may contribute to the differential responseof proximal and distal epicardial coronary arteries reported by the authors, the greater reactivity to ergonovine of the distal normal coronary segments could also reflect a less effective role of the endothelium of these distal segments in modulating the vasomotor tone of the underlying vascular smooth muscle. clsulls~,ass
did not speculate on the pathogenetic role of endothelium, because we considered that the evidence for reduced production or release of endothelium-derived relaxing factor(s) in the distal coronary tree in humans was scarce. The potential mechanisms suggestedin our article and that proposedby Hanet are still a matter of speculation but deserve careful attention. Dlmltris Tousaulis, Juan Cados Kaski,
Brussels, Belgium 31 July 1991
1. Tousoulis D, Kaski JC, BogatyP, Crea F, Gavrielides S, Galassi AR, Maseri A. Reactivity of proximal and distal angiographically normal and stenoticcoronary segmentsin chronic stable angina pectoris. Am J Cardiol 1991;67:1195-1200. 2. Shimokawa H, Flavahan NA, Shep herd JT, Vanhoutte PM. Endotheliumdependent inhibition of ergonovine-induced contraction is impaired in porcine coronary arteries with regeneratedendothelium. Circulation 1989;80:643-650. 3. Hanet C, Wijns W, DecosterP, Pouleur H, Dion R, RousseauMF. Angiographic evaluation of vasomotorproperties of internal mammary arteries before and after coronary artery bypassgrafting in men. Am J Cardiol 1990;65:918-921. 4. Griffith TM, Lewis MJ, Newby AC, HendersonAH. Endothelium-derivedrelaxing factor. J Am CON Cardiol 1988; 12:797-806. REPLY: Hanet proposes another possible explanation for our finding of a greater reactivity of distal than proximal coronary segments in response to ergonovine: “. . . differencesin endothelium-derived relaxing factor activity at different levels of the coronary artery tree.” This is an interesting possibility perhaps endorsed by recent reports112of a significantly larger constrictor responseof distal coronary segments to the administration of serotonin in patients with coronary artery diseasewho most probably had a damaged endothelium. In our study,3 however, the greater vasomotor responseof the distal vesselswas similar whether ergonovine or nitrates were administered, and whether angiographically normal or stenotic segmentswere considered. Thus, in our patients ergonovine appears to have had a direct coronary constrictor action. In view of these findings, in our article we 836
IYD MD
London, England
1. McFadden E, Clarke J, Davies G, Kaski JC, Haider A, Maseri A. Effects of intracoronary serotonin on coronary vesselsin patientswith stableangina and patients with variant angina. N Engl J Med 1991;324:648-654. 2. Golino P, Pi&one F, Willerson J, Bigazzi-Cappelli M, FocaccioA, Villari B, Indolfi C, Russolillo E, Condorelli M, Chiariello M. Divergent effectsof serotonin on coronary-artery dimensions and blood flow in patients with coronary atherosclerosisand control patients.N Engl J Med 1991;324:641-648. 3. TousoulisD, Kaski JC, BogatyP, Crea F, Gavrielides S, Galassi AR, Maseri A. Reactivity of proximal and distal angiographically normal and stenoticcoronary segmentsin chronic stable angina pectoris. Am J Cardiol 1991;67:1195-1200. Usefulneu of Nifedipine in Acute Myocardial Infarction
In the recent article by Yusuf et al,’ our INTACT study2 was mentioned, especially with regard to the mortality distribution among the 2 groups of patients receiving nifedipine (80 mg/day) and placebo, respectively. We believe that we were misquoted. First, the investigators included total and not cardiac mortality for their analysis of a possible negative effect of nifedipine on acute myocardial infarction. In our opinion this is not admissible, because there is no proof that nifedipine (or any other calcium antagonist) increases the incidence of noncardiac death. The investigators also do not mention any literature on this point. Second, although in INTACT there was a total of 10 cardiac deaths (8 in the nifedipine group and 2 in the placebo group), the p value (0.105) (Fisher’s exact t test, 2-sided) was not significant. Unfortunately, Yusuf et al do not
THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 69
MARCH 15, 1992
further differentiate the causes of cardiac death in the 2 groups, which (due to lack of space)were not specified in the original report. However, these data were published in a letter to the editors of Lancet in July 1990,3as well as in Cardiouuscular Drugs and Therapy.4 The data show that all 8 cardiac deaths in the nifedipine group were in patients with myocardial infarction before entry into the study. However, only 3 of these patients died, all from reinfarctions (2 without nifedipine for >2 months). In addition, there were 3 sudden cardiac deaths (1 without nifedipine for 2 months) and 2 unwitnessed deaths of unknown causes,which were included in the cardiac deaths. Of the 2 patients who died from myocardial infarction on placebo, only 1 had a prior myocardial infarction; the other died from a new myocardial infarction. Thus, the only new myocardial infarction in a patient entering the study with a normal left ventricle was in the placebo group! In all, there were 20 new myocardial infarctions: 15 nonfatal, 7 on placebo, and 8 on nifedipine (p = 0.814 for group differences [chi-square test]). In conclusion, INTACT did not show any group differences with regard to acute myocardial infarctions (9 with placebo including 2 deaths, and 11 on nifedipine including 3 deaths). We agree that statistical multiplicity has immense problems, but this is also true for any meta-analysis, because it includes studies of various designs with various end points. Also for INTACT, the design was not to test total mortality, but to test the antiatherosclerotic effect of nifedipine in patients with coronary artery disease, as seen during angiography over an interval of 3 years. Finally, with regard to the formation of new lesions it seemsto us somewhat arbitrary to quote only the nonsignificant difference among patients (17% less on nifedipine; p = 0.129; 2-sided) and not the significant difference of -28% in the number of new lesions per patient (p = 0.034; 2-sided; Cochran’s linear trend test). As there were more patients in the placebogroup with > 1 new lesion than in the nifedipine group, the number
of lesions per patient seems to us ple count without any element of also important for risk stratifica- judgment and, thus, essentially betion, because each coronary lesion ing free of bias. It is also available is a risk for myocardial infarction. from all trials. Since 80 to 90% of A patient with 5 new stenosesis cer- deaths in cardiovascular trials are tainly at a higher risk than a patient cardiac, and there are no data to suggest that nifedipine increases with only 1. Hence, if one is performing a the risk of noncardiac deaths, the meta-analysis, all the facts should difference between an overview rebe weighted very carefully. Finally, porting on all-cause or cardiac morif the possibility of an adverseeffect tality is small. We believe that this of a drug on myocardial infarction is worth the price of obtaining comis analyzed, one should primarily plete and bias-free data, especially deal with cardiac and not total mor- in secondary prevention trials. In small trials with few deaths, tality. Paul R. Lichtkn, MD such as INTACT, the play of Hannover, Germany 8 July 1991 chance may give the appearance of a distinction between various modes of deaths. The strength of 1. Yusuf S, Held P, Furberg CD. Update our overview is the inclusion of all of effects of calcium antagonists in myocardial infarction or angina in light of the known trials of calcium antagonists rather than focusing on a specific second Danish Verapamil Infarction Trial trial. The overall data suggest that (DAVIT-II) and other recent studies. Am J Cardiol 1991;67:1295-1297. mortality is increased with the use 2. Lichtlen PR, Hugenholtz PG, Rafflenof dihydropyridine calcium antagobet11W, Hecker H, Jost ST, Deckers JW, nists. However, there may be some on behalf of the INTACT group investiuncertainty concerning the magnigators. Retardation of angiographic protude of the adverse effect. gression of coronary artery disease by The secondissue raised by Lichtnifedipine; results of the International len regards the presentation of anNifedipine Trial on Antiatherosclerotic Therapy (INTACT). Lancer 1990;335: giographic data from his trial. It appears that the primary goal of 1109-1113. his trial was to assessangiographic 3. Lichtlen PR, on behalf of the INTACT study group. Reply to the letter of Drs. progression overall. There is no difM.R. Goldstein and F.W.A. Verheugt as ference between the active and conwell as a letter. from unknown authors trol groups on this end point. The to the article on INTACT, Lancet 1990; claim of an effect on prevention of 335:1109-1113. Lancet 1990;336:172new lesions is data-derived and, 174. moreover, utilizes inappropriate 4. Lichtlen PR, Hugenholtz PG, Rafflenbeul W, Hecker H, Jost ST, Nikutta P, statistical methods. Randomization in this trial is by patient (not by Deckers JW, on behalf of the INTACT group investigators. Retardation of coro- lesion), and therefore, the unit of nary artery disease in humans by the cal- analysis has to be the individual pacium-channel blocker nifedipine: results tient.2 From a physiologic point, the of the INTACT study (International Nirates of lesion changes in different fedipine Trial on Antiatherosclerotic parts of a coronary tree within an Therapy). Cardiovascular Drugs and individual patient are likely to be Therapy 1990;4(suppl 5):1047-1068. correlated. Moreover, the entire coronary tree within an individual is subjected to the same intervenREPLY: We thank Professor Lichtlen for his comments regarding our tion. article. He raises 2 issues:The first Therefore, the analysis of the deissue relates to our use of all-cause velopment of new lesionshasto take instead of cardiac mortality from the unit of randomization (which is the INTACT study.’ The large ma- the individual patient and not each jority of overviews of randomized lesion) into account. In the primary clinical trials in cardiovascular dis- paper,’ the only analysis that utieasehave focused on all-cause mor- lizes such an approach is the comtality, including our original arti- parison of the number of patients in cle* and the recent update.3 All- each group who develop new lesions cause mortality has the distinct (70 of 173 in the active vs 85 of 175 advantage of being based on a sim- in the control). This is not statisti-
cally significant. Furthermore, in 18% of patients, follow-up angiograms were not available. The potential influence of these missing data on the analysis requires clarification. The only analysis that includes all randomized patients and reaches statistical significance in INTACT is the adverse effect of nifedipine on total mortality. Whereas the apparent excessmay be exaggerated by the play of chance in this trial, the adversetrend is consistent with data from other related trials of dihydropyridines calcium antagonists. Therefore, the totality of evidence indicates that on balance these agents are likely to be more harmful than beneficial to patients with ischemic heart disease.There are no data indicating benefit to clinical end points in such patients. SaWm Yusuf, wu8,wee,DpN Bethesda, Maryland Peter Held, PhD
Goteborg. Sweden Curl Furberg,
MD
Winston Salem, North Carolina 20 September 1991
1. Lichtlen RP, Hugenhoitz PG, Rafflenbeul W, Hecker H, Jost S, Deckers JW. Retardation of angiographic progression of coronary artery disease by nifedipine: results of the International Nifedipine Trial on Antiatherosclerotic Therapy (INTACT). Lancer 1990;335:11091113. 2. Held PH, Yusuf S, Furberg CD. Calcium channel blockers in acute myocardial infarction and unstable angina: an overview. Br Med J 1989;299:1187-1192. 3. Yusuf S, Held P, Furberg CD. Update of effects of calcium antagonists in myocardial infarction or angina in light of the second Danish Verapamil Infarction Trial (DAVIT-II) and other recent studies. Am J Cardiol 1991;67:1295-1297. 4. Yusuf S, Garg R. Randomized trials to assessthe long term effects of therapies on angiographic endpoints. Chest 1991;99: 1243-1247.
Atrial Natriuretic Factor and Transfnural Pressures
Serra et al1 contribute important insights into the physiology of atria1 natriuretic factor (ANF) releaseby demonstrating attenuation in ANF production in patients with right ventricular (RV) infarction complicating inferior wall left ventricular infarction. They suggest plausible mechanisms,including RV necrosis READERS’ COMMENTS
837
did not speculate on the pathogenetic role of endothelium, because we considered that the evidence for reduced production or release of endothelium-derived relaxing factor(s) in the distal coronary tree in humans was scarce. The potential mechanisms suggestedin our article and that proposedby Hanet are still a matter of speculation but deserve careful attention. Dlmltris Tousaulis, Juan Cados Kaski,
Brussels, Belgium 31 July 1991
1. Tousoulis D, Kaski JC, BogatyP, Crea F, Gavrielides S, Galassi AR, Maseri A. Reactivity of proximal and distal angiographically normal and stenoticcoronary segmentsin chronic stable angina pectoris. Am J Cardiol 1991;67:1195-1200. 2. Shimokawa H, Flavahan NA, Shep herd JT, Vanhoutte PM. Endotheliumdependent inhibition of ergonovine-induced contraction is impaired in porcine coronary arteries with regeneratedendothelium. Circulation 1989;80:643-650. 3. Hanet C, Wijns W, DecosterP, Pouleur H, Dion R, RousseauMF. Angiographic evaluation of vasomotorproperties of internal mammary arteries before and after coronary artery bypassgrafting in men. Am J Cardiol 1990;65:918-921. 4. Griffith TM, Lewis MJ, Newby AC, HendersonAH. Endothelium-derivedrelaxing factor. J Am CON Cardiol 1988; 12:797-806. REPLY: Hanet proposes another possible explanation for our finding of a greater reactivity of distal than proximal coronary segments in response to ergonovine: “. . . differencesin endothelium-derived relaxing factor activity at different levels of the coronary artery tree.” This is an interesting possibility perhaps endorsed by recent reports112of a significantly larger constrictor responseof distal coronary segments to the administration of serotonin in patients with coronary artery diseasewho most probably had a damaged endothelium. In our study,3 however, the greater vasomotor responseof the distal vesselswas similar whether ergonovine or nitrates were administered, and whether angiographically normal or stenotic segmentswere considered. Thus, in our patients ergonovine appears to have had a direct coronary constrictor action. In view of these findings, in our article we 836
IYD MD
London, England
1. McFadden E, Clarke J, Davies G, Kaski JC, Haider A, Maseri A. Effects of intracoronary serotonin on coronary vesselsin patientswith stableangina and patients with variant angina. N Engl J Med 1991;324:648-654. 2. Golino P, Pi&one F, Willerson J, Bigazzi-Cappelli M, FocaccioA, Villari B, Indolfi C, Russolillo E, Condorelli M, Chiariello M. Divergent effectsof serotonin on coronary-artery dimensions and blood flow in patients with coronary atherosclerosisand control patients.N Engl J Med 1991;324:641-648. 3. TousoulisD, Kaski JC, BogatyP, Crea F, Gavrielides S, Galassi AR, Maseri A. Reactivity of proximal and distal angiographically normal and stenoticcoronary segmentsin chronic stable angina pectoris. Am J Cardiol 1991;67:1195-1200. Usefulneu of Nifedipine in Acute Myocardial Infarction
In the recent article by Yusuf et al,’ our INTACT study2 was mentioned, especially with regard to the mortality distribution among the 2 groups of patients receiving nifedipine (80 mg/day) and placebo, respectively. We believe that we were misquoted. First, the investigators included total and not cardiac mortality for their analysis of a possible negative effect of nifedipine on acute myocardial infarction. In our opinion this is not admissible, because there is no proof that nifedipine (or any other calcium antagonist) increases the incidence of noncardiac death. The investigators also do not mention any literature on this point. Second, although in INTACT there was a total of 10 cardiac deaths (8 in the nifedipine group and 2 in the placebo group), the p value (0.105) (Fisher’s exact t test, 2-sided) was not significant. Unfortunately, Yusuf et al do not
THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 69
MARCH 15, 1992
further differentiate the causes of cardiac death in the 2 groups, which (due to lack of space)were not specified in the original report. However, these data were published in a letter to the editors of Lancet in July 1990,3as well as in Cardiouuscular Drugs and Therapy.4 The data show that all 8 cardiac deaths in the nifedipine group were in patients with myocardial infarction before entry into the study. However, only 3 of these patients died, all from reinfarctions (2 without nifedipine for >2 months). In addition, there were 3 sudden cardiac deaths (1 without nifedipine for 2 months) and 2 unwitnessed deaths of unknown causes,which were included in the cardiac deaths. Of the 2 patients who died from myocardial infarction on placebo, only 1 had a prior myocardial infarction; the other died from a new myocardial infarction. Thus, the only new myocardial infarction in a patient entering the study with a normal left ventricle was in the placebo group! In all, there were 20 new myocardial infarctions: 15 nonfatal, 7 on placebo, and 8 on nifedipine (p = 0.814 for group differences [chi-square test]). In conclusion, INTACT did not show any group differences with regard to acute myocardial infarctions (9 with placebo including 2 deaths, and 11 on nifedipine including 3 deaths). We agree that statistical multiplicity has immense problems, but this is also true for any meta-analysis, because it includes studies of various designs with various end points. Also for INTACT, the design was not to test total mortality, but to test the antiatherosclerotic effect of nifedipine in patients with coronary artery disease, as seen during angiography over an interval of 3 years. Finally, with regard to the formation of new lesions it seemsto us somewhat arbitrary to quote only the nonsignificant difference among patients (17% less on nifedipine; p = 0.129; 2-sided) and not the significant difference of -28% in the number of new lesions per patient (p = 0.034; 2-sided; Cochran’s linear trend test). As there were more patients in the placebogroup with > 1 new lesion than in the nifedipine group, the number
of lesions per patient seems to us ple count without any element of also important for risk stratifica- judgment and, thus, essentially betion, because each coronary lesion ing free of bias. It is also available is a risk for myocardial infarction. from all trials. Since 80 to 90% of A patient with 5 new stenosesis cer- deaths in cardiovascular trials are tainly at a higher risk than a patient cardiac, and there are no data to suggest that nifedipine increases with only 1. Hence, if one is performing a the risk of noncardiac deaths, the meta-analysis, all the facts should difference between an overview rebe weighted very carefully. Finally, porting on all-cause or cardiac morif the possibility of an adverseeffect tality is small. We believe that this of a drug on myocardial infarction is worth the price of obtaining comis analyzed, one should primarily plete and bias-free data, especially deal with cardiac and not total mor- in secondary prevention trials. In small trials with few deaths, tality. Paul R. Lichtkn, MD such as INTACT, the play of Hannover, Germany 8 July 1991 chance may give the appearance of a distinction between various modes of deaths. The strength of 1. Yusuf S, Held P, Furberg CD. Update our overview is the inclusion of all of effects of calcium antagonists in myocardial infarction or angina in light of the known trials of calcium antagonists rather than focusing on a specific second Danish Verapamil Infarction Trial trial. The overall data suggest that (DAVIT-II) and other recent studies. Am J Cardiol 1991;67:1295-1297. mortality is increased with the use 2. Lichtlen PR, Hugenholtz PG, Rafflenof dihydropyridine calcium antagobet11W, Hecker H, Jost ST, Deckers JW, nists. However, there may be some on behalf of the INTACT group investiuncertainty concerning the magnigators. Retardation of angiographic protude of the adverse effect. gression of coronary artery disease by The secondissue raised by Lichtnifedipine; results of the International len regards the presentation of anNifedipine Trial on Antiatherosclerotic Therapy (INTACT). Lancer 1990;335: giographic data from his trial. It appears that the primary goal of 1109-1113. his trial was to assessangiographic 3. Lichtlen PR, on behalf of the INTACT study group. Reply to the letter of Drs. progression overall. There is no difM.R. Goldstein and F.W.A. Verheugt as ference between the active and conwell as a letter. from unknown authors trol groups on this end point. The to the article on INTACT, Lancet 1990; claim of an effect on prevention of 335:1109-1113. Lancet 1990;336:172new lesions is data-derived and, 174. moreover, utilizes inappropriate 4. Lichtlen PR, Hugenholtz PG, Rafflenbeul W, Hecker H, Jost ST, Nikutta P, statistical methods. Randomization in this trial is by patient (not by Deckers JW, on behalf of the INTACT group investigators. Retardation of coro- lesion), and therefore, the unit of nary artery disease in humans by the cal- analysis has to be the individual pacium-channel blocker nifedipine: results tient.2 From a physiologic point, the of the INTACT study (International Nirates of lesion changes in different fedipine Trial on Antiatherosclerotic parts of a coronary tree within an Therapy). Cardiovascular Drugs and individual patient are likely to be Therapy 1990;4(suppl 5):1047-1068. correlated. Moreover, the entire coronary tree within an individual is subjected to the same intervenREPLY: We thank Professor Lichtlen for his comments regarding our tion. article. He raises 2 issues:The first Therefore, the analysis of the deissue relates to our use of all-cause velopment of new lesionshasto take instead of cardiac mortality from the unit of randomization (which is the INTACT study.’ The large ma- the individual patient and not each jority of overviews of randomized lesion) into account. In the primary clinical trials in cardiovascular dis- paper,’ the only analysis that utieasehave focused on all-cause mor- lizes such an approach is the comtality, including our original arti- parison of the number of patients in cle* and the recent update.3 All- each group who develop new lesions cause mortality has the distinct (70 of 173 in the active vs 85 of 175 advantage of being based on a sim- in the control). This is not statisti-
cally significant. Furthermore, in 18% of patients, follow-up angiograms were not available. The potential influence of these missing data on the analysis requires clarification. The only analysis that includes all randomized patients and reaches statistical significance in INTACT is the adverse effect of nifedipine on total mortality. Whereas the apparent excessmay be exaggerated by the play of chance in this trial, the adversetrend is consistent with data from other related trials of dihydropyridines calcium antagonists. Therefore, the totality of evidence indicates that on balance these agents are likely to be more harmful than beneficial to patients with ischemic heart disease.There are no data indicating benefit to clinical end points in such patients. SaWm Yusuf, wu8,wee,DpN Bethesda, Maryland Peter Held, PhD
Goteborg. Sweden Curl Furberg,
MD
Winston Salem, North Carolina 20 September 1991
1. Lichtlen RP, Hugenhoitz PG, Rafflenbeul W, Hecker H, Jost S, Deckers JW. Retardation of angiographic progression of coronary artery disease by nifedipine: results of the International Nifedipine Trial on Antiatherosclerotic Therapy (INTACT). Lancer 1990;335:11091113. 2. Held PH, Yusuf S, Furberg CD. Calcium channel blockers in acute myocardial infarction and unstable angina: an overview. Br Med J 1989;299:1187-1192. 3. Yusuf S, Held P, Furberg CD. Update of effects of calcium antagonists in myocardial infarction or angina in light of the second Danish Verapamil Infarction Trial (DAVIT-II) and other recent studies. Am J Cardiol 1991;67:1295-1297. 4. Yusuf S, Garg R. Randomized trials to assessthe long term effects of therapies on angiographic endpoints. Chest 1991;99: 1243-1247.
Atrial Natriuretic Factor and Transfnural Pressures
Serra et al1 contribute important insights into the physiology of atria1 natriuretic factor (ANF) releaseby demonstrating attenuation in ANF production in patients with right ventricular (RV) infarction complicating inferior wall left ventricular infarction. They suggest plausible mechanisms,including RV necrosis READERS’ COMMENTS
837
