Symmetry™ Aortic Connector Devices and Acute Renal Injury: A Comparison of Renal Dysfunction After Three Different Aortocoronary Bypass Surgery Techniques

Abstract

Although the pathogenesis of acute renal injury after cardiac surgery is multifactorial, atherosclerosis of the ascending aorta and embolic burden are strong independent predictors. Use of the Symmetry™ aortic connector device (ACD) for proximal anastomosis of coronary grafts may reduce ascending aortic atheroembolism. Therefore, we tested the hypothesis that off-pump coronary artery bypass (OPCAB) surgery performed using an ACD is associated with less postoperative renal dysfunction compared with conventional OPCAB or on-pump coronary artery bypass graft (CABG) surgery. Three-thousand-three-hundred consecutive patients undergoing non-emergent aortocoronary bypass surgery were retrospectively divided into three groups by surgical procedure; Group A: OPCAB with ACD (n = 124), Group B: standard OPCAB (n = 313), Group C: on-pump CABG (n = 2863). Postoperative peak fractional change in creatinine compared with baseline was used as a measure of renal outcome. Multivariable analysis did not identify ACD use as an independent predictor of postoperative peak fractional change in creatinine (P = 0.71), although the relationships of several known renal risk factors with postoperative peak fractional change in creatinine were confirmed. We could not find evidence that OPCAB surgery using ACDs reduces acute renal injury compared with standard OPCAB or CABG surgery.

In Brief

IMPLICATIONS: We did not find that Symmetry™ aortic connector devices, as part of off-pump coronary artery bypass surgery, reduce renal injury. Similarly, we found no benefit of off-pump coronary artery bypass with an aortic connector device over conventional on-pump surgery. These data suggest that reduction of renal risk alone should not be an indication for aortic connector devices in off-pump coronary artery bypass surgery.

Postoperative renal dysfunction is a significant complication of cardiac surgery, associated with major increases in morbidity, mortality, and cost (^1–3). The pathogenesis is multifactorial (⁴), and risk factors include procedural characteristics and comorbidities such as ascending aortic atherosclerosis (⁵). Minimally invasive surgery techniques have become increasingly common in coronary artery bypass surgery; a proposed advantage of these procedures is reduced renal insult and improved overall outcome. However, retrospective multivariable analyses to assess renal outcome (^6,7) that account for known renal risk factors have not demonstrated that avoidance of cardiopulmonary bypass (CPB) reduces postoperative renal injury. Although alternate statistical approaches to retrospective analysis have questioned this conclusion (^8–12), a recent meta-analysis of randomized trials by Cheng et al. (¹³) combining findings of more than 1400 patients from 10 studies did not confirm an advantage of off-pump surgery with regard to renal outcome.

Clamp occlusion as well as cannulation of the ascending aorta are recognized as possible sources of particulate embolization and atheroembolic load to the kidneys; embolic burden is directly associated with postcardiac surgery renal dysfunction (¹⁴). Some interpretations of minimally invasive cardiac surgery have included a “minimal touch” approach to the ascending aorta (^15–18). The Symmetry™ aortic connector device system (ACD) includes a nickel-titanium connector (Fig. 1), an aortic cutter (to create an aortotomy), and a delivery device to facilitate implantation of the saphenous vein graft onto the aortic wall. The Symmetry™ ACD was developed to permit proximal coronary graft anastomosis with less manipulation of the aorta and may be a method by which the frequency of embolization and organ dysfunction can be reduced (^16,19). However, no studies have been performed to specifically test for an association of favorable renal outcome with the use of the ACD. Therefore, we tested the hypothesis that off-pump coronary artery bypass (OPCAB) surgery with the use of the Symmetry™ ACD is associated with reduced postoperative acute renal injury compared with standard OPCAB surgery and/or conventional on-pump coronary artery bypass graft (CABG) surgery.

Figure 1.: The appearance of the Symmetry™ device ex vivo (A), on fluoroscopy viewed perpendicularly to the aorta (B), and in an idealized depiction (C). The hooks attach to the saphenous vein graft (SVG) and the nickel-titanium struts anchor the device on the inner and outer walls of the aorta. Used with permission from Cavendish et al., J Am Coll Cardiol 2004;43:133–9.

Methods

After IRB approval, demographic, intraoperative, and renal data were obtained from the prospectively gathered Duke Cardiothoracic Surgery database for all primary elective aortocoronary surgery patients between June 14, 1999 and October 10, 2003.

Exclusion criteria included preoperative renal dysfunction requiring dialysis (n = 31), single vessel bypass surgery (n = 155), or conversion from off-pump to on-pump surgery (n = 30). Patients with postoperative acute renal failure requiring dialysis (n = 20) and patients who died within 2 days postoperatively (n = 5) were also excluded from the analysis, as their serum creatinine values did not accurately reflect perioperative renal injury, i.e., inadequate time for creatinine to increase.

Detailed patient and procedural information was gathered with reference to factors that were previously reported to be independently associated with postcardiac surgery renal dysfunction (^1–3), including age, gender, weight, African-American ethnicity, history of hypertension, diabetes, baseline renal dysfunction (preoperative serum creatinine [CrPre] >1.4 mg/dL or 133 μmol/L), obstructive lung disease, history of stroke or transient ischemic attacks, carotid bruit on physical examination, history of congestive heart failure (defined as history of paroxysmal or nocturnal dyspnea responsive to afterload reduction drug therapy), previous myocardial infarction, ASA physical status, preoperative ejection fraction, and unstable angina. Furthermore, procedural variables were gathered, including CPB duration, aortic occlusion duration, number of aortocoronary grafts, intraaortic balloon pump (IABP) use, inotrope use, and transfusion requirements. Date of surgery was also recorded for inclusion in the overall analysis.

Aortocoronary surgery was performed with CPB (CABG) or without CPB (OPCAB). Anesthesia was managed per the attending anesthesiologist’s preference and typically included IV propofol for induction, pancuronium or vecuronium bromide for muscle relaxation, and fentanyl and isoflurane for anesthesia maintenance. Use of drugs with potential diuretic effects (e.g., IV dopamine, furosemide) was not regulated; however, 12.5 g of 20% mannitol solution (Abbott Laboratories, North Chicago, IL) was routinely added to the CPB circuit priming solution. A standard epsilon-aminocaproic acid antifibrinolytic regimen was used for all CABG procedures during the study period, including a 10 g IV bolus before CPB immediately followed by 1 g/h IV infusion for 5 h.

OPCAB procedures included administration of an initial loading dose of 150 U per kg of heparin before coronary artery grafting to maintain a target activated clotting time (ACT) test of 300–400 s, use of a myocardial stabilization device, and placement of either a “sidebiter” clamp on the ascending aorta or use of the ACD for attachment of proximal vein grafts. Autotransfusion technology (Hemonetics Cell Saver, Braintree, MA) to return washed blood was used for all procedures, except for on-pump CABG procedures when blood loss exceeded 1500 mL, at which point blood salvage would change to direct return of heparin anticoagulated blood to the cardiotomy reservoir and CPB circuit. In all cases where an ACD was used, this was a Symmetry™ Bypass System Aortic Connector Device (St. Jude Medical, St. Paul, MN) (²⁰); this technology involves punching the aorta and deploying a saphenous vein graft loaded onto the ACD, thus creating a sutureless anastomosis of the vein graft to the proximal aorta with minimal aortic manipulation. ACD were used from July 2001 onwards. Patient selection for OPCAB with or without the use of the ACD was determined by the cardiothoracic surgeons. In our department, ACD were routinely used during OPCAB surgery by only 2 of the 10 cardiothoracic surgeons. Postoperative care was standardized for all aortocoronary bypass surgery patients, using an institutionally developed post-coronary bypass surgery care-map, with the goal of discharging all patients on postoperative day 4.

As part of CABG surgery procedures, CPB was performed with a Cobe Duo blood oxygenator with sealed hard-shell filtered venous reservoir (Cobe Laboratories, Lakewood, CO), a Cobe Century Perfusion System, and a heparin-coated 43 μm arterial line filter; (Cobe Sentry™, Cobe Cardiovascular, Inc., Arvada, CO). Nonpulsatile perfusion was maintained at 2–2.4 L/min/m². The CPB circuit was primed with 12.5 g of mannitol, crystalloid solution (Normosol R), and packed red blood cells, if required, to achieve a hematocrit of 20% or more during CPB. All patients were perfused during CPB through an ascending aortic cannula. The arterial carbon dioxide tension was maintained throughout CPB at 35–45 mm Hg (uncorrected for temperature), with the arterial oxygen tension maintained at 150–250 mm Hg. The mean arterial blood pressure was maintained between 50 mm Hg and 90 mm Hg during CPB by using increments of IV phenylephrine or sodium nitroprusside as required. Patients were cooled to a nasopharyngeal temperature of 34ºC to 28ºC during aortic cross-clamp and rewarmed to a nasopharyngeal temperature of 37ºC or a bladder temperature of at least 36ºC before separation from CPB. Heparin was administered before CPB for anticoagulation to a target ACT test of more than 480 s. Proximal vein grafting was by surgeon preference using a “single clamp” technique before removal of the aortic cross-clamp or placement of a partial occlusion clamp after cross-clamp removal.

Serum creatinine was measured as part of routine biochemical laboratory investigations for all patients using a dry slide enzymatic reflectance technique (VITROS 950; Johnson & Johnson, New Brunswick, NJ) with a range between 62 to 124 μmol/L considered as normal.

Renal function variables and choice of markers used in this study have been previously defined (²¹). Briefly, the creatinine data were assessed for the study from a quality assurance database. Baseline CrPre was defined as the value recorded closest to but not on the day of surgery. Peak in-hospital postoperative serum creatinine (CrmaxPost) was the highest of the daily in-hospital postoperative values. Peak fractional change in creatinine was the primary marker of renal filtration impairment and was defined as the difference between CrPre and CrmaxPost represented as a percentage of the preoperative value (%ΔCr). Creatinine clearance (CrCl) values were derived from CrePre and CrPost values using the Cockcroft Gault equation (²²). The perioperative change in CrCl (ΔCrCl) was defined as the difference between lowest postoperative CrCl (CrClPost) and preoperative (CrClPre) (ΔCrCl = CrClPre − CrClPost).

Patients were divided into 3 groups based on the surgical procedure, group A: OPCAB surgery with the use of the ACD, group B: standard OPCAB surgery (without the use of the ACD), and group C: on-pump CABG surgery. Univariate comparisons of demographic variables among the three groups were made with one-way analysis of variance for continuous variables and chi-square tests for categorical variables. Multivariable linear regression analysis was performed to determine the association of the type of surgery performed with %ΔCr.

Demographic variables previously reported to be renal risk factors (²) and date of surgery were included as covariates in the analyses. Significant variables were retained in the model. Because of the potential for non-normal distribution of serum creatinine values and to assess generalizability and robustness of results, these analyses were then repeated on ranked data; P value <0.05 was considered significant. All statistical analyses were performed using SAS statistical software version 8.0 (SAS Institute Inc., Cary, NC).

Results

A total of 3300 patients were eligible for analysis and assigned to groups according to surgical procedure: group A (n = 124), group B (n = 313), and group C (n = 2863).

Several patient and procedural characteristics were unevenly distributed among the groups (Table 1), with ACD patients having more frequent unstable angina and several other comorbid conditions. OPCAB patients had the fewest number of coronary grafts, whereas CABG patients received more transfusions.

Table 1: Patient and Procedural Characteristics

Among the three groups, there were no statistically significant differences in univariate comparisons of baseline and peak impairments of renal filtration function, reflected by serum creatinine, and estimated creatinine clearance values (Table 2, Fig. 2).

Table 2: Renal Function Variables

Figure 2.: Comparison of peak postoperative serum creatinine increase relative to baseline (%ΔCr) among three different coronary artery bypass graft surgery types. OPCAB with ACD, off-pump coronary artery bypass surgery with use of aortic connector device; OPCAB, standard off-pump coronary artery bypass surgery; CABG, on-pump coronary artery bypass graft surgery. There were seven on-pump patients above 400%ΔCr that are not depicted in this figure.

A multivariable analysis including surgical procedure type and accounting for patient and procedural renal risk factors and other variables with a univariate P value <0.05, also identified no relationship between ACD use and postoperative renal dysfunction (%ΔCr, P = 0.71, Table 3).

Table 3: Linear Regression Multivariable Analysis of Variables Associated with Preoperative to Maximal Postoperative Increase in Serum Creatinine (%ΔCr) After Three Different Coronary Artery Bypass Surgery Procedures

However, we identified independent associations of several previously known renal risk factors including age, weight, African-American ethnicity, baseline serum creatinine, history of diabetes, history of congestive heart failure, and perioperative intraaortic balloon counterpulsation (Table 3). An assessment of ranked renal data demonstrated similar results.

Dialysis-dependent patients were not included in the primary analysis because these individuals have truncated postoperative creatinine increases; we believe this would have a misleading effect in our analysis, which is dependent on assumptions of linearity. However, not including these patients in the primary analysis is a potential source of bias, particularly if dialysis patients are not equally distributed among the groups. Incidence rates of dialysis are presented in Table 2 to address concerns of potential for bias. In addition, findings from a secondary analysis including all postoperative dialysis patients support the findings of the primary analysis; no significant difference was detected among the 3 groups (P = 0.21).

We conducted a post hoc power analysis to assess what difference in renal function would be likely to be missed by our study. Given our sample size and an α error of 0.05, we could detect an effect size of 0.0029. For example, our sample size is large enough to have 80% power to identify a significant difference if the connector group had a %ΔCr of 24.0%, and the other 2 groups had %ΔCr of 24.5%. These findings indicate that our study should detect anything but relatively small differences in the magnitude of postoperative acute renal injury.

Finally, to address the role of baseline renal dysfunction, we conducted a secondary multivariable analysis in the subset of high risk patients with a CrPre level ≥133 μmol/L (≥1.5 mg/dL). We also identified no relationship between ACD use and postoperative renal dysfunction in the subgroup of patients with baseline renal dysfunction (%ΔCr, P = 0.12). The sample sizes for this analysis were 18, 39, and 407 for connector OPCAB, OPCAB, and CABG groups, respectively.

Discussion

In this multivariable analysis of 3300 consecutive primary non-emergent coronary bypass patients, we did not find that use of the Symmetry™ Aortic Connector System confers reduced renal risk compared with either standard OPCAB surgery or conventional CABG surgery. In addition, we did not see an advantage of OPCAB surgery over conventional on-pump CABG surgery with regard to renal risk, a finding that is consistent with conclusions from our previous study in a separate dataset (⁶). However, we confirmed significant independent associations of several previously identified renal risk factors (¹) with postoperative renal impairment in the study group, including increased CrPre, increasing age, diabetes mellitus, increased body mass, and use of IABP.

There has been speculation that the use of ACDs as part of off-pump procedures would reduce renal risk through reduced manipulation of the ascending aorta (^5,14,23). However, no previous studies have investigated the relationship of ACDs with postcardiac surgery renal dysfunction.

Many studies have compared renal outcome after OPCAB and on-pump CABG procedures (^{6–13,24–26}); yet despite this interest there is little consensus on the relative renal risks of the two procedures. Although three randomized trials have specifically investigated acute renal injury after on- and off-pump procedures, these studies have been small, using unproven surrogate markers for renal injury, and have not been conclusive (^24–26). In contrast, many retrospective studies have compared concurrent case series of on- and off-pump patients using different statistical approaches; these studies have drawn conflicting conclusions. A limitation of all retrospective studies, that is also applicable to the current study and to many published comparisons of OPCAB versus CABG surgery, is between-group differences in baseline renal risk factors; these differences presumably reflect bias in procedure assignment. Statistical approaches used to address between-group differences in studies specifically investigating renal outcome after CABG surgery have included multivariable analysis (^6,7), propensity scores (^9,10), case-matching (⁸), risk-stratification (¹²), and meta-analysis of retrospective studies (¹¹). Although two multivariable analyses that accounted for known renal risk factors identified no difference in renal risk between the two procedures (^6,7), other studies using alternate statistical approaches have favored OPCAB over CABG surgery (^8–12). Recently, Cheng et al. (¹³) highlighted concerns regarding confounding in retrospective analyses in a meta-analysis of randomized studies comparing OPCAB and CABG surgery; in their report, including more than 1400 patients from 10 trials, these authors found no reduction in the 30-day incidence of renal failure with OPCAB surgery. In the current study, we selected multivariable analysis as an acceptable analysis method to assess a retrospective dataset and found no advantage of the use of connector devices with regard to improved renal outcome.

Other limitations of our study must be addressed. The use of retrospective data for the study also presents concerns regarding accuracy. However, our data were drawn from a prospectively gathered computerized quality assurance database, and serum creatinine determinations are performed as part of routine preoperative and postoperative care. The study includes the first OPCAB procedures with the use of the ACD at our institution; it is possible that a period of learning may have influenced renal outcome. To assess for an influence of experience and the relationship of use of the Symmetry™ ACD with the timing of the surgery during the study period, we performed an analysis of Δ%Cr by date of surgery; we found no association between surgery date and the likelihood of acute renal injury in the whole patient group or the group receiving ACD devices. A further limitation is that the decision to use the ACD was not independent of the surgeon. Unfortunately, we do not have sufficient intraoperative data to analyze variables such as mean arterial blood pressure or the use of IV mannitol or dopamine in our study. However, we include several variables that would be expected to correlate with persistent low output states, a known predictor of postoperative acute renal injury (e.g., preoperative ejection fraction, history of congestive heart failure, need for intraaortic balloon counterpulsation, preoperative need for inotrope support) (^1–3). Other factors, such as mean arterial blood pressure during CPB and perioperative use of dopamine or mannitol, have not been found to be related to acute renal injury in the perioperative setting (^27–29). Another limitation of our study is the lack of aortic clamping data, specifically the influence of epiaortic scanning in placement of aortic clamps, and other deviations from the standard approach at our institution. Intraoperative transesophageal echocardiography is performed for all patients, and the subgroup deemed to be at high risk for embolic complications (as judged by descending and arch aortic plaque burden) also have epiaortic scanning to guide ascending aortic manipulation and selection of proximal grafting sites. In addition, this information is used in CABG surgery patients to guide the decision to use a single-clamp technique (high risk) versus both cross-clamp and partial occlusion clamps (low risk). The sidebiter clamp is used for all OPCAB procedures (without ACD), and no clamp is used for OPCAB with ACD.

The patient and procedural characteristics among the three groups were dissimilar with the OPCAB connector group (Group A) being at higher risk of renal injury. However, we used multivariable analysis including numerous known renal risk factors to account for the differences in renal risk factors and other demographic variables. Notably, we confirmed strong associations between many of these factors and renal injury in the overall group, including all three procedure types. Finally, evaluation of renal filtration function ignores other aspects of kidney function; serum creatinine is an adequate measure of renal filtration but it does not address the numerous other elements of renal function such as production and release of enzymes and hormones, regulation of osmolality, electrolyte and acid-base status, and excretion of metabolic end products and toxins.

Previous studies have identified atherosclerosis of the ascending aorta as a predictor of poor renal outcome (^5,23), whether because of atheroembolism to the kidneys or because the diseased aorta may be a marker of widespread atherosclerosis. Advocates of aortic connector technology have proposed that these devices may reduce renal risk as a result of reduced aortic manipulation compared with techniques involving aortic cross-clamping or side-clamping. Our findings do not support this conclusion, suggesting either that manipulation of an atherosclerotic aorta is less important than previously thought or that even the minimal manipulation of the ascending aorta associated with deployment of the ACD device is sufficient to initiate disruption and embolization of atheromatous lesions or other processes that may cause renal injury (e.g., inflammation). Interestingly, Scarborough et al. (¹⁹) found reduced cerebral atheroembolism with ACD devices. However, Martens et al. (³⁰) found equivalent embolic debris with the use of conventionally hand-sewn and ACD devices for proximal anastomosis. Other potential concerns related to the ACD have to be considered; case reports of increased early graft occlusion have been reported, and mid-term and long-term patency rates are yet to be assessed in large prospective trials (^20,31). Finally, the expense of connectors compared with traditional grafting techniques (approximately $450 US/graft) cannot be ignored (³²).

In summary, based on the results from our retrospective analysis of a patient cohort, we conclude that OPCAB surgery with the use of the ACD does not confer major protection from postoperative renal dysfunction compared to standard OPCAB or on-pump CABG surgery. These data suggest that reduction of renal risk alone should not be used as an indication for use of ACD in OPCAB surgery.