| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Cardiovascular Surgery, Gifu Prefectural Tajimi Hospital, Tajimi, Gifu
b Department of Cardiothoracic Surgery, Nagoya University Graduate School of Medicine, Showa-ku, Nagoya, Aichi, Japan
Accepted for publication July 12, 2007.
* Address correspondence to Dr Tokuda, Department of Cardiovascular Surgery, Gifu Prefectural Tajimi Hospital, 5-161 Maehata, Tajimi, Gifu, 507-8522, Japan (Email: tokuda{at}mxb.mesh.ne.jp).
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentReferences |
|---|
Methods: We analyzed a total of 261 grafts that were evaluated by intraoperative transit time flow measurement and underwent early postoperative coronary angiography within 3 months of surgery. Based on the control angiography, failing grafts were defined as occluded or patent grafts with greater than 50% stenosis or poor flow characteristics. Normal and failing graft indicators were compared according to the graft territories.
Results: According to the receiver operating characteristic curve analysis for the grafts to left coronary arteries, a mean flow of 15 mL/min or less, a pulsatility index of 5.1 or higher, and a backward flow of 4.1% or higher were found to be the optimal cutoff criteria to predict early graft failure. Similarly, for the grafts to right coronary arteries, the cutoff values were 20 mL/min, 4.7, and 4.6%, respectively. A systolic dominant flow curve pattern was a risk factor only in grafts to the left coronary arteries. Negative predictive values of these cutoff criteria ranged from 0.91 to 0.96, whereas positive predictive values ranged from 0.31 to 0.80.
Conclusions: Using these criteria, transit time flow measurement may be a useful method to predict early graft failure. However, surgeons should be aware of the low positive predictive values to avoid unnecessary graft revision.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
The approval of this retrospective study was obtained on the March 15, 2007, from the Ethics Committee of the Departments of Cardiovascular Surgery and Cardiology of Gifu Prefectural Tajimi Hospital. The Committee waived individual consent for the study.
Operative Procedures and Bypass Grafts
Coronary artery bypass graft surgery was performed in the standard manner, either with cardiopulmonary bypass (n = 85 [including 4 beating on-pump cases]) or without (n = 39). Combined procedures included mitral valve plasty (n = 9), ascending aortic replacement (n = 1), mitral valve replacement (n = 1), Dor procedure (n = 2), aortic valve replacement (n = 2), and tricuspid annuloplasty (n = 2).
Sequential anastomoses, and other anastomoses distal to the sequential anastomoses, were excluded from the analysis as the graft flow through the proximal sequential and distal anastomoses could affect each other. Also, the interpretation of the graft flow pattern was difficult and complicated. After excluding sequential anastomoses and their distal anastomoses, 261 grafts in 123 patients, evaluated by both intraoperative TTFM and early postoperative angiography, were submitted for analysis. These included 92 venous grafts and 169 arterial grafts. The type of conduits and the target territories of the grafts are shown in Table 1. The left or right internal thoracic artery, saphenous vein graft, radial artery, and gastroepiploic artery were used as the conduits.
|
The following values were obtained by TTFM analysis using the flowmeter: mean flow calculated across five cardiac cycles (Qmean), maximum flow recorded in one cardiac cycle (Qmax), minimum flow recorded through one cardiac cycle (Qmin), pulsatility index (PI) as the ratio between the difference (Qmax – Qmin/Qmean), and the percentage of backward flow (%BF) as the percentage of the flow through the graft directed backward across the anastomotic site (area below zero) compared with the total forward flow (area above zero) of the same cardiac cycle. The Qmin is a negative number if there is backward flow. The flow curve pattern was classified according to the dominancy based on the maximal flow value: systolic dominant when peak systolic flow exceeded peak diastolic flow by 10%; diastolic dominant or balanced when peak systolic flow did not exceed peak diastolic flow by 10% [5].
Postoperative Angiography
The postoperative angiography was performed 16.2 ± 12.6 days after surgery. Visual assessment of the angiography was made by two or more cardiologists and the results were classified according to the following system: (1) normal widely patent, less than 50% stenosis at any location in the graft, proximal anastomosis, distal anastomosis, or immediate 1 cm of target vessel, and normal Thrombolysis In Myocardial Infarction III (TIMI III) flow characteristics; (2) abnormal patent, with greater than 50% stenosis at any location in the graft, proximal or distal anastomosis, or immediate 1 cm of target vessel, or poor flow characteristics (non–TIMI III flow); or (3) occluded. Based on the angiography results, grafts were divided into two groups: group A (normal), and group B (failing grafts—abnormal or occluded) [4].
Statistical Analysis
Comparisons between the two groups were performed using unpaired t tests for continuous variables or the 
2 test or Fisher’s exact test for categorical variables. Univariate logistic regression was used to obtain odds ratios (and their 95% confidence limits) of early graft failure. Optimal cutoff values of Qmean, PI, and %BF, to predict early graft failure, were determined by means of the ROC curve analysis. The area under the curve, sensitivity, specificity, positive predictive value, and negative predictive value were also calculated. The optimal cutoff value was defined as providing maximal sensitivity and specificity. Comparisons among the three groups were performed using one-way analysis of variance. The Tukey-Kramer test was used as a post-hoc test. Statistical analysis was performed using JMP 6 software (SAS, Cary, North Carolina).
As differences have been reported in the normal flow patterns of grafts between the right coronary territories and the left coronary territories [5], TTFMs were analyzed according to the location of grafts, namely, the right and left coronary artery territories. The right coronary artery (RCA) territories included the right coronary artery and its branches. The left coronary artery (LCA) territories included the left anterior descending artery, diagonal branches, and the left circumflex artery, and its branches.
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
The target territories and the type of conduits of the failing grafts are shown in Table 1. There were no statistically significant differences in the incidence of graft failure among the different types of conduits (internal thoracic artery, radial artery, gastroepiploic artery, and saphenous vein graft), or between arterial and venous conduits and internal thoracic artery and other conduits. There were also no significant differences in the incidence of the failure between the grafts to RCA and LCA, and among the graft territories shown in Table1 (left anterior descending artery, diagonal branches, left circumflex artery, and RCA). Overall, within the 225 normal grafts, the Qmean was 45.5 ± 28.9 mL/min, the PI was 2.74 ± 2.27, and the %BF was 1.98 ± 4.16. In the 22 abnormal grafts, the values were 29.6 ± 20.8 mL/min, 4.21 ± 3.07, and 6.74 ± 10.2, respectively. In the 14 occluded grafts, Qmean was 15.1 ± 21.0 mL/min, the PI was 18.3 ± 28.9, and the %BF was 21.1 ± 28.7. According to the Tukey-Kramer test, Qmean, PI, and %BF differed between the normal and abnormal grafts, and also differed between the normal and occluded grafts. There was, however, no difference in these three values when abnormal and occluded grafts were compared.
During surgery, 6 grafts were revised after the initial TTFM (as described in Methods). The initial TTFM was not included in the analysis because the detailed data were not stored in the flowmeter system. However, according to surgical records, all of these initial measurements revealed a low Qmean (<10 mL/min), a high %BF (>5%), and a high PI (>10). Of the 6 revised grafts, anastomotic problems were found in 3 (1 proximal and 2 distal anastomoses), graft kinking was found in 1, and graft stretch was found in 1. These problems were subsequently corrected. In 1 graft, a revision was performed but no apparent problem was found. In this case, the TTFM did not improve, even after redoing the anastomosis. This gastroepiploic artery graft was found to be occluded in the postoperative angiography, and the target vessel was very small. The graft requiring kinking repair was found to be stenotic at the repair site (abnormal) and the other 4 were found to be normal in the angiography.
Comparison of TTFM Values Between the Two Groups
A comparison of TTFM values between the two groups (A and B) is shown in Table 2. Group A had a higher Qmean, lower PI, and lower %BF than group B, in both the RCA and LCA territories. Based on the univariate analysis, a low Qmean, lower Qmax, higher PI, and higher %BF were shown to be risk factors for predicting failing grafts in both territories. Odds ratios and their 95% confidence intervals are also shown. A systolic dominant flow curve pattern was shown to be a risk factor for graft failure in the LCA territories but not in the RCA territories. In fact, 56.7% of the grafts to the RCA had a systolic dominant flow curve pattern, significantly higher than the grafts to the LCA (p < 0.001). Although Qmax was higher in the grafts to the RCA (p = 0.015), Qmean, PI, and %BF were similar between the RCA and the LCA.
|
|
|
Although a lower Qmax in the LCA grafts was also associated with graft failure in the univariate analysis, we did not define the cutoff value of Qmax for predicting graft failure, as a high Qmax is reportedly not always a positive indicator [5]. A high Qmax can be due to rich graft flow, which generally indicates a patent graft, but occasionally it can be due to the spiked appearance of the TTFM curve. If the anastomosis is severely stenotic, the flow curve may appear spiky with a high maximal systolic velocity [5]. Thus, we did not perform an ROC analysis to obtain a cutoff value for Qmax.
Multiple Abnormal Measurements for Predicting Graft Failure
As the PPV was low for each abnormal criterion, we also introduced a criterion whereby the combination of multiple abnormal measurements was used to obtain a higher predictive value for detecting graft failure. For this criterion, when all the measurements listed in Table 3 were abnormal (according to the cutoff values), the test was regarded as positive. When we applied this criterion to the LCA, using the four abnormal measurements (a Qmean of 15 mL/min or smaller, a PI of 5.1 or higher, a %BF of 4.1% or higher, and a systolic dominant flow curve pattern), there were 11 positive and 190 negative results. Of these, there were 4 false positives and 19 false negatives. Therefore, the PPV was 0.64 and the NPV was 0.90. The 4 false positive grafts included 2 radial artery grafts and 2 internal thoracic artery grafts. Graft spasm was not noted in these grafts. In the case of the RCA, when using the criterion with the three abnormal measurements (Qmean of 20 mL/min or smaller, a PI of 4.7 or higher, and a %BF of 4.6 or higher), there were 7 positive and 53 negative results. Of these, there was 1 false positive and 4 false negatives. The PPV was 0.86 and the NPV was 0.93. The graft with the false positive result was a gastroepiploic artery graft to the RCA. This graft was particularly spastic, according to surgical observation.
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
The clear cutoff values derived by the ROC analysis are easy to use during surgery, and therefore very useful. There are, however, limitations of the TTFM. It has been reported that TTFM flow morphology is indistinguishable in anastomoses between mild to moderately severe stenosis (<75%) [11]. Less than critical stenoses cannot be detected by TTFM, as no modifications of the hemodynamic indicators of the grafts occur at this level, resulting in false negative test results. Also, even with abnormal TTFMs, grafts can, in some instances, be completely normal in the angiography [4]. Although the mechanism of that is not clear, it can cause false positive results. Transient graft spasm of arterial grafts may cause false positive results, as observed in this study. The existence of these false negative and false positive cases makes the percentage of true positives lower, since the prevalence of actual graft failure is low. That resulted in especially low sensitivity and lower PPV in this case. Even with optimizing the sensitivity and specificity of the ROC analysis, the sensitivities of the criteria, particularly in the graft to the LCA, were low and less than 0.5, consistent with previous reports [4].
Sensitivity is designed to ascertain the performance of the test; however, practically, it is not very helpful in determining the presence of graft failure. In this case, the PPV is more useful. The PPV is the percentage of observed failed grafts in those exhibiting positive test results. We found the PPV to be low, particularly for grafts to the LCA. That the relatively high NPV and low PPV observed in this study indicate a normal TTFM result is reassuring although a positive TTFM result may be falsely positive. Therefore, even if a surgeon sees an abnormal TTFM value, there is strong chance (1-PPV) that there is no anastomotic problem in the graft. Thus, TFFM values alone may prompt unnecessary revision of a graft. Balacumaraswami colleagues [12] similarly reported that graft patency assessment with TTFM alone may prompt unnecessary graft revision. They compared TTFM results with intraoperative fluorescence imaging and found that, in 3.8% of grafts, TTFMs can indicate poor flow despite satisfactory flow reflected by the intraoperative fluorescence imaging. They reported the 3.8% incidence of potential unnecessary revision to be greater than the 3% of necessary revision.
Therefore, abnormal TTFM values do not necessarily indicate immediate graft revision, particularly in the case of grafts to the LCA, and surgeons should exercise caution when interpreting abnormal TTFMs. Multiple TTFMs should be checked with repeated measurements before making the decision to undergo revision of anastomosis. Fortunately, when applying the criterion of using a combination of abnormal measurements (positive if all the values in Table 3 are abnormal), the PPV was reasonably high (0.64 for LCA and 0.86 for RCA territories). Thus, if all of the measurements (four for LCA and three for RCA) are abnormal, the graft should, in all likelihood, be revised. In addition, surgeons could evaluate information such as confidence in the anastomosis performed, the outside shape of the anastomosis, and the quality of target vessels or conduit. Other options suggested previously include repeated measurements with temporary occlusion of the stealing branch, or increasing blood pressure with inotropic agents [9]. To avoid unnecessary graft revision, the decision to undergo revision should be based on multiple TTFMs, as well as additional information (above), and not a single abnormal TTFM.
Limitations of the Study
There are several limitations to this study. First, as a retrospective study, the existence of bias could not be excluded. Second, we did not evaluate whether TTFM can better predict occluded grafts or abnormal grafts. We used both abnormal and occluded grafts as an endpoint and, as the numbers of occluded grafts were small (8 grafts to the LCA territories and 6 grafts to the RCA territories), we did not use occluded grafts alone as an endpoint. With their small numbers, we could not perform an ROC curve analysis to obtain cutoff values, or use their NPV and PPV to specifically predict graft occlusion. Similarly, previous studies have used both abnormal and occluded grafts as an endpoint [4, 5, 9]. We found no differences in the Qmean, PI, and %BF between the abnormal and occluded grafts. We therefore thought that it was appropriate to use both abnormal and occluded grafts as an endpoint of graft failure. Future studies will require a larger number of occluded grafts to assess the ability to predict them. Third, as discussed, the PPV of each meaurement included a wide range (0.31 to 0.8), indicating the inability of TTFM to confirm graft failure with a high degree of certainty. Although the criterion of using multiple abnormal measurements was helpful, the potential for false positives in the LCA graft (1-PPV = 0.36) still remained, and that should be recognized as a limitation of TTFM. If available, intraoperative indocyanine green angiography is a very good alternative as an intraoperative assessment. It has been reported to have better sensitivity and specificity than TTFM [4]. Moreover, it can provide information on the anatomic characteristics of the graft, allowing surgeons to identify the location of the problem [4]. Because of its high cost, however, this emerging technology is not as widely used as TTFM, and it is also not available in our institute. Fourth, we could not define why the ROC curve analysis appeared better for the RCA grafts than the LCA grafts. Because the LCA is compressed by the left ventricular contraction, factors such as the contractility or systolic pressure may affect the TTFM flow curve pattern in the LCA grafts, and consequently may affect the accuracy of the TTFM.
In conclusion, we have established specific cutoff values of criteria in TFFM for predicting early graft failure. A single abnormal value is indicative, but not always specific, for detecting graft failure, as the positive predictive value of each criterion is low. To avoid unnecessary graft revision, surgeons should exercise caution when interpreting abnormal TTFM results.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Related Article
This article has been cited by other articles:
|
|
 |
|
  Y. Takami, K. Tajima, S. Terazawa, N. Okada, K. Fujii, and Y. Sakai Transit-time flow characteristics of in situ right gastroepiploic arterial grafts in coronary artery bypass grafting J. Thorac. Cardiovasc. Surg., September 1, 2009; 138(3): 669 - 673. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Azmoun, R. Ramadan, N. Al-Attar, and R. Nottin Reply Ann. Thorac. Surg., April 1, 2009; 87(4): 1325 - 1326. [Full Text] [PDF]
|
|
|
|
|
|
F. Onorati, G. Santarpino, and A. Renzulli Reply Ann. Thorac. Surg., September 1, 2008; 86(3): 1053 - 1054. [Full Text] [PDF]
|
|
|
|
|
|
Y. Tokuda, M.-H. Song, H. Oshima, A. Usui, and Y. Ueda Predicting Midterm Coronary Artery Bypass Graft Failure by Intraoperative Transit Time Flow Measurement Ann. Thorac. Surg., August 1, 2008; 86(2): 532 - 536. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Herman, J. A. Sullivan, K. Buth, and J.-F. Legare Intraoperative graft flow measurements during coronary artery bypass surgery predict in-hospital outcomes Interactive CardioVascular and Thoracic Surgery, August 1, 2008; 7(4): 582 - 585. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Balacumaraswami Invited commentary Ann. Thorac. Surg., December 1, 2007; 84(6): 1934 - 1934. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
