F mary artery (IMA) graft carries a greater long-term

Internal Mammary Artery Grafts: The Shortest Route to the Coronarv Arteries J Thomas J. Vander Salm, MD, Sultan Chowdhary, MD,. N. Okike, MD, A. Thomas ezzella, MD, and Michael K. asque, MD University of Massachusetts Medical Center, Worcester, Massachusetts Inadequate length can limit the use of the internal mammary artery (IMA) for coronary revascularization. By following the shortest route from its origin to the recipient coronary artery, IMA use can be maximized. Seven cadavers were studied to determine that shortest route for the left and right IMAs. The shortest route for the left IMA to the left anterior descending coronary, diagonal, and circumflex coronary arteries was through the pericardium (p 5.1). For the right IMA, the significantly shortest routes were across the anterior heart for the left anterior descending and diagonal arteries, through the right pericardium for the right coronary artery or posterior descending artery, and through the pericardium and transverse sinus for the circumflex artery. Thus, any coronary artery can be reached with an in situ IMA, and the route through the pericardium is markedly shorter to ipsilateral coronary arteries. (Ann Thorac Surg ) or coronary artery bypass grafting, the internal mam- F mary artery (IMA) graft carries a greater long-term patency, a decreased need for reoperation, and, perhaps, a decreased recurrence of angina when compared with the saphenous vein graft [l-51. However, the number and length of IMAs limit their use. Numerous methods of extending the use of IMA grafts have been suggested, including augmenting their length by multiply transecting the fascia in the IMA pedicle [6, 71, constructing sequential grafts [7-111, using free IMA grafts by constructing an aorta-ima anastomosis [7, 11, 121, using both IMAs [7, 9, 13, 141, using both limbs of the IMA beyond its bifurcation [9], gaining length by dissecting the artery well down into the rectus muscle [7], and choosing the shortest route from the IMA origin to the recipient coronary artery [2, 7, 8, 1, 11, 151. This study was designed to find the shortest route for each IMA to each feasible recipient coronary artery. Material and Methods Seven autopsies were performed through the usual Rokitansky incision on 3 men and 4 women. Measurements were made of the patient's weight, height, and sternal lengths, and, on each side, the length of the IMA from its origin to the fifth intercostal space and to the costal margin (Table 1; Fig 1). The pericardium was opened in the midline. Then measurements were made from the origin of the IMA to each of the following feasible coronary arteries: right (RCA), posterior descending (DA), cicumflex marginal, diagonal, and left anterior descending (LAD). All measurements were made under the following conditions: Accepted for publication Sep 21, 1988 Address reprint requests to Dr Vander Salm, University of Massachusetts Medical Center, 55 Lake Ave N, Worcester, MA 1655. The cadaver was intubated with a cuffed endotracheal tube, usually through a tracheostomy. An Ambu bag was used to inflate the lungs to an extent approximating full inspiration before each measurement. Measurements were taken from a string draped without slack from the origin of the IMA to the coronary artery. Care was taken to avoid so much tension on the string that it would cut into tissues. On a given coronary artery, the measurement was always made to the same point, that to which a coronary graft would most commonly be performed. All statistical comparisons were performed using paired t tests (RS 1 statistical program) to assess differences between any two pertinent measurements on the same individual, thus obviating the need to normalize measurements between individuals. A p value of less than.5 was considered significant. Table I. Group Description Variable Age Weight (kg) Height (crn) S-G (cm) S-X (crn) LIMA-ICS (cm) LIMA-CM (cm) (cm) RIMA-CS (cm) Value" 59? 12.75 71.9 2 25.4 165.5? 11.4 16.4? 1.9 2.4? 1.9 17.4? 1.1 22.1? 1.1 17.5? 1.6 22.2?.6 a Data are shown as the mean 2 the standard deviation. CM = costal margin; ICS = fifth intercostal space; LIMA = origin of left internal mammary artery; RIMA = origin of right internal mammary artery; SG = sternal notch to junction between xiphoid and sternum; S-X = sternal notch to xiphoid tip. 1989 by The Society of Thoracic Surgeons 3-4975/89/$3.5

422 VANDER SALM ET AL Ann Thorac Surg Table 2. Distance From Internal Mamma y Arteries to Corona y Arteries robability of Null Hypothesis Distancea Route (cm) 1 2 3 4 LIM A-L AD M LIMA-Diag M LIM A-Cir M IJ RIMA-LAD T RIMA-Diag T RIMA-RCA RIM A - D A T 17.79 f 2. o,oo2 15.57 2 1.27 o,ol.2 14.14 f 1.3 18.41 * 2.24 o.ool 15.71 f 1.47 o.oo3.1 13.43 2 1.5 22.43 2 3.35 o.ool 19.21 f 3.86 o,ol.1 16.21 2 2.36 17.29 2 2.41 o.23 18.49 f 3.43 o,3.6.2 2.5 2 2.61 o.ool.4 21.86 2.5 16.7 f 1.62 o,oo8 18.1 2 2.61 o,41.1 18.71 f 2.14 o,16 19.5 2 2.94 16.7 2.89 o,lo 15.14 t 1.89 24.5 2 3.25 a Data are shown as the mean -C the standard deviation.22.4 Cir = circumflex marginal coronary artery; Diag = diagonal coronary artery; LAD = left anterior descending coronary artery; LIMA = left internal mammary artery; M = medial to lung; = over lung (LIMA) or pericardium (RIMA); = over pericardium, through transverse sinus; I' = through pericardium; DA = posterior descending coronary artery; T = through pericardium, through transverse sinus; RCA = right coronary artery; RIMA = right internal mammary artery; 1 = between adjacent values; 2 = between first and third values; 3 = between second and fourth values; 4 = between first and fourth values Fig 1. osterior view of sternal plate with internal mammary artery (IMA) measurements from origin to the fifth intercostal space (5th ICS) and to the costal arch (L = left; R = right.) Fig 2. Measurements of three routes from origin of left internal mammary artery to left anteriov descending coronary artery. (* = shortest route.) The following measurements were made from the IMA origin to the coronary artery (Tables 2, 3; Figs 2-9): Left IMA (LIMA) to LAD, diagonal, and circumflex over anterior surface of left lung. LIMA to LAD, diagonal, and circumflex medial to left lung. LIMA to LAD, diagonal, and circumflex medial to left lung and through a pericardial incision immediately anterior to the left phrenic nerve. Fig 3. Measurements of three routes from origin of left internal mammary artery to diagonal artery. (* = shortest route.)

Ann Thorac Surg 1989;47421-7 VANDER SALM ET AL 423 Table 3. Distance From Right Internal Mammary Artery to Circumflex Artery Route OR OL R L T Distance robability of Null Hypothesis (cm) OR OL R L T 24.64 2 4.97 21.7 2 3.27.12 23.79 & 3.92.44.9 19.75? 4.5.13.59.15 18.79? 2.8.2.13.3.67 2.71 2 3.96.16.81.78.72.4 a Data are shown as the mean 5 the standard deviation. OL = over pericardium, over heart to left side; OR = over pericardium, around right side of heart; = over pericardium, through transverse sinus; L = through pericardium, over heart to left side; R = through pericardium, around right side of heart; T = through pericardium, through transverse sinus. Right IMA (RIMA) to LAD and diagonal over the right pericardium and anterior to the heart. RIMA to LAD and diagonal through a pericardial incision immediately anterior to the right phrenic nerve and anterior to the heart. RIMA to LAD and diagonal through a pericardial incision immediately anterior to the right phrenic nerve and through the transverse sinus. RIMA to LAD and diagonal over the pericardium and through the transverse sinus. RIMA to circumflex over the right pericardium and along and under the right side of the heart. RIMA to circumflex over the right pericardium and over the aorta and pulmonary artery around the left side of the heart. RIMA to circumflex through the pericardium and along and under the right side of the heart. RIMA to circumflex through the pericardium and over the aorta and pulmonary artery around the left side of the heart. RIMA to circumflex through the pericardium and through the transverse sinus. RIMA to circumflex over the pericardium and through the transverse sinus. RIMA to RCA over the pericardium. RIMA to RCA through the pericardium. RIMA to DA over the pericardium. RIMA to DA through the pericardium. RIMA to DA through the pericardium and through the transverse sinus. RIMA to DA over the pericardium and through the transverse sinus. To determine the feasibility of various grafts, the measurements just listed were subtracted from the measured IMA lengths. All calculations were performed only on individual cadavers, but only the mean of the differences is expressed (Tables 4, 5). Results The IMA lengths to the fifth intercostal spaces were 17.4 cm and 17.5 cm on the left and right sides, respectively. Measured to the costal margins, the lengths were 22.1 cm pericardium sverse sinus Over pericardi u m transverse sinus I Fig 4. Measurements of three routes from origin of left internal mummay artery to circumflex artery. (* = shortest route.) wpericardium I *Over pericardium Fig 5. Measurements offour routes from origin of right internal mammary artery to left anterior descending coronary artery. (* = shortest route.)

424 VANDER SALM ET AL Ann Thorac Surg verse sinus *Thru \ ( pericardium Over pericardium ransverse sinus * d er pericardiurn Fig 6. Measurements of four routes from origin of right internal mammary artery to diagonal artery. (* = shortest route.) Fig 8. Measurements of two routes from origin of right internal mammary artery to right coronary artery. (* = shortest route.) and 22.2 cm for the left and right, respectively (see Table 1). The distances from the IMA origin to the coronary arteries, using the indicated different routes, are shown in Tables 2 and 3. For the LIMA to the LAD, the shortest route is through the pericardial incision; distances were 17.79 cm over the lung, 15.57 cm medial to the lung, and 14.14 cm through the pericardium. Each value differs significantly from each other ( p <.5). For the LIMA to the diagonal, the shortest route is also through the pericardial incision; distances were 18.41 cm over the lung, 15.71 cm medial to the lung, and 13.43 cm through the pericardial incision. Each of these values differs significantly from each other. For the LIMA to the circumflex, the shortest route is again through the pericardial incision; distances were 22.43 cm over the lung, 19.21 cm medial to the lung, and 16.21 cm through the pericardial incision. Each of these values differs significantly from each other. For the RIMA to the LAD, the shortest distance is the direct path over the pericardium to the artery; distances were 17.29 cm over the pericardium, 18.49 cm through the A pericardium Thru transverse sinus, Thru transverse sinus, thru pericardium Over great vessels, transverse sinus B Fig 7. (A) Measurements of four routes from origin of right internal mammary artery to circumflex artery. (B) Measurements of two routes anterior to the aorta from origin of right internal mamma y artery to circumflex artery. (" = shortest route.) Fig 9. Measurements of four routes from origin of right internal mammary artery to posterior descending coronary artery. (* = shortest route.)

Ann Thorac Surg 1989;47421-7 VANDER SALM ET AL 425 Table 4. Length of Left lnternal Mammary Artery Minus Distance to Coronary Arteries Route M Left Anterior Descending LIM A-ICS -.43 1.79 3.21 LIM A-CM 4.29 6.5 7.93 Diagonal LIMA-ICS -1.6 1.64 3.93 LIMA-CM 3.66 6.36 8.64 Circumflex LIMA-ICS -5.7-1.86 1.14 LIMA-CM -.36 2.86 5.86 a Data are shown as the mean value in centimeters. CM = costal margin; ICS = fifth intercostal space; LIMA = left internal mammary artery; M = medial to lung; = over lung; = through pericardium, medial to lung. right pericardial incision, 2.5 cm through the pericardium and the transverse sinus, and 21.86 cm over the pericardium and through the transverse sinus. The direct (anterior) path is significantly shorter than both routes through the transverse sinus but not significantly shorter than the anterior route through the pericardial incision. For the RIMA to the diagonal, the shortest distance is again the direct (anterior) path over the pericardium to the artery; distances were 16.7 cm over the pericardium, 18.1 cm through the pericardium, 18.71 cm through the pericardium and transverse sinus, and 19.5 cm over the pericardium and through the transverse sinus. The direct (anterior) path is significantly shorter than all the other routes. For the RIMA to the circumflex, the shortest distance is through the pericardium and transverse sinus. This route is significantly shorter than all other routes except the route anterior to the aorta and pulmonary artery and through the right pericardial incision ( p =.67). For the RIMA to the RCA, the slightly shorter distance is through the pericardial incision, 15.14 cm compared with 16.7 cm over the pericardium (p =.1). For the RIMA to the DA, the shortest route is through the pericardial incision and following the course of the native RCA; distances were 2.14 cm over the pericardium, 18. cm through the pericardium, 23.7 cm through the pericardium and transverse sinus, and 24.5 cm over the pericardium and through the transverse sinus. The shortest route is significantly so compared with all the other routes. We also compared the IMA lengths with distances to arteries. The measured distances to the arteries, subtracted from the measured lengths of the IMAs, are shown in Tables 4 and 5. When measured to the fifth intercostal space, the LIMA was always sufficiently long to reach the LAD or diagonal coronary artery provided that the shortest route (through the pericardium) was used. Even with this route, however, the LIMA was occasionally too short to reach the circumflex artery. Measured to the costal margin, the LIMA was always long enough to reach any of these three coronary arteries provided that the route through the pericardium was used. Using the RIMA measured to the costal margin and using the shortest route (over the pericardium and heart), the RIMA was always sufficiently long to reach the LAD or diagonal coronary artery. When measured to the fifth intercostal space, the RIMA was often too short to reach Table 5. Length of Right lnternal Mammary Artery Minus Distance to Coronary Arteries Route T OR OL R L Left Anterior Descending Diagonal Circumflex Right osterior Descending RIM A-ICS - RIM A-CM.21 4.93 1.43 6.14 1.43 6.14.2.64 2.7 a Data are shown as the mean value in centimeters. -.99-3. 3.73 1.71 -.6-1.21 4.11 3.5-1.29 3.43 2.36 7.7 -.5-5.57 4.21 -.86-4.36.36-2. 2.71-3.21 1.5-7. -2.29-7.14 3.57-6.29-2.88-2.43 1.14-1.57 2.5 CM = costal margin; ICS = fifth intercostal space; = over lung; OL = over pericardium and over heart to left side; OR = over pericardium and right side; = over pericardium, through transverse sinus; = through pericardium; L = through pericardium and over heart to left side; R = through pericardium and around right side of heart; T = through pericardium, through transverse sinus; RIMA = right internal mammary artery.

426 VANDER SALM ET AL Ann Thorac Surg either the LAD or the diagonal. Using any other than the shortest path, the RIMA to the costal margin length was often too short to reach the LAD or diagonal in individual cases, although the means of the differences were positive, thus indicating an average excess length of RIMA. For the circumflex artery, no combination of RIMA or route always gave a sufficient excess of RIMA length over distance to the coronary artery. However, the RIMA to the costal margin length routed through the pericardium and transverse sinus or routed through the pericardium and over the aorta and pulmonary artery was too short in only 1 patient each, and the mean of the differences was positive (RIMA more than long enough to reach) for both routes. Using the RIMA measured to the costal margin, the RIMA was always sufficiently long to reach the RCA, but did so with more artery to spare when routed through the pericardial incision. The only combination of RIMA that always reached the DA was RIMA to the costal margin length routed through the pericardial incision. Comment Compared with the saphenous vein, the IMA has a greater patency and produces a decreased long-term mortality when used as a coronary artery graft [l-51. In an effort to extend the use of the IMA, we evaluated the shortest course from IMA origin to the recipient coronary artery. All coronary arteries could be reached with one or both in situ IMAs. However, with only two arteries, regardless of methods to find the best route to the coronary artery, it is unlikely that more than two coronary beds can be revascularized with IMAs. Also, Tector (91 made the point that not all of the heart can be revascularized with IMA grafts, and suggested that they be used primarily to graft the left-sided vessels, thus leaving the right coronary arteries to be grafted with saphenous vein. His rationale for this allocation of IMAs is that the RIMA to the RCA or DA carries a higher occlusion rate than it does to the left system or than does the LIMA, in part (he speculates) because the right lung has a greater tendency to kink the RIMA than does the left lung to kink the LIMA. In the performance of the measurements for this study, it was obvious that any IMA crossing anterior to the heart to the opposite side had little or no risk of interference from the ipsilateral lung. However, any IMA coursing along the ipsilateral side of the heart can be in jeopardy from the lung on the same side with the danger of stretching the graft or kinking the anastomosis. By routing these grafts posterior and medial to the apex of the lung and then through the pericardium, one obtains the shortest course and the course with the least jeopardy from the adjacent lung. For revascularization of the left-sided coronary arteries with the LIMA, the route through the pericardium proves to be consistently and considerably shorter to the LAD, diagonal, and circumflex arteries than any other route. These results are confirmed by clinical studies by acific [15], Jones [7], and their colleagues. For the LAD and diagonal, and usually for the circumflex, the IMA transected at the fifth intercostal space will be sufficiently long to reach those coronary arteries. Sequential grafts require more length and hence require a more distal dissection of the IMA. It has been our clinical experience that the LIMA is usually of sufficiently large caliber below its bifurcation to be usable. To revascularize the LAD and diagonal with the RIMA, the shortest course, as would be expected, is the direct one over the heart. If transected at the level of the fifth intercostal space, the RIMA will often be too short; it should therefore be dissected down to and transected at the costal margin. As routing the RIMA across the anterior surface of the heart exposes it to considerable danger in coronary reoperations, the slightly longer course through a pericardial incision might be wiser, because it then allows for pericardial closure over the graft. If the RIMA is transected at the costal margin, it will almost always be sufficiently long to reach the circumflex branch arteries. Although passing it through the pericardium and through the transverse sinus proves to be the shortest course, the route over the aorta and pulmonary artery may be as good for proximal branches. For both the RCA and DA, the shortest course for the RIMA is through the pericardium. To consistently reach the DA, the RIMA must be divided near the costal arch. In the performance of the measurements for this study, several caveats must be raised. First, the measurements can be somewhat subjective. The lungs must be inflated to the same extent for each measurement. The same degree of tension must be kept in the measuring string. These variables were minimized by having the same individual (T. J.V.) make all measurements. Second, the conditions in a cadaver cannot reproduce those in a living patient. Cardiac fullness, for example, differs between life and death, and so does cardiac compressibility. Also, the heart and lungs of the cadaver lie quiescent. Third, the conclusions drawn from these measurements presume that the IMAs will be usable at the measured levels of transection (at the fifth intercostal space or costal arch). Finally, the decreased patency of RIMA grafts to the right coronary arterial system and the increased technical difficulty of peforming these anastomoses may create clinical reluctance to proceed, even though this study demonstrates that the RIMA should be sufficiently long. This study suggests other possible studies. The assessment of the coronary arteries accessible with free IMA grafts and the shortest route for them (or for saphenous vein grafts) could be made from a similar series of measurements originating at the aortic root. The feasible combinations of sequential IMA grafts could similarly be determined. The conclusions reached from this study are as follows: 1. Any coronary artery can be reached with an in situ IMA graft. 2. The shortest course to ipsilateral coronary arteries is medial to the lung and through the pericardium. 3. No conceivable frugality of IMA routing can allow the revascularization of more than two coronary beds; saphenous vein or another conduit such as gastroepi-

Ann Thorac Surg VANDER SALM ET AL 427 ploic artery still must be employed to revascularize the third. References 1. Olearchyk AS, Magovern GJ. Internal mammary artery grafting. Clinical results, patency rates, and long-term survival in 833 patients. J Thorac Cardiovasc Surg 1986;92:182. 2. Barner HB, Standeven JW, Reese J. Twelve-year experience with internal mammary artery for coronary artery bypass. J Thorac Cardiovasc Surg 1985;9:668. 3. Liddle HV, Gould BL, Jones D, et al. Conditional probability of multiple coronary graft failure. J Thorac Cardiovasc Surg 1984;87:526. 4. Singh RN, Sosa JA, Green GE. Long-term fate of the internal mammary artery and saphenous vein grafts. J Thorac Cardiovasc Surg 1983;86:359. 5. 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 1986;314:1. 6. Cosgrove DM, Loop FD. Techniques to maximize mammary artery length. Ann Thorac Surg 1985;4:78-9. 7. Jones EL, Lattouf, Lutz JF, King SB 111. Important anatomical and physiological considerations in performance of com- plex mammary-coronary artery operations. Ann Thorac Surg 1987;43:469-77. 8. McBride LR, Barner HB. The left internal mammary artery as a sequential graft to the left anterior descending system. J Thorac Cardiovasc Surg 1983;86:73. 9. Tector AJ. Fifteen years experience with the internal mammary artery graft. Ann Thorac Surg 1986;42(Suppl):S22-7. 1. Tector AJ, Schmahl TM, Canino VR. Expanding the use of the internal mammary artery to improve patency in coronary artery bypass grafting. J Thorac Cardiovasc Surg 1986;91:9. 11. Rankin JS, Newman GE, Bashore TM, et al. Clinical and angiographic assessment of complex mammary artery bypass grafting. J Thorac Cardiovasc Surg 1986;92:832. 12. Loop FD, Lytle BW, Cosgrove DM, et al. Free (aortacoronary) internal mammary artery graft. Late results. J Thorac Cardiovasc Surg 1986;92:827. 13. Galbut DL, Traad EA, Dorman MJ, et al. Twelve-year experience with bilateral internal mammary artery grafts. Ann Thorac Surg 1985;4:264-7. 14. Lytle BW, Cosgrove DM, Saltus GL, Taylor C, Loop FD. Multivessel coronary revascularization without saphenous vein: long-term results of bilateral internal mammary artey grafting. Ann Thorac Surg 1983;36:54-7. 15. acific AD, Sears NJ, Burgos C. Harvesting, routing, and anastomosing the left internal mammary artery graft. Ann Thorac Surg 1986;42:78-1.