Neurol Sci (2011) 32:133138 DOI 10.

1007/s10072-010-0457-4

ORIGINAL ARTICLE

The usefulness of albumin-adjusted ischemia-modied albumin index as early detecting marker for ischemic stroke
Jung Hwan Ahn Sang Cheon Choi Wee Gyo Lee Yoon Seok Jung

Received: 31 January 2008 / Accepted: 17 November 2010 / Published online: 11 December 2010 Springer-Verlag 2010

Abstract This study was conducted to investigate whether albumin-adjusted ischemia-modied albumin index (IMA index) is more sensitive and accurate than ischemiamodied albumin (IMA) as early detection marker of ischemic stroke, and to compare IMA and IMA index in progression and non-progression of ischemic stroke. This casecontrol study was done at an emergency medical center of a university hospital. 52 patients with neurological symptoms were enrolled (28 Ischemic Stroke Group and 24 Non-Stroke Group). In the ROC analysis of IMA index to diagnose stroke, area under the curve (AUC) was 0.990 (cutoff value 91.4; 95% CI: 0.9701.000; sensitivity: 95.8%; specicity: 96.4%). The AUC for IMA value was 0.928 (cutoff value 98 U/ml; 95% CI: 0.8570.999; sensitivity: 87.5%; specicity: 89.3%). The difference between progression (n = 12) and non-progression group (n = 16) in IMA and IMA index were statistically insignicant (p [ 0.01). IMA index was more sensitive than conventional IMA value as diagnostic biomarker of stroke, however, arguable as a predictive biomarker for progression of ischemic stroke. Keyword Ischemia Ischemia-modied albumin Ischemic stroke Serum albumin-adjusted IMA index

Introduction Acute ischemic stroke with typical symptoms and signs such as motor weakness is easily detected and it is treated with reperfusion therapy. However, it is often difcult to diagnose ischemic stroke in patients with vague neurological symptoms and signs such as vertigo [1, 2]. In this disease entity, the biomarker of ischemic stroke could be useful for decision making of acute ischemic stroke. In the two previous studies, time-dependent conventional ischemia-modied albumin (IMA) values were found to be elevated and a marker capable of reecting cerebrovascular accidents [3, 4]. IMA has been reported to play a rule-out role in selection of patients with acute coronary syndrome, through a negative result in the triple predictive test [5, 6]. IMA is a modied protein, generated under ischemic conditions with a decreased ability to bind cobalt. IMA is measured by albumin cobalt binding (ACB) test as an indirect measurement method. In this measurement method, the cobalt that does not bind to albumin is measured, therefore, IMA would be inuenced by the albumin concentration in the patients serum. Albumin-adjusted ischemia-modied albumin index (IMA index) has recently been introduced as a new marker that compensates the effect of serum albumin for screening of acute coronary syndrome and reported to be more sensitive and accurate than the conventional IMA value in the patients with acute coronary syndrome [7]. In this study, the usefulness of IMA index for detecting ischemic stroke was investigated, and compared it with conventional IMA value in regard to the accuracy to detect ischemic stroke. Moreover, the conventional IMA value and IMA index according to time after symptom onset were compared between progression and non-progression

J. H. Ahn S. C. Choi Y. S. Jung (&) Emergency Department, Ajou University School of Medicine, Woncheon-dong, Yeongtong-gu, Suwon 443-721, Republic of Korea e-mail: erdrajh@naver.com W. G. Lee Department of Clinicopathology, Ajou University School of Medicine, Woncheon-dong, Yeongtong-gu, Suwon 443-721, Republic of Korea

123

134

Neurol Sci (2011) 32:133138

stroke group to validate the possibility as a predictor of progression of ischemic stroke. Here, the progression group was dened by an increase in the National Institutes of Health Stroke Scale (NIHSS).

Patients with Troponin T and Electrocardiogram (EKG) abnormalities which suggested acute coronary syndrome, at admission were also excluded. Measurement

Materials and methods Study design This study was conducted prospectively over 3 months from September 1, 2006 at an emergency medical center of a university afliated hospital, after obtaining the approval by the Institutional Review Board of Ajou University Hospital (Suwon, South Korea). Enrolled patients The enrolled patients presented apparent brain dysfunction of abrupt onset: altering level of consciousness, dysarthria, aphasia, motor weakness of extremity, sensory decit, cranial nerve decit, limb and gait ataxia, and spinning sensation (vertigo). Fifty-two patients who were admitted to emergency center with sudden neurological symptoms within 6 h after symptom onset were nally selected. All of the enrolled patients underwent brain angiography, computed tomography (CT) with contrast, and brain magnetic resonance imaging (MRI). Patients diagnosed with stroke by brain perfusion MRI and brain diffusion weighted MRI were classied as brain ischemic stroke group (SG; n = 28). In SG, we classied progression (n = 12) and non-progression (n = 16). The progression group was dened by an increase in NIHSS. In non-progression SG, four patients were treated by reperfusion therapy such as the use of thrombolytic agents (n = 3) and percutaneous transluminal recanalization (n = 1). These patients were included in statistical analysis as SG. However, due to small number of patients, subgroup analysis was not conducted. Instead, the mean IMA value and IMA index before and after reperfusion therapy in these patients were described in the Discussion. The patients without special ndings in imaging studies and with specic diagnosis were classied as non-stroke group (NSG, n = 24). The types of neurological disorders in the NSG were peripheral vertigo (n = 15), intoxication (n = 4), hypoglycemia (n = 2), basilar migraine (n = 1), vasovagal syncope (n = 1), and conversion disorder (n = 1). Exclusion criteria included: cases in which 6 h had already passed from symptom onset and those with a past history of brain hemorrhage, cardiac arrest, myocardial ischemia, renal failure, and major or multiple surgeries. Patients with pulmonary embolism, normal pregnancy, or who did not give consent for the IMA test were excluded.

Demographic, clinical, NIHSS and vascular risk factors were compared between SG and NSG group. In progression stroke group, NIHSS was measured at admission in the emergency medical center and after progression. In SG, IMA was measured at the time of admission, 12 and 24 h after symptom onset. In NSG, patients were discharged within 12 h of symptom onset. IMA and albumin values were thus measured only on admission. Serum samples for IMA were immediately centrifuged and stored at -20C for later analysis. IMA was measured using the albumincobalt binding (ACB) test (Ischemia Technology, Denver, USA) on a TBA-200FR analyzer (Toshiba Medical Systems Co., Ltd., Tokyo, Japan). Within-assay (n = 10) imprecision, as expressed by the coefcient of variation, was B2%. IMA index was calculated according to the following formula: IMA index = serum albumin concentration (g/dl) 9 23 ? IMA (U/ml) - 100. Urine HCG, cardiac enzyme (myoglobin, CK, CK-MB, and Troponin T) and EKG were checked because normal pregnancy and acute coronary syndrome were excluded. Data analysis Continuous data are expressed as mean standard deviation. Data of 52 patients were analyzed by the SPSS 12 statistics program (SPSS Inc., USA). For general validation of statistics between the two groups and for IMA comparison and IMA index, MannWhitney U test and Chisquare test were performed. A Fishers exact test was used to compare vascular risk proles of SG and NSG. To investigate the correlation between conventional IMA value and stroke severity, assessed by the NIHSS at admission, Spearmans correlation coefcient was used. It was also used to nd out the correlation between IMA index and stroke severity, assessed by the NIHSS at admission. Cases with p values less than 0.01 were determined as statistically signicant. Data and receiver operating characteristic (ROC) curve were analyzed using the SPSS 12 statistics program (SPSS Inc., USA) and General ? Clinical laboratory statistics version 1.73 (Analyse-it software, Ltd., England, UK). We constructed an ROC curve for IMA and IMA index to identify the optimal threshold by maximizing the sum of sensitivity and specicity. We compared the sensitivity and specicity of conventional IMA at the cutoff value with IMA index. Sensitivity, specicity, positive predictive value (PPV) and negative predictive value (NPV) were calculated for the optimum

123

Neurol Sci (2011) 32:133138

135

cutoff point of 80 U/ml as recommended in the previous pilot study, and for the manufacturers cutoff point of 85 U/ml. In SG, two-sample KolmogorovSmirnov Z test was performed to compare brain stroke patients with and without progression.

Results The overall mean age was 60.2 16.3 years with male/ female ratio of 1:1.4:59.3 16.5 years and 1:1.3 in SG and 61.3 16.2 years and 1:1.4 in NSG, respectively. There was no signicant difference in mean age or gender ratio between the groups (p = 0.660, p = 0.870). The mean time of admission of enrolled patients was 164.6 (164.6 98.1 min), and there was no signicant difference in the mean time of admission between SG (151.6 107.3) and NSG (177.5 86.0) (p = 0.316). The NIHSS in SG and NSG was 4.9 4.3 and 4.0 5.9, respectively, and there was no signicant difference (p = 0.053). There was no correlation between admission IMA levels and NIHSS as well as IMA index and NIHSS. The Spearman correlation coefcients were 0.250 and 0.296 in SG (p = 0.218, p = 0.143). There was no signicant difference in vascular risk between SG and NSG (Table 1). Table 2 shows that the mean IMA index at the time of admission was 104.4 9.4 and 68.1 11.5 in SG and NSG, whereas the mean IMA value was 107.4 11.0 and
Table 1 Vascular risk factors in each group Vascular risk factors Patients Stroke group (%) Vascular history Past history of stroke Family history of stroke Hypertension Diabetes Hypercholesterolemia Atrial brillation Smoke 14.3 14.3 3.6 42.9 17.9 7.1 10.7 35.7 Non-stroke group (%) 8.3 4.2 12.5 33.3 29.2 4.2 4.2 45.8 0.669 0.351 0.340 0.399 0.514 1.000 0.611 0.775 p value

Fig. 1 Distribution of IMA index for each group

86.5 8.5 U/ml in SG and NSG, respectively (Fig. 1). Thus, IMA index and IMA value at the time of admission were signicantly different between the two groups (p \ 0.001, p \ 0.001). In the ROC analysis of IMA index, AUC was 0.990 (95% CI: 0.9701.000). When ischemic stroke was diagnosed with a cutoff value of 91.4, the sensitivity, specicity, and NPV and PPV were 95.8, 96.4, 95.8 and 96.4%, respectively (Fig. 2). In the ROC analysis of conventional IMA value, the area under the curve (AUC) was 0.928

Table 2 IMA level of each group Patients Stroke group IMA (U/ml) IMA index 107.4 11.0 104.4 9.4 Non-Stroke group 86.5 8.5 68.1 11.5 \0.001 \0.001 p value Fig. 2 Receiver operating characteristic curves for IMA index (bold line) and conventional IMA value (dotted line) for diagnosis of stroke, with areas under the curves of 0.990 (95% CI 0.9701.000; best cutoff value 91.4; sensitivity 95.8%; specicity 96.4%) and 0.928 (95% CI 0.8570.999; best cutoff value 98 U/ml; sensitivity 87.5%; specicity 89.3%)

IMA Ischemia-modied albumin

123

136 Table 3 Time-dependent IMA and IMA index in the stroke group Progression Initial IMA (U/ml) 12 h IMA (U/ml) 24 h IMA (U/ml) Initial IMA index 12 h IMA index 24 h IMA index 108.5 11.9 105.8 12.4 110.3 12.4 102.8 12.1 96.4 9.4 102.6 9.1 Non-progression 106.1 10.0 104.6 16.0 108.8 8.7 105.6 7.0 101.1 11.7 102.6 10.8 p value 0.970 0.500 0.830 0.696 0.356 0.913

Neurol Sci (2011) 32:133138

IMA Ischemia-modied albumin

(95% CI: 0.8570.999). A cutoff value of 98 U/ml appeared to be optimal, yielding a sensitivity of 87.5%, specicity 89.3%, PPV 87.5% and NPV 89.3% (Fig. 2). Using the cutoff value of 85 U/ml recommended by manufacturer, the sensitivity of IMA for the diagnosis of stroke decreased to 45.8%, with a specicity of 96.4%, PPV 91.7% and NPV 67.5%. Using the recommended previous pilot study cutoff of 80 U/ml, the specicity of IMA for the diagnosis of stroke decreased to 25%, with a sensitivity of 100%, PPV 100% and NPV 60.9%. IMA index at the time of admission, 12 and 24 h after symptom onset were 104.4 9.4, 99.1 10.8, and 102.6 10.0, respectively (Table 3). IMA values at the time of admission, 12 and 24 h after symptom onset were 107.4 11.0, 105.2 13.9, and 109.5 10.8 U/ml, respectively (Table 3). The IMA index and IMA value according to each time point were not signicantly different between the progression (n = 12) and non-progression (n = 16) SG. The NIHSS were 5.2 4.3 and 7.7 4.5 before and after progression, respectively, showing signicant difference in NIHSS between before and after progression (p = 0.002). The NIHSS at 12 and 24 h after symptom onset were 5.8 3.7 and 7.3 3.3. The NIHSS in the SG without progression was 4.7 4.5 at admission.

Discussion Albumin consists of 585 amino acids with molecular weight of 66,500 and a terminus characteristic in each organism [5]. Exposure to certain conditions such as ischemia, reperfusion and oxidation, whereby free radicals are generated, modies circulating albumin at its N-terminus by different mechanisms [4, 713]. When the body becomes an acidic state such as ischemia, proteins that migrate along the ischemic area in the blood ow release divalent copper, and the copper ions (Cu2?) are scavenged by albumin through tight binding to N-terminus. Furthermore, the released divalent copper ions are reduced to monovalent copper ions in the presence of reducing substances such as ascorbic acid, and the monovalent copper

ions produced by this mechanism react with oxygen to generate superoxide free radicals, which are converted to hydrogen peroxide (H2O2) by superoxide dismutase. Under normal conditions in vivo, hydrogen peroxide is converted to water and oxygen by catalase. However, in the presence of monovalent copper ions, H2O2 is converted to hydroxyl free radicals, -OH, through a reaction known as the Fenton reaction [413]. These hydroxyl free radicals modify albumin, which is referred to as IMA, and release more divalent copper ions, thus generating more hydroxyl free radicals through the Fenton reaction. The resulting vicious cycle induces a sudden increase in IMA [413]. Therefore, ischemia can be conrmed by a rapid increase of modied albumin level. Several studies have revealed that free radical production increases in stroke, particularly with ischemia and reperfusion [14], suggesting that IMA can be generated in ischemic stroke. A recent pilot study showed signicant increase of IMA after stroke [3, 4]. In this study, although the number of cases was small, thus limiting the statistical power of the conclusions, IMA value in SG was signicantly increased in comparison with NSG. These results strongly indicate the values of IMA as a good biological marker to diagnose brain stroke relatively rapidly. However, additional studies are required for further conrmation. IMA is measured by albumin cobalt binding (ACB) test as an indirect measurement method. In this measurement method, the cobalt that does not bind to albumin is measured. Therefore, IMA would be inuenced by the albumin concentration in the patients serum. The range of serum albumin concentration varies among individuals and even within an individual due to various reasons. Therefore, the interpretation of IMA value should be considered according to serum albumin concentration of individuals. One study reported that IMA index designed as albuminadjusted marker is more sensitive and accurate than the conventional IMA value in patients with acute coronary syndrome [7]. In this study, IMA index was also found to be more sensitive and accurate than conventional IMA value in patients with ischemic stroke. The mean- and cutoff-values of IMA in this study, however, were higher than those in the previous study [7]. For the mean value, the previous study of IMA usefulness in stroke suggested 83 U/ml (7986 U/ml). However, the mean IMA values of 107.4 U/ml in our study was higher, and even our NSG was higher at 86.5 U/ml. The cutoff value was also higher than the value for myocardial ischemia suggested by the manufacturer as well as that of the previous study. Using the cutoff value recommended by manufacturer and the previous study, sensitivity, specicity and NPV showed little merit for the diagnosis of ischemic stroke. Furthermore, 70.8% of NSG had higher values than the cutoff value of 80 U/ml. Three reasons could be

123

Neurol Sci (2011) 32:133138

137

considered for the difference in cutoff values between the previous study [7] and ours. First, ethnic background: the mean IMA value from two studies, both based on healthy Asians, was 91.2 5.4 U/ml (range 77105 U/ml), which is a higher reference value than for non-Asians [15]. Second, variation of equipments in each institution: according to Abadie et al. [16], both the results and equipments utilized vary in each institution. Therefore, it has been recommended that the cutoff value is analyzed through an ideal ROC curve in each examination room. The last, the different timing of patient selection: in the previous study, patients admitted within 3 h after acute ischemic stroke were enrolled. In this study, however, the enrolled patients were admitted within 6 h. In acute coronary syndrome, IMA is increased to the near peak level within a few minutes after onset of myocardial ischemia and continues to gradually increase for 6 h and then declines to its baseline value within 1224 h [10, 13]. Therefore, if the characteristics of IMA in stroke are similar to the ones in acute coronary syndrome, there would be no signicant difference in IMA values between 3 and 6 h. In our present study, the mean time of admission of the enrolled patients in SG was 151.6 (151.6 107.3) min, and the number of patients admitted within 3 h after presenting neurologic symptom was 16. The mean time of admission of patients within 3 h was 72.8 55.8 min and within 6 h was 256.7 54.2 min. Statistical analysis by MannWhitney U test revealed no signicant difference in conventional IMA value and IMA index between the patients admitted within 3 h after an acute ischemic stroke and within 6 h (p = 0.780, p = 0.125). However, it is still unknown whether the characteristics of IMA in stroke are similar to the ones in acute coronary syndrome. No reliable conclusions could be made because of limitations on the characteristics of IMA and subgroup analysis, and the small number of cases. Abboud et al. [4] reported that conventional IMA value on admission correlates with NIHSS. In our present series, however, conventional IMA value and IMA index did not correlate with NIHSS. It should be recognized that the analyses in the above two studies are limited by the small number of cases. Therefore, the association between IMA and stroke severity should be conrmed with larger numbers of subjects. In the present study, we observed that IMA values and IMA index in SG were not signicantly different at any time point, differing from the IMA characteristics in myocardial ischemia. In myocardial ischemia, IMA was increased to near peak level within a few minutes, continued to increase for up to 6 h, and subsequently disappeared after 1224 h [10, 14]. On the other hand, time-dependent IMA characteristics were not observed in ischemic stroke. Abboud et al. [4] suggested a possibility of an increase in oxidative stress

due to oligemia in the ischemic penumbra, and that reperfusion might result in an increase of IMA concentrations within the rst 24 h of onset in ischemic stroke. The timedependent changes in IMA value and IMA index, shown in Table 3, revealed no signicant difference between the progression and non-progression groups, indicating that IMA could not be used as a biomarker for the progression of ischemic stroke. An interesting fact uncovered in this study is that the IMA value and IMA index of patients with reperfusion were decreased after reperfusion therapy. The reperfusion therapy was done within 6 h after initial symptoms, therefore, the mean IMA was measured at admission and 12 h after symptom onset. The times of measurement of IMA value in each patient were at 6, 8, 10, and 10 h after symptom onset. The mean IMA value before and after reperfusion therapy was 114.2 and 100.2, and the mean IMA index was 109.1 and 87.1, respectively. The mean NIHSS scores were also decreased from 9.7 to 9.2 after reperfusion therapy. Because of the use of thrombolytic agents in only three cases and percutaneous transluminal recanalization in only one case in SG, any generalized conclusions could not be derived. Nevertheless, the above four cases indicate that IMA value and IMA index are decreased after reperfusion therapy, similar to ischemic cardiac disease. In one patient with reperfusion therapy, the IMA value increased after the reperfusion therapy was contrary to other patients, however, IMA index in all patients with reperfusion therapy decreased after reperfusion therapy. IMA index measurements in stroke patients before and after reperfusion therapy appear to be a possible method of evaluating the effect of reperfusion therapy. However, additional studies are required to conrm this. The limitations of our study are as follows: rst, although the level of signicance was set at 0.01, the number of cases is small, thus limiting the statistical power of the conclusions. Second, the study is unable to state the differences of IMA characteristics in various ethnic backgrounds because it was carried-out in an ethnically homogenous country. Third, the NSG may not be pure, because transient ischemic attack might have been included. However, according to Abboud et al. [7], IMA values in transient ischemic attack are lower than those in stroke. Additional study on transient ischemic attack is necessary, since the previous study was a pilot nature. At last, the range of the brain ischemic stroke group was very broad. In future studies, the usefulness of conventional IMA value and IMA index as a biomarker in acute stroke should further be examined by classifying subdivided study groups in terms of pathophysiology, location of cerebral infarction. We suggest that studies on IMA should be conducted for various ethnic groups. Moreover, measurement of the level

123

138

Neurol Sci (2011) 32:133138 7. Lee YW, Kim HJ, Cho YH, Shin HB, Choi TY, Lee YK (2007) Application of albumin-adjusted ischemia modied albumin index as an early screening marker for acute coronary syndrome. Clin Chim Acta 384:2427 8. Apple FS, Quist HE, Otto AP, Mathews WE, Murakami MM (2002) Release characteristics of cardiac biomarkers and ischemia-modied albumin as measured by the albumin cobalt-binding test after a marathon race. Clin Chem 48:10971100 9. Christenson RH, Duh SH, Sanhai WR, Wu AH, Holtman V, Painter P, Branham E, Apple FS, Murakami M, Morris DL (2001) Characteristics of an Albumin Cobalt Binding Test for assessment of acute coronary syndrome patients: a multicenter study. Clin Chem 47:464470 10. Garrido IP, Roy D, Calvino R, Vazquez-Rodriguez JM, Aldama G, Cosin-Sales J, Quiles J, Gaze DC, Kaski JC (2004) Comparison of ischemia-modied albumin levels in patients undergoing percutaneous coronary intervention for unstable angina pectoris with versus without coronary collaterals. Am J Cardiol 93:8890 11. Peacock F, Morris DL, Anwaruddin S, Christenson RH, Collinson PO, Goodacre SW, Januzzi JL, Jesse RL, Kaski JC, Kontos MC, Lefevre G, Mutrie D, Sinha MK, Uettwiller-Geiger D, Pollack CV (2006) Meta-analysis of ischemia-modied albumin to rule out acute coronary syndromes in the emergency department. Am Heart J 152:253262 12. Sinha MK, Roy D, Gaze DC, Collinson PO, Kaski JC (2004) Role of Ischemia modied albumin, a new biochemical marker of myocardial ischaemia, in the early diagnosis of acute coronary syndromes. Emerg Med J 21:2934 13. Zapico-Muniz E, Santalo-Bel M, Merce-Muntanola J, Montiel JA, Martinez-Rubio A, Ordonez-Llanos J (2004) Ischemiamodied albumin during skeletal muscle ischemia. Clin Chem 50:10631065 14. Zini I, Tomasi A, Grimaldi R, Vannini V, Agnati LF (1992) Detection of free radicals during brain ischemia and reperfusion by spin trapping and microdialysis. Neurosci Lett 138:279282 15. Kim J, Yoon J, Jung I (2005) A diagnostic value of ischemia modied albumin in patients with suspected acute coronary syndrome with normal EKG and cardiac markers. Korean Circ J 35:928933 16. Abadie JM, Blassingame CL, Bankson DD (2005) Albumin cobalt binding assay to rule out acute coronary syndrome. Ann Clin Lab Sci 35:6672

of IMA before and after the use of thrombolytic agents and percutaneous transluminal revascularization are expected to enable us to evaluate the effectiveness of reperfusion therapy. In conclusion, IMA and IMA index were increased in patients with ischemic stroke, and IMA index was more sensitive than conventional IMA value. Therefore, IMA and IMA index might be a new diagnostic biomarker in ischemic stroke. However, evidence indicates that IMA and IMA index should not be used at present as a biomarker for the progression of ischemic stroke.
Acknowledgment The authors thank Prof. Woon Ki Paik for his critical review of the manuscript.

References
1. Seemungal BM (2007) Neuro-otological emergencies. Curr Opin Neurol 20:3239 2. Johkura K (2007) Central paroxysmal positional vertigo: isolated dizziness caused by small cerebellar hemorrhage. Stroke 38:e26 e27 author reply e28 3. Gunduz A, Turedi S, Mentese A, Altunayoglu V, Turan I, Karahan SC, Topbas M, Aydin M, Eraydin I, Akcan B (2008) Ischemia-modied albumin levels in cerebrovascular accidents. Am J Emerg Med 26:874878 4. Abboud H, Labreuche J, Meseguer E, Lavallee PC, Simon O, Olivot JM, Mazighi M, Dehoux M, Benessiano J, Steg PG, Amarenco P (2007) Ischemia-modied albumin in acute stroke. Cerebrovasc Dis 23:216220 5. Sadler PJ, Tucker A, Viles JH (1994) Involvement of a lysine residue in the N-terminal Ni2? and Cu2? binding site of serum albumins. Comparison with Co2?, Cd2? and Al3?. Eur J Biochem 220:193200 6. Bar-Or D, Lau E, Winkler JV (2000) A novel assay for cobaltalbumin binding and its potential as a marker for myocardial ischemia-a preliminary report. J Emerg Med 19:311315

123

Copyright of Neurological Sciences is the property of Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.