of both HRs and ORs due to insufficient data, troponins presented as continuous variables rather than cut-points, or results that were not presented separately for dialysis patients. We did not include the remaining studies in HR meta-analysis, OR meta-analysis, or both. We presented a summary of these inclusion and exclusion reasons in Appendix E, Table 2.
Followup Time
All studies except one had a followup time for mortality of at least 1 year with time ranging from 1 to 5 years. Choy, 2003126 reported a followup time of only 6 months.
Assays and Cutpoints
The most common cardiac troponin I assays that studies used were the Beckman, Dade-Behring, and Abbott assays. Multiple studies compared two or more troponin I assays in the same study population.92, 96, 118, 139, 147-149 For the purpose of meta-analysis, we only used only one cardiac troponin I assay per population. The cut-points for elevation were extremely heterogeneous, ranging from 0.01 to 0.4 mcg/L.
Hazard Ratio for All-Cause Mortality Associated with Elevated Cardiac Troponin I
Ten studies provided HRs and 95 percent CIs suitable for meta-analysis. All of these studies suggested an increased risk of mortality associated with cardiac troponin I elevation (HR >1.0). However two of the studies did not meet statistical significance (CIs crossed 1.0). All except for one of these studies at least adjusted for age, and seven out of 10 additionally adjusted for CAD or CAD risk equivalent (CAD, cardiovascular disease, heart failure, diabetes).
We listed the pooled meta-analysis stratified by level of adjustment in Figure 9. The pooled HR of an elevated cardiac troponin I for all-cause mortality among studies that adjusted for at least age and CAD or risk equivalent was 2.7 (95% CI, 1.9 to 4.6), heterogeneity not-significant (I-squared = 27 percent). Again, we saw no apparent temporal trends when we sorted results by year of publication.
Odds Ratios for All-Cause Mortality Associated with Elevated Cardiac Troponin I
Nineteen studies provided enough data (i.e., number of events in each group) to be included in meta-analysis for ORs. The majority of studies showed a positive association between elevated cardiac troponin I and all-cause mortality. In two studies, the point estimate tended toward an inverse association, although not statistically significant. In fact, eleven of the 19 studies did not reach statistical significance, largely due to small sample size and small number of events in each group, as indicated in Figure 10.
The unadjusted pooled OR for all-cause mortality associated with elevated troponin I was 2.6 (95% CI, 1.9 to 3.6). Heterogeneity was not significant (I-squared=18 percent).
The Association of High-Sensitivity Cardiac Troponin T With All-Cause Mortality Among Patients on Dialysis
We found only one study that evaluated the association of a high-sensitivity troponin T assay with mortality. This study76 tested high-sensitivity troponin T (assayed by Roche E411 analyzer) on a continuous scale, rather than using a cut-point. These authors found that for every 2.72 ng/L increase in high-sensitivity troponin T level, the age-adjusted risk of all-cause mortality increased 1.4-fold (HR, 1.4; 95% CI, 1.0 to 2.0, P = 0.049]).
The Association of High-Sensitivity Cardiac Troponin I With All-Cause Mortality Among Patients on Dialysis
We found only one study that evaluated the risk of all-cause mortality for high-sensitivity cardiac troponin I among dialysis patients. Assa et al.157 evaluated the risk of high-sensitivity troponin I with all-cause mortality per 10 ng/L increase in troponin and did not find a statistically significant association; but the study may have been underpowered for this outcome.
The Association of Cardiac Troponin T With Cardiovascular Mortality Among Patients on Dialysis
Overview
We identified 20 studies representing 16 unique patient cohorts that reported results on the association of cardiac troponin T with cardiovascular-specific mortality.26, 89, 95, 98, 104, 111-113, 122, 124, 136-140, 144, 151, 152, 155, 158
We excluded two studies from meta-analysis of both HRs and ORs due to insufficient data. We presented a summary of these inclusion and exclusion reasons in Appendix E, Table 3.
Followup time ranged from 1 to 4.3 years.
Hazard Ratio for Cardiovascular-Specific Mortality Associated With Elevated Cardiac Troponin T
We identified seven studies that reported a HR with CIs. All of these studies suggested an increased risk, although three of seven studies did not meet statistical significance. All except one of the studies adjusted at least for age, and five of the seven studies additionally adjusted for CAD or CAD risk equivalent (CAD, cardiovascular disease, diabetes, or heart failure). We listed the pooled meta-analysis in Figure 11. Among the five studies that adjusted for at least age and CAD, there was a 3-fold increased risk (HR, 3.3; 95% CI, 1.8 to 5.4), there was substantial heterogeneity (I-squared, 66 percent), and when we sorted results by year, we saw no temporal trend.
Odds Ratio for Cardiovascular-Specific Mortality Associated With Elevated Cardiac Troponin T
Nine studies provided the number of events in each group, making it possible to determine unadjusted ORs. In one study (Duman 2005112), the authors reported an adjusted OR but did not report the number of events and sample sizes in each group. All of the studies suggested a positive association with increased risk, although three of the nine studies did not meet statistical significance.
We reported the pooled meta-analysis for the odds of cardiovascular mortality for elevated cardiac troponin T in Figure 12; it suggests a 4-fold increase in risk (OR, 4.3; 95% CI, 3.0 to 6.4); no heterogeneity found.
In a sensitivity analysis, including the one study with an adjusted OR, the pooled results were similar (OR, 4.5; 95% CI, 3.2 to 6.3).
The Association of High-Sensitivity Cardiac Troponin T With Cardiovascular Mortality Among Patients on Dialysis
One study158 presented results using a high-sensitivity cardiac troponin assay (Roche, 99th percentile 0.0135 mcg/L) but presented results as a continuous variable per 100 U increase (OR 1.5, 95 percent CI 1.2-1.9).
The Association of Cardiac Troponin I With Cardiovascular Mortality Among Patients on Dialysis
Overview
We identified 13 studies that reported the association of cardiac troponin I with cardiovascular-specific mortality.11, 65, 70, 88, 103, 112, 113, 119, 135, 138, 139, 151, 162
We excluded only one study from meta-analysis of both HRs and ORs due to insufficient data. We presented a summary of these inclusion and exclusion reasons in Appendix E, Table KQ4.
Followup time ranged from 1 to 4 years.
Hazard Ratio for Cardiovascular-Specific Mortality Associated With Elevated Cardiac Troponin I
We included three studies that adjusted for at least age and CAD or risk equivalent in the meta-analysis for HR (Figure 13). The pooled risk of the association of elevated cardiac troponin I with cardiovascular mortality by was 4.2 (95% CI, 2.0-9.2). Confidence intervals were wide, but there was not any significant heterogeneity among the studies (I-squared = 0%).
Odds Ratio for Cardiovascular-Specific Mortality Associated With Elevated Cardiac Troponin I
Nine studies reported the number of events in each group and we included them in our meta-analysis. Two studies103, 151 had very unusual odds ratios (OR 58 and 0.6, respectively). Both studies had zero events in one of the groups, and the Stata statistical program added 0.5 to 0 cells for calculations.
The overall pooled OR showed a 5-fold increased risk (OR, 5.2; 95% CI, 2.8 to 9.0),which was similar to results for elevated cardiac troponin T elevation (Figure 14). Heterogeneity I-squared was 0 percent.
Figure 14
Pooled odds ratio of the association of an elevated troponin I with cardiovascular mortality among patients on dialysis. * Study used a troponin assay manufactured by Beckman. † Study used a troponin assay manufactured by Dade.
One study135 used a very high cardiac troponin I cut-point of 2.3 mcg/L. In a sensitivity analysis excluding that study, the estimated risk was similar (OR, 4.5; 2.0 to 9.9).
The Association of High-Sensitivity Cardiac Troponin I With Cardiovascular Mortality Among Patients on Dialysis
We did not identify any studies that reported an association with a high-sensitivity troponin I assay and cardiovascular mortality among dialysis only patients.
The Association of Cardiac Troponin T With Major Adverse Cardiovascular Events Among Patients on Dialysis
Overview
Twelve studies reported results of the association of cardiac troponin T with MACE.26, 87, 91, 96-98, 107, 128, 142, 143, 146, 151.
We included all studies in our meta-analysis. We outlined the overview of inclusion/exclusion in Appendix E, Table 5. The followup time ranged from 0.3 to 5 years.
Hazard Ratio for Major Adverse Cardiovascular Events Associated With Elevated Cardiac Troponin T
Only one study presented a HR adjusted for at least age and age/CAD risk-equivalent (HR, 1.90; 95% CI 1.02 to 3.4) (Figure 15). One study adjusted for age but not CAD. One study107 only presented an adjusted HR per 0.01 mcg/L increase in cardiac troponin T as a continuous variable, rather than a cut-point. Therefore a meta-analysis could not be performed.
Odds Ratio for Major Adverse Cardiovascular Events Associated with Elevated Cardiac Troponin T
Nine studies provided results for number of events in each group making it possible to calculate an unadjusted OR. One study169 only presented an adjusted OR.
We listed the pooled meta-analysis in Figure 16, with an estimated 6-fold risk of MACE for elevated cardiac troponin T (OR, 6.0; 95% CI, 3.5 to 12.0) without significant heterogeneity (I-squared = 35 percent). In a sensitivity analysis including the study with an adjusted OR, the pooled meta-analysis association was slightly lower but still significant (OR, 5.1, 95% CI, 2.9 to 8.9). When we sorted results by year, it did appear that odds ratios were generally progressively larger in magnitude with more current years.
The Association of High-Sensitivity Cardiac Troponin T With Major Adverse Cardiovascular Events Among Patients on Dialysis
We did not identify any studies reporting an association of high-sensitivity cardiac troponin T assay with MACE among dialysis patients.
The Association of Cardiac Troponin I With Major Adverse Cardiovascular Events Among Patients on Dialysis
Overview
We identified 12 studies that reported an association of cardiac troponin I with MACE. We outlined these in Appendix E, Table 6.70, 93, 96, 103, 120, 123, 128, 129, 142, 143, 146, 151
Hazard Ratio for Major Adverse Cardiovascular Events Associated with Elevated Cardiac Troponin I
No study presented results for the association of cardiac troponin I with MACE using HRs.
Odds Ratio for Major Adverse Cardiovascular Events Associated with Elevated Cardiac Troponin I
The pooled meta-analysis, including all nine relevant studies (Figure 17), showed a greater than 6-fold association of troponin I with MACE (OR, 6.3; 95% CI, 3.5 to 13.2) without heterogeneity. We could not include Katerinis et al.93 in the meta-analysis because of zero events, and the inability to generate a log OR. Several studies were small with few events and large CIs; thus, there were widely ranging effect sizes—from OR of 0.7 to 87.0. Heterogeneity I-squared was 0 percent. One study reported an unadjusted OR but not the number of events, and two studies had qualitatively different descriptions of a troponin elevation. We performed sensitivity analyses as described below.
Results were similar in a sensitivity analysis including only the five studies that reported the number of events in each arm (so that unadjusted OR could be determined).96, 103, 142, 151 The pooled meta-analysis showed a three-fold association of troponin I with MACE with at least 1 year followup (OR, 3.8; 95% CI, 1.6 to 8.9). Heterogeneity I-squared was 6.9 percent.
In another sensitivity meta-analysis of six studies, which additionally included the study by Hung et al.120 and presented an unadjusted OR but not number of events, the results were similar (OR, 4.3; 95% CI, 2.1 to 8.9). Heterogeneity I-squared was 0 percent.
Finally, we performed an additional sensitivity analysis including two additional studies that had qualitatively different assessments of troponin I rather than a single baseline value. For Katerinis et al.,93 an elevated troponin included only those with troponin elevated for more than 3 months. For Beciani et al.,129 an elevated troponin included those with both consistent and variable elevated troponin levels. As mentioned above, Katerinis et al. had zero events and could not generate a log OR. The pooled meta-analysis was again similar (OR, 4.6; 95% CI, 2.4 to 8.7). Heterogeneity I-squared was 0 percent.
The Association of High-Sensitivity Cardiac Troponin I With Major Adverse Cardiovascular Events Among Patients on Dialysis
Assa et al.157 found a high-sensitivity troponin I per 10 ng/L increase to be associated with risk of adverse cardiac events [HR 1.21 (95 percent CI 1.06 – 1.38)].
The Association of Cardiac Troponin T or I With Outcomes Among Patients on Dialysis Other Than All-Cause Mortality, Cardiovascular Mortality, or Major Adverse Cardiovascular Events
Heart Failure
One study102 reported an approximate 3-fold increased risk for cardiovascular congestion (heart failure) for an elevated cardiac troponin T per 1 mcg/L increase in a multivariate model that also adjusted for age, left ventricular mass, and ejection fraction (HR, 3.0; 95% CI, 1.2 to 7.4). This evaluated troponin T on a continuous scale, not a cut-point.
Hospital Admissions
Another study11 did not find that dialysis patients with elevated troponin I (>0.03 mcg/L) had increased risk of hospital admissions for any cause or cardiac cause over a 2-year time period (P not significant).
Subsequent Acute Coronary Syndrome
Troyanov et al.109 evaluated risk of first ACS event. Both elevated cardiac troponin T and I predicted risk of ACS over a 3-year followup. For elevated cardiac troponin T (> 0.04 mcg/L; Roche Elecsys), the HR was 3.0 (95% CI, 1.0 to 8.6). For elevated cardiac troponin I (>0.3 mcg/L; Abbott AxSym), the HR was 3.4 (95% CI, 1.6 to 7.3). Both had similar areas under the curve for predicting ACS events at 1.5 years (0.73 vs. 0.77 for cardiac Troponin T and I, respectively).
KQ 4.3. Troponin Associations With Short- and Long-Term Outcomes by Subgroups
We presented results for dialysis, nondialysis, and kidney transplant subgroups of CKD patients separately as indicated in previous sections. Regarding dialysis-only cohorts, few studies stratified by other subgroups. Studies were too few to generate meta-analyses for subgroup type. Subgroups described were as follows:
KQ 4.4. Comparisons Between Troponin Assays To Predict Risk
While many studies evaluated multiple troponin assays in the same population (troponin T vs. troponin I, or multiple troponin I assays by different manufacturers compared with each other), they presented no formal interaction testing. They never included Troponin T and I levels in the same multivariate model adjusted for the other cardiac biomarker. Some studies hinted at a stronger association with troponin T than with troponin I among dialysis patients. However, in our pooled meta-analyses, the effect sizes of the association of adverse events for elevated cardiac troponin were similar for both T and I overall. Therefore, we are unable to draw any specific conclusion about which biomarker is better in the CKD patient. Both cardiac troponin markers T and I were similarly associated with an increased risk for adverse outcomes.
Strength of Evidence Among Patients on Dialysis
Tables 34 and 35 describe our strength of evidence grading for KQ4 among patients on dialysis.
Table 34
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of risk stratification among patients on dialysis: Strength of evidence domains.
Table 35
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of risk stratification among patients on dialysis: Details regarding strength of evidence domains.
Results for Nondialysis Chronic Kidney Disease Patients
Of the publications meeting criteria for Key Question (KQ) 4, 26 included nondialysis chronic kidney disease (CKD) patients as part or all of the study population.25, 68, 69, 71, 73, 78, 79, 82, 84, 90, 94, 99, 100, 105, 108, 115, 121, 125, 132, 147, 148, 154, 156
We described the results for those that analyzed a pre- or post-kidney transplantation population separately and included them with the results for KQ4.3.
KQ 4.2b. Troponin Associations With Short- and Long-Term Outcomes Among Nondialysis, Nontransplanted Chronic Kidney Disease Patients
Key Points
- Elevated troponin T and troponin I in nondialysis CKD patients predict all-cause mortality (strength of evidence: moderate). Pooled analysis from studies adjusted for at least age and CAD or risk equivalent found the following associations: (Troponin T: pooled HR, 3.1; 95% CI, 1.1 to 11.0. Troponin I: pooled HR 1.7; 95% CI 1.2 to 2.7.) Similar findings were seen for studies with less adjusted HRs and for pooled meta-analyses for the unadjusted ORs.
- Elevated troponin T is associated with an increased risk of composite cardiac outcome (MACE) in nondialysis CKD patients based on pooled analysis (pooled HR, 2.7; 95% CI, 1.1 to 7.6) (strength of evidence: moderate).
- Studies of MACE outcomes with elevated troponin I that included nondialysis patients also included dialysis patients, and ORs were not statistically significant (strength of evidence: insufficient).
- We identified no studies that examined high-sensitivity troponin I in asymptomatic, nondialysis CKD patients (strength of evidence: insufficient).
- The adjusted analyses in nondialysis CKD populations suggest that elevated high-sensitivity troponin T predicts adverse outcomes (strength of evidence: low).
The Association of Cardiac Troponin T With All-Cause Mortality Among Nondialysis Chronic Kidney Disease Patients
Troponin T was the most common troponin assay that studies analyzed in the nondialysis CKD population. Nine reports included an endpoint of all-cause mortality.79, 82, 99, 100, 108, 121, 125, 132, 154 Four reports were not included in pooled analysis due to inclusion of dialysis patients or troponin presented continuously. Two studies analyzed an identical population; therefore, we presented the results from the study reporting an adjusted analysis. We list results in Table 36.99, 108
Table 36
Summary of the associations of elevated troponin T with all-cause mortality in patients not on dialysis.
Two studies presented HR adjusted for at least age and CAD or risk equivalent. Four studies, each reporting a HR and CI, were similar enough to be included in a meta-analysis of HRs.99, 100 with pooled HR 3.4 (1.1 to 11.0) (Figure 18). Although we used the highest troponin T threshold value for the pooled analysis, one study using multiple cutoffs found a significant difference in mortality rate when it compared troponin T less than 0.03 mcg/L with values ranging from 0.03 to 0.09 mcg/L (HR, 4.3; 95% CI, 1.8 to 10.4, P < 0.001) and values greater than 0.1 mcg/L (HR, 5.5; 95% CI, 2.9 to 10.5, P < 0.001).100 One study presented an unadjusted analysis, which was not significant.121
A second pooled analysis included the five studies that presented ORs or numbers of events from which we could derive ORs (Figure 19).79, 99, 121, 125, 154 All results were unadjusted. Threshold values for troponin T ranged from 0.02 mcg/L to 0.1 mcg/L, although all used a troponin T assay from the same manufacturer. The pooled OR was significant and suggested that an elevated troponin T is a predictor of mortality in nondialysis CKD patients (OR, 3.0; 95% CI, 1.4 to 6.7). One study108 provided adjusted OR but not HR.
We did not include two reports of all-cause mortality in either pooled analysis due to the inclusion of dialysis patients. One of these found an elevated troponin T to be a predictor of all-cause mortality after adjustment (HR, 2.7; 95% CI, 1.1 to 11.0; P < 0.05),132 but the other reported a loss of significance when they adjusted data.82
A study by Lamb et al. compared two troponin T cutoff values and found sensitivity and specificity to be 67 and 62 percent, respectively, for a threshold of 0.01 mcg/L, and 51 and 80 percent, respectively, for a threshold of 0.03 mcg/L.99
The Association of Cardiac Troponin I With All-Cause Mortality Among Nondialysis Chronic Kidney Disease Patients
We found five studies that assessed troponin I with an outcome of all-cause mortality among nondialysis patients with CKD (Table 37).99, 108, 148, 154, 159 Two studies were used to perform a meta-analysis of HR adjusted for at least age and CAD or risk equivalent. The pooled HR was 1.73 (95% CI, 1.2 to 2.7) (Figure 20).
Table 37
Summary of the associations of elevated troponin I with all-cause mortality in patients not on dialysis.
One study108 provided adjusted OR but not HR. A small study of heart failure patients with CKD (n = 29) used a short-term followup period of 6 months and found no significant difference in mortality in an unadjusted analysis (OR, 1.4; 95% CI, 0.7 to 2.8).154 There was insufficient data to calculate a pooled OR.
One study identified troponin I as having a sensitivity of 60 percent and a specificity of 73 percent for death with an area under the curve of 0.75 (95% CI. 0.66 to 0.84, P < 0.001).99
Musso et al. studied a small cohort consisting of a combination of dialysis, nondialysis, and post-kidney transplant patients (n = 49), and therefore it was difficult to compare this study with the results from other analyses we presented here.148
The Association of Cardiac Troponin T With Major Adverse Cardiovascular Events Among Nondialysis Chronic Kidney Disease Patients
Nine studies evaluated Troponin T in the context of cardiac mortality and MACE outcomes (Table 38).25, 68, 69, 73, 82, 100, 125, 147, 148
Table 38
Summary of the associations of elevated troponin T with major adverse cardiovascular events in patients not on dialysis.
We pooled four comparable studies that adjusted for at least age and CAD or risk equivalent in an analysis of HRs (Figure 21).68, 69, 73, 100. Threshold values for troponin T ranged from 0.01 mcg/L to 0.1 mcg/L. When Hasegawa et al. separated the high-sensitivity troponin T values into four ranges, only the highest cutoff value of 0.033 mcg/L remained significant (HR, 6.2; 95% CI, 1.4 to 27.7).68. We used the highest cutpoint in our meta-analysis. The result of this pooled analysis was statistically significant (HR, 2.7; 95% CI, 1.1 to 7.6). One additional study presented an unadjusted HR.25
We did not include two studies with a MACE outcome in this meta-analysis because of inclusion of dialysis patients.147, 148 Neither of these found a significant association between elevated troponin T and MACE.
Two studies analyzed cardiac mortality; however, these results are difficult to compare as one study included both dialysis and nondialysis patients,82 and the other was comprised of predialysis patients, many of whom began dialysis during the followup period.125 Neither of these found troponin T to be a predictor of MACE in asymptomatic nondialysis patients.
The Association of Cardiac Troponin I With Major Adverse Cardiovascular Events Among Nondialysis Chronic Kidney Disease Patients
Two studies assessed the association with troponin I and composite MACE (Figure 22). Both studies combined dialysis and nondialysis patients in a small cohort (n = 49 and 40, respectively). One had a followup period of 18 months, and the other 9 months. The latter used two troponin I assays with different cutoff values (0.35 mcg/L for Dade Stratus, and 1.6 mcg/L for Behring OPUS Plus). Results were insignificant for both, despite different rates of elevated and nonelevated troponins within the population. Although results were not statistically significant, the study designs made it impossible to draw the conclusion that troponin I does not predict MACE in this population.147, 148
The Association of High-Sensitivity Troponin T With Risk Among Nondialysis Chronic Kidney Disease Patients
Quiroga et al.,156 using a sensitive cutpoint of troponin T >0.01 ng/L (0.00001 mcg/L), found that elevated troponin T was associated with a 2-fold increase risk of cardiovascular event (unadjusted OR, 2.08; 95 percent CI 1.03-4.16) (Figure 23).
The Association of High-Sensitivity Troponin I With Risk Among Nondialysis Chronic Kidney Disease Patients
No studies meeting criteria for KQ4 addressed high-sensitivity troponin I assays.
KQ 4.3b. Troponin Associations With Short- and Long-Term Outcomes by Subgroups of Nondialysis Patients
We presented results for dialysis patients and nondialysis (nontransplanted) CKD patients above in the respective sections.
We found some additional subgroup analyses investigating troponin associations in pre- and post-kidney transplant patients as follows:
Key Points
- We did not identify any studies that analyzed troponin I in pre-kidney transplant patients (strength of evidence: insufficient).
- In pre-kidney transplant populations, data suggested that elevated troponin T values are predictors of adverse outcomes. These studies included both dialysis and nondialysis patients (strength of evidence: moderate).
- Elevations in both troponin I and T are likely predictors of adverse outcomes in the post-kidney transplant period (strength of evidence: low).
- In nondialysis CKD patients with a history of CAD, an elevated troponin I is a predictor of adverse cardiac event (strength of evidence: low).
- Studies did not assess subgroups by age, sex, ethnicity, and comorbidities other than CAD in the asymptomatic, nondialysis CKD population (strength of evidence: insufficient).
Pre-Transplantation
We identified three reports of end-stage renal disease (ESRD) patients referred for kidney transplantation, some of whom had been on dialysis and some of whom had not.84, 90, 105 All of these evaluated troponin T (Table 39). Two studies by the same author considered a group of 644 ESRD patients with troponin T values that were measured upon referral for kidney transplant. The studies presented results for the entire population, regardless of whether the patient went on to receive transplantation. During a mean followup of 11.5 months, elevated troponin T of greater than 0.01 mcg/L was associated with death in a model adjusting for sex, age, albumin, history of stroke, body mass index, smoking status, cholesterol, hemoglobin, and time on dialysis (HR, 1.6; 95% CI, 1.1 to 2.5, P = 0.022).90
Table 39
Summary of the association with risk of elevated troponin T in pre-kidney transplantation populations.
In a subsequent study of only patients who underwent kidney transplantation, pre-transplant elevated troponin T of at least 0.01 mcg/L was associated with composite MACE (AMI, revascularization, peripheral vascular intervention, or stroke) during a mean followup period of 28.4 months. The study observed this association in a model adjusted for age, time on dialysis, ejection fraction, and delayed graft functioning (HR, 1.6; 95% CI, 1.1 to 2.2, P = 0.008).84
In a study of 117 patients, Sharma et al. found a troponin T of greater than 0.06 mcg/L to be associated with all-cause mortality in a 3-year followup (OR, 7.1; 95% CI, 5.7 to 10.2, P = 0.004), though results were not adjusted. The associated area under the curve was 0.82 (95% CI, 0.64 to 0.99; P =0.02), with a sensitivity of 75 percent and a specificity of 72 percent.105
Post-Transplantation
In the studies of post-kidney transplantation populations, three evaluated troponin I71, 78, 115 and one evaluated troponin T.94
Troponin I
We describe the results for studies of troponin I in Table 40. A cohort of 34 dialysis patients with troponin I measured prior to and following renal transplantation found that 47.1 percent of the patients had an increase in troponin I value after surgery as compared with pre-surgery levels, although none exceeded the cutoff value of 2.3 mcg/L. The study followed patients for 22 months, and none experienced cardiac events or died.115
Table 40
Summary of the association of elevated troponin I with risk in post-kidney transplantation populations.
Another study considering postoperative troponin I values following kidney transplant used a threshold value of 0.04 mcg/L. This reported in-hospital acute myocardial infarction (AMI), 1-year all-cause mortality, and 1-year coronary revascularization. Of 376 in-hospital patients, the study observed AMI in 6.3 percent of those with elevated troponin I, but did not observe AMI in patients with a nonelevated value (P < 0.001). Rates of in-hospital death and revascularization were not significant. At 1-year followup, the difference in mortality between the two groups was not significant, and the rate of revascularization (percutaneous coronary intervention or coronary artery bypass graft) was marginally significant at 5.3 percent of those in the elevated troponin I group compared with 1.4 percent of those in the nonelevated troponin I group (P = 0.49); however, neither percutaneous coronary intervention or coronary artery bypass graft was significant when assessed alone.71
A similar study of 331 post-kidney transplantation patients used a higher cutoff value of 0.07 mcg/L. The study defined MACE as AMI, revascularization, or death due to an ischemic event and reported after a 3-month followup. The study noted a significantly lower rate of outcome in those with a nonelevated troponin I when adjusted for a history of CAD (OR, 0.1; 95% CI, 0.03 to 0.4) or age (OR, 0.1; 95% CI 0.03 to 0.3).78
Troponin T
We listed the results of troponin T studies in post-kidney transplantation populations in Table 41. A study of 372 patients, who had received kidney transplant in the past 3 months, used troponin T measurements with a cutoff level of 0.03 mcg/L to analyze outcomes during a maximum followup period of 1,626 days. They found a higher rate of all-cause mortality in those with an elevated troponin T (57.1 percent) versus a nonelevated test (14.0 percent) (P < 0.001). The study found a similar result for an outcome of cardiac mortality (33.3 vs. 4.8 percent, P < 0.001). In a model adjusted for age, sex, smoking history, diabetes, blood pressure, cholesterol, body mass index, and blood biochemical levels, troponin T remained significantly associated with all-cause mortality (Exp(β) 2.7; 95% CI, 1.2 to 6.1, P < 0.001).94
Table 41
Summary of the association of elevated troponin T with risk in post-kidney transplantation populations.
Other Subgroups
In a subgroup of post-kidney transplantation patients (n = 78) with a history of CAD, Claes et al. found an increased risk of MACE for every 0.01 mcg/L increase in troponin I in an adjusted analysis (OR, 1.2; 95% CI, 1.0 to 1.4, P = 0.038).78
No other studies performed subgroup analysis in nondialysis populations.
Strength of Evidence (Nondialysis Chronic Kidney Disease Patients)
Tables 42 and 43 describe our strength of evidence grading for KQ4 among nondialysis patients. Tables 44 and 45 describe our strength of evidence grading for KQ4 among subgroups of nondialysis patients.
Table 42
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of risk stratification among nondialysis patients: Strength of evidence domains.
Table 43
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of risk stratification among nondialysis patients: Details regarding strength of evidence domains.
Table 44
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of risk stratification among subgroups of nondialysis patients: Strength of evidence domains.
Table 45
Association of elevated troponin T or I versus nonelevated troponin T or I in terms of risk stratification among subgroups of nondialysis patients: Details regarding strength of evidence domains.
- Search Results
- Use of Troponin for Diagnosis of Acute Coronary Syndrome Among Chronic Kidney Disease Patients
- Management of Acute Coronary Syndrome by Troponin Levels
- Short- and Long-Term Prognosis After Presentation with Acute Coronary Syndrome by Troponin Levels
- Use of Troponin for Risk Stratification Among Chronic Kidney Disease Patients Without Acute Coronary Syndrome
- Results - Cardiac Troponins Used as Diagnostic and Prognostic Tests in Patients ...Results - Cardiac Troponins Used as Diagnostic and Prognostic Tests in Patients With Kidney Disease
Your browsing activity is empty.
Activity recording is turned off.
See more...