**3. Promises of a new approach: TAVI**

In light of increasing life expectancy, the high incidence of AS in the elderly, the fact that many SAS patients are denied surgery, and because of improved endovascular technology, a new therapeutic approach using new devices is rapidly evolving, becoming widespread and would revolutionize AS treatment: TAVI.

After successful animal experimentation, Cribier et al. (2002) successfully performed the first human implantation of a balloon-expandable aortic valve prosthesis. Two TAVI devices are now under postmarketing surveillance in Europe: the balloon expandable Edwards SAPIEN™ prosthesis (Edwards Life-Sciences, CA) and the self-expandable CoreValve Revalving prosthesis™ (Medtronic Inc, MN). Two approaches are used: anterograde with transapical (TA) access or retrograde with transfemoral (TF) access. Two videos illustrating the TAVI procedures are available at www.NEJM.com.

## **3.1 Postoperative and short-term TAVI results**

The first step in the development of these devices was to demonstrate a high rate of successful implantation that is improving with time, from 88% to >98% (Grube et al., 2007; Tamburino et al., 2011; Webb et al., 2009). At 30 days, outcome was good with 9–18% mortality (Zajarias & Cribier, 2009) for high-risk patients for whom surgery was denied (octogenarians, high Logistic EuroSCORE ≥20% (Roques & Nashef, 2003), STS-PROM ≥10% (Shroeyr et al., 2003) or specific surgical contraindication(s) (e.g. porcelain aorta, history of mediastinal radiotherapy). In all studies, the observed mortality was below those predicted by STS-PROM or Logistic EuroSCORE.

Surgery relieved symptoms with systematically lower NYHA classification and impressive decreases of the mean transvalvular gradient from 46 mm Hg preoperatively to 10 mm Hg thereafter, associated with increased aortic surface from 0.6 to 1.6 cm2, respectively, for example, in the study by Webb et al. (2009). The NYHA classification and echocardiographically detected improvement of the mean gradient and aortic surface were sustained at 1 year (Tamburino et al., 2011; Webb at al., 2009). Survival at 1 year ranged from 73.8% (Webb et al., 2009) to 85% (Tamburino et al., 2011) and compared favorably with the spontaneous SAS evolution.

Prognosis at 1 year seems mostly related to comorbidities rather than cardiac status (Webb et al., 2009). In an Italian multicenter study, independent risk factors associated with late death were: prior stroke (hazard ratio (HR) 5.4; 95% confidence interval (CI) 1.47–20.39; p=0.01), prior acute pulmonary edema (HR 2.7; 95% CI, 1.1–6.7: p=0.03), chronic kidney disease (HR 2.5; 95% CI 1.0–6.4; p=0.048) and postprocedural paravalvular leak ≥2+ (HR 3.8; 95% CI 1.6–9.1: p=0.03) (Tamburino et al., 2011).

A landmark multicenter, prospective, randomized trial compared SAS patients, who surgeons considered unsuitable surgical candidates and were given standard treatment (including BAV) or TF transcatheter implantation of a balloon-expandable bovine pericardial valve (Leon et al., 2010). At 1 year, the death rate from any cause was 30.7% with TAVI and 50.7% with standard therapy (TAVI HR 0.55; 95% CI 0.4–0.74; p<0.001).

A principal concern of TAVI for octogenarians is its impact on QOL, for which data are scarce. At 3 months postTAVI, Krane et al. (2010) used SF-36, and found significantly improved QOL concerning the physical health summarized score, while the mental health summarized score remained unchanged. For a population ≥81 years, comparison of 6 months postTAVI to preoperative SF-36 data, Bekeredjian et al. (2010) obtained significantly improved physical and mental component summary scores from 28.4±10 to 46.8±9.2 (p<0.001) and from 37.3±10.8 to 50.6±10.1 (p<0.001), respectively.

#### **3.2 Specific TAVI-related complications**

174 Aortic Valve Surgery

In light of increasing life expectancy, the high incidence of AS in the elderly, the fact that many SAS patients are denied surgery, and because of improved endovascular technology, a new therapeutic approach using new devices is rapidly evolving, becoming widespread

After successful animal experimentation, Cribier et al. (2002) successfully performed the first human implantation of a balloon-expandable aortic valve prosthesis. Two TAVI devices are now under postmarketing surveillance in Europe: the balloon expandable Edwards SAPIEN™ prosthesis (Edwards Life-Sciences, CA) and the self-expandable CoreValve Revalving prosthesis™ (Medtronic Inc, MN). Two approaches are used: anterograde with transapical (TA) access or retrograde with transfemoral (TF) access. Two videos illustrating

The first step in the development of these devices was to demonstrate a high rate of successful implantation that is improving with time, from 88% to >98% (Grube et al., 2007; Tamburino et al., 2011; Webb et al., 2009). At 30 days, outcome was good with 9–18% mortality (Zajarias & Cribier, 2009) for high-risk patients for whom surgery was denied (octogenarians, high Logistic EuroSCORE ≥20% (Roques & Nashef, 2003), STS-PROM ≥10% (Shroeyr et al., 2003) or specific surgical contraindication(s) (e.g. porcelain aorta, history of mediastinal radiotherapy). In all studies, the observed mortality was below those predicted

Surgery relieved symptoms with systematically lower NYHA classification and impressive decreases of the mean transvalvular gradient from 46 mm Hg preoperatively to 10 mm Hg thereafter, associated with increased aortic surface from 0.6 to 1.6 cm2, respectively, for example, in the study by Webb et al. (2009). The NYHA classification and echocardiographically detected improvement of the mean gradient and aortic surface were sustained at 1 year (Tamburino et al., 2011; Webb at al., 2009). Survival at 1 year ranged from 73.8% (Webb et al., 2009) to 85% (Tamburino et al., 2011) and compared favorably with the

Prognosis at 1 year seems mostly related to comorbidities rather than cardiac status (Webb et al., 2009). In an Italian multicenter study, independent risk factors associated with late death were: prior stroke (hazard ratio (HR) 5.4; 95% confidence interval (CI) 1.47–20.39; p=0.01), prior acute pulmonary edema (HR 2.7; 95% CI, 1.1–6.7: p=0.03), chronic kidney disease (HR 2.5; 95% CI 1.0–6.4; p=0.048) and postprocedural paravalvular leak ≥2+ (HR 3.8;

A landmark multicenter, prospective, randomized trial compared SAS patients, who surgeons considered unsuitable surgical candidates and were given standard treatment (including BAV) or TF transcatheter implantation of a balloon-expandable bovine pericardial valve (Leon et al., 2010). At 1 year, the death rate from any cause was 30.7% with

A principal concern of TAVI for octogenarians is its impact on QOL, for which data are scarce. At 3 months postTAVI, Krane et al. (2010) used SF-36, and found significantly improved QOL concerning the physical health summarized score, while the mental health summarized score remained unchanged. For a population ≥81 years, comparison of 6 months postTAVI to preoperative SF-36 data, Bekeredjian et al. (2010) obtained significantly

TAVI and 50.7% with standard therapy (TAVI HR 0.55; 95% CI 0.4–0.74; p<0.001).

**3. Promises of a new approach: TAVI** 

and would revolutionize AS treatment: TAVI.

the TAVI procedures are available at www.NEJM.com.

**3.1 Postoperative and short-term TAVI results** 

by STS-PROM or Logistic EuroSCORE.

95% CI 1.6–9.1: p=0.03) (Tamburino et al., 2011).

spontaneous SAS evolution.

The results of all studies demonstrated that TAVI can be implanted safely, with intraprocedural mortality now a mean of 1% (Webb et al., 2009; Tamburino et al., 2011). Implantation failure is becoming rare, with the successful implantation rate ≥98% in the most recent studies (Bleiziffer et al., 2009; Tamburino et al., 2011). Changing the surgical approach during the intervention has also become very rare.

Coronary obstruction rarely complicates valve implantation (<1%). Valve positioning is challenging, particularly when the distance between the annulus and the coronary artery is short, as the native valve may be pushed against the coronary ostium (Lefèvre et al., 2011).

Some complications seem to be related to the access route used. With TF access, major vascular complications occur in 6.8–11.7% of the cases (Bleiziffer et al., 2009; Webb et al., 2009). However, technological advances have permitted sheath-size reduction from 24F to 18F, thereby allowing a percutaneous procedure with locoregional anesthesia and fewer vascular injuries. Improved screening, case selection and experience should surely lower the vascular injury rate further.

Second, 10% of the patients suffer clinical strokes (Grube et al. 2007). In a recent diffusionweighted, magnetic resonance imaging study on TAVI patients, the risk of diffuse cerebral embolism was 72.7%, with patients frequently having multiple new but clinically silent brain lesions. Although cerebral embolism was extremely common in the TF-TAVI cohort, the clinical stroke rate was 3.6% (Ghanem et al., 2010). It was suggested that TA-TAVI might be associated with fewer cerebral embolic events. However, results are controversial and further studies with larger cohorts are needed. Two mechanisms are involved: aortic atheroemboli and valvular calcific emboli. The elderly are at particularly high risk for perioperative neurological events because of advanced cerebral ischemic disease present preoperatively (Wang et al., 2010).

For both accesses, annulus measurement is challenging. Indeed, no gold standard currently exists for aortic annular measurement but transesophageal echocardiography provides accurate data to guide valve sizing before implantation (Messika-Zeitoun et al., 2010). Paravalvular aortic regurgitation is common after TAVI but remains stable at late follow-up (Webb et al., 2009). In the PARTNER study, moderate-or-severe perivalvular leakage was present in 11.8% of the patients at 30 days and 10.5% at 1 year (Leon et al., 2010). However, postprocedural paravalvular aortic regurgitation ≥2+ (HR 3.79) mainly affected later outcomes between 30 days and 1 year (Tamburino et al., 2011).

Conduction abnormalities are frequent after TAVI (Roten et al., 2010). The occurrence of atrioventricular block requiring pacemaker insertion at 30 days seems lower with the Edwards SAPIEN™ (<5%) (Lefèvre et al., 2011) than the CoreValve™ (up to 30%) (Jilaihawi et al., 2009). One possible explanation could be that the CoreValve™ was designed to be seated lower than the SAPIEN™ valve and might compress the underlying conduction system. However, based on a cohort of 67 patients (41 received CoreValve™ and 26 SAPIEN™), Roten et al. showed that the sole independent risk factor for complete atrioventricular block after TAVI was preexisting right bundle branch block (Roten et al., 2010). Because prosthesis sizing is a critical issue, to avoid perivalvular leakage and valve migration, "over sizing" of TAVI might have increased the risk of atrioventricular block (Bleiziffer et al., 2009). Further investigations are mandatory.

Relationship Between Aortic Valve Replacement and Old Age 177

underwent surgery and 274 were treated medically, 177 of whom had BAV. At 2 years, the mortality rate was significantly lower for the surgical group vs the medically treated group, respectively, 28.1% vs 53.4% (p<0.001) (Ben-Dor et al., 2010). Another TAVI limitation is also the high incidence of severe coronary artery disease among elderly SAS patients that also

Long-term durability of TAVI must now be demonstrated. TAVI provides acceptable hemodynamic results up to 3 years (Webb et al., 2009). At present, outcomes reflect comorbidities rather than cardiac status (Tamburino et al., 2011). Patient selection is a major concern for long-term durability evaluation of TAVI. Future studies should include patients with expected life expectancy unrelated to cardiac disease >5 years. At 2 years, survival was

With the learning curve, technological improvement and better patient selection, TAVIrelated morbidity should decline, particularly concerning vascular injuries, strokes,

The next step for TAVI for SAS management should be its direct comparison with surgery for high-risk patients without contraindication(s) for surgery, and then for intermediateand low-risk patients. Historical comparisons, case–control studies, propensity-score matching (Johansson et al., 2011) or comparison with spontaneous evolution are not able to answer these questions adequately. Multicenter, randomized–controlled trials should answer them (prognosis, durability, specific morbidity, etc…) in few years, thereby allowing

**4. Frailty and geriatric evaluation of octogenarians for SAS management** 

distinguish individuals in terms of consequences of hospitalization (Gill et al., 2004).

Regardless of the definition retained, the authors defined a certain number of common characteristics (Rockwood, 2005): frailty is a continuous state that is not simply present or absent; it is not the consequence of single organ involvement; the clinical manifestations are multiple. It can be recognized clinically, with that identification being considered the threshold of entry into frailty, the occurrence of the negative event marking the end of frailty in relationship to this factor (institutionalization or handicap, for example). The definitions of frailty are multiple (Ferrucci et al., 2004), but the most used are those that refer to the diminished physiological reserves, with a core event being the development of sarcopenia. This type of definition has shown its efficacy to predict the loss of autonomy, institutionalization or death in a cohort of patients with cardiovascular diseases but not initially handicapped and followed for 1 year (Fried et al., 2004). Its effect can be associated

To evaluate better an individual's situation on which the final therapeutic decision generally depends, the concepts of frailty or vulnerability are the themes of several international gerontology publications. We highlight that, at present, no consensual definition of frailty exists (Bergman et al., 2007; Karunananthan et al., 2009). However, the various authors agree that frailty is a state of susceptibility to aggression, which explains that, for a given health event, despite the same management and apparently sufficiently similar health status, the individuals will have very different outcomes. Thus, frailty is always defined as a function of the event that serves as the judgement criterion: falls, loss of autonomy, institutionalization, death… It is also defined by the time at which it is assessed. Therefore, an effective definition over the long term (Province et al., 1995) might not be operative to

60.9% in the study by Webb et al. (2009); what will the sample size be at 5 years?

influences prognosis and requires specific treatment.

pacemaker implantation and perivalvular leakage.

**4.1 Frailty** 

the widespread development of this very promising technique.

The results of some studies showed that the postoperative and short-term outcomes were worse for patients treated with TA than TF access, because the former were more severely ill and had more severe comorbidities. The Logistic EuroSCORE predicted risk was significantly higher for TA than TF, respectively: 35% vs 25% (p=0.01), as did STS-PROM 10.3% vs 8.7% (Webb et al., 2009). In-hospital mortality was 8% for the TF group vs 27% for TA group, with respective 1-year survival rates of 74±11% vs 60±13% (Al-attar et al., 2009). At 1 year, the European PARTNER study results were comparable (Lefèvre et al., 2011).

## **3.3 Risk assessment and patient selection**

TAVI development has highlighted the complexity of risk assessment and patient selection. At present, TAVI is indicated only for SAS patients ineligible for conventional AVR, but the definition of "ineligible" remains vague. TAVI was initially designed to treat old-old patients, mean age 81–86 years (Bleiziffer et al., 2009; Tamburino et al., 2011), with high Logistic EuroSCORE predicted risk ≥20%, mean range 22–29% (Grube et al., 2007; Webb et al., 2009) or STS-PROM score ≥10%, mean range 16–23% (Al-attar et al., 2009; Bleiziffer et al., 2009) or with contraindication(s) for surgery.

Logistic EuroSCORE and STS-PROM scores are increasingly being used to estimate operative mortality based on cardiac and extracardiac factors. The STS-PROM score appears to be more reliable than the EuroSCORE for predicting outcomes of high-risk AVR patients. However, STS-PROM tends to underestimate mortality (Dewey et al., 2007). A meta-analysis showed EuroSCORE to have low discrimination ability for valve surgery and it slightly over predicted risk (Parolari et al., 2010), particularly for octogenarians referred for AVR (Leontyev et al., 2009). Both scores share the same limitations: predictive ability is limited for high-risk patients, who represented only a small proportion of the population used to derive them. They do not take into account the surgical results in a given institution, relationship between volume and mortality, and the impact of progress concerning surgical techniques, cardiopulmonary bypass, anesthesia and intensive care. The scores also fail to evaluate fully the high-risk patient because they do not integrate a very important element: clinical judgment. Indeed, many factors that negatively influence prognosis are not considered: cirrhosis, impact of body mass index, chest irradiation, chest deformation, multivalve surgery, porcelain aorta, previous CABG and vascular tortuosity for TAVI. It is estimated that neither the Logistic EuroSCORE nor STS-PROM would have classed approximately one-quarter of the patients at high risk and yet they were refused surgery because of such factors (Webb et al., 2009).

For the specific SAS octogenarian population, it seems that the development of a new specific scoring classification is necessary (Florath et al., 2010), especially one including demographic variables, such as nutritional status, disability, dementia and frailty.

#### **3.4 Unanswered questions concerning TAVI**

TAVI is not suitable for all patients ineligible for conventional AVR and patients referred for TAVI are also highly selected. In a cohort of 469 SAS patients referred for participation in a TAVI trial, 362 (77.1%) patients did not meet the inclusion/exclusion criteria. The main exclusion criteria were low STS-PROM score <10% (72 patients), peripheral vascular or aorta disease (58 patients), aortic valve area >0.8 cm2 (54 patients), significant coronary artery disease (43 patients) and renal failure (25 patients). Among those 362 patients, 75 (20.7%) had 2 exclusion criteria and 26 (7.1%) had 3 exclusion criteria. Eighty-eight patients underwent surgery and 274 were treated medically, 177 of whom had BAV. At 2 years, the mortality rate was significantly lower for the surgical group vs the medically treated group, respectively, 28.1% vs 53.4% (p<0.001) (Ben-Dor et al., 2010). Another TAVI limitation is also the high incidence of severe coronary artery disease among elderly SAS patients that also influences prognosis and requires specific treatment.

Long-term durability of TAVI must now be demonstrated. TAVI provides acceptable hemodynamic results up to 3 years (Webb et al., 2009). At present, outcomes reflect comorbidities rather than cardiac status (Tamburino et al., 2011). Patient selection is a major concern for long-term durability evaluation of TAVI. Future studies should include patients with expected life expectancy unrelated to cardiac disease >5 years. At 2 years, survival was 60.9% in the study by Webb et al. (2009); what will the sample size be at 5 years?

With the learning curve, technological improvement and better patient selection, TAVIrelated morbidity should decline, particularly concerning vascular injuries, strokes, pacemaker implantation and perivalvular leakage.

The next step for TAVI for SAS management should be its direct comparison with surgery for high-risk patients without contraindication(s) for surgery, and then for intermediateand low-risk patients. Historical comparisons, case–control studies, propensity-score matching (Johansson et al., 2011) or comparison with spontaneous evolution are not able to answer these questions adequately. Multicenter, randomized–controlled trials should answer them (prognosis, durability, specific morbidity, etc…) in few years, thereby allowing the widespread development of this very promising technique.
