Early and adequate risk stratification is essential in patients with suspected acute coronary syndrome (ACS). The aim of the present study was to investigate whether glycogen phosphorylase BB (GPBB) could add prognostic information in the context of contemporary sensitive troponin I determination and B-type natriuretic peptide (BNP). Patients with suspected ACS were consecutively enrolled at 3 German study centers from January 2007 through December 2008. Troponin I, GPBB, and BNP were determined at admission. Follow-up information on the combined end point of death, myocardial infarction, revascularization, and hospitalization owing to a cardiovascular cause was obtained 6 months after enrollment. In total 1,818 patients (66% men) were enrolled of whom 413 (23%) were diagnosed as having acute myocardial infarction and 240 (13%) as having unstable angina pectoris, whereas in 1,165 patients (64%) an ACS could be excluded. Follow-up information was available in 98% of patients; 203 events were registered. GPBB measured on admission predicted an unfavorable outcome with a hazard ratio of 1.24 (p <0.05) in an unadjusted Cox regression model and showed a tendency with a hazard ratio of 1.13 (p = 0.07) in a fully adjusted model. Kaplan–Meier analysis revealed a poorer outcome in patients with increased GPBB levels amendatory to the information provided by troponin I or BNP. In conclusion, GPBB measurement provides predictive information on midterm prognosis in patients with chest pain in addition to BNP and troponin I.
To achieve a sufficient medical and interventional treatment of patients with suspected acute coronary syndrome (ACS), early and accurate diagnosis and risk stratification is essential. Glycogen phosphorylase BB (GPBB) has been discussed in recent years as a potentially useful biomarker in the early detection of ischemia. GP is an enzyme of cellular glucose metabolism found in the sarcoplasmic reticulum, where it catalyses the first step of glycogenolysis. GP occurs in various tissues but can be classified in 3 isoforms. These are GPMM (muscle), found in the skeletal and heart muscles; GPLL (liver), found initially in the liver tissue but is present ubiquitously except in the brain, skeletal muscle, and heart myocytes; and GPBB (brain), localized in relevant amounts solely in brain tissue and heart myocytes. GPBB is activated in conditions of ischemia and then released from its binding to glycogen molecules. Release of soluble GPBB into the circulation occurs rapidly with plasma levels increasing within the first hours after chest pain onset. Assuming an intact blood–brain barrier, this pattern of early release can be used as a diagnostic tool. Previous studies on GPBB in diagnosis are rare and have produced inconsistent results, although none of these studies has compared GPBB to a more sensitive troponin assay. Regarding the prognostic information of GPBB for use in the risk stratification of patients with suspected ACS, valid data are virtually absent. The aim of the present study was to investigate whether GPBB could add prognostic information in the context of contemporary sensitive troponin I determination and in the context of the established risk predictor B-type natriuretic peptide (BNP) in a large real-world cohort of patients with chest pain.
Methods
Patients with suspected ACS presenting consecutively to the chest pain unit of 1 of 3 recruiting centers (University Medical Center of Johannes Gutenberg University, Mainz; Federal Armed Hospital, Koblenz; or University Heart Center, Hamburg) from January 2007 through December 2008 were enrolled in this all-comers study as described previously. Participation was voluntary; each patient provided written informed consent. The study was authorized by the local ethics committees in the federal states of Rheinland-Pfalz and Hamburg and was performed according to the principles of the Declaration of Helsinki.
After admission all patients underwent a clinical assessment procedure including physical examination, medical history taking, noninvasive blood pressure, and pulse oxymetry. For diagnostic reasons, 12-lead electrocardiography was performed on admission and after 3 and 6 hours. Interpretation of all electrocardiograms for the study was carried out blinded to patient’s characteristics. Blood samples were collected at arrival and after 3 and 6 hours with the first blood draw before any invasive diagnostic procedure or treatment in all patients.
Two independent cardiologists blinded to GPBB values made the final diagnosis after a review of all available clinical findings and laboratory and imaging data. Acute myocardial infarction (AMI) was diagnosed according to the universal definition of MI. Myocardial necrosis was determined if ≥1 in-house troponin level assayed conventionally was increased above the corresponding cutoff in combination with a pattern of increasing or decreasing values of ≥20%. ST-segment elevation MI was diagnosed when increased troponin levels occurred in combination with relevant ST-segment elevation or new left bundle branch block documented on electrocardiogram at admission. Non–ST-segment elevation MI was diagnosed when increased troponin levels with an increasing or decreasing pattern were observed in addition to a nondiagnostic electrocardiogram. Unstable angina pectoris was diagnosed from the findings of negative troponin levels, nondiagnostic electrocardiogram, proved ischemia by stress testing, and/or angiographic findings of a culprit lesion and need for revascularization. Noncoronary chest pain (NCCP) was diagnosed when an ACS could be excluded. Follow-up information was obtained by standardized telephone interview and hospital or general practitioner charts. Outcome measurements were cardiovascular events as the combined end point of first occurrence of death, MI, revascularization, or hospitalization owing to cardiovascular causes within 6 months after initial enrollment.
All blood samples were processed in the hospital central laboratories and routine laboratory measurements including conventional in-house troponin determination were performed immediately. In addition, ethylenediaminetetraacetic acid plasma, citrate plasma, and serum samples were frozen at −80°C after centrifugation for further measurements. Analytic determination of evaluated plasma troponin I, BNP, and GPBB levels for all study centers was performed at the biomarker laboratory in Mainz by experienced technical assistants blinded to a patient’s characteristics. For adjudication of the final diagnosis of AMI, troponin T was measured at 2 study centers and troponin I at 1 study center as in-house troponin using conventional assays as described previously. In addition, troponin I was determined using a contemporary sensitive assay fulfilling the recommended imprecision criteria that were used for all further analyses presented in this article on the ADVIA Centaur XP system (TnI-Ultra; Siemens Healthcare Diagnostics, Erlangen, Germany). The assay detection limit was 0.006 ng/ml, the measuring range was 0.006 to 50 ng/ml, the 99th percentile was 0.04 ng/ml, and the 10% coefficient of variation was 0.03 ng/ml. Plasma GPBB levels were determined using a commercially available enzyme-linked immunoassay test kit (Diacordon GPBB-ELISA; Diagenics, Woburn, Massachusetts). Measurements were performed on a Tecan Sunrise Absorbance Reader (Tecan, Crailsheim, Germany). The assay has an analytic detection limit of 3 ng/ml and allows precise measurement of GPBB in a range of 3 to 100 ng/ml. According to the manufacturer, the threshold for diagnosis of MI is 10 ng/ml. A reference cohort cutoff of 15.7 ng/ml based on the 95th percentile reflecting the upper reference limit was determined in a group of 520 nonhospitalized apparently healthy subjects according to acute states of cardiovascular diseases in a pilot study of the Gutenberg Health Study. Baseline characteristics of this cohort are given in the supplementary material (available online).
Continuous variables were described by median and interquartile range if skewed and by arithmetic mean ± SD if symmetrical. To assess associations of GPBB with continuous variables and binary factors, Spearman rank correlation coefficients and Mann–Whitney test, respectively, were used to compare the median of the respective group. Receiver operating characteristic curve analyses were performed for single and combined biomarkers; the latter was achieved by logistic regression using logarithmically transformed biomarkers. Kaplan–Meier curves and different Cox proportional hazards regression models were used to investigate GPBB as a predictor of the combined end point. GPBB, troponin I, and BNP were entered in the Cox regression model as continuous variables with an increase by SD as the criterion after logarithmic transformation. To adjust for possible confounders the Global Registry of Acute Coronary Events (GRACE) score variables including heart rate, creatinine, ST-segment changes on electrocardiogram, age, systolic blood pressure, and Killip class were used. Need for defibrillation, also a variable in the GRACE score, was not included in the model because it presented too few cases. For Kaplan–Meier survival analyzes, biomarkers were dichotomized. All analyses were done with R 2.13.0 (R Foundation for Statistical Computing, Vienna, Austria).
Results
In total 1,818 patients (66% men) with suspected ACS were enrolled, of whom 413 (23%) were diagnosed as having AMI and 240 (13%) as having unstable angina pectoris. In 1,165 patients (64%) an ACS could be excluded and was classified as NCCP. Median times from chest pain onset to admission were 4.2 hours in patients with NCCP, 4.6 hours in those with unstable angina pectoris, and 4.3 hours in those with AMI. GPBB values were available in 1,679, troponin I in 1,789, and BNP in 1,781 patients. Patients with AMI had higher GPBB levels (5.74 ng/ml) compared to patients with NCCP (4.65 ng/ml, p <0.0001) or those with unstable angina pectoris (4.85 ng/ml, p <0.0001). Data on baseline characteristics according to diagnosis and on the time course of evaluated biomarkers after admission stratified by diagnosis are provided in the supplementary material . As published, application of the contemporary sensitive troponin I assay in discrimination of AMI from NCCP yielded an area under the receiver operating characteristic curve of 0.960, whereas GPBB yielded an area under the curve of 0.625. If restricting analyses to patients presenting within the first 3 hours after chest pain onset, troponin I yielded an area under the curve of 0.947 and GPBB of 0.647. The combination of GPBB and troponin I did improve the area under the curve of troponin I alone slightly to 0.963 in the overall cohort (p = 0.51) and 0.955 in patients presenting within 3 hours after chest pain onset (p = 0.46).
The influence of traditional categorical risk factors and of continuous variables including biomarkers on GPBB levels stratified by diagnosis is presented in Table 1 . In patients with exclusion of ACS, male gender, body mass index, waist-to-hip ratio, and prevalent diabetes mellitus were associated with higher GPBB levels. These associations lost their significance in patients with ACS. As expected, the strongest, still moderate, correlation (r = 0.15) was observed between GPBB and troponin I in patients with AMI. Interestingly, age and lipid status showed an association with GPBB only in patients with ACS without evident myocardial necrosis (unstable angina pectoris). Correlations of risk factors and GPBB levels in the evaluated reference population of 520 apparently healthy subjects are given in the supplementary material .
| Variable | NCCP | p Value | UAP | p Value | AMI | p Value |
|---|---|---|---|---|---|---|
| (n = 1,165) | (n = 240) | (n = 413) | ||||
| Levels of glycogen phosphorylase BB according to diagnosis and binary phenotypes | ||||||
| Men | 4.73 | 0.024 | 4.96 | 0.21 | 5.86 | 0.10 |
| Women | 4.44 | 4.64 | 5.14 | |||
| Body mass index (kg/m 2 ) | 0.30 | |||||
| >30 | 4.89 | <0.001 | 4.97 | 0.87 | 5.82 | |
| ≤30 | 4.56 | 4.82 | 5.66 | |||
| Diabetes mellitus | 0.28 | |||||
| Yes | 5.08 | 0.022 | 5.44 | 0.051 | 5.85 | |
| No | 4.62 | 4.84 | 5.70 | |||
| Current smoker | 0.60 | |||||
| Yes | 4.71 | 0.73 | 4.87 | 0.12 | 5.84 | |
| No | 4.64 | 4.82 | 5.69 | |||
| Hyperlipidemia | 0.89 | |||||
| Yes | 4.64 | 0.98 | 4.73 | 0.38 | 5.77 | |
| No | 4.68 | 5.13 | 5.64 | |||
| Known coronary artery disease ⁎ | 0.57 | |||||
| Yes | 4.65 | 0.45 | 4.95 | 0.72 | 5.64 | |
| No | 4.64 | 4.74 | 5.84 | |||
| Spearman correlations with continuous variables | ||||||
| Age | 0.00 | 0.97 | −0.15 | 0.024 | 0.05 | 0.33 |
| Low-density/high-density lipoprotein ratio | 0.01 | 0.68 | 0.15 | 0.038 | −0.09 | 0.087 |
| Waist-to-hip ratio | 0.08 | 0.045 | 0.13 | 0.14 | −0.01 | 0.87 |
| Glomerular filtration rate | −0.01 | 0.68 | 0.04 | 0.57 | 0.05 | 0.36 |
| Troponin I | 0.05 | 0.084 | −0.02 | 0.75 | 0.15 | 0.0040 |
| B-type natriuretic peptide | 0.06 | 0.040 | −0.04 | 0.56 | 0.00 | 0.98 |
⁎ Known coronary artery disease before the current episode of chest pain based on history taking and all available clinical information.
Within the 6-month follow-up period 203 events as first occurrences of the combined end point were registered. Follow-up data were available in 98% of patients (183-day median follow-up time as computed by Kaplan–Meier potential follow-up method). After 6 months 33 patients died, 28 developed nonfatal MI, 64 required percutaneous coronary intervention, and 182 were hospitalized from cardiovascular causes. Table 2 lists baseline characteristics of the study population according to event status. Patients with an event had significant higher median GPBB levels compared to patients with event-free survival (5.48 vs 4.76 ng/ml, p = 0.000 82) and compared to healthy controls (5.48 vs 4.19 ng/ml, p <0.001; Figure 1 ). Kaplan–Meier survival analyses showed an association between increased GPBB levels on admission and poorer 6-month outcome. An event was observed in 8.9% of patients in the lowest tertile, 9.6% of patients in the second tertile, and 15.2% of patients in the highest tertile. Increased GPBB levels provided additional information in the identification of high-risk patients amendatory to troponin I and might improve defining low-risk patients complementary to BNP determination as displayed in Figure 2 . In an unadjusted Cox regression model increased GPBB levels predicted an unfavorable 6-month outcome with a hazard ratio of 1.24 (p = 0.0003) comparable to troponin I and BNP with hazard ratios of 1.45 (p <0.0001) and 1.69 (p <0.0001), respectively. In a fully adjusted model using the GRACE score variables as established parameters in risk stratification of patients with ACS, GPBB increase showed a tendency with a hazard ratio of 1.13 (p = 0.067) to predict outcome ( Table 3 ). For the performed serial measurements, GPBB determined 3 or 6 hours after admission did not provide relevant more prognostic information than GPBB measured directly on admission ( supplementary material ).
