Lumen Loss in the First Year in Saphenous Vein Grafts Is Predominantly a Result of Negative Remodeling of the Whole Vessel Rather Than a Result of Changes in Wall Thickness
Background— The use of saphenous vein grafts (SVG) in coronary artery bypass surgery is established but little is known of SVG remodeling during the first year in vivo.
Methods and Results— The feasibility of measuring total vessel diameter (lumen plus wall), lumen diameter, and wall thickness by a novel computed tomography (CT) method was established in phantom model tubes (r=0.98 for lumen diameter and r=0.98 for wall thickness) and in an initial clinical study of 14 patients correlating CT and intravascular ultrasound measurements of SVG (r=0.88 for total vessel diameter, r=0.85 for lumen diameter and r=0.89 for wall thickness). In a separate group of 42 patients (aged 66±10 years; 36 male, 6 female) undergoing coronary artery bypass grafting, SVG total vessel diameter, lumen diameter, and wall thickness were determined prospectively with multi-slice CT angiography at 1 and 12 months postoperatively. Mean total vessel diameter decreased from 5.95±0.83 mm to 5.39±0.87 mm, P<0.001 (range, −39% to +8% change). Twenty-six patients (62%) had a decrease of SVG vessel diameter (negative remodeling) >5%. Mean lumen diameter decreased from 3.69±0.66 mm to 3.36±0.68 mm, P<0.001, (range, −40 to +11% change). Surprisingly, mean wall thickness decreased from 1.14±0.27 mm to 1.01±0.21 mm (P<0.001; range, −48 to +33% change).
Conclusions— Lumen loss in SVG between postoperative months 1 and 12 is predominantly caused by negative remodeling of the whole vessel rather than to changes in wall thickness. Therapies targeting negative remodeling may be required for optimal maintenance of SVG lumen in the first postoperative year.
Revascularization of multiple coronary arteries during coronary artery bypass graft surgery (CABG)1 requires conduits additional to the left internal thoracic artery, and saphenous vein grafts (SVG) are frequently required. However, SVG demonstrate a significant deterioration with up to 26% occluding by 18 months postoperatively.2
Whereas thrombosis and surgical factors are the predominant causes of SVG failure in the first month after implantation, intimal hyperplasia is thought to be the predominant cause of graft failure from postoperative months 1 to 12.3 Venous conduits may exhibit mild intimal or medial fibrosis pre-grafting, but most develop further intimal thickening within 4 to 6 weeks of arterial anastomosis. In angiographically normal SVG, intravascular ultrasound (IVUS) and pathological studies have shown a doubling of intimal thickness4 and total wall thickness5 by the end of the first postoperative year.
Surprisingly little is known regarding the natural history of SVG remodeling in the first postoperative year. Some studies have suggested absence of positive remodeling,6 whereas others have suggested that both positive and negative remodeling occur.7 Critically, most of these data have been taken from different patients, rather than serially following-up the same patients and vessels. This study was, therefore, undertaken to describe the natural history of SVG lumen and vessel remodeling during the first postoperative year. To undertake such serial evaluation of lumen and wall thickness noninvasively, a novel multi-slice computed tomography (CT) application was developed, validated, and applied prospectively to a surgical cohort.
Written informed consent and Institutional Review Board Human Research Ethics Committee approval were obtained for the IVUS validation and SVG remodeling studies.
Validation in Phantom Models of SVG
Perspex tubes were engineered with varying lumen diameters and wall thickness (Department of Biomedical Engineering, Concord Hospital, Sydney, Australia), filled with contrast (Ultravist 300; Schering AG, at 1:400 dilution with normal saline) and sealed. The tubes were placed in random order in a columnar styrene container placed at an arbitrary oblique orientation in the CT scanner and scanned. For each tube, the mean lumen diameter (internal diameter on contrast-enhanced CT) and total vessel (lumen plus wall) diameter (external diameter on non-contrast CT) were then measured perpendicular to the long-axis of the tube on 3 axial images and the results averaged. Wall thickness was calculated for each tube using the formula: wall thickness=(total vessel diameter−lumen diameter)/2. After changing the oblique orientation of the tubes within the scanner, the measurements were repeated. The measurements were then compared with the engineered specifications, which had been verified with a micrometer to the nearest 0.05 mm.
Validation Against IVUS
Sixteen patients undergoing conventional coronary angiography for clinical indications, with CABG >2 years previously, and without hemodynamically significant stenosis of their SVG within 10 cm of the origin, agreed to participate. Two patients withdrew before CT scanning, leaving 14 patients for analysis. IVUS was performed of their vein graft at the completion of the conventional coronary angiography (standard Judkin technique). A 2.5-French IVUS catheter (Atlantis SR Pro 40 MHz; Boston Scientific Scimed Inc, Minn) was placed in the proximal 10 cm of the SVG (the limit of the IVUS catheter pullback system) and automated pullback activated at 1 mm/s. To minimize risk of anastomotic injury, the guide was not advanced past the first graft-to-artery anastomosis, limiting the length of SVG analyzable in 8 cases. Intravenous nitrate was not given. There were no adverse events.
IVUS images were recorded onto S-VHS videotape for offline analysis. The distance from the ostium was calculated from the catheter pullback speed. With careful observation of the different movements and echo intensity of the SVG relative to the surrounding tissue, IVUS measurements of total vessel diameter are possible.7 Because the IVUS catheter may deform the cross-sectional shape of the vein graft to an “egg shape” (see Figure 1C), the average of measurements in the major and minor axes were used for analyses of total vessel diameter and lumen diameter.
CT angiography was performed at 8±4 days after the IVUS. To minimize the potential for mis-registration between CT slices, all measurements were performed on axial CT images, rather than multi-planar reconstructions. The lumen diameter was measured perpendicular to the long-axis of the vessel on the contrast-enhanced axial images (see online data supplement). Total vessel diameter (wall plus lumen) was measured from the corresponding noncontrast CT axial image, measured perpendicular to the vessel long axis. The distance from the ostium was calculated by measurement of 10-mm intervals in oblique views along the vessel.
For both IVUS and CT, lumen diameter and total vessel diameter were assessed every 10 mm from the ostium, with these “segmental” measurements averaged for each graft for analysis. Wall thickness was calculated with the formula given previously and this is illustrated in Figure 1.
Saphenous Vein Graft Remodeling Study
Patients scheduled to receive at least 1 SVG and attending preoperative assessment clinic between October 2002 and January 2004 were screened. Patients were ineligible if the serum creatinine was elevated, there was past iodine or intravenous contrast allergy, or if patients were unable to return for follow-up. Seventy-four consecutive patients in sinus rhythm undergoing elective CABG agreed to participate in the study. Within the first postoperative month, 2 patients died (pulmonary embolism, cardiac arrest after ventricular tachycardia), 1 experienced a stroke, renal impairment developed in 2, and 7 withdrew from the study. Sixty-two patients thus underwent CT angiography at 35±8 days postoperatively. Six patients had occluded SVG, leaving 56 with at least 1 patent SVG. Three other patients had 1 of their 2 SVG occluded, so only the patent graft was included in the analyses.
Between 1 and 12 months postoperatively, 2 patients died (cardiac arrest), serious comorbidity developed in 3 (2 had carcinoma and 1 had recurrent septicemia), and 9 patients withdrew from the study. Thus, 42 patients with patent SVG had CT angiography at both 1 and 12 months postoperatively and were included in the serial analyses. Positive remodeling was defined as an increase in total vessel diameter >5% (equivalent to a 10% increase in SVG area)7 and negative remodeling as a decrease in total vessel diameter >5%.
Baseline characteristics recorded included age, gender, body mass index, history of smoking, hypertension, hypercholesterolemia, and diabetes. There were no current smokers in this cohort. Hypertension was defined as a recorded preoperative blood pressure >140/90 mm Hg or pharmacological treatment for hypertension. Hypercholesterolemia was defined as a fasting cholesterol >5.5mmol/L (210 mg/dL) or treatment with a cholesterol-lowering agent.
Metoprolol 50 to 100 mg orally was administered before CT scanning to 30 patients at 1 month and to 24 patients at 12 months to achieve average heart rates for the whole cohort of 62±9 and 63±9 per minute, respectively.8 A noncontrast ECG-gated cardiac CT scan (Lightspeed Plus, GE Medical Systems; 4×2.5 mm collimation, 1.3 to 1.5 pitch, 0.5-second rotation time, 140 kV, 250 mA) was performed. This was followed by a timing scan using 20 mL of intravenous nonionic contrast (Ultravist 300, Schering AG, Germany) and a contrast-enhanced scan (4×2.5 mm collimation, 1.3 to 1.5 pitch, 0.5-second rotation time, 140 kV, 270 mA) using 150 mL of intravenous contrast at 3.5 mL/s. The mean scan delay was 26 seconds (range, 18 to 36) and mean breath-hold 27 seconds (range, 22 to 32). To ensure consistency between patients, CT scans were reconstructed using a soft convolution kernel9 from a data acquisition window centered at 70% of the RR interval.10 Multi-sector reconstructions were used for heart rates >65 beats per minute to improve temporal resolution (&125 ms). There were no adverse events.
For all CT and IVUS studies, 2 readers, blinded to clinical details and results, independently assessed each SVG for total vessel diameter, lumen diameter, and wall thickness (Card IQ and Advantage Workstation 4.0; GE Medical Systems and Clearview 2000, Boston Scientific, Minnesota). The measurements for the 2 readers were averaged for analyses. When a patient had >1 patent SVG, the results of the 2 grafts were averaged for the analyses. The authors had full access to the data and take full responsibility for their integrity. All authors have read and agree to the manuscript as written.
SPSS for Windows 10.0 (SPSS Inc, Chicago, Ill) was used and a 2-tailed P<0.05 was considered significant. Interobserver and intra-observer reliability was calculated using the intra-class correlation coefficients (ICC) (mixed 2-way model), in which ICC closer to 1.0 indicates greater reliability than ICC closer to 0.11 CT measurements were compared with IVUS measurements with Pearson correlations. Serial measurements were compared with paired t-tests. Data were expressed as mean±standard deviation. Power calculations were performed using PASS software (2005; Number Cruncher Statistical Systems, Kaysville).
Validation in Phantom Models of SVG
The transverse dimensions and CT densities of the phantom models overlapped with those of the SVG. Lumen diameter (range, 3.18 to 4.80 mm), wall thickness (0.75 to 1.57 mm), and measured CT densities (lumen, 282 to 394 Hounsfield units [HU]; wall, 18 to 45 HU) of the phantom tubes were comparable to those of SVG. For SVG (n =14), IVUS measurements of lumen diameters ranged from 2.95 to 5.08 mm, wall thickness ranged from 0.45 to 1.04 mm, and measured CT densities ranged from 202 to 314 HU for lumen and 19 to 77 HU for wall.
Nine tubes were scanned in 2 orientations, giving 18 measurements of lumen diameter and wall thickness. The proximal end (to the scanner) of the tubes were measured at left 23°, anterior 12° oblique orientation on the first scan and right 30°, posterior 70° oblique orientation on the second scan, with respect to the axis of the CT table. Lumen diameter (r=0.95 and r=0.97) and wall thickness (r=0.95 and r=0.96) measured by CT correlated closely with the engineered specifications of the phantom models for both readers. For CT lumen and wall thickness measurements, there was good intra-observer (ICC for lumen diameter, 0.97; wall thickness, 0.97) and inter-observer reliability (ICC for lumen diameter, 0.97; wall thickness, 0.98).
Validation Against IVUS Measurements
Fourteen patients (age 66±8 years, 12 male, 2 female, 10±3 years after CABG) were prospectively studied with IVUS and CT angiography. Nine patients had diabetes, 1 smoked, and 7 patients were ex-smokers. The mean systolic and diastolic blood pressures were 139±24 mm Hg and 78±11 mm Hg, respectively. All patients were taking statins.
On IVUS, lumen diameter and wall thickness measurements were available for 111 segments of SVG 1 cm apart (7.9 per patient), because 8 of 14 SVG were <10 cm length before the first anastomosis. Of these 111 segments, 108 (97%) were assessable for lumen diameter and 88 (79%) for total vessel diameter (and wall thickness) on CT. There was inadequate contrast between SVG and the surrounding tissue to permit measurement of total vessel diameter in 20 of the 23 nonassessable segments on noncontrast-enhanced CT. All SVGs studied by IVUS had a minimum of 6 corresponding segments assessable by CT, and thus IVUS and CT comparisons were available for all SVGs studied.
When segmental measurements were averaged for each graft, total vessel diameter, lumen diameter, and wall thickness measured by CT correlated closely with IVUS (r=0.88, 0.85 and 0.89, respectively; Figure 2). The mean difference between CT and IVUS for total vessel diameter was 0.13±0.29 mm, lumen diameter 0.06±0.28 mm and wall thickness 0.02±0.09 mm. This equates to CT measurement of total vessel diameter being within 0.57 mm of IVUS, lumen diameter within 0.55 mm, and wall thickness within 0.17 mm in 95% of cases (see Bland-Altman graphs in Figure 2).12 For CT, the intra-observer and inter-observer reliability was high (ICC 0.97, 0.90 for total vessel diameter; 0.86, 0.87 for lumen diameter; and 0.90, 0.90 for wall thickness). For IVUS measurements, the intra-observer and inter-observer reliability was also high (ICC 0.99, 0.97 for total vessel diameter; 0.98, 0.98 for lumen diameter; and 0.96, 0.94 for wall thickness). The mean IVUS and CT measurements of total vessel diameter were 5.12±0.60 mm and 5.26±0.63 mm, lumen diameter 3.75±0.42 mm and 3.81±0.52 mm, and wall thickness 0.69±0.16 mm and 0.71±0.18 mm, respectively.
SVG Remodeling Study
CT angiography was used to assess SVG dimensions prospectively in a cohort of 42 patients. There were 56 patent SVG (1.3 per patient) and 99 distal anastomoses (1.8 per SVG, 19 to left anterior descending artery territory, 39 to left circumflex artery territory, and 41 to right coronary artery territory). The mean target vessel stenosis proximal to the anastomotic site was 75±17% by quantitative coronary angiography. There were no new occlusions between the 1 and 12 month CT scans. The baseline patient characteristics are given in Table 1.
Between postoperative months 1 and 12, the mean total vessel diameter decreased from 5.95±0.83 mm to 5.39±0.87 (P<0.001; range, −39% to +8% change). Twenty-six patients (62%) had decreased SVG total vessel diameter >5% (negative remodeling), whereas only 1 patient (2%) had increased total vessel diameter >5% (positive remodeling). The mean lumen diameter decreased from 3.69±0.66 mm to 3.36±0.68 mm (P<0.001), with considerable inter-patient variability (range, −40% to +11%). Surprisingly, mean wall thickness decreased from 1.14±0.27 to 1.01±0.21 mm (P<0.001; range, −48% to +33%; Figure 3). Comparison of segments before and after the first anastomosis in the 26 sequential SVG showed nonsignificantly greater total vessel diameter (5.54±0.89 mm versus 5.44±0.77 mm; P=0.42) and lumen diameter (3.46±0.81 mm versus 3.28±0.67 mm; P=0.06) in the proximal compared with the distal portions, but there was no difference in wall thickness (1.06±0.26 mm versus 1.08±0.19; P=0.70). The ICC for CT measurement of total vessel diameter, lumen diameter, and wall thickness was, respectively, 0.89, 0.92, and 0.86 at 1 month, and 0.90, 0.92, and 0.80 at 12 months. Post hoc calculations demonstrated 99% power to detect 0.56 mm difference in total vessel diameter at a significance level (alpha) of 0.05 using a 2-tailed paired t test.
Lumen loss after CABG is important, because SVG with smaller lumen diameters are more prone to early graft failure.13 We have identified in nonoccluded SVG a mean loss of SVG lumen diameter of 9% (3.69 to 3.36 mm) between postoperative months 1 and 12, and a decrease in SVG wall thickness over this time.
Lumen loss can result from negative remodeling (loss of total vessel diameter) and/or wall thickening. This study shows that in the first postoperative year, lumen loss in SVG is predominantly caused by negative remodeling, with a mean decrease total vessel diameter of 0.56 mm, and the unexpected decrease in mean wall thickness of 0.13 mm. The pathophysiology of vascular remodeling is incompletely understood, but data suggest that remodeling is initiated by changes in hemodynamic conditions (flow, wall stretch, shear stress) and humoral factors (cytokines, vasoactive substances). These lead to signals that influence cell growth and migration and altered activity of matrix metalloproteinases.14 Furthermore, the change in lumen diameter is modifiable.15
IVUS and pathology studies have described an increase in SVG intima area16 and adventitia17 days to weeks after implantation.4 Pathology studies have shown that differences in the wall thickness of grafts are predominantly caused by differences in thickness of the intima, rather than of the media or adventitia.5 As most patients in this study had stable or reduced wall thickness between postoperative months 1 and 12, it is likely that wall thickening is most active within the first 4 to 6 weeks after SVG implantation. We postulate that early cellular and extracellular changes mediate increased wall thickness of the first 6 weeks, which then plateaus and, in the absence of atherosclerosis, may regress.
The wall thickness of SVG reference segments derived from previous studies (using the formula area=π · radius2) are &0.47 to 0.58 mm before implantation,5,6 1.01 mm 1 year after implantation,18 and 0.38 to 0.91 mm in chronic studies (average 8 to 11 years after implantation).7,19,20 These data support our observations. Moreover, some animal studies indicate that neointimal growth may stabilize after the first few weeks21 and human studies have not followed wall thickness serially in the same patients.6,18 The high usage of statins (81%) in this study may also have influenced wall thickness, because statins may reduce neointima formation in vein grafts via multiple mechanisms.22–24 Among the IVUS validation subjects (10±3 years after CABG), we found a mean wall thickness of 0.71 mm, consistent with previous long-term IVUS studies,7,19,20 and suggesting that in the absence of atherosclerotic plaque build-up, further but slowed progressive wall thinning may occur after the first year.
The clinical implications of these findings are that the major wall thickness changes in SVG are likely to occur within the first few weeks after implantation, and that strategies targeting remodeling, rather than intima-media thickness alone, may be critical for maximizing SVG lumen at 12 months. This may account for recent disappointing results with a E2F transcription factor decoy (an inhibitor of SVG neointima) to improve graft patency.2 With positive remodeling being more common in SVG with ruptured plaques than those with nonruptured plaques,20 the identification of early negative remodeling may also indicate that mild negative remodeling represents the normal SVG response to arterialization, with reduced risk of vulnerable plaque developing. The application of CT angiography to the study of SVG represents an exciting opportunity to understand the natural history of SVG remodeling and to identify factors determining vessel remodeling, patency, and changes in lumen and wall thickness.
The studies were commenced when 4-detector row CT scanners were commonly used. Current CT scanners with 64-detector rows, improved z-axis, and temporal resolution should have greater accuracy in patients with faster heart rates and in grafts that are not cylindrical (by reducing the volume averaging in an axial slice). However, the resolution in the axial plane, as used in the methodology of these studies, remains largely the same, namely 0.5×0.5 mm, and the validity of these results should not be affected.
Lumen loss in SVG between postoperative months 1 and 12 is predominantly caused by negative remodeling of the whole vessel rather than to changes in wall thickness. Further investigation of the determinants of early SVG remodeling will be required and attempts to maximize SVG lumen may need to target remodeling rather than intima-media thickness alone.
The authors thank John Ingall for assistance with CT scan protocols and Viet Nguyen for the mathematical proof.
Sources of Funding
The work was supported by the National Heart Foundation of Australia (L.K., G.L.) and a Pfizer Cardiovascular Lipid research grant.
Presented at the American Heart Association Scientific Sessions, Dallas, Tex, November 13–16, 2005.
Alexander JH, Hafley G, Harrington RA, Peterson ED, Ferguson TB Jr, Lorenz TJ, Goyal A, Gibson M, Mack MJ, Gennevois D, Califf RM, Kouchoukos NT. Efficacy and safety of edifoligide, an E2F transcription factor decoy, for prevention of vein graft failure following coronary artery bypass graft surgery: PREVENT IV: a randomized controlled trial. JAMA. 2005; 294: 2446–2454.
Hozumi T, Yoshikawa J, Yoshida K, Akasaka T, Takagi T, Honda Y, Okura H. Use of intravascular ultrasound for in vivo assessment of changes in intimal thickness of angiographically normal saphenous vein grafts one year after aortocoronary bypass surgery. Heart. 1996; 76: 317–320.
Nishioka T, Luo H, Berglund H, Eigler NL, Kim CJ, Tabak SW, Siegel RJ. Absence of focal compensatory enlargement or constriction in diseased human coronary saphenous vein bypass grafts. An intravascular ultrasound study. Circulation. 1996; 93: 683–690.
Nieman K, Rensing BJ, van Geuns RJ, Vos J, Pattynama PM, Krestin GP, Serruys PW, de Feyter PJ. Non-invasive coronary angiography with multislice spiral computed tomography: impact of heart rate. Heart. 2002; 88: 470–474.
Daley LE, Bourke GJ. Bias and measurement error. In: Daley LE, Boruke GJ, editors. Interpretation and Uses of Medical Statistics. London: Blackwell Science; 2000: 381–421.
Akowuah EF, Gray C, Lawrie A, Sheridan PJ, Su CH, Bettinger T, Brisken AF, Gunn J, Crossman DC, Francis SE, Baker AH, Newman CM. Ultrasound-mediated delivery of TIMP-3 plasmid DNA into saphenous vein leads to increased lumen size in a porcine interposition graft model. Gene Ther. 2005; 12: 1154–1157.
Shi Y, Pieniek M, Fard A, O’Brien J, Mannion JD, Zalewski A. Adventitial remodeling after coronary arterial injury. Circulation. 1996; 93: 340–348.
Pregowski J, Tyczynski P, Mintz GS, Kim SW, Witkowski A, Waksman R, Pichard A, Satler L, Kent K, Kruk M, Bieganski S, Ohlmann P, Weissman NJ. Incidence and clinical correlates of ruptured plaques in saphenous vein grafts: an intravascular ultrasound study. J Am Coll Cardiol. 2005; 45: 1974–1979.
Cooley BC. Murine model of neointimal formation and stenosis in vein grafts. Arterioscler Thromb Vasc Biol. 2004; 24: 1180–1185.