Aortic Valve Replacement Surgery: Comparison of Outcomes in Matched Sternotomy and PORT ACCESS Groups

William T. Brinkman, MD, William Hoffman, MD, Todd M. Dewey, MD, Dan Culica, MD, PhD, Syma L. Prince, RN, BSN, Morley A. Herbert, PhD, Michael J. Mack, MD, and William H. Ryan, MD Medical City Dallas Hospital, Dallas; Cardiopulmonary Research Science and Technology Institute, Dallas; The Heart Hospital, Plano; and Presbyterian Hospital of Dallas, Dallas, Texas

Background. In the past decade, minimally invasive approaches have been developed for aortic valve surgery. We reviewed our data to determine if the use of the PORT ACCESS technique has improved hospital morbidity and mortality. Methods. Data were collected on 90 patients who had a replacement of their aortic valve using PORT ACCESS procedures (PORT ACCESS aortic valve replacement [PAVR]). This group was then matched 1:4 to a control group having aortic valve replacement surgery using a standard sternotomy approach. Results. The two groups had no statistically significant differences in preoperative risk factors. The perioperative and 30-day outcomes from the matched AVR and PAVR groups showed no mortalities in the PAVR group and 3.1% in the AVR group. Mean length of stay was shorter for PAVR patients (7.2 ⴞ 5.0 days; median 6 days) compared with the mean stay in the sternotomy group (8.5 ⴞ

9.5 days; median 6 days), PAVR patients also had statistically significant shorter intensive care unit stays, and time on ventilator. The number of patients needing ventilator support postoperatively was significantly lower in the PORT ACCESS group. Cross-clamp and perfusion times were longer in the PAVR group. No other morbidity was significantly different between groups, except for postoperative tamponade (higher in PAVR group). Conclusions. In this analysis of matched patients, the patients having aortic valve replacement using PORT ACCESS procedures, spent a shorter time in the intensive care unit and had less need for postoperative ventilator usage (both number of patients using a ventilator and the mean time of use) in comparison with patients undergoing conventional sternotomy.

C

aortic valve replacement [PAVR]), and compares those results with a computer-matched control group of patients who underwent aortic valve surgery using standard median sternotomy.

onventional aortic valve surgery with sternotomy has evolved over the last 30 years and today provides improved patient survival, combined with acceptable morbidity and mortality [1]. In the past decade, aortic valve surgery utilizing minimally invasive approaches has been employed in an effort to decrease the “invasiveness” of the procedure. It has been difficult to consistently demonstrate objective benefits to minimally invasive techniques for aortic valve replacement (miniAVR). Reduced pain and hospital length of stay [2], decreased time until return to full activity, and decreased blood product use have been demonstrated [3, 4]. Other investigators have not been able to show any advantage to mini-AVR approaches except a smaller incision [5]. Aortic valve surgery using a PORT ACCESS approach, while not as common, has been shown to be safe and effective [6, 7]. This study reviews our experience with PORT ACCESS aortic valve surgery (PORT ACCESS Accepted for publication March 22, 2010. Presented at the Fifty-sixth Annual Meeting of the Southern Thoracic Surgical Association, Marco Island, FL, Nov 4 –7, 2009. Address correspondence to Dr Herbert, Medical City Dallas Hospital, 7777 Forest Ln, Ste C-740, Dallas, TX 75230; e-mail: morley.herbert@ hcahealthcare.com.

© 2010 by The Society of Thoracic Surgeons Published by Elsevier Inc

(Ann Thorac Surg 2010;90:131–5) © 2010 by The Society of Thoracic Surgeons

Patients and Methods Patient Population The authors’ Society of Thoracic Surgeons certified, audited database was queried for all patients undergoing isolated aortic valve replacement (AVR). Patients whose status was listed as “emergent” or “salvage,” had an intraaortic balloon pump preoperatively, or previous coronary artery bypass grafting or valve surgery were eliminated, leaving a dataset with 1,198 patients, 90 of whom had surgery using PORT ACCESS techniques between January 1996 and August 2009. All parameters were measured as defined in the Society of Thoracic Surgeons Adult Cardiac Database. Patients were considered potential candidates for PORT ACCESS AVR if they had not had prior cardiac surgery and were free of any significant aortic, iliac, or femoral arterial occlusive disease. This study was reviewed by the North Texas Institutional Review board at Medical City in Dallas and approved with a waiver of consent. 0003-4975/$36.00 doi:10.1016/j.athoracsur.2010.03.055

ADULT CARDIAC

Aortic Valve Replacement Surgery: Comparison of Outcomes in Matched Sternotomy and PORT ACCESS Groups

132

BRINKMAN ET AL PORT ACCESS AORTIC VALVE SURGERY

Operative Technique

ADULT CARDIAC

After induction of general anesthesia, patients receiving PAVR underwent endovascular cardiopulmonary catheter placement with the assistance of transesophageal echocardiography and intraoperative fluoroscopy. The minimal incision patients were prepared and draped in the usual manner. A transverse incision (2.0 to 3.0 cm) was made in the femoral crease, and the femoral artery and femoral vein were isolated. Through an anterior purse-string suture of 5-0 Prolene (Ethicon, Somerville, NJ), a QuickDraw (Edwards Lifesciences, Irvine, CA) venous cannula was passed into the superior vena cava under echocardiographic guidance and using the Seldinger technique. The femoral artery was then sized and cannulated through a transverse arteriotomy with a 21-Fr or 23-Fr EndoReturn (Edwards Lifesciences) arterial cannula. Retrograde cardioplegia was administered through a coronary sinus catheter (Cardiovations Inc, Somerville, NJ) placed percutaneously through the right internal jugular vein. The aorta was cross-clamped using a Cosgrove flexible aortic cross-clamp. Access to the aortic valve was obtained using a 4 to 5 cm right thirdinterspace minithoracotomy incision and a standard transverse aortotomy. In the AVR group, aortic valve surgery was performed using a conventional mediansternotomy approach with direct cannulation of the aorta and right atrium. Postoperatively, no fast track methods or techniques were employed in either group. The time of discharge was determined by conventional surgical judgment and patient threshold for discharge. As a fraction of the aortic replacement procedures, the PORT ACCESS percent randomly varied from 32% to 70% of the cases in any year.

Statistical Analysis All data were analyzed using SAS (v. 9.2; SAS Institute, Cary, NC). The data were analyzed as intent-to-treat. To create an equivalent control group of patients having their AVR procedure through a sternotomy, we used a computer matching algorithm to select equivalent patients. Matching was carried out using the GMATCH macro (created by Department of Biostatistics, Mayo Clinic,) that matched without replacement. To balance the datasets, it was necessary to match for age, New York Heart Association class, and heart failure. Patients were also matched by the date of surgery (within 5 years) to insure a balance of cases over time, and minimize comparing one group who may have been done early on with the second group that may have many more cases done later in the study. It also balances any changes in the surgical procedure over the period of the study. Groups were matched on the type (bioprosthetic or mechanical) of valve used to insure that there was no bias introduced from the differing postoperative regimens for the valve types. Matching was carried out with 4 sternotomy patients matched to each PORT ACCESS patient. All 90 PORT ACCESS patients were fully matched.

Ann Thorac Surg 2010;90:131–5

Categoric variables were analyzed using ␹2 statistics, while continuous variables were compared using t tests. Two-by-two tables with small cell counts in one or more cells were analyzed using the Fisher’s exact test. Preoperative risk factors and perioperative and 30-day outcomes were calculated for each patient group. Continuous data are presented as mean ⫾ standard deviation, while categoric variables are given as a count and percentage of patients. Continuous variables that are significantly skewed are given as median values and were tested using nonparametric tests. Learning curve data were analyzed using local weighted regression with a smoothing factor of 0.7. Three segments were detected and the data were analyzed using robust regression to minimize outlier influences. Analysis of variance with Tukey post comparison tests was used for a three way comparison of the surgical times.

Results Aortic Valve Replacement After matching the 90 PORT ACCESS patients with an equivalent group of 360 sternotomy patients, analysis of preoperative risk factors shows the groups are equivalent. As shown in Table 1, all the measured parameters were equally distributed in the two groups. The distribution of sizes for the implanted valves is shown in Figure 1. The distributions are not statistically different (p ⫽ 0.422). In both matched groups, 62.2% of the valves implanted were bioprosthetic models. The perioperative and 30-day outcomes from the matched AVR and AVR ⫹ PORT ACCESS groups are shown in Table 2. There were no mortalities in the PORT ACCESS group (0.0%), while in the AVR group the mortality rate was 3.1% (11 of 360). The mean length of stay was shorter for PORT ACCESS patients (7.2 ⫾ 5.0 days; median 6 days) compared with the mean stay in the sternotomy group (8.5 ⫾ 9.5 days; median 6 days), a statistically significant difference, p ⫽ 0.033. PORT ACCESS patients also had statistically significant shorter intensive care unit (ICU) stays, and time on ventilator. The number of patients needing to use a ventilator post-operatively was significantly lower in the PORT ACCESS group. Cross-clamp and perfusion times were slightly longer in the PORT ACCESS group. The surgical time (skin to skin) was calculated for each PORT ACCESS case and analyzed using local weighted regression. The results are shown in Figure 2. The data have been fit by three lines, one covering cases 1 to 22 and the next 22 to 47, with the remaining segment covering cases 47 to 90. In the first segment, the regression has a slope of ⫺4.3 ⫾ 2.4 minutes (improvement) per case; second segment improves at ⫺2.9 ⫾ 1.4 minutes per case, while the final segment is nearly flat with a slope of 0.27 ⫾ 0.37 per case. This suggests a learning curve that flattens after about 45 to 50 cases. The mean surgical times in groups 1, 2, and 3 are 302 ⫾ 75, 224 ⫾ 56, and 196 ⫾ 31 minutes, respectively. The decrease from group 1 to 2 is

Ann Thorac Surg 2010;90:131–5

BRINKMAN ET AL PORT ACCESS AORTIC VALVE SURGERY

133

Parameter

AVR (n ⫽ 360)

PAVR (n ⫽ 90)

p Value

Males Diabetes Dyslipidemia Renal failure Renal failure requiring dialysis Hypertension Stroke Previous myocardial infarction Angina Status – urgent Aortic stenosis Aortic insufficiency (2⫹) Mitral stenosis Mitral insufficiency (2⫹) Tricuspid stenosis Tricuspid insufficiency (2⫹) Heart failurea Current smoker NYHA class IVa Cerebrovascular disease Severe chronic lung disease Peripheral arterial disease Continuous variables: Age (years) a Ejection fraction Body surface area (m2) STS PROM (%)

55.0% (198) 20.8% (75) 39.7% (143) 2.2% (8) 0.6% (2) 60.6% (218) 5.8% (21) 4.4% (16) 19.9% (68/342) 24.2% (87) 85.0% (305/359) 29.3% (98/335) 2.0% (7/357) 4.7% (12/257) 0% 2.0% (5/249) 26.7% (96) 13.7% (49/359) 13.3 (48) 10.0% (36) 1.7% (6/347) 7.0% (25)

55.6% (50) 23.3% (21) 35.6% (32) 2.2% (2) 1.1% (1) 56.7% (51) 10.0% (9) 6.7% (6) 13.8% (12/87) 20.0% (18) 83.2% (74/89) 29.3% (24/82) 2.3% (2/88) 10.0% (5/50) 0% 0% 26.7% (24) 12.2% (11) 13.3% (12) 7.8% (7) 0% 6.7% (6)

0.925 0.605 0.468 1.000 0.562 0.501 0.156 0.382 0.193 0.403 0.672 0.998 0.852 0.132 — 0.322 1.000 0.722 1.000 0.521 0.220 0.921

68.8 ⫾ 12.5 0.537 ⫾ 0.116 1.94 ⫾ 0.28 3.43 ⫾ 2.91

71.5 ⫾ 11.6 0.523 ⫾ 0.112 1.94 ⫾ 0.25 3.62 ⫾ 2.59

0.065 0.304 0.896 0.572

a

Matching variable.

AVR ⫽ aortic valve replacement; NYHA ⫽ New York Heart Association; Society of Thoracic Surgeons predicted risk of mortality.

statistically significant (p ⬍ 0.05) while the change between groups 2 and 3 is not.

Complications There was one groin complication in the PAVR group. A 75-year-old male experienced a deep vein thrombosis

PAVR ⫽ PORT ACCESS aortic valve replacement;

STS PROM ⫽

after femoral vein cannulation and required long-term anticoagulation. No arterial insufficiency or embolic phenomena were noted related to femoral artery cannulation for PAVR. There was 1 case of deep sternal wound infection in the AVR group; no perioperative myocardial infarction in either group. One patient had to be converted from a PORT ACCESS approach to a full sternotomy. The procedure was successfully completed (as intent-to-treat, this patient was analyzed in the PORT ACCESS group). While perioperative stroke has been a concern in PORT ACCESS procedures, our analysis of the matched groups shows that its incidence is low, with a rate of 1.5% cases in the AVR group and 1.2% in the PORT ACCESS cohort (p ⫽ 0.839).

Comment

Fig 1. Distribution of valve sizes in sternotomy and PORT ACCESS groups.

Since the mid 1990s minimally invasive approaches to aortic valve surgery (mini-AVR) have been utilized. Most reports have involved partial upper sternotomy [8, 9] with a minority involving parasternal and right anterior thoracotomy approaches [10, 11]. This report focuses on the use of PORT ACCESS (PAVR) for replacement of the aortic valve. This technique, using endovascular cardio-

ADULT CARDIAC

Table 1. Preoperative Risk Factors in the Matched Patient Groups

134

BRINKMAN ET AL PORT ACCESS AORTIC VALVE SURGERY

Ann Thorac Surg 2010;90:131–5

Table 2. Mortality and Morbidity in the Matched Patient Groups AVR (n ⫽ 360)

Parameter

ADULT CARDIAC

Operative mortality Hospital stay (days) Cross-clamp time (minutes) Perfusion time (minutes) Ventilation time (hours) Prolonged ventilation (⬎24 hour) Required postop ventilator usage ICU stay (hours) Atrial fibrillation Blood products used Reoperation for bleeding Stroke Tamponade Readmit within 30 days of procedure Cardiac arrest Heart block Renal failure Renal failure requiring dialysis Pneumonia Septicemia AVR ⫽ aortic valve replacement;

3.1% (11) 8.5 ⫾ 9.5 (Median 6) 75 ⫾ 25 103 ⫾ 34 16.3 ⫾ 48.6 (Median 6.0) 10.3% (36/349) 76.0% (266/350) 65.7 ⫾ 71.5 (Median 45.6) 28.1% (99/353) 57.3% (205/360) 4.6% (16/347) 1.5% (5/345) 0.6% (2/345) 6.5% (21/321) 1.5% (5/345) 2.6% (9/345) 5.2% (18/345) 1.6% (5/313) 2.3% (8/349) 1.5% (5/344)

AVR ⫹ PORT ACCESS (n ⫽ 90)

p Value

0 7.2 ⫾ 5.0 (Median 6) 91 ⫾ 20 108 ⫾ 23 9.2 ⫾ 25.6 (Median 1.0) 10.3% (9/87) 51.8% (44/85) 44.4 ⫾ 44.3 (Median 25.0) 37.1% (33/89) 64.4% (58) 8.1% (7/86) 1.2% (1/86) 4.6% (4/87) 7.5% (5/67) 0 1.1% (1/87) 2.3% (2/86) 0 0 0

0.093 0.033 ⬍0.001 0.071 ⬍0.001 0.994 ⬍0.001 0.010 0.096 0.216 0.192 0.839 0.017 0.784 0.259 0.419 0.254 0.267 0.154 0.261

ICU ⫽ intensive care unit.

pulmonary bypass catheters, an endovascular pulmonary artery vent, and cardioplegia catheters has been described previously by our group and others [11, 12]. Objective benefits of these minimally invasive approaches have been inconsistently demonstrated. Prior studies have shown decreased length of stay [13, 14], decreased postoperative discomfort [3, 15], decreased time on the ventilator [2, 3, 14], and decreased use of blood products [2– 4, 14, 16] when compared with a conventional sternotomy. There also have been multiple reports that have demonstrated no objective benefit other than a smaller incision [5, 17]. Our data here support the contention that replacement of the aortic valve using a PORT ACCESS technique is safe and reproducible. Moreover, these data show significant advantages in the ventilator time, and ICU length of

stay. It is likely that less trauma to the chest wall can result in decreased postoperative ventilator use and shorter ICU stays. The time on cardiopulmonary bypass is equal in both groups, although the aortic cross-clamp times were significantly increased in the PORT ACCESS group. This has been a consistent finding in most minimally invasive valve surgery studies. The increase in cross-clamp time is modest and to our knowledge has not resulted in adverse consequences. Exposure to the aortic valve is excellent using PORT ACCESS techniques. No video assistance is necessary, but can be useful when teaching the technique. The key to good exposure rests in proper incision location and strategic placement of pericardial stay sutures. Conversion to full sternotomy was rare (1 of 90 patients) in this matched case series. There were no instances of paravalvular leak or valve dysfunction in either group. In this analysis of matched patients, the patients having aortic valve replacement using PORT ACCESS procedures spent a shorter time in the ICU and had less need for postoperative ventilator usage (both number of patients using a ventilator and the mean time of use) in comparison with patients undergoing conventional sternotomy.

References

Fig 2. Decrease in surgical time with surgeon experience using PORT ACCESS.

1. Shiono M, Sezai Y, Sezai A, Hata M, Iida M, Negishi N. Long-term results of the cloth-covered Starr-Edwards ball valve. Ann Thorac Surg 2005;80:204 –9. 2. Liu J, Sidiropoulos A, Konertz W. Minimally invasive aortic valve replacement (AVR) compared to standard AVR. Eur J Cardiothorac Surg 1999;16(suppl 2):S80 –3.

3. Mächler HE, Bergmann P, Anelli-Monti M, et al. Minimally invasive versus conventional aortic valve operations: a prospective study in 120 patients. Ann Thorac Surg 1999;67:1001–5. 4. Dogan S, Dzemali O, Wimmer-Greinecker G, et al. Minimally invasive versus conventional aortic valve replacement: a prospective randomized trial. J Heart Valve Dis 2003;12: 76 – 80. 5. Detter C, Deuse T, Boehm DH, Reichenspurner H, Reichart B. Midterm results and quality of life after minimally invasive vs. conventional aortic valve replacement. Thorac Cardiovasc Surg 2002;50:337– 41. 6. Kaur S, Balaguer J, Vander Salm TJ. Improved myocardial protection in minimally invasive aortic valve surgery with the assistance of port-access technology. J Thorac Cardiovasc Surg 1998;116:874 –5. 7. Kort S, Applebaum RM, Grossi EA, et al. Minimally invasive aortic valve replacement: echocardiographic and clinical results. Am Heart J 2001;142:476 – 81. 8. Tabata M, Aranki SF, Fox JA, Couper GS, Cohn LH, Shekar PS. Minimally invasive aortic valve replacement in left ventricular dysfunction. Asian Cardiovasc Thorac Ann 2007; 15:225– 8. 9. Tabata M, Umakanthan R, Cohn LH, et al. Early and late outcomes of 1000 minimally invasive aortic valve operations. Eur J Cardiothorac Surg 2008;33:537– 41. 10. Minale C, Reifschneider HJ, Schmitz E, Uckmann FP. Minimally invasive aortic valve replacement without sternotomy.

BRINKMAN ET AL PORT ACCESS AORTIC VALVE SURGERY

11. 12.

13.

14.

15. 16. 17.

135

Experience with the first 50 cases. Eur J Cardiothorac Surg 1998;14(suppl 1):S126 –9. Wheatley GH 3rd, Prince SL, Herbert MA, Ryan WH. Portaccess aortic valve surgery: a technique in evolution. Heart Surg Forum 2004;7:E628 –31. Gregoric ID, Tamez D, Karabulut MN, et al. Aortic valve implantation in the calf: a successful approach using Heartport cannulation and minimally invasive techniques. J Heart Valve Dis 2001;10:675– 80. Mihaljevic T, Cohn LH, Unic D, Aranki SF, Couper GS, Byrne JG. One thousand minimally invasive valve operations: early and late results. Ann Surg 2004;240:529 –34. Bonacchi M, Prifti E, Giunti G, Frati G, Sani G. Does ministernotomy improve postoperative outcome in aortic valve operation? A prospective randomized study. Ann Thorac Surg 2002;73:460 –5. Candaele S, Herijgers P, Demeyere R, Flameng W, Evers G. Chest pain after partial upper versus complete sternotomy for aortic valve surgery. Acta Cardiol 2003;58:17–21. Bakir I, Casselman FP, Wellens F, et al. Minimally invasive versus standard approach aortic valve replacement: a study in 506 patients. Ann Thorac Surg 2006;81:1599 – 604. Christiansen S, Stypmann J, Tjan TD, et al. Minimallyinvasive versus conventional aortic valve replacement– perioperative course and mid-term results. Eur J Cardiothorac Surg 1999;16:647–52.

DISCUSSION DR THORALF SUNDT (Rochester, MN): I appreciate the Association inviting me to discuss this paper. This is a very nicely performed study. Thanks very much for bringing this information to us. It is important for us to continue to press forward in making our surgical procedures less invasive. Indeed, even if you can’t prove it to be superior, if it is equivalent, that is clearly what the marketplace wants. My first question is about numbers. This represents 90 patients who had PORT ACCESS procedures performed over a considerable period of time. I think it represents about 7-1/2% of your total number of aortic valve replacements performed by your group during that same interval? DR BRINKMAN: There were about 1,200 cases. DR SUNDT: So they represent a pretty small fraction of the total cases. Who in your current practice has their AVR (aortic valve replacement) performed via this approach currently then? DR BRINKMAN: Well, there is, first, surgeon-specific preference, and there are certain surgeons in our group who do the PORT ACCESS technique, others who don’t. Body habitus is a consideration; in an obese patient generally we might shy away from the technique, or patients with severe aorto-iliac disease that would preclude femoral access, those are the main considerations, and a very isolated aortic valve problem, no aortic root problem, things like that. DR SUNDT: The second question relates to your choice of a thoracotomy approach rather than a hemisternotomy approach. What has your experience been in terms of pain postoperatively? Do you have patients with post-thoracotomy pain?

DR BRINKMAN: We have not looked at that specifically. Anecdotally, the pain between the upper hemisternotomy and the thoracotomy seemed to be similar, in my opinion. DR SUNDT: And then the last comment is that the principal difference that you showed seemed to be with regard to ICU (intensive care unit) time and ventilator time. Certainly as surgeons we tend to focus on technical interventions and their impact on these kinds of outcomes, but in fact there are probably process interventions that can have an equivalent, if not greater, impact on those very same outcomes. In our own institution we have recently instituted some clinical protocols with regard to vent weaning that allow the nurses to drive the process and have already seen a really dramatic impact on those exact same parameters. So I wonder if you could make a comment about what kinds of process interventions accompanied these PORT ACCESS patients and the relative contributions of the technical aspects of the procedure versus these protocols? DR BRINKMAN: Well, this was over a long period of time and some of the postop and ICU techniques have changed over that period. We use dedicated critical care doctors to see these patients afterwards. There was no difference in the doctors that were seeing the open versus the PORT ACCESS. There is no fast track on the PORT ACCESS versus the open. However, we do think that over time our length of intubation and ICU stay in general has decreased, too, with the use of critical care specialists. DR SUNDT: You don’t have any standardized vent weaning protocols or guidelines or pathways? DR BRINKMAN: There are some generalized pathways but nothing specific for the PORT ACCESS technique. DR SUNDT: Thank you.

ADULT CARDIAC

Ann Thorac Surg 2010;90:131–5