OUP user menu

Valve failure following homograft aortic valve replacement: does implantation technique have an effect?

Ayyaz Ali, Yasir Abu-Omar, Amit Patel, Ziad Ali, Ahmad Y. Sheikh, Asim Akhtar, Aleksandra Pavlovic, Panagiotis Theodorou, Thanos Athanasiou, John Pepper
DOI: http://dx.doi.org/10.1093/eurheartj/ehn174 1454-1462 First published online: 2 May 2008


Aims Structural valve deterioration (SVD) limits the long-term durability of homograft aortic valve replacement (AVR). Valves are implanted predominantly using two techniques, the free-hand sub-coronary (SC) technique or aortic root replacement (RR). Our objective was to identify risk factors associated with the development of SVD or ascending aortic dilatation. In particular we strived to determine whether the mode of implantation had an independent effect.

Methods and results Demographic and pre-operative clinical data were obtained retrospectively through case-note review. All operations were performed by a single surgeon. Actuarial freedom from ≥2+ AR (aortic regurgitation), elevated trans-valvular gradient (TVG) (≥25 mmHg) and ascending aortic dilatation (≥4.0 cm) were assessed using Kaplan–Meier curves and multivariable Cox proportional hazards regression. A propensity analysis was carried out using a non-parsimonius logistic regression model for implantation with SC vs. RR. Between 1 January 1991 and 1 January 2001, 215 patients underwent AVR with a homograft. The SC technique was used in 131 (61%) patients and 84 (39%) patients underwent RR. Technique was not an independent predictor for ≥2+ AR (adjusted hazard ratio 1.9; 95% CI 0.56–6.16, P = 0.31), elevated TVG (adjusted hazard ratio; 0.99; 95% CI 0.15–6.71, P = 0.99) or ascending aortic dilatation (adjusted hazard ratio 2.01; 95% CI 0.50–8.25, P = 0.33). One and 5 year actuarial freedom from ≥2+ AR (log-rank – P = 0.09) and ascending aortic dilatation (log-rank – P = 0.88) were not significantly different between groups.

Conclusion The incidence of SVD and ascending aortic dilatation is not affected by the method of implantation of the aortic homograft. All homografts are prone to SVD which is responsible for a progressive increase in the prevalence of these changes over time.

  • Aortic valve
  • Bioprosthesis
  • Valve replacement
  • Homograft


Homograft aortic valve replacement (AVR) is well established. This procedure has been demonstrated over four decades to be associated with excellent clinical performance.13 Two surgical techniques have predominantly been utilized for valve insertion: the freehand sub-coronary (SC) technique and total root replacement (RR), with re-implantation of the coronary ostia. Structural valve deterioration (SVD) limits the long-term durability of the homograft. This is manifested primarily by the development of post-operative aortic regurgitation (AR). It is suggested that when AR develops early in the post-operative course, that it may act as a nidus for accelerated SVD.4 Prior investigators have demonstrated that aortic insufficiency is more common if SC valve insertion is undertaken.57 With this technique considerable care needs to be taken to ensure that the pliable homograft is implanted in a manner which allows for restoration of the normal geometry of the aortic valve and its commissures. Failure to do so has been identified as an important cause of AR. Conversely, the RR that avoids the need to disrupt the valve’s internal anatomy, is logically considered by many to result in more normal aortic valve function and less AR.8 Another mode of valve failure is progressive immobility of the valve leaflets compromising normal valve opening and function. This is noted clinically as an increasing trans-valvular gradient (TVG) measured across the aortic valve using echocardiography. Finally, some patients develop ascending aortic dilatation following AVR. This has been suggested to be a specific risk associated with the use of the SC-implant technique, where the patients native aortic wall is retained.9 Using propensity scoring to adjust for other factors which may have influenced the choice of implant method, we undertook a comparison of the two techniques of implantation to determine whether either technique was associated with an increased incidence of mid-term valve failure.


This is a retrospective review of a consecutive series of patients who underwent AVR with a homograft from 1991 to 2001. Patients were not excluded from the analysis if they had additional concomitant cardiac procedures.

An established homograft cryopreservation protocol has been in place at our institution for over 15 years. Following sterile procurement, homografts are placed in nutrient antibiotic solution (Gaya 5) and stored in an incubator at 37°C for 24 h before refrigeration at 4°C. A cryoprotectant (DMSO) is added prior to the valve being placed in a heat-sealed double-sterile barrier comprising of a sterile nylon pouch and a sterile trilaminate aluminium sachet. The sachet is placed in a controlled rate freezer, which maintains a constant freezing rate of –1°C/min. At −60°C the sachet is transferred to an ultra low freezer and is stored at −130°C for up to 5 years. For the purpose of microbiological screening, two sets of tissue samples are removed from the valve at the time of dissection and subjected to identical treatment and conditions as that of the homograft. After 24 h of incubation the first set of tissue samples are cultured for bacterial and fungal contaminants, and the second set cultured when the valve is wrapped and packed.

Homovital homografts obtained at the time of orthotopic heart transplantation are similarly stored but in a weaker nutrient antibiotic solution for 72 h at 4°C. This allows time for preliminary microbiological screening results to become available; if these are negative the valve is free to be implanted. A maximum of 7 days of storage at 4°C is permitted before the valve is subjected to cryopreservation.

Surgical technique

Operations were performed by a single surgeon (J.P.). Although the choice of implant technique was tailored to the individual, homografts were used more frequently for endocarditis, aortic root disease, and for aortic valve re-operations. All procedures were performed on cardiopulmonary bypass with cooling at 28°C; myocardial protection was undertaken using antegrade and retrograde cold blood cardioplegia. With the SC technique the valve sinuses were excised. The valve was then implanted with separate inflow and outflow suture lines, with the latter anchoring the commissures and scalloped edge of the valve to the aortic wall. The proximal suture line was constructed with interrupted 3-0 ethibond sutures, the distal suture line was continuous and fashioned with 4-0 prolene. For the free-standing RR technique the homograft was anastomosed en-bloc to the aortic annulus proximally and the ascending aorta distally. The coronary ostia were then re-implanted as buttons within the homograft.

Data acquisition

Demographic, clinical and operative data were obtained from individual patient hospital records. Annual echocardiograms were used to assess cardiac and homograft valve function. Serial echocardiographic reports were scrutinized to assess the degree of AR. This was graded from 0 to 4, (0 = none, 1 = mild, 2=moderate, 3=moderately-severe, 4=severe). Additionally, TVG across the valve was documented and diameter of the proximal ascending aorta was measured. An ascending aortic diameter ≥4.0 cm was considered as significant aortic root dilatation.


Following AVR all patients were reviewed 6 weeks post-operatively in clinic. Following this initial visit, further follow-up appointments were scheduled annually at which time patients received a thorough clinical and echocardiographic evaluation. All patients were scheduled to attend these annual appointments to review their clinical condition and the performance of their aortic valve prosthesis. Clinical and echocardiographic follow-up data were collected at each clinic visit. Surviving patients remain under this annual follow-up protocol. The survival and freedom from re-operation commenced at the time of aortic valve operation and ended at the time of death/re-operation event or at last follow-up (censoring). With respect to echocardiographic data the freedom from ≥2+ AR, TVG ≥ 25 mmHg and aortic root dilatation commenced at the time of aortic valve operation and ceased at the time of echocardiographic diagnosis or at the time of last follow-up echocardiogram (censoring).

Structural valve failure

We defined structural valve failure as the presence of at least moderate (≥2+) AR or a TVG≥25 mmHg, noted on the last echocardiogram obtained during follow-up. Furthermore we documented the number of patients who developed ascending aortic dilatation of ≥4 cm.

Echocardiographic details

Transthoracic echocardiography was performed using the Hewlett Packard Sonos 5000. Two-dimensional (2D) images were obtained using standard apical five-chamber, parasternal short- and long-axis views. Continuous wave Doppler was used to determine the presence and severity of any regurgitation. Peak and mean systolic gradients across the homograft aortic valve were determined also from continuous wave Doppler using modified Bernoulli equation. Ascending aortic diameter was determined by 2D echocardiography.

Statistical methods

Patients were grouped according to whether they had SC implantation or RR. Comparability between the two groups is presented in Table 1. Continuous variables were presented as means ± SD or medians with interquartile ranges (IQR) for normally and non-normally distributed measures, respectively. Categorical variables were shown as percentages. Comparisons for continuous variables were performed with Student t- or Mann–Witney U- test and for binary variables with χ2 test or Fischer’s exact test, when appropriate. Actuarial outcomes were compared with Kaplan–Meier curves and multivariable Cox proportional hazards regression. A log-rank test was used to determine whether significant differences existed between curves. The proportional hazard assumption was tested for each covariate by two methods: first, correlating the corresponding scaled Schoenfeld residuals with the rank of time and secondly, using the procedure ‘stphtest’ in STATA which has been based on methods described by Grambsch and Therneau.10,11 The linearity assumption for each binary covariate was assessed graphically with Martingale residuals.

View this table:
Table 1

Baseline characteristics according to implantation technique

Sub-coronaryRoot replacementP-value
Age > 70, n (%)41 (31)25 (30)0.81
Male sex, n (%)93 (71)66 (79)0.21
Hypercholesterolemia, n (%)13 (10)9 (11)0.85
Diabetes, n (%)10 (8)3 (4)0.26
Hypertension, n (%)31 (24)19 (23)0.86
Creatinine > 150 µmmol/L, n (%)31 (24)17 (20)0.56
Left ventricular impairment, n (%)27 (21)19 (23)0.72
Pre-operative NYHA III/IV, n (%)78 (60)50 (60)0.99
Pre-operative sinus rhythm, n (%)32 (24)26 (31)0.29
ACE-inhibitor, n (%)31 (24)30 (54)0.06
Pre-operative aortic root diameter, cm (SD)3.4 (SD = 0.5)3.8 (SD = 0.8)<0.001
Concomitant CABG, n (%)34 (26)9 (11)0.006
Endocarditis, n (%)7 (5)15 (18)0.003
Aortic regurgitation, n (%)24 (18)45 (54)<0.001
Redo operation, n (%)16 (12)30 (36)<0.001
Emergency operation5 (4)9 (11)0.05
Bicuspid aortic valve56 (43)31 (37)0.39
Mixed aortic valve disease18 (14)13 (15)0.72
  • ACE, angiotensin-converting enzyme; CABG, coronary artery bypass grafting; NYHA, New York Heart Association.

The propensity score was used for adjustment as a linear term in Cox-regression (as a continuous variable). Following this we also categorized the variable into five dichotomous variables of equal units and then we inserted each of these in the multivariable analysis to calculate their coefficients. Finally the coefficients were graphed against the mid-point of each propensity score variable to assess linearity.

All statistical tests were two-sided and the significance level was set at 0.05. In all cases, corrections were not made for multiple comparisons. Statistical analysis was performed using SPSS 14.0 (SPSS Inc., Chicago, IL, USA) and Stata 9.0 (Stata Corporation, College Station, TX, USA).

The statistical analysis addressed confounding factors (patient selection) by use of a propensity score and heterogeneity (risk factors) by multivariable risk factor analysis. The following models were used in our analysis: (i) Unadjusted model; (ii) Adjusted for propensity score model; (iii) Adjusted for covariates model.

Propensity model

A propensity score for each patient was calculated to determine the probability of allocation to either of the two groups and compared (SC = 1 vs. RR = 0) using a non-parsimonious logistic regression model. The discrimination of the propensity model was assessed with calculation of c-statistic. All the variables listed in Table 2 were included in this model, along with clinically valid interactions. The selection of variables included in the propensity model was based first on clinical grounds (taking into account that the independent variable is a pre-operative factor which may be related to surgical technique selection), secondly on the percentage of missing data for each candidate variable (only variables with completeness of data were included) and finally on the value of the c-statistic to assess the model. The score was subsequently incorporated into a proportional-hazards model as a covariate. We used the propensity score for adjustment and not for matching in order to avoid reduction in study size. The propensity score analysis was performed as previously recommended.12,13

View this table:
Table 2

Results of non-parsimonious logistic regression modelling used to develop the propensity score

PredictorOR (95% CI)P-value
Male sex0.72 (0.22–2.36)0.59
Age > 70 years2.26 (0.69–7.36)0.18
Redo operation0.25 (0.08–0.81)0.02
Pre-operative sinus rhythm0.10 (0.02–0.52)0.006
Left ventricular impairment1.79 (0.21–15.18)0.59
Hypercholesterolemia1.02 (0.17–5.889)0.99
Pre-operative NYHA III/IV0.90 (0.27–3.01)0.87
Pre-operative hypertension1.64 (0.35–7.77)0.53
Concomitant CABG3.63 (0.69–19.1)0.13
Creatinine >150 µmmol/L1.61 (0.32–7.97)0.56
Diabetes mellitus0.75 (0.06–10.14)0.83
Aortic regurgitation0.32 (0.89–1.22)0.1
Pre-operative aortic diameter0.40 (0.19–0.86)0.02
Mixed aortic valve disease0.42 (0.08–2.10)0.29
Endocarditis0.13 (0.01–1.47)0.1
Bicuspid aortic valve0.48 (0.16–1.48)0.20
Emergency operation0.73 (0.51–10.63)0.82
ACE-inhibitor1.54 (0.26–9.01)0.63
Interaction term between ACE-inhibitor and NYHA III/IV2.25 (0.24–20.70)0.48
Interaction term between ACE-inhibitor and left ventricular impairment0.08 (0.002–2.47)0.15
Interaction term between ACE-inhibitor and pre-operative hypertension0.15 (0.02–1.58)0.12
Interaction term between concomitant CABG and left ventricular impairment2.67 (0.05–134.09)0.62
  • ACE, angiotensin-converting enzyme; CABG, coronary artery bypass grafting; NYHA, New York Heart Association.

Adjustment for covariates model

To elucidate associated causative factors to the outcomes of interest, a multivariable regression analysis was performed. Initially univariable regression analysis was used to determine all significant confounding variables (covariates). Those associated confounding variables were subsequently included in the multivariable regression model. Only statistically or clinically significant (P ≤ 0.1) causative factors from the univariable analysis were adjusted for (included) in the multivariable model. The model was assessed with the Hosmer–Lemeshow goodness of fit test comparing estimated with observed likelihood of outcome for groups of subjects.


Between 1 January 1991 and 1 January 2001, 215 consecutive patients underwent AVR with a homograft. The SC technique was used in 131 (61%) patients and 84 (39%) patients underwent RR. Patient characteristics according to implant method are demonstrated in Table 1. SC implantation was used more commonly for degenerative calcific aortic stenosis (P < 0.001). This variable was not included in the propensity score model due to the fact that the percentage of missing values in this variable was ∼12%. Alternately, RR was the preferred technique when AR was the predominant valve lesion or in patients with ascending aortic aneurysmal disease (P < 0.001). This is reflected in the pre-operative aortic root diameter as measured by echocardiography in the two groups. Aortic root diameter was significantly larger amongst patients undergoing RR. The RR technique was also used more frequently for endocarditis (P = 0.003) and redo aortic valve operations (P < 0.001), considerably increasing the risk profile of this group. Concomitant coronary artery bypass grafting (CABG) was more commonly performed in patients who had SC homograft implantation. The median follow-up time was 3.9 years (IQR 1.6–5.7). The degree of AR was evaluated annually using echocardiography. Eighty percent of patients (170/215) returned for serial echocardiographic evaluation. The median time to last echocardiogram obtained from the date of operation was 4.0 years (IQR 2.0–6.0).

Propensity analysis

In SC patients the median propensity score (0 = RR, 1 = SC) was 0.83 (IQR = 0.31) in RR patients, the median score was 0.34 (IQR = 0.45). The c-statistic was 0.81 (area under the curve) indicating satisfactory discrimination. Variables and clinically important interactions included in the propensity score model are listed in Table 2.

We compared the balance of all baseline covariates in Table 1 between treatment groups before and after propensity score matching using the absolute standardized difference (Figure 1), which directly quantifies the bias in the means (or proportions) of covariates across the groups.

Figure 1

Absolute standardized difference before and after propensity-matching-balance of covariates.


Thirty-day mortality was significantly higher amongst patients undergoing RR (14% RR vs. 2.2% SC, P < 0.001). The Kaplan–Meier 1- and 5- year mortality risks were 88.4 ± 4% and 88.4 ± 4% for RR, and 97 ± 1.7% and 93 ± 3% for SC (Figure 2). Kaplan–Meier analysis did not reveal a significant difference in survival between the two groups (log-rank = 1.87, P = 0.176).

Aortic regurgitation

Thirty-one patients had ≥2+ AR: 6 RR and 25 SC. The Kaplan–Meier 1- and 5- year freedom from ≥2+ AR were 91.8 ± 3.9% and 89.4 ± 4.4% for RR, and 94 ± 2.3% and 68 ± 6.3% for SC (Figure 3). This was not found to be a significant difference when the curves were compared using the log-rank test, P = 0.09. Factors selected for inclusion in Cox-regression are listed in Table 3. The unadjusted hazard ratio of technique for ≥2+ AR was 2.08 (95% CI 0.87–5.0), P = 0.10. The adjusted hazard ratio for implant technique after propensity scoring was 1.9; 95% CI 0.56–6.16, P = 0.31). Using the procedure ‘stphtest’ in STATA we calculated the global test to assess the proportionality assumption which was demonstrated to be non-significant, P = 0.88 (2 degrees of freedom). The results of the multivariable analysis (the covariates model) are presented in Table 3. The model had satisfactory goodness of fit (P = 0.86). The only significant independent predictor of ≥2+ AR was mixed aortic valve disease as the presenting valve lesion (hazard ratio 4.2; 95% CI 1.7–10.34; P = 0.002). Although failing to reach significance, there was a strong trend towards the SC technique being an independent predictor for ≥2+ AR (hazard ratio 2.4, 95% CI 0.94–6.1), P = 0.06. Using the procedure ‘stphtest’ in STATA the global test to assess the proportionality assumption was not significant, P = 0.08 (9 degrees of freedom). The smoothed Martingale residuals did not show any violation of the linearity assumption similarly to the graph of coefficients for the dichotomous categories of propensity score vs. mid-point of propensity score in each category. The occurrence of ≥2+ AR as detected by echocardiography over follow-up for both groups is documented in Table 4.

Figure 3

Freedom from ≥2+ AR (aortic regurgitation).

View this table:
Table 3

Predictors of ≥2+ AR (aortic regurgitation) adjusted for covariates Cox-regression analysis

PredictorHR (95% CI)P-value
Concomitant CABG1.35 (0.43–4.22)0.61
Sub-coronary technique2.4 (0.94–6.11)0.06
Pre-operative NYHA III/IV0.55 (0.25–1.18)0.12
Hypercholesterolemia0.49 (0.09–2.47)0.38
Left ventricular impairment1.51 (0.39–5.88)0.55
Creatinine >150 µmmol/L0.96 (0.34–2.71)0.94
Pre-operative sinus rhythm1.30 (0.31–5.42)0.72
Hypertension0.91 (0.39–2.09)0.81
Mixed valve disease4.19 (1.7–10.34)0.02
  • CABG, coronary artery bypass grafting; HR, heart rate; NYHA, New York Heart Association.

View this table:
Table 4

Patients with ≥2+ AR (aortic regurgitation) at echocardiography

Time from operationSub-coronaryRoot replacement
No.No. with echo%No.No. with echo%
1 year1610814.32742.1
2 year2310122.74824.8
4 year288234.18454.1
6 year235839.63348.9
8 year84119.021513.3

Trans-valvular gradient

A TVG ≥25 mmHg at follow-up echocardiography was noted in 13 patients: 9 RR and 4 SC. The unadjusted hazard ratio of technique for TVG ≥ 25 mmHg was 0.89 (95% CI 0.27–2.9), P = 0.84. The adjusted hazard ratio for implant technique after propensity scoring was 0.99; 95% CI 0.15–6.71, P = 0.99. There were no significant independent predictors for the presence of a TVG ≥ 25 mmHg on multivariable analysis. The prevalence of a TVG ≥ 25 mmHg at follow-up echocardiography for both groups is documented in Table 5.

View this table:
Table 5

Patients with ≥25 mmHg gradient at echocardiography

Time from operationSub-coronaryRoot replacement
No.No. with echo%No.No. with echo%
1 year1510813.82742.1
2 year71016.94824.8
4 year8829.864513.3
6 year185831.253414.7
8 year144134.131520.0

Aortic root dilatation

Finally, the presence of significant ascending aortic dilatation at follow-up echocardiography was analysed. Fifteen patients had ascending aortic diameter ≥4.0 cm: 5 RR and 10 SC. The Kaplan–Meier 1- and 5- year freedom from ascending aortic diameter ≥4.0 cm were 98% and 92% for RR, and 96% and 87% for SC (Figure 4). This was not significant on comparing the curves using the log-rank test, P = 0.88. Factors selected for inclusion in the Cox-regression covariates model are listed in Table 6. The unadjusted hazard ratio of technique for ascending aortic diameter ≥4 cm was 0.92 (95% CI 0.31–2.69), P = 0.88. After adjustment for propensity the hazard ratio for technique of implantation was 2.01; 95% CI 0.50–8.25, P = 0.33. The only independent predictors of ascending aortic dilatation (Table 6) were the presence of a pre-operative serum creatinine exceeding 150 μmol/L.

Figure 4

Freedom from ascending aortic dilatation ≥4 cm.

View this table:
Table 6

Variables entered into Cox hazards regression for ascending aortic diameter ≥4.0 cm

PredictorHR (95% CI)P-value
Creatinine >150 µmmol/L7.72 (2.48–23.99)0.001
Pre-operative sinus rhythm1.42 (0.67–5.1)0.62
Left ventricular impairment0.17 (0.18–1.59)0.12
Pre-operative hypertension1.87 (0.62–5.57)0.26
Pre NYHA III/IV0.47 (0.15–1.48)0.20
Hypercholesterolemia1.18 (0.24–5.79)0.84
  • HR, heart rate; NYHA, New York Heart Association.


In our series, nine patients required re-operation for valve failure secondary to severe aortic insufficiency. In eight of these patients there was a primary leaflet problem, with commissural leaflet tears being responsible for eventual valve failure in six patients. In the remaining two patients, leaflet prolapse was the cause of AR. There was no important early AR in our series secondary to technical error. Only one patient required re-operation due to the development of aortic stenosis secondary to progressive leaflet degeneration. The median time to re-operation was 5 years. No re-operation was required for any pathology related to excessive ascending aortic dilatation.


The aortic homograft as a valve substitute for AVR is associated with excellent haemodynamic performance, this is irrespective of whether the valve is implanted in the SC position or as total root-replacement.14 Residual aortic insufficiency is undesirable, indicating either a technical problem related to valve implantation or the presence of structural valve disease. The former is usually obvious as significant AR in the early post-operative period, whereas the latter is noted over the course of longer-term follow-up. Another important mode of valve failure is the development of progressive changes within the valve leaflets that impede normal motion. Such sclerotic changes can lead to progressive obstruction to trans-valvular blood flow, and correlate with an increasing trans-aortic gradient on echocardiography. Additionally changes within the native or homograft aortic wall may contribute to valve dysfunction. Significant aortic dilatation can lead to splaying and distortion of the valve commissures, promoting AR.9 Our aim was to observe the prevalence of these changes over time in patients undergoing homograft AVR, and to determine whether the mode of implantation had an impact on their incidence.

After adjustment for propensity of technique, implantation method was not an independent predictor of ≥2+ AR, TVG ≥ 25 mmHg or ascending aortic diameter ≥4 cm at follow-up echocardiography. However, there was a trend towards an association between SC implantation and residual AR. The actuarial freedom from this degree of AR or ascending aortic dilatation was not significantly different between groups at 5 years. Several previous studies have identified SC insertion as a risk factor for post-operative AR and re-operation.2,57 The concern is that residual AR may act as a substrate for the development and progression of structural valve failure. Alternately, technical errors in initial valve implantation can result in aortic insufficiency, this occurs almost exclusively in association with SC insertion.8 Failure to accurately size the homograft or abnormal orientation of the valve and its commissures within the aortic root are common causes of early post-operative AR. Previous investigators have demonstrated that there is a significant learning curve associated with the SC technique.8 Therefore early re-operation for important residual AR is much more common in the surgeon’s early experience with this technique. However, following this initial stage the frequency of technical errors relating to SC homograft insertion is considerably lower. It follows that the need for re-operation also decreases, as greater familiarity with this implant method is gained by the surgeon. Progressive aortic dilatation at the sinotubular junction has been described following SC implantation of stentless porcine bioprostheses. David et al.9 have concluded that this can be an important cause of post-operative aortic insufficiency and is associated with an increased risk of structural valve deterioration. In our study we failed to note any important relationship between SC implantation of homograft valves and an increased incidence of proximal ascending aortic dilatation. Furthermore, we analysed the influence of pre-operative bicuspid morphology of the aortic valve on post-operative ascending aortic dilatation. Bicuspidity of the aortic valve is widely believed to predispose to aneurysmal change of the ascending aorta.15,16 When bicuspid morphology was added as a covariate in the propensity score it was not found to be a predictor of either significant AR or ascending aortic aneurysmal dilatation.

Irrespective of whether aortic insufficiency complicates the early or late post-operative course, it is believed that it may self-perpetuate eventually culminating in valve failure necessitating re-operation. There is a loss of normal leaflet extensibility over time secondary to remodelling, degenerative changes, and possible immunological responses to leaflet tissue. Eventually these processes may lead to leaflet tearing, rupture or perforation. As mentioned earlier an increased re-operation risk following SC valve insertion has been reported, particularly if aortic root tailoring was undertaken to accommodate the scalloped homograft.4,8 A recent meta-analysis of 11 separate studies comparing the SC and RR techniques for homograft AVR, demonstrated that SC implantation was a risk factor for re-operation according to both a random and fixed effect model.17

The importance of follow-up echocardiography to detect homograft failure is now appreciated. Simple 2D echocardiography and colour Doppler echocardiography represent two essential tools required for such follow-up.57 A prior study focused on echocardiographic assessment of homograft AVR revealed that para-valvular leak complicated ∼15% of SC implants.5 Furthermore, this finding was essentially confined to patients who were operated using this technique. They also comment on an increased incidence of eccentric regurgitant jets in patients with SC insertion as opposed to RR. Central regurgitant jets are commonly encountered, with an initial eccentric trajectory at the level of the leaflets, which often contacts the left ventricular outflow tract and is then projected into the LV (left ventricle). Forty-six percent of patients in whom SC implantation was undertaken were noted to possess residual central AR.5 Continuous wave Doppler also allows for detection of an elevated trans-prosthetic gradient, which may suggest the development of degenerative stenotic disease. Additionally, the ascending aorta can also be evaluated for the presence of aortic disease. We undertake routine annual transthoracic echocardiography in all complying patients as part of our follow-up protocol. This allows for early detection of SVD and the capacity to monitor its progression and decide on whether re-operation may be indicated.

The aortic homograft has proven itself as a reliable prosthesis for replacement of the aortic valve. The operation is reproducible and is associated with very good clinical outcomes.3,4,1820 The SC method of insertion is widely believed to result in an increased incidence of residual AR compared with RR.5,6 We were unable to identify the free-hand SC technique as an independent predictor of ≥2+ AR. Furthermore, previous reports have suggested that there is an increased risk of early re-operation due to technical errors related to SC valve insertion.8 There were no early re-operations for SC homografts in our series following development of significant AR due to technical failure. A single surgeon, with extensive experience in both techniques, performed all operations in our series. This lends further credence to the findings of previous investigators, which suggest that the learning curve associated with SC insertion is a major factor associated with early technical complications.8 The method of implantation was not an independent risk factor for the development of AR, aortic stenosis or late ascending aortic dilatation. Actuarial freedom from AR and aortic dilatation was not different between patients undergoing SC insertion and RR. The majority of patients who undergo homograft AVR remain free from significant structural valve failure for many years. Technique of implantation should be tailored towards the need of each individual patient. Our analysis, however, suggests that the homograft can be implanted in either the SC position or as a RR without any predisposition towards an increased incidence of significant residual aortic insufficiency, progressive aortic stenosis or ascending aortic dilatation.

Study limitations

Our study has several limitations, primarily relating to our methodology and sample size. First, it is nearly impossible to account for all differences between patients and the selection process involved in determining method of homograft implantation. The propensity technique used in this study was responsible for the residual imbalance in covariates following propensity score matching. This needs to be considered as a limitation of our study although other techniques could not be applied due to sample size constraints. A principal difficulty is ascertaining whether patients in both groups were suitable for both procedures. This can only be guaranteed in the setting of a randomized controlled trial. Although propensity analysis is a powerful statistical technique, it is inherently limited by the number and accuracy of the variables evaluated. Furthermore, the low peri-operative mortality at our institution poses an important limitation with regard to the applicability of the data. Only 80% of survivors received follow-up echocardiograms. This attests the difficulty of obtaining long-term follow-up using this modality amongst a large number of patients over a long period of time. Our institution is a tertiary referral centre; accordingly the residence of many of our patients was a considerable distance away with a proportion even residing in foreign countries. This was an obstacle in obtaining complete echocardiographic follow-up for the entire cohort. However, we feel that our follow-up has allowed us to make a meaningful assessment of the issues discussed in this manuscript. Obtaining regular echocardiography in patients undergoing AVR proves to be difficult for many logistical reasons. The presence of studies with complete long-term echocardiographic follow-up is rare in the literature, with the majority focusing only on clinical outcomes such as survival and freedom from re-operation. Finally, as our median follow-up was <10 years, the very long-term implications for patients in our series remain undetermined.


The incidence of structural valve failure is not affected by the method of implantation of the aortic homograft. Both RR and SC insertion are associated with acceptably low rates of residual aortic insufficiency, progressive stenotic change and ascending aortic dilatation. All homografts are prone to structural valve deterioration that is responsible for a progressive increase in the prevalence of these changes over time.

Conflict of interest: There are no conflicts of interests for any of the authors in reporting of these data.


View Abstract