ORIGINAL ARTICLES. Great progress has been made in the last decade to make saphenous vein bypass grafting one of the safest major

Transcription

1 ORIGINAL ARTICLES Important Anatomical and Physiological Considerations in Performance of Complex Mammary-Coronary Artery Operations Ellis L. Jones, M.D., Omar Lattouf, M.D., Jerre F. Lutz, M.D., and Spencer B. King 111, M.D. ABSTRACT One or more internal mammary artery (IMA) anastomoses were performed in 87% of 692 consecutive coronary artery bypass operations performed over a 20- month period. One IMA was used in 68% (N = 469) and both IMAs were used in 19% (N = 130). Only saphenous vein grafts were used in 13% (N = 93). The mean number of anastomoses (all types) was 3.5. Fifty-seven patients were having a reoperation; bilateral IMA grafting was performed in 23% (N = 13). In 60 patients, 3 or more IMA anastomoses were performed: 3 IMA anastomoses, 50 patients; 4, 9 patients; and 5, l patient. In 27 patients, repeat coronary arteriography was performed within 30 days of operation to evaluate dynamics of IMA, saphenous vein, and native coronary artery flow. Major flow or all flow was through the graft (vs. the native coronary artery) in 62% of in situ IMA bypass grafts, 86% of free IMA grafts and 94% of saphenous vein grafts. Hospital mortality excluding patients having reoperation was 1.7% (11/635); it was less than 1% for patients having either single IMA grafting procedures (a4371 or bilateral IMA grafting procedures (l! 117). Hospital mortality for patients receiving only saphenous vein grafts was surprisingly high, 7.4% (6/81). Major determinants of flow through the in situ IMA sequential graft are the size and flow of the IMA, the degree of proximal native coronary artery narrowing, the distally grafted to proximally grafted coronary artery size ratio, and probably the size of the side-to-side anastomosis. Technical factors important for multiple IMA grafts are the length and diameter of the IMA, use of the diamond anastomosis, absence of coronary artery disease at side-to-side anastomoses, epicardial location of the coronary artery, and no angulation distal to the sites of sequential anastomoses. Unless ideal conditions of IMA flow and size are met, an IMA graft supplying more than two coronary arteries might better be done with a free IMA graft. Great progress has been made in the last decade to make saphenous vein bypass grafting one of the safest major From the Divisions of Cardiothoracic Surgery and Cardiology, Emory University School of Medicine, Atlanta, GA. Presented at the Thirty-third Annual Meeting of the Southern Thoracic Surgical Association, White Sulphur Springs, WV, Oct 30-Nov 1, Address reprint requests to Dr. Jones, The Emory Clinic, 1365 Clifton Rd, NE, Atlanta, GA operations performed in the world today. As the next decade begins, emphasis will be on the development of a superior revascularization procedure. Limitations of the traditional saphenous vein coronary bypass operation center on the marked susceptibility of this conduit to degeneration and atherosclerosis with reduced flow, return of symptoms, and subsequent serious cardiac events. Several excellent studies have noted the susceptibility of saphenous veins to late atherosclerotic changes in 50 to 60% of grafts implanted 10 to 12 years [l, 21. In a recently published report, Loop and co-workers [3] noted the significantly improved 10-year survival of patients with triple-vessel disease receiving internal mammary artery (IMA) grafts (83%) compared with those in whom only saphenous vein grafts were used (71%). The goal of performing complex mammary artery operations is to expand the use of an ideal vascular conduit to as many situations as possible without compromise of technical result or regional myocardial perfusion. The present study analyzes a consecutive series of patients undergoing coronary revascularization in whom the IMA was used in a wide variety of clinical situations. An attempt was made to extend its use whenever possible, but situations where the IMA was contraindicated were also carefully noted. To better understand the physiology of IMA flow under different conditions, a small number of patients having 3 or more IMA anastomoses underwent repeat coronary arteriography. Interaction between the IMA, saphenous vein, and native coronary flow was studied. The correlation between arteriographic anatomy and mammarykoronary flow patterns determined at the time of postoperative catheterization was used to improve operative technique whenever possible. Material and Methods In the 20-month period between January, 1985, and September, 1986, 692 consecutive patients underwent isolated coronary bypass grafting at Emory University Hospital. The procedures were all done by one of us. The mean age of the patients was 59.8 years. Eighty-one percent (561) of the patients were male. This study is a retrospective analysis of all patients undergoing myocardial revascularization as an isolated procedure and includes 57 patients (8.2% of the total) having a reopera tion. Of 60 patients having 3 or more mammary-coronary artery anastomoses using the two IMAs (3 anastomoses, 50 patients; 4 anastomoses, 9 patients; 5 anastomoses, Ann Thorac Surg 43: , May 1987

2 470 The Annals of Thoracic Surgery Vol 43 No 5 May 1987 patient), 27 agreed to and had satisfactory repeat coronary arteriography 7 to 30 days following operation. Serial electrocardiograms of all 27 patients were reviewed for evidence of transmural myocardial infarction as manifested by transient ST segment displacement followed by either the development of Q waves or the loss of R wave amplitude. Details of the methods used to evaluate the postoperative arteriograms have been described previously [4]. Basically, because opacification of the coronary arteries has been used as a determinant of collateral blood flow [5, 61, a similar grading system was used to compare flow through the native circulation with that supplied by the surgical conduits to the individual arterial segments. Grade 0 flow indicated no anterograde flow of contrast medium through the injected vessel to the bypass segment distal to the anastomotic site. Grade 1 flow signified minimal opacification distal to the anastomotic site with considerably less opacification of the vessel distal to the anastomosis compared with the selectively injected native vessel or conduit proximal to the anastomosis. There was incomplete visualization of the distal vessel. Grade 2 flow indicated distal opacification somewhat less than the vessel proximal to the anastomosis but good visualization of the entire distal vessel. Grade 3 flow indicated opacification of the vessel distal to the anastomosis equal to that seen in the native vessel or conduit being injected with full visualization of the entire length of the grafted vessel distal to the anastomotic site. For each anastomotic site, the following data were analyzed: size of the vessel at the anastomotic site, estimation of anterograde flow through the native circulation and the graft conduit, and degree of obstruction noted proximal to the anastomotic site on the preoperative arteriogram. Technique of Operation The right (RIMA) and left (LIMA) IMAs are dissected from the sternum with the electrocautery in usual fashion before heparinization and cannulation. The arteries are dissected proximally as high as possible without injuring the phrenic nerve (which may lie in close proximity to the proximal mammary pedicle in some patients), the subclavian vein, or the proximal mammary trunk on either side. The distal IMA is not divided until after the institution of cardiopulmonary bypass, at which time papaverine (Eli Lily and Co., Indianapolis, IN) (1 :4 dilution) is injected through a small No. 22 Silastic cannula into the artery. This maneuver affords an optimal diameter for the side-to-side anastomosis, but care must be taken to avoid dissection of the intima. To minimize this risk, the catheter is withdrawn a short distance before injection to eliminate impingement of the catheter tip on the side of the arterial wall. Systemic hypothermia to a nasopharyngeal temperature of 25 C is used in addition to cold crystalloid oxygenated cardioplegic arrest. This degree of hypothermia provides added protection to minimize myocardial in- jury from the slightly prolonged period of ischemia required to do the operation. The ischemic time can be reduced by performing most measurements and determining the sites and angles of the anastomoses before the aortic cross-clamp is applied. For side-to-side sequential anastomoses, a small incision (3 to 4 mm) is made in the fascia1 side of the distended IMA with a special diamond-tip knife (Codman & Shurtleff, Inc., Randolph, MA). This allows atraumatic entrance into the artery without tearing its side wall as is frequently seen with other knives. The incision is lengthened as necessary with specially designed micro-tip coronary vascular scissors (Codman & Shurtleff, Inc., Randolph, MA). Flow is measured at this time. Direct handling of the IMA wall, using IMA precisiontip forceps (Codman & Shurtleff, Inc., Randolph, MA) is now possible without fear of injuring the artery. Whether performing in situ or free grafts, the side-toside anastomosis is always performed before the end-toside anastomosis. When the most favorable position for the IMA is determined to be perpendicular rather than parallel to the long axis of the arterial segment to be grafted, a diamond anastomosis is performed. Care must be taken to keep both the coronary arteriptomy and the IMA opening small to avoid the "gull wing" stenosis seen with too large an incision in the coronary artery. It is interesting that in the performance of a number of these operations where incision in the IMA has been slightly longer than desired, the gull wing deformity is not created if the coronary arteriotomy is maintained at minimal length. Side-to-side anastomoses using the diamond technique must be made in areas of the coronary artery that are relatively free from atherosclerotic disease to avoid the need to extend the coronary arteriotomy incision. It has been noted that even when the terminal end of the IMA is used in areas of severe coronary disease, deep suturing to reapproximate intimal plaque produces telescoping of the distal end of the IMA into the coronary artery, thereby resulting in terminal IMA constriction and unacceptably reduced flow. The three types of side-to-side anastomoses with the IMA graft (in situ or free) are depicted in Figure 1. When the IMA approaches the coronary artery at a 45 degree rather than a 90 degree angle, better apposition of the two arteries is obtained by making an oblique incision in the coronary artery with the diamond knife, as is shown in Figure 1C. To accomplish this safely, however, the recipient coronary artery must be free from disease. Use of a given type of anastomosis is determined by the angle at which the IMA most naturally crosses the coronary artery. All side-to-side and many end-to-side anastomoses are performed in continuous fashion using 8-0 Deknatel monofilament suture (Howmedica, Inc., Queen's Village, NY) with a special ultrafine tipped-needle holder (Codman & Shurtleff, Inc., Randolph, MA). To avoid kinking of the IMA pedicle (especially at the first sequential anastomosis), a wedge of pericardium is re-

3 471 Jones, Lattouf, Lutz, King: Complex Mammary-Coronary Artery Operations Fig I. The three types of side-to-side anastomoses using the internal mammary artery (IMA) in situ or as a free graft. An oblique incision is made in the coronary artery when the approach of the IMA is at a 45 degree angle to the coronary artery. (LAD = left anterior descending coronary artery; MOM = middle obtuse marginal branch of circumflex; ALOM = anterolateral obtuse marginal branch of circumflex; DIAG = diagonal.) moved superiorly to allow the IMA to fall into the pleural space. On the right side, this incision frequently increases the length of the mammary pedicle to reach the distal right coronary artery or its posterior descending branch. Care is taken not to injure the phrenic nerve in the pericardial reflection. The pericardium had best be divided before performing the distal anastomosis to avoid traction on the IMA pedicle with the increased risk of anastomotic tearing and bleeding. Fixation sutures on the IMA pedicle to prevent torsion and tension are routinely used after each anastomosis is completed. When the IMA is selected as a free graft, it is dilated with papaverine as already described and amputated close to its origin from the subclavian artery; care is taken to avoid the phrenic nerve. For multiple side-toside anastomoses, careful orientation is required to avoid twisting the artery. When the IMA pedicle is used as a free graft for 3 distal anastomoses, orientation of the pedicle becomes extremely important to avoid an unrecognized 360-degree rotation. A surgical marking pen is useful to maintain alignment. When the artery is being dissected out of the free graft pedicle at the sites designated for side-to-side anastomoses, a 1.O-mm metallic probe is inserted to avoid cutting the artery wall. All side-to-side anastomoses are performed first and the end-to-side anastomosis, at the end. The proximal anastomosis is done last using a 2.7-mm disposable punch (Deknatel, Goosen, Orlando, FL) and 6-0 Prolene suture on a BV-1 needle (Ethicon, Inc., Johnson & Johnson, Somerville, NJ). No interposed segments of vein are necessary for this anastomosis (Figs 2, 3). Results Only saphenous vein grafts were used in 13% (93 patients) and some form of IMA combination, in the remaining 599 patients (87%). A single IMA was used in 68% (469 patients) and both IMAs in 19% (130 patients). Of the 57 patients having reoperative coronary bypass grafting, a single IMA was used in 56% (32 patients) and both IMAs were used in 23% (13 patients). The remaining patients had vein grafts only (21%). The IMA was used as a free graft to the aorta in 29 patients (5% of all IMAs used). The distribution was almost equal between the LIMA, 13 patients, and the RIMA, 16 patients. The in situ LIMA was used to graft the left anterior descending coronary artery (LAD) in 421 patients (92.5%) but was also used as a single graft to the first or second diagonal in 2% of patients (N = 8), the middle obtuse marginal branch of the circumflex coronary artery in 4% (N = 20), and the anterolateral obtuse marginal branch, the posterolateral obtuse marginal branch, or the posterior descending branch of the circumflex coronary artery in 1% (N = 6). The LIMA was used as a free

4 472 The Annals of Thoracic Surgery Vol 43 No 5 May 1987 Fig 2. lntermal mammary artery (IMA) free graft with 3 distal anastomoses to marginal branches of the circumflex coronary artery. (MOM = middle obtuse marginal branch; ALOM = anterolateral obtuse marginal branch; PLOM = posterolateral obtuse marginal brunch.) Fig 3. Free graft right internal mammary artery to middle and posterior marginal branches of the circumflex coronary artery. This demonstrates the added length and flexibility provided by the free graft. graft to only three arteries: the LAD in 9 patients, the diagonal in 2, and the anterolateral obtuse marginal branch of the circumflex in 2. Both the in situ and free graft RIMAs were used with greater diversity than was the case with the LIMA grafts (Fig 4). Sequential IMA Grafting Sequential IMA combinations (2 or 3 anastomoses per IMA) for both the LIMA and RIMA are depicted in Figures 5, 6, and 7. The LIMA was infrequently used as a free sequential graft (3 of 105 patients); when applied in this fashion, it was because the artery in situ had inadequate blood flow or was too short to accomplish 2 anastomoses. There were 60 patients in whom use of both IMAs was extended to accomplish 3 (N = 50), 4 (N = 9), and 5 anastomoses (N = 1). By far, the most frequent use of the in situ LIMA for sequential grafting was to perform 2 anastomoses to the LADldiagonal and middle or distal LAD (79% of patients) (see Fig 5). However, it was also used effectively to sequentially graft marginal branches of the circumflex. Although used early in the series, in situ sequential grafting to the anterolateral obtuse marginal branch Of the circumflex and LAD in which the limb of the IMA is taken anterior to the LAD is now limited to unusual anatomical circumstances because of

5 473 Jones, Lattouf, Lutz, King: Complex Mammary-Coronary Artery Operations RCA (4lo/o) In sltu RlMA (n=99 ptr) Diag (2%)' I aldm (91) LM (1%) Free Graft RlMA (11-16 ptr) alom. mom or plom (42%) (22%) (22%) pda'(ll%) Fig 4. Use of the right internal mammary artery (RIMA) as both an in situ and free graft. No sequential anastomoses are included. (RCA = right coronary artery; LAD = left anterior descending coronary artery; RCAd = distal right coronary artery beyond origin of posterior descending corona ry artery; Diag = diagonal; LM = left main coronary artery; alom = anterolateral obtuse marginal branch of circumflex; pda = posterior descending branch of RCA; mom = middle obtuse marginal branch of circumflex; plom = posterolateral obtuse marginal branch of circumflex.) a greater chance of producing acute angulation just distal to the first side-to-side anastomosis. Sequential grafting in which 2 or 3 anastomoses per IMA is desirable is more readily attainable with less chance of pedicle angulation when grafting obtuse marginal arteries rather than trying to combine a ramus or anterior marginal branch with a first diagonal or LAD. Postoperative Catheterization Evaluation In the 27 patients with 3 or more anastomoses using the two IMAs and restudied shortly after operation, the major portion or all of the coronary flow could be attributed to the in situ IMA in only 62% of the bypass grafts (44/ Fig 5. Sequential left internal mammary artery (IMA) combinations (2 or more anastomoses per IMA): distribution of in situ and free grafts. (LAD = left anterior descending coronary artery; D1 and Diag, = first diagoml; ALOM = anterozateral obtuse marginal branch of circumflex; and D2 and Diag, = second diagonal; MOM = middle obtuse marginal branch of circumflex; PLOM = posterolateral obtuse marginal branch of circumflex.) 71). Blood flow through the IMA graft was equal or nearly equal to that through the native coronary artery in 34% of anastomoses (24/71). Major flow or all flow occurred through the native artery in 3 (4%) of 71 bypass grafts. For those instances when most or all flow was through the in situ IMA graft, the average reduction in diameter of the proximal coronary artery to which the graft was placed was 83%. For those arterial segments in which there was equal flow through the in situ graft and the native circulation, the mean diameter reduction was 63% (p <.05). There were only three observations available for instances when major flow occurred through the native coronary artery; therefore statistical evaluation was rendered unreliable. However, it was interesting that in 2 of the 3 anastomoses when major flow occurred through the native coronary artery, proximal coronary diameter reduction was only 50%. Quantitatively, these findings contrasted with those for the IMA free graft and saphenous vein graft. The largest quantity of coronary flow in the former occurred through the graft, but with the saphenous vein, coronary flow was supplied almost exclusively by the graft rather than the native circulation (Table). With the vein graft, flow patterns were the same regardless of degree of proximal native artery obstruction. Patency without stenosis for the 95 bypasses with the in situ IMA, free IMA, or saphenous vein graft was 96% (73/76), 100% (7/7), and 100% (18/18), respectively. Perioperative infarction (new Q wave) occurred in 1 patient and was in the distribution of a saphenous vein graft. Mortality Hospital mortality excluding the 57 patients having reoperation was 1.7% (11/635 patients). For patients having single IMA grafting, hospital mortality was less than 1.0% (4/437 patients). For those having bilateral IMA grafting, hospital mortality was also less than 1.0% (11 117). For patients having only saphenous vein grafting, hospital mortality was 7.4% (6/81 patients). Among the 57 patients undergoing reoperation, there Left IMA In Situ (N.102) Free (N=3) LAD LAD D, ALOM LAD LAD LAD ALOM MOM MOM ALOM LAD LAD D, D, D, D, Diag LAD ALOM MOM PLOM MOM MOM Diag, Diag, ALOM D, PLOM (1) (1) (2) (2) (61) (20) (5) (6) (3) (1) (1) (1) (1)

6 474 The Annals of Thoracic Surgery Vol 43 No 5 May 1987 n In Situ RCA LAD Diag PDA Right IMA (N.9 pts) ALOM MOM Diag, ALOM PLOM MOM PLOM (1) (1) (1) Free ALOM AM of ALOM ALOM RCA MOM PDA MOM (1) (1) (1) Fig 6. Sequential right internal mammary artery (IMA) combinations (2 or more anastomoses per IMA): distribution of in situ and free grafts. (LAD = left anterior descending coronary artery; nag = diagonal; RCA = right corona ry artery; PDA = posterior descending branch of right coronary artery; ALOM = anterolateral obtuse marginal branch of circumflex; PLOM = posterolateral obtuse marginal branch of circumflex; MOM = middle obtuse marginal branch of circumflex; Diagz = second diagonal; AM of RCA = acute marginal branch of right coronary artery.) were 6 deaths. Two had grafting with a single IMA, 2 had grafting with both IMAs, and 2 had grafting with only saphenous veins. Comment The superior late patency of the IMA compared with the saphenous vein graft has been well documented in previous studies [3, 7-10]. Reports from the Montreal Heart Institute [2] and the Cleveland Clinic [l] have noted the susceptibility of saphenous vein grafts to atherosclerosis with approximately 50 to 60% of these grafts significantly infiltrated with obstructive atherosclerotic disease at the end of 10 to 12 years. Because of these observations, extensive clinical investigations have been undertaken to search for a better conduit for coronary artery bypass procedures. In recent years, there has been intensive interest in extending the mammary artery operation to graft as many coronary arteries as possible. Previous studies have indicated the adequacy of blood flow through the IMA to meet resting and increased myocardial demands, but the majority of these studies have referred to use of the IMA as a single graft [7, 9, 11-13]. Very little information has been provided concerning flow patterns through the IMA when it has been used as a conduit for two or three bypasses. The assumption that a single IMA can be used for more than one graft is strengthened by the findings in three series [3, 9, 121 of an increase in IMA size and blood flow with the passage of time. The clinical observations by McBride and Barner [lo], Myojin [14], Feruse [15], and Grondin [16], and their coworkers stressed certain guidelines to assure adequacy of IMA flow to a single coronary artery. The conclusion of all these studies was that this flow should be adequate if it is at least that of the bypassed coronary artery. Variables of flow were not stated. Previous studies, however, have not offered answers to the question of adequacy of mammary artery perfusion through 2 or possibly 3 anastomoses using the in situ graft. Guidelines for use in this situation are scant, and considerable judgment is now required to know when to use the IMAs for complex operations as well as when to abandon the IMA and use the saphenous vein graft. We believe our series is representative of a busy cardiac surgical practice. Although the series covers a 20- month period, use of the IMA for complex coronary grafting procedures has increased considerably during the last 6 months. Even within the time frame of the study, both IMAs were used in the revascularization procedure in approximately 1 in 5 patients. Careful observations in the operating room combined with a detailed analysis of the postoperative arteriograms have been of inestimable value in developing the operative technique. Use of complex mammary operations is ideal in younger patients in an effort to extend the life of the bypass procedure. It has been evident over the past year that these same patients who are the most ideal candidates for bilateral mammary artery procedures are also those who have become the most attractive candidates for percutaneous multivessel coronary angioplasty. It is likely that the two procedures will be compared in the future. Patients in our series were relatively young (mean age, 59.8 years). The majority of the older patients had either saphenous vein grafts only or a single IMA graft when the veins were extremely poor. Recently, the IMA has been used with increasing frequency as a single graft in many of our elderly patients because this same population often is found to have marginal or unsatisfactory saphenous veins.

7 475 Jones, Lattouf, Lutz, King: Complex Mammary-Coronary Artery Operations Fig 7. In situ sequential right internal mammary artery (IMA) graft to right coronary artery (RCA) (side-to-side) and posterior descending branch (PDA) (end-to-side). Because complex mammary artery grafting requires additional operative time, this procedure is not generally desirable for patients in unstable condition. Because of the extra time of myocardial ischemia, patients with good left ventricular function are preferred whenever possible, particularly when sequential grafting is contemplated. Because male patients have, in general, a larger body surface area and a larger chest wall size, they have been the most ideal candidates for sequential MA Flow Pattern for Coronary Artery Grafts Flow Equal between Major Flow Major Flow Graft and through through Native Native Graft Graft Coronary Coronary In situ IMA 62% (44/71) 34% (24/71) 4% (3/71) IMA free graft 86% (6/7) 14% (117) None sv 94% (17/18) 6% (1/18) None IMA = internal mammary artery; SV = saphenous vein grafting. Although not exclusively true, most elderly female patients with reduced body surface area and excessive body fat have been poor candidates, and extensive amounts of operative time may be lost in pursuing a complex mammary procedure. The three most common contraindications to use of the IMA regardless of other factors are inadequate artery diameter, length, and blood flow. In situations where the coronary arteries are extremely diseased, use of the IMA is much less flexible than that of the saphenous vein. Although the saphenous vein and terminal portion of the IMA can be used to bridge a long arteriotomy, this cannot be done with the IMA side-to-side anastomosis, particularly when performed with the diamond technique. We have found that considerable judgment is required to know when to and when not to use the IMA. Limitations of the IMA graft must be taken into account when planning the operation and prior to the anastomosis to prevent prolonged ischemic times. A contraindication to the use of the IMA as a sequential graft is the finding of an intramyocardial coronary artery (usually the LAD). In this situation, the secondary limb of the IMA becomes obstructed on exiting the epicardial groove on its course to the more distally grafted artery. Because of potential healing problems, use of both IMAs is contraindicated

8 476 The Annals of Thoracic Surgery Vol 43 No 5 May 1987 when the patient has had a previous mastectomy or irradiation to the parasternal areas. Postoperative arteriograms have been extremely informative in evaluating the operative procedure. They have demonstrated that blood flow through the distal limb of the IMA pedicle is reduced when a proximal sequential anastomosis has been performed to a native coronary artery with a noncritical stenosis. Whether this finding is a minor contraindication to sequential IMA grafting or whether these patterns of flow will change as proximal stenosis progresses is unknown at present. Conversely, early postoperative arteriograms have also shown that there appears to be a trend toward increased blood flow through the distal mammary segments when the first sequential anastomosis is performed to an artery that has a diameter less than that of more distally grafted arterial segments. It has not been possible to discern the importance of the size of the first side-to-side anastomosis itself, but this may have a substantial influence on blood flow through secondary and tertiary limbs of the IMA. These observations of flow using the in situ IMA graft were quantitatively different from those with the free IMA and saphenous vein grafts. There was usually greater flow through the graft than through the native artery with the free graft IMA, and there was flow almost exclusively through the graft rather than the native circulation with the saphenous vein. The latter observation for the vein graft was true regardless of the degree of proximal native coronary artery obstruction. From intraoperative observations and analyses of postoperative arteriograms, the best in situ sequential grafting situation appears to be an IMA with a diameter greater than 2.0 mm, IMA flow of at least 150 ml per minute (at 60 to 80 mm Hg), and a side-to-side anastomosis no greater than 3 mm in length. In the unusual situation where the in situ IMA is used for 3 anastomoses, IMA flow should be excessive (150 to 300 ml per minute), the diameter should be large (equal to or greater than 2.5 mm), and acute angles at all sequential anastomotic sites should be avoided entirely. Obstructive angulation of sequential grafts is most apt to occur when the in situ LIMA is taken anterior to the LAD, and is most apt to be distal to a sequential anastomosis performed to either a ramus intermedius or an anterolateral obtuse marginal branch of the circumflex. It is much safer for the IMA to pursue a posterior course to the other marginal branches of the circumflex. When using the IMA either individually or as a sequential combination, assessment of its flow is important to prevent regional myocardial wall hypoperfusion. Effective IMA flow is dependent on the diameter of the coronary artery bypassed, the degree of proximal native coronary obstruction, the viability and the size of the total myocardial muscle mass to be perfused, and the length of the IMA. The free graft IMA has been used to increase the effective length of the IMA, to provide more precise control of the angles formed at each sequential anastomosis, to increase the technical ease of performing side-to-side anastomoses particularly along the obtuse margin of the heart, and possibly to increase absolute flow through the IMA itself. The last is theoretically accomplished by decreasing the resistance determined by graft length (aorta-coronary anastomosis versus subclavian-coronary anastomosis), eliminating all chest wall branches, and possibly constructing a more flow-favorable angle from the aorta. Based on our experience and the excellent 5-year patency of 90% reported by Loop and colleagues [17], we believe the free graft IMA is an alternative that should be considered more often for multiple sequential grafts to the circumflex system. Use of both IMAs has become increasingly frequent for reoperative coronary bypass procedures in recent months. However, sequential anastomoses performed in this situation can be difficult and have not been used extensively in our experience. The IMA free graft is also more difficult when the anastomosis of the proximal end of the artery is made to the thickened aortic wall. This can be facilitated by suturing the proximal IMA to the old vein graft amputated close to the aorta. This is an area rarely affected by vein graft atherosclerosis. Technical modifications developed to extend the length and thus the use of the in situ IMA graft have been to dissect both the RIMA and LIMA far into the rectus sheath, divide the pericardial reflection down to the phrenic nerve, and, if necessary, divide the endothoracic fascia on the back of the IMA pedicle. In this way, the middle marginal branch of the circumflex, the LAD, the medial diagonal, and the posterior descending branch of the right coronary artery can be reached with the RIMA (see Fig 7). Although the anterolateral obtuse and middle obtuse marginal branches of the circumflex may be reached with the RIMA through the transverse sinus, we have found that the operation is safer and technically easier when the in situ RIMA is taken anterior to the aorta and pulmonary artery. If this approach cannot be accomplished because of inadequate length, the LIMA is used for the circumflex system and the RIMA, for the LAD or as a free graft. The LIMA is usually adequate for sequential grafting of all branches of the circumflex as well as the more traditional diagonal and LAD. Ideally, with the in situ IMA it is preferable to place the first side-to-side anastomosis to the most important artery (i.e., the LAD rather than its diagonal branch). Practically, however, it is often difficult or impossible to reach the first diagonal with the terminal limb of either IMA. Unlike the situation encountered with saphenous veins, fixation sutures to stabilize the IMA pedicle are important when using the IMA. This is particularly true when using the bifurcation branches of the RIMA where excessive tension on the anastomotic site by the weight of the pedicle may produce stretching and subsequent anastomotic stenosis. Although the overall hospital mortality of 1.7% (exclusive of patients having reoperation) and the hospital mortality for single and bilateral IMA procedures were satisfactory, we were somewhat surprised at the hospital mortality of 7.4% for patients having saphenous vein

9 477 Jones, Lattouf, Lutz, King: Complex Mammary-Coronary Artery Operations grafts only. The explanation for this difference is probably related to the use of the IMAs for the better surgical candidates, although this inherent selection bias was not consciously appreciated when the series was in progress. Extended use of the IMA for routine coronary artery bypass grafting should do much to reduce the 8- to 10- year failure rate seen with the saphenous vein graft. However, considerable judgment is necessary when selecting the operation in a variety of situations, and performance of multiple sequential IMA anastomoses is technically more difficult than when saphenous veins are used. From our experience, it has been just as important to know when to abandon a planned mammary operation as it is to know the indications for its use. The goals of the operation must be decided prior to the procedure itself. In situations of salvage, where expediency becomes the greatest priority, use of the IMA for more than one graft probably should not be considered. Special instruments have been devised to improve handling and suturing of the IMA and as with the saphenous vein graft operation, performance of these anastomoses will be improved with increased experience. Overall, extension of the use of the IMA to complex procedures offers substantial advantages to a group of patients who stand to benefit most by preventing both early and late conduit failure. The authors are indebted to Norma Shetterly for her invaluable research assistance in the preparation of the manuscript. References Lytle BW, Loop FD, Cosgrove DM, et al: Long-term (5 to 12 years) serial studies of internal mammary artery and saphenous vein coronary bypass grafts. J Thorac Cardiovasc Surg 89:248, 1985 Campeau L, Enjalbert M, Lesperance J, et al: Atherosclerosis and late closure of aortocoronary saphenous vein grafts: sequential angiographic studies at 2 weeks, 1 year, 5 to 7 years, and 10 to 12 years after surgery. Circulation 68:Suppl 2:107, Loop FD, Lytle BW, Cosgrove DM, et al: Influence of the internal mammary artery graft on 10-year survival and other cardiac events. N Engl J Med 314:1, Jones EL, Lutz JF, King SB, et al: Extended use of the internal mammary artery graft: important anatomic and physiologic considerations. Circulation 74:Suppl 3:42, Wainwright RJ, Maisey MN: Functional significance of coronary collateral circulation during dynamic exercise evaluation by thallium-201 myocardial scintigraphy. Br Heart J 43:47, McKeown TP, McClelland JS, Bone DK, et al: Nitroglycerin as an adjunct to hypothermic hyperkalemic cardioplegia. Circulation 68:Suppl 2:107, Tyras DH, Bamer HB, Kaiser GC, et al: Bypass grafts to the left anterior descending coronary artery. J Thorac Cardiovasc Surg 80:327, Lytle BW, Cosgrove DM, Saltus GL, et al: Multivessel coronary revascularization without saphenous vein: long-term results of bilateral internal mammary artery grafting. Ann Thorac Surg 36:540, Tector AJ, Schmahl TM, Janson 8, et al: The internal mammary artery graft. JAMA 246:2181, McBride LR, Bamer HB: The left internal mammary artery as a sequential graft to the left anterior descending system. J Thorac Cardiovasc Surg 86703, Kamath ML, Matysik LS, Schmidt DH, Smith LL: Sequential internal mammary artery grafts. J Thorac Cardiovasc Surg 89:163, Vogel HK, McFadden RB, Spence R, et al: Quantitative assessment of myocardial performance and graft patency following coronary bypass with the internal mammary artery. J Thorac Cardiovasc Surg 75:487, Schmidt DH, Balu F, Hellman C, et al: Isoproterenolinduced flow responses in mammary and vein bypass grafts. J Thorac Cardiovasc Surg 80319, Myojin K, Weiss G, Mee R, et al: Functional comparison of coronary bypass grafts of the saphenous vein and internal mammary artery. J Thorac Cardiovasc Surg 79:713, Feruse A, Klopp EH, Brawley RK, Gott VL: Hemodynamics of aorta-to-coronary artery bypass experimental and analytical studies: Ann Thorac Surg 14:282, Grondin CM, Lesperance J, Bourassa MG, Campeau L: Coronary artery grafting with the saphenous vein or internal mammary artery. Ann Thorac Surg 20:605, Loop FD, Lytle BW, Cosgrove DM: Free (aorto-coronary) internal mammary artery graft: late results. J Thorac Cardiovasc Surg 92:827, 1986