Introduction

The most common form of heart valve disease in western countries is aortic valve stenosis. [1] Its prevalence among patients older than 65 years of age is 2% to 7%. [2] Etiologically, the most common cause is valve calcification. [1] Surgical aortic valve replacement is currently the gold standard for treating aortic valve stenosis in low- or intermediate-risk patients. [3]

Transcatheter aortic valve implantation (TAVI) is a minimally invasive surgical procedure that was initially developed as an alternative treatment for patients with a high level of perioperative risk, and it has since established itself in this field. [4] TAVI involves the implantation of a prosthetic valve mounted on a stent and introduced with a catheter through transfemoral (TF), transapical (TA), transaxillary/subclavian or direct transaortic access. [5] Usually, the TF approach is preferred, because thoracotomy and penetration of the myocardium are not needed. The TA approach is common, if severe atherosclerotic disease does not allow retrograde insertion of the catheter. [6] The first TAVI was performed by Cribier et al. in 2002. [7] In 2010, 4,859 TAVIs were performed in Germany. In 2013, minimally invasive procedures (10,441) exceeded open surgical (9,899) procedures for the first time. [8] Because the number of surgical interventions remained virtually constant during this time, TAVI was used for patients for whom only a conservative approach had been previously available. While the first TAVI was performed under monitored anaesthetic care (MAC), general anaesthesia (GA) was typically utilized in the early years. [7] In recent years, TAVIs, through TF access, have often been performed under local anaesthesia and MAC in Germany. [8] For anaesthesiologists, TAVI is challenging for many reasons. First, many patients have a serious pre-existing condition and these are therefore high-risk patients. [5] Second, these patients are exposed to marked hemodynamic fluctuations during the TAVI procedure, for example, due to functional heart arrest through rapid pacing. To mitigate the risk of cardiopulmonary decompensation, a highly specific and interdisciplinary management approach is required, possibly including the deployment of extracorporeal membrane oxygenation (ECMO). [9]

For the present article, all TAVI interventions performed at our hospital using TF and TA access were evaluated retrospectively. The goal was to collect key information about the different types of operations, typical complications and procedural parameters in the context of benchmarking and to compare these results with the current literature.

Materials and methods

Results

During the time period investigated, 853 TF- and TA-TAVI patients were included in the study. Data for 100% of these patients were analysed. TF access was selected for 506 patients (59%), with TA chosen for 347 patients (41%). The demographic data and pre-existing conditions are shown in Table 1. The TA patients’ average logistic EuroScore (TA: 21 ± 15% vs. TF: 18 ± 12%; p = 0.002) and their ASA-class (p = 0.033) were significantly higher than of the TF patients’. The elective use of ECMO was similarly frequent in both groups (TF: n = 24 (5%), TA: n = 15 (4%); p = 0.868).

Download:

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

Table 1. Demographic data.

https://doi.org/10.1371/journal.pone.0193558.t001

Overall, TA patients had significantly more complications per patient than TF patients (0.7 vs. 0.4; p = 0.001). Subsequently, the complications are addressed in detail. All complications are summarized in Table 2.

Download:

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

Table 2. Complications.

https://doi.org/10.1371/journal.pone.0193558.t002

Mortality

The overall hospital mortality rate in our study was 6% (50 patients). There were 19 deaths among the TF patients and 31 deaths among the TA patients (n = 19 (4%) vs. n = 31 (9%); p = 0.002). The causes of death divided into TF and TA are shown in Fig 1. TF patients died significantly more often by bleedings (n = 9 (47%) vs. n = 4 (13%); p = 0.018). CVEs were also a more frequent death cause in TF patients, but without reaching statistical significance (n = 3 (16%) vs. n = 2 (7%); p = 0.355). TA patients died more often due to cardiac reasons (including MI, acute heart failure and cardiac arrest) or infections than TF patients, but also without reaching statistical significance.

Download:

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

Fig 1. Causes of death (in-hospital) with TF (transfemoral) and TA (transapical) access.

All data are presented as percentage. P-values were calculated from raw data: Cardiac (MI, acute heart failure and cardiac arrest): p = 0.148; Acute kidney injury (AKI): p = 1; cerebrovascular events (CVE): p = 0.355; bleeding: p = 0.018; infection: p = 0.134.

https://doi.org/10.1371/journal.pone.0193558.g001

Time specifications

The average anaesthesia coverage time for TA and TF was 174 ± 49 min. The mean anaesthesia coverage time for TA access (180 ± 50 min) was significantly longer than that for TF access (170 ± 48 min; p = 0.007). However, no difference was observed between TA and TF in terms of the induction time (average: 37 ± 12 min), the procedure time (average: 87 ± 44 min) or the emergence time (average: 7 ± 7 min). Significant differences between TF and TA were observed concerning the length of stay at ICU (TF: 35 ± 52 h, TA: 57 ± 84 h; p < 0.001), IMC (TF: 17 ± 51 h, TA: 60 ± 92 h; p < 0.001) and in hospital (TF: 10 ± 7 d, TA: 14 ± 11 d, p < 0.001) and the time to mobilization (TF: 20 ± 18 h, TA: 34 ± 37 h; p < 0.001) (Table 3).

Download:

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

Table 3. Time specifications.

https://doi.org/10.1371/journal.pone.0193558.t003

Discussion

This retrospective study enrolled data of 853 TAVI patients. TA access was associated with a higher ASA-status, EuroScore and complication rate. AKI, pneumonia sepsis and mortality were more often documented in TA group than TF group. In contrast, vascular complications occurred more frequently in the TF group. Length of ICU, IMC and total hospital stay was prolonged in the TA group.

The preoperative assessment revealed a higher ASA status and EuroScore for TA patients than for TF patients. A higher ASA status corresponds to a higher incidence of perioperative morbidity and mortality. [10] The EuroScore is the most widely used risk index for open cardiac surgery, and it may be seen as an indicator for the TA patients to have a higher probability to die due to the surgery. Although, this tool is primarily validated for open heart surgery, at this moment it is also accepted to evaluate the TAVI risk. [11] Taken together, the anaesthesiologic, as well as the operative risk for TA patients was higher than for TF patients. The increased length of stay on ICU, on IMC or in hospital may be attributed to this increased assessed risk by itself. However, in the TA group more complications were documented, which might even influence the length of the hospital stay. It has to be kept in mind, that the TF approach is commonly the first choice when a TAVI is under consideration. Only, if the condition of the patient’s peripheral vessels does not allow TF TAVI, the TA approach is chosen, leading maybe to an overall sicker patient population. [12]

In the TF group, vascular complications were 6%. In contrast, TA patients had a significantly lower rate of vascular complications (2%; p = 0,013). These numbers correspond to those described in others studies. There, vascular complications occurred in 4–16.4% of TF-TAVI procedures and 1–4% for TA-TAVI procedures [13–16]. Commonly, in TF access ilio-femoral vessels are damaged mainly due to the delivery system (6). However, except for vascular complications, in the TA group more complications were documented.

Most prominent was AKI in the TA group (TA: 18%, TF: 5%; p < 0.001). Whereas these number correspond to numbers presented in other studies. There, the incidence of AKI is up to 21.9% in TF patients and up to 44.4% in TA patients. [15,17] Barbash et al. have investigated the predictors for AKI, specifically for TAVI. [18] The sole independent risk factor identified by Barbash et al. was the transfusion of erythrocyte concentrates. In our study, the patients who developed an AKI had a preoperative haemoglobin level that was significantly lower than the level in patients who did not develop an AKI (11.7 g/dl vs. 12.4 g/dl, p = 0.004). In addition, the intraoperative and postoperative transfusions of erythrocyte concentrates were associated with a significantly increased incidence of AKI. Previous research has already shown that the transfusion of erythrocyte concentrates does not constitute an adequate treatment of perioperative anaemia in all cases. Indeed, the transfusion is associated with increased morbidity and mortality rates. Early diagnosis and prevention of anaemia is recommended in the context of patient blood management programs to ensure a more targeted and effective deployment of transfusions. [19] Overall, the AKI patients in our study were a subgroup featuring a particularly poor preoperative state of health (average EuroScore: 27% vs. 18%, p < 0.001).

In terms of pneumonia and TAVI, we identified only a single study from Covello et al., which reported a pneumonia rate of 7–8%. [20] Our incidence of pneumonia was also 7% (TF 6% vs. TA 10%; p = 0.044). According to Lynch et al., the risk factors for hospital-acquired pneumonia include mechanical ventilation for > 48 h and a long length of stay in the ICU and the hospital. [21] Our study confirmed these risk factors. According to Hausman et al., COPD patients are generally exposed to a higher risk of pneumonia for non-cardiosurgical interventions. [22] In our study, the COPD patients also tended to contract pneumonia more frequently (11% vs. 7%, p = 0.172), although the incidence was not statistically significant.

According to the literature, the rate of CVE in TF- and TA-TAVI patients is 1–3.8%. [13–17] Differences in the incidence of CVEs between TF and TA have not been described. In our study, 2% of TF patients and 3% of TA patients suffered a CVE, which is comparable to the literature data.

The incidence of MIs in TAVI patients is up to 1.5%. [13,14,17] In our study, the incidence was 1%, with no difference between TF (1%) and TA (1%). Although, due to the design of the study a distinction of the type of MI was not possible, there was a statistically significant association between the complications due to intervention and MI. The incidence of MIs in patients with complications caused by intervention was 3.7%, while only 0.5% of patients without interventional complications suffered an MI (p = 0.022).

According to the literature, 8.7–28.7% of TF and TA patients need a permanent PM. [13–15] Differences between TF and TA are not reported, which is supported by our findings (TF: 12%, TA: 14%; p = 0.46). Studies have cited existing conduction disorders as a risk factor for postoperative PM dependency for TAVI and for conventional aortic valve replacement. [23,24] In our study, this finding was supported by the association between an existing first-degree AV block, an RBBB and an incomplete LBBB and the implantation of a permanent PM. Typically, patients with a higher-degree AV block at the preoperative stage had already been fitted with a PM. It is assumed that the TAVI procedure may cause mechanical damage to the conduction system due to the anatomical proximity to the aortic valve of a prosthesis that is implanted too deeply. [25] Also, the valve model by itself has an influence on the PM incidence.

According to the literature, a procedure change from a TAVI to an open surgical intervention occurs in 0–1% of cases. [13,14] In our study procedure changes were necessary in 1% of cases and mainly resulting from surgical complications.

The mortality rate of TF patients ranges from 4.2% to 8.7%. [13,14,17] The mortality rate of our TF patients (4%) is comparatively low. The TA patient mortality rate was significantly higher (9%). In the literature, similarly high mortality rates for TA patients (8–8.8%) can be found. [15,16] In accordance with our results, Panchal et al. reported in their meta-analysis a higher mortality in TA. [26] As in our study, they found a higher logistic EuroScore in the TA group, which might have been linked to higher all-cause mortality. However, other possible explanations, e.g. a more invasive nature of this approach and its steeper learning curve might be attributable for increased mortality in TA approach.

A review of the literature for TF access revealed procedure times of 75 ± 40 min to 167.6 ± 5.2 min. [17] The major discrepancies between the studies can probably be attributed to the lack of a uniform definition of procedure time. In our study, the average procedure time was 87 ± 44 min (TF: 86 ± 45 min, TA: 91 ± 43 min). According to the literature, the postoperative stay for TF-TAVI patients ranges from 3 ± 4.45 d to 15.5 ± 10.34 d. [13,14,17] For our TF patients, the average length of hospital stay was 10 d. The TA patients spent a significantly longer time in the hospital (14 d) than the TF patients. Our average length of hospital stay appears to be relatively long, but it has to be considered, that the length of stay is strongly dependent on the local health care system, e.g. the DRG system.

Conclusions

In the present study, we showed that our patients’ outcomes are comparable to those reported in the available literature and that TA patients have an overall worse physical condition as well as a higher perioperative morbidity and mortality. Consequently TA patients need additional care and should only be operated in appropriately experienced medical centres.

Limitations

During the study period, the proportion of TF-patients in the collective continuously grew to 74.4% in 2015. Furthermore, operative techniques developed over the years and led to a significant reduction of complication rates and significant better patient outcome. Therefore, comparative interpretation of data between the two access groups must be cautious. Another limitation is that we did not use the standardized endpoint definitions for transcatheter aortic valve implantation (VARC-2). [27] urthermore, as we extracted the diagnosis of AKI, MI, CVE or pneumonia directly from our patient data management system, our retrospective study design did not allow us to discriminate between the different types. Additionally, only those data are presented that were documented in the files that were evaluated. Therefore, a comparison with other studies might be limited.