T attention as an alternative and durable graft in coronary

Transcription

1 Right Internal Mammary Artery for Myocardial Revascularization: Earlv Results and Indications J Johan Ramstrom, MD, Ole Lund, MD, Eduardo Cadavid, MD, Sten Oxelbark, MD, Johan B. Thuren, MD, and Axel C. Henze, MD, PhD Departments of Cardiothoracic Surgery and Thoracic Radiology, University Hospital, Uppsala, Sweden PhD, The right internal mammary artery (RIMA) was used for coronary artery bypass grafting in 258 patients from October 1985 to October The RIMA was inserted as the only graft in 8 patients and in combination with the left internal mammary artery (LIMA) in 231 patients, the right gastroepiploic artery in 19, and autologous vein in 184. The patients received a total of 1 to 8 distal anastomoses (mean number, 3.3). A total of 64% of the RIMAs were anastomosed to the left anterior descending coronary artery. The primary indication for use of the RIMA was small-vessel disease in 86 patients, repeat bypass grafting in 32, varicose or stripped saphenous veins in 61, and "selected routine case" in 79. The early (53 days postoperatively) mortality rate in these four groups was 8.1%, 6.3%, %, and %, respectively (p <.1). Independent risk factors (logistic regression analysis) for early mortality were small-vessel disease, insufficient grafting, repeat coronary artery bypass grafting, diabetes, history of smoking, age of 6 years or older, and family history of ischemic heart disease. Combined into a risk index, these risk factors identified six risk groups with early mortality of % in the four low-risk groups and 5.6% and 58.3% in groups V and VI, respectively (p <.1). RIMA-related variables were risk factors for significant postoperative myocardial enzyme release. Intraoperative electromagnetic flow measurements revealed no differences between the RIMA and LIMA. Early angiographic patency in 5 patients was 98% for the RIMA and 93% for the LIMA. The RIMA-related variables were risk factors for neither early mortality nor significant enzyme release. The RIMA can be used to revascularize any of the three coronary systems but is primarily suited for the left anterior descending coronary artery. Indications for use of the RIMA (in most instances with the LIMA) include any situation with exhausted venous reserves, small-vessel disease (an indication for primary arterial grafting), isolated right coronary artery disease in selected patients, and selected routine cases (with the LIMA). (Ann Thorac Surg ) he right internal mammary artery (RIMA) attracts T attention as an alternative and durable graft in coronary artery bypass grafting (CABG) because of the growing number of reoperations and in patients with varicose veins. As a complement to routine use of the left internal mammary artery (LIMA), the RIMA may also be warranted in patients with small-vessel disease in whom vein graft patency is not acceptable [l]. Possible higher graft patency of the RIMA relative to vein grafts as is the case for the LIMA [2] could also make routine double internal mammary artery (IMA) grafting indicated in selected patients without small-vessel disease or exhausted venous reserves. A potentially limiting factor in such double IMA grafting is the risk of sternal dehiscence, especially in diabetic patients [3-51. Contrary to numerous reports on the LIMA in CABG, use of the RIMA per se appears to have attracted little interest. The aim of this report was to analyze in detail our early results in 258 patients who, for various reasons, received the RIMA as an alternative conduit vessel. In addition to multivariate risk stratification of early mortality, sternal dehiscence, and significant enzyme release, Accepted for publication Sep 23, Address reprint requests to Dr Ramstrom, Department of Cardiothoracic Surgery, University Hospital, Uppsala, Sweden. we also analyzed intraoperative blood flow and early postoperative graft patency of the RIMA compared with the LIMA. Material and Methods A total of 4,48 patients underwent CABG at one center from October 1985 through September The present study group includes all patients (n = 258) in whom the RIMA was included as a bypass graft during this 6-year period. The study group included 195 men and 63 women with a mean age of 57 years (range, 21 to 77 years). Further patient data are presented in Table 1. Primary lndication for Use of RIMA It was our impression that small-vessel disease entailed increased operative risk. When this type of coronary artery disease was suspected on the coronary arteriogram, the RIMA in addition to the LIMA was used electively by two senior surgeons. The remaining four senior surgeons in our clinic used the RIMA when forced by lack of graft material. Small-vessel disease was recorded in the present study when it was confirmed intraoperatively using the following criteria: maximum diameter of a probe that could be passed distally to the arteriotomy was 1 mm, or the probe was maximally 1.5 mm in conjunction with 1993 by The Society of Thoracic Surgeons /93/$6.

2 1486 RAMSTROM ET AL IUMA GRAFTING Ann Thorac Surg Table 1. Preoperative and Intraoperative Data and Univariate Predictors of.early Mortality. of Early Mortality Variable Patients (%) Age (Y) <6 26 Sex Male Female Family history of CAD History of smokingb Diabetes, NYHA class I1 111 IV Coronary artery disease One-vessel Two-vessel Three-vessel Left mainstemd Small-vessel disease,f ne One Two Three Left ventricular ejection fraction c.3 Indication for use of RIMAS Small-vessel diseaseh Reoperation Varicose or stripped veins Routine Grafts RIMA RIMA + V RIMA + LIMA RIMA + LIMA + V RIMA + GEA RIMA + GEA + V RIMA + LIMA + GEA RIMA + LIMA + GEA + V Aortic occlusion time (miny >6 Insufficient grafting8 Cardioplegia infusion Single dose Repeated a Significance: 11 <.5. This catego includes smokers (n = 84) and ex-smokers. This includes atients 3 are insulin de endent (n = 6) and those who are not. $his category includes 18 gatients with two-vessel disease and 27 with three-vessel disease. This means involvement of one, two, or three main coronary artery systems (anterior descendin circumflex, and right) or none. Significance: p <.1. psignificance: g < ;ol. This means vessels receiving an artery graft and includes 5 pahents with varicose or stripped veins and 14 patients having reoperation. This mean only one cardioplegia infusion each with an aortic cross-clamp time that exceeded 3 minutes. CAD = coronary artery disease; GEA = astroepiploic artery; LIMA = left internal mammary artery; NYHW = New York Heart Association; RIMA = right internal mammary artery; V = vein. diffuse, severe distal atherosclerosis. Such vessels were primarily selected to receive an artery graft, which was accomplished in 86 patients. In 6 patients, small-vessel disease was recorded only for vessels that received a vein graft. The primary indication for use of the RIMA in the 258 patients was small-vessel disease of a vessel receiving an artery graft in 86, varicosity or previous stripping of the saphenous veins in 61, and repeat CABG because of recurrent angina in 32 (see Table 1). The remaining 79 patients, constituting a minority of routine cases during the study period, included 77 who underwent elective double IMA grafting because of expected higher graft patency of the RIMA relative to vein grafts, and 2 younger patients with isolated disease of the right coronary artery. The 86 patients with small-vessel disease as the primary indication included 45 with exhausted venous reserves (varicosity or previous stripping) and 14 who underwent repeat CABG. Opera tion After a midline sternotomy, a Pemco retractor (Pemco Instruments Inc, Cleveland, OH) facilitated exposure of both the RIMA and LIMA, which were dissected free as pedicles from their origin to the terminal division. General (3 C) and topical hypothermia and St. Thomas standard crystalloid cardioplegia were used in all patients. The NMA was used mainly as a pedicled graft (n = 25), but as a free graft in 8 patients because of inadvertent vascular injury or to reach a distant arteriotomy. The RIMA was inserted as the only graft in 8 patients and in combination with the LIMA in 231 patients, autologous vein from the greater or lesser saphenous system in 184, and the right gastroepiploic artery in 19. The RIMA was routed anterior to the ascending aorta for anastomosis to the left anterior descending coronary artery (LAD) or diagonal artery (n = 165). For anastomosis to the obtuse marginal or intermediate artery, the RIMA was pulled through the transverse sinus (n = 18). The right main coronary artery (n = 43), the posterior descending branch (n = 31), and the posterior lateral branch (n = 1) were reached by transpleural routing of the RIMA through a small pericardial window, usually anterior to the phrenic nerve (Fig 1). The LIMA was routed to the recipient coronary artery (LAD, n = 1; circumflex artery and its branches, n = 13; and posterior descending artery, n = 1) through a pericardial window. Concomitant procedures included aortic (n = 1) or mitral valve (n = 3) replacement, mitral valve repair (n = 2), and left ventricular aneurysmectomy (n = 4). Thromboendarterectomy was done on a vessel receiving an artery graft in 16 patients and on a vessel receiving a vein graft in 12. Insufficient grafting was defined as failure to graft a diseased main coronary artery system (LAD, circumflex, or right); in 48 patients, insufficient grafting was due to lack of graft material and in 16, to a technically nongraftable main system. Significant Enzyme Release We performed multivariate risk analysis of death within 3 days after operation (early mortality), sternal dehiscence, and significant postoperative enzyme release (ex-

3 Ann Thorac Surg RAMSTROM ET AL 1487 Fig 1. Routing of the pedicled right intcrnal rnaminary artery to (A) the right coronary or posterior descending artery through a pericardial window, (B) in front of ascending aorta to the left anterior descending or diagonal artery, and (C) through the transverse sinus to the circumflex or obtuse marginal artery. Perccrrtage of patients is shozon for each alternative. A B C cluding patients who died within 3 days). Serum levels of the myocardial-specific isoenzyme of creatine kinase [6] were measured 3 hours and 17 to 2 hours after release of the aortic cross-clamp, and serum levels of aspartate aminotransferase and alanine aminotransferase were measured on the first and second postoperative mornings. Significant enzyme release was recorded in patients who had maximum values of both the myocardial-specific isoenzyme of creatine kinase and aspartate aminotransferase that were more than three times the upper reference level for our laboratory and a maximum value of alanine aminotransferase that was less than two times the upper reference level. Our aim was to create a measure that was related beyond doubt to major ischemic myocardial injury. lntraopera tive Flow Measurement Blood flow in the RIMA graft was compared with that in the LIMA graft in the last 2 consecutive patients who had double IMA grafting. The IMA blood flow was measured during a stable hemodynamic state after decannulation and administration of protamine sulfate using a standard electromagnetic flowmeter (Transflow system; Skal Instruments Inc, Holland) according to standard methods. The probe size was selected to fit well around the graft with slight (1% to 15%) compression of the vessel. Stable and reliable measurements were obtained in 17 patients. Postoperative A ng iograph y Forty-three patients were selected for graft angiography (5 to 8 days postoperatively) before discharge. They included all 19 patients who received a right gastroepiploic artery graft and the last 24 patients in the series. Seven patients had angiography 2 to 4 years postoperatively because of recurrent angina during the study period, and all their IMA grafts were patent. The evaluation of early patency thus included a total of 5 patients. Angiography was performed using standard methods and equipment by selective contrast injection into the IMA ostium (n = 39) or into the subclavian artery (n = 11). To record a graft as patent, we required visualization of both the graft and the receiving coronary artery. Statistical Analysis The tests were computerized using the BMDP statistical software (71. Univariate comparisons between groups were done using a Pearson,$ test except where otherwise noted. Multivariate risk analysis of early mortality, significant enzyme release, and sternal dehiscence was performed using stepwise logistic regression analysis according to a previously published formalized analysis sequence [8]. The level of statistical significance was.5. Results The following variables were considered in the multivariate risk analysis: those shown in Table 1; those already mentioned (coronary artery receiving RIMA, LIMA, and gastroepiploic artery grafting; routing of RIMA; concomitant procedure; and thromboendarterectomy); number of artery grafts (1 to 3, with a mean of 2.); total number of grafts (1 to 5 with a mean of 3.); number of artery graft anastomoses (1 to 4 with a mean of 2.1); total number of anastomoses (1 to 8 with a mean of 3.3); sequential grafting (n = 44); previous myocardial infarction (n = 172); systemic hypertension (n = 14); duration of angina (.1 to 25 years with a mean of 5.3 years); hypertriglyceridemia (n = 9); hypercholesterolemia (n = 97); combination of the latter two (n = 56; both according to standard definitions); and surgeon. There were several important interactions between potential risk factors. Patients who underwent repeat CABG (n = 46) had the highest frequency of insufficient grafting (63% versus 17% for other patients; p < O.OOOl), whereas patients considered to be routine cases (n = 79) had the lowest frequency (8% versus 32%; p <.1) and the lowest prevalence of age greater than or equal to 6 years (25% versus 54%; p <.1). Patients with varicose or stripped saphenous veins (n = 16) had the highest prevalence of age 6 years or greater (63% versus 33%; p <.1). Patients who despite an aortic cross-clamp time in excess of 3 minutes were given only one cardioplegic infusion (n = 129) had the highest prevalence of smallvessel disease (5% versus 22%; y < O.OOOl), and only 22% (versus 4%; p <.1) were routine cases. Early Mortality and Significant Enzyme Release There were nine deaths within 3 days after operation, an early mortality rate of 3.5%. The deaths were related to myocardial infarction and low-output failure in 7 patients, aortic dissection in 1 patient, and capture of the disc of a Bjork-Shiley mitral valve in 1. Postmortem examination of

4 1488 RAMSTROM ET AL Ann Thorac Surg Table 2. Independent Risk Factors for Earlu Mortalitu Regression Coefficient Standard Risk Factof (b) Error u Value Small-vessel disease <.1 Insufficient grafting o.ooo1 Repeat CABGb Diabetes History of smoking Age 26 years Family history of CAD Risk index = b,z, + b2z b7z7: mean, (standard deviation); range, -25. Six equidistant risk-index groups: I I1 I11 IV V VI. of patients of deaths 2 7 Rate (W)c See text and Table 1 for definitions. ' This means repeat CABG as primary indication for use of right internal mammary artery (see Table 1). ie, excludes patients who also had small-vessel disease. ' Significance: p <.1. b,, b,,..., b, = regression coefficients; CABG = coronary artery bypass grafting; CAD = coronary artery disease; z,, z,..., z, = value of risk factors (equals 1 if risk level is present and if not; equals, 1, 2, or 3 for patients with small-vessel disease in none, one, two, or three main coronary arteries, respectively). 6 patients revealed that all artery graft anastomoses were open (small-vessel disease in 5) and that a vein graft had clotted in 2. Univariate risk factors for early mortality are shown in Table 1; Table 2 shows the final logistic regression model. Small-vessel disease gave the strongest stratification of the patients in terms of risk of early death, but insufficient grafting and repeat CABG also were strong independent risk factors. The patients with diabetes had a high mortality, but those who died (n = 3) all had small-vessel disease. The one death among patients hav- Table 3. lndependent Risk Factors for Significant Enzyme ReleasP Risk Factor Regression Standard Coefficient (b) Error p Value Aortic occlusion time vein graft anastomoses Ejection fraction of < Single-dose cardioplegia Risk index = b,z, + b,z2 + b,z, + b,z,: mean, 2.38 * 1.8 (standard deviation); range, Four equidistant risk-index groups: 1 I1 111 IV. of patients with enzyme leak Rate (%I)~ Patients who died early were excluded. ' Significance: 11 <.1. b,, b,, b,, b4 = regression coefficients; zl, z,, z3, z, = value of risk factors (equals 1 if risk level is present and if not; equals discrete value of aortic occlusion time). ing RIMA plus vein grafting occurred after repeat CABG, and the three deaths after RIMA plus LIMA grafting all occurred in patients with severe three-vessel coronary artery disease who had insufficient grafting because of lack of vein material. Significant enzyme release was noted in 24 (9.6%) of the 249 patients who survived the first 3 days after operation. The logistic regression model for these 249 patients is given in Table 3. Major Complications Sternal dehiscence occurred in 8 patients, none of whom died within 3 days postoperatively. Diabetes was the only independent risk factor. The frequency of sternal dehiscence (patients who died within 3 days excluded) was % in patients without double IMA grafting (n = 26), whereas patients with double IMA grafting had a frequency of 2% in the absence of diabetes (n = 196) and 15% when diabetes was present (n = 27) (p <.1). Reoperation for bleeding was necessary in 2 patients and was related to the RIMA pedicle or bed in 8. The latter group included the only iatrogenically damaged RIMA, which was accidentally divided at reoperation for bleeding. Respirator treatment for more than 48 hours was necessary in 13 patients. Two patients received hemodialysis treatment, and 2 needed intraaortic balloon pumping because of an intraoperative myocardial infarction (embolization of atheromatous material from the ascending aorta in 1 and during a complicated repeat CABG procedure in the other; both survived). patient sustained permanent neurological damage. RIMA Blood Flow and Early Patency Difficulties in obtaining a stable probe position resulted in oscillating and unreliable measurements in 3 of the 2 patients. The IMA blood flow values obtained in the remaining 17 are shown in Table 4. Using a Wilcoxon rank-sum test, there was no significant difference between RIMA and LIMA blood flow in any of the comparisons. Early patency in the 5 selected patients is shown in Table 5. There were no significant differences between RIMA and LIMA patency. Comment Until 1985, an IMA graft had been used in only 6 patients during 15 years of CABG at this center. The LIMA became Table 4. lntraoperative Blood Flow in Pedicled Internal Mammary Artery Grafts" RIMA Flow LIMA Flow Recipient Coronary Artery (mumin) (ml/min) All recipients (3-8) ~ 7 1 ( 2 ~ ~ 7) 1 Left anterior descending (3-7) [ll] (3-6) 151 Recipient with small-vessel disease ? 13 (4-7) [71 ( 2 ~ [51 ) a Data are shown as the mean t the standard deviation with the range in parentheses. Numbers in brackets are numbers of patients. LIMA = left internal mammary artery; mary artery. RIMA = right internal mam-

5 Ann Thorac Surg 19!33;55: RAMSTROM ET AL 1489 Table 5. Early Patency of Pedicled Internal Mammary Artery Grafts in 5 Patients" Variable RIM LIMA (n = 5) (n = 46) All grafts 98% (49/5) 93% (43/46) Grafts to LAD 96% (25/26) 1% (12/12) Grafts to small vessel 1% (16116) 89% (16118) All anastomoses 98% (52/53) 92% (48152) a Numbers in parentheses are the number of patent versus the total number of grafts or anastomoses. LAD = left anterior descending coronary artery; LIMA = left internal mammary artery; RIMA = right internal mammary artery. increasingly popular as a coronary bypass graft throughout the world after the publication in 1984 of a study by Grondin and colleagues [9] in Montreal. Since then, use of the LIMA in patients with a diseased LAD has become routine as a result of increased 1-year survival and better patency relative to vein grafts [lo]. Reports of appealing possibilities with double IMA grafting ( prompted us to try the method, primarily in repeat CABG procedures necessitated by recurrent angina pectoris. The smooth accommodation of a pedicled IMA to uneven, adhesion-covered surfaces impressed us, quite apart from our reluctance to replace a thrombosed vein graft with a new vein. We were also attracted by the fit of the IMA grafts to small-calibered coronary arteries, not least because the shortcomings of vein grafts in this type of coronary artery disease have been demonstrated [ 11. Our favorable results after double IMA grafting [14] soon led to routine use of the RIMA (in addition to the LIMA) in another group of patients long disadvantaged by lack of veins for grafting, namely, patients with varicosity or previous stripping of the saphenous veins. Our reluctance to accept insufficient grafting and rely on possible collateral circulation in patients with exhausted venous reserves has taken us further into the area of multiarterial grafting with an increasing use of the gastroepiploic artery [15]. It is our impression that the diameter of the RIMA exceeds that of the LIMA, at least in the proximal and middle parts in the majority of right-handed patients. Use of intraluminal papaverine hydrochloride and distal ligation of the pedicle after heparinization to achieve pulsatile pressure "beating into the graft" during the heart cannulation procedure are important in our view to give an impression of the true diameter of the IMA graft. Usually, the RIMA is well adapted in size to the LAD, and by routing the RIMA in front of the ascending aorta, it is possible to reach beyond the middle part of the LAD. Such speculations and observations led us to do elective double IMA grafting also in a number of routine cases with favorable results [14]. In the case of double IMA grafting, we believe that the LIMA is preferable to revascularize the circumflex artery, as use of the RIMA usually necessitates routing through the transverse sinus, which may compromise graft blood flow [16]. In our experience, only the most proximal parts of the circumflex artery can be reached by the RIMA through the transverse sinus; the risks of overstretching and rotating the RIMA and the v v difficulties in managing pedicle bleeding limit the usefulness of this route. In previous reports [ll, 171, there has been concern about the flow capacity and patency of the RIMA compared with the LIMA. In these reports, the RIMA usually served as a "secondary" graft. In the present series, the RIMA was predominantly used to revascularize the LAD, a selection that probably offers better opportunities to analyze the true capacity of the RIMA. The early mortality rate of 3.5% in this inhomogeneous series seems acceptable. Our mortality rate of 6% in patients who underwent repeat CABG is comparable with the results of Galbut and associates [IS], who used double IMA grafting in their reoperations. Our patients with small-vessel disease had the highest mortality rate (8%), which probably indicates the severity of their coronary artery disease, and small-vessel disease turned out to be an independent risk factor for early mortality. The poor function of vein grafts anastomosed to such vessels has already been mentioned [l]. However, postmortem examination showed that 5 of the 6 patients studied had small-vessel disease, all artery graft anastomoses were patent, and 2 patients with small-vessel disease had thrombosed vein grafts. Further, as IMA graft flow and patency were unrelated to small-vessel disease (see Tables 4, 5), it seems that our selection of patients with this type of coronary artery disease for multiarterial grafting is justified. It is important to note that there were no deaths in our other two main subgroups of patients who did not have the serious risk factors of repeat CABG or smallvessel disease: those with varicose or stripped saphenous veins and the 79 routine cases (77 with double IMA grafting). The routing of the RIMA and the coronary artery receiving the RIMA graft were risk factors for neither early mortality nor significant enzyme release. This indicates that the RIMA is a suitable bypass graft in almost any situation. The risk associated with repeat CABG in a patient with the RIMA routed anteriorly to the ascending aorta is evident. However, the present analysis indicates that intraoperative blood flow and early postoperative patency of the RIMA and LIMA are comparable, which has also been shown for long-term patency [19]. The increased longevity and the better patency associated with the LIMA as compared with vein grafting are thus probably also valid for the RIMA. In addition, because the IMAs are used to revascularize the most important coronary arteries, and because new critical coronary artery stenoses rarely are found distal to functioning grafts [2], it follows that a patient with repeat angina and a functioning RIMA graft (with or without a LIMA graft) will have new coronary artery stenoses (or occluded vein grafts) in less important vessels and at most only one- or two-vessel disease. Thus, the probability of repeat angina is reduced by using the RIMA (and LIMA) at the primary operation, and many patients with repeat angina and functioning IMA grafts are probably well suited for percutaneous transluminal angioplasty.

6 149 RAMSTROM ET AL Ann Thorac Surg It is rational to hypothesize that sealing off an anteriorly routed RIMA graft by closing the pericardium with a patch may reduce the risk associated with repeat sternotomy. In an animal model [21], it has been shown that an absorbable (polyhydroxybuturate) pericardial patch reduces the amount of sternal adhesions. A randomized study using such patches in CABG is in progress at our center. In accordance with other reports [%5], we found that diabetes was an independent risk factor for sternal dehiscence, which was a problem only after double IMA grafting. Chronic obstructive lung disease is also associated with increased risk of this complication [4]. Double IMA grafting is thus contraindicated in routine patients with one of these two risk factors. The logistic regression model for early mortality gave rise to a quite impressive risk stratification (see Table 2). In terms of identifying action parameters, the modification of which may directly improve the prognosis of the patients, the risk model for significant enzyme release (see Table 3) gives important complementary results. ne of the independent risk factors of these two models were related to the RIMA. In essence, only three of the risk factors are modifiable: insufficient grafting (failure to graft one of the three main coronary artery systems), three or more vein graft anastomoses, and failure to give repeated cardioplegia when cross-clamp time exceeds 3 minutes. Using standard crystalloid hyperkalemic cardioplegia, the latter two need no further explanation. The influence of insufficient grafting using the present definition is self-evident, and the implication is that every effort to succeed with at least one graft on a severely diseased artery or one of its branches should be tried. Another interpretation is that the influence of insufficient grafting, which also can be the result of truly technically nongraftable arteries, is secondary only to the consequences of severe ischemic heart disease per se. In the majority of our patients with insufficient grafting, however, the reason quite simply was a lack of graft material. This further underlines the importance of having several alternative graft vessels. The influence of a large number of vein graft anastomoses may seem at variance with that of insufficient grafting. However, in the present series, the patients received an average of two artery grafts, which were anastomosed to the most important of the diseased coronary arteries. The implication is that a surplus of vein grafts anastomosed to less important coronary artery branches when the main artery has been revascularized probably leads only to increased ischemic strain on the myocardium. We conclude that the RIMA used in CABG has qualities that make it comparable to the LIMA, and that the RIMA is the primary supplementary graft to the LIMA. The RIMA can be used to revascularize any of the three main coronary artery systems but is primarily suited for the LAD and right coronary artery and their branches. Indications for use of the RIMA include any situation with insufficient venous reserves, small-vessel disease, which in our opinion is an indication for primary arterial grafting, and isolated right coronary artery disease in selected patients. Elective double IMA grafting in patients without exhausted venous reserves or small-vessel disease can be performed with an early mortality approaching zero and a low risk of sternal dehiscence, but diabetes mellitus and possibly also chronic obstructive lung disease should serve as contraindications. We thank Patricia Melo Castan6 for drawing Figure 1. References 1. Grondin CM, Castonguay YR, Lesperance J, Bourassa MG, Campeau L, Grondin P. Attrition rate of aorta-to-coronary artery saphenous vein grafts after one year: a study in a consecutive series of 96 patients. Ann Thorac Surg 1972; Singh RN, Sosa JA, Green GE. Long-term fate of the internal mammary artery and saphenous vein grafts. J Thorac Cardiovasc Surg 1983;86: Grossi EA, Esposito R, Harris LJ. Sternal wound infections and use of internal mammary artery grafts. J Thorac Cardiovasc Surg 1991; Kouchoukos NT, Wareing TH, Murphy SF, Pelate C, Marshall WG Jr. Risks of bilateral internal mammary artery bypass grafting. Ann Thorac Surg 199;49: Loop FD, Lytle BW, Cosgrove DM, et al. Sternal wound complications after isolated coronary artery bypass grafting: early and late mortality, morbidity, and cost of care. Ann Thorac Surg 199; SCE (Scandinavian Committee on Enzymes) Evaluation 11: isoenzymer (isoenzymes). Helsinki: rdkem, Dixon WD (ed). BMDP statistical software reprinting. Berkeley, Los Angeles, London: University of California Press, Lund, Piegaard H, Nielsen TT, Knudsen MA, Magnussen K. Thirty-day mortality after valve replacement for aortic stenosis over the last 22 years. A multivariate risk stratification. Eur Heart J 1991;12: Grondin CM, Campeau L, Lesperance J, Enjalbert B, Bourassa MG. Comparison of late changes in internal mammary artery and saphenous vein grafts in two consecutive series of patients 1 years after the operation. Circulation 1984;7O(Suppl 1): Loop FD, Lytle BW, Cosgrove DM, et al. Influence of the internal mammary artery graft on 1-year survival and other cardiac events. N Engl J Med 1984;311:132% Barner HB. Double internal mammary-coronary artery bypass. Arch Surg 1974;19: Geha AS. Crossed double internal mammary-to-coronary artery grafts. Indications, techniques, and results. Arch Surg 1976;111: Lytle BW, Cosgrove DM, Saltus GL, Taylor PC, Loop FD. Multivessel coronary revascularization without saphenous vein: long-term results of bilateral internal mammary artery grafting. Ann Thorac Surg 1983;36: Henze A, l7amstrom J, Nystrom SO. Bilateral internal mammary artery for coronary revascularization. Early experience of 1 cases. Scand J Ihorac Cardiovasc Surg 1989;23: Ramstrom J, Jaramillo A, Cadavid E, Thuren J, Henze A. A "new" intraabdominal artery. The pedicled right gastroepiploic artery for myocardial revascularization. Eur J Surg 1992; Rankin JS, Newman GE, Bashore TM, et al. Clinical and angiographic assessment of complex mammary artery bypass grafting. J Thorac Cardiovasc Surg 1986;92: Huddlestone CB, Stoney WS, Alford WC Jr, et ap. Internal mammary artery grafts: technical factors influencing patency. Ann Thorac Surg 1986;42: Galbut DL, Traad EA, Dorman MJ, et al. Bilateral internal mammary artery grafts in reoperative and primary coronary bypass surgery. Ann Thorac Surg 1991;52:2-8.

7 Ann Thorac Surg RAMSTROM ET AL 1491 RIMA GRARING 19. Fiore AC, Naunheim KS, Dean P, et al. Results of internal thoracic artery grafting over 15 years: single versus double grafts. Ann Thorac Surg 199;49: Reul GJ, Cooley DA, Ott DA, Coelho A, Chapa L, Eterovic I. Reoperation for recurrent coronary artery disease. Causes, indications, and results in 168 patients. Arch Surg 1979; Malm T. Absorbable patches in cardiac surgery: an experimental study [Dissertation]. Uppsala: Acta Universitatis Upsaliensis, tice From the Southern Thoracic Surgical Association The Fortieth Annual Meeting of the Southern Thoracic Surgical Association will be held at Marriott s Bay Point Resort, Panama City Beach, Florida, vember 4-6,1993. The Postgraduate Course will be held the morning of Thursday, vember 4, 1993, and will provide in-depth coverage of thoracic surgical topics selected primarily as a means to enhance and broaden the knowledge of practicing thoracic and cardiac surgeons. Applications for membership should be completed by July 1, 1993, and forwarded to Marion R. Lawler, MD, Membership Committee Chairman, Southern Thoracic Surgical Association, 41 rth Michigan Avenue, Chicago, IL Hendrick B. Burner, M D Secretary-Treasurer Southern Thoracic Surgical Association 41 rth Michigan Avenue Chicago, IL