A MetalAnalysis of Randomized Trials Comparin

ii! Coronary Artery

9 ass

Grafting Wi Percutaneous bans uminal Coronary Angioplasty in Multivessel

Coronary Artery Disease Ida Sim, MD, Munish Gupta, BS, Kathryn McDonald, MM,

Ma&al G. Bourassa, ‘MD, and Mark A. Hlatky, MD

We Performed a meta-analysis of randomized trials that compared percutaneous transluminal corona angiopkrsty (PTCA) with coronary artery by x (CABG) surgery in patients with multivesse r

ss gm coronary

artery disease. The outcomes of death, combined death, and nonfatal myocardial infarction (MI), repeat revas- cularization, and freedom from an ina were analyzed. The ovemll risk of death and ft non tal MI was not dif- ferent over a follow-up of 1 to 3 years (CAB& PTCA odds mtio [ORI 1.03, 95% confidence interval 0.8 1 to 1.32, p = 0.8 1). Patients randomized to CABG tended to have a higher risk of death or MI in the early,

periproceduml period (OR 1.33, p = 0.091), but a low- er risk in subsequent follow-up (OR 0.74, p = 0.093). CABG Patients were much less likely to undergo anoth- er revascularization procedure (p <O.OOOOl), and were more likely to be angina free (OR 1.57, p <O.OOOOl). Thus, CABG and PTCA Patients have similar ovemll risks of death and nonfatal MI at 1 to 3 years of fol- low-up, but relative risk differences in mortality of up to 25% cannot be excluded. CABG Patients have sig- nificantly less angina and less repeat revascukrrization than PTCA patients.

(Am J Cardiol 1995;76: 1025-l 029)

c oronary artery bypass graft (CABG) surgery is a well-established method of coronary revasculariza-

tion that confers substantial survival benefit to patients with extensive multivessel disease.’ Whereas percuta- neous transluminal coronary angioplasty (PTCA) was tirst described by Gruentzig et al2 in 1977 as a treatment for l-vessel disease, improved equipment and greater operator experience have made it a treatment option for patients with multivessel disease as well. Several ran- domized trials have examined whether CABG or PTCA yields better outcomes, but none had a sample size large enough to exclude clinically significant mortality differ- ences. The technique of meta-analysis pools trials of similar design to detect risk differences with higher sta- tistical power and to achieve more precise estimates of outcomes. We performed a meta-analysis of all report- ed randomized trials that directly compared CABG with PTCA in patients with multivessel disease.

METHODS litemture search: We retrieved from Medline and

BIOSIS all English language studies of CABG versus PTCA for multivessel coronary artery disease published between 1985 and 1995. Randomized trials were iden- tified using the search strategy designed for the Cochrane

From the Department of Health Research and Policy, Stanford Uni- versity School of Medicine, Stanford, California; and the Montreal Heart Institute, Montreal, Quebec, Canada. This study was support- ed by Grant HS 08362 from the Agency for Health Care Policy and Research, Rockville, Maryland, and o grant from the Robert Wood Johnson Foundation, Princeton, New Jersey. Manuscript received April 24, 1995; revised manuscript received and accept- ed Au ust 8, 1995.

A dress for reprints: Mark A. Hlatky, MD, Stanford f Universi School of Medicine, HRP Redwood Building, Room 264, 1 Stanfor , California 943055092.

Collaboration.3 Cross-references of all articles were checked and professional contacts pursued to identify all relevant trials.

Data abstraction: The outcomes of interest were: (1) death, (2) combined death and nonfatal myocardial in- farction (MI), (3) freedom from angina, (4) repeat revas- cularization with CABG, and (5) repeat revasculariza- tion with PTCA. All outcomes were analyzed at the longest follow-up time reported for each trial. In addi- tion, death and nonfatal MI outcomes were separately analyzed for in-hospital (i.e., periprocedural) events and for those that occurred during follow-up after the initial revascularization procedure. Freedom from angina was defined as Canadian Cardiovascular Society4 class I1 angina.

Studies were abstracted by 2 independent reviewers and disagreements were resolved with a third party. We contacted all trial investigators for descriptions of their study populations.

Statistical methods: We used the Meta-AnalystTM soft- ware5 to combine all outcomes with both the Mantel- Haenszel fixed effects method6 and the DerSimonian and Laird random effects method.7 The odds ratio (OR) was used as the summary statistic. All reported p values are 2-tailed and were judged significant at the p ~0.05 level. Except where noted, the Mantel-Haenszel and Der- Simonian and Laird analyses gave similar levels of sta- tistical significance and we report only the Mantel-Haen- szel results.

RESULTS Description of the trials: We identified 241 relevant

articles of which 7 were randomized trial.~.*-~~ Six trials reported outcome data (Table I); the Bypass Angioplas- ty Revascularization Investigation* has reported only its

CORONARY ARTERY DISEASE/META-ANALYSIS OF CABG VERSUS PTCA 1025

TABLE I Description of Trials Included in the Meta-Analysis

Number of Coronary

Total Arteries Follow-Up (yr)

Trial Year Patients Narrowed Primary End Point Reported Planned

ERACI’O 1993 127 >2 Fridom from combined death, 1 5 Ml, repeat revascuiarization, and angina

RITA’ ’ 1993 1,011 51 Combined deoth and MI 2.5 25 CABRl12 1994 1 054 22 GAB113 1994 ‘359 22

Not available 1 >5 Freedom from angina 1 1

EASTI 1994 392 r2 Combindd death, Q-wave Ml, 3 8 or lorge thallium defect

se1 disease varied from 12% to 45% (Table II). Overall, 15% had l-vessel disease, 57% had 2-vessel disease, and 28% had 3-vessel disease. Patients with left main disease, recent MI, pri- or CABG, or prior PTCA were exclud- ed from all the trials.

CABRI = Coronary Angioplasty Versus Bypass Revoscularizotion Investigation; EAST = Emory Angioplasty Versus Surgery Trial; ERACI = Argentine Randomized Trial of Percutcmeobr Translumi- nol Coronary Angioplosty Versus Coronary Artery Bypass Surgery in Multivessel Disease; GABI = German Angioplasty Bypass Surgery Investigation; MI = myocordial infarction; RITA = Randomized Intervention Trial of Angina.

design and is not included in our me&analysis. The tri- al of Goy and co-workers9 enrolled only patients with l-vessel disease and was excluded from further analy- sis. The Randomized Intervention Trial of Angina (RITA) enrolled 456 patients with l-vessel and 555 patients with multivessel disease,” so we performed me&analysis both including and excluding the RITA trial. Because these methods yielded essentially identical results, we report the results that include the RITA trial.

Overoll results: At 1 to 3 years of follow-up, the overall risk of death was insignificantly lower in the CABG than in the PTCA patients (3.7% vs 3.9%; OR 0.92,95% confidence interval [CI] 0.63 to 1.34, p =. 0.67), whereas the combined risk of death or nonfatal MI was insignificantly higher in CABG patients (10.1% vs 9.8%; OR 1.03,95% CI 0.81 to 1.32, p = 0.81) (Figure 1).

CABG patients were significantly more likely to be angina-free (80.7% vs 73.1%; OR 1.57, 95% CI 1.32 to 1.87, p ~0.00001) (Figure 2). CABG patients were also substantially less likely to undergo either a subsequent CABG (1.0% vs 19.7%; OR 0.04, 95% CI 0.02 to 0.07, p <O.OOOOl) or subsequent PTCA (6.0% vs 22.9%; OR 0.21, 95% CI 0.16 to 0.27, p <O.OOOOl) (Figure 3).

The 5 trials randomized 2,943 patients, 1,449 to CABG, and 1,494 to PTCA. These 2,943 patients ac- counted for 5% of the approximately 55,560 patients screened for entry.

In-hospital results: The in-hospital death rate was slightly higher in CABG than in PTCA patients (1.4% vs 1.2%; OR 1.15,95% CI 0.60 to 2.18, p = 0.68). The com- bined risk of death or nonfatal MI was also insignificantly higher in CABG than in PTCA patients (5.7% vs 4.4%; OR 1.33, 95% CI 0.96 to 1.86, p = 0.091) (Figure 4).

Patient characteristics were not significantly different Results during 1 - to 3-years of follow-up: For patients between the 2 treatment groups. However, the propor- who survived the initial revascularization procedure, the tion of patients with unstable angina varied widely risk of death during 1 to 3 years of follow-up tended to among trials, and the proportion of patients with 3-ves- be less in CABG patients (OR 0.74,95% CI 0.45 to 1.20,

Overall Risk of Death Studv Year Total Patients

Overall Combined Risk of Death or Non-fatal Ml Studv Year Total Patients

Odds Ratio, 95% Conlidence Interval

ERACI 1993 127

RITA 1993 101;

CABRI 1994 1054

GABI 1994 359

EAST 1994 392

Total 2943

O.R. = 0.92. p = 0.67 0.1 1 10

Lower in CABG Lower in PTCA

Odds Ratlo, 95% Conildence lrlterval

ERACI 1993 127

RITA 1993 1011 --

CABRI 1994 1054

GABI 1994 359

EAST 1994 392

Total 2943

O.R. = 1.03. = p 0.81 0.1 1 10

Lower in CABG Lower in PTCA

FIGURE 1. O&rail risks at 1 to 3 years of follow-u ty Versus Bypass Revascularization Investigation; E L

CABG = coronary artery bypass graft surgery; CABRI = Coronary Angio@as- = Emory Angioplasty Venus Surgery Trial; ERACI = Argentine Randomtzed

Trial of PerWaneous Transluminal Coronary Angioplasty Versus Coronary Arte German Angioplasty Bypass Surgery Investigation; MI = myocardial inf&ction; 8

Bypass Surgery in Muhivessd Disease; GABI = .R.

nal coronary angioplasty; RITA = Randomized Intervention Trial of Angina. = odds ratio; PTCA = percutaneous translumi-

1026 THE AMERICAN JOURNAL OF CARDIOLOGY@ VOL. 76 NOVEMBER 15, 1995

TABLE II Description of Study Populations

ERACI’O RITA” CABRl’2 GABI’ EAW4

% Men 85 81 78* 80 74 Mean (yr) age 57 57 60* 59* 62 Mean ejection 61 - 63 56* 61

fraction (%) Prior MI (“A) 50 42 42 47 41 Angina (%)

o-2 - 41 34 35* 20 3-4 - 59 66 65* 77 Unstable 83 55 - 14 -

Number of coronary arteries narrowed (%)

1 0 45 0* 0 0 2 55 43 60* 82 60 3 45 12 40* 18 40

*Personal communication. Numbers may not odd to 100% due to missing data. -= no information available; other abbreviations (IS in Table I.

p = 0.22) (Figure 5). The combined rate of death or non- fatal MI at follow-up also tended to be lower in CABG than in PTCA patients (OR 0.74, 95% CI 0.52 to 1.05, p = 0.093). The OR f or ea an nonfatal MI was not d th d significantly correlated with the length of follow-up.

DISCUSSION The combined evidence from randomized trials sug-

gests that for patients with multivessel disease, the com- bined rate of mortality and nonfatal MI is not signifi- cantly different after PTCA or CABG at 1 to 3 years of follow-up, but that CABG provides better relief of angi- na and leads to fewer repeat revascularization proce- dures. Published data did not provide sufficient detail to allow analysis by number of diseased vessels, but obser- vational data suggest that PTCA yields better outcomes than CABG for 2-vessel disease, whereas PTCA vields

Freedom from Angina Study Year Total Patients

Odds Ratio, 95% Confidence Interval

ERACI 1993 127

RITA 1993 1011

CABRI 1994 1054

GABI 1994 359

EAST 1994 392

Total 2943

O.R. = 1.57, p < 0.00001 o , 1 10

Favors PTCA Favors CABG

FIGURE 2. Odds ratio for freedom from angina. Abbreviations as in Figure 1.

ing of primary data by the investigators of these trials would better define mortality risks in these and other subgroups of clinical interest.’

Although it is not surprising that CABG carried a slightly higher periprocedural risk, it is intriguing that risks after the initial procedure tended to be lower in CABG patients (OR for death, 0.74, p = 0.22; OR for death and MI, 0.74, p = 0.093). These findings under- score the need for long-term follow-up to determine whether differences in outcomes between CABG and PTCA will emerge over time. Extrapolation of short- term outcomes to the longer term may not be justified because bypass graft failure and progression of native coronary disease become more common after 25 years

worse outcomes than CABG for 3-vessel disease.15 Pool- of follow-up.

Risk of Revascularization with CABG Risk of Revascularization with PTCA Study Year Total Patients Study Year Total Patients

Odds Ratlo, 95% Conftdence Interval Odds Ratio. 95% Confidence Interval

ERACI 1993 127 ERACI 1993 127

RITA 1993 1011 RITA 1993 1011

CABRI 1994 1054 CABRI 1994 1054

GABI 1994 359 GABI 1994 359

EAST 1994 392 EAST 1994 392

Total 2943 - Total 2943

0.R.=0.04.p<0.00001 o.o, 0.1

O.R. = 0.21, p < o.oooo1 1 o o, 1

Lower in CABG Lowk’in CABG

FIGURE 3. Risk of repeat vascukwization. Abbreviations as in Figure 1.

CORONARY ARTERY DISEASE/META-ANALYSIS OF CABG VERSUS PTCA 1027

In-hospital Risk of Death Study Year Total Patients

Odds Ratio, 95% Confidence Interval

In-hospital Combined Risk of Death or Non-fatal Ml Study Year Total Patients

Odds Ratio. 95% Confidence Interval

ERACI 1993 127

RITA 1993 1011

CAERI 1994 1054

GABI 1994 359

EAST 1994 392

Total 2943

ERACI 1993 127

RITA 1993 1011 -

CbiBRl 1994 1054 - -

. .w.

- -

O.R.= 1.15,p=O.68 - 0.1 1 10 100

O.R. = 1.33. = p 0.091 0.1 1 10

Lower in CABG Lower in PTCA Lower in CABG Lower in PTCA

FIGURE 4. In-hospital risks. Abbreviations as in Figure 1.

Follow-up Risk of Death Study Year Total Patients

Follow-up Combined Risk of Death or Non-fatal Ml Study Year Total Patients

ERACI 1993 123

RITA 1993 1001

CABRI 1994 1040

GABI 1994 348

EAST 1994 388

Total 2900

Odds Ri 5% Confidence Interval Odds Ratio, 95% Confidence Interval

O.R.=O.74,~=0.22 I

0.01 0.1 1 10

Lower in CABG Lower in PTCA

ERACI 1993 123

RlTA 1993 1001

CABRI 1994 1040

GABI 1994 348

EAST 1994 388

Total 2900

O.R. =0.74, p=o.o93 o.ol” 0.1 1 10

Lower in CABG Lower in PTCA

--I-

--

-L.

FIGURE 5. Risks during l- to 3-year follow-up. Abbreviations as in Figure 1.

CAE3G patients had much better relief of angina than patients randomized to PTCA, despite the additional revascularization procedures in the PTCA group. The subjective reports of angina relief were supported by objective evidence of less ischemic burden in CABG patients, as provided by thallium scintigraphy in the Emory Angioplasty Versus Surgery Trial.14

FTCA patients were far more likely to undergo repeat revascularization in follow-up, largely because of re- stenosis and perhaps also because of a less complete ini- tial revascularization. However, repeat revascularization, when necessary, is part of the initial PTCA strategy, and we believe it should not be considered an adverse out- come. Although repeat revascularizations were most often

performed in the first year, the risk of repeat revascular- ization was consistently higher in the PTCA group at all time points in the follow-up period.

This meta-analysis of similar, good quality, random- ized trials should yield generally unbiased estimates of the relative efficacy of CABG and FTCA for patients who meet the entry criteria for the trials. Generalization of these results to other patient groups is limited by the stringent entry criteria, under which only 5% of all study site patients who required revascultization were en- rolled. The results should also not be generalized to com- paring CABG with atherectomies or stent procedures, which may have different outcomes than standard bal- loon angioplasty, l6

1028 THE AMERICAN JOURNAL OF CARDIOLOGY@ VOL. 76 NOVEMBER 15, 1995

Acknowledgment: We thank Phil Lavori for his help- ful advice, and Jean-Jaques Gay, Christian W Hamm, and John R. Hampton for providing additional details about the patients enrolled in their trials.

1. Yusuf S, Zucker D, Peduzai P, Fisher LD, Takam T, Kennedy JW, Davis K, Killip T, Passamani E, Norris R, Morris C, Mathur V, Vamauskas E, Chalmers TC. Effect of coronary artery bypass graft surgery on survival: overview of lo-year results from random&d trials by the Coronary Artery Bypass Graft Surgery Tri- alists Collaboration. Lancet 1994;344:563-570. 2. Gmentzig AR, Senning A, Siegenthaler WE. Nonoperative dilatation of coro- nary-artery stenosis. Percutaneous transluminal coronary angioplasty. N Engl J Med 1979;301:61-68. 3. Dickersin K, Lefebvre C. Establishing and maintaining an international register of RCTs. Hamilton: International Cochrane Collaboration, 1994: Section V, Appen- dix II. In: Hayward R, ed. Cochrane Collaboration Handbook. 4. Campeau L. Grading of angina pectoris. Circulation 1975;54:522-523. 5. Lau J. Meta-analyst. Boston: New England Medical Center, 1994. 6. Rotbman K. Modem Epidemiology. Boston: Little, Brown, 1986:358. 7. DerSimonian R, Laird N. Meta-analysis in clinical trials. Conrrol Clin Trials 1986;7:177-188. 8. BARI Investigators. Protocol for the Bypass Angioplasty Revascultization Investigation. Circulation 1991;84(suppl V):V-1-V-27. 9. Goy JJ, Eeckhout E, Bumand B, Vogt P, Stauffer JC, Humi M, Stumpe F, Ruchat P, Sadeghi H, Kappenberger L. Coronary angioplasty versus left internal mamma- ry artety grafting for isolated proximal left anterior descending tiery stenosis.

Lancet 1994;343:1449-1453. IO. Rodriguez A, Boullon F, Perez-B&no N, Paviotti C, Liprandi MIS, Palacios IF, ERACI Group. Argentine Randomized Trial of Percutaneous Transluminal Coronary Angioplasty Versus Coronary Artery Bypass Surgery in Multivessel Dis- ease (ERACI): in-hospital results and l-year follow-up. J Am CON Cardiol 1993; 22:106&1067. Il. RlTA Trial Participants. Coronary angioplasty versus coronary artery bypass surgery: the Randomised Intervention Treatment of Angina (RITA) trial. Lancer 1993;341:573-580. 12. Rickards A. Coronary Angioplasty Versus Bypass Revascularization Investi- gation (CABRI). Presented at the European Society of Cardiology XVth Congress. Nice, France, 1993. 13. Hamm CW, Reimers I, Ischinger T, Rupprecht HJ, Berger J, Bleifeld W. Ger- man Angioplasty Bypass Surgery Investigation (GABI). A randomized study of coronary angioplasty compared with bypass surgery in patients with symptomatic multivessel coronary disease. N Engl .I Med 1994;331:1037-1043. 14. King SB, Lemho NJ, Weintrauh WS, Kosinski AS, Barnhart HX, Kutner MH, Alazmki NP, Guyton RA, Zhao XQ. Emory Angioplasty Versus Surgery Trial (EAST). A randomized trial comparing coronary angioplasty with coronary bypass surgery. NEngl J Med 1994;331:1044-1050. 15. Mark DB, Nelson CL, Califf RM, Harrell FE, Lee KL, Jones RH, Fonin DF, Stack RS, Glower DD, Smith LR, DeLong ER, Smith PK, Reves JG, Jollis JG, Tcheng JE, Muhlbaier LH, Lowe JE, Phillips HR, Pryor DB. The continuing evo- lution of therapy for coronary artery disease: initial results from the era of coro- nary angioplasty. Circulation 1994:89:2015-2025. 16. Topol EJ, Leya F, Pinkerton CA, Whitlow PL, Hofling B, Simonton CA, Mas- den RR, Sermys PW, Leon MB, Williams DO. A comparison of directional atherec- tomy with coronary angioplasty in patients with coronary artery disease. The CAVEAT study group. N Engl J Med 1993;329:221-227.

CORONARY ARTERY DISEASE/META-ANALYSIS OF CABG VERSUS PTCA 1029