Author + information
- Received November 16, 2017
- Revision received January 11, 2018
- Accepted January 15, 2018
- Published online March 19, 2018.
- Edward Buratto, MBBSa,
- William Y. Shi, MBBS, PhDa,
- Rochelle Wynne, PhDa,b,
- Chin L. Poh, MBBSa,b,
- Marco Larobina, MBBSb,
- Michael O’Keefe, MBBSb,
- John Goldblatt, MBBSb,
- James Tatoulis, MDa,b and
- Peter D. Skillington, MBBSa,b,∗ ()
- aUniversity of Melbourne, Melbourne, Victoria, Australia
- bDepartment of Cardiothoracic Surgery, Royal Melbourne Hospital, Melbourne, Victoria, Australia
- ↵∗Address for correspondence:
Dr. Peter D. Skillington, Department of Cardiothoracic Surgery, Royal Melbourne Hospital, Grattan Parade, Parkville, Melbourne, Australia.
Background It is unclear whether the Ross procedure offers superior survival compared with mechanical aortic valve replacement (AVR).
Objectives This study evaluated experience and compared long-term survival between the Ross procedure and mechanical AVR.
Methods Between 1992 and 2016, a total of 392 Ross procedures were performed. These were compared with 1,928 isolated mechanical AVRs performed during the same time period as identified using the University of Melbourne and Australia and New Zealand Society of Cardiac and Thoracic Surgeons’ Cardiac Surgery Databases. Only patients between 18 and 65 years of age were included. Propensity-score matching was performed for risk adjustment.
Results Ross procedure patients were younger, and had fewer cardiovascular risk factors. The Ross procedure was associated with longer cardiopulmonary bypass and aortic cross-clamp times. Thirty-day mortality was similar (Ross, 0.3%; mechanical, 0.8%; p = 0.5). Ross procedure patients experienced superior unadjusted long-term survival at 20 years (Ross, 95%; mechanical, 68%; p < 0.001). Multivariable analysis showed the Ross procedure to be associated with a reduced risk of late mortality (hazard ratio: 0.34; 95% confidence internal: 0.17 to 0.67; p < 0.001). Among 275 propensity-score matched pairs, Ross procedure patients had superior survival at 20 years (Ross, 94%; mechanical, 84%; p = 0.018).
Conclusions In this Australian, propensity-score matched study, the Ross procedure was associated with better long-term survival compared with mechanical AVR. In younger patients, with a long life expectancy, the Ross procedure should be considered in centers with sufficient expertise.
Several large studies have demonstrated that young adult patients undergoing the Ross procedure exhibit similar survival to an age- and sex-matched population in the country of study (1–5). In contrast, survival after mechanical aortic valve replacement (AVR) falls well short of what would be expected in the general population (6,7).
Given the limited number of surgeons currently performing the Ross procedure and the long follow-up required to demonstrate a survival advantage, performing a randomized control trial comparing mechanical AVR and Ross procedure would pose a formidable challenge. There has been a single randomized trial comparing the Ross procedure with homograft AVR, which did demonstrate improved survival and freedom from reoperation for the Ross procedure (3). Conversely, there have only been 3 nonrandomized, matched analyses comparing Ross procedure with mechanical AVR, which have failed to show a survival benefit for the Ross procedure (8–10). However, these studies have included relatively small numbers of patients, potentially rendering them underpowered, and relatively short duration of follow-up (8–10).
Although the Ross procedure has clear advantages compared with a mechanical prosthesis, namely the avoidance of anticoagulation and the attendant hemorrhagic and thromboembolic complications, it has remained unclear whether the Ross procedure provides a survival advantage. Hence, in this study we performed a risk-adjusted analysis to compare the outcomes of the Ross procedure performed in younger patients, over a 25-year period, with a contemporaneous cohort of patients undergoing mechanical AVR.
All patients who underwent the Ross procedure between 1992 and 2016 were included in this study. The procedures were predominantly performed by the senior author (P.D.S.) at 3 hospitals within the University of Melbourne group. To generate a contemporaneous cohort of patients undergoing isolated AVR, we combined 2 Australian multi-institutional cardiac surgery databases, because no single database spanned the entire study period.
For the years 1992 to 2001, all patients undergoing isolated mechanical AVR were identified from the University of Melbourne cardiac surgery database, which included 7 cardiac surgery units, and has previously been described (11). In 2001 the University of Melbourne database was superseded by the Australian and New Zealand Society of Cardiac and Thoracic Surgeons database, a multi-institutional Australian database including 31 cardiac surgery centers, which has previously been described in detail (12). There are currently no New Zealand centers contributing to this database. As such, between 2001 and 2016, all patients undergoing isolated mechanical AVR were selected from the Australian and New Zealand Society of Cardiac and Thoracic Surgeons database. Only patients undergoing elective, isolated mechanical AVR were included in the comparison group. Patients who underwent urgent surgery, had other concomitant cardiovascular procedures, or a diagnosis of aortic dissection or endocarditis were excluded from the study. The enrolment periods for the 2 databases did not overlap, and as such no patient was included twice.
Baseline patient demographic and operative data were prospectively recorded for inclusion in the previously mentioned databases at the time of surgery. Survival data were obtained by matching with the Australian National Death Index, the date of censoring for mortality was December 1, 2016.
Ethics approval was obtained from the Royal Melbourne Hospital Human Research Ethics Committee (QA2013104), the Australian and New Zealand Society of Cardiac and Thoracic Surgeons database ethics committee, and the National Death Index (EO2016/2/260).
The techniques used in the 392 patients who underwent the Ross procedure have been described elsewhere (13). Pulmonary autograft root replacement using a modified inclusion cylinder method with autologous support of the autograft within the patient’s own aorta (root within root) was used in 92% (361 of 392). The pulmonary autograft was inserted inside a Valsalva Dacron graft in 5% (20 of 392) of patients (11 of 392). The remaining 3% had a mixture of techniques including unsupported root replacement and subcoronary implant technique. The pulmonary valve was replaced using a cryopreserved pulmonary homograft in all cases. Most mechanical aortic valve prostheses in the matched cohort were manufactured by St. Jude Medical (SJM) (St. Paul, Minnesota) (51.6%; 142 of 275). The individual valve models implanted in the matched patients included the American Thoracic Society standard in 16.7% (46 of 275), American Thoracic Society AP in 3.6% (11 of 275), Carbomedics in 8.0% (22 of 275), Carbomedics Reduced in 1.1% (3 of 275), Duromedics in 0.4% (1 of 275), Medtronic Advantage in 1.5% (4 of 275), Medtronic Hall in 1.8% (5 of 275), Medtronic Open Pivot Standard in 4.4% (12 of 275), On-X in 9.8% (27 of 275), SJM standard in 24.7% (68 of 275), SJM HP in 0.7% (2 of 275), SJM Masters in 9.8% (27 of 275), and SJM Regent in 16.4% (45 of 275). The prosthesis model was not recorded in 0.7% (2 of 275).
Pre-operative demographic and investigative data and 20-year survival data were compared between patients receiving a Ross procedure and mechanical aortic valve prosthesis. Categorical variables were expressed as frequencies and compared using the Fisher exact and chi-square tests. Continuous variables were expressed as mean ± SD and compared using the unpaired Student’s t-test. The Kaplan-Meier method was used to analyze unadjusted survival, which was compared using the log-rank test. Multivariable Cox regression was used to examine the association between Ross procedure and mortality. The variables used for multivariable analyses were age, sex, era of surgery, hypertension, diabetes mellitus, cerebrovascular disease, peripheral vascular disease, chronic obstructive pulmonary disease, myocardial infarction, dialysis, New York Heart Association functional class, ejection fraction <45%, aortic stenosis, aortic regurgitation, mixed aortic valve disease, reoperation, congestive heart failure, and Ross procedure.
Propensity-score matching was performed to correct for the bias associated with receiving a Ross procedure. A propensity-score was generated for each patient by performing a logistic regression with the Ross procedure as the dependent variable. Baseline clinical and investigative variables, which are expected to influence cardiac surgery patient outcomes, were included (Table 1). The C-statistic was calculated for the propensity model. Once generated, patients were matched 1 to 1 on their propensity score without replacement using the “greedy” matching method with a fixed caliper width of 0.05.
Following matching, standardized differences were used to assess the degree of baseline variable balance in the manner described by Austin (14). A high degree of balance is reflected by a standardized difference of ≤10%. Among the matched pairs, the test proposed by Klein and Moeschberger was used to compare long-term survival to calculate the p value (14).
The pre-operative and investigative profile of the entire study population is presented in Table 1. Those receiving the Ross procedure tended to be younger with a lower proportion exhibiting major cardiovascular risk factors.
Intraoperative data and 30-day mortality are presented in Table 2. Those in the Ross procedure group experienced longer cardiopulmonary bypass and aortic cross-clamp times. The proportion of 30-day mortality was similar (Ross, 0.3%; mechanical AVR, 0.8%; p = 0.50). The Ross procedure was not associated with increased risk of 30-day mortality on multivariable analysis (hazard ratio: 0.65; 95% confidence interval: 0.07 to 5.96; p = 0.70).
Survival data from the National Death Index were available with a mean follow-up of 10 ± 7 years (range: 0 to 25 years). Figure 1 displays the Kaplan-Meier survival analysis for the entire study cohort. The Ross procedure was associated with superior unadjusted 20-year survival. Multivariable analysis showed the Ross procedure to be associated with a reduced risk of late mortality (hazard ratio: 0.34; 95% confidence interval: 0.17 to 0.67; p < 0.001). The full multivariable model is shown in Table 3.
The propensity-score model performed well with a C-statistic of 0.88. We matched 275 Ross procedure patients to 275 patients receiving mechanical AVR, representing a 70% matching rate. Baseline variables were well balanced as displayed in Table 4.
Among propensity-score matched pairs, those receiving the Ross procedure experienced longer cardiopulmonary bypass and aortic cross-clamp times as displayed in Table 5. There was comparable 30-day mortality between the 2 groups after matching (Ross, 0%; mechanical AVR, 0.4%; p > 0.99).
Analysis of long-term survival among the 275 matched patient pairs showed those receiving the Ross procedure to experience better survival compared with those receiving mechanical AVR as displayed in the Central Illustration.
There have now been several large studies showing that young adult patients undergoing a Ross procedure exhibit similar survival to an age- and sex-matched population in the country of study (1–5). Also, many studies have shown that survival after mechanical AVR falls well short of what would be expected of the general population (6,7). However, it is true that the Ross operation is generally not offered to high-risk patients, such as those with extensive coronary artery disease, multivalvular heart disease, and other major comorbidities. Also, this young adult group is expected to have excellent survival, regardless of prosthetic choice, purely because of their young age and a few comorbidities.
Unfortunately, the challenges involved in conducting a randomized trial comparing the Ross operation with mechanical AVR are formidable. The treatment choices are quite different, with the necessity for life-long warfarin after mechanical AVR, whereas ongoing follow-up of both aortic and pulmonary valves is required after Ross procedure. Furthermore, with relatively few Ross procedures being performed and limited number of surgeons with experience in the procedure, the very long recruitment and follow-up periods are required before meaningful results are obtained. Impressively, El-Hamamsy et al. (3) have published a randomized control trial comparing Ross procedure with homograft AVR, demonstrating superior survival and freedom from reoperation in patients receiving the Ross procedure. However, there has never been a randomized control trial comparing the Ross procedure with mechanical AVR. Propensity-score matched studies, such as this, although less powerful, offer an alternative to determine a difference in survival between these 2 radically different prosthesis choices. Intuitively, one would expect worse survival after mechanical AVR, because of worse hemodynamic function, and the problem of thromboembolism, valve thrombosis, and bleeding, secondary to life-long anticoagulation. However, this has been difficult to demonstrate until now.
Thus far, 3 studies including matched analyses comparing Ross procedure with mechanical AVR have failed to demonstrate a significant difference in survival. Mokhles et al. (8) used propensity score matching to compare 236 Ross procedure patients with 252 patients undergoing mechanical AVR, and demonstrated equivalent survival. In this study the mean follow up was only 5.1 years, which was perhaps insufficient to demonstrate an advantage for Ross procedure. Furthermore, those receiving a mechanical prosthesis were drawn from a randomized control trial of state-of-the-art anticoagulation management, with patients being self-managed and telemonitored. The comparison in this study is thus biased, because those receiving mechanical valves likely experienced lower rates of bleeding and thromboembolism than would patients receiving standard monitoring of anticoagulation, which would potentially improve survival (15). Finally, the survival of the Ross cohort was somewhat lower than we have observed, and the patients were drawn from several different institutions, and in fact, from >1 country.
Mazine et al. (9) compared 208 propensity-matched pairs undergoing the Ross procedure with those undergoing mechanical AVR, demonstrating equivalent long-term survival. The smaller number of matched patients and the relative paucity of late deaths may have rendered the study underpowered to detect a significant difference in all-cause survival. However, there was a significant difference in late cardiac deaths at 20 years, favoring the Ross procedure.
Most recently, Sharabiani et al. (10) published a large study derived from the UK national database, comparing Ross procedure with mechanical and bioprosthetic AVR in a matched cohort of children and young adults (<40 years old). In a subgroup analysis of 16- to 40-year-old patients, survival was significantly better for Ross procedure patients than those receiving bioprosthetic AVR, but there was only a trend to improved survival when compared with patients receiving mechanical AVR. However, this study included a much younger group of patients, with overall Ross procedure patients having a mean age of only 13.1 years. Furthermore, in the subgroup analysis, 224 young adults underwent Ross procedure, and it is not clear what proportion was included in the matched analysis. Given a cohort of only 224 patients from which to draw matched pairs, it is quite likely this study was underpowered to detect a difference between the groups.
Despite the proven advantages of the Ross procedure, including excellent hemodynamic function, lack of need for permanent anticoagulation, lack of valve ticking noise, and low reoperation rate, when the operation is performed in an experienced center, its use is still confined to a relatively small number or centers worldwide (16). Until recently, this could be explained by the wide variety of different techniques used by various surgical groups to implant the pulmonary autograft in the aortic root. However, there has more recently been identification of perhaps 3 or 4 successful techniques, mostly involving some degree of support of the autograft, to prevent late dilation of the aortic root, which have led to more consistent and reproducible results (3,5,17). Even so, this has not led to an increase in the number of Ross procedures performed, despite evidence that this procedure, notwithstanding its complexity, can be taught to junior surgeons, without a concomitant increase in early mortality during the so-called “learning curve” (18). If survival post-Ross procedure can be proven to be superior to that of mechanical AVR, in this younger adult population group, this could provide the incentive for revival of the Ross procedure, possibly dictated by patient and cardiologist preference.
The long-term clinical outcomes of the Ross procedure performed in our centers has been previously reported (13,17). We have previously demonstrated low rates of early mortality, reoperation, and recurrent valvular dysfunction. This study represents the largest in the literature comparing the Ross procedure with mechanical AVR. In this paper, the Ross procedure is shown to be associated with a 10% adjusted survival benefit over 20 years. Also, there is nearly 3 times the incidence of late death at 20 years after mechanical valve versus Ross procedure (16% vs. 6%). This is in addition to the benefits of being free from anticoagulation after Ross procedure, and free of the other disadvantages of mechanical prostheses (19).
The survival benefit may be multifactorial. It could be partially explained by the absence of anticoagulation and its complications. However, it may also reflect the more favorable valve hemodynamics associated with a Ross procedure, whereby the effective orifice area is much greater than that which can achieved with a prosthesis.
Unsurprisingly, Ross procedure cases experienced longer cardiopulmonary bypass and aortic cross-clamp times. This may have a negative impact on post-operative complications. We were unable to analyze these given that early post-operative outcomes were not uniformly collected across both databases over the relatively long study period. However, the low early mortality rates observed in both groups are reassuring. Nevertheless, the long-term survival advantage afforded by the Ross procedure seems to outweigh any possible early morbidity differences between the groups.
One important consideration is that most Ross procedures in this study were performed by a single surgeon. Although we believe the Ross procedure—with our local modification—can be reproduced with appropriate training, we accept that this may limit generalizability. However, this is a reflection of the fact that few surgeons are currently performing the Ross procedure worldwide. Furthermore, this cohort includes our entire experience with the Ross procedure, including the learning curve, which we believe increases generalizability. By demonstrating favorable outcomes with the Ross procedure compared with mechanical AVR we hope that more surgeons will seek appropriate training, and similar results will be achieved in other centers.
Limitations of this study include its retrospective nature, whereby multivariable and propensity-score analyses are inherently subject to the presence of hidden and unquantifiable biases. In this study, we did not analyze the cause of death, or freedom from morbidity, such as thromboembolism, bleeding, reoperation, and prosthetic valve endocarditis, because these data were not uniformly available via the multicenter databases. Another limitation of using multicenter databases is that matching can only be performed on the data collected; some variables, such as left atrial size and left ventricular mass, would have ideally been included in the model, but are not in the database. It is possible that including some of these variables that could not be analyzed would have affected the significance of our findings. Multiple different mechanical prostheses were used throughout the study period, and although this may be considered a limitation, it reflects real-world practice in Australian institutions. We have not investigated the performance of bioprostheses compared with Ross procedure patients, and it is unclear how newer generation bioprostheses will perform. Other studies that have examined the survival after bioprosthetic AVR in young patients have shown it to be no better and in some studies, even worse than after mechanical AVR (20–22). A comparison of Ross procedure and bioprosthetic AVR will be the subject of a future report from our group.
In this long-term, propensity-score matched study, the Ross procedure is associated with better long-term survival in suitable patients compared with mechanical AVR. In younger patients, with a long life expectancy, the Ross procedure should be considered in centers with sufficient expertise.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: In young adults with aortic valve disease, the Ross procedure improved survival compared with aortic valve replacement with a mechanical prosthesis.
TRANSLATIONAL OUTLOOK: Further studies are required to determine whether the survival advantage associated with the Ross procedure can be generalized to patients operated in other surgical centers where techniques may differ.
Dr. Buratto is a recipient of a Reg Worcester Scholarship from the Royal Australasian College of Surgeons and a Postgraduate Scholarship from the National Health and Medical Research Council (1134340). Dr. Shi is supported by the Royal Australasian College of Surgeons Foundation for Surgery Peter King/Heart Foundation Research Scholarship in addition to the University of Melbourne Viola Edith Reid and the RG and AU Meade Scholarships. This project was supported by funding provided by the Ross Procedure and Cardiac Surgery Trust. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- aortic valve replacement
- Received November 16, 2017.
- Revision received January 11, 2018.
- Accepted January 15, 2018.
- Sievers H.,
- Stierle V.,
- Charitos E.,
- et al.
- Mokhles M.M.,
- Körtke H.,
- Stierle U.,
- et al.
- Mazine A.,
- David T.E.,
- Rao V.,
- et al.
- Sharabiani M.T.,
- Dorobantu D.M.,
- Mahani A.S.,
- et al.
- Buxton B.F.,
- Shi W.Y.,
- Newcomb A.E.,
- et al.
- Puskas J.,
- Gerdisch M.,
- Nichols D.,
- et al.
- Carrel T.
- Bouhout I.,
- Ghoneim A.,
- Poirier N.,
- et al.
- Wilson W.M.,
- Grigg L.E.,
- Gorelik A.,
- Skillington P.