A Comparative Synchronous Coronary Surgery Survival Study Robert J. Flemma, M.D., Donald C. Mullen, M.D., Derward Lepley, Jr., M.D., a n d Jacob Assa, B.A.
ABSTRACT Utilizing patient criteria published by the Veterans Administration Cooperative (VAC) Study, a cohort of 229 surgically treated patients was retrieved from the Milwaukee Cardiovascular Data Registry. These patients were all operated on by one surgeon during 1972 to 1974. Four-year survival of this group was compared with that of the medically treated cohort of 310 patients from the VAC Study. Operative mortality was included in all surgical groups. The cumulative 4-year survival of both groups revealed a 95 to 85% advantage for surgical therapy. In patients with three-vessel disease, the cumulative survival favored surgical therapy44% compared with 80% in the medically treated cohort-and in patients with triple-vessel disease and a normal left ventricle, surgical therapy again showed better results: 100% compared with 88%. Patients with twovessel disease and a normal left ventricle who underwent surgical intervention had slightly better 4-year survival than those who had medical treatment-100% versus 95%-and those with twovessel disease and an abnormal left ventricle had a 93% survival after surgical treatment compared with 84% for those with medical treatment. For patients with single-vessel disease, there was no difference in survival between the surgical and medical cohorts.
From St. Luke’s Hospital and the Department of Thoracic and Cardiovascular Surgery, The Medical College of Wisconsin, Milwaukee, WI. We thank the Milwaukee Cardiovascular Data Registry for the follow-up data and the life-table analysis forming the basis of this study, and Dr. Timothy Takaro, who provided us with data from the Veterans Administration Cooperative Study. We are appreciative of the work of the cardiologists at St. Luke’s Hospital who have continuously spent the time and effort to record for the computer data bank the characteristics of all patients catheterized so that meaningful analysis of our work is possible. Presented at the Twenty-fifth Annual Meeting of the Southern Thoracic Surgical Association, Nov 2-4, 1978, Marco Island, FL. Address reprint requests to Dr. Flemma, 9800 W Bluemound Rd, Milwaukee, WI 53226
Coronary bypass operation relieves angina pectoris. During the early years of coronary bypass operation, it was presumed that the operation also prolonged survival of patients affected by coronary artery disease. When this was questioned, comparisons were made between survival after coronary bypass operations done in the 1970s and medical therapy used in the late 1960s. These studies invariably showed coronary bypass operations to be superior by wide margins. However, we agree with critics who pointed out that the time difference and the many changes in medical therapy in the 1970s make these comparisons suspect, and indeed they are of historical interest only. The Veterans Administration Cooperative (VAC) Study, as presented by Murphy [21, Read [31 and their associates, shattered the scientific validity of those previous asynchronous studies and again raised the question of whether coronary bypass operation prolonged survival compared with optimal medical therapy of the 1970s. Randomized studies were touted as the only way to solve this dilemma. Surgeons looking at the VAC Study recognized the somewhat elevated operative mortality (5.8%) in the surgical cohort, and also noted that 12% of this cohort had no patent grafts. When these results were compared with the superior results of medical therapy, only left main coronary disease revealed statistically significant differences between the two methods of treatment. We believed it would be more pertinent to compare the medical therapy cohort of the VAC Study with a surgical cohort that had the same demographic and left ventricular (LV) function characteristics and that was operated on during the same period (1972 to 1974). The study would not be the ideal randomized study but would compare optimal surgical and optimal medical therapy in patients seen during the same years. It should be noted that the medical therapy cohort of the VAC Study represents the best
423 0003-49751791110423-06$01.25 0 1978 by Robert J. Flemma
424 The Annals of Thoracic Surgery Vol 28 No 5
November 1979
survival data ever reported for such treatment. Table 1. Demographic Characteristics The surgical data to be reported are commonly of the VAC Study Medical and Our reproducible in surgical centers performing Surgical Cohorts Seen from 1972 to 1974 many coronary bypass operations. To do this Surgical Cohort study, we analyzed the patients operated on by Medical Cohort one of us (R. J. F.) during 1972 to 1974. Our No left main coronary No left main coronary artery disease artery disease patient data were derived from computer tapes Stable + unstable anof the Milwaukee Cardiovascular Data Registry Stable angina gina established in 1970. Approximating the demoNo hypertension Hypertension graphic and LV function characteristics of the No aneurysm No aneurysm VAC Study as closely as possible, we retrieved Age < 68 yr (average, Age < 68 yr (average, data on 229 patients who underwent operation. 51 yr) 54 Yr) They are compared with the cohort of 310 medically treated patients from the VAC Study [2,3], VAC = Veterans Administration Cooperative. and survival is compared.
Methods The VAC Study has published its criteria for entrance into the randomized trial for stable chronic angina [2, 31. We used these criteria to retrieve a surgical cohort, catheterized and operated on at St. Luke’s Hospital during the same period that the patients in the VAC Study were seen (1972 to 1974). Patients with left main coronary disease and LV aneurysm were excluded. The surgical patients included those with stable and unstable angina and hypertension. The average age of the surgical patients was 54 years (range, 25 to 68 years), which is 3 years older than the VAC Study group, but we also used 68 years as the older age limit. Our patients included 196 men and 33 women. Vessel disease was defined as a 50% reduction in luminal diameter of one or more major coronary arteries, as in the criteria of the VAC Study. The criteria for the two groups of patients are shown in Table 1. The only differences between the two groups were that we included patients with unstable angina and hypertension, and that 15% were women. Our group is an accurate reflection of the spectrum of patients with ischemic heart disease usually seen in a community, and, if anything, these characteristics would be considered prejudicial to long-term survival rather than advantageous compared with those of the VAC Study. Forty patients had single-vessel, 79 have double-vessel, and 109 had triple-vessel coronary artery disease. All had objective evidence of ischemic coronary disease.
Table 2 compares LV function in the two cohorts. In both studies, patients with generalized poor LV function or a “quivering” ventricle were rejected. The VAC Study rejected patients with a marked elevation of left ventricular end-diastolic pressure (LVEDP) but this was not specifically defined. We rejected patients with LVEDP greater than 25 mm Hg. Patients with a large left ventricle were rejected by the VAC Study but we have never used this as a criterion for surgical rejection. The VAC Study also used a poor ejection fraction as a reason for rejection but since we did not have this as a routine measurement during the years covered by this review, we did not use it as a criterion for rejection. Poor ejection fraction alone is not even now a criterion for rejection for operation at St. Luke’s Hospital. The ejection fraction may be misleadingly low if done in only one projection. Biplane ventriculography was not mentioned in the VAC Study. Table 2 also shows that an abnormal left ventricle was considered to be present if a ventriculogram revealed one or more segments to be akinetic, hypokinetic or dyskinetic. This criterion was utilized by both the VAC Study and us, and was our sole criterion. The VAC Study included also an LVEDP greater than 14 mm Hg but this is subject to too many variables affecting the patient at the time of catheterization. Declaring a left ventricle to be abnormal because of an ejection fraction less than 45% or an LVEDP greater than 14 mm Hg is a liberal defi-
425 Flemma et al: Coronary Surgery Survival Study
Table 2 . Ventricular Function Criteria for Rejection and "Abnormality" of VAC S t u d y Medical and Our Surgical Cohorts Criteria Factor
Medical
Surgical
Rejection
Generalized poor LV function Marked elevation of LVEDP? Large LV Poor EF
Generalized poor LV function LVEDP > 25 mm Hg No Not routine at time
Abnormal LV
EF C 45% or 1 or more segments akinetic, hypokinetic, dyskinetic or LVEDP > 14 mm H g
No 1 or more segments akinetic, hypokinetic, dyskinetic No
~~~~
~~~
~~~
~
VAC = Veterans Administration Cooperative; LV = left ventricle; LVEDP = left ventricular end-diastolic pressure; EF = ejection fraction.
nition. The ventriculogram is more precise, and utilizing that criterion, 68% of patients in the VAC Study had an abnormal left ventricle versus 54% of our surgical cohort. Twenty percent of the surgical group had an LVEDP between 15 and 25 mm Hg. Thirty-one percent of the VAC Study cohort had an LVEDP greater than 14 mm Hg, but no upper limit was given. Ventricular function of the two groups is closely comparable considering this fact and considering that an ejection fraction less than 45% is a liberal definition of abnormal. The surgical cohort has been followed by us and by the Milwaukee Cardiovascular Data Registry, which provided the standard actuarial life-table analysis of the data. Only 2 patients have been lost to follow-up (less than l0/o). Operative mortality within 30 days of operation (1 of 229 patients) is included in all life tables for the surgical cohort.
Results Cumulative survival was calculated by standard actuarial life-table analysis. There were 229 patients in the surgical group, and the Milwaukee Cardiovascular Data Registry provided a 99% follow-up. We have reason to believe from personal contact that at least 1 of the 2 lost to follow-up is still alive, but will not respond for personal reasons. Figure 1 compares the cumulative survival for the two groups. Ninety-five percent of the sur-
NLVF
-
105 124 LVEDP< 1 5 ~ 1 7 3 2 1 . 5 1 46
ALVF
d !"
-
6
12
24
3k
i 0
60
i2
Months SERIES C U M U L A T I V E 229 PATIENTS 111110 V A C U M U L A T I V E M E D I C A L 310 PATIENTS
11111+PERSONAL
%
Fig 1 . Cumulative survival at 4 years of 310 medically treated patients in the Veterans Administration (VA) Cooperative Study compared with cumulative survival of 229 patients in the surgical cohort at 4 and 6 years. The figures in parentheses represent the number of patients alive at that time. Data on ventricular function of the surgical cohort are in the left lower corner. See text for details. WLVF = normal left ventricular function; ALVF = abnormal left ventricular function; LVEDP = left ventricular end-diastolic pressure.)
gical group was alive at 48 months; 177 of the 229 (77%) reached at least the 48-month follow-up. Thirty-five patients were followed for 72 months. Thus, more patients in the study were operated on earlier and followed longer than in the VAC Study in which only 94 of the 310 patients (30%) were followed for 48 months. Cumulative survival is not as revealing as the subset analysis. Figure 2 compares all patients with three-vessel disease cumulative, and all patients with three-vessel disease and normal
426 The Annals of Thoracic Surgery Vol 28 No 5 November 1979
--.....
c
..+mmmm+mm
- 90> 0
.-
8 8 % VA MID. 3 V D NLV
I NLV
MEDICAL
’z a0
>, 80a
70-
-> 90 0
v)
bp
100
3 V D NLVSURG.
(8l,l94% (17) *+mm.hmm+mmmm+ .‘Oo 94% 3 V D CUM.SURG.
SURGICAL3VD MEDICAL 3 V D SURGICAL 3 V D MEDICAL 3 V 0
v) J
NLV-(391PlS NLV-120) CUM. (109) CUM. (1521
w
60
70 60
6
12
24
36
48
60
72
Months
Fig 4 . Cumulative survival for the subsets of patients with double-vessel disease and a normal left ventricle (NLV).The surgical cohort, has a slightly better record at 4 years but medical therapy is nearly as effective.
60
’
SURGICAL ABN L V ( 4 3 ) PIS MEDICAL ABN L V I 7 5 ) P l S
6
12
24
36
40
60
72
Months V A MEDICAL-129
60
6
12
24
36
48
60
72
Months +m
=I
+ I
Su r g i c a I
ommmmmmo Medical
Fig 3. Cumulative survival of the 129 patients with triple-vessel disease and an abnormal left ventricle in the Veterans Administration (VA)Cooperative Study compared with thut of 70 patients in the same subset w h o had medical treatment. The surgical cohort has favorable longevity but it is 10% less than in the patients with triple-vessel disease and a normal left ventricle at 4 years.
LV function. It is interesting that patients with triple-vessel disease and a normal left ventricle in the surgical cohort had no operative or late deaths when followed for 72 months. This subset also did well when treated medically, with an 88% survival at 48 months. The 48-month survival for the surgical cohort shows an even greater difference when all patients with triple-vessel disease (cumulative) are compared: 94% in the surgically treated group and 80% in the medically treated group. Patients with three-vessel disease and an abnormal left ventricle were examined as a separate subset (Fig 3). Improved survival with sur-
Fig 5. Cumulative survival for the surgical and medical subsets of patients with double-vessel disease and an abnormal left ventricle (ABN LV). The more favorable results are found in the surgical cohort. Comparison with Figure 4 confirms the point that left ventricular function is a more important determinant of survival than the number of vessels diseased.
gical therapy became apparent at 2 years, and by 4 years there was a considerable difference favoring surgical therapy (90% versus 77%). The surgical survival has remained stable up to 6 years although the number of patients at 6 years is small. A comparison of Figures 2 and 3 shows that in both the surgical and medical subsets of patients with three-vessel disease, the presence of normal LV function greatly enhances 4-year survival. Figures 4 and 5 present life tables for patients with two-vessel disease. For both subsetspatients with a normal and those with an abnormal left ventricle-survival in the surgical cohort was better, but the difference in survival is less when there is a normal left ventricle. In the presence of abnormal LV function, the 93% survival in the surgical cohort has extended to 6 years. The difference became apparent at 2 years and persisted at 4 years.
427 Flemma et al: Coronary Surgery Survival Study
Since these factors may have distorted the reFor patients with single-vessel disease, there was no difference in survival between the sults of the VAC Study, we decided to use its medically and surgically treated groups, both demographic criteria to observe the results of being approximately 98% at 4 years. Reasons our surgical therapy during the same period, a for operating on patients with single-vessel synchronous study. This study is obviously not disease depend on considerations other than perfect, but the important survival comparisons survival. Obviously, the size and distribution involved identical subsets with respect to of an individual vessel are factors in the deci- number of vessels diseased and normal or abnormal ventricular function; for example, pasion. tients with triple-vessel disease and an abnorComment mal left ventricle in the surgical group were Studies on survival following coronary bypass compared with similar patients in the medical operation have been flawed by the fact that sur- €PUP. The mortality for all patients undergoing geons, anxious to prove the efficacy of their procedure, compared operative intervention in coronary bypass and operated on by the same the 1970s with medical therapy of the 1960s. surgeon at St. Luke’s Hospital, including outThese retrospective studies were asynchronous, side referrals and patients having vein bypass and there is little doubt that the comparison plus valve procedures or any procedure indoes not consider the advances in medical ther- cluding a coronary bypass was 1.2% (5 out of apy made in the 1970s. Until the VAC Study, 423 patients). The additional 194 patients resurgeons often compared their results with ferred to us were not included in the surgical cohort of 229 since they were not catheterized at those of Webster and co-workers [41. Medical therapy has changed, and the excel- St. Luke’s Hospital and the catheterization data lent results of medical therapy obtained in the were not entered into the computer. Even if VAC Study have rekindled the debate over sur- those patients were included, there would vival. The randomized prospective study may be no significant change in survival curves. be held as the paragon of scientific truth. How- Limiting the cohort to those studied at St. ever, such studies are not without flaws, as Luke’s Hospital ensured a uniform interpretapointed out by Hurst and associates [ll in 1978. tion and recording of data from ventriculoIn the VAC report, the study of chronic stable grams and angiograms that enhanced the reproangina is flawed mainly by the 17% of patients ducibility of results. The definition of LV dysfunction is imporwho were changed from medical to surgical therapy. If these patients were those with the tant. We consider the second of three criteria greatest angina, and hence the most precarious that the VAC Study employed as being most ischemia, removing them from the medical reproducible (see Table 2). Single-projection therapy cohort could account for increased sur- ejection fraction is not an accurate determinavival in this cohort and thus lessen the differ- tion and could well underestimate ejection ence in survival between medical and surgical fraction if left anterior oblique projection is not therapy. Should these crossovers be consid- performed. Selecting 45% as the lower limit of ered failures of medical therapy? Three other normal for ejection fraction is liberal in declarfactors in the VAC Study are cause for concern: ing a left ventricle abnormal. Also, declaring a its surgical group had a somewhat elevated left ventricle abnormal when the LVEDP is operative mortality (5.8%), 12% of patients had greater than 14 mm Hg is also liberal since all grafts closed, and the incidence of peri- that measurement can be subject to many varioperative infarction was 18%. These facts com- ables about the time of catheterization. Abnorbined with the high crossover from medical to malities seen on ventriculography can be unisurgical therapy lead to a high survival in the formly applied by experienced observers, and medical group, obscuring the differences be- we believe that the slightly greater percentage tween therapies and causing them to emerge of patients with an abnormal left ventricle in the medical cohort of the VAC Study could be only later in the study.
428 The Annals of Thoracic Surgery Vol 28 No 5
November 1979
due chiefly to the more liberal criteria for abnormality. The cumulative cohort analysis is not that revealing, but it does establish some of the overall characteristics of both groups and demonstrates increased longevity in the surgically treated group, despite a 17% incidence of crossover from the cohort undergoing medical therapy. The subsets all demonstrate that longevity in the surgical cohort is greater than in the medical cohort even though it is less distinct in patients with two-vessel disease. One interesting point is that patients with a normal left ventricle and two- or three-vessel disease have excellent long-term survival to at least 6 years with surgical therapy. This statistic reemphasizes the point that survival is more dependent on ventricular function than on the number of vessels diseased. Most informative is Figure 3 showing that in patients with triple-vessel disease and an abnormal left ventricle, the 4-year and 6-year surgical cumulative survival is impressive compared with the 4-year cumulative survival in the medical therapy cohort. The two cohorts are identical and provide further evidence concerning the increased survival with surgical therapy in patients with triple-vessel disease and a normal or abnormal left ventricle. Survival of the medical cohort at 5 and 6 years will be interesting, and if the trends continue, the difference should be exaggerated further with time.
Summary Based on the criteria of the VAC Study, a synchronous (1972 to 1974) surgical cohort of 229
patients with similar demographic and LV function characteristics was compared in terms of survival with the medical cohort of 310 patients from the VAC Study. The survival of patients who had surgical therapy was superior in the entire group. The subsets of patients with three- and 2-vessel disease and normal and abnormal LV function also revealed that survival of the surgical cohort at 4 and 6 years was greater than that of the medical cohort at 4 years. The presence of normal LV function led to greatest surgical survival at 4 and 6 years, emphasizing that LV function is the most important determinant of long-term survival after coronary bypass operation. The survival data of this synchronous study suggest strongly that three-vessel disease in the presence of a normal or abnormal left ventricle can be added to the indications for coronary bypass operation. Although less striking, survival for patients with two-vessel disease and normal or abnormal LV function is more favorable with surgical therapy.
References 1. Hurst JW, King SB, Logue RB, et al: Value of coronary bypass surgery. Am J Cardiol42:307, 1978. 2. Murphy ML, Hultgren HN, Detre K, et al: Treatment of chronic stable angina. N Engl J Med 297:621, 1977 3. Read RC, Murphy ML, Hultgren HN, et al: Survival of men treated for chronic stable angina pectoris: a cooperative randomized study. J Thorac Cardiovasc Surg 75:1, 1978. 4. Webster JS, Moberg C, Rincon G , et al: Natural history of severe proximal coronary artery disease as documented by coronary cineangiography. Am J Cardiol33:195, 1974
ABSTRACT Utilizing patient criteria published by the Veterans Administration Cooperative (VAC) Study, a cohort of 229 surgically treated patients was retrieved from the Milwaukee Cardiovascular Data Registry. These patients were all operated on by one surgeon during 1972 to 1974. Four-year survival of this group was compared with that of the medically treated cohort of 310 patients from the VAC Study. Operative mortality was included in all surgical groups. The cumulative 4-year survival of both groups revealed a 95 to 85% advantage for surgical therapy. In patients with three-vessel disease, the cumulative survival favored surgical therapy44% compared with 80% in the medically treated cohort-and in patients with triple-vessel disease and a normal left ventricle, surgical therapy again showed better results: 100% compared with 88%. Patients with twovessel disease and a normal left ventricle who underwent surgical intervention had slightly better 4-year survival than those who had medical treatment-100% versus 95%-and those with twovessel disease and an abnormal left ventricle had a 93% survival after surgical treatment compared with 84% for those with medical treatment. For patients with single-vessel disease, there was no difference in survival between the surgical and medical cohorts.
From St. Luke’s Hospital and the Department of Thoracic and Cardiovascular Surgery, The Medical College of Wisconsin, Milwaukee, WI. We thank the Milwaukee Cardiovascular Data Registry for the follow-up data and the life-table analysis forming the basis of this study, and Dr. Timothy Takaro, who provided us with data from the Veterans Administration Cooperative Study. We are appreciative of the work of the cardiologists at St. Luke’s Hospital who have continuously spent the time and effort to record for the computer data bank the characteristics of all patients catheterized so that meaningful analysis of our work is possible. Presented at the Twenty-fifth Annual Meeting of the Southern Thoracic Surgical Association, Nov 2-4, 1978, Marco Island, FL. Address reprint requests to Dr. Flemma, 9800 W Bluemound Rd, Milwaukee, WI 53226
Coronary bypass operation relieves angina pectoris. During the early years of coronary bypass operation, it was presumed that the operation also prolonged survival of patients affected by coronary artery disease. When this was questioned, comparisons were made between survival after coronary bypass operations done in the 1970s and medical therapy used in the late 1960s. These studies invariably showed coronary bypass operations to be superior by wide margins. However, we agree with critics who pointed out that the time difference and the many changes in medical therapy in the 1970s make these comparisons suspect, and indeed they are of historical interest only. The Veterans Administration Cooperative (VAC) Study, as presented by Murphy [21, Read [31 and their associates, shattered the scientific validity of those previous asynchronous studies and again raised the question of whether coronary bypass operation prolonged survival compared with optimal medical therapy of the 1970s. Randomized studies were touted as the only way to solve this dilemma. Surgeons looking at the VAC Study recognized the somewhat elevated operative mortality (5.8%) in the surgical cohort, and also noted that 12% of this cohort had no patent grafts. When these results were compared with the superior results of medical therapy, only left main coronary disease revealed statistically significant differences between the two methods of treatment. We believed it would be more pertinent to compare the medical therapy cohort of the VAC Study with a surgical cohort that had the same demographic and left ventricular (LV) function characteristics and that was operated on during the same period (1972 to 1974). The study would not be the ideal randomized study but would compare optimal surgical and optimal medical therapy in patients seen during the same years. It should be noted that the medical therapy cohort of the VAC Study represents the best
423 0003-49751791110423-06$01.25 0 1978 by Robert J. Flemma
424 The Annals of Thoracic Surgery Vol 28 No 5
November 1979
survival data ever reported for such treatment. Table 1. Demographic Characteristics The surgical data to be reported are commonly of the VAC Study Medical and Our reproducible in surgical centers performing Surgical Cohorts Seen from 1972 to 1974 many coronary bypass operations. To do this Surgical Cohort study, we analyzed the patients operated on by Medical Cohort one of us (R. J. F.) during 1972 to 1974. Our No left main coronary No left main coronary artery disease artery disease patient data were derived from computer tapes Stable + unstable anof the Milwaukee Cardiovascular Data Registry Stable angina gina established in 1970. Approximating the demoNo hypertension Hypertension graphic and LV function characteristics of the No aneurysm No aneurysm VAC Study as closely as possible, we retrieved Age < 68 yr (average, Age < 68 yr (average, data on 229 patients who underwent operation. 51 yr) 54 Yr) They are compared with the cohort of 310 medically treated patients from the VAC Study [2,3], VAC = Veterans Administration Cooperative. and survival is compared.
Methods The VAC Study has published its criteria for entrance into the randomized trial for stable chronic angina [2, 31. We used these criteria to retrieve a surgical cohort, catheterized and operated on at St. Luke’s Hospital during the same period that the patients in the VAC Study were seen (1972 to 1974). Patients with left main coronary disease and LV aneurysm were excluded. The surgical patients included those with stable and unstable angina and hypertension. The average age of the surgical patients was 54 years (range, 25 to 68 years), which is 3 years older than the VAC Study group, but we also used 68 years as the older age limit. Our patients included 196 men and 33 women. Vessel disease was defined as a 50% reduction in luminal diameter of one or more major coronary arteries, as in the criteria of the VAC Study. The criteria for the two groups of patients are shown in Table 1. The only differences between the two groups were that we included patients with unstable angina and hypertension, and that 15% were women. Our group is an accurate reflection of the spectrum of patients with ischemic heart disease usually seen in a community, and, if anything, these characteristics would be considered prejudicial to long-term survival rather than advantageous compared with those of the VAC Study. Forty patients had single-vessel, 79 have double-vessel, and 109 had triple-vessel coronary artery disease. All had objective evidence of ischemic coronary disease.
Table 2 compares LV function in the two cohorts. In both studies, patients with generalized poor LV function or a “quivering” ventricle were rejected. The VAC Study rejected patients with a marked elevation of left ventricular end-diastolic pressure (LVEDP) but this was not specifically defined. We rejected patients with LVEDP greater than 25 mm Hg. Patients with a large left ventricle were rejected by the VAC Study but we have never used this as a criterion for surgical rejection. The VAC Study also used a poor ejection fraction as a reason for rejection but since we did not have this as a routine measurement during the years covered by this review, we did not use it as a criterion for rejection. Poor ejection fraction alone is not even now a criterion for rejection for operation at St. Luke’s Hospital. The ejection fraction may be misleadingly low if done in only one projection. Biplane ventriculography was not mentioned in the VAC Study. Table 2 also shows that an abnormal left ventricle was considered to be present if a ventriculogram revealed one or more segments to be akinetic, hypokinetic or dyskinetic. This criterion was utilized by both the VAC Study and us, and was our sole criterion. The VAC Study included also an LVEDP greater than 14 mm Hg but this is subject to too many variables affecting the patient at the time of catheterization. Declaring a left ventricle to be abnormal because of an ejection fraction less than 45% or an LVEDP greater than 14 mm Hg is a liberal defi-
425 Flemma et al: Coronary Surgery Survival Study
Table 2 . Ventricular Function Criteria for Rejection and "Abnormality" of VAC S t u d y Medical and Our Surgical Cohorts Criteria Factor
Medical
Surgical
Rejection
Generalized poor LV function Marked elevation of LVEDP? Large LV Poor EF
Generalized poor LV function LVEDP > 25 mm Hg No Not routine at time
Abnormal LV
EF C 45% or 1 or more segments akinetic, hypokinetic, dyskinetic or LVEDP > 14 mm H g
No 1 or more segments akinetic, hypokinetic, dyskinetic No
~~~~
~~~
~~~
~
VAC = Veterans Administration Cooperative; LV = left ventricle; LVEDP = left ventricular end-diastolic pressure; EF = ejection fraction.
nition. The ventriculogram is more precise, and utilizing that criterion, 68% of patients in the VAC Study had an abnormal left ventricle versus 54% of our surgical cohort. Twenty percent of the surgical group had an LVEDP between 15 and 25 mm Hg. Thirty-one percent of the VAC Study cohort had an LVEDP greater than 14 mm Hg, but no upper limit was given. Ventricular function of the two groups is closely comparable considering this fact and considering that an ejection fraction less than 45% is a liberal definition of abnormal. The surgical cohort has been followed by us and by the Milwaukee Cardiovascular Data Registry, which provided the standard actuarial life-table analysis of the data. Only 2 patients have been lost to follow-up (less than l0/o). Operative mortality within 30 days of operation (1 of 229 patients) is included in all life tables for the surgical cohort.
Results Cumulative survival was calculated by standard actuarial life-table analysis. There were 229 patients in the surgical group, and the Milwaukee Cardiovascular Data Registry provided a 99% follow-up. We have reason to believe from personal contact that at least 1 of the 2 lost to follow-up is still alive, but will not respond for personal reasons. Figure 1 compares the cumulative survival for the two groups. Ninety-five percent of the sur-
NLVF
-
105 124 LVEDP< 1 5 ~ 1 7 3 2 1 . 5 1 46
ALVF
d !"
-
6
12
24
3k
i 0
60
i2
Months SERIES C U M U L A T I V E 229 PATIENTS 111110 V A C U M U L A T I V E M E D I C A L 310 PATIENTS
11111+PERSONAL
%
Fig 1 . Cumulative survival at 4 years of 310 medically treated patients in the Veterans Administration (VA) Cooperative Study compared with cumulative survival of 229 patients in the surgical cohort at 4 and 6 years. The figures in parentheses represent the number of patients alive at that time. Data on ventricular function of the surgical cohort are in the left lower corner. See text for details. WLVF = normal left ventricular function; ALVF = abnormal left ventricular function; LVEDP = left ventricular end-diastolic pressure.)
gical group was alive at 48 months; 177 of the 229 (77%) reached at least the 48-month follow-up. Thirty-five patients were followed for 72 months. Thus, more patients in the study were operated on earlier and followed longer than in the VAC Study in which only 94 of the 310 patients (30%) were followed for 48 months. Cumulative survival is not as revealing as the subset analysis. Figure 2 compares all patients with three-vessel disease cumulative, and all patients with three-vessel disease and normal
426 The Annals of Thoracic Surgery Vol 28 No 5 November 1979
--.....
c
..+mmmm+mm
- 90> 0
.-
8 8 % VA MID. 3 V D NLV
I NLV
MEDICAL
’z a0
>, 80a
70-
-> 90 0
v)
bp
100
3 V D NLVSURG.
(8l,l94% (17) *+mm.hmm+mmmm+ .‘Oo 94% 3 V D CUM.SURG.
SURGICAL3VD MEDICAL 3 V D SURGICAL 3 V D MEDICAL 3 V 0
v) J
NLV-(391PlS NLV-120) CUM. (109) CUM. (1521
w
60
70 60
6
12
24
36
48
60
72
Months
Fig 4 . Cumulative survival for the subsets of patients with double-vessel disease and a normal left ventricle (NLV).The surgical cohort, has a slightly better record at 4 years but medical therapy is nearly as effective.
60
’
SURGICAL ABN L V ( 4 3 ) PIS MEDICAL ABN L V I 7 5 ) P l S
6
12
24
36
40
60
72
Months V A MEDICAL-129
60
6
12
24
36
48
60
72
Months +m
=I
+ I
Su r g i c a I
ommmmmmo Medical
Fig 3. Cumulative survival of the 129 patients with triple-vessel disease and an abnormal left ventricle in the Veterans Administration (VA)Cooperative Study compared with thut of 70 patients in the same subset w h o had medical treatment. The surgical cohort has favorable longevity but it is 10% less than in the patients with triple-vessel disease and a normal left ventricle at 4 years.
LV function. It is interesting that patients with triple-vessel disease and a normal left ventricle in the surgical cohort had no operative or late deaths when followed for 72 months. This subset also did well when treated medically, with an 88% survival at 48 months. The 48-month survival for the surgical cohort shows an even greater difference when all patients with triple-vessel disease (cumulative) are compared: 94% in the surgically treated group and 80% in the medically treated group. Patients with three-vessel disease and an abnormal left ventricle were examined as a separate subset (Fig 3). Improved survival with sur-
Fig 5. Cumulative survival for the surgical and medical subsets of patients with double-vessel disease and an abnormal left ventricle (ABN LV). The more favorable results are found in the surgical cohort. Comparison with Figure 4 confirms the point that left ventricular function is a more important determinant of survival than the number of vessels diseased.
gical therapy became apparent at 2 years, and by 4 years there was a considerable difference favoring surgical therapy (90% versus 77%). The surgical survival has remained stable up to 6 years although the number of patients at 6 years is small. A comparison of Figures 2 and 3 shows that in both the surgical and medical subsets of patients with three-vessel disease, the presence of normal LV function greatly enhances 4-year survival. Figures 4 and 5 present life tables for patients with two-vessel disease. For both subsetspatients with a normal and those with an abnormal left ventricle-survival in the surgical cohort was better, but the difference in survival is less when there is a normal left ventricle. In the presence of abnormal LV function, the 93% survival in the surgical cohort has extended to 6 years. The difference became apparent at 2 years and persisted at 4 years.
427 Flemma et al: Coronary Surgery Survival Study
Since these factors may have distorted the reFor patients with single-vessel disease, there was no difference in survival between the sults of the VAC Study, we decided to use its medically and surgically treated groups, both demographic criteria to observe the results of being approximately 98% at 4 years. Reasons our surgical therapy during the same period, a for operating on patients with single-vessel synchronous study. This study is obviously not disease depend on considerations other than perfect, but the important survival comparisons survival. Obviously, the size and distribution involved identical subsets with respect to of an individual vessel are factors in the deci- number of vessels diseased and normal or abnormal ventricular function; for example, pasion. tients with triple-vessel disease and an abnorComment mal left ventricle in the surgical group were Studies on survival following coronary bypass compared with similar patients in the medical operation have been flawed by the fact that sur- €PUP. The mortality for all patients undergoing geons, anxious to prove the efficacy of their procedure, compared operative intervention in coronary bypass and operated on by the same the 1970s with medical therapy of the 1960s. surgeon at St. Luke’s Hospital, including outThese retrospective studies were asynchronous, side referrals and patients having vein bypass and there is little doubt that the comparison plus valve procedures or any procedure indoes not consider the advances in medical ther- cluding a coronary bypass was 1.2% (5 out of apy made in the 1970s. Until the VAC Study, 423 patients). The additional 194 patients resurgeons often compared their results with ferred to us were not included in the surgical cohort of 229 since they were not catheterized at those of Webster and co-workers [41. Medical therapy has changed, and the excel- St. Luke’s Hospital and the catheterization data lent results of medical therapy obtained in the were not entered into the computer. Even if VAC Study have rekindled the debate over sur- those patients were included, there would vival. The randomized prospective study may be no significant change in survival curves. be held as the paragon of scientific truth. How- Limiting the cohort to those studied at St. ever, such studies are not without flaws, as Luke’s Hospital ensured a uniform interpretapointed out by Hurst and associates [ll in 1978. tion and recording of data from ventriculoIn the VAC report, the study of chronic stable grams and angiograms that enhanced the reproangina is flawed mainly by the 17% of patients ducibility of results. The definition of LV dysfunction is imporwho were changed from medical to surgical therapy. If these patients were those with the tant. We consider the second of three criteria greatest angina, and hence the most precarious that the VAC Study employed as being most ischemia, removing them from the medical reproducible (see Table 2). Single-projection therapy cohort could account for increased sur- ejection fraction is not an accurate determinavival in this cohort and thus lessen the differ- tion and could well underestimate ejection ence in survival between medical and surgical fraction if left anterior oblique projection is not therapy. Should these crossovers be consid- performed. Selecting 45% as the lower limit of ered failures of medical therapy? Three other normal for ejection fraction is liberal in declarfactors in the VAC Study are cause for concern: ing a left ventricle abnormal. Also, declaring a its surgical group had a somewhat elevated left ventricle abnormal when the LVEDP is operative mortality (5.8%), 12% of patients had greater than 14 mm Hg is also liberal since all grafts closed, and the incidence of peri- that measurement can be subject to many varioperative infarction was 18%. These facts com- ables about the time of catheterization. Abnorbined with the high crossover from medical to malities seen on ventriculography can be unisurgical therapy lead to a high survival in the formly applied by experienced observers, and medical group, obscuring the differences be- we believe that the slightly greater percentage tween therapies and causing them to emerge of patients with an abnormal left ventricle in the medical cohort of the VAC Study could be only later in the study.
428 The Annals of Thoracic Surgery Vol 28 No 5
November 1979
due chiefly to the more liberal criteria for abnormality. The cumulative cohort analysis is not that revealing, but it does establish some of the overall characteristics of both groups and demonstrates increased longevity in the surgically treated group, despite a 17% incidence of crossover from the cohort undergoing medical therapy. The subsets all demonstrate that longevity in the surgical cohort is greater than in the medical cohort even though it is less distinct in patients with two-vessel disease. One interesting point is that patients with a normal left ventricle and two- or three-vessel disease have excellent long-term survival to at least 6 years with surgical therapy. This statistic reemphasizes the point that survival is more dependent on ventricular function than on the number of vessels diseased. Most informative is Figure 3 showing that in patients with triple-vessel disease and an abnormal left ventricle, the 4-year and 6-year surgical cumulative survival is impressive compared with the 4-year cumulative survival in the medical therapy cohort. The two cohorts are identical and provide further evidence concerning the increased survival with surgical therapy in patients with triple-vessel disease and a normal or abnormal left ventricle. Survival of the medical cohort at 5 and 6 years will be interesting, and if the trends continue, the difference should be exaggerated further with time.
Summary Based on the criteria of the VAC Study, a synchronous (1972 to 1974) surgical cohort of 229
patients with similar demographic and LV function characteristics was compared in terms of survival with the medical cohort of 310 patients from the VAC Study. The survival of patients who had surgical therapy was superior in the entire group. The subsets of patients with three- and 2-vessel disease and normal and abnormal LV function also revealed that survival of the surgical cohort at 4 and 6 years was greater than that of the medical cohort at 4 years. The presence of normal LV function led to greatest surgical survival at 4 and 6 years, emphasizing that LV function is the most important determinant of long-term survival after coronary bypass operation. The survival data of this synchronous study suggest strongly that three-vessel disease in the presence of a normal or abnormal left ventricle can be added to the indications for coronary bypass operation. Although less striking, survival for patients with two-vessel disease and normal or abnormal LV function is more favorable with surgical therapy.
References 1. Hurst JW, King SB, Logue RB, et al: Value of coronary bypass surgery. Am J Cardiol42:307, 1978. 2. Murphy ML, Hultgren HN, Detre K, et al: Treatment of chronic stable angina. N Engl J Med 297:621, 1977 3. Read RC, Murphy ML, Hultgren HN, et al: Survival of men treated for chronic stable angina pectoris: a cooperative randomized study. J Thorac Cardiovasc Surg 75:1, 1978. 4. Webster JS, Moberg C, Rincon G , et al: Natural history of severe proximal coronary artery disease as documented by coronary cineangiography. Am J Cardiol33:195, 1974
