- Email: [email protected]
Bedside markers of coronary reperfusion
JACC Vol. 22, No. 7 December 19!43:20§8-9
2058
4. Zabel M. Hohnloser SHq Parussel A. Just H. ST segment analysis for assessment of coronary artery patency: comparison of surface KG and Holter recordings. Eur Heart J 3992:13:1619-25.
The study of Shah and coworkers (1) on noninvasive bedside ~ken of coronary artery reperfusion in 82 patients undergoing bombo]ysjs for acute myocardialinfarctionwas carefully designed and conducted, hut we have some comments on their statistical analysis. The authors conclude that progressive decrease of pain and resolution of ST segment elevation are reliable markers for i&uct artery reopening. However, no data concerning sensitivity, specificityand positive or negative predictive value are given and the independentvalue of each of the three markers examinedwas not evaluated by multipleregressionanalysis. Instead, the conclusions are supported by seven figuresdepictingfindingsin selected patients. Becausethe statisticalanalysisof the data is inadequate,it is ditllcultto put their results into perspective with earlier reports. Shah et al. discuss three reasons why predicting coronary patency on the basis of ST elevationin the electrocardiogram(ECG) and relief of chest pain is better in their study than in previous investigations.1) They used a continuousassessment rather than analysisof two fixed time points. However, Krucoff and coworkers (2), utilizinga similarapproach of ECG analysis,found a sensitivity of g% and a specificityof 82%for the developmentof a steady state ST segmentelevation. Nevertheless, the reason for using only two time points(i.e., before and 90 13)or 120(4) minafter thrombolysis) is important: To salvage myocardium after failed tbrombolysis, accuratepredictionof coronary patency mustbe possiblewithinthis time frame. Electrocardiographicanalysis within this time period resulted in sensitivityand specificityvalues of 6fl%to 90%(3,4). 2) Shah et al. cite a lack of optimal ECG leads in previous studies. However, our group (4) recently demonstratedthat there is no significantadvantageof analyzingIZlead ECGs comparedwith only 2 Holter monitor leads in assessingST segment resolution. 3) Shah et al. state that Thromboiysis in Myocardial Infarction CTIMI)flow grade 2 does not produce reperfusion arrhythmiasor abruptincreasesin serum creatinekinaselevels.WhetherTIMIflow grade2 indicatessuccessfulor unsuccessfulreperfusionis at present unresolved.However, in our series (4), 5 of 10 patients with TIM1 flow grade 2 reperfusion exhibited accelerated idioventricular rhythmor sinusbradycardiawithin90 minafter lysis. Thus, the data providedby Shah and coworkers do not conclusivelysupport their statement that ST segment elevation and subjectiite chest pain constituteidealmarkersfor noninvasivepredictionofthesuccessor failureof thrombolysisin patients withacute myocardialinfarction. STEFANH. HOHNLOSER, MD MARKUSZABEL, MD Universily Cvospital Department
of Cardiology
HugstetterStrasse 79106 Freibwg,
55
Cennan~
References I. Shah PK. Cercek 8,
Lew AS, Ganz W. Angiographic validation of bedside markers of ~prfusion. I Am Coil Cardicl 1993;21:55-61. 2. ffiucoff MW, Green CE. Satler LF, et d. Noninvasive detection of coronary artery patency using continuous ST-segment monitoring. Am J Cardiol 1986:57:9L6-22. 3. Hohnloser SK, Zahel M, Kasper W. MeioertzT, Just H. Assessment of coronary :rtery patency after thmmbalytic therapy: amae prediction utilizing the wmbined analysis of three noninvasive markers. I Am Coil Cardiol 199l;lSA4-9.
8 1993 by the
American college
-I Cardiology
Among our 82 patients, 9 did not show a rapid and progressive decrease in ST segmentelevation and all had TIiU flowgrade 52 at angiography. Of the 73 patients with a rapid aud progressive decrease in ST segmentelevation, 69 had TIMI flow grade 3, and 4 had TIMI flow grade 52 at angiography. All of the lstter four patients had developed signs of reocclusion (recurrent chest pain and ree1evationof ST segments)hours after the initial decrease in ST elevation but before angiography. If we consider these four patients to have false positive findingsfor TIMI flow grade 3, the sensitivity,specificityand positive and negativepredictiveaccuracy of rapid ST segmentresolutionas a markerof reperfusionare IOQ%, 69%, 94%and laO%,respectively. As suggestedby HohnIoser, we subjected our data to a multivariate analysis considering the four patients with signs of reocclusion as failures. Of the four bedside variables(time to 50%ST segment decrease, time to relief of chest pain, occurrence of accelerated idioventricularrhythm, and reflex bradycardia),only the first two variableswere univariatepredictors. The ST variablewas always a significantpredictor on multivariate analysis and the chest pain variable was no longer significantonce the ST variable was entered into the model. We assessed the discriminantaccuracy of models based on the ST variable and the chest pain variableusingthe receiver operatingcharacteris:icIROC) curve area associatedwith the model. The univariateST model was significantlybetter than the univariate chest pain model in terms of the ROC area (p < 0.0001).The multivariatemodel based on both variables was only slightlybetter than the best univariate model in terms of the ROC area (p = 0.066).The results of these additiona! analyses support our originalconclusions. The reasons that Krucoff et al. (1) had an 89% sensitivity and 82%specificityin their study are not obvious to us. Their inclusion of patients up to 24 h after the onset of symptomsand initiationof treatment an average of 5.5 h after symptom onset may have influencedthe results. Because of the frequent fluctuations in the magnitude of ST segmentelevation that we observed in >50%of patients, sampling of ST segmentsat two points in time only (e.g., before and 90 mitt after initiationof therapy) can be misleading.In our experience, ST segmentelevationsfrequentlyincreaseduringthrombolytictherapy; consequently, the level from which they begin their rapid and progressivedecrease with reperfusion can be markedlyhigher than the pre!:eatment ST elevation, Moreover, frequent or continuous monitoringof the ST segments offers the additionaladvantage of providinginformationon the time of reperfusion. We agree that 12-leadmoni,toringoffers no significantadvantage. A single lead with distinct ST segment elevation is sufficient. However, using the same two leads (e.g., leads II and V_,)in all patients regard!essof infarct location (2) may be inadequate. Our conclusion that TIMI flow grade 2 does not represent adequate reperfusion is now supported by severai other studies (3-8). In summary,we believe, irt accord with other investigator&that
2058
4. Zabel M. Hohnloser SHq Parussel A. Just H. ST segment analysis for assessment of coronary artery patency: comparison of surface KG and Holter recordings. Eur Heart J 3992:13:1619-25.
The study of Shah and coworkers (1) on noninvasive bedside ~ken of coronary artery reperfusion in 82 patients undergoing bombo]ysjs for acute myocardialinfarctionwas carefully designed and conducted, hut we have some comments on their statistical analysis. The authors conclude that progressive decrease of pain and resolution of ST segment elevation are reliable markers for i&uct artery reopening. However, no data concerning sensitivity, specificityand positive or negative predictive value are given and the independentvalue of each of the three markers examinedwas not evaluated by multipleregressionanalysis. Instead, the conclusions are supported by seven figuresdepictingfindingsin selected patients. Becausethe statisticalanalysisof the data is inadequate,it is ditllcultto put their results into perspective with earlier reports. Shah et al. discuss three reasons why predicting coronary patency on the basis of ST elevationin the electrocardiogram(ECG) and relief of chest pain is better in their study than in previous investigations.1) They used a continuousassessment rather than analysisof two fixed time points. However, Krucoff and coworkers (2), utilizinga similarapproach of ECG analysis,found a sensitivity of g% and a specificityof 82%for the developmentof a steady state ST segmentelevation. Nevertheless, the reason for using only two time points(i.e., before and 90 13)or 120(4) minafter thrombolysis) is important: To salvage myocardium after failed tbrombolysis, accuratepredictionof coronary patency mustbe possiblewithinthis time frame. Electrocardiographicanalysis within this time period resulted in sensitivityand specificityvalues of 6fl%to 90%(3,4). 2) Shah et al. cite a lack of optimal ECG leads in previous studies. However, our group (4) recently demonstratedthat there is no significantadvantageof analyzingIZlead ECGs comparedwith only 2 Holter monitor leads in assessingST segment resolution. 3) Shah et al. state that Thromboiysis in Myocardial Infarction CTIMI)flow grade 2 does not produce reperfusion arrhythmiasor abruptincreasesin serum creatinekinaselevels.WhetherTIMIflow grade2 indicatessuccessfulor unsuccessfulreperfusionis at present unresolved.However, in our series (4), 5 of 10 patients with TIM1 flow grade 2 reperfusion exhibited accelerated idioventricular rhythmor sinusbradycardiawithin90 minafter lysis. Thus, the data providedby Shah and coworkers do not conclusivelysupport their statement that ST segment elevation and subjectiite chest pain constituteidealmarkersfor noninvasivepredictionofthesuccessor failureof thrombolysisin patients withacute myocardialinfarction. STEFANH. HOHNLOSER, MD MARKUSZABEL, MD Universily Cvospital Department
of Cardiology
HugstetterStrasse 79106 Freibwg,
55
Cennan~
References I. Shah PK. Cercek 8,
Lew AS, Ganz W. Angiographic validation of bedside markers of ~prfusion. I Am Coil Cardicl 1993;21:55-61. 2. ffiucoff MW, Green CE. Satler LF, et d. Noninvasive detection of coronary artery patency using continuous ST-segment monitoring. Am J Cardiol 1986:57:9L6-22. 3. Hohnloser SK, Zahel M, Kasper W. MeioertzT, Just H. Assessment of coronary :rtery patency after thmmbalytic therapy: amae prediction utilizing the wmbined analysis of three noninvasive markers. I Am Coil Cardiol 199l;lSA4-9.
8 1993 by the
American college
-I Cardiology
Among our 82 patients, 9 did not show a rapid and progressive decrease in ST segmentelevation and all had TIiU flowgrade 52 at angiography. Of the 73 patients with a rapid aud progressive decrease in ST segmentelevation, 69 had TIMI flow grade 3, and 4 had TIMI flow grade 52 at angiography. All of the lstter four patients had developed signs of reocclusion (recurrent chest pain and ree1evationof ST segments)hours after the initial decrease in ST elevation but before angiography. If we consider these four patients to have false positive findingsfor TIMI flow grade 3, the sensitivity,specificityand positive and negativepredictiveaccuracy of rapid ST segmentresolutionas a markerof reperfusionare IOQ%, 69%, 94%and laO%,respectively. As suggestedby HohnIoser, we subjected our data to a multivariate analysis considering the four patients with signs of reocclusion as failures. Of the four bedside variables(time to 50%ST segment decrease, time to relief of chest pain, occurrence of accelerated idioventricularrhythm, and reflex bradycardia),only the first two variableswere univariatepredictors. The ST variablewas always a significantpredictor on multivariate analysis and the chest pain variable was no longer significantonce the ST variable was entered into the model. We assessed the discriminantaccuracy of models based on the ST variable and the chest pain variableusingthe receiver operatingcharacteris:icIROC) curve area associatedwith the model. The univariateST model was significantlybetter than the univariate chest pain model in terms of the ROC area (p < 0.0001).The multivariatemodel based on both variables was only slightlybetter than the best univariate model in terms of the ROC area (p = 0.066).The results of these additiona! analyses support our originalconclusions. The reasons that Krucoff et al. (1) had an 89% sensitivity and 82%specificityin their study are not obvious to us. Their inclusion of patients up to 24 h after the onset of symptomsand initiationof treatment an average of 5.5 h after symptom onset may have influencedthe results. Because of the frequent fluctuations in the magnitude of ST segmentelevation that we observed in >50%of patients, sampling of ST segmentsat two points in time only (e.g., before and 90 mitt after initiationof therapy) can be misleading.In our experience, ST segmentelevationsfrequentlyincreaseduringthrombolytictherapy; consequently, the level from which they begin their rapid and progressivedecrease with reperfusion can be markedlyhigher than the pre!:eatment ST elevation, Moreover, frequent or continuous monitoringof the ST segments offers the additionaladvantage of providinginformationon the time of reperfusion. We agree that 12-leadmoni,toringoffers no significantadvantage. A single lead with distinct ST segment elevation is sufficient. However, using the same two leads (e.g., leads II and V_,)in all patients regard!essof infarct location (2) may be inadequate. Our conclusion that TIMI flow grade 2 does not represent adequate reperfusion is now supported by severai other studies (3-8). In summary,we believe, irt accord with other investigator&that