Postoperative Flow Characteristics of Left Internal Thoracic Artery Grafts

Transcription

1 Postoperative Flow Characteristics of Left nternal Thoracic rtery Grafts Michihiro Nasu, MD, Takashi kasaka, MD, Tsuyoshi Okazaki, MD, Masahiko Shinkai, MD, Hiroshi Fujiwara, MD, Jun Sono, MD, Yukikatsu Okada, MD, Satoru Miyamoto, MD, Sunao Nishiuchi, MD, Junichi Yoshikawa, MD, and Toyo Shomura, MD Departments of Thoracic and Cardiovascular Surgery and Cardiology, Kobe City General Hospital Kobe, Japan Twenty patients whose left internal thoracic artery (LT) was anastomosed to the left anterior descending artery (LD) underwent postoperative coronary angiography and Doppler ultrasound velocimetry. During angiography, the diameter of the LT conduit was measured at three points: proximal, mid, and distal. The degree of left anterior descending artery stenosis proximal to the anastomotic site was evaluated by densitometry. The LT flow velocity pattern was obtained at the three points to calculate the total, systolic, and diastolic flow volume. There were significant differences in the total LT flow among the three points (proximal, ml/min; mid, ml/min; distal, ml/min; p < between the proximal and the mid or distal portions). The degree of left anterior descending artery stenosis affected the distal LT flow and diameter (r = and 0.811, respectively). There were significant differences in the systolic LT flow among the three points (proximal, 13.2 _ 6.5 ml/min; mid, 8.1 ~- 4.7 ml/min; distal, 5.6 ± 3.4 ml/min; p < between the proximal and the mid or distal portions). However, there was no statistically significant difference in the diastolic LT flow among the three points (proximal, ml/min; mid, ml/min; distal, ml/min). We conclude that a lower degree of LD stenosis significantly reduces the LT flow, inducing the string phenomenon. dditionally, during the diastolic phase, the LT graft transports the blood primarily to the coronary artery but not to the side branches. Therefore, the steal phenomenon might not apply in the setting of an LT graft. (nn Thorac Surg 1995;59:154-62) he many clinical capabilities of the left internal tho- T racic artery (LT) as graft material have been established by many surgeons. However, some problems with its use still remain. lthough the biggest advantage of an LT graft is its long-term patency [1], flow reserve is limited during exercise despit:e the relief of angina [2]. The string phenomenon may occur when an LT graft is anastomosed to the left anterior descending artery (LD) with a low-grade proximal stenosis [3-6]. dditionally, the size of the LT*& graft depends on LT flow [2]. However, these observations are mainly based on angiographic findings and there are no clinical reports concerning the relationship of the size of the LT graft or the degree of LD stenosis with the actual flow of the LT graft. The steal phenomenon was observed in patients who received an LT graft that included a large remnant of the lateral costal branch [7-11]. When the steal phenomenon is generally discussed, the areas involved are in either the systemic or coronary circulation. On the contrary, an LT graft perfuses both the systemic and coronary circulations. However, there is a big difference in the perfusion pattern between the coronary and ccepted for publication July 21, ddress reprint requests to Dr Nasu, Department of Thoracic and Cardiovascular Surgery, Kobe City General Hospital, 4-6, Minatojima- Nakamachi, Chuo-ku, Kobe, Hyogo, 650, Japan. systemic circulations. lthough the systemic circulation is mainly perfused during the systolic phase, the coronary circulation is primarily perfused during the diastolic phase. Therefore, it seems that the same mechanism cannot account for the steal phenomenon affecting an LT graft. The present study focused on the following two issues: (1) whether proximal LD stenosis affects on the LT graft flow volume, and (2) whether side branch flow compromises coronary perfusion from an LT graft. Material and Methods Patients Twenty patients (19 male, 1 female) underwent coronary angiography and the Doppler ultrasound velocimetry after they had undergone coronary artery bypass grafting using an LT graft. Their age at examination ranged from 35 to 72 years (mean, 53.2 ± 11.1 years). Other patient characteristics are given in Table 1. The follow-up examination was performed from to 56 months after the grafting procedure (average, months). n 14 patients, angiography was performed before discharge within month postoperatively to confirm the success of the operative results. The remaining 6 patients were evaluated because they had shown an abnormal response during stress testing 10 to 56 months after the 1995 by The Society of Thoracic Surgeons /95/$ (94)

2 nn Thorac Surg NSU ET L ;59: LT GRFT FLOW CHRCTERSTCS Table 1. Patient Characteristics Characteristics Value No. of patients 20 Sex (male/female) 19/1 ge (y) (35-72) Body surface area (m 2) _ 0.15 ( ) Mean no. of bypass grafts 2.9 _ Two 8 Three 6 Four 4 Left ventricular end ( ) diastolic volume index (ml/m 2) Left ventricular end-systolic volume index (ml/m 2) ( ) Left ventricular ejection 0.46 _ ( ) fraction operation. Left ventriculography revealed normal motion of the anterior wall in 11 patients, hypokinesis in 5, and akinesis in 4. Coronary angiography showed no notable stenosis at the LT anastomotic site and various degrees of stenosis in the LD proximal to the anastomotic site. n 5 patients, 50% to 75% stenosis was observed; in 4, 76% to 90% stenosis; in 5, 91% to 99% stenosis; and, in 6, 100% stenosis. n addition, those patients in whom LT angiography revealed the presence of the string phenomenon were not included in this study because the Doppler guidewire interfered with LT flow in such patients, which made accurate assessment of flow impossible. ngiographic Measurements The postoperative diameter of the LT graft was quantitatively determined at three points: proximal, mid, and distal. The diameter was determined with a digital image-analyzing system using the catheter diameter as a reference. Some side branches, such as the pericardiophrenic and thymic branches, were distributed between the proximal and mid portion of the LT graft. The LD stenosis proximal to the anastomotic site was also evaluated with the same system, and expressed as the percentage of the diameter obstructed by stenosis. Flow Velocity Measurements in the LT Graft Details on the instrumentation used have been given in previous articles [12, 13]. ll flow velocity measurements were performed with a inch, 12-MHz, Doppler flow guidewire (Cardiometrics). The transducer has an estimated sample volume size of 2.5 mm and a diameter at range rate depth of 5 mm. The pulsed-doppler ultrasound velocimeter (Flowmap; Cardiometrics) consisted of a real-time spectral analysis system. The Doppler system can compute a variety of on-line spectral variables. fter angiographic study of the LT graft, a 5F catheter was positioned in the origin of the LT from the left subclavian artery. The Doppler guidewire was advanced through the catheter and into the LT graft and intro- duced to the anastomotic site. Thereafter, the guiding catheter was drawn out from the origin of the subclavian artery to avoid disturbing LT flow. Flow velocity in the LT graft was measured at the same three points where the diameter of the LT graft was measured. The Doppler system can compute a variety of online spectral variables, including the time-averaged spectral peak velocity and the systolic and diastolic time velocity integral. No patients suffered any complications related to the conduit cannulation or the Doppler guidewire instrumentation. Quantification of Flow Volume in the Left nternal Thoracic rtery quantitative estimate of LT flow volume was calculated from Doppler velocity data using the following equation: Q = (vrd2/4) (PV/2), where Q is the flow volume, D is the LT diameter, and PV is the time-averaged peak velocity [12]. n addition, systolic and diastolic LT flows were calculated using the following equations: Qs = Q. S/(S + D) and Qd = Q. D/(S + D), where Qs is the systolic LT flow, Q is the total LT flow, S is the systolic time velocity integral, D is the diastolic time velocity integral, and Qd is the diastolic LT flow. The total systolic, and diastolic LT flows were estimated at the three points. Statistical nalysis Simple regression analysis was used to evaluate the LT diameter and the total LT flow in relation to the severity of the proximal LD stenosis. One-way analysis of variance was used to estimate the total, systolic, and diastolic LT flow in relation to the sampling position; and repeated measures analysis of variance was used to compare the values at the three points. Between-group differences were compared by Scheff6's test, and a p value of less than 0.05 was considered statistically significant. ll data were expressed as the mean _+ standard deviation. Results Flow Velocity Patterns in the Left nternal Thoracic rtery Graft n a patient whose LD was totally occluded, typical biphasic velocity waveforms were consistently obtained along the length of the LT conduit, and the predominance of the diastolic component over the systolic one at the distal LT segment was significantly greater than that at the proximal segment (Fig 1). n a patient with less LD stenosis, a similar transformation was observed from the proximal segment to the mid segment of the

3 156 NSU ET L nn Thorac Surg LT GRFT FLOW CHRCTERSTCS 1995;59: Fig 1. Flow velocity patterns in a left internal thoracic artery graft at each position in a patient with 100% stenosis of the left anterior descending artery. Flow velocity patterns are consistently biphasic along the length of the graft. The predominance of the diastolic component over the systolic one is greater at the distal segment than at the proximal one. mid distal LT conduit (Fig 2). However, the systolic retrograde flow from the LD to the LT graft was noted at the distal part of the graft. This finding of retrograde flow corresponded to the back flow observed when a radiopaque substance was injected into the LD during angiography. Left nternal Thoracic rtery Diameter and Flow The proximal LT diameter averaged 2.5 ~ mm; the mid, mm; and the distal, 2.1 ± 0.39 ram. There was a statistically significant difference (p < 0.001) among the diameters of the three portions (Fig 3). Similarly, the proximal LT flow was ml/min; the mid, 29.9 ± 15.2 ml/min; and the distal, ml/min. There were statistically significant differences in the LT flow between the proximal portion and the mid (p < 0.01) or distal (p < 0.001) portion (Fig 3B). However, there was no statistically significant difference in the LT flow between the mid and distal portions. These data suggest that the side branches were distributed between the proximal and the mid portions of the LT conduit. Relationship Between the Degree of Left nterior Descending rtery Stenosis and the Diameter and Flow of the Left nternal Thoracic rtery The degree of stenosis significantly affected the distal diameter as well as the distal flow (r and 0.823, respectively) (Fig 4). These quantitative data suggest that the less the LD stenosis, the less the LT flow. Systolic and Diastolic Left nternal Thoracic rtery Flow at Each Position There was no significant difference in the diastolic flow among the three points (proximal, 22.9 ~ 11.0 ml/min; mid, ml/min; distal, 21.6 ± 10.8 ml/min) (Fig 5). Therefore, from this we can conclude that the diastolic LT flow is primarily drained into the coronary artery, but not into the side branches. Conversely, there Fig 2. Flow velocity patterns in a left internal thoracic artery graft at each position in a patient with 50% stenosis of the left anterior descending artery. Biphasic waveforms are seen at the proximal and mid segments of the left internal thoracic artery graft. t the distal segment, systolic reverse flow is observed. proximal mid distal

4 nn Thorac Surg NSU ET L ;59: LT GRFT FLOW CHRCTERSTCS O E mm < - 1 -J O<0.001 p<o.ool p<o.ool = i i 0.01), although the statistical power was much weaker. ngiography did not show the existence of any branches between the mid and distal portions. However, Daly and associates [14] speculated on the basis of the findings from their experiment that the vasa vasorum of LT grafts with an intact surrounding tissue pedicle is perfused by vessels in this tissue pedicle, which is perfused by small branches from the LT itself. Therefore, the flow difference between the mid and distal segments may result from the diversion of LT flow into its vasa vasorum. However, this interpretation is made with caution because of the limited accuracy of the flow calculation. ml/min 6o] J m B 10 0 px mid dis p<o.o01 p<0.01 n.s. px mid dis Fig 3. () Diameter of left internal thoracic artery CLT) graft at each position. There is a significant difference in the diameter among the three points (p < between the proximal and mid segments, between the proximal and distal segments, and between the mid and distal segments). (B) Flow volume of the left internal thoracic artery graft at each position. There is a significant difference among the three points (p < between the proximal and distal segments, and p < 0.01 between the proximal and mid segments). (dis = distab n.s. - not significant; px - proximal.) were significant differences in the systolic flow among the three points (proximal, 13.2 _+ 6.5 ml/min; mid, ml/min; distal, 5.6 _+ 3.4 ml/min; p < between the proximal and mid or distal portions) (Fig 5B). From this we can conclude that the side branches are perfused mainly during the systolic phase. small amount of systolic LT flow is drained into the coronary circulation. There was a significant difference in the systolic LT flow between the mid and distal segments (p Comment The LT has been widely used as the most reliable graft material because of its long-term patency [15, 16]. When anastomosed to the LD, an LT graft is associated with mm ~5 2.0 "~ 1.5 ~o ml/min ee r = % LD %stenosis r = , ~ ~ * * ** B e 0.r ~_ LD %stenosis % Fig 4. () Correlation of the degree of stenosis of the left anterior descending artery CLD % stenosis) with the distal diameter of the left internal thoracic artery (LT). The degree of stenosis significantly affected the distal diameter of the left internal thoracic artery graft (r ). (B) Correlation of the degree of stenosis of the left anterior descending artery with the distal flow volume of the left internal thoracic artery graft. The degree of stenosis signif@antly affected the ftow in the graft (r ).

5 158 NSU ET L nn Thorac Surg LT GRFT FLOW CHRCTERSTCS 1995;59: ml/min 5O 40 < k- "i 30._u o 20._~ 10 0 ml/min 60 v,,., -J._.9. 5O 40 3o 20 m 10 n.s. n.s. n.s. px mid dis )< <0.001 p<0.01 px mid dis Fig 5. () The diastolic flow volume at the three points in the left internal thoracic artery (LT) graft. There was no statistical significance among the three points. Therefore, the diastolic blood flow in the graft is primarily perfusing the coronary system. (B) The systolic flow volume at the three points in the left internal thoracic artery graft. There was a significant difference between the flow in the proximal and mid segments (p < 0.001), between the proximal and distal segments (p < 0.001), and between the mid and distal segments (p < 0.01). (dis - distal; n.s. = not significant; px ~ proximal.) a better long-term survival rate and angina-free rate compared with those associated with the use of a saphenous vein graft [1]. However, the string sign and steal phenomenon are two anomalous flow characteristics that can affect LT grafts. t was reported that the string sign occurred when an LT graft was anastomosed to an LD with low-grade proximal stenosis [3-6]. Kitamura and colleagues [17] showed angiographically that an LT graft is patent even when the string sign with no flow occurs, so-called no-flow patency. Seki and associates [3] contended that the string phenomenon could be regarded as a physiologic change reflecting the LT graft's response to the blood flow demands. ll of these views were based on angiographic findings. n our study, we calculated the LT flow from the LT size, determined during angiography, and the LT flow velocity profile, obtained using the Doppler guidewire, to evaluate quantitatively the relationship of the proximal LD stenosis to LT flow. Our data showed that the LT flow volume decreased proportionate to the decrease in the grade of the LD stenosis. n the patients with a less than 70% LD stenosis, the LT flow was less than 20 ml/min and the systolic reverse flow was detected. Theoretically, LT flow should be zero in patients with a less than 40% LD stenosis. n the patients with a completely obstructed LD, the LT flow rates were distributed widely because the perfusion areas and myocardial function status in our patients varied. f we could study a population of patients who had not suffered myocardial infarction and had only a single LD obstruction or stenosis at a similar segment, we would be able to obtain a more linear correlation between the extent of proximal LD stenosis and LT flow volume. However, it would be impossible to collect such a group of patients. Nevertheless, the p value is sufficiently high in our patients to indicate a linear correlation between the proximal LD stenosis and the LT flow volume. Recently, in a long-term experimental study conducted by Lust and colleagues [18], it was found that, even after 2 months of chronic flow competition from a fully patent native artery, there was still a recruitable flow reserve for the LT graft when the native vessels were occluded. n a patient of Kitamura and associates [17], the anatomic patency of a nonfunctioning LT graft was demonstrated when the native artery was occluded by a percutaneous transluminal coronary angioplasty balloon catheter. We encountered an interesting case that also shed light on this matter (Figs 6, 7). One month after operation in a patient of ours, LT arteriography revealed the presence of the string phenomenon together with backflow of the radiopaque substance into the LD. One year after operation, the LT conduit was observed to be functioning despite a totally occluded proximal LD, and the flow velocity pattern of the LT graft was normally biphasic. These findings indicate that a string sign represents a dynamic response to a sustained period of flow competition. However, basic research has shown that a direct product of the velocity of flowing blood, shear stress, is implicated as a cause of the vascular remodeling that occurs in native vessels [19]. Therefore, it remains open to question how long the anatomic patency and flow recruitability can be maintained in a nonfunctioning LT graft. s a surgical caution, it was reported that the side branches of the LT should be trimmed off [7-10]. On the other hand, findings from experimental studies have suggested that hypoperfusion to the phrenic nerve induces the phrenic nerve palsy that occurs after the pericardiophrenic artery is trimmed off [20]. Schmid [7],

6 nn Thorac Surg NSU ET L ;59: LT GRFT FLOW CHRCTERSTCS Fig 6. () ngiogram of the left anterior descending artery showing the string phenomenon 1 month after operation. t this time, an angiogram of the left anterior descending artery revealed good flow with backflow into the graft. (B) functioning left internal thoracic artery graft was observed 1 year after operation, with total occlusion of the proximal left anterior descending artery. B Wolfenden [8], and Pelias [9] and their associates reported observing the steal phenomenon produced by a large remnant side branch that perfused the systemic circulation. Wolfenden and colleagues [8] were not able to provide direct clinical evidence of the LT steal in their patients, however, because of LT injury that occurred after vein grafting to the LD at repeat coronary artery bypass grafting. n Pelias and Del Rossi's case, the patient had no symptoms for 8 years postoperatively and the lesion causing the recurrent angina was obscure because of the presence of vein graft occlusion and progression of disease. Schmid's group reported that their patient's postoperative angina was cured through successive embolization of the side branches. However, as much as we look at the LT angiographic findings after embolization, there seems to be anastomotic stenosis, possibly as much as 50%. dditionally, Seki and associates [3] reported that the remnant side branches do not appear to affect the LT size, and vert and colleagues [4] concluded that unligated side branches do not interfere with long-term graft patency. Therefore, the role of remnant side branches as a source of the steal phenomenon may be exaggerated. However, Singh and co-workers [11] reported on 4 patients with a remnant pericardiophrenic artery that had collaterals which drained into the pulmonary circulation. This kind of remnant artery might produce LT steal. Because pulmonary vascular resistance must be much less than the systemic peripheral resistance, it might be comparable to diastolic coronary resistance. n our patients, the diastolic LT blood flow was found to drain primarily into the coronary artery and the side branches were perfused mainly during the systolic phase. Certainly, some fraction of the systolic LT blood flow is drained into the coronary artery. This fraction is distributed to the epicardial coronary artery, but not to the intramuscular artery,

7 160 NSU ET L nn Thorac Surg LT GRFT FLOW CHRCTERSTCS 1995;59: B Fig 7. () Flow velocity pattern of a nonfunctioning left internal thoracic artery (LT) graft 1 month after operation. The waveform of the nonfunctioning graft is similar to that of the systemic artery. (B) Flow velocity pattern of a functioning left internal thoracic artery graft 1 year after operation. The waveform is biphasic and the diastolic component is highly predominant over the systolic one. because the intramuscular coronary resistance during the systolic phase is much higher than the epicardial one. Therefore, the diastolic steal phenomenon, which is due to diversion of the blood flow into remnant side branches, is not a realistic explanation for the flow pattern in the LT conduit observed in the resting state. Our study has three limitations. First, the changes in the LT diameter during a cardiac cycle are ignored. t the present time, there is no way to obtain a real-time measurement of LT diameters at the three points during a cardiac cycle. ngiography did not reveal any changes in the LT size. The second drawback to our study was that the LT flow velocity was only measured in the resting state. To investigate the steal phenomenon, we must perform the studies not only at rest but also during exercise. The third limitation is that the flow calculation does not yield entirely accurate measurements. n addition, the patients in our study did not have large branches, but it is unlikely that the diastolic coronary vascular resistance is greater than the diastolic systemic one under any circumstances. Therefore, we believe that the flow pattern in the LT graft does not change in any situation. Flemma and colleagues [21] observed LT flow to range from 10 to 65 ml/min and average 43 ml/min intraoperatively, whereas Barner [22] observed it to average 56 ml/min. Both investigators used an electromagnetic flow probe. n our study, the distal LT flow averaged 34 ml/min in 12 patients with a proximal LD stenosis that exceeded 90%. Therefore, the calculated LT flows in our patients are slightly lower than previously reported values. Doucette and associates [13] validated the accuracy of the flow measurement achieved using the Doppler guide wire method and quantitative angiography, both under in vivo and in vitro conditions. Their in vitro study showed that Doppler-derived flow, calculated using known tube diameters and estimating the mean velocity as 0.5 times the time-averaged peak velocity, was nearly identical to the flow measured by the electromagnetic flowmeter. t also showed that the tortuosity of the tube consistently caused the flow rate to be underestimated. Therefore, we positioned the guidewire tip at the straightly coursing segments of the LT grafts. n addition, their in vivo study revealed that the Dopplerderived native coronary artery flow rate was comparable to the value determined by electromagnetic flowmeter approximately 85% of the time. This difference is attributed to the fact that the spatial flow pattern in a straightly coursing artery should be parabolic. Therefore, our LT flow rates could be slightly lower than the real LT flow rates. Few articles about the LT flow velocity pattern have been published. Bach's group [23] found a consistent biphasic flow velocity pattern along the length of LT conduits and a gradual transition in the velocity pattern from a systemic to coronary artery flow velocity pattern. They attributed this finding to a particular intrinsic property of the internal mammary artery wall, possibly its high elastic tissue content. Certainly, van Son and associates [24] showed the tissue that makes up the proximal and distal segments is elastomuscular and that of the mid segment is elastic. n Guo-Wei's pharmacologic study [25], the reactivity of the distal LT section was found to be inversely correlated with the diameter of

8 nn Thorac Surg NSU ET L ;59: LT GRFT FLOW CHRCTERSTCS the artery. The findings from these studies suggest that the vascular compliance of the LT varies along its length. This variability might account for the gradual transition in the LT flow velocity pattern. However, as any surgeon knows, postoperatively the LT is tightly surrounded by hard connective tissue, and this might cause a highly compliant vessel to be less compliant. Bach and colleagues [23] missed the LT flow diversion to side branches and the coronary artery. We believe that LT flow diversion is a primary cause of the gradual transition in the flow velocity pattern along the length of the LT conduit. n summary, the presence of an LD stenosis proximal to the anastomotic site significantly affects the LT flow volume, and this validates the observations made previously on the basis of angiographic findings. dditionally, an LT graft transports the blood primarily to the coronary artery during the diastolic phase and the steal phenomenon might not apply in the setting of an LT graft. References 1. Boylan MJ, Lytle BM, Loop FD, et al. Surgical treatment of isolated left anterior descending coronary stenosis. Comparison of left internal mammary artery and venous autograft at 18 to 20 years of follow-up. J Thorac Cardiovasc Surg 1994;107: Kawasuji M, Tedoriya T, Takemuro H, Sakakibara N, Taki J, Watanabe Y. Flow capacities of arterial grafts for coronary artery bypass grafting. nn Thorac Surg 1993;56: Seki T, Kitamura S, Kawachi K, et al. quantitative study of postoperative luminal narrowing of the internal thoracic artery graft in coronary artery bypass surgery. J Thorac Cardiovasc Surg 1992;104: vert T, Huttunen K, Landou C, Bjork VO. ngiographic studies of internal mammary artery grafts 11 years after coronary artery bypass grafting. J Thorac Cardiovasc Surg 1988;96: Dincer B, Barner HD. The "occluded" internal mammary artery graft: restoration of patency after apparent occlusion associated with progression of coronary disease. J Thorac Cardiovasc Surg 1983;85: rice, Borras X, Ramino J. Patency of internal mammary grafts in no-flow situations. J Thorac Cardiovasc Surg 1987; 93: Schmid C, Heublein B, Reichelt S, Borst HG. Steal phenomenon caused by a parallel branch of the internal mammary artery. nn Thorac Surg 1990;50: Wolfenden HD, Newman DC. voidance of steal phenomena by thorough internal mammary artery dissection. J Thorac Cardiovasc Surg 1992;103: Pelias, Del Rossi. case of postoperative internal mammary steal. J Thorac Cardiovasc Surg 1985;90: Singh RN, Magovern GJ. nternal mammary graft: improved flow resulting from correction of steal phenomenon. J Thorac Cardiovasc Surg 1982;84: Singh RN, Pittsburgh P, Sosa J. nternal mammary arterycoronary artery anastomosis. nfluence of the side branches on surgical result. J Thorac Cardiovasc Surg 1981;82: Segal J, Kern MJ, Scott N, et al. lterations of phasic coronary artery flow velocity in humans during percutaneous coronary angioplasty. J m Coll Cardiol 1992;20: Docerre JW, Corl PD, Payne HM, et al. Validation of a Doppler guide wire for intravascular measurement of coronary artery flow velocity. Circulation 1992;85: Daly RC, McCarthy PM, Orzulak T, Schaff HV, Edwards WD. Histologic comparison of experimental coronary artery bypass grafts. Similarity of in situ and free internal mammary artery grafts. J Thorac Cardiovasc Surg 1988;96: Grondin CM, Campeau L, Lesperance J, Enjalbert M, Bourassa MG. Comparison of late changes in internal mammary artery and saphenous vein grafts in two consecutive series of patients 10 years after operation. Circulation 1984; 70(Suppl 1): cinapura J, Rose DM, Jacobowitz J, et al. nternal mammary artery bypass grafting: influence on recurrent angina and survival in 2100 patients. nn Thorac Surg 1989;48: Kitamura S, Kawachi K, Seki T, Sawabata N, Morita R, Kawata T. ngiographic demonstration of no-fow anatomical patency of internal thoracic-coronary artery bypass grafts. nn Thorac Surg 1992;53: Lust RM, Zeri RS, Spence P, et al. Effects of chronic native flow competition on internal thoracic artery grafts. nn Thorac Surg 1994;57: Glagov S, Zarins C, Giddens DP, Ku DN. Hemodynamics and atherosclerosis. nsight and perspectives gained from studies of human arteries. rch Pathol Lab Med 1988;112: O'Brien JW, Johnson SH, VanSteyn SJ, et al. Effects of internal mammary artery dissection on phrenic nerve perfusion and function. nn Thorac Surg 1991;52: Flemma RJ, Singh HM, Tector J, Lepley D Jr, Frazier BL. Comparative hemodynamic properties of vein and mammary artery in coronary bypass operations. nn Thorac Surg 1975;20: Barner HB. Blood flow in the internal mammary artery. m Heart J 1973;86: Bach RG, Kern MJ, Donohue TJ, guirre FV, Caracciolo E. Comparison of phasic blood flow velocity characteristics of arterial and venous coronary artery bypass conduits. Circulation 1993;88(Suppl 2): Van Son JM, Smedts F, de Wilde PCM, et al. Histological study of the internal mammary artery with emphasis on its suitability as a coronary artery bypass graft. nn Thorac Surg 1993;55: He GW. Contractility of the human internal mammary artery at the distal section increases toward the end. Emphasis on not using the end of the internal mammary artery for grafting. J Thorac Cardiovasc Surg 1993;106: NVTED COMMENTRY nternal thoracic artery (T) flow inadequacy, or the "'hypoperfusion syndrome," has received increasing attention as the number of patients who initially were revascularized with vein grafts exclusively are presenting for reoperation, and the vein graft is being replaced with the more preferred T conduit. Fourteen of the 20

9 162 NSU ET L nn Thorac Surg LT GRFT FLOW CHRCTERSTCS 1995;59: patients included in this study were asymptomatic and examined within 1 month of operation, solely to verify the technical success of the operation. Therefore, this study provides information not likely to be available from patient studies performed in the United States. The major points of this article are (1) that steal from undivided side branches of the proximal T on total graft flow is overestimated, and the influence of left anterior descending artery stenosis is underestimated, and (2) that the degree of proximal native coronary artery stenosis predicts the development of a "string sign." Regarding the latter, any conclusion having to do with the string sign should be disregarded. Nasu and associates did not demonstrate the string sign in any of the patients in their series, and any conclusion to that effect is speculative at best. They describe an additional case in the Comments section in which this sign was observed, but, in this patient, the T graft appeared to induce progression of the lesion in the left anterior descending artery, instead of the reverse, which argues against the initial premise. Finally, no data are provided on the native anatomic progression of the T diameters to make it possible to determine whether the observations made postoperatively are flow induced or anatomic. Nasu and associates assume that, because the vessels were restudied within 1 month of the operation, anatomic vascular remodeling could not have occurred and any changes must be flow dependent, but that assumption is not valid. Steal only has been reported as an important factor responsible for limiting T graft flow when any remaining intact proximal branches are large. Nasu and associates are correct in saying that in previous studies suggesting flow diversion through remaining proximal branches, the steal effect may have been overestimated and the effect of the native proximal lesion may have been underestimated. However, in all the patients in their series, only very small proximal branches with limited flow were left intact, so the presence of a steal, if in fact it was present, probably would not have been evident in these studies anyway. There are no data to indicate the relationship between the distal left T stenosis[flow and the steal fraction in each vessel. Therefore, although the extent of the lesion in the native coronary artery may have a more critical effect on the total T flow than proximal flow steal, the findings from this study hardly demonstrate that fact conclusively. variability in distal perfusion requirements, or the placement of additional grafts to partially occluded left anterior descending artery diagonal segments (10 patients received three or more grafts) could explain the differences in the graft flow requirements. Therefore, physiologic revision of the graft may as easily be the result of distal perfusion as of proximal stenoses. The validity of intraluminal Doppler echocardiography has been shown in several systems, and theoretically should not have been a problem in assessing the T flow, but in these patients the catheter system may have been obstructive nonetheless, thus causing flow to be underestimated. For example, in the 12 patients with stenosis of the left anterior descending artery exceeding 90%, the T flow averaged only 34 ml/min, despite the relatively proximal insertion of the graft, and flow patterns to segments with persistent hypokinesia or akinesia after grafting (n = 5) did not differ from those associated with normal wall motion (n = 4). This also suggests that flow may have been attenuated generally. n summary, Nasu and associates should be commended for their attempts to answer pertinent questions in a difficult setting, but the readers also should be advised to interpret the results cautiously, as this report raises more questions than it answers. Robert M. Lust, PhD Division of Cardiothoracic Surgery East Carolina University School of Medicine Brody 4S-22 Greenville, NC