Clinical and angiographic predictors of new total coronary occlusion in coronary artery disease: Analysis of 313 nonoperated patients

Clinical and Angiographic Predictors of New Total Coronary Occlusion in Coronary Artery Disease: Analysis of 313 Nonoperated Patients ALAIN MOISE, MD, JACQUES LESPI~RANCE, MD, PIERRE TH#ROUX, MD, YVES TAEYMANS, MD, CLAUDE GOULET, MD, and MARTIAL G. BOURASSA, MD

A new coronary artery occlusion was found in 98 of 313 consecutive patients (31% ) with coronary artery disease treated medically who underwent catheterization twice, 39 -4- 25 months apart. Multivariate logistic regression displayed 8 independent predictors of new occlusion. Four were available at the time of the second angiogram: the interval between the 2 studies (p = 0.005), a decrease in ejection fraction (p <0.01), the appearance of bundle branch block (p <0.01), and an interim myocardial infarction (p <0.05). Four other predictOrs were found at the time of the first angiogram: 2 anglographic characteristics, 1 related to the severity (presence of an 80% or greater luminal diameter narrowing of an artery supplying a non-akinetic left ventricular segment [p <0.005]) and 1 to the extent (count of the lesions narrowed 75 % or

less in luminal diameter in a 15-segment coding system [p <0.05]) of coronary artery disease, and 2 risk factors: smoking status (p <0.05) and male sex (p <0.05). The 140 male smokers with at least 80 % diameter stenosis or at least 4 segments with moderate (75 % or less) stenosis were at a higher risk of occlusion than the 173 other patients after intervals of less than 2 years (13 of 53 vs 7 of 74, p <0.001), 2 to 4 years (23 of 40 vs 10 of 47, p <0.005) and more than 4 years (27 of 47 vs 18 of 54, p <0.05). Thus, the appearance of a new occlusion, while strongly associated with new myocardial damage, can be predicted by a combination of 2 angiographic and 2 clinical characteristics at the time of the first evaluation.

As evaluated by angiography, the course of coronary artery disease (CAD) is progressive. 1 The definition of progression may vary, but usually includes the occurrence of a new occlusion as a qualifying criterion for progression. 2,3 Among the whole spectrum of arterial lumen patency, total occlusion is certainly a most identifiable and hemodynamically significant change. However, its pathophysiologic significance may differ from the other types of progression4: In patients with transmural myocardial infarction (MI), new occlusion is almost always found during the early hours5 and remains visualized in about 50% of patients in the chronic

stage. 6 Thus, the distinctive clinical and angiographic correlates of occlusion may be hidden by including occlusion in progression. A previous report from our institution summarized7 clinical and angiographic correlates of progression (without occlusion) in 313 medically treated patients who underwent catheterization twice. In the present investigation, we reanalyzed the data to identify clinical and angiographic patterns in patients with total occlusion.

(Am J Cardiol 1984;54:1176-1181)

Methods Patients: The files of patients who underwent catheterization twice, at least 3 months apart and without previous or interim coronary artery bypass surgery or angioplasty, were reviewed. Only patients with at least I narrowing of at least 50% of the luminal diameter of a coronary artery were considered. From January 1, 1970, to May 1, 1982, 362 such patients underwent repeat coronary angiography. Forty-nine patients were excluded from the study because of poor quality of the angiograms or inadequate matching of the view angulations. Thus, the data reported herein are from 313 patients who underwent repeat catheterization for various clinical

From the Montreal Heart Institute, Montreal, Qu6bec,Canada.This study was supported in part by a Fellowship award to Dr. Moise from the INSERM (Institut National de la Sant~ et de la Recherche IV~dicale),Paris, France, and the FRSQ (Fonds de la Recherche en Sante du Quebec), Quebec, Canada. Manuscript received May 31, 1984; revised manuscript received July 31, 1984, accepted August 1, 1984. Address for reprints: Martial G. Bourassa, MD, Department of Research, Montreal Heart Institute, 5000 east, B61angerStreet, Montreal, Quebec, Canada, H1T 1C8. 1176

December 1, 1984 THE AMERICAN JOURNAL OF CARDIOLOGY Volume54

reasons, including persistent stable angina (160 patients), unstable angina (52 patients), recent MI (15 patients) and research protocol (40 patients). For each patient, the cholesterol level, systemic blood pressure, anginal class s and electrocardiogram9 were recorded at the first and the second studies. The diagnosis of previous or interim MI was based 1° on the presence of at least 2 of the following 3 criteria: chest pain of at least 30 minutes in duration, appearance of new Q waves or of evolving ST-segment changes on the electrocardiogram, and elevation of the enzyme creatine kinase or glutamic oxaloacetic transaminase to at least twice the upper limit of normal. The 65 patients with an interim MI underwent repeat catheterization 2 days to 4 years after the clinical event (8 patients within I month and 57 patients after I month). Assessment of coronary artery disease: Selective coronary arteriography was performed by techniques previously described. 11 The luminal diameter narrowing was assessed in a predefined 15-segment coding system. 12 The severity of CAD was appraised along the number of arteries stenosed at least 70%, the Friesinger score, 13 the Gensini score 14 and the presence (vs absence) of an 80 to 99% stenosis jeopardizing a non-akinetic left ventricular (LV) segment. For this purpose, the diaphragmatic and posterobasal segments were related to the left circumflex and (dominant) right coronary arteries; the 3 anterior LV segments were related to the left anterior descending artery and its diagonal branches. The extent of CAD was quantified by the extent score,7 defined as the count of all segments stenosed 75% or less in the 15-coronary-segment coding system. Both films were simultaneously projected and compared. Occlusion was defined by the appearance of a new complete obstruction in a segment previously patent. Progression was considered separately and defined as at least a 20% increase in stenosis in a segment narrowed 50% or more, or at least a 30% increase in stenosis in a segment with less than 50% initial obstruction. 1.7 Left ventriculograms were recorded in 30 ° right anterior oblique projection and analyzed by dividing the LV contour in 5 segments, s Ejection fraction was calculated for all patients by the area-length method. Statistical methods: Univariate comparisons between the group of patients with occlusion and the group without occlusion were performed by the chi-square test and the t test. To identify the best set of independent predictors of occlusion, the data were fit by a multivariate logistic mode115: p = {1 + exp(-a - blxl - b2x2 . . . . . . -bnxn)} -I, where p is the probability of occlusion and xl, x2...Xn the n variables retained in a stepwise manner. Maximum likelihood method was used for the calculation of a, bl, b2...bn. At each step, the additional contribution of the next selected variable was assessed by a likelihood ratio test. Some interactions between variables were also tested. 16 Goodness of fit of the final model was checked by Hosmer's statistic. 17 Results

P a t i e n t characteristics (Table I): At the time of the first angiogram, the population consisted mainly of young patients (mean age 48 years) with good L V function (mean ejection fraction 59%). Thirty-five patients had moderate s (50 to 69% stenosis) CAD; 278 had significant 8 (->70% stenosis) CAD: 126 had 1-vessel, 110 had 2-vessel and 42 had 3-vessel CAD. On the first angiogram, 3,903 coronary segments could be analyzed (12.4 per patient), of which 3,465 were patent; 116 of these segments showed a new occlusion

TABLE I

1177

Characteristics of the 313 Patients at the Time of the First and Second Angiograms

Variable

Sex Male Female Age (yr)* Angina functional class t 0 1-2 3-4 Not exertional Previous myocardial infarction Interim myocardial infarction No, of vessels with > 7 0 % stenosis 0 1 2 3 Friesinger score" Gensini score" Extent score •t Presence of >1 80 to 99% stenosis§ Progression New occlusion Ejection fraction ° New akinesia

1st Angiogram

2nd Angiogram

263 (84 %) 50 (16%) 48 4- 8 51 4- 8 23 (7%) 146 (47%) 101 (32%) 43 (14%) 149 (48 % ) ...

18 (6%) 91 (29%) 168(54%) 36 (11%)

35(11%) 126 (40%) 110(35%) 42 (13%) 7.3 4- 2.9 36 -I- 32 3.4 4- 2.0 142 (45%)

21(7%) 91 (29%) 113(36%) 88 (28%) 8.7 4- 3,1 52 4- 37 3.1 4- 1.8 144 (46%)

...

139 (44%) 98 (31%) 55 4- 14 53 (17 % )

59"~-" 11 ...

65'(2:1% )

• Values are mean 4- standard deviation. 1"Canadian Cardiovascular Society definition,e Count of the _<75% stenosed segments on a 15-segment coding system. Supplying a non-akinetic ventricular wall segment.

on t h e r e p e a t angiogram. T h e 98 patients with at least 1 newly occluded segment constituted the "occlusion group."

Comparison of the occlusion and n o n o c c l u s i o n groups (Tables II and III): T h e chances for occlusion increased with the time interval between studies (p <0.001), b u t were unrelated to the year of the first angiogram. Occlusion was more frequent among male (92 of 263) t h a n female (6 of 50) patients (p <0.001). T h e mean age was the same in the occlusion and the nonocclusion groups. Also, the initial cholesterol levels in these groups did not differ significantly (243 vs 236 mg/dl, p ffi 0.25), nor did the frequency of a history of hypertension (31% vs 30%). However, occlusion was more f r e q u e n t in patients who were current smokers at the time of the first (77 of 202 vs 21 of 111, p <0.001) and second (67 of 179 vs 31 of 134, p <0.01) angiograms. Occlusion was strongly associated with interim MI, a p p e a r a n c e of h e a r t failure, electrocardiographic indexes of myocardial damage, new akinesia and decrease in ejection fraction. Occlusion was related neither to the worsening of anginal symptoms nor to the angiographic progression of CAD. N u m b e r of diseased vessels, Friesinger score and Gensini score were not informative a b o u t the chances of s u b s e q u e n t occlusion. Occlusion occurred more freq u e n t l y in patients with at least 1 stenosis of 80% or more in a vessel supplying a functional LV segment (56 of 142) t h a n in those without (42 of 171) (p <0.01). T h e e x t e n t score (count of the <75% stenoses) was higher in the occlusion t h a n in the nonocclusion group (3.92 ± 2.10 vs 3.20 ± 1.86, p <0.01); this difference could not

1178

PREDICTIONOF CORONARY OCCLUSION

T A B L E II

Clinical and Electrocardiographic Predictors of Occlusion Occlusion (98 pts)

Variable Age (yr)* Men Year of the first angiogram° Interval between studies (mo)* Smokers (1st catheterization) Smokers (2nd catheterization) Interim myocardial infarction Progression of angina functional class Heart failure (2nd catheterization) New abnorm01 Q wave New resting ~ST (>- 1 mm) New bundle branch block

47 4- 7 92 (94%) 1974.6 4- 2.7 46 4- 25 77 (78%) 67 (68%) 38 (39%) 43 (44 % ) 12 (12%) 26 (27%) 20 (20%) 10(10%)

No Occlusion (215 pts) 48 4- 8 171 (80%) 1974.8 4- 2.5 35 4- 24 125 (58%) 112 (52%) 27 (13%) 80 (37 %) 8 (4%) 12 (5%) 17 (8%) 3(1%)

p Value NS <0.001 NS <0.001 <0.001 <0.01 <0.001 NS <0.01 <0.001 <0.001 <0.001

• Values are mean -I- standard deviation. NS = not significant.

T A B L E III

Angiographlc Predictors of Occlusion

Variable No. of diseased vessels (>--70% stenosis) Friesinger score Gensini score Presence of >--1 stenosis 80 to 99% in diameter °S Extent scores Progression • New akinesia • Ejection fraction (1st cath) Change in ejection fraction

Occlusion No Occlusion (98 pts) (215 pts) (Mean -I- SD) (Mean 4- SD)

p Value

1.5 4- 0.8

1.5 4- 0.9

NS

7.4 4- 2.7 33 4- 29 56 (57%)

7.3 4- 3.0 38 4- 33 86 (40%)

NS NS <0.01

3.9 4- 2.1 49 (50%) 33 (34%) 60 4- 11 - 9 4- 12

3.2 4- 1.9 90 (42%) 20 (9%) 58 4- 12 --2 4- 10

<0.01 NS <0.001 NS <0.001

Number (%). 1 Supplying a non-akinetic ventricular wall segment. t Count of <75% stenosed segments in a 15-segment coding system. NS = not significant; SD = standard deviation. •

be explained by the number of 70 to 75% stenoses (0.59 4- 0.68 vs 0.55 + 0.59, difference not significant [NS]). Multivariate analysis (Table IV): Eight variables were retained by the stepwise logistic regression. The most predictive was an index of severity of the disease: the presence of an 80 to 99% stenosis jeopardizing a non-akinetic LV segment (p <0.005). Another initial angiographic characteristic, the extent score, was also retained, although it was less predictive (p <0.05). Attempts to model an interaction term between the presence of a stenosis of 80% or more and the extent score were not significant, suggesting that the additive model adequately describes the present data. The other variables retained in the logistic regression equation were the time interval between studies (p = 0.005), 3 indexes of myocardial damage (decrease in ejection fraction, new bundle branch block and interim MI) and 2 clinical baseline characteristics (smoking status and sex). The coefficients bi from the logistic model were used to compute a calculated probability, which was plotted against the observed proportion of occlusion (Fig. 1). Cases of occlusion were counted by fixed cutpoint intervals. Hosmer's statistic (X2 = 6.15, df = 8, p = 0.63, NS) suggested a good agreement between observed and predicted frequencies of occlusion.

Analysis with variables available at the first evaluation: The 4 baseline characteristics related to occlusion by multivariate analysis were combined to identify a high-risk group. The high-risk patients were defined as the male smokers with at least 1 stenosis of 80% or more jeopardizing a contractile LV segment, or an extent score of 4 or more. The high-risk group (140 patients) had a higher incidence of occlusion than the low-risk group (173 patients) for intervals of less than 2 years (13 of 53 vs 7 of 74, p <0.001), 2 to 4 years (23 of 40 vs 10 of 47, p <0.005) and more than 4 years (27 of 47 vs 18 of 54, p <0.05) (Fig. 2). Interim MI (42 of 140 vs 23 of 173, p <0.001), new akinesia (31 of 140 vs 22 of 173, p <0.05) and a 10% or greater decrease in ejection fraction (47 of 140 vs 35 of 173, p <0.01) were also more frequent in the high-risk group, suggesting that the predicted occlusions really contributed to the worsening of LV function. Occlusion by segment: The locations of the 116 occluded segments are listed in Table V. The incidence of occlusion according to the initial degree of diameter narrowing is listed in Table VI. Of the 98 patients who had occlusion, 56 had an 80 to 99% stenosis jeopardizing a non-akinetic LV segment; in these patients, the 69 recorded occlusions occurred in 32 cases at the site of the high-grade stenosis, in 10 cases at the segment immediately distal to the severe stenosis and in 27 cases at a remote lesion. Of the 24 new incidences of akinesia registered in patients with an 80 to 99% stenosis, 18 were located in an LV segment jeopardized by the severe stenosis. Discussion

This study focused on the correlates of new occlusions in a large series of patients who underwent recatheterization 39 4- 25 months after their first angiogram. Occlusion was related to several signs of interim myocardial damage. Risk for occlusion could be evaluated before such damage, using a combination of anglographic and clinical baseline characteristics. This result may have implications for risk stratification and therapeutic interventions in CAD. Limitations of the study: Repeated appraisal of the native coronary circulation was necessarily limited to survivors eventually suitable for bypass surgery, but

December 1, 1984 THE AMERICAN JOURNAL OF CARDIOLOGY Volume54

TABLE IV

1178

Summary of Stepwlse Logistic Regression X2

p Value

Bi

Odds Ratio

95 % CL for Odds Ratio

Difference Associated with Odds Ratio

10.44

0.002

0.458

2.50

1.42-4.39

7.87 7.72 7.17 6.53 5.87 5.63 5.09

0.005 0.006 0.007 0.011 0.015 0.018 0.024

0.0148 -0.039 1.046 0.443 0.171 0.423 0.530 -- 1.666

1.56 1.48 8.10 2.43 1.41 2.33 2.87

1.10-2.22 1.12-1.94 1.55-42.36 1.23-4.78 1.06-1.87 1.13-4.81 1.07-7.81

Presence of 1 or more stenosis vs absence of stenosis 60 vs 30 months 10% decrease vs unchanged Occurrence vs absence Occurrence vs absence 4 vs 2 Smoker vs nonsmoker Male vs female

Variable Entered 80 to 99% stanosls supplying a non-akinetlc ventricular segment (--1, -I-1) Time interval between studies (me) Change in ejection fraction (%) New bundle branch block ( - 1 , +1) Interim myocardial infarction ( - 1 , -I-1) Extent score* Smoking status ( - 1 , +1)* Sex (male) ( - 1 , +1) Constant a

• At the first catheterization. Dichotomic variables are noted when absent ( - 1 ) and when present (+ 1). Chl-square values are based on likelihood ratio test and should be compared with 1 degree of freedom chi-square dlstribuUon. CL = confidence limits.

who did not undergo operation, on the basis of their first angiogram. Among such patients, the physician requested a repeat catherterization, most often on the basis of a changing clinical pattern. Therefore, our patients may not be representative of the whole natural history of CAD. Although occlusion is almost always found in the early hours of an acute transmural MI, 5 it is less frequently found later. 6 Transient occlusion with subsequent repermeabilization could not be detected in the present study; conversely, some occlusions may have been preceded by 1 or several stages of disease progression. Serial (at least 3) cineangiographic studies may provide more detailed information on these issues.

N= 125 lee . .

14

94 .

52 .

28

14

.

75-

To determine the relation between coronary lesions and LV function, biplane angiography and quantitative is assessment of the LV wall motion are needed. The angiographic assessment of CAD was based on a visual appraisal of coronary stenoses and not on computerassisted measurements, z9 Also, eccentricity and length 2° of the stenoses could be taken into account in future studies. The robustness of our models should be tested in a prospective manner, in a new set of data. 21 Occlusion vs progression: In a previous analysis of the same set of data, 4 independent predictors of progression were identified7: age and extent score at the time of the first angiogram, time interval between studies, and unstable angina at the time of the second angiogram. Thus, occlusion and progression have distinctive correlates for risk factors (smoking status vs age), clinical events (MI vs unstable angina) and angiographic baseline characteristics (at least 80% stenosis and, to a lesser degree, extent score, vs extent score alone). The occurrence of progression and occlusion was statistically independent (Table III). These facts

m

0 B

S E R V E D

%

50-

0 C C L U

m

25p--

m

e

i

20

I

40

I

68

I

80

E

I

188

PREDTCTED X FIGURE 1. Predicted vs observed probabilities of occlusion. Using the estimated logistic function, a predicted probability of occlusion was computed for each patient. Patients were therefore ordered into classes defined by fixed cutpolnts (0.2, 0.4...) of the predicted probability of occlusion. In each class, patients with and without occlusion were counted, providing an observed rate of occlusion (hatched bars). The class size is Indicated at the top of each bar. This illustration is a graphic check of the accordance between observed and predicted probabilities of occlusion.

0 N

'°°I

LOW-RISK HIGH-RISK

p<.001

p<.005

p<.05

se-

O

N=

:40:

<2

2-4

:47:

>4 YEARS

FIGURE 2. Occlusion in low-risk and high-risk groups of patients. The groups were defined on the basis of the 4 baseline characteristics associated-by multivariate analysis--to subsequent occlusion. The high-risk group Included the 140 male smokers with an 80 to 99% stenosis supplying a non-akinetic ventricular segment, or an extent score of 4 or higher. The size of each group at each interval is indicated at the bottom of each bar.

1180

PREDICTIONOF CORONARY OCCLUSION

TABLE V

Coronary Arterial Occlusions In 98 PaUents No. of Occlusions

Proximal (preseptal) left anterior descending Mid-left anterior descending Distal left anterior descending First diagonal Second diagonal Proximal left circumflex Distal left circumflex First obtuse marginal Second obtuse marginal Proximal right Distal right Right posterior descending Crux area and posterolateral branches Total

TABLE Vl

10 21 2 6 1 7 12 9 4 30 11 2 1 116

Incidence of Occlusion According to the Initial Degree of Obstruction (Analysis by Segments)

% Diameter Stenosis on 1st Angiogram 0 (normal) _<75 % 80 to 99% supplying an akinetic LV segment 80 to 99% supplying a non-akineUc LV segment

No. of Segments at Risk

No. of Segments Occluded (2nd Angiogram)

Occlusion

2,155 1,073 38

17 66 1

0.8 % 6.2 % 2.6%

199

32

16.0%

Incidence

of

LV = left ventricular.

suggest, as a working hypothesis, that progression and occlusion may have distinctive pathophysiologic mechanisms. This possibility should be kept in mind for the choice of end points in intervention trials. Risk factors and occlusion: CAD is a multistage illness. 22 Although highly predictive of the presence of the CAD, the classic lifestyle-related risk factors, smoking status, hypertension and cholesterol level, were unrelated to the progression of coronary atherosclerosis in our study 7 and other studies. 2,3 In the present investigation, smoking status was associated with the occurrence of coronary occlusion. This is in accordance with experimental data23; also, a clinical study suggested that reinfarction was more frequent in persistent smokers. 24 Moreover, the Framingham study demonstrated a strong relation between smoking status and occurrence of MI, b u t a weak one, if it exists, with the appearance of angina pectoris. 25 We have no definite explanation for the lower incidence of occlusion in female patients. A bias selection is possible, as it is in any observational study. For example, differences in occupational status may reduce the chances for recatheterization in women likely to have occluded a coronary vessel. Further investigations are needed to clarify the relations between sex, hormonal status and occurrence of thrombosis in CAD. Occlusion and left ventricular function: In this study, the occurrence of a new occlusion was strongly associated with interim MI and with electrocardiographic 2s and angiographic indexes of myocardial damage. These findings are consistent with the ira-

paired global6 and regional 27 LV function in the chronic stage of MI when the coronary artery supplying the infarct zone remains occluded. Angiographic predictors of occlusion: In this study, the presence (vs absence) of a stenosis of 80% or greater jeopardizing a non-akinetic LV segment was the most powerful predictor of subsequent occlusion (p <0.005 by multivariate analysis). Although clustering by patients precludes a formal analysis on the population of segments, data from Table VI are also consistent with an increased risk of occlusion for 80% or greater stenoses in a vessel supplying a contractile LV segment. This enhanced risk of occlusion for high-grade stenosis is supported by previous reports. 2s,29 The extent score, previously found to be a powerful predictor of progression (p <0.001 by n~ultivariate analysis), 7 was also predictive of occlusion, b u t to a lesser degree (p <0.05). In a recent review, Seizer 3° stated that "the present classification of coronary artery disease on the basis of vessels involved by the disease is inadequate." Whether or not it is adequate for risk stratification (for survival) or indications for procedures is beyond the scope of the present paper. However, the classic indexes--number of diseased vessels, Gensini score and Friesinger score--were not helpful in predicting occlusion or progression. 2,3,7 All these indexes are based on hemodynamic concepts and overweight the already occluded vessels, including those supplying an akinetic LV wall; these lesions are usually stable and not progressive. So long as our findings will be confirmed, the presence (vs absence) of a stenosis of at least 80% in a vessel supplying a contractile LV segment and the extent score could provide valuable risk stratification in clinical trials with angiographic endpoints. Implications of this study: In this study, 2 anglographic and 2 clinical characteristics allowed the identification of patients at high risk of coronary occlusion. Intervention trials aimed at preventing occlusion would require a reasonable number of patients in this high-risk subset. 3° Since the high-risk patients had an accelerated worsening of their LV function, the end point of such trials could be either the occlusion itself or the residual LV function.

Acknowledgment: We thank Louise Lafrance (Universit~ de Montreal), Chantal Grenier (Ecole Polytechnique de Montreal) and Jean Gauthier (Institut de Cardiologie de Montreal) for their valuable help in the management of the data; we are indebted to C~line Germain, Louise Beaudry, L~opold Dupuis and the staff of our Archives department for their skillful assistance. We also thank Louise G. Demers and Monique Masse for their secretarial assistance.

References 1. Bourassa MG, Ryan TS, Goulet C, Abramovltz BA, Lesl~rance J, Faxon DP, EnJalbertM, MoCabe C. Early progressionof coronary disease in patients assigned to medical therapy (abstr). Circulation 1981;64:suppl IV: IV-83. 2. Bruschke AVG, WlJersTS, Kolsters W, LandmanJ. The anatomicevolution of coronary artery diseasedemonstratedby coronary arterlographyin 256 nonoparatedpatients. Circulation 1981;63:527-536. 3. Kramer JR, Kltazume H, Proudfll WL, MatsudaY, Williams GW, Sones FM Jr. Progressionand regression of coronary arteriosclerosis: relation to risk factors. Am Heart J 1983;105:134-144.

December 1. 1984 THE AMERICAN JOURNAL OF CARDIOLOGY Volume 54

4. Seizer A, Paslemak RC. Role of coronary arterlography in the evaluation of patients with coronary artery disease. Am J Meal 1981;70:747-751. 5. Deweed MA, Spores J, Notske R, Mouser LT, Burroughs R~Golden MS, Lang HT. Prevalence of total coronary occlusion during the early hours of transmural Infarction. N Engl J Med 1980;303:897-902. 6. Bertrand ME, Lefebwe JM, Lalane CL, Rousseau ME, Carre AG, Leklalfre JB. Coronary arterlography in acute transmural myocardial infarction. Am Heart J 1979;97:61-69. 7. Molas A, "llt~oux P, Taeymans Y, Lelp#rance J, FIn6s P, Waters DD, Delmolnga B, Robed P. Clinical and angiographlc factors associated with progression of coronary artery disease. JACC 1984;3:659-667. 8. Principal Investigators of CASS and Their Associates. The National Heart, Lung, and Blood Institute CoronaryArtery SurgeryStudy(CASS). Circulation 1981;63:suppl 1:1-1-1.31. 9. Rose GA, Blackburn H. Electrocardiographic grading codes. Circulation 1973;47:suppl 1:1-39-1-42. 10. Bloer CM, Roaske WR, O'Roud(e RA. Clinical pathologiccorrelates in acute myocardial infarction. Adv Cardlol 1978;23:14-24. 11. Bour__,J,~__MG, Leap~ance J, Campaeu L Selective coronary arteriography by the percutaneous femoral artery approach. Am J Rcontgenol 1969; 107:377-383. 12. Auslen WG, Edwards JE, Frye RL, Genslnl GG, Gott YL, Grlffllh LSC, McGoon DC, Murphy MC, Roe BIB.A reportingsystem on patientsevaluated for coronary artery disease. Report of the Ad Hoc Committee for Grading of Coronary Artery Disease, Council on CardiovascularSurgery, American Heart Association. Circulation 1975;51:suppl V:V-7-V-40. 13. Humphdas JOIN,Kuller L, Ross RS, Frleslnger GC, Page EE. Natural history of ischemic heart disease in relation to arterlographic findings. A twelve year study of 224 patients. Circulation 1974;49:489-497. 14. Genslnl GG. Coronary Arteriography. Mt. Kisco, NY: Futura, 1975:272. 15. Cox DR. The Analysis of Binary Data. London: Methuan, 1970:91. 16. Breslow NE, Day NE, eds. Statistical Methods In Cancer Research. Vol. I. The Analysis of Case-Control Studies. Lyon: International Agency for Research on Cancer, 1980:192-246. 17. Lemeshow S, Hoerner DW. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemlol 1982; 115:92-106. 18. Gelber9 HJ, Brundage BH, Glanz S, Parmley WW. Quantitative left ventflcular wall motion analysis: comparisonof area, chord and radial meb'lods.

1181

Circulation 1979;59:991-1000. 19. Brown BG, Bolso, EL, Dodge FIT.ArterJog~aphlcassessment of coronary arteriosclerosis. Review of current methods, their limitations and clinical applications. Arteriosclerosis 1982;2:2-15. 20. Rosch J, Antonovlc R, Treeeulh R$, Rahlmtoola H, Sire DN, Doffer CT. The natural history of coronary artery stonosls. A longitudinalanglographlc assessment. Radiology 1976;119:513-520. 21. McNeil BJ, Haeley JA. Statistical approachesto clinical predictions.N Engl J Mad 1981;304:1292-1294.

22. W..k,twan.E,Xrop~ .R.kPrye.RL, S~hSU(,T,~FE, K, apZ,ZT.Factors

23. 24.

2S. 26. 27. 28.

29.

30.

affecting me extent aria severity of coronary artery disease In patients enrolled in the coronary artery surgery study. Arteriosclerosis 1982;2: 208-215. Folta JD, Boeebrake FC, The effects of cigarette smoke and nicotine on platelet thr.omb.usformation In stenosad dog coronary arteries: Inhibition with phentoaamlne.Circulation 1982;65:465-470. Redmllzer PA, CunalnghamDA, DonnarAP, AndrewGM, SUd(CW, Jonas NL, Kavanagh T, Oldrldge NB, Parker JO, Shepherd RJ, Sutton JR. ~ ¢ t e r l s t l c s Ihat predicted recurrence of Infarction within 3 years in the Ontario Exercise Heart Collaborative StUdY.Can IViedAssoc J 1983;128: 1287-1290. Kennel WB. Cigarettes, coronary occlusions, and myocardial Infarction. JAMA 1981;246:871-872. Sullivan W, Vlodaver 7., Tuna N, Long L, Edwards JE. Correlation of elactrocardiographlcand pathologicfindings in healedmyocardialInfarction. Am J Cardlol 1978;42:724-732. Felld BJ, .Russell RO, Dowlln9 Jr, Rackley CE. Regional left vontricular performance in the year following myocardial Inferctk)n. Circulation 1972;46:679-689. Rafflenbeul W, Smith LR, Rogers WJ, Mealie JA, Racldey CE, Russell RO. Quantitative coronary arterlography. Coronary anatomy of patients with unstable ang na pactorls reexamined 1 year after optima reed cal therapy. Am J Cardlo11979;43:699-707. Kramer JR, K#azume H, Proudflt WL, Matsuda Y, GoonnaMtoM, W l l l l m GW, SoRes,h"FM. Segmentalanalysisof the rate of progressionIn patients with progressive coronary atherosclerosls. Am Heart J 1983;106:14271431. Selser A. On ~e limitation of therapeuticinterventiontrials In Ischernlc heart disease: a clinician's viewpoint. Am J Cerdlol 1982;49:252-255.