Burris, Nicholas; Schwartz, Kimberly; Brown, Jamie; Kwon, Michael; Pierson, Richard III; Griffith, Bartley; Poston, Robert
Objective: Strands of clot are frequently flushed out of saphenous vein grafts (SVG) during preparation for grafting, particularly those that are endoscopically harvested. However, saline distention at uncontrolled pressures increases graft thrombogenicity and the risk of early failure after coronary artery bypass grafting. The purpose of this prospective investigation was to define the incidence of intraluminal clot within endoscopically harvested SVG and the effect of attempted removal by saline distention.
Methods: Endoscopically harvested SVG were intraoperatively prepared for grafting by using saline distention at uncontrolled pressure (n = 24) or without distension (n = 20). Optical coherence tomography, a catheter-based infrared imaging system, was used to identify and characterize intraluminal clot strands in surplus SVG segments (average length for analysis, 4.9 ± 2.6 cm). These segments were also assessed for luminal tissue factor activity and percent endothelial integrity by CD31-directed immunohistochemistry.
Results: Clot strands were observed in 45.4% (20 of 44) of imaged SVG segments (severity of observed clots: 54%, mild; 32%, moderate; 14%, severe). Compared with grafts distended with saline, vein segments that were not distended displayed significantly higher endothelial integrity (60.1% ± 27.2% versus 24.7% ± 24.1%, P < 0.05) and lower tissue factor activity (1.28 ± 0.95 versus12.3 ± 5.5 U/cm2, P < 0.001) despite having a higher incidence of clot stands (65.0% versus 29.1%, P < 0.02, Fisher exact test). Static flow was observed in veins during endoscopic harvest.
Conclusions: Clot strands of varying severity are a common finding after endoscopic vein harvest. Saline distention is not completely effective in removing clot strands and increases overall graft thrombogenicity. Therefore, prevention of clot or less traumatic methods of removing clot are indicated.
Residual clot strands have been anecdotally reported after endoscopic harvest; however, without a method to detect and quantify intraluminal clot, this problem remains poorly understood. Optical coherence tomography (OCT) is a catheter-based imaging modality that can provide luminal images of the saphenous vein graft (SVG) at near-histologic resolution. Because images are provided in “real time,” OCT is an ideal tool for imaging clot strands in the operating room. The purpose of this study was to establish the incidence and severity of clot stands within the SVG after endoscopic harvest by using this highly sensitive imaging method.
Patient Enrollment and Data Management
After institutional review board approval was given (protocol H25350), all subjects provided informed consent before enrollment. Of 166 study subjects who underwent isolated coronary artery bypass grafting (CABG) at our institution from February 2005 until March 2006, an unselected subset of 41 (24.6%) had an OCT evaluation performed during surgery on surplus segments of 44 SVG. Demographics, preoperative risk factors and medications, and intraoperative and postoperative data were prospectively imported into a relational database.
Off-pump CABG (OPCAB) was performed by experienced surgeons through median sternotomy. The left internal thoracic artery (ITA) was harvested in all patients, and SVG were harvested endoscopically (VasoView6, Guidant Systems, Inc., Minneapolis, Minn). Heparin was given at the completion of the ITA takedown (but after endoscopic conduit harvest) at a dose calculated to obtain an ACT >300 seconds and heparin level >2 IU/mL, according to the protamine titration method (HMS heparin assay cartridges, Medtronic, Inc., Minneapolis, Minn). Preoperative aspirin (325 mg by mouth, daily) was continued and given within 6 hours after surgery. To prepare the conduits for grafting, procured SVG grafts were pressure-flushed with heparinized saline, using syringe injection with no methods used to control the distending pressure (n = 24). A separate subset of SVG were distended only at physiologic arterial pressure by creating the proximal anastomosis first and occluding the distal vessel with a bulldog clamp (n = 20). Both harvesters in this study had an experience level of more than 1,000 endoscopic vein harvests.
Before procurement, blood flow within the in situ saphenous vein was measured with the use of transit time techniques (VeriQ, Medistim, Inc.). Measurements were obtained with and without CO2 insufflation necessary to create visualization for the vein harvest in two locations: adjacent to the inflated trochar balloon and at the proximal end of the vein through a stab incision in the groin.
Ex Vivo OCT Analysis
Discarded segments of SVG were stored in Hanks Balanced Salt Solution (HBSS) at 4°C before analysis. Ex vivo evaluations by OCT (Lightlab, Inc., Westford, Mass), completed within 2 hours of removal from the operative field, were initiated by inserting the probe into the vessel though a Y-connector attached to a vein cannula. Occlusion of the other end of the segment allowed for gentle infusion of HBSS during imaging. Automated pullback images were graded by two separate reviewers for the presence and severity of clots strands. The infrared laser light emitted by OCT can be seen through the wall of the vein; therefore, the beginning and ending points of clot strands could be marked and measured externally. Clot were classified as (1) mild: single strand of clot <1 cm of the longitudinal length of the vein, (2) moderate: strand >1 cm of the longitudinal length of the vein, and (3) severe: multiple strands.
Biopsy specimens for histologic processing were procured at the completion of the ex vivo OCT scan. To exactly match the OCT images with the corresponding histopathologic sections, the vessel site where the biopsy was obtained was marked externally at the location of the catheter, visualized by the rotating infrared light at the catheter tip. These “image-guided” biopsy specimens were then stored in solution before being embedded and frozen in cutting compound (Tissue-Tek OCT, Redding, Calif). Additional formalin-fixed sections were embedded in paraffin, sectioned at 5-μm thickness, and stained with hematoxylin and eosin. Microscopic sections were analyzed by immunohistochemistry and with image analysis software (Bioquant Corp., Nashville, Tenn) to determine the percent of the luminal surface staining positive for the endothelial marker CD31 (R & D System, Inc., Minneapolis, Minn) as previously described.1 Gross confirmation of the clot strand on the luminal surface was obtained in a subset by opening the vein segments longitudinally to provide for direct inspection (n = 4).
Clots retrieved from the lumen of the SVG (n = 2) were analyzed for thrombin activity by incubation with the chromogenic substrate, S-2238 (333 μmol/L, Chromogenix) in a reaction buffer for 30 minutes. The absorption of the reaction buffer was assessed at 405 nm and then compared with a standard curve to determine thrombin activity. Immunohistochemical staining for p-selectin and CD41a (BD Pharmigen) was used to confirm the presence of platelets. A subset of discarded vein segments (n = 9) had their luminal surface exposed to a human plasmin solution (0.7 units/mL) (Sigma) for 1-hour incubation at 37°C, and the effusate was analyzed by enzyme-bound immunosorbent assay (Quadratech, Surrey, United Kingdom) for levels of d-dimer (soluble fibrin degradation product).
Tissue Factor Activity
Tissue factor activity on the luminal surface of the SVG was determined in conduits that were prepared for grafting with (n = 24) or without (n = 20) pressure distension. A 1-cm2 biopsy specimen was gently washed with HBSS, and its luminal surface was isolated in a custom-designed assay chamber. Tissue factor activity was determined by the addition of a reaction buffer (150 μL of 50 mmol/L Tris hydrochloride at pH 7.40, 2 mmol/L CaCl2) containing Factor VII (2 U/mL, American Diagnostica, Greenwich, Conn) and Factor X (2 U/mL, American Diagnostica). After a 20-minute incubation period, the reaction was stopped by the addition of EDTA (Sigma) and the generation of Factor Xa measured in the supernatant by adding a chromogenic substrate (Spectrozyme FXa, American Diagnostica). The absorption was measured in a spectrophotometer at 405 nm.
Early Angiographic Follow-up
Computed tomographic scans were acquired before discharge for all patients by using a 16-slice, multidetector-row CT scanner (MX8000IDT, Phillips Medical, Best, The Netherlands). Images were obtained with the use of a collimation of 0.75 mm × 16, thickness of 1 mm, scanning technique of 140 kVp, 350 to 500 mAs, and pitch 0.2 to 0.3, using 120 to 150 mL of iodinated contrast injected at 3 to 4 mL/s. Thrombosis was defined as complete occlusion of the graft on CT angiography scans, evidenced by the lack of contrast filling in the body of the graft.
Incidence, Severity, and Characterization of Residual Clot Strands
The average length of discarded segments available for analysis was 4.9 ± 2.6 cm. Luminal irregularities imaged on OCT (Fig. 1A) were confirmed to be clot by gross examination (Fig. 1B) and by immunohistochemical staining for p-selectin and CD41a (Fig. 1C). These clot strands were observed in 45% (20 of 44) of imaged SVG segments. The majority of clots were categorized as mild and focal (54%) (Fig. 2A). Clots of moderate severity were observed in 32% and severe clots with multiple strands (Fig. 2B) were observed in 14% of evaluated segments. Gentle flushing of the lumen during imaging revealed that the mild-moderate clot strands were largely mobile but remained fixed in discrete positions such as valve leaflets and ostium of vein branches for 68% of imaged clot strands. This suspicion about the site of clot attachment was confirmed by gross examination and biopsy. Vein segments in which clot was detected by OCT(n = 7) showed an average d-dimer lever of 1680.6 ± 2718.9 ng/mL in the effusate, confirming the presence of fibrin as compared with a d-dimer level of 805.1 ± 574.3 ng/mL in the effusate of segments that showed no OCT evidence of clot (n = 2).
Risk Factors for Clot
Veins that were distended were found to have a significantly lower incidence of residual clot strands than veins that were not distended (65% versus 29%, P < 0.02, Fisher exact test). Flow measurements taken during CO2 insufflation of the dissection tunnel showed average flow velocities of 1.1 ± 2.0 at the distal and 2.0 ± 1.7 mL/min at the proximal portion of the vein, with no flow in both portions noted in 71% of veins analyzed (Table 1). In contrast, vein flow remained constant in SVG that were analyzed during harvest using an open technique (n = 3). The average time of CO2 insufflation was 15 ± 4.8 minutes.
Biopsy specimens from clot strands showed an average thrombin activity of 210 ± 178 mU/mg clot. Immunohistochemical staining of vein segments containing clot revealed intraluminal strands that stained positive for both p-selectin and CD41a, confirming the presence of platelets. Despite the confirmed presence of platelets and thrombin in these clot strands, there was no detectable effect on patency in our preliminary analysis with follow-up CT angiography before hospital discharge. Only 1 of 44 evaluated grafts (2.3%) had early failure and was a graft that did not have clot strands detected during surgery. All of the SVG in this study cohort that were found to have clot were patent at this postoperative day 5 follow-up.
Effects of Saline Distention
Veins that were not distended showed a significantly higher percentage of endothelial integrity (60.1% ± 27.2% versus 24.7% ± 24.1%, P < 0.05) and lower tissue factor activity (1.28 ± 0.95 U/cm2 versus 12.3 ± 5.5 U/cm2, P < 0.001) than veins that were distended with saline during preparation for grafting.
Recent data show that endoscopic harvest of the saphenous vein is performed in the 65% of CABG cases in the United States (Guidant Corp., internal sales data). The rapid adoption and acceptance of this technology is the consequence of a dramatic reduction in leg wound complications.2,3 A single, randomized clinical trial comparing open versus endoscopically harvested SVG reported no difference in graft patency, implying that these benefits come without an apparent compromise in the quality of harvested conduits.4 However, one problem that appears to be unique to SVG that are procured endoscopically is the development of intraluminal clot strands. There is currently no convenient method for confirming the presence of these stands during surgery without destroying the conduit by opening up the vessel for direct gross assessment. These clot strands often come to the attention of the operative team after vigorous flushing of the SVG and observation of the clot in the venous effusate. This is not a sufficiently sensitive method to provide a meaningful assessment of the frequency, severity, or pathophysiologic importance of this issue. In this study, we imaged surplus vein segments by using OCT, a highly sensitive imaging technique,5 and established that clot strands are present in almost half of endoscopically harvested veins used as conduits for grafting. Our study was not adequately powered to address the pathophysiologic importance of these strands. It seems plausible there may be a risk of myocardial injury caused by distal embolization. In addition, the presence of thrombin and platelets raises concern that these clots may serve as a nidus for further thrombus formation within the newly grafted SVG.6,7
In the absence of anticoagulation, even a brief period of blood flow stasis is an established risk factor for clot formation within the venous system.8 To minimize bleeding during concurrent internal mammary artery harvest, heparin was not administered before initiating endoscopic harvest of the SVG in our subjects. The endoscopic procedure involves CO2 insufflation to a mean pressure of 8 to 10 mm Hg to create visualization of the surgical field.9 Using transit time flow readings of the in situ vein, we found that blood flow within the vein is stopped during the insufflation period, probably because of local compression within the tunnel by either the insufflation pressure or the trochar balloon, which presses on the vein at the knee. Although this compression is an advantage for minimizing venous bleeding during the harvest, the stasis of unheparinized blood within the SVG may have been an important factor in the development of clot strands in our cohort. A frequent practice is to give a small bolus of heparin (eg, 5000 U) before CO2 insufflation to prevent this stasis of unanticoagulated blood. However, the efficacy of a heparin dosing strategy that is not titrated by using point-of-care testing such as the ACT is uncertain. OCT scanning for clots may serve as an objective end point to determine the effectiveness of various anticoagulation strategies for this purpose. Alternative endoscopic vein harvest (EVH) systems, such as the one produced by Datascope Inc., do not use a trochar balloon or continuous insufflation and therefore may alleviate the stimulus for clot without needing to alter intraoperative anticoagulation.
A common response to clot strands has been vigorous saline flushing to remove the clot from the SVG. This technique was found to significantly decrease the incidence of residual clot strands in our cohort. However, there is an unintended trade-off of the intimal quality and thrombogenicity of the conduit. Veins frequently become distended during this saline flush, which compromises endothelial integrity, previously shown by our group to be an independent predictor of graft patency.1 Tissue factor activity, an established mediator of thrombosis,10 was also significantly increased within the luminal surface of SVG that were prepared using saline flush.11 The loss of endothelial integrity and subsequent exposure of subendothelium is the most likely cause of this increased tissue factor activity. Our data suggest that a less destructive way of eliminating these clots (eg, administration of ex vivo intraluminal fibrinolytics) is indicated to improve the overall quality of the SVG.
This study has several limitations. First, only the distal discarded segment of the vein was evaluated with OCT in this study. Although it seems reasonable to assume that the incidence of clot in discarded segments and grafted segments is similar, the only way to confirm this assumption is by scanning portions of vein that will be grafted. A study is underway that uses sterile probes to evaluate the entire length of conduit intraoperatively before grafting. Second, the incidence of clot that we detected in this study is higher than other assessments. It is possible that our estimate could have been exaggerated by ex vivo clot formation that occurred within the interval that the discarded vein segment was removed from the operative field until OCT imaging was performed. However, each vein segment was immediately flushed with heparinized saline to remove residual blood from the lumen while awaiting imaging, thereby minimizing the potential for clot formation during this interval. In contrast, the presence of static, unheparinized blood within the vein for periods of up to 40 minutes during EVH seems to be a far more likely stimulus for the observed intraluminal clots. Third, because vein harvesting by means of an open technique is unusual at our institution, we do not have an adequate control group of veins harvested with this traditional technique for comparison to the EVH group. Our suggestion that blood flow stasis from CO2 insufflation is a mechanism of this clot remains speculative in the absence of this control group. Last, because only one of the veins evaluated with OCT showed acute thrombosis, our study was insufficiently powered to relate clot strands with altered rates of early patency. Ongoing enrollment of patients will provide sufficient numbers to perform a multivariate analysis to determine the overall importance of clot strands in the context of other established risk factors for thrombosis. Although CT angiographic follow-up for our cohort is admittedly preliminary, excluding these data may provoke erroneous speculation that there is an extreme risk of thrombosis in SVG with clot strands; however, to date, our findings thus far do not support this concern.
In conclusion, clot within the SVG is common after endoscopic harvest. The current practice of aggressive saline flushing to remove these clot strands paradoxically increases the thrombogenicity of the graft and is therefore not recommended. Less traumatic means to prevent (earlier heparinization) or remove (local fibrinolytics) clot strands are worthy of investigation.
1. Manchio JV, Gu J, Romar L, et al. Disruption of graft endothelium correlates with early failure after off-pump coronary artery bypass surgery. Ann Thorac Surg. 2005;79:1991–1998.
2. Bonde P, Graham AN, MacGowan SW, et al. Endoscopic vein harvest: advantages and limitations. Ann Thorac Surg. 2004;77: 2076–2082.
3. Schurr UP, Lachat ML, Reuthebuch O, et al. Endoscopic saphenous vein harvesting for CABG: a randomized, prospective trial. Thorac Cardiovasc Surg. 2002;50: 160–163.
4. Yun KL, Wu Y, Aharonian VJ, et al. Randomized trial of endoscopic versus open vein harvest for coronary artery bypass grafting: six-month patency rates. Thorac Cardiovasc Surg. 2005;129:496–503.
5. Saia F, Schaar J, Regar E, et al. Clinical imaging of the vulnerable plaque in the coronary arteries: new intracoronary diagnostic methods. J Cardiovasc Med. 2006;7:21–28.
6. Poston RS, Gu J, Brown JM, et al. Endothelial injury and acquired aspirin resistance as promoters of regional thrombin formation and early vein graft failure after coronary artery bypass grafting. J Thorac Cardiovasc Surg. 2006;131:122–130.
7. Dale GL. Coated-platelets: an emerging component of the procoagulant response. J Thromb Haemost. 2005;3:2185–2192.
8. Morris RJ, Woodcock JP. Evidence-based compression: prevention of stasis and deep vein thrombosis. Ann Surg. 2004;239:162–171.
9. Morris RJ, Butler MT, Samuels LE. Minimally invasive saphenous vein harvesting. Ann Thorac Surg. 1998;66:1026–1028.
10. Muluk SC, Vorp DA, Severyn DA, et al. Enhancement of tissue factor expression by vein segments exposed to coronary arterial hemodynamics. J Vasc Surg. 1998;27:521–527.
11. Poston RS, Kwon MH, Gu J. Role of procurement-related injury in early saphenous vein graft failure after coronary artery bypass surgery. Future Cardiology. 2006;2:503–512.
© 2006 Lippincott Williams & Wilkins, Inc.