Literature search results
A total of 110 potentially relevant citations were identified and screened from the initial search. After the removal of duplicated studies, we retrieved 11 full-text articles for evaluation of which 4 observational studies satisfied our selection criteria [10,16-18]. Preferred Reporting Items for Systemic Reviews and Meta-Analyses (PRISMA) flow chart of the study selection is shown in Figure 1. Major excluded articles [19, 34-39] with reasons are reported in the Supplemental Appendix. The 4 included studies enrolled a total of 2054 patients; 164 patients with prior C-XRT, and 1890 patients without prior C-XRT. The summary of the included studies and their main findings are shown in Table 1 and the baseline characteristics of their population are shown in Table 2.
Risk of bias of the included studies
The included studies were together at moderate risk of bias according to the ‘Newcastle-Ottawa Scale’ assessment tool. The study of Dijos et al. has a very small number of patients in the C-XRT group as compared to the control group, and have not been adjusted adequately to the control group population in terms of age and peri-operative risk score [16]. The study of Bouleti et al. also has a small but equal number of patients in the comparison groups with a fair adjustment of the confounding factors between the comparison groups [17]. The studies of Agrawal et al. and Gajanana et al. have good quality selection with comparable patient-cohorts that are adjusted adequately for the confounders [10, 18]. The summary of the quality assessment domains from the included studies is shown in Table 3.
All-cause mortality
We analyzed the all-cause mortality at 30-day and 1-year follow-ups. The 30-day mortality outcome was reported in the four included studies and 1-year mortality was reported in the three included studies except Dijos et al. [16]. There was no statistically significant difference in the all-cause mortality at the 30-day follow-up when comparing the C-XRT group to the control group (OR 1.29, 95% CI 0.64 to 2.58, p=0.48). However, the C-XRT group showed statistically significant higher all-cause mortality at 1-year follow-up compared to the control group (OR 1.97, CI 1.15 to 3.39, p=0.01). The forest plots for the all-cause mortality at 30-day and 1-year follow-ups are shown in Figures 2A and 2B, respectively.
Safety outcomes (at 30-day follow-up)
Stroke (any): This outcome was reported in all four included studies. According to pooled analysis, the C-XRT group suffers similar rates of strokes compared to the control group (OR 2.87, 95% CI 0.83 to 9.93, p=0.10). The forest plot is shown in Figure 3A.
Major Bleed: This outcome was reported in all four included studies. There was no statistically significant difference in the major bleeding events between the comparison groups (OR 1.30, CI 0.72 to 2.33, p=0.38). The forest plot is shown in Figure 3B.
Access-related vascular complications: This outcome was reported in three included studies except for Agrawal et al. because it was not reported [10]. There was no statistically significant difference in access-related vascular complications in between the comparison groups (OR 1.15, CI 0.48 to 2.77, p=0.75). The forest plot is shown in Figure 3C.
Need for a pacemaker: This outcome was reported in all four included studies but we included data from three studies except for Agrawal et al. because it did not report the pacemaker implantation outcome at the 30-day follow-up [10]. According to pooled analysis, there was no statistically significant difference in the need for a pacemaker between the comparison groups (OR 0.95, CI 0.42 to 2.17, p=0.91). The forest plot is shown in Figure 3D.
Efficacy outcomes
Left ventricular ejection fraction: This outcome was reported in all four included studies. There is no statistically significant difference between the comparison groups (OR 1.23, CI -0.51 to 2.96, p=0.17). The forest plot shown in Figure 4A.
Mean aortic valve gradient: This outcome was reported in all four included studies but we analyzed data from three studies except for Gajanana et al. because it reported post-procedural mean aortic valve gradients as the difference in mean gradients with standard deviations [18]. We were unable to calculate appropriate data because co-variance was not reported by the author [18]. The pooled analysis showed no statistical significance between the comparison groups (OR -0.59, CI -1.42 to 0.24, p=0.17). The forest plot is shown in Figure 4B.
Post-procedural worsening of congestive heart failure
This outcome was reported in all four studies. The pooled analysis showed significantly higher rates of worsening congestive heart failure (CHF) in patients with prior C-XRT as compare to those without C-XRT (OR 2.03, CI 1.36 to 3.04, p=0.0006). The forest plot is shown in Figure 4C.