Progression of Coronary Artery Disease Predicts Clinical Coronary Events
Long-term Follow-up From the Cholesterol Lowering Atherosclerosis Study
Background Progression of coronary artery disease is assumed to be a surrogate end point for clinical coronary events. Because no single method or measure for a coronary angiographic end point is uniformly accepted as optimal, the utility and validity of surrogate end points for predicting clinical coronary events remain unsettled.
Methods and Results The Cholesterol Lowering Atherosclerosis Study randomized 162 nonsmoking, 40- to 59-year-old men with previous coronary artery bypass graft surgery to colestipol/niacin plus diet or placebo plus diet. Atherosclerosis change on 2-year coronary angiograms was evaluated by a consensus panel and by quantitative coronary angiography (average per-subject change in percent diameter stenosis [%S] and minimum lumen diameter [MLD]). With all three end points, the benefit of colestipol/niacin treatment on coronary artery atherosclerosis has been reported. Annual follow-up for an average of 7 years (range, 6.3 months to 10 years) has been carried out on all subjects who completed the 2-year angiogram. Clinical coronary events (need for revascularization, nonfatal acute myocardial infarction, and coronary death) have been documented. Risk of clinical coronary events was positively related to coronary lesion progression for all three surrogate end points (P<.05). New lesion formation in bypass grafts (P=.02) and progression of mild/moderate lesions (<50%S) were predictive of clinical coronary events (P<.01). Change in MLD contributed significantly to the prediction of clinical coronary events beyond a model with %S alone (P<.05).
Conclusions In this population of nonsmoking men with previous bypass surgery, both the consensus panel– and quantitative coronary angiography–based end points of coronary artery disease progression predict clinical coronary events. Subjects who demonstrate greater coronary artery lesion progression have an increased risk of future clinical coronary events. Design of shorter, smaller trials of antiatherosclerotic agents is justified.
The progression of CAD is generally assumed to be a surrogate end point for clinical coronary events. This has important implications because relative to morbidity/mortality end point trials, angiographic demonstration of CAD progression is possible with trials of substantially reduced duration, sample size, and expense. Although several coronary angiographic end points are available for trial outcome, no single method or measure has been uniformly accepted as optimal. Because of this, the utility and validity of CAD progression for predicting future clinical coronary events remain unsettled.
Long-term follow-up of the CLAS cohort permits examination of the appropriateness of angiographically defined coronary artery end points, assessed by both a consensus panel and by QCA, as predictors of subsequent clinical coronary events. CLAS, which randomized its first subject in 1980 and its last in 1984, tested colestipol/niacin therapy in post–coronary bypass nonsmoking men.1 Annual follow-up for an average of 7 years (range, 6.3 months to 10 years) has been carried out on all subjects who completed the 2-year angiogram.
In the present report, we explore the ability of coronary artery end points obtained from sequential coronary angiograms, acquired at baseline and at 2 years, to predict subsequent clinical coronary events. We compare indicators of atherosclerosis progression at 2 years after randomization as assessed by a consensus panel (the global change score) with QCA-derived per-subject changes in %S and MLD, as well as with the per-subject occurrence of new lesions, occurrence of progressing lesions, and occurrence of total occlusions. In addition, we contrast the lesion populations defined by location and determine which are predictive of progression of future clinical coronary events. The results are the first to conclusively demonstrate that %S and MLD based on an average per-subject change (the most commonly used end points in atherosclerosis angiographic trials) are predictive of future clinical coronary events.
A detailed description of the CLAS trial design and methodology has been published.1 Between September 1980 and October 1984, 188 nonsmoking, 40- to 59-year-old men with previous CABG underwent baseline coronary angiography and were randomized to receive colestipol/niacin plus cholesterol-lowering diet versus placebo plus diet; 162 subjects (86%) had a repeat angiogram after 2 years of intervention. Atherosclerosis change was evaluated by both a consensus panel and QCA. In addition, a total of 138 of 162 subjects (85%) continued their randomized treatment in an optional 2-year extension of CLAS (CLAS-II), in which 103 of 138 subjects (75%) had a repeat angiogram after 4 years of intervention. Benefit of colestipol/niacin treatment on coronary artery atherosclerosis has been reported at 2 years in 162 subjects by the consensus panel,2 at 2 years in a random sample of 85 subjects by QCA,3 and at 4 years in 103 subjects by the consensus panel.4 For this follow-up study, QCA at 2 years has been completed for the entire cohort.
All 162 film pairs, showing identical coronary artery views with treatment allocation and temporal order masked, were evaluated by an expert panel of two angiographers and a moderator.1 5 With the first film, a consensus opinion on lesion identification and %S was obtained and recorded by the moderator. With the second film, a consensus opinion on the amount of change for each lesion and the global change score was obtained. The GCS, an overall assessment of angiographic change that integrated all changes seen in the film pair, ranged from 0 (no demonstrable change) to 3 (extreme change). A direction for change (− for regression and + for progression) was subsequently assigned by the study statistician when the temporal blind was broken.
Of the 162 subjects with 2-year angiograms, 156 (96%) had evaluable end points by QCA. QCA analyses were performed masked to treatment allocation.1 6 7 Film pairs were processed in tandem using dual projectors to match frames for orientation and degree of contrast filling, and arterial segments were defined from branch to branch. Three sequential frames exposed during end diastole were digitized when possible; if not, three sequential frames from other phases of the cardiac cycle were digitized.6 The angiographic view (usually the right anterior oblique) that best demonstrated a lesion was selected. Both %S and MLD were measured for each panel-identified lesion and for lesions identified by the QCA analyst (which tended to be in smaller segments and were less severe) but not by the panel. Edge coordinates were corrected for pincushion distortion measured from the image of a 1×1-cm anteroposterior grid filmed at the beginning of each angiogram. Each end point was averaged over three sequential frames. The QCA end points were calculated as the average (per subject) change from baseline in %S and MLD.
Follow-up was carried out on all 162 subjects who completed the 2-year angiogram for an average of 7 years (range, 6.3 months to 10 years). Subjects were followed annually with either a clinic visit (90%) or a questionnaire (10%) and evaluated for major medical events. ECGs were routinely obtained as part of the annual examination. Data concerning lipid-lowering therapy subsequent to the last angiogram were also obtained. As of April 22, 1993 (the cutoff date for this analysis), only 6 subjects (4%) were lost to follow-up.
The following clinical coronary events were tabulated: requirement for revascularization due to recurrence or worsening of angina pectoris (PTCA, CABG), nonfatal acute MI, and coronary death. For the study, 2-year angiographically related revascularizations and silent MIs noted by ECGs obtained at annual follow-up examinations were not counted as clinical events. Hospital records were obtained to confirm all subject-reported events. Diagnosis of MI was made by a cardiologist who was masked to treatment group assignment and was defined as the presence of two of the following three criteria: typical chest pain, positive creatinine phosphokinase–MB, and presence of new Q wave on the ECG. Cause of death was determined from hospital records or from the family and verified by a death certificate.
Although some subjects had both 2- and 4-year angiograms, all analyses are based on angiographic changes noted at 2 years. Two clinical coronary end points were chosen as dependent variables: time from 2-year angiogram to first clinical coronary event (PTCA, CABG, nonfatal MI, or coronary death) and time from 2-year angiogram to nonfatal MI or coronary death. For descriptive purposes, clinical coronary event rates were calculated for categorical levels of the three surrogate angiographic end points. The GCS was categorized as regression (GCS<0), no change (GCS=0), or progression (GCS>0), and the average per-subject changes in %S and MLD determined by QCA were stratified into quartiles (<−2.4%S, −2.4%S to <0.8%S, 0.8%S to <4.4%S, and ≥4.4%S) and (≥0.10 mm, −0.02 to <0.10 mm, −0.15 to <−0.02 mm, and <−0.15 mm), representing regression to progression, respectively.
The six independent variables were (1) the three surrogate end points (GCS [by panel] and average change in %S and MLD [by QCA]) and (2) three dichotomous QCA-derived variables (the occurrence of new lesions, the occurrence of new total occlusions, and the occurrence of progressing lesions). A new lesion was defined as not present or <20%S at baseline, increased by ≥10%S, and was ≥20%S on the follow-up angiogram. A new total occlusion was defined as <100%S at baseline but increased to 100%S on the follow-up angiogram. A progressing lesion was defined as ≥20%S at baseline and increased by ≥10%S on the follow-up angiogram. Progressing lesions excluded new lesions but not new total occlusions.
Stratifying variables were (1) lesion location (native arteries, native arteries proximal to grafts, native arteries distal or unrelated to grafts, and bypass grafts) and (2) lesion severity at baseline—mild/moderate (<50%S) and severe (≥50%S).
The GCS and QCA independent variables were tested for predictiveness of clinical coronary event rates with the use of Cox regression analyses. Model covariates included indicator variables for treatment group assignment and whether the subject continued into CLAS-II. Because the 2-year angiographic change variables of interest were not related to the use of lipid-lowering medication during the follow-up period, this was not included as a covariate in the Cox regression analyses. Estimates of the RR for clinical coronary events were calculated for categorical levels of each surrogate end point (relative to the lowest level). Tests for trend used GCS- or QCA-determined average per-subject change in %S and MLD as continuous variables. We also estimated the RRs for clinical coronary events per 10%S change in %S and per 0.3-mm change in MLD. These analyses were carried out for all lesions and for lesion subpopulations defined by lesion location (native arteries, native arteries proximal to grafts, native arteries distal or unrelated to grafts, and bypass grafts) and lesion severity (mild/moderate and severe). The cutoffs of 10%S change and 0.3-mm change are twice the measurement error for %S and MLD on short-term repeat angiograms, respectively. Likelihood ratio χ2 tests were used for all significance tests. Finally, we estimated on a per-subject basis the RRs for clinical coronary events on three QCA-derived categorical variables: the occurrence of new lesions, the occurrence of new total occlusions, and the occurrence of at least one progressing lesion. For these analyses, each independent variable (new lesions, new total occlusions, and progressing lesions) was treated on a per-subject basis as a dichotomous variable (yes or no).
Baseline/2-Year Results by Treatment Group
In the cohort of 162 subjects with repeat angiograms at 2 years, the average±SD age at randomization was 54.1±4.5 years, the average±SD blood pressure was 122±13 mm Hg (systolic) and 80±9 mm Hg (diastolic), and 47 subjects (29%) had never smoked. The baseline average±SD lipid levels were total plasma cholesterol of 243±35 mg/dL, LDL-C of 170±31 mg/dL, HDL-C of 42±8 mg/dL, and triglycerides of 150±88 mg/dL. Eighty subjects were assigned to receive drug, and 82 were assigned to receive placebo. No significant differences were found between treatment groups in baseline characteristics. On-trial percentage changes in lipids (drug versus placebo) were −26% versus −4% for total plasma cholesterol, 37% versus 2% for HDL-C, −43% versus −5% for LDL-C, and −21% versus −5% for triglycerides (all P<.0001).
All 162 subjects had evaluable GCS end points; 156 of 162 subjects had evaluable end points by QCA. The distribution of angiographic change as determined by GCS (drug versus placebo) was regression (16% versus 4%), no change (45% versus 37%), and progression (39% versus 59%) (P=.004). The average±SD change in %S and MLD for native arteries plus grafts as determined by QCA (drug versus placebo) was 0.3±5.9%S versus 2.7±5.8%S (P=.02) and −0.01±0.22 mm versus −0.09±0.26 mm (P=.04).
Twenty-four subjects (15%) terminated CLAS after the 2-year angiogram, 103 (64%) continued their randomized treatment and terminated after the 4-year angiogram, and 35 (22%) continued their randomized treatment but terminated before the 4-year angiogram. On termination of CLAS, 57 of the subjects (73%) randomized to the drug group continued on some cholesterol-lowering medication, and 58 of the subjects (74%) randomized to the placebo group initiated some cholesterol-lowering medication (P=NS). Medications included niacin and/or colestipol/cholestyramine, lovastatin/pravastatin/simvastatin, or gemfibrozil. The average±SD change in %S for subjects continuing on lipid-lowering medication versus those who did not in the drug group were 0.4±6.1%S versus 0.1±5.2%S (P=NS). For subjects initiating lipid-lowering medication versus those who did not in the placebo group, these changes in %S were 2.7±4.9%S versus 2.7±8.0%S (P=NS).
Angiographic Results for the Combined Groups
Table 1⇓ presents the angiographic results for the two treatment groups combined. Shown in the left side of the table is the panel-based outcome for the 162 subjects with the GCS end point. According to the panel evaluation, 16 (10%) of the subjects regressed, 66 (41%) demonstrated no clinically significant change, and 80 (49%) subjects progressed. For the 156 subjects with evaluable QCA end points, the right side of Table 1⇓ is a summary of the baseline averages and 2-year changes in %S and MLD, overall and stratified by lesion location (all native arteries, native arteries proximal to grafts, native arteries distal or unrelated to grafts, and bypass grafts) and lesion severity (mild/moderate versus severe). The average±SD number of lesions per subject was 8.2±4.2 (5.8 lesions per subject in the native arteries compared with 2.4 lesions per subject in the bypass grafts). The average±SD %S was 36.2±7.0%S. On average, lesions in bypass grafts progressed more rapidly (3.2%S) than did lesions distal or unrelated to bypass grafts (0.3%S), with lesions proximal to bypass grafts being intermediate (2.5%S) (P=.03).
Table 2⇓ is a summary of the numbers of new lesions (overall and stratified by lesion location), progressing lesions (overall and stratified by lesion location and lesion severity) and new total occlusions (overall and stratified by lesion severity) for the combined treatment groups. New lesions were more prevalent in bypass grafts than they were in native arteries (10% versus 3%, P<.001). Fifteen (1%) of all QCA-evaluable lesions became totally occluded. Progression occurred more frequently in mild/moderate (<50%S) lesions compared with severe (≥50%S) lesions (11% versus 7%, P=.06). Of the 17 severe lesions that progressed, all were proximal to patent grafts (n=12), total occlusions at the second angiogram (n=7), or both.
Clinical Coronary Events
During follow-up of an average of 7 years (range, 6.3 months to 10 years) after the 2-year angiogram, 55 of 162 subjects (34%) (22 drug versus 33 placebo) had one or more clinical coronary events (drug versus placebo): PTCA (7 versus 14), CABG (16 versus 11), nonfatal MI (7 versus 19), or coronary death (2 versus 6). Subjects randomized to the drug group had lower event rates than did subjects randomized to the placebo group: all coronary events (RR=0.6; P=.04) and nonfatal MI or coronary death (RR=0.4; P=.02).
In the Cox regression analyses, treatment group remained a significant predictor of MI/coronary death after adjustment for angiographic change in the QCA end points (%S and MLD). On the other hand, treatment group was not a significant predictor of MI/coronary death after adjustment for GCS or for any coronary event after adjustment for angiographic change in any of the surrogate end points (GCS, %S, or MLD). For all analyses, the interaction between treatment group and angiographic change was not found to be a significant predictor of MI/coronary death or any coronary events.
Table 3⇓ is a summary of the event rates per 100 man-years stratified by the GCS (regression, no change, progression) and quartiles of average per-subject change in %S and MLD for native arteries plus grafts. The overall rates of clinical coronary events and nonfatal MI/coronary death were 5.8 and 2.7 per 100 man-years, respectively. Increase in event rates was associated with progression of the surrogate end points (ie, increasing values of GCS and %S and decreasing values of MLD). Table 3⇓ also presents relative risks of any coronary events and nonfatal MI/coronary death for categorical levels for each of the surrogate end points, adjusted for treatment group and CLAS-II status. The risk of clinical coronary events was positively related to coronary artery lesion progression for all three surrogate end points (P<.05).
Table 4⇓ presents the relative risks of any coronary events and nonfatal MI/coronary death per 10%S change in %S and per 0.3-mm change in MLD for all lesions and also for lesion subpopulations stratified by lesion severity and lesion location. Coronary artery lesion progression was significantly related to any coronary event (2.1≤RR≤2.2, P<0.001) and nonfatal MI/coronary death (1.8≤RR≤2.0, P<.03). Progression of mild/moderate lesions (<50%S), but not progression of severe lesions (≥50%S), was significantly related to coronary event rates (1.7≤RR≤2.1, P<.01 for any coronary event and nonfatal MI/coronary death). When the 60 new lesions (Table 2⇑) were excluded from the analysis of mild/moderate lesions, we found that progression of mild/moderate lesions was still related to coronary event rates (1.7≤RR≤1.8, P<.05). Progression in both native arteries (1.5≤RR≤1.8, P<.05) and bypass grafts (1.3≤RR≤1.6, P<0.05) was significantly correlated with coronary event rates. When stratified by lesion location relative to the bypass grafts, progression in lesions proximal to the bypass grafts (as determined by MLD only) was found to be significantly associated with nonfatal MI/coronary death rates (RR=1.5; P<.05).
Table 5⇓ presents relative risks of any coronary event and nonfatal MI/coronary death for the QCA-derived categorical end points of new lesions, new total occlusions, and progressing lesions. On a per-subject basis, the formation of at least one new lesion was related to the risk of any coronary event (RR=2.0; P=.03), and this risk was associated with new lesions in the bypass grafts (RR=2.3; P=.02) and not in the native arteries. Although the estimate of the relative risk was equivalent for nonfatal MI/coronary death, the association of new lesions to nonfatal MI/coronary death was of borderline significance (RR=2.1; P=.07). On a per-subject basis, the occurrence of new total occlusions was not significantly related to the risk of coronary events. On a per-subject basis, the occurrence of at least one progressing lesion was not significantly related to the risk of coronary events.
The relative predictive ability of the three coronary angiographic end points was determined from a multivariate analysis evaluating the added contribution of one surrogate end point above that of another surrogate end point. GCS did not contribute significantly beyond that for %S (χ12=0.41 for nonfatal MI/coronary death and χ12=2.21 for any coronary event) or for MLD (χ12=0.25 for nonfatal MI/coronary death and χ12=1.11 for any coronary event). %S or MLD did not contribute significantly beyond that for GCS for nonfatal MI/coronary death (χ12=2.58 for %S and 3.68 for MLD). In contrast, both %S and MLD contributed significantly beyond that for GCS for any coronary event (χ12=6.05 for %S and χ12=9.39 for MLD, P<.01). Finally, MLD contributed significantly beyond that for %S for any coronary event (χ12=4.49, P<.05).
The results of the present study indicate that progression of CAD determined by sequential coronary angiography separated by a 2-year interval is predictive of clinical coronary events. Assessment of CAD progression on a per-subject basis by QCA (%S or MLD) or by panel evaluation of overall angiographic change (GCS) predicts future clinical coronary events (Tables 3⇑ and 4⇑). For every 10%S increase in average %S, the risk for a clinical coronary event doubles over an average of 7 years of follow-up (Table 4⇑). CLAS results also demonstrate that formation of new lesions in bypass grafts is a precursor of poor clinical outcome (Table 5⇑). When stratified by lesion location, QCA-assessed progression in both native and bypass grafts was found to be predictive of clinical coronary events (Table 4⇑). When stratified by baseline lesion severity, QCA-assessed progression of mild/moderate lesions (<50%S) was significantly predictive of future clinical coronary events (even after excluding new lesions), whereas progression of severe lesions (≥50%S) was not (Table 4⇑).
Panel-Read and QCA-Read Angiographic End Points as Surrogate Outcome Measures for Subsequent Clinical Coronary Events
The assumption that angiographic progression is predictive of subsequent coronary events was suggested by Bemis et al8 and validated by Moise et al.9 Moise et al demonstrated that progression on sequential coronary angiograms added to the predictive power for acute MI over that of baseline status of the coronary anatomy and left ventricular function. The Program on the Surgical Control of the Hyperlipidemias (POSCH), which demonstrated benefit on coronary artery lesions with cholesterol lowering by partial ileal bypass surgery,10 has also demonstrated that the 3-year panel-based GCS (determined using identical methods as in CLAS) was predictive of subsequent clinical coronary events, fatal coronary events, and all-cause mortality.11 Average follow-up was 6.7 years after the 3-year angiogram. It is of interest, however, that CLAS results did not confirm the treatment/GCS interaction found in POSCH.
QCA assessment of coronary angiograms is increasingly used in coronary angiographic trials because of its enhanced reproducibility over human analysis. Numerous antiatherosclerotic trials using QCA end points have reported positive results for cholesterol-lowering drugs,3 12 13 14 15 16 cholesterol-lowering diet and other life-style changes,17 18 and calcium channel blockers.19 20 Recently, Waters et al20 presented a report from the nicardipine trial in which atherosclerosis progression, defined as a 2-year increase of at least 15%S in at least one coronary lesion, predicted coronary events and coronary death over a subsequent 4.5-year period.21 Although these results indicate that QCA change is predictive of clinical events, determination of change in individual lesions or coronary segments has not been used as a primary end point in angiographic trials.
CLAS follow-up indicates that per-subject assessment of coronary artery lesion progression by both panel (GCS) and QCA (%S and MLD) predicts subsequent clinical coronary events. In addition to validating POSCH results using the GCS end point, this is the first report to show average per-subject change in %S or MLD by QCA to be predictive of clinical coronary events. This is important because it is the average per-subject change that is commonly used as the end point in antiatherosclerosis angiographic trials with QCA and is most representative of overall CAD progression.
Examination of the Figure⇓ indicates that %S and MLD are highly correlated with GCS (P<.001). Consequently, we evaluated whether each angiographic change variable provided additional prediction of clinical coronary event rates beyond that provided by the other variables. From our multivariate analyses, we determined that once %S (or MLD) was in the model, GCS was no longer predictive. On the other hand, although the panel-based GCS is predictive of future clinical coronary events, either QCA end point (%S or MLD) offered additional predictive power beyond that of GCS alone. Because GCS is a seven-point categorical variable and %S and MLD are continuous measures of angiographic change, the range of QCA change within each GCS category provides additional predictive ability.
When the two QCA end points %S and MLD, were compared it was found that MLD contributed significantly beyond that for %S in predicting the rate for any clinical coronary event. This result may indicate that disease progression in normal reference segments has a greater impact on the predictability of the %S measure than does the dilation of coronary arteries (as a result of early atherosclerosis progression) on the predictability of the MLD measure.22 Alternatively, diseased reference diameters do not permit usable estimates of %S by QCA, whereas the MLD measure is available despite the lack of a nondiseased reference diameter.
Risk of Mild/Moderate Lesions
Subjects who demonstrated greater mild/moderate lesion progression over a 2-year interval were at increased risk for future clinical coronary events. This is in accord with previous reports indicating that clinical coronary events are associated with mild/moderate lesions that episodically progress to occlusion as a result of local anatomic factors.23 Although severe lesions are more likely to progress to occlusion than mild/moderate lesions, clinical coronary events more frequently occur as a result of acute occlusion of mild/moderate lesions. This may be a consequence of collateral vessel formation (or CABG), which may avert clinical coronary events when severe lesions occlude. Although there was a significant reduction in clinical coronary events over the subsequent follow-up of CLAS subjects treated with colestipol-niacin, we were unable to determine which specific lesions resulted in this reduction because angiography was not obtained when events occurred. However, it has been shown in the Familial Atherosclerosis Treatment Study (FATS)13 that mild/moderate lesion stability induced by LDL-C lowering significantly reduced on-trial clinical coronary events.24
Because acute coronary events most commonly arise from occlusion of lesions that are of mild/moderate severity, it is not surprising that higher rates of progression of mild/moderate lesions without occlusion result in a greater risk of future clinical coronary events. Our data (Table 1⇑) confirm reports that lesions that are initially the least severe are the most likely to progress.25 26 Low-grade plaques become unstable, fissure, thrombose, and ultimately occlude as they progressively accumulate extracellular lipid within the central core.23 Therefore, reduction in the amount of lipid in the central core, as a consequence of LDL-C reduction, should result in plaque stability with a decrease in lesion progression and resultant clinical coronary events.
Risk of Severe Lesions
Progression of severe lesions was not found to be associated with future clinical coronary events (Table 4⇑). This lack of association for severe lesions could be due to the fact that (1) only 17 (12%) of the 136 progressing lesions were severe (Table 2⇑) and (2) all progressing severe lesions were either proximal to patent grafts (n=12) and thus were protected from causing an event when progressing to occlusion and/or were total occlusions at the second angiogram (n=7), which cannot worsen to produce an event. In this study of subjects who underwent CABG, as well as in angiographic trials of other study populations, it is difficult to evaluate whether progression of severe lesions is associated with future clinical coronary events because such lesions are rarely left undisturbed when they are detected at coronary arteriography. However, this situation provides an interesting indirect confirmation of the hypothesis given above as to why progressing severe lesions rarely lead to clinical coronary events, namely, collateral vessel formation is a natural form of coronary artery revascularization.
Risk in Relation to Lesion Location
Progression rates, and the factors related to progression, are different in native arteries proximal to patent bypass grafts than in nonbypassed arteries or segments distal to sites of graft insertion.3 27 28 29 30 As shown in Table 1⇑, lesions in bypass grafts progressed more rapidly (3.2%S) than did lesions distal or unrelated to bypass grafts (0.3%S), with lesions proximal to bypass grafts being intermediate (2.5%S) (P=.03). In addition, as shown in Table 4⇑, lesion progression in both native arteries and bypass grafts is significantly related to the risk of any coronary event and the risk of nonfatal MI/coronary death. Therefore, progression of lesions per se, whether in bypass grafts or native coronary arteries, is predictive of clinical coronary events. Because we are unable to determine which events are due to bypass graft disease as opposed to native vessel disease, our findings should be extrapolated with caution to nonbypassed patients.
Implications for Future Antiatherosclerosis Clinical Trials
One limitation of our study is that the patient population included only nonsmoking, middle-aged men who had previously undergone CABG. Another limitation is that we are unable to determine which events are due to bypass graft disease as opposed to native vessel disease. However, taken together, findings from the 7-year CLAS follow-up as well as other studies11 21 strongly indicate that end points from sequential coronary angiographic films can be used as surrogate outcome measures for clinical coronary events, thus justifying the design of shorter and less expensive coronary antiatherosclerosis trials of lipid and nonlipid modifying therapies. Specifically, results from this study indicate that all of the traditional coronary angiographic end points—the QCA averaged per-subject end points (%S and MLD) and the panel-based GCS assessment of CAD progression—predict future clinical coronary events. Also, CLAS and POSCH, two independent trials with a common end point, GCS, indicate that lipid-lowering therapy that slows the progression of CAD results in a significant reduction in future clinical coronary events.
Selected Abbreviations and Acronyms
|CABG||=||coronary artery bypass graft surgery|
|CAD||=||coronary artery disease|
|CLAS||=||Cholesterol Lowering Atherosclerosis Study|
|GCS||=||global change score|
|MLD||=||minimum lumen diameter|
|PTCA||=||percutaneous transluminal coronary angioplasty|
|QCA||=||quantitative coronary angiography|
|%S||=||percent diameter stenosis|
This work was supported by National Heart, Lung, and Blood Institute grants RO1-HL-45005 and R01-HL-49885.
Reprint requests to Howard N. Hodis, MD, Atherosclerosis Research Unit, Division of Cardiology, University of Southern California School of Medicine, 2250 Alcazar St, CSC 132, Los Angeles CA 90033.
↵1 Deceased May 9, 1993.
- Received November 22, 1994.
- Revision received June 19, 1995.
- Accepted August 16, 1995.
- Copyright © 1996 by American Heart Association
Blankenhorn DH, Johnson RL, Nessim SA, Azen SP, Sanmarco ME, Selzer RH. The Cholesterol Lowering Atherosclerosis Study (CLAS): design, methods, and baseline results. Control Clin Trials. 1987;8:354-387.
Blankenhorn DH, Selzer RH, Mack WJ, Crawford DW, Pogoda J, Lee PL, Shircore AM, Azen SP. Evaluation of colestipol/niacin therapy with computer-derived coronary end point measures: a comparison of different measures of treatment effect. Circulation. 1992;86:1701-1709.
Azen SP, Cashin-Hemphill L, Pogoda J, Mack WJ, Sanmarco M, Wickham E, Blankenhorn DH. An evaluation of human panelists in assessing coronary atherosclerosis. Arterioscler Thromb. 1991;11:385-394.
Selzer RH, Hagerty C, Azen SP, Siebes M, Lee P, Shircore A, Blankenhorn DH. Precision and reproducibility of quantitative coronary angiography with applications to controlled clinical trials: a sampling study. J Clin Invest. 1989;83:520-526.
Mack WJ, Selzer RH, Pogoda J, Lee P, Shircore AM, Azen SP, Blankenhorn DH. Comparison of computer and human coronary angiographic end point measures for controlled therapy trials. Arterioscler Thromb. 1992;12:348-356.
Bemis CE, Gorlin R, Kemp HG, Herman MV. Progression of coronary artery disease: a clinical arteriographic study. Circulation. 1973;47:455-464.
Buchwald H, Varco RL, Matts JP, Long JM, Fitch LL, Campbell GS, Pearce MB, Yellin AE, Edmiston WA, Smink RD, Sawin HS, Campos CT, Hansen BJ, Tuna N, Karnegis JN, Sanmarco ME, Amplatz K, Castaneda-Ziniga WR, Hunter DW, Bissett JK, Weber FJ, Stevenson JW, Leon AS, Chalmers TC. Effect of partial ileal bypass surgery on mortality and morbidity from coronary heart disease in patients with hypercholesterolemia: report of the Program on the Surgical Control of the Hyperlipidemias (POSCH). N Engl J Med. 1990;323:946-955.
Buchwald H, Matts J, Fitch LL, Campos CT, Sanmarco ME, Amplatz K, Castaneda-Ziniga WR, Hunter DW, Pearce MB, Bissett JK, Edmiston WA, Sawin HS, Weber FJ, Varco RL, Campbell GS, Yellin AE, Smink RD, Long JM, Hansen BJ, Chalmers TC, Meier P, Stamler J. Changes in sequential coronary arteriograms and subsequent coronary events. JAMA. 1992;268:1429-1433.
Blankenhorn DH, Azen SP, Kramsch DM, Mack WJ, Cashin-Hemphill L, Hodis HN, DeBoer LWV, Mahrer PR, Masteller MJ, Vailas LI, Alaupovic P, Hirsch LJ, and the MARS Research Group. The Monitored Atherosclerosis Regression Study (MARS): coronary angiographic changes with lovastatin therapy. Ann Intern Med. 1993;1109:969-976.
Waters D, Higginson L, Gladstone P, Kimball B, LeMay M, Boccozzi SN, Lesperance J. Effects of monotherapy with an HMG-CoA reductase inhibitor on the progression of coronary atherosclerosis as assessed by serial quantitative arteriography: Canadian Coronary Atherosclerosis Intervention Trial. Circulation. 1994;89:959-968.
Waters D, Lesperance J, Francetich M, Causey D, Theroux P, Chiang Y-K, Hudon G, Lemarbre L, Reitman M, Joyal M, Gosselin G, Dyrda I, Macer J, Havel RJ. A controlled clinical trial to assess the effect of a calcium channel blocker on the progression of coronary atherosclerosis. Circulation. 1990;82:1940-1953.
Waters D, Craven TE, Lesperance J. Prognostic significance of progression of coronary atherosclerosis. Circulation. 1993;87:1067-1075.
Brown BG, Zhao XQ, Albers JJ. Lipid lowering and plaque regression: new insights into prevention of plaque disruption and clinical events in coronary disease. Circulation. 1993;87:1781-1791.
Stone PH, Gibson M, Pasternak RC, McManus K, Diaz L, Boucher T, Spears R, Sandor T, Rosner B, Sacks FM. Natural history of coronary atherosclerosis using quantitative angiography in men, and implications for clinical trials of coronary regression. Am J Cardiol. 1993;71:766-772.
Waters D, Lesperance J, Hudon G. Progression of coronary atherosclerosis: a prospective, quantitative, angiographic study. Circulation. 1990;82(suppl III):III-251. Abstract.
Mauer BJ, Oberman A, Holt JH, Kouchoukos NT, Jones WB, Russell RO, Reeves TJ. Changes in grafted and nongrafted coronary arteries following saphenous vein bypass grafting. Circulation. 1974;50:293-300.