- Email: [email protected]
Prognostic significance of progression of coronary artery disease
Prognostic SignificSmce of Progression of Coronary Artery Disease ALAIN MOISE, MD, MARTIAL G. BOURASSA,
MD, PIERRE THiROUX,
MD,
YVES TAEYMANS,
MD, LUCIEN CAMPEAU,
MD,
MD, ANDRE PASTERNAC,
MARC A. BOIS, MD, IHOR DYRDA, MD, MS, and PAUL ROBERT DAVID, MD
Angiographicaily documented coronary artery disease (CAD) can progress. Although progression itself is frequently recognized in patients who have undergone repeat cardiac catheterization, its prognostic significance remains unclear. To evaluate the influence of progression on survival, 313 patients with CA6 who underwent catheterization twice (39 f 25 months apart) Wre followed for 3 to 129 months (mean 41 f 30) after the second angiogram. At the time of the second angiogram, 21,. 91, 113 and 88. patients had 0-, I-, 2- and Wessel CAD, respectively. The mean,ejection fraction (EF) of the group was 55 f 13%. Progression was noted in 139 patients (44 % ). Qf the 3 13 patients, 33 died and 39 had acute myocardial infarction (AMI) durlng follow-up. Four-year survival was estimated at 94% and 83% in the nonprogression and progression groups, respectively. Progression was predictive of survival by (univariate) log-rank test (p
only EF (p
The state of our knowledge on survival in coronary artery disease (CAD) is based on univariatel and multi-
investigation was designed to assess the independent prognostic significance of CAD progression in a large series of patients, when taking into account other important baseline characteristics.
(Am J Cardiol 1985;55:941-946)
variate2B analyses of series of patients. Clinical (age2 and worsening angina3), electrocardiographic (ventricular arrhythmias2s) and angiographic (number of diseased vessels,1-3 ejection fraction [EFJ2and left main coronary artery stenosis1-3) characteristics carry in-
Methods Patients: From January 1,197Oto May 1,1982,313 patients underwent 2 cardiac catheterizations at our institution, without previous interim coronary bypass surgery or coronary angioplasty. The second angiogram was performed for persistent stable angina (181 patients), unstable angina (52 patients), research .protocol (40 patients), recent acute myocardial infarction (AMI) (15 patients) or multiple reasons (25 patients). The interval between the 2 studies varied between 3 and 118 months (mean 39 f 25). The presence of st least one 50% or greater diameter reduction of a major coronary artery at the time of the first angiogram was a requirement for inclusion in this study. Clinical and angiographii data: Previous AM1 was diagnosed if 2 of the 3 following criteria were present: (1) chest pain lasting 30 minutes or more; (2) elevation of creatine ki-
formation on survival. The angiographic variables in these studies are considered as fixed. In a previous study, Bemis et al4 showed that patients with progressive CAD were at higher risk of subsequent death. Little attention was given to this problem since their report. The present From the Department of Medicine and Research C&W of the Montreal Heart Institute, Montreal, Quebec, Canada. Manuscript received September 10, 1984; revised manuscript received January 2, 1985, accepted January 3. 1985. Address for reprints: Martial G. Bourassa, MD, Montreal Heart Instjtute, 5000 East Belanger Street, Montreal HlT lC8, Quebec, Canada. 941
PROGNOSTIC SIGNIFICANCE OF PROGRESSION
942
TABLE I
fg
Characteristics of the 313 Patients at the Time of the Second Angiogram
WI
Previous myocardial infarction Angina class’ 0 1 to 2 3 to 4 Not exertional Ejection fraction (%) Left ventricular aneurysm No6pf diseased coronary arteries
No.
(“/I
51 f8 283 176
(84%) (58%) (8%) (29%)
168
;;:;; 0
55*E 27
(9%) (7%)
1 z 3 Left main diameter reduction X0% 30-50 % Progression Occlusion n
9’: 113 88
:::a (28%)
15 18 139 98
Canadian Cardiovascular Society definition (see reference 13). + Patients with O-vessel disease had at least 1 segment initially narrowed by at least 50 % but less than 70 % . l
nase or glutamic oxaloacetic transaminase enzyme levels to twice the upper normal limit; and (3) characteristic, evolving ST-T changes or appearance of new Q waves.5 The angina functional class was evaluated according to the Canadian Cardiovascular Society definition.6 Coronary arteriography was performed using previously described techniques.7 The coronary angiograms were analyzed in the 15-segment coding system of the American Heart Associations; the percent luminal diameter stenosis was noted in each segment. Disposition between l-, 2- or 3-vesseldisease groups was based on the presenceof a 70%or greater diameter narrowing for each of the 3 main coronary vessels; patients with less than 70% but at least 50% diameter narrowings (moderate CAD according to the Coronary Artery Surgery Study [CASS] definition@) were counted as having 0 diseased coronary arteries. Left ventriculography was performed in the 30’ right anterior oblique projection; EF was computed by the area-length method6; the left ventricular contour was divided into 5 segments6p8;the motion of each segment was analyzed as normal, hypokinetic, akinetic or dyskinetic (aneurysmal).
TABLE II
Outcome Events No. of Pts
Analysis by death Unknown cause Nonvascular, noncardiac cause Vascular, noncardiac cause During coronary bypass surgery Acute heart failure Acute myocardial infarction (fatal) Sudden death Total Analysis by first occurrence of acute myoc :ardial infarction during follow-up Acute myocardial infarction (fatal) Acute myocardial infarction (nonfatal) Subsequently alive Subsequently died Total
4 25’ lo+ 39
Among the 25 AMIs, 7 were perioperative. + Among the 10 AMIs, 1 was perioperative. Three of these 10 patients died from a fatal recurrent acute myocardial infarction. l
Assessment of coronary artery disease progression: The first and second coronary angiograms being simultaneously projected, progression was assessedby a consensus9 of 3 observers on the following criterialo-12: a 20% or greater increase in stenosis in a segment initially narrowed by at least 50%; a 30% or greater increase in a segment with an initial obstruction of less than 50%. Occlusion-100% obstruction of a previously patent segment-was considered separately.” Follow-up: Time 0 for follow-up was defined as the day of the second cardiac catheterization (Fig. 1). Sudden death was defined as death that occurred within 1 hour after the onset of the symptoms,i3 in the absence of the autopsy evidence of very recent AMI. Criteria for AM1 that occurred during follow-up were the same as those for previous IvlI.5 However,. criteria for perioperative AMI was based on electrocardiographic evidence only.14 Death that occurred from acute heart failure without AMP5 was considered separately. Follow-up was completed for 312 of 313 patients for the vital status (99.7%); for 307 of 313 patients for the nature of any hospitalization (98%); and for 31 of 33 deaths for the mechanism of death. Statistical analysis: Survival curves were compared by log-rank test for discrete variables and Cox’s regression model with one P parameter for continuous variables.16 Variables significant by univariate analysis were jointly analyzed by Cox’s regression, which models the probability of survival S(t) at time t by: s(t) = {So(t)p+(flT$ where x is the vector of variables included in the model. The fli coefficients of Cox’s model allow the computation of a relative risk (supposed constant over time: “proportional hazard model”) for the occurrence of the analyzed events. Partial likelihood approach was used for ca1culations.l” Models were compared using a likelihood ratio test. The analyses were repeated for prediction of death, sudden death, AM1 (fatal or nonfatal) and survival without AMI. In the latest analysis, the first occurring event, AMI or death, was considered.
Results At the time of the second cardiac catheterization, most of the patients were symptomatic (Table I). CAD progression was documented in 44% of the patients. After the second angiogram, coronary bypass surgery was performed in 116 patients, and 197 patients were treated medically. Thirty-three patients died during the 3- to 129-month (mean 41 f 30) follow-up period (Table II). Four-year survival was estimated to be 89.5%. During the same follow-up period, 39 patients had AM1 (fatal or nonfatal) and 58 had a major event (AM1 or death). Some patients had 2 or more major events: for example, AMI, recurrent AM1 and next sudden death, but only the first major event was kept for the purpose of analysis. Four-year survival without AM1 was 82.5%. Prediction of survival: Survival was unrelated to the year of the second angiogram (p = 0.56), to (surgical or medical) therapy (p = 0.39) or to the occurrence of occlusion (p = 0.88). The univariate predictors of survival-2 clinical and 6 angiographic characteristics-are listed in Table III; they include progression (Fig. 2). Cox’s analysis displayed only EF (p
April I,1985
TABLE III
Univariate
Predictors of Survival
Variables
l’t
(3.129)
- Clinical data Angiographic findings Coronary vessels Left ventriculogram Progression vs no progression
- Mortality - Myocardial infarction
FtGURE 1. Design of the study. CABS = wronary artery bypass svgery; PTCA = percutaneous transluminal coronary angiOPlaSW
p Values
Variables Progression No. of akinetic left ventricular segments Left ventricular aneurysm No. of diseased coronary arteries Ejection fraction Left main stenosis (% ) 80 to 99% stenosis (n)’ Previous myocardial infarction Men Preseptal left anterior descending artery diameter stenosis (% ) Q wave on ECG Time between 1st and 2nd angiograms
‘00r-, 95 s ” -J
,
:..; I - I.......................1 :..........................i
90
-
?- 85 5 v,
-
80
-
1
l-
P<.01
i....: .. . . . .... ........ ..:
2
0.042 0.049
Supplying a nonakinetic left ventricular segment. p values are based on the log-rank test for dichotomous variables and on the Cox model with one p parameter for continuous variables. l
of diseased arteries (p
i... .. ..... .... -NO PROGRESSION (174 PTS) ..... ....PROGRESSION C 1 39 PTS>
0
1
2
YEARS (1741 (139)
(1491 (951
3
4
OF FOLLOW-UP (126) (751
PATIENTS
(107) (631
(821 (421
AT RISK
FIGURE 2. Cumulative survival of patients with and without progression of coronary artery disease. The follow-up begins at the time of the second catheterization. The comparison is performed by the log-rank test.
Progression was the most striking predictor of survival without AM1 by univariate analysis (p
Predictors of Mortality and Acute Infarction (AMI) (Cox Regression Analysls)
Predicted Events (no) Variable Ejection fraction (%) No. of diseased vessels % stenosis of the left main vessel Progression+
Death (n = 33) -O~&Ov$?OO1) 0:687 (O.Od5) o.oW$;O31)
Sudden Death (n = 15)
AMI (Fatal or Nonfatal) (n = 39)
Death or AMI (All Events) (n = 58)
Not Perloperative Death or AMI’ (n = 47)
-,pd,p;l,“~12) bi$8,93!:042)
P (p value) (0.062) 0.4;;o(O.p17)
-c&~~~o9, o:~~51~;jOo2)
-O$!4?;?$01) 0:~;80@“10)
(50.2)
1.06 (0.002)
0.822 (0.003)
Patients who died during a coronary bypass procedure or who experienced a perioperative myocardial infarction analysis. + Scored 1 when present and 0 when absent. P-estimators of the Cox regression coefficients in each analysis-are given for variables retained at p = 0.05. l
943
Volume 55
41 i 30 MO
313 PTS
Univarlate Predictors of Survlval Wlthout Acute Myocardlal Infarction
Independent
OF CARDIOLOGY
ANGlOGRAM
Znd ANGlOGRAM
p values are based on the log-rank test for dichotomous variables, and on the Cox model with one p parameter for continuous variables.
TABLE V
JOURNAL
p Values
Left ventricular aneurysm Left main stenosis (%) Ejection fraction No. of diseased coronary arteries No. of akinetic segments Q wave on ECG Previous myocardial infarction Progression
TABLE IV
THE AMERICAN
0.757 (0.012)
are not considered
in this
PROGNOSTIC SIGNIFICANCE OF PROGRESSION
944
AM1 (p = 0.59, NS) did not differ between the medical and surgical cohorts; in contrast, the predictive value of progression for occurrence of major events persisted (Table V), suggesting that the overall result was unlikely related to the balance between medical and surgical therapies.
100 2
i..: I..; i......, /....... ...... i ........ .. i
90 2 g
80
p=.0001
/ :
. i ............ i . ....
5
...
70 2
Discussion Progression and acute myocardial infarction: A previous studyi demonstrated that evidence of progression was more striking among patients who underwent repeat cardiac catheterization for an episode of unstable angina than it was among matched patients who had undergone repeat catheterization for stable angina. In the present investigation, progression was found to be the most important predictor of AM1 and, next to the number of diseased vessels,the second most powerful predictor for survival without AMI. This link between the anatomic progression of CAD and the occurrence of coronary events is noteworthy. Two working hypotheses are consistent with this strong relation. First, continuing progression may result in increasingly more severe (80 to 90%) stenoses, with a higher risk of occlusion and subsequent AMI.18 Second, both progression and AM1 could result from an hypothetical “active state” of the disease. Pathophysiologic studies aimed to investigate these hypotheses require the recognition of patients at high risk to progress. In such patients, the balance between risks and potential benefits could allow the performance of sequential angiographic studies on a prospective basis. Progression and death: The striking predictive value for AM1 of CAD progression could be contrasted with the apparent lack of prediction for sudden death. However, this does not mean that continuous evolution
NO PROGRESSION
Is >
60 ;i-,
2
0 1
0
YEARS
4
OF FOLLOW-UP 122
147
174 139
3
2
102 53
65
68 PATIENTS
ii
AT RISK
FIGURE 3. Cumulative survival withoul acute myocardiil infarction (AM) in patlents with and without progression. The comparison is performed by the log-rank test.
p
,Q8
”
,ONE
VESSEL
DISEASE
I
EJECTION I,‘,........ _
FRACTION
.
_
>58%
I
. :
L
. i .... ..- p<‘O’ :... ..
Cl31 PTS) .. . ...... <,@a PTS)
EJECTION
FRACTION
FIGURE 4. Cumulative survival without acute myocardial infarction (AMI) in various subsets of patients. In each subgrcup, survival without AMI was better in patients without than in patients with progression.
(50%
‘1 :....: PC.01 60f---.
Cl01 1100
PTS) PT;
/
1
i . ......... i__ F..... ..-: :. __ C43 PTS) --
0 0
1
2
3
YEARS
4
0
. .
<
39
. ..
PTS)
1
I
I
1
2
3
OF FOLLOW-UP
4
April I, 1985
of coronary atherosclerosis is necessarily of little vital consequence. Because of the smaller number of events (15 sudden deaths vs 39 patients with AMI), only huge differences in sudden death hazard rates in the progression and nonprogression groups could be detected. Also our follow-up was limited, and patients with (nonfatal) AM1 can be exposed to an enhanced risk of mortality in subsequent years. This hypothesis could be tested with a larger series of patients followed up for a longer time. Thus, the size and design of our study did not allow for definite conclusions to be drawn on the pathophysiologic mechanisms of sudden death.lg Limitations of this study: Only nonsurgically treated patients who survived until a second catheterization were eligible for this study and were followed up after the second angiogram. The representativity of this particular group is therefore limited. Progression was appraised by visual comparison, using a conservative definition, in accordance with the precision of the method.g Whether quantitative techniques20 could enhance the detectability of progression is debatable, because, as outlined by Kennedy et al,21 the problem is “how to quantitatively analyze paired films.” However, perfect matching of the angulations could be achieved only in prospective studies. Progression is strongly related to the time,11,22l23 and its occurrence may have been underestimated when the time between the 2 studies was short (less than 2 years).24 This limitation, however, should lead to a conservative estimate of the prognostic significance of progression. Also related to the time interval is the choice of an index of progression; although appealing, the definition of a progression rate (dividing the number of progressions by the time interval between studies) ,could mask the prognostic significance of progression, because the time interval between studies itself was a determinant of prognosis (Table IV) (the longer the time interval, the poorer the prognosis): This type of bias is related to the referral process for the second angiogram, and was previously stressed by Bruschke et al.23The definition of an optimal index of progression would require the repetition of angiograms at fixed time intervals. The year of the second angiogram was unrelated to prognosis: it is unlikely that our results could be biased by secular trends in CAD mortality.25 The nature of treatment (surgical vs medical) was unpredictive of outcome; moreover, discarding the perioperative deaths and AM1 did not substantially modify the results (Table V). The main result of the study persisted in various subgroups defined on the more conventional indexes of prognosis (Fig. 4), suggesting that our result was robust and unlikely to be related to a selection bias. The duration of the increased risk of AM1 in patients who progressed before the second angiogram remains to be evaluated: Longer follow-up periods and use of nonproportional hazard models26 may help to clarify this question. Progressive coronary artery disease-a diagnostic and therapeutic challenge: Spontaneous an-
gina or unstable angina12 are correlated with progression of CAD; recently, new onset angina (a variety
THE AMERICAN
JOURNAL
OF CARDIOLOGY
Volume 55
945
of unstable angina) was recognized as an independent long-term predictor of AMI, despite the low incidence of S-vessel disease in these patients28 With regard to the information provided by a first coronary angiogram, progression correlated better with the number of moderate lesion&s3 than with the severity of CAD. Thus, identification of patients at risk of progression is not necessarily out of range. It is unclear whether progressive and stable CAD warrant different therapeutic approaches. Progression is unrelated to the classic “risk factors,“11p22and perfect vs standard correction of these factors has given a disappointing result in a recent trial with clinical endpoints.2g Calcium antagonists may prevent some experimental models of coronary atherosclerosis.s” We are unaware of any clinical investigation on the effects of these drugs on angiographic endpoints. However, recognition of progressive CAD may provide a subset of patients at high risk of AMI: Design of trials aimed to prevent AM1 and/or slow down the process of atherogenesis should be considered in these patients, together with the reevaluation of risk stratification for conventional procedures and therapies. Acknowledgment: We are indebted to Chantal Grenier (Ecole Polytechnique de Mont&al) for many useful suggestions in the programmingwork. We alsothank JeanGauthier, Nicole Gibeault, Dominique Voyer (Institut de Cardiologie de Montreal), Pham Ngoc Dang and Robert Noel (Ecole Polytechnique de Montreal) for their valuable help in the managementof the data, and Monique Massefor her secreterial assistance.
References 1. Bruschke AVG, Proudflt WL, Scnres FM Jr. Progress study of 590 consecutive nonsurgical cases of coronary artery disease followed by 5-S years. I. Arteriographic correlations. Circulation 1973;47: 1147-I 153. 2. Hamtnerrnelster KE, DaRouen TA, Dodga HT. Variables predlcttve of survival in patlents with coronary disease. Selection by univartate and multivariate analyses from the clinical, electrocardiographic, exercise, arteriographic and quantitative angiographic evaluations. Circulation 1979;59:421-430. 3. Harris PJ, Harrell FE, Lee KL, Befrar VS, Rasatl RA. Survival in medically treated coronary artery disease. Circulation 1979;60:1259-1269. 4. Bemls ChE, Gorlln R, Kemp HO, Herman MV. Progression of coronary artery disease. A clinical arteriographic study. Circulation 1973;47:455463. 5. Bktor cn?, Roe&e WR, O’Rourke RA. Clinical pathologic correlates in acute myocardtal infarction. Adv Cardiol 1978;23:14-24. 6. Principal Investigators of CASS and their Associates. Natbnal Haart, Lung, and Blood Institute Coronary Artery Surgery Study. Ciilatkxt 1981;83:suppl 1:1-l-l-81. 7. Bouraasa MO, Lesp6rance J, Campeau L Selective corm &iogaphy by the percutaneous femoral artery approach. AJR 1969;107:377-383. 6. Austen WG, Edwards JE, Frye RL, Genslnl 00, Gott VL, Grlfflth LBC, McGoon DC, Merphy MC, Roe BB. A reporting system on patlents evakrated for coronary artery disease. Report of the ad hoc commMeeforgradingof coronary artery disease, Council on Cardlovascutar Strgsry. Amsrlcan Heart Association. Circulation 1975;5O:suppl V:V-7-V-40. 9. Ban Marco ME. Brooks BH. Blankenhorn DH. Reoroducibilttv of a consensus panel in the interpretation of coronary angiograms. Am Heart J 1978;96:430-437. 10. Bourassa MO, Ryan TJ, Goulet C, Akamowltz J, Lespbrance J, Faxon DP, Enjalbert M, McCabe CH. Early progression of coronary disease in patients assigned to medical therapy (abstr). Circulation 1981:64:suppl IV:IV-83. 11. Molse A, Theroux P, Taeymans Y, Waters DD, Lespbrance J, Flnes Ph, DescoMgs B, Robert P. Clinical and angiogr hit factors associated with progression of coronary artery disease. JAC 7 1984;3:859-867. 12. Yolse A, Theroux P, Taeymans Y, Descolngs B, Lesparance J, Waters DC, Pelletler GB. Beurasss MG. Unstable angina and progression of coronary atherosclerosis. N Engl J Mad 1983;309:885-689. 13. Goktsteh B. The necessity of a uniform definition of sudden coronary death: witnessed death within 1 hour of the onset of acute symptoms. Am Heart J 1982;103:156-159. 14. Namay DL, Hannnemw lster KE, Zta MB, DeRouen TA, Dodge HT, Namay K. Effect of parioperative myocardial infarction on late survival in patients undergoing coronary artery bypass surgery. Circulation 1982;65:1066-
948
PRDGNOST~C SIGNIFICANCE CF PROGRESSION
1071. 18. Clark LT, Garfeln DB, Dwyer EM. Acute puknonary edema dua to ischemic heat-i disease without accompanying rnyocardial Infarction. Natural history and clinical profile. Am J Med lS83;75:332-338. 18. z;-og Regression models and life tables. J R Stat Sot B 1972;34: 17. Cox DR. &rtial likelihood. Biometrika 1975;82:289-278. 18. Raffbtbul W, Smtth LR, Re9ara WJ, Mantle JA, Rackley CE. Russel RO. Ctuantftative coronary arterlography. Coronary anatomy of patfents with unstable angina pectorfs reexamined 1 yeer after optimal medical therapy. Am J Cardiol 1979;43:899-707. 1s. Davlea MJ, Thomas A. Thrombosis and acute coronary artery lesions in sudden cardiac ischemic death. N Engl J Med 1984;310:1137-1140. 20. Brown BG, Bolson EL, Dodge HT. Arteriogaphic assessment of coronary atherosclerosis. Review of current methods, their limitations, and clinical applications. Arteriosclerosis 1982;2:2-15. 21. Kennady JW, Flaher LD, Kllllp Th. Comnary argiogaphy quality control of the CASS study. In: Bond MG, lnsull W, 01 v S, Chandler AB. Comhill F. eds. Clinical Diagnosis of AtherosclerosT 8: Quantitative Methods of Evaluation. New York: Springer Verlag 1982:475-491. 22. Kramw JR, Matauda Y, Mull&an JC, Aronew M, Preadflt WL. Progression of coronary atherosclerosis. Circulation 1981;83:519-528. 22. ~eAVO,wijrnts,KoksersW,~J.Theanatomicevolution
24. 25. 25. 27. 28. 28. 20.
of coronary artery disease demcnstrated by coronary arteriogaphy in 258 nonoperated patients. Circulation 1981;83:527-538. Rafchten JS, Achuff SC, Burkley BH, Pearaoe TA. Risk factors and angiaphic progression of coronary artery disease (abstr). Circulation s1 82;88:suppl IkIt-283. RyarDB.H~FE,~KL,~rmRM,~RA.Animprwlngprogrosis over time in medically treated patients with coronary artery disease. Am J Cardiol 1983;52:444-448. Stabfefa DM, Carter WH, Wampfer GL. Suvival analysis of drug combinatktns using a hazards mods1with time-dspendent covariates. Biometrics 1980;36:537-548. Dyrda I, Fetltclerc R, Saltlel J, Buurassa MO. Clinical indices of angle graphic progession in coronary artery disease (abstr). Circulation 1973; 47:suppl IV:IV-58. Roberta KB, Caltff RM, Harrell FE, Lee KL, Pryor DB, Reaatl RA. The prognosis for patients with new-onset angina who have undsrgone cardiac catheterization. Circulation 1983;88:970-978. Multtple Risk Factor Intervention Trial Research Group. Muftipte Risk Factor lnterventlon Trial: risk factor changes and mortality resufts. JAMA 198% 248~1485-1477. Kramsch D, Aspen AJ, Rozler W. Atherosclerosis: prevention by a ~,;ffectlng abnormal level of blood llpkfs. Science 1981;213:1%-1 l-
MD, PIERRE THiROUX,
MD,
YVES TAEYMANS,
MD, LUCIEN CAMPEAU,
MD,
MD, ANDRE PASTERNAC,
MARC A. BOIS, MD, IHOR DYRDA, MD, MS, and PAUL ROBERT DAVID, MD
Angiographicaily documented coronary artery disease (CAD) can progress. Although progression itself is frequently recognized in patients who have undergone repeat cardiac catheterization, its prognostic significance remains unclear. To evaluate the influence of progression on survival, 313 patients with CA6 who underwent catheterization twice (39 f 25 months apart) Wre followed for 3 to 129 months (mean 41 f 30) after the second angiogram. At the time of the second angiogram, 21,. 91, 113 and 88. patients had 0-, I-, 2- and Wessel CAD, respectively. The mean,ejection fraction (EF) of the group was 55 f 13%. Progression was noted in 139 patients (44 % ). Qf the 3 13 patients, 33 died and 39 had acute myocardial infarction (AMI) durlng follow-up. Four-year survival was estimated at 94% and 83% in the nonprogression and progression groups, respectively. Progression was predictive of survival by (univariate) log-rank test (p
only EF (p
The state of our knowledge on survival in coronary artery disease (CAD) is based on univariatel and multi-
investigation was designed to assess the independent prognostic significance of CAD progression in a large series of patients, when taking into account other important baseline characteristics.
(Am J Cardiol 1985;55:941-946)
variate2B analyses of series of patients. Clinical (age2 and worsening angina3), electrocardiographic (ventricular arrhythmias2s) and angiographic (number of diseased vessels,1-3 ejection fraction [EFJ2and left main coronary artery stenosis1-3) characteristics carry in-
Methods Patients: From January 1,197Oto May 1,1982,313 patients underwent 2 cardiac catheterizations at our institution, without previous interim coronary bypass surgery or coronary angioplasty. The second angiogram was performed for persistent stable angina (181 patients), unstable angina (52 patients), research .protocol (40 patients), recent acute myocardial infarction (AMI) (15 patients) or multiple reasons (25 patients). The interval between the 2 studies varied between 3 and 118 months (mean 39 f 25). The presence of st least one 50% or greater diameter reduction of a major coronary artery at the time of the first angiogram was a requirement for inclusion in this study. Clinical and angiographii data: Previous AM1 was diagnosed if 2 of the 3 following criteria were present: (1) chest pain lasting 30 minutes or more; (2) elevation of creatine ki-
formation on survival. The angiographic variables in these studies are considered as fixed. In a previous study, Bemis et al4 showed that patients with progressive CAD were at higher risk of subsequent death. Little attention was given to this problem since their report. The present From the Department of Medicine and Research C&W of the Montreal Heart Institute, Montreal, Quebec, Canada. Manuscript received September 10, 1984; revised manuscript received January 2, 1985, accepted January 3. 1985. Address for reprints: Martial G. Bourassa, MD, Montreal Heart Instjtute, 5000 East Belanger Street, Montreal HlT lC8, Quebec, Canada. 941
PROGNOSTIC SIGNIFICANCE OF PROGRESSION
942
TABLE I
fg
Characteristics of the 313 Patients at the Time of the Second Angiogram
WI
Previous myocardial infarction Angina class’ 0 1 to 2 3 to 4 Not exertional Ejection fraction (%) Left ventricular aneurysm No6pf diseased coronary arteries
No.
(“/I
51 f8 283 176
(84%) (58%) (8%) (29%)
168
;;:;; 0
55*E 27
(9%) (7%)
1 z 3 Left main diameter reduction X0% 30-50 % Progression Occlusion n
9’: 113 88
:::a (28%)
15 18 139 98
Canadian Cardiovascular Society definition (see reference 13). + Patients with O-vessel disease had at least 1 segment initially narrowed by at least 50 % but less than 70 % . l
nase or glutamic oxaloacetic transaminase enzyme levels to twice the upper normal limit; and (3) characteristic, evolving ST-T changes or appearance of new Q waves.5 The angina functional class was evaluated according to the Canadian Cardiovascular Society definition.6 Coronary arteriography was performed using previously described techniques.7 The coronary angiograms were analyzed in the 15-segment coding system of the American Heart Associations; the percent luminal diameter stenosis was noted in each segment. Disposition between l-, 2- or 3-vesseldisease groups was based on the presenceof a 70%or greater diameter narrowing for each of the 3 main coronary vessels; patients with less than 70% but at least 50% diameter narrowings (moderate CAD according to the Coronary Artery Surgery Study [CASS] definition@) were counted as having 0 diseased coronary arteries. Left ventriculography was performed in the 30’ right anterior oblique projection; EF was computed by the area-length method6; the left ventricular contour was divided into 5 segments6p8;the motion of each segment was analyzed as normal, hypokinetic, akinetic or dyskinetic (aneurysmal).
TABLE II
Outcome Events No. of Pts
Analysis by death Unknown cause Nonvascular, noncardiac cause Vascular, noncardiac cause During coronary bypass surgery Acute heart failure Acute myocardial infarction (fatal) Sudden death Total Analysis by first occurrence of acute myoc :ardial infarction during follow-up Acute myocardial infarction (fatal) Acute myocardial infarction (nonfatal) Subsequently alive Subsequently died Total
4 25’ lo+ 39
Among the 25 AMIs, 7 were perioperative. + Among the 10 AMIs, 1 was perioperative. Three of these 10 patients died from a fatal recurrent acute myocardial infarction. l
Assessment of coronary artery disease progression: The first and second coronary angiograms being simultaneously projected, progression was assessedby a consensus9 of 3 observers on the following criterialo-12: a 20% or greater increase in stenosis in a segment initially narrowed by at least 50%; a 30% or greater increase in a segment with an initial obstruction of less than 50%. Occlusion-100% obstruction of a previously patent segment-was considered separately.” Follow-up: Time 0 for follow-up was defined as the day of the second cardiac catheterization (Fig. 1). Sudden death was defined as death that occurred within 1 hour after the onset of the symptoms,i3 in the absence of the autopsy evidence of very recent AMI. Criteria for AM1 that occurred during follow-up were the same as those for previous IvlI.5 However,. criteria for perioperative AMI was based on electrocardiographic evidence only.14 Death that occurred from acute heart failure without AMP5 was considered separately. Follow-up was completed for 312 of 313 patients for the vital status (99.7%); for 307 of 313 patients for the nature of any hospitalization (98%); and for 31 of 33 deaths for the mechanism of death. Statistical analysis: Survival curves were compared by log-rank test for discrete variables and Cox’s regression model with one P parameter for continuous variables.16 Variables significant by univariate analysis were jointly analyzed by Cox’s regression, which models the probability of survival S(t) at time t by: s(t) = {So(t)p+(flT$ where x is the vector of variables included in the model. The fli coefficients of Cox’s model allow the computation of a relative risk (supposed constant over time: “proportional hazard model”) for the occurrence of the analyzed events. Partial likelihood approach was used for ca1culations.l” Models were compared using a likelihood ratio test. The analyses were repeated for prediction of death, sudden death, AM1 (fatal or nonfatal) and survival without AMI. In the latest analysis, the first occurring event, AMI or death, was considered.
Results At the time of the second cardiac catheterization, most of the patients were symptomatic (Table I). CAD progression was documented in 44% of the patients. After the second angiogram, coronary bypass surgery was performed in 116 patients, and 197 patients were treated medically. Thirty-three patients died during the 3- to 129-month (mean 41 f 30) follow-up period (Table II). Four-year survival was estimated to be 89.5%. During the same follow-up period, 39 patients had AM1 (fatal or nonfatal) and 58 had a major event (AM1 or death). Some patients had 2 or more major events: for example, AMI, recurrent AM1 and next sudden death, but only the first major event was kept for the purpose of analysis. Four-year survival without AM1 was 82.5%. Prediction of survival: Survival was unrelated to the year of the second angiogram (p = 0.56), to (surgical or medical) therapy (p = 0.39) or to the occurrence of occlusion (p = 0.88). The univariate predictors of survival-2 clinical and 6 angiographic characteristics-are listed in Table III; they include progression (Fig. 2). Cox’s analysis displayed only EF (p
April I,1985
TABLE III
Univariate
Predictors of Survival
Variables
l’t
(3.129)
- Clinical data Angiographic findings Coronary vessels Left ventriculogram Progression vs no progression
- Mortality - Myocardial infarction
FtGURE 1. Design of the study. CABS = wronary artery bypass svgery; PTCA = percutaneous transluminal coronary angiOPlaSW
p Values
Variables Progression No. of akinetic left ventricular segments Left ventricular aneurysm No. of diseased coronary arteries Ejection fraction Left main stenosis (% ) 80 to 99% stenosis (n)’ Previous myocardial infarction Men Preseptal left anterior descending artery diameter stenosis (% ) Q wave on ECG Time between 1st and 2nd angiograms
‘00r-, 95 s ” -J
,
:..; I - I.......................1 :..........................i
90
-
?- 85 5 v,
-
80
-
1
l-
P<.01
i....: .. . . . .... ........ ..:
2
0.042 0.049
Supplying a nonakinetic left ventricular segment. p values are based on the log-rank test for dichotomous variables and on the Cox model with one p parameter for continuous variables. l
of diseased arteries (p
i... .. ..... .... -NO PROGRESSION (174 PTS) ..... ....PROGRESSION C 1 39 PTS>
0
1
2
YEARS (1741 (139)
(1491 (951
3
4
OF FOLLOW-UP (126) (751
PATIENTS
(107) (631
(821 (421
AT RISK
FIGURE 2. Cumulative survival of patients with and without progression of coronary artery disease. The follow-up begins at the time of the second catheterization. The comparison is performed by the log-rank test.
Progression was the most striking predictor of survival without AM1 by univariate analysis (p
Predictors of Mortality and Acute Infarction (AMI) (Cox Regression Analysls)
Predicted Events (no) Variable Ejection fraction (%) No. of diseased vessels % stenosis of the left main vessel Progression+
Death (n = 33) -O~&Ov$?OO1) 0:687 (O.Od5) o.oW$;O31)
Sudden Death (n = 15)
AMI (Fatal or Nonfatal) (n = 39)
Death or AMI (All Events) (n = 58)
Not Perloperative Death or AMI’ (n = 47)
-,pd,p;l,“~12) bi$8,93!:042)
P (p value) (0.062) 0.4;;o(O.p17)
-c&~~~o9, o:~~51~;jOo2)
-O$!4?;?$01) 0:~;80@“10)
(50.2)
1.06 (0.002)
0.822 (0.003)
Patients who died during a coronary bypass procedure or who experienced a perioperative myocardial infarction analysis. + Scored 1 when present and 0 when absent. P-estimators of the Cox regression coefficients in each analysis-are given for variables retained at p = 0.05. l
943
Volume 55
41 i 30 MO
313 PTS
Univarlate Predictors of Survlval Wlthout Acute Myocardlal Infarction
Independent
OF CARDIOLOGY
ANGlOGRAM
Znd ANGlOGRAM
p values are based on the log-rank test for dichotomous variables, and on the Cox model with one p parameter for continuous variables.
TABLE V
JOURNAL
p Values
Left ventricular aneurysm Left main stenosis (%) Ejection fraction No. of diseased coronary arteries No. of akinetic segments Q wave on ECG Previous myocardial infarction Progression
TABLE IV
THE AMERICAN
0.757 (0.012)
are not considered
in this
PROGNOSTIC SIGNIFICANCE OF PROGRESSION
944
AM1 (p = 0.59, NS) did not differ between the medical and surgical cohorts; in contrast, the predictive value of progression for occurrence of major events persisted (Table V), suggesting that the overall result was unlikely related to the balance between medical and surgical therapies.
100 2
i..: I..; i......, /....... ...... i ........ .. i
90 2 g
80
p=.0001
/ :
. i ............ i . ....
5
...
70 2
Discussion Progression and acute myocardial infarction: A previous studyi demonstrated that evidence of progression was more striking among patients who underwent repeat cardiac catheterization for an episode of unstable angina than it was among matched patients who had undergone repeat catheterization for stable angina. In the present investigation, progression was found to be the most important predictor of AM1 and, next to the number of diseased vessels,the second most powerful predictor for survival without AMI. This link between the anatomic progression of CAD and the occurrence of coronary events is noteworthy. Two working hypotheses are consistent with this strong relation. First, continuing progression may result in increasingly more severe (80 to 90%) stenoses, with a higher risk of occlusion and subsequent AMI.18 Second, both progression and AM1 could result from an hypothetical “active state” of the disease. Pathophysiologic studies aimed to investigate these hypotheses require the recognition of patients at high risk to progress. In such patients, the balance between risks and potential benefits could allow the performance of sequential angiographic studies on a prospective basis. Progression and death: The striking predictive value for AM1 of CAD progression could be contrasted with the apparent lack of prediction for sudden death. However, this does not mean that continuous evolution
NO PROGRESSION
Is >
60 ;i-,
2
0 1
0
YEARS
4
OF FOLLOW-UP 122
147
174 139
3
2
102 53
65
68 PATIENTS
ii
AT RISK
FIGURE 3. Cumulative survival withoul acute myocardiil infarction (AM) in patlents with and without progression. The comparison is performed by the log-rank test.
p
,Q8
”
,ONE
VESSEL
DISEASE
I
EJECTION I,‘,........ _
FRACTION
.
_
>58%
I
. :
L
. i .... ..- p<‘O’ :... ..
Cl31 PTS) .. . ...... <,@a PTS)
EJECTION
FRACTION
FIGURE 4. Cumulative survival without acute myocardial infarction (AMI) in various subsets of patients. In each subgrcup, survival without AMI was better in patients without than in patients with progression.
(50%
‘1 :....: PC.01 60f---.
Cl01 1100
PTS) PT;
/
1
i . ......... i__ F..... ..-: :. __ C43 PTS) --
0 0
1
2
3
YEARS
4
0
. .
<
39
. ..
PTS)
1
I
I
1
2
3
OF FOLLOW-UP
4
April I, 1985
of coronary atherosclerosis is necessarily of little vital consequence. Because of the smaller number of events (15 sudden deaths vs 39 patients with AMI), only huge differences in sudden death hazard rates in the progression and nonprogression groups could be detected. Also our follow-up was limited, and patients with (nonfatal) AM1 can be exposed to an enhanced risk of mortality in subsequent years. This hypothesis could be tested with a larger series of patients followed up for a longer time. Thus, the size and design of our study did not allow for definite conclusions to be drawn on the pathophysiologic mechanisms of sudden death.lg Limitations of this study: Only nonsurgically treated patients who survived until a second catheterization were eligible for this study and were followed up after the second angiogram. The representativity of this particular group is therefore limited. Progression was appraised by visual comparison, using a conservative definition, in accordance with the precision of the method.g Whether quantitative techniques20 could enhance the detectability of progression is debatable, because, as outlined by Kennedy et al,21 the problem is “how to quantitatively analyze paired films.” However, perfect matching of the angulations could be achieved only in prospective studies. Progression is strongly related to the time,11,22l23 and its occurrence may have been underestimated when the time between the 2 studies was short (less than 2 years).24 This limitation, however, should lead to a conservative estimate of the prognostic significance of progression. Also related to the time interval is the choice of an index of progression; although appealing, the definition of a progression rate (dividing the number of progressions by the time interval between studies) ,could mask the prognostic significance of progression, because the time interval between studies itself was a determinant of prognosis (Table IV) (the longer the time interval, the poorer the prognosis): This type of bias is related to the referral process for the second angiogram, and was previously stressed by Bruschke et al.23The definition of an optimal index of progression would require the repetition of angiograms at fixed time intervals. The year of the second angiogram was unrelated to prognosis: it is unlikely that our results could be biased by secular trends in CAD mortality.25 The nature of treatment (surgical vs medical) was unpredictive of outcome; moreover, discarding the perioperative deaths and AM1 did not substantially modify the results (Table V). The main result of the study persisted in various subgroups defined on the more conventional indexes of prognosis (Fig. 4), suggesting that our result was robust and unlikely to be related to a selection bias. The duration of the increased risk of AM1 in patients who progressed before the second angiogram remains to be evaluated: Longer follow-up periods and use of nonproportional hazard models26 may help to clarify this question. Progressive coronary artery disease-a diagnostic and therapeutic challenge: Spontaneous an-
gina or unstable angina12 are correlated with progression of CAD; recently, new onset angina (a variety
THE AMERICAN
JOURNAL
OF CARDIOLOGY
Volume 55
945
of unstable angina) was recognized as an independent long-term predictor of AMI, despite the low incidence of S-vessel disease in these patients28 With regard to the information provided by a first coronary angiogram, progression correlated better with the number of moderate lesion&s3 than with the severity of CAD. Thus, identification of patients at risk of progression is not necessarily out of range. It is unclear whether progressive and stable CAD warrant different therapeutic approaches. Progression is unrelated to the classic “risk factors,“11p22and perfect vs standard correction of these factors has given a disappointing result in a recent trial with clinical endpoints.2g Calcium antagonists may prevent some experimental models of coronary atherosclerosis.s” We are unaware of any clinical investigation on the effects of these drugs on angiographic endpoints. However, recognition of progressive CAD may provide a subset of patients at high risk of AMI: Design of trials aimed to prevent AM1 and/or slow down the process of atherogenesis should be considered in these patients, together with the reevaluation of risk stratification for conventional procedures and therapies. Acknowledgment: We are indebted to Chantal Grenier (Ecole Polytechnique de Mont&al) for many useful suggestions in the programmingwork. We alsothank JeanGauthier, Nicole Gibeault, Dominique Voyer (Institut de Cardiologie de Montreal), Pham Ngoc Dang and Robert Noel (Ecole Polytechnique de Montreal) for their valuable help in the managementof the data, and Monique Massefor her secreterial assistance.
References 1. Bruschke AVG, Proudflt WL, Scnres FM Jr. Progress study of 590 consecutive nonsurgical cases of coronary artery disease followed by 5-S years. I. Arteriographic correlations. Circulation 1973;47: 1147-I 153. 2. Hamtnerrnelster KE, DaRouen TA, Dodga HT. Variables predlcttve of survival in patlents with coronary disease. Selection by univartate and multivariate analyses from the clinical, electrocardiographic, exercise, arteriographic and quantitative angiographic evaluations. Circulation 1979;59:421-430. 3. Harris PJ, Harrell FE, Lee KL, Befrar VS, Rasatl RA. Survival in medically treated coronary artery disease. Circulation 1979;60:1259-1269. 4. Bemls ChE, Gorlln R, Kemp HO, Herman MV. Progression of coronary artery disease. A clinical arteriographic study. Circulation 1973;47:455463. 5. Bktor cn?, Roe&e WR, O’Rourke RA. Clinical pathologic correlates in acute myocardtal infarction. Adv Cardiol 1978;23:14-24. 6. Principal Investigators of CASS and their Associates. Natbnal Haart, Lung, and Blood Institute Coronary Artery Surgery Study. Ciilatkxt 1981;83:suppl 1:1-l-l-81. 7. Bouraasa MO, Lesp6rance J, Campeau L Selective corm &iogaphy by the percutaneous femoral artery approach. AJR 1969;107:377-383. 6. Austen WG, Edwards JE, Frye RL, Genslnl 00, Gott VL, Grlfflth LBC, McGoon DC, Merphy MC, Roe BB. A reporting system on patlents evakrated for coronary artery disease. Report of the ad hoc commMeeforgradingof coronary artery disease, Council on Cardlovascutar Strgsry. Amsrlcan Heart Association. Circulation 1975;5O:suppl V:V-7-V-40. 9. Ban Marco ME. Brooks BH. Blankenhorn DH. Reoroducibilttv of a consensus panel in the interpretation of coronary angiograms. Am Heart J 1978;96:430-437. 10. Bourassa MO, Ryan TJ, Goulet C, Akamowltz J, Lespbrance J, Faxon DP, Enjalbert M, McCabe CH. Early progression of coronary disease in patients assigned to medical therapy (abstr). Circulation 1981:64:suppl IV:IV-83. 11. Molse A, Theroux P, Taeymans Y, Waters DD, Lespbrance J, Flnes Ph, DescoMgs B, Robert P. Clinical and angiogr hit factors associated with progression of coronary artery disease. JAC 7 1984;3:859-867. 12. Yolse A, Theroux P, Taeymans Y, Descolngs B, Lesparance J, Waters DC, Pelletler GB. Beurasss MG. Unstable angina and progression of coronary atherosclerosis. N Engl J Mad 1983;309:885-689. 13. Goktsteh B. The necessity of a uniform definition of sudden coronary death: witnessed death within 1 hour of the onset of acute symptoms. Am Heart J 1982;103:156-159. 14. Namay DL, Hannnemw lster KE, Zta MB, DeRouen TA, Dodge HT, Namay K. Effect of parioperative myocardial infarction on late survival in patients undergoing coronary artery bypass surgery. Circulation 1982;65:1066-
948
PRDGNOST~C SIGNIFICANCE CF PROGRESSION
1071. 18. Clark LT, Garfeln DB, Dwyer EM. Acute puknonary edema dua to ischemic heat-i disease without accompanying rnyocardial Infarction. Natural history and clinical profile. Am J Med lS83;75:332-338. 18. z;-og Regression models and life tables. J R Stat Sot B 1972;34: 17. Cox DR. &rtial likelihood. Biometrika 1975;82:289-278. 18. Raffbtbul W, Smtth LR, Re9ara WJ, Mantle JA, Rackley CE. Russel RO. Ctuantftative coronary arterlography. Coronary anatomy of patfents with unstable angina pectorfs reexamined 1 yeer after optimal medical therapy. Am J Cardiol 1979;43:899-707. 1s. Davlea MJ, Thomas A. Thrombosis and acute coronary artery lesions in sudden cardiac ischemic death. N Engl J Med 1984;310:1137-1140. 20. Brown BG, Bolson EL, Dodge HT. Arteriogaphic assessment of coronary atherosclerosis. Review of current methods, their limitations, and clinical applications. Arteriosclerosis 1982;2:2-15. 21. Kennady JW, Flaher LD, Kllllp Th. Comnary argiogaphy quality control of the CASS study. In: Bond MG, lnsull W, 01 v S, Chandler AB. Comhill F. eds. Clinical Diagnosis of AtherosclerosT 8: Quantitative Methods of Evaluation. New York: Springer Verlag 1982:475-491. 22. Kramw JR, Matauda Y, Mull&an JC, Aronew M, Preadflt WL. Progression of coronary atherosclerosis. Circulation 1981;83:519-528. 22. ~eAVO,wijrnts,KoksersW,~J.Theanatomicevolution
24. 25. 25. 27. 28. 28. 20.
of coronary artery disease demcnstrated by coronary arteriogaphy in 258 nonoperated patients. Circulation 1981;83:527-538. Rafchten JS, Achuff SC, Burkley BH, Pearaoe TA. Risk factors and angiaphic progression of coronary artery disease (abstr). Circulation s1 82;88:suppl IkIt-283. RyarDB.H~FE,~KL,~rmRM,~RA.Animprwlngprogrosis over time in medically treated patients with coronary artery disease. Am J Cardiol 1983;52:444-448. Stabfefa DM, Carter WH, Wampfer GL. Suvival analysis of drug combinatktns using a hazards mods1with time-dspendent covariates. Biometrics 1980;36:537-548. Dyrda I, Fetltclerc R, Saltlel J, Buurassa MO. Clinical indices of angle graphic progession in coronary artery disease (abstr). Circulation 1973; 47:suppl IV:IV-58. Roberta KB, Caltff RM, Harrell FE, Lee KL, Pryor DB, Reaatl RA. The prognosis for patients with new-onset angina who have undsrgone cardiac catheterization. Circulation 1983;88:970-978. Multtple Risk Factor Intervention Trial Research Group. Muftipte Risk Factor lnterventlon Trial: risk factor changes and mortality resufts. JAMA 198% 248~1485-1477. Kramsch D, Aspen AJ, Rozler W. Atherosclerosis: prevention by a ~,;ffectlng abnormal level of blood llpkfs. Science 1981;213:1%-1 l-