Ethics approval

This randomized controlled trial was approved by the Institutional Review Board of Seoul National University Hospital (#1707-109-871, on September 11, 2017) and registered at clinicaltrials.gov (NCT03291210, on September 25, 2017) before patient enrollment. The study was conducted according to the Good Clinical Practice guidelines and Declaration of Helsinki. Written informed consent was obtained from all participants, who could withdraw at any time.

Study population and randomization

Adults (aged 20–99 years) with aortic stenosis (AS), undergoing elective TAVI under general anesthesia via the transfemoral approach in a single tertiary academic center (Seoul National University Hospital, South Korea), were eligible for the study. Eligibility for TAVI was determined based on the consensus of an institutional multidisciplinary heart team, including clinical cardiologists, cardiac interventionists, cardiac surgeons, imaging specialists, and anesthesiologists, and followed the current practice guidelines [12,13]. The predicted operative mortality risk was calculated using the Society of Thoracic Surgeons Predicted Risk of Mortality (STS-PROM) score, European System for Cardiac Operative Risk Evaluation (EuroSCORE) II, and logistic EuroSCORE. The heart team determined the anesthetic method (general anesthesia or conscious sedation) based on the patients’ comorbidities, preference, and ability to maintain a supine position without profound dyspnea or restlessness during the procedure. The exclusion criteria were a non-transfemoral approach, pre-procedural arterial partial pressure of oxygen (PaO₂) <65 mmHg or receiving oxygen treatment, severe pre-procedural renal dysfunction (defined as an estimated glomerular filtration rate [eGFR] <30 mL/min/1.73 m²), chronic pulmonary obstructive disease or symptomatic asthma, tuberculosis-destroyed lung, history of lung cancer, acute coronary syndrome within the past 6 months, documented pre-procedural cardiac troponin I (cTnI) or creatine kinase-myocardial band (CK-MB) elevation, stroke or transient ischemic attack within 6 months, pregnancy, and refusal to participate.

After enrollment and the informed consent process, the patients were randomized to receive F_IO₂ 0.3 or 0.8 during TAVI (1:1 allocation ratio) (Fig 1). Block randomization (blocks of four or six) was conducted using a computer-generated program (Random Allocation Software, ver. 2.0; software.informer.com) by an independent research nurse (Y.L) on the morning of the intervention. The group assignments according to the randomization list were concealed in an opaque envelop by an independent nurse, and the concealed envelop was opened by an anesthesiology in charge just before anesthesia induction while patients, interventionists, and data analyzers were blinded to the group allocations. Involved patients and interventionists could not see the F_IO₂ settings as they were blinded with a screen. However, anesthesiologists in charge of intraprocedural patient management could not be blinded as they monitored and controlled F_IO₂ during the procedure. Group allocation concealment was kept until data analyses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. CONSORT diagram.

CK-MB, creatine kinase-myocardial band; COPD, chronic obstructive pulmonary disease; F_IO₂, fraction of inspired oxygen; PaO₂, arterial partial pressure of oxygen.

https://doi.org/10.1371/journal.pone.0281232.g001

Study protocol

The routine monitoring techniques of our institution for patients under general anesthesia were applied, except F_IO₂ management. Without premedication, 12-lead electrocardiogram, pulse oxygen saturation (SpO₂), invasive and noninvasive arterial blood pressure, cerebral oxygen saturation (ScrbO₂), and bispectral index monitoring were performed. Left and right ScrbO₂ were measured using near-infrared spectroscopy (Somanetics INVOS oximeter; Covidien, Mansfield, MA, USA). Transesophageal or transthoracic echocardiography was performed to evaluate the valve position and presence of paravalvular regurgitation, as required by the interventionists.

Before inducing anesthesia, all participants were preoxygenated for 3 min using an anesthesia machine (Primus; Drägerwerk, Lubeck, Germany) with F_IO₂ 0.3 or 0.8 according to the group allocation. After stabilization, general anesthesia was induced and maintained by a target-controlled infusion of propofol (effect-site concentration [Ce], 2.5–4.0 μg/mL) and remifentanil (Ce, 1.0–4.0 ng/mL) using a commercial infusion pump (Orchestra, Fresenius Vial, Brézins, France). Neuromuscular blockade was established by administering rocuronium (0.6 mg/kg). Then the trachea was intubated and the lungs were ventilated in volume-controlled mode with a tidal volume of 0.6–0.8 mL/kg and ventilatory rate of 9–12 /min to maintain end-tidal CO₂ 35–45 mmHg. The alveolar recruitment maneuver was performed at 25 cmH₂O for 10 s after tracheal intubation, and a positive end-expiratory pressure of 5 cmH₂O was applied in all patients. According to the group assignment, F_IO₂ was maintained at 0.3 or 0.8 until the end of the TAVI procedure, unless the SpO₂ was <93%. If desaturation occurred, F_IO₂ was increased by 0.05–0.1, and an additional alveolar recruitment maneuver was performed as needed to maintain SpO₂ ≥93% by the attending anesthesiologists. Treatment of desaturation was triggered by low SpO₂ rather than PaO₂ from ABGA, as SpO₂ deterioration could be recognized promptly, and immediate intervention could be delivered. On completing the procedure, 100% O₂ was provided to all patients during anesthesia emergence. Patients were extubated in the intervention room, monitored in the cardiovascular care unit for 1–2 days, and then transferred to a general ward. Patients were discharged 5–7 days post-TAVI if they had no procedure-related complications.

The TAVI was conducted in accordance with the standard procedures in our institution. Using a transfemoral approach, a balloon-expandable Sapien III valve (Edwards Lifesciences, Irvine, CA, USA), self-expandable Evolut Pro or R valve (Medtronic, Minneapolis, MN, USA), or Lotus valve (Boston Scientific, Natick, MA, USA) was implanted at the diseased aortic valve. The valve was chosen by the heart team based on the size and structure of the native valve and sinus, heights of the coronary artery openings, and considerations regarding future coronary access, the risk of conduction disturbances, and annular calcification. The iliofemoral arteries were accessed under fluoroscopic guidance and closed percutaneously using Perclose ProGlide vascular suture-mediated closure devices (Abbott Vascular Devices, Redwood City, CA, USA). Before the procedure, the patients were given loading doses of dual antiplatelet agents: acetylsalicylic acid and clopidogrel (both 300 mg). During the procedure, the patients were heparinized with unfractionated heparin to achieve an activated clotting time >250 s. At completion of the valve implantation, the effects of heparin were reversed by protamine infusion.

During the procedure, arterial blood gas analysis (ABGA) was performed at four time points: baseline (before anesthesia induction, T1), after inducing general anesthesia (T2), after valve implantation (T3), and at the end of the procedure (T4). ABGA was performed using a GEM® Premier 3000 device (Model 5700; Instrumentation Laboratory, Lexington, MA, USA). Oxygenation variables, including PaO₂, arterial oxygen saturation (SaO₂), SpO₂, and ScrbO₂, were recorded at the same time with ABGA measurements. During the procedure, transfusion was triggered to maintain hematocrit 21–24% or by clinical judgement of anesthesiologists in charge based on ongoing blood loss and patients’ comorbidity. Trigger to treat hypotension was systolic blood pressure <90 mmHg or ≥20% drop from baseline.

Two serum cardiac biomarkers of myocardial injury, high-sensitivity cTnI (hs-cTnI) and CK-MB, were measured at baseline (before the procedure) and 1, 4, 8, 24, 48, and 72 h after TAVI. hs-cTnI was measured using an Abbott Architect Plus Analyzer (i2000SR; Flex, San Jose, CA, USA), which has a limit of detection of 0.0011 μg/L and limit of blank of 0.0007–0.0013 μg/L. An hs-cTnI concentration ≥99^th percentile in the normal population (0.028 μg/L) was deemed abnormal. Serum creatinine concentrations were calibrated using isotope dilution mass spectrometry (IDMS). To evaluate the baseline kidney function, we calculated eGFR using the modified diet in renal disease (MDRD) Eq [14] as a surrogate of GFR, which was routinely adopted in our institution during the study period. Although Chronic Kidney Disease Epidemiology (CKD-EPI) creatinine equation had better performance compared with MDRD equation for high levels of GFR, equal accuracy has been observed when GFR is <60 mL/min/1.73 m² [15].

Postprocedural acute kidney injury (AKI) was determined based on the serum creatinine level and urine output according to the Kidney Disease: Improving Global Outcomes Clinical Practice Guidelines criteria for AKI [16]. AKI was defined as an increase in serum creatinine ≥1.5 times the baseline level or by ≥0.3 mg/dL (≥26.5 μmol/L) [AKI_creatinine], or a urine output <0.5 mL/kg/h for ≥6 h within 7 days [AKI_{urine output}]. AKI occurring >7 days after the procedure was excluded because it might have been unrelated to the procedure. Acute kidney recovery (AKR) was defined as an increase in eGFR of ≥25% relative to baseline at 48 h post-TAVI [17]. Post-TAVI AKR has been acknowledged as a potential benefit following improvement of cardiac output with relief of aortic stenosis, and was observed in up to 1/3 of patients undergoing TAVI, occurring more frequently than AKI [18] and associated with improved survival than those who developed AKI [19].

The postprocedural development of new conduction abnormalities and incidence of permanent pacemaker insertion was assessed. Stroke was defined as an acute episode of a focal or global neurological deficit as a result of hemorrhage or infarction, based on the Valve Academic Research Consortium-2 (VARC-2) definition [20]. Periprocedural MI was defined based on a combination of new ischemic symptoms or signs and elevated cardiac biomarkers within 72 h following TAVI, according to the VARC-2 definition [20].

Postprocedural pulmonary complications, including reintubation, prolonged mechanical ventilation, and pneumonia, occurred within 7 days following the procedure or until discharge were assessed. Prolonged mechanical ventilation was defined as requirement of mechanical ventilation for more than 12 h after procedure. Pneumonia was defined using the Centers for Disease Control and Prevention (CDC) criteria [21]. The CDC definition includes progressive infilrates, consolidation, or cavitation on chest radiography; either fever (>38°C), leukopenia or luekoytosis, or altered mental status; and sputum changes suggesting infection, worsening cough or dyspnea, rales or bronchial breath sounds, or worsening gas exchange (hypoxemia, increased oxygen requirements, or increased ventilator demand) [21].

Study endpoints and sample size calculation

The primary study outcome was periprocedural myocardial injury, as reflected by the geometric area under the curve (AUC) for periprocedural serum hs-cTnI in the first 72 h post-TAVI, calculated according to the trapezoidal rule. Secondary outcomes were the AUC for serum CK-MB in the first 72 h post-TAVI, and the peak serum hs-cTnI and CK-MB levels in the same period. Post-procedural clinical outcomes were also evaluated, including AKI, AKR, new conduction abnormalities, permanent pacemaker insertion, stroke, MI, pulmonary complications, in-hospital cardiovascular mortality, and hospital length of stay.

To calculate the sample size, we conducted a pilot study of 10 patients undergoing transfemoral TAVI under general anesthesia. The AUC for serum hs-cTnI in the first 72 h after TAVI was 40.24 ± 28.16 ng/mL. To detect a 50% difference in hs-cTnI levels between the two treatment groups in the first 72 h, which was chosen to be clinically relevant by the study investigators, 32 patients were required for each group at 80% power and an alpha error of 5% when compared using an independent t-test using G*Power software package (ver. 3.1.9.2; Franz Faul, Universitat Kiel, Germany). Considering a 10% dropout rate, we recruited 36 patients to each group (a total of 72 patients).

Statistical analysis

Data are presented as the mean ± SD, median (interquartile range, IQR), or number (%) according to the data distribution. The primary endpoint, the AUC for serum hs-cTnI in the first 72 h post-TAVI, was analyzed using the independent t-test assuming the equal variance. For primary and secondary endpoints, sensitivity analysis was performed by using bootstrap inference for multiple imputation in the intention-to-treat dataset (n = 72). The bootstrap for multiple imputation was carried out with 2,000 bootstrap replicates and 10 multiple imputations by using functions bootMice and bootImputeAnalyse in the R package bootImpute. Missing values were replaced by using multivariate imputation by chained equations and an imputation model for each endpoint included the F_IO₂ group and baseline variables that achieved statistical significance at p value 0.2 via stepwise variable selection procedure in a linear regression model for each endpoint because of many variables (43 variables) compared to the number of observations.

Other continuous variables were analyzed using the independent t-test or Mann–Whitney U test according to the data distribution. Categorical variables were analyzed using Pearson’s chi square test or Fisher’s exact test. For repeated measures data, the groups were compared using a linear mixed-effects model, which included independent fixed effects for group, measurement time, and their interaction, and a random effect for subject (a random intercept), with a compound symmetry covariance structure. When the interaction between group and time was significant, mean difference between the groups at each measurement time (after induction, at valve implant, and at the end of procedure) was estimated by using linear contrast in the linear mixed-effects model and the p value from the linear contrast test was multiplied by 3 for Bonferroni correction for multiple comparisons. The analysis was done in an intention-to-treat manner. All analyses were performed using IBM SPSS Statistics (ver. 21.0; IBM Corp., Armonk, NY, USA) or R software (ver. 3.5.1; R Development Core Team, Vienna, Austria). A p value <0.05 was considered statistically significant.

Myocardial injury after TAVI

Even after successful TAVI, the cardiac biomarkers cTnI and CK-MB showed post-procedure increases despite prompt relief of transvalvular pressure overload, and periprocedural myocardial injury occurred along with transient deterioration in myocardial function [22]. Significant deterioration in the myocardial performance index, which implies both systolic and diastolic dysfunction, was observed immediately following TAVI [22]. New myocardial late enhancement with an ischemic pattern, indicating myocardial damage, was detected on cardiac magnetic resonance images following both balloon-expandable and self-expandable valve implantation [23].

Transient LV dysfunction and injury following TAVI seems to be partly influenced by procedural aspects of transcatheter valve deployment [24]. Procedure-related mechanical trauma during TAVI, including during balloon valvuloplasty, valve positioning, and prosthesis delivery, also plays a substantial role in myocardial damage [25]. Rapid ventricular pacing is used to temporarily reduce the LV output during pre-implantation balloon valvuloplasty and balloon-expandable valve implantation, and for post-implantation ballooning to reduce paravalvular leakage. Rapid ventricular pacing transiently reduces microvascular tissue perfusion and the flow index in small- and medium-sized vessels, and induces partial microcirculatory arrest and delayed recovery of microflow [26]. Subsequently, ventricular stunning and subsequent dysfunction may occur [24].

Hyperoxia and myocardial injury

The role of oxidative stress in reperfusion injury is relatively well established. Abrupt oxidative reactions following reperfusion produce reactive oxygen species (ROS) from cardiomyocytes and endothelial cells, which amplifies local inflammatory responses and leads to a vicious cycle of ROS production [27]. The biological mechanism underlying the adverse effects of hyperoxia is related to the generation of ROS, specifically the superoxide anion, which has a negative impact on coronary blood flow and LV distensibility [28]. Hyperoxia can exacerbate oxidative stress and thereby worsen coronary vasoconstriction and myocardial injury. Interestingly, hyperoxic reperfusion limited myocardial necrosis in rodents with cardiovascular risk factors more so than in a normoxemic reperfusion group, while the reverse occurred in healthy rodents [27]. Similarly, in a preliminary canine MI model, administering 100% oxygen had beneficial effects on the myocardium by reducing myocardial infarct size and improving the EF after reperfusion compared to room-air ventilation [29].

However, in the AVOID trial, patients presenting with acute MI were randomized to receive oxygen 8 L/min via face mask or ambient air, and oxygen treatment increased myocardial injury and infarct size in patients without hypoxia [30]. In patients admitted to the intensive care unit (ICU), conservative use of oxygen, which aimed to maintain arterial oxygen tension within the physiological range, reduced ICU mortality compared to conventional use of oxygen [31]. In the large randomized DETO2X-AMI study, there was no difference in 1-year mortality or the peak cardiac troponin level between patients with suspected MI who received supplemental oxygen versus ambient air [32]. During general anesthesia for major non-cardiac surgery, there was no difference in myocardial injury—assessed using the AUC for high-sensitive troponin in the first 3 postoperative days—between F_IO₂ 0.3 and 0.8 administered intraoperatively and for 2 h after surgery [3]. In the ICU-ROX study, conservative use of F_IO₂ (≤0.21) during mechanical ventilation in adult ICU patients resulted in no difference in the number of ventilator-free days compared to standard administration of F_IO₂ [33]. In a more recent nationwide registry trial, high oxygen supplementation (6–8 L/min by face mask) resulted in no significant difference in the 30-day or 1-year mortality rate in patients with suspected acute coronary syndrome compared to low oxygen treatment [34].

Myocardial injury and clinical outcomes after TAVI

Cardiac biomarkers elevation following surgery or intervention result from perioperative hemodynamic stress, inflammation, or oxygen supply and demand imbalance [35]. Periprocedural myocardial injury following TAVI was associated with a significantly increased risk of poor short- and long-term clinical outcomes, including 30-day and 1-year mortality, neurological events, and postprocedural permanent pacemaker implantation [36].

In our study, high oxygen tension during TAVI tended to increase the release of cardiac biomarkers in the first 72 h post-TAVI compared to the low oxygen tension group, but the difference was not significant. There was no clinical impact of the level of intraoperative oxygen tension during TAVI, except that recovery of kidney function was more common in the low- compared to the high-F_IO₂ group.

Acute kidney recovery

Acute kidney recovery, which is a relatively recently described phenomenon, has been observed more frequently than AKI after both TAVI and surgical aortic valve replacement (SAVR) [37]. Following TAVI or SAVR, normalization of the aortic valve area, prompt relief of the trans-aortic pressure gradient, and normalization of post-stenotic flow abnormality occur. Regarding renal blood flow, a rapid increase in cardiac output and reduced LV afterload may cause abrupt hemodynamic changes in the early postprocedural period, such as renal congestion. In a recent prospective registry analysis, both AKI and AKR early after TAVI were independent predictors of cardiovascular mortality [38]. In our study cohort, 10 of the 62 (16%) patients met the criteria for both AKI and AKR during the study period (Table 5). Rapid changes in renal hemodynamics could have occurred in these groups, and both post-TAVI AKI and AKR may reflect a cardiorenal aspect of the extra-cardiac damage characterizing severe AS. Further studies are required to assess the relationship between renal circulatory changes and clinical outcomes in AS patients following TAVI.

Study limitations

This study has some limitations. First, it was a single-center trial with limited number of included patients, and was only powered for one surrogate cardiac biomarker, hs-cTnI. To evaluate the effects of deteriorated power and bias on study data, we performed sensitivity analysis by using bootstrap inference for multiple imputation in the intention-to-treat dataset. Although we observed a trend toward reduced hs-cTnI release and better postprocedural kidney recovery, we cannot definitely conclude that arterial oxygenation is beneficial for patients undergoing TAVI in terms of periprocedural myocardial and renal protection. Second, we included patients undergoing general anesthesia for transfemoral TAVI in this study. However, many patients undergo TAVI under conscious sedation or even local anesthesia, unless they are at very high periprocedural risk due to severely compromised cardiopulmonary function or the inability to maintain a stable supine position, for example. Although conscious sedation is increasingly provided to patients undergoing transfemoral TAVI than general anesthesia [39], anesthetic protocol is determined considering patients’ comorbidity and practitioners’ preference and based on the institutional practice. As many centers, including ours, perform TAVI under general anesthesia [39], our results may contribute to establishing oxygen treatment strategy in this practical context. Future investigators could compare the oxygenation strategies of minimal supplemental oxygen and no supplemental oxygen, as in the treatment of acute MI patients without hypoxia, in terms of the likelihood of avoiding unnecessary periprocedural oxidative stress and protecting multiple organ systems, in patients undergoing TAVI under conscious sedation or local anesthesia. Third, we included relatively low-risk patients; we excluded those who were already hypoxemic or required supplemental oxygen, and those with acute coronary syndrome or renal failure. In high-risk patients with severe LV dysfunction or poor oxygenation, however, different supplemental oxygen strategies may have a differential impact on myocardial injury and other clinical outcomes. Therefore, further studies are required of high-risk patients, who may be more suitable candidates for TAVI. Lastly, there is lack of control subjects undergoing SAVR in evaluating oxygenation and perioprocedural myocardial injury in this study. As it is beyond the primary aim of the present study, future studies can be conducted regarding periprocedural oxygen content and myocardial injury in patients undergoing TAVI vs. SAVR.