Author + information
- Received June 17, 1997
- Revision received December 18, 1997
- Accepted January 19, 1998
- Published online April 1, 1998.
- Takashi Akasaka, MDA,* (, )
- Kiyoshi Yoshida, MD, FACCA,
- Takeshi Hozumi, MDA,
- Tsutomu Takagi, MDA,
- Shuichiro Kaji, MDA,
- Takahiro Kawamoto, MDA,
- Shigefumi Morioka, MDA,
- Michihiro Nasu, MDA and
- Junichi Yoshikawa, MD, FACCB
- ↵*Dr. Takashi Akasaka, Department of Cardiology, Kobe General Hospital, Minatojima-nakamachi 4-6, Chuo-ku, Kobe 650, Japan.
Objectives. This study sought to assess the flow dynamics of internal mammary artery grafts (IMAGs) in no-flow situations by use of a Doppler guide wire.
Background. Functionally no-flow and anatomically patent IMAGs have been reported by angiography in patients with a patent recipient coronary artery.
Methods. The study included 12 patients with an IMAG to the left anterior descending coronary artery (LAD) in whom no-flow patency of the graft was suspected angiographically. Thirteen patients with a normally functioning IMAG whose LAD was occluded in the proximal portion and was supplied only from the graft served as control patients. Phasic flow velocities were recorded in the distal portion of the graft and the recipient LAD using a 0.014-in., 15-MHz Doppler guide wire at rest and during hyperemia (0.14-mg/kg body weight per min intravenous adenosine infusion).
Results. There were no significant differences in systolic (15 ± 3 vs. 19 ± 6 cm/s, p = NS), diastolic (35 ± 11 vs. 37 ± 7 cm/s, p = NS) and time-averaged peak velocities at rest (20 ± 5 vs. 21 ± 5 cm/s, p = NS), during hyperemia (51 ± 12 vs. 54 ± 8 cm/s, p = NS) and in coronary flow velocity reserve (2.8 ± 0.9 vs. 2.7 ± 0.3, NS) in the native LAD in patients with a no-flow patent graft versus control patients. Within the graft, to and fro signals with systolic reversal and diastolic anterograde flow were seen in the no-flow patent grafts, although anterograde flow signals were recorded in systole and diastole in control patients. Systolic (−28 ± 19 vs. 22 ± 9 cm/s, p < 0.01), diastolic (18 ± 17 vs. 44 ± 14 cm/s, p < 0.01) and time-averaged (−2 ± 6 vs. 26 ± 9 cm/s, p < 0.01) peak velocities at rest were significantly smaller in the no-flow patent grafts than in control grafts. During hyperemia, anterograde flow became predominant, with a reduction in retrograde systolic flow signal and an increase in diastolic flow velocity and time-averaged peak velocity in the no-flow patent grafts, and no-flow situations disappeared temporarily.
Conclusions. Functionally no-flow situations of IMAGs manifesting to and fro signals with systolic flow reversal and diastolic antegrade low flow velocity are temporary conditions in certain hemodynamic circumstances, and these grafts function as conduits during hyperemic states.
Patency of internal mammary artery grafts (IMAGs) in no-flow situations has been demonstrated by angiography in patients whose recipient left anterior descending coronary artery (LAD) was patent [1–3]. Flow competition between the patent recipient LAD and the IMAG has been proposed as the cause of the no-flow condition of the IMAG [1–3]. However, the arterial graft would thrombose if there was actually no flow because blood stagnation has been reported as a cause of thrombus formation . Thus, some flow might be expected, to protect from thrombus formation and occlusion in the angiographically demonstrated no-flow but anatomically patent IMAG. In no-flow patent IMAGs, furthermore, antegrade flow has been demonstrated temporarily during recipient coronary artery occlusion by an angioplasty balloon , and a similar flow condition could be expected during an increase in blood flow demand that is more than the blood supply passing through the proximal stenosis of the native coronary artery. However, no reports have examined no-flow patent IMAGs in detail using techniques other than angiographic demonstration [1–3].
A recently developed Doppler guide wire, which can be advanced across stenotic coronary artery lesions [5–7], has been used for the assessment of flow dynamics of IMAGs and saphenous vein grafts (SVGs) [8, 9]. Different phasic flow patterns at rest and flow capacity during hyperemia have been demonstrated between the two grafts. Using this method, flow dynamics in no-flow patent IMAGs would be easily analyzed at rest and during hyperemia, without occlusion of the recipient LAD by an angioplasty balloon as reported previously .
The purpose of this study was to assess the flow characteristics of IMAGs in angiographically demonstrated no-flow situations by using a Doppler guide wire.
1.1 Study Patients
From April 1992 to December 1996, 289 patients who had previously undergone elective coronary artery bypass grafting to the LAD using a left IMAG were referred for evaluation of the graft patency to our department. Of these patients, 197 underwent coronary angiography 1 month after surgery in order to certify the successive operative results, and the remaining 92 underwent coronary angiography 1 year after operation for an annual check-up after bypass surgery. Of these patients, 20 were excluded from the study because of graft occlusion; 29 were also excluded because of significant diameter stenosis (>50%) of the graft at the site of anastomosis to the recipient LAD and 24 by the significant stenosis (>50%) of the recipient LAD distal to the graft anastomosis. Of the remaining 236 patients, 12 (5 patients 1 month and 7 patients 1 year after operation) showed no-flow patency of the graft by angiography, which was defined as interruption of contrast medium at the mid to distal portion of the graft and no inflow of the contrast into the LAD from the graft, although inflow of contrast medium from the recipient coronary artery up to the mid to proximal portion of the graft was observed during native coronary angiography, as reported previously [1–3]. The age of these patients ranged from 41 to 68 years with a mean of 57 ± 7. None showed collateral vessels from the LAD to the other coronary arteries or from the other vessels to the LAD. All had one to three additional SVGs to other coronary arteries, and the mean number of coronary artery bypass grafts was 2.6 ± 0.7 per patient. Two of them had a prior history of inferior myocardial infarction. Thirteen (4 patients 1 month and 9 patients 1 year after operation) with a normally functioning IMAG by angiography, in whom the recipient LAD was occluded at the proximal portion and was supplied only from the graft, and who underwent the subsequent Doppler study, served as control patients. The age of the control patients ranged from 37 to 67 years with a mean 53 ± 10. They also had one to three additional SVGs to other coronary arteries with a mean number of bypass grafts of 2.4 ± 0.6 per patient. Two of the controls also had a prior history of inferior myocardial infarction. The other patients were excluded from the study because of a prior history or electrocardiographic findings of anterior myocardial infarction, patency of the recipient coronary artery at the proximal site or refusal of the subsequent Doppler study. Patients with collateral vessels from the LAD to the other coronary arteries or from the other vessels to the LAD by angiography were also excluded from the study.
1.2 Cardiac Catheterization and Angiography
All medications, including beta-blocking agents, calcium-channel blocking agents and nitrates, were terminated at least 24 hours before cardiac catheterization. After sedation with 5 mg of diazepam administered orally, patients were taken to the catheterization laboratory. Any drugs likely to affect coronary hemodynamics were not used during the catheterization procedure before selective coronary angiography.
Cardiac catheterizations were performed by the femoral approach after local anesthesia with 0.5% lidocaine. The left ventricle was approached in a retrograde manner. Left and right heart pressure data were recorded using a fluid-filled catheter-transducer system, and cardiac output was measured by the thermodilution method. Biplane left ventriculography was performed to assess left ventricular (LV) wall motion and to measure LV volume by the area–length method. Selective coronary angiography was carried out by the Judkins’ technique after intravenous injection of 3 mg of isosorbide dinitrate. To assess the degree of the recipient LAD stenosis, and to measure the diameters of the distal IMAG and the native LAD at the position corresponding to the tip of the Doppler guide wire where flow velocity was recorded, coronary angiography was analyzed quantitatively by videodensitometric analysis using a commercially available system (CMS, Medical Imaging Systems, Inc.) according to previous reports [10, 11].
1.3 Coronary Flow Velocity Recordings
Phasic flow velocities were recorded in the proximal, mid and distal portion of the IMAG to the LAD and the recipient LAD distal to the graft anastomosis using a 0.014 in., 15-MHz Doppler guide wire (FloWire, Cardiometrics, Inc.), and a velocimeter (FloMap, Cardiometrics, Inc.) [5–9]. The pulse repetition frequency of the Doppler flowmeter was variable from 12 to 96 KHz within the velocity range selected.
The Doppler guide wire was advanced into the recipient LAD via the IMAG through a 5F coronary angiography catheter (Selecon, Clinical Supply, Inc.) using a technique similar to that of guide wire manipulation during percutaneous transluminal coronary angioplasty. The tip of the Doppler guide wire was advanced from the proximal to the distal portion of the bypass graft and then into the native LAD slightly distal to the graft anastomosis. An optimal Doppler signal was obtained in each portion by moving the guide wire slightly within the vessel lumen and adjusting the range gate control. The final position of the Doppler guide wire was confirmed by contrast injection. During the Doppler study, a 12-lead surface electrocardiogram and pressure waveform at the tip of the guiding catheter were monitored continuously.
Frequency analysis of the Doppler signals was carried out in real time by fast-Fourier transform using a velocimeter (FloMap, Cardiometrics, Inc.) [5–9]. Five minutes after contrast injection, Doppler signals were recorded on videotape and by a videoprinter along with an electrocardiogram and aortic pressure tracing. After Doppler signals under baseline conditions were recorded in each portion of the IMAG and the distal LAD, Doppler signals during hyperemic conditions were recorded in the recipient LAD and then in the distal portion of the bypass graft after 0.14 mg/kg body weight per min intravenous adenosine administration. Peak velocities in systole and diastole and the time average of the instantaneous spectral peak velocity (time-averaged peak velocity [APV]) during one cardiac cycle were obtained from the phasic coronary blood flow velocity recordings using off-line computerized planimetry. Coronary flow velocity reserve was obtained from the ratio of maximal to baseline APV, and flow volume was calculated from the APV and the diameter of the vessel at the point where flow was measured according to the proposals of Doucette et al., reported previously . Measurements were averaged over 5 beats.
1.4 Statistical Analysis
All data are expressed as mean values ± SD. Statistical analysis between no-flow patent graft and normally functioning graft was performed with an unpaired two-tailed ttesting. A p value of <0.05 was considered significant.
2.1 Clinical Characteristics and Angiographic Data
There were no significant differences between patients with a no-flow patent graft and control patients in age, time after bypass surgery, the total number of grafts and incidence of previous inferior myocardial infarction as indicated in Table 1. There were also no significant differences in the catheterization and angiographic data between these two groups, including heart rate, cardiac index, end-diastolic and end-systolic LV volume indexes, LV ejection fraction, LV end-diastolic pressure, mean pulmonary capillary wedge pressure and mean aortic pressure as shown in Table 1. Left ventriculography showed no asynergy in the anteroapical portion of the left ventricle, the territory of blood supply from the LAD in all patients.
The percent diameter stenosis of the proximal lesion of the LAD was significantly smaller in patients with a no-flow patent IMAG than in control patients (57 ± 5 vs. 100%, p < 0.01) (Table 1). Although the diameter of the distal LAD where phasic flow velocity was measured was not significantly different between the two groups, the diameter of the distal portion of the IMAG was significantly smaller in patients with a no-flow patent graft than in control patients (1.9 ± 0.2 vs. 2.5 ± 0.2 mm, p < 0.01) (Table 1).
2.2 Phasic Flow Velocity Data
In the recipient native LAD, peak velocities in systole and diastole, APV and calculated flow volume at baseline were not significantly different between no-flow patent grafts and normally functioning grafts (Fig. 1, Table 2). During hyperemia, APV was not different between the two groups, and as a result, coronary flow velocity reserve was also not different between the two groups in the recipient LAD distal to the anastomosis.
In the IMAG in control patients (Fig. 1, Table 2), predominant systolic peak and lower diastolic peak velocities were demonstrated in the proximal portion, and a gradual transition to native LAD flow velocity with increasing diastolic velocity and decreasing systolic velocity was observed from the mid to distal portion. Finally, systolic and diastolic antegrade flow velocity signals of a similar pattern to those in the recipient LAD were demonstrated in the distal IMAG. There were no significant differences in systolic (19 ± 6 vs. 22 ± 9 cm/s, p = NS) and diastolic (37 ± 7 vs. 44 ± 14 cm/s, p = NS) peak velocities at rest and a diastolic/systolic peak velocity ratio (2.2 ± 0.6 vs. 2.2 ± 0.6, p = NS) between the distal graft and the recipient LAD in control patients. Furthermore, APVs at rest (21 ± 5 vs. 26 ± 9 cm/s, p = NS) and during hyperemia (54 ± 8 vs. 46 ± 15 cm/s, p = NS) and, as a result, flow velocity reserve (2.6 ± 0.3 vs. 2.7 ± 0.3 cm/s, p = NS), were not significantly different between the distal graft and the recipient coronary artery in normally functioning IMAGs. In no-flow patent grafts (Fig. 1, Table 2), although predominant systolic peak and lower diastolic peak velocities were also demonstrated in the proximal portion, the systolic flow velocity decreased gradually to negative from the mid to distal portion while maintaining a similar diastolic velocity, and finally, to and fro signals with systolic reversal and diastolic anterograde flow were observed in the distal portion of the bypass grafts at baseline. As a result, in the distal portion of the graft, APV of the no-flow patent grafts became nearly zero (−2 ± 6 cm/s), and systolic and diastolic peak velocities (−28 ± 18 vs. 22 ± 9 cm/s, 18 ± 17 vs. 44 ± 14 cm/s, p < 0.01, respectively) and APV (−2 ± 6 vs. 26 ± 9 cm, p < 0.01, respectively) were significantly smaller in the no-flow patent grafts than in the normally functioning grafts. Furthermore, during hyperemia, in the distal portion of the graft, systolic flow reversal became diminished or disappeared and anterograde flow signals became predominant in patients with a no-flow patent graft (Fig. 2, Table 2), although an increase in both systolic and diastolic flow signals similar to those in the native coronary artery was observed in control patients. APV during hyperemia in the no-flow patent grafts was also significantly smaller than that in the normally functioning grafts (26 ± 12 vs. 46 ± 15 cm/s, p < 0.01).
The present study demonstrated to and fro signals with systolic reversal and diastolic anterograde flow in the distal portion of angiographically demonstrated no-flow patent IMAGs at baseline rest conditions by a Doppler guide wire. These no-flow patent grafts proved to be functioning in certain hemodynamic conditions with evidence of disappearance of retrograde flow and predominance of anterograde flow during hyperemia. This might be the first description of flow dynamics of angiographically no-flow patent IMAGs at rest and during hyperemia. This study also demonstrated the limitations of angiography for assessing patency of IMAGs and the superiority of flow velocity analysis by a Doppler guide wire for interpreting data of cardiac catheterization.
3.1 Hemodynamic Variables in LAD Distal to the Graft Anastomosis
The clinical characteristics and cardiac catheterization data of the patients in both no-flow patent and normally functioning IMAGs were not significantly different in the present study. Under similar hemodynamic conditions, the oxygen demand of the left ventricle should be similar for each patient group, because the determinants of myocardial oxygen demand are thought to be principally afterload, contractility and heart rate [12–14]. Coronary driving pressure and impedance have been proposed as physiologic factors governing coronary flow [14, 15]. Coronary capacitance and resistance, which might be the most important factors of coronary impedance [14, 15], could be similar in the LAD distal to the anastomosis in both patient groups if the geometry of the LAD was similar in the both groups, although there would obviously be some differences in coronary artery geometry among subjects. Heart rate and preload have been also reported to be factors that may influence coronary flow velocity measurements and flow reserve, whereas mean aortic pressure has been reported to not affect them . In this study, there were no significant differences in heart rate, LV end-diastolic pressure or mean pulmonary capillary wedge pressure, which might be indexes of LV preload and mean aortic pressure, between the two groups. Thus, similar coronary hemodynamic conditions might be expected in the two different groups, and the blood flow in the LAD distal to the anastomosis might be maintained similarly whether the flow was supplied from the graft or from the proximal native coronary artery. At baseline and during hyperemia there were no significant differences in the flow data for the LAD distal to the anastomosis between no-flow patent grafts and normally functioning grafts in this study.
3.2 Flow Characteristics in Normally Functioning and No-Flow Patent IMAGs
As previously reported, in the normally functioning IMAGs , although the mechanism of this transition in flow velocity pattern is still under investigation, a gradual longitudinal transition in the flow velocity pattern from a predominantly systolic velocity proximally to a predominantly diastolic velocity distally was also demonstrated in the present study, and phasic flow velocities in the distal portion of the normally functioning grafts at rest demonstrated systolic and diastolic antegrade flow signals similar to those in the native LAD [8, 9]. The difference in each parameter between the present study and the previous study might be related to the differences in the patient groups and the precise site of velocity measurement. Furthermore, maximal APV and calculated flow velocity reserve in the graft in this study were similar to previous descriptions [8, 9]. Also, calculated flow volumes in the graft were not significantly different from the previous report, although the sampling portion of the flow velocity might be slightly distal compared with the previous report . Similar flow volumes would be expected in different sampling sites if there were no big side branches and no blood loss in the graft. Furthermore, in controls, calculated flow volumes were greater in the graft than in the native LAD. This discrepancy of flow volume might be related to blood supply from the IMAG to the native LAD proximal to the graft anastomosis.
In no-flow patent grafts, a gradual longitudinal transition in the systolic flow velocity from antegrade proximally to retrograde distally was demonstrated, with lower antegrade diastolic velocity. To and fro signals with systolic reversal and diastolic antegrade flow were also observed in the distal portion of the bypass grafts at baseline in this study. This is the first description of flow dynamics at rest in IMAGs in angiographically demonstrated no-flow situations, and the to and fro situation of the flow was thought to be a mechanism of functionally no-flow conditions. In the native coronary artery, it has been reported that baseline flow remains normal until the coronary artery is narrowed by 80% to 85% diameter stenosis [17–19]. Therefore, the baseline flow of the native LAD distal to the graft anastomosis in the present study might have been sufficiently supplied from the proximal native coronary artery through the stenosis at baseline conditions because the diameter stenosis of the native coronary artery was 57% ± 5%, and no-flow was expected from the bypass graft to the recipient native coronary artery at rest in this situation. Because the observed to and fro IMAG flow might be the result of flow competition between the native LAD and the graft, the time difference between the flow passing through the proximal native coronary artery and the flow through the proximal to midportion of the bypass graft may influence the flow of the distal portion of the graft. This interpretation of the mechanism of no-flow situations would suggest that, in patients with more severely stenosed but patent proximal LADs, predominant antegrade flow with smaller or absent retrograde flow may be observed in the distal IMAG in relation to the hemodynamic severity of proximal LAD stenosis. Pressure recordings of the distal portion of the graft might give us some information to resolve this mechanism precisely, and flow velocity analysis in patients with IMAGs with various degrees of stenosis in the proximal LAD might give us important information.
As discussed earlier, although the graft should constitute the blood supply to the distal LAD in the normally functioning graft, the blood flow of the LAD distal to the graft anastomosis would mainly be supplied by the proximal native coronary artery at rest and no-flow situations of the graft would be expected in the no-flow patent grafts. The diameter of the graft should be smaller in cases with no-flow or low flow situations because the flow to the recipient LAD distal to the anastomosis should mainly come not from the graft but from the proximal native coronary artery. The diameter of the graft in the distal portion where the Doppler flow velocity was recorded was significantly smaller in patients with a no-flow patent graft than in normally functioning grafts in this present study, as previously reported [1–3].
Superior long-term patency of IMAGs to that of SVGs has been well established in coronary artery bypass graft surgery [20–22]. Excellent long-term patency would be expected if the bypass graft could be patent in no-flow or low flow conditions. The ability to keep patency in no-flow situations in IMAGs might be related to their excellent long-term patency. At this point, long-term follow-up of no-flow patent IMAGs or a comparative study of flow measurement between IMAGs and SVGs in no-flow or low flow situations would be useful.
3.3 Flow Dynamics during Maximal Hyperemia in the IMAGs
The maximal flow velocity and flow velocity reserve were measured at the distal portion of the bypass graft after a 0.14 mg/kg per min intravenous adenosine infusion. In control patients, flow velocity reserve measured in the distal portion of the graft was not significantly different from that in the native coronary artery because all blood flow to the distal native coronary artery was through the graft in this present study. This was also not significantly different from our previous report . In no-flow patent grafts, during hyperemia, systolic flow reversal became diminished or disappeared and antegrade flow signals became predominant and no-flow situations disappeared in the distal portion of the grafts. During hyperemia, blood flow demand in the native coronary artery distal to the graft anastomosis should be greater than the blood supply passing through the proximal stenosis of the native coronary artery, and the flow from the bypass graft would cover this insufficiency of blood supply from the proximal native coronary artery. Because it has been reported that coronary flow reserve begins to decrease at 40% to 50% diameter stenosis for a vasodilatory stimulus [17, 18]and the diameter stenosis of the proximal native coronary artery was 57 ± 5% in the present study, antegrade flow from the bypass graft to the recipient native coronary artery should be expected during hyperemia. Although a similar phenomenon was reported previously by angiography in association with progression of the proximal native coronary artery and during balloon angioplasty of the proximal native coronary artery , the present study may be the first report to demonstrate functioning of no-flow patent grafts as conduits during a short period of hemodynamic changes without intervention. To keep anatomic patency of the graft in no-flow situations, an increase of anterograde flow of the graft by increasing blood flow demand in the distal portion of the recipient LAD (by exercise and so on) might be important after bypass surgery.
3.4 Study Limitations
Some limitations to this study must be considered. 1) Although the angiographic finding of a no-flow situation was defined as previously reported [1–3], such a finding would be closely related to the force of the contrast injection. However, functionally no-flow conditions at rest were also confirmed by findings of to and fro signals with systolic reversal and diastolic anterograde flow by a Doppler guide wire, and these findings would indicate that an angiographic finding of a no-flow situation might rarely be related to the force of the contrast injection in this study. 2) Blood flow velocity in the proximal native coronary artery was not analyzed in the present study. Calculated flow volume at rest and during hyperemia and flow reserve in the proximal native LAD might be more suggestive in analyzing the mechanism of no-flow patency of the graft. However, it might be technically difficult to record flow velocity in the proximal LAD in patients with a normally functioning graft because of guide wire manipulations. 3) Flow volume for the LAD and bypass grafts was calculated using the average peak velocity of the phasic flow velocity recording and the vessel cross-sectional area at the point where the flow velocity was recorded according to the previous report by Doucette et al. . As described earlier in the discussion, bypass graft flow rates in the present study were similar to those in previous reports [9, 15]. Although it is not certain that the same formula could be applicable in the IMAG because the flow profile in IMAGs might be different from that in the native coronary artery , the flow reserve result as a ratio of the maximal to the baseline flow would not be affected. 4) Angiography at peak adenosine hyperemia was not performed in the present study. We speculate that anterograde flow would be demonstrated by angiography at peak adenosine hyperemia because flow velocity measurements might demonstrate flow dynamics of IMAGs more precisely compared with angiography and anterograde flow was demonstrated during hyperemia by Doppler guide wire. 5) The number of patients in the study was small. Further study of large numbers of patients should be done to confirm the mechanism of the no-flow patent graft.
To and fro signals with systolic reversal and diastolic anterograde flow were demonstrated in the distal portion of angiographically demonstrated no-flow patent IMAGs at baseline resting condition by a Doppler guide wire. This no-flow patent graft was proved to be functioning in certain hemodynamic conditions by the disappearance of retrograde flow and the predominance of anterograde flow during hyperemia. Thus, no-flow situations of IMAGs are temporary conditions in some hemodynamic circumstances, and these grafts function as conduits in other hemodynamic states.
We gratefully acknowledge the radiologic technicians and the nurses in our catheter laboratory for their assistance in the recording of coronary flow velocities in this study.
- time average of the instantaneous spectral peak velocity (time-averaged peak velocity)
- internal mammary artery graft
- left anterior descending coronary artery
- left ventricular
- saphenous vein graft
- Received June 17, 1997.
- Revision received December 18, 1997.
- Accepted January 19, 1998.
- The American College of Cardiology
- Segal J,
- Kern MJ,
- Scott NA,
- et al.
- Doucette JW,
- Corl PD,
- Payne HM,
- et al.
- Bach RG,
- Kern MJ,
- Donohue TJ,
- Aguirre FV,
- Caracciolo EA
- Akasaka T,
- Yoshikawa J,
- Yoshida K,
- et al.
- Parmley WW,
- Tyberg JV
- Klocke FJ,
- Ellis AK
- McGinn AL,
- White CW,
- Wilson RF
- Klocke FJ
- Fitzgibbon GM,
- Kafka HP,
- Leach AJ,
- Keon WJ,
- Hooper GD,
- Burton JR
- Barner HB,
- Standeven JW,
- Reese J
- Cameron A,
- Kemp HG,
- Green GE
- Fujiwara T,
- Kajiya F,
- Kanazawa S,
- et al.