Discussion
Operations in the setting of extended ascending aortic replacement can range in complexity from hemiarch to total arch replacement. The proponents of the hemiarch repair cite its technical simplicity, perceived durability, shorter circulatory arrest times, and potential for lower morbidity11,12. Those who prefer the total arch repair believe that the reintervention rate should be lower with complete replacement leading to improved long-term survival13-15.
There are two principle findings of this study. The first is that, total arch replacement at the time of extended ascending aortic replacement is associated with increased mortality and reintervention rates when compared with hemiarch replacement. Second, age is associated with increasing mortality regardless of the extent of repair. Collectively, our data suggests that mortality and reintervention rates are higher in the total arch group, making the hemiarch procedure the preferred procedure as long as it adequately addresses the aortic pathology, especially in those of advanced age.
The baseline characteristics of those undergoing hemiarch and total arch replacement were somewhat different in this series. Those undergoing hemiarch replacement were more commonly male, hypertensive, diabetic, had peripheral arterial disease and a prior myocardial infarction. In addition, those undergoing hemiarch repair were more likely to have been done emergently and have an aortic dissection as opposed to an aortic aneurysm as the operative indication. Our assessment is that the hemiarch group in this study appeared somewhat “sicker” than the total arch group, and that would be consistent with our philosophy of only performing a total arch replacement during extended ascending aortic replacement when clinically indicated.
Our in-hospital and 30 day mortality was 5.4 and 4.7% in the hemiarch group and 13.4 and 14.3% in the total arch groups. These results, as well as our overall survival at 1, 3, and 5 years of 83.6, 76.2, and 68.2% are comparable to previously published studies12,16-18. While our study did not focus on perioperative outcomes, we had relatively low stroke rates (3.4 vs 5.4%, hemiarch vs total arch, p=0.43), as well as similar rates of paraplegia, need for postoperative extracorporeal membrane oxygenation, and reoperations for bleeding between groups. These results are similar to others reported in the literature as well as rates of paraplegia12,19-23.
The primary objective of this study was to analyze preoperative and intraoperative factors which may be associated with survival in the setting of aortic arch surgery at the time of extended ascending aortic replacement. By Kaplan-Meier estimates, there was a clear difference in survival between those undergoing a hemiarch versus a total arch repair (log-rank p=0.012). To examine this relationship further, a multivariable Cox proportional hazard model was created which adjusted for baseline characteristics between groups. In this analysis, the presence of a total arch repair was associated with a 2.5 time increase in the likelihood of death during the follow up period as compared to the hemiarch group (HR 2.53, 95% CI 1.38-4.62, p=0.003). In addition to type of aortic repair, increasing age was also shown to be highly associated with mortality during the follow up period – with a 1.8 time increase per 10-year increase in age (HR 1.76 per 10 years of age, 95% CI 1.37-2.28, p<0.001). Other factors of significance or borderline significance on univariable modeling were not associated with mortality on multivariable modeling (gender, use of antegrade cerebral perfusion, previous myocardial infarction, renal failure, chronic lung disease, peripheral arterial disease, hyperlipidemia, hypertension, and concomitant coronary artery bypass grafting). It is also notable that the effect of age was constant in both the hemiarch and total arch groups, as assessment of an interaction term was not significant.
The central figure of this manuscript (Figure 3) demonstrates the powerful effect of type of repair and age. This figure puts the impact of the age and type of aortic repair estimates on mortality in perspective based on our multivariable model. The predicted hazard for mortality for a 50-year-old undergoing a hemiarch repair is 17.1 (HR 17.1, 95% CI 4.8 – 61.5), while the predicted hazard for mortality for a 70-year-old undergoing the same repair is 53.3 (HR 53.3, 95% CI 18.9 – 319.4). Similarly, the predicted hazard for mortality for a 50-year-old undergoing a total arch repair is 43.4 (HR 43.3, 95% CI 10-181), while the predicted hazard for mortality for a 70-year-old undergoing the same repair is 134.9 (HR 134.9, 95% CI 20-909). One can also compare across repair types – in a 50-year-old the hazard was 17.1 for a hemiarch and 43.4 for a total arch, and in a 70-year-old, the hazard was 53.3 with a hemiarch and 134.9 with a total arch.
It was not our intention in this study to examine indication in detail as we wanted to look at a comprehensive aortic experience in the setting of extended ascending aorta replacement. However, indication is known to be a strong risk factor for mortality in other studies17,22 and was included in our multivariable modeling. While Indication was not associated with overall survival in this analysis, this variable violated the proportionality assumptions of Cox proportional hazard modeling. Therefore, our final multivariable model referenced above was “stratified” by indication. However, to confirm our primary findings that age and type of repair were highly associated with mortality and that this was not influenced by indication, three sensitivity analyses were performed (Supplemental Table 4). As can be seen in Supplement Table 4, stratification by the indication variable in the model resulted in an improvement in model fit with a lower AIC and BIC as compared to the non-stratified model (AIC 562 to 448, BIC 608 to 487). The violation of proportionality occurred in the “other” indication for surgery (infections, porcelain aortas, and aorto-esophageal fistulas, n=16). As this group is admittedly a bit “unique” we performed the analysis excluding these observations, with similar finding for presence of a total arch repair and age per 10 years (both remained highly significant, and this model had worse fit than the indication stratified model). Finally, we fit our final model separately, in the cohorts only with aortic dissections and only with aneurysms, again with similar findings that total arch replacement and age per 10 years were both associated with increased mortality. Our findings on the effect of age on overall outcomes can be supported by recent literature publications13,24.
Type of aortic arch repair at time of extended ascending aortic replacement is thought to influence the need for subsequent aortic reintervention5,13,14,24,25. In this study, we utilized competing risk with death as a competing risk to determine estimates of aortic reintervention rates. As one cannot definitively determine that those who died would not have needed aortic reintervention, considering death as a competing risk is appropriate. However, analyses of these type are a bit difficult to interpret in the context of the literature as the vast majority of studies have not utilized this methodology.
Despite a more extensive aortic repair, those undergoing total arch replacement at the time of extended ascending aorta replacement had markedly higher reintervention rates (2.6, 2.6, and 4.4% at 1, 2, and 3 years in the hemiarch group and 5.0, 10.3, and 11.9% in the total arch group). After adjustment for age (the other strong predictor in the mortality model), those with a total arch repair were 3.2 times as likely to need aortic reintervention as compared to the hemiarch group (SHR 3.21, 95% CI 1.01 – 10.2, p=0.047). Age itself was not associated with the need for aortic reintervention (SHR 0.90, 95% CI 0.65 – 1.23, p=0.50).
There are numerous limitations to this study. First, this study is subject to all the limitation of a retrospective, non-protocoled study. Second, while presence of a total arch repair and age were variables of significance, the lack of a factor as significant could be due to small sample size and a type II error. Third, we lack power to examine different types of aortic arch repair. Fourth, the lack of accurate anatomic data from preoperative and postoperative computed tomography imaging makes assessment of these variables as it relates to mortality or need for aortic reintervention impossible. Fifth, there is a potential third group of interest we did not include in this study – those who had non-extended ascending aortic replacements with an aortic cross clamp in place. Sixth, the difference in survival may be due to higher perioperative mortality in the total arch group despite appropriately modeling to assess the continued adjusted risk over time. Seventh, surgeon bias and preference largely dictated when to intervene on aortic arch in extended ascending aortic replacement and type of aortic arch repair chosen. Last, the limited number of endovascular interventions in this study make any assessment of this technology in our practice difficult, although it is being increasingly utilized.
The optimal approach to aortic arch repair in the setting of ascending aortic disease remains controversial and continues to evolve. While some advocate for extensive replacement of the aortic arch during extended replacement of the ascending aorta, others prefer a more tempered approach absent true pathology. While this study cannot directly compare the two philosophical approaches of aortic arch repair outlined above, as aortic arch repair was only performed in this study when clinically indicated, if addressing all aortic pathology present at the time of operation is adequate, we have might expect to see similar survival and reintervention rates between these two groups. On the other hand, if more isolated pathology only requiring hemiarch replacement of the aorta truly progresses, then one may see a higher reintervention and possibly mortality rate in the less aggressively managed “hemiarch” group. The fact we found higher rates of mortality and reintervention in the total arch group, when we only addressed the aortic arch when pathology was present suggests that this group represents a group of patients with disease distinct from those with isolated ascending aortic pathology and that those with arch pathology inherently have more advanced disease which is likely to both decrease survival and increase the need for further aortic reintervention. This effect was particularly prominent in those of advanced age.
In conclusion, our data suggests that mortality and reintervention rates are higher with total arch replacement in the setting of extended ascending aortic replacement, making the hemiarch procedure the preferred procedure as long as it adequately addresses the aortic pathology, especially in those of advanced age. The rationale for extending ascending aortic repair to include a total arch replacement in order to decrease reintervention may not be valid.