Recently, several retrospective studies (1- 5) and the prospective studies (6- 7) demonstrated that angina attack before acute myocardial infarction (AMI) exerts a preconditioning (PC) effect in humans. However, collaterals, which play an important role on the infarct size, were not examined in the prospective studies (6- 7). Therefore, it has not been established whether preinfarction angina has a PC effect.

In experimental animal models (8- 9), transient myocardial ischemia before sustained myocardial ischemia reduced the infarct size to one-third to one-eighth of the ischemic risk area, which did not depend on collateral circulation. This classic PC effect is lost if the time interval of reperfusion before sustained ischemia becomes more than 35 to 60 min in anesthetized rabbits (10- 11), more than 2 to 4 h in conscious rabbits (12), more than 2 h in anesthetized dogs (13- 14) or more than 60 min in anesthetized pigs (15). However, 24 to 48 h after PC reperfusion, the myocardial protective effects against sustained ischemia again appear (the delayed PC effect or the second window of PC effect) (14,16- 18). However, this delayed PC effect on myocardial infarct size is still controversial (19- 20). Meanwhile, it has been established in pigs (21- 22) and rabbits (23) that the PC effect against myocardial stunning appears 24 h after PC reperfusion (the delayed PC effect).

The presence of a classic PC effect with a short time interval in the human heart has been reported using the percutaneous transluminal coronary angioplasty (PTCA) model (24) and the cardiac surgery model (25- 26), but even these findings are still controversial (27- 28). The concept of the short-duration classic PC effect followed by its disappearance, and then the reappearance of the delayed PC effect with a relatively long duration, has not been examined in any human studies. In other words, it is unknown whether the delayed PC effects on myocardial infarct size and stunning are present or not in human hearts. Our hypothesis is that, if the delayed PC effect is present, the enzymatic infarct size and/or the improvement of left ventricular function should be inversely correlated with the time interval between the first and/or last preinfarction angina attack and the onset of AMI, when that time interval is limited to between approximately 2 and 48 h. However, various factors that affect the enzyme release and left ventricular wall motion, such as the infarct-related artery, the grade of collateral flow, the time interval from the onset of AMI to reperfusion and the grade of recanalization, must also be considered in this analysis.

Thus, to determine whether the preinfarction angina has a PC effect independent of collateral flow, and whether the delayed PC effect exists in humans, the preinfarction angina attacks and their protective effect against AMI were examined prospectively in patients with first AMI. In whom, the infarct-related artery was the proximal left anterior descending artery; the grade of collateral flow at the time of AMI before the recanalization was Rentrop grade 0 or 1 (29); the culprit lesion was successfully recanalized by direct PTCA between 2 and 6 h after the onset of AMI, and there was no evidence of restenosis on coronary angiography performed one month after the onset of AMI.

From August 1, 1995, until June 31, 1997, 26 institutions in Japan participated in this study. Patients with AMI who met the inclusion criteria were enrolled in this study within two days after the onset of infarction. The presence of preinfarction angina was diagnosed by the patients’ doctors from the detailed clinical history within 24 h after the onset of infarction and was reconfirmed by an experienced doctor within two weeks. Angina was defined when the patients had symptoms (such as chest pain, chest discomfort, left arm pain or jaw pain) for no longer than 30 min which were considered to be of cardiac origin.

The inclusion criteria were: 1) patients with AMI who exhibited new-onset, typical angina within two days before the onset of AMI or who never exhibited angina before the onset of AMI, 2) a first AMI with typical chest pain lasting longer than 30 min, 3) the culprit lesion was in the proximal left anterior descending artery, which was totally occluded Thrombolysis in Myocardial Infarction flow grade 0 (TIMI 0) (30), with no other significant stenosis (≥75%) of the other vessels, 4) collateral circulation was Rentrop grade 0 or 1 (29) on coronary angiography during the acute phase (because when the collateral circulation was greater 30, the myocardium of the infarct area was better salvaged), 5) more than two leads with more than 2 mm ST elevation in the anterior precordial leads of the electrocardiogram, 6) a greater than threefold increase in serum creatine kinase (CK) level, 7) patients with shock were excluded, 8) the culprit lesion was successfully reperfused by direct PTCA between 2 and 6 h after the onset of AMI (because when the time interval from the onset of AMI to reperfusion was within 2 h, the myocardium of the infarct area was definitely better salvaged 31 and also because it is very difficult to recanalize the culprit lesion by direct PTCA within 2 h from the onset of AMI) and became less than 50% stenosis without the no-reflow phenomenon 32, and 9) no treatment with glibenclamide for diabetes (because it might have affected the results of this study 33).

During the follow-up period, patients were dropped from the analysis when there was: 1) more than 75% stenosis of the infarct-related artery during the chronic phase (four weeks after the onset of AMI), 2) ischemic events during the follow-up period, or 3) other events which would affect the left ventricular wall motion, including left ventricular rupture or pericardial effusion.

At the time of admission, all patients underwent coronary angiography, coronary reperfusion therapy and left ventriculography immediately after the coronary reperfusion. Then, during the chronic phase (four weeks after the onset), they again underwent coronary angiography and left ventriculography to evaluate the presence of restenosis and the left ventricular wall motion. Written informed consent for cardiac catheterization and coronary angioplasty was obtained from each patient.

Information about the patients’ age, gender, the presence of preinfarction angina, the time interval from the onset of AMI to coronary reperfusion and the clinical events during reperfusion therapy was collected. When preinfarction angina existed, the frequency, time, duration and strength of attacks and whether they occurred during effort or at rest were determined. The coronary risk factors (smoking, hyperlipidemia, diabetes and hypertension), the treatment and the prognosis were examined.

The plasma level of CK was measured on admission every 3 h for the first 18 h and every 12 h from 24 to 72 h after admission. The values at each institution were normalized by dividing the value by the upper limit of normal at each institution. The peak value of the plasma CK (units) and the cumulative plasma CK integrated over the first 72 h (multiples of upper limit of normal-hours) were calculated from a 72-h time-activity curve of plasma CK (34).

Cine-films, including coronary arteriograms and left ventriculograms, were separately analyzed by two physicians who were blinded to the clinical information. If there were differences, they were resolved by consensus between the two observers.

Coronary artery segments were identified according to the reporting system of the American Heart Association, and it was determined whether the culprit lesion was proximal or distal to the first major septal branch and whether it was proximal or distal to the first major diagonal branch. On the coronary arteriograms at the onset of AMI, the development of the angiographic collateral circulation was graded: 0 = none; 1 = filling of side branches of the artery to be dilated via collateral channels without visualization of the epicardial segment, 2 = partial filling of the epicardial segment via collateral channels, 3 = complete filling of the epicardial segment of the artery being dilated via collateral channels (29). In this study, only patients with grade 0 or 1 were enrolled.

The left ventriculograms were obtained in the 30° right oblique views immediately after reperfusion and during the chronic phase. The left ventriculogram was analyzed with a Kontron Cardio 500 system (Konton Electronik, Munich). The left ventricular endocardial borders were hand-traced, and volumes were calculated by the area-length method. End-diastolic and end-systolic images were defined as when the left ventricular volumes were maximal and minimal, respectively. The left ventricular ejection fraction (LVEF) (%) was calculated as: (left ventricular end-diastolic volume [LVEDV] − end-systolic volume) × 100/LVEDV.

Regional wall motion (RWM) was assessed from the left ventriculogram using the centerline method along 100 chords drawn perpendicular to a centerline constructed midway between the end-diastolic and end-systolic contours. The measured motion of the 100 chords was normalized for heart size by dividing by the length of the end-diastolic perimeter. This normalized chord motion was expressed in units of standard deviations (SD) from the mean wall motion of normal subjects (35). The regional wall motion index (RWMI) was calculated as the mean motion of chords (SD/chord) lying in the area of chords of RWM ≤−2 SD at acute phase (ischemic risk area).

All data are presented as mean ± SD. The paired Student t test was used to compare the two sequential values obtained during the acute and chronic phases. A two-way repeated-measures analysis of variance and Scheffé’s F test were used to determine the effects of phases and preinfarction angina on the angiographic parameters. One-way analysis of variance and Scheffé’s F test were used to compare continuous data between two groups. Chi-square analysis was used to compare categorical data. A value of p < 0.05 was considered statistically significant.

Thirty-seven patients were enrolled at the onset of AMI. Eighteen patients exhibited preinfarction angina (PA) (new-onset) within 48 h before the onset of AMI (PA[+] group) and 19 patients were free from PA (PA[−] group). In the PA(+) group, one patient had left ventricular rupture on the fourth day after infarction and had to be operated on urgently and one had pericardial effusion on the second day of infarction; two patients were found, after precise history-taking by an experienced doctor, to have angina attacks more than 48 h (79 h and two months) before the onset of AMI. In the PA(−) group, one patient suddenly died four days after the onset of AMI due to reinfarction, and the left ventriculogram of one patient during the acute phase was not suitable for analysis. During the chronic phase in the PA(+) group, one patient suffered an angina attack on the thirteenth day after infarction and had restenosis at the culprit lesion; one patient had restenosis at the culprit lesion and one had worsened chronic renal failure so that left ventriculography could not be performed. In the PA(−) group, one patient had restenosis on the culprit lesion and, in two patients, the cardiac catheterization could not be done four weeks after the onset of AMI. Therefore, 12 patients were excluded from the analysis and 11 patients in the PA(+) group and 14 in the PA(−) group were analyzed (Table le1).

There was no significant difference between the two groups in age, gender, the time interval from the onset of AMI to coronary reperfusion, the location of the culprit lesion, the collateral circulation during the acute phase, the clinical events during reperfusion therapy or the hemodynamics during the acute phase of AMI. After reperfusion therapy, patients were treated with antiplatelet agents (aspirin: 10 in the PA[+] group and 9 in the PA[−] group or ticlopidine hydrochloride: 4 in the PA[+] group and 3 in the PA[−] group) or potassium warfarin (2 in the PA[+] group and 2 in the PA[−] group), and there were no significant differences between any two groups.

The characteristics of the PA are summarized in (Table le2). The first PA attack occurred 2 to 48 h (21.5 ± 14.6 h) before the onset of AMI, and the last one occurred 1.5 to 24 h (10 ± 7.1 h) before the onset. The frequency of the PA was one occurrence in four patients, two occurrences in four, three occurrences in two and five occurrences in one (2 ± 1 occurrences). The duration of the PA was 1 to 20 min, with an average of 7.0 min for the first PA, 9 min for last one and 7 min for all episodes.

The peak values of the plasma CK levels were 22.1 ± 9.2 units in the PA(+) group and 40.0 ± 33.3 units in the PA(−) group. The cumulative plasma CK values during the first 72 h were 601 ± 198 units in the PA(+) group and 774 ± 459 units in the PA(−) group. There was no significant difference in the peak values of the plasma CK (p = 0.10) or the cumulative plasma CK values (p = 0.26) between the PA(+) and PA(−) groups.

The LVEF values in the PA(+) group and the PA(−) group, were, during the acute phase, 45.7 ± 9.7 and 51.4 ± 8.9%, respectively, and during the chronic phase, 56.2 ± 10.1 and 52.2 ± 10.2%, respectively. The LVEF was significantly increased in the PA(+) group (p < 0.05) but not in the PA(−) group from the acute to chronic phases. Two-way repeated-measures analysis of variance did not show significant differences in the LVEF between the PA(+) and the PA(−) groups (p = 0.33). Therefore, the LVEF during each phase or the change in the LVEF from the acute to chronic phases was not compared between the two groups.

The number of chords of RWM ≤ −2 SD at the acute phase (ischemic risk area) did not differ between the PA(+) and the PA(−) groups (42.8 ± 13.8 and 42.3 ± 10.2, respectively, p = NS). The RWMI in the PA(+) group and the PA(−) group were, during the acute phase, −3.36 ± 0.57 and −3.33 ± 0.58 SD/chord, respectively, and, during the chronic phase, −1.81 ± 1.25 and −2.63 ± 0.43 SD/chord, respectively. The RWMI was significantly increased in both the PA(+) and the PA(−) groups (p < 0.01 and p < 0.01, respectively) from the acute phase to the chronic phase. Two-way repeated-measures analysis of variance showed a significant difference in the RWMI between the PA(+) and the PA(−) groups (p < 0.05). The RWMI during the acute phase did not differ between the two groups, but the RWMI during the chronic phase was significantly larger in the PA(+) group than in the PA(−) group (−1.81 ± 1.25 and −2.63 ± 0.43, respectively, p < 0.05). The increase in the RWMI was larger in the PA(+) group than in the PA(−) group (1.55 ± 1.32 and 0.69 ± 0.75 SD/chord, respectively, p < 0.05).

The enzymatic infarct size was not correlated with the time interval from either the first or the last PA to the onset of AMI. Neither the LVEF nor RWMI during the chronic phase was correlated with the time interval from the first or last PA to the onset of AMI. However, the increases in both LVEF and RWMI from the acute to chronic phases were correlated with the time interval from the first PA to the onset of AMI (r = 0.622, p < 0.05 and r = 0.646, p < 0.05, respectively). The increase in the LVEF was correlated with the time interval from the last PA to the onset of AMI (r = 0.654, p < 0.05), but the increase in the RWMI was not. There was no correlation between the above time intervals and clinical variables such as age, gender, time interval from the onset of AMI to reperfusion or collateral grade.

This prospective study demonstrated that the increase in the RWMI between the acute and chronic phases of AMI was significantly larger in patients with PA than in those without, irrespective of collateral flow, and that the time interval from the first PA to the onset of AMI was positively correlated with the increases in both LVEF and RWMI.

In this study, the increase in angiographic RWMI from the acute phase to the chronic phase was significantly larger in patients with PA than in patients without PA. These data are consistent with those from previous retrospective studies (1- 2). However, the enzymatic infarct size did not differ between the patients with and without PA, which is not consistent with the results of previous retrospective and prospective studies (1,7). This discrepancy may be explained by the following facts. First, in two previous retrospective and prospective studies (1,7), the collateral flow at the onset of AMI, which has an effect on infarct size (30), was not evaluated. However, in this study, it was evaluated using the angiographic collateral grade (29), which was well correlated with the collateral flow velocity pattern assessed by a Doppler guide wire (36), and only the patients with no or poor collateral flow were selected. Second, the enzymatic infarct size is generally correlated with the pathological infarct size, but it is influenced by the time interval from the onset of AMI to reperfusion and the area perfused by the culprit artery (37). Furthermore, measurement of CK every 3 h would not be frequent enough to detect the peak of CK. Therefore, the angiographic infarct size, estimated from the improvement of left ventricular RWM, might be more sensitive for detecting small differences of infarct size than the enzymatic infarct size, estimated from the cumulative plasma CK values. Third, in previous retrospective studies (1,4) that showed a significant difference of enzymatic infarct size, the time from the onset of AMI to reperfusion was shorter than it was in this study. Nagao et al. (4) demonstrated, using the enzymatic infarct size, that the PC effect of PA was present only when the time interval from the onset of AMI to reperfusion was less than 1 h. Thus, if the time interval from the onset of AMI to reperfusion is shorter, the difference of infarct size may become larger and the enzymatic infarct size may also become different between the PA(−) and the PA(+) groups. However, this hypothesis remains unproved, because it has not been investigated in a prospective study in which collateral flow was considered. Fourth, beneficial effects on myocardial stunning have been shown to be related to the delayed PC effect, but not the classic PC effect, in rabbit and pig models (21- 22,38- 40). Those studies were performed using animal models without infarction (41). In addition, it is unclear whether left ventricular dysfunction due to stunning can continue approximately four weeks after prolonged ischemia, as seen in this study. However, a recent study revealed that postinfarction contractile dysfunction after reperfusion continues in the salvaged myocardium one month or more after AMI (maimed myocardium) (42) and then may cause incomplete recovery. Another problem with stunning is that a preceding angina attack may make the degree of stunning at the acute phase after infarction larger, and this may explain the better recovery in the PA(+) group at the one month follow-up. However, the similar levels of RWMI at the acute phase in the PA(+) and PA(−) groups disprove this possibility. Thus, the protective effect on left ventricular wall motion detected four weeks after AMI in this study may represent differences in the extent of infarction (even if the crude CK method did not reveal differences) and/or a protective effect against maimed myocardium.

The classic PC effect lasts from 35 to 120 min in the anesthetized animal models (10- 11) and from 2 to 4 h in conscious rabbits (12). This suggests that the classic PC effect may disappear within several hours in human hearts although we have no data in humans. Meanwhile, the delayed PC effect on infarct size is seen 24 h after the PC stimulus in dogs (14) and from 24 to 72 h after the PC stimulus in rabbits (16,43), although the precise time of reappearance is unclear. The mechanism of the delayed PC effect is still unclear, but the expression of heat shock protein (16,19) or MnSOD (44- 45) in myocardial tissues is thought to be involved in the mechanism. In human hearts, the classic PC effect using PTCA (24) and cardiac surgery models (25), and the PC effect using the PA (1- 2,4,7) have been reported. Among those studies, only one (7) investigated whether the delayed PC effect on infarct size was present or not although the collaterals were not examined in that study. In this study, the first PA attack occurred 2 to 48 h before the onset of AMI, and the last PA attack occurred 1.5 to 24 h before the onset of AMI. Although there was no significant difference in the enzymatic infarct size, the improvement of the LVEF and RWMI in patients with PA became larger when the time interval from the first PA to the onset of AMI became longer, up to 48 h. This strongly suggests the presence of the delayed PC effect in human hearts. An alternative explanation for the positive correlation between the left ventricular function and the time interval is that a deleterious effect may appear when the interval between angina and infarction is short compared with when it is long. However, in this study, the RWMI and LVEF at the acute and chronic phases and the changes between acute and chronic phases in patients of the PA(+) group, whose PA occurred a short time before AMI, were similar to those of the PA(−) group. This fact does not support the above scenario.

First, in this study as well as in previous studies (1- 7), the preinfarction ischemia for PC was diagnosed by the typical symptoms of angina before the onset of AMI. Therefore, any patients with silent myocardial ischemia before the onset of AMI were put into the PA(−) group in this study, and the number of any such patients is unknown.

Second, in all patients of this study, the culprit lesion was reperfused by direct PTCA, in contrast with the previous studies (1- 7). This procedure would make it difficult to collect the patients of AMI recanalized within 2 h because it is very difficult to recanalize the culprit lesion by direct PTCA within 2 h from the onset of AMI. Therefore, we excluded these patients. However, using this procedure, the time of reperfusion was accurate and we could confirm that the culprit lesion became less than 50% stenosis without the no-reflow phenomenon.

Third, in human hearts, the classic PC effect has been reported in the PTCA model (24) and surgical model (25- 26). However, we could not analyze the patients with the first PA within 2 h. Therefore, the classic PC effect on human infarct size remains unknown.

Fourth, in rabbits (11,46), preinfarction ischemia for 3 to 10 min produced the PC effects on infarct size, but ischemia of less than 2 min did not. Conversely, preinfarction ischemia for more than 15 min expanded the infarct size (46). It was also reported that the PC effect of a very brief period of ischemia was enhanced by repetition (11). However, Li et al. (9) reported that preconditioning with one brief period of ischemia was as effective as PC with 10 cycles of brief periods of ischemia. In this study, the effects of the duration and number of PA attacks could not be analyzed because of the small number of cases, and further investigation is needed to determine these effects.

Finally, the number of patients in this study is somewhat small compared with the numbers generally enrolled in multicenter studies. However, each patient was strictly selected as follows for this study: first AMI with single vessel disease, culprit lesion of the proximal left anterior descending artery which was totally occluded, no or poor collaterals, successfully reperfused by direct PTCA between 2 and 6 h after the onset of AMI and without restenosis or ischemic events during the follow-up period. Therefore, we believe that the data are reliable.

In conclusion, 1) PA did not significantly reduce the enzymatic infarct size, but can improve the increase in left ventricular wall motion, 2) the beneficial effects of PA on left ventricular wall motion are correlated positively with the time interval from the first PA attack to the onset of AMI when the time interval is limited from 2 to 48 h. These effects do not depend on collateral circulation. This suggests the presence of the delayed PC effect in human hearts. Appendix

We thank Miss Mihoko Kohrogi for her secretarial support. We also thank Kowa Co. Ltd. for the support of this study.

Clinical Centers (in Alphabetical order): Asahikawa City Hospital, Tachida K, MD; Gifu Municipal Hospital, Tanaka T, MD; Gifu University School of Medicine, Fujiwara H, MD; Hiroshima City Hospital, Sato H, MD; Iwate Medical University School of Medicine, Hiramori K, MD; Kansai Rosai Hospital, Nanto S, MD; Kushiro City General Hospital, Okada T, MD; Kyoto University, Tanaka M, MD; Matsunami General Hospital, Morita N, MD; Nagoya University School of Medicine, Ito T, MD; National Cardiovascular Center, Nonogi H, MD; National Toyohashi-Higashi Hospital, Suzuki T, MD; Nippon Medical School Hospital, Takano T, MD; Osaka City General Hospital, Haze K, MD; Osaka National Hospital, Hayashi T, MD; Osaka University School of Medicine, Hori M, MD, Tada M, MD; Prefectural Gifu Hospital, Watanabe S, MD; Rakuwakai Otowa Hospital, Kida M, MD; Sakurabashi Watanabe Hospital, Fujii K, MD; Sapporo Medical University School of Medicine, Miura T, MD; Shinnittetsu Muroran General Hospital, Matsuki T, MD; Social Welfare Juridical Person Hakodate Kouseiin Hakodate Goryoukaku Hospital, Oimatsu H, MD; Surugadai Nihon University Hospital, Kanmatsuse K, MD; Tokuyama Central Hospital, Ogawa H, MD; Tokyo Women’s Medical College, Hosoda S, MD; Yamaguchi University School of Medicine, Matsuzaki M, MD.