Gender Differences and Mortality Trends After Transcatheter Aortic Valve Implantation: A 10-Year Analysis From a Single Tertiary Center

Aim. To evaluate gender differences and mortality trends in a population undergoing transcatheter aortic valve implantation (TAVI) and to analyze the correlates to all-cause mortality at follow-up. Methods. The study comprises a prospective cohort of 592 TAVI patients (53.4% female) treated between 2008 and 2018. Mortality differences between genders at different timepoints were assessed according to log rank test. Predictors of all-cause mortality at follow-up were identified using a univariate model and were then analyzed through multivariate Cox proportional hazard models. Results. Compared with female patients, males were younger (81 ± 7.5 years vs 84.3 ± 5.3 years) and presented more comorbidities. Twelve female and 8 male patients (3.5%) died in the first 30 days after TAVI. Despite a higher Society of Thoracic Surgeons (STS) score in women, all-cause mortality rates at 30 days and 1 year were comparable. At long-term follow-up, female patients demonstrated better survival rates, despite a higher number of periprocedural complications. Correlates identified in men were the presence of diabetes and previous history of coronary artery bypass grafting, New York Heart Association class III/IV, pulmonary artery systolic pressure, and non-transfemoral access. None of these variables remained significant in the multivariable analysis. In females, only peripheral artery disease was associated with mortality. Shock and need for renal replacement were predictors of mortality in both genders, as was heart failure readmission after discharge. STS score was also shown to correlate with long-term mortality in both genders. Conclusion. Despite a higher STS score in women, 30-day mortality was not significantly different from men, while women present better clinical outcomes at long-term follow-up.

J INVASIVE CARDIOL 2021;33(6):E431-E442. Epub 2021 May 6.

Key words: aortic stenosis, gender differences, TAVI mortality, transcatheter aortic valve implantation

Aortic stenosis is the most common valvular heart disease in developed countries, and its impact on public health and healthcare resources is expected to increase due to the aging population.¹

Transcatheter aortic valve implantation (TAVI) is now an alternative to conventional cardiac surgery therapy for patients presenting symptomatic aortic stenosis in all settings of risk. In the European countries, predicted numbers point to 114,757 TAVI candidates annually and 58,556 in North America.²

Percutaneous valve treatment offers substantial reductions in mortality and improvement in quality of life compared with medical therapy^3,4 and has shown at least similar long-term outcomes compared with surgical aortic valve replacement (SAVR)^5-7 even in patients presenting low surgical risk.^8-10 Current mortality in TAVI trials with new-generation valves ranges from 1% to 2.4%^8,9 vs 3.5% in SAVR¹¹ in patients at low surgical risk, and 10%¹² in TAVI vs >12% in SAVR patients at high risk.¹³ Long-term follow-up of TAVI cases from real-world tertiary European centers is not available, particularly in cohorts in which female patients are the majority. Female gender has been associated with higher operative mortality after SAVR,¹⁴ yet previous reports are inconsistent regarding the effect of gender on outcomes,¹⁵ with the most frequent explanation being differences in preoperative risk profiles after propensity-matched assessment.^16-18 These findings do not seem to be extrapolatable for TAVI patients,¹⁹ so we aimed to evaluate mortality trends over time and to identify correlates to mortality, outlining the differences between female and male patients.

Population. A largely female prospective cohort of 592 TAVI patients (53.4% female), treated between April 2008 and December 2018 was evaluated. The patients were enrolled in the VCROSS (Valve Catheter Restorative Operation on Santa Cruz hospital) registry, which was created to obtain all demographic, clinical, procedural, and follow-up data from patients undergoing TAVI in our hospital, for severe aortic stenosis.

All patients signed procedural informed consent approved by the ethics committee. Mortality status was collected from electronic medical records and/or civil registries through November 2019. Complications and follow-up data were obtained from patient’s clinical files (handwritten and electronic records) or medical national platform records when available.

Patients were considered for TAVI if they had symptomatic severe aortic stenosis, defined by a jet velocity >4.0 m/s, mean gradient >40 mm Hg, and valve area <1.0 cm² or index <0.6 cm²/m², being referred to our hospital from primary or secondary centers. Indication for TAVI was assessed by a multidisciplinary heart team comprised of clinical and interventional cardiologists together with cardiac surgeons.

Procedure. Coronary artery disease was ruled in or out either by coronary computed tomography angiography or invasive coronary angiography. Revascularization was performed after the heart team’s decision for percutaneous approach, before TAVI procedure, whenever there was proof of ischemia or a >70% stenosis on a proximal or mid segment from one of the main branches (>50% for left main).

Transapical and subclavian procedures were performed by a hybrid team with interventional cardiologists and cardiac surgeons and transfemoral mainly by interventional cardiologists with background support from cardiac surgery.

Percutaneous valves included CoreValve, CoreValve Evolut R, and CoreValve Evolut Pro (Medtronic); Sapien 3, Sapien XT, Sapien 3 Ultra (Edwards Lifesciences); Portico (Abbott Vascular); and Lotus and Acurate Neo (Boston Scientific), and were implanted on both native and prosthetic valves.

After 2016, the procedures were performed mainly under conscious sedation and analgesia, with use of anesthesia for cases guided by transesophageal echocardiogram or when severe complications occurred during valve implantation. Pacing for balloon-expandable valves was performed on the left ventricular wire and through a temporary transvenous pacing in cases where conduction disturbances like right bundle-branch block were already present, or with the patient’s own pacemaker. Vascular closure devices used were Prostar and ProGlide (Abbott Vascular), AngioSeal (Terumo), and Manta (Teleflex).

After TAVI, patients were admitted to a cardiac care unit for the first hours until clinically stable to move to the cardiology ward with continuous electrocardiographic monitoring.

Endpoints. The primary endpoint was all-cause mortality at 30 days and 1 year, and at a median follow-up of 24 months. Secondary endpoints were in-hospital complications, heart failure readmission, myocardial infarction (MI), stroke, and bleeding > type 2 according to the Blood Academic Research Consortium (BARC) definition²⁰ after discharge.

Definitions. All definitions were obtained from PARTNER III⁸ or were otherwise specified. Chronic kidney disease was defined as baseline creatinine clearance <45 mL/kg/min as calculated by Cockroft-Gault equation and chronic pulmonary disease was defined as an established diagnosis of chronic obstructive pulmonary disease stage II or higher.

Reported in-hospital complications were: (1) emergent vascular surgery or percutaneous intervention due to vascular closure device failure; (2) femoral artery pseudoaneurysm diagnosed within 1 month after TAVI with or without the need for vascular surgery; (3) emergent cardiac surgery, defined as conversion to open sternotomy during the TAVI procedure secondary to any procedure-related complications, such as annular rupture or aortic dissection; (4) pericardial effusion, referring to evidence of a new pericardial effusion associated or not with hemodynamic instability and clearly related to the TAVI procedure; (5) > type 2 BARC bleeding with or without (6) hemorrhagic shock, defined by rapid and significant loss of intravascular volume, which may lead sequentially to hemodynamic instability, with decreases in oxygen delivery and tissue perfusion, cellular hypoxia, organ damage, and death; (7) cardiogenic shock, defined by decreased cardiac output (cardiac index <2.2) with evidence of end organ damage and increased filling pressures) in the presence of adequate intravascular volume, due to a cardiac cause; (8) myocardial infarction; (9) stroke; (10) need for mechanical invasive ventilation due to respiratory distress, pulmonary edema, or need of deep anesthesia due to procedure-related complications; (11) need for renal replacement therapy on patients with a previously preserved urinary output, caused by an acute renal injury (Acute Kidney Injury Network stage 3); (12) cardiac resuscitation due to cardiac arrest; and (13) in-hospital death, defined as death within the day the patient is admitted for valve implant procedure until the patient is discharged from the hospital.

Cardiovascular mortality was defined by proximate cardiac cause, all procedure- and device-related deaths including valve dysfunction, or other valve-related adverse events and non-coronary vascular conditions, such as neurological events, pulmonary embolism, ruptured or dissecting aneurysm, or death preceded by cardiovascular symptoms (such as chest pain). Non-cardiovascular mortality was defined as any death that was due primarily to an identifiable non-cardiovascular cause or etiology. Death of unknown cause was defined by unwitnessed death or unprovided information.

MI was defined as an ischemic event associated with clinically significant myocardial necrosis, such as new ischemic symptoms (eg, chest pain, shortness of breath) or signs (new ST-segment changes, imaging evidence of new loss of viable myocardium) and elevated cardiac biomarkers consisting of at least 1 sample with a peak value exceeding the 99th reference limit of high sensitivity troponin T and a significant 3-hour delta (∆) or a further increase of at least 50% post procedure for patients with baseline elevated values. Postprocedural MI was defined if the previous occurred within 72 hours from the procedure. Type 2 MI was defined according to the 4th Universal Definition of MI European Guidelines,²¹ requiring evidence of an imbalance between myocardial oxygen supply and demand unrelated to acute coronary atherothrombosis.

Heart failure readmission was defined as hospitalization >24 hours with clinical symptoms and objective signs of heart failure, including dyspnea, pulmonary edema, hypoperfusion, or anasarca, requiring a medical intervention like administration of intravenous diuretics, inotropic therapy, or institution of respiratory mechanical support. Bleeding was classified according to BARC criteria.²⁰

Statistical analysis. Continuous variables were summarized as mean ± standard deviation or median (1st quartile, 3rd quartile) and compared with analysis of variance or Kruskal-Wallis test based on data distribution. Binary and categorical variables were calculated as frequencies (proportions) and were compared with the Chi-square test or Fisher’s exact test if expected cell counts were <5. Survival curves were plotted for time-to-event variables with Kaplan-Meier estimates and compared using the log-rank test. Univariate and multivariate Cox proportional hazard models were used to calculate (un)adjusted hazard ratios (HRs) at follow-up. To determine covariates included in multivariate analysis, we performed univariate analyses for all patient-related baseline characteristics, procedural details, periprocedural complications, and postdischarge complications, namely, heart failure admission, stroke, MI, and bleeding. Variables evaluated are described in Table 1. We then used a Cox regression adjusted for covariates with a P-value <.20 at univariate analyses. For the Cox regression, 5 different models were used: (1) patient baseline and procedural characteristics; (2) including STS score and excluding overlapping characteristics (3) similar to the previous, using EuroSCORE II instead; (4) STS score; and (5) in-hospital complications. Model results are presented as HR, 95% confidence interval (CI), and P-values. Due to a low number of in-hospital complications resulting in wide CIs, HRs for these factors were not considered individually. The P-values were 2-sided and <.05 was considered statistically significant in all analyses. Statistical analyses were performed using SPSS Statistics, version 21 (IBM).

Patient characteristics are provided in Table 1. In the overall population, significant differences between gender were found with regard to age (81 ± 7.5 years in men vs 84.3 ± 5.3 years in women), as well as smoking habits and peripheral artery disease, chronic kidney and lung disease, previous MI and coronary artery bypass grafting, which were all significantly more prevalent in men. Notwithstanding, STS score was higher in women than in men (5.4% [IQ, 3.0%] vs 4.1% [IQ, 3.1%], respectively; P<.001).

30-day and 1-year mortality rates. Twenty patients died (3.5%; 12 women and 8 men) in the first 30 days after TAVI (Figure 2A and Figure 3) despite the higher STS score in women. One-year mortality was 13% (77 patients; 35 women and 42 men), a rate that is statistically comparable between both genders (log-rank P=.11) (Figure 1A and Figure 2A.)

Longer-term mortality rates. Over a median follow-up of 23 months (IQ, 21 months), women demonstrated better overall survival rates (70.3% vs 60.5% in men; P=.01; log-rank P<.01) (Figure 1B and Figure 2B) suggesting a benefit in this population despite the higher absolute number of periprocedural complications (P=.08) (Table 1).

Correlates to mortality. In the univariate analysis, different correlates to all-cause mortality were identified in men and women (Table 2). When using a Cox regression model, none of the independent variables was correlated to death and only STS score predicted long-term mortality (P<.001 for both women and men; HR, 1.12; 95% CI, 1.06-1.19 and HR, 1.11; 95% CI, 1.06-1.17, respectively).

When evaluating in-hospital complications and accounting for STS-predicted risk of mortality, the multivariable model identified hemorrhagic shock and cardiac resuscitation as independent predictors of long-term mortality in women, cardiogenic shock and need for prolonged invasive ventilation in men, and the need for renal replacement therapy in both genders.

Mortality trends. Thirty-day mortality seemed to decrease as the number of procedures increased (a 50% reduction in 30-day mortality from 4% in the first two-thirds of the patients to 2% in the last one-third) and it was lower than expected according to available risk predicting scores (Figure 2A).^22,23 The main 30-day mortality cause was cardiac (14/20 deaths; 70%), mostly because of refractory left ventricular  systolic dysfunction leading to multiorgan failure and death (5/20 deaths; 25%) and left ventricular rupture (3/20 deaths; 15%), but also because of massive aortic regurgitation (1/20 deaths; 5%), acute mitral regurgitation due to anterior leaflet take up by the device (1/20 deaths; 5%), acute MI (2/20 deaths; 10%), and arrhythmias (cardiac arrest preceded by ventricular fibrillation in 1/20 deaths; 5%) (Table 3).

In the first subset of 100 patients treated, corresponding to the first 3 years of the program, mortality at 1 year increased and reached 20%, which likely reflects the inclusion of higher-risk patients, as confirmed by the highest STS scores (Figure 3). After that landmark, similar to 30-day mortality, mortality at 1 year also decreased from 20.0% to 10.9% in the last 100 patients (18.2% in men and 4.2% in women).

Our results demonstrate that despite higher risk scores, female patients present short-term mortality similar to male patients and have better long-term survival after TAVI. Correlates to mortality identified in men and women are different, and contemporary risk scores may be inadequate to predict short- and long-term outcomes after TAVI in new lower-risk populations, especially when considering female patients. Trend evaluation shows a significant reduction in 30-day and 1-year mortality during the 10-year period of this study (Figure 3), despite no significant variation in the population’s predicted risk, reflecting the improved patient clinical assessment (not mirrored in scores), in the technology and in operator performances, contributing to a significant improvement on patient outcomes.

In a large cohort of 1011 patients undergoing TAVI over a period of 8 years, Kesteren et al²⁴ reported a reduction of 50% in 1-year mortality between 2009 and 2016. This change was mainly attributed to a reduction in 30-day mortality and on the fact that STS-predicted risk of mortality also significantly decreased. In our cohort, STS score did not reduce significantly as we included patients at extreme high risk during the entire registry, but the rates of 30-day and 1-year mortality did decrease (Figure 2). This suggests that improvements in patient clinical assessment and multidisciplinary treatment assignment contributed to better outcomes, as did technological advances in the valves and delivery systems, increasing operator experience and expertise, and procedure-related technical improvements.

To better understand this progress, one should start by analyzing short-term mortality. Terzian et al²⁵ reported a rate of 30-day mortality of twice the one observed in our cohort (7.5% vs 3.5%, respectively) in 600 patients between 2006 and 2014, with the deaths occurring in the catheterization laboratory or during the first 24 hours being the most frequent in agreement with our findings (33% and 38%, respectively) (Table 3). A decreasing rate of procedural death during the 8-year period was observed as in our experience, although no statistical analysis could be performed due to the low number of events. This improvement was mainly driven by a decrease in cases of postprocedural refractory congestive heart failure and major access complications, but also in the rate of cardiac rupture, reflecting better procedural planning and execution. To additionally enhance the role of operator experience on the outcomes, Beohar et al²⁶ compared the third tercile intervened vs the first tercile of a cohort with 1063 patients undergoing transfemoral TAVI in 2011. In our experience, device technology and procedural success improvement throughout the years allowed for a decrease in 30-day and long-term follow-up mortality, which was mainly attributed to better patient selection, valve sizing, and deployment. All-cause death at 1 year in this report was similar to the rate observed in our population (14.4% vs 13.0%, respectively).

Clinical assessment is a baseline step to define patient suitability and potential benefit for intervention, and is still a matter of debate. Different correlates have been identified in distinct populations. Recently, Overtchouk et al²⁷ studied a cohort of 11,469 patients and identified male sex, history of atrial fibrillation, and chronic renal failure as the strongest independent correlates to mortality. Beohar et al²⁸ identified serum creatinine, liver disease, coagulopathy, mental status, BMI, male gender, and STS score. Sabaté et al²⁹ also added peripheral vascular disease, lower ejection fraction, the need to convert to surgery, and at least moderate aortic regurgitation after TAVI as independent predictors of in-hospital and short-term (<1 year) mortality. These correlates diverge not only according to the population in the study, but also with the timing of the assessment (pre, peri, and post procedure), making it difficult to standardize risk definition tools. Furthermore, differences in gender have been outlined and start to suggest that different thresholds for TAVI should be established for female patients.^30,31

Chieffo et al³² described a 1-year mortality of around 12.5% in a female population at intermediate-to-high risk, which is not much higher than the rate observed in our female population, identifying EuroSCORE I, baseline atrial fibrillation, and prior percutaneous coronary intervention as independent predictors of 1-year death or stroke. The modest performance of contemporary risk scores in predicting early and long-term mortality after TAVI has been described,^33,34 and in the present study, only STS (and not EuroSCORE II) was an independent predictor of mortality in both men and women (HR, 1.11 for men; HR, 1.12 for women).

No gender differences were observed in our cohort regarding survival at 30 days³⁰ or 1-year mortality,^19,30,35 contrary to previous findings,^36,37 which is perhaps due to the low rate of events observed in our population. Multiple registries point toward a higher number of procedural complications in women and we also observed significantly higher rates of procedural vascular complications^32–35 and pericardial effusion,^37-40 although these findings are inconsistent in higher-risk populations.⁴¹ The differences in complications and cardiovascular events between men and women were nullified after hospital discharge. When considering longer-term follow-up, differences in mortality are apparent, with women surviving longer than men. Even considering the fact that, at any fixed age, women have a longer life expectancy than men, this would probably not explain the large significant difference between genders found in the study. Baseline characteristics differ between populations.^{19,31,35,40,42} In our population, women were older and had a lower prevalence of coronary artery disease, peripheral artery disease, and chronic kidney disease, as well as higher baseline left ventricular ejection fraction,^39,42 paradoxically presenting a higher mean STS.^36,37 The effect of age in contemporary risk scores may become obsolete as older patients with fewer comorbidities are treated, and younger patients with higher-risk profiles are considered for TAVI. In our analysis, age was not a significant predictor of mortality.³⁹ Correlations to lower mortality have been established, eg, between lower prevalence of coronary artery disease, diabetes, and atrial fibrillation,⁴³ and higher baseline left ventricular ejection fraction.⁴⁴ In our population, these variables were not independent predictors when interpreted out of a multiparametric assessment like the STS score. Non-transfemoral access was significantly different between patients deceased or alive at follow-up (Table 4), more frequently used in male patients, and it was an independent predictor for mortality (P<.01 in men).^39,42 No differences were seen in mortality regarding the mechanism of the valve.⁴²

Study limitations. After the 1-month follow-up visit, some of the study patients continued medical follow-up at referral hospitals. Due to patient data-protection laws, only live status could be assessed in every patient; thus, under-reporting of long-term follow-up cardiovascular events may have occurred. The cause of death was inaccessible in a percentage of patients, resulting in under-reporting of cardiovascular deaths.

Despite a higher STS score in women in a population undergoing TAVI, 30-day mortality was not significantly different between men and women, and it was largely lower than predicted. At long-term follow-up, women presented better outcomes.

Acknowledgments. The authors would like to thank Francisco Gama, MD; Gustavo de Sá Mendes, MD; Catarina Brízido, MD; Pedro Lopes, MD; Bruno Rocha, MD; Gonçalo Cunha, MD; João Presume, MD; Francisco António de Albuquerque, MD; and Sérgio Maltês, MD for their contributions to data collection.