Recurrent Myocardial Infarction: ClinicalPredictors and Prognostic Implications

ALON MARMOR, M.D., EDWARD M. GELTMAN, M.D., KENNETH SCHECHTMAN, PH.D.,

BURTON E. SOBEL, M.D., AND ROBERT ROBERTS, M.D.

SUMMARY Based on a prospective study in 200 consecutive patients with myocardial infarction, we

reported previously that early recurrent myocardial infarction is more frequent after nontransmural thantransmural infarction. Multiple logistic regression analysis using 14 clinical variables identified, in additionto type of infarction (nontransmural), three other risk factors for early recurrent infarction: obesity, femalegender and recurrent chest pain. Early recurrent infarction was documented by reelevation in plasma MBCK activity. The present study was performed to assess the accuracy of these variables as prospectivepredictors of early recurrent infarction. Studies were performed in a new population of 150 patientsadmitted consecutively with acute myocardial infarction (test set). The regression coefficients derived bymultiple logistic analysis from the training set population were applied by comparable analysis to the test setpopulation and the presence or absence of recurrent infarction was predicted correctly for 80% of thepatients. Patients were followed prospectively for 9 months (range 3-18 months) and life-table analysis wasperformed to assess the impact of recurrent infarction on short and long term survival. During the first 21days after infarction, mortality was 23% for patients with transmural and 10% for nontransmural infarc-tion (p < 0.01). However, among patients with nontransmural infarction, hospital mortality was 23%among those with early recurrent infarction and 8% for those without recurrence (n < 0.05). At theconclusion of follow-up, mortality was 34% among patients with nontransmural infarction and recurrence,

compared with 23% among patients with nontransmural infarction and no recurrence. Thus, early recur-

rent infarction has a distinct deleterious effect on survival, and a subset of patients more likely to experiencerecurrent infarction can be prospectively identified.

EARLY recurrent myocardial infarction, or extensionof infarction, has been recognized clinically for manyyears as manifested by severe prolonged recurrentchest pain and new ST-T-wave abnormalities on theECG.' Results of earlier studies suggested an inci-dence of early recurrent infarction of up to 86% ,2 butrecent studies using objective specific enzymatic crite-ria have reported an incidence of 17-25%).6 In a pre-vious report of 200 consecutive patients with acutemyocardial infarction, early recurrent infarction de-tected by reelevation of plasma MB creatine kinase(MB CK) was observed in 17% of patients with docu-mented myocardial infarction.7 8

Analyses were performed to determine whether cer-tain clinical characteristics of the population mightidentify a subset of patients at high risk for subsequentrecurrent infarction. Those who experienced an initialnontransmural myocardial infarction comprised onesubset of patients at high risk. The incidence of earlyrecurrent infarction was 43% after nontransmural in-farction, compared with only 8% among patients withtransmural infarction.7 Three other risk factors for ear-ly recurrent infarction were also identified: femalegender, obesity and recurrent chest pain. Using logis-tic regression analysis and combining the four varia-bles - type of infarction (subendocardial or trans-mural), gender, obesity and preceding recurrent chestpain we identified a subset of patients with a very

From the Cardiovascular Division, Washington University School ofMedicine, St. Louis, Missouri.

Supported in part by SCOR in Ischemic Heart Disease grant HL17646, NIH.Address for correspondence: Robert Roberts, M.D., Washington

University School of Medicine, Cardiovascular Division, 660 SouthEuclid Avenue, St. Louis, Missouri 63110.

Received August 6. 1981; revision accepted December 7, 1981.Circulation 66, No. 2, 1982.

high risk of developing early recurrent infarction. Obe-sity was defined as at least 25% in excess of ideal bodyweight for the person's height and recurrent pain wasdefined as chest pain that occurred at least 24 hoursafter the onset of the initial infarction on at least threeoccasions preceding the recurrent infarction.To assess the accuracy of these variables for predict-

ing recurrent infarction, the mathematical coefficientsderived from our initial study were applied to a newgroup of patients. In addition, since data describing theeffects of early recurrent infarction on both the in-hospital and the posthospital course of patients arelimited, follow-up was performed prospectively to de-termine the impact of early recurrent infarction onacute and long-term prognosis.

MethodsPatients

Three hundred fifty consecutive patients with acutemyocardial infarction were studied. The initial 200patients previously reported7 formed the training setand the next 150 patients the test set. Acute myocardialinfarction was documented by a history of typical pre-cordial chest pain, electrocardiographic manifestationsand serial elevations of plasma MB CK. Patients wereconsidered to have a transmural myocardial infarctionif new Q waves were observed on the ECG at least 30msec in duration in two or more anatomically adjacentleads within 24 hours of onset of infarction. Patients inwhom the new electrocardiographic changes were re-stricted to the ST-T segment and in whom plasma MBCK was elevated were classified as having nontrans-mural infarction. Patients with conduction defects inwhom development of Q waves on the ECG may bemasked, as in patients with left bundle branch block,were categorized as having infarction of undetermined

415

by guest on April 28, 2018

http://circ.ahajournals.org/D

ownloaded from

VotI 66. No 2. AtGUST 1982

type. The overall incidence of recurrent infarction re-fers to the incidence observed in all patients whetherclassified as subendocardial, transmural or of undeter-mined type. In the training set 9% and in the test set1 % of the infarctions were classified as undeterminedand could not be included in the analysis of subsets.

Creatine Kinase DeterminationsBlood samples were collected every 4 hours for the

first 72 hours and every 12 hours subsequently for 14days for the determination of plasma total and MB CKactivity. Samples were collected in EGTA (5 mM),centrifuged at 1000 g for 10 minutes, the plasma wasremoved and mercaptoethanol (10 mM) was added topreserve enzyme activity during storage at - 20TC.3Total CK activity was assessed spectrophotometrical-ly.'0 MB CK was assessed qualitatively by electro-phoresis' and quantitatively using the glass bead ad-sorption assay.'2

Early reinfarction (during hospital stay) was diag-nosed if plasma MB CK activity exhibited a reeleva-tion after a decline either to baseline or to values < 8IU/1. Diagnosis of late reinfarction (after discharge)was based on clinical criteria consisting of chest painand typical electrocardiographic changes and enzymat-ic criteria, if available.

Follow-up StudyAll 350 patients were followed prospectively for a

mean of 9 months (range 3-18 months). Major clinicalevents, including hospitalization, cardiac catheteriza-tion, coronary artery bypass surgery, myocardial in-farction and death, were confirmed by personal contactwith the patient, the family or the private physician andreview of hospital records. Throughout follow-up, thepresence or absence of angina was noted, as was itscharacter and frequency, presence of congestive heartfailure and use of medications. Sudden death was con-sidered cardiac in origin, but recurrent infarction wasdiagnosed only if a history of chest pain or infarctionwas confirmed at necropsy.

Statistical MethodsThe Washington University IBM 360/370 computer

system was used for analysis and storage of data. Datawere stored on disc and analyzed with the StatisticalAnalysis System (SAS)'3 14 and the BMDP series ofthe University of California in Los Angeles. Chi-square test for discrete variables and Wilcoxon tests forcontinuous variables provided information regardingbivariate relationships in the entire population.The training set of 200 patients was analyzed

through the use of an all possible subsets regressionusing 14 independent variables and early recurrent in-farction as the dependent variable.' This procedureyielded a subset of significant independent variablesthat were subjected to a stepwise logistic analysis,which provided four significant predictors of early re-current infarction (recurrent chest pain preceding therecurrence, type of infarction [subendocardial], gender[female], and obesity). The most influential predictors

were subendocardial infarction and recurrent chestpain. In a retrospective analysis of the training popula-tion, using all four variables, each patient who exhibit-ed a probability (p) of recurrence of greater than 0.5 ascomputed by the logistic regression analysis was iden-tified as having a recurrence. This process provided avery good retrospective differentiation for early recur-rent infarction and was successful in discriminatingbetween those who did and those who did not developan early recurrence with 80% accuracy.The validity of a stepwise logistic regression analy-

sis is traditionally evaluated by independent studies.Such repeat studies are intended to deal with the inher-ent mathematical bias that may be present when theregression coefficients of a logistic analysis are used inmaking predictions on the same set of data from whichthe coefficients were derived. Furthermore. repeatstudies assess the factor of chance that is inherent inchoosing a small subset of significant variables from alarge collection. Both of these pitfalls are inherent inan analysis of a single population regardless of samplesize. Thus, we sought to assess the accuracy of the fourvariables identified in the training set for the predictionof recurrent infarction by analyzing our test set of 150new consecutive patients by utilizing two independentmodalities. In the first instance, the test set was sub-jected to regression analysis of all possible subsets,followed by a stepwise logistic analysis to determine ifthe same predictive variables were significant in thetest set population and in the training set population.Second, the regression coefficients derived from thelogistic analysis of the training set were applied to thetest set in order to produce a mathematically unbiasedestimate of the predictive capacity of the four inde-pendent variables of interest. This provided a probabil-ity that any individual with a given set of values for thepredictive variables would extend, and if the probabil-ity for a given individual was greater than 0.5, wepredicted a recurrence.

The follow-up data were analyzed by life-table anal-ysis performed with a BMDP program to compare thesurvival curves of patients with transmural and suben-docardial infarction of patients with and without earlyrecurrent infarction in the entire population and in se-lected subsets.

ResultsComparison of Training and Test PopulationsThe test and the training populations were compared

on the basis of 14 clinical characteristics (table 1). Theoverall incidence of early recurrent infarction was al-most identical (17% and 19%). In the training set, theincidence of recurrent infarction was 43% after suben-docardial and only 8% after transmural infarction.These values are similar to those for the test set in 150new patients, in whom the incidence of recurrent in-farction was 40% after an initial subendocardial infarc-tion, compared with only 10% after an initial trans-mural infarction. The distribution of time of recurrenceshown for the overall populations in figure 1 was simi-lar with the maximum occurrence rate on the tenth day

CIRCULATION416

by guest on April 28, 2018

http://circ.ahajournals.org/D

ownloaded from

RECURRENT MI/Marmor et al.

TABLE 1. Clinical Characteristics of Patienit PopulationiTraining

Total Training Test vs testpopulation set set set (p)

n 350 200 150

Male:female 226:124 138:62 78:72 0.0455

Obesity* 111 (31.7%)* 50(25%)t 61 (40.7%)t 0.0018

Nontransmural infarction 125 (39.7%) 61 (33%) 64 (47.8%) 0.0076

Chest paint 130 (37.1%) 66 (33%7) 64 (42.7%) > 0.05

Transient ST-segment shift 106 (30.3%) 58 (29%7) 48 (32%) > 0.05

Previous myocardialinfarction-* 103 (29.4%) 49 (24.5%) 54 (36%) 0.0195

Diabetes§ 40 (11.5%) 21 (10.5%) 19 (12.7%) > 0.05

Smoking 132 (37.7%) 80 (40c%c) 52 (34.9%) > 0.05

Congestive heart failure 183 (52.4%) 101 (50.5%) 82 (55%) > 0.05

Lipid disorderW 66 (18.9%) 23 (11.5%) 43 (28.7%) 0.0001

Hypertensiontt 159 (45.4%) 76 (38%cl) 83 (55.3%7c) 0.0013

Mean age (years) 64.5 64.2 64.9 > 0.05

Mean peak CK(IU/1) 1106 1019 1222 > 0.05

Early recurrent infarction 64 (18.3%7) 35 (17.5%) 29 (19.3%) > 0.05

Died during follow-up 102 (29.1%) 62 (31%) 40 (26.7%) > 0.05

*Body weight > 25% above optimum weight for height.tAt least three episodes of anginal pain during hospitalization before extension.-,Based on documentation by ECG and enzyme changes.§Patients receiving treatment for previously diagnosed diabetes.¶Elevated serum cholesterol or elevated serum triglycerides or both as documented previously or during the present

hospitalization.ttSystolic blood pressure > 150 mm Hg or diastolic blood pressure > 90 mm Hg or both documented previously or

during the present hospitalization.

for the training set and on the ninth day for the test set.Forty-eight percent of the patients in the test set hadnontransmural infarction as opposed to 33% of thetraining set. There was a higher percentage of femalesin the test set and a higher incidence of obesity. Theincidence of previous infarction, lipid disorders andhypertension was also higher in the test population.There was no significant difference between the train-ing and test sets in the type or amount of medicationadministered.

All possible subsets multivariate regression analysiswas used in the test set of 150 patients to assess theassociation between each of the 14 variables previous-ly studied and the occurrence of early recurrent infarc-

Coz0

uzLu

w

D

6

_l

....-.....\.......\± \J S I S C. ^I \J 5 £ 5 [\WLXX1 KXN\\\4\\\\X\l\I2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

DAYS

FIGURE 1. The time distribution ofearly recurrent infarction.The majority of reinfarctions occurred during the first 10 daysafter the onset of symptoms.

tion. As in the training set, the same four variables (thetype of infarction, repeat episodes of chest pain pre-ceding the extension, gender and obesity) were foundto be significant (table 2). Thus, the two populationswere very similar with respect to the incidence of re-current infarction and clinical features associated withsignificant risk for reinfarction.

Prediction of Early Recurrent InfarctionA multiple logistic regression analysis was applied

to the training set, which yielded regression coeffi-cients for each of the four predictive variables. Toassess the accuracy of these four variables as prospec-tive predictors of recurrent infarction, the coefficientsso derived from the training set were applied to the testset producing an unbiased estimate of the probabilityof recurrence for each individual in the test set. Using aprobability of recurrence of 0.5 as the dividing pointfor prediction of recurrence or lack thereof, the pres-ence or absence of recurrent infarction was correctlyidentified prospectively in 80% of the test population.Early recurrent infarction was correctly predicted in 23of 28 patients (82%) who were documented to have arecurrence, but five patients were incorrectly predictedas not having a recurrence. One patient with recurrencecould not be classified as nontransmural or transmuraland thus was not considered in the analysis. In 106patients without early recurrent infarction, lack of re-currence was correctly predicted in 84 patients (82%).

417

8r

,\7

by guest on April 28, 2018

http://circ.ahajournals.org/D

ownloaded from

VoL 66. No 2, AuA(S-,s, I 1982

TABLE 2. The Relation2ship Between the Incidence (f EEarly Recurrent Infarction andi Selected Vcritiable.Incidence of recurrent infarction

Total population Transmural infarction Nontransmural infarction

In the In the In thepresence presence presencceof the Signifi- of the Signiti- of the Signifi-

variable cance* variable cance variable cance

Gender (female) 56.25%c 0.0001 42.86%c 0.4415 62.222 0.0001

Obesityt 54.69%c 0.00(1 28.57% 0.9131 63.27%7O.(0(001

Chest paint 85.94Yc 0.0001 92.86%c 0.0001 8 1.63% 0.0001

ECG changes§ 93.75% 0.0001 85.71%7( 0.000M1 95.92%c (0.(0)01History of angina 75.00% 0.0002 50.00%7 0.5432 83.67%l (0.0080History of hyperlipidemnia 28.13% 0.0360 7.14%cl 0.4471 34.69% 0.0912

Previous infarction 39.06% 0.0614 42.86%6 0.0455 36.23% )0.8726

History of cigarette smoking 37.50cc 0.1062 42.86 0(7 0.6137 36. 73%sc 0.87X26History of diabetes 17. 19% 0.1115 14.29%74 0.5572 18.37%: 0. 1424

Died during follow-up 34.38% 0(3.3082 28.57%7c 0.3170 18. 37%7 0,00

History of hypertension 48.44c%( 0.5928 50.00%(0.5159 46.94%- 1).6328

CHF 50.79% 0.7731 71.43% 0.4210 43.7 5c%c 0.2229

vAssociation between variable and incidence of early recurrent infarctiorn determined by multiple"Body weight - 25% above optimum weight for height..-At least three episodes of anginal chest pain during hospitalization prior to extension.§New ST elevation or depression accompanying episodes of chest pain.

We incorrectly predicted a recurrence in 22 patients.Thus, applying the regression coefficients from thetraining set to the test set, we identified correctly pa-tients who had a recurrence with a sensitivity of 82%and a specificity of 42%. However, overall, 80% ofthe patients in the test set were correctly classified withrespect to the presence or absence of recurrentinfarction.

Temporal Pattern of ReinfarctionIn the entire population of 350 patients, 75 patients

had reinfarctions. Sixty-four reinfarctions (85%) oc-curred during the initial hospitalization. The time ofoccurrence of reinfarction in relationship to the initialinfarction is illustrated in figure 1. Sixty-five percentoccurred between the third and tenth day. The inci-dence of reinfarction was much lower during the re-mainder of hospitalization, despite continued enzy-matic surveillance. After the initial hospital discharge,reinfarction was documented electrocardiographicallyand enzymatically in 11 patients. However, 15 patientsdied at home, and we could not determine whether thedeaths were associated with infarction. The time ofoccurrence of the 11 late episodes of documented rein-farction were evenly distributed throughout the follow-up.

Short- and Long-term PrognosisFor follow-up purposes, the total population of 350

patients is considered as one group. One hundred fourpatients died during the period of observation, whichaveraged 9 months (3-18 months). Fifty-seven pa-tients (55%) died during the first 3 weeks after theinitial infarction. Twenty-eight patients (13%) withtransmural infarction died during the first 72 hours.

locistic regression.

compared with only 11 patients (9%) with nontrans-mural infarction. This pattern continued during theduration of the initial hospitalization, with a mortalityrate of 23% for patients with transmural infarction at 3weeks and 10% among those with nontransmural in-farction (p < 0.01). Most of the deaths among patientswith transmural infarction occurred relatively early intheir course, after which mortality declined remark-ably (fig. 2). In contrast, the mortality rate remainedsubstantial during the follow-up interval among pa-tients with nontransmural myocardial infarction. Themortality for transmural and nontransmural infarctionwas approximately the same at the end of the 9-monthfollow-up, with a mortalitiy rate of 30% among pa-tients with transmural and 27% among patients withnontransmural infarction. Even when patients who ex-perienced a myocardial infarction predating their indexinfarction were excluded, results were similar (fig. 2).The mortality rate at the end of the follow-up intervalwas 30% for patients with transmural infarction and25% for patients with nontransmural infarction (NS).The overall mortality rate in patients with trans-

mural and nontransmural infarction was slightly higheramong patients who experienced reinfarction, but thetrend was not statistically significant, with a mortalityrate of 36% among patients with and 28% among pa-tients without recurrent infarction. Among patientswith transmural infarction, recurrence was infrequent(10%) and had no significant effect on either short- orlong-term survival. In contrast, among patients withnontransmural infarction, the incidence of early recur-rent infarction was high (42%) and the mortality wassubstantially higher among those with recurrent infarc-tion (fig. 3). Among patients with nontransmural in-farction who did not experience an early reinfarction,

CIRCULATION418

by guest on April 28, 2018

http://circ.ahajournals.org/D

ownloaded from

RECURRENT MI/Marmor et al.

mortality at 21 days was 8%, compared with 23%among patients with nontransmural infarction who hada reinfarction during the initial 3 weeks (p < 0.05).However, after the initial 3 weeks, mortality was com-parable for patients with and without recurrence. Atthe conclusion of the 9-month follow-up, mortalitywas 23% among patients with nontransmural infarc-

1.0

0.9

0.8

0.7

0.6

0.5

f AIIDU.4

z 0.3

c> 0.2D

0.1z

0

0 1.0oQa-0

C 0.9

LU 0.8

I-

< 0.7U r0.6

U-

0 60 120 180 240 300 360 420 480

TIME (Days)

FIGURE 2. Survival of patients with transmural (triangles)and nontransmural (circles) infarction. Each point on the sur-

vival curves reflects at least one new death, and the curves endwith the last recorded death in that group; thus, the survivalcurvesfor the two groups are ofdifferent durations in thisfigureand in figure 3. At the end of 180 davs, 56 patients with non-

transmural and 97 with transmural infarction were alive. Thelast point plotted in panel A represents 15 patients remainingalive with nontransmural and 67 patients with transmural in-farction. Panel A represents data from all patients; panel Brepresents patients without a prior infarction antedating theindex infarction. Patients with prior infarction were excludedfrom the analysis.

(D U.4z> 0.3-

> 0.2

0.-z0

0-

0.5

C 0.4

< 0.3

0.2

0.1 _

0o III0 50 100 150 200 250 300 350 400

TIME (Days)FIGURE 3. Survival ofpatients with nontransmural infarctionand early recurrent infarction (circles) and those with non-transmural infarction and no recurrence (squares). After 180days, 37 patients with nontransmural infarction and no recur-rence and 21 patients with nontransmural infarction with recur-rent infarction were alive. At the end offollow-up, seven pa-tients with nontransmural infarction and no recurrence and 15with nontransmural and recurrent infarction were alive. Datafrom all patients are included in panel A, but only patientswithout an infarction antedating their index infarction are in-cluded in panel B. Early mortality was greaterfor patients withearly recurrent infarction (p < 0. 05), but survival at] year wassimilar (NS).

tion who did not experience reinfarction, comparedwith 34% among those who did. Thus, a close tempo-ral relationship was evident between the pattern ofreinfarction (which usually occurred within the first 10days) and the pattern of mortality among patients withnontransmural infarction, with most deaths amongthose with reinfarction occurring within the first 30days.When patients with nontransmural infarction were

stratified according to the presence of early recurrent

419

by guest on April 28, 2018

http://circ.ahajournals.org/D

ownloaded from

VOL 66, No 2, AUIGLSI 1982

infarction, the pattern of mortality among patients withnontransmural infarction and early recurrence wasidentical to that of patients with transmural infarctionwithout recurrence (fig. 4). After patients with infarc-tion antedating their index infarction were excludedfrom analysis, the similarity persisted. These data indi-cate a marked influence of early recurrent infarction onearly mortality after nontransmural infarction despitethe relatively modest extent of the initial infarction.

1.0A

0.9

0.8

0.7

0.6

0.5

(. 0.4z> 0.3

D 0.2UO

Z 0.10O I00

0 .4B

QL 0.9Lu

>0.8 >_

_0.6

0.4

0.3-

0.1

0 1r III

0 60 120 180 240 300 360 420

TIME (Days)FIGURE 4. Comparison of the survival ofpatients with trans-mural infarction (triangles) to that of patients with nontrans-mural infarction with (circles) and without (squares) early re-current infarction. Early mortality was higher after transmuralthan after nontransmural infarction and no recurrence (A), butsurvival was identical for patients with transmural infarctionand those with nontransmural infarction and an early recurrentinfarction (B).

Among patients with nontransmural infarction and noearly recurrence, early survival was substantially bet-ter than that of patients with transmural infarction.However, patients with nontransmural infarction had asignificant late mortality, presumably because of con-tinuing reinfarction, a finding documented in 1 I pa-tients. Fifteen deaths occurred at home, and we couldnot determine whether they were associated withreinfarction.

DiscussionEarly recurrent myocardial infarction was detected

prospectively in our study based on reelevation of plas-ma MB CK after it had returned to baseline. ElevatedMB CK is a highly sensitive and specific index ofmyocardial infarction' 1-7 that provides an objectivemeans of detecting early recurrent myocardial injury.In our previous study, recurrent infarction7 was alsoassociated with elevated serum myoglobin and furtherimpairment in regional wall motion in the area contig-uous with the site of the antecedent infarction. In thepresent prospective study of 150 patients admitted con-secutively, the overall incidence of reinfarction was19% , with an incidence of 40% after an initial non-transmural infarction, compared with only 10% afteran initial transmural infarction. These results are verysimilar to those of the previous prospective study of200 patients' in which the overall incidence was 17%,43% after nontransmural and 8% after transmural in-farction. Since the 350 patients were studied prospec-tively and represent consecutive admissions, and sincethe increased incidence of reinfarction after nontrans-mural infarction is so striking and similar in both popu-lations, one must assume that it reflects some predomi-nate intrinsic feature that distinguishes transmuralfrom nontransmural infarction.

In our previous study, type of infarction, gender,obesity and preceding recurrent chest pain were shownto be risk factors as determined from multiple logisticregression analysis. Their high predictive value wasfurther confirmed: Application of the regression coef-ficients derived from the initial population had a sensi-tivity of 82% and a predictive accuracy of 80% forrecurrence of infarction in the new test population,with a specificity of 42%. The specificity of the predic-tive indexes refer to the population as a whole (trans-mural and nontransmural). Specificity, if applied tothe population with nontransmural infarction only, is88%, sensitivity is 76% and overall accuracy is 83%.Nevertheless, even when applied to the total popula-tion of transmural and nontransmural infarction, thenumber of patients required to assess therapy designedto reduce the incidence of reinfarction is at least oneorder of magnitude less than the number of patientsrequired to assess a reduction in the conventional end-point of mortality. Thus, based on four convenient andreadily available clinical measurements, one can iden-tify prospectively a subset of patients with a very highpropensity to develop reinfarction. The diagnostic andtherapeutic implications are self-evident, particularlyin assessing efficacy of prophylactic therapy.

420 CIRCULATION

by guest on April 28, 2018

http://circ.ahajournals.org/D

ownloaded from

RECURRENT MIlMarmor et al.

Nontransmural infarction is associated with modestnecrosis20 and the initial mortality is very low com-pared with that of transmural infarction. Despite theinitial favorable acute prognosis, prospective follow-up data on our patients confirmed previous retrospec-tive reports that long-term mortality of patients withnontransmural infarction is similar to that of trans-mural infarction. 8. 19 A major observation of this studyis the documentation by objective enzymatic criteria ofthe high incidence of early recurrent infarction, themajority within 14 days of the initial infarction. Duringthis interval, the mortality markedly increased in pa-tients with reinfarction and was significantly greater (p< 0.05) than in patients with nontransmural infarctionwho did not have a recurrence. Thus, the disparitybetween acute and late mortality in patients with non-transmural infarction may be due to the high incidenceof recurrent infarction. Objective documentation of a43% incidence of early recurrent infarction during theinitial 2 weeks after infarction should discourage earlydischarge for a seemingly uncomplicated and benignsubset of patients. The mechanism of the instability ofnontransmural infarction is not known.Our data indicate the unstable nature of patients who

have electrocardiographic evidence of nontransmuralmyocardial infarction. These patients are prone to rein-farction, which has a deleterious effect on prognosissuch that the mortality at 9 months was similar to thatof patients with transmural infarction. Of even greatersignificance was the finding that in addition to type ofinfarction, gender, obesity, and repeated episodes ofchest pain are also associated with increased risk ofdeveloping recurrent infarction. The validity of theclinical predictors was confirmed by the ability to iden-tify 80% of the patients with reinfarction correctly inthe test set based on regression coefficients derivedfrom the initial training population. With this ap-proach, one can assess the efficacy of prophylactictherapy in a new population with a relatively smallsample size.

AcknowledgmentThe authors thank Elizabeth Galie and Connie Large for their techni-

cal assistance and Joyce Kalaych and Charlene Brame for preparation ofthe manuscript.

References1. Rosenbaum FF, Wilson FN, Johnston FD: Changes in the precor-

dial electrocardiogram produced by extension of anteroseptal myo-cardial infarction. Am Heart J 30: 11, 1945

2. Reid PR, Taylor DR, Kelly DT, Weisfeldt ML, Humphries JO,Ross RS, Pitt B: Myocardial-infarct extension detected by precor-dial ST-segment mapping. N Engl J Med 290: 123, 1974

3. Rothkopf M, Boemer J, Stone MJ, Smitherman TG. Buja LM.Parkey RW, Willerson JT: Detection of myocardial infarction ex-tension by CK-B radioimmunoassay. Circulation 59: 268. 1979

4. Fraker TD. Wagner GS, Rosati RA: Extension of myocardial in-farction: incidence and prognosis. Circulation 60: 1126, 1979

5. Gillespie TA, Ambos HD, Sobel BE, Roberts R: Effects of dobuta-mine in patients with acute myocardial infaraction. Am J Cardiol39: 588, 1977

6. Strauss HD. Macdonald IL, Orr SA, Berman ND, Buda AJ: Myo-cardial infarct extension: diagnosis, incidence, and clinical signifi-cance. (abstr) Am J Cardiol 47: 418, 1981

7. Marmor A, Sobel BE. Roberts R: Factors presaging early recurrentmyocardial infarction ("extension"). Am J Cardiol 48: 603, 1981

8. Marmor A, Sobel BE, Roberts R: Prospective characterization ofextension of infarction detected with plasma MB CK. (abstr) Cir-culation 62 (suppl III): 111-81, 1981

9. Marmor A, Fukuyama T, Roberts R: Integrity of creatine kinaseisoenzymes maintained in a frozen state for up to three years.(abstr) Clin Res 27: 617A. 1979

10. Rosalki SB: Improved procedure for serum creatine phosphokinasedetermination. J Lab Clin Med 69: 696, 1967

11. Sherwin AL, Sieber GR, Elhilali MM: Fluorescence technique todemonstrate creatine phosphokinase isoenzymes. Clin Chim Acta17: 245, 1967

12. Henry PD, Roberts R, Sobel BE: Rapid separation of plasma cre-atine kinase isoenzymes by batch adsorption with glass beads. ClinChem 21: 844, 1975

13. Miller JP, McCrate MM. Province MA, Wette R: Maximum likeli-hood estimation of the multivariate logistic. Proceedings of the 3rdInternational Congress of the S.A.S. Users Group, 1978, pp 303-305

14. Halperin M, Blackwelder WC, Werter J: Estimation of a multivar-iate logistic risk function: a comparison of the discriminant func-tion and maximum likelihood approaches. J Chronic Dis 24: 125,1971

15. Konttinen A, Sommer H: Specificity of serum creatine kinaseisoenzymes in diagnosis of acute myocardial infarction. Br Med J1: 386, 1973

16. Roberts R, Gowda KS, Ludbrook PA. Sobel BE: The specificity ofelevated serum MB CPK activity in the diagnosis of acute myocar-dial infarction. Am J Cardiol 36: 433, 1975

17. Roe CR, Limbird LE, Wagner GS, Nerenberg ST: Combinedisoenzymes analysis in the diagnosis of myocardial injury: applica-tion of electrophoretic methods for the detection and quantitation ofthe creatine phosphokinase MB isoenzyme. J Lab Clin Med 80:577, 1972

18. Cannon DS, Levy W, Cohen LS: The short and long-term progno-sis of patients with transmural and non-transmural myocardial in-farction. Am J Med 61: 452. 1976

19. Madigan NP, Rutherford BD, Frye RL: The clinical course, earlyprognosis and coronary anatomy of subendocardial infarction. AmJ Med 60: 634, 1976

20. Geltman EM, Ehsani AA, Campbell MK, Schechtman K, RobertsR, Sobel BE: The influence of location and extent of myocardialinfarction on long-term ventricular dysrhythmia and mortality. Cir-culation 60: 805. 1979

421

by guest on April 28, 2018

http://circ.ahajournals.org/D

ownloaded from

A Marmor, E M Geltman, K Schechtman, B E Sobel and R RobertsRecurrent myocardial infarction: clinical predictors and prognostic implications.

Print ISSN: 0009-7322. Online ISSN: 1524-4539 Copyright © 1982 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation doi: 10.1161/01.CIR.66.2.415

1982;66:415-421Circulation. 

http://circ.ahajournals.org/content/66/2/415.citationthe World Wide Web at:

The online version of this article, along with updated information and services, is located on

  http://circ.ahajournals.org//subscriptions/

is online at: Circulation Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer information about this process is available in the

located, click Request Permissions in the middle column of the Web page under Services. FurtherEditorial Office. Once the online version of the published article for which permission is being requested is

can be obtained via RightsLink, a service of the Copyright Clearance Center, not theCirculationpublished in Requests for permissions to reproduce figures, tables, or portions of articles originallyPermissions:

by guest on April 28, 2018

http://circ.ahajournals.org/D

ownloaded from