Transapical versus transfemoral approach and risk of acute kidney injury following transcatheter aortic valve replacement: a propensity-adjusted analysis

Abstract Background: The aim of this study was to compare the incidence of post-procedural acute kidney injury (AKI) and other renal outcomes in patients undergoing transapical (TA) and transfemoral (TF) approaches for transcatheter aortic valve replacement (TAVR). Methods: All consecutive adult patients undergoing TAVR for aortic stenosis from 1 January 2008 to 30 June 2014 at a tertiary referral hospital were included. AKI was defined based on Kidney Disease Improving Global Outcomes (KDIGO) criteria. Logistic regression adjustment, propensity score stratification, and propensity matching were performed to assess the independent association between procedural approach and AKI. Results: Of 366 included patients, 171 (47%) underwent TAVR via a TA approach. AKI occurrence in this group was significantly higher compared to the TF group (38% vs. 18%, p < .01). The TA approach remained significantly associated with increased risk of AKI after logistic regression (OR 3.20; CI 1.68–4.36) and propensity score adjustment: OR 2.83 (CI 1.66–4.80) for stratification and 3.82 (CI 2.04–7.44) for matching. Nonetheless, there was no statistically significant difference among the TA and TF groups with respect to major adverse kidney events (MAKE) or estimated glomerular filtration rate (eGFR) at six months post-procedure. Conclusion: In a cohort of patients undergoing TAVR for aortic stenosis, a TA approach significantly increases the AKI risk compared with a TF approach. However, the TAVR approach did not affect severe renal outcomes or long-term renal function.


Study population
This was a single-center retrospective cohort study conducted at a tertiary referral hospital. All consecutive adult patients (age 18 years or older) who underwent TAVR via TA or TF approach for aortic stenosis from 1 January 2008 to 30 June 2014 at Mayo Clinic in Rochester, MN, USA were included. For patients with multiple aortic valve replacement procedures, only the first intervention during the study period was included in the analysis. Patients with end-stage renal disease, received any dialysis modalities within 14 days prior to intervention, or did not provide research authorization were excluded. Almost all TAVR procedures (99%) in our institution during study period were performed using balloon-expandable valve under general anesthesia. This study was approved by the Mayo Clinic Institutional Review Board (IRB). Informed consent was waived for patients who provided research authorization.

Data collection
Clinical characteristics, laboratory data, echocardiographic data, procedural data, and post-procedural data were collected using manual and automated retrieval from the electronic medical record. The risk of operative mortality was calculated based on patient demographics, pre-operative clinical characteristics, and procedures utilizing the Society of Thoracic Surgeons (STS) risk score. [19][20][21] The estimated glomerular filtration rate (eGFR) was derived using the Chronic Kidney Disease Epidemiology Collaboration equation. 22

Clinical outcomes
The primary outcome was post-operative AKI. AKI was identified and staged based on the serum creatinine (SCr) criterion of the Kidney Disease Improving Global Outcomes (KDIGO) definition. 23 AKI was defined as an increase in SCr of 0.3 mg/dL (26.5 lmol/L) within 48 h of intervention or a relative increase of 50% from baseline within seven days after intervention. The baseline values for SCr were obtained as follows: if outpatient SCr measurement between 180 and seven days before intervention was available, the most recent value was used. Otherwise, the lowest value within five days prior to intervention was used. Secondary outcomes included in-hospital mortality, need for renal replacement therapy (RRT) at any time during hospitalization, the composite of major adverse kidney events (MAKE), and six-month mortality. MAKE were defined as the composite of in-hospital death, use of RRT, or persistence of renal dysfunction (defined by SCr 200% of reference) at hospital discharge.
Patient vital status was reviewed using registration data and the electronic medical record. For patients with unknown vital status at six months after ICU admission, the Social Security Death Index was used. 24 Among surviving patients at six months after intervention, renal function was assessed by eGFR calculation at six months and evaluation of the need for chronic dialysis. eGFR values of 5 mL/min/1.73 m 2 at six months were imputed for patients who needed chronic dialysis.

Statistical analysis
All continuous variables were reported as mean ± standard deviation (SD). All categorical variables were reported as counts with percentages. The differences in clinical characteristics and outcomes of patients in the TA and TF groups were tested using the Student's t-test for continuous variables and the Chi-squared test, or Fisher's exact test, for categorical variables, as appropriate. A two-sided p values of <.05 was considered statistically significant. Unless specified, analyses were performed using JMP statistical software (SAS, Cary, NC).
When estimating the effect of the TA approach on post-procedural AKI compared to the TF approach, three separate statistical analyses to account for preoperative differences between these two groups: (1) Multivariate logistic regression adjusting for variables with statistically significant differences in the univariate analysis: age, body mass index (BMI), dyslipidemia, stroke, peripheral vascular disease, chronic lung disease, previous percutaneous coronary intervention (PCI), previous cardiac surgery, previous coronary bypass graft surgery (CABG), and pre-operative statin use. (2) Propensity score stratification into quintiles. The pooled risk of AKI after stratification by propensity score was assessed using the Cochran-Mantel-Haenzel method.
(3) Matching on a 1:1 basis on the logit of propensity score with a caliper width equal to 0.25 of the SD of the logit of the propensity score. Propensity scores were estimated using logistic regression with the TA approach as the dependent variable on the following covariates: STS risk score, eGFR, and variables with statistically significant differences in univariate analysis between the two groups, as described above. Propensity score generation, stratification, and matching for patients in the TA and TF groups were performed using R statistical software version 3.1.1 (Vienna, Austria). 25

Results
A total of 366 patients were included in the initial cohort prior to matching: 171 in the TA group and 195 in the TF group. In the unmatched cohort, the TA group was significantly older, had higher BMI, and more prevalence of dyslipidemia, as well as stroke, peripheral vascular disease, history of previous PCI, cardiac surgery, CABG, and pre-operative statin use (Table 1). Intraoperatively, there was no significant difference in procedure duration or the need for intra-aortic balloon pump. However, the TA group more frequently required RBC transfusion than the TF group ( Table 2).
In the unmatched cohort, post-operative AKI occurred in 38% (N ¼ 65) of the TA group: 32% stage 1, 2% stage 2, and 4% stage 3. In contrast, post-operative AKI occurred in 18% (N ¼ 36) of the TF group: 14% stage 1, 2% stage 2, and 3% stage 3 (Table 3). In unadjusted analysis, AKI occurrence in the TA group was significantly higher than in the TF group (OR ¼ 2.71; 95% CI 1.68-4.36; p < .001). There was no significant difference in the need for RRT, in-hospital mortality, MAKE, stroke/TIA, vascular complication, six-month mortality, as well as eGFR at six months. In adjusted analysis, TA approach remained significantly  associated with increased risk of AKI: OR 3.20 (CI 1.87-5.59) (Figure 1). Propensity score matching created a matched cohort of 108 patients in each group. The matching rigorously reduced the difference in covariates between two groups ( Figures S1 and S2). In this matched cohort, baseline characteristics and pre-operative variables were comparable between the TA and TF groups (Table 1). There was no significant difference in procedure duration, need for RBC transfusion, or need for intra-aortic balloon during procedure between the two groups ( Table 2).
In the matched cohort, AKI occurred in 42% (N ¼ 45) of the TA group: 34% stage 1, 2% stage 2, and 6% stage 3. In contrast, AKI occurred in 16% (N ¼ 17) of the TF group: 13% stage 1, 1% stage 2, and 2% stage 3 ( Table 3). The risk of AKI was significantly higher in the TA group (OR ¼ 3.82; CI 2.04-7.44; p < .001) (Figure 1). A similar association was observed after propensity score stratification with OR of 2.83 (CI 1.66-4.80). There was no significant difference in the need for the RRT, in-hospital mortality, MAKE, six-month mortality, as well as eGFR at six months.

Discussion
Comprehensive propensity score analysis in a cohort of severe AS patients undergoing TAVR was used to demonstrate a TA approach significantly increased the risk of AKI compared to a TF approach with an overall 3.82fold increased the risk of AKI. However, there were no significant differences between these two approaches in the setting of severe AKI with respect to RRT, in-hospital mortality, or six-month mortality. Before matching, patients in the TA group were significantly older, had a higher BMI, and more prevalence of dyslipidemia, as well as stroke, peripheral vascular disease, history of previous PCI, cardiac surgery, CABG, and pre-operative statin use.
Despite comparable patient populations in propensity score match analysis, TA-TAVR still carried a significantly higher AKI risk compared to TF-TAVR. Therefore, the mechanisms of this observation of higher AKI risk are likely beyond the fact that patients are undergoing TA-TAVR usually have more severe peripheral vascular disease. It is possible that the instrumentation of the aorta during the TA-TAVR may lead to cholesterol crystal emboli and the dislodgement of calcium plaques, resulting in AKI. 10 TF-TAVR can also be performed under moderate sedation and local anesthetics, while TA-TAVR is usually performed under general anesthesia, which can contribute to renal hypoperfusion. 26 Although patients undergoing TA-TAVR developed AKI more frequently compared to TF-TAVR, the primary outcome difference was in stage 1 AKI and did not translate into more major adverse events or six-month mortality. Comparable mortality risk among patients undergoing TA-TAVR and TF-TAVR was recently demonstrated by Schymik et al. 27 Using propensity score matching, no difference was reported in mortality risk between patients undergoing TA and TF approaches in a high volume treatment center. An experienced heart  (11) 14 (8) .40 10 (9) 8 (7) .62 Continuous variables are reported as mean ± SD; Categorical variables are reported as count (percentage). a Analysis was limited in survivor six months after procedure. team was postulated to be an important explanation for this observation. However, in PARTNER 2 trial, 28 which studied intermediate-risk patients, TAVR resulted in a lower rate of death or disabling stroke than surgery in the TF-access cohort, whereas outcomes were similar in the two groups in the transthoracic-access cohort.
There are several limitations to our study. (1) As we utilized data from a prospectively collected database and electronic health records, our study has an observational design. Our results establish an association, not causality. Additionally, this study was conducted in a tertiary referral center serving a predominantly Caucasian population, potentially introducing selection bias and limiting the generalizability of the results. (2) We did not include urine output criterion for AKI diagnosis. The inclusion of urine output criterion would increase the sensitivity of AKI definition. However, an indwelling urinary catheter was not used to obtain accurate hourly urine output data for a significant percentage of patients in our cohort. Also, diuretics were commonly utilized in TAVR patients enrolled in our study. Therefore, it could reduce the accuracy of urine output criterion for AKI diagnosis.
In conclusion, TA-TAVR is associated with higher incidence of AKI compared to TF-TAVR. However, this increased frequency of mild AKI associated with TA-TAVR does not translate into higher mortality, need for RRT, or long-term renal function.