|Home | About | Journals | Submit | Contact Us | Français|
Successful adaptation of a vein graft to an arterial environment is incompletely understood. We sought to investigate whether early vein graft remodeling is predictive of subsequent patency.
A prospective longitudinal study of 67 patients undergoing lower extremity bypass with autologous vein between February 2004 and April 2008. Preoperative blood samples were drawn for biomarkers. During the bypass operation, a 5 cm index segment of the graft was registered for serial lumen diameter measurements (0, 1, 3, 6, 9, and 12 months) using duplex ultrasound. Patients with at least 2 study ultrasounds were included in this analysis.
The median age was 70 years (IQR 59-76y) and the median follow-up was 32 months (IQR 15-47mo). Over half (55%) of the subjects were male, 40% had diabetes mellitus, 49% had critical limb ischemia, and most were on a statin (75%) and antiplatelet medication (91%). The median baseline high-sensitivity C-reactive protein level (hsCRP) for the cohort was 3.2mg/L (IQR 1.4-9.7mg/L). The average intraoperative, post-implantation vein lumen diameter was 3.9 ± 1.0mm, increasing to 4.7 ± 1.1mm at 1 month, an average 24 ± 27% change per subject. By 3 months, the average lumen diameter was 5.1 ± 1.6mm, with little subsequent change observed to 12 months. Non-white race, baseline hsCRP ≥5mg/L, statin use and initial lumen diameter were significantly associated with early (0-1 month) vein remodeling in a multivariable regression model. The primary patency rate for the cohort was 60 ±6.3% at 2 years. Initial lumen diameter of the index segment was not associated with primary patency, whereas larger lumen diameter achieved at 1 month (≥5.1mm) was positively associated with primary patency (P=.03, log-rank). Early (30 day) remodeling behavior was used to divide subjects into “poor remodelers” (< −5% lumen diameter change, N=6), “modest remodelers” (−5 to +25% change, N=29) and “robust remodelers” (>+25% change, N=30). Early remodeling category was significantly associated with primary patency rate at 2 years P=.02, log-rank). A multi-variable Cox proportional hazards model showed that modest (HR 3.9, 95% CI 1.02-15, P=.04) and poor (HR 13, 95% CI 1.9-89, P=.008) remodelers had significantly higher hazard ratios for graft failure than robust early remodelers.
Early remodeling of the arterialized vein appears to predict mid-term bypass graft patency. In addition to baseline diameter, race, inflammation, CRP, and statin use are associated with early adaptive remodeling, but the mechanism for these observations are not understood.
Vein bypass grafts are an effective surgical intervention for patients with advanced peripheral and coronary artery occlusive disease. However, vein graft disease is common, causing clinical failure in over a third of patients undergoing lower extremity bypass surgery within five years.1-5 Most of these failures are secondary to the development of de novo lesions within the first 2 years of implantation. Despite decades of research, our understanding of the pathophysiology of vein graft failure remains limited. The adaptation of veins to the arterial environment induces requisite changes in wall structure in response to biomechanical forces, namely wall tension and shear stress.6, 7 The factors governing the variability in this remodeling response, both between individual patients and spatially within a vein, are likely to be important predictors and treatment targets for vein graft disease.
In prior reports, we described early remodeling changes in lower extremity vein grafts utilizing serial, high resolution ultrasound imaging of a defined region of the graft beginning from the time of implantation.8-10 We documented substantial variability in the early remodeling response, particularly within 30 days of surgery and a correlation between early remodeling and biomarkers of systemic inflammation, particularly high-sensitivity C-reactive protein (hsCRP).10 The mechanisms driving this early post-implantation response, and the association between early venous remodeling and long-term clinical outcomes, remained unclear.
We sought to further examine these questions by integrating imaging and clinical outcomes data from a well-characterized prospective cohort of patients who underwent lower extremity vein bypass grafting at a single institution. The primary objectives of the study were to determine significant baseline factors associated with vein graft remodeling, and to determine if remodeling changes observed in a discrete, pre-specified region of the venous conduit were associated with graft patency.
This report focuses on a prospective imaging substudy nested within a larger prospective cohort study investigating markers of systemic inflammation and clinical outcomes in patients undergoing lower extremity bypass surgery with autogenous vein. The inclusion/exclusion criteria of this cohort have been described elsewhere,10, 11 but briefly, patients undergoing primary or redo lower extremity bypass surgery solely with autogenous vein for lifestyle-limiting claudication or critical limb ischemia were eligible for enrollment. Importantly, patients were excluded if, in the 30 days preceding bypass surgery, they had a condition effecting systemic inflammation including myocardial infarction, stroke, major illness, major operation or evidence of a severe foot infection (i.e. deep space infection, need for operative debridement, etc.).
The imaging substudy was prospectively designed to investigate the patterns of vein graft remodeling and the relationship between structural changes in the vein graft, markers of systemic inflammation, and clinical outcomes. All subjects in the parent study were eligible. Study ultrasounds were performed in the operating room and at 1, 3, 6, 9 and 12 months postoperatively (see below). The protocol was designed to study remodeling using repeated measurements for each subject, thus only patients with two or more study ultrasounds were included in this analysis.
Enrollment began in February 2004 and the follow-up closed in December 2008. All patients provided written informed consent prior to enrollment. The study protocol was approved by the Institutional Review Board of the Brigham and Women’s Hospital.
A preoperative blood sample was drawn to assess inflammatory markers as previously described.10, 11 Blood was collected in EDTA and citrate vacutainer tubes and immediately iced. Tubes were spun at 3000 rpm for 20 minutes at 4C and baseline plasma samples were stored at −80C. The hsCRP measurements were performed as a batch in a core facility using a validated, high sensitivity assay.
The vein graft configuration and performance of a completion angiogram or duplex ultrasound was left to the discretion of the attending surgeon. In the majority of cases, veins were tunneled superficially for ease of graft surveillance.
Identification and ultrasound imaging of the index segment has been described previously.9, 10 Briefly, after completion of the proximal and distal anastomoses, a 5cm, superficial, straight, valveless segment of the vein bypass graft was identified and registered as the index segment. Surgical clips were left as a reference and the distance from the proximal anastomosis was recorded for subsequent identification.
Five high-resolution M-mode cross-sectional systolic lumen diameter images were recorded at 1cm intervals along the index segment using an ATL HDI 3000 ultrasound machine (Advanced Technology Laboratories, Bothell, Washington) with a 10-MHz transducer and cardiac-gating. The “index segment diameter” is the average of these 25 measurements. For volume flow and shear stress calculations (Hagen-Pouseille equation), the Doppler velocity spectral waveform from a sagittal view of the centerline of the index segment was captured over several cardiac cycles as previously described.9,10 Postoperative imaging of the index segment occurred at 1, 3, 6, 9 and 12 months using the same protocol. All imaging analysis was performed using the Brachial Analyzer (Medical Imaging Applications, Coralville, Iowa) software package.
All patients were followed for clinical and graft-related events with clinical surveillance duplex ultrasounds 1, 3, 6, 9, 12, and every 6 months thereafter. All patients in the study were followed for a minimum of one year after the index procedure. Overall, 96% of subjects had ≥2 clinical ultrasounds, 37% had >5 ultrasounds and 6% had >10 ultrasounds.
Clinical vein graft surveillance included measurement of ankle-brachial index and a standard DUS scan with peak systolic velocity mapping performed in an Intersocietal Commission for the Accreditation of Vascular Laboratories (ICAVL) approved vascular laboratory. Clinical or graft-related events were recorded during the postoperative visits, including hospitalizations, major adverse cardiovascular events, amputations, graft revisions, or graft occlusions. Grafts that were intervened upon or occluded (i.e. lost primary patency) were removed from the imaging substudy. Primary, primary-assisted, and secondary graft patency were defined in accordance with accepted SVS guidelines.12
All data are reported as the mean ± standard deviation (SD) or median with interquartile range (IQR) depending on the distribution. To examine changes in average lumen diameter over time, we used a random-effects model to account for multiple diameter measurements per subject. This allowed for adjustment of the variance to include both within- and between-subject variability. Our primary fixed predictor was time at which the measurement was made with subject as the random effect. Missing data were treated as “Missing at Random.” Since lumen diameter had a non-linear relationship with time, linear-splines with knots at 1, 3 and 6 months were used to the fit of the model. Spline models account for departures from linearity by allowing for a regression line that is continuous at each of the knots and linear in the intervals between them.
Linear regression models were used to model change in lumen diameter from 0-1 month, reducing our analysis from multiple to one measurement per subject. Baseline variables significant in univariable analysis at the p<.25 level were included in multivariable regression models, using backward selection to derive a final model. Of note, hsCRP was modeled as both a continuous and dichotomous variable (<5mg/L vs ≥5mg/L) as previously reported.10,11,24
Patients were categorized into remodeling groups based on the extent of 0-1 month remodeling. Comparisons of characteristics between types of remodelers were made using ANOVA for normally distributed variables, and the Kruskal-Wallis test for variables that were not normally distributed. Post-hoc pairwise testing was done using the t-test after ANOVA, and Wilcoxon rank-sum test after Kruskal-Wallis. No adjustments were made for multiple comparisons. Kaplan-Meier techniques were used to compare median time to loss of patency across these three groups, using the log-rank rest. Multivariable Cox proportional hazard models for graft patency were created using statistically significant (p<.15) factors identified on univariable analysis with the addition critical limb ischemia and initial lumen diameter, two well-known predictors of lower extremity vein graft primary patency. All data were analyzed with STATA/SE version 11 (StataCorp, College Station, Texas).
Overall, 103 patients consented to participate in this prospective imaging substudy. Thirty-six were excluded from this analysis because they had less than 2 complete study ultrasounds. The 67 subjects in this analysis had at least two study DUS measuring the vein graft index segment, the first of which was intra-operative. The mean number of study ultrasounds per subject was 3.7 ±1.4. The median clinical follow-up time was 32 months (IQR 15-47mo).
The subjects in this cohort were notable for a median age of 70 years (IQR 59-76y), 45% were women, 40% were diabetic, 75% were taking a statin, 49% had a bypass for critical limb ischemia (CLI) and 81% had a single segment greater saphenous vein graft conduit (Table I). As reported previously,10,11 there was a significant positive correlation between hsCRP level and CLI, r=.37 P=.002.
The mean initial (intraoperative, post-implantation) diameter of the vein graft index segment was 3.9 ±1.0mm and increased to 4.7 ±1.1mm by one month (Table II). The average percent change in lumen diameter was +24 ±27% from 0-1 month, falling to +8 ±19% from 1-3 months and near 0% from 3-12 months (Figure 1). On average, 69% of the average total diameter increase occurred within the first postoperative month. A striking observation from these measurements is that very little average change in graft lumen diameter was evident beyond 3 months.
Changes in index segment lumen diameter over the course of the study appeared to be associated with baseline patient factors, most notably race and hsCRP value. Whereas average initial lumen diameter was not significantly different by race (4.0 ±.98mm white vs. 3.4 ±1.0mm non-white, P=.07), by 1 month non-white racial status was associated with a significantly smaller index segment lumen diameter (4.9 ±1.0mm white vs. 3.8 ±1.0mm non-white, P=.006). Using a random-effects model, non-white race was associated with a .9mm (95% CI −1.5 to −.34mm, P=.002) smaller graft diameter on average over the 12-month imaging study period (Figure 2A). Similarly, although baseline hsCRP level was not associated with initial graft diameter, over the 12 months of follow-up subjects with baseline hsCRP ≥5mg/L had graft lumen diameters that were on average .5mm smaller (95% CI −.98 to −.012mm, P=.045, Figure 2B).
Similar to previous observations, we found that the calculated mean shear stress in these lower extremity vein grafts decreased over the first year, from 23 ±21dynes/cm2 at the time of surgery to 12 ±5.2dynes/cm2 at one year, similar to the reported shear stress in normal superficial femoral arteries.13, 14
Because both the greatest absolute change and variability in lumen diameter occurred within the first postoperative month, we focused our analysis on this critical period of venous adaptation. Of the 67 subjects, 55 had a 1 month study DUS, while the remaining 12 had their first study DUS at 3 months or later. On univariate analysis, sex, critical limb ischemia, non-white race, baseline hsCRP ≥5mg/L, statin use and initial diameter were significantly associated with 0-1 month remodeling. In a multivariable model, only non-white race, baseline hsCRP ≥5mg/L and initial diameter remained significant predictors of early remodeling (Table III). Of note, when initial calculated shear stress is substituted in the model for initial vein diameter (with which it is intrinsically related), it also remains significant, as reported previously.10
Both intraoperative, baseline lumen diameter (P=.01) and the change in diameter from 0-1 month (P=.01) were significant positive predictors of the diameter achieved at 1 month. Additional factors associated with index segment diameter attained at 1 month included sex, race, diabetes mellitus, statin use and baseline hsCRP ≥5mg/L. Significant factors in a multivariable model were initial lumen diameter (P<.0005), non-white race (P=.02), and baseline hsCRP ≥5mg/L (P=.04).
To further investigate the potential relationship between lumen remodeling and inflammation, we asked the question of what happens to relatively smaller veins when they are used for bypass grafts. We examined a subset of 19 subjects with an initial lumen diameter <3.5mm (range 2.3-3.4mm, average 2.9 ±.34mm). By 1 month, these veins dilated on average +47 ±29%, a significantly greater proportional change than their larger counterparts (P=.001), achieving a mean lumen diameter of 4.1 ±.8mm. However, if the initial post-implant vein diameter was <3.5mm in a patient with an elevated hsCRP (≥5mg/L; N=6), the remodeling from 0-1 month was reduced to +25 ±12%, P=.01.
In this cohort, 25 patients lost primary patency during follow up. The primary patency rate was 68 ±5.8% at one year and 60 ±6.3% at two years. Importantly, 10 grafts (40%) lost primary patency due to a perianastomotic stenosis, 11 (44%) had a mid-graft stenosis at a location other than the index segment and 4 grafts occluded without identification of a culprit lesion.
Initial, post-implantation index segment lumen diameter was not associated with primary patency. Specifically, the baseline index segment diameter of vein grafts that eventually failed was not significantly different from those that remained patent, (4.0 ±1.1mm vs. 3.8 ±.80mm, P=.33). However, when modeled over the 12 month imaging study period, the average diameter of grafts that eventually lost primary patency was .5mm smaller (95% CI −1.0 to −.052mm, P=.03, Figure 3). In contrast, the index segment lumen diameter achieved at 1 month, when modeled as a categorical variable (tertiles: <4.2mm, 4.2-5.1mm or >5.1mm), was significantly associated with primary patency rate at 2 years (P=.03, log-rank). Thus there was a significant relationship observed between the lumen diameter changes within the index segment and clinical outcome, even though the index segment itself was never the site of a critical lesion.
To further explore the relationship between venous remodeling and patency, subjects were categorized based on clustering of the change in index segment lumen diameter from 0-1 month as “poor remodelers” (<-5% change in lumen diameter, N=6), “modest remodelers” (−5% to +25% change, N=29) or “robust remodelers” (>25% change, N=20) (Table IV). Poor remodelers tended to have higher hsCRP values and were more likely to have critical limb ischemia than either modest or neutral remodelers, although these differences were not significant. Remodeling category was significantly associated with primary patency rate at 2 years (P=.02, log-rank; Figure 4). A multivariable Cox proportional hazards analysis showed that the modest (HR 3.9, 95% CI 1.02-15, P=.04) and poor (HR 13, 95% CI 1.9-89, P=.008) remodelers had significantly higher hazard ratios for graft failure than robust early remodelers (Table V). In contrast, when a similar analysis was applied to changes in lumen diameter observed over later (1-3mo or after 3mo) time intervals, there were no significant associations with primary patency. Early remodeling was also not associated with primary-assisted patency (86 ±4.2% at 1y) or secondary patency (89 ±3.8% at 1y) rates.
Since initial vein graft diameter is an established predictor of clinical outcome for lower extremity bypass surgery, we examined the relationship between systemic inflammation and patency for initially smaller grafts (<3.5mm). In patients with a low baseline hsCRP (<5mg/L) there was no difference in the observed 2-year primary patency rates between grafts having an initial index segment diameter <3.5 vs ≥3.5mm (P=.97 log-rank). However, the primary patency rate of smaller grafts in subjects with a baseline hsCRP ≥5mg/L (33 ±19%) tended to be lower than the primary patency rates of smaller grafts with a baseline hsCRP <5mg/L (67 ±14%, P=.25) or larger grafts with a baseline hsCRP ≥5mg/L (54 ±12%, P=.46).
In this report we update and extend prior work describing lumen diameter changes in lower extremity vein grafts over time,8-10 and their relationship to patient factors and clinical outcomes. Lumen remodeling is largely complete by 3 months, with most of the caliber change observed within 30 days. Early venous remodeling is strongly associated with hemodynamic forces particularly shear stress, but also with patient-specific factors including race and systemic inflammation. This study, for the first time, demonstrates that early vein remodeling is associated with subsequent graft patency, suggesting that factors which globally influence vein adaptation may also modulate the incidence and/or progression of critical lesions. Stated another way, although the pathology of vein graft failure is almost always segmental, the systemic and regional milieu of the entire conduit appears to be a critical element of the healing response and the clinical outcome.
The relationship between initial vein diameter and patency of lower extremity bypass grafts has been established in a number of studies.15-17 The fundamental differences between previous reports and this study must be highlighted for correct interpretation of our findings. In most of the prior studies, such as Schanzer et al,17 the reported conduit diameter was an estimate of the minimal lumen diameter of the entire vein at the time of surgery. In other reports, initial vein graft imaging was performed up to one week following bypass surgery.7 In our study, we deliberately identified an “ideal” index segment of the graft intraoperatively for serial imaging. To facilitate later imaging, the segment was selected because it was a uniform diameter, valveless, without major branches, and at some distance from either anastomosis, without regard for the absolute minimum or maximum graft diameter. Furthermore, the selected cohort reported here may not be representative of a broader surgical series, particularly in relation to the use of higher risk conduits. Thus our observations on the relative importance of initial versus subsequent vein diameter must be appreciated within this context, and do not represent a departure from the well-established criteria by which surgeons have judged vein quality intraoperatively. Instead, the results highlight that early postoperative remodeling is distinct from initial size, and appears to correlate with graft patency. For example, a small vein that undergoes robust remodeling would be predicted to have better patency than a larger vein that fails to remodel.
These results extend and validate our prior findings by focusing on a carefully characterized subset of patients with multiple ultrasound measurements and detailed clinical follow-up for a median of nearly 3 years. The factors associated with early graft remodeling have remained consistent: initial lumen diameter/shear stress, baseline hsCRP, race, and statin use. The mechanisms underlying these observations remain to be defined, however we conjecture that endothelial dysfunction and vascular inflammation are the underlying key modifiers that influence biomechanical adaptation of the grafted vein. Ongoing and future studies hope to directly address these questions.
Systemic inflammation, represented here by plasma hsCRP level, has been demonstrated as a risk factor for incident cardiovascular disease and related clinical events.18-21 The relationship between inflammatory biomarkers and remodeling following vascular interventions has been less well studied,22-24 but we previously described an association between elevated hsCRP and outcomes of lower extremity bypass surgery.11, 24 Multiple studies have also observed a negative association between non-white racial status and vein graft patency.25-27 This study suggests that a common link may be found in the early postoperative remodeling response of the venous conduit. Our statistical analysis suggests that these factors have independent significance, however it should be noted that African-Americans, for example, have higher hsCRP levels across a number of large cohort studies, even when controlling for demographic and comorbid factors.28-31 We hypothesize that endothelial (dys-)function may be the common mechanism behind these associations, but additional research is required in this area.
Collectively these data suggest a conceptual evolution regarding the pathophysiology of vein graft failure. The observation that early caliber changes in a non-diseased section of the conduit—i.e. an “imaging biopsy”—correlate with the incidence of critical lesions elsewhere in the graft, implies that purely local factors (e.g. technical issues, surgical injury, local geometry) are only part of the story. This is further accented by the importance of patient-level factors such as race and hsCRP level, which have persisted in this and other studies. Thus we conjecture that systemic factors (genetic, biochemical, immunologic) and regional factors (hemodynamics) influence healing of the entire venous conduit and act as an important overlay to local determinants in graft lesion development. At present, the clinical relevance of these findings is limited by the modest size and single-center nature of the cohort. One intriguing hypothesis is that therapeutic modulation of inflammation may improve early remodeling and clinical outcomes of vein grafts; this requires formal testing in a prospective trial design.
In addition to sample size, our study is limited by a relative lack of racial and ethnic diversity, a mixture of indications and bypass configurations, as well as incomplete compliance with the longitudinal imaging protocol. Due to a variety of factors including clinical events, missed visits, and inadequate quality of some individual scans many patients did not have measurements available at all of the predetermined time points. Limitations of ultrasound in resolving the outer border of the graft limit our ability to characterize the structural changes in the graft wall i.e. neointima formation, wall composition, and mechanical compliance. Other imaging modalities32-34 may be more useful in this regard.
In conclusion, early remodeling of lower extremity vein bypass grafts is associated with hemodynamics, systemic inflammation, and patient-level risk factors such as race. Vein grafts that remodel poorly within the first month appear to be at increased risk for failure. Further investigation is needed to understand the mechanisms, and define potential clinical relevance for these findings.
We would like to thank Marie D. Gerhard-Herman, MD and Nicole Wake, BS for their technical assistance.
This study was supported by funding from the National Heart, Lung and Blood Institute (R01 HL 75771; CDO, MAC, MSC).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Presented as a Rapid Paced Paper at the 65th Annual Vascular Annual Meeting, Chicago, IL, June 16-18, 2011.