Eur Radiol (2005) 15: 1994–1999DOI 10.1007/s00330-005-2766-1 CARDIAC

J. DorgeloT. P. WillemsP. M. A. van OoijenG. F. V. PandayP. W. BoonstraF. ZijlstraM. Oudkerk

Received: 25 June 2004Revised: 15 March 2005Accepted: 29 March 2005Published online: 20 May 2005# Springer-Verlag 2005

A 16-slice multidetector computed tomographyprotocol for evaluation of the gastroepiploicartery grafts in patients after coronary arterybypass surgery

Abstract Arterial coronary bypassgrafts [internal mammary arteries andgastroepiploic artery (GEA)] are inwidespread use for coronary surgery.Since selective catheterisation of theGEA graft to monitor patency, is oftenunsuccessful, a non-invasive protocolto visualise the GEA-graft from originto anastomosis is presented using 16-slice multidetector computed tomog-raphy (MDCT). Twenty-six malepatients (mean age 58.1±6.7 years)with GEA grafts were scanned ac-cording to a protocol of an ECG-synchronised cardiac scan followed bya thoracoabdominal scan. To terminatethe scan at the correct anatomicallevel, the lowest level of the GEAwascoded based on the lumbar vertebraelevel. Scores ranging from one (ex-cellent) to four (bad) were assigned toevaluate visualisation quality of the

grafts. GEA grafts were assessable in62% of the thoracoabdominal scansand 69% of the cardiac scans. Onaverage, the lowest part of the GEAcorresponded with a level between L1and L2, in two cases in the upper partof L3. Mean visualisation score in thethoracoabdominal scans and cardiacscans was good (respectively 1.4±0.6and 1.4±1.0). Sixteen-slice MDCT is apromising alternative for catheterisa-tion in evaluating patency of GEAgrafts, using the presented protocolwith thoracoabdominal scan includingL3 for complete coverage of the GEAgraft.

Keywords Multidetector computedtomography . Non-invasive coronaryimaging . Coronary artery bypassgraft . Gastroepiploic artery

Introduction

Arterial coronary artery bypass grafts (CABG), especiallythe left and right internal mammary artery (LIMA andRIMA), have been in widespread use for coronary surgeryfor 30 years. These grafts have improved their durabilityand have reduced postoperative complications without in-crease of perioperative complications compared with ve-nous bypass grafts [1, 2].

In patients with triple vessel disease with a coronarylumen diameter of less than 50%, the two internal mam-mary arteries (IMAs) cannot revascularise the entire myo-cardium. Since the late 1980s, the right gastroepiploicartery (GEA) has also been used as a bypass graft to over-come this problem [1, 2].

Since 1989, in the University Medical Center Groningenapproximately 1300 arterial CABG operations have beenperformed also using GEA (Fig. 1). Patients require fre-quent follow-up to monitor the patency of the bypass graftsand the condition of the coronary arteries. Sometimes,follow-up includes evaluation of the graft with the ref-erence standard, coronary angiography (CAG). However,selective catheterisation of the GEA graft is difficult and isunsuccessful in 22% of examinations [3]. Furthermore,CAG has significant morbidity and mortality [4, 5]. Non-invasive diagnostic modalities have a successful diagnos-tic improvement, with the main advantages being shorterexamination times, implementation in an outpatient setting,lower costs and lower risks. Evaluation of IMA and venousbypass grafts using single slice spiral CT [6], electron beam

J. Dorgelo (*) . T. P. Willems .P. M. A. van Ooijen . M. OudkerkDepartment of Radiology,University Medical Center Groningen,Hanzeplein 1,PO Box 30001, 9700Groningen, The Netherlandse-mail: j.dorgelo@rad.umcg.nlTel.: +31-50-3614260

G. F. V. Panday . P. W. Boonstra .F. ZijlstraThoraxcenter,University Medical Center Groningen,Groningen, The Netherlands

CT (EBT) [7], four-slice multidetector CT (MDCT) [8–11]and 16-slice MDCT [12, 13] have been described ex-tensively. However, no CT data have been reported on non-invasive evaluation of gastroepiploic bypass grafts.

In this article, a protocol to visualise the entire GEA graftfrom origin to anastomosis, including an ECG gated scanof the heart for detailed evaluation of the anastomosis, ispresented. The correct anatomical level of terminating thescan without missing the GEA origin will be discussedmore specifically.

Patients, materials and methods

Patients

MDCT imaging was performed prospectively on 26 con-secutive, asymptomatic male patients (mean age 58.1±6.7years) with a gastroepiploic bypass graft. Twenty-one ofthese patients also had a RIMA and LIMA graft, three anadditional LIMA graft and one an additional RIMA graft.The mean period of time since the CABG operation was9.4±1.7 years. Mean heart rate during cardiac scanning was64.7±13.0 bpm. Informed consent was obtained from allpatients.

Multidetector computed tomography

Data acquisition

Scans were performed on a 16-slice multidetector sys-tem (Somatom Sensation 16; Siemens AG, Forchheim,Germany). The scan protocol consisted of an ECG syn-chronised cardiac scan and a standard contrast enhancedthoracoabdominal scan. Both scans were made in cranio-caudal direction with patients in supine position. No addi-tional beta-blockers were administered prior to scanning.

The procedure started with automated bolus detectionwith the region of interest placed in the descending aorta atthe level of the carina, followed by an inspiratory breath-hold cardiac scan of approximately 20 s with a 12×0.75mm collimation. ECG data were recorded simultaneously.Lastly, an inspiratory breathhold scan (duration approxi-mately 6 s with a collimation of 16×1.5 mm) at the tail ofthe bolus was made to visualise the complete route of allbypasses (IMAs and GEAs). This scan included the thoraxand upper third of the abdomen to visualise the total routeof the IMAs and gastroepiploic artery.

Iomeprol 400 mg/ml (Iomeron 400, Bracco, Italy) wasused as contrast agent. This was injected through a 20-Gvenflon placed in a cubital vein as a biphasic injection, 100ml with a rate of 4.0 ml/s for the cardiac scan and 80 ml at2.5 ml/s for the thoracoabdominal scan. Scan parametersare shown in Table 1.

Fig. 1 Impression of the three arterial coronary artery bypass grafts.The right gastroepiploic artery originates from the gastroduodenalartery, a side branch of the hepatic artery

Table 1 Scan parameters of the cardiac scan and thoracoabdominalscan

Parameter Cardiac scan Thoracoabdominalscan

Tube current (mAs) 500 90Tube voltage (kV) 120 120Collimation 12×0.75 mm 16×1.5 mmGantry rotation time (ms) 420 500Table feed per gantryrotation

2.8 30

Synchronizationto the ECG

Retrospectively No

Effective slicethickness (mm)

1 2

Reconstructionincrement (mm)

0.6 1.2

Field of view (mm) 220 380Contrast agent Iomeprol

400 mg/mlIomeprol400 mg/ml

Amount (ml) 80 100Rate (ml/s) 4.0 2.5

1995

Data reconstruction

Retrospective gating was performed at an optimal recon-struction window. This window was determined by choos-ing the best image of a slice through the mid-segments ofthe three main coronary arteries (right coronary artery, leftanterior descending coronary artery and circumflex coro-nary artery), reconstructed in 20 phase steps of 50 ms.

Data were evaluated using dedicated post-processinghardware and software (Vitrea 2; Vital Images, Plymouth,Minn., USA). Post-processing time was approximately 15min per patient.

Data evaluation

The GEA was visualised from origin to anastomosis toprevent missing stenoses at the origin of the vessel. Thelowest slice of the thoracoabdominal scans and the lowestlevel of the GEAwere coded based on the lumbar vertebraelevel to determine the scan length for complete imaging ofthe GEA. The vertebrae were divided into three parts,upper, middle and lower. The upper part of L1 was coded 1,the middle part of L1 was coded 2 continuing up to thelower part of L4, which was coded 12 (Fig. 2).

The vessel assessability was recorded for both coronaryarteries and bypass grafts. A visualisation score of 1 (ex-cellent visualisation), 2 (good), 3 (poor) or 4 (bad vis-ualisation) was used to grade the visualisation of the bypassgrafts. All bypass grafts (from origin to anastomosis) werescored for visibility in the thoracoabdominal scans. Thescanned parts of the grafts and coronary arteries before andafter the anastomoses were scored in the cardiac scans. Theanastomoses segments were scored separately.

In both cardiac and thoracoabdominal scans, the axialslices and, if necessary, multiplanar reformations (MPR),maximum intensity projections (MIP) and volume render-ing (VR) images were used for scoring the vessels. Thecause of non-assessability of vessels was recorded.

Results

In the thoracoabdominal scans, 16 out of 26 GEA bypassgrafts could be assessed (62% assessability). In five cases,the grafts could not be evaluated because the origin was notdisplayed due to premature termination of the scan. Poor orno contrast filling was observed in another five cases. Nodifferentiation could be made as to whether these findingswere caused by occlusion of the vessel or low contrastconcentrations.

The thoracoabdominal scans were terminated at a meanlevel of 4.8, corresponding with the mid-level of L2. In thefive premature terminated scan, mean level of terminationwas 3.4, corresponding with a level between L1 and L2.The lowest part of the vessel in the 16 scans with completeGEAs, was 3.8, corresponding with a level between L1 andL2. In two cases the lowest level was in the upper part ofL3.

In the cardiac scans, 18 out of 26 bypass grafts wereassessable (69%). Five grafts could not be evaluated due topoor or absent contrast filling, as in the thoracoabdominalscans. Patient movement, vascular clip scattering and theabsence of the graft due to a termination of the scan abovethe level of the anastomosis caused unassessability of threeother grafts.

The mean visualisation score of the bypass grafts in thethoracoabdominal scans was good to excellent (1.4±0.6).No stenoses were observed in the assessable grafts. In thecardiac scans, mean scores for the grafts and their anas-tomoses were respectively 1.4±1.0 and 1.9±1.1 (Figs. 3, 4and 5). Scattering of vascular clips often decreased thequality of the images of the anastomoses and reduced thescore.

Maximum intensity projections (MIP) were used twicein addition to the axial slices, to obtain a better impressionof the bypass grafts in the thoracoabdominal scans. In fourcardiac scans, additional techniques had to be used, namelythree MIPs and one volume rendering.

Fig. 2 The encoding of the level of the lowest slice of thethoracoabdominal scans and the lowest level of the gastroepiploicartery, based on the lumbar vertebrae. The vertebrae were divided inan upper, middle and low part. The encoding ranged from 1 (upperpart of L1) to 12 (lower part of L4)

1996

Discussion

Patients with coronary artery bypass grafts require frequentfollow-up to monitor the patency of grafts and condition ofthe coronary arteries, sometimes including coronary angi-ography. Selective catheterisation of the gastroepiploicartery can be difficult, resulting in unsuccessful catheter-isation of up to 22% [3].

Multidetector CT may be a suitable candidate to replacediagnostic coronary catheter angiography (CAG) [14, 15].Follow-up of coronary artery bypass grafts seems to be agood indication. Excellent visualisation of bypass graftshas been described in literature for four-slice MDCT [9]and 16-slice MDCT [12]. Both Ropers et al. [8] andNieman et al. [10] reported good accuracy in detecting graft

Fig. 5 Volume rendering images of a thoracoabdominal scan withsegmentation of the gastroepiploic artery graft, arising from theceliac trunk via the common hepatic artery and gastroduodenal ar-tery. The anastomosis and the coronary arteries cannot be evaluatedsufficiently on this scan due to the larger collimation and absence ofECG synchronisation

Fig. 3 Volume rendering image of a cardiac scan showing a gas-troepiploic artery connected to a calcified right coronary artery and aleft internal mammary artery connected to a calcified and narrowedleft anterior descending coronary artery

Fig. 4 Volume rendering image showing a detail of a gastroepiploicartery connected to a posterior descending branch

1997

patency and graft stenoses using MDCT with catheterangiography as reference standard.

Our results show good to excellent visualisation ofgastroepiploic artery bypass grafts with a 16-slice MDCT.The scan protocol consists of a cardiac scan, followed by athoracoabdominal scan and a biphasic contrast materialinjection to enable evaluation of the entire route of thebypass graft. The thoracoabdominal scan has to include L3to be able to visualise the entire GEA graft in all patients.This is important, since in the case of a non-visualisedorigin of the graft, one cannot differentiate between nor-mal antegrade contrast filling and retrograde contrast fill-ing caused by an occlusion of the origin of the vessel.However, in our protocol, the presence of antegrade flowremains an assumption since MDCT only provides infor-mation on the vessels at a certain position in the RR-interval and does not provide flow information. A possiblesolution to this problem has been reported by Tello et al.[16]. They describe that both flow grade in the vessels andits direction can be predicted by generating three-dimen-sional flow images of bypass grafts using multiple 3Dimages in the same projection, each at a threshold 10Hounsfield Units higher than the previous image. Thistechnique was not used in our patient group but might beuseful in case of doubt.

Both the GEA grafts as well as the anastomoses of thegrafts on the coronary arteries can be visualised adequatelyin the cardiac scan. However, the anastomoses were scoredless well than the grafts, primarily due to local scatteringartefacts of metallic vascular clips.

In this study, thoracoabdominal scans included the entirethorax in 25 out of the 26 patients since they had one or twointernal mammary artery grafts. To prevent to scan theheart twice, an alternative protocol could be as follows: toscan the upper thoracic part (RIMA and LIMA grafts) with16×1.5 mm collimation, the heart with 12×0.75 mm col-limation and ECG synchronisation and again the upperabdominal part (the GEA graft) with 12×0.75 mm colli-mation. However, it is not possible to perform this in onecontinuous scan. Furthermore, in avoiding missing parts ofthe grafts one would have to overlap the scans slightly. Itwas calculated (ImPACT CT Patient Dosimetry Calculator;Impact group, London, UK) that a dose reduction of only6% (maximum 14 mSv instead of 14.9 mSv) would beestablished using this alternative protocol. Moreover, anadditional contrast administration would be necessary andno continuous image reconstructions could then be made.In patients with a single GEA graft, one can reduce ra-

diation dose and scan time by limiting the thoracoabdom-inal scan from the base of the heart to the lowest level ofL3. Extending the cardiac scan to the lowest level of L3 isnot recommended. The absence of cardiac motion in thisregion does not necessitate ECG-synchronised scanningand doubles scan length and radiation dose.

In five patients, poor or no contrast filling of the GEAgraft was observed. This was observed in the same patientsin both cardiac scan and thoracoabdominal scan and allneighbouring vessels showed good contrast filling. There-fore, this was rather thought to be a problem of graft pa-tency than of poor contrast administration. This idea issupported by findings in literature of relative frequentabsence of GEA patency in complaint free patients. Forexample Hirose et al. [17] described 1-, 3-, and 5-year GEAgraft patency rates of 98.7, 91.1, and 84.4%, respectively.

The main limitation of this study is that no comparisoncould be made between 16-slice MDCT and coronary an-giography as reference standard to determine the accuracyof the technique. However, it should be stressed that theprimary goal was to evaluate the quality of the GEA graftsand anastomoses using MDCT and not to determine thediagnostic accuracy of this technique in this asymptomaticpatient group.

Patients did not receive additional beta-blockers prior toMDCT scanning in this study. The current scan protocolprescribes that patients receive an oral dose of 50 mgmetoprolol 1 h prior to scanning in case of heart rateshigher than 70 bpm. In this way cardiac motion artefacts inthe ECG-gated scans are reduced and scan quality isimproved.

Conclusion

Catheterisation of gastroepiploic bypasses can be difficultand is not possible in 22% of the patients. Results of thisstudy show that 16-slice MDCT is a promising alternativefor coronary angiography in the follow-up of gastroepi-ploic arterial coronary artery bypass grafts patency using aprotocol consisting of an ECG-synchronised cardiac scanand a standard thoracoabdominal scan including L3 forcomplete coverage of the GEA graft. Beta-blocker admin-istration prior to the scan is recommended in case of highheart rates.

Acknowledgements We thank Dr. Estelle Noach for criticalreading of the manuscript and Wim Tukker for technical support.

1998

References

1. Bergsma TM, Grandjean JG, Voors AAet al. (1998) Low recurrence of anginapectoris after coronary artery bypassgraft surgery with bilateral internalthoracic and right gastroepiploic arter-ies. Circulation 97:2402–2405

2. Grandjean JG, Voors AA, Boonstra PWet al. (1996) Exclusive use of arterialgrafts in coronary artery bypass opera-tions for three-vessel disease: use ofboth thoracic arteries and the gastro-epiploic artery in 256 consecutive pa-tients. J Thorac Cardiovasc Surg112:935–942

3. Isshiki T, Yamaguchi T, Nakamura M etal. (1990) Postoperative angiographicevaluation of gastroepiploic arterygrafts: technical considerations andshort-term patency. Catheter Cardio-vasc Diagn 21:233–238

4. de Bono D (1993) Complications ofdiagnostic cardiac catheterisation: re-sults from 34,041 patients in the UnitedKingdom confidential enquiry intocardiac catheter complications. TheJoint Audit Committee of the BritishCardiac Society and Royal College ofPhysicians of London. Br Heart J70:297–300

5. Laskey W, Boyle J, Johnson LW (1993)Multivariable model for prediction ofrisk of significant complication duringdiagnostic cardiac catheterization. Theregistry committee of the society forcardiac angiography and interventions.Catheter Cardiovasc Diagn 30:185–190

6. Ueyama K, Ohashi H, Tsutsumi Y et al.(1999) Evaluation of coronary arterybypass grafts using helical scan com-puted tomography. Catheter CardiovascInterv 46:322–326

7. Achenbach S, Moshage W, Ropers D etal. (1997) Noninvasive, three-dimen-sional visualization of coronary arterybypass grafts by electron beam tomog-raphy. Am J Cardiol 79:856–861

8. Ropers D, Ulzheimer S, Wenkel E et al.(2001) Investigation of aortocoronaryartery bypass grafts by multislice spiralcomputed tomography with electrocar-diographic-gated image reconstruction.Am J Cardiol 88:792–795

9. Frohner S, Wagner M, Schmitt R et al.(2002) Multi-slice CT of aortocoronaryvenous bypasses and mammary arterybypasses: evaluation of bypasses andtheir anastomoses. Rontgenpraxis54:163–173

10. Nieman K, Pattynama PM, Rensing BJet al. (2003) Evaluation of patients aftercoronary artery bypass surgery: CTangiographic assessment of grafts andcoronary arteries. Radiology 229:749–756

11. Marano R, Storto ML, Maddestra N etal. (2004) Non-invasive assessment ofcoronary artery bypass graft with ret-rospectively ECG-gated four-rowmulti-detector spiral computed tomog-raphy. Eur Radiol 14:1353–1362

12. Gurevitch J, Gaspar T, Orlov B et al.(2003) Noninvasive evaluation of arte-rial grafts with newly released multi-detector computed tomography. AnnThorac Surg 76:1523–1527

13. Panday GF, Grandjean JG, Ho KYet al.(2003) A rare case of herniation ofomentum into the pericardial cavityafter using the right gastro-epiploicartery for coronary bypass grafting.Interact Cardiovasc Thorac Surg2:154–155

14. Nieman K, Cademartiri F, Lemos PAet al. (2002) Reliable noninvasivecoronary angiography with fast sub-millimeter multislice spiral computedtomography. Circulation 106:2051–2054

15. Ropers D, Baum U, Pohle K et al.(2003) Detection of coronary arterystenoses with thin-slice multi-detectorrow spiral computed tomography andmultiplanar reconstruction. Circulation107:664–666

16. Tello R, Hartnell GG, Costello P et al.(2002) Coronary artery bypass graftflow: qualitative evaluation with cinesingle-detector row CT and comparisonwith findings at angiography. Radiolo-gy 224:913–918

17. Hirose H, Amano A, Takanashi S et al.(2002) Coronary artery bypass graftingusing the gastroepiploic artery in 1,000patients. Ann Thorac Surg 73:1371–1379

1999