This retrospective cohort study included all patients with severe symptomatic AS who underwent TAVR between February 2010 and July 2022 at the Kaplan Medical Center, Rehovot, Israel. Patients were categorized into two groups, those with or without DM.

Inclusion Criteria: Patients aged 18 years and older who underwent TAVR for severe AS were included in the study. Eligible patients receiving self-expandable valve (SEV) or balloon-expandable valve (BEV), had complete periprocedural data and follow-up data.

Exclusion Criteria: We excluded patients without available blood tests, those who experienced periprocedural death (up to 24 hours post-TAVR), and individuals who underwent non-transfemoral access TAVR. Additionally, we excluded 12 patients who had other types of valves implanted. Other exclusion criteria included patients with medical conditions that could affect blood glucose levels, such as malignancies, concurrent infections, chronic inflammatory diseases, and those receiving glucocorticoid therapy within three months before admission.

The data were collected retrospectively from the hospital database, which included the baseline characteristics, lab tests, procedural data, and clinical outcomes, and are illustrated in Tables 1,2 below.

The decisions regarding TAVR vs. AVR, valve type and size, and access were made by a multidisciplinary group consisting of an interventional cardiologist, a cardiac surgeon, an echocardiographic specialist, and a radiologist. The decision to perform a pre-TAVR balloon aortic valvuloplasty was left to the interventional cardiologist at the start of the procedure. The procedures were performed in a hybrid room with a cardiac surgeon on standby. Before undergoing TAVR, all patients received a standard coronary preoperative evaluation and provided informed consent for the procedure and the data collection and analysis.

The SEV-treated patients were implanted with the following valves: corevalve, evolute-R, or evolute-PRO (Medtronic, Inc., Minneapolis, MN, USA). BEV-treated patients were implanted with the following valves: Sapien, Sapien XT, or S3 (Edwards Lifesciences, Irvine, CA, USA). The percutaneous approach and the safety wire technique, along with the Prostar XL vascular closure device (Abbott Vascular, Redwood City, CA, USA), were used for trans-femoral artery access and closure. We considered the procedure duration to be “skin to skin”. The start time was when arterial access was obtained, and the end time was when this access was closed. The standard approach was local anesthesia with conscious sedation. After inserting the femoral sheet, we used heparin to maintain a minimum active clotting time (ACT) of over 250 seconds. If needed, a protamine-heparin antagonist (1 mg for each 100 U of heparin) was administered at the time of vascular closure.

Blood samples were obtained using a sterile syringe without stasis. Laboratory analyses were done pre-procedure, daily post-procedural during the intensive care unit (ICU) stay, and during the physician’s visit to the cardiology ward. The blood test data were retrospectively collected.

Fasting pre-procedural blood glucose (BG) levels were obtained from the blood tests drawn on the morning of the procedure.

The standard follow-up visits were 30 days and six months after being discharged from the hospital. All follow-ups were performed at our medical center. The post-procedural events were defined using the criteria from the Valve Academic Research Consortium 3 (VARC-3). Data on mortality were collected from medical records after a 12-year follow-up period.

We used a two-sample means independent t-test to determine the sample size required for our study comparing controlled and uncontrolled hyperglycemic states in TAVR patients. Our study aims to compare diabetic and nondiabetic patients undergoing TAVR and to examine the impact of periprocedural blood glucose levels on clinical outcomes. Our sample size calculations were based on parameters estimated from previous studies that reported mean creatinine values and variances [9, 10]. This analysis indicated that a total sample size of 39 participants is needed to achieve a power of 0.80 at a significance level of 0.05. Specifically, this includes 13 participants in the hemoglobin A1c (HbA1c) 5.7 group and 26 participants in the HbA1c $\ge$7 group, maintaining a sample size ratio of 2:1.

The expected effect size of –0.33 suggests a moderate difference in means between the groups, with the HbA1c 5.7 group having a mean creatinine of 0.82 and the HbA1c $\ge$7 group having a mean creatinine of 1.15. Additionally, the standard deviations for the HbA1c 5.7 group and the HbA1c $\ge$7 group are 0.19 and 0.37, respectively, reflecting the unequal variances that will be addressed using Satterthwaite’s t-test.

This sample size will ensure adequate power to detect differences in outcomes as we compare diabetic and nondiabetic patients undergoing TAVR while also examining the impact of periprocedural blood glucose levels on clinical outcomes.

The categorical variables are presented as frequencies and percentages, and continuous-type variables are presented as mean $\pm$ standard deviation (SD). We tested normality between the various study groups’ continuous variables using a Shapiro-Wilk test. A Mann-Whitney non-parametric test was performed when abnormal distributions were found. When appropriate, we performed a Pearson’s chi-square test for categorical variables. Main effect estimates are presented with their 95% confidence intervals.

We divided the patients into two groups; patients with DM were identified based on whether the patient had this diagnosis in previous records or was on prescribed antidiabetic treatment (oral medications or insulin). The other group was patients free of DM. We further subdivided the DM patients into several groups according to pre-procedural BG levels (BG 100 mg/dL, BG 101–125 mg/dL, and BG $\ge$126 mg/dL). These values were determined based on the universally accepted definition of hyperglycemia in clinical trials and according to baseline HbA1c levels (HbA1c 5.7 and HbA1c $\ge$7), which indicate controlled and uncontrolled hyperglycemic states respectively (Fig. 1) [11, 12, 13, 14, 15, 16].

Fig. 1.

Study flowchart of patients from 2010 to 2022. A total of 683 patients were treated with TAVR. After excluding 68 patients, a total of 615 patients were finally included in the analysis. BG, blood glucose; DM, diabetes mellitus; TAVR, trans-aortic valve replacement; HbA1c, hemoglobin A1c.

We used the Kaplan Meier test for post-six-month cumulative survival analysis to evaluate the survival between the groups. A log-rank test was used when appropriate. p values 0.05 were defined as statistically significant. We use logistic regression multivariable analysis to adjust for confounding factors.

All statistical analyses were done using the SPSS program (Statistical Package for the Social Sciences) (IBM, version 27.0) (IBM Corp., Armonk, NY, USA). This study was approved by Kaplan Medical Center’s local Helsinki Committee.

Fig. 1 shows the study’s flowchart. Six hundred and eighty-three patients were identified for the twelve-year study period. Sixty-eight participants were excluded from the study. The analyzed population included 615 patients (46% male, mean age 81.4 $\pm$ 7.3 years) with symptomatic severe AS (mean trans-aortic pressure gradient 47.1 $\pm$ 16.3 mmHg) and prohibitive or high operative risk (society of thoracic surgeons (STS) score of 7.93 $\pm$ 1.6). The baseline, procedural aspects of the study population, and the two subgroups are summarized in Table 1. Diabetic patients were younger (80.7 $\pm$ 6.8 vs. 82.0 $\pm$ 6.8 years, p = 0.001) and had a higher body mass index (29 $\pm$ 5.1 vs. 27.5 $\pm$ 4.7 kg/m², p 0.001) than non-diabetic patients. DM patients had more cardiovascular risk factors, including peripheral vascular disease (20.1% vs. 14.4% in non-diabetics, p = 0.059), previous coronary bypass surgery (11.5% vs. 5.4%, p = 0.017), and higher cholesterol levels (153 $\pm$ 41 mg/dL vs. 162 $\pm$ 39 mg/dL in non-diabetics, p = 0.003). There was no significant difference in blood pressure between the two groups. Patients with DM had a higher serum creatinine level of 1.31 $\pm$ 0.93 mg/dL compared to 1.15 $\pm$ 0.90 mg/dL in those without DM (p = 0.001). Additionally, the estimated glomerular filtration rate (e-GFR) was lower in the DM group at 52.42 $\pm$ 24.3 mL/min/1.73 m² versus 64.10 $\pm$ 17.2 mL/min/1.73 m² for non-DM patients (p = 0.012).

The risk of 30-day mortality (24 hours to 30 days) was identical among patients with and without DM (8.8% for DM versus 12.7% for no DM, p = 0.310; Table 2). Patients with DM exhibited a higher one-year mortality rate (10.4% vs. 5.8%; p = 0.036).

Comparing the 30-day outcome according to VARC 3 criteria, patients with and without DM had comparable stroke rates (4.8% vs. 4.1%, RR 0.9, p = 0.649), major or life-threatening bleeding (8.9% vs. 9.3%, RR 1.1, p = 0.869), myocardial infarction (1.2% vs. 2.1%, RR 1.4, p = 0.374), and arrhythmias (26.7% vs. 30.2%, RR 1.0, p = 0.657). The median length of stay was two days in both groups (p = 0.96). Even though the patients with DM had decreased kidney function at baseline compared with patients without DM, there was no increase in acute kidney injury or the need for dialysis in this group. Table 2 presents clinical outcomes in the study population. In Fig. 2, the p-value of 0.12, derived from the Kaplan-Meier curve comparing mortality rates between diabetic and non-diabetic patients, indicates that there is no statistically significant difference in survival between the two groups at the significance level of 0.05.

Fig. 2.

Kaplan Meir curve based on all cause mortality for patients with and without DM, showing no difference in survival between the two groups. DM, diabetes mellitus.

In diabetic patients, 91 (33%) were uncontrolled (HbA1c $>$7), and 46 (17%) were well-controlled (HbA1c 5.7) (Fig. 3). We divided the DM patients into subgroups: BG 100 mg/dL (54 patients (20%)) and BG $\ge$126 mg/dL (170 patients (62%)). We found no differences in the baseline characteristics, basic lab tests, and procedural indexes between the subgroups (Table 3).

Fig. 3.

Simple boxplot for differences between subgroups in thirty-day outcomes of diabetic patients (n = 270) after TAVR according to VARC 3 criteria and pre-procedural blood glucose (BG) and hemoglobin A1c (HbA1c) levels. (A) Mortality, (B) Arrhythmia, (C) Stroke, (D) Major vascular complications, (E) Acute kidney injury. TAVR, transcatheter aortic valve replacement; VARC 3, Valve Academic Research Consortium 3; AKI, acute kidney injury.

Two hundred and seventy of the included participants suffered from diabetes mellitus with a mean HbA1c of 7.4 $\pm$ 2.5. We measured the level of pre-procedural BG for all diabetic patients (mean 134 $\pm$ 45 mg/dL). Patients who experienced mortality had a median HbA1c of 6.9 [6.4–8.1], compared to 6.8 [6.1–7.7] for those who did not, with a p-value of 0.108, indicating no significant difference. Conversely, arrhythmia was associated with a higher HbA1c of 7.1 [6.3–8.6] compared to 6.7 [6.2–7.5] (p = 0.061), and BG levels were significantly elevated in those with arrhythmia (177 [139–228] vs. 155 [129–197], p = 0.026). For stroke, the median HbA1c was significantly higher at 7.6 [6.7–8.5] compared to 6.8 [6.2–7.7] (p = 0.048), while BG levels showed a trend without significance (165 [157–240] vs. 160 [129–203], p = 0.085). Major vascular complications did not show significant differences in HbA1c (7.0 [6.4–8.3] vs. 6.8 [6.2–7.7], p = 0.427) or BG (160 [134–218] for both groups, p = 0.987). However, for acute kidney injury (AKI), the median HbA1c was significantly higher at 7.6 [6.5–8.8] compared to 6.7 [6.2–7.6] (p = 0.029), and BG levels were also significantly higher (190 [157–210] vs. 158 [128–199], p = 0.015). Overall, these findings indicate that elevated HbA1c and BG levels are linked to complications such as arrhythmia, stroke, and AKI, underscoring the importance of glycemic control in patients undergoing TAVR (Fig. 3, Table 4).

Table 5 presents the results of a multivariable logistic regression analysis examining the impact of blood glucose levels and HbA1c on outcomes such as arrhythmia, stroke, and AKI in patients with diabetes mellitus (N = 270). The analysis shows that for arrhythmia, the odds ratio for HbA1c is 1.1039 (p = 0.23), indicating no significant association, while blood glucose has an odds ratio of 0.998 (p = 0.76), also showing no significant effect. In the case of stroke, HbA1c presents an odds ratio of 1.194 (p = 0.36), suggesting no meaningful impact, but blood glucose is significantly associated with stroke, exhibiting an odds ratio of 1.020 (p = 0.013). Notably, for AKI, the odds ratio for HbA1c is 2.304 (p = 0.02), indicating a significant link between higher HbA1c levels and increased risk of AKI. Blood glucose levels also trend toward significance with an odds ratio of 1.0137 (p = 0.061).